Báo cáo khoa học: MicroRNA-23b mediates urokinase and c-met downmodulation and a decreased migration of human hepatocellular carcinoma cells

MicroRNA-23b mediates urokinase and c-met downmodulation and a decreased migration of human hepatocellular carcinoma cells Alessandro Salvi1, Cristiano Sabelli1, Silvia Moncini2, Marco Venturin2, Bruna Arici1, Paola Riva2, Nazario Portolani3, Stefano M. Giulini3, Giuseppina De Petro1 and Sergio Barlati1

1 Division of Biology and Genetics, Department of Biomedical Sciences and Biotechnology, IDET Centre of Excellence, University of Brescia, Italy 2 Department of Biology and Genetics, Medical Faculty, University of Milan, Italy 3 Department of Medical and Surgical Sciences, University of Brescia, Italy

Keywords c-met; hepatocellular carcinoma cells; microRNA-23b; urokinase

Correspondence G. De Petro, Department of Biomedical Sciences and Biotechnology, Division of Biology and Genetics, University of Brescia, Viale Europa n. 11, 25123 Brescia, Italy Fax: +39 30 3701157 Tel: +39 30 3717 264 241 E-mail: depetro@med.unibs.it

(Received 19 December 2008, revised 18 March 2009, accepted 19 March 2009)

doi:10.1111/j.1742-4658.2009.07014.x

Urokinase-type plasminogen activator (uPA) and c-met play a major role in cancer invasion and metastasis. Evidence has suggested that uPA and c-met overexpression may be coordinated in human hepatocellular carci- noma (HCC). In the present study, to understand whether the expression of these genes might be coregulated by specific microRNAs (miRs) in human cells, we predicted that Homo sapiens microRNA-23b could recog- nize two sites in the 3¢-UTR of uPA and four sites in the c-met 3¢-UTR by the algorithm pictar. The miR-23b expression analysis in human tumor and normal cells revealed an inverse trend with uPA and c-met expression, indicating that uPA and c-met negative regulation might depend on miR- 23b expression. Transfection of miR-23b molecules in HCC cells (SKHep1C3) led to inhibition of protein expression of the target genes and caused a decrease in cell migration and proliferation capabilities. Further- more, anti-miR-23b transfection in human normal AB2 dermal fibroblasts upregulated the expression of endogenous uPA and c-met. Cotransfection experiments in HCC cells of the miR-23b with pGL4.71 Renilla luciferase reporter gene constructs, containing the putative uPA and c-met 3¢-UTR target sites, and with the pGL3 firefly luciferase-expressing vector showed a decrease in the relative luciferase activity. This would indicate that miR- 23b can recognize target sites in the 3¢-UTR of uPA and of c-met mRNAs and translationally repress the expression of uPA and c-met in HCC cells. The evidence obtained shows that overexpression of miR-23b leads to uPA and c-met downregulation and to decreased migration and proliferation abilities of HCC cells.

Abbreviations GAPDH, glyceraldehyde 3-phosphate dehydrogenase; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; HGF, hepatocyte growth factor; hsa-miR-23b, Homo sapiens microRNA-23b; miR, microRNA; NM, nitrocellulose membrane; PT, peritumoral; uPA, urokinase-type plasminogen activator.

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2966

MicroRNAs (miRs) are small (21–25 nucleotide) non- protein-coding RNAs implicated in negative gene expression regulation [1,2]. More than 500 human miRs have been identified (http://microrna.sanger. ac.uk, version 13.0, updated March 2009), and over 1000 miRs are predicted to exist in the vertebrate and human genome [3,4]. The biogenesis of miRs involves a complex protein system. They are generally tran- scribed by RNA polymerase II or III into pri-miR transcripts that are processed by the RNase III enzyme

A. Salvi et al.

miR-23b downmodulates uPA and c-met

In addition, in the case of cancer

for expression of these genes [20,21]. uPA can be regu- lated at the transcriptional and post-transcriptional levels (i.e. promoter methylation ⁄ demethylation, and distinct adenylate ⁄ uridylate AU-rich element-binding proteins, which can increase ⁄ decrease the mRNA sta- bility) [22,23]; c-met expression can be negatively mod- ulated at the transcriptional level by the transcription repressor Daxx [24]. For uPA and c-met in human HCC, one of the major cancers worldwide, with a poor prognosis [25], we have shown that the mRNA expression levels are unfavorable prognostic factors for HCC patients, and that uPA mRNA overexpression is strictly associated with c-met mRNA levels [26,27]. Thus, in this study, we have verified the hypothesis of miR-mediated gene regulation of uPA and c-met. We identified by bioinformatics, and subsequently vali- dated, Homo sapiens miR-23b (hsa-miR-23b) as a putative miR mediating both uPA and c-met down- regulation. Overexpression of miR-23b led to uPA and c-met silencing, and to decreased migration ability of SKHep1C3 HCC cells. inhibition of endogenous miR-23b by anti-miR-23b molecules led to upregulated uPA and c-met expression in human normal AB2 fibroblasts.

Results

Bioinformatic prediction of miRs targeting uPA and c-met 3¢-UTRs [15],

Drosha to release the precursor product, pre-miR, which consists of about 70 nucleotides. Exportin 5 transports these molecules into the cytoplasm, where they are processed by the RNase III enzyme Dicer into a transient 22 nucleotide linear duplex. Only one strand of the duplex, called the mature miR, is loaded into the miR-associated multiprotein RNA-induced silencing complex. Most miRs bind to the 3¢-UTR of the target mRNAs with imperfect complementarity, and they direct translational repression and ⁄ or mRNA target degradation [5]. It has been predicted that miRs target up to 30% of protein-coding genes in humans [6]. Accumulating evidence indicates that miRs are involved in development, differentiation, apoptosis, proliferation, and several diseases, including cancer [7]. During recent years, there have been several micro- array studies that have allowed the identification of new miRs and the determination of global expression profiles under certain biological conditions [8,9]. Array platforms have also been useful for the definition of differential miR signatures between neoplastic tissues [10–12]. and healthy ones, Although the number of discovered miRs has rapidly increased, data on possible mRNA targets for most miRs remain elusive. Experimental validation of miRs predicted by bioinformatic analysis could be funda- mental for the definition of a biological function of a given miR [13,14]. Therefore, the main purpose of this study was to validate the miRs predicted by pictar targeting uro- (http://www.pictar.bio.nyu.edu) kinase-type plasminogen activator (uPA) and c-met genes.

is (HGF)], responsible for

Bioinformatic prediction, determined using the algo- rithm pictar (also verified by targetscan), showed putative target sites for six and 13 miRs in the uPA and c-met 3¢-UTR sequences, respectively (Table 1). Among the predicted miRs, we focused on hsa-miR- 23b, as it recognizes binding sites in both uPA and c-met transcripts (two and four binding sites for uPA and c-met respectively). Furthermore, hsa-miR-23a may possibly target the uPA and c-met 3¢-UTRs, but the free energy of the binding sites (two and four for uPA and c-met, respectively) is generally higher than that for hsa-miR-23b. hsa-miR-23b is an intronic miR located in intron 12 of the host gene C9orf3 on chro- mosome 9, and the predicted binding sites in the uPA and c-met 3¢-UTRs are conserved, to varying degrees, across species (Fig. 1A,B).

Expression of mature miR-23b and uPA and c-met target genes in human tumor and normal cells

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2967

To evaluate the possible effects of miR-23b on the two targets, uPA and c-met, we first verified their uPA is a serine protease that, when bound to its receptor, plays an important role in proteolytic cas- cades, and governs the extracellular matrix turnover and the migration and ⁄ or proliferation of several types of tumor cells [16]. The tyrosine protein kinase recep- tor c-met, when bound to its ligand [hepatocyte growth factor several cellular responses, such as invasive growth in development and cancer, tissue regeneration, angiogenesis, proliferation, and migration; c-met may also be activated in an HGF-independent manner [17]. uPA and c-met are involved in human cancer, through the development and progression of several types of malignant tumor. An essential role of uPA and c-met in the migration and proliferation of human hepatocellular carcinoma (HCC) cells has been assessed by functional studies carried out using RNA interference technology [18,19]. The evidence suggests that the two systems, uPA–uPA receptor and HGF–c-met, might cooperate in the acquisition of malignant phenotypes of cancer cells, but little is known about the regulatory mechanisms

A. Salvi et al.

miR-23b downmodulates uPA and c-met

Table 1. PICTAR miR prediction.

No. binding sites

Free energies (kcalÆmol)1)

uPA miR

hsa-miR-193b hsa-miR-193a hsa-miR-23b hsa-miR-23a hsa-miR-199a hsa-miR-199b

2 2 2 2 1 1

)21.9, )21.1 )23.7, )21.1 )19.4, )19.6 )17.0, )15.7 )18.2 )19.1

c-met miR

hsa-miR-34a

6

hsa-miR-34c hsa-miR-1 hsa-miR-142-5p hsa-miR-34b hsa-miR-198 hsa-miR-206 hsa-miR-199a* hsa-miR-144 hsa-miR-23a hsa-miR-23b hsa-miR-302c* hsa-miR-101

3 4 4 2 2 2 3 3 4 4 4 1

)21.7, )20.3, )21.7, )19.3, )21.2, )22.8 )23.4, )19.3, )27.2 )14.4, )13.8, )17.5, )18.3 )10.9, )14.8, )13.4, )20.2 )23.9, )27.5 )22.0, )21.3 )18.2, )17.9 )14.1, )26.0, )16.0 )15.6, )14.2, )14.4 )17.9, )16.3, )17.9, )21.8 )20.2, )17.4, )15.2, )21.8 )15.6, )17.7, )10.8, )25.6 )15.9

*, the opposite strand to the mature miRNA.

the

Fig. 1. The uPA and c-met 3¢-UTRs, respectively, harbor two and four putative binding sites for miR-23b. (A) The location of sites 1 and 2 in the uPA 3¢-UTR, and complementarity between miR-23b and the putative uPA 3¢-UTR target sites. The conserved bases of the putative miR-23b target sequence are also shown. (B) The loca- tion of sites 1, 2, 3 and 4 in the c-met 3¢-UTR, and complementarity between miR-23b and the putative c-met 3¢-UTR target sites. Con- served bases of the putative miR-23b target sequence present in the c-met 3¢-UTR are also shown. has, Homo sapiens; ptr, P. trog- lodytes (chimpanzee); mmu, M. musculus (mouse); rno, R. norvegi- cus (rat); cfa, C. familiaris (dog).

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2968

expression in human normal and tumor cell lines. As shown in Fig. 2A,B, real-time RT-PCR and northern blot data revealed that miR-23b was highly expressed in AB2 human dermal fibroblasts, whereas it was detectable at lower and variable levels in five human lines. Regarding uPA expression, HCC-derived cell detected by real-time-PCR, western blotting, and zymography, undifferentiated HCC-derived SKHep1C3, SKHep1C3.69.2 and HA22T ⁄ VGH cells had a considerable level of uPA mRNA as well as high uPA protein and enzymatic activity (Fig. 2A,C). These three cell lines also displayed a higher amount of c-met protein (Fig. 2C). In contrast, lower levels of uPA and c-met proteins and mRNAs were detectable in AB2 normal cells and in the differentiated HCC cells, HepG2 and HuH6 cells (Fig. 2A,C). These data

A. Salvi et al.

miR-23b downmodulates uPA and c-met

Fig. 2. Expression determination of mature miR-23b, uPA and c-met in human normal and HCC-derived cells. (A) Real-time RT-PCR detec- tion of miR-23b, uPA and c-met mRNAs. The amounts of miR-23b, uPA and c-met mRNAs were evaluated as described in Experimental procedures. (B) Northern blot detection of miR-23b and U6 RNA (as control). (C) Western blot and zymographic analysis of conditioned media from normal and tumor cells for the detection of uPA protein and its enzymatic activity, and western blot for c-met detection in cell lines. Lane 1: AB2. Lane 2: SKHep1C3. Lane 3: SKHep1C3.69.2. Lane 4: HuH6. Lane 5: HepG2. Lane 6: extracts from the same cell Ha22T ⁄ VGH. RQ, relative quantification.

miR-23b decreased uPA and c-met protein expression in SKHep1C3 cells

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2969

levels as low as indicate an inverse trend between the expression levels of miR-23b and of uPA and c-met. AB2 control cells, which did not express uPA and c-met, displayed a high level of miR-23b, whereas the most aggressive HCC cells, which produced high levels of uPA and c-met, had a lower expression level of miR-23b. The differentiated HuH6 and HepG2 cells showed a differ- ent profile of uPA and c-met expression, as well as of miR-23b. The HuH6 cells, with undetectable levels of uPA and c-met, expressed a certain amount of miR-23b, less than that expressed by normal AB2 cells. The HepG2 cells with a c-met mRNA amount comparable to that displayed by SKHep1C3 cells expressed miR-23b at those of SKHep1C3 cells. To assess the effects of miR-23b on uPA and c-met expression, we transiently transfected miR-23b in SKHep1C3 cells to evaluate target gene expression. Western blot and zymographic analysis of the condi- tioned media of transfected cells revealed a significant reduction of uPA expression and its enzymatic activity. The data shown in Fig. 3 clearly show, respectively, 62% and 71% inhibition of uPA protein expression 48 transfection of 100 nm miR-23b and 72 h after (Fig. 3A). Maximum inhibition of uPA enzymatic activity (by 51%) was obtained at 72 h after transfec- tion (Fig. 3B).

A. Salvi et al.

miR-23b downmodulates uPA and c-met

Fig. 3. miR-23b inhibits uPA and c-met protein expression in SKHep1C3 cells. (A) Western blot analysis of uPA in the conditioned media of control cells (lanes 1 and 4) and miR-23b-transfected cells, at 48 h and 72 h after transfection. Lane 2 : 50 nM miR-23b, 48 h. Lane 3 : 100 nM miR-23b, 48 h. Lane 5 : 50 nM miR-23b, 72 h. Lane 6 : 100 nM miR-23b, 72 h. (B) Zymographic detection of the corresponding uPA enzymatic activity. (C) Western blot detection of GAPDH in control and miR-23b-transfected cells. (D) Western blot detection of c-met and GAPDH in cell extracts from control and miR-23b-transfected cells. The protein amount of the housekeeping gene GAPDH was compa- rable in all samples tested.

produce very low amounts of miR-23b, the high level detected in transfected cells may be due to the transfected molecules.

Anti-miR-23b transfection in normal AB2 cells leads to upregulated uPA and c-met protein expression

Forty-eight hours after transfection, c-met protein expression was inhibited by 48% at 100 nm miR-23b, as detected by a semiquantitative western blotting analysis. Seventy-two hours after transfection, the 170 kDa c-met precursor form was inhibited by 89% (Fig. 3D), indicat- ing that the translatability of c-met mRNA was strongly affected. The increased amount of the 145 kDa form (produced by proteolytic processing of the precursor) may be due to c-met protein turnover.

steady-state mRNA levels (Fig. 4A,B), providing of evidence

expression

72 h (Fig. 5A), and the activity

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2970

To investigate whether miR-23b targeted uPA and c-met mRNAs for degradation, a semiquantitative RT-PCR evaluation was carried out on RNA isolated from control and transfected cells. The data showed uPA comparable and c-met that miR-23b in SKHep1C3 cells can lead to decreased uPA and c-met protein expression without affecting the amounts of their mRNA. Furthermore, regarding miR expression, transfected cells displayed higher amounts of miR-23b (Fig. 4C, lanes 2 and 3) both at 48 and at 72 h after transfection, with a decline at 72 h (Fig. 4C, lanes 5 and 6). As the SKHep1C3 cells To investigate whether the silencing of miR-23b might lead to upregulation of uPA and c-met, AB2 fibro- blasts (at a high expression level of miR-23b) were transfected with 100 nm antisense RNA oligonucleo- tides complementary to miR-23b. Forty-eight and 72 h after transfection, the conditioned media, cell lysates and total RNA were examined by zymography, wes- tern blot and RT-PCR to analyze uPA and c-met expression. The levels uPA protein increased 2.25 ± 0.84-fold at 48 h and 6.68 ± 1.11- fold at corresponding increased 3.1 ± 0.33-fold and enzymatic 9.55 ± 2.12-fold, as compared with the control, at 48 and 72 h (Fig. 5B, lanes 3 and 6, and lanes 2 and 5,

A. Salvi et al.

miR-23b downmodulates uPA and c-met

cells increased 2.56 ± 0.28-fold

(Fig. 6A,

transfection of

the annealing of

Fig. 6, for uPA the mRNA expression in anti-miR-23b- transfected and 6.18 ± 1.11-fold, respectively, at 48 and 72 h after transfection as compared with control cells (cells plus lanes 3 and 6, P < 0.05). The DOTAP) expression of c-met mRNA in anti-miR-23b-transfected cells increased 1.72 ± 0.27-fold and 2.84 ± 0.64-fold at 48 and 72 h (Fig. 6B, lanes 3 and 6, P < 0.05). As the detectable miR-23b amount shown in Fig. 6C, decreased in transfected cells by 91% and 79%, respec- tively, at 48 and 72 h. All together, these results showed that synthetic anti-miR-23b oligo- ribonucleotides leads to uPA and c-met upregulation, both at the mRNA level and at the protein level, and that transfected anti-miR with endogenous miR decreases the level of detectable miR expression in transfected cells.

miR-23b interacts with the uPA 3¢-UTR

(A,B) RT-PCR evaluation of uPA and c-met mRNAs on total Fig. 4. RNA isolated from control cells (lanes 1 and 4) and miR-23b-trans- fected cells. Lane 2 : 50 nM miR-23b, 48 h. Lane 3 : 100 nM miR- 23b, 48 h. Lane 5 : 50 nM miR-23b, 72 h. Lane 6 : 100 nM miR-23b, 72 h. miR-23b did not affect the steady-state amounts of uPA and c-met mRNAs. (C) Real-time RT-PCR evaluation of miR-23b in control and in transfected cells. RQ.

same fragments

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2971

respectively). The c-met protein expression level increased 2.3 ± 0.51-fold at 72 h as compared with controls (cells plus DOTAP) (Fig. 5C, lanes 5 and 6). Next, we assessed the mRNA expression levels of the target proteins and miR-23b expression. As shown in To verify the putative direct interaction between miR- 23b and the uPA 3¢-UTR, the regions of the 3¢-UTR of human uPA mRNA containing the two putative hsa-miR-23b-binding sites were cloned into the Renilla luciferase report plasmid construct pGL4.71. In partic- ular, a 150 bp sequence (pGL4.71 uPA-3¢-UTR-1S) containing miR-23b-binding site 1 (1S construct) and a 102 bp sequence (pGL4.71 uPA-3¢-UTR-2S) containing miR-23b-binding site 2 (2S construct) were cloned. The respective control constructs were obtained by cloning the in an antisense orientation (pGL4.71 uPA-3¢-UTR-1AS and pGL4.71 uPA-3¢- UTR-2AS). The Renilla constructs were cotransfected with a control firefly luciferase reporter plasmid into SKHep1C3 cells. After 48 h of transfection with 100 nm miR-23b (Fig. 7A), the 1S and 2S constructs inhibited the luciferase relative activity, respectively, by 32% (P < 0.01) and 36% (P < 0.001) (Fig. 7A). Dose–response experiments showed that the inhibition of Renilla luciferase by the 1S construct was 20% and 27%, respectively, at 75 nm and 100 nm miR-23b; for the 2S construct, the maximum inhibition obtained was 29% at 100 nm miR-23b (Fig. 7B). There was no inhibition by the corresponding antisense sequences (Fig. 7C). The data obtained showed that the two putative binding sites of the uPA mRNA 3¢-UTR were targets for hsa-miR-23b, and thus miR-23b could modulate gene expression directly at the uPA 3¢-UTR. As well as assessment of the possible direct inter- action between miR-23b and the c-met 3¢-UTR, the entire c-met 3¢-UTR (2278 bp) was cloned into Renilla luciferase-expressing plasmids and then cotransfected with the firefly luciferase-expressing plasmids. The

A. Salvi et al.

miR-23b downmodulates uPA and c-met

Fig. 5. Anti-miR-23b enhances uPA and c-met protein expression in AB2 cells. Western blot analysis (A) of uPA in the conditioned media of control cells (lanes 1, 2, 4, and 5) and anti-miR-23b-transfected cells (lanes 3 and 6) at 48 and 72 h after transfection, and the corresponding uPA enzymatic activity (B) verified by zymography. (C) Western blot detection of c-met expression in cell extracts from control cells (lanes 1, 2, 4, and 5) and transfected cells (lanes 3 and 6) at 48 and 72 h after transfection.

(Fig. 8B). Furthermore, activity

results showed a greater decrease in luciferase activity as compared with controls, both in miR-23b-transfected and untransfected SKHep1C3 cells. Inhibition was 73% without exogenous miR-23b and 65% (P < 0.001) with miR-23b (Fig. 8A). These results would suggest that the c-met 3¢-UTR is an important target for control of the protein expression level.

UTR-3S (179 bp), and pGL4.71 c-met-3¢-UTR-4S (157 bp), were also tested. Constructs 1S, 2S and 4S inhibited the luciferase relative activity by, respectively, 26%, 20%, and 10%. Construct 3S did not affect the luciferase results obtained with SKHep1C3 cells treated or untreated with miR-23b showed that the 1S, 2S and 4S con- structs were miR-23b sensitive (Fig. 8C). Furthermore, as shown in Fig. 8E, the c-met-3¢-UTR-1S construct was not sensitive to the control miR-107.

In Hela cells, the c-met 3¢-UTR was a target that was responsive to miR-23b (Fig. 8B, P < 0.01), indi- cating that SKHep1C3 rather than Hela cells produce different molecules (including miRs) that may interact with the c-met 3¢-UTR.

miR-23b reduced the migration and proliferation abilities of SKHep1C3 cells

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2972

pGL4.71 called c-met-3¢-UTR-1S To assess the biological effects of miR-23b, we exam- ined its effects on SKHep1C3 cell migration and the migration proliferation. As shown in Fig. 9A, Subsequently, four regions of the 3¢-UTR of c-met mRNA containing putative hsa-miR-23b-binding sites cloned into the pGL4.71 Renilla luciferase report plas- mid, (175 bp), pGL4.71 c-met-3¢-UTR-2S (166 bp), pGL4.71 c-met-3¢-

A. Salvi et al.

miR-23b downmodulates uPA and c-met

that

abilities of miR-23b-transfected cells were significantly affected, by 45% (P < 0.05) and 39% (P < 0.01), respectively, as compared with controls, at 50 nm and 100 nm miR-23b. Regarding cell proliferation, miR- 23b decreased the proliferative ability of SKHep1C3 cells by up to 25% (P < 0.05) at 24 h after seeding, as compared with controls (Fig. 9B). These results suggest that migration and proliferation of SKHep1C3 cells can be modulated by expression of miR-23b. On the other hand, the anti-miR-induced inactivation of miR- 23B in AB2 fibroblasts did not modulate the migration ability of these normal cells (data not shown). This anti-miR-induced uPA and c-met suggests expression is not sufficient to induce motility in these normal cells.

Dysregulation of miR-23b expression in human HCC

In order to determine the potential role of miR-23b in human HCC, we assessed the miR-23b expression level in HCC biopsy specimens and corresponding peritu- moral (PT) tissues. As shown in Fig. 10, among the samples derived from 17 HCC patients, there are two subsets: one (14 ⁄ 17, 82%) showing miR-23b down- regulation in HCC specimens as compared with PT tis- sues; and another showing miR-23b (3 ⁄ 13, 18%) upregulation in HCC specimens (Fig. 10A). This indi- cates dysregulated expression of miR-23b in HCC specimens. The average expression level of miR-23b in HCC specimens is about two-fold lower than in PT tissues (P < 0.01) (Fig. 10B).

Discussion

Fig. 6. Anti-miR-23b enhanced uPA and c-met mRNA expression in AB2 cells. AB2 cells were transfected with the anti-miR oligonu- cleotides as described in Experimental procedures. Forty-eight hours and 72 h after transfection, total RNA isolations from control cells (lanes 1, 2, 4, and 5) and transfected cells (lanes 3 and 6) were examined by RT-PCR to detect uPA mRNA (A) and c-met mRNA (B), and by real-time PCR to detect miR-23b expression (C). RQ.

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2973

MicroRNAs have emerged as important negative regu- lators of gene expression that primarily block transla- tion of target mRNAs. Many miR genes and several tissue ⁄ organ-specific miRs have been identified, and the miR signature in several human diseases, including cancer, has been assessed, but little is known about target mRNAs. As uPA and c-met overexpression in human HCC are strictly associated [27], and uPA regu- lation can also occur at the post-transcriptional level [23], we examined the possibility that a given miR might negatively coregulate uPA and c-met gene expression in human cells. Among the putative miRs targeting uPA or c-met (six and 13, respectively; see Table 1), specific computational searches have shown target sites in the uPA and c-met 3¢-UTRs for miR-23a and miR-23b, indicating two and four target sites in the uPA and c-met 3¢-UTRs, respectively. Therefore,

A. Salvi et al.

miR-23b downmodulates uPA and c-met

A

Fig. 7. miR-23b interacted with uPA 3¢-UTR mRNA. The 150 bp sequence pGL4.71 uPA-3¢-UTR-1S containing miR-23b-binding site 1 and the 102 bp sequence pGL4.71 uPA-3¢-UTR-2S containing miR-23b-binding site 2 were transfected in SKHep1C3 cells to assess the luciferase activity. The respective antisense constructs were used as controls (pGL4.71 uPA-3¢-UTR-1AS and pGL4.71 uPA-3¢-UTR-2AS). The Renilla constructs were cotransfected with a control firefly luciferase report into SKHep1C3 cells. Both site 1 and site 2 inhibited the luciferase relative activity, by 32% (P < 0.01) and 36% (P < 0.001), respectively, 48 h after 100 nM miR-23b transfection (A). Dose–response experiments showed that the inhibition of Renilla luciferase for the 1S region was 20% and 27%, respectively, at 75 and 100 nM miR-23b; for the 2S region, the maximum inhibition obtained was 29% at 100 nM miR-23b (B). the corresponding antisense No inhibition was obtained for sequences (C).

B

were physiologically downregulated. Lower levels of miR-23b expression were detected in the three undif- ferentiated HCC cell lines, SKHep1C3, SKHep1 C3.69.2, and HA22T ⁄ VGH, with high levels of uPA (at both the mRNA and enzymatic activity levels).

C

cells (as

We then examined the role of miR-23b in targeting uPA and c-met following transfection of miR-23b mol- ecules into SKHep1C3 cells. In the transfected cells, the endogenous uPA and c-met protein were downreg- ulated without affecting the steady-state amount of the corresponding mRNAs. Delivery of anti-miR-23b oli- goribonucleotides into AB2 normal cells increased the detectable levels of endogenous uPA and c-met at the mRNA and protein levels. These findings show that uPA and c-met downregulation can be induced by ectopic miR-23b expression in SKHep1C3 malignant cells. The mechanisms by which the endogenous miR- 23b might play a biological role in the physiological negative regulation of uPA and c-met expression in normal AB2 anti-miR-23b transfection induced uPA and c-met upregulation) remain to be elucidated. These findings suggest a regulatory mecha- nism involving the molecules miR-23b, uPA, and c-met, which until now have not been known to interact.

according to thermodynamic parameters, we decided to use miR-23b for experimental validation.

lines. All cell

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2974

The fact that miR-23b ectopic expression in HCC induced endogenous uPA downregulation via cells translation repression, whereas the endogenous miR- 23b in AB2 cells seemed to target uPA mRNA for deg- radation, may be due to the presence of different aden- ylate ⁄ uridylate AU-rich element-binding proteins in these tumor cells and in normal cell lines, respectively increasing or decreasing the stability of uPA mRNA and therefore impairing or favoring their degradation. As the uPA mRNA of certain human tumor cell lines is more stable than that produced by normal cells [20], we speculate that miR-23b may target the uPA mRNA First, we ascertained the expression of miRN-23b, uPA and c-met in AB2 human dermal fibroblasts and in four HCC-derived human cell lines expressed variable levels of miR-23b and the putative uPA and c-met mRNA targets. mR-23b was highly expressed in AB2 normal cells, whereas uPA and c-met

A. Salvi et al.

miR-23b downmodulates uPA and c-met

Fig. 8. miR-23b interacts with c-met 3¢-UTR mRNA. The Renilla luciferase plasmids containing the entire c-met 3¢-UTR (2278 bp) were tested for luciferase activity in SKHep1C3 cells (A) and in Hela cells (B). The luciferase inhibition obtained in SKHep1C3 cells was 73% in the absence of exogenous miR-23b, and 65% (P < 0.001) in the presence of miR-23b (A). The luciferase inhibition obtained in Hela cells was 22% in the presence of miR-23b (B). The Renilla luciferase plasmids pGL4.71 c-met-3¢-UTR-1S (175 bp), pGL4.71 c-met-3¢-UTR-2S (166 bp), pGL4.71 c-met-3¢-UTR-3S (179 bp), and pGL4.71 c-met-3¢-UTR-4S (157 bp), harboring, respectively, putative miR-23b-binding sites 1, 2, 3, and 4, were analyzed for luciferase activity. Sites 1, 2 and 4 inhibited the luciferase relative activity by 26%, 20%, and 10%, respectively. Binding site 3 did not affect luciferase activity (C). Furthermore, results obtained with SKHep1C3 cells treated or untreated with miR-23b showed that sites 1, 2 and 4 were miR-23b-sensitive (D). Site 1 was found to be responsive to miR-23b and not to miR-107, used as control (E).

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2975

molecules at the translation and ⁄ or degradation level, depending on the high or low stability of the mRNAs. Regarding the interaction of miR-23b with target mRNA, the decreased luciferase activity obtained with uPA 3¢-UTR constructs (1S and 2S) and with c-met 3¢-UTR constructs (1S, 2S, 3S, and 4S) indicated that miR-23b can interact with the 1S and 2S target sites within the uPA 3¢-UTR, and with the 1S and 2S target sites within the c-met 3¢-UTR. Furthermore, the inter- action of the 1S target site was sensitive only to miR- 23b and not to miR-107. Hence, regulation of human uPA and c-met expression could be directly mediated by miR-23b. In addition, other miRs targeting the c-met 3¢-UTR (i.e. miR-34a and miR-199a*) [33,34] and ⁄ or unknown regulatory factors may interact with the c-met 3¢-UTR in SKHep1C3 cells, as indicated by the strong inhibition of luciferase activity occurring with or without the presence of cotransfected miR-23b. Furthermore, we cannot exclude an indirect influence of miR-23b on c-met, even if it is very unlikely, at least through uPA. Evidence has been provided that c-met activation may be concomitant with the induc- tion of the uPA proteolysis network [20,21], but not vice versa.

A. Salvi et al.

miR-23b downmodulates uPA and c-met

A

B

Fig. 10. Evaluation of miR-23b expression in human HCC. (A) The miRNA profiling carried out by real-time PCR showed downregula- tion of miR-23b in HCC specimens as compared with PT tissues (14 ⁄ 17, 82%). (B) The average expression level of miR-23b in HCC specimens was decreased by about two-fold as compared with PT tissues (P < 0.01). RQ.

Fig. 9. The migration and proliferation abilities of miR-23b-trans- fected SKHep1C3 cells. Cell migration was evaluated using a 24-well transwell chamber with 8 lm pore inserts. The migration abilities of miR-23b-transfected cells were significantly affected, by 45% (P < 0.05) and 39% (P < 0.01), as compared with controls, at 50 and 100 nM miR-23b, respectively [(A) histograms 3 and 4]. miR- 23b also decreased the proliferative ability of SKHep1C3 cells as compared with controls (B), as evaluated by cell growth experi- ments.

target

to explain why

It

tides 2–7 (this is the case for the 2S uPA target site, the 2S and 4S c-met sites), or with seed matches (even longer than six nucleotides) containing a single bulge, one or a double mismatch, or GU wobbles [this may also be the case for the 1S uPA target site containing a 7mer seed match (to the 5¢-miR nucleotides 2–8) with one GU wobble at posi- tion 7; and for the 1S c-met target site with a 10mer seed match (to the 5¢-miR nucleotides 1–10) with two GU wobbles at positions 7 and 8 and one mismatch at position 9, but with correct pairing at positions 1 and 2)] [15,30–33]. We are aware that miR target recognition principles are a much debated topic, and that, in animals, miR–mRNA target site functional duplexes can be more variable in structure. At pres- ent, there is insufficient understanding of mRNA tar- geting by miRs seed matches of different types might be relevant in miR–target site interaction [31–33]. is commonly believed that bioinformatic prediction, together with experimental biological validation of target sites, could refine the general understanding of target selection and regula- tion by miRs [31,33]. Concerning miR-23b expression in other

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2976

species and the conservation across species of its target sites (a) (http://microrna.sanger.ac.uk), is known that: it Regarding the lack of function of the 3S construct of the c-met 3¢-UTR, this may be due to the fact that the 3S c-met predicted target site has an 11mer seed match (to the 5¢-miR nucleotides 1–11), with one GU wobble at position 2 and three mismatches at posi- tions 8, 9, and 10. The GU wobble probably cannot be tolerated at position 2 (because, usually, the seed sequence starts from the 5¢-miR nucleotide 2). How- ever, it is very well known that, even in human cells, target genes can be regulated by miRs through target sites with perfect seed matches at the 5¢-miR nucleo-

A. Salvi et al.

miR-23b downmodulates uPA and c-met

Experimental procedures

The analysis of miR predicted targets was performed using the algorithm pictar [40].

miR target prediction

cells

(SKHep1: ATCC HTB-52),

Cell cultures

c-met occurs mainly

regulation,

[41], selected from human SKHep1Clone3 (SKHep1C3) HCC-derived the SKHep1C3-transfected cells and the SKHep1C3.69.2 cells [18] obtained from SKHep1C3 cells after two passages in nude mice as HCC xenografts were maintained in Earle’s MEM (Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (Invitrogen) at 37 (cid:2)C in a 5% CO2 incubator. AB2 human dermal fibroblasts [42] and AB2-transfected cells were also cultured under the same conditions. Differentiated human HCC-derived cells (HepG2, ATTC HB-8065; HuH-6) and HA22T ⁄ VGH undifferentiated HCC-derived cells [43] were maintained in supplemented with 10% fetal RPMI-1640 (Invitrogen) bovine serum at 37 (cid:2)C in a 5% CO2 incubator. The HuH-6 and HA22T ⁄ VGH cells were kindly provided by N. D’Alessandro (University of Palermo, Italy). The Hela cells (ATTC CCL-2), derived from human cervical cancer, were cultured in DMEM (Invitrogen) supplemented with 10% fetal bovine serum (Invitrogen) at 37 (cid:2)C in a 5% CO2 incubator.

the miR-23b target sites harbored in the 3¢-UTR of uPA mRNA are highly conserved across species (except site 1 for the uPA Rattus norvegicus 3¢-UTR), whereas the miR-23b target sites 1 and 2 within the c-met 3¢-UTR are mainly conserved in humans and chimpanzees; and (b) miR-23b is expressed in several species (humans, Mus musculus, R. norvegicus, and Canis familiaris), but no data are available on its expression in Pan troglodytes (at least to our know- ledge). It is therefore possible to speculate that the miR-23b-mediated regulation of uPA may occur in humans, mice, dogs, and rats, whereas the miR-23b- in mediated regulation of humans. As mentioned above, the bioinformatic anal- ysis predicts that uPA and c-met levels might be regu- lated by at least six and 13 distinct miRs, respectively (Table 1). Importantly, all other predicted miRs, not experimentally tested in this study, with target sites in the uPA or c-met 3¢-UTR mRNA may independently regulate uPA or c-met expression levels. As already proposed for other genes, this possibility argues in in favor of combinatorial miR target which the translatability of uPA and c-met mRNAs may be subject to combinatorial regulation by multi- ple miRs, including miR-34a and miR-199a*, which are known to target several genes, as well as c-met [28,29].

samples,

the corresponding PT nontumor

Tissues and clinicopathological features of HCC

All of the samples from human HCCs (n = 17), as well as resected 1–2 cm from the malignant tumor, were obtained from the surgically resected specimens for pathological examina- tion. Each biopsy specimen was obtained with the patient’s informed consent under standard conditions of sampling and processing, as previously described [26]. Each specimen was considered to be HCC or PT after pathological examination.

that

that of other molecules

In this study, 17 HCC subjects underwent surgical resec- tion, including 14 men and three women (all Italian), ranging from 49 to 79 years of age (mean age, 64.8 ± 7.8 years). None had any apparent distant metastasis, and none had been previously treated for HCC. The PT tissues of HCCs showed different background disease (11 cirrhosis, two hepatitis, three steatosis, one hemosiderosis), and most of the patient sera were submitted for common analysis for the presence of hepatitis B virus (HBV) and hepatitis C virus (HCV), as well as for the determination of a-feto- protein. Regarding hepatitis viruses, seven patients were positive for HCV, three for HBV, and three for HBV and HCV, and four were found to be negative. Among the 15 the a-fetoprotein level was lower than patients tested,

It should be added that miR-23b may also be rele- vant in human disease, as miR-array studies have shown that: (a) miR-23b is downregulated in Alzhei- mer’s disease brains [34]; and (b) it is one of the 10 upregulated miRs in bladder cancer and one of the top five dysregulated miRs in uterine leiomyoma [35,36]. Furthermore, we have shown here with real- time data that miR-23b is downregulated in a subset of HCC cells. The fact the endogenous uPA and c-met expression in HCC cells can be downregu- lated by transfected miR-23b molecules, also reduc- ing cell migration and proliferation abilities, raises the issue of a new approach to silence uPA and c-met of human HCC xenografts in nude mice to determine the effects on tumor growth and metas- regulation of uPA and tasis. As pharmacological c-met has proved to be a challenge in cancer, this regulatory mechanism may potentially be a new tool with which to alter the expression of uPA and (e.g. c-met, as well as shRNA, uPA inhibitors, and tyrosine protein kinase inhibitors) [37–39].

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2977

In summary, our results suggest that ectopic miR- 23b overexpression in HCC cells can lead to uPA and c-met downregulation and to a decreased migratory ability of these malignant cells.

A. Salvi et al.

miR-23b downmodulates uPA and c-met

10 ngÆmL)1 in eight patients, between 11 and 100 ngÆmL)1 in six, and above 100 ngÆmL)1 in one. Each HCC was also examined for several pathological characteristics, but none of those considered (i.e. size of the tumor, degree of differ- entiation, and the presence of pseudocapsule invasion) was related to miR-23b expression in HCC.

Transient transfection of SKHep1C3 cells with miR-23b and luciferase construct

SKHep1C3 cells were seeded at a confluency of 60–80%, and then transfected with 100 nm double-stranded hsa-miR-23b, 24 h after seeding, and with the luciferase reporter constructs (0.5 lg), 48 h after seeding, using Lipofectamine 2000 trans- fection reagents according to the manufacturer’s instruction. Seventy-two hours after seeding, the cells were washed with NaCl ⁄ Pi and lysed with passive lysis buffer (Promega, San Diego, CA, USA), and firefly luciferase (f-luc) and Renilla luciferase (r-luc) activities were determined using the dual-luciferase reporter assay system (Promega) and a luminometer. The relative reporter activity was obtained by normalization to the f-luc activity. For the dose–response experiments, the cellular conditions and transfection times were the same, and the concentration of miR-23b was increased from 0 to 100 nm. The transient transfections of Hela cells with miR-23b and luciferase construct were performed under the same experimental conditions.

the human uPA and MET 3¢-UTRs were Segments of PCR-amplified from cDNA from SKHep1C3 cells, using primers containing flanking XbaI recognition sequences (Table S1). PCRs were performed using PFU Taq polymer- ase (Promega, San Diego, CA, USA) with proofreading activity. The PCR products were ligated in the XbaI restric- tion site downstream of the Renilla luciferase coding region of the pGL4.71 vector (Promega), in which the simian virus region from the pGL3-Promoter vector 40 promoter (Promega) was previously cloned to obtain the pGL4.71P plasmid. The correct orientation of the insert was verified by sequencing.

3¢-UTR constructs

Real-time quantification of mature miR-23b by stem–loop RT-PCR

and

single-strand

The total RNAs from tissue samples, transfected and non- transfected cells were isolated using TRIzol reagent (Invi- trogen), according to the manufacturer’s instructions. For a quantitative analysis of mature miR-23b, a two-step Taq- Man real-time PCR analysis was performed, using primers and probes obtained from Applied Biosystems (Foster, CA, USA).

(internal

serum. The

Molecules of dsRNAs that mimic endogenous hsa-miR- 23b mature miR (5¢-AUCACAUUGCCAGGGAUUACC- anti-miR-23b miR inhibitor 3¢) (oligoribonucleotides), designed to inhibit endogenous hsa-miR-23b, were purchased from Ambion (Austin, TX, USA). SKHep1C3 cells were seeded in complete medium at 80% confluence in a 24-well Costar (8 · 104 cells per well). The cells were transfected into serum-free Earle’s MEM, 24 h after seeding, at 50 and 100 nm of double- stranded hsa-miR-23b, using DOTAP liposomal transfec- tion reagents according to the manufacturer’s instruction (Invitrogen). The transfection medium was replaced, after 24 h, with Earle’s MEM supplemented with 10% conditioned media and cell fetal bovine extracts were prepared for analysis 48 and 72 h after the transfection.

cDNA was synthesized from total RNA (50 ng) in 15 lL reactions, using reverse transcriptase and the stem–loop pri- mer for miR-23b (Applied Biosystems; PN 4373073) or control; Applied Biosystems; PN RNU66 4373382) contained in the TaqMan MicroRNA Reverse Transcription kit (Applied Biosystems, Foster, CA, USA). The reverse transcriptase reaction was performed by incu- bating the samples at 16 (cid:2)C for 30 min, 42 (cid:2)C for 30 min, and 85 (cid:2)C for 5 min. The PCR reaction (20 lL) contained 1.3 lL of reverse transcriptase product, 10 lL of Taq-Man 2· Universal PCR Master Mix, and 1 lL of the appropri- ate TaqMan MicroRNA Assay (· 20) containing primers and probes for the miR of interest. The PCR mixtures were incubated at 95 (cid:2)C for 10 min, and this was followed by 40 cycles of 95 (cid:2)C for 15 s and 60 (cid:2)C for 60 s. PCRs were per- formed in triplicate using a 7500 real time PCR system. The expression of miR-23b was based on the DDCT method, using RNU66 as an internal control.

Transient transfection of SKHep1C3 cells with miR-23b and transient transfection of AB2 cells with anti-miR-23b

serum. The

For the transient transfection of AB2 cells with anti- miR-23b, the AB2 cells were seeded in complete medium at 80% confluence in a 24-well Costar (5 · 104 cells per well) and then transfected in the serum-free Earle’s MEM, 24 h after seeding, with 100 nm single-stranded anti-miR-23b miR inhibitor, using DOTAP transfection reagents according to the manufacturer’s instruction (Invi- replaced, after trogen). The transfection medium was 24 h, with Earle’s MEM supplemented with 10% fetal bovine conditioned media and cell extracts were prepared for analysis 48 and 72 h after the transfection.

Total RNA of transfected and nontransfected cells was extracted with Trizol reagent according to the manufacturer’s

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2978

RT-PCR analysis (Agilent) and real-time PCR

A. Salvi et al.

miR-23b downmodulates uPA and c-met

Denmark) corresponding to the complementary sequences of the mature miR-23b (miR-23b probe: 5¢-ggtaatccctggcaa tgtgat-3¢). As a control, the blots were hybridized with an oligonucleotide probe complementary to the U6 RNA (U6 probe: 5¢-cacgaatttgcgtgtcatcctt-3¢). The blots were incu- bated with streptavidin-coupled horseradish peroxidase (Signosis), and this was followed by enhanced chemilumi- nescence (Thermo Scientific, Rockford, IL, USA) detection. results were visualized on X-ray film (Thermo The Scientific).

(nucleotides

rabbit

immunoreacted using

instructions (Invitrogen), and quantified with a NanoDrop ND-1000 spectrophotometer (NadoDrop Tech., Wilming- ton, DE, USA). Total RNA (1 lg) was first treated with DNase (0.02 UÆlL)1) for 5 min at room temperature, and then inactivated for 10 min at 65 (cid:2)C; subsequently, total RNA was reverse-transcribed using random hexamers. A PCR was performed using glyceraldehyde 3-phosphate dehy- drogenase (GAPDH1 ⁄ 2) specific primers [44] to verify the integrity of RNA and adequate cDNA synthesis. Endoge- nous hsa-miR-23b precursor was detected by PCR analysis with miR-23b P forward and miR-23b P reverse specific primers (according to the hsa-miR-23b precursor sequence the Sanger Institute; http:// posted in the miRBase of forward, microrna.sanger.ac.uk/sequences): miR-23b P 5¢-GGTGCTCTGGCTGCTTGG-3¢ 5–22); miR-23b P reverse, 5¢-GCCAAGGTCGTGGTTGCG-3¢ (nucleotides 80–97). The endogenous uPA expression was detected using uPA1 and uPA2 specific primers [44]. Endoge- nous c-met expression was determined using MET-1 and MET-2 specific primers [19]. The amplification reactions for a given gene were performed using a limited number of PCR cycles in a range far from plateau values, which were previ- ously determined [26] (data not shown). The relative amounts of the PCR products were determined using an Agilent 2100 Bioanalyzer (Agilent Tech., Palo Alto, CA, USA) [18]. The Agilent 2100 Bioanalyzer allowed us to quantify up to 12 samples on a chip consisting of wells connected by micro- channels filled with a specific gel–dye mix. The sample and ladder wells were filled with a DNA marker solution before addition of 1 lL of sample. The chip was spun and placed into the Bioanalyzer for subsequent analysis. The relative amounts (expressed in ngÆlL)1) of target gene PCR products were referred to their respective GAPDH amounts. The experiments were performed twice for reproducibility.

The expression of uPA and c-met mRNAs in the cell lines shown in Fig. 2 was evaluated using a TaqMan Gene Expression Assay (Applied Biosystems). GAPDH was used as an internal standard. The PCR mixture (25 lL) contain- ing 1 lL of the specific probe, 11.25 lL of cDNA (synthe- sized as described above) and 13.75 lL of Taq-Man 2· Universal PCR Master Mix were incubated initially at 55 (cid:2)C for 2 min, and then at 95 (cid:2)C for 10 min; this was followed by 40 cycles of 95 (cid:2)C for 15 s and 60 (cid:2)C for 60 s. Determination of the expression of uPA and c-met mRNAs was based on the DDCT method.

The 48 and 72 h conditioned media were collected from cultures of both nontransfected and transfected cells. Con- stant amounts of proteins were loaded, under nonreducing conditions, on a Novex NuPAGE (4–12%) Bis ⁄ Tris gel (Invitrogen), or on an 8% SDS polyacrylamide gel, which was blotted onto a nitrocellulose membrane (NM). NMs were anti-(human uPA) (1 : 1000 in 1% BSA) and alkaline phosphatase-conjugated anti-rabbit IgG (1 : 7500 in 0.3% BSA); for zymography, NMs were overlayed onto casein agar containing 2 lgÆmL)1 human plasminogen (Technoclone GmbH, Vienna, Austria) to evaluate uPA activity (EC 3.4.21.73) [41]. The cell extracts were collected from 48 h and 72 h cultures of non- transfected and transfected cells by adding 0.05% SDS [41]. Constant amounts of proteins were loaded, under reducing conditions, on a Novex NuPAGE (4–12%) Bis ⁄ Tris gel or on an 8% SDS polyacrylamide gel, and were then trans- ferred to NMs. The blots were immunoreacted using rabbit anti-(human c-met) (1 : 1000 in 0.3% BSA) and alkaline phosphatase-conjugated anti-rabbit IgG (1 : 7500 in 0.3% BSA), or with the mouse monoclonal antibodies anti-GAP- DH (1 : 300 in 1% BSA) and alkaline phosphatase-conju- gated anti-mouse IgG (1 : 7500 in 0.3% BSA). The results of the immunoreaction were detected with Nitroblue tetra- zolium and bromochloroindolyl phosphate (Promega). The bands corresponding to c-met (170 and 145 kDa) and GAPDH were scanned and analyzed using a digital system (IAS), and the integrated optical density (IOD) values were expressed in pixels. The expression percentages were the sum of the IODs of the bands corresponding to the precur- sor (170 kDa) and the b-chain (145 kDa) forms of c-met protein [19]. The amount of target protein in SKHep1C3 control cells treated with transfection reagents was consid- ered to be 100%.

Western blotting and zymography

Northern blotting

For the motility assay, cells (8 · 104 cells per well) were loaded on transwell 8.0 lm polycarbonate membrane inserts (Corning Incorporated, Lowell, MA, USA) in tripli- cate wells. Cells were seeded in the upper chamber in

The total RNA samples (7.5 lg each) were electrophoresed on 15% TBE ⁄ urea precast gels (Invitrogen) and transferred to a Nylon+ membrane (Invitrogen). The hybridization buffer, wash buffer, blocking buffer and detection wash buffer were supplied by Signosis (Sunnyvale, CA, USA). Membranes were hybridized (42 (cid:2)C for 16 h) with 5¢-biotin- labeled mercury LNA detection probes (Exiqon, Vedbaek,

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2979

Cell migration and cell proliferation assays

A. Salvi et al.

miR-23b downmodulates uPA and c-met

5 Filipowicz W, Bhattacharyya SN & Sonenberg N (2008)

Mechanisms of post-transcriptional regulation by microRNAs: are the answers in sight? Nat Rev Genet 9, 102–114.

6 Lewis BP, Burge CB & Bartel DP (2005) Conserved

seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets. Cell 120, 15–20.

7 Zhang B, Pan X, Cobb GP & Anderson TA (2007)

serum-free Earle’s MEM (nontranfected cells) or in serum- free Earle’s MEM with transfection reagent and hsa-miR- 23b, at concentrations of 50 and 100 nm (transfected cells). Serum-free Earle’s MEM (0.5 mL) was added to the bot- tom chamber, and the plates were incubated at 37 (cid:2)C in a 5% CO2 incubator. After 24 h, the medium in the bottom chamber was removed, and 0.5 mL of AB5 human fibro- blast serum-free conditioned medium was used as chemo- attractant. After 24 h at 37 (cid:2)C in a 5% CO2 incubator, the cells that had migrated to the bottom chamber were trypsi- nized and counted.

MicroRNAs as oncogenes and tumor suppressors. Dev Biol 302, 1–12.

8 Lakshmipathy U, Love B, Goff LA, Jornsten R,

For cell proliferation assays, SKHep1C3 and SKHep1C3 transfected cells were seeded (5 · 104 cells per dish) in trip- licate in 5 cm diameter Petri dishes. After 24 h, 48 h and 72 h of incubation at 37 (cid:2)C, the cells were trypsinized, and counted using a Burker chamber.

Graichen R, Hart RP & Chesnut JD (2007) MicroRNA expression pattern of undifferentiated and differentiated human embryonic stem cell. Stem Cell Dev Dec 16, 1003–1016.

9 Laurent LC, Chen J, Ulitsky I, Mueller FJ, Lu C,

Shamir R, Fan JB & Loring JF (2008) Comprehensive microRNA profiling reveals a unique human embryonic stem cell signature dominated by a single seed sequence. Stem Cells 26, 1506–1516.

10 Rosenfeld N, Aharonov R, Meiri E, Rosenwald S,

Spector Y, Zepeniuk M, Benjamin H, Shabes N, Tabak S, Levy A et al. (2008) MicroRNAs accurately identify cancer tissue origin. Nat Biotechnol 26, 462–469. 11 Budhu A, Jia HL, Forgues M, Liu CG, Goldstein D,

Each experiment was carried out at least twice. Histograms represent the mean values, and bars indicate standard errors of the mean. The statistical significance of the results was determined using Student’s t-test, with data considered significant when P £ 0.05. Statistical analysis was per- formed with kyplot (v.2.0b15, http://www.woundedmoon. org/win32/kyplot.html).

Statistical analysis

Acknowledgements

Lam A, Zanetti KA, Ye QH, Qin LX, Croce CM et al. (2008) Identification of metastasis-related microRNAs in hepatocellular carcinoma. Hepatology 47, 897–907. 12 Schetter AJ, Leung SY, Sohn JJ, Zanetti KA, Bowman ED, Yanaihara N, Yuen ST, Chan TL, Kwong DL, Au GK et al. (2008) MicroRNA expression profiles associated with prognosis and therapeutic outcome in colon adenocarcinoma. JAMA 30, 425–436.

13 Kuhn DE, Martin MM, Feldman DS, Terry AV Jr,

Nuovo GJ & Elton TS (2008) Experimental validation of miRNA targets. Methods 44, 47–54.

14 Kru¨ tzfeldt J, Poy MN & Stoffel M (2006) Strategies to determine the biological function of microRNAs. Nat Genet 38(Suppl. S1), 4–19.

15 Mazie` re P & Enright AJ (2007) Prediction of

This work was supported by the Ministero dell’Istruzi- one, dell’Universita` e della Ricerca (MIUR), MIUR PRIN 2007 (MWCEAL__003), the Centro Eccellenza IDET, Fondazione Cariplo and Consiglio Nazionale delle Ricerche and MIUR within the context of the Progetto Strategico Oncologia. The authors would like to thank E. Frassi MD for collecting clinical data of HCC patients, and Dr R. Coates (Centro Linguistico, University of Brescia), medical writer, for his linguistic revision of the manuscript.

References

microRNA targets. Drug Discov Today 12, 452–458. 16 Andreasen PA, Kjoller L, Christensen L & Duffy MJ

1 Bartel DP (2004) MicroRNAs: genomics, biogenesis,

mechanism, and function. Cell 116, 281–297.

2 Bushati N & Cohen SM (2007) MicroRNAs functions.

(1997) The urokinase-type plasminogen activator system in cancer metastasis: a review. Int J Cancer 72, 1–22. 17 Gao CF & Vande Woude GF (2005) HGF ⁄ SF-Met sig-

Annu Rev Cell Dev Biol 23, 175–205.

naling in tumor progression. Cell Res 15, 49–51. 18 Salvi A, Arici B, Alghisi A, Barlati S & De Petro G

(2007) RNA interference against urokinase in hepatocel- lular carcinoma xenografts in nude mice. Tumor Biol 28, 16–26.

3 Bentwich I, Avniel A, Karov Y, Aharonov R, Gilad S, Barad O, Barzilai A, Einat P, Einav U, Meiri E et al. (2005) Identifications of hundreds of conserved and non- conserved human microRNAs. Nat Genet 37, 766–770. 4 Berezikov E, Guryeev V, van de Belt J, Wienholds E,

19 Salvi A, Arici B, Portolani N, Giulini SM, De Petro G & Barlati S (2007) In vitro c-met inhibition by antisense RNA and plasmid-based RNAi down-modulates

Plasterk RH & Cuppen E (2005) Phylogenetic shadowing and computational identification of human microRNA genes. Cell 120, 21–24.

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2980

A. Salvi et al.

miR-23b downmodulates uPA and c-met

32 Friedman RC, Kai-How Farh K, Burge CB & Bartel

migration and invasion of hepatocellular carcinoma cells. Int J Oncol 31, 451–460.

20 Jeffers M, Rong S & Woude GFV (1996) Enhanced

DP (2009) Most mammalian mRNAs are conserved tar- gets of microRNAs. Genome Res 19, 92–105.

33 Bartel DP (2009) MicroRNAs: target recognition and

regulatory functions. Cell 136, 215–233.

tumorigenicity and invasion-metastasis by hepatocyte growth factor ⁄ scatter factor-met signalling in human cells concomitant with induction of the urokinase prote- olysis network. Mol Cell Biol 16, 1115–1125.

34 Wang WX, Rajeev BW, Stromberg AJ, Ren N, Tang G, Huang Q, Rigoutsos I & Nelson PT (2008) The expression of microRNA miR-107 decreases early in Alzheimer’s disease and may accelerate disease pro- gression through regulation of beta-site amyloid pre- cursor protein-cleaving enzyme 1. J Neurosci 28, 1213–1223.

21 Lee KH, Choi EY, Kim MK, Hyun MS, Jang BI, Kim TN, Kim SW, Song SK, Kim JH & Kim JR (2006) Regulation of hepatocyte growth factor-mediated uroki- nase plasminogen activator secretion by MEK ⁄ ERK activation in human stomach cancer cell lines. Exp Mol Med 38, 27–35.

22 Pakneshan P, Tetu B & Rabbani SA (2004) Demethyla- tion of urokinase promoter as a prognostic marker in patients with breast carcinoma. Clin Cancer Res 10, 3035–3041.

23 Nagamine Y, Medcalf R & Munoz-Canoves P (2005) Transcriptional and post-transcriptional regulation of the plasminogen activator system. Thromb Haemost 93, 661–675.

35 Gottardo F, Liu CG, Ferracin E, Calin GA, Fassan M, Bassi P, Sevignani C, Byrne D, Negrini M, Pagano F et al. (2007) Micro-RNA profiling in kidney and blad- der cancers. Urol Oncol Semin Orig Invest 25, 387–392. 36 Wang T, Zhang X, Obijuru L, Laser J, Aris V, Lee P, Mittal K, Soteropoullos P & Wei JJ (2007) A micro- RNA signature associated with race, tumor size, and target gene activity in human uterine leiomyosarcomas. Genes Chromosome Cancer 46, 336–347.

37 Schmitt M, Harbeck N, Thomssen C, Wilhelm O,

24 Morovoz VM, Massol NA, Vlamidirova OV,

Maulo GG & Ishov AM (2007) Regulation of c-met expression by transcription repressor Daxx. Oncogene 22, 1–10.

Magdolen V, Reuning U, Ulm K, Hofler H, Janicke F & Graeff H (1997) Clinical impact of the plasminogen activation system in tumor invasion and metastasis: prognosis relevance and target for therapy. Thromb Haemost 78, 285–296.

38 Guo YJ, Mazar AP, Lebrun JJ & Rabbani SA (2002)

25 Thorgeirsson SS & Grisham JW (2002) Molecular path- ogenesis of human hepatocellular carcinoma. Nat Genet 31, 339–346.

An antiangiogenic urokinase-derived peptide combined with tamoxifen decreases tumor growth and metastasis in a syngeneic model of breast cancer. Cancer Res 62, 4678–4684.

26 De Petro G, Tavian D, Copeta A, Portolani N, Giulini SM & Barlati S (1998) Expression of urokinase-type plasminogen activator (u-PA), u-PA receptor, and tis- sue-type PA messenger RNAs in human hepatocellular carcinoma. Cancer Res 58, 2234–2239.

39 Corso S, Comoglio PM & Giordano S (2005) Cancer therapy: can the challenge be MET? Trends Mol Med 11, 284–292.

40 Krek A, Gru¨ n D, Poy MN, Wolf R, Rosenberg L,

27 Tavian D, De Petro G, Benetti A, Portolani N, Giulini SM & Barlati S (2000) u-PA and c-MET mRNA expression is co-ordinately enhanced while hepatocyte growth factor mRNA is down regulated in human hepatocellular carcinoma. Int J Cancer 87, 644–649.

Epstein EJ, MacMenamin P, da Piedade I, Gunsalus KC, Stoffel M et al. (2005) Combinatorial microRNA target prediction. Nat Genet 37, 495–500.

28 Sun F, Fu H, Liu Q, Tie Y, Zhu J, Xing R, Sun Z &

Zheng X (2008) Downregulation of CCND1 and CDK6 by miR-34a induces cell cycle arrest. FEBS Lett 582, 1564–1568.

41 Salvi A, Arici B, De Petro G & Barlati S (2004) Small interfering RNA urokinase silencing inhibits invasion and migration of human hepatocellular carcinoma cells. Mol Cancer Ther 3, 671–678.

42 De Petro G, Copeta A & Barlati S (1994) Urokinase- type and tissue-type plasminogen activators as growth factors of human fibroblasts. Exp Cell Res 213, 286– 294.

43 Notarbartolo M, Poma P, Perri D, Dusonchet L,

29 Kim S, Lee UJ, Kim MN, Lee E-J, Kim JY, Lee MY, Choung S, Kim YJ & Choi YC (2008) MicroRNA miR-199a* regulates the MET proto-oncogene and the downstream extracellular signal-regulated kinase 2 (ERK2). J Biol Chem 283, 18158–18166.

30 Brennecke J, Stark A, Russel RB & Cohen SM (2005) Principles of microRNA-Target recognition. PLoS Biol 3, 404–418.

Cervello M & D’Alessandro N (2005) Antitumor effects of curcumin, alone or in combination with cisplatin or doxorubicin, on human hepatic cancer cells. Analysis of their possible relationship to changes in NF-kB activation levels and IAP gene expression. Cancer Lett 224, 53–65.

31 Brodersen P & Voinnet O (2009) Revisiting the princi- ples of microRNA target recognition and mode of action. Nat Rev Mol Cell Biol 10, 141–148.

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2981

A. Salvi et al.

miR-23b downmodulates uPA and c-met

44 Tavian D, Salvi A, De Petro G & Barlati S (2003) Sta- ble expression of antisense urokinase mRNA inhibits the proliferation and invasion of human hepatocellular carcinoma cells. Cancer Gene Ther 10, 112–120.

This supplementary material can be found in the online version of this article.

Supporting information

Please note: Wiley-Blackwell is not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corre- sponding author for the article.

FEBS Journal 276 (2009) 2966–2982 ª 2009 The Authors Journal compilation ª 2009 FEBS

2982

used to The following supplementary material is available: the generate Table S1. Oligonucleotides 3¢-UTR constructs of the human uPA and MET genes.