Experimental and Molecular Pathology 96 (2014) 195–199

Contents lists available at ScienceDirect

Experimental and Molecular Pathology

j ourna l homepage: www.e lsev ie r .com/ locate /yexmp

microRNA-146 up-regulation predicts the prognosis of non-small celllung cancer by miRNA in situ hybridization

Jiangchao Li a,1, Hong Yang b,1, Yan Li c, Ying Liu a, Shupeng Chen c, Cuiling Qi a, Qianqian Zhang a, Tian Lan a,Xiaodong He a, Xin-Yuan Guan c,⁎, Lijing Wang a,⁎⁎a Vascular Biology Research Institute, Guangdong Pharmaceutical University, Guangzhou 510006, Chinab Department of Thoracic Surgery, Sun Yat-Sen University Cancer Center, Guangzhou, Chinac State Key Laboratory of Oncology in Southern China, Sun Yat-Sen University Cancer Center, Guangzhou, China

⁎ Correspondence to: X.-Y. Guan, State Key LaboratorySun Yat-Sen University Cancer Center. Dongfeng E. RoaTel:. +86 20 87343165.⁎⁎ Correspondence to: L. Wang, Vascular Biology RPharmaceutical University, Guangzhou 510006, China. Tel

E-mail addresses: xyguan@hku.hk (X.-Y. Guan), wang1 Authors contributed equally to this work.

0014-4800/$ – see front matter. Crown Copyright © 2013http://dx.doi.org/10.1016/j.yexmp.2013.11.004

a b s t r a c t

a r t i c l e i n f o

Article history:Received 11 June 2013and in revised form 1 November 2013Available online 18 January 2014

Keywords:microRNA-146NSCLCISHPrognosis

Non-small cell lung cancer (NSCLC) accounts for approximately 70% of all lung cancer-related deaths worldwide.Prognostic markers are essential for the early detection of lung cancer in patients. In this study, we first identifiedmicroRNA146 (miR-146) expression in cancer cell lines using miRNA in situ hybridization (MISH) and confirmedthe accuracy of MISH using q-RT-PCR. In addition, two different systems, BCIP/NBT and ELF, were used to detectthe signal for a comparative analysis of the specificity ofMISH. Compared to the BCIP/NBT system, the ELF detectionsystem was more effective for MISH. Furthermore we detected the expression of miR-146 in NSCLC tissues (43cases) and normal tissues (32 cases). Based on our results, we can conclude that miR-146 is more highly expressedin cancer tissue than normal tissue (t-test, P b 0.05) and that miR-146 can predict the prognosis of NSCLC byMISH.Our findings preliminary demonstrate that MISH can be applied as amolecular diagnostic tool to determine the ex-pression and localization ofmiRNAs in cancer tissues and thatmiR-146, determined byMISH, predicts the prognosisof NSCLC patients.

Crown Copyright © 2013 Published by Elsevier Inc. All rights reserved.

Introduction

Non-small cell lung cancers (NSCLCs) are the leading cause of cancermortality, and the overall 5-year survival of NSCLC patients is notmore than 15% (Jemal et al., 2009). If one can diagnose NSCLC at anearly stage, the survival rate of patients can be effectively improved.Therefore, some specific early molecular markers are urgently need-ed for the detection of lung tumors. microRNAs (miRNAs) may openthe door to the early molecular diagnosis of NSCLC and allow riskstratification based on miRNA profiles (Cho, 2011; Mendell, 2005;Xing et al.).

miRNAs are small non-coding RNAs of approximately 20–23 nucleo-tides that can regulate cancer cell processes by annealing to the 3′ UTRsof target genes (Griffiths-Jones et al., 2006). Some miRNAs are impor-tant as oncogenes or anti-oncogenes (Cho et al., 2011; He et al., 2005);many of these can facilitate the development of lung cancer as hasbeen reported in previous studies (Bishop et al.). Meanwhile, lungcancer-related miRNA profiles have been established (Gee et al.;Hayashita et al., 2005; Vaporidi et al., 2012). For example, the increased

of Oncology in Southern China,d, Guangzhou, 510070, China.

esearch Institute, Guangdong:. +86 20 39352126.lijing@gdpu.edu.cn (L. Wang).

Published by Elsevier Inc. All rights

expression of miR-146 was correlated with poor prognosis in lung can-cer patients.

Currently, analyses of miRNA expression are usually performedby q-RT-PCR, Northern blotting using miRNA probes and microchiparray technology (Hu et al.; van Rooij, 2011). However, these tech-niques cannot detect the localization of miRNAs in a tissue or singlecell; miRNA in situ hybridization (MISH) can accomplish this(Neely et al., 2006). Perhapsmore importantly, there are great differ-ences between the expression profiles of miRNAs in cancer cells andnon-cancer cells. Our previous study demonstrated that MISH can beapplied to detect the expression of miRNA375 in paraffin-embeddedesophageal cancer tissue (J. Li et al.; Y. Li et al.). In this work, we over-came the technical difficulty that the melting temperature (Tm) ofthemiRNA probe hybridization complex is too low to allow for its de-tection. Additionally, we further analyzed miR-146 to confirm itsvalue. Several studies have reported that miR-146 is associatedwith lung cancer (Perry et al., 2009), but further details on the specif-ic subcellular compartmentalization of miR-146 have not been re-ported. It is not possible to use RT-PCR or other methods to detectthe expression of a miRNA in single cells or at low copy expression(Lu and Tsourkas, 2009).

Therefore, our study focuses on MISH and verifies the accuracy ofthis technique using q-RT-PCR. Briefly, we describe the MISH methodto determine miR-146 expression in cultured cells and paraffin-embedded tissues and use it to detect the expression of miR-146 inNSCLC. The primary objective of this work is to enrich themolecular de-tection of miRNAs in cancer tissues.

reserved.

196 J. Li et al. / Experimental and Molecular Pathology 96 (2014) 195–199

Materials and methods

Cell culture and tissue treatment

Non-small-cell lung cancer cell lines ACC212102 and SCC211441(established by our lab) and KYSE140, KYSE180, KYSE510, and HKESC1cell lineswere cultured in DMEMwith 10% FBS supplementedwith L-glu-tamine. The cells were cultured on glass cover slides at 37 °C in a 5% CO2

environment.When the cells reached70%–80% confluence, theywere im-mediately fixed in 10% formalin for 2 h andMISHwas performed on glasscover slides as described below.

Clinical specimen collection

The expression of miR-146 was evaluated in a total of 43 non-smallcell lung cancer and 32matched normal adjacent lung tissue samples by

Fig. 1.MISH results are consistentwith RT-PCR. (1)miRNA expressionwas analyzed in a series oabove. The gray mean is shown in C. (2) In addition, the expression of miR-146 in these cell linMISH detection is positively correlated with the data obtained from RT-PCR. The correlation co

miRNA in situ hybridization (MISH). The samples included 20 lungadenocarcinoma samples, 17 squamous cell carcinoma samples, 6adeno-squamous carcinomas, and no large-cell carcinomas. All thespecimens were collected from patients in the Department of ThoracicSurgery, Sun Yat-Sen University Cancer Center, Guangzhou, Chinafrom October 2009 to July 2012 with patient consent and institutionalreview board approval.

ISH probes

An oligonucleotide probe, complementary to the hsa-miR-146b-5p(miR-146) probe, was purchased from Exonbio Lab (Guangzhou,China). The sequence of the probe is 5′-AGCCTATGGAATTCAGTTCTCA-3′; the 5′ and 3′ ends were modified with digoxigenin (DIG). Someof these bases were modified with 2-fluorine. A probe with the se-quence5′-AGCGTATGGAATTCAGATCTCA-3′ served as control probe.

f cell lines byMISH, according to the data transformation as shown inA and B, as describedes was quantified using RT-PCR, shown in a histogram (D). The expression of miR-146 byefficient (R2 = 0.93) of the two methods is significant (P b 0.05, E and F).

197J. Li et al. / Experimental and Molecular Pathology 96 (2014) 195–199

MISH

In situ hybridization (ISH) for miR-146 was performed on fixedparaffin-embedded sections as previously described (J. Li et al.; Y. Liet al.). The MISH procedure was carried out as follows: The paraffinwax was removed from the tissue sections with fresh xylene threetimes, and then the samples were washed in PBS and RNase-freewater for 3 min. Next, the tissues were digested with protease in 0.1 NHCl for 10 min and 0.1% PBS for 5 min. We recommend proteinase Kto treat tissue that has beenfixed formore than sixmonths. The protein-ase K concentration is 20 mg/ml, and working dilutions are 1:100 inPBS. Notably, proteinase K may lead to over-digestion, which cancause loss of tissue morphology. After digestion, the samples were im-mersed in RNase-free water and air-dried, and then they were pre-hybridized and hybridized withmiRNA probe or control probe at a con-centration of 0.5 μM. The sampleswere incubated at 37 °C for 24 h; thentwo stringent washes were carried out in 2 × SSC containing Tween-20at 25 °C for 2 min and 2 × SSC for 2 min at room temperature. ISH wasperformed twice on each slide.

Detection systems

The ELF detection (Paragas et al., 1997) kit (Invitrogen, San Diego,CA) utilizes substrate cleavage by phosphatase to produce a yellow–

green fluorescent signal at the site of enzymatic activity. The ELF precip-itate is up to 40 times brighter than common probes directly labeledwith fluorophores. The probe was labeled by DIG combined with anti-Dig conjugated biotin, and detection was carried out according to themanufacturer's specifications. The slides were counterstained withHoechst 33342 (1 μg/ml).

The NCIP/NBT detection system was purchased from Roche Diagnos-tics. The probe was labeled with DIG combined with anti-Dig conjugatedalkaline phosphatase (AP), and the cells were stainedwith NCIP/NBT. Thenuclei of cells were stained by Nuclear Fast Red.

Fig. 2. Comparison between the specificity of the ELF and BCIP/NBT detection systems. (1) The uand pericancerous tissues, respectively. (2) D shows high expression of miR-146 obtained by ththe arrow indicates a few cells specifically expressingmiR-146. C and F are controls. All images athe miR-146 than the BCIP/NBT detection system.

Real-time RT-PCR of miRNA

Total RNA was isolated from tissues using Trizol reagent (Invitrogen,San Diego, CA). According to the instructions of Ribobio primer (theBulge-loop™), the miRNA qRT-PCR primer sets included one RT primerand a pair of qPCR primers that are specific formiR-146b-5p (Guangzhou,China) for the RT-PCR. Briefly, 100 ng RNA was added to each reversetranscription reaction. The PCR system was as follows: 62.5 nM U6 or62.5 nM miR-146b-5p RT primer (2 μl at 5 μM), 5 μl 5× RT buffer, 2 μlof a 2.5mMdNTPmix, 100 U of reverse transcriptase, and 20 U of reversetranscriptase inhibitor (Invitrogen, San Diego, CA). Reactions were incu-bated at 42 °C for 60 min and 70 °C for 15 min. The real-time PCR wasperformed as follows: 10 μl of 2× SYBR mix (ABI system, USA), 2 μl ofRT product, and 2 mM miR-146b-5p or U6 primer (2 μl at 5 μM). ThePCR conditions were as follows: pre-denature at 95 °C for 30 s; 95 °Cfor 10 s, 60 °C for 20 s, and 70 °C for 10 s; 40 cycles. The PCR experimentswere performed and analyzed on an ABI Prism 7900HT Detection System(Applied Biosystems, USA). U6 was used as an internal control. The datawere normalized to internal controls. Quantitation of the PCR data wascarried out using the ΔΔCT method as described previously (J. Li et al.;Y. Li et al.).

Quantification of MISH images, analysis and statistics

To investigate the accuracy of MISH in cell lines, a qRT-PCR methodwas adopted to quantify the expression of miRNA. To validate if MISHaccurately reflects the expression of miR-146 in cell lines, we wantedto analyze whether the expression level of miR-146 detected by MISHcorrelated with the RT-PCR results. To accomplish this, we comparedthemean gray value of miR-146 MISH per cell with the values obtainedby quantitative RT-PCR. The MISH images were analyzed using imageJ software using the Tool/image/split of the color menu option. The re-sults are shown in Fig. 1; the green signals were used to analyze thegray value, and the gray signals were then measured using the mean

pper panels show the BCIP/NBT detection system results; A and B show lung cancer tissuee ELF detection system; the photographs in E and B display pericancerous tissues, inwhichremagnified 400×. These results demonstrate that the ELF-MISH system can better detect

198 J. Li et al. / Experimental and Molecular Pathology 96 (2014) 195–199

gray value program in the software. The cells were imaged with anOlympus BX51 fluorescence microscope. The nucleus was stained byHoechst to evaluate the co-localization of miRNA.

Results

Analysis of the accuracy of miR-146 ISH using qRT-PCR

To investigatewhetherMISH can be applied to accurately determinethe expression of miRNAs and whether MISH results accurately reflectthe real expression of miRNAs (Fig. 1A), we compared the results ofMISH experiments to those of RT-PCR. First, we measured the MISHspots, which corresponded to an individual miR-146, and next, wecounted 20–40 cells in cell culture and obtained the mean gray valueof miR-146 expression using image J software (the format of Fig. 1A isidentical to that of B). A quantification of miR-146 using RT-PCR in thesame cell lines (Fig. 2C) revealed thatMISH could reflect the relative ex-pression of miRNA in single cells (Fig. 1D). However, in cases with highmiR-146 expression,we could not determine the actual expression levelbecause it was difficult to discern the fluorescent gray value using thesoftware. Nonetheless, the approximate miRNA copy numbers couldstill be estimated in the single cells by MISH because there is a linearcorrelation between the expression of miRNA by RT-PCR and themean gray value (the total cellular fluorescence in a single cell).The R2 value measuring this correlation is significant (Figs. 1E andF, P b 0.05, R2 = 0.9329).

Fig. 3. High expression of miR-146 is associated with poor prognosis. (1) A representative phocancer nest, which is separated from the peritumoral tissue by the dotted line (A); miR-146expression of miR-146 in cancer tissue by MISH. Among 43 cases, 31 cases show high miR-146 ecarcinoma tissues (A and B). (3) The up-regulation of miR-146 expression was significantly assoc

Confirming the specificity and sensitivity ofmiRNA ISHusing EF andBCIP/NBTdetection systems

To further validate the specificity and sensitivity of the miR-146probe for MISH, we evaluated two detection systems: the EF and BCIP/NBT detection systems, which were previously used for mRNA ISH(Millet et al., 1996) or IHC. Both adapted detection systems can revealthe miR-146 expression signals and detect single cells expressing miR-146, as shown in Fig. 2. The representative images in Figs. 2A and Bwere obtained with the BCIP/NBT detection system, and the images inFigs. 2D and E were obtained with the ELF detection system for miRNAISH. From the data, we can deduce that the BCIP/NBT detection systemis less effective than the ELF detection system. Alternatively, the resultssuggested that MISH with the ELF kit is superior to the conventionalBCIP/NBT detection system in cancer tissue and that the sensitivity ofthe ELF detection system is dramatically higher than that of BCIP/NBT.Unfortunately, the shortcoming of ELF is also noticeable; the ELF kit hasa high background (Fig. 2D), which could lead to difficulties in differenti-ating the miRNA expression. Figs. 2C and F represent control samples.

The up-regulation of miR-146 in NSCLC is associated with poor prognosis

To further apply and optimize themethod for diagnosis, we examinedthe expression of miR-146 in 43 lung NSCLC and 32 matched adjacentnormal lung tissue samples using MISH. The expression of miR-146 ishigher in NSCLC than in adjacent normal lung tissue (Fig. 3A). It has

to of miR-146 expression in cancer tissue and the peritumoral tissue. The “Ca” indicates ais more highly expressed in the cancerous tissue than in the peritumoral tissue. (2) Thexpression. The results demonstrate that the expression of miR-146 is up-regulated in NSCLCiated with poor overall survival (C, P b 0.05, Log risk = 10.56) in NSCLC patients.

199J. Li et al. / Experimental and Molecular Pathology 96 (2014) 195–199

been reported that the expression ofmiR-146 (analyzed by RT-PCR) is as-sociated with the clinical prognosis of NSCLC, and this is consistent withour results. The score standardwas defined as “high” and “low” accordingto the representative Figs. 3A and B. The correlation between miR-146down-regulation and clinical–pathological features was analyzed bySPSS 16.0 software. The up-regulation ofmiR-146 is significantly correlat-edwith the overall survival rate (one-wayANOVA, P b 0.05) but is not as-sociated with other clinical characteristics (table not shown). To furtherinvestigate the prognostic value of miR-146 in NSCLC, Kaplan–Meieranalysis was carried out; the high expression of miR-146 can be asso-ciated with overall survival in the NSCLC cohort (Fig. 3C, P = 0.003, Logrisk = 10.56). The median expression of miR-146 was used to classifythe patients into two halves. The group with high miR-146b expressionhad a significantly worse overall survival rate (16.8 months) comparedto the group with low miR-146b expression (57.7 months).

Discussion

The expression ofmiRNAs varieswith the development of lung cancer(Izzotti et al., 2009), and most of these miRNA variations emerge duringtumorigenesis (Landi et al.; Lu et al., 2005). The methods of determiningmiRNAs included q-RT-PCR, Northern blotting and miRNA chip arrays.FISH is a traditional method to detect DNA localization or mRNA expres-sion. However, there are few reports using FISH for diagnosis in cancertissue (Hermansen et al.; Sempere and Korc). However, the advantagesof subcellular localization are currently being re-evaluated in light ofnew molecular factors such as miRNAs. In our previous study, we opti-mized a MISH method by testing miR-375 in ESCC (J. Li et al.; Y. Liet al.). Herein, we further verified the expression of miR-146 in lung can-cer tissue. Consistent with our expectations, miR-146 is highly expressedin NSCLC and can specifically detect the high expression of miR-146 inlung cancer but not in normal tissue.

The data fromMISH have shown that miR-146 is highly expressed inNSCLC tissue, which is consistent with previous studies using RT-PCR.We compared the results of two different MISH detection systems in-cluding ELF andBCIP/NBT; the resultswere similar, but the ELF detectionmethod is better suited for MISH because it can identify single express-ing cancer cells. In addition, miR-146 was also tested using RT-PCR incell lines in this study, and the results are consistent withMISH. Accord-ing to this study, high expression ofmiR-146 is associatedwith poor out-comes in NSCLC. Therefore, a MISHmethod to detect miR-146 may be avalid tool for predicting the patient survival rate in lung cancer.

MiRNAs are more stable than mRNAs, which enables them to bereadily detected in formalin-fixed paraffin-embedded tissue (Konget al.). Moreover, our study shows that miRNA ISH is an effective wayto determine the expression of a miRNA. In summary, we have demon-strated that miR-146 expression can be accurately detected by MISHand these experimental data can be used to analyze patient prognosis.MISHmay provide a unique method for miRNA detection and may pre-dict the prognosis of lung cancer (Raponi et al., 2009). In addition, weprovided support for the use ofMISH in studyingmiRNAor inmoleculardiagnosis. Additional studies should continue work on these aspects ofMISH technology.

Conflicts of interest statement

The authors disclose no potential conflicts of interest.

Acknowledgments

This work is supported by the Medical Scientific Research Founda-tion of Guangdong Province (No. A2013312) and the National NaturalScience Foundation of China (No. 30971606, No. 81172338 and No.81000863).

References

Bishop, J.A., et al., 2010. Accurate classification of non-small cell lung carcinoma using anovel microRNA-based approach. Clin Cancer Res. 16, 610–619.

Cho, W.C., 2011. Promises and challenges in developing miRNA as a molecular diagnostictool for lung cancer. Expert. Rev. Mol. Diagn. 11, 763–766.

Cho, W.C., et al., 2011. MiR-145 inhibits cell proliferation of human lung adenocarcinomaby targeting EGFR and NUDT1. RNA Biol. 8, 125–131.

Gee, G.V., et al., 2010. Downregulated MicroRNAs in the differential diagnosis of malig-nant pleural mesothelioma. Int J Cancer.

Griffiths-Jones, S., et al., 2006. miRBase: microRNA sequences, targets and gene nomen-clature. Nucleic Acids Res. 34, D140–D144.

Hayashita, Y., et al., 2005. A polycistronic microRNA cluster, miR-17–92, is overexpressedin human lung cancers and enhances cell proliferation. Cancer Res. 65, 9628–9632.

He, L., et al., 2005. A microRNA polycistron as a potential human oncogene. Nature 435,828–833.

Hermansen, S.K., et al., 2012. MiR-21 expression in the tumor cell compartment holds un-favorable prognostic value in gliomas. J Neurooncol. 111, 71–81.

Hu, Z., et al., 2010. Serum microRNA signatures identified in a genome-wide serummicroRNA expression profiling predict survival of non-small-cell lung cancer. J ClinOncol. 28, 1721–1726.

Izzotti, A., et al., 2009. Downregulation of microRNA expression in the lungs of rats ex-posed to cigarette smoke. FASEB J. 23, 806–812.

Jemal, A., et al., 2009. The convergence of lung cancer rates between blacks andwhites underthe age of 40, United States. Cancer Epidemiol. Biomarkers Prev. 18, 3349–3352.

Kong, K.L., et al., 2012. MicroRNA-375 inhibits tumour growth and metastasis in oesoph-ageal squamous cell carcinoma through repressing insulin-like growth factor 1 recep-tor. Gut.

Landi, M.T., et al., 2010. MicroRNA expression differentiates histology and predicts surviv-al of lung cancer. Clin Cancer Res. 16, 430–441.

Li, J., et al., 2013. Cell-specific detection of miR-375 downregulation for predicting theprognosis of esophageal squamous cell carcinoma by miRNA in situ hybridization.PLoS One 8, e53582.

Li, Y., et al., 2011. Downregulation of RBMS3 is associated with poor prognosis in esoph-ageal squamous cell carcinoma. Cancer Res.

Lu, J., Tsourkas, A., 2009. Imaging individual microRNAs in single mammalian cells in situ.Nucleic Acids Res. 37, e100.

Lu, J., et al., 2005. MicroRNA expression profiles classify human cancers. Nature 435,834–838.

Mendell, J.T., 2005. MicroRNAs: critical regulators of development, cellular physiologyand malignancy. Cell Cycle 4, 1179–1184.

Millet, S., et al., 1996. The caudal limit of Otx2 gene expression as a marker of the mid-brain/hindbrain boundary: a study using in situ hybridisation and chick/quailhomotopic grafts. Development 122, 3785–3797.

Neely, L.A., et al., 2006. A single-molecule method for the quantitation of microRNA geneexpression. Nat. Methods 3, 41–46.

Paragas, V.B., et al., 1997. The ELF-97 alkaline phosphatase substrate provides abright, photostable, fluorescent signal amplification method for FISH. J. Histochem.Cytochem. 45, 345–357.

Perry, M.M., et al., 2009. Divergent intracellular pathways regulate interleukin-1beta-induced miR-146a and miR-146b expression and chemokine release in human alve-olar epithelial cells. FEBS Lett. 583, 3349–3355.

Raponi, M., et al., 2009. MicroRNA classifiers for predicting prognosis of squamous celllung cancer. Cancer Res. 69, 5776–5783.

Sempere, L.F., Korc, M., 2013. A method for conducting highly sensitive microRNA in situhybridization and immunohistochemical analysis in pancreatic cancer. Methods MolBiol 980, 43–59.

van Rooij, E., 2011. The art of microRNA research. Circ. Res. 108, 219–234.Vaporidi, K., et al., 2012. Pulmonary microRNA profiling in a mouse model of

ventilator-induced lung injury. Am. J. Physiol. Lung Cell. Mol. Physiol. 303,L199–L207.

Xing, L., et al., 2010. Early detection of squamous cell lung cancer in sputum by a panel ofmicroRNA markers. Mod Pathol. 23, 1157–1164.