Biology Reference
In-Depth Information
Based on these crystal structures, we surmised that a regulatory helix, most
likely in UCR2, might close across the active site and contact the aromatic arms of
the atypical PDE4 inhibitors. This proved to be correct as we were able to obtain
cocrystal structures of RS25344 (PDB ID: 3G4G) and PMNPQ (PDB ID: 3G45) in
contact with an
a
-helical region of UCR2 that closed across the active site (Fig.
2
).
The protein constructs used for cocrystallization experiments were highly engi-
neered using our Gene Composer
software (Lorimer et al.
2009
). We modified
surface residues on the catalytic domain to promote crystal packing, evaluated the
effect of different linker sequences between UCR2 and the catalytic domain on
crystallization, and evaluated literally hundreds of different protein constructs.
Gene Composer
™
also allowed us to design synthetic genes with codon usage
optimized for either bacteria, insect or mammalian expression systems (Raymond
et al.
2009
). Indeed, we explored upward of 180 synthetic gene constructs before
obtaining cocrystals of RS25344 or PMNPQ with forms of PDE4D and PDE4B
containing UCR2 expressed and purified from insect cells. We found in these
experiments that crystallization was ligand dependent.
Our PDE4B UCR2 cocrystal structure with PMNPQ reveals nearly a complete
structure of the truncated UCR2 seen in supershort PDE4 isoforms (Fig.
2
). In this,
an N-terminal
a
-helix of UCR2 lies over one wall of the PDE4B catalytic domain.
This is joined by a sharp beta-turn to a second
a
-helix that caps the active site and
contacts the two aromatic arms of the otherwise buried PMNPQ. In the PDE4D
UCR2 cocrystal structure, although the first 27 residues are disordered and not
visible in the crystal lattice, a well-defined helix spanning residues Asn191 to
Asp203 was observed over the active site. Unfortunately, in both structures, the
LR2 linker region between UCR2 and the N-terminal portion of the catalytic
domain is disordered and cannot be modeled. It is therefore not possible to
unequivocally determine from these structures whether the UCR2-catalytic domain
interaction is intramolecular or occurs between two different molecules in a dimer
formation, for example.
The UCR2
a
-helix closing across the PDE4 active site is identical in sequence
between isoforms, except for one key residue which contacts the PDE4 inhibitor
(Table
1
). This residue is a phenylalanine (Phe) in PDE4D and a tyrosine (Tyr) in
PDE4A-C. Fascinatingly, the phenylalanine polymorphism in PDE4D arose in the
mammalian lineage giving rise to primates, as PDE4D contains the ancestral
tyrosine in mice, rats, and dogs. This then has important consequences for drug
discovery when trying to extrapolate the pharmacology and tolerability of atypical
PDE4 selective inhibitors from, as is invariably done, rodents to man.
In addition to the interaction with the bound inhibitor, it is important to note that
closure of the UCR2 helix across the active site is stabilized by direct interactions
between the helix and the PDE4 catalytic domain. For example, there is good shape
complementarity between the UCR2 helix and the catalytic domain groove, Gln192
is positioned to make a hydrogen bond to the main chain carbonyl of Asn528, and
Phe201 fits within a hydrophobic cleft made by Ile376 and Met439 (numbering
based on PDE4D3 isoform, GenBank accession No. AAA97892). Mutation of
Phe201 alters inhibitory potency of atypical inhibitors interacting with UCR2
™
