Download Viruses, Genes and Cancer1 One person in every four in the United

AMER. ZOOL., 29:653-666 (1989)
Viruses, Genes and Cancer1
J. MICHAEL BISHOP
Department of Biochemistry and Biophysics, University of California,
San Francisco, California 94143
SYNOPSIS. Cancer takes many forms and has many causes. But it is possible to unite these
many forms and causes with a single hypothesis: that cancer may be a malady of genes,
that abnormalities of genes usually lie at the heart of the disease. Recent research has
uncovered evidence that this hypothesis may be correct. Many human tumors contain
genetic damage that can account for cancerous growth. The damage affects genes that
are normally vital to normal growth and development, but that have run amok in cancer
cells. The prevention and treatment of cancer has until now been based on trial and error.
The identification and characterization of damaged genes in human tumors points the
way to new and more rational strategies for the diagnosis, prognosis and therapy of cancer.
INTRODUCTION
they have elaborate internal structure that
One person in every four in the United allows them to live and breathe; they move
States will develop cancer, one in every five from one place to another with purpose;
will die of the disease. These are tragic they have distinctive personalities and
dimensions, but they are no larger that the assignments; they converse by means of
intellectual challenge cancer presents. chemical and molecular languages. The
Every minute, ten million cells divide in greatest wonder of cells, though, is that
our bodies. The divisions usually occur in each knows what it is to do, and when and
the right place, at the right time, governed where. Cancer is a failure of this wonder;
by mechanisms for which we usually have the cancer cell violates its social contract
names but often no explanations. When with other cells, proliferating and spreadthe governance fails, cancer may arise. Why ing in an unfettered way.
does the governance fail? How does it fail?
The instructions that dictate the strucWhat hope do we have of penetrating the ture and activities of our cells are inscribed
complexities of cancer cells, in which the on DNA. We call the vocabulary of those
changes of structure and function seem too instructions genes. Genes are interpreted
numerous to count?
by a familiar chemical pathway that first
These questions have been before the copies DNA into smaller molecules known
human mind for a very long time, and until as RNA, and then RNA into even smaller
now, the answers had seemed distant molecules known as proteins (Strickberindeed. But over the past decade, a great ger, 1986). Proteins are the handmaidens
change has occurred in how we think of of genes: the molecules that get most of
cancer. Where once we sought only the the jobs done, the pieces from which much
innumerable factors in our lives that might of the cell is built, the engines that drive
cause cancer, now we seek with equal dil- the chemical reactions of life. It now
igence a single unifying explanation of how appears that cancer results from mistakes
those causes might work. The search began in this chain of command, mistakes that
originate from damage to DNA.
with cells, but it has led us to genes.
Biologists have long nurtured the belief
CANCER IS A MALADY OF GENES
that cancer is at its heart a malady of genes
Our bodies are built with bricks called (Bishop, 1987). The belief grew from
cells. But these are not ordinary bricks: diverse roots: the recognition of hereditary
predispositions to cancer; the presence of
damaged chromosomes in cancer cells; the
1
From the Symposium on Science as a Way of Know- connection between susceptibility to caning—Cell and Molecular Biology presented at the Annual
Meeting of the American Society of Zoologists, 27- cer and impaired ability to repair damaged
DNA—individuals who inherit such
30 December 1988, at San Francisco, California.
653
654
J. MICHAEL BISHOP
impairments also inherit a predisposition
to cancer; and the evidence that relates the
mutagenic potential of substances to their
carcinogenicity—the ability to damage
DNA seems often to subsume the ability
to cause cancer. Now these roots have been
conjoined by demonstrations that cancer
cells harbor both dominant and recessive
genetic damage, arising from point mutations and large rearrangements of DNA
alike, distorting both the expression and
biochemical function of genes. In this essay,
I will review how the damage has been
found, the nature of the genes that it affects,
and the ways that it may figure in the genesis of human cancer.
SIMPLIFICATIONS IN CANCER RESEARCH
The genetic mistakes that lead to cancer
occur in single cells. Most cancers begin as
single cells, but grow into colonies composed of innumerable cells, all derived from
the original founder (Nowell, 1986). The
early steps in the genesis of cancer probably occur in many of our cells during a
lifetime. But only rarely does the course of
events continue to its lethal end—a homogeneous colony of cancer cells with the
potential to expand its size without surcease. Thus, most of us will not develop
more than one cancer in our lifetime; and
the cancer usually will have originated from
a single wayward cell, whose progeny have
successfully eluded the many defenses that
our body can mount against both errant
natives and foreign intruders.
How are we to find the crucial errors
within cancer cells, the damaged genes that
set the cells on the route to cancerous
growth? To answer this question, we have
turned to two simplifications. The first is
to view cancer as a disease of individual
cells and to study this disease not in the
body of an animal but in a petri dish. The
second is to exploit agents that will rapidly
and reproducibly convert cells to cancerous growth under experimental circumstances; prominent among these agents are
viruses that cause tumors in animals. Let
us look at our simplifications in greater
detail.
You doubtless think of cancer as a tumor,
a relentlessly expanding mass of tissue. But
whole tumors are not easy objects for
experimental study. So we resort to the
belief that the properties of individual cancer cells probably explain the behavior of
tumors. We can define these properties by
growing the cancer cells outside of the animal, using an artificial mixture of nutrients
to feed the cells, and glass or plastic vessels
to contain the cells. In this setting of "tissue
culture," cancer cells misbehave exactly as
we might have expected from the properties of whole tumors (Hanafusa, 1977):
they continue to grow when they should
not—when they have become crowded by
their neighboring cells; they develop a different appearance from their normal counterparts; and they retain many other properties we attribute to cancer cells. In sum,
they remain convincing caricatures of the
cells in an invasive cancer.
But to describe a cancer cell is not to
understand it. To understand how cancer
arises, we need to track the events that
occur from the moment a cell is first set
on the path to cancerous growth. We cannot do this with human cancer—the process is too complex. But we can do this by
using viruses that cause tumors in animals
and convert cells to cancerous growth in
the test tube.
We have known for more than fifty years
that viruses can cause cancer in birds and
animals. From this knowledge, there sprang
two schools of thought. One school argued
that we should search for viruses in human
cancer, that viruses must cause the disease.
The other school held that since there may
be many causes of cancer (even celery has
made the lists), we would be better off to
seek the central molecular mechanisms by
which the disease arises: tumor viruses
should be used to ferret out the genetic
and chemical processes that cause the cancer cell to run amok. Now both views
appear to have been vindicated.
Viruses have been found in some human
cancers and it seems likely that they contribute to the genesis of the disease (Table
1). The list of current suspects resembles
a police line-up because guilt is often more
in the eye of the beholder than in the evidence. The evidence is most persuasive for
hepatitis B virus (Beasley, 1988), which is
VIRUSES, GENES AND CANCER
655
TABLE 1. Candidate human tumor viruses.
Virus
A. DNA genomes
Hepatitis B virus (HBV)
Human papilloma viruses (HPV)
Epstein-Barr virus (EBV)
Herpes simplex viruses (HSV)
B. RNA genomes
Human T-cell leukemia virus I (HTLV-I)
a principal cause of liver cancer—probably
the most prevalent cancer on the globe,
although relatively uncommon in the
U.S.A. owing to limited frequency of infection with HBV. A vaccine against HBV is
now being widely deployed around the
world. The vaccine is likely to prevent not
only Hepatitis B, but most cancers of the
liver as well. This will represent the fulfillment of a much-cherished dream:
immunisation against a human cancer.
But the revelations I seek to explain do
not concern viruses as causes of cancer.
Instead, I tell of how tumor viruses have
been used as tools to lay bare the secrets
of the cancer cell. The utility of viruses as
experimental reagents rests on a genetic
simplication. The DNA of human cells is
large enough to accommodate one million
genes. For reasons that are not yet entirely
clear, we do not have one million genes;
but we do have tens of thousands, most of
them not yet even identified. (Recall that
ignorance when next you hear the irresponsible claim that we are on the brink of
creating life in a test tube and of altering
the human species for all time; we do not
even have names for most of what we would
have to alter.) Each gene has its own specific chore, and among these chores, there
must be many that are important in the
genesis of cancer. By contrast, viruses generally have less than a dozen genes, and
often only one of these genes is required
to produce cancer. So in the extreme,
viruses can simplify the study of cancer by
more than a thousand fold.
Over the recent decades, scientists have
exploited the simplifications offered by
viruses to find a set of human genes whose
Liver carcinoma
Carcinoma of cervix, vulva, penis and skin
Warts
Burkitt's Lymphoma
Carcinomas of urogenital tract (?)
Acute T-cell leukemia
activities may lie at the heart of many cancers, no matter what their causes. We view
these genes as the keyboard on which many
different carcinogens might play. An enemy
has been found, and we have begun to
understand its lines of attack.
RETROVIRUSES AND CANCER GENES
We owe much of our story to the retroviruses, so called because their life cycle
reverses the normal flow of genetic information (Fig. 1) (Temin, 1972). As in many
other viruses, the genes of retroviruses are
carried by RNA rather than DNA. But
retroviruses are unique because their RNA
genes are copied into DNA by an enzyme
called reverse transcriptase during viral
growth. The newly made viral DNA is then
inserted into the chromosomal DNA of the
host cell, and the cell in turn uses its own
machinery to express viral genes.
The discovery of reverse transcriptase
was a momentous event in the history of
biology: it revised our view of the directions in which genetic information can flow;
it provided a vital tool for the technology
of recombinant DNA; and it revealed a
new means by which our genetic dowry
retains is plasticity. The genesis of that
plasticity merits a digression.
We have known for some time that genes
can move from one place to another within
the cell, and we have learned that our own
genetic dowry has been incessantly
reshaped by the shuffling of genes and
other pieces of DNA (Fig. 2). At first, we
thought that genes moved only in the form
of DNA. Now we know that the vehicle
can also be RNA, copied back into DNA
by reverse transcriptase (Weiner et al.,
656
J. MICHAEL BISHOP
5' 5'
ALTERNATE MODES FOR
GENETIC TRANSPOSITION
DNA
DNA
FIG. 2. Transposition of genes within living cells.
Genes can transpose either directly from one position
in DNA to another, or indirectly by means of reverse
transcription of RNA and subsequent integration.
Replication of RNA can also occur directly, mediated
by protein enzymes of autocatalysis. Replication of
RNA is now prominent only as part of certain viral
life cycles, but it may have been of more general
importance in earlier versions of the biosphere.
Translation
1 Cleavage
FIG. 1. The molecular life cycle of retroviruses. The
incoming genome of the virus is a diploid singlestranded RNA, which is transcribed into doublestranded DNA by reverse transcriptase. The capital
letters denote viral genes; the lower-case t, a terminal
redundancy in viral DNA; and the remaining lowercase letters, viral gene products.
1986). The DNA of our chromosomes is
littered with genes that have been moved
into new positions by this process; the
movement may represent a powerful
engine of evolution; and it is connected to
another recent and astonishing discovery—that RNA can reduplicate itself, at
least sluggishly, in an epicycle that some
observers believe is an artifact of the very
beginning of life (Alberts, 1986). RNA may
have preceded DNA in the primieval soup,
and if so, reverse transcriptase may have
been midwife to the DNA version of the
biosphere. Much of this remains speculative, but it serves to dramatize how the
study of a creature so simple as a virus can
change our view of life itself.
The lesson extends to cancer. The study
of retroviruses has led us to some of the
genes that must figure in the genesis and
growth of cancer cells. It has done so in
two ways.
First, the integration of retroviral DNA
is potentially mutagenic (Varmus, 1983): it
can damage important cellular genes, and
it can influence their expression by bringing them under the sway of powerful viral
signals. We call this "insertional mutagenesis," and we will see later that it may indeed
cause tumors.
Second, some retroviruses carry genes
which can themselves be carcinogenic:
expression of any of these single genes is
sufficient to give rise to cancerous growth
(Bishop, 1982). We call these genes "oncogenes" for obvious reasons, and their value
to us is beyond measure: they are cancer
genes incarnate; they are the keys with
which we hope to unlock the closet in which
cancer has sheltered its ugly secrets.
THE ONCOGENE SRC
Although he could not have known it at
the time, Peyton Rous gave us the first
oncogene when, in 1910, he isolated a
tumor virus from a chicken (Pitot, 1983).
The virus discovered by Rous was the first
VIRUSES, GENES AND CANCER
of its kind by many years, the first cancer
virus ever glimpsed. For this remarkable
discovery, Rous was at first criticized and
disparaged; his finding was simply beyond
the ken of most scientists of the time. He
eventually despaired of convincing anyone
that he was right, gave up the pursuit of
his historical discovery, and entered
another field of research. Fortunately, Rous
had good genes: he lived long enough to
receive the Nobel Prize after he had passed
the age of 80; the rest of the scientific world
had finally caught up with him.
We now know that the virus Peyton Rous
discovered is a retrovirus, and that it has
but four genes. Three of these genes are
used to reproduce the virus, the fourth is
an oncogene that causes cancer. We call
this oncogene src because it induces tumors
of the connective tissue known as sarcomas.
We first learned of src's existence from
experiments with genetics (Bishop, 1982).
It is possible to damage src in such a way
that the action of the gene becomes vulnerable to heat; at one temperature (usually 35°C), the gene is active; at a higher
temperature (usually 39°-41°C), the gene
is inactive. If cells are affected by a temperature-sensitive src at the lower temperature, they convert to cancerous growth
within 12 hr. By merely shifting the temperature to a higher level, the cells can be
"cured"—returned to normal growth
because src has been inactivated. These
events are little short of the miraculous:
they occur so quickly that they can be followed in real time by microscopy. In truth,
the cells at the higher temperature are only
in remission; as soon as the temperature is
lowered, the activity of src returns and the
cells renew cancerous growth. From these
experiments, we learn that at least one viral
gene is responsible for cancerous growth;
that the gene must be continuously active
if cancerous growth is to persist; and that
the gene probably works by directing the
synthesis of a protein.
As an aside, I note that when I describe
these marvelous events to general audiences, I am often asked whether the properties of temperature-sensitive mutants
underlie the use of heat to treat cancer—
657
a dubious therapy in most settings, but one
that seems to have an endless attraction for
the public eye. I then attempt to explain
that the temperature-sensitive mutant is an
artifice, an invention whose properties
nevertheless tell us something of nature's
real ways. I often fail. Our culture is
blighted by the fact that the general public
is so innocent of how science proceeds.
The wonders of recombinant DNA have
made src more palpable (Bishop, 1982). The
gene has been isolated as a tiny piece of
DNA bearing only src. When introduced
into cells and expressed, src in this form
can transform cells to neoplastic growth.
When injected into chickens, the DNA
elicits malignant tumors. Here is graphic
evidence that the action of src alone is sufficient to cause cancer, and here is eloquent
testimony to the experimental value of
oncogenes. We need identify only a single
gene product, a single protein, a single biochemical activity, to get a view of how a
cancer cell might arise. We have that view
for src, and it is a vista full of promise and
puzzles.
THE SRC ONCOGENE ENCODES A MEMBRANE
PROTEIN THAT CATALYZES PROTEIN
PHOSPHORYLATION
We first of all know where in the cell the
src protein attacks (Bishop, 1982). Diverse
experimental strategies have located the
protein at the inner surface of the membrane that encloses the cell. For the
moment, at least, it remains a mystery how
chemical reactions at the surface of the cell
can control the behavior of the cell, and
we are furthermore ignorant of how these
events might lead to cancerous growth. But
src has given us an invaluable clue, because
the protein that src encodes is an enzyme,
a molecule that catalyzes a chemical reaction.
The reaction that src catalyzes is seemingly simple: the transfer of phosphate ions
to other proteins, a process known as "protein phosphorylation" (Hunter, 1984). The
src protein catalyzes the phosphorylation
of only one amino acid within proteins—
tyrosine. The phosphorylation of tyrosine
was unknown until src and other oncogenes
came under study, yet now we know that
658
J- M I C H A E L
BISHOP
PHOSPHOTYROSINE
0
IIM
HO
TYROSINE
OH
|
P
OH
()
H
;^
H
H'
H
C
•N
C
H
I
C
I
I II
H
H
O
H
H
FIG. 3. The phosphorylation of tyrosine. Modification of tyrosine in the manner illustrated here plays a
diverse and important role in the regulation of how proteins perform their functions in living cells.
diverse forms of this enzymatic activity are
ubiquitous among vertebrate cells, and that
it is vital to the control of cellular phenotype.
The addition of phosphate to tyrosine
appears simple in a chemical formula (Fig.
3), but when this property of the src protein
was discovered, it sent the shiver of
recognition down the spines of biochemists
because we have come to know protein
phosphorylation as one of the chief means
by which the activities of cells are governed
(Hunter, 1987). The governance is
achieved because phosphorylation can
change the structure and function of proteins, turning them on or off by remolding
their form. Thus, by phosphorylating
numerous cellular proteins, the product of
src could rapidly change myriad aspects of
cell structure and function. Now we need
to know which cellular proteins are phosphorylated, which are the targets for the
deadly onslaught. Alas, we do not know
these proteins yet; they remain one of the
great enigmas of cancer research. But a
door has opened!
THE ONCOGENES OF RETROVIRUSES ARE
MANY AND DIVERSE
Many doors will follow because, as more
and more retroviruses have been isolated
and studied, more and more oncogenes
have come to view. We now know of more
than twenty different oncogenes in retroviruses (Varmus, 1984). Taken together,
these oncogenes can cause most of the
major forms of cancer that afflict human
beings. And these are not artifacts of the
laboratory: each virus has been isolated
from a naturally occurring tumor in an animal or bird.
Retroviral oncogenes act in diverse ways,
deploying their protein products to virtually all reaches of the cell—the nucleus,
the cytoplasm, the cell surface, even the
exterior of the cell (Hunter, 1984). We are
excited by this daunting variety. The
growth of cells is controlled by an elaborate circuitry that extends from the surface
of the cell to its deepest interior (Fig. 4).
Each oncogene product touches the circuitry at one of its junction boxes, so each
VIRUSES, GENES AND CANCER
659
Cellular Phenotype: A Regulatory Circuit
FIG. 4. The biochemical circuitry that regulates the phenotype of vertebrate cells. The diagram illustrates
how some of the identified proto-oncogenes fit into the regulatory circuitry of vertebrate cells. Most of the
abbreviations represent acronyms for individual genes. Other terms include: Ptd I, phosphotidyl inositol;
PKC, protein kinase C; S6, phosphorylation of the S6 protein of ribosomes; p-ser, phosphoserine; p-tyr,
phosphotyrosine; R, generic receptor; G, GTP-binding proteins.
will probably tell us something of how the
circuitry works. One of the cardinal tenets
of medical science is that we learn about
normal things by studying the abnormal.
Here we see that tenet exemplified.
To understand the circuitry, we must
learn the biochemical mechanisms that
operate at the junction boxes. In outline,
here is what we now know or suspect.
Roughly half of the known oncogenes act
through the phosphorylation of proteins,
either by inducing phosphorylation or,
more commonly, by catalysing it (Hunter,
1984). A smaller gene family encodes analogues of the G-proteins that transduce signals to effectors such as adenyl cyclase,
although the effector targets for the products of these oncogenes remain a mystery
(Barbacid, 1987). A third group of oncogene products may serve as regulators of
transcription; the case is particularly strong
for the oncogenes/oj and jun (Curran and
Franza, 1988). And there are hints that
some oncogenes play on DNA replication,
although these remain provisional and
controversial.
THE DISCOVERY OF PROTO-ONCOGENES
At first it seemed only a dim hope that
the lessons learned from retroviral oncogenes could prefigure the abnormalities
that engender human cancer. Then the
unexpected happened, as it does so often
in science.
In the early 1970s, Robert Huebner and
George Todaro proposed that all cells contain the seeds of their own destruction in
the form of retroviruses whose oncogenes
could be activated by carcinogens of many
different forms (Todaro and Huebner,
1972). Huebner and Todaro called this
theory the "oncogene hypothesis" (they in
fact invented the term "oncogene" to
advertise their proposal). The oncogene
hypothesis prompted a search for the src
gene in the DNA of normal cells (Bishop,
1982). The search was a wild grasp for at
least two reasons: first, few observers
believed the oncogene hypothesis—
although disbelief is never a reason to avoid
experiment; and second, there was no reason to expect that src, an oncogene of a
660
J. MICHAEL BISHOP
chicken virus, would represent one of the
all-important intrinsic oncogenes postulated by Huebner and Todaro. But the
experiment was performed, since src was
the only oncogene then in hand; and against
all odds, src was discovered in the DNA of
normal birds and mammals. Moreover, it
quickly became clear that the findings had
turned the oncogene hypothesis on its head:
the src in normal DNA is a cellular gene,
not a viral gene, and it is from the cell that
the virus of Peyton Rous first obtained its
oncogene (Bishop, 1982). The virus is a
pirate; the booty is a cellular gene with the
potential to become a cancer gene.
We now know that almost all of the
numerous retroviral oncogenes are but
wayward copies of cellular genes that we
call "proto-oncogenes" (Bishop, 1983). We
can find each of the proto-oncogenes in
many difFerent species—mammals, birds,
insects, arrayed across one thousand million years of evolution. What accounts for
this remarkable success story? Why have
proto-oncogenes survived the hazards of
evolution over such great lengths of time?
We presume that they serve vital functions
for the creatures in which they are found.
Clues to those functions have come from
several quarters.
First, serendipity led us to the realization
that a still growing number of proto-oncogenes encode either growth factors or their
receptors on the cell surface (Helkin and
Westermark, 1984). Among these have
been the growth factor PDGF and the
receptors for thyroid hormone, EGF, and
the hemopoietic growth factor CSF-I.
Moreover, the study of proto-oncogenes
has fingered several factors and receptors
never glimpsed before and still in search
of purpose.
Second, a number of proto-oncogenes
encode proteins found in the nucleus.
Those that have been adequately characterized have proven to be factors that participate in the regulation of transcription
(Curran and Franza, 1988).
And third, mutant alleles have been
identified for the counterparts of four
proto-oncogenes in the fruitfly,Drosophila
melanogaster (Shilo, 1987). All of the muta-
tions elicit profound disturbances of development. In one instance (the abl protooncogene), the search for mutations was
deliberate; in the other instances, mutations recognized first for their effects on
development later proved to be in Drosophila counterparts of mammalian protooncogenes.
So it is now clear that the proteins
encoded by proto-oncogenes represent
junction boxes in the elaborate circuitry
that controls cellular phenotype (Fig. 4):
polypeptide hormones that act on the surface of the cell; receptors for these hormones; proteins that convey signals from
the receptors to the deeper recesses of the
cell; and nuclear functions that may
orchestrate the genetic response to afferent commands. We can view the action of
oncogenes as "short circuits" at the junction boxes and, thus, imagine how the sustained or augmented influence of an otherwise normal gene product might trigger
mayhem.
RETROVIRAL ONCOGENES EXEMPLIFY
GENETIC DAMAGE T H A T
CAN BE TUMORIGENIC
Whatever the functions of proto-oncogenes might be, we presume that evolution
did not install these genes to cause cancer.
Why then does their transfer into retroviruses give rise to oncogenes? The answer
lies in the elaborate molecular gymnastics
by which proto-oncogenes are copied into
the genomes of retroviruses (Bishop, 1983).
During the copying, proto-oncogenes suffer damage that can convert them to oncogenes—from Dr. Jekyll to Mr. Hyde. We
also imagine that any other influence that
can damage a proto-oncogene might give
rise to an oncogene, even if the damage
occurred without the gene ever leaving the
cell, without the gene ever confronting a
virus.
Piracy of proto-oncogenes by retroviruses is an accident of nature, serving no
purpose for the virus. But the event has
profound implications for cancer research.
In an act of benevolence, retroviruses have
brought to view cellular genes whose activities may be vital to many forms of carci-
661
VIRUSES, GENES AND CANCER
B-Cell Lymphomas in Chickens:
Induction of a Proto-Oncogene by Insertional Mutagenesis
Proto-Oncogene
(c-myc)
Cell
Cell
DNA
*
4
4
Viral Inserts
DNA
Viral Inserts
RNA
\
p58c-my_c
FIG. 5. Integration of retroviral DNA in the vicinity of the proto-oncogene myc. In chicken lymphomas
induced by retroviruses, integration of viral DNA in the vicinity of the proto-oncogene myc (c-myc) activates
transcription from the gene and, thus, causes sustained production of the gene product. These events are
believed to be the first step in the genesis of the lymphomas.
nogenesis. It might have required many
decades more to find these genes by other
means amongst the morass of the human
DNA; instead, we have the genes made
manifest in retroviruses, excerpted from
amid the morass and made available for
our closest scrutiny.
RETROVIRUSES CAN INITIATE
TUMORIGENESIS BY ACTIVATING
PROTO-ONCOGENES
There was good fortune in these observations: the fact that the mutated gene was
already known to us as a proto-oncogene
added logical force to the discovery; myc is
a gene whose tumorigenic potential had
already been demonstrated by transduction into retroviruses.
The number of viruses that apparently
cause tumors in this manner is large, and
some of these viruses have led us to new
proto-oncogenes. Imagine that the cellular
gene in this scheme was not previously
known to us. Because we can track the viral
DNA to its residence in the DNA of the
tumor cell, we can find and isolate the harried cellular gene. More than two dozen
new proto-oncogenes have been discovered in this way, genes not glimpsed before
by any means (Varmus, 1984). Each of these
genes has been implicated in the genesis of
particular kinds of tumors induced by
retroviruses, each is likely to tell us something new about how normal and cancer
cells conduct their daily affairs.
Transduction of proto-oncogenes into
retroviruses has taught us many lessons.
But the possibility that proto-oncogenes
might be the targets for carcinogens of various sorts remained speculative until the
unexpected revelations of insertional
mutagenesis came into view.
Some retroviruses do not have oncogenes, but they can nevertheless cause cancer. They do so by exploiting proto-oncogenes (Varmus, 1983). The first example
of this emerged from the study of chicken
lymphomas in which the previously recognized proto-oncogene myc has been actiABNORMALITIES OF CHROMOSOMES
vated by insertions of retroviral DNA
CAN
ACTIVATE PROTO-ONCOGENES
upstream, within, or even downstream of
Fruitful though the study of retroviruses
the gene (Fig. 5). The activation of transcription from myc is generally viewed as has been, we still know of no instance where
the first of several steps in tumorigenesis. a retrovirus induces human tumors through
662
J. MICHAEL BISHOP
t (9;22) Chronic Myeloid Leukemia
22
9q+
22q"
Ph1
bcr/abl
FIG. 6. The translocation that creates the Philadelphia chromosome. In the cells of human chronic myelogenous leukemia, a reciprocal translocation between chromosomes 9 and 22 changes the structure of the
proto-oncogene abl and the activity of the gene product. A gene known as bcr is the reciprocal partner in the
translocation. The abnormal chromosome 22 that results (22q~) is known by cytogeneticists as the Philadelphia
chromosome (Ph'), in recognition of the city where the abnormality was discovered.
the agency of either a transduced oncogene or insertional mutagenesis. We turn
therefore to genetic damage found directly
in human cancer cells to give our story
immediacy.
Four kinds of damage have been encountered: translocations between or within
chromosomes; deletions affecting discrete
portions of chromosomes; abnormal
amplification of large domains within chromosomes; and point mutations within
genes. Translocations, amplification, and
point mutations have typically affected
proto-oncogenes of the conventional sort,
whereas deletions characteristically signal
the existence of a different sort of genetic
element also involved in tumorigenesis. We
will look first at damage affecting protooncogenes.
In one form of damage, portions of two
separate chromosomes break away and
exchange places (Fig. 6). The exchange is
called translocation; it has now been found
in a wide variety of human cancers. In some
examples, previously recognized protooncogenes lie near the points of breakage
in the chromosomes; in others, molecular
dissection in the vicinity of the breaks has
uncovered novel proto-oncogenes. The
harvest from these enquiries has been
nearly a dozen proto-oncogenes, each a
potential cancer gene (Varmus, 1984).
Translocations can affect the functions
of proto-oncogenes in either of two ways
(Bishop, 1987). First, translocations move
proto-oncogenes from one place to another
and thus change the molecular context of
the genes. When this happens, the genes
are unleashed from their usual controls.
We presume (but cannot yet prove) that
the unrestrained activity of the protooncogene is an ingredient of cancerous
growth. Second, translocations can fuse
portions of two genes, engendering an
abnormal protein whose functional activity
may be greatly augmented and/or altered
in specificity.
The second form of damage to chromosomes arises when a large region of
DNA is duplicated abnormally to give many
copies of itself, an overgrowth we refer
to as "gene amplification" (Fig. 7). As
amplification proceeds, two chromosomal
abnormalities can arise: independently
replicating units called double-minute
chromosomes; and homogeneously-staining regions within chromosomes. Doubleminute chromosomes and homogeneouslystaining regions alert the microscopist to
the presence of amplified genes in cancer
cells. By rooting about in the amplified
DNA of human cancer cells, molecular
biologists have uncovered both familiar and
novel proto-oncogenes (Varmus, 1984).
Amplification of a proto-oncogene provides the cell with the ability to make more
VIRUSES, GENES AND CANCER
663
Recombination
8
Amplification
FIG. 7. Amplification of DNA in vertebrate cells. Occasional amplification of DNA within vertebrate chromosomes gives rise to two karyotypic abnormalities, double-minute chromosomes (DMs) and homogeneouslystaining regions (HSR). The amplification is abnormal, but its mechanism is not known—the version shown
here is only one of several possibilities.
of the protein encoded by the gene than is
normal: each additional copy of the gene
increases the amount of protein that can
be manufactured. Once again, the pathogenic influence appears to be an ungovernable amount of an otherwise normal
gene function.
POINT MUTATIONS IN PROTO-ONCOGENES
Many cancer cells contain no visible
damage to chromosomes for us to explore.
So experimentalists have forced the issue
by showing that DNA taken from some
human tumors can elicit cancerous growth
of cells in tissue culture, and that the cells
can in turn grow into tumors when
implanted into experimental animals
(Weinberg, 1983). These findings indicate
that the DNA from the tumors contains a
full-blown oncogene of the sort we were
once accustomed to finding only in viruses.
When the responsible genes were isolated from tumor DNA, they proved to be
proto-oncogenes we had encountered
before in the study of retroviruses, from a
family of genes known generically as ras
(Weinberg, 1983). Moreover, the genes
were damaged: they contained mutations
that made single changes in the protein
encoded by the proto-oncogenes. As a
result, the genes now had the biological
properties of viral oncogenes, even though
they still resided in the DNA of tumor cells
and had nothing to do with viruses.
In the few years that have ensued since
this discovery, mutations have been found
in ras proto-oncogenes from many different kinds of human cancers, in perhaps
twenty percent of all the tumors examined.
These findings have greatly enhanced the
credibility of at least two major themes in
modern cancer research: first, that protooncogenes discovered originally by the
study of retroviruses are likely to play a
role in the genesis of human cancer; and
second, that mutations of the sort detected
by the famous Ames test are likely to be
important in the pathogenesis of cancer.
How do point mutations affect the function of ras genes? The proteins encoded by
these genes transmit signals within the cell
by means that remain obscure. All we know
is that the binding of GTP to the proteins
is essential to signal transmission, and that
the proteins normally limit their own activity by slowly hydrolysing GTP once it has
been bound (Barbacid, 1987). The point
mutations found in the ras genes of cancer
cells reduce the hydrolysis of GTP by the
gene products and, thus, allow the products to remain active for long periods of
time. Thus, the cell is subjected to sustained signalling that may lead to abnormal
proliferation.
664
J. MICHAEL BISHOP
RECESSIVE MUTATIONS IN
CANCER CELLS
In 1866, the French neurosurgeon and
anthropologist Paul Broca sketched the
pedigree of his wife's family and discerned
an hereditary predisposition to cancer. We
now know that Broca was correct, that predispositions to cancer can be inherited
(Ponder, 1980). In most instances, we do
not understand the mechanisms of this
inheritance, and in particular, we have yet
to implicate proto-oncogenes. But in some
few cases, we have the beginnings of an
explanation that has added new dimensions to the genetics of cancer. The explanation goes as follows.
We are diploid organisms: all of our cells
possess two copies of most of our genes.
The oncogenes we have discussed until now
are genetically dominant; their abnormalities are expressed even when a normal copy
of the same gene is also present in the cell.
But it now appears likely that most or all
human tumors also bear genetic lesions that
are recessive, that make their presence
known only when no normal counterpart
of the damaged gene is present in the cell.
One gene affected by recessive genetic
damage has been identified in human retinoblastomas and isolated by molecular
cloning, and numerous others have at least
been mapped to their approximate locations on human chromosomes (Klein,
1987). The search for recessive cancer
genes and the means by which they act
represents one of the newest frontiers in
cancer research. It will be a difficult frontier to tame.
Two sorts of genetic elements may
therefore figure in the genesis of cancer.
One is pathogenic only if it produces an
active protein, the other may play an etiological role when it is inactive or absent.
Damage to these two sorts of genes apparently combines in the genesis of human
cancer (Bishop, 1987).
ONCOGENES AT THE BEDSIDE
Both dominant and recessive damage to
genes has now been found in a wide variety
of human cancers (Bishop, 1987). The lists
impress because of the diversity of tumors
involved; because of their identities—several can be counted among the principal
nemeses of humankind; and because the
lists have been assembled after only a few
years of pursuit, with still imperfect tools.
There is doubtless more to come.
How might these findings aid in the management of cancer? It is too early to give
a decisive answer to this question, but there
are ways in which proto-oncogenes and
oncogenes may soon be used in the diagnosis and prognosis of cancer. I will illustrate three.
i) Neuroblastoma is the most common
cancer of children under the age of two,
although it also occurs in older youngsters.
The tumors are assigned to four stages,
according to severity: I and II, relatively
localized and usually curable disease; and
III and IV, more widespread and inevitably
fatal disease. A proto-oncogene called
N-myc is often amplified in neuroblastomas,
but only in those of stages III and IV—
and within those stages, in 30 to 50% of
all the tumors examined (Schwab, 1985).
Thus, it appears that amplification of N-myc
is a dire prognostic sign that can be used
in the counselling of patients and their
families, and in the choice of therapy.
ii) The disease myelodysplasia is a disorder of the human bone marrow that
begins as a benign ailment. After several
years, however, between 10 and 30% of
the affected individuals develop leukemia
and then rapidly die. Mutant ras genes have
been found in the blood cells of some individuals with myelodysplasia, especially in
those individuals who eventually develop
leukemia (Liu et al, 1987). These results
encourage the hypothesis that a search for
the damaged proto-oncogenes may provide a means to identify those cases of myelodysplasia that are likely to develop leukemia and thus require aggressive
therapeutic management.
iii) The gene affected by recessive damage in human retinoblastomas has been
identified and isolated by molecular cloning (Weinberg, 1988). Thus, it is now possible to chart the inheritance of damaged
copies of this gene in families afflicted with
VIRUSES, GENES AND CANCER
665
congenital retinoblastoma and, thus, to cancer growth; and the discovery of src in
identify children at risk for the disease normal cells, the first sighting of protooncogenes. Here is a familiar but ofteither before or after birth.
What of treatment? Are we likely to neglected lesson: the proper conduct of sciacquire new antidotes for cancer from our ence lies in the pursuit of nature's puzzles,
study of oncogenes? There is little likeli- wherever they may lead. We cannot prehood that we will be able to repair or judge the utility of any scholarship, we can
replace damaged proto-oncogenes in the only ask that it be sound. We cannot always
foreseeable future: we have not yet learned assault the great problems of biology at
how to operate with surgical precision and will; we must remain alert to nature's clues
efficiency on the DNA of living human cells and seize on them whenever and wherever
(Hogan and Lyons, 1988). But if we focus they may appear—even if it be in a chicken.
on the protein handmaidens of genes,
Bit by bit, the inner workings of the caninstead, we can see more cause for hope. cer cell are coming under our sway. With
Given sufficient information about how this knowledge, we seek devices for diagthese proteins act, the pharmaceutical nosis, for prognosis, and for rational designs
chemist (or perhaps the immunotherapist) of therapy and prevention of cancer. But
can probably invent ways to interdict their a greater intellectual adventure overshadaction, even to exploit the specificity of ows even these: the quest to understand
genetic damage and thus to reverse the the cancer cell in all of its particulars, to
effects of oncogenes. We are not close to know what keeps us whole and what renimplementing this strategy, but it is a rea- ders us asunder. The French mathematisonable hope.
cian and physicist, Henri Poincare, phrased
the point well:
CONCLUSION
"The scientist does not study nature
Cancer may be caused by many things, because it is useful; he studies it because
but each of these might act to damage our he delights in it, and he delights in it
DNA—by evoking translocations, by trig- because it is beautiful. If nature were not
gering amplification, by causing mutations. beautiful, it would not be worth knowing,
When any of these affect proto-oncogenes, and if nature were not worth knowing, life
there may be trouble. It appears that a set would not be worth living."
of cellular genes may contribute to the genWhen we have finally solved the riddle
esis of cancer, no matter what its cause. We of cancer, we will see beauty in the eleshould soon have the means to learn how gance of design by which the lives of our
these genes act, and in so doing, we will cells and, hence, of ourselves are governed.
seek to solve not only the riddles of the
cancer cell, but the riddles of normal
REFERENCES
growth and development as well. After Alberts, B. 1986. The function of the hereditary
centuries of bewilderment, the human
materials: Biological catalyses reflect the cell's
evolutionary history. Amer. Zool. 26:781-796.
intellect has finally laid hold of cancer with
a grip that may eventually extract the Barbacid, M. 1987. Ras genes. Ann. Rev. Biochem.
56:779-827.
deadly secrets of the disease.
Beasley, R. P. 1988. Hepatitis-B virus—the major
etiology of hepatocellular carcinoma. Cancer 61:
We learned much of this from chickens,
1942-1956.
beasts not renowned for glamour. The
Bishop,
J. M. 1982. Oncogenes. Sci. Amer. 246(3):
chicken virus discovered by Peyton Rous
80-93.
70 years ago was the first cancer virus to Bishop, J. M. 1983. Cellular oncogenes and retrobe isolated. Decades later, the oncogene
viruses. Ann. Rev. Biochem. 52:301-354.
src of this virus became the first retroviral Bishop, J. M. 1987. The molecular genetics of cancer. Science 235:305-311.
oncogene to be identified; the product of
src, the first retroviral "oncoprotein" to be Curran, T. and B. R. Franza, Jr. 1988. Fos and Jun:
The AP-1 connection. Cell 55:395-397.
identified; the action of the src protein, the Hanafusa,
H. 1977. Cell transformation by RNA
first biochemical mechanism implicated in
tumor viruses. Comprehensive Virol. 10:401-483.
666
J. MICHAEL BISHOP
Helkin,C.-H.andB. Westermark. 1984. Growth factors: Mechanism of action and relation to oncogenes. Cell 37:9-20.
Hogan, B. and K. Lyons. 1988. Gene targeting: Getting nearer the mark. Nature 336:304-305.
Hunter, T. 1984. The proteins of oncogenes. Sci.
Amer. 251(2):70-80.
Hunter, T. 1987. A thousand and one protein kinases. Cell 50:823-829.
Klein, G. 1987. The approaching era of the tumor
suppressor genes. Science 238:1539-1545.
Liu, E., B. Hjelle, R. Morgan, F. Hecht, and J. M.
Bishop. 1987. Mutationsof the Kirsten-ras protooncogene in human preleukaemia. Nature 330:
186-188.
Nowell, P. C. 1986. Mechanisms of tumor progression. Cancer Res. 46:2203-2208.
Pitot, H. C. 1983. The Rous sarcoma virus. J. Amer.
Med. Assoc. 250:1447.
Ponder, B. A. J. 1980. Genetics and cancer. Biochim.
Biophys. Acta 605:369-410.
Schwab, M. 1985. Amplification of N-myc in human
neuroblastomas. Trends in Genetics 1:271-275.
Shilo, B.-Z. 1987. Proto-oncogenes in Drosophila melanogaster. Trends in Genetics 3:69—73.
Strickberger, M. W. 1986. The structure and organization of genetic material. Amer. Zool. 26:769780.
Temin, H. M. 1972. RNA-directed DNA synthesis.
Sci. Amer. 226:24-34.
Todaro, G. J. and R. J. Huebner. 1972. The viral
oncogene hypothesis: New evidence. Proc. Natl.
Acad. Sci. U.S.A. 69:1009-1015.
Varmus, H.E. 1983. Retroviruses. In J.Shapiro (ed.),
Transposable elements, pp. 411-503. Academic
Press, New York.
Varmus, H. E. 1984. The molecular genetics of cellular oncogenes. Ann. Rev. Biochem. 18:553612.
Weinberg, R. A. 1983. A molecular basis of cancer.
Sci. Amer. 249(4): 126-143.
Weinberg, R. A. 1988. Finding the anti-oncogene.
Sci. Amer. 259(3):44-54.
Weiner, A. M., P. L. Deininger, and A. Efstratiadis.
1986. Nonviral retroposons: Genes, pseudogenes, and transposable elements generated by
the reverse flow of genetic information. Ann.
Rev. Biochem. 55:631-661.