(Received for publication, July 30, 1996, and in revised form, November 20, 1996)
From the Danish Cancer Society, Department of Molecular Cancer Biology, Strandboulevarden 49, DK-2100 Copenhagen, Denmark and the ¶ Institute for Gene Biology, Russian Academy of Sciences, Vavilov Str. 34, Moscow, Russia
The transcription of the mts1 gene
correlates with the metastatic potential of mouse adenocarcinomas. Here
we describe strong enhancer whose location coincides with the DNase I
hypersensitivity area in the first intron of the mts1 gene.
The investigation of the transcriptional activity of a series of
plasmids bearing deletions in the first intron sequences revealed that
the observed enhancer has a composite structure. The enhancer activity
is partially formed by the B-related element: GGGGTTTTTCCAC. This
sequence element was able to form several sequence-specific complexes
with nuclear proteins extracted from both Mts1-expressing CSML100 and Mts1-non-expressing CSML0 adenocarcinoma cells. Two of these complexes were identified as NF-
B/Rel-specific p50·p50 homo- and p50·p65 heterodimers. The third complex was formed by the 200-kDa protein. Even
though the synthetic
B-responsible promoter was active in mouse
adenocarcinoma cells, a mutation preventing NF-
B binding had no
effect on the mts1 natural enhancer activity. On the
contrary, the mutation in the
B-related element, which abolished the
binding of the 200-kDa protein, led to the functional inactivation of this site in the mts1 first intron. The mts1
B-like element activated transcription from its own mts1
gene promoter, as well as from the heterologous promoter in both CSML0
and CSML100 cells. However, in vivo occupancy of this site
was observed only in Mts1-expressing CSML100 cells, suggesting the
involvement of the described element in positive control of
mts1 transcription.
The mts1 gene encodes a Ca2+-binding protein belonging to the S100 subfamily (1-4). Even though several molecular targets of Mts1 protein, including the heavy chain of non-muscle myosin (5), tropomyosin (6), and some unidentified proteins (7), have been characterized, the precise biological function of the Mts1 protein remains unknown. The possible involvement of this protein in cell motility and signal transduction is under discussion (5, 8, 9). mts1-specific mRNA was detected in some types of normal mouse cells such as lymphocytes, macrophages, neutrophils (10), and immortalized fibroblasts (2, 3). mts1 transcription was shown to be inducible in several cell systems (3, 4, 10).1
It was shown in direct transfection experiments that constitutive overexpression of the Mts1 protein in a benign rat epithelial cell line (11) and in src-transformed rat fibroblasts (9) influenced in vitro cell invasion and promoted tumor progression. In addition, the expression of the mts1 gene correlated with the metastatic phenotype of mouse and human carcinoma cells (12, 13). Therefore, the study of the regulation of mts1 transcription can give knowledge leading to a better understanding of the changes that occur in transcription controlling mechanisms in carcinoma cells during tumor progression.
Transcriptional regulation of the mts1 gene has been studied
in the mouse metastatic adenocarcinoma cell line CSML100, where this
gene is highly expressed. It was demonstrated that the 5-flanking area
of the mts1 gene up to
1890
bp2 did not contain any sequences that
positively regulate mts1 gene expression in these cells
(14). On the other hand, a positive regulatory element was identified
in the first intron of the mts1 gene between positions +294
and +309 (15).
The constructs that contained the CAT gene, under the control of the mts1 regulatory sequences, revealed some level of transcriptional activity when they were transiently transfected into mouse adenocarcinoma cells CSML0, even though the endogenous mts1 transcription was not detectable in those cells (14). Furthermore, mts1 gene transcription can be induced by azadeoxycytidine in two human lymphoma cell lines (16). Thus it was speculated that DNA methylation is involved in transcriptional control of the mts1 gene (14, 16).
Strutz et al. (17) have described a cell-specific enhancer
that was present in the 1800-bp upstream region of the mts1
gene. The 1800 to +65 mts1 upstream region placed in front
of luciferase reporter gene revealed transcriptional activity in 3T3
and DBF fibroblasts. This activity composed 14-21% of the strength of SV40 early enhancer. On the other hand, the mts1 upstream
region was not active in mouse epithelial MCT, embryonic PYS-2, and
mesangial MHC cells.
In our research we have continued to study the transcriptional control
of the mts1 gene. We show that the enhancer activity is
associated with the mts1 first intron sequences in mouse
adenocarcinoma cells. We report that the previously described
regulatory element located between bp +294 and +309 does not exhaust
the positive regulatory system which controls mts1 gene
expression in CSML100 cells. The deletion of the
SacI-PstI fragment (positions +258 and +646) of
the mts1 first intron sequence decreased but did not abolish
the transcriptional activity of reporter-containing plasmids
transiently transfected in CSML100 cells. Another enhancer was found to
be located in the DNase I hypersensitive area in the first intron of
the mts1 gene between positions +750 and +950 bp. Two
regions of DNA-protein interactions were identified in this area. One
of these regions represents a B-related site that binds p50, p65,
and another protein with molecular mass of 200 kDa in vitro.
The occupancy of this site in vivo was observed only in
Mts1-expressing cells. The mutagenesis analysis of the
B-related
site allowed us to conclude that the interaction with 200-kDa protein
was functional in the nucleotide context of the mts1
enhancer. The second type of proteins that are likely to be involved in
formation of DHSs in both CSML0 and CSML100 cells apparently represent
Msbps (minisatellite binding proteins) (18, 19). These proteins protect
intron DNA between +811 and +841 bp and are not involved in the control
of mts1 transcription.
Nuclei were isolated as described in Ref. 16 and incubated at 37 °C for 5 min in 200-µl aliquots with different concentrations of DNase I (0.1, 0.3, 1, 3, and 6 units per aliquot). DNA was purified, digested with EcoRI or HindIII, and analyzed by Southern blot hybridization.
The Hybridization ProbesTo map the DHSs we used two probes. The ExIII probe corresponded to the sequence 1-145 bp from the beginning of the mts1 third exon; the mi-probe corresponded to the +294 to +442 bp from the mts1 gene transcription start. The highly radioactive probes were synthesized in 11 PCR cycles. P-41CATInt plasmid was used as a template.
PlasmidsP-41CATInt and p-41CATInt plasmids have been
described previously under the names p-41CAT and p-41CAT
(15). The
structure of the p-41CATInt plasmid contained the
41 to +1254
mts1 sequences and was as follows: 41 bp of mts1
gene 5
-flanking area, first exon, first intron sequences, 14 bp of the
second exon linked to the bacterial CAT gene. p-41CATInt
corresponded to the p-41CATInt with the exception of the deleted
SacI-PstI restriction fragment. p-41LUCInt,
p-1890LUC, and p-41LUC were made by cloning of the
41 to +1254,
1890 to +65, and
41 to +65 sequences of the mts1 gene,
upstream of the Photinius pyralis luciferase gene obtained from pfLUC plasmid (20) kindly provided by Dr. M. Jäättelä. Promoterless pLUC plasmid was made by the
excision of the c-fos minimal promoter from pfLUC.
CAT-containing deletion mutants were obtained as described
below. Pdel1 and pdel4 constructs were as follows: p-41CATInt DNA was
digested with PstI and, after Bal31 nuclease
treatment, self-ligated. Plasmids pdel5 to pdel8 were constructed as
follows: after PstI and Bal31 nuclease treatment, p-41CATInt DNA was digested with EcoRI. The obtained
fragments were subcloned into PstI/EcoRI-treated
p-41CATInt plasmid. Pdel9 plasmid DNA was synthesized on p-41CATInt
template DNA in 12 cycles PCR according to Ref. 21. As primers we used
oligonucleotides AGCAGCCGCGCCCAACGCTGGGAG (coding strand; positions
+831, +854) and GTGGAAAAACCCCAGCTGCCTAATAGGAG (noncoding strand;
positions +800, +772). This amplification generated deletion of 30 bp
between +800 and +831. To generate the plasmids pmut1LUC and pmut2LUC, we introduced the mutations into p-41LUCInt plasmid DNA by amplifying the whole plasmid using non-mutated 5
-end and mutated 3
-end primers:
5
-end, AGTAAGATGAAGTGGCAGAGG; 3
-end, CCTATTAGGCTGCCTGGTTTTTTCCAC (pmut1LUC) or CCTATTAGGCAGCTGCTATTTTTACAC (pmut2LUC). The
borders of all deletions as well as point mutations were verified by
DNA sequence analysis. PGL2 control plasmid that contained the
luciferase gene driven by SV40 promoter/enhancer elements was purchased
from Promega Inc. To create pM5+ and pM5
plasmids, we cloned five copies of the Sb sequence, GCAGCTGGGGTTTTTCCACTT, in direct (pM5+) and
reverse (pM5
) orientations, downstream of the CAT gene in the pCAT plasmid (Promega), where the CAT gene was driven by
minimal SV40 promoter. The plasmid pN5 contained five copies of
B
oligonucleotide, AGTTGAGGGGACTTTCCCAGGC, from the mouse Ig
enhancer
(22), downstream of the CAT gene in the pCAT plasmid.
Mouse non-metastatic
adenocarcinoma CSML0 and mouse metastatic adenocarcinoma CSML100 cell
lines (12) were maintained in Dulbecco's modified Eagle's medium
supplemented with 10% fetal calf serum and 50 µg/ml each of
streptomycin and penicillin. 2 × 106 CSML0 or CSML100
cells in 100 µl of phosphate-buffered saline were transfected by
electroporation with a single pulse of 250 V, 250 microfarads using the
Bio-Rad electroporation system "Gene Pulser" and seeded on 10-cm
dishes. The efficiency of each transfection was monitored by a
cotransfected -galactosidase expression vector pSV
-gal. Two days
post-transfection the cells were lysed, and luciferase activity was
measured by a luminometer (Promega luciferase assay system). The
obtained lysates were also tested for their
-galactosidase activity
by using o-nitrophenyl
-D-galactopyranoside (Sigma) as a chromogenic substrate. In some
experiments the transfected cells were subjected to RNA isolation.
RNA from transfected cells was isolated as
described (23), treated with DNase I, extracted by phenol/chloroform,
and analyzed in RNase protection assay or by primer extension
technique. Commercial plasmids pTRIB and pTRIC (Ambion) were used for
riboprobes preparation. The intensity of CAT- and
-gal-specific signals was determined using an UltraScan apparatus
(LKB). For primer extension analysis CAT-specific primer
CAACGGTGGTATATCCAGTG was annealed to the RNA and extended using avian
myeloblastosis virus reverse transcriptase (Promega) according to the
manufacturer's recommendations.
Nuclear extract preparation, EMSA, in vitro DNase I
footprinting analysis, and methylation interference assay were
performed as described (24, 25). The +691 to +891 fragment of the
mts1 first intron sequence studied in EMSA was amplified in
15 cycles PCR using one end-labeled and one non-labeled primer:
GAACATATAGCACCTAGG and TCAACCACAAGCACAGG. The sequences of end-labeled
double-stranded Sb-containing and B consensus-containing
oligonucleotides (22) analyzed in EMSA are given under "Plasmids."
Oct-1 was detected with the double-stranded oligonucleotide
TGTCGAATGCAAATCACTAGAA (26). All anti-NF-
B/Rel antibodies were
purchased from Santa Cruz Biotechnology.
CSML100 nuclear proteins were UV cross-linked in situ to the Sb containing oligonucleotide, where all Ts were substituted on 5-bromodeoxyuridine, and analyzed as described (27).
In Vivo Genomic FootprintingThe dimethyl sulfate treatment of cell cultures, DNA preparation, and ligation-mediated PCR were performed as described (28). The chain-specific primers were as follows: primer 1, CTGAAACCTGCAAAG; primer 2, CTACCCCTCAAACTCTGGAGATC; primer 3, CTGGAGATCCATGCCAGCACCTTTC.
Previously, we have searched in the mts1
5-flanking area for cis-elements controlling mts1
transcription in CSML100 mouse adenocarcinoma cells. No such elements
were found (14). All plasmids studied contained the mts1
first intron sequence and differed by the length of upstream regions.
However, the enhancer activity of the first intron of the
mts1 gene could mask the function of some regulatory
elements located upstream of the gene. Therefore, at the initial step
of the search for cis-elements regulating mts1
transcription, we compared the abilities of the mts1 first intron and 5
-flanking sequences to enhance transcription in mouse adenocarcinoma CSML0 and CSML100 cells. The cells were transiently transfected by the constructs whose structure is presented in Fig.
1. p-1890LUC and p-41LUC contained 1890 and 41 bp
upstream from the mts1 first exon, the whole sequence of the
first exon (38 bp), and 27 nucleotides of the first intron placed in
front of luciferase reporter gene. p-41LUCInt construct was generated from the previously described p-41CAT and contained the whole mts1 first intron. The activity of these plasmids was
studied in CSML100 and CSML0 adenocarcinoma cells compared with the
activity of SV40 promoter/enhancer (pGL2 control construct), taken as
100%. The results of these experiments showed that there were no
sequences that significantly influenced mts1 promoter
activity between
1890 and
41 bp, neither in CSML100 nor in CSML0
cells (compare p-41LUC and p-1890LUC). The comparison of p-41LUC and
p-1890LUC plasmids with the promoterless pLUC revealed that these
plasmids were weakly expressed in CSML100 and inert when transfected in
CSML0 cells (Fig. 1, A and B). On the other hand,
the presence of the mts1 first intron sequence considerably
increased the activity of the reporter plasmid in CSML100 (Fig.
1A) and to the lesser extent in CSML0 cells (Fig.
1B).
1226 bp of the mts1 first intron contain several potential
TATA boxes, which can work as artificial promoters in transient transfection assay and thereby determine the enhanced level of activity
of p-41LUCInt. To exclude this possibility, as well as to check if the
first intron enhancer activity can be exclusively addressed to the
previously described +294 to +309 positive element (15), we transfected
CSML100 cells by p-41CATInt, p-41CATInt, and pdel4 plasmids (Fig.
1C). RNA was isolated and normalized to the transfection
standard, and the 5
-ends of CAT-containing RNA were mapped
by the extension of the CAT-specific primer. The main bands
obtained in lanes 1 and 2 corresponded to the
138-nucleotide cDNA and coincided with the predicted size of RNA
synthesized from the mts1 gene natural promoter. The
deletion of the previously described +294 to +309 element led to the
40% reduction of the mts1 natural promoter activity (Fig.
1C, lane 2). The deletion of the +43 to +1134 sequence in
pdel4 plasmid abolished the transcription directed by the
mts1 promoter (Fig. 1C, lane 3). The obtained data allowed us to make two conclusions. First, several potential TATA
boxes located in the mts1 first intron were inactive in
transient transfection experiments in CSML100 cells. Therefore, the
constructed plasmids created an adequate model to study transcription
from the natural mts1 promoter. Second, in addition to the
described +294 to +309 functional sequence, a novel positive element(s) was present in the first intron of the mts1 gene.
The presence of DHSs correlates with the binding of transcription factors to nearby DNA sequences in many tissue-specific genes (for review see Ref. 29). Therefore, we have searched for DHSs in and around the mts1 gene as potential indicators for regions involved in the transcriptional control.
Nuclei were isolated from CSML0 and CSML100 cells and treated with
DNase I in different concentrations. After DNA purification and
digestion with EcoRI, the samples were blot-hybridized to the ExIII probe, which corresponded to the third exon sequence of the
mts1 gene, close to the EcoRI site (Fig.
2A). The 2.1-kb fragment was detected in DNA
isolated from DNase I-treated CSML100 nuclei. The appearance of this
fragment identified a DHS 2.1 kb upstream from the EcoRI
site, hence in the vicinity of the transcription start. Another DNase
I-hypersensitive area, located 1.2-1.4 kb upstream from the
EcoRI site of the third exon, was revealed in DNA isolated
from both CSML0 and CSML100 nuclei. This area, which contained at least
two DHSs located next to each other, corresponded to the first intron
sequences (Fig. 2A).
In order to map the first intron DHSs more precisely, DNA from DNase I-treated CSML0 and CSML100 nuclei was digested with HindIII and hybridized to the mi probe. This hybridization revealed a cell-specific location of DHSs in the first intron of the mts1 gene mapped 100-300 bp downstream from PstI site between + 750 and +950 (Fig. 2B). The different positions of DHSs in the chromatin of mts1-expressing and mts1-non-expressing cells might indicate both quantitative and qualitative differences in the composition of proteins (possibly including transcription factors) involved in their formation.
Functional Analysis of the mts1 Gene First Intron SequencesTo determine whether a transcription enhancer element
was present within the mts1 first intron DNase
I-hypersensitive area, two series of deletion-bearing constructs were
obtained and examined by transient transfection of CSML0 and CSML100
cells (Fig. 3). To obtain more reliable statistical
data, we have used two different genes as reporters in deletion
constructs: CAT (analyzed by RNase protection) and
luciferase (analyzed by detection of enzymatic activity in cell
extracts). The deletion of the sequence between positions +470 and +780
(plasmid pdel1) did not influence CAT transcription in
CSML100 or CSML0 cells (Fig. 3). On the other hand, the +646 to +789
deletion in the composition of the pdel5 construct caused a 2.5-fold
reduction of transcriptional activity in CSML100 cells (Fig. 3). Beside
that, the +646 to +789 deletion apparently affected activity of
CAT-containing constructs transfected in CSML0 cells (Fig.
3A). However, in these cells where the activity of the
p-41Int plasmid was relatively low, the sensitivity of the RNase
protection method was insufficient to measure the influence of the +646
to +789 deletion (Fig. 3A). The application of luciferase as
a reporter gene allowed us to conclude that this deletion caused approximately equal effects on activities of plasmids transfected in
both CSML0 and CSML100 cells (Fig. 3B).
The expansion of the deleted region to +850 at the 3-end (construct
pdel6) did not change the level of the reporter's expression. However,
in CSML100 cells the next deletion mutant, pdel7, with the 3
-end
border of the deletion mapped at the position +880, revealed further
reduction of the enhancer activity from 35 to 16% compared with the
p-41CATInt and p-41LUCInt plasmids (Fig. 3). Pdel7 had approximately
the same level of transcriptional activity as the pdel8 construct
(3
-end border at +1031). Therefore, two elements located in the DNase
I-hypersensitive area in the mts1 first intron composed
enhancer activity in CSML100 cells. One element is mapped in the
vicinity of +789, and the other is located between +850 and +880 (Fig.
3, compare the transcriptional activities of pdel1 with pdel5 and pdel6
with pdel7). The +850 to +880 region seemed to be not active in
mts1-non-expressing CSML0 cells (Fig. 3B).
Functional analysis of the mts1 first intron sequences
showed that in CSML100 cells the enhancer located in the vicinity of
DHSs has stronger activity than the previously described +294 to +309
positive regulatory element (15) (Fig. 3B, compare the
activities of pdel8 with p-41Int
).
To further characterize the mts1 first intron enhancer we searched for proteins interacting with the mts1 first intron sequence in the area of DHSs.
In Vitro Study of DNA-Protein Interactions within the +691 to +891-bp RegionGel retardation analysis of the +691 to +891-bp
DNA fragment revealed two types of sequence-specific complexes in both
CSML100 (Fig. 4A) and CSML0 (data not shown)
nuclear extracts: rapidly migrating group of complexes, Cr, and the
slowly migrating Cs. Using the DNase I footprinting technique, as well
as methylation interference analysis, we found that Cr complexes were
formed by the extensive sequence +811 to +841 bp (designated as Sa)
which contains the minisatellite "core" element, CTGGGCAGGCAG (data not shown). The minisatellite core sequence was shown to be able to
interact with the ubiquitously expressed minisatellite-binding proteins, Msbps (18, 19), which form rapidly migrating complexes that
appear to be similar to Cr complexes. Cr complexes were not functional
in transient transfection; the deletion of the +801 to +830 as well as
the +789 to +850 intron sequences did not influence mts1
enhancer activity neither in CSML100 nor in CSML0 cells (compare p-41CATInt and p-41LUCInt to pdel9, and pdel5 to pdel6, Fig. 3, A and B). On the other hand, Msbps strongly interacting
with the +811 to +841-bp sequence in vitro could be involved
in the formation of DHSs in vivo in both CSML0 and CSML100
cells.
The methylation interference analysis of Cs revealed two guanines with
the coordinates +789 and +790 (sense strand) and three guanines with
the coordinates +797, +798, and +800 (antisense strand) which were
directly involved in the formation of this complex (Fig.
4B). The positions of these guanines directly coincided with
the location of B-related sequence. Moreover, +788 and +789 guanines
were deleted in pdel5 construct which revealed the decreased expression
in both CSML100 and CSML0 cells, but they remained intact in the
structure of pdel1 plasmid with a wild-type level of enhancer activity
(Fig. 3).
The B-related sequence was designated as Sb. A more detailed study
of proteins interacting with the Sb sequence was performed.
The end-labeled 21-mer oligonucleotide that contained Sb
sequence was mixed with 15 µg of CSML0 (Fig. 5,
lanes 1-3) or 15 µg of CSML100 nuclear extracts
(lanes 6-8) and analyzed in EMSA. We observed three
sequence-specific complexes in both extracts used. In addition, another
complex designated as C2 was observed in some CSML0 and CSML100 nuclear
extracts. This complex was poorly reproducible, and its appearance
could result from C1 degradation. Slowly migrating C1 seemed to be
identical with Cs formed by the Sb sequence in the composition of the
+691 to +891 fragment (Fig. 4); two more rapidly migrating complexes
were NF-B-specific; their formation was abolished when 100-fold
excess of the nonlabeled
B consensus oligonucleotide was added (Fig.
5A, lanes 3 and 8). To determine the composition
of these complexes, we used specific antibodies to all known members of
the NF-
B/Rel family. Gel supershift analysis revealed that rapidly
migrating complexes represent p50·p50 homo- and p50·p65
heterodimers in both CSML0 and CSML100 cells (data not shown). C1 and
C2 complexes were not influenced by any of the anti-NF-
B antibodies.
In order to determine the number and molecular weights of the proteins
comprising the C1 complex, we UV irradiated the C1 complex in
situ (27), followed by SDS-PAGE. The C1 complex was found to be
formed by the single protein, whose molecular mass was determined as
200 kDa (Fig. 5B).
The qualitative composition of NF-B proteins interacting with Sb in
CSML0 cells was identical with that observed in CSML100 nuclear
extracts. But there were significant quantitative differences in the
abundance of p50·p50, p50·p65, and C1 complexes between CSML0 and
CSML100 cells (Fig. 5A). When normalized to the expression level of the Oct-1 binding factor, the differences varied from 10.8 times for C1 to 2.5 times for p50·p65 complex. The ubiquitous Oct-1
binding factor was previously shown to be equally expressed in normal
and in different types of malignant mammary epithelium (30).
Five copies of the 21-mer that contained Sb sequence
were subcloned in both orientations downstream of the CAT
gene in the composition of the pCAT plasmid, which contained minimal
SV40 promoter (pM5+ and pM5 constructs). The oligomerized Sb sequence could activate transcription driven by this promoter 11-13-fold in
CSML100 cells and 8-10-fold in CSML0 cells (Fig. 6).
The activation occurred in an orientation-independent manner. The
enhancer activity of the oligomerized five copies of the Ig
B (pN5
construct) was even greater in both cell lines, about 28-fold
enhancement in SV40 promoter activity in CSML100 and 15-fold activation
in CSML0 cells (Fig. 6A, lane 2; B, lane
2).
The Binding of the 200-kDa Protein Is Sufficient for Maximal Functional Activity of the Sb Sequence in the Nucleotide Context of the mts1 Enhancer
NF-B was shown to be functionally active in both
CSML100 and CSML0 mouse adenocarcinoma cells (Fig. 6A, lane
2; B, lane 2). Sb interacts in vivo with
NF-
B as well as with the 200-kDa protein (Fig. 5A) and
activates transcription from the heterologous promoter (Fig. 6A,
lanes 3-4; B, lanes 3-4). In order to
examine the functional activity of Sb in the nucleotide context of the
mts1 enhancer and to address this activity to the specific
protein, we designed two mutated oligonucleotides. mut1
bound 200 kDa but failed to interact with p50·p50 and p50·p65 (Fig.
7A, lanes 3-4). The mutations in
mut2 abolished the binding of all proteins (Fig. 7A,
lanes 5-6). The replacement of wild-type Sb by the
mut1 sequence in the p-41LUCInt construct revealed no
changes in enhancer activity of resulting plasmid pmut1LUC. On the
contrary, the second type of mutation (pmut2LUC plasmid) decreased the
enhancer activity to 32-37% of the wild-type enhancer activity in
both CSML100 and CSML0 cells (Fig. 7, B and C).
Therefore, the formation of the C1 complex mediates the function of Sb
in the context of the mts1 first intron sequences.
In Vivo DNA-Protein Interaction at the Sb Site in CSML0 and CSML100 Cells
In transient transfection experiments, 200-kDa protein
mediates activation of reporter constructs via the interaction with the
Sb sequence in both CSML100 and CSML0 cell lines, even though this
factor is 5.6-10.8 times more abundant in CSML100 compared with CSML0
cells. However, using in vivo genomic footprinting, we found
the Sb sequence to be occupied only in CSML100 cells where the
mts1 gene is highly expressed (Fig. 8). Three
guanine residues with coordinates +797, +798, and +800 (antisense
strand) were contacted by proteins in CSML100 cells. Noteworthily, the same Gs were involved in the formation of the Cs complex in
vitro (compare Fig. 4C and 8). Since Cs had the same
motility in EMSA as C1 complex (data not shown), it was likely formed
by the 200-kDa protein. Therefore, Sb sequence seems to be occupied by
200-kDa factor in CSML100 cells in vivo.
In our present work we have continued the study of the positive control of mts1 gene transcription. We describe the composite enhancer located in the first intron of the gene. Transient transfection experiments revealed that the activity of this enhancer composed 180 and 12% of SV40 early enhancer activity in mts1-expressing CSML100 and mts1-non-expressing CSML0 cells correspondingly. We found two DHSs located in the first intron of the mts1 gene between +750 and +950 bp. The location of these DHSs was quite different in CSML0 and CSML100 cells and coincided with the main enhancer activity revealed by deletion analysis of the mts1 first intron sequences (Fig. 3).
The DNA-protein interactions in the vicinity of the mts1 first intron DHSs were studied in both cell lines. Two sequences designated as Sa (+811 to +841) and Sb (+788 to +800) were found to bind nuclear proteins extracted from both CSML0 and CSML100 cells. Sa includes minisatellite core sequence CTGGGCAGGCAG (+817 to +828) which was shown previously to interact with the 40-kDa murine protein Msbp-1 ubiquitously expressed in mouse tissues (18). We suppose that the proteins involved in the interaction with Sa in CSML0 and CSML100 cells were identical with Msbps (18, 19), because of 1) their sequence specificity and 2) the distinctive pattern of their motility in EMSA (Cr complexes, Fig. 4A). The deletion of the Sa sequence did not influence the transcriptional activity of the pdel9 construct transiently transfected in CSML100 or CSML0 (Fig. 3) cells.
Sb represented the sequence GGGGTTTTTCCAC, which differed from the B
consensus site GGGRNNYYCC (22) by R (purine) to T substitution at the
fourth position (where Y indicates pyrimidine). This sequence was able
to form complexes in vitro with the bacterially expressed
p65, c-Rel, and p50 proteins (31). However, to our knowledge this
sequence has never been described as a functional
B site before. The
functional
B sites most similar to the Sb sequence have been found
in the
-interferon intronic enhancer (GAATTTTCC) (32) and in the
promoter of the major histocompatibility complex class II associated
invariant chain gene (GGGTATTTCC) (33).
Apart from p50·p50 homo- and p50·p65 heterodimers Sb formed stable complex C1 with the 200-kDa protein. This complex was 5.6-10.8 times more abundant in CSML100 then in CSML0 nuclear extracts (Fig. 5A).
The functional importance of the Sb sequence was shown. Five copies of Sb sequence cloned downstream from the CAT gene created an artificial enhancer that was active in CSML0 and CSML100 cells (Fig. 6). Furthermore, we have studied the effect of mutagenesis of the Sb site on the mts1 natural enhancer activity. The mutation (mut1) which abolished the interaction with the p50·p50 and p50·p65 dimers but did not influenced the 200-kDa protein binding had no significant effect on the mts1 enhancer activity. On the other hand, the second type of mutation (mut2) that made the Sb site completely inactive with respect to any protein binding resulted in a 63-68% loss of enhancer activity.
We performed in vivo dimethyl sulfate footprinting analysis
of the occupancy of the antisense strand of the Sb sequence in mts1-expressing CSML100 and mts1-non-expressing
CSML0 cells. In CSML100 cells, the decreased intensity of the bands
corresponding to the +797, +798 and +800 guanine residues indicated
their direct involvement in DNA-protein interaction (Fig. 8). However,
+800 guanine is located outside of the region homologous to the B consensus sequence. On the other hand, this residue was contacted by
the protein in the composition of the Cs complex (Fig. 4B). Since the Cs complex had the same motility in EMSA as the C1 complex (data not shown), and the only one protein component of this complex was the 200-kDa protein, it is highly likely that the protein bound by
Sb in CSML100 cells in vivo was identical with the 200-kDa factor. Taken together, the mutagenesis and in vivo
footprinting data allowed us to conclude that the interaction of the
200-kDa protein with Sb was important for the functioning of the
mts1 natural enhancer. This was supported by the study of
the human mts1 first intron sequence homologous to the mouse
Sb. In the human genome this sequence contains the thymidine stretch
which is one nucleotide longer than that in the mouse Sb,
CTGGTTTTTTCCAC. This difference led to the inability of the human Sb to
interact with NF-
B in vitro. On the other hand, the human
Sb formed the stable C1 complex with the 200-kDa
protein.1
The nature of the 200-kDa transcription factor remains obscure. To our
knowledge, three high molecular weight proteins that exhibited a
sequence specificity similar to or identical with that of NF-B have
been described earlier. All of them belong to the family of
metal-finger transcription factors that activate transcription of the
number of tissue-specific genes and viruses and have overlapping but
distinctive functional properties. MBP1/PRDII-BF1 (34, 35) and
MBP2/AGIE-BP1/AT-BP1/MIBP1 (36-39) have molecular masses of 298 and
275 kDa, respectively, and are therefore different from the protein
that activates transcription of the mts1 gene. The molecular
mass of the third known protein from this group HIV-EBP2 (40) roughly
coincides with 200 kDa. On the other hand, in vitro
translated HIV-EBP2, as well as all other members of this family, had
affinity to the Ig
B site (40), whereas C1 complex was unaffected
even by a 100-fold excess of the unlabeled Ig
B sequence (Figs. 5
and 7). Thus, the 200-kDa protein may perhaps represent a new member of
this family.
P-41CATInt, P-41LUCInt, pM5+, and pM5 constructs were functional when
transfected in CSML0 cells. Why the Sb sequence neither interacts with
the 200-kDa protein in vivo (Fig. 8) nor activates transcription of the endogenous mts1 in CSML0 cells is not
clear. The different occupancies of the Sb site in CSML0 and CSML100 cells in vivo might due to the different concentrations of
the 200-kDa protein in these cell lines (Fig. 5A). The
repression of the mts1 gene in CSML0 cells might be caused
by DNA hypermethylation (13, 14, 16) or/and by some other negative
regulatory systems which remain obscure.
The mts1 gene was shown to be inducible in mouse macrophages
by lipopolysaccharide treatment (10) and in mouse
fibroblasts in response to phorbol 12-myristate 13-acetate1
and serum (2, 3). In many cases the cellular response to these stimuli
is achieved through the activation of the NF-B (as reviews see Refs.
41-46). The
B-like sequence element found in the first intron of
the mts1 gene could provide the transcriptional response of
this gene to phorbol 12-myristate 13-acetate and
lipopolysaccharide.
We thank Dr. Kriajevska and Dr. Ambartsumian for a critical reading of the manuscript, Dr. Jäättelä for the pfLUC plasmid, and M. Wieser for help with the manuscript preparation.