Of ticks, mice and men: understanding the dual-host lifestyle of Lyme disease spirochaetes

Key Points

• Lyme disease first came to public attention in the 1970s as a result of an epidemic of oligoarthritis in children and adults living in the vicinity of the town of Lyme, Connecticut, USA. The observation that infected individuals had a skin lesion called erythema migrans was crucial to identifying the causative agent (the spirochaete Borrelia burgdorferi) and its arthropod vector (Ixodes scapularis).Lyme disease is now known to be worldwide in distribution and is a major public health problem in the United States.

• B. burgdorferi is perpetuated in an enzootic cycle in which uninfected larval ticks acquire the spirochaete by feeding on an infected reservoir host, usually a small mammal (for example, the white-footed mouse, Peromyscus leucopus). The infected larvae moult to become nymphs, which then transmit the bacterium to an uninfected animal with the next blood meal. The outcome of infection is variable and dependent on the mammalian host. Humans are an incidental, dead-end host.

• B. burgdorferi has an unusual genome consisting of a 1 Mb linear chromosome and numerous linear and circular plasmids. The plasmids are a primary repository for differentially expressed lipoproteins. The bacterium is an auxotroph for all amino acids, nucleotides and fatty acids; it also lacks genes encoding enzymes for the tricarboxylic acid cycle and oxidative phosphorylation.

• B. burgdorferi resembles a Gram-negative bacterium in that it contains both outer and inner membranes. However, the architecture and composition of its outer membrane differ markedly from those of Gram-negative bacteria. Most notably, decorating the surface of B. burgdorferi are differentially expressed outer-surface lipoproteins. As with all spirochaetes, the organelles of motility (the flagella) are located in the periplasmic space.

• With the onset of the nymphal blood meal, spirochaetes in flat nymphal ticks undergo extensive changes in gene and protein expression that enable transmission of B. burgdorferi to the mammalian host. Many of these transcriptional changes are regulated by the response regulatory protein 2 (Rrp2)–RpoN–RpoS and histidine kinase Hk1–Rrp1 pathways.

• Following transmission, borrelial virulence determinants (for example, OspC) act in concert with tick salivary components (SALPs) to enable the bacterium to establish a foothold at the bite site and subsequently disseminate.

• The presence of spirochaetes locally and in tissues following dissemination triggers innate immune pathogen-sensing mechanisms (for example, Toll-like receptors), which recruit circulating leukocytes and orchestrate the development of the adaptive response. Macrophages are thought to be crucial innate immune effectors for bacterial clearance.

• Clearance of organisms is dependent on the appearance of specific antibodies. Spirochaetes counter the humoral response of the host by downregulating target surface antigens and activating the recombinatorial surface lipoprotein system (vls) for antigenic variation.

Abstract

In little more than 30 years, Lyme disease, which is caused by the spirochaete Borrelia burgdorferi, has risen from relative obscurity to become a global public health problem and a prototype of an emerging infection. During this period, there has been an extraordinary accumulation of knowledge on the phylogenetic diversity, molecular biology, genetics and host interactions of B. burgdorferi. In this Review, we integrate this large body of information into a cohesive picture of the molecular and cellular events that transpire as Lyme disease spirochaetes transit between their arthropod and vertebrate hosts during the enzootic cycle.

Main

Since its original description, Lyme disease (Lyme borreliosis) has risen from relative obscurity to become a prototypical emerging infectious disease¹. The path to notoriety began with a little noticed epidemic of oligoarthritis in the mid 1970s, mainly in children, in several rural communities clustered about the town of Lyme in southeastern Connecticut, USA². Physicians misdiagnosed many of these children as having juvenile rheumatoid arthritis, leading two astute mothers to seek the assistance of investigators at nearby Yale University (New Haven, Connecticut)¹. The observation that approximately one-quarter of patients developed an expanding, annular skin lesion before the onset of arthritis proved the key to solving the medical mystery. The pathognomonic 'bull's eye' rash, which is now called erythema migrans (Fig. 1), had been closely associated with the bite of the sheep tick (castor bean tick) Ixodes ricinus in Northern Europe and was widely believed to be caused by an unknown infectious agent, possibly a spirochaete¹. These early field investigations prompted an intensive hunt that led to the isolation of the causative organism from a species of North American deer tick, Ixodes scapularis³. Shortly thereafter, a novel spirochaete was isolated from skin, blood and cerebrospinal fluid specimens obtained from patients with erythema migrans^4,5. DNA–DNA hybridization studies revealed the isolate to be a new species within the genus Borrelia⁶, and the species was subsequently named Borrelia burgdorferi after its co-discoverer, Willy Burgdorfer.

Figure 1: The enzootic cycle of Borrelia burgdorferi.

Ixodes spp. ticks undergo a three-stage life cycle — larva, nymph and adult — with one blood meal per stage. Although some Borrelia spp. that cause relapsing fever can be passed from adult to egg (transovarial transmission), this does not occur with Borrelia burgdorferi sensu lato, so each generation of tick must acquire a B. burgdorferi infection anew. Larval ticks feed on many different animals, including Peromyscus spp. mice, squirrels and birds. B. burgdorferi infection is acquired by feeding on an infected reservoir animal, and the bacterium is retained during the subsequent stages (that is, trans-stadially) after each blood meal and moult^7,10. Nymphs feed on a similar range of hosts to larvae; transmission of spirochaetes to a competent reservoir host by a feeding nymph perpetuates the enzootic cycle for the next generation of larval ticks. Although small mammals are usually thought of as the primary reservoirs for Lyme disease spirochaetes, studies have called attention to the importance of migratory birds as disseminators of spirochaetes over large distances^7,10. Adult ticks are not generally important for maintenance of B. burgdorferi in the wild, as they feed predominantly on larger animals such as deer, which are incompetent hosts for B. burgdorferi⁷. However, deer are important for maintenance of the tick population because adult ticks mate on them. Although all three stages of Ixodes spp. can feed on humans, nymphs are responsible for the vast majority of spirochaete transmission to humans. It is unknown whether infected humans can transmit spirochaetes to feeding larvae, and humans are generally considered dead-end hosts. Dogs are probably incidental hosts and not part of the enzootic cycle.

Lyme disease spirochaetes are widely distributed throughout the temperate zones of the Northern Hemisphere, and their range continues to expand as lands that were denuded for agriculture become reforested, creating new habitats for deer, ticks and infection-susceptible vertebrates⁷. Lyme disease accounts for >90% of all vector-borne disease in the United States, with nearly 30,000 confirmed cases reported in 2008 (Ref. 8). As many as 60,000 cases occur each year within the range of I. ricinus, the primary vector in Europe⁹.

The expanding universe of Borrelia burgdorferi

Borrelia spp. fall into two major phyletic groups, one containing the causative agents of Lyme disease and the other containing the spirochaetes responsible for relapsing fever⁷. Phylogenetic analyses have led to the division of Lyme disease spirochaetes into numerous species, collectively referred to as B. burgdorferi sensu lato (s.l.). Among the more than 20 named and unnamed species in the B. burgdorferi s.l. complex, three genospecies predominate as human pathogens: B. burgdorferi sensu stricto (s.s.) in the United States and Western Europe, and Borrelia garinii and Borrelia afzelii in Eurasia⁷. Four other species, Borrelia valaisiana, Borrelia lusitaniae, Borrelia spielmanii and Borrelia bissettii, have also been isolated or detected by PCR in specimens from small numbers of patients⁷. Hereafter, B. burgdorferi is used to refer to B. burgdorferi s.l.

The enzootic cycle

The life cycle of Lyme disease spirochaetes is depicted in Fig. 1. B. burgdorferi is transmitted principally by four species of hard tick within the I. ricinus complex: I. scapularis and Ixodes pacificus (the western black-legged tick) in eastern and western North America, respectively, I. ricinus in Europe and Ixodes persulcatus (the tiaga tick) in Asia^7,10. The tick larvae are uninfected when they hatch, as there is no transovarial transmission, and B. burgdorferi is acquired after feeding on an infected reservoir host. After moulting to the nymphal stage, the ticks transmit the pathogen to the animal that provides its next blood meal. Larvae of hard ticks other than Ixodes spp. inefficiently acquire Lyme disease spirochaetes from a blood meal and/or fail to maintain them through the trans-stadial moult¹¹. The Ixodes spp. that transmit B. burgdorferi to humans tend to feed on diverse species of vertebrates, accounting for the geographical breadth of transmission cycles within the Northern Hemisphere. Their generalist feeding behaviour is also responsible for the incidental infection of humans^7,10. Field studies in North America and Eurasia have identified a variety of small-mammal and avian reservoirs in enzootic transmission cycles⁷. The white-footed mouse, Peromyscus leucopus, is considered to be the main reservoir in the northeastern United States, whereas rodents and migratory birds are the principal reservoirs in Europe for B. afzelii and B. garinii, respectively.

Outcomes of infection

Spirochaetes are deposited into the bite wound along with the tick saliva during tick feeding¹². Infection rarely, if ever, occurs during the first 24 hours of nymphal feeding but becomes increasingly likely after the tick has been attached for 48 hours or longer¹³. The outcome of the inoculating event depends on the recipient. White-footed mice remain persistently infected without evidence of inflammation¹⁴. Although all laboratory strains of inbred mice are susceptible to infection, the development of organ system pathology (that is, arthritis and carditis) in immunocompetent mice is highly strain dependent^14,15 (Box 1).

The B. burgdorferi genome does not encode any known toxins or the machinery that would be required to secrete them¹⁶; tissue damage, and hence disease, is believed to be mediated by the inflammatory response elicited in the mammalian host^15,17. Although the natural history of Lyme disease in humans is variable and poorly defined¹⁷, seroprevalence surveys in endemic areas have revealed that asymptomatic or subclinical infections occur frequently¹⁸. Erythema migrans, the most common clinical manifestation of borrelial infection, develops following an incubation period of 3–32 days¹⁷. Low-level spirochaetaemia probably occurs in the majority of untreated patients¹⁹, occasionally affecting the peripheral or central nervous system, joints or heart¹⁷. In Europe, most dermatological manifestations are attributed to B. afzelii, and neuroborreliosis is caused mostly by B. garinii²⁰. The relative infrequency of B. burgdorferi s.s. as a cause of disease in Europe is believed to explain why carditis and arthritis are less common sequelae of infection here than in North America²⁰. Differences in infectivity have also been noted between isolates of B. burgdorferi s.s.^21,22.

The three decades since the discovery of B. burgdorferi have witnessed an impressive accumulation of knowledge on the molecular biology of this pathogen, its interactions with its arthropod vector, its virulence determinants, the immune responses it elicits during mammalian infection and the mechanisms involved in immune evasion. In the remainder of this Review, we integrate this knowledge into a cohesive picture of the molecular and cellular events that sustain the B. burgdorferi enzootic cycle.

Genomics and cellular architecture

Genomic complexity coupled with metabolic parsimony. Members of the genus Borrelia possess the most complex genomes of all known bacteria^16,23. More than 20 distinct, naturally occurring genetic elements have been identified in the B31 type strain of B. burgdorferi. These include the ∼1 Mb linear chromosome and an assortment of linear and circular plasmids totalling approximately 600 kb^16,23. The chromosome carries the majority of housekeeping genes and is fairly constant in organization and content across the genus²³. The plasmids, which encode most of the differentially expressed outer-surface lipoproteins, exhibit much greater variability in gene content, are not equally represented in all strains and, in laboratory models, are not all essential for maintenance of the enzootic cycle²³. One of the most curious features of the borrelial genome is its degree of redundancy. This is reflected not only in the extraordinarily large number of paralogous gene families distributed mainly among its plasmid complement, but also in the cp32 plasmid family, members of which contain large stretches of essentially identical DNA interrupted by sequence-variable lipoprotein genes²³. Table 1 summarizes the key genes that are currently believed to contribute to maintenance of the B. burgdorferi enzootic cycle.

Table 1 Key genes thought to contribute to maintenance of the Borrelia burgdorferi sensu lato enzootic cycle

The metabolic abilities of B. burgdorferi are extremely limited as a result of reductive evolution following the adoption of a lifestyle that involves parasitizing nutrients from its hosts^16,24. The bacterium is an auxotroph for all amino acids, nucleotides and fatty acids. It also lacks genes encoding enzymes for the tricarboxylic acid cycle and oxidative phosphorylation, deriving energy instead from the fermentation of sugars to lactic acid via the Embden–Meyerhof pathway^24,25. One of the metabolic oddities of B. burgdorferi is that it does not seem to require iron. Spirochaetes grown in vitro have no measurable iron content²⁶ and, with the possible exception of BB0690, a Dps-like bacterioferritin orthologue²⁷ (formerly known as NapA), the genome does not encode orthologues for any known iron-requiring metalloproteins, such as cytochromes, catalase or superoxide dismutase^16,26.

Cell envelope. The structure, composition and physical properties of the cell envelope of B. burgdorferi diverge markedly from those of prototypical proteobacteria such as Escherichia coli²⁸ (Figs 2,3). The outer membrane is a fluid and fragile bilayer that is devoid of lipopolysaccharide^16,29. Its major lipid constituents are phosphatidylcholine, phosphatidylglycerol and the highly immunogenic but non-inflammatory glycolipids cholesteryl-6-O-acyl-β-D-galactopyranoside and cholesteryl-β-D-galactopyranoside³⁰, none of which is found in the typical proteobacterial outer membrane. The cholesterol glycolipids form raft-like microdomains that are required for the complement-independent bactericidal activity of antibodies directed against lipoproteins on the outer surface of the spirochaete²⁹. The borrelial outer membrane also contains a low density of proteins with membrane-spanning domains²⁸. Some of these proteins possess channel-forming properties, although they lack sequence similarity to the better studied bacterial porins²⁸. BesC, a TolC orthologue encoded by BB_0142, forms part of an efflux system that is required for virulence and antimicrobial resistance³¹.

Figure 2: Cellular architecture of Borrelia burgdorferi.

a | Freeze-fracture electron micrographs showing the convex and concave leaflets of the inner and outer membranes (IM and OM, respectively). Integral membrane proteins (particles) are considerably more abundant in the IM; the density of OM particles is much lower than that in a prototypical Gram-negative bacterium. Scale bars represent 500 nm. b | Cryoelectron tomograms of the ends of borrelial cells, showing IM, OM, peptidoglycan (PG), flagellar motor and filament assemblies (FMFA), chemoreceptor arrays (CR) and an external layer comprising outer-surface lipoproteins (Osps). Upper scale bar represents 1 μm, lower scale bar represents 100 nm. c | Cryoelectron tomographic cross-section showing the IM, OM, periplasmic space (PS) and periplasmic flagella (PFs). Scale bar represents 50 nm. d | Longitudinal cryoelectron tomographic slice showing a ribbon of nine flagellar filaments wrapping around the IM in a right-handed helix. Scale bar represents 200 nm. Part a is reproduced, with permission, from Ref. 130 © (1994) American Society for Microbiology (ASM). Part b is reproduced, with permission, from Ref. 34 © (2009) ASM, and from Ref. 36 © (2011) ASM. Parts c,d are reproduced, with permission, from Ref. 35 © (2009) ASM.

Figure 3: The borrelial cell envelope.

This schematic of the borrelial cell envelope shows the outer membrane, flagellar filaments, peptidoglycan and inner membrane. The outer membrane contains outer-surface lipoproteins (Osps) in high density and β-barrel outer-membrane-spanning proteins such as BamA in low density. The inner membrane is rich in integral membrane proteins, many of which are transporters. BbCRASP, complement regulator-acquiring surface protein; OppA1, oligopeptide permease A1; PTS, phosphotransferase system.

From the standpoint of pathogenesis, the most notable difference between the outer membranes of B. burgdorferi and proteobacteria is the number and variety of lipoproteins that adorn the spirochaete surface²⁸ (Figs 2,3). Characterization of differentially expressed outer-surface lipoproteins has been a major focus of research, including vaccine development. How Lyme disease spirochaetes direct nascent lipoproteins to their surface is one of the mysteries of borrelial cell biology. B. burgdorferi contains a lipoprotein outer-membrane localization (LOL) system similar to that of proteobacteria for directing newly exported lipoproteins to the outer membrane, although the borrelial LOL system uses entirely different sorting rules to the E. coli prototype²⁸. However, B. burgdorferi lacks any of the known pathways for translocating lipoproteins across the outer membrane to the outer surface²⁸. A β-barrel assembly machinery A (BamA) orthologue³² seems to be the central component of the B. burgdorferi outer-membrane assembly machinery. The outer membranes of BamA-depleted spirochaetes are deficient in outer-surface lipoproteins as well as membrane-spanning β-barrels, suggesting that an as-yet-unidentified integral outer-membrane protein is involved in directing lipoproteins to the spirochaete surface.

Motility. B. burgdorferi moves by posteriorly propagating planar waves³³. A defining morphological feature of B. burgdorferi, as with all spirochaetes, is that the organelles of motility — the flagella — are contained entirely within the periplasmic space³³ (Fig. 2d). The filaments are attached at each cell pole to linearly arranged flagellar motors (7–11 at each end), which are elaborate nanomachines that convert chemiosmotic potential into motive force³⁴. Cryoelectron tomography has revealed that the flagellar filaments form flat ribbons that wind along the cell cylinder on top of the peptidoglycan layer and below the outer membrane³⁵ (Fig. 2d). Parallel and adjacent to the motors are arrays of methyl-accepting chemotaxis proteins³⁶ (Fig. 2b) that bind attractant and repellent molecules through their periplasmic domains, transmitting environmental signals to the motors by modulating the phosphorylation state of the chemotaxis (Che) two-component system, CheA–CheY³³.

Regulation of differential gene expression

Considering the extensive transcriptional changes that the spirochaete undergoes throughout the enzootic cycle, the apparatus for controlling gene expression seems surprisingly sparse. However, it is becoming increasingly apparent that layers of regulatory complexity do in fact exist^37,38. Many genes that encode proteins required for transmission of spirochaetes between vector and vertebrate hosts are transcribed by an alternative RNA polymerase σ-factor, RpoS, which is transcribed by another alternative σ-factor, RpoN (also known as σ⁵⁴ and NtrA)^{37,38,39,40,41} (Fig. 4a). RpoN-dependent transcription of rpoS requires the formation of an open rpoS promoter complex, mediated by phosphorylated response regulatory protein 2 (Rrp2) and presumed metal-dependent DNA binding by a Fur–PerR orthologue, Borrelia oxidative stress regulator (BosR; also known as Fur), upstream of the rpoS promoter^{37,38,42,43,44,45,46,47} (Fig. 4a). The phosphate group for Rrp2 phosphorylation seems to come from acetyl phosphate, the intermediate of the acetate kinase–phosphate acetyltransferase pathway that converts acetate to acetyl-CoA⁴⁸ (Fig. 4a). In addition, rpoS expression is post-transcriptionally regulated by the small RNA DsrA⁴⁹, the RNA chaperone Hfq⁵⁰ and the RNA-binding protein CsrA^51,52.

Figure 4: Regulation of gene expression in Borrelia burgdorferi.

a | The histidine kinase 1 (Hk1)–response regulatory protein 1 (Rrp1) and the alternative RNA polymerase σ-factor RpoS global regulatory systems. Binding of unidentified ligands to the periplasmic sensor domains (D1 and D2) of the hybrid histidine kinase Hk1 initiates a phosphorelay that activates the diguanylyl cyclase activity of Rrp1, resulting in the production of cyclic di-GMP (c-di-GMP)^56,57,58,131. Phosphodiesterase A (PdeA) and PdeB degrade c-di-GMP to 5′-phosphoguanylyl-(3′–5′)-guanosine (pGpG) and GMP, respectively^59,60. Activation of Rrp2 in vitro and in vivo occurs via the high-energy phosphoryl donor acetyl-phosphate rather than by its presumptive cognate histidine kinase, Hk2 (Ref. 48. The function of Hk2 is currently unknown. Phosphorylated Rrp2, Borrelia oxidative stress regulator (BosR) and RpoN initiate transcription of rpoS^{37,38,42,43,44,45,46,47}. This is depicted as a trimeric complex, but the precise interactions between these proteins have yet to be determined. Putative BosR-binding sites (BSs) containing the direct repeat sequence TAAATTAAAT are shown⁴⁷; −24/−12 is the RpoN-binding site in the rpoS promoter⁴⁷. RpoS in turn induces the expression of genes that are required during the mammalian-host phase of the spirochaete life cycle and represses the expression of tick-phase genes. b | Expression of the Hk1–Rrp1 and RpoS global regulatory systems during the B. burgdorferi life cycle^{37,38,56,57,58,68,80}. In the flat nymph, both the Hk1–Rrp1 and the Rrp2–RpoN–RpoS systems are inactive and only tick-phase genes are expressed. The nymphal blood meal activates both the Hk1–Rrp1 and Rrp2–RpoN–RpoS pathways. Expression of mammalian-phase genes begins in concert with downregulation of tick-phase genes. Following inoculation into a mammalian host, the spirochaetes complete the process of adaptation; the Hk1–Rrp1 pathway is inactive, the Rrp2–RpoN–RpoS pathway is active, mammalian-phase genes are expressed and tick-phase genes are repressed. During larval acquisition of spirochaetes, Hk1–Rrp1 is activated, probably at the feeding site, whereas the Rrp2–RpoN–RpoS system is inactivated. Mammalian-phase genes are repressed, expression of tick-phase genes begins and ingested spirochaetes bind to the larval midgut epithelium via OspA and possibly other receptors^65,66,67. GGDEF, a conserved motif present in diguanylyl cyclases¹³¹; Hpt, histidine-containing phosphotransfer domain⁵⁶; HTH, helix–turn–helix domain; N, amino; PAS, putative sensor domain for Hk2; Rec, receiver domain.

Two DNA-binding proteins — BpaB (encoded by multiple alleles per isolate) and EbfC — are involved in regulation of the cp32-encoded ospE, ospF and elp lipoprotein gene families (also collectively known as erp genes)⁵³. EbfC is also involved in regulating the expression of other borrelial genes, including those encoding components of the oligopeptide permease ABC transporter⁵⁴. The DNA-binding protein Hbb has a role in regulating the expression of the gene encoding p66, a porin and adhesin⁵⁵. As a final example, a two-component system composed of histidine kinase 1 (Hk1) and Rrp1 controls production of the second messenger molecule cyclic di-GMP (c-di-GMP) in response to stimuli received during both the larval and nymphal blood meals^56,57,58 (Fig. 4a). The fine-tuning of c-di-GMP levels is mediated by two phosphodiesterase (Pde) proteins, PdeA and PdeB^59,60 (Fig. 4a).

The Borrelia –tick interface

Larval acquisition. Lyme disease spirochaetes disseminate in their reservoir hosts and are then acquired by a naive larva taking a blood meal. Live imaging of disseminated borreliae in the dermis of mice revealed bacteria that appear to be searching randomly for the chemotactic stimuli that are provided, or elicited intracutaneously, by larval feeding⁶¹. Real-time visualization has also caught spirochaetes in the act of migrating directly towards the feeding site and rapidly entering the hypostome, whereas other spirochaetes nearby seem oblivious to the same chemoattractants (L. Bockenstedt, personal communication). Larval acquisition of spirochaetes occurs rapidly. The bacteria can be detected in larval midguts by immunofluorescence within 24 hours of attachment and are plentiful by 48 hours, before significant amounts of blood have been imbibed⁶².

During infection of the mammalian host, the borrelial Rrp2–RpoN–RpoS pathway is fully activated, promoting the expression of genes that are required within the mammalian host while repressing genes that are required within the tick^37,38,41 (Fig. 4b). By the time the tick larvae are fully engorged, this pathway is inactive, or nearly so, resulting in the upregulation of tick-phase genes, such as ospA, which are repressed by RpoS^38,63. B. burgdorferi specifically binds the neuroendocrine stress hormones adrenaline and noradrenaline; it has been proposed that OspA expression is upregulated in response to the production of these hormones in the skin, which is in turn induced by the combined mechanical and pharmacological assault of larval feeding⁶⁴. OspA-deficient spirochaetes cannot bind to tick receptor for OspA (TROSPA) on tick midgut epithelial cells and are eventually expelled from the larval digestive tract with the blood meal waste^65,66,67. In addition, activation of the Hk1–Rrp1 pathway during larval feeding (Fig. 4b), with consequent production of c-di-GMP (Fig. 4a), is required for successful bacterial colonization of the larvae^56,57,58. Hk1-deficient and Rrp1-deficient mutants are destroyed early during feeding^56,57, perhaps because they cannot remodel their surfaces to protect themselves against antimicrobial substances elaborated by the larval midgut.

The ingress of blood triggers a burst of spirochaete replication that continues for weeks through the moult⁶⁸. Following tick drop-off, the spirochaete must turn to alternative carbon sources for glycolysis and phospholipid biosynthesis, as the glucose supply within the tick midgut lumen diminishes. The expression of genes within the glp operon, which encodes proteins involved in the uptake and utilization of glycerol, is upregulated during the tick phase of the spirochaete life cycle^57,69. The induction of glp expression during larval acquisition of spirochaetes is driven by activation of the Hk1–Rrp1 pathway⁵⁷ and by the loss of RpoS-mediated gene repression⁶⁹ (Fig. 4b). Ixodes spp. ticks surround the blood meal with an acellular matrix of glycoproteins and chitin, called the peritrophic membrane⁶⁸. N-acetylglucosamine and chitobiose, secreted by epithelial cells for synthesis and remodelling of the peritrophic membrane, may be alternative carbon sources in addition to being essential for peptidoglycan biosynthesis^25,70.

The flat nymph. Spirochaetes within flat, or unfed, nymphs exist in a poorly understood metabolic state that enables them to endure prolonged periods of nutrient deprivation⁶⁸. In this state, both the Rrp2–RpoN–RpoS and Hk1–Rrp1 pathways are inactive, as are mammalian-phase genes, while tick-phase genes are maximally expressed^37,38,68 (Fig. 4b). Despite appearances, spirochaetes residing within flat ticks are not dormant; the bacteria in dissected unfed midguts are motile, albeit sluggish⁷¹. Moreover, microarray analyses of spirochaetes cultivated in vitro under 'unfed-tick' conditions, as well as quantitative reverse transcriptase PCR (qRT-PCR) of selected genes in borreliae residing in flat ticks, suggest that many borrelial genes are preferentially expressed in the anoxic environment of the unfed midgut^68,72,73,74. A stringent response analogous to that observed in E. coli during amino acid starvation has been suggested to occur during this phase, as the levels of transcripts encoding a RelA–SpoT orthologue are higher in bacteria cultivated in vitro under conditions that mimic unfed ticks than in bacteria cultivated under conditions that mimic fed ticks and mammals⁷². Because the spirochaetes in unfed ticks are not replicating, they do not require N-acetylglucosamine. They do, however, require a carbon source to maintain subsistence levels of glycolysis. Glycerol, which is thought to be produced as a natural antifreeze by overwintering ticks⁶⁹, could permeate the outer membrane of the spirochaete and enter the bacterial cytoplasm via glycerol uptake facilitator GlpF, which is expressed by borreliae in unfed ticks⁶⁹. In addition to OspA, the surface-exposed lipoprotein BptA⁷⁵ and the Dps-like bacterioferritin orthologue BB0690 (Ref. 27) are essential for prolonged residence in flat ticks.

The nymphal blood meal. The nymphal blood meal ends a long nutritional drought but also presents the bacterium with a new and complex set of challenges. To take advantage of the influx of nutrients, the spirochaete must sense the new environment and extensively revamp its substrate uptake mechanisms and intermediary metabolism from famine to feast mode^24,68. Whereas a temperature shift during cultivation in vitro induces many of the antigenic changes associated with nymphal feeding in borreliae^37,38,41,68, temperature shifting of infected unfed ticks does not⁷⁶. Metabolic changes in the spirochaete that are associated with nutrition from the blood meal, and possibly components of the blood meal itself, are required to fully activate the genetic programmes associated with transmission^48,74.

During the first 24 hours of nymphal feeding (the preparatory phase of the digestive process), little to no blood enters the midgut, and spirochaete numbers remain essentially unchanged from numbers in the unfed-tick state^71,77. Activation of the Rrp2–RpoN–RpoS pathway at the outset of feeding (Fig. 4b) results in the transcription of genes that are required for mammalian infection and simultaneously begins the slow repression of tick-phase genes^37,38,63,68. Nymphal feeding also induces the expression of genes which are regulated independently of RpoS and are dependent instead on σ⁷⁰ (also known as RpoD). The products encoded by these genes include the outer-membrane-spanning protein p66, which acts as a porin and adhesin^78,79,80, and BbCRASPs (complement regulator-acquiring surface proteins), which are outer-surface lipoproteins that protect the bacterium against complement-mediated lysis by binding complement factor H and complement factor H-like protein 1 (Refs 81, 82, 83, 84, 85). As is the case during larval feeding, spirochaetes require the two-component system Hk1–Rrp1 to avoid being destroyed at the start of nymphal feeding^56,57 (Fig. 4b).

By 48 hours, bacterial replication and dissemination are well under way^{68,71,77,86,87}. Although it has been suggested that spirochaetes traverse the tick midgut by downregulating OspA, detaching from epithelial cells, and then using their motility to migrate between cells and penetrate the basement membrane, two lines of evidence argue that this scenario is overly simplistic. First, OspA is highly expressed by spirochaetes distributed throughout the midgut, even at late time points during feeding^62,63,87. Second, confocal microscopy has shown that the bacteria maintain a close association with the epithelial cells as they replicate, coalescing into networks of non-motile bacteria that literally surround the hypertrophying epithelial cells and progress towards the basement membrane^71,86. Spirochaetes that reach the basolateral surface of the epithelium transition into motile organisms that cross the basement membrane and enter the haemocoel^71,88. BBE31, a surface-exposed borrelial lipoprotein, may be required for spirochaetes to penetrate the midgut through its interaction with TRE31, a tick protein that is secreted by epithelial cells during feeding⁸⁹. Binding of host-derived molecules, such as plasminogen, to the bacterial surface probably facilitates penetration of the collagenous matrix of the basement membrane⁹⁰. The salivary glands pose the final mechanical barrier to be overcome; only a handful of spirochaetes reach the interior of the gland, where they access the salivary stream^71,77,87.

Within the mammalian host

The bite site. The host factors in tick saliva that enhance the survival of B. burgdorferi and are inoculated into the cutaneous bite site (Fig. 5) include molecules that can impede various mammalian responses, such as the generation of reactive oxygen species (ISL 1373), activation of complement (ISAC and SALP20), chemotaxis of neutrophils (sialostatin L), and antibody-mediated killing, dendritic cell-mediated priming of T cells and keratinocyte-mediated release of cytokines and antimicrobial peptides (SALP15)^68,91,92,93. Expression of the bacterial protein OspC is crucial for the establishment of early infection^94,95,96, although there is controversy as to whether OspC promotes the penetration of salivary glands by spirochaetes disseminating within the tick or the survival of spirochaetes deposited at the bite site through binding of SALP15. As in the feeding nymph, borrelial proteins that bind to complement factor H- and complement factor H-like protein 1 protect the bacterium against complement-mediated killing during this early window of vulnerability^{80,81,82,83,84}.

Figure 5: The tick–mammal interface.

The tick creates a feeding pit with its mouthparts, using its hypostome (a barbed protuberance) as an anchor to the skin of its host. Initial salivary secretions form a cement cone around the hypostome that further anchors the tick during feeding. Subsequently, the tick produces copious amounts of saliva containing a plethora of bioactive agents that Borrelia burgdorferi sensu lato exploits to help establish infection. The diagram shows just a few of these bioactive agents, including SALP15 (which binds to the spirochaetes and inhibits killing of the bacteria by antibodies, as well as T cell priming within the lymph nodes) and sialostatin L (which blocks neutrophil chemotaxis)^68,92,93. A group of borrelial surface lipoproteins, collectively referred to as BbCRASPs (complement regulator-acquiring surface proteins), bind complement factor H, preventing activation of the alternative complement pathway81–84. B. burgdorferi cells are recognized by innate immune effector cells such as dendritic cells (DCs), neutrophils and macrophages, initially via surface-exposed pattern recognition receptors; activation of these cells increases following internalization and degradation of spirochetes within phagolysosomes. DCs that have taken up spirochaetes migrate to the lymph nodes, where they present processed borrelial antigens to T cells and B cells. Sensitized T cells enter the circulation and are recruited to the site of infection. Plasma cells secrete specific antibodies that can kill B. burgdorferi via complement-dependent and -independent pathways. Production of pro-inflammatory cytokines by activated macrophages results in the recruitment of additional neutrophils, T cells, macrophages and DCs to the bite site, and eventually the development of erythema migrans¹⁵. OspC, outer-surface lipoprotein C; TLR, Toll-like receptor.

Initial sensing of B. burgdorferi by the mammalian host probably occurs through pattern recognition receptors, such as Toll-like receptors (TLRs) and NOD-like receptors (NLRs), on dendritic cells and sentinel macrophages within the dermis (Fig. 5). TLR2, which heterodimerizes with TLR1 to recognize triacylated lipoproteins, seems to be central to the induction of many inflammatory cytokines in response to B. burgdorferi. Engagement of TLR2–TLR1 heterodimers by borrelial lipoproteins results in activation of mitogen-activated protein kinase (MAPK) pathways, translocation of nuclear factor-κB (NF-κB) into the nucleus, and subsequent production and release of inflammatory mediators¹⁵. Other TLRs, such as TLR5, TLR7, TLR8 and TLR9, cooperate with TLR2–TLR1 to induce pro-inflammatory molecules, including type I interferons (IFNs)^97,98,99,100. TLR-mediated release of chemokines and cytokines by resident phagocytes presumably attracts other inflammatory cells to the tick bite site. In mice, neutrophils are recruited early but disappear within 16 hours of inoculation and are followed by monocytes and macrophages, both of which are capable killers of B. burgdorferi¹⁵. However, B. burgdorferi is a highly motile organism and can move at speeds of up to 4 μm sec⁻¹ in tissue⁶¹, enabling it to evade capture by these comparatively sluggish professional phagocytes. Spirochaetes that are unable to evade phagocytes are ingested and rapidly degraded within phagosomal vacuoles⁹⁸. The phagocytosis of borreliae and subsequent degradation within phagolysosomes further amplifies the release of inflammatory cytokines through activation of TLR signalling within phagolysosomes⁹⁸. Internalization of spirochaetes in the absence of antibodies is mediated by at least two different integrins, αMβ2 integrin (also known as CR3 or CD18–CD11b) and α3β1 integrin^15,101. Although the internalization of B. burgdorferi results in activation of the inflammasome, neither the course of disease nor the spirochaete burden is worsened appreciably in caspase 1-deficient mice¹⁵. In humans, recruitment of T cells heralds the development of the characteristic erythema migrans rash (Fig 1). Biopsies of erythema migrans lesions show infiltration by T cells (CD8⁺ cells as well as CD4⁺ cells), macrophages, plasmacytoid and monocytoid dendritic cells, and neutrophils in varying proportions, but no B cells^102,103; a number of inflammatory cytokines and chemokines have also been identified in erythema migrans lesions^102,104.

Dissemination and immune evasion. After a delay of up to 2 days, B. burgdorferi begins to spread to distant tissues¹⁰⁵. To disseminate, B. burgdorferi penetrates the matrix between cells and enters capillary beds. The spirochaete circumvents its inability to produce enzymes that are capable of digesting extracellular matrix components by appropriating host proteases such as plasminogen and its activator, urokinase^88,106. B. burgdorferi also induces multiple host matrix metalloproteinases (MMPs), the major class of host proteases involved in the degradation of extracellular matrix components, from both phagocytic and non-phagocytic cells^15,80. Entry into capillaries provides B. burgdorferi with access to the bloodstream. Egress from the circulation into tissues involves tethering and adhesion to vascular endothelium followed by extravasation⁶¹. B. burgdorferi expresses a variety of adhesins that could mediate attachment to host tissues through diverse receptors such as integrins (which can bind p66 and the outer-surface protein encoded by the locus BB_B07)^78,107, proteoglycans (which can bind decorin-binding protein A (DbpA) and DbpB)¹⁰⁸, glycosaminoglycans¹⁰⁹, laminin (which can bind ErpX and BmpA)^110,111 and fibronectin (which can bind the protein encoded by the locus BB_K32 and RevA)^80,112,113.

Carditis in mice and humans is predominantly caused by infiltration of monocytes or macrophages, with a minority of lymphocytes and neutrophils^15,17. Arthritis in susceptible strains of mice, by comparison, shows a predominance of neutrophils with smaller numbers of macrophages and lymphocytes¹⁵. IFNγ appears to mediate carditis, but not arthritis, in susceptible mice^114,115, whereas type I IFNs may be particularly important for the development of arthritis⁹⁹. B. burgdorferi is unique among pathogens in that its diacylglycerol glycolipid can directly activate invariant natural killer T cells¹¹⁶. These T cells might participate in clearing spirochaetaemia after being recruited by Kupffer cells that recognize B. burgdorferi in the liver¹¹⁷. In mice, invariant natural killer T cells are recruited to the heart during B. burgdorferi infection, where they could have a role in controlling infection and inflammation through augmenting phagocytosis¹¹⁴.

Both the innate and adaptive immune systems are important for controlling infection and inflammation during the disseminated phase, as mice that are deficient in either system have greater bacterial burdens than wild-type mice¹⁵. T cells do not appear to be required for resolution of either arthritis or carditis in mice; in fact, the presence of T cells without B cells can worsen both¹¹⁸. However, CD4⁺ T cells can hasten resolution of carditis in the presence of B cells¹¹⁸. The development of the specific humoral response to B. burgdorferi is crucial for clearing the pathogen^15,119. Early T cell-independent production of antibodies, predominantly immunoglobulin M, is crucial for the initial reduction of spirochaete burdens^120,121. T cell-dependent production of immunoglobulin G by B cells is typically detectable by the second week of infection¹⁵. Antibody production does not seem to require TLR signalling, as both TLR2-deficient and MYD88-deficient mice develop normal humoral responses to infection with B. burgdorferi^121,122,123. Antibodies against many different borrelial proteins elicit strain-dependent reductions in pathogen burden or a reduction in carditis or arthritis¹⁵. In response, the bacterium is thought to downregulate lipoproteins, such as OspC, which are no longer required to establish or maintain infection^96,124. B. burgdorferi also uses a system of antigenic variation to evade antibodies. VlsE is a 35 kDa lipoprotein that undergoes antigenic variation through the recombination of sequences from silent cassettes into the expressed vlsE locus^84,125,126. The spirochaete may also exploit mechanisms that are used by the host to suppress inflammation and therefore limit tissue damage, in effect making the mammalian host an unwitting accomplice to the spirochaete's strategy for persistence¹²⁷.

Concluding remarks and future directions

Two facets of the life cycle of B. burgdorferi — the bacterium's ability to adapt to markedly divergent host environments and its ability to evade the defences of its mammalian reservoir — account, to a large extent, for the extraordinary zoonotic success of this spirochaete and its continued expansion as a threat to public health. We are only just beginning to understand the mechanisms whereby this versatile bacterium coordinates changes in its transcriptome, cellular architecture and metabolism with the feeding behaviour and physiology of its arthropod vector. Marginally better understood are the stratagems that enable the spirochaete to navigate within the mammal, regularly gaining entry to niches that are impassable for other bacteria, when establishing a persistent infection. Despite its small genome, the Lyme disease spirochaete possesses deceptively complex machinery for tightly regulating gene and protein expression, with an unusual combination of components identified to date. The small number of orthologues for known transcription factors hints at the existence of novel post-transcriptional control pathways; it is noteworthy that the burgeoning interest in small RNAs among microbiologists has only recently made its influence felt in the field of borrelial research. Co-evolution of the spirochaete with its arthropod and mammalian hosts has enabled the bacterium to take advantage of host physiological processes and make compensatory reductions in its own biosynthetic machinery. In so doing, the bacterium has had to develop ingenious and parsimonious strategies to obtain and utilize the nutrient sources available within the 'feast and famine' confines of its life cycle. Indeed, the nutrients themselves, especially carbon sources, and/or their metabolic by-products seem to provide regulatory as well as chemotactic signals that guide the spirochaete as it moves between hosts.

Beginning with the discovery that spirochaete lipoproteins are major inflammatory agonists, we have learned a great deal about the mechanisms by which the mammalian host senses the presence of live spirochaetes and mobilizes cellular and humoral defences to combat the intruder. By contrast, little is understood about the processes that occur late in infection and the mechanisms that enable the bacterium to persist in the face of the robust cellular and humoral immune responses that it elicits. In the past, the emphasis has been on immune evasion by the spirochaete, and this topic needs far more scrutiny. Nevertheless, attention also needs to be paid to the possibility that host-mediated modulation of pathogen-sensing pathways and of resultant responses contributes to spirochaete persistence. We are just beginning to understand the polymorphisms in specific mouse genes that result in some inbred strains of mice being more prone to manifestations of infection than the natural P. leucopus host. Whether these genes are identical to those responsible for disease susceptibility and expression in humans is unknown. However, one can envision translational studies to answer the question of why some infected patients have only subclinical disease, whereas others develop overt manifestations. These and other advances will hopefully lead to a better understanding of the determinants of vector and host specificity for B. burgdorferi, and to the manipulation of these determinants to interrupt the cycle of transmission. Although eliminating Lyme disease spirochaetes from nature is unrealistic, diminishing their threat to humans would seem to be an achievable goal.

Box 1 | Animal models of Lyme disease

One of the important early advances in the study of Lyme disease was the development of animal models of infection. As mice and other rodents are natural reservoirs for Borrelia burgdorferi sensu lato, the causative agent, it is not surprising that wild Peromyscus spp. mice show no observable changes during infection with B. burgdorferi, as symptomatic disease could leave the infected animal with a survival disadvantage and/or limit the opportunities for transmission of spirochaetes to new cohorts of vectors¹²⁸. However, when infected with B. burgdorferi, specific inbred strains of the laboratory mouse Mus musculus have been found to exhibit features similar to those of human Lyme disease^14,15. For example, C3H and BALB/c mice develop both ankle joint arthritis and carditis on infection with B. burgdorferi, whereas C57BL/6 and DBA mice are more resistant to developing signs of infection and typically have only minimal inflammation in the heart and joints. The greater disease susceptibility of C3H mice does not seem to be due to greater control of the infection, as the spirochaete numbers are comparable to those in disease-resistant mouse strains. By contrast, in BALB/c mice, spirochaete numbers and signs of inflammation directly correlate with the inoculum size, suggesting that the mechanisms of arthritis and carditis are different in this mouse strain. No strain of inbred mouse develops erythema migrans, meningitis or encephalitis, and thus all are imperfect models for human Lyme disease. Also unknown is whether the genetic determinants controlling the development of arthritis and carditis in inbred mice are related to those responsible for the corresponding manifestations in humans.

Other animals have been used to model specific features of human Lyme disease. Dogs infected with B. burgdorferi can develop arthritis and facial nerve palsies¹²⁹. Rhesus monkeys can also be infected with B. burgdorferi and are used as a model of neuroborreliosis because of their propensity for central nervous system infection, particularly when immunosuppressed with corticosteroids¹⁴. Rhesus monkeys also develop erythema migrans, mononeuritis multiplex and arthritis, making them the animal model that is most similar to humans for this disease¹⁴. However, for reasons of cost and ease of genetic manipulation, animal models other than mice have been studied sparingly.