Therapeutic Effect of Proteinase-Activated Receptor-1 Antagonist on Colitis-Associated Carcinogenesis

Abstract

Background & Aims

Inflammatory bowel disease is associated with carcinogenesis, which limits the prognosis of the patients. The local expression of proteinases and proteinase-activated receptor 1 (PAR₁) increases in inflammatory bowel disease. The present study investigated the therapeutic effects of PAR₁ antagonism on colitis-associated carcinogenesis.

Methods

A colitis-associated carcinogenesis model was prepared in mice by treatment with azoxymethane (AOM) and dextran sulfate sodium (DSS). PAR₁ antagonist E5555 was administered in long- and short-term protocol, starting on the day of AOM injection and 1 week after completing AOM/DSS treatment, respectively. The fecal samples were collected for metagenome analysis of gut microbiota. The intestinal myofibroblasts of the Crohn’s disease patients were used to elucidate underlying cellular mechanisms. Caco-2 cells were used to investigate a possible source of PAR₁ agonist proteinases.

Results

AOM/DSS model showed weight loss, diarrhea, tumor development, inflammation, fibrosis, and increased production of inflammatory cytokines. The β-diversity, but not α-diversity, of microbiota significantly differed between AOM/DSS and control mice. E5555 alleviated these pathological changes and altered the microbiota β-diversity in AOM/DSS mice. The thrombin expression was up-regulated in tumor and non-tumor areas, whereas PAR₁ mRNA expression was higher in tumor areas compared with non-tumor areas. E5555 inhibited thrombin-triggered elevation of cytosolic Ca²⁺ concentration and ERK1/2 phosphorylation, as well as IL6-induced signal transducer and activator of transcription 3 (STAT3) phosphorylation in intestinal myofibroblasts. Caco-2 cell-conditioned medium contained immunoreactive thrombin, which cleaved the recombinant protein containing the extracellular domain of PAR₁ at the thrombin cleavage site.

Conclusions

PAR₁ antagonism is proposed to be a novel therapeutic strategy for treatment of inflammatory bowel disease and its associated carcinogenesis.

Graphical abstract

Summary.

Inflammatory bowel disease is frequently associated with carcinogenesis, which limits the prognosis of patients. The present study demonstrates that PAR₁ antagonist ameliorated inflammation, fibrosis, tumor genesis, and microbiota changes in colitis-associated cancer mouse models.

Inflammatory bowel disease (IBD), represented by Crohn’s disease and ulcerative colitis, is characterized by chronic, recurrent inflammation in the digestive tract.¹ Over the past decade, IBD has emerged as a global public health challenge worldwide, and incidence among children and older adults has increased dramatically.^2,3 Pathogenesis of IBD is associated with the environment, lifestyle, infections, antibiotic use, and gut microbiota.4, 5, 6, 7, 8 The risk of colorectal tumors increases because of repeated remission and relapse of inflammation, which limits the prognosis of the patients with IBD.⁹ Better understanding of the underlying mechanism of carcinogenesis is a prerequisite for developing a novel therapeutic strategy.

Protease-activated receptors (PARs), a unique family of G protein-coupled receptors, are activated by proteolytic cleavage at the N-terminal extracellular domain by the agonist proteinases that unmask the receptor-activating tethered ligand, which binds intramolecularly and causes conformational changes, thus eliciting downstream intracellular signaling.¹⁰ PAR₁ is expressed in cancer-related fibroblasts, epithelial cells, and macrophages in the tumor microenvironment.¹¹ PAR₁ serves as a receptor for thrombin and plays important roles in carcinogenesis.12, 13, 14 PAR₁ activation can impact nociception, motility, proliferation, secretion, barrier function, mucosal defense, and vascular remodeling.15, 16, 17 Activated PAR₁ disrupts mucosal barrier function by promoting excessive epithelial cell apoptosis and thereby induces the translocation of luminal content into intestinal wall.¹⁸ The activity of proteinases including thrombin is notably elevated in inflamed intestinal tissues in animal models and Crohn’s disease patients.^19,20 Vorapaxar, a PAR₁ antagonist, has been reported to inhibit the thrombin-induced proinflammatory effects in colonic tissues.¹² Therefore, PAR₁ could be a potential target for treatment of IBD and its associated carcinogenesis.

In the present study, the therapeutic effects of a PAR₁ antagonist E5555 on colitis-associated carcinogenesis in the mouse model were investigated. The intestinal myofibroblasts were used to investigate the cellular mechanisms underlying the therapeutic effect of E5555, whereas Caco-2 cells were used to investigate a possible source of agonist proteinases. Microbial colonization²¹ and antibiotic treatment²² profoundly impact the transcriptional profile of intestinal epithelial cells, leading to the up-regulation or down-regulation of hundreds of genes.²³ When compared with germ-free mice, the vascular remodeling of villous structures in colonized mice appears to be under the control of the microbiota’s up-regulation of PAR₁.¹⁸ The up-regulation and activation of PAR₁ could trigger abnormal inflammation and cell signaling, potentially increasing the risk of intestinal diseases including IBD and cancer development. PAR₁ plays a key role in linking intestinal bacteria and gastrointestinal function. It is thus necessary to examine how PAR₁ suppression affects the intestinal microbiota. Therefore, the impact of PAR₁ antagonist on the gut microbiota was also investigated in the present study.

Results

E5555 Treatment Ameliorated the Inflammation and Carcinogenesis in the AOM/DSS Model

The azoxymethane/dextran sulfate sodium (AOM/DSS) group exhibited a significant decrease in body weight compared with the control group. Both short- and long-term treatments with E5555 did not affect body weight (Figure 1A and B). The prolapse of anal neoplasms was frequently observed in the AOM/DSS group but never in the control or E5555 treatment groups (Figure 2B). The AOM/DSS group showed an increase in the scores of stool consistency (Figure 2C) and occult/gross bleeding at the end of the protocol (Figure 2D). Both short-term and long-term treatments with E5555 significantly reduced these scores. The AOM/DSS group developed a significant number of colon tumors, whereas both long- and short-term treatments with E5555 significantly reduced the number of tumors (Figure 2E and F). Histologically, both E5555-treated and non-treated AOM/DSS groups showed moderately to well-differentiated tubular adenocarcinoma. (Figure 2G). The mRNA expression of PAR₁ (f2r) and PAR₄ (f2rl3) was significantly up-regulated in the tumor area in the AOM/DSS group, whereas that of PAR₂ (f2l1r) and PAR₃ (f2rl2) mRNA was down-regulated, compared with the non-tumor area (Figure 2H). Furthermore, positive staining of thrombin was observed in the mucosa and submucosal area of the tumor area (Figure 2I), whereas the expression of thrombin mRNA was up-regulated in the non-tumor areas, but not the tumor area, of the AOM/DSS group (Figure 2J).

Figure 1.

Body weight changes in the experimental groups. Time course of changes in body weight (A) and body weight at week 21 (B) of each experimental group. n = 8. ∗∗∗P < .001.

Figure 2.

Effects of E5555 on carcinogenesis in AOM/DSS model. (A) Experimental protocol. (B) Representative pictures of anal area of indicated mice at time of fecal sampling. (C–E) Summaries of stool consistency score (C; n = 8), occult/gross bleeding score (D; n = 8), and number of tumors (E; n = 8). (F) Representative pictures of colon, showing the tumor development. (G) Representative image of HE stains of tumor areas of the indicated mice. (H) Summary of mRNA expression of PAR₁ (f2r), PAR₂ (f2rl1), PAR₃ (f2rl2), and PAR₄ (f2rl3) in non-tumor and tumor areas of colon of AOM/DSS group (n = 4). (I) Representative pictures of HE stains of tumor area and immunofluorescence staining for thrombin.¹⁰Dotted squares indicate area enlarged. Nuclei were labeled with To-Pro-3 (blue). (J) Summary of expression of thrombin mRNA in colon of control mice (CTR) and AOM/DSS mice (non-tumor and tumor area) (n = 5). ∗P < .05, ∗∗P < .01, ∗∗∗P < .001.

The E5555 Treatment Ameliorated Inflammation, Fibrosis, and Proliferation of Non-Tumor Areas in the AOM/DSS Model

AOM/DSS group showed more edematous changes and fibrosis in the stroma and more inflammatory cell infiltration, mainly lymphoid cells, than the control group (Figure 3A). Interstitial edematous changes and inflammatory cell infiltration were reduced in the E5555 treatment group (Figure 3A). Masson’s trichrome staining revealed an extensive increase in collagen fibers in the AOM/DSS group. The fibrosis was suppressed in both long- and short-term treatments with E5555 groups (Figure 3A). The AOM/DSS group showed significantly higher scores of inflammations and fibrosis than those seen with the control (Figure 3B). Both long- and short-term treatments with E5555 mitigated the inflammation and fibrosis and decreased the respective scores (Figure 3A and B). The mRNA expression of transforming growth factor−β (TGF-β), a key inducer of fibrogenesis, was up-regulated in the non-tumor area of the colon in the AOM/DSS group, whereas both long-term and short-term treatments with E5555 inhibited this up-regulation (Figure 3C). The mRNA expression of some proinflammatory cytokines (tumor necrosis factor alpha [TNF-α], interleukin [IL] 6, and interferon gamma) was also up-regulated in the non-tumor area of the colon in the AOM/DSS group, whereas both long- and short-term treatments with E5555 inhibited this up-regulation (Figure 3D). The epithelial proliferative activity was evaluated by the level of how high proliferating cell nuclear antigen (PCNA)-positive cells reached the luminal side of mucosa (Figure 3E and F). The PCNA-positive cells reached a higher level in the AOM/DSS group than that seen in the control. The levels that PCNA-positive cells reached in the groups of long- and short-term treatments with E5555 were both significantly lower than those seen in the AOM/DSS group (Figure 3E and F).

Figure 3.

Effects of E5555 on inflammation and fibrosis in AOM/DSS model. (A) Representative image of HE and Masson’s trichrome staining at non-tumor areas. (B) Summaries of inflammation score (n = 8) and fibrosis score (n = 8). (C) The mRNA expressions of TGF-β1 in non-tumor areas (n = 8). (D) The mRNA expression of TNF-α, IL6, and interferon γ in non-tumor areas (n = 8). (E) Representative image of immunofluorescence staining for PCNA (purple or pink)²⁴ in non-tumor areas. To-Pro-3 (blue) was used to label the nuclei (n = 8). (F) Summary of ratio of length of PCNA-positive area to length of mucosal layer (n = 8). Scale bar: 100 μm. ∗P < .05, ∗∗P < .01, ∗∗∗P < .001.

E5555 Treatment Significantly Alters the Structure of the Gut Microbiota

On day 0 of the experimental protocol (Figure 2A), there was no significant difference in both α-diversity (Shannon-Wiener diversity index and Chao1 richness) and β-diversity (principal coordinates analysis based on Bray–Curtis distance) of gut microbiota between the control mice and the mice that were later subjected to AOM/DSS administrations either with or without E5555 treatment (Figure 4A). At the end of the protocol (week 21), there was a significant difference in β-diversity, but not α-diversity, between the control and AOM/DSS groups (Figure 4B). Both long- and short-term treatments with E5555 caused a significant difference in β-diversity, but not α-diversity, from both AOM/DSS and control groups (Anosim test, P < .05) (Figure 4B).

Figure 4.

Microbial diversity of mice microbiota at the start and end point. (A) Summaries of Shannon-Wiener index and Chao 1 richness and principal coordinates analysis score plots of the fecal microbiome on day 0 in control mice (CTR; n = 5) and mice (A/D; n = 5), which were later subjected to AOM/DSS administrations either with or without E5555 treatment. P value and R value obtained with analysis of similarities (ANOSIM, 999 permutations) are shown in a key of the principal coordinates analysis score plots. (B) Summaries of Shannon-Wiener index and Chao 1 richness and principal coordinates analysis score plots of the fecal microbiome in the indicated mice groups (CTR, E(L), A/D, A/D+E(S), A/D+E(L)) at the end of the protocol, week 21. P values and R values obtained for the indicated comparison with the analysis of similarities (ANOSIM, 999 permutations) are shown in the table.

E5555 Treatment Modulated Gut Microbiota-Related Colitis-Associated Carcinogenesis Modules

To further explore changes in the gut microbiota at a finer level after E5555 treatment, the relative abundance of species between inter-groups was calculated, and a total of 33 responsive species-level genome bins (SGBs) were identified. The relative abundance of Lachnospiraceae bacterium (COE1 sp003513705, COE1 sp009774375, COE1 sp000403335, TF01-11 sp003612285, and UBA3282 sp003611805) and Borkfalkiaceae bacterium (UMGS1004 sp900547645) in the AOM/DSS group was significantly lower than that in control group (Wilcoxon test, P < .05). The relative abundance of Ruminococcaceae bacterium UBA7177 sp002491225, Akkermansia muciniphila, Phocaeicola sartorii, Prevotellamassilia sp009775705, and Lachnospiraceae bacterium (COE1 sp002490665, UBA3282 sp009774175) in the AOM/DSS group was significantly higher than that in control group (Wilcoxon test, P < .05). Both long- and short-term treatment with E5555 significantly increased the relative abundance of Ruthenibacterium lactatiformans compared with the AOM/DSS group (Wilcoxon test, P < .05), whereas it decreased the relative abundance of Oscillospiraceae bacterium UMGS1872 sp009774055, Lawsonibacter sp900550705, and Flavonifractor sp002159455 compared with the AOM/DSS group (Wilcoxon test, P < .05). Short-term treatment with E5555 decreased the relative abundance of Haliangiaceae bacterium WYBA01 sp011526095, Borkfalkiaceae bacterium UBA11940 sp003503945, Lawsonibacter sp000492175, Alistipes sp900546065, Duncaniella muris, Parabacteroides sp014287585, Bacteroides fluxus, and Intestinimonas timonensis (Wilcoxon test, P < .05). The long-term treatment with E5555 increased the relative abundance of Christensenellales bacterium RACS-045 sp013316135, Duncaniella sp001689575, Anaeroplasmataceaebacterium UMGS268 sp900540705, Dysosmobacter sp009774015, and Muribaculaceae bacterium CAG-485 sp009775365 compared with the AOM/DSS group (Wilcoxon test, P < .05), whereas it decreased the relative abundance of Lachnospiraceae bacterium UBA7050 sp002493905, Duncaniella dubosii, and Muribaculaceae bacterium CAG-485 sp002493045 compared with the AOM/DSS group (Wilcoxon test, P < .05). The relative abundance of Christensenellales bacterium RACS-045 sp013316135, Duncaniella sp001689575, Clostridia bacterium UBA7597 sp900767195, Alistipes sp900546065, and Intestinimonas timonensis was higher, whereas the relative abundance of Muribaculaceae bacterium CAG-485 sp002493045 was lower in long-term treatment compared with short-term treatment (Wilcoxon test, P < .05) (Figure 5, Tables 1 and 2).

Figure 5.

Profiles of significantly different species and gut metabolic modules among groups at week 21. The heatmap displays the results of comparisons of reads per kilobase million (RPKM) of indicated species (left panel) and abundance of indicated metabolic modules (lower panel) between the indicated groups. ∗P < .05, ∗∗P < .01. Circles on the grid indicate distribution of indicated metabolic modules, which are related to enteritis-related cancer, across the indicated species. Raw data are listed in Tables 1 and 2.

Table 1.

Detailed Information on the Differential Metabolic Modules Shown in Figure 5

Differential modules		Abundance of modules				P value, Wilcoxon.test
Modules ID	Metabolic modules	Mean_Ca	Mean_Cc	Mean_Cd	Mean_Ce	Ca vs Cc	Cd vs Cc	Ce vs Cc	Cd vs Ce
MF0025	Tryptophan degradation I	170.9854941	105.1412004	96.60646202	72.80978106	.016293604	>.05	>.05	>.05
MF0022	Galacturonate degradation I	225.0027935	167.8649282	168.761176	158.8532287	.009023439	>.05	>.05	>.05
MF0019	Rhamnose degradation	157.2640078	118.3412759	97.74014966	97.88024286	.047201768	>.05	>.05	>.05
MGB019	GABA degradation	2.710054372	6.537104951	1.698082982	2.433539259	>.05	.028280123	>.05	>.05
MGB006	Glutamate synthesis I	380.9198817	365.9615331	376.614974	332.290935	>.05	>.05	.009023439	.009023439
MGB007	Glutamate synthesis II	380.7243942	362.6713766	373.3700252	331.4737267	>.05	>.05	.009023439	.009023439
MF0085	Urea degradation	169.3110035	136.921046	96.57169634	73.00481779	>.05	>.05	.009023439	>.05
MF0086	Acetyl-CoA to acetate	392.5572142	374.5695461	390.5667034	342.1328502	>.05	>.05	.028280123	.016293604
MF0006	Lactose degradation	359.1510279	348.7359444	347.2692078	319.3121955	>.05	>.05	.009023439	.047201768
MF0089	Butyrate production II	13.15625469	12.0544267	10.42123381	4.36334591	>.05	>.05	.016293604	>.05
MF0002	Fructan degradation	141.3192878	156.5582326	154.2348961	180.1974749	>.05	>.05	.047201768	.028280123
MF0092	Lactate production	206.4830665	145.4908565	145.1559707	142.8913683	.016293604	>.05	>.05	>.05
MF0017	Galactose degradation	284.0354572	274.3322049	273.9540056	246.9252845	>.05	>.05	.028280123	.016293604
MF0012	Trehalose degradation	35.75825997	44.1287524	21.99046991	42.7246804	>.05	.016293604	>.05	>.05
MF0098	Hydrogen metabolism	271.427465	211.1191102	183.3835793	181.7043419	.009023439	>.05	>.05	>.05
MF0011	Sucrose degradation II	20.48289052	9.697062473	13.15651599	14.74317489	.047201768	>.05	>.05	>.05
MF0009	Melibiose degradation	320.4705953	308.6470816	312.6860029	292.144795	>.05	>.05	.028280123	.028280123
MGB057	Corrinoid dependent enzymes	394.3986422	375.6885547	393.0226785	344.4872581	>.05	>.05	.028280123	.016293604
MGB053	Butyrate synthesis II	13.15625469	12.0544267	10.42123381	4.36334591	>.05	>.05	.016293604	>.05
MGB049	Tryptophan degradation	170.9854941	105.1412004	96.60646202	72.80978106	.016293604	>.05	>.05	>.05
MGB041	Menaquinone synthesis (vitamin K2) II (alternative pathway: futalosine pathway)	27.99674797	22.511588	19.65920297	8.325459131	>.05	>.05	.028280123	.016293604
MGB043	Acetate synthesis I	341.8662155	342.9676702	363.8411132	325.9423946	>.05	.009023439	.028280123	.009023439
MF0078	Lactaldehyde degradation	185.2069177	127.7497298	97.61548116	98.35978963	.016293604	>.05	>.05	>.05
MF0076	4-aminobutyrate degradation	1.655552554	3.09126546	0.869561884	0.53069585	>.05	>.05	.047201768	>.05
MGB035	Isovaleric acid synthesis II (KADC pathway)	36.41001535	36.45607484	35.81877077	18.62192959	>.05	>.05	.028280123	.047201768
MGB036	S-Adenosylmethionine (SAM) synthesis	382.0185457	365.7463117	379.5768599	332.8285384	>.05	>.05	.016293604	.009023439
MF0047	Glutamine degradation II	380.7243942	362.6713766	373.3700252	331.4737267	>.05	>.05	.009023439	.009023439
MF0044	Cysteine degradation I	184.8558395	236.5702117	271.7297324	236.8598359	.047201768	>.05	>.05	>.05
MF0050	Threonine degradation II	284.5531118	251.9744724	232.3810317	194.2441589	>.05	>.05	.028280123	>.05
MF0058	Lysine degradation II	3.798591702	2.302546622	0.478389848	0.352129212	>.05	>.05	.047201768	>.05
MF0054	Arginine degradation IV	11.15676632	11.23908853	9.42228484	3.565764323	>.05	>.05	.016293604	>.05

Table 2.

Detailed Information of Significantly Different SGBs Shown in Figure 5

Differential SGBs		Abundance of RPKM				P value, Wilcoxon.test
SGBs ID	Taxonomy	Mean_Ca	Mean_Cc	Mean_Cd	Mean_Ce	Ca vs Cc	Cd vs Cc	Ce vs Cc	Cd vs Ce
SGB.76	Akkermansia muciniphila	0.57570553	1.7935802	12.14610053	12.5093473	.028280123	>.05	>.05	>.05
SGB.25	Alistipes sp900546065	4.48454184	3.44327494	3.44410964	1.46854182	>.05	>.05	.009023439	.016293604
SGB.139	Anaeroplasmataceaebacterium UMGS268 sp900540705	0.123372621	0.10252299	1.438363208	0.81590992	>.05	.009023439	>.05	>.05
SGB.14	Bacteroides fluxus	2.21301258	2.32983562	1.92749925	0.763930716	>.05	>.05	.047201768	>.05
SGB.99	Borkfalkiaceae bacterium UBA11940 sp003503945	1.115836426	1.228250376	0.171906117	0.097817752	>.05	>.05	.028280123	>.05
SGB.110	Borkfalkiaceae bacterium UMGS1004 sp900547645	0.134712877	0.071457573	0.08171492	0.065886616	.047201768	>.05	>.05	>.05
SGB.133	Christensenellales bacterium RACS-045 sp013316135	0.131474378	0.097896411	1.306334518	0.111252593	>.05	.028280123	>.05	.047201768
SGB.58	Clostridia bacterium UBA7597 sp900767195	0.153253961	0.077237505	0.15940472	0.064333002	>.05	>.05	>.05	.047201768
SGB.3	Duncaniella dubosii	8.17608936	11.08834046	6.35169772	13.12030624	>.05	.047201768	>.05	>.05
SGB.26	Duncaniella muris	0.750809538	1.00155057	0.78290125	0.498717014	>.05	>.05	.028280123	>.05
SGB.131	Duncaniella sp001689575	0.218952022	0.22608845	3.82214474	0.312983506	>.05	.009023439	>.05	.009023439
SGB.135	Dysosmobacter sp009774015	2.1899677	1.479601832	7.5199514	10.59970459	>.05	.016293604	>.05	>.05
SGB.128	Flavonifractor sp002159455	3.88388916	5.55735414	0.632789642	0.475702966	>.05	.009023439	.009023439	>.05
SGB.100	Haliangiaceae bacterium WYBA01 sp011526095	2.407987544	0.841403274	0.134696437	0.110354202	>.05	>.05	.047201768	>.05
SGB.93	Intestinimonas timonensis	0.629803844	0.98768186	0.722072012	0.300436356	>.05	>.05	.028280123	.047201768
SGB.98	Lachnospiraceae bacterium COE1 sp000403335	1.055863904	0.179266889	0.190336026	0.184826467	.028280123	>.05	>.05	>.05
SGB.108	Lachnospiraceae bacterium COE1 sp002490665	0.52559075	7.0658122	6.804202448	6.140131086	.009023439	>.05	>.05	>.05
SGB.103	Lachnospiraceae bacterium COE1 sp003513705	3.00083407	0.500193982	0.16913107	9.658861164	.047201768	>.05	>.05	>.05
SGB.22	Lachnospiraceae bacterium COE1 sp009774375	0.404017824	0.176445897	0.194715195	0.209031312	.047201768	>.05	>.05	>.05
SGB.136	Lachnospiraceae bacterium TF01-11 sp003612285	0.621992134	0.346481436	1.566985162	1.11369049	.009023439	>.05	>.05	>.05
SGB.81	Lachnospiraceae bacterium UBA3282 sp003611805	0.907807342	0.157418653	0.317064736	0.204813928	.047201768	>.05	>.05	>.05
SGB.107	Lachnospiraceae bacterium UBA3282 sp009774175	1.1464108	7.79913394	3.54005966	4.915588938	.047201768	>.05	>.05	>.05
SGB.9	Lachnospiraceae bacterium UBA7050 sp002493905	0.281436992	1.073464494	0.16292603	0.16063817	>.05	.047201768	>.05	>.05
SGB.122	Lawsonibacter sp000492175	1.655552554	3.09126546	0.869561884	0.53069585	>.05	>.05	.047201768	>.05
SGB.63	Lawsonibacter sp900550705	2.374367074	1.678132698	0.548200652	0.546429356	>.05	.028280123	.047201768	>.05
SGB.18	Muribaculaceae bacterium CAG-485 sp002493045	16.21413172	28.5877202	9.0068678	27.2471438	>.05	.016293604	>.05	.028280123
SGB.42	Muribaculaceae bacterium CAG-485 sp009775365	1.623231348	1.005713442	6.31329712	1.82819698	>.05	.047201768	>.05	>.05
SGB.129	Oscillospiraceae bacterium UMGS1872 sp009774055	1.022430308	5.31850175	0.228450774	0.287353544	>.05	.009023439	.009023439	>.05
SGB.15	Parabacteroides sp014287585	2.0923636	2.25275784	2.009848794	1.024280206	>.05	>.05	.016293604	>.05
SGB.77	Phocaeicola sartorii	0.834169176	3.12820282	4.23705236	5.487186728	.016293604	>.05	>.05	>.05
SGB.7	Prevotellamassilia sp009775705	7.99422466	26.8491572	23.2977896	19.9282972	.009023439	>.05	>.05	>.05
SGB.113	Ruminococcaceae bacterium UBA7177 sp002491225	0.174480412	2.239454996	13.02246746	5.979055872	.009023439	>.05	>.05	>.05
SGB.140	Ruthenibacterium lactatiformans	0.312945174	0.413257246	2.630973808	1.253457982	>.05	.047201768	.016293604	>.05

The metabolic modules of gut microbiota were profiled by mainly focusing on the modules closely related to enteritis-related cancer. Thirty-one significantly differential metabolic modules have been identified according to 33 significantly differential SGBs among the 4 groups by pairwise comparison, which mainly included threonine degradation, arginine degradation, lysine degradation, galactose degradation, vitamin K2 synthesis, and glutamate synthesis. Specifically, the abundance of galactose degradation, menaquinone synthesis (vitamin K2) II, threonine degradation II, lysine degradation II, and arginine degradation IV seen with the short-term treatment with E5555 were significantly lower than those seen in the AOM/DSS group (Wilcoxon test, P < .05). The abundance of glutamate synthesis I seen with long-term treatment with E5555 was significantly higher, whereas the abundance of galactose degradation and menaquinone synthesis (vitamin K2) II seen with long-term treatment with E5555 were significantly lower than those seen in the AOM/DSS group (Wilcoxon test, P < .05). Overall, the E5555 treatment significantly altered the abundance of amino acid degradation and synthesis and vitamin synthesis in the gut microbiota-related enteritis-cancer modules.

Because there was a significant difference in β-diversity between the control and the long-term treatment with E5555 (Figure 4B), the effect of E5555 (long-term treatment) on the metabolic modules was further analyzed (Figure 6, Tables 3 and 4). The analysis detected 33 metabolic pathways that significantly differed between the control and the E5555 (L) groups (Wilcoxon test, P < .05). The 6 pathways (tryptophan degradation 1, rhamnose degradation, lactate production, hydrogen metabolism, sucrose degradation 2, and lactaldehyde degradation) were significantly increased in the AOM/DSS group while significantly decreased in the E5555 (L) group compared with the control group (Figure 6).

Figure 6.

Profiles of significantly different species and gut metabolic modules between the control and E5555 (L) groups at week 21. The heatmap displays the results of the abundance of reads per kilobase million (RPKM) of the indicated metabolic modules (left panel) and the indicated species (lower panel) of the 2 groups. ∗P < .05, ∗∗P < .01. Circles on the grid indicate the distribution of the indicated metabolic modules across the indicated species. Raw data are listed in Tables 3 and 4.

Table 3.

Detailed Information on Differential Metabolic Modules Shown in Figure 6

Differential SGBs	Abundance of RPKM		P value, Wilcoxon.test
Taxonomy	Mean_CTR	Mean_E(L)	CTR vs E (L)
Akkermansia muciniphila	0.57570553	9.21199704	.009023439
Alistipes shahii	113.491262	21.8957294	.009023439
Desulfovibrio sp000403945	1.71998862	0.370976708	.009023439
Intestinimonas timonensis	0.728333714	0.252604656	.009023439
Lachnospiraceae bacterium COE1 sp002490665	0.52559075	13.71808266	.009023439
Lachnospiraceae bacterium COE1 sp009774375	9.70603733	0.084473268	.009023439
Lachnospiraceae bacterium TF01-11 sp003612285	11.73399908	0.140281054	.009023439
Muribaculaceae bacterium CAG-485 sp002361215	1.15177492	3.63277784	.009023439
Anaeroplasmataceae bacterium UMGS268 sp900540705	2.44960722	0.068656683	.016293604
Bacteroides fluxus	2.21301258	0.764452468	.016293604
Borkfalkiaceae bacterium UBA11940 sp003503945	1.115836426	0.166568287	.016293604
Haliangiaceae bacterium WYBA01 sp011526095	2.407987544	0.075637433	.016293604
Lachnospiraceae bacterium CAG-81 sp009917545	0.437356588	0.158290836	.016293604
Muribaculaceae bacterium CAG-873 sp009775225	3.99041358	15.525813	.016293604
Prevotellamassilia sp009775705	7.99422466	21.8167154	.016293604
Ruthenibacterium lactatiformans	0.312945174	0.77691509	.016293604
Alphaproteobacteria bacterium CAG-495 sp001917125	0.165737711	1.082254834	.028280123
Lachnospiraceae bacterium COE1 sp000403335	1.055863904	0.160580652	.028280123
Lachnospiraceae bacterium COE1 sp003513705	3.00083407	0.116323033	.028280123
Lachnospiraceae bacterium UBA3282 sp003611805	0.907807342	0.115914114	.028280123
Lachnospiraceae bacterium UBA3282 sp009774585	0.311735008	0.096850728	.028280123
Anaerotruncus massiliensis	1.0592687	0.445136368	.047201768
Bacteroidales bacterium RC9 sp009774765	1.193805752	0.434070264	.047201768
Borkfalkiaceae bacterium UBA11940 sp013316055	2.497212334	0.139497368	.047201768
Duncaniella muris	3.97091308	9.28539876	.047201768
Dysosmobacter sp009774015	0.807467778	0.413691016	.047201768
Eubacterium_J plexicaudatum	2.185886902	0.135572338	.047201768
Flavonifractor sp002159455	4.57440864	2.35495512	.047201768
Lachnospiraceae bacterium UBA3282 sp002492525	0.239936142	3.650466316	.047201768
Lachnospiraceae bacterium UBA7050 sp002493905	0.281436992	0.136938348	.047201768
Lawsonibacter sp000492175	1.655552554	0.566718656	.047201768
Lawsonibacter sp900550705	2.374367074	0.720928248	.047201768
Muribaculaceae bacterium CAG-485 sp009775365	1.623231348	6.32590564	.047201768
Oscillospiraceae bacterium UMGS1872 sp009774055	1.022430308	0.372877676	.047201768
Acetatifactor sp902796105	0.170364398	0.083191972	.047201768

Table 4.

Detailed Information on Significantly Different SGBs Shown in Figure 6

Differential modules	Abundance of modules		P value, Wilcoxon.test
Metabolic modules	Mean_CTR	Mean_E(L)	CTR vs E (L)
Aspartate degradation I	372.8434876	337.0511632	.009023439
Butyrate production I	175.9523341	140.2537107	.009023439
Butyrate synthesis I	175.9523341	140.2537107	.009023439
Cortisol degradation	181.5706266	131.0424522	.009023439
Galacturonate degradation I	225.0027935	143.0568704	.009023439
Hydrogen metabolism	271.427465	170.7852937	.009023439
Lactate production	206.4830665	108.7775012	.009023439
Maltose degradation	115.9746571	23.28622766	.009023439
Methionine degradation I	278.1402281	219.1124698	.009023439
Pectine degradation II	119.7796058	56.18301677	.009023439
Pentose phosphate pathway	300.9784602	274.1876268	.009023439
PUFAs synthesis (AA, EPA, DHA)	323.5861714	260.4881067	.009023439
Arabinose degradation	2.524747588	0.861303196	.016293604
Lactaldehyde degradation	185.2069177	118.0169728	.016293604
Proline degradation	7.700087326	14.8666858	.016293604
Sulfate reduction (dissimilatory)	4.587931907	1.541854264	.016293604
Arabinoxylan degradation	324.9772108	279.7742396	.028280123
Arginine degradation IV	11.15676632	4.169898375	.028280123
Bifidobacterium shunt	26.2122084	10.85036241	.028280123
4-aminobutyrate degradation	1.655552554	0.566718656	.047201768
Acetate degradation	94.54559176	122.1342652	.047201768
Acetate to acetyl-CoA	94.54559176	122.1342652	.047201768
Aspartate degradation II	67.55092745	91.31731572	.047201768
Glycolysis (pay-off phase)	266.4023123	224.0844715	.047201768
Glycolysis (preparatory phase)	381.2240782	360.0024597	.047201768
Lysine degradation II	3.798591702	25.0004255	.047201768
p-Cresol synthesis	166.7180344	230.2394621	.047201768
Rhamnose degradation	157.2640078	114.7163359	.047201768
S-Adenosylmethionine (SAM) synthesis	382.0185457	363.8245029	.047201768
Sucrose degradation II	20.48289052	9.172510027	.047201768
Tryptophan degradation I	170.9854941	107.0893819	.047201768
Tryptophan degradation	170.9854941	107.0893819	.047201768
Xylose degradation	202.7636361	131.357596	.047201768

Functional PAR₁ Expression in the Cultured Intestinal Myofibroblasts

In the colon mucosa of the patients with Crohn’s disease, some of the PAR₁-positive cells were also positive for αSMA (Figure 7A). The cultured myofibroblasts established from the colon of Crohn’s disease patients were used to further explore the functional role of PAR₁ in myofibroblasts (Figure 7B and C). Thrombin (1 U/mL) and a PAR₁-activating peptide TFLLR-NH₂ (10 μmol/L) induced a transient elevation of the cytosolic Ca²⁺ concentrations ([Ca²⁺]_i), presumably because of the Ca²⁺ release from the store sites, in the absence of extracellular Ca²⁺. Subsequent replenishment of 2 mmol/L extracellular Ca²⁺ induced a peak and sustained elevation presumably due to Ca²⁺ influx from extracellular space. However, the level of Ca²⁺ influx component obtained with thrombin and TFLLR-NH₂ did not differ significantly from that seen without any stimulation. The pretreatment with 1 μmol/L E5555 abolished the transient elevation of [Ca²⁺]_i induced by thrombin and TFLLR-NH₂ and significantly inhibited the sustained components (Figure 7D–G). In the cells transfected with 3 different PAR₁-targeted small interfering RNAs (siRNAs), thrombin-induced transient [Ca²⁺]_i elevation was significantly inhibited (Figure 7H and I). The Ca²⁺ influx component was slightly but significantly inhibited by 1 of 3 siRNAs targeted PAR₁ (Figure 8). The degree of suppression of the thrombin-induced Ca²⁺ release seen with 3 siRNAs correlated with their efficacy of knockdown of the PAR₁ expression (Figure 7H and I, Figure 8).

Figure 7.

Functional PAR₁expression in cultured intestinal myofibroblasts. (A) Co-immunostaining for PAR₁ and α-smooth muscle actin (αSMA) in colonic mucosa of patients with Crohn’s disease. Nuclei were labeled with To-Pro-3 (blue). White arrows indicate double-positive cells for PAR₁ and αSMA. (B) Representative immunoblots of PAR₁ in intestinal myofibroblast cells, transfected with negative control siRNA (NCsi) and PAR₁-targeted siRNA #1 (PAR₁si). (C–I) Representative recordings (C, D, F, and H) and summaries (E, G, and I; n = 4) of changes in cytosolic Ca²⁺ concentrations ([Ca²⁺]_i) induced by 1 U/mL thrombin and 10 μmol/L PAR₁-activating peptide, TFLLR-HN₂, in intestinal myofibroblast cells without (C–G) and with transfection of siRNA (H and I). When pretreated, 1 μmol/L E5555 was added 5 minutes before and during stimulation with thrombin and TFLLR-NH₂. Transient elevation of [Ca²⁺]_i seen in absence of extracellular Ca²⁺ was evaluated as a Ca²⁺ release component, and sustained elevation of [Ca²⁺]i seen after replenishment of extracellular Ca²⁺ was evaluated as a Ca²⁺ influx component. ∗P < .05, ∗∗∗P < .001, ns, not significant.

Figure 8.

PAR₁knockdown in the cultured intestinal myofibroblast. (A) Quantitative polymerase chain reaction analysis of knockdown efficiency of 3 siRNAs targeted PAR₁ (PAR₁si-1, PAR₁si-2, PAR₁si-3) and negative control siRNA (NCsi) in the cultured intestinal myofibroblasts (n = 4). (B–D) Representative recordings of changes in cytosolic Ca²⁺ concentrations ([Ca²⁺]i) in intestinal myofibroblast cells, transfected with NCsi, PAR₁si-2, and PAR₁si-3 (n = 4). ∗∗∗P < .001.

E5555 Suppressed the Phosphorylation of ERK and STAT3 in the Cultured Intestinal Myofibroblasts

Seeing that E5555 inhibited inflammation, carcinogenesis, and epithelial proliferative activity in the AOM/DSS model mice, the effects of E5555 on the key signals related to inflammation and cell proliferation were investigated using the cultured intestinal myofibroblasts. The stimulation with 1 U/mL thrombin for 5 minutes significantly increased the phosphorylation level of extracellular signal-regulated kinase (ERK)1/2. The ERK1/2 phosphorylation was inhibited by pretreatment with 1 μmol/L E5555 and siRNA-mediated knockdown of PAR₁ (Figure 9A–D). Thrombin had no significant effect on the phosphorylation of STAT3; however, knockdown of PAR₁ expression significantly decreased the basal and thrombin-stimulated phosphorylation of ERK1/2 (Figure 9E and F). IL6 significantly increased the STAT3 phosphorylation, whereas 1 μmol/L E5555 prevented the IL6-induced, but not basal, phosphorylation of STAT3 (Figure 9G and H, Figure 10).

Figure 9.

Effect of E5555 on phosphorylation of ERK1/2 and STAT3 in cultured intestinal myofibroblasts. (A and B) Representative immunoblot image (A) and summary (B; n = 4) showing effects of 1 U/mL thrombin and 1 μmol/L E5555 on phosphorylation of ERK1/2 in cultured intestinal myofibroblasts. (C–F) Representative immunoblot image (C and E) and summaries (D and F; n = 4) showing effect of 1 U/mL thrombin on phosphorylation of ERK1/2 (C and D) and STAT3 (E and F) in cultured intestinal myofibroblasts transfected with negative control (NCsi) and PAR₁-targeted (PAR₁si) siRNA. (G and H) Representative immunoblot image (G) and summary (H; n = 4) showing effects of 5 ng/mL IL6 and 1 μmol/L E5555 on phosphorylation of STAT3 in cultured intestinal myofibroblasts. The raw data are shown in Figure 12, Figure 13, Figure 14, Figure 15∗P < .05, ∗∗P < .01, ∗∗∗P < .001.

Figure 10.

Effects of E5555 on inflammatory cytokines and phosphorylation of STAT3 in the cultured intestinal myofibroblasts. (A–C) The mRNA expressions of IL1β, IL6, and IL8 in cultured intestinal myofibroblasts treated with 1 U/mL thrombin or 1 μmol/L E5555 (n = 4). (D) Immunoblot image (C, n = 1) showing effects of 5 ng/mL IL6, 1 U/mL thrombin, and 1 μmol/L E5555 on phosphorylation of STAT3 in cultured intestinal myofibroblasts. ∗P < .05, ∗∗P < .01, ∗∗∗P < .001.

Secretion of Functional Thrombin From Caco-2 Cells

Thrombin was detected in the colon mucosa of the AOM/DSS model mice (Figure 2I). To explore the possible source of thrombin in mucosa, the ability of epithelial cells to secrete thrombin was investigated using Caco-2 cells. The conditioned media of Caco-2 cells contained immunoreactive bands that represent prothrombin, meizothrombin, and α-thrombin (Figure 11A). The conditioned media cleaved the extracellular region of PAR₁ (PAR₁E) that contains the thrombin cleavage site and yielded 2 major cleavage products, which were similar to those seen with 1 U/mL thrombin. Thrombin completely cleaved PAR₁E within 5 minutes, whereas the conditioned media took 50 minutes to give a similar effect (Figure 11B and C). The conditioned media induced a significant increase in [Ca²⁺]_i in the absence of extracellular Ca²⁺ in the cultured myofibroblasts; however, it has no significant effect on the increase in [Ca²⁺]_i after replenishment of extracellular Ca²⁺, as in the case of thrombin. The conditioned media–induced [Ca²⁺]_i elevation seen in the absence, but not presence, of extracellular Ca²⁺ was significantly inhibited by pretreatment with 1 μmol/L E5555 (Figure 11D–F).

Figure 12.

Original immunoblot images and data source forFigure 9A.

Figure 13.

Original immunoblot images and data source forFigure 9C.

Figure 14.

Original immunoblot images and data source forFigure 9E.

Figure 15.

Original immunoblot images and data source forFigure 9G.

Figure 11.

Secretion of functional thrombin from Caco-2 cells. (A) Representative immunoblot image of detection with anti-thrombin antibody in conditioned medium of Caco-2 cells. (B) Schematic presentation of the recombinant protein containing a part of the extracellular region of PAR₁ (PAR₁E), which contains the thrombin cleavage site, and a C-terminal GST tag. (C) Representative image of SDS-PAGE gel showing time course of cleavage of PAR₁E-GST by conditioned media of Caco-2 cells. Thrombin digestion was shown as a positive control. (D–F) Representative recordings (D and E) and summaries (F) showing effect of conditioned media of Caco-2 and 1 μmol/L E5555 on cytosolic Ca²⁺ concentrations ([Ca²⁺]_i). Transient elevation of [Ca²⁺]_i seen in absence of extracellular Ca²⁺ was evaluated as Ca²⁺ release, and sustained elevation of [Ca²⁺]i seen after replenishment of extracellular Ca²⁺ to 2 mmol/L was evaluated as Ca²⁺ influx. ∗∗∗P < .001.

Discussion

The present study demonstrated for the first time that PAR₁ antagonist E5555 exerted anti-inflammatory and anti-carcinogenic effects in AOM/DSS model. The inhibition of inflammatory cytokines, fibrosis, and epithelial proliferative activity by E5555 was observed in mice colon tissue, with further insights into the underlying cellular mechanisms obtained through the analysis of cultured intestinal myofibroblasts derived from patients with IBD. E5555 treatment inhibited the thrombin-induced Ca²⁺ signaling and ERK phosphorylation, which were also corroborated by PAR₁ knockdown. PAR₁ knockdown decreased the basal level of STAT3 phosphorylation, whereas E5555 inhibited IL6-induced STAT3 phosphorylation. The analysis of the gut microbiome revealed that E5555 caused a shift of β-diversity in a new direction that was different from both control and AOM/DSS model. The detailed analysis of bacterial species and related metabolic module suggested an intriguing contribution of gut microbiota to the pathogenesis in AOM/DSS model as well as the therapeutic effects of E5555. Our findings suggest that PAR₁ antagonist could be a potential strategy for mitigating intestinal inflammation and related carcinogenesis in IBD patients.

It remains unclear that PAR₁ of which cell type contributes to pathogenesis in AOM/DSS model and the therapeutic effects of E5555. PAR₁ is widely expressed in various types of cells, including fibroblasts, epithelial cells, smooth muscle cells, endothelial cells, and immune cells within the intestinal tract, whereas PAR₁ is known to induce cell proliferation, production of inflammatory cytokines, proinflammatory phenotypic conversion of the cells, and impairment of intestinal epithelial barrier function, depending on cell types.^19,25, 26, 27 It is readily conceivable that there are multiple targets for E5555 exerting the therapeutic effects. At least, the intestinal myofibroblasts of the IBD patients express the functional PAR₁, which mediates the signal of cell proliferation and inflammation. The analysis of mRNA expression revealed the higher levels of PAR₁ in tumor area than in non-tumor area. Therefore, E5555 might have directly inhibited the development of tumor by acting on the PAR₁ of tumor cells in addition to mitigating the inflammation. Such direct inhibition of epithelial cell proliferation might be related to the decrease in the PCNA-positive cells in AOM/DSS model. Because of a lack of reliable antibody for mice PAR₁, the histochemical analysis of the PAR₁-expressing cells was halted. The contribution of other cell types such as macrophage remains elusive.

The activity of proteinases including thrombin, which potentially act as agonist of PAR₁, is elevated in IBD patients.^20,28 PAR₁ is expressed in fibroblasts, epithelial cells, smooth muscle cells, endothelial cells, and immune cells within the human intestinal tract.¹⁸ The expression of PAR₁ is up-regulated in IBD patients.^19,29,30 Therefore, the therapeutic effects of E5555 observed in the AOM/DSS model in the present study could be extrapolated to the therapeutic application of PAR₁ antagonism to the treatment of IBD patients. In fact, the ameliorating effects of PAR₁ antagonists on the damaged intestinal mucosa and increased level of thrombin at the site of inflammation were observed in clinical studies.^18,19 The findings of the clinical studies are consistent with those obtained with the animal experiments in the present study. The present study provides further insights into the potential translational implications to clinical applications.

The therapeutic effect of E5555 may be not only dependent on but also independent of PAR₁, because E5555 inhibited the IL6-induced STAT3 phosphorylation in the cultured myofibroblasts. The inhibitory effects of E5555 on the [Ca²⁺]_i increases induced by thrombin and PAR₁-activating peptide were recapitulated by PAR₁ knockdown by siRNA. This observation verifies the PAR₁ antagonism of E5555. However, E5555 has also been reported to induce apparent off-target effects, such as inhibition of the vascular contraction induced by membrane depolarization without PAR₁ stimulation in rabbits^31,32 and prolongation of the QT interval in humans.³³ Such off-target effect might contribute to anti-inflammatory effect of E5555; however, precise mechanism underlying the inhibition of IL6-induced STAT3 phosphorylation remains to be elucidated.

It remains elusive precisely how PAR₁ is activated in the AOM/DSS model as well as in IBD patients. Luminal proteinases primarily originated from host digestive enzymes, proteinases produced by the gut microbiota, mucosal proteinases released by ectopically expressed clotting factors, and proteinases secreted from resident or infiltrating immune cells are possible sources of PAR₁ agonist.²⁵ PAR₁ is activated by various proteinases, including thrombin, a high affinity agonist.²⁸ The serine protease activity was 10-fold higher in IBD fecal samples compared with healthy controls.³⁴ Thrombin activity significantly increased in the Crohn’s disease patients and the rat enteritis model.¹⁹ Inflammation of the intestinal mucosa is accompanied by different degrees of mucosal destruction and hemorrhage, which also serves as a source of thrombin in addition to the local production of thrombin by intestinal epithelial cells.³⁵ Furthermore, the IBD patients show an increased coagulability.³⁶ The present study demonstrated the deposition of immunoreactive thrombin in mucosa and increased mRNA expression of thrombin in both non-tumor and tumor areas. Furthermore, the present study demonstrated that the conditioned media obtained from Caco-2 cells contained immunoreactive prothrombin, meizothrombin, α-thrombin, and γ-thrombin as previously reported¹⁹ and exhibited a proteinase activity to cleave the extracellular region of PAR₁ in the same pattern as that seen with purified thrombin. The local production of thrombin can contribute to pathogenesis of inflammation and carcinogenesis, whereas thrombin produced by tumor cells in turn contribute to exacerbation of pathology of IBD. Furthermore, the gut microbiota is known to secrete proteinases capable of activating PARs, leading to inflammation and potentially contributing to cancer development.^37,38 However, the metagenome analysis did not provide useful information regarding the proteinase-producing species.

The results of metagenome analysis of gut microbiota also suggest the contribution of changes in diversity of microbiota to the pathogenesis as well as the therapeutic effect of E5555. The AOM/DSS group of mice had a significantly lower abundance of the core setter Lachnospiraceae bacterium, which is an important short-chain fatty acids (SCFAs)–producing family in the gut.^39,40 SCFAs are the key substances to maintain the normal function of gut, the morphology of intestinal epithelial cells, and epithelial barrier function and inhibit the expression of proinflammatory factors IL1β, IL6 and TNFα.^41,42 SCFAs also prevent the formation of colitis-associated cancer by inhibiting nuclear factor kappa B activity and cyclooxygenase 2 expression.⁴³ Moreover, short-term treatment with E5555 significantly reduced Alistipes sp. Alistipes is a potential pathogen in colorectal tumors, which could promote the development of colitis-associated cancer through IL6/STAT3 pathway.⁴⁴ Long-term treatment with E5555 significantly increased Muribaculaceae sp. Muribaculaceae has been proved to help intestinal barrier function and anti-inflammation.⁴⁵ It is worth noting that Alistipes sp. and Muribaculaceae sp. are involved in the regulation of amino acid metabolism, because amino acids metabolism is a key factor affecting molecular biosynthesis and cancer cell proliferation.⁴⁶ Overall, the formation of a new balance of gut microbiome may underlie the therapeutic effects of E5555 in improving the intestinal inflammation and colitis-associated carcinogenesis.

DSS disrupts intestinal mucosal barriers and induces intestinal inflammation by intervening in the metabolism of gut microbiota, whereas changes in microbiota diversity might be closely associated with the therapeutic effects of E5555 in AOM/DSS model mice. Among the numerous strains altered in the AOM/DSS group, the abundance of the 2 strains (Oscillospiraceae bacterium UMGS1872 sp009774055 and Flavonifractor sp002159455) are noted to be significantly decreased by both long-term and short-term E5555 treatment. Oscillospiraceae family is involved in damaging intestinal mucosa,⁴⁷ whereas Flavonifractor family is found higher in patients with Crohn’s disease and promotes the development of colorectal cancer.^48,49 The decrease in their abundance might be related to the therapeutic effect of E5555. Moreover, the treatment with E5555 per se caused a significant difference in β-diversity and in the 33 metabolic pathways (Figure 6). Of note were 2 pathways that were particularly associated with the development of intestinal inflammation and cancer: a pathway of tryptophan degradation I, which contribute to facilitating colorectal carcinogenesis,⁵⁰ and a pathway of hydrogen metabolism, which is associated with immunity, inflammation, and oxidative stress and protective role in ulcerative colitis.⁵¹ However, pinpointing which specific strains altered by E5555 are closely associated with its therapeutic effects remains elusive.

In conclusion, the present study proposes a potential of E5555, a PAR₁ antagonist, to prevent and inhibit IBD and the associated carcinogenesis. The therapeutic effects may not only be dependent on but also independent of PAR₁ antagonism. The alteration of diversity of gut microbiota is also suggested to contribute to the therapeutic effects of E5555. The therapeutic effects of E5555 were observed even in the short-term treatment protocol, which started 1 week after the completion of AOM and DSS treatment. The intestinal inflammation reaches a substantial level at this time point, whereas the carcinogenesis requires further incubation to exhibit an apparent development of tumors.⁵² Therefore, PAR₁ antagonist primarily inhibits intestinal inflammation, thereby secondary to prevent carcinogenesis. However, the findings of the present study suggest the capability of PAR₁ antagonist in inhibiting carcinogenesis by directly affecting the tumor development in the later stage. Such inhibitory effect of PAR₁ antagonist remains to be established when translating the findings of the present study into clinical application.

Methods

Animal Care

The experimental protocol was approved by the Kagawa University Institutional Animal Care and Use Committee and was performed in compliance with the guidelines for conducting animal experiments at Kagawa University (approval number: 21679).

Chemicals and Solutions

Azoxymethane (A5486; Sigma-Aldrich Co, St Louis, MO) and dextran sulfate sodium (M.W. 36,000–50,000; CAS Number: 9011-18-1; MP Biomedicals, Inc, Santa Ana, CA) were used to establish a colitis-associated carcinogenesis mouse model. TFLLR-NH₂, a PAR₁-activating peptide, was synthesized by Eurofins Genomics (Tokyo, Japan). Thrombin (bovine plasma; T7513) was purchased from Sigma-Aldrich. Calcium ionomycin was purchased from LKT Laboratories (St Paul, MN). Atopaxar (E5555) was purchased from Axon Medchem (Groningen, the Netherlands). Fura-2 acetoxymethyl ester (fura-2-AM) was purchased from Dojindo (Kumamoto, Japan). Calcium ionomycin (I5753) was purchased from LKT Laboratories. Phosphate-buffered saline (PBS) was composed of 8 mmol/L Na₂HPO₄·12H₂O, 2.1 mmol/L NaH₂PO₄·2H₂O, and 140 mmol/L NaCl. Hepes-buffered saline solution (HBS) was composed of 10 mmol/L Hepes, pH 7.4, 135 mmol/L NaCl, 5 mmol/L KCl, 1 mmol/L CaCl₂, 1 mmol/L MgCl₂, and 5.5 mmol/L D-glucose.

Animal Model and Experimental Conditions

The design protocol for animal experiments is shown in Figure 2A. Six-week-old female C57BL/6NCrSlc mice (SLC Inc, Hamamatsu, Japan) were used. The mice were able to accommodate for more than 1 week under a 12-hour dark-and-light cycle at 25°C room temperature with ad libitum access to food and water until each mouse weighed >20 g. The mice were randomly divided into 5 experimental groups (10 mice per group): control group, E5555 group (E), AOM and DSS molding group (A/D), AOM/DSS with E5555 long-term treatment group (A/D+E(L)), and AOM/DSS with E5555 short-term treatment group (A/D+E(S)). The control group was intraperitoneally administered saline with normal food and water. Group E was administered saline intraperitoneally on day 0 and fed regular food containing E5555 and normal drinking water. Group A/D was administered AOM (12 mg/kg body weight) intraperitoneally on day 0 and fed regular food and regular drinking water except for weeks 2, 4, and 6, when they were fed drinking water containing 2% DSS. A/D+E(L) and A/D+E(S) groups were prepared in the same protocol as that used for the A/D group except that they were fed the food containing E5555 throughout 21 weeks of the experimental protocol (A/D+E(L)) and during the last 14 weeks of the experimental protocol (A/D+E(S)), respectively (Figure 2A).

Stool samples (5 samples for each group at each time point) were collected for metagenome analysis of gut microbiota on day 0 and at the end of the experimental protocol. Samples were then stored at –80°C until DNA was extracted. The stool consistency was scored using the following scale: 0, normal; 1, slightly loose stool; 2, severely loose stool; 3, diarrhea. At the end of the experimental protocol, mice were euthanized by cervical dislocation. The intestine between the cecum and anus was excised, and its length was measured. The intestine was then opened longitudinally, and the number of polyps was counted. Portions of the colon were placed in a lysis buffer FARB from the Tissue Total RNA Mini Kit (FAVORGEN, Taipei, Taiwan, China) for RNA extraction or quenched in liquid nitrogen for protein extraction and then stored at –80°C until the next procedure. The remaining intestine was fixed on a rubber sheet using a needle, fixed in 10% formalin-containing neutral buffer for 24 hours.

Histology Evaluation

The colon tissues of mice, fixed in 10% formalin, were embedded in paraffin. The human colon specimens were obtained from inflammation areas of the patients with Crohn’s disease with consent under colonoscopies (Fukuoka University Hospital Ethics Committee approval number: 12-9-11).⁵³ After deparaffinization and rehydration, the 3-μm-thick tissue sections were subjected to hematoxylin-eosin (HE), Masson’s trichrome, and immunohistochemical staining.

The degrees of inflammation and fibrosis of mice colon were scored according to the criteria shown in Tables 5 and 6 by a pathologist observing HE and Masson’s trichrome staining, respectively, at ×200 magnification in a blinded manner.

Table 5.

Inflammation Score Criteria

Item	Abbreviation	Score
		0	1	2	3	4
Inflammation	I	None	Mild	Moderate	Severe
Extend	E	None	Mucosa and submucosa	Transmural
Regeneration	R	Complete regeneration	Almost complete regeneration	Regeneration with crypt deletion	Surface epithelium not intact	No tissue repair
Crypt damage	C	None	1/3 of basal damaged	2/3 of basal damaged	Only surface epithelium intact	Entire crypt and epithelium lost
Percent involvement	P	None	1%–25%	26%–50%	51%–75%	76%–100%

The total histologic score was calculated as I+E+R+C+P.

Table 6.

Fibrosis Score Criteria

Score	Item
1	Mild fibrosis	Focal mucosal/submucosal collagen deposition without architectural distortion
2	Moderate fibrosis	Significant mucosal/submucosal collagen deposition with modest distortion of the mucosal/submucosal architecture but without obscuring the mucosal/submucosal border
3	Severe fibrosis	Extensive mucosal/submucosal collagen deposition with marked architectural distortion obscuring the mucosal/submucosal border

Immunofluorescence Staining

Formalin-fixed, paraffin-embedded specimens obtained from the mice and the patients with Crohn’s disease were deparaffinized and rehydrated, followed by a procedure of heat antigen retrieval with citrate acid (pH 6) for 20 minutes at 100°C. Then, the specimens were blocked with a blocking One Histo solution (Nacalai Tesque, Kyoto, Japan) for 10 minutes at room temperature and then incubated with primary antibody overnight at 4°C. The subsequent 1-hour incubation with secondary antibodies was then followed by nuclei counterstain using TO-PRO-3 iodide (642/661) (1/1000 dilution in PBS; Invitrogen, Thermo Fisher Scientific). All antibodies used in the present study are listed in Table 7.

Table 7.

List of the Antibodies

Protein	Dilution	Species	Catalog number	Purpose
Actin, α-Smooth Muscle	1:1000	Sigma-Aldrich Co LLC	A2547	IH
β-Actin	1:3000	FUJIFILM	010-27841	WB
ERK1/2	1:1000	BioLegend	686901	WB
Mouse IgG Alexa Fluor 488	1:1000	Cell Signaling	#4408s	IH
Mouse IgG Alexa Fluor 546	1:1000	Thermo Fisher	A11018	IH
Mouse IgG HRP	1:3000	Sigma-Aldrich Co LLC	A4416	WB
PAR1/Thrombin Receptor	1:500	Abcam plc	ab32611	IH WB
Phospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204)	1:1000	Cell Signaling	#4370	WB
Phospho-Stat3 (Tyr705) (3E2)	1:1000	Cell Signaling	9138S	WB
Proliferating Cell Nuclear Antigen	1:200	Dako	M0879	IH
Rabbit IgG Alexa Fluor 488	1:1000	Cell Signaling	#4412s	IH
Rabbit IgG Alexa Fluor 546	1:1000	Thermo Fisher	A11071	IH
Rabbit IgG HRP	1:3000	Sigma-Aldrich Co LLC	A6154	WB
Rat IgG HRP	1:3000	BIO-RAD	STAR72	WB
Stat3 (124H6)	1:1000	Cell Signaling	9139S	WB
Thrombin (F-1)	1:250	Santa Cruz Biotechnology	sc-271449	IH WB

Cell Culture and Medium Collection

Intestinal myofibroblast cells, isolated from patients with Crohn’s disease as described,⁵³ were cultured in SmBM media (Lonza, Basel, Switzerland) supplemented with 5% fetal bovine serum (cat: CC-4102D), human fibroblast growth factor-B (cat: CC-4068D), 0.1% gentamicin sulfate/amphotericin-B (CA-1000, cat: CC-4081D), 0.1% recombinant human insulin (cat: CC-4021D), and epidermal growth factor (cat: CC-4230D) at 37°C in a humidified 5% CO₂ environment. Intestinal myofibroblast cells were plated at 4000 cells/well in 6-well plates until they reached 80%–90% confluency and then incubated in serum-free culture for 24 hours before stimulation with 1 U/mL thrombin to evaluate the effects of E5555 on phosphorylation of ERK1/2 and STAT3 (Figure 9).

Caco-2 cells, a human colon carcinoma cell line, were purchased from the RIKEN BRC (RCB-0988) and were cultured in low-glucose Dulbecco modified Eagle medium (Sigma-Aldrich) supplemented with 1% nonessential amino acids, 1% penicillin/streptomycin, and 10% fetal bovine serum (Biological Industries, Kibbutz Beit HaEmek, Israel) at 37°C in a humidified 5% CO₂ environment. To obtain the conditioned media of Caco-2 cells, the cells were plated on the upper chamber of the Trans-well dish, cultured for 21 days with medium renewal every 2–3 days, and then incubated with serum-free medium for 24 hours before collection of the conditioned media. The conditioned media were concentrated approximately 10-fold with Pierce Protein Concentrator PES (10 K MWCO, 5–20 mL, 24pk, Thermo Fischer Scientific, UK) by centrifuging with centrifuge CR22N (Himac, China) at 5000g for 20 minutes at 4°C.

Fura-2 Fluorometry for Measurement of Cytosolic Ca²⁺ Concentrations

Intestinal myofibroblast cells were seeded at 2.0 × 10⁴ cells/dish in 35-mm dishes and cultured for 3 days. On day 4, cells were subjected to fura-2 fluorometry. The cells were loaded with fura-2 by incubating them in SmBM containing 5 μmol/L fura-2/AM for 60 minutes under cell culture conditions. After fura-2 loading, the dishes were set on the stage, which was kept at 37°C, and the cells were washed thrice and equilibrated in 1 mL HBS at 37°C. Changes in the intensities of fura-2 fluorescence at 340 nm (F340) and 380 nm (F380) excitation and their ratio (F340/F380) were monitored using a front-surface fluorometer CAM-230-OF1 (JASCO, Tokyo, Japan), as previously described.⁵⁴ In this protocol, cells were first exposed to a Ca²⁺-free HBS containing 2 mmol/L EGTA for 10 minutes and then stimulated with 1 U/mL thrombin or 10 μmol/L TFLLR-NH₂. After the transient [Ca²⁺]_i elevation returned to the pre-stimulation level, extracellular Ca²⁺ was replenished to 2 mmol/L in the continuous presence of stimulation. After sustained [Ca²⁺]_i elevation was recorded for 10 minutes, the cells were exposed to 50 μmol/L ionomycin in the presence of 1 mmol/L extracellular Ca²⁺. The transient [Ca²⁺]_i elevation seen in the absence of extracellular Ca²⁺ and the sustained [Ca²⁺]_i elevation seen after the replenishment of extracellular Ca²⁺ were evaluated as Ca²⁺ release and Ca²⁺ influx components, respectively. The area under the curve of the Ca²⁺ release component was quantified from the time of stimulation to the time when [Ca²⁺]_i returned to the pre-stimulation levels. The area under the curve of the Ca²⁺ influx component was quantified using a 10-minute trace, starting from the time of replenishment of extracellular Ca²⁺. The ionomycin response was recorded for 10 minutes at the last of recording, and the area under the curve of ionomycin response was used for the normalization. The area under the curve was quantified using a JMP Pro15 software program.

Real-Time Polymerase Chain Reaction Analysis

RNA was purified using the Tissue Total RNA Mini Kit (Favorgen Biotech Corp, Taiwan, China). After measuring RNA concentration, reverse transcription was conducted using Prime Script RT Master Mix (Takara Bio Inc, Shiga, Japan). TaqMan Fast Advanced Master Mix (Applied Biosystems, Foster City, CA) was used for quantitative polymerase chain reaction on a MicroAmp Fast 96 well Reaction Plate (Applied Biosystems). The primer probe sets used are listed in Table 8.

Table 8.

List of the Quantitative Polymerase Chain Reaction Assays

Gene name	Vendor	Prime Time qPCR Assay no.
ACTB	IDT	Mm.PT.39a.22214843.g
f2	IDT	Mm.PT.58.5715422
f2r	Thermo Fisher	Mm00438851_m1
f2rl1	IDT	Mm.PT.58.6456606
f2rl2	IDT	Mm.PT.58.13455176
f2rl3	IDT	Mm.PT.58.45910106.g
il6	Thermo Fisher	Mm00446190_m1
ifng	Thermo Fisher	Mm01168134_m1
tgfβ1	Thermo Fisher	Mm01178820_m1
tnf	Thermo Fisher	Mm00443258_m1

Extraction of DNA and Metagenomic Analysis

DNA in fecal samples was extracted using the QIAamp Fast DNA Stool Mini kit (Qiagen GmbH, Hilden, Germany) in accordance with the manufacturer’s instructions. The quality, purity, and integrity of the extracted DNA were then measured by Nanodrop spectrophotometer, 1.0% agarose gel electrophoresis, and the Qubit dsDNA Assay Kit with a Qubit 2.0 fluorometer (Life Technologies, CA). High-quality DNA with DNA concentration greater than 20 ng/μL and OD260/280 between 1.8 and 2.0 were selected. The DNA library was constructed with NEBNext Ultra DNA Library Prep Kit for Illumina (New England Biolabs, Inc) according to the manufacturer’s instructions. Finally, the DNA libraries were sequenced on an Illumina NovaSeq platform to generate paired-end reads (Tianjin Novogene Technology Co, Ltd, Tianjin, China).

Quality Control of Metagenomic Data

A total of 50 fecal samples were shotgun sequenced, generating 316.89 Gb of high-quality paired-end reads (6.34 ± 0.83 Gb per fecal sample; range, 4.54–9.10 Gb). KneadData (http://huttenhower.sph.harvard.edu/kneaddata), Trimmomatic, and Bowtie2 (v2.3.5.1) were used to remove low-quality sequences, mouse contaminating reads, and adaptor sequences. Finally, the 240.36 Gb of clean data (4.81 ± 1.31 Gb per fecal sample; range, 3.53–6.24 Gb) remained for downstream analysis.

Metagenomic Assembly, Contig Binning, Genome Dereplication

To obtain bacterial SGBs from the metagenomic data set after quality control, MEGAHIT (HKU-BGI, China) was first used to assemble reads for each sample into contigs. Then, MetaBAT2 (DOE JGI, CA) was used to select bin contigs greater than 2000 bp to obtain metagenome-assembled genomes. Second, reads were mapped back to the corresponding contigs using BWA-MEM2,⁵⁵ and the read depth was calculated using Samtools (Sanger Institute, Hinxton, Cambridge, UK) and the jgi_summarize_bam_contig_depths function in MetaBAT2. Third, the levels of completeness and contamination of metagenome-assembled genomes were evaluated by CheckM (ACE, Queensland, Australia), and the metagenome-assembled genomes were classified as high quality (completeness ≥80%, contamination ≤5%); partial quality (completeness ≥50%, contamination ≤5%); and medium quality (completeness ≥70%, contamination ≤10%). Finally, dRep (v3.0.1; Jillian F. Banfield, UCB, USA) was used to cluster and obtain SGBs for the high-quality genomes with the parameter settings -pa 0.95 and -sa 0.95. This study yielded a total of 145 SGBs.

Taxonomic Annotation, Abundance, Prediction of Gut Metabolic Modules of SGBs

The SGBs were annotated using Kraken2 (JHU CS, USA) and NCBI nonredundant Nucleotide Sequence Database with default settings. The putative genes were searched by UniProt Knowledgebase (UniProtKB, release 2020.11) using the Blastp function of DIAMON with default options, using CoverM (https://github.com/wwood/CoverM) to calculate each SGB abundance with parameter “–min-read-percent-identity 0.95 –min-covered-fraction 0.4”. Gene abundance was expressed by reads per kilobase per million mapped reads. The predicted open reading frames for each SGB were compared with the Kyoto Encyclopedia of Genes and Genomes to predict the metabolic modules.

Statistical Analyses of Metagenomic Data

All statistical analyses were performed using the R software (v.4.2.1; University of Auckland, New Zealand.), and data were expressed as mean ± standard deviation. Shannon index and Chao 1 index were calculated using R packages (including vegan, ggpubr, dplyr, etc) to evaluate the changes in microbial α-diversity in fecal samples, and the principal coordinates analysis (based on Bray–Curtis distance) was used to evaluate the changes in microbial structure (β-diversity) in fecal samples. The difference in species and metabolic modules inter-group was calculated by Wilcoxon tests (cutoff level, P < .05). All graphical presentations were generated with R and Adobe Illustrator (AI) environment.

Deposition of Metagenome Data

The raw sequence data reported in this article have been deposited in the Genome Sequence Archive (Genomics, Proteomics & Bioinformatics 2021) in National Genomics Data Center (Nucleic Acids Res 2022), China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (GSA: CRA014331) that are publicly accessible at https://ngdc.cncb.ac.cn/gsa.

Transfection of siRNA

Intestinal myofibroblast cells were transfected with siRNA using lipofectamine RNAiMAX (Life Technologies, Carlsbad, CA) according to the manufacturer’s instructions. The cells were then subjected to fura-2 fluorometry or phosphorylation experiments 2 days after transfection. Knockdown efficiency was evaluated using real-time polymerase chain reaction 24 hours after transfection for immunoblot and 48 hours after transfection for fura-2 fluorometry (Dojindo, Kumamoto, Japan), as described above. The siRNA sequences used are listed in Table 9.

Table 9.

List of the siRNAs

Target	Sense strand(5′-3′)	Antisense strand (5′-3′)
PAR-1①	GAGGUAAGACUUAGUACU	CACAGAUAGUACUAAGUC
PAR-1②	GCAAGUAAAAUGGAUACC	UAGAGCAGGUAUCCAUUU
PAR-1③	GCCUCUAUCUUGCUCAUG	UGACUGUCAUGACCAAGA

Immunoblot Analysis

The cell lysates were prepared in RIPA buffer (50 mmol/L Tris-HCl, adjusted to pH 7.6 by Tris base, 150 mmol/L NaCl, 1% Nonidet P40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate with proteinase inhibitor cocktail; Nacalai Tesque, Kyoto, Japan). Protein was quantified using the Protein Assay Bicinchoninate Kit according to the manufacturer’s instructions (Nacalai Tesque), and protein content was adjusted to 0.3–0.5 μg/μL in 4× Laemmli Sample buffer (BIO-RAD, USA). Samples were heated at 95°C for 5 minutes before being used for Western blot analysis. An equal amount of protein (3–5 μg) was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on 8% polyacrylamide gels and transferred onto a polyvinylidene difluoride (0.45 μm, Sigma-Aldrich) membrane using the electroblotting system (BIO-RAD, Power Pac HC; Berkeley, CA). The membrane was blocked for 30 minutes with Blocking One (Nacalai Tesque) and then incubated with primary antibody diluted in the Can Get Signal Immunoreaction Enhancer solution 1 (Toyobo, Osaka, Japan) overnight at 4°C, and then incubated for 1 hour with secondary antibodies diluted in Can Obtain Signal Immunoreaction Enhancer solution 2 (Toyobo, Osaka, Japan) at room temperature. Immune complexes were detected using an ECL Western blot detection system (LAS4000, Fujifilm, Japan).

In Vitro Digestion of PAR₁E by Concentrated Caco-2 Culture Medium

The PAR₁E-GST protein was prepared by removing the SUMOstar-tag with SUMOstar protease (LifeSensors, Malvern, PA) from SUMO-PAR₁E-GST protein, prepared as previously described.⁵⁴ The PAR₁E-GST protein was diluted with a buffer consisting of 50 mmol/L Hepes, pH 7.4, 200 mmol/L NaCl, and 10% (wt/vol) glycerol at a final concentration of 10 μmol/L (0.49 mg/mL) and then preincubated for 5 minutes at 25°C. After preincubation, proteolytic digestion was initiated by adding the concentrated conditioned medium at 10-fold dilution or thrombin (1 U/mL). The reaction was terminated by adding p-amidinophenylmethylsulfonylfluoride at 50 μmol/L and then mixed with 1 volume of 2× Laemmli sample loading buffer. The digests were subjected to SDS-PAGE on 12% polyacrylamide gel and then visualized by staining with Bio-safe Coomassie G-250 stain (BIO-RAD).

To examine thrombin secretion from Caco-2 cells, a PAGE clean-up kit (06441, Nacalai Tesque) was used to concentrate the conditioned medium further 10 times. The immune-reactive thrombin was evaluated by immunoblot analysis as described above.

Statistical Analyses

The n number indicates the number of independent experiments or the number of animals. Differences in numerical variables among groups were evaluated using analysis of variance, followed by the Tukey–Kramer test for multiple comparisons. A value of P <.05 was considered significant. All statistical analyses were performed using JMP software (ver. 16.1.0; SAS Institute, Cary, NC), Image Studio Lite ver. 4.0 (LI-COR, Inc, NE), Image J (1.47 V, National Institutes of Health, Bethesda, MD), and Prism-GraphPad ver9.5.1 (GPS-2910207-RAVN-2A765). P <.05 was considered to indicate statistical significance.