Cis‐acting DNA elements flanking the variable major protein expression site of Borrelia hermsii are required for murine persistence

Abstract In Borrelia hermsii, antigenic variation occurs as a result of a nonreciprocal gene conversion event that places one of ~60 silent variable major protein genes downstream of a single, transcriptionally active promoter. The upstream homology sequence (UHS) and downstream homology sequence (DHS) are two putative cis‐acting DNA elements that have been predicted to serve as crossover points for homologous recombination. In this report, a targeted deletion/in cis complementation technique was used to directly evaluate the role for these elements in antigenic switching. The results demonstrate that deletion of the expression site results in an inability of the pathogen to relapse in immunocompetent mice, and that the utilized technique was successful in producing complemented mutants that are capable of antigenic switching. Additional complemented clones with mutations in the UHS and DHS of the expressed locus were then generated and evaluated for their ability to relapse in immunocompetent mice. Mutation of the UHS and inverted repeat sequence within the DHS rendered these mutants incapable of relapsing. Overall, the results establish the requirement of the inverted repeat of the DHS for antigenic switching, and support the importance of the UHS for B. hermsii persistence in the mammalian host.

Human infection with relapsing fever spirochetes results in a series of febrile episodes that are interrupted by periods of apparent wellness.
The waxing and waning fever that is characteristic of relapsing fever is a direct result of spirochetal antigenic variation and sequential immune evasion. Febrile episodes correspond to outgrowths of spirochetal populations that possess a predominant serotype. Bacterial density during these episodes can be >10 7 Borrelia per ml of blood, while afebrile periods are marked by serotype-specific immune responses that greatly reduce or eliminate the bacterial load in blood (Dworkin, Schwan, & Anderson, 2002;Raffel, Battisti, Fischer, & Schwan, 2014).
The serotypes of relapsing fever Borrelia are characterized by antigenically distinct immunodominant outer surface lipoproteins (Barbour, Tessier, & Stoenner, 1982;Barstad, Coligan, Raum, & Barbour, 1985). Thus, newly "switched" spirochetes are not yet recognized by the host's immune system, and serve as progenitors for the next febrile relapse. The lagging host humoral response will clear the new serotype in time, but the cyclic proliferation of immune escape variants and subsequent immune clearance of these populations will be repeated due to the process of antigenic variation (Barbour & Restrepo, 2000).
Not only is antigenic switching linked to the clinical course of disease, it serves to prolong the time in which the bacteria are found in blood. This persistence is an essential feature of the lifecycle of these pathogens, as it increases the chances for tick acquisition (Barbour & Restrepo, 2000;Lopez, Mccoy, Krajacich, & Schwan, 2011). Moreover, it has been demonstrated that without an antigenic variation system, relapsing fever Borrelia are unable to persist in mammalian hosts, and therefore, unable to cause febrile relapses (Raffel et al., 2014).
Together, the lipoproteins are known as variable major proteins (Vmp), and the B. hermsii genome harbors at least 59 silent vmp gene copies . Only one locus is transcriptionally active during infection of the mammalian host, and this expression site (termed vmp Ex for the purposes of this study) is located near the telomere end of an approximately 27.8 kb linear plasmid named lpE27 (formerly lp28-1) (Barbour, 2016;Barbour, Burman, Carter, Kitten, & Bergstrom, 1991;Kitten & Barbour, 1990;Plasterk, Simon, & Barbour, 1985). Antigenic variation occurs through a nonreciprocal gene conversion event that duplicates one of the silent, promoterless vmps downstream of the active promoter at vmp Ex Plasterk et al., 1985). Two cis-acting DNA elements have been implicated as crossover points for homologous recombination, and are found flanking both vmp Ex and silent vmp loci Dai et al., 2006;Kitten & Barbour, 1990). The first site, the upstream homology sequence (UHS), is 61-62 bp in length, and is partially intragenic extending 7 bp upstream of the transcriptional start site and 26 bp into the Vmp coding region Dai et al., 2006).
The second element, the 214 bp downstream homology sequence (DHS), is entirely extragenic and lies a variable distance away from the Vmp stop codon Kitten & Barbour, 1990;Restrepo et al., 1992). Within the DHS, a 31-bp inverted repeat sequence is found between positions 47 and 77 . Inverted repeat (IR) sequences form putative secondary stem loop structures in DNA, and these structures are known to be highly recombinogenic (Barbour, Dai, Restrepo, Stoenner, & Frank, 2006). Dai et al. (2006) proposed the following mechanism for antigenic switching in B. hermsii based on sequencing analysis: recombination is initiated in the distal portion of the DHS, and the replication fork for repair extends upstream using a silent vmp as a template until it reaches the UHS, where replication is terminated. This 3′-5′ mechanism is supported by two key findings. First, a chimeric vmp Ex site was found in a relapse variant where the upstream portion remained the same as the starting serotype and the downstream portion represented a "switched" vmp, indicative of premature replication fork termination . Second, the initiation of recombination in the DHS is supported by the finding that the vmp Ex is oriented such that the DHS lies near the telomere of lpE27, a structure that is known to be inherently recombinogenic (Chaconas, Stewart, Tilly, Bono, & Rosa, 2001). A role for the UHS and DHS in the rate of genetic recombination has also been verified. Greater sequence similarity in the UHS and a shorter distance from the stop codon of the vmp to its downstream DHS predict a higher recombination rate of the archived vmp into the expression site .
The objective for the study presented herein was to verify the importance of these cis-acting DNA elements for gene conversion.
Thus, we hypothesized that the UHS and DHS are required for the recombination event that leads to antigenic variation. The underlying principle behind our method is that without a functional antigenic variation system, B. hermsii is unable to persist in the murine host. Indeed, Raffel et al. (2014) demonstrated that when the vmp Ex promoter was disrupted, B. hermsii was no longer capable of relapsing in the murine host. To start, we employed a direct, mutational approach whereby the vmp Ex locus was deleted from lpE27 in wild-type B. hermsii using a targeted deletion technique. As expected, the vmp Ex deletion mutant was not capable of persistence in an immunocompetent murine host, but did persist in severe combined immunodeficient (SCID) mice, indicating that immune clearance in the former is a direct result of the absence of antigenic switching. Next, a novel targeted deletion/in cis complementation technique was used to directly evaluate the role of the putative cis-acting sites in antigenic switching. The results from this work strongly support the requirement of the DHS-IR for antigenic switching in B. hermsii, and suggest that the UHS is important for both proper expression and gene conversion of the vmp Ex locus.
Importantly, the findings corroborate, through mutational analysis, the mechanism for vmp Ex recombination proposed by previous investigators .

| Ethics statement
The experimental procedures involving strains of inbred mice were carried out in accordance with the American Association for

| Murine infection
Male, C.B-17/IcrHsd-Prkdcscid (SCID) or C3H/HeNHsd (C3H) mice 4-6 weeks of age were purchased from Harlan (Indianapolis, IN). While the SCID mice are from a different background than the wild-type mice, various strains of SCID mice infected with B. hermsii demonstrate similar, persistently high levels of spirochetemia following inoculation (Alugupalli et al., 2003;James, Rogovskyy, Crowley, & Bankhead, 2016;Raffel et al., 2014). The animals were subcutaneously inoculated with a B. hermsii clone with 1 × 10 6 total spirochetes per mouse in 100 μl. All clones, including wild type, were obtained by serial dilution and plating. After plating, they were passaged no more than two times in vitro from frozen glycerol stocks prior to murine inoculation. To confirm infection, 15 μl of blood was drawn from mice via the saphenous vein each day over the course of 10 days. A total of 5 μl of blood was immediately diluted 1:3 with BSK-II medium (Barbour, 1984), and then 10 μl of diluted blood was examined under a dark field microscope. The remaining 10 μl of whole blood was cultured in 1 ml of BSK-II containing Borrelia antibiotic cocktail (0.02 mg/ml phosphomycin, 0.05 mg/ml rifampicin, and 2.5 mg/ml amphotericin B). Blood cultures were incubated at 35°C under 1.5% CO 2 for 3-4 weeks. The blood cultures were periodically examined via dark-field microscopy for the presence of viable Borrelia cells. A mouse whose blood sample(s) showed viable spirochetes via culture and/or microscopy was considered infected.

| Bacterial strains, culture conditions, and DNA extraction
The isolation and characterization of B. hermsii DAH has been described previously (Porcella et al., 2005;Schwan, Schrumpf, Hinnebusch, Anderson, & Konkel, 1996), and was acquired as a gift from George Chaconas, who obtained it from Tom Schwan. The DAH strain is nearly identical to isolate HS1, with the latter being the subject of many previous studies on antigenic variation in B. hermsii (Barbour, 2016;Raffel et al., 2014). All wild-type and mutant clones of B. hermsii were cultivated at 35°C under 1.5% CO 2 in modified Barbour-Stoenner-Kelly medium (BSK-II) supplemented with 12% rabbit serum (Accurate Chemical and Scientific Corp., Westbury, NY).
Cell densities and growth phase were monitored under dark-field microscopy and enumerated using a Petroff-Hausser counting chamber. DNA from B. hermsii used in Southern blotting, field inversion gels, and PCR analysis was extracted from in vitro grown cultures using a plasmid midi kit (Qiagen, Valencia, CA).
All plasmids generated herein were propagated in EC19 Escherichia coli cells (Hove, Haldorson, Magunda, & Bankhead, 2014). To obtain clonal colonies after transformation into E. coli, plates were incubated at 35°C overnight. All cultures in liquid broth were incubated with shaking overnight at 30°C. Plasmids for further cloning or sequence verification were extracted using a plasmid midi kit (Qiagen). Plasmids for transformation into B. hermsii were extracted using a midi kit (Qiagen), and further ethanol precipitated to concentrate the DNA.

| Complementation and UHS/DHS mutant plasmid construction
The plasmid pAE160 was generated for in cis restoration of vmp Ex on native plasmid lpE27 using the targeted deletion technique. Utilizing the XhoI and AscI restriction enzyme sites on pAE12, the target sequence was replaced by cloning a PCR amplified downstream portion of lpE27 with primers P626 and P627, producing pAE16. This 986 base pair target sequence lies between base pairs 10,339 and 11,324 on the GenBank annotated B. hermsii DAH lpE27 plasmid (accession number CP000273). Next, the variable membrane protein expression site, vmp Ex , was amplified with P541 and P542, and cloned into pAE16 using existing NheI and FspI sites, respectively. The resultant plasmid, pAE160, was then transformed into wild-type B. hermsii, producing the complemented mutant.
To generate plasmids that have mutated UHS and DHS, vmp Ex sites with different sequence alterations were commercially synthesized (GenScript, Piscataway, NJ) and cloned into pAE160 using NheI and FspI sites. By replacing the native vmp Ex with the mutated versions, plasmids pAE160UHS and pAE160DHS were generated. All plasmids were sequenced and verified prior to transformation into wild-type B. hermsii.

| Borrelia hermsii transformation and mutant screening
Borrelia hermsii DAH electrocompetent cells were prepared and electroporated according to methods described previously (Fine, Earnhart, & Marconi, 2011;Samuels, 1995). Following electroporation, the cells were treated as outlined in James et al. (2016) to obtain clonal populations. Briefly, electroporated cells were immediately placed in 5 ml of 35°C BSK-II, where they were allowed to recover overnight. The 5 ml recovery culture was then added to 45 ml of fresh, preincubated BSK-II under kanamycin selection (200 μg/ml). Once viable spirochetes were observed in this polyclonal culture, clonal isolation of transformants was achieved through serial dilution in fresh BSK-II with kanamycin, and plating in 96-well culture plates (Sarstedt, Newton, NC).
Clonal populations of mutants were initially identified by a change of media color from red to yellow. Positive wells were PCR screened for the presence of kan R , and correct clones were subcultured in 50 ml fresh BSK-II with kanamycin for DNA extraction. DNA from the mutants was subjected to PCR for the expected lpE27 integration site. Primers P233 and P537 were used to amplify and sequence the integration site of Bh∆vmp Ex ; primers P854 and P855 were used for Bh::Comp, Bh::UHS AS , and Bh::DHS ΔIR . All mutant DNA was also PCR amplified with P308 and P309 to verify the absence of vmp Ex in Bh∆vmp Ex , and to sequence vmp Ex in the other mutants.

| Antibody generation
The gene of the current Vmp expressed in our stock culture of B. hermsii DAH, vlpA7, was amplified with P558 and P559, possessing NdeI and BamHI 5′ tails, respectively. The 1,016 base pair product was cloned into pET15b (Novagen-Merck Millipore), and recombinant protein was obtained by Ni-NTA agarose column protein purification (Qiagen).
Proteins were concentrated with Pierce Protein Concentrators 9K MWCO and quantified with a Micro BCA Protein Assay kit (Thermo Scientific, Marietta, OH). Recombinant VlpA7 (58.5 μg) was mixed with TiterMax Gold Adjuvant (Sigma Aldrich, St. Louis, MS) following the manufacturer's instructions, and injected subcutaneously into three C3H/HeNHsd mice (Harlan, Indianapolis, IN). Thirty-four days post inoculation, the mice were exsanguinated and humanely euthanized.
Collected blood was centrifuged at 6,000g for 10 min, serum was pooled, and stored at −80°C until use in immunoblotting.

| Western blotting
Wild-type or mutant B. hermsii were grown to late-log phase, counted, and centrifuged (6,000g for 15 min) to achieve duplicate pellets of 1 × 10 9 cells. Pellets were resuspended in 450 μl PBS + 5 mmol/L MgCl 2 . Eight units of proteinase K were added to one sample and incubated at room temperature for 40 min. The second aliquot of cells served as an untreated control. Ten microliter of phenylmethanesulfonylfluoride (0.2 mol/L) was added to inhibit further proteinase K activity, and the cells were washed in 500 μl of PBS-MgCl 2 two times. Cells were resuspended in 950 μl PBS-MgCl 2 and 500 μl of sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer (Rogovskyy & Bankhead, 2013). Lysates were stored at −20°C until use.
Ten microliter of each lysate (~1 × 10 7 cells) was heated at 100°C for 5 min prior to loading on a 15% acrylamide gel. Samples were electrophoresed in a Tris-glycine buffer containing 0.01% SDS.
Resolved proteins were transferred to a 0.45μm pore nitrocellulose membrane (Bio-Rad, Hercules, CA). The membranes were blocked in 5% nonfat dry milk in TBS overnight at 4°C, then incubated with either 1:750 diluted rabbit anti-B. burgdorferi FlaB antibody (Rockland Immunochemicals, Gilbertsville, PA) or 1:1,000 anti-VlpA7 antibody at room temperature for 1 hr. The membranes were washed three times with TBST, then incubated at room temperature for 1 hr with a 1:5,000 dilution of either donkey anti-mouse or anti-rabbit horseradish peroxidase-conjugated secondary antibody (Jackson ImmunoResearch Laboratories, West Grove, PA). After washing two times with TBST and once with TBS, the blots were developed using an enhanced chemiluminescent substrate (Bio-Rad).

| Field inversion gel electrophoresis
Plasmid DNA (500 ng) was electrophoresed on a 0.7% SeaKem LE Agarose gel (Lonza, Basel, Switzerland) with 0.5× TBE buffer. DNA was initially electrophoresed without inversion at 100 V for 15 min, followed by 24 hr of inversion with recirculation of buffer at 4°C. The reverse-pulse electrical field was supplied by a PPI-200 Programmable Power Inverter (MJ Research, Watertown, MA) on Program 3.

| In vitro growth assays
Wild-type B. hermsii and all mutant clones were grown to late-log phase and subcultured in triplicate to a cell density of 1 × 10 5 spirochetes per ml. Spirochetes were enumerated at 24-hr intervals and expressed as mean densities with standard deviation. Cell densities and growth phase were monitored by visualization under dark-field microscopy and enumerated using a Petroff-Hausser counting chamber.

| Sequencing of vmp Ex
The vmp Ex locus of spirochetes recovered from exponential-phase murine blood cultures were sequenced to assess antigenic switching.
Blood cultures were diluted 1:5 in PBS, and subjected to standard PCR with primers P308 and P309 (Table S1). The resultant vmp Ex amplicon was cleaned using the QIAquick PCR Purification Kit (Qiagen), and submitted for sequencing using the "Power Read" option for difficult templates (Eurofins Operon, eurofinsgenomic.com). For inconclusive sequencing results, the vmp Ex amplicon was cloned into pJET1.2/blunt (Thermo Fisher Scientific, Waltham, MA), transformed into DH5α Escherichia coli, and plated to obtain individual colonies. Five individual colony-forming units from each blood culture were inoculated into Luria broth media and incubated overnight at 37°C. The pJET 1.2/vmp Ex amplicon constructs were extracted with a QIAprep Mini kit (Qiagen), and submitted for sequencing as described above.

| Statistical analyses
All analyses were performed using the R statistical platform (version 3.4.0, R Core Team, 2017). Linear regression and analysis of covariance was applied for statistical comparison of growth curves. Analysis of variance (ANOVA) was used to compare the mean maximum cell density of each strain to one another. Fisher's exact test was used in cases where data could be applied to contingency tables (murine infection data). The ps < .05 were considered statistically significant.

| Targeted deletion of the B. hermsii vmp Ex locus
A recent study by Raffel et al. (2014) demonstrated that vmp Ex can be disrupted through the replacement of the promoter and 5′ region of vmp Ex with an antibiotic resistance cassette. In order to verify the importance of the UHS and DHS in antigenic switching for this study, a construct with an intact promoter region was required. Thus, our methods differed from the previously published study in that we employed a targeted deletion technique, and later, a similar in cis complementation technique to manipulate vmp Ex . To validate our methods, we first genetically deleted the vmp Ex locus. Targeted deletion has been previously used in B. burgdorferi, the causative agent of Lyme disease, but has never been applied to relapsing fever Borrelia spp. (Bankhead & Chaconas, 2007;Chaconas et al., 2001).
To utilize targeted deletion for the generation of a vmp Ex -B. hermsii clone (Bh∆vmp Ex ), the deletion plasmid pAE12 was produced. pAE12 was developed by cloning a target sequence identical to a 1,005 bp region on lpE27 immediately downstream of the genes for autonomous replication into a plasmid construct containing a replicated telomere and an aphI gene conferring kanamycin resistance (kan R ; Figure 1a).
Replicated telomeres (rtels) are naturally produced during linear plasmid replication in Borrelia spp., and DNA cleavage and rejoining of the rtel by a telomere resolvase (ResT) results in completion of the linear plasmid duplication event (Kobryn & Chaconas, 2001

| Bh∆vmp Ex does not establish persistent infection in immunocompetent mice
In principle, B. hermsii spirochetes that are unable to express and antigenically switch vmp Ex cannot persist in an immunocompetent host.
In order to establish a proof of principle and to determine whether a B. hermsii mutant with a genetically deleted vmp Ex would be cleared in immunocompetent mice, two groups of five C3H mice were inoculated with 1 × 10 6 wild-type B. hermsii or BhΔvmp Ex . While the infectious dose of wild-type B. hermsii can be as low as 1 organism, an inoculum of 10 6 provided the ability to consistently detect spirochetes in the blood of C3H mice by day 1-2 post inoculation .
Detection of the onset of spirochetemia was longer at increasingly lower doses, and much more variable, a finding confirmed by other investigators (Alugupalli et al., 2003). Blood samples were taken each day from each mouse throughout a 10-day experimental period. Mice whose blood samples revealed spirochetes either by direct microscopy or culture were considered positive for the day sampled. The results demonstrated that the wild-type clone was able to persist in all five mice throughout the 10-day study period, whereas mice infected with BhΔvmp Ex remained positive only up to day 5 post inoculation (Table 1). Similarly, two groups of five SCID mice that lack an acquired immune response were inoculated with both clones. BhΔvmp Ex mutant spirochetes were capable of persistence in SCID mice throughout the entire study period indicating the mutant does not have any inherent defects in its ability to infect and persist in the absence of an antibody response (Table 1). Consistent with a previously published study (Raffel et al., 2014), the data indicate that the lack of a functional vmp Ex results in clearance of infection from the blood in immunocompetent hosts, but not in immunodeficient mice. These results further verify the importance of vmp Ex for evasion of the acquired immune response.

| Construction of an in cis vmp Ex complement clone
In order to test the importance of the UHS and DHS for recombination, it was necessary to ensure that replacement of the native vmp Ex with an in vitro-generated locus did not, in itself, disrupt antigenic switching.

| Construction of UHS/DHS mutants
Sequencing analysis of infection versus relapse variants has implicated the conserved UHS and DHS elements as crossover points for the recombination event that leads to antigenic variation .

| Verification of vmp Ex complement and mutant clones
After transformation with pAE160, pAE160/UHS, or pAE160/DHS, clones were initially screened for the presence of kan R . The expected integration site of each mutant was PCR amplified and sequenced.
Correct clones were then subjected to Southern blotting, surface proteolysis, and immunoblotting. For all complemented mutants, DNA hybridization with kan R and lpE27-specific probes revealed that integration of pAE160 constructs occurred on the B. hermsii lpE27 plasmid (Figure 5a).
Immunoblotting of Bh::Comp and Bh::DHS ΔIR lysates with an Anti-VlpA7 antibody demonstrated binding before, but not after, proteinase K treatment (Figure 5b). Interestingly, the results also showed that the The vmp Ex sites in all mutants were found to be correct.
To determine if the mutant clones had replication defects, in vitro growth curves were generated ( Figure 6). All mutants were grown to late exponential phase and then subcultured to 1 × 10 5 spirochetes per ml at the start of the growth curve. During exponential growth (days 1-4 post subculture), the doubling times (and 95% confidence intervals) for Bh::Comp, Bh::UHS AS , and Bh::DHS ΔIR were 9.3 (8.8-9.9), 9.0 (8.3-9.8), and 10.5 (8.6-13.6) hr, respectively. Linear regression analysis of the slopes during these 72 hr of exponential growth revealed that none of the mutants had significantly different growth than the wild-type strain, nor one another (Table S2). Moreover, all mutants reached similar maximum spirochetal densities at day 6 post inoculation, and none of the mutant mean maximum cell densities were significantly different from wild type (data in Table S3). In sum, the data indicate that all B. hermsii mutants were successfully generated, and that none of the mutants displayed in vitro growth defects.

| A B. hermsii mutant with an in ciscomplemented vmp Ex locus persists in immunocompetent mice and undergoes antigenic variation
In order to determine the importance of the UHS and DHS in an-  Bh::Comp, the same vmp Ex PCR products from spirochetes recovered on day 10 were submitted for sequencing followed by BLAST analysis.
Spirochetes from mice 2, 4, and 5 had switched to VlpA18, while those from mouse three switched to VlpC19 (Data S1 depicts sequencing and alignment results for mice 3 and 4). Additionally, spirochetes from mouse 2 were also detected to switch to VlpD17. No results were obtained from mouse 1 due to a sequencing failure. Together, the results from the restriction fragment analysis and sequencing demonstrate that Bh::Comp is capable of antigenic variation and persistence during infection of an immunocompetent host.
(**p < .01, *p < .05). While the mechanisms that regulate the genetic recombination of the vmp Ex locus are unknown, it has been shown that antigenic switching occurs spontaneously in culture, and thus presumably, in SCID mice . Therefore, as a final attempt to verify the absence of antigenic switching, the experiment was independently repeated twice more with an additional two groups of three SCID mice

| Deletion of the DHS-resident inverted repeat results in immune clearance
As previously mentioned, a 31 bp IR sequence is found within the DHS between positions 47 and 77 . IR sequences can form secondary stem-loop structures in DNA, and these structures are known to be recombinogenic. To evaluate the importance of the IR sequence found within the vmp Ex DHS for antigenic switching, groups of five C3H and SCID mice were infected with Bh::DHS ΔIR as described above. To monitor for the presence of spirochetes, blood was collected from each mouse every day for 10 days post inoculation, and examined by direct microscopy or blood culture. The results showed that Bh::DHS ΔIR is able to initially infect immunocompetent C3H mice, but is cleared in all five mice by day 4 post inoculation (

| DISCUSSION
While previous studies have implicated the importance of the UHS and DHS for vmp Ex recombination Dai et al., 2006;Kitten & Barbour, 1990) Upstream of the vmp Ex promoter lies a 13 bp run of T-residues in the B. hermsii strain DAH . Within the next approximately 6 kb immediately upstream of the poly-T tract lie three IR sequences, each ~1 kb, that form putative secondary stem-loop structures . Together, the poly-T run and IR sequences are proposed to serve some role in the regulation of Vmp expression, or perhaps in vmp Ex recombination, because both are only found in the B. hermsii genome upstream of the vmp Ex promoter on lpE27 Sohaskey, Zuckert, & Barbour, 1999). In fact, it has been demonstrated that the B. hermsii poly-T run enhances transcription of a reporter construct in B. burgdorferi; no role for the IR sequences have been elucidated to date (Sohaskey et al., 1999). While the results herein did not seek to establish a role for the 5′ IR sequences, it is interesting that the complemented mutant, Bh::Comp, was able to undergo vmp Ex recombination, despite a physical separation of 1,169 bp between the poly-T run and the most proximal IR due to insertion of the kan R cassette (Figure 9). In wild-type B. hermsii, the most 3′ IR begins only 1 bp from the poly-T sequence . These findings suggest that the poly-T and IR sequences do not exert their influence on vmp Ex by spatial proximity, if they regulate Vmp expression and/or recombination at all.
For the Bh::UHS AS mutant, the major conclusion that can be drawn from the results is that some element in this region, other than the transcriptional start site and ribosomal binding site, is required for transcription or translation of vmp Ex . The −10 and −35 promoter elements lie outside of the UHS boundaries and were not altered, thus whatever sequence is required for Vmp expression seems to lie within the UHS. One possibility may be the conserved 4-mer and 6-mer palindromes depicted in Figure 4. Conceivably, if antigenic variation were possible in Bh::UHS AS , it cannot be excluded that a gene conversion event would replace the defective UHS with a functional UHS that would restore Vmp surface expression in vivo. The failure to detect any vmp Ex switching in Bh::UHS AS after 10 days of infection in SCID mice lends support to the possibility that disruption of the UHS results in an absolute obliteration of vmp Ex gene conversion, particularly in light of the finding that wild-type B. hermsii switches in SCID mice by day 10 post inoculation. A caveat, however, is that it is not currently known whether the gene conversion process requires vmp Ex transcription and/or translation. Moreover, further caution must be taken when F I G U R E 9 Schematic of the lpE27 plasmids in wild-type Borrelia hermsii and Bh::Comp. The poly-T run, inverted repeat sequences (depicted by arrows), vmp Ex promoter (p, arrow), and vmp Ex are depicted in wild-type (WT) and complemented strains (Bh::Comp). A kanamycin resistance (kan R ) selection marker is shown in Bh::Comp. Schematic is not to scale interpreting the Bh::UHS AS vmp Ex sequence analyses from SCID mice as it is unknown how the alterations in the UHS and DHS elements affect the kinetics of switching. Here, switching may have occurred, but at a lower rate than wild type and therefore, was not detected.
Future investigations can take advantage of the targeted deletion/in cis complementation technique to evaluate which sequences within the UHS are specifically required for Vmp expression.
The data presented herein indicate that the IR sequence is required for the recombination event that leads to relapses. This requirement provides a partial explanation for the conservation of the archived DHS sites that are scattered throughout the B. hermsii genome. A major question remains, however, and that is, "is the DHSresident IR structure sufficient for vmp Ex recombination?" Based on these results, it is unclear if the Bh::DHS ΔIR mutant did not persist in immunocompetent mice because it lacked the secondary hairpin structure characteristic of IR sequences, or if deleting the IR destroyed enough homology within the DHS that genetic recombination was no longer possible. To address this, it will be interesting to generate a mutant with the DNA homology in the DHS destroyed while leaving the IR intact, and another with only half of the IR sequence disrupted to obliterate secondary structure formation.
What role do IR sequences, such as the 31 bp DHS-resident IR, serve in recombination? The secondary stem-loop structures that IR's form in DNA are known to induce genetic instability (reviewed in Wang & Vasquez, 2006;Zhao, Bacolla, Wang, & Vasquez, 2010). Indeed, secondary DNA structures are associated with recombination in a number of prokaryotic and eukaryotic organisms. The exact mechanisms vary, but these secondary structures may stall replication forks, leading to collapse of the fork and double-stranded breaks. Another possibility is that stemloop structures attract DNA repair proteins that recognize the secondary structure as "damaged," and induce double-stranded breaks (Zhao et al., 2010). While the exact mechanism underlying the non-reciprocal gene conversion event that results in antigenic variation in B. hermsii remains unknown, the results presented herein verify the importance of the DHSresident IR for efficient vmp Ex recombination.
In addition to confirming the requirement of the DHS-resident IR for antigenic switching, the data provide evidence for the importance of the UHS in Vmp expression and suggest that the homology in the UHS may be an essential feature for antigenic variation. Using the targeted deletion/in cis complementation technique, ongoing studies will investigate other features of the UHS and DHS elements, such as the role for the conserved 4-mer and 6-mer palindromes as identified by Dai et al. (2006). Nevertheless, the results from this study have revealed further insight for the mechanism of antigenic variation at the molecular level, and have provided new tools for the genetic study of B. hermsii.

ACKNOWLEDGMENTS
We thank Tim Casselli and Yvonne Tourand for providing their technical expertise on the methodology described herein. This work was supported by a NIH/NIAID grant (R21 AI101230; T.B.). The authors declare no conflicts of interest.