Mucosal Lymphoid Infiltrate Dominates Colonic Pathological Changes in Murine Experimental...

13
136 JID 2005:192 (1 July) Martino et al. MAJOR ARTICLE Mucosal Lymphoid Infiltrate Dominates Colonic Pathological Changes in Murine Experimental Shigellosis Maria Celeste Martino, 1 Giacomo Rossi, 2 Irene Martini, 1 Ivan Tattoli, 1 Damiana Chiavolini, 3 Armelle Phalipon, 4 Philippe J. Sansonetti, 4 and Maria Lina Bernardini 1 1 Dipartimento di Biologia Cellulare e dello Sviluppo, Universita ` La Sapienza, Rome, 2 Facolta ` di Medicina Veterinaria, Universita ` di Camerino, Matelica, and 3 Dipartimento di Microbiologia, Universita ` di Siena, Siena, Italy; 4 Pathoge ´nie Microbienne Mole ´culaire, Institut Pasteur, Paris, France Background. Shigella species are invasive human pathogens that cause acute rectocolitis by triggering a dysregulated inflammatory reaction in the colonic and rectal mucosa. Because mice are naturally resistant to shigellosis, there is no mouse model that mimics human disease. We explore the susceptibility of intestinal flora– depleted mice to shigellosis after intragastric infection with Shigella strains. Methods. Mice given 5 g/L streptomycin as a beverage were infected intragastrically with cfu of either 8 1 10 invasive or noninvasive Shigella strains. Results. We found that invasive Shigella strains persist up to 30 days in feces, whereas the persistence of noninvasive Shigella strains was reduced. Colonization primarily involves the colon and the cecum and, to a lesser extent, the ileum. The hallmark of inflammation in the intestinal tissue is a dramatic expansion of the lymphoid follicles, in which a high apoptotic index is recorded. Conclusions. We provide a murine model in which shigellae are able to reach their natural tissue target: the colon. Moreover, the absence of polymorphonuclear leukocyte recruitment and of epithelial cell lesions reveal some aspects of shigellosis that are usually hidden by the prevalence of this cell population. This novel model may contribute to the identification of new targets for vaccines and therapies. Shigellosis is an acute rectocolitis [1] provoked by the ingestion of as few as 10–100 Shigella colony-forming units [2], and 1150 million cases occur every year [3]. Shigellae invade the colonic mucosa by translocating through M cells of the follicle-associated epithelium that covers lymphoid nodules dispersed on the colo- rectal surface [4]. From there, shigellae penetrate in- testinal epithelial cells (IECs), in which they inject ef- fector proteins via a type III secretion apparatus [5]. These events initiate inflammation through the apop- totic killing of infected macrophages [6], due to the activation of caspase-1, resulting in the parallel release Received 5 October 2004; accepted 11 February 2005; electronically published 31 May 2005. Financial support: European Commission (grant QLK2-1999-00938); Italian Ministero dell’Istruzione, Universita ` e Ricerca (grant PRIN2002). Reprints or correspondence: Dr. Maria Lina Bernardini, Dipartimento di Biologia Cellulare e dello Sviluppo, Universita ` La Sapienza, Via dei Sardi 70, 00185 Rome, Italy ([email protected]). The Journal of Infectious Diseases 2005; 192:136–48 2005 by the Infectious Diseases Society of America. All rights reserved. 0022-1899/2005/19201-0019$15.00 of mature interleukin (IL)–1b [7] and, possibly, a de- creased expression of IL-1 receptor antagonist [8]. In- vaded IECs produce proinflammatory cytokines and chemokines, particularly IL-8, a potent activator of poly- morphonuclear leukocytes (PMNLs) [9, 10], after in- tracellular sensing of a peptidoglycan motif [11] via nucleotide-binding oligomerization domain protein 1 [12]; inflammation thus increases in intensity and ex- tends according to the progression of shigellae in the epithelium. This is an essential feature, because mucosal inflammation disrupts the epithelial barrier and facil- itates further invasion of the epithelium by shigellae [13–15]. PMNLs eventually eliminate invasive shigellae [16] by blocking their escape from the phagocytic vac- uole. In summary, eradication of shigellae at the innate stage of the immune response occurs at the expense of massive inflammatory destruction of the epithelium. Although mice are the ideal animals to use in the ex- perimental study of human infectious diseases, because of their natural resistance to intestinal infection by Shi- gella species, to date, alternative animal systems have by guest on May 23, 2016 http://jid.oxfordjournals.org/ Downloaded from

Transcript of Mucosal Lymphoid Infiltrate Dominates Colonic Pathological Changes in Murine Experimental...

136 • JID 2005:192 (1 July) • Martino et al.

M A J O R A R T I C L E

Mucosal Lymphoid Infiltrate DominatesColonic Pathological Changes in MurineExperimental Shigellosis

Maria Celeste Martino,1 Giacomo Rossi,2 Irene Martini,1 Ivan Tattoli,1 Damiana Chiavolini,3

Armelle Phalipon,4 Philippe J. Sansonetti,4 and Maria Lina Bernardini1

1Dipartimento di Biologia Cellulare e dello Sviluppo, Universita La Sapienza, Rome, 2Facolta di Medicina Veterinaria,Universita di Camerino, Matelica, and 3Dipartimento di Microbiologia, Universita di Siena, Siena, Italy;4Pathogenie Microbienne Moleculaire, Institut Pasteur, Paris, France

Background. Shigella species are invasive human pathogens that cause acute rectocolitis by triggering adysregulated inflammatory reaction in the colonic and rectal mucosa. Because mice are naturally resistant toshigellosis, there is no mouse model that mimics human disease. We explore the susceptibility of intestinal flora–depleted mice to shigellosis after intragastric infection with Shigella strains.

Methods. Mice given 5 g/L streptomycin as a beverage were infected intragastrically with cfu of either81 � 10invasive or noninvasive Shigella strains.

Results. We found that invasive Shigella strains persist up to 30 days in feces, whereas the persistence ofnoninvasive Shigella strains was reduced. Colonization primarily involves the colon and the cecum and, to a lesserextent, the ileum. The hallmark of inflammation in the intestinal tissue is a dramatic expansion of the lymphoidfollicles, in which a high apoptotic index is recorded.

Conclusions. We provide a murine model in which shigellae are able to reach their natural tissue target: thecolon. Moreover, the absence of polymorphonuclear leukocyte recruitment and of epithelial cell lesions revealsome aspects of shigellosis that are usually hidden by the prevalence of this cell population. This novel model maycontribute to the identification of new targets for vaccines and therapies.

Shigellosis is an acute rectocolitis [1] provoked by the

ingestion of as few as 10–100 Shigella colony-forming

units [2], and 1150 million cases occur every year [3].

Shigellae invade the colonic mucosa by translocating

through M cells of the follicle-associated epithelium

that covers lymphoid nodules dispersed on the colo-

rectal surface [4]. From there, shigellae penetrate in-

testinal epithelial cells (IECs), in which they inject ef-

fector proteins via a type III secretion apparatus [5].

These events initiate inflammation through the apop-

totic killing of infected macrophages [6], due to the

activation of caspase-1, resulting in the parallel release

Received 5 October 2004; accepted 11 February 2005; electronically published31 May 2005.

Financial support: European Commission (grant QLK2-1999-00938); Italian Ministerodell’Istruzione, Universita e Ricerca (grant PRIN2002).

Reprints or correspondence: Dr. Maria Lina Bernardini, Dipartimento di BiologiaCellulare e dello Sviluppo, Universita La Sapienza, Via dei Sardi 70, 00185 Rome,Italy ([email protected]).

The Journal of Infectious Diseases 2005; 192:136–48� 2005 by the Infectious Diseases Society of America. All rights reserved.0022-1899/2005/19201-0019$15.00

of mature interleukin (IL)–1b [7] and, possibly, a de-

creased expression of IL-1 receptor antagonist [8]. In-

vaded IECs produce proinflammatory cytokines and

chemokines, particularly IL-8, a potent activator of poly-

morphonuclear leukocytes (PMNLs) [9, 10], after in-

tracellular sensing of a peptidoglycan motif [11] via

nucleotide-binding oligomerization domain protein 1

[12]; inflammation thus increases in intensity and ex-

tends according to the progression of shigellae in the

epithelium. This is an essential feature, because mucosal

inflammation disrupts the epithelial barrier and facil-

itates further invasion of the epithelium by shigellae

[13–15]. PMNLs eventually eliminate invasive shigellae

[16] by blocking their escape from the phagocytic vac-

uole. In summary, eradication of shigellae at the innate

stage of the immune response occurs at the expense of

massive inflammatory destruction of the epithelium.

Although mice are the ideal animals to use in the ex-

perimental study of human infectious diseases, because

of their natural resistance to intestinal infection by Shi-

gella species, to date, alternative animal systems have

by guest on May 23, 2016

http://jid.oxfordjournals.org/D

ownloaded from

A Murine Model to Study Human Shigellosis • JID 2005:192 (1 July) • 137

Figure 1. Bacterial persistence and distribution throughout the organs.A, Recovery of Shigella flexneri 5a strain M90T from cecum, colon, ileum,and liver. Mice were infected with cfu of M90T and euthanized81 � 10at day 1, 3, 5, 6, or 7 ( mice for each time point), and relevantn p 3organs were removed. Data are the geometric (error bars)mean � SEof 3 experiments ( mice for each time point). B, Recovery of M90Tn p 9and strain BS176 colony-forming units from feces of infected mice. Mice( for each strain) were infected with cfu of M90T or8n p 20 1 � 10BS176. Data shown are of 1 representative experiment in which colony-forming unit values were calculated in fecal pellets of 3 groups of 5randomly chosen mice. Results obtained in 3 identical experiments hadan SD !20%. The difference between BS176 and M90T colony-formingunits per fecal pellet from 20 mice in both groups at days 9 (P p .0091)and 15 after infection ( , Mann-Whitney U test) was statisticallyP ! .0001significant. The value of cfu/g of feces (dotted line) was con-31 � 10sidered to be the limit of bacterial persistence.

been used [17–20]. It is unclear why mice are resistant to de-

veloping shigellosis. Three major hypotheses have been put

forth to account for this species specificity: (1) Shigella species

cannot compete in the mouse colonic environment, which con-

sists of commensal flora, mucins, and associated antimicrobial

products; (2) mice lack a receptor that allows binding to and

invasion of IECs; and (3) the murine epithelium is unable to

relay and amplify the inflammatory reaction initiated by solitary

nodules. In an attempt to understand the respective relevance

of these hypotheses in the pathogenesis of the natural disease,

we have developed a novel experimental mouse model. In this

model, mice are intragastrically (ig) infected with S. flexneri 5a

in the presence of the antibiotic streptomycin (Sm). Under

these conditions, the murine colonic mucosa senses the pres-

ence of shigellae and mounts an inflammatory reaction char-

acterized by a paucity of recruitment of PMNLs and destruction

of the epithelial lining.

MATERIALS AND METHODS

Bacteria, growth conditions, and genetic procedures. The S.

flexneri 5a strains used were the Sm-resistant (SmR) fully vir-

ulent wild-type strain M90T [21] and its avirulent Sm-suscep-

tible (SmS) variant BS176, which lacks the virulence plasmid

pWR100 [22, 23]. M90T SmR is a spontaneous SmR variant

[21] of the original clinical isolate M90T [22]. BS176 is the

original plasmidless variant of M90T [22]. To obtain BS176

SmR, BS176 was the receiver in an experiment of generalized

P1-mediated transduction that was performed as described else-

where [24] following Miller’s protocol [25] and using M90T

SmR as the donor. M90T SmR and BS176 SmR were able to

survive in media containing up to 5 mg/mL of Sm. Shigella

strains were grown on trypticase soy broth (Becton Dickinson)

or Hektoen enteric agar (HEA) (Oxoid). Sm (Sigma) was used

at a concentration of 100 mg/mL.

Infection of mice. A minimum of 20 five-week-old female

BALB/c mice (Charles River) were used per experimental group

and received a 5 g/L Sm solution as a beverage for 48 h. After

a 6-h starvation period, mice were infected ig by a polyethylene

feeding tube with cfu of S. flexneri 5a strain M90T SmR81 � 10

or strain BS176 SmR in 100 mL of sodium bicarbonate (1.4%).

The antibiotic treatment was maintained until the end of the

experiment. A group of uninfected mice, which was otherwise

treated as above, was used as a control.

Recovery of shigellae from feces and organs of infected mice.

Feces were collected daily and examined for the presence of

shigellae. Feces from 5 mice (0.2–0.3 g) were individually col-

lected and homogenized in a sample tube containing 5 mL of

saline solution (0.9% NaCl), serially diluted, and plated on plates

with HEA and Sm. Bacterial counts were reported as colony-

forming units per gram of feces. Bacterial persistence was con-

sidered to have ended when counts dropped to !1�103 cfu/g

of feces. When this threshold was reached, the fecal pellets were

collected, analyzed by individual mouse, and reported as colony-

forming units per gram of feces as above. The minimum de-

tectable value was 5 cfu/fecal pellet.

Relevant organs from mice euthanized at the desired time

points were removed and prepared for histopathological ex-

amination, reverse-transcription polymerase chain reaction

(RT-PCR) analysis, and counts of viable shigellae. For bacterial

by guest on May 23, 2016

http://jid.oxfordjournals.org/D

ownloaded from

138 • JID 2005:192 (1 July) • Martino et al.

Table 1. Histopathological scores for lesions in intestinal tis-sues from mice infected with Shigella flexneri 5a strain M90T orBS176 and from uninfected control mice.

Group,hours afterinfection,organ

Score for

Transcytosisa/epithelialdamageb

Interstitial areas

NeutrophilscMononuclear

cellsdLymphoidfolliclese

M90T24

Colon 1 2 1 1Cecum 0 0 0 1

72Colon 4 1 3 3Cecum 2 0 1 2

120Colon 3 1 4 3Cecum 2 0 2 2

BS17624

Colon 0 0 0 0Cecum 0 0 0 0

72Colon 1 0 1 1Cecum 0 0 0 0

120Colon 0 0 2 1Cecum 0 0 1 0

Control24

Colon 0 0 0 0Cecum 0 0 0 0

72Colon 0 0 1 0Cecum 0 0 0 0

120Colon 0 0 0 1Cecum 0 0 0 0

NOTE. A total of 6 mice were used in each group. for epithelialP p .0022damage in colon from M90T-infected mice vs. BS176-infected mice at 72 and120 h after infection and for lymphoid follicle activation of colon of M90T-infected mice vs. BS176-infected mice at 72 and 120 h after infection (Mann-Whitney U test).

a Scoring for transcytosis across the epithelium: 0, !10 cells; 1, 10–19 cells;2, 20–49 cells; 3, 50–100 cells; 4, 1100 cells.

b Scoring for epithelial damage: 0, absence of cellular damage; 1, presenceof cytoplasmic vacuolization in some epithelial cells; 2, diffuse cellular vacu-olization; 3, diffuse vacuolization and necrosis/detachment of singular/sporadicepithelial cells; 4, diffuse and severe cellular vacuolization with areas of cellulardetachment (groups of cells with microerosive status).

c Scoring for no. of neutrophils per high-power field: 0, �4 cells; 1, 5–19cells; 2, 20–49 cells; 3, 50–100 cells; 4, 1100 cells.

d Scoring for no. of mononuclear cells per high-power field: 0, �49 cells;1, 50–100 cells; 2, 101–200 cells; 3, 201–600 cells; 4, 1600 cells.

e The mean areas of 10 different randomly selected follicles in gut-associ-ated lymphoid tissues for each sample was measured. Scoring for mean areaof activated lymphoid follicles: 0, none; 1, up to 0.17 mm2; 2, up to 0.32 mm2;3, up to 2.07 mm2.

counts, tissue sections were placed in ice-cold saline solution

(5 mL for liver, 4 mL for colon and ileum, and 2 mL for spleen

and cecum) and homogenized, and serial dilutions were plated

on plates with HEA and Sm. Bacterial counts were reported as

colony-forming units per organ. The minimal detectable values

were 50 cfu/liver, 40 cfu/ileum, 40 cfu/colon, 20 cfu/spleen, and

20 cfu/cecum.

Histopathological analysis. Samples for histopathological

and immunohistochemical (IH) analysis were taken from the

same areas of the organ of interest from 6 randomly chosen

mice in each experimental group. For histopathological ex-

amination, 3-mm sections were deparaffinized, rehydrated, and

stained with hematoxylin-eosin. The following monoclonal an-

tibodies (MAbs) were used: mouse anti–S. flexneri 5a lipopoly-

saccharide (LPS; dimeric IgA; 6 mg/mL) [18], rat anti–murine

CD3+ T lymphocytes, rat anti–murine CD8+ T lymphocytes

(KT15), rat anti–murine macrophages (F4/80 antigen), rat anti–

dendritic/interdigitating cells (Midc-8) (all from Serotec), rat

anti–murine CD5+ B lymphocytes, mouse anti–murine CD21+

B lymphocytes, rat anti–murine CD32+CD16+ NK cells, and

rabbit polyclonal IgG anti–mouse IL-2Ra (CD25) (all from Santa

Cruz Biotechnology). For IH analysis, tissue sections were

treated as described elsewhere [26]. Binding of antibodies was

shown by use of avidin-biotin immunoperoxidase or avidin-

biotin alkaline phosphatase (both from Vector Laboratories)

with biotin-conjugated rabbit anti–rat IgG (Vector Laborato-

ries) and biotinylated goat anti–mouse IgG (AO433; DAKO)

(both at 1:200), respectively. The reactions were visualized with

3-1-diaminobenzidine (DAB; Sigma), Very Intense Purple, or

Vector Blue and Vector Red (all from Vector Laboratories). For

multiple stainings, the first chromogen used was DAB, with or

without nickel addition, the second was VIP, the third was

Vector Blue, and the last was Vector Red, with Meyer’s he-

matoxylin used as a nuclear counterstain.

The number of inflammatory cells and of lymphoid aggre-

gates was assessed at �400 and �100 magnification, respec-

tively, was scored as described by Dixon et al. [27], and was

customized for murine samples as described by Happonen et

al. [28]. The proapoptotic effect induced by S. flexneri was

highlighted through TUNEL colorimetric staining (DeadEnd;

Promega), in accordance with the manufacturer’s instructions.

RNA extraction and RT-PCR. Total RNA from homoge-

nized ileum or colon of 3 mice per experimental group was

extracted using Trizol solution (Invitrogen), in accordance with

the manufacturer’s instructions. RT of total RNA (1 mg) and

cDNA PCR was performed using the Super-Script One-Step

RT-PCR with Platinum Taq (Invitrogen), in accordance with

the manufacturer’s instructions. PCR conditions, primers, and

densitometry are described elsewhere [24].

ELISA. Peripheral blood was obtained from the caudal vein

at 24, 48, 72, and 96 h after infection. IL-6 levels in serum were

determined by solid-phase ELISA (R&D Systems). The absor-

bance was measured at 450 nm, and concentrations were de-

termined by interpolation of a standard calibration curve.

Analysis of data. The exact Mann-Whitney U test was used

for statistical analysis of bacterial counts in feces and the cal-

by guest on May 23, 2016

http://jid.oxfordjournals.org/D

ownloaded from

A Murine Model to Study Human Shigellosis • JID 2005:192 (1 July) • 139

Figure 2. Immunolocalization of Shigella flexneri 5a lipopolysaccharide (LPS) in colonic mucosa of mice infected with strain M90T (A, C, E) or BS176(B, D, F) at 24, 72, and 120 h after infection. Sections of colonic tissue were immunostained with an IgA mouse anti–S. flexneri 5a LPS monoclonalantibody. Black arrows indicate LPS staining. Open arrows in E point to a large amount of LPS diffused in the lymphoid infiltrate. Bars, 50 mm.

culation of individual histopathological scores. was con-P ! .05

sidered to be statistically significant.

RESULTS

Establishment of optimal conditions for infection. The ex-

perimental conditions were determined on the basis of the

results of several experiments aimed at evaluating the role that

the resident flora play in establishing infection, the impact that

different doses of Sm have on colonization of S. flexneri, and

the distribution of microorganisms in the organs.

Mice were infected ig with cfu of S. flexneri 5a SmR91 � 10

wild-type strain M90T. Shigellae persisted up to 3 days in feces.

Bacterial counts in feces were cfu/g of feces at day 1,65 � 10

cfu/g of feces at day 2, and cfu/g of feces at3 27 � 10 1 � 10

day 3 after infection.

To assess how resident bacterial flora could affect coloni-

zation of S. flexneri, an antibiotic treatment was applied to

eliminate local flora. We found that 5 g/L Sm given as a beverage

48 h before infection was sufficient to abolish all detectable

aerobic flora from feces. M90T was administered at 3 doses

( , , and cfu), and the inoculum of 1 �7 8 91 � 10 1 � 10 1 � 10

108 cfu was selected for use in the final experiments, because

it allowed for an improved bacterial persistence in the feces,

compared with that of the other inocula. The following protocol

was applied in the final experiments: at 48 h before infection,

a minimum of 20 mice per experimental group were given Sm

(5 g/L) as a beverage, and the treatment was maintained until

shigellae were cleared from the feces. After a 6-h starvation

period, mice were infected with cfu of S. flexneri 5a81 � 10

strain M90T or BS176. Feces were collected daily. At different

time points, depending on the experiment, at least 3 mice were

euthanized, and the relevant organs were removed.

From day 1, most shigellae were present in the colon, cecum,

ileum, and liver. This trend remained unvaried during the

course of 7 days of infection, with a significant increase in the

number of shigellae in the liver at day 7, compared with that

by guest on May 23, 2016

http://jid.oxfordjournals.org/D

ownloaded from

140

Figu

re3.

Hist

opat

holo

gica

lcha

ract

eriza

tion

ofco

loni

cep

ithel

ium

and

lam

ina

prop

ria(L

P).A

–C,H

emat

oxyl

in-e

osin

(HE)

stai

ning

ofco

loni

cse

ctio

nsfro

mm

ice

infe

cted

with

Shig

ella

flexn

eri5

ast

rain

M90

T(A

)or

BS17

6(B

)and

from

unin

fect

edco

ntro

lmic

e(C

)at

120

haf

ter

infe

ctio

n.In

A,ar

row

head

spo

int

totra

nscy

tosi

sof

poly

mor

phon

ucle

arle

ukoc

ytes

(PM

NLs

)thr

ough

out

the

epith

eliu

m.I

nB,

the

arro

whe

adin

dica

tes

aPM

NL

unde

rth

eLP

.D–F

,HE

stai

ning

ofch

orio

nse

ctio

nsfro

mm

ice

infe

cted

with

M90

T(D

)at

72h

afte

rin

fect

ion

orBS

176

(E)a

t12

0h

afte

rinf

ectio

nan

dfro

mun

infe

cted

cont

rolm

ice

(F).

G–I,

Imm

unoh

isto

chem

ical

char

acte

rizat

ion

ofth

ein

ters

titia

lmon

ocyt

ein

filtra

tein

sect

ions

ofth

eLP

from

mic

ein

fect

edw

ithM

90T

(G)o

rBS

176

(H)a

ndfro

mun

infe

cted

cont

rolm

ice

(I)at

120

haf

teri

nfec

tion.

Avid

in-b

iotin

imm

unop

erox

idas

ela

belin

gof

F4/8

0-po

sitiv

em

acro

phag

esim

mun

osta

ined

with

anIg

Am

ouse

anti–

S.fle

xner

i5a

LPS

mon

oclo

nala

ntib

ody

(bro

wn

and

arro

whe

ads)

and

CD3+

Tly

mph

ocyt

es(b

lue)

.A–C

,Bar

s,25

mm

.D–I

,Bar

s,50

mm

.

by guest on May 23, 2016

http://jid.oxfordjournals.org/D

ownloaded from

141

Figure 4. Immunohistochemical (IH) characterization of lymphoid follicles at 120 h after infection. A and G, Hematoxylin-eosin staining of thecolonic mucosa (CM) showing the dramatic expansion of the lymphoid follicles in mice infected with Shigella flexneri 5a strain M90T (A) with respectto those in mice infected with strain BS176 (G). B and H, IH labeling showing CD3+ T lymphocytes (brown) in the follicles of mice infected with M90T.C and I, IH labeling showing a subpopulation of CD3+CD25+ lymphocytes in the follicles of mice infected with BS176. D and L, IH staining showingCD32+CD16+ NK cells (red and arrowheads) and CD8+ T lymphocytes (brown and arrows) around the follicles in the CM of mice infected with M90T(D) or BS176 (L). E and M, IH staining showing CD5+ B lymphocytes (purple-violet), interspersed dendritic cells (black), CD21+ B lymphocytes (brown),CD3+ T lymphocytes (blue), and CD32+CD16+ NK cells (red) in lymphoid follicles of mice infected with M90T (E) or BS176 (M). F and N, Schematicrepresentation of E and M, respectively. Bars, 50 mm.

by guest on May 23, 2016

http://jid.oxfordjournals.org/D

ownloaded from

142 • JID 2005:192 (1 July) • Martino et al.

Figure 5. Apoptosis assessment in the colonic tissue. TUNEL analysis of sections of the colonic mucosa from mice infected with Shigella flexneri5a strain M90T (A) or BS176 (B) and from uninfected control mice (C). Open arrows point to the stained cells. Bars, 50 mm.

Table 2. Apoptotic index in gut-associated lymphoid tissue (GALT) follicles and the lamina propria (LP).

Group

Apoptotic index, mean � SE, %

In GALT folliclesa In the LPb

24 h 72 h 120 h 24 h 72 h 120 h

M90T 0.48 � 0.12 2.20 � 0.65 5.44 � 0.37c 6.60 � 0.32 10.96 � 0.23 9.64 � 0.17BS176 0.53 � 0.21 0.72 � 0.19 0.89 � 0.23 8.87 � 0.15 7.42 � 0.18 8.58 � 0.31Control 0.56 � 0.11 0.41 � 10.09 0.53 � 0.17 6.28 � 0.19 6.44 � 0.21 8.56 � 0.22

a The apoptotic index was calculated as the percentage of TUNEL-positive cells (i.e., the number of TUNEL-positive mono-nuclear cells divided by the total number of lymphoid cells in GALT follicles, � 100) at 24, 72, and 120 hours after infection. A

of cells was counted for each sample.mean � SE 2203� 212b The apoptotic index was calculated as the percentage of TUNEL-positive cells (i.e., the number of TUNEL-positive mono-

nuclear cells divided by the total number of lymphoid cells in diffuse infiltrate in the LP, � 100) at 24, 72, and 120 hours afterinfection. A of cells was counted for each sample.mean � SE 656� 97

c for values at 120 h vs. values at 24 h after infection.P ! .0001

at day 1 (figure 1A). Histopathological analysis confirmed rel-

evant morphological changes at early time points in the liver,

colon, and, to a lesser extent, cecum (data not shown).

Bacterial persistence. To assess whether invasiveness plays

a role in bacterial persistence and distribution, we compared

ig infection with M90T and infection with its noninvasive var-

iant SmS strain BS176. M90T persisted in feces for a mean�

SD of days, whereas BS176 persisted for a mean30.2 � 2.5

�SD of days (figure 1B). After 7 days, the number19.0 � 2.2

of BS176 colony-forming units in feces rapidly decreased,

whereas the number of M90T colony-forming units remained

high. Similarly, during the first 7 days of infection, bacterial

counts in the colon and cecum did not vary for either strain,

with the number of colony-forming units of BS176 being ∼1

log lower than the number of colony-forming units of M90T

in these organs (data not shown).

Histopathological features. Histopathological analysis was

performed on tissue sections removed from the colon and ce-

cum of mice euthanized at 24, 72, and 120 h after infection.

The severity of tissue inflammation was evaluated by recording

qualitative and quantitative alterations of the intestinal mucosa

as reported elsewhere [27, 28]. These alterations were quantified

and recorded as the mean of scores calculated for each lesion,

and these scores are reported in table 1.

In the intestinal tissue of M90T-infected mice, we observed a

relative paucity of PMNLs, compared with the massive numbers

of mononuclear cells and the dramatic expansion of the lym-

phoid follicles of gut-associated lymphoid tissues (GALT). In the

intestinal tissue of BS176-infected mice, only a moderate number

of mononuclear cells were found at 120 h after infection.

Localization of the shigellae in the tissues. The distribu-

tion of shigellae in the gastrointestinal tract was highlighted by

immunostaining for S. flexneri LPS. The cecal and colonic con-

tents of M90T-infected mice were strongly stained for S. flexneri

LPS. However, although a scant positive result for S. flexneri

LPS was found in the cecal mucosa, a stronger presence was

detected in the colonic mucosa (CM). In the CM, at 24 h after

infection, LPS was mainly found in the cytoplasm of macro-

phages located in the central portion of villi beyond the epi-

thelial layer (figure 2A). At 72 h after infection, LPS was de-

tected in the deep chorion (figure 2C), whereas, at 120 h after

infection, a large amount of LPS was found diffused within the

CM and in scattered lymphoid cells of the activated follicles

(figure 2E). In BS176-infected mice, small quantities of LPS

by guest on May 23, 2016

http://jid.oxfordjournals.org/D

ownloaded from

A Murine Model to Study Human Shigellosis • JID 2005:192 (1 July) • 143

Figure 6. Assessment of the expression of proinflammatory factors and production of interleukin (IL)–6. A, Agarose gel showing reverse-transcriptionpolymerase chain reaction (RT-PCR) and densitometry of products of RNA extracted from colon and ileum. For each time point, 3 mice were euthanized,and mRNA was extracted from the intestinal tissue. The RT-PCR analysis was performed using primers for b-actin, IL-1b, IL-6, IL-12, interferon (IFN)–g, tumor necrosis factor (TNF)–a, and inducible nitric oxide synthase (iNOS). In uninfected tissues, no expression of mRNA for these cytokines wasobserved (data not shown). Similar results were obtained in 3 additional identical experiments. Results obtained in 3 identical experiments had anSD within 18% of the values of the arbitrary units shown here. B, Serum IL-6 levels in mice infected with M90T or BS176 at 24, 48, 72, and 96 hafter infection. Data are the geometric (error bars; mice) for each time point.mean � SE n p 3

were detected in colonic and cecal contents. At 24, 72, and 120

h after infection, LPS was immunostained in mucosal mac-

rophages of the CM (figure 2B, 2D, and 2F).

Histopathological features of the epithelium and the struc-

ture and cell populations of the lamina propria (LP). In

M90T-infected mice, at 72 h after infection, several mononu-

clear leukocytes and PMNLs were interspersed exclusively in

the epithelial cell layer of the CM. At 120 h after infection,

transcytosis of PMNLs across the vacuolated colonic epithelium

could still be observed (figure 3A). However, in contrast to

what was described for ig infection of rabbits [29] and guinea

pigs [30], no destruction of the epithelial lining was detected.

In BS176-infected mice, at 120 h after infection, minor mor-

phological changes characterized by vacuolization of a few ep-

by guest on May 23, 2016

http://jid.oxfordjournals.org/D

ownloaded from

144 • JID 2005:192 (1 July) • Martino et al.

Figure 7. Immunohistochemical (IH) analysis of tissue sections of liver. Hematoxylin-eosin staining of tissue sections of liver from mice infectedwith Shigella flexneri 5a strain M90T (A) or BS176 (B) and from uninfected control mice (C) at 72 h after infection are shown. IH staining shows S.flexneri lipopolysaccharide in the liver from mice infected with M90T (D) or BS176 (E) and from uninfected control mice (F) at 120 h after infection.In A, open arrows indicate a microgranuloma resulting in mononuclear cell aggregates with few neutrophils. In B, open arrows point to small aggregatesof lymphocytes. Bars, 50 mm.

Table 3. Histopathological scores for liver frommice infected with M90T.

Hours afterinfection

Score

Microgranulomasa Degenerationb

24 0 072 1 0120 2 2

NOTE. Liver from both BS176-infected mice and unin-fected control mice had scores of 0 in both parameters ana-lyzed at all time points.

a Microgranulomas were defined as well-circumscribed cellaggregates composed of �5 mononuclear phagocytes. Scor-ing for mean area of granuloma extension: 0, none; 1, up to0.008 mm2; 2, up to 0.042 mm2; 3, up to 0.4 mm2.

b Scoring for no. of hepatocytes showing degeneration (i.e.,cloudy swelling and/or vacuolar degeneration) per high-powerfield: 0, none; 1–19 cells; 2, 20–49 cells; 3, 50–200 cells.

ithelial cells in the absence of transcytosis of inflammatory cells

were observed (figure 3B).

At 24 h after infection, the LP of M90T-infected mice showed

an inflammatory infiltrate characterized by a few PMNLs and a

significant number of mononuclear cells, consisting of CD3+ T

lymphocytes, a few plasma cells, and macrophages (data not

shown). At 72 h after infection, the mononuclear cell–infiltrating

population increased (figure 3D), and a small number of PMNLs

persisted. At 120 h after infection, the colonic villi were enlarged,

and the LP appeared infiltrated by a large number of CD3+ T

lymphocytes (histopathological score, 4) and scattered macro-

phages that were positively immunostained by mouse anti–S.

flexneri 5a LPS (figure 3G). In BS176-infected mice, at 120 h

after infection, the CM showed no significant infiltrate within

the LP (figure 3E). Only a few CD3+ T lymphocyte infiltrates

(histopathological score, 1) in association with a moderate pres-

ence (histopathological score, 2) of interspersed LPS-positive

macrophages (figure 3H) were observed. No alteration in the

epithelium (figure 3C), no cellular infiltrate (figure 3F), and a

few or no CD3+ T lymphocytes were observed in the corre-

sponding tissue from the uninfected control mice (figure 3I).

Characterization of the expanded follicles in the CM. The

main sign of the presence of shigellae in the colonic tissue was

the dramatic expansion of the lymphoid follicles. At 24 h after

infection, we observed the presence of lymphoid aggregates in

the CM of M90T-infected mice. The size of these well-structured

follicles reached a peak at 120 h after infection (figure 4A). Their

mantle area was covered by CD3+ T lymphocytes, primarily rep-

resented by CD3+CD25+ lymphocytes (figure 4B and 4C) and by

many CD3+CD8+ lymphocytes (figure 4D) interspersed with a

large number of CD32+CD16+ NK cells. The central area con-

tained large arachnid-shaped dendritic cells (DCs) interspersed

in a large population of CD5+ B lymphocytes and a few CD3+

T lymphocytes. An area of CD21+ B lymphocytes was immu-

nostained around this central portion. These well-structured fol-

licles, localized in the deep chorion, were consistently accom-

panied by CD32+CD16+ NK cells (figure 4E and 4F) and by a

by guest on May 23, 2016

http://jid.oxfordjournals.org/D

ownloaded from

A Murine Model to Study Human Shigellosis • JID 2005:192 (1 July) • 145

Figure 8. Schematic representation of different steps of Shigella flexneri colonization of the gut mucosa and the liver. IFN, interferon; IL, interleukin;LPS, lipopolysaccharide; TNF, tumor necrosis factor.

mixed population of CD3+ T lymphocytes that formed a diffuse,

superficial, and subepithelial infiltrate, as described above. In the

CM of BS176-infected mice, there were small to moderate lym-

phoid aggregates (figure 4G) not organized as a structured follicle.

As was observed in M90T-infected mice, at 120 h after infection,

a certain number of CD3+CD25+ lymphocytes covered the mantle

area of the lymphoid aggregates (figure 4H and 4I). At this time

point, CD3+CD8+ lymphocytes constituted the prevalent lym-

phocyte population, whereas few or no CD32+CD16+ NK cells

were detected (figure 4L). The lymphoid aggregates mainly re-

sulted in CD3+ T, CD21+ B, and CD5+ B lymphocytes (figure

4M and 4N).

Apoptosis of infected cells in the lymphoid follicles of the LM.

The amount and the localization of apoptotic cells in the CM

was assessed by a TUNEL analysis (figure 5 and table 2). In the

M90T-infected mice, from 24 to 120 h after infection, an in-

creasing number of TUNEL-positive cells was present in the

activated GALT (figure 5A). In contrast, in the BS176-infected

by guest on May 23, 2016

http://jid.oxfordjournals.org/D

ownloaded from

146 • JID 2005:192 (1 July) • Martino et al.

mice, a few TUNEL-positive lymphoid cells were detected (figure

5B). In uninfected control mice, a few TUNEL-positive cells were

observed in the rare solitary lymphoid follicles of the CM (figure

5C), whereas a few apoptotic cells were counted in unstimulated

lymphoid cells of the GALT (table 2).

Expression of proinflammatory cytokines in colonic tissue

at early time points of infection and serum IL-6 production.

We assessed the expression of IL-1b, IL-6, IL-12, interferon

(IFN)–g, tumor necrosis factor (TNF)–a, and inducible nitric

oxide synthase (iNOS) mRNAs by RT-PCR in the colon and

ileum of mice infected with M90T or BS176 and of uninfected

control mice at 24, 48, and 72 h after infection (figure 6A). In

the ileum, a considerable expression of mRNA for IFN-g, TNF-

a, and IL-12 was seen exclusively in M90T-infected mice only

at 72 h after infection.

At 24 h after infection, in M90T-infected mice, a measurable

expression of all these cytokines was observed in colonic tissues,

whereas, in BS176-infected mice, only low levels of IL-12, TNF-

a, and iNOS were detected. In M90T-infected mice, this trend

remained unchanged, with a progressive increase in the expres-

sion of mRNA for these cytokines—with the exception of mRNA

for IL-6, whose low expression slightly decreased from that ob-

served at 48 h after infection—until 72 h after infection. In

BS176-infected mice, at 48 h after infection, weak expression of

mRNA for IL-12, TNF-a, IFN-g, and iNOS was present, whereas,

at 72 h after infection, no or minimal expression of mRNA for

these cytokines was observed (figure 6A). In M90T-infected mice,

serum IL-6 production was high at 24 h after infection and then

disappeared at 72 h after infection (figure 6B).

Damage by shigellae and localization in the liver. In M90T-

infected mice, at 120 h after infection, the liver showed hepatic

cell death and well-defined hepatic microgranulomas with mono-

nuclear cell aggregates consisting of epithelioid and lymphoid

cells (figure 7A). LPS was detected within the cytoplasm of many

macrophages located in some central areas of these microgran-

ulomatous lesions and in Kupffer cells (figure 7D). In BS176-

infected mice, the liver did not show significant morphological

changes, and little or no infiltration of mononuclear cells was

observed (figure 7B). A positive response to LPS was detected

only in Kupffer cells, in the absence of significant lesions (figure

7E). The lesions are scored in table 3.

DISCUSSION

In this study, we addressed 3 key hypotheses about the inability

of mice to develop shigellosis. First, we examined whether the

endogenous flora could protect the intestinal tissue from Shi-

gella infection. We found that treatment of mice with Sm is a

prerequisite for colonization of the intestinal tract by shigellae,

which is in accordance with the role played by the intestinal

flora in preventing oral infection with various enteropathogens,

such as Salmonella species [31–33]. Then, we investigated the

ability of Shigella species to invade the intestinal tissue. As in

humans, the colon and eventually the cecum of the Sm-treated

mice were the main targets for the invasion of a wild-type

Shigella strain. However, in contrast to human shigellosis, no

lesions or only a few abscesses were observed in the mice, de-

spite the presence of a high number of shigellae. Finally, we

considered the hypothesis that the murine epithelium may be

unable to respond to intracellular shigellae by recruiting an

inflammatory infiltrate. Interestingly, the presence of a mono-

nuclear infiltrate and a dramatic expansion of lymphoid folli-

cles dispersed along the CM and the cecal mucosa were the

main features of the infected intestine, thus demonstrating that

the intestinal epithelium is able to sense shigellae and to mount

an inflammatory reaction. However, in contrast to the situation

in human shigellosis, and in accordance with the paucity of

lesions observed, a moderate number of PMNLs was observed

in the mice.

In the rabbit ligated ileal loop model of Shigella infection,

inhibition of IL-8 prevents the transmigration of PMNLs, thereby

reducing the number of tissue lesions at the expense of bacterial

dissemination in deeper layers of the CM [9]. In mice, the

delivery of recombinant human IL-8 together with invasive

Shigella strains through intraluminal instillation induces mu-

cosal lesions similar to those seen in human disease [20]. In

the present study, at 24 h after infection, shigellae were isolated

from the liver, in which they induced microgranulomas, as is

schematically shown in figure 8. These findings suggest that the

absence of a murine equivalent of human IL-8 [34] might con-

tribute to the prevention of the deleterious consequences of

inflammation and favor the septic dissemination of shigellae

into the liver. This confirms the results of our study showing

that, in mice infected intravenously, shigellae reach the liver,

where they induce microgranulomatous lesions [26]: again, the

absence of PMNLs is a feature of these lesions.

In the CM of Sm-treated mice, several cell populations par-

ticipate in the mononuclear infiltrate. CD32+CD16+ NK cells are

localized superficially and subepithelially in the CM of M90T-

infected mice. NK cells producing IFN-g play an essential role

in the resistance in mice after primary Shigella intranasal infection

[35], whereas, in humans, high levels of mRNA for IFN-g are

found in the LP only during late-phase shigellosis [36].

In M90T-infected mice, only mature DCs migrating to sec-

ondary lymphoid organs, such as the expanded follicles, were

observed. In M90T-infected mice, apoptosis of macrophages

and the subsequent release of shigellae or their material in the

expanded follicles might contribute to the maturation of DCs.

Mature DCs can produce IL-12 and stimulate NK cells again

[37], thus amplifying their activation and proliferation, the

release of cytokines such as TNF-a and IFN-g, and their bac-

teriolytic activity [38]. In accordance with this hypothesis, mRNA

for both IL-12 and IFN-g was detected only in colonic tissues

by guest on May 23, 2016

http://jid.oxfordjournals.org/D

ownloaded from

A Murine Model to Study Human Shigellosis • JID 2005:192 (1 July) • 147

of M90T-infected mice. In the rectal mucosa of humans with

shigellosis [39], a large expansion of CD3+ T lymphocytes is

observed. In mice, the large CD3+ T lymphocyte population

primarily consists of a subset of CD3+CD25+ lymphocytes.

CD4+CD25+ lymphocytes are a subset of naturally activated

lymphocytes, CD45RBlow, which control inflammatory re-

sponses to commensal bacteria [40] by acting as a negative

regulator of the immune response [41]. This cell population

might play an important role in controlling the magnitude and

the deleterious consequences of the murine intestinal tissue

response to shigellae. Whether CD3+CD25+ lymphocytes play

a role during natural shigellosis is still unknown. In the present

study, a large number of CD3+CD8+ lymphocytes was present

in the LP, particularly in the cortical areas of the expanded

follicles of M90T-infected mice. This cell population is the T

cell subset predominantly involved in the inflammation [39]

of human rectal mucosa. Finally, in the present study, CD5+ B

lymphocytes were diffusely distributed in the expanded central

areas of follicles. In mice, CD5+ B lymphocytes predominate

in the gut mucosa [42] and play a crucial role in the eradication

of pathogens by stimulating the innate immune system via the

classic pathway of complement activation [43].

In conclusion, it must be stressed that ig infection of Sm-

treated mice is, to our knowledge, the first model of shigellosis

in which the pathogens spontaneously interacted with the co-

lonic tissue that represents the natural target of their infection.

Furthermore, in accordance with the results of studies of in-

tragastric infection of the macaque Macaca mulatta, the dis-

tribution of bacteria in this tissue is essentially concentrated

around lymph nodes [4].

The study of infected murine colonic tissue in which PMNLs

are poorly recruited helps us to understand the other cell pop-

ulations involved in inflammation, their interactions, and their

interplay with bacteria. This may facilitate the search for new

vaccines and therapeutic approaches to shigellosis.

Acknowledgments

We thank Giorgina Levi and Stanley Stewart, for help in manuscriptpreparation, and Giuseppe Bertini, Piero Piccoli, and Giancarlo Cortese,for care of the mice.

References

1. Mathan VI, Mathan MM. Intestinal manifestations of invasive diar-rheas and their diagnosis. Rev Infect Dis 1991; 13(Suppl 4):S311–3.

2. DuPont HL, Levine MM, Hornick RB, Formal SB. Inoculum size inshigellosis and implications for expected mode of transmission. J InfectDis 1989; 159:1126–8.

3. Kotloff KL, Winickoff JP, Ivanoff B, et al. Global burden of Shigellainfections: implications for vaccine development and implementationof control strategies. Bull World Health Organ 1999; 77:651–6.

4. Sansonetti PJ, Arondel J, Fontaine A, d’Hauteville H, Bernardini ML.ompB (osmo-regulation) and icsA (cell-to-cell spread) mutants of Shi-

gella flexneri: vaccine candidates and probes to study the pathogenesisof shigellosis. Vaccine 1991; 9:416–22.

5. Buttner D, Bonas U. Port of entry: the type III secretion translocon.Trends Microbiol 2002; 10:186–92.

6. Zychlinsky A, Prevost MC, Sansonetti PJ. Shigella flexneri induces ap-optosis in infected macrophages. Nature 1992; 358:167–9.

7. Zychlinsky A, Fitting C, Cavaillon JM, Sansonetti PJ. Interleukin 1 isreleased by murine macrophages during apoptosis induced by Shigellaflexneri. J Clin Invest 1994; 94:1328–32.

8. Arondel J, Singer M, Matsukawa A, Zychlinsky A, Sansonetti PJ. In-creased interleukin 1 (IL-1) and imbalance between IL-1 and IL-1receptor antagonist during acute inflammation in experimental shi-gellosis. Infect Immun 1999; 67:6056–66.

9. Sansonetti PJ, Arondel J, Huerre M, Harada A, Matsushima K. Inter-leukin-8 controls bacterial transepithelial translocation at the cost ofepithelial destruction in experimental shigellosis. Infect Immun 1999;67:1471–80.

10. Philpott DJ, Yamaoka S, Israel A, Sansonetti PJ. Invasive Shigella flexneriactivates NF-kappaB through a lipopolysaccharide-dependent innateintracellular response and leads to IL-8 expression in epithelial cells.J Immunol 2000; 165:903–14.

11. Girardin SE, Boneca IG, Carneiro LA, et al. Nod1 detects a uniquemuropeptide from gram negative bacterial peptidoglycan. Science 2003;300:1584–7.

12. Girardin SE, Tournebize R, Mavris M, et al. CARD4/Nod1 mediatesNF-kappaB and JNK activation by invasive Shigella flexneri. EMBORep 2001; 2:736–42.

13. Perdomo OJ, Cavaillon JM, Huerre M, Ohayon H, Gounon P, San-sonetti PJ. Acute inflammation causes epithelial invasion and mucosaldestruction in experimental shigellosis. J Exp Med 1994; 180:1307–19.

14. Sansonetti PJ, Arondel J, Cavaillon JM, Huerre M. Role of IL-1 in thepathogenesis of experimental shigellosis. J Clin Invest 1995; 96:884–92.

15. Sansonetti PJ, Phalipon A, Arondel J, et al. Caspase-1 activation of IL-1beta and IL-18 are essential for Shigella flexneri-induced inflammation.Immunity 2000; 12:581–90.

16. Weinrauch Y, Drujan D, Shapiro SD, Weiss J, Zychlinsky A. Neutrophilelastase targets virulence factors of enterobacteria. Nature 2002; 417:91–4.

17. Voino-Yasenetsky MV, Voino-Yasenetskaya MK. Experimental pneu-monia caused by bacteria of Shigella group. Acta Morphol Acad SciHung 1962; 11:439–54.

18. Phalipon A, Kaufmann M, Michetti P, et al. Monoclonal immuno-globulin A antibody directed against serotype-specific epitope of Shi-gella flexneri lipopolysaccharide protects against murine experimentalshigellosis. J Exp Med 1995; 182:769–78.

19. Fernandez MI, Thuizat A, Pedron T, Neutra M, Phalipon A, SansonettiPJ. A newborn mouse model for the study of intestinal pathogenesisof shigellosis. Cell Microbiol 2003; 5:481–91.

20. Singer M, Sansonetti PJ. IL-8 is a key chemokine regulating neutrophilrecruitment in a new mouse model of Shigella-induced colitis. J Im-munol 2004; 173:4197–206.

21. Allaoui A, Mounier J, Prevost C, Sansonetti PJ, Parsot C. icsB: a Shigellaflexneri virulence gene necessary for the lysis of protrusions duringintercellular spread. Mol Microbiol 1992; 6:1605–16.

22. Sansonetti PJ, Kopecko DJ, Formal SB. Involvement of a plasmid inthe invasive ability of Shigella flexneri. Infect Immun 1982; 35:852–60.

23. Buchrieser C, Glaser P, Rusniok C, et al. The virulence plasmid pWR100and the repertoire of proteins secreted by the type III secretion ap-paratus of Shigella flexneri. Mol Microbiol 2000; 38:760–71.

24. Cersini A, Martino MC, Martini I, Rossi G, Bernardini ML. Analysisof virulence and inflammatory potential of Shigella flexneri purinebiosynthesis mutants. Infect Immun 2003; 71:7002–13.

25. Miller JH. A short course in bacterial genetics. Cold Spring Harbor,NY: Cold Spring Harbor Laboratory Press, 1992.

26. Martino MC, Rossi G, Tattoli I, et al. Intravenous infection of virulentshigellae causes fulminant hepatitis in mice. Cell Microbiol 2005; 7:115–27.

by guest on May 23, 2016

http://jid.oxfordjournals.org/D

ownloaded from

148 • JID 2005:192 (1 July) • Martino et al.

27. Dixon MF, Genta RM, Yardley JF, Correa P. Classification and gradingof gastritis: the updated Sydney System. Am J Surg Pathol 1996; 20:1161–81.

28. Happonen I, Linden J, Saari S, et al. Detection and effects of helico-bacters in healthy dogs and dogs with signs of gastritis. J Am Vet MedAssoc 1998; 213:1767–74.

29. Etheridge ME, Hoque AT, Sack DA. Pathologic study of a rabbit modelfor shigellosis. Lab Anim Sci 1996; 46:61–6.

30. Bernardini ML, Arondel J, Martini I, Aidara A, Sansonetti PJ. Param-eters underlying successful protection with live attenuated mutants inexperimental shigellosis. Infect Immun 2001; 69:1072–83.

31. Brown KJ, Tannock GW, Eyres RA, Elliott RB, Lines R. Colonizationof Salmonella typhimurium and Shigella flexneri III of the gastrointes-tinal tract of mice treated with beta-2-thienylalanine and streptomycin.Antonie Van Leeuwenhoek 1979; 45:531–46.

32. Que JU, Casey SW, Hentges DJ. Factors responsible for increased sus-ceptibility of mice to intestinal colonization after treatment with strep-tomycin. Infect Immun 1986; 53:116–23.

33. Barthel M, Hapfelmeier S, Quintanilla-Martinez L, et al. Pretreatmentof mice with streptomycin provides a Salmonella enterica serovar Ty-phimurium colitis model that allows analysis of both pathogen andhost. Infect Immun 2003; 71:2839–58.

34. Zhang Z, Jin L, Champion G, Seydel KB, Stanley SL Jr. Shigella infection

in SCID-HU-INT mice: role for neutrophils in containing bacterialdissemination in human intestine. Infect Immun 2001; 69:3240–7.

35. Way SS, Borczuk AC, Goldberg MB. An essential role for gammainterferon in innate resistance to Shigella flexneri infection. Infect Im-mun 1998; 66:1342–8.

36. Raqib R, Ljungdahl A, Lindberg AA, Andersson U, Andersson J. Localentrapment of interferon gamma in the recovery from Shigella dysen-teriae type 1 infection. Gut 1996; 38:328–36.

37. Moretta A. Natural killer cells and dendritic cells: rendezvous in abusedtissues. Nat Rev Immunol 2002; 2:957–60.

38. Gerosa F, Baldani-Guerra B, Nisii C, Marchesini V, Carra G, TrinchieriG. Reciprocal activating interaction between natural killer cells anddendritic cells. J Exp Med 2002; 195:327–33.

39. Islam D, Veress B, Bardhan PK, Lindberg AA, Christensson B. In situcharacterization of inflammatory responses in the rectal mucosae ofpatients with shigellosis. Infect Immun 1997; 65:739–49.

40. Singh B, Read S, Asseman C, et al. Control of intestinal inflammationby regulatory T cells. Immunol Rev 2001; 182:190–200.

41. Shevach EM. CD4+CD25+ suppressor T cells: more questions thananswers. Nat Rev Immunol 2002; 2:389–400.

42. Kantor AB, Herzemberg LA. Origin of murine B-cell lineages. AnnuRev Immunol 1993; 11:501–38.

43. Carroll MC, Prodeus AP. Linkages of innate and adaptive immunity.Curr Opin Immunol 1998; 10:36–40.

by guest on May 23, 2016

http://jid.oxfordjournals.org/D

ownloaded from