INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Infection by pathogenic Leptospira species is an important and frequently life-threatening cause of human disease characterized by hematogenous dissemination to multiple organs including the brain, aqueous humor, liver, lungs, and kidneys. Leptospirosis occurs in a variety of urban and rural settings, and is considered to be the most widespread zoonosis in the world (10, 22, 43, 46). The wide distribution of Leptospira species results from their ability to colonize the renal tubules of a diverse group of wild and domestic animals. After urinary shedding, Leptospira species are transmitted directly to a new host or indirectly through contact with organisms contaminating moist environments. The ability to survive as free-living organisms is unique among the invasive spirochetes and presumably reflects differential expression of proteins involved in adaptation to the environment outside the mammalian host. Based upon these biological considerations, it is anticipated that certain leptospiral proteins expressed in cultivated organisms may or may not be expressed during infection (5). Proteins expressed during infection may serve as determinants in leptospiral pathogenesis and as targets for the host immune response. To develop a more comprehensive understanding of leptospiral protein expression, we have used the humoral immune response during human leptospirosis as a reporter of protein antigens expressed during infection.

The identification of leptospiral antigens expressed during infection has potentially important implications for the development of new serodiagnostic and immunoprotective strategies. Most research on leptospiral antigens has been focused on lipopolysaccharide (LPS). Variations in the carbohydrate side chains of LPS are responsible for the antigenic diversity observed among leptospiral serovars, of which over 250 have been identified (10). As a result of the immunodominance of LPS, leptospiral vaccines consisting of inactivated whole-cell immunogens, termed bacterins, are based largely on inducing antibodies against carbohydrate epitopes within this moiety. For this reason, currently used vaccines often do not provide cross-protection against serovars not contained in vaccine preparations. In contrast, leptospiral protein extracts can induce protection against challenge with heterologous serovars in experimental animal models (39).

The antigenic variability of leptospiral LPS is also a limitation for serodiagnosis. The microscopic agglutination test (MAT) has been the "gold standard" confirmatory test for the past 70 years and is most likely based on seroreactivity with the LPS antigens. The need to assess agglutination by dark-field microscopy and maintain a large battery of live leptospiral antigens in culture restricts the use of the MAT to a few reference laboratories worldwide. More widely accessible serologic approaches have been developed, approaches which take advantage of cross-reactive antigens in crude extracts which are shared among diverse leptospiral serovars. These cross-reactive antigens could include proteins and/or components of leptospiral LPS (30). Currently available serologic assays include the macroscopic agglutination (31), indirect hemagglutination (28), and microcapsule agglutination (2) tests, all of which are less sensitive than the MAT and identify less than 50% of patients presenting with early-phase leptospirosis. Assays that detect immunoglobulin M (IgM) and are based upon crude antigen (1, 14, 27, 38, 44) appear to be more sensitive for serodiagnosis but may be subject to variations in specificity.

The need to develop better serodiagnostic strategies has become even more critical now that leptospirosis has been recognized as an emerging cause of epidemics such as the 1995 outbreak of severe pulmonary hemorrhage syndrome in Nicaragua (42). In the rest of Latin America, large epidemics occur annually among impoverished populations in major urban centers and are associated with case fatality rates of over 15% (22, 29). In order to respond to this emerging public health problem, case identification needs to be performed promptly so that rapid outbreak investigations and timely administration of antibiotic therapy can be implemented. However, the broad spectrum of clinical presentations associated with leptospirosis hampers case identification. In several outbreak situations, leptospirosis was initially confused with dengue (22, 26, 35, 42). Therefore, early diagnosis must rely on an efficient laboratory test that can be easily implemented in the field without dependence on reference laboratory settings.

For the purpose of developing a diagnostic test that can be applied to the variety of epidemiological situations associated with human and veterinary leptospirosis, ideally an antigen which is highly conserved among diverse pathogenic leptospiral strains should be selected. The amino acid sequences of leptospiral proteins, such as the major outer membrane protein, LipL32, appear to be highly conserved across leptospiral species (16). To identify candidate protein antigens for serodiagnosis, we characterized the humoral immune response in leptospirosis by studying the immunoblot reactivity of a large number of patients and by characterizing the recognized protein antigens. Earlier one-dimensional immunoblot studies used clinical sera to identify the relative mobility of several immunogenic proteins but were unable to further characterize these antigens (7, 8). Recent molecular characterization of leptospiral proteins such as GroEL (4, 33), DnaK (3), the OmpL1 porin (15, 36), and the lipoproteins LipL41 (37) and LipL32/MOMP (16) has provided the antibody reagents needed to definitively identify many of the major protein antigens. Sera from leptospirosis patients from Barbados and Brazil were used to perform one- and two-dimensional immunoblot analyses of leptospiral proteins. The major outer membrane protein, LipL32, and heat shock proteins GroEL and DnaK were found to be the dominant immunoreactive protein antigens. The human immune response also identified the surface-exposed lipoprotein, LipL41, the outer membrane porin, OmpL1, and a series of other less well characterized membrane-associated proteins. On the other hand, sera from patients did not recognize the previously described protein, LipL36, which is a prominent component of the leptospiral outer membrane in organisms cultured in vitro. We believe that these data provide useful insights into the pathogenesis of leptospirosis and the identification of candidate protein antigens for serodiagnosis and immunoprotection.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and media. Leptospira kirschneri strain RM52 (41) and other leptospiral strains were obtained from the National Leptospirosis Reference Center (National Animal Disease Center, Agricultural Research Service, U.S. Department of Agriculture, Ames, Iowa). Most of the leptospiral strains used in this publication are described in a recent DNA relatedness study (6). Leptospiral strains analyzed by 1D and 2D electrophoresis were clinical isolates from Salvador, Brazil (Leptospira interrogans serogroup Icterohaemorrhagiae serovar copenhageni) (22) and Barbados (L. kirschneri serogroup Autumnalis serovar bim) (21). Leptospires were cultivated in Johnson-Harris Bovine Serum Albumin Tween 80 medium (Bovuminar PLM-5 Microbiological Media; Intergen) (20). Escherichia coli BLR(DE3)pLysS [F ompT hsdS_B (r_B m_B) gal dcm (srl-recA)306::Tn10(TcR) (DE3) pLysS(CmR)] (Novagen) was used as the host strain for the pRSET expression vector (Invitrogen). E. coli cells were routinely grown in Luria-Bertani broth or on Luria-Bertani agar (34).

Patients and control individuals. During active hospital-based surveillance for epidemic leptospirosis in the city of Salvador, Brazil, consecutive patients were identified between March 1996 and February 1998 according to a clinical definition based on the presence of characteristic severe manifestations (jaundice and acute renal failure) without laboratory or radiological evidence for a disease other than leptospirosis (22). According to the surveillance routine, a first, acute-phase serum sample was collected at the time of hospital admission. A second, convalescent-phase serum sample was collected 14 or more days after the collection of the acute-phase sample, typically during outpatient evaluation after hospital discharge. Informed consent was obtained from patients or their guardians, and the guidelines of the Brazilian Ministry of Health, Barbados Ministry of Health, the New York Presbyterian Hospital, and the U.S. Department of Health and Human Services were followed in the conduct of the clinical research.

Rabbit antisera. Leptospiral GroEL serum was a generous gift of B. Adler (Monash University, Clayton, Victoria, Australia). Leptospiral DnaK serum was a generous gift of J. Timoney (Gluck Equine Institute, Lexington, Ky.). Antisera to OmpL1 (15), LipL32 (16), LipL36 (17), LipL41 (37), and LipL45/31 (J. Matsunaga, M. Mazel, T. Young, and D. A. Haake, unpublished data) were prepared by immunizing New Zealand White rabbits with purified His6 fusion proteins.

Gel electrophoresis and immunoblotting. For one-dimensional sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), samples were solubilized in a final sample buffer composed of 62.5 mM Tris hydrochloride (pH 6.8), 10% glycerol, 5% 2-mercaptoethanol, and 2% SDS and were separated on a discontinuous buffer system (23). Two sets of molecular mass standards were used in SDS-PAGE: for quantitative immunoblot analyses with individual patients' sera, prestained high-range protein standards (Gibco BRL) which contained rabbit skeletal muscle myosin H-chain (200 kDa), rabbit muscle phosphorylase B (97.4 kDa), bovine serum albumin (68 kDa), hen egg white ovalbumin (43 kDa), bovine carbonic anhydrase (29 kDa), beta lactoglobulin (18.4 kDa), and hen egg white lysozyme (14.3 kDa) were used; for qualitative analyses with pooled human sera and two-dimensional electrophoresis, protein standards (Bio-Rad) which contained rabbit skeletal muscle myosin (200 kDa), E. coli beta-galactosidase (116 kDa), rabbit muscle phosphorylase B (97 kDa), bovine serum albumin (66.2 kDa), hen egg white ovalbumin (45 kDa), bovine carbonic anhydrase (31 kDa), soybean trypsin inhibitor (21.5 kDa), and hen egg white lysozyme (14.4 kDa) were used. Two-dimensional gel electrophoresis was performed according to the method of O'Farrell (32) modified by Görg et al. (11, 12). Samples for 2D gel electrophoresis were solubilized in a rehydration solution composed of 8 M urea, 2% Triton X-100, 20 mM dithiothreitol, and 2% carrier ampholyte mixture (IPG Buffer; Pharmacia). Immobiline DryStrips (Pharmacia) were rehydrated overnight in rehydration solution containing leptospiral material. Isoelectric focusing was performed using a Pharmacia Multiphor II system. After isoelectric focusing, SDS-PAGE was performed as described above. Gels were stained with Coomassie brilliant blue or were transferred to 0.45-µm Immobilon-P membranes (Millipore) for immunoblotting.

Scoring of immunoblots and statistical analysis. A pilot study was performed to identify the spectrum of antigen bands recognized by 30 convalescent-phase sera from Leptospirosis Group 1 patients. Relative mobility (M_r) was estimated for identified antigens based on comparisons with prestained high-range protein molecular mass standards (Gibco BRL). Two serum samples which in combination recognized all identified antigen bands were chosen and used in subsequent analyses as quality control standards to identify the positions of antigen bands in each immunoblot. For the purpose of determining the proportion of sera that react to individual antigen bands, two investigators used a scale based on visual intensity (1 [barely visible] to 4 [intense staining]) to score immunoblots. After performing independent observations, the investigators jointly reviewed discordant results and assigned final values after arriving at an agreement. Positive reactions to a particular antigen band were defined by scores of 2. Data were entered into EpiInfo (version 6.04, Centers for Disease Control and Prevention) and analyzed using the SAS system (version 6.11, SAS Institute). The frequencies of band recognition of sera from leptospirosis cases were compared with those of sera from healthy community controls using the chi-square test with Yates' correction. Logistic regression analysis was used to assign an order to the antigens and sequential combinations were graphed in a receiver-operator characteristic curve.

Cell fractionation studies. (i) Soluble and total membrane fractions. A leptospiral culture containing 4 × 10¹⁰ L. kirschneri isolates was washed twice in 5 mM MgCl₂-PBS at 4°C and resuspended in 6 ml of lysis buffer (20 mM Tris [pH 8]-150 mM NaCl-2 mM EDTA-2 mg of lysozyme per ml). The bacterial suspension was subjected to three cycles of freezing, thawing, and tip sonication, followed by centrifugation at 100,000 × g for 30 min to separate the soluble supernatant fraction from the membrane pellet fraction. The supernatant was precipitated with acetone.

(ii) Triton X-114 fractions. L. kirschneri organisms were also fractionated by solubilization with 1% Triton X-114 by a modification of the method described previously (16). In brief, a leptospiral culture containing 4 × 10¹⁰ L. kirschneri isolates was washed twice in 5 mM MgCl₂-PBS and extracted in the presence of 1% protein grade Triton X-114 (Calbiochem), 150 mM NaCl, 20 mM Tris (pH 8), and 2 mM EDTA at 4^°C. The insoluble material was removed by centrifugation at 17,000 × g for 10 min. After centrifugation, 20 mM CaCl₂ was added to the supernatant. Phase separation was performed by warming the supernatant to 37°C and subjecting it to centrifugation for 10 min at 1,000 × g. The detergent and aqueous phases were separated and precipitated with acetone.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Quantitative analysis of humoral immune response to leptospirosis. In initial one-dimensional SDS-PAGE and Western blot analyses, sera from patients with laboratory-confirmed leptospirosis (Group 1) recognized up to 25 distinct leptospiral antigen bands with M_r greater than 20 kDa from L. interrogans serovar copenhageni, the etiologic agent of urban epidemics in Salvador, Brazil (22). The predominant humoral response against these antigens during infection was IgG antibodies, regardless of whether samples were analyzed from the acute or convalescent phase of illness (mean interval and standard deviation between onset of illness and sample collection, 8.4 ± 4.2 and 32.1 ± 10.6 days, respectively) (Fig. 1). Although an IgM antibody response was consistently detected against low M_r species corresponding to leptospiral LPS, little or no detectable IgM response to higher M_r antigens was identified during the acute or convalescent phase of illness (Fig. 1). Two exceptions were observed: IgM reactivity to p37 and a doublet of antigen bands that has M_r values (35 to 36 kDa) consistent with those for leptospiral flagellar proteins (Fig. 1, lanes 2, 4, and 6).

View larger version (89K): [in this window] [in a new window]   FIG. 1.   Representative immunoblots of sera from patients with leptospirosis during urban epidemics in the city of Salvador, Brazil. SDS-PAGE (10% polyacrylamide) was used to separate antigen extracts of a clinical isolate of L. interrogans serovar copenhageni. Immunoblots were incubated with serum samples of individual patients (patients 235 [lane 1], 66 [lanes 2 and 3], 7 [lanes 4 and 5], 134 [lanes 6 and 7], 205 [lane 8], and 66 [lane 9]) and probed with either anti-IgM (IgM) or anti-IgG (IgG) secondary antibody conjugate. Lane 10 was incubated with serum from a healthy control individual. The locations of molecular mass markers (in kilodaltons) and selected leptospiral proteins are shown on the right and left, respectively. In addition, the positions of leptospiral proteins are marked () within the immunoblot and correspond to the following, in descending order: p76/82, p70, p62, p48, p45, the p41/42 complex, p37, p32, and p31 in lane 3; p44 and the p41/42 complex in lane 5; p58 and the p41/42 complex in lane 8; and the p41/42 complex and p25 in lane 9.

View larger version (16K): [in this window] [in a new window]   FIG. 2.   Receiver-operator characteristic curve showing the sensitivity and false-positive rate for immunoblot seroreactivity to combinations of leptospiral proteins. Sensitivity and 1-specificity were defined as the proportion of sera from confirmed cases and community control individuals, respectively, which recognized at least one leptospiral antigen in a combination. The best order of antigen combinations for acute-phase sera was as follows: p32 alone (a); p32 or p62 (b); p32, p62, or p76 (c); p32, p62, p76, or p41 (d); p32, p62, p76, p41, or p45 (e). For convalescent-phase sera the best order was as follows: p32 (A); p32 or p76 (B); p32, p76 or p41 (C); p32, p76, p41, or p45 (D); p32, p76, p41, p45, or p62 (E).

Conservation of protein antigens among leptospiral strains. Immunoblot analysis in Fig. 3 shows that pooled convalescent-phase sera from leptospirosis patients from Salvador, Brazil, recognized a consistent pattern of immunodominant antigens in leptospiral strains other than L. interrogans serovar copenhageni. Sera from patients who acquired leptospirosis in Barbados produced similar immunoblot reactivity (data not shown). Up to 14 distinct bands were detected in a single leptospiral strain. Among different serovars, small polymorphisms were observed in the M_r of particular immunodominant antigens such as the p37, p45, p48, and p58 proteins. The patterns of reactivity to these antigens fell into four classes. The p62 and p76 proteins were Class I antigens, detected in all organisms within the genus Leptospira, including pathogenic and nonpathogenic strains. There were at least five Class II antigens, p31, p32, p41/42, p45, and p82, which were consistently identified among all pathogenic strains. Class III antigens, which were found in most pathogens, included the p48 and p58 proteins and the high-molecular-weight p160 protein. Class IV antigens, such as p37 and p25, were expressed in one or only a few strains. On the basis of these immunoblot results, Leptospira inadai and Leptospira weilii appear to share few protein antigens (only the p32, p62, and p76 proteins) with other pathogenic Leptospira species. The expression of the immunodominant proteins by leptospiral strains is summarized in Table 3.

View larger version (101K): [in this window] [in a new window]   FIG. 3.   Immunoblot panel of Leptospira species probed with pooled convalescent-phase sera from leptospirosis patients in Salvador, Brazil. The locations of molecular mass markers (in kilodaltons) and selected leptospiral proteins are indicated to the right and left, respectively. Most proteins were conserved among pathogenic Leptospira species L. interrogans, L. kirschneri, L. borgpetersenii, L. noguchii, L. santarosai, L. weilii, and L. inadai. Two proteins, p62 and p76, which were identified as heat shock proteins GroEL and DnaK, respectively, were exceptions to this pattern. Their expression is conserved in all Leptospira species including the nonpathogens L. biflexa, L. meyeri, and L. wolbachii and the related organism Leptonema illini.

Identification of protein antigens by 2-D gel electrophoresis. A comparison of silver-stained 2-D gels prepared using pH 3 to 10 versus pH 4 to 7 Immobiline DryStrips showed that virtually all leptospiral proteins had isoelectric points within the pH 4 to 7 range (data not shown). In addition, there was better protein separation with 2-D gels prepared using pH 4 to 7 Immobiline DryStrips. On the basis of these preliminary experiments, a decision was made to use pH 4 to 7 Immobiline DryStrips in the 2-D immunoblot studies. Specific antisera were used to identify the location, in 2-D immunoblots, of DnaK, GroEL, OmpL1, LipL32, LipL36, LipL41, and LipL45/31. 2-D immunoblots of leptospiral strain L. interrogans serovar copenhageni from and endemic to Brazil or L. kirschneri serovar bim from and endemic to Barbados were probed with pooled convalescent-phase sera from leptospirosis patients from the same region. Significant differences were not observed with respect to the antigen patterns recognized by sera from patients of the two epidemiologically distinct regions. As shown in immunoblot analyses of serovar copenhageni antigens (Fig. 4 and Table 3), the electrophoretic mobilities of p76, p62, one antigen band within the p41/42 complex, and p32 allowed identification of these proteins as DnaK, GroEL, LipL41, and LipL32, respectively. Probing of 2-D immunoblots with pooled patient sera also demonstrated reactivity with the 33-kDa form of OmpL1 (Fig. 4). Reactivity of pooled patient sera to other previously characterized leptospiral proteins was either weak (LipL45/31) or not identified (LipL36). In addition, pooled patient sera reacted with several uncharacterized proteins, including the p45, p25, and p22 proteins. The latter two proteins were not well visualized in 1-D immunoblots (Fig. 1 and 3), probably because they were obscured by reactivity to leptospiral LPS antigens migrating in the lower-molecular-mass region of immunoblots. This demonstrates that in addition to providing definitive identification of protein antigens, a second advantage of the isoelectric focusing step of 2D immunoblots is the resultant separation of leptospiral protein antigens, such as the 25- and 22-kDa proteins, from LPS.

View larger version (40K): [in this window] [in a new window]   FIG. 4.   A two-dimensional immunoblot of L. interrogans serovar copenhageni proteins probed with pooled convalescent-phase sera from leptospirosis patients in Salvador, Brazil. The locations of DnaK, GroEL, LipL41, OmpL1, LipL32, and three as yet unidentified antigens (p45, p25, and p22) are indicated. The locations of molecular mass markers (in kilodaltons) and reference isoelectric points (pI) are indicated on the right and top of the figure, respectively.

Localization of protein antigens by leptospiral fractionation. We analyzed the behavior of protein antigens in two complementary leptospiral fractionation procedures. The first technique separated organisms into total membrane (cytoplasmic membrane and outer membrane) and soluble (cytoplasm and periplasm) fractions. The second technique involved separation of organisms into Triton X-114 soluble and insoluble fractions, followed by phase partitioning of the Triton X-114 soluble fraction into detergent (hydrophobic) and aqueous (hydrophilic) phases. Previous leptospiral fractionation studies have demonstrated that the Triton X-114 insoluble material consists of the protoplasmic cylinder, including the cytoplasm, cytoplasmic membrane, and peptidoglycan cell wall, including penicillin-binding and flagellar proteins (19). The Triton X-114 detergent phase has been shown to contain outer membrane components, including leptospiral LPS, OmpL1 (an outer membrane porin), and several lipoproteins, including LipL32 (the major outer membrane protein), LipL36, and LipL41, while Triton X-114 aqueous phase would be expected to contain soluble periplasmic proteins (16, 17, 37).

View larger version (96K): [in this window] [in a new window]   FIG. 5.   Localization of leptospiral protein antigens by fractionation of cellular components. Extracts of whole organisms (W) were used as starting material. The total membrane fraction (MP) contains both the cytoplasmic membrane and outer membrane, while the soluble fraction (MS) contains cytoplasmic and periplasmic material. The detergent Triton X-114 selectively releases the outer membrane, leaving the cytoplasm and cytoplasmic membrane in the Triton X-114 insoluble pellet (TP). Phase partitioning of the Triton X-114 extract separates periplasmic material in the aqueous phase (TA) from outer membrane components in the detergent phase (TD). The locations of molecular mass markers (in kilodaltons) and leptospiral protein antigens are shown on the left and right, respectively. In addition, the positions of leptospiral proteins are marked by arrows () within the immunoblot and correspond to, in descending order, p58, p48, p45, and p42 in the p41/42 complex in lane TP and LipL41, part of the p41/42 complex, and LipL45/31 in lane TD.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Pathogenic Leptospira species possess a number of protein antigens that are expressed during infection of mammalian hosts and become targets for the host immune response. The goals of this study were to perform quantitative and qualitative analyses of the protein antigens recognized by antibodies induced during human leptospirosis. The work presented here builds upon earlier immunoblot studies using leptospirosis patient sera from Australia and New Zealand (7) and from Barbados (8). Our efforts were facilitated by several important technical developments. First, significant improvements in the sensitivity and specificity of immunoblotting techniques have become available (24). Secondly, the formation of the Salvador Leptospirosis Study Group made it possible to obtain large numbers of well-characterized patient sera (22). Thirdly, many of the major leptospiral protein antigens have now been characterized on a molecular basis, and monospecific antibody reagents for these protein antigens have become available to assist in the interpretation of immunoblot studies (15-17, 37).

Our findings indicate that p32 is the immunodominant protein antigen recognized by the humoral response during natural infection. This conclusion is consistent with those from earlier immunoblot studies of sera from leptospirosis patients in Barbados, which identified a 32-kDa outer membrane protein as a major immunoreactive antigen (8). Our results extend those findings both in terms of frequency analysis based upon much larger numbers of sera and by the identification of p32 as the major outer membrane lipoprotein, LipL32. Convalescent-phase sera from Brazilian patients with confirmed leptospirosis reacted with LipL32 more frequently (84%) than with any other antigen (Table 1). Reactivity to p32 also demonstrated high specificity: 0 to 5% background reactivity was observed in all but one of the control groups, including those from regions with both high and low prevalence of leptospirosis (Table 1). The p32 antigen was consistently observed in all leptospiral pathogens tested (Fig. 3), and its identity as LipL32 was confirmed by 2D electrophoresis (Fig. 4).

These results are consistent with the recent characterization of LipL32 as an outer membrane lipoprotein, which is expressed at high levels by pathogenic Leptospira species (16). A comparison of LipL32 amino acid sequences from six leptospiral serovars, representing five leptospiral species, found a 97.8% average amino acid sequence identity. The high degree of LipL32 sequence identity indicates that serodiagnostic strategies based upon this antigen would be effective regardless of the infecting serovar. The specificity of this protein antigen for leptospiral infection is supported by BLAST searches of the GenBank database in which no significant homologues of the LipL32 sequence were identified. The antigenicity of LipL32 is presumably enhanced by its high level of expression in leptospiral pathogens and by lipid modification of its amino terminus, a property known to enhance the antigenicity of other spirochetal lipoproteins, such as OspA of Borrelia burgdorferi (9).

Two other prominent leptospiral antigens, p62 and p76, were identified in this study to be molecular chaperones GroEL and DnaK, respectively. Expression of bacterial heat shock proteins, including leptospiral GroEL and DnaK, is upregulated at the elevated temperatures encountered within the mammalian host (3, 4, 40). Both GroEL and DnaK are recognized by significant numbers of acute- and convalescent-phase sera from patients with confirmed leptospirosis (Table 1). In immunoblots with acute-phase sera, GroEL (45%) was the only antigen recognized more frequently than LipL32 (37%). However, only 16% of confirmed cases demonstrated seroconversion to GroEL between the acute and convalescent phases of illness (opposed to 50% for LipL32), suggesting that the immunoreactivity observed during acute-phase illness may have been due to preexisting, possibly cross-reactive, antibodies or a vigorous memory response. Furthermore, significant seroreactivity was observed among control sera to GroEL and, to a lesser degree, DnaK, probably reflecting the ubiquitous expression of these proteins in eubacteria (13) and the fact that many different types of infections are associated with an immune response to heat shock proteins (47). A recent study found that the dominant antigenic determinant in leptospiral GroEL is a 20-amino-acid region that is highly conserved among prokaryotes (33). This finding indicates that cross-reactivity with GroEL proteins from other bacteria could limit the feasibility of using leptospiral GroEL as a specific marker for leptospiral seroreactivity.

LipL41 was the fourth previously characterized antigen that we identified to be a target of the humoral immune response during leptospiral infection. Like LipL32, LipL41 is lipidated at its amino terminus and is located in the leptospiral outer membrane (37). A significant fraction of LipL41 appears to be exposed on the leptospiral surface, making it a potential target of a protective antibody response. When used in combination with OmpL1, immunization with recombinant lipidated LipL41 protects hamsters from challenge with virulent L. kirschneri (18). Our results indicate that in contrast to LipL32 and LipL41, other leptospiral lipoproteins would have limited usefulness in the serodiagnosis of human leptospirosis. For example, LipL36 is expressed by most leptospiral pathogens grown in culture, including the strains isolated from Salvador, Brazil, and Barbados; however, it was not detected by sera from patients acquiring leptospirosis in those locations in one- or two-dimensional immunoblot analyses (Fig. 4).

OmpL1 is a transmembrane outer membrane protein with porin activity which has been shown to be a protective immunogen (15, 18, 36). Although OmpL1 is expressed during mammalian infection (5), immunoblot reactivity using clinical leptospirosis sera could be demonstrated only by 2-D immunoblotting (Fig. 4). The difficulty in demonstrating OmpL1 reactivity could be due, in part, to this protein's unusual electrophoretic mobility pattern. In its undenatured form, OmpL1 migrates in SDS-PAGE with an apparent molecular mass of 25 kDa. In its denatured form OmpL1 migrates closer to its true molecular mass of 33 kDa (36). Neither the denatured nor undenatured form of native OmpL1 was detectable by one-dimensional immunoblotting with sera from leptospirosis patients (data not shown). At least two explanations could account for this result. First, reactivity with the 25- and 33-kDa forms of OmpL1 is likely to be obscured on immunoblots by reactivity with LPS and LipL32, respectively. Secondly, OmpL1 is expressed at low levels by Leptospira species, so there would be relatively less OmpL1 on immunoblots using native proteins.

An important advantage of the present study is that it uses sera from patients with naturally occurring leptospiral infections that probably result from relatively small infectious doses. For this reason, the immune response would be expected to exclusively target antigens expressed by leptospiral organisms within the mammalian host, not antigens expressed exclusively on environmental organisms at the time of inoculation. Therefore, recognition of lipoproteins LipL32 and LipL41 and heat shock proteins GroEL and DnaK by the humoral immune response to leptospirosis is a strong indication that these proteins are expressed during infection. These results confirm earlier immunoblotting and immunohistochemistry studies involving the hamster model of leptospirosis, which found that LipL32 and LipL41 are expressed by organisms within the proximal renal tubule, while LipL36 expression is down regulated during infection (5, 16). In those studies, efforts were made to avoid exposure to environmental organisms by inoculating hamsters with L. kirschneri obtained directly from infected hamster tissues. Immunoblots using sera from hamsters challenged with host-derived organisms recognized OmpL1, LipL41, and moieties that appear to be the p22, p37, and p45 antigens identified in this study (5). Interestingly, LipL32, GroEL, and DnaK were not well recognized by the hamster sera, suggesting that the results were biased by the artificial nature of experimental infection.

A second important advantage of studying sera from patients with clinical leptospirosis is the robust immune response, which allows identification of a much larger number of protein antigens than could be identified using infection-derived hamster sera (5). The diversity of recognized leptospiral protein antigens and the heterogeneous patterns of the antibody response observed among infected individuals are evident in Fig. 1 and 3. Furthermore, immunoreactive proteins were found to be shared among groups of genetically diverse leptospiral strains. The immunoblot pattern in Fig. 3 shows that most strains can be categorized as either pathogens or nonpathogens based upon the immunoreactive proteins which they express. Previous phylogenetic studies have indicated that L. inadai is an intermediate between leptospiral pathogens and nonpathogens (25). This observation was confirmed in the present study: among pathogen-specific protein antigens, only LipL32 was detectable in L. inadai.

There is an urgent need to address emerging epidemics of leptospirosis, particularly in medically underserved populations in developing countries, but surveillance and diagnosis have been hampered by the lack of an effective, widely available laboratory tool for case confirmation. The results of the present study serve as a guide to develop new strategies for serodiagnosis. The anti-LipL32 response was identified as the most important serologic marker of infection in immunoblot analyses. Seroreactivity against other leptospiral protein antigens did not significantly enhance the diagnostic sensitivity and specificity observed for the anti-LipL32 response alone (Table 2 and Fig. 2). LipL32 seroreactivity had sensitivity levels of 37 and 84% in detecting leptospiral infection during the acute and convalescent phases, respectively, of illness. In addition, an anti-LipL32 response was detected in 73 and 25% of convalescent-phase sera from probable (according to the MAT) and unconfirmed cases, respectively, of suspected leptospirosis, suggesting that LipL32 seroreactivity may be capable of capturing cases not identified by the standard laboratory confirmation method. The low frequency of reactivity in healthy individuals and patients with syphilis, hepatitis, and dengue from regions where leptospirosis is endemic (Table 1) indicates that the anti-LipL32 response is highly specific and therefore useful in differentiating leptospirosis from other causes of acute febrile illness. Application of more sensitive and rapid detection formats such as recombinant protein-based immunoassays will be the next step in evaluating the usefulness of this marker of infection for laboratory case confirmation in the field.

In addition to their use in serodiagnosis, leptospiral proteins expressed during mammalian infection may also have immunoprotective potential. The present study identified more than 20 immunoreactive proteins, several of which appear to be surface exposed and therefore serve as targets of a protective immune response. It has been demonstrated recently that immunization with whole leptospiral protein preparations confers protection in experimental animal models (39). In contrast to anti-LPS responses, those against leptospiral proteins were found to protect against challenge with heterologous, as well as homologous, leptospiral serovars. Furthermore, leptospiral outer membrane proteins OmpL1 and LipL41 have been shown to induce synergistic immunoprotection when expressed as membrane-associated recombinant antigens in E. coli (18). It is anticipated that additional leptospiral proteins identified in this study will be evaluated as immunoprotective antigens leading to the development of improved vaccines for the prevention of leptospirosis.