MATERIALS AND METHODS

Twenty six carcinoid tumours from 26 patients were identified from the pathology archive. Twelve were located in the small intestine, 10 in the appendix, three in the colon, and one in the stomach. A representative paraffin wax block containing tumour was selected for each case. In 11 cases, more than one paraffin wax block was investigated (total of 42 paraffin wax blocks). Immunohistochemistry was performed on paraffin wax sections using a biotinylated second antibody (swine antirabbit IgG; Dako, Ely, Cambridgeshire, UK) at a 1/200 dilution for 30 minutes at room temperature. The primary antibody used for detecting GRPR was a novel polyclonal rabbit antiserum raised against a synthetic peptide (residues 243–260), corresponding to a region of the third intracellular loop of human GRPR. This was used at a 1/100 dilution for one hour at room temperature. To detect GRP, a commercially available anti-GRP polyclonal antibody was used (Dako rabbit antihuman polyclonal; A0429). Before incubation with the primary antibody, endogenous peroxidase was blocked by immersion in hydrogen peroxide/methanol for 10 minutes; antigen retrieval was performed by pressure cooking for one minute in citrate buffer (pH 6), and non-specific binding sites were blocked by incubation for 30 minutes in a 1/15 dilution of normal swine serum/phosphate buffered saline. To detect bound antibody, the sections were incubated for 30 minutes with streptavidin–biotin complex–horseradish peroxidase (strept ABC complex–HRP; Dako). Diaminobenzidine was used as chromogen. Omission of the primary antibody was used as a negative control.

Immunohistochemical staining was scored semiquantitatively according to intensity and distribution in the following way: 1, no staining; 2, weak staining; 3, strong staining; 1, no cells stained; 2, < 10% cells stained; 3, 10–50% cells stained; 4, 50–90% cells stained; and 5, all cells stained. Staining was interpreted by one investigator (NS). For the purposes of data presentation, tumours were considered positive if the sum score of intensity and distribution was ⩾ 6—that is, strong staining in at least 10% of cells or weak staining in over half of the tumour cell population.²⁸ All 26 tumours were stained for GRPR. Nineteen of these were also examined for GRP. Insufficient tissue remained for GRP staining in seven cases. In the 11 tumours where more than one block of tissue was stained, the highest score was recorded.

The Wilcoxon rank sum test for non-paired samples was used to compare scores between tumours at different sites.

RESULTS

Staining for both GRP and GRPR was cytoplasmic (figs 1 and 2). No membrane staining was seen. In total, nine of 19 carcinoids expressed GRP, whereas 22 of 26 tumours expressed GRPR. Staining for GRP was often patchy and focal, whereas GRPR staining was usually more diffuse (figs 1 and 2). GRP staining was also seen in the nerve endings of the myenteric plexus. GRP and GRPR were not detected in normal endocrine cells of the small bowel or colon. Table 1 describes the results of immunostaining for individual tumour sites. In our study, GRP was less intensely expressed in appendix and colonic tumours than in small bowel neoplasms (mean scores, 2.86 and 1.00 v 5.53). This did not reach significance however. GRPR was more strongly expressed in appendix and colonic tumours than in small bowel carcinoids (mean scores, 11 and 7.67 v 5.79) and this was significant (p < 0.05). Using our cutoff value of 6, nine of the 12 small bowel tumours were positive compared with 12 of 13 colonic and appendix carcinoids. Furthermore, if only strength of staining was considered, 0 of 12 small bowel carcinoids showed strong (3+) reactivity compared with eight of 13 appendix/colonic tumours. The single gastric carcinoid had a score of 6 for GRPR and 1 for GRP. Where more than one block from a tumour was immunostained, the immunoscore was concordant in 65% of cases for GRPR and 79% of cases for GRP. In 19 cases where immunostaining was performed with both antibodies, only seven carcinoids were positive for both GRP and its receptor. This was mainly the result of absent or weak GRP expression in GRPR positive tumours.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Gastrin releasing peptide expression in a small bowel carcinoid. Staining is heterogeneous and cytoplasmic (original magnification, ×100).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Gastrin releasing peptide receptor expression in a carcinoid of the appendix. Staining is diffuse and cytoplasmic (original magnification, ×100).

Because the gene encoding GRPR is located on the X chromosome, we also tested whether receptor status was different between male and female patients; however, no significant difference was found (mean scores 6 and 5.9, respectively).

DISCUSSION

Our study shows for the first time that GRP, the mammalian homologue of bombesin, and its high affinity receptor GRPR, are expressed simultaneously in a proportion of gastrointestinal carcinoid tumours. GRP is a 27 amino acid peptide, encoded on chromosome 18, and synthesised as a larger precursor protein (prepro-GRP) with receptor binding activity localised to the C-terminal end.²⁹ The antibody that we used in our study recognises the mature amidated peptide, rather that its precursor. GRPR is a 384 amino acid transmembrane protein, encoded on the X chromosome, which belongs to a family of heptahelical G protein coupled receptors.³⁰ For our study, we used a new rabbit polyclonal antibody raised against a synthetic peptide corresponding to a region of the third intracellular loop of human and mouse GRPR. Western blotting has shown that this antibody recognises a single 95 kDa protein in mouse and human cell lines, corresponding to the expected size of fully glycosylated GRPR. Preincubation of this antibody with synthetic peptide has been shown to abolish immunocytochemical staining. GRP has several biological actions, including the stimulation of gastrointestinal hormone production, stimulation of pancreatic enzyme secretion, and regulation of intestinal motility.¹⁸ In addition, it has been shown to be mitogenic in a variety of non-transformed and neoplastic, tumour derived cell lines.^21–24 In vitro studies show that this growth stimulatory effect is mediated through the high affinity GRPR, which activates several intracellular signalling pathways after ligand binding.³¹ These include activation of phospholipase C, release of intracellular calcium, activation of protein kinase C, and tyrosine phosphorylation.

The normal tissue distribution of GRP is restricted to the central nervous system, neuroendocrine cells of the fetal lung, and nerve fibres in the myenteric plexus of the gastrointestinal tract.¹⁸ In our study, we confirm the presence of immunoreactive GRP in the nerve cells of the myenteric plexus in the small bowel and appendix. As in previous studies, we were unable to demonstrate expression in the normal endocrine cell population of the gut. The normal distribution of the GRPR is more controversial. Although some workers report GRPRs throughout the gastrointestinal tract, including the myenteric plexus, smooth muscle coat, and normal colonic epithelium, other groups have consistently failed to show receptor expression outside the gastric antrum.^32,33 Some of this variability may result from the different techniques used to demonstrate the receptor, which have included immunohistochemistry, mRNA in situ hybridisation, reverse transcriptase polymerase chain reaction, and ligand binding studies. In a previous immunocytochemical study of the small bowel, appendix, and colon, we were able to demonstrate GRPR expression in Paneth cells but not neuroendocrine or surface epithelial cells (E J Cartwright, et al, unpublished data, 2003). In contrast to the absence of GRP/GRPR immunoreactivity in normal endocrine cells, we have found that nine of 19 carcinoid tumours aberrantly express GRP and 22 of 26 expressed the GRPR. In 1984, Bostwick et al described a series of 20 intestinal carcinoid tumours that were immunopositive for GRP.¹⁷ They found that 60% of tumours were GRP positive, although immunoreactivity was less common in colonic neoplasms (20%) than in tumours of the appendix and ileum (73%). In a subsequent paper, they described the staining as cytoplasmic, with intensity varying from cell to cell.³⁵ Our findings are similar, with frequent, strong expression of GRP in small bowel carcinoids, but no staining in two colonic tumours. Unlike Bostwick et al, we also found only a low level of expression in tumours of the appendix. The distribution of staining within positive tumours was often heterogeneous, with single cells or small groups of cells being strongly positive but surrounded by unstained tumour. There was no morphological difference between the positive and negative tumour cells. In particular, there was no evidence of increased mitotic activity in these immunopositive foci. In future studies, however, it would be interesting to double label these cells with a proliferation marker, such as Ki67 or proliferating cell nuclear antigen.

“Our results suggest that although increased gastrin releasing peptide (GRP) synthesis may be involved in autocrine/paracrine stimulation of some carcinoids, overexpression of GRP receptor on its own may be sufficient to drive proliferation in others”

As far as we are aware, there have been no previous immunocytochemical studies of GRPR expression in human carcinoid tumours, although the aberrant expression of GRPR has been demonstrated in gastric and colonic adenocarcinomas.^27,36 In our current study, 22 of 26 carcinoid tumours expressed GRPR. Immunoreactivity was usually more diffuse within the tumour than the corresponding GRP staining. An interesting and perverse finding was that expression tended to be stronger in appendix and colonic tumours than in small bowel carcinoids, the reverse of the pattern seen for GRP. The correlation coefficient between GRP and GRPR immunoscores was negative (p  =  0.207; not significant) Only one third of tumours strongly expressed both proteins. This suggests that although increased GRP synthesis may be involved in autocrine/paracrine stimulation of some carcinoids, overexpression of GRPR on its own may be sufficient to drive proliferation in others. This would be analogous to the role of c-neu (c-erb B2) in transgenic mice, where abnormal expression of a growth factor receptor is sufficient to cause mammary carcinogenesis.³⁷ Alternatively, another ligand may activate GRPR in GRP negative neoplasms.

The staining for GRPR was uniformly granular and cytoplasmic, identical to that described in an immunohistochemical study of colon carcinomas.²⁷ Because this protein is a cell surface receptor that normally spans the plasma membrane, one would expect a membranous pattern of immunostaining. It may be that the receptor detected by this antibody is abnormally distributed within the cell as a result of either overproduction or mutation. The last option is certainly possible, because Carroll et al recently demonstrated multiple mutations of the receptor in 75% of gastric carcinomas expressing the protein.³⁸ They found that whereas some of these mutations caused constitutive activation of the receptor, others rendered it non-functional. Unfortunately, the intracellular location of the mutated receptor was not described. Some growth factor receptors become internalised after ligand binding and activation (for example, vascular endothelial growth factor receptor 2 (KDR)).³⁹ Therefore, cytoplasmic staining may also reflect translocation to the cell interior as a result either of constitutive activation (an activating mutation) or binding of GRP.

In conclusion, our study has confirmed the synthesis of the GRP peptide in almost half of the gastrointestinal carcinoid tumours studied, and shown for the first time that some of these tumours coexpress the GRPR. Although it was not confirmed that the receptor was functional in these cells, our results suggest that autocrine/paracrine stimulation of tumour cell division could exist in intestinal carcinoid tumours, similar to that seen in small cell anaplastic carcinoma.