Cellular Signalling 26 (2014) 1420–1426

Contents lists available at ScienceDirect

Cellular Signalling

j ourna l homepage: www.e lsev ie r .com/ locate /ce l l s ig

MiR-135a functions as a tumor suppressor in epithelial ovarian cancerand regulates HOXA10 expression

Weiwei Tang a,b, Yi Jiang a, Xiaoxin Mu c, Lei Xu a,d, Wenjun Cheng a,⁎, Xinru Wang e,f,⁎⁎a Department of Gynecology, The First Affiliated Hospital of Nanjing Medical University, Nanjing, Chinab Department of Gynecology and Obstetrics, Jiangsu Province Hospital on Integration of Chinese and Western Medicine, Nanjing, Chinac Liver Transplantation Center, The First Affiliated Hospital of Nanjing Medical University, Nanjing, Chinad Department of Gynecology and Obstetrics, Nanjing Maternity and Child Health Care Hospital, Nanjing, Chinae State Key Laboratory of Reproductive Medicine, Institute of Toxicology, Nanjing Medical University, Nanjing, Chinaf Key Laboratory of Modern Toxicology of Ministry of Education, School of Public Health, Nanjing Medical University, Nanjing, China

Abbreviations: EOC, epithelial ovarian cancer; HOXmicroRNAs; UTRs, untranslated regions.⁎ Correspondence to:W. Cheng, Department of Gynecol

of Nanjing Medical University, Nanjing 210029, China. Te⁎⁎ Correspondence to: X. Wang, State Key LaboratoInstitute of Toxicology, Nanjing Medical UniversiTel./fax: +86 25 86862863.

E-mail addresses: chengwenjundoc@163.com (W. Che(X. Wang).

http://dx.doi.org/10.1016/j.cellsig.2014.03.0020898-6568/© 2014 Elsevier Inc. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 17 February 2014Accepted 2 March 2014Available online 6 March 2014

Keywords:Epithelial ovarian cancerMiR-135aHOXA10ApoptosisAdhesion

The activation of homeobox A10 (HOXA10) has been proved to be an important event in epithelial ovariancarcinogenesis, yet its regulation in epithelial ovarian cancer (EOC) is still not fully understood. Here, weaimed to reveal the mechanism that a predicted target miRNA regulates HOXA10 expression and the associationof its expression with progression of EOC. Here, by using computer-assisted algorithms from PicTar, TargetScan,and miRBase, we identified that the predicted target miRNA of HOXA10 was miR-135a. MiR-135a expression inEOC tissues and controls was measured with quantitative RT-PCR. The role of miR-135a and HOXA10 in thegrowth and survival of several EOC cell lines was determined with several in vitro approaches. We found thatmiR-135a expressionwas downregulated in an EOC patient cohort. Also, patients with lowmiR-135a expressionhad shorter overall survival and progression-free survival durations than those with high expression. Functionalanalysis of three EOC-derived cell lines (SKOV-3, HEY, and OVCAR-3) demonstrated that miR-135a directlyregulated HOXA10 expression by targeting its 3′-UTR. Inhibition of HOXA10 expression with miR-135a mimicsand HOXA10 siRNA consistently resulted in cell apoptosis with concomitant enhancement of caspase-3, increaseof p53 expression and reduction of Bcl-2 expression, and also suppressed cell growth and adhesion. Thesefindings suggest that ubiquitous loss of miR-135a expression is a critical mechanism for the overexpression ofHOXA10 in EOC cells, which is implicated in epithelial ovarian carcinogenesis. Furthermore, miR-135a may bepredictive of EOC prognosis.

© 2014 Elsevier Inc. All rights reserved.

1. Introduction

Epithelial ovarian cancer (EOC) is themost prevalent type of ovariancancer and a leading cause of cancer deaths inwomenworldwide. In theUSA, researchers estimated that in 2014, 21,980 new ovarian cancercases would be diagnosed and 14,270 patients would die of this disease[1]. Surgery and chemotherapy are increasingly used to treat EOC, butthey have not resulted in significant improvements in EOC survivaldurations. Therefore, identifying the molecular mechanisms of EOC

A10, homeobox A10; miRNAs,

ogy, The First AffiliatedHospitall./fax: +86 25 86211033.ry of Reproductive Medicine,ty, Nanjing 210029, China.

ng), xrwang@njmu.edu.cn

carcinogenesiswill help improve the understanding of the pathogenesisof EOC and ultimately identify novel diagnostic and/or prognosticmarkers for it.

MicroRNAs (miRNAs) are small (19–25 nt), endogenous noncodingRNAs that are novel posttranscriptional regulators of gene expression.These regulators mediate their own effects on all aspects ofgrowth and differentiation by binding to complementary sequences in3′-untranslated regions (UTRs) of their target genes via their ownRNA-induced silencing complexes [2]. The actions ofmiRNAs are varied,as they play essential roles in regulation of glucose metabolism [3], cellproliferation, apoptosis, differentiation, and biological development[4–7]. Additionally, miRNAs are involved in tumorigenesis and cancermetastasis by regulating the expression of their target oncogenes ortumor suppressor genes [8–12]. In recent years, researchers hypothe-sized that a large number of miRNAs have abnormal expression pat-terns in many different cancers including epithelial ovarian cancer[13,14], and meanwhile, many miRNAs have been proven to be predic-tive of ovarian cancer prognosis and metastasis andmay be biomarkersfor ovarian cancer detection [15,16].

1421W. Tang et al. / Cellular Signalling 26 (2014) 1420–1426

HOXA10 is an abdominal-B-like HOX gene that encodes forhomeodomain regulatory proteins that control cell growth and dif-ferentiation during embryonic and uterine development [17,18]. Inaddition, a number of reports on the functions of HOXA10 in tumorigen-esis have emerged in the literature over the past decade [19,20],which in-clude the relation between HOXA10 and ovarian cancers [19–21].Although HOXA10 is a modulator rather than the driving force of tumor-igenesis and does not act identically to classic oncogenes, some investiga-tors have conclusively shown its oncogenic potential in the promotion ofmalignant phenotypes [19]. Previously, we showed thatHOXA10 induceddifferentiation of EOC [20]. However, the mechanism of HOXA10 activa-tion of EOC pathogenesis remains poorly understood.

In this study, by using PicTar, TargetScan, and miRBase database,we find out that one miRNA, miR-135a, is a regulator of HOXA10expression, and its aberrant regulation has been demonstrated in a widevariety of tumors including breast cancer [22], hepatocellular carcinoma[23], colorectal cancer [24,25], gastric cancer [26], renal cell carcinoma[27], classic Hodgkin lymphoma [28], andmalignant glioma [29]. Howev-er, the role of miR-135a in EOC development, especially regarding its linkwith HOXA10, has not been explored yet. The analysis of both tumor cellsand their microenvironment can be an effective approach to understand-ing the behavior of EOC. Therefore, our objective in this study was todetermine howmiR-135a is involved in regulation of HOXA10 expressionand cell function phenotype in epithelial ovarian cancer.

2. Materials and methods

2.1. Patients and tissue samples

Tissue samples were collected from 73 subjects who underwentsurgery from July 2008 to December 2010 for the treatment of epithelialovarian cancers at the Department of Gynecology of the First AffiliatedHospital of Nanjing Medical University. All EOC patients had beenhistopathologically diagnosed with primary ovarian cancer. The Inter-national Federation of Gynecology andObstetrics (FIGO) staging systemwas used to stage cases. Patients' characteristics were summarized inTable 1. Samples from 55 cases involving other ovarian tumors orovarian cystadenomas were collected as controls. All enrolled patientsgave written informed consent and the study was approved by ourinstitutional ethics committee. All EOC patients' follow-up data wereacquired, and overall survival and progression-free survival durationswere calculated from the date of diagnosis to the date of death andthe date of disease progression or last follow-up, respectively.

2.2. Cell lines and cell culture conditions

OVCAR-3was purchased from the American Type Culture Collection.SKOV-3 and HEY cells were obtained from Fudan University, Shanghai,China. These three human EOC-derived cell lines and HEK 293T werecultured in DMEM (Gibco-BRL, USA) supplemented with 10% FBS

Table 1Clinical characteristics of the 73 EOC patients.

Patients characteristics Patients

No. of patients 73Age, year (median, range) 54, (18–83)SurgeryPrimary debulking 60Interval debulking 13

FIGO stage (I + II/III + IV) 27/46Histological type (serous/endometrial/mucinous/clear cell/others) 35/3/23/2/10Histological grade (G1 + G2/G3) 17/56Intraoperative residual tumor, cm (none/≤1/N1) 17/53/3Lymph node metastasis (yes/no) 57/16Chemotherapy (platinum-based/non-platinum-based/none) 67/2/4Elevated CA125 levels, U/ml (b5 × 105/≥5 × 105) 14/59Relapse (yes/no) 19/54

(Gibco-BRL, USA), 100 U/ml penicillin, and 100 μg of streptomycinat 37 °C in 5% CO2.

2.3. Transfection

MiR-135amimics, carboxyfluorescein (FAM)-labeledmiRNAmimics(miR-135a mimic negative control), miR-135a inhibitor, FAM-labeledmiRNA inhibitor (miR-135a inhibitor negative control), HOXA10 siRNA,and HOXA10 siRNA negative control were purchased fromGenePharma,Shanghai. The sequences are as follows: MiR-135a mimics: 5′-UAUGGC UUU UUA UUC CUA UGU GAA CAU AGG AAU AAA AAG CCAUAU U-3′, miR-135a mimic negative control: forward, 5′-UUC UCCGAA CGU GUC ACG UTT-3′; reverse, 5′-ACG UAC ACG UUC GGAGAA TT-3′, miR-135a inhibitor: 5′-UCA CAU AGG AAU AAA AAGCCA UA-3′, FAM-labeled miRNA inhibitor negative control: 5′-CAGUAC UUU UGU GUA GUA CAA-3′, HOXA10 siRNA: 5′-CAC GGA CAGACA AGU CAA ATT UUU CAC UUG UCU GUC CGU GTT-3′, and HOXA10siRNA negative control: forward, 5′-UUC UCC GAA CGU GUC ACGUTT-3′; reverse, 5′-ACG UAC ACG UUC GGA GAA TT-3′. Thesethree EOC cells were allowed to grow to 70% to 90% confluence in6-well plates before transfection. Cells were transfected with theseoligonucleotides by using Lipofectamine 2000 as recommended by themanufacturer's instructions.

2.4. RNA preparation and quantitative real-time PCR

Total RNA was isolated from epithelial ovarian cancer tissues orEOC cell lines using TRIzol reagent (Invitrogen, USA) according to themanufacturer's instructions. The extracted RNA was dissolved in DEPC-treated ddH2O and subjected to DNAse I treatment (Fisher Scientific,USA) to avoid genomic DNA contamination. Reverse transcription ofmRNA with 1 μg of RNA was performed to make cDNA using the FirstStrand cDNA synthesis kit (Fermentas, USA) and oligo(dT) primers oran miR-135a-specific reverse transcription primer (5′-CCC TGC TCGCAG TAT TTG AGT CAC ATA GGA-3′).

Quantitative real-time polymerase chain reaction (PCR) was usedto precisely quantify miR-135a and HOXA10 expression in humanepithelial ovarian tumor samples and EOC cell lines. QuantitativeRT-PCR was performed with SYBR Green reagents (Takara, Japan)in a 7300 real-time PCR system from Applied Biosystems. The 2−ΔΔCT

method was used to measure the HOXA10 and miR-135a gene expres-sion comparedwith the endogenous controls (β-actin or U6 noncodingsmall nuclear RNA). Amplification of HOXA10 andmiR-135a genes wasperformed with an initial step at 94 °C for 5 min followed by 40 cyclesof denaturation at 94 °C for 30 s, annealing at 63 °C for 30 s, and exten-sion at 72 °C for 10 s. All primers for HOXA10, miR-135a, GAPDH andthe U6 genes were designed by Primer Premier 5.0 and synthesizedby GenePharma: HOXA10, forward: 5′-GGG TAA GCG GAA TAA ACT-3′and reverse: 5′-GCA CAG CAG CAA TAC AAT A-3′; miR-135a, forward:5′-AAC CCT GCT CGC AGT ATT TGA G-3′ and reverse: 5′-GCG GCA GTATGG CTT TTT ATT CC-3′; GAPDH, forward: 5′-GAC CTG ACC TGC CGTCTA-3′ and reverse: 5′-AGG AGT GGG TGT CGC TGT-3′; U6, forward:5′-CTC GCT TCG GCA GCA CA-3′ and reverse: 5′-AAC GCT TCA CGAATT TGC GT-3′.

2.5. Protein extraction and Western blot analysis

Total proteins were collected from EOC cell lines using RIPA buffercontainingprotease inhibitors according to themanufacturer's protocol.Cell lysates were washed with cold PBS and incubated on ice in lysisbuffer for 30 min. The lysates were centrifuged at 25,000 g for 30 minat 4 °C and the protein concentrations in supernatants were measuredusing a Bradford protein assay kit (Galen Biopharm InternationalCo., China). Primary antibodies used in this study were specific toHOXA10, Bcl-2 (Santa Cruz), p53 (Proteintech Group) and GAPDH(KangChen Bio-tech).

1422 W. Tang et al. / Cellular Signalling 26 (2014) 1420–1426

2.6. Luciferase reporter assays

Awild-type 3′-UTR and amutant 3′-UTR of HOXA10were amplifiedfrom the cDNA of SKOV-3 cells using PCR. The primers for the wild-type3′-UTR were 5′-GGC GAG CTC TTC AGG CTC TGC CCA GGA ACT CG-3′(forward) and 5′-CCC AAG CTT CGG CTC CTT TGC ACC ATT GAC CT-3′(reverse), and the primers for the mutant 3′-UTR were 5′-GGC GAGCTC TTC AGG CTC TGC CCA GGA ACT CG-3′ (forward) and 5′-CCC AAGCTT GCA TGT ATA GGC TTT TTC CCC CAG A-3′ (reverse). Four and twohundred-base pair fragments of either the wild-type 3′-UTR or themutant 3′-UTR of HOXA10 were incorporated into a luciferase miRNAexpression reporter vector (pMIR-REPORT). The inserts were confirmedby using DNA sequencing.

HEK 293T cells were plated in 12-well plates in triplicate 24 h beforetransfection. PLemiR-135a, β-galactosidase and pMIR-REPORT withwild-type or mutant HOXA10 3′-UTR were transiently co-transducedinto cells by using Lipofectamine 2000 as recommended by themanufacturer's instructions. The luciferase activity was measuredafter 24 h of incubation according to the manufacturer's protocolusing a luciferase reporter assay system (Promega). The firefly lucifer-ase activity was normalized to the β-galactosidase activity values foreach sample. The experiment was performed at least three times andin triplicate.

2.7. Cell proliferation assay

SKOV-3 cells were transiently transfected with miR-135a mimics(100 pmol) and HOXA10 siRNA (100 nmol) and their respective con-trols by using Lipofectamine 2000 for 48 h in 6-well plates. To deter-mine the growth of SKOV-3 cells, cells were seeded in 96-well platesat 10,000 cells/well after 48-hour incubation. 20 μl of MTT solution(5 mg/ml) was added into each well and cells were incubated at37 °C. After 4-hour incubation, the supernatants were removed and200 μl of DMSO was added to each well. The plate was gently shakenon a shaker for 10 min and the absorbance was measured at 490 nmusing an automatic multi-well spectrophotometer (Bio-Rad, USA). Theexperiment was performed at least three times and each treatmentincluded five wells.

2.8. Measurement of caspase-3 activity

SKOV-3 cells were plated in 6-well plates and transfected withmiR-135a mimics and HOXA10 siRNA and their respective controlsas in Cell proliferation assay. After 48-hour incubation, cells wereharvested and lysed with lysis buffer on ice for 15 min. The celllysates were centrifuged at 20,000 g for 10 min at 4 °C. The proteinconcentration was measured with Bradford protein assay kit. Activa-tion of caspase-3 activities in SKOV-3 cells was determined by usingCaspase-3 Activity Assay kit (Beyotime Biotech, China) according tothe manufacturer's instructions. The experiment was performed atleast three times and in triplicate.

2.9. Annexin V-FITC assay for cell apoptosis

SKOV-3 cells were seeded onto 6-well plates and transientlytransfected with miR-135a mimics and HOXA10 siRNA and theirrespective controls as above. Cells were harvested, washed withcold PBS and resuspended in 500 μl of Binding Buffer at a density of5 × 105 cells/ml. According to the manufacturer's instructions ofAnnexin V-FITC Assay Kit (BioVision), cells were then incubatedwith 5 μl of Annexin V-FITC and 5 μl of propidium iodide (PI) atroom temperature for 5 min in the dark. Cells were analyzed byusing flow cytometry (excitation at 488 nm and emission at 530 nm).

2.10. Cell adhesion assay

SKOV-3 cells were seeded onto 6-well plates and transientlytransfected with miR-135a mimics and HOXA10 siRNA and theirrespective controls as above. Ninety-six-well plates were precoatedwith 100 μl of 0.2 μg/μl Matrigel (BD Biosciences) dissolved anddiluted in serum-free DMEM at 37 °C overnight. After aspirating,the wells were blocked with 50 μl of 2% bovine serum albumin(BioSharp) for 1 h at 37 °C. 1 × 105 cells in 100 μl of serum-freeDMEM containing 0.1% bovine serum albumin were seeded into theprecoated wells. After incubation at 37 °C for 1 h, the cells werewashed with PBS and the MTT assay was done as in Cell proliferationassay. The experiment was repeated three times.

2.11. Statistical analysis

MiR-135a expression was dichotomized using the maxstat softwareprogramof a language for data analysis and graphics (R) [30]. The statis-tical significance was determined using two-tailed Student's t-testsbetween themeans of control and experimental groups. The correlationbetween overall survival and progression-free survival and miR-135awas analyzed using the Kaplan–Meier method and log-rank test.All statistical calculations were performed and all graphs weregenerated by using the Graphpad Prism 5.0 software. Differenceswere considered significant when p b 0.05.

3. Results

3.1. Downregulation of miR-135a expression occurs in EOC tissues andindicates poor progression-free and overall survival

We identified the predicted target miRNA of HOXA10 usingcomputer-aided algorithms from PicTar (http://pictar.mdc-berlin.de/),TargetScan (http://www.targetscan.org/), and miRBase (http://microrna.sanger.ac.uk/). The algorithms indicated that this targetmiRNA was found to be miR-135a. By using quantitative real-timePCR, we demonstrated that the expression of miR-135awas significant-ly downregulated in EOC tissues compared with controls (p= 0.0024)(Fig. 1A). Meanwhile, low level of miR-135a coincided withhigh expression of HOXA10 in EOC tissues compared with controls(p = 0.0022) (Fig. 1B).

Themedian age of the EOC patients was 54 years. In the 73 subjects,60 cases underwent primary debulking surgery, 46 cases had highInternational Federation of Gynecology and Obstetrics stage disease(III/IV), 57 cases had lymph node metastasis, and 67 cases receivedplatinum-based chemotherapy (shown in Table 1). By using the maxi-mal standardized log-rank statistic at the cut-off point of 1.49 meangene expression normalized to U6 noncoding small nuclear RNA, wedichotomized the 73 patients into two categories according tomiR-135a expression level: 58 patients with low miR-135a expressionhad a worse progression-free survival than did 15 patients with highmiR-135a expression (Fig. 1C, p = 0.0015); meanwhile, the analysisof these two groups of patients showed that the patients with highmiR-135a had better overall survival than those with low miR-135a(Fig. 1D, p = 0.0050).

3.2. MiR-135a regulated HOXA10 expression in EOC cell lines

To determine whether miR-135a is a negative regulator of HOXA10expression, we transfected miR-135a mimics or miR-135a inhibitorinto EOC cell lines (SKOV-3, HEY, and OVCAR-3). Transfection withmiR-135a mimics successfully led to a greater increase in miRNA (datanot shown) and miR-135a inhibitor led to a significant reduction asshown in Fig. 2 (*p b 0.001). Transfection with miR-135a mimics for48 h significantly reduced HOXA10 expression in SKOV-3, HEY andOVCAR-3, and 72-hour incubation demonstrated a much lower value

A B

C D

Control EOC0

1

2

3 p=0.0024

n=55 73

Rel

ativ

e ex

pres

sion

of m

iR-1

35a

com

pare

d to

U6

0 10 20 30 400

20

40

60

80

100Low miR-135a (n=58)

High miR-135a (n=15)

p=0.0015

Months after diagnosis

Pro

babi

lity

of P

rogr

essi

on-F

ree

Sur

viva

l

0 20 40 60 800

20

40

60

80

100Low miR-135a (n=58)

High miR135a (n=15)

p=0.0050

Pro

babi

lity

of O

vera

ll S

urvi

val

0.0

0.5

1.0

1.5

2.0 p=0.0022

Rel

ativ

e ex

pres

sion

of H

OX

A10

com

pare

d to

GA

PD

H

Control EOCn=55 73

Months after diagnosis

Fig. 1.MiR-135a expression in EOC tissues and its correlation with the progression-free and overall survival of EOC patients. (A) Relative expression of miR-135a in all samples of EOC andcontrols normalized to the U6 noncoding small nuclear RNA in quantitative real-time PCR analysis: miR-135a expression was significantly down-regulated in EOC tissues as comparedwith controls (p = 0.0024). (B) Higher level of HOXA10 mRNA in EOC compared with controls (p = 0.0022). (C) Correlation of miR-135a expression with progression-free survival ofEOC patients (p = 0.0015). (D) Correlation of miR-135a expression with overall survival of EOC patients (p = 0.0050).

1423W. Tang et al. / Cellular Signalling 26 (2014) 1420–1426

in HEY and OVCAR (by about 1-fold) as shown in Fig. 3 (left, **p b 0.01).On the contrary, transfection with miR-135a inhibitor for 48 h resultedin increased levels of HOXA10 mRNA and a much higher value for 72 h(***p b 0.001).

We further investigated HOXA10 protein expression by usingWestern blot analysis after these three EOC cell lines were transfectedwith miR-135a mimics or miR-135a inhibitor for 24, 48 and 72 h.Cells transfected with miR-135a mimics for 48 h had lower levels ofHOXA10 expression and further for 72 h. On the contrary, cellstransfected with miR-135a inhibitor had higher expression of HOXA10for both 48 and 72 h in comparison with those transfected with controlmiRNA (Fig. 3 right).

SKOV-3HEY

OVCAR-30.0

0.5

1.0

1.5

2.0

Negative controlmiR-135a Inhibitor

**

* *

Rel

ativ

e ex

pres

sion

of m

iR-1

35a

com

pare

d to

U6

Fig. 2. MiR-135a inhibitor suppresses miR-135a level in three EOC cell lines. Cells wereseeded onto 6-well plates and grown to 70%–90% confluence before transfection, andthen cells were transfected with miR-135a inhibitor for 48 h. Cells were harvested andmiR-135a expression was assessed by real-time PCR. The experiment was performed atleast three times and in triplicate. *p b 0.05 and **p b 0.01.

3.3. MiR-135a directly targets the 3′-UTR of HOXA10

To determine whether miR-135a regulated HOXA10 throughthe predicted binding sites in its 3′-UTR, we designed two luciferaseconstructs by incorporating either wild-type or mutant 3′-UTR ofHOXA10, which constitutively expresses luciferase unless repressed bythe incorporated 3′-UTR. Cotransfection of HEK 293T cells with thispMIR-REPORT construct containing mutant HOXA10 3′-UTR andPLemiR-135a did not show much difference compared with control.Cotransfection with this luciferase construct containing wild-typeHOXA10 3′-UTR and PLemiR-135a resulted in a higher luciferase activ-ity than control, leading to a nearly 60% decline in the luciferase activitycompared with control (**p b 0.01) (Fig. 4).

3.4. MiR-135a mediates EOC cell proliferation, apoptosis, and adhesion

Of the three EOC cell lines used in our experiments, SKOV-3 cells hadthe highest level of HOXA10 expression as shown in Fig. 5A, which wasused in cell proliferation, apoptosis, and adhesion assays. To determinewhether downregulation of HOXA10 expression induced by miR-135aaffects EOC cell growth, MTT assay was performed. Transfection witheither miR-135a mimics or HOXA10 siRNA consistently inhibitedSKOV-3 cell proliferation compared with each control. In addition,miR-135a mimics were more effective in suppressing tumor cellproliferation than HOXA10 siRNA (**p b 0.01) (Fig. 5B).

To clarify the role of miR-135a in EOC tumorigenesis, we also inves-tigated the expression or activity of three apoptosis-associated genesincluding p53, Bcl-2, and caspase-3. Downregulation of HOXA10 ex-pression by either miR-135a mimics or HOXA10 siRNA decreased theexpression of Bcl-2 but increased the expression of p53 and caspase-3activity in EOC cells (Fig. 5C and D). Similarly, miR-135a mimics weremore effective than HOXA10 siRNA on regulating the expression ofcaspase-3, p53 and Bcl-2 activity. These results demonstrated that

A

B

C

24h

48h

72h

0

1

2

3

4miR-135a mimicsmiR-135a inhibitorInhibitor negative controlmimics negative controlcontrol

****

******

SKOV-3

Rel

ativ

e ex

pres

sion

of H

OX

A10

com

pare

d to

GA

PD

H

0

1

2

3

4miR-135a mimicsmiR-135a inhibitorInhibitor negative controlmimics negative controlcontrol***

***

******

HEY

Rel

ativ

e ex

pres

sion

of H

OX

A10

com

pare

d to

GA

PD

H

0

1

2

3miR-135a mimicsmiR-135a inhibitorInhibitor negative controlmimics negative controlcontrol

******

****

OVCAR-3

Rel

ativ

e ex

pres

sion

of H

OX

A10

com

pare

d to

GA

PD

H

24h

48h

72h

24h

48h

72h

Fig. 3.HOXA10 expression is modulated bymiR-135a. (A), (B) and (C) Relative HOXA10mRNA (left) and protein expression levels (right) in SKOV-3, HEY and OVCAR-3 cells. Cells weregrown in the 6-well plates and transfectedwith miR-135a mimics and miR-135a inhibitor and their respective negative controls. After 24, 48 or 72-hour incubation, cells were harvestedand assayed using real-time PCR and Western blot analysis. **p b 0.01 and ***p b 0.001.

1424 W. Tang et al. / Cellular Signalling 26 (2014) 1420–1426

miR-135-mediated downregulation of HOXA10 expression resulted in achange in the expression of its downstream apoptosis-associated genesthat may control epithelial ovarian tumor growth.

0.0

0.5

1.0

1.5ControlMutantWide-type

**

Rel

ativ

e lu

cife

rase

aci

tivity

Fig. 4.MiR-135a directly regulates HOXA10 by targeting HOXA10 3′-UTR. HEK 293T cellswere transfected with the pMIR-REPORT construct, mutant 3′-UTR or wild-type 3′-UTRplus β-galactosidase and PLemiR-135a. After 24-hour incubation, cells were harvestedand assayed by using a luciferase reporter assay system. Luciferase activity wasnormalized to β-galactosidase activity. **p b 0.01.

To further validate that miR-135a mediates EOC cell apoptosis, wedetected apoptosis of SKOV-3 cells by using Annexin-V FITC stainingassay following transfection with the miR-135a mimics or HOXA10siRNA (Fig. 5E). Apoptosis was shown to be higher in cells transfectedwith HOXA10 siRNA andmuchmore in cells with themiR-135amimicscompared with their respective controls.

In addition, we also investigated whether reduction of HOXA10expression by miR-135a mimics or HOXA10 siRNA affects EOC celladhesion. As shown in Fig. 5F, transfectionwith eithermiR-135amimicsor HOXA10 siRNA resulted in lower ability of SKOV-3 cells to bind toMatrigel in comparison with controls.

4. Discussion

The current knowledge aboutmicroRNAs in cancer is still controver-sial. MiR-135a has been proven to contribute to colorectal pathogenesisthrough its regulation of APC gene [24], promote the growth and inva-sion of colorectal cancer cell via metastatic suppressor 1 (MTSS1) [25],induce breast cancer cell migration and invasion by targeting HOXA10and contribute to the development of portal vein tumor thrombus[22,23]. However, miR-135a was also reported to inhibit cancer cell

0.0

0.2

0.4

0.6

0.8

1.0mimics control

miR135a mimics

siRNA control

HOXA10 siRNA

*** **

Cel

l pro

lifer

aito

n ac

tivity

0

20

40

60

80

100 mimics control

miR-135a mimics

control

HOXA10 siRNA***

***

Adh

esio

n(%

)

A

B

E

43KD

53KD

26KD

36KD

HOXA10

P53

Bcl-2

GAPDH

C

F

10.81%

miR-135a mimics

5.04%

mimics control

5.31%

siRNA control

8.25%

HOXA10 siRNA

43kD

36kD

HOXA10

GAPDH

D

0

100

200

300 mimics controlmiR-135a mimics

siRNA control

HOXA10 siRNA

***

***

Cas

pase

-3 a

ctiv

ity(%

cont

rol)

Fig. 5. Inhibition of HOXA10 expression with miR-135a and HOXA10 siRNA suppresses EOC cell proliferation, promotes apoptosis, and attenuates cell adhesion potential. (A) Westernblot analysis for HOXA10 protein levels in EOC cell lines. (B) MTT assay for SKOV-3 cell after transfected with miR-135a mimics or HOXA10 siRNA. **p b 0.01; ***p b 0.001.(C) Caspase-3 activity was measured in SKOV-3 cells after transfected with miR-135a mimics or HOXA10 siRNA. Each data point represents mean ± SD from three independent wells.***p b 0.001. (D)Western blot analysis for HOXA10, p53, and Bcl-2 protein levels in SKOV-3 cells after transfected with miR-135amimics or HOXA10 siRNA. (D) Flow cytometry analysisof apoptosis in the SKOV-3 cells after transfectionwithmiR-135amimics orHOXA10 siRNA. The experimentwas repeated at least three times. (E) Cell adhesion assay for SKOV-3 cell. Cellswere pretreated with miR-135a mimics or HOXA10 siRNA for 48 h and assayed by MTT. The experiment was repeated at least three times. ***p b 0.001.

1425W. Tang et al. / Cellular Signalling 26 (2014) 1420–1426

proliferation through its target gene Janus Kinase 2 (JAK2) in gastriccancer and by targeting c-MYC in renal cell carcinoma [26,27]. In ourstudy, miR-135a is playing a critical role in EOCs by regulatingHOXA10 expression through directly targeting its 3′-UTR. Takentogether, these findings suggest that miR-135a functions widelydifferently in some specific tumor lineages.

A significant observation from our work is that low level miR-135aindicated both poor progression-free and overall survival rates of thesubjects with EOC. This coincided with high expression of HOXA10.Previous research has suggested that HOXA10 was overexpressed inhuman ovarian clear cell adenocarcinoma and correlated with poorsurvival [21]. However, the mechanism of upregulation of HOXA10expression during tumor progression remains unclear. In this study,we demonstrate that miR-135a level is downregulated in EOCs andidentify HOXA10 as a functional target of miR-135a. By analyzingthree EOC cell lines in vitro, our current study provided the evidenceto demonstrate that miR-135a is playing a role in cell growth, apoptosisand cell adhesion by directly regulating HOXA10 through targeting its3′-UTR. Moreover, several downstream targets of HOXA10, such as

p53 and p21 [31,32], were observed to play multiple roles in severalpathways, acting as transcription factors and regulators of their ownsubsets of genes, including Bcl-2 and caspase-3 [33,34]. These findingssupport our hypothesis that aberrant expression of HOXA10 is pro-moted during carcinogenesis in a variety of cell signaling pathways.

Our results also demonstrated that increased expression of miR-135a leads to decreased HOXA10 expression and increased p53 expres-sion in EOC cells. These results provide the first specific link betweenHOXA10 and p53, the most frequently mutated tumor suppressorgene in high-grade ovarian serous carcinoma cells. We observed an in-verse correlation between HOXA10 and p53 expression in EOC cells.The increased p53 expression resulted in increased cell adhesion andapoptosis rates and decreased proliferation via regulation of Bcl-2 andcaspase-3 expression. In addition, miR-135a mimics were more effec-tive than HOXA10 siRNA in changing cancer cell biology and expressionof p53-associated proteins. These findings demonstrated that miR-135a-mediated regulation of HOXA10 expression is at least partiallymediated by p53, thus linking miR-135a with the p53 regulatory net-work. However, of note is that this is unlikely to be the only pathway

1426 W. Tang et al. / Cellular Signalling 26 (2014) 1420–1426

explaining the phenotype that miR-135a mediates EOC cell prolifer-ation, apoptosis, and adhesion since a single miRNA may targetmultiple genes. Further study is required to identify other targetsof miR-135a and thus advance our understanding of miR-135a-mediated tumorigenesis.

Taken together, our results identifiedmiR-135a as a potential tumorsuppressor in patients with EOC in that it downregulates expressionof the oncogene HOXA10. But the detailed mechanisms regardingthe target sequence in HOXA10 and additional signaling involved inthe process that miR-135a regulated the pathogenesis of EOC remainto be identified. MiR-135a may identify novel therapeutic targets andprognostic markers in patients with EOC.

Conflicts of interest statement

The authors declared no conflicts of interest.

Author contributions

WWT, YJ, XXM and LX performed the experiments. WJC and XRWdesigned the experiments. WWT and WJC analyzed the data andwrote the manuscript.

Acknowledgments

This work was supported by the National Natural ScienceFoundation of China (81272871 to W Cheng), Nature ScienceFoundation of the Science and Technology Department of JiangsuProvince (BK2010576 to W Cheng), Jiangsu Province Six SummitTalent Foundation (303070774IB09 to W Cheng), and Jiangsu ProvinceMedical Key Talent Grant (2011 to W Cheng). We thank Dr. BinghuaJiang for his valuable assistance in cell experiments.

References

[1] R. Siegel, J. Ma, Z. Zou, A. Jemal, CA Cancer J. Clin. 64 (1) (Jan. 2014) 9–29.[2] D.P. Bartel, Cell 116 (2) (Jan. 23 2004) 281–297.[3] M.N. Poy, L. Eliasson, J. Krutzfeldt, et al., Nature 432 (7014) (Nov. 11 2004) 226–230.[4] J. Brennecke, D.R. Hipfner, A. Stark, R.B. Russell, S.M. Cohen, Cell 113 (1) (Apr. 4

2003) 25–36.[5] P. Xu, M. Guo, B.A. Hay, Trends Genet. 20 (12) (Dec. 2004) 617–624.[6] C.Z. Chen, L. Li, H.F. Lodish, D.P. Bartel, Science 303 (5654) (Jan. 2 2004) 83–86.[7] A.M. Cheng, M.W. Byrom, J. Shelton, L.P. Ford, Nucleic Acids Res. 33 (4) (2005)

1290–1297.[8] L. He, J.M. Thomson, M.T. Hemann, et al., Nature 435 (7043) (Jun. 9 2005) 828–833.[9] J. Lu, G. Getz, E.A. Miska, et al., Nature 435 (7043) (Jun. 9 2005) 834–838.

[10] N. Pencheva, S.F. Tavazoie, Nat. Cell Biol. 15 (6) (Jun. 2013) 546–554.[11] A. Esquela-Kerscher, F.J. Slack, Nat. Rev. Cancer 6 (4) (Apr. 2006) 259–269.[12] O.A. Kent, J.T. Mendell, Oncogene 25 (46) (Oct. 9 2006) 6188–6196.[13] L. Zhang, J. Huang, N. Yang, et al., Proc. Natl. Acad. Sci. U. S. A. 103 (24) (Jun. 13 2006)

9136–9141.[14] M.V. Iorio, R. Visone, L.G. Di, et al., Cancer Res. 67 (18) (Sep. 15 2007) 8699–8707.[15] S. Vang, H.T. Wu, A. Fischer, et al., PLoS One 8 (3) (2013) e58226.[16] H. Zheng, L. Zhang, Y. Zhao, et al., PLoS One 8 (11) (2013) e77853.[17] A. Chariot, J. Gielen, M.P. Merville, V. Bours, Biochem. Pharmacol. 58 (12) (Dec. 15

1999) 1851–1857.[18] M. Kelly, G. Daftary, H.S. Taylor, Am. J. Obstet. Gynecol. 194 (4) (Apr. 2006)

1100–1107.[19] C. bate-Shen, Nat. Rev. Cancer 2 (10) (Oct. 2002) 777–785.[20] W. Cheng, J. Liu, H. Yoshida, D. Rosen, H. Naora, Nat. Med. 11 (5) (May 2005) 531–537.[21] B. Li, H. Jin, Y. Yu, et al., Int. J. Gynecol. Cancer 19 (8) (Nov. 2009) 1347–1352.[22] Y. Chen, J. Zhang, H. Wang, et al., BMC Cancer 12 (2012) 111.[23] S. Liu, W. Guo, J. Shi, et al., J. Hepatol. 56 (2) (Feb. 2012) 389–396.[24] R. Nagel, S.C. le, B. Diosdado, et al., Cancer Res. 68 (14) (Jul. 15 2008) 5795–5802.[25] W. Zhou, X. Li, F. Liu, et al., Acta Biochim. Biophys. Sin. (Shanghai) 44 (10) (Oct.

2012) 838–846.[26] H.Wu,M. Huang, P. Cao, T.Wang, Y. Shu, P. Liu, Cancer Biol. Ther. 13 (5) (Mar. 2012)

281–288.[27] Y. Yamada, H. Hidaka, N. Seki, et al., Cancer Sci. 104 (3) (Mar. 2013) 304–312.[28] A. Navarro, T. Diaz, A. Martinez, et al., Blood 114 (14) (Oct. 1 2009) 2945–2951.[29] S. Wu, Y. Lin, D. Xu, et al., Oncogene 31 (34) (Aug. 23 2012) 3866–3874.[30] T. Hothorn, B. Lausen, Comput. Stat. Data Anal. 43 (2003) 121–137.[31] V.C. Bromleigh, L.P. Freedman, Genes Dev. 14 (20) (Oct. 15 2000) 2581–2586.[32] M.C. Chu, F.B. Selam, H.S. Taylor, Cancer Biol. Ther. 3 (6) (Jun. 2004) 568–572.[33] S. Haldar, M. Negrini, M. Monne, S. Sabbioni, C.M. Croce, Cancer Res. 54 (8) (Apr. 15

1994) 2095–2097.[34] T. Miyashita, S. Krajewski, M. Krajewska, et al., Oncogene 9 (6) (Jun. 1994)

1799–1805