Department of
BIOCHEMISTRY AND BIOPHYSICS

 

 

 

 

fs
or
©
QO
Lid
oC
I
<l
=)
=
=<
<I

 

 

 

Fe

UNIVERSITY OF CALIFORNIA, SAN FRANCISCO

 
HAROLD E. VARMUS, M.D., Professor

MOLECULAR STUDIES OF RETROVIRUSES AND HEPATITIS-B
VIRUSES

PAUL BATES, Postdoctoral Fellow
BRUCE BOWERMAN, Graduate Student
SHANTANU BASU, Postdoctoral Fellow

DELICIA CABALLERO, Lab Helper

MARIO CHAMORRO, Staff Research Associate
LUNG-JI CHANG, Postdoctoral Fellow
TITIA DE LANGE, Postdoctoral Fellow

HISAMARU HIRAL Postdoctoral Fellow
JOSH KAPLAN, Graduate Student (shared with Bishop)
JAN KITAJEWSKI, Postdoctoral Fellow
TYLER JACKS, Graduate Student
HITEN MADHANI, Graduate Student
JANINE MARINOS, Administrative Assistant
JOHN MASON, Postdoctoral Fellow
EFFIE MEREDITH, Secretary
DAVID MORGAN, Postdoctoral Fellow
TUANTU NGUYEN, Staff Research Associate
SUZANNE ORTIZ, Staff Research Associate
PETER PRYCIAK, Graduate Student
MARY ANNE SCHOFIELD, Graduate Student
RAYMOND SCOTT, Lab Helper
GREG SHACKLEFORD Postdoctoral Fellow
ANN TSUKAMOTO, Postdoctoral Fellow
JAN TUTTLEMAN, Postdoctoral Fellow
MICHAEL VERDERAME, Postdoctoral Fellow
JOHN YOUNG, Postdoctoral Fellow

In this laboratory, we use two intriguing and medically-important classes of animal
-viruses--the retroviruses and hepatitis B-type viruses---as points of departure for
studying various aspects of the behavior of eukaryotic cells at the molecular level.
Several properties of these viruses extend our concerns beyond the usual confines of
virology: (i) the oncogenes of retroviruses are derived from normal cellular genes
(proto-oncogenes) that are themselves targets for mutations implicated in many types
of cancers; (ii) the DNA form of the retroviral genome (the provirus) is structurally
similar to transposable elements from all types of living organisms; (iii) RNA-directed
DNA synthesis is central to the life cycles of both classes of virus, transposition by
various mobile genetic elements, and the generation of certain pseudogenes and
abundantly repeated sequences in eukaryotic genomes; (iv) the complex pathological
consequences of persistent infections by both virus classes involve a wide range of
interactions between viral and host macromolecules; and (v) the regulation of viral
gene expression displays many features characteristic of cellular genes, but more
conveniently addressed with viral reagents. Consideration of such issues brings us to
grips with several fundamental questions in eukaryotic biology: what is the molecular
basis of cancer and of the normal regulation of growth? What is the nature of somatic
mutation? What devices are used to regulate gene expression? What is the mechanism
by which eukaryotic transposable elements integrate into chromosomes?

183
RETROVIRUSES AS TRANSPOSABLE
GENETIC ELEMENTS

The Mechanism of Proviral Integration

The central event in the retrovirus life cycle is
the covalent integration of a DNA copy of the RNA
genome into host chromosomes. Although a great
deal is known about the synthesis of unintegrated
double stranded viral DNA by the virus-coded
enzyme, reverse transcriptase, we know well only the
structural features of integrated (proviral) DNA and
few of the functional properties of the integrative
mechanism. Like many transposable elements
from plants, bacteria, yeast, and insects, proviruses
can be found at many different sites in host
genomes but are always joined to host DNA at the
same sites in viral DNA. The provirus contains viral
genes arranged as they are in viral RNA (most
commonly: 5'-gag-pol-env-3'), flanked by long
terminal repeats (LTRs) that are generated during
reverse transcription and used for regulation of
transcription. The LTRs terminate with short
inverted repeats, and the entire provirus is flanked
by short direct repeats of cellular origin generated
during the integration step.

We are using biochemical and genetic
techniques to characterize the components required
for successful establishment of the provirus. A few
years ago, we used site-directed mutagenesis to
reveal a function (integrase) encoded by the 3' end
of the pol gene and absolutely required for
integration. More recently, we have found that
synthesis and integration of retroviral DNA appear
to be mediated during infection by subviral

nucleoprotein complexes derived from entering .

virus particles; once synthesis of viral DNA is
completed, these complexes can carry out the
integration reaction when placed in a cell free
system as described below.

Bruce Bowerman has been characterizing the
nucleoprotein complexes responsible for synthesis
and integration of viral DNA. He has found that
cytoplasmic complexes containing linear DNA are
smaller than virus particles, but sediment more
rapidly (ca. 160S) than free DNA. DNA in the
complexes can be precipitated with antisera against
the major core (gag) proteins of virus nucleocapsid,
and the complexes can be separated from most
cellular material by density centrifugation or gel
chromatography. Complexes from the nucleus of
infected cells are similar in physical properties but
contain closed circular DNA with one or two LTRs,
as well as linear DNA.

184

Through a collaboration with Pat Brown, a post-
doctoral fellow in Mike Bishop's lab, Vira]
nucleoprotein complexes have been shown to
mediate correct retroviral integration in vitro, Pat
and Bruce developed a sensitive genetic assay for
integration by preparing complexes from cells
infected with a strain of murine leukemia virus
(MLV) carrying an amber suppressor tRNA gene
(supF). Extracts containing the complexes are
incubated with naked lambda phage DNA bearing
multiple amber mutations, and, after packaging,
phage are tested for their ability to grow on a non-
suppressor bacterial host.

Under these assay conditions, 0.1 to 1% of the
complexes mediate correct integration of MLV
DNA into bacteriophage DNA. All of the
components required for the integration activity are
tightly associated with the complexes, since little or
no activity is lost when the complexes are prepared
by gel filtration or rate zonal centrifugation.
Cytoplasmic complexes are at least as effective as
nuclear complexes, implying that linear DNA serves
aS precursor to the integrated form in vitro. We are
uncertain whether there is a transient circular
intermediate in vitro, but the absence of a
requirement for ATP as an energy source argues
against a circularization step. Development of a
physical assay that allows analysis of intermediates
in the reaction will now be required to examine such
issues more closely.

Several other approaches are being taken to
evaluate the role of the integrase and its putative
viral target sequence in the integration reaction. (i)
Frameshift and missense mutants of the integrase
have been used as helper viruses to form viral
pseudotypes with genomes carrying selectable
markers. Kimiko Hagino-Yamagishi found that the
rare proviruses formed by these pseudotypes are
mostly aberrant and include oligomers of viral DNA
generated by unknown mechanisms. (ii) The
integrase coding domain has been expressed at high
levels in yeast by Shantanu Basu. The protein binds
non-specifically to DNA and can be separated from
yeast proteins and nucleolytic activities by
conventional chromatographic procedures.
Preparations of integrase will be used for structural
studies (in collaboration with C. Craik and R. Stroud),
for attempts to complement integrase-deficient
complexes in the in vitro integration assay, and for
efforts to develop a soluble integration system. (iii)
The claim that closed circular DNA is the proximal
precursor to the integrated provirus depends heavily
upon an experiment by Panganiban and Temin; in
that experiment, the region of circular viral DNA
formed by joining the ends of linear DNA was
claimed to act as an integration site when relocated
at internal sites in the viral genome. We are
attempting to reassess this result using the in vitro
integration assay. (iv) Chemical inhibitors of the
integrase should block integration, an essential step
in the virus life cycle, and hence might be potent
therapeutic agents. We are adapting the in vitro
integration assay to screen for candidate inhibitors
of the human immunodeficiency virus (HIV), the
causative agent of AIDS.

Thus far we have no evidence that host factors
are required for integration, but there may be a host
factor that impedes integration. The mouse locus
called Fv-1 is known to house dominant alleles that
restrict replication of MLV at one or more early
steps in the life cycle. Peter Pryciak is using
susceptible and resistant celi types, in concert with
the in vitro integration assay, to determine the
mechanism by which Fv-1 alleles mediate
resistance.

Retroviral Mechanisms for Gene Expression:
Ribosomal Frameshifting

Retroviruses use idiosyncratic mechanisms for
synthesis and integration of viral DNA, but they
exploit the machinery of the host cell to perpetuate
the provirus and express its genes. As described in
previous reports, proviruses harbor signals for host
RNA polymerase I] (promoters, enhancers), for
RNA processing (splicing and polyadenylation), for
hormonally-regulated transcription (e.g. by
glucocorticoids), and for translation. One unusual
regulatory phenomenon that operates at the
translational level is described in this report.

Ribosomal frameshifting appears necessary for
expression of the pol gene of most retroviruses.
Tyler Jacks first demonstrated this in efforts to
explain how the Rous sarcoma virus (RSV) provirus,
which contains gag and pol in adjacent, briefly
overlapping reading frames, can direct synthesis of a
gag-poly polyprotein. A spliced mRNA for this
protein has not been found, suggesting that a
translational mechanism might allow bypass of the
gag terminator. Tyler used SP6 RNA polymerase to
make a single RNA species from cloned RSV DNA
in vitro and showed that the gag-pol RNA yields both
gag and gag-pol proteins in a reticulocyte lysate, in
proportions similar to those observed in infected
cells. Thus it seemed likely that ribosomal
frameshifting within the overlap region is used to
circumvent the gag terminator at a fixed frequency
(ca.3-5%).

Tyler then extended these observations to other
retroviruses, the human immunodeficiency virus
(HIV) and the mouse mammary tumor virus

(MMTV). The gag-pol overlap in HIV is relatively
long (205 nt) and the frameshift occurs near its 5

.end, whereas the RSV frameshift occurs near or at

the 3' end of a 58 nt overlap. MMTV (and a few other
retroviruses) have an arrangement of reading
frames that requires two frame shifts to enter the pol
frame: one shift from gag to a reading frame for a
viral protease (pro), and another from pro to pol.
The anticipated frameshifts occur within short
overlap regions at relatively high frequency (15-25%)
in vitro and in vivo to allow synthesis of adequate
amounts of the full-length fusion protein. The
MMTV overlap sequences are themselves unable to
mediate frameshifting, implying that sequence
context is important.

Tyler's recent work shows that two elements in
viral RNA are essential for efficient frameshifting: a
short sequence at the frameshift site and an inverted
repeat (capable of forming a hairpin structure)
several nucleotides downstream of the site. A few
sequences that can serve as frameshift sites---U
UUA, A AAA AAC, U UUA AAC, and A AAU UUU
(with triplets from the initial reading frame)-—-were
first identified by sequence gazing. Use of the U
UUA site in RSV and HIV RNA has been confirmed
by aminoacid sequencing of the "transframe"
protein and by site-directed mutagenesis of the site.
Tyler and Hiten Madhani demonstrated the
importance of the potential hairpin by showing that
3' deletion mutations impair frameshifting when
they invade the stem; frameshifting is also impaired
by a five base mutation that destabilizes the stem
and restored by a compensatory mutation. The
presence of the hairpin sequence correlates with a
translational pause near the frameshift site,
suggesting that retardation of polyribosomes is
required for slippage into the -1 reading frame.
Hiten is now attempting to gauge the allowable
spacing between the site and the hairpin. The
tRNAs involved in the frameshift events have yet to
be identified.

We have naturally sought other contexts in
which frameshifting might occur. Frameshifting in
the +1 direction almost certainly occurs during
translation of RNA of the yeast retrotransposon Ty
and perhaps during synthesis of the hepatitis B virus
reverse transcriptase (see below). Robin Colgrove
has written a computer program to seek genes likely
to mediate -1 frameshifts based upon Tyler's
consensus sequences; several candidates will soon
be subjected to experimental test.

Retroviruses as Genetic Vectors

Retroviruses have demonstrated their ability to
transduce cellular genes and ferry them as part of

185
viral genomes into new cells, where they are
efficiently expressed. We and many others are
exploiting this property of retroviruses to study
genes (see section II), principally with MLV vectors
that are suitable for expressing two cloned genes, at
least one of which is selectable. These vectors can
be propagated as helper-free, high titre, replication-
incompetent stocks, they can be used to remove
introns from cellular genes, and their proviruses can
be readily recovered by molecular cloning.

The construction of cell lines expressing subsets
of retroviral genes also permits close inspection of
hitherto elusive viral interactions, including
recombination, heterozygosis, and phenotypic
mixing. Jan Tuttleman has been attempting to
determine whether specific recognition signals
between core and envelope components mediate
the formation of virus particles; she is doing this by
measuring the efficiency of formation of viral
pseudotypes in which the core components are
provided by RSV and the envelope proteins by MLV,
RSV, or hybrid genes.

Greg Shackleford has recently succeeded in
devising a vector system based upon the use of
MMTV and has produced a vector that can ferry
foreign genes into cells when complemented by
wild-type MMTV. In addition, he has generated the
first infectious clone of helper DNA, and virus
produced from this clone can induce mammary
tumors in mice. Components of these clones are
being used by Greg and John Mason to develop an
MMTV helper cell for propagation of helper-free
vector stocks. In addition, mutagenesis of the helper
clone will allow us to assess the hitherto elusive role
of an unassigned open reading frame in the MMTV
LTR during replication and tumorigenesis.

Cloning Genes for Retroviral Receptors

Entry of retroviruses into cells depends upon
host-encoded transmembrane proteins that serve as
virus receptors. The remarkable specificity of virus-
host interactions has been known for over twenty
years, from studies of the polymorphic envelope
proteins of avian retroviruses and their
corresponding receptors, yet little biochemical
information is available about the receptors or about
the nature of their interactions with viral envelope
glycoproteins. Over the past few years, receptors for
several animal viruses have been identified or their
genes cloned; perhaps the most remarkable
example is the first known receptor for a retrovirus,
the lymphocyte cell surface antigen CD4 that is
required for attachment of the AIDS virus, HIV, to
target cells.

186

Paul Bates, John Young, and Suzanne Ortiz are
attempting to clone the chicken genes that encode
the receptors for multiple subgroups of avian
retroviruses. Genomic DNA and cDNAs in
expression vectors are being introduced into avian
and mammalian cell lines that lack the appropriate
receptors (e.g. for subgroup A or subgroup B
viruses); cells that have acquired instructions for
making the receptors are identified by an acquired
susceptibility to avian retroviruses carrying a marker
that can be selected or screened for. (These
markers include antibiotic resistance genes and
genes encoding cell surface markers for which good
antisera are available.)

We ultimately expect these experiments to
reveal the nature of the receptors, the basis for
subgroup specificity and polymorphism (does it
reside in protein sequence or carbohydrate
modifications?), the sites of significant interaction
between the receptors and the various envelope
proteins, and perhaps the normal function of the
host-encoded receptors.

RETROVIRUSES AS ONCOGENIC AGENTS

Retroviruses competent to induce tumors are
conveniently grouped in two categories: those highly
oncogenic agents that carry oncogenes transduced
from normal cells and those less efficient agents
that lack their own oncogenes. Among the first
group are viruses employing about twenty
distinctive oncogenes (v-onc's); our studies have
been confined almost exclusively to the most
intensively examined member, v-src, the oncogene
of Rous sarcoma virus (RSV). v-sre encodes a
phosphoprotein of 60,000 daltons (pp60v-src) that
displays protein kinase activity in vitro and induces
phosphorylation of tyrosine residues in several
putative target proteins in vivo. Each v-onc is
derived from a cellular proto-oncogene, and these in
turn are often members of gene families (e.g. c-src is
one of at least seven closely related genes, the latest
of which, hck, was recently discovered by Nancy
Quintrell in Mike Bishop's lab). With the exception
of a few proto-oncogenes that have proved to
encode growth factors or growth factor receptors, the
normal functions of proto-oncogenes are unknown,
though mutant (putatively oncogenic) versions of
these genes frequently surface in a variety of
neoplasias in man and animals.

In the second group of oncogenic retroviruses,
lacking their own oncogenes, are a large number of
viruses producing a wide spectrum of diseases. Over
the past several years we have worked principally
with the avian leukosis virus (ALV) and
myeloblastosis-associated virus (MAV), which
induce B cell lymphomas and nephroblastomas,
and with the mouse mammary tumor virus (MMTV),
which induces mammary carcinomas. Viruses of
this second type appear to act as insertional
mutagens, enhancing the expression of adjacent
cellular oncogenes as an initial step in
tumorigenesis. The cellular protooncogenes
affected by these mutations (e.g. c-myc, c-Har-ras,
and int-1) have come under scrutiny as a
consequence. All of our recent work in this area has
been focused upon the int-1 gene activated by
MMTV.

Two related issues are also briefly addressed in
this section. We have begun to explore the
possibility of using homologous recombination in
animal cells to inactivate proto-oncogenes and
thereby evaluate their normal functions. We have
also initiated efforts to study genes that appear to
act in a recessive manner during oncogenesis
(recessive oncogenes"),

The Function of the src Gene of RSV:
Analysis of Mutants of src

We have isolated a large group of non-
conditional] src mutants arising spontaneously in an
RSV provirus integrated in the genome of an
infected rat cell. These mutants make full sized
products without protein kinase activity (presumably
due to missense mutations) or truncated proteins
(presumably due to nonsense or frameshift
mutations), and they have proven to be useful
reagents for examining the nature of eukaryotic
mutations, as well as the functions of the v-sre gene.

In the past, we have used conventional cloning
and sequencing procedures to analyze selected
mutations and secondary mutations that restore the
wild type phenotype. Some of these have produced
surprising information about a large amino terminal
portion of pp60v-sre that lies between an essential
signal for myristylation at the amino terminus and
the tyrosine kinase domain in the carboxyterminal
half of the protein. Mary Anne Schofield has shown
that large substitutions and duplications near the
aminoterminus are compatible with full
transforming potential, and Mike Verderame has
found that loss of ability to transform rat cells (with
retention of transforming potential in chicken cells)
is due to loss of one codon in a highly conserved
sequence upstream of the kinase domain. (The
latter mutant, v-sre-L, is discussed again below.)
Earlier analysis of our mutants indicated that a
second short (7 Kd) protein was expressed from v-src
at low levels; site-directed mutagenesis by Paul
Bates has shown that the second protein is not
required for transformation of mammalian cells.

One deletion mutant previous sequenced by
Graeme Mardon suggested that the seventh amino
acid encoded by v-sre might be crucial for
recognition by the mammalian myristoyl
transferase; Josh Kaplan has used site-directed
mutagenesis to show that lysine and arginine are
functional in the seventh position, whereas
asparagine is not. (The asp-7 allele is mentioned
again below.)

Mike Verderame has been using the so-called
“polymerase chain reaction“ to amplify the v-src
gene in over a dozen cell lines containing non-
conditionally mutant genes encoding 60 Kd proteins
without detectable kinase activity. Sequences of the
amplified DNAs should provide a survey of lesions
that inactivate the v-sre kinase.

Genetic Approaches to Host Proteins that
interact with v-sre Protein

It is likely that several host components are
essential to the function of v-src; among these are
targets for the v-src kinase activity and proteins that
modify the v-src gene product and determine its
locale in the cell. We have developed three genetic
assaults on the problem of isolating host genes for
such factors.

a) Suppressing non-transforming src alleles.
Josh Kaplan has developed a system for isolation of
host mutants that suppress v-src mutations. He has
built recombinant MLVs carrying the E.coli
guanosine phosphoribosyl transferase (Ecogpt) and
various src alleles (e.g. those that produce non-
transformed or partially transformed phenotypes or
that are host-dependent for transformation). After
infection of HPRT- rat cells and mutagenesis,
transformed cells could arise by secondary
mutations in src, by src-independent host mutations,
or by src-dependent host mutations; the
experimental set-up permits tests that distinguish
among these possibilities. In theory, DNA-
mediated transformation techniques can ultimately
be used to isolate any host gene in which the desired
src-dependent mutations arise.

This system has been put to its first
experimental test with cells infected by MLV
carrying the c-sre coding sequence. Mutagenesis of
non-transformed cells producing high levels of c-src
protein has yielded a number of transformants
currently being analyzed. Additional tests (e.g. with
the myristylation signal mutant of v-src, asp-7) are
also in progress.

b) Isolating cells resistant to v-sre. Mary Anne
Schofield has used MLV-based vectors carrying v-

187
src and metabolic markers to build a rat cell line
containing three expressed wild type v-src loci. She
has mutated this line and selected at least one
revertant to a non-transformed phenotype. In this
setting, restoration of the normal phenotype is more
likely to occur as a result of mutations in host genes
than in viral genes. Attempts are being made to
characterize the resistance phenomenon and to
identify and isolate the host genes involved.

c) Suppressing host range differences. v-src-L
appears to have lost transforming activity in rat cells,
while retaining it in chicken cells. With the
assumption that the protein product of this allele
can distinguish between rat and chicken versions of
some cellular factor(s) required for transformation,
Mike Verderame has been attempting to identify
the relevant chicken gene. The central strategy
involves co-transfection of mammalian cells
carrying the v-src-L allele with chicken DNA and a
metabolic marker, scoring for transformed colonies.

Oncogenesis by Retroviruses Without Viral
Oncogenes: MMTV and the int-1 Gene

Our current work on proto-oncogenes activated
by proviral insertion mutations is confined to the int-
1 gene; accounts of recently completed studies of c-
myc and c-Ha-ras genes can be found in last year's
report and the references.

The int-1 gene was discovered in 1982 when Roel
Nusse, then a post-doctoral fellow in our group, used
the technique known as “transposon tagging” to
identify genes that serve as targets for insertion
mutation during mammary tumor induction by
MMTV. Int-1 is unrelated to known retroviral
oncogenes or to any other cloned mammalian gene,
and it has been highly conserved during evolution.
Recently, Nusse’s group in Amsterdam discovered
that the Drosophila homologue of int-1 is the
segment polarity gene, wingless. About 75% of
tumors in C3H mice have MMTV insertion
mutations that activate expression of int-1 a gene
that appears to be unexpressed in all adult tissues
save testis (see below). The proviruses are present
on both sides of a transcriptional unit; they are
generally pointed away from the transcribed region
and hence act as enhancers rather than promoters
of int-1.

The nucleotide sequence of int-1 cDNA predicts
a protein of 370 amino acids, with a hydrophobic
amino terminus containing a consensus site for
signal peptidase cleavage, a cysteine-rich
carboxyterminus, and four potential N-linked
glycosylation sites. Tony Brown and Jackie Papkoff
have examined the properties of int-1 proteins made

_ epithelial cells.

188

proteins made in vitro or in cultured cells after
introduction of a transcriptionally activated gene,
using poly- and monoclonal sera raised against
synthetic peptides. The primary product is subject
to multiple modifications, including Proteolytic
cleavage and several glycosylations, producing at
least five distinguishable forms of int-1 protein. The
protein(s) appear to enter a secretory pathway
according to several criteria, but we have yet to
detect the protein in extracellular fluids. Further
efforts are being made to understand the processing
pathway, to identify the biological active product of
the gene, and to seek the anticipated cell surface
receptor for int-1 protein. To assist in these tasks, int-
1 proteins have been produced in E.coli by Ann
Tsukamoto and Mario Chamorro, and in insect cells
infected with Baculovirus vectors by Mario.

Last year, in a collaborative study with Gail
Martin's lab, we showed that expression of int-1 is
confined to the midgestational embryo and the
mature testis. Greg Shackleford has recently shown
that int-1 RNA is present only in post-meiotic cells of
the testis, most abundantly in early spermatids.
Expression in the embryo is confined to the neural
tube, with transcripts detectable in all portions of the
developing central nervous system save that
destined to become the forebrain. (In situ
hybridizations by McMahon's group in England
further restricts expression to a small subset of cells
within the neural tube.) Greg is attempting to
ascertain the function of int-1 in developing neural
tissue by observing the behavior of the gene in
cultured cells from neural lineages.

Last year Tony Brown developed a biological
assay for int-1, using MLV vectors to induce a
dramatic change in the phenotype of an established
line of mouse mammary epithelial cells (C57MG).
The infected cells form dense, refractile foci,
permitting a direct quantitative assay for the gene
and mutant versions of it. Ann Tsukamoto is
creating a library of linker insertion mutations and
John Mason is making several site directed
mutations (e.g. to affect potential cleavage and
glycosylation sites) to map the functional domains
required for production of biologically active protein.
Since int-1 does not render the C57MG line
tumorigenic, Ann is also testing the effects of the
gene in other lines and in primary mammary
In a collaborative effort with Rudi
Grossched], Ann is also preparing transgenic mice
carrying the int-1 gene governed by the MMTV LTR
(as found in virus-induced mammary tumors), by a
histocompatibility gene promoter (to incite
expression in many tissues), and by the elastase
promoter (to confine expression to the exocrine
pancreas). These experiments should tell us
whether the oncogenic potential of the int-1 locus
can be demonstrated in epithelial tissues other than
breast or in non-epithelial cells; they should also
provide us with excellent materials for seeking
additional oncogenes that participate in mammary
tumorigenesis.

Tony Brown has found that, although MLV
vectors carrying int-1 do not transform fibroblast cell
lines, infected fibroblasts induce morphological
transformation of adjacent C57MG cells. This
apparent "paracrine" effect could be due to secreted
int-1 protein itself or to some other factor produced
by cells under the influence of the int-1 gene. We
are attempting to distinguish between these
alternatives, since the assay is a potentially useful
one in a search for the int-1 receptor and in
purification of biologically active int-1 protein.

Recessive Oncogenes and Wilms' Tumor

All of the well-studied oncogenes appear to
operate in a dominant fashion, yet studies of several
tumor types over the past fifteen years, particularly
retinoblastoma and Wilms' tumor (a
nephroblastoma), indicate that inactivation of
certain loci on both chromosomes can play a
decisive role in tumorigenesis. Cytogenetic
techniques localized some of these recessive loci to
chromosomal bands (e.g. the Wilms' tumor gene
{WT] to band 13 on the short arm of human
chromosome 11), and a major event in the past year
was the serendipitous isolation of part of the
retinoblastoma (RB) locus by Dryja, Weinberg, and
colleagues. However, virtually nothing is yet known
about the product of the RB gene or about its
normal and abnormal functions.

Titia de Lange has been taking three
approaches to isolation of the WT gene: functional
tests for genes that suppress the oncogenic
phenotype of Wilms’ cell lines, cloning of
translocation breakpoints near or within 11p13, and
differential cDNA cloning. The last approach is
based upon the observation that some
retinoblastomas cannot make RB RNA whereas
others make a apparently normal] transcript
presumed to contain small mutations; the presence
or absence of WIT RNA might then be the only
major difference between the RNA populations in
two Wilms’ tumors.

Inactivating Proto-Oncogenes: A Test Case
with c-fos

Understanding the normal functions of proto-
oncogenes would be easier if it were possible to
make mutations that inactivate both copies of such

189

genes in mammalian cell lines or animals. To gauge
the feasibility of this sort of approach, Jan Tuttleman

_has begun experiments, in collaboration with Mario

Capecchi's lab in Utah, that are designed to
interrupt the proto-oncogene c-fos by homologous
recombination. Clones of the first exon of the gene
are being constructed to contain promoterless
coding sequences that produce secreted proteins
(e.g. interleukin-2) or confer a selectable phenotype;
such clones will be microinjected into rat fibroblast
lines shown to have two copies of c-fos accompanied
by a second selectable marker. After further
screening by Southern blotting of cells that express —
the promoterless sequence, we expect to identify
successfully microinjected cells that have inserted
the interrupted c-fos exon into the homologous
region of the chromosome. This procedure will be
performed twice to disrupt both copies of c-fos. To
avoid lethal effects of the double knock-out, a
retrovirus vector carrying an intact fos coding
sequence will be used to infect some lineages;
dependence upon the "rescuing" provirus can later
be tested by selecting against another marker in the
provirus, thereby curing the cell of the ancillary fos
sequence. If successful, this method will be applied
to other genes and to cells (embryonic stem cells)
capable of repopulating the mouse germ line.

HEPATITIS B VIRUSES AS ANALOGUES OF
RETROVIRUSES

The hepatitis B viruses of man and other
animals---woodchucks, ground squirrels, and ducks-
--have some striking resemblances to retroviruses.
First, these viruses replicate their DNA genomes
through RNA intermediates, using a viral enzyme to
synthesize viral DNA from an RNA template in the
final phase of the life cycle. Second, the hepatitis B
viruses frequently establish a chronic infection, and
their genomes are sometimes integrated into the
host genome, though not with the specificity for viral
sequences manifest by retroviruses. Third, the
human and woodchuck viruses seem to have a
major role in the generation of primary
hepatocellular carcinoma (hepatoma), a tumor
arising after long latency and often carrying
integrated viral DNA, though the viral genome is not
believed to contain an oncogene.

We have been attempting to understand the
intricacies of the unusual replication cycle, the
peculiarities of expression of the highly compact
genomes, and the mysterious connection between
infection and carcinogenesis characteristic of this
virus class. Much of our work on these problems
over the past few years is summarized in last year's
report, and additional detail can be found in the
report by Don Ganem, with whom all of these
studies are jointly conducted. Discussion here is
confined to relatively few topics.

Biogenesis of Hepatitis B Virus Reverse
Transcriptases

Although the synthesis of hepatitis B virus DNA
from an RNA template has been studied in viral
cores isolated from infected livers, virtually nothing
is known about the physical properties of the reverse
transcriptase or about the manner in which it is
generated. The enzyme is presumed to be encoded
mainly by the longest open reading frame in the
hepatitis B viral genome, since the predicted
product of that frame shares some amino acid
sequence with retroviral polymerases. However, no
mRNA dedicated to translation of that frame has
been identified, no effective anti-polymerase
antisera are available, and the enzyme has not been
obtained in soluble form from virus particles.

To generate better reagents for studying the
enzyme, Lung Chang has expressed the long open
reading frame of duck, rodent, and human hepatitis
B viruses in vitro and in bacterial and baculovirus
vector systems. These proteins are being used to
raise effective antisera, to screen sera from infected
patients and animals for titre against the reverse
transcriptase, and to seek enzymatic activities.

Since all hepadnaviral genomes contain
overlapping core and polymerase reading frames, it
is possible that ribosomal frameshifting is used to
produce core-polymerase fusion proteins during
infection. Unlike the situation with retroviruses,
however, the frameshift would occur in the +1 rather
than in the -1 direction. We have been testing this
possibility in two ways. Peter Pryciak has asked
whether in vitro translation systems from
reticulocytes or rodent livers can mediate a core-
polymerase frameshift; thus far the answer appears
to be negative. Lung Chang has been mutagenizing
the core-polymerase overlap region of the duck
hepatitis B virus genome in order to distinguish
between models for production of the polymerase
that involve frameshifting and those that involve
translational initiation in the polymerase frame; the
most decisive mutants are still under test.

Determinants of the Host Range of Hepatitis p
Viruses

Hepatitis B viruses generally manifest relatively
narrow host and tissue tropisms, but the viral and
cellular factors responsible for these tropisms (je.
host receptors and their viral ligands) have yet to be
identified. We showed a couple of years ago that
the genomes of two closely related rodent viruses
the ground squirrel and woodchuck hepatitis viruses,
can be recombined in vitro without loss of viability,
Since the woodchuck virus will not grow in ground
squirrels, it should be possible to assign host range
determinants to a component of the viral genome by
testing several recombinants for their ability to grow
in ground squirrels, an exercise started by Christoph
Seeger and being extended by Joel Lavine. (Both
parental viruses grow in woodchucks, so the viability
of all recombinants can be checked in the
permissive host.) If, as anticipated, host range is
determined by pre-surface sequences, more refined
mutagenesis will be performed in this region.

The Oncogenic Role of Hepatitis B Viruses

We have been attempting to take advantage of
the very high incidence of hepatoma in woodchucks
chronically infected with WHV, as compared with
the low incidence in ground squirrels chronically
infected with GSHV. In collaboration with Christoph
Seeger (a former post-doctoral fellow here) and Bud
Tennant at Cornell, we have found, as mentioned in
the preceding section, that GSHV (and GSHV-WHV
recombinants) can establish sustained, high-titre
infection in woodchucks. A large cohort of newborn
woodchucks has been chronically infected with
GSHV to ascertain whether the high frequency of
tumorigenesis is determined by virus or host. If the
virus is crucial, GSHV-WHV recombinants will be
used to identify the region of the viral genome that
influences oncogenic potency.

We are continuing to characterize the collection
of human hepatomas received some years ago from
collaborators in Hong Kong. Joel Lavine is
extending work begun by Teddy Fung to seek
rearrangements of known proto-oncogenes and to
ask whether hepatitis B viral DNA is integrated in
the same region of the genome in independent
tumors, as might be expected if a cellular locus was a
favored target for insertional activation during
tumorigenesis.

190
1986-1987 PUBLICATIONS

VARMUS, H.E. and BISHOP, J.M. Proteins encoded by oncogenes. Cancer Surveys 5: 153-158, 1986.

VARMUS, H.E. Cellular and viral oncogenes. Chapter 9 in "Molecular Basis of Blood Diseases”
(Stamatoyannopoulos, Nienhuis, Leder, and Majerus, eds.), W.B. Saunders Co., New York, pp. 271-346, 1987.

GANEM, D. and VARMUS, H.E. The molecular biology of the hepatitis B viruses. Ann. Rev. Biochem. 56: 654-693,
1987.

VARMUS, HLE. Reverse transcription. Scientific American, in press, 1987.

SEEGER, C., GANEM, D. and VARMUS, H.E. Biochemical and genetic evidence for the hepatitis B virus
replication strategy. Science 232: 477-484, 1986.

WESTAWAY, D., PAPKOFF, J., MOSCOVICI, C. and VARMUS, H.E. Identification of a provirally activated c-Ha-
ras oncogene in an avian nephroblastoma via a novel procedure: cDNA cloning of a chimaeric viral-host
transcript. EMBO J. 5: 301-309, 1986.

NOTTENBURG, C. and VARMUS, H.E. Features of the chicken c-myc gene that influence the structure of c-myc
RNA in normal cells and bursal lymphomas. Mol. Cell. Biol. 6: 2800-2806, 1986.

JAKOBOVITS, A., SHACKLEFORD, G.M., VARMUS, H.E. and MARTIN, G.M. G.M. Two proto-oncogenes
implicated in mammary carcinogenesis, int-1 and int-2, are independently regulated during mouse development.
Proc. Natl. Acad. Sci. USA 83: 7806-7810, 1986.

PERSING, D.H., VARMUS, H.E. and GANEM, D. Antibodies to pre-S and X determinants arise during natural
infection with ground squirrel hepatitis virus. J. Virol. 60: 177-184, 1986.

BROWN, A.M.C., WILDIN, R.S., PRENDERGAST, T.J. and VARMUS, HE. A retrovirus vector expressing the
putative mammary oncogene inf-1 causes partial transformation of a mammary epithelial cell line. Cell 46: 1001-
1009, 1986.

PERSING, D.H., VARMUS, H.E. and GANEM, D. Inhibition of secretion of hepatitis B surface antigen by a
related presurface polypeptide. Science 234: 1388-1392, 1986.

SCHWAB, M., KLEMPNAUER, K-H., ALITALO, K., VARMUS, H. and BISHOP, M. Rearrangement at the 5’ end
of amplified c-myc in human COLO 320 cells is associated with abnormal transcription. Mol. Cell. Biol. 6: 2752-
' 2755, 1986.

ENDERS, G.H., GANEM, D. and VARMUS, H.E. 5° terminal sequences influence the segregation of ground
squirrel hepatitis virus RNAs into polyribosomes and viral core particles. J. Virol. 61: 35-41, 1987.

NOTTENBURG, C., STUBBLEFIELD, E. and VARMUS, H.E. An aberrant ALV provirus inserted downstream
from the chicken c-myc coding sequence in a bursal lymphoma results from intrachromosomal recombination
between two proviruses and deletion of cellular DNA. J. Virol. 61: 1828-1833, 1987.

BROWN, P.O., BOWERMAN, B., VARMUS, H.E. and BISHOP, J.M. Correct integration of retroviral DNA in
vitro. Cell, 49: 347-356, 1987.

STONE, J., DE LANG, T., RAMSAY, G., JAKOBOVITS, E., BISHOP, J.M., VARMUS, H. and LEE, W. Definition of
regions in human c-myc that are involved in transformation and nuclear localization. Mol. Cell. Biol. 7: 1697-1709,
1987.

SHACKLEFORD, G.M. and VARMUS, H.E. Expression of the proto-oncogene int-1 is restricted to postmeiotic
male germ cells and the neural tube of mid-gestational embryos. Cell, in press, 1987.

191
JACKS, T., TOWNSLEY, K., VARMUS, H.E. and MAJORS, J. Two efficient ribosomal frameshifting events are

required for synthesis of mouse mammary tumor virus gag-related polyproteins. Proc. Natl. Acad. Sci. USA 84.
4298-4302, 1987. ,

QUINTRELL, N. R. LEBO, VARMUS, H., BISHOP, J.M., PETTENATI, MJ., LE BEAU, M.M., DIAZ, Mo. and
ROWLEY, J.D. Identification of a human gene (HCK) that encodes a protein-tyrosine kinase and is expressed jn
hemopoietic cells. Mol. Cell. Biol. 7: 2267-2275, 1987.

SEEGER, C., GANEM, D. and VARMUS, HLE. In vitro recombinants of ground squirrel and woodchuck hepatitis B
viral DNA produce infectious virus in squirrels. ]. Virol., submitted, 1987.

PAPKOFF, J., BROWN, A.M.C. and VARMUS, HE. The int-1 proto-oncogene products are glycoproteins that
appear to enter the secretory pathway. Mol. Cell. Biol., submitted, 1987.

BROWN, A.M.C., PAPKOFF, J., FUNG, Y.K.T., SHACKLEFORD, G.M. and VARMUS, HE. Identification of
protein products encoded by the proto-oncogene int-1. Mol. Cell. Biol., submitted, 1987.

MAJOR RESEARCH SUPPORT
Source: National Institutes of Health
Title: Molecular Analysis of Retroviruses and Oncogenes.
Total period
of award: August 1, 1985 to July 31, 1992.
Source: National Institutes of Health
Title: The Molecular Biology of Hepatitis B-type Viruses.
Total period
of award: April 1, 1985 to March 31, 1989.
Source: National Institutes of Health
Title: Oncogenic Potential of the Hepatitis B-type Viruses
Total period
of award: March 1, 1984 to February 28, 1988.
Source: American Cancer Society - Research Professorship
Total period
of award: 1984 -.

PERSONNEL WHO LEFT THE LABORATORY 1986-1987 EMDNMG&

1. Anthony Brown, Assistant Professor, Department of Cell Biology and Anatomy, Cornell U. Medical

College

2, Greg Enders, Medical Student, UCSF Medical School

3. Kimiko Hagino-Yamagishi, Staff Researcher, Department of Viral Oncology, Cancer Institute, Tokyo

4. Carol Nottenburg, Staff Researcher, Fred Hutchinson Cancer Research Center

5. Jacqueline Papkoff, Staff Researcher, Syntex Research

6. Christoph Seeger, Assistant Professor, School of Veterinary Medicine, Cornell University

7. Hans von Dam, Graduate Student, Department of Molecular Biology, Agricultural University in

Wageningen

192