Introduction

Pseudomyxoma peritonei (PMP) is a rare disorder characterized by progressive accumulation of viscous, mucinous ascites in the peritoneal cavity. The gelatinous ascites is produced by disseminated implants of mucinous epithelium on the peritoneal surfaces and omentum. The incidence of PMP is two times as high in women as in men. Therefore, there has been some controversy about the role of ovarian mucinous neoplasms as a potential origin of PMP. Recent immunohistochemical and molecular genetic studies indicate, however, that a dissemination of epithelium from mucinous cystadenomas of the appendix is a common cause of PMP in women. The occurrence of synchronous ovarian mucinous tumors in PMP can, in most instances, be regarded as implants of mucinous epithelium of appendiceal derivation (for review, see: [3, 7]). In rare cases, as recently described by Ronnett and Seidman [13], ovarian teratomas containing elements of mucinous appendiceal epithelium may give rise to PMP. The appendiceal origin is further supported by the high expression of the intestinal mucin MUC2 and the intestinal-specific homeobox gene CDX2 in PMP [10, 16].

The mucinous change in appendix or mucinous cystadenoma, previously called mucocele, is histologically characterized by extensive mucosal replacement of abundant mucin-producing goblet cells. Despite a usually well-differentiated morphology, cytogenetic studies have revealed monoclonal loss of heterozygoity, indicating the true neoplastic nature of appediceal mucinous cystadenomas [3, 7].

The regenerating (Reg) proteins belong to a group of calcium-dependent, c-type lectins. The 17 Reg proteins of various mammalian species recognized so far are small (about 20 kD) secreted proteins of four superfamilies, now called Regs I–IV. The Reg proteins have been functionally implicated as growth factors and antiapoptotic “survival factors” for pancreatic islets and neural cells, as acute phase reactants modifying inflammation, and as regeneration factors stimulating mucosal repair of the gastrointestinal tract.

The genes for the three previously known human Reg proteins (Reg I–III, also called Reg1a or Lithostatin A, Lithostatin B, and pancreatitis-associated protein) are clustered in tandem on chromosome 2p12 (reviewed in [18]). Hartupee et al. [2] and Kämäräinen et al. [4] independently identified, cloned, and characterized a fourth member of the human Reg family called Reg IV or Reg-like protein (RELP). The gene for RELP has a different chromosomal location on p 12–13 and encodes a secreted protein of 158 amino acids that shows about 45% similarity with the Reg I–III proteins [4].

In situ hybridization revealed a restricted expression of RELP in the neuro-endocrine cells of normal intestinal epithelium. By immunohistochemical staining of RELP with rabbit antibodies, we demonstrated a coexpression of RELP and chromogranin A in selected cells in the crypt epithelium of small intestinal mucosa [4]. A strongly up-regulated expression of RELP was found in the goblet cells in areas of intestinal metaplasia in the gastric mucosa and in Barrett’s esophagus [4]. Hartupee et al. [2] reported an accumulation of Reg IV message in colon mucosa from patients with inflammatory bowel diseases. We used immunohistochemistry and found a strong expression of RELP protein in the goblet cells of inflamed intestinal mucosa in Crohn’s disease and in ulcerative colitis [4].

In situ hybridization has revealed up-regulated expressions of RELP message in colorectal adenomas and colorectal cancer [15, 19]. Oue et al. [11] recently performed a serial analysis of gene expression on samples from primary gastric cancers and found that RELP was one of the most prominently up-regulated genes in the schirrous type of gastric cancer [11].

Taken together, previous findings indicate that the up-regulated expression of RELP associates with preneoplastic and neoplastic changes in the gastrointestinal mucosa. Given our previous observations on the accumulation of RELP in metaplastic goblet cells in the gastric mucosa, we asked whether RELP is also expressed in the neoplastic goblet cells of appendiceal mucinous cystadenomas and PMP. Here we report the expression of RELP in appendiceal mucinous cystadenomas and in PMP in relation to other intestinal marker proteins.

Materials and methods

Formalin fixed and paraffin-embedded specimens were collected from the archives of the Department of Pathology, Haartman Institute, University of Helsinki. All patients were of Finnish, Caucasian origin. The PMP samples conformed to the classification of Ronnett et al. [12] of disseminated peritoneal adenomucinosis type. Cases representing peritoneal mucinous carcinosis with or without intermediate/discordant features were not included. The data are summarized in Table 1.

Table 1 Patient data and summary of the staining results
Full size table

Monoclonal antibodies to RELP

Monoclonal antibodies to RELP have not been reported previously. The RELP coding complementary DNA (cDNA) sequence was amplified by polymerase chain reaction (PCR) and cloned into an expression vector by replacing human immunoglobulin G1 (IgG1) variable region sequence with RELP coding sequence. The human IgG1 heavy chain secretion signal and the N-terminal amino acids of the variable region, glutamine (Gln)-isoleucine (Ile)-glutamine (Gln), are retained on the recombinant protein. Human IgG1 hinge, calponin homology 2 (CH2), and CH3 domains are C-terminal to the RELP coding sequence. SP2/0 myeloma cells were transfected by electroporation with the RELP plasmid, and clonal cell lines that stably secrete the protein were isolated. The recombinant RELP protein was purified by protein A column chromatography.

Mice were immunized with 50 μg of RELP/Fc protein mixed with adjuvant until IgG titers of 1:50,000 were achieved. Three days prior to fusion the mice were injected with 10 mg of RELP in phosphate buffered saline (PBS), administered intravenously. The splenocytes were harvested and fused at a 1:1 ratio with FO murine myeloma cells with a polyethylene glycol 3,000 solution. To assess hybridoma supernatant activity, solid phase enzyme immunoassays (EIA) were performed. RELP/Fc protein or a control Fc construct were coated overnight on to 96-well EIA plates. After washing, the wells were then blocked with 1% bovine serum albumin in PBS. Undiluted supernatants from the growth-positive hybridomas were incubated on both Fc protein coated plates and then probed with horseradish peroxidase (HRP)-labeled goat antimouse IgG Fc-specific antibody. HRP substrate solution was added, and the absorbance was measured at 490 nm by an automated plate spectrophotometer. Only clones specifically reactive to RELP/Fc and not to the negative-control Fc protein were expanded in tissue culture and subcloned. Monoclonal supernatants were purified using protein A.

Mouse monoclonal antibodies to the human mucins-MUC2, mucin 5 subtypes A and C (MUC5AC), and MUC6 were obtained from Novocastra Laboratories Ltd (Newcastle upon Tyne, UK). Monoclonal anti-CDX2 antibody was obtained from BioGenex (San Ramon, CA). Secondary antibody, HRP-conjucated Rabbit antimouse immunoglobulin (IgG), was obtained from Dako (Glostrup, Denmark).

Immunohistochemistry

Four-μm sections from formalin-fixed, paraffin-embedded tissues were mounted on 3-aminopropyl-triethoxysilane-coated slides (Sigma), deparaffinized, and heated twice for 5 min in a microwave oven (650 W) before exposure to the first antibody (monoclonal mouse anti-RELP antibody, dilution 1:500). EliteKit (Vectastain, Vector Laboratories, Burlingame, CA) was used for immunoperoxidase staining, visualized with 3-amino-9-ethylcarbazole (Sigma).

In situ hybridization

We prepared probes by using the TA cloning kit (Invitrogen, San Diego, CA) ligating a polymerase chain reaction-amplified 0.4-kilobase RELP cDNA insert into the pCR-II vector (Invitrogen). The templates for RELP antisense or sense RNA probes were generated by linearizing the appropriate vector construct (3′ to 5′, or 5′ to 3′, respectively). The RNA Labeling Kit of Boehringer–Mannheim (Mannheim, Germany) was used according to the manufacturer’s instructions to generate digoxigenin-labeled RNA probes by in vitro transcription.

Treatment of sections for in situ hybridization

Tissue sections were deparaffinized as before and specially prepared for in situ hybridization. Prior to hybridization, the sections were treated with 0.3% triton X-100 and then with 10 mg/ml proteinase K for 15 min at 37°C. The sections were then postfixed with 4% paraformaldehyde for 5 min. After acetylation of the section [2×5 min with triethylamine buffer containing 0.25% acetic anhydride (v/v)], they were incubated with a prehybridization buffer [4× sodium chloride–sodium citrate (SSC) containing 50% deionized formamide] for 15 min at 37°C.

The prehybridization buffer was drained and sections were overlaid with hybridization buffer (Sigma, St. Louis, MI) containing 10 ng of digoxigenin-labeled RNA probe. Sections were covered with plastic coverslips and incubated in a humid chamber at 42°C overnight.

After hybridization, the coverslips were removed and the sections were washed at 37°C twice in 2× SSC for 15 min each then twice with 1× SSC for another 15 min each. Unbound RNA probe was digested with a NaCl–Tris HCl–ethylenediaminetetraacetic acid buffer containing 20 μg/ml of ribonuclease A at 37°C; the sections were washed twice with 0.1× SSC for 30 min each. The labeled probe was detected using the digoxigenin nucleic acid detection kit (Roche, Germany) according to the manufacturer’s instructions.

Results

Staining of sections from normal appendix with monoclonal antibody to RELP revealed exclusive expression in scattered cells in the crypt epithelium (Fig. 1b). We have earlier shown that the cells strongly expressing RELP in normal intestinal mucosa codistribute with cells expressing chromogranin A as a marker of neuro-endocrine differentiation [4]. The nuclei of the normal appendiceal epithelium stained strongly for CDX2. Normal goblet cells stained for MUC2 but not for RELP. The MUC2 staining was confined to the perinuclear region and along the cellular membrane (Fig. 1d). A weak staining for MUC5AC was seen locally in a morphologically normal mucosa in one case, while expression of MUC6 was not evident in the sections from normal appendix (Fig. 1e,f).

Fig. 1
figure 1

Sections from proximal appendix with normal mucosa (case number 5 in Table 1) (a–f), and from mucinous cystadenoma in the distal end of the same appendix (g–l), stained with van Gieson trichrome stain (a and g) and immunohistochemically with antibodies to RELP (b, arrow heads, and h), CDX2 (c and i), MUC2 (d and j), MUC5AC (e and k), and MUC6 (f and l). Original magnification ×100

Full size image

A strong expression of RELP was seen throughout in the mucin-containing goblet cells of appendiceal mucinous cystadenomas (Fig. 1h). Higher magnifications revealed a slightly granular staining of the goblet content for RELP. The nuclei of the adenomatous mucosa stained for CDX2. MUC2 was also strongly expressed in all cells of the adenomatous epithelium (Fig. 1j). The intracellular staining for MUC2 was mainly seen in the perinuclear, basal, and lateral portions of the goblet cells, and hence, did not fully codistribute with the staining for RELP.

In contrast to the normal appendiceal mucosa, the adenomatous epithelium stained strongly for MUC5AC. The antibodies to MUC5AC decorated the goblet cells of the upper, luminal portion of the adenomatous mucosa, while the deepest glands did not contain the antibodies to MUC5AC (Fig. 1k). Staining for MUC6 was restricted to the basal, MUC5AC-negative glands in five out of six adenomas that came from women. The epithelium of the basal glands in three appendiceal cystadenomas from male patients did not stain for MUC6 (Fig. l, Table 1). Immunohistochemical staining of epithelial implants from PMP also revealed a strong expression of RELP in the mucin-producing cells (Fig. 2b). The staining pattern was virtually identical to that seen in the epithelium of appendiceal mucinous cystadenoma. As reported earlier by O’Connell et al. [10], the epithelial cells of PMP stained throughout for MUC2. As in appendiceal mucinous cystadenomas, the intracellular distribution of MUC2 appeared mainly at the perinuclear and baso-lateral areas of the goblet cells (Fig. 2c,d). The PMP epithelium also stained for MUC5AC (Fig. 2e), while no staining was seen for MUC6 in the nine cases of PMP investigated in this study (Table 1).

Fig. 2
figure 2

Sections of epithelial implants of PMP stained with van Gieson’s trichrome stain (a) and immunohistochemically with antibodies to RELP (b), CDX2 (c), MUC2 (d), and MUC5AC (e). Original magnification ×100. In situ hybridization of PMP epithelium with antisense RELP probe (f) and with control, sense probe (g). Original magnification ×200

Full size image

To confirm the local production of RELP in the PMP epithelium, in situ hybridization was carried out with antisense and sense RNA probes to RELP. As shown in Fig. 2f, a strong signal indicating accumulation of RELP messenger RNA was seen in PMP epithelium, while no signal was obtained from consecutive sections using the sense (control) probe.

Discussion

We have employed a novel monoclonal antibody and in situ hybridization to demonstrate a strong expression of RELP in the neoplastic goblet cells of appendiceal cystadenoma and epithelial implants of PMP. However, it appears unlikely that the robust expression of RELP is a consequence of neoplastic transformation because we have previously reported a strong staining for RELP in nonneoplastic, reactive mucosal goblet cells in inflammatory bowel diseases and in metaplastic goblet cells. Goblet cells of normal intact intestinal mucosa did not express RELP at levels discernible by immunohistochemistry or by in situ hybridization [4]. RELP and MUC5AC displayed a similar expression profile, albeit with a different intracellular distribution. Therefore, convergent mechanisms appear to regulate the de novo acquisition of goblet cell phenotype and activation of expression of the RELP and MUC5AC genes. The molecular details of these events, however, are still poorly understood.

Yamamoto et al. [16] recently provided experimental data to show that the homeodomain protein CDX2 acts as a transcriptional activator to regulate the expression of MUC2 in transfected COS-7 cells. Moreover, Mutoh et al. [8] generated transgenic (TG) mice in which ectopic expression of CDX2 was targeted to the gastric mucosa under the promoter of the noncatalytic subunit of the rat H+/K+-adenosine triphosphatase gene. They found that the TG mice developed a progressive loss of parietal cells and eventually showed intestinal metaplasia of the entire gastric mucosa. The same authors recently reported a progress from intestinal metaplasia to adenomatous polyps and to invasive gastric cancer in the same TG mice with targeted gastric overexpression of CDX2 [9]. However, it seems unlikely that CDX2 alone regulates the transcription of the RELP gene in appendiceal mucinous adenomas and PMP, because a strong nuclear staining for CDX2 is seen in normal intestinal epithelium, including normal goblet cells, which lack RELP expression.

RELP is abundantly present in normal intestinal neuro-endocrine cells. Therefore, we performed immunohistochemistry of appendiceal carcinoid tumors and found expression of RELP in solid and, in particular, in goblet cell variants of the tumor (manuscript in preparation). Interestingly, appendiceal carcinoid tumors show a similar female preponderance as mucinous adenomas and PMP do [13]. Given the expression of RELP in both tumors, it is tempting to speculate about a common cellular origin of these neoplasms.

Coexpression of MUC5AC and RELP was seen in the goblet cells of PMP. In the appendiceal mucinous cystadenomas, however, RELP homogeneously stained the entire transformed mucosa, while positivity for MUC5AC was seen in its upper, luminal portion, leaving the deepest glands unstained. Adenomatous epithelium in the basal glands stained for MUC6 in five out of six appendiceal mucinouos cystadenomas from women, while no reactivity for MUC6 was seen in the adenomas from men. Although the material is too limited to allow any firm conclusions, it is interesting to notice that steroid hormones influence the expression of MUC6 in breast cancer cells [1]. This might have some bearing on the higher female incidence of appendiceal mucinous cystadenomas and PMP.

MUC6 was absent in all ten cases of PMP, while a strong expression of MUC5AC was detected. This might indicate that subclones expressing MUC5AC have a higher propensity for intraperitoneal implantation. The MUC5AC-positive subclones may also be better adjusted for growth and/or survival under poorly oxygenated conditions, like inside the mucous masses of PMP. The MUC5AC-containing goblet cells are also located in the upper portion of the appendiceal adenomatous mucosa, furthest away from the circulation.

The functional significance of the strongly up-regulated expression of RELP in inflammatory, metaplastic, and neoplastic goblet cells remains to be clarified. Because RELP, like other known members of the Reg protein family, is a secreted protein, it may be a ligand for a still unidentified receptor mediating signals in a paracrine or autocrine fashion. A receptor for the homologous Reg I, RegR was cloned by Kobayashi et al. [6]. Reg IR is a 919 amino acid transmembrane protein, which shows homology to the human multiple exostoses-like gene. RegR has been found on beta cells of pancreatic islets and in rat gastric mucosa where it mediates proliferative signals [5]. A corresponding receptor for RELP remains to be identified.

Taken together, we show here a strong de novo expression of RELP in the neoplastic goblet cells of appendiceal mucinous cystadenomas and PMP. RELP may serve as a clinical and immunohistochemical marker for these disorders. Our recent observations indicate that RELP displays relative specificity for neoplastic goblet cells of intestinal derivation. Immunohistochemical staining of 73 ovarian tumors with mucinous differentiation revealed some degree of positivity in 15 cases (20%) (unpublished observations). Further studies are in progress to elucidate the pathogenetic role of RELP expression in neoplastic and metaplastic goblet cells in the gastrointestinal tract.