Cyclooxygenase-2 Expression and Role of Vasoconstrictor Prostanoids in Small Mesenteric Arteries From Patients With Crohn’s Disease
Background— The present study investigates the vascular reactivity and the involvement of nitric oxide and prostanoids in regulating vasoconstriction of small mesenteric arteries from patients with Crohn’s disease (CD) to understand the vascular component of this pathology.
Methods and Results— An increased production of proinflammatory cytokines (tumor necrosis factor-α and interleukins 1β, 6, and 8) has been observed in biopsy specimens of inflammatory intestinal mucosa. However, contractile responses of small mesenteric arteries from CD patients in response to norepinephrine were not changed ex vivo when compared with controls. Exposure to either the nitric oxide synthase inhibitor NG-nitro-l-arginine or the cyclooxygenase (COX) inhibitor indomethacin did not modify contractions induced by norepinephrine in either control or CD patients. However, in the latter, the specific COX-2 inhibitor N-(2-cyclohexyloxy-4-nitrophenyl) methanesulfonamide significantly attenuated norepinephrine-induced vasoconstriction. Furthermore, immunohistochemical analysis showed marked COX-2 expression in the whole arterial wall of vessels from CD patients. Vessels from control patients exhibited weak COX-2 staining in the adventitial and endothelial layers only.
Conclusions— The above results provide direct evidence for COX-2 expression in small mesenteric arteries from CD patients. They also shed new light on the involvement of vasoconstrictor metabolites of COX in regulating contraction of these arteries. Of particular interest is the balance between vasoconstrictor products from COX-2 and unidentified vasodilatory products that maintained vascular reactivity in a physiological range despite an increase of circulatory cytokines in patients with CD.
Received October 25, 2002; accepted December 6, 2002.
Among the proposed hypothesis for Crohn’s disease (CD) pathogenesis, a vascular cause has been suggested, mostly in reference to anatomic and pathological studies. Indeed, Wakefield et al1 found intravascular thrombi and granulomas associated with the gut vasculature in both inflamed and noninflamed intestine in CD. Thus, granulomatous vasculitis may, at least in part, participate in the primary events leading to mucosal injury and subsequent histological and endoscopic lesions. Recently, Desreumaux et al2 described in situ alteration of fibrinolysis in the vascular lesions in the ileum of CD patients. Inflammatory cytokines, in particular tumor necrosis factor-α (TNF-α), have been suggested to be involved in the onset and/or perpetuation of mucosal lesions in CD.3 High cytokine levels result in upregulation of inducible nitric oxide synthase (iNOS) and cyclooxygenase-2 (COX-2). Both NO and COX products are greatly involved in the vascular response to inflammatory injury.4 In CD, enhanced production of NO, increased NO synthase activity,5 and enhanced prostaglandin production6 in inflamed intestinal mucosa have been reported. Nevertheless, the potential role of such mediators in the mesenteric vascular abnormalities in CD has not been explored and was presently investigated in small mesenteric arteries (SMA).
The protocol was approved by the local ethics committee (CCPPRB d’Alsace No. 1). Seven patients (2 men and 5 women aged 33±4 years) with CD were included. The patients demonstrated moderate to severe disease (mean Van Hees activity index, 193±15; range, 150 to 261). Four patients had ileocolonic, 2 had colonic, and one had isolated ileal involvement. Except for one patient, all received steroids (10 to 50 mg/d). The control group included 6 patients aged 43±4 years (3 women and 3 men), who underwent surgery for colorectal cancer (n=4) or diverticular disease in a chronic noninflammatory state (n=2).
Cytokine Immunoassays and Histological Inflammation Scoring
Fragments of inflamed intestinal mucosa in CD patients and normal mucosa in controls were removed and washed in Ca2+-Mg2+-free Hanks’ medium supplemented by 100 IU/mL penicillin and 100 μg/mL streptomycin. They were blotted carefully, weighed, and individually placed in tissue culture plates. Cytokine assays were performed after 24 hours.
In parallel, tissue samples from all specimens were assigned an overall inflammation score (ranging from 5 to 20), as previously used.3
Culture supernatants were assayed using 2-site ELISAs specific for human interleukin (IL)-1β, -6, and -8 and TNF-α (Antibody Solutions). Quantitative evaluation of secreted cytokines was achieved as previously described.3 Results of cytokine concentrations were expressed according to the weight of biopsy specimens.
SMA (200 to 400 μm internal diameter) were harvested from biopsy specimens of inflamed colon. Segments (2-mm-long) were removed and mounted in a myograph filled with physiological salt solution kept at 37°C and gassed with 95% O2/5% CO2. Mechanical activity was recorded isometrically as previously described.4
The vessels were first challenged with KCl (100 mmol/L). The functionality of the endothelium was assessed by the ability of bradykinin (1 μmol/L) to induce >60% relaxation of vessels precontracted with U46619 at 80% of their maximal responses. Cumulative concentration-effect curves to norepinephrine (NE) were performed in the absence or presence of either the NO synthase inhibitor NG-nitro-l-arginine (L-NA; 100 μmol/L), the COX inhibitor indomethacin (10 μmol/L), or the selective COX-2 inhibitor N-(2-cyclohexyloxy-4-nitrophenyl) methanesulfonamide (NS-398; 1 μmol/L).
COX-2 and iNOS Immunostaining
Arterial segments were immersed in 4% paraformaldehyde in 0.1 mol/L Na-phosphate buffer for 6 hours. Then, they were soaked overnight in buffer containing 20% sucrose. The samples were then frozen in isopentane (−50°C) and sectioned on a cryostat microtome. Specific polyclonal rabbit antibody anti-human-PGHS-2 (C-terminus; Oxford Biomedical Research; 1/50) or mouse anti-iNOS antibody (Transduction Laboratories; 1/50) was incubated overnight. Then, 12-μm sections were incubated with the secondary fluorochrome Alexa 488 (Molecular Probes) anti-rabbit or fluorescein isothiocyanate (FITC) anti-mouse conjugated antibody (Chemicon; 1/200) for 1 hour. The specificity of COX-2 labeling was tested using a negative control obtained by coincubation of Alexa 488 anti-rabbit antibodies with the preimmune serum (Oxford Biomedical Research). Control experiments for iNOS immunostaining were realized without primary antibody. Confocal images of labeled vessels were acquired as previously described.4
Results are expressed as mean±SEM of n experiments in which n represents the number of patients. ANOVA and the nonparametric Spearman rank test were used for statistical analysis. P<0.05 was considered statistically significant.
Concentrations of TNF-α (30±10 versus 2±0.15 pg/mg of tissue per mL; P<0.01), IL-1β (320±158 versus 20±10 pg/mg of tissue per mL; P<0.05), IL-6 (405±179 versus 37.5±25 pg/mg of tissue per mL; P<0.01), and IL-8 (3966±1599 versus 56±17.5 pg/mg of tissue per mL; P<0.01) were significantly higher in the culture medium of CD intestinal mucosa (mean histological score, 14.8; range, 11 to 18) than in controls. TNF-α concentration in organ culture supernatants showed a trend toward correlation with histological score (r=0.77, P=0.08).
Vascular reactivity to KCl depolarization (control: 6.9±0.3 mN/mm, n=6; CD: 7.5±0.8 mN/mm, n=7) and maximal relaxation to bradykinin (control: 75.2±4.7%, n=6; CD: 76.7±6.4%, n=7) were not significantly different between groups of patients. The increases in tension produced by NE were not significantly different in SMA from control and CD patients (Figure 1A). In vessels from control patients, neither L-NA nor indomethacin significantly modified contractile responses to NE (Figure 1B). In arteries taken from patients with CD, L-NA was without effect on responses induced by NE (Figure 1C). In arteries from CD, contraction to NE seemed to be lower in the presence of the nonspecific inhibitor of COX-isoforms indomethacin (Figure 1C). Interestingly, the selective inhibitor of inducible COX-2, NS-398, significantly reduced NE-induced contraction in arteries from CD patients (Figure 1E) but not in vessels from controls (Figure 1D).
COX-2 and iNOS Immunostaining
Whereas weak or no staining of COX-2 was found in the smooth muscle cell layer of arteries from the 3 control patients studied (Figure 2A), marked COX-2 labeling was observed in the medial layer of arteries from the 3 patients with CD (Figure 2B). The adventitia of vessels from all patients displayed specific staining. Finally, the endothelial layer of vessels from all patients displayed COX-2 staining even though the labeling was more pronounced in arteries from CD patients. The negative control obtained by coincubation of COX-2 antibodies with the preimmune control serum did not display any staining demonstrating the specificity of our labeling (Figure 2C). No specific iNOS immunostaining could be detected in arteries from controls or CD patients (data not shown).
The above results provide direct evidence for COX-2 expression of SMA from control and patients with CD. However, the medial layer of vessels from patients with CD exhibits marked COX-2 expression, even though this enzyme is found in the adventitial and endothelial layer of vessels from both control and CD patients. Furthermore, the results also shed new light on the involvement of vasoconstrictor metabolites of COX in regulating contractions of these arteries. Of particular interest is the balance between vasoconstrictor products from COX-2 and unknown vasodilatory products that maintained vascular reactivity in a physiological range despite an increase of circulatory cytokines in patients with CD.
As previously shown,3 increased spontaneous release of proinflammatory cytokines Il-1β, IL-6, IL-8, and TNF-α was found in CD mucosa, thus confirming the development of inflammatory process in the intestine of these patients.
Elevated levels of inflammatory cytokines have been reported to cause attenuation of endothelium-dependent vasodilatation and/or decreased responsiveness of vascular smooth muscle to vasoconstrictor stimuli.7 In the present study, the endothelium-dependent relaxation to bradykinin and the contractile response to NE were not impaired in vessels from patients with CD. Thus, both endothelial function and reactivity of SMA to vasoconstrictor agonists are preserved despite increased cytokine production in CD patients. Such observations have already been reported in small omental arteries from septic shock patients4 and may be a general feature of small arteries in inflammatory diseases.
Blockade of NO synthase with L-NA did not modify contractile responses in arteries from control patients. This is consistent with previous reports showing that endothelium-derived NO does not play a major role in small human arteries.4 In arteries from patients with CD, L-NA did not affect vasoconstrictor responses to NE, and no iNOS staining could be detected under the experimental conditions used. In contrast, inflammatory states with high cytokine production associated with induction of iNOS and amplification of NO-mediated responses are well documented in CD.8,9 Possible reasons for these discrepancies may be that iNOS is induced in SMA but the experimental conditions used did not allow its detection because the native form of the protein may be less accessible for the anti-iNOS antibody. Furthermore, the difference might result from lower NO production, increased NO breakdown as a consequence of an increased oxidative status, or an increased effect of vasoconstrictor factors that unmasked the inhibitory effect of NO in SMA from patients with CD.
One of the most important findings was the upregulation of COX-2 in the medial layer of SMA from CD patients. Although COX-2 expression was observed in the adventitial and endothelial layers of SMA from both control and CD patients, marked immunolocalization of COX-2 was observed in the medial layer containing smooth muscle cells of vessels taken from inflamed intestine only. Recent works have reported that COX-2 is induced in colonic epithelial cells and in lamina propria mononuclear cells,10 as well as in the myenteric plexus11 from CD specimens. The present findings provide evidence for COX-2 expression in SMA. Elevated levels of IL-1β and TNF-α, which are both normal activators of the nuclear factor κB, may account for enhanced expression of COX-2 as well as IL-8 in inflamed tissues from CD patients. In conjunction with these results, enhanced expression of COX-2 but not COX-1 has been reported in inflamed colon.10
The use of the selective COX-2 inhibitor NS-398 unmasked reduced responses to NE of arteries from CD patients, suggesting that vasoconstrictor products from COX-2 play a role in agonist-induced vasoconstriction. In contrast to results obtained in arteries from CD patients, NS-398 was without effect on NE-induced contraction in control patients. These data suggest that either COX-2 metabolites derived from the adventitia and/or the endothelium play a minor role in regulating vasoconstriction in SMA, or its effects on contraction are blunted by other products from other sources, including NO from iNOS.12 We cannot distinguish among these possibilities. Nevertheless, the present study showed COX-2 expression in the medial layer of arteries from CD associated with the production of vasoconstrictor products that regulate contractions to NE in these arteries. It is noteworthy that when both constitutive and inducible COX isoforms were blocked with indomethacin, only a slight diminution of contraction to NE was observed. These results suggest that metabolites from COX-1 also play a role in regulating vascular reactivity in CD patients. Probably, COX-1 releases mainly vasodilatory metabolites, which counteract the effects of vasoconstrictor prostanoids from COX-2 in CD.
Altogether, the present study highlights production of vasoconstrictor prostanoids, probably from COX-2, to preserve vascular contractility in SMA from CD patients. Prolonged or excessive inflammatory injury may overwhelm this compensatory mechanism, leading to vasoplegia, microthrombi formation, and subsequent induction or worsening of intestinal lesions. Furthermore, a decrease in NE-induced contractile response in SMA from CD patients in the presence of indomethacin and NS-398, a selective COX-2 inhibitor, may explain the potential,13 although controversial, toxicity of nonsteroidal antiinflammatory drugs in CD patients and provide a strong pathophysiological hypotheses for this particular drug-related side effect.
J.M. Reimund was supported by the “Association François Aupetit” and A. Tabernero by the “Fondation pour la Recherche Médicale.”
Wakefield AJ, Sawyerr AM, Dhillon AP, et al. Pathogenesis of Crohn’s disease: multifocal gastrointestinal infarction. Lancet. 1989, ii: 1057–1062.
Reimund JM, Wittersheim C, Dumont S, et al. Increased production of tumour necrosis factor-α, interleukin-1β, and interleukin-6 by morphologically normal intestinal biopsies from patients with Crohn’s disease. Gut. 1996; 39: 684–689.
Stoclet JC, Martinez MC, Ohlmann P, et al. Induction of nitric oxide synthase and dual effects of nitric oxide and cyclooxygenase products in regulation of arterial contraction in human septic shock. Circulation. 1999; 100: 107–112.
Rachmilewitz D, Stamler JS, Bachwich D, et al. Enhanced colonic nitric oxide generation and nitric oxide synthase activity in ulcerative colitis and Crohn’s disease. Gut. 1995; 36: 718–723.
Stenson WF. Role of eicosanoids as mediators of inflammation in inflammatory bowel disease. Scand J Gastroenterol. 1990; 25 (suppl 172): 13–18.
Greenberg S, Xie J, Wang Y, et al. Tumour necrosis factor-alpha inhibits endothelium- dependent relaxation. J Appl Physiol. 1993; 74: 2394–2403.
Roberts PJ, Morgan K, Miller R, et al. Neuronal COX-2 expression in human myenteric plexus in active inflammatory bowel disease. Gut. 2001; 48: 468–472.
Kleschyov AL, Muller B, Keravis T, et al. Adventitia-derived nitric oxide in rat aortas exposed to endotoxin: cell origin and functional consequences. Am J Physiol. 2000; 279: H2743–H2751.