Download c-fos Protein Can Induce Cellular Transformation: A Novel

Cell, Vol. 36, 51-60
January 1984, Copyright
0 1964 by MIT
0092.8674/84/010051-IO
c-fos Protein Can Induce Cellular
Transformation:
A Novel Mechanism
Activation of a Cellular Oncogene
A. Dusty Miller, Tom Curran,
and lnder M. Verma
Molecular Biology and Virology Laboratory
The Salk Institute
P. 0. Box 85800
San Diego, California 92138
Summary
The FBJ murine osteosarcoma virus (FBJ-MuSV)
induces tumors in vivo and transformation in vitro.
Transformation is due to the expression of a single
viral protein (p55’-f0*) which is encoded by sequences derived from mouse genetic material. The
homologous cellular gene (~40s) does not transform
cultured cells after introduction by transfection. We
show that even though the c-fos protein is completely different from the v-fos protein at its C terminus, it is capable of transforming cultured fibroblasts. However, activation of the transforming potential of the c-fos gene requires two manipulations-a transcriptional enhancer sequence must be
linked to the gene and an interaction at the 3’ end
of the gene, which inhibits transformation, must be
disrupted. Our studies show that normal cellular
protein can induce transformation when expressed
in an inappropriate cell type.
Introduction
Retroviruses capable of inducing tumor formation after a
relatively short latency often contain and express normal
cellular genetic information (Weiss et al., 1982). The protein
products of these viruses contain domains homologous to
normal cellular proteins. It is the general consensus that
neoplastic transformation is due either to expression of a
normal cellular protein at inappropriately
high levels or to
the expression of an altered form of a normal cellular
protein.
The FBJ murine osteosarcoma virus (FBJ-MuSV) is an
acutely oncogenic retrovirus that arose by recombinational
events rnvolving the FBJ murine leukemia virus and mouse
genetic material (Curran et al., 1982). FBJ-MuSV causes
osteosarcomas in mice and can transform tissue culture
cells (Finkel et al., 1966; Levy et al., 1973; Curran and
Teich, 1982a). The virus encodes a single 55,000 dalton
protein (~55”“‘“) whose coding sequences are derived
entirely from the mouse genome (Curran and Teich, 1982b;
Van Beveren et al., 1983). DNA sequences homologous
to the viral transforming gene (v-fos) have been molecularly
cloned from both mouse and human genomes (Curran et
al., 1983). These clones appear to contain complete cellular fos (c-fos) genes, including consensus sequences for
RNA promotion and polyadenylation. The organizations of
the mouse and human c-fos genes are similar in that they
both have lengths of about 3.5 kb and contain three introns
$02.00/O
of
(Van Beveren et al., 1983; van Straaten et al., 1983). The
predicted 2.2 kb mRNA product of the c-fos gene is found
in normal mouse and human tissues and is expressed at
the highest levels in extraembryonal membranes (Muller et
al., 1982, 1983).
DNA sequence data predict that the mouse and human
c-fos proteins are 94% homologous, and are very closely
related to ~55”.“” with the exception of the C-terminal
region. The v-fos gene has undergone an out of frame
deletion such that its C-terminal 49 amino acids are completely different from the 48 C-terminal amino acids of the
c-fos protein (Van Beveren et al., 1983). While molecularly
cloned FBJ-MuSV proviral DNA containing the v-fos gene
is able to transform cultured fibroblasts, neither the mouse
c-fos nor the human c-fos gene induces transformation.
We describe a series of experiments designed to determine the factor(s) responsible for the difference in transforming potential of the c-fos and v-fos genes. The results
show that the c-fos protein itself can transform fibroblasts
and that activation of the c-fos gene requires two manip
ulations, linkage of transcriptional enhancer elements and
disruption of interacting sequences at the 3’ end of the
gene.
Results
Construction of v-fos/c-fos(mouse)
Molecules
Chimeric
We wanted to determine the factor(s) responsible for the
difference in the transforming ability of the v-fos and c-fos
genes. Our strategy involved construction of chimeric
plasmids containing portions from both genes. We describe the construction and confirmation of the plasmid
structures in detail because, as will be shown below,
activation of the c-fos gene is complex, and analysis of
activation would be difficult without the use of well defined
homogeneous plasmid DNA. The v-fos and mouse c-fos
genes can be arbitrarily divided into three regions (Figure
1) by convenient restriction enzyme sites: sequences 5’
of the Nco I site, including promoter sequences and 83%
of the fos protein coding region; sequences between the
Nco I and Sal I sites, spanning the v-fos gene deletion and
coding regions for the C termini of both the viral and
mouse fos proteins; and sequences 3’ of the Sal I site,
Including 3’ noncoding regions and putative polyadenylation srgnals. Out of 316 amino acids there are only 5
differences between the proteins coded by c-fos(mouse)
and v-fos in the region 5’ of the Nco I site, but there are
48 differences out of 65 amino acids in the Nco I to Sal I
region. The C-terminal 49 amino acids of the v-fos protein
are completely different from the C-terminal 48 amino acids
of the mouse c-fos protein because of a 104 bp deletion
in the v-fos gene (Van Beveren et al., 1983). In the human
c-fos gene there is no Sal I site, because of a single base
change, so that the C-terminal and 3’ untranslated sequences cannot be easily separated (Figure 1).
Chrmeric molecules containrng regions from both the v-
Cell
52
Nco I,
I
\
/Sal I
FBJ-MuSV
c - fos (mouse]
c - for
H
Frgure 1 Organizatron
IH
I
I
(human]
H
of v-fos and c-fos genes
The top line deprcts the FBJ-MuSV provirus. The v-fos coding regron IS
shown by an open box rn the mrddle of the provrrus. Rat DNA sequences
surroundtng the provrrus are shown by wavy Irnes. The middle and lower
lanes deprct the mouse and human c-fos genes, respectively,
wrth the cfos codrng regtons shown by closed boxes. The coding regions of each of
the c-fos genes are separated by three introns. The restnctron endonucleases Nco I and Sal I divide the v-fos and mouse c-fos genes into three
regions, and the human c-fos into two regions, as shown. RNA 5’ cap and
polyadenylatron
signals are shown. The origins of these plasmids are
described rn Experimental Procedures. The arrows in the FBJ-MuSV provrrus
and in the c-fos gene indicate the positions of recombination between the
mouse gene and the helper retrovirus that generated FBJ-MuSV.
fos (pFBJ-2) and mouse c-fos (pc-fos(mouse)-3)
genes
were constructed as shown in Figure 2. These hybrid
molecules were given three-letter code names signifying
the origin (from 5’ to 3’) of each region of the hybrid; M
indicates mouse and V indicates viral sequences. For
example, MVV signifies a plasmid with 5’ fos sequences
(up to the Nco I site) derived from c-fos(mouse), and Cterminal-coding
(Nco I to Sal I) and 3’ untranslated sequences (3’ of the Sal I site) derived from v-fos. We
confirmed the structures of the constructs by digesting
each chimeric plasmid with a combination of restriction
enzymes Xho I, Sal I, and Nco I, which yield DNA fragments
diagnostic for each of the mouse c-fos and v-fos gene
regions (Figure 3).
Transforming
Ability of v-fos/c-fos(mouse)
Hybrid
Molecules
The ability of recombinant plasmids to induce transformation was monitored using 208F rat fibroblasts as the
recipient cells. Uncut plasmid DNAs were transfected with
carrier DNA into the cells, and transformed foci were
scored approximately 3 weeks later. No foci were seen in
cells transfected with carrier DNA alone. Many of these
results were confirmed using NIH/3T3 cells as the indicator
cells, but foci observed on NIH/3T3 cells were less distinct
and therefore more difficult to quantitate. Table 1 A shows
the transforming potential of the v-fos/c-fos(mouse) recombinant molecules.
Mouse c-fos Gene Promoter is Functional
As already indicated, construct VVV (v-fos) transformed
cells, whereas MMM (c-fos(mouse)) did not (Table 1A).
MVV transformed cells with an efficiency similar to that of
VVV (Table 1 A), showing that the 5’ promoter for transcription in the c-fos(mouse) clone is functional. We also conclude from this result that the portion of the mouse c-fos
protein 5’ of the Nco I site can transform cells when linked
to the C terminus of the viral fos protein. The 3’ long
terminal repeat (LTR) in the viral portion of MVV provides
a polyadenylation
signal and contains enhancer sequences that have been shown to elevate gene expression
in other systems (Oskarsson et al., 1980; Blair et al., 1981;
Chang et al., 1982).
Altered C Terminus Does Not Affect Transformation
Both MVV and MMV transformed cells (Table IA), demonstrating that the middle fragment of these constructs
can be derived from either c-fos(mouse) or v-fos without
altering their transforming ability. Since the middle fragment
contains the coding region for the C terminus of the fos
protein, the results suggest that the transforming activity
of fos is unaffected by the large out of frame deletion in vfos which results in an altered C terminus. This conclusion
is supported by the observation that, like VVV, VMV also
induces transformation (Table 1 A). Taken together these
results show that any combination of viral or mouse sequences, coding for amino and carboxy regions of the fos
protein, will induce transformation provided that a 3’ LTR
is present. In particular, no difference was detected in
focus size or morphology between MMV and VVV transformed cells. Since the entire protein coding region in MMV
is derived from the mouse gene, we conclude that the
unaltered mouse fos protein can induce transformation.
Mouse c-fos Gene Contains Transcription
Terminator
Construct VVM induces transformation (Table IA), which
implies that the 3’ fragment of the mouse c-fos gene
contains signals required for proper termination of transcription. Size analysis of RNA in VVM transformed cells
(see below) revealed that transcription terminated at the
predicted c-fos polyadenylation site (Van Beveren et al.,
1983).
Cooperative
Inhibition by 3’ Mouse Sequences
Surprisingly, construct VMM, which contains viral 5’ promoter and enhancer elements as well as proper transcription termination signals, did not transform cells (Table IA).
Since VVM, VMV, and VVV induced transformation, we
conclude that there is an interaction between the middle
and 3’ mouse sequences that inhibits transforming activity,
presumably by reducing expression of the fos protein. To
be certain of this result, additional independent isolates of
VMM and VVM were tested for their transforming ability,
with identical results; VVM induced transformation whereas
VMM did not.
LTR Is Required for Transformation
No transformation was observed following transfection with
the plasmid MVM (Table 1A). MVM contains a functional
promoter and transcription termination signals and can
encode a transforming protein. Also, cooperative inhibition
of transformation by 3’ mouse sequences is disrupted in
c-fos Gene Actrvatron
53
Figure 2. Construction
menc Molecules
of v-fos/c-fos(mouse)
Chi-
In all DNA molecules depicted plasmid vehicle
sequences are shown as crrcular, while fos gene
and proviral sequences
are shown horizontally
above the vehrcle sequences. N. Nco I; H. Hind Ill;
B, Barn HI; X, Xho I; S, Sal I; Ss. Sst I; E, Eco RI.
Plasmids from which all hybrid molecules are derived are shown at the top. The mouse c-fos gene
is denoted by a heavy lkne wrth coding regions
shown by closed boxes. The FBJ-MuSV provrrus
IS denoted by a lrght line, the open box In the
middle of the provirus Indicates the v-fos coding
region, and the wavy lines Indicate surroundrng rat
genome-derived
DNA. Expected
mRNAs
are
shown above the constructs,
and sizes include
about 150 nucleotides to account for polyadenylation. All derivative constructs
were made using
two-way ligatrons rn which the DNA fragment containrng the ongin of plasmrd replication was treated
with bacterial alkaline phosphatase before Irgation.
Points of ligation are shown by breaks rn the
molecule and dashes between vehicle and fosrelated sequences. Constructs MMV, MW, VVM,
and VMM’ were denved drrectly from MMM and
VVV. Construct VMV was derived from VVV and
MMV, construct MVM from MW and MMM, and
construct VMM from VMM’ and VVM. All restrictron enzyme sates used in these constructions
are
shown on the parental plasmids, but only key sites
are shown on the constructs.
Whenever an endonuclease site IS shown on a plasmrd, all other
sates where that enzyme cleaves are also shown.
MVM because we have shown that both middle and 3’
untranslated mouse sequences are required for inhibition
to occur. However, unlike the transforming constructs
MMV, MVV, and VVM, MVM lacks an LTR. Only constructs
containing LTFis transform cells. LTRs have been shown
to contain transcriptional enhancer elements (Jolly et al.,
1983a; Luciw et al., 1983). We presume that the mouse cfos promoter is efficiently utilized in fibroblasts only when
the enhancer elements in the LTR are present.
Transformation
by v-fos/c-fos(human)
Hybrid
Molecules
Hybrid molecules containing human and viral sequences
(Figure 4) displayed a pattern of transformation (Table 16)
identical with that of the v-fos/c-fos(mouse)
chimeras.
Since the human fos gene does not have a Sal I site (Figure
1) only chimeric molecules utilizing the Nco I site could be
constructed. However, to be consistent we retained the
three-letter code to describe these molecules as if there
was a Sal I cleavage site. As previously noted, HHH did
not transform cells. HVV, however, did transform cells,
showing that the 5’ human sequences are functional and
can generate a transforming protein. VHH’ and VHH”,
which contain different lengths of cellular sequences 3’ of
the polyadenylation
signal (Figure 4) did not transform
cells, The C termini of c-fos(mouse) and c-fos(human)
proteins are very similar, since there are only 2 differences
out of 65 amino acids. Thus it seems unlikely, in view of
the v-fos/c-fos(mouse)
hybrid results, that the C terminus
of the human c-fos protein alone inhibited the transforming
ability of VHH’ and VHH”. The DNA sequence of the
human c-fos gene shows a high degree of homology with
the mouse c-fos gene, and furthermore, both VHH’ and
VHH” have polyadenylation
signals which are similar to
that of the mouse c-fos gene. Hence it is unlikely that lack
of transformation
by VHH’ and VHH” is due to the
absence of proper polyadenylation signals. Thus it seems,
as in the case of the mouse c-fos gene, that there is some
interaction between the middle and 3’ human fos sequences. Since the human c-fos C-terminal-coding and 3’
untranslated sequences cannot be easily separated, this
hypothesis was not confirmed. The hybrid construct HVM
(Figure 4) did not transform cells (Table 1B). By analogy
to construct MVM, HVM presumably does not transform
Table 1 Transformrng
23.0
-6.8
- 5.1
J 3.1
2.3 2.0 -
72.8
-
Transformatron
vvv
MMM
MVV
MMV
VMV
VVM
VMM
MVM
+
-
HHH
HVV
VHH’
VHH”
HVM
-
VW%
VMM’
VMM’(A),
MM(A).
+
-
1.8
-1.3
0.56 -
0.3
0.2
Frgure 3. Conftrmatron
of v-fos/c-fos(Mouse)
Hybrid Plasmrd Organizatton
Characteristic DNA fragments expected from each region of the c-fos and
v-fos clones after cleavage with Xho I, Sal I, and Nco I are shown below.
The locations of these restnction sates in the parental plasmrds are shown
In Figure 2.
Region
Expected
M-V-~
-M-V-
5.1, 1.8
6.8
0.3
0.2
2.8, 1.3
3.1
M
V
Clone
A
l-
9.4
6.6
4.4
Fragments
v-fos/c-fos(mouse)
hybrid plasmids were cleaved with Xho I, Sal I, and Nco
I, and about 2 Ag of DNA from each was electrophoresed
rn a 1% agarose
gel. The gel was stained with ethidrum bromide and photographed
under
UV Irght. Hind Ill cleaved X phage DNA was also electrophoresed
at both
sides of the gel to provide size markers; the sizes of the marker bands are
ltsted to the left of the gel. Srzes of the DNA fragments from the v-fos/cfos(mouse) chimeric molecules are shown to the right of the gel. The gel
was intentronally overloaded so that the 0.2 and 0.3 kb fragments could be
detected.
cells because it lacks an LTR.
Activation of VMM
We reasoned that VMM did not transform cells because
of an inhibitory interaction between the C-terminal-coding
sequences (Nco I to Sal I) and untranslated c-fos(mouse)
sequences 3’ of the Sal I site. Since the mouse c-fos
protein coding regions are all located 5 of the Sal I site,
this hypothesis can be tested by removing the mouse
sequences in VMM which are 3’ or the Sal I site. To
compensate for the loss of the polyadenylation
signal in
the 3’ mouse sequences, an SV40 early polyadenylation
signal was added. This new clone, VM(A), (Figure 5)
tranformed cells (Table 1C) with an efficiency similar to
that of VVV (Table IA). This result supports the hypothesis
Potential of Hybrrd Molecules
Panel
C
+
+
+
+
-
+
-
Clones were tested for transformrng abrlrty by transfectron onto rat 208F
cells, as described rn Expenmental Procedures
(+) Indicates transformatron
effrcrency of about 200 focr/pg DNA. (-) Indicates transformrng effrcrency
of <IO foc+g DNA.
that there is some interaction between the middle and 3’
c-fos sequences which interferes with fos expression.
We also constructed a hybrid molecule, VMM’ (Figure
2) which contains viral sequences 5’ of the Nco I site and
mouse sequences 3’ of the Nco I site up to a Barn HI
endonuclease site. This molecule contains only a portion
of the mouse sequences 3’ of the polyadenylation signal,
as compared with VMM. The shorter 3’ sequence is
denoted M’. VMM’ does not transform cells (Table IC),
indicating that the interaction between middle and 3’
mouse sequences can be confined, in the 3’ fragment, to
about 1 kb between the Sal I and Barn HI sites. VMM’(A),
(Figure 5) which contains the SV40 polyadenylation signal
attached to VMM’ beyond and in addition to the cfos(mouse) polyadenylation signal, also does not transform
cells. We therefore conclude that the reason VMM’ does
not transform cells is not because of the absence of a
functional polyadenylation signal. Since VM(A), transforms
cells and VMM’(A), does not induce transformation, it is
clear that insertion of the M’ sequences-i.e.,
mouse fos
gene sequences from Sal I to Barn HI-inhibits
the transforming ability of the DNA molecule. In addition, since VVM
transforms cells, this inhibition is due to some interaction
between C-terminal-coding and 3’untranslated
mouse sequences.
Another construct, MM(A)” (Figure 5) in which the 3’
mouse c-fos sequences were replaced with the SV40 early
polyadenylation signal, did not transform cells (Table 1 C).
Thus elimination of the interaction between middle and 3’
sequences is not sufficient to confer transforming ability
to the c-fos(mouse) gene. Compared to MMV, which does
transform cells (Table 1A), construct MM(A),, lacks enhancer sequences present in the LTR. We conclude that two
changes are required to activate the c-fos gene to transform cultured cells. Both disruption of the interaction be-
c-fos Gene Actrvatron
55
0
1
1
I
2
I
3
I
4
I
5
I
6
I
7
I
8 kbp
I
PC-loslhumanl-1
Figure 4 Constructron
of v-fos/c-fos(human)
Chimeric
Molecules
Recombtnant plasmtds and restrictron sites are displayed as described In Frgure 2. The molecular clone of the human fos gene, pc-fos(human)-1,
IS shown at
the top. c-fos genes (human and mouse) are shown by heavy lanes wrth codtng regrons Indicated by closed boxes. Viral sequences are shown by light lanes
and codrng regrons by open boxes. Hand Ill partial digests were employed rn the constructron of HVV, VHH”, and VHH’. HVV and VHH” were derived from
HHH and VVV, HVM from HW and MMM, and VHH’ from VHH” and HHH. All recombinants were checked for proper structure by restriction analysis.
tween the C-terminal-coding
and 3’ untranslated c-fos
sequences and insertion of an enhancer region near the
c-fos gene are necessary.
Low Level Transforming Activity of Some Chimeric
Clones
There is a clear difference between the transforming constructs (200 foci/yg DNA) and the nontransforming
constructs (<lo foci/pg DNA) shown in Table 1. However, a
low level transformation frequency is seen with certain
constructs denoted transformation
negative. Although
MMM, MVM, HHH, HVM, and MM(A), have never been
observed to transform cells, occasional transformed foci
are seen after transfection with VMM, VMM’, VMM’(A),,
VHH’, and VHH’ ‘. These results are further described
below.
Cotransfection of MMM and VMM with a Selectable
Marker
We have shown that both the viral and cellular fos proteins
can transform cells, and moreover, that exchanging the C
termini of these proteins does not affect their transforming
ability. Thus DNA molecules that code for fos proteins but
do not transform cells presumably do not transform because of insufficient expression of the fos protein. To
determine at what level in protein biogenesis this inhibition
occurs, we isolated cell lines containing these nontransforming DNAs by cotransfection with a selectable marker.
208F cells are hypoxanthine
phosphoribosyltransferase
(HPRT) deficient (Quade, 1979; Miller et al., 1983) so an
HPRT expression vector (p4aA8; Jolly et al., 1983b) was
used as the selectable marker. The HPRT expression
vector was mixed with a 20.fold excess of the nontransforming constructs MMM or VMM, or with the transforming
construct VVM, and transfected with carrier DNA into 208F
cells. HPRT+ cells were selected in medium containing
hypoxanthine, amethopterin, and thymidine (HAT). Such a
transfection protocol has been shown to yield a high
percentage of clones that contain the cotransfected DNA
as well as the selectable marker (Wigler et al., 1979). The
number of HPRT+ colonies with transformed morphology
was monitored visually (Table 2). Cotransfection of cells
with MMM never gave rise to a colony containing transformed cells. Cotransfection with VMM gave rise to colo-
Cell
56
Figure 5. Construction of Hybrids Containrng
SV40 Early Polyadenylation Srgnal
the
Constructs
are displayed as described in Figure
2. The SV40 polyadenylation srgnat is contained
In an SV40 DNA fragment of 0.26 kb rn length
from SV40 map posrtrons 0.14-0.19. The SV40
and pBR322 sequences rn these constructs were
derived from cDNA expressron
vector pcDV1
(Okayama and Berg, 1983). The plasmrd used in
the constructs
described here was p4aA8, the
HPRT expression vector (Jolly et al., 1983b). The
Barn HI and Xho I sites used to attach the SV40
sequences to mouse sequences are within the 10
bp of each other rn a DNA linker. Sal I/Xho I
lrgatrons occur because of a complementary
DNA
5’ overhang, but such ligated molecules cannot
be cleaved with Sal I or Xho I. VM(A), was denved
from VMM and p4aA8, VMM’(A). from VMM’ and
p4aA8, and MM(A),, from MMM and p4aA8. All
plasmrds were checked for proper restnctron enzyme patterns.
nies containing transformed cells at lo-fold lower frequency than cotransfection with VVM.
Analysis of lransfected
DNA
We isolated DNA from clonal cell lines cotransfected with
the HPRT expression vector and construct MMM, VMM,
or VVM. Cellular DNA was digested with Barn HI and Hind
Ill, which cleave just outside of the transfected fos genes,
and should reveal the presence of intact fos genes by
virtue of the size of the fos-containing
DNA fragment
produced (4.8 kb from MMM and VVM, 4.9 kb from VMM).
DNA fragments with sizes expected from intact fos genes
were identified in every clone, although many rearranged
copies were also detected (Figure 6A).
Analysis of RNA Transcripts
To determine if the inability of constucts MMM and VMM
to transform rat fibroblasts was at the level of transcription,
we analyzed the fos-specific RNA produced by these
clonal cell lines in comparison to that of cells transformed
by cotransfection with construct VVM (Figure 6B). fosspecific RNA of the expected size (2.7 kb) was detected
in VVM cotransfected
cells (lane 6). Transcripts of the
expected size (2.2 kb) which hybridized to a fos probe
were found in cells cotransfected with MMM (lanes l-3);
however, the amount of RNA was more than 15 times
lower than that in cells cotransfected with the transforming
construct VVM (lane 6). Therefore, MMM may not transform cells simply because it is not efficiently transcribed.
fos-specific RNA of the expected size (2.8 kb) was also
found in cell clones cotransfected with VMM that displayed
a nontransformed
morphology
(lane 4). Levels of fosspecific RNA in several nontransformed
VMM cotransfected clonal cell lines were only 2-3 times lower than
those in VMM cotransfected cells that displayed a transformed morphology (for example, compare lanes 4 and 6
in Figure 6B). Since we know that the protein synthesized
by the VMM construct can transform cells (because VM(A),
transform cells), and since cells cotransfected with VMM
express fos-specific RNA at levels similar to those of
transformed cells cotransfected with construct VVM, we
conclude that translation of fos-specific RNA in VMM
Table 2. Percentages of Colonres Contarnrng
Cotransfectron
of Construct DNAs
Transfected
DNA
Transformed
Transformed
HPRT only
co5
MMM + HPRT
co.5
VMM + HPRT
4
VVM + HPRT
40
Cells followrng
Colonres (%)
Two hundred HPRT+ colonres produced after transfectron wrth the rndrcated
DNAs were screened for transformed
morphology
The percentage
of
colonres showrng transformed morphology IS rndrcated.
cotransfected cells is inhibited. It is interesting to note that
the levels of fos-specific RNA in the nontransformed VMM
cotransfected cells (Figure 6B, lane 4) are higher than
those in the VMM cotransfected cells that display a transformed morphology (Figure 6B, lane 5). Thus the step in
protein biogenesis that is affected by the presence of
interacting 3’ mouse c-fos gene sequences appears to be
at the level of translation.
As noted above, some of the VMM cotransfected colonies appeared to be transformed (Table 2). One of these
transformed clones was analyzed along with the nontransformed clones. The fos-specific RNA detected in this clone
(Figure 6B, lane 5) does not have the size expected from
the VMM construct, and probably represents a transcript
from a rearranged gene. We have obtained similar results
in the case of VHH” transfected cells. Occasional transformed foci obtained after transfection of 208F cells with
VHH” were isolated by micromanipulation,
and fos-specific RNA produced by two clones was analyzed (Figure
6C). On the same gel, we have analyzed FBJ-MuSV
nonproducer cell RNA (3.6 kb expected size, lane 1) and
RNA from HVV transformed cells (2.9 kb expected size,
lane 2). The major fos-specific RNA from VHH” cotransfected cells should be about 2.8 kb in size. One VHH”
transformed clone displayed an RNA species of the expected size, while the other clone showed a larger RNA
species of 3.5 kb (Figure 6C, lanes 3 and 4). Thus occa-
c-ios Gene Actlvatlon
57
A
B
1234567
C
1234567
12
D
34
123456
78910
23-
9.4.
6.61.8-
1.8-
2.32.0.
Figure 6 fos-Specific
DNA, RNA, and Protern rn Cotransfected
Cell Clones
(A) DNA was extracted from cells, cleaved with Hind Ill and Barn HI, electrophoresed
I” an agarose gel (10 pg DNA per lane), transferred to nitrocellulose
paper, and hybridized to a fos-speclftc
probe. Lanes 1-3, DNA from three Independent clones of construct MMM cotransfected
cells, lane 4, DNA from a
nontransformed
clone of VMM cotransfected
cells; lane 5, DNA from a transformed clone of VMM cotransfected
cells; lane 6, DNA from a transformed clone
of VVM cotransfected
cells; lane 7, DNA from control 208F rat cells. The positlons of DNA size standards are shown. (B and C) Total RNA was extracted
from clonal cell Ilnes. electrophoresed
(IO pg RNA per lane), transferred to nitrocellulose paper, and hybridized to a fos-speclflc probe. The posItIons of the
rlbosomal RNAs are Indicated. (B) Lanes l-7 show RNA from the same cell lines used in lanes 1-7 of (A). (C) Lane 1. RNA from RS2 cells (FBJ-MuSV
nonproducer 208F cells); lane 2. RNA from cells transformed by transfection with HVV; and lanes 3 and 4, RNA from occaslonal transformants
Isolated after
transfectlon
with VHH”.
(D) Cotransfected
cells were labeled with %methlonlne
for 20 mm, and radlolabeled fos-speclflc
protein was analyzed by
lmmunopreclpdatlon
using tumor-bearing
rat serum (odd-numbered
lanes) and normal rat serum as a control (even-numbered
lanes) as described (Curran
and Telch, 1982a; T Curran. A. D Miller, L. Zokas, and I. Verma. manuscript submitted). In addition to ~55’“” proterns. cellular p39 proteins are copreclpltated
in thus procedure (Curran and Telch, 1982b). Lanes 1 and 2. 208F FBJ-MuSV nonproducer cells (RS2 cells); lanes 3 and 4, MMM cotransfected
cells (same
as In lane 1 of A and B); lanes 5 and 6, VVM cotransfected
cells (same as In lane 6 of A and 6); lanes 7 and 8, nontransformed
VMM cotransfected
cells
(same as in lane 4 of A and B), and lanes 9 and IO. another clone of nontransformed
VMM cotransfected
cells Molecular weight markers are listed to the left
of the gel.
sional transformed foci produced by constructs with a 5’
viral LTR linked to middle and 3’ c-fas sequences may be
due to rearrangements of the gene, resulting in disruption
of the 3’ inhibitory sequences.
Protein Analysis
Since the nontransforming construct VMM was adequately
transcribed in cotransfected cells, and the size of the
transcript matched the expected size, we analyzed these
cells for the presence of fos protein. Figure 6D shows that
while fos-specific protein can be seen in VVM cotransfected cells (lane 5), no corresponding
protein can be
detected in MMM (lane 3 ) or VMM (lanes 7 and 9)
cotransfected cells. lmmunofluorescence
staining of cells
using fos-specific antisera (T. Curran, A. D. Miller, L. Zokas,
and I. M. Verma, manuscript submitted), supports this
result (data not shown). We conclude from these data that
at least 5fold less fos protein is synthesized in cells
containing constructs VMM and MMM than in cells transformed by VVM.
Discussion
The c-fos Protein Induces Transformation
Nucleotide sequence analysis of the v-fos and c-fos genes
shows that the respective protein products differ from
each other at their C termini. Thus the protein product of
the v-fos gene is qualitatively altered from that of its
progenitor (c-fos). Initial experiments in which the 3’ sequences of the human c-fos gene were substituted such
that both the C terminus of the fos protein and transcriptional termination signals were provided by viral DNA suggested that the transforming potential of the fos protein
was dictated by the altered C terminus (Van Beveren et
al., 1983). However, a detailed analysis of v-foslc-fos
chimeric molecules shows that the c-fos protein can itself
transform fibroblasts, and that the mechanism of activation
of the c-fos gene is substantially more complex.
The entire protein coding region of the MMV construct
is derived from the c-fos gene. This chimeric molecule is
as efficient at inducing transformation of fibroblasts in vitro
as cloned FBJ-MuSV proviral DNA (Table 1 A). In addition,
proteins containing any mixture of v-fos and c-fos N- and
C-terminal regions can induce transformation, since MVV
and VMV also transform fibroblasts. The major difference
between the two proteins is at their C termini. Furthermore,
fos proteins expressed from constructs containing the cfos C terminus are extensively modified compared with
those containing the v-fos C terminus (T. Curran, A. D.
Miller, L. Zokas, and I. M. Verma, manuscript submitted).
Cell
58
We conclude that although c-fos and v-fos proteins are
qualitatively different, they are both capable of inducing
transformation. The contention that the C terminus of the
fos protein is not crucial for induction of transformation is
supported by sequence analysis of another murine osteosarcoma virus, FBR-MuSV, which also encodes a fosrelated transforming protein, The p75 gag-fos fusion protein of FBR lacks 98 fos C-terminal amino acids, which are
replaced with 8 amino acids derived from transcribed
mouse sequences not adjacent to the fos gene (C. Van
Beveren, S. Enami, T. Curran, and I. M. Verma, manuscript
submitted).
Requirements
for Activation
of the c-fos Gene
Linkage of an LTR
The normal c-fos protein can transform cells, but the c-fos
gene itself does not induce transformation. From the results presented here, it is apparent that the c-fos gene
promoter is functional in fibroblasts since the constructs
MVV, MMV, and HVV induce transformation. All three
constructs, however, contain viral LTRs at their 3’ ends,
which in addition to transcription termination signals contain transcription enhancer elements. With constructs like
MVM, HVM, or MM(A),, in which the 3’ LTR sequences
are not present and alternative poly(A) addition signals are
provided, no transformation of transfected cells was observed. We assume that the c-fos gene promoter requires
transcriptional enhancer elements to be fully functional in
fibroblasts. This suggests that a tissue-specific “enhancerlike” element in the c-fos gene may operate in normal
amnion cells, such as those described for immunoglobulin
genes (Banerji et al., 1983; Gillies et al., 1983; Queen and
Baltimore, 1983). Activation of the c-mos and c-H-ras-I
oncogenes by addition of either 5’ or 3’ LTRs has previously been reported (Oskarsson et al., 1980; Blair et al.,
1981; Chang et al., 1982).
Disruption of Interacting
3’ Sequences
The other change required for activation of the c-fos gene
is the disruption of interacting 3’ sequences. The construct
VMM does not transform cells even though an enhancer
is present in the 5’ LTR. It should be noted that the only
difference between mouse C-terminal-coding
sequences
(0.3 kb) and viral C-terminal-coding
sequences (0.2 kb)
are the 104 bp that have been deleted in the viral DNA
fragment, Hence the region in the mouse C-terminal coding
sequences that interacts with 3’ untranslated mouse sequences is well defined. Results obtained using the polyadenylation site from SV40 show that the 3’ untranslated
c-fos(mouse) sequences which interact with the mouse Cterminal-coding sequences are located within the 1 kb 3’
untranslated fragment present in VMM’(A),.
can transform cultured cells. Introduction of the c-fos gene
itself, however, does not transform these cells. Presumably
the normal c-fos gene does not transform cells because
expression of the fos protein is inhibited in these cells. This
negative control occurs at two levels.
On one level, transcription from the c-fos promoter is
reduced in cutured cells. Cultured cells into which the cfos(mouse) gene (construct MMM) has been introduced
by cotransfection with a selectable marker contain 15 times
less fos-specific RNA than cells cotransfected with the
transforming construct VVM (Figure 6B). This low level of
transcription can be substantially increased by linkage of
enhancer sequences. It should be noted that MMM was
introduced into cells by cotransfection
with an HPRT
expression vector that contains SV40 enhancer
sequences. It is possible that the enhancer sequences present in this vector could elevate expression of cotransfected MMM sequences. Thus transcription from MMM
may appear higher than it actually would be in the absence
of the cotransfected DNA.
The second level of control of c-fos gene expression is
due to the interaction of 3’ fos sequences This interaction
does not necessarily occur at the DNA level, but could be
at any of several steps in the production of the c-fos
protein from the c-fos gene. We have shown that levels of
fos-specific RNA in cells cotransfected with VMM, which
generally does not transform cells because of inhibitory 3’
sequences, are only about 2-3 times lower than those in
cells cotransfected with VVM, which does transform cells.
However, no fos protein was detected in VMM cotransfected cells. This result suggests that transformation by
VMM is inhibited at the level of translation. A precedent
exists for this type of regulation in the case of translational
inhibition of endogenous MMTV expression (Vaidya et al.,
1983).
We have identified two mechanisms for the control of
c-fos gene expression. One of these is unusual and seems
to operate by cooperative action between two different
regions of the gene. In another system, DNA sequences
near the mouse c-mos gene have recently been identified
which inhibit transformation by the c-mos gene linked to
an LTR (T. Wood and G. Vande Woude, personal communication). Perhaps such potent transforming
genes
must of necessity be held under rigorous restraint, involving multiple levels of control, since inadvertent activation
could result in neoplasia. However, the c-fos gene is
expressed at high levels in normal amnion cells. We therefore propose an alternative model to the quantitative or
qualitative models of neoplastic transformation, that oncogene expression in an inappropriate cell type can lead
to transformation.
Experimental
Regulation
of c-fos Expression
Since the c-fos gene is expressed during development in
a stage- and tissue-specific manner, its expression is
closely regulated. We have shown that the c-fos protein
Procedures
Recombinant
DNA
The isolation of the biologically active molecular clone of FBJ-M&V
as a
Hnd III insert in pBR322, called pFBJ-2, has been descrbed (Curran et al.,
1982). Isolation of the human c-f&s gene as an Eco RI Insert I” pBR322.
c-fos Gene Actrvatron
59
denoted pc-fos(human)-1.
has been reported (Curran et al., 1983). The
molecular clone of the mouse c-fos gene, pcfos(mouse)3,
was derived
from clones of portrons of the c-fos(mouse)
gene as follows. A cloning
vehicle contarnrng two Sst I sttes was constructed
by clonrng a 2.0 kb Pst
I fragment from A. MLV,,-I (Berns et al.. 1980) Into the Pst I site of pBF1322
to make pMLV,-105 (Fuhrman et al., 1981) which contarns mouse cellular
sequences and the 5’ portion of a Moloney-MuLV provirus. The orientatron
of the Insert IS such that the mouse sequences are farther from the Eco RI
sate of pBR322 than the vrral sequences. pMLV,-105 was digested wrth Sst
I and Eco RI and the 4.2 kb fragment containing pBR322 sequences and
0.6 kb of mouse cellular sequences was isolated and treated with bactertal
alkaltne phosphatase. Thus DNA fragment was ligated to an Sst I to Eco RI
fragment, containtng 5’ c-fos(mouse)
codtng sequences, derived from Xcfos(mouse)-2 (Curran et al., 1983) to make pc-fos(mouse)-2A.
The Eco RI
Insert of pc-fos(mouse)-I
(Curran et al., 1983) contarning 3’ c-fos(mouse)
coding sequences
was then cloned Into the stngle Eco RI site of pcfos(mouse)2A
The resultrng clones were screened for the correct orrentatron of the Eco RI Insert. and a clone was chosen that contained the
complete c-fos(mouse)
gene, pc-fos(mouse)-3.
Part of the reason for this
compltcated constructron was to construct the complete c-fos(mouse)
gene
from sequenced DNA fragments, thus the complete nucleottde sequence
of the c-fos gene In pc-fos(mouse)-3
IS known (Van Beveren et al., 1983).
Construction of chtmenc clones from these parental clones was carried out
usrng standard techntques
(Manratrs et al., 1982). and plasmrds were
propagated rn E colt strarn DH-1.
Transformation
Assay
208F rat cells (Ouade. 1979) were grown In Dulbecco-Vogt
modified Eagle
medtum supplemented wrth 10% fetal bovine serum. 208F cells were plated
at IO6 cells per 5 cm dash 1 day before transfectron.
The cells were
transfected wrth 0.1-l .O rg of construct DNA plus 10 fig of NIH/3T3 cell
DNA (carner DNA) as prevrously described (Graham and van der Eb. 1973;
Corsaro and Pearson, 1981). The next day the cells were trypstnized and
seeded Into three 5 cm dashes and fed every 3 days thereafter wrth medium
contarnrng 5% fetal bovrne serum plus 2 X 1O-6 M dexamethasone.
whrch
tends to enhance the contrast between transformed and nontransformed
208F cell morphology. Transformed
focr were vtstble after about 1 week
The plates were statned and foci counted after about 3 weeks.
Cotransfection
HPRT- 208F cells were plated at 5 X IO5 cells per 5 cm dash 1 day before
transfectton. Cells were transfected wrth 1 pg construct DNA plus 50 ng
p4aA8 HPRT expressron vector DNA (Jolly et al.. 1983b) and 10 pg NIH/
3T3 cell tamer DNA. The medrum was replaced 1 day after transfectton,
and 2 days after transfectton HPRT+ cells were selected In HAT (30 PM
hypoxanthtne. 1 MM amethoptenn. and 20 NM thymrdrne) medrum. Colonres
were Isolated about 1 week later ustng cloning rungs.
Nucleic Acid Analysis
Purtfrcatron and analysis of cellular DNA by Southern transfer (Southern,
1975) and hybndrzatron has been prevrously described. Cellular RNA was
purrfred by guanrdrne-throcyanate
extractron (Chtrgwrn et al., 1979). and
analyzed by electrophoresrs
In agarose/formaldehyde
gels, transfer to
nrtrocellulose, and hybrtdrzatron to a radroactrve probe (Thomas, 1980;
Curran et al.. 1983). The fos-specific probe conststed of a n&translated
(Rtgby et al., 1977) 1 .O kb Pst I DNA fragment from pfos-I (Curran et al.,
1982).
Acknowledgments
A. D. M. IS a postdoctoral fellow of the Leukemta Socrety of America; T. C.
IS supported by a Damon Runyon-Walter
Wrnchell Cancer Fund fellowshtp
(DRG 551). We thank Ltza Zokas and Joanne Tremblay for technrca.
assistance, and Sandy Haight for her artwork. This work was supported by
Nattonal lnstttutes of Health and Amencan Cancer Society grants to I. M. V.
The costs of publrcation of this article were defrayed In part by the
payment of page charges Thus artrcle must therefore be hereby marked
“advertisement”
rn accordance
with 18 U.S.C. Section 1734 solely to
indicate thus fact
Recerved September
12, 1983: revised
October
20. 1983
References
Banerjr, J.. Olson, L., and Schaffner, W. (1983). A lymphocyte-speciftc
cellular enhancer is located downstream of the Joining region in rmmunoglobulin heavy charn genes. Cell 33, 729-740
Blatr, D. G., Oskarsson.
T. G.. Wood, W. L., McClements, P. J., Ftschinger,
G.. and Vande Woude, G. (1981). Acttvatron of the transforming potential
of a normal cell sequence: a molecular model for oncogenesrs.
Sctence
212,941~943.
Chang, E. H.. Furth, M. E.. Scolnrck, E. M., and Lowy, D. R. (1982).
Tumorigenic transformation of mammalian cells induced by a normal human
gene homologous to the oncogene of Harvey murine sarcoma virus. Nature
297,479-483
Chrrgwrn, J. M., Przybyla. A. E.. MacDonald, R. J., and Rutter, W J. (1979).
lsolatton of brologcally active ribonuclerc acrd from sources enriched in
ribonuclease. Biochemrstry 18, 5294-5299
Corsaro, C. M.. and Pearson, M. L. (1981). Enhancing the effrcrency
DNA-mediated gene transfer in mammalran cells. Somatic Cell Genet.
603-616.
of
7,
Curran. T., Peters, G.. Van Beveren. C., Tetch, N. M., and Verma, I. M.
(1982). FBJ murtne osteosarcoma vrrus: identificatton and molecular clonrng
of btologlcally active provlral DNA. J. Vrrol. 44, 674-682.
Curran. T., MacConnell, W. P.. van Straaten, F., and Verma, I. M. (1983).
Structure of the FBJ munne osteosarcoma
virus genome: molecular cloning
of Its associated helper vrrus and the cellular homolog of the v-fos gene
from mouse and human cells. Mol. Cell. Brol. 3, 914-921,
Curran, T, and Tech, N. M. (1982a). ldenttfrcatron of a 39,000.dalton
protean In cells transformed by the FBJ munne osteosarcoma
virus. Virology
116, 221-235.
Curran. T., and Tetch, N. M (1982b). Candidate
osteosarcoma
virus oncogene: characterization
phoprotern. J. Vlrol. 42, 114-122.
product of the FBJ murine
of a 55,000.dalton phos-
Frnkel, M. P., Btskrs, B. O., and Jinktns, P. B. (1966).
osteosarcomas
rn mace. Scrence 157, 698-701.
Virus inductton
of
Fuhrman, S , Van Beveren, C., and Verma. I. M. (1981). ldentrfication of an
RNA polymerase II rnrtratron site in the long termrnal repeat of Moloney
murrne leukemia vrral DNA. Proc. Nat. Acad. Sci. USA 78, 541 l-5415
Grllres, S. D., Mornson. S. L., 01, V. T.. and Tonegawa, S. (1983). A trssuespectfic transcrlptron enhancer element IS located in the major rntron of
rearranged rmmunoglobulrn heavy charn gene. Cell 33, 717-728.
Graham. R., and van der Eb, A. (1973). A new techntque for the assay of
rnfecttvrty of human adenovtrus 5 DNA. Vrrology 52, 456467.
Jolly, D. J., Esty, A C., Subramanr, S., Fnedmann. T., and Verma, I. M.
(1983a) Elements In the long termtnal repeat of murine retrovrruses enhance
stable transformatron by thymrdine kinase gene. Nucl. Acids Res 7 1. 18551872.
Jolly, D. J.. Okayama, H., Berg, P., Esty A. C.. Filpula. D , Bohlen, P.,
Johnson, G. G., Shrvely, J. E Hunkapillar, T.. and Friedmann, T. (1983b).
lsolatron and charactertzatton
of a full-length expressible cDNA for human
hypoxanthtne phosphoribosyltransferase.
Proc. Nat Acad. SCI. USA 80.
477-481.
Levy J A.. Hartley, J. W., Rowe, W. P., and Huebner. R. J. (1973). Studtes
of FBJ osteosarcoma
virus In tissue culture. I. Brologtcal charactensttcs
of
the ‘C-type vrruses. J. Nat. Cancer Inst. 57. 529-539.
LUCIW. P. A., Brshop, J. M.. Varmus, H. E., and Capecchr. M. R (1983).
Locatton and functron of retroviral and SV40 sequences
that enhance
brochemrcal transformation after micrornfectron of DNA. Cell 33, 705-716
Manta& I, Fritsch, E. F., and Sambrook,
Laboratory Manual. (Cold Spring Harbor,
Laboratory).
J. (1982) Molecular Clonrng: A
New York. Cold Spring Harbor
Miller, A. D , Jolly, D J.. Frtedmann, T., and Verma, I. M. (1983). A
transmrssrble
retrovlrus
expressing
human phosphoribosyltransferase
(HPRT). gene transfer into cells obtatned from humans deftcrent in HPRT.
Proc. Nat. Acad Sci. USA 80, 4709-4713.
Muller, R.. Slamon, D. J., Tremblay, J. M., Clone, M. J., and Verma, I M.
(1982). Drfferentral expressron of cellular oncogenes durtng pre- and postnatal development of the mouse Nature 299, 640-644
Cell
60
Muller. R., Verma, I. M., and Adamson, E. D. (1983). Expression of c-one
genes: c-fos transcripts accumulate to high levels during development of
mouse placenta, yolk sac and amnron. EMBO J. 2, 679-684.
Okayama,
expressron
H., and Berg. P. (1983). A cDNA clonrng vector that permits
of cDNA Inserts in mammalian cells. Mol. Cell Biol. 3, 280-289.
Oskarsson,
M.. McClements, W., Blair, D. G.. Mabel, J. V., and Vande
Woude, G. (1980). Properties of a normal mouse cell DNA sequence (sarc)
homologous to the src sequence of Moloney sarcoma virus. Science 207,
1222-1224.
Quade, K. (1979). Transformatron of mammalran cells by avian myelocytomatosrs virus and avian erythroblastosis
virus. Virology 98, 461-465.
Queen, C., and Balkmore, D. (1983). lmmunoglobulrn gene transcnptron
activated by downstream sequence elements. Cell 33, 741-748.
is
Rrgby, P. W.. Dteckmann. M., Rhodes, C., and Berg, P. (1977). Labeling
deoxynbonuclerc
acrd to hrgh specific activity in vitro by neck translation
wrth DNA polymerase I, J. Mol. Brol. 7 13, 237-251.
Southern,
fragments
E. M. (1975). Detection of specific sequences
among DNA
separated by gel electrophoresis.
J. Mol. 8101. 98, 503-517
Thomas, P S. (1980). Hybridizatton
fragments transferred to nrtrocellulose.
5205.
of denatured RNA and small DNA
Proc. Nat. Acad. Sc USA 77, 5201-
Vaidya, A. B., Taraschi, N. E.. Tancrn, S L, and Long, C. A.(1983).
Regulatron of endogenous
murrne mammary tumor virus expression in
C57BL mouse lactatrng mammary glands: transcrrption of functional mRNA
with a block at the translation level. J. Vrrol. 46, 818-828.
Van Beveren, C., van Straaten, F., Curran, T., Mtiller, R., and Verma, I. M.
(1983). Analysrs of FBJ-MuSV provrrus and c-fos (mouse) gene reveals that
vrral and cellular fos gene products have different carboxy termrnr Cell 32,
1241-1255
van Straaten. F., Muller. R.. Curran, T., Van Beveren. C., and Verma. I. M.
(1983). Complete nucleotide sequence of a human c-one gene: deduced
ammo acid sequence of the human c-fos gene protern. Proc. Nat. Acad.
SCI. USA 80. 3183-3187.
Weiss, R. A, Terch, N. M., Varmus, H. E and Coffin, J. M (1982). RNA
Tumor Vrruses. (Cold Sprrng Harbor, New York: Cold Spring Harbor Laboratory).
Wrgler, M.. Sweet, R., Sm. G. K , Wold, B., Pellrcer, A , Lacy, E., Manratrs.T ,
SlIverstern. S., and Axel, R.(1979). Transformation of mammalran cells wrth
genes from procaryotes and eucaryotes. Cell 16, 777-785.
Note Added
in Proof
The work referred to throughout the text as T. Curran, A D. Mrller, L. Zokas,
and I. M. Verma, manuscrrpt submrtted, IS in press (Cell 36, Number 2,
1984).