Borrelia burgdorferi, a diderm motile spirochete bacterium, is the causative agent of Lyme disease in the U.S. Each year greater than 25,000 confirmed cases of Lyme disease are reported to the United States Centers for Disease Control and Prevention with about 96% of those cases reported from only 14 states in the Midwest and East (1). Lyme disease is also a significant health problem in parts of Europe and Asia where it is more commonly caused by Borrelia afzelii and Borrelia garinii than B. burgdorferi.

Borrelia spp. are maintained in nature in a tick–mammal life cycle. B. burgdorferi is carried by several species of the Ixodes genus of tick and is transmitted to mammals through tick saliva (2). The spirochetes are maintained in the tick midgut as the tick progresses through its life stages, but the bacteria are not passed transovarially to its offspring (2). The primary mammalian reservoir for B. burgdorferi is the white-footed mouse, Peromyscus leucopus (2). This reservoir is not known to be physically affected by the infection (3). Additional small animals and birds can also serve as reservoirs, whereas large animals and humans can be accidental hosts for tick feeding and subsequent infection.

Human infection with Borrelia spp. often results in a number of generic symptoms including headache, fatigue, and general malaise, and for this reason, the infection is often misdiagnosed or goes untreated. A large percentage of individuals infected with B. burgdorferi will display a rash, termed erythema migrans, at the site of a tick bite (4). An untreated infection with B. burgdorferi can result in late stage symptoms including arthritis, carditis, and neurologic issues (5–7). The CDC reported from 2001 to 2010 that 31% of confirmed Lyme disease cases presented with Lyme arthritis, 14% with neurologic symptoms, and 1% with cardiac involvement (1). The results of a late stage Borrelia infection vary depending on the infecting species, with B. garinii most often associated with neurologic symptoms and B. afzelii infection commonly associated with a skin rash called acrodermatitis chronica atrophicans (8–10).

Borrelia spp. are able to exist in the tick–mammal life cycle due to their ability to adapt to the environment in which they reside. In in vitro studies, Borrelia spp. are able to respond to changes in pH and temperature of the environment, as well as cell density of spirochetes, to differentially regulate the production of many of their outer surface proteins (Figure 1) (11–14).

One way in which B. burgdorferi is able to respond to changes in these environmental conditions is through the RpoN–RpoS signaling system (14, 16). RpoS, RpoN, Rrp2, and BosR are considered the master regulators of virulence gene expression in B. burgdorferi (17–23). RpoS and Rrp2 have been shown to be required for mouse infectivity (18, 24). One example of such control is the reciprocal expression of outer surface protein A (ospA) and outer surface protein C (ospC) tightly regulated by RpoN, RpoS, and Rrp2 (19–21, 25, 26).

Borrelia burgdorferi produces OspA on its surface while in the unfed tick (11). Upon the uptake of blood into the midgut of the tick, ospA expression is maintained until transmission into the mammal when expression is decreased and ospC expression is increased in conjunction with many other genes that encode outer surface proteins, to aid in survival within the mammal (15, 27, 28). Interestingly, OspC production is not necessary and is, in fact, detrimental to survival of the bacteria, likely due to the high immunogenicity of the OspC protein (29). In addition to regulation of surface protein production by RpoN, RpoS, Rrp2, or BosR, B. burgdorferi also utilizes other mechanisms to rapidly change the epitopes available on the surface inside the mammalian host, but not within the tick (30). For example, B. burgdorferi encodes a variable membrane protein-like sequence (Vls) antigenic variation system that enables evasion of recognition by the host-adaptive immune system by continual recombination of silent vls gene segments encoding different vlsE sequences into the expression site (31–34).

In order to colonize tissue sites distal to the site of tick bite in the mammal, it is thought that Borrelia travel within the vascular system of the host. Specific tissue colonization by the spirochetes is thought to occur by targeted exit from the vasculature dictated by the binding specificity of the outer surface proteins of Borrelia spp. This hypothesis is plausible, as characteristics of vascular beds differ depending on the tissue they are associated with as well as the size and type of vessel (116, 117). Tissue bed-specific endothelial cell surface receptors have been documented including VCAM1 in liver, CD36 in lung and heart, L-selectin (CD62L/SELL) in the spleen, and CD133 in the skin, brain, eye, and testicular microvasculature (116). Vascular beds can also vary in their production of proteoglycans, GAGs, and ECM components [as reviewed in Ref. (118)]. The ability of B. burgdorferi to bind to GAGs and ECM components has been specifically associated with effects on bacterial virulence and tissue colonization for Borrelia outer surface proteins such as DbpA and BBK32 (36, 39, 42–44, 51, 71, 77, 113, 119).

It is possible that B. burgdorferi are able to interact with differentially available endothelial cell surface proteins and utilize them to preferentially bind different tissues. Endothelial cell binding by B. burgdorferi has been observed to occur in vitro through interaction with cell surface proteoglycans such as dermatan sulfate, fibronectin, and heparin (120, 121). The binding of B. burgdorferi to the vasculature has been observed, in vivo, to occur in three stages (35, 104). Using intravital microscopy techniques on mouse flank skin, Norman et al. observed both tethering and dragging interactions to occur during the first stage of B. burgdorferi interactions with the vasculature (35). Using a flow chamber, tethering and dragging interactions with the endothelium were confirmed to occur both at the cell surface and at endothelial cell junctions (121). Stationary interactions, which occur after tethering and dragging, may be controlled by B. burgdorferi integrin-binding proteins such as P66 and BBB07. After the spirochete is tightly associated with the endothelial cell junction, it is able to transmigrate into the tissue where it is presumed to colonize and replicate.

By inoculating mice with a library of filamentous phage-expressing B. burgdorferi N40 D10/E9 gene fragments, a number of outer surface proteins of B. burgdorferi that have a tropism for particular mouse tissues were identified (122). Several of the identified proteins in these experiments have been shown to bind to a small number of host proteins, as described in Table 1. This redundancy in protein function suggests a high importance for tissue tropic binding by the bacteria during a mammalian infection. The tissue targeting effects of the different outer surface proteins of Borrelia may be due to the presence of nutrients required for survival of the bacteria at those particular sites. The cumulative data suggest an association between in vitro ligand affinity and the ability to bind to particular tissues in vivo. For example, the GAG-binding proteins BBK32 and DbpA from B. burgdorferi have been shown to be tropic for joint tissue in mouse models of Borrelia infection (40, 114). Detailed work with BBK32 has shown the joint targeting effects of the protein to be mediated specifically by the GAG-binding domain (40). This association between ligand affinity and tissue targeting does not hold true with Fn-binding adhesins. Fn-binding proteins, BB0347 and BBK32, have been shown to preferentially bind joint tissue (40, 62), whereas the Fn-binding protein, RevA, was found to have a tropism for heart tissue in vivo (62). These data highlight the inherent difficulties in using in vitro binding data between two proteins to infer in vivo function of a protein. It is becoming increasingly clear that the outer surface proteome of Borrelia spp. during mammalian infection functions collaboratively to direct the bacteria to specific tissue sites. Further experiments using in vivo models will need to be performed to determine functions for each outer surface protein of Lyme disease Borrelia.

Interestingly, tissue tropism of Borrelia spp. appears not only to be determined by different bacterial proteins but also by allelic differences in a gene encoding a given protein taken from different Borrelia species. For example, dbpA gene sequences are variable between strains as well as species of Borrelia, and this variation not only affects ligand binding, as mentioned earlier, but also tissue tropism (114). It was shown that DbpA from B. burgdorferi strains B31-A3 and N40 D10/E9 is tropic for joint tissue adhesion and colonization, while DbpA from B. afzelii and B. garinii is not (114). DbpA from B. afzelii strain VS461 showed a tropism for skin, while DbpA from B. garinii strain PBr was uniquely tropic for heart tissue colonization among the dbpA alleles tested (75). These differences in tissue tropism of allelic gene variants from different strains and species of Borrelia could explain the differences observed in the symptoms of a late stage infection with Borrelia spp. DbpA may contribute to these symptoms as B. burgdorferi infection often results in arthritis symptoms, while late stage B. afzelii infection commonly presents with skin lesions.

Rapid dissemination of B. burgdorferi during an infection is thought to be facilitated by the spirochetes traveling in the bloodstream of the mammalian host. During this dissemination process, the bacteria are exposed to the innate immune system of the host, designed to detect and clear invading pathogens. B. burgdorferi is also exposed to these blood products inside the midgut of a feeding tick during a blood meal. Resistance to killing by innate immune mechanisms in the blood and host tissues is, therefore, essential for maintenance of the enzoonotic lifestyle of B. burgdorferi.

One of the major innate immune components in the host blood and tissues is the complement cascade. The complement cascade is a series of proteolytic cleavage events in which inactive precursors are converted into active enzymes in the host serum and tissues. These proteolytic cleavage events are activated by three distinct pathogen-recognition mechanisms. The first is termed the “classical” complement cascade, mediated by antibodies that recognize the surface of the pathogen and recruit complement component C1q molecules, which then activate the cascade of enzymatic cleavage events. Mannose-binding lectin molecules present in the serum can bind to sugar moieties on the surface of an invading pathogen, activating the complement cascade by the “lectin” pathway. A third pathway known as the “alternative” pathway activates the complement cascade by random deposition of complement component C3 molecules onto the surface of the pathogen, activating the cascade. All three of the complement pathways converge on the activation of complement component C3, ultimately resulting in formation of the membrane attack complex pore in the pathogen membrane, and pathogen lysis.

The host has adapted many different regulatory mechanisms to control the activation of this pathway to minimize harm to host tissues. One such mechanism is the C1 inhibitor protein (C1-INH), which acts on the C1s and C1r molecules of the first step of the classical pathway (123). Additionally, the host produces C4-binding protein (C4BP), which inhibits the classical and lectin pathways by binding to C4b and inducing its proteolytic cleavage and subsequent inactivation [reviewed in Ref. (124)]. The host also has a number of different mechanisms that it employs for regulating the alternative complement cascade. The host produces Factor H, Factor H-like protein 1 (FHL-1), and complement Factor H-related (CFHR) proteins, which all inhibit the alternative complement cascade by acting on C3b. Factor H can act to sterically hinder the interactions of C3b with Factor B, compete with Factor Bb for binding to C3b, and induce its proteolytic degradation and inactivation into iC3b [reviewed in Ref. (125)]. Regulation of the complement cascade can occur at the end stages of the cascade by employing factors such as clusterin (ApoJ), vitronectin, and CD59. Vitronectin is known to inhibit the binding of complement factor complex C5–C7 to the pathogen membrane, as well as inhibit C9 oligomerization (126). CD59 functions in a similar fashion to vitronectin, intercalating into the membrane attack complex, and inhibiting its polymerization. Clusterin is a complement regulatory glycoprotein associated with apolipoprotein AI, a protein component of high-density lipoprotein (HDL) cholesterol molecules (125). Clusterin inhibits insertion of the membrane attack complex into the pathogen membrane by forming attack complex aggregates away from the pathogen surface (125).

Borrelia spp. produce several proteins on their surfaces, which are proposed to allow them to evade clearing by the complement cascade. Many of the proteins act by recruiting complement regulatory factors to the surface of B. burgdorferi, such as the surface proteins CspA (86, 92–95), CspZ (82, 90, 91), ErpP (94), ErpA (81, 98), and ErpC (127–129) as determined in vitro. Full length CspA and ErpP were found to be required for binding to purified human Factor H and FHL-1 in vitro using a series of C-terminal truncation proteins in a solid phase binding assay (92). The binding of Factor H and FHL-1 by CspA on the surface of B. burgdorferi was confirmed by far western blot and immunofluorescence assays (95) and contributes to cleavage and inactivation of complement component C3b (93). CspA has also been shown, in vitro, to interact with complement components C7 and C9, at a distinct location from the site of Factor H binding on CspA (92, 93, 95). Binding of CspA to C7 and C9 inhibits assembly of the membrane attack complex at the spirochete surface when incubated in active human serum as determined by immunofluorescence microscopy, contributing to bacterial resistance to lysis by human serum proteins in vitro (92, 93, 95). Interestingly, cspA was found to be expressed only in an unfed tick and not during mammalian infection, suggesting a necessary role for complement resistance of Borrelia within the tick (130, 131). Similar to CspA, another outer membrane protein of B. burgdorferi, CD59-like protein, also binds complement component C9 as well as C8β and inhibits the insertion of the membrane attack complex in vitro (132).

Borrelia burgdorferi also produces several OspE-related protein family members including ErpA, ErpP, and ErpC, which have been shown to recruit Factor H and CFHR proteins to the bacterial surface in vitro (81–83). By solid phase-binding assay and immunoblot, ErpA, ErpP, and ErpC were found to bind to CFHR-1, unlike CspA and CspZ, which showed no binding to CFHR-1 in active human serum in vitro (82). Additionally, recombinant ErpP and ErpA were found to bind full length recombinant Factor H with a low dissociation constant, suggesting a potential role for this interaction in vivo, though this has not yet been established (81, 83).

CspZ, a gene encoding another Factor H and FHL-1-binding protein on the surface of B. burgdorferi, has a reciprocal expression pattern to cspA (90). CspZ was found to be expressed primarily during mammalian infection but was not found to be required for establishment of mammalian infection (91). Consistent with this result, CspA and CspZ on the surface of B. burgdorferi were found to interact with human Factor H and FHL-1 in vitro, although production of CspZ was not found to be required for survival in active human serum in vitro (82, 91).

Recently, BBK32 was found to interact with a component of the classical complement cascade in vitro (80). BBK32 was shown to bind C1r, a member of the first protein complex in the classical complement cascade, whereby inhibiting its zymogen activity and effectively blocking attack complex formation in vitro (80). A C-terminal portion of BBK32 (residues 206–354) outside of the Fn- and GAG-binding domains, was found to be necessary for the interaction with C1 in vitro (80). We would predict based on the in vitro binding capabilities of BBK32 to C1, that BBK32 cooperates in vivo with additional Borrelia surface proteins to ensure successful inhibition of multiple branches of the complement cascade.

Experiments have been performed to discern the role of complement in the clearance of Borrelia during mammalian infection using C5-deficient mice (133). Complement component C5 is a point of convergence of all three pathways of the complement cascade. The activation of C5 ultimately results in the formation of the membrane attack complex pore in the pathogen membrane. B. burgdorferi recovery by culture from tissues of several infected strains of mice naturally deficient in C5, including A/J, AKR/J, B10.D2/oSnJ, DBA/2J, and SWR/J, was not found to be different than recovery from infected C5 sufficient C3H/HeJ mouse tissues at 2, 4, and 12 weeks p.i. (133). The authors concluded that complement activity is not required for clearance of B. burgdorferi during a mouse infection. Similar results were seen in experimentally infected C3-deficient mice where complement sensitive B. garinii was not recovered from complement-deficient mice (134). Additionally, it was shown that recruitment of Factor H to the surface of B. burgdorferi in vivo is not necessary for serum resistance of the bacteria as evidenced by similarities in bacterial burdens of WT and Factor H-deficient mice (135). Given what we now know about the ability of B. burgdorferi to interact with several host complement regulatory factors upstream of C3 and C5 in vitro, one would not predict to see a difference in WT bacteria survival in these experiments.

The evasion of the alternative complement cascade is not unique to B. burgdorferi and has been observed in Borrelia hermsii, a causative agent of relapsing fever (136, 137), Borrelia bavariensis (138), as well as other bacterial pathogens such as Streptococcus pyogenes (139), Bordetella pertussis (140, 141), Neisseria gonorrhoeae (142–145), Escherichia coli (146–148), Leptospira interrogans (149), Yersinia pseudotuberculosis (150, 151), Salmonella enterica serovar Typhimurium (152), and Moraxella catarrhalis (153). Many of these organisms obtain resistance to host complement by coating themselves in C4-binding protein (C4BP) and Factor H, similar to B. burgdorferi [reviewed in Ref. (124)].

Borrelia spp. are highly adept at preventing clearance by the innate immune system of the host while adhering to and colonizing host tissues. The success of this mammalian pathogen can, so far, be attributed to the presences of a few bacterial proteins produced on the pathogen surface though there are many more adhesins whose in vivo functions have yet to be determined. For many years, researchers in the Borrelia field have identified and studied outer surface proteins of Borrelia spp. that may contribute to the efficiency of Borrelia as a pathogen. The host proteins that are specifically bound by different outer surface proteins are beginning to be revealed. Borrelia outer surface proteins BBK32, DbpA, RevA, CspA and Erps A, C, and P, have all been shown to bind to multiple macromolecules from the host cell surface and ECM. In addition to binding to multiple host cell surface proteins, Borrelia proteins BBK32, CspA, and Erps A, C, and P have all been found to bind to host complement components. The multifunctional and redundant capabilities of many of the proteins on the surface of Borrelia provide the bacteria with a higher probability of successfully infecting the arthropod and mammalian hosts. As research of Borrelia outer surface proteins progresses, it becomes increasingly clear that redundancy in ECM and host protein-binding specificity acts to guarantee that the bacteria will successfully colonize host tissues while evading detection by the host immune system.

As suggested in this review, there is a strong correlation between ligand specificity of different bacterial outer surface proteins and the tissue tropism of those proteins. When examining the literature it becomes apparent that the binding affinity for host macromolecules such as GAGs by BBK32 and DbpA may be the necessary interactions required for joint tropism and colonization (40, 114). This is further evidenced by the documented species to species differences in GAG-binding capacity and joint colonization seen with the production of DbpA from B. afzelli and B. garinii, Borrelia species, which do not commonly cause arthritis in humans (114).

As discussed in this review, complement protein resistance is achieved by Borrelia spp. through the action of a few outer surface proteins with multiple levels of redundancy. One aspect of redundancy is the ability of multiple Borrelia proteins to bind to and recruit the same complement inhibitory protein. This functional redundancy ensures the successful inhibition of the complement cascade at different stages of the pathway by multiple Borrelia proteins. Redundancy in protein function is a common theme for both the tissue binding and complement protein recruitment aspects of Borrelia pathogenesis, and likely contribute largely to the pathogenic success of this bacterial genus.

JAC and JC both contributed to the writing of this manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.