Journal

of
Microbiological
Journal of Microbiological Methods 37 (1999) 267–295
Methods

First International Conference on Bartonella as Emerging Pathogens

March 5–7, 1999
Max Planck Institute for Biology
¨
Tubingen, Germany

Organizers: have provided to help ensure the success of this

meeting:
Christoph Dehio (Germany)
Anna Sander (Germany) ¨
Max-Planck-Gesellschaft, Munchen (Germany)
¨ Hygiene und Mikro-
Deutsche Gesellschaft fur
Scientific Committee: biologie (Germany)
Burt Anderson (U.S.A.) ¨ Allgemeine und Angewandte
Vereinigung fur
Siv G. Andersson (Sweden) Mikrobiologie (Germany)
Christoph Dehio (Germany)
´
Yves Piemont (France) The organizers gratefully acknowledge the contribu-
Didier Raoult (France) tions from the following companies in support of this
Russell L. Regnery (U.S.A.) conference:
Anna Sander (Germany)
BAYER AG, Pharma Forschung, Leverkusen (Ger-
Local Organisation: many)
Michaela Dehio (Germany) BAYER AG, Bayer Vital, Leverkusen (Germany)
¨
BIOS GmbH, Grafelfing ¨
/ Munchen (Germany)
Conference Office: ¨
Genzyme Virotech, Russelsheim (Germany)
¨
Barbel Zeller (Germany) Hoffmann-La-Roche, Grenzach-Wyhlen (Germany)
Karger AG, Basel (Switzerland)
Under the auspices of: Leica Microsystems, Bensheim (Germany)
Department of Infection Biology, Max Planck MRL Diagnostics, Cypress / California (USA)
¨
Institute for Biology, Tubingen in cooperation with Pfizer GmbH, Karlsruhe (Germany)
¨ Hygiene und
the ‘‘Deutsche Gesellschaft fur Sanofi Diagnostics, Freiburg (Germany)
¨
Mikrobiologie’’ and the ‘Vereinigung fur
Allgemeine und Angewandte Mikrobiologie’
2. Abstracts

1. Acknowledgements Abstracts are listed by sessions (I-VI) or keynote

presentations as they appear in the program.
The organizers are indebted to the following In each session, abstracts of main speakers and
organizations for the generous financial support they short oral presentations appear in the chronological

0167-7012 / 99 / $ – see front matter  1999 Elsevier Science B.V. All rights reserved.
PII: S0167-7012( 99 )00073-1
268 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
order of the program and are followed by poster
abstracts in alphabetical order by authors.
The numbers of poster abstracts refer to the
numbers on the poster boards.
Keynote Lecture
Title: The spectrum of Bartonellae host / vector
evolution: challenges and opportunities
for public health intervention
Author: Russell L. Regnery
Institution: Viral and Rickettsial Zoonoses Branch,
Centers for Disease Control and Preven-
tion, Atlanta, GA, U.S.A.
Advances in understanding the diversity and natural
histories of members of the genus Bartonella under-
score the diverse spectra of bartonellae zoonotic
potential and human disease. One extreme can be
represented by the growing list of enzootic agents,
which have no obvious pathologic effect on their
sylvan hosts, may be closely associated with arthro-
pod vectors, and are not recognised to pose large-
scale threats to human health. However, at least
some of these bartonellae can, under the appropriate
circumstances, cause human infections and may be
considered examples of previously unrecognised,
potentially emergent human disease (e.g., Bartonella
elizabethae). Bartonella henselae, the agent of cat-
scratch disease, has a well recognised non-human
vertebrate host (the cat) and apparent arthropod
transmission capability; however, this organism has
well established potential for human infection. Cat-
scratch disease exemplifies a ubiquitous zoonosis.
Lastly, Bartonella quintana and perhaps Bartonella
bacilliformis have no apparent requirement for non-
human vertebrate reservoirs and appear to be agents
of human-to-human, arthropod-transmitted disease.
These diseases are characterised by long-term human
bacteremias and, without the prerequisite for non-
human vertebrate reservoirs, these agents are not (no
longer?) strictly zoonotic. It is clear that Bartonella
quintana and B. bacilliformis qualify for truly epi-
demic potential in the presence of the appropriate
arthropod vectors. No recognised Bartonella spp. are
known to exist without either a non-human reservoir
or an arthropod vector.
These examples appear to represent different
evolutionary themes of a diverse and remarkably
common bacterial genus. Better understanding the
roles of bartonellae zoonotic reservoirs, arthropod
vectors, pathologies, physiologies, and the genetic
diversity of the genus should help to anticipate the
potential for future emergent human disease and
hopefully will provide creative opportunities for
interrupting human disease transmission.
Session I: Clinical Manifestions
Main Speaker
Title: Cat-scratch disease – not only a chil-
dren’s disease
Authors: Anna Sander 1 and Gerd Jurgen
¨ Ridder 2
1
Institutes: Institute for Medical Microbiology and
Hygiene and 2 Department of Otor-
hinolaryngology, University of Freiburg,
Germany
The clinical manifestation of cat-scratch disease
(CSD) was first described more than 50 years ago by
Debre´ in Paris, who reported several patients with
suppurative adenitis following cat scratches. Typical
CSD follows by a characteristic clinical course. The
first clinical manifestation is a primary inoculation
lesion, which appears 3 to 10 days after contact with
a cat and manifests as a vesicular, erythematous or
red-brown papular skin lesion. Usually about 2
weeks after inoculation one or more regional lymph
nodes enlarge up to several centimeters in 2 to 3
weeks. Fever, especially high fever up to 408C is
rather rare, and occurs only in about one third of the
cases. The lymph nodes remain stationary for
another 2 to 3 weeks and resolve in the following
weeks. The usual duration of the disease is about 2
to 4 months. In the late course of infection lymph
nodes may suppurate in about 10 to 15% of CSD
patients. Atypical CSD occurs in 5 to 25% of cases.
Many different organs can be affected: eyes, liver,
central nervous system, skin and bones.
Most cases of CSD are reported by pediatricians,
but CSD is not only a children’s disease. In a
retrospective study, we investigated lymph nodes of
60 patients with histopathologically suspected CSD.
 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
In a total of 42 patients (70%) CSD could be
confirmed by PCR or serologically. More than 50%
of the patients were adults and lymph nodes most
often involved were located on the neck and jaw.
Therefore we investigated in a prospective study
218 patients with unclear infectious masses on the
head and the neck. Thirtynine (18%) of the patients
suffered from CSD, 35 (16%) had other infectious
diseases, 20 (9%) patients had malignant diseases
and 37 (17%) suffered from other non malignant
diseases. Forty percent of the cases cured, but
remained unfortunately undiagnosed. Most cases of
CSD appeared in adults (85%) and 64% of patients
were female.
Even nowadays, a large number of CSD-cases
remain undiagnosed, especially if the infected lymph
nodes are rather small and no other symptoms
apparent. Since CSD is a self limiting disease, lymph
node swelling often disappear without being recog-
nized or diagnosed as a Bartonella infection.
Session I: Clinical Manifestions
Main Speaker
Title: Bacillary angiomatosis and other mani-
festations of Bartonella infections in the
immunocompromised patient
Authors: J.E. Koehler 1 , M. Sanchez 1 , S. Tye 1 ,
C. Garrido 1 , F. Chen 1 , K. Hadley 1 , A.L.
Reingold 1 , R.L. Regnery 2 , L. Cooper 2 ,
J. Olson 2 and J.W.Tappero 2
Institutions: 1 University of California at San Fran-
cisco and Berkeley, San Francisco,
U.S.A. 2 Centers for Disease Control and
Prevention, Atlanta, U.S.A.
Manifestations of Bartonella infection in the im-
munocompromised include lymphadenitis, meningi-
tis, bacteremia, endocarditis and unusual, vascular
proliferative lesions of bacillary angiomatosis (BA)
and bacillary peliosis. BA usually occurs in HIV-
infected patients with CD4 , 50, accompanied by
bacteremia and fever; diagnosis occurs after de-
tection and biopsy of the striking BA lesions.
Isolated Bartonella bacteremia probably occurs
much more frequently than BA, especially in HIV-
269
infected patients with fever of unknown origin
(FUO), but in the absence of cutaneous disease,
either remains unsuspected, or undetected due to
difficult culture techniques. We thus hypothesized
that Bartonella infections represent an under-recog-
nized cause of fever in late-stage HIV infection.
We conducted a study in SF to determine the
prevalence of Bartonella infections among prospec-
tively identified, HIV-infected patients with 1) acute
febrile illness (T . 388C) without a known source, or
2) FUO, defined by persistent or recurrent fever of
. 2 weeks. Patient sera were tested for presence of
Bartonella antibodies by IFA at CDC. Blood cul-
tures were performed by the lysis-centrifugation
method.
We evaluated 433 patients: 61 (14%) had evidence
of Bartonella infection by culture, IFA or both.
Either B. henselae or B. quintana was cultured or
DNA amplified from the blood, tissue or both, of 12
patients (3%), 5 of whom did not have BA or focal
disease. MAC was recovered from 14% and Histop-
lasma from 1.6% of patients’ blood. These data
indicate that the prevalence of Bartonella infection is
much greater than previously documented and Bar-
tonella infection should be considered in the dif-
ferential diagnosis of HIV-infected patients with
FUO.
Session I: Clinical Manifestions
Main Speaker
Title: Clinical aspect of Bartonella quintana
including endocartitis
Author: Didier Raoult
Institution: Faculte´ de Medecine,
´ Unite´ des Ricket-
tsies, Marseille, France
Bartonella quintana is the agent of trench fever. It is
transmitted by body lice and determines several
clinical forms of infectious diseases. The disease was
forgotten until 1990 when it was reported as a cause
of Bacillary angiomatosis. In patients infected with
lice it could determine a classic trench fever, with
fever headaches, leg pains, which could relapse. It
could also determine, after or not a typical trench
fever, a chronic bacteremia being pauci or asympto-
270 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
matic without fever. This chronic bacteremia could
lead to an endocarditis even in patients without
previous valvulopathy. Currently 15% of asympto-
matic homeless in Marseille are bacteremic in win-
ter. Bartonella quintana has been found in lice in
France, Zimbabwe, Burundi, Peru and Russia and
therefore the disease is prevalent every where. In
patients with trench fever serology is usually posi-
tive, it is negative during chronic bacteremia and
positive at high level during endocarditis. In patients
with AIDS, Bacillary angiomatosis could be ob-
served but not peliosis and in our laboratory serology
is always negative include cases.
We observed two cases of chronic adenopathies in
patients bite by cat fleas with negative serology in
whom B. quintana was isolated.
Finally the spectrum of the disease caused by B.
quintana is extending and is probably incompletely
known.
Session I: Clinical Manifestions
Oral Presentation / Poster [1
Title: Ocular manifestations of Bartonella in-
fections
T. Ness 1 , M. Punder 1
Authors: ¨ and A. Sander 2
1
¨
Institution: Universitats-Augenklinik Freiburg,
Freiburg, Germany; 2 Institut fur
¨ Med.
Mikrobiol. und Hygiene der Universitat¨
Freiburg, Freiburg, Germany
Purpose: Parinaud oculoglandular syndrome and
neuroretinitis are classical clinical entities caused by
Bartonella species. We report on clinical mani-
festations of Bartonella infections based on the
retrospective review of the records of seven patients.
Case reports: Serologic findings of all 7 patients
revealed the evidence of Bartonella infection by a
typical IgG course and / or by the detection of IgM
antibodies. The age of our patients ranged from 10
years to 30 years. Only one female patients suffered
from the typical Parinaud ocularglandular syndrome
with facial lymphadenopathy, follicular conjunctivi-
tis, and a chronic course. Symptoms resolved within
3 months. Ptosis of the affected eye was the only
sequel.
The ocular findings in the other six patients were
completely different. All of them suffered from
decrease of visual acuity in one eye. Opthalmoscopic
findings included mild vitritis (4 / 6), disc edema
(6 / 6), retinal edema (5 / 6), intraretinal infiltrates
(3 / 6), dilated retinal vessels (5 / 6), and macular star
figure (5 / 6). Three of the six patients reported a
flu-like illness 1 to 2 weeks before onset of the
ocular symptoms. Associated findings were impaired
color vision and afferent pupillary defect in all
patients. Fluorescein angiography showed late leak-
age of the optic disc and the retinal vessels. After
treatment with doxycyclin or acithromycin visual
acuity recovered and symptoms resolved in 3 of the
six patients. In two patients the clinical course was
self-limiting without any therapy. One patient was
lost for follow-up.
Conclusion: Unilateral neuroretinitis is the most
common ocular feature of an infection with Bar-
tonella species. It occurs in young-to-middle-age
adults with acute loss of visual acuity. The prognosis
is excellent with antibiotic treatment but may also be
good without treatment.
Session I: Clinical Manifestions
Oral Presentation / Poster [2
Title: Epidemiology of Bartonellosis in Peru:
prospective population-based study
Authors: L. Laughlin, J. Chamberlin, C. Ponce, H.
Gonzales, S. Romero, A. Gonzalo, D.
Watts and C. Carillo
Institutions: Uniformed Services University of the
Health Sciences, Bethesda, MD, USA;
Peruvian Ministry of Health, Peru;
Naval Medical Research Institute De-
tachment, Peru
Bartonellosis is a highly fatal epidemic and endemic
infectious disease that occurs throughout the medi-
cally underserved communities of the Andes Moun-
tains in South America. Disease manifests as a 3-
stage clinical infectious disease caused by Bartonella
bacilliformis (Bb), an aerobic, motile alpha-2
Proteobacteria that intracellularly infects the human
red blood cell and reticuloendothelial cells. The
emergence of bartonellosis in new geographic areas,
 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
the increase in number of reported cases, and the
changing clinical manifestations justify further study.
A prospective, population-based study to define
the mechanism(s) of disease transmission in bartonel-
losis was initiated in January 1997 in a small
mountain village in the Caraz District of Peru.
Annual house to house surveys were carried out in
four contiguous villages with a population of over
1600. All volunteers answered questionnaires and
contributed a blood sample. 41% of the individuals
in the study population participated, representing
60% of study households. 42% gave a history of
either the acute febrile form or the chronic verrucous
form of bartonellosis at some time in the past. 12.5%
gave a history of bartonellosis during the past 1 year.
Two-thirds of recent cases occurred in children less
than 10 years old. 89% of cases occurred during the
time period of January to June. In an effort to
discover if the human was a silent reservoir host,
patients without any manifestations of disease were
blood cultured to detect asymptomatic bacteremia.
Four of over 600 (0.6%) cohort patients were
positive for Bb; 14% (3 / 21) of patients who had
been treated were bacteremic 2-12 months after
clinical cure; 22% (6 / 27) of verrucous rash patients
were bacteremic. We believe that the asymptomatic,
but bacteremic patient plays an important role in
disease transmission.
Session I: Clinical Manifestions
Poster [3
Title: Bartonella vinsonii subspecies Berkhoffii
and other alpha proteobacteria associated
with cardiac arrhythmiasis, endocarditis
or myocarditis in dogs
Authors: Edward B. Breitschwerdt, Clarke E.
Atkins and Talmage T. Brown
Institutions: College of Veterinary Medicine, North
Carolina State University, Raleigh, NC
27606, U.S.A.; University of Florida,
Gainesville, FL 32610, U.S.A.
Cardiac arrhythmias, endocarditis or myocarditis
were identified in 16 dogs, of which 15 were
seroreactive to Bartonella vinsonii subspecies ber-
271
khoffii antigens. Historical abnormalities were highly
variable among these dogs, but frequently included
substantial weight loss, syncope, collapse or sudden
death, vomiting and diarrhea, or lameness. Fever was
an inconsistently detected abnormality. Cardiac dis-
ease was diagnosed following an illness of short
duration in most dogs, but a protracted illness of at
least 6 months duration was reported in 5 dogs.
Valvular endocarditis was diagnosed echocardiog-
raphically and / or histologically in 12 dogs, 2 of
which also had moderate to severe, multifocal
myocarditis. Three dogs, lacking definitive evidence
of endocarditis, were included because of uncharac-
terized heart murmurs and arrythmias and another
dog had valvular endocardiosis accompanied by
occasional left ventricular myocardial inflammatory
foci. Thrombocytopenia, mild neutrophilia, and
monocytosis were the most frequent hematologic
abnormalities. Hemoglobin and protein were de-
tected in the urine of most dogs. Alpha proteobac-
teria were not isolated from the blood using either
conventional or lysis centrifugation blood culture
techniques. Using PCR amplification and DNA
sequencing of a portion of the 16S rRNA gene B.
vinsonii was identified in the blood or heart valve of
3 dogs. DNA sequence alignment of PCR amplicons
derived from blood or tissue samples from 7 dogs
clustered among members of the alpha subdivision of
proteobacteria, however, isolation efforts were not
successful and the limited quantity of sequence
obtained did not allow for genus identification.
Serologic or molecular evidence of concurrent in-
fection with other tick-transmitted pathogens, includ-
ing Ehrlichia canis, Babesia canis, Babesia gibsonii,
or spotted fever group rickettsiae was obtained for 8
dogs. We conclude that Bartonella vinsonii sub-
species berkhoffii and other alpha Proteobacteria
species are an important, previously unrecognized,
cause of arrhythmias, endocarditis, myocarditis,
syncope and sudden death in dogs.
Session I: Clinical Manifestions
Poster [4
Title: Detection of two strains of Bartonella
henselae (Bh) by cultural and PCR-
based methods: first reports in Italy
272 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
Authors: A. Fabio, L. Bonazzi, A. Tuzzi, L. Serra,
P. Terenziani, B. Casali, E. Farnetti and
U. Fabio
Institution: Arcispedale S. Maria Nuova, Reggio
Emilia, University of Modena, Italy
We report the recovery of two Bh strains associated
with Bacillary Angiomatosis (BA) (1st case) and
Cat-Scratch-Disease (CSD) (2nd case). ]]] 1st case: a
37-years old drug-user HIV-infected male, presented
high recurrent fever and multiple nodular papules
over the whole integument. Histological examination
of a skin biopsy specimen showed characteristic
feature of BA; Warthin-Starry staining revealed
cluster of small bacilli. Hepatic ecography showed a
visceral involvement. Patient serum yelded an high
diagnostic titre (1:512), when tested in an indirect
IFA to Bh. Three blood samples showed bacterial
growth after 35 days of incubation in Bactec bottles
(Bactec Alert NR730, Becton-Dickinson) and the
organisms were subcultured 4 days in 5%CO 2 at
378C. 2nd case: a 34 years-old female was admitted
]]]
to hospital presenting a high (39–408C) recurrent
pyrexia without any clinical sign except a skin rash.
Since she owed a kitten and had been frequently
scratched, CSD was suspected. A pelvic-abdominal
TAC put in evidence ipodense lesions of flogistic
nature. Two serum samples presented a fourfold rise
in Bartonella antibody titers. Three patient’s blood
samples were negative after 40 days of incubation,
while a kitten blood sample showed microbial
growth under the same condition reported for the
first.
The two strains recovered from blood samples
(patient: 1st case ; cat: 2nd case) were identified by
morphological, biochemical and molecolar analysis.
DNA was extracted from agar-grown bacterial cells
and PCR was performed by using
CAT1 59-GATTCAATTGGTTTGAAG-
GAGGCT-39 and CAT2 59-TCACATCACCAG-
GACGTATTC-39 oligonucleotide primers under the
conditions described by Anderson et al. (J. Clin.
Microbiol. 32:942-948, 1994). Positive control con-
sisted of reaction mixture containing DNA of Bh
TA-2 gift by Dr. Avidor, Tel Aviv, Israel. In both
samples was visualized with electrophoresis a 414-bp
band product, Bh specific. To detect DNA of Bh in
the feline blood sample, we set up an semi-nested
PCR (Mouritsen et al., Hum. Pathol. 28:820-826,
1997) where CAT1 and CAT2 were used as outer
oligonucleotide primers and RH1 (59-
GGTGCGTTAATTACCGATCC-39) and CAT2
were used as inner oligonucleotide primers. DNA
from feline blood sample showed an amplified
product of 390 bp Bartonella henselae specific.
Session I: Clinical Manifestions
Poster [5
Title: Clinical manifestations of cat-scratch
disease in the head and neck
Authors: ¨
Gerd Jurgen Ridder 1 , Anna Sander 2 ,
Bernhard Richter 1
Institutions: 1 Department of Otorhinolaryngology,
and 2 Institute of Medical Microbiology
and Hygiene, University of Freiburg,
Germany
Bartonella henselae is the causative agent of cat-
scratch disease (CSD), an inflammatory infection of
the lymph nodes. The predominantly affected three
anatomic sites in CSD are the axillary (46%), the
head and neck (26%) and the inguinal (17%) lymph
nodes. Enlarged cervical, submandibular, or preau-
ricular lymph nodes are typical for CSD in the head
and neck region. So far, only few cases of atypical
manifestations have been reported. The aims of this
study were: (1) to investigate the frequency of CSD
by systematical screening of all patients having
unclear infectious masses in the head and neck; (2)
which kinds of clinical manifestations of CSD other
than typical lymphadenitis colli can be detected. In a
prospective clinical study (January 1997 to De-
cember 1998) the frequency and the clinical pre-
sentations of CSD were analyzed in 218 patients
with suspected infectious head and neck masses.
CSD was diagnosed serologically by an indirect
immunofluorescence test, histologically and / or by
detection of B. henselae DNA in extirpated lymph
nodes and was found in 39 cases (18%). Cervical,
submandibular or preauricular lymphadenitis
occurred in 60%, 20% and 20%, respectively. Acute
lymphadenitis was diagnosed in 22 patients, chronic
lymphadenitis in 6 patients and abscessed
lymphadenitis in 11 patients, respectively. Seven of
 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
the 39 patients had fever. In 14 patients unusual
manifestations were found, as enlargement of the
parotid gland (n 5 6), pharyngitis (n 5 2), Parinaud’s
oculoglandular syndrome (n 5 3), swelling of the
submandibular gland (n 5 1), perichondritis of the
auricle (n 5 1), or glossitis (n 5 1). The results of our
study revealed a higher than expected prevalence of
CSD (18%) in patients with unidentified enlarged
lymph nodes in the head and neck. About one third
of the patients with CSD presented unusual symp-
toms in addition to the typical lymphadenopathy.
Based on these findings, B. henselae should no
longer be considered a rare cause of infections in the
head and neck. CSD should be included in the
diagnostic approach in unclear infectious masses in
the head and neck.
Session I: Clinical Manifestions
Poster [6
Title: Disseminated cat-scratch disease in chil-
dren: report of 2 cases
Authors: M. Ruess 1 , A. Sander 1 , M. Brandis 2 , R.
Berner 2
Institution: 1 Institute for Medical Microbiology and
Hygiene; 2 Children’s Hospital, Universi-
ty of Freiburg, Germany
Cat-scratch disease (CSD) is a common cause of
subacute regional lymphadenopathy with typical
clinical manifestations and a history of cat contact.
Atypical (disseminated) CSD occurs in 5-25% of
cases with various symptoms and may be difficult to
diagnose. It often presents as prolonged fever ( . 2
weeks), malaise, fatigue, night sweat, myalgia, ar-
thralgia, weight loss, skin eruptions, hepato-
splenomegaly, generalized lymphadenopathy, pleural
effusion, osteolytic lesions, thrombocytic purpura
and hemolytic anemia.
Case 1 : A previously healthy, 12-year-old girl
developed 3 weeks after measle infection fever (38.5
to 40.08C for 4 weeks), enlarged submandibular and
axillar lymph nodes, pharyngitis, abdominal pain,
back pain, diffuse myalgia, weight loss and hepato-
splenomegaly. At admission the leukocyte count was
5900 / ml and the C-reactive protein 7.7 mg / dl. An
273
abdominal MRI-scan of the abdomen showed multi-
ple abscesses in liver and spleen, and a technetium
isotope scan of the skeleton revealed multiple de-
posits in a lumbar vertebra, corresponding to
spondylitis. A serum sample taken at the time of
admission showed high titres of antibodies to Bar-
tonella henselae by immunofluorescence (IgM 1024
and IgG 8000). After therapy of three weeks with
erythromycin the patient improved, was afebrile and
in a good general condition. Three months later B.
henselae titers had decreased (IgM 128, IgG 4000).
Six months later no IgM antibodies were detectable,
IgG antibodies were 2000.
Case 2 : A 2.5 year old boy suffering from
abdominal pain presented with a great enlargement
of the spleen. A CT-scan revealed thrombosis of the
portal vein and several intra-abdominal lymph nodes.
Three weeks later the boy developed ascites. A
serum sample obtained 6 weeks after onset of illness
presented high titers of antibodies to B. henselae
(IgM 128, IgG 1024). The abdominal lymph nodes
were resected and B. henselae DNA was detected by
PCR. After four months the ascites and the spleno-
megaly disappeared, and clinically the boy had
recovered completely. In a second serum sample
taken at this time, antibodies to B. henselae de-
creased (IgM , 64, IgG 512).
Conclusion: Disseminated CSD has to be taken
into account in children presenting with a systemic
illness which can resemble hematological malig-
nancy or systemic juvenile chronic arthritis.
Session II: Diagnostic Procedures
Main Speaker
Title: Bartonella isolation and identifications
Author: Russell L. Regnery
Institution: Viral and Rickettsial Zoonoses Branch,
Centers for Disease Control and Preven-
tion, 1600 Clifton Road, Atlanta, GA
30333, U.S.A.
Isolation and identification of members of the
genus Bartonella are necessarily interdependent;
without isolation there is no agent in pure culture to
identify and, without the means to identify the
274 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
isolate, the isolate remains unrecognised. This pre-
sentation summarises observations regarding bar-
tonellae isolation and identification.
Isolations of bartonellae are typified by relatively
long incubation periods. Rich, hemin-containing
media appears to be required for axenic growth. A
variety of bacteriologic media have been used for
Bartonella spp. isolation, however, colony morphol-
ogy may vary with the media. Some agents (e.g.,
Bartonella henselae) require a CO 2 enriched atmos-
phere for growth whereas others do not (e.g., Bar-
tonella bacilliformis). Growth of some species is
relatively temperature insensitive (e.g., B. henselae)
whereas growth of B. bacilliformis in culture is
temperature-sensitive. Data suggest that there is
differential expression of B. henselae antigens when
the organism is cultivated with eukaryotic cells as
compared to when grown axenically on agar. Blood
lysis greatly enhances recovery of B. henselae from
infected blood and significantly reduces the incuba-
tion period prior to visualisation of colonies. All
Bartonella spp. tested to date are acutely sensitive to
antibiotics in vitro, and although anti-fungal com-
pounds can be used during isolation, further attempts
to develop antibiotic-based selective media have not
been successful.
Although relatively few Bartonella spp. are
known to infect humans, a wide variety of bartonel-
lae exist in nature as potential sources of emergent
human disease. Bartonellae isolates are best defini-
tively identified with genotypic methods. The
genotypic methods used have been varied, however,
in general, the most information-rich methods rely
on DNA sequence analysis of polymerase chain
reaction-amplified gene fragments. Although such
methods have made possible informative Bartonella
phylogenies, the diversity of bartonellae complicates
development of rapid, reliable, and specific identifi-
cation methods for routine, hospital laboratory-based,
identification of bartonellae genotypes.
Session II: Diagnostic Procedures
Main Speaker
Title: Serological diagnosis of cat scratch dis-
ease
Author: Reinhard Zbinden
Institution: Department of Medical Microbiology,
University of Zurich, Zurich, Switzer-
land
Serology to Bartonella henselae is the simplest
way to diagnose cat scratch disease (CSD). We
earlier described an in-house indirect immuno-fluo-
rescence assay (IFA) based on a 2-day cocultivation
of B. henselae with Vero cells on chamber slides;
transmission electron microscopy revealed clusters of
multiple intracellular organisms around the nuclei of
Vero cells. This in-house IFA revealed in 17 of 20
patients with clinically diagnosed CSD IgG titers of
1:512 or higher. The other 3 patients revealed titers
of 1:1024 three weeks later. In contrast, all but one
of the 321 controls had lower IgG titers. In the
meantime, we have replaced our in-house test system
with commercial slides with Vero-cell associated
bacilli for detection of IgG to B. henselae and found
in our mixed urban / rural population a sensitivity of
84.6% and a specificity of 93.4% using a cutoff titer
of 1:256.
Since we could show that our in-house Vero cell-
associated IFA was not useful to detect IgM to B.
henselae because of false-positive results in blood
donors and in patients with lymphadenopathy not
due to B. henselae, we now use commercial slides
with agar-derived B. henselae that yield a sensitivity
of 70% for the detection of IgM in patients with
CSD. Furthermore, we could demonstrate by Western
blot that sera from patients with EBV-VCA-IgM
showed strong reactions against B. henselae coculti-
vated with Vero cells but less against agar-derived B.
henselae.
Serodiagnosis is still hampered by cross-reactive
antibodies and improvement of antigen preparations
is needed. Further investigations of the use of
recombinant-expressed antigens or immunoaffinity-
purified antigens in ELISAs as well as incorporation
of different B. henselae serotypes might be of value
for the serologic diagnosis of Bartonella infections.
Despite those pitfalls, detection of IgG and IgM to B.
henselae could replace traditional diagnostic criteria
for the diagnosis of CSD in patients with
lymphadenitis and prevent them from unnecessary
surgery, but histology and PCR may still be neces-
sary in atypical clinical situations.
 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
Session II: Diagnostic Procedures
Main Speaker
Title: Diagnosis of Bartonella infections: a
molecular approach
Authors: F. Dutly and M. Altwegg
Institution: Department of Medical Microbiology,
¨
University of Zurich, ¨
Zurich, Switzer-
land
Laboratory diagnosis of infections due to B. quintana
as well as B. henselae can be performed by culture,
serology, and / or molecular methods, all of them
having advantages and disadvantages. With amplifi-
cation methods either of two completely different
approaches can be used. Usually sterile samples may
be analyzed using a broad-spectrum PCR with
primers complementary to conserved regions of the
16S rRNA gene followed by sequencing. This tech-
nique is extremely helpful to detect rare or un-
suspected cases of Bartonella infections such as B.
quintana-associated cases of endocarditis. Using this
approach Bartonella species and at least two sub-
types within B. henselae (genotypes I and II) were
found. Bartonella specific-PCRs are more sensitive
than broad-spectrum PCR but require that Bartonella
infection is suspected by the clinician. They involve
the use of genus-, species- or type-specific primers
for amplification followed by (semi-) nested reampli-
fication or hybridization with specific oligonucleo-
tides. Amplification of rRNA targets has also been
used in epidemiological investigations concerning
the prevalence of B. henselae genotypes I and II in
human and animal specimens. It is not yet estab-
lished whether or not these two types differ in their
clinical and / or epidemiological significance except
for the tendency towards a more severe disease in
patients infected with B. henselae type I.
Session II: Diagnostic Procedures
Oral Presentation / Poster [7
Title: Diagnosis of human bartonellosis by
indirect fluorescence antibody test
275
Authors: J. Chamberlin 1 , S. Gordon 1 , L.
Laughlin 1 , R. Regnery 2 , J. Olson 2 , S.
Romero 3 and A.Gonzalo 3
Institutions: 1 Uniformed Services University of the
Health Sciences, Bethesda, MD, U.S.A.
2
CDC, Atlanta, GA, U.S.A.; 3 Naval
Medical Research Institute Detachment,
Lima, Peru
Background: Diagnosis of South American human
bartonellosis is problematic and is based on clinical
impression and demonstration of the in-
traerythrocytic bacilli on Giemsa or Wright stained
thin blood smear. Culture of Bartonella species is
difficult, requiring special media and techniques and
up to 8 week’s incubation time. There are no
generally accepted serologic assays available to
confirm clinical suspicion of disease. We report the
development of an indirect fluorescence antibody
(IFA) test for detection of antibodies to Bartonella
bacilliformis.
Methods: A Peruvian isolate of B. bacilliformis
was co-cultivated with Vero cells. Antigen and
antisera were prepared for IFA testing by standard
techniques. Fluorescein-labeled affinity purified anti-
body to human IgG(H 1 L) was used in all tests.
Sera from 43 patients with slide or culture-confirmed
B. bacilliformis infections, from 101 healthy con-
trols, and from 356 volunteers from an area of Peru
endemic for bartonellosis were tested. Sera from
patients with 8 other diseases were assessed for cross
reactive antibodies.
Results and Interpretation: Using a serum dilution
cut-off point of 256 or higher, 32 (74%) of the
Bartonella patients and 38% of 356 volunteers were
seropositive for B. bacilliformis. 93 (92%) of healthy
control sera were seronegative. One of 2 patients
with cat-scratch disease, and one of 2 patients with
secondary syphilis had titers . 256. The IFA is 74%
sensitive and 92% specific in detecting antibodies to
B. bacilliformis as compared to a gold standard of
culture-confirmed infection. Since many patient sera
were drawn during the acute phase of disease,
demonstration of a rise in titer during convalescence
may help improve the sensitivity of the test. The IFA
may be a useful diagnostic screening tool in com-
munity surveys. As measured by IFA, the point
276 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
prevalence of infection in an endemic area of Peru is
38%.
Session II: Diagnostic Procedures
Oral Presentation / Poster [8
Title: Bartonella henselae IFA serology: ob-
server concordance and the effect of
utilizing wild-type substrate
Authors: N. Cimolai, L. Benoit, A. Hill and C.
Lyons
Institution: Children’s and Women’s Health Centre
of British Columbia, Vancouver, Canada
With the implementation of IFA serology for B.
henselae in the context of cat scratch disease, it
should be anticipated that the general problems
which have been historically experienced with other
IFA serodiagnostic techniques would be revisited.
Among the many potentially relevant factors, inter-
observer variability and the source of the antigen
may be problematic. We have developed an IgG-IFA
using acetone-fixed whole cells which are obtained
from growth on blood-based media. We applied this
technique to the screening of 333 sera from both
adult and pediatric sources, and then blinded two
observers to the sources and to the alternate interpre-
tations. In addition, we assessed the use of two
different antigen substrates: B. henselae ATCC
49793 (Oklahoma) and a wild-type low passage
isolate B. henselae-129 which was acquired from the
Pacific Northwest of Canada. Among the 333 sera,
the observer variability was nil, 2-fold, 4-fold, and
8-fold for 59.2%, 33.3%, 7.2%, and 0.3% respective-
ly. The two B. henselae substrates were assessed
with the use of 68 adult donor sera. In total, there
were 12 / 68 (17.7%) that had $ 4-fold variation and
1 / 68 (1.5%) that demonstrated $ 8-fold variation.
For 24 of the latter sera, there was no difference in
IFA titre; for 16 / 68 (23.5%), the titre was greater
with the ATCC strain substrate versus 28 / 68
(41.2%) for which the titre was greater with the
wild-type substrate. The reproducibility of the IFA is
consistent with that which is seen in other infectious
disease IFA serodiagnostic assays. There is a trend
towards increasing IFA titres when locally acquired
wild-type substrate is utilized.
Session II: Diagnostic Procedures
Poster [9
Title: Will endemic seropositivity for Bar-
tonella henselae lead to considerable
overdiagnosis?
Authors: N. Cimolai, L. Benoit, A. Hill and C.
Lyons
Institution: Children’s and Women’s Health Centre
of British Columbia, Vancouver, Canada
The application of serodiagnostic assays is compli-
cated by the potential for endemic seropositivity, the
magnitude of which may have great impact on the
perceived positive predictive value. We assessed
seroprevalence in a pediatric population that suffered
from well-defined episodes of an illness which was
regarded predominantly as an upper and lower
respiratory disease. Sera were acquired prospectively
from 54 children, age range 1-16 (mean 5.5 yrs.).
Sera were evaluated with an IgG-IFA which included
a wild type B. henselae low passage ([129, British
Columbia, Canada) strain. The bacterium was ace-
tone fixed. Final diagnoses for these patients were
mainly of an infectious disease and included the
following spectrum: respiratory syncytial virus (10),
influenza (9), parafinfluenza (3), pertussis (4),
rhinovirus (4), adenovirus (4), Epstein-Barr virus
(6), bacterial pneumonia (6), and other (8). Overall,
there was a high frequency of elevated titres among
sera but none . 1 / 512. For titres $ 1 / 128, $ 1 /
256, and $ 1 / 512, the frequencies of reactive sera
were 33.3%, 18.5%, and 7.4% respectively. Age-
dependent variation was not appreciable (e.g. IFA $
1 / 256 in 20.7% and 16.0% for those # 5 and . 5
yrs. respectively). Sera from patients with cat scratch
disease have ranged from 1 / 128 to 1 / 2048 when the
same antigen and methods are used. The high
frequency of IFA titres among this patient population
implies that there is the potential to considerably
overdiagnose B. henselae infection if serology is
frequently applied to patients who have a low pre-
 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
test probability for active infection. Confirmatory
serological tests should be developed with the aim of
enhancing specificity.
Session II: Diagnostic Procedures
Poster [10
Title: Serological and epidemiological analysis
of the prevalence of Bartonella spp.
antibodies among swedish elite orien-
teers 1992-93
Authors: Svena McGill 1,2 , Lars Wesslen´ 1 , Eva
1 1
Hjelm , Martin Holmberg and Goran ¨
Friman 1
Institution: 1 Department of Infectious Diseases and
Clinical Microbiology, Uppsala Uni-
versity Hospital, Uppsala, Sweden; 2
Division of Viral and Rickettsial Dis-
eases, Centers for Disease Control and
Prevention, Atlanta, Georgia, U.S.A.
The popular, physically-demanding, and highly na-
ture-interactive sport of orienteering in Sweden was
marked by an elevated rate of sudden unexpected
deaths (SUDs) among its competitors during the
years 1979-1992, with (a) common underlying
cause(s) suspected. Subsequently, sera was collected
during 1992-93 from among the athletes comprising
the elite segment of orienteers holding a nationally-
ranked position, and a corresponding suvey compil-
ing various epidemiological data was obtained. In
this study, a total of 1136 sera, representing 766
male and 370 female elite orienteers (a proportion
reflective of the overall male: female ratio within the
sport), were scrologically analyzed by indirect-
fluorescent antibody (IFA) assay for the presence of
IgG antibodies against 3 Bartonella spp.: B. hen-
selae, B. elizabethae and B. quintana. Of the sera
tested, 31% (355 / 1136) were seropositive for at least
one species of Bartonella, with titers ranging up to
1 / 512. Three-hundredfifty of 1136 (30,8%) had
antibodies against B. elizabethae; 34 / 1136 (3%)
against B. henselae, and 16 / 1136 (1,5%) against B.
quintana. A 5% seropositivity to Bartonella spp.
among a control group of 100 blood donor sera was
used for comparison. Despite the unequal composi-
277
tion of males and females within the sample group,
both evidenced an equal rate of seropositivity to
Bartonella spp., with 31% of males (184 / 355) and
31% of females (114 / 1136) manifesting antibodies
against Bartonella. Among the orienteers with posi-
tive Bartonella serology, 53% (184 / 355) reported to
have been bitten by a tick.
The found markedly high prevalence of Bartonella
spp. antibodies in Swedish orienteers may be indica-
tive of a connection with increased susceptibility to
SUD or other risk factors and renders further in-
vestigation of importance.
Session III: Treatment & Prevention
Main Speaker
Title: Evaluation of antimicrobial treatment of
Bartonella henselae infections
Authors: James W. Bass 1 , Bonnie Cary Freitas 1 ,
Alexander D. Freitas 1 , Cheryl L. Sisler 1 ,
Debora S. Chan 1 , Judy M. Vincent 1 ,
Donald A. Person 2 , John R. Claybaugh,
Robert R. Wittler, Martin E. Weisse 3 ,
Russell L. Regnery 4 , Leonard N. Slater 5
Institutions: 1 Tripler Army Medical Center, Hon-
olulu, HI; U.S.A.; 2 University of Kansas
School of Medicine; West Virginia;
3
University School of Medicine, U.S.A.;
4
Viral and Rickettsial Zoonoses Branch,
Centers for Disease Control and Preven-
tion; U.S.A. 5 University of Oklahoma
Health Science Center and Veterans
Affairs Medical Center; U.S.A.
The most common clinical presentation of Bartonella
henselae infection is now considered ‘‘typical’’ cat-
scratch disease (CSD). Several reports have shown
rifampin, trimethoprim / sulfamethoxazole, amino-
gylcosides, fluoroquinolones, tetracyclines, and
macrolides to be effective in the treatment of typical
CSD; however, most have been anecdotal observa-
tions involving only a few patients. The macrolide,
azithromycin, has been shown to be uniformly active
in vitro at very low concentrations against B. hen-
selae, its pharmacologic properties allow for once
daily dosing for a five day course, and the drug
278 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
concentrates in the tissues where B. henselae organ-
isms are found in greatest number in CSD.
We studied the efficacy of azithromycin in the first
prospective, double blind, placebo-controlled, ran-
domized trial for the treatment of patients with
typical CSD. Eligible patients included healthy milit-
ary members or their dependents with laboratory-
confirmed, clinically typical CSD. Endpoint evalua-
tions were predetermined as time in days to 80%
resolution of the initial total lymph node volume as
measured by three dimensional ultrasonography.
An 80% decrease in initial lymph node volume
was documented in 7 / 14 azithromycin-treated pa-
tients compared with 1 / 15 placebo-treated patients
during the first 30 days of observation (P 5 0.026).
We concluded that treatment of patients with typical
CSD with oral azithromycin affords significant clini-
cal benefit as measured by total decrease in lymph
node volume within the first month of treatment.
The effect of antibiotics in the treatment of typical
CSD is still controversial. Additional studies are
needed to confirm our findings and to further de-
lineate the effect of antibiotics in the treatment of
typical CSD.
Session III: Treatment & Prevention
Main Speaker
Title: Antibiotic therapy of Bartonella infec-
tions: current knowledge from in vitro
and in vivo data.
Authors: Max Maurin and Didier Raoult
Institution: Unite´ des Rickettsies, Faculte´ de
´
Medecine, Marseille, France
Five Bartonella species are considered potential
human pathogens: Bartonella bacilliformis, the agent
of Carrion disease which is restricted to the Andean
valleys of South America, and species of wider
distribution including Bartonella quintana, Bartonel-
la henselae, Bartonella elizabethae, and Bartonella
clarridgeiae. Infection with Bartonella species may
be responsible for acute manifestations with bac-
teremia (referred as Oroya fever for B. bacilliformis),
and may evolve to chronic infection of the skin
(cutaneous bacillary angiomatosis and verruga
peruana) and / or internal organs. Bartonella sp. may
be grown on blood-enriched agar, but may infect
erythrocytes and endothelial cells both in vitro and in
vivo. The in vitro antibiotic susceptibilities of nine B.
quintana strains, three B. henselae strains, B.
elizabethae F9251, and four B. bacilliformis strains
were tested, using a modified version of the anti-
biotic dilution method adapted to the fastidious
growth of these bacteria. All Bartonella strains
displayed similar susceptibility patterns and were
highly susceptible to most antibiotics tested. How-
ever, only the aminoglycosides displayed significant
bactericidal activity against these pathogens grown
either in axenic medium or in endothelial cells. In
vivo, data on the efficacy of antibiotics to treat
Bartonella infections are disappointing and may vary
according to the disease considered. Cat scratch
disease is highly resistant to antibiotic therapy.
Penicillin G, chloramphenicol, cotrimoxazole, tetra-
cyclines, macrolides, and fluoroquinolones have been
used successfully to treat patients with bacillary
angiomatosis or Oroya fever. However, therapeutic
failures and relapses on antibiotic withdrawal have
been reported in immunocompromised- and / or
chronically infected-patients, suggesting that anti-
biotics were only bacteriostatic. In such patients,
MICs may not be predictive of in vivo efficacy of the
antibiotic therapy, both because Bartonella species
are facultative intracellular bacteria, and because
antibiotics with bactericidal activity may be needed
to eradicate Bartonella sp. infection. In vitro, only
the aminoglycosides display such properties. Strep-
tomycin is the first line antibiotic to treat verruga
peruana and gentamicin has been used successfully
in combination with another antibiotic in many
patients involved with Bartonella endocarditis.
Session III: Treatment & Prevention
Main Speaker
Title: Feline vaccination for the prevention of
cat scratch disease
Authors: J.A. Rooney 1 , K.A. Dubois 1 , J.G.
Olson 2 and R.L. Regnery 2
Institutions: 1 Heska Corp., Ft. Collins, CO, U.S.A.
2
Viral & Rickettsial Zoonoses Branch,
Centers for Disease Control and Preven-
tion, Atlanta, GA, U.S.A.
 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
There is substantial epidemiologic evidence that cats
are the zoonotic reservoir for cat-scratch disease and
vectors for human infections with B. henselae. Most
feline Bartonella infections show no overt clinical
signs, although recent reports have described tran-
sient fever and lymphadenopathy. Cats can remain
bacteremic for prolonged periods of time. Antibiotic
treatment of cats, under controlled conditions, has
not proven effective in clearing B. henselae bac-
teremias.
Prior studies indicate that convalescent cats ex-
perimentally infected with B. henselae are resistant
to subsequent re-challenge with live B. henselae.
Experiments were designed to evaluate whether a
feline B. henselae bacterin would prevent the estab-
lishment of bacteremia, subsequent to challenge.
Preliminary studies were performed to ascertain the
minimal infectious dose of live B. henselae in cats
and determine the range of inactivated antigen
required to elicit an immune response. Cats from the
antigenic dose study, as well as 2 non-immunized
control cats, were subsequently challenged with live
organisms to determine whether immunization with
inactivated B. henselae antigen protected against
infection. The majority became bacteremic, although
high- and medium-antigen dose animals remained
abacteremic for longer periods than did control and
low-antigen dose cats. Since killed whole cell bac-
terin alone did not stimulate a long term protective
response, six different adjuvants with a variety of
immune stimulating functions were formulated with
the inactivated bacterin and screened for efficacy in
enhancing protection of cats against challenge with
live B. henselae. One of these adjuvants, poly[di-
(carboxylatophenoxy)phosphazene] (PCPP), was se-
lected for further evaluation. Results of two experi-
ments demonstrated 96% efficacy against challenge
in cats immunized with the PCPP adjuvanted, killed
bacterin formulation, as determined by the sub-
sequent absence of B. henselae bacteremia.
Session III: Treatment & Prevention
Poster [11
Title: Rifampin-resistant Bartonella henselae
(Bh)
279
Authors: M. Giladi, S. Abulafia, Y. Kletter, R.
Orni-Wasserlauf, Y. Golan, B. Avidor, I.
Shalit, M. Ephros
Institution: Tel Aviv Med. Ctr., Tel Aviv, and
Carmel Med. Ctr. Haifa, Israel
Bh is a principal etiologic agent of cat scratch
disease (CSD). Rifampin is frequently used for
treating this condition. Herein, we report on the first
Bh isolate which acquired resistance to rifampin.
Two tubes with lyophilized B. henselae ATCC
49793 were obtained from the University of Ok-
lahoma, U.S. (D.F. Welch), in 1993. One tube was
kept in -808 C (original strain), and the other (lab.
strain) was continuously passaged in the laboratory
on chocolate agar plates without antibiotics. In vitro
susceptibility of the lab. strain and 4 other B.
henselae isolates from patients with CSD was as-
sessed using E-test. MIC’s of 20 antimicrobials were
similar for all isolates tested, except for rifampin
which had MICs of 0.016-0.32 mg / ml for the 4 CSD
isolates and . 256 mg / ml for the lab. strain. MIC of
the original strain was 0.016 mg / ml. Resistance of
the lab. strain to rifampin was also confirmed by a
time-kill assay. 2 3 10 5 CFU / ml of Bh was incu-
bated in broth for 5 days with various concentrations
of rifampin (0.12 to 8.0 mg / ml). Even 8.0 mg / ml of
rifampin resulted in less than 1-log decrease of Bh
lab. strain CFU / ml, as compared with growth con-
trol, without rifampin, which increased to 5 3 10 6
CFU / ml. Rifampin was bactericidal (a 3-log-CFU /
ml decrease) at a concentration of 0.25 mg / ml when
the original strain was tested by the same assay,
indicating that resistance to rifampin had been
acquired in the laboratory. This report emphasizes
the importance of susceptibility testing of Bartonella.
It also suggests that using rifampin as monotherapy
might result in resistance to this antibiotic.
Session IV: Epidemiology & Reservoirs
Main Speaker
Title: Distribution of Bartonella in european
wild mammals
Authors: R.Heller 1 , D. Bermond 1 , G. Delacour 2 ,
H.J. Boulouis 3 , B.Chomel 4 and Y.
Piemont 1
280 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
Institutions: 1 Instit. Bacteriol., Univ. Louis Pasteur,
Strasbourg, France; 2 Office National de
la Chasse, Gerstheim, France; 3 Ecole
´ Maisons Alfort, France; 4 Sch. Vet.
vet.
Med., Univ. Calif., Davis, U.S.A
Objectives: To explore the diversity of Bartonella
reservoirs in European wildlife, because at least one
Bartonella species (B. henselae) is known as respon-
sible for human zoonoses. The animals tested were:
47 field-voles (Clethrionomys and Microtus sp.), 34
field-mice (Apodemus sp.), 2 water-voles (Rattus
norvegicus), 28 muskrats (Ondrata zibethicus), 23
coypus (Myocastor coypus), 2 shrew-mice (Sorex
araneus), 30 warren rabbits (Oryctolagus cuniculus),
8 hares (Lepus europaeus), 58 roe-deer (Capreolus
capreolus) and 4 fallow-deer (Dama dama).
Methods: Animal blood samples were collected by
intravenous or intracardiac puncture and injected into
lysis-centrifugation tubes. Cultures were performed
in Bactec  liquid medium and on Columbia agar
with 5% rabbit blood which were incubated for 2
months in 5% CO 2 at 358C. Characterization of the
bacterial isolates was performed by amplification of
the 16S rRNA and citrate synthase genes followed
by direct sequencing of the amplicons, by PCR /
RFLP and by total DNA hybridization for some
strains.
Results: Among 236 mammals blood samples, 135
(57%) contained cultivable Bartonella spp. About
80% of cultures were obtained within 4 to 11 days.
Bartonella isolates were obtained from field-voles
(33 / 47, 70%), field-mice (29 / 34, 85%), water-voles
(2 / 2), 28 muskrats (4 / 28, 14%), shrew-mice (1 / 2),
warren rabbits (9 / 30, 30%) and roe-deer (57 / 58,
98%). No Bartonella isolates could be obtained from
the 23 coypus, the 8 hares and the 4 fallow-deer. The
Bartonella isolates corresponded either to species
formerly described or to new species. We described
in particular B. tribocorum in water-voles and B.
alsatica in warren rabbits. These two new species
can be useful for the developpement of animal
models for natural infection. For the isolates ob-
tained from rodents we could evidence the presence
of 5 and 8 different clusters as determined by 16S
rRNA and citrate synthase gene sequence respective-
ly.
Conclusion: Most of the European wild mammals
tested harbor cultivable Bartonella spp. in their
blood. Some Bartonella species are preferentially
associated with defined mammals species. The actual
relevance of the Bartonella infection in human and
animal pathology remains still unknown. So the
species diversity within the Bartonella genus re-
quires further investigation, in order to update the
diagnostic tools (culture, serology and gene amplifi-
cation) for bartonelloses.
Session IV: Epidemiology & Reservoirs
Main Speaker
Title: Epidemiology of Bartonella infections
in domestic and wild carnivores and
ruminants
Authors: Bruno B. Chomel, R.W. Kasten, K.
Yamamoto, C. Chang, T.E. Honadel and
Y. Kikuchi
Institutions: Department of Population Health and
Reproduction, School of Veterinary
Medicine, University of California,
Davis. Davis, CA 95616, U.S.A.
Bartonella are emerging pathogens, several of them
being zoonotic agents. Domestic carnivores have
been identified in recent years as the main reservoir
of several Bartonella species, including the recently
identified B. koehlerae in California cats. Domestic
cats represent the main reservoir of at least two
Bartonella species involved in cat scratch disease,
primarily B. henselae and possibly B. clarridgeiae.
Cats can be bacteremic for very long period of time
and suffer relapses of bacteremia, especially when
infected with B. clarridgeiae. They can be co-infec-
ted with different Bartonella species or types. Pre-
valence of these Bartonella species or types varies
from one geographical area to another. For instance,
B. clarridgeiae is commonly isolated from European
cats or from cats in the Philippines (about 1 / 3 of all
isolates), whereas it is less frequent in North
America (approximately 10%) and has not been
isolated yet from Japan. Bartonella henselae type II
is highly dominant in Europe and in the Western
USA, whereas it is in equal proportion in the Eastern
USA, and has not been isolated in Filipino cats. Wild
felids are also infected with Bartonella. We reported
a 30% antibody prevalence in various captive felids
 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
from California and in, respectively, 35% and 53%
of free-ranging pumas (Felis concolor) and bobcats
( Lynx rufus). Testing of 274 pumas from North,
Central and South America indicates that Bartonella
infection is widespread in wild felids in the Western
Hemisphere. In Pumas from the southwestern USA,
antibody prevalence was 43% (30 / 69), whereas it
was 9% (7 / 76) in pumas from northwestern USA
and 0% in 14 animals from British Columbia,
Canada. Seroprevalence for Florida panthers was 6%
(2 / 36). In Central and South America, antibody
prevalence were respectively 32% (9 / 28) and 21%
(11 / 51). Bartonellae were isolated from 3 of 8
(37.5%) pumas and 7 of 19 (37%) bobcats from
California.
In canids, B. vinsonii subsp. berkhoffii was iso-
lated from a dog suffering endocarditis. Antibody
prevalence in dogs from North Carolina and Virginia
was 3.6%, but was much higher in dogs living in
rural areas and seropositive for Ehrlichia and
Babesia. A tick-borne transmission was suggested, as
seropositive dogs were 14 times more likely to have
a history of heavy tick exposure. In California, we
isolated B. vinsonii subsp. berkhoffii from 17 (32%)
of 54 coyotes, suggesting a possible wildlife reser-
voir. A serosurvey conducted on 869 coyote samples
collected between 1994 and 1998 indicated a 35%
prevalence, the highest prevalence being observed in
coyotes from the coastal regions, especially in cen-
tral and southern California. In a sero-survey of
1,873 US government owned dogs, 163 dogs (9%)
were found to be seropositive. The highest preval-
ence was observed in the south of the USA and in
the northern plains.
New Bartonella have been recently isolated from
roe-deer (Capreolus capreolus) in Europe and we
isolated similar strains from black-tailed deer
(Odocoileus hemionus) from California. More than
90% of these deer were bacteremic, whereas only
10% were seropositive. We also isolated Bartonella
from 15 of 100 elk (Cervus canadensis) from
California and Oregon, but none from 84 bighorn
sheep (Ovis canadensis). Furthermore, Bartonella
were isolated from 49% of 128 cattle from California
and Oklahoma. Bacteremia was observed in 89%
(47 / 53) of the beef cattle, but in only 17% (11 / 63)
of the dairy cattle. PCR / RFLP profiles of these
domestic and wild ruminant isolates indicate possi-
bility of inter-species transmission.
281
Session IV: Epidemiology & Reservoirs
Main Speaker
Title: Epidemiological investigation of Bar-
tonella bacteraemias in UK woodland
rodents
Authors: R.J. Birtles, S. Feore, K. Bown, M.
Begon, D. Raoult and M. Bennett
Institutions: Unite´ des Rickettsies, Universite´ Aix-
Marseille II, France, and CCID, Uni-
versity of Liverpool, UK
An understanding of the epidemiology of bartonellae
in rodents is important from several standpoints; in
medical and veterinary terms, the list of species
implicated as agents of disease now includes some of
those that naturally infect rodents. Furthermore, as
there is, as yet, no indication of virulence variation
among Bartonella strains and species, it is sensible
to consider all as having pathogenic potential. From
an ecological perspective, data derived from the
study of bartonella infections in rodents may allow
investigation of infectious disease dynamics within
naturally susceptible populations and the role of such
infections in the regulation of host abundance.
Horizontal surveys of rodent populations have begun
to provide an insight into the diversity of Bartonella
species encountered together with details of their
distribution and prevalence. In the UK, such surveys
have shown that populations of bank voles and field
mice concurrently sustain at least four distinct bar-
tonellae. However, at present nothing is known about
the epidemiologies of each species and the degree of
the ecological interaction between them. In this study
we used PCR-based detection and characterisation
methods to longitudinally monitor bartonella bac-
teraemias in woodland rodents. Twelve animals were
repeatedly trapped over a 12-month period and blood
samples were collected by tail clipping. The presence
and identity of Bartonella DNA in each of the 109
samples was determined using a PCR based which
amplified a fragment of the 16S / 23S rRNA inter-
genic spacer. Bartonella DNA was detected in 70
samples, from which 12 different nucleotide se-
quences (genotypes) were obtained. These genotypes
fell into five clusters, only two of which corres-
ponded to recognised Bartonella species, and citrate
282 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
synthase sequence analysis of unrecognised clusters
indicated their phylogenetic distinction. With the
exception of one bank vole, DNA was detected in
multiple samples from each animal and for each,
different genotypes were always detected. In some
animals, a genotype detected in samples collected
early in the study was re-encountered in later sam-
ples, irrespective of whether interim samples con-
tained other genotypes or no detectable bartonella
DNA. These results suggest there to be a complex
and dynamic epidemiology of Bartonella species in
woodland mammal communities and that most ani-
mals appear to be bacteraemic most of the time,
specific infections are often superseded.
Session IV: Epidemiology & Reservoirs
Oral present. / Poster [12
Title: Preliminary determination of the vector
of human bartonellosis in Caraz, Peru
through the use of ELISA, PCR, and
remote sensing technologies
Authors: Richard Andre,1 Scott Gordon,1 Penny
Masuoka,1 Caroline Korves,1 Roberto
Fernandez,2 Eliska Rejmankova,3 Donald
Roberts,1 Faustino Carbajal,2 Larry
Laughlin,1 Michael Zyzak,2 Hubel
Gonzalis 4 and Doug Watts 2
Institutions: 1 Uniformed Services University of the
Health Sciences, Bethesda, MD, U.S.A.;
2 2
Naval Medical Research Institute De-
tachment, Lima, Peru; 3 University of
California, Davis, CA, U.S.A.; 4 Ministry
of Health, Caraz, Peru
Infected sand flies of the genus Lutzomyia are
believed to transmit Bartonella bacilliformis to
humans living in the high mountain valleys of Peru;
however, the use of modern technologies such as the
ELISA, PCR, and remote sensing have not been used
to verify this. During the last two years, we made
collections of sand flies in the small villages sur-
rounding the town of Caraz, Peru. The ELISA was
used to examine over 500 bloodfed sand flies, which
were primarily captured indoors; 65% tested positive
for human bloodmeals. DNA was isolated from field
collected sand flies using a commercial extraction kit
and PCR primers for B. henselae citrate syntase gene
(gltA). Over 800 sand flies were placed in 100 pools
for Bartonella testing. Almost 30 pools were positive
for Bartonella organisms by PCR screening and half
of these were successfully sequenced. We have
isolated and identified B. bacilliformis in L. ver-
rucarum for the first time using PCR technology.
Two sequences had 99% agreement with B. quin-
tana, and one sequence was identical to B. henselae.
Using remote sensing images and aerial photos, we
have classified villages, rivers, major vegetation
zones, and agricultural areas. From a plot of patient
houses, it is apparent that the bartonellosis patients
primarily live in the agricultural areas, even though
most of the population lives in the town. We also
found that there are not greater numbers of patients
living near the major river in the study area. Higher
populations of sand flies were caught in agricultural
areas.
Session IV: Epidemiology & Reservoirs
Poster [14
Title: Bartonella in Portugal: An overview
Authors: Fatima Bacellar 1 , Eric Marston 2 , Raquel
Santos 3 , Barbara Ellis 2 and Armindo
Filipe 1
Institutions: 1 Centro de Estudos de Vectores e
Doenças Infecciosas, Instituto National
´
de Saude, ´
Aguas de Moura, Portugal;
Viral and Rickettsial Zoonoses Branch,
Centers for Disease Control & Preven-
tion, Atlanta, GA, USA; 3 Dermatologia,
Hospital de Curry Cabral, Lisbon, Por-
tugal
Bartonellae have been recognized in Portugal since
the beginning of the century and clinical cases of cat
scratch disease (CSD) have been described since the
1950s. Since identification of the agent responsible
for CSD in 1992, we have employed the indirect
immunofluorescence test (IFAT) for routine diag-
nosis of patients and for general population serosur-
veys in humans and non-human animals. Results of
Bartonella-related research done over the last six
years are presented here.
Antigens used in IFATs were obtained from
 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
Bartonella henselae and B. quintana grown in Vero
E6 cell culture. In total, sera from 219 patients with
signs and symptoms compatible with CSD and
bacillary angiomatosis (BA) were tested against both
antigens. Cut-off values for the IFA test were
determined using 100 blood donor samples as con-
trols. Other studies included genetic characterization
of Bartonella isolated from whole blood from
humans, urban cats and rats (Rattus rattus and R.
norvegicus), and Bartonella detected in angioma
biopsies. DNA extracts from isolates or biopsy
material were used in a PCR amplifying 338 bp of
the gltA gene. PCR products were sequenced using
an automated sequencer.
Overall, 36% (n 5 78) of the patients with a
clinical diagnosis of CSD and BA had antibodies
reactive to B. henselae and B. quintana. Bartonella
quintana was confirmed by nucleotide sequence
analysis from an angioma of an HIV-positive in-
dividual. Serosurveys of rats and cats indicated that
19.5% and 17.6%, respectively, had positive serolo-
gy. Four Bartonella isolates from R. rattus and R.
norvegicus were obtained, including two novel geno-
types.
Session IV: Epidemiology & Reservoirs
Poster [15
Title: Cat scratch disease in Belgium: A 4
years survey
Authors: G. Bigaignon 1 , C. Gusbin 1 , A. Van
Lint 1 , M. Delmee
´ 1 , L. Boon-Falleur 1 , T.
M’Bilo , A. Aeby 2 , P. Lepage 2 , C.
1
Potvliege 2 , M.-L. Delforge 3 , C. Rossi 3
and B. Sztern 4
Institutions: 1 Cliniques Universitaires St Luc, 1200
Bruxelles, Belgium; 2 Centre Hospitalier
de Tivoli, 7100 La Louviere, ` Belgium;
3
ˆ
Hopital Universitaire Erasme, 1070
Bruxelles, Belgium; 4 Centre Hospitalier
`
Moliere Longchamp, 1190 Bruxelles,
Belgium
The main causal agent of Cat Scratch Disease is a
bacterium named Bartonella henselae : special slides
have been adapted for serodiagnosis by indirect
immunofluorescence assay (MRL Diagnostics,
283
Cypress, California), following the recommadations
of the CDC, with the dilution at 1 / 64 to detect IgG
and IgM.
Over a 4 years period, between 23 / 12 / 1993 and
31 / 12 / 1997, the Laboratory of Serology of the
University Hospital St Luc received 2221 serum
samples from patients with clinical signs suggestive
of CSD, mainly in pediatrics and predominantly in
male patients (sex ratio M / F : 1.086).
The presence of antibodies IgG and / or IgM anti-
Bartonellahenselae was mainly observed among
boys (in the first 4 decades: 14%, 18%, 16% and
20% of these ones and 12%, 12%, 10% and 13%
among the girls).
Among many classical manifestations of CSD, this
serology was an help to diagnose several complica-
tions, as two cases of endocarditis, one encephalitis
and one oculoglandular syndrome of Parinaud.
The percentages of positivity were higher in the
Western and Northern parts of Belgium (22% and
16% in male and female patients respectively), lower
in the province of Luxembourg (12% and 10%) and
intermediate in the Brabant (15% and 10%).
These 276 positive results in 4 years, or 69 case
reports per year, give an annual rate of infection with
Bartonella henselae in Belgium of 0.69 per 100,000
residents, very near the rate of 0.77 in the population
of the United States.
Session IV: Epidemiology & Reservoirs
Poster [16
Title: Epidemiology of Bartonellosis in Peru:
Animal reservoir studies
Authors: W.P. Carney 1 , S.W. Gordon 1 , G.S.
Hohenhaus 1 , A. Gozalo 2 , D. Watts 2 , A.
Gonzales 3 , R.L. Regnery 4 , J.C. Olson 4 ,
J. Rooney 4 , and E.L. Marston 4
Institution: 1 Department of Preventive Medicine and
Biometrics, Uniformed Services, Uni-
versity of the Health Sciences, Bethesda,
MD, U.S.A.; 2 Naval Medical Research
Institute Detachment, Lima, Peru;
3
Ministry of Health, Caraz, Chavin Re-
gion, Peru; 4 Division of Viral and Ric-
kettsial Diseases, Centers for Disease
Control, Atlanta, GA, U.S.A.
284 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
In the Andes mountains of South American Bar-
tonella bacilliformis causes distinct acute and
chronic disease syndromes, respectively termed
oroya fever and verruga peruana. As part of a larger
effort to understand the epidemiology of this disease,
we have been investigating the possible role of wild
and / or domestic animals as reservoir hosts of B.
bacilliformis in the Chavin region of Peru. Over
1000 animals have been sampled and 750 have been
examined for evidence of infection using blood
samples obtained by venipuncture for bacterial cul-
ture and / or PCR testing. Bartonella infections were
identified in Mus musculus, Phyllotis andium,
Oryzomys xantheolus, Akodon mollis, guinea pigs,
domestic dogs and a toad. Preliminary sequencing
results from a sub-sample of PCR and / or culture
positive animals have identified several different
strains of Bartonella. A BLAST search of GenBank
showed greatest similarity to Bartonella isolates
from rodents in southeastern United States and Peru.
Phylogenic analyzes of partial citrate synthase gene
sequences indicate that multiple strains of Bartonella
are circulating in the Chavin region of Peru. Al-
though not all of the samples have been analyzed, no
human Bartonella strains have yet been detected
from wild or domestic animals in the Chavin region
of Peru.
Session IV: Epidemiology & Reservoirs
Poster [17
Title: Nanobacteria, extraordinary relative of
Bartonella
Author: E. Olavi Kajander, Kristin Ehrbar, Mia
¨ ¨
Monkkonen, ¨
Mikael Bjorklund and
Neva Çiftçioglu
Institution: Department of Biochemistry and Bio-
technology, University of Kuopio,
Kuopio, Finland
Nanobacteria are cytotoxic, intra- and extra-cellular,
fastidious gram-negative minute bacteria. Their
ecological niche is blood, and they cause calcifica-
tion and stone formation in long-lasting chronic
infection in the mammalian host. Standard mi-
crobiological diagnostic methods cannot detect them
because of their extraordinary properties. 16S rRNA
genomic sequence of nanobacteria place them into
the alpha-2 subgroup of Proteobacteria, near Bar-
tonella. Nanobacteria and Bartonella do share sever-
al extraordinary properties: they survive in blood for
extended periods causing bacteremia with relatively
mild symptoms, possibly due to their weakly cyto-
toxic endotoxin that has been shown to be similar to
that of Chlamydia by our collaborative work with the
Hjelles’ group. Both interact with erythrocytes and
have specific invasion mechanisms to enter other
host cells. We have produced monoclonal and poly-
clonal antibodies against these bacteria which recog-
nized similar epitopes in Nanobacteria and several
Bartonella species. These bacteria are stained poorly
with standard microbial stains, but stain well with
silver methods. They can be diagnosed by using cell
culture methods. Presence in blood, clinically special
relationships, serological cross-reactivity and highly
homologous 16S rRNA between these relatives bring
additional difficulties to the diagnosis of Bartonel-
losis. Nanobacteria and Bartonella are opening a
novel aspect to the understanding of host-microbe
relationship in the form of bacteremia. They illus-
trate that entire animal species can be chronically
infected: cows with Nanobacteria and cats with B.
henselae. Inadvertent human infection can cause
several diseases which were not previously known to
be bacterially associated.
Session IV: Epidemiology & Reservoirs
Poster [18
Title: Prevalence of Bartonella species among
pet cats in Japan
Authors: Soichi Maruyama, Shigeo Tanaka,
Takeo Sakai, and Yasuji Katsube
Institution: College of Bioresource Sciences, Nihon
University, 1866 Kameino, Fujisawa,
Kanagawa 252-8510, Japan
The causative agent of cat scratch disease (CSD) is
recognized to be Gram-negative bacteria, Bartonella
henselae. Recently, it is reported that CSD was also
attributed to new Bartonella species, B. clarridgeiae.
The authors investigated the prevalence of Bartonel-
la species among pet cats as a basic epidemiological
study of CSD in Japan.
 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
A total of 450 blood samples were collected from
pet cats. Of them, 200 samples were derived from
central area (Kanagawa Prefecture), 150 from west-
ern area (Kyoto, Osaka and Hyogo Prefectures), and
100 from southwestern area (Kagoshima and
Okinawa Prefectures). The blood samples were taken
to EDTA tubes and sent to our laboratory under
freezing condition. The isolation of bartonella was
followed by the method reported previously. The
Bartonella species was identified by PCR-restriction
fragment polymorphism in the citrate synthase gene
with TaqI and HhaI. Species of B. clarridgeiae was
confirmed by pulsed-field gel electrophoresis with
SmaI and AscI digestions.
Bartonella species were isolated from 43 (9.5%)
of 450 cats of Japan. The isolation rate varied from
2% in Hyogo to 20% in Okinawa. The rate in central
area (5.0%) was significantly lower than that in
western area (11.3%) and that in southwestern area
(16.0%) (p , 0.01). B. clarridgeiae was isolated
from 2 of 50 cats in Kyoto, 3 of 50 in Osaka and 1 of
50 in Okinawa. None of B. clarridgeiae was detected
from Kanagawa, Hyogo and Kagoshima. Though the
cats of Kanagawa, Kagoshima, Kyoto and Okinawa
were infected with only B. henselae or B. clar-
ridgeiae, one of 8 positive-cats in Osaka harbored
both B. henselae and B. clarridgeiae in the blood.
Keynote Lecture
Title: Rupture and invasion of the intestinal
epithelial barrier by Shigella: Molecular
and cellular approaches
Author: Philippe J. Sansonetti
Institution: Unite´ de Pathogenie
´ Microbienne Mol-
éculaire, INSERM U 389, Institut Pas-
teur, 28, rue du Dr. Roux, F-75724 Paris
´
Cedex 15, France
The pathogenesis of bacillary dysentery can be
studied at different levels of integration of the
cellular components that constitute the colonic
mucosal barrier. We have considered the interaction
of Shigella flexneri in three experimental systems
that bring complementary information and provide a
scheme of the events occurring in the human colo-
rectal mucosa as Shigella invasion proceeds. Inter-
285
action of S.flexneri with individual epithelial cells
show a series of events in which the bacterium, upon
contact with the cell surface, releases a complex of
Ipa proteins (i.e. invasins) through a specialized,
activable, type III secretory apparatus (i.e. Mxi / Spa).
Via a complex signaling process that involves both
the cascade of signals elicited by the three GTPases
of the Rho family (Cdc42, Rac and Rho) and
pp60 c-src , these invasins cause major rearrangment of
the subcortical cytoskeletal network thereby allowing
bacterial entry by a macropinocytic event. Invasion
by the bacterium turns the epithelial cell to a strongly
pro-inflammatory entity, due to activation of NFKB.
Then the bacterium lyses its phagocytic vacuole and
initiates intracytoplasmic movement, due to polar
assembly of actin filaments caused by a bacterial
surface protein, IcsA. The cytoskeletal-associated
protein N-WASP appears to play an essential role in
initiation of actin polymerization. This allows very
efficient colonization of the host cell cytoplasm and
passage of adjacent cells via protrusions which are
engulfed by a cadherin-dependent process. However,
when invasive Shigella are deposited on the apical
side of polarized monolayers of human colonic cells,
they are unable to invade, indicating that bacteria
need to reach the subepithelial area to invade the
epithelium. In this system, it has been shown that
transepithelial signaling caused by apical bacteria
induces adherence and transmigration of basal poly-
morphonuclear leucocytes (PMN), thus disrupting
the monolayer’s permeability and facilitating bacteri-
al invasion. LPS accounts for a large part of this
transepithelial signalization to PMN. Such process
could account for invasion in intestinal crypts.
Finally, models of infection such as the rabbit ligated
intestinal loop show that initial bacterial entry occurs
essentially via M cells of the follicular associates
epithelium. It then causes apoptosis of macrophages
located in the follicular dome, inducing release of
IL-1ß, which in turn initiates inflammation, leading
to destabilization of the epithelial stucture as
modeled above.
Session V: Pathogenesis
Main Speaker
Title: Bartonella – endothelial interactions
286 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
Authors: Marlene Meyer 1 , Markus Roger
¨ 1 , Anja
Seubert , T. Korff , Christa Lanz 1 and
1 2
Christoph Dehio 1
Institutions: 1 Dept. of Infect. Biol., Max Planck
¨
Institute for Biology, Tubingen, Ger-
many 2 University of Gottingen
¨ Medical
¨
School, Gottingen, Germany
Bartonella infections of immunocompromised pa-
tients often involve specific interactions of the
bacteria with the endothelium. Colonised endothelial
cells become activated resulting in vascular
neoformation (i.e. in bacillary angiomatosis). Using
cultured human umbilical vein endothelial cells
(HUVECs) infected with B. henselae, selected as-
pects of the colonisation as well as the bacterium-
stimulated activation and proliferation of endothelial
cells could be investigated. Using this in vitro
infection model, we were able to demonstrate a new
mechanism of cellular invasion by bacteria. This
process involves a unique sequence of pathogen-host
cell interactions. First, the leading lamella of endo-
thelial cells establishes cellular contact with bacteria.
Bacterial aggregation on the cell surface is then
mediated by rearward transport and capping. The
bacterial aggregate is engulfed by host cell mem-
brane protrusions and finally internalised resulting in
the singular and well-defined invasome structure.
Invasome-mediated invasion is an actin-dependent
but microtubuli-independent process. Gentamicin-
protection assays demonstrate the viability of B.
henselae inside the invasome.
Live B. henselae as well as crude bacterial extracts
have previously been reported to trigger the prolifer-
ation and migration of HUVECs in vitro. We have
further fractionated B. henselae and localised the
mitogenic activity to the outer membrane prepared
by sucrose gradient centrifugation. The mitogenic
activity is concentration dependent and heat- and
trypsin-sensitive, indicating a proteinaceous nature of
the responsible factor. Total and outer membrane
fractions were also found to activate endothelial cells
arranged in spheroids to form sprouts, demonstrating
that these membrane preparations can stimulate
angiogenesis in vitro. The in vivo angiogenic activity
of the outer membrane of B. henselae remains to be
elucidated. Purification of the angiogenic activity
from the B. henselae outer membrane is under
investigation.
Session V: Pathogenesis
Main Speaker
Title: Mechanisms of entry of B. Bacilliformis
into red cells and endothelial cells
Authors: Anita Verma 1 ; Steve Derrick 1 ; George
Davis 2 ; Garret Ihler 1
Institutions: 1 Dept. of Medical Biochemistry an Ge-
netics, Texas A&M College of Medi-
cine, College Station, TX 77843, U.S.A.
2
Dept. of Medical Pathology and Lab-
oratory Medicine, Texas A&M, College
of Medicine, College Station, TX 77843,
U.S.A.
Bartonella bacilliformis enters both erythrocytes and
endothelial cells by an endocytotic process. A pro-
tein, deformin, which is found mostly in the superna-
tant, but also in bacterial membrane fractions, is
capable of reproducing the morphological effects of
the bacterium on the red cell membrane. This
includes deformation of the membrane (trenches,
conical depressions, invagination), and formation of
endocytic vacuoles in the red cells. Curiously, this
protein has no observable morphological effects on
endothelial cells, as far as we have been able to learn
thus far, and in particular it does not induce vacuola-
tion. We have extensively restudied the purification
of this protein with results similar to those reported
previously (Biochim. Biophys. Acta 12234 (1995)
173-183). Some useful new aspects of the purifica-
tion will be reported. Unfortunately, we still have not
been able to obtain an adequately large sample for
sequence determination.
B. bacilliformis and B. henselae are similar in
many aspects. However infection of human red cells
is not an aspect of the human disease caused by B.
henselae, although the bacterium was found in cat
red cells. We hoped to use this observation to
identify the human red cell receptor for the bac-
terium. We were not successful in this, but we did
find that both bacteria bind several proteins from red
cell membrane preparations, including actin and
spectrin.
Actin and spectrin are not receptors of course,
being on the inside, and it is not clear how to
interpret this result for red cells. However, infection
 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
of endothelial cells with B. bacilliformis produces a
major change in the actin organization in the cell and
in their migratory behavior in an in vitro scratch
]]
assay. Also the distribution of a cell surface marker,
PECAM, is markedly altered, with disruption of
cell-cell adhesions, resulting in loss of junctional
staining.
Session V: Pathogenesis
Main Speaker
Title: Genetic manipulation of Bartonella vir-
ulence determinants
Author: Michael F. Minnick
Institution: The University of Montana, U.S.A.
Due to the lack of a system for site-specific genetic
manipulation, few reports have been published con-
cerning the molecular mechanisms involved in
pathogenesis, growth, and antibiotic resistance of
Bartonella species. We approached this problem by
molecularly characterizing the DNA gyrase of Bar-
tonella bacilliformis. DNA gyrase is a type II
topoisomerase that introduces negative supercoiling
into DNA and is the target of several types of
antimicrobial agents. Point mutations in the gyrase B
subunit gene ( gyrB) can confer resistance to
coumarin antibiotics, thus providing a locus and
selectable phenotype for investigating allelic ex-
change. We isolated and characterized twelve
coumermycin A1-resistant strains of B. bacilliformis
and characterized their respective gyrB genes. Single
nucleotide transitions at three separate loci in gyrB
conferred resistance to the drug and resulted in
amino acid substitutions of Gly124 . Ser, Arg184
. Gln, Thr214 . Ile or Thr214 . Ala. These data
were used to design single-stranded and double-
stranded oligonucleotides which contained the lesion,
in hopes of generating allelic exchange mutants.
Repeated attempts at using linearized DNA strategies
to mutagenize gyrB failed, leading us to believe that
a formidable restriction modification system is pres-
ent in the bacterium. To alleviate this problem, we
isolated a highly competent, DCM methylase mutant
of B.bacilliformis (strain JB584) that demonstrates a
2000% increase in transformation efficiencies rela-
tive to wild-type strains. In addition, we developed a
287
suicide vector, termed pUB1, containing a pMB1
replicon and kanamycin resistance cassette (nptI) to
attempt allelic exchange via a single crossover event.
To date, two virulence genes have been successfully
mutagenized using the JB584 strain and pUB1
constructs, including the flagellin ( flaA) and the
invasion-associated locus B (ialB) genes. Com-
plementation of these loci in trans was also achieved
using the broad host-range vector pBBR1-MCS
containing cloned, wild-type genes. Preliminary data
with flagellin mutants and transcomplemented strains
suggests that flagellin is an important virulence
determinant. The combined system of the JB584
strain, pUB1 suicide vector and PBBR1-MCS shuttle
vector is a significant breakthrough, and will enhance
our ability to study the molecular biology of Bar-
tonella.
Session V: Pathogenesis
Oral Presentation / Poster [19
Title: Isolation, sequencing and expression of
the Bartonella henselae gene encoding
the 43-KDA outer membrane endothelial
cell binding protein
Authors: Andrew W.O. Burgess and Burt E. An-
derson.
Institute: University of South Florida College of
Medicine, Dept. of Medical Microbiol-
ogy and Immunology, Tampa, FL
33612, U.S.A.
Members of the genus Bartonella are unique in that
they are bacteria that cause proliferation of mi-
crovascular endothelial cells and neovascularization
(angiogenesis). The mechanisms by which Bartonel-
la henselae causes these processes are unknown.
Given the importance of surface-exposed determi-
nants in the pathogenesis of many organisms, outer
membrane proteins (OMPs) of B. henselae were
identified. Enrichment of the outer membrane frac-
tion of B. henselae by sarkosyl treatment of total
membranes, together with radioiodination and
biotinylation of intact organisms, suggest that at least
nine proteins, with molecular weights of 28, 30, 35,
43, 58, 61, 79, 92 and 171 kDa, are located in the
outer membrane. Binding assays using (1) Triton
288 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
X-100-extracted biotinylated human umbilical vein
endothelial cell (HUVEC) surface proteins incubated
with immobilized B. henselae OMPs, and (2)
biotinylated B. henselae surface proteins incubated
with intact HUVECs, show that the 43-kDa protein
(Omp43) is the major adhesin for endothelial cells.
The gene encoding Omp43 was cloned and se-
quenced. Sequence analysis revealed an open reading
frame of 1206 nucleotides coding for a protein of
402 amino acids. Analysis of the deduced amino acid
sequence shows 38% identity over the entire se-
quence to a Brucella spp. porin, as well as shows a
potential signal sequence and peptidase cleavage site,
further supporting the outer membrane location of
Omp43. Cleavage of the signal peptide would result
in a mature 380-aa polypeptide with a predicted MW
of 42 kDa. Expression of Omp43, without the signal
sequence, in Escherichia coli revealed a histidine-
tagged fusion protein of 45 kDa on Western blots
using antibody to the N-terminal Xpress tag. Binding
assays revealed that purified recombinant Omp43, at
concentrations of 11 and 2.75 mg / ml, bound to intact
HUVECs, while negative control LacZ fusion pro-
tein did not bind. The identification of B. henselae
OMPs, as well as the major adhesin for endothelial
cells, should provide a basis for further investigation
of the role of adherence in the pathogenesis of B.
henselae.
Session V: Pathogenesis
Oral Presentation / Poster [20
Title: In vitro effects of LPS from Bartonella
quintana on inflammator mediators
Authors: A. Foca,` G. Matera, M.C. Liberto, R.
Diana, A. Pollio, M. Martucci, E. Gullet-
ta
Institutions: Chair of Microbiology, Chair of Clinical
Chemistry and Chair of Clinical Pathol-
ogy, University ’Magna Graecia’, Catan-
zaro, Italy
Bartonella quintana has been reported as the
cause of endocarditis and persistent bacteriaemia.
One of the main pathogenic factor of gram-negative
bacteria, including Bartonella quintana, is repre-
sented by the lipopolysaccharide (LPS). However,
up to our knowledges, very little information is
available on the pathophysiological role of Bartonel-
la LPS.
The present study was undertaken to explain the
relationship between Bartonella quintana LPS and
inflammatory response from the host, by means of
the ’in vitro’ human whole blood model of sepsis.
Here we evaluated: i) TNF as a general and early
inflammatory mediator; ii) bactericidal / permeability-
increasing protein (BPI) as a specific tool for addres-
sing neutrophil (PMN) activation; iii) IL-8 as an
important chemokine and a late inflammatory medi-
ator playing an important role in the interaction
between monocytes and PMN. Our data suggest that
Bartonella LPS might not be able to stimulate a very
potent and early inflammatory reaction at the same
extent of the LPS from other gram-negative bacteria.
This is particularly true looking at the earliest
mediator of most bacterial sepsis (e.g. TNF). How-
ever monocytes may still be able to stimulate PMN,
which should be activated, as indicated by the IL-8
and BPI levels. The IL-8 increase was the most
prominent biological effect of Bartonella LPS in our
model and this feature may fit with some pathogenic
aspects of Bartonella quintana infections.
Session V: Pathogenesis
Poster [21
Title: Cellular immune response to Bartonella
henselae in the murine infection model
Authors: M. Arvand, Th. Regnath, M. E. A.
Mielke, H. Hahn
Institution: Institute for Medical Microbiology and
¨
Immunology, Universitatsklinikum Ben-
jamin Franklin, Free University of Ber-
lin, Germany
Clinical presentation of infection with Bartonella
henselae ranges from a relatively mild lymphadenop-
athy with few additional symptoms, seen in cat
scratch disease, to life-threatening systemic disease
in immunocompromised individuals, e.g. patients
with AIDS. The more severe clinical manifestation
 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
in the immunocompromised host points towards a
role of T cells in the pathogenesis of B. henselae
induced diseases.
We studied the cellular immune response to B.
henselae in the murine model of B. henselae in-
fection. C57BL / 6-mice were infected with B. hen-
selae intraperitoneally, mononuclear cells of the
spleen were isolated and restimulated with B. hen-
selae in vitro. The cellular immune response was
investigated by measuring i) the lymphocyte prolifer-
ation rate, and ii) the g-interferon secretion.
We found a specific cellular immune response in
mice infected with B. henselae.
Session V: Pathogenesis
Poster [22
Title: A mouse model of Bartonella infection:
influence of mouse breed, age and
gravidity on bacteremia levels
Authors: F. Barrat 1 , D. Bermond 3 , C. Bouillin 2 ,
C. Gandoin 2 , D. Thibault 1 , B. Chomel 4 ,
R. Heller 3 , Y. Piemont 3 and H.-J.
Boulouis 1,2
Institutions: 1 INRA / Microbiologie, Ecole Veter- ´´
inaire, Maisons-Alfort, France;
2
´´
Microbiologie / IIAC, Ecole Veterinaire,
Maisons-Alfort, France; 3 Inst. Bac-
teriologie, Univ. L. Pasteur, Strasbourg,
France; 4 Depart. Popul. Health & Re-
prod., School of Vet. Med., UC Davis,
Davis, CA, USA.
Bartonella infection is usually characterized in its
animal reservoir by a persistent bacteremia without
specific clinical signs of disease. In some species,
such as cats and dogs, prolonged bacteremia occurs
despite the presence of circulating antibodies, where-
as a very limited antibody response is detected
during bacteremia in some other species, such as
rodents or ruminants. We have developped a model
of Bartonella infection in laboratory mice to study
the pathogenesis and the immune response to Bar-
tonella infection in animals.
Laboratory mice were inoculated intraveinously
with a Bartonella strain isolated from a field mouse
289
(Apodemus spp.). Blood samples (100 ml) were
collected in EDTA tubes by cardiac puncture just
prior to inoculation and 7, 14, 21, 28, 49, 60 and 90
days after inoculation (PI). After osmotic shock with
distilled water and centrifugation, blood was imme-
diately plated on 5% defibrinated rabbit blood BHI
agar in a 5% CO 2 atmosphere. CFU were counted
fifteen days after plating. Each assay was performed
on groups of 5 to 8 animals.
In 10–12 week-old Balb / C mice, bacteremia
appeared on day 7, reached a maximum of 100,000
CFU / ml between day 14 and 21 and desappeared
between day 49 and day 60 PI. No relapse of
bacteremia was observed during the 4 weeks follow-
ing the last positive culture. Level-but not length-of
bacteremia varied with the doses of CFU and the
number of subculture of Bartonella strains. The
mean level of bacteremia was significatively higher
when Bartonella was inoculated to Balb / C mice
than to C57Bl / 6 mice (p , 0,001) or Swiss mice
(p , 0,0001).
The comparison of bacteremia between young
Balb / C mice (10-week old) and old Balb / Cmice
(18-month old) showed a significatively elevated
bacteremia (two fold; p , 0,04) in the aged mice.
The comparison of bacteremia between virgin mice
and gravid mice showed that gestation amplified the
number of CFU / ml by two folds if the infection
occurred before mating (p , 0,001) and five folds if
the infection happened after mating (p , 0,007). All
four embryos from one of eight Balb / C mice
inoculated 10 days before mating and harvested by
aseptic ceasarean operation (at 18 days of gestation)
yielded 10-15 CFU after culture of their spleen and
liver.
In conclusion, we have established a murine
model of Bartonella infection. It allowed us to
demonstrate the influence of the reproductive status
on the course of Bartonella infection and to confirm
experimentally the transmission of Bartonella in
utero in the murine species.
Session V: Pathogenesis
Poster [23
Title: Bartonella henselae induces upregula-
tion of adhesion molecules in human
290 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
endothelial cells-involvement of nuclear
factor-kB
Authors: O. Fuhrmann 1 , N. Schwarzer 1 , H.
Geisel 1 , M. Arvand 2 , C. Dehio 3 and N.
Suttorp 1
Institutions: 1 Dept. of Internal Medicine, Justus-
Liebig-University, Giessen, Germany
2
Dept. for Med. Microbiol. and Infec-
tionsimmunol., FU Berlin, Germany
3
Dept. of Infection Biology, Max Planck
¨
Institute for Biology, Tubingen, Ger-
many
Introduction: Infection of endothelial cells by Bar-
tonella henselae induces cell migration and prolifer-
ation and is an important step in the pathogenesis of
bacillary angiomatosis (BA). Adhesion of leukocytes
to endothelial cells is an essential step in an in-
flammatory reaction. We studied infection and activa-
tion of cultured human umbilical vein endothelial
cells (HUVEC) and focussed on signal transduction
pathways, transcription factors and recruitment of
immune cells.
Methods: Use of three cultivated wild-type Bar-
tonella-strains (ATCC 49882, ATCC 49793, and a
spontaneous rif-mutant) and of B. henselae isolated
from bacillary angiomatosis lesions in a german
HIV-patient. Expression of adhesion molecules (cell
surface ELISA, Northern blot), EMSA, NF-kB-re-
porter gene assay; leukocyte-‘‘rolling’’ (under de-
fined shear stress), -adhesion, and transmigration on /
through HUVEC monolayer.
Results: Expression of E-selectin, intracellular
adhesion molecule-1 (ICAM-1), and vascular cell
adhesion molecule (VCAM-1) was potently upregu-
lated within 5-8 h in Bartonella infected HUVEC as
indicated by Northern hybridization and cell surface
ELISAS. The freshly isolated Bartonella strain
turned out to be substantially more effective than
cultivated strains. B. henselae induced activation of
multiple host cell signal-transduction pathways.
Bandshift assays with a NF-kB consensus oligo-
nucleotide showed a time-dependent activation of
NF-kB. Transfection assay experiments involving a
pGLNf-kB luc reporter gene plasmid also induced
significant NF-kB-activation in Bartonella henselae-
infected HUVEC.
Conclusion: Infection of HUVEC with Bartonella
henselae (freshly isolated strain, cultivated strains)
resulted in endothelial cell activation (upregulation
of adhesion molecules, NF-kB-translocation) and
overall induced a proinflammatory endothelial cell
phenotype.
Session V: Pathogenesis
Poster [24
Title: Hemin is essential for growth and a
main iron source of Bartonella henselae
Authors: Stefan Kretzer, Stefan Bereswill and
Anna Sander
Institute: Institute of Medical Microbiology and
Hygiene, University of Freiburg,
Freiburg, Germany
Bartonella henselae, the etiologic agent of cat
scratch disease (CSD), is usually grown on blood
containing media. No growth occurs in Brucella
broth (or Brucella agar) alone, but the organisms
remain viable up to 3 days. B. henselae (strain
G5436) was tested for its ability to utilize different
sources of iron and porphyrin. Combinations of 7
different supplements were evaluated for their
capacity to restore growth of B. henselae in Brucella
broth and / or Brucella agar. Of the tested substances
ferrous chloride, ferric chloride, protoporphyrin IX,
hemin, hemoglobin, transferrin and lactoferrin only
hemin and hemoglobin at a concentration of more
than 0.03 mM and 0.01 mM respectively, restored
growth, whereas protoporphyrin IX even in combina-
tion with ferric chloride (each at up to 1 mM) had no
effect. Growth with hemin as a source of iron was
not notably diminished in presence of concentrations
of EDDHA, an iron chelator, suggesting that the
ionic iron present in unsupplemented Brucella agar is
not utilized by B. henselae in addition to the hemin
offered. Photometric quantification of growth in
Brucella broth and detection of colony forming units
(cfu / ml) on Brucella agar plates supplemented with
hemin demonstrated lack of growth at concentration
of less than 0.003 mM and 0.03 mM respectively.
Best growth occured at hemin concentrations of 0.1
 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
to 0.2 mM, but growth did not markedly increase
with higher concentrations.
In conclusion, hemin seems to be essential for
growth of B. henselae. The fact that ferrous chloride,
ferric chloride, protoporphyrin IX, lactoferrin and
transferrin failed to support growth provides evi-
dence that hemin is used as the main iron source.
The finding that protoporphyrin IX and ferric chlo-
ride, two main components of hemin, could not
restore growth of B. henselae, might indicate that B.
henselae is not able to synthesize hemin.
Session V: Pathogenesis
Poster [25
Title: Murine model of subcutaneous Bartonel-
la henselae infection
Authors: T. Regnath, M. Arvand, H. Hahn and
M.E.A. Mielke
Institution: Institute of Infectious Diseases, Depart-
ment of Medical Microbiology, Uni-
versity Hospital Benjamin Franklin, Free
University of Berlin, Germany
Progress in understanding the pathogenesis of and
the immune response to Bartonella henselae in-
fection has been limited by the lack of a suitable
animal model. Therefore, we have developed a
murine model of subcutaneous infection with Bar-
tonella henselae.
Female C57BL / 6 mice at an age of 10 to 12
weeks were infected by s.c. injection of 5 3 10 5
viable B. henselae (ATCC 49882) into the hind
footpad. At different times from 4 days to 20 weeks
postinfection, popliteal lymph nodes were harvested,
weighed and examined for histopathological altera-
tions. In parallel serum was collected for detection of
B. henselae-specific antibodies by means of an in-
house ELISA.
The course of infection was characterised by a
pronounced swelling of the draining lymph nodes
reaching its highest degree of intensity 6 weeks
postinfection. Four weeks later lymph node swelling
began to decrease spontaneously, and at the end of
the 16th week p.i. lymph nodes showed normal size.
Lymph node tissue sections showed numerous ac-
291
cumulations of CD11b 1 monocytes. B. henselae-
specific IgG antibodies were detected from 2 weeks
postinfection throughout the experiment.
This model is applicable to the study of patho-
genesis of B. henselae infection in the immuno-
competent host and to the development and evalua-
tion of diagnostic procedures.
Session VI: Genetics & Genomics
Main Speaker
Title: Green fluorescent protein as a tool to
study the pathogenesis of Bartonella
spp. in vitro and in vivo
1
Authors: ¨
Ralf Schulein , Christian Gille 1 , Anja
Seubert , Yves Hansmann 2 , Remy
1
Heller 2 , Yves Piemont 2
´ , Siv Andersson
3
, Christa Lanz , Christoph Dehio 1
1
Institutions: 1 Dept. of Infection Biology, Max Planck
¨
Institute for Biology, Tubingen,Ger-
many; 2 Instit. Bacteriol., Univ. Louis
Pasteur, Strasbourg, France; 3 Mol. Evo-
lution and Evolutionary Biology, Upp-
sala University, Sweden
We are using the gene encoding green fluorescent
protein (GFP) as a visual marker of gene expression
for (i) in vivo fluorescent tagging of Bartonella spp.
to trace the hemotropic infection in an appropriate
animal model by means of flow cytometry and
fluorescence microscopy and (ii) cloning the genes
of Bartonella spp. which are differentially expressed
during the infection of endothelial cells in vitro by
means of flow cytometry and bacterial cell sorting
(differential fluorescence induction technology, DFI).
We have established in vivo fluorescent tagging
with GFP for the rat pathogen B. tribocorum. A
conjugative suicide vector bearing a composite trans-
poson has been constructed and used to deliver a
GFP-expression cassette into the chromosome of B.
tribocorum. A resulting recombinant strain express-
ing GFP stably and at high levels was used to study
the colonisation of rats following intravenous in-
fection. Bacteria are cleared from blood within hours
but reappear in circulating erythrocytes on day 4
post-infection. The bacteria then grow exponentially
292 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
inside erythrocytes over a few days until they reach a
critical cell density (10-15 bacteria per erythrocyte).
Apparently, bacteria colonise the erythrocytes with-
out further replication over a prolonged period in a
perfectly culturable state (up to 10 7 bacteria per ml
blood). Bacteremia ceases at 8–10 weeks post-in-
fection. The constant number of infected erythro-
cytes throughout the entire bacteremic phase (1
infected erythrocyte in 1000–5000 erythrocytes)
suggests that erythrocyte invasion in this animal
infection model is an early and highly synchronised
process.
We used the techniques of DFI to identify genes of
B. henselae which are differentially expressed during
specific interaction with endothelial cells. For this
purpose, a promoter-probe library of the B. henselae
chromosome was constructed by fusing subgenic
fragments to a promoter-less gfp gene in a conjuga-
tive broad-host-range vector. After conjugation into
B. henselae, the resulting library was used to infect
endothelial cells. Following bacterial uptake, infected
host cells were lysed and the bacteria released were
sorted by FACS for high fluorescence. The recovered
bacterial pool was then axenically cultivated on
blood agar and sorted by FACS for weak fluores-
cence to obtain bacterial clones which differentially
express GFP under the two culture conditions used.
By this means, we have identified a promoter which
is up-regulated approximately 10-fold during endo-
thelial infection as compared to axenic growth on
agar plates. The corresponding gene encodes a
putative outer membrane autotransporter protein,
which may represent a potential novel virulence
determinant of B. henselae.
Session VI: Genetics & Genomics
Main Speaker
Title: Assembly and packaging of the bac-
teriophage particle of Bartonella hen-
selae
Authors: Burt Anderson 1 , Christoph Dehio 2 and
Terri Bowers 1
Institution: 1 University of South Florida, College of
Medicine, Dept. of Medical Microbiol.
and Immunol., Tampa, Florida, 33612,
U.S.A.; 2 Dept. of Infection Biology,
¨
Max Planck Institute for Biology, Tub-
ingen; Germany
Previous studies have shown that B. henselae
contains 14-kilobase extra-chromosomal DNA frag-
ments that are packaged into bacteriophage-like
particles. The packaged DNA appears to be a
random (or near-random) collection of DNA frag-
ments excised from the B. henselae genome. Our
laboratory has attempted to elucidate the processes
by which the DNA fragments are packaged into
bacteriophage particles and exported to the outside of
the bacteria. It is our hypothesis that, either in the
bacterial chromosome or packaged in low abundance
into particles, there exists a set of genes responsible
for the processes of excision, assembly and exporta-
tion. To ascertain the location of these genes we
performed Southern blot analysis. Total DNA from
B. henselae was isolated and digested with several
restriction endonucleases that produce fragments
predominantly between 2 and 30 kilobases. Dupli-
cate sets of the resolved DNA fragments were
transferred to nylon membranes and hybridized with
two probes corresponding to two different particle-
associated genes. The first probe corresponded to
major particle-associated protein gene pap31 and the
second probe, pap36, a protein that has been shown
to react with antibody raised to the particle. No
common sized bands to hybridized with both probes
for any of the restriction endonucleases tested. The
results suggest that these two genes associated with
particle production do not reside on a contiguous
fragment of DNA less than 14-kilobases. It is likely
that particle production is the result of two or more
genes located on distal regions of the B. henselae
genome and resulting in assembly of defective viral
particles.
Session VI: Genetics & Genomics
Main Speaker
Title: Genome sequencing and analysis of
Bartonella henselae
Authors: S.G.E. Andersson, C.M.U. Alsmark, B.
¨
Canback, and C.G. Kurland
 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
Institution: Molecular Evolution and Evolutionary
Biology, Uppsala University, Sweden
The powerful combination of genomics and bioinfor-
matics has provided a wealth of information about
the 2.0 Mb genome of Bartonella henselae, the
causative agent of cat scratch disease. Genome
sequence data has been obtained with a whole
genome random shotgun strategy, using small insert
(2 kb) genome DNA libraries from B. henselae
prepared in M13 vectors. At the date of this writing,
97% of the genome has been sequenced at a 5-fold
coverage. Here, a first preliminary analysis of the
genome sequence data obtained will be presented.
The genes identified have been classified according
to their functional features and a comparative analy-
sis of the metabolic capacities of Rickettsia
prowazekii and B. henselae will be discussed. Both
genomes encode components of the tricarboxylic
acid cycle, the NADH dehydrogenases and the ATP
synthase complexes. However, B. henselae has a
complete set of genes for glycolysis that are not
present in the R. prowazekii genome. Another strik-
ing difference is that B. henselae has a significantly
higher fraction of the genome devoted to biosyn-
thetic and regulatory functions than R. prowazekii.
The comparison of virulence genes provide the
potential to recognize the genetic basis for successful
human infections by the obligate and facultative
intracellular parasites R. prowazekii and B. henselae,
respectively.
Session VI: Genetics & Genomics
Oral Presentation / Poster [26
Title: Characterization of the regulatory region
of Bartonella henselae (Bh) involved in
htrA gene expression
Authors: Sandra I. Resto-Ruiz, Debra Sweger,
and B. E. Anderson
Institution. Dept. of Medical Microbiology and Im-
munology, University of South Florida,
College of Medicine, Tampa, FL, U.S.A.
Studies suggest that the synthesis of stress re-
293
sponse (SR) proteins in prokaryotes is enhanced as a
result of environmental insults. The high temperature
requirement A (htrA) gene is among the bacterial SR
families responsible for the regulation of the ex-
tracytoplasmic heat shock response. Cloning and
sequencing of the htrA gene of Bh traced a 5’
regulatory region harboring a promoter (P1) which
bears high homology with the sigmaE-type heat
shock promoters (Gene 178:35-38). In this study we
report the site of transcription initiation (STI) eli-
cited by P1 using primer extension methodology. An
oligonucleotide primer located approximately 100
bases downstream of P1 was designed and the STI
was mapped to the adenine found four bases down-
stream of P1. Furthermore, we report the finding of a
second promoter P2 within the htrA regulatory
region. Unlike other htrA genes, it was observed that
the DNA fragment extending from P1 to the transla-
tion start codon consisted of approximately 195
bases. This distance appeared to be unusually long
when compared to the htrA gene of Escherichia coli
( Ec); therefore, we searched for additional segments
which could be involved in the regulatory process.
Primer extension reactions, using a primer located
beyond the start of translation, demonstrated a
second STI which mapped to the adenine located in
positon 165. Parallel reactions using RNA isolated
from Ec harboring the construct pBlue /BhhtrA
resulted in analagous sites as those determined with
total RNA from Bh. Homology search showed high
similarity between P2 and the htrA regulatory region
of Brucella abortus. We are presently evaluating the
promoter activity of P1 and P2 and comparing it to
that of other known promoters such as: EclacZ and
EcgroEL. We anticipate these results will assist in
the future evaluation of these promoters as tools in
vaccination and diagnostics.
Session VI: Genetics & Genomics
Oral Presentation / Poster[27
Title: Cloning and characterization of genes
associated with pilin expression in Bar-
tonella henselae
Authors: Michael Schmiederer and Burt E. An-
derson.
294 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295

Institution: Dept. of Medical Microbiology and Im- A system for efficient single-crossover gene disrup-
munology, University of South Florida tion for general use in Bartonella spp. has been
College of Medicine, Tampa, FL, U.S.A. established using the ialB gene of the rat pathogen B.
tribocorum as a model. The ialB gene of B. tri-
bocorum was cloned by means of PCR using primers
Several lines of evidence indicate that B. henselae
encoding the conserved sequences which flank the
express Type IV pili. Pili expression has been
human erythrocyte invasion-associated locus ialB of
observed by electron microscopy. Additional evi-
B. bacilliformis. An internal fragment of the ialB
dence of pili expression by B. henselae can be
gene from B. tribocorum was cloned into pCD394, a
attributed to the phase variation seen in subcultured
minimal vector containing an oriT, a kanamycin
isolates. These smooth phenotypes of B. henselae
resistance cassette and the ori ColE 1 , which allows
lack the ability, of their rough phenotypic counter-
replication in E. coli but not in Bartonella spp. The
parts, to attach and invade human epithelial cells.
resulting plasmid, pCG011, was transferred by
The smooth phenotype also demonstrates a signifi-
conjugation from E. coli to B. tribocorum. By
cant decrease of immunoreactivity as demonstrated
selecting for kanamycin resistance, ex-conjugants
by indirect fluorescent assay. We hypothesize that
were isolated which have integrated this plasmid into
pili play a major role in pathogenesis and also act as
their chromosome as a result of a single-crossover
an immunodominant antigen. Degenerate primers
event, thereby disrupting the chromosomal ialB gene
were constructed with consensus sequences from
by incomplete duplication. PCR analysis allowed us
organisms that express Type IV pili. PCR was
to demonstrate the stability of the generated ialB 2
performed using B. henselae Hou-1 genomic DNA
mutant over several passages on non-selective media.
as a template. A 200 base pair product was amplified
Preliminary data indicate that the ialB 2 mutant has
and cloned into pCR2.1-TOPO vector (Invitrogen,
completely lost the ability to cause bacteremia in
Carlsbad CA). The insert was sequenced revealing an
infected rats, suggesting that this type of gene
open reading frame. The deduced amino acid se-
disruption generates mutants of sufficient stability for
quence has 58% identity over 74 residues to the pilin
in vivo analysis of the resultant mutant phenotype.
transcriptional regulator PILA of N. meningitidis.
Moreover, our data suggest that the ialB locus
PCR was used to label the nucleotide sequence with
plays an important role in the invasion of B. tri-
digoxigenin. This probe was used to screen a B.
bocorum into rat erythrocytes and support the previ-
henselae Hou-1 genomic DNA library to recover the
ous findings that the heterologous expression of the
rest of the gene. A 4.5kb fragment was recovered
two-gene locus ialA /ialB of B. bacilliformis in
from the library and cloned into pBluescript. The
minimally invasive E. coli leads to an increase in
resulting plasmid, pPIL3, is currently being se-
capacity to invade human red blood cells in vitro.
quenced in order to identify the remaining open
reading frame coding this gene.
Session VI: Genetics & Genomics
Session V: Pathogenesis

Poster [28 Poster [29

Title: Site-specific mutagenesis of the ialB Title: Differentiation of Bartonella species by

locus of Bartonella tribocorum by sin- using DNA sequence information of
gle-crossover gene disruption genes involved in riboflavin synthesis

Authors: Christian Gille, Christa Lanz and Authors: Silke Hinkelmann, Anna Sander, Man-
Christoph Dehio fred Kist, and Stefan Bereswill
Institutions: Dept. of Infection Biology, Max Planck Institution: Institute of Medical Microbiology and
¨
Institute for Biology, Tubingen; Ger- Hygiene, University of Freiburg,
many Freiburg, Germany
 Abstracts / Journal of Microbiological Methods 37 (1999) 267 – 295
Introduction and aims: Riboflavin (Rf, vitamin B 2 )
is the precursor for the coenzymes Flavin-Mononu-
cleotide (FMN) and Flavin-Adenine-Dinucleotide
(FAD) which are essential for cell metabolism. The
Rf biosynthesis pathway in E. coli consists of five
enzymes designated as RibA to -E. The conservation
of the Rf synthesis genes in bacteria and the absence
of this biosynthesis pathway in humans provides
evidence that DNA sequence information of the Rf
synthesis genes could be used for detection and
differentiation of invasive pathogens.
It was the aim of this study to investigate whether
Rf synthesis genes are present in Bartonella species
and whether the corresponding DNA sequence in-
formation allows differentiation of Bartonella
species or individual isolates.
Materials and Methods: A plasmid based genomic
DNA library of B. henselae was constructed and
screened for functional complementation of a ribC
deficient mutant strain of E. coli as indicated by
growth on LB agar without Rf.
Results: A DNA fragment carrying the genes ribD,
295
ribC, and ribE was isolated from a DNA library of
Bartonellahenselae. The same fragment could be
amplified by PCR from DNA of B. clarridgeiae, B.
quintana, and B.bacilliformis. The DNA sequence of
the Rf synthesis genes was found to be considerably
different among Bartonella species (78-87%) but
highly conserved among isolates of B. henselae ( .
98%).
Discussion: The results indicate that Bartonella
species carry Rf synthesis genes and that the corre-
sponding DNA sequence information can be used for
differentiation on the species level, but is not suited
for typing of individual strains.
The development of a PCR system for species-
specific detection and differentiation of Bartonella
species on the basis of sequence information of the
ribC gene is currently under investigation.
The complementation of the ribC mutation in E.
coli confirmed that the B. henselaeribC gene is
functionally active in E. coli and encodes a subunit
of the enzyme Rf synthetase which catalyses the
terminal step in Rf synthesis.