Immunobiol., vol. 185, pp. 453-465 (1992)

1 Institute

of Clinical Microbiology, University of Erlangen-Nurnberg, Erlangen, and of Electron Microscopy, University of Bayreuth, Bayreuth, Germany

2 Department

Differences of Two Borrelia burgdorferi Strains in Complement Activation and Serum Resistance VOLKER BRADE1, INGRID KLEBER 1, and GEORG ACKER2 Received December 4, 1991 . Accepted in Revised Form March 3, 1992

Abstract Complement activation and serum resistance of the Borrelia burgdorferi strains B31 (American strain) and PKo (European strain) were compared. In 25 % (v/v) normal human serum (NHS) free of B. burgdorferi-specific antibodies the cells of the PKo strain were high activators of complement as indicated by rapid and strong C9 consumption, by deposition of up to 336763 C9 molecules per cell and by the formation of the terminal complement complex on the cell surface. By comparison, complement activation by the B31 strain was low with 5.4fold less C9 deposited per cell. The addition of B. burgdorferi-specific antibodies to NHS either as purified IgG or heat-inactivated patient sera, had no influence on the results with both strains. After an incubation period of 2 h at 37°C in 25 % (v/v) NHS most cells of the PKo strain had lost their viability as indicated by cell immobilization and failure to multiply in subcultures. In addition, extensive cell fragmentation and bleb formation were observed in the electron microscope. In contrast, the B31 strain remained alive and morphologically intact after the same incubation with NHS. We conclude from our results that complement activation and serum resistance are properties which differ considerably between isolated strains of B. burgdorferi.

Introduction Borrelia burgdorferi has been identified as being a cause of a disease known as Lyme borreliosis which primarily involves the skin, the central nervous system and the joints (reviewed in 1). The outbreak of this disease may occur months or even years after infection, which suggests that the organisms can persist in the infected host. Indeed, it was possible to grow B. burgdorferi in culture medium from material obtained from patients with a late manifestation of the disease (2). Survival and multiplication of the bacteria in the infected host may be facilitated by many mechanisms including resistance against natural defense systems, a delayed immune response or resistance against effector mechanisms built up during the immune response. With respect to humoral factors, it was described that B. burgdorferi is resistant against serum in the absence of specific antibody (3).

454 . V. BRADE, 1. KLEBER, and G. ACKER

Borrelia burgdorferi-specific antibodies are formed very slowly after infection which gives the bacteria a chance of multiplication. However, once formed, antibodies seem to playa protective role. This was concluded from successful passive immunization experiments in animal models (4,5) as well as from in vitro studies in which sera reduced the mobility and viability of B. burgdorferi (6,8). Based on these data, it was claimed that specific antibodies and complement are harmful to B. burgdorferi (8). The published work, however, does not allow to conclude that B. burgdorferi strains are uniformly sensitive to complement in the presence of antibody. Recent studies have demonstrated a great heterogeneity within this bacterial species concerning its protein profile and genetic background (9, 10). In our present study we add new information on the heterogeneity within the species B. burgdorferi by demonstrating complement sensitivity or resistance in two different strains. Materials and Methods 1. Diluents Isotonic veronal-(barbital- )buffered saline (VBS) and isotonic VBS-dextrose (VBS-G) were prepared according to the methods described by RAPP and BORSOS (11). These buffers contained 0.1 'Yo/gelatin, 1 mM Mg2+ and 0.15 mM Ca2+.

2. Borrelia burgdorferi strains and growth conditions In all experiments the B. burgdorferi strains B31 (ATCC 35210) and PKo (12) were used. Both strains have been maintained in our laboratory for more than two years. A stock of both bacteria was kept frozen at -70°C in BSK medium (12). The protein profile of the B31 and PKo strains according to SDS-PAGE analysis is shown in Figure 1. For all experiments, samples were thawed and cultured at 33°C in BSK medium (12 ml tubes) for approximately one week. Subsequently each strain was subcultured 3-5 times for the same period of time in order to obtain highly mobile and viable bacteria. The bacteria cultured in this manner were then centrifuged for 10 min at 1500 x g, suspended in buffer (VBS-G), placed on a Neubauer counting chamber (0.02 mm depth) and counted by darkfield microscopy. The final bacterial suspension was adjusted to contain approximately 5 x 10 8 microorganisms/ml VBS-G. In this buffer, the viability and mobility of B. burgdorferi was maintained for at least 2 h at 33°C as well as at 37 0c.

3. Source of human serum and Borrelia burgdorferi-specific IgG A pool of normal human serum (NHS) was prepared from serum samples which were free of antibodies against B. burgdorferi according to negative results with the indirect immunofluorescence test (1FT). Samples of this pool were stored at -70°C. Serum from a patient with stage 3 Lyme borreliosis (Acrodermatitis chronica atrophicans) served as a source of purified IgG antibodies. After absorption with Treponema phagedenis this serum had a titer of 1:4096 in the 1FT. For preparation of IgG, serum samples (2 ml) were run through a protein G Sepharose 4 (Pharmacia) column. Eluted IgG was adjusted to 1.25 mg protein/ml VBS. The 1FT titer of this preparation was 1:64. Six other sera from patients with Lyme borreliosis were used in addition as antibody source after heat inactivation (30 min at 56°C) and filter sterilization. These sera were from three patients with arthritis (1FT titers 1:256, 1:512 and 1:1024), from two patients with meningitis (1FT titers 1:128 and 1:512) and from one patient with Erythema chronicum migrans (1FT titer 1:20). All sera were tested in the 1FT after absorption with Treponema phagedenis.

Complement Activation by Borrelia burgdorferi Strains . 455

MW

-

MW

-- 69K---.

...-60K

58K---'

40K-' -

...- 41K(flagellin)

-33K (Osp B) ...-31K (Osp A)

26K-'

B31 PKo Figure 1. SDS-PAGE of the B. burgdorferi strains B31 and PKo. Samples containing 5 lAg B. burgdorferi protein were subjected to 5DS-PAGE (10% polyacrylamide) under reducing conditions followed by silver staining. Arrows on the left side indicate major differences between the B31 and PKo strains. 4. Purification and radioiodination of C9

C9 purified as described by BIESECKER and MOLLER-EBERHAR[) (13) was radiolabeled with Na l2sJ us ing Jodobeads (Pierce) to a specific activity of 4 x lOs to 1.5 x 106 cpm/ lAg C9.

5. C9 consumption and C9 deposition B. burgdorferi grown as described were incubated in the followin g manner. Sample one contained 0.1 ml B. burgdorferi (5 x 107 microorganisms), 0. 1 ml NHS, 0.2 ml VBS-G and 0. 28 lAg radiolabeled C9. The second sample differed from sample one by addition of 0.1 IAI B. burgdorferi-specific IgG (125 lAg) instead of 0.1 ml VBS-G . Controls included tubes 3 and 4 in which NHS was replaced by CH56 and tubes 5 nad 6 in which B. burgdorferi was replaced by VBS-G. The controls served for the calculation of specific C9 binding (tubes 3 and 4) and C9 consumption (tubes 5 and 6), respectively. Incubation of all six samples was performed for 2 h at 3rc fo llowed by immediate centrifugation. The supernatants were collected and assayed for functi onal C9 activity (14). Bacterial pellets were washed twice in YBS and resuspended in 50 IAI VBS. One half was used to measure pellet-associated radioactivity in a gamma scintilla-

456 . V. BRADE, 1. KLEBER, and G. ACKER tion counter (Auto Gamma 500, Packard Instr. Co. Inc.). The total number of bound C9 molecules (labeled plus unlabeled) was calculated according to JOINER et al. (15) with CH56 as the control. The unspecific binding of radioactive C9 in CH56 was between 1 and 2 % of the total input counts and was subtracted from all samples. The second half was diluted with the same volume of SDS sample buffer for subsequent SDS-PAGE and autoradiography. 6. SDS-PAGE and autoradiography Polyacrylamide gels (10 %) were prepared according to the method of LAEMMLI (16). For application to the gel, samples were boiled for 7 min in SDS sample buffer containing 10 % ~­ mercaptoethanol and electrophoresed at 60 rnA. For analysis of the protein profile of B. burgdorferi 5 flg bacterial protein was applied to each lane. After the electrophoretic separation, the gels were subjected to silver staining (17). For autoradiography the gels were dried with a slab gel drier after electrophoresis of the sample. Autoradiography was performed at -70°C with Kodak XAR-5 film. 7. Bactericidal assay Test samples of 0.4 ml contained either 0.1 ml B. burgdorferi (5 x 10 7 microorganisms), 0.1 ml NHS and 0.2 ml VBS-G, or 0.1 ml B. burgdorferi, 0.1 ml NHS, 0.1 ml VBS-G and 0.1 ml specific antibodies (either purified IgG (125 flg) or undiluted, heat-inactivated patient serum). Two additional controls contained CH56 instead of NHS. All samples were incubated for 2 h at 37°C. After incubation, all tubes were centrifuged for 2 min at 9980 x g and the pellets were resuspended in 1.4 ml BSK. After vigorous mixing, a small portion (50 fll) was used for darkfield microscopy where the total cell number was counted as an average of four random fields. Furthermore, the percentage of mobile cells and the frequency of blebs and fragments were assessed for each sample. The major portion (1.35 ml) of the bacterial suspension was added to 10.65 ml fresh BSK medium and further incubated for 48 hat 33°C. Evaluation of all samples was then performed in the same manner as that following the 2 h incubation period. 8. Electron microscopy For thin-section electron microscopy, cells of the PKo strain and of the strain B31 were incubated for 2 h at 37°C in NHS or in CH56 (control), as described in the previous paragraph. Serum-treated samples were fixed with glutaral-dehyde-Os04 and embedded in Spurr's resin as described (18). Thin sections were cut with an ultramicrotome, model ULTRACUT (Reichert-Jung, Vienna, Austria), and mounted on Formvar-carbon-coated copper grids. The sections were stained with 2 % (w/v) aqueous uranyl acetate and lead citrate (19) and examined in a TEM, EM 109 (Zeiss, Oberkochen, Germany), at 80 kV.

Results

1. C9 consumption We first explored whether the PKo and B31 strains differed with respect to C9 consumption. These studies revealed very strong C9 consumption by

the PKo strain after a 2 h incubation period. Under the same conditions, the B31 strain consumed about 2-fold less C9 (Table 1). The presence of specific IgG in the assay did not enhance C9 consumption by the B31 strain.

Com plement Activation by Borrelia burgdorferiStrains

457

Table 1. C9 consumption and C9 deposition with B. burgdorferi strains B31 and PKo

B31 B31/IgG PKo PKo/ lgG

C9 consumption I (%)

C9 deposition1 (C9 molecules/cell x 10- 3 )

45.5±14.0 44.0 ± 7.8 94.8 ± 5.3 98.6 ± 1.2

54.9 ± 7.2 61.8 ± 4.3 277.4 ± 66.0 277.2 ± 69.9

Incubation of 0. 1 ml B. burgdorferi (5 x 107 cells) in 0.1 ml NHS and 0.2 ml VBS-G or 0. 1 ml VBS-G plus 0.1 ml specific IgG (125 fig) for 2 h at 37 °C. Haemoly tic C9 determined in the supernatants (values represent mean ± SD of 4 experiments). 1 Incubation of 0.1 ml B. burgdorferi (5 x 10 7 cells) in 0.1 ml NHS, 0.28 fig radiolabeled C9 and 0.2 ml VBS-G or 0.1 ml VBS-G plus 0.1 ml specific IgG (125 fig) for 2 h at 37 CO. Pelletassociated radioactivity determined and specific C9 deposition calculated (values represent mean ± SD of 4 experiments). 1

2. C9 deposition In these experiments we tested whether both strains differed with respect to the amount of surface bound C9 after complement activation. For the PKo strain approximately five times more C9 molecules were determined as compared to the B31 strain (Table 1). With both strains the presence of specific IgG had no effect on the outcome of the experiment. The time course of C9 deposition was also analyzed in kinetic experiments (Fig. 2). C9 binding on the B31 strain was slow and increased gradually over the entire incubation period. In contrast, C9 binding to the PKo strain was very rapid with maximal values after a 15-30 min incubation period. mol. C9 / cell (x10-3) 350 300 250 200 150 100 50 O ~~~~~~~~~~~~~Lr~~~~~~J

0.00

1.00

2.00

time (h) Figure 2. Kinetics of C9 deposition on the B. burgdorferi strains B31 and PKo. Bacteria (5 x 10 7) were incubated for 2 h at 37 °C in 25% (v/v) NHS containing 0.28 fig 115J_C9. At the indicated times samples were removed and centrifuged. After washing the bacterial pellets the number of specifically bound C9 molecules was determined.

458 . V. BRAVE, 1. KLEBER, and G. ACKER

1

TCC

2

3 4

5 6

7 8 910

••

C9 Figure 3. Demonstration of C5b-9 (m) after incu bation of the B. burgdorferi strains B31 and PKo in NHS. Bacteria (5 x 10 7 ) were incubated for 2 h at 37"C in 25% (v/v) NHS or CH56 (control) plus 0.28 ~g 125J -C9 in the presence or absence of B. burgdorferi-specific IgG. After incubation the washed pellets were subjected to SDS-PAGE (10% polyacrylamide) under reducing conditions and to autoradiography. Lanes: 1, NHS/ 125J-C9; 2, CH56/ 12SJ-C9; 3, B31 + NHS/ 125J-C9; 4, B31/IgG + NHS/ 125J_ C9; 5, B31 + CH56/ 12SJ-C9; 6, B31/IgG + CH56/ 12SJ-C9; 7, PKo + NHS/ 12SJ-C9; 8, PKo/ IgG + NHS/ 125J-C9; 9, PKo + CH56/ 12SJ-C9; 10, PKo/IgG + CH56/ 12SJ-C9. TCC = terminal complement complex.

3. Structure of bound C9 In the following experiments, we analyzed to what extent C9 bound to the surface of the B. burgdorferi strains was incorporated into the C5b-9 complex. After incubation with serum containing radiolabeled C9 the bacteria were subjected to SDS-PAGE and autoradiography. As shown in Figure 3, a terminal complement complex (TCC) was formed on both strains, but to a greater extent on the PKo strain. 4. Serum resistance

Both strains were compared with respect to their serum resistance in 25 % human serum in the presence and absence of added B. burgdorferiFigure 4. Thin-section electron micrographs of cells of the B. burgdorferi strain PKo after ~ incubation with CH56. Bacteria (5 x 107 ) were incubated for 2 h at 37 °C in 25% (v/ v) CH56. Serum-treated samples were then fixed and prepared for electron microscopy. A: Intact cells, occasionally blebs (arrow). Amorphous material at the cell surface (5). Magnification x 83000. Dimensions in ~m in all micrographs. B: Cross-sectioned cell wi th protoplasmic cylinder (P), cytoplasmic membrane (CM), outer membrane (OM), and flagella (F, arrows). Magnification x 181000.

Complement Activation by Borrelia burgdorFeriStrains . 459

J

OM

0.1

460 . V. BRADE, 1. KLEBER, and G. ACKER Table 2. Bactericidal assay with the B. burgdorferi strains B31 and PKo No. Bactericidal assayt

1. 2. 3. 4. 5. 6. 7. 8.

B31 + NHS B311 Ab4 + NHS B31 +CH56 B311 Ab4 + CH56 PKo+NHS PKolAb s + NHS PKo+CH56 PKolAb s + CH56

Approximate Cell cell number fragments (x 10 7)2 and blebs

Cell clumps

Cell multiplication in subcultures 3

4.9 ± 0.7 4.5 ± 0.5 5.1 ± 0.7 5.0 ± 0.4 2.7 ± 1.2 1.8 ± 0.9 5.1 ± 0.7 5.3 ± 0.8

B31: Frequent and large. No difference in samples 1-4

yes yes yes yes no no yes yes

rare rare very rare very rare frequent frequent rare rare

PKo: Less frequent and smaller than with B31. No difference in samples 5-8

t Standard assay with 5 x 107 B. burgdorferi in 25 % (v/v) NHS or CH56 for 2 hat 3rc (results of 4 experiments). 2 More than 95 % of the counted spirochetes were mobile. 3 Subcultures in BSK medium for 48 h after the bactericidal assay. 4 Source of specific antibodies (Ab) were either purified IgG (1 patient) or heat-inactivated sera (six patients). 5 Source of specific antibodies (Ab) was purified IgG (1 patient).

specific antibodies. After a 2 h incubation period the B31 strain was unaffected with respect to its cell number and mobility. Occasionally cell damage was indicated by the detection of fragments and blebs (Table 2). However, these events were rare and also seen in the controls (CH56). Viability of the B31 strain was proven by subcultures which were judged after 48 h incubation time by the same criteria. Again, the bacteria appeared mobile and intact and the cell number almost doubled within this incubation period. In the same experiment the PKo strain behaved totally differently. In serum, but not in the control (CH56), the number of the bacteria decreased significantly and there was extensive fragment and bleb formation (Table 2). Those bacteria which seemed unaffected or demonstrated little bleb formation after the 2 h incubation period did not multiply in 48 h subcultures (Table 2). Thus, irreversible damage had occurred to the vast majority of cells of the PKo strain. It should be noted that bacterial clumps of variable size were observed with both strains. This clumping phenomenon was independent of the presence or absence of antibody and more pronounced with the B31 strain (Table 2) indicating thin the B31 and PKo strains differ with respect to their surface properties.

Figure 5. Thin-section electron microscopy after incubation of cells of the B. burgdorferi strain ~ PKo with NHS. Bacteria (5 x 10 7) were incubated for 2 h at 37°C in 25% (v/v). Serum-treated samples were then fixed and prepared for electron microscopy. A: Numerous blebs, cell fragments and spheroblasts. Magnification x 24000. B: Blebs usually bound by the trilaminar outer cell membrane (OM). Fuzzy appearance of the outer leaflet of the OM due to deposition of serum proteins (arrow). Protoplasmic cylinder (P) less electron dense than in the control (Fig. 4). Magnification x 83000.

Complement Activation by Borrelia burgdorleriStrains . 461

A

462 . V.

BRADE,

1.

KLEBER,

and G.

ACKER

5. Thin-section electron microscopy The morphological changes occurring to cells of the PKo strain by the action of NHS were also visualized by electron microscopy. After the 2 h incubation period in CH56 (control) the ultrastructural architecture of the cells of the PKo strain remained intact showing the typical trilaminar structure of the outer cell membrane, a dense protoplasmic cylinder and the periplasmic location of flagella (Fig. 4B). Outside the outer cell membrane amorphous, in patches distributed material was observed which represents most probably the already described slime layer at the cell surface of B. burgdorferi (22, Fig.4A). However, after incubation in NHS numerous detached blebs of different size or blebs still connected to the bacterial body were seen (Fig. 5A and B). The blebs were usually surrounded by a trilaminar structure (Fig. 5B) which is the typical appearance of the outer cell membrane of the intact bacterial cells (Fig.4B). In parallel experiments with cells of the strain B31, no alteration of the cell morphology could be observed (data not shown).

Discussion In our studies the B. burgdorferi strains B31 and PKo differed with respect to their serum resistance. The American strain B31 (ATCC 35210) was a low activator of complement as determined by C9 consumption and C9 deposition (Table 1, Fig. 2) and terminal complement complex (TCC) formation (Fig. 3). Failure of complement to damage this strain under the conditions of our bactericidal assay was indicated by full mobility and viability after the 2 h incubation period (Table 2). Additional morphological studies with the electron microscope showed the characteristic ultrastructure of intact cells of B. burgdorferi (data not shown). Surprisingly the addition of specific antibody from patients with Lyme borreliosis did not make any difference in the complement activation and viability experiments (Table 1 and 2). There are very recent reports that specific antibodies together with complement are detrimental to B. burgdorfen·. In these studies, immune sera from infected rats (6), from hamsters (7,20) or from patients with Lyme borreliosis (8,21) were found to immobilize and kill B. burgdorferi. Thus, in contrast to our experiments, several investigators presented evidence that complement and specific antibodies are bactericidal to B. burgdorferi. At present, these different findings cannot be explained. It should be realized, however, that the other investigators (6-8, 20, 21) used a different B. burgdorferi strain 297, which was originally isolated years ago from the cerebrospinal fluid of a patient. As shown in our present experiments the properties of a given B. burgdorferi isolate have a strong influence on the outcome of the experiments. Furthermore, the various bactericidal assays (6-8, 20, 21) differed considerably

Complement Activation by Borrelia burgdorferi Strains . 463

with respect to the complement source, cell concentration, incubation time and addition of extra lysozyme. Based on the results of our experiments we conclude that antibody and complement are not uniformly bactericidal to all B. burgdorferi strains. The European PKo strain (12) was a high activator of complement even in the absence of specific antibody (Table 1, Fig. 2). Complement-mediated damage was proven in our bactericidal assay by drastically reduced mobility and viability after the 2 h incubation period (Table 2). In addition, extensive membrane damages were visualized in the electron microscope. The most prominent findings were cell fragmentation and formation of blebs of different size, either detached or still connected to the bacterial body (Fig. 5). Blebs already observed earlier in context with studies on the ultrastructure of B. burgdorferi (22) most likely consist of protoplasm a surrounded by the bacterial trilaminar outer membrane. Complement and lysozyme are the lytic agents which allow the efflux of protoplasm a from the bacterial cells. Interestingly, the complement activation by the investigated strains of B. burgdorferi and killing of the PKo strain were independent of the presence of specific antibodies (Table 1, Fig. 2). We therefore conclude that surface properties of B. burgdorferi determine whether the strain is a low or high activator of complement and whether the bacteria are killed. It is known that B. burgdorferi strains differ considerably with respect to their phenotype (9) and genotype (10) and that the strains B31 and PKo belong to different groups (23). In our study we confirmed that the B31 and PKo strains have a different protein pattern according to SDS-PAGE analysis (Fig. 1). Differences in surface properties between both strains became apparent in the darkfield microscope. Both isolates tended towards spontaneous cell agglomeration. However, cell clumps were more prominent in number and size with the B31 than with the PKo strain. Specific antibodies did not enhance this phenomenon (Table 2). For the future, it remains an interesting task to identify those surface constituents of B. burgdorferi strains which are important for low complement activation and for serum resistance. The analysis of surface components in this context should include a previously described (22) and in Figure 4A demonstrated slime cover of spirochetes which may be important for protection against complement. Finally, further studies with serum resistant and serum sensitive B. burgdorferi strains should reveal whether serum resistance belongs to the virulence factors of pathogenic B. burgdorferi. References 1. STEERE, A. 1989. Lyme Disease. N. Engl. J. Med. 321: 586-596. 2. ASBRINK, E. and A. HOVMARK. 1985. Successful cultivation of spirochetes from skin lesions of patients with erythema chronica migrans afzelius und acrodermatitis chronica atrophicans. Acta Path. Microbiol. Immunol. Scand. 93: 161-163.

464 . V. BRADE, 1. KLEBER, and G. ACKER 3. KOCHI, S. K. and R. C. JOHNSON. 1988. Role of immunoglobulin G in killing of Borrelia burgdorferi by the classical complement pathway. Infect. Immun. 56: 314-32l. 4. SCHMITZ, J. L., R. F. SCHELL, A. G. HEjKA, and D. M. ENGLAND. 1990. Passive immunization prevents induction of Lyme arthritis in LSH hamsters. Infect. Immun. 58: 144-148. 5. SIMON, M. M., U. E. SCHAIBLE, R. WALLlCH, and M. D. KRA?;lER. 1991. A mouse model for Borrelia burgdorferi infection: approach to a vaccine against Lyme disease. Immunol. Today 12: 11-16. 6. PAVIA, C. S., V. KISSEL, S. BrrTKER, F. CABELLO, and S. LEVINE. 1991. Antiborrelial activity of serum from rats injected with the Lyme disease spirochete. J. Infect. Dis. 163: 656-659. 7. LOVRICH, S. D., S. M. CALLISTER, J. L. SCHMITZ, J. D. ALDER, and R. F. SCHELL. 1991. Borreliacidal activity of sera from hamsters infected with the Lyme disease spirochete. Infect. Immun. 59: 2522-2528. 8. KOCHI, S. K., R. C. JOHNSON, and A. P. DALMASSO, 1991. Complement-mediated killing of the Lyme disease spirochete Borrelia burgdorferi. Role of antibody in formation of an effective membrane attack complex. J. Immunol. 146: 3964-3970. 9. WILSKE, B., V. PREAC-MuRSIC, G. SCHIERZ, R. KUHBECK, A. G. BARBOUR, and M. KRAMER. 1988. Antigenic variability of Borrelia burgdorferi. Ann. N.Y. Acad. Sci. 539: 126-143. 10. BARBOUR, A. G. 1988. Plasmid analysis of Borrelia burgdorferi, the Lyme disease agent. J. din. Microbiol. 26: 475-478. 11. RAPP. H. J. and T. BORSOS (eds.). 1970. Molecular basis of complement action. AppletonCentury-Crafts, New York, pp. 75-109. 12. PREAC-MuRSIC, V., B. WILSKE, and G. SCHIERZ. 1986. European Borrelia burgdorferi isolated from humans and ticks. Culture conditions and antibiotic susceptibility. Zbl. Bakt. Hyg. A 263: 112-118. 13. BIESECKER, G. and H. J. MULLER-EBERHARD. 1980. The ninth component of human complement: Purification and physicochemical characterization. J. Immunol. 124: 1291-1296. 14. PILZ, D., T. VOCKE, J. HEESEMANN, and V. BRADE. 1992. Studies on the mechanism of YadA-mediated serum resistance of Yersinia enterocolitica serotype 03. Infect Immun. 60: 189-195. 15. JOINER, K. A., C. A. HAMMER, E. J. BROWN, R. J. COLE, and M. M. FRANK. 1982. Studies on the mechanism of bacterial resistance to complement-mediated killing. 1. Terminal complement components are deposited and released from s. minnesota S218 without causing bacterial death. J. Exp. Med. 155: 797-907. 16. LAEMMLI, U. K. 1970. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227: 680-685. 17. HOCHSTRASSER, D. F., M. G. HARRINGTON, A. C. HOCHSTRASSER, M. J. MILLER, and C. R. MERRIL. 1988. Methods for increasing the resolution of two-dimensional protein electrophoresis. Analyt. Biochem. 173: 424-435. 18. SPURR, A. R. 1969. A low-viscosity epoxy resin embedding medium for electron microscopy. J. Ultrastruct. Res. 26: 31-43. 19. REYNOLDS, D. S. 1963. The use of lead citrate at high pH as an electron opaque stain in electron microscopy. J. Cell. BioI. 17: 208-212. 20. SCHMITZ, J. L., R. F. SCHELL, S. D. LOVRICH, S. M. CALLISTER, and J. E. COL 1991. Characterization of the protective antibody response to Borrelia burgdorferi in experimentally infected LHS hamsters. Infect. Immun. 59: 1916-1921. 21. CALLISTER, S. M., R. F. SCHELL, and S. D. LOVRICH. 1991. Lyme disease assay detects killed Borrelia burgdorferi.]. Clin. Microbiol. 29: 1773-1776. 22. BARBOUR, A. G. and S. F. HAYES. 1986. Biology of Borrelia species. Microbiol. Rev. 50: 381-400.

Complement Activation by Borrelia burgdorferiStrains . 465 23. ADAM, T., G. S. GASSMANN, C. RASIAH, and U. B. GOBEL. 1991. Phenotypic and genotypic analysis of Borrelia burgdorferi isolates from various sources. Infect. Immun. 59: 2579-2585.

Dr. VOLKER BRADE, Zentrum der Hygiene, Abteilung fur Medizinische Mikrobiologie, Paul-Ehrlich-Str. 40, 6000 Frankfurt am Main 70, Germany