Oncogene (1997) 15, 1805 ± 1813
 1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00
Distinct sensitivity of neuroblastoma cells for retinoid receptor agonists:
evidence for functional receptor heterodimers
Antoine Carpentier1, Nicole Balitrand1, CeÂcile Rochette-Egly2, Braham Shroot3, Laurent Degos4
and Christine Chomienne1
1
Laboratoire de Biologie Cellulaire HeÂmatopoõÈeÂtique, Universite Paris VII, CNRS EP107 - Institut d'HeÂmatologie, HoÃpital SaintLouis, 1 av. Cl. Vellefaux 75475 Paris cedex 10; 2IGBMC, Parc d'Innovation - 1, rue Laurent Fries 67404 Illkirch cedex; 3CIRDGALDERMA - 635 route des Lucioles 06902 Sophia Antipolis cedex and 4Institut d'HeÂmatologie, HoÃpital Saint-Louis, 1 av. Cl.
Vellefaux 75475 Paris cedex 10, France
Retinoic acid (RA) plays a major role in embryogenesis
of the nervous system and has been reported to induce
di€erentiation in neuroblastoma cell lines. To identify
RA signaling pathways involved in such di€erentiation
processes, two RA-sensitive neuroblastoma cell lines
(LA-N-5 and SH-SY5Y) were extensively studied.
Northern blot experiments determined that of the three
RAR mRNAs, only RARa was signi®cantly expressed,
with respectively weak or undetectable levels of RARg
and RARb. RXRs (a and b) receptors were weakly
expressed. Western blotting analysis con®rmed the
constitutive expression of RARa and absence of RARb
and weak levels of RXRa. Treatment with all-trans-RA
up-regulated RARa and induced a drastic increase of
RARb (both at the RNA and protein level). To further
characterize the function of RARa, RARb and RXRa in
NB cells, nuclear extracts from LA-N-5 cells were
analysed by EMSA studies. Three speci®c retarded
complexes were observed which were signi®cantly
decreased or shifted in the presence of monoclonal
antibodies to RARa, RARb and RXRa. RA treatment
dramatically induced a DR5-binding RXRa-RARb
heterodimer. Treatment with combinations of RARa or
RARb agonists with a RXRa agonist or with a RARa
agonist alone, induced neurite-outgrowth supporting the
probability that both RXRa-RARa or RXRa-RARb
heterodimers are involved in RA-mediated di€erentiation
of NB cells. The availability of novel synthetic RAspeci®c receptor ligands should provide the possibility of
tissue speci®c therapeutic regimes.
Keywords: neuroblastoma; retinoic acid;
receptor agonist; retinoic acid receptor
retinoid
Introduction
Retinoic Acid (RA) and related vitamin A derivatives
(retinoids) exert profound e€ects on many biological
processes, including embryonic pattern formation, cell
growth and di€erentiation (De Luca, 1991; Gudas,
1992). Various retinoids have been shown to modulate
growth and di€erentiation of a number of human
cancers (reviewed in Gudas et al., 1994), or in
leukemias such as acute promyelocytic leukemia
(Chomienne et al., 1991). Originating in neural crest
Correspondence: C Chomienne
Received 18 November 1996; revised 5 June 1997; accepted 5 June
1997
cells of developing sympathetic nervous system,
neuroblastoma (NB) is a common solid tumor of
childhood with poor prognosis, which sometimes
spontaneously di€erentiates in patients (Evans et al.,
1971). RA along with other retinoids has been
demonstrated to be e€ective in the di€erentiation and
inhibition of the growth of NB cell lines in vitro
(Sidell, 1982; Sidell et al., 1983; Reynolds et al., 1994),
with concomitant regulation of homeobox, thrombospondin and laminin genes (Castle et al., 1992; Manohar
et al., 1996). Subsequent clinical trials with 13-cis RA
have shown to be of some but not complete e€ect
(Finklestein et al., 1992; Smith et al., 1992; Villablanca
et al., 1995).
Investigations of the RA-signaling pathways involved
in growth and di€erentiation of normal and malignant
neuronal cells may provide insights to achieve maximal
therapeutic ecacy. RA controls target gene transcription through nuclear receptors which belong to the
steroid/thyroid hormone nuclear receptor superfamily,
RARs (RARa, b and g) and RXRs (RXRa, b and g)
(reviewed in Leid et al., 1992; Mangelsdorf and Evans,
1995). RARs and RXRs act through direct association
with speci®c DNA sequences (RA-response elements or
RAREs) which consist of a direct repeat of PuG(G/
T)TCA motifs with 1 ± 5 bp spacing (DR1 to DR5)
(Mader et al., 1993; Mangelsdorf and Evans, 1995).
These di€erent RAREs, present in the promoter region
of target genes, dictate in a given tissue the mode of
receptor binding (hetero or homodimers), the type of
receptors involved and consequently, the speci®c
retinoid-receptor ligand(s) e€ective for signaling.
RA-receptor genes are di€erentially expressed during
embryonic development and in adult tissues. Thus
RARa is ubiquitous having been found in all tissues
and cells examined, while RARb shows more restricted
pattern: for example expression is high in brain and
heart but undetectable in muscle and intestine (De The
et al., 1989; Rees et al., 1989). In mouse embryos, in situ
hybridization studies have demonstrated that neural
crest cells abundantly express RARa, and RARb;
RARg expression is spatio-temporally restricted
(Ruberte et al., 1991). Concerning the RXRs receptors, studies performed in fetal neural tissues reported
that RXRb is the most predominant, whereas RXRa is
rarely found and RXRg absent (to the exception of
the basal ganglia) (Mangelsdorf et al., 1992). Human
NB tumor cells express all three RARs isoforms,
although at various levels (Lovat et al., 1993;
Marshall et al., 1993; Li et al., 1994; Wuarin et al.,
1994; Redfern et al., 1994) and of the RXRs, RXRb
Neuroblastoma sensitivity for retinoid receptor agonists
A Carpentier et al
Western blotting with anti-RARs or anti-RXRs
antibodies were used to study protein expression in
nuclear extracts of LA-N-5 and SH-SY5Y cell lines.
RARa was easily detected as a 55 kDa protein, while
RARb was not seen and RXRa barely detected (Figure
2a, lanes 1, 3 and data not shown), con®rming mRNA
studies.
was reported to be predominant in one NB cell line
(Plum and Clagett-Dame, 1995).
However, to what extent mRNA levels determine
protein concentrations and in¯uence the nature of
functional heterodimers is unknown. In an e€ort to
study RA signaling pathways, LA-N-5 and SH-SY5Y,
two NB cell lines which have been shown to be
sensitive to RA (Sidell et al., 1983; Marshall et al.,
1993), were studied to determine the constitutive
expression of RARs (a, b, g) and RXRs (a, b, g),
their modulation and capacity to bind RARE and
con®rm in vivo their response to receptor speci®c
agonists.
E€ects of RA, alone or in combination with IFNa, on
growth and morphologic di€erentiation in LA-N-5 and
SH-SY5Y cell lines
RA-mediated growth inhibition and morphologic
di€erentiation with respect of long neurite formation
were assessed in LA-N-5 and SH-SY5Y cell lines. The
ability of IFNa (100 U/ml) to potentiate RA-ecacy
was also tested, since IFNa has been reported to
enhance RA-induced di€erentiation and to decrease
cell growth in some NB cell lines (Higuchi et al., 1991;
Rosolen et al., 1992).
In both cell lines, RA induced a growth inhibition
after 7 days (42%+19 for LA-N-5, and 44%+7 for
SH-SY5Y) (Figure 3a) and induced morphological
changes with formation of long neurites (Figure 3b, a
to d), as previously reported (Sidell et al., 1983;
Marshall et al., 1993). These changes were already
noticeable after 3 days of treatment and were much
more marked by day 7. Treatment with IFNa alone
(100 U/ml) did not result in signi®cant growth
inhibition (12%+17 for LA-N-5 and 19%+3 for
SH-SY5Y) (Figure 1a) or morphological changes. In
Results
Di€erential constitutive expression of RARs and RXRs
in LA-N-5 and SH-SY5Y cell lines
We analysed the expression of RARa, b, g and RXR a,
b, g genes by Northern blot analysis. RARa mRNAs
were constitutively expressed in both cell lines as two
transcripts of 3.5 and 2.5 kb (Figure 1a, lanes 1).
RARb mRNAs were undetectable, and RARg mRNAs
were either undetectable in LA-N-5 cells or barely seen
in SH-SY5Y cells as a single 3.0 kb transcript. For
genes of the RXR group, RXRb was weakly detected
as a 2.8 kb transcript (Figure 1b) and RXRa
expression evidenced only by RT ± PCR (data not
shown).
a
1h
1
2
3
4
1
2
3
b
48h
24h
4
1
2
3
RARα
LA-N-5
2.5 —
3.5
3.1
RARγ
RXRα
2.7 —
RXRβ
1.8 —
RXRγ
GAPDH
GAPDH
1
3.5 —
2.5 —
2
48h
24h
3
4
1
2
2
5.5 —
RARβ
1h
24h
1
4
3.5 —
SH-SY5Y
1806
3
4
1
2
3
4
RARα
3.5
3.1
RARβ
3.0 —
RARγ
5.5 —
RXRα
2.7 —
RXRβ
1.8 —
RXRγ
GAPDH
GAPDH
Figure 1 (a) Northern blot analysis of RARs in SH-SY5Y and LA-N-5 cell lines. Total RNAs (20 mg) were isolated from cells
cultured for 1, 24 and 48 h without (lane 1), or with 561076 M RA (lane 2), 100 U/ml IFNa (lane 3), or both (lane 4). The same
blots were stripped and rehybridized with a 32P-labeled GAPDH probe. (b) Northern blot analysis of RXRs in SH-SY5Y and LAN-5 cell lines after 24 h culture (1 and 48 h gave the same levels). Total RNAs (20 mg) were isolated from cell lines cultured without
(lane 1) or with RA (lane 2). The same blots were stripped and rehybridized with a 32P-labeled GAPDH probe
Neuroblastoma sensitivity for retinoid receptor agonists
A Carpentier et al
contrast, association of IFNa enhanced both RAinduced growth inhibition (63%+8 for LA-N-5, and
56%+9 for SH-SY5Y) (Figure 3a) and distinct neurite
formation (data not shown).
Up-regulation of RARa and RARb by RA alone or in
association with IFNa in LA-N-5 and SH-SY5Y cell
lines
Regulation of RARs and RXRs by RA or IFNa was
studied in LA-N-5 and SH-SY5Y cell lines. Cells were
grown for 1 h, 24 h and 48 h in the presence or
a
LA-N-5
1
2
SH-SY5Y
LA-N-5
3
1
4
2
absence of RA (561076M) and/or IFNa (100 U/ml),
RNA extracted and analysed for RARs and RXRs
gene expression.
LA-N-5 and SH-SY5Y cell lines showed respectively
a 2.8- and 1.9-fold increase in RARa mRNAs levels
after 24 h incubation with RA. Interestingly, the
increase was both delayed and transient, being
discernible after 1 h incubation, with return to basal
levels within 48 h (Figure 1, lane 2). This increase
concerned both the two 3.5 and 2.5 kb transcripts. In
contrast, a dramatic, more prolonged increase in
RARb levels was detected in both cell lines after 24
SH-SY5Y
3
b
4
LA-N-5
175 —
1
2
24h
83 —
Nucl
62 —
48h
47 —
RARα
32 —
RARα
RARβ
Figure 2 Western blot analysis of LA-N-5 and SH-SY5Y cell lines. (a) 50 mg of nuclear extracts were isolated from both cell lines
cultured for 48 h without (lanes 1, 3) or with 561076 M RA (lanes 2, 4). RARa and RARb proteins (arrows) were revealed with
speci®c RARa and RARb polyclonal antibodies. The position of prestained molecular weight standards (New England, Biolab) is
indicated in kilodaltons. (b) 50 mg of nuclear (Nucl) LA-N-5 cell extracts were isolated from cell lines cultured for 24 or 48 h
without (lane 1) or with 561076 M RA (lane 2). RARa protein levels were revealed with a speci®c anti-RARa polyclonal
antibodies. The di€erence in RARa constitutive signals between 24 and 48 h is due to di€erent exposure times
A
B
Figure 3 (A) Cell growth in LA-N-5 and SH-SY5Y cell lines, after 7 days of culture without (control), or with 561076 M RA (RA), 100 U/
ml IFNa (IFNa), or both (RA/IFNa). Experiments were done in triplicate, and results are expressed as percentage of control culture
(Mean+s.e.m.). (B) RA-induced di€erentation of LA-N-5 (a ± b) and SH-SY5Y (c ± d) cells. Cells were cultivated for seven days in presence
(b), (d), or absence (a), (c) of 561076 M RA. Neurite formation between clusters are seen upon RA treatment (b), (d) 6125
1807
Neuroblastoma sensitivity for retinoid receptor agonists
A Carpentier et al
1808
and 48 h incubation (Figure 1, lane 2). Transcripts of
3.5 and 3.1 kb were detected with a slight preponderance of the smaller transcript. No modulation of
RARg or RXRs were seen (Figure 1, lane 2). It is
noteworthy that although IFNa alone did not modify
the constitutive expression pattern of the receptors
(Figure 3a, lane 3), association with RA resulted in a
1.8-fold increase of RARa levels at 1 h (Figure 1, lane
4). There was no di€erence in expression patterns of
RARb, RARg and RXRs in the presence of IFNa and
compared with RA alone (data not shown).
Immunoblotting analysis on nuclear extracts con®rmed the up-regulation of both RARa and RARb
upon RA treatment. RARb induction after 48 h
treatment was dramatic in both cell lines, suggesting
that protein levels re¯ect mRNA modulation. Induction of RARa levels was weak in SH-SY5Y (1.6-fold)
(Figure 2a) and evident in LA-N-5 (2.9-fold) (Figure
2b). No signi®cant modulation of RXR protein levels
were noted (data not shown).
RA-signaling pathways in NB cell lines
To further determine functionality of receptors in RAsensitive NB cells, we performed electrophoresis
mobility shift assays (EMSA) with LA-N-5 nuclear
extracts, after incubation with a 32P-labeled DR5
oligodeoxynucleotide probe. Two major speci®c nuclear complexes (A1 and A2), and one minor one (A3)
were detectable (Figure 4a, lanes 1, 3) and were
abolished when incubated with excess of unlabeled
DR5 (Figure 4a, lane 2). No other speci®c binding
complexes were observed after RA treatment, but A1,
A2 and A3 complexes were signi®cantly enhanced
(Figure 4a, lane 7). Experiments with speci®c antiRARs and anti-RXRs monoclonal antibodies allowed
to further analyse the partners of these nuclear
complexes. First, incubation with an anti-pan-RXR
monoclonal antibody showed a complete shift of the
three complexes suggesting that all of them involved an
RXR partner (Figure 4a, lanes 4, 8). Second,
incubation with an anti-RARb reduced the formation
of only the A1 complex in control cells, and produced
a speci®c shifted complex upon RA treatment (Figure
4b). Third, anti-RXRa antibody induced disappearance
of the A1 complex with a speci®c shifted complex, best
seen after RA-treatment (Figure 4a, lanes 5, 9).
Incubation with both anti-RARb and RXRa monoclonal antibodies resulted in a speci®c double-shifted
complex and reduced binding of A1 (Figure 4c, lane 4)
indicating that in NB cells, RARb binds to DR5elements as a heterodimer with RXRa. Anti-RARa
antibodes speci®cally reduced the formation of the A1
(Figure 4, lanes 6, 10). These data indicate that nuclear
extracts of NB cells which speci®cally interact with the
RARE from the RARb promoter contain at least
RARa and b and RXRa.
Treatment of LA-N-5 cells with RAR and RXR speci®c
agonists
To support the hypothesis that both RARa and RARb
may participate in RA's signaling pathways in NB,
cells were cultured for ten days in the presence of
speci®c RARa (1076, 1077 or 1078 M) and RARb (1076
or 1077 M) agonists, alone or in association with an
RXRa agonist (1076 M). Cells cultured with or without
5 1076 M RA were used as positive and negative
controls (Figure 5a and b). On the tenth day, cell
viability was more than 95%. The RARa agonist
induced a signi®cant morphological di€erentiation at
all concentrations (1076 to 1078 M) (Figure 5c), whereas
growth inhibition occurred only at 1076 and 1077.
Treatment with RARb (1076 and 1077 M) or RXRa
agonists (1076 M) alone did not result in any signi®cant
morphological changes or growth inhibition (Table 1,
Figure 5d).
A synergistic e€ect was however obtained when the
RXRa agonist was associated to either RARa or
RARb agonists (Table 1, Figure 5e and f). Thus
di€erentiation induction by an RARa agonist alone,
and of the combination of either RARa or RARb
agonists with an RXRa agonist was similar to or
indeed superior all-trans RA 1076 alone.
Discussion
This study determined novel features of RARa and
RARb pathways in neuroblastoma (NB). During RAinduced di€erentiation of NB cell lines, RARa and
RARb are induced both at the mRNA and protein
levels. Gel retardation assays con®rmed the presence of
an RXRa/RARb heterodimer and of RARa in a
speci®c DR5-RARE bound nuclear extract. Treatment
with speci®c RAR and RXR agonists further
demonstrates that both RARa and RARb pathways
can separately induce morphological di€erentiation of
NB cells.
In our study, the RA-induced increase in RARa
mRNA and the ecacy of the RARa agonist alone to
trigger morphological di€erentiation of LA-N-5 cells
suggest the participation of RARa in RA-induced
neuronal di€erentiation. Interestingly, low RARa
expressions have been associated to bad prognosis in
NBs (Marshall et al., 1993). In the light of the present
results low RARa expression in NB tumors may re¯ect
a speci®c stage of tumor di€erentiation and in¯uence
therapeutic ecacy. RARa has previously been
implicated in the di€erentiation of human tissues. In
hematopoietic cells, where RARa is preferentially
expressed, alteration of the RARa receptor through
the t(15;17) and t(11;17) translocations has been linked
to blockage of di€erentiation in hematopoietic cells
(Grignagni et al., 1993; Rousselot et al., 1994) and to
leukemogenesis (de The et al., 1990; Chen et al., 1994).
The most compelling evidence for the role of RARa in
di€erentiation programmes was provided by studies of
HL-60 cells which are resistant to RA and harbour a
mutant RARa; transfection using normal RARa
expression vectors was found to restore the differentiation response to RA (Collins et al., 1990). This was
further stressed by the superior ecacy of RARa
speci®c agonists to induce di€erentiation in APL and
HL-60 cells (Calabresse et al., 1990; Chen et al., 1996).
RA-treatment is known to increase RARa mRNA in
acute promyelocytic leukemic cells (Chomienne et al.,
1991) and in RA-sensitive NB cell lines (Lovat et al.,
1993; Li et al., 1994; Wuarin et al., 1994; Redfern et
al., 1994). These up-regulations are likely RA-induced
as RARa2 gene promoters are known to have a RARE
motif (Leroy et al., 1991).
Neuroblastoma sensitivity for retinoid receptor agonists
A Carpentier et al
1809
a
A3
A2
A1
1
2
3
4
5
6
7
8
10
RA
Control
b
9
c
A3
A2
A1
A2
A1
RARβ Ab
RXRα Ab
RARβ Ab
0
+
Control
0
+
0
0
0
+
+
0
+
+
RA
Ra
Figure 4 Competition and supershift analysis of DR5-binding complexes. 3 mg of nuclear extracts of LA-N-5 were isolated from
cell lines cultured for 24 h without (control) or with 561076 M retinoic acid (RA). (a) Nuclear extracts were incubated for 20 min
with 32P-labeled DR5 alone (lanes 3, 7), or associated with an unrelated oligonucleotide (lane 1), or with a 200-fold excess of
unlabeled DR5 (lane 2). Three speci®c constitutive complexes were seen (A1, A2, A3). Incubation with monoclonal anti-pan-RXR
(lanes 4, 8) and anti-RXRa antibodies (lanes 5, 9) resulted in shifted complexes (L). Incubation with anti-RARa antibodies (lanes 6,
10) only inhibited formation of the A1 complex in control extracts. (b) Nuclear extracts were incubated for 20 min with 32P-labeled
DR5, without (0) or with monoclonal anti-RARb antibodies (+). Incubation with anti-RARb antibodies only inhibited formation
of the A1 complex in control extracts while a shifted complex was seen on RA-treated nuclear extracts (small arrow on the left). (c)
Nuclear extracts were incubated for 20 min with 32P-labeled DR5, without (0) or with anti-RARb antibodies (+), or anti-RXRa
(3), or both (4). Incubation with either anti-RXRa or anti-RARb antibodies induced shifted complexes (small arrow on the left),
while incubation with both resulted in a doubled shifted complex (large arrow on the left)
Neuroblastoma sensitivity for retinoid receptor agonists
A Carpentier et al
1810
a
b
c
d
e
f
Figure 5 Treatment of LA-N-5 cells with RAR and RXR speci®c agonists. Cells cultured in the absence of RA or agonists were
used as a negative control (a). Cells were cultured for ten days in the presence of 561076 M RA (b) RARa agonist (1078 M) (c),
RARb (1076 M) (d) agonists, and combination of RARa (1078 M) and RXRa agonists (1076 M) (e) or combination of RARb
(1076 M) and RXRa agonists (1076 M) (f). Treatment with a RXRa agonist alone did not result in any signi®cant morphological
changes (data not shown). 6250
Furthermore, this study brings evidence that RARb
may also be implicated in RA signaling pathways
leading to di€erentiation. This is clearly evidenced by
the identi®cation of an RXRa-RARb functional
heterodimer in EMSA studies and the striking
ecacy of the association of RARb and RXRa
speci®c agonists to induce di€erentiation. Synergistic
actions of selective RAR and RXR agonists have
recently been reported in P19 and F9 cells (Roy et al.,
1995) and in APL cells (Chen et al., 1996). This has
suggested that in the heterodimer, both RAR and
RXR may activate transcription, although cooperation
between independent events induced by RAR and
RXR ligands could not be ruled out. RARb has been
less clearly implicated in di€erentiation processes and
oncogenesis than RARa. Nevertheless, absence of
RARb expression is noted in small cell lung and
ovarian cancers where deletion of chromosome 3 is
frequent (Carney and de Leij, 1988), and transgenic
mice bearing an excess of a dysfunctional RARb
isoform have increased lung tumors (Berard et al.,
1994). Interestingly, in NB cell lines, RARb is either
decreased or absent and inversely correlated to the
ampli®cation of N-myc (Li et al., 1994), a clinical
prognosis parameter of neuroblastomas (Brodeur and
Nakagawara, 1992). If indeed RARb participates in
neuronal di€erentiation, then absence of RARb
expression in NB cells may be one of the events
leading to the malignant process. Its up-regulation by
RA could represent the functional pathway explaining
RA-sensitivity. RARb2 gene promoters are also
known to possess a RARE motif (De The et al.,
1990), and indeed, RA-treatment increases RARb
mRNA in a variety of cell lines such as hepatoma
(De The et al., 1989), embryonal carcinoma (Hu and
Gudas, 1990; Nervi et al., 1990) and NB cell
Neuroblastoma sensitivity for retinoid receptor agonists
A Carpentier et al
1811
Table 1
A
RARa
CD 336
CD 2314
CD 2809
8
>3750
>6100
Kd values (nM)
RARb
131
145
>1700
B
Agonist(S)
Control
RA (5 1076 M)
CD2809 (1076 M)
CD336 (1076 M)
CD336 (1077 M)
CD336 (1078 M)
CD2314 (1076 M)
CD2314 (1077 M)
CD2809 (1076 M)
CD2809 (1076 M)
CD2809 (1076 M)
CD2809 (1076 M)
CD2809 (1076 M)
and
and
and
and
and
CD336 (1076 M)
CD336 (1077 M)
CD336 (1078 M)
CD2314 (1076 M)
CD2314 (1077 M)
RARg
Transactivations RXRa
45
no binding
>5200
no
no
EC50*50 nM
Morphological e€ect
Growth inhibition
0
++
0
++
++
++
0/+
0
+++
+++
+++
++
++
none
40%
none
30%
25%
none
none
none
75%
70%
35%
40%
none
(A) ligand binding properties of the agonists, (B) cell growth inhibition and morphological di€erentiation
induced by selective synthetic retinoids in LA-N-5 cells. The morphological di€erentiation of the cells was
evaluated on the tenth day with respect to long neurite formation, and denoted 0 (no e€ect when compared to
control), +(intermediary e€ect), ++(e€ect similar to RA alone), or +++(e€ect greater than RA alone)
lines (Lovat et al., 1993; Li et al., 1994; Wuarin et al.,
1994; Redfern et al., 1994).
This RA-induced transcription of both RARa and
RARb, which has been shown to be critical in RAsensitivity of NB cells (Wuarin et al., 1994), could
either suggest that both RARa and RARb are needed
in the di€erentiation process of neuroblastoma cells on
their own or that there is a threshold level of ligandactivated RARs which must be reached for ecient
di€erentiation. That both RARa and RARb pathways
can separately induce morphological di€erentiation
argues in favor of the latter hypothesis. Functional
redundancy has been reported in knock-out mice
(Kastner et al., 1995), and in RARg7/7 null F9
cells where over expression of RARa and RARb can
clearly mediate the RA-induction of early target genes
whose expression is abrogated in RARg7/7 cells
(Taneja et al., 1995). Yet, di€erentiation and late gene
modulation may not be completely restored strongly
suggesting that di€erent receptors may induce speci®c
molecular events (Marshall et al., 1995). The distinct
ecacy of retinoid agonists in this study raises the
possibility that target gene transactivation through
RARa or RARb in NB cells must di€er as only the
RARa agonist was e€ective alone, RARb requiring the
presence of the RXR agonist. As recently suggested,
this would imply a direct control through the AF-2
domain of RARa in the RXR/RARa heterodimer and
through the AF-2 domain of RXRa for the RXRa/
RARb heterodimer (Chen et al., 1996; Lala et al.,
1996). Thus di€erential expression and also di€erential
biological function may account for the presence of
various retinoic acid receptors in a given tissue.
As NB cell lines are thought to arise in neural crest
cells, the preferential expression of RXRb we detect in
LA-N-5 and SK-SY5Y cells is in agreement with the
high levels of RXRb in fetal neural tissues (Mangelsdorf et al., 1992). Yet, only RXRa showed evidence of
functional involvement in DR5-binding experiments,
which may suggest that RXRa is more dedicated to
RARE binding than RXRb.
Treatment of acute promyelocytic leukemia is the
®rst successful application of di€erentiation therapy for
malignant disease and e€orts are being made to
evaluate retinoic acid therapy in other cancers. Our
results provide strong molecular basis for RA ecacy
in neuroblastoma cell lines and further stress the
interest of targeted therapy with speci®c ligands,
which might extend the therapeutic use of retinoids
and avoid RA's numerous side-e€ects.
Materials and methods
Cell cultures
The human neuroblastoma cell lines LA-N-5 (Sidell et al.,
1983), and SH-SY5Y (Ross et al., 1983) were kindly
provided respectively by Dr Sidell (LA) and Muriel Rigolet
(UCSF). Cells were grown in RPMI 1640 (Life Technology, Cergy-Pontoise, France), supplemented with 10%
Fetal Calf Serum (Boehringer, Meylan, France), 2 mM
Glutamine, 50 U/ml penicillin, 50 mg streptomycin and
1 mg/ml fungizone (Life Technology, Cergy-Pontoise,
France). All-trans-RA (RA) and IFNa were kindly
provided by Ho€man-Laroche Laboratories (Basel, Switzerland). Speci®c agonists CD336 (RARa agonist), CD2314
(RARb agonist) and CD2809 (RXRa agonist) were kindly
provided by B Shroot (CIRD Galderma, Sophia-Antipolis,
France). The powdered retinoids were dissolved in
dimethyl-sulfoxide at an initial stock concentration of
1072 M, stored at 7208C, and further diluted in RPMI
1640 before use. For IFNa, the initial stock concentration
of 104 unit/ml was stored at 7208C, and further diluted in
RPMI 1640 before use. For growth and di€erentiation
assays, 400 000 cells were plated in 25 cm2 culture ¯asks,
RA (5 1076 M) or agonists (1076 to 1078 M) were added at
day 1 of culture, and the media (with or without RA or
speci®c agonists) was changed on day 4 as previously
reported (Sidell, 1982). Morphological di€erentiation was
Neuroblastoma sensitivity for retinoid receptor agonists
A Carpentier et al
1812
assessed daily with a phase-contrast microscope, with
respect of long neurites formation. Cells were harvested
on the indicated day by trypsinization, and counted on a
Malassez hematocytometer. Cell viability was checked by
trypan blue exclusion. Cell counts were performed in
triplicate, and results expressed as the mean+s.e.m.
Northern blot analysis, DNA probes and hybridization
techniques
Con¯uent cells were treated for 1 h, 24 h or 48 h with RA
(5 1076 M), IFNa (100 U/ml) or both, and 3061076 cells
were collected and used for each extraction. RNA was
extracted by the guanidium isothiocyanate method,
precipitated through a CsCl gradient (Chirgwin et al.,
1979), and stored at 7708C until use. RNA (15 ± 20 mg)
was denatured and electrophoresed on a 1% agaroseformaldehyde gel in formamide loading bu€er, and then
transferred to a nitrocellulose ®lter (Hybond C, Amersham, Les Ulis, France). Filters of Northern transfers were
hybridized with cDNA probes labeled by the Radprime
DNA labeling system (Gibco, Life Technology, CergyPontoise, France) to speci®c activities of 109 c.p.m./mg. A
rat glyceraldehyde-3-phosphate dehydrogenase probe was
used to rehybridize the ®lters and normalize RNA
expression. Prehybridizations and hybridizations were
performed
in
solutions
containing
6XSSC
(16SSC=0.15 M sodium chloride, 0.15 M sodium citrate),
0.5% sodium dodecyl sulfate (SDS), 50% formamide, 56
Denhart's solution and 100 mg/ml Herring sperm DNA at
428C. The ®lters were then washed under stringent
conditions in 26SSC, 0.1% SDS for 20 min at room
temperature, and for another 5 min in 0.26SSC, 0.1%
SDS at 428C. The ®lters were subjected to autoradiography
using Kodak-Biomax ®lms for 7 days. Quantitative
densitometry was performed with a scanning laser
densitometry (Appligene, Illkirch, France).
RARa, RARb, RARg, RXRa and RXRg probes were
obtained by EcoRI digestion from clones pSG5-RARa, p17RARb (generous gifts from Dr De TheÂ), pBS-RARg, pRSXR3-1 and pSK-mRXRg (generous gifts from Drs Mangelsdorf and Evans). The RXRb probe was obtained by ECoRI ±
SacI digestion from clone pBS-KS TR24-3 (generous gift
from Dr Dihenzo).
Preparation of nuclear extracts
At di€erent time intervals, cells were collected by trypsineEDTA (Life Technology, Cergy-Pontoise, France) and
washed twice with cold phosphate bu€er saline (PBS)
containing 10 mM NaF, 1 mM levamisole, 1 mM benzamidine, 10 mM sodium metabisul®te. Pellets were gently
resuspended on ice in 0.4 ml of bu€er A (20 m M HEPES,
pH 8/5 mM MgCl2/0.5% Nonidet P-40/10 mM NaF/1 mM
levamisole/1 mM benzamidine/10 mM sodium metabisul®te/
0.1 mM phenylmethylsulfonyl ¯uoride and 1% aprotinin).
After centrifugation, the supernatant was kept at 7808C
and used as cytoplasmic extracts. The pellet containing the
nuclei was further extracted in 0.4 ml of bu€er B (10 mM
HEPES, pH8/25% glycerol/5 mM MgCl2/0.1 mM CaCl2/
0.1 mM EDTA/0.1 mM EGTA/500 mM NaCl/10 mM
Naf/0.5 spermidine/0.15 mM spermine/7 mM b-mercaptoethanol/1 mM levamisole/1 mM benzamidine/1 mM sodium
metabisul®te/0.1 mM phenylmethylsulfonyl ¯uoride and
1% aprotinin). The suspension was incubated for 15 min
on ice and then centrifuged at 10 000 g for 15 min. The
resulting supernatant was referred to as the nuclear extract
and rapidly frozen at 7808C. Protein concentrations were
determined with a BCA-protein assay kit (Pierce, Rockford, USA).
Western immunoblotting
Discontinuous sodium dodecyl sulfate polyacrylamide gel
electrophoresis (SDS ± PAGE) was performed using a 10%
separating gel, and a 5% stacking gel. Prestained molecular
weight markers (New-England Biolab, Montigny, France)
were loaded in adjacent lanes. Nuclear extracts corresponding to 50 mg protein were diluted 1 : 1 with 26 sample
bu€er (2% SDS, 0.125 M Tris HCl, 20% glycerol, 0.02%
bromophenol blue, and 10% b-mercaptoethanol). Equal
loadings was checked by Coomassie blue and Red Ponceau
stainings. After electroblotting, nitrocellulose membranes
(Hybond-ECL, Amersham, Les Ulis, France) were immunoprobed overnight with RARa and RARb polyclonal
antibodies directed against the F region (Gaub et al., 1992;
Rochette-Egly et al., 1992), and detected with a chemiluminescence kit and a peroxydase conjugated goat anti
rabbit/anti-mouse antibody (Boehringer-Mannheim, Meylan). Blots were washed, autoradiographed and developed
according to the manufacturer's instructions.
Electrophoretic mobility shift assay (EMSA)
3 mg of nuclear extracts were incubated for 20 min at room
temperature with 0.2 ng of 32P-labeled double-stranded
oligonucleotide (50 000 c.p.m.) and 2 mg of poly(dI-dC)
(Pharmacia Biotech, Orsay, France) in 20 ml of binding
bu€er (20 mM KCl, 1 mM MgCl2, EGTA 0.1 mM, DTT
0.5 mM, Ficoll 0.4%, 20 mM HEPES, pH 7.5). The
oligonucleotide used in this study was a synthetic DR5element (5'-GATCAGGGTTCACCGAAAGTTCACTCGCATATATTAG-3'), which corresponds to the native
RARb response element (de The et al., 1990) and presents
a high anity for RXR/RAR heterodimers (Mader et al.,
1993; Mangelsdorf and Evans, 1995). The DNA-protein
complexes were separated by electrophoresis on 4%
polyacrylamide gels in 0.46 Tris-borate/EDTA electrophoresis bu€er (TBE). The gels were dried and subsequently analysed by autoradiography. For competition
analysis, nuclear extracts were preincubated with 200-fold
excess of unlabeled double-stranded DR5 or unrelated
oligonucleotide (5'-GATCTTCTAGAAGAGTCCAGGTGGACAGGTAAG-3') in binding bu€er before the addition
of 32P-labeled DR5. For supershift analysis, nuclear
extracts were incubated with 2 mg of monoclonal anti
pan-RXR (DE region) 4RX1D12, anti RXRa (DE region)
4RX3A2, anti RXRb (AB region) 16RX3E8, anti RARb (F
region) Ab8B(F) and anti-RARa antibodies (F region)
Ab9a, (Gaub et al., 1992; Rochette-Egly et al., 1992, 1994).
Acknowledgements
This work was supported by grants from Association
contre le Cancer and was part of Biomed I programme,
AC-SSV and INSERM Network. All the antibodies used in
this study were kindly provided by CeÂcile Rochette-Egly
and Pierre Chambon (Strasbourg, France). We thank
Pierre Chambon for his interest in this work.
References
Berard J, Gaboury L, Landers M, De Repentigny Y, Houle
B, Kothary R and Bradley WE. (1994). EMBO J., 13,
5570 ± 5580.
Brodeur GM and Nakagawara A. (1992). Am. J. Pediatry.
Hematol. Oncol., 14, 111 ± 116.
Neuroblastoma sensitivity for retinoid receptor agonists
A Carpentier et al
Calabresse C, Barbey S, Venturini L, Balitrand N, Degos L,
Fenaux P, Chomienne C. (1995). Leukemia, 9, 2049 ± 2057.
Carney DN and De Leij L. (1988). Semin. Oncol., 15, 199 ±
214.
Castle VP, Ou X, O'Shea S and Dixit VM. (1992). J. Clin.
Invest., 90, 1857 ± 1863.
Chen Z, Guidez F, Rousselot P, Agadir A, Chen SJ, Wang
ZY, Degos I, Waxman S, Zelent A and Chomienne C.
(1994). Proc. Natl. Acad. Sci., 91, 1178 ± 1182.
Chen JY, Cli€ord J, Zusi C, Starrett J, Tortolani D,
Ostrowski J, Reczek PR, Chambon P and Gronemeyer
H. (1996). Nature, 382, 819 ± 822.
Chirgwin JM, Przybyla AE, MacDonald RJ and Rutter WJ.
(1979). Biochemistry, 18, 5294.
Chomienne C, Balitrand N, Ballerini P, Castaigne S, De TheÂ
H and Degos L. (1991). J. Clin. Invest., 88, 2150 ± 2154.
Collins SJ, Robertson KA and Mueller L. (1990). Mol. Cell.
Biol., 10, 2154 ± 2163.
De Luca LM. (1991). FASEB J., 5, 2924 ± 2933.
De The H, Marchio A, Tiollais P and Dejean A. (1989).
EMBO J., 8, 429 ± 433.
De The H, Chomienne C, Lanotte M, Degos L and Dejean A.
(1990). Nature (London), 347, 558 ± 561.
Evans AE, D'Angio GJ and Randolph J. (1971). Cancer, 27,
374 ± 378.
Finklestein JZ, Krailo MD, Lenarsky C et al. (1992). Med.
Ped. Oncol., 20, 307 ± 311.
Gaub MP, Rochette-Egly C, Lutz Y, Ali S, Matthes H,
Scheuer I and Chambon P. (1992). Exp. Cell Res., 201,
335 ± 346.
Grignagni F, Ferruci PF, Testa U, Talamo G, Fagioli M,
Alcalay M, Mencarelli A, Grignani F, Peschle C, Nicolleti
L and Pellici PG. (1993). Cell, 74, 423 ± 431.
Gudas LJ. (1992). Cell Growth Di€., 3, 655 ± 662.
Gudas LJ, Sporn MB and Roberts AB. (ed) (1994). The
retinoids: Cellular Biology and Biochemistry of the
Retinoids. Raven Press: New York, pp 443 ± 520.
Higuchi T, Hannigan GE, Malkin D, Yeger H and Williams
BR. (1991). Cancer Res., 1, 3958 ± 3964.
Hu L and Gudas L. (1990). Mol. Cell Biol., 19, 391 ± 396.
Kastner P, Mark M and Chambon P. (1995). Cell, 15, 859 ±
869.
Lala DS, Mukherjee R, Schulman IG, Koch SSC, Dardashti
LJ, Nadzan AM, Croston GE, Evans RM and Heyman
RA. (1996). Nature, 383, 450 ± 453.
Leid M, Kastner P and Chambon P. (1992). Trends Biochem.
Sci., 176, 427 ± 433.
Leroy P, Nakshatri H and Chambon P. (1991). Proc. Nat.
Acad. Sci. USA, 87, 4804 ± 4808.
Li C, Einhorn PA and Reynolds CP. (1994). Progress in Clin.
Biol. Research, 385, 221 ± 227.
Lovat PE, Pearson ADJ, Malcolm A and Redfern CPF.
(1993). Neurosci Lett., 162, 109 ± 113.
Mader S, Chen JY, Chen Z, White J, Chambon P and
Gronemeyer H. (1993). EMBO J., 15, 5029 ± 5041.
Mangelsdorf DJ, Borgmeyer U, Heyman RA, Zhou JY, Ong
ES, Oro AE, Kakizuka A and Evans RM. (1992). Genes
and Development, 6, 329 ± 344.
Mangelsdorf DJ and Evans RM. (1995). Cell, 83, 841 ± 850.
Manohar CF, Salwen HR, Furtado MR and Cohn SL.
(1996). Tumour Biology, 17, 34 ± 47.
Marshall GM, Cheung B, Stacey KP, Norris MD and Haber
M. (1993). Anticancer Res., 13, 437 ± 442.
Marshall GM, Cheung B, Stacey KP, Camacho ML,
Simpson AM, Kwan E, Smith S, Haber M and Norris
MD. (1995). Oncogene, 11, 485 ± 491.
Nervi C, Vollberg TM, Grippo JF, Lucas DA, George MD,
Sherman MI, Shudo K and Jetten AM. (1990). Cell
Growth Di€., 1, 535 ± 542.
Plum LA and Clagett-Dame M. (1995). Arch. Biochem.
Biophys., 319, 457 ± 463.
Redfern CPF, Lovat PE, Malcolm AJ and Pearson ADJ.
(1994). Biochem. J., 304, 147 ± 154.
Rees J, Daly A and Redfern C. (1989). Biochem. J., 259,
917 ± 919.
Reynolds CP, Schindler PF, Jones DM, Gentile JL, Prott
RT and Einhorn PA. (1994). Advances in Neuroblastoma
Research. Evans AE, Biedler JL, Brodeur GM (eds).
Wiley-Liss: New York, 4, 237 ± 244.
Rochette-Egly C, Gaub MP, Lutz Y, Ali S, Scheuer I,
Chambon P. (1992). Mol. End., 6, 2197 ± 2209.
Rochette-Egly C, Lutz Y, P®ster V, Heyberger S, Scheuer I,
Chambon P, Gaub MP. (1994). B.B.R.C., 525 ± 536.
Rosolen A, Colomonici OR, Pfe€er LM, Whitesell L,
Nordan R and Nekers LM. Eur. Cytokine Netw. (1992),
2, 81 ± 88.
Ross RA, Spengler BA, Biedler JA. (1983). J. Natl. Cancer
Inst., 71, 741 ± 747.
Rousselot P, Hardas B, Castaigne S, Dejean A, De The H,
Degos L, Farzaneh F, Chomienne C. (1994). Oncogene, 9,
545 ± 551.
Roy B, Taneja R, Chambon P. (1995). Mol. Cell Biol., 15,
6481 ± 6487.
Ruberte E, Dolle P, Chambon P, Morriss-Kay G. (1991).
Development, 111, 45 ± 60.
Sidell N. (1982). J.N.C.I., 68, 589 ± 596.
Sidell N, Altman A, Haussler MR, Seeger RC. (1983). Exp.
Cell Res., 148, 21 ± 30.
Smith MA, Adamson PC, Balis FM, et al. (1992). J. Clin.
Oncol., 10, 1666 ± 1673.
Taneja R, Bouillet P, Boylan JF, Gaub MP, Roy B, Gudas L,
Chambon P. (1995). Proc. Natl. Acad. Sci. USA, 92,
7854 ± 7858.
Villablanca JG, Khan AA, Avramis VI, Seeger RC, Matthay
KK, Ramsay NKC, Reynolds CP. (1995). J. Clin. Oncol.,
13, 894 ± 901.
Wuarin L, Chang B, Wada R, Sidell N. (1994). Int. J. Cancer,
15, 840 ± 845.
1813