Lipoprotein succession in Borrelia burgdorferi: similar but distinct roles for OspC and VlsE at different stages of mammalian infection

Borrelia burgdorferi alternates between ticks and mammals, requiring variable gene expression and protein production to adapt to these diverse niches. These adaptations include shifting among the major outer surface lipoproteins OspA, OspC, and VlsE at different stages of the infectious cycle. We hypothesize that these proteins carry out a basic but essential function, and that OspC and VlsE fulfil this requirement during early and persistent stages of mammalian infection respectively. Previous work by other investigators suggested that several B. burgdorferi lipoproteins, including OspA and VlsE, could substitute for OspC at the initial stage of mouse infection, when OspC is transiently but absolutely required. In this study, we assessed whether vlsE and ospA could restore infectivity to an ospC mutant, and found that neither gene product effectively compensated for the absence of OspC during early infection. In contrast, we determined that OspC production was required by B. burgdorferi throughout SCID mouse infection if the vlsE gene were absent. Together, these results indicate that OspC can substitute for VlsE when antigenic variation is unnecessary, but that these two abundant lipoproteins are optimized for their related but specific roles during early and persistent mammalian infection by B. burgdorferi.


Introduction
Borrelia burgdorferi cycles between mammalian and tick environments, each of which varies over time (see Radolf et al., 2012 for a recent review). When the spirochaetes are deposited in mammalian skin by a feeding tick, they must combat the host innate immune defences while replicating and spreading throughout the body of the mammal. As the mammal mounts an acquired immune response against B. burgdorferi, the spirochaete evades immune surveillance, at least in part, by changing components of its lipoprotein coat (Zhang et al., 1997;Crother et al., 2004;Liang et al., 2004). When a naive tick feeds on an infected mammal, the spirochaetes are acquired by the tick and establish infection in the midgut. As the tick digests the blood meal, the nutritional environment changes for the bacteria.
Borrelia burgdorferi genes important for nutrient acquisition and host adaptation are distributed throughout the complex B. burgdorferi genome, which is composed of a linear chromosome and multiple linear and circular plasmids (Fraser et al., 1997;Casjens et al., 2000;. In addition to essential chromosomal genes, many genes encoding products important for growth in the mouse-tick infectious cycle are located on plasmids (Purser et al., 2003;Byram et al., 2004;Jewett et al., 2007b;. Plasmid genes of unknown function are also important for the natural spirochaete life cycle (Labandeira-Rey et al., 2003;Grimm et al., 2004;Bankhead and Chaconas, 2007;Jewett et al., 2007a), but some entire plasmids are dispensable for bacterial growth in culture (Purser and Norris, 2000;Labandeira-Rey and Skare, 2001) and the loss of others may lead to small or negligible effects on infectivity in mice and ticks (Elias et al., 2002;Dulebohn et al., 2011).
Survival in the diverse mouse and tick environments requires adaptation by altering the gene expression and protein composition of the bacteria (Hodzic et al., 2002;. Among the adaptations observed is a characteristic succession of outer surface lipoproteins (Schwan et al., 1995;Ohnishi et al., 2001;Liang et al., 2002b;Crother et al., 2004), whose genes are located on circular and linear plasmids (cp and lp respectively). Within infected ticks, the spirochaetes produce OspA (outer surface protein A) (Fingerle et al., 1995;Schwan et al., 1995), which contributes to maintaining tick colonization (Pal and Fikrig, 2003;Yang et al., 2004;Battisti et al., 2008) and is encoded on the 54 kb linear plasmid lp54. When nymphal ticks feed, the spirochaetes begin to synthesize OspC and reduce OspA synthesis (Schwan et al., 1995;Ohnishi et al., 2001). The ospC gene is located on the circular plasmid cp26 in the B. burgdorferi genome, and is required for the bacteria to initiate infection of a naive mammal (Grimm et al., 2004;Stewart et al., 2006;Tilly et al., 2006). OspC, however, is immunogenic and targeted by mammalian antibodies (Wilske et al., 1986;1988), so bacteria that persist within an immunocompetent mammal must subsequently downregulate OspC (Liang et al., 2002a;Crother et al., 2004). Concomitant with the development of the host acquired immune response and OspC downregulation is increased synthesis of VlsE by B. burgdorferi (Hodzic et al., 2003;Crother et al., 2004). VlsE is a third lipoprotein, whose amino acid sequence does not resemble either OspC or OspA, but is abundantly present on the spirochaete surface during persistent infection (Crother et al., 2003). The vlsE locus is found on the 3′ end of the linear plasmid lp28-1, and has the key characteristic of undergoing antigenic variation by an error-prone gene conversion-like mechanism that utilizes the 15 silent vlsE cassettes found upstream of the vlsE locus on lp28-1 (Zhang et al., 1997;Zhang and Norris, 1998;Coutte et al., 2009). Because it is antigenically variable, VlsE protein can be present on the spirochaete surface during persistent infection of an immunocompetent host. Spirochaetes acquired by a larval tick feeding on a persistently infected small mammal reduce VlsE production in the tick midgut (Indest et al., 2001;Bykowski et al., 2006) and resume synthesis of OspA (Schwan et al., 1995;Hodzic et al., 2002).
Although this pattern of lipoprotein succession has been described, the functions of these outer surface proteins remain undefined. OspA plays a role in tick colonization by B. burgdorferi , and studies suggest that it is a tick midgut adhesin (Pal et al., 2000) and shields spirochaetes from mammalian antibodies in the incoming blood meal of feeding ticks (Battisti et al., 2008). OspC has been suggested to have roles in host selectivity (Brisson and Dykhuizen, 2004), plasminogen binding (Lagal et al., 2006;Onder et al., 2012), invasion (Lagal et al., 2006;Wormser et al., 2008), dissemination (Seemanapalli et al., 2010), salivary gland migration in the tick , evasion of innate immunity (Xu et al., 2008a), and recognition of the mammalian environment (Earnhart et al., 2010). Conflicting data exist as well for each of these proposed roles, and no well-supported function for OspC has emerged. ospC mutants have the clear phenotype of being defective at the initial phase of mammalian infection following needle inoculation or tick bite (Tilly et al., 2007;Dunham-Ems et al., 2012). An extensive study involving complementation of an ospC mutation with the genes for lipoproteins VlsE, OspA, and DbpA suggested that OspC provides a somewhat nonspecific function, since substituting ospC with the other (unrelated) lipoprotein genes allowed spirochaetes to persist for several weeks after inoculation into SCID mice (Xu et al., 2008a). Further studies in immunocompetent mice suggested that the functions of these lipoproteins, while overlapping, may be optimized to the times at which they are naturally expressed, as constitutive expression of OspA and VlsE reduced infectivity of B. burgdorferi in mice (Xu et al., 2008b).
A simple model to explain these and other results is that these major B. burgdorferi outer surface lipoproteins serve a common function at different stages of the bacterial mouse-tick life cycle. In this model, that function is fulfilled by OspC during initial mammalian infection, by VlsE during persistent infection, and by OspA during the tick stage of the life cycle. The nature of that function may be protection against mammalian and tick innate immune defences. The present study describes experiments designed to further test the model that OspC and VlsE play similar roles at different stages of mammalian infection.

Determining if OspC can substitute for VlsE
Mouse infection for assessing pCpospC retention. The first set of experiments was designed to test whether OspC production must be sustained during infection when VlsE production is not possible. We pursued this question with an experimental design that was based on our previous studies with spirochaetes that contain lp28-1 and, therefore, produce VlsE . In those studies, initial infection with an ospC mutant spirochaete required complementation with the ospC gene on a shuttle vector (which we call pCpospC, to denote that it carries the ospC gene transcribed from the ospC promoter). The complementing plasmid was lost during persistent infection, indicating that ospC expression was not required at later times. In immunodeficient SCID mice, which lack B and T cells, the ospC-containing shuttle vector was more stable, most likely because these mice do not produce antibodies against OspC that could lead to clearance of spirochaetes that retained the plasmid and continued to express ospC. Nevertheless, significant shuttle vector loss was detected, indicating that sustained OspC production was not necessary during persistent infection of SCID mice either. In the present study, we reasoned that if VlsE normally fulfils the role of OspC during persistent infection, spirochaetes lacking VlsE would be unable to tolerate the absence of OspC during persistent infection. A measure of this intolerance would be strong positive selection for shuttle vector maintenance during persistent infection with a B. burgdorferi mutant lacking both the endogenous vlsE and ospC genes, but complemented with a shuttle vector carrying the ospC gene. Consequently, we screened for retention of the ospC-containing shuttle vector at a time at which we had previously found significant loss of the same shuttle vector by ospC mutant spirochaetes that retained lp28-1 and, therefore, vlsE. We predicted that spirochaetes lacking VlsE would retain the ospC-containing shuttle vector, in order to continue producing OspC in the absence of VlsE. Consistent with this hypothesis, a previous study with lp28-1-deficient spirochaetes found that these spirochaetes maintained OspC production during persistent infection of an immunodeficient host (Embers et al., 2008).
Wild-type (WT, C3H) and C3H-SCID mice were inoculated with 10 4 spirochaetes per mouse of a strain lacking lp28-1 (vlsE − ), a double mutant lacking both vlsE and ospC (vlsE − ΔospC), and the same two strains containing the ospC shuttle vector pCpospC (Table 1A). Six weeks post-inoculation, the mice were euthanized and spirochaete isolation from tissues was attempted. As expected, all strains were highly attenuated for persistent infection in WT mice (Table 2). Isolates were obtained from ankle joints from only two WT mice infected with vlsE − /pCpospC and one WT mouse infected with vlsE − ΔospC/pCpospC (Table 2). However, all 10 WT mice inoculated with spirochaetes containing pCpospC were seropositive (Table 2), indicating that they had been at least transiently infected. These data are consistent with previous studies showing that lp28-1 (carrying vlsE) is required for persistent infection of WT mice.
In contrast, spirochaetes lacking vlsE but containing ospC either at its normal location or on a shuttle vector (vlsE − , vlsE − /pCpospC, and vlsE − ΔospC/pCpospC) persisted in all tissues in all SCID mice inoculated, whereas those lacking both ospC and vlsE (vlsE − ΔospC) were not isolated from any SCID mouse tissues (Table 2). These findings were also consistent with previous studies, which demonstrated that lp28-1 was dispensable for persistent infection by B. burgdorferi in SCID mice (Lawrenz et al., 2004;Bankhead and Chaconas, 2007). Surprisingly, in this experiment, spirochaetes lacking just ospC (ΔospC) were isolated from most SCID mice in which they were inoculated. This result differed from previous work and other experiments in this study, in which ospC was found to be an essential gene for normal infection of both immunocompetent and immunodeficient mice (Grimm et al., 2004).
Retention of pC pospC in spirochaetes lacking vlsE during mouse infection. If sustained OspC production is required when VlsE production is not possible, spiro- chaetes lacking both vlsE and ospC but containing the ospC-carrying shuttle vector pCpospC should retain that shuttle vector, even during infection of SCID mice. To ascertain if this were true, isolates obtained from mice infected with bacteria containing pCpospC were plated for single colonies and colonies were screened by PCR for the presence of pCpospC. In SCID mice, we obtained isolates from all tissues of all mice infected with vlsEdeficient or vlsE-and ospC-deficient spirochaetes containing pCpospC, and compared shuttle vector retention in the presence and absence of the native ospC gene in bacteria lacking vlsE. We found significant loss of pCpospC in SCID mice infected with spirochaetes lacking VlsE alone, but complete retention of the shuttle vector when the vlsE − spirochaetes also lacked the endogenous ospC gene ( Fig. 1) (P < 0.0001, as assessed by a twotailed Mann-Whitney test or one-way ANOVA). This result supports our hypothesis that continued OspC production would be required by bacteria lacking VlsE, if these two outer surface proteins fulfil the same essential function during infection of the mammalian host.

Determining if VlsE or another outer-surface protein, OspA, can substitute for OspC
Generating B. burgdorferi with increased vlsE and ospA expression. Although OspA is neither required for nor typically produced during mammalian infection, studies by Xu et al. suggested that an ospC mutant could be partially complemented for mouse infectivity by several major outer-surface lipoproteins, including VlsE and OspA (Xu et al., 2008a,b). To confirm these results, we constructed several plasmids to determine if VlsE or another unrelated major outer-surface protein, OspA, could complement ospC mutant spirochaetes when establishing mammalian infection. Shuttle vectors pC pvlsE1 (ospC promoter-vlsE1) and pCpospA (ospC promoter-ospA; Table 1B) encode vlsE and ospA expression from the ospC promoter. Shuttle vectors pFpvlsE1 (flaB promoter-vlsE1) and pFpospA (flaB promoter-ospA) encode constitutive vlsE and ospA expression from the flaB promoter. Finally, shuttle vector pCpvlsE2 (ospC promoter-vlsE2) has the ospC promoter and signal sequence fused to the VlsE coding sequence (lacking its native signal sequence), to test if lipoprotein processing and membrane localization affects how well VlsE can fulfil the role of OspC (Table 1B). These shuttle vectors were introduced into ospC mutant B. burgdorferi (ΔospC) to determine if OspA or VlsE, appropriately regulated or constitutively expressed, could fulfil the requirement for OspC in establishing mammalian infection. The shuttle vectors were also introduced into WT B. burgdorferi to investigate if inappropriate expression of these lipoproteins in the presence of native ospC expression would be deleterious for the spirochaetes. Finally, we verified that expression of the C pvlsE and CpospA genes on the shuttle vector paralleled that of the endogenous ospC gene by introducing these constructs into the high-passage strain B313 (Table 1A) (Sadziene et al., 1993), which lacks native copies of the ospA and vlsE genes and constitutively produces OspC (Fig 2). We also confirmed the expression a. Both immunocompetent (WT, C3H) and immunodeficient (C3H-SCID) mice were injected with 10 4 spirochaetes of the indicated B. burgdorferi strain. Six weeks post-inoculation, WT mice were retroorbitally bled and all mice were euthanized to assess infection by isolation of spirochaetes from ear, bladder, and ankle joints. Seroconversion of WT mice toward B. burgdorferi proteins is shown in parentheses, and indicates at least transient infection. b. Isolation of these strains from WT mice was only successful from ankle joints. c. ND, not determined, because the infection was not attempted in this experiment.  of OspA and VlsE from the flaB promoter fusion genes on the shuttle vector in B. burgdorferi, using strain B31-A34, which lacks the native vlsE gene (Jewett et al., 2007b) ( Fig. 2). Although these experiments do not guarantee that the promoters on the shuttle vectors will behave identically to the endogenous ospC and flaB promoters when the spirochaetes infect a mammal, production of OspA and VlsE in cultured spirochaetes with these constructs correlated with what we know about expression of the endogenous promoters in the same culture conditions.
OspA and VlsE do not fulfil the role of OspC in establishing mammalian infection in immunocompetent or immunodeficient SCID mice. Since the above results suggest that OspC is able to fulfil the role of VlsE (in the absence of an acquired immune response), we tested whether the converse were true, and VlsE could fulfil the role of OspC in establishing mammalian infection. We first used shuttle vectors carrying the VlsE or OspA open reading frames (ORFs) under the regulation of the ospC promoter. Groups of WT and C3H-SCID mice were intradermally injected with an inoculum of 10 4 spirochaetes of WT or ΔospC clones harbouring these shuttle vectors (Tables 1A and B). Immunocompetent mice were bled and all were euthanized three weeks later, and tissues were cultured for isolation of spirochaetes (Table 3). All WT mice that were positive by serology were also positive for B. burgdorferi isolation from all harvested tissues. Neither VlsE nor OspA was able to complement the ΔospC mutant (Table 3). Overproduction of VlsE and OspA was inadequate for even transient infection, since uninfected animals were also seronegative. Furthermore, we found that inappropriate expression of vlsE (i.e. from an addi-tional copy on the shuttle vector and under control of the ospC promoter) attenuated WT B. burgdorferi at a dose of 10 4 in both WT and SCID mice (Table 3). This finding was similar to that of Xu et al., in which increased expression of vlsE led to clearance of the spirochaetes in immunocompetent mice (Xu et al., 2008b). To confirm that the reduced infectivity of spirochaetes carrying the ospC promoter-vlsE fusion on a shuttle vector was caused by inappropriate vlsE expression, we displaced the pCpvlsE1 shuttle vector with the incompatible plasmid pKFSS1 (Frank et al., 2003). The resultant WT/pKFSS1 strain exhibited WT infectivity (Table 3), indicating that the attenuated infectivity of WT/pCpvlsE1 was due to the ospC promoter-driven expression of vlsE from the shuttle vector.
Although neither VlsE nor OspA was capable of fulfilling the role of OspC when under control of the ospC promoter, the previous study of Xu et al. found that constitutive expression of these proteins in an ospC mutant background could restore infectivity to varying degrees in both WT and SCID mice (Xu et al., 2008a). Since our previous experiments utilized constructs with expression from the ospC promoter, we attempted to replicate the Xu et al. studies by introducing shuttle vector constructs with vlsE and ospA under control of the constitutive flaB promoter into both WT and ΔospC B. burgdorferi (Table 4). VlsE production did not complement the ΔospC mutant in either WT or SCID mice (Table 4), and uninfected animals were seronegative also, indicating that they had not been even transiently infected. OspA production was not able to complement the ΔospC mutant in WT mice (0/5 mice infected and seronegative), as was observed by Xu et al. (2008a,b), but we did find a significant difference between the ΔospC and the ΔospC/pF pospA strains in SCID mice (3/10 vs 5/5 mice infected respectively) (Table 4). However, in a separate experiment (Table 2), surprisingly, we found infection with the ΔospC strain in 4 out of 5 SCID mice at a dose of 10 4 , so infection by ΔospC/ pFpospA may not be a consequence of OspA production. Constitutive expression of vlsE did not attenuate WT B. burgdorferi, as we had seen with ospC promoter-driven expression of vlsE (Table 3). This may be the result of a lower expression level of VlsE from the pFpvlsE1 construct, since the ospC promoter is highly induced during early infection as compared with the flaB promoter (Tokarz et al., 2004).
Assaying the effect of replacing the endogenous ospC gene on cp26 with the vlsE gene expressed from the ospC promoter. In the experiments described above, complementation of an ospC mutant was attempted with genes located on shuttle vectors, which generally have slightly higher copy number than the endogenous genomic plasmids of B. burgdorferi . To   Fig. 2. Immunoblot demonstrating VlsE and OspA production by spirochaetes carrying the shuttle vector constructs and the ΔospC::CpvlsE strain. Antisera to VlsE, OspA and FlaB were used to probe B. burgdorferi lysates containing approximately 10 7 spirochaetes. The pCpvlsE1, pCpospA, pCpvlsE2, pFpvlsE1, and pFpospA shuttle vectors were introduced into B. burgdorferi strains B313, which lacks native copies of vlsE and ospA, or B31-A34, which lacks a native copy of vlsE, to confirm production of these lipoproteins from their respective shuttle vector constructs. Wild type B31-A3 (WT), and ΔospC::CpvlsE, which expresses vlsE from the ospC locus on cp26, were grown in pH 6.8 BSKII medium to late log phase to induce vlsE production from the ospC promoter.
address the possibility that the lack of complementation (or attenuation of infection by WT B. burgdorferi) was a result of excess protein production due to higher copy number, we constructed a strain in which the ospC gene located on cp26 was replaced with the vlsE coding sequence, with the fusion between the ospC promoter at the start codon of the vlsE ORF (see Experimental procedures). Appropriate expression of the vlsE gene from the ospC promoter was assessed by growing spirochaetes in vitro at pH 6.8, in which conditions the WT ospC gene is typically induced. Although not entirely conclusive, a lysate of ΔospC::C pvlsE (which has the normal vlsE gene, in addition to the ospCpvlsE fusion located on cp26) grown in these conditions appeared to contain more VlsE protein than was found in WT spirochaetes, as expected [ΔospC::CpvlsE (pH 6.8), Fig. 2]. The lysate of WT bacteria grown in parallel contained more OspC than a lysate of uninduced WT B. burgdorferi (data not shown), confirming that the ospC promoter was induced in the conditions used. When tested for the ability to infect WT or SCID mice at an inoculum of ∼ 10 3 per mouse, the ΔospC::CpvlsE bacteria were non-infectious in both cases (Table 5). Not only were we unable to isolate the bacteria from any tissue, but the serology of the WT animals was negative (data not shown), indicating that the mice were not even transiently infected. In contrast, all WT and SCID mice were infected and colonized by WT B. burgdorferi. Again, we found that VlsE was unable to substitute for the absence of OspC, even when production of VlsE was directed by the ospC promoter at the normal locus on cp26 and, therefore, most likely to have been appropriate.

Discussion
Although OspC has been shown to fulfil a critical role early in mammalian infection by B. burgdorferi, its function has remained undefined. This paper presents data addressing the model in which the function that OspC provides initially is required throughout infection by B. burgdorferi, and VlsE subsequently fulfils that function during persistent infection. We provide strong evidence that OspC production can compensate for VlsE deficiency in an immunodeficient host. We demonstrate that the ospC Table 3. Infectivity of B. burgdorferi expressing vlsE and ospA from the ospC promoter.

B. burgdorferi strain
No. of infected mice/ no. of mice injected a WT SCID WT 6/8 9/10 WT/pCpvlsE1 2/10 0/5 WT/pCpvlsE2 1/5 ND c WT/pCpospA 4/5 5/5 WT/pKFSS1 b 5/5 ND ΔospC 0/10 0/10 ΔospC/pCpvlsE1 0/5 0/5 ΔospC/pCpvlsE2 0/5 ND ΔospC/pCpospA 0/5 0/5 a. Both immunocompetent WT mice and immunodeficient C3H-SCID mice were injected intradermally with 10 4 of the designated B. burgdorferi strain. At three weeks post-injection, WT mice were retroorbitally bled to test for infection by immunoblot for seroconversion to B. burgdorferi proteins, and all mice were euthanized for attempted isolation of spirochaetes from the ear, bladder, and tibiotarsal joint. All WT mice positive by serology were also positive by isolation from all three tissues. b. The pBSV2G-ospCp-vlsE shuttle vector (pCpvlsE1) was displaced from the WT/pCpvlsE1 strain by introduction of the incompatible pKFSS1 shuttle vector. This strain was then injected into wild-type mice at a dose of 10 4 to confirm that the loss of the pBSV2G-ospCp-vlsE shuttle vector restored infectivity to wild-type levels. c. No data. These strains were only tested in WT mice.

B. burgdorferi strain
No. of infected mice/ no. of mice injected a WT SCID WT 9/10 10/10 WT/pFpvlsE1 4/5 3/5 WT/pFpospA 2/5 5/5 ΔospC 0/10 3/10 ΔospC/pFpvlsE1 0/5 0/5 ΔospC/pFpospA 0/5 5/5 b a. Both immunocompetent WT mice and immunodeficient C3H-SCID mice were injected intradermally with 10 4 organisms of the designated B. burgdorferi strain. At three weeks post-injection, WT mice were retro-orbitally bled to test by immunoblot for seroconversion to B. burgdorferi proteins, and all mice were euthanized for attempted isolation of spirochaetes from the ear, bladder, and tibiotarsal joint. All seropositive WT mice were also positive for isolation from all three tissues. b. The number of SCID mice infected by the ΔospC/pFpospA strain was significantly different from that of ΔospC (P = 0.026, as determined by Fisher's two-tailed exact probability test). a. Both immunocompetent (IRW) and immunodeficient C3H-SCID mice were injected with ∼ 10 3 organisms of the indicated B. burgdorferi strain. At three weeks post-inoculation, WT mice were retroorbitally bled and tested for infection by immunoblot for seroconversion to B. burgdorferi proteins. All mice were euthanized and isolation of spirochaetes attempted from the ear, bladder, and ankle joint.
gene must be retained, and presumably expressed, when vlsE is missing. These data imply that OspC can carry out the function of VlsE during persistent infection, provided it is not targeted by acquired immunity. However, although OspC is able to substitute for VlsE in an immunodeficient host, we found that VlsE could not substitute for OspC in any host background. Therefore, our data indicate that the two proteins are not strictly interchangeable, so the simplest model of redundant function with reciprocal expression is not correct.
Although this study does not support the idea that VlsE can also substitute for OspC, some data do suggest that other B. burgdorferi proteins can take the place of OspC during initiation of mammalian infection. One of the strongest pieces of evidence is apparently normal infection of mice by an ospC deletion mutant when tissue from a persistently infected animal was transferred to a naive animal . We previously speculated that VlsE, which is made by the hostadapted spirochaetes, allows infection of the naive animal in this scenario. Xu et al. (2008a) partially restored infectivity of an ospC mutant by complementation with VlsE, and also suggested that the proteins have similar functions. However, we were not able to duplicate their findings (A. Bestor, unpubl. results), even when using the same vlsE allele as Xu et al. We also found that ospA was unable to fully complement an ospC mutation. Although we found apparent complementation of the ospC mutation by ospA driven by the flaB promoter in SCID mice, the ability of ΔospC spirochaetes to infect SCID mice on occasion (see, e.g. Table 2 and below) means that this result may have been spurious. Clearly, there are differences between our system and that of Xu et al., who did find complementation of an ospC mutation by overexpression of the vlsE and ospA genes (Xu et al., 2008a). For example, the typical ID50 of in vitro-grown WT B. burgdorferi is almost 100-fold higher in our experiments than in theirs. Also, the ospC mutant used by Xu et al. (2008a) is an insertion, rather than a deletion of the entire coding sequence, and it lacks lp25 but carries the essential gene pncA on the ospC-complementing plasmid, which requires that the shuttle vector be retained during mammalian infection. Nevertheless, in our experiments, VlsE and OspA did not fulfil the role of OspC for initiation of infection.
It remains unclear why OspC can substitute for VlsE but not vice versa. OspC may perform a unique role early in infection, in addition to a common function that VlsE typically assumes as the spirochaetes persist in a mammal. Without invoking an additional unknown role, VlsE could be unable to take the place of OspC because the lipoprotein requirement during initial infection by B. burgdorferi is more stringent, since few bacteria are present and they are all located in the tick bite (or needle inoculation) site.
Because it appears that OspC and VlsE are not fully interchangeable, we propose that these two proteins require both appropriate context and appropriate timing to be fully functional. By context, we encompass the roles of other proteins produced at the same time and also variations in membrane composition and arrangement, which are influenced by lipid availability and temperature. Although the OspC and VlsE sequences are not related, the tertiary structures of the OspC dimer and VlsE monomer are similar in size and shape, with variable regions forming surface-exposed loops (Eicken et al., 2001;Kumaran et al., 2001), which may allow them to carry out a common function. As to the question of what their common function could be, OspC and VlsE might serve to protect the bacterium from the particular host environment in which it finds itself, or they might stabilize the bacterial structure, again in the particular host context. If OspC and VlsE protect B. burgdorferi against host defences, which defences those might be remains undefined. We have looked without success for differences between ospC mutant and WT spirochaetes with respect to phagocytosis by mouse and human neutrophils, phagocytosis by mouse macrophages, susceptibility to human or mouse complement, or susceptibility to mouse natural killer cells (K. Tilly, A. Porter, F. DeLeo, C. Checroun, and P. Rosa, unpubl. data). Although these studies were not exhaustive, they emphasized the complexity of the interaction between host and bacterium and the inherent limitations of investigating that interaction either with isolated components or in vivo. If OspC and VlsE stabilize the bacteria in the host environment, large amounts of a particular lipoprotein may be an essential component of the recently described B. burgdorferi lipid rafts (LaRocca et al., 2010). Given that the B. burgdorferi membrane structure and composition vary depending on the host environment, it would not be surprising if an essential lipoprotein component would also need to vary.
Although we did not succeed in heterologous complementation of the ospC deletion mutation, we did occasionally observe infection of SCID mice by ΔospC spirochaetes. Furthermore, in recent experiments we have sometimes obtained infection in both WT and SCID mice with ΔospC spirochaetes when inoculating at a dose ≥ 10 7 spirochaetes, confirmed by isolation of the mutant from all cultured tissues (A. Bestor, K. Tilly and P. Rosa, unpubl. results). This is in contrast to the complete lack of infectivity by ospC mutant spirochaetes that we found in earlier studies (Grimm et al., 2004;Stewart et al., 2006;Tilly et al., 2006). ospC mutant spirochaetes grown directly from stocks frozen since the strain was originally isolated exhibit a similar phenotype (K. Tilly, unpubl. results), so we do not think that the ΔospC mutant used in this study has undergone a compensatory mutation. Since the earlier tissue transfer studies suggest that a B. burgdorferi protein pro-duced during persistent infection, when appropriately regulated and in the correct context, can successfully fulfil the ospC requirement (Tilly et al., 2009), our culture conditions may have shifted subtly to increase production of that product and its appropriate context by ΔospC spirochaetes. We have tested several batches of medium, and growth to different densities, but have not pinpointed a tangible variable that affects infectivity of the ospC mutant. Despite occasional infection by the ospC mutant at high dose, we have been unable to demonstrate consistently increased infectivity at a standard inoculum when the mutant was complemented with either vlsE or ospA.
This and other studies are delineating a succession of lipoproteins that coat B. burgdorferi and provide essential functions in the mouse and tick hosts at various stages of the bacterial life cycle. OspC is typically on the bacterial surface when spirochaetes initiate mammalian infection, and then subsequently downregulated. VlsE is produced and undergoes antigenic variation during persistent infection. OspA is required during bacterial colonization of ticks. Each has unique expression and protein characteristics, yet their roles may be related. The bacteria are exposed to different degrees of immune surveillance throughout the infectious cycle, from dermal inoculation of the mammal, to blood stream dissemination and peripheral tissue colonization, and acquisition with the blood of a seropositive host by the feeding tick. The spirochaetes also will have different protein and lipid composition at these stages, which could affect bacterial stability and survival in the host environment. The changing spirochaete surface, of which the major lipoproteins are essential components, could be considered to be analogous to the more complex developmental changes found in vector-borne parasites during their mammalian and arthropod phases. Further defining the shared or specific roles of major outer surface lipoproteins will help elucidate the details of the B. burgdorferi infection process and, in so doing, reveal potentially widespread adaptations to the host and vector environments.
To make strains for testing whether OspC could fulfil the role of VlsE, the ospC gene in vlsE − , which lacks lp28-1 and the encoded vlsE locus (Table 1A), was inactivated by allelic exchange using plasmid pJK109 . This plasmid includes ospC flanking sequences surrounding flgBpkan, which replaces the ospC gene. We chose a transformant (vlsE − ΔospC) that had the same plasmid content as vlsE − , but lacked the ospC ORF. vlsE − and vlsE − ΔospC were transformed by electroporating with pBSV2GospC (referred to as pCpospC), a shuttle vector carrying the ospC gene and its own promoter (Table 1B) that had been methylated , and selecting for gentamicin-resistance. Transformant clones with the same plasmid content as vlsE − and containing pCpospC were used for subsequent experiments.

Construction of shuttle vectors carrying lipoprotein genes fused to various promoters
The pCpvlsE1 shuttle vector, expressing vlsE under control of the ospC promoter, was constructed as follows. The ospC promoter was amplified from ospC7 (Grimm et al., 2004) genomic (g) DNA with Vent polymerase (New England Biolabs) using primers 1 and 2 (Table 6) and ligated into the KpnI and XbaI sites in the multiple cloning site (MCS) of pBSV2G (Elias et al., 2003), yielding pBSV2G-ospCp. The region of lp28-1 encompassing the vlsE ORF was amplified with Taq polymerase, using primers 3 and 4 (Table 6), from WT gDNA that had been treated with Mung Bean nuclease (New England Biolabs) to nick hairpin ends, and ligated into BspHI-digested pBSV2G-ospCp, yielding pBSV2G-ospCp-vlsE (pCpvlsE1).
The pCpvlsE2 shuttle vector, expressing and transporting VlsE under the control of the ospC promoter and carrying the ospC signal sequence, was constructed as follows. The region of cp26 from nucleotides 16647-16971, encompassing the ospC promoter and ospC signal sequence, was amplified from WT gDNA with Vent polymerase using primers 1 and 5 (Table 6), and ligated into the KpnI and XbaI sites found in the MCS of pBSV2G, yielding pBSV2G-ospCp(+ss). The vlsE locus was amplified without the predicted signal sequence from pCpvlsE1 with Vent polymerase using primers 6 and 7 (Table 6), and ligated into XbaI/HindIII-digested pBSV2G-ospCp(+ss), yielding pBSV2G-ospCp(+ss)-vlsE (pCpvlsE2).
All constructs were sequenced upon completion to confirm they were as designed and free of any mutations before transforming them into the WT and the ΔospC strains (see Table 1), and selecting with gentamicin. B. burgdorferi plasmid contents of the transformants were confirmed to be the same as the parental strains by PCR (Akins et al., 1998).

Plasmid and strain construction to replace the ospC ORF on cp26 with the vlsE ORF
To make a plasmid for replacing the ospC ORF with the vlsE ORF at the ospC locus on cp26, we began with pJK109, a plasmid previously used to inactivate ospC by allelic exchange . The ospC upstream fragment was modified at the 3′ end by addition of a BspHI site that included the ATG start codon of OspC, and was generated by PCR using primers 2 (Table 6) and JK170 . We amplified the vlsE gene from the pCpvlsE1 plasmid using primers 3 and 13 ( Table 6). Both of these fragments were cloned into PCR2.1Topo. The amplified fragments were digested from the vector using SacI-BspHI and BspHI-NotI, respectively, and ligated (in a three-way ligation) with SacI-NotI digested JK109. A plasmid with the correct fusion was named pOVK.
The ospC promoter-vlsE ORF fusion was moved into B. burgdorferi by transforming WT bacteria with approximately 5 μg of methylated pOVK . Three colonies were obtained, of which one had the appropriate allelic replacement at the ospC locus on cp26. When the plasmid content of this strain was checked, it was found to have lost lp25. Accordingly, the clone was transformed with 10 μg of total genomic DNA from A3 lp25-Sm, which has a flgBp-aadA fusion (Frank et al., 2003) inserted into lp25 at the site used in A3 lp25-Gm (Grimm et al., 2005). A strep R colony in which lp25 had been restored was picked and used for subsequent experiments.

Mouse infections
All animal experiments were performed using protocols approved by the Animal Care and Use Committee of the Rocky Mountain Laboratories and according to the guidelines of the National Institutes of Health. Rocky Mountain Laboratories is accredited by the International Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). C3H/HeN and C3H/HeN SCID mice were purchased from Harlan Sprague-Dawley (Indianapolis, IN). RML mice are outbred, derived from Swiss-Webster mice, and bred at the Rocky Mountain Laboratories. IRW mice are inbred, derived from RML mice, and also bred at the Rocky Mountain Laboratories. For infection studies, mice were injected with 80% of the inoculum intraperitoneally and 20% of the inoculum subcutaneously. In some cases (as indicated in the Results), bacteria were inoculated into the intradermal-subcutaneous compartment, with the appropriate number of bacteria provided in a single 100 μl injection with a 27 ga needle. The mice were euthanized three to six weeks post-inoculation, and ears, bladders, and ankle joints harvested and cultured in 10 ml BSKII medium. Isolates from mice infected with vlsE − /pCpospC or vlsE − ΔospC/pCpospC were plated in solid medium and resultant colonies were screened by PCR with primers 14 and 15 (Table 6), to assess retention of pCpospC (Tilly et al., 2009).