Molecular Brain Research 115 (2003) 93–103www.elsevier.com/ locate/molbrainres

Research report

T ranscription regulator LMO4 interferes with neuritogenesis in humanSH-SY5Y neuroblastoma cells

a a a aDung Vu , Pascale Marin , Claude Walzer , Maria Margherita Cathieni , Estellea a b a a ,*¨ `N. Bianchi , Farouk Saıdji , Genevieve Leuba , Constantin Bouras , Armand Savioz

aDepartment of Psychiatry, University of Geneva School of Medicine, CH-1225 Geneva, SwitzerlandbUniversity Psychogeriatrics Hospital, University of Lausanne, CH-1008 Lausanne, Switzerland

Accepted 11 March 2003

Abstract

LMO4 is a transcription regulator interacting with proteins involved, among else, in tumorigenesis. Its function in the nervous system,and particularly in the adult nervous system, has however still to be elucidated. We decided to modify its expression in a neuronal model,human SH-SY5Y neuroblastoma cells, by permanent transfection of sense or anti-senseLmo4 cDNAs. Generated clones overexpressingthe Lmo4 transcript in sense orientation tended to aggregate. They showed significantly reduced average number of neurites per cell andaverage neuritic length per cell. The opposite was observed with clones overexpressing the anti-senseLmo4 transcript. Furthermore,selected clones were subjected to 72 h long-term treatments with retinoic acid and phorbol ester (TPA), two biochemicals known tostimulate differentiation of non-transfected SH-SY5Y cells and other neuroblastoma cells. Neuritogenesis occurred after retinoic acidstimulation in all cases. The inhibitory effect of senseLmo4 RNA overexpression on neuritic outgrowth was indeed prevented. Theprotein kinase C activator TPA could not induce neuritogenesis in SH-SY5Y cells overexpressing senseLmo4 RNA. Thus, senseLmo4RNA overexpression, notLmo4 endogenous transcription, overrides the stimulatory effect of TPA upon neuritic outgrowth. We alsoshowed thatLmo4-dependent neuritic retraction and outgrowth correspond to altered phosphorylation of cytoskeletal proteins. Overall,Lmo4 RNA overexpression interferes with neuritic outgrowth, whereas anti-senseLmo4 RNA expression favors neuritogenesis inSH-SY5Y cells. Consequently, changes inLmo4 RNA expression levels might alter the rate of neuritic outgrowth in the developing andadult nervous system. 2003 Elsevier B.V. All rights reserved.

Theme: Cellular and molecular biology

Topic: Gene structure and function: general

Keywords: LMO4; Neurofilament; Phorbol ester; Retinoic acid; SH-SY5Y cells; Tau

1 . Introduction groups [5,9,20,29]. LMO proteins are built up almostexclusively by two tandem LIM domains, themselves

LMO4 belongs to the LMO (‘LIM-only’) protein fami- formed by 2 zinc-finger motifs. These domains allowly. The corresponding gene has been identified by several protein–protein interactions[19,33,36] and can be found

not only alone, as in LMO proteins, but also in associationwith other domains, such as e.g., homeodomains[3,8].Abbreviations: bp, base pair; FCS, fetal calf serum; mRNA, messengerLMO proteins act via other proteins as transcriptionribonucleic acid; NF, neurofilament; PBS, phosphate-buffered saline;

PCR, polymerase chain reaction; PKC, protein kinase C; RA, all-trans- regulators[36].retinoic acid; RT–PCR, reverse transcriptase–polymerase chain reaction; HumanLmo4 is located on chromosome 1p22.3[31]. ItTBE, Tris–borate–EDTA electrophoresis buffer; TPA, 12-O-tetrade- codes for a 165 amino acid protein[20] that shares onlycanoyl phorbol 13-acetate (phorbol ester); U, unit

about 50% amino acid sequence within the LIM domain*Corresponding author. Tel.:141-22-305-5310; fax:141-22-305-regions of LMO1, 2 and 3. LMO4 appears to be the most5350.

E-mail address: armand.savioz@hcuge.ch(A. Savioz). distant relative in this protein family[9]. Clim-1 (or Ldb-

0169-328X/03/$ – see front matter 2003 Elsevier B.V. All rights reserved.doi:10.1016/S0169-328X(03)00119-0

94 D. Vu et al. / Molecular Brain Research 115 (2003) 93–103

2) and Clim-2 (Ldb-1 or NLI) are interacting proteins that tomycin, Trizol, DNA standard molecular weights,might retain LMO4 in the cell nucleus[9]. The protein Neurobasal� medium and B27 supplement were pur-complex LMO4/Clim-2 has been shown to interact with chased from Life Technologies/ Invitrogen. Heat-inacti-the DNA binding protein mDEAF-1 (mouse deformed vated FCS was from Oxoid. MP agarose,Pwo DNAepidermal autoregulatory factor-1) in mice. This observa- polymerase, restriction enzymes (BamHI, EcoRI, PvuI),tion led to the speculation that LMO4 might be implicated shrimp alkaline phosphatase, T4 ligase, Pefabloc SC andin the transcriptional regulation of proenkephalines[29]. Complete were provided by Roche, and RA, TPA as wellRecently, Sum et al.[30] found that LMO4 interacts with as endonuclease (Benzonase) were from Sigma. SH-SY5YBRCA1 (breast cancer 1), a tumor suppressor gene and neuroblastoma cells at passage number 12 were purchasedthat it represses the transcriptional activity of BRCA1 in from the European Collection of Cell Culture (ECACC,breast tissue. Moreover, LMO4 is often overexpressed in Salisbury, UK). Kits were from Invitrogen, Qiagen orprimary invasive breast carcinomas[34] as well as primary AMS Ambion, and are specified whenever used. Mono-prostate cancer[12]. Thus, LMO4 appears to be an clonal mouse Tau-1 (Chemicon) recognizes tau when alloncogene protein as are LMO1 and 2[1,17–19,23,24]. serines at positions 195, 198, 199 and 202 (numberedFinally, Lmo4 transcription was increased 3 h after according to the longest tau isoform of 441 amino acids)carbachol stimulation of human embryonic kidney cells are dephosphorylated[37]. SMI 31 is a mouse monoclonal293 stably transfected with the M1 muscarinic acetyl- IgG1 that detects phosphorylated neurofilaments NF-H andcholine receptor[35]. In the human,Lmo4 shows a broad NF-M (Steinberger Monoclonals, Lutherville, USA).spectrum of expression[20]. However, the roles of LMO4 Monoclonal mouse anti-neurofilament NF-L 68 reacts(e.g.,[7]) in the human brain, and particularly in the adult specifically with the 68 kDa isoform (Clone NR4, Sigma).brain, have still to be investigated. Secondary sheep anti-mouse antibody was purchased from

Thus, to elucidate functions of LMO4 in the nervous Amersham Biosciences and the BCA protein assay kitsystem, we decided to modify its expression in human from Pierce. Human NFs purified according to Tokutake etSH-SY5Y neuroblastoma cells by permanent transfection al.[32] are a gift from Dr B.M. Riederer (Institute ofof sense or anti-senseLmo4 cDNAs. Indeed, the SH-SY5Y Cellular Biology and Morphology, Lausanne, Switzerland).cell line is a well-known model for neuronal differentia-tion. It allows the monitoring of possible alterations in 2 .2. Sense and anti-sense LMO4 cDNA cloning incellular morphology in response to various stimuli known pIRESneo2to induce neuritic outgrowth. Low serum concentration(0.5%) can promote neuritogenesis in various neuroblas- Total RNA was extracted from a human cerebellartoma cell lines, SH-SY5Y cell line included[38]. SH- sample (Department of Geriatrics, University Hospital,SY5Y cells also differentiate and acquire processes upon Geneva) according to a commercial version (Life Tech-long-term treatment with TPA, at a time when PKC is nologies/ Invitrogen) of the method of Chomczynski anddownregulated[10,14,15].In the short-term, TPA is able to Sacchi[2].activate PKC, probably by substituting for diacylglycerol. After a standard reverse transcription reaction (10 min,Finally, RA is a morphogen and a regulator of differentia- 258C; 15 min, 428C; 5 min, 998C) using random hexa-tion during embryogenesis. It also promotes significant mers, PCR were carried out withLmo4 primers F1 (59-GT-neuritic outgrowth in SH-SY5Y cells[6,11,26,38]. CAGGATCCAGACCATGGTGAATCCGGGCA-39) and

]]]Thus, in the present study, we investigated whether R1 (59-GTCAGGATCCTTAGCAGACCTTCTGGTCTG-

]]]sense and anti-senseLmo4 RNA overexpression in SH- GC-39) designed withBamHI sites (underlined sequences)SY5Y cells act upon cellular aggregation or neuritogenesis, at their 59 ends. These primers correspond to the first 16and interfere with cellular responses to neuritic outgrowth and the last 22 nucleotides of theLmo4 coding sequence,promoting stimuli. respectively[20]. Pwo DNA polymerase, with proofread-

ing activity, was used during 35 cycles (15 s, 948C; 30 s,61 8C; 45 s, 728C) after a 2 min denaturation step (948C).

2 . Materials and methods In the last cycleTaq DNA polymerase was used to add anadenosine at both 39 DNA ends to allow subsequent

2 .1. Materials topoisomerase I TA cloning ([27], Invitrogen). After a finalelongation step (7 min, 728C) the amplified 523 bp

Sequencing primers and PCR primers were synthesized fragment was cloned into the linearized plasmid pCR4-by Microsynth, Switzerland. Their specificity was verified TOPO and transformed intoE. coli TOP10 according toby sequence comparison (NCBI, Blast). Random hexamers the manufacturer9s recommendations (TOPO TA cloningwere purchased from Amersham Biosciences. Geneticin kit for sequencing, Invitrogen).disulfate (G418) and plasmid pIRESneo2[22] were from Subsequently, 1mg of pCR4-TOPO with theLmo4Clontech.Taq polymerase was from Eurobio.L-glutamine, insert was digested with 2 UBamHI (1 h 30 min, 378C).PBS, penicillin, plasmid pCR4-TOPO, RPMI-1640, strep- The gel extractedLmo4 insert (Qiex II DNA extraction

D. Vu et al. / Molecular Brain Research 115 (2003) 93–103 95

from agarose gels, Qiagen) was then ligated (5 U/ml T4 18S rRNA signal to obtain a standardized value ofligase, overnight, 128C) in 100 ngBamHI linearized and expression, thus allowing a relative quantitative compari-dephosphorylated (1 U/ml shrimp alkaline phosphatase, 1 son ofLmo4 expression levels between the selected SH-h, 378C) pIRESneo2 vector. After cloning, sense or anti- SY5Y clones.sense orientation of theLmo4 insert was verified byBamHI and EcoRI digestions, respectively. Sequencing 2 .4. Culture of SH-SY5Y cellsusing primers F1 or R1 (Automated DNA sequencing,Applied Biosystems EBI 3100 DNA sequencer) was Human SH-SY5Y neuroblastoma cells were routinelycarried out to exclude any mutation in the clonedLmo4 grown in a 378C incubator containing 5% CO /95%2

inserts. The Blast program (NCBI) was used to verify the humidified air in full medium RPMI-1640 with 10% FCS,sequences. 2 mML-glutamine, 100 U/ml penicillin G and 100mg/ml

streptomycin. Cells were passaged every 5–6 days. For2 .3. Relative quantitative RT–PCR this purpose confluent cells were released in DPBS (150

mM NaCl, 3 mM KCl, 1.5 mM KH PO , 7.9 mM2 4.Relative quantitative RT–PCR was carried out accord- Na HPO?2H O, 0.1 mM EDTA; pH 7.4), centrifuged2 4 2

ing to the instruction manual ‘‘Quantum RNA 18S Internal (1100 r.p.m., 5 min) and resuspended in full medium at theStandards’’ (AMS Ambion). For the reverse transcription desired dilution.47.5 ml RNAse free H O was added to Ready-To-Go2

RT–PCR beads (Amersham Biosciences). In each experi-2 .5. Transfection of SH-SY5Y cellsment one tube with heat-inactivated reverse transcriptase(10 min, 958C) was used as a negative control. Sub- For the transfection 5mg of pIRESneo2 constructssequently, 1mg total RNA extracted from SH-SY5Y cells without or with theLmo4 insert in sense and anti-senseusing Trizol as described before and 0.5ml random orientation were linearized with 12.5 UPvuI in 25 mlhexamers (0.5mg/ml) were added in each tube. After restriction buffer for 2.5 h at 378C. Success of theprimer annealing (10 min, RT), reaction was carried out digestion was verified by visualizing the plasmids on afor 30 min at 428C and inhibited for 5 min at 958C. The 0.9% TBE MP agarose gel. The linearized plasmids werereverse transcription reaction mix was divided in two for then purified using the ‘High pure PCR product purifica-the PCR amplification of theLmo4 and the 18s rRNA tion kit’ (Roche) and at the end dissolved in 2 to 3ml H O.2

transcripts, used as invariant endogenous control. Seventy-five to 85% confluent cell cultures were trans-For the amplifications of theLmo4, 0.3ml of 15 mM F2 fected with the ‘Effectene transfection reagent’ (Qiagen).

(59-GACAGCTATTGGCACA-39) and 0.3ml of 15 mM Briefly 1.8 mg linearized plasmid was diluted in 400mlR2 primers (59-TTCTGGTCTGGCAGTAGT-39) were DNA EC condensation buffer. Enhancer (12.8ml) wasadded to the reverse transcription mix. PCR reaction added. After 2 to 5 min at room temperature, 40ml ofstarted for 2 min at 958C. This first step was followed by Effectene transfection reagent was added, the solution was27 or 30 cycles (30 s, 958C; 60 s, 498C; 60 s, 728C) and vortexed and the DNA–Effectene complexes incubated ata final elongation step (7 min at 728C). The amplified room temperature for 5–10 min. Then, 1.2 ml of fullfragment is 370 bp long. Note that 35 cycles have to be medium was added. A volume of 400ml of this mix wascarried out to visualize endogenousLmo4 transcription in distributed in each well containing the cells previouslySH-SY5Y cells. washed in PBS and resuspended in 1.8 ml full medium.

The 324 bp 18S rRNA product was PCR amplified as After 24 h incubation cells were washed in PBS anddescribed for Lmo4 except that 0.4ml of a 5 mM resuspended in full medium. Twenty-four hours latercompetimer to 5mM 18S rRNA primer mixture (3 to 7 geneticin was added at 200mg/ml, a concentration thatratio) was used instead of theLmo4 primers and that the eliminates non transfected cells within 1 week. After 16 toannealing temperature was 50.68C. The 20 days selection clones were isolated and the rest of theQuantumRNA�18S internal standards with the Com- cells kept as pools.petimer technology (AMS Ambion) was used to preventsignal saturation. 2 .6. Cell culture stimulations and differentiation assays

Five microliters of amplified transcripts were visualizedon 1% TBE (90 mM Tris–borate, 2 mM EDTA) MP For each pIRESneo2 construct, one clone was grown inagarose gels by ethidium bromide staining. Pictures were RPMI-1640 full medium in presence of geneticin (200taken (Gelprint workstation, MWG-Biotech). Signal inten- mg/ml). As 0.5% FCS stimulates neuritogenesis of non-sities of the PCR products were measured with a GS-700 transfected SH-SY5Y cells grown in RPMI-1640, TPA anddensitometer (Bio-Rad) and were expressed as optical RA were applied with this low serum concentration to

2density unit3area (mm ) using the molecular analyst possibly strengthen differences in phenotype. When cellssoftware program (Bio-Rad). The intensity of aLmo4 reached 50% confluency the medium was changed withsignal was divided by the intensity of the corresponding 0.5% FCS only and the cells were incubated with geneticin

96 D. Vu et al. / Molecular Brain Research 115 (2003) 93–103

as well as RA (500 nM) or TPA (100 nM). A second 6 M urea (modified from[25]); Miniprotean II dual slabstimulation (with changes of growth medium) was carried cell, Bio-Rad) and then electroblotted (Semi-dry electro-out 48 h later. Alterations in cellular morphology were blotter, Model HEP-1, Owl Scientific; 1 h 40 min, 2

2regularly monitored (4 h, 48 h, 72 h). Alternatively, RPMI- mA/cm , 10–14 V) to a nitrocellulose membrane (Protran,1640 with 10% FCS was replaced by Neurobasal medium Schleicher and Schuell). Acrylamide gels (7%) were usedsupplemented with serum-free B27, 2 mML-glutamine, with antibodies NF-L 68 and SMI 31, and 10% gels with100 U/ml penicillin G, 100mg/ml streptomycin and 200 Tau-1. Non-specific binding sites were blocked by incubat-mg/ml geneticin. After 24 h, TPA (32 nM) was added and ing the membrane in 1% nonfat dry milk, 1% BSA, 0.05%cells were harvested for Western blotting when they were Tween-20, PBS 13 (pH 7.4) for 45 min. The membraneconfluent, i.e. about 6 days later. In all cases photographs was then incubated for 1 h 30 min with the primary mousewere taken after 72 h. Each experiment was repeated at monoclonal antibody diluted in blocking solution (at aleast three times. 1:2000, 1:3000 and 1:10,000 dilution for NF-L 68, Tau-1

and SMI 31, respectively), and subsequently for 30 min2 .7. Data analysis with a horseradish peroxidase-labeled sheep anti-mouse

IgG secondary antibody diluted in blocking solutionFor each cell group, i.e. pIRESneo2 transfected cells (1:4000). Specific bands were detected by chemilumines-

without or with Lmo4 insert in sense and anti-sense cence (SuperSignal West Pico Cheminescent Substrate,orientation, stimulated or not, 2 to 4 black and white 9313 Pierce) on film (Hyperfilm ECL, Amersham). Their inten-cm photographs (film T-MAX 100 pro; Kodak) were taken sities were measured by densitometry as described forwith the digital photo apparatus KR-10M (Ricoh) coupled relative quantitative RT–PCR (Section 2.3). The secondaryto an inversed Televal 31 microscope (Zeiss). Cells were antibody has been shown not to contribute to the observedbetween 40 and 90% confluency depending on the applied signals.treatment. For each clone or pool of cells two to fourrandomly chosen fields were analyzed manually. For eachfield, the numbers of cells (A) and neurites (B) were 3 . Resultscounted, and the total neuritic length (inmm; C) wasmeasured. In this case measured neuritic segments were3 .1. Sense and anti-sense Lmo4 cDNA transfection insummed and converted to micrometers by appropriate SH-SY5Y neuroblastoma cellsconversion factors. For the experiment with TPA stimula-tion the number of branchings (D) was also counted. For A 523 bp cDNA fragment corresponding to the full-the statistical analyzes a Fisher–Snedecor test (ANOVA) length humanLmo4 coding sequence was cloned into thewas applied to the average number of neurites per cell pIRESneo2 transfection vector following sub-cloning in(B /A), the average length of a neurite (inmm; C /B), the the vector pCR4-TOPO. Transfections in SH-SY5Y neuro-average neuritic length per cell (inmm; C /A), or the blastoma cells were carried out withPvuI linearizedaverage number of neuritic branchings per cell (D/A). Note plasmid pIRESneo2, used as negative control, pIRESneo2that the valuesC /B and C /A do not take the degree of with theLmo4 insert in sense orientation to overexpressneuritic arborisation into consideration and that no distinc- Lmo4 RNA, as well as in anti-sense orientation to inhibittions were made between nascent and differentiated neur- the endogenous mRNALmo4 translation. For each trans-ites. fection six clones were isolated and the rest of the cells

kept as pools. Clones 2 and 3 for pIRESneo2, clones 8 and2 .8. Western blotting 10 for pIRESneo2 with anti-senseLmo4 and clones 13 and

16 for pIRESneo2 with senseLmo4, as well as theCells were rinsed twice in Tris buffer (50 mM Tris, 1 respective pooled cells were further analyzed.

mM EDTA, 2 mM MgCl , 0.25 M sucrose, pH 7.5). They Expression of the sense and anti-senseLmo4 RNAs was2

were subsequently scraped and collected by centrifugation assessed in the transfected cells by relative-quantitative(10 min, 2000 g, 4 8C) in the presence of protease RT–PCR (Fig. 1A) and subsequent calculation of the ratioinhibitors Pefabloc SC and Complete, and resuspended in of the intensity of theLmo4 signal divided by the intensitysolubilization buffer (50 mM Tris, 2% SDS, 3 M Urea, pH of the 18S rRNA signal. The 18s rRNA was used as an6.8). Sonication (Bransonic, 1 to 2 min) and addition of invariant internal control to confirm that similar amounts200–300 U endonuclease were carried out to destroy of total RNA were used in each RT–PCR. With theDNA. Subsequently, protein concentrations were deter- exception of one 18S rRNA amplification that likely failedmined by the BCA protein assay. The proteins were (Fig. 1A; 7), all other 18S rRNA reactions showed cleardenatured in Laemmli loading buffer (50 mM Tris–HCl, signals in control (Fig. 1A; 1–4) and transfected cells (Fig.pH 6.8, 10% glycerol, 2.5%b-mercaptoethanol, 2% SDS, 1A; 5, 6, 8–10). Obtained signal ratios were the following:0.01% Bromophenol Blue) at 988C for 6 min. untransfected SH-SY5Y, pIRESneo2 (pool), clone 2 and 3:

Equal amounts of proteins were separated by SDS– 0; pIRESneo2 with senseLmo4 (pool): 2.5; clone 13: 0.5;polyacrylamide gel electrophoresis (7 or 10% acrylamide, pIRESneo2 with anti-senseLmo4 (pool): 0.4; clone 8: 0.9

D. Vu et al. / Molecular Brain Research 115 (2003) 93–103 97

Fig. 1. (A) Lmo4 mRNA and 18S rRNA expression in SH-SY5Y neuroblastoma cells transfected with pIRESneo2 without (clones 2 and 3) or withLmo4insert in anti-sense (clones 8 and 10) or sense orientation (clones 13 and 16): a RT–PCR semi-quantitative study. After reverse transcription a 370 bpLmo4and a 324 bp 18S rRNA fragment were PCR amplified for 30 cycles and visualized on a 1% MP agarose gel. Total RNA was extracted from nontransfected SH-SY5Y cells (1), a pool of cells with pIRESneo2 (2), clone 2 (3), clone 3 (4), a pool of cells with pIRESneo2 with senseLmo4 (5), clone 13(6), clone 16 (7), a pool of cells with pIRESneo2 with anti-senseLmo4 (8), clone 8 (9) and clone 10 (10). (B) EndogenousLmo4 expression in nontransfected SH-SY5Y cells was detectable after 35 PCR cycles. Lengths of standard molecular weights (St) are in base pairs (bp).

and clone 10: 0.17. Thus, under the chosen conditions of about 50 to 70% confluency representative pictures wereamplification Lmo4 transcription was visualized only in taken (Fig. 2A). For each clone and pool, 211 to 440 cellsthe cells transfected with the pIRESneo2 with cloned sense were counted (A5number of cells) and analyzed. Theand anti-senseLmo4 cDNA (Fig. 1A, 5–10). Indeed, number of neurites (B) and the total length of neurites perendogenousLmo4 transcription was not detected in the analyzed picture field (C) were measured, and the averagecontrol SH-SY5Y cells, transfected or not with number of neurites per cell (B /A), the average length of apIRESneo2, despite similar levels of 18S rRNA and total neurite (C /B) and the average neuritic length per cellRNA used as start material (Fig. 1A, 1–4). Endogenous (C /A) were subsequently calculated (Fig. 2B). Despite theLmo4 transcription can be visualized by increasing the fact that clones transfected with the same construct maynumber of PCR cycles (35 cycles) leaving all other behave differently, possibly due to various chromosomalexperimental conditions unchanged (Fig. 1B). However, to insertion sites [e.g., clone 3 (Fig. 2B,4) compared to cloneallow comparison of expression between clones, relative 2 (Fig. 2B, 3)], three main observations could be made:quantitative RT–PCR have to be carried out with less PCRcycles, thus preventing signal saturation and diminution ofsignal differences. Amplification after heat inactivation of • There was a significant increase (P,0.001) of all threereverse transcriptase generated no signals demonstrating parameters studied (B /A, C /B and C /A) in the cellsthat genomic DNA was not amplified (Result not shown). overexpressing anti-senseLmo4 RNA (Fig. 2B, 5–7;Verification of changes in LMO4 protein expression by Fig. 2A, clone 10) compared to the cells with theWestern blotting was inconclusive. Indeed commercial transfected vector without insert (Fig. 2B, 2–4) or(polyclonal) and home-made (monoclonal) antibodies did overexpressingLmo4 RNA in sense orientation (Fig.not result in the expected 18 kDa LMO4 protein signal but 2B, 8–10).in higher molecular bands, often varying in size between • On the contrary, in the cells overexpressing senseantibodies (see Discussion). Lmo4 RNA (Fig. 2B, 8–10;Fig. 2A, clone 13), there

Overall, these results showed that transfected sense and was a significant decrease in the average number ofanti-senseLmo4 cDNAs are indeed transcribed and over- neurites per cell (B /A; P,0.001) and the averageexpressed compared to the endogenousLmo4 gene. neuritic length per cell (C /A; P,0.01). However, the

diminution of the average neuritic length (C /B; Fig.3 .2. Cellular differentiation of SH-SY5Y neuroblastoma 2B middle) was less important for clones 13 (P,0.05)cells transfected or not with sense and anti-sense Lmo4 and 16 (P,0.01), and not significant for the pooledcDNA cells. Comparisons were carried out with the control

cells transfected with the pIRESneo2 vector (Fig. 2B,All SH-SY5Y cells, pools or clones, were grown in 2).

RPMI-1640 with 10% FCS. Only for the transfected clones • It is also worth noting that cells that overexpressedgeneticin (200mg/ml) was added. When cells were at Lmo4 RNA in sense orientation tended to aggregate.

98 D. Vu et al. / Molecular Brain Research 115 (2003) 93–103

Fig. 2. (A) Illustration of morphological differences of SH-SY5Y neuroblastoma cells (SH-SY5Y), clone 2 (pIRESneo2), clone 10 (anti-senseLmo4) andclone 13 (senseLmo4). Scale bar550 mm. (B) Histograms presenting average values and standard errors of the number of neurites (B) per cell (A) (left,B /A), the length of one neurite (middle,C /B in mm) and the neuritic length per cell (right,C /A in mm). Non transfected SH-SY5Y cells (1), SH-SY5Ycells transfected with pIRESneo2 (pool of cells (2), clones 2 (3) and 3 (4)), with pIRESneo2 with anti-senseLmo4 (pool of cells (5), clones 8 (6) and 10(7)), and with senseLmo4 (pool of cells (8), clones 13 (9) and 16 (10)).Values in 1, 2, 3, 4, and 8, 9, 10 are always significantly different (P,0.001) fromthose in 5, 6 and 7. Values in 1, 2, 3 are significantly different from those in 8, 9, 10 (P,0.001 forB /A, P,0.01 for C /A). For C /B the difference issignificant between 1, 2, 3 and 9, 10 (P,0.01 andP,0.001, for 9 and 10, respectively), but not between 1, 2, 3 and 8. Differences between 4 and 8, 9, 10are usually non-significant.

Thus, it appears that both sense and anti-senseLmo4 still inhibiting neuritogenesis in presence of neuritic out-RNA are involved in the regulation of neuritic length and growth promoting stimuli, clone 13 (senseLmo4), cloneoutgrowth. 10 (anti-senseLmo4) as well as clone 2 (pIRESneo2) were

treated with RA or TPA in 0.5% FCS. A total of 238 to3 .3. Responses of SH-SY5Y neuroblastoma cells 357, and 81 to 159 cells were analyzed in two randomlytransfected or not with sense and anti-sense Lmo4 cDNA chosen fields for each RA- and TPA-stimulated clones,to neuritic outgrowth promoting stimuli respectively. Numbers of cells (A), neurites (B), total

neuritic length (C) and neuritic branchings (D) wereTo verify if the overexpression of senseLmo4 RNA was measured. The following observations were noteworthy:

D. Vu et al. / Molecular Brain Research 115 (2003) 93–103 99

also overrides the inhibitory effect of senseLmo41. The RA treatment resulted for all three clones in RNA overexpression (clone 13) on neuritic outgrowth.

similar neuritic outgrowth and absence of massive cell 2. The TPA stimulation (Fig. 3A, TPA) resulted indeath (Fig. 3A, RA). Indeed, there was no significant similar neuritic outgrowth (difference non-significant)(P.0.05) difference in the average number of neurites in clone 2 and clone 10, not in clone 13. Indeed, thereper cell (B /A), the average length of a neurite (C /B) was a significant difference (P,0.05) between cloneand the average neuritic length per cell (C /A; Fig. 3B, 13, and clones 2 and 10 for the average number ofleft) between the clones. Thus, not only RA induces neurites per cell (B /A), the average length of a neuriteneuritogenesis, as already known (e.g.,[38]), but it (C /B) and the average neuritic length per cell (C /A,

Fig. 3. (A) Retinoic acid (RA) and phorbol ester (TPA) treatments (in 0.5% FCS) of clone 2 (pIRESneo2), clone 10 (anti-senseLmo4) and clone 13 (senseLmo4). Scale bar585mm. (B) Histograms presenting average values and standard errors of the neuritic length (C) per cell (A) after RA (left,C /A in mm)or TPA (middle) treatment, as well as of the number of neuritic branchings (D) per cell after TPA treatment (right,D/A). Clone 2 (1), clone 10 (2), clone 13(3). C /A values are significantly different (*P,0.05) between clone 13 and clones 2 and 10 only after TPA treatment (middle), not between clones 2 and10 or after RA treatment (left).D/A values are significantly different (*P,0.05) between clones 10 and 13, not between clones 2 and 10 (right).

100 D. Vu et al. / Molecular Brain Research 115 (2003) 93–103

Fig. 3B, middle). Thus, senseLmo4 RNA overexpres-sion (clone 13), notLmo4 endogenous expression • Alterations of unphosphorylated tau isoforms signals(clone 2), overrides the stimulatory effect of TPA upon (antibody Tau-1) in clones 10 and 13 compared toneuritic outgrowth. control clone 2. Moreover, the TPA treatment resulted

in decreased signals in clones 10 (247%) and 13(275%), less so in clone 2 (222%). This correspondsAverage numbers of neuritic branchings per cell (D/A)to an increase in phosphorylation of tau epitopes. Inwere significantly different (P,0.05) between clones 13these cases, only signal intensities of the more intenseand 10 only (Fig. 3B, right). In this last case branchingsbands were measured.are slightly more numerous than in clone 2 and varicosities

• Increased phosphorylated NF-H and NF-M signals inare particularly noteworthy. However, the difference be-clone 10 compared to the other clones. A highertween these two clones was below significance level (P.molecular weight NF-H signal (upper band), likely0.05).corresponding to post-translational modified, hyper-Similar observations were made with clones 2, 10 andphosphorylated NF-H, is also visible with clone 10.13 grown in neurobasal medium as in RPMI-1640 full

• Increased levels of the 68 kDa NF-L signal in clone 2medium.(5.33) and 13 (2.93) compared to clone 10 inabsence of TPA treatment. Following TPA treatment

3 .4. Effects of sense and anti-sense Lmo4 transcription weaker alterations in NF-L or even NF-M signalupon cytoskeletal proteins intensities are possible.

Western blottings were carried out to determine possible Overall, these data suggest that the level ofLmo4alterations at the level of cytoskeletal proteins that are transcription participates or interferes with the regulationinvolved in neurite outgrowth and retraction (Fig. 4). of the metabolism of cytoskeletal proteins by modulatingComparisons between clones 2, 10 and 13 revealed: kinases or phosphatases activities.

Fig. 4. Western blots carried out with clone 2 (pIRESneo2), clone 10 (anti-senseLmo4) and clone 13 (senseLmo4) treated (1) or not (2) with TPA.Equal amounts (20mg) of proteins were loaded in duplicate. The antibodies used reacted against proteins of the cytoskeleton (tau, neurofilaments: NF-H,NF-M and NF-L). Molecular weights as well as purified human neurofilaments (NF) of 200 kDa (NF-H) and 160 kDa (NF-M) are shown on the right side.

D. Vu et al. / Molecular Brain Research 115 (2003) 93–103 101

4 . Discussion phosphorylation, e.g., those observed in tau at serines 195,198, 199 or 202 suggest that GSK-3b, JNK, cdk5, MAPK

In the present studyLmo4 cDNA was cloned in sense or phosphatases might depend onLmo4 expression. Inand anti-sense orientation in the vector pIRESneo2. After addition we observed for clone 10 and 13 TPA-mediatedpermanent transfection of the constructs in human SH- phospho-tau increased immunoreactivity. PKC is known toSY5Y neuroblastoma cells and verification of transcription phosphorylate tau via the MAP kinase pathway[4]. Theof the Lmo4 inserts, their effects on cellular differentia- state of phosphorylation appears however lower than intion, with or without neuritic outgrowth promoting treat- control clone 2 (Fig. 4). In this case microtubules might bements, have been monitored. less stable due to the dissociation of phosphorylated tau

Transfection experiments demonstrated that overexpres- from the tubulins. In clone 10, that shows longer neurites,sion of senseLmo4 RNA interferes with neuritogenesis NF-H and NF-M phosphorylations are strongly increased.and reduces the average number of neurites per cell, and, Phosphorylation of these proteins has been implicated into a lesser extent, the neuritic length per cell, whereas the the process responsible for the radial expansion of axonsaverage length of a neurite was significantly reduced in [16]. Thus, in the SH-SY5Y cells with longer neurites,clones 13 and 16, but not in the pooled cells. Overexpres- phosphorylated NF-M and NF-H might be needed tosion of the anti-senseLmo4 transcript favored higher increase the neuritic caliber providing clear spacing foraverage numbers of neurites and neuritic length per cell, as organelle transport and proper functioning of the neurites.well as a higher average length of a neurite. These analyses However, the precise transduction pathways involved inare valid when comparisons were made with the SH-SY5Y Lmo4-mediated regulation of neuritogenesis have still tocells or the pooled cells transfected with the vector without be investigated. The mechanism that prevents neuriticLmo4 insert. Note that the mean neuritic length corre- outgrowth in clone 13 following TPA stimulation has alsosponds to the data of other studies (e.g.,[21]). Overexpres- to be elucidated. It might be related to the state of tau-,sion of senseLmo4 RNA also favored aggregation of the NF-phosphorylation or content that differ between thistransfected cells, a marker of tumoral cells. Overall, these clone and the other ones. To further define molecular linksdata corroborate the observations made by Visvader et al. betweenLmo4 and neuritogenesis, identification (e.g., by[34] that showedLmo4 mRNA overexpression in human microarrays) of altered gene expression in SH-SY5Y cellsbreast cancer cell lines and suggested an inhibitory effect overexpressing sense or anti-senseLmo4 RNA would alsoof Lmo4 on differentiation (i.e., mRNA expression of milk be worth investigating.proteins such asb-casein) in mammary epithelial cells. This study is limited by our difficulties to monitor

Clones 2, 10 and 13 corresponding to the clones with LMO4 proteins by Western blotting. The available com-plasmid pIRESneo2, pIRESneo2 withLmo4 cDNA in mercial and home-made antibodies did not result in theanti-sense and sense orientation, respectively, were grown expected 18 kDa LMO4 protein signal but in other notin culture conditions promoting neuritic outgrowth (RA or easily interpretable signals (results not shown). In thisTPA in 0.5% FCS). The following main conclusions could context, it is worth mentioning that LMO4 posttranslation-be drawn: First, the inhibitory effect of senseLmo4 RNA al modifications, such as ubiquitination[13] are possibleoverexpression on neuritogenesis can be overidden by RA. and must be considered in future experiments.Thus, this biochemical compound likely acts on pathways In conclusion,Lmo4 RNA overexpression interferesthat must be dominant over theLmo4 signaling pathway. with SH-SY5Y neuroblastoma cell differentiation, favorsSecondly, the inhibitory effect of senseLmo4 RNA cell division and aggregation, whereas anti-senseLmo4overexpression on neuritogenesis cannot be overriden by RNA overexpression mediates neuritogenesis.Lmo4TPA, a PKC activator, as observed with clone 13. The fact changes in expression might thus play a role not only inthat senseLmo4 RNA overexpression resulted in reduced regulation of neuritogenesis but also during brain develop-average number and size of neurites appears to be the most ment as suggested by Kenny et al.[9], or even duringobvious reason why TPA could not induce branchings in neurodegenerative diseases such as Alzheimer’s disease.this case as well. TPA induced branchings in cells trans- Indeed, we have observed by immunohistochemistry car-fected with the vector alone and with the anti-senseLmo4 ried out in Alzheimer compared to control brains acDNA. In this later case, branchings appeared to be consistent decrease of LMO4 expression in two of theslightly (difference below level of significance) more most vulnerable regions in this disease, the entorhinalnumerous than in the clone with the vector alone. Thus, it cortex and the CA1 hippocampal region (unpublishedis possible, but not proven, that a long-term stimulation results). In this context, it is worth mentioning thatwith TPA (72 h) might slightly increase neuritic branchings decreasedLmo4 transcription was observed by microarrayin clone 10 compared to clone 2. analysis in the hippocampus of a transgenic mouse model

We have also shown by Western analysis that theLmo4- of Alzheimer’s disease overexpressing a mutant amyloiddependent neuritic retraction, or outgrowth correspond to precursor protein[28]. Overall, our experiments show thatalterations of cytoskeletal proteins (tau, NF-H, NF-M and changes inLmo4 expression levels might alter the rate ofNF-L). These are at least partly due to changes in neuritic outgrowth in the developing and adult nervous

102 D. Vu et al. / Molecular Brain Research 115 (2003) 93–103

of candidate genes associated with prostate cancer progression in thesystem. The significance ofLmo4 misregulation hasCWR22 model system using tissue microarrays, Cancer Res. 62already been shown in tumorigenesis[12,30,34] and the(2002) 1256–1260.

effects of Lmo4 transcription levels have to be further [13] H .P. Ostendorff, R.I. Peirano, M.A. Peters, A. Schluter, M. Bossenz,investigated in the development of the central nervous M. Scheffner, I. Bach, Ubiquitination-dependent cofactor exchangesystem as well as in the adult and the aging brain. on LIM homeodomain transcription factors, Nature 416 (2002)

99–103.[14] S . Pahlman, A.I. Ruusala, L. Abrahamsson, M.E. Mattsson, T.

Esscher, Retinoic acid-induced differentiation of cultured humanA cknowledgements neuroblastoma cells: a comparison with phorbolester-induced dif-

ferentiation, Cell Differ. 14 (1984) 135–144.[15] S . Pahlman, A.I. Ruusala, L. Abrahamsson, L. Odelstad, K. Nilsson,The authors thank Dr B.M. Riederer for the NFs, Ms B.

Kinetics and concentration effects of TPA-induced differentiation ofPastori and C. Aubry for technical assistance as well as Drcultured human neuroblastoma cells, Cell Differ. 12 (1983) 165–

C. Rossier for automated sequencing and J. D. Laughton 170.for helpful comments and criticism. [16] H .C. Pant, Veeranna, Neurofilament phosphorylation, Biochem. Cell

Biol. 73 (1995) 575–592.[17] T .H. Rabbitts, Chromosomal translocation master genes, mouse

models and experimental therapeutics, Oncogene 20 (2001) 5763–R eferences 5777.

[18] T .H. Rabbitts, H. Axelson, A. Forster, G. Grutz, I. Lavenir, R.Larson, H. Osada, V. Valge-Archer, I. Wadman, A. Warren, Chromo-[1] T . Boehm, L. Foroni, Y. Kaneko, M.F. Perutz, T.H. Rabbitts, Thesomal translocations and leukaemia: a role for LMO2 in T cell acuterhombotin family of cysteine-rich LIM-domain oncogenes: distinctleukaemia, in transcription and in erythropoiesis, Leukemia 11members are involved in T-cell translocations to human chromo-(Suppl.) (1997) 271–272.somes 11p15 and 11p13, Proc. Natl. Acad. Sci. USA 88 (1991)

[19] T .H. Rabbitts, K. Bucher, G. Chung, G. Grutz, A. Warren, Y.4367–4371.Yamada, The effect of chromosomal translocations in acute[2] P . Chomczynski, N. Sacchi, Single-step method of RNA isolation byleukemias: the LMO2 paradigm in transcription and development,acid guanidinium thiocyanate–phenol–chloroform extraction, Anal.Cancer Res. 59 (Suppl.) (1999) 1794–1798.Biochem. 162 (1987) 156–159.

[20] J . Racevskis, A. Dill, J.A. Sparano, H. Ruan, Molecular cloning of[3] I .B. Dawid, J.J. Breen, R. Toyama, LIM domains: multiple roles asLMO4, a new human LIM domain gene, Biochim. Biophys. Actaadapters and functional modifiers in protein interactions, Trends1445 (1999) 148–153.Genet. 14 (1998) 156–162.

[21] M . Raghunath, R. Patti, P. Bannerman, C.M. Lee, S. Baker, L.N.[4] F .J. Ekinci, T.B. Shea, Free PKC catalytic subunits (PKM) phos-Sutton, P.C. Phillips, C. Damodar Reddy, A novel kinase, AATYKphorylate tau via a pathway distinct from that utilized by intactinduces and promotes neuronal differentiation in a human neuro-PKC, Brain Res. 850 (1999) 207–216.blastoma (SH-SY5Y) cell line, Mol. Brain Res. 77 (2000) 151–162.[5] G . Grutz, A. Forster, T.H. Rabbitts, Identification of the LMO4 gene

[22] S . Rees, J. Coote, J. Stables, S. Goodson, S. Harris, M.G. Lee,encoding an interaction partner of the LIM-binding protein LDB1/Bicistronic vector for the creation of stable mammalian cell linesNLI1: a candidate for displacement by LMO proteins in T cell acutethat predisposes all antibiotic-resistant cells to express recombinantleukaemia, Oncogene 17 (1998) 2799–2803.protein, BioTechniques 20 (1996) 102–104.[6] F . Grynspan, W.B. Griffin, P.S. Mohan, T.B. Shea, R.A. Nixon,

[23] B . Royer-Pokora, U. Loos, W.D. Ludwig, TTG-2, a new geneCalpains and calpastatin in SH-SY5Y neuroblastoma cells duringencoding a cysteine-rich protein with the LIM motif, is over-retinoic acid-induced differentiation and neurite outgrowth: com-expressed in acute T-cell leukaemia with the t(11;14)(p13;q11),parison with the human brain calpain system, J. Neurosci. Res. 48Oncogene 6 (1991) 1887–1893.(1997) 181–191.

[24] I . Sanchez-Garcia, T.H. Rabbitts, LIM domain proteins in leukaemia[7] O . Hermanson, T.M. Sugihara, B. Andersen, Expression of LMO-4and development, Semin. Cancer Biol. 4 (1993) 349–358.in the central nervous system of the embryonic and adult mouse,

[25] N .M. Scherer, M.-J. Toro, M.L. Entman, L. Birnbaumer, G-proteinCell Mol. Biol. 45 (1999) 677–686.distribution in canine cardiac sarcoplasmic reticulum and sacrolem-[8] O . Hobert, H. Westphal, Functions of LIM-homeobox genes, Trendsma: comparison to rabbit skeletal muscle membranes and to brainGenet. 16 (2000) 75–83.and erythrocyte G-proteins, Arch. Biochem. Biophys. 259 (1987)[9] D .A. Kenny, L.W. Jurata, Y. Saga, G.N. Gill, Identification and431–440.characterization of LMO4, an LMO gene with a novel pattern of

[26] T .B. Shea, I. Fischer, V.S. Sapirstein, Effect of retinoic acid onexpression during embryogenesis, Proc. Natl. Acad. Sci. USA 95growth and morphological differentiation of mouse NB2a neuro-(1998) 11257–11262.blastoma cells in culture, Brain Res. 353 (1985) 307–314.[10] U . Leli, T.B. Shea, A. Cataldo, G. Hauser, F. Grynspan, M.L.

[27] S . Shuman, Novel approach to molecular cloning and polynucleotideBeermann, V.A. Liepkalns, R.A. Nixon, P.J. Parker, Differentialsynthesis using vaccinia DNA topoisomerase, J. Biol. Chem. 269expression and subcellular localization of protein kinase C alpha,(1994) 32678–32684.beta, gamma, delta, and epsilon isoforms in SH-SY5Y neuro-

[28] T .D. Stein, J.A. Johnson, Lack of neurodegeneration in transgenicblastoma cells: modifications during differentiation, J. Neurochem.mice overexpressing mutant amyloid precursor protein is associated60 (1993) 289–298.with increased levels of transthyretin and the activation of cell[11] G . Lopez-Carballo, L. Moreno, S. Masia, P. Perez, D. Barettino,survival pathways, J. Neurosci. 17 (2002) 7380–7388.Activation of the phosphatidyl-inositol-3 kinase/Akt signaling path-

[29] T .M. Sugihara, I. Bach, C. Kioussi, M.G. Rosenfeld, B. Andersen,way by retinoic acid is required for neural differentiation of SH-Mouse deformed epidermal autoregulatory factor 1 recruits a LIMSY5Y human neuroblastoma cells, J. Biol. Chem. 277 (2002)domain factor, LMO-4, and CLIM coregulators, Proc. Natl. Acad.25297–25304.Sci. USA 95 (1998) 15418–15423.[12] S . Mousses, L. Bubendorf, U. Wagner, G. Hostetter, J. Kononen, R.

Cornelison, N. Goldberger, A.G. Elkahloun, N. Willi, P. Koivisto, W. [30] E .Y. Sum, B. Peng, X. Yu, J. Chen, J. Byrne, G.J. Lindeman, J.E.Ferhle, M. Raffeld, G. Sauter, O.P. Kallioniemi, Clinical validation Visvader, The LIM domain protein LMO4 interacts with the

D. Vu et al. / Molecular Brain Research 115 (2003) 93–103 103

cofactor CtIP and the tumor suppressor BRCA1 and inhibits BRCA1 cells in vitro and is overexpressed in breast cancer, Proc. Natl. Acad.activity, J. Biol. Chem. 277 (2002) 7849–7856. Sci. USA 98 (2001) 14452–14457.

[31] E . Tse, G. Grutz, A.A. Garner, Y. Ramsey, N.P. Carter, N. Copeland, [35] H . von der Kammer, C. Demiralay, B. Andresen, C. Albrecht, M.D.J. Gilbert, N.A. Jenkins, A. Agulnick, A. Forster, T.H. Rabbitts, Mayhaus, R.M. Nitsch, Regulation of gene expression by muscarinicCharacterization of the Lmo4 gene encoding a LIM-only protein: acetylcholine receptors, Biochem. Soc. Symp. 67 (2001) 131–140.genomic organization and comparative chromosomal mapping, [36] I .A. Wadman, H. Osada, G.G. Grutz, A.D. Agulnick, H. Westphal,Mamm. Genome 10 (1999) 1089–1094. A. Forster, T.H. Rabbitts, The LIM-only protein Lmo2 is a bridging

[32] S . Tokutake, S.B. Hutchison, J.S. Pachter, R.K. Liem, A batchwise molecule assembling an erythroid, DNA-binding complex whichpurification procedure of neurofilament proteins, Anal. Biochem. includes the TAL1, E47, GATA-1 and Ldb1/NLI proteins, EMBO135 (1983) 102–105. J. 16 (1997) 3145–3157.

[33] V .E. Valge-Archer, H. Osada, A.J. Warren, A. Forster, J. Li, R. Baer, [37] J . Wang, Y.C. Tung, Y. Wang, X.T. Li, K. Iqbal, I. Grundke-Iqbal,T.H. Rabbitts, The LIM protein RBTN2 and the basic helix–loop– Hyperphosphorylation and accumulation of neurofilament proteinshelix protein TAL1 are present in a complex in erythroid cells, Proc. in Alzheimer disease brain and in okadaic acid-treated SY5Y cells,Natl. Acad. Sci. USA 91 (1994) 8617–8621. FEBS Lett. 507 (2001) 81–87.

[34] J .E. Visvader, D. Venter, K. Hahm, M. Santamaria, E.Y. Sum, L. [38] G . Wu, Y. Fang, Z.H. Lu, R.W. Ledeen, Induction of axon-like andO’Reilly, D. White, R. Williams, J. Armes, G.J. Lindeman, The LIM dendrite-like processes in neuroblastoma cells, J. Neurocytol. 27domain gene LMO4 inhibits differentiation of mammary epithelial (1998) 1–14.