FEBS Letters 585 (2011) 2965–2971

journal homepage: www.FEBSLetters .org

Ago1 and Ago2 differentially affect cell proliferation, motility and apoptosiswhen overexpressed in SH-SY5Y neuroblastoma cells

Chiara Parisi a, Corinna Giorgi b, Enrico Maria Batassa a, Laura Braccini a, Giovanna Maresca c,Igea D’agnano c, Viviana Caputo a, Annamaria Salvatore c, Flavia Pietrolati a, Carlo Cogoni a,b,Caterina Catalanotto a,b,⇑a Department of Cellular Biotechnology and Hematology, Sapienza University, Rome, Italyb European Brain Research Institute ‘‘Rita Levi-Montalcini’’, Rome, Italyc CNR-Institute of Cell Biology and Neurobiology, Rome, Italy

a r t i c l e i n f o

Article history:Received 1 July 2011Revised 1 August 2011Accepted 1 August 2011Available online 11 August 2011

Edited by Tamas Dalmay

Keywords:ArgonauteCell proliferationSH-SY5YCancermiRNA

0014-5793/$36.00 � 2011 Federation of European Biodoi:10.1016/j.febslet.2011.08.003

⇑ Corresponding author. Address: Department ofHematology, Viale Regina Elena 324, Sapienza Univer+39 06 4457731.

E-mail address: catala@bce.uniroma1.it (C. Catalan

a b s t r a c t

Argonaute are a conserved class of proteins central to the microRNA pathway. We have highlighted anovel and non-redundant function of Ago1 versus Ago2; the two core factors of the miRNA-associ-ated RISC complex. Stable overexpression of Ago1 in neuroblastoma cells causes the cell cycle toslow down, a decrease in cellular motility and a stronger apoptotic response upon UV irradiation.These effects, together with a significant increase in p53 levels, suggest that Ago1 may act as atumor-suppressor factor, a function also supported by GEO Profiles microarrays that inversely cor-relate Ago1 expression levels with cell proliferation rates.� 2011 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.

1. Introduction sufficient to induce its repression [2], demonstrating the centrality

Argonaute (Ago) proteins belong to a conserved gene familyfound in plants, yeast, Drosophila and mammals. The mammalianprotein family consists of the Ago and Piwi subfamilies, eachrepresented by four members [1]. Ago proteins are ubiquitouslyexpressed and their structure consists of a less conserved amino-terminal domain and three highly conserved domains: the Mid do-main and the signature PAZ and Piwi domains. Despite their highsequence similarity, the endonuclease activity is unique of Ago2.

In mammals, the best-studied members of this family are Ago1and Ago2 as they represent the major component of RISC, the micr-oRNA (miRNA) effector complex.

miRNAs are genome encoded small RNA molecules derivingfrom processing of longer double stranded precursors, which con-trol gene expression post-transcriptionally. miRNAs interact withAgo proteins and guide the RISC complex on bound mRNA targets,leading to their degradation or translational repression. Artificialtethering of Ago proteins to the 30UTR of a reporter mRNA is

chemical Societies. Published by E

Cellular Biotechnology andsity, 00161 Rome, Italy. Fax:

otto).

of these factors in miRNA mediated regulation.By regulating the post-transcriptional expression of a large

number of genes, miRNAs control a wide range of physiologicalprocesses, including proliferation, differentiation, cell metabolismand migration. Indeed, their deregulation has been associated withseveral human diseases, including neurodegenerative processesand cancer [3,4]. In particular, miRNAs can function as oncogenesor tumor suppressors by targeting genes which are key regulatorsof tumor progression and tumorigenesis is often accompanied by ageneral downregulation of miRNAs [3]. Additionally, both up- anddown-regulation of proteins involved in miRNA biogenesis, such asDrosha and Dicer, are associated with cancer progression [5,6].Being so central to miRNA function, misregulation of Ago proteinsshould likely affect cell proliferation and differentiation. However,little is known about either direct or miRNA-mediated effects ofthese proteins on the cell cycle. Further, PAR-CLIP experimentsand profiles of Ago1 and Ago2-associated miRNAs seem to indicatethat Ago proteins bind the same miRNAs and mRNA targets, sug-gesting a functional redundancy of these two factors [7,8].

To challenge this view and investigate potential non-redundantroles of mammalian Ago1 and Ago2, we overexpressed both pro-teins in a neuroblastoma cell line. We observed that Ago1 has aclear inhibitory effect on cell proliferation and motility while

lsevier B.V. All rights reserved.

2966 C. Parisi et al. / FEBS Letters 585 (2011) 2965–2971

eliciting apoptosis, a behaviour evocative of most tumor suppres-sor factors.

2. Materials and methods

Quantitative data are represented as the mean ± SD. The Stu-dent’s t-test was used for statistical analysis (⁄P < 0.05 and⁄⁄P < 0.001).

DNA constructs, primer sequences, buffers, cell culture proto-cols, antibodies and sources are described in Supplemental Data.

2.1. Stable cell lines generation

SH-SY5Y cells were transfected with pCMV-Flag-Ago1, pCMV-Flag-Ago2 or pCMV-Flag-empty vector using Lipofectamine 2000.Cells were diluted to clonal density 24 h post-transfection, and se-lected based on their resistance to G418 (300 lM for 10 days). Singleclones were picked and independently grown under selection, to ob-tain monoclonal cell lines; the remaining clones from each platewere expanded to generate polyclones. Monoclones and polycloneswere further selected based on their Flag-Ago expression.

2.2. Induction of neuronal differentiation

Eighty percentage confluent SH-SY5Y cells were exposed to 1%FBS and 10 lM Retinoic Acid for five days. At 0, 1, 2, 3 and 5 daysof treatment, images of the differentiating cells were taken withNikon ECLIPSE TE2000-E inverted microscope with 10–20 � lenses.Neurite outgrowth was measured using NIS software. At least 100neurites were measured for each sample.

2.3. Western blot analysis and RT-PCR

Cells were lysed with lysis buffer, spun 10 min at 10000g, andsamples were subjected to western blot analysis. Semi-quantita-tive RT-PCR was performed using One-step SuperscriptIII RT-PCRfollowing manufacturer protocols with oligos Ago1f, Ago1r, Ago2f,and Ago2r.

2.4. Trypan blue exclusion and FACS analyses

2 � 104 cells were seeded in 48-well plates and trypsinized, try-pan blue stained and counted at 24, 48 and 72 h. In FACS analyses,7.5 � 105 cells were plated in 10 cm dishes, trypsinized 24 h laterand fixed at a concentration of 1 � 106 cells/ml. 500.000 cells werePBS washed and resuspended in RNaseA (75 KU/ml) and propidiumiodide (50 lg/ml) overnight. Samples were measured with a FAC-Scan cytofluorimeter and 20,000 events/samples were acquiredusing CELLQuest BD software. Cell cycle phases percentages wereestimated using ModFit software.

2.5. Wound healing and motion analyses

A confluent plate of cells was wounded with a tip to induce theedge of the monolayer of cells to heal the empty space. Cells buf-fered with 15 mM HEPES (pH 7.2) at 37 �C and a time-lapse moviewas taken acquiring images every 5 min for 18 h. A Zeiss Axiovert-35 microscope equipped with a JVC digital CCD camera (50� objec-tive) and the IAS2000 software (Deltasistemi) was used. The speedof migration was calculated dividing the average distances coveredby individual nuclei by the time of observation.

2.6. Cell invasion assay

In vitro cell invasion assays were performed using Matrigel-coated inserts according to manufacturer protocols, with

2.5 � 105 cells/insert and10% FBS as chemoattractant. After 24 h,migrated cells were ethanol fixed and stained with 0.1% crystal vio-let in 20% methanol. Images of five different fields (10� objective)for each filter were taken and cells were counted. The mean of fourexperiments done in duplicate was evaluated for each clone.

2.7. UV irradiation

Eighty percentage confluent cells were exposed to UV irradia-tion (8 mJ/cm2) 1 h post-medium replacement, and incubated for24 h. After fixing with 4% PFA, apoptotic and non-apoptotic nucleiwere Hoechst-stained and counted. In each experiment apoptoticnuclei were counted in 10 different areas (40 � objective) of eachcoverslip and normalized against the total number of cells in thefield.

3. Results

3.1. Ago1 overexpressing cells show a proliferation defect

We generated stable SH-SY5Y neuroblastoma cell linesoverexpressing Ago1 and Ago2 proteins to test whether they par-ticipate in modulating cell cycle and differentiation pathways.SH-SY5Y cells were chosen because both proteins are normally ex-pressed in this cell type [9,10].

We choose to overexpress rather than downregulate these fac-tors, to avoid compensatory effects of related members of thishighly conserved family of proteins. To this aim, we have stablytransfected SH-SY5Y cells with plasmids expressing either FLAG-Ago1, FLAG-Ago2 or a control empty vector. Three independentmonoclones and a polyclone were selected for each Ago proteinbased on their expression levels detected by western blot assaysand semi-quantitative RT-PCR (Fig. 1).

We initially tested potential differences between the prolifera-tion rates of our clones by comparing their growth curves over a72 h period. As shown in Fig. 2A, Ago1 overexpression results ina significant reduction of the monoclones’ proliferation rate(mAgo1). In particular, already 48 h after plating we registered a�50% reduction of mAgo1’s growth rate compared to both controland Ago2 monoclones (mAgo2). Similarly, Ago1-overexpressingpolyclone (PolAgo1) shows a comparable proliferative defect withrespect to both control and Ago2 polyclone (PolAgo2) (Fig. 2B).

We next asked whether the different proliferation rates of ourclones correlated with a different cell cycle distribution. Propidiumiodide and FACS analyses of both monoclones and polyclones re-vealed that Ago1 overexpression causes a 20% increase in G1/G0cells concomitantly with a 15% decrease of cells in S phase(Fig. 3). Ago2, on the other hand, does not seem to affect cell cycledistribution when overexpressed.

Moreover, lack of sub-G1 peaks in the FACS analysis, indicatethat Ago1 overexpression does not induce cell death of SH-SY5Y(data not shown).

These results indicate that Ago1 overexpression slows downSH-SY5Y cell cycle, likely as a consequence of a delay in G1/S tran-sition, while upregulation of Ago2 does not exert any effect.

3.2. Both Ago1 and Ago2 overexpression induce SH-SY5Ydifferentiation

SH-SY5Y cells can be induced to differentiate into neuronal-likecells upon five days of all-trans retinoic acid (RA) treatment [11].Thus, we examined whether Ago1 and Ago2 overexpression can in-duce differentiation per se or have an effect on RA-induced differ-entiation. Neurite length and expression of neuronal proteinmarkers, such as gap-43 and bIII-tubulin, were chosen as parame-ters of the differentiation status. Even before RA treatment, neurite

A

B

C

Fig. 1. Overexpression of Ago1 and Ago2 proteins in SH-SY5Y. Western Blotanalyses with anti-Flag antibody, showing the expression of Flag-Ago1 (A) and Flag-Ago2 (B) in three monoclones (mAgo) and in a polyclone (PolAgo) normalizedagainst b-actin. Ctr: cells transfected with empty plasmid. (C) Semi-quantitative RT-PCR of Ago1 and Ago2 mRNAs using total RNA from Ctr, PolAgo1, PolAgo2 and fromtwo representative monoclones (mAgo1c and mAgo2c).

A

B

Fig. 2. Cell growth analysis. (A) Cell growth curves of mAgo1, mAgo2 and Ctr clonesdetermined by trypan blue assays. Three experiments were done in triplicate onthree different mAgo1, mAgo2 and Ctr monoclones. (B) The same analysis wasperformed on PolAgo1, PolAgo2.

C. Parisi et al. / FEBS Letters 585 (2011) 2965–2971 2967

length of PolAgo1 and PolAgo2 cells increased by �50% whencompared to control cells (Fig. 4A, B). Additionally, during the firstthree days of RA induction, PolAgo1 and PolAgo2 clones maintain asimilar increment of neurite length with respect to control. How-ever, after longer exposures to RA, all clones reach the same differ-entiation status with comparable neurite outgrowth (Fig. 4B, T5time point). Consistently, the neuronal marker bIII-tubulin is sig-nificantly more expressed in PolAgo1 and PolAgo2 clones at day3 of RA induction than in control clones, while at day 5 all clonesshow similar expression levels of this protein. Interestingly, Gap-43 is up-regulated by the overexpression of Ago proteins bothbefore and after RA treatment (Fig. 4C), in line with the differenti-ation phenotype of PolAgo1 and PolAgo2. Both Ago1 and Ago2monoclones behaved similarly to polycones in our differentiationanalyses (data not shown), confirming the eliciting effect of Ago1and Ago2 overexpression on SH-SY5Y neuronal differentiation.

3.3. Ago1 overexpression results in a less invasive phenotype of SH-SY5Y

The growth and differentiation phenotypes observed upon Ago1overexpression are reminiscent of those induced by a tumor sup-pressor factor. Given that a characteristic feature of cancer cells istheir ability to invade surrounding tissues, we next investigatedwhether Ago1 and Ago2 can affect neuroblastoma cells invasiveness.In a wound healing assay, the migration rate of all polyclones was re-corded and reported as the mean value registered for at least 20 cellsfor each clone in three independent experiments. In Fig. 5A, repre-sentative photographs show the healing process at 0 and 18 h afterwounding, a time point chosen because shorter than SH-SY5Y dou-bling time. We observed that PolAgo1 cells cover an area�20% smal-ler than the one healed by control and PolAgo2 cells. Additionally,the average rate of migration of Ago1 cells (150.32 ± 2.27 lm/h) isat least 25% lower than control (200.48 ± 9.72 lm/h) and Ago2 cells(232.55 ± 5.43 lm/h) (Fig. 5B). These data suggest that Ago1 overex-pression, in addition to modulating growth rate and differentiation,also leads to a decrease in the motility of cells.

To further test the metastatic potential of our clones, we per-formed an invasion assay, in which cells are induced to migratethrough matrigel-coated filters. As shown in Fig. 6, Ago1 overex-pression reduces the invasiveness of SH-SY5Y cells by about 50%compared to both control and PolAgo2 cells. Overall these analysesindicate that Ago1 overexpressing cells are characterized by a re-duced metastatic potential, in agreement with the tumor suppres-sor phenotype suggested above.

3.4. Ago1 overexpression results in the accumulation of p53

To gain insight into the molecular mechanisms that mediate thephenotypic effects of Ago1 overexpression in SH-SY5Y cells, weanalysed the expression levels of key regulatory proteins involvedin cell cycle and proliferation. Overexpression of Ago1 leads to asignificant increase in p53 protein levels, which are unchanged inAgo2 overexpressing cells (Fig. 7). Semi-quantitative RT-PCR anal-yses show that p53 mRNA levels are unaffected by the overexpres-sion of either Ago protein (data not shown), indicating that p53protein enrichment in PolAgo1 is not due to an increase in tran-scription or stability of its mRNA, but rather to an increase in sta-bility or translation efficiency of the protein.

A widely used method to assess p53 activation involves anevaluation of the response to genotoxic stimuli, such as UV irradia-tion. Apoptotic nuclei were counted by Hoechst staining 24 h afterUV-exposure and, consistently with increased p53 levels, Ago1-overexpressing cells showed an augmented sensitivity to UV-irradiation, more than doubling the number of apoptotic nucleiper field (Fig. 8A). Conversely, Ago2 overexpressing cells were not

A

B

Fig. 3. Cell cycle analysis. Cell cycle phases distribution of mAgo1, mAgo2, PolAgo1, PolAgo2 and Ctr clones obtained by propidium iodide cell staining and FACS analysis asreported in Section 2. For each clone a representative DNA histogram (A) and the mean percentage of the cell population in each cell cycle phase (B) are shown. n = 4 Formonoclones and n = 2 for polyclones.

2968 C. Parisi et al. / FEBS Letters 585 (2011) 2965–2971

significantly affected by UV irradiation. These findings were furthersupported by an increase of cleaved Caspase-3 levels in PolAgo1 ex-tracts, a characteristic feature of cells undergoing apoptosis (Fig. 8B).

4. Discussion

We have stably overexpressed Ago1 and Ago2 proteins in a SH-SY5Y neuroblastoma cell line to identify novel and specific functionsof these two orthologs. Both Ago1- and Ago2-overexpressing clonesdisplay more prominent neurite outgrowth than their control coun-terparts, even before activation of the differentiation pathway withretinoic acid, suggesting that the over-expression of either protein insufficient per se to induce neural differentiation. Interestingly, it hasbeen previously shown that differentiation by RA does not dependon an increase of endogenous Ago proteins as their abundance isunaffected by RA treatment of a neuroblastoma cell line [9].

Biochemically, our result is paralleled by the expression pattern ofthe differentiation marker Gap43, up-regulated in both PolAgo1and PolAgo2 even before RA induction. The phenotypic and bio-chemical effects exerted by the overexpression of these factors ishowever lost at late time points (5 days) after RA treatment. Wes-tern blots of bIII-tubulin reveal a similar profile; its levels are up-reg-ulated in the context of Ago1 or Ago2 overexpression at 3 days of RAtreatment but merge with control levels at 5 days. These data sug-gest an overlapping function, likely miRNA-mediated, of Ago1 andAgo2 in neuronal differentiation. Redundancy of Ago1 and Ago2function in miRNA-regulated pathways is indicated by profiles ofthe miRNAs associated with each Ago protein and recent PAR-CLIPexperiments [8], showing that these two factors bind the same poolof miRNAs and mRNA targets.

However, when analyzing the proliferation profiles of ourclones, we observed a non-redundant effect of Ago1 versus Ago2;

A

B C

Fig. 4. Differentiation analysis. (A) Representative images of cells before (day 0) and after 3 and 5 days of RA induction. (B) The mean length at least 100 neurites is shown foreach clone at the indicated time points. Scale bar: 100 lm. (C) Representative western blot of bIII-tubulin and Gap-43 expression in Ctr, PolAgo1 and PolAgo2 clones beforeand after RA induction.

A

B

Fig. 5. Motility assay. (A) Representative images of Ctr, PolAgo1 and PolAgo2 clonesat time 0 and 18 h after wounding. (B) Histogram depicting average speed of thedifferent clones from three independent experiments.

C. Parisi et al. / FEBS Letters 585 (2011) 2965–2971 2969

Ago1 overexpression causes the cell cycle to slow down, probablydue to a delay in the G1/S transition. Additionally, high levels ofAgo1 decrease cellular motility, in wound healing and invasion as-says, and induce a stronger apoptotic response upon UV exposure.Thus, Ago1 overexpression rescues four hallmark features of tumorcells, suggesting that Ago1 behaves as a tumor suppressor factorwhen up-regulated. Parallel analyses of Ago2 overexpressingclones reveal no significant effect in any of the phenotypical fea-tures tested, differently from what was observed in human breastcancer cells. In these cells, high levels of Ago2 correlated with anenhanced proliferation and migration ability [12].

This unexpected effect of Ago1 is supported by data from thepublic functional genomics data repository GEO Profiles [13]. Sev-eral microarray analyses of differentiating and tumorigenic cellsreveal an inverse correlation between Ago1 expression levels andcellular division rates. In several tumor cell lines Ago1 levels aresignificantly lower, while Ago2 is unperturbed (GEO Accession:GDS1375/218287 and GDS1375/ 213310 [14]). Accordingly, region1p34–35 of chromosome 1, which includes Ago1 gene, is often de-leted in Wilms’ tumors and has been associated with neuro-ectodermal tumours [10,15]. Ago1’s potential role in modulatingthe cell cycle is also corroborated by its expression profiles inmicroarray analyses of differentiating neurons during brainembryogenesis; being very low when mostly undifferentiatedneural progenitors are present, but growing significantly whenthey begin to differentiate into neurons (GEO Accession:GDS3442/1434331 and GDS3442/1426366 [16]). Additionally, it

A

B

Fig. 6. Cell invasion assay. (A) Representative images of cells migrated though matrigel or control inserts. (B) ‘‘Invasion index’’ is plotted for each clone, representing thenumber of matrigel-invading cells normalized against control insert-migrating cells. n = 5.

Fig. 7. p53 expression levels. Western blot of PolAgo1, PolAgo2, Ctr, mAgo1(c) andmAgo2(c) lysates with p53 and b-actin antibodies, representative of four indepen-dent experiments.

A B

Fig. 8. Apoptosis induction. (A) Histogram representing the percentage of apoptoticnuclei upon UV-irradiation. n = 6. (B) Western blot analysis for Caspase-3 on cloneslysed 24 h after UV irradiation. The antibody recognizes both full length (35 kDa)and cleaved (17 kDa) Caspase 3.

2970 C. Parisi et al. / FEBS Letters 585 (2011) 2965–2971

is noteworthy that Ago1 levels peak in brain tissues, correlatingagain its abundance with cell cycle arrest. These data, togetherwith our results, suggest that Ago1 likely plays a role in down-regulating cell proliferation in multiple cellular contexts, raisingseveral complex questions regarding Ago proteins functionalredundancy and mode of action. It has been reported that thelength of G1 may be a primary modulator of neuronal differentia-tion [17], thus in principle Ago1 over-expression may induce dif-ferentiation as a mere consequence of its effects on the cell cycle.However, it should be noted that Ago2 over-expression has a sim-ilar influence on neuronal differentiation without noticeablyaffecting cell cycle progression. This suggests that, at least in thiscase, the effects observed on cell differentiation are uncoupledfrom those observed on cell proliferation.

Two alternative scenarios can be imagined in interpreting ourdata: Ago1 effects on the cell cycle could be caused by misregula-tion of miRNAs or could be due to a miRNA-independent processthat involves Ago1.

In the miRNA dependent scenario, Ago1 unique effects could beonly explained by a non-redundant function of Ago1 versus Ago2.A possible solution in this direction is offered by differential rolesplayed by Ago proteins in miRNA biogenesis [18,19]. In particular,Ago1 was identified as part of a large multiprotein complex com-mitted to miRNA maturation from pre-miRNA in a neuronal cellline. Moreover, the non-redundant phenotype of Ago proteinsoverexpression may be due to Ago2’s unique endonucleolyticactivity, which may influence both the loading of miRNAs on RISCcomplexes and the degradation of specific target mRNAs.

It may otherwise be that these effects depend on accessory fac-tors specifically associated with Ago1. One such example isTRIM32, a protein that induces neuronal differentiation by bindingAgo1 and increasing the activity of specific miRNAs [20].

In the alternative scenario, the tumor suppressor-like effectsobserved only upon Ago1 overexpression, and not in Ago2 clones,

C. Parisi et al. / FEBS Letters 585 (2011) 2965–2971 2971

could be explained by functions of Ago1 that are independent ofmiRNAs.

Studies in S. Pombe, have shown that Ago1 is required for a nor-mal G1/S transition, cytokinesis and the response to DNA damage[21]. Interestingly, miRNAs are not expressed in S. Pombe andAgo1 associates with factors directly involved in regulating the cellcycle [22].

miRNA-independent functions of Ago1 protein are also sup-ported by Ago1 mutants isolated in Arabidopsis, either defectivein RNAi but functional in development or viceversa, indicating thatcontrol of these two processes by Ago1 can be uncoupled [23].

It should be noted that the effects of Ago1 on the cell cycle atthe G1/S transition are in accordance with both the miRNA-dependent and independent scenarios. Indeed, most cell cycle-targeting miRNAs modulate this specific checkpoint [24]. Likewise,Ago1 mutants are defective in G1 arrest in S. Pombe, where miR-NAs are absent [21].

An indication of the molecular mechanism underlying the phe-notypic alterations of Ago1 overexpressing cells comes from theevidence that p53 levels are strongly upregulated in these clones.

A connection between Ago1 and p53 levels is in accordancewith recent studies linking p53 to the miRNA pathway; p53 is amiRNA target, a transcriptional regulator of miRNAs and is in-volved in miRNA biogenesis [25,26]. However, a pathway in whichp53 is part of the tumor suppressor cascade activated by Ago1 iscompatible with both miRNA-dependent and independentscenarios.

In conclusion, our data provide insights that further the under-standing of the Argonaute family of proteins, suggesting a novelrole for Ago1 in downregulating cell proliferation and motility. Fu-ture work should unravel the molecular cascade that underlies theeffects of Ago1 overexpression and test potential applications incancer treatments.

Acknowledgments

This work was supported by Regione Lazio, Miur Firb RBAP10L8TY-007 to Carlo Cogoni and Research University Project (C26A0935LN) to Caterina Catalanotto.

We are grateful to Silvia Anna Ciafrè and Silvia Galardi for help-ful discussion.

The authors declare no conflict of interest.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.febslet.2011.08.003.

References

[1] Hutvagner, G. and Simard, M.J. (2008) Argonaute proteins: key players in RNAsilencing. Nat. Rev. Mol. Cell. Biol. 9, 22–32.

[2] Pillai, R.S., Artus, C.G. and Filipowicz, W. (2004) Tethering of human Agoproteins to mRNA mimics the miRNA-mediated repression of proteinsynthesis. RNA 10, 1518–1525.

[3] Garzon, R., Calin, G.A. and Croce, C.M. (2009) MicroRNAs in cancer. Annu. Rev.Med. 60, 167–179.

[4] Hebert, S.S. and De Strooper, B. (2009) Alterations of the microRNA networkcause neurodegenerative disease. Trends Neurosci. 32, 199–206.

[5] Lin, R.J. et al. (2010) MicroRNA signature and expression of Dicer and Droshacan predict prognosis and delineate risk groups in neuroblastoma. Cancer Res.70, 7841–7850.

[6] Torres, A., Torres, K., Paszkowski, T., Jodlowska-Jedrych, B., Radomanski, T.,Ksiazek, A. and Maciejewski, R. (2011) Major regulators of microRNAsbiogenesis Dicer and Drosha are down-regulated in endometrial cancer.Tumour Biol. 32, 769–776.

[7] Landthaler, M. et al. (2008) Molecular characterization of human Argonaute-containing ribonucleoprotein complexes and their bound target mRNAs. RNA14, 2580–2596.

[8] Hafner, M. et al. (2010) Transcriptome-wide identification of RNA-bindingprotein and microRNA target sites by PAR-CLIP. Cell 141, 129–141.

[9] Potenza, N., Papa, U. and Russo, A. (2009) Differential expression of Dicer andArgonaute genes during the differentiation of human neuroblastoma cells.Cell. Biol. Int. 33, 734–738.

[10] Koesters, R. et al. (1999) Human eukaryotic initiation factor EIF2C1 gene:cDNA sequence, genomic organization, localization to chromosomal bands1p34–p35, and expression. Genomics 61, 210–218.

[11] Reynolds, C.P. (2000) Differentiating agents in pediatric malignancies:retinoids in neuroblastoma. Curr. Oncol. Rep. 2, 511–518.

[12] Adams, B.D., Claffey, K.P. and White, B.A. (2009) Argonaute-2 expression isregulated by epidermal growth factor receptor and mitogen-activated proteinkinase signaling and correlates with a transformed phenotype in breast cancercells. Endocrinology 150, 14–23.

[13] Barrett, T. et al. (2010) NCBI GEO: archive for functional genomics data sets –10 years on. Nucleic Acids Res. 39, D1005–1010.

[14] Talantov, D., Mazumder, A., Yu, J.X., Briggs, T., Jiang, Y., Backus, J., Atkins, D. andWang, Y. (2005) Novel genes associated with malignant melanoma but notbenign melanocytic lesions. Clin. Cancer Res. 11, 7234–7242.

[15] Dome, J.S. and Coppes, M.J. (2002) Recent advances in Wilms tumor genetics.Curr. Opin. Pediatr. 14, 5–11.

[16] Hartl, D. et al. (2008) Transcriptome and proteome analysis of early embryonicmouse brain development. Proteomics 8, 1257–1265.

[17] Salomoni, P. and Calegari, F. (2010) Cell cycle control of mammalian neuralstem cells: putting a speed limit on G1. Trends Cell Biol. 20, 233–243.

[18] Cheloufi, S., Dos Santos, C.O., Chong, M.M. and Hannon, G.J. (2010) A dicer-independent miRNA biogenesis pathway that requires Ago catalysis. Nature465, 584–589.

[19] Wulczyn, F.G. et al. (2007) Post-transcriptional regulation of the let-7microRNA during neural cell specification. Faseb J. 21, 415–426.

[20] Schwamborn, J.C., Berezikov, E. and Knoblich, J.A. (2009) The TRIM-NHLprotein TRIM32 activates microRNAs and prevents self-renewal in mouseneural progenitors. Cell 136, 913–925.

[21] Carmichael, J.B., Provost, P., Ekwall, K. and Hobman, T.C. (2004) Ago1 and dcr1,two core components of the RNA interference pathway, functionally divergefrom rdp1 in regulating cell cycle events in Schizosaccharomyces pombe. Mol.Biol. Cell 15, 1425–1435.

[22] Stoica, C., Carmichael, J.B., Parker, H., Pare, J. and Hobman, T.C. (2006)Interactions between the RNA interference effector protein Ago1 and 14-3-3proteins: consequences for cell cycle progression. J. Biol. Chem. 281, 37646–37651.

[23] Morel, J.B., Godon, C., Mourrain, P., Beclin, C., Boutet, S., Feuerbach, F., Proux, F.and Vaucheret, H. (2002) Fertile hypomorphic Argonaute (ago1) mutantsimpaired in post-transcriptional gene silencing and virus resistance. Plant Cell14, 629–639.

[24] Bueno, M.J. and Malumbres, M. (2011) MicroRNAs and the cell cycle. Biochim.Biophys. Acta 1812, 592–601.

[25] Suzuki, H.I. and Miyazono, K. (2010) Dynamics of microRNA biogenesis:crosstalk between p53 network and microRNA processing pathway. J. Mol.Med. 88, 1085–1094.

[26] Boominathan, L. (2010) The tumor suppressors p53, p63, and p73 areregulators of microRNA processing complex. PLoS One 5, e10615.