Phage display systems

Dom Spinella dspinella at
Tue Oct 21 10:15:44 EST 1997

Keith Jolley writes:
> We are considering using a phage display system for antibody epitope
> mapping studies and are considering which kit to purchase. If anyone
> has experience with either Invitrogen's FliTrx system or NEB's Ph.D.
> kits I would be interested in hearing of any potential pitfalls or
> shortcomings as well as any successes. The main concern is with the
> 12-mer library kits, in that the stated complexity of the systems is
> 2x10^8 (Invitrogen) to 2x10^9 (NEB) independent primary clones,
> compared to a total 20^12 (4x10^15) possible combinations. Clearly
> only a tiny proportion of dodecapeptides will be represented in these
> systems. NEB claim that "by spreading the equivalent diversity of a
> 7-mer library over a window of 12 residues permits affinity selection
> of peptide ligands for targets requiring more than 7 ligand residues
> for tight binding". Does anyone's experience bear this out?
> I would also be interested in hearing of experiences with
> other similar systems.
> Thank you,
> Keith Jolley


There are two considerations here.  First off, NEB is correct: their
dodecamer library affords a "sliding window" effect such that each phage
displays 5 (i.e. 12-7) overlapping heptameric peptides.  Thus, you get
more peptide diversity for your money.  We have built libraries of up to
43 random amino acids which have even greater advantages along the same
lines, despite the fact that, as you point out, it is impossible to make
"complete" libraries representing every possible peptide sequence. These
advantages are particularly relevant when the binding "epitope" is long
or discontinuous. (See: McConnell et al, "Construction and Screening of
M13 Phage Libraries Displaying Long Random Peptides", J. Molecular
Diversity 1:165-176 for a description and more complete discussion).  We
have also used the NEB system and found it to be fine.  

The alternative consideration however is that many antibody epitopes are
in fact 6-7 residues long, and are simple linear (i.e.
non-discontinuous) sequences.  Hence, for the longer displayed peptides,
once you isolate a binding phage and infer the amino acid sequence of
the displayed peptide, you still won't know immediately which of the
overlapping peptides constitutes the epitope without doing some
additional work (unless there is substantial sequence homology to some
natural antigen and you can figure out where the homology begins and
ends). The point is that there is a tradeoff:  the maximum information
is obtained from the smallest possible binding peptide, and yet the
probability of finding a binding sequence is higher in libraries
displaying longer peptides. So, the choice is yours.  In our case, we
are not looking for antibody epitopes but receptor binding molecules. 
Nevertheless, the problem is similar and our solution is to screen ALL
our libraries simultaneously and use the data from the shortest

BTW, there is also a Molecular Diversity forum on the Bionet which,
while not as well subscribed as this user group, does often deal with
phage display issues. If you haven't already done so, I would post your
question there as well -- perhaps you'll get a different opinion (for
one thing, I've never used the Invitrogen libraries and don't know how
good they are).

I hope this rambling discussion is of some help!  Good luck.
-- Dom Spinella

More information about the Methods mailing list