The role of VlsE antigenic variation in the Lyme disease spirochete: persistence through a mechanism that differs from other pathogens


*E-mail; Tel. (+1) 403 210 9692; Fax (+1) 403 270 2772.


The linear plasmid, lp28-1, is required for persistent infection by the Lyme disease spirochete, Borrelia burgdorferi. This plasmid contains the vls antigenic variation locus, which has long been thought to be important for immune evasion. However, the role of the vls locus as a virulence factor during mammalian infection has not been clearly defined. We report the successful removal of the vls locus through telomere resolvase-mediated targeted deletion, and demonstrate the absolute requirement of this lp28-1 component for persistence in the mouse host. Moreover, successful infection of C3H/HeN mice with an lp28-1 plasmid in which the left portion was deleted excludes participation of other lp28-1 non-vls genes in spirochete virulence, persistence and the process of recombinational switching at vlsE. Data are also presented that cast doubt on an immune evasion mechanism whereby VlsE directly masks other surface antigens similar to what has been observed for several other pathogens that undergo recombinational antigenic variation.


Borrelia burgdorferi is the causative agent of the multisystemic disorder Lyme disease, which can last from months to years due to the ability of this spirochete to successfully evade host immune defences (Barbour, 2001; Steere et al., 2004). It has long been presumed that a key mechanism to immune evasion by B. burgdorferi is recombination at the vls (Vmp-like sequence) locus located on the 28 kb linear plasmid, lp28-1 (Zhang et al., 1997; Zhang and Norris, 1998a; Norris, 2006). The vls locus is composed of an expression site (vlsE) encoding the 35 kDa lipoprotein VlsE, and a contiguous array of 15 silent (unexpressed) cassettes. These cassettes show strong homology to the central region of VlsE, and contain six variable regions flanked by highly conserved sequences (Zhang et al., 1997). Cassette segments act as a source of non-reciprocal recombination events with the expression locus in order to produce large numbers of distinct antigenic variants during infection (Zhang and Norris, 1998b). By changing the surface epitopes of VlsE in this way, the spirochete is thought to outmaneuver the host antibody response resulting in persistent infection. Studies have shown the production of antibodies specific for the variable regions of VlsE during experimental infection of mice (McDowell et al., 2002), and the three-dimensional structure of VlsE supports the idea that these variable regions are easily accessible to antibodies while invariable regions are obscured (Eicken et al., 2002).

A number of early studies have demonstrated the importance of the vls-resident plasmid, lp28-1, for infectivity and persistence by B. burgdorferi (Purser and Norris, 2000; Labandeira-Rey and Skare, 2001). Clones in which lp28-1 is missing exhibit an intermediate infectivity phenotype where spirochetes are able to disseminate to various tissue sites, but display reduced virulence due to an inability to persist in a mouse host. However, it has been shown that lp28-1 minus clones are capable of long-term survival in severe combined immunodeficient mice which lack an effective antibody response (Labandeira-Rey et al., 2003; Purser et al., 2003). Moreover, lp28-1 mutants grow normally in a dialysis membrane chamber implanted in the peritoneal cavity of rats where exposure to either antibodies or immune cells is restricted (Purser et al., 2003). These findings indicate that lp28-1 is required for persistence only when in the presence of an effective humoral immune response. Nevertheless, generating genetic knockouts of the vls locus has been unattainable in the past, and thus definitive evidence for the role of vls antigenic variation in virulence has been elusive.

In addition to VlsE, there is evidence to indicate that a number of other surface proteins, including the decorin binding protein A (DbpA) (Hagman et al., 2000), the fibronectin binding protein BBK32 (Fikrig et al., 2000) and the outer surface protein C (OspC) (Wilske et al., 1993), are also immunogenic. This differs from other antigenic variation systems, such as with the variable surface glycoprotein of Trypanosoma brucei (Taylor and Rudenko, 2006) which is the only antigenic structure exposed on the surface, or with the variable major protein in Borrelia hermsii that dominates the antibody response to this pathogen (Barbour, 2003). How B. burgdorferi survives the robust antibody response directed to antigens other than VlsE has been enigmatic.

Many pathogens utilize one or more of a number of evasion strategies in addition to antigenic variation, and several models have been suggested for how VlsE might become the primary target for the host immune response. One idea is similar to the protozoan pathogens Plasmodium falciparum and T. brucei, where the variable surface protein essentially coats the surface of the organism (Dzikowski et al., 2006; Taylor and Rudenko, 2006). In the case of B. burgdorferi, where other surface antigens exist, VlsE may act as a shield to obscure other epitopes. Indeed, coexpression studies of P66 and a number of Osp proteins have shown that colocalization of proteins on the membrane surface can serve to hide epitopes from antibody recognition (Bunikis and Barbour, 1999). Another proposal is that VlsE might be a T-cell-independent antigen that may directly stimulate B cells to produce antibodies, thereby directing the humoral response to VlsE (Philipp et al., 2001). Finally, it has been suggested that other surface antigens might be downregulated in the mammalian host, leaving VlsE as the dominant presence on the bacterial surface (Liang et al., 2002).

We report here the use of telomere resolvase-mediated targeted deletions in lp28-1 to remove either the entire vls locus on the right side of lp28-1 or loci BBF01BBF19 on the left side. Subsequent studies revealed an essential role for vlsE in persistent mouse infection. They also demonstrated the absence of any factors on the left side of lp28-1 involved in virulence, persistence or recombinational switching in the mouse, and call into question a mechanism for immune evasion by which other surface antigens are cloaked by VlsE.


Construction of a vls knockout through targeted deletion in lp28-1

Previous studies have clearly demonstrated the importance of lp28-1 for persistence associated with Lyme disease (Purser and Norris, 2000; Labandeira-Rey and Skare, 2001; Labandeira-Rey et al., 2003; Lawrenz et al., 2004). However, it has remained unclear whether genes other than vlsE and the silent vls cassettes on lp28-1 are responsible for the inability to persist, as generating vls knockouts has thus far proven difficult. In order to assess whether antigenic switching at vlsE is required for full infectivity and persistence in vivo, we employed a method by which only the vls locus from the lp28-1 plasmid in B. burgdorferi would be deleted. This method is based on previous work in our laboratory that utilizes targeted deletion through integration of a plasmid containing a replicated telomere (Chaconas et al., 2001; Beaurepaire and Chaconas, 2005). Insertion of a replicated telomere in a linear plasmid leads to subsequent deletion of DNA by the telomere resolvase, ResT, an enzyme that processes replicated telomeres to generate covalently closed hairpin ends at the termini of the linear replicons in Borrelia species (Kobryn and Chaconas, 2002; Chaconas, 2005).

To obtain a vls knockout in B. burgdorferi, the plasmid pTB44 was constructed (see Experimental procedures). This plasmid, which cannot replicate in B. burgdorferi, carries a kanamycin resistance gene, a replicated telomere from the left end of lp17, and a stretch of target DNA homologous to lp28-1 (see Fig. 1A). The infectious B. burgdorferi clone B31-A3 (Elias et al., 2002) was transformed with this construct, followed by selective growth in kanamycin. As shown in Fig. 1A, insertion of the plasmid construct into lp28-1 was expected to occur at a target site just upstream of the vls locus, resulting in loss of both the silent cassettes and vlsE after telomere resolution. Twelve transformants were recovered and screened by polymerase chain reaction (PCR) for the presence of the kan gene and the absence of vlsE (see Table 1 for primers used in this work). Seven clones matched these criteria and were further analysed by field inversion gel electrophoresis of mini-genomic DNA preps. All seven contained a deletion of lp28-1 of the expected size, and one such clone is shown in Fig. 1B. The identity of the truncated constructs was verified by excision from a field inversion gel, followed by PCR reactions with primers that give unique products based upon the ResT-mediated deletions. Deletion of the vls locus at the appropriate location was confirmed by sequencing across the deletion junction (see Experimental procedures), and two isolates (A3Δvls-clone 2 and A3Δvls-clone 9) were arbitrarily selected for further study. Plasmid content was analysed in these clones by PCR screening to ensure that no plasmids had been lost during the strain construction. Both clones showed identical plasmid profiles compared to the parental B31-A3 with only the non-essential cp9 missing. A list of the B. burgdorferi clones used in this study is presented in Table 2.

Figure 1.

Deletion of vlsE and silent vls cassettes by insertion of a replicated telomere in lp28-1.
A. Schematic of the construction strategy for the vls knockout plasmid. The target sequence for insertion (green; co-ordinates 17 296–18 800 of the complete lp28-1 sequence – see Experimental procedures) was chosen so that only the silent cassettes and vlsE would be deleted from lp28-1. The genes encoding paralogous family proteins 49, 32, 50 and 57 (BBF23, 24, 25 and 26 arranged from left to right) that have been previously shown to allow autonomous replication in an lp28-1-derived shuttle vector (Stewart et al., 2003) are shown as black lined arrows. Hairpin telomeres are shown as red hatched regions.
B. Analysis of the completed construct by field inversion gel electrophoresis. Genomic DNA from B31-A3 (lane 2) and a B31-A3Δvls clone (lane 3) are shown. Field inversion gel electrophoresis was carried out as described in Experimental procedures. A molecular weight marker (M) is shown in lane 1. The positions of the lp28 plasmids 1–4 as well as vls knockout plasmid are shown on the right.

Table 1.  Oligonucleotide primers (5′−3′) used in this study.
B248GCGATATAAGTAGTACGACGGGGAAACCAGPCR screen for vlsE gene and antigenic switching
B348CGCAGCAGCAACGATGTTACPCR screen for gent gene
B513CCGGGGTACCGCTGTATAATGTCAAATGGCTAGGAmplify target site for vls deletion
B846CTGCACTACCACAAGAGATTGCAPCR screen for left-end sequence of lp28-1
B848CATTTCTAGTCTAGATTGCAGTTATTTCTAAAATTAACTAmplify target site for lp28-1 left-end deletion
B945GTTTAA ATCATCATAT TTGTTGTGATCTAmplify junction at vls deletion
B947ACTTTTTTTGTTTCATTTGATTCCAATCAmplify junction at lp28-1 left-end deletion
B949CTCGCGAAAGCTCCAATACGCASequencing of junction at vls deletion
B950TGGCAGAGCATTACGCTGACTSequencing of junction at lp28-1 left-end deletion
Table 2.  Bacterial strains used in this study.
B. burgdorferi B31 cloneMissing plasmidsvls locusvlsEReference
A3cp9++Elias et al. (2002)
A1cp9, lp28-1Elias et al. (2002)
A3Δvls-clone 2cp9This study
A3Δvls-clone 9cp9This study
A3Δvls + pBSV2::vlsEacp9+This study
A3 lp28-1Δleft end-clone 2cp9++This study
A3 lp28-1Δleft end-clone 4cp9++This study

Deletion of vls leads to a loss of persistence in vivo by B. burgdorferi

To determine whether loss of the vls locus alone is responsible for the lack of infectivity and persistence previously observed with lp28-1 minus clones of B. burgdorferi B31, groups of six C3H/HeN-immunocompetent mice were inoculated with either B31-A3 (wild-type parental strain), B31-A1 (lp28-1 minus), B31-A3Δvls-clone 2, B31-A3Δvls-clone 9, or B31-A3Δvls + pBSV2::vlsE (a complemented vls knockout clone), and sample sites were tested for the presence of spirochetes at various times during a 4 week period. It is important to note that the VlsE-supplemented clone (B31-A3Δvls + pBSV2::vlsE) does not represent true complementation, as this strain lacks the silent cassettes involved in antigenic switching at the vlsE locus and, therefore, cannot persist in the mouse host (Lawrenz et al., 2004).

As shown in Table 3, blood samples taken from all mice at days 4 and 7 produced positive cultures for spirochetes. Spirochetes from week 1 blood cultures were grown in the presence or absence of appropriate antibiotic to assess the presence of the lp28-1Δvls construct and the pBSV2::vlsE complementing plasmid. The resulting growth curves were indistinguishable regardless of whether or not antibiotic was present (data not shown). The fact that comparable growth patterns were observed indicates that the vast majority of spirochetes cultured at day 7 post infection still retained their respective plasmids, after both in vitro propagation to generate the inoculum, and proliferation during the first week in the mouse.

Table 3.  Effect of vls deletion on B. burgdorferi infection in C3H/HeN mice.
 B. burgdorferi B31 clonevlsBloodaEarHeartBladderJointTotal sitesbTotal micec
  • a. 

    Values listed correspond to number of cultures positive/number tested at either blood, ear, heart, bladder or joint.

  • b. 

    Number of positive tissue sites/number tested.

  • c. 

    Number of mice giving positive cultures/number tested.

  • At day 14, three mice in each group were sacrificed and three continued onto day 28.

Day 4A3 (wild type)+6/6      
A1 (lp28-1 minus)6/6      
A3/(lp28-1Δvls-clone 2)6/6      
A3/(lp28-1Δvls-clone 9)6/6      
A3/(lp28-1Δvls + vlsE)6/6      
Day 7A3 (wild type)+6/6      
A1 (lp28-1 minus)6/6      
A3/(lp28-1Δvls-clone 2)6/6      
A3/(lp28-1Δvls-clone 9)6/6      
A3/(lp28-1Δvls + vlsE)6/6      
Day 12A3 (wild type)+ 6/6     
A1 (lp28-1 minus) 3/6     
A3/(lp28-1Δvls-clone 2) 3/6     
A3/(lp28-1Δvls-clone 9) 4/6     
A3/(lp28-1Δvls + vlsE) 2/6     
Day 14A3 (wild type)+ 6/63/33/33/315/156/6
A1 (lp28-1 minus) 2/61/32/32/37/156/6
A3/(lp28-1Δvls-clone 2) 2/61/32/32/37/156/6
A3/(lp28-1Δvls-clone 9) 3/62/32/31/38/156/6
A3/(lp28-1Δvls + vlsE) 0/61/31/33/35/155/6
Day 21A3 (wild type)+ 3/3     
A1 (lp28-1 minus) 0/3     
A3/(lp28-1Δvls-clone 2) 0/3     
A3/(lp28-1Δvls-clone 9) 0/3     
A3/(lp28-1Δvls + vlsE) 0/3     
Day 28A3 (wild type)+ 3/33/33/33/312/123/3
A1 (lp28-1 minus) 0/30/30/30/30/120/3
A3/(lp28-1Δvls-clone 2) 0/30/30/30/30/120/3
A3/(lp28-1Δvls-clone 9) 0/30/30/30/30/120/3
A3/(lp28-1Δvls + vlsE) 0/30/30/30/30/120/3

While all six mice infected with the positive control clone B31-A3 gave productive cultures for spirochetes from ear biopsies taken at day 12, only partial recovery was obtained from either the lp28-1 minus-, lp28-1Δvls-, or lp28-1Δvls + vlsE-infected mice (Table 3). At day 14, three mice from each group were euthanized, and ear, heart, bladder and joint tissues were harvested and cultured for the presence of spirochetes. As expected, the wild-type B31-A3 clone was recovered from all three mice at 15 of 15 sites tested (Table 3). Cultures of tissues from the lp28-1 minus clone as well as A3Δvls- and A3Δvls + vlsE-infected mice resulted in spirochete recovery from between only 5 and 8 of the total of 15 sites cultured at 14 days. The mutant spirochetes were, therefore, either compromised at their efficiency of infection relative to wild type, or alternatively, in the process of being cleared by day 14.

By 3 weeks post inoculation, ear biopsies from the remaining three B31-A3-infected mice (wild type) produced positive cultures for spirochetes, while neither the lp28-1 minus nor the lp28-1Δvls clones were recovered from this site (Table 3). Finally, complete clearance from all locations of all spirochetes lacking the vls locus was observed by week 4. All of these mice had clearly been infected as shown by the presence of B. burgdorferi in the blood at 7 days, and by seroconversion for P39 at 28 days (data not shown). In contrast, 12 of 12 sites from mice infected with the wild-type B31-A3 clone produced positive cultures at 28 days.

To rule out the possibility that the lp28-1Δvls clones were attenuated in any virulence properties other than that associated with vlsE, four immunodeficient C3H/Smn.Clcr.Hsd-scid (SCID) mice were inoculated with an A3Δvls knockout clone. It has been reported previously that lp28-1 minus clones exhibit full infectivity and can be cultured from all examined tissue sites in SCID mice (Labandeira-Rey et al., 2003; Purser et al., 2003). After 4 weeks post infection, positive cultures were recovered from 16 of 16 sites cultured (ear, heart, bladder and joint; data not shown). Taken together with the results from the immunocompetent C3H/HeN mouse infections, these results offer definitive evidence that the vls locus in B. burgdorferi is required for persistence within the mammalian host.

The left side of lp28-1 is not required for persistence or for antigenic switching at vlsE

Although the work described above provided strong evidence for the role of the vls locus in persistence, it remained unclear whether vlsE and the silent cassettes were the sole virulence determinants on lp28-1, and whether any other region on lp28-1 was required to promote antigenic switching at vlsE. In an attempt to answer these questions, we set out to obtain lp28-1 mutants which would contain only the vls locus and the necessary genes for autonomous replication of the plasmid. As shown in Fig. 2A, B31-A3 cells were transformed with the plasmid pTB55 that resulted in the deletion of the left side of the lp28-1 plasmid (BBF01–BBF19), leaving the right side of lp28-1 with the replication functions (BBF23–BBF26) and an intact vls locus. In addition, three small potential open reading frames of 56, 43, and 82 amino acids (BBF27, BBF28 and BBF30, respectively) with no predicted function remained between vls and the plasmid replication functions on the lp28-Δleft construct.

Figure 2.

Deletion of the left-end portion of lp28-1 by insertion of a replicated telomere.
A. Schematic of the construction strategy for the lp28-1Δleft plasmid. The target sequence for insertion (green; co-ordinates 7751–9037 of the complete lp28-1 sequence – see Experimental procedures) was chosen so that only the genes encoding for paralogous family proteins 49, 32, 50 and 57 that allow autonomous replication of lp28-1 (BBF23, 24, 25 and 26 shown as arrows arranged from left to right), the silent cassettes and vlsE would remain. Hairpin telomeres are shown as red hatched regions.
B. Analysis of the completed construct by field inversion gel electrophoresis. Genomic DNA from B31-A3 (lane 2) and a B31-A3/lp28-1Δleft clone (lane 3) are shown. A molecular weight marker (M) is shown in lane 1. The positions of the lp28 plasmids 1–4 as well as the left-end knockout plasmid are shown on the right.
C. Restriction analysis of the vlsE gene sequence from spirochetes cultivated in vitro or cultured from mice. The amplified vlsE genes from either B31-A3 (wild type), B31-A3/lp28-1Δleft-clone 2, or B31-A3/lp28-1Δleft-clone 4 were analysed undigested or digested with HphI both before and after inoculation in a C3H/HeN mouse as indicated. DNA fragments were resolved on a 2% MetaPhor agarose gel and visualized with ethidium bromide staining. The native B31-A3 vlsE sequence contains an HphI restriction site which results in a 400 bp band (lanes 7–9). Segmental gene conversion during antigenic variation of vlsE can result in removing this site and/or introduction of additional sites (lanes 10–12) due to its prevalence throughout the silent vls cassettes. A 100 bp DNA marker is shown on the left.

In total, 15 transformants were recovered and screened by PCR for the presence of the kan gene, vlsE, and the absence of DNA sequence corresponding to the left end of lp28-1. Six clones matched these criteria and were further analysed under the same conditions previously described for analysis of the vls deletion clones. Five of these six clones contained a deletion of lp28-1 of the expected size, and one such clone is shown in Fig. 2B. Plasmid content was analysed in these clones by PCR screening, and two clones (A3 lp28-1Δleft-clone 2 and A3 lp28-1Δleft-clone 4) that showed identical plasmid profiles compared to the parental B31-A3 were chosen for further study.

Two groups of four C3H/HeN mice were inoculated with either the B31-A3 lp28-1Δleft-clone 2 or B31-A3 lp28-1Δleft-clone 4, and sample sites were tested for the presence of spirochetes at various times during a 4 week period. As shown in Table 4, blood samples from day 7 and ear samples from days 14 and 21 produced positive cultures from all of the mice infected with either of the two left end-deletion clones, and at day 28 post infection, spirochetes were recovered from ear, heart, bladder and joint tissue sites from all mice.

Table 4.  Effect of lp28-1 left-end deletion on B. burgdorferi infection in C3H/HeN mice.a
B. burgdorferi B31 cloneDay 7Day 14Day 21Day 28Total mice
BloodEarEarEarHeartBladderJointTotal sites
  • a. 

    Values listed correspond to number of cultures positive/number tested.

A3/(lp28-1Δleft-clone 2)4/44/44/44/44/44/44/416/164/4
A3/(lp28-1Δleft-clone 4)4/44/44/44/44/44/44/416/164/4

To confirm that vls recombination was occurring in the two left end-deletion isolates, cultures from day 21 were tested for antigenic switching at the vlsE locus. This was carried out by subjecting PCR products corresponding to the variable regions of vlsE to restriction analysis (Ohnishi et al., 2003) to identify new HphI restriction endonuclease cleavage sites introduced or removed by switching at vlsE. As shown in Fig. 2C, recombination at vlsE on lp28-1Δleft occurred at day 21 (lanes 11 and 12) as efficiently as in the wild-type B31-A3 clone (lane 10) with a full-length lp28-1 plasmid. Together, these results indicate that the silent cassettes and vlsE are indeed the only lp28-1-encoded genes required for persistence of B. burgdorferi in the mouse host and for antigenic switching at the vlsE locus, with the possible exception of the three small open reading frames BBF27, BBF28 and BBF30, which are likely too small for a functional role in infectivity or switching at vlsE.

The effect of VlsE on the ability of B. burgdorferi to reinfect a previously infected host

The vls recombination system of B. burgdorferi is similar to that found in the human pathogens causing African sleeping sickness (Cross, 1996; Donelson, 2003), malaria (Dzikowski et al., 2006) and gonorrhoea (Kline et al., 2003; Criss et al., 2005), where variation of surface antigens promotes persistence of the pathogen through evasion of host immune responses. These organisms are also capable of reinfecting previously afflicted individuals, presumably due to their antigenic variation systems. The absolute requirement for the presence of the vls locus in B. burgdorferi persistence led us to question whether the presence of VlsE might also impart a capacity for host reinfection. Specifically, we asked whether mice initially infected with the VlsE-deficient clone, which is completely cleared by 28 days, could be reinfected with wild-type B31-A3, which produces VlsE.

Our experimental scheme (Fig. 3) involved infection of 12 C3H/HeN mice with B31-A3Δvls. Progress of the infection was followed (see Fig. 3) with the expectation of clearance by week 3 based upon the results gathered from the A3Δvls infectivity experiments presented in Table 3. Positive cultures were obtained for all 12 mice at day 4 and day 10 post inoculation (data not shown). As expected, ear biopsies from all 12 mice at day 21 produced negative cultures for spirochete growth. After an additional week (day 28), the mice were split into three groups of four mice each and inoculated with blood-stimulated spirochetes (grown in the presence of 3.5% blood for 48 h). The infections were performed with either B31-A3, B31-A3Δvls, or B31-A3Δvls + vlsE clones. Because about 150 B. burgdorferi genes are turned on by growth in Barbour–Stoenner–Kelly II (BSK)-II medium supplemented with blood relative to growth in BSK-II alone (Tokarz et al., 2004), there may exist the need to upregulate genes that may modify the arrangement of surface antigens, including VlsE. Indeed, it has been reported that VlsE expression levels increase dramatically during infection within the mouse, and this increase correlates closely with a rise in antibody production in the host (Liang et al., 2004). In spite of blood stimulation, all of the mice failed to produce any positive cultures for spirochetes from any of the tissue sites tested after the 4 week period (Fig. 3, bottom). The inability of the B31-A3Δvls + vlsE and the wild-type B31-A3 clones to reinfect VlsE-naïve mice suggests that antibodies directed against B. burgdorferi antigens other than VlsE can prevent reinfection.

Figure 3.

Experimental strategy for assaying reinfection in C3H/HeN mice that had previously been infected with and cleared B. burgdorferi B31-A3Δvls. Initial infection was carried out on 12 mice. Spirochetes were cultured from blood (days 4 and 7) or ear biopsies (days 14 and 21) to confirm infection and monitor clearance. One hundred per cent of the mice cleared the B. burgdorferi B31-A3Δvls infection by day 21. They were divided into three groups and reinfected at day 28 with either wild-type B31-A3, A3Δvls, or A3Δvls + pBSV2::vlsE. Blood samples, ear biopsies and organs were cultured for spirochetes at the indicated times to monitor reinfection.

Does VlsE shield surface antigens of B. burgdorferi?

A daunting question regarding the role of VlsE in evading the host immune response has been how such a feat is accomplished through sequence variation of this single lipoprotein, despite the presence of a substantial number of additional antigens residing on the bacterial surface. One possible mechanism, similar to what is used by the protozoan pathogens causing African sleeping sickness (Taylor and Rudenko, 2006) and malaria (Kraemer and Smith, 2006), is that VlsE hides other surface antigens by coating the surface of the pathogen. To determine whether B. burgdorferi utilizes VlsE to mask surface antigens, we performed bacteriacidal assays using immune serum from mice infected with B31-A3 or B31-A3Δvls. Antiserum from the vls minus infection was observed to be directed against a variety of surface antigens other than VlsE as shown by Western blotting (data not shown). The expectation was that this antiserum would effectively kill spirochetes lacking VlsE. In the case of wild-type B. burgdorferi, however, if VlsE obscures other surface antigens, then a protective effect would be expected from its presence. Similarly, addition of VlsE back to the vls deletion strain would be expected to exhibit the same protective effect.

Figure 4 shows the results of such an experiment. Serum from an uninfected mouse resulted in about 15% non-specific killing of all three spirochetes tested in the assay. Under conditions where about 60% of the Δvls strain was killed, no protective effect was observed either with the wild-type B. burgdorferi strain or with the complemented vls mutant. Serum from a wild-type infection also showed no difference in killing between the wild-type, vls deletion and the complemented vls deletion.

Figure 4.

Survival of B. burgdorferi B31 clones in a borreliacidal assay with sera from mice infected with either B31-A3 or B31- A3Δvls. Immune sera concentrations were adjusted such that an approximate 50% survival rate was observed, to maximize any differences in response by the different B. burgdorferi strains as described in Experimental procedures. The experiment was performed for both (A) spirochetes (grown in BSK-II) and (B) blood-stimulated spirochetes (grown in BSK-II + 3.5% mouse blood). Bar graphs represent the mean and standard deviation from experiments run in triplicate.

These results suggested that VlsE does not impart a protective masking of other B. burgdorferi surface antigens. However, we repeated the experiment in Fig. 4A with blood-stimulated spirochetes, where the gene expression profile would more closely simulate that observed in the mouse and perhaps modify the pattern of surface antigens observed in cultured spirochetes. Figure 4B shows the killing assay repeated with these blood-stimulated spirochetes. Once again, no difference in killing was observed in the presence or absence of VlsE, suggesting that masking of surface antigens by VlsE may not be the mechanism by which persistence is imparted to B. burgdorferi by VlsE.


The vls locus is the only lp28-1-encoded determinant required for persistent infection and for switching at vlsE

Mounting evidence over the years has suggested that the lp28-1 plasmid of B. burgdorferi, which contains the vls antigenic variation locus, is important for the persistence associated with Lyme disease (Purser and Norris, 2000; Labandeira-Rey and Skare, 2001; Labandeira-Rey et al., 2003; Lawrenz et al., 2004), and complementation studies have suggested that VlsE variation is important for virulence (Lawrenz et al., 2004). Despite these indications, it has remained unclear whether genes other than vlsE and the silent cassettes on lp28-1 are responsible for the lp28-1 minus phenotype, as generating vls knockouts has thus far proven difficult. The results presented in this study demonstrate for the first time that the vlsE gene and silent cassettes that make up the vls locus are absolutely required for persistence. Furthermore, the long-term survival of the vls knockout clones in immunodeficient SCID mice is consistent with the findings that the lp28-1 plasmid is not required for persistent infection in the absence of an adaptive immune response (Labandeira-Rey et al., 2003; Purser et al., 2003; Lawrenz et al., 2004), and confirms the idea that vls recombination functions to evade the humoral immune response in the mouse host (Norris, 2006; Zhang et al., 1997).

Another continuing question concerning the lp28-1 minus phenotype observed in immunocompetent mice has been whether any of the non-vls genes on lp28-1 play a necessary role in spirochete persistence. A previous study showed that a number of clinical isolates of B. burgdorferi from Lyme disease patients were missing the lp28-1-encoded BBF01, suggesting that the gene product is not required for infectivity (Iyer et al., 2003). Our results extend this study and show that B. burgdorferi lacking BBF01–BBF19 are fully infectious and persistent in immunocompetent mice, and strongly suggest that the silent vls cassettes and vlsE are the sole lp28-1 components required for spirochete virulence. Moreover, our findings that the lp28-1 left end-deletion clones carried out vls recombination in immunocompetent C3H mice, indicate that protein factors required for antigenic switching at vlsE are not carried on lp28-1 but must be encoded elsewhere in the B. burgdorferi genome.

VlsE does not appear to confer reinfection ability

The absolute requirement for the vls locus in persistence associated with Lyme disease led us to question whether mice originally infected with spirochetes lacking VlsE could be reinfected with a fully infectious clone that has an intact vls recombination system. Previous studies have shown that cured mice are immune to reinfection when challenged by intradermal inoculation or with an autograft from infected mice (Barthold, 1993; Piesman et al., 1997). However, these studies involved challenging actively immunized mice previously infected with a fully infectious clone of B. burgdorferi with autologous isolates, compared with the present study, where mice naturally cleared of partially infectious spirochetes are challenged with fully infectious clones. While the former deals with natural immunity to antigenically similar bacteria previously encountered by the host immune response, the latter involves immunity to spirochetes that have a clear antigenic difference in the form of the VlsE protein important for immune evasion and persistence.

We found that C3H/HeN mice immunized by intradermal inoculation with spirochetes lacking vlsE are resistant to reinfection with a vls recombination-competent clone, suggesting that vlsE alone is not sufficient for complete immune evasion and reinfection by B. burgdorferi. It is entirely possible that expression levels of vlsE and/or other important factors involved in establishing infection are insufficient in an inoculum cultivated in vitro. Thus, VlsE may require expression levels greater than those occurring during propagation in vitro in order for it to offer protection from the host immune response. It has been reported that VlsE expression levels increase dramatically in some tissues during infection within the mouse (Crother et al., 2003; Liang et al., 2004). It is noteworthy, however, that the complementing plasmid pBSV2::vlsE promotes VlsE expression at a level several times higher than in a wild-type spirochete grown in culture (Lawrenz et al., 2004). This high level of VlsE expression was verified in our complemented construct by Western blotting (data not shown) using rabbit antiserum to the C6 peptide (Liang et al., 1999). This increase in VlsE did not result in any increase in reinfection ability (Fig. 3) or any resistance to killing by bactericidal antibody (Fig. 4). While the results from our reinfection experiments are by no means conclusive, they support previous conclusions that natural immunity to reinfection with B. burgdorferi is complex and likely involves immunosuppression by the spirochete that is difficult to re-establish in an immunized mouse host (Piesman et al., 1997; Diterich et al., 2001; 2003). Nonetheless, B. burgdorferi clearly does not possess the ability to reinfect a previously infected host in the way that T. brucei (Taylor and Rudenko, 2006), P. falciparum (Dzikowski et al., 2006) and N. gonorrhea (Kline et al., 2003; Criss et al., 2005) do, suggesting differences in the mechanism by which immune evasion is accomplished.

Mechanism of VlsE-promoted escape from immune surveillance

The inability of VlsE to confer reinfection ability to B. burgdorferi suggested that VlsE may not coat the surface of the spirochete as has been observed for the antigenically variable proteins of some protozoal parasites, such as trypanasomes (Taylor and Rudenko, 2006) and Plasmodium (Dzikowski et al., 2006). We tested whether VlsE acts to shield other surface antigens using borreliacidal assays. Neither the presence of the vlsE gene in its natural environment with low-level expression (data not shown), nor on a complementing plasmid with higher expression levels (Lawrenz et al., 2004; data not shown), conferred protection to borreliacidal antiserum. A caveat with these experiments is that, despite in vitro cultivation in the presence of 3.5% blood, it is entirely possible that altered expression of VlsE and/or other B. burgdorferi proteins during infection in the mouse host might alter the surface topology of the spirochete. We cannot rule out this possibility at present. Nonetheless, our experiments raise the possibility that a mechanism other than VlsE shielding of surface antigens may be involved in making B. burgdorferi antigenically inert.

It has been proposed that during infection, other surface antigens may be downregulated, leaving VlsE as the only surface-exposed protein (Liang et al., 2002). A number of studies have demonstrated that the major outer surface protein, OspC, must be downregulated shortly after being introduced into a mouse host in order to avoid recognition by host antibodies (Liang et al., 2004; Xu et al., 2006; Tilly et al., 2007). On the other hand, expression levels of other lipoproteins, such as DbpA and BBF01, either remain unchanged or are increased under the influence of immune pressure in immunocompetent mice during infection (Liang et al., 2004). Thus it seems unlikely that global downregulation of surface antigens other than VlsE would play a role in the immune evasion mechanism in B. burgdorferi.

Another suggested method by which surface antigens other than VlsE escape recognition is that VlsE might be a T-cell-independent antigen that could directly stimulate B cells to produce antibodies, and the potent humoral response generated by VlsE may serve to override antibody production against other potential surface antigens (Philipp et al., 2001). Interestingly, we have observed that B31-A3Δvls + pBSV2::vlsE clones are cleared faster in C3H/HeN-immunocompetent mice relative to non-VlsE-complemented B31-A3Δvls clones (this study and unpublished results). As shown in Table 3, the A3Δvls + pBSV2::vlsE clone cleared at a faster rate than either the vls knockout clones or the lp28-1 minus control, and we have reproduced this effect with both this same clone and an additional VlsE-complement clone in separate experiments. It is conceivable that this outcome could be the result of direct stimulation of B cells by VlsE, and this modulating ability would perceivably result in more effective clearance of these spirochetes due to the absence of vls recombination in these clones. For this reason, we favour the idea that VlsE may function to confine the host adaptive response, and experiments are currently underway to test this hypothesis. Regardless, what is certain is that the generation of vls knockouts of B. burgdorferi will serve to aid future studies on the various aspects of antigenic variation in the Lyme disease spirochete.

Experimental procedures

Bacterial strains

Borrelia burgdorferi B31-A3 and B31-A1, kind gifts of Patti Rosa, are clonal isolates from B31 MI, and their respective infectivities and plasmid profiles were determined in a previous study (Elias et al., 2002). B31-A3 was found to be fully infectious and missing only the cp9 plasmid, while B31-A1 is an intermediate-infectivity clone missing both cp9 and lp28-1 (Table 2). All B. burgdorferi clones were cultivated in liquid BSK-II medium supplemented with 6% rabbit serum (Cedarlane Laboratories) and incubated at 35°C in a 1.5% CO2 environment. B31-A3Δvls-clone 2 was transformed with a modified form of pMBL20 (Lawrenz et al., 2004), where the kanamycin gene was replaced by cloning in a gentamycin cassette at the unique NdeI and NcoI sites. pMBL20 was a generous gift from Steven Norris, and consists of the pBSV2 shuttle vector and the vlsE gene including the native promoter and leader sequence. An E. coli sbc strain was utilized to maintain plasmid constructs containing replicated telomeres as described previously (Chaconas et al., 2001), while E. coli DH5α cells were used to maintain pMBL20.

Plasmid construction

In construction of the vls deletion plasmid pTB44, the region from 17 296 to 18 800 of the complete lp28-1 sequence (see was PCR amplified using primers B513 and B514 (Table 1) and cloned into the pCR BluntII-TOPO vector (Invitrogen). The insert was recovered by cleavage with KpnI and XhoI, and cloned into KpnI–XhoI-digested pGCL47-4, which carries a flgBp-driven kanamycin gene and a 70 bp replicated telomere from the left end of lp17. The lp28-1 left end-deletion plasmid, pTB55, was constructed by PCR amplifying the region from 7751 to 9037 using the primers B848 and B849, followed by TOPO cloning. The cloned insert was retrieved by digestion with XbaI and EagI, and cloned into the respective sites in pGCL47-4.

Borrelia burgdorferi transformation

Borrelia burgdorferi cells were electroporated as described previously (Samuels, 1995) with a total of 50 µg of DNA. After electroporation, bacterial cells were immediately resuspended in 1 ml of prewarmed (35°C) BSK-II liquid medium and then pipetted into 9 ml of prewarmed BSK-II. Following overnight recovery at 35°C, the transformation was diluted into a total of 100 ml BSK-II supplemented with 200 µg ml−1 kanamycin and distributed into 96-well plates (250 μl each well), and incubated at 35°C for 1–3 weeks. Wells that exhibited a change in colour from red to yellow as an indication of growth were verified by dark-field microscopy for the presence of viable spirochetes and chosen for PCR analysis.

Mouse infections

All animal infections were carried out in accordance with approved protocols from the University of Calgary Animal Research Centre. Four-week-old male C3H/HeN (wild-type) and C3H.C-Prkdcscid/IcrSmnHsd (SCID) mice (Harlan, Indianapolis, IN) were infected by both intraperitoneal (1 × 104 cells ml−1) and subcutaneous (1 × 103 cells ml−1) needle inoculation. Blood samples were obtained by bleeding at the saphenous vein, and a total of 50 μl was used to inoculate 1.75 ml of BSK-II containing Borrelia antibiotic cocktail (0.02 mg ml−1 phosphomycin, 0.05 mg ml−1 rifampicin and 2.5 μg ml−1 amphotericin B). For growth curve measurements, a total of 50 μl of blood was taken from the saphenous vein 7 days post infection, and 25 μl was used to inoculate a 5 ml BSK-II culture with or without the indicated antibiotic. Ear, heart, bladder and joint tissues were obtained aseptically and cultured in 1.5 ml of BSK-II containing Borrelia antibiotic cocktail. Dark-field microscopy was used to determine the presence or absence of viable spirochetes for each cultured tissue sample.

vlsE switching assay

A 775 bp segment containing the six variable regions of vlsE was PCR amplified using the primers B248 and B249 (Table 1). PCR products were purified using QIAquick PCR spin columns (Qiagen), quantified by absorbance at 260 nm, and digested in reactions containing 200 ng DNA and 2 units of HphI (NEB) as per the manufacturer's instructions at 37°C for 1 h. Reactions were analysed on a 10 cm 2% MetaPhor agarose gel in 1× Tris-acetate-EDTA buffer at 67 V for 2.5 h.

Borreliacidal assay

An assay to test the borreliacidal effects of mouse sera was adapted from a previously described assay (Piesman et al., 1997). Serum was obtained from a cardiac puncture of mice infected with either B31-A3 (wild type) or B31-A3Δvls-clone 2 at day 15 post infection. Mouse serum (10 μl) was diluted 1:2 in a buffer containing 10 mM HEPES pH 7.4, 1 mM MgCl2, and 150 mM NaCl to produce a final volume of 20 μl. Spirochetes grown in BSK-II with appropriate drug selection to a concentration of 108 cells ml−1 were pelleted, washed three times in HEPES buffer, and resuspended to a final concentration of 6 × 106 cells ml−1. For cultures grown in the presence of blood, spirochetes were cultured in BSK-II containing 3.5% mouse blood and incubated at 35°C for 48 h before centrifugation. Equal volumes (20 μl) of resuspended spirochetes were added to diluted sera and incubated at 35°C for 15 h. The number of viable cells was counted using a Petroff-Hauser counting chamber under dark field. Only those cells that were motile and retained a characteristically spiral shape (i.e. no membrane blebbing) were determined viable. The per cent survival was calculated by comparing the number of viable cells in each experimental sample with that of a control sample containing spirochetes incubated in HEPES buffer alone.


We are grateful to Yvonne Tourand for her efforts to establish the HphI RFLP assay for vlsE switching that was used in Fig. 2C. We would like to thank Steven Norris (University of Texas, Houston) for the gift of the pBSV2 shuttle vector containing the vlsE gene and for comments on the manuscript. We would also like to thank Patti Rosa for kindly providing the B. burgdorferi B31-A1 and B31-A3 clones used in this study, and Mario Philipp for providing antiserum to the VlsE C6 peptide. Our gratitude is also extended to laboratory members for helpful comments on the manuscript. This research was undertaken, in part, thanks to funding from the Canadian Institutes of Health Research (CIHR), the Canada Research Chairs Program, and from the Alberta Heritage Fund for Medical Research. G.C. was supported by a Scientist Award from the Alberta Heritage Foundation for Medical Research and holds a Canada Research Chair in the Molecular Biology of Lyme Disease.