Abstract
The Lyme disease agent, Borrelia burgdorferi, is able to persistently infect humans and animals for months or years in the presence of an active immune response. It is not known how the organisms survive immune attack in the mammalian host. vlsE, a gene localized near one end of linear plasmid lp28-1 and encoding a surface-exposed lipoprotein in B. burgdorferi B31, was shown recently to undergo extensive genetic and antigenic variation within 28 days of initial infection in C3H/HeN mice. In this study, we examined the kinetics of vlsE sequence variation in C3H/HeN mice at 4, 7, 14, 21, and 28 days and at 7 and 12 months postinfection. Sequence changes were detected by PCR amplification and sequence analysis as early as 4 days postinfection and accumulated progressively in both C3H/HeN and CB-17 severe combined immunodeficient (SCID) mice throughout the course of infection. The sequence changes were consistent with sequential recombination of segments from multiple silent vls cassette sites into the vlsE expression site. No vlsE sequence changes were detected in organisms cultured in vitro for up to 84 days. These results indicate that vlsE recombination is induced by a factor(s) present in the mammalian host, independent of adaptive immune responses. The possible inducing conditions appear to be present in various tissue sites because isolates from multiple tissues showed similar degrees of sequence variation. The rate of accumulation of predicted amino acid changes was higher in the immunologically intact C3H/HeN mice than in SCID mice, a finding consistent with immune selection of VlsE variants.
Lyme disease in humans and animals is a multisystemic disorder caused by infection by a genetically diverse group of spirochetes that includes Borrelia burgdorferi (9), Borrelia afzelii (12), and Borrelia garinii (1). These pathogenic spirochetes are transmitted to individuals through the bite of an infected ixodid tick (5). In untreated individuals, Lyme disease spirochetes can persist for months or years in human patients and other mammalian hosts in the presence of an active immune response (26). Mechanisms for long-term survival of Lyme disease spirochetes in mammalian hosts are not well understood.
A genetic locus designated vls for vmp-like sequence in B. burgdorferi B31 clone 5A3 (B31-5A3) was recently identified (Fig. 1A) (33). The vls locus shares sequence homology and recombinatory features with the vmp system for variation of variable major proteins (VMPs) in the relapsing fever agent, Borrelia hermsii (4, 33). VMPs have been divided into small variable protein (Vsp) and large variable protein (Vlp) families based on size and sequence differences (10, 17). The Vsp family also includes B. burgdorferi OspC, due to sequence homology (10, 17). VlsE (the protein product from the B. burgdorferi vls expression site) is similar to large VMPs (33) and therefore belongs to the Vlp family. Plasmids hybridizing to a B31-5A3 vlsE probe were present in all high-infectivity B. burgdorferi, B. afzelii, and B. garinii strains tested (33). Recent analyses verify that other B. burgdorferi strains contain vls sequences, although significant heterogeneity is present (18, 19).
FIG. 1.
Overall experimental strategy for examining the kinetics of vlsE variation. (A) The overall structure of the vlsE and silent vls cassette loci in B31-5A3 as previously described (33). (B) Infection of C3H/HeN mice with low-passage B. burgdorferi B31-5A3. Organisms were cultured by using skin biopsies obtained on days 4, 7, 14, 21, and 28 after initial infection. Clonal populations were prepared by subsurface plating and used as a source of DNA templates for PCR amplification and sequence analysis of the vlsE cassette region. DR, 17-bp direct repeats.
The vls system is located on a 28-kb linear plasmid (lp28-1) in B. burgdorferi B31. The vls system consists of the 1-kb vlsE gene and 15 silent vls cassettes of 474 to 594 bp (Fig. 1A) (33). The recently completed genome sequence of B. burgdorferi B31 corroborated the sequence for the silent cassette region, but vlsE was not present due to the underrepresentation of telomeric sequences (15). During experimental mouse infection, the vlsE cassette region undergoes extensive segmental recombination with the vls silent cassettes via a gene conversion mechanism (33, 34). The purpose of the present study was to determine the earliest detectable occurrence and the frequency of vlsE sequence variation in B. burgdorferi B31 during the course of experimental infection in mice.
MATERIALS AND METHODS
Bacterial strains and cultures.
The high-infectivity B. burgdorferi B31 clone 5A3 (B31-5A3) was originally isolated from low-passage strain B31 and characterized according to infectivity by Norris et al. (22). The vls system in strain B31-5A3 was subsequently identified and characterized by Zhang et al. (33). B. burgdorferi clones M1e4A and M1e4C were isolated from an ear biopsy specimen from a C3H/HeN mouse infected 28 days previously with B31-5A3 (33).
Animal studies.
Eight-week-old, female C3H/HeN mice (Harlan Sprague-Dawley, Houston, Tex.) and CB-17 severe combined immunodeficient (SCID) mice (Charles River Laboratories, Wilmington, Mass.) were housed in microisolator cages and provided with antibiotic-free food and water ad libitum. For mouse inoculation, frozen stocks of B. burgdorferi strains that had previously undergone no more than three in vitro passages since cloning were cultured in BSK II broth (3) at 34°C for 7 days as previously described (22). The cultures were diluted in BSK II broth to a concentration of 106 cells/ml as determined by dark-field microscopy, and 0.1 ml (105 organisms) was injected subcutaneously at the base of the tail. For analysis of vlsE cassette sequence variation during in vitro culture, the original stock of B31-5A3 was consecutively subcultured every 7 days for up to 12 passages. Aliquots of the cultures from each passage were stored in 15% (vol/vol) glycerol at −70°C. In the kinetics study, each of five C3H/HeN and five CB-17 SCID mice were inoculated with 105 B31-5A3. Skin biopsies (∼2 mm in diameter) from the back of each mouse were acquired under aseptic conditions at different time points for culture of B. burgdorferi in 6 ml of BSK II broth with rifampin and phosphomicin (33). All mice were sacrificed on day 28 postinfection, and samples from four tissue sites (bladder, heart, joint, and skin) were cultured. Aliquots of all cultures from mouse tissues were preserved as stocks in 15% glycerol at −70°C. Clonal populations of B. burgdorferi cultures were prepared by subsurface plating and selection of colonies as described previously (22).
For the chronic infection study, we kept eight C3H/HeN mice that were infected with 105 cells of B31-5A3 in our previous study (33) for 12 months. Skin biopsies were obtained from each of these mice at 7 months after initial infection and cultured for B. burgdorferi in 6 ml of BSK II broth. All mice were sacrificed 12 months after the initial infection, and samples from four tissue sites (bladder, heart, joint, and skin) were cultured.
For infection studies with B. burgdorferi clones M1e4A and M1e4C, four C3H/HeN mice were inoculated with 105 cells as described above. All mice were sacrificed 28 days after infection, and samples from four tissue sites (bladder, heart, joint, and skin) were cultured.
Amplification and sequencing of vlsE cassette region.
The cassette regions of different vlsE alleles were amplified by PCR and sequenced as described previously (33). Primers F4120 (5′-GCGGATCCAGTACGACGGGGAAACCAG-3′) and R4066 (5′-CTTTGCGAACGCAGACTCAGCA-3′) were used for PCR amplification. First-passage cultures of B. burgdorferi clones were used as the sources of template DNA by scraping a small quantity from the surface of the frozen culture, allowing it to thaw, and adding 3 μl (∼105 organisms) per PCR reaction. The PCR products were purified by using Wizard PCR purification columns (Promega, Madison, Wis.) and sequenced on an ABI 377 DNA-Sequencer (Perkin-Elmer/ABI, Foster City, Calif.) at the University of Texas—Houston Microbiology and Molecular Genetics DNA Core Facility. The PCR products were sequenced directly (without cloning the products) to minimize the chances of sequence errors due to PCR infidelity. Sequence analyses were performed with Genetics Computer Group (Madison, Wis.) programs. PILEUP and BOXSHADE programs were used to generate graphic output of the sequence alignments.
Quantitative assessment of vlsE sequence variation.
The nucleotide and predicted amino acid sequences of vlsE alleles from B. burgdorferi clones were aligned with the corresponding sequences of the parental allele vlsE1 by using the PILEUP and BOXSHADE programs. Individual nucleotides or predicted amino acid residues that were different from those of the parental allele vlsE1 were counted as sequence changes and summed for each clone. Identical sequences from multiple clones from the same isolate were considered siblings and counted as a single clone. The sum of the sequence changes for each group of clones at appropriate time points was divided by the total number of clones included to provide a measure of the average number of sequence changes at different time points.
Nucleotide sequence accession numbers.
The sequences of vlsE (allele vlsE1) and the 15 silent vls cassette sequences of B31-5A3 are contained in GenBank entries U76405 and U76406, respectively. The complete coding sequences of the vlsE alleles m1e4A and m1e4C are available under GenBank entries U84554 and U84556. The DNA sequences of vlsE variants presented in this study were deposited in GenBank under the following entries: AF030080 (1396A), AF030081 (1396B), AF030082 (1396D), AF030083 (1396E), AF030084 (1396G), AF034485 (1249C), AF034486 (1250C), AF034487 (1250D), AF034488 (1250E), AF034489 (1251C), AF034490 (1251E), AF034491 (1255C), AF034493 (1360B), AF034494 (1364A), AF034495 (1365D), AF034496 (1373A), AF034497 (1373B), AF034498 (1373D), AF034499 (1373G), AF034500 (1374A), AF034501 (1375A), AF034502 (1377A), AF034503 (1379A), AF034504 (1379D), AF034505 (1379E), AF034506 (1380C), AF034507 (1394B), AF034508 (1394C), AF034509 (1394D), AF034510 (1394E), AF034511 (1394F), AF034512 (1395A), AF034513 (1395B), AF034514 (1395D), AF034515 (1395E), AF034516 (1395F), AF034517 (1410A), AF034518 (1411A), AF034519 (1413A), AF034520 (1414A), AF034521 (1421A), AF034522 (1498A), AF034523 (1500B), AF034524 (1501C), AF034525 (1502C), AF034526 (1502D), AF034527 (1502E), AF034528 (1502F), and AF034529 (1502G).
RESULTS
Kinetic analysis of vlsE variation in C3H/HeN mice.
In our previous study (33), the vlsE cassette region of B. burgdorferi isolated 28 days postinfection of C3H/HeN mice was found to undergo extensive sequence changes consistent with the replacement of the parental vlsE sequence with short segments of sequences from multiple silent vls cassettes. Based on these results, we postulated that sequence changes resulted from multiple independent recombination events occurring during the course of infection between several different silent vls cassettes and the vlsE expression site. To determine if this model was correct, the previously characterized B. burgdorferi clone B31-5A3 (33) was used to infect a group of five C3H/HeN mice, and skin biopsies from each mouse were cultured for organisms on days 2, 4, 7, 14, 21, and 28 postinfection (Fig. 1B). Among the relatively noninvasive methods examined (skin biopsy, blood, and ear punch), skin biopsy yielded the most consistent culture results in preliminary experiments (data not shown). Thus, our analysis of vlsE variation was focused on mouse skin isolates.
Attempts to isolate B. burgdorferi on day 2 postinoculation were not successful, but organisms were cultured from skin biopsies of all five mice on days 4, 7, 14, 21, and 28. Ten clones for each of the isolates were prepared by subsurface plating (Fig. 1B) and designated with a combination of the isolate number (e.g., 1360) and clone designation (A through J). To provide an assessment of vlsE variation in a single mouse during the course of infection, sequential isolates from one mouse (mouse A) were examined in detail. Five clones of mouse A skin biopsy cultures were included at each time point to determine vlsE sequence changes within single biopsy samples. In addition, one skin biopsy clone from each of the four other mice was analyzed for each time point in this study. The vlsE cassette regions of the selected clones were PCR amplified and sequenced with a pair of primers specific for the flanking, conserved regions of vlsE (Fig. 1B) (33).
vlsE sequence variation could be detected at the earliest time point at which positive cultures were obtained (i.e., 4 days postinfection). When compared with the parental allele vlsE1, six of nine 4-day clones examined exhibited sequence variation (Fig. 2), whereas three others had vlsE cassette region sequences identical to that of the parental vlsE allele. Five clones from a single skin isolate (isolate 1360) had sequence changes identical to each other (Fig. 2), possibly due to outgrowth of a single organism from the initial skin culture (33). At 7 days postinfection, 5 of 12 clones tested had occasional sequence changes in variable regions II, III, and V, but 7 others maintained the sequence of the parental allele vlsE1 (data not shown). Collectively, while some organisms in the 4- and 7-day isolates still maintained the parental vlsE1 allele, many had initiated sequence variation.
FIG. 2.
Kinetics of vlsE sequence variation during infection of C3H/HeN mice with B. burgdorferi B31-5A3. PCR products corresponding to the regions shown were amplified from B. burgdorferi clones isolated on days 4, 7, 14, 21, and 28 postinfection in C3H/HeN mice. The predicted amino acid sequences for days 4, 14, and 28 are shown; all nucleotide sequences and predicted amino acid sequences are available through GenBank. The 4-day clones 1360A to 1360G, the 14-day clones 1373A to 1373G, and the 28-day clones 1394B to 1394F were prepared from a single skin biopsy culture from mouse A at each time point. The additional clones shown were isolated from other infected mice in the same experiments. Residues identical to the VlsE cassette region of B31-5A3 (vlsE1) are marked as dashes (−); dissimilar and similar amino acids are shown in uppercase and lowercase, respectively; gaps are indicated by periods. The six variable regions (VR) VR-I through VR-VI are shaded.
Dramatic increases in vlsE sequence variation were observed among the clones isolated on day 14. All of the nine 14-day clones examined showed sequence changes compared with the parental allele vlsE1 (Fig. 2), and all but two of the 14-day clones were unique in terms of vlsE cassette sequence. At 21 days postinfection, all eight B. burgdorferi clones tested exhibited sequence variation in all six variable regions (data not shown). The 28-day clones had extensive sequence changes within all six variable regions (Fig. 2), paralleling the results obtained in the previous study (33).
Consistent with our hypothesis (33), nearly all sequence changes in these vlsE variants could be identified in the 15 silent vls cassettes. For B. burgdorferi clones isolated on days 4 (Fig. 2) and 7 (data not shown), the majority of sequence changes in each of the variable regions apparently resulted from one or two recombination events. Sequences of these segments could be found easily in the corresponding regions of 15 silent vls cassettes. For example, the sequences of the variable regions IV and V in clone 1360B were identical to the corresponding sequences of the silent vls cassettes 16 and 6, respectively. For the clones isolated after 14 days, sequence changes in each of the six variable regions became more complex and mostly appeared as hybrids of multiple silent vls cassettes (Fig. 2). For instance, the vlsE allele 1374A apparently acquired a DNA segment in VR-IV encoding amino acids AASEADQE from vls11, whereas adjacent sequences could be found only in other silent cassettes (Fig. 2).
vlsE variation during chronic infection.
In a previous report, we showed extensive vlsE sequence variation in eight C3H/HeN mice on day 28 postinfection (33). In the present study, the same group of mice were sampled at 7 and 12 months after initial infection. At 7 months, the organisms were isolated from skin biopsies of all six mice examined. Organisms were isolated from all tissue sites tested (bladder, heart, joint, and skin) in all eight mice 12 months after initial infection. These results are consistent with persistent infection by B. burgdorferi (6).
The vlsE cassette regions from a total of eight clones from both 7- and 12-month isolates were amplified and sequenced. All vlsE alleles analyzed had extensive vlsE sequence changes in all six variable regions (Fig. 3); none of these vlsE alleles tested were identical to any of the 15 vlsE alleles analyzed on day 28 postinfection in the previous study (33). However, no sequence variation was detected in the conserved regions of the vlsE cassette in any of the 7- and 12-month clones examined (Fig. 3).
FIG. 3.
Predicted peptide sequence alignment of the parental B31-5A3 vlsE cassette region (vlsE1) and the derivative alleles of progeny clones obtained 7 and 12 months postinfection. The sequence similarity is depicted as described for Fig. 2.
Lack of detectable vlsE variation in vitro.
During the preparation of B. burgdorferi inoculum for mouse infection, the organisms had to undergo 7 days of in vitro culture. In addition, the organisms were usually cultured for ∼2 to 3 more weeks for organism isolation from mouse tissues and for cloning by subsurface plating. Thus, it was possible that vlsE sequence variation detected in C3H isolates during the early days of infection (days 4 and 7) could simply reflect sequence changes generated during in vitro cultivation. To test this possibility, we consecutively subcultured B31-5A3 cultures in BSK II broth every 7 days for up to 12 passages (84 days) in parallel with the mouse infection studies (Fig. 2). Because our previous study did not reveal detectable vlsE sequence changes in the original stock or the organisms undergoing 14-day in vitro cultivation (33), our analysis in this study was focused on the organisms undergoing longer in vitro culture.
After 4 weeks of in vitro cultivation, two of the eight B. burgdorferi clones analyzed lost the vls-carrying plasmid lp28-1, as determined by PCR amplification and Southern hybridization (data not shown). A similar analysis revealed that by passage 12 (84 days), only one of 25 clones tested retained lp28-1. These observations are consistent with the instability of lp28-1 during in vitro cultivation (33), although all of the B. burgdorferi isolates from mice examined contained this plasmid even at 12 months after initial infection (data not shown).
Six clones from the 4-week culture were used as sources of template DNA to amplify and sequence the vlsE cassette region. Consistent with the previous results obtained from the organisms undergoing 14-day in vitro cultivation (33), all six clones of the organisms cultured 28 days had an amplified vlsE sequence identical to that of the parental B31-5A3. The one lp28-1-positive clone from the passage-12 (84-day) culture contained the same vlsE cassette sequence as the passage-1 B31-5A3. In agreement with the previous study (33), these results indicate that vlsE recombination does not occur at a high frequency under standard culture conditions.
Kinetic analysis of vlsE variation in SCID mice.
Immune responses, particularly antibody responses, have been shown to select and thus enrich antigenic variants of African trypanosomes (14) and B. hermsii (2). To examine the possible role of adaptive immune responses in the occurrence of vlsE variation, B31-5A3 was used to infect a group of five CB-17 SCID mice, which are incapable of T- and B-cell development. To minimize the potential loss of the SCID mice during the process of consecutive sampling, only three time points (days 4, 14, and 28) were included in these studies. All five mice survived this process without complications, and positive cultures were obtained from all skin biopsy specimens. Seven to nine clones isolated at each of three time points were used as a source of template DNA for PCR amplification of the vlsE cassette region and subsequent sequence analysis.
Similar to the results from the C3H mouse kinetic study (Fig. 2), sequence analysis revealed the occurrence of vlsE sequence variation as early as 4 days after the initial infection in the SCID mice. Clone 1365D had sequence changes in the variable regions IV, V, and VI (Fig. 4), although eight additional clones retained the parental vlsE cassette sequence. Each of these variant sequences could be found in corresponding positions of silent vls cassettes. Although more changes were found in the 14- and 28-day clones compared with the 4-day clones (Fig. 4), the extent of vlsE sequence variation in SCID mice was lower during this time period than in C3H/HeN mice.
FIG. 4.
Occurrence of vlsE sequence variation in CB-17 SCID mice. The predicted peptide sequence alignment of the parental B31-5A3 allele vlsE1 and derivative vlsE alleles obtained from B. burgdorferi clones isolated on days 4, 14, and 28 postinfection in SCID mice are shown. The 4-day clones 1365A to 1365F, the 14-day clones 1379A to 1379E, and the 28-day clones 1395A to 1395F were each prepared from a single skin biopsy culture from SCID mouse A. The additional clones shown were isolated from other infected mice in the same experiment. The predicted stop codons are marked by asterisks. The sequence similarity is shown as described for Fig. 2.
Interestingly, clones 1379A and 1379C had identical vlsE cassette sequences containing a putative stop codon (TAG) in the region between variable regions III and IV. Since this putative stop codon and adjacent nucleotides encoding glycine (G) and cysteine (C) are present only in silent cassette vls11 (33), this segment of the sequence was apparently acquired from vls11 through recombination. Despite this putative stop codon, clone 1379A can establish infection in C3H/HeN mice. Organisms could be isolated from skin biopsies of each of five C3H/HeN mice 7 days postinfection with 105 cells of clone 1379A. All six 7-day clones examined in a single skin isolate retained the putative stop codon and its adjacent sequences. Immunoblot analysis of clone 1379A did not reveal a detectable protein band reactive with rabbit antiserum against the GST-Vls1 fusion protein (data not shown).
Progression of vlsE variation during infection.
The numbers of vlsE nucleotide and predicted amino acid changes were used as a crude measure of the number and the extent of the recombination events occurring during infection. In this analysis, identical variants from the same culture (as in Fig. 2, day 4) were considered to be siblings because of the extremely low probability that independent recombination events would yield the same sequence. As such, one representative clone from each of these groups of identical variants was included in each comparison.
This quantitative assessment revealed progressive increases in vlsE sequence variation with increased time of infection in both C3H/HeN and CB-17 SCID mice (Table 1). The proportion of variants compared to parental clones increased at a greater rate in C3H/HeN mice than in CB-17 mice, a finding suggestive of selective elimination of nonvariant organisms in immunocompetent mice. As indicated previously, the number of observed nucleotide changes accumulated at a higher rate in C3H/HeN mice than in SCID mice, again suggesting the occurrence of immune selection. In addition, the numbers of predicted amino acid changes correlated well with the changes at the nucleotide level (Table 1). In fact, based on standard genetic codes for prokaryotes, the nucleotide sequence changes in vlsE alleles examined so far have resulted in changes higher than expected in predicted amino acid sequences (data not shown).
TABLE 1.
Progressive vlsE sequence variation during the course of infection of C3H/HeN and CB-17 (SCID) mice with B. burgdorferi B31-5A3a
| Time postinoculation | No. of variants/totalb
|
Avg no. of nucleotide changes/variant (±SE)c
|
Avg no. of predicted amino acid changes/variant (±SE)c
|
|||
|---|---|---|---|---|---|---|
| C3H/HeN | SCID | C3H/HeN | SCID | C3H/HeN | SCID | |
| 4 days | 3/7 | 1/8 | 16.7 ± 3.5 | 10 | 11.0 ± 2.4 | 8 |
| 7 days | 4/11 | 8.8 ± 0.5 | 6.0 ± 1.1 | |||
| 14 days | 9/9 | 4/6 | 34.7 ± 4.7 | 12.0 ± 2.9 | 19.3 ± 2.1 | 6.8 ± 2.3 |
| 21 days | 6/6 | 35.8 ± 2.2 | 21.7 ± 0.8 | |||
| 28 days | 19/19 | 8/8 | 41.5 ± 2.3 | 22.6 ± 4.8 | 24.6 ± 0.9 | 12.4 ± 2.9 |
| 7 mo | 7/7 | 43.1 ± 1.8 | 27.9 ± 1.6 | |||
| 12 mo | 8/8 | 39.6 ± 1.6 | 25.4 ± 1.7 | |||
Combined results from Fig. 2, 3, and 4 and additional experiments. Mice were inoculated with 105 B31-5A3, and B. burgdorferi was cultured from skin biopsies at the indicated time points. Clones isolated by subsurface plating were used as a source of template DNA for PCR, and the products were subjected to sequence analysis. Changes relative to the parental clone B31-5A3 were calculated as described in the text. In this analysis, identical variant clones from the same initial culture were considered siblings and were counted as a single variant.
B. burgdorferi clones containing DNA sequence changes in the vlsE cassette region over the total clones examined. Sibling variants were counted as a single clone.
Average nucleotide or predicted amino acid sequence changes relative to B31-5A3 per unique variant ± standard error (SE).
vlsE variation in multiple tissue sites.
Detection of vlsE variation in vivo but not in vitro indicates that vlsE recombination is induced in the mammalian host. Because the results described above were all based on skin isolates, we wanted to determine whether the induction of vlsE recombination is specific for certain tissues or organs. During the kinetic analysis of vlsE sequence variation in C3H/HeN mice (Fig. 2), we obtained B. burgdorferi cultures of bladder, heart, and joint specimens from the same group of mice 28 days postinfection. To minimize possible variation among different mice, we focused on the isolates of bladder, heart, and joint specimens from mouse A, from which the skin isolates 1360, 1373, and 1394 were also obtained (Fig. 2). Sequence analysis showed that B. burgdorferi clones isolated from bladder, heart, and joint tissues had a similar degree of vlsE sequence variation when compared with the clones isolated from a skin biopsy of the same mouse (Table 1). The average numbers of nucleotide (and amino acid) changes for vlsE alleles from bladder, heart, and joint isolates are 33.3 (24.7), 33 (22.8), and 44 (25.3), respectively, compared with the average of 49.8 (29.0) changes for five clones of the skin isolate (Table 1). These results suggest that the factor(s) responsible for induction of vlsE recombination is present in multiple tissue sites.
vls recombination can occur in other B. burgdorferi B31 variants.
All of the aforementioned results demonstrating vlsE recombination were obtained with B. burgdorferi B31-5A3 (Fig. 2 to 4) (33). We wanted to determine whether this recombination is an allele-specific phenomenon. To test this hypothesis, B. burgdorferi clone M1e4C (105/mouse) was used to infect three C3H/HeN mice by needle inoculation. The organisms were isolated from the infected mice 28 days postinfection, and each isolate was given a serial number. Mouse skin isolate 1396 derived from a M1e4C-infected mouse was selected to generate 10 clonal populations designated A through J. The vlsE cassette regions of five of these clones were amplified by PCR and sequenced.
All five vlsE alleles derived from M1e4C had deduced amino acid sequence changes (Fig. 5). Consistent with the sequence variation among the vlsE1-derived vlsE alleles (33), all varied segments could be found in the corresponding regions in silent vls cassettes. Parallel studies were also conducted with the strain M1e4A, and similar results were obtained (data not shown). Therefore, vls recombination is an ongoing process, in which each vlsE variant is capable of generating progeny with substantial vlsE sequence changes during mammalian infection.
FIG. 5.
Continued variation of vlsE following infection of mice with B. burgdorferi clone M1e4C. The vlsE cassette region predicted amino acid sequence of the parental allele m1e4C was aligned with those of seven derivative alleles obtained from skin biopsies 28 days after infection of C3H/HeN mice. The sequence of VlsE1 from strain B31-5A3 (from which M1e4C was derived) has been included for comparison. The predicted amino acid sequence similarity is shown as described for Fig. 2.
DISCUSSION
We had shown previously that promiscuous recombination at the vlsE site of B31-5A3 could lead to extensive sequence variation in the cassette regions of the surface-exposed lipoprotein VlsE by 28 days postinfection in C3H/HeN mice (33). In the present study, over 50 B. burgdorferi clones isolated at different time points were analyzed to determine the kinetics of vlsE sequence variation. At an early time point (day 4), some clones exhibited occasional sequence changes in the variable regions, while others retained an intact parental vlsE allele (Fig. 1 and 3). With the progression of infection in both C3H/HeN and SCID mice (Table 1), more variable sequences accumulated throughout the vlsE cassette region, i.e., in each of the six variable regions. Thus, vlsE variation began within the first few days after the organisms were introduced into the mammalian host, and additional recombination events accumulated during the course of infection.
Because the number of sequence changes was ∼40 nucleotides for the clones of 28-day, 7-month, and 12-month infections (Table 1), nearly all variable sites in the initial vlsE cassette region appear to be changed by 28 days postinfection in C3H/HeN mice (Fig. 2). This value actually exceeds 34 nucleotides, the average number of differences between the B31-5A3 silent cassettes and the vlsE expression sites. Each clonal variant was unique beyond 14 days of infection, and new variants were observed at 7 and 12 months postinfection. Therefore, recombination appeared to continue at the same rate throughout the course of infection.
Infecting mice with the mouse isolate M1e4C further substantiated that vls recombination is a never-ending process (Fig. 5). The recombination observed in B. burgdorferi clones isolated from M1e4C-infected mice was just as extensive as that obtained with its parental strain B31-5A3 (Fig. 5). Therefore, sequence variation appears to continue as long as the spirochete is in the mouse environment. We believe that similar vlsE sequence variation can occur in other infectious B. burgdorferi strains and possibly other Lyme disease spirochetes, such as B. afzelii and B. garinii. All of the tested infectious strains of B. burgdorferi, B. afzelii, and B. garinii have been shown by hybridization to possess vls sequences (18, 33).
The results presented here indicate that vlsE recombination in B. burgdorferi is induced in the mammalian host as suggested by the previous study (33). vlsE sequence variation could be detected as early as 4 days postinfection in the infected C3H/HeN mice (Fig. 2). In contrast, no vlsE variation was detected in seven clones isolated from B31-5A3 cultured in vitro for 28 to 84 days. Detection of vlsE variation in vivo but not in vitro cannot be explained by adaptive immune selection in C3H/HeN mice because vlsE variation could also be detected in SCID mice (Fig. 4). All eight clones isolated from the CB-17 mice at 28 days postinfection exhibited vlsE variation (Fig. 4; Table 1) despite the lack of immune selective pressure. Furthermore, these clones contained between one and six independent recombination events, as indicated by comparisons of their nucleotide sequences to the silent cassette sequences (data not shown). Another possible argument against the in vivo induction concept is that B. burgdorferi may have a higher growth rate in vivo than in vitro, such that more vlsE variants would be generated in mice in a given period of time. B. burgdorferi has an estimated doubling time of approximately 12 to 16 h in BSK II broth (3, 22). Based on previous results (22), 12 in vitro passages correspond to 100 generations. Although the in vivo growth rates of B. burgdorferi strains have not been determined, the organisms are not present in large quantities in the tissues of the infected animals (6). It is unlikely that the number of generations occurring within 4 days in vivo exceeds that of 12 in vitro passages. Thus, any difference in growth rate alone cannot explain our observations.
In vivo induction of virulence factors is a common theme among pathogens (25, 30, 31). Many factors in host tissues have been well documented to regulate in vivo gene expression (16, 21). Temperature has been demonstrated to modulate transcription of virB in Shigella flexneri (29) and of yop loci in Yersinia species (7). Recently, Schwan et al. (24) have shown that B. burgdorferi expresses the outer surface protein C (OspC) at 24°C but not at 37°C. Expression of other B. burgdorferi proteins, including OspE and OspF, appears to be upregulated in a similar fashion (13, 28). In the present study, we cultured the organisms at 34°C, which is similar to the body temperature of mice (i.e., 37°C). Further studies will be necessary to determine whether temperature is a factor in the induction of vlsE recombination. Elevated CO2 levels induce expression of capsule and toxin genes in Bacillus anthracis (20) and of enterotoxin genes in Vibrio cholerae (27). Our culture conditions included 5% CO2 and 1.5 to 3% O2 with balance N2 (22), under which the CO2 level is similar to that in mammalian tissues. Thus, CO2 is unlikely to be an inducing factor for vlsE recombination.
Another obvious difference between BSK II medium and animal tissues is iron concentration. Most of the iron in mammalian tissues is sequestered within cells or bound to transferrin and lactoferrin. The extracellular free iron in human serum is estimated to be approximately 10−18 μM (32). Low iron levels can induce expression of genes for diphtheria toxin in Corynebacterium diphtheriae (8) and for Shiga-like toxin type I in Escherichia coli (11). The standard BSK II medium (3) is estimated to contain approximately 1.6 μM iron (23). We are investigating the possible role of iron in the induction of vlsE recombination. Other factors, including calcium, magnesium, pH, and osmolarity, have been shown to modulate gene expression in a number of other pathogenic microorganisms (16, 21). The potential role of these factors in in vivo induction of vlsE recombination remains to be investigated.
Direct comparison of vlsE sequence changes in C3H/HeN mice with those in SCID mice revealed differences in the numbers and patterns of sequence changes (Table 1). It is reasonable to believe that the selection of sequence variants occurs in immunocompetent C3H/HeN mice. In this model, clones expressing the initial VlsE “clonotype” or minor variants thereof would activate the immune system during the early stages of infection, resulting in the rapid elimination of these clones. This selection would explain the higher number of changes in the surviving clones in C3H/HeN mice (Fig. 2) compared to SCID mice (Fig. 4).
Identification of a putative stop codon within the coding region of the vlsE allele in clone 1379A (Fig. 4) is of interest. The infectious phenotype and lack of immunoreactive VlsE protein in this clone raise a question about whether VlsE expression is required for infectivity. We have not yet determined the 50% infective dose (22) to see whether infectivity is reduced in this clone. Also, our immunoblotting results do not rule out the possibility that the VlsE protein was expressed in clone B. burgdorferi 1379A. In our previous study, the antiserum against the parental VlsE cassette region could not detect other VlsE variants by immunoblotting analysis (33). Further studies, including the disruption of vlsE by mutagenic means, will be necessary to further define the possible role of the VlsE protein in the pathogenesis of Lyme disease.
ACKNOWLEDGMENTS
We thank Daimin Zhao, Jerrilyn Howell, John Hardham, and Dachun Wang for providing technical assistance and helpful suggestions.
This work was supported by grant AI37277 from the National Institutes of Health.
REFERENCES
- 1.Baranton G, Postic D, Saint Girons I, Boerlin P, Piffaretti J-C, Assous M, Grimont P A D. Delineation of Borrelia burgdorferi sensu stricto, Borrelia garinii sp. nov., and group VS461 associated with Lyme borreliosis. Int J Syst Bacteriol. 1992;42:378–383. doi: 10.1099/00207713-42-3-378. [DOI] [PubMed] [Google Scholar]
- 2.Barbour A G. Antigenic variation of a relapsing fever Borrelia species. Annu Rev Microbiol. 1990;44:155–171. doi: 10.1146/annurev.mi.44.100190.001103. [DOI] [PubMed] [Google Scholar]
- 3.Barbour A G. Isolation and cultivation of Lyme disease spirochetes. Yale J Biol Med. 1984;57:521–525. [PMC free article] [PubMed] [Google Scholar]
- 4.Barbour A G. Linear DNA of Borrelia species and antigenic variation. Trends Microbiol. 1993;1:236–239. doi: 10.1016/0966-842x(93)90139-i. [DOI] [PubMed] [Google Scholar]
- 5.Barbour A G, Fish D. The biological and social phenomenon of Lyme disease. Science. 1993;260:1610–1616. doi: 10.1126/science.8503006. [DOI] [PubMed] [Google Scholar]
- 6.Barthold S W, Persing D H, Armstrong A L, Peeples R A. Kinetics of Borrelia burgdorferi dissemination and evolution of disease after intradermal inoculation of mice. Am J Pathol. 1991;139:263–273. [PMC free article] [PubMed] [Google Scholar]
- 7.Bergman T, Håkansson S, Forsberg Å, Norlander L, Macellaro A, Bäckman A, Bölin I, Wolf-Watz H. Analysis of the V antigen lcrGVH-yopBD operon of Yersinia pseudotuberculosis: evidence for a regulatory role of LcrH and LcrV. J Bacteriol. 1991;173:1607–1616. doi: 10.1128/jb.173.5.1607-1616.1991. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Boyd J, Oza M N, Murphy J R. Molecular cloning and DNA sequence analysis of a diphtheria tox iron-dependent regulatory element (dtxR) from Corynebacterium diphtheriae. Proc Natl Acad Sci USA. 1990;87:5968–5972. doi: 10.1073/pnas.87.15.5968. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Burgdorfer W, Barbour A G, Hayes S F, Benach J L, Grunwaldt E, Davis J P. Lyme disease, a tick-borne spirochetosis? Science. 1982;216:1317–1319. doi: 10.1126/science.7043737. [DOI] [PubMed] [Google Scholar]
- 10.Cadavid D, Pennington P M, Kerentseva T A, Bergström S, Barbour A G. Immunologic and genetic analyses of VmpA of a neurotropic strain of Borrelia turicatae. Infect Immun. 1997;65:3352–3360. doi: 10.1128/iai.65.8.3352-3360.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Calderwood S B, Mekalanos J J. Iron regulation of Shiga-like toxin expression in Escherichia coli is mediated by the fur locus. J Bacteriol. 1987;169:4759–4764. doi: 10.1128/jb.169.10.4759-4764.1987. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Canica M M, Nato F, du Merle L, Mazie J C, Baranton G, Postic D. Monoclonal antibodies for identification of Borrelia afzelii sp. nov. associated with late cutaneous manifestations of Lyme borreliosis. Scand J Infect Dis. 1993;25:441–448. doi: 10.3109/00365549309008525. [DOI] [PubMed] [Google Scholar]
- 13.de Silva A M, Fikrig E. Arthropod- and host-specific gene expression by Borrelia burgdorferi. J Clin Invest. 1997;99:377–379. doi: 10.1172/JCI119169. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Donelson J E. Mechanisms of antigenic variation in Borrelia hermsii and African trypanosomes. J Biol Chem. 1995;270:7783–7786. doi: 10.1074/jbc.270.14.7783. [DOI] [PubMed] [Google Scholar]
- 15.Fraser C M, Casjens S, Huang W M, Sutton G G, Clayton R, Lathigra R, White O, Ketchum K A, Dodson R, Hickey E K, Gwinn M, Dougherty B, Tomb J F, Fleischmann R D, Richardson D, Peterson J, Kerlavage A R, Quackenbush J, Salzberg S, Hanson M, van Vugt R, Palmer N, Adams M D, Gocayne J, Venter J C, et al. Genomic sequence of a Lyme disease spirochaete, Borrelia burgdorferi. Nature. 1997;390:580–586. doi: 10.1038/37551. [DOI] [PubMed] [Google Scholar]
- 16.Guiney D G. Regulation of bacterial virulence gene expression by the host environment. J Clin Invest. 1997;99:565–569. doi: 10.1172/JCI119196. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Hinnebusch B J, Barbour A G, Restrepo B I, Schwan T G. Population structure of the relapsing fever spirochete Borrelia hermsii as indicated by polymorphism of two multigene families that encode immunogenic outer surface lipoproteins. Infect Immun. 1998;66:432–440. doi: 10.1128/iai.66.2.432-440.1998. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Iyer R, Hardham J M, Schwartz I, Norris S J. Abstracts of the 98th General Meeting of the American Society for Microbiology 1998. Washington, D.C: American Society for Microbiology; 1998. Conservation and diversity of vlsE among Borrelia burgdorferi isolates from upstate New York, abstr. D-139; p. 233. [Google Scholar]
- 19.Kawabata H, Myouga F, Inagaki Y, Murai N, Watanabe H. Genetic and immunological analyses of Vls (VMP-like sequences) of Borrelia burgdorferi. Microb Pathog. 1998;24:155–166. doi: 10.1006/mpat.1997.0183. [DOI] [PubMed] [Google Scholar]
- 20.Leppla S H. Production and purification of anthrax toxin. Methods Enzymol. 1988;165:103–116. doi: 10.1016/s0076-6879(88)65019-1. [DOI] [PubMed] [Google Scholar]
- 21.Mekalanos J J. Environmental signals controlling expression of virulence determinants in bacteria. J Bacteriol. 1992;174:1–7. doi: 10.1128/jb.174.1.1-7.1992. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Norris S J, Howell J K, Garza S A, Ferdows M S, Barbour A G. High- and low-infectivity phenotypes of clonal populations of in vitro-cultured Borrelia burgdorferi. Infect Immun. 1995;63:2206–2212. doi: 10.1128/iai.63.6.2206-2212.1995. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Sambri V, Cevenini R, Placa M L. Susceptibility of iron-loaded Borrelia burgdorferi to killing by hydrogen peroxide and human polymorphonuclear leukocytes. FEMS Microbiol Lett. 1991;81:67–72. doi: 10.1016/0378-1097(91)90473-n. [DOI] [PubMed] [Google Scholar]
- 24.Schwan T G, Piesman J, Golde W T, Dolan M C, Rosa P A. Induction of an outer surface protein on Borrelia burgdorferi during tick feeding. Proc Natl Acad Sci USA. 1995;92:2909–2913. doi: 10.1073/pnas.92.7.2909. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Seifert H S, Wright C J, Jerse A E, Cohen M S, Cannon J G. Multiple gonococcal pilin antigenic variants are produced during experimental human infections. J Clin Invest. 1994;93:2744–2749. doi: 10.1172/JCI117290. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Seiler K P, Weis J J. Immunity to Lyme disease: protection, pathology and persistence. Curr Opin Immunol. 1996;8:503–509. doi: 10.1016/s0952-7915(96)80038-0. [DOI] [PubMed] [Google Scholar]
- 27.Shimamura T, Watanabe S, Sasaki S. Enhancement of enterotoxin production by carbon dioxide in Vibrio cholerae. Infect Immun. 1985;49:455–456. doi: 10.1128/iai.49.2.455-456.1985. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Stevenson B, Schwan T G, Rosa P A. Temperature-related differential expression of antigens in the Lyme disease spirochete, Borrelia burgdorferi. Infect Immun. 1995;63:4535–4539. doi: 10.1128/iai.63.11.4535-4539.1995. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Tobe T, Nagai S, Okada N, Asler B, Yoshikawa M, Sasakawa C. Temperature-regulated expression of invasion genes in Shigella flexneri is controlled through the transcriptional activation of the virB gene on the large plasmid. Mol Microbiol. 1991;5:887–893. doi: 10.1111/j.1365-2958.1991.tb00762.x. [DOI] [PubMed] [Google Scholar]
- 30.Turner C M. The rate of antigenic variation in fly-transmitted and syringe-passaged infections of Trypanosoma brucei. FEMS Microbiol Lett. 1997;153:227–231. doi: 10.1111/j.1574-6968.1997.tb10486.x. [DOI] [PubMed] [Google Scholar]
- 31.Waldor M K, Mekalanos J J. Lysogenic conversion by a filamentous phage encoding cholera toxin. Science. 1996;272:1910–1914. doi: 10.1126/science.272.5270.1910. [DOI] [PubMed] [Google Scholar]
- 32.Williams P H, Griffiths E. Bacterial transferrin receptors—structure, function, and contribution to virulence. Med Microbiol Immunol. 1992;181:301–322. doi: 10.1007/BF00191543. [DOI] [PubMed] [Google Scholar]
- 33.Zhang J-R, Hardham J M, Barbour A G, Norris S J. Antigenic variation in Lyme disease borreliae by promiscuous recombination of vmp-like sequence cassettes. Cell. 1997;89:275–285. doi: 10.1016/s0092-8674(00)80206-8. [DOI] [PubMed] [Google Scholar]
- 34.Zhang J-R, Norris S J. Genetic variation of the Borrelia burgdorferi gene vlsE involves cassette-specific, segmental gene conversion. Infect Immun. 1998;66:3698–3704. doi: 10.1128/iai.66.8.3698-3704.1998. [DOI] [PMC free article] [PubMed] [Google Scholar]