Differential Expression of COX-1 and COX-2 in the Gastrointestinal Tract of the Rat

Abstract

The aim of this study was to use immunohistochemistry with morphometry to investigate COX-1 and COX-2 expression in the normal rat gastrointestinal (GI) tract and examine if sites of ulceration previously observed with long-term COX-2 inhibitor administration in mice correlate with differential COX-1/COX-2 expression. COX-2 positive cells were observed predominantly in the apical lamina propria of intestinal villi with fewer cells in the mucosal epithelium. The highest level of COX-2 expression was observed at the ileocaecal junction (ICJ). COX-2 expression was also present in parasympathetic ganglia of the submucosa and muscularis. In the stomach, the highest grade of COX-2 expression was observed in the apical lamina propria of the fundus adjacent to the junctional ridge. In contrast, COX-1 positive cells within the lamina propria were evenly distributed along the GI tract but were present in higher numbers than COX-2 positive cells. The mean level of COX-1 expression at the ICJ was not significantly different from the ileum and caecum. Evidence that the highest level of COX-2 expression in normal rats is located on the ileal side of the ICJ provides the first mechanism to explain spontaneous ulceration and perforation of the distal ileum in COX-2−/− animals.

Introduction

Cyclooxygenase (COX) enzymes catalyse the conversion of arachidonic acid to prostaglandins and exist in 2 genetically different isoforms, constitutive COX-1 and inducible COX-2 (Mitchell et al., 1995). Sustained inhibition of both COX isoforms by nonsteroidal anti-inflammatory drugs (NSAIDS) can cause intestinal ulceration in humans and laboratory animals (Bjarnason et al., 1987). However, clinical studies indicate that selective COX-2 inhibitors are less ulcerogenic than nonspecific COX inhibitors (e.g., indomethacin) (Shah et al., 2001). Selective COX-2 inhibitors do not induce gastric lesions in rats (Schmassmann et al., 1998) and cause less gastrointestinal side effects than conventional NSAIDS in humans (Hawkey, 1999).

However, the received wisdom that COX-1 inhibition is the sole cause of NSAID-induced GI side effects has been challenged by the observation that long-term COX-2 deficiency or inhibition is associated with significant intestinal pathology despite normal intestinal PGE-2 levels in the mouse. This indicates a role for COX-2 in the maintenance of small intestinal integrity (Sigthorsson et al., 2002). Also, inhibition of both COX-1 and COX-2 is required for NSAID-induced gastric injury, indicating a role for COX-2 as well as COX-1 in maintaining the integrity of the stomach (Wallace et al., 2000). COX-2 inhibition leads to small bowel damage in COX-1-deficient mice. Therefore COX-2 may be more important for maintenance of small bowel integrity than COX-1. COX-2 products may modulate the mucosal reaction to luminal antigens or delay spontaneous ulcer healing, similar to their role in dermal wound healing (Lee et al., 2003). Treatment of rats with potent inhibitors of COX-2 has resulted in a delay in gastric ulcer healing and COX-2 has been shown to contribute to PGI-2 synthesis in the rat stomach (Tegeder et al., 2000). COX-1 inhibition up-regulates COX-2 expression and this may be the key to NSAID-induced intestinal damage (Tanaka et al., 2002). Selective COX-2 inhibitors, e.g., Celecoxib, have been shown to decrease the epithelial proliferative response and delay healing of cryoprobe-induced gastric ulcers. Similar findings are reported with multiple compounds, strongly suggesting that the lesion is a result of pharmacological action.

The rat caecum is particularly sensitive to long-term, low-dose indomethacin, both in terms of chronic intestinal inflammation and changes in prostanoid metabolism (Nygard et al., 1995). These authors suggested that site-dependent susceptibility to intestinal injury in the rat may reflect the relative importance of local prostanoid homeostasis in different parts of the intestine.

Despite the interest in this area, the cellular expression of COX-1 and COX-2 in the GI tract and their relative contributions to wound healing and toxicity are largely unknown. The aim of this study was to use immunohistochemistry with morphometry to investigate COX-1 and COX-2 expression in the normal rat GI tract and examine if sites of ulceration previously observed with long-term COX-2 inhibitor administration in mice correlate with differential COX-1/COX-2 expression. A 2-stage process was adopted that used a grading system to evaluate the entire GI tract and then morphometric and statistical analysis was applied to the ileum, ICJ, and caecum. In addition, to better characterise the COX-expressing cells, an antibody (ED-1) that recognises a subset of tissue macrophages was also used. Improved understanding of COX expression in the rat GI tract will enable more accurate prediction and understanding of NSAID-induced pathology in preclinical safety assessment and enhance risk assessment for clinical use.

Materials and Methods

Animals and Tissues

The entire GI tract from 6 individual 12-week-old Han Wistar rats (3 male and 3 female); and ileocaecal junction (ICJ) area from 35 control rats (25 male and 10 female) was used. The oesophagus, duodenum, jejunum, ileum, colon, and rectum were placed in the block in a “Swiss-roll” style (Figure 1A). Stomach and ICJ were placed longitudinally in the block. Sections of ICJ contained approximately 5–10 mm of distal ileum, ileocaecal junction and 2–5 mm of proximal caecum. Caecum was placed transversly in the block.

Histology Procedures

All sections for light microscopy were fixed in 10% buffered formalin, embedded in paraffin wax, sectioned, and stained with haematoxylin and eosin (H&E). Sections were also stained with Periodic Acid Schiff, Giemsa, Martius scarlet blue, and Toluidene blue. Histopathological examination of H&E-stained sections revealed no morphological evidence of background pathology in the GI tract in any of the animals used in this study.

Immunohistochemistry

A Streptavidin ABC/HRP-DAB (Avidin Biotin Complex/Horseradish Peroxidase-Diaminobenzidine) system was used with the following primary antibodies: COX-2 goat polyclonal raised against a peptide mapping at the carboxy terminus of COX-2 of rat origin (Santa Cruz sc-1747; 1:20,000), COX-1 goat polyclonal raised against a peptide mapping at the carboxy terminus of COX-1 of human origin (Santa Cruz sc-1752; 1:5,000). Tissue sections were autoclaved for 10 minutes using a citrate buffer (pH 6.0; H-3300, Vector Labs, USA) to retrieve antigens prior to adding the primary antibody. Sections of rat brain containing foci of chronic inflammation were included as a positive control for each of the primary antibodies used. Preabsorption studies using the specific peptide immunogens (sc-1747-p and sc-1752-p; Santa Cruz) demonstrated that the staining was specific. ED-1 mouse monoclonal (Serotec MCA341R) that binds a single chain glycoprotein expressed on the lysosomal membrane of myeloid cells of the rat, was used as a macrophage marker at 1:10,000 to further characterise COX-expressing cells within the lamina propria. A goat polyclonal raised against hepatitis B antigens (Dako B0560; 1:5000), and an isotype-matched mouse monoclonal antibody (Dako 0931) used at matched IgG concentration acted as negative control antibodies. All of the slides stained for any particular primary antibody were stained as part of a single run. Staining of each anatomical compartment of the entire GI tract was graded on a 5-point scale (1–5 corresponding to very slight, slight, moderate, marked, very marked) according to the approximate number and intensity of positive cells observed. At each site the grade was scored for each animal and the mean score calculated for all animals.

Morphometry and Statistical Analysis

A detailed morphometric analysis was applied only to data for the ileum, ICJ, and caecum. Mean scores of positive cells and cell density were calculated for each of these anatomical compartments and the 2 relevant histological compartments: epithelium and lamina propria. Using a grid system on the projected image of the histological slide on a computer monitor, the areas of these compartments were estimated. For each slide the following information was recorded: tissue area for lamina propria, tissue area for epithelium, number of positive cells in the lamina propria, number of positive cells in the epithelium. The positive cell density was calculated by dividing the observed number of positive cells in each histological compartment by the tissue area. All calculated density observations were square-root transformed. This is standard practice for all data involving, or related to counts, and was used so that the data are normally distributed, enabling parametric analysis. In this analysis all p-values less than 0.05 were regarded as statistically significant. Statistical analysis was performed using JMP version 4.0.2 software (Cary NC:SAS Institute Inc).

Results

COX-2 Expression

In all regions of the GI tract examined, there was COX-2 expression of cells within the parasympathetic ganglia of the submucosa and muscularis (Figure 1D). In the øesophagus there was an absence of COX-2 within the lamina propria and epithelium. The apical lamina propria of the gastric fundus adjacent to the junctional ridge was the region with the highest COX-2 expression in the stomach (Figure 1B). The majority of the fundic region had lower levels of lamina propria COX-2 expression than the pylorus. Positive COX-2 cells were observed at low incidence in the glandular epithelium. In the nonglandular stomach there was an absence of COX-2 expression within the epithelium, however positive cells were observed at low incidence in the lamina propria and submucosa.

In the small intestine (duodenum, jejunum, and ileum) COX-2 positive cells with round to oval morphology and cytoplasmic expression, were observed in the epithelium and also within the lamina propria, predominantly at the apex of the villi. Higher levels of expression were present in the jejunum and ileum relative to the duodenum (Figure 2). In-traepithelial positive cells were distributed at an incidence of approximately 1 positive cell per villus. Peyer’s patches either had a few positive cells located subepithelially or were negative. COX-2 staining was cytoplasmic and was graded as very slight to slight.

Foci of densely stained COX-2 positive cells were observed in the apical lamina propria of the villi at the ICJ, predominantly on the ileal side of the junction (Figure 1D and 1F). These cells were frequently more densely stained than positive cells situated in the ileum or caecum of the same animal. The mean score of COX-2 positive cells was higher in both sexes at the ICJ relative to the ileum and the caecum (Figure 2) and this reflected an increase in local density and intensity of staining of positive cells. The ICJ was the region of highest COX-2 expression in 15 out of 36 animals examined (Figure 3). In a further 10 animals the ICJ and caecum together constituted the regions with the highest COX-2 expression (result not shown). Statistical analysis of data from the ileum, ICJ, and caecum was possible because of the higher number of animals examined (n = 35). This demonstrated that COX-2 expression in the lamina propria was significantly higher at the ICJ than the ileum (p < 0.0001) and caecum (p < 0.0001) (Figure 4). In contrast the density of COX-2 positive cells within the ICJ epithelium was not significantly different from that in the epithelium of the caecum and ileum. Expression of COX-2 within the muscularis was not increased at the ICJ relative to the ileum.

COX-2 positive cells within the caecum were located in the lamina propria and the epithelium, predominantly in the apical lamina propria. COX-2 staining was cytoplasmic, and the mean score of COX-2 expression was similar to that observed in the ileum (Figure 2). Positive cells were located at the same location in the lamina propria of the colon and rectum. Intraepithelial positive cells were observed at low incidence in the colon, but higher incidence in the rectum. No sex difference in COX-2 expression was noted. Sections of rat brain containing foci of chronic inflammation were included as a positive control for COX-2 staining and a subset of mononuclear cells were positive (results not shown).

COX-1 Expression

COX-1 positive cells were observed in the lamina propria, submucosal vascular endothelium and within the muscularis and muscularis mucosae of the entire GI tract. Intraepithelial COX-1 staining was absent in the oesophagus. Weak staining of cells within the lamina propria and endothelial cells of vessels was present. COX-1 positive cells were located in the nonglandular region of the stomach within the epithelium at the limiting ridge and also within the lamina propria. In the fundic and pyloric regions, positive cells were located throughout the lamina propria but predominantly near the apex (adluminal). This expression was graded as slight to moderate and the cells were more abundant than COX-2 positive cells at the same site.

COX-1 positive cells within the lamina propria of the small intestine were located throughout the length of the villus and were graded as slight to moderate. These cells were located throughout the villar lamina propria but were predominantly within the apical two-thirds of the villus. Intraepithelial COX-1 positive cells were graded as very slight. Peyer’s patches at these sites had positive cells located subepithelially and fewer numbers of positive cells scattered diffusely throughout the lymphoid tissue. COX-1 staining was cytoplasmic and was graded as moderate in all animals. Although there was a population of small positive cells within the submucosa, staining of the parasympathetic plexuses was absent.

COX-1 positive cells, graded as slight to moderate, were observed in the lamina propria of the villi at the ICJ, with the same morphology and location as those observed in the ileum (Figure 1E). In contrast to COX-2 expression, the mean score of COX-1 positive cells at the ICJ was similar or lower to that observed in the ileum and the caecum (Figure 2). Intraepithelial COX-1 positive cells were graded as very slight. In 27 out of 36 animals examined (i.e., 75%) (Figure 3), there was equal COX-2 expression across the ileum, ICJ, and caecum.

COX-1 positive cells were observed within the lamina propria of the caecum, with the same morphology and location as those observed in the ileum. This staining was graded from slight to marked. Lamina propria COX-1 positive cells within the colon and rectum were located predominantly in the apical region (cf., other regions of GI tract) and were graded as slight. Intraepithelial COX-1 positive cells were graded as very slight. No sex difference in COX-1 expression was noted. Sections of rat brain containing foci of chronic inflammation were included as a positive control for COX-1 staining and a subset of mononuclear cells were positive (results not shown).

ED-1 Expression

In the ileum, ED-1 positive granular cytoplasmic staining of tissue macrophages within the lamina propria of the villi was observed. These cells were located throughout the length of the villus, extending further apically than COX-1 positive cells. ED-1 staining was graded as very slight to moderate. There were no positive cells observed within the epithelium. In those animals in which lymphoid follicles or aggregations were observed in the lamina propria, there were numerous ED-1 positive cells scattered diffusely through the follicle. The ileum is the region of highest ED-1 expression in 16 out of 36 animals examined (i.e., 44%) (Figure 3). There was no staining of cells noted within the parasympathetic plexusus. Occasional positive cells were observed in the submucosa, muscularis, and serosa.

At the ICJ, ED-1 positive cells were similar in morphology to those observed in the ileum, were located in the same region of the villus and were graded as very slight to moderate. The mean score of ED-1 positive cells was lower in both sexes at the ICJ relative to the ileum and this reflected a decrease in local density of positive cells. In the caecum, ED-1 positive cells were similar in morphology to those observed in the ileum, were located in the same region of the villus and were graded as very slight to moderate. The mean score of ED-1 positive cells was lower in both sexes at the caecum relative to the ileum (results not shown) and this reflected a decrease in local density of positive cells. No sex difference in ED-1 expression was noted.

Discussion

The level of COX-2 expression varies along the length of the rat GI tract and the observation that the site of highest expression within the lamina propria is the ICJ is a novel finding. A distinct pattern of COX-2 expression within the lamina propria of the ICJ was observed with positive cells being located predominantly on the ileal side at the apex of the villi. COX-2 expression was cytoplasmic, with the strongest staining located perinuclearly. This finding is consistent with reports that COX-2 is located within the endoplasmic reticulum and nuclear membrane (Otto and Smith, 1994; Morita et al., 1995). COX-2 protein can be expressed in a wide range of cell types, for example monocytes, macrophages, fibroblasts, and endothelial cells in ulcerated gastric mucosa (Schmassmann et al., 1998).

COX-2 positive mononuclear cells in this study have a morphology and location consistent with some of them being macrophages, which have previously been reported to express COX-2 in the rat colon (Reuter et al., 1996). In addition, the location of the COX-2 cells is in a region of the lamina propria containing ED-1 positive macrophages. However, there were no positive ED-1 cells observed within the epithelium, indicating that those COX-1 and COX-2 positive cells at that site were not expressing ED-1. ED-1 is not expressed on all tissue macrophage populations and therefore this finding does not necessarily exclude these intraepithelial cells from being macrophages, but this does indicate a range of cellular differentiation within the villus. The highest density of ED-1 cells was found in the ileum rather than the ICJ and therefore high levels of COX-2 expression at the ICJ cannot simply be explained by high numbers of macrophages at the site. Future studies on the identity of the COX-2 expressing cells will use a broad panel of antibodies, including those that recognise different macrophage subpopulations (e.g., ED-2).

None of the animals examined in this study showed morphological evidence of pathology at the ICJ and yet there was an increase in both the number and intensity of staining of COX-2 positive cells at this site. COX-2 is classically considered an inducible enzyme but others have demonstrated COX-2 protein to be constitutively expressed, for example in the smooth muscle cells of the muscularis mucosae and in fibroblast cells of the gastric mucosa in both rabbit and man (Zimmermann et al., 1998). However, adequate controls to validate this expression were not described. COX-2 protein is expressed constitutively in the lamina propria of the GI tract in mice (Hull et al., 1999). A recent study in the dog found that all gastrointestinal tissues expressed COX-1 and COX-2 mRNA, although protein expression was only observed for COX-1 (Wilson et al., 2004). There is conflicting evidence on the level of COX-2 expression in the gastrointestinal mucosa of control rats, with Reuter et al. (1996) finding evidence of COX-2 mRNA in the colon, but no evidence of protein expression by IHC. Other workers have demonstrated COX-1 and COX-2 staining in normal rat colon in the peri-nuclear region and cytoplasm of the interstitial tissue and epithelial cells (Vogiagis et al., 2001).

An interesting observation in our study was the high levels of COX-2 expression observed at the junctional ridge of the stomach, relative to the fundus and pylorus. The junctional ridge and ICJ share some anatomical features in that they are functional interfaces between dissimilar parts of the GI tract. The junctional ridge is known to be subject to local damage resulting in inflammation and epithelial vacuolation. Similar susceptibility to damage and inflammation may explain the high levels of COX-2 expression at the ICJ. The finding that COX-2 expression is increased in the lamina propria in experimental colitis in the rat (Reuter et al., 1996) tends to support this theory. Therefore interanimal variation in ICJ COX-2 expression may be the result of variable endogenous factors, such as GI bacterial flora or stress. Within the lamina propria there are a number of cell types capable of producing prostaglandins: myofibroblasts, leucocytes, smooth muscle cells, endothelial cells, and fibroblasts.

Prostaglandins derived from either COX-1 or COX-2 have fundamental regulatory roles in GI barrier function, inflammation, immunophysiology of electrolyte transport, and GI motility (Powell et al., 1999). Currently the cellular sources of prostaglandin in normal and diseased intestine are poorly defined. It is known that LPS upregulates COX-2 protein in rat muscularis resident macrophages (Hori et al., 2001). Non–bone-marrow-derived lamina propria stromal cells have basal COX-2 expression and COX-2 dependent PGE-2 production by these cells is spontaneous and continuous and could potentially contribute to the hyporesponsiveness of the intestinal immune response (Newberry et al., 2001).

There is a site-dependent gradient of prostaglandin E2 concentration in the rat intestine with highest levels in the small intestine and caecum. COX-2 is expressed at higher levels in enterocytes derived from the distal relative to the proximal bowel in MIN mice and azoxymethane-treated F344 rats (Roy et al., 2001). Our data supports this trend of higher COX-2 expression in distal parts of the intestine. Furthermore, the distal predominance of COX-2 expression appears to be important in human sporadic colon carcinogenesis, with rectal cancers overexpressing COX-2 in 90% of cases, whereas more proximal tumours up-regulate COX-2 only 22% of the time (Dimberg et al., 1999). There is evidence that COX-2 is involved in the repair of intestinal mucosal damage, and that the animals that develop lesions are somehow predisposed by other factors responsible for mucosal insult. In this study, COX-2 expression was noted within the plexusus between the inner circular and outer longitudinal muscularis (Auerbach’s or myenteric plexus), and within the submucosa (Meissner’s plexus) in all regions of the GI tract examined. These cells are parasympathetic ganglion cells and give rise to postganglionic fibres that supply the mucosal glands and the smooth muscle of the muscularis and muscularis mucosae, indicating a role for COX-2 in gut secretion and motility.

In contrast to COX-2, the expression of COX-1 is relatively uniform along the GI tract, with no increase at the ICJ or junctional ridge of the stomach. The number of positive cells within the lamina propria is higher than for COX-2 expression at the same site and the cells are more diffusely located within the villus. COX-1 expression was cytoplasmic and predominantly perinuclear. The stomach and large intestine showed relatively low levels of expression of COX-1 relative to other parts of the GI tract. In contrast to COX-2 expression, COX-1 was also observed in the submucosal vascular endothelium and within the muscularis and muscularis mucosae of the entire GI tract. It is possible that COX-2 positive cells represent a subset of those expressing COX-1. The grade of lamina propria staining in the 3 anatomical regions examined was highly consistent, with 75% of animals showing equal staining across the ileum, ICJ, and caecum. Previous work has demonstrated that COX-1 can also be expressed in epithelial cells of the intestinal crypts and is increased following injury by irradiation (Cohn et al., 1997). ED-1 positive cells were located throughout the villus and were found further proximally than COX-1 positive cells. This indicates that not all ED-1 positive macrophages express COX-1.

It has been suggested that the gastrointestinal toxicity of conventional NSAIDs, which are lipid-soluble weak acids, is due to a topical effect in addition to inhibition of the COX-1 enzyme within the mucosa (Bjarnason et al., 2003). Evidence in the rat indicates that the high intestinal tolerability of Celecoxib, which does not produce intestinal ulcers, is due to a combination of the absence of a topical damaging effect and selective COX inhibition (Tibble et al., 2000). In vitro work has demonstrated that COX-2 inhibitors, in contrast to nonspecific COX inhibitors, do not interfere with basal reepithelialisation (Giap et al., 2002) but earlier work showed that COX-2 inhibitors can inhibit normal epithelial cell proliferation and in some cases induce apoptosis (Erickson et al., 1999). In wound-healing studies, COX-2 inhibitors delay reepithelialisation and inhibit angiogenesis (Futagami et al., 2002). COX-1 knockout mice do not spontaneously develop gastric lesions, demonstrating that the absence of COX-1 alone is not sufficient to induce gastric damage (Langenbach et al., 1995). COX-2 expression is increased in tissues undergoing neovascularisation in parallel with increases with VEGF (Majima et al., 2000). Although COX-1 and COX-2 mRNA are expressed at similar levels in control normal rat stomachs (Vogiagis et al., 2000), increased levels of COX-2 mRNA have been demonstrated in acetic acid-induced rodent gastric ulcers (Mizuno et al., 1997). Therefore COX-2 inhibitors could potentially reduce healing of lesions. In acute studies, COX-2 inhibitors had no effect on the generation of prostaglandins, resulting in no damage, whereas co-administration of COX-1 and COX-2 inhibitors provoked intestinal damage.

This study provides new information and important insight into the biology of COX enzymes in the control rat GI tract. Anatomic sites that act as interfaces or junctions between functionally dissimilar compartments of the GI tract, such as the gastric junctional ridge or ICJ, are subject to local injury that may account for the locally high levels of COX-2 expression. Here we report that the highest level of COX-2 expression in normal rats is located on the ileal side of the ICJ, which provides the first mechanism to explain spontaneous ulceration and perforation of the distal ileum in COX-2 −/−animals (Sigthorsson et al., 2002). Improved understanding of differential COX-1 and COX-2 expression in the rat GI tract will enable more accurate prediction and mechanistic understanding of NSAID-induced pathology in preclinical safety assessment and enhance risk assessment for new drugs in this class prior to clinical use.