miR-29a Is Repressed by MYC in Pancreatic Cancer and Its Restoration Drives Tumor-Suppressive Effects via Downregulation of LOXL2

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is an intractable cancer with a dismal prognosis. miR-29a is commonly downregulated in PDAC; however, mechanisms for its loss and role still remain unclear. Here, we show that in PDAC, repression of miR-29a is directly mediated by MYC via promoter activity. RNA sequencing analysis, integrated with miRNA target prediction, identified global miR-29a downstream targets in PDAC. Target enrichment coupled with gene ontology and survival correlation analyses identified the top five miR-29a-downregulated target genes (LOXL2, MYBL2, CLDN1, HGK, and NRAS) that are known to promote tumorigenic mechanisms. Functional validation confirmed that upregulation of miR-29a is sufficient to ablate translational expression of these five genes in PDAC. We show that the most promising target among the identified genes, LOXL2, is repressed by miR-29a via 3'-untranslated region binding. Pancreatic tissues from a PDAC murine model and patient biopsies showed overall high LOXL2 expression with inverse correlations with miR-29a levels. Collectively, our data delineate an antitumorigenic, regulatory role of miR-29a and a novel MYC-miR-29a-LOXL2 regulatory axis in PDAC pathogenesis, indicating the potential of the molecule in therapeutic opportunities. IMPLICATIONS: This study unravels a novel functional role of miR-29a in PDAC pathogenesis and identifies an MYC-miR-29a-LOXL2 axis in regulation of the disease progression, implicating miR-29a as a potential therapeutic target for PDAC. VISUAL OVERVIEW: http://mcr.aacrjournals.org/content/molcanres/18/2/311/F1.large.jpg.

Figure 1:. Knockdown/ inhibition of MYC results in increased miR-29a expression.

(A) qPCR analysis showing pri-miR-29a/b1 expression in normal human ductal epithelial cell lines (HPNE and HPDE) and human pancreatic cancer cell lines (Panc-1, MIA PaCa-2, BxPC-3, AsPC-1); n=4. (B) cBioportal analysis depicting the genetic alterations in the putative pri-miR-29a/b1 promoter binding transcription factors MYC, SMAD4, GLI3, NF- ĸB (NF- ĸB1) and YY1 in pancreatic adenocarcinoma patients from the TCGA database. (C) Regression plot indicating negative correlation between miR-29a expression (log2 normalized+1) and MYC (z-score) in TCGA pancreatic adenocarcinoma patients with alterations in MYC. (D) Western blot analysis of Panc-1 cells transfected with siCTRL, siSMAD4, siMYC and siGLI3 and assessed for SMAD4, MYC and GLI3 with GAPDH as the loading control. (E) Mature miR-29a expression as observed by qPCR analysis for Panc-1 cells transfected with siCTRL, siSMAD4, siMYC and siGLI3; n=4. (F) qPCR analysis of total RNA from Panc-1 cells transfected with siCTRL and siMYC showing pri-miR-29a/b1 expression; n=4. (G) Panc-1 cells were treated with various concentrations (1 μM, 10 μM and 100 μM) of small molecule MYC inhibitor 10058-F4. Subsequently, total RNA was subjected to qPCR analysis for mature miR-29a expression levels; n=3. Numerical data are represented as average fold change (ΔΔCT) ± standard error of the mean (SEM); *p< 0.05, **p< 0.01.

Figure 2:. MYC nuclear localization negatively correlates with miR-29a and represses miR-29a by directly binding to the promoter region of the molecule in PDAC.

(A) Representative images for immunofluorescence (IF) staining of normal human ductal epithelial cell lines (HPNE and HPDE) and human pancreatic cancer cell lines (Panc-1, MIA PaCa-2, BxPC-3, AsPC-1) for MYC. Scale bar is 50 μm, 20X magnification. (B) Average percentage of nuclear co-localization plotted for relative pri-miR-29a/b1 expression for the cell lines presented as ± SEM; n=4. Co-localization was calculated based on DAPI nuclear staining and MYC IF. (C) Western blot analysis of MYC expression in nuclear and cytoplasmic fractions of HPNE, Panc-1, MIA PaCa-2, BxPC-3, AsPC-1 cell lines. Quantification of band intensities were normalized to LAMB1 for nuclear and GAPDH for cytosolic fractions respectively. (D) Schematic representation of the two MYC binding sites at the pri-miR-29a/b1 promoter region. (E) Luciferase reporter constructs containing miR-29a/b1 promoter region with MYC binding sites were co-transfected in Panc-1 cells with siCTRL or siMYC and renilla luciferase expression plasmid. All readouts were normalized to renilla luciferase activity and average relative luminescence normalized to respective controls is presented as ± SEM; n= 5, *p< 0.05. (F) Real-time PCR analysis of DNA fragments precipitated in a CHIP assay using Panc-1 cell line. Two primer pairs (C1 and C2) designed within conserved MYC binding sites at miR-29a/b1 promoter and a primer pair for a validated MYC-binding region of CDKN1A were used to detect MYC- specific binding. Fold enrichment is represented as the signal obtained for MYC immunoprecipitation relative to that with IgG. Data presented as ± SEM; n= 3.

Figure 3:. RNA-seq analysis and identification of differentially expressed miR-29a target transcripts from the Panc-1 and MIA PaCa-2 datasets.

(A) Schematic representation of the RNA-seq analysis pipeline used to identify differentially expressed miR-29a target genes from the Panc-1 and MIA PaCa-2 datasets. (B) Volcano plot depicting the differentially expressed genes obtained from RNA-seq analysis for miR-29a overexpressing (OE) and control Panc-1 cells. (C) Correlation between differential expression (log₂FC) of the transcripts identified by RNA-seq in the two different PDAC cell lines of Panc-1 and MIA PaCa-2. (D) Venn diagram of downregulated transcripts in Panc-1 and/or MIA PaCa-2 datasets. Genes with logFC < 1, FDR<0.05 and p< 0.05 were only included. (E) The most enriched biological processes (GO terms) for the four ontologies (ECM Matrix Related, Metabolism, Migration/ Invasion or Cancer/Growth/Proliferation) associated with the overlapping miR-29a downregulated targets identified by RNA-seq are shown. The number on or outside each horizontal axis represents the gene number for a particular GO term. The false discovery rate value is shown as q-value for each GO. (F) Heatmap of the 43 overlapping downregulated miR-29a targets alongside the four associated GO categories.

Figure 4:. Functional validation of miR-29a downstream targets.

(A) Kaplan-Meier plots assessing the correlations between the seven miR-29a candidate gene (MYBL2, LOXL2, CLDN1, HGK, NRAS, FNDC5, and TUBD1) expressions and overall survival of TCGA pancreatic adenocarcinoma patients. (B) Panc-1 cells were transfected with different concentrations (5nM, 10nM and 20nM) of control (CTRL) or miR-29a mimics. Total protein was harvested from the cells 48 hrs post-transfection and subjected to western blot analysis for miR-29a candidate targets of MYBL2, LOXL2, CLDN1, HGK, and NRAS. GAPDH was used as the loading control. Quantification of band intensities normalized to GAPDH and relative to respective controls are represented as ± SEM; n=3, *p< 0.05, **p< 0.01, ***p< 0.001 (right). (C) Migration capacity of Panc-1 cells transfected with siCTRL, siMYBL2, siLOXL2, siCLDN1, siHGK, and siNRAS, or CTRL and miR-29a mimic were assessed by transmembrane cell migration assays. Relative cell migration was determined by the average number of migrated cells normalized to control (siCTRL and CTRL respectively) per 5 random fields and the data is presented as ± SEM; n=3, *p< 0.05, **p< 0.01, ***p< 0.001. (D) Total protein from Panc-1 cells transfected with siCTRL, siMYBL2, siLOXL2, siCLDN1, siHGK, and siNRAS, or CTRL and miR-29a mimic were subjected to western blot analysis for E-cadherin and vimentin. GAPDH was used as the loading control.

Figure 5:. miR-29a directly downregulates LOXL2 in PDAC cell lines.

(A) Schematic representation of putative wild-type (WT) and mutated (Mut) binding sites of the miR-29 family members at the 3’-UTR region of LOXL2 used in luciferase assay. (B) Relative luciferase activity of LOXL2 3’-UTR WT and Mut reporter constructs co-transfected with control (CTRL) or miR-29a mimics in Panc-1 cells. All readouts were normalized to renilla luciferase activity for each well, and average relative luminescence normalized to respective controls is presented as ± SEM; n= 6, **p< 0.01. (C) Western Blot for LOXL2 and MYC in Panc-1 cells transfected with siCTRL or siMYC. Relative protein levels were measured and normalized to GAPDH levels (indicated below). (D) Panc-1 and MIA PaCa-2 cells were transfected with LNA miRNA inhibitor control (siCT) or LNA miR-29a inhibitor (si-29a), and co-transfected with siCT and siCTRL (siCT1), or siMYC and si-29a. Total protein was subjected to western blot for LOXL2, and expression levels were normalized to GAPDH. (E) Panc-1 cells transfected with CTRL or miR-29a mimic, and siCTRL or siLOXL2 were cultured in serum free media for 48 hrs. Conditioned media thus obtained were subjected to ELISA for detection of secreted LOXL2. Data is presented as ± SEM; n= 4. ***p< 0.001. (F) Newly cross-linked pepsin-soluble collagen in ECM of Panc-1 cells transfected with CTRL or miR-29a mimic, and siCTRL and siLOXL2. Data is presented as ± SEM; n= 3. *p< 0.05; **p< 0.01 siCTRL (for siLOXL2). (G) Hydroxyproline content representing heavily cross-linked insoluble collagen in ECM of Panc-1 cells transfected with CTRL or miR-29a mimic, and siCTRL or siLOXL2. Data is presented as ± SEM; n= 3. *p< 0.05; **p< 0.01.

Figure 6:. Elevated LOXL2 levels inversely associate with miR-29a expression in KPC mice pancreas and human PDAC tumors.

(A) Upper panel: Representative photographs for immunohistochemical staining of LOXL2 in pancreatic sections from control C57BL/6, and Kras^LSL.G12D/+; p53^R172H/+; Pdx1-Cre (KPC) mice at 4.5– 6 months of age (original magnification X20). While negative staining was observed for control C57BL/6, KPC mice pancreata with PanIN lesions (black arrows) and PDAC stained positive for LOXL2 with significantly higher LOXL2 expression. LOXL2 positivity from immunohistochemistry analysis was quantified and presented as ± SEM; n= 15 animals per group (right), ***p< 0.001. Lower panel: Representative photographs for immunohistochemical staining of LOXL2 in PDAC clinical specimens. Positive staining for LOXL2 was observed around PanIN (black arrows) and PDAC (red arrows) lesions in PDAC tumor specimens with little or no staining for normal patient pancreatic specimens. LOXL2 positivity was quantified and presented as ± SEM (right); n= 6 for normal controls and n= 4 for patient PDAC tumors, **p< 0.01. (B) Total RNA from frozen pancreatic tissue sections of C57BL/6 (n=7; solid blue) or KPC mice (n=8; solid purple) were isolated and subjected to qPCR analysis to determine miR-29a and LOXL2 expressions. miR-29a expression is represented by inverse triangles and LOXL2 expression is represented by triangles. Mean expressions of miR-29a and LOXL2 for each group are indicated as horizontal lines, **p< 0.01. (C) Correlation analysis between LOXL2 and miR-29a expressions in PDAC patients from the TCGA database (n= 178).