Neutrophil-epithelial crosstalk at the intestinal lumenal surface mediated by reciprocal secretion of adenosine and IL-6

Abstract

Adenosine is formed in the intestinal lumen during active inflammation from neutrophil-derived 5' AMP. Using intestinal epithelial cell line T84, we studied the effect of adenosine on the secretion of IL-6, a proinflammatory cytokine involved in neutrophil degranulation and lymphocyte differentiation. Stimulation of T84 monolayers with either apical or basolateral adenosine induces A2b receptor-mediated increase in IL-6 secretion, which is polarized to the apical (luminal) compartment. In addition, Salmonella typhimurium, TNF-alpha, and forskolin, known inducers of IL-6 secretion in intestinal epithelial cells, also stimulate IL-6 secretion into the apical compartment. We show that IL6 promoter induction by adenosine occurs through cAMP-mediated activation of nuclear cAMP-responsive element-binding protein (CREB). We also show that IL-6 released in the luminal (apical) compartment achieves a sufficient concentration to activate neutrophils (from which the adenosine signal originates), since such IL-6 is found to induce an intracellular [Ca(++)] flux in neutrophils. We conclude that adenosine released in the intestinal lumen during active inflammation may induce IL-6 secretion, which is mediated by cAMP/CREB activation and occurs in an apically polarized fashion. This would allow sequential activation of neutrophil degranulation in the lumen -- a flow of events that would, in an epithelium-dependent fashion, enhance microbicidal activity of neutrophils as they arrive in the intestinal lumen.

Full text

Introduction
Adenosine is a critical extracellular signaling molecule
that is produced by various cells during normal meta-
bolic activity and released in a regulated fashion (1–3).
For example, adenosine is released during inflamma-
tory conditions and acts as a paracrine factor with
diverse effects on a variety of organ systems including
cardiovascular, nervous, urogenital, respiratory, and
digestive systems (2, 4, 5). Once released, adenosine
interacts with an adenosine receptor belonging to the
family of seven-transmembrane G protein–coupled
cell-surface receptors (2, 4). On the basis of initial
pharmacological criteria, adenosine was thought to
interact with one of several adenosine-receptor sub-
types A1, A2a, A2b, and A3 (2, 4, 6, 7), all of which
have now been cloned. One of the well-known effects
of adenosine is the modulation of inflammation in
many tissues through its potent and selective regula-
tion of proinflammatory or anti-inf lammatory
cytokine production. For example, adenosine induces
IL-10 and suppresses IL-12 secretion in monocytes
(8), induces IL-8 release by mast cells (9), and stimu-
lates IL-6 secretion by astrocytes (10).
In the intestine, neutrophil transepithelial migration
and arrival in the lumen to form crypt abscesses is the
pathologic hallmark of the active phase of many intes-
tinal disorders, including inflammatory bowel disease
(1, 10). We have shown previously that neutrophils,
upon transmigration into the intestinal lumen, release
5′AMP (1, 10). Adenosine is derived from 5′AMP when
it is converted by the intestinal apical membrane 5′
ectonucleotidase (CD 73) (11) and can subsequently
interact with the intestinal adenosine receptor. In this
way, as neutrophils arrive in the lumen from which a
pathogenic microbial threat originates, these cells not
only directly defend the surface but also stimulate a
secretory flush. Other biological roles of stimulation of
A2b receptors in intestinal lumen are not known. Using
molecular, pharmacologic, and biochemical approach-
es, we characterized the intestinal adenosine receptor
as the A2b subtype in both T84 cells, a model intestin-
al epithelial cell line, and intact human intestinal
epithelia (12). Furthermore, the A2b receptor appears
to be the only adenosine receptor present in T84 cells
and is present in both apical and basolateral mem-
branes (12–14). The A2b receptor is functionally cou-
The Journal of Clinical Investigation | April 2001 | Volume 107 | Number 7 861
Neutrophil-epithelial crosstalk at the intestinal
lumenal surface mediated by reciprocal secretion
of adenosine and IL-6
Shanthi V. Sitaraman,1,2 Didier Merlin,1Lixin Wang,1Michelle Wong,1
Andrew T. Gewirtz,1Mustapha Si-Tahar,1and James L. Madara1
1Epithelial Pathobiology Unit, Department of Pathology, and
2Division of Digestive Diseases, Department of Medicine, Emory University, Atlanta, Georgia, USA
Address correspondence to: Shanthi V. Sitaraman, 1639 Pierce Drive, Room 2101 WMRB, Atlanta, Georgia 30322, USA.
Phone: (404) 727-2430; Fax: (404) 727-5767; E-mail: ssitar2@emory.edu.
Mustapha Si-Tahar’s present address is: Unité de Pharmacologie Cellulaire,
Unité Associée/Institut National de la Santé et de la Recherche Médicale 285, Institut Pasteur, Paris, France.
Received for publication November 14, 2000, and accepted in revised form February 7, 2001.
Adenosine is formed in the intestinal lumen during active inflammation from neutrophil-derived
5′AMP. Using intestinal epithelial cell line T84, we studied the effect of adenosine on the secretion
of IL-6, a proinflammatory cytokine involved in neutrophil degranulation and lymphocyte differ-
entiation. Stimulation of T84 monolayers with either apical or basolateral adenosine induces A2b
receptor–mediated increase in IL-6 secretion, which is polarized to the apical (luminal) compart-
ment. In addition, Salmonella typhimurium, TNF-α, and forskolin, known inducers of IL-6 secretion
in intestinal epithelial cells, also stimulate IL-6 secretion into the apical compartment. We show that
IL6 promoter induction by adenosine occurs through cAMP-mediated activation of nuclear cAMP-
responsive element-binding protein (CREB). We also show that IL-6 released in the luminal (apical)
compartment achieves a sufficient concentration to activate neutrophils (from which the adeno-
sine signal originates), since such IL-6 is found to induce an intracellular [Ca++] flux in neutrophils.
We conclude that adenosine released in the intestinal lumen during active inflammation may induce
IL-6 secretion, which is mediated by cAMP/CREB activation and occurs in an apically polarized fash-
ion. This would allow sequential activation of neutrophil degranulation in the lumen — a flow of
events that would, in an epithelium-dependent fashion, enhance microbicidal activity of neutrophils
as they arrive in the intestinal lumen.
J. Clin. Invest. 107:861–869 (2001).

pled to Gαs and stimulation of the apical or basolater-
al surface with adenosine results in increased cAMP
(12). After preliminary observations, we have made use
of the apical and basolateral expression of the A2b
receptor in T84 cells to examine whether adenosine
induces the secretion of IL-6, a proinflammatory
cytokine that is induced by the cAMP-mediated signal-
ing pathway, and whether this induction is polarized.
IL-6 is a 21- to 28-kDa glycoprotein that, under spec-
ified conditions, may be secreted by monocytes,
macrophages, lymphocytes, and epithelial cells, includ-
ing the intestinal epithelial cells (15, 16). Expression of
IL-6 is important for the host response to a number of
infections, and excessive secretion of IL-6 is thought to
contribute to the pathogenesis of many diseases,
including rheumatoid arthritis and inflammatory
bowel disease. IL-6 receptors are present in a variety of
cells including monocyte, macrophage, lymphocyte,
neutrophil, and epithelial cells (including intestinal
epithelial cells) (15–17).
IL-6 is present in very high levels in both serum and
intestinal tissue from patients with Crohn’s disease
(18–20). Il-6 has been shown to play major role in the
pathogenesis of inflammatory bowel disease and is
required for the development of Th1 cell–mediated
colitis (22, 23). Intestinal epithelial cells and lamina
propria mononuclear cells have been shown to be the
major source of IL-6 seen in inflammatory bowel dis-
ease (21). In addition to inflammatory bowel disease,
IL-6 has been shown to play a major role in the induc-
tion of B-cell differentiation, monocyte proliferation,
and neutrophil recruitment to sites of inflammation
(15). IL-6 has been also shown to interact with recep-
tors on endothelial cells and decrease endothelial bar-
rier function (24). In epithelial cells, IL-6 expression has
been shown to be induced by various cytokines such as
IL-1β(25), TNF-α(26), chemokines (17, 27), and by Sal-
monella typhimurium (28). It is not known if IL-6 is
secreted by epithelial cells in a polarized fashion.
We demonstrate here that adenosine released in the
intestinal lumen (as provided by the transmigrating
neutrophils) via activation of intestinal A2b receptors
causes substantial and polarized IL-6 secretion. In con-
trast with secretion of many other cytokines exempli-
fied by IL-8, IL-6 secretion is polarized to the apical
compartment regardless of whether the cells are stim-
ulated by apical or basolateral adenosine. We also show
that intestinal IL-6 secretion is transcriptionally medi-
ated by activated transcription factor (ATF) and cAMP-
responsive element binding protein (CREB) elements.
Finally, IL-6 can stimulate [Ca++] signaling in neu-
trophils, presumably the basis for the degranulating
effect of this cytokine. These data demonstrate, for the
first time to our knowledge, polarized release of an
epithelial proinflammatory cytokine into the luminal
compartment. They also suggest that the activation of
adenosine A2b receptors may provide epithelial-derived
paracrine signals to neutrophils positioned apically
after transmigration. Paracrine signaling of luminal
immune cells provides an additional means of regula-
tion of inflammatory responses by intestinal epithelia.
Methods
Reagents. All tissue-culture supplies were obtained from
Life Technologies (Grand Island, New York, USA).
Adenosine and 5′-(N-ethylcarboxamido) adenosine
(NECA) were obtained from Research Biochemicals
International (Natick, Massachusetts, USA). Forskolin
and carbachol were obtained from Sigma Chemical Co.
(St. Louis, Missouri, USA). Reagents for SDS-PAGE and
nitrocellulose membranes (0.45-µM pores) were from
Bio-Rad Laboratories Inc. (Hercules, California, USA).
Anti-CREB and anti–phospho-CREB were purchased
from New England Biolabs Inc. (Beverly, Massachusetts,
USA). Other Ab’s include FITC-labeled goat anti-rabbit
Ab and peroxidase-conjugated mouse anti-rabbit Ab
obtained from Pierce Chemical Co. (Rockford, Illinois,
USA) and New England Biolabs Inc., respectively. IL-6
and monoclonal anti–IL-6 was obtained from R&D Sys-
tems Inc. (Minneapolis, Minnesota, USA).
Cell culture. T84 cells (American Type Culture Col-
lection, Rockville, Maryland, USA) were grown and
maintained in culture as described previously (10) in a
1:1 mixture of DMEM and Ham’s F-12 medium sup-
plemented with penicillin (40 mg/l), ampicillin (8
mg/l), streptomycin (90 mg/l), and 5% newborn calf
serum. Confluent monolayers were prepared as
described previously (1). Experiments were done on
cells plated for 7–8 days on permeable supports of 0.33
cm2(inserts). Inserts with rat tail collagen–coated
polycarbonate membrane filter (5-µm pore size; Corn-
ing-Costar Corp., Cambridge, Massachusetts, USA)
rested in wells containing media until steady-state
resistance was achieved, as described previously (1).
This permits apical and basolateral membranes to be
separately interfaced with apical and basolateral
buffer, a configuration identical to that developed pre-
viously for various microassays (1). The T84 cells had
a high electrical resistance (1,200–1,500 Ω·cm2). All
experiments were performed on T84 cells between pas-
sages 69–83. Cos-7 cells (American Type Culture Col-
lection) were grown in DMEM containing 10% FCS
supplemented with penicillin and streptomycin.
IL-6 assay. Cells were placed in HBSS (250 µl apical
and 300 µl basolateral) and apical and basolateral solu-
tion was collected at various times after stimulation
and passed through 0.33-µm filters to remove debris.
IL-6 was measured using ELISA (Quantikine IL-6
immunoassay kit; R&D Systems Inc.).
Plasmids, transient transfections, and chloramphenicol
acetyl transferase assays. Various IL-6 promoter con-
structs were the generous gift of J.L. Harcourt and
M.K. Offerman (Department of Medicine, Winship
Cancer Center, Emory University, Atlanta, Georgia,
USA) and have been characterized previously (29). The
wild-type construct is the full-length IL-6 promoter
–15 to –435 subcloned into pCAT3-Basic (Promega
Corp., Madison, Wisconsin, USA). Mutants were gen-
862 The Journal of Clinical Investigation | April 2001 | Volume 107 | Number 7

erated using wild-type IL-6 promoter-chlorampheni-
col acetyl transferase (IL-6 promoter–CAT) in the
NF–IL-6 site, ATF/CRE site, or NF-κB sites. Cells were
transiently transfected using DEAE-dextran. Briefly,
COS 7 cells were passaged from one p100 to four six-
well plates at a density of 2 ×105cells/well for trans-
fection the following day. The plasmid (5 µg) was
diluted in 380 µl of 10 mM Tris-HCl, pH 7.5, contain-
ing 100 mM NaCl. Twenty microliters of DEAE-dex-
tran (10 mg/ml in PBS) was added to this mixture.
Cells were washed with PBS, dextran-DNA mixture
was added to the cells, and they were incubated at
37°C for 30 minutes. Serum-free media containing 0.2
µM chloroquine was added to the cells, and after 2.5
hours of incubation at 37°C, cells were washed with
10% DMSO in serum-free media. Cells were then
washed thoroughly with PBS, incubated for 66 hours
in regular media, and subsequently stimulated for 4
hours with adenosine. CAT assays were performed
according to manufacturer’s protocol (Promega-CAT
assay kit). Cells were lysed and protein quantitation
was done according to either the Bradford or Lowry
assay kit (Bio-Rad Laboratories Inc.). The cell viability
was not impaired by transfection or other aspects of
these treatments as judged by cell survival.
Confocal microscopy. Monolayers of T84 cells were
washed in HBSS, fixed in buffered formaldehyde for
20 minutes, permeabilized with saponin, incubated
with respective primary Ab’s overnight in a humidity
chamber, washed with HBSS, and subsequently incu-
bated with fluoresceinated secondary Ab’s (Jackson
ImmunoResearch Laboratories, West Grove, Pennsyl-
vania, USA). Monolayers, mounted in p-phenylenedi-
amine glycerol (1:1), were analyzed by
confocal microscopy (Zeiss dual laser
confocal microscope; Carl Zeiss Inc.,
Thornwood, New York, USA).
SDS-PAGE and Western blot. T84 cells
were lysed with PBS containing 1% Tri-
ton X-100 and 1% Nonidet P-40 (NP-40;
vol/vol), protease-inhibitor cocktail,
EDTA, SDS, 1 mM sodium orthovana-
date, and 1 mM sodium fluoride. SDS-
PAGE was performed according to the
Laemmli procedure using 10% acry-
lamide gel. Proteins were electrotrans-
ferred to nitrocellulose membranes and
probed with anti-CREB or anti–phos-
pho-CREB Ab’s (diluted 1:1,000). Subse-
quently, membranes were incubated with
the corresponding peroxidase-linked sec-
ondary Ab diluted 1:2,000, washed, and
were subsequently incubated with
enhanced chemiluminescence (ECL)
reagents (Amersham Pharmacia Biotech,
Piscataway, New Jersey, USA) before
exposure to high-performance chemilu-
minescence films (Amersham Pharmacia
Biotech). For mol wt determination,
polyacrylamide gels were calibrated using standard
proteins (Bio-Rad Laboratories Inc.) with mol wt
markers within the range 7,700 to 214,000.
Neutrophil isolation. Polymorphonuclear neutrophils
(PMNs) were isolated from whole blood (anticoagu-
lated with citrate/dextrose) obtained from healthy
volunteers, using a gelatin sedimentation technique
described previously (10). PMNs were resuspended in
modified HBSS devoid of [Ca++] and Mg2+ (HBSS) at
a concentration of 4 ×107cells/mL (4°C) and used
for subsequent experiments.
Determination of intracellular [Ca++]. Intracellular [Ca++]
was determined as described previously (30). Neu-
trophils were loaded with Indo-I and placed into a spec-
trofluorometer (at 37°C). Fluorescence excitation was
read at 355 nm while the excitation wavelength was
alternated between 405 and 485 nm four times per sec-
ond via Intracellular Cation Software (Hitachi, Sunny-
vale, California, USA). After reading fluorescence for 3
minutes, fMLP, IL-6 ± IL-6 Ab was added, as indicated
in the figure legends. Values of intracellular [Ca++] were
calculated as described previously (30).
Results
Adenosine induces polarized secretion of IL-6 mediated by the
A2b receptor. Adenosine added either to the apical or
basolateral side of T84 cells elicited IL-6 secretion that
was apically polarized to the apical side, irrespective of
the side of stimulation. As shown in Figure 1a, adeno-
sine (100 µM) added to the apical compartment of T84
cells induced approximately a ninefold increase in IL-6
secretion compared with T84 cells treated with vehicle
alone. In contrast, there was no significant elevation of
The Journal of Clinical Investigation | April 2001 | Volume 107 | Number 7 863
Figure 1
IL-6 induction by adenosine. T84 monolayers were prewashed in HBSS and equili-
brated at 37°C for 20 minutes. Adenosine (Ado; 100 µM) was added to the apical
(Ap) or basolateral (Bs) compartment. After an incubation period of 5.5–6 hours at
37°C, IL-6 was measured in the apical or basolateral compartment as described in
Methods. Values are expressed as pg/ml. Data represent the responses observed in
three separate experiments plotted as mean ± SD, n= 3 samples per treatment group.
Footnotes represent values significantly different from monolayers treated with vehi-
cle alone, AP< 0.001, BP< 0.05. (a) Monolayers treated with either vehicle or adeno-
sine. (b) Monolayers treated with either adenosine (100 µM) or adenosine plus 8-SPT
(100 µM). Ap, apical; Bs, basolateral.

IL-6 levels in the basolateral media. We have shown pre-
viously that basolaterally secreted cytokines are readily
measured by such approaches (30). Similar polarized
secretion of IL-6 was obtained when the T84 cells were
stimulated with basolateral adenosine. IL-6 in the baso-
lateral compartment was mildly but significantly
increased upon stimulation with basolateral adenosine.
We next studied the effect of 8-(p-sulfophenyl theo-
phylline) (8-SPT), an adenosine-receptor antagonist, on
the IL-6 secretion induced by adenosine. As seen in Fig-
ure 1b, IL-6 secretion induced by apical or basolateral
treatment of T84 cells with adenosine was abolished
when cells were also exposed to 8-SPT (100 µM) at the
time of adenosine stimulation, suggesting that the IL-6
secretion is mediated by the adenosine receptor.
Adenosine induces IL-6 secretion in a dose-dependent fash-
ion. A dose response of IL-6 secretion to adenosine is
shown in Table 1. Apical adenosine induced signifi-
cant IL-6 secretion at a dose of 1 µM, while basolat-
eral adenosine induced significant IL-6 secretion only
at higher concentrations. The rightward shift of the
dose response to basolateral adenosine is consistent
with our previous observation and is due to the pres-
ence of adenosine deaminase on the basolateral
membrane of the epithelial cells (12). The addition of
apical or basolateral NECA, a nonmetabolized ana-
logue of adenosine, induced an identical dose-
response curve (data not shown).
Time course of IL-6 induction by adenosine. The time
course of induction of IL-6 by adenosine (Figure 2)
showed that significant IL-6 secretion into the apical
media occurred 3 hours after exposure to adenosine
with maximal secretion at 6 hours. Apical and basolat-
eral adenosine induced a twofold increase in IL-6 secre-
tion in the apical media at 3 hours, with maximal stim-
ulation (ninefold to tenfold increase) between 3 and 6
hours after adenosine stimulation. We also performed
experiments to determine the length of exposure to
adenosine required to induce maximal IL-6 secretion.
To do this, cells were exposed to adenosine for 10, 60,
120 minutes, washed, and apical and basolateral media
were collected for a total of 6 hours. Our results showed
that a 10-minute exposure to adenosine was sufficient
to induce maximal IL-6 secretion (data not shown).
Stimulation of IL-6 by other stimuli is polarized. To see if
IL-6 secretion is polarized after stimulation by other
proinflammatory stimuli, T84 cells were stimulated
with S. typhimurium and TNF-α(both of which have
been shown to induce IL-6 transcripts in intestinal
epithelial cells) and forskolin, a direct activator of
adenylate cyclase (28). As shown in Figure 3, forskolin
(10 µM), S. typhimurium, and TNF-α(100 ng/ml)
induced a greater than tenfold increase in IL-6 secre-
tion, which also was polarized to the apical compart-
ment. All of these stimuli also increased IL-6 secretion
in the basolateral compartment albeit to much lesser
extent (vehicle, forskolin, S. typhimurium, and TNF-α
respectively, in picograms per milliliter: 1.9 ± 0.5, 4.9
± 0.4, 12.1 ± 4.1, 33.4 ± 4.2). Carbachol, an agent that
increases intracellular [Ca++], did not increase IL-6
secretion in the apical or the basolateral media.
Adenosine-induced IL-6 secretion is transcriptionally mediat-
ed. IL-6 promoter activity has been localized to –1 to
–435 upstream of the start site. The IL-6 promoter con-
tains two overlapping regions — second messenger (also
called multiple responsive elements or MREs) and
cytokine responsive element — defined by their roles in
IL-6 induction by a variety of inducers (15, 29) (Figure
4a). Critical consensus elements within the MREs have
been identified. MRE I contains an ATF/CREB site and
MRE II contains a NF–IL-6 site (also called c/EBPβ).
Agents that increase intracellular cAMP, as adenosine
and forskolin are known to do, induce IL-6 secretion by
the activation of both MRE I and MRE II. Additional
sites that are important for enhanced IL-6 expression in
response to various inducers include the AP-1 site (–283
to –277) and the NF-κB site (–73 to –63). In addition, the
IL-6 promoter also contains several trans-acting elements
that downregulate IL-6 transcription, such as the gluco-
corticoid responsive element and IFN regulatory factor.
We studied the effect of adenosine on the induction
of the IL-6 promoter linked to the CAT reporter gene.
864 The Journal of Clinical Investigation | April 2001 | Volume 107 | Number 7
Table 1
IL-6 secretion: dose response to adenosine
Ap adenosine Bs adenosine
Ap media Bs media Ap media Bs media
No addition 3.0 ± 0.4 2.0 ± 0.2 3.6 ± 0.8 0.9 ± 0.6
0.01 µM 3.6 ± 1.0 0.9 ± 0.1 3.9 ± 0.2 1.2 ± 0.4
0.1 µM 3.4 ± 1.2 1.5 ± 0.1 3.7 ± 0.5 1.1 ± 0.7
1.0 µM 8.4 ± 2.0A1.8 ± 0.4 5.2 ± 1.0 1.9 ± 0.2
10 µM 27.0 ± 3.0B2.0 ± 0.6 10.9 ± 1.9A3.2 ± 0.3A
100 µM 30.0 ± 3.4B4.5 ± 1.5 29.6 ± 3.2B3.1 ± 1.3A
IL-6 in picograms per milliliter. Significantly different from controls, AP< 0.05;
BP< 0.001. Ap, apical; Bs, basolateral.
Figure 2
Time course of IL-6 secretion. T84 monolayers were prewashed in
HBSS. After equilibration at 37°C for 20 minutes, cells were stimulat-
ed with adenosine (apical or basolateral; 100 µM), HBSS was collect-
ed for 1, 2, 3, or 6 hours, and IL-6 was measured as described in Meth-
ods. Values are expressed as pg/ml IL-6. Data represent the responses
observed in three separate experiments plotted as mean ± SD, n= 2
samples per treatment group. Footnotes represent values significantly
different from the respective 0 time point, AP< 0.001, BP< 0.05.

COS-7 cells that possess A2b receptors were used in
view of the difficulty of transfecting T84 cells. It has
been shown previously that the A2b receptors in COS-
7 cells, like those in the T84 cells, positively couple to
adenylate cyclase(31). Stimulation of COS-7 cells with
adenosine thus results in increased levels of intracellu-
lar cAMP, and this forms the major signaling pathway
induced by adenosine via the A2b receptor. In addition,
COS 7 cells secrete IL-6 (unstimulated: 36.8 ± 3.5;
adenosine 100 µM: 155.5 ± 25.0 pg/ml). As seen in Fig-
ure 4b, adenosine induced approximately a 40-fold
increase in CAT activity in the cells transfected with the
full-length IL-6 promoter construct (+15 to –435).
Mutations in either the region of the ATF/CREB or
NF–IL-6 elements of the IL-6 promoter abolished
responsiveness to adenosine (Figure 4b) while muta-
tion of the NF-κB site did not affect the CAT activity in
response to adenosine. Such data imply that both the
ATF/CREB and the NF–IL-6 sites are important for
adenosine induced IL-6 secretion.
Activation of ATF/CREB mediates adenosine-induced IL-6
secretion. CREB and ATF are nuclear transcription fac-
tors that are activated by cAMP-mediated signaling
pathways. Since the only known signaling molecule
induced by adenosine is cAMP, we next sought to see if
adenosine induced activation of CREB in T84 cells. As
seen in Figure 5, apical or basolateral stimulation with
adenosine (100 µM) resulted in phosphorylation of
CREB while the total CREB remained unchanged.
Using Ab’s to phosphorylated CREB at ser-133, we
showed that adenosine induced phosphorylated CREB
5 minutes after stimulation, and maximal induction
was seen at 1 hour after stimulation. In addition, both
apical and basolateral adenosine induced phosphory-
lated ATF-1, a transcription factor that belongs to the
CREB family. The phosphorylated ATF-1, like CREB,
was detected at 5 minutes after adenosine stimulation
with maximal levels seen at 1 hour (Figure 5). Both the
phosphorylated CREB and ATF-1 induction returned
to baseline by 6 hours (Figure 5).
To study the cellular localization of phosphorylated
CREB, cells were stimulated with adenosine for 5, 45,
or 60 minutes and subsequently stained for immunolo-
calization of CREB. While control cells showed no
The Journal of Clinical Investigation | April 2001 | Volume 107 | Number 7 865
Figure 3
Effect of various proinflammatory agents on IL-6 secretion. T84 cells
were washed with HBSS. After equilibration of 20 minutes at 37°C,
forskolin (10 µM), TNF-α(100 ng/ml), and carbachol (1 µM) were
added basolaterally. Apical and basolateral media were collected for
5.5–6 hours after stimulation. S. typhimurium was added apically for 1
hour for surface colonization (30), after which the monolayers were
rinsed thoroughly with HBSS and incubated for an additional 5 hours,
and IL-6 assays were performed. Data represent the responses observed
in two separate experiments plotted as mean ± SD, n= 2 per treatment
group. ASignificantly different from the unstimulated control, P< 0.001.
BSignificantly different from unstimulated control, P< 0.05.
Figure 4
Effect of adenosine on IL-6 promoter activity and analysis of adeno-
sine-mediated regulation of IL-6 promoter mutants. (a) Most ele-
ments of the IL-6 promoter that have been characterized lie within
the 300 bp proximal to the start site (+1) of transcription. These ele-
ments include multiple AP-1–binding sites, an IRF-1–binding site,
three glucocorticoid response elements (GRE), two second-messen-
ger and cytokine-responsive elements (MREs), the first of which con-
tains ATF/CREB binding sites and the second a binding site for
C/EBPβor NF-IL-6–binding site, and a NF-κB–binding element. The
promoter also contains two transcription start sites, one major and
one minor. Mutants that were generated in the ATF/CRE, NF–IL-6,
and NF-κB sites are indicated in Methods. (b) COS-7 cells were tran-
siently transfected with wild-type (WT) or various mutant IL-6 CAT
constructs. Cells were stimulated with adenosine 66 hours after
transfection, and a CAT assay was performed 4 hours after stimula-
tion. Data represent fold increase over transfected and unstimulat-
ed control (expressed as nanograms CAT/µg protein) observed in
two separate experiments plotted as mean ± SD from duplicate
determination of two samples per group, n= 2.

staining of phosphorylated CREB at any of the time
points tested, cells treated with apical or basolateral
adenosine for 5 minutes showed nuclear staining of
phosphorylated CREB (Figure 6, a, b, and c). Control
cells are outlined with rhodamine-phalloidin staining
in Figure 6a. Similar results were obtained at 30 and 60
minutes after adenosine stimulation (data not shown).
IL-6 induces neutrophil activation. Transmigrated neu-
trophils can release 5′AMP into the lumen (10), which
is converted to adenosine (11) and recognized by the
apical A2b receptors on intestinal epithelia (12). Api-
cal adenosine stimulation, as shown above, stimulates
apical IL-6 secretion. Thus, we next sought to deter-
mine if IL-6 might, in turn, modify neutrophil
responses, thus indicating potential of neutrophil-to-
epithelial cell and epithelial-to-neutrophil paracrine
loops after transepithelial migration. Because calcium
mobilization is part of many of the pathways by which
neutrophils are activated, we measured intracellular
[Ca++] flux in response to grad-
ed doses of IL-6 (25–100
pg/ml) with or without
anti–IL-6 mAb. As seen in Fig-
ure 7, IL-6 increased intracel-
lular [Ca++] in neutrophils in a
dose-dependent fashion. IL-6
(100 pg/ml) induced a [Ca++]
flux that was comparable to
that induced by fMLP (10–5 M).
The effect of IL-6 on the [Ca++]
flux was completely abolished
by the anti–IL-6 mAb, thus
indicating that [Ca++] mobi-
lization is due to IL-6. Thus
these data demonstrate a
paracrine loop in which there
is cross-talk between epithelial
cells and neutrophils after
neutrophil transmigration.
Discussion
In this study we used polarized model human intestinal
epithelia to examine the role of adenosine, a paracrine
factor generated by the conversion of neutrophil-derived
5′AMP during active intestinal inflammation, in the
epithelial orchestration of immune response. The cell
line used, T84, was used previously to identify mecha-
nisms of neutrophil-derived 5′AMP conversion to
adenosine (CD73), to identify the adenosine receptor
(A2b), and to define its signaling mechanism (cAMP).
These observations were shown subsequently to accu-
rately predict events in natural human intestinal epithe-
lia (1, 12, 13). We now demonstrate that adenosine
induces intestinal epithelial cells to secrete significant
levels of IL-6 in a polarized manner. Specifically, adeno-
sine induces IL-6 secretion into the apical (luminal) com-
partment, whether the cells are stimulated apically or
basolaterally. In addition, known inducers of IL-6,
including forskolin, S. typhimurium, and TNF-αalso
866 The Journal of Clinical Investigation | April 2001 | Volume 107 | Number 7
Figure 5
Expression of phosphorylated ATF/CREB in response to
adenosine. T84 cells were washed with HBSS, equili-
brated for 20 minutes at 37°C, and stimulated with
adenosine (100 µM, apical or basolateral) for the vari-
ous times indicated. Whole-cell detergent lysates
(approximately 15 µg protein/lane) were resolved by
SDS-PAGE and immunoblotted for phosphorylated
ATF/CREB (a) and total CREB (b). Scanning densitom-
etry of the blot is shown in bdepicting a four- and eight-
fold increase in ATF-1 and CREB, respectively compared
with 0 time. The induction was maximal at 60 minutes
and declined at 180 and 360 minutes after stimulation.
Data represent the responses observed in two separate
experiments with two filters per time point.
Figure 6
Confocal microscopy of phospho-CREB. T84 monolayers were incubated with apical or baso-
lateral adenosine (100 µM) for 5 minutes. Nuclear staining was determined by immunofluores-
cence labeling and confocal microscopy. The en face images document the presence of activat-
ed phospho-CREB staining in cells treated with apical or basolateral adenosine (band c,
respectively). Control cells double-stained with phospho-CREB and rhodamine-phalloidin are
shown in a. Similar results were obtained at 30 and 60 minutes after adenosine stimulation.

induce IL-6 secretion predominantly in the apical com-
partment. Moreover, baseline IL-6 secretion in unstimu-
lated epithelial monolayer is also significantly higher
apically than basolaterally. Similar apically polarized
secretion has been demonstrated recently for secretory
leukocyte protease inhibitor (SLPI) in intestinal epithe-
lial cells (32). The induction of basolateral IL-6 secretion,
though still far less than apical secretion, is significant
following TNF-α stimulation. One explanation for
apparent basolateral IL-6 secretion in response to
TNF-αis related to the drop in barrier function (transep-
ithelial resistance), which also occurs in response to
TNF-α(33) and likely diminishes the ability to rapidly
separate apical and basolateral compartments.
The induction of IL-6 by adenosine is likely mediated
by the A2b receptor (the only adenosine receptor sub-
type present in this cell line) because 8-SPT, an adeno-
sine-receptor antagonist, abolishes IL-6 induction by
both apical and basolateral adenosine. The dose
response of adenosine-induced IL-6 secretion is also
suggestive of a receptor-mediated process with satura-
tion of IL-6 induction seen at 10 µM for apical adeno-
sine and 100 µM for basolateral adenosine. The IL-6
induction by adenosine appears to be a transcription-
mediated process and not release of preformed or stored
IL-6, since IL-6 induction is seen 3 hours after stimula-
tion with adenosine. Consistent with this notion is our
data on the transcriptional activation of IL-6 promoter
by adenosine. We demonstrate that adenosine induces
the activation of the full-length IL-6 promoter and the
ATF/CREB and NF–IL-6 sites, but not the NF-κB site of
the IL-6 promoter.
We have shown previously that the only demonstrable
signaling pathway for adenosine is mediated by
cAMP/protein kinase A (PKA), and adenosine does not
induce calcium or PKC activity (ref. 12; S.V. Sitaraman
and J.L. Madara, unpublished observation). Adenosine
induces a polarized increase in cAMP in the intestinal
epithelial cells. While basolateral adenosine stimulation
results in several-fold increase in cAMP, apical stimula-
tion results in small but significant increase in cAMP in
response to the same dose of adenosine (12). We demon-
strate here that adenosine-induced IL-6 secretion is
mediated by a time-dependent activation of CREB, a
nuclear transcription factor whose activation by the
cAMP/PKA pathway is well established (34). Interest-
ingly, despite disparate cAMP levels induced by apical
versus basolateral adenosine, the kinetics of CREB
induction and the phosphorylation of nuclear CREB
did not differ. This suggests that the signaling at the
apical membrane perhaps occurs in microdomains that
contain the A2b receptor and its signaling effectors.
The kinetics of IL-6 induction shows that a 10-
minute exposure to adenosine is sufficient for maximal
IL-6 induction, suggesting that the signal transduction
process is complete within this time frame. Consistent
with this notion is the phosphorylation of CREB,
which occurs by 5 minutes. Based on our observation
that the only known signaling pathway for adenosine
in T84 cells is cAMP/PKA mediated and that adenosine
induces a time-dependent activation of phosphorylat-
ed CREB and its nuclear translocation in T84 cells, we
conclude that adenosine induces IL-6 via transcrip-
tional activation mediated by cAMP. This is also sup-
ported by the inability of carbachol, an agent that
increases intracellular [Ca++], to induce IL-6.
The Journal of Clinical Investigation | April 2001 | Volume 107 | Number 7 867
Figure 7
Effect of IL-6 on neutrophil calcium response. Neutrophils (106) were
loaded with the [Ca++] indicator Indo-1 and placed in a thermostat-
ed spectrofluorometer, and intracellular [Ca++] was measured as
described in Methods. Stimuli were added at 60 seconds. Intracellu-
lar [Ca++] was measured in response to fMLP (10–6M), IL-6 in the
indicated doses, and monoclonal anti–IL-6 added along with IL-6
(100 pg/ml) (IL-6 + IL-6 Ab).
Figure 8
Schematic representation of epithelial-neutrophil interaction in
a crypt abscess. An epithelial monolayer with neutrophils trans-
migrating to the luminal compartment is shown at the left.
Inset shows that adenosine is derived from the enzymatic con-
version (mediated by epithelial ectonucleotidase, CD 73) of 5′
AMP released in the lumen by the neutrophils. Adenosine thus
released interacts with the adenosine A2b receptor, a G pro-
tein–coupled receptor, resulting in an increase in intracellular
cAMP that may be involved in the transcriptional activation of
IL-6 secretion. IL-6 is preferentially released in the apical com-
partment and induces intracellular [Ca++] flux in neutrophils,
which may be involved in the release of oxygen radicals, elas-
tase, etc., from the neutrophils.

CREB and ATF are nuclear transcription factors that
are activated upon phosphorylation of Ser-133 residue
in the PKA-inducible domain (34). NF–IL-6 is distinct
from the CREB family of transcription factors and is
activated by cAMP and calcium-mediated signaling
pathways (10). The promoter region of NF–IL-6 has
CRE and is induced by cAMP-mediated signaling path-
ways (35). Activated NF–IL-6 forms heterodimers with
activated ATF/CREB to bind to the promoter region of
target genes (16). Therefore, the cAMP-mediated sig-
naling pathway may be a common mechanism for the
activation of IL-6 transcription through CRE and
NF–IL-6 elements by adenosine. This is consistent with
the observation by others that CRE and NF–IL-6 are
the major sites for transcriptional activation of IL-6
induced by agents that increase intracellular cAMP
such as forskolin and cholera toxin(10, 36).
IL-6 is an important proinflammatory cytokine that
is consistently seen in high levels in the serum and tis-
sue of patients with active inflammatory bowel dis-
ease, and one of the sources for the IL-6 is the lamina
propria cells, including macrophages and monocytes
(18–20, 37). In this article we provide further evidence
that the epithelial cell production of chemokines may
also contribute to immune regulation at the mucosal
surface. Here we have shown that IL-6 is secreted into
the luminal compartment. In considering the poten-
tial role of this cytokine, neutrophils seem a logical
target, not only because they transmigrate into the
lumen during active inflammation, but because they
also provide paracrine apical adenosine signals and
possess IL-6 receptors (15). Our data suggest that api-
cally secreted IL-6 may indeed increase neutrophil
intracellular [Ca++] in neutrophils — a classic early sig-
nal in neutrophil activation. This specific signaling
event is consistent with studies of downstream events
in which IL-6 exposure has been shown to induce elas-
tase release, PAF production, and production of oxy-
gen-free radicals by neutrophils (38). IL-6 has also
been shown to interact with intestinal epithelial cells
via IL-6 receptors (18, 27). For example, IL-6 induces
the sodium-glucose cotransporter in the small intes-
tinal cells causing an increased uptake of glucose (39).
The other potential effect of IL-6 on epithelial cells is
being explored currently in our laboratory.
In conclusion, we believe this is the first demonstra-
tion of apically polarized cytokine secretion by intes-
tinal epithelial cells in response to an effector, here
adenosine, an endogenous nucleoside that is generat-
ed in situ during active inflammation. The effect of
epithelial-derived IL-6 on induction of intracellular
[Ca++] in neutrophils strongly suggests that this
cytokine plays a relevant physiological role in the inter-
action between the epithelial cells and neutrophils
(Figure 8). Future studies will involve confirming our
in vitro observation in natural human intestinal
epithelia. Taken together with its effect on chloride
secretion, the event underlying secretory diarrhea,
adenosine thus may act as a proinflammatory agent,
and adenosine antagonists may therefore be useful
agents to ameliorate particular aspects of intestinal
inflammation such as IL-6–mediated neutrophil acti-
vation in the crypt lumen.
Acknowledgments
This work was supported by NIH grant K08 DK-02802
(to S. Sitaraman), CCFA career development awards (to
S.V. Sitaraman and D. Merlin), and NIH grants K01
DK-02792 (to A.T. Gewirtz), and DK-35932 and DK-
47662 (to J.L. Madara). We would like to thank Michael
Hobart for his excellent graphic art and Anjali Rao for
a critical reading of the manuscript.
1. Nash, S., Parkos, C., Nusrat, A., Delp, C., and Madara, J.L. 1991. In vitro
model of intestinal crypt abscess: a novel neutrophil-derived secreta-
gogue (NDS) activity. J. Clin. Invest. 87:1474–1477.
2. Palmer, T.M., and Stiles, G.L. 1995. Adenosine receptors. Neurophar-
macology. 34:683–694.
3. Roman, R.M., and Fitz, J.G. 1999. Emerging roles of purinergic sig-
nalling in gastrointestinal epithelial secretion and hepatobiliary func-
tion. Gastroenterology. 116:964–979.
4. Ralevic, V., and Burnstock, G. 1998. Receptors for purines and pyrim-
idines. Pharmacol. Rev. 50:413–492.
5. Van Belle, H., Gossens, F., and Wynants, J. 1987. Formation and release
of purine catabolites during hypoperfusion, anoxia, and ischemia. Am.
J. Physiol. 252:H886.
6. Feoktistov, I., and Biaggioni, I. 1997. Adenosine A2B receptors. Phar-
macol. Rev. 49:381–402.
7.Linden, J., Auchampach, J.A., Xiaowei, J., and Figler, R.A. 1995. The
structure and function of A1 and A2B adenosine receptors. Life Sci.
62:1519–1524.
8. Link, A.A., et al. 2000. Ligand-activation of adenosine A2a receptors
inhibits IL-12 production by human monocytes. J. Immunol.
164:436–442.
9. Feoktistov, I, and Biaggioni, I. 1995. Adenosine A2b receptors evoke
interleukin-8 secretion in human mast cells. An enprofylline-sensitive
mechanism with implications for asthma. J. Clin. Invest. 96:1979–1986.
10. Madara, J.L., et al. 1993. 5’AMP is the neutrophil-derived paracrine fac-
tor that elicits chloride secretion from T84 intestinal epithelial mono-
layers. J. Clin. Invest. 91:5716–5723.
11.Strohmeier, G.R., et al. 1997. Surface expression, polarization, and
functional significance of CD73 in human intestinal epithelia. J. Clin.
Invest. 99:2588–2601.
12. Strohmeier, G.R., Reppert, S.M., Lencer, W.I., and Madara, J.L. 1995.
The A2b adenosine receptor mediates cAMP responses to adenosine
receptor agonists in human intestinal epithelia. J. Biol. Chem.
270:2387–2394.
13.Barrett, K.E., Cohn, J.A., Huott, P.A., Wasserman, S.I., and Dharm-
sathaphorn, K. 1990. Immune-related intestinal chloride secretion. II.
Effect of adenosine on T84 cell line. Am. J. Physiol. 258:C902–C912.
14.Sitaraman, S.V., Si-Tahar, M., Merlin, D., Strohmeier, G.R., and
Madara, J.L. 2000. Polarity of A2b adenosine receptor expression deter-
mines characteristics of receptor desensitization. Am. J. Physiol. Cell.
Physiol. 278:C1230–C1236.
15. Keller, E.T., Wanagat, J., and Ershler, W.B. 1996. Molecular and cellu-
lar biology of interleukin-6 and its receptor. Front. Biosci. 1:d340–d357.
16. Simpson, R.J., Hammacher, A., Smith, D.K., Matthews, J.M., and Ward,
L.D. 1997. Interleukin-6: structure-function relationships. Protein Sci.
6:929–955.
17.Shirota, K., LeDuy, L., Yuan, S.Y., and Jothy, S. 1990. Interleukin-6 and
its receptor are expressed in human intestinal epithelial cells. Virchows
Arch. B Cell Pathol. Incl. Mol. Pathol. 58:303–308.
18.Hosokawa, T., et al. 1999. Interleukin-6 and soluble interleukin-6
receptor in the colonic mucosa of inflammatory bowel disease. J. Gas-
troenterol. Hepatol. 14:987–996.
19. Reinisch, W., et al. 1999. Clinical relevance of serum interleukin-6 in
Crohn’s disease: single point measurements, therapy monitoring, and
prediction of clinical relapse. Am. J. Gastroenterol. 94:2156–2164.
20. Louis, E., et al. 1997. A high serum concentration of interleukin-6 is
predictive of relapse in quiescent Crohn’s disease. Eur. J. Gastroenterol.
Hepatol. 9:939–944.
21. Kusugami, K., et al. 1995. Elevation of interleukin-6 in inflammatory
bowel disease is macrophage- and epithelial cell-dependent. Dig. Dis.
Sci. 40:949–959.
22. Atreya, R., et al. 2000. Blockade of interleukin 6 trans signaling sup-
presses T-cell resistance against apoptosis in chronic intestinal inflam-
868 The Journal of Clinical Investigation | April 2001 | Volume 107 | Number 7

mation: evidence in crohn disease and experimental colitis in vivo. Nat.
Med. 6:583–588.
23. Yamamoto, M., Yoshizaki, K., Kishimoto, T., and Ito, H. 2000. IL-6 is
required for the development of Th1 cell-mediated murine colitis. J.
Immunol. 164:4878–4882.
24.Yan, S.F., Ogawa, S., Stern, D.M., and Pinsky, D.J. 1997. Hypoxia-
induced modulation of endothelial cell properties: regulation of bar-
rier function and expression of interleukin-6. Kidney Int. 51:419–425.
25. Moon, M.R., et al. 2000. Interleukin-1beta induces complement com-
ponent C3 and IL-6 production at the basolateral and apical mem-
branes in a human intestinal epithelial cell line. Shock. 13:374–378.
26.McGee, D.W., Beagley, K.W., Aicher, W.K., and McGhee, J.R. 1993.
Transforming growth factor-beta and IL-1 beta act in synergy to
enhance IL-6 secretion by the intestinal epithelial cell line, IEC-6. J.
Immunol. 151:970–978.
27. Molmenti, E.P., Ziambaras, T., and Perlmutter, D.H. 1993. Evidence for
an acute phase response in human intestinal epithelial cells. J. Biol.
Chem. 268:14116–14124.
28. Weinstein, D.L., O’Neill, B.L., and Metcalf, E.S. 1997. Salmonella typhi
stimulation of human intestinal epithelial cells induces secretion of
epithelial cell-derived interleukin-6. Infect. Immun. 65:395–404.
29. Harcourt, J.L., and Offermann, M.K. 2000. Interferon alpha synergis-
tically enhances induction of interleukin-6 by double stranded RNA in
HeLa cells. Eur. J. Biochem. 267:1–11.
30. Gewirtz, A.T., et al. 2000. Salmonella typhimurium induces epithelial
IL-8 expression via Ca(2+)-mediated activation of the NF-kappaB path-
way. J. Clin. Invest. 105:79–92.
31.Clancy, J.P., Ruiz, F.E., and Sorscher, E.J. 1999. Adenosine and its
nucleotides activate wild-type and R117H CFTR through an A2B
receptor-coupled pathway. Am. J. Physiol. 276:C361–C369.
32. Si-Tahar, M., Merlin, D., Sitaraman, S., and Madara, J.L. 2000. Consti-
tutive and regulated secretion of secretory leukocyte proteinase
inhibitor by human intestinal epithelial cells. Gastroenterology.
118:1061–1071.
33. Schmitz, H., et al. 1999. Tumor necrosis factor-alpha (TNFalpha) reg-
ulates the epithelial barrier in the human intestinal cell line HT-29/B6.
J. Cell. Sci. 112:137–146.
34. Shaywitz, A.J., and Greenberg, M.E. 1999. CREB: a stimulus-induced
transcription factor activated by a diverse array of extracellular signals.
Annu. Rev. Biochem. 68:821–861.
35. Roesler, W.J. 2000. What is a cAMP response unit? Mol. Cell. Endocrinol.
162:1–7.
36. Zidek, Z. 1999. Adenosine: cyclic AMP pathways and cytokine expres-
sion. Eur. Cytokine Netw. 10:319–328.
37.Reinisch, W., et al. 1999. Clinical relevance of serum interleukin-6 in
Crohn’s disease: single point measurements, therapy monitoring, and
prediction of clinical relapse. Am. J. Gastroenterol. 94:2156–2164.
38. Biffl, W.L., et al. 1996. Interleukin-6 stimulates neutrophil production
of platelet-activating factor. J. Leukoc. Biol. 59:569–574.
39. Hardin, J., Kroeker, K., Chung, B., and Gall, D.G. 2000. Effect of proin-
flammatory interleukins on jejunal nutrient transport. Gut.
47:184–191.
The Journal of Clinical Investigation | April 2001 | Volume 107 | Number 7 869