ABSTRACT
The human granulocyte-macrophage colony stimulating factor (GM-CSF) gene promoter binds a sequence-specific single-strand DNA binding protein termed NF-GMb. We previously demonstrated that the NF-GMb binding sites were required for repression
of tumor necrosis factor-[alpha]
(TNF-[alpha]) induction of the proximal GM-CSF promoter sequences in fibroblasts. We now describe the isolation of
two different cDNA clones that encode cold shock domain (CSD) proteins with NF-GMb binding characteristics. One is identical to the previously reported CSD
protein dbpB and the other is a previously unreported variant of the dbpA CSD
factor. This is the first report of CSD factors binding to a cytokine gene.
Nuclear NF-GMb and expressed CSD proteins have the same binding specificity for the
GM-CSF promoter and other CSD binding sites. We present evidence that CSD
factors are components of the nuclear NF-GMb complex. We also demonstrate that overexpression of the CSD proteins
leads to complete repression of the proximal GM-CSF promoter containing the NF-GMb/CSD binding sites. Surprisingly, we show that CSD overexpression can also directly repress a region
of the promoter which apparently lacks NF-GMb/CSD binding sites. NF-GMb/CSD factors may hence be acting by two different mechanisms. We
discuss the potential importance of CSD factors in maintaining strict
regulation of the GM-CSF gene.
The expression of cytokines in specific cell types and in response to specific
stimuli is tightly regulated. Control of gene transcription plays a major role
in this regulation (
1
). Numerous transcription factors have been identified that mediate the
activation of cytokine genes, including C/EBP, NF-[kappa]B, AP1, NF-AT and ets factors (
2
-
4
). As well as responding to activators, it is important that cytokine genes can
be maintained in a repressed state in the absence of activation signals and
that these genes can be completely inactivated upon withdrawal of activation
signals. The mechanisms of cytokine gene repression have not been well
characterized. Repressor elements have, however, been identified in a number of
cytokine and immune function gene promoters, such as in the interleukin-1[beta] (IL-1[beta]), IL-2, IL-3, IL-8, tumor necrosis factor-[alpha] (TNF-[alpha]), IL-2
receptor-[alpha], interferon-[beta] (IFN-[beta]), LD78, VCAM-1 and ELAM-1 promoters (
5
-
11
). In most cases little is known about the nature or function of the proteins
binding to these elements. Cloning and characterization of the IRF-2 repressor of IFN-[beta]- and IFN-inducible genes has however demonstrated that it
functions by competition with an activator (
12
). It has also been demonstrated that the repressor of the ELAM-1 gene (ATF-a0) acts by heterodimerizing with an activator and preventing its function (
10
).
The gene for the cytokine granulocyte-macrophage colony stimulating factor (GM-CSF), is expressed in a wide variety of cells, including fibroblast
and endothelial cells, in response to TNF-[alpha] and IL-1 and in activated T cells (
3
,
13
-
15
). The proximal promoter region of the GM-CSF gene can be divided into two functional domains (see Fig.
4
). The first, containing repeated 5'-CATTA/T-3' elements (-65 to -31), has been shown to bind AP1, ets and
NF-AT transcription factors and to respond to activation signals in
fibroblasts, endothelial cells and Jurkat T cells (
3
,
11
,
13
,
14
,
16
-
19
). The second domain (-114 to -66) contains CK-1 and CK-2 sequence elements, which are conserved in many
cytokine genes, and is responsive to the HTLV-1 activator, tax and to CD28 co-stimulation in Jurkat T cells (
3
,
18
,
20
). This domain binds a number of nuclear proteins, including activators from the
NF-[kappa]B family (
3
,
18
,
20
,
21
). In contrast to the situation in T cells, we have determined that the -114 to -66 domain of the human GM-CSF promoter has repressor activity in fibroblasts (
11
). This region of DNA can partially repress the TNF-[alpha]-inducible response of the downstream -65 to -31 region. The ability of this domain to act as
a repressor element correlates with the binding of a nuclear factor called NF-GMb and mutation of the NF-GMb binding sites (within -114 to -79) results in relief of repression of the downstream -65 to -31 region. We also previously presented
evidence that NF-GMb binding could be involved in repressing the NF-[kappa]B sites across the -114 to -66 region (
11
). An NF-[kappa]B site overlapping the 3'-end of the NF-GMb binding sites (see Fig.
4
) was found however to be weakly responsive to TNF-[alpha] in fibroblasts (
11
).
We found that NF-GMb from fibroblast and HUT78 T cell nuclear extracts bound only to single-strand DNA and contacted two repeated 5'-CCTG-3' sequences on the non-coding strand of the GM-CSF -114 to -79 region (see
Fig.
2
a). Both repeats were required for full NF-GMb binding. We could not detect any binding of nuclear NF-GMb to double-strand DNA. In addition to NF-GMb, we detected a faster migrating nuclear complex, NF-GMc, which results from protein binding to one or
other of the repeated 5'-CCTG-3' sequences on the non-coding strand of the -114 to -79 region. UV cross-linking demonstrates that
the NF-GMb complex is composed of a 42.5 and a 22 kDa protein, while the NF-GMc complex is composed of only a 22 kDa protein (
11
). We previously interpreted this such that NF-GMb represented the binding of a dimer of 22 kDa, with NF-GMc representing the binding of a single 22 kDa monomer (
11
). It is also possible that NF-GMb represents the binding of two separate protein complexes, both a 42.5
kDa protein and a dimer of the 22 kDa protein. We determined that NF-GMb/c binding activity was constitutively present in nuclear extracts of T
cells and fibroblasts (
11
, unpublished data).
We now report the isolation of cDNA clones which encode proteins with NF-GMb-like binding activity. The cDNAs code for factors belonging to an
unusual group of proteins called cold shock domain (CSD) factors. Sequences for
two different CSD factors were cloned, one identical to dbpB and the other a
potentially new variant of dbpA. We show that the nuclear NF-GMb complex is composed of CSD factors and that overexpression of CSD
factors leads to repression of the GM-CSF promoter.
A total of 5 * 10
5
phage from a [lambda]gt11 cDNA expression library, made from HUT78 T cell RNA (Clontech, CA),
were screened as previously described (
22
) with an NF-GMb binding site probe. The binding conditions used were those described
for detecting nuclear NF-GMb complex formation in gel retardation assays (
11
) except that binding was performed at 4oC. The probe was a single-strand DNA oligonucleotide (2*GM-) containing a double copy of the -114 to -79 region of the non-coding (-) strand of the human GM-CSF promoter, which
contains the repeated NF-GMb binding sites. The sequence of the 2*GM- probe is: 5'-aattcAACTA
For phage dot blots, purified [lambda]gt11 phage were dotted onto bacterial lawns and probed with
32
P-labeled oligonucleotides exactly as for the initial library screening.
Eco
RI inserts from positive [lambda]gt11 cDNA clones were cloned into the pSP72 vector (Promega) and inserts
were then sequenced in both directions by the dideoxy sequencing procedure
using progressive oligonucleotide primers. Sequences were compared to the
GenBank database. Five cDNA clones encoded dbpB and two clones encoded a
variant of dbpA (dbpAv). The dbpB [lambda]gt11 and pSP72 clones are designated [lambda]B1, [lambda]B2, [lambda]B3, [lambda]B4 and [lambda]B5 and pB1, pB2, pB3, pB4 and
pB5 respectively. The dbpAv clones are named [lambda]A1 and [lambda]A2 and pA1 and pA2.
Eco
RI inserts from pB5 and pA2 containing full-length dbpB and dbpAv coding sequences were cloned in sense orientation
into expression vector pSG5 (Stratagene), with a modified polylinker, to
generate pSGdbpB and pSGdbpAv respectively. To generate the dbpB insert, pB5
was partially digested with
Eco
RI, as the dbpB coding region has an internal
Eco
RI site. pSGdbpBdel1 was created by digesting pSGdbpB with
Nar
I, which removes the CSD and the C-terminal domain of dbpB, followed by religation. pSGdbpBdel1 contains the
first 47 amino acids of dbpB. pGM-41 and pGM-43 have previously been described and contain respectively the -114 to -66 region and the -65 to -31 region of the human GM-CSF promoter in the pBLCAT2
reporter vector (
11
). pSV2CAT contains the SV40 early promoter on the chloramphenicol
acetyltransferase (CAT) reporter gene (
23
).
Human embryonic lung fibrobasts (HEL; Commonwealth Serum Laboratories,
Australia) were grown in DMEM and 10% fetal calf serum. These cells were used
from passage 14 to 20 in all experiments. HEL fibroblasts were co-transfected with 15 [mu]g of reporter constructs and 5 [mu]g of expression plasmids using DEAE-dextran as described (
11
). Twenty four hours following transfection, cells were stimulated with TNF-[alpha] (100 U/ml) or left untreated for an additional 24 h. Cells were
then harvested and CAT assays performed as described (
11
). Percentage [
14
C]chloramphenicol conversion to acetylated forms via CAT activity in extracts was determined using phosphorimager analysis (Molecular Dynamics).
All oligonucleotides were synthesized on an Applied Biosystems model 381A DNA
synthesizer. Full-length oligonucleotides were purified from non-denaturing polyacrylamide gels (
22
). Single-strand DNA probes for gel retardation asays or library screening were
prepared by end-labeling coding (+) or non-coding (-) strand oligonucleotides with [[gamma]-
32
P]ATP and T4 polynucleotide kinase followed by gel purification.
Nuclear extracts from HUT78 T cells were prepared as previously described (
24
). Gel retardation assays were performed using 0.25 ng of single-strand
32
P-labeled probe in a 10 [mu]l reaction mix of 0.5* TM (
25
,
11
) buffer containing 200 mM KCl, 0.4 [mu]g poly(dI[middot]dC) and 1.0 [mu]g of crude HUT78 T cell nuclear extract. Reactions were
incubated at room temperature for 20 min and analyzed on 12% non-denaturing polyacrylamide gels in 0.5* TBE (
25
). Competition with unlabeled single-strand oligonucleotides was performed by addition of nuclear extract and
unlabeled probe, followed by immediate addition of the labeled single-strand probe.
Previous studies demonstrated that HUT78 T cell nuclear extracts contained high
levels of protein capable of forming NF-GMb complexes (
11
). We therefore screened a [lambda]gt11 cDNA expression library, made from HUT78 T cell RNA, for clones
expressing proteins with NF-GMb binding activity. The library was screened with a single-strand oligonucleotide (2*GM-) containing a double copy of the non-coding (-) strand of the -114 to -79 region of the human GM-CSF promoter. The non-coding strand
of the -114 to -79 region contains the two repeated NF-GMb binding sites (5'-CCTG-3') (
11
). After three rounds of screening seven positive clones were isolated. Inserts
from the cDNA clones were subcloned into pSP72 plasmid vectors and sequenced.
All of the clones contained open reading frames encoding CSD proteins. CSD
proteins, also known as Y-box proteins, represent an unusual family of factors which have been shown
to bind single-strand DNA, double-strand DNA and RNA and to be involved in transcriptional activation
and repression, mRNA packaging and translational regulation (
26
-
28
). CSD proteins are divided into three domains, the central domain representing
the CSD domain, which is highly conserved across this family of proteins (
26
-
28
).
Five of these clones ([lambda]B1-[lambda]B5, Fig.
1
) contained DNA sequences identical to the cDNA sequence reported for the human
CSD protein called dbpB (
29
). These sequences are also similar, but not identical, to the first reported
sequence for human YB-1 (
30
). The differences in the YB-1 sequence are presumably due to sequence error and it is generally
assumed that human dbpB and YB-1 are identical (
31
,
32
). The [lambda]B5 clone contains the full dbpB coding region and encodes a protein of
324 amino acids.
To determine the binding specificity of the proteins expressed from dbpAv and
dbpB cDNA clones, the clones were screened with a panel of probes in phage dot
blots (Fig.
2
). The clones were first screened with wild-type (GM) or mutant (GMm19, m21, m23 or m25) coding (+) or non-coding (-) single-strand oligonucleotides covering the NF-GMb binding sites in the -114 to -79 region of the GM-CSF promoter (Fig.
2
a). As expected, all cDNA clones expressed protein that bound to the screening
probe (2*GM-; Fig.
2
a). No binding or only a relatively small amount of binding was observed to the
DNA strand complementary to the screening probe (2*GM+). As previously reported for binding of NF-GMb from fibroblast and HUT78 T cell nuclear extracts (
11
), protein expressed from all the cDNA clones bound to the wild-type GM (-114 to -79) non-coding (-) strand but did not bind to the coding (+)
strand. We observed that the binding of protein from dbpB clones was
consistently stronger than the binding of dbpAv protein to the GM- oligonucleotide. Consistent with NF-GMb binding activity, binding of protein from the dbpB and dbpAv
clones was reduced or abolished by mutation of one or other of the NF-GMb binding sites (GMm19 or m21) and completely abolished when both sites
were mutated (GMm23). Binding was not reduced by the irrelevant GMm25 mutation
and, as we have previously observed for NF-GMb binding (
11
, unpublished data), the GMm25 mutant resulted in an increase in binding. This
mutant increases the C content of the GM- sequence. As CSD proteins can have a preference for CT-rich sequences (
28
), this may explain the increased binding to the GMm25 mutant.
The proteins expressed from CSD cDNA clones were next screened with single-strand oligonucleotides previously reported to bind CSD proteins (Fig.
2
b). The oligonucleotides represented the coding (+) and non-coding (-) strands of sequences from the HPV 18 enhancer (HPV;
32
), the c-
myc
nuclease-sensitive element (NSE) (myc;
34
) and the EGF receptor NSE (EGFR;
34
). The HPV coding (+) sequence has been shown to bind human YB-1 (dbpB;
32
). The myc coding (+) sequence binds dbpB and a human CSD protein called NSEP-1 (
34
,
35
) and the EGFR coding (+) sequence binds NSEP-1 (
34
). The coding (+) CSD binding strands are all CT-rich in nature. Consistent with these observations, the dbpAv and dbpB
proteins expressed from cDNA clones bound specifically to the HPV+, myc+ and
EGFR+ single-strand oligonucleotides, demonstrating no observable binding to the
complementary non-coding (-) GA-rich strands. dbpB proteins bound all CSD oligonucleotides
with similar affinity, whereas dbpAv protein had the highest affinity for the
most highly CT-rich myc+ and EGFR+ oligonucleotides.
The expressed cDNA clones were also screened with the coding (+) and non-coding (-) strands of the Y-box motif (5'-CTGATTGGCCAA-3') from the HLA DR[alpha] promoter (DR[alpha];
30
). This sequence was initially reported to bind YB-1 (dbpB) (
30
). There have been subsequent conflicting reports as to the ability of YB-1 (dbpB) to bind to Y-box sequences in genomic genes of higher organisms in either double-strand or single-strand form (
33
-
38
). In addition, NSEP-1 does not bind to a double-strand or single-strand Y-box sequence (
34
) and dbpA was found not to bind to a double-strand Y-box motif of the MHC gene I-A[beta] (
37
). Consistent with these later observations, we find that dbpAv and dbpB expressed protein cannot bind to single-strand HLA DR[alpha] Y-box (DR[alpha]) coding (+) or non-coding (-) sequences (Fig.
2
b).
The above results demonstrate that protein expressed from the dbpB and dbpAv
cDNA clones have the appropriate binding characteristics previously described
for CSD proteins and that they can bind to the GM-CSF promoter with the same specificity as that observed for nuclear NF-GMb.
Given the above results, nuclear NF-GMb from HUT78 T cells was analyzed for its binding to the single-strand CT-rich CSD oligonucleotides in gel retardation assays. These
assays will also detect any binding of the NF-GMc complex, which appears to be a component of the NF-GMb complex as determined from its binding characteristics and
apparent protein content (
11
), as discussed above. In Figure
3
a it can be seen that NF-GMb/c-like complexes bind to the HPV+ and EGFR+ sequences and an NF-GMc-like complex binds to the myc+ sequence. No binding was
observed to the HLA DR[alpha] Y-box sequences (DR[alpha]), even on long exposure. Consistently these apparent NF-GMb/c complexes on the
32
P-labeled HPV+, myc+ and EGFR+ CSD binding site sequences were competed for
by the unlabeled GM- oligonucleotide, whereas the extent of competition by GMm23- (NF-GMb binding site mutant) was considerably less (Fig.
3
b). Competition was approximately five times greater with GM- than with GMm23-. Conversely NF-GMb/c complexes on the GM- oligonucleotide were competed for with the CSD
binding site oligonucleotide HPV+ but not with a non-specific oligonucleotide (Fig.
3
c). An additional complex marked x (Fig.
3
a-c) was also seen in gel retardations. The nature of this complex is
unknown, but we have demonstrated that it does not contact the repeated 5'-CCTG-3' NF-GMb binding sites of the GM-CSF promoter (unpublished data). These
data demonstrate that nuclear NF-GMb/c shows the same binding specificity for CSD oligonucleotides as the
expressed CSD proteins. Consistently we also observed that NF-GMb/c complex formation on GM- or HPV+ was reduced by inclusion of a polyclonal antibody to a
Xenopus
CSD protein (a gift from A.Wolffe; data not shown). These data, taken together,
indicate that CSD proteins are components of the nuclear NF-GMb/c complex.
Since we have shown in a previous report (
11
) that the NF-GMb/CSD binding sites were involved in repression of the -65 to -31 region of the GM-CSF promoter, we wished to determine the effect of
overexpression of the CSD proteins on GM-CSF promoter function. To do this the pGM-41 (-65 to -31) and pGM-43 (-114 to -31) reporter constructs were co-transfected with the dbpAv
(pSGdbpAv) and dbpB (pSGdbpB) expression clones into HEL fibroblasts and
treated with or without TNF-[alpha] (Fig.
4
). Consistent with the suggested function of NF-GMb/CSD, overexpression of both full-length dbpAv and dbpB repressed the 3-fold TNF-[alpha]-inducible activity from the pGM-43 construct, containing NF-GMb binding sites, by 67 and
60% respectively. Surprisingly, overexpression also repressed expression from
the pGM-41 reporter construct. We are not aware of any NF-GMb/CSD binding sites in the -65 to -31 region (unpublished data). The TNF-[alpha]-inducible expression was reduced by 76 and 70% respectively by
dbpAv and dbpB and the uninduced expression was reduced by 50% with both
proteins. These results suggest that NF-GMb/CSD factors can also function in the absence of apparent binding
sites.
The repressive effects discussed above were not however observed when either (i) pGM-43 and pGM-41 were co-transfected with pSGdbpBdel1, containing only the N-terminal sequences of dbpB, or (ii) when full-length dbpAv and dbpB constructs were co-transfected with a heterologous SV40 early
promoter construct, pSV2CAT. As shown in Figure
4
, pSGdbpBdel1 had little effect on pGM-43 expression and activated both pGM-41-induced (33%) and uninduced (50%) expression. dbpAv had little
effect on induced SV40 promoter expression and a slight repressive effect (23%)
was observed with dbpB (Fig.
4
). As there are no reported studies on the function of CSD protein N-terminal domains, the reason for the activation effect of pSGdbpBdel1 is
at present unclear. Given these results, however, it is clear that the high
levels of repression of the GM-CSF promoter by full-length dbpAv and dbpB represent a specific repression of the GM-CSF promoter and are not just due to toxic affects of protein
overexpression. It has also been reported that
Xenopus
CSD proteins can affect translation (
27
,
28
). The minor effects of dbpAv and dbpB overexpression on pSV2CAT expression
rules out a major general effect on CAT reporter protein levels due to changes
in translation.
We have cloned human cDNAs encoding proteins that bind to a single-strand DNA NF-GMb binding site probe derived from a novel repressor element in the
human GM-CSF promoter (
11
). The cDNAs code for the CSD proteins dbpAv and dbpB. The dbpB sequence has
previously been reported (
29
), while the dbpAv sequence is a new variant of the reported human dbpA sequence
(
29
).
CSD proteins are characterized by a central domain of ~100 amino acids (
26
-
28
) which is highly conserved throughout evolution and is found to be 43%
identical to bacterial cold shock proteins (
26
). Eukaryotic CSD proteins include dbpB (YB-1 and EF1a) (
29
-
33
,
35
,
36
,
39
-
43
), dbpA (
29
,
31
,
35
,
44
), NSEP-1 (
34
), FRGY1 and FRGY2 (
45
). In addition to binding single-strand DNA, CSD proteins can also bind double-strand DNA and RNA; the CSD domain is necessary for all these
binding activities (
31
-
35
,
46
-
48
). By virtue of these varied nucleotide binding activities, CSD proteins have
been shown to be involved in transcriptional activation and repression, mRNA
packaging and translational regulation (
27
,
28
,
36
,
37
,
39
). The changes in protein sequence between dbpA and dbpAv are outside the CSD
domain, resulting in a single amino acid substitution in the N-terminal region and a C-terminal extension on dbpAv. The functions of CSD protein N-terminal regions and the most C-terminal portions of these proteins have yet to be
determined.
As for nuclear NF-GMb, dbpAv and dbpB expressed from cDNA clones required the repeated 5'-CCTG-3' sequences on the non-coding (-) strand of the -114 to -79 GM-CSF promoter
region for binding. Conversely, NF-GMb and the related complex NF-GMc bind to the same set of defined single-strand DNA CSD binding sites as the expressed CSD proteins.
Competition and antibody assays indicated that CSD proteins are components of
the nuclear NF-GMb/c complex. Accordingly, the identification of a 42.5 kDa protein in
the NF-GMb complex, by UV-crosslinking (
11
), is consistent with the expected size for dbpAv (41 kDa) as deduced from the
DNA sequence and with the observed size for CSD proteins detected by antibodies
in HeLa and Raji cells (
32
,
38
).
Xenopus
and bacterial CSD proteins (
26
,
28
,
49
) can bind a motif called a Y-box sequence (5'-CTGATTGGCCAA-3'). Mammalian/avian CSD proteins can also bind Y-box sequences in certain viral promoters (
36
,
39
,
41
). Mammalian/avian CSD proteins generally, however, have a preference for CT-rich sequences and their binding to Y-box sequences in genomic genes is controversial (
28
,
34
,
35
). Consistently we observed NF-GMb/CSD binding to three CT-rich single-strand DNA CSD binding sites, but not to a Y-box element. A common feature of the three CT-rich sequences investigated is the presence of
repeated CT pairs (
32
,
34
). Six CT pairs can also be found in the GM-CSF promoter oligonucleotide (GM-), two of these in the repeated 5'-CCTG-3' sequences involved in NF-GMb/CSD binding. We have also
identified a similar repeated sequence in the G-CSF promoter that acts as a CSD binding site (unpublished data) and have
reported potential NF-GMb (CSD) binding sites in the repressor elements of a number of other
immune function genes (
11
).
We previously demonstrated that the -65 to -31 region of the GM-CSF promoter is TNF[alpha] responsive and that addition of the -114 to -66 region resulted in partial
repression of this response (
11
). Mutation of the NF-GMb/CSD binding sites resulted in relief of this repression, indicating
that the binding of NF-GMb/CSD factors was required for repressor element function. The binding
of NF-GMb to the repressor element also appeared to prevent transcription factor
action across the -114 to -66 region. The potential repressive action of NF-GMb factors is now confirmed here by our observation that
overexpression of the CSD proteins dbpAv and dbpB results in near complete
repression of the remaining TNF-[alpha]-inducible activity of the -114 to -31 construct (pGM-43). Repression of transcription by CSD
proteins has also been reported for the HLA-DR[alpha] and I-A[beta] genes and CSD binding sites have been located in a
repressor element of the [gamma]-globin genes (
35
,
37
,
50
). As NF-GMb (CSD) binding to the GM-CSF promoter is single-strand specific (
11
), we previously proposed that binding to single-strand DNA will affect local DNA structure, preventing the binding of double-strand DNA-specific activating factors across the repressor element and
the downstream -65 to -31 region. A similar mechanism of action has recently been
suggested for the action of CSD proteins on the MHC class II DR[alpha] promoter and [gamma]-globin genes (
35
,
38
). Consistently, single-strand regions have been detected in CSD binding sites
in vitro
, but have not yet been examined
in vivo
(
34
,
35
,
38
).
Transcriptional regulation by CSD factors may also be brought about via protein-protein interactions, as dbpA has recently been shown to form a complex
with a subunit of the positive NF-Y transcription factor resulting in transcriptional repression of MHC
class II genes (
37
). This complex formation occurs in the absence of a CSD binding site.
Similarly, we observed here that overexpression of dbpAv and dbpB could
directly repress the -65 to -31 region in the absence of the upstream NF-GMb/CSD binding site repressor element. We have not detected
any NF-GMb/CSD binding sites in the -65 to -31 region (unpublished data). Mutation across the -65 to -31 region alone does not reveal any repressor
elements, indicating that direct repression of this region does not involve a
DNA sequence-protein interaction (unpublished data).
The relative contribution of repression of the GM-CSF promoter by the upstream repressor element (-114 to -66) and the direct action of CSD factors on the -65 to -31 region is difficult to estimate. It is of
interest however that the addition of the upstream repressor element results in
greater repression of the -65 to -31 region than overexpression of dbpAv and dbpB. The amount of
repression by the -114 to -66 region is even greater when the most 3' NF-[kappa]B site (Fig.
4
), which is a weak activation site within this region, is mutated (
11
). It is possible that the binding of CSD factors to the -114 to -66 repressor element is the primary means of repression and that
this is backed up by a second mechanism of repression acting directly on the -65 to -31 region. Such a dual mechanism of repression by CSD factors
should ensure that there is no inappropriate expression of the endogenous GM-CSF promoter.
We thank Dr A.Wolffe for the gift of the
Xenopus
CSD antibody. We also thank Anna Sapa for excellent technical assistance and Dr
Peter Cockerill for database analysis. This project was funded by a National
Health and Medical Research Council Project Grant to MFS.
REFERENCES
Return