Full Papers
tion, Invitrogen) was diluted 1:2000 in standard buffer (50 mm
Tris·HCl pH 8.0, 200 mm NaCl, 1 mm DTT) before addition of impor-
tin-a to a concentration of 2 mm and filling assay wells to a volume
of 100 mL. The standard Tris buffer was selected from a buffer
screen, as it showed the greatest thermal stabilisation of importin-
a. Fragments were screened at a final concentration of 5 mm with
that of murine DIBB-importin-a1 in the Xenopus–mouse TPX2–im-
portin-a structure; PDB ID 3KND) to enable rigid body refine-
[14]
ment, after which coordinates were refined using iterative cycles of
[25]
[26]
[27]
PHENIX, CCP4, and COOT. Electron density maps were then
inspected for difference density corresponding to the soaked li-
gands, which were then built into the model manually using COOT,
with restraints generated in PHENIX. The resulting models were fur-
ther refined with CCP4 or PHENIX, as well as adding waters and
ions from the crystallisation conditions. The stereochemistry of the
5
% DMSO. The fluorescence readout was monitored from 35–608C
À1
(
LightCycler) or 25–608C (Thermocycler), both at 0.018Cs . All
samples were measured in duplicate. Fragments with a negative
thermal shift greater than twice the negative standard deviation of
each plate’s negative controls wells (5% DMSO) were removed
from the enriched library.
[28]
structure was finally assessed using MolProbity. Molecular graph-
ics were generated using the PyMOL Molecular Graphics System,
version 1.6.0.0, Schrçdinger LLC. Data collection and refinement
statistics are shown in Table S3 of the Supporting Information. At-
tempts at co-crystallisation are described in the Supporting Infor-
mation.
Ligand-observed NMR screening: The three ligand-observed tech-
niques—STD, CPMG, and WaterLOGSY—were performed on either
a Bruker Avance 500 MHz with TCI cryoprobe or on a Bruker
Avance 700 MHz spectrometer with a TXI cryoprobe. All samples
were made to a total volume of 200 mL and added to 3 mm capilla-
ry tubes (Bruker) before inserting into standard or thick-walled
NMR tubes (Bruker). All samples were made to volume in standard
Compound docking: All compounds were docked into an unli-
ganded structure of mouse DIBB-importin-a1. The protein was pre-
pared in Discovery Studio Visualizer (Release 3.5, Accelrys Software
Inc., San Diego, CA, USA) by removal of crystallographic waters
and addition of hydrogens before selecting the coordinates of the
key minor site residue Glu396. Compounds to be docked were
loaded as .mol files into Discovery Studio Visualizer where hydro-
gens and relevant charges were added before the structures were
minimised to relax bond lengths and fix bond angles. Ligands
were docked into the protein in a radius of 13 around Glu396
protein buffer with 10% D O and 20 mm TSP. Screens were con-
2
ducted with fragments at 2 mm from [D ]DMSO stock plates,
6
giving a final DMSO concentration of 2%. Binding was determined
on observation of a change in fragment signals in the presence of
2
0 mm importin-a, relative to a control sample containing only
buffer and no protein. If found to bind importin-a, a displacer
sample was made by addition of 40 mm TPX2 (IIKPFNLSKGKKRTF-
DEAAS) peptide (Designer Bioscience) direct to the protein sample
and displacement noted if the fragment signals showed a return
to those seen in the control sample.
[29]
using GOLD 5.1 using default slow docking settings with 25 runs
for each ligand. Figures of the ten lowest-energy docked positions
of each were exported and subjected to visual analysis.
Synthetic chemistry: General directions for the synthesis of the
compounds described as well as their spectroscopic analyses are
supplied in the Supporting Information. The purity of all com-
pounds tested for binding to importin-a was assessed by LCMS or
HPLC and is >95% unless otherwise stated.
Isothermal titration calorimetry: All experiments were performed
on a Microcal ITC200 instrument (GE Healthcare) at 258C. All ITCs
were performed at pH 6.0 using ITC buffer (100 mm citrate pH 6.0,
2
00 mm NaCl, 10% DMSO). Fragment ITCs were performed using
the competitive ITC technique following the procedure and formu-
la described by Zhang et al. (Supporting Information). Briefly, the
titration of 200 mm 20-amino acid TPX2 to 25 mm importin-a 10%
[20]
Accession numbers: The coordinates and structure factors for the
mouse–mouse TPX2–importin-a structure used to inform com-
pound synthesis, as well as importin-a compound structures 1, 11,
DMSO at pH 6.0 (average K =1.0 mm) was compared directly with
d
1
2, 13, 14, 15, 16, and 17 have been deposited in the RCSB Protein
U5N, 4U5S, 4U58, 4U5O, 4U5U, and 4U5V respectively.
the same titration in the presence of the fragment at 5 or 10 mm.
Direct ITCs were performed for elaborated compounds at pH 6.0 in
ITC buffer titrating 10 mm compound into 25 mm protein. Com-
pounds that showed double binding in crystal structures (13, 16,
and 17) had their ITCs fitted using a one-site binding model, as
this gave the best curve fitting, suggesting the second binder re-
mains relatively weak. Experimental parameters are described fully
in the Supporting Information.
4
Acknowledgements
This work was supported by the University of Cambridge Cancer
Research UK Medicinal Chemistry training program and Medical
Research Council grant U105178939 (to M.S.). We thank Shintaro
Aibara and Florian Gruessing for help with molecular biology;
Fleur Ferguson, Steffen Lang, Christopher Stubbs, Paweł Sled z´ ,
and Christina Spry for help with biophysical assays; and Anthony
Coyne, Jorge Gómez-Magenti, Jamal el-Bakali, and Rajaval Srini-
vasen for help with compound synthesis. We thank the beamline
staff at Diamond Light Source and ESRF for support with crystal-
lographic data collection.
Crystallisation and crystallographic methods: Crystals of unli-
ganded DIBB-importin-a were grown using hanging-drop vapour
diffusion in 24-well Linbro plates (Hampton Research) with 4 mL
drops of 100 mm DIBB-importin-a1 at 1:1 protein/precipitant
volume ratio (optimal precipitant conditions: 0.1m citrate pH 6.0,
1
.05–1.20m ammonium sulfate, 20–40 mm DTT) and streak-seeding
from an DIBB-importin-a seed crystal plate 18 h after sealing the
wells. Crystals reached their maximum size after 1–2 weeks, when
they were soaked in solutions of fragment (100 mm or saturating
concentration) in crystallisation buffer supplemented with 10%
DMSO for between 30 min and 5 h. Crystals were then briefly ex-
posed to crystallisation buffer (with half fragment concentration
and an additional 10% glycerol) as a cryoprotectant and flash
cooled at 100 K in liquid nitrogen. X-ray data were collected at the
Diamond Light Source or European Synchrotron Radiation Facility.
Keywords: cancer · fragment-based ligand design · nuclear
transporters · protein–protein interactions · structure-guided
ligand design
[23]
All datasets were processed and reduced using XDS and AIM-
[24]
LESS. The unit cells of these crystals were sufficiently close to
[
ChemMedChem 2015, 10, 1232 – 1239
1238ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim