LMNA Knock-Down Affects Differentiation and Progression of Human Neuroblastoma Cells

Background Neuroblastoma (NB) is one of the most aggressive tumors that occur in childhood. Although genes, such as MYCN, have been shown to be involved in the aggressiveness of the disease, the identification of new biological markers is still desirable. The induction of differentiation is one of the strategies used in the treatment of neuroblastoma. A-type lamins are components of the nuclear lamina and are involved in differentiation. We studied the role of Lamin A/C in the differentiation and progression of neuroblastoma. Methodology/Principal Findings Knock-down of Lamin A/C (LMNA-KD) in neuroblastoma cells blocked retinoic acid-induced differentiation, preventing neurites outgrowth and the expression of neural markers. The genome-wide gene-expression profile and the proteomic analysis of LMNA-KD cells confirmed the inhibition of differentiation and demonstrated an increase of aggressiveness-related genes and molecules resulting in augmented migration/invasion, and increasing the drug resistance of the cells. The more aggressive phenotype acquired by LMNA-KD cells was also maintained in vivo after injection into nude mice. A preliminary immunohistochemistry analysis of Lamin A/C expression in nine primary stages human NB indicated that this protein is poorly expressed in most of these cases. Conclusions/Significance We demonstrated for the first time in neuroblastoma cells that Lamin A/C plays a central role in the differentiation, and that the loss of this protein gave rise to a more aggressive tumor phenotype.


Introduction
Neuroblastoma is an embryonic tumor of the sympathetic nervous system and is one of the most common cancers in childhood. The clinical presentation of neuroblastoma is extremely variable, ranging from benign to highly aggressive tumors [1]. MYCN amplification, which occurs in about 25% primary tumors, is considered to be the most important oncogenic marker in neuroblastoma, correlating with unfavorable prognosis [2]. Additionally, a high differentiation stage has been associated with a favorable outcome [3]. Nevertheless, the mechanisms governing neuroblastoma cell differentiation are not completely understood.
The ability of many neuroblastoma cell lines to differentiate and to present unique markers of neuronal differentiation in the presence of various agents has led to the use of these cell lines as model systems to study human neuronal differentiation [4,5].
Lamins, type V intermediate filaments, are the major components of the nuclear lamina. They are divided into A and B types based on similarities in their primary sequence and biological properties. In mammals, two major A-type (Lamin A and C) and two major B-type (Lamin B1 and B2) lamins have been characterized. The most important and well-established function of nuclear lamins is to provide shape and mechanical stability to the nucleus [6]. Furthermore, several lines of evidence have supported a role for the lamins in maintaining the mechanical properties of the entire cell. Indeed, Lamin A/C deficiency causes impaired mechanotransduction and decreased mechanical stiffness [7].
Recent models of nuclear architecture describe lamins as determining factor for chromosome positioning throughout the nucleus [8,9], anchoring chromatin to the nuclear lamina and functioning as a nucleoplasmic scaffold for nuclear chromatin. Indeed, many aspects of nuclear activity, such as DNA replication and transcription, are affected by modifications of the nuclear lamina [10,11].
The individual types of lamins are differentially expressed during development [12][13][14]. A role of these proteins in the differentiation processes has been demonstrated in the muscle and adipocyte differentiation [15,16] Additionally, the expression of the A-type lamins is often reduced or absent in subsets of cells with a low degree of differentiation and/or cells that are highly proliferative, including various human malignancies [17][18][19][20][21], even though some authors have reported in colorectal tumors that Atype lamins may play a role as risk biomarker [22].
The main objective of this work was to investigate whether Lamin A/C is involved in neuroblastoma differentiation. Moreover, we studied a possible role of Lamin A/C in the progression of this neuronal cancer.

LMNA Gene Knock-down Inhibits Retinoic Acid (RA)induced Differentiation in Neuroblastoma Cells
We chose a neuroblastoma model that can differentiate in vitro and that has high expression level of Lamin A/C. As a differentiating stimulus, we used RA. Exposure of SH-SY5Y cells to RA increased LMNA gene expression ( Figure 1A).
We silenced the LMNA gene in this cell line and studied the effects on the differentiation processes. LMNA knock-down (KD) was obtained by infecting the cells with a lentiviral vector that allows the simultaneous expression of the EmGFP reporter gene and of an artificial miRNA targeting the LMNA mRNA. The LMNA-KD cells showed a reduction in the Lamin A/C protein of approximately 70% compared to Mock cells, infected with the same lentiviral vector carrying a non-targeting artificial EmGFP-miRNA ( Figure 1B). The reduction of Lamin A/C expression resulted in a cell population exhibiting a flatter shape without significant neurite outgrowths, indicating an impairment of neuronal differentiation ( Figure 1C and D). Consistent with the impairment of neuritis outgrowth is the increase of DNA synthesis as evaluated by BrdU incorporation during the differentiation process in LMNA-KD compared to Mock cells ( Figure 1E). Additionally, changes in some cell cycle related molecules were also evidenced by an increase of cyclins A and D1, and a decrease of p27 kip1 in LMNA-KD compared to Mock cells ( Figure 1F).
Consistent with the impairment in neurite formation in LMNA-KD cells, we observed decreased induction of the primary neuronal differentiation markers ( Figure 2). Figure 2A shows fluorescent images of the expression of neurofilament-200 (NF200) in LMNA-KD compared to Mock cells. The inhibition of the NF200 expression is evident in LMNA-KD cells. The impairment of differentiation in LMNA-KD cells was further confirmed by the loss of induction of some neuronal differentiation markers such as tyrosine hydroxylase (TH), dopamine-b-hydroxylase (DBH) and enolase 2 (ENO2), as evaluated by qRT-PCR ( Figure 2B) and Western blot (TH and DBH, Figure 2C).
We have also studied the effect of LMNA knock-down in other two neuroblastoma cell lines, IMR-32 and SMS-KCNR. We chose these two cell lines because of their different expression of Lamin A/C protein ( Figure 3A). We silenced LMNA gene and the level of gene knock-down for the two cell lines is shown in Figure 3B. We demonstrated that also in these cell lines silencing of LMNA gene resulted in the inhibition of cell differentiation as evidenced by the loss of induction of the indicated markers after 3 and 5 days of differentiation ( Figure 3C).
Gene Expression Profiling of LMNA-KD Cells Confirms an Impairment of Differentiation and Evidences the Establishment of a More Aggressive Phenotype We compared the gene expression profile of Mock and LMNA-KD cells, both in RA-treated and untreated cells, using Agilent microarray technology. We verified that the LMNA expression was induced by RA in Mock cells (approximately 1.6 folds), while the silencing effect in LMNA-KD cells was 0.2 fold. We identified a set of 1320 genes differentially expressed at least 2.0-fold by RA in both Mock and LMNA-KD cells. We defined two sets of genes that were obtained by comparing the RA-treated samples with the untreated samples in the Mock (Mock RAvsNT ), or LMNA-KD (LMNA-KD RAvsNT ) cells. The Venn diagram in Figure S1  The functional analysis of the differentially expressed genes, performed using the DAVID web tool, evidenced significant modulation (p,0.05) of functional categories in the Mock cells which includes categories related to neuron differentiation among the up-regulated genes and to cell cycle among the down-regulated genes (Table S1). Similarly, the categories modulated by RA only in LMNA-KD cells include categories related to cell migration for the up-regulated genes and to neuron differentiation for the downregulated genes (Table S2).
We examined whether the impairment of cell differentiation that occurs in LMNA-KD cells is associated with a more aggressive phenotype. We defined a list of genes associated with specific cancer features using the Cancer Gene Index database of the NCI. Analyzing the LMNA-KD versus Mock cells whole gene expression profiles using the gene annotations of Cancer Gene Index, we identified 14 genes (eight up-regulated and six downregulated) of which differential expression is related to a more aggressive phenotype. Among these, we found some tumor suppressor genes (e.g. H19, ACTN4) that were down-regulated and some genes related to tumor progression and metastasis formation (e.g. ENPP2, GRIA3, TPM3) that were up-regulated ( Figure 4A and Table S3). The expression levels of the four most modulated genes (up-or down-regulated) were confirmed by qRT-PCR ( Figure 4B). In addition, ''cancer'' was the top category found, even when the list of differentially modulated genes between Mock and LMNA-KD cells were analyzed by Ingenuity Pathway Analysis (IPA) (368 genes, 5.38610 243 ,p,9.63610 26 ). In this analysis (Table S4)   To validate also at protein level the results obtained in the microarray experiments, a comparative proteome profiling of Mock and LMNA-KD cells was performed. A total of 442 proteins were qualitatively identified across both conditions. Only proteins identified in at least two out of three injections with a fold-change greater than 1.3 were considered in the subsequent analysis. Applying these filtering criteria, a total of 68 significant, differentially expressed proteins were selected (Table S5 and S6). In order to highlight the possible key candidates responsible for the different cellular properties of Mock and LMNA-KD cells, we performed an unsupervised bioinformatic analysis using the proteomic dataset of modulated proteins. The complete list of modulated proteins was loaded into IPA software and analysis was performed without any pre-established criteria of inclusion or exclusion. Data investigation revealed a muddled association of interactions among identified proteins and their direct or indirect interactors.
The resulting network with the highest score (67) is shown in Figure 5 (See Table S7 for proteins specifications). Among the most modulated proteins, we noted the up-regulation of the heterogeneous nuclear ribonucleoprotein L (HNRPNL) and the down-regulation of the alpha-actinin 4 (ACTN4). The expression changes of these two proteins correlate with a similar modulation at the RNA level (LMNA-KD/Mock HNRPNL expression ratio, 1.4; ACTN4 expression ratio, 0.4).
Additionally, members of 14-3-3 scaffold proteins family such as b, c and f were found to be slightly increased in LMNA-silenced cells. The changes in expression of these proteins correlate with a similar modulation at the RNA level (LMNA-KD/Mock YWHA gene expression ratio, 1.4).

The Knock-down of LMNA Increases Cell Motility and Invasion in vitro, and the Tumorigenicity in vivo
We examined the motility and the invasion ability of Mock and LMNA-KD cells in response to the conditioned medium of NIH3T3 cells ( Figure 6A). The migratory ability was enhanced in LMNA-KD (approximately 1.5-fold) compared to Mock cells. Similarly, the ability to invade matrigel-coated filters was enhanced in LMNA-KD (approximately 1.7-fold).
We then examined how LMNA silencing could affect the expression and activity of matrix metalloproteases (MMPs). A zymographic analysis of the conditioned medium from Mock and LMNA-KD cells showed an increase in the secretion of the gelatinase 72-kDa MMP2 after LMNA silencing ( Figure 6B) compared to Mock cells. A western blot analysis of MMP2 expression in the same samples, also demonstrated an increased expression of MMP2 protein by approximately 1.6-fold in LMNA silenced compared to Mock cells ( Figure 6C). Consistent with these results is the increased expression of MMP2 protein in the lysates of the Mock and LMNA-KD cells from the same experiment ( Figure S2). The activity of MMP9 was undetectable in our experimental conditions. Because cell invasion involves cytoskeleton rearrangements, we studied the intracellular organization of actin filaments by staining actin with phalloidin ( Figure 6D). Knock-down of LMNA increased the number and tickness of cytoplasmic actin-containing fibers. Noteworthy, this effect was particularly evident in the LMNA-KD cells showing the lowest staining for Lamin A/C.
To evaluate whether LMNA silencing can affect tumorigenesis, LMNA-KD and Mock cells were injected in nude mice and tumor weight was monitored. LMNA-KD tumors grew significantly faster than Mock tumors, the volume of LMNA-KD tumors being about 4-fold increased when compared to Mock tumors at day 29 after implantation (p,0.001), thus demonstrating an increased aggressiveness after reduction of LMNA expression ( Figure 6E). The expression levels of the most modulated genes related to tumor aggressiveness, previously identified in the microarray analysis, were evaluated in Mock and LMNA-KD xenografts, showing that the expression levels of these aggressiveness-related genes were similar to those obtained in the in vitro analyses ( Figure 6F).
We analyzed the effect of the knock-down of LMNA also in the IMR-32 and SMS-KCNR neuroblastoma cell lines. In Fig. 7A and C are reported the changes of the expression levels (up-or down-regulated) of the genes which were related to a more aggressive phenotype in the SH-SY5Y cells, evaluated by qRT-PCR in the IMR-32 and SMS-KCNR, Mock and LMNA-KD cells. We found a similar behavior of the indicated genes also in these neuroblastoma cell lines compared to the SH-SY5Y cells. EDN1 gene was found not expressed in the SMS-KCNR cell line. In Fig. 7B is shown that the IMR-32 LMNA-KD cells have an enhanced ability to invade matrigel-coated filters (approximately 1.6-fold).

In vivo Expression of Lamin A/C in Early Stage NB
Nine primary, therapy-free, early stage (I and II) neuroblastomas (NB) were obtained upon informed parents' or guardian's consent [23]. Tumours were classified according to the International Neuroblastoma Pathology Classification (INPC) [24] Lamin A/C specific nuclear stain was observed in scattered interstitial cells and in some vessels in eight out of the nine tumours analyzed. On the contrary neuroblastoma cells displayed very rarely nuclear stain. Only in two cases, namely a stage I and a stage II lesion, scattered areas displayed nuclear stain of the tumor cells.

Reduced Lamin A/C Expression is Associated with Increased Drug Resistance
We studied the effects of different doses of etoposide (VP-16), doxorubicin (DOXO) and paclitaxel (PTX) on the cell survival of the Mock and LMNA-KD cells. The silencing of LMNA resulted in a significant increase in cell survival for all of the compounds employed ( Figure 8A). Because the three drugs are all substrates of the P-glycoprotein, we also studied whether LMNA knock-down affects the expression and/or function of this membrane pump. Indeed, the expression of the P-gp protein was increased by approximately 3.4-fold in LMNA-KD compared to Mock cells ( Figure 8B). In addition, the activity of the pump was analyzed by using a functional assay which employs as probe the fluorescent dye precursor calcein-acetoxymethylester (calcein AM). This precursor is an excellent substrate of the P-gp transporter and therefore is suitable as model to determine in vitro drug resistance   of cells, to screen transporter substrates, and to monitor the P-gp mediated efflux of drugs which are substrate of the pump. As shown in Fig. 8C the P-gp activity was decreased by approximately 50% in LMNA-KD compared to Mock cells. The P-gp pump inhibitor Verapamil was used as control to completely block the activity of the pump in both conditions.

Discussion
Our results demonstrate for the first time in neuroblastoma cells that Lamin A/C expression was necessary for differentiation and the loss of this protein gave rise to a more aggressive tumor phenotype.
In agreement with studies on the development of mammals as well as Xenopus and Drosophila, which showed that the nuclear lamina composition correlates with cell differentiation [14,[25][26][27], we showed that Lamin A/C expression increases in three different neuroblastoma cell lines during the differentiation program induced by retinoic acid.
Silencing the LMNA gene inhibited the differentiation program as first evidenced by the reduced ability of the cells to form mature neurite-networks and to increase their proliferative potential. The impairment of the differentiation program was further confirmed by the decreased levels of several known differentiation markers such as the proteins NF200, TH, DBH, and the genes ENO2 and MAPT. The genome-wide gene expression profile data also confirmed the inhibition of cell differentiation observed in LMNA-silenced cells. Noteworthy, the number of RA induced genes is considerably lower in LMNA-KD compared to the Mock cells. Indeed, the cell cycle-related genes down-regulated in Mock cells to favor differentiation, were not modulated in LMNA-KD cells. Notably, the signaling pathway engaged by RA is still functional in these cells since the expression of the RA receptors as well as of the retinoid-binding proteins [28], is not changed in LMNA-KD cells (data not shown; see the microarray data deposited in the Gene Expression Omnibus (GEO) repository, accession GSE30677).  (Table S7): the coloured features indicate proteins identified (green = down-regulated and red = up-regulated in LMNA-KD compared to Mock), whereas the un-coloured features indicate additional members of this network that were not detected by the proteomic analysis. The node shapes indicate function: enzymes (diamonds), transcription regulators (ovals), nuclear receptors (rectangles), cytokines (squares), transporter (trapezoids), and ''other'' (circles). Protein-protein associations are indicated by edges containing single lines, whereas proteins acting upon another protein (controlling their expression) are indicated by arrows. Continuous or dotted lines indicate direct or indirect protein interactions, respectively. doi:10.1371/journal.pone.0045513.g005 Analyzing the subset of LMNA-KD genes that is the most diagnostic of the differences between untreated and RA-treated samples we demonstrated an increased expression of genes related to tumor aggressiveness and not to differentiation, as would have been expected given the differentiation stimulus. Indeed, focusing on the samples not induced to differentiate, we clearly demonstrate that Lamin A/C contributes to the maintenance of normal cellular processes, highlighting the up-regulation of genes related to cancer progression and the down-regulation of some tumor suppressor genes in the LMNA-KD cells. Specifically, we observed the downregulation of the ACTN4 gene, which was also confirmed by the proteomic analysis. Even though the roles of ACTN4 in tumors reported in the literature are contradictory [29,30] our data are in agreement with those of authors who demonstrated the role of ACTN4 gene in suppressing tumorigenicity in different tumor types [31][32][33]. Additionally, our data, showing the increase of some members of the 14-3-3 protein family, are in agreement with those of other authors, who demonstrated a positive role of such proteins in the induction of oncogenic transformation and in promoting cancer cell survival [34,35].
We demonstrated that alterations in the expression of genes involved in tumor aggressiveness resulted in the augmented ability of the cells to migrate increasing their proteolytic activity. Cell migration is a dynamic process that involves cyclical cytoskeletal rearrangements, and the formation of actin stress fibers could be responsible for changing the physical location of the cell and increasing its migratory ability [36]. Indeed, an increased presence of actin filament bundles (stress fibers), particularly at the periphery of the cell, was evident. Our data are in contrast with those of Lee et al. obtained in mouse embryonic fibroblasts showing that Lamin A/C deficiency reduces the speed of cell migration [37]. However, the assay and the model used to analyze cell migration is evidently different from that we used in this work. The more aggressive phenotype of the LMNA-KD cells was also demonstrated in vivo since tumors grown from injected LMNA-KD cells were significantly more tumorigenic than their control counterparts and exhibited the same modifications in gene expression previously observed in vitro.
The increased expression of the HNRNPL protein is consistent with an impairment of the nuclear pore complex activity (data not shown). HNRNPs are shuttling proteins that participate in the export of mature mRNA from the nucleus to the cytoplasm [38]. It is possible that the increase in the HNRNPL protein represents an attempt of the cells to compensate for the nucleo-cytoplasm transport altered by the loss of Lamin A/C. A-type lamins have also been reported to function as nucleoplasmic scaffolds anchoring heterochromatin to the nuclear lamina, possibly intervening in the epigenetic regulation of gene transcription [8,9,39,40]. It is possible that the reduced expression of Lamin A/ C disrupts the peripheral organization of the transcriptionally silent chromatin thus allowing some of the genes previously repressed to be transcribed. Indeed, Shevelyov et al. [41] demonstrated that the transcriptional activation of the laminabound testis-specific gene cluster in the male germ line is coupled with its translocation away from the nuclear envelope, thus corroborating our hypothesis.
We demonstrated increased expression levels and activity of the P-gp as indicated by FACS analyses in the LMNA-KD cells. This could be responsible for increased resistance of the LMNA-KD cells toward substrates of the efflux pump P-gp. However, no modification in the expression of genes coding for P-gp was observed, strongly indicating that a post-transcriptional event regulated the expression of this protein as recently reported by others [42].
Importantly, our preliminary data on the expression of Lamin A/C in NB tumor samples show that this protein is poorly expressed in most of the first stages cases analyzed, suggesting that an accurate analysis of Lamin A/C expression in early stage NB could provide important information on the tumor development.
In conclusion, our data demonstrate that Lamin A/C is required for the differentiation program of neuroblastoma cells. In addition, our data strongly suggest that low levels of Lamin A/ C could favor a more aggressive tumor phenotype. The characterization of neuroblastoma tumors based on A-type lamin expression may help to develop rational therapeutic protocols based on the molecular profile of the individual tumor.

Ethics Statement
All procedures involving animals and their care were authorized and certified by the decree n. 67/97A of the Italian Minister of Health and protocol 2560/97 of the Rome Health Service Unit (ASL -RMB).

Cell Line Maintenance, Differentiation and Treatment
The three human malignant neuroblastoma (SH-SY5Y, IMR-32 and SMS-KCNR) cell lines were purchased from the American Type Culture Collection (ATCC) in November 2007 and the cells maintained the characteristics of neuroblastoma and the ability to differentiate in vitro. SH-SY5Y and IMR-32 cells were grown in a 1:1 mixture of Eagle's Minimum Essential Medium and F12 medium (Gibco) or Eagle's Minimum Essential Medium, respectively, supplemented with 0.5% non-essential amino acids, 0.5% sodium pyruvate. SMS-KCNR cells were grown in RPMI 1640. All the mediums were supplemented with 10% fetal bovine serum (FBS, Hyclone), 2 mM L-glutamine and 1% penicillin and streptomycin and the cells maintained in a fully-humidified incubator containing 5% CO 2 at 37uC. For all the differentiation experiments, SH-SY5Y and SMS-KCNR cells were seeded at a density of 5610 3 cells/cm 2 . The following day cells were induced to differentiate by 10 mM all-trans retinoic acid (ATRA, named RA throughout the paper; Sigma Chemical) in a ''differentiation medium'' (DM, composed of 50% fresh and 50% conditioned culture medium). For the differentiation experiments, IMR-32 cells were seeded at a density of 1.5610 3 cells/cm 2 , the following day cells were induced to differentiate by 10 mM RA (Sigma Chemical) in a DM composed of medium plus 1% FBS. The cells were fed with DM containing fresh RA after 3 days. RA was dissolved in dimethyl sulfoxide (DMSO, Sigma Chemical). The concentration of DMSO in each experiment was always #0.01%, which was not toxic and did not induce differentiation. For the drug experiments, the SH-SY5Y cells were seeded at a density of 5610 3 cells/cm 2 . Twenty-four hours after plating, fresh medium was added, and after another 24 hours (48 hours of growth), the cells were treated for 2 hours with different doses of etoposide (VP- 16) or paclitaxel (PTX), both dissolved in DMSO, or doxorubicin (DOXO), dissolved in water.

Generation of Lentiviral Vectors
The LMNA-knockdown SH-SY5Y, SMS-KCNR and IMR-32 cell lines were generated by using a lentiviral expression vector containing a double-strand oligo encoding a pre-miRNA sequence as the ''knockdown cassette'', thus enabling the expression of an engineered miRNA sequence directed against the LMNA mRNA. The lentiviral expression vector was produced using the BLOCK-iT Lentiviral Pol II miR RNAi Expression System (Invitrogen) to obtain a GatewayH-adapted lentiviral destination vector named pLenti6/V5-GW/EmGFP-miR-LMNA from pcDNA6.2-GW/ EmGFP-miR-LMNA (Invitrogen), according to the manufacturer's instructions. As a negative control, we generated the pLenti6/V5-GW/EmGFP-miR-Neg control plasmid from the pcDNA TM 6.2-GW/EmGFP-miR-Neg control plasmid (Invitrogen), which contains an insert that can form a hairpin structure that is processed into mature miRNA but is not predicted to target any known vertebrate gene.

Generation of Recombinant Viruses and Lentiviral Infection of Cells
To produce lentiviral stocks, the lentiviral vector (3 mg) and the ViraPower TM Packaging Mix (9 mg) were cotransfected into 6610 6 packaging 293FT cells using Lipofectamine 2000 (Invitrogen), according to the manufacturer's instructions. The supernatants were collected after 48 and 72 hours, and were used for infection. The cells were seeded into 6-well plates at a concentration of 2610 5 cells/well. The following day, the cells were infected for 18 hours with the virus-containing supernatant from pLenti6/ V5-GW/EmGFP-miR-LMNA or pLenti6/V5-GW/EmGFP-miR-Neg, and then fresh medium was added. After 6 hours, the cells were subjected to a second round of infection for an additional 18 hours. Stably transduced cells were selected for by culturing the cells in the presence of blasticidin at 3 mg/mL (Invitrogen). The transduced cells were screened using western blotting and realtime RT-PCR assays to determine the levels of LMNA expression and the miRNA expression was monitored by checking the simultaneous coexpression of the EmGFP reporter gene by fluorescence microscopy.

Animal Experiments
Female CD-1 nude (nu/nu) mice were purchased from Charles River Laboratories (Calco). All procedures involving animals and their care were authorized and certified by the decree n. 67/97A of the Italian Minister of Health and protocol 2560/97 of the Rome Health Service Unit (ASL -RMB). Each experimental group included 8 mice. Cells in the exponential growth phase were harvested from the culture, washed with medium and resuspended in Matrigel (2.5 mg/ml; BD Biosciences) and 10 7 cells implanted subcutaneously in the dorsal region of the mice. The tumor weight was calculated as previously reported [43]. Two different experiments were performed.  Table S3 for specification).

Tissues Immunohistochemistry
Nine primary, therapy-free, early stage (I and II) neuroblastomas (NB) were obtained upon informed parents' or guardian's consent [23]. Tumours were classified according to the International Neuroblastoma Pathology Classification (INPC) [24]. Upon collection specimens were snap frozen in liquid nitrogen. Cryostat sections (4 mm thick) were fixed in cold absolute acetone and stained by indirect immunoperoxidase method using the murine monoclonal antibody (clone JOL2) to lamin A/C (Chemicon International) according to the manufacturer's instruction. Staining was revealed by a supersensitive immunohistochemistry kit as described [44]. Positive and negative controls of the stain were acetone fixed cytospins of Mock and LMNA-KD SH-SY5Y cells. While in the former over 40% of the cells displayed a clear nuclear stain, in the latter only isolated nuclear stain was detectable.

Whole Genome Expression Profiling
The microarray data are MIAME compliant and have been deposited in the Gene Expression Omnibus (GEO) repository (Accession number GSE30677).         Figure 4.

Supporting Information
(DOC) Materials and Methods S1 Supporting materials and methods and supporting references. (DOC)