Lyme disease, caused by tick-borne Borrelia burgdorferi, is a multi-systemic and multi-phasic disease in humans, which includes pauciarticular arthritis in up to 60% of untreated patients [1, 2]. In the absence of antibiotic treatment, arthritis and other lesions undergo resolution with variable bouts of recurrence over the course of months to years of persistent infection [3]. Laboratory mice develop arthritis and carditis that follow a similar multi-phasic course as humans, with resolution and periodic bouts of recurrence over the course of persistent infection [4]. The mouse model has implicated the humoral immune response as a critical factor in arthritis and carditis resolution. Infection of T-cell deficient (Tcr α/βnull, Tcr γ/δ-null), but not B-cell deficient (Igh6-null) or severe combined immunodeficient (SCID) or Rag1-null mice follows a course of resolution that is similar to fully immunocompetent mice [5], and passive transfer of serum from actively infected immunocompetent mice that have undergone disease resolution (immune serum) into infected SCID mice results in complete resolution of arthritis and carditis, but not clearance of infection [6–8].

Identification of the B. burgdorferi antigens targeted by antibodies that mediate disease resolution is complicated by the fact that B. burgdorferi grown in culture medium does not reflect the antigenic profile of spirochetes during mammalian infection [9, 10]. As a means to identify vulnerable antigenic targets that are expressed in the mammalian host that are responsible for antibody-mediated disease resolution, immune serum from actively infected mice has been used to probe B. burgdorferi genomic expression libraries or outer membrane extracts. These efforts revealed arthritis-related protein (BBF01/Arp) as well as decorin binding protein A (DbpA), among other antigens expressed during infection [8, 11–13]. Antiserum generated in mice hyperimmunized with non-lipidated recombinant Arp or DbpA induced arthritis and carditis resolution, but did not eliminate infection, when passively transferred to actively infected SCID mice [8, 12]. Immunization with DbpA was found to induce protective immunity against cultured spirochetes [11, 14], but not tick-borne spirochetes [15], whereas Arp immunization was ineffective at eliciting protective immunity against cultured spirochetes [16]. Outer surface protein C (OspC), another immunogenic protein expressed during infection, has also been shown to be vulnerable to passively transferred OspC antibody in SCID mice, but is down-regulated in response to specific antibody, thereby avoiding immune clearance in immunocompetent mice [17, 18].

Based upon the presence of a leader sequence, arp encodes a putative lipoprotein that is expressed in ticks and throughout the course of mammalian infection and has been shown to stimulate the production of antibody in both experimentally infected mice and naturally infected humans [13, 16, 19–21]. Genomic comparison among several B. burgdorferi sensu stricto (s.s.) strains reveals highly conserved BBF01/arp sequences (95-100% identity from GenBank Blast). Curiously, the genomes of other B. burgdorferi sensu lato strains that are available in GenBank, such as B. afzelii and B. garinii, do not appear to have an arp homolog. In contrast to arp conservation in B. burgdorferi s.s. strains, dbpA and ospC, which also encode immunogenic antigens that are expressed during infection [19, 21–23], have considerable variation (81-85% identity) among the same B. burgdorferi s.s. strains (GenBank). As noted, both Arp and DbpA stimulate an arthritis-resolving immune response [8], and DbpA and OspC elicit protective immune responses against challenge [11, 14, 24]. It is therefore curious that Arp has such a conserved sequence among B. burgdorferi s.s. strains, when it is so obviously subjected to immune selection pressure. The present study explored the biological behavior of B. burgdorferi devoid of, or complemented with, Arp. Arp was found to be non-essential for infectivity, but it influenced infectious dose, spirochete burdens in tissues, arthritis severity, and tick infection kinetics, underscoring its biological significance.

Seven B. burgdorferi B31-arp deletion mutants (Δarp) were created, and found to grow equally well in BSKII medium as B31 (wild-type) spirochetes. The 7 Δarp mutants were initially tested for infectivity in infant ICR mice, which serve as an inexpensive system for titrating infectivity [5]. All seven mutants were determined to be flagellin B (flaB) DNA-positive and arp DNA-negative by polymerase chain reaction (PCR), following growth selection in streptomycin. Four 2-day-old mice were inoculated with 10⁶ of each Δarp mutant or wild-type spirochetes, and sub-inoculation site and urinary bladder were cultured to determine infectivity and ability to disseminate at 7 and 21 days after inoculation. All were infectious, and all disseminated to the urinary bladder. Spirochetes cultured from the inoculation site and urinary bladder were tested by PCR for presence of flaB and arp. Urinary bladder isolates from mice that were flaB-positive and arp-negative were selected for further analysis and confirmed to be arp-null. Upon subsequent inoculation of infant ICR mice with wild-type or each of the seven Δarp mutants, arthritis was of equivalent severity as mice infected with B31 among all groups of mice, indicating that B. burgdorferi devoid of arp were not only infectious, but also equally pathogenic as wild-type B. burgdorferi in susceptible infant mice. One arp isolate (Δarp3) was selected for further analysis.

Four C3H-scid mice were each inoculated with 10⁶ wild-type and five C3H-scid mice were each inoculated with 10⁶ Δarp3 spirochetes, and then necropsied at 60 days of infection to compare the full range of pathogenicity of each inoculum, unencumbered by acquired immunity. All inoculation sites and urinary bladders were culture-positive in both groups. Spirochetes were isolated from blood of 4/4 wild-type inoculated mice, whereas only 2/4 (one sample not collected) Δarp3 inoculated mice were bacteremic. All mice in both groups had severe (mean arthritis score 3.0 ± 0 SD) arthritis in tibiotarsal joints, as well as arthritis in both knees, and all mice had carditis. Despite equally severe disease, spirochete burdens in sub-inoculation, heart base, and tibiotarsal tissues, based upon flaB quantitative PCR (Q-PCR), were significantly lower (P ≤ 0.05) in Δarp3 infected C3H-scid mice compared to wild-type infected mice (Figure 1). Spirochete burdens were also lower in ventricular muscle and quadriceps muscle, but differences were not statistically significant.

A confirmatory experiment was performed in which 5 C3H-scid mice were each inoculated with 10⁶ wild-type and 5 C3H-scid mice were each inoculated with 10⁶ Δarp3 spirochetes, and necropsied on day 28 after inoculation. Inoculation sites and urinary bladders in all mice from both groups were culture-positive, and all mice in both groups were bacteremic. Arthritis severity scores were equivalent in both groups (mean 2.8 ± 0.4 SD wild-type vs. mean 2.4 ± 0.5 SD Δarp3). Significantly lower flaB Q-PCR spirochete burdens (P ≤ 0.05) were found in tissues of Δarp3 infected C3H-scid mice compared to wild-type infected C3H-scid mice, including inoculation site (320 ± 486 SD vs 249,660 ± 187,097 SD), heart base (983 ± 1,353 SD vs 514,400 ± 171,404 SD), ventricular muscle (9,656 ± 5,911 SD vs 432,260 ± 374,173 SD), and tibiotarsus (201 ± 187 SD vs 163,520 ± 89,127 SD). Spirochete burdens were also reduced in quadriceps muscle (3,730 ± 1,412 SD vs 58,640 ± 74,839 SD), but differences were not statistically significant (P = 0.07).

Next, groups of 5 immunocompetent C3H mice were inoculated with 10⁵ wild-type or Δarp3 spirochetes, and then necropsied on days 14, 28 and 42. Tissues were examined for arthritis and carditis, and flaB Q-PCR was performed on sub-inoculation site, heart base, ventricular muscle, tibiotarsus and quadriceps muscle. Inoculation sites of all mice were culture-positive at each interval tested, but none of the urinary bladders of mice inoculated with Δarp3 were culture-positive at 14 days, suggesting delayed dissemination, or reduced sensitivity due to lower tissue burdens (Table 2). Compared to inoculation site, urinary bladders were less consistently culture-positive in both groups of mice, underscoring the greater accuracy of PCR for assessing dissemination and tissue burdens (Table 3). At 14 days, 1/5 wild-type inoculated mice had 1+ inflammation of the tibiotarsus and 5/5 had carditis, whereas none of the Δarp3 inoculated mice had inflammatory lesions in joints or heart at this interval. At 28 days, 5/5 wild-type inoculated mice had both arthritis (1.6 ± 0.5 SD severity) and carditis, whereas only 1/5 Δarp3 inoculated mice had carditis and none had arthritis. At 42 days, 3/3 wild-type inoculated mice continued to have arthritis (1.5 ± 0.5 SD) and carditis, and 1/5 Δarp3 mice had arthritis (1+ severity) and 2/5 had carditis. PCR-positive tissue samples at all intervals indicated that wild-type infected mice had higher spirochete burdens in tissues compared to Δarp3 infected mice (Figure 2). At day 14, most tissues from wild-type inoculated mice were PCR-positive, whereas very few tissues from Δarp3 inoculated mice were PCR-positive (Table 3). The rate of PCR-positive tissues increased in the Δarp3 inoculated mice on days 28 and 42 to rates similar to wild-type infected mice, but flaB DNA copy numbers were consistently lower.

	Culture			Inflammation
Day	Inoculum	Inoc. site	Urinary bladder	Tibiotarsus	Knee	Heart
14	wild-type	5/5*	3/5	1/5	0/5	5/5
	Δarp3	5/5	0/5	0/5	0/5	0/5
28	wild-type	5/5	4/5	5/5	4/5	5/5
	Δarp3	5/5	2/4**	0/5	0/5	1/5
42	wild-type	3/3	2/3	3/3	0/3	1/3
	Δarp3	4/4**	5/5	1/5	0/5	2/5

* number positive/number tested.

** one culture sample contaminated.

Day	Inoculum	SubIN	Hrt base	Vent M	Quad M	Tibiotarsus
14	wild-type	4/5*	5/5	5/5	4/5	4/5
	Δarp3	2/5	1/5	0/5	0/5	0/5
28	wild-type	5/5	ND**	5/5	5/5	4/5
	Δarp3	4/5	ND	3/5	3/5	5/5
42	wild-type	3/3	3/3	3/3	3/3	3/3
	Δarp3	3/5	5/5	4/5	3/5	2/5

* number positive/number tested.

** ND = not done.

A confirmatory experiment was performed in which groups of 4 C3H mice were inoculated with 10⁶ wild-type or Δarp3 spirochetes, and then necropsied on day 28 to verify the difference in tissue spirochete burdens in heart base, ventricular muscle, quadriceps muscle, and tibiotarsal tissue. Tissues were not collected for histopathology. In wild-type infected mice, 4/4 inoculation sites and 3/4 urinary bladders were culture-positive, and 3/3 inoculation sites (one sample contaminated) and 0/4 urinary bladders were culture-positive in Δarp3 infected mice. Spirochete burdens were significantly lower (P ≤ 0.05) in tissues of Δarp3 infected mice compared to wild-type infected mice, including sub-inoculation site (139 ± 266 SD vs. 1,761 ± 1,682 SD), heart base (45 ± 54 SD vs. 2,333 ± 1,400 SD), ventricular muscle (28 ± 26 SD vs 448 ± 276 SD), and quadriceps muscle (15 ± 23 SD vs 367 + 291 SD). Spirochete burdens were also lower in tibiotarsus tissue of Δarp3 infected mice (13 ± 11 SD vs 16,171 ± 29,765 SD), but differences were not statistically different (P = 0.16). Based upon these observations, it was determined that both C3H-scid mice as well as C3H mice infected with Δarp3 had lower spirochete burdens in tissues.

Sera from C3H mice that were confirmed to be culture-positive at 60 days of infection with wild-type or Δarp3 spirochetes were determined to be appropriately sero-reactive against recombinant Arp antigen (Arp seropositive or seronegative, respectively). Serum antibody titers from Δarp3 infected mice were equivalent to antibody titers in mice infected with wild-type infected mice when tested against B. burgdorferi lysate antigen (≥1:24,300), and antibody titers to recombinant Arp antigen were verified to be either negative (Δarp3) or positive (Δarp3 + lp28-1G), with titers equivalent to Arp titers in wild-type immune sera (1:2,700).

Larval ticks were fed upon the before-mentioned wild-type or Δarp3 infected C3H mice 3 days before necropsy at day 42. Replete ticks were allowed to molt and harden into nymphs, and then tested by Q-PCR for flaB and arp DNA. Among ticks that fed upon wild-type infected mice, 30/30 were PCR positive for both flaB and arp, with 53,950 mean ± 84,668 SD flaB copy numbers per tick. In contrast, 40/50 ticks that fed upon Δarp3 infected mice were flaB-positive, with significantly lower spirochete loads per positive tick (1,384 mean ± 1,780 SD flaB copy numbers, P = 0.0002). Tick cohorts from individual Δarp3 infected mice contained 9/10, 5/10, 10/10, 6/10 and 10/10 positive ticks. Results demonstrated that Δarp3 can be acquired by ticks from infected C3H mice, but ticks that acquired Δarp3 harbored fewer organisms compared to wild-type.

The ability of Δarp3 spirochetes to be transmitted from infected ticks to naïve C3H mice was next evaluated by placing 10 nymphal ticks from the wild-type and Δarp3 positive tick cohorts (above) onto each recipient mouse. Mice were necropsied at 3 weeks following tick feeding, and ear, heart base, ventricular muscle, tibiotarsus and quadriceps muscle were tested by flaB Q-PCR. Among 5 mice fed upon by ticks carrying wild-type spirochetes, 4/5 mice became infected, and all tissue sites from the 4 positive mice were PCR-positive, with high copy numbers of flaB DNA in tissues (Figure 3). In contrast, 2 of the 7 mice that were fed upon by Δarp3 infected ticks were positive, but only a single tissue in each of the positive mice contained low copy numbers of flaB DNA. Results indicated that Δarp3 spirochetes are capable of tick-borne transmission. Since ticks infected with Δarp3 spirochetes had significantly fewer spirochete loads compared to ticks infected with wild-type spirochetes, it could not be concluded that there was less efficient transmission.

This study examined the effect of targeted deletion of BBF01/arp on infectivity of B. burgdorferi B31. The median infectious dose of B. burgdorferi B31 with an arp null mutation was elevated approximately ten-fold compared to wild-type spirochetes, and restored by complementation. Therefore, it is apparent that BBF01/arp is not essential for infectivity of the mammalian host. This is supported by indirect results of others, who demonstrated diminished infectivity in B. burgdorferi spirochetes lacking linear plasmid 28–1 (lp28-1), which encodes only two unique and functional genes, vlsE and arp[25–29]. Furthermore, clones of B. burgdorferi B31 with a deletion of the left side of lp28-1, which contains arp, remained infectious and capable of persistence, similar to wild-type spirochetes [25].

Examination of the pathogenicity of various B. burgdorferi B31 clones lacking lp28-1 has shown that clones lacking lp28-1 were infectious in BALB/c-scid mice and reached similar tissue burdens as wild-type spirochetes, but were incapable of inducing arthritis [29]. Results of the current study indicate that infection of C3H-scid mice with arp null spirochetes resulted in lower spirochete burdens in tissues with development of equally severe arthritis as that induced by wild-type spirochetes. Thus, it appears that arp null spirochetes are equally (if not more) arthritogenic than wild-type B. burgdorferi in C3H-scid mice. The lack of effect on tissue burdens and arthritis in BALB/c-scid mice infected with B. burgdorferi devoid of the entire lp28-1 plasmid, but reduced burdens in infections with arp null spirochetes observed in the current study are likely due to the experimental variations in B. burgdorferi strains (B31-5A11 vs. B31-A3), mouse strains (BALB/c-scid vs. C3H-scid), or a number of other possible genetic variables.

Lack of lp28-1 has been associated with failure to persist in immunocompetent mice. This has been attributed to vlsE, since clones lacking a region of the plasmid that encodes arp are capable of persistent infection [25, 29]. The current study examined persistence in immunocompetent C3H mice up to 42 days after inoculation, and demonstrated that arp null spirochetes were indeed capable of persistence. In the present study, we also infected mice for antibody evaluation at 60 days of culture-confirmed infection, and thus verified persistence for up to 60 days. As in C3H-scid mice, arp null spirochete burdens were lower in C3H mouse tissues compared to wild-type spirochetes. Notably, arthritis severity was markedly reduced in C3H mice infected with arp null spirochetes. Since arp null spirochetes are fully arthritogenic in SCID mice, these results suggest that the lower pathogenicity of arp null spirochetes in immunocompetent mice is a consequence of susceptibility of the arp null mutant to immune response. Other antigens that are expressed during infection have also been shown to be susceptible to arthritis-resolving antibody responses, including DbpA [8], BmpA, and BmpB [30]. In the absence of Arp, these or other antigens may be targets of immune-mediated phenotypic effects noted in the present study.

Although arp null spirochetes are capable of surviving in the murine host, their ability to do so appears to be compromised, since arp null spirochete burdens were 2 logs fewer in tissues of SCID mice compared to wild type spirochetes, and were even lower in immunocompetent mice. Thus, arp null spirochetes appear to be either less fit to grow or are more vulnerable to innate and acquired immune factors compared to wild type spirochetes. This lack of fitness is likely responsible for the additional phenotypic effect of arp deletion that was observed in acquisition and transmission by vector ticks. Larval ticks were fed upon mice infected with wild-type or arp null spirochetes, and allowed to molt into nymphs. Ticks became infected with both types of spirochetes, but following molting, nymphal ticks that were colonized with arp null spirochetes had significantly lower spirochete loads per tick compared to ticks colonized with wild type spirochetes. The lower arp null spirochete loads were likely influenced by the lower spirochete burdens in tissues of mice that were fed upon by the ticks. This effect has been demonstrated by others [31] in which ticks that fed upon MyD88 deficient mice infected with B. burgdorferi had higher spirochete burdens compared to ticks that fed upon wild-type mice. MyD88 deficient mice have significantly higher spirochete tissue burdens compared to wild-type mice. The lower rate of transmission of arp null spirochetes from infected nymphal ticks to naïve mice could also have been influenced lower spirochete burdens in arp null colonized ticks. Further studies are needed to examine dynamics within ticks, but there is normally a significant burst of replication of spirochetes within fed ticks [32] that did not appear to occur in ticks colonized with arp null spirochetes. Nevertheless, results indicated that arp null spirochetes could be acquired and transmitted by vector ticks, albeit at diminished levels.

Deletion of the arp gene resulted in a modest phenotypic effect, including reduced infectious dose, reduced fitness of B. burgdorferi for growth in the mammalian host, and reduced ability for acquisition and transmission by the vector tick. Deletion of a number of B. burgdorferi genes has been found to have only mild phenotypic effects upon infectivity and persistence of B. burgdorferi (reviewed in [33]). This is likely due in large part to compensatory up-regulation of other genes. Although the function of Arp remains unknown, the current study in which arp was deleted with relatively modest phenotypic effects underscores the complexity of B. burgdorferi biology and emphasizes caution in attributing phenotype or lack thereof to the role of a single gene alteration.