Abstract Chronic ingestion of non-steroidal anti-inflammatory medication is reported to delay or, in part,reverse development of polyps in the colon, but themechanism for this effect is unknown. UsingmRNA and immunoglobulin probes, specific for prostanoid receptors and for prostaglandin endoperoxide synthase(COX 1 and 2), we sought to define, by in situ and in vitro techniques, changes in PGE2 receptors and synthe-sis in cell populations of precancerous familial adenoma-tous polyposis (FAP) colonic mucosa. In FAP, expressionof prostanoid receptors EP3 and EP4 among colonic lam-ina propria mononuclear and lateral crypt epithelial cellswas robust, with 53.9±5.3% of mononuclear cells stain-ing EP4

+. When sections of normal colonic mucosa wereexamined by similar techniques, prostanoid receptor EP4was expressed on only 21.3±1.2% of lamina propriamononuclear cells (including CD4+ T lymphocytes), aswell as on surface and lateral crypt epithelium, and thisdistribution was found at the mRNA level as well. Whenreceptor expression was quantitated by densitometry, im-munoreactive EP3 protein on deep basolateral (but notother) FAP crypt epithelium was enhanced 2.8-fold overnormal, and the number of prostanoid receptor EP4

+

mononuclear cells by 2.5-fold. On the other hand, whileCOX 1 expression in mononuclear cells was prominentin normal and FAP mucosa, densitometric analysisshowed immunoreactive prostaglandin endoperoxidesynthase levels were further increased in FAP, due to agreater than fourfold elevation of COX 2 expressionamong mononuclear cells and epithelia. Our data suggestenhanced cell-specific prostanoid receptor expressionand increased prostanoid synthesis in precancerous FAPmucosa.

Keywords Prostanoid receptors · COX · Familial adenomatous polyposis (FAP) · EP receptors · Colon cancer

Introduction

Tumors of the colon and rectum account for approxi-mately 15% of all visceral cancers in the United States,making adenocarcinoma of the colon one of the most im-portant malignancies in Western society (Landis et al.1999). Epidemiological studies in humans using non-steroidal anti-inflammatory drugs (NSAIDS) as well asexperiments in animals with chemically induced tumorssuggest a role for prostanoids in the progression of tu-mors arising from the colonic epithelium (LaBayle et al.1991; Reddy et al. 1993; Suh et al. 1993; Thun et al.1993; Giovannucci et al. 1994, 1995; Muscat et al. 1994;Schreinemachers and Everson 1994). Thus, studies in-volving long-term use of NSAIDS, which have a markedeffect on two crucial enzymes [prostaglandin (PG) endo-peroxide synthase 1 and 2, designated cyclooxygenase(COX) 1, 2] in the pathway committed to prostanoidsynthesis, have demonstrated a substantial reduction (ashigh as 48%) in colonic tumors, both spontaneously aris-ing colonic polyps (Suh et al. 1993; Thun et al. 1993; Giovannucci et al. 1994,1995; Muscat et al. 1994; Schreinemachers and Everson 1994) and in a familialpolyposis syndrome (LaBayle et al. 1991). In paralleland confirmatory studies, the intestinal adenocarcinoma-predisposed min mouse, when rendered genetically defi-cient for COX 2, has a substantially reduced incidence ofintestinal tumors (Oshima et al. 1996). Taken together,these investigations suggest an enhancing effect of pros-tanoids on colonic epithelial cell tumorigenesis.

Several mechanisms could account for the apparenttumor-promoting effects of prostanoids. On the onehand, prostanoids may be responsible for more rapid cellgrowth and an expansion of the proliferative zone of epi-thelial cells in the intestinal crypt, reported in persons atincreased risk for colon cancer, i.e., those with familialpolyposis or Gardener’s syndrome (Qiao et al. 1995;

V. Takafuji · D. Lublin · K. Lynch · J.K. Roche (✉ )Departments of Internal Medicine and Pharmacology, Box 801317, MR-4 Building, University of Virginia Health System, Charlottesville, VA 22908, USAe-mail: jkr7m@virginia.eduTel.: +1-804-2432655, Fax: +1-804-2436169

Histochem Cell Biol (2001) 116:171–181DOI 10.1007/s004180100287

O R I G I N A L PA P E R

Vivian Takafuji · Dariene Lublin · Kevin Lynch James K. Roche

Mucosal prostanoid receptors and synthesis in familial adenomatous polyposis

Accepted: 27 April 2001 / Published online: 7 June 2001© Springer-Verlag 2001

Itzkowitz and Kim 1998). On the other hand, an increasein COX 2, as reported (Eberhart et al. 1994; DuBois et al. 1996), may be the primary event in enhancing pros-tanoid-driven proliferation, reducing apoptosis, and de-creasing intracellular adhesion of epithelial cells in co-lonic crypts of cancer-predisposed individuals. A pro-gressive accumulation of mutations, including the APC,p53, and ras genes, may follow, and this sequence hasnow been shown to be strongly associated with the de-velopment of carcinoma in the colonic epithelium (Kinzler and Vogelstein 1996).

We became interested in whether there are additionalpredisposing factors to explain the epidemiological dataof NSAIDS in colonic tumors; specifically, to ascertainwhether an altered density or type of receptor for pros-tanoids on colonic epithelia facilitates the developmentof the adenomatous polyp. Eight prostanoid receptorsnow have been described and cloned (PGD DP,PGE EP1, EP2, EP3, EP4, PGF2α FP, PGI2 IP, andTxA2 TP; Hirata et al. 1991, 1994; Sugimoto et al. 1992;Bastien et al. 1994; Katsuyama et al. 1994; Lake et al.1994; Regan et al. 1994; Funk et al. 1998). cDNAs andimmunoglobulin to unique domains of these receptorshave given the opportunity to test the hypothesis that in-herited differences in the number and distribution of cel-lular receptors for prostanoids, perhaps operating in con-cert with increased levels of COX 2, are associated withepithelial tumorigenesis. Of special interest in investigat-ing this proposition would be tissue from patients with astrong predisposition to development of frank adenocar-cinoma of the colon, such as those in the precancerousstage of familial adenomatous polyposis (FAP). Howev-er, at present, the type and distribution of prostanoid re-ceptors and of important enzymes for prostanoid synthe-sis in these individuals are not known. Using probes spe-cific for all four described human receptors for an impor-tant prostanoid in the colon, PGE2, as well as for COX 1and 2, we sought to determine whether the normal distri-bution and level of expression of prostanoid receptorsand of COX in human colonic mucosa is altered in FAP.

Materials and methods

Human tissues examined

Human colonic specimens from intestine resected at the Universi-ty of Virginia Health System were studied under a protocol ap-proved by the Human Investigations Committee (number 1638).The diagnosis in each case was confirmed histologically by theDepartment of Pathology, University of Virginia, and included his-tologically normal colon from three patients with diverticulosis orchronic constipation, and full-thickness colon from three cases ofFAP. Each specimen was immediately fixed (Hollende Bouin’sfixative) for 8–14 h and subsequently irrigated overnight. Afterdehydration, tissue samples were paraffin-embedded, and 5-µm-thick sections were cut on a microtome (American Optical, Buf-falo, N.Y., USA), placed on glass slides coated with Vectabond re-agent (Vector Laboratories, Burlingame, Calif., USA), and heatedto 55°C for 30 min prior to use.

Immunocytochemistry of human tissues

Antibodies

Monoclonal antibodies to human COX 1 and to human COX 2,supplied as culture supernatants, were a kind gift of Dr. Peter Isak-son (Searle Laboratories, St. Louis, Mo., USA). Their specificityand lack of crossreactivity were confirmed by enzyme-linked im-munosorbent assay (ELISA; Fig. 7B). Monospecific antisera toprostanoid receptors EP1, EP2, EP3, and EP4 were agift from JeffJohnson of Cayman Laboratories, Ann Arbor, Mich., USA: theywere elicited in rabbits after injection with 17–26 amino acid-longsynthetic peptides, representing unique sequences within each receptor. The amino acid sequence constituting each immuno-gen was: EP1, GLTPSAWEASSLRSSRHSGLSHF (C-terminus);EP2, SLRTQDATQTSCSTQSDASKQADL (C-terminus); EP3,NQMSVEQCKTQMGKERE (third extracellular domain); EP4,MSTPGVHSSASLSPDRLNSPVTI (N-terminus). For each antise-rum, there was no crossreactivity to non-homologous peptidesconstituting other receptors of the same (EP) family, as shown byELISA performed as we have reported (Planchon et al. 1999); re-ceptor-specific serum was evaluated at 0.2–1 µg/ml, and peptideswere plated in triplicate at 0.5 µg/ml.

Tissue staining

Sections were deparaffinized and immersed in 20% v/v acetic acidin water (2 min) to inactivate any endogenous intestinal phospha-tases, blocked with 10% normal goat serum in PBS (20 min), andincubated with 50–100 µl primary antibody to EP1, EP2, EP3, orEP4 (3.0 µg/ml IgG) or monoclonal antibodies to COX 1 orCOX 2 (0.15 µg/ml IgG) for 45 min. Control slides were exposedin parallel to normal rabbit serum or to isotype-matched mousemonoclonal immunoglobulin (Sigma Chemical, St. Louis, Mo.,USA) at the same concentration of antibody (3.0 and 0.15 µg/ml,respectively). The sections were subsequently washed in PBS andincubated with the biotinylated goat anti-rabbit or goat anti-mouseimmunoglobulin (Vector Laboratories) for 30 min. The enzymeavidin DH biotinylated complex was diluted in PBS and applied tothe sections (45 min). The substrate (BCIP-NBT; Vector Laborato-ries) was diluted in a 100 mM TRIS-HCl buffer (pH 9.5) and theslide visually monitored for staining (10–20 min). The antigenicspecificity of the reactions was confirmed by absorption of immu-noglobulin with whole antigen (ovine COX 1, 2) or with the origi-nal immunizing peptide (for EP2, EP3, and EP4) in 50- to 100-foldantigen excess. Completeness of antibody absorption by antigenwas assessed by supernatant testing using ELISA methodology,described previously (Planchon et al. 1999).

Tissue sections were also sequentially exposed to a series of an-tibodies to identify the nature of EP4- and COX 1- and 2-positivemucosal cells. Antibodies used included monoclonal immunoglob-ulin to CD3 (Dako, Glostrup, Denmark; 1 µg/ml), CD4 and CD8(Ortho Diagnostics Corporation, Raritan, New Jersey, USA; used1:500), and cytokeratin (Novocastra Laboratories Ltd., Newcastleupon Tyne, UK; 0.3 µg/ml). For these studies, a second substrate(alkaline phosphatase substrate red; Vector Laboratories) was usedin addition to BCIP-NBT. Preliminary double-antibody studies re-vealed that the prostanoid receptor EP4 was present on T lympho-cytes (example in Fig. 1A) with 63±13% of cells that express EP4displaying cell surface CD3 as well. Further studies showed that, ofthe two major subsets of CD3+ T lymphocytes, most EP4-positivecells (71.4%) were CD4-positive (example in Fig. 1B); noEP4/CD8 double-stained cells were detected. Control sections ex-posed to normal immunoglobulin followed by labeled second anti-body showed no staining (Fig. 1C). Thus, in situ, EP4 was presenton CD3+ mucosal T lymphocytes, which expressed CD4 as well.

Densitometry

Stained slides were assessed under a Leitz microscope with a 12-V, 60-W halogen lamp. The Neurolucida softwareand Ludl mo-tor-driven stage (MicroBrightField, Colchester, Vt., USA), previ-

172

ously described (Planchon et al. 1999), were used for image analy-sis, with a Leitz oil-immersion 40×/1.3 NA objective. Forms thesize of mononuclear cells or larger were counted, while smallerstained objects were judged as background. Five to ten visualfields representative of the staining pattern seen were chosen fromeach section. Measurements of brightness (units of pixel lumi-nance in gray scale with black=0 and white=250) were performedand results (designated the relative intensity) were expressed asthe reciprocal of the net brightness number (experimental minuscontrol), where “experimental” represents the value obtained withan antigen-specific antibody and “control” is that obtained with anirrelevant immunoglobulin, matched for concentration.

Flow cytometry

Isolation of human intestinal lamina propria mononuclear cells (LPMC)

Macroscopically normal mucosa was obtained from the surgicalmargins of resected intestinal specimens, where the pathological

lesion was ≥5 cm away. LPMC were isolated by the method ofBull and Bookman (1977) with some modification reported fromour laboratory (Roche et al. 1985).

Cell staining

Single-cell suspensions obtained from human colon were incubat-ed with primary unlabeled antibody to EP receptors (namedabove, 1 µg/ml, 4°C, 15 min) and then with R-FITC-labeled goatanti-rabbit IgG (H+L) (30 µg/ml, 15 min at 4°C; Rockland, Gil-bertsville, Pa., USA). The samples were analyzed on a fluores-cence-activated cell sorter (FACScan; Becton Dickinson, FranklinLakes, Calif., USA) and 6–8×103 gated living cells were acquiredin list mode for each sample. Concentration-matched normal rab-bit serum served as the negative control. Preliminary studiesshowed that, with unfractioned normal colonic LPMC (>95%mononuclear cells), 20.2±4.7% were EP4

+ while <1% were EP3+

by flow cytometry. When only the CD3+ population of LPMCwas examined by the same methodology, 32.5% of the cells wereEP4

+. Overall, then, the prostanoid receptor EP4 was evident onone-third of lamina propria CD3+ T lymphocytes and constituted20% of all mononuclear cells in the lamina propria compart-ment.

In situ hybridization

Source and sequence of probes

Prostanoid receptors (complete translation open reading frames)were isolated by polymerase chain reaction. The oligonucleotidesequences used for primers for human EP4 were (5′ GCCAGC-CACTATCAGTC and 5′ TAGCCCTTCTGAGCACAG). Bothwere subcloned into the plasmid vector pcDNA1 (Invitrogen, SanDiego, Calif., USA).

Labeling of RNA

RNA probes were transcribed using the digoxigenin RNA labelingtechnique for in vitro transcription (Boehringer Mannheim, Mann-heim, Germany) as recently described by us (Cosme et al. 2000).

Hybridization

Tissue preparation and prehybridization procedures were as re-ported (Nicholas et al. 1993). Exposure of tissue sections toprobes and to substrate was as we have reported (Cosme et al.2000). In preliminary studies, the antisense EP4 mRNA probehybridized with epithelial cells from the base to the tip of eachcrypt (Fig. 2A). Material corresponding to the location of mono-nuclear cells in the lamina propria was hybridized as well, incontrast to the control, in which “sense” strand alone was used(Fig. 2B).

Data management

Data from densitometry measurements are shown as relative inten-sity (mean ±1 SD), as defined above. For in situ studies, all exper-imental slides were quenched at the same time as correspondingcontrol slides. The latter were stained with control IgG monoclo-nal immunoglobulin, control normal rabbit serum (immunocyto-chemistry), or “sense” probe (in situ hybridization). Densitometryof control and experimental slides was performed at the same sit-ting to eliminate possible differences in background illuminationlevels between reading periods. Spectrophotometer readings inELISA assays are shown as absorbance at 405 nm over time. Sig-nificant differences were determined using the non-paired Stu-dent’s t-test (P<0.05).

173

Fig. 1A–C Identification of lamina propria mononuclear cell(LPMC) subset markers on cells expressing the prostanoid recep-tor EP4. Normal human colonic mucosa were sequentially exposedto immunoglobulin with specificity for EP4 and then for markers(CD3, CD4) of human T lymphocytes. The specificity of each im-munoglobulin used and the color of their substrates were as fol-lows: A EP4 (black), CD3 (red); B EP4 (black), CD4 (red); C irrel-evant antigens (black and red). Arrows indicate double-stainedcells in A and B. Magnification ×1000 (A–C)

Results

EP receptor expression in FAP mucosa

Immunocytochemistry

Prostanoids operate through ligand-specific EP receptorson cells’ plasma membranes to effect a biologicalchange. We were interested to know, therefore, what pat-tern of EP prostanoid receptor expression would be dem-onstrated by the intestinal mucosa of patients with a syn-drome, FAP, predisposed to an accelerated onset of co-lonic adenocarcinoma. When colonic FAP tissue sectionswere incubated with prostanoid receptor-specific immu-noglobulin, characterized below (Figs. 4, 5, 6), localiza-tion of receptor expression was found to be distinctive inseveral respects. First, the extent of EP3 receptor immu-noreactivity was broad in FAP, with expression on epi-thelium extending from the surface to the lower one-third of the crypt and into the crypt base (Fig. 3A). Sec-ond, EP4 receptor immunoreactivity was found on themajority of LPMC in FAP (Fig. 3B): by direct enumera-tion of >200 cells, 53.9±5.2% were EP4 positive. Third,EP2 receptor expression was restricted primarily to epi-thelium at the crypt apex in FAP (Fig. 3E). Findings

were similar in three consecutive FAP cases available tous. Thus, in FAP, the distribution of two EP receptors(EP3 and EP4) was prominent, with each receptor typeexpressed on most epithelial and/or lamina propria cells.

In situ hybridization for EP4

To determine whether the presence of immunoreactivereceptor protein was associated with detectable andcorresponding mRNA in situ, we hybridized digoxige-nin-labeled EP4 mRNA with tissue sections of FAP mu-cosa. “Antisense” mRNA hybridized with epithelialcells on the apical, lateral, and basal portions of epithe-lium, as well as with cells in the lamina propria(Fig. 3D). These results were consistent with thoseabove using methodology for detecting immunoreactiveprotein, indicating FAP epithelium and LPMC both ex-press prostanoid receptor EP4 at the protein and mRNAlevels.

Expression of EP receptors in normal human colonic mucosa

In addition to examining pathological tissues, it is impor-tant to know the normal distribution of receptors forPGE2 in colonic mucosa. Using receptor-specific immu-noglobulin with the avidin-biotin technique, we foundthat epithelium at the apex of crypts in normal humancolonic mucosa strongly expressed EP3 (Fig. 4A, E).Less immunoreactive EP3 was present on the lateral epi-thelium, with little or no receptor expression at the cryptbase. No cells in the lamina propria or muscularis muco-sa were positive; controls (Fig. 4C, D) were negative,and the antibody probe was specific for EP3 only(Fig. 4B). On the other hand, using immunoglobulinelicited to EP4 – shown to have no crossreactivity withpeptides from other related receptors (EP1, EP2, andEP3), as determined by ELISA (Fig. 5A) – the EP4 re-ceptor protein was found to be expressed on mononucle-ar cells scattered throughout the lamina propria(Fig. 5B). Direct enumeration, comparing EP- and H&E-stained sections, showed that only 21.7±1.2% of all cellsamong colonic LPMC were EP4 positive, compared with53.9±5.3% in FAP mucosa. In addition, epithelium ex-pressed EP4 receptor protein in a universal manner; forexample, staining of the lateral epithelium extendeddeeply into the crypt. Controls (Fig. 5C, D) were nega-tive, with only a low level residual stain remaining afterabsorption with EP4. Further studies showed EP2 recep-tor was expressed on epithelial cells at the crypt apex on-ly, with basolateral epithelium demonstrating no stainingabove background (Fig. 6A). Mononuclear cells in thelamina propria did not express EP2, and controls werenegative (Fig. 6C, D). In each case, for example, EP2,EP3, EP4, the immunoglobulin probe was shown by im-munoassay to be specific for a single member of the EP

174

Fig. 2A, B In situ hybridization to localize prostanoid receptorEP4 mRNA in human colonic mucosa. After prehybridization, 5-µm-thick sections of normal human mucosa were exposed to la-beled “antisense” and “sense” RNA probes specific for the EP4 re-ceptor (A and B, respectively). Arrows denote positively stainedepithelium (surface and basolateral). Arrowheads indicate cells inthe lamina propria that were EP4 positive. Magnification ×250

receptor family (Figs. 4B, 5A, 6B). With EP1-specificimmunoglobulin, on the other hand, no cells in normalcolonic mucosa, either LPMC or epithelium, expressedthis receptor protein (Fig. 6E). Thus, taking these studiestogether, there are at least three distinct receptors forPGE2 in normal human colon (for example, EP2, EP3,and EP4), each demonstrating its own pattern of localiza-tion among mucosal cells.

A comparison was made of prostanoid receptor ex-pression in FAP and normal colonic mucosa. While pros-tanoid receptor EP3 and EP4 immunoreactivity was pres-ent in both histologically normal and FAP colon (above),it remained possible that the degree of receptor proteinexpression per cell differed according to the histologicaldiagnosis. Using the Neurolucida 2.1 system to quant-itate the intensity of prostanoid receptor immunoreactivi-ty on individual cells in consecutive FAP and normal co-lons (three each), EP3 expression on all crypt epithelia

(surface and basolateral) was similar (3.2±1.4 versus3.0±0.5 for FAP and normal, respectively). However,deep crypt epithelium underlying polyps in FAP sectionsdemonstrated 2.8-fold more immunoreactive EP3 recep-tor, compared with normal crypt epithelium (P<0.01; Table 1). On the other hand, the relative intensity of EP3expression was similar at the crypt apex for epitheliumin FAP and normal mucosa (ratio of FAP/normal=0.94).Turning to prostanoid receptor EP4, immunoreactivity onFAP and normal epithelium in toto was similar (2.2±0.1versus 2.2±0.3, examining three colons with FAP andthree that were normal). Deep in FAP mucosa, however,EP4

+ epithelial expression was evident, while that of nor-mal (non-FAP) epithelia was less (Table 1; P<0.02). Incontrast, the relative intensity of EP4 immunoreactivityamong LPMC was nearly the same (FAP and normal).Overall, through quantifying immunoreactive protein onindividual cells in mucosa, the data suggest that the

175

Fig. 3A–E Localization ofprostanoid receptors EP2, EP3,and EP4 in familial adenoma-tous polyposis (FAP) colonicmucosa. Five-micron-thick sec-tions of colonic mucosa wereincubated with immunoglobu-lin having the following speci-ficities: A EP3; B EP4; C irrele-vant antigen; E EP2. In D, insitu hybridization with EP4“antisense” RNA was per-formed on FAP colonic muco-sa. In A, arrowheads mark EP3receptor protein in deep baso-lateral epithelium. In B and D,arrows indicate deep lateralcrypt epithelium and arrow-heads indicate individualLPMC, both expressing EP4.Inset in B shows detailed viewof epithelial cell staining withEP4-specific immunoglobulin.Magnification ×400

amount of two EP immunoreactive receptor proteins ex-pressed on deep basolateral epithelia in FAP is increased,indicating that particular epithelial cells in FAP mucosahave an increasednumber of receptors available to be li-gated by prostanoids.

Enzymes for synthesis of prostanoids in FAP

Lamina propria

While an increase in the distribution and intensity of im-munoreactive prostanoid receptor protein was found inFAP epithelium in the studies above, it is not knownwhether mucosal enzymes responsible for prostanoidsynthesis are altered in FAP as well. Comparing FAP andhistologically normal colonic mucosa (three each), therewere several major findings. In the lamina propria com-partment, COX 1 immunoreactivity was expressed bymononuclear cells in both FAP and normal colonic mu-cosa (Fig. 7A, C). Remarkably, most of the COX 1-posi-tive LPMC were localized to the subepithelial region atthe apex of each crypt; others were distributed through-out the lamina propria. In both instances (FAP and nor-mal), cells with a marked expression of COX 1 werepresent as a thick band in the muscularis mucosa. Theantigenic specificity of these findings was shown by theability of exogenous COX 1 antigen to inhibit immuno-globulin binding both in situ (Fig. 7F) and in vitro(Fig. 7B1). However, COX 2, not evident among LPMCin normal colonic tissue (Fig. 7D), was widely expressedin FAP on cells in the lamina propria (Fig. 7E).

Epithelium

In the colonic epithelial compartment, COX 2 immuno-reactive protein was evident from base to tip of FAP tis-

176

Fig. 4A–E Localization of the prostanoid receptor EP3 in normalhuman colonic mucosa. Five-micron-thick sections of histologi-cally normal human colon were incubated with sera having thefollowing specificities: A prostanoid receptor EP3 peptide; C irrel-evant antigen; D EP3 peptide, preabsorbed with 100:1 excess ofthe peptide; E same as A, but normal colonic mucosa from anothercase. Immunostaining intensity was substantially reduced whenEP3-specific immunoglobulin was preabsorbed with excess EP3peptide (D). Control immunoglobulin demonstrated no reactivitywith human colonic mucosa (C). Inset in A shows detailed view ofepithelial cell staining with EP3-specific immunoglobulin. Magni-fication ×250 (A, C–E); inset ×1000. B Specificity of the studyimmunoglobulin elicited, by EP3 peptide, to all EP receptor pep-tides (identified by symbol in the figure), measured as net absor-bance at 405 nm by enzyme-linked immunosorbent assay (ELISA)methodology. The immunoglobulin probe demonstrated minimalcrossreactivity with immunogenic peptides constituting other re-ceptors in the EP family, which can bind PGE2 (for example, EP1,EP2, and EP4); P was <0.05 for all points after 20 min, comparingreactivity to EP3 with that to other peptides

Table 1 Comparison of the intensity of prostanoid receptor EP ex-pression on individual cell types in mucosa of patients with a his-tologically normal colon and those with familial adenomatouspolyposis (FAP). (LPMC Lamina propria mononuclear cells)

Receptor Cell type Relative intensity of EP receptor expression

FAPa Normal colon Fold increase

EP3 EpitheliumSurface 3.2±0.4 3.0±0.5 0.94Basolateral 4.5±1.3* 1.6±0.1 2.8

EP4 EpitheliumSurface 2.2±0.6 1.9±0.1 1.2Basolateral 2.3±0.5** 0.8±0.2 2.9LPMC 5.8±0.6 5.9±0.9 1.04

*P<0.01, comparing relative intensity of EP receptor expressionon deep basolateral epithelium of FAP with that in normal mucosa**P<0.02, comparing deep basolateral epithelial cells in FAP andnormal mucosaa Reported as mean ±1 SD

sue (Fig. 7E), while only epithelium at the crypt apexwas modestly and intermittently positive in histological-ly normal tissue (Fig. 7D). The antigenic specificity ofthis finding was demonstrated by the ability of exoge-nous COX 2 to completely inhibit binding by immuno-globulin to the homologous study antigen (Fig. 7B2).

177

Fig. 5A–D Prostanoid receptor EP4 expression in the lamina pro-pria and epithelial compartments. A Antigenic specificity of anti-serum elicited by EP4, as determined by ELISA described inFig. 4B. All data points for immune reactivity to EP4 peptide after2 h were statistically different (P<0.05) from those of the otherpeptides (EP1, EP2, EP3). B–D In situ study to identify immunore-active EP4 prostanoid receptors in normal human colon using im-munoglobulin characterized in A. Controls consisted of colonicsections exposed to EP4-specific serum, preabsorbed with a 100:1excess of the homologous peptide (EP4; C), or to preimmune rab-bit serum (D). In B, arrows indicate EP4

+ lateral crypt epitheliumand arrowheads indicate positively stained LPMC. Magnification×250

Fig. 6A–E Presence of the prostanoid receptor EP2 but not EP1 onepithelial cells in normal human colonic mucosa. Five-micron-thick sections were incubated with a primary immunoglobulin,which had these antigenic specificities: A EP2; C irrelevant anti-gen; D EP2, but after preabsorption with a 100-fold excess of thehomologous antigen; E EP1. Arrows in A indicate epithelium atthe apex of crypts which strongly expressed EP2. B Antigen-spe-cific reactivity of an anti-EP2 immunoglobulin determined byELISA as described in Fig. 4B, and shown as the net absorbanceat 405 nm. All points for immunoglobulin reactivity to EP2 pep-tide after 1 h were statistically significant, P<0.05, compared withreactivity to other peptides

Table 2 Relative intensity of cyclooxygenase (COX) expressionin the colonic mucosa of patients with FAP compared to thatwhich is histologically normal. (EC Basolateral (non-surface) epi-thelia)

Tissue Cell type Relative intensity of COX expressiona

COX 1 COX 2b

FAP LPMC 20.7±6 22.7±11.1 (4.3)EC 0.5±0.5 8.9±1.6 (22.3)

Normal LPMC 21±7 5.3±2.4EC <0.1 0.4±0.2

a Reported as the mean ±1 SDb Number in parentheses represents fold increase of FAP value forCOX 2 over that in histologically normal tissue

COX 1 was not detected in epithelium of histologicallynormal or FAP-derived tissue (Fig. 7A, C). Thus, an en-zyme for synthesis of prostanoids (COX 2) appears to beinduced in FAP epithelia, both in the base and in the lat-eral aspects of crypts.

To determine by more quantitative means whetherdifferences exist in the degree of COX expression amongindividual cells in histologically normal and FAP colonic

mucosa, a densitometric analysis was performed on theimmunohistochemical sections described above. FAP tis-sue clearly demonstrated an enhanced COX 2 expres-sion, on a per cell basis, among LPMC and among baso-lateral epithelia (greater than fourfold increase of each;Table 2; comparing cells in Fig. 7E with those inFig. 7D). COX 1 expression among LPMC in FAP mucosa was similar in intensity to that of normal muco-

178

Fig. 7A–F Immunocytochemi-cal localization of cyclooxyge-nase (COX) 1 and COX 2 incolonic mucosa. For FAP tis-sue, the antigenic specificity of the immunoglobulins wasCOX 1 (A) and COX 2 (E). Inparallel studies, histologicallynormal colonic mucosa was ex-posed to immunoglobulin spe-cific for COX 1 (C), COX 2(D), and COX 1 preabsorbedwith a 100-fold excess ofCOX 1 (F). Arrows in A and Cindicate the subepithelial loca-tion of COX 1+ LPMC, and mmshows the location of the mus-cularis mucosa. Arrows in Dand E indicate COX 2+ epithe-lium. Magnification×400. In B,the ELISA was used to deter-mine the specificity of immu-noglobulin used in these stud-ies. The immunogen used foreliciting immunoglobulin wasCOX 1 (in B1) and COX 2 (inB2). **Indicates P<0.05 forbinding by immunoglobulin toCOX 1 or to COX 2, comparedwith control IgG (●–● ); *indi-cates P<0.05, comparing COXreactivity of immunoglobulinunabsorbed (▲––▲) versus ab-sorbed (■■–■■ ) with homologousantigen; for example, COX 1 inB1 and COX 2 in B2

sa (ratio of relative intensities of FAP/normal=0.98; Table 2; Fig. 7A compared with Fig. 7C). Thus, FAP co-lonic mucosa demonstrated an increased COX 2 expres-sion among cells in the lamina propria and in the baso-lateral epithelial compartment.

Discussion

Chronic ingestion of medications which block enzymaticactivity of COX and, consequently, inhibit prostanoidsynthesis, substantially reduce the incidence of colonicpolyps that arise spontaneously from human epitheliumas well as those that occur in familial syndromes (LaBayle et al. 1991; Suh et al. 1993; Thun et al. 1993;Giovannucci et al. 1994, 1995; Muscat et al. 1994;Schreinemachers and Everson 1994). Further, Yang re-cently reported that tissue levels of PGE2 were elevatedin adenomas compared with normal-appearing mucosa inFAP patients (Yang et al. 1998). These and other data(Reddy et al. 1993; Oshima et al. 1996) suggest a rolefor prostanoids in tumorigenesis of colonic epithelium.However, the biology of prostanoid synthesis and the lo-cation of cellular receptors for prostanoids in the colon isnot well understood. For example, it is not clear whichcolonic cell types have receptors to bind prostanoids,whether the expression of COX 1 or COX 2 is increasedin the mucosa of a precancerous polyposis syndrome,and which cell type in colonic mucosa expresses thesesynthase enzymes. In FAP, where tissue is available priorto the development of frank adenocarcinoma, the poten-tial exists to detect and investigate predisposing and en-hancing factors for tumor formation. In this paper, we re-port major changes in prostanoid biology in the FAP co-lonic mucosa which may have importance for epithelialtumorigenesis.

The final effect of PGE2 on epithelial biology in FAPis likely dependent on the expression of an EP receptor,the receptor density on the plasma membrane, and the in-tracellular events it can initiate once the receptor is ligat-ed. In investigating this, we found prostanoid receptorEP3 expression down the lateral wall of crypts in FAPcolonic mucosa by immunohistochemistry (Fig. 3A).This extension of receptors increases the number of co-lonic epithelial cells which can be ligated by PGE2 overa total colonic surface area of many meters. Ligation ofthe EP3 prostanoid receptor has been shown to result in adecrease in intracellular cAMP (Coleman et al. 1994).Further, the density of immunoreactive EP3 receptors onepithelia, deep in FAP crypts, was 2.8-fold that in normalcolonic mucosa. These elements (extension of receptorexpression, increased receptor density, and effective in-tracellular signaling) may allow prostanoids to drive pro-liferation of human intestinal epithelial cells in vivo asreported in vitro (Qiao et al. 1995), and may predisposeto cell transformation.

The finding that the T lymphocyte in the lamina pro-pria of the human colon is a predominant site of PG en-doperoxide synthase in normal and FAPmucosa was not

expected, and suggests that lymphocytes have a centralrole in synthesis of prostanoids in gut. In normal colon, aportion of the CD4+ subset of T lymphocytes expressedCOX 1, and these cells were often located at the apex ofcrypts, adjacent to epithelium. In FAP, immunoreactiveCOX 2 was newly expressed in lamina propria T lym-phocytes, in addition to expression of COX 1 in the up-per lamina propria. In fact, by densitometry, the quantity(relative intensity) of immunoreactive COX 1 and 2 ex-pressed per LPMC in FAP mucosa was >20, among thehighest values reported in our study. PGs generated byLPMC in the colon have only a short distance to diffuse,the width of one cell in some instances, to reach an ap-propriate receptor on epithelium; thispathway indeedmay be enhanced in FAP, where the number of COX-containing LPMC is increased. Prior studies of localiza-tion of COX in the normal intestinal tract have been limited, with one concluding that the major source ofPGE2 was mononuclear cells, most likely lymphocytes (Mikkelsen et al. 1991), and the other concluding thatCD68 (ostensibly a marker for macrophages) was coex-pressed with a minority of COX 2+ cells (Zifronic et al.1982). Given these findings by ourselves and others, onecould suggest that the major effect of aspirin is to reducethe flow of prostanoids to the epithelial compartment, asgenerated by mononuclear cells in the lamina propria.

Prostanoid synthesis by LPMC in human mucosalikely has a profound effect on the local mucosal im-mune response. Makaul and coworkers (1985) describeda marked inhibition by PGE2 of lymphokine secretion bymononuclear cells; the biochemical intermediary for thischange was a greater than tenfold elevation of lympho-cyte intracellular cAMP, which occurred within 25 minof exposure to PGE2. Barrera and coworkers (1996) re-ported a similar reduction by PGE2 of cytokine release(IL-2, -3) from freshly isolated human LPMC and frommouse T cell hybridomas derived from intestine. To theextent that the cytokine-driven portion of the immune re-sponse is broadly inhibited by PGE2, immune mecha-nisms for detection and removal of abnormal or trans-formed cells maybe depressed, thus increasing the poten-tial for tumorigenesis.

The limitations of our study are several. First, thesource of immunoglobulin used to detect peptides consti-tuting individual receptors in the EP family was hyper-immune serum and not monoclonal immunoglobulin.Nevertheless, the specificity of each serum was shownby its ability in ELISA to recognize the original (immu-nizing) peptide, a constituent of one EP receptor, and notheterologous peptides belonging to another member ofthe same receptor family (Figs. 4B, 5A, 6B), as well asby the ability of preabsorption of each serum with indi-vidual peptides to specifically diminish tissue staining byimmunoglobulin (Figs. 4D, 5C, 6D). Second, limitationsimposed on the morphological quality of tissue used forin situ hybridization was a concern, with prehybridiza-tion requiring exposure of tissue to Triton X-100 andother tissue-injurious reagents. Third, the potential for anunintended selection of cells used for imaging studies

179

and densitometry could have resulted in findings thatwere non-representative. To avoid selective bias in thisprocess, we obtained data on the universe of epithelialcells and LPMC in each field rather than on a sample. In-homogeneity in the staining of some cells may reflect thedistribution of the study antigen in the cell. Fourth, weconcentrated on preneoplastic conditions and predispos-ing factorsin largely histologically normal tissue. Studiesof COX in adenomas and frankly neoplastic tissue arereported (Eberhart et al. 1994; DuBois et al. 1996). In-vestigation of neoplasms of varying sizes, histologicaltypes, and stages of progression was clearly beyond thegoals of the current work, which focused on preneoplas-tic mucosa.

Overall, then, we have identified in FAP colonic mu-cosa an expanded distribution and density of prostanoidEP on epithelium, and a marked increase in COX expres-sion as well. Remarkably, the augmentation in COX wasfound not only on epithelium, but in the lamina propriaT lymphocytes, a site from which locally synthesizedprostanoids are well-positioned to influence adjacentepithelia. The evidence suggests that machinery for bothsynthesis of and cellular ligation to prostanoids in muco-sa is increased in FAP, and that lymphocytes serve as theprincipal and unexpected source of PG in colonic muco-sa, both normal and FAP. Future therapeutic options forretarding the development of colonic polyps could takeadvantage of these observations, by designing means tofurther reduce endogenous PG synthesis, by modifyingthe number and type of prostanoid receptors on epitheliaor LPMC, or byusing receptor antagonists that blockpharmacologically the effects of PGs on epithelia.

Acknowledgements This work was supported in part by a grantfrom the National Institute of Health (CA68226) to K.L. andJ.K.R. The authors wish to thank Ida Garrison for her excellenttyping of this manuscript.

References

Barrera S, Lai J, Fiocchi C, Roche JK (1996) Regulation by prostaglandin E2 of interleukin release by T lymphocytes inmucosa. J Cell Physiol 166:130–137

Bastien, L, Sawyer N, Grygorczyk R, Metters KM, Adam M(1994) Cloning, functional expression and characterization ofthe human prostaglandin E2 receptor EP2 subtype. J Bio Chem269:11873–11877

Bull DM, Bookman MA (1977) Isolation and functional character-ization of human intestinal mucosal lymphoid cells. J Clin In-vest 59:966–974

Coleman RA, Smith WL, Narumiya S (1994) Classification ofprostanoid receptors: properties, distribution and structure ofthe receptors and their subtypes. Pharmacol Rev 46:205–229

Cosme R, Lublin D, Takafuji V, Lynch K, Roche JK (2000) Pros-tanoids in human colonic mucosa: effects of inflammation onPGE2 receptor expression. Hum Immunol 61:684–696

DuBois RN, Radhika A, Reddy BS, Entingh AJ (1996) Increasedcyclooxygenase-2 levels in carcinogen-induced rat colonic tu-mors. Gastroenterology 110:1259–1262

Eberhart CE, Coffey RJ, Radhika A, Giardiello FM, Ferrenbach S,DuBois RN (1994) Up-regulation of cyclooxygenase-2 geneexpression in human colorectal adenomas and adenocarcino-mas. Gastroenterology 107:1183–1188

Funk CD, Furci L, Fitzgerald GA, Grygorczyk R, Rochette C,Bayne MA, Abramovitz M, Adam M, Metters KM (1998)Cloning and expression of a cDNA for the human prosta-glandin E receptor EP1 subtype. J Biol Chem 268:26767–26772

Giovannucci E, Rimm EB, Stampfer MJ, Colditz GA, Ascherio A,Willet WC (1994) Aspirin use and the risk for colorectal can-cer and adenoma in male health professionals. Ann Intern Med121:241–246

Giovannucci E, Egan K, Hunter D, Stampfer MJ, Colditz GA,Willet WC, Speizer FE (1995) Aspirin and the risk of colorec-tal cancer in women. N Engl J Med 333:609–614

Hirata M, Hayashi Y, Ushikubi F, Yokota Y, Kageyama R, Narumiya S (1991) Cloning and expression of a cDNA for a human thromboxane A2 receptor. Nature 349:617–620

Hirata M, Kakizuka A, Aizawa M, Ushikubi F, Narumiya S (1994)Molecular characterization of a mouse prostaglandin D recep-tor and functional expression of the cloned gene. Proc NatlAcad Sci USA 91:11192–11196

Itzkowitz SH, Kim YS (1998) Colonic polyps and polyposis syn-dromes. In: Feldman M, Scharschmidt BF, Sleisenger MH(eds) Gastrointestinal and liver disease, 6th edn. Saunders,Philadelphia, pp 1892

Katsuyama M, Sugimoto Y, Namba T, Irie A, Negishi M, Ichikawa A (1994) Cloning and expression of a cDNA for thehuman prostacyclin receptor. FEBS Lett 344:74–78

Kinzler KW, Vogelstein B (1996) Lessons from hereditary coloncancer. Cell 87:159–170

LaBayle D, Fischer D, Vielh P, Drouhin F, Pariente A, Bories C,Duhamel O, Trousset M, Attali P (1991) Sulindac causes re-gression of rectal polyps in familial adenomatous polyposis.Gastroenterology 101:635–639

Lake S, Gullberg H, Wahlqvist J, Sjogren AM, Kinhult A, Lind P,Hellstrom-Lindahl E, Stjernschantz J (1994) Cloning of the ratand human prostaglandin F2 alpha receptors and the expres-sion ofthe rat prostaglandin F2 alpha receptor. FEBS Lett355:317–325

Landis SH, Murray T, Bolden S, Wingo PA (1999) Cancer statis-tics. CA 49:8–31

Makoul GT, Robinson DR, Bhalla AK, Glimcher LH (1985) Prostaglandin E2 inhibits the activation of clones T cell hy-bridomas. J Immunol 134:2645–2650

Mikkelsen HB, Rumessen JJ, Qvortrup K (1991) Prostaglandin Hsynthase immunoreactivity in human gut. Histochemistry 96:295–299

Muscat JE, Stellman SD, Wynder EL (1994) Non-steroidal anti-inflammatory drugs and colorectal cancer. Cancer 74:1847–1854

Nicholas AP, Pieribone V, Hokfelt T (1993) Distributions ofmRNAs for α2 adrenergic receptor subtypes in rat brain: an insitu hybridization study. J Comp Neurol 328:575–594

Oshima M, Dinchuk JE, Kargman SL, Oshima H, Hancokc B,Kwong E, Trzaskos JM, Evans JF, Taketo MM (1996) Sup-pression of intestinal polyposis in AOCE (Ä–716) knockoutmice by inhibition of cyclooxygenase 2 (COX 2). Cell87:308–309

Planchon S, Fiocchi C, Takafuji V, Roche JK (1999) Transforminggrowth factor-β1 preserves epithelial barrier function: identifi-cation of receptors, biochemical intermediates, and cytokineantagonists. J Cell Physiol 181:55–66

Qiao L, Kozoni V, Tsioulias GJ, Koutsos MI, Hanif R, Shiff SJ(1995) Selected eicosanoids increase the proliferation rate ofhuman colon carcinoma cell lines and mouse colonocytes invivo. Biochim Biophys Acta 1258:215–223

Reddy BS, Rao CV, Rivenson A, Kelloff G (1993) Inhibitory ef-fect of aspirin on azoxymethane-induced colon carcinogenesisin F344 rats. Carcinogenesis 14:1493–1497

Regan JW, Bailey TJ, Pepperl DJ, Pierce KL, Bogardus AM, Donello JE, Fairbairn CE, Kedzie KM, Woodward DF, GilDW (1994) Cloning of a novel human prostaglandin receptorwith characteristics of the pharmacologically defined EP2 sub-type. Mol Pharmacol 46:213–220

180

Roche JK, Fiocchi C, Youngman K (1985) Sensitization to epithe-lial antigens in chronic mucosal inflammatory disease. J ClinInvest 75:522–530

Schreinemachers DM, Everson RB (1994) Aspirin use and lung,colon and breast cancer incidence in a prospective study. Epi-demiology 5:138–146

Sugimoto Y, Namba T, Honda A, Hayashi Y, Negishi M, IchikawaA (1992) Cloning and expression of a cDNA for mouse pros-taglandin E receptor EP3 subtype. J Biol Chem 267:6463–6466

Suh O, Mettlin C, Petrelli N (1993) Aspirin use, cancer and polypsof the large bowel. Cancer 72:1171–1177

Thun M, Namboodiri M, Calle E, Flanders W, Heath C (1993) Aspirin use and risk of fatal cancer. Cancer Res 53:1322–1327

Yang VW, Shidus JM, Hamilton SR, Spannhake EW, HubbardWC, Hylind LM, Robinson CR, Giardiella FM (1998) Size-dependent increase in prostanoid levels in adenomas of pa-tients with familial adenomatous polyposis. Cancer Res88:1750–1753

Zifronic A, Treves AJ, Sachar DB, Rachmilewitz D (1982) Pros-tanoid synthesis by cultured intestinal epithelial and mononu-clear cells in inflammatory bowel disease. Gut 24:659

181