miR-21-mediated regulation of 15-hydroxyprostaglandin dehydrogenase in colon cancer

Elevated prostaglandin E2 (PGE2) levels are observed in colorectal cancer (CRC) patients, and this increase is associated with poor prognosis. Increased synthesis of PGE2 in CRC has been shown to occur through COX-2-dependent mechanisms; however, loss of the PGE2-catabolizing enzyme, 15-hydroxyprostaglandin dehydrogenase (15-PGDH, HPGD), in colonic tumors contributes to increased prostaglandin levels and poor patient survival. While loss of 15-PGDH can occur through transcriptional mechanisms, we demonstrate that 15-PGDH can be additionally regulated by a miRNA-mediated mechanism. We show that 15-PGDH and miR-21 are inversely correlated in CRC patients, with increased miR-21 levels associating with low 15-PGDH expression. 15-PGDH can be directly regulated by miR-21 through distinct sites in its 3′ untranslated region (3′UTR), and miR-21 expression in CRC cells attenuates 15-PGDH and promotes increased PGE2 levels. Additionally, epithelial growth factor (EGF) signaling suppresses 15-PGDH expression while simultaneously enhancing miR-21 levels. miR-21 inhibition represses CRC cell proliferation, which is enhanced with EGF receptor (EGFR) inhibition. These findings present a novel regulatory mechanism of 15-PGDH by miR-21, and how dysregulated expression of miR-21 may contribute to loss of 15-PGDH expression and promote CRC progression via increased accumulation of PGE2.

MicroRNAs (miRNAs) are small non-coding RNAs, 18-22 nucleotides in length, that suppress gene expression post-transcriptionally 16 . miRNAs are fundamental regulators of gene expression and are predicted to regulate more than 60% of protein-coding genes 16 . miRNA expression has been shown to be dysregulated during tumorigenesis, thereby contributing to cancer progression due to mis-regulation of target mRNAs 17 . miR-21 is one of the most frequently upregulated miRNA in various human tumors, and is highly up-regulated miRNA in colorectal tumors [18][19][20][21] . Through this marked upregulation, circulating miR-21 has been shown to be an effective biomarker for early detection of CRC, along with a prognostic marker for aggressive disease, as it is associated with poor patient survival, advanced stage colorectal cancer, and reduced responsiveness to chemotherapy [21][22][23] . Through its pleiotropic upregulation in cancer and ability to regulate multiple tumor suppressor genes, miR-21 has gained much attention as a target for therapeutic inhibition 19,24-27 . Resistance to chemotherapies poses a significant clinical hurdle, as most patients who initially respond to therapy eventually develop resistance 28 . One such mechanism is dysregulation of miRNAs 29 . Intriguingly, miR-21 has been shown to be upregulated in resistant cells, and inhibition of miR-21 resensitized cells to chemotherapeutics 26 . Additionally, miR-21 overexpression was associated with resistance to epidermal growth factor receptor (EGFR) inhibitors, including erlotinib 30 . It has also been shown that COX-2 and PGE 2 contribute to resistance in CRC 31 . Targeted suppression of PGE 2 production through NSAIDs and selective COX-2 inhibitors are effective in modulating prostaglandin levels and in preventing colon polyp recurrence; however, these treatment regimens are associated with thrombotic events and cardiovascular side effects 32 . Through its ability to regulate PGE 2 levels, 15-PGDH is now being recognized as an alternative therapeutic target 33,34 , and elucidating the pathways that regulate its expression will be helpful in developing effective therapeutic strategies. The objective of this study was to determine if elevated miR-21 promotes constitutive repression of 15-PGDH expression, allowing for increased PGE 2 levels observed in CRC tumors. These findings provide insights into the role of miR-21 as a novel post-transcriptional link between EGFR signaling and 15-PGDH expression that may lead to alternative therapeutic interventions to improve on CRC patient outcomes.

Results
miR-21 and 15-PGDH expression are inversely correlated. miR-21 has been shown to be a consistently upregulated miRNA in all solid tumors, including CRC 21 . 15-PGDH serves as a tumor suppressor in gastrointestinal cancers, and is underexpressed in the majority of colon adenocarcinomas with low expression correlating to an aggressive disease phenotype 13 . Our prior work had identified that the 15-PGDH 3′UTR harbors putative miR-21 binding sites 2 , suggesting that elevated miR-21 in CRC may serve to inhibit 15-PGDH expression. Illustrated in Fig. 1a,b, three potential miR-21 binding sites were predicted within the 15-PGDH 3′UTR by in silico target prediction using RNAhybrid, microRNA.org, and microCOSM [35][36][37][38] .
In order to determine if an inverse correlation between miR-21 and 15-PGDH expression existed in vivo, normal human colonic tissue and colon tumor samples were analyzed for 15-PGDH and miR-21 expression levels ( Fig. 1c). 15-PGDH mRNA levels were attenuated in 94% of tumor samples with an average 17-fold decrease in 15-PGDH mRNA expression in tumor samples as compared to their matched normal colonic tissue (left panel). Conversely, miR-21 levels were shown to be elevated in 100% of tumor samples with an average 20-fold increase in miR-21 expression in tumor samples as compared to its matched normal colonic tissue (right panel). In normal colonic tissue, a significant inverse correlation of miR-21 levels with 15-PGDH mRNA expression was observed (r = −0.4499, *P = 0.0405) (Fig. 1d). 15-PGDH and miR-21 expression levels in tumor samples showed a similar trend of being inversely correlated, but was not found to be statistically significant (data not shown). To determine if this inverse correlation was also observed in colon cancer cell lines, a panel of colon cancer cell lines were analyzed for 15-PGDH mRNA expression and miR-21 levels. CRC cell lines (Moser, HCT-116, HT-29, HCA-7, HCT-15, Caco2) revealed a similar trend, although miR-21 levels and 15-PGDH mRNA expression were not significantly negatively correlated (Fig. 1e, r = −0.62, P = 0.19). Additionally, RNA-Seq and miR-Seq data from The Cancer Genome Atlas (TCGA) revealed miR-21 expression was significantly negatively correlated with 15-PGDH mRNA expression in colorectal cancer (r = −0.23, P < 0.00001, n = 287) (Fig. 2). Based on this data, we next examined if miR-21 could regulate 15-PGDH expression in CRC.

miR-21 regulates prostaglandin accumulation. Increased levels of PGE 2 observed in CRC tumors have
been shown to influence several cancer-associated pathways. Based on 15-PGDH's role in catabolizing PGE 2 and other prostaglandins (PGs) 9 , and that miR-21 can attenuate 15-PGDH protein expression, we hypothesized that miR-21 could promote elevated PG levels. To test this, HCT-15 cells that show low endogenous miR-21 levels and miR-21-mediated suppression of 15-PGDH expression (Figs 1e and 3), were transfected with synthetic miR-21 or control miR followed by assessment of prostaglandin levels in culture media. In miR-21-transfected HCT-15 cells, an approximate 3-fold and 2-fold increase in PGE 2 and PG levels were observed, respectively ( Fig. 4a,b), indicating that miR-21 can modulate prostaglandin production through regulating 15-PGDH protein expression.
EGFR signaling regulates miR-21 and 15-PGDH. 15-PGDH has been shown to be reactivated in colon cancer cell lines by inhibiting EGFR signaling 13,15,40 . Interestingly, other studies have shown that miR-21 expression is attenuated in response to EGFR inhibition 41 , indicating a potential miR-21-mediated mechanism for the observed increased 15-PGDH expression in response to EGFR tyrosine kinase inhibitors. To evaluate this, HT-29 cells were treated with the EGFR inhibitor erlotinib or vehicle control, and assayed for 15-PGDH protein levels and miR-21 expression. In cells treated with erlotinib, increased 15-PGDH protein levels as well as decreased miR-21 expression were observed ( Fig. 5a,b). Additionally, erlotinib decreased COX-2 protein levels ( Fig. 5c). To determine if erlotinib-dependent induction of 15-PGDH is due to its effect on decreased miR-21 levels, HT-29 cells were transfected with miR-21 prior to treatment with erlotinib or vehicle control. As shown in Fig. 5d, erlotinib-dependent induction of 15-PGDH was attenuated in the presence of miR-21, further implicating miR-21 as a post-transcriptional link influencing 15-PGDH levels downstream of EGFR signaling.

miR-21 inhibition regulates 15-PGDH levels and cell proliferation rates. To establish if inhibition
of miR-21 can reciprocally restore 15-PGDH expression, a miR-21 sponge was used in HT-29 and HCT-116 cells, as they express intermediate and high levels of miR-21, respectively. miRNA sponges act as decoy 3′ UTRs to sequester specific miRNAs, preventing the miRNA from binding to its endogenous targets 42 . The miR-21 sponge construct contains the GFP open reading frame under the control of the CMV promoter, while the 3′ UTR has multiple tandem miR-21 binding sites inserted 43 . These miR-21 binding sites have a three base mismatch within the middle portion of the sequence, producing a bulge that protects against endonucleolytic cleavage. In miR-21 sponge-transfected HCT-116 cells, elevated expression of both miR-21 target mRNAs PTEN and 15-PDGH were observed (Fig. 6a). Additionally, the miR-21 sponge increased 15-PGDH and PTEN protein levels (30% and 40%, respectively) in HT-29 cells, but did not change COX-2 expression (Fig. 6b). Lastly, we measured differences in HT-29 cellular proliferation after inhibition of miR-21 and EGFR signaling. Expression of the miR-21 sponge significantly reduced proliferation in HT-29 cells, whereas transfection of miR-21 or erlotinib alone did not impact proliferation significantly (Fig. 6c). Combination of erlotinib and the miR-21 sponge showed an additive effect in suppressing proliferation rates, suggesting potential synergy of EGFR and miR-21 inhibition in modulating CRC cell growth.  www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
The relevance of PGE 2 in colorectal cancer is well established, as elevated levels of PGE 2 from colon cancer patients correlates with poor prognosis 7,8 . While overexpression of COX-2 is considered a primary means for increased PG synthesis in CRC, loss of 15-PGDH expression in colonic tumors is now recognized as a contributing factor to increased PGE 2 levels. These changes in expression can result from various regulatory mechanisms, including the differential expression of miRNAs observed in colorectal tumors. COX-2 has been shown to be regulated through miRNA-mediated mechanisms 2,44-47 , however the potential of 15-PGDH to be regulated by miRNAs in CRC is not known. The findings presented here provide an additional level of understanding how control of PGE 2 levels can occur through miR-21-mediated regulation of 15-PGDH.
Expression patterns of miRNAs are altered in colon adenocarcinomas, and miR-21 is the highest expressed miRNA in a wide range of solid tumor including CRC [18][19][20][21]48 . A direct correlation between elevated tumor miR-21 expression and CRC is associated with worse clinical outcome 49 . Using clinical CRC samples and TCGA data sets, we show that miR-21 and 15-PGDH mRNA expressions were significantly negatively correlated in paired tumors, and in normal colonic tissue the inverse of this was observed (Fig. 1). We also analyzed miR-21 and 15-PGDH expression levels in colon cancer cell lines and saw a similar trend with our clinical data.
Colorectal tumors arise as a result of the activation of oncogenes as well as inactivation of tumor suppressor genes 50 . miR-21 is considered an oncomiR due to its constitutive overexpression in solid malignancies presumably through its ability to downregulate several tumor suppressors, metastatic, and apoptotic genes that have been identified as miR-21 targets 51 . Our work extends on this list to include 15-PDGH as a miR-21 target in CRC, and are consistent with current observations in other tumor types where miR-21 was identified to be a regulator of 15-PGDH [52][53][54][55] .
Inhibition of COX-2 activity has been associated with adverse cardiovascular side effects, highlighting the importance of developing alternative therapeutic strategies to regulate prostaglandin levels 32 . 15-PGDH expression is lost in the majority of colorectal adenocarcinomas, and this loss is associated with altered prostaglandin levels and poor prognosis [12][13][14] , indicating that re-establishment of 15-PGDH expression may provide a therapeutic benefit. We show that inhibition of miR-21 using a deliverable miRNA sponge increased 15-PGDH levels and significantly reduced CRC cell proliferation rates (Fig. 6). This decrease in proliferation is consistent with observations using COX-2 specific inhibitors 56,57 and provides an alternative therapeutic strategy to specifically target PGE 2 levels in CRC tumors. Clinically, miR-21 has been implicated in resistance to chemotherapies, including EGFR inhibitors 26,30 . COX-2 and PGE 2 also have a role in CRC resistance, and prostaglandin inhibition synergizes with EGFR inhibitors 31,32,58 . Our work supports these findings by showing a miR-21 sponge used in combination with erlotinib decreased CRC cellular proliferation greater than either therapy alone (Fig. 6). It would be worthwhile to investigate whether miR-21 inhibition can overcome erlotinib resistance, as this is a major obstacle with EGFR inhibitors in the clinic. Furthermore, selective inhibition of miR-21 may offer a greater effect on suppressing tumor growth, based on its ability to downregulate various tumor suppressor genes and promote resistance mechanisms. Therefore, it is appealing to determine the ability of miR-21 inhibition to synergize with conventional chemotherapies both in vitro and in vivo. This may be a more potent intervention than combination therapies using COX-2 specific inhibitors, as miR-21's oncogenic qualities are not limited to arachidonic acid signaling. www.nature.com/scientificreports www.nature.com/scientificreports/ There are several hurdles to climb before miRNA inhibitors can be used in CRC patients. The major obstacle is delivery, as it is crucial to get miRNA inhibitors into the tumor to ensure limited off-target effects at therapeutic concentrations 59 . Recently, clinical trials have provided hope for RNA therapeutic potential in cancer, and novel tumor targeting approaches are enhancing delivery of RNA molecules 60,61 . It is also critical to understand what effects miR-21 inhibition will have within the tumor microenvironment. While COX-2 and PGE 2 's roles in tumor microenvironment are fairly well-defined 62 , much less is known about miR-21 and 15-PGDH. Current research has begun to develop therapeutics that modulate the tumor microenvironment composition, with some therapies including miRNAs 63,64 . Thus, it will be important to study how miR-21 functions within the tumor microenvironment, and how modulation of miR-21 could potentially reprogram the tumor microenvironment to influence tumor growth.
Loss of 15-PGDH expression and increased levels of miR-21 have both been reported to occur at the adenoma stage 12,13,[65][66][67] , suggesting these changes may be an early cellular event during colorectal tumorigenesis occurring concomitantly. Previous work has shown that loss of 15-PGDH expression in CRC can be through transcriptional mechanisms, and 15-PGDH expression can be reactivated by correcting TGF-β or EGFR signaling 13,15,40 . Here we show that miR-21 expression is attenuated in response to EGFR signaling inhibition, indicating a potential mechanism for the observed increased 15-PGDH expression in response to EGFR inhibition 41 . PGE 2 has also been shown to transactivate EGFR allowing for increased prostaglandin signaling and tumor cell growth 68 , which could potentiate the effect of dysregulated 15-PGDH and miR-21 expression. We propose a model by which Figure 5. EGFR signaling regulates miR-21 and 15-PGDF. HT-29 cells were serum starved for 16 hr, following exposure to 100 ng/mL EGF for 1 hr, after which cells were treated with 5 μM Erlotinib or vehicle control for 5 hours and analyzed for (a) 15-PGDH protein, (b) relative miR-21 levels by qPCR, and (c) 15-PGDH and COX-2 protein expressions. Actin and tubulin served as protein loading controls; RNU6B served as an RNA internal control. (d) HT-29 cells were transfected with miR-21 or control miR for 48 hours, following serum starvation for 16 hr. Cells were then exposed to 100 ng/mL EGF for 1 hr, after which cells were treated with 5 μM Erlotinib or vehicle control for 5 hours and analyzed for 15-PGDH protein expression. Actin served as a loading control.
Scientific RepoRts | (2019) 9:5405 | https://doi.org/10.1038/s41598-019-41862-2 www.nature.com/scientificreports www.nature.com/scientificreports/ miR-21, the EGF pathway, and arachidonic acid signaling interact to promote CRC tumorigenesis (Fig. 7). We acknowledge our model's simplification, given the complex cross-talk between EGFR and arachidonic acid signaling, along with the pleiotropic effects of miR-21. Further work in completely defining this mechanism, as well as other miR-21-governed regulatory networks will further our understanding of this molecular pathway. Taken together, our findings indicate that miR-21 sits at the center of a molecular network and contributes to colon cancer progression in part through its ability to modulate PGE 2 levels through regulating 15-PGDH expression.
Protein, PGE 2 , and cell proliferation analysis. Cells were washed with PBS and lysed in RIPA buffer (50 mM Tris at pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 0.1% protease inhibitor). 25 μg of protein were loaded onto 10% SDS-PAGE gels and transferred onto PVDF membrane (VWR, Radnor, PA). Western blots were performed using antibodies against 15-PGDH (HPGD; HPA005679; Sigma-Aldrich, St. Louis, MO), anti-Luciferase (Promega, Madison, WI), COX-2 (160112, Cayman Chemical, Ann Arbor, MI), and PTEN (9559, Cell Signaling Technology, Danvers, MA). Membranes were stripped and re-probed using β-actin (Clone C4; MP Biomedicals, Solon, OH), tubulin (HRP-66031, Proteintech, Rosemont, IL) or GAPDH (HRP-60004, Proteintech, Rosemont, IL). antibodies. Cells transfected with luciferase reporter constructs were lysed in reporter lysis buffer (Promega, Madison, WI) and assayed using the Luciferase Assay System (Promega). Reporter gene activities were normalized to total protein; all results represent the average of triplicate experiments. Prostaglandin E 2 (PGE 2 ) levels and prostaglandin (PG) levels in cell culture media were analyzed by PGE 2 ELISA (R&D Systems, Minneapolis, MN), and Prostaglandin screening EIA kit (Cayman), respectively. Media was removed and cells were incubated for 20 min with serum-free media containing 10 μM arachidonic acid (Cayman) in serum-free media. Relative PGE 2 and PG levels were normalized to total protein www.nature.com/scientificreports www.nature.com/scientificreports/ levels and are an average of three experiments. WST-1 Cell Proliferation Assay (Cayman Chemical, Ann Arbor, MI) was performed in 96-well plates per the manufacturer's protocol, normalized to total protein, and represent the average of 3 biological replicates. miRNA Ribonucleoprotein immunoprecipitations. Immunoprecipitation of miRNA ribonucleoprotein complexes (mRNP-IP) was performed as described 71 , using a polyclonal anti-HA antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) or control IgG pre-coated to protein A/G PLUS agarose (Santa Cruz Biotechnology). Total RNA was isolated from immunoprecipitates using 1 ml Trizol per IP reaction and Taqman or SYBR green qPCR analysis of mRNA in RNP-IP samples was performed as described above.
Human tissue samples. Human colon tumors and histologically normal tissue were obtained from the Center for Colon Cancer Research (CCCR) Tissue Biorepository at the University of South Carolina (USC) with oversight and approval from the USC Institutional Review Board. Primary colorectal cancers were collected at the time of surgery by the CCCR Biorepository and immediately snap-frozen in liquid nitrogen and stored at −80°. Two independent pathologists confirmed diagnosis of all samples used in the study. Total RNA was isolated using Trizol from approximately 50 mg of tissue and converted to cDNA as described above. The quality of the resulting cDNA were determined for each sample by quantitative real-time PCR using GAPDH Taqman probes (for mRNA quality) and RNU6B (U6 RNA) primers (for small RNA quality). Samples were consented for use in biomedical research at the time of surgery. All the tissue samples and associated data obtained from the biorepository are fully deidentified.
Bioinformatic & statistical analysis. The Cancer Genome Atlas (TCGA) was mined using the TCGA-assembler 2 R software package 72 . Colorectal (COAD) RNA-Seq (gene.normalized_RNAseq, gene_RNAseq)and miR-Seq (mir_GA.hg19mirbase20, mir_HiSeq.hg19.mirbase20) was downloaded by TCGA-assembler 2 and analyzed on R using internal lab written software. Inquiries about lab written code can be emailed to carollutzlab@gmail.com. The data are expressed as the mean +/− SEM. Student's t-test and one-way ANOVA were used to determine significant differences. Where indicated, the Mann-Whitney U test was used to determine statistical significance. Inverse correlation studies used the Pearson product-moment correlation coefficient (PMCC) to determine the correlation value, r, and P-value was determined by using the correlation value, r, and the sample size. P-values less than 0.05 were considered significant.