ABSTRACT
We have identified and characterized a novel retinoic acid (RA) response element
(Hi-RARE) in the second intron of the mouse major histocompatibility
H2K
b
gene. The Hi-RARE sequence is conserved in all mouse classical and Q class I genes, in
MHC class I genes of the rat, Rhesus macaque, cat and in the vast majority of
human classical and non-classical class I genes. The Hi-RARE sequence lies within a regulatory element responsible for
constitutive expression of a 5
'
enhancerless
H2K
b
gene in the Ltk
-
fibroblasts. Hi-RARE consists of two inverted palindromic RARE consensus sites (5
'
-PuGGTCA-3
'
) separated by an 8 nt spacer. Mutational analysis revealed that both inverted
palindromic hexanucleotide motifs are indispensable functional sites for the 9-
cis
RA response. The Hi-RARE sequence confers 9-
cis
RA inducibility to a heterologous promoter. The inducibility is further
augmented in embryonal carcinoma cells by the expression of recombinant
retinoic acid receptors (RARs) and the retinoid X receptors (RXRs).
In vitro
, the recombinant RAR/RXR heterodimer creates DNA-protein complex with the Hi-RARE sequence. Treatment of P19 embryonal carcinoma cells with 9C-RA induces the Hi-RARE binding activity of nuclear proteins that proved
to be RAR (or RAR-like)/RXR heterodimer. Thus the Hi-RARE represents a new type of RA response element with a role in the
modulation of the expression of MHC class I family genes.
The family of the mouse MHC class I genes is comprised of >25 genes, pseudogenes
and gene fragments but only three of them represent the classical MHC class I
genes both in mice (
H2K, -D, -L)
and in humans (
HLA-A, -B, -C)
(
1
). The variable number of related functional class I genes,
H2M, Q
and
Tla
in mice and
HLA-E, -F, -G,
and
-M
in humans, show a different pattern of expression (
2
-
4
).
The classical class I major histocompatibility complex (MHC) genes play a
central role in the cellular immune response. Their products bind endogenously
processed foreign polypeptides and present them on the cell surface. Thus cells
carrying foreign antigen in the context of MHC molecules can be effectively
discriminated from cells bearing self antigen during immunological surveillance
by CD8
+
cytotoxic T lymphocytes (
5
,
6
). The expression of classical MHC class I genes varies greatly among cell types
of an adult organism, being high in lymphoid tissues, liver and lung and low or
absent on the surface of brain cells, acinar cells of the pancreas and mature
sperm (
7
).
It is known that crucial control over the classical MHC class I gene expression
is transcriptional (
5
,
8
). The region at nucleotides -213 and -61 relative to the transcription start site harbors three
elements, enhancer A or class I regulatory element (CRE), enhancer B, and
interferon response sequence (IRS) which partially overlaps the CRE. In the
adult mouse, these elements control transcription of classical
H2
class I genes in both constitutive and inducible fashion. Enhancer A core
region has been confirmed as a target for the binding of at least four
transcription factors that may govern constitutive transcription of the
H2
class I genes: KBF1, H2TF1 and NFkB (reviewed in
9
).
Inducible
H2
transcription by tumor necrosis factor and interferon was shown to be
controlled from the CRE region and from the overlapping IRS in the 5' flank of the
H2
class I genes (
7
). The upstream part of the CRE binds different members of the nuclear hormone
receptor family (
10
) and acts as a response element for retinoic acid (
11
,
12
).
Although the majority of studies of MHC class I gene regulation have focused on
the 5' upstream regulatory sequences, it appears that these sequences do not account for all
cis
-regulation of these genes (
13
-
15
). In particular, we have shown that a deletion mutant of the
H2K
b
gene that lacks the entire 5' enhancer sequence is fully expressed when transfected into L
fibroblasts. We demonstrated further that its transcription was supported from
a strong downstream regulatory element (H2DRE) located in the second intron and
flanking exon sequences, +272 to +806 relative to the transcription start site
(
15
).
Here we describe a new retinoic acid response element (referred to as Hi-RARE, histocompatibility-intron retinoic acid response element) within the second intron of
the
H2K
b
gene DRE region. We show that retinoids induce binding of the nuclear proteins
to Hi-RARE and that in embryonal carcinoma cells transcription of a reporter
gene carrying the Hi-RARE sequence can be considerably augmented by simultaneous coexpression
of the RAR-RXR transactivator heterodimers. Moreover, we demonstrate that Hi-RARE sequence binds recombinant RAR/RXR heterodimer
in vitro
and that monoclonal antibodies anti-RXR and anti-RAR nuclear receptors specifically supershift the complexes formed
between Hi-RARE DNA and nuclear proteins of embryonal carcinoma cells. The Hi-RARE consensus sequence is conserved in mouse, rat, Rhesus macaque
and human, thus supporting its functional role in various MHC class I genes.
The following double-stranded oligonucleotides, shown as coding strands, were used: Hi-RARE(W), wild-type form of the intronic (+559 to +584) sequence of the
H2K
b
gene, 5'-gatcGAGTGACCCCGGGTCGGAGGTCACGA-3'; Hi-RARE (M1), 5'-gatcGAGTGACCCCGGGTCG-AGGTCACGA-3'; Hi-RARE (M2), 5'-gatcGAGTGACCCCGccTCGGAGGTCACGA-3'; Hi-RARE
(M3), 5'-gatcGGGTCGGAGGTCACGA-3'; Hi-RARE (M4), 5'-gatcGAGTGACCCCGGGTCG-3'; (Hi-RARE mutant
forms 1-4). H-2RII, region II of the 5' enhancer of
H2L
d
gene (-204 to -180), 5'-gatcAGGCGGTGAGGTCAGGGGTGGGGAA-3'. The double-stranded oligonucleotide RARE[beta]2 encompasses retinoic acid response
element from the promoter of RAR[beta]2 gene, 5'-TCGACGGGTAGGGTTCACCGAAAGTTCACTCGC-3'.
The
Dde
I-
Hin
fI (+517 to +684) and
Hin
fI-
Kpn
I (+684 to +806) fragments of the
H2K
b
gene DRE region were derived from p
Dde
I-
Kpn
I289conCAT plasmid (
15
) and were subcloned into the
Bam
HI site of the pUC18 cloning vector. The reporter plasmids pHiRAREconCAT,
pHiRAREtconCAT were obtained by cloning a double-stranded oligonucleotide Hi-RARE as a monomer or tetramer, respectively, into the
Bam
HI restriction site of the pconCAT vector. Similarly have been prepared reporter
plasmids carrying mutated Hi-RARE sequences. To construct the pDS245conCAT reporter, the
Dde
I-
Sau
3AI fragment (+517 to +762) overlapping the second intron of the
H2K
b
gene was subcloned into the
Bam
HI site of pconCAT (
16
). The pSS193conCAT reporter plasmid containing the
Sau
3AI fragment (-264 to -61) from the 5' flank of the
H2K
b
gene was described previously (
15
).
Murine embryonal carcinoma P19 and F9 cells were grown in Dulbecco's modified
Eagle's medium (DMEM) supplemented with 10% calf serum (FCS). Murine fibroblast
Ltk- cells and human cervical carcinoma HeLa cell line were cultured in DMEM
containing 5% FCS. All media were supplemented with 2 mM L-glutamine, 100 U/ml penicillin and 100 [mu]g/ml streptomycin.
Cells were transfected 1 day after plating using the calcium phosphate
coprecipitation technique with 5 [mu]g of the chloramphenicol acetyltransferase (CAT) reporter plasmid and 2 [mu]g [beta]-galactosidase expression vector pCH110 (Pharmacia) as an
internal control for transfection efficiency (
16
). The most efficient combination of transactivators was determined by
cotransfection of the reporter plasmid with various combinations of pSG5
vectors expressing mRAR[alpha]1, mRAR[beta]2, mRAR[gamma]1 and/or mRXR[alpha], mRXR[beta], mRXR[gamma] (a generous gift of P. Chambon). The
cDNA of cloned receptors (
17
,
18
) was inserted into a cloning site of the pSG5 eukaryotic expression vector. The
total amount of DNA used for transfection was adjusted to 10 [mu]g with pUC18 DNA. After transfection the medium was replaced by a medium
supplemented with all-
trans
retinoic acid or 9-
cis
retinoic acid (a kind gift of J. Grippo) at a final concentration of 10
-8
to 10
-5
M or with a vehicle (ethanol) and incubated for 20 h. CAT assays were performed
as described previously (
15
) and normalized for transfection efficiency by measuring the activity of [beta]-galactosidase. Nuclear extracts of the cell lines were prepared
according Dignam and co-workers (
19
). Whole-cell extracts derived from cultured murine cell lines Ltk- and P19 were prepared according to the protocol of Schöler and co-workers (
20
). The extracts were used directly in electrophoretic mobility-shift assay or were stored at -80oC.
The Hi-RARE double stranded oligonucleotide or DNA fragments were end-labeled by filling-in with [[alpha]-
32
P]dATP using Klenow. Three to six [mu]g of the cell extracts were incubated with 20-50 fmol radiolabelled probe (0.5-2 * 10
4
c.p.m.) and with 1-2 [mu]g of non-specific competitor poly(dI-dC) at room temperature for 20 min in a 20 [mu]l reaction mixture containing 20 mM HEPES (pH 8.0),
70 mM KCl, 0.2 mM EDTA, 1 mM DTT, 0.5 mM PMSF, 5% Ficoll (type 400). For
competition an 100-fold excess of a specific oligonucleotide was added 5 min prior to the
addition of the labeled probe. The EMSA experiments with the recombinant His-RAR[alpha]/RXR[alpha] were carried out as described (
21
). Human 6H-RAR[alpha]/RXR[alpha] (a gift of H. Stunnenberg) was expressed in HeLa cells from
recombinant vaccinia virus. 6H-RAR[alpha]/RXR[alpha] was copurified through Ni
2+
-NTA chromatography (
21
). In supershift experiments a 1:40 dilution of monoclonal antibodies anti-RAR[alpha] (Ab9[alpha]) and anti-RXR (mouse [alpha],[beta],[gamma] - 4RX-1D12), a gift from
Pierre Chambon, was added to the binding reaction on ice, 15 min before loading
on the gel. Samples were electrophoresed on 6% polyacrylamide gels in 0.25* TBE buffer at 25 mA at 4oC and DNA-protein complexes were then visualized by autoradiography.
The
Dde
I-
Hin
fI fragment (+517 to +684) of the
H2K
b
gene subcloned in the
Bam
HI site of the pUC18 plasmid was radiolabeled at the 3' end of either the coding or the non-coding strand and partially methylated with dimethyl sulfate (0.5%)
at room temperature for 4 min as described (
22
). The probe (5 ng, 2 * 10
5
c.p.m.), twice ethanol precipitated, was incubated with 20 [mu]g Ltk- nuclear extract proteins and with 5 [mu]g poly(dI-dC) for 20 min and subjected to EMSA. Gel bands with bound
and free probe were excised, eluted, extracted with chloroform, precipitated by
ethanol and cleaved with 10% piperidine at 100oC for 30 min. After ethanol precipitation the probes were resolved on 8%
polyacrylamide-8 M urea gel along with G/A reaction prepared according to the protocol
of Maxam and Gilbert (
22
).
The 5' enhancerless
H2K
b
gene reveals a strong transcriptional activity dependent on an intragenic
sequence called
H2
downstream regulatory element (H2DRE) (
15
). The
Dde
I-
Kpn
I fragment (+517 to +806) encompassing parts of the second and third exons and
the whole second intron (+557 to +742) was shown to bear the main
transcriptional activation potential of the H2DRE. To identify the DNA sequence
that binds the Ltk- nuclear proteins we employed the methylation interference assay. Nine
guanines on the coding strand and seven guanines on the non-coding strand interfered with the complex formation when methylated. The
tightest contacts with DNA binding protein(s) were displayed by guanines 577
and 578 on the coding strand and guanines 580 and 582 on the non-coding strand (Fig.
1
). Inspection of this DNA region for potential
cis
-acting regulatory sequences using the SIGNAL SCAN program (
17
) revealed the sequence motif AGGTCA (+576 to +581) which represents the half
site of the consensus retinoic acid response element (RARE): PuG(G/T)TCA(N
1-5
)PuG(G/T)TCA (
23
). Furthermore, the sequence immediately upstream creates together with the
above mentioned motif an everted repeat and a direct repeat, respectively,
which we termed Hi-RARE (histocompatibility-intron RARE): 5'-
To test the functionality of the Hi-RARE sequence, we constructed two reporter plasmids. The whole second
H2K
b
intron with adjacent sequences (
Dde
I-
Sau
3AI fragment) was inserted in front of the CAT gene equipped with the chicken
conalbumin promoter. The 9C-RA inducibility (10
-8
and 10
-6
M) of the resulting pDS245conCAT reporter construct was assayed after
transfection into mouse P19 and F9 embryonal carcinoma cell lines and human
cervical carcinoma HeLa cell line. In contrast to HeLa cells and F9 cells, P19
cells showed maximum CAT expression at the lower dose of 9C-RA (Fig.
2
). The inducibility of the same construct in P19 cells was higher after 9C-RA treatment than after all-
trans
RA at the same concentration (10
-8
or 10
-7
M). This finding implies participation of the retinoid X receptors (RXR) in the
transactivating event (
24
,
25
).
To elucidate whether RA receptors are involved in transactivation, we
cotransfected the pHi-RAREmconCAT reporter plasmid (with single copy of the Hi-RARE sequence) with various combinations of mouse RARs and RXRs
expression vectors into P19 EC cells. The greatest response to 9C-RA induction was observed with the following combinations of RXR and RAR
vectors: RAR[beta]2-RXR[gamma] > RAR[alpha]1-RXR[gamma] > RAR[alpha]1-RXR[alpha] > RAR[beta]2-RXR[alpha]
(Fig.
3
). While the expression of RAR homodimers did not elicit a significant response,
the RXR homodimers induced small but reproducible increase of the reporter gene
expression (Fig.
3
, lanes 16-18). The pSS193conCAT reporter construct containing the 5'
H2K
b
gene enhancer (-254 to -61) with the previously described RARE (
10
) was co-expressed with the most efficient receptor combination (
12
) RAR[beta]2/RXR[beta] (Fig.
3
, lane 1) to serve as a positive control for the CAT activity. The control set
of P19 transient transfectants with the same plasmid combinations as in Figure
3
, but not treated with RA, did not exceed the values of the CAT activity found
in negative controls (data not shown). The plasmid pconCAT did not show any
response to 9C-RA when coexpressed with nuclear receptors RAR[alpha]1 and RXR[alpha] (Fig.
3
, lane 18). This finding implies that the Hi-RARE sequence inserted into the reporter plasmid (pHi-RAREconCAT) is responsible for the transactivation of the reporter
CAT gene through binding of retinoic acid receptors.
Mutations in the internal repeat or a deletion of the spacer guanine between the
internal and 3' repeat did not interfere with binding of 9C-RA-stimulated nuclear proteins, nor with the 9C-RA mediated response of CAT reporter vectors carrying
these mutated Hi-RAREs (M1, M2, Fig.
4
A). However, deletion of either 5' or 3' external repeat (M3 and M4 respectively, Fig.
4
A) totally eliminated the nuclear protein binding to the mutated Hi-RAREs (data not shown) and their functional response to 9C-RA (Fig.
4
B). The results strongly indicate that the Hi-RARE external half-sites are critical for its regulatory function. Evolutionary
conservation of the external half-sites but not of the internal repeat (see below) seems to support such a
conclusion.
Since the Hi-RARE sequence confers RA inducibility to a heterologous promoter and this
inducibility is further augmented by expression of RX and RA nuclear receptors,
we reasoned that binding of endogenous nuclear proteins to Hi-RARE might increase upon treatment with 9C-RA. To test this assumption, an electrophoretic mobility-shift assay was employed using the radiolabeled Hi-RARE oligonucleotide and nuclear extracts derived from
treated and control P19 EC cells. Nuclear extract from untreated cells yielded
three DNA-protein complexes in EMSA (Fig.
5
A, lane 2). Following the 9C-RA treatment, the middle complex increased 5.7-fold in quantity and a new more rapidly moving faint complex
occurred. In contrast, two major bands were not affected. Three out of four
bands were entirely competed with 100 molar excess of cold Hi-RARE oligonucleotide over the radiolabeled Hi-RARE probe. The H-2RII oligonucleotide with the 5' RARE of the
H2L
d
gene (
10
) displayed lower ability to compete. However, the RA-induced complexes were abolished by 100 molar excess of RARE[beta]2 oligonucleotide (Fig.
5
A, lane 4), which represents natural RAR/RXR binding site of the RARE[beta]2 gene promoter (
26
). Since the mobilities of RA-induced complexes with Hi-RARE probe and with the radiolabeled RARE[beta]2 probe were identical (data not shown), it seemed likely that
RA-induced complexes with Hi-RARE probe were created by RAR and/or RXR nuclear factors. The
conclusion was confirmed by supershift experiments with antibodies against RXR
and RAR. The RA-inducible complexes from P19 cells formed with Hi-RARE or RARE[beta]2 probes were specifically recognized by anti-RXR antibody as shown by the supershifts of specific
complexes (Fig.
5
B, lanes 4 and 9). Anti-RAR antibody decreased the amounts of RA-inducible complexes with both probes. The results confirm the
presence of the RXR and the RAR or a related cross-reacting factor in the RA-inducible P19/Hi-RARE complexes.
If Hi-RARE plays a significant role in the regulation of the MHC class I genes
then its sequence should be more conserved during evolution than adjacent,
supposedly function-less intronic sequences. To test this possibility we aligned an 81 bp
sequence from the second intron of the
H2K
b
gene that includes Hi-RARE plus 61 flanking nt, (+559 to +739) with the cognate intronic
sequences of other mammalian MHC class I genes available from the GenBank
database. The conservation of the Hi-RARE motifs, especially of the outer everted repeats (TGACCC and AGGTCA),
was quite striking. None out of 16 functional classical mouse class I (
H2K
,
-D
,
-L
) and
H2Q
genes revealed a mutation which would violate the canonical PuG(G/T)TCA
sequence of the outer repeats. A corollary to this finding is the relaxed
conservation of the same sequences found in the mouse class I pseudogenes as
five out of seven pseudogenes carried mutations that disrupted the consensus
sequence of the Hi-RARE repeats. The Hi-RARE degenerate GGGTCG motif internal to the everted repeats is less
preserved since it is disrupted in four mouse genes by single base
substitutions. Thus the mouse Hi-RARE consensus sequence appears to be TGACCC(N8)AGGTCA.
The 3' intron sequences flanking the Hi-RARE were more divergent than Hi-RARE and this tendency was even more obvious when human MHC
class I genes were examined. The outer everted repeat was preserved in 14 out
of 15 examined classical human class I genes and in seven out of nine non-classical class I genes and pseudogenes. The internal GGGTCG motif
preserved in many mouse
H2
class I genes was altered in all their human homologues with the exception of
HLA-A1
. Moreover, all HLA class I genes examined contain an additional pentanucleotide
insert, GCCNPu, inside the Hi-RARE so that the
HLA
Hi-RARE consensus sequence displays an 11-13 nt spacer: TGACCC(N11-13)AGGTCA.
The intact Hi-RARE was also conserved in MHC class I genes of the rat (
27
), rhesus macaque (
28
) and cat (
29
). In the rabbit (
30
) everted repeat was partially altered: TGACCC (N12) AGcTCg. In the dog (
31
), pig (
32
) and chicken (
33
) the Hi-RARE was mutated in both everted half sites but was still traceable. The
choice of species examined was delimited by the availability of class I genes
with the 2nd intron sequences in the GenBank database. A detailed sequence
comparison of the 80 bp fragment carrying the
H2K
b
Hi-RARE with all studied MHC class I genes is available from the authors
(jforejt@biomed.cas.cz).
In this report we identified and characterized a new putative retinoid response
element, referred to as Hi-RARE, in the second intron of the mouse
H2K
b
gene. High steady-state levels of transcription had been observed previously from the 5'-enhancer-less
H2K
b
gene transfected into the Ltk- cell line. A novel enhancer-like activity responsible for this transcription was localized in
the second intron and its flanking sequences (
15
). To analyze this activity further, we used nuclear extracts from the Ltk- cells in methylation interference assay and found protected guanines in
the sequence: 5'-
The 5'-flanking regions of
H2
class I genes harbor the sequence functioning as RAR[beta]/RXR[beta] response element (
10
-
12
,
35
) in embryonic cells and embryonal carcinoma cells. The intronic Hi-RARE also function as a retinoid response element as shown here by several
independent experiments.
Retinoid hormones control gene expression by binding specific nuclear receptors
which then function as ligand-activated transcription factors. These proteins bind to
cis
-acting RA response elements, RAREs, present in RA inducible genes. Two
families of such receptors, RARs [alpha], [beta], [gamma] and RXRs [alpha], [beta], [gamma] and their isoforms have been
characterized in great detail (
36
,
37
). They bind as homodimers or heterodimers to RARE which consists of two or more
directly repeated hexanucleotide motifs PuG(G/T)TCA. Binding of receptor
homodimers follows, to a certain extent, the `5-4-3-1' rule (
38
-
40
) which requires a 5 nt spacer (DR5) between two directly repeated RARE motifs
for binding the RARs, DR4 for thyroid receptors, DR3 for vitamin D3 receptor
and DR1 for RXR homodimers. Thus, the internal degenerate motif together with
the 3' motif of the Hi-RARE may function as binding sites for the RXR homodimers. However,
most of the receptors of the nuclear hormone superfamily form heterodimers, in
many instances with RXR as an auxiliary protein (
10
,
12
,
40
,
41
). They generally require larger spacers and
in vitro
they bind well to synthetic RAREs including inverted repeats (
42
), thus potentially fitting in with the structure of the Hi-RAREs in the mouse and human MHC class I genes.
The capacity of Hi-RARE to mediate an RA response via retinoid receptors was suggested by the
fact that the responsiveness to RA by the Hi-RARE driven reporter CAT gene was dramatically increased by cotransfection
of RAR and RXR expression vectors. The most potent heterodimers were those
involving RXR[gamma] and RAR[beta]2 or RAR[alpha]1. RXR homodimers induced only a moderate response. This
analysis strongly suggests that the Hi-RARE belongs to the family of retinoid response elements. This is further
supported by the results of electrophoretic mobility shift and supershift
assays which revealed that RA treatment of P19 cells induces nuclear proteins
that bind to the Hi-RARE oligonucleotide and that these proteins are recognized by anti-RAR and -RXR antibodies. Moreover, the recombinant RAR/RXR heterodimer
binds to the Hi-RARE with the affinity comparable with the RARE, its natural target in the
promoter of the RAR[beta] gene.
Taken together, the experimental data provide strong evidence that the Hi-RARE functions as a retinoid response element defined by its ability to
confer RA responsiveness to a heterologous promoter. However, this evidence is
not sufficient for determining the physiological function of the Hi-RARE at its autochthonous position in the
H2K
b
gene. The superfamily of nuclear receptors includes, besides RARs and RXRs,
many other closely related proteins which can use each other's response
elements rather promiscuously (
17
). Also, there is a number of orphan receptors, such as COUP, PPAR, MB67 (see
43
for review) for which the ligands are not known and which alone or with RXR as
an auxiliary protein could contribute to the
H2K
b
gene regulation.
To attribute a functional role to a non-coding DNA sequence, its evolutionary conservation or diversification can
be a critical clue. Based on this criterion, the fact that none out of 15
examined mouse class I genes carried any mutation in the 5'-TGACCC(N8)AGGTCA-3' Hi-RARE motif uncovers the functional importance of
this sequence element. In addition, five out of seven mouse class I pseudogenes
show mutations that disrupted one or both inverted palindromic hexanucleotide
motifs which is another piece of evidence that conservation of this sequence is
necessary for a function of the
H2K
b
gene and not, for example, for some unrelated structural feature of a
particular DNA domain. The computer assisted analysis revealed that a Hi-RARE with the same inverted palindromic core motifs is highly conserved
also in human HLA genes, since the motifs are intact in 21 out of 24 HLA class
I genes. The Hi-RARE sequence is also preserved in the second introns of MHC class I genes
of the rat, cat and rhesus macaque but not in the class I genes of rabbit, dog
and pig. All non-rodent MHC class I genes analyzed in this study carried an insertion of
the 5'-GCCNG-3' pentanucleotide in their Hi-RAREs so that the consensus included the longer
spacer: 5'-TGACCC(N11-13)AGGTCA-3'. The internal, partially degenerate motif 5'-GGGTCG-3' is much less
preserved in evolution. Thus the comparison of the second introns clearly
points to the external repeats as being the most conserved parts of the Hi-RARE. This conclusion fits nicely with the mutational analysis of the Hi-RARE which showed that the deletion of either of the external
inverted palindromic repeats results in loss of function of the mutated Hi-RARE.
Such a consensus, including two inverted repeats and a long internal spacer is
far from the classical structure of RARE with two or more direct repeats and 1-5 bp long spacers (
39
,
40
). On the other hand, it closely resembles the structure of the [gamma]F-RARE of the [gamma]F-crystallin gene (
34
). Moreover, synthetic palindromic and inverted palindromic repeats are
efficient transactivating response elements for RARs, T3Rs and VDRs and require
longer spacers (3-12 nt) for optimal function (
42
).
Evidence is accumulating that genes other than MHC class I have inverted
palindromic or palindromic RA response elements. Inverted palindromic repeats
were described in the RARE of the promoter regions of the mouse [gamma]F-crystallin gene (
34
), the human medium chain acyl-coenzyme A dehydrogenase gene (
44
) and as a part of the chicken lysozyme silencer (
45
) (Fig.
7
).
We thank Pierre Chambon for the generous gift of expression vectors of murine
retinoic acid, retinoid X receptors, and antibodies against RAR and RXR. We
thank Joseph Grippo and Arthur Levin for providing us with 9-
cis
retinoic acid, and Hendrik Stunnenberg and Rafael Valcarcel for purified His-tagged RAR[alpha]/RXR[alpha] proteins and for RARE[beta]2 oligonucleotide. We also gratefully acknowledge
Shirley Tilghman, Nancy Garvey, Tom Vogt and Ingrid Grummt for critically
reading the manuscript and Jarmila Kralova and Jiri Hatina for helpful
comments. This work was supported by grant 55273 from the Academy of Sciences
of Czech Republic. J.F. is the International Research Scholar of the Howard
Hughes Medical Institute.
REFERENCES
Return