Prostaglandin E2 Induces miR675-5p to Promote Colorectal Tumor Metastasis via Modulation of p53 Expression

Journal Pre-proof Prostaglandin E2 Induces MIR675-5p to Promote Colorectal Tumor Metastasis via Modulation of p53 Expression Bo Cen, Jessica D. Lang, Yuchen Du, Jie Wei, Ying Xiong, Norma Bradley, Dingzhi Wang, Raymond N. DuBois PII: DOI: Reference:

S0016-5085(19)41559-X https://doi.org/10.1053/j.gastro.2019.11.013 YGAST 63003

To appear in: Gastroenterology Accepted Date: 1 November 2019 Please cite this article as: Cen B, Lang JD, Du Y, Wei J, Xiong Y, Bradley N, Wang D, DuBois RN, Prostaglandin E2 Induces MIR675-5p to Promote Colorectal Tumor Metastasis via Modulation of p53 Expression, Gastroenterology (2019), doi: https://doi.org/10.1053/j.gastro.2019.11.013. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 by the AGA Institute

PGE2

COX1/2

AKT NF-κB

β-catenin

NF-κB β-catenin H19 mRNA

MIR675-5p

p53

Myc mRNA

Myc

Myc

RhoA

Cell Migration and Invasion

Gastroenterology

Prostaglandin E2 Induces MIR675-5p to Promote Colorectal Tumor Metastasis via Modulation of p53 Expression

Short title: PGE2 Promotes Metastasis via MIR675-5p

Bo Cen1#, Jessica D. Lang2#, Yuchen Du3, Jie Wei1, Ying Xiong1, Norma Bradley1, Dingzhi Wang1, and Raymond N. DuBois1, 4* 1

Department of Biochemistry and Molecular Biology, Medical University of South

Carolina, Charleston, Charleston, SC 29425. 2

Biodesign Institute of Arizona State University, Tempe, AZ85287, and Integrated

Cancer Genomics Division, Translational Genomics Research Institute (TGen), Phoenix, AZ 85004. 3

Biodesign Institute of Arizona State University, Tempe, AZ85287 (current address:

Department of Pediatrics-Oncology, Baylor College of Medicine, Houston, TX77030) 4

Department of Research and Division of Gastroenterology, Mayo Clinic, Scottsdale, AZ

85259

Grant Support: This work is supported by NIH R01 DK047297, CA184820, and P01 CA077839 (to R.N.D.). The project described is also supported by the NIH National Center for Advancing Translational Sciences (NCATS) through Grant Number UL1 TR001450 (to B.C.).

1

Abbreviations: PG, Prostaglandins; CRC, miRNA, microRNA; colorectal cancer; UTR, untranslated region; ChIP, chromosome immunoprecipitation; OE, over-expression; qPCR, quantitative polymerase chain reaction; CRISPR, clustered regularly interspaced short palindromic repeats. *Correspondence to: Raymond N. DuBois, MD. Ph.D., 601 Clinical Science Building, 96 Jonathan Lucas Street, Suite 601, Charleston, SC 29425, Tel: 843-792-2842 and Fax: 843-792-2967, [email protected]

Disclosures: The authors declare no competing interests.

Author contributions: R.N.D. supervised the entire study. B.C. and D.W. designed the experiments. B.C., J.D.L., Y.D., J.W., Y.X., and N.B. performed experiments. J.D.L. performed bioinformatics analysis. B.C. and J.D.L. wrote the manuscript. D.W. and R.N.D. edited the manuscript.

#These authors contribute equally.

2

BACKGROUND & AIMS: Prostaglandin E2 (PGE2) promotes colorectal tumor formation and progression by unknown mechanisms. We sought to identify microRNAs (miRNAs) might mediate the effects of PGE2 on colorectal cancer (CRC) development. METHODS: We incubated LS174T colorectal cancer cells with PGE2 or without (control) and used miRNA-seq technology to compare expression patterns of miRNAs. We knocked down levels of specific miRNAs or proteins in cells using small interfering RNAs or genome editing. Cells were analyzed by immunoblot, quantitative PCR, chromosome immunoprecipitation, cell invasion, and luciferase reporter assays; we measured gene expression, binding activity, cell migration and invasion, and transcriptional activity of transcription factors. NSG mice were given injections of LS174T cells and growth of primary tumors and numbers of liver and lung metastases were quantified and analyzed by histology. We used public databases to identify correlations in gene expression pattern with patient outcomes. RESULTS: We identified miRNA 675-5p (MIR675-5p) as the miRNA most highly upregulated by incubation of colorectal cancer cells with PGE2. PGE2 increased expression of MIR675-5p by activating expression of Myc, via activation of AKT, NF-κB, and β-catenin. PGE2 increased the invasive activities of cultured CRC cells. LS174T cells incubated with PGE2 formed more liver and lung metastases in mice than control LS174T cells. We identified a 3’UTR in the TP53 mRNA that bound MIR675-5p; binding resulted in loss of the p53 protein. Expression of MIR675-5p or its precursor RNA, H19, correlated with expression of cyclooxygenase-1 and cyclooxygenase-2 and shorter survival times of patients with CRC.

3

CONCLUSIONS: We found incubation of CRC cells with PGE2 to increase their invasive activity and ability to form liver and lung metastases in mice. PGE2 downregulates expression of p53 by increasing expression of MIR675-5p, which binds to and prevents translation of TP53 mRNA. These findings provide insight into the mechanisms by which PGE2 promotes tumor development and progression. Strategies to target the PGE2 might be developed for treatment of CRC. Key Words: gene regulation, tumor suppressor, post-transcriptional modification, inflammation.

4

Introduction Prostaglandins (PGs) are bioactive lipids that impact normal development, tissue homeostasis, inflammation, and cancer progression

1

. PGs are produced from

arachidonic acid by action of prostaglandin-endoperoxide synthases 1 (PTGS1) and/or prostaglandin-endoperoxide synthases 2 (PTGS2), also referred to as cyclooxygenase 1 (COX-1) and cyclooxygenase 2 (COX-2). COX-1 helps maintain gastric epithelial cytoprotection and homeostasis 2. COX-2, induced by inflammatory stimuli, hormones and growth factors, is elevated in inflammatory and proliferative diseases, such as cancer2. COX-2 levels are increased in multiple cancer types and are associated with decreased survival among cancer patients

3-5

. The activity of COX-1/2 can be inhibited

by nonsteroidal anti-inflammatory drugs (NSAIDs) which are known to reduce the risk of CRC 6. PGE2 is a commonly studied PG and the most abundant one found in various types of human malignancies and higher levels correlate with a poor prognosis

7-9

. A

urinary PGE2 metabolite (PGE-M) levels are associated with an increased risk of multiple human cancers

10

. PGE2 exerts its effects by binding to cell surface EP

receptors that belong to the family of G protein-coupled receptors. Each EP receptor activates distinct downstream signaling pathways that transduce a particular biologic effect. PGE2 has been shown to promote tumor epithelial cell proliferation, survival, and migration/invasion via multiple signaling pathways 1. However, all of the mechanisms by which PGE2 accelerates cancer formation, progression, and metastasis in complex living systems have not been fully elucidated. A growing number of reports suggest that the H19 long intergenic noncoding RNA (lincRNA) expression is linked to carcinogenesis

11

. H19 was found to be the precursor

5

of two distinct microRNAs (miRNAs), MIR675-5p and MIR675-3p suggested that these miRNAs partially confer functionality of H19

12

. It has been

12, 13

. The levels of

MIR675 (by convention, MIR675 refers to the MIR675-5p.) are significantly increased in several cancer tissues such as glioma cancer

17

, hepatocellular cancer

14

15

, gastric cancer

18

, and breast cancer

, CRC

16

, non-small cell lung

19

. It is not clear how MIR675 is

regulated in cancers, and like H19, the role of MIR675 in cancer development and metastasis was observed in separate studies

17, 19-22

. Here, we investigate the role of

MIR675-5p in PGE2-induced CRC invasion and metastasis.

Materials and Methods MiRNA Total RNA from cells was extracted using the miRNeasy Mini Kit (Qiagen) and reverse transcribed by the miRCURY™ LNA™ Universal RT microRNA PCR Universal cDNA Synthesis Kit II (Exiqon/Qiagen). Expression of miRNAs was examined by quantitative polymerase chain reaction (qPCR) using iQ SYBR Green Supermix (BioRad) and miRCURY microRNA assays (Exiqon/Qiagen) on a QuantStudio 7 Flex Real-Time PCR System (Thermo Fisher Scientific). Human SNORD44 was measured as a control. Clustered regularly interspaced short palindromic repeats(CRISPR)/Cas9 system To generate CRISPR/Cas9 plasmids targeting MIR675, p53, AKT1, AKT2, and Myc, primers of single-guide RNAs were synthesized at Integrated DNA Technologies, annealed, ligated into LentiCRISPRv2 (Addgene #52961) and transduced to cells by lentivirus using Lenti-X packaging single shots (VSV-G, Clontech #631275) and 293T

6

cells. Transduced cells were selected with puromycin and pooled. The targeted sequences are listed in the supplementary Materials and Methods. Animal Experiments Animal studies were performed in compliance with institutional guidelines under an IACUC approved protocol at MUSC or Arizona State University. NOD-scidIL-2Rg-/(NSG) mice were obtained from Jackson Laboratory (Bar Harbor, ME) and bred inhouse. For the orthotopic mouse model, 25,000 cells derived from LS174T cells were injected into the cecal wall of male NSG mice at the age of 8 weeks. Four days later, PGE2 (Cayman Chemical #14010) or vehicle were given to mice through intraperitoneal injection at 300 µg/mouse/day. PGE2 was dissolved in 100% ethanol and diluted to desired concentration with PBS immediately before the injection. The mice were sacrificed at 8 weeks for experiment with p53 overexpression and at 6 weeks for experiment with p53 CRISPR depletion after the start of the treatment. Metastatic nodules on livers were visually inspected and counted. Metastatic nodules on lungs were revealed by India ink staining and counted. Primary cecal tumors were isolated and weighted. Haematoxylin and eosin stain was performed on sectioned liver and lung to confirm tumor histology. Statistical Analysis The results of quantitative studies are reported as mean ± SD or Mean ± SEM (for animal experiments). The SEM/SD was calculated based on the number of independent experiments. Differences were analyzed by two-sided Student's t test. P values of < .05 were regarded as significant.

7

Procedures of several methods, including in vitro cell invasion assay, luciferase reporter assays, human CRC patient specimens and tissue microarray, miR-Seq analysis, cell culture and transfection, immunoblotting, siRNAs, RhoA pull-down activation assay, RNA-ChIP

and

qPCR,

p53

pathway

RT2

Profiler

PCR

array,

chromatin

immunoprecipitation, and public data mining, are available in the supplementary Materials and Methods.

Results PGE2 induces MIR675-5p expression in CRC cells To identify miRNAs that mediate effects of PGE2 on CRC cells, we conducted a miRNA-Seq experiment on LS174T cells treated with PGE2. Next generation sequencing data analysis identified a total of 403 miRNAs among which 56 miRNAs (FDR < 0.4) were responsive to PGE2 at all three time points (Figure 1A and Supplementary Table 1). We confirmed changes of some of the top candidates by qPCR assays (Supplementary Figure 1). The most strongly induced miRNA was MIR675-5p. The induction of MIR675-5p was obvious even when the PGE2 concentrations was as low as 0.01 µM (Figure 1B). Dose-dependent expression of MIR675-3p and lincRNA H19 were also observed (Figure 1B). The induction of MIR6755p and H19 was reproducible in multiple CRC cell lines treated with 1 µM PGE2 except for HCT116 which required higher doses of PGE2 (Figure 1C). More importantly, treatment of primary CRC cells isolated from CRC patient specimens with PGE2 robustly induced MIR675-5p expression (Figure 1D). We analyzed public databases and found that expression of MIR675-5p was significantly upregulated in the CRC

8

tissues compared with normal tissues (Figure 1E). This finding was confirmed by evaluation of samples from a different cohort of patients in a CRC tissue microarray (Figure 1F). Similarly, we observed upregulation of H19 in the CRC tissues when compared with normal tissues in two separate public databases (Figure 1G). As expected, H19 expression strongly correlated with MIR675-5p expression in CRC patient samples (Supplementary Figure 2). NF-κB and β-catenin converge on the Myc promoter to control H19/MIR675-5p expression PGE2 activated AKT by inducing phosphorylation of AKT in LS174T and SW48 cells (Supplementary Figure 3 A and B). PGE2 also activated the NF-κB by inducing phosphorylation of NF-κB and degradation of I-κBα via reducing its half-life (Supplementary Figure 3 A, B and C). Furthermore, PGE2 activated a NF-κB responsive element (Supplementary Figure 3D, left panel) and caused nuclear translocation of NFκB (Supplementary Figure 3E). Meanwhile, PGE2 also activated the β-catenin by inducing

increased

levels

of

active

β-catenin

(non-phosphorylation

on

Ser33/Ser37Thr41) (Supplementary Figure 3 A and B). PGE2 also stimulated the activity of a TOPflash (TCF reporter) (Supplementary Figure 3D, right panel) and induced nuclear translocation of β-catenin (Supplementary Figure 3F). To determine whether these signaling molecules are involved in PGE2 induction of MIR675-5p, CRISPR/Cas9 and siRNA technologies were deployed. CRISPR depletion of AKT1 or AKT2 inhibited PGE2-induced expression of MIR675-5p (Figure 2A) and H19 (Figure S3G, left panel). Similarly, knockdown of NF-κB (Figure 2C) or β-catenin (Figure 2D)

9

with siRNAs attenuated the effect of PGE2 on induction of MIR675-5p and H19 (Figs. 2A and Supplementary 3G). To further determine how these signaling molecules regulate the expression of H19 and MIR675-5p, we found that depletion of AKT1, AKT2, NF-κB, or β-catenin, or treatment with their inhibitors blocked PGE2 induction of Myc expression at both protein (Figure 2 B-D and Supplementary Figure 3 H and I) and mRNA levels (Figure 2E). These results demonstrate that Myc is a downstream target of these molecules. Moreover, the AKT inhibitor MK2206, I-κB inhibitor BAY11-7085, or a β-catenin inhibitor FH535 blocked Myc promoter activity induced by PGE2 in the luciferase reporter assays (Figure 2F). Additionally, analysis of ChIP assays showed that PGE2 induced the binding of β-catenin and NF-κB to the promoter of the Myc gene (Figure 2G). In agreement with previous findings

23

, our ChIP assays showed that PGE2 increased the

binding of Myc to the promoter of the H19 gene, which can be suppressed by the inhibitors of AKT, I-κB, or β-catenin (Figure 2H). Depletion of Myc inhibited PGE2 induction of MIR675-5p (Figure 2I) and H19 (Supplementary Figure 3J), demonstrating that Myc is required for PGE2-induced expression of MIR675-5p and H19. We further examined whether activation of AKT activates NF-κB. MK2206 blocked PGE2-induced degradation of I-κBα and phosphorylation of NF-κB (Figure 2J), suggesting that activation of AKT leads to activation of NF-κB pathway. In support of this notion, MK2206 inhibited the phosphorylation of Thr23 on IKKα (Figure 2K). Phosphorylation of Thr23 on IKKα by AKT is a prerequisite for the phosphorylation of NF-κB at Ser534 by IKKα

24

. Moreover, knockdown of IKKα inhibited PGE2-induced

degradation of I-κBα and phosphorylation of NF-κB (Supplementary Figure 3K).

10

A previous study suggested that PGE2 induces β-catenin stabilization/activation partially through AKT-GSK-3β cascade

25

. Indeed, we found that treatment of cells with

PGE2 led to the rapid phosphorylation of GSK-3β on serine 9, which requires AKT activity as it was blocked by AKT or PI3K inhibitors (Supplementary Figure 3H). It does not appear that PGE2 treatment increases receptor-mediated WNT signaling as we did not detect increased phosphorylation of LRP6 co-receptor (Supplementary Figure 3L). While MK2206 or BAY11-7085 effectively blocked PGE2-induced luciferase activity of a NF-κB response element, a β-catenin inhibitor, FH535, failed to do so (Figure 2L). Similarly, MK2206 or FH535 but not BAY11-7085 blocked PGE2-induced TOPflash activity (Figure 2M). Furthermore, MK2206, BEZ235, or BAY11-7085 but not FH535 inhibited the binding of NF-κB to the MYC promoter (Supplementary Figure 3M). Additive effects on expression of Myc (Supplementary Figure 3N) or MIR675-5p or H19 (Supplementary Figure 3O) was observed when both NF-κB and β-catenin were blocked. The tumor suppressor p53 is a direct target of MIR675-5p To identify novel downstream targets of MIR675-5p in response to PGE2 treatment, we used bioinformatics predictive approaches. RNAhybrid 26 analysis showed that there is a potential MIR675 binding site on the 3’ untranslated region (UTR) of TP53 mRNA with the minimum free energy of -31.7 kcal/mol (Supplementary Figure 4A). We employed RNA-ChIP assays to examine whether the TP53 mRNA is selectively enriched in the Ago2/RISC complex after PGE2 treatment or MIR675-5p overexpression. Indeed, PGE2 treatment or MIR675-5p overexpression caused significant enrichment of TP53 mRNA compared with controls (Figure 3A). We also assessed several previously

11

identified targets of MIR675-5p, including IGF1R, GPR55, MITF, RB, and CDC6, and found only increases of IGF1R and GPR55 mRNAs incorporated into RISC when MIR675-5p was overexpressed (Supplementary Figure 4B), confirming the previous notion that miRNAs suppress their targets in a context-dependent manner. Further, we found that PGE2 treatment caused downregulation of p53 and its direct target p21 in LS174T (Figure 3B) and SW48 (Supplementary Figure 4C) cells, both of which express wild-type p53 protein. CRISPR depletion of MIR675-5p not only increased the basal level of p53 but also blocked PGE2-induced downregulation of this protein. Re-expression of MIR675-5p (Supplementary Figure 4D) in depleted cells reversed this effect (Figure 3B and Supplementary Figure 4C). Overexpression of MIR675-5p reduced the expression of p53 and p21 (Figure 3C). In addition, upregulation of p53 and p21 expression caused by MIR675 depletion can be reversed by only MIR675-5p but not MIR675-3p rescue expression (Supplementary Figure 4E), suggesting MIR675-3p does not target p53. To further explore whether MIR675-5p directly suppresses p53 through the putative binding site in the 3’UTR of TP53 transcript, a luciferase reporter was obtained in which the full-length TP53 3’UTR (1207 bp)

27

with

wild-type or mutated MIR675-5p binding site were embedded downstream of the firefly luciferase. First, PGE2 treatment reduced luciferase expression of wild-type TP53 3’UTR, but did not affect expression of luciferase in tandem with the mutated TP53 3’UTR (Figure 3D). Moreover, a MIR675-5p inhibitor inhibited the effect of PGE2 on reduction of TP53 promoter activity (Figure 3D). Secondly, the luciferase expression was elevated by the MIR675-5p inhibitor and repressed by MIR675-5p mimic, respectively (Figure 3 E and F). In contrast, neither MIR675-5p inhibitor nor mimic significantly altered the

12

expression of the luciferase with mutated TP53 3’UTR (Figure 3 E and F). To further explore whether MIR675-5p and PGE2 modulate the p53 signaling pathway, we employed a PCR array which profiles the expression of 84 genes related to p53mediated signal transduction, including many direct targets of p53 such as p21. We first validated this array by comparing the expression of these genes in control and p53 CRISPR depletion LS174T cells. The expression of most genes including p21 was suppressed in p53 depletion cells compared with control cells (Supplementary Figure 4F). CRISPR depletion of MIR675 largely enhanced, whereas PGE2 treatment reduced, the expression of these genes (Figure 3 G and H). PGE2 promotes CRC cell invasion and metastasis through MIR675-5p and p53 To determine whether MIR675-5p is involved in PGE2-induced CRC cell migration/invasion and metastasis

28-30

, we first created three CRISPR constructs

targeting different loci of MIR675 genomic sequence. All three sgRNAs effectively reduced the expression of MIR675-5p (Supplementary Figure 5A). Depletion of MIR675-5p strongly inhibited PGE2-induced cell invasion in both LS174T (Figure 4A) and SW48 (Supplementary Figure 5B) cells in vitro. Re-expression of MIR675-5p or CRISPR depletion of p53 in MIR675 CRISPR depletion cells reversed this effect (Figure 4B). Overexpression of wild-type p53 blocked PGE2-induced cell invasion (Figure 4C). These results reveal that PGE2 enhances cell invasion through a MIR675-5p-p53 axis. Wild-type p53 negatively regulates cell migration and invasion

31

. Loss of p53 leads to

increased levels of GTP bound (active) RhoA and activated ROCK, its main effector protein. Most signals leading to altered cell migration and invasion converge on the Rho family of small GTPase

32

. We found that a ROCK inhibitor Y27632 inhibited PGE2-

13

induced cell invasion in LS174T (Figure 4D) and SW48 (Supplementary Figure 5C) cells, suggesting that RhoA is involved in PGE2 promotion of cell invasion. Indeed, a pulldown assay revealed that PGE2 treatment led to increased levels of GTP bound RhoA in LS174T (Figure 4E) and SW48 (Supplementary Figure 5D) cells. CRISPR depletion of MIR675-5p decreased level of GTP bound RhoA both in the presence and absence of PGE2. Re-expression of MIR675-5p or depletion of p53 in MIR675-5p depletion cells reversed this effect (Figure 4E and Supplementary Figure 5D). The activation of RhoA signaling by PGE2 was further supported by increased phosphorylation of MYPT1, a substrate of ROCK, which also requires MIR675-5p (Figure 4F and Supplementary Figure 5E). Overexpression of p53 effectively blocked PGE2-induced RhoA activation (Figure 4G and Supplementary Figure 5F). These in vitro results prompted us to explore whether MIR675-5p regulates CRC metastasis in vivo. We employed a CRC orthotopic model in which cells were injected into the cecal wall of NSG mice (Figure 5A). The mice were then treated with vehicle or PGE2. PGE2 treatment increased the number of metastatic lesions both in the liver and lung (Figs. 5B, Supplementary Figure 6A). PGE2 promotion of metastasis was not seen in mice carrying MIR675-5p depleted cells or wild-type p53 overexpressing (p53OE) cells (Figure 5B), demonstrating that the MIR675-5p-p53 pathway is required for PGE2induced metastatic spread. The primary tumor weight in MIR675-5p depletion or p53OE group was slightly reduced compared to the control group (Supplementary Figure 6B). Tumor cells in the PGE2 group displayed increased expression of MIR675-5p compared with the vehicle group (Supplementary Figure 6C), indicating that PGE2 induced MIR675-5p in vivo. PGE2 also reduced expression of p53 and p21 in vivo (Figure 5C).

14

Similarly, we performed an independent experiment using anti-MIR675-5p (miRZip) sequences, these sequences inhibited the effect of PGE2 on liver metastatic lesions (Supplementary Figure 6D). In support of these results, overexpression of MIR675-5p increased the number of liver metastastatic lesions compared to control cells (Supplementary Figure 6E). Lastly, to demonstrate the crucial role of p53 in CRC metastasis, we injected cells with p53 CRISPR depletion. p53 depletion promoted liver and lung metastases to a level that was comparable to that of PGE2 treatment (Figure 5D).

Expression of MIR675-5p and H19 correlates with that of COX-1 and COX-2 in CRC patients and predicts poor colorectal cancer patient survival To evaluate the clinical relevance of the expression of MIR675 and its precursor H19 to the prognosis of CRC patients, we analyzed multiple public datasets and found that high MIR675-5p (Figure 6 A and B) or H19 (Figure 6C) expression in CRC patients is associated with poorer disease-free survival and/or overall survival. Importantly, expression of PTGS1 (COX-1) or PTGS2 (COX-2) positively correlated with H19 expression (Figure 6D). Moreover, the protein expression of COX-2 was also positively correlated with that of MIR675-5p in a distinct tissue microarray cohort of CRC patients (Figure 6E). Interestingly, high H19 expression is associated with poorer overall survival in patients with high but not low COX-2 expression (Figure 6F). The smaller cohort of patients with available MIR675-5p expression data reveal a similar association with survival (Supplementary Figure 7). In cultured CRC cells, inhibition of COX-2 by celecoxib, or by genetic COX-2 CRISPR depletion increased the expression of p53 and

15

p21, which was reversed by PGE2 treatment (Figure 6G and Supplementary Figure 8), consistent with the notion that COX-2 derived PGE2 induces the expression of MIR6755p which suppresses p53 expression.

Discussion There have been controversial reports as to whether the H19/MIR675 duo are prooncogenic or tumor-suppressive. Most studies have indicated that they are associated with growth, migration, invasion, and metastasis in many cancers; however, the reported functional mechanisms vary (reviewed in 11). In CRC cells, H19 can function as a miRNA sponge to promote epithelial to mesenchymal transition and tumor growth and MIR675 can increase cell growth via targeting the tumor suppressor gene, Rb

33 16

.

Their role in CRC invasion and metastasis has not been previously investigated. In a publicly available database of RNA-seq performed in matched normal colon, primary CRC, and metastatic CRC, the expression of Myc and H19 increase from normal colon epithelium to CRC primary tumors to metastatic tumors (Supplementary Figure 9). Our study establishes that PGE2 can exert its biological functions, at least partially, through regulation of a miRNA. We identified MIR675-5p as a highly inducible miRNA that mediates the effect of PGE2 on CRC invasion and metastasis by regulating the expression of tumor suppressor protein p53 and subsequent modulation of RhoA activity. Importantly, the expression of COX-1 and COX-2 correlated well with that of H19, the precursor of MIR675-5p, and with MIR675-5p expression, in CRC patients. High COX-2 expression in colorectal cancers, as well as elevated H19 or MIR675-5p expression predicts poor survival. Due to the unavailability of paired data for p53 protein

16

expression levels and MIR675 or H19 expression, we were unable to analyze the correlation of p53 with MIR675-5p/H19 expression in CRC patients. This type of study is also complicated by the complex regulation of p53 protein expression including the effects of frequent mutations of this tumor suppressor protein in human cancer. MiRNA expression in response to PGE2 treatment has not previously been examined. We have identified more than 50 miRNAs whose expression might be altered significantly in response to a surge of PGE2 in CRC cells. This sheds new light on our knowledge of the biological role of the pro-inflammatory bioactive lipid, PGE2. Interestingly, Hur et al. recently identified five miRNAs as potential markers of the development of liver metastases in CRC patients

34

. Two of the downregulated miRNAs

(miR-320, miR-221) in liver metastases identified in this study were also downregulated in PGE2-treated cells (Supplementary Table 1). Our study focused on MIR675-5p because it was most highly induced by PGE2. However, the significance of changes of other miRNAs identified here needs to be further evaluated and explored. As an important inflammatory mediator, PGE2 plays a critical role in guiding and governing various aspects of the inflammatory response in humans

35

. Not surprisingly,

another upregulated miRNA in our screen was miR-146a (Supplementary Figure S1), which can be induced by different pro-inflammatory stimuli

36

as well. As is the case with

MIR675-5p upregulation, the induction of this miRNA by PGE2 is likely mediated by activation of NF-κB which is essential for the induction of miR-146a transcription as well 37

. Besides protein regulatory factors, miRNAs have emerged as key regulators of

inflammation, and it is likely that they modulate signaling of onset and termination of inflammation

38, 39

. Here we propose MIR675-5p is a new player in inflammatory

17

responses. Its precise role in regulation of inflammation associated with other cancers is yet to be determined. As a direct target of MIR675-5p, wild-type p53 may link PGE2 with inflammation and cancer. A causative relationship between chronic inflammation and cancer is a wellaccepted paradigm 40. However, the underlying mechanisms remain largely hypothetical. The link between chronic inflammation and cancer suggests that inflammation, largely controlled by NF-κB, must interfere with natural tumor suppressor mechanisms, of which wild-type p53 is extremely important. Hence, an obvious hypothesis is that the carcinogenicity of chronic inflammation is due to p53 suppression downstream of activated NF-κB. Indeed, this hypothesis is supported by findings from many studies (reviewed in

41

). A direct molecular link initiated from NF-κB to p53 has not been

established. Our study shows that PGE2-activated NF-κB can promote transcription of the Myc oncogene, in turn, as a transcription factor, Myc drives the upregulation of H19, leading to increased expression of MIR675-5p, which represses the protein translation of wild-type p53 in CRC cells. This adds a new tier of regulation to the interaction between wild-type p53 and NF-κB. Other inflammatory mediators can regulate the expression of p53 as well. The pro-inflammatory cytokine Interleukin 6 downregulates p53 expression and activity by stimulating ribosome biogenesis in CRC cells

42

. These

results suggest the establishment of inflammation may create a situation in which p53 can no longer effectively exert its function as an eradicator of transformation-prone cells. Interestingly, a recent study demonstrated that MIR675 can directly target p53 that impacts apoptosis induced by brain injury in mice, corroborating our findings and expanding the role of MIR675 43.

18

TP53 gene hemizygous deletion and point mutations frequently occur in CRC 44. The metastatic potential of tumor cells can be positively influenced by loss of wild-type p53 or expression of p53 gain-of-function mutants

31

. p53 directly influences transcription of

genes involved in metastasis by binding promoters of a variety of genes known to be involved in regulating cell motility and adhesion, processes that are important for metastatic spread of disease

45

. Most signals leading to altered cell migration and

invasion converge on the Rho family of small GTPases. Members of this family, including Rac, cdc42, and Rho, control actin dynamics and are integral to cytoskeletal changes accompanying tumor cell migration and invasion

46, 47

. Loss of p53 leads to

increased levels of GTP-bound (active) RhoA and activated ROCK, its main effector protein

48

. Overexpression and activation of RhoA is widely believed to contribute to the

metastatic process in multiple tumor types overexpressed in CRC tumors

54

49-53

. It has been shown that RhoA is

and RhoA silencing was reported to suppress the

growth of colon cancer xenografts in immunodeficient mice

55

. However, a group of

researchers argue that RhoA inactivation can enhance Wnt signaling and promote colorectal cancer56. Although this study has not been independently confirmed, one cannot rule out the possibility of context-dependent regulation. Our study supports a positive role for RhoA in promoting CRC invasion and metastasis. We reveal a novel PGE2-initiated axis in which upregulation of MIR675-5p is capable of suppressing the expression of wild-type p53, leading to activation of RhoA-ROCK pathway and consequently increased CRC cell invasion and metastasis to liver and lung. It is possible that MIR675-5p is also capable of suppressing mutant p53 expression assuming their 3’UTR sequences are mutation free. A recent report demonstrated that

19

multiple single nucleotide variants (SNVs) were present in the TP53 3’UTR in tumor specimens from 244 patients with diffuse large B-cell lymphoma 1. Most of the newly identified 3’UTR SNVs were located at sites that are complementary to seed sequences of miRNAs that are predicted or experimentally known to target TP53. Three SNVs disrupt the seed match between miR-125b and the TP53 3’UTR, thereby impeding suppression of p53 by this miRNA. Therefore, it is likely that mutant p53 may be protected from suppression by MIR675-5p due to the presence of SNVs. Interestingly, there is a SNV (G>A) that disrupts the seed match between MIR675-5p and the TP53 3’UTR (Supplementary Figure 10A) in patients with diffuse large B-cell lymphoma 1. This mutation effectively blocked the effect of MIR675-5p mimics or inhibitors on the TP53 3’UTR in a luciferase reporter assay (Supplementary Figure 10B). In agreement, we found that MIR675-5p is capable of targeting mutant p53 in DLD1 cells whose 3’UTR remains mutation free (confirmed by sequencing) in the seed sequence of MIR675-5p (Supplementary Figure 10C). In conclusion, while targeting PGE2 production remains a promising clinical direction in cancer therapy, finding a selective therapeutic window might be challenging due to its critical role in mucosal homeostasis and wound repair

35

. Our findings provide a unique

opportunity to target MIR675-5p, an important downstream effector of PGE2 to inhibit CRC metastasis. It is less likely to cause cardiovascular toxicity because the levels of PGE2 are not affected. The role of MIR675-5p in CRC metastasis is supported by the work reported here, but requires further study to assess its value in diagnostic, prognostic, or therapeutic clinical management. Given its role in promoting metastasis through p53 suppression, expression of MIR675-5p may have value as a prognostic

20

biomarker to predict the metastatic potential of primary tumors. MiRNAs have also been used as diagnostic biomarkers circulating in the bloodstream. Further work is needed to test whether MIR675-5p is detectable in the bloodstream, as well as the specificity of its expression for advanced CRC. Our findings not only reveal a novel mechanism by which PGE2 promotes CRC metastasis, but also provide a rationale to target miR675-5p in CRC prevention and treatment.

21

References 1. 2. 3. 4.

5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

18. 19.

20. 21. 22.

Wang D, Dubois RN. Eicosanoids and cancer. Nat Rev Cancer 2010;10:181-93. Dubois RN, Abramson SB, Crofford L, et al. Cyclooxygenase in biology and disease. FASEB J 1998;12:1063-73. de Groot DJ, de Vries EG, Groen HJ, et al. Non-steroidal anti-inflammatory drugs to potentiate chemotherapy effects: from lab to clinic. Crit Rev Oncol Hematol 2007;61:52-69. Newcomb PA, Baron J, Cotterchio M, et al. Colon Cancer Family Registry: an international resource for studies of the genetic epidemiology of colon cancer. Cancer Epidemiol Biomarkers Prev 2007;16:2331-43. Ogino S, Kirkner GJ, Nosho K, et al. Cyclooxygenase-2 expression is an independent predictor of poor prognosis in colon cancer. Clin Cancer Res 2008;14:8221-7. Fischer SM, Hawk ET, Lubet RA. Coxibs and other nonsteroidal anti-inflammatory drugs in animal models of cancer chemoprevention. Cancer Prev Res (Phila) 2011;4:1728-35. McLemore TL, Hubbard WC, Litterst CL, et al. Profiles of prostaglandin biosynthesis in normal lung and tumor tissue from lung cancer patients. Cancer Res 1988;48:3140-7. Rigas B, Goldman IS, Levine L. Altered eicosanoid levels in human colon cancer. J Lab Clin Med 1993;122:518-23. Wang D, Dubois RN. Cyclooxygenase-2: a potential target in breast cancer. Semin Oncol 2004;31:64-73. Wang D, DuBois RN. Urinary PGE-M: a promising cancer biomarker. Cancer Prev Res (Phila) 2013;6:507-10. Raveh E, Matouk IJ, Gilon M, et al. The H19 Long non-coding RNA in cancer initiation, progression and metastasis - a proposed unifying theory. Mol Cancer 2015;14:184. Cai X, Cullen BR. The imprinted H19 noncoding RNA is a primary microRNA precursor. RNA 2007;13:313-6. Keniry A, Oxley D, Monnier P, et al. The H19 lincRNA is a developmental reservoir of miR-675 that suppresses growth and Igf1r. Nat Cell Biol 2012;14:659-65. Shi Y, Wang Y, Luan W, et al. Long non-coding RNA H19 promotes glioma cell invasion by deriving miR-675. PLoS One 2014;9:e86295. Li H, Yu B, Li J, et al. Overexpression of lncRNA H19 enhances carcinogenesis and metastasis of gastric cancer. Oncotarget 2014;5:2318-29. Tsang WP, Ng EK, Ng SS, et al. Oncofetal H19-derived miR-675 regulates tumor suppressor RB in human colorectal cancer. Carcinogenesis 2010;31:350-8. He D, Wang J, Zhang C, et al. Down-regulation of miR-675-5p contributes to tumor progression and development by targeting pro-tumorigenic GPR55 in non-small cell lung cancer. Mol Cancer 2015;14:73. Yu YQ, Weng J, Li SQ, et al. MiR-675 Promotes the Growth of Hepatocellular Carcinoma Cells Through the Cdc25A Pathway. Asian Pac J Cancer Prev 2016;17:3881-5. Vennin C, Spruyt N, Dahmani F, et al. H19 non coding RNA-derived miR-675 enhances tumorigenesis and metastasis of breast cancer cells by downregulating c-Cbl and Cbl-b. Oncotarget 2015;6:29209-23. Zhou YW, Zhang H, Duan CJ, et al. miR-675-5p enhances tumorigenesis and metastasis of esophageal squamous cell carcinoma by targeting REPS2. Oncotarget 2016;7:30730-47. Zhu M, Chen Q, Liu X, et al. lncRNA H19/miR-675 axis represses prostate cancer metastasis by targeting TGFBI. FEBS J 2014;281:3766-75. Liu C, Chen Z, Fang J, et al. H19-derived miR-675 contributes to bladder cancer cell proliferation by regulating p53 activation. Tumour Biol 2016;37:263-70.

22

23.

24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36.

37.

38. 39. 40. 41. 42.

43.

Barsyte-Lovejoy D, Lau SK, Boutros PC, et al. The c-Myc oncogene directly induces the H19 noncoding RNA by allele-specific binding to potentiate tumorigenesis. Cancer Res 2006;66:53307. Bai D, Ueno L, Vogt PK. Akt-mediated regulation of NFkappaB and the essentialness of NFkappaB for the oncogenicity of PI3K and Akt. Int J Cancer 2009;125:2863-70. Castellone MD, Teramoto H, Williams BO, et al. Prostaglandin E2 promotes colon cancer cell growth through a Gs-axin-beta-catenin signaling axis. Science 2005;310:1504-10. Rehmsmeier M, Steffen P, Hochsmann M, et al. Fast and effective prediction of microRNA/target duplexes. RNA 2004;10:1507-17. Chen J, Kastan MB. 5'-3'-UTR interactions regulate p53 mRNA translation and provide a target for modulating p53 induction after DNA damage. Genes Dev 2010;24:2146-56. Buchanan FG, Wang D, Bargiacchi F, et al. Prostaglandin E2 regulates cell migration via the intracellular activation of the epidermal growth factor receptor. J Biol Chem 2003;278:35451-7. Kim JI, Lakshmikanthan V, Frilot N, et al. Prostaglandin E2 promotes lung cancer cell migration via EP4-betaArrestin1-c-Src signalsome. Mol Cancer Res 2010;8:569-77. Lu X, Han J, Xu X, et al. PGE2 Promotes the Migration of Mesenchymal Stem Cells through the Activation of FAK and ERK1/2 Pathway. Stem Cells Int 2017;2017:8178643. Powell E, Piwnica-Worms D, Piwnica-Worms H. Contribution of p53 to metastasis. Cancer Discov 2014;4:405-14. Muller PA, Vousden KH, Norman JC. p53 and its mutants in tumor cell migration and invasion. J Cell Biol 2011;192:209-18. Liang WC, Fu WM, Wong CW, et al. The lncRNA H19 promotes epithelial to mesenchymal transition by functioning as miRNA sponges in colorectal cancer. Oncotarget 2015;6:22513-25. Hur K, Toiyama Y, Schetter AJ, et al. Identification of a metastasis-specific MicroRNA signature in human colorectal cancer. J Natl Cancer Inst 2015;107. Nakanishi M, Rosenberg DW. Multifaceted roles of PGE2 in inflammation and cancer. Semin Immunopathol 2013;35:123-37. Taganov KD, Boldin MP, Chang KJ, et al. NF-kappaB-dependent induction of microRNA miR-146, an inhibitor targeted to signaling proteins of innate immune responses. Proc Natl Acad Sci U S A 2006;103:12481-6. Larner-Svensson HM, Williams AE, Tsitsiou E, et al. Pharmacological studies of the mechanism and function of interleukin-1beta-induced miRNA-146a expression in primary human airway smooth muscle. Respir Res 2010;11:68. Tahamtan A, Teymoori-Rad M, Nakstad B, et al. Anti-Inflammatory MicroRNAs and Their Potential for Inflammatory Diseases Treatment. Front Immunol 2018;9:1377. O'Connell RM, Rao DS, Baltimore D. microRNA regulation of inflammatory responses. Annu Rev Immunol 2012;30:295-312. Gudkov AV, Gurova KV, Komarova EA. Inflammation and p53: A Tale of Two Stresses. Genes Cancer 2011;2:503-16. Gudkov AV, Komarova EA. p53 and the Carcinogenicity of Chronic Inflammation. Cold Spring Harb Perspect Med 2016;6. Brighenti E, Calabrese C, Liguori G, et al. Interleukin 6 downregulates p53 expression and activity by stimulating ribosome biogenesis: a new pathway connecting inflammation to cancer. Oncogene 2014;33:4396-406. Yang S, Tang W, He Y, et al. Long non-coding RNA and microRNA-675/let-7a mediates the protective effect of melatonin against early brain injury after subarachnoid hemorrhage via targeting TP53 and neural growth factor. Cell Death Dis 2018;9:99.

23

44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56.

Baker SJ, Fearon ER, Nigro JM, et al. Chromosome 17 deletions and p53 gene mutations in colorectal carcinomas. Science 1989;244:217-21. Wei CL, Wu Q, Vega VB, et al. A global map of p53 transcription-factor binding sites in the human genome. Cell 2006;124:207-19. Bishop AL, Hall A. Rho GTPases and their effector proteins. Biochem J 2000;348 Pt 2:241-55. Heasman SJ, Ridley AJ. Mammalian Rho GTPases: new insights into their functions from in vivo studies. Nat Rev Mol Cell Biol 2008;9:690-701. Xia M, Land H. Tumor suppressor p53 restricts Ras stimulation of RhoA and cancer cell motility. Nat Struct Mol Biol 2007;14:215-23. Faried A, Nakajima M, Sohda M, et al. Correlation between RhoA overexpression and tumour progression in esophageal squamous cell carcinoma. Eur J Surg Oncol 2005;31:410-4. Kamai T, Kawakami S, Koga F, et al. RhoA is associated with invasion and lymph node metastasis in upper urinary tract cancer. BJU Int 2003;91:234-8. Chan CH, Lee SW, Li CF, et al. Deciphering the transcriptional complex critical for RhoA gene expression and cancer metastasis. Nat Cell Biol 2010;12:457-67. Horiuchi A, Imai T, Wang C, et al. Up-regulation of small GTPases, RhoA and RhoC, is associated with tumor progression in ovarian carcinoma. Lab Invest 2003;83:861-70. Mardilovich K, Olson MF, Baugh M. Targeting Rho GTPase signaling for cancer therapy. Future Oncol 2012;8:165-77. Fritz G, Just I, Kaina B. Rho GTPases are over-expressed in human tumors. Int J Cancer 1999;81:682-7. Wang H, Zhao G, Liu X, et al. Silencing of RhoA and RhoC expression by RNA interference suppresses human colorectal carcinoma growth in vivo. J Exp Clin Cancer Res 2010;29:123. Rodrigues P, Macaya I, Bazzocco S, et al. RHOA inactivation enhances Wnt signalling and promotes colorectal cancer. Nat Commun 2014;5:5458.

Authors names in bold designate shared co-first authorship.

24

Figure Legends Figure 1. Identification of MIR675-5p as a top candidate induced by PGE2 in CRC cells. (A) A Venn diagram representation of miR-Seq results showing the numbers of miRNAs significantly up- or down-regulated [False discovery rate (FDR) < 0.4] in LS174T cells treated with PGE2 for 8, 24, and 48 h relative to control cells. (B) Expression of MIR6755p, MIR675-3p, and H19 was examined by qPCR in LS174T cells. *P < .05, ** P < .02, compared with no PGE2 treatment. (C) Expression of MIR675-5p and H19 was examined by qPCR in CRC cell lines treated with 1 µM PGE2 for 24 h. Results are relative to the respective cell line treated with DMSO. * P < .05, ** P < .02; ns, not significant, compared with the DMSO treatment. (D) Expression of MIR675-5p was examined by qPCR in human CRC patient specimens treated with 1 µM PGE2 for 24 h. * P < .05, compared with the DMSO treatment. (E) Expression of MIR675-5p in tumor tissues of CRC patients (n=20) compared with normal tissues (n=20), as analyzed from published dataset GSE83924. (F) Expression of MIR675-5p in tumor tissues of CRC patients (n=50) compared with adjacent normal tissues (n=50) in a CRC tissue microarray revealed by in situ hybridization. (G) Expression of H19 in tumor tissues of CRC patients compared with normal tissues, as analyzed from published datasets TCGA (n=19, 3, and 101 for normal colon, normal rectum, and CRC, respectively) and Hong Colorectal Cancer (n=12 and 70 for normal colon and CRC, respectively). All data are presented as the mean ± SD unless otherwise specified.

25

Figure 2. NF-κB and β-catenin converge on Myc promoter to control MIR675-5p expression. (A) Expression of MIR675-5p was examined by qPCR in LS174T cells with reduced AKT1 (CRISPR), AKT2 (CRISPR), NF-κB (siRNA), or β-catenin (siRNA) expression compared with control cells. Cells were treated with 1 µM PGE2 or DMSO for 24 h. * P < .05. (B) Protein expression was examined by Western analysis. Cells were treated with 1 µM PGE2 or DMSO for 6 h. (C, D) Protein expression was examined by Western blot analyses. LS174T cells were transfected with siRNAs targeting NF-κB (C) or βcatenin (D) or a control siRNA prior to treatment with PGE2 or DMSO for 8 h. (E) Expression of Myc mRNA was examined by qPCR in cells indicated in (A). Cells were treated with 1 µM PGE2 or DMSO for 8 h. * P < .05. (F) The activity of Myc promoter was examined by a dual-luciferase reporter assay in LS174T cells. Cells were treated with 1 µM MK2206, 10 µM BAY11-7085, or 10 µM FH535 for 30 min before adding PGE2 (1 µM) for 8 h. * P < .05, compared with the PGE2 treatment. (G) The binding of βcatenin or NF-κB to the Myc promoter in response to PGE2 treatment was examined by PCR following a ChIP assay. (H) The binding of Myc to the H19 promoter in response to PGE2 treatment (upper panel); or PGE2 treatment in the absence or presence of inhibitors (lower panel) was examined by PCR following a ChIP assay. (I) Expression of MIR675-5p was examined by qPCR in cells with Myc CRISPR depletion. The effect of Myc depletion was demonstrated with Western analysis. Cells were treated with PGE2 or DMSO. * P < .05. (J, K) Protein expression was examined by Western analyses in LS174T cells. Cells were treated with inhibitors for 30 min before adding PGE2. (L, M) The activity of NF-κB or β-catenin was examined by a luciferase reporter containing

26

repeated NF-κB responsive element (L) or TCF binding sites (M). Cells were treated with inhibitors for 30 min before PGE2 was added. * P < .05; ns, not significant, compared with the PGE2 treatment. All data are presented as the mean ± SD.

Figure 3. Identification of p53 as a direct target of MIR675-5p. (A) RNA-ChIP analysis was conducted to detect levels of TP53 mRNA that were incorporated into the Ago2-RISC complex from LS174T cells treated with or without PGE2 or cells expressing indicated plasmids. ** P < .02. (B) Protein expression was examined by Western analysis in LS174T cells with indicated modifications: MIR675 CrR, MIR675-5p was re-expressed in cells with MIR675 CRISPR (Cr) depletion. (C) Protein expression was examined by Western analyses in LS174T cells with indicated modifications: MIR675-OE, MIR675 overexpression. (D) Alignment of MIR675-5p and its binding site in the 3’UTR of TP53. The nucleotides in the 3’UTR that were replaced to make the TP53 mutant are marked in red. Luciferase activity of the reporter construct containing the wild-type or MIR675-5p-binding mutant 3’UTR of TP53 was measured after transfection with or without a MIR675-5p inhibitor into LS174T cells, which were treated with PGE2 or DMSO. * P < .05; ns, not significant. (E, F) Luciferase activity of the reporter constructs in (D) was measured after transfection with or without a MIR6755p inhibitor (E) or MIR675-5p mimic (F). * P < .05. (G) MRNA levels of p53-regulated genes embedded in a PCR array in MIR675 CRISPR depletion cells relative to control cells were measured by qPCR. (H) The PCR array in (G) was used to measure mRNA levels of p53-regulated genes in PGE2 treated cells relative to DMSO treated cells. All data are presented as the mean ± SD.

27

Figure 4. MIR675-5p mediates PGE2-induced cell invasion through suppressing p53 expression and activation of RhoA signaling. (A) In vitro invasion of LS174T cells. Cells carrying a control construct or CRISPR (Cr) constructs targeting different loci in the MIR675 genomic sequence were treated with PGE2 or DMSO. * P < .05, compared with the eGFP treated with PGE2. (B) Invasion of LS174T cells carrying indicated constructs was measured as in (A). * P < .05; ns, not significant, compared with the eGFP treated with PGE2. (C) Effect of wild-type (WT) p53 overexpression (OE) on cell invasion of LS174T cells. * P < .05, compared with the control treated with PGE2. (D) PGE2-induced cell invasion of LS174T cells was evaluated with or without Y27632. * P < .05, compared with the treatment with PGE2 alone. (E) Activation of RhoA in LS174T cells. Cells carrying indicated constructs were treated with PGE2 or DMSO. (F) Phosphorylation of MYPT1 was examined by Western analysis. LS174T cells were treated with PGE2 or DMSO. (G) Levels of RhoA-GTP and expression of indicated protein in whole cell lysates (WCL) of LS174T cells were detected by Western analysis. All data are presented as the mean ± SD. Figure 5. MIR675 mediates PGE2-induced liver and lung metastases of CRC cells in vivo. (A) Schematic description of the metastatic model of CRC. (B) Mice carrying indicated cells (n=6) were treated with PGE2 or vehicle. Tumor nodules on liver and lung were counted separately. Representative images of liver or lung with metastatic nodules are shown. Data are presented as the mean ± SEM (*P < .05). (C) Protein expression in primary cecal tumors from control cells treated with PGE2 or vehicle was examined by

28

Western analysis. (D) Mice carrying indicated cells (n=6) were treated with PGE2 or vehicle. Tumor nodules on liver and lung were counted separately. Representative images of liver or lung with metastatic nodules are shown. Data are presented as the mean ± SEM (*P < .05). Figure 6. Clinical correlation of COXs-H19-MIR675-5p expression profiles and prognosis of CRC patients. Kaplan-Meyer analysis of the TCGA CRC provisional dataset (2016 version) revealed that patients with high MIR675 (A-B) and H19 (C) expression demonstrate poorer disease-free survival (A) and overall survival (B-C). (D) Expression of COX-1 RNA (PTGS1) and COX-2 RNA (PTGS2) correlate with H19 expression in the TCGA CRC dataset. (E) Expression of MIR675 correlates with COX-2 expression as revealed by a CRC tissue microarray (n=50). (F) High H19 expression is associated with poorer overall survival in patients with high but not low COX-2 expression in the TCGA dataset. (G) Protein expression was examined by Western analysis in LS174T cells with reduced expression of COX-2 or MIR675 achieved by CRISPR depletion or control cells. Cells were treated with PGE2 or Celecoxib.

29

Supplementary Materials and Methods Reagents and Antibodies The small molecule inhibitors MK2206, BEZ235, BAY117085, and FH535 were obtained from Selleck Biochemicals. PGE2 was from Cayman Chemical. The following antibodies were purchased from Cell Signaling Technology: anti-Myc (#13987), anti-β-catenin (#8480), anti-active β-catenin (#8814), anti-phospho-β-catenin (S552) (#5651), anti-p53 (#2527), anti-phospho-NFκB (S536) (#3033), anti-p21 (#2947), anti-AKT (#4691), anti-phospho-AKT (S473, #4058), anti-AKT1 (#2938), anti-AKT2 (#3063), anti-phospho-GSK3β (S9) (#5558), anti-IKKα (#2682), anti-phospho-LRP6 (#2568), and anti-GAPDH (#8884S). Anti-phospho-IKKα (S23) (#SAB4300106), anti-βactin (#A3854) and anti-β-tubulin (#T4026) antibodies were purchased from Sigma. Anti-NFκB (#SC372), anti-IκBα (#SC371), anti-phospho-IκBα (#SC8404), and antiLamin A (#SC56137) were from Santa Cruz Biotechnology. Anti-RhoA antibody was from Cytoskeleton. HRP-linked enhanced chemiluminescence (ECL) mouse (#NA934) and rabbit IgG (#NA931) were purchased from GE Healthcare Life Sciences. CRISPR/Cas9 targeting sequences MIR675-1-AGCGTCACCAAGTCCACTGT, MIR675-3-GAGAGGGCCCACAGTGGACT,

MIR675-2-CAGCGTCACCAAGTCCACTG, p53-

CCCCGGACGATATTGAACAA,

AKT1-TCACGTTGGTCCACATCCTG, AKT2- TCTCGTCTGGAGAATCCACG, Myc TGCGTAGTTGTGCTGATGTG. LentiCRISPR-eGFP (Addgene #51760) was used as the control. In Vitro Cell Invasion Assay

1

5 × 104 cells were suspended in 500 µl of serum free McCoys's 5A media and placed in Growth Factor Reduced Matrigel Invasion Chamber (Corning, #354483). The lower chamber was filled with serum-free McCoy's 5A media containing 1 µM PGE2. After overnight incubation, the cells were fixed in Diff Quick Fixative from DADE Diagnostics (Deerfield, IL) and stained in 0.1% crystal violet. Cells were removed from the upper surface of the filter with a cotton swab. Cells were counted on the undersurface of the filter to determine invasion. Luciferase reporter assays A luciferase reporter containing the 3’UTR of human p53 (145-pGL3ctrl-3' UTR) was a gift from Michael Kastan (Addgene plasmid # 28175) 2. The predicted binding sites were mutated from CCGTACTA to AATTAACG by PCR to generate the mutant 3’UTR. LS174T cells were co-transfected with 0.3 µg firefly luciferase constructs, 0.03 µg Renilla luciferase constructs and 40 ng MIR675-5p inhibitor or mimics in 24- well plate. Cells were serum-starved for 48 hrs before treated with PGE2. Firefly and Renilla luciferase activities were measured using the Dual-Luciferase Reporter Assay (Promega) and the results were normalized as relative luciferase activity (Firefly luciferase/Renilla luciferase). Similarly, TOPflash and FOPflash plasmids (EMD Millipore) were used to measure β-catenin activity. pGL4.32[luc2P/NF-κB-RE/Hygro] Vector (Promega) was used to measure NF-κB activity. MYC promoter (Del 1) was a gift from Bert Vogelstein (Addgene plasmid # 16601). Human CRC patient specimens and tissue microarray

2

The CRC patient specimens were obtained from Mayo Clinic and MD Anderson Cancer Center. The study was approved by the Institutional Review Board of both institutions. Informed consent was obtained from all patients. A human CRC tissue microarray (#BC05118a) was obtained from Biomax. The in situ hybridization of MIR675 and immunohistochemical staining of COX-2 were performed according to manufacturer’s instructions. miR-Seq analysis LS174T cells were serum-starved and treated with 1 µM PGE2 for 0, 8, 24, and 48 h before harvest. MiRNAs were isolated with miRNeasy Mini Kit (Qiagen). Purified miRNA (50 ng) was used to prepare small RNA-Seq libraries using the Small TruSeq RNA Sample Prep Kit following the protocol as described by the manufacturer (Illumina, San Diego, CA). Following library preparation, libraries were purified using the Pippin Prep (Sage Science, Beverly, MA). Libraries were clustered on the cBot as described by the manufacturer (Illumina, San Diego, CA). Clustered small RNA-seq libraries were single read sequenced using version 4 with 1X50 cycles on an Illumina HiSeq2500. Demultiplexing was performed utilizing bcl2fastq-1.8.4 to generate Fastq files. miR-Seq data were analyzed on CAP-miRSeq by The Center for Cancer Genomics Bioinformatics Core at Medical University of South Carolina (MUSC). Cell culture and transfection Colon cancer cell lines LS174T, SW48, DLD-1, HT29, and CT26 were from American Type Culture Collection. All cell lines were authenticated by providers utilizing Short Tandem Repeat (STR) profiling. Cells were grown in McCoy’s 5A medium with L3

glutamine (Corning Cellgro, #10-050-CV) and 10% fetal bovine serum (GE Healthcare, #SH30071.3) at 37°C under 5% CO 2. MIR675-5p overexpressing cells were established by lentiviral infection using shMIMIC Lentiviral microRNA (GE Healthcare) viral particles. LS174T/ZipNeg and Zip675 cells were generated by lentiviral infection of parental LS174T cells with pGreenPuro Scramble Hairpin Control (miRZip) or miRZip-675 anti-MIR675 (System Biosciences), and a pooled population was selected using puromycin. Cells were transfected with HiPerFect transfection reagents (Qiagen) or Lipofectamine 3000 reagent according to manufacturer’s instructions. Immunoblotting Cells were harvested in lysis buffer consisting of 50 mM Tris pH 7.4, 150 mM NaCl, 1% NP-40, 5 mM EDTA. Following 30 min incubation in lysis buffer at 4°C, lysates were cleared by centrifugation at 16,000 × g for 10 min at 4°C, then protein concentrations were determined by DC Protein Assay (BioRad). Peroxidase conjugated donkey antirabbit and sheep anti-mouse (1:10,000; GE Healthcare NA934 and NA931, respectively) antibodies were incubated for 1 h at room temperature. ECL prime kit (GE Healthcare) was used to detect chemiluminescence. siRNAs: siRNAs targeting NF-κB (#6261S), and IKKα (#6372S), were purchased from Cell Signaling

Technology.

siRNAs

targeting

β-catenin

were

obtained

from

Horizon/Dharmacon.

4

RhoA pull-down activation assay RhoA Pull-Down Activation Assay Kit (#BK036) from Cytoskeleton, Inc was used. Briefly, cells were lysed and lysates were incubated with rhotekin-RBD beads at 4ºC on a rotator for 1 h. After washing, beads were mixed with 2× Laemmli sample buffer and boiled for 2 min. Samples were subject to Western blot analysis with anti-RhoA antibody. RNA-ChIP and qPCR To isolated RNA from RISC, experiments were conducted as described 3. LS174T cells treated with PGE2 or DMSO or LS174T cells with MIR675-5p overexpression or vector control were fixed with 1% formaldehyde, followed by chromatin fragmentation. Cells were lysed and incubated with Protein A (Invitrogen) together with anti-pan Ago, clone 2A8 antibody (Millipore) or IgG control for immunoprecipitation. One-fourth of the immunoprecipitated RNA was reverse transcribed using iScript cDNA synthesis kit (BioRad), followed by qPCR analysis. p53 pathway RT2 Profiler PCR array Human p53 Signaling Pathway PCR array (#PAHS-027ZA-6) was obtained from Qiagen. PCR was performed on a Quantstudio 7 Flex real-time PCR machine. Data were analyzed with Qiagen’s web-based analysis center. Chromatin immunoprecipitation Chromatin Immunoprecipitation was performed using a ChIP assay kit (Upstate USA, Inc., Charlottesville, VA, USA) on LS174T cells treated with or without PGE2. Briefly, cells were cross-linked by incubating them in 1% (vol/vol) formaldehyde-containing

5

medium for 10 min at 37°C and then sonicated to make soluble chro matin with DNA fragments between 200 and 1000 bp. Samples of total chromatin were taken at this point to use as a positive control in the PCR (input chromatin). The cell lysates were precleared by incubation with G-Sepharose beads and then incubated with the antiactive β-catenin (8814S) or anti-NF-κB (sc-8008) antibody overnight at 4°C. DNA– protein complexes were collected with G-Sepharose beads followed by several rounds of washing, eluted and reverse cross-linked. Following treatment with Protease K, the samples were extracted with phenol-chloroform and precipitated with ethanol. The recovered DNA was resuspended in TE buffer and used for the PCR amplification. The primer sequences used for β-catenin and NF-κB were previously published (β-cateninforward: 5′-GTG AAT ACA CGT TTG CGG GTT AC-3′; β-catenin-reverse: 5′-AGA GAC CCT TGT GAA AAA AAC CG-3′; NF-κB-forward: 5′-ACT TTG CAC TGG AAC TTA CAA CAC-3′; NF-κB-reverse: 5′-CGA AAA AAA TCC AGC GTC TAA G-3′)

4, 5

. A distal (2 kb)

upstream region for both loci was monitored as a control. The PCR products were fractionated on 3% agarose gels and stained with ethidium bromide. Public Data Mining: Oncomine expression analysis: H19 expression between normal colorectal tissues and CRCs was examined using datasets available through Oncomine 67. Two datasets were examined: Hong Colorectal (GSE9348) and TCGA Colorectal. The Hong Colorectal dataset consists of a Chinese population, including 12 normal colon samples from healthy patients and 70 primary CRC tissues. The TCGA Colorectal dataset 68 consists of 19 normal colon samples, 3 normal rectum samples, and 101 colon adenocarcinoma samples. Normal tissues in this dataset are matched normal samples 6

from CRC patients. Default Oncomine analysis parameters were used to examine H19 expression in these datasets. Gene expression and survival analysis: TCGA provisional RNA-Seq data (COAD, version from 05/27/2016) was used to examine the expression of H19 and MIR675 in CRC and their association with patient survival. This dataset consists of 461 cases, for which RNA-Seq data is available for 382 samples and miRNA-Seq data is available for 255 samples. Clinical data, including overall survival and disease-free survival times, are available for 391 colon adenocarcinoma patients. Expression of genes is presented as median RNA Seq V2 RSEM as calculated by TCGA standards. Kaplan-Meyer curves were generated using the survminer and mvtnorm packages in R. MIR675 expression in normal and CRC tissues: Micro-RNA gene expression datasets comparing normal colon mucosal tissue to CRC tumor tissue were obtained from the Gene Expression Omnibus (GEO). The dataset (GSE83924) compares 20 normal colorectal biopsies to 20 CRC primary tumors using a microRNA microarray 69.

Supplementary References 1.

2. 3. 4. 5.

Li Y, Gordon MW, Xu-Monette ZY, et al. Single nucleotide variation in the TP53 3' untranslated region in diffuse large B-cell lymphoma treated with rituximab-CHOP: a report from the International DLBCL Rituximab-CHOP Consortium Program. Blood 2013;121:4529-40. Chen J, Kastan MB. 5'-3'-UTR interactions regulate p53 mRNA translation and provide a target for modulating p53 induction after DNA damage. Genes Dev 2010;24:2146-56. Wang H, Sun T, Hu J, et al. miR-33a promotes glioma-initiating cell self-renewal via PKA and NOTCH pathways. J Clin Invest 2014;124:4489-502. Sierra J, Yoshida T, Joazeiro CA, et al. The APC tumor suppressor counteracts beta-catenin activation and H3K4 methylation at Wnt target genes. Genes Dev 2006;20:586-600. Papanikolaou V, Athanassiou E, Dubos S, et al. hTERT regulation by NF-kappaB and c-myc in irradiated HER2-positive breast cancer cells. Int J Radiat Biol 2011;87:609-21.

7

Supplementary Table 1 Fold changes of miRNA expression compared to the 0 h treatment with PGE2 (FDR<0.4) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40

miRNA hsa-let-7a-5p hsa-let-7b-3p hsa-let-7d-5p hsa-let-7e-5p hsa-let-7f-1-3p hsa-let-7f-5p hsa-miR-1246 hsa-miR-130b-5p hsa-miR-141-3p hsa-miR-146a-5p hsa-miR-150-3p hsa-miR-15a-5p hsa-miR-15b-3p hsa-miR-16-2-3p hsa-miR-181a-3p hsa-miR-185-5p hsa-miR-194-5p hsa-miR-197-3p hsa-miR-200a-5p hsa-miR-200c-3p hsa-miR-20a-5p hsa-miR-215-3p hsa-miR-215-5p hsa-miR-221-3p hsa-miR-22-3p hsa-miR-25-5p hsa-miR-26a-1-3p hsa-miR-26a-2-3p hsa-miR-26a-5p hsa-miR-30c-5p hsa-miR-320a hsa-miR-324-5p hsa-miR-326 hsa-miR-330-3p hsa-miR-331-5p hsa-miR-335-5p hsa-miR-338-5p hsa-miR-340-3p hsa-miR-421 hsa-miR-4449

8h 1.271253 -1.50951 1.200141 1.374665 -1.25172 1.304462 -1.34902 1.268058 1.170841 1.234652 -1.45503 1.18059 -1.69883 1.190398 -1.27038 -1.88599 1.151493 -2.17047 1.200646 -1.17871 1.169521 1.507901 1.187763 -1.28254 1.172193 -1.31997 -2.09662 -1.76963 1.138084 -1.42419 -1.39669 -1.70274 -2.11106 -1.32075 1.345171 1.215225 -2.10738 1.432834 1.200168 1.254676

24 h 1.276652 -1.84837 1.189358 1.418426 -1.85291 1.330627 -1.83212 1.544952 1.251112 1.839628 -1.93962 1.208254 -1.92654 1.275358 -1.19846 -4.91923 1.300139 -3.44207 1.179006 -1.35373 1.102046 1.628782 1.475826 -1.49654 1.211297 -1.23766 -2.15984 -1.32317 1.196207 -1.66293 -1.50108 -2.4259 -2.39291 -1.28005 1.327514 1.174728 -3.31092 1.451457 1.166293 1.466089

48 h 1.29795 -2.58962 1.097863 1.28446 -2.38857 1.371367 -2.04996 1.30312 1.756971 3.586591 -2.142 1.371284 -1.76884 1.369583 -1.50984 -5.66895 1.344453 -5.24445 1.353615 -1.47254 1.10293 1.968734 1.787581 -1.57257 1.280898 -1.96796 -2.0518 -1.19862 1.426171 -3.27599 -1.94251 -2.53832 -3.10146 -1.9136 1.238402 1.302854 -2.95498 1.536778 1.118138 1.361163 8

41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56

hsa-miR-486-5p hsa-miR-500a-5p hsa-miR-501-3p hsa-miR-501-5p hsa-miR-548i hsa-miR-552-5p hsa-miR-574-5p hsa-miR-6087 hsa-miR-629-5p hsa-miR-652-3p hsa-MIR675-5p hsa-miR-769-3p hsa-miR-877-5p hsa-miR-92a-3p hsa-miR-941 hsa-miR-99b-5p

-2.15796 -1.40099 -1.30916 -1.41197 1.270612 1.413052 -1.42486 -1.2683 -1.40148 -1.3146 3.56678 -1.43159 -1.29017 -1.25186 -1.27609 -1.52161

-2.35086 -1.67129 -1.48322 -2.18364 1.248099 1.293319 -1.66196 -2.02171 -1.47608 -1.54408 6.12335 -1.45442 -1.83739 -1.29149 -1.46533 -1.44117

-4.61521 -2.3622 -2.27026 -2.86255 1.303709 1.35292 -1.6514 -1.82463 -1.47355 -1.78001 8.51104 -1.42396 -2.64884 -1.68155 -1.83866 -2.16978

9

A

B

D

E

G

C

F

A

D

B

C

F

E

G

H

I

J

K

L

M

A

B

D

E

G

H

C

F

A

D

G

B

E

C

F

A

B

C

D

B 0.75

++

++

n=39

+ +++ +++ ++++ + +

0.25

0.00

+

+

n=265

0.50

MIR675 high MIR675 low p = 0.015 0

30

60 90 Time (Months)

1.00

C ++ + +

0.75

++++ ++ +++++++++++ ++ +++++ ++ +++++++++++ ++ ++ + + +++ + ++ ++++++ ++++ + ++++ +++++++

0.75

+

0.50

+

++

n=294

++ +

0.25

0.00

120

MIR675 high MIR675 low p = 0.0047 0

30

n=53

0.50

0.00

60

90

D

120

E

10000

H19 (TPM)

10000

100

1

Spearman's rho = 0.139 p = 0.00699 1

F

10 100 PTGS1 (TPM)

Overall Survival, PTGS2 high H19=high 1.00 0.75

+++++ +++++ ++++++++++ +++++++++++++ ++++ ++++++ + ++ ++ + ++ ++++++ + ++

0.50 0.25 0.00

p = 0.0021

+ ++

n=191

50

+

+

Time

100

1

10 100 PTGS2 (TPM)

H19=high

n=28 0

Spearman's rho = 0.1531 p = 0.00270

150

0.75 0.50

H19=low

+++ ++++ ++++++ +++++++++ ++ + ++ +++++++ ++ +++ ++++ + + ++ ++++ ++ +++ + + + +

0.25

p = 0.33

n=25

++ ++++ ++ +

n=205

0.00 0

50

Time

100

1000

G

Overall Survival, PTGS2 low

H19=low 1.00

+++ +++

100

1

1000

Survival probability

H19 (TPM)

+

150

+++++ +++ + +++++++ + + +++++++++++++ ++ ++ +++++++ +++++ +++ + + +++++ ++++ +++ +++ ++ ++ ++++ +++ + + +++ ++++ + ++++ +++++ ++ + +

++ +++++++++ +

n=396

0.25

Time (Months)

Survival probability

1.00

Overall Survival Probability

1.00

+++ +++ +++ +++ + ++++ ++++ ++ ++++ ++++++ +++ + +++++ +++++ +++ + ++

Overall Survival Probability

Disease-free Survival Probability

A

H19 high H19 low p = 0.0057 0

+

+

n=53

50 100 Time (Months)

150

Supplementary Figure 1. Validation of candidate miRNAs identified in miRNA-Seq analysis. Expression of indicated miRNAs in LS174T cells treated with PGE2 was examined by qPCR assay.

10

Supplementary Figure 2. Correlation of expression of H19 and MIR675 in CRC patients in the TCGA CRC provisional dataset.

11

12

Supplementary Figure 3. NF-κB and β-catenin independently enhance MYC transcription to control H19/MIR675-5p expression.(A, B) Protein expression in LS174T (A) or SW48 (B) cells was examined by Western analysis. (C) Protein expression in LS174T cells was examined by Western analysis. Cells were pretreated with DMSO or PGE2 before adding cycloheximide (CHX, 50 µg/ml) for indicated times. (D) Activity of βcatenin or NF-κB was measured in LS174T or SW48 cells treated with PGE2 or DMSO by a dual-luciferase assay. * P < .05, compared with the DMSO treatment. (E, F) Western analysis was carried out following a cell fractionation assay to examine the nuclear translocation of NF-κB (E) or β-catenin (F) in response to PGE2 treatment. (G) Expression of H19 was examined by qPCR in LS174T cells with reduced AKT1 (CRISPR), AKT2 (CRISPR), NFκB (siRNA), or β-catenin (siRNA) expression compared with control cells. All cells were treated with 1 µM PGE2 or DMSO for 24 h. * P < .05. (H) Protein expression in LS174T cells treated with PGE2 or DMSO in the absence or presence of MK2206 (1 µM), GSK690693 (10 µM), or BEZ235 (0.5 µM) was examined by Western analysis. (I) Protein expression in LS174T cells treated with PGE2 or DMSO in the absence or presence of FH535 (10 µM) was examined by Western analysis. (J) Expression of H19 was examined by qPCR in LS174T cells with reduced Myc (CRISPR) expression compared with control cells. All cells were treated with 1 µM PGE2 or DMSO for 24 h. * P < .05. (K) Protein expression in LS174T cells was examined by Western analysis. LS174T cells were transfected with siRNAs targeting IKKα or a non-targeting control siRNA prior to treatment with PGE2 or DMSO for 8 h. (L) Protein expression in LS174T cells treated with PGE2 was examined by Western analysis. (M) The binding of NF-κB to the MYC promoter in response to PGE2 treatment in the presence of various inhibitors was examined by PCR following a ChIP assay. (N) Expression of Myc in LS174T cells treated with PGE2 and indicated inhibitors was examined by Western analysis. (O) Expression of MIR675-5p and H19 was examined by qPCR in LS174T cells treated with PGE2 and indicated inhibitors. Data are presented as the mean ± SD.

14

Supplementary Figure 4. MIR675-5p directly targets p53. (A) The binding of MIR675-5p to the 3’UTR of TP53 mRNA predicted by the RNAHybrid 2.2 is shown. (B) RNA-ChIP analysis was conducted to detect levels of a panel of mRNAs whose 3’UTRs reportedly harbor MIR675 binding sites that were incorporated into the Ago2-RISC complex from LS174T cells expressing indicated plasmids. (C) Protein expression was examined by Western analysis in SW48 cells with indicated modifications. (D) Expression of MIR6755p in LS174T cells with indicated modifications was measured by qPCR. (E) Protein expression was examined by Western analysis in LS174T cells with MIR675 CRISPR (MIR675 Cr), MIR675-5p rescue expression (MIR675 Cr-5p), and MIR675-3p rescue expression (MIR675 Cr-3p). (F) mRNA levels of 84 p53-regulated genes embedded in a PCR array in TP53 CRISPR depletion cells relative to control cells were measured by real-time PCR. The nineteen most downregulated genes are graphed and shown. 15

Supplementary Figure 5. MIR675-5p mediates PGE2-induced cell invasion through suppressing p53 expression and activation of RhoA signaling. (A) Expression of MIR675-5p was measured by PCR. LS174T cells carrying a control construct or CRISPR constructs targeting different loci of MIR675 genomic sequence were seeded and treated with PGE2 or DMSO. * P < .05, compared with the eGFP treated with PGE2. (B) In vitro invasion of SW48 cells was evaluated with a growth factor-reduced matrigel Boyden chamber assay. Cells carrying a control construct or CRISPR KO construct targeting MIR675 were seeded and treated with PGE2 or DMSO. * P < .05, compared with the control treated with PGE2. (C) PGE2 induced cell invasion of SW48 cells was evaluated with or without ROCK inhibitor Y27632. * P < .05, compared with the treatment with PGE2 alone. (D) Activation of RhoA in SW48 cells was measured by a rhotekin-RBD pull-down assay. Cells carrying indicated constructs were treated with PGE2 or DMSO. (E) Phosphorylation of MYPT1 was examined by Western analysis. SW48 cells carrying indicated constructs were treated with PGE2 or DMSO. (F) Levels of RhoA-GTP following a pull-down assay and expression of indicated protein in whole

16

cell lysates (WCL) of SW48 cells were detected by Western analysis. Data are presented as the mean ± SD.

17

Supplementary Figure 6. MIR675 mediates PGE2-induced liver and lung metastases of CRC cells in vivo. (A) Haematoxylin and eosin stain (H&E stain) of liver and lung sections from mice experiment in Fig. 5. (B) Comparison of weight of primary cecal tumor from mice experiment in Fig. 5. Data are presented as the mean ± SEM (*P < .05, compared with the control treated with the vehicle). (C) Expression of MIR675-5p in primary tumors from experiment described in Fig. 5A was examined by qPCR (n=5). Data are presented as the mean ± SEM (*P < .05, compared with the vehicle treatment). (D) LS174T cells with a MIR675 inhibitor expressing construct (Zip675) or the empty construct (ZipNeg) were injected into the cecal wall of NSG mice. The mice (n=6) were treated with PGE2 or vehicle and the liver metastasis lesions were counted. Data are presented as the mean ± SEM (*P < .05). (E) LS174T cells with a MIR675 expressing construct or the empty construct were injected into the cecal wall of NSG mice (n=6). The liver metastasis lesions were counted. Data are presented as the mean ± SEM (*P <.05).

18

Supplementary Figure 7. High MIR675 expression is associated with poorer overall survival, regardless of COX-2 expression. All data are derived from the TCGA CRC provisional dataset (2016 version).

19

Supplementary Figure 8. Protein expression of indicated proteins was examined by Western analysis in SW48 cells with reduced expression of MIR675 achieved by CRISPR depletion or control cells. Cells were treated with PGE2 or Celecoxib.

Supplementary Figure 9. Expression of H19 and Myc in normal (n=18), primary tumor (n=18), and metastatic tumor tissues (n=18) of CRC patients, as analyzed from published dataset GSE50760. 20

Supplementary Figure 10. Targeting of mutant p53 by MIR675-5p. (A) Alignment of MIR675-5p and its binding site in the 3’UTR of TP53 is shown. The nucleotide in the seed sequence of MIR675-5p that was identified as a SNV is marked in red. (B) Luciferase activity of the reporter construct containing the wild-type or SNV mutant 3’UTR of TP53 was measured after transfection with or without a MIR675-5p mimics or inhibitor into LS174T cells. (C) Protein expression of indicated proteins was examined by Western analysis in DLD-1 cells. Cr, CRISPR; R, MIR675-5p rescue expression. 21

BACKGROUND AND CONTEXT: Prostaglandin E2 (PGE2) promotes colorectal tumor formation and progression by unknown mechanisms.

NEW FINDINGS: We found incubation of CRC cells with PGE2 to increase their invasive activity and ability to form liver and lung metastases in mice. PGE2 downregulates expression of p53 by increasing expression of MIR675-5p, which binds to TP53 mRNA and prevents its translation.

LIMITATIONS: These studies were performed in cell lines and mice. Studies in humans are needed.

IMPACT: These findings provide insight into the mechanisms by which PGE2 promotes tumor development and progression. Strategies to target the PGE2 might be developed for treatment of CRC.

LAY SUMMARY The pro-inflammatory prostaglandin E2 induces MIR675-5p by induction of Myc expression via activation of AKT, NF-κB and β-catenin. MIR675-5p targets p53 to enhance RhoA activity to promote colorectal cancer metastasis.