272 Rev. sci. tech. Off.
int.
Epiz.,
17 (1)
allows
the detection of interacting proteins expressed from
complementary DNAs
(cDNAs)
in yeast. These attempts have
not met with success,
perhaps
because the CA 'bait' used in
these experiments did not assume the correct conformation
for
interaction with the Fvl gene
product.
An alternative approach, which ultimately proved successful,
was to
adopt
a genetic approach known as positional cloning.
In
essence, positional cloning comprises three steps. Firstly,
the gene in question is mapped as accurately as possible to a
small,
defined genetic interval. Then, the DNA from this
interval is cloned. Finally, the genes encoded within these
clones
are identified and differences between different alleles
are sought. Though labour-intensive, this approach has met
with considerable success in cloning genes associated with
inherited diseases in man and in a variety
of
mouse mutations,
in which little was known about the function of the mutated
gene
(11).
The
possibility of cloning Fvl in this way arose as a result of
observations made in the course of genetic studies with the
endogenous
MLVs
of
mice.
If a retrovirus infects a germ
cell,
the resulting provirus can
Become
part
of the germ line,
provided that
undue
harm
to the host is not caused by virtue
of
expression or by the position of the provirus in the genome.
These
inherited elements, which are known as endogenous
retroviruses, are associated with a plethora of biological
phenomena (7). The endogenous retroviruses of a given
species
can be classified into a few groups, each containing up
to ten thousand individual members which show a high
degree of nucleotide sequence similarity but differ in their
specific
integration sites within the genome. A lower degree of
sequence
similarity is seen between different groups. There
are between
fifty
and one
hundred
endogenous
MLVs
in mice
(7).
In the course of
mapping
one MLV sub-family, four
proviruses, called Xmv8, Xmv9,
Xmvl4
and
Xmv44,
were
shown to be tightly linked to Fvl (15).
Since
the Fv4
restriction gene was known to correspond to a fragment of a
retrovirus, which prevented infection by competing for the
receptor for virus binding
(24),
it was suggested that one or
more of these endogenous proviruses might correspond to
Fvl
(15). As described below, this suggestion,
though
not
correct
in detail, contained an element of
truth.
To
test the possibilities that one of these proviruses might be
Fvl
(or
if
not Fvl itself, might lie
close
enough to Fvl to allow
positional cloning of the
gene),
a detailed genetic analysis of
the region of distal chromosome 4 known to contain Fvl was
performed
(37).
This
study
ruled
outXmv8,
Xmv9 and
Xmv44
as candidates for Fvl. However, Xmv9 and two other markers
(Nppa
and
lap3rcl
1) showed no recombination with Fvl, and
the researchers argued that these markers must lie within
1.2 megabase (Mb) of one another and Fvl, confirming the
feasibility
of a positional approach.
The
research team set out to clone the chromosomal interval
spanning these markers. Genomic libraries containing large
fragments of DNA derived from
C57BL
mice (Fvl
b/b)
cloned
as yeast artificial chromosomes
(YACs)
were screened for
Nppa
and Xmv9. A variety of different
YACs
were isolated.
Unfortunately, attempts to generate an ordered array of
overlapping
YACs
failed because many of the
YACs
were
unstable, probably as a result of multiple copies of a repeated
zinc
finger gene.
At
this point, the research team decided to abandon a
systematic
gene-finding approach and. concentrate on testing
the cloned
YACs
for Fvl activity in a functional test (Fig. 3),
based on the argument that since Fvl was a dominant gene,
introduction of a
YAC
containing Fvl into
cells
should result
in an alteration of their susceptibility to
MLV
infection. Mouse
cells
were fused with yeast carrying Xmv9 or
Nppa
YACs
and
stable
clones were isolated. These
cell
clones were expanded
and infected with N- and
B-tropic
viral vectors carrying a gene
for
resistance to the antibiotic puromycin. By enumerating
puromycin-resistant colonies, the
cells
could be typed for Fvl
phenotype. Using this approach, a YAC carrying the
Nppa
gene was shown also to contain Fvl activity
(5).
As expected,
given the
Fvlb/b
origin of the
YAC,
the introduced resistance
gene affected N-tropic virus replication but did not inhibit an
NB-tropic
virus.
To
isolate the Fvl gene, a cosmid library was
prepared
from
the
YAC
and individual cosmids carrying Fvl were identified
in the functional assay. Fragments of a positive cosmid were
retested and a 6.5 kilobase (kb) Fvl-positive clone was
isolated.
The DNA sequence of this clone was determined and
a
1.4 kb open reading frame with a predicted protein
product
of
459 amino acids, subsequently shown to possess Fvl
activity
restricting the replication of B-tropic virus, was
identified. The corresponding fragment was cloned from a
library
prepared
from an Fvln/n mouse restricted B-tropic
MLV,
thus
confirming that the research team had succeeded
in cloning Fvl. The predicted
product
of the n allele is
nineteen amino acids shorter
than
that of the b
allele;
the
alleles
also differ at two other amino acid positions (5).
Origin
of
Fv1
Analysis
of the structure and distribution of Fvl revealed a
number of
unusual
features (Fig. 4) (5, Le
Tissier
et al, in
preparation). Firstly, most genes have multiple exons,
whereas Fvl comprises a single long exon. Secondly, the
presumptive polyadenylation signal is provided by the second
of
a pair of short, interspersed nuclear repeats (B2 elements)
located
downstream of the open reading frame, rather
than
by
a
'natural' gene-specific signal. Thirdly, the Fvl gene is present
only
in mice and not in rats or humans, which implies a
relatively
recent evolutionary origin.
Sequence
data
bank comparisons provide an explanation for
these findings. The Fvl coding sequence is related to the gag
gene of the endogenous retrovirus family, called
HERV-Ls
or