ABSTRACT
Regulation of the human MHC class I HLA-A11 promoter is governed by a complex array of regulatory elements. One of
these elements, shown here to be critical for the transcriptional activity of
the promoter, was used to screen a [lambda]gt11 library and allowed the identification of a cDNA which coded for the zinc
finger protein ZFX. ZFX was shown to bind the sequences AGGGCCCCA and
AGGCCCCGA, located respectively at positions -271 to -263 and -242 to -234 of the HLA-A11 promoter, with similar affinities through
its three C-terminal zinc fingers. ZFX
575
, a short isoform of ZFX, activates transcription from the HLA-A11 promoter in a Leydig cell line.
Major histocompatibility complex (MHC) class I genes encode highly polymorphic
molecules present on the surface of most nucleated vertebrate cells with the
exception of some specialised cell types, such as neurones, corneal cells,
pancreatic acinar cells and mature sperm cells (
1
).
In many instances, the presence of functional products of the human HLA-A and HLA-B loci is critical when a cell must be recognised as abnormal
(either malignant or virally infected) in order to be eliminated by the immune
system. Intracellular association of endogenous peptides which originate from
proteolytic degradation with the polymorphic heavy chain of MHC class I
molecules and the monomorphic light chain ([beta]2-microglobulin) is accomplished during routing to the cell membrane,
where the complex can eventually be recognised by cytotoxic T lymphocytes via
their T cell receptor (
2
).
Some viruses have evolved specific mechanisms leading to down-regulation of MHC class I in infected cells (
3
) and this has been shown to be critical for the elimination or survival of
virally infected cells (
4
). Occasional evidence for the expression of MHC class I favouring the expansion
of tumour cells by promoting escape from NK cell lysis has been reported (
5
). However, the frequent loss of MHC class I expression in many tumours (
6
) and the selective loss of expression of specific alleles in some of them (
7
,
8
) is consistent with the escape of class I-negative tumour cells from an immune process operating against transformed
cells which express HLA class I genes. Indeed, therapies aimed at indirectly
stimulating or at re-expressing MHC class I molecules on the surface of defective cancerous
cells have proved successful (
9
). The expression level of MHC class I genes varies among different tissues and
can be regulated by various effectors, including retinoic acid (
10
), TNF (
11
), IFN (
12
) and hormones (
13
).
In many instances these effectors have been shown to act ultimately at the
transcriptional level. Some proteins, such as members of the rel and STAT
families (
14
-
16
), have been repeatedly implicated in the transcriptional control of MHC class I
genes. They act through DNA elements which are fairly well conserved both intra- and inter-specifically. However, the rapid evolution and extensive
polymorphism of MHC class I genes is not restricted to the coding portion of
the genes and is also observed in their control regions (
17
). Although at present the implications of this fact cannot be fully
appreciated, it is already known that it can account for locus- or allele-specific regulation by effectors such as IFN (
18
,
19
).
Previous analysis of transcriptional regulation of the HLA-A11 gene indicated that in addition to a conserved region which is the
target of rel, STAT and IRF family factors, five other control elements within
the first 337 nucleotides (nt) of the promoter also exert significant effects
on transcription (
20
).
We report here that the mutation of one of these elements dramatically reduces
transcription from the HLA-A11 promoter. Using an oligonucleotide spanning this region, we screened a
HeLa [lambda]gt11 library and isolated a partial cDNA corresponding to the C-terminal portion of the ZFX protein. The binding mode of
recombinant ZFX fusion proteins to the HLA-A11 element was also characterised. We demonstrate that a short isoform of
ZFX, ZFX
575
(
21
), can be a potent transactivator of the HLA-A11 promoter. In addition to the identification of a novel transcriptional
effector of the HLA class I genes, this work constitutes the first
demonstration that the ZFX gene indeed encodes a
bona fide
transcription factor, as initially proposed (
21
,
22
).
A cDNA library of HeLa cells in [lambda]gt11 (Clontech) was screened essentially as described (
23
), except that poly(dI[middot]dC)[middot](dI[middot]dC) was used as a non-specific competitor in 500-fold excess along with an unrelated
oligonucleotide in 100-fold excess in 0.5% BSA, 10 mM HEPES, pH 7.5, 50 mM KCl, 0.5 mM MgCl
2
, 10 [mu]M ZnCl
2
, 5% glycerol and hexamerised -272/-233 double-stranded oligonucleotides, labelled with [[gamma]-
32
P]ATP and used at 5 ng/ml: -272/-233C, 5'-GATCCAGGGCCCAGGCGTGGCTCAGGGTCTCAGGCCCCGAA-3'; -272/-233NC, 5'-GATCTTCGGGGCCTGAGACCCTGAGAGCCAGCCTGGGGCCCTG-3'. Spots superimposed on replicates were processed through secondary and tertiary screening and their inserts identified after subcloning. The full-length ZFX
575
was amplified on human ovary cDNA (Clontech) with the 5' primer oligonucleotide 1 (5'-GGGAATTCGCCGCCAACCGTATGATGTTCCAGACTATGCTGACACAGAGCGGAAATTGAT-3'), which introduces a nine amino acid HA tag at
the N-terminus, and the 3' primer oligonucleotide 2 (5'-CGGAATTCGGATCCAAGGCCAATATCTCACAAACG-3') with a mixture of 1/25 Pfu (Stratagene)
and Taq (BRL) (
24
). Amplified fragments cloned in PSG5 (Stratagene) were sequenced and found to
bear a mean of two mutations per clone. A full-length cDNA was reconstituted by conventional methods, using unique
internal restriction sites for
Pml
I and
Xcm
I. During the course of this work we obtained a cDNA that was lacking nucleotide
2071 (numbered according to
21
); this frameshift mutation precisely located after the end of the ninth zinc
finger introduces a nonsense codon at position 2111. The resulting cDNA encodes
a protein deleted of its four C-terminal zinc fingers, referred to as ZFX
575
[Delta]C. GST fusion vectors were constructed after amplification of different
fragments using 100 ng of cloned ZFX cDNA as the substrate. The inserts of the
various constructs were amplified as follows: using oligonucleotides 3 and 7
for GST4ZFX, oligonucleotides 4 and 7 for GST3ZFX, oligonucleotides 5 and 7 for
GST2ZFX and oligonucleotides 4 and 6 for GST3ZFX[Delta]C: oligonucleotide 3, 5'-GGGTGATCAGGATCCAGATCTAGCCCGTCAGTATCGGCCGA- 3'; oligonucleotide 4, 5'-GGGAATTCGGATTCCGGCACAAAGGATCTTCCATTTAG-3'; oligonucleotide 5,
5'-GGGA- TTCGGATCCGGCAGTGGCAGGAAAGT-3'; oligonucleotide 6, 5'-GGGGAATTCT- TAGTCTTTCGTGTGAATGGAAATAACG- TGC-3';
oligonucleotide 7, 5'-CGGAATTCGGATCCAAGGCCAATATCTCACAAACG-3'. The 6ZFX insert was amplified using oligonucleotides
8, 5'-GCGGGATCCACCAAGAAAGCAAAAC-3'; and 9: 5'-GCTCTAGAAGCTTAGGGCAGGCCAACTTC-3', digested with
Bam
HI and
Hin
dIII and cloned into 24KYH digested with
Bam
HI and
Hin
dIII (24KYH is a high copy number prokaryotic expression vector, in which
expression is driven by a T7 promoter controlled by the lac repressor; details
concerning its organisation are available separately on request). The resulting
vector, 24KYH6ZFX, enables production of a protein beginning with MKYHHHHHHGS,
followed by the amino acids translated from the sequence of the insert. An
Eco
RI fragment amplified in two steps with the oligonucleotides 10 (5'-GATGTTCCAGACTATGCTCCCAAGAACGTCGTCGTCCTGATGGATTCCGGCACAAAGGATCTTCCATTTAG-3') and 7 followed by amplification with
oligonucleotides 11 (5'-GGAATTCAGATCTGCCACCATGGAGTATCCGTATGAGTTCCAGACTATGCTCC-3') and 7 was cloned in the PSI vector (Promega). This
produced the vector PSIN3ZF, which allowed translation of a protein beginning
with MEYPYDVPDYAPKKRRRPDGS, followed by the three C-terminal zinc fingers. Generation of the mutated HLA-A11 promoters was achieved by two PCR steps. The first step
consisted of amplifying the -337/-260 region while introducing a
Bgl
II site at -266 with oligonucleotides 12 (5'-AAACTGCAGGACTCAGGGAGACATTGAGACA-3') and 13 (5'-GAGAGCACCCGACG
BL 21 DE3 bacteria bearing PGSTZFX fusions or 24KYH6ZFX were grown in L broth
supplemented with 10 [mu]M ZnCl
2
to an OD of 0.8 at 550 nm; 0.2 mM IPTG was then added to the medium and growth continued at 37oC for 3 h. For PGSTZFX, the bacteria were centrifuged and resuspended in a 1/20 vol. of 50 mM Tris, ph 7.5, 150 mM NaCl, 0.5 mM MgCl
2
, 10 [mu]M ZnCl
2
, 0.1 mM PMSF, and 50 mM benzamidine, sonicated on ice and centrifuged at 12 000
g
for 15 min. The supernatant was adjusted to 2 mM CHAPS and applied to GSH-agarose (1 ml:10 ml supernatant); followed by extensive washing with 100
ml 20 mM Tris, pH 7.5, 200 mM NaCl, 10 [mu]M ZnCl
2
(buffer A), then 20 ml buffer A adjusted to 1.6 M NaCl, then again with 10 ml
buffer A. The column was then resuspended in 3 vol. 50 mM glycine, pH 9.2, 1 mM
CaCl
2
, 10 [mu]M ZnCl
2
and 0.1 U/ml nuclease S7 (Boehringer). Reactions were checked by monitoring the
OD variations of the supernatant (stable after 10 min) and allowed to proceed
for 1 h. The column was then washed again by alternating buffer A, buffer A +
1.6 M Nacl and buffer A. The proteins were then batch eluted in 20 mM Hepes, 200 mM NaCl, 0.5mM MgCl
2
, 10 [mu]M ZnCl
2
, 1 mM CHAPS and 5 mM GSH (1 vol.). The eluate was adjusted to 50% glycerol and
stored at -20oC. In the case of 24KYH6ZFX, the cells were centrifuged, resuspended
in 1/20 vol. of Tris 50 mM pH 7.5, 1 M NaCl, 5 mM benzamidine and 0.1 mM PMSF and sonicated on ice. The homogenate was centrifuged at 4000
g
for 5 min. The supernatant was discarded and the pellet sonicated in 50 mM Tris
pH 8.9, 1% Triton X-100 and 5 mM EDTA. After centrifugation, the supernatant was discarded and
the pellet re-extracted twice in the same way with 1.5 M urea, then solubilized in buffer B (6 M guanidine and 10 mM imidazole pH 7.5). The 6 M guanidine extract was loaded on NI
++
-agarose equilibrated in buffer B (1/10 v/v extract) and the column was
washed extensively with 10 vol. buffer B, followed by 10 vol. buffer B ajusted
to 40 mM imidazole pH 7.5. Finally, the proteins were eluted with 3 vol. buffer
B ajusted to 120 mM imidazole pH 7.5, and 10 [mu]M ZnCl
2
. The proteins adjusted to an OD of 0.4 at 280 nm were successively dialysed
against 3 M guanidine, 20 mM Tris pH 7.5, 0.5 mM MgCl
2
and 10 [mu]M ZnCl
2
and then against the same buffer with 1.5, 0.75 and 0.4 M guanidine and finally
without guanidine at 4oC. The proteins were stored at -20oC with 50% glycerol.
EMSA and DMS interference were carried out essentially as described (
20
). For DMS protection, DNA binding was allowed to proceed for 15 min on ice in
the presence of 1 [mu]g poly(dI[middot]dC)[middot](dI[middot]dC), 500 ng unlabelled heterologous oligonucleotide
and 1 ng labelled fragment; 0.25 [mu]l DMS was then added to the 20 [mu]l reaction, along with a 500-fold excess of unlabelled homologous fragment. The reaction was
allowed to proceed for 30 s and stopped by separation of the components by PAGE
at a field strength of 10 V/cm. After 90 min migration, the free and complexed
DNAs were eluted and processed through piperidine cleavage as described (
25
). The oligonucleotides used in these experiments were: -282/-224C, 5'-GGCGAAGTCCCAGGGCCCCAGGCGTGGCTCTCAGGGTCTCAGGCCCCGAAGGCGGTGT-3'; -282/-224 NC, 5'-TACACCGCCTTCGGGCCTGAGACCCTGAGAGCCACGCCTGGGGCCCTGGGACTTCGC-3'; -255/-224C, 5'-GCTCTAGGGTCTCAGGCCCCGAAGGCGGTGT-3';-255/-224 NC, 5'-TACACCGCCTTGGGGCCTGAGACCCTAGAG-3'; -282/ -250C, 5'-GGCGAAGTCCCAGGGCCCCAGGCGTGGCTCT-3'; -282/-250NC, 5'-TAGAGCCACGCCTGGGGCCCTGGGACTTCGC-3'; -282/-250BIIC, 5'-ggcgaagtcccagg-
gcagatctcgtggctct-3'; -282/-250BIINC, 5'-tagagccacgagatctgccctgggacttcgc-3'; -255/ -224ATC, 5'-GCTCTAGGGTCTCTGGCCCCGAAGGCGGTTGT-3'; -255/-224TANC, 5'-TACACCGCCTTCCGGGGGCCAGAGACCCTAGAG-3'; -255/-224XbaC, oligonuc-
leotide 16; -255/-224XbaNC, oligonucleotide 17. The results of DMS interference and
protection were quantified using ImageQuant software and the Molecular Dynamics
Personal Densitometer.
The JF (B-EBV) cell line was maintained and used in RPMI with 10% FCS. The TM3 cell
line was obtained from ATCC and passaged in DMEM/F12 medium, 10% FCS and 5%
horse serum. The cells were seeded in 10 cm
2
plates 24 h prior to transfection in DMEM medium without phenol red and supplemented with 2 [mu]M ZnCl
2
plus 10% foetal calf serum pre-treated with activated charcoal (FCSC). Transfection was performed using
DOTAP (Boehringer), as recommended by the manufacturer, in a seeding medium
adjusted to 2.5% FCSC. For each transfection, 2 [mu]g of the expression vector, 0.2 [mu]g of the luciferase reporter and 0.04 [mu]g Tk [beta]Gal or 0.02 [mu]g pSV[beta]Gal vector were used. After 6 h incubation with liposomes, the transfection medium
was removed and replaced by the seeding medium for 36 h. Cells were lysed with 200 [mu]l of reporter lysis buffer (Promega) and luciferase activity was determined
using the luciferase assay system (Promega). [beta]-Galactosidase activity was determined according to the manufacturer,
using the Galactolight kit (Tropix) downscaled 3-fold, and was used for normalisation of the luciferase activity. CAT
activity was measured as previously described (
20
) and quantified with a [beta] Imager 3600, using [beta]-Vision software (Biospace). All transactivation assays were
repeated several times using at least two different plasmid preparations.
A previous study undertaken in our laboratory has shown that the HLA-A11 promoter region -273/-205 binds effectors that activate transcription. Analysis of
DNA binding activities from JF, a B-EBV cell line, and from a HeLa cell line demonstrated that several
proteins come into contact with DNA in the -273/-233 region. Six bases involved in contacts with at least two
different proteins in the -266 to -261 region were mutated by introduction of a
Bgl
II restriction site (Fig.
1
). CAT constructs driven by the mutated promoter -337/+2BII were compared with CAT constructs bearing the wild-type -337/+2 promoter.
A [lambda]gt11 cDNA library from HeLa cells was screened with a radiolabelled and
hexamerised -272/-233 oligonucleotide encompassing the site identified by mutation.
One clone was identified and found to contain an insert with 100% identity to
the C-terminal region of ZFX, starting in the middle of the ninth zinc finger of
ZFX (Fig.
2
A). This indicated that four complete zinc fingers at the C-terminus of ZFX, which contains 13 zinc fingers, are sufficient for
binding to the promoter element. Subcloning the coding portion of the [lambda]gt11 cDNA in a PGEX vector allowed purification of the fusion protein on
GSH-agarose and facilitated subsequent analysis of its DNA binding activity,
whose specificity was ascertained by competition with homologous and
heterologous oligonucleotides (Fig.
2
B). The minimal number of zinc fingers required for DNA binding was established
by constructing various fusion products. Deleting the partial ninth and the
tenth zinc fingers (PGST3ZFX) significantly increased the specific activity of
the DNA binding protein. Deletion of the eleventh zinc finger (PGST2ZFX)
resulted in a protein of low activity, whereas retaining the eleventh and
twelfth zinc fingers and deleting the thirteenth (PGST3ZFX[Delta]C) abolished DNA binding (Fig.
2
B).
The interactions of ZFX with the HLA-A11 promoter were more precisely determined by probing the interactions of
GST3ZFX with DNA by means of methylation interference. The DNA molecules eluted
from the two retarded complexes were compared with the free DNA. Analysis of
the upper strand from the faster migrating complex indicated a weak
interference of two groups of bases separated by 25 nt, namely AGGG, from -271 to -268, and AGG, from -242 to -240 (Fig.
3
A, lanes 5, 7 and 8). The lower strand of the same complex showed contacts at GG
(-267, -266) and 26 nt away at GGG (-239 to -237), (Fig.
3
A, lanes 1, 3 and 4). Considering what is already known about the organisation
of zinc fingers and their binding to DNA, we hypothesised that two sites are
present in this region, one at -265 and the other at -238. Quantification of several interference results indicated that
binding at the two sites in the faster complexes resulted in a near 50%
interference at each of the two sites. This is consistent with a 1 protein/1
DNA complex in which the two DNA sites are bound with equivalent affinity.
Quantification of the interference pattern of the slower complexes compared
with the free DNA, which was in large excess, is fully consistent with a 2
protein/1 DNA complex, since the interference values for bases of the two sites
exceeded 80% (Fig.
3
A, lanes 1, 2 and 4 and 5, 6 and 8). We further investigated the relative
affinities of the two sites by means of the DMS protection assay, which was
carried out on the 1 protein/1 DNA complex under conditions which disfavour the
2/1 complex according to the EMSA analysis. The DNA located at the positions of
the free probe (which was in large excess) and of the 1/1 complex were eluted,
cleaved with piperidine and analysed on a denaturing gel (as expected, the 2/1
complex was not observed in this preparative gel; not shown). Analysis of
protection data (Fig.
3
B) indicated that the protected bases (-268, -241 and -240 on the upper strand and -267, -266, -239 and -238 on the lower strand)
overlapped with the bases identified by interference. More surprisingly,
quantification of the protection indicated that the 50% maximal protection at
each site for a 1/1 complex with two equivalent binding sites was exceeded. The
reactivity of the two most protected bases, -239 and -266 of the lower strand, were reduced 3.6-fold and 4.1-fold, respectively. This indicated that the 2/1
complex was most probably formed in solution at the time of contact with DMS,
but that it was too unstable to be effectively caged by the gel and was
dissociated into the 1/1 complex when loaded on the gel. Two bases of the non-coding strand were hyper-reactive: -236 (2.9-fold) and -264 (1.7-fold). The similar reactivity of the N7 of
one guanine 5' of the two sites could have been due to a similar mobility or
deformation introduced in the DNA backbone by the zinc fingers (
26
).
A cDNA which codes for ZFX
575
, a short isoform of ZFX, was amplified by PCR of human ovary cDNA. A full-length cDNA was cloned in the eukaryotic expression vector PSG5. The
effects of the PSGZFX
575
and PSG5 vectors on transcription were compared by co-transfection assay with HLA-A11 promoter-driven reporters in various cell lines (JF, Cos-7, Cos-6, HeLa, 293, T47D, TM3 and TM4; data not shown).
Only the murine Leydig cell line TM3 allowed reproducible analysis of the
transcriptional effects of ZFX. In this cell line, PSGZFX
575
was found to stimulate transcription of the -337/+2, -273/+2 and -242/+2 promoter constructs by a mean value of ~20-fold (Fig.
5
). PSGZFX
575
had a moderate effect on the -205/+2 promoter construct, increasing its activity 2-fold. Introduction of the mutation, which was shown to abolish
binding of ZFX to the -282/-250 region (Fig.
4
A, lanes 8-10) in the context of the -337/+2 promoter, reduced its basal activity by 8-fold in TM3 cells and precluded transactivation by PSGZFX
575
(Fig.
5
). On the other hand, modification of the downstream site in the context of the -337/+2 promoter by a mutation shown to abolish binding of ZFX to the -255/-224 region significantly affected neither the basal
activity nor the stimulated activity of the promoter (Fig.
5
, -337/+2Xba). As expected, the stimulatory activity of ZFX
575
was lost when the four C-terminal zinc fingers were deleted. Normalised promoter activities were
similar with the empty PSG5 vector and with the PSGZFX
575
[Delta]C expression vector bearing the deleted cDNA. However, a 2- to 3-fold co-stimulation of [beta]-galactosidase and luciferase activities
were reproducibly noted with this construct (not shown) and may reflect
increased cell survival or increased expression of the co-introduced genes.
We show here that mutation of an element centred at -265 of the HLA-A11 promoter has a profound effect on its transcriptional activity.
Oligonucleotides encompassing this region were used to screen a [lambda]gt11 library; a partial cDNA was isolated and found to correspond to the
C-terminal portion of ZFX. Several years ago, ZFX was identified as an X
chromosome gene that was highly homologous to ZFY, formely thought to be a
strong candidate for testis-determining factor (TDF) (
30
). The role of ZFY as TDF was definitively ruled out when genetic analysis and
transgenic techniques demonstrated that in fact SRY was TDF (
31
). The functions of both ZFY and ZFX remained elusive, although a role for them
as transcription factors has been proposed on the basis of their gene sequences
(
21
,
22
), and later on the basis of the transcriptional activity of constructs which
fused the N-termini of ZFX and ZFY to the GAL4 DNA binding site (
32
). Mammalian ZFX transcripts are ubiquitous (
22
,
33
) and derive from at least four untranslated 5' exons and seven coding exons. The acidic N-terminal domain and the nuclear localisation signal are encoded by
exons 5-10, while all 13 zinc fingers are encoded by exon 11. Transcripts
containing all exons from 5 to 11 encode the largest ZFX protein, ZFX
804
, while alternative transcripts missing exon 5 encode ZFX
575
, a protein initiated in exon 7 (
21
). Other alternative transcripts have been observed in the mouse, in which exons
7 or 10 are spliced out (
34
). We have focused on the functional study of ZFX
575
and our study conveys two messages, that, as previously hypothesised, ZFX indeed
regulates transcription and that ZFX is capable of regulating HLA-A11 class I gene transcription.
The identification of a DNA binding site specifically recognised by the C-terminal zinc fingers of ZFX was the first step in obtaining experimental
evidence of the involvement of ZFX in transcription. The terminal three zinc
fingers of ZFX were shown to be sufficient for binding two sites on the HLA-A11 promoter, 7 and 8 bases of which respectively were shown to be
affected in their reactivity toward DMS. This is consistent with the view that
a single C
2
H
2
zinc finger frequently recognises three contiguous bases (
28
,
29
). Efforts to understand the relationship between the amino acid composition of
the conserved helix of C
2
H
2
zinc fingers and the nature of the bases recognised have been made in several
laboratories (
35
-
40
). A perfectable code is now emerging which, when applied in the three C-terminal zinc fingers of ZFX (and of ZFY, whose DNA binding properties
were found to be indistinguishable from those of ZFX; not shown), predicts the
A
1
C
2
G
3
G
4
X
5
C
6
X
7
C
8
A
9
recognition site in preference to T
1
C
2
G
3
G
4
X
5
C
6
X
7
C
8
T
9
(Table
1
). The site determined at -265, AGGGCCCCA, departs from this prediction at position 2, while the
site at -238, AGGCCCCGA, departs at positions 2, 4 and 8, but was experimentally
found to be of similar affinity when compared with the -265 site. Interestingly, the nonamer AGGGTCTCA, found between the two
sites, better matches the prediction, but no evidence of binding to this
sequence was observed. Using EMSA, efficient specific DNA binding was observed
with purified fusion proteins containing either six (not shown) or fewer C-terminal zinc fingers. However, we have been unable to detect the DNA
binding activity of ZFX
575
by EMSA of crude nuclear and cytoplasmic extracts of transfected TM3 cells or
other cells (COS-7 and 293). The same holds true for the product of the PSIN3ZF construct.
We suspect that an abundant inhibitor masks the DNA binding site of ZFX when
the cells are disrupted. Similarly,
in vitro
transcription/translation did not allow detection of ZFX
575
DNA binding activity, while the DNA binding activity of the three C-terminal zinc finger proteins was detected only very transiently soon
after translation. RNase treatment of cellular extracts as well as of
in vitro
translation products were ineffective in rescuing the DNA binding activity (not
shown). Nevertheless, the demonstration of transcriptional activation, which
necessitates both the DNA binding site characterised in the HLA-A11 promoter and overproduction of ZFX
575
in TM3 (Fig.
5
), implies that, in undisrupted cells, at least a portion of the ZFX
575
molecules are capable of recognising their DNA targets. In the context of the
HLA-A11 promoter, the PSIN3ZF construct, which has no transactivation domain,
was shown to repress transcription. However we have also identified two other
promoters (not shown) that are regulated by ZFX, one of which is stimulated by
PSIN3ZF and PSGZFX
575
, the other repressed by PSGZFX
575
. Thus, the ZFX gene might be able to influence transcription by a variety of
mechanisms, depending on the promoter and the context provided by the cell line
or tissue. In this respect, it is noteworthy that the magnitude of
transactivation was much higher in TM3 than in any other cell line. The
exacerbated activity of a gene borne by a sex chromosome in specialised cells
of a sexual tissue suggests its involvement in sex-specific regulation. However the physiological significance of this
observation remains to be clarified, and it is possible that the ubiquitous
distribution of ZFX transcripts reflects a general function in cell
maintenance.
Table 1
Whether ZFX
575
can interact in TM3 cells with an activating partner that is specifically
abundant in such cells or whether a widespread inhibitor of ZFX is lacking
there is at present unknown, but this issue is amenable to investigation. Our
results suggest that the DNA binding activities of ZFX proteins are
undetectable in EMSA of cellular extracts under the usual conditions. However,
we believe that the transactivation assay conditions described here can provide
a basis for comparing the transcriptional activities of the various ZFX and ZFY
gene family members and their isoforms.
The -273/-250 region of the HLA-A11 promoter has been previously described as a binding site
for AP2 and at least one other protein (
20
). We show here that mutating this DNA binding site results in a profound
weakening of the promoter's activity in all the cell lines tested. Some of the
interactions which take place at this site are therefore important for
controlling transcription of the HLA-A11 promoter. We show that ZFX is able to contact some of the bases
identified in our previous studies, but produces an interference pattern
clearly different from any protein so far identified. ZFX also binds a second
site centred at position -238. The two binding sites identified in the -273/-205 region of the HLA-A11 gene do not contribute to ZFX transcriptional
activation of the various promoter constructs in the same way. In the -337/+2 promoter context, mutation of the -265 site strongly reduced transcriptional activity and no longer
allowed transactivation by ZFX. Therefore, in this case, the remaining intact
site at -238 appeared to be silent. Conversely, comparison of the transactivation
magnitude of the -273/+2 construct containing the two sites and the -242/+2 construct containing only the downstream site led to the
conclusion that ZFX can transactivate the promoter through interactions with
the -238 site and that further adjunction of the -265 site within the -273/+2 construct results in no additional increase in the
transactivation magnitude. Thus, in this case, it is the -265 site which seems to be silent, although this is only the simplest
interpretation of the data. These observations challenge the simple conceptions
of transcriptional control regions as modular additive elements and emphasise
the importance of stereospecific interactions in the assembly of transcription
complexes. This matter was recently reviewed and investigated and led to the
proposal of the enhanceosome as a cooperatively built up and highly ordered
transcription complex (
41
). It is possible that sequences between -337 and -273 bind effectors which contribute to shaping the transcription
complex in a different way when compared with the transcription complex
assembled on the -273/+2 or -242/+2 promoters, hence the different contributions of the two
binding sites in these different contexts, despite remarkably similar
magnitudes of transactivation by ZFX. On the other hand, it is also possible
that the
Bgl
II mutation at -265 has pleiotropic effects, not only precluding ZFX binding at this site
but also other key effectors, or perturbing a critical local bending of DNA.
The interference pattern of ZFX obtained here at the -265 site and the AP2 interference pattern previously described at the
same site (
20
) strongly suggest that their binding at this site is mutually exclusive. It
will be of interest to study the functional consequences of AP2 and ZFX
interactions on the HLA-A11 promoter by means of co-transfection studies.
The AGGGCCCCA sequence centred at -265 is well conserved among HLA-A alleles and is also found in HLA-G and HLA-E genes. However, it is less significantly conserved
among HLA-B and HLA-C alleles, in which this region is quite polymorphic. This sequence
is not known in the homologous position of the murine MHC class I genes.
However, the whole spectrum of DNA binding recognition by ZFX is not known at
present, thus it cannot be decided whether ZFX exerts an allele- or locus-specific regulation or whether it acts more generally on the various
promoters of HLA Class I genes.
We thank Thierry Salomon of Molecular Dynamics for quantification of DMS
interference and protection data, Claude Hennion (Biospace) for help with
quantification of CAT assays, Noah Hardy for re-reading the manuscript and Cynthia Calabresse for help with designing the
figures. We also wish to acknowledge the contribution of Laura Brown and David
Page for sharing material with us. This work was supported by the Association
Franaise contre les Myopathies (AFM) and by the Institut National de la Santé et de la Recherche Médicale (INSERM).
REFERENCES
Return