Effects of all-trans-retinoic acid on human SH-SY5Y neuroblastoma as in vitromodel in neurotoxicity research

Yuen-Ting Cheung a,1, Way Kwok-Wai Lau a,1, Man-Shan Yu a, Cora Sau-Wan Lai a,Sze-Chun Yeung a, Kwok-Fai So a,b,c, Raymond Chuen-Chung Chang a,b,c,*a Laboratory of Neurodegenerative Diseases, Department of Anatomy, The University of Hong Kong, Pokfulam, Hong Kong SAR, Chinab Research Centre of Heart, Brain, Hormone and Healthy Aging, LKS Faculty of Medicine, The University of Hong Kong, Pokfulam, Hong Kong SAR, Chinac State Key Laboratory of Brain and Cognitive Sciences, The University of Hong Kong, Pokfulam, Hong Kong SAR, China

NeuroToxicology 30 (2009) 127–135

A R T I C L E I N F O

Article history:

Received 8 July 2008

Accepted 3 November 2008

Available online 14 November 2008

Keywords:

Differentiation

Neuroblastoma

Neurotoxicity

Retinoic acid

6-Hydroxydopamine

MPP+

A B S T R A C T

Human neuroblastoma SH-SY5Y is a dopaminergic neuronal cell line which has been used as an in vitro

model for neurotoxicity experiments. Although the neuroblastoma is usually differentiated by all-trans-

retinoic acid (RA), both RA-differentiated and undifferentiated SH-SY5Y cells have been used in

neuroscience research. However, the changes in neuronal properties triggered by RA as well as the

subsequent responsiveness to neurotoxins have not been comprehensively studied. Therefore, we aim to

re-evaluate the differentiation property of RA on this cell line. We hypothesize that modulation of signaling

pathways and neuronal properties during RA-mediated differentiation in SH-SY5Y cells can affect their

susceptibility to neurotoxins. The differentiation property of RA was confirmed by showing an extensive

outgrowth of neurites, increased expressions of neuronal nuclei, neuron specific enolase, synaptophysin

and synaptic associated protein-97, and decreased expression of inhibitor of differentiation-1. While

undifferentiated SH-SY5Y cells were susceptible to 6-OHDA and MPP+, RA-differentiation conferred SH-

SY5Y cells higher tolerance, potentially by up-regulating survival signaling, including Akt pathway as

inhibition of Akt removed RA-induced neuroprotection against 6-OHDA. As a result, the real toxicity cannot

be revealed in RA-differentiated cells. Therefore, undifferentiated SH-SY5Y is more appropriate for studying

neurotoxicity or neuroprotection in experimental Parkinson’s disease research.

� 2008 Elsevier Inc. All rights reserved.

Contents lists available at ScienceDirect

NeuroToxicology

1. Introduction

Parkinson’s disease (PD) is an aging-related disease with noeffective treatments (Kedar, 2003). In studies of the properties ofneurotoxins and development of new therapeutic compoundsfor disease management, in vitro cell culture models are oftenused (Segura-Aguilar and Kostrzewa, 2006), among whichneuroblastoma SH-SY5Y cell line has commonly been chosento study the pathogenesis of neurodegeneration (Chang et al.,2002; Xue et al., 2006; Zheng et al., 2006) and for drug screening(Levites et al., 2002a). 6-Hydroxydopamine (6-OHDA) and1-methyl-4-phenyl-pyridinium ion (MPP+) are commonly usedneurotoxins in experimental PD.

As derived from neuroblastoma, SH-SY5Y cells are ofteninduced to differentiate by All-trans-retinoic acid (RA) to obtainmore neuron-like properties, including neurite outgrowth and

* Corresponding author at: Department of Anatomy, LKS Faculty of Medicine, The

University of Hong Kong. Rm. L1-49, Laboratory Block, Faculty of Medicine Building,

21 Sassoon Road, Pokfulam, Hong Kong SAR, China. Tel.: +852 2819 9127;

fax: +852 2817 0857.

E-mail address: rccchang@hkucc.hku.hk (R.-C. Chang).1 These two authors have equal contribution to this study.

0161-813X/$ – see front matter � 2008 Elsevier Inc. All rights reserved.

doi:10.1016/j.neuro.2008.11.001

morphological changes (Pahlman et al., 1984), so as to mimicresponses of neurons in studies. RA has also been shown to down-regulate the mRNA and protein levels of the differentiation-inhibiting basic helix-loop-helix (Id) transcription factors (Lopez-Carballo et al., 2002). RA also activates survival signaling inneuroblastoma (Paillaud et al., 2002; Lee et al., 2006), promotingcell survival and reducing cell susceptibility to neurotoxins(Cavanaugh et al., 2006; Fernandez-Gomez et al., 2006). Therefore,we hypothesize that RA-mediated differentiation affects cellsusceptibility to neurotoxicity study in PD research.

While RA-differentiated SH-SY5Y cell model has long been usedfor studies in neuroscience, undifferentiated SH-SY5Y has also beenchosen as model cell line (Levites et al., 2002b; Levites et al., 2003;Xue et al., 2006; Lee et al., 2006). However, a report has shown thatundifferentiated SK-N-SH cells (the sister cell line of SH-SY5Y cells)do not exhibit significant difference in their neuronal propertiescompared to that of the RA-differentiated SK-N-SH cells (Lombetet al., 2001). This raises a question on choosing betweendifferentiated and undifferentiated cells in neuroscience research.It is still controversial about the necessity of differentiating SH-SY5Ycells with RA. Therefore, we aim to re-evaluate the changes ofneuronal properties of human neuroblastoma SH-SY5Y cells upon

Y.-T. Cheung et al. / NeuroToxicology 30 (2009) 127–135128

RA-induced differentiation and cellular responses to differentneurotoxins with or without differentiation.

To address this question, we compared immunoreactivity ofdifferent neuronal markers in undifferentiated and RA-differ-entiated SH-SY5Y cells by Western-blot and immunocytochem-ical analysis. Next, we determined intracellular signaling alteredby RA by Western-blot analysis. Finally, we examined thedifferential responses to 6-OHDA and MPP+. The results haveimplication in using RA-differentiated SH-SY5Y cells for in vitro

study of neurotoxicity.

2. Experimental procedures

2.1. Materials

6-OHDA was purchased from Sigma (Saint Louis, USA) whereasMPP+ was purchased from RBI (Wayland, MA, USA). Materials usedfor neuronal cell culture were purchased from Gibco-BRL(Invitrogen, NY, USA). Other chemicals used in this study werepurchased from companies listed as follow: RA, dimethylsulphoxide (DMSO), Triton X-100, 1,4-dithiotreitol (DTT), paraf-ormaldehyde, protease inhibitor cocktail, phosphatase inhibitorcocktail, Tween-20, Temed, 30% acrylamide, mouse anti-b-actinmonoclonal antibody and rabbit anti-SAP97 from Sigma; cyto-toxicity detection kit (LDH) and cell proliferation kit (MTT) fromRoche Diagnostics (Mannheim, Germany); colorimetric caspase-3substrate (Ac-DEVD-pNA) from Calbiochem, Inc. (La Jolla, CA, USA);caspase-3 activity kit from Biosource (Camarillo, CA, USA); AlexaFluor-488-conjugated goat anti-mouse IgG antibody, Alexa Fluor-488-conjugated goat anti-rabbit IgG antibody and prolong goldantifade reagent from Invitrogen (Molecular Probes, Oregon, USA);mouse anti-neuronal nuclei (NeuN), mouse anti-neurofilamentand mouse anti-human gamma neuron specific enolase (NSE),mouse anti-synaptophysin, rabbit anti-tyrosine hydroxylase fromChemicon (Temecula, CA, USA); rabbit polyclonal antibodies forJNK, phospho-JNK, Akt, phospho-Akt, Erk, phospho-Erk, phospho-PKC, mTOR, phospho-mTOR, MAP2, Akt inhibitor LY600124 andErk1/2 inhibitor U0126 from Cell Signaling Technology (Beverly,MA, USA); rat anti-dopamine transporter (DAT) from Abcam(Cambridge, MA, USA); rabbit anti-inhibitor of DNA binding 1 (Id1)from Santa Cruz (CA, USA); horseradish peroxidase-conjugatedgoat anti-rabbit and goat anti-mouse antibodies from DAKO(Glostrup, Denmark); PVDF membrane and protein assay kit werefrom Bio-Rad (Richmond, CA, USA); Biomax X-ray film from Kodak(Tokyo, Japan); enhanced chemiluminescence (ECL) detection kitfrom Amersham (Buckinghamshire, UK).

2.2. Cell culture

The procedures for growing SH-SY5Y cells has been describedelsewhere (Chang et al., 2002; Suen et al., 2003). Briefly, SH-SY5Ycells were cultured with 10% complete medium (minimumessential medium, MEM, 10% heat inactivated fetal bovine serum,L-glutamine (2 mM), penicillin (50 U/ml), streptomycin (50 mg/ml)) in a humidified, 5% CO2, 37 8C incubator. Forty-eight hourafter seeding, serum levels of the medium were reduced to 3%with RA (10 mM) for differentiation for seven days prior totreatment. For Western-blot analysis, LDH and caspase-3 activityassay, differentiated cells at 2 � 105 cells/well were culturedonto 6-well plate. For MTT assay, differentiated cells at 1 � 104

cells/well were cultured onto 96-well plate. For immunofluor-escent staining, differentiated cells at 2.5 � 104 cells/well werecultured onto 4-well chamber slide. Cells were cultured ontoculture plate without coating. For comparison between undiffer-entiated and differentiated SH-SY5Y cells, sister cultures wereplated at the same time.

2.3. Treatments

All treatments were performed under dark condition unlessotherwise stated. A final concentration of 6-OHDA (25 mM) orMPP+ (1 mM) was used to treat SH-SY5Y cells with or withoutdifferentiation. To investigate the role of Akt in attenuating 6-OHDA neurotoxicity after differentiation, undifferentiated or RA-differentiated SH-SY5Y cells were pre-treated with LY600124(5 mM) for 1 h, before subsequent treatment of 6-OHDA (25 mM).

2.4. Measurement of the release of LDH assay

General toxicity was measured by LDH assay. The procedures ofthe assay were followed according to the methods publishedelsewhere (Suen et al., 2003; Yu et al., 2005; Yu et al., 2006a; Hoet al., 2007). Briefly, culture medium was collected after treatmentfor the assay. The reaction underwent in dark environment for30 min prior to measurement. Changes in absorbance weremeasured at 492 nm by a multiplate reader (Labsystem). Resultswere expressed as percentage of control.

2.5. Measurement of 3-(3,4-dimethylthiazol-2-yl)-2,5-

diphenyltetrazolium bromide (MTT) assay

Cell viability was determined by a mitochondria enzyme-dependent reaction of MTT as described elsewhere (Fang et al.,2005). Briefly, MTT was added to SH-SY5Y cells after treatment in96-well plates. Metabolic active cells cleaved the yellow tetra-zolium salt MTT to purple formazan crystals. The formazan formedwas solubilized and the absorbance was measured by a multiplatereader at 570 nm. Results were expressed as percentage of control.

2.6. Measurement of caspase-3-like activity assay

Apoptosis was determined by caspase-3-like activity assay. Theprocedures of caspase-3-like activity assay has been describedelsewhere (Suen et al., 2003; Lin et al., 2004; Lai et al., 2006; Yuet al., 2006b; Yu et al., 2007b). Briefly, cellular proteins wereharvested in lysis buffer after treatment. Proteins were separatedby centrifugation at 14,000 g for 30 min at 4 8C. Supernatant wascollected and protein concentration was determined by proteinassay kit (Bio-Rad). Equal amount of protein from each sample wasincubated with caspase-3 substrate for 2 h at 37 8C. The caspase-3-like activity was determined by the absorbance at 405 nm of theyellowish product (pNA) cleaved from the substrate. Specificactivity (s.a., unit = pmol/min/mg) were calculated and reported intext. Results were expressed as percentage of control.

2.7. Western-blot analysis

Procedures of Western-blot were described elsewhere (Yu et al.,2004; Yu et al., 2007a; Lai et al., 2008). After treatment, SH-SY5Ycells were harvested and lysed in ice-cold lysis buffer containingTris (10 mM, pH 7.4), NaCl (100 mM), EDTA (1 mM), EGTA (1 mM),NaF (1 mM), Na4P2O7 (20 mM), Na3VO4 (2 mM), Triton X-100 (1%),glycerol (10%), SDS (0.1%), deoxycholate (0.5%). Phenylmethylsul-fonyl fluoride (1 mM), protease inhibitor cocktail, and phosphataseinhibitor cocktail were added. Protein extracts (50 mg) wereseparated in 6 or 12.5% SDS-PAGE gel and then transferred onto aPVDF membrane. The membrane was blocked by 5% non-fat drymilk with BSA in Tris-buffered saline (pH 7.4) containing 0.1%Tween-20. It was then incubated with rabbit anti-Akt (1:1000dilution), rabbit anti-phosphorylated Akt at serine 473 (1:1000dilution), rabbit anti-phosphorylated PKCpan (1:1000 dilution),rabbit anti-JNK (1:1000 dilution), rabbit anti-phosphorylatedJNK (1:1000 dilution), rabbit anti-Erk1/2 (1:1000 dilution), rabbit

Y.-T. Cheung et al. / NeuroToxicology 30 (2009) 127–135 129

anti-phosphorylated Erk1/2 (1:1000 dilution), rabbit anti-mTOR(1:1000 dilution), rabbit anti-phosphorylated mTOR (1:1000dilution), rabbit anti-Id1 (1:1000 dilution), mouse anti-humanNSE (1:500 dilution), mouse anti-NF (1:1000 dilution), rabbit anti-MAP2 (1:1000 dilution), mouse anti-NeuN (1:500 dilution), rabbitanti-Th (1:500 dilution), rat anti-DAT (1:500 dilution), mouse anti-synaptophysin (1:5000 dilution), rabbit anti-SAP97 (1:1000dilution), or mouse anti-b-actin (1:5000 dilution) for 4 h at roomtemperature. Subsequently the membrane was incubated withhorseradish peroxidase-conjugated secondary antibodies (1:1000,1:2000 or 1:10000 dilution) for 45 min at room temperature.Bands were visualized on a Biomax X-ray film (Kodak) using anenhanced chemiluminescence (ECL) kit.

2.8. Immunocytochemical analysis of neuronal specific markers

Effects on neuronal characteristics by differentiation wereinvestigated by immunocytochemical analysis of neuronalmarkers. The staining procedures were based on previouspublication with modification (Chang et al., 2000; Chang et al.,2001). Briefly, cells were washed with Tris-buffered salinefollowed by fixation in 4% paraformaldehyde on ice or methanolin �20 8C for 20 min. Non-specific binding of antibody wasblocked by 1% BSA in Tris-buffered saline. The cells weresubsequently incubated with mouse anti-NF (1:200 dilution),mouse anti-human NSE (1:200 dilution), rabbit anti-Th (1:200dilution), mouse anti-synaptophysin (1:400 dilution) or rabbitanti-SAP97 (1:200 dilution) at 4 8C overnight. Cells were thenincubated with Alexa Fluor-488-conjugated goat anti-mouse oranti-rabbit IgG antibody. Fluorescent intensity was examinedunder a confocal microscope (Bio-Rad, Radiance2000). Laserscanning was performed under a 600�magnification to measurethe average fluorescent intensity. Photos were taken from threerandomly selected fields for each experiment and at least threeindependent experiments were performed.

2.9. Measurement of nitro blue tetrazolium (NBT) assay

Generation of intracellular reactive oxygen species wasdetermined by a NBT as described elsewhere (Vrablic et al.,2001). Briefly, 200 ml NBT (1.0 mg/ml) was added to the medium ofSH-SY5Y cells at the end of the treatment period, followed byadditional incubation of 2 h at 37 8C. Intracellular superoxide anioncatalyzed the conversion of NBT to purple formazan. Metabolicactive cells cleaved the yellow tetrazolium salt MTT to purpleformazan crystals. Cells were washed with PBS, and then harvestedwith 100 ml DMSO. The lysates were then dissolved in 100 ml 2 MKOH, and the absorbance at 570 nm was determined by usingspectrophotometric method.

2.10. Statistical analysis

The results are expressed by mean � S.E.M. from at least threeindependent experiments. For statistical comparisons, quantitativedata was analyzed by one-way analysis of variance (ANOVA) followedby Tukey-test according to the statistical program SigmaStat1 (JandelScientific, Chicago, IL, USA). A p-value less than 0.05 was regarded assignificant.

3. Results

3.1. Differentiation properties of retinoic acid

To demonstrate that SH-SY5Y cells can be differentiated by RAas in other reports, we first showed that extension of neurites, atypical neuronal phenotype, was observed 24 h after application of

RA (data not shown) and such phenomenon was retained till day 7(Fig. 1B), while undifferentiated cells maintained relatively shortneurites (Fig. 1A). Protein expression of Id1 was significantlyreduced after differentiation by RA, as revealed by Western-blot(Fig. 1C) and immunocytochemical analysis (Fig. 1D). Moreover,without differentiation, Id1 was shown to distribute throughoutthe cell but re-localized in cytoplasm after RA-induced differentia-tion (Fig. 1D).

3.2. Western-blot analysis on the changes in neuronal

marker levels by RA

After showing successful differentiation by RA, we next soughtto examine the effects of RA on neuronal properties. We chose toinvestigate neuronal markers including NSE, synaptophysin, post-synaptic associated protein-97 (SAP97), NeuN, NF, MAP2 anddopaminergic neuronal markers such as DAT and Th. Protein levelsof neuronal markers in undifferentiated and RA-differentiated SH-SY5Y cells were compared by Western-blot analysis (Fig. 2). Weobserved a significant increase in the expression of NSE,synaptophysin, SAP97 and NeuN in RA-differentiated cells. Therewas no significant change in the expression of NF, MAP2 and thedopaminergic neuronal markers DAT and Th between undiffer-entiated and RA-differentiated cells.

3.3. Immunocytochemical analysis on the changes in neuronal

marker levels by RA

To confirm the changes in the expression of neuronal markersrevealed by Western-blot analysis, immunocytochemical analysiswas then performed. Fluorescent intensity of NSE, synaptophysin,SAP97, NF, and Th was compared between undifferentiated andRA-differentiated SH-SY5Y (Fig. 3). In undifferentiated SH-SY5Y,NSE immunoreactivity was mainly detected in the cytoplasm. Afterdifferentiation, there was a significant increase in fluorescentintensity of NSE, both in soma and neurites. Significant increases influorescent intensity in soma and neurities after differentiationwere also observed in synaptophysin and SAP97. In agreementwith the results from Western-blot, no significant change wasobserved in fluorescent intensity of NF and Th (Fig. 3).

3.4. Activation of survival signaling after differentiation by

retinoic acid

To elucidate whether RA regulates survival signaling path-ways, the phosphorylation of Akt, mTOR, Erk1/2 and PKC in SH-SY5Y cells was examined by Western-blot analysis (Fig. 4A). Weobserved a marked increase in the phosphorylation of Akt atserine 473 after differentiation. However, there was no sig-nificant change in its downstream target mammalian target ofrapamycin (mTOR). Furthermore, RA stimulated an increasein Erk 1/2 phosphorylation and reduced the phospho-PKC levelin SH-SY5Y cells.

3.5. Activation of c-Jun N-terminal kinase (JNK) pathway after

differentiation by retinoic acid

Recent reports have shown that JNK pathway is involved inRA-mediated differentiation (Yu et al., 2003). Therefore, weinvestigated the effects of RA on JNK pathway by Western-blotanalysis (Fig. 4B). In undifferentiated SH-SY5Y, phosphorylatedform of both JNK1 and JNK2 was barely detectable. However,phospho-JNK2 level was markedly increased 7 days afterdifferentiation, while the increase in phospho-JNK1 was mar-ginal. The phosphorylation state of the downstream c-Jun wasalso increased.

Fig. 1. Differentiation properties of retinoic acid. (A) Undifferentiated SH-SH5Y cells cultured in 10% complete medium for 3 days. (B) Differentiated SH-SY5Y cells cultured

with RA (10 mM) for 7 days. Scale bar = 50 mm. (C) Western-blot analysis of Id1 expression in SH-SY5Y cells with (RA) or without (Control) RA-differentiation. Numbers 1–3

indicated three independent experiments. (D) Immunocytochemical analysis of Id1 expression in SH-SY5Y cells with (RA) or without (Control) RA-differentiation.

Y.-T. Cheung et al. / NeuroToxicology 30 (2009) 127–135130

3.6. Differentiation effects of retinoic acid on the susceptibility

of SH-SY5Y cells to 6-OHDA-induced neurotoxicity

We have demonstrated that RA-induced differentiation reg-ulates both survival and stress signaling. Therefore, we sought toexamine whether this also modulates the susceptibility of SH-SY5Ycells to neurotoxins. Cells with or without RA-induced differentia-

Fig. 2. Western-blot analysis of changes in expression of neuronal markers.

Expression of NSE, synaptophysin, SAP97, NeuN, NF, MAP2, Th and DAT in SH-SY5Y

cells with (RA) or without (Control) RA-differentiation. Numbers 1–3 indicates

three independent experiments.

tion were exposed to 6-OHDA (25 mM) for 24 h and then assayedfor cell viability by LDH release, MTT assay and caspase-3-likeactivity. Release of LDH triggered by 6-OHDA was reduced by twofolds in differentiated group (Fig. 5A). We also observed acorresponding increase in cell viability. After differentiation,MTT assay showed a near 0.3-fold increase in the number ofviable cells compared to undifferentiated cells 24 h after exposureto 6-OHDA (Fig. 5B). In addition, the caspase-3-like activity wasreduced by 0.5-fold in differentiated group (Fig. 5C).

3.7. Differentiation effects of retinoic acid on the susceptibility

of SH-SY5Y cells to MPP+-induced neurotoxicity

The differentiation effects on another Parkinsonism mimetictoxin MPP+ were also investigated. Similarly, SH-SY5Y cells with orwithout differentiation were exposed to MPP+ (1 mM) for 24 h andthen assayed for cell viability by LDH release, MTT assay andcaspase-3-like activity. A significant reduction of nearly 0.5-fold inMPP+-triggered LDH release was observed after differentiation(Fig. 6A). Accordingly, MTT assay showed a 0.1-fold increase indifferentiated cells after challenged by MPP+ (Fig. 6B). MPP+-induced caspase-3-like activity was significantly reduced by 0.7-fold after differentiation with RA (Fig. 6C).

3.8. Effect of retinoic acid on the generation of 6-OHDA-induced

reactive oxygen species in SH-SY5Y cells

6-OHDA can induce oxidative stress by generating reactiveoxygen species (Soto-Otero et al., 2008). We sought to investigate

Fig. 3. Immunocytochemical analysis of neuronal markers in SH-SY5Y cells with (RA) or without (Control) RA-differentiation. Changes in expression of NeuN, SAP97, NSE,

synaptophysin, NF and Th were determined by their fluorescent intensity using confocal microscope. Scale bar = 50 mm.

Y.-T. Cheung et al. / NeuroToxicology 30 (2009) 127–135 131

Fig. 4. Western-blot analysis of survival signaling molecules and JNK pathway in

SH-SY5Y cells with (RA) or without (Control) RA-differentiation. (A) Activation of

Akt, Erk1/2, mTOR and PKC were detected by their phosphorylation state. (B)

Activation of JNK1/2 and c-Jun were detected by their phosphorylation state.

Numbers 1–3 indicates three independent experiments.

Fig. 5. Changes in cells susceptibility to 6-OHDA-induced neurotoxicity after

differentiation by retinoic acid (RA). SH-SY5Y cells with (RA) or without (Control)

RA-differentiation were exposed to 6-OHDA (25 mM) for 24 h. (A) General

cytotoxicity measured by the release of LDH. (B) Cell viability measured by MTT

assay. (C) Apoptotic cells measured by caspase-3-like activity assay. Results are

expressed in percentage of control (mean � S.E.M., n � 3, *p < 0.05, **p < 0.001

[Control vs 6-OHDA treated group]; #p < 0.05, ##p < 0.001 [undifferentiated vs RA-

differentiated group]).

Y.-T. Cheung et al. / NeuroToxicology 30 (2009) 127–135132

the effect of retinoic acid differentiation on 6-OHDA-inducedoxidative stress. By NBT assay, we observed a significant increasein the production of intracellular reactive oxygen species in SH-SY5Y cells after treatment of 6-OHDA. In RA-differentiated SH-SY5Y cells, there was a slight decrease in oxidative stress in both 6-OHDA and MPP+ treated group. However, such effect was notstatistically significant (Fig. 7).

3.9. Attenuation of the RA-mediated neuroprotection in response

to 6-OHDA toxicity by inhibiting Akt but not Erk1/2

We have shown that RA up-regulated Akt signaling, so wesought to examine whether Akt inhibitor would attenuate RA-mediated protection to neurotoxins. Undifferentiated or RA-differentiated SH-SY5Y cells were exposed to Akt inhibitorLY600124 (5 mM) for 1 h prior to 6-OHDA (25 mM) treatmentfor 24 h. Neurotoxicity in term of apoptosis was measured bycaspase-3-like activity assay (Fig. 8). Inhibitor alone did not causesignificant change in caspase-3-like activity. 6-OHDA causedmarked increase in caspase-3-like activity in both undifferentiated(1.53 folds) and RA-differentiated cells (1.25 folds). Pre-treatmentof Akt inhibitor exaggerated the toxicity of 6-OHDA by showing a0.5-fold increase in undifferentiated group and 1-fold increase indifferentiated group. More importantly, RA-mediated neuro-protection against 6-OHDA was removed, as pre-treatment ofinhibitor led to a comparable toxicity in both differentiated andundifferentiated cells.

4. Discussion

The study demonstrates that RA-induced differentiation of SH-SY5Y cells modulates their responsiveness to neurotoxins byaltering survival signaling pathways. Since both differentiatedand undifferentiated SH-SY5Y cells have been widely used in

neuroscience research, our study has high implication in using thiscell line for neurotoxicity and neurotoxin studies.

The differentiation property of RA is well established (Goodallet al., 1997; Tosetti et al., 1998; Guarneri et al., 2000; Hashemiet al., 2003). It has been stated that neuroblastoma cells have to bedifferentiated in vitro for at least 7 days for experiment (Sarkanenet al., 2007). We re-confirm the differentiation property of RA in

Fig. 6. Changes in cells susceptibility to MPP+-induced neurotoxicity after

differentiation by retinoic acid (RA). SH-SY5Y cells with (RA) or without

(Control) RA-differentiation were exposed to MPP+ (1 mM) for 24 h. (A) General

cytotoxicity measured by the release of LDH. (B) Cell viability measured by MTT

assay. (C) Apoptotic cells measured by caspase-3-like activity assay. Results are

expressed in percentage of control (mean � S.E.M., n � 3, *p < 0.05, **p < 0.001

[Control vs MPP+ treated group]; #p < 0.05, ##p < 0.001 [undifferentiated vs RA-

differentiated group]).

Fig. 7. Changes in neurotoxin-induced intracellular oxidative stress after

differentiation by retinoic acid (RA). SH-SY5Y cells with (RA) or without

(Control) RA-differentiation were exposed to 6-OHDA (25 mM) or MPP+ (1 mM)

for 24 h. Generation of reactive oxygen species measured by NBT. Results are

expressed in percentage of control (mean � S.E.M., n � 3, *p < 0.05).

Fig. 8. Role of survival signaling pathways in RA-mediated neuroprotection against

6-OHDA-induced apoptosis. Cell were pre-treated with LY600124 (5 mM) for 1 h

subsequent to 6-OHDA (25 mM) for 24 h. Caspase-3-like activity was examined.

Results are expressed in percentage of control (mean � S.E.M., n � 3).

Y.-T. Cheung et al. / NeuroToxicology 30 (2009) 127–135 133

SH-SY5Y cells by investigating the changes in morphology,differentiation markers and neuronal markers. Morphologically,we observed extensive neurites outgrowth as a typical neuronalphenotype in SH-SY5Y cells by RA (Fig. 1B), which is similar toother reports (Vesanen et al., 1994; Clagett-Dame et al., 2006). RA-differentiation showed a down-regulation and re-localization inId1, which has been demonstrated both by our result and others(Lopez-Carballo et al., 2002).

Previously, Lombet and co-workers have demonstrated nosignificant change in expression of gamma subunit of NSE after RA-

mediated differentiation on SK-N-SH cells, the sister cell line of SH-SY5Y cells (Lombet et al., 2001). However, in our system, weobserved a significant increase in NSE protein expression after RA-mediated differentiation by both immunocytochemical andWestern-blot analyses. The difference may be due to the lowconcentration of RA (3 mM) that they used for differentiationcompared to our system.

To address whether RA-differentiated SH-SY5Y cells arefunctionally mature, protein expression of pre-synaptic proteinsynaptophysin has been studied. Synaptophysin is an abundantintegral membrane protein on synaptic vesicles involved in therelease of neurotransmitters (Rehm et al., 1986). It has beenreported to be re-localized after differentiation by RA andcholesterol (Sarkanen et al., 2007). Similarly, we observed anincrease in synaptophysin expression along neurites after differ-entiation by RA.

We showed that RA-mediated differentiation affects neuronalproperties by further regulating other well established neuronalmarkers. Among the studied protein markers such as NeuN, NF,

Y.-T. Cheung et al. / NeuroToxicology 30 (2009) 127–135134

MAP2 and dopaminergic neuronal markers DAT and Th, NeuN isexpressed in neurons upon maturation neuron and thereforeserved as a marker for mature neurons (Mullen et al., 1992; Weyerand Schilling, 2003). In agreement with what have been reported,we found that RA-differentiation increased the expression ofNeuN. Protein levels of NF and MAP2 have been reported toincrease in short period (3–6 days) of RA-mediated differentiation(Sharma et al., 1999; Li et al., 2000; Pan et al., 2005). However,Encinas and co-workers obtained contrasting observations byshowing a comparable expression pattern of NF and MAP2between RA-differentiated and undifferentiated SH-SY5Y cells(Encinas et al., 2000). In agreement to their results, we did notobserve significant change in the expression of NF and MAP2 withor without differentiation.

SH-SY5Y cells express dopaminergic markers so that they arewidely used in PD research (Kitao et al., 2007; Kyratzi et al., 2007;Sung et al., 2007). We showed that the expression of DAT and Th isindependent to the differentiation by RA. Together, our resultssuggest that there are limited changes in neuronal markerexpressions after RA-mediated differentiation of SH-SY5Y cells.

The signaling involved in cell differentiation has been widelystudied (Lopez-Carballo et al., 2002; Takeda and Ichijo, 2002; Leeand Kim, 2004; Moran et al., 2005; Song et al., 2005; Fontan-Gabaset al., 2007). Regarding the signaling involved in RA-mediateddifferentiation, PI3k/Akt, JNK and p35/cdk5 have been reported toplay important roles in neurite outgrowth, cell cycle arrest anddifferentiation (Yu et al., 2003; Miloso et al., 2004). We also showedthat RA-differentiation regulates some of these proteins. Interest-ingly, although mTOR is one important downstream protein ofPI3K/Akt pathway, our results demonstrated that mTOR does notinvolve in RA-mediated differentiation process. Instead, weobserved a significant decrease of active PKC after differentiationby RA, which has been shown to negatively regulate Akt signalingpathway in mouse keratinocytes (Li et al., 2006). Furthermore,inhibition of PKC has been reported to promote neuronal cellssurvival through Akt-dependent pathway (Zhu et al., 2004).

Apart from cell differentiation, RA can trigger survival signalingin different cell types (Paillaud et al., 2002; Lee et al., 2006). Thisleads to a suggestion that RA-mediated differentiation mayalso affect cellular response to neurotoxins. Our results show thatRA-differentiated cells are less susceptible to Parkinsonismmimetic than undifferentiated cells. The resistance to Parkinson-ism mimetic after differentiation by RA in our system was not dueto dopaminergic neuronal markers DAT as we did not observe anysignificant change in DAT expression by RA. We also found thatdifferentiation did not protect the cells from 6-OHDA-inducedoxidative stress but it increased the mitochondrial activity, asrevealed by NBT assay and MTT assay, respectively. However, thereare reports showing that up-regulation of survival signaling (Aktand Erk1/2) can protect neurons from Parkinsonism mimetic (Jiangand Yu, 2005; Cavanaugh et al., 2006; Fernandez-Gomez et al.,2006; Fujita et al., 2006; Huang et al., 2007). We focus on the role ofAkt in RA-mediated neuroprotection in response to 6-OHDA. Afterinhibition of Akt, caspase-3-like activity triggered by 6-OHDA inRA-differentiated cells is comparable to that of the undifferen-tiated cells, suggesting that Akt plays a role in RA-mediatedneuroprotection in response to 6-OHDA toxicity during differ-entiation.

Taken together, RA differentiates SH-SY5Y neuroblastoma cellsby altering specific neuronal markers. However, with the considera-tion that there is no significant change in dopaminergic propertiesand RA modulates the Akt pathway resulting in higher tolerance to6-OHDA toxicity, undifferentiated SH-SY5Y may be more appro-priate for studying neurotoxicity or neuroprotection in experi-mental PD research. Cautions should be taken if one attempts touse this cell type for investigating neurotoxic responses of drugs.

Conflict of interest statement

There is no potential conflict of interest or competing interest.

Acknowledgements

This work is supported by competitive earmarked researchgrant (7552/06M) and NSFC/RGC Joint Research Scheme(N_HKU 707/07 M) from Research Grant Council, Hong Kong,University Strategic Research Theme on Drug Discovery from TheUniversity of Hong Kong, and HKU Seed Funding for BasicResearch (200711159028).

References

Cavanaugh JE, Jaumotte JD, Lakoski JM, Zigmond MJ. Neuroprotective role of ERK1/2and ERK5 in a dopaminergic cell line under basal conditions and in response tooxidative stress. J Neurosci Res 2006;84:1367–75.

Chang RCC, Chen W, Hudson P, Wilson B, Han DS, Hong JS. Neurons reduce glialresponses to lipopolysaccharide (LPS) and prevent injury of microglial cells fromover-activation by LPS. J Neurochem 2001;76:1042–9.

Chang RCC, Hudson PM, Wilson BC, Liu B, Abel H, Hong JS. High concentrations ofextracellular potassium enhance bacterial endotoxin lipopolysaccharide-inducedneurotoxicity in glia-neuron mixed cultures. Neuroscience 2000;97:757–64.

Chang RCC, Suen KC, Ma CH, Elyaman W, Ng HK, Hugon J. Involvement of double-stranded RNA-dependent protein kinase and phosphorylation of eukaryotic initia-tion factor-2alpha in neuronal degeneration. J Neurochem 2002;83:1215–25.

Clagett-Dame M, McNeill EM, Muley PD. Role of all-trans retinoic acid in neuriteoutgrowth and axonal elongation. J Neurobiol 2006;66:739–56.

Encinas M, Iglesias M, Liu Y, Wang H, Muhaisen A, Cena V, et al. Sequential treatment ofSH-SY5Y cells with retinoic acid and brain-derived neurotrophic factor gives rise tofully differentiated, neurotrophic factor-dependent, human neuron-like cells. JNeurochem 2000;75:991–1003.

Fang X, Yu MM, Yuen WH, Zee SY, Chang RCC. Immune modulatory effects of Prunellavulgaris L. on monocytes/macrophages. Int J Mol Med 2005;16:1109–16.

Fernandez-Gomez FJ, Pastor MD, Garcia-Martinez EM, Melero-Fernandez dM. Gou-Fabregas M, Gomez-Lazaro M, et al. Pyruvate protects cerebellar granular cellsfrom 6-hydroxydopamine-induced cytotoxicity by activating the Akt signalingpathway and increasing glutathione peroxidase expression. Neurobiol Dis2006;24:296–307.

Fontan-Gabas L, Oliemuller E, Martinez-Irujo JJ, de Miguel C, Rouzaut A. All-trans-retinoic acid inhibits collapsin response mediator protein-2 transcriptional activityduring SH-SY5Y neuroblastoma cell differentiation. FEBS J 2007;274:498–511.

Fujita H, Ogino T, Kobuchi H, Fujiwara T, Yano H, Akiyama J, et al. Cell-permeable cAMPanalog suppresses 6-hydroxydopamine-induced apoptosis in PC12 cells throughthe activation of the Akt pathway. Brain Res 2006;1113:10–23.

Goodall AR, Danks K, Walker JH, Ball SG, Vaughan PF. Occurrence of two types ofsecretory vesicles in the human neuroblastoma SH-SY5Y. J Neurochem1997;68:1542–52.

Guarneri P, Cascio C, Piccoli T, Piccoli F, Guarneri R. Human neuroblastoma SH-SY5Ycell line: neurosteroid-producing cell line relying on cytoskeletal organization. JNeurosci Res 2000;60:656–65.

Hashemi SH, Li JY, Ahlman H, Dahlstrom A. SSR2(a) receptor expression and adrener-gic/cholinergic characteristics in differentiated SH-SY5Y cells. Neurochem Res2003;28:449–60.

Ho YS, Yu MS, Lai CS, So KF, Yuen WH, Chang RCC. Characterizing the neuroprotectiveeffects of alkaline extract of Lycium barbarum on beta-amyloid peptide neurotoxi-city. Brain Res 2007;1158:123–34.

Huang HY, Lin SZ, Kuo JS, Chen WF, Wang MJ. G-CSF protects dopaminergic neuronsfrom 6-OHDA-induced toxicity via the ERK pathway. Neurobiol Aging2007;28:1258–69.

Jiang Z, Yu PH. Involvement of extracellular signal-regulated kinases 1/2 and (phos-phoinositide 3-kinase)/Akt signal pathways in acquired resistance against neuro-toxin of 6-hydroxydopamine in SH-SY5Y cells following cell-cell interaction withastrocytes. Neuroscience 2005;133:405–11.

Kedar NP. Can we prevent Parkinson’s and Alzheimer’s disease? J Postgrad Med2003;49:236–45.

Kitao Y, Matsuyama T, Takano K, Tabata Y, Yoshimoto T, Momoi T, et al. Does ORP150/HSP12A protect dopaminergic neurons against MPTP/MPP(+)-induced neurotoxi-city? Antioxid Redox Signal 2007;9:589–95.

Kyratzi E, Pavlaki M, Kontostavlaki D, Rideout HJ, Stefanis L. Differential effects ofParkin and its mutants on protein aggregation, the ubiquitin-proteasome system,and neuronal cell death in human neuroblastoma cells. J Neurochem2007;102:1292–303.

Lai CS, Yu MS, Yuen WH, So KF, Zee SY, Chang RCC. Antagonizing beta-amyloid peptideneurotoxicity of the anti-aging fungus Ganoderma lucidum. Brain Res2008;1190:215–24.

Lai SW, Yu MS, Yuen WH, Chang RCC. Novel neuroprotective effects of the aqueousextracts from Verbena officinalis Linn. Neuropharmacology 2006;50:641–50.

Lee JH, Kim KT. Induction of cyclin-dependent kinase 5 and its activator p35 throughthe extracellular-signal-regulated kinase and protein kinase A pathways during

Y.-T. Cheung et al. / NeuroToxicology 30 (2009) 127–135 135

retinoic-acid mediated neuronal differentiation in human neuroblastoma SK-N-BE(2)C cells. J Neurochem 2004;91:634–47.

Lee JH, Shin SY, Kim S, Choo J, Lee YH. Suppression of PTEN expression duringaggregation with retinoic acid in P19 mouse embryonal carcinoma cells. BiochemBiophys Res Commun 2006;347:715–22.

Levites Y, Amit T, Mandel S, Youdim MB. Neuroprotection and neurorescue againstAbeta toxicity and PKC-dependent release of nonamyloidogenic soluble precursorprotein by green tea polyphenol (�)-epigallocatechin-3-gallate. FASEB J2003;17:952–4.

Levites Y, Amit T, Youdim MB, Mandel S. Involvement of protein kinase C activation andcell survival/cell cycle genes in green tea polyphenol (�)-epigallocatechin 3-gallate neuroprotective action. J Biol Chem 2002a;277:30574–80.

Levites Y, Youdim MB, Maor G, Mandel S. Attenuation of 6-hydroxydopamine (6-OHDA)-induced nuclear factor-kappaB (NF-kappaB) activation and cell death bytea extracts in neuronal cultures. Biochem Pharmacol 2002b;63:21–9.

Li BS, Zhang L, Gu J, Amin ND, Pant HC. Integrin alpha(1) beta(1)-mediated activation ofcyclin-dependent kinase 5 activity is involved in neurite outgrowth and humanneurofilament protein H Lys-Ser-Pro tail domain phosphorylation. J Neurosci2000;20:6055–62.

Li L, Sampat K, Hu N, Zakari J, Yuspa SH. Protein kinase C negatively regulates Aktactivity and modifies UVC-induced apoptosis in mouse keratinocytes. J Biol Chem2006;281:3237–43.

Lin KF, Chang RCC, Suen KC, So KF, Hugon J. Modulation of calcium/calmodulin kinase-IIprovides partial neuroprotection against beta-amyloid peptide toxicity. Eur JNeurosci 2004;19:2047–55.

Lombet A, Zujovic V, Kandouz M, Billardon C, Carvajal-Gonzalez S, Gompel A, et al.Resistance to induced apoptosis in the human neuroblastoma cell line SK-N-SH inrelation to neuronal differentiation. Role of Bcl-2 protein family. Eur J Biochem2001;268:1352–62.

Lopez-Carballo G, Moreno L, Masia S, Perez P, Barettino D. Activation of the phospha-tidylinositol 3-kinase/Akt signaling pathway by retinoic acid is required for neuraldifferentiation of SH-SY5Y human neuroblastoma cells. J Biol Chem 2002;277:25297–304.

Miloso M, Villa D, Crimi M, Galbiati S, Donzelli E, Nicolini G, et al. Retinoic acid-inducedneuritogenesis of human neuroblastoma SH-SY5Y cells is ERK independent andPKC dependent. J Neurosci Res 2004;75:241–52.

Moran CM, Donnelly M, Ortiz D, Pant HC, Mandelkow EM, Shea TB. Cdk5 inhibitsanterograde axonal transport of neurofilaments but not that of tau by inhibitionof mitogen-activated protein kinase activity. Brain Res Mol Brain Res 2005;134:338–44.

Mullen RJ, Buck CR, Smith AM. NeuN, a neuronal specific nuclear protein in vertebrates.Development 1992;116:201–11.

Pahlman S, Ruusala AI, Abrahamsson L, Mattsson ME, Esscher T. Retinoic acid-induceddifferentiation of cultured human neuroblastoma cells: a comparison with phor-bolester-induced differentiation. Cell Differ 1984;14:135–44.

Paillaud E, Costa S, Fages C, Plassat JL, Rochette-Egly C, Monville C, et al. Retinoic acidincreases proliferation rate of GL-15 glioma cells, involving activation of STAT-3transcription factor. J Neurosci Res 2002;67:670–9.

Pan J, Kao YL, Joshi S, Jeetendran S, Dipette D, Singh US. Activation of Rac1 byphosphatidylinositol 3-kinase in vivo: role in activation of mitogen-activatedprotein kinase (MAPK) pathways and retinoic acid-induced neuronal differentia-tion of SH-SY5Y cells. J Neurochem 2005;93:571–83.

Rehm H, Wiedenmann B, Betz H. Molecular characterization of synaptophysin, a majorcalcium-binding protein of the synaptic vesicle membrane. EMBO J 1986;5:535–41.

Sarkanen JR, Nykky J, Siikanen J, Selinummi J, Ylikomi T, Jalonen TO. Cholesterolsupports the retinoic acid-induced synaptic vesicle formation in differentiatinghuman SH-SY5Y neuroblastoma cells. J Neurochem 2007;102:1941–52.

Segura-Aguilar J, Kostrzewa RM. Neurotoxins and neurotoxicity mechanisms, an over-view. Neurotox Res 2006;10:263–87.

Sharma M, Sharma P, Pant HC. CDK-5-mediated neurofilament phosphorylation inSHSY5Y human neuroblastoma cells. J Neurochem 1999;73:79–86.

Song JH, Wang CX, Song DK, Wang P, Shuaib A, Hao C. Interferon gamma inducesneurite outgrowth by up-regulation of p35 neuron-specific cyclin-dependentkinase 5 activator via activation of ERK1/2 pathway. J Biol Chem 2005;280:12896–901.

Soto-Otero R, Mendez-Alvarez E, Sanchez-Iglesias S, Zubkov FI, Ovskressensky LG,Varlamov AV, et al. Inhibition of 6-hydroxydopamine-induced oxidative damageby 4,5-dihydro-3H-2-benzazepine N-oxides. Biochem Pharmacol 2008;75:1526–37.

Suen KC, Lin KF, Elyaman W, So KF, Chang RCC, Hugon J. Reduction of calcium releasefrom the endoplasmic reticulum could only provide partial neuroprotectionagainst beta-amyloid peptide toxicity. J Neurochem 2003;87:1413–26.

Sung JY, Lee HJ, Jeong EI, Oh Y, Park J, Kang KS, et al. Alpha-synuclein overexpressionreduces gap junctional intercellular communication in dopaminergic neuroblas-toma cells. Neurosci Lett 2007;416:289–93.

Takeda K, Ichijo H. Neuronal p38 MAPK signalling: an emerging regulator of cell fateand function in the nervous system. Genes Cells 2002;7:1099–111.

Tosetti P, Taglietti V, Toselli M. Functional changes in potassium conductances of thehuman neuroblastoma cell line SH-SY5Y during in vitro differentiation. J Neuro-physiol 1998;79:648–58.

Vesanen M, Salminen M, Wessman M, Lankinen H, Sistonen P, Vaheri A. Morphologicaldifferentiation of human SH-SY5Y neuroblastoma cells inhibits human immuno-deficiency virus type 1 infection. J Gen Virol 1994;75(Pt 1):201–6.

Vrablic AS, Albright CD, Craciunescu CN, Salganik RI, Zeisel SH. Altered mitochondrialfunction and overgeneration of reactive oxygen species precede the induction ofapoptosis by 1-O-octadecyl-2-methyl-rac-glycero-3-phosphocholine in p53-defective hepatocytes. FASEB J 2001;15:1739–44.

Weyer A, Schilling K. Developmental and cell type-specific expression of the neuronalmarker NeuN in the murine cerebellum. J Neurosci Res 2003;73:400–9.

Xue S, Jia L, Jia J. Hypoxia and reoxygenation increased BACE1 mRNA and protein levelsin human neuroblastoma SH-SY5Y cells. Neurosci Lett 2006;405:231–5.

Yu MS, Ho YS, So KF, Yuen WH, Chang RCC. Cytoprotective effects of Lycium barbarumagainst reducing stress on endoplasmic reticulum. Int J Mol Med 2006a;17:1157–61.

Yu MS, Suen KC, Kwok NS, So KF, Hugon J, Chang RCC. Beta-amyloid peptides inducesneuronal apoptosis via a mechanism independent of unfolded protein responses.Apoptosis 2006b;11:687–700.

Yu MS, Lai CS, Ho YS, Zee SY, So KF, Yuen WH, et al. Characterization of the effects ofanti-aging medicine Fructus lycii on beta-amyloid peptide neurotoxicity. Int J MolMed 2007a;20:261–8.

Yu MS, Wong AY, So KF, Fang JN, Yuen WH, Chang RCC. New polysaccharide fromNerium indicum protects neurons via stress kinase signaling pathway. Brain Res2007b;1153:221–30.

Yu MS, Lai SW, Lin KF, Fang JN, Yuen WH, Chang RCC. Characterization of polysacchar-ides from the flowers of Nerium indicum and their neuroprotective effects. Int J MolMed 2004;14:917–24.

Yu MS, Leung SK, Lai SW, Che CM, Zee SY, So KF, et al. Neuroprotective effects of anti-aging oriental medicine Lycium barbarum against beta-amyloid peptide neuro-toxicity. Exp Gerontol 2005;40:716–27.

Yu YM, Han PL, Lee JK. JNK pathway is required for retinoic acid-inducedneurite outgrowth of human neuroblastoma, SH-SY5Y. Neuroreport 2003;14:941–5.

Zheng L, Roberg K, Jerhammar F, Marcusson J, Terman A. Autophagy of amyloid beta-protein in differentiated neuroblastoma cells exposed to oxidative stress. NeurosciLett 2006;394:184–9.

Zhu D, Jiang X, Wu X, Tian F, Mearow K, Lipsky RH, et al. Inhibition of protein kinase Cpromotes neuronal survival in low potassium through an Akt-dependent pathway.Neurotox Res 2004;6:281–9.