Biomedical Engineering Reference
In-Depth Information
the gene or whole gene clusters. Fragments averag-
ing about 4 kb are likely to be inconveniently short.
Alternatively, the gene may be contained on an
Eco
RI fragment that is larger than the vector can
accept. In this case the appropriate gene would not
be cloned at all.
These problems can be overcome by cloning
ran-
dom
DNA fragments of a large size (for
High mol. wt eukaryotic
DNA (>100 kb)
λ
Charon 4A DNA
(replacement vector)
Cleave with a mixture
of
Hae
III
and
Alu
I
(very partial digest)
Size fractionate
Anneal
natural
cohesive
ends of
λ
~20 kb
Eco
R
I
methylase to
block
Eco
R
I
sites
replacement
vectors, approximately 20 kb). Since the DNA is
randomly fragmented, there will be no systematic
exclusion of any sequence. Furthermore, clones will
overlap one another, allowing the sequence of very
large genes to be assembled and giving an oppor-
tunity to 'walk' from one clone to an adjacent one
(p. 107). Because of the larger size of each cloned
DNA fragment, fewer clones are required for a com-
plete or nearly complete library. How many clones
are required? Let
n
be the size of the genome relative
to a single cloned fragment. Thus, for the human
genome (2.8
λ
Me Me
Eco
R
I
Me Me
Blunt-end ligation with
Eco
R
I
linker molecules
Me Me
+
Internal fragments
Size fractionate
to remove
internal
fragments
Me Me
Eco
R
I
Me Me
Me Me
Anneal
Eco
R
I
cohesive ends. Ligate
10
6
kb) and an average cloned frag-
ment size of 20 kb,
n
×
Me Me
Me Me
10
5
. The number of
independent recombinants required in the library
must be greater than
n,
because sampling variation
will lead to the inclusion of some sequences several
times and the exclusion of other sequences in a
library of just
n
recombinants. Clarke and Carbon
(1976) have derived a formula that relates the
probability (
P
) of including any DNA sequence in a
random library of
N
independent recombinants:
=
1.4
×
Me Me
Me Me
Packaging
in vitro
Phage particle
Fig. 6.2
Maniatis' strategy for producing a representative
gene library.
ln (
1
−
P
)
but insertion of the resulting fragments into vectors
requires additional steps. The more commonly used
procedure involves restriction endonucleases. In the
strategy devised by Maniatis
et al
. (1978) (Fig. 6.2),
the target DNA is digested with a mixture of
two
restriction enzymes. These enzymes have tetra-
nucleotide recognition sites, which therefore occur
frequently in the target DNA and in a complete
double-digest would produce fragments averaging
less than 1 kb. However, only a partial restriction
digest is carried out, and therefore the majority of
the fragments are large (in the range 10 -30 kb).
Given that the chances of cutting at each of the
available restriction sites are more or less equi-
valent, such a reaction effectively produces a
random set of overlapping fragments. These can be
size-fractionated, e.g. by gel electrophoresis, so as
to give a random population of fragments of about
N
=
1
ln
1
−
n
Therefore, to achieve a 95% probability (
P
0.95) of
including any particular sequence in a random
human genomic DNA library of 20 kb fragment size:
=
ln (
1095
−
.
)
5
N
=
.
=×
42
10
1
ln
1
.
−
14
×
10
5
Notice that a considerably higher number of recom-
binants is required to achieve a 99% probability, for
here
N
10
5
.
How can appropriately sized random fragments
be produced? Various methods are available. Random
breakage by mechanical shearing is appropriate
because the average fragment size can be controlled,
=
6.5
×
