immunoelectrophoretic characterization and cross-reactivity of rochalimaea henselae, rochalimaea...

APMIS 102. 931-942, I994 Printed in Denmark . All rights re.served

Copyright 0 A P M I S I994

ISSN OYO3-4641

Immunoelectrophoretic characterization and cross-reactivity of Rochaliwa henselae,

R o c h d i w a quintana and Afipia felis

KRESTEN ENGBEK and CLAUS KOCH

Department of Immunology, Section of Applied Immunology, Statens Seruminstitut, Copenhagen, Denmark

Engbzk, K. & Koch, C. Immunoelectrophoretic characterization and cross-reactivity of Rochalimaea henselae, Rochalimaea quintana and Afpia felis. APMIS 102: 931-942, 1994.

The soluble antigens of Rochalimaea henselae, Rochalimaea quintana and AJipia felis were character- ized by crossed immunoelectrophoresis using bacterial sonicates as antigens against pooled hyperimmune rabbit sera. A precipitin pattern was drawn for each bacterium and shown to be reproducible and stable even when normal or preimmune rabbit serum was incorporated in the intermediate gel. By this technique 56 antigens were identified from R. henselae, 49 from R. quintana, and 39 from A . felis. The serological cross-reaction between R. henselae, R. quintana and A . felis, and between these 3 bacteria and 32 pathogenic bacteria was analysed by rocket-line immunoelectrophoresis, crossed-line immunoelectrophoresis, and tandem-crossed electrophoresis. It was concluded that (i) 4-7 antigens distinguish R. henselae, R. quintana and A . felis from each other, (i i) both Gram-positive and Gram- negative bacteria cross-react with R. henselue, R. quintana and A. felis antisera, (iii) the cross-reacting antigens of Gram-negative bacteria have both precipitating and non-precipitating specificities, whereas Gram-positive bacteria have mainly non-precipitating specificities, (iv) the cross-reacting antigens are common to several species, and ( v ) fewer cross-reacting antigens are found in phylogenetically disparate species than in more closely related species.

Key words: Rochalimaea henselae; Rochalimaea quintana; Afpia felis; serological cross-reactions; immunoelectrophoresis.

K. Engbzk, Department of Immunology, Section of Applied Immunology, Statens Seruminstitut, Artillerivej 5, DK-2300 Copenhagen S, Denmark.

The aetiological agent of cat-scratch disease (CSD) has not been indisputably identified. Two species of bacteria, Ajipia felis and Rochalimaea henselae, have been proposed as the causative organism. English et al. (6) isolated A. felis from lymph nodes of 10 patients with CSD. Three of seven patients with recent CSD had a four-fold or higher rise in antibody titre to A. felis in an in- direct immunofluorescence assay. On the other hand, Regnery et al. (15) found that serum samples from 36 of 41 patients with suspected

Received August 12, 1994. Accepted October 27, 1994.

CSD had antibody titres of 64 or more t o R. henselae, whereas only 10 of these patients had raised but low titres to A. felis. Similar results were obtained by enzyme immunoassay (2). In spite of these conflicting findings, no studies have been published on the serological cross-reactivity of A. felis and R. henselae. We have included R. quintana in the present study as this organism has been isolated from immunocompromised patients with bacillary angiomatosis (BA), a disease normally caused by R. henselae (10).

Ajipia and Rochalimaea genera belong to the a-2 subgroup of the class Proteobacteria, but further systematization has not been attempted (16). The genus Rochalimaea was formerly

93 1

ENGBKK & KOCH

classified in the Rickettsiaceae family of the order Rickettsiales. Phylogenetic data, however, have shown that Rochalimaea species are less related to organisms normally classified in the Rickettsiales than to Bartonella bacilliformis, a bacterium that causes Carrion’s disease in resi- dents of the high valleys of the Andes. Recently it has been proposed that the genus Rochali- maea be removed from the order Rickettsiales and that the two genera be united in the Bar- tonellaceae family, a proposal not followed in this article (5 ) .

There have been few immunochemical studies of Rochalimaea and Ajipia. Hollingdale et al. (8) found that the outer membrane of R. quintana consisted of protein and lipopolysaccharide components. The protein, but not the lipopolysaccharide, reacted with sera from patients in- fected by Rickettsia tsutsugamushi. Reed et al. (14) produced polyclonal antibodies against a protein fraction from R. henselae of molecular weight > 14.3 kDa. The antiserum cross-reacted weakly with cultures of R. quintana, but this cross-reactivity could be removed by specific ab- sorption. The antiserum did not cross-react with A . felis, Treponema pallidum, Borrelia burgdorferi or Helicobacter pylori. Knobloch (9) identified 24 antigens of Bartonella bacilliformis by immunoblotting and immunoprecipitation with rabbit hyperimmune sera. The crude antigen cross-reacted with convalescent sera containing antibody against Chlamydia psittaci, but not with Brucella, Coxiella, Legionella, Myco- plasma, Neisseria, Salmonella, Shigella, Ascaris, Schistosoma, or Echinococcus antisera. The crude antigen also reacted with an unknown antibody in human sera, whereas immunoblotting and immunoprecipitation of Triton-soluble antigens showed no reactions.

The purpose of the present study was three- fold: (1) to characterize the soluble antigens of R. henselae, R. quintana and A. felis, (2) to study the serological cross-reactivity between these bacteria, and (3) to identify specific antigens that may be used in the diagnosis of CSD and BA.

MATERIALS AND METHODS

Strains Rochalimaea henselae ATCC 49882, Rochalimaea

quintana ATCC VR 358 and Ajipia felis ATCC 53690

932

were procured from the American Type Culture Col- lection, Rockville, MD, USA and Brucella abortus biotype 1 ATCC 23448 came from the State Veterin- ary Serum Laboratory, Denmark. The other strains mentioned in Table 1 were obtained from the national collection at the Departments of Clinical Micro- biology and Tuberculin, Statens Seruminstitut, Denmark.

Antigen preparation Rochalimaea and Streptococcus species were cul-

tured on blood agar base no. 2 (Difco 0696) with 5% horse blood, Meisseria gonorrhoea, Haemophilus and Capnocytophaga on chocolate agar, Ajipia and Le- gionella on buffered charcoal-yeast extract (BCY E) agar (State Serum Institute product number 262104), and the remaining species on brain-heart infusion agar (Difco 0418). The growth was scraped off the plates and washed three times in saline. After the last centrifugation the pellet was resuspended in 150 mM NaCl containing 10 mM EDTA, 50 mM Tris-HCI, 0.1% Triton X-100, pH 7.6 (to a total volume of 5 ml per g wet weight of pellet), and ultrasonicated in a Branson Sonifer model 250 (Branson Ultrasonics Co., Danbury, CT, USA) for 5 min in ice. The sonicated suspension was centrifuged at 45,000 g for 60 min and the supernatant divided into 1 .O ml samples and stored at -80°C. The protein concentration was determined by the dye-binding assay described by Bradford (4).

Hyperimmune sera Three groups of 10 rabbits were each immunized

with A. felis, R. henselae and R. quintana antigen re- spectively (200 pg bacterial protein per rabbit per immunization) given intracutaneously at multiple sites. The immunogen was prepared by adsorbing 2 ml sonicated antigen onto 1.5 ml aluminium hydroxide suspension (6 mg aluminium per ml) in 1.5 ml saline and emulsifying the mixture with 5 ml Freund’s adju- vant (complete for the first and incomplete for subsequent immunizations). After eight immunizations at fortnightly intervals the antisera were tested in crossed immunoelectrophoresis and five rabbits were selected for further immunizations at monthly intervals. Ten days after each subsequent immunization the rabbits were bled 3040 ml, the blood was allowed to clot, and the serum was recovered (7).

Immunoelectrophoresis Crossed immunoelectrophoresis (XIE), crossed-

line immunoelectrophoresis (XLIE), tandem-crossed immmunoelectrophoresis (TCIE) and rocket-line immunoelectrophoresis (RLIE) were carried out as described by Axelsen & Bock (1) and Kr~lll(11-13) with equipment from Daela, Copenhagen. Electrophoresis was performed in gels of 1% agarose of low electroen- dosmosis (Hispanagar D-1 LE, lot HO 147/87, E- 09080 Burgos, Spain) in Tris-barbitone buffer, pH 8.6, ionic strength 0.02, on glass plates.

ANTIGENS OF ROCHALIMAEA AND AFIPIA

I

...

0 Rochalimaea henselae

0 Afipia jelis

0 Rochalimaea quintana

Fig. 1. Crossed immunoelectrophoresis reference diagram of R. henselae, R. quintana and A . felis precipitin patterns with blank intermediate gel. (A) 18 pg R. henselae sonicate against 40 pl cm-2 anti-R. henselae serum. (B) 18 pg R. quintana sonicate against 40 p1 cmP2 anti-R. quintana serum. (C) 18 pg A. felis sonicate against 40 pl cmp2 anti-A. felis serum.

933

ENGBRK & KOCH

Fig. 2. Identification of cross-reacting antigens of R. henselae, R. quintunu and A. felis by rocket-line immunoelectrophoresis. A rectilinear gel containing 50 pl cm-2 sonicate of R. henselue, and a blank gel with three circular wells containing 10 p1 sonicate of R. henselae (Rh), R. quintana (Rq) and A. felis (Af) electrophoresed against a gel containing 17 p1 cmp2 rabbit anti-R. henselue at 2 V cmp ' for 18 h with the anode at the top.

RH

Fig. 3. Number of antigens of R. henselue cross-reacting with antigens of 13 bacterial species by rocket-line immunoelectrophoresis. The rectilinear gel contained 10 pg cmP2 R. henselae (RH) sonicate and a blank gel with 13 circular wells containing 10 p1 sonicates of Escherichia coli, Yersiniu enterocolitica, Vibrio cholerae, Pasteurellu multocida, Huemophilus injuenzae, Flavobacterium meningosepticum, Legionella pneumophila, Cap- nocytophaga canimorsus, Afipia jelis, Brucellu abortus, Agrobacterium tumefaciens, Streptococcus pneumoniue and Staphylococcus aureus was electrophoresed against a gel containing 9 pl cm-' rabbit anti-R. henselue (anti- RH) at 2 V cm-' for 18 h with the anode at the top.

934

ANTIGENS OF ROCHALIMAEA A N D AFIPIA

For XIE, 20 p1 of the sample antigens and a bromophenol-albumin marker were applied in circular wells in an agarose gel and separated electrophor- etically at a potential gradient of 10 V cm-'. Single lanes of the separated antigens were cut out and transferred to 5 X 7 cm glass plates. A 1.5 X 5 cm intermediate gel and a 4 x 5 cm top gel containing antiserum to the sample antigen were poured onto the plate, and electrophoresis was carried out at right angles to the first direction at a potential gradient of 2 V cm-' for 18 h. The intermediate gel contained either 200 p1 heterologous bacterial antiserum or 200 p1 saline as control. After electrophoresis, the gel was

pressed, washed, dried, and stained with Coomassie brilliant blue R-250 (17).

The procedure for XLIE followed that described for XIE, except that 200 p1 heterologous bacterial sonicate was added to the intermediate gel and electrophoresed together with the first-dimension gel.

For TCIE, two circular wells were punched 3 mm apart. A sonicate of individual bacterial species whose antigens were to be compared was added to each well and allowed to soak completely into the gel. The wells were then filled with agarose and the gels were electrophoresed. Electrophoresis in the second dimension was performed as described for XIE. For

TABLEI. Cross-reaction of R. henselae, R. quintana and A. felis antisera with antigens from 37 different bacterial species determined by rocket-line immunoelectrophoresis. The number of precipitating antigens is shown in

brackets Bacterial species Strain Preimmune Hyperimmune serum against

no. serum R. henselae R. auintana A. felis Brucella abortus ATCC 23448 0 + (7) +(5) + (4) Agrobucterium tumefaciens AB 2715 0 + (7) + ( 5 ) + (4) AJipia clevelandensis ATCC 49720 0 +(3) +(3) +(5) Pseudomonas diminuta AB 1266 0 +(3) +(3) +(3) Alcaligenes xylosoxidans U 105 0 + (2) + (2) + (2) Bordetella pertussis 12916 +(I) + (2) +(I) + (2) Eikenellu corrodens U 173 0 +u> +(I) 0 Kingella kingae ATCC 23330 0 +(I) +(I) 0 Acinetobacter calcoaceticus u 10 0 + (2) + ( a + ( a Pseudomonas aeruginosa ATCC 27853 0 +(I) +(2) 0 Vibrio parahaemolyticus ATCC 17802 0 +(I) +(I) +(I) Aeromonas hydrophila U28IK25 0 +(I) + ( a 0 Cardio bacter ium hom in is ATCC 15826 0 +(I) +(a +(I) Escherichia coli ATCC 25922 0 + ( a +(2) 0 Enterobacter aerogenes U 23 0 +(I) +(2) +(I) Salmonella typhimurium u 7 0 +(I) + ( a 0 Citrobacter koseri u 33 0 + ( a +(a + (2) Pro t eus mirab ilis U 27 0 + ( I ) +(a +(I) Yersinia enterocolitica UX 29 0 +(I) + ( 2 ) 0 Haemophilus injuenzae ATCC 9795 0 + ( I ) +(2) 0 Legionella pneumophila ATCC 33153 0 +(I) +(2) 0 Pasteurella multocida U 158 0 + ( I ) +(2) 0 Flavobacterium meningosepticurn U 1 10 0 + (2) + ( a + ( 2 ) Capnocytophaga canimorsus P 1357 0 + (2) +(2) + ( a

Staphylococcus aureus ATCC 29923 0 0 0 0 Streptococcus pyogenes group A ATCC 12353 0 0 0 0 Streptococcus pneumoniae CCNG 8438 0 0 0 0 Enterococcus faecalis ATCC 29212 0 0 0 0 Corynebacterium diphtheriae ATCC 10356 0 0 0 +(I) Corynebact. pseudotuberculosis ATCC 3450 0 +(I) +(I) + ( I ) Listeria monocytogenes NCTC 7973 0 +(I) +(I) 0 Streptomyces somaliensis MNC 1219 0 +(I) 0 +(I) Mycobacterium tuberculosis H37RV 0 0 0 +(2) Mycobacterium fortuitum MNC 20 0 +(I) +(I) +(I) Nocardia asteroides MNC 81 0 0 0 +( I ) Oerskovia turbata MNC 11 13 + ( I ) +(I) +(I) + (2)

Micrococcus luteus ATCC 9341 0 0 0 0

935

ENGBWK & KOCH

936

ANTIGENS OF R O C H A L I M A E A AND AFIPIA

Fig. 4. Identification of cross-reacting antigens of R. henselae, R. quintana and A. felis by crossed immunoelectrophoresis. (A) 18 pg R. henselae sonicate against 40 pl cm-2 anti-R. henselae serum with blank intermediate gel; (B) 18 pg R. henselae sonicate against 40 pl cm-2 anti-R. henselae serum and 200 p1 rabbit anti-R. quintana serum in the intermediate gel; ( C ) 18 pg R. quintana sonicate against 40 p1 cmP2 anti-R. quintana serum with blank intermediate gel; (D) 18 pg R. quintana sonicate against 40 pl cm-* anti-R. quintana serum and 200 p1 anti-R. henselae serum in the intermediate gel; (E) 18 pg A. felis sonicate against 40 pl cm-2 rabbit anti-A. felis serum with blank intermediate gel; (F) 18 pg A. felis sonicate against 40 p1 cmp2 anti-A. felis serum and 200 p1 anti-R. henselae serum in the intermediate gel. Electrophoresis was carried out at 10 V cm-' for 45 min in the first dimension with the anode to the right and at 2 V cm-' for 18 h in the second dimension with the anode at the top. Precipitin lines unaffected by the presence of heterologous antiserum in the intermediate gel are indicated by their reference numbers.

each comparison, two controls were included in which one of the bacterial sonicates was replaced with phosphate-buffered saline.

For RLIE, a 2 mm-thick blank agarose gel was cast on a 10x20 cm glass plate. A single trough (20x1 cm) was cut in the agar 1 cm from the cathodic mar- gin and filled with sonicate of one of the species R. henselae, R. quintana and A. felis in melted agarose. A 1 cm-wide contact gel on the anodic side of the antigen trough was left blank. The remaining gel on the anodic side was removed and replaced with antibody-containing agarose. A row of 13 circular wells was punched in the contact gel and 10 p1 samples of the bacterial sonicates to be tested for cross-reactivity were added to the wells.

The optimal Ag/Ab ratio for immunoelectrophoresis was defined as the lowest concentration of

the antigen and antiserum that gave the maximum number and sharpest presentation of precipitin lines obtainable. This was attained in crossed immunoelectrophoresis, crossed-line immunoelectrophoresis and tandem-crossed immunoelectrophoresis when 18 pg sonicated bacterial antigen was electrophoresed into gel containing 40 p1 cm-2 antiserum and in rocket- line immunoelectrophoresis 10 pg per cmp2 sonicated bacterial antigen against 9 p1 cm-2 antiserum. These ratios were used in all succeeding experiments.

RESULTS

Reference diagrams of crossed immunoelectrophoretic precipitin patterns of R. henselae, R.

937

ENGBAK & KOCH

quintana and A. felis (the reference strains) obtained at the optimal Ag/Ab ratio with a blank intermediate gel are shown in Fig. 1. Each reference diagram was traced on the basis of five experiments with three different batches of antigen. A solid line indicates a distinct precipitin line present in all experiments, whereas a broken line denotes a weak precipitin line not necessarily present in all experiments. When preimmune rabbit serum, pooled rabbit serum, pooled mouse serum or normal human serum was added to the intermediated gel, all precipitin lines shown in the reference diagrams could still be identified, even though some had a different appearance. By this method the maximum number of precipitin lines detected was 56 for R. henselae, 49 for R. quintana and 39 for A. felis, each corresponding to a detectable antigen. The precipitin lines were numbered according to their electrophoretic mobility in the first dimension, with no. 1 as the most anodic line.

The numbers of antigens from each of the three species R. henselae, R. quintana and A. felis that cross-reacted with antisera against the other two species and the numbers of cross-reacting antigens in 37 pathogenic bacteria were studied by RLIE. Two types of change in precipitin pattern were detected: i) a precipitin “rocket” in the antibody-containing gel above the well with bacterial sonicate indicated the presence of a cross-reacting antigen that could be precipitated by the antibody, or ii) an anodic deflection of one of the horizontal precipitin lines above the well with bacteria sonicate indicated the presence of a cross-reacting epitope that did not form a precipitate.

The most extensive cross-reaction (Fig. 2) was found between R. henselae and R. quintana, as all but six R. henselae and four R. quintana antigens cross-reacted with antiserum against the other species. The cross-reaction between A. felis antigens and R. henselae or R. quintana antisera was less conspicuous, as only six A. felis antigens cross-reacted with these antisera. Testing of the other 37 bacterial species demonstrated that all species possessed 3-7 cross-reacting antigens without precipitating properties, and with the ex- ception of 6 species (Micrococcus luteus, Staphy- lococcus aureus, Streptococcus pyogenes group A , Streptococcus pneumoniae, Enterococcus faecalis and Neisseria gonorrhoeae) 1-7 cross-reacting antigens that produced precipitation lines (Fig.

938

3). Fewer (<3) cross-reacting antigens were found in phylogenetically disparate species than in more closely related species such as Brucella abortus, Agrobacterium tumefaciens and Pseudo- monas diminuta, and the cross-reacting antigens of the distant species had no apparent connec- tion to a precipitin line from the reference strains. This may indicate either a high concentration of the homologous antigen that has run out of the gel or that the cross-reacting antigen forms a precipitin line while the homologous antigen forms a non-precipitating complex with the antiserum (Table 1).

To identify the antigens of R. henselae, R. quintana and A. felis that demonstrated a mu- tual cross-reaction, antiserum against the heterologous species was incorporated in the intermediate gel of the three reference systems (Fig. 4). The changes in the precipitation lines were analysed according to the principles described by Axelsen & Bock (1) and Bock & Axelsen (3). No changes were observed when preimmune serum was added to the intermediate gel. If R. quintana antiserum was incorporated in the intermediate gel of the R. henselae reference system, several precipitin lines were deflected to- wards the intermediate gel and only lines RH- 2, RH-17, RH-45 and RH-47 were unchanged. In the reverse situation, when R. henselae antiserum was added to the intermediate gel of the R. quintana reference system, only lines RQ-2, RQ-8, RQ-9, RQ-36 and RQ-47 were not deflected. Similarly, when R. henselae or R. quin- tuna antiserum was added to the A. felis reference system, only lines AF-5 (only with R. quin- tuna antiserum), AF-6, AF-10, AF-11, AF-16, AF-17, AF-21 and AF-23 were stationary.

The identity of antigens in R. henselae, R. quintana and A. felis cross-reacting serologically with antigens in other bacterial species was in- vestigated in TCIE and XLIE by comparison with sonicates of Brucella abortus, Agrobacteri- um tumefaciens, Pseudomonas diminuta, Alcali- genes xylosoxidans, Acinetobacter calcoaceticus, Citrobacter koseri, Flavobacterium meningosepticum, and Capnocytophaga canimorsus (Fig. 5 , Table 2). The numbers of cross-reacting antigens demonstrated by these techniques were close to those found in RLIE, but the identity of some of the weaker precipitin lines could not be established by these methods, although different concentrations of antigens were used.

ANTIGENS OF ROCHALIMAEA A N D AFIPIA

Fig. 5. Crossed line immunoelectrophoresis of (A) 20 p1 R. quintana sonicate against 750 p1 rabbit anti-R. quintana with 200 p1 R. henselae sonicate incorporated in the intermediate gel; (B) 20 p1 R. quintana sonicate against 750 p1 rabbit anti-R. quintana with 200 p1 Brucella abortus sonicate incorporated in the intermediate gel; (C) 20 p1 R. henselae sonicate against 750 p1 rabbit anti-R. henselae with 200 pl Agrobacterium tumefaciens sonicate incorporated in the intermediate gel. First-dimension electrophoresis was carried out at 10 V cm-' for 45 min with the anode to the right; second-dimension electrophoresis at 2 V cm-' for 18 h with the anode at the top. Pre- cipitin peaks fusing with horizontal lines are indicated by their reference numbers.

939

ENGBEK & KOCH

TABLE 2. IdentiJication of R. henselae, R. quintana and A. felis antigens cross-reacting with Brucella abortus, Agrobacterium tumefaciens, Pseudomonas diminuta, Alcaligenes xylosoxidans, Acinetobacter calcoaceticus, Citrobacter koseri, Flavobacterium meningosepticum and Capnocytophaga canimorsus determined by crossed- line immunoelectrophoresis and tandem-crossed immunoelectrophoresis. The numbering refers to the respective

reference diagrams. Bold numbers indicate a common antigen for all tested species R. henselae R. quintana A. felis

Brucella abortus 16, 21, 30, 49, 52" 15, 29, 30, 36, 44, 46,47

Agrobacterium tumefaciens 16, 21, 30, 49, 52" 15, 44, 46, 47" Pseudomonas diminuta 21, 52" 15, 46, 47" 16, 22, 29. 38 Alcaligenes xylosoxidans 21, 52 15, 46 22, 38 Acinetobacter calcoaceticus 21, 52 15, 46 22, 38 Citrobacter koseri 21, 52 15, 46 22, 38 Flu vobacter ium men ingosep t icum 21, 52 15,46 22, 38 Capnocytophaga canimorsus 21, 52 15, 46 22, 38 " +additional unidentified cross-reacting antigens.

18, 20, 22, 25, 38

14, 18, 22, 35, 38

Four cross-reacting antigens were identified for the Rochalimaea species and three for A . felis. Two antigens from each of these species were found to cross-react serologically with all eight comparison species, RH-21 and RH-52 for R. henselae, RQ-15 and RQ-46 for R. quintana, and AF-22 and AF-38 for A . felis.

DISCUSSION

The uncertainty about the cause of CSD prompted us to investigate the serological cross- reactivity of R. henselae, R. quintana and A . felis. One of the main difficulties in developing reliable serodiagnostic assays in bacterial infections lies in the complex nature of the antigens used for such tests. Since bacterial sonicates are complex mixtures of large numbers of antigens, it is to be expected that they possess epitopes that are shared by other bacteria. Accordingly, phylogenetically related species, or even unre- lated bacteria, may induce the production of antibodies that cross-react because of similar- ities in the constituents present in different organisms or as a result of the exposure of new antigenic determinants due to the preparation procedure.

The chief aim of our investigation was to characterize the antigens of R. henselae, R. quin- tuna and A . felis, and to define their antigenic relationship. At least 56 antigens in R. henselae, 49 in R. quintana and 39 in A . felis were identified, which are the highest numbers of antigens so far demonstrated in these organisms. Not all the precipitin lines were equally distinct or pres-

ent in all experiments, but the precipitation patterns of the three bacteria were sufficiently sensitive, specific and reproducible to serve as a reference system for studies of serological cross- reactions and antibody responses in infections with these organisms.

Although the rabbits were immunized with the same antigen doses and immunization pro- tocol, the maximum number of precipitin lines obtained for R. henselae was higher than for R. quintana and A . felis, suggesting that in rabbits certain R. henselae antigens were more immuno- genic than those of the other two species.

XIE with antiserum against heterologous bacterial species incorporated in the intermediate gel is an accepted method of detecting serological cross-reactivity between bacteria and identifying the antigens involved. However, lack of deflection of a particular precipitin line does not exclude cross-reactivity, and deflection of a line does not prove cross-reactivity, as non-im- munological interactions may be involved. To overcome these problems we have used RLIE and XLIE to study the serological cross-reactivity between the bacteria.

The RLIE experiments show that R. henselae and R. quintana antigens cross-react extensively with antibodies raised against components of the other species and only four and six antigens, re- spectively, distinguish the two species from each other, whereas A . felis antigen cross-reacts much less with antibodies raised against the two Roch- alimaea species (Fig. 2). In the XLTE experiments the serological cross-reaction became more evi- dent and only 6-7 antigens discriminate A . jdis from R. henselae and R. quintana (Fig. 4). In ad-

940

ANTIGENS OF ROCHALIMAEA AND AFIPIA

dition, the study of the cross-reactivity of R. henselae, R. quintana and A. felis antisera with sonicates of 37 Gram-positive and Gram-negative bacterial species showed that (i) both Gram-positive and Gram-negative bacteria contain components that share epitopes with R. henselae, R. quintana and A. felis antigens, (ii) the cross-reacting antigens of Gram-negative bacteria show both precipitating and non-precipitating properties, whereas those of Gram-positive bacteria are mainly non-precipitating, (iii) the cross-reacting antigens are common to several species, and (iv) fewer cross-reacting antigens are found in phylogenetically less related species than in closely related species such as Brucella abortus and Agrobacterium tumefaciens. Though our results show that R. henselae, R. quintana and A . felis contain several antigens that share epitopes with those of other bacteria, these antigens may still have specific antigenic determinants which distinguish them from the corresponding antigens in other bacteria.

XLIE and TLIE were used to identify five antigens of R. henselae and R. quintana, and seven antigens of A . felis that cross-reacted serologically with antigens of eight other bacterial genera. Two antigens from each of the reference strains (RH-21, RH-52, RQ-15, RQ-46, AF-22 and AF-38) were shown to be related to antigens in all of the eight other organisms tested.

In conclusion, our studies show that R. henselae, R. quintana and A. felis contain both specific antigens and antigens that share determinants with those of other Gram-positive and Gram-negaive bacteria. For all three strains, some of the specific antigens (RH-17, RH-45, RQ-9, RQ-36, AF-6, AF-22) formed well-isolated, distinct precipitin lines, and may thus be useful for the production of monospecific antibodies. Which of these participate in the immunization process during the natural infection will be identified in a further study.

We are grateful to Dr L. 0. Uttenthal for his sugges- tions regarding the manuscript.

REFERENCES

I . A.uelsen, N. H. & Bock, E.: Antigenic identity.

2. Barka, N. E., HadJield, T., Patnaik, M . . Scand. J. Immunol. 17: 177-182, 1983.

Schwartzman, W. A. & Peter, J . B.: EIA for de- tection of Rochalimaea henselae-reactive IgG, IgM, and IgA antibodies in patients with suspected cat-scratch disease. J. Infect. Dis. 167:

3. Bock, E. & Axelsen, N. H.: Partial antigenic identity. Scand. J. Immunol. 17: 183-187, 1983.

4. Bradford, M . M . : A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72: 248-254, 1976.

5. Brenner, D. J . , O’Connor, S. P., Winkler, H. H. & Steigerwalt, A. G.: Proposal to unify the genera Bartonella and Rochalimaea, with description of Bartonella quintana comb. nov., Bartonella vin- sonii comb. nov., Bartonella henselae comb. nov., and Bartonella elizabethue comb. nov., and to re- move the family Bartonellaceae from the order Rickettsiales. Int . J. Syst. Bact. 43: 777-786, 1993.

6. English, C. K., Wear, D. J . , Margileth, A . M., Lissner, C. R. & Walsh, G. P.: Cat-scratch disease. Isolation and culture of the bacterial agent. J. Am. Med. Assoc. 259: 1347-1352, 1988.

7. Harboe, N. M . G. & Ingild, A,: Immunization, isolation of immunoglobulins and antibody titre determination. Scand. J. Immunol. 17: 345-35 1, 1983.

8. Hollingdale, M . R., Vinson, J . W. & Herrmann, J . E.: Immunochemical and biological properties of the outer membrane-associated lipopolysaccharide and protein of Rochalimaea quintana. J. Infect. Dis. 141: 672-678, 1980.

9. Knobloch, J.: Analysis and preparation of Bar- tonella bacillformis antigens. Am. J. Trop. Med.

10. Koehler, J . E.. Quinn, F. D., Berger, T. G., Le- Boit, P. E. & Tappero, J . W.: Isolation of Roch- alimaea species from cutaneous and osseous lesions of bacillary angiomatosis. N. Engl. J. Med. 327: 1625-1631, 1992.

1 1. Krdl, J. : Crossed-line immunoelectrophoresis. Scand. J. Immunol. 17, Suppl. 10: 171-173, 1983.

12. Kr011, J.: Rocket-line immunoelectrophoresis. Scand. J. Immunol. 17: 165-169, 1983.

13. Kr011, J. : Tandem-crossed immunoelectrophoresis. Scand. J. Immunol. 17, Suppl. 10: 135- 139, 1983.

14. Reed, J . A . , Brigati, D. J . , Flynn, S. D., McNutt, S. N., Min, K., Welch, D. F. & Slater, L. N.: Immunocytochemical identification of Rochali- maea henselae in bacillary (epithelioid) angiomatosis, parenchymal bacillary peliosis, and per- sistent fever with bacteremia. Am. J. Surg. Pathol. 16: 6540-6547, 1992.

15. Regnery, R. L., Olson, J. G., Perkins, B. A. & Bibb, W. : Serological response to “Rochalimaea henselae ” antigen in suspected cat-scratch disease. Lancet 339: 1443-1445. 1992.

1 503- 1 504, 1993.

Hyg. 39: 173-178, 1988.

94 1

ENGBEK & KOCH

16. Relman, D. A . , Lepp, P. W., Sadler, K. N . & 11. Svendsen, P. J. , Weeke, B. & Johunsson, B. G.: Schmidt, T. M. : Phylogenetic relationships Chemicals, solutions, equipment and general among the agent of bacillary angiomatosis, Bar- tonellu bacilliformis, and other alpha-proteobacteria. Mol. Microbiol. 6: 1801-1807, 1992.

procedures. Scand. J. Immunol. 17: 3-20, 1983.

942

immunoelectrophoretic characterization and cross-reactivity of rochalimaea henselae, rochalimaea...

Documents