Borrelia burgdorferi infection causes an initial skin lesion called erythema migrans (EM) in human Lyme disease and in models of monkey and rabbit borreliosis. EM results from the inflammatory response triggered by spirochete replication and likely develops to contain the initial infection but allows bacterial dissemination to occur. The essential lack of neutrophil involvement in EM histopathology prompted us to examine the consequence of increasing their recruitment in the inflammatory response to the Lyme disease agent. B. burgdorferi was modified genetically to constitutively express and secrete the chemokine KC, a neutrophil chemoattractant. After inoculation into the dermis of the murine host, control spirochetes induced an infiltration of macrophages, neutrophils, and basophils within 6 h; however, the recruited neutrophils and basophils were quickly substituted by eosinophils, and the inflammatory response became macrophage dominant by 16 h. Such a response failed to contain the initial infection and allowed the spirochetes to disseminate. In contrast, B. burgdorferi with KC secretion induced an intensive neutrophil infiltration at the inoculation site, and as a result, the host’s ability to control the initial infection was greatly enhanced. Taken together, this study suggests that the failure of sufficient neutrophil recruitment and activation during the initial inflammatory response may allow B. burgdorferi to effectively colonize the mammalian host.
The immune system consists of billions of motile cells that protect the host against infection. The development, trafficking, and adhesion of immune cells depends largely on a sophisticated communication system composed of chemokines and cytokines (1). Chemokine messages are decoded by specific receptors that initiate signal transduction events leading to various effector responses. The best-known chemokine effector mechanism is chemotaxis, in which leukocytes migrate in response to a chemotactic gradient and accumulate in the focus, where immune cells are most needed. During a microbial infection, tissue macrophages recognize and engulf pathogens or simply recognize their products by means of surface receptors that are able to bind common constituents of microorganisms and subsequently are activated to release cytokines and chemokines, thereby initiating an inflammatory response (2, 3). Inflammatory chemokines recruit leukocytes such as neutrophils and monocytes from the bloodstream into the site of infection (1). Both macrophages and neutrophils are professional phagocytes capable of effectively killing foreign invaders; such an inflammatory response is able to clear most infections. However, some microbial pathogens evade initial elimination, disseminate to distal tissues, and establish a chronic infection.
The inability of the initial inflammatory response to contain infection reflects the ability of microbial pathogens to escape innate immunity. The Lyme disease spirochete Borrelia burgdorferi is one of the most invasive bacterial pathogens; infection with it results in a multisystem disorder, such as arthritis, neurological abnormalities, carditis, and acrodermatitis chronica atrophicans in humans (4). In nature, the enzootic life cycle of B. burgdorferi is maintained via cycling between mammals and the vector tick Ixodes scapularis (5). Although it is a slow-growing bacterium with a doubling time of ∼8 h in the best in vitro conditions, its ID50 value in the murine host is <100 organisms (6, 7). B. burgdorferi infection in humans induces a significant inflammatory response at the site of the tick bite, leading to the development of an early manifestation called erythema migrans (EM)3 (4). Histopathologically, EM is marked by perivascular infiltration of lymphocytes and plasma cells with few macrophages and scattered mast cells (8). Similar skin lesions were reported in both monkey and rabbit models of Lyme borreliosis (9, 10). In the monkey, the histopathology of EM consists of macrophage and lymphocyte infiltration, accompanied by a prominent eosinophil and an occasional basophil response (9). In the rabbit EM lesion, the infiltration is a mixed lymphocyte population with occasional macrophages and plasma cells (10). Apparently, EM results from the inflammatory response triggered by spirochete replication and develops to contain infection. Interestingly, tick saliva inhibits neutrophil function and may aid in establishing B. burgdorferi infection (11, 12). The failure of the inflammatory response to control initial infection, in conjunction with the essential lack of neutrophil involvement in EM development, prompted us to examine the consequence of increasing neutrophil recruitment and activation during B. burgdorferi infection.
To modulate the initial inflammatory response, B. burgdorferi was modified genetically to constitutively express and secrete the chemokine KC, a potent neutrophil attractant and activator (13). KC is the murine homolog of human growth-related oncogene-α; the functional receptor for KC has been identified as CXCR2 (14). KC is expressed at high levels in the joints of B. burgdorferi-infected susceptible mice, and blocking the activation and recruitment of neutrophils via CXCR2 prevented the development of Lyme arthritis (15). In this study, enhancing the expression of KC at the inoculation site led to an intensive neutrophil infiltration and dramatically increased the host’s ability to control the initial infection.
Materials and Methods
Construction of plasmid pBBE22-flaBp-bba74s-kc-ospCt
The plasmid was constructed as illustrated in Fig. 1⇓; the sequences of primers used in the study are listed in Table I⇓. A 251-bp fragment of the flaB promoter region (flaBp) and a 262-bp downstream sequence (ospCt) starting from the stop codon of the ospC gene were amplified, respectively, with primers P1F and P1R and P2F and P2R. Spirochetal DNA was used as a template; resulting amplicons were purified, digested with AflII, repurified, ligated, and used as template for nested PCR with use of primers P3F and P3R and cloned into the TA cloning vector pNCO1T as previously described (16), creating an intermediate vector designated pNCO1TBC, which can replicate in Escherichia coli but not in B. burgdorferi.
RNA was extracted from a joint specimen of a SCID mouse that had been infected with B. burgdorferi and developed severe arthritis; then, the RNA was converted into cDNA using the primer KC-RT, which is specific for the murine chemokine KC, as described previously (17). A 383-bp fragment (bba74s), including the signal peptide-coding region and upstream noncoding sequence of the bba74 gene, was amplified using spirochetal DNA as a template and P4F and P4R as primers. A 279-bp sequence (kc), including the 77-aa KC-coding region and a short downstream sequence, was generated using the mouse cDNA preparation as a template source and P5F and P5R as primers. Purified PCR products were subjected to blunt-end ligation and then used as a template for nested PCR with the use of primers P6F and P6R, generating a 319-bp amplicon coding exactly for the signal peptide of BBA74 and the 77-aa KC. Following purification, the PCR product was digested with NdeI and AflII, repurified, and cloned into the intermediate vector pNCO1TBC to create a second intermediate plasmid designated pNCO1TBCKC, which can replicate in E. coli but not in B. burgdorferi. The entire insert was sequenced to ensure it was as designed; then, it was cleaved off and finally cloned into the recombinant plasmid pBBE22 (a gift from S. Norris, Houston, TX), following digestion with BamHI and XbaI. The constructed plasmid was designated pBBE22-flaBp-bba74s-kc-ospCt, which can replicate in both E. coli and B. burgdorferi systems (7).
Generation of variants with KC expression and secretion
The B. burgdorferi B31 5A13 culture (a gift from S. Norris) was subjected to serial dilution as described previously (18); a single clone, which lost lp25 and lp56 but contained the 19 remaining plasmids, was obtained and designated 13A. Bacteria were harvested from 3.0 ml of the 13A culture and transformed either with the recombinant plasmid pBBE22 or pBBE22-flaBp-bba74s-kc-ospCt as described previously (18). Transformants were identified by PCR using a primer pair specific for the kanamycin cassette (18). Their plasmid contents were first surveyed by PCR and then confirmed by microarray hybridization using primers and protocols reported previously (18).
In vitro growth rate study
Spirochetes grown to late logarithmic phase were transferred to fresh BSK-H complete medium (Sigma-Aldrich) at a cell density of 5 × 105/ml. The culture was maintained at 33°C, and spirochetes were counted daily under a darkfield microscope. Cell densities were calculated using the formula (bacterial number per milliliter = cell number/field (×400) × dilution fold × 3 × 105).
ELISA determination of KC concentrations
Spirochetes were grown in 1.4 ml of BSK-H complete medium at 33°C to early- and late-log and stationary phases, then harvested by centrifugation at 13,000 × gg for 5 min, and 100 μl of the resulting sample was used in the ELISA. Each individual sample was run in triplicate wells. The KC level associated with the spirochete pellet was converted to KC content per milliliter of culture.
C56BL/6J mice at ages of 4–6 wk (The Jackson Laboratory) received an i.p. injection of a mixture of B. burgdorferi culture supernatant (100 μl) and 400 μl of PBS. Recombinant KC (1 ng) expressed in E. coli
SCID mice on a BALB/c or C3H background at ages of 6–8 wk (provided by the Louisiana State University Division of Laboratory Animal Medicine, Baton Rouge, LA) each received two intradermal/s.c. injections at two inoculation sites in the chest. The breeding pairs (BALB/cJHanHsd-Prkdcscid and C3H.C-Prkdcscid/IcrSmnHsd) were initially ordered from Harlan Sprague Dawley. One site was inoculated with 50 μl of BSK medium containing 105 spirochetes with KC expression; the second site was injected with the same number of bacteria without KC expression as a control. The distance between the two inoculation sites was at least 2 cm. Injection sites were labeled with a marker immediately after inoculation. Mice were euthanized 6, 16, 24, or 72 h later; inoculation-site skin and associated s.c. tissues were collected for histopathological study using H&E staining as described previously (18).
BALB/c SCID mice 4–5 wk of age received a single intradermal/s.c. injection of 107 spirochetes. Mice were sacrificed 30 days postinoculation; heart, tibiotarsal joint, and skin specimens were collected aseptically for spirochete culture as described previously (18).
Determination of ID50 values
Spirochetes were grown at 33°C to late-log phase (108 cells/ml) and 10-fold serially diluted with BSK-H complete medium. C3H SCID mice 4–5 wk of age each received one single intradermal/s.c. injection of 100 μl of spirochetal suspension. Mice were euthanized 30 days postinoculation; heart, tibiotarsal joint, and skin (not from inoculation site) specimens were harvested for bacterial culture as described previously (18). The ID50 value was calculated as described by Reed and Muench (19).
Passive immunization with KC mAb
n = 3); each mouse was intradermally/s.c. inoculated with 10–105 spirochetes on day 1 as described above. The injections of KC mAb and rat IgG were repeated on days 3 and 14. Mice were sacrificed 1 mo after initial inoculation; heart, tibiotarsal joint, and skin specimens were aseptically collected for spirochete culture, and ID50 values were determined as described above.
Generation of B. burgdorferi variants expressing KC
The presence of lp25 and lp56, the two plasmids that may carry restriction enzymes, negatively affects the transformation efficiency of B. burgdorferi (20). A highly transformable clone designated 13A, which lost both lp25 and lp56, was subcloned via serial dilution from the B. burgdorferi B31 clone 5A13 that was lacking lp25 (7). The plasmid lp25, but not lp56, is required for infection because it carries bbe22, a gene that encodes a nicotinamidase essential for the survival of B. burgdorferi in the mammalian environment (7). The 13A spirochetes, which contained 19 of the 21 known borrelial plasmids, were transformed with either the recombinant plasmid pBBE22 or pBBE22-flaBp-bba74s-kc-ospCt; 15 and 12 transformants were obtained with each plasmid. Transformation of 13A with pBBE22 is expected to restore infectivity as the recombinant plasmid contains a copy of the bbe22 gene (7). One advantage of this system is that pBBE22 and its associated genes are firmly maintained during mammalian infection. The plasmid pBBE22-flaBp-bba74s-kc-ospCt was constructed from pBBE22 as illustrated in Fig. 1⇑. Plasmid content analyses by PCR and microarray hybridization identified two transformants receiving each plasmid, namely 13A/E22/A, 13A/E22/B, 13A/KC/A, and13A/KC/B, for further analysis. These transformants had the same plasmid content and all were lacking cp9, lp28-1, lp21, and lp5, in addition to lp25 and lp56; however, cp9, lp21, and lp5 are not important for mammalian infection (21). Due to an initial biosafety consideration that B. burgdorferi armed with KC secretion might dramatically increase pathogenicity, only clones lacking lp28-1, the plasmid that is essential for the infection of immunocompetent animals (7) were chosen; therefore, all animal studies were done in SCID mice.
KC expressed by B. burgdorferi is highly secreted
KC expression and successful secretion were first confirmed by a chemokine ELISA. Spirochetes were grown to different cell densities; KC associated with cells and secreted into the supernatant was separately quantified. As expected, the clone 13A/E22/A did not make any KC, either associated with the cells or secreted into the medium (Table II⇓). In contrast, the 13A/KC/A and 13A/KC/B produced KC in all growth phases and accumulated the chemokine at concentrations of >30 ng/ml, when grown to stationary phase, indicating that the fused flaB promoter is able to drive constitutive KC expression. More importantly, >97% of the total KC was detected in the supernatant, indicating that the fused signal peptide of bba74 leads to a successful secretion of KC into the extracellular environment.
KC secreted by B. burgdorferi is biologically active
The biological activity of secreted KC was confirmed by a chemotaxis assay, in which a stationary phase supernatant of the 13A/E22/A culture and recombinant KC were used as negative and positive controls, respectively. Like recombinant KC, all the supernatants harvested from the early- and late-log, and stationary phases of the 13A/KC/A culture increased neutrophil counts at least 12-fold in peritoneal leukocytes (p values ranging from 3.9 × 10−51 to 2.5 × 10−47; Table III⇓), indicating a high activity of KC secreted by genetically modified B. burgdorferi.
KC expression and secretion does not affect in vitro spirochetal growth
The doubling time during the log phase and the cell density at stationary phase were assessed to reflect KC expression and secretion effects on spirochete growth. When grown in BSK-H complete medium at 33°C, the transformants 13A/E22/A, 13A/E22/B, 13A/KC/A, and 13A/KC/B completed a generation approximately every 8 h at log phase and reached maximum cell densities of 2.0 × 108/ml at stationary phase (data not shown), indicating that the addition of KC expression and secretion does not affect in vitro Borrelia growth. Furthermore, spirochetal morphology and motility were not influenced by KC expression and secretion, as assessed under a darkfield microscope.
B. Burgdorferi with KC secretion modulates the inflammatory response to neutrophil dominance
To investigate the effects B. burgdorferi with KC secretion has on recruiting neutrophils in the tissue, SCID mice were intradermally/s.c. inoculated with 105 spirochetes and euthanized 6, 16, 24, and 72 h later. Histopathological examination of the inoculation site with the control 13A/E22/A spirochetes revealed a significant inflammatory infiltration composed of macrophages, neutrophils, and basophils within 6 h (Fig. 2⇓A, left panel). By 16 h, both neutrophils and basophils disappeared; the inflammatory response became macrophage dominant, accompanied with a significant number of eosinophils (Fig. 2⇓B, left panel). The response continued to develop for 24 h and then maintained its intensity to the end of the 72-h study (Fig. 2⇓, C and D, left panels). Throughout the study, recruited cell types underwent dynamic changes; neutrophils and basophils were involved within the first 6 h and were then replaced by eosinophils within the next 10 h. The ratio of eosinophils to macrophages gradually declined from 16 to 72 h postinoculation.
In contrast, the 13A/KC/A spirochetes recruited a massive neutrophil infiltration as early as 6 h postinoculation (Fig. 2⇑A, right panel), reaching a peak infiltration at 16 h (Fig. 2⇑B, right panel). Then, the inflammatory response underwent a resolving process (Fig. 2⇑C, right panel); by the end of the 72-h study, nearly all involved neutrophils were cleared, probably by macrophages, which were recruited to repair damaged tissues (Fig. 2⇑D, right panel). Similar results were obtained when C3H SCID mice were used (data not shown).
KC secretion severely attenuates infectivity potential
Early and sustained recruitment of neutrophils might have a significant effect on the ability of B. burgdorferi to establish and maintain an infection. A high dose of bacteria was used in a preliminary study to examine whether intensive neutrophil infiltration can contain the initial infection. Groups of five SCID mice on a BALB/c background were challenged with a single dose of 107 spirochetes of the clone 13A/E22/A, 13A/E22/B, 13A/KC/A, or 13A/KC/B. One month later, mice were euthanized; heart, joint, and skin samples were used for bacterial culture. The 13A/E22/A and 13A/E22/B bacteria were recovered from each specimen harvested from all of the 10 inoculated mice (Table IV⇓). In contrast, the 13A/KC/A and 13A/KC/B spirochetes were grown only from 6 of the 10 animals, indicating that chemokine secretion severely attenuates but does not abolish the infectivity of B. burgdorferi.
Next, the ID50 values were determined to reflect how much the B. burgdorferi infectivity potential was reduced. C3H SCID mice were inoculated with different doses of bacteria. Like BALB/c SCID mice, the immunodeficient animals on a C3H background were successfully infected only with the dose of 107 bacteria with KC expression. Overall, the ID50 values of the clones 13A/E22/A and 13A/E22/B were 75 and 50 organisms, respectively, in contrast to 13A/KC/A and 13A/KC/B with the values of 5.2 × 106 and 7.5 × 106 (Table V⇓), reflecting a 105-fold decrease in infectivity. This severely attenuated virulence most likely resulted from the enhanced host’s ability to contain initial infection, attributed to the intensive neutrophil infiltration in the skin tissue at the inoculation site (Fig. 2⇑, right panel).
Anti-KC Ab treatment dramatically increases infectivity of KC-secreting B. burgdorferi
To determine whether the attenuated infectivity of B. burgdorferi with KC secretion resulted indirectly from the recruitment of neutrophils, we completed a KC neutralization study. Treatment of mice with a KC mAb decreased the ID50 value of the clone 13A/KC/A to 500 organisms (Table VI⇓) from ∼5.2 × 106 (Table V⇑), reflecting in a 10,000-fold increase in infectivity. This ID50 value was still 10-fold higher than that of B. burgdorferi without KC expression, probably because the mAb treatment was unable to completely eliminate the high level of bacterium-secreted KC.
The inflammatory response constitutes the first line of host defense against microbial infection. Its quality, measured as the composition and magnitude of immune cells recruited to infected tissues, largely determines the fate of an infection. Although inflammatory cells as a whole can fight various types of infection, only the macrophage and neutrophil are able to directly kill bacterial pathogens. All leukocytes are effectively recruited by chemokines, some of which are cell specific, whereas others attract multiple types. In this study, B. burgdorferi was modified genetically to constitutively express and secrete the CXCR2 neutrophil chemoattractant KC. The chemokine, secreted into the extracellular environment, was highly active in a chemotaxis assay. B. Burgdorferi with KC secretion recruited an intensive neutrophil infiltration at the inoculation site; as a result, the host’s ability to contain the initial Borrelial infection was enhanced greatly.
The flaB promoter and the ospC terminator of B. burgdorferi were used to initiate and stop transcription, respectively, of the introduced chemokine gene. Like Gram-negative bacteria, B. burgdorferi has an inner and outer membrane, between which is the periplasmic space. The signal peptide of BBA74, an outer membrane protein of B. burgdorferi that can be secreted into the extracellular environment of bacteria (22), was added to direct recombinant KC to cross the inner membrane (cytoplasmic membrane), where the signal peptide might have been cleaved off by a signal peptidase; such a process might allow the biologically active KC portion to be released into the periplasmic space. Although it is unknown how KC crossed the outer membrane into the extracellular environment, >97% of the chemokine synthesized by B. burgdorferi was indeed detected in the supernatant. More importantly, the chemotaxis assay showed the secreted KC actively recruited neutrophils. The expression and delivery system was so efficient that the resulting neutrophil infiltration was very intense, demonstrating the feasibility of genetically engineering pathogenic bacteria to secrete chemokines in a biologically active form. This may provide a powerful tool for studying mechanisms that microbial pathogens use to manipulate the inflammatory response and evade innate immunity.
As a novel entity, the pathogenic bacterium modified with chemokine secretion was completely unknown with unknown biohazard potential before the study. Therefore, as a containment measure, only transformants lacking lp28-1 were used. Although this plasmid is essential for the ability of B. burgdorferi to evade adaptive immunity, the lack of lp28-1 does not affect the infectivity potential in SCID mice (7, 23). The advantage of using immunodeficient mice in the current study is to completely eliminate potential interferences resulting from adaptive immune responses, thus allowing the investigation to focus on the inflammatory response solely contributed by innate immunity.
B. burgdorferi infection induces a significant inflammatory response in the skin tissue at the initial entry site, leading to the genesis of EM lesions in human Lyme disease and in models of monkey and rabbit borreliosis (8, 9, 10). One common feature of EM histopathology in these hosts is the lack of a neutrophil presence. Although an EM clinical sign is not producible in the murine model, an EM-like histopathology has been induced consistently in the current study. This allowed us to thoroughly investigate the kinetics of the initial inflammatory response to B. burgdorferi infection. Such a study has been lacking in the EM histopathology of human Lyme disease and in models of monkey and rabbit borreliosis. Within the first 6 h after inoculation, neutrophils were indeed recruited to the site of infection but were quickly excluded from the inflammatory response. By 16 h postinoculation, the response became macrophage dominant, accompanied with a significant number of eosinophils. As SCID mice were used in the study, lymphocytes and plasma cells, which abundantly accumulate in the EM lesion of human Lyme disease and in models of monkey and rabbit borreliosis (8, 9, 10), were absent. Both basophils and eosinophils have been reported in the EM lesion of monkey Lyme borreliosis (9). Neither the basophil nor the eosinophil is expected to directly kill bacterial pathogens; however, they are recruited in the inflammatory response to B. burgdorferi infection. In contrast, the neutrophil, a major professional bacterial killer, was involved only during the first hours. Such an inflammatory response is unable to effectively contain initial infection and allows spirochetal dissemination to occur.
The macrophage is a primary phagocyte capable of effectively engulfing many microbial pathogens. Despite the large number of macrophages involved, however, the initial inflammatory response allows B. burgdorferi to escape elimination, raising questions about the role of macrophages in directly eliminating Lyme disease spirochetes. The neutrophil is also a phagocyte capable of killing bacteria but is apparently involved only within the first hours and then excluded from the initial inflammatory response to B. burgdorferi infection in the skin. The polarization of the response to neutrophil dominance, in which other types of inflammatory cells were essentially absent, by modifying B. burgdorferi to constantly secrete KC dramatically increased the host’s ability to control initial infection, clearly demonstrating that the phagocyte can directly eliminate Lyme disease spirochetes. Like macrophages, the neutrophil can phagocytose and kill microorganisms intracellularly. In addition, it is equipped with several antimicrobial systems, which allow it to combat a broad spectrum of bacteria and other microbial pathogens in different ways (24, 25). After an encounter with bacteria, neutrophils may be activated to release specific granules to the exterior of the cell to target bacteria extracellularly (24). A study by Lusitani et al. (26) showed that B. burgdorferi is susceptible to in vitro killing by various human polymorphonuclear leukocyte components. However, it has also been suggested that neutrophil killing of B. burgdorferi is inefficient unless the spirochetes are opsonized with specific Ab (27). The current study clearly indicates that the neutrophil can effectively kill Lyme disease spirochetes in the murine host in the complete absence of Ab; whether the killing is in an intracellular and/or extracellular manner remains to be investigated.
The failure of macrophages recruited at the inoculation site to contain the initial infection suggests that the phagocyte may inefficiently engulf or kill Lyme disease spirochetes in the murine host. The chemokine expression and secretion system should help address this issue. By using the same strategy, B. burgdorferi should be easily modified to secrete a monocyte/macrophage chemoattractant such as the chemokine C10 (28). If a more intensive recruitment of monocytes from the bloodstream into infected tissues is unable to improve the host’s ability to contain the initial infection, the inability of macrophages to eliminate the spirochetes would be confirmed. The same strategy can be used to attract other innate immune cells, study their roles in combating bacterial infection, and explore mechanisms microbial pathogens use to evade the innate immune response. Moreover, the concept of genetically engineering pathogenic bacteria to express and secrete chemokines and even cytokines may be further extended to recruit adaptive immune cells and improve the quality of specific immune responses against microbial infection, thus facilitating vaccine development.
Increasing neutrophil recruitment by modulating the inflammatory response greatly enhances the host’s ability to contain the initial infection, indicating that this phagocyte is able to effectively eliminate the Lyme disease agent. Interestingly, this major professional bacterial killer is involved only within the first hours after B. burgdorferi infection. Although it remains to be determined why neutrophils are not more highly recruited and activated during an initial B. burgdorferi infection, our study suggests that this inefficient neutrophil response may allow B. burgdorferi to effectively avoid elimination and establish an infection in the mammalian host.
We thank S. Norris for providing the B. burgdorferi B31 clone 5A13 and M. T. Kearney for assistance with statistical analysis.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was in part supported by National Institutes of Health AR052748 (to C.R.B.), and a National Institutes of Health/National Institute of Arthritis and Musculoskeletal and Skin Diseases Career Development Award and an Arthritis Foundation Investigators Award (both to F.T.L.).
↵2 Address correspondence and reprint requests to Dr. Fang Ting Liang, Department of Pathobiological Sciences, Louisiana State University, Skip Bertman Drive, Baton Rouge, LA 70803. E-mail address:
↵3 Abbreviation used in this paper: EM, erythema migran.
- Received November 29, 2006.
- Accepted January 26, 2007.
- Copyright © 2007 by The American Association of Immunologists