Combinatorial Phage Antibody Libraries
problems associated with this type of non-
immunized library is ensuring that the
gene pool from which the library is made
is truly naive. In fact, most V
and V
genes taken from IgM and light chain
mRNA have already been biased in the
host, precluding the selection of antibody
against many antigens.
Nevertheless, some highly speciFc anti-
bodies against diverse targets have been
selected from nonimmune libraries. In
one example, a nonimmunized library was
screened against a diverse panel of anti-
gens, and speciFc antibodies (sc±v) were
isolated in every case, including antibodies
against botulinum neurotoxin,
elementary bodies, and the tu-
mor marker Erb-B2; antibodies to the latter
of which had dissociation constants be-
low 1 nM. Another nonimmune library
independent clones has been
described from which has been isolated
both cross-reactive and highly speciFc ±abs
against the highly related glycoprotein hor-
mones, human chorionic gonadotropin,
human luteinizing hormone, and human
follicle-stimulating hormone. Yet another
nonimmune library was engineered using
p9 display, and it performed comparably
to many p3-displayed libraries.
Semisynthetic Libraries
Another source for nonimmunized li-
braries consists of antibodies that are
encoded strictly by germ line genes. Such
semisynthetic libraries are constructed
from naturally occurring antibody genes
in which some or all of the CDRs are
composed of synthetic gene segments,
and typically the CDR3 region of the
heavy chain is composed of a random-
ized oligonucleotide sequence. A synthetic
library has been described in which 49 V
germ line genes were used as a starting
point for library construction. These germ
line heavy chains were given randomized
CDR3 regions of Fve or eight residues
and were combined with a single germ
line lambda light chain gene. The library
size of 2
clones represented only a
fraction of the potential diversity in this
system. However, this limitation must be
considered in conjunction with the knowl-
edge that immune systems themselves can
represent only a fraction of the potential
diversity at any one time, and that even
germ line antibody genes are themselves
biased by evolutionary pressures. To date,
many synthetic antibody phage libraries
have been described, some with consid-
erable diversity (e.g. ‘‘GrifFths library,’’
The importance of antigen recognition
of CDR sequences, particularly the CDR3
of the heavy chain, makes them an ideal
target for introducing segments of syn-
thetic sequence to generate very large
numbers of different speciFcities. Since
naturally occurring heavy-chain CDR3 seg-
residues, the diversity that may be gen-
erated by randomization of this region
alone is virtually limitless. In this sense,
semisynthetic libraries may be considered
to be more truly diverse and unbiased
than those created from a naive source.
However, although this aspect of library
technology offers many possibilities, here
too there are limits to diversity. ±or ex-
ample, the complete randomization of 16
amino acids in a single heavy-chain CDR
will alone require a library of more than
clones to fully represent it. This Fgure
is much greater than the transformation
efFciency of
E. coli
clones per
microgram of vector DNA). The proba-
bility of identifying desired speciFcities
may, however, be increased by custom-
designing semisynthetic libraries using
previous page 1414 Encyclopedia of Molecular Cell Biology and Molecular Medicine read online next page 1416 Encyclopedia of Molecular Cell Biology and Molecular Medicine read online Home Toggle text on/off