www.elsevier.com/locate/intimp

International Immunopharmacology 3 (2003) 853–864

Bartonella quintana lipopolysaccharide effects on leukocytes,

CXC chemokines and apoptosis: a study on the human whole blood

and a rat model

Giovanni Materaa,*, Maria Carla Libertoa, Angela Quirinoa, Giorgio Settimo Barrecaa,Angelo Giuseppe Lambertia, Michelangelo Iannoneb, Eliana Mancusoc,

Ernesto Palmac, Francesco Antonio Cufaric, Domenicantonio Rotirotib,c, Alfredo Focaa

a Institute of Microbiology, Department of Medical Sciences, Faculty of Medicine, University of Catanzaro,

Via T. Campanella 115, I-88100 Catanzaro, Italyb ISN-CNR Section of Pharmacology, Roccelletta di Borgia, Catanzaro, Italy

cFaculty of Pharmacy, University of Catanzaro, Catanzaro, Italy

Received 2 August 2002; received in revised form 3 December 2002; accepted 20 February 2003

Abstract

Bartonella quintana, an emerging gram-negative pathogen, may cause trench fever, endocarditis, cerebral abscess and

bacillary angiomatosis usually with the absence of septic shock in humans. B. quintana lipopolysaccharide (LPS), a deep rough

endotoxin with strong reactivity in the limulus amebocyte lysate (LAL)-assay, was studied in human whole blood and in a rat

model. A significant (P< 0.05) increase of interleukin-8 (IL-8) concentration, comparable to the level induced by enterobacterial

LPS, was stimulated in the human whole blood by B. quintana LPS. Isolated human neutrophils delayed their apoptotic behavior

in the presence of B. quintana LPS. In the rat, B. quintana LPS induced a significant (P < 0.001) increase in white blood cell

count, both 30 and 60 min after intravenous injection. Such leukocytosis was inhibited by pretreatment with prazosin, an a-

adrenergic antagonist. B. quintana LPS did not significantly change heart rate (HR), hematocrit (HCT) and platelet count in the

above reported in vivo model, and regarding mean blood pressure (MAP) only a very early (5 min after LPS) and mild (yet

significant) hypotension was observed. In contrast, a long-lasting decrease of MAP was found in Salmonella minnesota R595

LPS-treated animals. Blood TNFa levels did not change significantly from the baseline in rats injected with either saline or with

B. quintana LPS, on the contrary S. minnesota R595 LPS-injected animals showed substantial increase of TNFa levels up to

2924 pg/ml at 60 min after LPS injection. B. quintana LPS as well as Salmonella LPS-injected rats exhibited an increase of the

blood levels of GRO/CINC-1, particularly at 240 min after LPS administration. Apical part of rat gut villi showed several

TUNEL-positive cells in tissue sections from B. quintana LPS-treated animals. Taken together, our data demonstrates that B.

quintana LPS is able to selectively stimulate some inflammatory mediators. B. quintana LPS-induced leukocytosis appears

mediated by an a-adrenergic receptor. The delayed apoptotic process of leukocytes and the chemokine increase may explain the

apoptotic cells found in the rat gut and the inflammatory reactions in some human Bartonella diseases. This peculiar

1567-5769/03/$ - see front matter D 2003 Elsevier Science B.V. All rights reserved.

doi:10.1016/S1567-5769(03)00059-6

* Corresponding author.

G. Matera et al. / International Immunopharmacology 3 (2003) 853–864854

inflammatory pattern induced by B. quintana LPS, may partially account for the lack of severe septic shock, observed in human

B. quintana infections.

D 2003 Elsevier Science B.V. All rights reserved.

Keywords: Neutrophils; Endotoxins; Interleukins

1. Introduction tis and in several others disease due to Bartonella)

Together with well-characterized acute and chronic

diseases such as trench fever, bacteremia, endocardi-

tis, bacillary angiomatosis, osteitis, present knowledge

about Bartonella quintana infection includes a broad

range of non-specific clinical symptoms and signs

[1,2]. Cases reported such as suspected bacteremia

due to B. quintana might represent only a small

percentage of the patients actually infected, as sug-

gested by the results of the Seattle seroprevalence

survey [3]. Chronic B. quintana bacteremia associated

with severe headaches and leg pain as well as with

low platelet counts, has been reported very recently in

homeless patients [4]. Ileitis has been associated with

Bartonella infection in humans [5].

Pathogenic mechanisms of Bartonella infections

are poorly understood, particularly in vivo.

Endothelial changes play a crucial role in bacil-

lary angiomatosis and in the heart valves pathology

of patients suffering from B. quintana endocarditis

[6,7].

Also, monocytes, macrophages and neutrophils

might be involved in engulfment of Bartonella spp.

[8].

LPS found in the outer membrane of gram-

negative envelope, may be released as cell wall

blebs [9], and represents one of the bacterial com-

ponents mainly responsible for pathogenic mecha-

nisms [10]. However, very little information is

available on the pathogenic role of Bartonella LPS

[11]. In our laboratory, we have extracted and

characterized B. quintana LPS and we have also

investigated the in vitro features of LPS interaction

with an endothelial cell line, as a major intracellular

replicative site of B. quintana [12]. Up to our

knowledge, it was not addressed the question why

a bacterium, that may produce periodic bacteremias

(e.g. as in trench fever) and that stimulates clinical

features (e.g. again as in trench fever, in endocardi-

quite typical for a conventional endotoxin bearing

organism, does not usually cause septic shock

[1,4,8].

The aim of the present paper is to investigate some

effects of B. quintana LPS after in vivo administra-

tion in the rat and following in vitro stimulation of

the human whole blood and of the isolated human

PMN. Evaluation and neutralization of endotoxic

activity of this LPS using the LAL-assay were also

reported.

2. Materials and methods

2.1. Evaluation and neutralization of endotoxic

activity of B. quintana LPS

The endotoxic activity of extracted B. quintana

LPS was evaluated using the limulus amebocyte

lysate test (LAL test) (QCL-1000; BioWhittaker Bio-

products, Walkerville, MD, USA). As reference com-

pounds for LAL experiments, one rough Re che-

motype LPS (from Salmonella minnesota R595;

Sigma, St. Louis, MO, USA) and one smooth type

LPS (Proteus mirabilis LPS, RIBI, Montana, USA)

were used. Also, its in vitro neutralization by anti-

biotics was evaluated by LAL reactivity inhibition, as

previously reported [13]. Results are shown as O.D. at

405 nm.

2.2. Induction of IL-8 release in human whole blood

after stimulation with B. quintana LPS

Blood samples were obtained from healthy volun-

teers and aliquoted. The first aliquot was added with

0.03 ml of 15% EDTA and tested for differential cell

count using a MAX-M cell counter (Coulter, Luton,

UK). Blood samples exhibited meanF S.E.M. values

for HCT of 46F 1%, for total leukocyte count of

G. Matera et al. / International Immunopharmacology 3 (2003) 853–864 855

7200F 190 cells/Al and for monocyte count of

360F 20 cells/Al. Endotoxin-free heparin (50 IU/ml

final concentration), instead of EDTA, was added to

all the remaining aliquots of each blood sample. A

second aliquot of 1.8 ml of each blood sample

(control) was added with 0.2 ml of sterile saline and

processed immediately as reported below. Remaining

aliquots of each blood sample were distributed into

sterile polypropylene tubes (1.8 ml/tube), and 0.2 ml

volume of purified LPS (LPS-treated; final concen-

tration 1000 ng/ml) or the same volume of sterile

saline (control) was added. Thus, 648� 104 WBC/ml

was used. Whole blood incubation was performed in

non-rotating tubes at 37 jC. After 2, 4 and 24 h of

incubation, plasma was separated from blood cells by

centrifugation for 5 min at 1500� g, and stored at

� 20 jC in different aliquots. IL-8 concentration was

evaluated by enzyme-linked immunosorbent assay

(ELISA).

2.3. Inflammatory mediators detection by ELISA

Human IL-8, as well as rat TNFa and GRO/CINC-

1 concentrations were measured by commercially

available ELISA kits (Amersham Pharmacia Biotech,

Italy). The lower limit of sensitivity was 4 pg/ml for

IL-8 assay. The sensitivity of rat TNFa and GRO/

CINC-1 assays were 10 and 4.7 pg/ml, respectively.

Concerning IL-8 concentration in human whole

blood, data obtained from each assay were corrected

for blood dilution value in order to reflect the IL-8

concentration as pg/ml of undiluted blood. Since

values of HCT and differential WBC showed only

negligible differences among samples from different

donors, we did not try to adjust IL-8 values for

monocyte and/or white cell count.

2.4. Human PMN apoptosis

Neutrophils were purified by dextran sedimentation

from blood samples obtained by human volunteers.

Cells were counted by hemocytometer, adjusted to

5� 105 cells/ml and cultures were carried out into

a 3-cm petri dish, containing a coverslip and 2 ml

of RPMI 1640, with 10% fetal calf serum. Then,

B. quintana LPS or S. minnesota R595 LPS, after

dilution in RPMI 1640, was added to treated cultures

in order to reach a final concentration of 10 ng/ml.

Incubation was carried for different times in a 5% CO2

atmosphere. Before microscopic evaluation of apop-

tosis, the coverslip of the PMN culture to be examined

was picked up with sterile forceps and placed face-up

down on a glass slide, bearing 10 Al of kit buffer plus1Al annexin and 1 Al propidium (Annexin V-EGFP

staining kit, MBL, NaKa-Ku Nagaya, Japan). Appli-

cation of a DNA staining, propidium iodide (PI),

allowed to discriminate apoptosis from necrosis. Cells

in earlier stages of apoptosis, which have a normal

appearance and display a green fluorescence on outer

leaflet, are named annexin-positive cells (A cells).

Cells in later stages of apoptosis begin to lose mem-

brane integrity and show an additional red fluores-

cence on strongly condensed nuclei. These cells are

reported as annexin-propidium-positive cells (AP

cells) and may be easily distinguished from necrotic

cells which appear diffusely stained with PI, and are

reported as propidium-positive cells (P cells), with

nuclei conserving their shapes. Percentages of apop-

totic cells were determined evaluating at least 500 cells

per sample, and counting A, AP and P cells, using a

fluorescence microscope (Leica Microsystem Wetzlar

GmBH, Wetzlar, Germany).

2.5. Rat hemodynamic, hematological and cytokine

measurements

Male Wistar rats (200–250 g b.w., Charles River,

Italy) were anesthetized with sodium pentobarbital

(50 mg/kg; i.p.), and instrumented for measurement

of arterial blood pressure and heart rate. Using a

Gould-Statham P23Db pressure transducer attached

to a PowerLab (AD Instruments) and to a Columbus

Instruments Computer, blood pressure and heart rate

were continuously displayed and recorded. After

completion of surgical procedures, animals were

allowed to stabilize their cardiovascular parameters

for 20 min. Then, baseline hemodynamic values were

recorded, a blood sample was obtained for white

blood cell and platelet counts and for hematocrit

evaluation and an equal volume of sterile saline

was returned to the animal. After 5 min, animals

received LPS from B. quintana (10 mg/kg, i.v.; in 0.2

ml/100 g b.w. of sterile saline) or LPS from S.

minnesota R595 (10 mg/kg, i.v.; in 0.2 ml/100 g

b.w. of sterile saline) or 0.2 ml/100 g b.w. of sterile

saline (controls) as a slow injection over 2 min.

Table 1

Reactivity of LPS from B. quintana, P. mirabilis and S. minnesota

R595 measured by LAL test

LPS O.D. 405 nm

(pg/ml)B. quintana P. mirabilis S. minnesota R595

1.5 N.D. N.D. N.D.

3.1 0.149F 0.058 N.D. N.D.

6.2 0.351F 0.060 N.D. N.D.

G. Matera et al. / International Immunopharmacology 3 (2003) 853–864856

Further, blood samples were obtained 30 and 60 min

after LPS or saline administration, unless otherwise

specified.

Separate groups of anesthetized animals received

either an a-adrenergic antagonist, prazosin (1 mg/kg,

i.v.; prazosin hydrochloride, Sigma), or saline admin-

istered 20 min before the LPS from B. quintana (10

mg/kg, i.v.) or 0.2 ml/100 g b.w. of sterile saline

(controls), in order to evaluate the role of a-adrenergic

mechanism in the leukocytosis induced by B. quin-

tana LPS.

Hematological measurements were carried out by a

automatic blood cell counter (MAX-M, Coulter).

Information on rat TNF a and GRO/CINC-1 blood

levels were obtained from different groups of anes-

thetized animals treated either with one of the above

LPSs (10 mg/kg, i.v.; in 0.2 ml/100 g b.w. of sterile

saline), or with sterile saline (0.2 ml/100 g b.w., i.v.).

2.6. Histology

In a separate group of experiments, the rats (n = 4

per each group) were anesthetized with sodium pen-

tobarbital 50 mg/kg, i.p. and treated with B. quintana

LPS (10 mg/kg, i.v.; in 0.2 ml/100 g b.w. of sterile

saline) or S. minnesota R595 LPS (10 mg/kg, i.v.; in

0.2 ml/100 g b.w. of sterile saline) or 0.2 ml/100 g

b.w. of sterile saline (controls) as a slow injection over

2 min.

After 3 h, the rats were transcardially perfusion-

fixed with 4% buffered formaldehyde pH 7.4 after a

brief rinse with saline and heparin (0.1%) at room

temperature. Gut, lung, liver and kidney were re-

moved, kept in cold fixative for 2 h, and stored in

phosphate-buffered saline (PBS) overnight. Tissues

were dehydrated in ethanol and embedded in paraffin.

Sections (5 Am) were cut in a microtome and stained

with hematoxylin–eosin (H&E) for light microscopic

examination. Terminal deoxynucleotidyltransferase-

mediated dUTP nick end labelling (TUNEL) was

performed according to the producer’s protocol (MK

500; Takara Biomedical Group, Japan).

12.5 0.587F 0.029 0.160F 0.018 0.042F 0.012

25 0.694F 0.042 0.245F 0.029 0.268F 0.020

50 1.095F 0.048 0.385F 0.021 0.671F 0.061

100 1.297F 0.050 0.831F 0.018 0.941F 0.031

Data are meansF S.E.M. of five experiments, which were carried

out in duplicate.

N.D.: not detectable.

3. Statistical analysis

The data are reported as meansF standard error of

the mean (S.E.M.), unless otherwise specified. One-

way ANOVA was used to analyse the data and

significant differences between groups were deter-

mined by Fisher’s Protected Least Significant Differ-

ence (PLSD) test. A P < 0.05 was considered to be

statistically significant.

4. Results

4.1. Evaluation and neutralization of endotoxic

activity of B. quintana LPS

Data concerning LAL-test reactivity are reported in

Table 1. When compared to other reference endotox-

ins, B. quintana LPS showed a strong endotoxic

activity, as measured by LAL-test. Furthermore, this

activity was also evident when concentrations of 3.1

and 6.2 pg/ml of endotoxin were tested, while other

reference endotoxins resulted undetectable at these

two low concentrations.

The direct in vitro neutralization of the LPS from

B. quintana by antibiotics was evaluated using the

technique of the inhibition of the reactivity of bacte-

rial LPS in the LAL-test and the data are reported in

Table 2. A significant and dose-dependent reduction

of the LPS reactivity due to polycationic antibiotics

and particularly to polymyxin B can be observed. On

the contrary, no significant modifications of the LPS

reactivity was caused by the beta-lactam used (cefta-

zidime). A comparable reduction of LPS reactivity in

the LAL test was found using one reference endotoxin

(from S. minnesota R595) plus the above polycationic

Table 2

Effect of polymyxin B (PB), tobramycin (TOB), isepamicin (ISE)

and ceftazidime (CTZ) on the reactivity of B. quintana, S.

minnesota R595 and P. mirabilis LPS (Control, 50 pg/ml) measured

by LAL test

Antibiotic O.D. 405 nm

(pg/ml)B. quintana S. minnesota R595 P. mirabilis

Control 1.095F 0.048 0.806F 0.030 0.398F 0.25

PB (10) 0.875F 0.006 0.513F 0.087* 0.391F 0.016

PB (100) 0.560F 0.033* 0.150F 0.067* 0.354F 0.026

PB (1000) 0.381F 0.046* 0.118F 0.001* 0.381F 0.006

TOB (10) 0.789F 0.020 0.554F 0.014* 0.365F 0.007

TOB (100) 0.732F 0.027* 0.322F 0.064* 0.372F 0.001

TOB (1000) 0.697F 0.025* 0.171F 0.001* 0.386F 0.014

ISE (10) 0.841F 0.018 0.327F 0.036* 0.397F 0.018

ISE (100) 0.708F 0.039* 0.178F 0.021* 0.394F 0.013

ISE (1000) 0.650F 0.038* 0.104F 0.019* 0.372F 0.007

CTZ (10) 0.866F 0.003 0.791F 0.053 0.352F 0.026

CTZ (100) 0.784F 0.016 0.890F 0.032 0.374F 0.011

CTZ (1000) 0.789F 0.007 0.861F 0.022 0.380F 0.007

Data are meansF S.E.M. of five experiments, which were carried

out in duplicate. *P< 0.01 vs. antibiotic-free LPS Control, by

Fisher’s PLSD test.

G. Matera et al. / International Immuno

antibiotics. On the contrary, a lack of LPS neutraliza-

tion was observed when we used the well-known

polymyxin-resistant LPS from P. mirabilis.

Fig. 1. Kinetics of the release of IL-8 in human whole blood stimulated w

R595 LPS (1000 ng/ml, i.v.; open circles). Instead of LPS, sterile saline w

meansF S.E.M. of data from three healthy donors tested in duplicate. Each

time, by Fisher’s PLSD test.

4.2. Induction of IL-8 release in human whole blood

after stimulation with B. quintana LPS

The activity of B. quintana LPS on IL-8 release is

shown in Fig. 1. We found a significant (P < 0.05 vs.

control values) increase of IL-8 along with the time

(2–4–24 h, after 1000 ng/ml of LPS). The LPS from

S. minnesota R595 (1000 ng/ml) stimulated a slower,

but significant, increase of IL-8 at 4 h after LPS,

whereas after 24 h both LPS induced almost the

same levels of this chemokine. Lower dosages of

both LPS did not achieve a statistically significant

IL-8 increase in comparison to control (data not

shown).

4.3. Human PMN apoptosis

Human neutrophil apoptosis was studied by An-

nexinV–Propidium staining. PMN were scored as

‘‘A +AP’’ (Annexin plus Annexin/Propidium posi-

tive, that is early apoptotic cells plus late apoptotic

ones) or ‘‘P’’ (Propidium-only—positive, that is nec-

rotic cells).

In control PMN, a rapid increase of A +AP per-

centage was found up to the sixth hour, then the

pharmacology 3 (2003) 853–864 857

ith B. quintana LPS (1000 ng/ml, open triangles), or S. minnesota

as added to control human whole blood (open squares). Values are

experiment was carried out twice. *P < 0.05 vs. control at the same

Fig. 2. Apoptosis induced by B. quintana LPS on PMN evaluated by the Annexin-V Propidium staining. Kinetics of percentages of early + late

apoptotic (A +AP) cells observed in control human PMN (open squares), in B. quintana LPS-stimulated human PMN (open lozenges), or in S.

minnesota R595 LPS-stimulated human PMN (open circles). It is also shown the kinetic of percentages of necrotic (P) cells found in control

human PMN (crossed squares), in B. quintana LPS-stimulated human PMN (crossed lozenges), or in S. minnesota R595 LPS-stimulated human

PMN (crossed circles). Data are shown as mean percentageF S.E.M. of positive cells, out of at least 500 counted ones, during three experiments

carried out in duplicate. See the text for details.

G. Matera et al. / International Immunopharmacology 3 (2003) 853–864858

percentage of A +AP cells decreased to a value lower

than that observed at 2 h; on the contrary, the curve

representing the necrotic PMN rose slowly but con-

stantly and crossed the descending A+AP cell curve

between 6 and 24 h (Fig. 2). Other investigators

reported steadily rising proportions of necrotic PMN

after 18–30 h of in vitro culture. On the contrary, the

percentage of early apoptotic cells that had showed an

Table 3

Changes (%) of mean arterial pressure (MAP) of Wistar rats at 5, 15, 30, 45

mg/kg) and S. minnesota R595 LPS (10 mg/kg), as well as sterile saline

Treatment Time (min)

5 15

B. quintana LPS 75.2F 5.8 ( P= 0.013) 95.6F 5.4

S. minnesota R595 LPS 70.1F 4.8 ( P= 0.006) 76.8F 9.1

Control 103.9F 5.0 92.9F 1.9

Values are meansF S.E.M. of data from six animals.

P vs. control at the same time, by Fisher’s PLSD test.

increase until 18–30 h, then exhibited a progressive

decrease [37].

B. quintana LPS (10 ng/ml) caused a delay of the

apoptotic course of human neutrophils (Fig. 2). The

concentration of 10 ng/ml of LPS was selected during

preliminary experiments, as the lowest dosage of LPS

producing a substantial difference in the apoptosis

percentage when compared to the control PMN. On

and 60 min after intravenous administration of B. quintana LPS (10

(control)

30 45 60

83.9F 4.7 87.8F 7.4 79.4F 6.2

(P= 0.04) 73.6F 8.3 81.6F 17.6 70.5F 17.1

81.9F 5.0 88.5F 13.0 79.8F 8.5

Table 5

Effect of prazosin pretreatment (1 mg/kg, i.v., 20 min before LPS or

saline) on changes (%) of WBC counts of Wistar rats at 30 and 60

min after intravenous administration of B. quintana LPS (10 mg/

kg), or sterile saline

Pretreatment Treatment Time (min)

30 60

Saline B. quintana

LPS

144F 15.4 149.4F 16.7

Prazosin B. quintana

LPS

96.8F 15.1

(P= 0.008)

93.8F 18.4

(P= 0.002)

Prazosin Saline 108F 14.1 104.5F 5.0

Values are meansF S.E.M. of data from six animals.

P vs. saline plus B. quintana LPS-injected rats at the same time, by

Fisher’s PLSD test.

G. Matera et al. / International Immunopharmacology 3 (2003) 853–864 859

the contrary, higher concentrations did not modify the

PMN apoptotic behavior already observed following

10 ng/ml of LPS.

PMN added with B. quintana LPS showed a

slowly increasing A +AP curve with a reduced

slope for all the experimental period, while the P

curve was always close to the baseline values; the

two curves never crossed each other within 24 h.

We observed a similar trend using S. minnesota

R595 LPS, however, the percentage values were

all greater than in B. quintana LPS-treated PMN

and the corresponding curves were shifted to the

left.

Therefore, the time-course of the A+AP and P

curves of the control PMN demonstrate that the

whole process of apoptosis lasted less than 24 h

for a large percentage of PMN. On the contrary, a

delay of apoptotic process appeared in the PMN

treated with both LPS used, since the A +AP curve

and the P curve did not cross each other during the

first 24 h.

4.4. Rat hemodynamic, hematological and cytokine

measurements

B. quintana LPS did not significantly change

heart rate (HR), hematocrit (HCT) and platelet count

in the rat, however, mean arterial pressure (MAP)

values decreased significantly (P= 0.013 vs. control

at same time) at 5 min after LPS injection (Table 3).

As expected the LPS from Salmonella induced a

significant hypotension at 5 and 15 min after LPS

injection.

Salmonella LPS caused a significant thrombocyto-

penia at 30 and 60 min and a significant leukopenia

after 60 min since its administration (Table 4). B.

Table 4

Changes (%) of WBC and platelet counts of Wistar rats at 30 and 60 min af

minnesota R595 LPS (10 mg/kg), as well as sterile saline (control)

Treatment Time (min)

30

WBC Platelets

B. quintana LPS 154.2F 14.5 ( P< 0.001) 105.1F 0.9

S. minnesota R595 LPS 70.4F 9.8 74.2F 5.2 ( P

Control 89.1F 4.6 114.3F 15.3

Values are meansF S.E.M. of data from six animals.

P vs. control at the same time, by Fisher’s PLSD test.

quintana LPS consistently induced a significant in-

crease in white blood cell (WBC) counts after intra-

venous injection in Wistar rats (Table 4). It was noted

that 30 and 60 min after B. quintana LPS, WBC

increased up to 154.2F 14.5% and 158.4F 15.8%

compared with basal level (P < 0.001 vs. control at

the same time).

Using separate groups of animals, we found that in

prazosin-pretreated B. quintana LPS-injected rats,

WBC were found 96.8F 15.1% and 93.8F 18.4%

of the basal level at 30 and 60 min after LPS. On the

contrary, in saline-pretreated B. quintana LPS-injected

animals, WBC changes overlapped those high values

found in unpretreated B. quintana LPS-injected rats.

In prazosin-pretreated LPS vehicle-injected animals,

WBC showed only negligible changes from baseline

at both times (Table 5). Therefore, prazosin blunted

the B. quintana LPS-induced increase of WBC, with-

out producing any WBC change in LPS vehicle-

treated animals.

ter intravenous administration of B. quintana LPS (10 mg/kg) and S.

60

WBC Platelets

158.4F 15.8 ( P < 0.001) 114.1F11.4

< 0.001) 49.2F 6.8 ( P= 0.004) 85.1F 5.2 ( P= 0.04)

82.9F 0.5 104.8F 1.0

Fig. 3. Kinetics of blood levels of TNFa (A) and GRO/CINC-1 (B)

in the rat injected with B. quintana LPS (10 mg/kg, i.v.; open

lozenges), or S. minnesota R595 LPS (10 mg/kg, i.v.; open circles).

Data are meansF S.E.M. from groups of five rats injected with

either LPS (from B. quintana or from S. minnesota R595 LPS) or

with saline (open squares). *P< 0.05 vs. control at the same time,

by Fisher’s PLSD test.

Fig. 4. (A) Apical part of rat gut villi exhibited apoptotic cells

(mainly villi) in B. quintana LPS-injected (10 mg/kg, i.v.) animals,

as shown by TUNEL-processed tissue sections. (B) Animal injected

with LPS from S. minnesota R595 (10 mg/kg, i.v.) showed similar

apoptotic modifications of gut villi. (C) Absence of apoptotic cells

in TUNEL-processed tissue sections obtained from gut of saline-

injected control rats. Representative findings from one of the four

animals used for each treatment.

G. Matera et al. / International Immunopharmacology 3 (2003) 853–864860

Data from experiments carried out in parallel with

different groups of rats injected with either LPS (from

B. quintana or from S. minnesota R595 LPS) or with

saline, showed that blood TNFa levels did not move

from baseline in saline-injected rats. Similarly, in the

rats stimulated by B. quintana LPS, TNFa did not

change from background levels. On the contrary, S.

minnesota R595 LPS-injected animals showed a sharp

increase of TNFa levels (Fig. 3A).

G. Matera et al. / International Immunopharmacology 3 (2003) 853–864 861

Samples from the above animals showed negligible

modifications over the time in the blood levels of

GRO/CINC-1, after saline injection. However, in

samples of B. quintana LPS-stimulated rats we found

an increase of GRO/CINC-1, which achieved a sig-

nificant (P < 0.05) difference vs. control at 60 min and

at 240 min after the challenge. Also, Salmonella LPS-

injected rats exhibited a significant (P < 0.05) GRO/

CINC-1 elevation at the same times after the stimulus

(Fig. 3B).

4.5. Histological studies

Apical part of rat gut (ileum) villi showed apop-

totic cells (mainly villi; very rarely goblet cells) in B.

quintana LPS-injected animals, as shown by TUNEL-

processed tissue sections (Fig. 4A). Animals injected

with LPS from S. minnesota R595 showed similar

apoptotic modifications of gut villi (Fig. 4B). No

Tuneel-positive cells were evident in gut villi of rats

treated with saline (Fig. 4C). Apoptotic features

showed, refer to samples obtained from rats 3 h after

LPS administration (10 mg/kg, i.v.). The morpholog-

ical examination of H&E-stained sections from B.

quintana LPS-treated animals did show the presence

of an infiltrate of neutrophils in the gut villi (data not

shown).

No morphological modifications were seen in lung,

liver and kidney in all the animals examined.

5. Discussion

This is the first report on the effects of B. quintana

LPS on IL-8 in human whole blood and on human

PMN apoptosis. Also, we evaluate previously not

addressed activities of B. quintana LPS on the car-

diovascular system, hematological values and on

tissue apoptosis in the rat. For this purpose we used

B. quintana LPS isolated from aqueous phase follow-

ing Westphal and Jann procedure, as modified by

Minnick [14]. Other investigators reported data on

B. quintana LPS extracted from phenol phase [11].

The two products are quite different and we used the

aqueous phase extract because KDO percentage and

the protein content are more satisfactory. In particular,

KDO amount is comparable with that found in the

enterobacterial LPS and the protein percentage of our

product is similar to that reported by other authors in

water–phenol extracted LPS [15].

In our laboratory, we have previously demonstrated

that lipopolysaccharide extracted from B. quintana

Oklahoma strain showed a molecular weight of about

5000 with a ‘‘deep rough’’ chemotype [12]. This

finding is consistent with results reported by others

dealing with the LPS isolated from Bartonella bacil-

liformis [14].

B. quintana LPS exhibited marked endotoxicity

and was significantly neutralized with polycationic

antibiotics in an LAL-assay model (see Tables 1 and

2). This observation may contribute to explaining in

vitro bactericidal activity of aminoglycosides against

many species of genus Bartonella [16] and the suc-

cess of therapeutic aminoglycoside use in some clin-

ical cases of bartonellosis [8]. The absence of Proteus

LPS neutralization by polycationic antibiotics has

been reported to correlate with the esterification of

the core-lipid A phosphates and core KDO, which has

been found in Proteus LPS [17]. On the contrary, LPS

from B. quintana, as well as from S. minnesota R595

might make such acidic groups (due to the lack of

esterification) available to polycationic antibiotics.

Although our B. quintana LPS showed a strong

endotoxic activity in LAL-assay (where it appeared

more potent than other reference LPS tested), it was

not able to induce a substantial increase of TNFa

levels in human whole blood (data not shown) and in

rats (see Results).

In B. quintana LPS-challenged human endothelial

cells, we previously demonstrated both transcriptional

and protein low levels of TNFa [12]. These data

underlie that the high endotoxic reactivity, showed

by LAL-test, is not associated to a potent proinflam-

matory activity in very different models and this is

particularly true looking at TNFa levels in the rat.

Such discrepancies have been demonstrated for other

LPS [18].

However, in this study we show that IL-8 is

increased in B. quintana LPS-stimulated human

whole blood. In a previous study, IL-8 has been found

to be the major chemokine induced by B. quintana

LPS on endothelium [19]. Moreover, endothelial IL-8

has been demonstrated to be also upregulated at tran-

scriptional level, as indicated by mRNA semiquanti-

tative PCR findings [12]. In addition, IL-8 has been

reported to be a crucial factor in LPS-mediated angio-

G. Matera et al. / International Immunopharmacology 3 (2003) 853–864862

genesis [20] and may contribute to the proliferative

lesions found in both human and animal bartonellosis.

IL-8 level may be even greater than we found, due to

interactions of cytokines and chemokines with gran-

ulocytes, erytrocytes and plasma soluble receptors.

Such interactions may be overcome by using a deter-

gent such as Triton X-100, but such detergent may

interfere [21] with non-electrochemiluminescence

assay of cytokines (e.g. ELISA).

Blood levels of TNFa did not change in our rats

after B. quintana LPS treatment, while reference

Salmonella LPS caused a substantial increase. The

low but detectable levels of TNFa found in baseline

blood samples and in samples from control animals

may be explained with a moderate stress due to the

anesthesia and to the effect of intravascular catheter

used for removal of blood samples [22].

Our data showed that GRO/CINC-1, the rat’s

chemokine equivalent to human IL-8 [23], is en-

hanced by B. quintana LPS treatment. GRO/CINC-1

has been demonstrated to play a potential role in

neutrophil activation during the rat’s response to

various inflammatory stimuli including LPS [23]. It

has been reported that in IL-8-treated primates, as well

as in CINC-administered rats, TNFa levels do not

increase, and that co-injection of LPS and anti-CINC

antibody, causes a significant increase of TNFa in the

rat [24]. Rats challenged with B. quintana LPS

showed only a very early and mild (yet significant)

hypotension. On the contrary, a sharp and long-lasting

decrease of MAP was observed in S. minnesota R595

LPS-treated animals. TNFa is well known for its

direct hypotensive effect, which may take place within

minutes to hours, both in human [25] and in the rat

[26]. On the other hand, the CXC chemokine IL-8 has

been reported to exhibit beneficial antishock effects

[27].

Thus, the lack of TNFa increase and the beneficial

cardiovascular effects of the rat CXC chemokine,

GRO/CINC-1 might explain the lack of a long-lasting

hypotension in B. quintana LPS-treated rats. How-

ever, the very early effect on the MAP, produced by

both LPS used may suggest a mediator, different from

TNFa [26], rapidly released after LPS and with a very

quick time course. Other authors demonstrated that

the maximal hypotensive response was observed at 10

min after LPS injection in anesthetized rats. Such

early reduction in blood pressure was associated with

enhanced plasma concentration of norepinephrine and

cardiac functional alterations [28]. It has already been

reported that pretreatment of LPS-injected rats with a-

adrenergic antagonist ameliorated survival and shock

sequelae [29]. In preliminary experiments, we found

that rat pretreatment with the a-adrenergic antagonist

prazosin abolished the very early hypotensive effect

of Salmonella and Bartonella LPSs (data not shown).

Overall, an a-adrenergic mechanism may play a role

in the early hypotension caused by both Bartonella

and Salmonella LPSs.

Moreover, in the present study we demonstrate a

consistent and significant increase of PMN in blood

samples of rats injected with B. quintana LPS. Hyper-

leukocytosis has also been observed during human B.

quintana infection [8]. Altenburg et al. [30] reported

that LPS-stimulated neutrophilia in the rat is mediated

by an a1-adrenoceptor. Our data on prazosin-mediated

inhibition of B. quintana LPS -induced PMN change,

suggest that such an event is probably mediated by the

catecholamine norepinephrine through an a-adrener-

gic receptor activation. Quick neutrophilia by cate-

cholamine-mediated demargination might fit well

with early neutrophilia observed after B. quintana

LPS. On the other hand, TNFa, a major cytokine

inducing neutropenia by PMN margination [31], did

not appear stimulated by B. quintana LPS in this

study and was found to be downregulated by the same

LPS in a previous investigation [12].

To the extent of our knowledge, no data are

available from the literature on chemokine levels

during in vitro or in vivo experiments using B.

quintana products and in particular LPS. Regarding

TNFa it has been demonstrated that this cytokine was

released by J774, a murine macrophage cell line, after

in vitro stimulation with heat-killed Bartonella hense-

lae [32].

The pathological study of tissues from rats treated

with B. quintana LPS also demonstrated that such

LPS is able to induce apoptosis when injected in rats

and that gut villi show a strong sensitivity to endo-

toxin-induced cellular damage. This is probably due,

even in absence of a substantial hypotension, to a

damage caused by chemokines-stimulated leukocytes.

This hypothesis is consistent with the presence,

reported in this study, of neutrophil infiltration at the

gut villi level in treated-LPS, but not in untreated

animals.

G. Matera et al. / International Immunopharmacology 3 (2003) 853–864 863

Despite the fact that evidence of a modulatory

effect of LPS on cellular apoptosis in vivo and in

vitro abound, many investigators report divergent

results. Even among phagocytes, LPS has been dem-

onstrated to increase macrophage apoptosis and delay

neutrophil death [33].

Our data are consistent with a positive influence of

LPS on rat villi cells apoptosis, that is apparently in

contrast with the delay observed in the neutrophil

apoptosis. However, the ‘‘sensitivity’’ of the gut to

LPS could be explained with the experimental obser-

vation of Schmidt et al. [34] that a reduction of villus

blood flow, due to a vasoconstriction in the central

villus arterioles, occurs during normotensive endotox-

emia in LPS-treated rats. Recently, ileitis has been

associated with Bartonella infection in humans [5].

Moreover, we observed an increased lifespan of

LPS-stimulated human PMN; this delay of PMN

apoptosis is consistent with previous reports evaluat-

ing the same parameter after neutrophil stimulation

with conventional LPS from enterobacteria, as well as

with the much less potent LPS from Helicobacter,

Porphyromonas and Bacteroides [35,36]. Bacteria

bearing a low-potency LPS may take advantage of a

lack of a very effective inflammatory and immune

response. Therefore, they can maintain a fair inflam-

matory activity for a much longer time, than bacteria

with conventional LPS, because the LPS tolerance is

substantially absent, thus leading to a chronic inflam-

mation [35,36].

In conclusion, our results (obtained in two different

models) are evidence that B. quintana LPS is able to

act selectively on certain inflammatory mediators (e.g.

chemokines) and this could be a very important

pathogenic mechanism in bartonellosis. Indeed, the

weak stimulus for TNFa release induced by B. quin-

tana LPS in comparison to conventional LPS from

enterobacteria may account for the lack of acute

inflammatory response and the absence of long-last-

ing hypotension evident both in our experimental

model and in clinical reports of bartonellosis. This

pathogenetic feature may favor a more chronic evo-

lution of the Bartonella-caused diseases probably due

to the lack of quick clearance of the bacteria; such a

picture has been proposed [35,36] for other long-

lasting bacterial diseases associated with low TNFa

levels, high IL-8 release and a delay of apoptosis of

different cell types including neutrophils.

Acknowledgements

We are indebted to Dr. D. Raoult, University of

Marseille, for the kind gift of B. quintana, Oklahoma

strain, used in present work. Our thanks go to Mr.

Giovanni Politi, D. Saturnino, A. Macri and S. Frustaci

for excellent technical support.

References

[1] Drancourt M,Moal V, Brunet P, Dussol B, Berland Y, Raoult D.

Bartonella (Rochalimaea) quintana infection in a seronegative

hemodialyzed patient. J Clin Microbiol 1996;34:1158–60.

[2] Raoult D, Drancourt M, Carta A, Gastaut JA. Bartonella (Ro-

chalimaea) quintana isolation in a patient with chronic aden-

opathy, lymphopenia, and a cat. Lancet 1994;343:977.

[3] Jackson LA, Spach DH, Kippen DA, et al. Seroprevalence of

Bartonella quintana among patients at a community clinic in

downtown Seattle. J Infect Dis 1996;173:1023–6.

[4] Brouqui P, Lascola B, Roux V, Raoult D. Chronic Bartonella

bacteremia in homeless patients. N Engl J Med 1999;340:

184–9.

[5] Massei F, Massimetti M, Messina F, Macchia P, Maggiore G.

Bartonella henselae and inflammatory bowel disease. Lancet

2000;356:1245–6.

[6] Conley T, Slater L, Hamilton K. Rochalimaea species stimu-

late human endothelial cell proliferation and migration in vi-

tro. J Lab Clin Med 1994;124:521–8.

[7] Brouqui P, Raoult D. Bartonella quintana invades and multi-

plies within endothelial cells in vitro and in vivo and forms

intracellular blebs. Res Microbiol 1996;147:719–31.

[8] Maurin M, Raoult D. Bartonella (Rochalimaea) quintana in-

fections. Clin Microbiol Rev 1996;9:273–92.

[9] Devoe IW, Gilchrist JE. Release of endotoxin in the form of

cell wall blebs during in vitro growth of Neisseria meningi-

tidis. J Exp Med 1973;138:1156–67.

[10] Parrillo JE. Pathogenetic mechanisms of septic shock. N Engl

J Med 1993;328:1471–7.

[11] Hollingdale MR, Vinson JW, Herrmann JE. Immunochemical

and biological properties of the outer membrane associated

lipopolysaccharide and protein of Rochalimaea quintana. J

Infect Dis 1980;141:672–9.

[12] Liberto MC, Matera G. Pathogenic mechanisms of Bartonella

quintana. New Microbiol 2000;23:449–56.

[13] Matera G, Berlinghieri MC, Barreca GS, Foca A. In vitro

effects of novel glycopeptide antibiotics on the reactivity of

the lipopolysaccharide (LPS) of S. minnesota R595. Micro-

biologica 1995;18:325–30.

[14] Minnick MF. Identification of outer membrane proteins of

Bartonella bacilliformis. Infect Immun 1994;62:2644–8.

[15] Mergenhagen SE, Hampp EG, Scherp HW. Preparation and

endotoxic activity of endotoxins from oral bacteria. J Infect

Dis 1961;108:304–10.

[16] Maurin M, Birtless R, Raoult D. Current knowledge of Bar-

G. Matera et al. / International Immunopharmacology 3 (2003) 853–864864

tonella species. Eur J Clin Microbiol Infect Dis 1997;16:

487–506.

[17] StSwierzko A, Kirikae T, Kirikae F, et al. Biological activities

of lipopolysaccharides of Proteus spp. and their interactions

with polymyxin B and an 18-kD cationic antimicrobial protein

(CAP-18)-derived peptide. J Med Microbiol 2000;49:127–38.

[18] Hurley JC. Endotoxemia: methods of detection and clinical

correlates. Clin Microbiol Rev 1995;8:268–92.

[19] Liberto MC, Matera G, Lamberti A, Stella A, Pollio A, Pa-

terno R, et al. Bartonella quintana effects on endothelial

(ECV304) apoptosis and expression of inflammatory media-

tors. 100th General Meeting of American Society for Micro-

biology, May 21–25. Los Angeles, California; 2000.

[20] Koch AE, Polverini PJ, Kunkel SL. Interleukin-8 as a macro-

phage-derived mediator of angiogenesis. Science 1992;258:

1798–801.

[21] Puren AJ, Razeghi P, Fantuzzi G, Dinarello CA. Interleukin-

18 enhances lipopolysaccharide-induced interferon-g produc-

tion in human whole blood cultures. J Infect Dis 1998;178:

1830–4.

[22] Mastronardi CA, Yu WH, McCann SM. Comparison of the

effects of anesthesia and stress on release of tumor necrosis

factor-alpha, leptin, and nitric oxide in adult male rats. Exp

Biol Medicine (Maywood) 2001;226:296–300.

[23] Iidia M, Watanabe K, Tsurufuji M, Takaishi K, Iisuka Y,

Tsurufuji S. Level of neutrophil chemotactic factor CINC/

GRO a member of the interleukin-8 family, associated with

lipopolysaccharide-induced inflammation in rats. Infect Im-

mun 1992;60:1268–72.

[24] Ulich TR, Howard SC, Remick DG, et al. Intratracheal admin-

istration of endotoxin and cytokine: VI. Antiserum to CINC

inhibits acute inflammation. Am J Physiol 1995;268:L245–50.

[25] Zwaveling JH, Maring JK, Moshage H, et al. Role of nitric

oxide in recombinant tumor necrosis factor-alpha-induced cir-

culatory shock: a study in patients treated for cancer with

isolated limb perfusion. Crit Care Med 1996;24:1806–10.

[26] Heidemann SM, Sarnaik AP. The role of tumor necrosis factor

in endotoxic shock. Ann Clin Lab Sci 1996;26:60–3.

[27] Wang LZ, Su JY, Lu CY, Zhou BH, Ma DL. Effects of re-

combinant human endothelial-derived interleukin-8 on hem-

orrhagic shock in rats. Zhongguo Yao li Xue bao 1997;18:

434–6.

[28] Iwase M, Yokota M, Kitaichi K, et al. Cardiac functional and

structural alterations induced by endotoxin in rats: importance

of platelet-activating factor. Crit Care Med 2001;29:609–17.

[29] Armstrong J, Tempel GE, Cook JA, Wise WC, Halushka PV.

The effects of alpha adrenergic blockade on arachidonic acid

metabolism and shock sequelae in endotoxemia. Circ Shock

1986;20:151–9.

[30] Altenburg SP, Martins MA, Silva AR, Cordeiro RS, Castro-

Faria-Neto HC. LPS-induced blood neutrophilia is inhibited

by alpha1-adrenoceptor antagonists: a role for catecholamines.

J Leukoc Biol 1997;61:689–94.

[31] Ulich TR, del Castillo J, Ni RX, Bikhazi N, Calvin L. Mech-

anisms of tumor necrosis factor alpha-induced lymphopenia,

neutropenia, and biphasic neutrophilia: a study of lymphocyte

recirculation and hematologic interactions of TNFa with en-

dogenous mediators of leukocyte trafficking. J Leukoc Biol

1989;45:155–67.

[32] Musso T, Badolato R, Ravarino D, et al. Interaction of Barto-

nella henselae with the murine macrophage cell line J774: in-

fection and proinflammatory response. Infect Immun 2001;69:

5974–80.

[33] Power C, Fanning N, Redmond HP. Cellular apoptosis and

organ injury in sepsis: a review. Shock 2002;18:197–211.

[34] Schmidt H, Secchi A, Wellmann R, Bach A, Bohrer H, Geb-

hard MM, et al. Effect of endotoxemia on intestinal villus

microcirculation in rats. J Surg Res 1996;61:521–6.

[35] Hofman V, Ricci V, Mograbi B, et al. Helicobacter pylori

lipopolysaccharide hinders polymorphonuclear leukocyte

apoptosis. Lab Invest 2001;81:375–84.

[36] Preshaw PM, Schifferle RE, Walters JD. Porphyromonas gin-

givalis lipopolysaccharide delays human polymorphonuclear

leukocyte apoptosis in vitro. J Periodontal Res 1999;34:

197–202.

[37] Ren Y, Stuart L, Lindberg FP, Rosenkranz AR, Chen Y, Maya-

das TN, Savill J. Nonphlogistic clearance of late apoptotic

neutrophils by macrophages: efficient phagocytosis independ-

ent of beta 2 integrins. J Immunol 2001;166:4743–50.