(Received for publication, June 20, 1996, and in revised form, October 2, 1996)
From the Laboratory of Chemoprevention, NCI, National Institutes of Health, Bethesda, Maryland 20892-5055
Cellular transformation driven by the E1A
oncogene is associated with the development of cellular resistance to
the growth inhibitory effects of transforming growth factor-
(TGF-
). We demonstrate that development of resistance occurs
simultaneously with decreased expression of TGF-
type II receptor
(TGF-
RII) mRNA and protein. To determine whether changes in
transcriptional regulation are responsible for the decreased receptor
expression in E1A-transformed cells, a series of mobility shift assays
was performed utilizing nuclear extracts from E1A-transformed and untransformed murine keratinocytes using radiolabeled positive regulatory elements (PRE1 and PRE2) of the TGF-
RII promoter. The
results from these assays suggest that E1A-transformed cells express
markedly lower levels of nuclear proteins that bind specifically to
PRE1 and PRE2. Transfection of both E1A-transformed and untransformed cell lines with a series of mutant promoter constructs confirmed that
both PREs contribute significantly to basal expression of TGF-
RII
and that inactivation of either element leads to markedly reduced
promoter activity. We conclude that development of TGF-
resistance
in E1A-transformed cells is achieved in part through transcriptional
down-regulation of the TGF-
RII gene and that this down-regulation
is the result of decreased expression of unidentified transcription
factor complexes that interact with PRE1 and PRE2.
Transforming growth factor-
(TGF-
)1 is considered the prototypical
multifunctional cytokine, playing a central role in various vital
cellular processes such as growth, differentiation, synthesis of matrix
components, and apoptosis (1, 2, 3, 4). Appropriate cellular responses to
extracellular TGF-
depend on a system of multiple TGF-
-specific
cell surface receptor proteins and are initiated with binding of
TGF-
ligand directly to the TGF-
type II receptor, which is a
constitutively active serine-threonine kinase (5, 6, 7, 8). When bound to
ligand, the type II receptor forms a heteromeric complex with the
TGF-
type I receptor, which then leads to phosphorylation and
activation of the type I receptor serine-threonine kinase and allows
intracellular signaling to proceed.
One of the most prominent effects of TGF- in vitro is
pronounced inhibition of growth of epithelial cells (9). However, it is
well known that epithelial-derived cancers typically demonstrate resistance to the growth inhibitory effect of TGF-
(10). Acquisition of such TGF-
resistance by clonal populations of tumor cells may
represent an essential step in the process of carcinogenesis. Preliminary evidence indicates that TGF-
resistance develops relatively late in the evolution of cancer, coinciding with a period
when tumor phenotype first becomes recognizably malignant (11).
Sporn and Roberts (12) first predicted that receptor defects would be
identified as the basis of TGF- resistance, a prediction soon
confirmed by the finding of a strong correlation between structural
abnormalities of the TGF-
RII gene and development of TGF-
resistance in human gastric cancer cell lines (13). Subsequently, a
specific TGF-
RII mutation has been identified and associated with
defective DNA mismatch repair in colon cancer cells from patients with
the HNPCC (hereditary nonpolyposis colon cancer) syndrome (14). A
similar association has been found in gastric cancer as well (15).
It is becoming clear that the TGF- signaling pathway subserves a
vital tumor suppressor function in various cell lines (16, 17).
However, we have identified several TGF-
-resistant cancer cell
lines, which express decreased levels of cell surface receptors and yet
contain no recognizable mutations in the TGF-
receptor genes.2 This observation has suggested to
us the possibility that transcriptional regulation or posttranslational
mechanisms may lead to development of TGF-
resistance in some
cases.
A high level of structural complexity for the promoter region of
TGF- RII enhances the potential for significant transcriptional regulation of this gene. Preliminary characterization of the promoter region of TGF-
RII has revealed the presence of two positive regulatory elements (PRE1 and PRE2) and at least one negative regulatory element (NRE) in addition to the core promoter element (18).
Discrete target sequences have been identified within each of these
transcriptional regulatory elements, and each appears to interact with
multiple nuclear proteins, including six putative novel transcription
factor complexes.
Cellular transformation driven by the E1A oncogene is a popular system
for modeling the genetic events of carcinogenesis. This model was first
introduced in 1962 when it was discovered that adenovirus infection
leads to sarcoma formation in newborn hamsters (19). Since then,
considerable effort has been directed toward dissecting the genetic
events associated with the development of malignancy in this system
(20, 21, 22, 23, 24, 25). The adenovirus genome has been completely sequenced, and
numerous separable genetic functions have been identified and
categorized as either `early' or `late' according to their time of
expression relative to the onset of DNA replication (26). Expression of one of the adenovirus early region genes known as E1A is required for
cellular transformation in vitro and has also been
associated with several other cellular effects including induction of
resistance to growth inhibition by TGF-.
To investigate the mechanism underlying development of E1A-mediated
TGF- resistance, a series of experiments was performed utilizing a
dl799N mouse keratinocyte cell line stably transfected with an E1A
expression vector (27). A second mouse keratinocyte cell line
transfected with an enhancerless SV2 promoter was used as a control.
Northern blot analysis demonstrated a marked decrease in TGF-
RII
mRNA expression from E1A-transfected dl799N cells compared with
control. Receptor cross-linking assay further revealed that decreased
mRNA levels were associated with diminished expression of TGF-
RII protein. Chloramphenicol acetyltransferase transfection assays
demonstrated that the regulatory activity of PRE1, PRE2, and NRE was
intact in dl799N cells although levels of transcription were
consistently lower than in SV2neo cells transfected with the
same constructs. Electrophoretic mobility shift assays revealed a
significant reduction in DNA binding by the transcription factor complexes interacting with both PRE1 and PRE2. This study suggests that
TGF-
resistance in E1A-transformed mouse keratinocytes is mediated
in part by transcriptional down-regulation of the TGF-
RII gene and
that this negative regulation occurs at both the PRE1 and PRE2
loci.
Pam212(SV2neo) and Pam212(dl799N) mouse keratinocyte cell lines were obtained from Paolo Dotto (27). Pam212(SV2neo) cells have been stably transfected with an SV2 promoter and neomycin resistance gene construct. Pam212(dl799N) cells have been stably transfected with an E1A expression vector. Both cell lines were maintained in Dulbecco's modified Eagle's medium containing low glucose, L-glutamine, 110 mg/liter sodium pyruvate, and supplemented with 10% (v/v) calf serum. dl799N cells additionally contained 100 µg/ml neomycin.
Northern Blot analysisTotal RNA was isolated from cells
with guanidium isothiocyanate/phenol/chloroform. 10 µg of RNA was
electrophoresed on a 1.0% agarose gel containing 0.66 M
formaldehyde, transferred to a Duralon-UV membrane, and cross-linked
with a UV Stratalinker (Stratagene). Blots were prehybridized and
hybridized in 1% bovine serum albumin, 7% (w/v) SDS, 0.5 M sodium phosphate, 1 mM EDTA at 65 °C.
Hybridized blots were washed in 0.1% SDS, 1 × SSC for 20 min
periods at 50 °C before film exposure. RNA blots were hybridized
with 32P-labeled cDNA probes for human TGF-1 and
TGF-
RII. A probe for GAPDH was used to control for sample
loading.
Cells were plated at a density of
1 × 106 cells/well in 6-well dishes. Cells were
washed twice with cold binding buffer containing 1 × minimal
essential medium, 25 mM HEPES, pH 7.4, and 1 mg/ml bovine
serum albumin fraction V. Binding was carried out with 100 pM 125I-labeled TGF- in the presence and
absence of 100-fold molar excess of unlabeled TGF-
, and cells were
incubated on a rotating platform at 4 °C for 2.5 h. Cells were
washed twice with cold wash buffer 1 containing 1 × minimal
essential medium and 25 mM HEPES, pH 7.4. 1.0 ml of 300 µM disuccinimidyl suberate was added to cross-link
associated proteins, and cells were incubated for 15 min at 4 °C.
Cells were washed twice with cold wash buffer 2 containing 250 mM sucrose, 10 mM Tris, pH 7.4, and 1 mM EDTA. Cellular protein was solubilized with buffer
containing 1% Triton X-100, 10 mM Tris, pH 7.4, 1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride,
and 1 µg/ml each of pepstatin and leupeptin while incubating on a
rotating platform at 4 °C. Lysate was centrifuged, and the supernatant was electrophoresed on a 4-10% linear SDS-polyacrylamide gel (Novex).
DNA
constructs were generated by polymerase chain amplification using
genomic DNA containing the 5-untranslated region of TGF-
RII as a
template. Amplified DNA fragments were cloned into a promoterless CAT
expression plasmid (pGEM4-SV0CAT) (28) using HindIII and
KpnI or XbaI restriction sites built into the
oligonucleotides used for amplification. The sequences of the
polymerase chain reaction-generated portions of all constructs were
verified by DNA sequencing. The plasmid containing the CAT gene alone
was used as the control. All CAT construct plasmids were purified by
two sequential CsCl banding steps.
For transient expression assays, cells were plated at 1.2 × 106/100 cm2 dish and cultured for 24 h before transfection by the calcium phosphate coprecipitation method with 5-10 µg of the appropriate plasmids purified by banding in CsCl. Cells were harvested 48 h after the addition of DNA. The extracts were then assayed for CAT activity. All transfections were repeated a minimum of three times. For normalization of transfection efficiencies in SV2neo and dl799N cells, a growth hormone expression plasmid (pSVGH) was included in cotransfections. Growth hormone expression was quantified using a growth hormone detection kit (Nichols Institute).
Nuclear ExtractsNuclear extracts of SV2neo and dl799N cells were prepared as described (29) with minor variation. Monolayers of cells (3 × 106 to 5 × 106) were harvested by scraping, washed in cold phosphate-buffered saline, and incubated in 2 packed-cell volumes of buffer A (10 mM HEPES, pH 8.0, 1.5 mM MgCl2, 10 mM KCl, 0.5 mM dithiothreitol, 200 mM sucrose, 0.5 mM phenylmethanesulfonyl fluoride, 1 µg of both leupeptin and aprotinin/ml, and 0.5% Nonidet P-40) for 5 min at 4 °C. The crude nuclei released by lysis were collected by microcentrifugation, rinsed once in buffer A, and resuspended in a two-thirds packed-cell volume of buffer C (20 mM HEPES, pH 7.9, 1.5 mM MgCl2, 420 mM NaCl, 0.2 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, 1.0 mM dithiothreitol, and 1.0 µg of both leupeptin and aprotinin/ml). Nuclei were incubated on a rocking platform at 4 °C for 30 min and clarified by microcentrifugation for 5 min. The resulting supernatants were diluted 1:1 with buffer D (20 mM HEPES, pH 7.9, 100 mM KCl, 0.2 mM EDTA, 20% glycerol, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and 1 µg of both leupeptin and aprotinin/ml).
Electrophoretic Mobility Shift AssayDouble-stranded
oligonucleotides representing the first and second enhancer regions as
well as a series of mutant oligonucleotides for each region were
generated using an oligonucleotide synthesizer. Two oligonucleotides,
TRII-(
219/
172) and T
RII-(+1/+50), were labeled using a
fill-in reaction with [
-32P]dCTP (50 µCi at 3,000 Ci/mmol) and the Klenow fragment of Escherichia coli DNA
polymerase I. The fragments were then gel purified using a 6%
polyacrylamide gel and autoradiography to locate the specific fragment.
Binding reactions contained 10 µg of nuclear extract protein, buffer
(10 mM Tris, pH 7.5, 50 mM NaCl, 1 mM dithiothreitol, 1 mM EDTA, and 5%
glycerol), 2 µg of poly(dI-dC), and 30,000 cpm of
32P-labeled DNA in a volume of 10 µl. Reactions were
incubated at room temperature for 20 min. Competition reactions were
performed by adding an unlabeled double-stranded oligonucleotide to the reaction mixture. Reactions were electrophoresed on a 6% NOVEX precasted nondenaturing polyacrylamide gel at 100 V for 1 h in a
100 mM Tris borate-EDTA buffer. Gels were vacuum dried and
analyzed by autoradiography.
Northern blot analysis was utilized to analyze mRNA
expression levels for the TGF- RII and TGF-
1 genes. Fig.
1 shows that expression of TGF-
RII mRNA is
significantly lower in E1A-transformed dl799N cells compared with the
control SV2neo cells. In contrast, expression of TGF-
1
mRNA is equivalent between the two cell lines. GAPDH mRNA was
used to control for sample loading and indicates that the sample
quantity was slightly greater for dl799N indicating that the reduction
in TGF-
RII mRNA expression was even greater than that
visualized.
RNA was isolated from cell samples treated with 1 nM
retinoic acid, 1 nM 1,25-(OH)2D3,
or a combination of both retinoic acid and
1,25-(OH)2D3. A previous study involving human
myeloid leukemia cell lines reported an increase in expression of the
TGF- type II receptor following treatment with retinoic acid (30).
Retinoic acid and/or 1,25-(OH)2D3 treatment of
the murine keratinocyte cell lines used in our study, however, had no
apparent effect on the level of TGF-
1 or TGF-
RII mRNA
expression.
A receptor cross-linking assay using
125I-labeled TGF- was performed to determine whether
decreased mRNA levels for TGF-
RII translated into decreased
protein expression. Receptor cross-linking was performed three times,
and Fig. 2 shows representative results. Levels of
TGF-
type III and type I receptor cross-linked protein were
minimally decreased in the E1A-transformed dl799N cells compared with
control. The level of TGF-
type II receptor protein, however, was
markedly decreased.
Transcriptional Activity of the TGF-
In order to determine the functional activity of the
two reported positive regulatory elements in the TGF- RII promoter
region, progressively shorter fragments of the 5
-flanking region fused with the coding region of the bacterial CAT gene in the plasmid pGEM-SV0CAT were transfected into both SV2neo and dl799N
cells. As seen in Fig. 3, transcriptional activity was
less in dl799N cells than in SV2neo cells for each construct
tested. In both cell lines, deletion of the sequence between
1000 and
274 led to a marked increase in transcriptional activity, suggesting
the presence of an as yet uncharacterized negative regulatory element upstream of PRE1. Deletion of the sequence between
274 and
137 removed PRE1 and led to a predictable decrease in CAT activity. Deletion of nucleotides
137 to
47 eliminated the sequence for a
putative NRE (18) and led to a marked increase in transcription levels.
Deletion of the core promoter element between
47 and +2, as expected,
significantly decreased transcriptional activity. A relatively high
level of CAT activity remained with transfection of the smallest
construct, +2 to +50, indicating that the second positive regulatory
element, PRE2, was active in both cell lines. Transcriptional activity
was consistently higher in SV2neo cells compared with
the E1A-transformed dl799N cells, varying from three to more than six
times greater. Regulatory activity of PRE1, PRE2, and NRE appears to be
intact in both E1A-transformed and control cell lines.
Identification of a Single DNA Binding Protein from SV2neo Cells That Interacts with PRE1 Target Sequence
Previous
characterization of the TGF- RII promoter region identified two
discrete DNA binding protein complexes, an AP1/CREB-like transcription
factor complex and a putative novel transcription factor complex. To
determine whether these two protein complexes were similarly present in
mouse keratinocyte cells, we synthesized a series of mutant
oligonucleotides derived from PRE1 (Fig. 4A). Each mutant oligonucleotide contained a 4-base pair substitution mutation. EMSA was then performed using a radiolabeled PRE1 probe incubated with SV2neo nuclear extract in competition with
the series of mutant oligonucleotides. As shown in Fig. 4B,
specific binding was observed with a single protein complex that
migrated at the same position as complex a, the previously
identified AP1/CREB-like transcription factor complex. Competitive
binding was abolished by mutation of the complex a target
sequence. To confirm the identity of the single protein complex,
another EMSA was performed with radiolabeled PRE1 oligonucleotide probe
incubated with SV2neo nuclear extract in competition with
double-stranded oligonucleotides representing the consensus sequences
for several known transcription factors. Only the unlabeled AP1 and CRE
sequences demonstrated competition with the labeled PRE1 probe for
binding to the protein complex, indicating that it represents the
AP1/CRE-like transcription factor (data not shown). Complex
b (18) interactions were extremely low or undetectable.
Expression of E1A Is Associated with Decreased Interaction of Complex "a" with PRE1
To determine whether E1A expression
affects interaction of transcription factor complex a with
PRE1, another EMSA was performed utilizing a labeled PRE1 probe and
nuclear extract from both dl799N and SV2neo cells. As shown
in Fig. 5, interaction of complex a with PRE1
is markedly reduced in the E1A-expressing dl799N cell line compared
with control. Lanes 1 and 3 represent reaction
mixtures containing 5 random-labeled oligonucleotide representing PRE1 (
219 to
172) incubated with 10 µg of purified SV2neo
or dl799N nuclear extract, respectively. Lanes 2 and
4 demonstrate specificity of binding and represent identical
reaction mixtures in competition with unlabeled PRE1 oligonucleotide.
The level of activity represented by the minor bands is equivalent in
both cell lines, demonstrating equal loading of protein. Semiquantitive
analysis of the absolute levels of activity represented by the
individual bands in this assay using a scintillation counter suggests a
relative decrease of approximately 5-fold, which correlates well with
the observed decrease in TGF-
RII transcriptional activity in
E1A-expressing cells. However, direct quantitive comparisons of
DNA-protein complex binding and transcriptional activity should be
cautiously interpreted given the likelihood of additional factors
influencing overall levels of transcription.
Identification of Three Nuclear Protein Complexes from SV2neo Cells Interacting with PRE2
We have previously described three discrete
nuclear protein complexes that interact specifically with target
sequences within PRE2. To demonstrate that the mouse keratinocyte cell
line SV2neo expressed the same complement of proteins, we
synthesized another series of mutant oligonucleotides based on the PRE2
DNA sequence (Fig. 6A). Each mutant
oligonucleotide contained a 5-base pair substitution mutation. EMSA was
performed using a 32P-labeled PRE2 double-stranded
oligonucleotide probe incubated with SV2neo nuclear extract
in competition with each mutant oligonucleotide. As shown in Fig.
6B, lane 1, three major protein complexes were observed in the absence of competing oligonucleotides. Mutation of
nucleotides +16 to +20 (AAGTG, M4) led to decreased
competition for binding to complex c, while competition for
binding to complexes d and e was diminshed by
mutations through a longer sequence from +11 to +29 (AGTTTCCTGTTTCCC,
M3-M6). These target sequences are equivalent to
those previously identified (Figs. 6B and 7) in other cell
lines.
Expression of E1A Is Associated with Decreased Interaction of Complexes "c," "d," and "e" with PRE2
Another EMSA
was performed using a 32P-labeled PRE2 oligonucleotide
probe and nuclear extract from both dl799N and SV2neo cells. Fig. 7 shows that binding of all three complexes,
a, b, and c, to PRE2 is significantly reduced in
E1A-expressing dl799N cells compared with control. Lanes 1 and 3 contain reaction mixtures consisting of 5
random-labeled oligonucleotide representing PRE2 (+1 to +50) incubated
with 10 µg of purified SV2neo or dl799N nuclear extract,
respectively. Lanes 2 and 4 contained identical reaction mixtures in competition with unlabeled PRE2 oligonucleotide to
demonstrate specificity of binding. Activity of the rapidly migrating
minor band in lanes 1 and 3 is equivalent,
indicating equivalent protein loading.
To compare the activity of PRE1 and PRE2
in both E1A-transformed and untransformed cells, we created a series of
CAT constructs containing various combinations of mutations and
deletions in the X and Y target sequences of PRE1
as well as the Z target sequence of PRE2. Fig.
8 presents a schematic representation of the constructs used. In each case, presence of a bar indicates that the wild-type sequence is intact while absence signifies that the sequence has been
mutated. These constructs were transfected into both SV2neo and dl799N cells, and CAT activity was assayed as a measure of transcriptional activity. Fig. 8 shows that basal transcriptional activity of the TGF- RII promoter in both cell lines is principally dependent on the Y target sequence on PRE1 and the
Z target sequence on PRE2. A comparison of the activity of
constructs
219/+35 and
219M7/+35 reveals that mutation of the
Y sequence markedly reduces measured CAT activity following
transfection into both the SV2neo control keratinocyte cell
line as well as the d1799N E1A-transformed cell line. Similarly,
comparing
219/+35 to
219/+35M3 demonstrates that mutation of the
Z sequence leads to an equally sharp decrease in basal
transcription rates in both cell lines. In contrast, mutation of the
X target sequence, which interacts with a binding protein
that appears not to be expressed in mouse keratinocytes, failed to
inhibit transcriptional activity to any significant degree
(compare activity of construct
219/+35 with construct
219M5/+35).
The highest level of transcriptional activity was obtained following
transfection of constructs containing intact Y and
Z target sequences (constructs 219/+35 and
219M5/+35),
and conversely, the lowest level of activity was observed with the use
of constructs containing mutations or deletions of the Y and
Z target sequences (constructs
219M7/+35M3 and
219M7/+2). Overall transcriptional activity was markedly reduced in
E1A-transformed cells (range from 0.6 to 5.2% acetylation) compared
with control (range from 1.3 to 23.5% acetylation). However, the
pattern of CAT activity was identical for the two different cell lines,
indicating that both PRE1 and PRE2 regulatory elements are functional
in dl799N cells despite their lower activity levels.
A growing body of experimental evidence supports the concept that
TGF- RII operates as a tumor suppressor gene (16, 17). Expression of
TGF-
RII is required for the growth inhibitory effects of TGF-
ligands on proliferating epithelial cells. Mutation of the TGF-
RII
gene has been observed in a number of different human malignancies,
including colon, gastric, and endometrial cancers, and is highly
correlated with the development of TGF-
resistance (15).
Transfecting human breast cancer and colon cancer as well as hepatoma
cells lacking type II receptor with wild type TGF-
type II receptor
restores sensitivity to TGF-
and decreases tumorigenicity in
transplanted breast and colon cancer cells (31, 32). Moreover,
transfection of antisense TGF-
1 into a TGF-
-sensitive FET colon
cancer cell line enhances its tumorigenicity (33).
Similarly, numerous studies indicate that transcriptional regulation
makes an important contribution to determining the level of expression
of TGF- RII. Transfection of TGF-
-sensitive murine myeloid cells
with the src oncogene results in development of TGF-
resistance, which correlates with decreased expression of TGF-
RII
protein and mRNA (34). Conversely, transfection of TGF-
-resistant esophageal epithelial cells with cyclin D1 leads to
enhanced expression of TGF-
RII mRNA and protein along with increased TGF-
sensitivity (35). Our previous study of the TGF-
RII promoter reported the presence of two separate positive regulatory
elements in addition to the core promoter element (18), and we have
since confirmed the presence of at least two additional negative
regulatory elements.3 Each regulatory
element interacts with multiple nuclear DNA binding proteins, including
several putative novel transcription factors, in a sequence-specific
manner. Such a complex array of interdependent regulatory elements and
binding proteins easily allows for the possibility of multiple parallel
transcriptional regulatory pathways.
Our findings, as presented in the current study, demonstrate that
transformation of keratinocytes with the E1A oncogene is associated
with decreased expression of TGF- RII protein and an absolute
reduction in steady-state levels of TGF-
mRNA resulting from
down-regulation of TGF-
RII promoter activity. Specific binding of
nuclear protein complexes to the first and second positive regulatory
elements of TGF-
RII is markedly decreased in E1A-transformed cells
compared with control, suggesting that transcription of TGF-
RII is
being repressed at the level of transcription factor expression or DNA
interaction. Quantitatively, the decrease in TGF-
RII transcript
level demonstrated in Fig. 1 and the lower cell-surface expression of
TGF-
RII shown in Fig. 2 may not appear to correlate with the 3- to
6-fold decrease in transcriptional activity from the reporter construct
revealed in Fig. 3. The influence of additional uncharacterized
enhancer- and/or repressor-like elements on the specific cellular
expression levels of TGF-
RII cannot be ruled out. Moreover,
absolute levels of transcript and protein are likely influenced by
additional variably active cellular processes that make a direct
quantitative analysis impossible.
Similar transfection experiments have been performed utilizing SNU
gastric cancer cells resistant to growth inhibition by TGF-.
Preliminary data suggest that PRE2 may have a more significant role in
determining TGF-
RII transcription levels in SNU
cells.4 In this study, it also appears as
if the interaction of complexes c, d, and
e with PRE2 demonstrate a greater overall decrease compared with the interaction of complex a and PRE1. Moreover, Fig. 8
reveals that the greatest difference in transcriptional activity occurs when PRE2 alone is altered. These data suggest that the nuclear factors
interacting with PRE2 may constitute the more important targets in
transcriptional regulation of TGF-
RII expression.
A previous study by Missero et al. (27) also examined the
phenomenon of TGF- resistance resulting from transformation of keratinocytes with the E1A oncogene. Their data revealed that development of resistance to TGF-
1 growth inhibition most closely correlated with binding of E1A proteins to the retinoblastoma gene
product, pRb, and three other unidentified cellular proteins, p60,
p107, and p300. These results led them to conclude that TGF-
resistance in cells transformed with E1A developed via a postreceptor mechansim involving negative interactions with downstream elements of
the TGF-
signaling pathway.
Although it is conceivable that a postreceptor mechanism also
contributes to the observed decrease in TGF- sensitivity that occurs
in E1A-transformed cells, our data strongly support the existence of an
additional mechanism that results in direct transcriptional down-regulation of TGF-
type II receptor expression as well. Moreover, our finding that E1A expression results in decreased binding
of specific nuclear proteins to the TGF-
RII promoter suggests an
alternative interpretation of the results previously reported by
Missero et al. The p60 protein has been identified as a
human cyclin A protein (37); however, one or both of the other two
unidentified proteins, which are negatively regulated by E1A oncogene
products, may be identical to the putative novel transcription factors
responsible for activating transcription of TGF-
RII.
E1A-mediated effects in other cell systems have been demonstrated to occur through transcriptional down-regulation of specific gene expression. Transfection of mouse C2 myocytes with the E1A oncogene inhibits myogenic differentiation of these cells (37). E1A exerts its effect in these cells through two separate mechanisms, inhibition of expression of a myogenic regulatory factor, MyoD, and repression of MyoD-activated transcription (38). Both effects require the presence of two separate conserved regions of the E1A oncogene. These regions are identical to the domains of E1A required for transformation of keratinocytes, supporting the likelihood that the transforming properties of E1A similarly involve transcriptional regulation.
Adenovirus-host cell interactions have also served as an informative
model for studying a particular type of transcriptional regulation
involving nucleotide-specific DNA methylation. This mechanism for
repressing gene transcription occurs in mammalian DNA most commonly at
deoxycytodine residues within dinucleotide couplets known as CpG
islands. One such CpG island is located in the inactive target sequence
of TGF- RII PRE1, and multiple CpG islands can also be found
immediately flanking the PRE2 target sequence. Insertion of adenovirus
DNA has been shown to alter methylation patterns of adjacent host DNA
and, consequently, has clear potential to affect transcriptional
activity of nearby promoter elements (39). Although DNA methylation is
unlikely to contribute to the TGF-
RII down-regulation observed in
our study, it is interesting to speculate that enhanced methylation of
PRE1 and/or PRE2 sequences plays a complementary role in repressing
TGF-
RII expression as well as viral oncogenesis, and we are
currently performing a preliminary analysis of TGF-
RII promoter
methylation patterns.
In summary, this study demonstrates that E1A transformation of
keratinocytes is associated with down-regulation of TGF- type II
receptor expression resulting from a specific decrease in
transcriptional activation of the TGF-
RII promoter region. The
mechanism underlying this decreased expression has been shown to
involve decreased interaction between several putative novel
transcription factor proteins and their specific target sequences
within the two positive regulatory regions, PRE1 and PRE2, of the
TGF-
RII promoter. Further research is being performed to determine
whether decreased binding is a consequence of decreased expression of
the transcriptional factors, decreased activation of an inactive form,
or competitive binding. The findings from our study provide additional
support for the concept that the TGF-
type II receptor subserves an
important tumor suppressor function in mammalian cells. Inactivation of the receptor in E1A-transformed keratinocytes disables the earliest stage of the TGF-
signaling pathway and abolishes TGF-
-mediated growth inhibition, and insofar as this model reproduces key general molecular events of carcinogenesis, transcriptional down-regulation of
TGF-
RII may play a corresponding role in development of the TGF-
-resistant, aggressive cell growth demonstrated by various human
carcinomas.
We thank Paolo Dotto for mouse keratinocyte cell lines and John Letterio and Anita B. Roberts for critical reading of the manuscript and helpful suggestions.