Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Open Access

Peer-reviewed

Research Article

miR-376c promotes carcinogenesis and serves as a plasma marker for gastric carcinoma

  • Pei-Shih Hung,

    Affiliation Department of Education and Medical Research, National Yang-Ming University Hospital, Yilan, Taiwan

    ⨯
  • Chin-Yau Chen,

    Affiliation Department of Surgery, National Yang-Ming University Hospital, Yilan, Taiwan

    ⨯
  • Wei-Ting Chen,

    Affiliation Department of Surgery, National Yang-Ming University Hospital, Yilan, Taiwan

    ⨯
  • Chen-Yu Kuo,

    Affiliation Department of Medicine, National Yang-Ming University Hospital, Yilan, Taiwan

    ⨯
  • Wen-Liang Fang,

    Affiliations Division of General Surgery, Veterans General Hospital–Taipei, Taipei, Taiwan, Department of Dentistry, National Yang-Ming University Hospital, Yilan, Taiwan

    ⨯
  • Kuo-Hung Huang,

    Affiliations Division of General Surgery, Veterans General Hospital–Taipei, Taipei, Taiwan, Department of Dentistry, National Yang-Ming University Hospital, Yilan, Taiwan

    ⨯
  • Peng-Chih Chiu,

    Affiliation Department of Dentistry, National Yang-Ming University Hospital, Yilan, Taiwan

    ⨯
  • Su-Shun Lo

    sslo@ymuh.ym.edu.tw

    Affiliations Department of Surgery, National Yang-Ming University Hospital, Yilan, Taiwan, School of Medicine, National Yang-Ming University, Taipei, Taiwan

    ⨯

Abstract

Gastric carcinoma is highly prevalent throughout the world. Understanding the pathogenesis of this disease will benefit diagnosis and resolution. Studies show that miRNAs are involved in the tumorigenesis of gastric carcinoma. An initial screening followed by subsequent validation identified that miR-376c is up-regulated in gastric carcinoma tissue and the plasma of patients with the disease. In addition, the urinary level of miR-376c is also significantly increased in gastric carcinoma patients. The plasma miR-376c level was validated as a biomarker for gastric carcinoma, including early stage tumors. The induction of miR-376c was found to enrich the proliferation, migration and anchorage-independent growth of carcinoma cells and, furthermore, the repression of the expression of endogenous miR-376c was able to reduce such oncogenic phenotypes. ARID4A gene is a direct target of miR-376c. Knockdown of endogenous ARID4A increased the oncogenicity of carcinoma cells, while ARID4A was found to be drastically down-regulated in tumor tissue. Thus, expression levels of miR-376c and ARID4A mRNA tended to be opposing in tumor tissue. Our results demonstrate that miR-376c functions by suppressing ARID4A expression, which in turn enhances the oncogenicity of gastric carcinoma cells. It seems likely that the level of miR-376c in plasma and urine could act as invaluable markers for the detection of gastric carcinoma.

Citation: Hung P-S, Chen C-Y, Chen W-T, Kuo C-Y, Fang W-L, Huang K-H, et al. (2017) miR-376c promotes carcinogenesis and serves as a plasma marker for gastric carcinoma. PLoS ONE 12(5): e0177346. https://doi.org/10.1371/journal.pone.0177346

Editor: Javier S. Castresana, University of Navarra, SPAIN

Received: January 3, 2017; Accepted: April 26, 2017; Published: May 9, 2017

Copyright: © 2017 Hung et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are within the paper and its Supporting Information files.

Funding: This study was supported by grant MOST-102-2314-B-010-043-MY3 from Ministry of Science and Technology, and National Yang-Ming University Hospital grants RD2014001 and RD2015-001.

Competing interests: The authors have declared that no competing interests exist.

Introduction

The International Agency for Research on Cancer (IARC) has reported that gastric carcinoma (GC) to be the fifth most common malignancy worldwide. More than 50% of GC cases occur in Eastern Asia, which includes China, Japan, Korea and Taiwan. In 2014, GC ranked as the seventh largest cause of cancer death among all malignancies and sixth highest cause of cancer death among the male population of Taiwan. Surgical resection is the principal treatment modality of GC [1–3]. In general, the 5-year survival rate of resectable GC, with three out of five cases being resectable, is about 45% after the curative surgical operation, which implies that about 50% of patients suffer from recurrence of the tumor at a later time. Although the 5-year survival rate of late-stage patients is much lower than that of early-stage patients, the 10-year survival rate of resectable early GC patients is higher than 90%. Therefore, an early diagnosis of the disease is critical for successful surgical interception of GC.

Gastric carcinogenesis is a multistep neoplastic process that involves many genetic and environmental factors [4–7]. Despite the fact that environmental factors play important roles in GC carcinogenesis, it is still important to identify both genetic susceptibility factors as well as the others factors that predispose a patient to early metastasis. It is also very important to validate the mechanisms that underlie the metastasis and recurrence of this malignancy. Unfortunately an early diagnosis of GC is not feasible for most GC patients due to a requirement for diagnostic confirmation by gastroscopy and the lack of convenient non-invasive biomarkers that can be used for routine population screening. It would be of great benefit to identify novel biomarkers that are both non-invasive and convenient that will allow the simple detection of GC; these would greatly aid early diagnosis.

MicroRNAs (miRNAs; miRs) are short, non-coding endogenous RNAs; they are involved in the regulation of gene expression, which they do by controlling mRNA translation. The function of miRNAs is to repress gene activity and they mainly do this by binding with imperfect complimentarity to conserved sites in the 3'-untranslated regions (3'-UTRs) of a given gene, which blocks the translation of the mRNA transcript into protein [8]. Previous studies have reported that aberrant miRNA expression contributes to the neoplastic process in many types of cancers including GC, either via the control of oncogenes or via the control of tumor suppressors [9–12]. Research, including ours, has shown that miR-21, the miR-17-92 cluster, miR-146a, miR-200, miR-370 and a number of other miRNAs are up-regulated in GC, which means that they may act as oncogenic miRNAs [9, 13–17]. Circulating miRNAs have been found to be stable in the bio-fluids of cancer patients and many reports have demonstrated that these circulating miRNAs can indeed be used as biomarkers to identify and monitor a variety of cancers [18–21]. In addition, urine samples, which contain fewer proteins, seem to be better at preserving RNA than blood and this is becoming an important body fluid for the development of RNA biomarkers [22].

In our previous miRNA microarray screening [9], miR-376a and miR-376c were found to be up-regulated in GC tissue samples. Studies have shown that the miR-376 family is likely to play crucial roles in tumorigenesis. In human cervical and liver cancer cell lines, miR-376a has been shown to directly target the matrix extracellular phosphoglycoprotein gene that reduces G2 arrest and sensitizes cells to DNA damage-induced killing [23]. Furthermore, miR-376a overexpression has been found to significantly inhibit the arsenic trioxide-induced apoptosis of retinoblastoma cells [24]. It has also been reported that miR-376c targets activin receptor-like kinase 7 and that this enhances the survival of ovarian cancer cells [2]. Using a K-ras transgenic rats model, the serum level of miR-376a was suggested as a potential novel pre-clinical biomarker for pancreatic ductal carcinoma [25]. In addition, serum miR-376c level has been strongly associated with the poor differentiation of GC [26]. Circulating miR-376c has also been shown to be conspicuously up-regulated in the plasma of breast cancer patients and this miRNA seems to have the potential for being developed as part of multi-marker blood-based tests for the early detection of breast cancer [27]. These reports when taken together imply that miR-376a and miR-376c are likely to be important tumor markers.

The human ARID4A gene belongs to the AT-rich interactive domain (ARID) gene family, which consists of 15 genes. All ARID proteins bind to DNA either at AT-rich sites or in a non-specific manner. The ARID gene family can be further divided into seven distinct subfamilies, namely the ARID1, ARID2, ARID3, ARID4, ARID5, JARID1 and JARID2 subfamilies [28]. Aberrant expression of ARID proteins is a fairly common feature of tumor cells [29]. Cancer genome sequencing has identified the presence of recurrent somatic mutations that inactivate ARID1A in a various neoplasms including GC [28, 30–32]. ARID4A is known to be a tumor suppressor in a variety of malignancies [33–35]. However, there was no report up to the present has ever explored ARID4A and its relationship with GC.

In this study, we demonstrated that miR-376c is up-regulated in GC tissue as well as the plasma and urine from GC patients. The present of miR-376c in plasma could be a potential novel clinical diagnostic biomarker for early stage GC. In addition, miR-376c was found to able to enhance the oncogenicity of GC cells via targeting of ARID4A.

Materials and methods

GC tissue, plasma and urine samples

Resected GC tumor tissue samples together with paired non-cancerous mucosa (NCM) samples were harvested from 35 patients (group 1 in Table 1) who underwent curative resection of a previously untreated GC in National Yang-Ming University Hospital, between 2013 and 2015. About 0.5cm3 of the resected tissue was ground up and total RNA isolated from the lysate. Pre-surgical plasma samples of GC subgroup were obtained from 65 patients who were diagnosed GC (group 2 in Table 1) between 2013 and 2016. The control plasma samples were obtained from outpatients (group 3 in Table 1) who underwent gastroscopic examination between 2015 and 2016 to exclude the presence of neoplastic lesion in the upper gastrointestinal (GI) tract. Each sample from the above subjects consisted of 5 ml of whole blood that was collected in a heparin-coated tube [9]. Furthermore, both pre-surgical plasma and post-surgical plasma, sampled later than 1 month after surgery, were obtained from eight patients who were part of group 2. Urine samples were also collected from the patients who formed groups 2 and 3. The age of subjects ranges from 20 to 94 years old. Patients having previous history of GC or the presence of GI neoplasms other than GC were excluded for analysis. This study was approved by a hospital institute review board (Approval No.: 2013-B003 and 2014-A019) and all subjects provided written informed consent before the sampling took place.

thumbnail
Download:
Table 1. Clinical parameters of the subjects.

https://doi.org/10.1371/journal.pone.0177346.t001

No gender difference was noted between GC patients and control individuals (Fisher’s exact test; detailed analysis not shown), but control individuals were younger than GC patients (p<0.0001, un-paired t-test).

Reagents

Chemically modified miR-376 mimics (Cat. No.: MC11031 and MC12260) and an appropriate scrambled control (Cat. No.: 4464058), as well as miR-376 locked nucleic acid (LNA) blockers (Cat. No.: MH11031 and MH12260) and an appropriate scrambled control (Cat. No.: 4464076) were purchased from Applied Biosystems (Thermo Fisher Scientific, Waltham, MA, USA). A small interference oligonucleotide, siARID4A (Cat. No.: sc-6290), which was used to knock down ARID4A gene expression, as well as a control oligonucleotide (siControl; Cat. No.: sc-37007), were both purchased from Santa Cruz Biotech (Santa Cruz, CA, USA). TransFectinTM (Cat. No.:170–3351, Bio-Rad, Hercules, CA, USA) lipid reagent was used for transfection. Unless specified otherwise, all other reagents were purchased from Sigma-Aldrich (St Louise, MO, USA).

Quantitative RT-PCR (qRT-PCR)

Total RNA was purified using TRI reagent®. Applied Biosystems® TaqMan MicroRNA Reverse Transcription Kits (Thermo Fisher Scientific) and TaqMan® MicroRNA Assay Kits (Thermo Fisher Scientific) were used to detect the expression of miR-376a (Cat. No.: 000565) and miR-376c (Cat. No.: 002122). These reactions were carried out using iQ5 real-time PCR detection systems (Bio-Rad). U6B (Cat. No.: 001093), miR-16 (Cat. No.: 000391) or miR-1228 (Cat. No.: 002919) were used as controls for the analysis of cell, tissue and plasma samples [36, 37]. The mRNA expression levels of ARID4A (Cat. No.: Hs00243140_m1) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH, internal control, Cat. No.: Hs02758991_g1) were analyzed using the Applied Biosystems® TaqManq PCR System (Thermo Fisher Scientific). The mean threshold cycle (ΔCt) was achieved by normalization of the tested miRNA or mRNA signal against an internal control. The ΔΔCt was used to measure the relative changes in expression between tumor and NCM tissue samples and between the various experimental groups [9]. A negative control without a template was run in parallel to assess the overall specificity of the reaction. The 2ΔΔCt values were used to represent the fold changes in miRNA or mRNA expression.

Cell culture and proliferation assays

SCM-1 cell line originally established from a Taiwanese GC patient was grown in PRMI-1640 medium supplemented with 10% heat-inactivated fetal bovine serum (FBS, Biological Industries, Kibbutz Beit Haemek, Israel) at 37°C in a humidified atmosphere of 5% CO2 [9, 38]. The cell line was passaged at a density ratio of 1:5 when 80% confluence was reached [9]. Cell proliferation was analyzed by the MTT assay or by counting of cell numbers directly. For the MTT assay, the cells were seeded at 1000 cells/well using 96-well plates in triplicate. Growth curves were drawn using the values for OD570nm. For the cell number counts, the cells were seeded at 5000 cells/well using 48-well plates in triplicate. Cell growth curves were drawn based on the total number of cells present when counted at different time points.

Transwell migration assay

Transwells (Corning, Acton, MA, USA) were pre-coated with fibronectin as a chemotactic attractor to induce cell migration. A total of 1x105 cells were seeded into each Transwell containing serum-free medium in the upper chamber. Then 1 μM hydroxyurea was added to arrest cellular proliferation. After 24 hours, the migrated cells on the lower surface of membrane were stained with Hoechst 33258 (Thermo Fisher Scientific) and counted under fluorescence microscopy.

Anchorage-independent colony forming assay

Using 6-well plates, lower agar layer was created using a 1:1 mixture of 1.8% agarose and 30% FBS. Next, an upper agar layer was formed using a 1:1 mixture of methylcellulose and 30% FBS such that it covered the lower agar layer. Then, 1x105 cells were seeded into upper layer and the plates cultivated for 7 days. Finally, the colonies present on the plates were stained with 0.05% crystal violet and those colonies with a diameter larger than 50 μm in 15 randomly selected fields were counted using an inverted microscope.

Reporter constructs and activity assay

The 3′-UTR sequence of the ARID4A gene which contains predicted miR-376c binding site was cloned into the pMIR-REPORTER vector (Applied Biosystems) to generate the reporter construct designated pMIR-ARID4A. The plasmid pRL-TL, which expresses the renilla luciferase gene, was used as a transfection control. After co-transfecting of miR-376c mimic with pMIR-REPORTER (vector alone; VA) or pMIR-ARID4A, as well as pRL-TL for 24 hours, the lysate were collected and assayed in accordance with the protocol of Dual-Luciferase® reporter assay system (Promega, WI, USA). Firefly luciferase activity was normalized against renilla luciferase activity to indicate the reporter activity.

Western blot analysis

A total of 30 μg of protein from the various whole cell lysates were resolved by electrophoresis using a 10% denaturing polyacrylamide gel; the approach followed previously described protocols [9]. Primary antibodies against ARID4A (Cat. No.: sc-81640), GAPDH (Cat. No.: sc-47724) and α-Tubulin (Cat. No.: sc-5286) as well as appropriate secondary antibodies were purchased from Santa Cruz Biotech. The protein signals were detected using a Western Lightning Chemiluminescence Reagent Plus Kit (Perkin-Elmer, Wellesley, MA, USA) and a Chemi-Smart 3000 image acquisition system (Viber Lourmat, Collégien, France).

Statistics

The Fisher’s exact test, Mann–Whitney test, t-test, two-way ANOVA, linear regression analysis, receiver operating characteristic (ROC) analysis, and various bioinformatic modules were used for the statistical analysis. The results were considered to be statistically different when p< 0.05.

Results

miR-376a and miR-376c are up-regulated in GC

The expression of miR-376a and miR-376c in 35 paired GC and NCM tissue samples were analyzed. Both miR-376a and miR-376c were significantly up-regulated in the GC tissue compared to the NCM tissue (Fig 1A and 1B). ROC analyses showed that the expression levels of miR-376a and miR-376c in the GC tissue had predictive powers of 0.64 and 0.60, respectively, when distinguishing GC tissue from NCM tissue (Fig 1C and 1D) using area under curve (AUC) analysis. Although no statistically significant difference was found between the GC patients and control individuals for plasma miR-376a levels (Fig 1E), the amount of miR-376c in plasma was significantly higher in GC patients compared to control healthy individuals (Fig 1F). A positive correlation between the expression levels of miR-376a and miR-376c in the tissue (Fig 1G) and plasma (Fig 1H) of GC patients was found.

thumbnail
Download:
Fig 1. miR-376a and miR-376c in GC tissue and plasma.

(A, B) Before-and-after plot. The–ΔCt of miR-376a (A) and miR-376c (B) in 35 pairs of GC and NCM. Paired t-test. (C, D) ROC analysis of miR-376a (C) and miR-376c (D).AUC, area under curve. (E, F) Whisker plot. The–ΔCt of miR-376a (E) and miR-376c (F) in plasma samples from 108 healthy control donors and 65 GC patients. +, mean value. Numbers within parenthesis, numbers of tissue pairs or cases. Un-paired t-test. (G, H) Linear regression analysis between the expression of miR-376a and miR-376c in GC tissues (G) and the expression of miR-376a and miR-376c in the patients’ plasma (H).

https://doi.org/10.1371/journal.pone.0177346.g001

The ROC analyses indicated that plasma level of miR-376c in GC patients had a predictive power of 0.77 when separating GC patients from the control population (Fig 2A). Among the eight GC patients who had both pre-surgical and post-surgical plasma samples available, it was noted that there was a reduction in the plasma levels of miR-376c after tumor resection (Fig 2B). In addition, it was also found that the urinary levels of miR-376c were also significantly higher in GC patients compared to the control population (Fig 2C). The prediction power of urinary miR-376c with respect to GC was found to be about 0.70 (Fig 2D).

thumbnail
Download:
Fig 2. Plasma and urinary miR-376c in GC.

(A) ROC analysis of plasma miR-376c. (B) Before-and-after plot. Changes of miR-376c in eight paired pre-surgical and post-surgical samples. Paired t-test. (C) Whiskers plot. Urinary miR-376c in 11 healthy controls and 20 GC patients. +, mean value. Numbers within parenthesis, numbers of cases. Un-paired t-test. (D) ROC analysis of urinary miR-376c. AUC, area under curve.

https://doi.org/10.1371/journal.pone.0177346.g002

Plasma miR-376c is up-regulated in early stage GC

We investigated the plasma levels of miR-376c in GC patients relative to the control individuals according to the clinicopathological features of the disease. The plasma levels of miR-376c were up-regulated in both male patients and female patients relative to controls of the same gender (Fig 3A). In addition, patients with intestinal type and well differentiated GC tumors had increased levels of miR-376c compared to controls (Fig 3B and 3C). It should be noted that plasma levels of miR-376c were also up-regulated in T1, N0 and Stage I patients (Fig 3D–3F). When tumors became more poorly differentiated, became larger, underwent nodal metastasis or underwent progression to a more advanced clinical state, the plasma level of miR-376c was found not to increase further (Fig 3B–3F). Thus, our findings imply that the up-regulation of miR-376c in plasma occurs during the early establishment stage of human GC tumorigenesis and that the plasma of miR-376c level does not seem to reflect either tumor aggressiveness or tumor burden.

thumbnail
Download:
Fig 3. Plasma level of miR-376c and clinicopathological parameters.

Whisker plots. (A) Gender. (B) Lauren classification. (C) Cellular differentiation. (D) Tumor size. (E) Nodal status. (F) Clinical stage. +, mean value. Numbers within parenthesis, numbers of tissue pairs or cases. Un-paired t-test.

https://doi.org/10.1371/journal.pone.0177346.g003

As the average age of control individuals was younger than that of GC patients for more than ten years (Table 1), concerns were raised to wonder that the age difference might underlie the discrepancies of miR-376c in plasma. We thus divided group 2 and group 3 subjects into younger and older subgroups according to the median ages. The analysis showed that the plasma miR-376c in younger patients was higher than that in controls (S1A Fig). In addition, the plasma miR-376c in patients was also higher than that in older controls (S1B Fig). It is unlikely that age may confound the plasma miR-376c across sample groups.

miR-376c promotes the oncogenicicity of the SCM-1 cell line

The molecular functions of miR-376c were analyzed by carrying out transient transfections withmiR-376c mimic and miR-376c blocker. At 24 hours after transfecting with mimic, miR-376c was found to be up-regulated in SCM-1 cells (Fig 4A). Exogenous miR-376c expression seemed to have significantly increased the proliferation (Fig 4B), migration (Fig 4C) and anchorage-independent colony formation (Fig 4D) of SCM-1 cells. By way of contrast, at 24 hours after transfecting with blocker, there was a decreased in the endogenous expression of miR-376c in SCM-1 cells (Fig 4E). Furthermore, in the presence of blocker, there were also a decrease in proliferation (Fig 4F), migration (Fig 4G) and anchorage-independent colony formation (Fig 4H), all relative to the control.

thumbnail
Download:
Fig 4. miR-376expression and the oncogenic phenotypes of SCM-1 cells.

(A-H) miR-376c. (I-K) miR-376a. (A-D) Cells transfected with miR-376c mimic. (E-H) Cells transfected with miR-376c LNA. (I-K) Cells transfected with miR-376a mimic or miR-376a LNA. (A, E, I) qRT-PCR analysis to detect the expression of miR-376; Mann-Whitney test. (B, F). Proliferation assay; Two-way ANOVA test. (C, G, J) Migration assay; Mann-Whitney test. (D, H, K). Anchorage-independent growth; Mann-Whitney test. Each assay is independently carried out for three times. Data shown are mean ± SE in one representative assay.

https://doi.org/10.1371/journal.pone.0177346.g004

The molecular functions of miR-376a were also addressed by transfecting miR-376a mimic and miR-376a LNA blocker. Treatment with miR-376a mimic significantly increased exogenous miR-376a expression level of the SCM-1 cell line (Fig 4I, Lt) as well as increasing migration (Fig 4J) and anchorage-independent colony formation by this cell line (Fig 4K, Lt). However, treatment with miR-376a LNA blocker decreased the level of endogenous expression of miR-376a (Fig 4I, Rt) together with bringing about a reduction in the anchorage-independent colony formation of SCM-1 cells (Fig 4K, Rt).

miR-376c negatively regulates ARID4A expression in SCM-1 cells

The TargetScan (v7.1) in silico module predicted that ARID4A is a putative target of miR-376c. The alignment of miR-376c and the 3’-UTR of the ARID4A gene, which contains the target sequences of miR-376c, is illustrated in the Upper panel of Fig 5A. The activity of pMIR-ARID4A reporter was slightly lower than control reporter in cells transfected with control oligonucleotide. After transfection with the miR-376c mimic for 24 hours, the presence of exogenous miR-376c was able to further suppress the activity of pMIR-ARID4A reporter (Fig 5A, Lower). In addition, the expression of ARID4A mRNA and protein in SCM-1 cells were also suppressed (Fig 5B, Lt). However, the reduction in miR-376c expression brought about by treatment with miR-376c LNA blocker increased the expression of ARID4A at both the mRNA and protein level (Fig 5B, Rt). To elucidate the functions of ARID4A, siARID4A oligonucleotide and its equivalent scrambled control (siControl) were transiently transfected into SCM-1 cells. At 24 to 48 hours after transfecting with the active siRNA, the levels of ARID4A mRNA and protein expression were found to have decreased (Fig 5C). This decrease in endogenous ARID4A expression seemed to be associated with a slight up-regulation in cell proliferation (Fig 5D), and this in turn brought about a resultant up-regulation of migration (Fig 5E) and anchorage-independent colony formation (Fig 5F) by SCM-1 cells.

thumbnail
Download:
Fig 5. miR-376c decreases the expression of ARID4A and ARID4A acts as a suppressor in SCM-1 cells.

(A) Upper, the predicted target sequence present in the ARID4A 3’UTR (red) and the seed sequence (green) of miR-376c. Lower, reporter activity. VA, control reporter. Decreased activities are noted in pMIR-ARID4A reporter in relation to VA reporter. The decrease is particularly drastic with the treatment of miR-376c mimic. (B) Cells transfected with miR-376c mimic or miR-376c LNA. (C) Cells transfected with siARID4A. (B, C) ARID4A expression. Upper, qRT-PCR analysis to detect mRNA expression; Lower, Western blot analysis to detect protein expression. The numbers are normalized values. The pictures are from one representative analysis. (D-F) Phenotypic changes following the knockdown of ARID4A. (D) Proliferation. (E) Migration. (F) Anchorage-independent colony formation. Mann-Whitney test or two-way ANOVA. Each analysis is independently repeated twice. Data shown are mean ± SE in one representative analysis.

https://doi.org/10.1371/journal.pone.0177346.g005

Down-regulation of ARID4A expression in GC tumors

ARID4A mRNA expression was analyzed in randomly selected group 1 sample pairs to try to unravel the reasons behind the drastic down-regulation of ARID4A expression in the GC tumors (16/18, 88.9%; Fig 6A). The increased expression of miR-376a and miR-376c in this subset of tumors was highly correlated with a decrease in ARID4A expression (Fig 6B) and this was consistent with the findings outlined in Fig 1G. Although the correlation between expression of miR-376a and expression of ARID4A (Fig 6C) or between expression of miR-376c and expression of ARID4A (Fig 6D) was not significant, the profiling was able to demonstrate the presence of a trend towards a reverse correlation between miR-376 expression and ARID4A expression.

thumbnail
Download:
Fig 6. Down-regulation of ARID4A expression in GC.

Before-and-after blot. ARID4A mRNA expression in 18 pairs of GC and NCM tissues. Numbers within parenthesis, numbers of tissue pairs. Paired t-test. (B-D) Linear regression analysis in 18 pairs of GC and NCM tissues. (B) The analysis between the expression of miR-376a and miR-376c. (C) The analysis between the expression of miR-376a and the expression of ARID4A mRNA. (D) The analysis between the expression of miR-376c expression and the expression of ARID4A mRNA.

https://doi.org/10.1371/journal.pone.0177346.g006

The loss of ARID1A is one of the key factors that confounds GC survival [32]. According to a large scale bioinformatics study encompassing 4,623 tumors from multiple anatomical sites [39], ARID4A is one of the genes found within the chromatin regulatory element group that exhibits potential driver alterations in one or more cancer types (detailed analysis not shown). Although the average mutation frequency of ARID4A in all tumors is only 1.1%, the mutation frequency of ARID4A in GC is 13.6% (3/22); this is despite the fact that the mutation frequency of ARID1A compared to ARID4A is much higher (27.3%; 6/22) in tumors. Furthermore, in a TCGA-based cBioPortal database (http://www.cbioportal.org/), it has also been revealed that there is much more frequent ARID1A mutation (28%; 111/393) than ARID4A (6%; 28/393). However, an analysis of the 478 GC patients in the cBioPortal showed an immensely lower abundance of the ARID4A transcript (expression average value~800) relative to the ARID1A transcript (expression average value ~3700) (S2A Fig). Finally, the RNA-sequencing results from EMBL-EBI (http://www.ebi.ac.uk/) also show that there is a lower amount of the ARID4A transcript than ARID1A transcript present in eleven GC cell lines (S2B Fig). Taking all this informed together, it seems likely that the expression of ARID4A is low in GC.

The searching of cBioPortal database revealed the absence of miR-376c mutation in GC and the rare occurrence of miR-376c up-regulation in less than 3% GC tumors (S1 Table).

Discussion

In our previous study, we found that miR-370 was up-regulated and that it contributed to the progression of GC [9]. As an extension of our previous work, this study has confirmed that miR-376a and miR-376c are up-regulated in GC tissue. The miR-376 family is located on 14q32 and consists of several related miRNAs, including miR-376a1-a2, miR-376b, miR-376c (previously named as miR-368), miR-654, miR-B1, miR-B2 and so on. These miRNAs have highly similar sequences [40]. Studies have revealed that the miR-376 family plays important roles in cancer progression. It had been reported that miR-376b and miR-654 are down-regulated in human lung cancer [41, 42]. By way of contrast, miR-376a and miR-376c are up-regulated in breast cancer and miR-376a may be useful as a breast cancer detection marker [27, 43]. This study identifies for the first time that miR-376c is up-regulated in both primary GC and in the plasma of GC patients. The findings of increased miR-376c in the plasma of GC patients are in agreement with a previous study [26]. Furthermore, since the plasma level of miR-376c decreases after tumor resection, it is likely that the miR-376c found in the plasma originates from the tumor. However, the discrepancy between our cohort and cBioPortal database in the frequency of miR-376c up-regulation needs to be further clarified by adjusting multiple etiological, racial or clinical factors.

It is worth noting that our preliminary assays also show that miR-376c is up-regulated in the urine of GC patients. An ROC analysis further specified the level of miR-376c in urine as a means of identifying the presence of malignant tissue. Since the sampling of urine is more convenient and non-invasive relative to the sampling of plasma [22, 44], this assessment needs to be further extended in order to assist with the development of a simple clinical diagnosis system that can be used for GC screening.

Human chromosome 14q32 locus has been shown to be amplified in head and neck carcinoma cell lines [45] and in GC, especially in metastatic submucosal-invasive GC [46]. Furthermore, the translocation, loss of heterozygosity, and deletion of chromosome 14q32 has been reported in melanoma [47] and prostate cancer [48] and these changes may underlie the silencing of the miR-376 cluster. Our findings suggest that 14q32 amplification might be a possible reason for the up-regulation of both miR-376a and miR-376c in GC since they are up-regulated concordantly. Although miR-376a expression is high in tumor tissue, it was not found to be elevated in the patients’ blood. This might be due to barriers either affecting the secretion of miR-376a from neoplastic cells to their microenvironment or blocking the release of miR-376a from tumors into the circulation; this dichotomy needs to be addressed [49–51]. Alternatively, the relatively high abundance of miR-376a in the plasma of control individuals needs to be investigated in more detail.

Apart from the potency of plasma miR-376c as a means of distinguishing cancer patients from control individuals, the level of plasma miR-376c was also found to be up-regulated in tumors at an early stage of their neoplastic development. This implies that miR-376c could be used as a marker for early GC tumorigenesis. Although a previous study was unable to identify an increase in miR-376c in the serum of dysplastic gastric lesions [26], there remains a need to further explore the expression status of miR-376c during the various stages of gastric carcinogenesis. Our analysis verified that even patients carrying relatively better differentiated tumors also displayed an increased level of miR-376c in their plasma. These conflicts with a previous study indicating that poorly differentiated tumors had a higher serum level of miR-376c than tumors displaying better differentiation [26].

We have also shown that an induction of miR-376c expression increases the proliferation, migration and anchorage-independent growth ability of SCM-1 cells; furthermore, an inhibition of miR-376c expression repressed these phenotypes. These findings are important clues supporting the hypothesis that miR-376c is oncogenic in gastric carcinogenesis. Our in silico analysis using databases substantiated the fact that there is a low abundance of the ARID4A transcript in GC tissue and cell lines [32]. The reporter activity showed that miR-376c directly bound to the 3’-UTR of ARID4A gene and repressed the expression of luciferase gene. Furthermore, our functional assays were able to demonstrate that ARID4A has suppressor activity in the SCM-1 cell line. Decreased and increased expression levels of ARID4A were noted in GC cells following either miR-376c induction or inhibition, respectively. These findings, together with the fact that there is a trend of towards opposite expression levels patterns of miR-376c and ARID4A in tumor tissue samples, suggest the presence of a miR-376c-ARID4A regulation axis in GC. The clinical findings also suggest that the down-regulation of ARID4A in GC tumors may be modulated by various other factors in addition to miR-376c.

The survival rate of the late-stage GC patients is much lower than that of early-stage patients [1, 3]. However, the early diagnosis of GC is somewhat impeded by the lack of markers available for population screening. This study provides clues supporting the idea that plasma miR-376c may act as a potential marker for early GC. If urinary miR-376c can also be further validated, it could become an invaluable early non-invasive marker for GC. Finally, the fact that ARID4A down-regulation is driven by miR-376c up-regulation strongly suggests that this novel mechanism contributes to GC pathogenesis.

Supporting information

S1 Fig. Plasma level of miR-376c as related to age.

Whisker plots. (A) Analysis across control and GC patient subgroups divided by median age. (B) Analysis across GC patients and control subgroups divided by median age. The median ages for control individuals (group 3) and GC patients (group2) are years 70 and 57, respectively. +, mean value. Numbers within parenthesis, numbers of cases. Un-paired t-test.

https://doi.org/10.1371/journal.pone.0177346.s001

(DOCX)

S2 Fig. Analysis of ARID4A expression using bioinformatic domains.

(A) cBioPortal. ARID1A (Upper) and ARID4A (Lower) mRNA expression in 478 GC tumor tissues (TCGA, Provisional). (B) EMBL-EBI. ARID1A (Upper) and ARID4A (Lower) transcripts in 11 GC cell lines detected by RNA-sequencing.

https://doi.org/10.1371/journal.pone.0177346.s002

(DOCX)

S3 Fig. Original uncropped and unadjusted blots of Fig 5B—α-Tubulin blot.

Lt lane, Control treatment; Rt lane, miR-376c mimic treatment.

https://doi.org/10.1371/journal.pone.0177346.s003

(TIF)

S4 Fig. Original uncropped and unadjusted blots of Fig 5B—ARID4A blot.

Lt lane, Control treatment; Rt lane, miR-376c mimic treatment.

https://doi.org/10.1371/journal.pone.0177346.s004

(TIF)

S5 Fig. Original uncropped and unadjusted blots of Fig 5B—α-Tubulin blot.

Lt lane, Control treatment; Rt lane, miR-376c LNA treatment.

https://doi.org/10.1371/journal.pone.0177346.s005

(TIF)

S6 Fig. Original uncropped and unadjusted blots of Fig 5B–ARID4A blot.

Lt lane, Control treatment; Rt lane, miR-376c LNA treatment.

https://doi.org/10.1371/journal.pone.0177346.s006

(TIF)

S7 Fig. Original uncropped and unadjusted blots of Fig 5C–GAPDH blot.

Lt lane, siControl treatment; Rt lane, siARID4A treatment.

https://doi.org/10.1371/journal.pone.0177346.s007

(TIF)

S8 Fig. Original uncropped and unadjusted blots of Fig 5C–ARID4A blot.

Lt lane, siControl treatment; Rt lane, siARID4A treatment.

https://doi.org/10.1371/journal.pone.0177346.s008

(TIF)

S1 Table. miR-376c alterations in gastric carcinoma documented in cBioPortal database.

https://doi.org/10.1371/journal.pone.0177346.s009

(DOCX)

S2 Table. The clinical parameters and the qRT-PCR analysis of group 1 samples.

https://doi.org/10.1371/journal.pone.0177346.s010

(DOCX)

S3 Table. The clinical parameters and the qRT-PCR analysis of group 2 samples.

https://doi.org/10.1371/journal.pone.0177346.s011

(DOCX)

S4 Table. The clinical parameters and the qRT-PCR analysis of group 3 samples.

https://doi.org/10.1371/journal.pone.0177346.s012

(DOCX)

S5 Table. The plasma miR-376c expression in eight paired pre-surgical and post-surgical samples.

https://doi.org/10.1371/journal.pone.0177346.s013

(DOCX)

S6 Table. The urinarymiR-376c expression in 11 healthy controls and 20 GC patients.

https://doi.org/10.1371/journal.pone.0177346.s014

(DOCX)

S7 Table. The clinical parameters, miR-376 and ARID4A △△Ct values in 18 pairs of GC and NCM tissues.

https://doi.org/10.1371/journal.pone.0177346.s015

(DOCX)

Acknowledgments

We acknowledge the technical services provided by Clinical Medicine Research Laboratory of National Yang-Ming University Hospital. This study was supported by grant MOST-102-2314-B-010-043-MY3 from Ministry of Science and Technology, and National Yang-Ming University Hospital grants RD2014-001 and RD2015-001.

Author Contributions

  1. Conceptualization: SSL PSH.
  2. Data curation: PSH.
  3. Formal analysis: PSH CYK.
  4. Funding acquisition: SSL.
  5. Investigation: PSH PCC.
  6. Project administration: SSL.
  7. Resources: CYC WTC CYK WLF KHH.
  8. Supervision: SSL.
  9. Validation: PSH.
  10. Visualization: SSL PSH.
  11. Writing – original draft: SSL PSH.
  12. Writing – review & editing: SSL PSH.

References

  1. 1. Wu CW, Hsiung CA, Lo SS, Hsieh MC, Chen JH, Li AF, et al. Nodal dissection for patients with gastric cancer: a randomised controlled trial. Lancet Oncol. 2006;7(4):309–15. pmid:16574546
  2. 2. Ye G, Fu G, Cui S, Zhao S, Bernaudo S, Bai Y, et al. MicroRNA 376c enhances ovarian cancer cell survival by targeting activin receptor-like kinase 7: implications for chemoresistance. J Cell Sci. 2011;124(Pt 3):359–68. pmid:21224400
  3. 3. Lo SS, Wu CW, Chen JH, Li AF, Hsieh MC, Shen KH, et al. Surgical results of early gastric cancer and proposing a treatment strategy. Ann Surg Oncol. 2007;14(2):340–7. pmid:17094028
  4. 4. Gonzalez CA, Agudo A. Carcinogenesis, prevention and early detection of gastric cancer: where we are and where we should go. Int J Cancer. 2012;130(4):745–53. pmid:21918974
  5. 5. Figueiredo C, Garcia-Gonzalez MA, Machado JC. Molecular pathogenesis of gastric cancer. Helicobacter. 2013;18 Suppl 1:28–33.
  6. 6. Calcagno DQ, Gigek CO, Chen ES, Burbano RR, Smith Mde A. DNA and histone methylation in gastric carcinogenesis. World J Gastroenterol. 2013;19(8):1182–92. pmid:23482412
  7. 7. Conteduca V, Sansonno D, Lauletta G, Russi S, Ingravallo G, Dammacco F. H. pylori infection and gastric cancer: state of the art (review). Int J Oncol. 2013;42(1):5–18. pmid:23165522
  8. 8. Brennecke J, Stark A, Russell RB, Cohen SM. Principles of microRNA-target recognition. PLoS Biol. 2005;3(3):e85. pmid:15723116
  9. 9. Lo SS, Hung PS, Chen JH, Tu HF, Fang WL, Chen CY, et al. Overexpression of miR-370 and downregulation of its novel target TGFbeta-RII contribute to the progression of gastric carcinoma. Oncogene. 2012;31(2):226–37. pmid:21666718
  10. 10. Volinia S, Calin GA, Liu CG, Ambs S, Cimmino A, Petrocca F, et al. A microRNA expression signature of human solid tumors defines cancer gene targets. Proc Natl Acad Sci U.S.A. 2006;103(7):2257–61. pmid:16461460
  11. 11. Sassen S, Miska EA, Caldas C. MicroRNA: implications for cancer. Virchows Arch. 2008;452(1):1–10. pmid:18040713
  12. 12. Wu Q, Jin H, Yang Z, Luo G, Lu Y, Li K, et al. MiR-150 promotes gastric cancer proliferation by negatively regulating the pro-apoptotic gene EGR2. Biochem Biophys Res Commun. 2010;392(3):340–5. pmid:20067763
  13. 13. Zeng Z, Wang J, Zhao L, Hu P, Zhang H, Tang X, et al. Potential role of microRNA-21 in the diagnosis of gastric cancer: a meta-analysis. PLoS One. 2013;8(9):e73278. pmid:24023850
  14. 14. Zhang Z, Li Z, Gao C, Chen P, Chen J, Liu W, et al. miR-21 plays a pivotal role in gastric cancer pathogenesis and progression. Lab Invest. 2008;88(12):1358–66. pmid:18794849
  15. 15. Mendell JT. miRiad roles for the miR-17-92 cluster in development and disease. Cell. 2008;133(2):217–22. pmid:18423194
  16. 16. Xiao B, Zhu ED, Li N, Lu DS, Li W, Li BS, et al. Increased miR-146a in gastric cancer directly targets SMAD4 and is involved in modulating cell proliferation and apoptosis. Oncol Rep. 2012;27(2):559–66. pmid:22020746
  17. 17. Ahn SM, Cha JY, Kim J, Kim D, Trang HT, Kim YM, et al. Smad3 regulates E-cadherin via miRNA-200 pathway. Oncogene. 2012;31(25):3051–9. pmid:22020340
  18. 18. Gao X, Wang Y, Zhao H, Wei F, Zhang X, Su Y, et al. Plasma miR-324-3p and miR-1285 as diagnostic and prognostic biomarkers for early stage lung squamous cell carcinoma. Oncotarget. 2016.
  19. 19. Qiu X, Zhu H, Liu S, Tao G, Jin J, Chu H, et al. Expression and prognostic value of microRNA-26a and microRNA-148a in gastric cancer. J Gastroenterol Hepatol. 2016.
  20. 20. Tsai MM, Wang CS, Tsai CY, Huang CG, Lee KF, Huang HW, et al. Circulating microRNA-196a/b are novel biomarkers associated with metastatic gastric cancer. Eur J Cancer. 2016;64:137–48. pmid:27420607
  21. 21. Stuopelyte K, Daniunaite K, Bakavicius A, Lazutka JR, Jankevicius F, Jarmalaite S. The utility of urine-circulating miRNAs for detection of prostate cancer. Br J Cancer. 2016.
  22. 22. Hanke M, Hoefig K, Merz H, Feller AC, Kausch I, Jocham D, et al. A robust methodology to study urine microRNA as tumor marker: microRNA-126 and microRNA-182 are related to urinary bladder cancer. Urol Oncol. 2010;28(6):655–61. pmid:19375957
  23. 23. Sheng J, Luo W, Yu F, Gao N, Hu B. MicroRNA-376a sensitizes cells following DNA damage by downregulating MEPE expression. Cancer Biother Radiopharm. 2013;28(7):523–9. pmid:23570370
  24. 24. Zhang Y, Wu JH, Han F, Huang JM, Shi SY, Gu RD, et al. Arsenic trioxide induced apoptosis in retinoblastoma cells by abnormal expression of microRNA-376a. Neoplasma. 2013;60(3):247–53. pmid:23373993
  25. 25. Yabushita S, Fukamachi K, Tanaka H, Sumida K, Deguchi Y, Sukata T, et al. Circulating microRNAs in serum of human K-ras oncogene transgenic rats with pancreatic ductal adenocarcinomas. Pancreas. 2012;41(7):1013–8. pmid:22513294
  26. 26. Song MY, Pan KF, Su HJ, Zhang L, Ma JL, Li JY, et al. Identification of serum microRNAs as novel non-invasive biomarkers for early detection of gastric cancer. PLoS One. 2012;7(3):e33608. pmid:22432036
  27. 27. Cuk K, Zucknick M, Heil J, Madhavan D, Schott S, Turchinovich A, et al. Circulating microRNAs in plasma as early detection markers for breast cancer. Int J Cancer. 2013;132(7):1602–12. pmid:22927033
  28. 28. Cajuso T, Hanninen UA, Kondelin J, Gylfe AE, Tanskanen T, Katainen R, et al. Exome sequencing reveals frequent inactivating mutations in ARID1A, ARID1B, ARID2 and ARID4A in microsatellite unstable colorectal cancer. Int J Cancer. 2014;135(3):611–23. pmid:24382590
  29. 29. Wilsker D, Patsialou A, Dallas PB, Moran E. ARID proteins: a diverse family of DNA binding proteins implicated in the control of cell growth, differentiation, and development. Cell Growth Differ. 2002;13(3):95–106. pmid:11959810
  30. 30. Wang K, Kan J, Yuen ST, Shi ST, Chu KM, Law S, et al. Exome sequencing identifies frequent mutation of ARID1A in molecular subtypes of gastric cancer. Nat Genet. 2011;43(12):1219–23. pmid:22037554
  31. 31. Wiegand KC, Shah SP, Al-Agha OM, Zhao Y, Tse K, Zeng T, et al. ARID1A mutations in endometriosis-associated ovarian carcinomas. N Engl J Med. 2010;363(16):1532–43. pmid:20942669
  32. 32. Yang L, Wei S, Zhao R, Wu Y, Qiu H, Xiong H. Loss of ARID1A expression predicts poor survival prognosis in gastric cancer: a systematic meta-analysis from 14 studies. Sci Rep. 2016;6:28919. pmid:27354232
  33. 33. Chittenden T, Livingston DM, Kaelin WG Jr. The T/E1A-binding domain of the retinoblastoma product can interact selectively with a sequence-specific DNA-binding protein. Cell. 1991;65(6):1073–82. pmid:1828394
  34. 34. Rodrigues-Lisoni FC, Peitl P Jr., Vidotto A, Polachini GM, Maniglia JV, Carmona-Raphe J, et al. Genomics and proteomics approaches to the study of cancer-stroma interactions. BMC Med Genomics. 2010;3:14. pmid:20441585
  35. 35. Wu MY, Eldin KW, Beaudet AL. Identification of chromatin remodeling genes Arid4a and Arid4b as leukemia suppressor genes. J Natl Cancer Inst. 2008;100(17):1247–59. pmid:18728284
  36. 36. Davoren PA, McNeill RE, Lowery AJ, Kerin MJ, Miller N. Identification of suitable endogenous control genes for microRNA gene expression analysis in human breast cancer. BMC Mol Biol. 2008;9:76. pmid:18718003
  37. 37. Hu J, Wang Z, Liao BY, Yu L, Gao X, Lu S, et al. Human miR-1228 as a stable endogenous control for the quantification of circulating microRNAs in cancer patients. Int J Cancer. 2014;135(5):1187–94. pmid:24488924
  38. 38. Chin LT, Meng CL, Chang TM, Chen TK, Shen KL. Establishment and characterization of human gastric cancer cell line. J Surg Asso ROC 1987;19:27.
  39. 39. Gonzalez-Perez A, Jene-Sanz A, Lopez-Bigas N. The mutational landscape of chromatin regulatory factors across 4,623 tumor samples. Genome Biol. 2013;14(9):r106. pmid:24063517
  40. 40. Kawahara Y, Zinshteyn B, Sethupathy P, Iizasa H, Hatzigeorgiou AG, Nishikura K. Redirection of silencing targets by adenosine-to-inosine editing of miRNAs. Science. 2007;315(5815):1137–40. pmid:17322061
  41. 41. Kim SC, Jung Y, Park J, Cho S, Seo C, Kim J, et al. A high-dimensional, deep-sequencing study of lung adenocarcinoma in female never-smokers. PLoS One. 2013;8(2):e55596. pmid:23405175
  42. 42. Pang AL, Title AC, Rennert OM. Modulation of microRNA expression in human lung cancer cells by the G9a histone methyltransferase inhibitor BIX01294. Oncol Lett. 2014;7(6):1819–25. pmid:24932239
  43. 43. Cuk K, Zucknick M, Madhavan D, Schott S, Golatta M, Heil J, et al. Plasma microRNA panel for minimally invasive detection of breast cancer. PLoS One. 2013;8(10):e76729. pmid:24194846
  44. 44. Zhou J, Gong G, Tan H, Dai F, Zhu X, Chen Y, et al. Urinary microRNA-30a-5p is a potential biomarker for ovarian serous adenocarcinoma. Oncol Rep. 2015;33(6):2915–23. pmid:25962395
  45. 45. Singchat W, Hitakomate E, Rerkarmnuaychoke B, Suntronpong A, Fu B, Bodhisuwan W, et al. Genomic Alteration in Head and Neck Squamous Cell Carcinoma (HNSCC) Cell Lines Inferred from Karyotyping, Molecular Cytogenetics, and Array Comparative Genomic Hybridization. PLoS One. 2016;11(8):e0160901. pmid:27501229
  46. 46. Kuroda A, Tsukamoto Y, Nguyen LT, Noguchi T, Takeuchi I, Uchida M, et al. Genomic profiling of submucosal-invasive gastric cancer by array-based comparative genomic hybridization. PLoS One. 2011;6(7):e22313. pmid:21811585
  47. 47. Zehavi L, Avraham R, Barzilai A, Bar-Ilan D, Navon R, Sidi Y, et al. Silencing of a large microRNA cluster on human chromosome 14q32 in melanoma: biological effects of mir-376a and mir-376c on insulin growth factor 1 receptor. Mol Cancer. 2012;11:44. pmid:22747855
  48. 48. Formosa A, Markert EK, Lena AM, Italiano D, Finazzi-Agro E, Levine AJ, et al. MicroRNAs, miR-154, miR-299-5p, miR-376a, miR-376c, miR-377, miR-381, miR-487b, miR-485-3p, miR-495 and miR-654-3p, mapped to the 14q32.31 locus, regulate proliferation, apoptosis, migration and invasion in metastatic prostate cancer cells. Oncogene. 2014;33(44):5173–82. pmid:24166498
  49. 49. Chen X, Liang H, Zhang J, Zen K, Zhang CY. Secreted microRNAs: a new form of intercellular communication. Trends Cell Biol. 2012;22(3):125–32. pmid:22260888
  50. 50. Turchinovich A, Weiz L, Langheinz A, Burwinkel B. Characterization of extracellular circulating microRNA. Nucleic Acids Res. 2011;39(16):7223–33. pmid:21609964
  51. 51. Creemers EE, Tijsen AJ, Pinto YM. Circulating microRNAs: novel biomarkers and extracellular communicators in cardiovascular disease? Circ Res. 2012;110(3):483–95. pmid:22302755