Báo cáo khoa học: All-trans-retinoic acid inhibits collapsin response mediator protein-2 transcriptional activity during SH-SY5Y neuroblastoma cell differentiation pptx

Identification of functional AP-2 and Pax-3 elements within the CRMP-2 promoter We further decided to clarify which transcription fac-tors were responsible for CRMP-2 transcriptional act

protein-2 transcriptional activity during SH-SY5Y

neuroblastoma cell differentiation

Lorena Fonta´n-Gaba´s1, Erik Oliemuller2, Juan Jose´ Martı´nez-Irujo1,2, Carlos de Miguel1and

Ana Rouzaut1,2

1 Department of Biochemistry, University of Navarra, Pamplona, Spain

2 Division of Oncology, Center for Applied Medical Research, University of Navarra, Pamplona, Spain

Neural cells migrate and differentiate during brain

development in a highly regulated fashion [1] During

this process, neurons orient their axons towards their

functional effectors in a coordinated way The family

of semaphorin proteins (previously called collapsins)

play a significant role in axonal pathfinding, through their action as chemorepellents [2] Collapsin response mediator protein-2 (CRMP-2), also known as dihydro-pyrimidinase-like 2, is a member of a family of cyto-plasmic proteins [3] that were originally identified as

Keywords

CRMP-2; gene regulation; neuroblastoma;

promoter; retinoic acid

Correspondence

A Rouzaut, Center for Applied Medical

Research, University of Navarra, Av Pı´o XII

55, 31008 Pamplona, Spain

Fax: +34 948 19 47 18

Tel: +34 948 19 47 00

E-mail: arouzaut@unav.es

(Received 14 July 2006, revised 9 October

2006, accepted 16 November 2006)

doi:10.1111/j.1742-4658.2006.05597.x

Neurons are highly polarized cells composed of two structurally and func-tionally distinct parts, the axon and the dendrite The establishment of this asymmetric structure is a tightly regulated process In fact, alterations in the proteins involved in the configuration of the microtubule lattice are fre-quent in neuro-oncologic diseases One of these cytoplasmic mediators is the protein known as collapsin response mediator protein-2, which interacts with and promotes tubulin polymerization In this study, we investigated collapsin response mediator protein-2 transcriptional regulation during all-trans-retinoic acid-induced differentiation of SH-SY5Y neuroblastoma cells All-trans-retinoic acid is considered to be a potential preventive and thera-peutic agent, and has been extensively used to differentiate neuroblastoma cells in vitro Therefore, we first demonstrated that collapsin response medi-ator protein-2 mRNA levels are downregulated during the differentiation process After completion of deletion construct analysis and mutagenesis and mobility shift assays, we concluded that collapsin response mediator protein-2 basal promoter activity is regulated by the transcription factors AP-2 and Pax-3, whereas E2F, Sp1 and NeuroD1 seem not to participate

in its regulation Furthermore, we finally established that reduced expres-sion of collapsin response mediator protein-2 after all-trans-retinoic acid exposure is associated with impaired Pax-3 and AP-2 binding to their con-sensus sequences in the collapsin response mediator protein-2 promoter Decreased attachment of AP-2 is a consequence of its accumulation in the cytoplasm On the other hand, Pax-3 shows lower binding due to all-trans-retinoic acid-mediated transcriptional repression Unraveling the molecular mechanisms behind the action of all-trans-retinoic acid on neuroblastoma cells may well offer new perspectives for its clinical application

Abbreviations

ActD, actinomycin D; ATRA, all-trans-retinoic acid; CRMP-2, collapsin response mediator protein-2; EMSA, electrophoretic mobility shift assay.

mediators of semaphorin-induced growth cone collapse

[4] This protein is abundant in the distal part of the

rising axon [5], where it interacts with tubulin

heterodi-mers, allowing its polymerization in such a way that it

seems crucial for axonal growth and for the

determin-ation of axon–dendrite fate Moreover, overexpression

of CRMP-2 induces the formation of multiple axons,

whereas a dominant negative mutant of CRMP-2

inhibits the formation of the primary axon,

compromi-sing normal cell development [6]

The molecular mechanisms through which CRMP-2

associates preferentially with tubulin heterodimers have

recently been described [7,8] CRMP-2 is sequentially

phosphorylated at several serine and threonine

resi-dues, leading to a reduced affinity for tubulin dimers

However, CRMP-2 can regulate axonal growth

through different mechanisms; the data available in the

literature point to a panoply of CRMP-2-interacting

proteins, such as Numb, a protein associated with

endocytosis and membrane recycling [9] It has also

been shown that CRMP-2 can interact with actin in

a phosphorylation status-independent manner [10]

Therefore, CRMP-2 seems to be involved in neuronal

differentiation, modulating cytoskeletal organization

and endocytosis of neuronal cell adhesion molecules at

the growth cone

As CRMP-2 activity is mainly controlled through

protein phosphorylation, it might well be assumed that

its transcriptional regulation is not of physiologic

rele-vance Nevertheless, there are differences in its mRNA

expression levels in different tissues and developmental

stages [11,12] Also, deregulated CRMP-2 mRNA

expression has been related to several important

neuro-logic disorders, such as Alzheimer’s disease and

schizo-phrenia [13,14] Furthermore, differences in CRMP-2

mRNA and phosphorylation levels between normal

and neuroblastoma cells have been recently reported

by Tahimic et al [15] A relationship between altered

transcriptional regulation of CRMP-2 expression and

neuro-oncologic maladies seems possible, as cell

polari-zation is one of the main points in neuronal cell

differ-entiation, and involution towards dedifferentiated

phenotypes is one of the main events that contribute

to the settlement of some of these pathologies [16] In

fact, members of the CRMP family of proteins have

been reported to act as tumor and metastasis

suppres-sors: CRMP-1 for lung cancer [17], and CRMP-5 in

paraneoplastic disorders [18]

Therefore, we considered it relevant to study the

regu-lation of CRMP-2 gene expression during

all-trans-reti-noic acid (ATRA)-induced differentiation of SH-SY5Y

neuroblastoma cells We opted for ATRA as

differenti-ating agent because it has been widely employed in both

in vitro and in vivo studies on neuroblastoma and has proven useful as a coadjuvant in chemotherapy [19]

In this article, we present data supporting reduced CRMP-2 expression during neuroblastoma cell differ-entiation We provide strong experimental evidence for the involvement of the Pax-3 and AP-2 transcription factors in CRMP-2 transcriptional regulation Both of these have been widely reported to be implicated in neuronal cell development Finally, we have estab-lished that CRMP-2 transcriptional repression during this process is mediated through impaired binding of AP-2 and Pax-3 to their consensus sequences

Results and Discussion

ATRA downregulates CRMP-2 mRNA expression CRMP-2 is a cytoskeleton-interacting protein that has been linked to neuronal cell differentiation, as it binds

to tubulin heterodimers and promotes microtubule assembly during this process [5] It has been shown that after ATRA treatment, SK-H-SH neuroblastoma cells develop broad lamellipodia containing radial actin fibers, reorganizing their cellular scaffold [20]

Interestingly, Butler et al showed how ATRA treat-ment of SH-SY5Y neuroblastoma cells caused great modifications in microtubule distribution without changing the expression levels of tubulin [21] These findings pointed to a possible regulation of the pro-teins that participate as microtubule adaptors, stimula-ting interest in the study of CRMP-2 transcriptional regulation during this process

Therefore, we started by using northern blot to measure CRMP-2 mRNA expression levels during the differentiation of SH-SY5Y neuroblastoma cells As shown in Fig 1B, decreased levels of CRMP-2 mRNA are seen 48 h after ATRA treatment and are main-tained for at least 6 days

As gene expression can be modulated through both transcriptional and post-transcriptional mechanisms, we sought to assess whether ATRA treatment affected CRMP-2 gene expression at the promoter level or at the level of its mRNA stability To achieve this goal, SH-SY5Y cells were treated with ATRA 10 lm for 48 h, and actinomycin D (ActD), an inhibitor of transcrip-tion, was added afterwards RNA samples were collec-ted at different time intervals after ActD treatment As shown in Fig 1C, ATRA did not reduce the half-life of CRMP-2 mRNA, indicating that it has no effect on mRNA stability, but rather triggers its transcriptional regulation Western blot analyses revealed a reduction

in CMRP-2 protein levels, starting 72 h after ATRA treatment and lasting for at least 15 days (Fig 1D)

Isolation and characterization of the regulatory

region of the human CRMP-2 gene

As a second step in our study, we aimed to evaluate

the presence of regulatory consensus sequences in the

human CRMP-2 promoter The CRMP-2

transcrip-tion start site was previously identified by RACE

[22] It was located 282 bp upstream from the first base of the ATG translation initiating codon There-fore, a sequence of 949 bp upstream from the first transcribed base was introduced in the matinspector program [23] for bioinformatic analysis The results

of this examination are given in Fig 2, and show the presence of several cis-acting elements involved in

A

B

C

D

Fig 1 SH-SY5Y neuroblastoma cells experience changes in CRMP-2 gene expression during ATRA-induced differentiation (A) Neurite induc-tion revealed SH-SY5Y neuroblastoma cell differentiainduc-tion 2 and 6 days after exposure to 10 l M ATRA Scale bar 50 lm (B) Northern blot analyses of CRMP-2 mRNA expression in differentiating cells Total RNA was obtained at the indicated times, and 10 lg of each sample was analyzed by northern blotting The 18S ribosomal RNA gene was used as loading control (C) CRMP-2 mRNA stability was measured by adding 5 lgÆmL)1of ActD to SH-SY5Y neuroblastoma cells exposed to ATRA for 48 h RNA was extracted 0, 1, 3, 6, 9 and 12 h afterwards, and CRMP-2 mRNA levels were analyzed by northern blot mRNA expression level was calculated as the percentage of the density of the control sample at time 0 h (100%), and plotted as a function of time (D) Western blot study of CRMP-2 protein levels in ATRA-treated SH-SY5Y cells Analysis by densitometry showed that CRMP-2 mRNA expression was decreased by approximately 30% after ATRA expo-sure, whereas the reduction in protein content was slower and less pronounced (10–20%) ‘C’ corresponds to nontreated cells, and ‘T’ cor-responds to cells treated with 10 l M ATRA Results represent the average from three independent experiments The statistical test used was Student’s t-test.

Fig 2 Phylogenetic analyses show the existence of conserved regions in the CRMP-2 promoter The CRMP-2 promoter regions from five different mammals were aligned using BLAT software Conserved binding regions for several transcription factors involved in neurogenesis and differentiation are highlighted in boxes + 1 indicates the first transcribed nucleotide of the CRMP-2 mRNA.

neuronal cell differentiation that could be implicated

in its transcriptional regulation, such as Sp1, AP-2,

E2F, Pax-3 and NeuroD1 Typical TATA and CAAT

boxes were also identified in this region We also

searched for conserved regions between different

spe-cies using blat software (http://genome.ucsc.edu), and

found that the CRMP-2 promoter sequence is

strongly conserved, especially in the region located

between ) 535 and + 91 from the first transcribed

base Phylogenetically conserved regions serve as a

reliable predictor of regulatory elements, and the

detection of these significantly reduces the number of

candidate transcription factors to be tested in

func-tional assays [24]

In order to further identify the minimal 5¢ region responsible for transcriptional regulation, we per-formed four serial deletions of the CRMP-2 promoter region and cloned them into the luciferase pGL3 basic vector Deletion constructs were transiently transfected into SH-SY5Y neuroblastoma cells, and their relative luciferase activity was assayed 48 h later As can be seen from Fig 3A, deletion of the most distal part of the promoter (from) 941 to ) 768) led to a nonsignifi-cant decrease in constitutive promoter activity An additional deletion of the region located between ) 768 and ) 354 of the CRMP-2 promoter resulted in an increase in promoter activity In fact, two putative repressor elements were located in this region: HIC-1,

A

Fig 3 Characterization of the CRMP-2 minimal promoter region (A) Deletion constructs of the CRMP-2 promoter were generated and used

in luciferase reporter gene assays 48 h after transfection The relative luciferase values are shown as percentage of the mean ± SD activity

of each construct relative to CRMP-2 whole promoter (+ 91 to ) 949) Statistical significance was obtained by use of the Mann–Whitney sta-tistical test *P < 0.05; ns, not significant Results represent averages from at least three independent experiments (B) EMSA showing DNA-binding activity of SH-SY5Y nuclear extracts Oligonucleotides corresponding to the regions ) 229 ⁄ ) 151 (lanes 1–8) and ) 130 ⁄ ) 63 (lanes 9–13) from the human CRMP-2 promoter were labeled A 100-fold molar excess of the unlabeled oligonucleotides containing the con-sensus-binding sequence of AP-2 (lane 3), E2F (lane 4), Sp1 (lane 8) and Pax-3 (lane 11) or NeuroD1 (lane 12) was added for competition experiments (C) For the supershift assays, SH-SY5Y nuclear extracts were preincubated with 2 lg of anti-AP-2 serum (lane 3) or anti-Pax-3 serum (lane 7) SS, supershifted band in the presence of specific antibody.

a tumor suppressor essential for mammalian

develop-ment [25], and a neuron-restrictive silencer eledevelop-ment

recognized by a transcription factor known as the

neu-ron-restrictive silencer factor, a zinc finger containing

a transcriptional repressor [26] The neuron-restrictive

silencer element and neuron-restrictive silencer factor

have also been reported to function as direct

transcrip-tional repressors in the promoter of the semaphorin

receptors [27], indicating possible coordinated

regula-tion of genes involved in the same signaling pathway

Finally, deletion of the most proximal fragment

spanning bases) 354 to ) 130 of the CRMP-2

promo-ter caused a sharp 90% decrease in promopromo-ter activity

This region, which includes two Sp1, E2F and AP-2

potential binding sites, was considered to be

respon-sible for CRMP-2 minimal promoter activity

Identification of functional AP-2 and Pax-3

elements within the CRMP-2 promoter

We further decided to clarify which transcription

fac-tors were responsible for CRMP-2 transcriptional

activity in SH-SY5Y neuroblastoma cells For this

pur-pose, we analyzed the binding of the following

tran-scriptional regulators: E2F, which participates in the

control of cell cycle progression [28]; and Sp1, a

ubi-quitous transcription factor involved in constitutive

gene expression [29] The involvement of these two

transcription factors in CRMP-2 transcriptional

regu-lation in TGW neuroblastoma cells was suggested by

Kodama et al [30] We also chose to study the

func-tionality of other transcription factor-binding sites,

based upon their relationship with the process of cell

differentiation, phylogenetic conservation, and highest

matrix similarity values They were: AP-2, a retinoid

responsive factor that regulates the expression of many

mammalian genes during vertebrate development

[31,32]; NeuroD1, a transcription factor specifically

involved in neurogenesis [33], whose expression has

been reported to be increased in SH-SY5Y

neurobla-stoma cells after ATRA treatment [34]; and Pax-3, a

member of the paired box (PAX) family of

transcrip-tion factors implicated in neural crest differentiatranscrip-tion

during embryogenesis [35,36]

Therefore, labeled oligonucleotides spanning the

regions) 229 to ) 151 (for AP-2-, Sp1- and E2F-binding

assays) and ) 130 to ) 63 (for Pax-3- and

NeuroD1-binding assays) of the CRMP-2 minimal promoter

sequence were used as probes for electrophoretic

mobil-ity shift assays (EMSAs) in combination with nuclear

extracts from SH-SY5Y cells (Fig 3B)

DNAÆprotein complexes were detected on the labeled

) 229 ⁄ ) 151 oligonucleotide (Fig 3B, lanes 2 and 7) A

100 molar excess of unlabeled AP-2 consensus oligonu-cleotide (Fig 3B, lane 3) abolished the formation of a shifted band, pointing at the specificity of AP-2 binding

On the other hand, a 100 molar excess of unlabeled E2F and Sp1 consensus oligonucleotides (Fig 3B, lanes 4 and 8, respectively) or a nonspecific competitor (Fig 3B, lane 5) did not inhibit the formation of proteinÆDNA complexes Moreover, supershift EMSAs employing an AP-2-selective antibody demonstrated the direct binding

of AP-2 to its consensus sequence in the CRMP-2 pro-moter (Fig 3C, lane 3), whereas a nonspecific antibody did not (Fig 3C, lane 4)

To study putative Pax-3- and NeuroD1-binding sites

in the CRMP-2 promoter region, a 100 molar excess

of DNA oligonucleotides containing the binding sites for Pax-3 or NeuroD1 were used as competitor probes

As shown in Fig 3B, Pax-3 consensus oligonucleotide competed with protein binding to the CRMP-2 promo-ter region ) 130 ⁄ ) 63 (lane 11), but a NeuroD1 con-sensus oligonucleotide did not (lane 12) An excess of

a nonspecific competitor failed to compete with the formation of the Pax-3 shifted band (lane 13)

Supershift assays of Pax-3 binding to the CRMP-2 promoter region were performed to confirm the involve-ment of this transcription factor in the regulation of CRMP-2 promoter basal activity (Fig 3C, lane 7)

To gain further insights into the contribution of each transcription factor to CRMP-2 promoter activ-ity, point mutations were introduced in the AP-2 and Pax-3 consensus-binding sites of the pGL3-CRMP-2 ) 354 ⁄ + 91 minimal promoter, and their luciferase reporter activities were tested in SH-SY5Y cells To test the suitability of the mutations AP-2⁄) 213, AP-2⁄) 166 and Pax-3 ⁄ ) 113, labeled

CRMP-2)229 ⁄ ) 151 or Pax-3 ⁄ ) 113M oligonucleotides were tried in EMSA using the mutated sequences AP-2⁄ ) 213M, AP-2 ⁄ ) 166M or the ) 130 ⁄ ) 63 oligonucleo-tide as cold probes (supplementary Fig S1)

As can be seen in Fig 4A, the CRMP-2 promoter constructs harboring mutations for AP-2⁄) 213 and Pax-3⁄) 133 elements showed reduced promoter activ-ity (55% and 40%, respectively) Mutation of the AP-2 element at ) 166 had no significant effect on the constitutive promoter activity of the CRMP-2 promo-ter construct (Fig 4A)

To confirm the involvement of AP-2 and Pax-3 as transcriptional activators of the CRMP-2 gene, AP-2 (pS-AP-2) and Pax-3 (pcDNA3-PAX-3) expression vectors were cotransfected with the CRMP-2 promoter

in SH-SY5Y neuroblastoma cells The reporter activity

of CRMP-2 promoter constructs pGL3-CRMP2 ) 941 ⁄ + 91 and pGL3-CRMP2 ) 354 ⁄ + 91 increased after transfection with 0.2 lg of the AP-2 and Pax-3

expression vectors No effect was observed when the

expression vectors were transfected along with the

AP-2⁄) 213M or Pax-3 ⁄ ) 113M mutated constructs

(Fig 4B), providing more experimental evidence for

the involvement of AP-2 and Pax-3 in the regulation

of CRMP-2 expression

These data indicate that AP-2 and Pax-3 bind to the

CRMP-2 minimal promoter region, whereas neither

E2F nor NeuroD1 or Sp1 seem to be involved in

CRMP-2 basal promoter activity in SH-SY5Y

neurob-lastoma cells AP-2 and Pax-3 transcription factors

have been reported to be significantly involved in the

regulation of genes implicated in cell development and

cytoskeletal reorganization, e.g the genes for the

tyro-sine kinase receptor gene c-kit and E-cadherin [37,38]

AP-2 and Pax-3 show impaired binding to the

CRMP-2 promoter after ATRA treatment of

SH-SY5Y cells

It has been previously reported that both mRNA levels

and phosphorylation status of several

microtubule-related proteins are critical for neurons to maintain

normal cytoskeletal architecture In fact, the data

available in the literature show the regulation of these

proteins during the differentiation of SH-SY5Y

neuro-blastoma cells [39,40] For this reason, we were interes-ted in studying how CRMP-2 promoter activity could

be affected by ATRA treatment

Luciferase reporter experiments demonstrated that ATRA treatment caused a 50% reduction in CRMP-2 promoter activity compared with nontreated SH-SY5Y cells (Fig 5A) These results were in agreement with the decrease observed in CRMP-2 mRNA expression levels shown in Fig 1B We also noticed that the val-ues for the reporter activity of AP-2⁄) 213 mutated construct were the same as those obtained from ATRA-treated cells, raising the issue of ATRA treat-ment affecting AP-2 transactivation of the CRMP-2 promoter (Fig 5A)

Therefore, to determine whether there was alteration

of the binding activity of AP-2 and Pax-3 to their consensus sites after ATRA treatment, we performed EMSAs using SH-SY5Y nuclear protein extracts obtained 24 and 48 h after ATRA exposure As shown

in Fig 5B, AP-2 DNA binding to the CRMP-2 promo-ter decreased 24 and 48 h afpromo-ter treatment, whereas bind-ing of Pax-3 to its consensus sequence was reduced only

48 h after treatment Pax-3 is a transcription factor that

is finely regulated during nervous system development [41] The fall in Pax-3 expression seems to be a necessary prerequisite for the onset of morphologic differentiation and⁄ or for cessation of cell proliferation [42]

Our results are different from those published by Kodama et al [30], who pointed out the involvement of the transcription factors E2F, Sp1 and GATA1⁄ 2 in CRMP-2 transcriptional regulation after glial cell-derived neurotrophic factor exposure of neuroblastoma cells In fact, in their work, glial cell-derived neuro-trophic factor treatment increased CRMP-2 mRNA expression levels, whereas we found a reduction in CRMP-2 transcriptional activity after ATRA treatment This apparent discrepancy can be explained by the use

of different cell lines and stimuli, and by the fact that glial cell-derived neurotrophic factor has been reported

to induce cell proliferation and inhibit ATRA-induced neuritogenic and growth inhibitory effects in neurobla-stoma cells [43] It seems reasonable that the expression

of a tubulin adaptor protein in actively proliferating cells would be increased, whereas the level should decrease in those cells that have their cellular scaffold

‘fixed’, as is the case with differentiated cells

Two different mechanisms to explain the impaired transcription factor binding to CRMP-2 promoter

Having demonstrated decreased AP-2 and Pax-3 DNA binding to their consensus sequences in

ATRA-differ-A

B

Fig 4 Mutation of AP-2- and Pax-3-binding sites in the CRMP-2

promoter impaired its transcriptional activity (A) pGL-3-CRMP-2

minimal promoter constructs harboring mutations in their

AP-2 ( ) 213) or Pax-3 () 113) consensus-binding sites showed

decreased luciferase reporter activity in SH-SY5Y cells The mean

luciferase activities from three separate transfection experiments

are expressed as percentage relative to the minimal promoter

region (+ 91 to ) 354) Statistical significance was obtained by use

of the Mann–Whitney statistical test ns, not significant; *P < 0.05;

**P < 0.01 (B) Cotransfection of AP-2a or Pax-3 expression

vec-tors with the pGL3-CRMP-2 minimal promoter resulted in increased

reporter activity, whereas the use of their corresponding mutated

promoter constructs did not induce any change.

entiated neuroblastoma cells, we sought to identify

whether these changes were caused by alterations in

their expression or in their cellular availability

The study of AP-2a protein levels demonstrated, as

reported by others [44], an increase in the total AP-2

content in the cell in response to ATRA treatment, but

this increase was not statistically significant (Fig 6A)

Therefore, this could not explain its reduced binding to

the CRMP-2 promoter seen with EMSAs (Fig 5B) We therefore searched for changes in its subcellular distribu-tion after ATRA exposure, and found a significant decrease in the nuclear content of AP-2a protein (Fig 6B) that would explain the reduced AP-2a DNA binding to the CRMP-2 promoter after the treatment (Fig 5B), and therefore the increase in AP-2a cytoplas-mic content This is not the first time that this has been

A

Fig 5 Effects of ATRA treatment on CRMP-2 promoter activity and transcription factor binding (A) Exposure of SH-SY5Y cells for 48 h to

10 l M ATRA reduced the transcriptional activity of all the CRMP-2 promoter constructs tested except for those mutated in the AP-2 ( ) 213)

or Pax-3 ( ) 113) consensus-binding sites Results are given as the mean ± SD luciferase activities from three separate transfection experi-ments expressed as the percentage relative to the minimal promoter region (+ 91 to ) 354) in control cells ns, not significant; **P < 0.01;

*P < 0.05 a Treated versus nontreated b Treated versus pGL3-CRMP-2-354 ⁄ + 91 (ATRA) Statistical significance was obtained by use of the Mann–Whitney test (B) EMSA assay illustrating a reduction in AP-2 DNA binding to the ) 229 ⁄ ) 151 region of the CRMP-2 promoter after ATRA treatment (C) Pax-3 EMSA assays demonstrate that 48 h of ATRA exposure leads to a smaller amount of Pax-3 DNA being bound to its consensus sequence in the ) 130 ⁄ ) 63 region of the CRMP-2 promoter ‘C’ corresponds to nontreated cells, and ‘T’ corresponds to cells treated with 10 l M ATRA.

reported; for example, Popa et al [45] demonstrated a

loss of AP-2 transcriptional activity in differentiated

keratinocytes, associated with a reduction in its binding

to DNA This failure was caused by increased

accumu-lation of phosphorylated AP-2 in the cytoplasm

On the other hand, mRNA expression levels of Pax-3

were reduced 48 h after ATRA treatment (Fig 6C)

This decrease could be responsible for the lower

amounts of protein bound to the CRMP-2 promoter

We tried to measure Pax-3 protein content by western

blot, using the same antibody employed for the EMSA,

but were not successful

There are reports in the literature of ATRA

indu-cing cell differentiation through the inhibition of c-myc

and N-myc expression [46] Pax-3 transcription factor expression is regulated by N-myc during neural cell development [47] Therefore, we looked for a link between the expression of c-myc, N-myc and Pax-3 at the mRNA level in neuroblastoma cell differentiation ATRA treatment induced a fast decrease in the mRNA level of the c-myc proto-oncogene as measured

by semiquantitative PCR (as early as 6 h, and main-tained for the whole differentiation process) (Fig 6C)

In contrast, we found a similar reduction in the expres-sion levels of the N-myc and Pax-3 genes 48 h after ATRA treatment Therefore, it seemed reasonable to propose that, as ATRA treatment is mediating N-myc gene inhibition, Pax-3 mRNA expression is probably

A

C

B

Fig 6 ATRA treatment alters AP-2 subcellular distribution and Pax-3 expression (A) Western blot analyses of total AP-2 protein expression levels in control (‘C’) and ATRA-treated (‘T’) SH-SY5Y cells (B) ATRA treatment alters AP-2 subcellular distribution Western blot analysis of nuclear and cytoplasmic protein extracts purified from ATRA-treated cells demonstrated significantly reduced levels of AP-2 in the nuclear fraction, whereas it was increased in the cytoplasm (C) Effect of ATRA treatment on the mRNA expression levels of Pax-3 and its transcrip-tional regulators c-myc and N-myc, as measured by semiquantitative PCR The b-actin gene was amplified as loading control Analysis by densitometry showed that Pax-3, CRMP-2 and N-myc mRNA levels decreased significantly (P < 0.05) in differentiating cells, along with a major descent in c-myc expression (P < 0.001) The statistical test used was Student’s t-test **P < 0.01; *P < 0.05 ‘C’ corresponds to nontreated cells, and ‘T’ corresponds to cells treated with 10 l M ATRA.

being compromised, resulting in less transcription

fac-tor being available to modulate CRMP-2 promoter

activity

In conclusion, we have demonstrated that the

human CRMP-2 promoter contains AP-2 and Pax-3

functional elements in its minimal promoter These

two transcription factors seem to regulate basal

CRMP-2 promoter activity and expression ATRA

treatment induces a decrease in CRMP-2 expression

during the differentiation of neuroblastoma cells

through impaired binding of the aforementioned

tran-scription factors to their consensus sequences in the

CRMP-2 promoter We finally demonstrated that the

reduced binding of Pax-3 is due to a decreased

tran-scriptional rate, whereas loss of specific DNA-bound

AP-2 complexes results from a reduced level of AP-2

in the nucleus

Given that the family of collapsin response mediator

proteins has been reported to be mainly regulated by

post-translational mechanisms, the notion of them being

also regulated at the transcriptional level gives us

another example of a coordinated cell response at both

the mRNA synthesis and protein activation stages

Experimental procedures

Materials

ATRA and ActD were purchased from Sigma (St Louis,

MO, USA) The pSP(RSV)-NN and pSP(RSV)-AP-2a

expression vectors were kindly provided by T Williams

(University of Colorado Health Sciences Center, Denver,

CO, USA) Rabbit anti-(human Pax-3) serum and the

pcDNA3-Pax-3 expression vector were generously provided

by F J Rauscher III (The Wistar Institute, Philadelphia,

PA, USA) Antibody to CRMP-2 (C4G) was a generous

gift of Y Ihara (Faculty of Medicine, University of Tokyo,

Japan) Rabbit anti-(human AP-2a) serum and secondary

goat anti-(mouse IgG) and goat anti-(rabbit IgG)

conju-gated to horseradish peroxidase were purchased from Santa

Cruz Biotechnology (Santa Cruz, CA, USA) Anti-b-actin

serum (A5441: monoclonal) was purchased from Sigma

Cell culture

The human neuroblastoma cell line SH-SY5Y was

pur-chased from ATCC (CRL-2266) and cultured at 37C

under 5% CO2in DMEM: Ham’s F12 (1 : 1) medium

sup-plemented with 10% (v⁄ v) heat-inactivated fetal bovine

serum, 100 UÆmL)1 penicillin, 100 lgÆmL)1 streptomycin,

250 ngÆmL)1 fungizone and nonessential amino acids, all

from Invitrogen (Paisley, UK) Cells were plated at 200

cellsÆmm)2for all the experiments

RNA extraction

SH-SY5Y cells were serum-starved for 24 h and then trea-ted with 10 lm ATRA for different time periods Total RNA was isolated using TRIzol (Invitrogen), following the manufacturer’s instructions

Northern blot

Total RNA (10 lg) was resolved by electrophoresis in a 1% formaldehyde⁄ agarose gel, transferred to a nylon mem-brane (Amersham Biosciences, Uppsala, Sweden), UV crosslinked, and prehybridized at 42C in 5 · NaCl ⁄ Cit (1· NaCl ⁄ Cit: 150 mm NaCl, 0.15 mm sodium citrate,

pH 7.0), 50% formamide, 50.4 mm phosphate buffer

0.1 mgÆmL)1salmon sperm DNA for 8 h at 42C Hybrid-ization was carried out overnight at 42C in prehybridiza-tion soluprehybridiza-tion with 1· Denhart’s solution ⁄ 12.5% dextran sulfate, using a final concentration of 106c.p.m labeled probe per milliliter Radiolabeling of specific probes was carried out with [a-32P]dCTP (specific activity 3000 CiÆmmol)1) by random priming using the Klenow frag-ment of DNA polymerase I (Bioline, London, UK) After hybridization, blots were washed three times with

2· NaCl ⁄ Cit ⁄ 0.1% SDS for 15 min at 65 C Blots were autoradiographed with intensifying screens Quantification

of the mRNA levels in the autoradiograms was performed using imagemaster software (Amersham Biosciences)

Promoter–luciferase constructs and site-directed mutagenesis

The region from the human CRMP-2 promoter spanning bases between ) 941 and + 91 from the transcription start site was PCR amplified from human genomic DNA The primers used for the amplification reaction were as follows: CRMP-2 sense, 5¢-CCATTCCTCCGCCCTACTAAGTT-3¢; and CRMP-2 antisense, 5¢-TTCTTCCTCTCCTCCAACA CAGC-3¢ The PCR product was ligated into the pGEMT easy vector (Promega, Madison, WI, USA), sequenced, and subcloned into the SmaI and SacI sites of the luciferase reporter vector pGL3-basic (Promega) Constructs harboring 5¢ serial deletions derived from the 1 kb pGL3-CRMP-2 con-struct were obtained either by restriction enzyme digestion (construct pGL3-CRMP-2 ) 768 ⁄ + 91 digested with PstI),

or by PCR (constructs pGL3-CRMP-2 ) 354 ⁄ + 91 and pGL3-CRMP-2 ) 130 ⁄ + 91) using primers

TG-3¢, and KpnI-CRMP-2 ⁄) 130, 5¢-CTGGTACCATCG CTGCTCGTCTCTCTCG-3¢, as forward primers, and CRMP-2 antisense as the reverse primer

Site-directed mutations were generated by PCR using the QuickChange Site-Directed Mutagenesis Kit from Strata-gene (Cedar Creek, TX, USA) The pGL3-CRMP-2