Discovery of small molecule RIP1 kinase inhibitors for the treatment of pathologies associated with necroptosis

Article abstract of DOI:10.1021/ml400382p

Potent inhibitors of RIP1 kinase from three distinct series, 1-aminoisoquinolines, pyrrolo[2,3-b]pyridines, and furo[2,3-d]pyrimidines, all of the type II class recognizing a DLG-out inactive conformation, were identified from screening of our in-house ki

Full text of DOI:10.1021/ml400382p

Letter
pubs.acs.org/acsmedchemlett
Discovery of Small Molecule RIP1 Kinase Inhibitors for the Treatment
of Pathologies Associated with Necroptosis
Philip A. Harris,*,† Deepak Bandyopadhyay,‡ Scott B. Berger,† Nino Campobasso,‡ Carol A. Capriotti,†
Julie A. Cox,‡ Lauren Dare,† Joshua N. Finger,† Sandra J. Hoffman,† Kirsten M. Kahler,§ Ruth Lehr,‡
John D. Lich,† Rakesh Nagilla,† Robert T. Nolte,§ Michael T. Ouellette,‡ Christina S. Pao,‡
Michelle C. Schaeffer,† Angela Smallwood,‡ Helen H. Sun,† Barbara A. Swift,† Rachel D. Totoritis,‡
Paris Ward,‡ Robert W. Marquis,† John Bertin,† and Peter J. Gough†
‡
†Pattern Recognition Receptor DPU and Platform Technology & Science, GlaxoSmithKline, Collegeville Road, Collegeville,
Pennsylvania 19426, United States
§Platform Technology & Science, GlaxoSmithKline, Research Triangle Park, North Carolina 27709, United States
S
* Supporting Information
ABSTRACT: Potent inhibitors of RIP1 kinase from three
distinct series, 1-aminoisoquinolines, pyrrolo[2,3-b]pyridines,
and furo[2,3-d]pyrimidines, all of the type II class recognizing
a DLG-out inactive conformation, were identified from
screening of our in-house kinase focused sets. An exemplar
from the furo[2,3-d]pyrimidine series showed a dose propor-
tional response in protection from hypothermia in a mouse
model of TNFα induced lethal shock.
KEYWORDS: RIP1, type II kinase inhibitors, necroptosis
ecrosis is a highly inflammatory form of cell death that is
thought to be involved in the pathogenesis of a wide
kinase activity are not limited to cell death, as RIP1 kinase
activity has also been implicated as a direct driver of
proinflammatory cytokine production.14
Recently, Shi et al. have reported the cocrystal structures of
several necrostatins bound in a hydrophobic pocket between
the N- and C-lobes of the Rip1 kinase domain. These inhibitors
interact with highly conserved amino acids in the activation
loop and surrounding structural elements to stabilize RIP1 in
an inactive conformation.15
The reported Necrostatins have moderate potency and poor
pharmacokinetic properties rendering them unsuitable for
development as therapeutics. In order to identify novel
inhibitors of RIP1 kinase activity, we developed a fluorescence
polarization (FP) biochemical assay. Since expression of the full
length protein containing the death domain led to insoluble
protein, not suitable for assay development, we used the kinase
domain of RIP1 (1−375).
Screening of the GSK kinase inhibitor libraries identified a
number of inhibitors of RIP1 characteristic of the type II kinase
inhibitor class that target the inactive DFG-out conformation.16
This conserved DFG (Asp-Phe-Gly) sequence (or more rarely
D[LWY]G) is located immediately before the activation loop
and it adopts a different conformation in which the aspartate
and phenylalanine side chains change positions and point in
N
variety of human diseases.1 The inflammation induced by
necrotic cell death is primarily due to disruption of the plasma
membrane and the unregulated leakage of danger associated
molecular patterns (DAMPs), intracellular components that
potently activate the innate immune system when released into
the extracellular matrix.2 Historically, necrosis has been
considered an uncontrolled form of cell death that is refractory
to therapeutic intervention. However, more recent work has
shown that necrosis can be a programmed and tightly regulated
event, which offers a novel opportunity for the treatment of
diseases driven by necrotic cell death.
One form of orchestrated necrotic cell death, termed
programmed necrosis (or necroptosis), is induced by TNFα
and dependent upon RIP1 (receptor interacting protein 1)
kinase activity.3 This pathway was initially identified following a
cell based screen in which Degterev et al. identified a series of
small molecules, termed Necrostatins, that blocked the necrotic
death of human monocytic U937 cells induced by treatment
with TNFα and the caspase inhibitor zVAD.fmk.4 These were
subsequently identified as RIP1 kinase inhibitors5 and were
shown to have efficacy in animal models of ischemia/
reperfusion injury in the brain,6 retina,7 and kidney,8 as well
as models of myocardial infarction9 and retinal detachment.10
Recent data from genetically manipulated mouse models has
further highlighted the role for RIP1-dependent necroptosis as
a key driver of the pathogenesis of inflammation and disease in
the intestine and the skin.11−13 Additionally, the effects of RIP1
Received: September 25, 2013
Accepted: November 4, 2013
Published: November 4, 2013
© 2013 American Chemical Society
1238
dx.doi.org/10.1021/ml400382p | ACS Med. Chem. Lett. 2013, 4, 1238−1243

ACS Medicinal Chemistry Letters
Letter
opposite directions compared to the active “DFG-in”
orientation. The phenylalanine moves inward to partially
obstruct the ATP-binding site. This rearrangement creates a
new hydrophobic allosteric binding pocket, adjacent to the
ATP-binding site. In RIP1 this Asp-Phe-Gly sequence on the
activation loop is Asp-Leu-Gly (DLG).
Table 1. 1-Aminoisoquinolines
Some common features are present in these type II kinase
inhibitors as shown in Figure 1. They contain a heterocycle that
Figure 1. Binding modes of type II kinase inhibitors.
forms one or two hydrogen bonds with the kinase hinge residue
and a hydrophobic aryl moiety occupying the new allosteric
hydrophobic pocket. In addition, the inhibitors contain an aryl
urea (or amide) moiety linking the hinge-binding heterocycle
to the hydrophobic moiety, which makes additional hydrogen
bond interactions to an aspartate-backbone NH and a highly
conserved glutamate side chain of the α-C helix.17
Several series that are characteristic of type II kinase
inhibitors were identified from the GSK kinase screening sets;
whereas only a single type I series was identified with a very
poor kinase selectivity profile. The superior kinase selectivity
profile of the type II inhibitors is thought to be due to the
amino acid residues around the newly formed hydrophobic
binding pocket being less conserved compared to the residues
that form the ATP binding pocket.18 The type II series profiled
in Table 1 have a 1-aminoisoquinoline as the hinge binder and
a meta-(trifluoromethyl)phenyl as the hydrophobic aryl
moiety.19 The meta (trifluoromethyl)-phenyl is a common
moiety found in multiple type II kinase inhibitors, such as the
VEGFR2/b-Raf inhibitors Sorafenib and Regorafenib, and is
thought to target a conserved hydrophobic subpocket in the
allosteric site.17 In addition to the trifluoromethyl meta
substituent, fluoro-substitution around the aryl ring led to
modest changes in potency except at the 2-position (compound
5) where activity was largely lost. Substitution at the 6-position
by chlorine also led to a large drop off in potency (compound
7). However, replacement of the meta-(trifluoromethyl)-phenyl
with a tert-butylisoxazole (compound 8) showed equivalent
efficacy.
To measure cellular activity we examined the ability of our
active compounds to prevent the necrotic death of human
monocytic U937 cells induced by treatment with TNFα and
the caspase inhibitor zVAD.fmk, as described by Degterev et
al.4 In general the potency of the compounds in the cell based
assay were about 10−50-fold weaker than the biochemical
potencies. In this series, the optimal inhibitors in terms of both
biochemical and cellular potencies were compounds 1 and 8.
However, the moderate cellular potencies observed did not
support advancing this series into in vivo models.
To confirm the binding mode of this series in RIP1, we
obtained a cocrystal structure of compound 8 bound in RIP1
(1−324) as shown in Figure 2. Structure determination was
done by molecular replacement using the atomic coordinates of
In addition to the FP binding assay we also utilized an ADP-
Glo kinase activity assay, a luminescent kinase assay measuring
the ADP formed during autophosphorylation of RIP1. In
general the potency of the compounds in both the FP binding
assay and the ADP-Glo activity assay correlated reasonably well.
1239
dx.doi.org/10.1021/ml400382p | ACS Med. Chem. Lett. 2013, 4, 1238−1243

ACS Medicinal Chemistry Letters
Letter
Table 2. 5-Phenylpyrrolo[2,3-b]pyridines
Figure 2. Co-crystal structure of Rip1 (1−324) and 1-aminoisoquino-
line 8.
LIMK1 kinase domain (PDB code 4NEU) as a search model
(see Supporting Information for details). The X-ray structure of
the RIP1 kinase domain was refined to 2.6 Å resolution
(Figures 2 and S1−3 and Table S1, Supporting Information).
There are two molecules in the asymmetric unit referred to as
molecule A and B, respectively. Both chains are similar enough
for discussion of one chain. RIP1 (1−324) model has the
canonical kinase fold with an N-lobe and a C-lobe connected by
an intervening hinge. Residues 166−187 of the activation loop
and the N- and C-termini are disordered and not modeled. The
model starts at residue 7 and ends at residue 313.
The X-ray structure confirms compound 8 is a type II kinase
inhibitor and binds to a DLG-out inactive form of RIP1,
distinct than the inactive conformation observed for Rip1
cocrystallized with the necrostatins.15 The 1-aminoisoquinoline
heterocycle of 8 is a two point hinge binder and forms H-bonds
to the backbone N and CO of residue Met 95. The isoquinoline
is sandwiched between Tyr 94 and Ile43 on one side and
Leu145 on the opposite face. A bifurcated H-bond occurs
between the side chain of Glu63 and the urea portion of 8.
Glu63 is part of the c-helix, common to kinases. A fourth H-
bond occurs between the backbone amide of Asp156 and the
carbonyl oxygen of the urea. Asp156 is part of the DLG motif.
The isoxazole heterocycle sits deep in the allosteric pocket
created by the DLG-out conformation and partially defined by
the C-helix. The pocket is extremely hydrophobic and defined
by amino acids Met66, Met67 (C-helix), Leu70, Val75 (loop
following C-helix), Leu129, Val134 (catalytic loop), and
Leu159 (after DLG). The tert-butyl group resides in a mostly
hydrophobic subpocket comprising Leu70, Val75, Val76,
Leu129, Val134, His136, Ile154, and Ala155.
A second series identified from screening utilized the 5-
phenylpyrrolo[2,3-b]pyridine heterocycle as the hinge binder,
as shown in Table 2.20 The unsubstituted urea (compound 9)
had only weak potency, but small substitutions at the 3-position
(entries 10−14) gave up to 1000-fold increase in activity.
These small groups are presumably targeting the same
lipophilic subpocket where the tert-butyl group of the 1-
aminoisoquinolines series was shown to occupy in the cocrystal
structure shown in Figure 2. With regards to both enzymatic
and cellular potencies, the 3-trifluoromethyl substitution was
optimal. The nonaromatic cyclopentyl urea (compound 16)
had weak activity.
Table 3 shows the SAR resulting from substitution at the aryl
ring at the 5-position of the pyrrolo[2,3-b]pyridine, with a 5-
fluoro-3-(trifluoromethyl)phenyl ring at the urea occupying the
hydrophobic pocket. In general, activity was well maintained
with the meta-pyridyl (compound 18) giving excellent
potencies in both the biochemical ADP-Glo (4 nM) and
cellular U937 (6.3 nM) assays. Despite their excellent in vitro
potencies, both the pyrrolo[2,3-b]pyridines 18 and 20 had
minimal systemic oral exposures when examined in both rat
and mouse which would not support evaluation in an in vivo
model.
A third series we identified with a more promising
pharmokinetic profile was the furo[2,3-d]pyrimidines as
detailed in Table 4.21 The SAR around the urea shows similar
preference for small groups at the meta position as observed in
the previous series. As observed with the 1-aminoisoquinolines
series, fluoro substitution at the 2 position (compound 28) or
chloro substitution at 6 position (compound 30) were
detrimental to potency. Similar to compound 8 in the 1-
aminoisoquinoline series, incorporation of the tert-butylisox-
azole urea 35 was favorable and quite comparable to the 5-
fluoro-3-(trifluoromethyl)phenyl urea 27.
1240
dx.doi.org/10.1021/ml400382p | ACS Med. Chem. Lett. 2013, 4, 1238−1243

ACS Medicinal Chemistry Letters
Letter
Table 3. 5-Arylpyrrolo[2,3-b]pyridines
Table 4. Furo[2,3-d]pyrimidines
The kinase selectivity profiles of representative exemplars
from each of the series were obtained against 300 kinases using
a P33 radiolabeled assay at Reaction Biology Corporation
(RBC).22 Each compound was tested at 1 μM concentration in
duplicate against each kinase with 10 μM ATP and >50%
inhibition was defined as being active. RIP1 kinase was not
included in this RBC selectivity panel. The pyrrolo[2,3-
b]pyridine series showed the highest kinase selectivity, with
two exemplars 17 and 18 inhibiting 7 and 15 of the 300 kinases,
respectively. The furo[2,3-d]pyrimidine 27 inhibited 25 kinases
of the 300 asssayed at 1 μM. The least selective series was the
1-aminoisoquinolines, with compounds 1 and 8 showing
activity against 30 and 46 kinases, respectively.
The furo[2,3-d]pyrimidine 27 possessed good systemic
exposure in the mouse following oral administration with an
AUC of 14 7 μg·h/mL and a Cmax of 1100 ng/mL at 4 h
(Figure 3). Compound 27 was therefore selected for further
evaluation in vivo using a mouse model of TNFα induced lethal
shock. In this model, injection of TNFα leads to a systemic
inflammatory response, characterized by hypotension, hepatitis,
hypothermia, and bowel necrosis. Duprez et al. have reported
that the RIP1 kinase inhibitor necrostatin, administered
intravenously, showed significant protection from hypothermia
and death in this model.23 Compound 27 was dosed orally 15
min prior to TNFα injection and showed 15%, 22%, and 77%
protection from body temperature loss over 7 h, compared to
TNFα alone, at doses of 0.2, 2.0, and 20 mg/kg, respectively
1241
dx.doi.org/10.1021/ml400382p | ACS Med. Chem. Lett. 2013, 4, 1238−1243

ACS Medicinal Chemistry Letters
Letter
AUTHOR INFORMATION
■
Corresponding Author
*(P.A.H.) E-mail: philip.a.harris@gsk.com.
Author Contributions
All authors have given approval to the final version of the
manuscript.
Notes
All studies involving the use of animals were conducted after
review by the GlaxoSmithKline (GSK) Institutional Animal
Care and Use Committee and in accordance with the GSK
Policy on the Care, Welfare, and Treatment of Laboratory
Animals.
The authors declare the following competing financial interests:
All authors are current employees and stockholders of
GlaxoSmithKline.
Figure 3. Average ( standard deviation) blood concentrations of
compound 27 following oral administration (2.0 mg/kg) in the male
C57BL/6 mouse.
(Figure 4). Compound 27 has a relatively moderate molecular
weight (431.3) and ClogP (4.2) and good permeability (210
ABBREVIATIONS
■
RIP, receptor interacting protein; TNFα, tumor necrosis factor
alpha; DFG, aspartic acid-phenylalanine-glycine; DLG, aspartic
acid-leucine-glycine; zVAD.fmk, N-benzyloxycarbonyl-Val-Ala-
Asp(O-Me) fluoromethyl ketone; FP, fluorescence polar-
ization; VEGFR, vascular endothelial growth factor receptor;
b-RAF, rapidly accelerated fibrosarcoma; ADP, adenosine
diphosphate; ATP, adenosine triphosphate; AUC, area under
the curve
REFERENCES
■
(1) Iyer, S. S.; Pulskens, W. P.; Sadler, J. J.; Butter, L. M.; Teske, G. J.;
Ulland, T. K.; Eisenbarth, S. C.; Florquin, S.; Flavell, R. A.; Leemans, J.
C.; Sutterwala, F. S. Necrotic cells trigger a sterile inflammatory
response through the Nlrp3 inflammasome. Proc. Natl. Acad. Sci. U.S.A.
2009, 106, 20388−20393.
(2) Bianchi, M. E. DAMPs, PAMPs and alarmins: all we need to
know about danger. J. Leukocyte Biol. 2007, 81, 1−5.
(3) Galluzzi, L.; Kroemer, G. Necroptosis turns TNF lethal. Immunity
2011, 35, 849−851.
Figure 4. Evaluation of compound 27 in mouse TNFα induced lethal
shock model measuring reduction in body temperature loss.
nm/sec), but requires further lead optimization to address its
low solubility (13 μM) and high protein binding (0.04% free
fraction in mouse).
(4) Degterev, A.; Huang, Z.; Boyce, M.; Li, Y.; Jagtap, P.; Mizushima,
N.; Cuny, G. D.; Mitchison, T. J.; Moskowitz, M. A.; Yuan, J. Chemical
inhibitor of nonapoptotic cell death with therapeutic potential for
ischemic brain injury. Nat. Chem. Biol. 2005, 1, 112−119.
(5) Degterev, A.; Hitomi, J.; Germscheid, M.; Ch’en, I. L.; Korkina,
O.; Teng, X.; Abbott, D.; Cuny, G. D.; Yuan, C.; Wagner, G.; Hedrick,
S. M.; Gerber, S. A.; Lugovskoy, A.; Yuan, J. Identification of RIP1
kinase as a specific cellular target of necrostatins. Nat. Chem. Biol.
2008, 4, 313−321.
(6) Xu, X.; Chua, K. W.; Chua, C. C.; Liu, C. F.; Hamdy, R. C.; Chua,
B. H. Synergistic protective effects of humanin and necrostatin-1 on
hypoxia and ischemia/reperfusion injury. Brain Res. 2010, 1355, 189−
194.
In conclusion, we have identified 3 distinct type II inhibitors
of RIP1 kinase (1-aminoisoquinolines, pyrrolo[2,3-b]pyridines,
and furo[2,3-d]pyrimidines) with potent activity in both
biochemical and cellular assays. A cocrystal structure of a 1-
aminoisoquinoline in the RIP1 kinase domain shows the
inhibitor occupying the ATP binding pocket with the kinase
adopting a DLG-out conformation. An exemplar from the
furo[2,3-d]pyrimidine series showed excellent mouse exposure
dosed orally and gave a dose proportional response in
protection from hypothermia in a mouse model of TNFα
induced lethal shock. Further exploration of these promising
RIP1 kinase inhibitors is underway to determine their utility in
the clinic.
(7) Rosenbaum, D. M.; Degterev, A.; David, J.; Rosenbaum, P. S.;
Roth, S.; Grotta, J. C.; Cuny, G. D.; Yuan, J.; Savitz, S. I. Necroptosis, a
novel form of caspase-independent cell death, contributes to neuronal
damage in a retinal ischemia-reperfusion injury model. J. Neurosci. Res.
2010, 88, 1569−1576.
ASSOCIATED CONTENT
* Supporting Information
■
S
Details of the synthesis and spectroscopic characterization of all
compounds and intermediates, kinase selectivity data together
with protocols for biological experiments, and crystallographic
information. This material is available free of charge via the
Internet at http://pubs.acs.org.
(8) Linkermann, A.; Brasen, J. H.; Himmerkus, N.; Liu, S.; Huber, T.
̈
B.; Kunzendorf, U.; Krautwald, S. Rip1 (receptor-interacting protein
kinase 1) mediates necroptosis and contributes to renal ischemia/
reperfusion injury. Kidney Int. 2012, 81, 751−761.
(9) Smith, C. C.; Davidson, S. M.; Lim, S. Y.; Simpkin, J. C.;
Hothersall, J. S.; Yellon, D. M. Necrostatin: a potentially novel
cardioprotective agent? Cardiovasc. Drugs Ther. 2007, 21, 227−233.
(10) Trichonas, G.; Murakami, Y.; Thanos, A.; Morizane, Y.; Kayama,
M.; Debouck, C. M.; Hisatomi, T.; Miller, J. W.; Vavvas, D. G.
Receptor interacting protein kinases mediate retinal detachment-
Accession Codes
The atomic coordinates and structure factors of RIP1 bound to
compound 8 have been deposited in the Protein Data Bank
with accession code 4NEU.
1242
dx.doi.org/10.1021/ml400382p | ACS Med. Chem. Lett. 2013, 4, 1238−1243

ACS Medicinal Chemistry Letters
Letter
induced photoreceptor necrosis and compensate for inhibition of
apoptosis. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 21695−21700.
(11) Gunther, C.; Martini, E.; Wittkopf, N.; Amann, K.; Weigmann,
̈
B.; Neumann, H.; Waldner, M. J.; Hedrick, S. M.; Tenzer, S.; Neurath,
M. F.; Becker, C. Caspase-8 regulates TNF-α-induced epithelial
necroptosis and terminal ileitis. Nature 2011, 477, 335−339.
́
(12) Welz, P. S.; Wullaert, A.; Vlantis, K.; Kondylis, V.; Fernandez-
Majada, V.; Ermolaeva, M.; Kirsch, P.; Sterner-Kock, A.; van Loo, G.;
Pasparakis, M. FADD prevents RIP3-mediated epithelial cell necrosis
and chronic intestinal inflammation. Nature 2011, 477, 330−334.
(13) Bonnet, M. C.; Preukschat, D.; Welz, P. S.; van Loo, G.;
Ermolaeva, M. A.; Bloch, W.; Haase, I.; Pasparakis, M. The adaptor
protein FADD protects epidermal keratinocytes from necroptosis in
vivo and prevents skin inflammation. Immunity 2011, 35, 572−582.
(14) Lukens, J. R.; Vogel, P.; Johnson, G. R.; Kelliher, M. A.; Iwakura,
Y.; Lamkanfi, M.; Kanneganti, T. D. RIP1-driven autoinflammation
targets IL-1α independently of inflammasomes and RIP3. Nature
2013, 498, 224−227.
(15) Xie, T.; Peng, W.; Liu, Y.; Yan, C.; Maki, J.; Degterev, A.; Yuan,
J.; Shi, Y. Structural basis of RIP1 inhibition by necrostatins. Structure
2013, 21, 493−499.
(16) Kufareva, I.; Abagyan, R. Type-II kinase inhibitor docking,
screening, and profiling using modified structures of active kinase
states. J. Med. Chem. 2008, 51, 7921−7932.
(17) Liu, Y.; Gray, N. S. Rational design of inhibitors that bind to
inactive kinase conformations. Nat. Chem. Biol. 2006, 2, 358−364.
(18) Davis, M. I.; Hunt, J. P.; Herrgard, S.; Ciceri, P.; Wodicka, L. M.;
Pallares, G.; Hocker, M.; Treiber, D. K.; Zarrinkar, P. P.
Comprehensive analysis of kinase inhibitor selectivity. Nat. Biotechnol.
2011, 29, 1046−1051.
(19) Washio, Y. Preparation of Aryl Substituted Isoquinoline
Derivatives As Inhibitors of VEGFR2, TIE-2, and EphB4. PCT Int.
Appl. WO2005049576, 2005.
(20) Patnaik, S.; Stevens, K. L.; Gerding, R.; Deanda, F.; Shotwell, J.
B.; Tang, J.; Hamajima, T.; Nakamura, H.; Leesnitzer, M. A.; Hassell,
A. M.; Shewchuck, L. M.; Kumar, R.; Lei, H.; Chamberlain, S. D.
Discovery of 3,5-disubstituted-1H-pyrrolo[2,3-b]pyridines as potent
inhibitors of the insulin-like growth factor-1 receptor (IGF-1R)
tyrosine kinase. Bioorg. Med. Chem. Lett. 2009, 19, 3136−3140.
(21) Miyazaki, Y.; Tang, J.; Maeda, Y.; Nakano, M.; Wang, L.; Nolte,
R. T.; Sato, H.; Sugai, M.; Okamoto, Y.; Truesdale, A. T.; Hassler, D.
F.; Nartey, E. N.; Patrick, D. R.; Ho, M. L.; Ozawa, K. Orally active 4-
amino-5-diarylurea-furo[2,3-d]pyrimidine derivatives as anti-angiogen-
ic agent inhibiting VEGFR2 and Tie-2. Bioorg. Med. Chem. Lett. 2007,
17, 1773−1778.
(22) Reaction Biology Corp. http://www.reactionbiology.com.
(23) Duprez, L.; Takahashi, N.; Van Hauwermeiren, F.;
Vandendriessche, B.; Goossens, V.; Vanden Berghe, T.; Declercq,
W.; Libert, C.; Cauwels, A.; Vandenabeele, P. RIP kinase-dependent
necrosis drives lethal systemic inflammatory response syndrome.
Immunity 2011, 35, 908−918.
1243
dx.doi.org/10.1021/ml400382p | ACS Med. Chem. Lett. 2013, 4, 1238−1243