Measurement of Fecal Calprotectin Improves Monitoring and Detection of Recurrence of Crohn's Disease Following Surgery

ORIGINAL ARTICLE

Prostaglandin E2 and Polyenylphosphatidylcholine ProtectAgainst Intestinal Fibrosis and Regulate Myofibroblast Function

Angela C. Baird • Frances Lloyd • Ian C. Lawrance

Received: 18 November 2014 / Accepted: 19 January 2015

� Springer Science+Business Media New York 2015

Abstract

Background Intestinal fibrosis is a serious and often recurrent

complication of inflammatory bowel disease despite surgical

intervention. The anti-fibrotic potential of prostaglandin E2

(PGE2) and polyenylphosphatidylcholine (PC) was investi-

gated using the murine model of 2,4,6-trinitrobenzene sulfonic

acid (TNBS)-induced chronic intestinal inflammation and

fibrosis, and murine and human intestinal myofibroblasts.

Methods Mice were treated with TNBS enemas weekly for

2 or 6 weeks ± PGE2 (10 mg/kg/day orally) or PC (200 mg/

kg/day orally). Inflammation and fibrosis were histologically

assessed and scored. Pro-inflammatory cytokines, TLR4, and

ECM-related gene expression from the colonic tissue and

cultured myofibroblasts were assessed by RT-qPCR. The

levels ofa-SMA? staining and endogenous PGE2 in vivo were

also assessed.

Results Both PGE2 and PC treatment significantly

decreased TNBS-induced intestinal inflammation and excess

collagen deposition in vivo. This was accompanied by

decreased a-SMA? staining in the lamina propria and lower

collagen type I (COL1a1) expression. Endogenous PGE2

levels demonstrated that PC was not being converted into

PGE2, thus mediating its effects primarily via PGE2-

independent pathways. Both PGE2 and the PC isoform, 1,2-

dilinoleoylphosphatidylcholine (DLPC), regulated primary

mouse myofibroblast and CCD-18co COL1a1 production,

and induced lower collagen type I to III and TGF-b1 to TGF-

b3 ratios, demonstrating their ability to induced normal

healing in the presence of phorbol 12-myristate 13-acetate

(protein kinase C-dependent inducer of collagen production).

Conclusion PGE2 and PC both have potent anti-fibrogenic

potentials in their ability to regulate inflammatory cell and

myofibroblast accumulation within inflamed tissue, to

decrease pro-inflammatory cytokine expression and to

maintain normal healing in an inflammatory environment.

Keywords Prostaglandin E2 (PGE2) �Polyenylphosphatidylcholine (PC) � Intestinal fibrosis �Inflammatory bowel disease (IBD) � Extracellular matrix

(ECM) � Myofibroblasts

Introduction

Intestinal fibrosis is a common and serious complication of

inflammatory bowel disease (IBD) [1], which is a conse-

quence of chronic inflammation that leads to excessive

extracellular matrix (ECM) [2] deposition. In Crohn’s disease

(CD), most patients initially present with uncomplicated

inflammatory disease, but within 10 years, more than 70 %

will develop stricturing or perforating complications [3] and

more than a third will develop intestinal narrowing and

obstruction, which frequently results in surgical intervention

[4]. Surgery, however, does not prevent recurrence of

inflammation or fibrosis. Irrespective of the remarkable pro-

gress in the medical management of IBD, limited work has

been carried out on anti-fibrotic therapies despite the need for

therapeutic options. The ability to regulate ECM homeostasis

Electronic supplementary material The online version of thisarticle (doi:10.1007/s10620-015-3552-9) contains supplementarymaterial, which is available to authorized users.

A. C. Baird (&) � F. Lloyd � I. C. Lawrance

Fremantle Unit, School of Medicine and Pharmacology, Level 6,

T Block, Fremantle Hospital, University of Western Australia,

Alma Street, Fremantle, WA 6010, Australia

e-mail: [email protected]

A. C. Baird � F. Lloyd � I. C. Lawrance

Centre of Inflammatory Bowel Diseases, Fremantle Hospital,

Fremantle, WA, Australia

123

Dig Dis Sci

DOI 10.1007/s10620-015-3552-9

by controlling collagen production and deposition by intesti-

nal myofibroblasts may be one such option.

Intestinal myofibroblasts are the main effector cells of

gastrointestinal fibrosis that play a critical role in tissue

repair and scar formation [5]. They secrete the major

parenchymal collagens, type I and type III. Depending on

the physiological and pathological state of the tissue, the

absolute amount and relative proportions of these two types

of collagens vary greatly. Defects in collagen metabolism

can affect fibril formation (size and degree of cross-link-

age) and can disrupt tissue architecture [6].

Maintaining an appropriate balance between collagen

deposition/degradation is imperative for ECM homeostasis;

therefore, stringent regulation of matrix metalloproteinases

(MMPs), tissue inhibitor metalloproteinases (TIMPs),

growth factors, and pro-inflammatory cytokines is para-

mount. Two lipid family members that may play an impor-

tant role in the regulation of these factors are prostaglandin

E2 (PGE2) and polyenylphosphatidylcholine (PC).

PGE2 is an anti-fibrotic lipid mediator, derived from

arachidonic acid metabolism by cyclooxygenase (COX).

Deficiencies in PGE2 production [7–9] and responsiveness

[10, 11] have been reported in IBD [12] as well as in

human idiopathic pulmonary fibrosis and animal models of

fibrosis. It inhibits fibroblast proliferation, migration, and

differentiation into myofibroblasts, collagen synthesis, and

induces lung fibroblast apoptosis via the activation of the E

prostanoid 2 and 4 receptors [8, 13].

PCs are phospholipids that belong to the lipid family of

biological polymers and are the major component of bio-

logical membranes. PC is thought to play an important

anti-inflammatory role, as IBD patients have significantly

lower levels of PC in their intestinal mucosa compared to

healthy controls [14]. Slow-release oral PC supplementa-

tion can also reduce disease activity in 90 % of IBD

patients, with clinical remission achieved in 50 % [15–17].

Disease activity reduction may be attributable to PC’s

ability to influence mucosal signalling to decrease pro-

inflammatory cytokine production [15].

This study investigated the proficiency of PGE2 and PC

to regulate ECM homeostasis in vivo, using the TNBS-

induced murine model of chronic inflammation and fibrosis

and their ability to regulate murine and human intestinal

myofibroblast function in order to maintain ECM homeo-

stasis and inhibit their fibrogenic potential in vitro.

Methods and Materials

Animals

Twelve-week-old female C57bl/6 9 129Svj (Bl/6;129)

inbred mice were housed and fed standard mouse chow and

tap water ad libitum following protocols approved by the

Animal Ethics Committee, University of Western

Australia.

Mouse Model

The 2,4,6-trinitrobenzene sulfonic acid (TNBS) murine

model of chronic intestinal inflammation and fibrosis was

used as previously described [18]. All mice were examined

daily for signs of colitis and systemic inflammation. The

mice were randomized into (1) controls, 0.1 ml water/sal-

ine (n = 5); (2) TNBS alone (n = 5); (3) TNBS/PGE2

(n = 5); (4) TNBS/PC (n = 5); (5)TNBS/Indo (n = 5); (6)

TNBS/Indo/PGE2 (n = 5). PGE2 (Sigma-Aldrich Pty Ltd,

NSW, Australia) was reconstituted in absolute ethanol at a

concentration of 500 mg/ml and used to supplement the

drinking water at 10 mg/kg per day. PC (Spectrum

Chemicals, CA, USA) was added to standard AIN93G

mouse chow (Specialty Feeds, WA, Australia) at 3 mg/g to

make the custom AIN93G-042 chow and fed at a rate of

2 g of chow per mouse, per day to give a final concentra-

tion of 200 mg/kg per day for each mouse. Indomethacin

(Sigma-Aldrich) was suspended in absolute ethanol at a

concentration of 20 mg/ml and used to supplement the

drinking water at 0.2 mg/kg per day. The water with PGE2

and Indo was changed on a daily basis. Additional groups

of animals treated with PGE2 (n = 5) and PC (n = 5)

alone were also included into the study as experimental

controls.

Tissue Processing

After 2 and 6 weeks of treatment, the colons were removed

intact from the anus to the ileocecal junction, cleaned,

weighed, and the length was measured. At least three

transverse sections of the colons were taken from the distal

5 cm of the involved colon. The middle section was for-

malin-fixed for histological examination, and the sections

from either side were immediately frozen in liquid nitrogen

for mRNA isolation.

Scoring of Histological Changes and a-SMA Staining

Serial sections of mouse colon were stained with hema-

toxylin and eosin (H&E), Masson’s Trichrome (TC), and

Sirius red with fast green counterstain to assess inflam-

mation and fibrosis. Each cross section was divided into 3

regions, blindly assessed by an independent pathologist

(CHF), and scored semi-quantitatively. The degree of

inflammation was scored from 0 to 3 (0, no inflammation;

1, very low level; 2, low level of leukocyte infiltration; and

3, high level of leukocyte infiltration, loss of goblet cells,

thickening of the colon wall and high vascular density [19,

Dig Dis Sci

123

20]. Depending on the density and extent of TC positive

connective tissue staining and disruption of tissue archi-

tecture compared to untreated control mice, fibrosis was

also scored from 0 to 3 (0, absent; 1, mild; 2, moderate; and

3, severe) [18]. The sums of the 3 regions from each animal

were averaged with the sum of the other animals to give a

score out of 9. Error bars represent mean ± SD.

Serial sections were also stained with anti-a-smooth

muscle actin (a-SMA) (1:2,000; Sigma-Aldrich) antibody,

and the number of positive pixels per unit area was quan-

titated using ImageScope, version 11.2 (Aperio Technolo-

gies, Inc. USA).

PGE2 Immunoassay

Colonic tissue from each mouse was homogenized in

0.1 M PO4 buffer (pH 7.4) containing 1 mM EDTA and

10 lM indomethacin, freeze/thawed twice, supernatant

collected, and assayed using the PGE2 EIA Kit (Enzo Life

Sciences International Incorp, PA, USA) as per manufac-

turer’s instructions.

Myofibroblast Isolation and Culture

Primary colonic myofibroblasts (pMFBs) were derived by

explant outgrowth in culture from wild-type Bl/6;129 mice

as previously described [21]. Cells were split 1:4 when

70–80 % confluent and used between passages 3 and 4.

Cells were routinely stained with anti-a-SMA IgG2a

(1:2,000), anti-vimentin IgG2a (1:25; Novocastra, Leica

Biosystems Pty Ltd, NSW, Australia), and anti-desmin

IgG1 (1:100, Novocastra Leica Biosystems) to confirm that

the cells were myofibroblasts (Supp Fig. 1).

CCD-18co cells (CRL 1459) at passage 9 were obtained

from the American Type Culture Collection (Rockville,

MD, USA) and used between passages 11 and 12. Cells

were routinely split 1:3 at 70–80 % confluence and main-

tained in DMEM (GIBCO, Life Technologies Australia,

Pty, Ltd, VIC, Australia), supplemented with 10 % fetal

bovine serum (GIBCO, Life Technologies Australia),

200 mM L-glutamine (GIBCO, Life Technologies Austra-

lia), and 1.1 % 100 9 Antibiotic–Antimycotic (GIBCO,

Life Technologies Australia), under 5 % CO2 at 37 �C.

In vivo, the PC used was a mixture of polyunsaturated

phosphatidylcholines extracted from soy beans. DLPC,

which constitutes 43–50 % of the PC extract, was used to

investigate the specific effects of PC on myofibroblast

function in vitro. Cells were pre-treated with phorbol

12-myristate 13-acetate (PMA) (Sigma-Aldrich Pty, Ltd)

for 24 h at 10 ng/ll to induce protein kinase C (PKC)-

dependent activation of NFjB and subsequent collagen

production, prior to the addition of PGE2 (50 lM), DLPC

(50 lM) and Indo (50 lM) alone or in combination with

PGE2 for a further 24 h.

RNA Isolation and RT-qPCR

Total RNA was isolated using Tri Reagent� (Ambion, Life

Technologies Australia, VIC, Australia) for cell cultures

and RNeasy Mini Kits (QIAGEN, Pty Ltd, VIC, Australia)

for colonic tissue according to manufacturer’s instructions.

IL-1b, IL-6, TNFa, TLR4, TGF-b1, TGF-b3, COL1a1,

COL3a1, TIMP-1, Mm MMP-12, Hs MMP-1, and COX-2

mRNA expression was assessed by RT-qPCR using the

RotorGene Q 5-plex (QIAGEN). Gene expression was

normalized to the internal control, GAPDH, and expressed

as a fold change relative to control group (Table 1).

MTT Cell Proliferation Assays

pMFBs and CCD-18co cells were cultured in 24-well

dishes, maintained, and treated as previously described for

24 h and 6 days. Following treatment, 5 mg/ml of MTT

(thiazolyl blue tetrazolium bromide) was added to each

well and incubated for 3.5 h at 37 �C. The culture medium

containing MTT was removed, and the cells were lysed

with MTT solvent (4 mM HCl, 0.1 % Igepal CA 630 in

isopropanol). Absorbance was read at 590 nM on a FLU-

Ostar Optima (BMG Labtech Pty Ltd, VIC, Australia).

Statistical Analysis

All results were expressed as mean ± standard error of the

mean (SEM). Comparisons between the different treatment

groups were analyzed by Mann–Whitney nonparametric

t tests using the Graphpad Prism 4.0 software package for

Windows PC (Graphpad, San Diego, CA, USA). For all

analyses, p \ 0.05 was considered to be statistically

significant.

Results

Macroscopic and Microscopic Changes

At 2 weeks, inflammation was markedly increased in

TNBS-treated animals (Fig. 1ad) compared to controls

(Fig. 1aa), with increased infiltration of inflammatory cells

(long arrow) and architectural disruption down to the

muscularis mucosae (short arrow; Fig. 1ad). Excess col-

lagen deposition was also noted in these animals (arrow-

head; Fig. 1ae–af) compared to controls (Fig. 1ab–ac).

Treatment with TNBS/PGE2 (Fig. 1ag) and TNBS/PC

(Fig. 1aj) protected against the TNBS-induced changes, as

demonstrated by the preservation of colonic architecture,

Dig Dis Sci

123

absence or low levels of inflammatory cell infiltrate, and

the lack of excess collagen deposition (Fig. 1ah–ai, ak–al,

respectively).

The TNBS-induced changes in inflammation and fibro-

sis corresponded to significantly higher colon weight/

length ratios, a measurement indicative of colonic edema

and/or hyperplasia, compared to controls (p \ 0.0001;

Fig. 1b) and both TNBS/PGE2- and TNBS/PC-treated

animals (p \ 0.0001 for both) at 2 and 6 weeks. This

together with the macroscopic and microscopic changes

clearly demonstrates the protective effects of PGE2 and PC

against TNBS-induced inflammation and fibrosis.

To investigate the effects of oral PGE2 on endogenous

PGE2 levels in the colonic tissue and to determine whether

PC was being converted to PGE2, PGE2 immunoassays

were performed on control, TNBS-, TNBS/PGE2-, and

TNBS/PC-treated colonic tissue. TNBS/PGE2-treated

animals had significantly higher PGE2 levels compared to

controls, TNBS alone, and TNBS/PC (p \ 0.0001 for all;

Supp Fig. 2). Treatment with TNBS/PC induced a small

but nonsignificant increase in PGE2, suggesting that the

majority of PC ingested was not being converted into PGE2

and that its anti-inflammatory and anti-fibrogenic effects

were primarily mediated via PGE2-independent pathways.

Colonic Inflammation and Fibrosis Scoring and a-SMA

Staining

At 2 and 6 weeks, the control, PGE2, and PC alone groups

did not demonstrate any increase in either inflammation or

fibrosis scoring (Fig. 2aa–ab), indicating that oral supple-

mentation of PGE2 or PC alone neither induce mucosal

injury and inflammation nor alter ECM deposition.

Table 1 Primer sequences used for measuring expression levels of key ECM-related proteins by RT-qPCR

Target gene Primer sequence Annealing temp (�C)

Mm GAPDH F: 50-GTG GTG GAC CTC ATG GCC TAC-30

R: 50-CGA GTT GGG ATA GGG CCT CTC-3058

Hs GAPDH F: 50-TGC CCC CTC TGC TGA TGC C-30

R: 50-CCT CCG ACG CCT GCT TCA CCA C-3059

Hs/Mm Collagen Type I a1 F: 50-GGC GGC CAG GGC TCC GAC CC-30

R: 50-AAT TCC TGG TCT GGG GCA CC-3059

Hs/Mm Collagen Type III a1 F: 50-CCC AGA ACA TYA CAT AYC AC-30

R: 50-GAT TAR AAC AAG AKG AAC ACA-3058

Hs MMP-1 F: 50-GGT GAT GAA GCA GCC CAG-30

R: 50-CAG TAG AAT GGG AGA GTC-3060

Mm MMP-13 (Mm MMP-1) F: 50-TGAACATCCATCCCGTGACC-30

R: 5-GGCATGACTCTCACAATGCG-3053

Hs/Mm TIMP-1 F: 50-ART CAA CSA GAC CAC CTT ATA CCA-30

R: 50-ASC TGR TCC GTC CAC AAR CA-3056

Hs/Mm TGF-b1 F: 50-GCC CTG GAC ACC AAC TAT TGC-30

R: 50-AGC TGC ACT TGC AGG AGC GCA C-3061

Hs/Mm TGF-b3 F: 50-GCT CTT CCA GAT MCT TCG RC-30

R: 50-AGC AGT TCT CCT CCA RGT TG-3055

Hs COX-2a F: 50-GTT CCA CCC GCA GTA CAG-30

R: 50-GGA GCG GGA AGA ACT TGC-3051

Mm COX-2 F: 50-AAAACCGTGGGGAATGTATGAGCAC-30

R: 50-AAACTTCGCAGGAGGGGGATGTTG-3062

Mm IL-1b F: 50-GTGGCTGTGGAGAAGCTGTG-30

R: 50-GAAGGTCCACGGGAAAGACAC-3058

Mm IL-6 F: 50-GTATGAACAACGATGATGCACTTG-30

R: 50-ATGGTACTCCAGAAGACCAGAGGA-3058

Mm TNF F: 50-CTGTAGCCCACGTCGTAGC-30

R: 50-TTGAGATCCATGCCGTTG-3060

Mm TLR4 F: 50-AGGAAGTTTCTCTGGACTAACAAGTTAGA-30

R: 50-AAATTGTGAGCCACATTGAGTTTC-3058

Dig Dis Sci

123

TNBS-treated mice had significantly higher inflamma-

tion scores compared to control animals at both 2 and

6 weeks (p \ 0.0001 for both; Fig. 2aa). Treatment with

both TNBS/PGE2 and TNBS/PC resulted in significantly

lower inflammation scores compared to TNBS alone

(p \ 0.001 for both at 2 and 6 weeks) and were not dif-

ferent to controls, PGE2, or PC alone (Fig. 2aa). Similarly,

TNBS-treated mice had significantly higher levels of

fibrosis at both 2 and 6 weeks compared to controls

(p \ 0.0001), TNBS/PGE2 (p \ 0.0001 for 2 weeks, and

p \ 0.001 for 6 weeks) and TNBS/PC (p \ 0.0001 for

both 2 and 6 weeks; Fig. 2ab). Again, the fibrosis scores of

TNBS/PGE2 and TNBS/PC did not differ from controls,

PGE2, or PC treatment alone.

To determine whether the differences in inflammation

and fibrosis scoring correlated with the number of

localized myofibroblasts, the tissues were stained with

anti-a-SMA antibody and analyzed using a positive pixel

Fig. 1 aa–al Representative

images of macroscopic and

microscopic changes after

2 weeks of TNBS, TNBS/PGE2,

and TNBS/PC treatment and in

untreated controls. Serial

sections of mouse colon were

stained with hematoxylin and

eosin (H&E), Masson’s

Trichrome (TC), and Sirius red

with fast green counterstain to

assess the degree of

inflammation and fibrosis.

Influx of inflammatory infiltrate

(long arrows), disruption of

lamina propria (short arrows),

and collagen deposition

(arrowhead). Scale bar 100 lm,

9200 magnification. b Mouse

colon weight-to-length ratio.

The colon weight-to-length

ratios were measured after 2 and

6 weeks of TNBS, TNBS/PGE2,

and TNBS/PC treatment and

normalized to control (n = 5).

Results were expressed as

weight/length (g/cm) with

mean ± SD. *p \ 0.05 versus

control. a p \ 0.05 versus

TNBS alone

Dig Dis Sci

123

algorithm. TNBS-treated animals (Fig. 2bb) had more

positive pixels (arrows) per unit area compared to con-

trols (p \ 0.001; Fig. 2ba), while TNBS/PGE2- and

TNBS/PC-treated animals had significantly fewer posi-

tive pixels per unit area compared to TNBS alone

(p \ 0.0001 for both; Fig. 2bc–bd), suggesting that PGE2

and PC both may have an effect on myofibroblast

migration, differentiation, or proliferation within the

colonic tissue. No differences between controls, TNBS/

PGE2, or TNBS/PC were observed.

The Effects of PGE2 and PC on Pro-inflammatory

Cytokine and TLR4 Expression

In the setting of chronic inflammation, prolonged secretion

of pro-inflammatory cytokines may induce tissue damage,

aberrant wound healing, and fibrosis. IL-1b, IL-6, and

TNFa are key pro-inflammatory cytokines that are secreted

early in the inflammatory response that can increase the

number of collagen producing cells, as well as the level of

collagen production and secretion. It was therefore of

Fig. 2 a Scoring of histological

changes. The degree of

(aa) inflammation and

(ab) fibrosis was semi-

quantitated by dividing each

cross section into three regions

and scoring each 0–3. The sum

of each tissue section was

averaged with the sum of the

other animals to give a total

score out of 9. n = 5 animals

per treatment, and results were

expressed as mean ± SD.

*p \ 0.0001 compared to

control. a p \ 0.001 compared

to TNBS alone. b a-SMA?

staining in the colon. Colonic

sections taken from (ba) control

and (bb) TNBS-, (bc) TNBS/

PGE2-, and (bd) TNBS/PC-

treated mice were stained with

anti-a-SMA antibody, and the

number of positive pixels (black

arrows) per unit area was

quantitated. (be) Results were

expressed as mean ± SD.

n = 5 animals per treatment

group. *p \ 0.001 versus

control. a p \ 0.001 versus

TNBS alone

Dig Dis Sci

123

interest to examine the effects of PGE2 and PC on the

levels of IL-1b, IL-6, and TNFa gene expression within the

colonic tissue.

IL-1b, IL-6, and TNFa mRNA expression was analyzed

from whole colonic tissue isolated from controls and

TNBS-, TNBS/PGE2-, and TNBS/PC-treated animals. At

2 weeks, TNBS/PGE2- and TNBS/PC-treated animals had

significantly lower levels of colonic IL-1b compared to

TNBS alone (p \ 0.0001 for both; Fig. 3a), but were not

different from controls. This trend continued through to

6 weeks for TNBS/PGE2 compared to TNBS (p \ 0.0001;

Fig. 3b). Treatment with TNBS/PC also resulted in lower

levels of IL-1b compared to TNBS alone (p \ 0.01);

however, the levels were significantly higher compared to

controls (p \ 0.05; Fig. 3b), suggesting that its ability to

regulated TNBS-induced IL-1b production was limited by

6 weeks. IL-6 expression was significantly lower in both

TNBS/PGE2- and TNBS/PC-treated animals compared to

TNBS alone at both 2 (p \ 0.0001; Fig. 3c) and 6

(p \ 0.0001; Fig. 3d) weeks, with no difference compared

to controls. Significantly lower levels of TNFa expression

were also noted in TNBS/PGE2- and TNBS/PC-treated

animals compared to TNBS alone at both time points

(p \ 0.0001 for all; Fig. 3e, f). These data demonstrate that

oral supplementation of PGE2 and PC both lead to a

decrease in TNBS-induced pro-inflammatory cytokine

expression.

Considering their ability to regulate pro-inflammatory

cytokine expression, it was of interest to investigate whe-

ther PGE2 and PC could regulate TLR4 expression, as it is

one of the key receptors responsible for pro-inflammatory

cytokine production upon mitogen recognition. Animals

treated with TNBS alone had significantly higher TLR4

expression compared to controls at 2 (p \ 0.001; Fig. 3g)

and 6 (p \ 0.0001; Fig. 3h) weeks. In contrast, TLR4

expression in TNBS/PGE2- and TNBS/PC-treated animals

were not different compared to controls but were, however,

significantly lower compared to TNBS alone at 2 and

6 weeks (p \ 0.01 for all; Figs. 3g, 3h).

The Effects of PGE2 on Colonic ECM-Related Gene

Expression

To determine whether PGE2 and PC had an effect on the

levels of collagen production within the tissue, the

expression of four key ECM-related genes, TGF-b1, col-

lagen type I (COL1a1), MMP-13, TIMP-1, and the

inducible cyclooxygenase, COX-2, was measured. At

2 weeks, a significant decrease was observed in TGF-b1,

COL1a1, and TIMP-1 expression in TNBS/PGE2

(p \ 0.01; Fig. 4a, p \ 0.001; Fig. 4c, p \ 0.001; Fig. 4e,

respectively) and TNBS/PC (p \ 0.05; Fig. 4a, p \ 0.05;

Fig. 4c, p \ 0.001; Fig. 4e, respectively)-treated animals

compared to TNBS alone. Although TGF-b1, COL1a1,

and TIMP-1 expression was numerically lower in TNBS/

PGE2- and TNBS/PC-treated animals compared to TNBS

alone at 6 weeks, the decreases were not significant

(Fig. 4b, d, f). MMP-13 (murine equivalent of human

MMP-1) expression was not significantly different between

any of the groups with the exception of TNBS/PGE2 at

2 weeks (p \ 0.05; Fig. 4g) compared to TNBS alone, and

TNBS compared to controls (p \ 0.05; Fig. 4h) at

6 weeks. COX-2 expression was not different between any

of the groups at any time (Fig. 4i, j). These data demon-

strated that PGE2 and PC are capable of significantly

inhibiting TNBS-induced production of COL1a1 during

acute inflammation (2 weeks), possibly via inhibition of

TGF-b1 production, and are taking on a more regulatory

role by 6 weeks as opposed to inhibiting production alto-

gether in vivo.

ECM-Related Gene Expression by Myofibroblasts

In Vitro

The effects of PGE2 and PC specifically on the ECM-

producing cells, myofibroblasts, were next investigated.

For this isolated pMFBs and the non-transformed human

intestinal myofibroblast cell line CCD-18co, were treated

with PMA alone, PMA/PGE2, or with the major PC iso-

type, DLPC (PMA/DLPC), and analyzed for TGF-b1,

COL1a1, TIMP-1, MMP-13/MMP-1, and COX-2 mRNA

expression.

In pMFBs, PMA treatment resulted in significantly

higher TGF-b1, COL1a1, TIMP-1, and MMP-13 expres-

sion compared to untreated controls (p \ 0.01 for all;

Fig. 5a–i). PMA/PGE2 treatment resulted in numerically

lower levels of TGF-b1 (Fig. 5a) and significantly lower

levels of COL1a1 compared to PMA alone (p \ 0.001;

Fig. 5c). There were no differences in TIMP-1 levels

compared to PMA alone (Fig. 5e), but was found to be

significantly higher compared to controls (p \ 0.0001).

MMP-13 expression was significantly higher in PMA/

PGE2-treated pMFBs compared to both PMA alone and

controls (p \ 0.0001 for both; Fig. 5g), as was the

expression of COX-2 (p \ 0.001 compared to PMA alone

and untreated controls; Fig. 5i).

Treatment with PMA/DLPC resulted in significantly

lower TGF-b1, COL1a1, and TIMP-1 expression com-

pared to PMA alone (p \ 0.0001 for both; Fig. 5a, c,

p \ 0.001; Fig. 5e). There were no differences in MMP-13

or COX-2 expression between PMA/PC- and PMA-treated

pMFBs (Fig. 5g, i).

PMA/PGE2 treatment had a similar effect on CCD-18co

cells, with significantly lower COL1a1 expression

(p \ 0.0001; Fig. 5d) and significantly higher MMP-1

(p \ 0.0001; Fig. 5h) and COX-2 (p \ 0.01; Fig. 5j)

Dig Dis Sci

123

expression compared to PMA alone. TGF-b1 and TIMP-1

expression was not different compared to PMA treatment

alone (Fig. 5b, f). Although PMA/PC treatment also resulted

in significantly lower levels of COL1a1 compared to PMA

alone (p \ 0.001; Fig. 5d), TGF-b1 expression was not dif-

ferent (Fig. 5b), while TIMP-1 and MMP-1 levels were sig-

nificantly higher (p \ 0.05; Fig. 5f, p \ 0.001; Fig. 5h). The

findings demonstrate that PC induces ECM homeostasis via

different pathways in murine and human myofibroblasts.

To determine whether PGE2 and PC had any effect on

cell proliferation, MTT assays were performed on pMFB

and CCD-18co cells treated with PMA, PMA/PGE2 and

PMA/DLPC. In pMFBs, PMA/PGE2 and PMA/DLPC

treatment inhibited PMA-induced cell proliferation after

24 h (p \ 0.05; Supp Fig. 3a) and 6 days (p \ 0.0001;

Supp Fig. 3c). In CCD-18co cells, PMA/PGE2- and PMA/

DLPC-treated cultures had numerically fewer viable cells

compared to PMA alone at 24 h (Supp Fig. 3b) but were

Fig. 3 Colonic pro-

inflammatory cytokine and

TLR4 expression. Colonic RNA

isolated from controls and

TNBS-, TNBS/PGE2-, and

TNBS/PC-treated mice at 2 and

6 weeks was analyzed for (a,

b) IL-1b, (c, d) IL-6, (e,

f) TNFa, and (g, h) TLR4

mRNA expression normalized

to the housekeeping gene,

GAPDH. Results were

expressed as mean ± SD and

normalized back to control.

n = 3–5 animals per treatment

group for both 2 and 6 weeks.

*p \ 0.05 versus control.

a p \ 0.01 versus TNBS alone

Dig Dis Sci

123

Fig. 4 Colonic gene expression

of ECM-related proteins. RNA

isolated from controls and

TNBS-, TNBS/PGE2-, and

TNBS/PC-treated mice at 2 and

6 weeks was analyzed for (a,

b) TGF-b1, (c, d) COL1a1, (e,

f) TIMP-1, (g, h) MMP-13, and

(i, j) COX-2 mRNA expression

normalized to the housekeeping

gene, GAPDH. Results were

expressed as mean ± SD and

normalized back to control.

n = 3–5 animals per treatment

group for both 2 and 6 weeks.

*p \ 0.05 versus control.

a p \ 0.05 versus TNBS alone

Dig Dis Sci

123

Fig. 5 Myofibroblast gene

expression of ECM-related

proteins. RNA isolated from

untreated controls and PMA-,

PMA/PGE2-, and PMA/DLPC-

treated pMFBs and CCD-18co

cells was analyzed for (a,

b) TGF-b1, (c, d) COL1a1, (e,

f) TIMP-1, (g, h) MMP-13/

MMP-1, and (i, j) COX-2

mRNA expression normalized

to the housekeeping gene,

GAPDH. Results were

expressed as mean ± SD and

normalized back to controls.

n = 3 replicated per treatment/

experiment, and each

experiment was performed in

duplicate. *p \ 0.01 versus

control. a p \ 0.05 versus PMA

alone

Dig Dis Sci

123

not significant. By 6 days, treatment with PMA/PGE2 and

PMA/DLPC significantly inhibited PMA-induced cell

proliferation (p \ 0.0001 for both; Supp Fig. 3d). PMA/

DLPC-treated cultures also had significantly fewer viable

cells than the untreated controls (p \ 0.001).

The Effect of PGE2 and PC on Collagen and TGF-bRatios

The balance between the different collagen and TGF-bsubtypes during tissue repair is imperative for normal

healing. Dysregulation of this fine balance may lead to

fibrosis. To investigate the effects of PGE2 and PC on the

COL1a1 to COL3a1 (COL I/III) and TGF-b1 to TGF-b3

(b1/b3) expression by myofibroblasts, pMFBs and CCD-

18co cells were treated with PMA alone, PMA/PGE2, or

PMA/PC, and COL1a1, COL3a1, TGF-b1, and TGF-b3

mRNA expression was measured and compared.

Untreated controls had a numerically lower COL I/III

ratio in both pMFBs and CCD-18co cells, while treatment

with PMA resulted in significantly higher levels of COL

I/III ratios in both pMFBs and CCD-18co cells (p \ 0.0001

for both; Fig. 6a, b). PMA/PGE2-treated pMFBs and CCD-

18co cells had significantly lower COL I/III ratios

(p \ 0.05 for pMFBs and p \ 0.001 for CCD-18co;

Fig. 6a, b). These ratios were similar to the COL I/III ratios

expressed by the controls, suggesting that PGE2 induces a

healing process that is more regimented, while PMA-

induced collagen synthesis is dysregulated. Similarly,

PMA/PC-treated cells resulted in significantly lower COL

I/III ratios (p \ 0.0001 for pMFBs and p \ 0.05 for CCD-

18co).

PMA-treated cells had significantly higher b1/b3 ratios

in both pMFBs (p \ 0.05; Fig. 6c) and CCD-18co cells

(p \ 0.001; Fig. 6d). In pMFBs, PMA/PGE2 treatment

resulted in an approximate 1:1 ratio of b1/b3, while in

CCD-18co cells a significantly lower b1/b3 ratio was

observed (p \ 0.001; Fig. 6d). The opposite was noted for

PMA/PC-treated cells, where a significantly lower b1/b3

ratio was observed in pMFBs (p \ 0.001; Fig. 6c) and an

approximate 1:1 ratio of b1/b3 (Fig. 6d).

Exogenous PGE2 Protects Against the Pro-

inflammatory and Pro-fibrotic Effects of TNBS/Indo

Indomethacin (Indo) is a NSAID that can cause gastric

bleeding and induce ulcerations in the gastric lining,

leading to inflammation and fibrosis. As Indo is a COX-2

inhibitor, it was of interest to see whether exogenous PGE2

could protect against Indo-induced mucosal damage. Ani-

mals treated with a combination of TNBS/Indo/PGE2 had

significantly lower levels of inflammatory cell infiltrate and

collagen deposition compared to TNBS/Indo-treatment

alone (Supp Fig. 4aa–af), which translated into

Fig. 6 Cellular ratios of COL

I/III (a, b) and b1/b3 (c,

d) mRNA expression in pMFBs

and CCD-18co cells. RNA was

isolated from cells following

treatment with PMA alone,

PMA/PGE2, and PMA/DLPC

for 24 h. COL1a1, COL3a1,

TGF-b1, and TGF-b3

expression was measured and

normalized to the housekeeping

gene, GAPDH. Results were

expressed as mean ± SD.

n = 3 replicated per treatment/

experiment, and each

experiment was performed in

duplicate. *p \ 0.05 versus

COL1a1 or TGF-b1,

respectively

Dig Dis Sci

123

significantly lower inflammation and fibrosis scoring at 2

and 6 weeks (p B 0.004; Supp Fig. 4ag). These animals

also had significantly lower a-SMA staining compared to

TNBS/Indo (p = 0.02; Supp Fig. 4b), suggesting that

PGE2 was capable of inhibiting myofibroblast recruitment

and localization. When looking at PGE2 effects in vitro, it

was observed that in both pMFBs and CCD-18co cells

PGE2 was able to significantly inhibit COL1a1, TGF-b1,

and TIMP-1 expression (p \ 0.0001; Supp Fig. 4ca–cf),

while significantly increasing MMP-13/MMP-1 and COX-

2 (p B 0.009; Supp Fig. 4cg–cj). When comparing the

COL I/III and b1/b3 ratios, it was observed that in both

pMFBs and CCD-18co cells TNBS/Indo/PGE2 treatment

results in significantly lower COL I/III (p \ 0.05; Supp

Fig. 4d) and lower/equal b1/b3 ratios (p \ 0.0001 for

pMFBs only). These data demonstrate that PGE2 plays an

important role in protecting against NSAID-induced

mucosal damage.

Discussion

Intestinal fibrosis is a well-described complication of

longstanding CD resulting from chronic inflammation and

dysregulated wound healing [22, 23]. Although various

anti-inflammatory and immunosuppressive agents have

demonstrated efficacy for active CD, they have little

impact on fibrosis once it occurs [23].

In this study, it was demonstrated that both PGE2 and

PC are capable of protecting against the development of

intestinal fibrosis in TNBS-induced murine models of

chronic inflammation via several different avenues that

include the maintenance of mucosal integrity, reducing the

influx of mononuclear cells, regulating immune response,

and preventing myofibroblast accumulation within the

lamina propria. Their ability to regulate intestinal myofi-

broblast function was also documented, suggesting that

they also play a key role in the maintenance of ECM

homeostasis providing possible therapeutic targets for the

treatment or prevention of IBD related fibrosis.

Aberrant epithelial barrier function plays a central role

in the pathophysiology of IBD. PGE2 is known to promote

epithelial restitution and inhibit the destabilisation of the

epithelial barrier function [24], while PC has been shown to

decrease mucosal permeability to reestablish surface

hydrophobicity [25]. In accordance with this, the intro-

duction of PGE2 and PC to TNBS-treated animals in this

study resulted in preservation of the colonic architecture,

decreased leukocyte accumulation, and significantly less

inflammation and fibrosis compared to animals treated with

TNBS alone.

The consistently low inflammation and fibrosis scores,

low numbers of a-SMA? staining and decreased leukocyte

accumulation over 2 and 6 weeks of TNBS treatment,

suggest that both PGE2 and PC are capable of promoting

and maintaining normal tissue healing, thereby preventing

significant intestinal fibrosis over time. In line with this, the

initial decreases in ECM-related gene expression in PGE2-

and PC-treated animals at 2 weeks were not as marked by

week 6, which infers significant inhibition of excess type I

collagen (COL1a1) production in the early phases of

healing to prevent the formation of scar tissue, followed by

controlled collagen production during the remodeling

phase to restore normal dermal collagen composition. A

complete resolution of inflammation was not observed,

which is thought to be a result of the animals only receiving

a fixed concentration of PGE2 and PC over the 6 week

period, while TNBS was administered weekly in escalating

doses, therefore to achieve complete resolution, increasing

concentration of PGE2 and PC over time may be required.

MTT assays performed on pMFBs, together with the

low numbers of a-SMA? staining in the lamina propria,

suggest that PGE2 and PC are also capable of regulating

myofibroblast accumulation by decreasing their prolifera-

tion. PGE2’s ability to inhibit myofibroblast chemotaxis

[26], epithelial-to-mesenchymal transition [27], and endo-

thelial-to-mesenchymal transition [28] may also be con-

tributing to the lower levels of a-SMA? staining in the

lamina propria.

The intestinal epithelium is also actively involved in

antigen processing and immune cell regulation and sig-

nalling via transmembrane pattern recognition receptors

(TLRs) [29]. TLR signalling drives basal immune mecha-

nisms essential for host barrier integrity and maintaining

tolerance. Aberrant or dysfunctional TLR signalling may

impair commensal-mucosal homeostasis, contributing to

the amplification and perpetuation of tissue injury resulting

in the chronic inflammation and fibrosis observed in IBD.

Treatment with PGE2 inhibited TNBS-induced TLR4

mRNA expression at 2 and 6 weeks, which was also

accompanied by decreases in the key pro-inflammatory

cytokine, IL-1b, IL-6, and TNF, expression. Down-regu-

lation of TLR4 expression has also been reported in lung

alveolar macrophages in rats [30], where PGE2 was found

to dampen the innate immune response by inhibiting TLR4

mRNA translation into protein as opposed to down-regu-

lating TLR4 mRNA expression itself. Regardless of the

mechanism, this demonstrates that PGE2 can regulate the

innate immune response.

Treatment with PC also resulted in significantly lower

TLR4 mRNA expression and decreased IL-1b, IL-6, and

TNF expression. PC has the ability to integrate into spe-

cialized membrane microdomains known as lipid rafts to

displace the assembly, activation, and subsequent signal-

ling of certain membrane receptors including TLRs [31,

32], which may explain the decrease in pro-inflammatory

Dig Dis Sci

123

cytokine production. Together this suggests that the anti-

inflammatory and anti-fibrotic effects of PC may be med-

iated not only by its membrane-associated properties, but

also via its ability to regulate the innate immune response.

Oral PGE2 administration resulted in significantly

increased PGE2 levels in the colonic tissue, which is

important because IBD patients that have deficiencies in

endogenous PGE2 synthesis due to the expression of the

dominant negative G-protein a-subunit 2 by intestinal

myofibroblasts, have more severe inflammation [24]. By

being able to counteract these deficiencies with oral PGE2

is thus paramount for the modulation of intestinal immune

response and epithelial restitution.

This increase coupled with the observed increase in

COX-2 levels may also be contributing to its ability to

inhibit the NSAID-induced mucosal damage. PC can be

degraded by phospholipases, such as cytosolic A2a, into

arachidonic acid, then converted to PGE2 via COX-2 [15],

and can therefore mediate its protective effects via PGE2.

With PC having no effect on endogenous PGE2 levels, it

clearly indicates that the anti-inflammatory and anti-fibro-

genic effects were mediated via PC itself and not by its

derivative PGE2.

Further investigation of the anti-fibrogenic potential of

PGE2 and PC using pMFBs and CCD-18co cells clearly

demonstrated that they both play a pivotal physiological

role in maintaining ECM homeostasis by regulating col-

lagen production. Decreases in COL1a1 mRNA were

accompanied by the up-regulation of MMP-13/MMP-1 and

COX-2 in pMFBs and CCD-18co cells when treated with

PGE2. MMP-13/MMP-1 is the interstitial collagenase

responsible for specifically breaking down collagen types I,

II, and III; therefore, its up-regulation aids in the deposition

of newly synthesized collagens. This suggests that in the

presence of an inflammatory mediator, PGE2 is capable of

regulating collagen production and deposition. Although

PC inhibited collagen production in both pMFBs and CCD-

18co cells, it was demonstrated that in pMFBs the decrease

in COL1a1 was accompanied by a decrease in TGF-b1 and

TIMP-1 expression, suggesting that it inhibits TGF-b1-

induced collagen production and induces MMP production

in a paracrine fashion. In contrast to this, in CCD-18co

cells, PC had no effect on TGF-b1 mRNA levels but sig-

nificantly up-regulated MMP-1. This suggests that PC can

regulate collagen deposition as well as production by

human myofibroblasts, but only production in murine

myofibroblasts, which may be important when investigat-

ing specific targets for inhibiting aberrant collagen

production.

Further evidence of PGE2 and PC’s anti-fibrogenic

effects was their ability to regulate the COL I/III and b1/b3

ratios produced by pMFBs and CCD-18co cells. During the

initial inflammatory phase, there is a low COL I/III ratio,

which then shifts to a high COL I/III ratio during the

regeneration phase followed by equal proportions of COL I

and COL III during the final remodeling phase [33]. These

changes usually occur in a predictable manner; however, if

the healing process is disrupted or impaired, there is a

persistent shift toward a high COL I/III ratio, leading to

fibrogenesis. TGF-b1 promotes ECM deposition with

scarring and fibrosis [34], while the addition of TGF-b3 to

these wounds reduced ECM deposition and scarring [35].

In humans, highly inflamed and fibrosed intestinal tissue

from UC and CD patients have significantly higher TGF-b1

gene expression compared to normal tissue [36] and

reduced TGF-b3 expression by intestinal myofibroblasts

[37]. Having the ability to regulate these ratios suggests

that PGE2 and PC can both promote a microenvironment

that does not support persistent inflammation and sub-

sequent fibrosis.

In conclusion, this study demonstrated that both PGE2

and PC have potent anti-fibrogenic potentials that are

mediated, not only by their ability to protect the gastric

mucosal integrity in order to limit inflammation, but also

by their ability to directly regulate fibroblast function. By

doing so, both PGE2 and PC can inhibit the production of

pro-inflammatory cytokines, inhibit the recruitment and

accumulation of leukocytes, promote epithelial restitution,

and most importantly maintain ECM homeostasis to pre-

vent fibrogenesis. These multiple modes of action give rise

to a plethora of novel targets that may be further investi-

gated in an effort to develop therapeutic approaches to

preventing the initial development and post-surgical

recurrence of fibrosis in patients in IBD.

Acknowledgments The authors wish to thank Dr Cynthia H., For-

rest (CHF) from the Department of Histopathology, Fremantle Hos-

pital, Fremantle, WA, Australia, for her histopathological expertise.

This research was undertaken with support from the Raine Medical

Research Foundation and the Fremantle Hospital Medical Research

Foundation.

Conflict of interest The authors declare that they have no conflict

of interest.

References

1. Rieder F, et al. Predictors of fibrostenotic Crohn’s disease. In-

flamm Bowel Dis. 2000;17:2000–2007.

2. Lawrance IC, et al. Altered response of intestinal mucosal

fibroblasts to profibrogenic cytokines in inflammatory bowel

disease. Inflamm Bowel Dis. 2001;7:226–236.

3. Lawrance IC, et al. Inflammation location, but not type, deter-

mines the increase in TGF-beta1 and IGF-1 expression and col-

lagen deposition in IBD intestine. Inflamm Bowel Dis.

2001;7:16–26.

4. Van Assche G, et al. Medical therapy for Crohn’s disease stric-

tures. Inflamm Bowel Dis. 2004;10:55–60.

Dig Dis Sci

123

5. Luna J, et al. Mesenchymal cell proliferation and programmed

cell death: key players in fibrogenesis and new targets for ther-

apeutic intervention. Am J Physiol Gastrointest Liver Physiol.

2011;300:G703–G708.

6. Dale PD, et al. A mathematical model for collagen fibre forma-

tion during foetal and adult dermal wound healing. Proc Biol Sci.

1996;263:653–660.

7. Hodges RJ, et al. Severity of lung injury in cyclooxygenase-2-

deficient mice is dependent on reduced prostaglandin E(2) pro-

duction. Am J Pathol. 2004;165:1663–1676.

8. Huang SK, et al. Prostaglandin E(2) induces fibroblast apoptosis

by modulating multiple survival pathways. FASEB J.

2009;23:4317–4326.

9. Keerthisingam CB, et al. Cyclooxygenase-2 deficiency results in

a loss of the anti-proliferative response to transforming growth

factor-beta in human fibrotic lung fibroblasts and promotes ble-

omycin-induced pulmonary fibrosis in mice. Am J Pathol.

2001;158:1411–1422.

10. Huang SK, et al. Variable prostaglandin E2 resistance in fibro-

blasts from patients with usual interstitial pneumonia. Am J

Respir Crit Care Med. 2008;177:66–74.

11. Moore BB, et al. Bleomycin-induced E prostanoid receptor

changes alter fibroblast responses to prostaglandin E2. J Immu-

nol. 2005;174:5644–5649.

12. Wallace JL. Prostaglandin biology in inflammatory bowel dis-

ease. Gastroenterol Clin North Am. 2001;30:971–980.

13. Iwanaga K, et al. Prostaglandin E2 promotes wound-induced

migration of intestinal subepithelial myofibroblasts via EP2, EP3,

and EP4 prostanoid receptor activation. J Pharmacol Exp Ther.

2012;340:604–611.

14. Ehehalt R, et al. Phosphatidylcholine and lysophosphatidylcho-

line in intestinal mucus of ulcerative colitis patients. A quanti-

tative approach by nanoElectrospray-tandem mass spectrometry.

Scand J Gastroenterol. 2004;39:737–742.

15. Schneider H, et al. Lipid based therapy for ulcerative colitis-

modulation of intestinal mucus membrane phospholipids as a tool

to influence inflammation. Int J Mol Sci. 2010;11:4149–4164.

16. Stremmel W, et al. Phosphatidylcholine for steroid-refractory

chronic ulcerative colitis: a randomized trial. Ann Intern Med.

2007;147:603–610.

17. Stremmel W, et al. Retarded release phosphatidylcholine benefits

patients with chronic active ulcerative colitis. Gut.

2005;54:966–971.

18. Lawrance IC, et al. A murine model of chronic inflammation-

induced intestinal fibrosis down-regulated by antisense NF-kappa

B. Gastroenterology. 2003;125:1750–1761.

19. Klopcic B, et al. Indomethacin and retinoic acid modify mouse

intestinal inflammation and fibrosis: a role for SPARC. Dig Dis

Sci. 2008;53:1553–1563.

20. Neurath MF, et al. Antibodies to interleukin 12 abrogate estab-

lished experimental colitis in mice. J Exp Med.

1995;182:1281–1290.

21. Ng YL, et al. Secreted protein acidic and rich in cysteine

(SPARC) exacerbates colonic inflammatory symptoms in dextran

sodium sulphate-induced murine colitis. PLoS One.

2013;8:e77575.

22. Fiocchi C. Inflammatory bowel disease: etiology and pathogen-

esis. Gastroenterology. 1998;115:182–205.

23. Froehlich F, et al. Fibrostenotic Crohn’s disease. Digestion.

2007;76:113–115.

24. Edwards RA, Smock AZ. Defective arachidonate release and

PGE2 production in Gi alpha2-deficient intestinal and colonic

subepithelial myofibroblasts. Inflamm Bowel Dis.

2006;12:153–165.

25. Fabia R, et al. Effects of phosphatidylcholine and phosphatidyl-

inositol on acetic-acid-induced colitis in the rat. Digestion.

1992;53:35–44.

26. Kohyama T, et al. Prostaglandin E(2) inhibits fibroblast chemo-

taxis. Am J Physiol Lung Cell Mol Physiol. 2001;281:L1257–

L1263.

27. Flier SN, et al. Identification of epithelial to mesenchymal tran-

sition as a novel source of fibroblasts in intestinal fibrosis. J Biol

Chem. 2010;285:20202–20212.

28. Horie-Sakata K, et al. Role of cyclooxygenase 2 in hepatocyte

growth factor-mediated gastric epithelial restitution. J Clin

Gastroenterol. 1998;27:S40–S46.

29. Szabo G, et al. TLR4, ethanol, and lipid rafts: a new mechanism

of ethanol action with implications for other receptor-mediated

effects. J Immunol. 2007;178:1243–1249.

30. Degraaf AJ, et al. Prostaglandin E2 reduces Toll-like receptor 4

expression in alveolar macrophages by inhibition of translation.

Am J Respir Cell Mol Biol. 2014;51:242–250.

31. Anes E, et al. Selected lipids activate phagosome actin assembly

and maturation resulting in killing of pathogenic mycobacteria.

Nat Cell Biol. 2003;5:793–802.

32. Miranda DT, et al. Soy lecithin supplementation alters macro-

phage phagocytosis and lymphocyte response to concanavalin A:

a study in alloxan-induced diabetic rats. Cell Biochem Funct.

2008;26:859–865.

33. Ruszczak Z. Effect of collagen matrices on dermal wound heal-

ing. Adv Drug Deliv Rev. 2003;55:1595–1611.

34. Shah M, et al. Neutralising antibody to TGF-beta 1,2 reduces

cutaneous scarring in adult rodents. J Cell Sci.

1994;107:1137–1157.

35. Shah M, et al. Neutralisation of TGF-beta 1 and TGF-beta 2 or

exogenous addition of TGF-beta 3 to cutaneous rat wounds

reduces scarring. J Cell Sci. 1995;108:985–1002.

36. Lawrance IC, et al. Ulcerative colitis and Crohn’s disease: dis-

tinctive gene expression profiles and novel susceptibility candi-

date genes. Hum Mol Genet. 2001;10:445–456.

37. McKaig BC, et al. Differential expression of TGF-beta isoforms

by normal and inflammatory bowel disease intestinal myofibro-

blasts. Am J Physiol Cell Physiol. 2002;282:C172–C182.

Dig Dis Sci

123