E.Y. Cotrina et al.
Bioorganic & Medicinal Chemistry 28 (2020) 115794
interesting fact is that no mismatches between the efficiency maps have
occurred in case of compounds 5, 6, 8 and 10 which, in turn, are most
potent than iododiflunisal.
inhibitor prepared by mixing different volumes of a stock solution of the
compound in H2O/DMSO (1:1) to give a range of final compound con-
centrations of 0–40 μM. DMSO content was adjusted to a final 5% (v/v),
Appearing above IDIF in the nBEI-NSEI map and on top of the pro-
gression line of NPOL = 3, product 8 is the most efficient candidate of
where all ligands tested are soluble. After 30 min incubation at 37 ◦C
with 15 s shaking every minute, 100 L of 400 mM KAcO, 100 mM KCl,
and 1 mM EDTA buffer at pH 4.2 were added to each well. The final
μ
this series. The potency of this compound (IC50 = 3.6 μM), compares
with the most potent TTR amyloidogenic inhibitor known up to date,
mixture, containing 0.4 mg/mL TTR, 0 to 40 μM ligand, and 5% DMSO,
triiodophenol61 (IC50 = 3.2
μ
M), and provides an approximate idea of
was incubated at 37 ◦C with 15 s shaking every minute. Absorbance at
340 nm was monitored for 1.5 h at 1 min intervals. A control solution of
the level of optimization achieved. This optimization capacity of the
LEIs is reinforced when the comparison between the biological activities
of 8 and Tafamidis, the drug marketed in Europe for the treatment of
FAP, shows that product 8 has an enhanced potency and efficiency.
Thus, under the kinetic turbidity assay conditions, Tafamidis presents an
the ligands at the highest concentration (40 μM) following the same
procedure in the absence of TTR was also monitored, showing that the
ligands remained soluble and no turbidity due to colloidal aggregation
was observed. Initial rates of protein aggregation (ν0) were obtained
from the linear plot absorbance versus time. The dependence of ν0 on
inhibitor concentration is defined as:
IC50 = 6.59
μ
M, and values of 6.48 and 1.30 for nBEI and NSEI respec-
tively. Besides 8, two more compounds, 6 (3.9
μ
M, NPOL = 3) and 5
(4.3 μM, NPOL = 4) have been identified and characterized as better and
υ0 = A + B⋅eꢀ C[I]
more efficient candidates than IDIF. The main feature of this triad (5, 6
and 8) is that their three crucial LEI variables (potency, size and po-
larity) have been optimized at the same time.
where ν0 is the initial rate of fibril formation (in absorbance units per
hour, AU⋅hꢀ 1) and [I] the concentration of the inhibitor (
μM). From the
Finally, it is worth noting that the non-halogenated derivatives at the
position 5, that were also included among the test compounds, show
lower potencies than their halogenated counterparts, and are found
outside the preferred candidate efficiency region. A particular case is 6A
adjustable parameters, the IC50 (inhibitor concentration at which the
initial rate of protein aggregation is half than that in absence of inhib-
itor) and RA(%) (percentage reduction of amyloidosis at high inhibitor
concentration) were calculated (See SI, Fig. S3)
which shows a potency of 21.7 μM and a very poor efficiency behavior,
Crystallographic Complex. 3D atomic coordinates of the TTR-
iododiflunisal complex (PDB ID: 1Y1D)29 used in the present work
were obtained from the structural information available in the Protein
The asymmetric crystal unit of TTR complexes is formed by a dimer, two
ligand molecules (one for each binding site) and water molecules; taking
this into account, coordinates for the tetrameric form of TTR were ob-
tained by applying the crystallographic symmetry transformations
described in the pdb file. For residues with multiple conformations, we
considered the one with the highest occupation factor.
far from its iodinated counterpart 6 (3.9 μM); this reinforces our hy-
pothesis on the important role that halogen atoms play on TTR tetramer
stabilization and prevention of fibril formation.
3. Conclusions
Here we presented the first experimental evidence of a novel appli-
cation of the LEI formulation for prospective lead optimization by using
the iododiflunisal chemico-biological space as example. The results also
suggest that the LEI methodology, both retrospective and prospective,
may be easily combined and integrated with computational workflows
such as pharmacophore modeling and docking experiments. This pro-
spective LEI approach has allowed us to identify a triad of compounds
with optimized properties (potency, size and polarity) with respect to
iododiflunisal. Significantly, compound 8 that maps in the extreme
North-East corner of the efficiency plane, has the best combination of
IC50 and physico-chemical properties (size, polarity) as ‘combined’ in
the corresponding NSEI, nBEI values (Table 2). Compound 8 compares
Hydrogen Atoms Refinement. Added hydrogen atoms were
energy-minimized by using the Protonate 3D package implemented in
MOE 2013.08.54 Ligand partial charges were obtained by computing the
electrostatic-potentials around the optimized structures using MOE
2013.08. Minimization was carried out using a distance dependent
dielectric constant and a cutoff distance of 10 Å for Van der Waals in-
teractions. Hydrogen atoms refinement was accomplished using 1000
cycles of steepest descents followed by conjugate gradient until the
maximum gradient of the energy was smaller than 0,05 kcal/mol∙Å2.
Data Set Selected. A set of 2300 biphenylic compounds was
extracted from MMsINC Database.53 LigX package (MOE 2013.08) was
used for the hydrogen addition and ligand preparation.
very favorably with triiodophenol (3.2 μM), one of the most potent TTR
fibrillogenesis inhibitor known up to date, which may be of interest for
future drug developments in the field of TTR-related amyloid diseases
treatment. In contrast, compound 10 with better IC50 than IDIF and
slightly larger values of efficiency per size (IC50 = 3.7 vs. 4.2;
nBEI = 6.85 vs. 6.66, respectively) exhibits a significantly lower value of
the polarity efficiency NSEI (NSEI = 1.09 vs. 1.79) (Table 2) making it
significantly more polar and less suitable for further development.
Docking Experiments. MOE 2013.08 package was used to perform
the docking studies of the data set selected with the crystallographic TTR
complex. Alpha Triangle was used as placement method, Alpha HB as
score function and MMF94 as forcefield in the refinement step of the
docking solutions.
3.1. Experimental section
Author contributions
3.1.1. Synthesis and characterization of compounds.
All diflunisal and IDIF analogues were prepared following the
Schemes 1, 2 and 3. The synthesis and characterization are described in
the SI.
Overall research design and writing of the manuscript: G.A., C.A.Z.,
A.P., N.B.C., J.Q., D.B. Computational studies: D.B. Chemistry experi-
ments: G.A. Biological experiments: E.Y.C., M.V.
Protein and inhibitors. The human TTR variant Y78F protein was
recombinant expressed in E. coli and purified as already reported.60 All
assays were performed in buffers containing a final 5% (v/v) DMSO
concentration for solubilization of the ligands.
Declaration of Competing Interest
The authors declare no competing financial interest.
Acknowledgements
Kinetic Turbidimetric Assay. Inhibition of fibrillogenesis was
determined by the kinetic turbidimetric assay previously reported.60 In
seven different wells of a 96-well microplate, 20 μL of a 4 mg/mL TTR
variant Y78F solution in 20 mM potassium phosphate buffer, 100 mM
KCl, and 1 mM EDTA at pH 7.6, was mixed with an 80 L solution of
μ
We thank Dr. Lluís Bosch for help on the synthesis work. Funding
6