INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Relapsing fever is reported to have been described by Hippocrates in the 4th century B.C. (30). The name relapsing fever has been attributed by several workers to Craigie (18, 30, 57), who, along with Henderson, described an epidemic fever of humans in Edinburgh, Scotland, in 1843 (17, 35). Jenner also presented a clinical description of the disease in 1850 (39). The spirochetal agent of louse-borne relapsing fever (LBRF) was first observed in the blood of patients by Obermeier during an outbreak of the disease in Berlin, Germany, in 1868 (7). However, the role of the human body louse (Pediculus humanus) in spirochete transmission was not described until 1907 (45). The causative agent of LBRF is now known as Borrelia recurrentis. The organism has no wild animal reservoir and is transmitted solely among humans by the body louse (14). LBRF was once widespread globally when human body lice were much more abundant than they are today (30). In the 19th century, substantial outbreaks occurred in the British Isles (17), Europe (7), and the United States (49). Outbreaks also were documented in some parts of Europe, India (19), China (16), the Andean region of South America, and several African countries in the first half of the 20th century. In recent decades, LBRF has been recorded only in northeastern and central Africa, especially Ethiopia, Somalia, and Sudan, where infestations of human body lice remain prevalent (1-3, 8, 10, 15, 24, 34, 48, 60, 66, 68, 69; P. L. Perine and D. F. Reynolds, Letter, Lancet ii:1324-1325, 1974).

The clinical manifestations of LBRF include the classical recurrence of acute episodes of fever (13, 56), sometimes complicated by bleeding associated with thrombocytopenia (25). Antibiotic treatment can initiate a rapid and fatal Jarisch-Herxheimer reaction (72, 74). Historically, laboratory confirmation of all relapsing fevers, including both louse-borne and tick-borne forms, has relied on the identification of spirochetes in patient blood in the febrile episodes (11). However, spirochetes frequently are not identified because of the cyclic nature of the spirochetemia and low sensitivity of detection by light microscopy. As a consequence, various types of serological tests have been developed to detect antibodies produced during infection with relapsing-fever spirochetes (11, 30, 31) and thereby enhance diagnosis. However, the utility of these assays has been limited by a lack of sensitivity and specificity.

Previously, we reported that the enzyme glycerophosphodiester phosphodiesterase (GlpQ) is absent in Lyme disease spirochetes but is present and immunogenic in the tick-borne relapsing-fever spirochete, Borrelia hermsii (62). We found that GlpQ was recognized by antisera obtained from humans and other animals after infection with tick-borne relapsing-fever spirochetes. In contrast, serum samples taken from humans with a diagnosis of Lyme disease were nonreactive.

Cutler and coworkers (21) demonstrated in 1994 that Kelly's medium (40) supported the continuous growth of B. recurrentis. This ability to culture LBRF spirochetes creates opportunities to perform in vitro studies and to develop new diagnostic tests. Four isolates of B. recurrentis were cultured from the blood of acutely ill, spirochetemic patients in a recent outbreak of LBRF in southern Sudan. The glpQ gene from each of these isolates and those from four other Borrelia species were sequenced. Recombinant B. recurrentis GlpQ protein was used for serological testing of acute- and convalescent-phase serum samples from LBRF patients. Our findings demonstrate the utility of GlpQ to serologically confirm LBRF. This antigen will also be useful for retrospective serological surveys when the presence of LBRF is suspected.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Borrelia strains and cultivation. B. recurrentis isolates 107, 115, 119, and 132 were obtained from the blood of four LBRF patients living in Rumbek County of southern Sudan during an epidemiological investigation in April 1999. B. hermsii strain DAH was isolated at Rocky Mountain Laboratories (RML) from the blood of a human with relapsing fever in eastern Washington (62). Borrelia coriaceae CO53 (ATCC 43381) was isolated from Ornithodoros coriaceus collected in California (42). Borrelia parkeri RML, Borrelia turicatae RML, and Borrelia anserina RML were isolated from Ornithodoros parkeri, Ornithodoros turicata, and a domestic chicken, respectively, and are part of the RML bacterial reference collection. Borrelia crocidurae CR2A was provided by Sven Bergström, Umeå University, Umeå Sweden. B. burgdorferi B31 was isolated from Ixodes scapularis collected on Shelter Island, New York (12). Lysates of Treponema pallidum were provided by Steven Norris, University of Texas Health Science Center, Houston.

PCR and DNA sequence analysis. Total genomic DNA was purified from 100-ml cultures of each B. recurrentis isolate or 500-ml cultures of the other Borrelia species, quantified by UV spectroscopy, and diluted to approximately 0.1 µg for use in each 100-µl PCR (50, 51). Taq enzyme and reaction constituents were used as recommended by the manufacturer (Perkin-Elmer, Roche Molecular Systems, Inc., Branchburg, N.J.). Primers used for amplifying glpQ DNA fragments were manufactured by Life Technologies, Baltimore, Md. (Table 1). PCRs were performed under mineral oil for 25 cycles with a Perkin-Elmer thermocycler. Each cycle consisted of denaturation at 94°C for 1 min, annealing at 50°C for 30 s, and extension at 72°C for 2 min. After the 25th cycle, an additional 7-min extension was done at 72°C.

PCR amplification and glpQ cloning. The glpQ gene and presumed ribosomal binding site were amplified by PCR from B. recurrentis isolate 115 with primers formulated on the basis of the glpQ gene region of B. hermsii (Table 1). After amplification, 10 µl of the total 100-µl reaction mixture was examined in a 0.7% agarose electrophoresis gel stained with ethidium bromide. An amplification product of the correct size was cloned into the pCR2.1 vector of the TA Cloning System (Invitrogen Corp., San Diego, Calif.) and transformed into Escherichia coli. The sequence of the entire DNA insert in this recombinant plasmid (pTA-115) was determined with the M13 universal primers and an internal set of primers (Table 1), which confirmed the identity, integrity, and proper orientation of glpQ.

Production of rabbit antiserum. Antisera to recombinant GlpQ were produced in two rabbits with different antigen preparations. For one preparation, a 100-ml overnight culture of E. coli TA-115 was centrifuged and suspended in 10 ml of phosphate-buffered saline (PBS), and the cells were disrupted by sonication. An equal volume of the lysate was emulsified in Ribi adjuvant (Corixa Montana, Hamilton, Mont.), and 5 ml of the mixture was injected subcutaneously at five sites. For the second preparation, a whole-cell lysate of E. coli TA-115 was separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and the unfixed gel was stained with a water-based Coomassie blue to visualize the GlpQ band. This band was excised from the gel, triturated in 5 ml of PBS with a mortar and pestle, and injected subcutaneously at five sites. Both rabbits were bled prior to the primary immunizations and boosted subcutaneously on day 28 postimmunization with a preparation identical to that used for the primary immunization. Blood from both rabbits was collected again 33 and 57 days after the first booster injection.

Human serum. Acute- and convalescent-phase serum samples obtained from 42 patients hospitalized in Addis Ababa, Ethiopia, with fever and documented spirochetemia during the acute phase of illness have been described previously (54). Although the spirochetes infecting these patients were not identified, LBRF is hyperendemic in Ethiopia (8; G. Borgnolo, B. Hailu, and F. Chiabrera, Letter, Lancet 338:827, 1991), and no recent reports of tick-borne relapsing fever are known for this region, strongly suggesting that B. recurrentis was the cause of the infections. These samples were collected from December 1970 to June 1971 (54) and were stored frozen at 20°C at RML until used in this study. The duration between the collection of acute- and convalescent-phase samples ranged from 5 to 17 days (mean of 11 days). No information regarding antibiotic treatment of these patients was available.

One-dimensional gel electrophoresis. Whole-cell lysates of spirochetes were prepared as described previously (61). One-dimensional SDS-PAGE using Laemmli buffer (41) and a vertical gel apparatus (Bethesda Research Laboratories-GIBCO, Gaithersburg, Md.) was used to separate proteins.

Western blot analysis. Whole-cell lysates were electrophoresed in one-dimensional acrylamide gels and blotted onto nitrocellulose membranes with Towbin buffer (71) and a Trans-Blot cell (Bio-Rad Laboratories, Hercules, Calif.). The membranes were blocked overnight at room temperature with TSE-Tween (50 mM Tris [pH 7.4], 150 mM NaCl, 5 mM EDTA, 0.05% Tween 20) and incubated with either antiflagellin monoclonal antibody H9724 (6), rabbit antisera to the recombinant GlpQ diluted 1:500, or human antisera diluted 1:100. Bound antibodies were detected by ¹²⁵I-labeled protein A autoradiography (61).

Indirect immunofluorescence assay (IFA). B. recurrentis 132 was cultured in BSK-H medium, harvested by centrifugation, rinsed twice with PBS, and mixed with fresh, washed sheep red blood cells. These sheep cells provide a negative background for comparison and an internal check for nonspecific binding of conjugate. Thin smears of the cell suspensions were made on glass microscope slides, dried at room temperature, fixed with methanol, wrapped with foil, and stored at 20°C until used. Human serum samples were tested with twofold serial dilutions ranging from 1:16 to 1:2,048. Bound antibodies were detected with a 1:100 dilution of goat anti-human immunoglobulin G (heavy plus light chains)-fluorescein isothiocyanate (Kirkegaard & Perry, Gaithersburg, Md.) and epifluorescence microscopy. The intensity of fluorescence was scored as 1+ to 4+, and the endpoint was defined as the highest dilution that provided reactivity greater than the background with the red blood cells. The geometric mean titers were determined by Perkins's method (55).

ELISA with whole-cell antigen. A 500-ml culture of B. recurrentis containing approximately 10⁸ cells per ml was centrifuged (12,500 × g), rinsed and suspended twice in PBS, and diluted to an optical density of 0.05 at 600 nm. This suspension was sonicated on ice in 20-ml portions for 2 min at an output setting of 5 using a Branson Sonifier-Cell Disruptor 185 (VWR Scientific, San Francisco, Calif.). The protein concentration of the sonicated suspension was determined with the Bradford assay (Bio-Rad Laboratories). The sonicate was stored at 4°C prior to use. To test sera by enzyme-linked immunosorbent assay (ELISA), Immulon-2 96-well, flat-bottomed microdilution plates (Dynatech Laboratories, Inc., Alexandria, Va.) were coated with 100 µl of the sonicated spirochetal suspension per well (229 µg per ml) and were dried overnight at 37°C. Wells were blocked to inhibit nonspecific binding with 200 µl of diluent (PBS, 5% horse serum, 0.05% Tween 20, 0.001% dextran sulfate) for 1 h at 37°C and then washed once with PBS-0.05% Tween 20. One pair of acute- and convalescent-phase serum samples were tested at eight twofold serial dilutions (1:32 to 1:4,096) by incubating 100 µl of each dilution per well for 1 h at 37°C. After three washes, 100 µl of a 1:2,500 dilution of goat anti-human immunoglobulin G (heavy and light chains) conjugated to horseradish peroxidase (Kirkegaard & Perry Laboratories) was added per well and incubated for 1 h at 37°C. After three washings, a substrate of 50% 2,2'-azino-di-(3-ethyl-benzthiazoline sulfonate) was added and left for 20 min before analysis at 405 nm with a Labsystems Multiskan Plus microtiter plate reader (Fisher Scientific, Pittsburgh, Pa.). This assay showed good discrimination between the negative and positive samples at dilutions above 1:128. A dilution of 1:250 was chosen to test the 84 Ethiopian serum samples and 12 normal serum samples by the procedure described above. Each serum sample was tested in triplicate, and the mean absorbance value was determined. Samples were considered positive if their mean absorbance was greater than the mean plus 3 standard deviations (SDs) of the absorbance of normal control sera tested at the same dilution.

Synthesis and purification of recombinant GlpQ for ELISA. The glpQ gene of B. recurrentis isolate 115 was amplified with PCR using primers that contained XhoI (Br GlpQ fus 5') and BamHI (Br GlpQ fus 3') sites (Table 1) for cloning in frame with an amino-terminal histidine (His) tag in the pET-15b expression vector (Novagen Inc., Madison, Wis.). Total genomic DNA (40 ng) of the spirochete was used as the template in the PCR with an initial heating at 94°C for 2 min; 25 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 2 min; and a final extension at 72°C for 7 min. The 5' fusion primer began with the codon for the first amino acid immediately downstream of the lipidated cysteine, and the 3' primer corresponded to a region of DNA sequence downstream of the B. recurrentis glpQ stop codon. The PCR fragment was ligated in the pCR2.1 vector (Invitrogen) according to the instructions of the manufacturer, and the plasmid was transformed into E. coli. Bacterial colonies were screened with PCR, and plasmid DNA was purified from a positive clone with a miniprep kit (Qiagen Inc., Valencia, Calif.) and digested with BamHI and XhoI. The restricted DNA fragment was purified with a QiaexII gel purification kit (Qiagen) and quantitated by UV spectroscopy. The pET-15b vector was digested with BamHI and XhoI, purified with a Quick Step clean-up kit (Edge BioSystems, Gaithersburg, Md.), quantitated by UV spectroscopy, treated with HK-phosphatase (Epicentre, Madison, Wis.) for 1 h at 30°C, and heat inactivated at 65°C for 15 min. The B. recurrentis glpQ PCR fragment was ligated to the pET-15b vector, transformed into E. coli XL1-Blue cells (Stratagene, La Jolla, Calif.) by electroporation, and grown on Luria broth plates with ampicillin (100 µg/ml). A single recombinant was picked and checked by PCR using the PCR conditions described above and the Br GlpQ fus 5' and Br GlpQ fus 3' primers. Vector DNA was purified from this recombinant with a Qiagen miniprep kit, quantitated, and transformed into chemically competent E. coli BLR(DE3) cells. A single recombinant was examined with PCR using the Br GlpQ fus 5' and Br GlpQ fus 3' primers and used for expression of the GlpQ fusion protein.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Identification of B. recurrentis. The four spirochete isolates cultured from patients in southern Sudan were presumptively identified as B. recurrentis on the basis of clinical and epidemiological history, presence of human body lice, detection of spirochetes in blood smears, ability of modified Kelly's medium to support spirochetal growth, and history of prior LBRF outbreaks in this region. Immunoblot analysis of whole-cell lysates with the genus-specific monoclonal antibody H9724 (6) identified the spirochetes as Borrelia (data not shown). The plasmid profiles of these four isolates were distinct from those obtained from all other species of Borrelia in our collection, were different from those described for Borrelia duttonii (20), and were similar to those described for B. recurrentis (22). The DNA sequences of the 16S rRNA and flagellin genes from the four isolates were identical to B. recurrentis sequences deposited in GenBank (data not shown), thereby confirming the species identification.

Identification and DNA sequence analysis of the glpQ genes of B. recurrentis and other Borrelia species. Genomic DNA preparations obtained from the four B. recurrentis isolates and single isolates of B. crocidurae, B. turicatae, B. parkeri, and B. coriaceae yielded PCR amplification products of the appropriate size for glpQ. The amplified products were sequenced, and BLAST searches of the eight open reading frames confirmed that the sequences were glpQ.

View larger version (9K): [in this window] [in a new window]   FIG. 1.   Dendrogram showing the relatedness of the glpQ DNA sequences from B. recurrentis (four isolates) and B. crocidurae, B. parkeri, B. turicatae, B. hermsii, and B. coriaceae (one isolate each). The scale represents the percentage of nucleotide substitutions observed out of the total number of nucleotides compared.

View larger version (51K): [in this window] [in a new window]   FIG. 2.   Alignment of the deduced GlpQ amino acid sequences, including those from B. recurrentis 115, B. crocidurae CR2A, B. parkeri RML, B. turicatae RML, B. hermsii DAH, and B. coriaceae CO53. Consensus amino acid residues are framed, and dashes indicate gaps introduced to maximize alignment.

Immunological detection of GlpQ. The glpQ gene of B. recurrentis isolate 115 was cloned into the pCR2.1 vector. DNA sequence analysis confirmed that glpQ had been cloned and that no spurious mutations were present. This transformant of E. coli, designated TA-115, was inoculated into a rabbit to produce antibodies to GlpQ. The preinoculation serum was nonreactive, but serum obtained at 28 days after the primary rabbit immunization reacted strongly to a single 39-kDa protein in the B. recurrentis lysate (data not shown). A protein with a slightly greater apparent molecular weight was identified in E. coli TA-115 but was absent in the lysate made from E. coli containing vector only. The difference in electrophoretic mobility between the native and recombinant GlpQ may be due to differential processing in E. coli and B. recurrentis. The rabbit immunized with the suspension of acrylamide containing recombinant GlpQ had very weak serological reactivity 28 days after the primary immunization. However, at 33 days following the booster injection, this serum reacted strongly by immunoblot analysis to GlpQ in lysates of B. recurrentis and E. coli TA-115. In addition, this serum identified the homologous GlpQ protein in B. crocidurae, B. hermsii, B. parkeri, B. turicatae, B. coriaceae, and B. anserina (Fig. 3). As expected, this serum did not react with B. burgdorferi, which lacks GlpQ (32, 62). The serum also did not react with T. pallidum, which has a GlpQ with only 38.7 to 42.9% amino acid identity with GlpQ of B. recurrentis (the variation in identity values is due to the use of different algorithms).

View larger version (71K): [in this window] [in a new window]   FIG. 3.   Immunoblot analysis with rabbit antiserum collected 4 weeks after booster injection with GlpQ excised from the gel preparation of E. coli TA-115, showing the presence of GlpQ in whole-cell lysates of B. recurrentis and other Borrelia species except B. burgdorferi. No reactivity with GlpQ made by T. pallidum was detectable. Lysates of E. coli TA-115 expressing GlpQ and E. coli with the vector only were used as controls. Molecular mass standards (MMS) are shown on the left in kilodaltons.

Reactivity of LBRF patient serum samples to B. recurrentis and recombinant GlpQ. IFA is a useful test for the initial serological screening of antibodies resulting from Borrelia infections (5, 63), but the assay has not been applied previously to LBRF. Paired serum samples from 42 human LBRF patients hospitalized with acute fever and spirochetemia in Addis Ababa, Ethiopia, were tested by IFA with whole cells of B. recurrentis isolate 132. The IFA geometric mean titers for the acute- and convalescent-phase serum samples were 1:83 and 1:575, respectively. Thirty-three patients (78.5%) had a fourfold or greater increase in titer in the convalescent-phase sample, five patients (12%) had a twofold increase, and four patients (9.5%) had no increase in titer. None of the serum pairs had an acute-phase titer greater than that of the convalescent-phase sample. Seventeen (40%) of the 42 acute-phase serum samples had an IFA titer equal to or greater than 1:128, suggesting that (i) these patients had seroconverted prior to hospitalization, (ii) the patients were infected prior to the current illness, or (iii) the titers were falsely positive.

View larger version (78K): [in this window] [in a new window]   FIG. 4.   Immunoblots with representative paired acute-phase (lanes A) and convalescent-phase (lanes C) human serum samples from six LBRF patients (a to f) showing seroconversion to GlpQ (arrow). Whole-cell lysates of B. recurrentis, E. coli TA-115 with recombinant GlpQ, and E. coli vector only were used as controls. The asterisk shows reactivity to a putative 22-kDa variable small protein. Convalescent-phase samples were obtained 10 to 13 days after the acute-phase samples were drawn. Molecular mass standards (MMS) are shown on the right in kilodaltons.

View larger version (102K): [in this window] [in a new window]   FIG. 5.   Immunoblot analysis of four serum samples obtained from the same individual collected 17 to 27 years after documented infection with B. recurrentis. The patient was infected in 1973 in Addis Ababa, Ethiopia. Each serum sample was tested against whole-cell lysates of B. recurrentis, E. coli TA-115 expressing recombinant GlpQ, and E. coli containing the vector only. The top arrow indicates reactivity with GlpQ in E. coli TA-115, which is weak compared to the reactivity shown in Fig. 4 with serum samples obtained from patients only 10 to 13 days after hospitalization. The lower arrow indicates strong reactivity to a putative 22-kDa variable small protein. Molecular mass standards (MMS) are shown on the right in kilodaltons. p.i., postinfection.

View larger version (12K): [in this window] [in a new window]   FIG. 6.   Scatter plot showing reactivity of each sample of acute-phase () and convalescent-phase () serum determined by IFA to the entire spirochete and by immunoblotting to recombinant GlpQ (n = 84). An IFA titer equal to or greater than 1:128 was considered to be positive. The horizontal dashed line separates values judged to represent positive and negative titers. A positive immunoblot result was considered to be any reactivity to recombinant GlpQ. The vertical dashed line separates positive and negative immunoblot samples. X indicates the IFA geometric mean titer for each group of samples with equal reactivity to recombinant GlpQ. Reactivity to GlpQ was quantified by measuring the thickness of the band in the blot.

View larger version (16K): [in this window] [in a new window]   FIG. 7.   ELISA absorbance values of acute- and convalescent-phase LBRF patient serum samples with two antigen preparations. (A) Whole-cell sonicate of B. recurrentis as antigen. All serum samples were tested in triplicate at 1:250, and the average value was used. The mean absorbance for each group is shown by the horizontal solid bar, 1 SD above and below the mean is shown by the vertical open bar, and the range is shown by the vertical line. The horizontal dashed line represents the threshold for determining a positive sample, determined by the mean absorbance plus 3 SDs of the absorbance values of the normal serum samples. (B) His-GlpQ purified protein as antigen and serum samples tested in triplicate at a 1:100 dilution.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Previously, we identified, cloned, and characterized the B. hermsii glpQ gene, made recombinant GlpQ protein, and produced rabbit anti-GlpQ antibody (62). The rabbit antiserum reacted specifically with single proteins of approximately 38 to 42 kDa in B. hermsii, B. parkeri, B. turicatae, the Florida canine borrelia, B. crocidurae, B. coriaceae, B. anserina, and recombinant E. coli. Mice infected with B. hermsii, B. turicatae, the Florida canine borrelia, B. crocidurae, and B. duttonii and chickens infected with B. anserina seroconverted to recombinant B. hermsii GlpQ. Humans with tick-borne relapsing fever in North America also had antibodies to GlpQ, whereas Lyme disease patients did not. These serological data suggest that all species of Borrelia other than those assigned to the B. burgdorferi species complex (sensu lato) make a GlpQ homolog. In the present work, we characterized the glpQ genes in B. recurrentis, B. crocidurae, B. turicatae, B. parkeri, and B. coriaceae. We also confirmed that B. anserina makes a putative GlpQ homolog. We have been unable to amplify glpQ in this species and, as a consequence, have sequenced only about 50% of the gene. The protein appears to be larger than homologs in all other species by immunoblot analysis (Fig. 3). GlpQ was not detected in B. burgdorferi or T. pallidum by immunoblotting with the rabbit anti-B. recurrentis GlpQ antibody, results identical to our earlier observations with antiserum produced to B. hermsii GlpQ (62). The absence of glpQ in B. burgdorferi was confirmed when it was shown to be absent from the genome sequences (32). In contrast, T. pallidum has a glpQ (=gpd) homolog (33, 65, 67). Its lack of reactivity with anti-Borrelia GlpQ antisera is likely due to the low level of identity (38%) between the amino acid sequences.

Glycerophosphodiester phosphodiesterase (GlpQ) was first described for E. coli as the product of a member of the glp regulon (43). The proteins encoded by genes in this regulon participate primarily in the salvage of glycerol released when phospholipids and triglycerides are degraded (44). Specifically, GlpQ in E. coli hydrolyzes deacylated phospholipids to form an alcohol and glycerol-3-phosphate (70). Glycerol-3-phosphate is either used in the synthesis of new phospholipids or converted to dihydroxyacetone phosphate by GlpD and then to glyceraldehyde-3-phosphate by glyceraldehyde-3-phosphate dehydrogenase for use in glycolysis (44). Homologs of glpQ also have been identified in Haemophilus influenzae (38, 52), Bacillus subtilis (53), and several other species by genome sequencing projects.

The function of GlpQ in Borrelia spirochetes is unknown, and its subcellular localization in B. hermsii is uncertain. Previously we found no reduction in the amount of this protein detected following proteinase K treatment of intact B. hermsii, suggesting that GlpQ is not located on the outer surface (62). However, Shang and coworkers (64) reported that GlpQ (=Gpd) was present in outer membrane preparations and therefore may be associated with the inner surface of the outer membrane. Regardless of its location, this putative enzyme in relapsing-fever spirochetes is quite immunogenic.

GlpQ of B. recurrentis rapidly stimulates a strong antibody response in humans that is detectable by immunoblotting 1 to 2 weeks after clinical presentation. However, this protein is produced by many species of Borrelia that achieve significant densities in blood and are transmitted by argasid ticks. In Africa, the tick-borne relapsing-fever spirochetes B. duttonii and B. crocidurae may occur where their respective tick vectors, Ornithodoros moubata and Ornithodoros erraticus sonrai, also occur. Therefore, in geographic regions where the ranges of these infected ticks overlap with the presence of humans infested with body lice, there could be difficulties in serological confirmation of infection with these closely related species. We found that the GlpQ proteins of B. recurrentis and B. crocidurae differ by only one amino acid. Hence, retrospective serological testing with only GlpQ in the absence of a clinical history or epidemiological information will not distinguish serologically between exposure to these two species of spirochetes. The geographical distribution of O. erraticus in Africa suggests possible overlap with regions of LBRF endemicity in central and eastern Africa (23, 29), as is true for O. moubata (27, 37). Although we were unable to investigate B. duttonii, we expect that this species has a GlpQ with high amino acid sequence identity to the B. recurrentis and B. crocidurae GlpQs. Hence, identifying the probable vector as either the human body louse or Ornithodoros ticks will support serological testing for anti-GlpQ antibodies to identify the causative agent.

IFA titers with serum samples from LBRF patients obtained soon after infection were strongly cross-reactive with B. hermsii and B. burgdorferi. We and others have observed this phenomenon with serum samples from tick-borne relapsing-fever and Lyme disease patients (46, 47, 62), emphasizing the importance of immunoblotting with specific recombinant antigens. However, the serum samples obtained from the patient infected with B. recurrentis 17 to 27 years earlier were nearly or completely nonreactive with these other species but were still positive with B. recurrentis. These data, although limited, suggest that serological cross-reactivity by IFA may wane with time after exposure.

In summary, we have characterized the glpQ genes in four isolates of B. recurrentis from Sudan and in single isolates of B. crocidurae, B. turicatae, B. parkeri, and B. coriaceae. The gene from B. recurrentis was cloned and expressed in E. coli, and the recombinant GlpQ protein was used to test sera from human LBRF patients. Paired acute- and convalescent-phase serum samples from these patients demonstrated seroconversion to this antigen 1 to 2 weeks after hospitalization. Immunoblotting with recombinant GlpQ was more sensitive than the ELISA with purified His-tagged GlpQ, although the sensitivity of this assay may be improved by increasing the concentration of antigen. The IFA with fixed, whole spirochetes and the ELISA with sonicated, whole-cell borrelia antigen performed equally but are less specific in their reactivity for borrelioses. Thus, GlpQ will significantly increase the specificity of serological testing for LBRF in regions, like Africa, where patients infested with body lice present with recurrent febrile disease.