Roles of Myofibroblasts in Prostaglandin E2–Stimulated
Intestinal Epithelial Proliferation and Angiogenesis
Jinyi Shao,1George G. Sheng,
2Randy C. Mifflin,
3Don W. Powell,
3and Hongmiao Sheng
1
1Department of Surgery and Cancer Center, Indiana University School of Medicine, Indianapolis, Indiana; 2Department of Surgery,University of Cincinnati, Cincinnati, Ohio; and 3Department of Medicine, University of Texas Medical Branch, Galveston, Texas
Abstract
Prostaglandins (PG) are produced throughout the gastroin-testinal tract and are critical mediators for a complex arrayof physiologic and pathophysiologic processes in the intestine.Intestinal myofibroblasts, which express cyclooxygenase(COX) and generate PGE2, play important roles in intestinalepithelial proliferation, differentiation, inflammation, andneoplasia through secreting growth factors and cytokines.Here, we show that PGE2 activated human intestinal sub-epithelial myofibroblasts (18Co) through Gs protein–coupledE-prostanoid receptors and the cyclic AMP/protein kinase Apathway. 18Co cells and primary colonic myofibroblast iso-lates expressed a number of growth factors; several of themwere dramatically regulated by PGE2. An epidermal growthfactor–like growth factor, amphiregulin (AR), which was notexpressed by untreated cells, was strongly induced by PGE2.Expression of vascular endothelial growth factor A (VEGFA)was rapidly increased by PGE2 exposure. Hepatocyte growthfactor (HGF) was elevated in PGE2-treated myofibroblasts atboth mRNA and protein levels. Thus, PGE2-activated myo-fibroblasts promoted the proliferation and migration of in-testinal epithelial cells, which were attenuated by neutralizingantibodies to AR and HGF, respectively. Moreover, in thepresence of PGE2, myofibroblasts strongly stimulated the mi-gration and tubular formation of vascular endothelial cells.Neutralizing antibody to VEGFA inhibited the observedstimulation of migration. These results suggest that myofibro-blast-generated growth factors are important mediators forPGE2-induced intestinal epithelial proliferation and angio-genesis, which play critical roles in intestinal homeostasis,inflammation, and neoplasia. (Cancer Res 2006; 66(2): 846-55)
Introduction
Prostaglandins (PG) are generated throughout the gastrointes-tinal tract and play critical roles in an array of physiologic andpathophysiologic processes (1, 2). PGs exert a trophic effect onsmall intestinal mucosa and stimulate intestinal epithelial cellproliferation (3). Short-term administration of PGE2 causessignificant stimulation of DNA synthesis; prolonged PGE2 treat-ment markedly increases the weight and DNA content of theintestinal mucosa (4). PGE2 and prostacyclin stimulate intestinalepithelial cell migration and therefore promote intestinal restitu-tion (5). Moreover, PGE2 exerts growth-stimulatory effects onintestinal tumors, and administration of PGE2 provides a growth
advantage to intestinal neoplasms (6, 7). In contrast, genetic dis-ruption of the cyclooxygenase-2 (COX-2) gene or the E-prostanoidreceptor 2 (EP2) results in a substantial reduction of polyps inAPC knockout mice (8, 9). Further evidence shows that PGE2promotes intestinal neoplasia through enhancing tumor angiogen-esis (9–11). Knockout of the EP2 receptor or inhibition of COX-2enzyme results in a reduction of neoangiogenesis in APCD716
mouse tumors (9, 12). Understanding the precise mechanisms bywhich PGE2 promotes intestinal epithelial growth and angiogenesisremains a significant challenge. It has been shown that PGE2directly stimulates the proliferation of transformed intestinalepithelial cells (6, 13, 14) and increases the expression ofproangiogenic growth factors in colon cancer cells (15, 16).However, the effects of PGE2 on cell growth and angiogenesisin vivo are considerably complex. Interactions between intestinalepithelial cells and stromal cells, which include fibroblasts,myofibroblasts, endothelial cells, and other cell types, maydramatically influence the homeostasis and transformation of theintestinal epithelium (17).A large body of studies has shown that intestinal subepithelial
myofibroblasts (ISEMF) play crucial roles in intestinal organo-genesis (18–21), proliferation, and differentiation of intestinalepithelial cells (19), mucosal protection, and wound healing (22).ISEMFs are located in the lamina propria throughout thegastrointestinal tract (23, 24) and act through the secretion ofgrowth factors, cytokines, and chemokines. ISEMFs express andproduce a large number of growth factors, including hepatocytegrowth factor (HGF; ref. 25), insulin-like growth factor (26),basic fibroblast growth factor (bFGF; ref. 27), platelet-derivedgrowth factor (PDGF; ref. 28), transforming growth factor-h(TGF-h; ref. 29), colony stimulating factor (30), nerve growthfactor (30), and stem cell factor (31). Furthermore, immunohis-tochemical studies reveal that fibroblasts are the predominantcell type in the lamina propria of normal colon; however, inboth hyperplastic and neoplastic polyps, interstitial fibroblasts arereplaced by myofibroblasts, suggesting that myofibroblasts playcritical roles in colorectal neoplasia (32).18Co cells were derived from human colonic mucosa and exhibit
many properties of intestinal subepithelial myofibroblasts (33).Expression of COX-1 is constitutive in 18Co cells, whereas COX-2can be induced by a variety of stimuli (34). Interleukin-1-activated18Co cells produce a significant amount of PGE2 (35). Given thecritical functions of both ISEMF and PGE2 in the intestine, wehypothesized that PGE2 may induce the production of certaingrowth factors by ISEMF, which, in turn, stimulate the growth andtransformation of intestinal epithelium. In the present study, weshow that PGE2 exposure increased the expression and secretionof amphiregulin (AR), HGF, and vascular endothelial growth factor(VEGF) in 18Co cells and as well as in primary human colonicmyofibroblasts. PGE2-activated 18Co cells stimulated the prolifer-ation and migration of intestinal epithelial cells. Conditioned
Requests for reprints: Hongmiao Sheng, Department of Surgery, IndianaUniversity, Indianapolis, IN 46202. Phone: 317-274-2630; E-mail: [email protected].
I2006 American Association for Cancer Research.doi:10.1158/0008-5472.CAN-05-2606
Cancer Res 2006; 66: (2). January 15, 2006 846 www.aacrjournals.org
media from PGE2-activated 18Co cells promoted the migration andtubular formation of vascular endothelial cells. Thus, our studysuggests that myofibroblasts may play critical roles in PGE2-induced intestinal growth and transformation.
Materials and Methods
Cell culture and reagents. 18Co cells were purchased from American
Type Culture Collection (Manassas, VA) and grown in MEM supplemented
with 10% fetal bovine serum (FBS) and nonessential amino acids. 18Co cells
used for this study were passages 12 to 14. Primary colonic myofibroblast(CMF, passages 4-10) cultures were established from histologically normal
margins of surgically resected colonic tissue using the outgrowth method
described by Mahida et al. (36, 37). The myofibroblast phenotype was
verified by immunohistochemistry and flow cytometry. All are positive fora-SMA, myosin heavy chain, and vimentin but are negative for cytokeratin
(epithelial cell marker), desmin (smooth muscle cell marker), factor VIII
(endothelial cell marker), CD45 (bone marrow–derived hematopoetic cellmarker), CD83 and ILT3 (dendritic cells), lysozyme and MAC 387 (both for
macrophages), and other markers of dendritic cells, B cells, or endothelia.
Rat intestinal epithelial (RIE) cells were a generous gift from Dr. Susan
Kirkland (University of London) and grown in DMEMwith 10% FBS. Humanumbilical vein endothelial cells (HUVEC) were purchased from Cascade
Biologics (Portland, OR) and grown in Medium 200 supplemented with
Sulprostone, and PGE1 alcohol were purchased from Cayman Chemical(Ann Arbor, MI). H-89, LY-294002, and PD-98059 were purchased from
Calbiochem (San Diego, CA). AR, HGF, and neutralizing antibodies were
purchased from R&D Systems (Minneapolis, MN).Growth factor array. To determine the relative expression levels of
growth factors, GEArray Q Series Human Growth Factor Gene Array
(SuperArray Bioscience Corporation, Frederick, MD) was carried out
according to the manufacturer’s instructions. Biotin-labeled probe wassynthesized from total RNA and hybridized with a nylon membrane printed
with cDNAs of 96 growth factors and cytokines. The array image was
captured with chemiluminescence detection and analyzed using the
software of GEArray Expression Analysis Suite.RIE cell-18Co cell coculture system and DNA synthesis. RIE cells
(5 � 103) suspended in 400 AL complete medium were placed in Transwell
chambers (0.4 Am, Corning Costar Co., Cambridge, MA) and then grown in
serum-free medium for 24 hours. Separately, confluent 18Co cells weregrown in a 24-well plate and treated with 0.5 Amol/L PGE2 for 24 hours.
Subsequently, Transwell chambers containing RIE cells were inserted into
the 24-well plate and grown for 24 hours. [3H]thymidine (1 ACi) was addedto the lower chambers 5 hours before harvest. The Transwell chambers
were washed thrice with 10% trichloroacetic acid times, and the filters
were collected from the chambers. Incorporation of [3H]thymidine was
determined using a scintillation counter.Cell migration assay. 18Co cells were grown in 24-well plates, serum
starved for 24 hours, and treated with vehicle or PGE2. RIE or HUVEC cells
suspended in 400 AL serum-free McCoy’s 5A medium were placed in
uncoated Transwell chamber (8 Am, Corning Costar). The Transwellchambers were then inserted into the 24-well plate containing 18Co cells.
After an incubation period of 5 hours at 37jC, cells on the upper surface of
the filter of Transwell chambers were removed with a cotton swab. Thefilters were fixed and stained with 0.5% crystal violet solution. Three
microscope fields (�200) from each Transwell chamber were randomly
selected, and cells adhering to the undersurface of the filter were counted.
HUVEC tube formation. HUVECs were suspended in 0.1 mL ofindicated conditioned media and placed on growth factor reduced Matrigel
(Collaborative Biomedical Products, Bedford, MA) in 96-well plates.
Morphology of the cells was documented using a digital camera attached
to an inverted microscope. Three photographs from random fields of eachmicrotiter well (quadruplicate wells for each group) were analyzed. Tubes
were defined as straight cellular extensions joining two cell masses (38).
Tube formation was assessed by the numbers of tubular structures and thelength of tubes.
RNA extraction and Northern blot analysis. Extraction of total cellularRNA was carried out as previously described (39). RNA samples were
separated on formaldehyde-agarose gels and blotted onto nitrocellulose
membranes. Blots were hybridized with cDNA probes labeled with
[a-32P]dCTP by random primer extension (Stratagene, La Jolla, CA). Afterhybridization and washes, the blots were subjected to autoradiography.
Reverse transcription-PCR. Expression of EP receptors in 18Co cells
was determined using reverse transcription-PCR (RT-PCR) as described
previously (14). Human HGF and VEGF primer pairs were purchased fromR&D Systems. RT-PCR was carried out using ProStar RT-PCR system
(Stratagene) according to the manufacturer’s instructions.
ELISA. Levels of human HGF, AR, and VEGF proteins in cell culture
media were quantified using ELISA kits (R&D Systems). Cells were seeded in24-well plate, and serum was deprived for 24 hours before PGE2 treatment.
Culture media were collected and stored at �80jC until assays.
Transient transfection and luciferase assay. Assays to determinetranscriptional activity were described previously (39). Briefly, 18Co cells
were transfected with 0.5 Ag of AR reporter plasmid (�850 to �87) or VEGF
reporter plasmid (�2279 to +54) along with 0.1 Ag of the pRL-SV plasmid,
containing the Renilla luciferase gene (Promega, Madison WI), using theFuGENE 6 procedure (Roche, Indianapolis, IN) as described in the
manufacturer’s protocol. Transfected cells were lysed at indicated times
for luciferase assay. Firefly and Renilla luciferase activities were measured
using a Dual-Luciferase Reporter assay system (Promega) and a lumin-ometer. Firefly luciferase values were standardized to Renilla values.
Immunoblot analysis. Immunoblot analysis was done as previously
described (14). Anti–phosphorylated extracellular signal-regulated kinase(pERK) and anti-phosphorylated Akt (pAkt) antibodies were purchased
from Cell Signaling Technology (Beverly, MA).
Data analysis. All statistical analyses were done on a personal computer
with the StatView 5.0.1 software (SAS Institute, Inc., Cary, NC). Analysesbetween two groups were determined using the unpaired Student’s t test.
Differences with P < 0.05 were considered as statistically significant.
Results
PGE2 induced stellate transformation of 18Co myofibro-blasts. In response to increased levels of intracellular cyclic AMP(cAMP), myofibroblasts undergo stellate transformation (40).Agents that increase cAMP levels including forskolin, choleratoxin, and PGE2 induce stellate morphology in 18Co cells (33).Confluent 18Co cells were grown in serum-deprived medium for24 hours and then treated with 0.5 Amol/L PGE2. Cells acquired astellate shape with dendritic-like processes by 2 hours following theaddition of PGE2 (Fig. 1A). By 24 hours, most stellate-transformedcells returned to their regular fibroblastoid morphology. A selectiveprotein kinase A (PKA) inhibitor, H-89 (5 Amol/L), completelyattenuated the PGE2-induced stellate transformation of 18Co cells.PGE2 acts via specific transmembrane G protein-coupled receptors;four E type prostaglandin (EP) receptor subtypes have beenidentified (41). EP2 and EP4 are known to increase intracellularcAMP levels and activate the PKA pathway. Expression of all offour EP receptors was detected in 18Co cells by RT-PCR (Fig. 1B).To determine which EP receptor mediated the PGE2-inducedstellate transformation, selective agonists were employed. Butap-rost, a selective EP2 agonist, strongly transformed 18Co cells;almost all cells acquired a stellate appearance by 2 hours afterthe treatment. Activation of EP4 receptor modestly inducedstellation; a small portion of 18Co cells was transformed by PGE1alcohol (Fig. 1B). As expected, EP1 and EP3 agonists (17-phenyl-trinor-PGE2 and Sulprostone) did not induce any morphologictransformation of 18Co cells (data not shown).The functional role of PGE2-induced stellation of 18Co cells is
complex (42). We found that PGE2 did not alter the proliferation of
PGE2 Activation of Myofibroblasts
www.aacrjournals.org 847 Cancer Res 2006; 66: (2). January 15, 2006
18Co cells in the absence of serum (data not shown); however,PGE2 significantly stimulated the migration of 18Co cells,determined by a modified Boyden chamber assay (Fig. 1C).Numbers of migrating 18Co cells increasedf1-fold in the presenceof 0.5 Amol/L PGE2 compared with the cells treated with vehicle.18Co cells expressed an array of growth factors. To
determine which growth factors were expressed by 18Co cells, wecarried out targeted cDNA arrays using GEArray Human GrowthFactor Array. 18Co cells were treated with either vehicle or PGE2for 1, 2, or 4 hours. The expression profile of 96 growth factors,cytokines, and chemokines was analyzed. Changes in geneexpression z2-fold were shown. As summarized in Table 1, 18Cocells expressed a number of growth factors, which belong to theepidermal growth factor (EGF), FGF, PDGF, HGF, TGF-h, andneuronal growth factor families. Expression of several growthfactors was significantly regulated by PGE2, including amphiregulin(AR), HGF, neuregulin 1, and VEGFA. Although AR was notexpressed under regular circumstances, AR expression was stronglyinduced by PGE2. A 27-fold increase was detected after 18Co cellswere exposed to PGE2 for 1 hour. HGF was constitutively expressedby 18Co cells, which was increased 5.8-fold at 4 hours after PGE2exposure. VEGFA was expressed at low levels in vehicle-treated18Co cells, which was increased 5-fold at the earliest time point.PGE2 increased the expression of AR. To validate the findings
made from the Growth Factor cDNA Array, we first investigated theregulation of AR by PGE2. AR mRNA was not detected in vehicle-treated 18Co cells; however, PGE2 rapidly increased the levels of ARmRNA, noted by Northern blot (Fig. 2A) and real-time RT-PCRanalysis (Fig. 2B). The PGE2-induced expression of AR remained atleast 24 hours. Secretion of AR was not detected in untreated 18Cocell culture medium. AR secretion, however, was robustly increasedin PGE2-stimulated 18Co cells, reaching f100 pg/mL by 24 hours
(Fig. 2C). The stimulatory action of PGE2 on AR production requiredactivation of the cAMP/PKA pathway; inhibition of PKA by H-89completely attenuated the PGE2-induced AR (Fig. 2D). In contrast,a mitogen-activated protein (MAP)/ERK kinase (MEK) inhibitor(PD-98059) and a phosphatidylinositol 3-kinase (PI3K) inhibitor(LY-294002) did not block PGE2-induced AR expression. Activationof either EP2 receptor or EP4 receptor increased the production ofAR; however, the EP2 signaling seemed to be the predominantpathway mediating AR induction (Fig. 2E). To determine theregulatory mechanism mediating PGE2 induction of AR, an ARpromoter-driven reporter plasmid was introduced into 18Co cells.PGE2 induced the activity of the AR promoter f7-fold, which wascompletely blocked by H-89 (Fig. 2F). Although Butaprost exerteda similar stimulatory effect on the AR promoter, PGE1 alcohol didnot have any effect on AR transcription in 18Co cells.PGE2 induced the expression of HGF. The Growth Factor
cDNA Array showed that HGF was constitutively expressed by18Co cells and significantly increased by 4 hours after PGE2exposure. RT-PCR analysis revealed that PGE2 did not significantlychange the expression of HGF at 1 hour; however, levels of HGFmRNA were robustly increased at 2 and 4 hours after PGE2treatment (Fig. 3A). HGF protein was detected in 18Co culturemedia at a concentration of f0.4 ng/mL. Addition of PGE2increased the levels of HGF f5-fold by 24 hours (Fig. 3B).Inhibition of the PKA pathway by H-89 attenuated thePGE2-induced production and secretion of HGF. Addition of eitherButaprost or PGE1 alcohol increased HGF production, suggestingthe involvement of both EP2 and EP4 signals in PGE2 inductionof HGF.PGE2-activated 18Co cells stimulated intestinal epithelial
proliferation and migration. To determine whether PGE2-activated 18Co cells were able to stimulate the proliferation of
Figure 1. PGE2 induction of stellate transformation of 18Co cells. A, confluent 18Co cell cultures were serum deprived for 24 hours before addition of vehicle (V ) or0.5 Amol/L PGE2 (E2 ). H-89 (5 Amol/L) was added 15 minutes before PGE2 treatment. Cells were photographed at the indicated times. B, expression of EP receptorswere analyzed by RT-PCR (top ). Butaprost or PGE1 alcohol at 0.5 Amol/L was added to serum-deprived 18Co cells. Morphologic alterations were documentedusing a digital camera attached to an inverted microscope (bottom ). C, 2 � 105 18Co cells were suspended in serum-free medium and seeded in 8-Am Transwellchambers. Vehicle or 0.5 Amol/L PGE2 were added into the lower chambers. After a 24-hour incubation, filters were fixed and stained with 0.5% crystal violet solution.Cells adhering to the undersurface of the filter were photographed (left and middle ) and counted (right ). Columns, mean of cell numbers done in triplicate; bars, SD.*, P < 0.05. Cell migration assays were done at least three times independently.
Cancer Research
Cancer Res 2006; 66: (2). January 15, 2006 848 www.aacrjournals.org
intestinal epithelial cells, we carried out experiments usingnontransformed RIE cells. Previous studies have shown that theEGF receptor (EGFR) is restricted to the basolateral compartmentof intestinal epithelial cells (43, 44). Therefore, when EGFR ligandsare added to the apical compartment, no mitogenic response isobserved. In contrast, basolateral administration of EGFR ligandsto intestinal epithelial cells grown on Transwell filters resultsin proliferation (44). RIE cells were insensitive to the growth-stimulatory effect of AR and HGF when grown on plastic dishes. ARslightly increased RIE cell proliferation only at relatively highconcentrations (10-100 ng/mL; Fig. 4A). However, when RIE cellswere plated on Transwell filters (0.4 Am); addition of AR to thelower chamber strongly stimulated the proliferation of RIE cells(Fig. 4B). Based on these results, RIE cells and 18Co cells werecocultured in a similar system, in which RIE cells were grown in theupper chamber and 18Co cells were grown in the bottom chamber.Neither PGE2 nor 18Co cells stimulated the proliferation of RIEcells; however, PGE2-activated 18Co cells increased DNA synthesisof RIE cells by f100% (Fig. 4C). Addition of anti-AR–neutralizingantibody significantly attenuated the growth advantage of RIE cellsthat were stimulated by PGE2-treated 18Co conditioned media,indicating the mitogenic effect of 18Co cell–generated AR (Fig. 4D).We next investigated if PGE2-stimulated 18Co cells modulate the
motility of intestinal epithelial cells. In a modified Boyden chamberassay, PGE2 did not stimulate the migration of RIE cells and thecells that migrated through the polycarbonate membrane retained
a cuboidal appearance. Placing 18Co cells in the bottom chambersignificantly increased the motility of RIE cells. However, whenactivated by PGE2, 18Co cells strongly stimulated the migrationof RIE cells; the number of migrating cells increased by f100%(Fig. 5A, right). Additionally, it was noted that the migrating RIEcells acquired a widely stretched morphology when coculturedwith PGE2-activated 18Co cells (Fig. 5A, left, d, arrows). Todetermine which growth factor mediated the promigratory action
Figure 2. PGE2 induction of AR expression. A, 18Co cells were serum deprivedfor 24 hours before PGE2 treatment. Levels of AR mRNA were analyzed byNorthern blot. B, 18Co cells were serum deprived for 24 hours before PGE2
treatment. Levels of AR mRNA were analyzed by real-time RT-PCR. C, 18Cocells were serum deprived for 24 hours and treated with vehicle (V ) or PGE2
(E2 ) for the indicated times. Levels of AR protein in cell culture media weredetermined by ELISA assay. Columns, mean of AR content done in triplicate;bars, SD. *, P < 0.05. ELISA assays were done at least three timesindependently. D, 18Co cells were treated with vehicle, 5 Amol/L H-89 (H ),25 Amol/L PD-98059 (P), or 10 Amol/L LY-294002 (L) for 15 minutes before theaddition of 0.5 Amol/L PGE2. After a 24-hour incubation, levels of AR in cellculture media were determined by ELISA assay. Columns, mean of AR done intriplicate; bars, SD. *, P < 0.05. E, 18Co cells were serum deprived for 24 hoursbefore treatments (V = ethanol, E2 = 0.5 Amol/L PGE2, EP1/3 = 0.5 Amol/L17-phenyl-trinor-PGE2, EP2 = 0.5 Amol/L Butaprost, EP3 = 0.5 Amol/LSulprostone, and EP4 = 0.5 Amol/L PGE1 alcohol). After 24 hours, levels of ARprotein were determined by ELISA assay. Columns, mean of AR content done intriplicate; bars, SD. *, P < 0.05. F, 18Co cells were transiently transfectedwith an AR promoter reporter vector. Cells were then treated with ethanol (V ),0.5 Amol/L PGE2 (E2 ), 0.5 Amol/L Butaprost (But ), or 0.5 Amol/L PGE1 alcohol (E1 )(E1 ) along with 5 Amol/L H-89 or DMSO (V ) for 6 hours. Firefly and Renillaluciferase activities were measured and standardized. Columns, mean of Renillaadjusted luciferase activity done in quadruplicate; bars, SD. *, P < 0.05.Representative of three separate experiments.
Table 1. PGE2 regulation of 18Co-expressed growthfactors
of PGE2-activated 18Co, RIE cells were stimulated with AR or HGF.We found that AR did not increase RIE cell migration, but HGFsignificantly stimulated the migration of RIE cells (Fig. 5B). Inagreement with this observation, anti-HGF–neutralizing antibodycompletely blocked the PGE2/18Co–induced RIE cell migration(Fig. 5C).Proliferation and migration of intestinal epithelial cells require
activation of the MAP kinase and the PI3K pathways (45). Todetermine whether PGE2-stimulated 18Co cells activated thesesignaling pathways in intestinal epithelial cells, levels of pERK andpAkt in RIE cells were analyzed. Addition of PGE2 did not changethe levels of pERK and pAkt in RIE cells (data not shown). 18Coconditioned media rapidly increased the levels of pERK and pAkt.
However, PGE2-activated 18Co conditioned media exerted signif-icantly stronger effects on activation of both MEK/ERK and PI3K/Akt pathways (Fig. 5D).PGE2 increased the production of VEGF in 18Co cells. COX-
2/PGE2 mediates hypoxic induction of VEGF in hepatic stellatecells (46). Our results of Growth Factor Array showed that PGE2exposure induced the expression of VEGFA, suggesting that VEGFAis a PGE2 target gene in 18Co cells. Exposure to PGE2 rapidlyincreased the levels of VEGFA mRNA in 18Co cells, noted byNorthern analysis (Fig. 6A). Similar results were observed usingreal-time RT-PCR; levels of VEGFA mRNA increased f3.5-foldafter the 18Co cells were treated with PGE2 for 1 hour (Fig. 6B).Moreover, PGE2 treatment increased the production and secretionof VEGFA protein. Levels of VEGFA protein were elevated f3.5-fold in PGE2-stimulated 18Co culture media (Fig. 6C). Induction ofVEGFA production was mediated by both EP2 and EP4 receptors,because both Butaprost and PGE1 alcohol increased the levels ofVEGFA mRNA (Fig. 6D) and protein (Fig. 6E). Addition of H-89completely attenuated the PGE2-induced expression of VEGFA atboth mRNA and protein levels. Furthermore, PGE2 stimulatedVEGF transcription through both EP2 and EP4 signaling. Additionof PGE2, Butaprost, and PGE1 alcohol increased the activity ofVEGF promoter f1-fold (Fig. 6F).PGE2-activated 18Co cells enhanced angiogenesis. Because
PGE2 induced the expression of VEGF in 18Co cells, it was ofinterest to determine whether PGE2-activated 18Co cells enhancedangiogenesis. Proangiogenic factors, including VEGF, stimulateneoangiogenesis by inducing endothelial cell proliferation, migra-tion, and tubular organization. The effects of 18Co cells on themigration of endothelial cells were evaluated in a coculture systemsimilar to the system described in Fig. 4D . HUVECs were seeded inthe Transwell; 18Co cells were placed in the bottom chamber.Addition of PGE2 into the bottom chamber without 18Co cellsslightly stimulated the migration of HUVECs. The presence of 18Cocells in the bottom chamber significantly increased the motility ofHUVECs. When cocultured with PGE2-activated 18Co cells,HUVECs acquired a widely stretched morphology (Fig. 7A), and
Figure 4. RIE cell proliferation in coculture with 18Co cells. A, RIE cells (2.5 � 104) were seeded in 24-well plates and subjected to serum deprivation for 24 hours. ARor HGF at indicated concentrations were added. After a 24-hour incubation, DNA synthesis was analyzed by [3H]thymidine incorporation. Columns, mean of CPMdone in quadruplicate; bars, SD. Representative of three separate experiments. B, 4 � 103 RIE cells were seeded in a Transwell (0.4 Am) and serum deprived for24 hours. AR or HGF at the indicated concentrations were added to the lower chambers. After 24 hours, [3H]thymidine (1 A Ci) was added. The filters containing RIEcells were collected, and [3H]thymidine incorporation was measured using a scintillation counter. C, in a coculture system, 4 � 103 RIE cells were seeded in theTranswell (0.4 Am) and serum deprived for 24 hours before being inserted into a 24-well plate, where serum-starved 18Co cells were stimulated with vehicle (V) or0.5 Amol/L PGE2 for 24 hours. [3H]thymidine incorporation in RIE cells was determined. D, 18Co cells were treated with vehicle (V ) or PGE2 (E2) for 24 hours, andthen 18Co conditioned media were collected. Normal goat IgG (IgG , 20 Ag/mL), anti-AR–neutralizing antibody (aAR , 20 Ag/mL), or anti-HGF–neutralizing antibody(aHGF , 20 Ag/mL) was added to the 18Co conditioned media and incubated at 4jC for 30 minutes. RIE cells (4 � 103) were seeded in the Transwell (0.4 Am) and serumdeprived for 24 hours before being inserted into a 24-well plate, which contained pretreated 18Co conditioned media as indicated. [3H]thymidine incorporation inRIE cells was determined after a 24-hour incubation and a 5-hour pulse.
Figure 3. Expression of HGF in 18Co cells. A, 18Co cells were serum deprivedfor 24 hours before addition of ethanol (V ) or 0.5 Amol/L PGE2 (E2). TotalRNA was extracted at the indicated time points, and the expression of HGFmRNA was analyzed by RT-PCR. B, 18Co cells were serum deprived for24 hours before treatments (V = ethanol, E2 = 0.5 Amol/L PGE2, EP1/3 =0.5 Amol/L 17-phenyl-trinor-PGE2, EP2 = 0.5 Amol/L Butaprost, EP3 = 0.5 Amol/LSulprostone, and EP4 = 0.5 Amol/L PGE1 alcohol). Supernatants of the cellcultures were collected after 24 hours, and levels of HGF protein weredetermined by ELISA assay. Columns, mean of HGF content done in triplicate;bars, SD. *, P < 0.05. ELISA assay was repeated at least three times.
Cancer Research
Cancer Res 2006; 66: (2). January 15, 2006 850 www.aacrjournals.org
Figure 5. RIE cell migration in coculture with 18Co cells.A, 2 � 105 RIE cells suspended in serum-free medium wereseeded in Transwell chambers (8 Am). The chambers were theninserted into 24-well plates, where 18Co cells were grown andstimulated with vehicle (V ) or 0.5 Amol/L PGE2 for 24 hours. After a5-hour incubation, filters were fixed and stained with 0.5% crystalviolet solution. Cells adhering to the undersurface of the filterwere photographed (left ). Cell numbers in three microscope fields(�200) from each Transwell were counted. Columns, mean ofmigrating cells done in triplicate; bars, SD. B, 2 � 105 RIE cellssuspended in serum-free medium were seeded in Transwellchambers (8 Am). Vehicle (V ), 100 ng/mL AR or 10 ng/mL HGFwere added to the lower chambers. RIE cell migration wasdetermined after a 5-hour incubation as above. C, 2 � 105 RIEcells suspended in serum-free medium were seeded in Transwellchambers (8 Am). Vehicle (V ) or PGE2 (E2 )–stimulated 18Coconditioned media that were pretreated with normal IgG (IgG ) or20 Ag/mL anti-HGF–neutralizing antibody (aHGF ) were added tothe lower chambers. RIE cell migration was determined after a5-hour incubation. D, 18Co cells were serum deprived for 24 hoursand then stimulated with vehicle (V ) or 0.5 Amol/L PGE2 (E2) for24 hours before collecting the conditioned media. To eliminatethe direct effect of PGE2, 0.5 Amol/L PGE2 was added tovehicle-treated conditioned media. 18Co conditioned media werethen added to serum-deprived RIE cells, and cellular protein wasextracted at the indicated time points. Levels of pERKs andpAkt were determined by Western analysis.
Figure 6. Expression of VEGF in 18Co cells. A and B, 18Co cells were serum deprived for 24 hours before addition of ethanol (V ) or 0.5 Amol/L PGE2 (E2). Total RNAwas extracted at the indicated time points, and levels of VEGF mRNA were analyzed by Northern blot (A) and real-time RT-PCR (B ). C, 18Co cells were serumdeprived for 24 hours before PGE2 exposure. After 24 hours, conditioned media were collected, and levels of VEGF protein were determined by ELISA assay.Columns, mean of VEGF content done in triplicate; bars, SD. *, P < 0.05. D and E, 18Co cells were serum deprived for 24 hours before treatments (V = ethanol,E2 = 0.5 Amol/L PGE2) in the presence of a PKA inhibitor (5 Amol/L H-89) or DMSO (V ). Cells were also treated with EP agonists (EP1/3 = 0.5 Amol/L17-phenyl-trinor-PGE2, EP2 = 0.5 Amol/L Butaprost, EP3 = 0.5 Amol/L Sulprostone, and EP4 = 0.5 Amol/L PGE1 alcohol). D, levels of VEGF mRNA weredetermined by RT-PCR after the cells were treated for 2 hours. E, after a 24-hour incubation, levels of VEGF protein in culture media were determined by ELISA assay.Columns, mean of VEGF content done in triplicate; bars, SD. *, P < 0.05. F, 18Co cells were transiently transfected with a VEGF promoter reporter vector. Cellswere subjected to the indicated treatments for 6 hours (V = ethanol, E2 = 0.5 Amol/L PGE2, But = 0.5 Amol/L Butaprost, and E1 = 0.5 Amol/L PGE1 alcohol). Firefly andRenilla luciferase activities were measured and standardized. Columns, mean of Renilla -adjusted luciferase activity done in quadruplicate; bars, SD.*, P < 0.05. Representative of three separate experiments.
PGE2 Activation of Myofibroblasts
www.aacrjournals.org 851 Cancer Res 2006; 66: (2). January 15, 2006
the number of migrating cells was robustly increased (Fig. 7B, left),which, however, was significantly attenuated by the addition ofanti-VEGFA–neutralizing antibody (Fig. 7B, right).To determine the effects of 18Co cell–released growth factors on
tubular organization, HUVECs were placed on growth factor–reduced Matrigel. HUVECs spontaneously form tubular structureson extracellular matrix. Addition of PGE2 stimulated HUVEC tubeformation by 8 hours, as quantitated by the numbers and thelength of tubes (Fig. 7C). Conditioned media that collected fromPGE2-stimulated 18Co cells robustly increased the number andlength of tubes of HUVECs. Moreover, HUVEC-formed tubes weredissociated by 24 hours (Fig. 7D). Addition of PGE2-treated 18Coconditioned media prevented HUVEC tube dissociation. Interest-ingly, the presence of anti-VEGFA–neutralizing antibody did notsignificantly reduce the PGE2/18Co–induced HUVEC tube forma-tion (data not shown), suggesting the involvement of a complexmechanism.PGE2 induction of growth factors in primary myofibro-
blasts. To determine whether the observations from 18Co cells
may actually occur in the human intestine, the induction of growthfactors by PGE2 was evaluated in human primary subepithelialCMF. Three CMF primary isolates (4, 5, and 7) were treated withPGE2; levels of AR, HGF, and VEGFA were determined by RT-PCR.PGE2 exposure robustly increased the expression of AR mRNA in allCMF cultures, which normally did not express AR (Fig. 8A). HGFand VEGFA mRNAs were induced by PGE2 in a majority of CMFisolates. Moreover, ELISA assay revealed that whereas HGF andVEGF proteins were increased f1-fold in PGE2-stimulated CMFculture media, the presence of PGE2-induced protein levels of AR>10-fold in all CMF isolates (Fig. 8B).
Discussion
COX-2 is not expressed in normal intestinal mucosa; its activityincreases dramatically in inflammation, injury, and neoplasia of theintestine (47). In studies of human colorectal cancer, COX-2 levelsare increased in about 90% of cancers and f50% of premalignantcolorectal adenomas, but the enzyme is not usually detected in
Figure 7. HUVEC migration and tubularformation in coculture with 18Co cells.A and B, 1 � 105 HUVECs suspended inserum-free medium were seeded inTranswell chambers (8 Am). The chamberswere then inserted into 24-well plates,where 18Co cells were grown andstimulated with vehicle (V ) or 0.5 Amol/LPGE2 for 24 hours. After a 5-hourincubation, filters were fixed and stainedwith 0.5% crystal violet solution. A, cellsadhering to the undersurface of thefilter were photographed. B, numbers ofmigrating cells in three microscope fields(�200) from each Transwell were counted.Columns, mean of migrating cells done intriplicate (left ); bars, SD. HUVEC migrationwas assessed when vehicle (V ) or PGE2
(E2)–stimulated 18Co conditioned mediathat were pretreated with normal IgG(IgG ) or anti-VEGF–neutralizing antibody(aVEGF , 20 Ag/mL) were added to thelower chambers (right ). C, 1 � 104 HUVECsuspended in serum-free mediumcontaining vehicle (V ), 0.5 Amol/L PGE2
(E2), vehicle-treated 18Co conditionedmedia, or PGE2-activated 18Coconditioned media were placed onto growthfactor–reduced Matrigel. After an 8-hourincubation, cells were photographed(left , �100). Numbers of tubes werecounted (middle ), and the relative lengthof the tubular structure was measured(right ). D, 5 � 103 HUVEC cellssuspended in serum-free mediumcontaining vehicle (V ), 0.5 Amol/L PGE2
(E2), vehicle-treated 18Co conditionedmedia, or PGE2-activated 18Coconditioned media were placed onto growthfactor–reduced Matrigel. After a 24-hourincubation, cells were photographed(left , �100).
Cancer Research
Cancer Res 2006; 66: (2). January 15, 2006 852 www.aacrjournals.org
adult intestinal tissues (48, 49). Although there are conflicting dataregarding which cell types express COX-2 in intestinal tumors,COX-2 expression increases in both the epithelial and the stromalcompartments (50). Indeed, PGs can be produced by a variety ofcell types, including normal and transformed intestinal epithelialcells, myofibroblasts, and macrophages (34, 51, 52). PGs serve asautocrine or paracrine lipid mediators to signal changes withintheir immediate environment, suggesting that PGE2 may mediateinteractions between intestinal epithelial cells and stromal cellsthrough both autocrine and paracrine mechanisms. PGE2 derivedfrom both stromal and epithelial compartments may stimulatestromal cells to release growth factors, which, in turn, provide apro-neoplastic environment for the intestinal epithelium. In thepresent study, we found that exogenous PGE2 induced theexpression and secretion of several pro-proliferative and proangio-genic growth factors in intestinal subepithelial myofibroblasts,providing the evidence that myofibroblasts may be a criticalmediator for COX-2/PGE2–mediated intestinal epithelial growth,transformation, and neoangiogenesis.Our results show that exogenous PGE2 induced the expression of
AR at both mRNA and protein levels in 18Co cells. AR is a memberof the EGF growth factor family and a ligand of the EGFR. It hasbeen shown that AR is a primary mitogen for hepatocytes andcritical in the early steps of liver regeneration; AR-null mice displayan impaired proliferative response after partial liver resection (53).Moreover, AR exerts tumor-promoting effects on colorectalcarcinomas. AR mRNA is expressed in 60% to 70% of primary andmetastatic human colorectal carcinomas but in only 2% to 7% ofnormal colonic mucosa samples studied (54). AR plays critical rolesin colon cancer cell proliferation and transformation that arerequired for the growth of human colon carcinoma xenografts (55).We have reported that in response to PGE2 exposure, the expressionof AR is significantly increased in transformed intestinal epithelialcells, which stimulates the growth of colon cancer cells via anautocrine mechanism (39, 56). Recent studies have further stressedthe critical role of AR in the transformation of a variety of epithelialcell types. Chang et al. (57) showed that COX-2 overexpression in themammary gland of transgenic mice induced mammary cancer.Interestingly, genetic deletion of the EP2 receptor significantlyreduced the COX-2-induced mammary cancer, which is associated
with a dramatic down-regulation of AR. In contrast, an EP2-specificagonist strongly increases the expression of AR in mammary cancercell lines, suggesting that AR is a mediator for COX-2/PGE2–inducedmammary gland hyperplasia. In another study, Moraitis et al. (58)reported that tobacco smoke stimulates the expression of COX-2through activating the EGFR signaling system in human oralmucosa. Overexpression of AR and TGF-a was determined to be themechanism for the tobacco smoke–induced EGFR activity andCOX-2 expression, suggesting a positive loop between the COX-2/PGE2 pathway and AR/EGFR signaling.The growth of solid tumors requires a blood supply that is
achieved through neoangiogenesis. VEGF is one of the majorregulators for neoangiogenesis, which induces endothelial cellproliferation, migration, and tubular organization (19). PGE2induces the expression of VEGF in colon cancer cells and inAPCmin/+ polyps (16). EP2-mediated PGE2 signaling plays criticalroles in neoangiogenesis. Homozygous deletion of the EP2 receptorsignificantly reduces the number and size of intestinal polyps inAPCD716 mice that is associated with a reduction of VEGFexpression, suggesting that PGE2/EP2 signaling is critical forincreased levels of VEGF in intestinal neoplasm (9). Our data showthat PGE2 increased the expression, production, and secretion ofVEGF in 18Co cells and that PGE2-activated 18Co cells promotedthe migration and tubular formation of HUVEC. This suggests thatmyofibroblasts may provide proangiogenic factors for intestinalremodeling and transformation. In addition to VEGF, a number ofmembers of the FGF family and the TGF-h family were expressedby 18Co cells and regulated by PGE2; their functional roles in PGE2proangiogenic actions are still under investigation.HGF is a known myofibroblast-derived growth factor that
regulates epithelial cell proliferation, differentiation, motility, andmorphology (59). HGF is expressed in the stomach, small intestine,and colon (60). The biological functions of HGF in the smalland large intestines are not clear, but it is likely that HGF exertsgrowth-stimulatory effects on intestinal epithelial cells. Whencolon cancer T84 cells are grown in type I collagen gel, HGF doesnot induce their differentiation but stimulates their growth (61).Our results showed that PGE2-activated myofibroblasts increasedthe production of HGF, which predominantly stimulated themigration of intestinal epithelial cells.
Figure 8. PGE2 induction of growth factors in human colonicmyofibroblasts. A, human primary colonic myofibroblast cellisolates CMF4, CMF5, and CMF7 were serum deprived for24 hours before PGE2 treatment. Total RNA was extracted at theindicated time points. Levels of AR, HGF, VEGF, and h-actinwere analyzed by RT-PCR. B, CMF4, CMF5, and CMF7 cells wereserum deprived for 24 hours before vehicle (V ) or PGE2 (E2)treatment. After a 24-hour incubation, protein levels of AR, HGF,and VEGFA in conditioned media were measured by ELISA assay.
PGE2 Activation of Myofibroblasts
www.aacrjournals.org 853 Cancer Res 2006; 66: (2). January 15, 2006
PGE2 signals through specific receptors, including EP1, EP2, EP3,and EP4. Our results showed that PGE2-induced expression andproduction of growth factors was mediated by the EP2,4/cAMP/PKA pathway, because both Butaprost and PGE1 alcohol repro-duced PGE2 actions on induction of HGF, AR, and VEGF. Insupport of these findings, a selective PKA inhibitor attenuated thestimulatory effects of PGE2. Although both EP2 and EP4 pathwaysacted quite similarly, differences were observed. For example, ARwas regulated by PGE2 at the transcriptional level, which wasmediated by the EP2/cAMP/PKA pathway only. PGE1 alcohol didnot stimulate the transcription of AR but modestly increased thelevels of AR protein in 18Co culture media, suggesting theinvolvement of a post-transcriptional regulation. Interestingly,activation of both EP2 and EP4 increased the transcription ofVEGF . Because PGE1 alcohol also binds to other EP receptors (62),its specificity to EP4 is relative. Further experiments are requiredto determine the precise function of EP4 in PGE2 activation of18Co cells.
In summary, our studies suggest that myofibroblasts are apotential mediator of the growth-stimulatory effect and pro-neoplastic action of PGE2 in the intestine. Myofibroblasts mayreceive PGE2 stimulation via autocrine and paracrine pathways.Upon activation by PGE2, myofibroblasts increasingly produce andsecrete growth factors, which stimulate intestinal epithelial cellgrowth and promote angiogenesis. Given the special localization ofmyofibroblasts in normal intestine and intestinal neoplasia, theinteraction between myofibroblasts and intestinal epithelial cellsmay play important roles in intestinal epithelial growth andtransformation.
Acknowledgments
Received 7/25/2005; revised 10/10/2005; accepted 11/8/2005.Grant support: NIH grants DK-065615, DK-064593 (H. Sheng), and DK-55783
(D.W. Powell).The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.
Cancer Research
Cancer Res 2006; 66: (2). January 15, 2006 854 www.aacrjournals.org
References1. Mohajer B, Ma TY. Eicosanoids and the smallintestine. Prostaglandins Other Lipid Mediat 2000;61:125–43.
2. Krause W, DuBois RN. Eicosanoids and the largeintestine. Prostaglandins Other Lipid Mediat 2000;61:145–61.
3. Uribe A, Alam M, Midtvedt T. E2 prostaglandinsmodulate cell proliferation in the small intestinalepithelium of the rat. Digestion 1992;52:157–64.
4. Dembinski A, Konturek SJ. Effects of E, F, and I seriesprostaglandins and analogues on growth of gastroduo-denal mucosa and pancreas. Am J Physiol 1985;248:G170–5.
5. Blikslager AT, Roberts MC, Rhoads JM, et al.Prostaglandins I2 and E2 have a synergistic role inrescuing epithelial barrier function in porcine ileum.J Clin Invest 1997;100:1928–33.
6. Pai R, Soreghan B, Szabo IL, et al. Prostaglandin E2transactivates EGF receptor: a novel mechanism forpromoting colon cancer growth and gastrointestinalhypertrophy. Nat Med 2002;8:289–93.
7. Wang D, Buchanan FG, Wang H, et al. ProstaglandinE2 enhances intestinal adenoma growth via activation ofthe Ras-mitogen-activated protein kinase cascade.Cancer Res 2005;65:1822–9.
8. Oshima M, Dinchuk JE, Kargman S, et al. Suppressionof intestinal polyposis in ApcD716 knockout mice byinhibition of cyclooxygenase 2 (COX-2). Cell 1996;87:803–9.
9. Sonoshita M, Takaku K, Sasaki N, et al. Accelerationof intestinal polyposis through prostaglandin receptorEP2 in Apc(Delta 716) knockout mice. Nat Med 2001;7:1048–51.
10. Tsujii M, Kawano S, Tsuji S, et al. Cyclooxygenaseregulates angiogenesis induced by colon cancer cells.Cell 1998;93:705–16.
11. Williams CS, Tsujii M, Reese J, et al. Hostcyclooxygenase-2 modulates carcinoma growth. J ClinInvest 2000;105:1589–94.
12. Oshima M, Murai N, Kargman S, et al. Chemo-prevention of intestinal polyposis in the Apcdelta716mouse by rofecoxib, a specific cyclooxygenase-2 inhib-itor. Cancer Res 2001;61:1733–40.
13. Sheng H, Shao J, Kirkland SC, et al. Inhibition ofhuman colon cancer cell growth by selective inhibitionof cyclooxygenase-2. J Clin Invest 1997;99:2254–9.
14. Sheng H, Shao J, Washington MK, et al. ProstaglandinE2 increases growth and motility of colorectal carcino-ma cells. J Biol Chem 2001;276:18075–81.
15. Fukuda R, Kelly B, Semenza GL. Vascular endothelialgrowth factor gene expression in colon cancer cellsexposed to prostaglandin E2 is mediated by hypoxia-inducible factor 1. Cancer Res 2003;63:2330–4.
16. Shao J, Jung C, Liu C, Sheng H. Prostaglandin E2stimulates the b-catenin/T cell factor-dependenttranscription in colon cancer. J Biol Chem 2005;280:26565–72.
17. Elenbaas B, Weinberg RA. Heterotypic signalingbetween epithelial tumor cells and fibroblasts incarcinoma formation. Exp Cell Res 2001;264:169–84.
18. Finch PW, Cunha GR, Rubin JS, et al. Pattern ofkeratinocyte growth factor and keratinocyte growthfactor receptor expression during mouse fetal develop-ment suggests a role in mediating morphogeneticmesenchymal-epithelial interactions. Dev Dyn 1995;203:223–40.
19. Fritsch C, Simon-Assmann P, Kedinger M, Evans GS.Cytokines modulate fibroblast phenotype and epithelial-stroma interactions in rat intestine. Gastroenterology1997;112:826–38.
20. Powell DW, Mifflin RC, Valentich JD, et al. Myofibro-blasts. I. Paracrine cells important in health and disease.Am J Physiol 1999;277:C1–9.
21. Powell DW, Mifflin RC, Valentich JD, et al. Myofibro-blasts. II. Intestinal subepithelial myofibroblasts. Am JPhysiol 1999;277:C183–201.
22. Wallace JL, Granger DN. The cellular and molec-ular basis of gastric mucosal defense. FASEB J 1996;10:731–40.
23. Kaye GI, Pascal RR, Lane N. The colonic pericryptalfibroblast sheath: replication, migration, and cytodiffer-entiation of a mesenchymal cell system in adult tissue. 3.Replication and differentiation in human hyperplasticand adenomatous polyps. Gastroenterology 1971;60:515–36.
24. Joyce NC, Haire MF, Palade GE. Morphologic andbiochemical evidence for a contractile cell networkwithin the rat intestinal mucosa. Gastroenterology 1987;92:68–81.
25. Bradbury J. A two-pronged approach to the clinicaluse of HGF. Lancet 1998;351:272.
26. Miller SM, Farrugia G, Schmalz PF, et al. Hemeoxygenase 2 is present in interstitial cell networks ofthe mouse small intestine. Gastroenterology 1998;114:239–44.
27. Berndt A, Kosmehl H, Mandel U, et al. TGF h andbFGF synthesis and localization in Dupuytren’s disease(nodular palmar fibromatosis) relative to cellularactivity, myofibroblast phenotype and oncofetal variantsof fibronectin. Histochem J 1995;27:1014–20.
28. Bostrom H, Willetts K, Pekny M, et al. PDGF-Asignaling is a critical event in lung alveolar myofibroblastdevelopment and alveogenesis. Cell 1996;85:863–73.
29. Border WA, Noble NA. Transforming growth factor hin tissue fibrosis. N Engl J Med 1994;331:1286–92.
30. Pfeilschifter J. Mesangial cells orchestrate inflamma-tion in the renal glomerulus. News Physiol Sci 1994;9:271–6.
31. Klimpel GR, Langley KE, Wypych J, et al. A role forstem cell factor (SCF): c-kit interaction(s) in theintestinal tract response to Salmonella typhimuriuminfection. J Exp Med 1996;184:271–6.
32. Adegboyega PA, Mifflin RC, DiMari JF, et al.Immunohistochemical study of myofibroblasts in nor-mal colonic mucosa, hyperplastic polyps, and adeno-matous colorectal polyps. Arch Pathol Lab Med 2002;126:829–36.
33. Valentich JD, Popov V, Saada JI, Powell DW.Phenotypic characterization of an intestinal subepi-thelial myofibroblast cell line. Am J Physiol 1997;272:C1513–24.
34. Mifflin RC, Saada JI, Di Mari JF, et al. Regulation ofCOX-2 expression in human intestinal myofibroblasts:mechanisms of IL-1-mediated induction. Am J PhysiolCell Physiol 2002;282:C824–34.
35. Hinterleitner TA, Saada JI, Berschneider HM, et al.IL-1 stimulates intestinal myofibroblast COX geneexpression and augments activation of Cl- secretion inT84 cells. Am J Physiol 1996;271:C1262–8.
36. Mahida YR, Galvin AM, Gray T, et al. Migrationof human intestinal lamina propria lymphocytes,macrophages and eosinophils following the loss ofsurface epithelial cells. Clin Exp Immunol 1997;109:377–86.
37. Mahida YR, Beltinger J, Makh S, et al. Adult humancolonic subepithelial myofibroblasts express extracellu-lar matrix proteins and cyclooxygenase-1 and -2. Am JPhysiol 1997;273:G1341–8.
39. Shao J, Lee SB, Guo H, et al. Prostaglandin E2stimulates the growth of colon cancer cells via inductionof amphiregulin. Cancer Res 2003;63:5218–23.
40. Furuya S, Furuya K. Characteristics of culturedsubepithelial fibroblasts of rat duodenal villi. AnatEmbryol (Berl) 1993;187:529–38.
44. Damstrup L, Kuwada SK, Dempsey PJ, et al.Amphiregulin acts as an autocrine growth factor intwo human polarizing colon cancer lines that exhibitdomain selective EGF receptor mitogenesis. Br J Cancer1999;80:1012–9.
www.aacrjournals.org 855 Cancer Res 2006; 66: (2). January 15, 2006
Phosphatidylinositol 3-kinase mediates proliferativesignals in intestinal epithelial cells. Gut 2003;52:1472–8.
46. Olaso E, Salado C, Egilegor E, et al. Proangiogenicrole of tumor-activated hepatic stellate cells inexperimental melanoma metastasis. Hepatology 2003;37:674–85.
47. Williams CS, Mann M, DuBois RN. The role ofcyclooxygenases in inflammation, cancer, and develop-ment. Oncogene 1999;18:7908–16.
48. Eberhart CE, Coffey RJ, Radhika A, et al. Up-regulation of cyclooxygenase 2 gene expression inhuman colorectal adenomas and adenocarcinomas.Gastroenterology 1994;107:1183–8.
49. Kargman SL, O’Neill GP, Vickers PJ, et al. Expressionof prostaglandin G/H synthase-1 and -2 protein inhuman colon cancer. Cancer Res 1995;55:2556–9.
50. Chapple KS, Cartwright EJ, Hawcroft G, et al.Localization of cyclooxygenase-2 in human sporadiccolorectal adenomas. Am J Pathol 2000;156:545–53.
52. Ribardo DA, Crowe SE, Kuhl KR, et al. Prostaglan-din levels in stimulated macrophages are controlledby phospholipase A2-activating protein and by acti-vation of phospholipase C and D. J Biol Chem 2001;276:5467–75.
53. Berasain C, Garcia-Trevijano ER, Castillo J, et al.Amphiregulin: an early trigger of liver regeneration inmice. Gastroenterology 2005;128:424–32.
54. Ciardiello F, Kim N, Saeki T, et al. Differentialexpression of epidermal growth factor-related proteinsin human colorectal tumors. Proc Natl Acad Sci U S A1991;88:7792–6.
55. De Luca A, Arra C, D’Antonio A, et al. Simultaneousblockage of different EGF-like growth factors results inefficient growth inhibition of human colon carcinomaxenografts. Oncogene 2000;19:5863–71.
56. Shao J, Evers BM, Sheng H. Prostaglandin E2synergistically enhances receptor tyrosine kinase-dependent signaling system in colon cancer cells. J BiolChem 2004;279:14287–93.
57. Chang SH, Ai Y, Breyer RM, et al. The prostaglandinE2 receptor EP2 is required for cyclooxygenase 2-
mediated mammary hyperplasia. Cancer Res 2005;65:4496–9.
58. Moraitis D, Du B, De Lorenzo MS. Levels ofcyclooxygenase-2 are increased in the oral mucosa ofsmokers: evidence for the role of epidermal growthfactor receptor and its ligands. Cancer Res 2005;65:664–70.
59. Matsumoto K, Nakamura T. Hepatocyte growthfactor: molecular structure, roles in liver regeneration,and other biological functions. Crit Rev Oncog 1992;3:27–54.
60. Di Renzo MF, Narsimhan RP, Olivero M, et al.M. Expression of the Met/HGF receptor in normal andneoplastic human tissues. Oncogene 1991;6:1997–2003.
61. Halttunen T, Marttinen A, Rantala I, et al. Fibroblastsand transforming growth factor h induce organizationand differentiation of T84 human epithelial cells.Gastroenterology 1996;111:1252–62.
62. Kiriyama M, Ushikubi F, Kobayashi T, et al. Ligandbinding specificities of the eight types and subtypes ofthe mouse prostanoid receptors expressed in Chinesehamster ovary cells. Br J Pharmacol 1997;122:217–24.
![RoleofPGE inAsthmaandNonasthmatic EosinophilicBronchitis2) by COXs, and metabolism of prostaglandin H 2 to prostaglandin E 2 via prostaglandin E synthase [12]. There are three enzymes](https://static.documents.pub/doc/80x56/60d522031e41432a8f254505/roleofpge-inasthmaandnonasthmatic-eosinophilicbronchitis-2-by-coxs-and-metabolism.jpg)
