CLINICAL AND DIAGNOSTIC LABORATORY IMMUNOLOGY, Nov. 1997, p. 731–735 1071-412X/97/$04.0010 Copyright © 1997, American Society for Microbiology

Vol. 4, No. 6

Western Immunoblotting Analysis of the Antibody Responses of Patients with Human Monocytotropic Ehrlichiosis to Different Strains of Ehrlichia chaffeensis and Ehrlichia canis SHENG-MIN CHEN,1† LOUIS C. CULLMAN,2

AND

DAVID H. WALKER1*

Department of Pathology, University of Texas Medical Branch, Galveston, Texas 77555-0609,1 and MRL Diagnostics, Cypress, California 906302 Received 17 March 1997/Returned for modification 12 May 1997/Accepted 29 August 1997

of the 120- and 29/28-kDa proteins for diagnostic sensitivity and specificity, and to compare the sensitivity of Western immunoblotting and recombinant dot blot-enzyme-linked immunosorbent assay (ELISA) for the detection of antibodies to the 120-kDa protein.

Ehrlichia chaffeensis has been isolated from the blood of three patients with human monocytotropic ehrlichiosis (HME) (6, 7, 10). Prior to the first isolation (Arkansas strain) in 1991, Ehrlichia canis, the etiologic agent of canine tropical pancytopenia (canine monocytic ehrlichiosis), was employed as a surrogate antigen for serologic diagnosis by indirect immunofluorescence assay (IFA) (8, 9). Subsequently, nearly all patients have been diagnosed by seroconversion or demonstration of a single titer of 64 or greater by IFA with E. chaffeensis (Arkansas strain) as an antigen, a more sensitive method than E. canis IFA (7, 13, 14, 23). The major immunodominant antigens of E. chaffeensis are 120-, 66-, 58-, 55-, 29-, 28-, and 22-kDa proteins (4). A previous report suggested that detection of serum reactivity with E. chaffeensis (Arkansas strain) proteins of 28 or 29 kDa by Western immunoblotting was diagnostic of HME (3). There has been no comparison of the reactivity of sera from HME patients with the three strains of E. chaffeensis (Arkansas, 91HE17, and Sapulpa strains). A recent study demonstrated that detection of antibodies to the 120-kDa protein by a recombinant 120-kDa protein dot blot-enzyme immunoassay is potentially both sensitive and specific (23). Thus, this investigation of the reactivity of sera from HME patients and healthy controls with three strains of E. chaffeensis and one strain of E. canis was undertaken by Western immunoblotting and IFA. The purpose of the study was to determine the protein targets that serve as the basis for the IFA diagnosis of HME, to evaluate the four ehrlichial strains for differences in reactivity of the different proteins, to compare the detection

MATERIALS AND METHODS Cultivation and purification of ehrlichiae for Western immunoblotting. E. chaffeensis, Arkansas strain, was obtained from Jacqueline Dawson, Viral and Rickettsial Zoonoses Branch, Centers for Disease Control and Prevention, Atlanta, Ga. E. chaffeensis 91HE17 and Sapulpa were isolated in our laboratory from patients from Arkansas and Oklahoma, respectively (6, 10). E. canis, Oklahoma strain, was also kindly provided by Jacqueline Dawson. The ehrlichiae were cultivated in DH82 cells, a canine macrophage cell line, with Eagle minimum essential medium containing 10% bovine calf serum and 2 mM L-glutamine at 37°C. The cells were harvested when they were 90 to 100% infected. The ehrlichiae were purified from infected DH82 cell components by Renografin density gradient centrifugation as reported previously (4). DH82 antigens were prepared from uninfected cell monolayers. The protein concentration of the DH82 antigen was adjusted to the same concentration as the ehrlichial protein. Human sera. Twenty-seven patients’ sera were demonstrated to contain antibodies reactive with E. chaffeensis, Arkansas strain, by IFA at the Microbiological Reference Laboratory (Cypress, Calif.), which provides the major source of E. chaffeensis serologic diagnoses in the United States. Sera were obtained from patients in Arkansas, Texas, Missouri, Georgia, South Carolina, Kentucky, Indiana, Virginia, New Jersey, and New York. The available demographic data included gender (21 of 23 patients were male) and age (median, 55 years; range, 23 to 84). Clinical data were provided for 18 of the patients as follows: 7 (41%) of 17 patients had rashes, 15 (94%) of 16 patients were febrile (two serum samples were designated past infection), 15 (83%) of 18 patients reported headaches, 13 (76%) of 17 patients were thrombocytopenic, and 8 (47%) of 17 patients were leukopenic. Twelve negative control serum samples were obtained from blood collected by the American Red Cross in Los Angeles, Calif., and 15 negative control serum samples were obtained from Burlington, Vt., where no rickettsial diseases occur. Sera were stored at 2 to 8°C in 0.1% sodium azide prior to being shipped to Galveston, Tex. Subsequently the sera were stored at 220°C. IFA. The antigen slides used for IFA were prepared with DH82 cells infected with E. chaffeensis. The slides were fixed in acetone for 10 min at room temperature and stored at 4°C with dessicant before use. The human sera were diluted in phosphate-buffered saline (PBS). Fluorescein isothiocyanate-conjugated goat anti-human heavy chain-specific immunoglobulin G (Jackson Immunoresearch

* Corresponding author. Mailing address: Department of Pathology, University of Texas Medical Branch, 301 University Blvd., Galveston, TX 77555-0609. Phone: (409) 772-2856. Fax: (409) 772-2500. E-mail: [email protected]. † Dr. Sheng-Min Chen, the primary author of this paper, was killed tragically in an automobile accident on 12 January 1997. 731

Downloaded from http://cvi.asm.org/ on June 8, 2018 by guest

In order to evaluate the relative sensitivity of the detection of antibodies against various antigenic proteins of Ehrlichia chaffeensis for the diagnosis of the emerging infectious disease human monocytotropic ehrlichiosis, Western immunoblotting was performed with 27 serum samples from convalescent patients with antibodies, as demonstrated by indirect immunofluorescence assay. Among 22 patients with antibodies reactive with the 120-kDa protein, 15 showed reactivity with the 29/28-kDa protein(s) and the proteins in the 44- to 88-kDa range. Two of the serum samples with this pattern reacted with the 29/28-kDa protein(s) of only the 91HE17 strain, and one sample reacted with only that of the Arkansas strain, indicating that the antibodies were stimulated by strain-specific epitopes. Overall, antibodies to the 29/28-kDa protein(s) were detected in only 16 patients’ sera, suggesting that this protein is less sensitive than the 120-kDa protein. Two of 12 serum samples from healthy blood donors had antibodies reactive with the 120-kDa protein; one of these samples reacted also with the 29/28-kDa protein(s) of Ehrlichia canis, suggesting that unrecognized ehrlichial infection might have occurred, including human infection with E. canis. A high correlation between reactivity with the 120-kDa protein by Western immunoblotting and the recombinant 120-kDa protein by dot blot supports the potential usefulness of this recombinant antigen in diagnostic serology.

732

CHEN ET AL.

CLIN. DIAGN. LAB. IMMUNOL.

TABLE 1. Western immunoblot reactivity of serum samples from 27 patients with IFA-confirmed E. chaffeensis infection Result for protein:

No. of serum samples

120 kDa

12 2 1 1 1 5 1 1 1 2

1 1 1 2 2 1 1 1 2 2

a

29/28 kDa

Geometric mean titer

1 1 (91HE17 only) 1 (Arkansas only) 1 2 2 2 2 2 2

1,024 512 1,024 1,024 512 257 256 1,024 256 362

44 to 88 kDa

1 1 1 1 (55 1 (58 1 1 (44 1 (58 2a 2

and 88 kDa only) kDa only) kDa only) kDa only)

Reactive only with a 38-kDa ehrlichial protein.

FIG. 1. Western immunoblotting of sera from two patients with HME (A and B) and one healthy blood donor (C) with the antigens of E. chaffeensis Arkansas (lanes a), 91HE17 (lanes b), and Sapulpa (lanes c), E. canis (lanes d), and uninfected DH82 cells (lanes e). The lanes labeled MW are molecular weight standards (in thousands).

of 64 or greater is considered the diagnostic “gold standard,” Western immunoblotting detection of reactivity with the 120kDa band had a sensitivity of 81% and a specificity of 83%, and reactivity with the 29/28-kDa band(s) of E. chaffeensis had a sensitivity of 59% and a specificity of 100% (Tables 2 and 3). Comparison of reactivity with the 120-kDa protein by Western immunoblotting and by recombinant polypeptide dot blotELISA showed a strong concordance (Table 2), with 20 serum samples yielding positive results and 2 serum samples yielding negative results by both assays of the 27 IFA-positive serum samples.

RESULTS Serum samples from the 27 patients diagnosed as having HME by IFA showed several different patterns of reactivity with the protein bands of E. chaffeensis (Table 1). The most frequently observed pattern, reactivity with the 120-, 44- to 88-, and 29/28-kDa proteins, was observed in 15 patients (Fig. 1). Five patients’ serum samples reacted with the 120- and 44- to 88-kDa proteins but not with the 29/28-kDa protein(s) (Fig. 2). One patient’s serum reacted with the 120-kDa protein and only the 44-kDa protein, and another patient’s serum reacted with the 120-kDa protein and only the 58-kDa protein. Two of the patients’ sera did not react with any of the ehrlichial proteins by Western immunoblotting, and at the time of study in the research laboratory in Galveston, these two serum samples were no longer reactive by IFA at a dilution of 64. One patient had reactivity with the 55-, 88-, and 29/28-kDa proteins, and two other patients’ sera reacted only with the 58-kDa or 38kDa protein, respectively. Twenty-five of the 27 control sera were negative by immunoblotting. One healthy donor’s serum reacted with the 120- and 44- to 88-kDa proteins, but not with the 29/28-kDa protein(s). Another control serum reacted with the 120-kDa proteins of E. chaffeensis and E. canis and the 28-kDa protein of E. canis, but not with the 29/28-kDa proteins of E. chaffeensis (Fig. 3). If the reference laboratory IFA result of reactive at a titer

FIG. 2. Western immunoblot of serum from a patient with HME with the antigens of E. chaffeensis Arkansas (lane a), 91HE17 (lane b), and Sapulpa (lane c), E. canis (lane d), and uninfected DH82 cells.

Downloaded from http://cvi.asm.org/ on June 8, 2018 by guest

Lab, Inc., West Grove, Pa.) with Evans blue counterstain was used at a dilution of 1:50 in PBS. An antibody titer of 1:64 or greater was considered positive. Western immunoblotting. Sodium dodecyl sulfate (SDS)-polyacrylamide slab gel electrophoresis was performed by a modification of the method described by Laemmli (15), with a 4% stacking gel and a 10% separating gel. The purified ehrlichial antigens and DH82 cell antigens were dissolved in final sample buffer (5% 2-mercaptoethanol–10% glycerol–2% SDS–0.08% bromophenol blue in 62.5 mM Tris buffer [pH 6.8]) at room temperature without being subjected to heat denaturation. An aliquot containing 25 mg of protein was loaded into each lane. Biotinylated SDS-polyacrylamide gel electrophoresis high-range and lowrange molecular mass standards were the product of Bio-Rad Laboratories, Richmond, Calif. Electrophoretic separation of ehrlichial or DH82 cell proteins was performed at 18 mA for the stacking gel and 36 mA for the separating gel. Electrophoretic transfer of the proteins in the SDS-polyacrylamide gel to nitrocellulose sheets was performed at 195 mA for 3 h at 4°C. Nitrocellulose membranes (0.22-mm pore size; Micron Separations Inc., Westboro, Mass.) were incubated overnight in blocking buffer consisting of 5% (wt/vol) nonfat dry milk in PBS at 4°C. The blots were then incubated with human sera at room temperature for 1 h. The immune sera were diluted 1:150 in blocking buffer. After being washed in three changes of PBS containing 0.1% Tween-20 (10 min each time), the immunoblots were incubated at room temperature for 1 h with peroxidaselabeled goat anti-human immunoglobulin G (Kirkegaard & Perry, Gaithersburg, Md.) diluted 1:2,500 in PBS. The blots were washed again as described above. Peroxidase activity was demonstrated with 3,39-diamino-benzidine as the substrate.

VOL. 4, 1997

ANTIBODY RESPONSES OF PATIENTS WITH HME

733

TABLE 3. Comparison of detection of antibodies to Ehrlichia chaffeensis by IFA and by Western immunoblotting for the 29/28-kDa protein(s) Result for: No. of patientsa

IFA

Western blot assay for 29/28-kDa protein(s)

16 11 0b

1 1 2

1 2 1

a

Twenty-seven healthy control subjects were negative by both assays. There were no cases in which the Western blot results were positive when the IFA results were negative. b

DISCUSSION FIG. 3. Western immunoblots of sera from two healthy subjects (A and B) with the antigens of E. chaffeensis Arkansas (lanes a), 91HE17 (lanes b), and Sapulpa (lanes c), E. canis (lanes d), and uninfected DH82 cells (lanes e). The lanes labeled MW are molecular weight standards (in thousands).

IFA serologic results with antigens of each strain of E. chaffeensis did not indicate any sera that reacted with one or more strains of E. chaffeensis and yet failed to react with the other strains at the diagnostically significant cutoff titer of 64. In general, this cross-reactivity was reflected in the similarity of Western immunoblotting results with individual sera and the three E. chaffeensis strains. There were three exceptions to this general rule, particularly regarding reactivity with the 29/28kDa protein(s). One HME patient’s serum reacted with the 29/ 28-kDa protein(s) of only the Arkansas strain, and two patients’ sera reacted with the 29/28-kDa protein(s) of only strain 91HE17 (Fig. 4). Often a patient’s serum reacted either with the 28- or with the 29-kDa protein or reacted strongly with one protein and weakly with the other. Previous studies suggested that the 28- and 22-kDa proteins contain common antigens (5). Our results showed that reactivity with the 22-kDa protein was not observed in the absence of reactivity with the 28-kDa protein. When reactivity with the 28-kDa protein was present, reactivity with the 22-kDa protein band may or may not be observed. Sera of some patients reacted intensely with the 22-kDa proteins of certain strains of E. chaffeensis, while reactivity with the 22-kDa proteins of

The human humoral immune response to the electrophoretically separated proteins of E. chaffeensis has been reported previously for 20 patients (3, 4, 19). Our immunoblots have demonstrated reactivity with a 120-kDa protein of E. chaffeensis, Arkansas and 91HE17 strains, and the analogous 97-kDa protein of the Sapulpa strain in 24 of 29 patients’ sera examined in this and previous studies (4, 6). Neither Rikihisa et al. nor Brouqui et al. described the detection of an immunoreac-

TABLE 2. Comparison of IFA serology for E. chaffeensis and antibodies to the 120-kDa protein determined by Western immunoblotting and recombinant dot blot–ELISA No. of patients

IFA

Western blot assay

Protein dot blot–ELISA

20 2 3 2

1 1 1 1

1 1 2 2

1 2 1 2

FIG. 4. Western immunoblots of sera from two patients (A and B) with HME that are reactive with a protein of 29/28-kDa of only E. chaffeensis Arkansas (A, lane a) and 91HE17 (B, lane b). Lanes: MW, molecular weight standards (in thousands); a, Arkansas strain; b, 91HE17 strain; c, Sapulpa strain; d, E. canis; and e, uninfected DH82 cells.

Downloaded from http://cvi.asm.org/ on June 8, 2018 by guest

other strains was weak or even absent. In some instances of variability among the reactivities of the 22-kDa proteins of different strains with sera, the 28-kDa proteins of all the different strains of E. chaffeensis and E. canis were equally reactive, suggesting the presence of an epitope(s) on the 22-kDa protein which is not shared by the 22-kDa proteins of the other strains.

734

CHEN ET AL.

have HME, the diagnostic sensitivity of the immunoblotting detection of antibodies to the 120- and 29/28-kDa proteins would be 95 and 70%, respectively. This situation further emphasizes the current lack of a sensitive gold standard for diagnosis. Although most of the patients’ sera that were reactive with the 120-, 29-, or 28-kDa protein cross-reacted with the analogous proteins of E. canis and with all three strains of E. chaffeensis, and all sera that were positive by IFA at the diagnostically significant cutoff titer of 64 for the Arkansas strain of E. chaffeensis were also positive for the other two strains, three patients’ 29/28-kDa protein immunoblotting results suggested infection with a particular strain of E. chaffeensis. The use of immunoblotting with a different strain of E. chaffeensis with the criterion of 29/28-kDa protein reactivity would have classified these three patients as not having HME, which clearly was not the case. This situation emphasizes the need to define further the antigenic diversity of E. chaffeensis as well as to validate the diagnostic methods by the comparison of multiple assays for substantial numbers of clinically and epidemiologically wellcharacterized patients. ACKNOWLEDGMENTS We thank Josie Ramirez for expert assistance in the preparation of the manuscript and Thomas Bednarek for assistance in the preparation of the illustrations. This study was supported by a grant from the National Institute of Allergy and Infectious Diseases (AI31431). REFERENCES 1. Anderson, B. E., J. W. Sumner, J. E. Dawson, T. Tzianabos, C. R. Greene, J. G. Olson, D. B. Fishbein, M. Olsen-Rasmussen, B. P. Holloway, E. H. George, and A. F. Azad. 1992. Detection of the etiologic agent of human ehrlichiosis by polymerase chain reaction. J. Clin. Microbiol. 30:775–780. 2. Barenfanger, J., P. G. Patel, J. S. Dumler, and D. H. Walker. 1996. Identifying human ehrlichiosis. Lab. Med. 27:372–374. 3. Brouqui, P., C. Lecam, J. Olson, and D. Raoult. 1994. Serologic diagnosis of human monocytic ehrlichiosis by immunoblot analysis. Clin. Diagn. Lab. Immunol. 1:645–649. 4. Chen, S.-M., J. S. Dumler, H.-M. Feng, and D. H. Walker. 1994. Identification of the antigenic constituents of Ehrlichia chaffeensis. Am. J. Trop. Med. Hyg. 50:52–58. 5. Chen, S.-M., V. L. Popov, H.-M. Feng, and D. H. Walker. 1996. Analysis and ultrastructural localization of Ehrlichia chaffeensis proteins with monoclonal antibodies. Am. J. Trop. Med. Hyg. 54:405–412. 6. Chen, S.-M., X.-J. Yu, V. L. Popov, E. L. Westerman, F. G. Hamilton, and D. H. Walker. 1997. Genetic and antigenic diversity of Ehrlichia chaffeensis: comparative analysis of a novel human strain from Oklahoma and previously isolated strains. J. Infect. Dis. 175:856–863. 7. Dawson, J. E., B. E. Anderson, D. B. Fishbein, J. L. Sanchez, C. S. Goldsmith, K. H. Wilson, and C. W. Duntley. 1991. Isolation and characterization of an Ehrlichia sp. from a patient diagnosed with human ehrlichiosis. J. Clin. Microbiol. 29:2741–2745. 8. Dawson, J. E., Y. Rikihisa, S. A. Ewing, and D. B. Fishbein. 1991. Serologic diagnosis of human ehrlichiosis using two Ehrlichia canis isolates. J. Infect. Dis. 163:564–567. 9. Dawson, J. E., D. B. Fishbein, T. R. Eng, M. A. Redus, and N. R. Greene. 1990. Diagnosis of human ehrlichiosis with the indirect fluorescent antibody test: kinetics and specificity. J. Infect. Dis. 162:91–95. 10. Dumler, J. S., S. Chen, K. Asanovich, E. Trigiani, V. L. Popov, and D. H. Walker. 1995. Isolation and characterization of a new strain of Ehrlichia chaffeensis from a patient with nearly fatal monocytic ehrlichiosis. J. Clin. Microbiol. 33:1704–1711. 11. Dumler, J. S., J. E. Dawson, and D. H. Walker. 1993. Human ehrlichiosis: hematopathology and immunohistologic detection of Ehrlichia chaffeensis. Hum. Pathol. 24:391–396. 12. Dunn, B. E., T. P. Monson, J. S. Dumler, C. C. Morris, A. B. Westbrook, J. L. Duncan, J. E. Dawson, K. G. Sims, and B. E. Anderson. 1992. Identification of Ehrlichia chaffeensis morulae in cerebrospinal fluid mononuclear cells. J. Clin. Microbiol. 30:2207–2210. 13. Everett, E. D., K. A. Evans, R. B. Henry, and G. McDonald. 1994. Human ehrlichiosis in adults after tick exposure. Ann. Intern. Med. 120:730–735. 14. Fishbein, D. B., J. E. Dawson, and L. E. Robinson. 1994. Human ehrlichiosis in the United States, 1985 to 1990. Ann. Intern. Med. 120:736–743. 15. Laemmli, U. K. 1970. Cleavage of structural proteins during the assembly of

Downloaded from http://cvi.asm.org/ on June 8, 2018 by guest

tive 120-kDa protein in 17 HME patients’ sera (3, 19). It is presumed that the explanation is the apparently smaller quantity of this protein relative to those of proteins in the 44- to 88-kDa range and proteins in the 20- to 30-kDa range. Not only was detection of antibodies to the 120-kDa protein more sensitive than that to the 20- to 30-kDa proteins (81 and 59%, respectively) but also a recombinant 120-kDa polypeptide has previously demonstrated 86% sensitivity and 100% specificity in the dot blot-ELISA format (23). It appears that the 120-kDa protein is at least as good a candidate (if not better) for Western immunoblot confirmation of the serologic diagnosis as the proteins in the 20- to 30-kDa group. Indeed, a higher IFA titer (geometric mean titer, 939) was observed among the sera containing antibodies to the 29/28-kDa protein(s) than among those containing antibodies to the 120-kDa protein (geometric mean titer, 645). This correlation suggests that the 120-kDa protein might be more immunogenic or that antibodies to the 120-kDa protein might appear earlier in the course of illness. The latter hypothesis must be tested in a prospective clinical study with a complete collection of data regarding the onset of illness. If detection of antibodies to the 120-kDa and the 29/ 28-kDa proteins is a valid criterion for confirmation of the diagnosis of HME by Western immunoblotting, then two of the healthy subjects who were investigated likely had a subclinical or undiagnosed infection with E. chaffeensis, E. canis, or an antigenically related organism, such as Ehrlichia ewingii. Indeed, asymptomatic human infection with E. canis and asymptomatic seroconversion to the antigens of E. chaffeensis have been described previously (18, 21). The fundamental problem in the evaluation of the sensitivity and specificity of diagnostic assays for HME is the lack of a sensitive gold standard diagnostic method. It is unfortunate that the ideal standard, cultivation of E. chaffeensis, is so insensitive as well as laborious and slow (6, 7, 10). PCR with E. chaffeensis-specific primers for the 16S rRNA gene or primers for other genes such as the 120-kDa protein offers an alternative specific diagnostic approach, but the potential occurrence of false-positive results due to DNA contamination with amplicons from other sources in the laboratory makes it less than an absolute gold standard assay. The report of Everett et al. that 7 (35%) of 20 HME patients diagnosed by 16S rDNA PCR did not develop IFA antibodies to E. chaffeensis 5, 10, 14, 16, 26, 33, and 35 days after onset of illness suggests that IFA serology might be relatively insensitive (13). In a study of five HME patients diagnosed by E. chaffeensis IFA serology and six controls, Anderson et al. observed that the 16S rDNA PCR had a sensitivity of 100% and a specificity of 100% (1). Although most HME sera were reactive in Western immunoblots with shared antigens of E. canis, 4 of 30 previously reported patients’ sera that were positive by IFA for antibodies to E. chaffeensis were negative for E. canis (7, 13). Indeed, IFA with E. canis (Oklahoma strain) detected only 88% of sera identified by IFA with E. canis (Florida strain) (8). Even lesssensitive diagnostic approaches are the identification of morulae in peripheral blood smears (only 7%) or cerebrospinal fluid and the immunohistologic detection of E. chaffeensis in bone marrow biopsy (40%) (2, 11–13, 16, 17, 20). The results of this investigation also suggest that IFA serology may detect antibodies that were not stimulated by E. chaffeensis. Two IFA-diagnosed patients’ sera did not react with any protein bands of E. chaffeensis or E. canis. Furthermore, the patient whose serum reacted only with a 58-kDa protein, presumably the GroEL analog, and the patient whose serum reacted only with an unidentified 38-kDa band are unlikely to have been infected with an organism of the E. chaffeensisE. canis group. If these four patients were considered not to

CLIN. DIAGN. LAB. IMMUNOL.

VOL. 4, 1997 the head of bacteriophage T4. Nature 227:680–685. 16. Maeda, K., N. Markowitz, R. C. Hawley, M. Ristic, D. Cox, and J. E. McDade. 1987. Human infection with Ehrlichia canis, a leukocytic rickettsia. N. Engl. J. Med. 316:853–856. 17. Paddock, C. D., D. P. Suchard, K. L. Grumbach, W. K. Hadley, R. L. Kerschmann, N. W. Abbey, J. E. Dawson, B. E. Anderson, K. G. Sims, J. S. Dumler, and B. G. Herndier. 1993. Brief report: fatal seronegative ehrlichiosis in a patient with HIV infection. N. Engl. J. Med. 329:1164–1167. 18. Perez, M., Y. Rikihisa, and B. Wen. 1996. Ehrlichia canis-like agent isolated from a man in Venezuela: antigenic and genetic characterization. J. Clin. Microbiol. 34:2133–2139. 19. Rikihisa, Y., S. A. Ewing, and J. C. Fox. 1994. Western immunoblot analysis of Ehrlichia chaffeensis, E. canis, or E. ewingii infections in dogs and humans. J. Clin. Microbiol. 32:2107–2112.

ANTIBODY RESPONSES OF PATIENTS WITH HME

735

20. Roland, W. E., G. McDonald, C. W. Caldwell, and E. D. Everett. 1995. Ehrlichiosis—a cause of prolonged fever. Clin. Infect. Dis. 20:821–825. 21. Yevich, S. J., J. L. Sanchez, R. F. DeFraites, C. C. Rives, J. E. Dawson, I. J. Uhaa, B. J. B. Johnson, and D. B. Fishbein. 1995. Seroepidemiology of infections due to spotted fever group rickettsiae and Ehrlichia species in military personnel exposed to areas of the United States where such infections are endemic. J. Infect. Dis. 171:1266–1273. 22. Yu, X.-J., P. Brouqui, J. S. Dumler, and D. Raoult. 1993. Detection of Ehrlichia chaffeensis in human tissue by using a species-specific monoclonal antibody. J. Clin. Microbiol. 31:3284–3288. 23. Yu, X.-J., P. Crocquet-Valdes, L. C. Cullman, and D. H. Walker. 1996. The recombinant 120-kilodalton protein of Ehrlichia chaffeensis, a potential diagnostic tool. J. Clin. Microbiol. 34:2853–2855.

Downloaded from http://cvi.asm.org/ on June 8, 2018 by guest