(Received for publication, July 2, 1996, and in revised form, October 11, 1996)
From the Max-Planck-Institute of Biochemistry,
Department of Molecular Biology, Am Klopferspitz 18A,
82152 Martinsried, Federal Republic of Germany,
§ SUGEN, Inc., Redwood City, California 94063, and
the ¶ Cattedra di Oncologia Medica, Universita G. D'Annunzio, Via dei Vestini 6, 66100 Chieti, Italy
90K is a secreted glycoprotein with tumor
suppressive functions, which is up-regulated in various types of cancer
and in AIDS. In order to understand the regulation of its expression,
the mouse 90K gene was isolated and analyzed. The gene spans about
8.8-kilobase pairs and consists of 6 exons and was localized on
chromosome 11, region E. RNase protection identified one major
transcription start site (+1) and three minor ones (3, +32, +34). The
mouse 90K gene was found to have a TATA-less promoter of unusual
structure. The 2.3-kilobase pair 5
-flanking region exhibited strong
promoter activity in NIH 3T3 cells; however, it contained neither a
TATA-box nor a SP1 site and was not GC-rich. No known initiator motif
was found around the transcription start site. 5
- and 3
-deletions defined a minimal promoter of 51 base pairs (
66
16), not
including the start site, essential and sufficient for promoter
activity. This minimal promoter showed increased activity after
stimulation with interferon-
or poly(I·C), a substance mimicking
viral infection. Essential for both inductions was the integrity of an
interferon regulatory factor element within this sequence, a potential
binding site for the anti-oncogenic transcription factor interferon
regulatory factor-1.
90K is a secreted glycoprotein of 90 kDa, originally identified in the supernatant of human breast cancer cells (1, 2). Purification and cloning revealed its identity to the binding protein of human Mac-2 (3), a soluble and membrane-associated lectin thought to play a role in cell adhesion and immune response, and high similarity to a mouse cyclophilin C-associated protein (4, 5). Besides a scavenger receptor cysteine-rich domain (6), no similarity was observed to other proteins.
90K was detected in serum of healthy persons and, at elevated levels, in serum of patients suffering from various types of cancer (1, 10), AIDS (7-9), or autoimmune diseases.1 In many mammary carcinomas an inverse correlation was found between expression of 90K and mRNA levels of the HER2 receptor tyrosine kinase, a marker for poor prognosis for patients with breast cancer (2).
Activation of the immune system was suggested by increased natural killer and lymphokine-activated killer cell activity of peripheral blood lymphocytes after treatment with 90K (2). Natural killer cells destroy cells lacking major histocompatibility complex class I molecules, a common consequence of virus infection or malignant transformation (11). Furthermore, an increase of 90K expression in mouse mammary carcinoma and glioblastoma cells significantly reduced their tumorigenicity in nude mice (12). The tumor suppressive activity of 90K was local as well as systemic, since a 90K expressing tumor could inhibit the growth of a distally implanted tumor. Perhaps as part of the tumor suppressive mechanism, 90K expression led to strong induction of intercellular adhesion molecule-1 and vascular adhesion molecule-1 in the tumor endothelium (12). Based on these observations, we hypothesized that 90K plays an important role in the body defense against cancer, viral infections, and possibly other pathogens.
How expression of 90K is regulated is only poorly understood. Previous
studies reported up-regulation of its secretion in cancer patients by
exogenous IFN2 (13) and in human tumor
cell cultures by IFN
and IFN
(13, 14). In mouse macrophages the
90K mRNA was shown to be induced by IFN
, tumor necrosis factor
, and adherence (5).
In order to define the molecular basis of transcriptional regulation of
90K, we isolated in this study the mouse 90K gene and analyzed its
structure. The promoter region was characterized, and elements
mediating induced transcriptional induction after treatment with IFN
or poly(I·C), a polynucleotide mimicking the double-stranded RNA of
viruses, were identified.
An
oligo(dT)-primed cDNA library of 3T3 L1 adipocytes in UniZap
(Stratagene, prepared by Dr. I. Sures) was screened using a full-length
human 90K cDNA probe, random-primed labeled with [-32P]dATP (Amersham Int., UK) (15). Hybridization was
at relaxed stringency in 30% formamide, 5 × SSC at 42 °C,
washing at 42 °C in 0.2 × SSC, 0.1% SDS. A full-length
cDNA clone of mouse 90K was obtained and checked by sequencing
using the dideoxy chain termination method (16), comparing it with the
published sequence (5). This cDNA was random-primed labeled with
[
-32P]dATP and applied as a probe screening a genomic
mouse library of 129 SVJ mice in
FixII (Stratagene) at stringent
hybridization conditions (50% formamide, 5 × SSC). Positive
clones were isolated and further checked by polymerase chain reaction
(PCR; Ref. 21), amplifying short sequences from 5
-noncoding, coding,
and 3
-noncoding regions of the mouse 90K cDNA to confirm isolation
of a complete genomic clone. Restriction mapping and Southern blot
analysis were carried out according to standard procedures (15). Phage DNA was digested with EcoRI and NotI. A
0.8-kilobase pair (kb) and a 12-kb NotI-EcoRI
fragment were subcloned into pBluescript II KS (+) (Stratagene). A
4.6-kb EcoRI-EcoRI fragment was introduced into
pLXSN (17). All exons, the exon-intron borders, and the 5
-flanking
region were sequenced using the dideoxy chain termination method.
Intron sizes were confirmed by PCR, using oligonucleotide primer pairs
flanking the introns.
Mouse chromosomes were prepared according to the published procedure (18). Chromosome slides were made by conventional method (hypotonic treatment, fixation, and air dry). The DNA probe, the 12-kb EcoRI-NotI fragment of the mouse 90K genomic clone, was biotinylated using the BRL BioNick labeling kit (19). The procedure for FISH detection was performed according to Heng et al. (19) and Heng and Tsui (19). FISH signals and DAPI banding pattern were recorded separately by taking photographs, and the assignment of the FISH mapping data with chromosomal bands was achieved by superimposing FISH signals with DAPI-banded chromosomes.
RNase ProtectionThree sequences of the 5-flanking region
of the mouse 90K gene, overlapping the putative transcription start
region, were amplified by PCR and subcloned into pBluescript II KS (+)
(RMD2:
66
126, 192 base pairs (bp); RMD5:
66
91, 157 bp;
RMD7:
112
91, 203 bp). To facilitate subcloning, the PCR primers contained 5
-overhangs, creating a HindIII site at the
upstream, and a XbaI site at the downstream end of the
amplified fragment. Plasmid DNA was prepared using Qiagen midi-prep
columns (Diagen, Germany), linearized by HindIII, and
purified by Qiaex (Diagen, Germany).
For preparation of RNA probes, 1 µl of linearized template
(approximately 200 ng) was mixed with 1 µl of 10 × transcription buffer (Boehringer Mannheim, Germany), 1 µl of 40 units/µl RNasin (Boehringer Mannheim, Germany), 1 µl of 125 µM UTP (Pharmacia, Sweden), 1 µl of 5 mM
(GTP + ATP + CTP) (Pharmacia, Sweden), 4 µl of
[-32P]UTP (3000 Ci/mmol; Amersham, UK) and 1 µl of
20 units/µl T7 RNA polymerase (Boehringer Mannheim, Germany) and
incubated for 1 h at 37 °C. Then, 100 µl of H2O
and 1 µl of 10 mg/ml Escherichia coli tRNA (Boehringer
Mannheim, Germany) were added. After extraction with 1 volume of
Chloropane (CHCl3/phenol, 1:1, saturated with 10 mM sodium acetate, pH 6.0, 0.1 M NaCl, 1 mM EDTA), nucleic acids were precipitated by adding 0.7 volumes of 5 M ammonium acetate and 2.5 volumes of ethanol.
Samples were incubated for 30 s in liquid nitrogen and centrifuged
for 10 min at 10,000 × g. The pellet was washed with
70% ethanol, resuspended in 5 µl of formamide dye, denatured for 2 min at 95 °C, and applied on a sequencing gel (5% polyacrylamide, 8 M urea). After running, the RNA probe was located by
autoradiography, excised, and extracted from the smashed gel piece by
shaking for 2 h in 200 µl of elution buffer (0.5 M
ammonium acetate, 0.1 M EDTA, 0.1% SDS, 50 µg/ml E. coli tRNA). Gel pieces were removed by spinning through a
mini glasswool column, and the RNA was precipitated as described
above.
200,000 cpm of radiolabeled RNA probe were mixed with 30 µg of total RNA or 0.5-2 µg of poly(A)+ RNA and dried completely in a vacuum centrifuge. The pellet was resuspended in 10 µl of hybridization buffer (80% formamide, 40 mM PIPES, pH 6.7, 0.4 M NaCl, 1 mM EDTA), heated for 5 min at 90 °C, and incubated overnight at 50 °C. Then 100 µl of RNase digestion buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA, 200 mM NaCl, 100 mM LiCl, 40 µg/ml RNase A, 2 µg/ml RNase T1 (both enzymes from Boehringer Mannheim, Germany)) were added. After 60 min incubation at 37 °C 2 µl of 10% SDS and 3 µl of 10 mg/ml proteinase K (Boehringer Mannheim, Germany) were added for another 15 min incubation at 37 °C. The digest was transferred to a fresh tube, extracted with chloropane, and precipitated as described above. The pellet was resuspended in 5 µl of formamide loading buffer and analyzed on a denaturing sequencing gel. The length of the protected fragments was estimated by comparison to DNA fragments of a sequencing reaction run in parallel. Size estimation of several undigested RNA probes of known length by this method was accurate (data not shown).
Construction of 90K CAT Fusion GenesA 4.6-kb
EcoRI-EcoRI fragment containing exon 1 and exon 2 was subcloned into pLXSN. Deletion mutants were generated from this
clone by PCR (21). To facilitate subcloning of PCR fragments, upstream
primers contained a 5-overhang creating a HindIII site and
downstream primers a 5
-overhang creating a XbaI site. The PCR fragments were cloned into pCAT-Basic (Promega) digested with HindIII and XbaI. The sequence of the amplified
fragments was checked by DNA sequencing (16). To generate MD3B, a
0.62-kb HindIII-EcoRI fragment of the mouse 90K
gene, directly upstream of the 4.6-kb EcoRI-EcoRI
piece, was introduced into a pBluescript construct containing already
1.7 kb of the upstream side of the 4.6-kb fragment, beginning at the
EcoRI site. The fused 2.3-kb fragment was subcloned into
pCAT-Basic. Sequencing of a PCR fragment derived from DNA of the
genomic phage, which overlapped the fusion EcoRI site,
proved the validity of the construct. Constructs with point mutations
were generated by PCR using mismatch primers.
NIH 3T3 cells (ATCC CRL 1658) and CRL
6468 (ATCC CRL 6468) were maintained in Dulbecco's modified Eagle's
medium high glucose (Life Technologies, Inc., UK) supplemented with
10% fetal calf serum (Sigma), 1 mM sodium pyruvate, 100 units/ml penicillin, and 0.1 mg/ml streptomycin at 37 °C, 5%
CO2. All plasmids used for transfection were purified by
CsCl gradient (21). Subconfluent cultures (approximately 80%
confluent) were transfected using the BBS method (15), applying 3.5-5
µg of 90K CAT construct and 0.8 µg of pCMV (Clontech) per 35-mm
well. After 72 h cells were washed once with PBS (137 mM NaCl, 2.7 mM KCl, 1.5 mM
KH2PO4, 8.1 mM
Na2HPO4, pH 7.4) and lysed with 100 µl of CAT
lysis buffer supplied with the CAT enzyme-linked immunosorbent assay
kit (Boehringer Mannheim, Germany). 10-50 µl of the lysate were
tested for CAT protein in the CAT enzyme-linked immunosorbent assay
according to the instructions of the manufacturer.
-Galactosidase
activity was assessed following the method of by Nolan et
al. (22), adjusted to microtiter plates. Briefly, 10 µl of
lysate were mixed in a microtiter plate with 190 µl of PM-2 buffer
(20 mM NaH2PO4, 80 mM
Na2HPO4, 0.1 mM MnCl2,
2 mM MgSO4, 40 mM
-mercaptoethanol, pH 7.3). Then 50 µl of 4 mg/ml
o-nitrophenyl-
-D-galactopyranoside (Sigma) in
PM-2 were added to each well. After incubation for 10-30 min at room
temperature, optical density was measured at 420 nm. The CAT lysis
buffer did not influence the
-galactosidase activity as tested in
independent experiments previously. The relative promoter activity was
calculated by dividing the amount of CAT protein by the
-galactosidase activity.
For the stimulation experiments, cells of two 35-mm culture dishes were
trypsinized after transfection, combined, and reseeded into two 15-mm
culture dishes. After overnight incubation one of the dishes was
treated with 375 units/ml mu IFN (Sigma) or poly(I·C), whereas the
other dish was used as a control. For treatment with poly(I·C), cells
were washed two times with PBS+ (PBS supplemented with 0.9 mM CaCl2 and 0.5 mM
MgCl2) and incubated for 1 h with serum-free
Dulbecco's modified Eagle's medium containing 100 µg/ml DEAE-dextran (Pharmacia, Sweden) and 10 µg/ml poly(I·C) (Sigma). Afterwards, cells were washed again two times with PBS+ and incubated further for 72 h with normal growth medium. Control cells were treated with growth medium without IFN
or with serum-free
Dulbecco's modified Eagle's medium without DEAE-dextran and
poly(I·C). DEAE-dextran alone did not induce 90K message (data not
shown). Cells were lysed and assayed for CAT protein as described
above. Stimulation was calculated by dividing amount of CAT enzyme of
treated cells by that of untreated cells.
10-cm culture
dishes with confluent NIH 3T3 were stimulated as described above. After
indicated times cells were transferred on ice, washed once with PBS,
and lysed by adding 1 ml of GTC solution (4 M guanidine
thiocyanate, 25 mM sodium citrate, 0.5% sarcosyl, 100 mM -mercaptoethanol, pH 7.0). The lysate was refluxed through a 27-gauge needle to shear genomic DNA. Then, 0.1 volumes of 2 M sodium acetate, pH 4.0, 1 volume of phenol equilibrated with H2O, and 0.2 volumes of CHCl3 were added
sequentially, mixing after each addition. After 15 min incubation on
ice, the samples were centrifuged for 10 min at 10,000 × g. The aqueous phase was transferred to a new tube. RNA was
precipitated by addition of an equal volume of isopropyl alcohol,
incubation for 30 s in liquid nitrogen, and centrifugation for 10 min at 10,000 × g. The RNA pellet was resuspended in
300 µl of GTC solution, and RNA was precipitated again with 1 volume
of isopropyl alcohol. After washing with 70% ethanol, RNA was taken up
in 50 µl of H2O. Northern blotting was carried out
following standard procedures (15). As a probe, the complete mouse 90K
cDNA was used, random primed labeled with [
-32P]dATP. To control the amount of RNA applied, the
blot was hybridized with a TaqI-StyI fragment of
the mouse glyceraldehyde-3-phosphate dehydrogenase cDNA.
Screening of a mouse genomic
DNA library identified six different recombinant clones. After initial
characterization by restriction mapping and PCR amplification of short
sequences from the 5-noncoding, coding, and 3
-noncoding region of the
mouse 90K cDNA, a clone was chosen containing all sequences
represented in the cDNA clone (2) and more than 2 kb of 5
-flanking
sequence. This clone was analyzed further by restriction mapping,
Southern blotting, and PCR (Fig. 1). Following digestion
with EcoRI and NotI, three fragments of 0.8, 4.6, and 12 kb were subcloned, and the complete sequences represented in the
90K mRNA and 5
-flanking sequence were characterized.
Chromosomal Localization
Using FISH mapping, the chromosomal
location of the mouse 90K gene was determined. As a probe, a 12-kb
fragment of the 90K gene was used. Under the conditions used, the
hybridization efficiency was 60% for the probe (among 100 checked
mitotic figures, 59 of them showed signals on one pair of the
chromosomes). Upon identification of the chromosomes by DAPI banding
and evaluation of 10 preparations, the 90K gene was located on
chromosome 11, region E (Fig. 2). This corresponds well
to the location of human 90K to 17q25 (23).
Exon-Intron Organization
The murine 90K gene was determined
to contain 6 exons separated by 5 introns, defining a gene of about 8.8 kb in size (Fig. 1). Exon sizes ranged from 74 bp to 1.4 kb and intron
sizes from 0.13 to 2.4 kb (Table I). All exon-intron
borders were consistent with the reported consensus sequence for
5-splice donor and 3
-splice acceptor junctions in mammalian genes
(24, 25). While exon 1 contains noncoding sequences, exon 2 is composed
of a noncoding stretch of nucleotides, the initiation codon, and most
of the leader peptide coding region. The only identified subdomain of 90K, the scavenger receptor cysteine-rich repeat, is encoded by exons 3 and 4, which are separated by a small intron of 0.13 kb.
|
Transcription start sites were
assessed using RNase protection assays. Three different probes
overlapping the putative start region (RMD2, RMD5, and RMD7;
Fig. 3A) were applied on poly(A)+
RNA prepared from two unstimulated mouse mammary tumor cell lines (ATCC
CRL 6374 and ATCC CRL 6468). In addition, total RNA derived from
unstimulated NIH 3T3 cells, showing a very low level of 90K mRNA,
and from NIH 3T3 treated for 24 h with IFN or poly(I·C), as
described under "Materials and Methods," was used. Both agents induced high levels of 90K message in NIH 3T3 cells (cf.
Fig. 6). RMD2 resulted in protected fragments of 126 nucleotides (nt; major signal), 93, 95, and 129 nt (Fig. 3B, left panel: lane
1 (CRL 6468), lane 2 (CRL 6374); right panel:
lane 10 (poly(I·C), lane 11 (IFN
)). RMD5 and RMD7
showed an identical pattern of protected fragments of 91 (major
signal), 58, 60, and 94 nt (Fig. 3B, left panel: lanes 4 and
7 (CRL 6468), lanes 5 and 8 (CRL
6374), of RMD5 and RMD 7, respectively; right panel: lanes 4 and 7 (poly(I·C), lanes 5 and 8 (IFN
), of RMD7 and RMD 5, respectively). Since the probes RMD5 and
RMD7 share the same 5
-end, which is hybridizing to the mRNA of 90 kDa, this identity was expected (Fig. 3A). The 5
-end of
RMD2 is 35 nt further downstream. On that account, the protected
fragments detected with RMD2 should be 35 nt longer than the fragments
detected with RMD5 or RMD7, which exactly agrees with the observed size
differences. Induction of 90K mRNA in NIH 3T3 correlated with
increased amounts of protected fragments, demonstrating the specificity
of the signals (Fig. 3B, right panel: compare lanes 6, 9, and 12 (untreated) to lanes 4, 7, and
10 (poly(I·C) and lanes 5, 8, and 11 (IFN
)). tRNA, used as negative control, did not yield any protected
fragments (Fig. 3B, left panel: lanes 3, 6, and
9).
All three probes indicated one major site of transcription initiation
(designated +1) and three minor ones 3 bp upstream and 31 and 33 bp
downstream of the major start site (Fig. 4). These start sites were
identical in unstimulated mammary tumor cells and in NIH 3T3 cells
stimulated with IFN or poly(I·C), suggesting the same promoter was
responsible for transcription of the 90K gene in induced and noninduced
cell lines (Fig. 3B, left panel: compare lanes
1-3 (CRL 6374) to lanes 4, 5, 7, 8, 10, and
11 (NIH 3T3)). This notion was confirmed by long exposures
of the RNase protection gels, which revealed a protected fragment
corresponding to the major transcription start site even with RNA of
untreated NIH 3T3 cells.
The single major transcription start site suggested the presence of an
initiator element guiding the transcription machinery to that
nucleotide (26). However, alignment of the DNA sequence surrounding the
major transcription start site of the mouse 90K gene with known
initiator sequences did not reveal strong similarities. Only a certain
degree of similarity was detected to the downstream part of the
porphobilinogen deaminase (PBGD) initiator element (Fig. 3C;
Ref. 26). The sequence "TCCTGG" is found in the human PBGD
initiator element and, 6 bp downstream of the major start site, in the
90K gene. Mutations in this sequence prevented accurate transcription
initiation of the human PBGD gene in vitro. The upstream
part of the human PBGD initiator element, however, is not found in the
90K gene. Most importantly, the "C" at position 1 of the PBGD
initiator, shown to be essential for initiator function, was replaced
by an "A" in the 90K gene.
2.3 kb of 5-flanking
DNA of the mouse 90K gene were sequenced and screened for potential
transcription factor binding sites by comparison to the data base
"TFSITES" using the GCG program "FASTA" (Fig.
4). Several TATA box-like sequences were found, all of
them, however, in reverse orientation and located far upstream of the
major transcription start site and therefore unlikely to be functional.
It was shown that the sequence TACAAA can behave like a low affinity
TATA box (28). This sequence motif is present exactly at the major
transcription initiation site. However, since mRNA synthesis is
usually initiated 25-30 bp downstream of the TATA box, this TACAAA
sequence is very likely not responsible for the major transcription
start. Still, it might result in mRNA synthesis beginning at the
two minor start sites downstream (+32, +34).
The 90K gene, therefore, seems to belong to the class of TATA-less promoters. The majority of these promoters characterized to date display a very high GC content, possess multiple SP1 binding sites, and have a CpG island covering the transcription start (29). None of these features was found in the mouse 90K gene (GC content and CpG islands were checked using the GCG program "BASE PAIRPLOT").
Sequence analysis identified potential binding sites for NF-I (2103;
Ref. 30), AP-1 (
1512,
815; Ref. 31), AP2 (
491; Ref 32), and
progesterone receptor (
1455; Ref. 33). There is no CAAT box close to
the transcription start. At position
28 a sequence was detected shown
to bind the transcription factors IRF-1 and IRF-2 in vitro
(IRF-E; Ref. 34). IRF-1 is an activating transcription factor induced
by IFN
as well as by IFN
and IFN
. The transcription factor
IRF-2 is mostly inhibiting but can be activating as well, as shown
recently. Overlapping with this motif is a sequence "TTTCTGAAA"
(
33
25), similar to the GAS consensus element "TTNCNNNAA"
(35). GAS elements mediate transcriptional activation by IFN
. These
sites are interesting, since 90 kDa is known to be up-regulated by
IFN
and IFN
(5, 13, 14). Also tumor necrosis factor
was shown
to induce 90K gene expression (5). However, no NF
B site was found in
the 2.3-kb 5
-flanking region.
To
examine the promoter activity of the 5-flanking region of the mouse
90K gene, a series of CAT expression plasmids were constructed
containing various deletions of the 5
-flanking region. Using PCR,
fragments were generated and cloned into the plasmid pCAT-Basic
immediately upstream of the CAT encoding sequence. Promoter activity
was assessed measuring the amount of CAT enzyme in NIH 3T3 cells
transfected with these constructs. To compensate for different
transfection efficiencies, CAT values were normalized with
-galactosidase activity, as described above.
The 2.3-kb piece of 5-flanking region demonstrated strong promoter
activity (MD3B; Fig. 5), being significantly above the negative control of pCAT-Basic. This activity was comparable with that
of the SV40 early promoter. Interestingly, however, 90K gene expression
is expressed in NIH 3T3 cells only at a very low level. This apparent
discrepancy might suggest the existence of additional negative elements
in the 90K gene or of other regulatory mechanisms controlling 90K
mRNA levels.
5-Deletion up to
1235 (MD8) led to a small increase in promoter
activity, whereas further deletion to
133 (MD21) had no significant
effect on promoter strength. While deletion to
67 (MD2) increased
activity about 2-fold, indicating negatively acting elements in the
region
132
67, removal of additional 51 bp led to complete loss
of promoter activity (MD22). Since this mutant still contains the
sequence around the major transcription start site, this region
apparently does not possess any promoter activity. This is in contrast
to several described initiator sequences, which by themselves can
initiate low level transcription (26).
3-Deletions from +126 to +92 (MD5) and to
15 (MD1N), including the
transcriptional start site, decreased promoter activity, pointing to
positive elements in these parts of the 90K gene. The reduced activity
of MD1N compared with MD2 could also be due to the loss of the
endogenous transcriptional start site. Further deletion up to
98
(MD12) completely abolished promoter activity, confirming the
nonfunctionality of the upstream TATA box-like sequences.
Together the 5- and 3
-deletions allowed to identify a short region
from
66 to
16 to be essential for promoter activity. Testing this
sequence by itself, a significant promoter activity was measured
(MD1N). This minimal promoter, therefore, seems to be essential as well
as sufficient for transcriptional activity. Remarkably, this sequence
does not comprise the major transcription start site.
Previous works
reported induction of 90K mRNA in various cell lines by IFN and
IFN
(5, 13, 14). Poly(I·C), a substance mimicking the
double-stranded RNA of viruses, was shown to elicit gene induction
similar to that observed after infection with Newcastle disease virus,
probably via a common pathway triggered by double-stranded RNA (36).
Since we speculated that 90K might play a role in host defense against
viruses, we tested IFN
and poly(I·C) for their effect on 90K
mRNA in NIH 3T3 cells. As shown in Northern blotting (Fig.
6), both substances could induce 90K message levels. The
kinetics of the induction was similar. After 8 h an increase was
clearly detectable, becoming rather pronounced the following 16 h,
especially with poly(I·C).
To test
whether the isolated 5-flanking region of the mouse 90K gene can
mediate an increased promoter activity after stimulation with IFN
and poly(I·C), NIH 3T3 cells were transfected with various 90K gene
CAT fusion constructs. Transfected cells were split, and one-half was
treated with the stimulating agents, while the other was used as a
control. The extent of stimulation was determined by dividing the
amount of CAT protein of treated cells by that of untreated.
IFN increased the promoter activity of the 2.3-kb 5
-flanking region
of the 90K gene 3-5-fold, in different experiments (Fig. 7). 5
-Deletions up to
67 (MD2) and 3
-deletion to
15 (MD1N) did not significantly change this induction. Poly(I·C)
enhanced promoter activity of MD3B 6-10-fold, in different experiments (Fig. 7). 5
-Deletion to
1235 decreased inducibility by half, indicating a poly(I·C) responsive element in the region
2183
1235. Further 5
-deletions up to
66 and 3
-deletions to
16 had no influence on the inducibility. These results show that the
minimal promoter is able to mediate enhanced transcription after
stimulation with poly(I·C) or IFN
.
Functional Importance of IRF Element within the Minimal Promoter
The minimal promoter contains at its downstream end an
IRF-E, which could mediate stimulation by IFN or poly(I·C),
e.g. via the anti-oncogenic transcription factor IRF-1. To
assess the function of this IRF-E several mutants of the minimal
promoter have been prepared (Fig. 8). MD1 contains two
point mutations not interfering with the IRF-E consensus, and MD24 four
point mutations destroying the IRF-E motif. The IRF-E is deleted
partially in MD23 and completely in MD25 and MD26.
Comparison with the wild type minimal promoter (MD1N) revealed that
destruction of the IRF-E significantly reduced basal promoter activity
(Fig. 9A, MD23-MD26). Still, even
the biggest truncation of the minimal promoter tested (MD25) retained
transcriptional activity at least 3-fold above background (Basic).
Integrity of the IRF-E consensus sequence was also essential for the
inducibility by IFN or poly(I·C), since disruption of the IRF-E
motif leads to complete loss of induction (Fig. 9B,
MD23-MD26). Stimulation values below 1, as observed for
MD23-MD26, might be due to a general inhibitory effect of the
treatments on protein synthesis. Modification of the IRF-E sequence
without destroying the consensus (MD1) did not influence stimulation by
poly(I·C), while a slightly reduced activation was observed with
IFN
. In conclusion, these data indicate that the IRF-E within the
minimal promoter contributes to the basal activity and is mediating
transcriptional activation by IFN
or poly(I·C).
Promoter Elements Functional Also in Mammary Carcinoma Cell Line CRL 6468
We found recently that overexpression of 90K protein in
mouse mammary cell lines significantly reduced their ability to form tumors in nude mice (12). In order to see whether the mouse 90K
promoter elements characterized in NIH 3T3 are similarly functional also in other cell lines, we transfected four promoter constructs (MD3B, MD10, MD1N, and MD23) into the easily transfectable mouse mammary carcinoma cell line CRL 6468. This cell line has a basal amount
of 90K mRNA higher than NIH 3T3, which is about 5-fold increased by
treatment with poly(I·C) or IFN (data not
shown;).3
As in NIH 3T3, the minimal promoter (MD1N) was able to mediate induced
transcription after stimulation with poly(I·C) (Fig. 10C) or IFN (Fig. 10B). This
induction was dependent on the IRF-E within the minimal promoter since
partial deletion of this sequence completely abolished stimulation
(Fig. 10, B and C, MD23). A
poly(I·C) responsive element not responsive to IFN
was detected
far upstream of the minimal promoter (Fig. 10, B and
C; compare MD3B with MD10). The
relatively low degree of induction might be explained partly by the
fact that for technical reasons (reseeding, growing to complete
confluency) the stimulation was started only 36 h after transfection. The relative promoter activity of MD3B and MD10 compared
with MD1N was higher than in NIH 3T3, which could be related to the
higher basal level of 90K mRNA observed in CRL 6468. As in NIH 3T3,
partial deletion of the IRF-E significantly reduced promoter activity
(Fig. 10 A, compare MD1N with MD23). Taken together, these data indicate that 90K promoter elements found in
NIH 3T3 are similarly functional in the mammary carcinoma cell line CRL
6468.
The promoter of the mouse 90K gene is a TATA-less promoter of unusual structure. Promoters without a TATA box have previously been found in constitutively expressed "housekeeping" genes, which in addition are GC-rich, contain several SP1 sites, and a CpG-rich island, and initiate transcription at multiple sites (29). A second group of TATA-less genes is not GC-rich and possesses only one or few tightly clustered transcription start sites. These discrete transcription initiation points are encompassed by an initiator element, containing all information necessary for determining specific initiation of transcription, both in vivo and in vitro (26). The 90K promoter, in contrast, is not GC-rich, initiates transcription mainly at one single site, and is lacking so far characterized initiator elements. The partial similarity to the PBGD initiator (27) does not include the upstream part of the motif, shown to be essential for initiation for that gene. Furthermore, deletion of the region around the transcription start site did not abolish promoter activity in the mouse 90K gene (Fig. 5, MD1N), in contrast to the PBGD gene, where the initiator is essential for strong promoter activity (27).
Few promoters of mammalian genes have been analyzed that like the 90K
promoter do not fit into the groups described above, being TATA-less,
not GC-rich, and without known initiator sequences: the promoters of
cytosolic phospholipase A2 (37), which is important for prostaglandin
and leukotriene synthesis of c-mos (38), a germ cell
restricted proto-oncogene of C4 (39), a complement component, and of
mouse DNA methyltransferase (Ref. 40). Whereas the cPLA2 promoter has a
long run of CA repeats, the methyltransferase promoter contains a high
amount of GT dinucleotides. Both the CA repeats and the GT-rich region
could be deleted without affecting the promoter activity. In all of
these genes a small region around the transcriptional start site was
essential for promoter activity. No further similarities could be
identified. The 5-flanking region of the mouse 90K gene did not
contain a long run of CA repeats or a region rich in GT dinucleotides.
There is a stretch of 88 bp (
923
836), enriched in G and T
(28.5 and 62.5%, respectively). Deletion of this region had no
influence on transcriptional activity. In contrast to the other
promoters, the region around the transcription start of 90 kDa was
dispensable for promoter activity.
The lack of features in the 90K promoter common to other genes suggests a rather distinct way of linking the 90K promoter to the basal transcription machinery. Conceivably, this could be a feature defining specific and characteristic regulation of 90K expression.
A small sequence of 51 bp (66
16) upstream of the major start
site was found to be essential and sufficient for transcriptional activity. Sequences upstream (
2183
1235,
132
113,
112
67) decreased, and sequences downstream (
17
+91, +92
+ 126) increased promoter activity. To this minimal promoter proteins should bind that directly or indirectly interact with RNA polymerase II, similar to the TATA box binding protein in TATA box containing promoters (26). TFII-I, a 120-kDa protein specifically binding to the
terminal deoxynucleotidyltransferase initiator, might play this role in
certain TATA-less promoters (41).
The minimal promoter also mediated transcriptional induction by IFN
or poly(I·C). Point mutations and deletions within the minimal
promoter showed the integrity of the IRF-E at
28
16 to be
essential for the stimulation. This activation may involve binding of
IRF-1 to the IRF-E. Similar to 90K protein the expression of this
transcription factor is increased by treatment of cells with IFNs,
poly(I·C), and by viral infection (36, 42, 43) and also by other
stimuli such as tumor necrosis factor
, interleukin-1, and
interleukin-6 (44-46). Experiments with IRF-1 knockout mice, however,
suggested that at least poly(I·C) can induce the binding of
activating transcription factors besides IRF-1 to the IRF-E (47).
Moreover, IRF-2, normally an inhibiting transcription factor binding to
the IRF-E, could be involved in stimulation as reported lately (48). It
was recently shown that specific activation of the IFN
gene requires
a defined set of regulatory elements, including two IRF-1 binding
sites, which are precisely arranged on the DNA surface (49). It will be
interesting to test if similar cooperative effects take place in the
minimal promoter of 90 kDa.
The isolated 2.3-kb 5-flanking region of the 90K gene possessed strong
promoter activity and mediated inducibility by IFN
and poly(I·C).
Still, additional regulatory regions upstream or downstream, or
different regulatory mechanisms, like chromatin structure or mRNA
stability, for example, might modulate the amount of 90K mRNA. The
high promoter activity in low expressing NIH 3T3 cells suggests a
negatively acting mechanism in these cells independent from the 2.3-kb
5
-flanking region.
Induced transcription by poly(I·C), a double-stranded RNA mimicking
viral infections, indicated that virus infection may increase 90K
levels not only via induction of IFNs but also possibly directly involving IRF-1. Another pathway seemed to be IRF-1-independent, since
the region 2183
1235 increased poly(I·C) induction, while not
containing an IRF-E. Moreover, this element did not respond to
IFN
.
We recently reported the inhibitory effect of 90K protein on tumor growth of mammary cell lines in nude mice (12). Based on this result it would be conceivable that certain oncogenes block, and certain anti-oncogenes stimulate, 90K expression. Interestingly, IRF-1 has been described as a tumor suppressor gene (50-52). Embryonic fibroblasts from mice with an inactivated IRF-1 gene can be transformed by expression of c-Ha-ras oncogene. Expression of IRF-1 reverted this phenotype. Furthermore, overexpression of IRF-2, an antagonist of IRF-1, in NIH 3T3 cells enhanced their tumorigenicity in nude mice. Concomitant overexpression of IRF-1 prevented this effect. IRF-1 is very likely activating a set of genes whose products are required for the negative regulation of cell growth. It is tempting to speculate that 90K belongs to this group of genes.
In summary, the functional analysis of the 90K promoter appears to be in line with our hypothesis of 90K as a part of the body defense against viral infections and cancer. Characterization of proteins interacting with the 90K promoter and comparison of promoter activity in normal and malignant cells will be carried out to further our understanding on how 90K expression is regulated.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) U76646[GenBank].
We thank Dr. K. Wiebauer for her kind help in establishing the RNase protection assay. We are grateful to Professor M. Beato for his advice and to Dr. S. Harroch for discussion and encouragement. The poly(A)+ RNA for the RNase protection assay was kindly provided by Dr. E. Imyanitov. FISH mapping was carried out by Dr. H. Heng, SeeDNA Biotech Inc.