Fragment screening is frequently used for hit identification. However, there was no report starting from a small fragment for the development of an anaplastic lymphoma kinase (ALK) inhibitor, despite the number of ALK inhibitors reported. We began our research with the fragment hit F-1 and our subsequent linker design, and its docking analysis yielded novel cis-1,2,2-trisubstituted cyclopropane 1. The fragment information was integrated with a structure-based approach to improve upon the selectivity over tropomyosin receptor kinase A, leading to the potent and highly selective ALK inhibitor, 4-trifluoromethylphenoxy-cis-1,2,2-trisubstituted cyclopropane 12. This work shows that fragments become a powerful tool for both lead generation and optimization, such as the improvement of selectivity, by combining them with a structure-based drug design approach, resulting in the fast and efficient development of a novel, potent, and highly selective compound.
Fragment screening is frequently used for hit identification. However, there was no report starting from a small fragment for the development of an anaplastic lymphoma kinase (ALK) inhibitor, despite the number of ALK inhibitors reported. We began our research with the fragment hit F-1 and our subsequent linker design, and its docking analysis yielded novel cis-1,2,2-trisubstituted cyclopropane 1. The fragment information was integrated with a structure-based approach to improve upon the selectivity over tropomyosin receptor kinase A, leading to the potent and highly selective ALK inhibitor, 4-trifluoromethylphenoxy-cis-1,2,2-trisubstitutedcyclopropane 12. This work shows that fragments become a powerful tool for both lead generation and optimization, such as the improvement of selectivity, by combining them with a structure-based drug design approach, resulting in the fast and efficient development of a novel, potent, and highly selective compound.
Fragment-based drug
design (FBDD) was
reported first as “structure–activity
relationship (SAR) by NMR” by Fesik et al. at Abbott Pharmaceutical in 1996.[1] During
the past 2 decades, various kinds of FBDD approaches have been reported.
In hit identification, fragment screening has the advantage over high-throughput
screening (HTS) because it generally has higher hit rates, and the
hit compounds from fragment screening tend to show good solubility
and high affinity with low molecular weight.[2] Also for hit identification, the fragment is a powerful tool in
the lead generation process. Several methods are reported for fragment-based
lead generation, for example, fragment-based growing and linking,[3] molecular hybridization,[4] and scaffold hopping.[5] In the optimization
phase, a cocrystal structure of a compound with a target protein can
be obtained more easily than in the lead generation step because of
the high affinity of the compound; therefore, a structure-based drug
design (SBDD) is predominantly adopted in that phase. As for anaplastic
lymphoma kinase (ALK) inhibitors, there had been no report on the
use of a fragment-based approach for the development of an ALK inhibitor,
despite the number of them in use for the treatment of ALK-positive
nonsmall cell lung cancer (NSCLC),[6−10] brain metastasis ALK-positive lung cancer,[10,11] and
central nervous system (CNS) disorders.[10,12] All of the
reports began with HTS or virtual screening[13] and used SBDD in the optimization process.[10] In this work, we describe the discovery process of cyclopropane
derivatives as novel, potent, and highly selective ALK inhibitors
from a small fragment. We also show that a fragment-assisted approach
is effective in improving not only activity but also selectivity in
lead optimization.
Results and Discussion
Fragment Identification
and Analysis of Cocrystal
Structures
We began our studies with high-concentration (0.3–300
μM) biochemical assays of a small fragment library. Among those
fragments, 3-(3-methyl-1H-pyrazol-5-yl)pyridine (F-1) had a half-maximal inhibitory concentration (IC50) value of 220 μM with a high ligand efficiency (LE) [−ΔG/heavy atom count (HAC), defined as the free energy of
binding divided by the number of nonhydrogen heavy atoms[14] of 0.42. The cocrystal structure of fragment F-1 with the ALK protein shown in Figure A (PDBcode 7JYT) suggested that the pyrazole moiety of
this fragment acted as the “hinge” binder by making
two hydrogen bonds; one was formed between the pyrazoleNH and Glu1197
O=C and the other between the pyrazolenitrogen with amideNH of Met1199. The pyridine moiety of fragment F-1 made
the CH−π interaction with Gly1202 (Figure A). Crizotinib is the first launched ALK
inhibitor for the treatment of NSCLC, and its ALK inhibitory activity
was highly potent under our enzymatic and cellular assays with IC50 values of 1.8 and 37 nM, respectively. X-ray crystallographic
analysis of the cocrystal structure of crizotinib with ALK revealed
several key interactions (Figure B, PDBcode 2XP2). The 2-aminopyridine NH and pyridinenitrogen form
two hydrogen bonds to the protein backbone of the kinase hinge region.
The pyrazole ring makes a CH−π interaction with the Gly1202
(3.7 Å), and the 2,6-dichloro-3-fluorophenyl group contributes
to the potency as a hydrophobic interaction, in which fluorine at
C3 plays a crucial role by filling a small hydrophobic pocket, formed
by the Gly1269 and Gln1254 residues, and polarizing the adjacent C4–H
bond to productively interact with Arg1253carbonyl oxygen.[6] The superposition of the cocrystal structure
of fragment lead F-1/ALK on that of crizotinib/ALK (Figure C) implied that fragment F-1can be connected to the 6-cholorine atom on the benzene
ring of crizotinib at the 4-position of the pyridine ring of F-1 with the shortest distance. The distance between 4-C on
the pyridine ring of F-1 and 6-Cl on the benzene ring
of crizotinib was 3.88 Å and corresponded to the three-bond length.
Figure 1
(A) Cocrystal
structure
of fragment hit F-1 with ALK (PDB code 7JYT) and (B) cocrystal
structure of crizotinib with ALK (PDB code 2XP2). Hydrogen bonds and CH−π
interactions are indicated as red dashed lines. (C) Structure overlay
of crizotinib/ALK (cyan) and fragment hit F-1/ALK (magenta).
(A) Cocrystal
structure
of fragment hit F-1 with ALK (PDBcode 7JYT) and (B) cocrystal
structure of crizotinib with ALK (PDBcode 2XP2). Hydrogen bonds and CH−π
interactions are indicated as red dashed lines. (C) Structure overlay
of crizotinib/ALK (cyan) and fragment hit F-1/ALK (magenta).
Design of Novel Scaffold
According
to an analysis of the binding mode of crizotinib in the ALK protein,
the biaryl structure and the lipophilicbenzene ring were considered
to be important to the high binding affinity. Therefore, we planned
to connect the fragment hit F-1 with the lipophilicbenzene
ring in the following three steps: (1) identification of the attachment
point, (2) determination of the length of a linker based on the distance,
and (3) design of the linkers.The design for the novel scaffold
began by setting the oxygen atom at the 4-position of the pyridine
ring in order to provide the attachment point of the linker. Although
a biaryl structure is normally twisted by the introduction of a substituent
at the ortho-position,[15] we assumed that
the oxygen atom at the ortho-position might assist to keep the two
rings, the pyrazole ring and the pyridine ring, coplanar by forming
an intramolecularhydrogen bond with the pyrazoleNH or the pyrazoleCH. An unchanging conformation was thought to be necessary to maintain
the potent kinase inhibitory activity because the angle of the pyridine
ring was important to form a CH−π interaction with the
Gly1202. In addition, we replaced the chlorine atom at the 6-position
of crizotinib with a carbon atom in order to provide an attachment
point on the aryl moiety. As a second step, the 2,6-dichloro-3-fluorobenzene
group was simplified into a 4-fluorobenzene ring to make the calculation
simple. As mentioned earlier, the distance between the pyridine ring
of fragment F-1 and the chlorine atom of crizotinib was
3.88 Å; thus, the distance between the oxygen atom and the replaced
carbon atom was 2.7 Å, which corresponded to the two-bond-length
linkers. In the last phase, we exhaustively designed the two-bond-length
linkers to restrict the conformations by cyclization (Figure ). We expected the cyclization
to be particularly effective on an adjustment of the angle of the
4-fluorobenzene ring.
Figure 2
Strategy
for linker design.
Strategy
for linker design.Ethyl and cyclized linkers
were designed
as two-bond-length linkers. The ring size in the cyclization ranged
from a three-membered ring to a six-membered ring, and each cyclized
linker had four stereoisomers; therefore, 17 structures were generated
in total. The linkers were named designs D-1 to D-17 as shown in Figure .
Figure 3
Designed linkers (D-1 to D-17).
Designed linkers (D-1 to D-17).
Computational Evaluation
of Designed Compounds
In order to select the best scaffold,
we performed a docking analysis of the designed compounds using the
software packages Maestro and Glide SP (Schrödinger Inc). For
the docking analysis, we calculated three scores: ΔE, root-mean-square deviation (rmsd), and ΔG. The ΔE values express the conformational
energy penalties (ΔE) possessed by each design
compound in the ALK protein. The rmsds are the measures of the average
distances between the benzene ring of crizotinib in the cocrystal
structure with the ALK and the benzene ring of D-1 to D-17. The rmsd value is usually used to assess how well a
submitted structure matches the known target structure. Thus, a lower
rmsd value indicates a better fit of the designed compound to crizotinib.
The ΔG values indicate the sum of the binding
energies, such as hydrogen bonding, van der Waals forces, and electrostatic,
hydrophobic, and solvation contributions to binding.The results
of the calculations for all the designed linkers (D-1 to D-17) are summarized in Table . It has been reported that if the conformational
energy penalties (ΔE) of the ligand are >3
kcal/mol, the calculated conformations would be uncertain.[16] Thus, we set the cutoff value of ΔE as 3 kcal/mol. According to the cutoff value of ΔE, D-1, D-2, D-3, D-4, D-5, D-8, and D-12 were selected. Among the seven compounds, D-1, D-3, D-8, and D-12 showed similarly
low rmsd (<1) and ΔG values. Last, we synthesized cis-1,2-cyclopropane D-3, regarding it as the
best compound based on the following two reasons: (1) the binding
of compound D-3 is entropically favored over those of
compounds D-1 and D-12 because the conformation
of D-3 is more rigid and (2) the synthetic tractability
of D-3 was better than that of D-8.
Table 1
Docking Scores of Designs 1–17a
design no.
ΔE (kcal)
rmsd (Å)
ΔG (kcal)
D-1
1.42
0.32
–47.6
D-2
2.87
1.38
–45.5
D-3
2.19
0.34
–48.4
D-4
1.30
3.58
–41.5
D-5
2.05
6.39
–41.5
D-6
3.90
1.56
–46.8
D-7
8.67
1.79
–42.2
D-8
1.39
0.74
–51.0
D-9
7.37
3.11
–39.9
D-10
3.84
1.48
–47.5
D-11
3.98
0.29
–48.9
D-12
2.44
0.53
–49.1
D-13
11.6
2.44
–41.5
D-14
4.31
2.03
–47.4
D-15
4.54
0.35
–47.3
D-16
14.8
1.83
–32.2
D-17
4.69
1.78
–37.8
All calculations were performed by MAESTRO and Glide SP.
All calculations were performed by MAESTRO and Glide SP.
Fragment-Assisted
SAR Analysis
Before
synthesizing the designed cis-1,2-cyclopropane derivative,
we prepared structurally related fragments to the initial hit F-1 by combinatorial chemistry. The SARaround the pyridine
ring of the fragment F-1 may be rapidly and easily analyzed
using the additional fragment library, which can be prepared in just
two steps: a Suzuki coupling and a deprotection of the tetrahydropyranyl
(THP) ethers group. The second fragment library consisted of mono-substituted
phenyl groups at ortho, meta, or para positions and heteroaromatic
rings such as pyrimidine (Figure ).
Figure 4
Preparation
of the second fragment library by combinatorial
chemistry.
Preparation
of the second fragment library by combinatorial
chemistry.A high-concentration assay against
the second
fragment library was conducted. The conversion into other heteroaromatic
rings, such as 2-pyridine or 5-pyrimidine, did not improve the potency.
The introduction of substituents at the 2- and 4-positions also diminished
the potency compared with the initial fragment hit 1 (data
not shown), whereas the substituents at the 3-position had a great
impact on the ALK inhibitory activity. The results are shown in Table . The electron-withdrawing
groups at the 3-position tended to increase the ALK inhibitory activities
(F-5, F-7, and F-9). Among
them, the 3-cyano-phenyl fragment F-7 showed the most
potent ALK inhibitory activity with an IC50 of 51 μM.
Because 3- and 4-pyridine fragments F-1 and F-2 were more potent than the phenyl fragment F-3, it was
expected that the introduction of the cyano group on the pyridine
fragment might show better potency compared with fragment F-7. However, the combination of a cyano group with a 4-pyridine ring
resulted in a decrease in ALK inhibitory activity (F-13, IC50 = 130 μM). Consequently, we decided to incorporate
the selected scaffold, cis-1,2-cyclopropane, into
the 3-cyano-phenyl fragment F-7.
Table 2
Results
of Fragment
Screening at High Concentration (0.3–300 μM)
compd (T-no)
Ar
ALK IC50 (μM)a
F-1
3-pyridine
220 (170–290)
F-2
4-pyridine
240 (200–290)
F-3
Ph
>300
F-4
3-F-Ph
240 (170–350)
F-5
3-Cl-Ph
66 (49–88)
F-6
3-Me-Ph
230 (160–330)
F-7
3-CN-Ph
51 (40–66)
F-8
3-MeO-Ph
170 (130–220)
F-9
3-CF3-Ph
60 (47–78)
F-10
3-t-Bu-Ph
270 (230–310)
F-11
3-CF3O-Ph
95 (77–120)
F-12
3-MeSO2-Ph
120 (95–160)
F-13
3-CN-4-pyridine
130 (100–160)
IC50 values and 95% confidence intervals (CIs) (given
in parentheses) were calculated by nonlinear regression analysis of
percent inhibition data (n = 2).
IC50 values and 95% confidence intervals (CIs) (given
in parentheses) were calculated by nonlinear regression analysis of
percent inhibition data (n = 2).
Evaluation of Designed
Compounds
The installation of
the cis-1,2-cyclopropoxy group on the 3-cyanophenyl
fragment F-7 generated 5-cyano-2-cyclopropoxy derivative 1 and 3-cyano-2-cyclopropoxy derivative 2 as
shown in Figure A;
therefore, it was necessary to discuss the direction of the cyano
group at the meta-position of the benzene ring in the pocket of ALK
protein because of its asymmetry. We compared the cocrystal structure
of fragment F-1/ALK with a newly obtained cocrystal structure
of fragment F-5/ALK (Figure B). In the two cocrystal structures, both
the chloro group and the pyridinenitrogen pointed toward the E0 region. It was to be expected that the chlorine atom could
interact with the hydrophobic residues Ala1200 and Gly1201 in the
E0 region; however, it was interesting that the polarpyridinenitrogen also sat in the hydrophobic E0 region. On the
basis of the observations, it was hypothesized that the hydrophiliccyano group might also point toward the E0 region and the
5-cyano-2-cyclopropoxy derivative would properly reflect the fragment
activity. Therefore, we synthesized a chiral (1S,2S)-2-(2,4-difluorophenyl) cyclopropoxy-4-cyano-benzene (1) to evaluate the profiles of the novel scaffold. We referred
our previous report that explored the SAR of substitutions on the
fluorobenzene ring[12] because the binding
mode of the benzene ring in the designed compound was estimated to
be similar to that of the benzene ring in the reported 1H-pyrrolo[2,3-b]pyridine analogues. We replaced 4-fluorobenzene
in the designed compound into 2,4-difulorobenzene that showed the
strongest ALK inhibitory activity among the various substituted benzene
or pyridine, although the data was not shown in the report. The 4-fluorobenzene
ring was utilized only in the simulation to simplify the calculations
due to its symmetry.
Figure 5
(A) Two possible structures by addition of the cis-1,2-cyclopropoxy group; 5-cyano-2-cyclopropoxy derivative 1 and 3-cyano-2-cyclopropoxy derivative 2. (B)
Cocrystal structures of fragments F-1 (yellow, PDB code 7JYT) and F-5 (cyan, PDB code 7JYS).
(A) Two possible structures by addition of the cis-1,2-cyclopropoxy group; 5-cyano-2-cyclopropoxy derivative 1 and 3-cyano-2-cyclopropoxy derivative 2. (B)
Cocrystal structures of fragments F-1 (yellow, PDBcode 7JYT) and F-5 (cyan, PDBcode 7JYS).As expected from the calculations,
the
chiral (1S,2S)-2-(2,4-difluorophenyl)cyclopropoxy-4-cyano-benzene
derivative (1) showed potent ALK enzymatic inhibitory
activity with an IC50 value of 0.080 μM and more
than 30-fold greater selectivity over 37 kinases in a corporate kinase
panel assay while exhibiting low ALKcellular inhibitory activity
with an IC50 value of >1.0 μM and moderate tropomyosin
receptor kinase (Trk)A inhibitory activity (Table ). The LE value of compound 1 for ALK was also remarkably high at 0.37, demonstrating that compound 1 was an attractive lead compound. On the other hand, the
enantiomer 3 corresponding to D-2 exhibited
a 16-fold decrease in ALK inhibitory activity with an IC50 value of 1.3 μM. The absolute configuration of compound 1 was determined as the 1S, 2S form by X-ray single diffraction analysis after derivatization to
aminomethyl compound 4 as shown in Figure . The assay results of 1 and 3 were consistent with the calculations for D-2 and D-3.
Table 3
Enzymatic
Kinase
Activity Profiles of 1 and 3
ALK enzyme
ALK cell
TrkA enzyme
calculation
compd
stereo chemistry
IC50 (μM)a
IC50 (μM)a
IC50 (μM)a
LEb for ALK
rmsd (Å)
ΔE (kcal)
1
1S, 2S
0.08
>1.0
0.95
0.37
0.747
1.08
3
1R, 2R
1.3
NTc
NTc
0.31
1.62
1.86
IC50 values and 95% CIs (given in parentheses) were calculated
by nonlinear regression analysis of percent inhibition data (n = 2).
LE (ligand
efficiency) = −1.36 log IC50/number of heavy atoms.
Not tested.
Figure 6
ORTEP of compound 4. Thermal ellipsoids
are drawn at
30% probability.
ORTEP of compound 4. Thermal ellipsoids
are drawn at
30% probability.IC50 values and 95% CIs (given in parentheses) were calculated
by nonlinear regression analysis of percent inhibition data (n = 2).LE (ligand
efficiency) = −1.36 log IC50/number of heavy atoms.Not tested.The cocrystal structure of compound 1 with ALK was solved and superimposed with the structure
of crizotinib extracted from the cocrystal structure with ALK (Figure A). As expected from
the binding mode of fragment F-1, compound 1 also bound to the ALK in the adenosine triphosphate pocket, and
the pyrazolenitrogen and NH of compound 1 made hydrogen
bonds with the backbone carbonyl of hinge residue Glu1197 and the
amideNH of hinge residue Met1199. As expected, the observed coplanarity
of the biaryl structure in the cocrystal supported our strategy 1
in Figure . In addition,
compound 1 also formed several key interactions. The
cyanobenzene ring was interacting with a hydrophobic residue, Gly1202.
Leu1256 participated in two CH−π interactions with the
2,4-difluorophenyl group and the methyl pyrazole ring, leading to
the L-shape conformation that is thought to be important to ALK potency.[6] Intriguingly, the 2,4-difluorophenyl ring of
compound 1 overlapped with the 2,6-dichloro-3-fluorophenyl
group of crizotinib with an rmsd value of 0.60 Å. The value was
approximately equal to the rmsd value of 0.34 Å in the calculated
conformation of D-3 as shown in Table . The cocrystal structure of 1 with ALK was also superimposed with the docking structure of D-3 (Figure B). The position of 2,4-difluorobenzene ring of compound 1 was almost the same as the 4-fluorobenzene ring of D-3, demonstrating that the calculated binding mode accurately reflected
the observed binding conformation.
Figure 7
(A) Overlay of cocrystal structure of 1 (green)/ALK
(gray) (PDB code 7JYR) and structure of crizotinib (magenta) extracted from the cocrystal
structure with ALK (PDB code 2XP2). (B) Superposition of cocrystal structure of 1 (green, PDB code 7JYR)/ALK and docking structure of D-3 (cyan)/ALK.
The p-loop was depicted by a string in yellow. The key amino acid
residues for binding are highlighted in orange.
(A) Overlay of cocrystal structure of 1 (green)/ALK
(gray) (PDBcode 7JYR) and structure of crizotinib (magenta) extracted from the cocrystal
structure with ALK (PDBcode 2XP2). (B) Superposition of cocrystal structure of 1 (green, PDBcode 7JYR)/ALK and docking structure of D-3 (cyan)/ALK.
The p-loop was depicted by a string in yellow. The key amino acid
residues for binding are highlighted in orange.Additional polar substituents
at the 2-position of 1-phenoxy-2-phenyl-cyclopropane were explored
in order to improve the enzymatic and cellularactivities by filling
a small hydrophilic pocket underneath the p-loop. The results are
summarized in Table . Among the evaluated compounds 5–11, aminomethyl compound 9 displayed the most potent ALK
enzymatic and cellular inhibitory activities with an IC50 of 0.029, 0.16 μM, respectively, the equivalent LE of 0.37,
and the improved ligand lipophilicity efficiency of 4.3 compared to
the nonsubstituted compound 5, and had the best selectivity,
being a 17-fold increase over TrkA.
Table 4
SAR of Substituents
at the 2-Position of Cyclopropane Derivatives
ALK enzyme
ALK cell
TrkA enzyme
compd
R
stereo chemistry
IC50 (μM)a
IC50 (μM)a
IC50 (μM)a
LEb for ALK
LLEc for ALK
1
H
1S, 2S
0.080
>1.0
0.95
0.37
2.9
5
H
racemic
0.13
NTd
1.2
0.36
2.7
6
CO2Me
racemic
0.041
0.45
0.085
0.34
3.3
7
CH2OH
racemic
0.034
NTd
0.14
0.37
4.3
8
CH2OMe
racemic
0.043
0.89
0.38
0.35
3.4
9
CH2NH2
racemic
0.029
0.16
0.50
0.37
4.3
10
CH2NHMe
racemic
0.049
NTd
0.92
0.35
3.8
11
CH2NMe2
racemic
0.11
NTd
1.6
0.32
2.8
IC50 values and 95% CIs (given in parentheses)
were calculated by nonlinear regression analysis of percent inhibition
data (n = 2).
LE (ligand efficiency) = −1.36 log IC50/number
of heavy atoms.
LLE = pIC50 – clog P.
Not tested.
IC50 values and 95% CIs (given in parentheses)
were calculated by nonlinear regression analysis of percent inhibition
data (n = 2).LE (ligand efficiency) = −1.36 log IC50/number
of heavy atoms.LLE = pIC50 – clog P.Not tested.
Fragment-Assisted Kinase
Selectivity Improvement
Next, we
demonstrated a design for the improvement of the kinase selectivity
using fragment information. The selectivity of the Trk family (TrkA,
TrkB, and TrkC) in ALK programs had already been surpassed in several
reports. The reported inhibitors utilized the single difference of
the amino acid residue between ALK and the Trk family. The ALK residue
Leu1198 was replaced by a tyrosine from the Trk family (Figure A). The residues were close
to the cyano group of compound 1 according to the cocrystal
structure of 1/ALK. The fragment assay results shown
in Table reminded
us of the trifluoromethyl benzene fragment F-8, whose
ALK inhibitory activity was 60 μM, similar to F-7 at 51 μM (Figure B). As for fragment activity, the trifluoromethylphenoxy cyclopropane
derivative was expected to present nearly equivalent ALK inhibitory
activity to cyanophenoxy cyclopropane derivative 9. Conversion
from a cyano group into a trifluoromethyl group was also predicted
to improve the selectivity over TrkA based on two factors: electronic
and steric aspects. First, it is generally recognized that the trifluoromethyl
group is highly electronegative,[17] thereby
leading to electronic repulsion with a polar amino acid residue represented
by a tyrosine. Second, it is considered that the trifluoromethyl group
is sterically bulkier than the cyano group, as supported by a web-based
calculation using the freely available Molinspiration online service,[18] with a gap in volumes of the substituents of
14.4 Å3. The fact provided an opportunity for the
bulkier trifluoromethyl group to crash with a tyrosine in the Trk
family. The trifluoromethylphenoxy cyclopropane derivative 12 was synthesized and evaluated.
Figure 8
Strategy to
improve the
selectivity over TrkA. (A) Cocrystal structure of 1/ALK
(green, PDB code 7JYR) and TrkA (cyan, PDB code 4F0I) with selectivity residues highlighted. (B) Comparison
between fragments F-7 and F-9 and the design
of compound 12 from compound 9.
Strategy to
improve the
selectivity over TrkA. (A) Cocrystal structure of 1/ALK
(green, PDBcode 7JYR) and TrkA (cyan, PDBcode 4F0I) with selectivity residues highlighted. (B) Comparison
between fragments F-7 and F-9 and the design
of compound 12 from compound 9.The profiles of compound 12 are summarized in Figure . As speculated,
compound 12 showed potent
ALK enzymatic inhibitory activity with an IC50 value of
0.018 μM and a 128-fold increase in selectivity over TrkA but
unexpectedly exhibited the decreased ALKcellular inhibitory activity
with an IC50 value of 0.95 μM. Compound 12 was chiral, and the absolute configuration was determined as 1S, 2S form by X-ray single diffraction
analysis after the derivatization to a p-nitrosulfonamidecompound 13. We confirmed that 1S, 2Scompound 12 was the eutomer because the other
enantiomer 1R, 2R exhibited a weaker
ALK enzymatic inhibitory activity with an IC50 value of
0.40 μM. A cocrystal structure of 12 bound to ALK
was obtained and indicated almost the same binding mode as that of
compound 1. Interestingly, the amino group of 12 forms a direct hydrogen bond with His1124 in the p-loop and a water-mediated
hydrogen bond network involving in Asn1254, contributing to an increase
in the binding affinity. A KINOMEscan screening platform of 456 kinases,
using an active site-directed competition binding assay, revealed
that compound 12 inhibited only five kinases other than
ALK; LTK and ULK3 with 5–10% control, DCAMKL1, FAK, and ULK1
with 10–35% control at 300 nM compound concentration, demonstrating
remarkably high kinase selectivity, whereas compound 12 also inhibited mutated ALK (L1198M) with 10–35% control at
300 nM, and ALK (C1156Y) with 1–5% control at 300 nM.
Figure 9
Data of compound 12. (A) In vitro profiles.
(B) ORTEP, thermal
ellipsoids are drawn at 30% probability. (C) Cocrystal structure with
ALK (green, PDB code 7JY4) and with the interaction residues highlighted in orange and the
p-loop highlighted in yellow. (D) KINOMEscan screening dendrograms
at 300 nM compound. The assays were performed by DiscoveRx Corporation.
Images were generated using TREEspot software.
Data of compound 12. (A) In vitro profiles.
(B) ORTEP, thermal
ellipsoids are drawn at 30% probability. (C) Cocrystal structure with
ALK (green, PDBcode 7JY4) and with the interaction residues highlighted in orange and the
p-loop highlighted in yellow. (D) KINOMEscan screening dendrograms
at 300 nM compound. The assays were performed by DiscoveRx Corporation.
Images were generated using TREEspot software.
Chemistry
Scheme shows the
synthesis of 4-cyanophenoxy-cis-1,2-disubstituted
cyclopropane derivatives 1 and 3. A commercially
available 2-bromo-4-cyanophenol 14 was converted into
vinyl ether 15 with an
iridiumcatalyst. The vinyl ether moiety was coupled with tosylhydrazone 22 separately prepared from aldehyde 20 with
a rhodium catalyst to afford cyclopropane derivative 16. The methylpyrazole moiety was introduced by Suzuki reaction with
boronic acid pinacol ester 25, followed by the deprotection
of the THP group to afford racemiccyclopropane compound 17. Enantiomers 18 (tR1) and 19 (tR2) were
obtained by a preparative chiral high-performance liquid chromatography
(HPLC) separation of racemiccompound 17 using a CHIRALCEL
OJ column. The three free compounds, 17, 18, and 19, were converted to hydrochloride salts 5, 1, and 3, respectively. The chiral
aminobenzylcompound 4 was prepared from the chiral compound 18via reduction of the cyano group. The
tosylhydrazone 22 was prepared from commercially available
benzaldehyde 20 and tosylhydrazine 21. The
pyrazole boronic ester 25 was prepared in a two-step
reaction: THP protection and regioselective borylation from the commercially
available material 23.
Scheme 1
Synthesis of 4-Cyanophenoxy-cis-1,2-disubstituted
Cyclopropane Derivatives
Reagents
and conditions: (a) vinyl acetate, [Ir(1,5-cod)Cl]2, Na2CO3, toluene, 100 °C, 92%; (b) 22, Rh2(OAc)4, LHMDS, BnEt3NCl, dioxane,
40 °C, and 30%; (c) 25, Pd(PPh3)4, Na2CO3, DME–H2O,
and reflux; (d) 4 N HCl/EtOAc, EtOAc, 0 °C, and 84% in two steps;
(e) optical resolution; (f) LAH, THF, 50 °C, and 89%; (g) MeOH,
rt, and 89%; (h) dihydropyrane, TFA, toluene, reflux, and 96%; and
(i) n-BuLi, triisopropyl borate, THF, −78
°C, then pinacol, acetic acid, rt, and 67%.
Synthesis of 4-Cyanophenoxy-cis-1,2-disubstituted
Cyclopropane Derivatives
Reagents
and conditions: (a) vinyl acetate, [Ir(1,5-cod)Cl]2, Na2CO3, toluene, 100 °C, 92%; (b) 22, Rh2(OAc)4, LHMDS, BnEt3NCl, dioxane,
40 °C, and 30%; (c) 25, Pd(PPh3)4, Na2CO3, DME–H2O,
and reflux; (d) 4 NHCl/EtOAc, EtOAc, 0 °C, and 84% in two steps;
(e) optical resolution; (f) LAH, THF, 50 °C, and 89%; (g) MeOH,
rt, and 89%; (h) dihydropyrane, TFA, toluene, reflux, and 96%; and
(i) n-BuLi, triisopropyl borate, THF, −78
°C, then pinacol, aceticacid, rt, and 67%.Scheme shows the
synthesis of 4-cyanophenoxy-cis-1,2,2-trisubstituted
cyclopropane derivatives 6–11. Vinyl
ether 15 was
converted to the trisubstituted cyclopropane 26 by diazocompound 34 and a catalytic amount of rhodium acetate.
The methylpyrazole moiety was introduced by Suzuki reaction with pyrazole
boronicester 25 to afford the intermediate 27. The ester 27 was selectively reduced to alcohol intermediate 28 by calcium borohydride generated in situ under the coexistence of the cyano group. The hydroxy group of 28 was methylated by sodium hydride and iodomethane to afford
methoxyintermediate 29. The primary amine 30 was prepared by Mitsunobu reaction with phthalimide, followed by
deprotection with hydrazine. Secondary amine 31 and tertiary
amine 32 were prepared from alcohol intermediate 28 in two steps: (1) Dess–Martin oxidation and (2)
reductive amination. The deprotection reaction of the THP group of 27–32 by acid afforded 4-cyanophenoxy-cis-1,2,2-trisubstituted cyclopropane derivatives 6–11. The diazo compound 34 was prepared from 2,4-difluorophenyl acetate 33 by
diazo transfer reaction under basicconditions.
Scheme 2
Synthesis of 4-Cyanophenoxy-cis-1,2,2-trisubstituted
Cyclopropane Derivatives
Reagents and conditions: (a) 34, Rh2(OAc)4, CH2Cl2, 45 °C,
and 49%; (b) 25, Pd(PPh3)4, Na2CO3, DME–H2O, reflux, and quant.;
(c) NaBH4, CaCl2, EtOH–THF, 0 °C
to rt, and 73%; (d) NaH, MeI, THF, rt, and 75%; (e) phthalimide, Ph3P, DIAD, THF, and 0 °C to rt; (f) hydrazine monohydrate,
EtOH, rt, and 89% in two steps; (g) Dess–Martin reagent, MeCN,
and 0 °C; (h) amine, NaBH(OAc)3, TEA, MeOH, rt, and
69–100% in two steps; (i) 4 N HCl/EtOAc or 2 N HCl/MeOH, MeOH,
0 °C, and 49–98%; and (j) 4-AcNHPhSO2N3, DBU, MeCN, 0 °C to rt, and 94%.
Synthesis of 4-Cyanophenoxy-cis-1,2,2-trisubstituted
Cyclopropane Derivatives
Reagents and conditions: (a) 34, Rh2(OAc)4, CH2Cl2, 45 °C,
and 49%; (b) 25, Pd(PPh3)4, Na2CO3, DME–H2O, reflux, and quant.;
(c) NaBH4, CaCl2, EtOH–THF, 0 °C
to rt, and 73%; (d) NaH, MeI, THF, rt, and 75%; (e) phthalimide, Ph3P, DIAD, THF, and 0 °C to rt; (f) hydrazine monohydrate,
EtOH, rt, and 89% in two steps; (g) Dess–Martin reagent, MeCN,
and 0 °C; (h) amine, NaBH(OAc)3, TEA, MeOH, rt, and
69–100% in two steps; (i) 4 NHCl/EtOAc or 2 NHCl/MeOH, MeOH,
0 °C, and 49–98%; and (j) 4-AcNHPhSO2N3, DBU, MeCN, 0 °C to rt, and 94%.Scheme shows the
synthesis of 4-trifluoromethylphenoxy-cis-1,2,2-trisubstitutedcyclopropane derivatives 12 and 13. The
vinyl ether intermediate 36 was prepared by the same
method, as shown in Scheme . The cyclopropanation followed
by the introduction of the pyrazole moiety was conducted under the
same condition in Scheme to afford the cyclopropyl ester intermediate 38. The ester was reduced to afford the alcohol 39 by
lithium aluminum hydride (LAH). The racemic primary amino compound 40 was synthesized through the same reactions in Scheme . The optical resolution
of compound 40 afforded the chiral amine 41 and its enantiomer 42. The chiral amine 41 was converted to salt form 12 by hydrogen chloride.
The sulfonamide 13 was prepared by sulfonylation reaction
from the chiral amino compound 41.
Scheme 3
Synthesis of 4-Trifluoromethylphenoxy-cis-1,2,2-trisubstituted
Cyclopropane Derivatives
Reagents and conditions: (a)
vinyl acetate, [Ir(1,5-cod)Cl]2, Na2CO3, toluene, 100 °C, and
53%; (b) 34, Rh2(OAc)4, Et2O, rt, and 62%; (c) 25, Pd(PPh3)4, Na2CO3, DME–H2O,
reflux, and 80%; (d) LAH, THF, 0 °C, and quant.; (e) phthalimide,
Ph3P, DIAD, THF, and 0 °C to rt.; (f) hydrazine monohydrate,
EtOH, and rt.; (g) 2 N HCl/MeOH, EtOAc, rt, and 38% in three steps;
(h) optical resolution; (i) 4 N HCl/EtOAc, MeOH, rt, and 97%; and
(j) p-NsCl, pyridine, rt, and 54%.
Synthesis of 4-Trifluoromethylphenoxy-cis-1,2,2-trisubstituted
Cyclopropane Derivatives
Reagents and conditions: (a)
vinyl acetate, [Ir(1,5-cod)Cl]2, Na2CO3, toluene, 100 °C, and
53%; (b) 34, Rh2(OAc)4, Et2O, rt, and 62%; (c) 25, Pd(PPh3)4, Na2CO3, DME–H2O,
reflux, and 80%; (d) LAH, THF, 0 °C, and quant.; (e) phthalimide,
Ph3P, DIAD, THF, and 0 °C to rt.; (f) hydrazine monohydrate,
EtOH, and rt.; (g) 2 NHCl/MeOH, EtOAc, rt, and 38% in three steps;
(h) optical resolution; (i) 4 NHCl/EtOAc, MeOH, rt, and 97%; and
(j) p-NsCl, pyridine, rt, and 54%.
Conclusions
Several
ALK inhibitors were developed and
approved as potential treatments for ALK-positive cancers such as
NSCLC. These studies began with hit compounds with relatively large
molecular weights, identified from their own corporate libraries,
and utilized SBDD approaches for the optimization of activity and
selectivity. However, there was no report adopting a fragment-based
approach for an ALK program. In this study, we have described the
strategic design of a novel scaffold from small fragments and the
efficient improvement of both potency and kinase selectivity using
fragment information. We began our research with a high concentration
assay of the fragment library, and the subsequent linker design and
its docking analysis afforded a novel cis-1,2,2-trisubstituted
cyclopropane structure with potent ALK inhibitory activity. Furthermore,
the fragment information was integrated with a structure-based approach
to overcome the selectivity over TrkA. The resulting compound 12 showed potent ALK inhibitory activity and remarkably high
kinase selectivity. However, the compound 12 traded off
the ALKcellular inhibitory activity against the selectivity over
TrkA. Further optimization to increase the cellularactivity, such
as an introduction of the solubilizing moiety like a piperidine of
crizotinib, and subsequent in vivo evaluation will
be required for developing a tool of the pharmacological evaluation.A fragment-based approach is often used to generate the lead compound
in the development of a new chemotype by connecting two fragments,
optimizing the fragment itself, or adding functional groups onto the
fragment. This work indicates that a fragment-based approach becomes
a powerful tool for selectivity, even in the optimization phase, by
combining it with an SBDD approach. This mixed approach could also
make the process simple and rational and, eventually, could contribute
to the development of a novel, potent, and highly selective compound,
in a short period, without synthesizing a large number of compounds.
This approach can be a useful tool for chemotype hopping in various
kinase projects in cases where there is a cocrystal structure of the
known ligand with the target protein.
Experimental Section
General
Methods
General Chemistry Information
All
solvents and reagents were obtained from commercial sources and were
used as received. Yields were not optimized. All reactions were monitored
by thin layer chromatography analysis on Merck Kieselgel 60 F254 plates
or Fuji Silysia NH plates or liquid chromatography–mass spectrometry
(LC–MS) analysis. LC–MS analysis was performed on a
Shimadzu liquid chromatography–mass spectrometer system, operating
in the APCI (+ or −) or ESI (+ or −) ionization mode.
Analytes were eluted using a linear gradient of 0.05% trifluoroaceticacid (TFA) containing water/acetonitrile or 5 mM ammonium acetatecontaining the water/acetonitrile mobile phase and detected at 220
nm. Column chromatography was carried out on silica gel [(Merck Kieselgel
60, 70–230 mesh, Merck) or (Chromatorex NH-DM 1020, 100–200
mesh, Fuji Silysia Chemical, Ltd.)] or on prepacked Purif-Pack columns
(SI or NH, particle size: 60 μm, Fuji Silysia Chemical, Ltd.).
Analytical HPLC was performed with Corona Charged Aerosol Detector
(CAD) or a photodiode array detector. The column was a Capcell Pak
C18AQ (50 mm × 3.0 mm i.d., Shiseido, Japan) or L-column 2 ODS
(30 mm × 2.0 mm i.d., CERI, Japan) with a temperature of 50 °C
and a flow rate of 0.5 mL/min. Mobile phases A and B under neutral
conditions were a mixture of 50 mmol/L ammonium acetate, water, and
acetonitrile (1:8:1, v/v/v) and a mixture of 50 mmol/L ammonium acetate
and acetonitrile (1:9, v/v), respectively. The ratio of mobile phase
B was increased linearly from 5 to 95% over 3 min, 95% over the next
1 min. Mobile phases A and B under acidicconditions were a mixture
of 0.2% formic acid in 10 mmol/L ammonium formate and 0.2% formicacid in acetonitrile, respectively. The ratio of mobile phase B was
increased linearly from 14 to 86% over 3 min and 86% over the next
1 min. Elemental analyses were carried out by Takeda Analytical Laboratories,
and the results were within ±0.4% of theoretical values. All
final test compounds were purified to >95% chemical purity as measured
by analytical HPLC. Proton nuclear magnetic resonance (1HNMR) spectra were in all cases consistent with the proposed structures. 1HNMR spectra were recorded on a Varian Mercury-300 (300 MHz)
or a Bruker DPX300 (300 MHz) instrument. All proton shifts are given
in parts per million (ppm) downfield from tetramethysilane (δ)
as the internal standard in a deuterated solvent, and coupling constants
(J) are in hertz (Hz). NMR data are reported as follows:
chemical shift, integration, multiplicity (s, singlet; d, doublet;
t, triplet; q, quartet; quint, quintet; m, multiplet; dd, doublet
of doublets; td, triplet of doublets; ddd, doublet of doublet of doublets;
and brs, broad singlet), and coupling constants.
The title compound was prepared in 41% yield as a colorless
crystal from 19 using the procedure analogous to that
described for the synthesis of 1. 1HNMR (300
MHz, DMSO-d6): δ 1.38–1.55
(2H, m), 2.09 (3H, s), 2.52–2.57 (1H, m), 4.35 (1H, td, J = 6.3, 3.3 Hz), 5.44 (1H, d, J = 0.6
Hz), 6.89–6.97 (1H, m), 7.21–7.36 (2H, m), 7.51 (1H,
d, J = 8.7 Hz), 7.75 (1H, dd, J =
8.6, 2.2 Hz), 8.10 (1H, d, J = 2.3 Hz), 2H not detected.
MS (ESI/APCI) m/z: 352.0 [M + H]+. mp 178–184 °C; Anal. Calcd for C20H16N3OClF2: C, 61.94; H, 4.16; N,
10.84. Found: C, 61.97; H, 4.20; N, 10.78. HPLC purity: 100%.
To a solution
of 18 (106 mg, 0.30 mmol) in tetrahydrofuran (THF) (dry)
(3 mL) was added LAH (45.8 mg, 1.21 mmol) at room temperature. The
mixture was stirred at 50 °C under the Ar atmosphere for 2 h.
The mixture was cooled to room temperature, diluted with THF, and
then quenched with water (0.046 mL), 15% NaOH aq (0.046 mL), and water
(0.138 mL). The mixture was stirred overnight and filtered by a Celite
pad. The filtrate was concentrated in vacuo to afford 4 (95.0 mg, 0.268 mmol, 89%) as a pale yellow gum. 1HNMR (300 MHz, DMSO-d6): δ 1.22–1.32
(1H, m), 1.39–1.49 (1H, m), 1.79 (2H, dd, J = 19.5, 9.7 Hz), 2.07 (3H, s), 2.39–2.48 (1H, m), 3.31 (2H,
s), 4.11–4.21 (1H, m), 5.48 (1H, br s), 6.93 (1H, t, J = 8.5 Hz), 7.15–7.36 (4H, m), 7.69 (1H, br s),
12.35 (1H, br s). MS (ESI/APCI) m/z: 356.2 [M + H]+.
The title compound was prepared in 71% yield as colorless
crystals from 17 using the procedure analogous to that
described for the synthesis of 1. 1HNMR (300
MHz, DMSO-d6): δ 1.37–1.56
(2H, m), 2.09 (3H, s), 2.52–2.55 (1H, m), 4.36 (1H, td, J = 6.2, 3.4 Hz), 5.44 (1H, s), 6.93 (1H, td, J = 8.4, 2.3 Hz), 7.20–7.39 (2H, m), 7.51 (1H, d, J = 8.6 Hz), 7.75 (1H, dd, J = 8.6, 2.2 Hz), 8.10
(1H, d, J = 2.2 Hz), 2H not detected. MS (ESI/APCI) m/z: 352.0 [M + H]+. mp 198–206
°C. HPLC purity: 100%.
To a solution of 41 (51.4
mg, 0.12 mmol) in pyridine (1 mL) was added 4-nitrobenzenesulfonyl
chloride (28.3 mg, 0.13 mmol) at room temperature. After stirring
for 2 h, the mixture was purified by column chromatography (NH-silica
gel, hexane/ethyl acetate, 40:60 to 0:100) to afford 13 (39.6 mg, 0.065 mmol, 54%) as (a) colorless crystals. 1HNMR (300 MHz, DMSO-d6): δ 1.14–1.20
(1H, m), 1.49 (1H, t, J = 6.2 Hz), 2.06 (3H, s),
2.83 (1H, dd, J = 14.0, 5.3 Hz), 3.54 (1H, dd, J = 13.8, 7.4 Hz), 4.34 (1H, dd, J = 5.9,
3.2 Hz), 5.21 (1H, s), 6.96 (1H, td, J = 8.6, 2.5
Hz), 7.19 (1H, ddd, J = 11.1, 9.0, 2.3 Hz), 7.27–7.38
(1H, m), 7.50–7.58 (1H, m), 7.59–7.69 (1H, m), 7.87–7.95
(2H, m), 8.02 (1H, br s), 8.26–8.40 (3H, m), 12.52 (1H, br
s). MS (ESI/APCI) m/z: 608.9 [M
+ H]+.
3-Bromo-4-(vinyloxy)benzonitrile
(15)
A mixture of 14 (5.20 g, 26.3
mmol), vinyl acetate (7.26 mL, 78.8 mmol), iridium (I) cyclooctadienechloride (0.176 g, 0.26 mmol), and Na2CO3 (1.67
g, 15.8 mmol) in toluene (25 mL) was stirred at 100 °C overnight.
The solvent was removed under reduced pressure. The residue was dissolved
in EtOAc and washed with water. The separated aqueous layer was extracted
with EtOAc again. The combined organic layers were washed with brine,
dried over MgSO4, and concentrated in vacuo. The residue was purified by column chromatography (silica gel,
hexane/ethyl acetate, 100:0 to 80:20) to afford 15 (5.40
g, 24.1 mmol, 92%) as a pale yellow solid. 1HNMR (300
MHz, DMSO-d6): δ 4.77 (1H, dd, J = 6.1, 1.9 Hz), 4.98 (1H, dd, J = 13.4,
2.1 Hz), 7.00 (1H, dd, J = 13.4, 5.9 Hz), 7.36 (1H,
d, J = 8.3 Hz), 7.86–7.93 (1H, m), 8.26 (1H,
d, J = 1.9 Hz).
To a solution of 22 (0.388
g, 1.25 mmol) in dry THF (5 mL) was added LHMDS (1.31 mL, 1.31 mmol)
at −78 °C under the Ar atmosphere. The mixture was stirred
at the same temperature for 15 min, gradually warmed to room temperature,
and concentrated in vacuo. The residue was mixed
with 1,4-dioxane (7 mL). To the resulted slurry was added 15 (1.12 g, 5.00 mmol), benzyltriethylammonium chloride (0.028 g, 0.13
mmol), and rhodium(II) acetate dimer (0.055 g, 0.13 mmol) at room
temperature. The mixture was stirred at 40 °C under Ar overnight.
The mixture was added water and extracted with EtOAc twice. The combined
organic layers were washed with brine, dried over Na2SO4, and concentrated in vacuo. The residue
was purified by column chromatography (silica gel, hexane/ethyl acetate,
99:1 to 88:12) to afford 16 (131 mg, 0.374 mmol, 30%)
as a colorless oil. 1HNMR (300 MHz, CDCl3):
δ ppm 1.40 (1H, td, J = 7.4, 3.3 Hz), 1.48–1.57
(1H, m), 2.48–2.60 (1H, m), 4.17 (1H, td, J = 6.3, 3.4 Hz), 6.63–6.78 (2 H, m), 7.09 (1H, td, J = 8.5, 6.3 Hz), 7.20 (1H, d, J = 8.6
Hz), 7.50 (1H, dd, J = 8.5, 2.0 Hz), 7.67 (1H, d, J = 2.0 Hz).
A
mixture of 16 (126 mg, 0.36 mmol), 25 (631
mg, 2.16 mmol), Pd(Ph3P)4 (125 mg, 0.11 mmol),
and Na2CO3 (0.360 mL, 0.72 mmol) in 1,2-dimethoxyethane
(DME) (3.6 mL) was stirred under the Ar atmosphere at 80 °C for
11 h. The mixture was diluted with water and extracted with EtOAc
twice. The combined organic layers were washed with brine, dried over
Na2SO4, and concentrated in vacuo. The residue was purified by column chromatography (silica gel,
hexane/ethyl acetate, 95:5 to 75:25) to afford a THP-protected intermediate.
This material was dissolved in MeOH and was added 4 NHCl in EtOAc
(1 mL, 4.00 mmol) at room temperature, and then the reaction mixture
was concentrated in vacuo. The resulted solid was
dissolved in water and sat. NaHCO3 aq and then extracted
with EtOAc twice. The combined organic layers were washed with brine,
dried over Na2SO4, and concentrated in vacuo. The residue was purified by column chromatography
(silica gel, hexane/ethyl acetate, 95:5 to 60:40) to afford 17 (107 mg, 0.304 mmol, 84%) as a pale yellow oil. 1HNMR (300 MHz, DMSO-d6): δ 1.35–1.56
(2H, m), 2.09 (3H, s), 2.52–2.59 (1H, m), 4.35 (1H, d, J = 3.1 Hz), 5.32–5.62 (1H, m), 6.85–7.00
(1H, m), 7.19–7.39 (2H, m), 7.50 (1H, d, J = 8.6 Hz), 7.74 (1H, d, J = 8.6 Hz), 8.00–8.18
(1H, m), 12.45–12.71 (1H, m). MS (ESI/APCI) m/z: 352.1 [M + H]+.
Optical
Resolution of 17
Compound 17 (100
mg, 0.304 mmol) was separated by preparative
HPLC (CHIRALCEL OJ, 50 mmID × 500 mmL, mobile phase: hexane/ethanol
700/300, 80 mL/min, UV220 nm) to afford 18 (38.8 mg,
45%) as a white solid and 19 (43.1 mg, 42%) as a white
solid. 18: tR = 6.75 min
(CHIRALPAK AD, 4.6 mmID × 250 mmL, mobile phase: hexane/2-propanol,
1.0 mL/min, UV220 nm); 99.7% ee. 19: tR = 8.86 min (CHIRALCEL OJ, 4.6 mmID × 250 mmL, mobile
phase: hexane/ethanol 700/300, 1.0 mL/min, UV220 nm); 99.5% ee. Optical
rotation was not determined.
A mixture of aldehyde 20 (1.421
g, 10 mmol) and tosylhydrazine 21 (2.049 g, 11.00 mmol)
and in MeOH (30 mL) was stirred at room temperature overnight and
then concentrated in vacuo. The residue was crystallized
in hexane–EtOAc. The crystals were collected by filtration
and dried in vacuo to afford 22 (2.77
g, 8.93 mmol, 89%) as a colorless crystal. 1HNMR (300
MHz, DMSO-d6): δ 2.37 (3H, s), 7.13
(1H, td, J = 8.6, 2.4 Hz), 7.32 (1H, ddd, J = 11.3, 9.1, 2.5 Hz), 7.41 (2H, d, J =
7.9 Hz), 7.67–7.82 (3H, m), 8.02 (1H, s), 11.63 (1H, br s).
To a mixture of 3-methyl-1H-pyrazole 23 (25.0 g, 304 mmol) in toluene
(152 mL) were added 3,4-dihydro-2H-pyrane (30.6 mL,
335 mmol) and TFA (1.17 mL, 15.2 mmol)
at room temperature. The mixture was stirred under reflux for 1 day.
The mixture was diluted with brine and NaHCO3 aq at room
temperature, extracted with EtOAc, dried over MgSO4, filtered,
concentrated in vacuo, and purified by column chromatography
(silica gel, hexane/ethyl acetate, 100:0 to 50:50) to yield 24 (48.5 g, 292 mmol, 96%) as a yellow oil. 1HNMR (300 MHz, DMSO-d6): δ 1.40–1.73
(3H, m), 1.78–2.11 (3H, m), 2.15 (3H, s), 3.52–3.69
(1H, m), 3.80–3.96 (1H, m), 5.26 (1H, dd, J = 10.4, 2.4 Hz), 6.06 (1H, d, J = 2.3 Hz), 7.70
(1H, d, J = 2.3 Hz).
To a solution of 24 (23.0 g, 138 mmol) in dry THF (173 mL) was added n-BuLi (91 mL, 145 mmol) dropwise at −78 °C.
The reaction mixture was stirred at the same temperature for 2 h.
To this solution was added triisopropyl borate (35.1 mL, 152 mmol)
dropwise at −78 °C. The reaction mixture was gradually
warmed to room temperature and stirred for 2 h. To this mixture was
added pinacol (18.0 g, 152 mmol) and aceticacid (16.0 mL, 277 mmol)
at the same temperature. The mixture was stirred overnight and then
concentrated in vacuo. The residue was diluted with
water and extracted with EtOAc twice. The combined organic layers
were washed with brine, dried over MgSO4, and concentrated in vacuo. The residue was purified by column chromatography
(silica gel, hexane/ethyl acetate, 100:0 to 85:15) to afford 25 (26.9 g, 92.0 mmol, 67%) as a yellow oil. 1HNMR (300 MHz, DMSO-d6): δ 1.28 (12H, d, J = 3.9 Hz), 1.41–1.74 (3H, m),
1.84 (1H, dd, J = 12.7, 2.4 Hz), 1.91–2.04
(1H, m), 2.16 (3H, s), 2.18–2.33 (1H, m), 3.45–3.59
(1H, m), 3.81–3.96 (1H, m), 5.61 (1H, dd, J = 10.1, 2.4 Hz), 6.42 (1H, s). MS (ESI/APCI) m/z: 293.3 [M + H]+.
The title compound was prepared in a quantitative
yield as a colorless amorphous powder from 26 using the
procedure analogous to that described for the synthesis of 17. 1HNMR (300 MHz, DMSO-d6): δ 1.33–1.70 (4H, m), 1.70–1.80 (1H, m), 1.80–1.95
(1H, m), 2.00–2.14 (4H, m), 3.21–3.38 (1H, m), 3.61
(3H, s), 3.77 (1H, d, J = 11.4 Hz), 4.68–4.86
(2H, m), 5.06–5.49 (1H, m), 6.84–7.19 (2H, m), 7.32
(2H, qd, J = 8.9, 6.5 Hz), 7.58–7.68 (2H,
m), 7.95 (1H, dd, J = 8.7, 2.2 Hz). MS (ESI/APCI) m/z: 494.2 [M + H]+.
To a suspension of calcium chloride (0.222
g, 2.00 mmol) in EtOH (1 mL) was added NaBH4 (0.151 g,
4.00 mmol) at room temperature. After stirring for 30 min, a solution
of 27 (0.493 g, 1.00 mmol) in dry THF (1.5 mL) was added
thereto at 0 °C. The reaction mixture was gradually warmed to
room temperature and stirred overnight. The reaction mixture was concentrated in vacuo. The residue was diluted with water and extracted
with EtOAc twice. The combined organic layers were washed with brine,
dried over MgSO4, and concentrated in vacuo. The residue was purified by column chromatography (silica gel,
hexane/ethyl acetate, 85:15 to 50:50) to afford 28 (0.337
g, 0.724 mmol, 73%) as a colorless amorphous powder. 1HNMR (300 MHz, DMSO-d6): δ 0.93–1.03
(1H, m), 1.30–1.71 (5H, m), 1.83 (1H, d, J = 14.2 Hz), 2.00–2.32 (4H, m), 3.19–3.37 (1H, m),
3.40–3.64 (2H, m), 3.76 (1H, d, J = 10.7 Hz),
4.23 (1H, ddd, J = 9.6, 6.2, 3.1 Hz), 4.63–4.86
(1H, m), 4.86–4.95 (1H, m), 5.10–5.61 (1H, m), 6.77–7.15
(2H, m), 7.16–7.32 (1H, m), 7.57–7.70 (2H, m), 7.91–8.00
(1H, m). MS (ESI/APCI) m/z: 466.3
[M + H]+.
To a solution of 28 (100 mg,
0.21 mmol) in dry THF (2 mL) was added NaH (12.9 mg, 0.32 mmol) at
0 °C. After stirring for 15 min, iodomethane (0.040 mL, 0.64
mmol) was added thereto at the same temperature. The reaction mixture
was gradually warmed to room temperature, stirred for 1 h, and then
concentrated in vacuo. The residue was diluted with
water and extracted with EtOAc twice. The combined organic layers
were washed with sat. brine, dried over MgSO4, and concentrated in vacuo. The residue was purified by column chromatography
(silica gel, hexane/ethyl acetate, 85:15 to 70:30) to afford 29 (77.0 mg, 0.160 mmol, 75%) as a colorless gum. 1HNMR (300 MHz, DMSO-d6): δ 1.01–1.14
(1H, m), 1.30–1.95 (6H, m), 2.01–2.36 (4H, m), 3.15–3.22
(3H, m), 3.22–3.38 (2H, m), 3.49–3.65 (1H, m), 3.76
(1H, d, J = 11.0 Hz), 4.23–4.36 (1H, m), 4.62–4.88
(1H, m), 5.10–5.59 (1H, m), 6.77–6.99 (1H, m), 6.99–7.15
(1H, m), 7.25 (1H, dtd, J = 12.7, 8.7, 6.8 Hz), 7.59
(1H, dd, J = 8.7, 1.8 Hz), 7.65 (1H, dd, J = 3.7, 2.2 Hz), 7.96 (1H, dt, J = 8.7,
1.8 Hz). MS (ESI/APCI) m/z: 480.2
[M + H]+.
To a solution of 28 (0.465
g, 1.00 mmol), isoindole-1,3-dione (0.220 g, 1.50 mmol) and PPh3 (0.393 g, 1.50 mmol) in dry THF (5 mL) was added diisopropyl
azodicarboxylate (DIAD) (0.789 mL, 1.50 mmol) at 0 °C. The reaction
mixture was gradually warmed to room temperature, stirred for 2 h,
and then concentrated in vacuo. The residue was purified
by column chromatography (silica gel, hexane/ethyl acetate, 90:10
to 65:35) to afford a phthalimide intermediate (0.544 g, 0.915 mmol,
92%) as a colorless amorphous powder. The intermediate was dissolved
in EtOH (10 mL) and added hydrazine monohydrate (0.444 mL, 9.15 mmol)
at room temperature. The mixture was stirred at 60 °C for 3 h
and then diluted with diethyl ether adequately. The resulted slurry
was filtered by a Celite pad. The filtrate was concentrated in vacuo. The residue was purified by column chromatography
(NH-silica gel, hexane/ethyl acetate, 70:30 to 20:80) to afford 30 (380 mg, 0.818 mmol, 89%) as a colorless amorphous powder. 1HNMR (300 MHz, DMSO-d6): δ
0.90 (1H, ddd, J = 11.7, 6.3, 3.2 Hz), 1.28–1.74
(7H, m), 1.74–1.94 (1H, m), 2.00–2.32 (4H, m), 2.56–2.67
(1H, m), 2.69–2.90 (1H, m), 3.19–3.41 (1H, m), 3.77
(1H, t, J = 8.5 Hz), 4.25 (1H, ddd, J = 11.1, 6.2, 3.3 Hz), 4.59–4.91 (1H, m), 5.07–5.59
(1H, m), 6.78–7.15 (2H, m), 7.15–7.31 (1H, m), 7.55–7.68
(2H, m), 7.89–8.00 (1H, m). MS (ESI/APCI) m/z: 465.2 [M + H]+.
To a solution of 28 (0.372
g, 0.80 mmol) in CH3CN (4.00 mL) was added Dess–Martin
periodinane (0.441 g, 1.04 mmol) at 0 °C. After stirring for
1 h, the reaction was quenched with sat. Na2S2O3 aq at room temperature. The mixture was quenched with
EtOAc twice. The combined organic layers were washed with sat. NaHCO3 aq and sat. brine, dried over MgSO4, and concentrated in vacuo. The residue was purified by column chromatography
(silica gel, hexane/ethyl acetate, 90:10 to 70:30) to afford an aldehyde
intermediate (0.346 g) as a colorless amorphous powder. To a mixture
of aldehyde intermediate (0.250 g, 0.540 mmol) and methylamine (0.275
mL, 2.70 mmol) in MeOH (2.5 mL) was added NaBH(OAc)3 (0.572
g, 2.70 mmol) at room temperature. After stirring overnight, the mixture
was concentrated in vacuo. The residue was partitioned
between EtOAc and water. The separated aqueous layer was extracted
with EtOAc again. The combined organic layers were washed with sat.
NaHCO3 aq and sat. brine, dried over MgSO4,
and concentrated in vacuo. The residue was purified
by column chromatography (NH-silica gel, hexane/ethyl acetate, 90:10
to 70:30) to afford 31 (0.192 g, 0.400 mmol, 69% in two
steps) as a yellow gum. 1HNMR (300 MHz, DMSO-d6): δ 0.98 (1H, ddd, J = 16.9,
6.3, 3.4 Hz), 1.30–1.73 (5H, m), 1.73–2.39 (9H, m),
2.39–2.47 (1H, m), 2.74–2.93 (1H, m), 3.14–3.44
(1H, m), 3.77 (1H, t, J = 9.4 Hz), 4.14–4.30
(1H, m), 4.62–4.89 (1H, m), 5.05–5.57 (1H, m), 6.76–7.14
(2H, m), 7.14–7.32 (1H, m), 7.59 (1H, d, J = 8.8 Hz), 7.64 (1H, dd, J = 3.9, 2.1 Hz), 7.89–8.00
(1H, m). MS (ESI/APCI) m/z: 479.3
[M + H]+.
The title compound was prepared in a quantitative
yield as a colorless amorphous powder from 28 using the
procedure analogous to that described for the synthesis of 31. 1HNMR (300 MHz, DMSO-d6): δ 1.09–1.53 (6H, m), 1.60–1.92 (2H, m), 2.01–2.16
(10H, m), 2.76–3.07 (1H, m), 3.21–3.41 (1H, m), 3.78
(1H, t, J = 11.4 Hz), 4.10–4.24 (1H, m), 4.63–4.90
(1H, m), 4.99–5.53 (1H, m), 6.74–7.13 (2H, m), 7.19–7.37
(1H, m), 7.57 (1H, d, J = 8.7 Hz), 7.64 (1H, dd, J = 5.4, 2.2 Hz), 7.88–7.98 (1H, m). MS (ESI/APCI) m/z: 493.3 [M + H]+.
Methyl
Diazo(2,4-difluorophenyl)acetate
(34)
To a solution of a benzylester 33 (2.78 g, 13.2 mmol) and 4-acetamidobenzenesulfonyl azide (3.48 g,
14.5 mmol) in CH3CN (35 mL) was added 1,8-diazabicyclo[5.4.0]undec-7-ene
(DBU) (2.36 mL, 15.8 mmol) at 0 °C. The reaction mixture was
gradually warmed to room temperature, stirred for 1 h, and concentrated in vacuo. The residue was dissolved in EtOAc and washed
with water. The separated aqueous layer was extracted with EtOAc again.
The combined organic layers were washed with sat. brine, dried over
MgSO4, and concentrated in vacuo. The
residue was purified by column chromatography (silica gel, hexane/ethyl
acetate, 100:0 to 90:10) to afford 34 (2.62 g, 12.4 mmol,
94%) as an orange oil. 1HNMR (300 MHz, DMSO-d6): δ 3.78 (3H, s), 7.22 (1H, tdd, J = 8.6, 8.6, 2.7, 1.1 Hz), 7.39 (1H, ddd, J = 11.5,
9.1, 2.7 Hz), 7.69 (1H, td, J = 8.9, 6.4 Hz).
The title compound was prepared in 80%
yield as a yellow gum from 37 using the procedure analogous
to that described for the synthesis of 17. 1HNMR (300 MHz, DMSO-d6): δ 1.32–1.58
(4H, m), 1.58–1.92 (2H, m), 2.00–2.18 (5H, m), 3.19–3.40
(1H, m), 3.59–3.63 (3H, m), 3.76–3.94 (1H, m), 4.61–4.86
(2H, m), 4.88–5.55 (1H, m), 6.81–7.25 (2H, m), 7.25–7.45
(1H, m), 7.56–7.74 (2H, m), 7.83 (1H, dd, J = 8.8, 2.2 Hz). MS (ESI/APCI) m/z: 537.2 [M + H]+.
The title compound was prepared in quantitative
yield as a colorless amorphous powder from 38 using the
procedure analogous to that described for the synthesis of 4. 1HNMR (300 MHz, DMSO-d6): δ 0.91–1.09 (1H, m), 1.29–1.93 (6H, m), 2.02–2.37
(4H, m), 3.15–3.67 (4H, m), 3.83 (1H, t, J = 11.9 Hz), 4.15–4.30 (1H, m), 4.59–4.86 (1H, m),
4.95–5.66 (1H, m), 6.77–7.39 (3H, m), 7.58–7.70
(2H, m), 7.83 (1H, dd, J = 8.8, 2.2 Hz). MS (ESI/APCI) m/z: 509.2 [M + H]+.
The title compound was prepared in 38%
yield in three steps as an orange amorphous solid from 39 using the procedures analogous to that described for the synthesis
of 30 and the deprotection of THP under the acidiccondition. 1HNMR (300 MHz, DMSO-d6): δ
1.02 (1H, dd, J = 5.9, 2.9 Hz), 1.53 (1H, t, J = 6.1 Hz), 2.10 (3H, s), 2.63 (1H, d, J = 13.2 Hz), 3.02 (1H, d, J = 13.0 Hz), 3.30 (2H,
s), 4.34 (1H, dd, J = 5.7, 3.0 Hz), 5.32 (1H, br
s), 7.06 (1H, td, J = 8.5, 2.1 Hz), 7.30 (1H, ddd, J = 10.8, 9.3, 2.5 Hz), 7.36–7.52 (1H, m), 7.54–7.71
(2H, m), 8.06 (1H, br s), 12.2–12.89 (1H, m). MS (ESI/APCI) m/z: 424.1 [M + H]+.
Optical
Resolution of 40
The compound 40 (68.5 mg, 0.162 mmol) was separated
by preparative HPLC (CHIRALPAK AD, 50 mmID × 500 mmL, mobile
phase: hexane/ethanol/diethylamine 800/200/1, 80 mL/min, UV254 nm)
to afford 41 (21.0 mg, 31%) as a colorless gum and 42 (29 mg, 42%) as a colorless gum. 41: tR = 7.10 min (CHIRALPAK AD, 4.6 mmID ×
250 mmL, mobile phase: hexane/ethanol/diethylamine 800/200/1, 1.0
mL/min, UV254 nm); 99.9% ee. 42: tR = 10.2 min (CHIRALPAK AD, 4.6 mmID × 250 mmL, mobile
phase: hexane/ethanol/diethylamine 800/200/1, 1.0 mL/min, UV254 nm);
99.9% ee. Optical rotation was not determined.
Calculation
Conditions
The docking study
was performed using the protein structure of the cocrystal structure
of ALK in the Protein Data Bank (PDB ID 2XP2). The protein structure was prepared
by the Protein Preparation Workflow in Maestro 11.5 (version 11.5.001,
Schrödinger, New York, USA) remaining all crystallographicwaters and the disorder loops were constructed by Prime. The 3D structures
of designed compounds from D-1 to D-17 were
built with the default parameters by LigPrep 4.0 version. The docking
study was performed with the default parameters by using Glide SP
7.3 version. The grid-enclosing box was placed on the centroid of
the ligand based on the crystal structures of ALK. In this docking
study, the position of the methylpyrazole moiety was restricted to
one of the cocrystal structure of ALK in complex with F-1 within 0.5 Å rmsd tolerance of the reference core. The conformational
energy penalties were calculated by using the MacroModel module in
Maestro and the energy penalties indicate the difference of the ligand
potential energies between tightly restrained minimization and unrestraint
minimization of the docking pose in water. The binding energies of
the docking pose were calculated with the default parameters in Prime
MMGBSA 3.0 version.
Kinase Panel
Protocol
Activity-based kinase profiling was performed by
either radioactivity assays using [γ-33P] ATP or AlphaScreen
assays (PerkinElmer) with anti-phosphotyrosine antibody.[19] The kinase profiling was performed by KINOMEscan
Profiling Service of DiscoverX corporation.[20] Compound 13 was screened at 0.3 μM, and assay
results are described as % Ctrl.
In Vitro Inhibitory Activity against ALK and TrkA
Kinase
Protein Preparation
The expression plasmid for humanALK kinase domain (residues 1058–1620,
Genebank Accession no. NM_004304) was constructed in a pFastBacGST
vector (Thermo Fisher Scientific). Recombinant proteins of ALK were
prepared using a baculovirus expression system. The GST-tagged TrkA
kinase domain (residues 436–790, Genebank accession no. NM_001012331.1) was purchased from Carna Biosciences (Kobe, Japan).
Cell-Based
ALK Autophosphorylation Assay
The full-length cDNA encoding
humanALK (NM_004304) construct was
amplified by PCR and cloned into the pcDNA3.3 vector (Thermo Fisher
Scientific). Humanembryonic kidney293 (HEK293)cell line was cultured
with Dulbecco’s modified Eagle’s medium (Thermo Fisher
Scientific) containing 10% fetal bovine serum (FBS), 100 U/mL of penicillin-G,
and 100 μg/mL of streptomycin sulfate (Thermo Fisher Scientific).
For evaluation of compound inhibitory activity in cells, the suspended
HEK293cells were transfected with expression plasmids encoding the
ALK full-length construct using the Fugene HD reagent (Roche Diagnostics,
Basel, Switzerland) according to the manufacturer’s instructions.
The cell suspension was seeded into 96-well culture plates at 3.0
× 104 cells/well using Dulbecco’s modified
Eagle’s medium (DMEM) supplemented with 10% FBS and incubated
two overnights in a 5% CO2 atmosphere at 37 °C. After
incubation for two overnights, cells were treated with various concentrations
of test compounds for 60 min in 10% FBS containing DMEM. The cells
were then lysed by adding cell lysis buffer (Cell Signaling Technology,
Danvers, MA). Quantification of phopho-ALK (Tyr1604) was performed
by using Phospho-ALK (Tyr1604) Sandwich ELISA Kit (Cell Signaling
Technology) according to the manufacturer’s instructions. Percentage
inhibition was calculated from the signal intensity by using the following
formulaA: compound nontreatment, B: positive
control (10 μM NVP-TAE684) treatment, and X: test compound.The IC50 values and 95% CIs were estimated in the same
way as in vitro kinase assays.
In Vitro Kinase Assays
For ALK enzyme assays, the
HTRF detection system was utilized. The
kinase buffer consisted of 50 mM Tris-HCl (pH 7.5), 5 mM MgCl2, 5 mM MnCl2, 2 mM dithiothreitol (DTT), 0.01%
Tween20, and 0.01% bovine serum albumin (BSA). The optimal concentrations
of ATP and biotin-poly-GT for ALK enzyme assay were 1 μM and
2.5 μg/mL, respectively. After 10 min of preincubation of the
enzyme, poly-(GT)-biotin, and test compounds, the kinase reaction
was started by adding ATP and incubated for 60 min. The reaction was
terminated by adding the detection mixture containing crypate-labeled
PT66 (Cisbio, Codolet, France) and streptavidin-Xlent! (Cisbio) dissolved
in detection buffer (Cisbio). After incubation at room temperature
for 60 min, the plates were read with an EnVision 2102 Multilabel
Reader (PerkinElmer, Waltham, MA). Assays for TrkA enzyme were performed
using the AlphaScreen assay system (PerkinElmer). The kinase buffer
consisted of 50 mM Tris-HCl (pH 7.5), 5 mM MgCl2, 5 mM
MnCl2, 2 mM DTT, and 0.01% Tween20. After preincubation
of the enzyme, poly-(GT)-biotin, and compounds for 5 min at room temperature,
the kinase reaction was initiated by the addition of final 1 μM
ATP. After 60 min of incubation, the reaction was terminated by adding
the AlphaScreen phosphotyrosine (P-Tyr-100) acceptor and streptavidin
donor beads (PerkinElmer) in 62.5 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonicacid (pH 7.5) with 250 mM NaCl, 0.1% BSA, and 100 mM ethylenediaminetetraaceticacid (EDTA). The assay plates were incubated at room temperature in
the dark for more than 12 h and then read with an EnVision 2102 Multilabel
Reader. Percentage inhibition was calculated from the signal intensity
of either HTRF or AlphaScreen assay by using the following formulaA: total reaction, B: without enzyme,
and X: test compound.The IC50 values and 95% CIs
were estimated using a four-parameter logisticcurve using XLfit software
(IDBS, London, UK).
Molecular Biology,
Protein Purification, and Crystallization of human ALK
A
clone corresponding to the kinase catalytic domain region of humanALK kinase residues 1090 to 1406 was cloned into the pFastBacHTb vector
(Thermo Fisher Inc.). It contained an N-terminal 6X Histidine tag
with a cleavable TEV protease site and a polyhedron replication origin
site. It was subcultured in the Baculovirus insect cell strain SF9
(Spodoptera frugiperda, Thermo Fisher
Inc.) for large-scale expression and purification. Cells were lysed
by sonication, followed by centrifugation at 14,000 rpm for 1 h. The
supernatant was loaded on a nickel-charged NTAcolumn and purified
by linear gradient of increasing nickel sulfate (20–200 mM).
The eluate was checked for purity and treated with TEV protease. The
final step was size exclusion chromatography on a Superdex 75 column.
After confirmation of the purity of the fractions and their monodispersity,
protein was concentrated to 25 mg/mL in buffer containing 20 mM Tris
pH 8.0, 150 mM NaCl, 0.5 mM EDTA, 1 mM benzamidine, and 0.25 mM TCEP
and used for structural studies. Initial screening in the Hampton
PEG Ion screen yielded several hits both with apo protein and in the
presence of 1 mM concentration of compounds. Best diffracting crystals
appeared when optimized under the following conditions: 19–25%
PEG 3350, 2.0% Tacsimate pH 7.0, and a 7:3 or 8:2 ratio of Tris chloride:
Tris sodium (100 mM final concentration). Crystals were harvested
after cryoprotection with either 15% ethylene glycol or 15% glycerol
for 15 min.
Crystal Structure Determination
Data was collected at beamline 5.0.3 of the Advanced Light Source
at Berkeley, USA, and beamline X6A of the National Synchrotron Light
Source at Brookhaven, USA. The structure was solved by molecular replacement
using MOLREP,[21] a part of the CCP4 package.[22] The structure of ALK kinase (PDB ID 3L9P) was used as the
starting model. This was followed by iterative rounds of manual model
building in Coot[23] and restrained refinement
using Refmac5[24] and Phenix.Refine.[25] The ligand atoms and water molecules were fit
manually after interpreting the difference map and accounting for
the protein atoms. Crystallographic data collection and refinement
statistics are summarized in the Supporting Infomation.
Authors: J Jean Cui; Michelle Tran-Dubé; Hong Shen; Mitchell Nambu; Pei-Pei Kung; Mason Pairish; Lei Jia; Jerry Meng; Lee Funk; Iriny Botrous; Michele McTigue; Neil Grodsky; Kevin Ryan; Ellen Padrique; Gordon Alton; Sergei Timofeevski; Shinji Yamazaki; Qiuhua Li; Helen Zou; James Christensen; Barbara Mroczkowski; Steve Bender; Robert S Kania; Martin P Edwards Journal: J Med Chem Date: 2011-08-18 Impact factor: 7.446
Authors: Thomas H Marsilje; Wei Pei; Bei Chen; Wenshuo Lu; Tetsuo Uno; Yunho Jin; Tao Jiang; Sungjoon Kim; Nanxin Li; Markus Warmuth; Yelena Sarkisova; Frank Sun; Auzon Steffy; AnneMarie C Pferdekamper; Allen G Li; Sean B Joseph; Young Kim; Bo Liu; Tove Tuntland; Xiaoming Cui; Nathanael S Gray; Ruo Steensma; Yongqin Wan; Jiqing Jiang; Greg Chopiuk; Jie Li; W Perry Gordon; Wendy Richmond; Kevin Johnson; Jonathan Chang; Todd Groessl; You-Qun He; Andrew Phimister; Alex Aycinena; Christian C Lee; Badry Bursulaya; Donald S Karanewsky; H Martin Seidel; Jennifer L Harris; Pierre-Yves Michellys Journal: J Med Chem Date: 2013-06-26 Impact factor: 7.446
Authors: Iwan J P de Esch; Daniel A Erlanson; Wolfgang Jahnke; Christopher N Johnson; Louise Walsh Journal: J Med Chem Date: 2021-12-20 Impact factor: 7.446