Enhanced production of IL-8 in chronic but not in early ileal lesions of Crohn's disease (CD)

Abstract

Distinct Th1/Th2 patterns have been observed during the evolution of CD. The aim of this study was to compare neutrophil involvement and IL-8 mRNA and protein expression during early recurrent lesions and chronic phases of CD. Twenty-nine patients with CD having ileocolonic resection with anastomosis were studied. Biopsies were obtained during surgery from the non-inflamed ileal mucosa and from chronic ileal lesions. Endoscopic ileal biopsies were also taken from early recurrent ileal lesions occurring 3 months after surgery. Neutrophil counts were performed and mucosal IL-8 levels were evaluated by competitive reverse transcriptase-polymerase chain reaction and immunohistochemistry. Early recurrent ileal lesions were characterized by low neutrophil counts and IL-8 production at the mRNA and protein levels compared with the ileal chronic lesions. The main cellular sources of IL-8 in the early recurrent lesions were neutrophils, while in chronic lesions the majority of IL-8-stained cells were CD3⁺ T cells and macrophages. These results confirmed that the nature of the inflammatory infiltrate and the expression of cytokine profiles may differ between the acute and chronic phases of CD.

Introduction

CD is a chronic intestinal inflammatory disease characterized by multiple recurrences that may be endoscopic or clinical. Disturbances in mucosal immunoregulation have been demonstrated previously for CD and may represent an important component in the pathogenesis of chronic intestinal inflammation and tissue injury [1,2]. Although the study of altered cytokine expression has been an active area of investigation, a clear consensus regarding potential pathological mediators of mucosal ulcerations during the evolution of the disease has not been reached [1–5]. In the chronically inflamed mucosa of patients with CD, high levels of proinflammatory cytokines and a type-1 cytokine pattern (IL-2, interferon-gamma (IFN-γ)) have been demonstrated [2–5]. In contrast, we recently showed that early post-surgical recurrence of ileal CD was associated with significantly elevated IL-4 mRNA levels and a type-2 profile, suggesting distinct patterns in acute versus chronic inflammation [5]. This may have an important impact on the way we approach treatment of CD and therefore merits further investigation.

IL-8 is the main chemokine involved in neutrophil recruitment and activation [6]. A major role for IL-8 in mucosal ulcerations in inflammatory bowel diseases (IBD) has been suspected [7]. This hypothesis is supported by the functional characteristics of IL-8 [6], as well as a good correlation between endoscopic/histological grades of intestinal inflammation and IL-8 mucosal concentrations in patients with IBD [8,9]. In addition, several experimental models of colitis demonstrate high concentrations of IL-8 in the intestinal mucosa, the disease severity of which is improved after in vivo neutralization of IL-8 [10].

In this study, we attempted to compare neutrophil involvement and IL-8 mRNA and protein expression during early and chronic phases of CD in order to further document cytokine profile changes during the clinical evolution of the disease.

PATIENTS and METHODS

Patients and study protocol

Twenty-nine patients with CD (14 women and 15 men; mean age 31 years) having had a first ileocolectomy for CD and ileocolonic anastomosis were included in the study. All patients gave their informed consent after local ethical committee approval was obtained. In all patients, the diagnosis of CD was made using well-established criteria [11]. Eighteen patients had pure ileal involvement and 11 had ileocolonic CD at surgery. The mean duration of CD at the time of surgery was 7·5 years (range 2–13 years). All patients underwent surgery because of disease-related complications, i.e. symptomatic stenosis (n = 25) and fistula (n = 4). No patient received immunosuppressive treatment. If patients were under steroid therapy before surgery, the drug was tapered and discontinued within 4 weeks before surgery. During the surgical procedure, an ileoscopy was systematically performed to assess the macroscopic and histological integrity of the ileal mucosa 30 cm above the future anastomosis as previously described [12]. Eight biopsies were taken in the operating room from macroscopically non-inflamed ileal mucosa and chronic resected ileal lesions with endoscopic forceps. These biopsies were immediately frozen in liquid nitrogen. No further medical treatment (i.e. steroids, salicylates, antibiotics, immunosuppressive drugs) was prescribed after surgery except for anti-diarrhoeal and anti-spasmodic drugs, whenever needed.

Three months after surgery, an ileocolonoscopy was performed in all patients. Endoscopic findings were scored in the i1 to i4 grades according to the criteria of Rutgeerts et al. [13]. Four endoscopic biopsies were taken from the early recurrent ileal lesions from 23 selected patients with endoscopic recurrence scoring i3 (diffuse aphtous lesions) or i4 (diffuse inflammation and large ulcers).

All biopsy specimens were taken with the same endoscopic forceps and all samples had a similar size and a mean weight of 4 ± 1 mg.

Histological study

Diagnostic evaluation, activity assessment and immunohistochemical analyses were performed on cryostat sections. The sections were examined by two independent observers without prior knowledge of the clinical status of the patients. The inflammatory activity was assessed on haematoxylin–eosin- and May–Grünwald–Giemsa-stained sections using a scoring system graded from 0 to 10 which included architectural and inflammatory changes as previously described [14]. Neutrophil counts within the lamina propria were systematically performed at a magnification × 400 in five different areas that appeared to have the most severe inflammatory change. For each patient, results were expressed as the mean number of cells per area, achieving a coefficient of variation < 10% [15].

IL-8 mRNA quantification by reverse transcriptase-competitive polymerase chain reaction

Two frozen biopsies were immediately placed in RNAzol (Bioprobe, Montreuil, France), and homogenized by mechanical dispersion. Total RNA was extracted as previously described [5]. After treatment at 37°C for 30 min with 20–50 U of DNase I RNase-free (Boehringer, Mannheim, Germany), the quantity and purity of RNA were estimated by determining absorbance at 260/280 nm. Total RNA (5–10 μg) was reverse transcribed into complementary DNA (cDNA) using Moloney Murine Leukaemia virus reverse transcriptase (Gibco BRL, Cergy Pontoise, France), 5 μm oligo-dT16 and 2·5 mm of each of the four dNTP (Pharmacia, Orsay, France) in a final reaction volume of 20 μl in the presence of 1 U/μl of human placenta Ribonuclease Inhibitor (Promega, Lyon, France). Samples were incubated at 42°C for 60 min followed by heating for 5 min at 95°C to inactivate the enzyme and finally stored at −20°C until use.

Quantification is based on competitions between unknown amounts of mucosal cDNA and serial dilution of a known amount of pQB, an internal multispecific control [16]. The reaction mixture was amplified by polymerase chain reaction (PCR) using sense and antisense primers specific for β-actin (5′ GGG TCA GAA GGA TTC CTA TG 3′; 5′ GGT CTC AAA CAT GAT CTG GG 3′) or IL-8 (5′ TTG GCA GCC TTC CTG ATT TC 3′; 5′ AAC TTC TCC ACA ACC CTC TG 3′) [16]. All samples were systematically subjected to 40 PCR cycles, consisting of denaturation for 1 min at 94°C, primer annealing for 1 min at 53°C (IL-8) or 60°C (β-actin), and primer extension for 1·5 min at 72°C using a Thermolyne Amplitron II thermal cycler (Barnstead Thermolyne, Dubuque, IA). After PCR amplification, quantification of β-actin and IL-8 mRNA was performed by electrophoresis in 2–3% agarose gel as previously described [5,17]. As accurate quantification of RNA in a sample is often difficult and imprecise [15–17], the number of IL-8 mRNA molecules was expressed compared with the number of 10⁶ mRNA molecules of an internal control, β-actin, in the same sample.

Immunohistochemistry

The presence of neutrophils was confirmed with an immunohistochemical staining for myeloperoxidase on serial frozen sections using a rabbit anti-human immunoglobulin polyclonal antibody (Dako, Glostrup, Denmark; code A398) in a dilution of 1:400 as primary antibody for 30 min. The primary antibody was applied following pretreatment of the sections with swine serum for 7 min. After washing the primary antibody the sections were further treated with swine anti-rabbit biotin for 30 min and with the avidin–biotin complex for 30 min.

Immunohistochemical staining for IL-8 was performed as previously described [14] using an avidin–biotin–peroxidase technique with rat MoAbs directed against human IL-8 (SanverTECH, Boechout, Belgium). Briefly, sections were fixed in acetone for 10 min, rinsed in PBS with 0·1% of saponin and incubated overnight at 4°C with the antibody directed against IL-8. After endogenous peroxidase was blocked using a 0·3% H₂O₂ buffer with 0·1% saponin for 30 min, the avidin–biotin–peroxidase complex was added (Dako) for 30 min and the reaction product was developed using a solution of 0·05 m acetate buffer pH 4·9 containing 0·05% 3-amino-9-ethylcarbazole (AEC; Janssen, Beerse, Belgium) and 0·01% H₂O₂ as substrates. As negative controls, primary antibody was omitted or replaced with an irrelevant IgG antibody.

As the main cellular sources of IL-8 in the mucosa are reported to be monocytes/macrophages, neutrophils, T cells and epithelial cells, double stainings using mouse IgG1 MoAbs directed against human CD3 (UCHT1; Dako) and CD68 (KP1; Dako) were used to identify, respectively, T cells and monocytes/macrophages. Neutrophils, eosinophils and epithelial cells were identified solely by morphological appearance. The double staining was performed using a combination of the peroxidase technique (brown reaction product) and the alkaline phosphatase anti-alkaline phosphatase technique (blue reaction product). Endogenous alkaline phosphatase was blocked using levamisole. The brown reaction product was obtained using DAB salt (diaminobenzidine; Sigma, Saint-Quentin Fallasier, France) and the blue reaction product was obtained with fast blue BB salt (4-benzoylamino 2,5 diethyoxy benzene-diazonium chloride; Sigma). The quantification of stained cells was performed by two blinded observers on identical areas of the biopsies. Cell counts were performed at a magnification of × 400 in three different areas. For each patient, the results were expressed as the mean percentage of labelled cells per area when a coefficient of variation < 10% was obtained.

Statistical analysis

PCR results were expressed as the median ± s.e.m. numbers of IL-8 mRNA molecules per 10⁶ mRNA molecules of β-actin. Comparisons between the median absolute numbers of neutrophils per area, the median absolute number of IL-8-stained cells by immunohistochemistry and the IL-8 mRNA levels were analysed by the non-arametric Kruskal–Wallis one-way anova test. Differences were statistically significant if P < 0·05.

Results

Histological results

Inflammatory scores and neutrophil counts in healthy and involved ileal biopsies of patients with CD are summarized in Table 1. The biopsies taken during surgery from macroscopically non-nflamed ileal mucosa (n = 13) were histologically normal with an inflammatory score of 0 and no mucosal infiltration by neutrophils. Biopsies performed in the early recurrent ileal lesions during endoscopic recurrence at 3 months (n = 14), showed signs of inflammation, as assessed by the inflammatory scores (median: 3) and the presence of infiltrating neutrophils (median: 3). Both the inflammatory scores (median: 6·5) and the neutrophil counts (median: 11) were significantly higher in the chronically inflamed ileal lesions (n = 10) than in the early recurrent ileal lesions (P < 0·005) and the non-inflamed ileal mucosa (P < 0·001). Differences between the inflammatory scores and the infiltrating neutrophils in the early recurrent ileal lesions and the non-inflamed mucosa were also significant (P < 0·001).

Table 1.

Inflammatory score and lamina propria neutrophil counts during the different phases of CD

	Terminal ileum			Statistics
	Non-inflamed n = 13	Early recurrent L. n = 14	Chronic L. n = 10	Non-inflam versus early L	Non-inflam versus chronic L	Early L versus chronic L
Inflammatory score	0	3	6·5	P < 0·001	P < 0·001	P < 0·005
Neutrophil counts*	0	3	11	P < 0·001	P < 0·001	P < 0·001

Results were expressed as the median number of cells per area at a magnification of × 400.

L, Lesion; inflam, inflamed.

^*Similar results were obtained using standard staining and immunohistochemistry using antibody directed against myeloperoxidase.

Mucosal IL-8 mRNA concentrations during the different phases of CD

High tissue concentrations of β-actin mRNA (109 ± 152 × 10⁶ molecules) were detected in all biopsy specimens. IL-8 mRNA were detected in all specimens taken from non-inflamed ileum of the 18 patients with CD. IL-8 mRNA levels were only at the detectable rate (one IL-8 molecule for 10⁶ β-actin molecules) in the early recurrent ileal lesions of patients with CD. Significantly higher concentrations of IL-8 mRNA molecules (P < 0·001) were detected in chronic ileal lesions (n = 15; median 1114) compared with non-inflamed ileal mucosa (n = 18; median 149) and with early recurrent ileal lesions (n = 9; median 1) (Fig. 1). Concentrations of IL-8 mRNA were lower in the early recurrent ileal lesions than in the non-inflamed ileal mucosa of patients with CD (P < 0·001).

Fig. 1.

IL-8 mRNA concentrations during the different phases of CD. Individual values and medians (horizontal bars) of IL-8 mRNA expressed in number of molecules per 10⁶ mRNA molecules of β-actin. IL-8 mRNA molecules in the early recurrent ileal lesions (n = 9) were significantly lower than in non-inflamed mucosa (n = 18) and chronic ileal lesions (n = 15) of patients with CD (P < 0·001). IL-8 mRNA molecules were significantly higher in chronic ileal lesions compared with non-inflamed ileal mucosa of patients with CD (P < 0·01).

Number of IL-8-positive cells per area during the different phases of CD

In the lamina propria of patients with CD, the absolute numbers of IL-8-positive cells per area were in accordance with the levels of IL-8 mRNA detected by quantitative PCR in the different tissue specimens (Figs 2 and 3). In chronically inflamed ileal mucosa (n = 8) (Fig. 3C), the number of positive cells per area (median 26) was significantly increased compared with early recurrent ileal lesions (median 3) (Fig. 3B) (P < 0·001) and with non-inflamed ileal mucosa (median 5) (Fig. 3A) (P < 0·001). Furthermore, the number of IL-8-positive cells was significantly higher in non-inflamed ileal mucosa (n = 13) than in early recurrent ileal lesions (n = 10) (P < 0·05).

Fig. 2.

Number of IL-8-stained cells during the different phases of CD. Individual values and medians (horizontal bars) of the number of IL-8-positive cells per area (mag. × 400). The number of IL-8-positive cells is lower in early recurrent ileal lesions of CD compared with the non-inflamed ileal mucosa (P = 0·003) and chronic ileal mucosa (P < 0·001). The number of IL-8-stained cells was higher in the chronic ileal lesions compared with non-inflamed ileal mucosa (P < 0·001).

Fig. 3.

IL-8-positive cells in the ileal mucosa of patients with CD. (A) Non-inflamed ileal mucosa ( × 125): the villous architecture is well preserved with rare IL-8-positive cells in the lamina propria of the villi. (B) Early recurrent ileal lesion ( × 125): oedema of the mucosa which induces widening of the intercryptal area with occasional IL-8-positive cells. (C) Chronic ileal lesion ( × 125): irregular villi with mononuclear infiltration. Most IL-8-positive cells were detected in the lamina propria.

Nature of IL-8-stained cells in early and chronic ileal CD lesions

All positive cells in non-inflamed mucosa were CD68⁺ macrophages. The few lamina propria-positive cells in early recurrent CD lesions were mainly invasive neutrophils (median 54%), but also CD3⁺ T cells (median 28%) and CD68⁺ macrophages (median 13%). In chronic ileal lesions, the nature of IL-8-producing cells was different, with most mononuclear stained cells corresponding to CD3⁺ T cells (median 42%) and macrophages (median 27%) (Table 2). At this stage of the disease, neutrophils only represented about 27% of the IL-8-positive cells. None of the numerous eosinophils present in the lamina propria was stained using the MoAb directed against IL-8 (Table 2). In the epithelium, the positive stained cells were not epithelial cells but intraepithelial lymphocytes at any stage of the disease.

Table 2.

Characterization of the IL-8-positive cells in early recurrent and chronic ileal lesions of patients with CD

	Early recurrent lesions (%)	Chronic lesions (%)
CD3+ T cells	28	42
CD68+ macrophages	13	27
Neutrophils	54	27
Eosinophils	0	0
Epithelial cells	0	0

Results were expressed as the median percentage of labelled cells per area at a magnification of × 400.

Discussion

The major finding of this study is the significant difference in neutrophil counts and IL-8 synthesis in the early recurrent ileal lesions occurring 3 months after surgery compared with the chronic lesions of patients with CD. These results confirm and extend the hypothesis of different cytokine profiles being associated with different stages of CD.

IL-8 mRNA levels and IL-8 immunoreactive cells were significantly increased in the chronically inflamed ileal mucosa compared with non-inflamed ileum. Other studies have investigated the inflammatory cytokine pattern at the chronic stage of CD [7–9,18–22]. However, their interpretation is difficult due to the variable nature of the clinical specimens and the methods used to measure cytokines. Here, we used a specific and quantitative method to measure IL-8 mRNA which reflects the in situ production of cytokines rather than measurement of IL-8 in the serum [18–20], qualitative assessment of IL-8 [8,9,22] or analysis of IL-8 produced in vitro after intestinal tissue culture or cell extraction from the affected mucosa [21]. Our results are in accordance with Izutani et al., who showed using quantitative PCR a 20-fold increase in IL-8 mRNA level in inflamed colon compared with healthy colon of patients with CD [23]. Immunohistochemical observations were well correlated with PCR analysis and confirmed the up-regulation of IL-8 synthesis at the protein level in chronic ileal lesions.

The cellular sources of IL-8 mRNA in the mucosa of patients with CD remain debated [7–9,22–24]. In vitro, numerous cells such as macrophages [24], T cells [25,26], neutrophils [8,24,27], eosinophils [28], mast cells [29], endothelial [30] and epithelial [31–33] cells have the capacity to synthesize IL-8. In our study, the major sources of IL-8 in the chronic ileal lesions were CD3⁺ T cells, CD68⁺ macrophages, and neutrophils. The involvement of neutrophils and monocytes/macrophages in the mucosal production of IL-8 has been previously demonstrated in CD [8,24], but our results show that CD3⁺ lamina propria and intraepithelial T cells are also an important source of IL-8, at least in chronic lesions. This finding is in accordance with in vitro data showing IL-8 synthesis from intraepithelial lymphocytes freshly isolated from the ileum [25]. These results suggest that the recruited neutrophils, macrophages and T cells are responsible for the production of IL-8 in the chronic lesions of CD, which may contribute through autocrine and paracrine mechanisms to prolonged augmentation and perpetuation of intestinal inflammatory cell attraction. In vitro experiments have suggested an increased IL-8 production by epithelial cells in CD [22,24,31]. We were unable to confirm this hypothesis in vivo, in agreement with others using in situ hybridization or immunohistochemistry [22,24].

Early recurrent ileal lesions were characterized by low neutrophil counts and IL-8 production at the mRNA and protein level compared with chronic lesions. The present PCR studies were performed in homogeneous groups of untreated patients with CD, quantified for a similar amount of β-actin using the same run of competitive PCR, in endoscopic samples having a similar size and weight. These results were confirmed by immunohistochemistry using a MoAb directed against IL-8, which showed a smaller number of IL-8-stained cells mainly produced by neutrophils in the early recurrent ileal lesions compared with chronic ileal lesions of patients with CD. Taken together, these results suggest that tissue cytokine profile and the nature of the inflammatory infiltrate may differ considerably between the acute and the chronic phases of CD. Interestingly, IL-8 synthesis was also decreased in early recurrent lesions associated with a moderate neutrophil infiltrate compared with non-inflamed ileum. The most likely explanation for this apparent discrepancy is that the mechanisms of neutrophil recruitment in the early recurrent lesions are multifactorial and may also involve mediators other than IL-8 such as IL-1, tumour necrosis factor-alpha (TNF-α), platelet-activating factor (PAF) and the CXC chemokines including epithelial neutrophil activating protein 78 (ENA-78), GRO α, β, γ, GCP-2, NAP-2, 4 and SDF-1 [34]. The mechanisms leading to this ‘down-regulation’ are unknown. We have previously reported that the early recurrent ileal lesions of patients with CD were associated with a significant increase in IL-4 mRNA and a decrease in IFN-γ mRNA compared with the normal mucosa of patients with CD or controls [5]. In vitro, IL-4 has been described as a potent anti-inflammatory cytokine able to down-regulate IL-8 production by neutrophils [35,36]. In vivo, similar observations have been published in animal models of inflammation. In an experimental colitis induced in rat by intrarectal administration of trinitrobenzene sulphonic (TNBS) acid, exogenous IL-4 introduced by a recombinant human vector had a therapeutic effect inhibiting intestinal tissue damage, neutrophil infiltration in the mucosa and myeloperoxidase activity [37]. Similar results have been obtained in glomerular inflammation induced by injection of anti-glomerular basement membrane antibodies in mice where endogenous IL-4 is reported to suppress neutrophil influx and to reduce tissue damage [38]. Taken together, these results could be consistent with a down-regulation by IL-4 of IL-8 and neutrophil infiltration in the early recurrent ileal lesions of CD.

These observations further suggest that mechanisms leading to early recurrent lesions of CD differ from the first stage of a classical inflammatory response characterized by a neutrophilic infiltration of damaged tissue mediated at least in part by chemokines such as IL-8, ENA-78, and GRO α, β, γ [39]. In the early recurrent mucosal lesions of CD, neutrophils are not the main cells involved in the inflammatory mechanisms; eosinophils represent > 80% of the inflammatory cell population, which correlates with increased IL-4, IL-5 and IgE production [5,40]. The kinetics of the intestinal inflammatory infiltrate and of cytokine profiles observed in CD may thus be more reminiscent of that observed in allergic diseases such as atopic dermatitis, in which the chronic inflammatory skin disease results from sequential activation of Th2- and Th1-type cells [41]. If confirmed, this could lead to different therapeutic schemes at the early and chronic phases of CD and open new avenues in the understanding of the primary events at the origin of intestinal lesions.