Enhanced mucosal cytokine production in inflammatory bowel disease

GASTROENTEROLOGY

1992:102:529-537

Enhanced Mucosal Cytokine Production in Inflammatory Bowel Disease WILLIAM ANDREW Division Canberra,

E. PULLMAN, SUSAN ELSBURY, J. HAPEL, and WILLIAM F. DOE

of Clinical Australia

Sciences,

John

Curtin

School

of Medical

Proliferation, maturation, chemotaxis, and activation of neutrophils and monocytes are mediated largely by cytokines, including colony-stimulating factors and lymphokines. Cytokines produced in the intestinal mucosa contribute to the increased migration of neutrophils and monocytes into the lesion of inflammatory bowel disease and to the activation of these inflammatory cells. Lamina propria mononuclear cells isolated from colon tissue from 14 patients with inflammatory bowel disease (IBD) and from histologically normal controls were studied. Cells from IBD-affected tissue produced significantly more colony-stimulating factor activity (1402+ 252 U) per 2 X lo6 cellsthan those from normal mucosa (362+ 85 U), mainly because of the increased production of granulocyte colony-stimulating factor and interleukin 1.This was accompanied by increases in the amount of specific messenger RNA for these two cytokines in lamina propria mononuclear cells from mucosa of patients with Crohn’s disease (CD) compared with normal controls. By confrasf, there was a substantial reduction in interleukin 3 production in CD and in ulcerative colitis lamina propria mononuclear cells, and this was reflected in significantly reduced expression of interleukin 3 messenger RNA in CD cells. Of the agents used in the therapy of IBD, hydrocortisone and 5-aminosalicylic acid, but not cyclosporin A, markedly suppressed in vitro production of cytokines by lamina propria mononuclear cells, suggesting that their therapeutic efficacy in vivo may be due in part to down-regulation of cytokine production in the inflamed mucosa. he mucosal lesions of Crohn’s disease (CD) and ulcerative colitis (UC) are characterized by the presence of infiltrates of myeloid cells composed predominantly of neutrophils and macrophages. ggmTcphagocyte scanning has shown the localization of these cells, particularly neutrophils, to the active foci of inflammatory bowel disease (IBD).‘,’ Neutrophil and monocyte proliferation, matura-

T

MASANOBU

Research,

Australian

KOBAYASHI, National

University,

tion, and activation are regulated in part by cytokines, including the colony-stimulating factors (CSF) and interleukins (IL), which also have chemotactic properties.3-5 The study of these hormonelike glycoproteins has recently been greatly facilitated by the cloning of genes for some of these substances and production of recombinant mo1ecules.68 As a result, many of the physicochemical and biological properties of IL-3, granulocyte-macrophage CSF (GM-CSF), granulocyte CSF (G-CSF), and IL-l have been elucidated.g-” To determine the relevance of these factors to the increased neutrophil and macrophage infiltrate in IBD and to the activation of these cells in the inflamed mucosa, the levels of cytokine production by lamina propria mononuclear cells (LPMCs) were studied in IBD-affected colonic mucosa and histologically normal controls. The results indicate that LPMCs from IBD-affected mucosa produce greatly enhanced levels of CSF activity that are mainly attributable to increased production of G-CSF and IL-l.

Materials and Methods Patients and Mucosal Specimens Histologically normal mucosa obtained after surgical resection of intestine from 13 patients with nonmalignant, noninflammatory conditions including sigmoid volvulus (n = 3), angiodysplasia (n = a), megacolon (n = I), adenoma (n = 4), and Meckel’s diverticulum (n = 1)were used as the source of control LPMCs. For messenger RNA studies, histologically normal mucosa from cancer-bearing colon also served as controls. Intestinal tissue resected from 14 patients with IBD, comprising 10 specimens of CD (5 ileal) and 4 of UC, formed the diseased group. The diagnoses of CD and UC were made on standard clinical, endoscopic, radiological and histological criteria. At the time of surgery, all IBD patients were taking corticosteroids and/ or azathioprine. Five patients were also being treated with

0

1992 by the American Gastroenterological Association 6616-5065/92/$3.66

530 PULLMAN ET AL.

sulphasalazine. Mucosal involvement by IBD was confirmed histologically in all the resected specimens. This project has been approved by the Australian Capital Territory Hospital’s Ethics Committee. Mucosal

Disaggregation

The intestinal mucosa was disaggregated using an enzymic method based on modifications of previously published mefhods.‘2’13 Briefly, freshly resected mucosa was treated with sequential washes of 0.75 mmol/L (EDTA) at 37°C to remove all the epithelial cells and the mucus. The tissue was then minced finely and incubated overnight with gentle stirring in the disaggregation solution containing RPM1 1640 (Flow Labs, Melbourne, Australia), 10% heat-inactivated fetal calf serum (FCS; CSL Laboratories, Melbourne, Australia), 2 U/mL purified collagenase (CLSPA type, Worthington Biochemical Corp., Freehold, NJ), 5 U/mL gentamicin (50 pg/mL), nystatin (100 U/mL), 20 mmol/L HEPES buffer, and glutamine (2 mmol/L). Cells were recovered by centrifugation of the digest (400 g, 10 minutes at 4’C) followed by resuspension of the cells in RPM1 1640 containing 10% FCS and LPMCs isolated by centrifugation (400 g for 30 minutes) onto a Ficoll-paque cushion (1.077 g/mL; Pharmacia Fine Chemicals AB, Uppsala, Sweden) following the method of Boyum.14 The viability of LPMCs was determined by trypan blue (0.1%) exclusion and always exceeded 92%. The cells were cultured at a density of 2 X lo6 cells/ml in RPM1 1640 containing 10% FCS, gentamicin, penicillin, and HEPES (20 mmol/L, pH 7.4) in plastic tissue culture flasks in a humidified atmosphere of 5% CO, in air at 37°C for 24 hours to deplete adherent cells before investigation of cytokine production as outlined below. To assess the distribution of cell types in the LPMC population, cytospins (Shandon Centrifuge, 500 rpm, 5 minutes) were made before and after the 24-hour culture period used to deplete adherent cells. Cytospins were stained with Wright Giemsa, and cell types were scored as a percentage of total cell numbers. Fluorescence analysis was also carried out as outlined below to assess the proportion of T cells and activated cells in LPMCs from control and IBD-affected tissue. Antibodies

and Immunofluorescence

Indirect immunofluorescence was carried out using the following first-layer monoclonal antibodies (mAbs) diluted in phosphate-buffered saline (PBS) as indicated: 0KT3, l/5 (anti-CD3, donated by Dr. H. S. Warren), antiTat, l/20 (anti-IL-2 receptor IL2-R; Becton Dickinson, Sunnyvale, CA). LPMCs (1 X 107/mL) were prepared in PBS with 5% FCS at 4°C. The first-layer antibody was added for 20 minutes at 4’C. Cells were then washed twice, and the second-layer antibody, fluorescein-conjugated antimouse F(ab’), Ig (Silenius, Melbourne, Australia), was added for a further 20 minutes at 4°C. After washing twice, 200 cells were counted, and those showing positive membrane fluorescence were expressed as a percentage of the total cell number.

GASTROENTEROLOGY Vol. 102,No. 2

Induction

of Cytokine

Production

by LPMCs

The nonadherent LPMCs were washed and resuspended at a density of 4 X 106/mL in RPM1 supplemented by 1% heat-inactivated human AB serum (HABS). The cultures were then stimulated with a 2-hour pulse of phytohemagglutinin (PHA, CSL Labs) at a concentration of 40 pg/mL. Control cultures received no PHA. Cells were then washed and resuspended in RPM1 with 1% HABS and cultured in 2-mL volumes in 24-well flat-bottom plates (Linbro; Flow Labs, McLean, VA) for a further 22 hours. The cytokine-containing supernatants were then recovered by centrifugation (4OOg, 5 minutes), filtered (0.22 pm; Millipore Corp., Bedford, MA), and stored at 4°C before assay.

Assay and Characterization

of CSF Activity

Total CSF activity in the culture supernatants generated from 4 X lo6 cells/pL was determined using 50yL aliquots in a bone marrow stem cell (BMSC)-proliferation assay.ls Briefly, human myeloid cell precursors, obtained by density-gradient centrifugation of long-term bone marrow cultures, were added to twofold serial dilutions of test samples in Iscove’s minimal essential medium (IMEM) in flat-bottom 96-well plates (Nunc, Roskilde, Denmark) at a density of 2 X lo4 cells/well. Cultures were incubated for 48 hours before being pulsed overnight with 0.5 pCi of tritiated thymidine and harvested onto glass-fiber filter paper (Titertek, McLean, VA). Each assay included standard titrations of GM-CSF, G-CSF, IL-3, IL-2, and IL-l. Results were expressed as end-point units (U), defined as the reciprocal of the sample dilution producing proliferation equal to the mean plus 3 standard deviations (SD) of the spontaneous proliferation for each assay.

Fractionation

of Conditioned

Media

Conditioned media from LPMCs obtained from two control subjects, two CD patients, two UC patients were concentrated lo-fold using Minicon B15 filters (Amicon Corp., Danvers, MA) and filtered (0.22 pm) before separation by hydrophobic interaction chromatography using FPLC phenyl-superose (Pharmacia) and conditions chosen to optimize the separation of the standards, rGM-CSF, rIL3, and purified G-CSF.” Supernatants were dialyzed overnight against 0.05 mol/L phosphate buffer (pH 6) containing 1.7 mol/L (NH,),SO,. Aliquots of sample (500 pL) were then loaded onto a phenyl-superose column (HR5/5, Pharmacia) and eluted using a descending linear salt gradient (23 mL) followed by 4 mL of 0.05 mol/L phosphate buffer and an ascending linear gradient to 70% ethylene glycol (23 mL). Elution was controlled by the programmed LCC-500 processor (Pharmacia), and 25 X 2-mL fractions were collected. The protein profile was determined by continuous monitoring of the eluate by absorbance at 280 nm. Each fraction was dialyzed against PBS (pH 7.4) and assayed for BMSC proliferation activity. To confirm the presence of hemopoietic colony-forming activity, the fractions of the three major peaks of prolifera-

MUCOSAL CYTOKINE PRODUCTION IN IBD

February 1992

tive activity (five fractions of each) were pooled, concentrated lo-fold using Minicon Bl5 filters, and assayed in a conventional colony-forming assay.” Briefly, purified BMSCs were cultured in the wells of leukocyte migration plates containing 400 PL of growth medium (IMEM, 20% FCS, and 0.3% agar) and 2 X lo4 cells. Each assay was performed in triplicate, and BMSCs were grown in the presence of different CSFs at saturating doses (50 ED,, units) determined by titration. After 7 and 14 days, the colonies were counted by dark-ground illumination, and the gels were placed on glass microscope slides and dried at 37°C. Gels were then fixed in methanol before staining with May-Grunwald Giemsa to identify granulocyte colonies, and alphanapthyl acetate esterase and napthol AS-D chloroacetate esterase (Sigma Chemical Co.) for identification of granulocyte and macrophage colonies. To enable direct comparison of component CSF activities in conditioned media generated with equivalent numbers of cells and to allow for variation between BMSC assays, the following calculation was made. The proliferative response for each sample (in counts per minute), adjusted relative to the peak proliferation to a GM-CSF standard for that assay, was calculated according to the formula Test Fraction cpm - Spontaneous cpm x loo ’ Maximum cpm - Spontaneous cpm

mercially available rIL-2 (Janssen, Beerse, Belgium) and rIL-1 (Roche, Basel, Switzerland) were used as standards. Molecular Probing RNA Expression

of IL-1 and IL-2

IL-l activity was assayed using the concanavalin A (Con A)-activated thymocyte assay. Briefly, 10’ thymocytes from C3H/HeJ mice were cultured in 96-well microtiter plates (Nunc) containing 10 pg/mL of Con A and serial dilutions of culture supernatant in Dulbecco’s modified Eagle medium with additives. After 3 days of culture, the cells were pulsed with 0.5 pCi of tritiated thymidine for 6 hours, harvested on glass paper, and counted in a liquid scintillation counter. IL-l was also assayed using a commercially available IL-l enzyme-linked immunosorbent assay kit (Cistron NJ) according to the manufacturer’s instructions. IL-2 was assayed using the IL-2-dependent mouse cell line CTLL-2 as previously described.17 Standard batches of recombinant human IL-2 (rIL-2) were simultaneously assayed and served as controls. Human

Recombinant

and Purified

Cytokines

Recombinant human IL-3 and GM-CSF were made by transfection of COS-1 cells” using appropriate expression vectors containing either the genomic clone of IL-3 (Dr. B. Van Leeuwen, John Curtin School of Medical Research)g or a complementary DNA clone of GM-CSF (G. Wong, Genetics Institute, Cambridge, MA).7 The IL-3 genomic clone was sequenced and found to be identical to that isolated by Yang et aL6 Purified G-CSF was prepared from medium conditioned by the human bladder carcinoma cell line (HBC)-5637 by previously reported methods.‘g~zo G-CSF fractions were tested for the presence of IL-1 activity using the Con A-stimulated thymocyte assay. Com-

for Cytokine

Messenger

Total RNA was extracted by centrifugation of cell lysates through CsCl using a method modified from that of Chirgwin et al.*l LPMCs (2 X 107) obtained from cell cultures after removal of the cytokine rich supernatant were lysed in 4 mL of 4 mol/L guanidine isothiocyanate and layered onto a 1.5-mL cushion of 5.7 mol/L CsCl in 5 mL polyallomer tubes (Beckman Corp., Palo Alto, CA) for centrifugation at 36,000 rpm for 20 hours at 20°C. The RNA pellet was purified by successive washes and precipitations in cold absolute alcohol, 7.5 mol/L guanidine hydrochloride, and finally cold absolute alcohol. The RNA pellet was resuspended in 50 FL of water and assayed by spectrophotometry (OD, 260). Absorbance at OD 280 was also measured to ensure that there was no significant contamination with protein. Twenty micrograms of total RNA was placed in each slot of a 1% agarose gel slab, electrophoresed, and then blotted onto nitrocellulose filters (Schleicher and Schuell, Dassel, Germany) as outlined by Davis et al.” using a slot-blot apparatus (Schleicher and Schuell). The NC filter was then sequentially probed with 32P-labeled probes specific for IL-3,’ GM-CSF,’ and G-CSF.’ Drug Regulation

Assay

531

of Cytokine

Production

Cyclosporin A, a gift from Sandoz (Basel, Switzerland), was supplied as a 50-mg/mL concentrate for intravenous infusion containing cremophore EL (65% wt/vol) and ethanol (26% wt/vol). A l/1000 dilution was made in Hank’s buffered salt solution (HBSS), and fresh dilutions made from this 50 Fg/mL stock were prepared for each experiment in a final concentration of 100 ng/mL. Stock solutions of hydrocortisone (10 mg/mL; Glaxo, Melbourne, Australia) were freshly prepared by dissolving the powder in HBSS containing 20 mmol/L HEPES (pH 7.4). Subsequent dilutions were made in medium containing 1% HABS to yield a final working concentration of 0.2 mg/ mL. A stock solution of 5-aminosalicylic acid (5-ASA; 5 mg/mL; Pharmacia, Sydney, Australia) was freshly prepared by dissolving in RPM1 1640 medium containing 20 mmol/L HEPES and adjusted to pH 7.4 with 0.1 mol/L NaOH. Final concentrations of 5-aminosalicylic acid used were 0.1-2.5 mg/mL (0.67-16.5 mmol/L). Replicate experiments for drug-treated and mediumonly control LPMC cytokine production were conducted in 96-well flat-bottom microtiter plates (Nunc). Both control and drug-exposed LPMCs were stimulated with PHA (40 pg/mL). Culture supernatants were harvested after 24 hours by centrifuging the plate (4OOg, 5 minutes) and aspirating the contents with Pasteur pipettes. Supernatants were stored at 4°C before assay for CSF activity. Statistics For statistical analysis, frequency distributions of the data were constructed. The data for CSF production by LPMCs were normally distributed, and the unpaired t test was applied. Data are given as the mean 4 SEM.

532 PULLMAN ET AL.

GASTROENTEROLOGY Vol. 102.No. 2

Results

4000-

LPMC Yields and Distribution

of Cell Types

Cell yields from the control colons, resected for nonmalignant/noninflammatory conditions, were 14.6 + 2.9 X lo6 cells/g mucosa, and from IBD colons, 30.6 +- 6.7 X lo6 cells/g mucosa. The cell yields for ileal tissue affected by CD were 27.5 f 7.9 X 10” cells/g mucosa, and for normal (control) ileal tissue, 18.9 -t 5.5 X lo6 cells/g mucosa. The proportion of adherent cells in the LPMC populations was relatively constant in the range of 10% f 3% for both the control and IBD colon groups. Very few adherent cells were found during the second 24-hour culture period, i.e., during cytokine production, and no significant difference was observed between LPMC, from IBD or control mucosa. This was consistent with previously reported data from this laboratory.” The proportion of lymphocytes was also similar between the two groups (control, 89% f 9%, IBD, 92% + 5%). CD3+ T lymphocytes constituted the major population of LPMCs, accounting for 45% k 3% of control and 48% f 5% of IBD-derived LPMCs. The proportion of activated T cells was, however, markedly increased in the IBD population as assessed by surface expression of IL-2R (IBD vs. control, 36% vs. 5%) and TFR (29% vs. 16%), which served as markers of T-lymphocyte activation. CSF Production

by Cultured LPMCs

Total CSF activity in conditioned medium from cultures of LPMCs obtained from nonmalignant/noninflammatory mucosa (control group) and IBD-affected tissue was determined using the BMSC assay. Constitutive CSF activity detected in the supernatants of unstimulated cultures was significantly higher for IBD-affected tissue (287 + 82 U) than for the control nonmalignant/noninflammatory group (23 + 6 U) (P < 0.5). The T-cell mitogen phytohemagglutinin (PHA) was used to further stimulate cultured LPMC cytokine production. PHAstimulated LPMCs from IBD-affected tissue expressed significantly greater levels of CSF activity (1402 f 252 U) than those from the control group (362 k 85 U) (P < 0.05) (Figure 1). When the IBD group was divided into UC and CD, there was significantly elevated total CSF production by PHA-stimulated LPMCs from the UC (1378 f 355 U; P < 0.05) and CD (1411 + 343 U; P < 0.05) groups compared with control LPMCs. Slightly higher, but not statistically significant, CSF production was found for the CD group compared with the UC group. There were no significant differences between LPMCs from ileal and coionic tissue (data not shown). These results indicate that CSFs are produced constitutively by LPMCs, which are capable of much

3000

??

2 z z 2ooom 5 %

”

1000

1.

_I___

IL A t

04

CONTROL

IBD

Figure 1. CSF activity produced by LPMCs. LPMCs from control normal (m, n = 13)and IBD-affected (A [CD], n = 10; ??[UC], II = 4) were cultured and stimulated with a a-hour pulse of PHA (40 pg/mL). Supernatants were collected after a further 22 hours, and 50-pL aliquots were assayed in the BMSC assay. The results are expressed as end-point units for 4 X 10’ LPMCs/mL conditioned medium. Significantly greater CSF production was found for LPMCs from both IBD groups (P < 0.05) compared with controls. - - -, Mean and SE.

greater CSF production after PHA stimulation. Furthermore, LPMCs from IBD mucosa generated significantly greater levels of CSF than noninflamed mucosa both constitutively and after PHA stimulation. Because IL-2 increases CSF production by some cell lines,23 the effect of IL-2 on LPMC CSF production was studied. CSF production increased markedly in a concentration-dependent manner when LPMCs were exposed to increasing amounts of IL-2 (Figure 2). Characterization by LPMCs

of Specific CSFs Produced

Preliminary separation and characterization of CSF activity was achieved using FPLC phenyl-superose hydrophobic interaction chromatography. Replicate elution profiles were obtained for PHAstimulated LPMC supernatants from two normal control colons, two UC- and two CD-affected colons. Each fraction was assayed for CSF activity in the BMSC assay. Individual CSFs were measured by comparing the elution profile obtained by FPLC of each sample with the profiles obtained using known amounts of each CSF standard. The FPLC profiles of the CSF standards varied by no more than one frac-

February

MUCOSAL CYTOKINE PRODUCTION

1992

IN IBD

533

Analysis of CSF-Specific Messenger RNA in LPMCs

IL-2

(U/ml)

Figure 2. IL-2 induced LPMC CSF production. LPMCs were cultured at 4 X lO’/mL with stimulating doses of rIL-2 at concentrations of 20,100, and 500 U/mL. The control cultures were given no IL-2. The culture supernatants were collected after 24 hours and assayed for CSF activity (n = 6).

To ensure that the measured increases in CSF production in LPMC supernatants from IBD mucosa were due to de novo synthesis of messenger RNA (mRNA) for G-CSF, IL-3, and GM-CSF, and were not the result of artifacts due to cell separation and culture, slot-blot analysis of RNA using complementary DNA probes for G-CSF, IL-3, and GM-CSF was performed on LPMC RNA preparations from CD. (Figure 4). Using equivalent amounts of total RNA (20 pg), densitometry scanning revealed that substantially

-\ \ 80 100 f

\

\

\

\

\

60-

tion throughout these experiments, showing the reproducibility of the FPLC-fractionation procedure. Representative elution profiles for PHA-stimulated culture supernatants are shown in Figure 3. The activity in these peaks (counts per minute incorporated) was up to fourfold greater than for the unstimulated supernatants, which had similar profiles. The elution profile of the culture supernatants for LPMCs from control normal colon resected for noninflammatory/nonmalignant conditions revealed three distinct peaks corresponding, from left to right, to GM-CSF, IL-3, and G-CSF (Figure 3). Representative elution profiles for LPMC supernatant from IBDaffected tissue, including UC and CD, showed four substantial peaks. The three largest peaks in the elution profile were identified as GM-CSF, IL-3, and GCSF by comparison with the positions of CSF standards using rGM-CSF, rIL-3, and purified G-CSF. For equivalent numbers of LPMCs, the standardized proliferative activity (counts per minute) for G-CSF was substantially greater in culture supernatants from CD and UC LPMCs than in the corresponding supernatants from noninflamed tissue LPMCs. The GMCSF peaks were broader in the CD and UC profiles, and maximum proliferation appeared to be slightly decreased especially in CD, whereas IL-3 activity was markedly depressed in both the IBD profiles. The fourth peak, which was not evident in control LPMC supernatants, was identified as IL-l using an enzyme-linked immunosorbent assay. Like G-CSF, IL-l was more pronounced in supernatants from CD LPMCs than in supernatants of UC LPMCs (Figure 3). The peaks of proliferative activity were also assayed for colony forming activity (Table 1). Peak 1 and peak 2 activities were consistent with GM-CSF and IL-3, whereas peak 3 generated pure granulocyte colonies, confirming the presence of G-CSF.

\

\

\

\

40.

20.

07 0

100.--,

5

10

15

20

25

5

10

15

20

25

\

.A

0

1001-,

U.C.

.

t 1oc

a0 60

40

20

0 0

5

10 FRACTION

15

20

25

NUMBER

Figure 3. Representative elution profiles for PHA-stimulated LPMCs from control and IBD-affected tissue. PHA-stimulated LPMC supernatants derived from controls (Con, n = 2) and patients with CD (n = 2) and UC (n = 2) were separated on FPLC phenyl-superose. CSF proliferative activity in the fractions was measured using the BMSC assay. The profiles represent the mean of the results for each fraction. The peaks were identified by the use of r-CSF standards and by colony-forming assays. In the elution profiles, major peaks are apparent, corresponding (left to right) to GM-CSF (1) IL-3 (2) G-CSF, (3) and IL-1 (4).

GASTROENTEROLOGYVol. 102.No.2

534 PULLMAN ET AL.

Table

1. Colony Formation FPLC Separations

by Pooled

Fractions

From

% Colony type Peak

Colony numbers”

G

M

GM

38.0f 1.7 36.3 k 3.4 24.3 rt 1.2

40 77 100

26 15 0

34 8 0

1 2 3

“Per 2 X 104, expressed

G, granulocyte;

as mean + SEM (n = 3). M, macrophage; GM, granulocyte-macrophage.

more G-CSF mRNA was expressed by LPMCs from CD LPMC (area under curve, 2.17) than by the LPMC RNA from histologically normal control mucosa [area under curve, 1.11). A reciprocal change was observed for the amount of IL-3, which was markedly reduced in CD compared with normal LPMC (Figure 4). The amounts of mRNA were slightly reduced for GM-CSF in CD. These findings therefore support the data from chromatographic separation of CSF activities obtained in the biological assays, showing increased levels of G-CSF and decreased amounts of IL-3. Drug Regulation

of CSF Production

Three drugs, 5-ASA, hydrocortisone, closporin A, were examined for their

and cypotential

effects on LPMC CSF expression. 5-ASA and corticosteroids are used in IBD therapy, whereas the therapeutic value of cyclosporin A, which suppresses IL-2 and IL-3 production, is currently being investigated. Hydrocortisone at a dose of 0.2 mg/mL (representing the mean serum level for a TO-kg man given 100 mg intravenously 4 times a day) produced complete suppression of CSF production by LPMCs (Figure 5). A substantial reduction in CSF activity was also induced by 5-ASA at a dose of 0.5 mg/mL (3.3 mmol/L) to 31% of control levels. This dose represents a midrange therapeutic level for fecal 5-ASA concentration.24 Cyclosporin A at a dose of 100 ng/mL, which represents a midrange serum therapeutic concentration, barely affected LPMC supernatant CSF levels (96% of control levels). None of the three drugs impaired the proliferative responses in the BMSC assay when tested at the same concentrations as those used in the assays of CSF production by LPMCs, indicating that the observed reductions in LPMC cytokine production were directly attributable to the actions of hydrocortisone and 5-ASA. The suppression induced by 5-ASA at 0.5 mg/mL was partial, but higher concentrations, up to the upper limit of expected 5-ASA concentrations (2.5 mg/mL) in fecal fluid, suppressed CSF production in a dose-dependent manner. Complete suppression was observed at concentrations in excess of 2 mg/mL (Figure 6). Discussion

LPMC Source GM-CSF

This article reports that intestinal LPMCs from tissue affected by CD and UC produce significantly elevated levels of G-CSF activity, on a per-cell

100

Figure 4. Slot-blot analysis of RNA from control normal and CD-affected tissue. Total RNA was extracted from equivalent numbers of PHA-stimulated LPMCs from histologically normal mucosa (control) and CD-affected tissue. Twenty micrograms of RNA was loaded into each slot on the agarose gel, electrophoresed, and then applied to a nitrocellulose filter. The presence of mRNA for CSFs was detected by probing with 32P-labeled complementary DNA probes specific for G-CSF, IL-3, and GM-CSF. The intensity of the autoradiograph “slot” indicates the amount of specific mRNA present.

-

!i-ASA

HC DRUG TREATMENT

Figure 5. Effect of drugs on LPMC CSF production. LPMCs (4 X lO’/mL) were treated with 5-ASA, 0.5 mg/mL; hydrocortisone, 0.2 mg/mL, or cyclosporin A, 100 ng/mL, and stimulated with PHA. The culture supernatants were collected after 24 hours and assayed for CSF production, which is expressed as a percent of the non-drug-treated control.

MUCOSAL

February 1992

09000 7

0.500

+\.-. 1 .ooo

CONCENTRATION

1.500

OF 5-ASA

2.000

2.500 *

( mg/ml )

Figure 6. Dose-response curve for 5-ASA effect on CSF production. LPMCs were treated with a range of 5-ASA concentrations from 0.1 to 2.5 mg/mL and stimulated with PHA. The culture supernatants collected after 24 hours were assayed for CSF activity and expressed as a percent of the control, non-drug-treated group (n = 6).

basis, compared with histologically normal intestinal mucosa both constitutively and after stimulation by PHA treatment. This enhanced production occurs in the absence of any significant alteration in the composition of LPMCs, particularly with respect to the proportions of adherent cells and T lymphocytes between the IBD and the control noninflamed tissue groups. These findings of similar proportions of mucosal lymphocytes between IBD-affected mucosa and control mucosa are supported by published data on disaggregated mucosa25*26*27 and by immunohistochemical studies of intact mucosa.27 Thus, the results reflect the altered ability of LPMCs from IBD tissue to produce CSFs and implies an enhanced state of activation of LPMCs in the IBD-affected group. Coupled with the overall increase in numbers of inflammatory cells observed in IBD tissue, these results imply that in situ production of CSF is markedly elevated in IBD intestinal lamina propria. The increased CSF levels are mainly the result of increases in G-CSF and in IL-l production. By contrast, both GM-CSF and IL-3 levels appear to be decreased in IBD LPMCs, whereas GM-CSF levels appear to be slightly reduced. These findings are confirmed by slot-blot analysis of specific mRNA. The altered shape of the GM-CSF peak in both UC and CD may represent the activities of other growth factors such as IL-4 or IL-5, which have not been assayed in this study. Among the cytokines identified as products of LPMCs, GM-CSF and G-CSF can act synergistically to induce proliferation and maturation of myeloid stem cells to form colonies of mature granulocytes and macrophages in vitro and in vivo.3.28 These cytokines also activate mature granulocytes, eosinophils,

CYTOKINE

PRODUCTION

IN IBD

535

and macrophages, generate chemotactic activity and induce phagocytic activity, granulocyte cytotoxicity, antibody-dependent cellular cytotoxicity, superoxide generation, and release of leukotrienes and prostaglandins.2g-3B CSFs produced in peripheral tissues may also contribute to circulating CSF levels and hence may act on bone marrow precursors.3g IL-l stimulates monocytes to produce hemopoietic growth factors including G-CSF and M-CSF, resulting in substantial activation, accumulation, and production of granulocytes and macrophages. IL-l, which is produced predominantly by activated monocytes and macrophages, plays a central part in inflammatory responses by releasing endogenous pyrogen, inducing acute-phase protein synthesis, activating T lymphocytes, increasing antibody synthesis, and enhancing collagen production by fibroblasts.40 The increased levels of LPMC-produced G-CSF and IL-l, therefore, have the potential to maintain circulating myeloid cell numbers and to attract peripheral blood monocytes, neutrophils, and eosinophils to the inflammatory lesion of IBD. Once localized to the intestinal mucosa, these cells may be activated by locally produced CSFs. Although the generation of cytotoxic and antimicrobial functions by neutrophils and monocytes contributes to the host defenses, the release of tumor necrosis factor, free oxygen radicals, and IL-l by activated neutrophils and monocytes may also contribute to mucosal damage in IBD. The substantial levels of constitutive (unstimulated) secretion of CSFs by LPMCs found in IBD may reflect production by activated macrophages and lymphocytes induced by the presence of endotoxin and by the mucosal immune response to antigens crossing the damaged epithelial barrier. The LPMCs from IBD-affected tissue are in a state of enhanced activation compared with noninflamed control LPMCs, as supported by the data in this study indicating greater surface expression of the activation markers IL-2R and TFR in the IBD group. These findings of enhanced activation in IBD mucosa have been described for a range of markers including the expression of the 4F2 marker of lymphocyte activation.41,42 The striking induction of CSF expression by mitogens and by IL-2 reveals the capacity of the LPMC system to respond to immune and inflammatory challenge. This phenomenon has been shown to occur in other systems in response to specific antigen4 and endotoxin exposure.43 Induction of CSF production by IL-Z further implicates the activated T cell as having a regulatory role in the mucosal inflammatory response. The production of GM-CSF and greatly enhanced G-CSF by LPMCs from IBD-affected tissue correlates with the influx of neutrophils into the IBD lesion as

536

PULLMAN ET AL.

shown by the ““Tc-phagocyte and “‘In-neutrophil scans.2’44 The finding of substantial amounts of IL-l in LPMC supernatants from CD and from UC tissue reported previously45 and recently confirmed,45 together with the enhanced levels of G-CSF, another mononuclear phagocyte product, also correlate with the observed recruitment of monocytes to the lesion of IBD46 and consequently the enhanced numbers of monocytes and macrophages shown immunohistologically in inflamed tissue.47,48 If these cells are activated in the mucosa by exposure to endotoxin or to locally produced cytokines, further cytokine release including IL-l and G-CSF may occur, amplifying further the recruitment of both monocytes and granulocytes to the lesion and their activation, resulting in the sustained, chronic damage characteristic of the intestinal lesion of IBD. The production of macrophage CSF by macrophages, fibroblasts, and endothelial cells may also be relevant to monocyte-macrophage activation and turnover in IBD but has not been investigated in this study. The finding that 5-ASA and hydrocortisone at therapeutic levels suppress CSF production by LPMCs may be important to the mechanisms by which these agents exert their therapeutic effect in IBD. To our knowledge, this is the first report that 5-ASA influences cytokine production. The failure of cyclosporin A to suppress CSF production significantly is not unexpected, because in other systems, cyclosporin A suppresses IL-Z and IL-3 production but not that of GM-CSF or G-CSF.4g-51 Moreover, the T cells found in the LPMCs may already be activated so that inhibition of IL-2 production can have little effect. This selective cytokine suppression may well limit any potential therapeutic efficacy of cyclosporin A in IBD, particularly if G-CSF plays a major role in the mediation of mucosal injury. The specificity of action of each drug on the production of individual CSFs deserves further study because of its relevance to therapeutic efficacy. The results presented in this study show that intestinal lamina propria cells from inflamed IBD mucosa produce substantially increased amounts of CSF activity, particularly of the macrophage-derived cytokines G-CSF and IL-1 rather than those of T-cell origin such as GM-CSF and IL-3, which are decreased compared with in LPMCs from normal noninflamed mucosa. These findings are consistent with an inflammatory reaction driven by endotoxin entry though a damaged mucosal epithelium resulting in increased G-CSF and IL-l levels that are amplified further by IL-Z. T cell-derived GM-CSF and IL-3, together with macrophage-derived G-CSF and IL-l, therefore, contribute to a complex network of cytokine-induced signals that is capable of inducing and maintaining the chronic inflammatory lesion of IBD.

GASTROENTEROLOGY Vol. 102, No. 2

References 1. Pullman WE, Hanna R, Sullivan P, Booth JA, Lomas F, Doe WF. Technetium-99m autologous phagocyte scanning-a new imaging technique for inflammatory bowel disease. Br MedJ 1986;293:171-174. 2. Pullman WE, Sullivan P, Barratt PJ, Lising J, Booth JA, Doe WF. Assessment of inflammatory bowel disease activity by technetium 99m phagocyte scanning. Gastroenterology 1988;95:989-996. 3. Metcalf D. The molecular biology and functions of the granulocyte-macrophgage colony stimulating factors. Blood 1986; 67:257-267. 4. Metcalf D. The role of the colony-stimulating factors in resistance to acute infections. Immunol Cell Biol 1987;65:35-43. 5. Clark SC, Kamen R. The human haematopoietic colonystimulating factors. Science 1987;236:1229-1237. 6. Yang Y-C, Ciarletta AB, Temple PA, Chung MP, Kovacic S, Witek-Giannotti JS, Leary AC, Kriz R, Donahue RE, Wong GG, Clark SC. Identification by expression cloning of a novel haematopoietic growth factor related to murine IL-3. Cell 1986;47:3-10. 7. Wong GG, Witek JS, Temple PA, Wilkens KM, Leary AC, Luxenberg DP, Jones SS, Brown EC, Kay RM, Orr EC, Shoemaker C, Golde DW, Kaufman RJ, Hewick RM, Clark SC, Wang EA. Human GM-CSF. Molecular cloning of the cDNA and purification of the natural and recombinant products. Science 1985;228:810-815. 8. Nagata S, Tsuchiya M, Asano S, Kaziro Y, Vakmazaki T, Yamamoto D, Hirata Y, Kubota N, Oheda M, Homura H, Ono M. Molecular cloning and expression of cDNA for human granulocyte colony-stimulating factor. Nature 1986;319:415-417. 9. Kobayashi M, van Leeuwen BH, Elsbury S, Martinson ME, Young IG, Hapel AJ. Interleukin-3 is significantly more effective than other colony stimulating factors in long-term maintenance of human bone marrow-derived colony forming cells in vitro. Blood 1989;73:1836-1841. 10. Cannistra SA, Griffin JD. Regulation of the production and function of granulocytes and monocytes. Semin Oncol 1988;25:173-188. 11. Metcalf D. Haemopoietic growth factors. Med J Aust 1988; 148:516-519. 12. Golder JP, Doe WF. Isolation and preliminary characterization of human intestinal macrophages. Gastroenterology 1983;84:795-802. 13. Bull DM, Bookman MA. Isolation and preliminary characterization of human intestinal mucosal lymphoid cells. J Clin Invest 1977;59:966-974. 14. Boyum A. Separation of leukocytes from blood and bone marrow. Stand J Clin Lab Invest 1968;21:31-50. 15. Pullman WE, Elsbury S, Kobayashi M, Hargreaves J, Preston L, Van Leeuwen B, Hapel AJ. Rapid assay and identification of human haemopoietic growth factors. J Immunol Methods 1989:117:153-161. 16. Bradley TR, Metcalf D. The growth of mouse bone marrow cells in vitro. Aust J Exp Biol Med Sci 1966;44:287-299. 17. Gillis S, Ferm M, Ou W, Smith K. T cell growth factor: parameters of production and a quantitative microassay for activity. J Immunol 1978;120:2027-2032. 18. Parker BA, Stark GR. Regulation of Simian Virus 40 transcription: sensitive analysis of the RNA species present early in infections by virus or viral DNA. J Viral 1979;31:360-369. 19. Nicola NA, Metcalf D, Johnson GR, Burgess AW. Separation of functionally distinct human granulocyte-macrophage colony-stimulating factors. Blood 1979;54:614-626. 20. Welte K. Platzer E. Lu L. Gabrilove IL. Loui E. Mertelsmann R. Moore MAJ. Purification and biochemical characterization of

February

1992

human pluripotent hematopoietic colony-stimulating factor. Proc Nat1 Acad Sci USA 1985;82:1526-1530. 21. Chirgwin JM, Przybyla AE, MacDonald RJ, Rutter WJ. Isolation of biologically active ribonucleic acid from sources enriched in ribonuclease. Biochemistry 1979;18:5294-5299. 22. Davis LG, Dibner MD, Battey JF. Basic methods in molecular biology. New York: Elsevier, 1986. regulation of 23. Kelso A, Metcalf D, Gough NM. Independent granulocyte-macrophage colony-stimulating factor and multi-lineage colony-stimulating factor production in T lymphocyte clones. J Immunol 1986;137:764-769. of 24. Klotz U, Maier KE. Pharmacology and pharmacokinetics 5-aminosalicylic acid. Dig Dis Sci 1987;32:46S-503. of mucosal lympho25. Clancy R. Isolation and characterization cytes in Crohn’s disease. Gastroenterology 1976;70:177-180. 26. Fiocchi C, Battisto JR, Farmer RB. Gut mucosal lymphocytes in inflammatory bowel disease. Isolation and preliminary functional characterization. Dig Dis Sci 1979;24:705-717. 27. Selby WS, Janossy G, Goldstein G, Jewel1 DP. Intestinal lymphocyte subpopulations in inflammatory bowel disease: an analysis by immunohistological and cell isolation techniques. Gut 1984;25:32-40. 28. Begley CG, Nicola NA, Metcalf D. Proliferation of normal human promyelocytes and myelocytes after a single pulse stimulation by purified GM-CSF or G-CSF. Blood 1988;71:640-645. 29. Socinski MA, Cannistra SA, Elias A, Antman AH, Schnipper L, Griffin JD. Granulocyte-macrophage colony stimulating factor expands the circulating haemopoetic progenitor cell compartment in man. Lancet 1988;1:1194-1198. factor30. Becker S, Warren MK, Haskill S. Colony-stimulating induced monocyte survival and differentiation into macrophages in serum-free cultures. J Immunol 1987;139:37033709. 31. Gasson JC, Weisbart RM, Kaufman SE, Clark SC, Hewich RM, Wong GG, Golde DW. Purified human granulocyte macrophage colony-stimulating factor: direct action on neutrophils. Science 1984;226:1399-1342. 32. Handman E, Burgess AW. Stimulation of granulocyte macrophage colony stimulating factor of Leishmania tropica killing by macrophages. J Immunol1979;112:1134-1137. 33. Silberstein DS, Owen WF, Gasson JC, Di Persion JF, Golde DW, Bina JC, Soberman R, Austen KF, David JP. Enhancement of human eosinophil cytotoxicity and leukotriene synthesis by biosynthetic (recombinant) granulocyte-macrophage colony stimulating factor. J Immunol 1986;137:3290-3294. 34. Vadas MA, Varigos G, Nicola N, Pincus S, Dessein A, Metcalf D, Battye FL. Eosinophil activation by colony stimulating fadtor in man: metabolic effects and analysis by flow-cytometry. Blood 1983;61:1232-1241. 35. Vadas MA, Nicola NA, Metcalf D. Activation of antibody dependent cell-mediated cytotoxicity of human neutrophils and eosinophils by separate colony stimulating factors. J Immunol 1983;130:795-799. 36. Weisbart RH, Golde DW, Clark SC, Wong GG, Gasson JC. Human granulocyte-macrophage colony-stimulating factor is a neutrophil activator. Nature 1985;314:361-363. 37. Weisbart RM, Kwan L, Golde DW, Gasson JC. Human GM-CSF primes neutrophils for enhanced oxidative metabolism in response to the major physiological chemoattractants. Blood 1987;69:18-21. 38. Weisbart RM, Kacena A, Schuh A, Golde DW. GM-CSF induces human neutrophil @-mediated phagocytosis by an

MLJCOSAL CYTOKINE PRODUCTION

IN IBD

537

IgA Fc receptor activation mechanism. Nature 1988;332:647648. 39. Sheridan JW, Metcalf D. Studies on bone marrow colony stimulating factor (CSF): relation of tissue CSF to serum CSF. J Cell Physiol 1972;80:129-140. 40. Dinarello CA. Interleukin-1 and its biologically related cytokines. Adv Immunol 1989;44:153-205. 41. Fiocchi C, Battisto JR, Farmer RG. Studies on isolated gut mucosal lymphocytes in inflammatory bowel disease. Detection of activated T cells and enhanced proliferative response to staphylococcus aureus and lipopolysaccharides. Dig Dis Sci 1981;36:728-736. 42. Pallone F, Fais S, Squarcia 0, Biancone L, Pozzilli P, Boirivant M. Activation of peripheral blood and intestinal lamina propria lymphocytes in Crohn’s disease. In vivo state of activation and in vitro response to stimulation as defined by the expression of early activation antigens. Gut 1987;28:745-753. 43. Nicola NA, Burgess AW, Metcalf D. Similar molecular properties of granulocyte-macrophage colony-stimulating factors produced by different mouse organs in vitro and in vivo. J Biol Chem 1979;254:5290-5299. 44. Saverymuttu SH, Peters AM, Crofton ME, Rees H, Lavender JP, Hodgson HJ, Chadwick VS. “‘Indium autologous granulocytes in the detection of inflammatory bowel disease. Gut 1985;26:955-960. 45. Pullman WE. Analysis of leucocyte functions in inflammatory bowel disease. Ph.D. thesis, Australian National University, 1988. 46. Mahida YR, Wu K, Jewel1 DP. Enhanced production of interleukin lb by mononuclear cells isolated from mucosa with active ulcerative colitis and Crohn’s disease. Gut 1989; 30:835-838. 47. Hume DA, Allan W, Hogan PG, Doe WF. Immunohistochemical localisation of the macrophage antigen 25F9 in intestinal mucosa and associated lymphoid tissue. J Leukoc Biol 1987;42:474-484. 48. Mahida YR, Pate1 S, Gionchetti P, Vaux D, Jewel1 DP. Macrophage subpopulations in lamina propria of normal and inflamed colon and terminal ileum. Gut 1989;30:326-834. 49. Bickel M, Tsuda M, Amstad P, Evequoz V, Mergenhagen SE, Wahl SM, Pluznik DM. Differential regulation of colony stimulating factors and interleukin 2 production by cyclosporin A. Proc Nat1 Acad Sci USA 1987;84:3274-3277. 50. Granelli-Piperno A, Andrus L, Steinman RM. Lymphokine and nonlymphokine mRNA levels in stimulated human T cells. Kinetics, mitogen requirements and effects of Cyclosporin A. J Exp Med 1986;163:922-937. 51. Orosz C, Roopenian D, Widmer M, Bach FH. Analysis of cloned T cell function II. Differential blockade of various cloned T cell functions by cyclosporine. Transplantation 1983;36:706-711. Received April 26, 1990. Accepted July 29, 1991. Address requests for reprints to: William F. Doe, Division of Clinical Sciences, John Curtin School of Medical Research, Australian National University, Box 334, Canberra, Australian Capital Territory 2601,Australia. Supported by research fellowships from the Saw Medical foundation (University of Western Australia) and the National Health and Medical Research Council (to W.E.P.). The authors thank the surgeons and pathologists from Woden Valley and Royal Canberra Hospitals for providing samples of human colon, Lynette Preston for her technical help with the hybridization studies, and Anabel Walters for typing the manuscript.