JOURNAL OF PATHOLOGY, VOL. 178: 201-206 (1996)

DIFFERENTIAL IN SITU EXPRESSION OF THE

RANTES IN HUMAN INFLAMMATORY BOWEL DISEASE

GENES ENCODING THE CHEMOKINES MCP-1 AND

LUCA MAZZUCCHELLI*, CHANTAL HAUSER*, KASPAR ZGRAGGEN~, HANS E. WAGNER$, MAX w. HESS*, JEAN A. LAISSUE* AND CHRISTOPH MUELLER*

*Institute of Pathology and $Department of Surgery, University of Bern, and ?Department of Surgery, Tiefenauspital, Bern, Switzerland

SUMMARY Two chemotactic cytokines, monocyte chemoattractant protein-1 (MCP-1) and RANTES, possibly contribute to the recruitment and

activation of leukocytes in inflamed tissues. The expression of these cytokine genes was evaluated in tissue sections from resected bowel segments of 14 patients with inflammatory bowel disease (IBD) and seven control patients by use of "S-labelled antisense RNA probes. MCP-1 and RANTES transcripts were generally increased in the intestinal mucosa of patients with IBD, compared with controls. Whereas MCP-1 gene expression in the mucosa was restricted to the lamina propria, the gene coding for RANTES was expressed in intraepithelial lymphocytes and in the subepithelial lamina propria. Furthermore, MCP-1 mRNA, but not RANTES mRNA, was abundant in vessel-associated cells, such as endothelial cells, medial smooth muscle cells, and intraluminal cells; in smooth muscle cells of the intestinal tunica muscularis; and in cells of the myenteric plexus. Compared with controls, a significant increase of MCP-1-expressing cells was observed in tissue specimens from patients with IBD, in endothelial cells of venules, and in cells present in the lumen of intestinal vessels. Conversely, the expression of MCP-1 mRNA in smooth muscle cells and myenteric plexus cells appeared to be comparable in control and diseased intestines. The increased number of MCP-1 and RANTES mRNA-expressing cells in mucosa from patients with IBD suggests that these cytokines play a role in the pathogenesis of mucosal inflammation. Furthermore, the expression of the MCP-1 gene in vessel-associated cells may indicate its involvement in mechanisms regulating the adhesion of blood monocytes to endothelial cells.

KEY WORDS-inflammatory bowel disease; cytokine; in situ hybridization

INTRODUCTION

Over the past few years, a new superfamily of chemo- tactic peptides with a rather restricted target cell specifi- city has been discovered.' These pro-inflammatory cytokines, called chemokines, are grouped into two structurally distinct families according to the position of the first two cysteine residues in the conserved motif, which are either adjacent (C-C family) or separated by one amino acid (C-X-C family). Most members of the C-X-C family, which include interleukin-8 (IL-8)/ neutrophil-activating protein- 1 (NAP- l), NAP-2, macrophage inflammatory protein-2 (MIP-2), and ENA 78, are rather specific chemoattractant and activating cytokines for neutrophils.2 Conversely, C-C chemokines such as monocyte chemoattractant protein-1 (MCP-1),3 MCP-2 and MCP-3,4 RANTES (regulated upon acti- vation, normal T expressed, and presumably ~ecreted),~ human macrophage inflammatory protein (HuMIP-a) and HuMIP-P, 1-309, and C102 were originally described as chemotactic peptides for different mononuclear cell types, such as, macrophages/monocytes, eosinophils, and different lymphocyte subsets. They do not affect neutrophil chemotaxis.

Addressee for correspondence: Luca Mazzucchelli, MD, Institute of Pathology, University of Bern, Murtenstr. 31, CH-3010 Bern, Switzerland.

CCC 0022-341 7196lO2020 1-06 0 1996 by John Wiley & Sons, Ltd.

MCP-1 exerts specific chemotactic activity for mono- cytes and lymphocytes.&* Furthermore, it is a potent histamine-releasing factor for basophils.' MCP- 1 is secreted constitutively, or upon stimulation by inflam- matory mediators such as IL-1 or tumour necrosis factor-a (TNF-a), by many malignant tumour cell lines and by different leukocyte subsets, as well as by endo- thelial cells, smooth muscle cells, and keratinocyte~.~

Like MCP-1, RANTES is a chemoattractant for monocytes" and causes the release of histamine from basophils." The RANTES protein also exerts in vitro chemotactic activity for eosinophils'2 and for T-cells of the memory/helper phenotype (CD4+/CD45RO)." Conversely, RANTES does not appear to influence B-lymphocyte chemotaxis.

Previous studies revealed ex ression of RANTES mRNA in synovial lining cellJ3 and synovial fibro- b l a s t ~ . ' ~ In cell-mediated transplant rejection of the kidney, it is expressed in mononuclear cells, renal tubular epithelium, and vascular end~thelium.'~

The main aims of this study were to analyse the expression of MCP-1 and RANTES genes in situ in intestinal specimens from controls and from patients with inflammatory bowel disease (IBD); to investigate whether these cytokines with overlapping biological properties are differentially expressed in normal and diseased intestine; and to identify possible cellular sources of these chemotactic peptides in situ.

Received 26 January 1995 Accepted 2 May 1995

202 L. MAZZUCCHELLI ET AL.

MATERIALS AND METHODS Tissues

Seven patients with Crohn's disease (CD; three male and four female; mean age 35.7 years) and seven patients with ulcerative colitis (UC; five male and two female; mean age 48.3 years) were analysed. In the group with CD, the site of disease was ileal (n=2), ileocolonic (n=4), and colonic (n=l) . The indication for bowel resection was the presence of strictures or fistulae. In the group with UC, six patients had a severe pancolitis refractory to medical management and one patient had colorectal cancer and chronic pancolitis. Control tissues (five specimens from the small bowel and six from the colon) were obtained from patients with colon cancer (n=4), stomach cancer (n= l), acute pancreatitis (n= l), and diverticular disease of the sigmoid colon (n= l ) . All tissue specimens were obtained at surgery. Blood supply was interrupted approximately 3 0 4 5 min before the resected bowel segments were sampled. Each case was documented with one to four transmural specimens (mean 3.2) taken from bowel segments with active disease. In patients with cancer, macroscopically normal small intestinal and colonic samples were chosen in areas remote from neoplastic lesions. Tissue specimens for in situ hybridization (ISH) were snap-frozen immediately after removal and stored at - 70°C; adjacent tissue specimens were processed for conventional diagnostic histopathological assessment.

Immunohistochemistry To study the phenotype of cells expressing the MCP-1

and RANTES genes, semi-serial frozen sections from selected cases were immunostained. Monoclonal anti- bodies directed against the T-cell markers CD3, CD4, and CD8; the macrophage marker CD68; and the B-cell marker CD20; and biotinylated secondary antibodies and avidin-biotin complex reagents were used, accord- ing to the instructions provided by the manufacturer (Dakopatts, Copenhagen, Denmark).

Preparation of 35S-labelled RNA probes and in situ hybridization

A 650 bp cDNA fragment of the MCP-1 gene (kindly provided by Dr Yoshimura, NCI, Frederick, MD, U.S.A.) and a 400 bp cDNA fragment of the RANTES gene (obtained from Genentech South, San Francisco, CA, U.S.A.) were subcloned into the transcription vec- tors pBluescript-KS+ and -SK -, respectively, by stan- dard techniques. After linearization of the plasmids with appropriate restriction endonucleases, "S-labelled anti- sense and sense RNA probes of comparable specific activity were prepared using T7 and T3 RNA polymer- ases Boehringer Mannheim, Germany), respectively, and 3'S-labelled CTP (Amersham, Arlington Heights, IL, U.S.A.).16 In situ h bridizations were performed as previously described. I6.h

Controls and evaluation of in situ hybridization As the negative control, non-hybridizing 35S-labelled

sense probe was used for each tissue section. Cells

hybridized with an antisense RNA probe were consid- ered positive for gene expression when they had at least twice as many silver grains as cells hybridized with the corresponding sense RNA probe. All mRNA-expressing cells were counted in ten different fields per tissue section (0.16 mm2 per field) chosen at random in the mucosa. To assess gene expression in blood vessels, ten randomly chosen cross-sections of venules, arterioles, or capillaries were examined per tissue section.

Statistical analysis

The Kruskal-Wallis test was used to test the overall significance of the differences of the results obtained in the three groups, controls, CD, and UC." If the test yielded significant differences (P<O.O5), a testing ana- logue of the Bonferroni pairwise comparison procedure based on the ranks of the observations was used to obtain information about the comparison of the single groups, i.e. CD vs. control and UC vs. control.''

RESULTS

MCP-I mRNA expression In the intestinal mucosa of control tissue, MCP-I

mRNA gene expression was occasionally found in mononuclear cells of the lamina propria (Table I). Cells expressing the MCP-1 gene showed a focal distribution pattern and were frequently located around mucosal lymphoid follicles (Fig. 1A). Compared with control tissues, the density of MCP- 1 -expressing cells in the lamina propria was increased in patients with CD and UC (Table I). The MCP-1 gene expression was not strictly related to histological signs of active inflam- mation. Labelled cells were located at the base of ulcers and in areas with severe neutrophilic infiltrates (Fig. lB), but also in intestinal mucosa with minimal signs of active (neutrophilic) inflammation. Tissue specimens from two patients with CD and one patient with UC showed only a few labelled cells in the mucosa, despite the presence of florid inflammation. Epithelial cells and intraepithelial lymphocytes (IELs), either in control tissues or in intestinal specimens from patients with IBD, completely lacked detectable MCP-1 mRNA.

In control tissues and in tissues from patients with IBD, approximately 60 per cent of the intestinal arteri- oles and venules showed MCP-1 gene expression in the vessel wall, in smooth muscle cells and/or in endothelial cells (Table I). Among labelled vessels within each group examined, positive endothelial cells were preferentially located in venules (Figs 1C and 1D). The percentage of venules with positive endothelial cells in patients with IBD was significantly higher than in controls (Table I). Furthermore, in tissue specimens from patients with CD and UC, the number of intravascular mononuclear cells expressing MCP- 1 mRNA was increased, compared with controls (Table I). Most of these cells were ident- ified as monocytes/macrophages (CD68-positive cells) by immunostaining of semi-serial sections. Finally, prominent MCP-1 gene expression was found in cells of the myenteric plexus region (Figs 1E and 1F) and in

MCP-1 AND RANTES GENES EXPRESSION IN IBD 203

Table I-MCP-1 or RANTES-expressing cells in intestinal specimens from control patients and patients with CD or UC*

Controls Crohn’s disease Ulcerative colitis P (7 patients) (7 patients) (7 patients) (Kruskal-Wallis test)

mRNA MCP-1 and RANTES- expressing cells in the mucosa (ceIls/mrn*)

MCP- 1 RANTES

Percentage of cross-sectioned vessels with mRNA MCP-1 -expressing cells

Percentage of arterioles and venules with positive endothelial cells

Arterioles Venules

mRNA MCP-1-positive cells in the lumen of vessels (per 100 cross-sectioned vessels)

0.9 (0-3.7) 6.3 (3.9-17)

64.3 (0-100)

0 (0-50) 37.5 (0-80)

0 (0-16.6)

8.5 (1.0-17.8)t 20.6 (049) t

59.4 (21.4-67.5)

40 (0-100) 83 (lo-loo)?

10 (0-200)

15.7 (1 ’2-77)t 0.007 98 (0-242)f 0.03

58.9 (20-93.7) n s .

0 (0-100) n.s. 83 (lO-lOO)? 0.02

30 (10-210)t 0.035

*Median value (range) of MCP-1 or RANTES-expressing cells in the mucosa and the MCP-1-expressing cells in intestinal vessels, endothelial cells

tP<O.05 (CD or UC vs. control); calculated with Bonferroni adjusted significance of difference. of arterioles and venules, and vascular lumen.

smooth muscle cells of the tunica muscularis (data not shown). No significant differences were seen between normal and diseased intestine in this respect.

RA NTES mRNA expression

RANTES mRNA-expressing cells were present in the intestinal mucosa of all control specimens (Table I). Thirty-one five per cent of labelled cells in the mucosa were located in the epithelial layer. They were morpho- logically identified on hybridized slides as IELs. They were predominantly located within the epithelium at the top of the villi in the small bowel and at the colonic mucosal surface. The remaining positive cells of the mucosa were located in the lamina propria, but rarely in its deep layer (Figs 2A and 2B). RANTES tran- scripts were consistently found in cells within mucosal lymphoid follicles.

Compared with controls, the density of RANTES- expressing cells in the mucosa was generally increased in specimens from patients with CD and in tissues from patients with UC (Table I). The intestinal mucosa of one patient with CD and one patient with UC, how- ever, completely lacked cells with RANTES mRNA at a detectable level. The ratio of IELllamina propria labelled cells was similar to that observed in control tissues. Cells containing RANTES mRNA were pre- dominantly found in intestinal mucosa displaying signs of chronic inflammation and minimal active (neutro- philic) inflammation but, in contrast to MCP-1, rarely in mucosal areas with florid inflammation. Furthermore, in intestinal specimens from patients with CD and to a lesser extent in specimens from patients with UC, RANTES-expressing cells were located in the lymphoid follicles (predominantly composed of CD4-positive cells

with scattered CD8- and CD68-positive cells), present in the entire intestinal wall; scattered RANTES mRNA- expressing mononuclear cells were seen in the interstitial spaces of the tunica muscularis. Finally, prominent RANTES mRNA expression was observed in epithe- lioid cell granulomas present in tissue sections from two patients with CD (Figs 2C and 2D).

In contrast to MCP-1 mRNA, RANTES mRNA was never detected in smooth muscle cells, in vessel- associated cells, or in cells of the myenteric plexus either in control tissues or in specimens from patients with IBD.

Controls

None of the intestinal sections hybridized with the sense probes yielded a focal reactivity. Smooth muscle cells of the tunica muscularis and mucosal lymphoid follicles provided an excellent positive control, as they consistently expressed the MCP-1 and RANTES gene, respectively. No evidence was found for non-specific binding of the radiolabelled RNA probes to particular cell populations in the intestine.

DISCUSSION

These results show a numerical increase of cells expressing the MCP-1 and RANTES genes in the mucosa of patients with active IBD, compared with controls. These findings are in accordance with very recent studies2’ that found increased mRNA levels of these cytokines in the mucosal lesions of patients with IBD, after total cellular RNA extraction from colono- scopic biopsies. However, the in situ hybridizations of

204 L. MAZZUCCHELLI ET AL.

Fig. I-in sir i i hybridization with MCP-I antisense probe. (A) Normal colonic mucosa from a patient with colonic cancer: MCP- I -expressing cells are preferentially located around mucosal lymphoid follicles. x 400. (B) Inflammatory pseudopolyp of the mucosa from a patient with active UC: MCP-I-expressing cells are present in the subepithelial lamina propria. x 100. (C) Intestinal venules (top) and arterioles (bottom) of the tela subserosa from a patient with florid ulcerative colitis (H & E stain). (D) Note the preferential location of MCP-I transcripts in endothelial cells of venules. x 200. (E) Section of the myenteric plexus from a patient with mild active ulcerative colitis (H & E stain). (F) Prominent MCP-1 gene expression in cells of the myenteric plexus even in the absence of inflammatory infiltrates. x 400

the present study reveal that the distribution pattern of labelled cells in the mucosa is not strictly related to the histological signs of active (neutrophilic) inflammation. Thus, both cytokines presumably play a more important role in perpetuating chronic mucosal inflammation than in the initiation of the acute stages of disease.

The increased RANTES gene expression in the intes- tinal mucosa of patients with IBD, namely in lamina propria cells and in particular in IELs, is of special interest, since the RANTES-mediated chemotactic activ- ity for CD4 + /CD45RO + 'memory' T-lymphocytes may substantially contribute to the pathogenesis of IBD. The CD4-positive T-cell population in the lamina propria, through its production of various soluble mediators,

modulates the activation and function of other cell types, such as monocytes/macrophages, mast cells, basophils, and eosinophils. Thus, the recruitment of peripheral CD4+ T-cells may lead to up-regulation of mucosal immune responses and may intensify chronic inflammation. The observed expression patterns of the RANTES gene in granulomas in patients with CD most likely also indicates the involvement of this cytokine in delayed-type inflammatory reactions, as demonstrated in other granulomatous diseases.2'

The weak or absent MCP-1 or RANTES gene expres- sion in the mucosa of a few patients with active IBD is not easily understood; the prominent expression of both genes in extramucosal compartments of the intestinal

MCP-1 AND RANTES GENES EXPRESSION IN IBD 205

Fig. 2-In situ hybridization with RANTES antisense and sense probes. (A) The colonic mucosa from a patient with mild active Crohn’s disease with irregular surface and minimal signs of active (neutrophilic) inflammation shows positive cells predominantly located in the subepithelial lamina propria. (B) Dark-field image of the same section. x 100. (C) Epithelioid cell granuloma in the tela submucosa of an ileal section from a patient with Crohn’s disease: section hybridized with RANTES antisense probe and counterstained with nuclear fast red, showing a signal in centrally located multinuclear giant cells. (D) Semi-serial section hybridized with RANTES sense probe as a negative control and counterstained with haematoxylin to highlight histological details. x 400

wall of these patients suggests that the lack of gene expression in the mucosa is not an artefact and that mucosal chemokine expression may be absent in some stages of IBD. It was not possible to analyse the potential influence of drug therapy on mucosal gene expression, because of insufficient clinical data. This point is of importance since recent studies22 have dem- onstrated a down-regulation of MCP-1 expression in the colonic tumour cell line Caco-2 after treatment with dexamethasone. In addition, the same authors detected MCP-1 by immunohistochemistry in the surface epithe- lial cells of normal intestine and specimens from patients with IBD. Hence, MCP-1 possibly generates a chemo- tactic gradient for macrophages in the lamina propria. The apparent discrepancy between these immunohisto- chemical findings and the absence of detectable MCP- 1 mRNA by ISH in epithelial cells described in the present report may be explained by (i) the presence of MCP-1 transcripts below the detection level for ISH in epithelial cells, (ii) drug therapies, (iii) the binding of MCP-I to cell surface proteoglycans and/or receptodmediated

endocytosis of MCP- 1 by epithelia cells, (iv) modulation of MCP- 1 expression by infecting invasive bacteria,23 and (v) the presence in the intestinal epithelial cells of immunologically cross-reactive chemokines such as MCP-2 or MCP-34 that are not detected by ISH.

A further finding of our study is the expression of MCP-1 and RANTES mRNAs in the intestinal mucosa of control patients. This indicates that local expression of the two genes is not specific for IBD, but may allow the continued recruitment of cells responsible for an efficient immunological surveillance of these anatomical sites with a high antigenic burden.

The preferential expression of MCP- 1 transcripts in endothelial cells of venules, as well as in intravascular mononuclear cells, suggests that this cytokine is also involved in mechanisms regulating the adhesion of blood monocytes to endothelial cells in vivo. Previous studies24 have indicated that MCP-1 can induce up-regulation of two heterodimers of the p2 integrin family, Mac-1 (CDllb/CDIS) and p150,95 (CDllc/ CD18), on human peripheral blood monocytes. It is

206 L. MAZZUCCHELLI

likely that some aspects of the adhesion of blood mono- cytes to endothelial cells are mediated by p2 integrin- ICAM-1 interactions. Thus, the contribution of MCP-1 to the recruitment of monocytes/macrophages may not be restricted to its chemotactic effects on these target cells, but may affect interactions between ICAM-1 on the surface of endothelial cells and its ligand on leuko- cytes. Interestingly, both ICAM-1 and MCP-1 expres- sion are stimulated by similar inflammatory cytokines, including IL-1 and T N F - c ~ . ~ ~

We have shown that the number of MCP-1-expressing cells in medial smooth muscle cells of intestinal arterioles and venules in controls and in patients with IBD is comparable. In particular, vascular smooth muscle cells express the MCP-1 gene at high levels, even in the absence of vascular lesions or inflammation. Indeed, the MCP-1 gene seems to be generally expressed in smooth muscle cells, since we found a comparable density of MCP- 1 -expressing cells in the tunica muscularis of con- trol intestinal tissues and in the tunica muscularis of diseased intestines. The biological significance of these findings is not clear. In the absence of macrophage and lymphocyte accumulation in smooth muscle, MCP- 1 is unlikely to have played an active role in the modulation of an inflammatory process at this site. By contrast, these results seem to call into question the significance of previous report^'^.^^ indicating that MCP-1, produced in vascular smooth muscle cells, plays a role in early events i n atherogenesis.

Finally, we report that the MCP-1 gene is expressed in the inyenteric plexus region, regardless of the occurrence of local inflammatory infiltrates. We have not deter- mined the cellular source of MCP-1 at this site, but nerve cells and MHC class 11-expressing enteroglial cellsZx are likely candidates. The question of whether MCP-1 synthesis in these cells links the nervous and immune system, for instance to regulate local neuro- peptide production, remains to be answered.

I n conclusion, our data suggest that there is an increased, but differential expression of MCP- 1 and RANTES genes in intestinal specimens from patients with IBD, compared with controls. This could con- tribute to the increased recruitment and activation of monocytes/macrophages and lymphocytes to distinct sites of the bowel wall. These cytokines are thus likely to play an important role in the pathogenesis of IBD.

ACKNOWLEDGEMENTS

We thank T. Ptrinat and S. Trachsel for expert technical assistance and Professor F. Nothiger (Tiefenauspital, Bern, Switzerland) for the generous supply of tissue samples. This study was supported by a Swiss National Science Foundation grant (No. 31-33977.92 to CM).

REFERENCES I . Oppenheim JJ. Zachariae COC. Mukaida N, Matsushima K. Propertics of

the novel proinflammatory supergene 'intercrine' cytokine family. Annu RCJV lltlrIful7o/ 1991; 9 617448.

2. Miller MD, Krangel MS. Biology and biochemistry of the chemokines: a faiiiily of chemotactic and inflammatory cytokines. Crif Rev Immunol 1992; 12: 17 46.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

2n.

21.

22.

23

24

25

26

27

28

ET AL.

Yoshimura T, Leonard EJ. Human monocyte chemoattractant protein-I: structure and function. Cytokines 1992; 4 131-1 52. Van Damme BJ, Proost P, Lenaerts JP, Opdenakker <i. Structural and functional identification of two human, tumor-derived monocyte chemotiic- tic proteins (MCP-2 and MCP-3) belonging to the chemokinc family. J Evp Med 1992; 176 5 9 6 5 . Schall TJ, Jongstra J , Dyer BJ, i!r ol. A human T cell-specific molecule is ii

member of a new gene family. J Immunol 1988; 141: 1018 1025. Yoshimura T, Robinson EA, Tanaka S, Appella E. Leonard EJ. Purifi- cation and amino acid analysis of two human glioma-derived monocyte chemoattractants. J E.xp Mtd 1989; 169: 1449-1459. Furutani Y, Nomura H, Notake M, et ol. Cloning and sequencing of the cDNA for human monocyte chemoattractant and activating factor (MCAF). Biochcm Bioph,vs Res Commun 1989; 1 5 9 249-255. Carr MW, Roth SJ, Luther E. Rose SS, Springer TA. Monocyte chemo- attractant protein-l acts as a T lymphocyte chemoattractant. Prrx M t f l Acad Sci USA 1994; 91: 3652-3656. Alam R, Lett-Brown MA, Forsythe PA, ct nl. Monocyte chemotactic and activating faclor is a potent histamine-rclcasing factor for biisophils. J Clrn

Schall TJ, Bacon K, Toy KJ, Goeddel DV. Selective attraction of mono- cytes and T lymphocytes of the memory phenotype by cytokine RANTES. ,Vufure 1990; -747: 669 67 I . Kuna P, Reddigan SR, Schall TJ. Rucinski MY. Viksmann MY, Kaplan AP. RANTES. a monocyte and T lymphocyte chemotactic cytokine relcases histamine from human basophils. J Imniunol 1992; 149 636442. Rot A. Krieger M, Brunner T, Bischoff SC. Schall TJ. Dahinden CA. RANTES iind macrophage inflammatory protein I alpha induce the migration and aclivalion of normal human eosinophil granulocytes. J Erp Mi4 1992; 176 1489-1495. Schall TJ, Lu LH. Gillet N. Ameiito EP. RANTESlSlS cytokine gene expression in rheumatoid synovium a s analyzcd by in sifu hybridization. Arthrirls Rhrwm 1991; 3 4 S1 17. Rathanaswami P, Hachicha M, Sadick M. Schall T.1. McColl SR. Expres- sion of the cyrokine RANTES in human rheumatoid syiiovi;il fibroblasts. J B i d C1ii.m 1993; 268: S834 S839. Pauison J, Nelson PJ, Huie P, ['I o/. RANTES cheniokine expression in cell-mediated transplant rejection of the kidney. krnwf 1994; 343: 209-7-1 I . Mueller C , tiershenfeld HK. Lobe CG. 0kad;i CY, Rleiickley RC. Weissman 1L. A high proportion of T lymphocytcs that infiltratc H-2- incompatible heart allografts in vivv express p i e r encoding cytotoxic cell-specific serine proteases. but do nu t cxprcs thc Mcl-ICdclincd lymph nodc homing receptor. J Exp Med 1988; 167: 1124-1 136. Ma7n1cchelli L, Hauser C. Zgraggcn K. cf ( I / . Expression of interleukin-8 gene in inflaininatoiy bawd disease is related 10 the hirtological gradc of active inflammation. At11 J Purlid 19994: 1 4 4 997 1007. Altman DG. Praclcil Statistics fcx Media l Researcli. London: Chapman & Hall. 1991. Neter J, Wassei-man W. Kutner MH. Applied ILineiir Staliatical Models-Regression. Analysis of Variance and Experimcutal Design. Homewood. IL: Irwin, 1985. Mullin GE, Schmidt A, Braun-Elwert L. ('I ul. KAN'TES, MCAF and MIP-2n mRNA levels are increased i i i the miic~rsal Iesi~iis of IBD. Gu.jfro- cnr i vdrqy 1994; 106: A741. Devergne 0. Marfaing-Koka A. Schall TT, "I ol l'roduction 01' the RANTES chemokine in delayed-type hypersensitivily rcaclionr: i n r o l w ment of m;icroph;iger and endothclial cclls. J E r p h l d 1994: 179 1689 1694. Reinecker HC. Loh EJ. Ringler DJ, Mehta A. Kombeau JL, MacDermott RP. Moiiocyte~clieinoattract~int protein 1 gcnc cxprcssion in intestinal epithelial cells and inflammatory bowel disease mucnsic. ~ ; i r s / r r i r t i r i ~ r o l ~ ~ ~ )

Jung HC. Eckmann L, Yang SK, cf ol. A distinct array of proinflammatory cytokines is expressed in human colon epithelial cells in response to bacterial invasion. J C/in fnvesr 1995; 9 5 55-65. Jiang Y, Beller DI, Frendl G. Graves DT. Monocyte chemoattractant protein-I regulates adhesion molecule expression and cytokine production in human monocytes. .I Immunol 1992; 148: 2423-2428. Sica A, Wang JM, Colotta F, ef ol. Monocytc chemotactic and activating factor gene expression induced in endothelial cells by IL-l and tumor necrosis factor. J lmmunal 1990; 144: 3034 3038. Cushing SD. Berlinger JA. Valente AJ, el ( I / . Minimally modified low density lipoprotein induces monocyte chemotactic prolein I in human endothelial cells and smooth muscle cells. Proc Null Acad S c i USA 1990; 87: 5134-5 138. Navah M, lmes SS, Hama SY, er a/. Monocyte transmigration induced by modification of low density lipoprotein in cocultures or human aortic wall cells IS due to induction of monocyte chemotactic protein 1 synthesis and ir abolished by high density lipoprotein. J Clin Invest 1991; 88: 2039- 2046. Geboes K. Rutgeerts P, Ectors N, et a/. Major histocompatibility class I I expression on the small intestinal nervous system in Crohn's disease. Gastroenfcrolog); 1992; 103 439447.

Invest 1992; 8% 723-728.

1995; ion: 40-50.