Borrelia burgdorferi, the Lyme disease pathogen, employs several immune-evasive strategies to survive in mammals. Unlike mice, major reservoir hosts for B. burgdorferi, rabbits are considered to be nonpermissive hosts for persistent infection. Antigenic variation of the VlsE molecule is a probable evasion strategy known to function in mice. The invariable region 6 (IR6) and carboxyl-terminal domain (Ct) of VlsE elicit dominant antibody responses that are not protective, perhaps to function as decoy epitopes that protect the spirochete. We sought to determine if either of these characteristics of VlsE differed in rabbit infection, contributing to its reputed nonpermissiveness. VlsE recombination was observed in rabbits that were given inoculations with either cultured or host-adapted spirochetes. Early observations showed a lack of anti-C6 (a peptide encompassing the IR6 region) response in most rabbits, so the anti-Ct and anti-C6 responses were monitored for 98 weeks. Anti-C6 antibody appeared as late as 20 weeks postinoculation, and the anti-Ct response, evident within the first 2 weeks, oscillated for prolonged periods of time. These observations, together with the recovery of cultivable spirochetes from tissue of one animal at 98 weeks postinoculation, challenge the notion that the rabbit cannot harbour a long-term B. burgdorferi infection.
Borrelia burgdorferi, the spirochete that causes Lyme disease, is thought to have many strategies at its disposal to avoid the host immune response (reviewed in Embers 2004). These include suppression of innate and acquired immune responses, seclusion into immune-privileged sites, and antigenic variation. In rodents, the principal reservoir of B. burgdorferi in nature, this spirochete is able to persist for months in a manner that is detectable by cultivation of organ tissue (Barthold et al., 1993). While B. burgdorferi spirochetes have been readily isolated from cottontail rabbits in North America (Anderson et al., 1989), European rabbits do not appear to be competent hosts for Borrelia afzelii spirochetes (Matuschka et al., 2000). When B. burgdorferi was experimentally injected into New Zealand White (NZW) rabbits, no organisms could be recovered from cultures of skin, lymph node, joint and spinal cord specimens at 8 weeks postinoculation (Foley et al., 1995). Regardless of whether this atypical infection results from the host's response or the organism's phenotype, these findings imply that one or more of the reputed defence mechanisms of B. burgdorferi may not function in rabbits.
A prominent evasion mechanism in many organisms is antigenic variation. In B. burgdorferi, antigenic variation of an outer membrane lipoprotein, expressed by the variable major protein-like gene (vlsE), is produced by segmental gene conversion within the variable regions of the vlsE central cassette region (Zhang et al., 1997; Zhang & Norris, 1998). Changes in sequence at the vlsE locus throughout the course of infection (Sung et al., 2001) and subsequent antibody responses directed to the variable regions (McDowell et al., 2002) have been demonstrated in mice. While the specific mechanisms that control vlsE recombination have only begun to be delineated (Bykowski et al., 2006), cues provided by the mammalian host environment, particularly immune signals, have been implicated in the onset and rate of sequence changes. vlsE recombination does not occur in vitro (Zhang et al., 1997), in the tick (Indest et al., 2001; Ohnishi et al., 2003; Nosbisch & de Silva, 2007), or in dialysis-bag peritoneal implants (Norris, 2006), an immune-protected host environment. The pro-inflammatory mediator interferon-gamma appears to promote recombination at the vlsE locus and subsequent survival of B. burgdorferi in mice (Anguita et al., 2001). The necessity of vlsE for infectivity is implied by the fact that all infectious strains carry this gene (Iyer et al., 2000; Purser & Norris, 2000). The requirement for recombination, however, has not been empirically determined. In earlier studies, we examined the antibody response to VlsE in various host species. Mice, dogs, monkeys and humans that were infected with B. burgdorferi were almost unfailingly able to respond vigorously to IR6, the invariable region ‘six’ of VlsE (Liang & Philipp, 2000). The central domain of this protein contains six variable and six invariable regions interspersed with each other (Zhang et al., 1997). Compared with the antigenicity of other invariable regions, for example IR2, 3, 4, and 5, that of IR6 is dominant. Infected animals (humans, monkeys, dogs, and mice) either respond to IR6 and not to the other IRs, or respond more vigorously to IR6 (Liang & Philipp, 1999). As with IR6, the carboxyl-terminal domain of VlsE is immunodominant in these animal species (Liang et al., 2001a). The carboxy-terminal and amino-terminal domains of VlsE remain invariant as infection proceeds (Zhang & Norris, 1998). The variable regions of VlsE are indeed immunogenic (McDowell et al., 2002), but the dominant epitopes, interestingly, lie within invariant segments of the molecule.
The immunodominance of invariant segments is uncommon in an antigenic variation molecule. The variable regions are predictably immunodominant in other antigens that undergo variation, as exemplified by the variable surface glycoprotein of African trypanosomes, the pilin of Neisseria gonorrhoeae, and the major surface protein 2 of Anaplasma marginale. In these antigens, the invariable portions are comparatively nonimmunogenic and are thought to contribute essential conformational features to these molecules (Borst, 1991; Abbott et al., 2004).
What could be the role, if any, of the immunodominance of IR6 and the C-terminal domain (Ct) of VlsE? Evidence indicates that the Ct and portions of IR6 are exposed on the surface of the VlsE molecule, but the epitopes are not accessible to antibodies on live spirochetes (Liang et al., 1999, 2001b; Eicken et al., 2002). The immunodominance of these sections affords no risk to the spirochete and thus, we hypothesized, it could serve as a decoy vis-à-vis relevant epitopes of VlsE. Decoy epitopes are defined here as antigenic determinants that elicit dominant nonprotective B- or T-cell responses. When such epitopes have been made to be subdominant by changes in their structure or in the immunization/infection procedure, they are replaced by other, more protective, epitopes of the parent molecule by means of a qualitative shift in antigenic repertoire (Scheerlinck et al., 1993). This pathogen-defence phenomenon has been described for both bacteria and viruses (Garrity et al., 1997; Novotny & Bakaletz, 2003).
An initial assessment of the antibody response to IR6 in rabbits infected with B. burgdorferi yielded no detectable response, as measured by antibody binding to the IR6 peptide (C6) in an enzyme-linked immunosorbent assay (ELISA). Immunoblot assays with whole-cell extracts of B. burgdorferi revealed that these same animals responded well to other antigens of B. burgdorferi. This result (unpublished) contrasted starkly with those obtained by us in other host species. It could be the result simply of an absence of VlsE synthesis — a possibility dispelled by a recent publication (Crother et al., 2004) − or, more interestingly, the result of a selective inability of the rabbit B- or T-helper lymphocyte repertoire to recognize C6 (and Ct) epitopes. Thus the decoy effect of the anti-C6 (or Ct) response would be lost. The absence of vlsE recombination in rabbit, as was observed in spirochetes cultured in medium supplemented with rabbit serum in vitro (Zhang et al., 1997), while unrelated to the observed lack of anti-C6 antibody response, also could contribute to the apparent susceptibility of the spirochete to immune elimination in this host.
To examine these possibilities, rabbits were infected with either in vitro-cultured or host-adapted organisms. Skin biopsies were collected over 8 weeks and cultured; recombination of the vlsE gene was assessed in spirochetes isolated from those biopsies. Serum was collected for 98 weeks. Through the analysis of long-term serum antibody responses to VlsE, C6 and Ct in the rabbit model, the results of this study provide new insight into B. burgdorferi infection in this host.
Materials and methods
Spirochete strain, rabbits, and rabbit infection with B. burgdorferi
The spirochete strain used for all of the experiments was the clonal isolate 5A3 of B. burgdorferi B31. This is an infectious clone that was obtained by subsurface plating of strain B31 and lacks plasmids cp32-3, lp28-2, and lp56, none of which is infectivity-associated (Purser & Norris, 2000; Jacobs et al., 2006). Rabbits were 12–14-week-old New Zealand White (NZW) females. For needle inoculations (animals U039–U042), rabbits were given intradermal injections along the shaved back at six sites with 106 spirochetes per site. On day 11 postinoculation (PI), three 4-mm-diameter pieces of skin within the erythema were resected from each animal. Two of these biopsies from each of the needle-inoculated animals were transplanted onto each of four other rabbits (U043–U046) via small incisions in the recipient's skin and were then sutured into place. The remaining biopsy was mechanically disrupted and injected subcutaneously into CD-1 mice via a 16-gauge syringe in order to verify the presence of infectious organisms. All four of the mice were positive for infection as assessed by culture of ear biopsies that were taken at 4 weeks PI. Skin from rabbit U039 was transplanted onto rabbit U043, that of U040 onto U044, U041 onto U045, and U042 onto U046. All animals were initially bled at weeks 0, 2, 4, 6, 8, and 11 PI, and every 3 weeks thereafter. U040 died of a gastric trichobezoar by week 73, and U041 died during week 93. Periodic blood collection continued for the other rabbits until the 98th week PI, when the remaining animals were euthanized.
Skin biopsies, obtained by a punch of 2 mm diameter within the zone of erythema migrans, were taken every 2 weeks during the first 8 weeks PI. Similar biopsies were taken from all of the rabbits between weeks 41 and 45 PI.
Practices in the housing and care of animals conformed to the regulations and standards established to implement the Animal Welfare Act. Animal care facilities were fully accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care-International.
Peptide and VlsE ELISAs
The ELISAs for determination of C6 and Ct reactivity were performed essentially as previously described (Liang et al., 2001a), except that the C6 peptide was derived from strain B31, sequence MKKDDQIAAAIALRGMAKDGKFAVK. The Ct peptide, which was 52 amino acids long, reproduced the sequence of the C-terminal invariable domain of VlsE, also from B. burgdorferi B31 (Liang et al., 2001a). Rabbit serum samples were added, in duplicate, to peptide-bound wells of the ELISA plates at a dilution of 1 : 400. For VlsE ELISAs, recombinant B31 VlsE, cloned, expressed, and purified as described previously (Bacon et al., 2003), was bound to 96-well plates overnight in 0.1 M Na2HPO4, pH 9.0 at 50 ng well−1. Following a wash in PBS, blocking buffer (5% nonfat dry milk in PBS) was added and allowed to block each plate for 2 h. Rabbit serum was then added to wells, in triplicate, at a 1 : 200 dilution in 5% milk and incubated for 1 h at room temperature in a rotatory shaker set at 150 r.p.m. Specimens from most, but not all, time points of collection were tested for anti-VlsE antibodies. Following four washes in PBS, goat anti-rabbit IgG-horseradish peroxidase secondary antibody (Sigma Chemical Co., St Louis, MO) was added to each well at 0.3 µg mL−1 in 100 µL of blocking buffer. This secondary antibody was used for both peptide and VlsE ELISAs. After 1 h of incubation and four washes in PBS, the bound antibodies were detected by the 3,5′ 5,5′tetramethylbenzidrine (TMB) microwell peroxidase substrate system (Kirkegaard & Perry Laboratories, Gaithersburg, MD) and colour was allowed to develop for 10 min. The enzymatic reaction was stopped by the addition of 100 µL of 1 M H3PO4 per well, and the OD was measured at 450 nm with an ELx-800 microplate reader (Bio-tek Instruments, Winooski, VT). Every ELISA plate contained duplicate samples of preimmune serum from all of the rabbits to arrive at a cut-off value equal to the mean OD value of 8 preimmune specimens plus three times the SD of that mean.
PCR amplification and sequencing of amplified cassette fragments
Isolates from rabbit skin were grown in BSK-H medium (Sigma). When grown to late logarithmic phase, the cultures were diluted in BSK-H medium and subsurface-plated in BSKII agarose as described previously (Norris et al., 1995). Ten separate clones from each isolate were selected and inoculated into BSKII medium, grown to late logarithmic phase, and frozen at −70°C in BSKII medium with 10% glycerol. PCR amplification for detection of lp28-1 was performed using 2 µL of each frozen stock culture with primers F4120 (5′-GCGGATCCAGTACGACGGGGAAACCAG-3′) and R4066 (5′-CTTTGCGAACTGCAGACTCAGCA-3′). These primers are in the invariable regions flanking the vlsE cassette region and amplify a 660-bp product. Amplification was carried out for 35 cycles with initial denaturation at 94°C for 2 min, then denaturation at 94°C for 45 s, annealing at 52°C for 45 s, and extension at 72°C for 1 min using Dynazyme II DNA polymerase (Finnzyme, New England Biolabs, Ipswich, MA). PCR products were analysed by electrophoresis on 1% agarose gels, and visualized with ethidium bromide. PCR amplification for detection of the lp25 plasmid was performed as described for plasmid lp28-1, with the exception that the primer pair F4511 (5′-AGAATTATGTCGGTGGCGTTGT-3′) and R4484 (5′-ATTAAAGCCGC CTTTTCCTTGGT-3′) was used; the amplification product was 258 bp long.
The PCR products from clones positive for lp28-1 were sequenced to detect sequence variation in the vlsE expression cassette using primers F4120 and R4066. The PCR products were prepared for sequencing using a MoBio PCR clean-up kit (MO BIO Laboratories, Inc., Carlsbad, CA). Sequencing was performed at the University of Texas Microbiology Core facility using an ABI 377 automatic DNA sequencer. The vlsE cassette region sequences were deposited at the US National Center for Biotechnology Information with accession numbers EF044276–EF044300.
Evidence of infection in animals that were inoculated with cultured or host-adapted spirochetes
An inoculum of cultured B. burgdorferi was delivered using needle and syringe to each of four rabbits. Eleven days after the initial inoculation, skin biopsy fragments were obtained from the needle-inoculated animals and transplanted onto four additional rabbits so as to provide infection with host-adapted spirochetes. Infection was assessed by in vitro culture of skin biopsies, and the antibody responses to C6, Ct and VlsE were monitored initially for an 11-week period, and then for a total of 98 weeks PI. Animals were then euthanized, and tissue fragments that were obtained from several organs were cultured to assess the residual presence of viable spirochetes. Recombination of the vlsE gene was examined in spirochetal clones obtained by subsurface plating of isolates recovered from skin biopsy cultures at 2–6 weeks PI.
Evidence of infection was sought in all of the animals by cultivation of skin biopsy samples. Regardless of the source of infecting spirochetes, i.e. cultured or host-adapted, all of the animals showed evidence of infection by either week 2 or week 4 PI (Table 1). By week 8, however, skin biopsies from the same group of animals were consistently culture-negative. Skin biopsies taken from all of the rabbits between weeks 41 and 46 PI also yielded no detectable spirochetes. Rabbits U040 and U041 died before necropsy. Following euthanasia, the other six animals were subjected to necropsy at week 98 and the following specimens were collected: blood, heart, bladder, spleen, and tendons of thoracic and pelvic limb joints. Tissues were cut into small c. 2 mm2 pieces, placed into BSK-H medium, cultured at 34°C, and examined over 8–9 weeks. At the time of necropsy, the left radiohumeral tendon of rabbit U045 was culture-positive, whereas all remaining cultures were negative for the presence of spirochetes.
|Rabbit||Inoculated spirochetes||Week 2||Week 4||Week 6||Week 8|
|Rabbit||Inoculated spirochetes||Week 2||Week 4||Week 6||Week 8|
Cultured: rabbits inoculated by needle and syringe with 106Borrelia burgdorferi B31-5A3 at each of six intradermal inoculation sites. Host-adapted: rabbits implanted with skin tissue from rabbits inoculated 11 days earlier with cultured organisms, as described in the text. Skin biopsies were incubated in BSK-H medium at 34°C for 7–20 days before determination of active infection by the presence (indicated by a +) of live spirochetes in cultures.
Recombination of the vlsE gene by spirochetes reisolated from animals that had been inoculated with either cultured or host-adapted organisms was examined by comparison of the sequence of their variable domain with that of the original clonal isolate used for the inoculations. Spirochetal clones were generated from skin biopsy isolates obtained at 2 weeks PI from needle-inoculated animal U041 and at 2 and 6 weeks PI from needle-inoculated animal U042 (Table 2). Rabbit U045 was used as a source of spirochetes derived from a host-adapted inoculum. Clones were generated from isolates obtained at 2 and 4 weeks PI. For all of the isolates, including the clone originally used to needle-inoculate the rabbits, 10 clones (or subclones) were generated (Table 2).
|Source of clones||Type of inoculum||Weeks postinfection||Clones||lp25||lp28-1||Mean number of VlsE sequence changes|
|Source of clones||Type of inoculum||Weeks postinfection||Clones||lp25||lp28-1||Mean number of VlsE sequence changes|
Rabbits (Rbt) were inoculated with either cultured or host-adapted Borrelia burgdorferi, as described in Table 1.
Changes, deletions, and insertions within the VlsE variable-domain deduced amino acid sequence, relative to the sequence in the parental clone, B. burgdorferi B31-5A3.
No sequence variation was observed in 10 clones derived from the 5A3 clone used as the inoculum, consistent with the lack of detectable vlsE recombination during in vitro culture of B. burgdorferi (Table 2). vlsE recombination was readily apparent at 2 weeks PI in animals inoculated with cultured organisms. The average number of changes in the sequence of the 6 variable regions was 16 in rabbit U041 and 22 in rabbit U042 (Table 2). This number increased to 35 by 6 weeks PI. In the animal inoculated with host-adapted spirochetes (U045), who received spirochetes that had been in the skin of rabbit U041 for 10 days, the average number of sequence changes was 30 at 2 weeks PI, and 34 at 4 weeks PI.
Most of the clonal isolates contained lp25, as would be expected, as this plasmid was shown to be essential for infectivity in mice (Purser & Norris, 2000). Surprisingly, lp28-1, the plasmid that encodes vlsE, was absent in about one-half of the clones recovered from rabbit skin biopsies. These plasmids could have been lost during the time the clones were amplified in culture (Schwan et al., 1988), but a similar analysis of B. burgdorferi clones isolated from mouse tissues in the same manner yielded much higher lp28-1 retention values (Labandeira-Rey & Skare, 2001; McDowell et al., 2001 and Table 2). A statistical comparison of lp28-1 retention frequency between clones isolated from mice (Labandeira-Rey & Skare, 2001) and from rabbits in this study, 19/20 and 25/50, respectively, resulted in a significant difference (P=0.0004, test of difference between two proportions, one-sided).
Antibody response to VlsE, C6 and Ct during the first 11 weeks of infection
The IgG antibody responses to C6 and Ct were monitored in all animals initially during the first 11 weeks following infection. All of the animals responded vigorously to Ct. As early as 2 weeks PI, the response to Ct, as measured by the ELISA OD value, was 20 (U041) to 43 (U044) times the cut-off OD value. In contrast, the response to C6 was no higher than between 1.8 (U043) and 12.5 (U044) times the cut-off value (Fig. 1) in the first 11 weeks following infection. Levels of serum antibodies that recognized the recombinant VlsE molecule were also quantified and found essentially to mirror the C6 response: except for one animal (U046), anti-VlsE antibody levels remained either undetectable or low early in infection (<11 weeks PI). These results confirmed our preliminary observation of a poor, and most frequently absent antibody response to IR6 in rabbits, but dispelled the possibility that this might be compounded by a poor response to other immunodominant epitopes, as the marked response to Ct clearly indicated.
Antibody response to VlsE, C6 and Ct beyond 11 weeks of infection
The response to Ct appeared to decline in most animals by week 11 PI, but, surprisingly, in rabbit U044 the C6 response increased (Fig. 1). Thus, we examined the VlsE, C6 and Ct antibody responses in the weeks following this time-point, until rabbits were euthanized at 98 weeks PI. The response to VlsE antigens was dissimilar among individual hosts, but all rabbits generated an antibody response to Ct early in infection (Fig. 1). A continual decline in anti-Ct antibodies over time was observed only in one rabbit (U045). In other animals, most notably U040 and U043, the Ct response declined initially, but rose after 40 weeks of infection. A periodic rise and fall in Ct response over time was common, and did not reflect single-point spikes, but genuine trends in the responses. No correlation between the antibody responses and method of inoculation (host-adapted vs. cultured organisms) was apparent.
As mentioned, a response to the C6 region of VlsE was present, albeit at a very low level, in early infection for most rabbits. A rise in anti-C6 antibodies as infection proceeded, which occurs characteristically earlier in other host species, was observed in three of eight rabbits (U040, U044 and U045, Fig. 1); one of these animals (U045) did not generate antibodies to C6 until very late (22 weeks) into infection. Two additional observations with respect to C6 and reactivity to the recombinant VlsE molecule are of interest: (1) only those animals with an early (before week 20) C6 response also had appreciable anti-VlsE IgG antibody responses; and (2) the trend, in terms of inception and duration, of VlsE reactivity paralleled that of C6 more closely than that of Ct.
Borrelia burgdorferi infectivity has long been known to depend on both spirochetal isolate and host species, but the underlying mechanisms have only recently begun to be evaluated. The initial driving force for the realization of the experiments presented in this paper was, on the one hand, the standing paradigm that the NZW rabbit was a nonpermissive host for persistent infection by B. burgdorferi, and on the other, our preliminary observation that the antibody response to IR6 was essentially silent in this host. This observation was in contrast to what we had previously observed in mice, dogs, monkeys and humans, animals in which B. burgdorferi infection evoked strong and early anti-C6 antibody responses. Notably, these are also host species regarded as permissive for B. burgdorferi long-term infection. With the decoy hypothesis in mind, and the contention that a plausible explanation for the rabbit nonpermissiveness for persistent infection was a selective inability of the rabbit B- or T-helper lymphocyte repertoire to recognize C6 (and Ct) epitopes, we set out to ascertain the antibody response to VlsE using as antigens not only IR6 (C6) but also Ct and VlsE. The results we obtained not only exclude the possible role of C6 in providing decoy epitopes for VlsE in this host, but also may dictate the need for a shift in the paradigm that regards the rabbit as a host that does not permit persistent infection with B. burgdorferi.
All animals showed a marked early antibody response to Ct. As observed by us previously, the anti-C6 response was either very low or undetectable during the first 6–11 weeks, but after this time its level became very high in three of the animals. We contend, but cannot prove except for the result obtained with animal U045, which was culture-positive at 98 weeks PI, that the rabbit can be a permissive host for persistent, but probably low-level, B. burgdorferi infection. Our contention is supported by two observations. First, animal U045 experienced a late and dramatic increase in C6 antibody, up from a level that remained negligible for 26 weeks PI. Second, the levels of anti-Ct antibodies oscillated widely in several animals over a prolonged period of time.
The long-term production of specific antibodies could result from low-grade chronic infection, from repeated exposure to antigen or to antigen–antibody complexes, or from cross-reactivity to self or environmental antigens (Slifka & Ahmed, 1998). In a low-level infection, such persistent antibody production may be propelled by stimulation of memory B cells with antigen–antibody complexes on the surfaces of follicular dendritic cells (Wu et al., 1996). Long-lived plasma cells may also provide a mechanism for maintaining persistent antibody production (Slifka & Ahmed, 1998). However, in the absence of active infection, antigen would eventually be consumed and should result in a slow decline of specific antibodies, unlike the periodic rise and fall of specific antibody levels that we observed in several of the animals. Likewise, cross-reactivity to other antigens would either plateau or dissipate. The variable long-term antibody responses to VlsE epitopes must be a result of periodic alteration and/or reintroduction of antigen.
A series of increases and decreases in anti-VlsE response, specifically Ct, was observed to some degree in seven of eight rabbits. Such a trend would be expected in response to an antigenic variation molecule, such that alterations in the antigen bring about an antibody response that wanes when the exact composition of that antigen ceases to exist and it is replaced by a new variant. However, Ct antigenic regions are invariant during infection; thus, such changes in antibody response cannot be explained by changes in the Ct segment itself. If C6 or Ct epitope recognition by either B cells or helper T cells is dependent upon amino acids within the variable region — as with conformational, discontinuous epitopes that encompass invariant and variable segments in the case of B cells, and adjacent variable segments in the case of helper T cells — then slight sequence changes may allow for either the activation or silencing of B cells when new variants arise.
We thus propose a scenario in which B. burgdorferi engages in a low-grade chronic infection (Barthold et al., 1993) with periodic antigenic restimulation owing to small alterations in the vlsE coding sequence and/or replenishment of immune complexes to follicular dendritic cells by changes in the quantity of antigen as a function of bacterial load.
The early rise of anti-C6 antibody responses in hosts such as mice and humans and the relatively later onset of this response in rabbits may have resulted from the manner in which these animals generate antibody diversity. In contrast to humans and mice, rabbits use only a few variable-region gene segments in the production of immunoglobulin receptors on immature B cells. B-cell receptor diversity is generated by gene conversion with variable-region pseudogenes in the gut-associated lymphoid tissue (GALT) at 4–8 weeks of age. In both cases, further diversity is generated by somatic mutation, but for the rabbit this occurs in GALT following presentation of intestinal antigens, a process that may be slowed in germ-free laboratory environments (Lanning et al., 2000). The late onset of a C6 response may have resulted from these differences in B-cell development that produce the rabbits' antibody receptor repertoire.
We also had hypothesized that the putative refractory nature of rabbit infection was the result of the limited or absent recombination of the vlsE gene of B. burgdorferi in this host environment. This notion was disproved by our results, as ostensible sequence changes in the vlsE gene were present in clonal isolates recovered from skin biopsy cultures at 2 weeks PI. The average number of sequence changes (30 changes) was comparable to, and even higher than, that observed in mice at 4 weeks PI (Zhang et al., 1997). Unlike the case for mice, however, and perhaps significantly, lp28-1, the plasmid that encodes vlsE, was absent in 25 of the 50 clones recovered from rabbit skin biopsies. Although these plasmids could have been lost during the time the clones were amplified in culture (Schwan et al., 1988), similar analyses of B. burgdorferi clones isolated from mouse tissues have yielded much higher lp28-1 retention values than those observed by us in rabbits (Labandeira-Rey & Skare, 2001; McDowell et al., 2001). While we have no insight into the mechanism whereby loss of lp28-1 may occur more frequently in rabbits than in mice, the difference may explain the contrast in permissiveness of persistent B. burgdorferi infection in these hosts. No lp28-1-negative isolate has been recovered from cultures of skin biopsies or blood specimens from humans infected with B. burgdorferi (Iyer et al., 2003), and it is likely that, as was shown in mice (Purser & Norris, 2000; Labandeira-Rey & Skare, 2001), lp28-1-negative organisms are also not viable in rabbits. When lp28-1-deficient spirochetes are inoculated into mice, the organisms survive for <4 weeks (Labandeira-Rey et al., 2003). In the present experiment, it appeared as though such spirochetes could be recovered from rabbit skin biopsy cultures for up to 6 weeks PI (Table 1). However, the rabbits were not infected with lp28-1-negative spirochetes. All of the 10 clones isolated from the B. burgdorferi needle inoculum harboured lp28-1 (Table 2), and thus loss of lp28-1 plasmid must have occurred at different times PI. Therefore, recovery of spirochetal clones lacking this plasmid as late as 6 weeks PI does not imply that such organisms survived in rabbits for that length of time. No spirochetes were cultivable from the skin by week 8 PI (Table 1).
The recovery and culture of organisms from rabbit tissue nearly 2 years after infection, although arguably a rare event, proves that B. burgdorferi can persist in the rabbit. The periodic rise and fall of specific antibody levels that we observed in several of the animals is consistent with this tenet. We thus speculate that the rabbit is a permissive host in which the infectivity-associated plasmid lp28-1 may be lost more readily than in mice. This plasmid loss could account for a decline in spirochetal burden over time.
B.J.B.J. is a US Federal Government employee. No claim to original US Government works.
This work was supported in part by NIH grants RR00164-41 (M.T.P.) and R01 AI37277 (S.J.N.). The authors would like to thank Maury Duplantis for rabbit necropsies.