Full text of DOI:10.1111/j.1348-0421.2002.tb02704.x

Microbiol. Immunol., 46(5), 333―342, 2002
Differential Expression and Polarized Secretion of
CXC and CC Chemokines by Human Intestinal
Epithelial Cancer Cell Lines in Response to
Clostridium difficile Toxin A
Jung Mogg Kim1, Joo Sung Kim2, Hyun Chae Jung*, 2, Yu-Kyoung Oh3, In Sung Song2,
and Chung Yong Kim2
1Department of Microbiology and Institute of Biomedical Science, Hanyang University College of Medicine, Seoul 133
―791,
Korea, 2Department of Internal Medicine and Liver Research Institute, Seoul National University College of Medicine, and Clin-
3
ical Research Institute, Seoul National University Hospital, Seoul 110
―
744, Korea, and Department of Microbiology,
Pochon CHA University College of Medicine, Kyunggi-do 487―800, Korea
Received December 17, 2001. Accepted February 19, 2002
Abstract: Intestinal epithelial cells are the initial sites of host response to Clostridium difficile infection and
can play a role in signaling the influx of inflammatory cells. To further explore this role, the regulated expres-
sion and polarized secretion of CXC and CC chemokines by human intestinal epithelial cells were inves-
tigated. An expression of the CXC chemokines, including IL-8 and growth-related oncogene (GRO)-ꢀ, and
the CC chemokine monocyte chemoattractant protein (MCP)-1 from HT-29 cells increased in the 1―6 hr fol-
lowing C. difficile toxin A stimulation, assessed by quantitative RT-PCR. In contrast, the expression of neu-
trophil activating protein-78 (ENA-78) was delayed for 18 hr. The up-regulated mRNA expression of
chemokines was paralleled by the increase of protein levels. However, the expression of macrophage
inflammatory protein (MIP)-1ꢀ, RANTES (regulated on activation normal T cells expressed and secreted),
and interferon-ꢁ-inducible protein-10 (IP-10) was not changed in HT-29 or Caco-2 cells stimulated with toxin
A. Upon stimulation of the polarized Caco-2 epithelial cells in a transwell chamber with toxin A, CXC and
CC chemokines were released predominantly into the basolateral compartment. Moreover, the addition of
IFN-ꢁ and TNF-ꢀ to toxin A stimulated Caco-2 cells increased the basolateral release of CC chemokine
MCP-1. In contrast, IFN-ꢁ and TNF-ꢀ had no effect on the expression of the CXC chemokines IL-8 and
GRO-ꢀ. These results suggest that a CXC and CC chemokine expression from epithelial cells infected with
C. difficile may be an important factor in the mucosal inflammatory response.
Key words: Clostridium difficile toxin A, Chemokine, Intestinal epithelial cells
Clostridium difficile (C. difficile) causes antibiotic-
associated diarrhea and pseudomembranous colitis in
Abbreviations: dATP, deoxyadenosine 5'-triphosphate; dCTP,
humans (17). The pathogenic strains of C. difficile
deoxycytidine 5'-triphosphate; dGTP, deoxyguanosine 5'-triphos-
phate; dTTP, deoxythymidine 5'-triphosphate; ELISA, enzyme-
linked immunosorbent assay; ENA-78, neutrophil activating pro-
tein-78; FBS, fetal bovine serum; GRO, growth-related oncogene;
release two kinds of large exotoxins: toxin A (308 kDa)
and toxin B (269 kDa). In animal models, only toxin A
is enterotoxic and causes fluid secretion, mucosal edema,
and villous disruption by inducing massive acute inflam-
mation with neutrophil infiltration (2, 17). The first
host cells that C. difficile toxins interact within the colon
are epithelial cells. In vitro studies have shown that
purified toxin A has potent effects on human colonic
HEPES, N-(2-hydroxyethyl)piperazine-N'-ethanesulfonic acid;
IFN-γ, interferon-gamma; IL, interleukin; IP-10, interferon-γ-
inducible protein-10; LDH, lactate dehydrogenase; LPS,
lipopolysaccharide; MCP, monocyte chemoattractant protein;
MIP, macrophage inflammatory protein; NF-κB, nuclear factor-
kappa B; PBMC, peripheral blood mononuclear cells; RANTES,
regulated on activation normal T cells expressed and secreted; RT-
PCR, reverse transcription-polymerase chain reaction; SD, stan-
dard deviation; SDS-PAGE, sodium dodecyl sulfate-polyacryl-
amide gel electrophoresis; SEM, standard error of the means;
TNF-α, tumor necrosis factor-alpha.
*Address correspondence to Professor Hyun Chae Jung,
Department of Internal Medicine, Seoul National University
College of Medicine, 28 Yongon-dong Chongno-gu, Seoul
110
―744, Korea. Fax:＋82―2―743―6701. E-mail: hyunchae
@plaza.snu.ac.kr
333

334
J.M. KIM ET AL
epithelial cells (6, 8, 26). Although intestinal epithelial
cells can produce the neutrophil chemoattractant inter-
leukin (IL)-8 in response to C. difficile toxin A (1, 26),
the exact pathogenic mechanism of C. difficile toxin A-
induced inflammatory cell infiltration has not yet been
fully studied.
(ATCC HTB 38) and Caco-2 (ATCC HTB 37) human
ileocecal epithelial cells were grown in DME (pH 7.4,
Sigma Chemical Co., St. Louis, Mo., U.S.A.) supple-
mented with 10% fetal bovine serum (FBS, Hyclone
Laboratories, Logan, Ut., U.S.A.), 2 mM glutamine, 25
mM HEPES, and antibiotics (100 units/ml of penicillin
and 100 µg/ml of streptomycin), as described previous-
ly (18).
The polarized monolayers of Caco-2 cells were pre-
pared by seeding 5×105 Caco-2 cells on top of collagen-
coated microporous supports (0.4-µm pore size; 4.7-
cm2 growth area) in transwell chambers (Costar, Cam-
bridge, Mass., U.S.A.). The medium was replaced every
2 days, and the cells were allowed to differentiate for 14
days before use (21, 24). The volumes of medium in the
apical and basal compartments were 1.5 ml and 2.6 ml,
respectively. The formation of tight junctions was func-
tionally assessed by the measurement of the electrical
resistance across the monolayers with a Millicell elec-
trical resistance system (Millipore, Bedford, Mass.,
U.S.A.) (21, 24). The electrical resistance of the non-
stimulated control in this study ranged from 430 to 495
Ω×cm2, after the subtraction of the resistance across a
cell-free filter.
To isolate human peripheral blood mononuclear cells
(PBMC), heparinized peripheral blood was obtained
from healthy volunteers. PBMC was separated from
the whole blood by Ficoll-Hypaque (Amersham Phar-
macia Biotech, Uppsala, Sweden) density gradient cen-
trifugation, washed three times in Hanks’ balanced salt
solution, and resuspended in complete RPMI-1640 medi-
um (Sigma) at 5×106 cells/ml. Human neutrophils
were obtained from healthy volunteers by dextran sedi-
mentation and Ficoll-Hypaque density gradient separa-
tion, as previously described (22). PBMC and neu-
trophils were washed, resuspended, and cultured at a
concentration of 5×106 cells/ml in RPMI-1640 media
supplemented with 10 mM HEPES (Sigma) and 10%
FBS; a polypropylene tube was used to prevent the cells
from adhering.
Chemokines are low-molecular-weight proteins with
pleiotropic effects on the recruitment and activation of
leukocytes at the sites of inflammation. Based on the
arrangement of the conserved cysteine residues (28),
chemokines have been grouped into four distinct families,
CC, CXC (where X can be any amino acid), C, and
CX3C. The CXC chemokine family can be further
divided, depending on whether its members have an
ELR (Glu-Leu-Arg) amino acid motif, which is important
for the chemoattraction and activation of neutrophils
[e.g., epithelial-neutrophil activating protein-78 (ENA-
78), growth-related oncogene (GRO) family members,
and IL-8], or whether they lack this motif [e.g., inter-
feron-γ-inducible protein-10 (IP-10)] (4, 28). Members
of the CC family of chemokine function mainly as
chemoattractants of monocytes, eosinophils, lympho-
cytes, and basophils (28). Several reports have shown the
up-regulated expression of chemokines in human intesti-
nal epithelial cells after pathogenic microbial infection
(13, 14, 19, 20, 24). These studies have suggested that
epithelial cells, which line the human intestinal mucosa,
can interact with C. difficile toxin A and may provide
early or late signals for the initiation and maintenance of
the mucosal inflammatory response. Although several
papers have demonstrated that C. difficile toxin A stim-
ulates the expression of IL-8 in human intestinal epithe-
lial cells (1, 26) or monocytes (7, 25), it is still not
known whether chemokines other than IL-8 are also
produced by human intestinal epithelial cells in response
to C. difficile infection, and if so whether cytokines pro-
duced by immune cells could influence the chemokine
expression by C. difficile toxin stimulated epithelial
cells.
For a better understanding of the mechanisms by
which epithelial cells can participate in the mucosal
inflammatory response in the intestine following toxi-
genic C. difficile infection, we assessed the expression of
CXC and CC chemokines in human intestinal epithelial
cancer cell lines. Here we demonstrate that stimula-
tion with C. difficile toxin A resulted in an up-regulated
expression and a basolateral secretion of the CXC
chemokines IL-8, GRO-α, and ENA-78 and the CC
chemokine MCP-1 in human intestinal epithelial cells.
For the immunostimulation of epithelial cells, recom-
binant human interferon (IFN)-γ (PharMingen, San
Diego, Calif., U.S.A.) and tumor necrosis factor (TNF)-
α (R&D Systems, Minneapolis, Minn., U.S.A.) were
used.
Purification of C. difficile toxin A. C. difficile toxin A
was purified from a toxigenic strain (ATCC 9689), as
described previously (15). Briefly, toxigenic C. difficile
was cultured anaerobically at 37 C in dialysis tubing, and
culture filtrates were applied to a thyroglobulin affinity
column. Fractions showing cytotoxicity (as assessed
by the rounding of Vero cells) and hemagglutinating
activity (determined by using a 1% rabbit erythrocyte
Materials and Methods
Cell cultures. HT-29 human colon epithelial cells

335
CHEMOKINE EXPRESSION BY C. DIFFICILE TOXIN A
suspension) were subsequently subjected to two sequen-
tial anion exchange chromatographic steps with Q
Sepharose FF and Mono Q columns (both obtained from
Pharmacia Biotech, Brussels, Belgium) incorporated
into a fast protein liquid chromatography apparatus
(Pharmacia Biotech). The purity of toxin A was then
monitored by a commercial monoclonal ELISA kit for
toxin A of C. difficile (BioWhittaker, Inc., Walkersville,
Md., U.S.A.). To determine the purity of C. difficile
toxin A, we tested whether the toxin A-induced patho-
genic morphology of HT-29 epithelial cells could be
recovered by the treatment of anti-toxin A antibody.
The toxin A pretreated unhealthy HT-29 cells recov-
ered the normal morphology when anti-toxin A (Tech-
Lab, Blacksburg, Va., U.S.A.) was added. Moreover, the
purified toxin also showed a single band in SDS-PAGE
(data not shown). Buffers were prepared by the use of
LPS-free water (Baxter Healthcare Corp., Deerfield, Ill.,
U.S.A.), and lipopolysaccharide (LPS) activity in toxin
A solutions (1 mg/ml) was less than one endotoxin
unit/ml (quantitative chromogenic limulus amebocyte
lysate; BioWhittaker, Walkersville, Md., U.S.A.) (32).
Aliquots of the purified toxin A (at concentrations of 50
to 100 µg/ml) were frozen at －70 C until used.
RNA extraction and RT-PCR analysis. Confluent
monolayers of HT-29 or Caco-2 cells in 24-well plates
were stimulated with C. difficile toxin A for the indicat-
ed times. The total cellular RNA was then extracted from
the cells by the acid guanidinium thiocyanate-phenol-
chloroform method. Oligonucleotide primers used for
PCR amplification and the sizes of the PCR products
obtained from target cellular RNA are described in the
previous report (12).
Quantitative RT-PCR using an internal standard was
employed to quantify cytokine mRNA levels according
to our previous methods (12, 13). Plasmids for syn-
thetic standard RNAs were kindly provided by Dr. Kag-
noff of the University of California, San Diego. Briefly,
serial dilutions of standard RNA molecules (from 103 to
108) were mixed with 1 µg of extracted total RNA from
the cells and reverse transcribed at 37 C for 60 min
under the previously described conditions (12, 13). Sub-
sequently, 5 µl of the cDNA mixture was amplified by a
thermal cycler (GenAmp PCR system 9600; Perkin
Elmer Cetus, Norwalk, Conn., U.S.A.) in 50 µl of 10 mM
Tris, pH 8.3; 50 mM KCl; 2 mM MgCl2; 200 µM con-
centrations each of dATP, dCTP, dGTP, and dTTP; and
25 pM each of 5' and 3' primer. PCR amplification con-
sisted of 35 cycles of 1-min denaturation at 95 C, 2.5-min
annealing, and extension at 60 C (IL-8, GRO-α, MCP-1,
and IP-10), 65 C [ENA-78, macrophage inflammatory
protein (MIP)-1α, and regulated on activation normal T
cells expressed and secreted (RANTES)], or 72 C (IFN-
γ). For an amplification of β-actin, PCR was performed
for 32 cycles of 1-min denaturation at 95 C and 2.5-min
annealing and extension at 72 C. PCR products were
separated in 2% NuSieve agarose gel (FMC Bioproducts,
Rockland, Me., U.S.A.) and visualized by the use of
ethidium bromide stain. Chemokine mRNA levels of
5×103 molecules/µg of total RNA were considered pos-
itive. Although lower levels could be detected and quan-
tified, they were not considered to be meaningful bio-
logically because they would reflect less than 1 mRNA
transcript per 20 cells on average (20). Each experiment
included negative controls in which RNA was omitted
from the reverse transcription mixture, and cDNA was
absent in the PCR reaction. RNA from phytohemag-
glutinin-M (Sigma)-activated PBMC was used as a pos-
itive control.
Chemokine ELISA and lactate dehydrogenase (LDH)
assays. The CXC and CC chemokines in culture super-
natants were assayed by enzyme-linked immunosorbent
assay (ELISA). Before chemokine proteins were mea-
sured, the supernatants were filtered through a 0.22-µm
filter to remove any contaminants. The levels of human
IL-8, GRO-α, ENA-78, and MCP-1 were determined
by the use of Quantikine immunoassay kits (R&D Sys-
tems) as described by the manufacturer.
The LDH activity in culture supernatants was assayed
by a colorimetric method based on the reduction of
pyruvate to lactate in the presence of LDH and NADH
(21, 24). The remaining pyruvic acid was colorimetri-
cally detected after a reaction with 2,4-dinitrophenylhy-
drazine to form a colored hydrazone (LDH-LD, Sigma).
The absorbance was determined at 450 nm.
Statistical analysis. Results are presented as mean ±
standard deviation (SD) for quantitative RT-PCR and
mean ± standard error of the means (SEM) for ELISA.
Wilcoxon’s rank sum test was used for statistical analy-
sis. A P value of less than 0.05 was considered statisti-
cally significant.
Results
Expression of CXC and CC Chemokine Genes in Human
Intestinal Epithelial Cells in Response to C. difficile
Toxin A
IL-8, GRO-α, and ENA-78 are CXC chemokines that
carry an amino terminal ELR motif and act as potent
chemoattractants and activators of neutrophils (28). HT-
29 cells constitutively expressed low levels of IL-8 and
GRO-α mRNA expression, but the expression of these
CXC chemokines increased after stimulation with C.
difficile toxin A (Fig. 1). The numbers of specific tran-
scripts were quantified from the RNA extracts of C. dif-
ficile toxin A stimulated and non-stimulated cells. As

336
J.M. KIM ET AL
Fig. 2. Time course of chemokine mRNA. The confluent HT-29
monolayers in 24-well plates were incubated with C. difficile
toxin A (10 ng/ml) for the indicated period. For the quantification
of CXC and CC chemokine and β-actin transcripts, the total
RNA was reverse-transcribed by the use of an oligo(dT) primer
and synthetic internal RNA standards and amplified by PCR.
The values were expressed as the mean±SD of five separate
experiments. ●, IL-8; □, GRO-α; ▲, ENA-78; ◇, MCP-1.
Fig. 1. RT-PCR analysis of CXC and CC chemokine mRNA in C.
difficile toxin A stimulated HT-29 colon epithelial cells. The
confluent monolayers of HT-29 cells were stimulated with C. dif-
ficile toxin A (10 ng/ml) for the indicated hours, after which
cellular RNA was extracted by the guanidinium thiocyanate-
phenol-chloroform method. The expression of mRNA for each
chemokine and β-actin was assessed by RT-PCR by the use of
specific primers. The data are from representative experiments.
The (＋) and (－) represent positive and negative controls,
respectively. The “C” represents the bands of RT-PCR products for
cellular RNA extracted from non-stimulated HT-29 cells, and the
“T” represents those from C. difficile toxin A stimulated cells.
C. difficile toxin A stimulation of another human intesti-
nal epithelial cell line, Caco-2 (Table 1). However, the
expression of CXC chemokine IP-10 and CC chemokine
MIP-1α (a chemoattractant of monocytes and T cells)
and RANTES (a chemoattractant of monocytes, T cells,
and eosinophil) was not altered in toxin A stimulated HT-
29 cells (IP-10, ＜5×103 transcripts/µg RNA; MIP-1α,
＜5×103 transcripts/µg RNA; RANTES, ＜5×103 tran-
scripts/µg RNA) or Caco-2 cells (IP-10, ＜5×103 tran-
scripts/µg RNA; MIP-1α, ＜5×103 transcripts/µg RNA;
RANTES, ＜5×103 transcripts/µg RNA). The mRNA
levels of β-actin in stimulated cells remained relatively
constant throughout the same period (~ 5×106 tran-
scripts/µg RNA).
shown in Fig. 2, the expression of mRNA transcripts for
IL-8 was first noted to increase at 1 hr after stimulation,
compared with non-stimulated control (5.6-fold increase).
The expression peaked at 3 hr poststimulation (IL-8,
~ 50-fold increase) or 6 hr poststimulation (GRO-α, ~ 32-
fold increase) and decreased to baseline thereafter. In
contrast, the kinetics of ENA-78 mRNA expression was
delayed relative to the other CXC chemokines tested
(peaked at 18 hr poststimulation, ~ 26-fold increase),
implying the difference in the regulatory mechanism
that controls ENA-78 expression after C. difficile toxin A
stimulation.
CC chemokine MCP-1, a potent chemoattractant of
monocytes, was also up-regulated in HT-29 stimulated
with C. difficile toxin A, as shown in Fig. 1. An
increased expression of mRNA transcripts for MCP-1
was first noted at 3 hr after stimulation, compared with
non-stimulated control (6.6-fold increase) (Fig. 2). The
expression peaked at 6 hr poststimulation (12-fold
increase) and decreased to baseline thereafter. Similar
increases were noted in mRNA expressions of CXC
(ENA-78, GRO-α, and IL-8) and CC (MCP-1) following
CXC and CC Chemokine Secretion by Intestinal Epithe-
lial Cells in Response to C. difficile Toxin A Stimulation
To determine whether increased chemokine mRNA
levels were accompanied by increased protein secre-
tion, we measured the amount of chemokine proteins in
culture supernatants. The kinetics of CXC and CC
chemokine secretion was in parallel with the mRNA
expression (Fig. 3). For example, HT-29 cells stimulat-
ed with C. difficile toxin A for 24 hr produced ~ 8-fold
higher amounts of IL-8 compared with non-stimulated
control. Similarly, Caco-2 cells treated with toxin A
(10 ng/ml) for 24 hr showed several-fold increases in the
secretion of CXC chemokines [ENA-78, 2.8±0.9; GRO-
α, 11.5±3.2; IL-8, 5.2±1.6 (the mean of fold increase
±SEM, n＝5)] and CC chemokines [MCP-1, 3.5±1.2
(the mean of fold increase±SEM, n＝5)].
The magnitude of the chemokine response was depen-

337
CHEMOKINE EXPRESSION BY C. DIFFICILE TOXIN A
Table 1. Chemokine mRNA levels in Caco-2 intestinal epithelial cells stimulated with C.
difficile toxin A
Chemokine mRNA (×104 transcripts/µg of RNA)
Chemokine
Control
Toxin A stimulated
Ratio of
stimulated/control
CXC
IL-8
GRO-α
ENA-78
7.0±2.2
1.3±0.5
＜0.5
112±33
48±25
2.3±1.8
17
40
＞5
CC
MCP-1
3.2±0.5
35±19
11
β-Actin
290±800
288±50
1.0
Caco-2 cells were stimulated with C. difficile toxin A (10 ng/ml). Chemokine mRNA lev-
els were determined by the use of quantitative RT-PCR at 3 hr (IL-8), 6 hr (GRO-α and
MCP-1), or 18 hr (ENA-78) after stimulation. Values are expressed as mean±SD of five
separate experiments.
Fig. 3. CXC and CC chemokine secretion by HT-29 cells stimu-
lated with C. difficile toxin A. The confluent HT-29 monolayers
in 24-well plates were incubated with C. difficile toxin A (10
ng/ml) for the indicated period, and the protein levels of each
chemokine were determined by ELISA. The data are the mean±
SEM of five separate experiments. ●, IL-8; □, GRO-α; ▲,
ENA-78; ◇, MCP-1.
dent on the concentration of stimulating C. difficile toxin
A per epithelial cell. The stimulation of HT-29 cells with
increasing concentrations of toxin A was paralleled by
increased IL-8 and GRO-α release. At the concentrations
of 0.1, 1, 10, and 100 ng/ml, IL-8 release increased 1.9,
3.3, 6.7, and 5.6-fold at 24 hr after stimulation, respec-
tively, relative to those of non-stimulated controls (Fig. 4).
Fig. 4. IL-8 and GRO-α production of HT-29 cells stimulated with
C. difficile toxin A. The confluent HT-29 monolayers in 24-well
plates were incubated with the indicated concentration of C. dif-
ficile toxin A for 24 hr, and the protein levels of IL-8 and GRO-
α were determined by ELISA. The data are the mean±SEM of
five separate experiments.
Kinetics of ENA-78 and IL-8 Secretion by Intestinal
Epithelial Cells after Stimulation with C. difficile Toxin A
Since the kinetics of ENA-78 mRNA expression was
delayed relative to the other chemokines tested (Fig. 2),
the secretion kinetics of ENA-78 and IL-8 was studied in
HT-29 cells stimulated with C. difficile toxin A. The
protein secretion was assayed at 6, 12, 18, and 24 hr after
stimulation. As shown in Table 2, the maximal rate of
IL-8 secretion occurred during the first 6 hr after stimu-
lation, with little additional IL-8 secreted after 12 hr. In
contrast, the kinetics of ENA-78 secretion was delayed in
comparison with that of IL-8. These results are compa-
rable with the delayed mRNA expression of ENA-78 in
response to C. difficile toxin A (Fig. 2).

338
J.M. KIM ET AL
Table 2. Kinetics of ENA-78 and IL-8 secretion by HT-29 cells following stimulation with C.
difficile toxin A
Control
Cumulative
Toxin A stimulation
Cumulative Secretion
secretion rate
Time after
infection (hr)
Secretion
rate
Chemokine
ENA-78
secretion
(pg/ml)
(pg/ml·hr)
(pg/ml)
(pg/ml·hr)
6
12
18
24
98±15
98±12
102±14
103±16
―
―
―
―
92±7
96±12
126±11
512±62
―
1
5
64
IL-8
6
12
18
24
92±12
205±27
270±36
276±25
15
19
11
3
840±26
1,512±287
1,734±266
1,816±251
140
112
37
14
Culture supernatants were harvested at the indicated times after stimulation with C. difficile
toxin A (10 ng/ml), and the cumulative secretion of ENA-78 and IL-8 during the indicated time
period was assayed by ELISA. The secretion rates for each culture were calculated by sub-
tracting the cumulative secretion of the prior culture period from the total cumulative secretion
of each respective period and dividing the net value by the number of hours in the respective
period. Values are expressed as mean±SEM of five separate experiments.
Fig. 5. Basolateral chemokine secretion and apical LDH released by polarized Caco-2 cells following stimulation with C. difficile toxin
A. The polarized monolayers of Caco-2 cells in transwell chambers were stimulated with C. difficile toxin A (10 ng/ml) for the indicated
period, and the supernatants were obtained from upper and lower chambers. The chemokine secretion was determined by ELISA, and
the LDH activity was determined by enzymatic assay. The data are the mean±SEM of seven separate experiments.

339
CHEMOKINE EXPRESSION BY C. DIFFICILE TOXIN A
Basolateral Secretion of CXC and CC Chemokines by
Polarized Caco-2 Intestinal Epithelial Cells after Stim-
ulation with C. difficile Toxin A
Intestinal epithelial cells are structurally and func-
tionally polarized in vivo (9). If the chemokines pro-
duced by C. difficile toxin A stimulated epithelial cells
play a physiological role in the influx of inflammatory
cells into the mucosa, they would be released predictably
from the basolateral surfaces of the bacterial infected
epithelial cells. To determine the polarity of chemokine
secretion by C. difficile toxin A stimulated epithelial
cells, we cultured Caco-2 cells as polarized monolayers
on transwell chambers that form tight junctions. The
impermeability of monolayers to IL-8, GRO-α, or MCP-
1 was established by the addition of a relatively high dose
of IL-8 (10 ng/ml), GRO-α (10 ng/ml), or MCP-1 (10
ng/ml) to the basolateral compartment of non-stimulated
monolayers. This resulted in the appearance of only
~2―3% of the added IL-8, GRO-α, or MCP-1 in the apical
compartment 24 hr later. After apical stimulation with C.
difficile toxin A, the levels of IL-8, GRO-α, MCP-1,
and LDH were determined in both apical and basolater-
al compartments at the indicated times. As shown in Fig.
5, more than 85% of IL-8, GRO-α, and MCP-1 were
found in the basolateral compartment, whereas LDH
was released predominantly into the apical compart-
ment. In this system, the electrical resistance across
monolayers decreased only slightly after stimulation
(430 to 495 Ω×cm2 in controls and 410 to 365 Ω×cm2
at 24 hr after stimulation with 10 ng/ml of toxin A). In
up to 24 hr of poststimulation, the concentration of LDH
was higher in the apical compartment than in the baso-
lateral compartment. However, by 48 hr poststimulation
the apical levels of LDH were approximately equal to the
basolateral levels. The electrical resistance across mono-
layers did not decrease significantly at 18 hr after stim-
ulation with 10 ng/ml of toxin A (451.0±35.8 Ω×cm2
in controls; 417.2±30.6 Ω×cm2 in toxin A stimula-
tion; mean±SEM, n＝5, P＞0.05). These results indi-
cate that the secreted amounts of IL-8, GRO-α, and
MCP-1 in the basolateral compartment do not include the
diffusion of those from the apical compartment within at
least 18 hr after stimulation. Thus it appears that C. dif-
ficile toxin A stimulated epithelial cells secrete IL-8,
GRO-α, and MCP-1 predominantly from the basolater-
al surface. Similarly, basolateral secretion was observed
in the production of ENA-78 after the stimulation of
Fig. 6. IFN-γ and TNF-α increase basolateral MCP-1 release in C.
difficile toxin A stimulated polarized Caco-2 epithelial cells.
The polarized monolayers of Caco-2 cells in transwell chambers
were apically stimulated with C. difficile toxin A (10 ng/ml) in the
basolateral presence of IFN-γ (50 ng/ml) and TNF-α (10 ng/ml)
for 24 hr, and the supernatants were obtained from the upper and
lower chambers. The MCP-1, IL-8, and GRO-α secretion were
determined by ELISA. The data are the mean±SEM of seven
separate experiments. The asterisk indicates statistical significance
with P＜0.05 in comparison with the toxin A stimulated group
without added IFN-γ and TNF-α.

340
J.M. KIM ET AL
Caco-2 cells with C. difficile toxin A (basolateral com-
partment, 384±42 pg/ml; apical compartment, 58±12
pg/ml; 18 hr poststimulation, n＝5).
cells could express and secrete several CXC and CC
chemokines. The epithelial response to C. difficile toxin
A appears to be well suited for contributing to the accu-
mulation and activation of inflammatory cells, including
neutrophils and monocytes, at the site of infection by
establishing a chemotactic gradient within the underlying
mucosa.
IFN-γ and TNF-α Increase the Basolateral Release of
CC Chemokine MCP-1 from C. difficile Toxin A Stimu-
lated Polarized Epithelial Cells
IFN-γ or TNF-α induces differential up-regulation
of CXC and CC chemokine secretions in colonic epithe-
lial cell lines (23, 31). In this study, the stimulation
with C. difficile toxin A did not up-regulate IFN-γ expres-
sion in HT-29 and Caco-2 intestinal epithelial cells.
Unlike epithelial cells, PBMC stimulated with toxin A
(10 ng/ml) for 6 hr resulted in higher levels of IFN-γ
mRNA than in control (toxin A stimulated, 9.6×106
IFN-γ transcripts/µg of RNA; control, 5.2×105 IFN-γ
transcripts/µg of RNA; mean values of three separate
experiments). Neutrophils, however, did not express
IFN-γ, as assessed by quantitative RT-PCR. Meanwhile,
the combined treatment of human neutrophils with toxin
A (10 ng/ml) and IFN-γ (50 ng/ml) for 6 hr significant-
ly increased the expression of TNF-α, whereas toxin A
alone was unable to induce TNF-α expression (1.7×106
TNF-α mRNA transcripts/µg of RNA in toxin A ＋
IFN-γ; 1.5×105 TNF-α mRNA transcripts/µg of RNA in
toxin A; 9.8×104 TNF-α mRNA transcripts/µg of RNA
in IFN-γ-treated; 2.1×105 TNF-α mRNA transcripts/µg
of RNA in non-treated control; these are the mean values
of three separate experiments). These results suggest that
the immune cells can produce IFN-γ or TNF-α in
response to C. difficile infection, and those produced
cytokines may influence the neighboring mucosal epithe-
lial cells. Since a high population of lymphocytes is
found in intestinal mucosa infected with C. difficile (2,
11, 17), we tested whether IFN-γ and TNF-α stimulation
could influence CXC and CC chemokine release from C.
difficile toxin A stimulated polarized epithelial cells.
As shown in Fig. 6, the basolateral addition of IFN-γ (50
ng/ml) and TNF-α (10 ng/ml) to toxin A stimulated
cells showed an increase of CC chemokine MCP-1 pro-
duction into the basolateral compartment, as compared
with toxin A stimulated Caco-2 cells in the absence of
IFN-γ and TNF-α. In marked contrast, IFN-γ and TNF-
α showed no significant effects on the secretion of CXC
chemokines IL-8 and GRO-α in C. difficile stimulated
epithelial cells.
The secretion of CXC and CC chemokines after stim-
ulation with C. difficile toxin A occurred mostly from the
basolateral surface in polarized monolayers of intestinal
epithelial cells. Our data suggest that increased baso-
lateral chemokine secretion did not simply result from
cell lysis, since LDH, a marker of cell lysis, was pre-
dominantly found in the apical compartment during the
24 hr after toxin A stimulation. In general, secreted
proteins not specifically targeted to the apical surfaces of
polarized epithelial cells appear to be predominantly
released at the basolateral surfaces of those cells (27).
Within 3 hr of poststimulation with C. difficile toxin A,
intestinal epithelial cells rapidly up-regulated the expres-
sion of the CXC chemokines IL-8 and GRO-α, which are
potent neutrophil chemoattractants. The kinetics of IL-
8 mRNA expression by epithelial cells coincides with the
previous results of the other group (1). However, the
kinetic of other chemokines except IL-8 has not been
demonstrated in the human colon epithelial cells stimu-
lated with C. difficile toxin A. The expression of CXC
chemokines observed in this study suggests that intesti-
nal epithelial cells can provide signals for the mucosal
influx of neutrophils in response to C. difficile infec-
tion. Moreover, the rapid expression of CC chemokine
MCP-1 by intestinal epithelial cells indicates that muco-
sal epithelial cells may also produce signals for the
chemoattraction of a monocyte population. In this
regard, the expression of CXC and CC chemokines by
intestinal epithelial cells could, in part, explain the pres-
ence of different types of inflammatory cells during the
course of the acute mucosal inflammatory response.
ENA-78 is also important in the recruitment of neu-
trophils from the lamina propria into the epithelial layers
(16). Unlike the other chemokines studied, however, the
up-regulated expression and production of ENA-78
showed a delayed time course. ENA-78 mRNA expres-
sion did not reach maximal levels for 12 hr after stimu-
lation with C. difficile toxin A, and high levels of ENA-
78 secretion persisted for at least 24 hr in the intestinal
epithelial cell lines. Invasive bacterial infection also
showed a similar pattern of the delayed kinetics of ENA-
78 production in intestinal epithelial cells (33). This sug-
gests that intestinal epithelial cells, besides their role in
initiating acute mucosal inflammation through the rapid
production of neutrophil and monocyte chemoattrac-
tants, may play a role during the later phases of the
Discussion
C. difficile infection supports the infiltration of many
numbers of inflammatory cells such as neutrophils and
lymphocytes into the mucosal layer (2, 11, 17). The pres-
ent study demonstrated that human intestinal epithelial

341
CHEMOKINE EXPRESSION BY C. DIFFICILE TOXIN A
mucosal inflammatory response. The mechanisms under-
lying the delayed expression of ENA-78, relative to the
other chemokines, in the intestinal epithelial cells remain
to be further elucidated. However, the differences in their
upstream promoter regions and/or activation of their
relevant transcription factors may provide an explanation,
since other cell types are known to express these
chemokines with delayed kinetics (3, 29). IL-8, which is
secreted earlier and in relatively larger quantities than
ENA-78, was 10 to 100-fold more potent than ENA-78 in
its ability to chemoattract neutrophils in vitro (30).
These results imply that the nature of the chemotactic
gradients that are established within the mucosa by dif-
ferent chemokines with similar activities (e.g., IL-8 and
ENA-78 are both neutrophil chemoattractants) could
have an influence on mucosal inflammatory cell migra-
tion and localization.
It is quite unlikely that the up-regulation of CXC and
CC chemokines in colon epithelial cell lines is due to
contaminants such as LPS. In this study, all buffers
were prepared with LPS-free water during the toxin
purification steps, and LPS activity in toxin A solutions
(1 mg/ml) was found to be less than one endotoxin
unit/ml.
cells infected with invasive bacteria (5) or IL-8 produc-
tion by human monocytes stimulated with C. difficile
toxin A (10). Therefore further studies are undergoing to
determine if the interaction of transcription factors such
as NF-κB with cytokines may account for the syner-
gistic effects of IFN-γ and TNF-α observed here.
In conclusion, the present study provides a possibili-
ty that intestinal epithelial cells may be an important
source of CXC and CC chemokines in response to toxi-
genic C. difficile infection, leading to the provocation of
the mucosal inflammatory response.
References
1) Branka, J.E., Vallette, G., Jarry, A., Bou-Hanna, C., Lemarre,
P., Van, P.N., and Laboisse, C.L. 1997. Early functional
effects of Clostridium difficile toxin A on human colonocytes.
Gastroenterology 112: 1887―1894.
2) Burakoff, R., Zhao, L., Celifarco, A.J., Rose, K.L., Donovan,
V., Pothoulakis, C., and Percy, W.H. 1995. Effects of purified
Clostridium difficile toxin A on rabbit distal colon. Gas-
troenterology 109: 348―354.
3) Corbett, M.S., Schmitt, I., Riess, O., and Walz, A. 1994.
Characterization of the gene for human neutrophil-activating
peptide 78 (ENA-78). Biochem. Biophys. Res. Commun.
205: 612―617.
It is interesting that the stimulation with a combination
of IFN-γ and TNF-α showed a differential up-regulation
of CC and CXC chemokines in C. difficile toxin A stim-
ulated epithelial cells. In this study, we found that IFN-
γ was up-regulated in human PBMC stimulated with
toxin A and that TNF-α was up-regulated in neutrophils
treated with toxin A and IFN-γ. On a single cell basis,
neutrophils and lymphocytes may secrete small quanti-
ties of IFN-γ or TNF-α in response to toxin A stimula-
tion. However, they constitute a majority of infiltrating
cells in C. difficile-infected intestinal mucosa (2, 11,
17) and may be the major source of IFN-γ and TNF-α.
This led us to hypothesize that the IFN-γ or TNF-α pro-
duced by the infiltrating immune cells may influence
the neighboring mucosal epithelial cells. Thus the addi-
tion of IFN-γ and TNF-α to C. difficile toxin A stimulated
intestinal epithelial cells showed an increased basolateral
release of CC chemokine MCP-1, whereas IFN-γ or
TNF-α had no effect on the release of CXC chemokine
IL-8 and GRO-α in toxin A stimulated epithelial cells.
Although the mechanism of up-regulation of MCP-1
by co-stimulation with IFN-γ and TNF-α is still unclar-
ified, it is possible that these cytokines can regulate the
transcription factors in intestinal epithelial cells. IFN-γ
is known to synergistically enhance TNF-α-induced IL-
8 production in a human gastric cancer cell line through
a synergistic activation of transcription factors (34).
Nuclear factor-kappa B (NF-κB) activation induces IL-
8 and MCP-1 production by human colon epithelial
4) Dewald, B., Moser, B., Barella, L., Schumacher, C., Baggi-
olini, M., and Clark-Lewis, I. 1992. IP-10, a gamma-inter-
feron-inducible protein related to interleukin-8, lacks neu-
trophil activating properties. Immunol. Lett. 32: 81―84.
5) Elewaut, D., DiDonato, J.A., Kim, J.M., Truong, F., Eck-
mann, L., and Kagnoff, M.F. 1999. NF-kappa B is a central
regulator of the intestinal epithelial cell innate immune
response induced by infection with enteroinvasive bacteria.
J. Immunol. 163: 1457
6) Fiorentini, C., and Thelestam, M. 1991. Clostridium difficile
toxin A and its effects on cells. Toxicon 29: 543 567.
―1466.
―
7) Flegel, W.A., Muller, F., Daubener, W., Fischer, H.G.,
Hadding, U., and Northoff, H. 1991. Cytokine response by
human monocytes to Clostridium difficile toxin A and toxin
B. Infect. Immun. 59: 3659―3666.
8) Hecht, G., Pothoulakis, C., LaMont, J.T., Carlson, S., and
Madara, J.L. 1998. C. difficile toxin A perturbs cytoskeletal
structure and tight junction permeability of cultured human
intestinal epithelial monolayers. J. Clin. Invest. 82: 1516
―
1524.
9) Hofer, D., Jons, T., Kraemer J., and Drenckhahn, D. 1998.
From cytoskeleton to polarity and chemoreception in the
gut epithelium. Ann. N.Y. Acad. Sci. 859: 75―84.
10) Jefferson, K.K., Smith, M.F., Jr., and Bobak, D.A. 1999.
Roles of intracellular calcium and NF-kappa B in the
Clostridium difficile toxin A-induced up-regulation and
secretion of IL-8 from human monocytes. J. Immunol. 163:
5183
11) Johnson, S., and Gerding, D.N. 1998. Clostridium difficile-
associated diarrhea. Clin. Infect. Dis. 26: 1027 1034.
―5191.
―
12) Jung, H.C., Eckmann, L., Yang, S.K., Pamja, A., Fierer, J.,
Morzycka-Wroblewska, E., and Kagnoff, M.F. 1995. A dis-

342
J.M. KIM ET AL
derived cytokines. Eur. J. Immunol. 29: 530―536.
tinct array of proinflammatory cytokines is expressed in
human colon epithelial cells in response to bacterial invasion.
J. Clin. Invest. 95: 55 65.
24) Laurent, F., Eckmann, L., Savidge, T.C., Morgan, G., Theo-
dos, C., Naciri, M., and Kagnoff, M.F. 1997. Cryptosporid-
ium parvum infection of human intestinal epithelial cells
induces the polarized secretion of C-X-C chemokines. Infect.
―
13) Jung, H.C., Kim, J.M., Song, I.S., and Kim, C.Y. 1997.
Helicobacter pylori induces an array of pro-inflammatory
cytokines in human gastric epithelial cells: quantification of
mRNA for interleukin-8, -1α/β, granulocyte-macrophage
colony-stimulating factor, monocyte chemoattractant protein-
1 and tumour necrosis factor-α. J. Gastroenterol. Hepatol. 12:
Immun. 65: 5067―5073.
25) Linevsky, J.K., Pothoulakis, C., Keates, S., Warny, M.,
Keates, A.C., Lamont, J.T., and Kelly, C.P. 1997. IL-8 release
and neutrophil activation by Clostridium difficile toxin-
473
14) Kagnoff, M.F., and Eckmann, L. 1997. Epithelial cells as sen-
sors for microbial infection. J. Clin. Invest. 100: 6 10.
15) Kamiya, S., Reed, P.J., and Borriello, S.P. 1989. Purification
and characterization of Clostridium difficile toxin A by
bovine thyroglobulin affinity chromatography and dissocia-
tion in denaturing conditions with or without reduction. J.
―
480.
exposed human monocytes. Am. J. Physiol. 273: G1333―
G1340.
26) Mahida, Y.R., Makh, S., Hyde, S., Gray, T., and Borriello,
S.P. 1996. Effect of Clostridium difficile toxin A on human
intestinal epithelial cells: induction of interleukin 8 produc-
―
tion and apoptosis after cell detachment. Gut 38: 337―347.
27) Rindler, M.J., and Traber, M.G. 1998. A specific sorting
signal is not required for the polarized secretion of newly syn-
thesized proteins from cultured intestinal epithelial cells. J.
Med. Microbiol. 30: 69―77.
16) Keates, S., Keates, A.C., Mizoguchi, E., Bhan, A., and
Kelly, C.P. 1997. Enterocytes are the primary source of the
chemokine ENA-78 in normal colon and ulcerative colitis.
Cell Biol. 107: 471
―479.
28) Rollins, B. 1997. Chemokines. Blood 90: 909
―928.
Am. J. Physiol. 273: G75
17) Kelly, C.P., and LaMont, J.T. 1998. Clostridium difficile
infection. Annu. Rev. Med. 49: 375 390.
―
G82.
29) Schnyder-Candrian, S., and Walz, A. 1997. Neutrophil-acti-
vating protein ENA-78 and IL-8 exhibit different patterns of
expression in lipopolysaccharide- and cytokine-stimulated
―
18) Kim, J.M., Eckmann, L., Savidge, T.C., Lowe, D.C., Wit-
thoft, T., and Kagnoff, M.F. 1998. Apoptosis of human
intestinal epithelial cells after bacterial invasion. J. Clin.
human monocytes. J. Immunol. 158: 3888
30) Walz, A., Strieter, R.M., and Schnyder, S. 1993. Neutrophil-
activating peptide ENA-78, p. 129 139. In Lindley, I.J.D.,
―3894.
―
Invest. 102: 1815
―
1823.
Westwick, J., and Kunkel, S. (eds), The chemokines, Biolo-
gy of the inflammatory peptide supergene family II, Plenum,
New York.
19) Kim, J.M., Jung, H.C., Im, K.I., Song, I.S., and Kim, C.Y.
1998. Synergy between Entamoeba histolytica and
Escherichia coli in the induction of cytokine gene expression
31) Warhurst, A.C., Hopkins, S.J., and Warhurst, G. 1998. Inter-
feron gamma induces differential up-regulation of alpha
and beta chemokine secretion in colonic epithelial cell lines.
in human colon epithelial cells. Parasitol. Res. 84: 509
512.
―
20) Kim, J.M., Jung, H.C., Jin, D.Z., Im, K.I., Song, I.S., and
Kim, C.Y. 1997. Cytokine genes are down-regulated when
attachment of Entamoeba histolytica to HT-29 colon epithe-
Gut 42: 208―213.
32) Warny, M., Keates, A.C., Keates, S., Castagliuolo, I., Zacks,
J.K., Aboudola, S., Qamar, A., Pothoulakis, C., LaMont,
J.T., and Kelly, C.P. 2000. p38 MAP kinase activation by
Clostridium difficile toxin A mediates monocyte necrosis, IL-
lial cells is prevented. Scand. J. Immunol. 45: 613―617.
21) Kim, J.M., Oh, Y.K., Kim, Y.J., Oh, H.B., and Cho, Y.J.
2001. Polarized secretion of CXC chemokines by human
intestinal epithelial cells in response to Bacteroides fragilis
enterotoxin: NF-kB plays a major role in the regulation of IL-
8 production, and enteritis. J. Clin. Invest. 105: 1147―1156.
33) Yang, S.K., Eckmann, L., Panja, A., and Kagnoff, M.F.
1997. Differential and regulated expression of C-X-C, C-C,
and C-chemokines by human colon epithelial cells. Gas-
8 expression. Clin. Exp. Immunol. 123: 421―427.
22) Kim, J.S., Jung, H.C., Kim, J.M., Song, I.S., and Kim, C.Y.
1998. Interleukin-8 expression by human neutrophils acti-
vated by Helicobacter pylori soluble proteins. Scand. J.
troenterology 113: 1214―1223.
34) Yasumoto, K., Okamoto, S., Mukaida, N., Murakami, S.,
Mai, M., and Matsushima, K. 1992. Tumor necrosis factor
alpha and interferon gamma synergistically induce inter-
leukin 8 production in a human gastric cancer cell line
through acting concurrently on AP-1 and NF-κB-like bind-
ing sites of the interleukin 8 gene. J. Biol. Chem. 267:
Gastroenterol. 33: 1249―1255.
23) Kolios, G., Wright, K.L., Jordan, N.J., Leithead, J.B.,
Robertson, D.A., and Westwick, J. 1999. C-X-C and C-C
chemokine expression and secretion by the human colonic
epithelial cell line, HT-29: differential effect of T lymphocyte-
22506―22511.