..................................................................................................................................................................................................
Modulation of Cytokine Release by Differentiated
CACO-2 Cells in a Compartmentalized Coculture Model
with Mononuclear Leucocytes and Nonpathogenic
Bacteria
A. Parlesak*, D. Hallery, S. Brinz*, A. Baeuerlein* & C. Bode*
*Department of Physiology of Nutrition, Institute
for Biological Chemistry and Nutrition,
Hohenheim University (140e), Stuttgart; yCentre
for Nutrition and Food Research, Immunobiology
of Nutrition, Technical University of Munich,
Freising, Germany
Received 23 February 2004; Accepted in revised
form 17 June 2004
Correspondence to: Alexandr Parlesak, PhD,
Hohenheim University (140e), Department of
Physiology of Nutrition, Garbenstraße 28, D-70593
Stuttgart, Germany. E-mail: parlesak@uni-
hohenheim.de
Introduction
Enteric bacteria are in close proximity to the intestinal
epithelium as well as specialized, lymphoid and accessory
cells in the lamina propria and the follicles of the gut-
associated lymphoid tissue (GALT). The maintenance of
local homeostasis in the intestinal mucosa to enteric bac-
teria and their antigens requires a fine-tuned balance
between immune activation and regulation, a process
that, if impaired, may lead to intestinal inflammation
[1, 2]. Intestinal epithelial cells (IEC) are thought to con-
tribute to immunomodulation of mucosal leucocytes by at
least two different mechanisms. Firstly, they separate bac-
teria from intraepithelial lymphocytes and lamina propria
mononuclear cells (LPMC). Secondly, IEC can trigger the
immune response in GALT by the secretion of cytokines,
the constitutive or inducible expression of costimulatory
# 2004 Blackwell Publishing Ltd. Scandinavian Journal of Immunology 60, 477–485
Abstract
To further investigate the interaction between human mononuclear leucocytes
[peripheral blood mononuclear cells (PBMC)] and enterocytes, the effect of a
confluent layer of differentiated CACO-2 cells on cytokine kinetics during
challenge with bacteria in a compartmentalized coculture model was investi-
gated. Nonpathogenic Escherichia coli were added either to the apical or the
basolateral compartment of this transwell cell culture system, the latter of which
contained human leucocytes. The synthesis of tumour necrosis factor (TNF-a)
and interleukin (IL)-12 was significantly suppressed by CACO-2 cells when
leucocytes were stimulated directly with bacteria. This suppression was not
paralleled by changes in the production of IL-10, IL-6 and transforming growth
factor (TGF)-b. When the bacteria were applied apically to the CACO-2 cell
layer, the production of TNF-a, IL-12, IL-1b, IL-8, IL-6, IL-10, TGF-b and
interferon-g was pronouncedly lower as compared to the bacterial stimulation of
leucocytes beneath the CACO-2 cells. In the latter experiments, IL-6, IL-8 and
TNF-a were the cytokines being mostly induced by apical addition of E. coli.
Quantitative mRNA expression analysis revealed that IL-8 gene expression was
equally induced in both CACO-2 and PBMC after apical stimulation with
bacteria. Of note, bacteria-stimulated CACO-2 cells produced little or no
cytokines in the absence of leucocytes, supporting the concept of leucocyte–
epithelial cell cross-talk in modulating cytokine responses in the gut mucosa.
molecules and their capability to present antigens by com-
ponents of the human major histocompatibility complex
[3–7]. We have previously shown that colonization of
germ-free rats with Gram-negative, nonpathogenic, enteric
bacteria leads to an early but transient activation of colonic
epithelial cells [8, 9], demonstrating the physiological rele-
vance for commensal bacteria-induced signalling to the
mucosal epithelium.
Increasing efforts are made on research dealing with
effects of nutrients and drugs on the immune system
affecting the local immune response in the intestine. To
investigate the interaction among bacteria, enterocytes,
and leucocytes, we developed a coculture model in which
a confluent layer of differentiated enterocytes (CACO-2
cells) separates an apical compartment with nonpathogenic
bacteria from a basolateral compartment with
477
478 Coculture of Bacteria, Enterocytes and PBMC
..................................................................................................................................................................................................
mononuclear leucocytes that were isolated from human
peripheral venous blood [peripheral blood mononuclear
cells (PBMC)] [10]. In this model, leucocyte–epithelial
cell cross-talk became evident [8–11].
The enterocyte-like cell line CACO-2 used in the pre-
sent model mimics important properties of the mucosal
epithelium; after confluence, cells stop to proliferate and
spontaneous polarized differentiation occurs [12]. After
differentiation, microvilli, desmosomes and tight junctions
are formed [13], and enzymes of the brush border mem-
brane are expressed [14]. The permeability of a CACO-2
cell monolayer to macromolecules [13] and bacterial pro-
ducts such as endotoxin [10] was shown to be absent or
negligible after confluence and differentiation.
In the present study, we tested the hypothesis whether
or not the presence of IEC affects the cytokine production
of human leucocytes and, if so, to which extent this
immunomodulation occurs. Furthermore, the role of spa-
tial separation of bacteria and leucocytes by the IEC layer
was investigated. For this purpose, the kinetics of 11
different cytokine concentrations were monitored for
64 h (in some experiments up to 72 h) in the basal com-
partment of the described model. Out of these cytokines,
three can be contributed to acute inflammation [tumour
necrosis factor (TNF)-a, interleukin (IL)-1b and IL-8],
three to activation of TH1-type lymphocytes [IL-12, inter-
feron (IFN)-g, IL-2] and two to activation of TH2-type
lymphocytes (IL-4, IL-5) [15]. Furthermore, the kinetics
of three cytokines [IL-6, IL-10 and transforming growth
factor (TGF)-b1] was measured, which can act either
stimulating or suppressive on immunocompetent cells,
depending on the kind and state of cell being affected.
Materials and methods
Cell culture. The human enterocyte-like cell line CACO-2
was obtained from the ‘Deutsche Sammlung von Mikroor-
ganismen und Zellkulturen’ (Braunschweig, Germany:
DSMZ number ACC169) and was cultured in Dulbecco’s
Modified Eagle Medium (DMEM) containing 17% inac-
tivated (56
C for 30 min) fetal calf serum (FCS) and 1%
concentrated nonessential amino acid solution (all
components: Gibco BRL/Life Technologies, Karlsruhe,
Germany). FCS was controlled for endotoxin content
(<0.2 ng/ml: LAL Test, Chromogenix, Mölndal, Sweden).
For transwell cultures, 1.0  106 cells were suspended in
2 ml cell culture medium and cultured in inserts on a
semipermeable polyethylene terephthalate membrane
(4.3 cm2 surface, 0.45 mm pore diameter, 1.6  106
pores/cm2: Falcon, Becton-Dickinson, Le Pont De Claix,
France). Inserts were placed into cavities of six-well plates
(Greiner, Frickenhausen, Germany). During cell growth
and differentiation, medium in both compartments was
changed three times a week. Confluence was controlled by
measurement of transepithelial electrical resistance (TEER;
# 2004 Blackwell Publishing Ltd. Scandinavian Journal of Immunology 60, 477–485
A. Parlesak et al.
Millicell ERS Ohmmeter, Millipore, Eschborn, Germany)
and visual control of cell layer integrity under the micro-
scope. Cells became confluent after 6  1 days. Within this
time, the TEER exceeded the cut-off point of 450 O  cm2
[10]. After complete differentiation, which lasted for further
10  2 days, coincubation experiments were performed.
Bacteria culture. Nonpathogenic Escherichia coli K12
were precultured (37
C) on Luria broth (1% Trypton:
Oxoid, Wesel, Germany; 0.5% yeast extract: Difco,
Heidelberg, Germany; 0.5% NaCl) agar plates. Thereafter,
2  10 ml of liquid Luria broth was inoculated from the
agar plates. After an additional 12 h, 100 ml of the incubation
medium were transferred into fresh Luria broth medium
and incubated further for 24 h to bring the bacteria into
the stationary growth phase. Thereafter, the bacteria sus-
pension was centrifuged and bacteria were washed twice
with phosphate-buffered saline. To obtain the final bac-
teria stock solution, cell concentration was adjusted to
5.0  109 colony-forming units (cfu)/ml in cell culture
medium. Gentamicin (120 mg/ml; Gibco BRL/Life Tech-
nologies) was added to the medium to avoid uncontrolled
growth [10].
Leucocyte preparation. Mononuclear leucocytes from
peripheral venous blood (PBMC) of healthy donors (two
women, eight men; age: 42  10 years) were isolated from
buffy coats (cell-enriched plasma) that were obtained from
the Institute of Blood Transfusion Medicine, Katharinen-
hospital Stuttgart. Equal volumes (15 ml) of cell-enriched
plasma and RPMI 1640 medium were mixed and layered
over 15 ml Ficoll solution (r ¼ 1.077 g/ml; both solutions:
Seromed, Berlin, Germany). After centrifugation (400  g,
30 min), mononuclear leucocytes were taken from the
interface, washed three times with RPMI 1640 medium
and adjusted to 2.0  107 cells/ml.
Cocultivation and stimulation with bacteria. To investi-
gate the interaction among bacteria, differentiated CACO-
2 cells and human leucocytes, six different experimental
approaches were conducted. Each experiment was repeated
10 times in triplicate with PBMC (4.0  106) from 10
healthy donors. In two further experiments, the differen-
tiated CACO-2 cell layer was incubated either with or
without bacteria in the absence of leucocytes. Total
volume in the basal and apical compartment was always
2.0 ml. Bacterial stimulation was performed with E. coli
K12 (2.0  107 cfu per compartment). In detail, experi-
ments were as follows (Fig. 1):
1. To obtain the maximum stimulation of the applied leuco-
cytes under comparable conditions, PBMC were stimu-
lated by direct contact with E. coli K12 (no CACO-2 cells
were present).
2. To assess the effect of CACO-2 cells on bacteria-induced
secretion of cytokines by leucocytes, the E. coli
K12–PBMC suspension was coincubated beneath inserts
with confluent CACO-2 cells.
A. Parlesak et al. Coculture of Bacteria, Enterocytes and PBMC 479
..................................................................................................................................................................................................

A B

C D

Figure 1 Experimental arrangements in the transwell system for the assessment of leucocyte–enterocyte interactions under different kinds of stimulation
by nonpathogenic Escherichia coli. Leucocytes were challenged with bacteria either in the absence of CACO-2 cells (A) or beneath a differentiated
monolayer of these cells (B). CACO-2 cells were stimulated apically with E. coli either in the presence of PBMC (C) or without leucocytes (D).

3. Bacteria in the apical compartment were separated from
PBMC in the basolateral compartment by a confluent
CACO-2 monolayer.
4. To assess the basal secretion of cytokines, PBMC were
cultured beneath the differentiated CACO-2 cells on inserts
in the absence of bacteria (negative control experiments).
5. The contribution of the CACO-2 cells to the production
of cytokines was assessed by stimulating differentiated
CACO-2 cells on inserts with bacteria as described above
in the absence of leucocytes.
6. The basal production of cytokines by CACO-2 cells (with-
out bacteria and leucocytes) and, separately, by PBMC
(without CACO-2 cells and bacteria) was monitored as
well.
The incubations with PBMC were aborted after 4 h, 8 h,
16 h, 24 h, 32 h and 64 h, those without leucocytes after
12 h, 24 h and 72 h. For the estimation of the reduction of
cytokine production by sole spatial separation by the semi-
permeable membrane, bacteria were added to the apical
side of the inserts (without CACO-2 cells) which were
placed above the PBMC suspension. This experiment
was abrogated after 16 h.
Permeability of CACO-2 cell monolayers. To assess the
integrity of the confluent CACO-2 cell layer, fluorescein-
labelled dextran Mr 4400 (Sigma-Aldrich, Steinheim,
Germany) was added to the apical compartment in
experimental approaches no. 3, 4, 5 and 6. Relative perme-
ability of this marker was measured in the basal medium
with a microplate fluorescence reader (BMG, Freiburg,
Germany).
RNA isolation and Light Cycler analysis. RNA was isolated
using Trizol (Invitrogen, Karlsruhe, Germany) and 1 mg of
total RNA was reverse transcribed as described previously
[9]. Quantitative polymerase chain reaction (PCR) was done
in the Light CyclerTM using LC-FastStart DNA Master
KitTM (Roche, Penzberg, Germany) and the following pri-
mers: IL-8-A; (50 ) 5-ATGACTTCCAAGCTGGC-3 and
IL-8-B; (30 ) 5-ACTTCTCCACAACCCT-3, 18SrRNA-A;
(50 ) 5-AAGTCTTTGGGTTCCGGG-3 and 18SrRNA-B
(30 ) 5-GGACATCTAAGGGCATCACA-3. PCR amplifi-
cation produced a 274 and 205 bp product, respectively.
The PCR program was one cycle of denaturation at
95
C for 10 min followed by 50 cycles of 95
C for 15 s,
annealing at 60
C for 10 s and extension at 72
C for 20 s.
The amplified product was detected by the presence of a

# 2004 Blackwell Publishing Ltd. Scandinavian Journal of Immunology 60, 477–485

480 Coculture of Bacteria, Enterocytes and PBMC
..................................................................................................................................................................................................
SYBR1 green fluorescent signal. Melting curve analysis and
gel electrophoresis were used to document the amplicon
specificity. Calibration curves were generated by measuring
serial dilutions of stock cDNA. For the relative quantifica-
tion of IL-8 gene expression in CACO-2 and PBMC, the
crossing point (Cp) of the log-linear portion of the ampli-
fication curve was determined. IL-8/18S ribosomal RNA
ratios were used to normalize the data. The data are pre-
sented as fold increase of IL-8 mRNA in relation to the
nonstimulated controls including CACO-2 and PBMC,
respectively.
Measurement of cytokine concentrations. In all experi-
ments with compartmentalization, cytokines were meas-
ured in the basal medium, because in previous
experiments, relevant cytokine concentrations occurred
only in the basolateral compartment [10]. After centrifu-
gation of the medium (2000  g, 2 min), the supernatant
was frozen in aliquots (80
C). The following cytokines
were measured: TNF-a, IL-1b, IL-2, IL-4, IL-5, IL-6,
IL-8, IL-10, IL-12, IFN-g and TGF-b1. Out of the cyto-
kines measured, TGF-b1 was the only one being already
present in the FCS used for incubation of cells. Therefore,
only the increase in concentration of this cytokine was
regarded in the evaluation. All cytokine concentrations
were measured with OptEIA ELISA sets (BD Pharmingen,
Heidelberg, Germany) according to the instruction of the
manufacturer.
Statistics. If not indicated otherwise, values are given
as means  standard error of the mean (SEM). Before
evaluation of cytokine profiles with ANOVA (within-sub-
ject design, sixfold-repeated measurements), values
were transformed logarithmically to obtain normal dis-
tribution and homogeneity of variances. Experiments
without kinetic measurements were evaluated with a
one-way ANOVA and Tukey’s post hoc test. Only if
statistical difference (P < 0.05) between cytokine con-
centrations occurred, they were termed as ‘higher’ or
‘lower’.
Results
Bacterial stimulation of PBMC
Four different types of cytokine secretion kinetics were
distinguishable in bacteria-stimulated leucocyte cocul-
tures. TNF-a was the only cytokine that reached max-
imal concentration between 8 h and 16 h, and the
concentration of which decreased from that time stea-
dily. IL-1b, IL-10, TGF-b and IL-12 secretion
increased continuously after initial stimulation. Maxi-
mal concentrations of these cytokines were reached after
24–32 h of stimulation and remained relatively constant
until the end of the incubation. The concentration of
# 2004 Blackwell Publishing Ltd. Scandinavian Journal of Immunology 60, 477–485
A. Parlesak et al.
IL-6 and IL-8 increased continuously after bacterial
stimulation, the latter having a biphasic course
(Fig. 2). The onset of the production of the TH1-type
cytokines IFN-g and IL-2 was delayed compared to the
other cytokines. In all experiments, IL-8 and IL-6 were
the dominant mediators reaching by far the highest
concentrations out of the measured cytokines (Fig. 2,
Table 1). Production of cytokines by nonstimulated
PBMC in the absence of CACO-2 cells was either not
detectable [TGF-b (32 h); IL-12 (32 h)] or very low
[TNF-a (12 h): 7.3  3.9 pg/ml; IL-10 (32 h):
5.2  1.0 pg/ml; IL-8 (72 h): 4.48  1.03 ng/ml; IFN-g
(72 h): 23.8  3.3 pg/ml; IL-2 (72 h): 13.4  3.4 pg/
ml)]. IL-4 and IL-5 were not detectable in any of the
samples mentioned above.
Bacterial stimulation of PBMC in the presence of CACO-2 cells
The addition of differentiated CACO-2 cells on transwell
inserts above the PBMC–E.coli suspension resulted in a
significantly reduced production of TNF-a and IL-12
(Fig. 2). Moreover, IL-1b synthesis was transiently sup-
pressed early after bacterial stimulation (12 h). The differ-
entiated CACO-2 cell layer also induced a shift in the
biphasic IL-8 secretion course, demonstrating a delayed
onset of IL-8 production in bacteria-stimulated PBMC
(Fig. 2). In contrast, the kinetics of IL-10, IL-6, TGF-b,
IFN-g and IL-2 were not significantly affected by the
CACO-2 cells, suggesting differential effects of the epithe-
lial cell monolayer on bacteria-induced cytokine secretion
patterns in PBMC.
Apical stimulation of CACO-2 cells by bacteria in the
basolateral presence of PBMC
In experiments where leucocytes and bacteria were separ-
ated by the semipermeable membrane and the confluent
CACO-2 cell layer, the final concentration of all cytokines
mentioned above (besides IL-2) was significantly reduced
as compared to experiments where leucocytes were stimu-
lated by direct contact with bacteria (Fig. 2). Remarkable
reductions were achieved for the production of TNF-a
(8 h: 98%), IL-1b (24 h: 98%), IL-6 (64 h: 96%),
IL-10 (32 h: 97%), IFN-g (64 h: 95%), IL-12 (24 h:
89%) and IL-8 (24 h: 82%). The smallest difference
was obtained for TGF-b (32 h:  45%; Fig. 2). Interest-
ingly and in contrast to all cytokines mentioned above, the
production of IL-2 after 64 h was significantly higher as
compared to experiments where leucocytes had direct con-
tact with bacteria (64 h: þ600%) (Fig. 2).
Of note, in experiments with bacterial challenge of
enterocytes, the concentrations of TNF-a, IL-1b, IL-8,
IFN-g, IL-10 and IL-6 were significantly higher as com-
pared to experiments without bacteria (Table 1). Cytokines
being especially induced by this challenge were IL-6 (32 h:
A. Parlesak et al. Coculture of Bacteria, Enterocytes and PBMC 481
..................................................................................................................................................................................................

6 400

20
5
300
15 4

IL-12 (pg/ml)
TNF-α (ng/ml)

IL-10 (ng/ml)
3 200
10

2
100
5
1

0 0 0
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70

Time (h) Time (h) Time (h)

15 200 5

12 160 4
IL-1β (ng/ml)

IFN-γ (ng/ml)
IL-6 (ng/ml)

9 120 3

6 80 2

3 40 1

0 0 0
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70

Time (h) Time (h) Time (h)

1600
1400 180
Newly synthesized TGF-β1 (pg/ml)

1200 150
1200
1000
120
IL-2 (pg/ml)
IL-8 (ng/ml)

800
800 90
600
60
400 400

200 30

0 0 0
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70

Time (h) Time (h) Time (h)

Figure 2 Concentrations of nine different cytokines [tumour necrosis factor-a, interleukin (IL)-1b, IL-8, IL-10, IL-6, transforming growth factor-b1,
IL-12, interferon-g and IL-2] within the first 64 h of cocultivation of enterocytes, bacteria (Escherichia coli K12) and mononuclear leucocytes [peripheral
blood mononuclear cells (PBMC)]. Bacteria were added either into the apical (&, dashed line) or the basolateral (*, solid line) compartment with respect
to the differentiated CACO-2 cell monolayer. In the third experimental arrangement (*, dotted line), bacteria and PBMC had direct contact and the
CACO-2 cell monolayer was omitted.

about 880-fold), TNF-a (8 h: about 51-fold) and IL-8 both CACO-2 cells (relative fold increase 15.4) and
(24 h: 36-fold). Cytokines being only moderately induced PBMC (relative fold increase 10.4) after 12 h of stimula-
by apical stimulation in the compartmentalized model tion. Nevertheless, Cp values significantly differed in
were IL-12, IL-2 and TGF-b (Table 1). CACO-2 cells (Cp ¼ 24.0) and PBMC (Cp ¼ 18.3),
To investigate the cellular source of cytokines, we cocul- suggesting that the absolute quantity of IL-8 mRNA is
tured CACO-2 with PBMC from five different donors much higher in PBMC. After 4 h of bacterial stimulation,
and stimulated the CACO-2 cell monolayer with E. coli IL-8 mRNA expression was only induced in PBMC (rela-
K12 for 4 and 12 h. Quantitative analysis of IL-8 mRNA tive fold increase 3.1) but not in CACO-2 cells (relative
expression in CACO-2 cells and PBMC was performed by fold increase 0.9).
using Light Cycler analysis. As shown in Fig. 3, IL-8 In control experiments where bacteria and PBMC were
mRNA expression was comparably strongly induced in separated by the semipermeable membrane only (without

# 2004 Blackwell Publishing Ltd. Scandinavian Journal of Immunology 60, 477–485

482 Coculture of Bacteria, Enterocytes and PBMC A. Parlesak et al.
..................................................................................................................................................................................................

Table 1 Time course of the concentration of nine cytokines in the basolateral compartment of the coculture model in the presence of PBMC from 10
donors with and without Escherichia coli K12 used as an apical challenge of the CACO-2 cell monolayer

Cytokine Challenge (yes/no) 4h 8h 16 h 24 h 32 h 64 h

TNF-a (pg/ml) Control 31 53 18  8 12  3 82 51
E. coli K12 123  27 257  32 221  35 76  14 27  7 92
IL-1b (pg/ml) Control 80  17 78  12 51  14 87  23 70  7 66  8
E. coli K12 62  22 63  19 230  78 206  62 259  59 158  58
IL-8 (ng/ml) Control 0.66  0.23 1.37  0.30 8.79  4.50 3.26  0.85 3.14  0.94 4.56  0.99
E. coli K12 3.20  0.80 9.23  1.73 50.1  10.8 118  27.7 86.0  13.5 157  18.7
IL-12 (pg/ml) Control ND ND ND ND ND ND
E. coli K12 15  4 93 12  4 19  4 12  3 15  5
IFN-g (pg/ml) Control 73 73 82 10  1 13  3 16  4
E. coli K12 51 51 12  3 20  5 31  6 156  33
IL-2 (pg/ml) Control 20  4 22  5 32  3 41  9 46  10 98  24
E. coli K12 41 72 27  5 42  6 50  8 147  25
IL-10 (pg/ml) Control 31 82 23  13 16  5 72 14  4
E. coli K12 61 27  4 119  19 164  25 139  20 129  21
IL-6 (pg/ml) Control 25  12 16  4 11  8 81  53 12  5 22  6
E. coli K12 380  103 2240  520 8210  1880 11640  2950 10520  2220 9320  2260
TGF-b1 (pg/ml) Control 171  53 181  64 278  62 317  81 479  107 342  114
E. coli K12 237  44 254  45 329  49 520  88 582  97 600  119

TNF, tumour necrosis factor; IL, interleukin; TGF, transforming growth factor; ND, not detectable.

CACO-2 cells), the synthesis of TNF-a and IL-10 were separated by the confluent CACO-2 cell layer
(6.64  1.06 ng/ml and 1.58  0.20 ng/ml, respectively) (Table 1).
after 16 h was about 38 and 65% compared to that pro-
duced by suspensions of leucocytes and bacteria without
CACO-2 cells, but still 30 and 13 times higher, respect- Bacterial stimulation of CACO-2 cells in the absence of PBMC
ively, compared to incubations where PBMC and bacteria In the absence of leucocytes, the nonchallenged CACO-2
cell layer secreted negligible amounts of TNF-a, IL-10,
IL-6 and IL-8 into the basal compartment (Fig. 4). The
22 concentration of all these cytokines increased significantly
20 after apical stimulation of the enterocyte-like cells with
18 bacteria (Fig. 4), but the concentrations remained far
16 below those occurring after bacterial challenge of the apical
(Fold increase)

14 surface of the CACO-2 cell layer in the basolateral pre-

IL-8 mRNA

12 sence of leucocytes (Table 1). After 72 h, IL-2 and IFN-g

10 were below detection limit in incubations without leuco-
8 cytes in the basolateral compartment.
6
4
2 Transepithelial electrical resistance/permeability to dextran Mr
0 4400
4 12 4 12 Time (h)
TEER of the CACO-2 cell monolayer increased during
CACO-2 PBMC development of confluence and differentiation by
641  85 O  cm2. The lowest increase (522 O  cm2)
Figure 3 Quantitative mRNA expression analysis for interleukin (IL)-8 in
CACO-2 cells and peripheral blood mononuclear cells (PBMC). CACO- clearly exceeded the suggested cut point of 450 O  cm2
2 cells were stimulated with Escherichia coli K12 for 4 h and 12 h in the [10]. The microscopic investigation of the basolateral
presence of PBMC in the basolateral compartment of the transwell cell medium revealed that neither bacteria nor CACO-2 cells
culture system. RNA was isolated and reverse transcribed as described in were able to migrate through the PET membrane. The
Material and methods. IL-8 mRNA expression was quantified using Light
integrity of the CACO-2 cell monolayer remained fairly
Cycler analysis and presented as fold increase relative to nonstimulated
control cells. The data in the bar chart represent the mean  SEM from stable for at least 24 h, as assessed by its permeability to
five different experiments with PBMC from five different healthy dextran Mr 4400 (Fig. 5). Surprisingly, the permeability of
volunteers. the CACO-2 cell layer increased significantly after 72 h

# 2004 Blackwell Publishing Ltd. Scandinavian Journal of Immunology 60, 477–485

A. Parlesak et al.
..................................................................................................................................................................................................
300
250
Concentration (pg/ml)
200
150
100
50
0
TNF (12 h) IL-10 (24 h) IL-6 (24 h)
Figure 4 Secretion of tumour necrosis factor-a, interleukin (IL)-10, IL-6
and IL-8 by the CACO-2 layer without and with bacteria on the apical
surface after 12 h/24 h in experiments without leucocytes (*P < 0.05).
only if both bacteria and PBMC were present, but not
after the addition of bacteria alone.
Discussion
We have previously shown that nonpathogenic commensal
enteric bacteria have the potential to initiate innate host
responses in IEC in vivo and in vitro [8, 9]. To mimic the
complex interaction of the epithelium with adjacent
immune cells in the intestinal mucosa, we established a
reductionist epithelial–immune cell coculture system,
demonstrating a role for PBMC in initiating and regulat-
ing bacteria-mediated activation of IEC [8–11]. Previous
studies showed that IEC profoundly inhibit monocyte
activation [11, 16, 17] and T-cell function [18] through
both secreted mediators and cell–cell interactions. In the
present study, we show immunosuppressive effects of
CACO-2 cells on E. coli-stimulated PBMC. The secretion
of the pro-inflammatory cytokines TNF-a, IL-12 and, to
some extent, IL-1b was significantly inhibited. On the
other hand, cytokines such as IL-10 and TGF-b1, which
can suppress the synthesis of pro-inflammatory cytokines
[19–21], were not significantly affected by the presence of
CACO-2 cells. Direct contact between CACO-2 cells and
PBMC was prevented by the use of 0.4 mm pore size
membranes in transwell inserts, suggesting that immuno-
suppressive factors were secreted from CACO-2 cells.
However, from the present experiments, the factor respon-
sible for the suppression of TNF-a, IL-12 and IL-1b
synthesis remains unknown.
Translocated bacteria and their products are thought to
participate in initiating and perpetuating chronic mucosal
inflammation [22–24], suggesting that the direct bacterial
# 2004 Blackwell Publishing Ltd. Scandinavian Journal of Immunology 60, 477–485
Coculture of Bacteria, Enterocytes and PBMC 483
8
*
Permeability of dextran (%)
6
4
2
0
0 20 40 60 80
Time (h)
IL-8 (24 h)
Figure 5 Relative permeability of fluorescein isothiocyanate-labelled
dextran Mr 4400 through the CACO-2 cell monolayer over 72 h. The
permeability was measured in the CACO-2 cell layer without other cells
(*, dotted line), in the CACO-2 cell layer with peripheral blood
mononuclear cells (PBMC) in the basal compartment (&, dashed line), in
the CACO-2 cell layer with bacteria on the apical side (, dashes and
dots) and in the CACO-2 cell layer with bacteria on the apical side and
PBMC in the basal medium (&, solid line) (*P < 0.05).
stimulation of PBMC in our model system may reflect
more pathological conditions. Consequently, with the
exception of IL-2, IL-4 and IL-5, the separation of bacteria
and PBMC by the confluent layer of differentiated
CACO-2 cells resulted in a pronounced reduction of
cytokine secretion into the basolateral compartment
(Fig. 2). Only a small part of this effect can be contributed
to sole spatial separation by the semipermeable PET mem-
brane, which reduced the production of TNF-a and IL-10
by the factor of 2-3 (after 16 h). Owing to the specifica-
tions of the manufacturer, the semipermeable PET mem-
brane has pores with a diameter <0.4 mm, which does not
allow the permeation of intact bacteria through the mem-
brane, but enables the diffusion of bacterial fragments and
products. This diffusion seems to be strongly reduced by
the CACO-2 cell monolayer, as the permeation of mol-
ecules with a molecular mass of Mr 4400 is clearly limited,
at least within the first 24 h of incubation (Fig. 5). There-
fore, the suppression of diffusion of macromolecular pro-
ducts from the apical compartment is likely to contribute
substantially to the reduction of cytokine production
resulting from the separation of bacteria and leucocytes
by the CACO-2 cell monolayer. The combined barrier of
membrane and CACO-2 cells reduced the synthesis of
these cytokines by factors up to 82 (TNF-a), supporting
the concept that the intestinal epithelium contributes to
maintain an immunosuppressive environment in the nor-
mal gut [25]. Surprisingly, neither the apical addition of
bacteria alone nor the basal addition of PBMC alone
484 Coculture of Bacteria, Enterocytes and PBMC
..................................................................................................................................................................................................
seemed to affect the integrity of the CACO-2 cell mono-
layer within the applied incubation time (Fig. 5). The
increase in permeability in the presence of both bacteria
and PBMC after 72 h was paralleled by an increase of the
concentration of IL-8, TNF-a and IFN-g. In previous
studies, these cytokines were shown to impair epithelial
integrity [26–29], which makes them possible mediators of
the increased permeability measured in the present study.
We have previously shown that Gram-negative bacteria
induce nuclear factor (NF)-kB signalling and kB-depend-
ent gene expression in IEC. Considering that the stimula-
tion of CACO-2 cells with bacteria in the presence of
PBMC still induced significant IL-8 and IL-6 protein
production and that both genes are regulated at least in
part by the transcription factor NF-kB [30] and are
expressed in IEC [9, 10], we next evaluated the cellular
source of cytokine production by performing quantitative
mRNA expression analysis in CACO-2 and PBMC. Inter-
estingly, IL-8 gene expression was induced in CACO-2
and PBMC at an early stage of cocultivation (12 h), the
total amount of IL-8 mRNA being distinctly higher in
PBMC. Taken together, the complex nature of leucocyte–
epithelial cell cross-talk makes it difficult to speculate
about the absolute contribution of each cell type in the
production of IL-8 protein in the coculture model. Con-
versely, CACO-2 cells are not capable of IFN-g and IL-2
production within the applied model [10], and the
presence of PBMC did not induce synthesis of the latter
cytokines in IEC (data not shown).
Enterocyte-like cell lines were used as models to obtain
important information about the involvement of IEC in
the immune response of the intestinal mucosa, especially
to evaluate the pathology of enteropathogens [4, 31, 32]. It
is our understanding that epithelial–immune cell interac-
tions contribute to the homeostasis of normal intestinal
mucosa [25]. In the present model, although some tissue-
specific cells present in the GALT such as lamina propria
lymphocytes, intraepithelial lymphocytes, mast cells, den-
dritic cells and other cells are missing, the model allows the
assessment of leucocyte–enterocyte interaction. Previous
studies showed that freshly isolated nonstimulated LPMC
secrete IL-6 (approximately 5 ng/106 cells/48 h of incuba-
tion) [33], TNF-a (215 pg/106 cells/24 h) [34] and IL-8
(approximately 4.5 ng/106 cells/24 h) [35] in similar quan-
tities compared to the concentrations measured in the
basolateral compartment of CACO-2/PBMC cocultures
after apical stimulation with bacteria (Table 1). Our recent
data also indicate similar function of LPMC and PBMC in
the regulation of epithelial cell responses to bacterial sti-
muli ex vivo in the coculture system [36]. However, the
processes within the GALT leading to specific immunity
cannot be mimicked by the present model. This holds true
especially for the production of IL-2 which is generally
considered to require specific T-cell receptor activation.
According to the negligible permeation of the permeability
# 2004 Blackwell Publishing Ltd. Scandinavian Journal of Immunology 60, 477–485
A. Parlesak et al.
marker through the layer of CACO-2 cells, a transport
of intact bacterial antigens through this barrier seems
unlikely. Hence, the mechanism for the enhanced IL-2
production by PBMC in the presence of CACO-2
cells remains unknown. Noteworthy in this context are
the occurrence of intestinal inflammation in mice with a
disrupted IL-2 gene [37] and the promotion of intestinal
cell restitution by IL-2 [38]. The current opinion that
activation of TH1 cell results in the parallel shedding of
characteristic cytokines (IL-12, IL-2 and IFN-g) [15]
seems not to be met within the presented model, as IL-2
on the one hand and IFN-g together with IL-12 on the
other are differently regulated.
Evidently, the applied strain of E. coli can be considered as
a standard stimulus only and cannot meet the complex
properties of the intestinal microflora. Therefore, long-term
interactions between intestinal flora and the epithelium of
the intestine, which are also affected by bacterial metabolism,
cannot be investigated in the described model, mainly
because the intestinal flora cannot be cultivated outside the
gut for obvious reasons. Instead, in the present model, a
standardized, short-termed immunological stimulation
occurs, which regards leucocyte–enterocyte interactions, the
modulation of which by drugs and nutrients can be
measured.
Acknowledgments
We thank Dr Luz from the Institute of Blood Transfusion
Medicine, Katharinenhospital Stuttgart, who made the
buffy coats available. We are indebted to Daniela Distel
for culturing cells and bacteria and to Susanne Scharr and
Sabine Fischer for their support in the measurement of
cytokines. This work was supported by the Wissenschafts-
förderung der Deutschen Brauwirtschaft e. V. (Project
B67) and by the young investigator award for Dirk Haller
in the Emmy Noether program by the Deutsche For-
schungsgemeinschaft (HA 3148/1-2).
References
1 Blumberg RS, Saubermann LJ, Strober W. Animal models of muco-
sal inflammation and their relation to human inflammatory bowel
disease. Curr Opin Immunol 1999;11:648–56.
2 Sartor RB. Pathogenesis and immune mechanisms of chronic
inflammatory bowel diseases. Am J Gastroenterol 1997;92 (12
Suppl.):5S–11S.
3 Sartor RB. Cytokines in intestinal inflammation: pathophysiological
and clinical considerations. Gastroenterology 1994;106:533–9.
4 Kagnoff MF, Eckmann L. Epithelial cells as sensors for microbial
infection. J Clin Invest 1997;100:6–10.
5 Blumberg RS, Terhorst C, Bleicher P et al. Expression of a
nonpolymorphic MHC class I-like molecule, CD1D, by human
intestinal epithelial cells. J Immunol 1991;147:2518–24.
6 Panja A, Barone A, Mayer L. Stimulation of lamina propria lym-
phocytes by intestinal epithelial cells: evidence for recognition of
nonclassical restriction elements. J Exp Med 1994;179:943–50.
A. Parlesak et al.
..................................................................................................................................................................................................
7 Hershberg RM, Framson PE, Cho DH et al. Intestinal epithelial
cells use two distinct pathways for HLA class II antigen processing. J
Clin Invest 1997;100:204–15.
8 Haller D, Russo MP, Sartor RB, Jobin C. IKK beta and phospha-
tidylinositol 3-kinase/Akt participate in non-pathogenic Gram-nega-
tive enteric bacteria-induced RelA phosphorylation and NF-kappa B
activation in both primary and intestinal epithelial cell lines. J Biol
Chem 2002;277:38168–78.
9 Haller D, Holt L, Kim SC, Schwabe RF, Sartor RB, Jobin C.
Transforming growth factor-beta 1 inhibits non-pathogenic Gram
negative bacteria-induced NF-kappa B recruitment to the interleu-
kin-6 gene promoter in intestinal epithelial cells through modulation
of histone acetylation. J Biol Chem 2003;278:23851–60.
10 Haller D, Bode C, Hammes WP, Pfeifer AM, Schiffrin EJ, Blum S.
Non-pathogenic bacteria elicit a differential cytokine response by
intestinal epithelial cell/leucocyte co-cultures. Gut 2000;47:79–87.
11 Haller D, Serrant P, Peruisseau G et al. IL-10 producing CD14low
monocytes inhibit lymphocyte-dependent activation of intestinal
epithelial cells by commensal bacteria. Microbiol Immunol
2002;46:195–205.
12 Chantret I, Barbat A, Dussaulx E, Brattain MG, Zweibaum A.
Epithelial polarity, villin expression, and enterocytic differentiation
of cultured human colon carcinoma cells: a survey of twenty cell
lines. Cancer Res 1988;48:1936–42.
13 Hidalgo IJ, Raub TJ, Borchardt RT. Characterization of the human
colon carcinoma cell line (CACO-2) as a model system for intestinal
epithelial permeability. Gastroenterology 1989;96:736–49.
14 Pinto M, Robine-Leon S, Appay M-D et al. Enterocyte-like differ-
entiation and polarization of the human colon carcinoma cell line
CACO-2 in culture. Biol Cell 1983;47:323–30.
15 Mosmann TR, Cherwinski H, Bond MW, Giedlin MA,
Coffman RL. Two types of murine helper T cell clone. I. Definition
according to profiles of lymphokine activities and secreted proteins. J
Immunol 1986;136:2348–57.
16 Nathens AB, Rotstein OD, Dackiw AP, Marshall JC. Intestinal
epithelial cells down-regulate macrophage tumor necrosis factor-
alpha secretion: a mechanism for immune homeostasis in the gut-
associated lymphoid tissue. Surgery 1995;118:343–50.
17 Spottl T, Hausmann M, Kreutz M et al. Monocyte differentiation in
intestine-like macrophage phenotype induced by epithelial cells. J
Leukoc Biol 2001;70:241–51.
18 Christ AD, Colgan SP, Balk SP, Blumberg RS. Human intestinal
epithelial cell lines produce factor(s) that inhibit CD3-mediated
T-lymphocyte proliferation. Immunol Lett 1997;58:159–65.
19 de Waal Malefyt R, Abrams J, Bennett B, Figdor CG, de Vries JE.
Interleukin 10 (IL-10) inhibits cytokine synthesis by human mono-
cytes: an autoregulatory role of IL-10 produced by monocytes. J Exp
Med 1991;174:1209–20.
20 Fiorentino DF, Zlotnik A, Mosmann TR, Howard M, O’Garra A.
IL-10 inhibits cytokine production by activated macrophages. J
Immunol 1991;147:3815–22.
21 Chantry D, Turner M, Abney E, Feldmann M. Modulation of
cytokine production by transforming growth factor-beta. J Immunol
1989;142:4295–300.
# 2004 Blackwell Publishing Ltd. Scandinavian Journal of Immunology 60, 477–485
Coculture of Bacteria, Enterocytes and PBMC 485
22 Sedman PC, Macfie J, Sagar P et al. The prevalence of gut transloca-
tion in humans. Gastroenterology 1994;107:643–9.
23 Duchmann R, May E, Heike M, Knolle P, Neurath M, Meyer
zum Buschenfelde KH. T cell specificity and cross reactivity
towards enterobacteria, bacteroides, bifidobacterium, and antigens
from resident intestinal flora in humans. Gut 1999;44:812–8.
24 Sartor RB. The influence of normal microbial flora on the develop-
ment of chronic mucosal inflammation. Res Immunol
1997;148:567–76.
25 Strober W. Interactions between epithelial cells and immune cells in
the intestine. Ann N Y Acad Sci 1998;859:37–45.
26 Sansonetti PJ, Arondel J, Huerre M, Harada A, Matsushima K.
Interleukin-8 controls bacterial transepithelial translocation at the
cost of epithelial destruction in experimental shigellosis. Infect
Immun 1999;67:1471–80.
27 Seydel KB, Li E, Zhang Z, Stanley SL Jr. Epithelial cell-initiated
inflammation plays a crucial role in early tissue damage in amebic
infection of human intestine. Gastroenterology 1998; 115:1446–53.
28 Zareie M, McKay DM, Kovarik GG, Perdue MH. Monocyte/
macrophages evoke epithelial dysfunction: indirect role of tumor
necrosis factor-alpha. Am J Physiol 1998;275:C932–9.
29 Satake M, Watanabe H, Miyamoto Y, Shimizu M. Induction of
nitric oxide synthase and subsequent production of nitric oxide not
involved in interferon-gamma-induced hyperpermeability of
CACO-2 intestinal epithelial monolayers. Biosci Biotechnol Bio-
chem 2001;65:428–30.
30 Pahl HL. Activators and target genes of Rel/NF-kappaB transcrip-
tion factors. Oncogene 1999;18:6853–66.
31 Eckmann L, Jung HC, Schurer-Maly C, Panja A, Morzycka-
Wroblewska E, Kagnoff MF. Differential cytokine expression by
human intestinal epithelial cell lines. regulated expression of inter-
leukin 8. Gastroenterology 1993;105:1689–97.
32 Jung HC, Eckmann L, Yang SK et al. A distinct array of proin-
flammatory cytokines is expressed in human colon epithelial cells in
response to bacterial invasion. J Clin Invest 1995;95:55–65.
33 Hosokawa T, Kusugami K, Ina K et al. Interleukin-6 and soluble
interleukin-6 receptor in the colonic mucosa of inflammatory bowel
disease. J Gastroenterol Hepatol 1999;14:987–96.
34 Noguchi M, Hiwatashi N, Liu Z, Toyota T. Secretion imbalance
between tumour necrosis factor and its inhibitor in inflammatory
bowel disease. Gut 1998;43:203–9.
35 Daig R, Rogler G, Aschenbrenner E et al. Human intestinal epithe-
lial cells secrete interleukin-1 receptor antagonist and interleukin-8
but not interleukin-1 or interleukin-6. Gut 2000;46:350–8.
36 Haller D, Holt L, Parlesak A et al. Differential effect of immune cells
on non-pathogenic Gram-negative bacteria-induced nuclear factor-
kappaB activation and pro-inflammatory gene expression in intest-
inal epithelial cells. Immunology 2004;112:310–20.
37 Sadlack B, Merz H, Schorle H et al. Ulcerative colitis-like disease in
mice with a disrupted interleukin-2 gene. Cell 1993;75:253–61.
38 Cario E, Becker A, Sturm A et al. Peripheral blood mononuclear
cells promote intestinal epithelial restitution in vitro through an
interleukin-2/interferon-gamma-dependent pathway. Scand J
Gastroenterol 1999;34:1132–8.