The avid binding of HIV-1 Nef to the Src homology-3 (SH3) domain of Hck (K-
D 250 nM) has been shown to involve an interaction between the RT-loop of H
ck-SH3 and residues in Nef outside of its prototypic polyproline type II (P
PII) helix-containing SH3-ligand region. Such distinctive interactions are
thought to provide specificity and affinity for other SH3/ Ligand protein c
omplexes as well. Here, we have constructed and successfully displayed on t
he surface of M13 bacteriophage particles a complex library of SH3 domains,
which are derived from Hck but carry a random hexapeptide substitution in
their RT-loops (termed RRT-SH3). Using this strategy we have identified ind
ividual RRT-SH3 domains that can bind to Nef up to 40-fold more avidly than
Hck-SH3. Some of these high-affinity RRT-SH3 domains resembled Hck-SH3 in
that they bound much less well to a Nef variant containing an engineered F9
0R mutation that interferes with docking of the native Hck RT-loop. In addi
tion, we could also select RRT-SH3 domains with an opposite specificity, wh
ich were dependent on the Arg90 residue for strong binding, and bound 100-f
old less well to unmodified Nef. These results demonstrate the utility of p
hage-display in engineering of signaling protein interaction domains, and e
mphasize the importance of the RT-loop in SH3 ligand selection, thus sugges
ting a general strategy for creating SH3 domains with desired binding prope
rties. (C) 1999 Academic Press.