Selective inhibitors of the protein tyrosine phosphatase SHP2 block cellular motility and growth of cancer cells in vitro and in vivo

Article abstract of DOI:10.1002/cmdc.201500015

Selective inhibitors of the protein tyrosine phosphatase SHP2 (src homology region 2 domain phosphatase; PTPN11), an enzyme that is deregulated in numerous human tumors, were generated through a combination of chemical synthesis and structure-based ration

Full text of DOI:10.1002/cmdc.201500015

DOI: 10.1002/cmdc.201500015
Full Papers
Selective Inhibitors of the Protein Tyrosine Phosphatase
SHP2 Block Cellular Motility and Growth of Cancer Cells
in vitro and in vivo
Stefanie Grosskopf,[a, b, c] Chris Eckert,[b, c] Christoph Arkona,[a] Silke Radetzki,[b]
Kerstin Bçhm,[c] Udo Heinemann,[c] Gerhard Wolber,[a] Jens-Peter von Kries,[b]
Walter Birchmeier,[c] and Jçrg Rademann*[a, b]
Selective inhibitors of the protein tyrosine phosphatase SHP2
(src homology region 2 domain phosphatase; PTPN11), an
enzyme that is deregulated in numerous human tumors, were
generated through a combination of chemical synthesis and
structure-based rational design. Seventy pyridazolon-4-ylidene-
hydrazinyl benzenesulfonates were prepared and evaluated in
enzyme assays. The binding modes of active inhibitors were
simulated in silico using a newly generated crystal structure of
SHP2. The most powerful compound, GS-493 (4-{(2Z)-2-[1,3-
bis(4-nitrophenyl)-5-oxo-1,5-dihydro-4H-pyrazol-4-yliden]hydra-
zino}benzenesulfonic acid; 25) inhibited SHP2 with an IC50
value of 71Æ15 nm in the enzyme assay and was 29- and 45-
fold more active toward SHP2 than against related SHP1 and
PTP1B. In cell culture experiments compound 25 was found to
block hepatocyte growth factor (HGF)-stimulated epithelial–
mesenchymal transition of human pancreatic adenocarcinoma
(HPAF) cells, as indicated by a decrease in the minimum neigh-
bor distances of cells. Moreover, 25 inhibited cell colony forma-
tion in the non-small-cell lung cancer cell line LXFA 526L in
soft agar. Finally, 25 was observed to inhibit tumor growth in
a murine xenograft model. Therefore, the novel specific com-
pound 25 strengthens the hypothesis that SHP2 is a relevant
protein target for the inhibition of mobility and invasiveness of
cancer cells.
Introduction
The reversible phosphorylation of tyrosine residues in proteins
is the key to controlled activation of cellular signal transduc-
tion pathways. Accordingly, tyrosine phosphorylation and de-
phosphorylation is regulated carefully by two enzyme classes:
protein tyrosine kinases (PTK) and protein tyrosine phosphatas-
es (PTP), respectively. Although numerous small-molecule in-
hibitors of PTKs are available and have established PTKs as
pharmacological targets in clinical cancer therapy,[1] only few
potent, specific and intracellularly active inhibitors of PTPs
have been described so far. This lack of chemical tools for PTPs
hampers the investigation of their significance in signal trans-
duction and thus the development of inhibitor lead com-
pounds for pharmacological use. Among the human PTPs, the
protein SHP2 has attracted considerable interest due to its dis-
tinct functional roles.[2,3] In contrast to most other phosphatas-
es, which down-regulate the activity of dependent signal trans-
duction cascades via dephosphorylation of tyrosine kinases
and their corresponding substrate proteins, SHP2 activity has
been identified as essential for signal amplification.
The protein tyrosine phosphatase SHP2 (also called SH-PTP2,
Syp, PTP1D, and PT-P2C) is a ubiquitously expressed 68 kDa
non-receptor PTP. Structurally, SHP2 consists of two consecu-
tive phosphotyrosine binding domains, which are termed N-
and C-terminal Src homology 2 domains (N-SH2 and C-SH2, re-
spectively), and a catalytic PTP domain.[2] In the inactive state,
the SH2 domains of SHP2 auto-inhibit the catalytic PTP
domain by binding intramolecularly to defined acidic resi-
dues.[4,5] Upon activation, several receptor tyrosine kinases, in-
cluding MET, EGF, and PDGF, recruit adaptor proteins such as
GAB and GRAB to form a protein complex. Phosphotyrosine
residues of the adaptor complex are then recognized by the
C-SH2 and N-SH2 domains of SHP2; this, in turn, leads to the
activation of the catalytic PTP domain. As a consequence,
SHP2 dephosphorylates a number of target proteins including
RAS, and this results in the up-regulation of the ERK/MAPK
pathway.[6]
[a] Dr. S. Grosskopf, Dr. C. Arkona, Prof. Dr. G. Wolber, Prof. Dr. J. Rademann
Institut für Pharmazie/Institut für Chemie, Freie Universität Berlin
Kçnigin-Luise-Str. 2+4, 14195 Berlin (Germany)
E-mail: joerg.rademann@fu-berlin.de
Homepage: http://www.bcp.fu-berlin.de/ag-rademann
[b] Dr. S. Grosskopf, C. Eckert, Dr. S. Radetzki, Dr. J.-P. von Kries,
Prof. Dr. J. Rademann
Leibniz-Institut für Molekulare Pharmakologie (FMP)
Robert-Rçssle-Str. 10, 13125 Berlin (Germany)
[c] Dr. S. Grosskopf, C. Eckert, Dr. K. Bçhm, Prof. Dr. U. Heinemann,
Prof. Dr. W. Birchmeier
Max Delbrück Center for Molecular Medicine (MDC)
Robert-Rçssle-Str. 10, 13125 Berlin (Germany)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201500015: Tables containing crystallo-
graphic data for SHP2 structures, supplemental figures of cell-scattering
experiments, and protocols for the synthesis and analytical data for com-
pounds 1–19. PDB IDs: 3ZM0, 3ZM1, 3ZM2, 3ZM3.
This specific functional role of the SHP2 protein is supported
by considerable genetic evidence. The gene encoding SHP2,
ChemMedChem 2015, 10, 815 – 826
815
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
PTPN11, has been identified as an oncogene in the develop-
ment of several types of leukemia and solid tumors.[3] Gain-of-
function mutants of SHP2 have been found in childhood hem-
atological malignancies such as juvenile myelomonocytic leu-
kemia (JMML), a progressive myelodysplastic/myeloprolifera-
tive disorder characterized by overproduction of tissue-infiltrat-
ing myeloid cells.[7] Somatic mutations in PTPN11 account for
about 35% of JMML patients. Somatic gain-of-function muta-
tions have been identified in several types of leukemia includ-
ing acute myeloid leukemia (AML), myelodysplastic syndrome
(MDS), and acute lymphoid leukemia (ALL).[7,8] In addition,
PTPN11 mutations occur in solid tumors, including lung adeno-
carcinoma and colon cancer. It has also been reported that
SHP2 is a key mediator of the oncogenic CagA protein of Heli-
cobacter pylori, which causes gastric cancer.[9,10] Beyond the
cancer field, several activating mutations of SHP2 have been
found in Noonan syndrome, a developmental disorder charac-
terized by facial anomalies, short stature, heart disease, skeletal
defects, and hematological disorders. About 50% of all cases
of Noonan syndrome are caused by germline PTPN11 muta-
tions.[11,12] While the majority of pathological SHP2 mutations
enhance the catalytic activity of the protein, mostly by attenu-
ating the auto-inhibition of the N-SH2 domain, deactivating
mutations of the PTPN11 gene have also been described in
a disease denominated as LEOPARD syndrome, which displays
symptoms similar to those of Noonan syndrome.[13–15] Although
the occurrence of similar disease phenotypes for activating
and deactivating mutations appears to be contradictory at first
glance, it has been explained by the finding that mutations de-
activating the catalytic domain in vitro might still result in
functional activation in vivo if they enhance the adaptor func-
tion. In summary, the collected findings suggest that SHP2
might be a molecular target for anticancer and leukemia thera-
py, as it plays a decisive role in the regulation of cell motility,
cell migration, and angiogenesis.
Results and Discussion
The development of potent and selective PTP inhibitors has
been challenging in drug research[20,2] due to the high degree
of homology found in the phosphotyrosine substrate binding
pockets.[21] Recently, second-site binding fragments have been
detected in a ligation assay, and these fragments were convert-
ed into moderately active, but highly specific, PTP inhibitors.[22]
These results encouraged further studies on fragment-based,
specific PTP inhibitors. Besides the specificity issue, most of the
potent small-molecule inhibitors of PTPs contained di-anionic
phosphotyrosine bioisosteres such as the phenyl difluorome-
thylphosphonate isostere, rendering them inefficient in cross-
ing membrane barriers and thus impairing the pharmacokinet-
ics of potential drug candidates. As a result, only a few com-
pounds are currently in clinical trials.[23]
Developing a selective SHP2 inhibitor is furthermore difficult
due to the similarity between SHP2 and SHP1, which share
60% overall primary sequence identity and ~75% similarity in
their catalytic PTP domains.[24] The starting point for the cur-
rent study was the inhibitor 1, a pyrazolonylidene hydrazinyl-
benzenesulfonate. This inhibitor has already been character-
ized functionally in two important aspects.[16] Inhibitor 1 is a mi-
cromolar competitive inhibitor of the catalytic domain of SHP2
and has only little specificity for SHP2 over PTP1B and SHP1.
Moreover, 1 was found to be active in MDCK cells, as demon-
strated by its effects on ERK phosphorylation. Immunoblotting
of protein extracts from MDCK cells revealed that inhibitor
1 blocks the phosphorylation of ERK in a concentration-depen-
dent manner. This effect could be overcome by overexpressing
cellular RAS. Because SHP2 is required for RAS/ERK pathway ac-
tivation (phosphorylation), the above findings are in line with
the hypothesis that inhibitor 1 crosses the cell membrane, suc-
cessfully inhibits intracellular SHP2, and thereby interferes with
RAS signaling to ERK.[6]
In preceding work we identified a low-micromolar SHP2 in-
hibitor in a virtual screening approach using a homology
model of SHP2.[16] The best compound of this initial work, the
pyrazolon-4-ylidene hydrazinylbenzenesulfonate 1, was identi-
fied as a competitive inhibitor of SHP2. Although the com-
pound showed only moderate potency and low selectivity
toward SHP2, it was able to down-regulate the hepatocyte
growth factor (HGF)-induced MAP/ERK pathway in cells and
displayed cellular activity in simple metastasis models. Huang
et al. described PTP1B inhibitors that cross-inhibit SHP2, but
none of them has demonstrated activity in cell cultures.[17]
Chen et al. as well as Zhang et al. have reported other mole-
cules that inhibit SHP2 at similar micromolar concentrations
and with limited selectivity.[18,19] Therefore, the development
and primary validation of potent and selective inhibitors of
SHP2, which will allow us to study the role of SHP2 in normal
and pathological processes, became the aim of our study.
Results thereof are presented in the following.
With these two principal findings in hand, we attempted to
further improve the potency and selectivity of inhibitor 1 while
conserving the principal lead structure for which the desired
intracellular activity had been successfully demonstrated. Con-
sidering the above-mentioned findings on the selectivity of
second-site binding fragments in PTPs, we hypothesized that
a careful optimization of those parts of inhibitor 1 pointing
outside the phosphotyrosine binding pocket might lead to this
goal.
Chemical optimization of SHP2 inhibitors
The starting point of this study, pyrazolonylidene hydrazinyl-
benzenesulfonate 1 (Scheme 1c), contains three substituent
positions (R1, R2, and R3; Scheme 1b). Whereas position R2, ac-
cording to molecular modeling studies (Figure 1), binds to the
phosphotyrosine pocket, substituents at R1 and R3 are available
for targeting of secondary sites (Scheme 1b). Therefore, a pro-
tocol was devised for the variation of substituent positions R1–
R3 in four steps starting from three fragment building blocks:
one carboxylic acid chloride (F1) and two hydrazines (F2 and
F3) (Scheme 1a). The phosphorus ylide, triphenylphosphorany-
ChemMedChem 2015, 10, 815 – 826
www.chemmedchem.org
816
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Variations of the R2 group were
scrutinized containing the ben-
zenesulfonic acid as a reported
phosphotyrosine mimetic, which
was predicted to bind the pri-
mary phosphotyrosine pocket of
SHP2 (Figure 1b). Here, the sub-
stitution of R2 with a sulfonamide
group (in 7) maintained inhibito-
ry activity at low micromolar
levels. Addition of two chlorines
yielding the 2,6-dichloro benze-
nesulfonic acid 8 even improved
inhibition twofold. For the R3 po-
sition, molecular modeling had
predicted available space for
sterically demanding residues,
which might contribute addi-
tional binding interactions with
the surface of SHP2. Accordingly,
several bulky groups were intro-
duced at R3; however, 4-cyclo-
hexylphenyl 9 decreased the ac-
tivity twofold and 4-benzylphen-
yl 11 by a factor of four. In con-
trast, smaller substituents at R3
provided the first sub-micro-
molar inhibitors of the series, in-
cluding the 4-toluyl- (14), 4-me-
thoxyphenyl- (15), and 4-trifluor-
omethylphenyl derivative (16),
while the 4-chlorophenyl residue
at R3 (compound 17) did not im-
prove the activity.
Scheme 1. a) Synthesis of the pyrazolone scaffold VI enabling variation of substituents R1–R3 in a four-step proto-
col starting from three fragment building blocks, carboxylic acid chloride F1 (or carboxylic acid) and two aryl hy-
drazines, F2 and F3, respectively. Reagents and conditions: step a: methyl 2-(triphenylphosphoranylidene)acetate I
(1 equiv), R1PhCOCl F1 (1.2 equiv), lutidine (1.4 equiv), CH2Cl2, RT, 16 h, or: methyl 2-(triphenylphosphanylidene)a-
cetate I, 4-nitrobenzoic acid (1.2 equiv), 1-(mesitylene-2-sulfonyl)-3-nitro-1H-1,2,4-triazole (MSNT, 1.2 equiv), luti-
dine (1.4 equiv), CH2Cl2, RT, 16 h; step b: methyl 3-aryl-3-oxo-2-triphenylphosphoranylidenepropanoate II, dime-
thyldioxirane (DMDO) in acetone (3 equiv), CH2Cl2, RT, 1 h; step c: methyl 2,2-dihydroxy-3-oxo-propanoate III
(1 equiv), R2PhNHNH2 F2 (1.2 equiv), EtOH, HCl (1% v/v), 858C, 16 h; step d:+R3NHNH2 F3 (1.2–1.4 equiv), 858C,
16 h. b) General structure of SHP2 inhibitors. c) Chemical structure of compound 1. d) Chemical structure of com-
pound 25 (GS-493).
lidene acetate I, was first acylated with the acid chloride F1 to
furnish keto-phosphoranylidene acetate II (step a).[25,26] Second-
ly, this intermediate was cleaved by oxidation with 2,2-dimeth-
yl dioxirane to generate the 2,3-diketo ester III (step b).[27]
Compound III comprises a triselectrophile, which was coupled
with two different aryl hydrazines F2 and F3 in a “two-step-
one-pot” synthesis. The first hydrazine fragment was con-
densed at the strongest electrophilic 2-carbonyl position (C2)
activated by the addition of 1% of concentrated hydrochloric
acid, providing the 2-hydrazone IV (step c). Sustained acidic
conditions were required to activate the 3-carbonyl position,
giving rise to a condensation with the second hydrazine frag-
ment followed by the cyclizing cleavage of methyl ester, lead-
ing to a pyrazolone ring (step d). Variation of the three frag-
ment building blocks F1, F2, and F3 delivered a focused selec-
tion of potential SHP2 inhibitors, which were validated in phos-
phatase inhibition assays with proteins SHP2, PTB1B, and SHP1
(see Experimental Section below).
Next, heterocyclic substituents were tested at R3. The 2-ben-
zoxazolyl and 3-chloropyridazinyl residues (compounds 10 and
12) showed about 40- and 2-fold decreases in activity, respec-
tively, whereas the N-morpholinylethyl residue (compound 13)
was threefold less active. A threefold improvement in inhibito-
ry activity into the sub-micromolar range was observed for
compounds 20–22, containing 2-tetrazolo[b]pyridazinyl, 2-qui-
noxalinyl, and 2-benzothiazolyl substituents. Even more power-
ful, with respective IC50 values of 0.37 and 0.15 mm, were the
inhibitors carrying 3,4-ethylendioxyphenyl (24, GS-458) or 3,4-
methylendioxyphenyl at R3 (23, GS-447). The most active SHP2
inhibitor obtained thus far was 25 (GS-493, IC50 =0.071 mm),
bearing 4-nitrophenyl at the R3 position (Scheme 1d), corre-
sponding to a 20-fold improved inhibitory activity against
SHP2 relative to 1 (Table 1).
Specificity of the compounds was evaluated by comparing
inhibition in the enzyme assay against SHP2 with its activity
against SHP1 and PTP1B, two closely related PTPs. In all cases
the analyzed compounds showed preferred inhibition of the
targeted phosphatase SHP2. The strongest effects were identi-
fied for activating derivatizations at the R3 position. GS-447
(23) and GS-458 (24) showed 22- and 31-fold higher activity
against SHP2 than against SHP1 and 18- and 36-fold higher ac-
First, substitutions of the 4-nitro group at the R1 position of
compound 1 were investigated (Scheme 1c). Replacing the
nitro group by chloride, trifluoromethoxy, trifluoromethylsul-
fonyl, and aminosulfonyl or shifting the nitro group to the 3-
position decreased inhibition by at least 30-fold (Table 1, 2–6).
ChemMedChem 2015, 10, 815 – 826
www.chemmedchem.org
817
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Table 1. Structures of SHP2 inhibitors and their inhibition of enzyme activity against SHP2, SHP1, and PTP1B.
Compd
Modification
R2
IC50 [mm][a]
SHP1
R1
R3
SHP2
PTP1B
1
2
3
4
5
6
7
8
9
–4-NO2
–3-NO2
–4-Cl
–4-O-CF3
–4-SO2-CF3
–4-SO2NH2
–4-NO2
–4-NO2
–4-NO2
–4-SO3H
–4-SO3H
–4-SO3H
–4-SO3H
–4-SO3H
–4-SO3H
–4-SO2NH2
–2,5-Cl2-4-SO3H
–4-SO3H
–C6H5
–C6H5
–C6H5
–C6H5
–C6H5
–C6H5
–C6H5
–C6H5
1.38Æ0.28
10.0Æ1.32
ND
62.9Æ25.3
NA
NA
26.6Æ4.63
ND
13.1Æ2.48
ND
24
72.6Æ32.66
NA
NA
NA
NA
NA
NA
42.5Æ9.01
5.00Æ0.85
0.61Æ0.09
20.3Æ5.48
ND
5.23Æ0.75
43.1Æ9.86
7.29Æ1.23
81.3Æ39.2
–C6H11
10
–4-NO2
–4-SO3H
29.1Æ6.22
11.6Æ1.60
NA
11
12
–4-NO2
–4-NO2
–4-SO3H
–4-SO3H
–CH2-Ph
5.42Æ1.35
2.30Æ0.52
22.0Æ4.32
14.6Æ2.47
83.2Æ38.6
26.7Æ9.51
13
–4-NO2
–4-SO3H
4.34Æ0.94
12.8Æ1.61
NA
14
15
16
17
18
–4-NO2
–4-NO2
–4-NO2
–4-NO2
–4-NO2
–4-SO3H
–4-SO3H
–4-SO3H
–4-SO3H
–4-SO3H
–C6H4-4-Me
–C6H4-4-OMe
–C6H4-4-CF3
–C6H4-4-Cl
0.41Æ0.08
0.50Æ0.1
0.94Æ0.2
1.16Æ0.31
2.05Æ0.35
9.83Æ1.62
2.14Æ0.34
14.4Æ3.73
10.2Æ1.72
15.2Æ3.06
6.56Æ1.49
6.80Æ1.44
10.6Æ3.0
8.93Æ2.27
16.9Æ4.0
–C6H4-3-Cl
19
–4-NO2
–4-SO3H
4.77Æ.0.72
7.66Æ1.08
26.3Æ6.8
20
21
22
–4-NO2
–4-NO2
–4-NO2
–4-SO3H
–4-SO3H
–4-SO3H
0.4Æ0.12
0.30Æ0.06
0.30Æ0.07
8.52Æ1.67
3.73Æ0.68
6.37Æ0.99
5.80Æ1.3
3.04Æ0.57
4.42Æ0.9
23 (GS-447)
–4-NO2
–4-SO3H
0.37Æ0.08
7.99Æ1.31
6.48Æ1.31
24 (GS-458)
25 (GS-493)
–4-NO2
–4-NO2
–4-SO3H
–4-SO3H
0.15Æ0.04
4.66Æ0.69
2.08Æ0.31
5.38Æ1
4-NO2-C6H4–
0.071Æ0.02
3.17Æ0.41
[a] IC50 values were determined using an enzymatic DiFMUP assay and are the meanÆSD calculated from three independent experiments; NA: not active,
ND: not determined.
ChemMedChem 2015, 10, 815 – 826
www.chemmedchem.org
818
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
tivity than against PTP1B. The best compound 25 (GS-493) ex-
hibited higher specificity against SHP2 than compound 1 and
other reported SHP2 inhibitors; its activity was 30-fold higher
for SHP2 compared with the closely related phosphatase SHP1
and 45-fold higher over activity against PTP1B (Table 1).
All novel compounds were soluble in buffer at significantly
higher concentrations than those finally used in the biological
experiments. Inhibition assays were conducted in the presence
of 0.01–0.1% (w/v) bovine serum albumin (BSA) in the buffer
without any decrease in inhibition, indicating that the active
compounds including 25 are not prone to unspecific protein
binding. Addition of the surfactant Tween-20 at 0.03% (v/v) to
the assay did not alter inhibition, excluding simple aggregation
of compounds as a source of enzyme inhibition. These bio-
physical data for the specific inhibitor 25 encouraged its fur-
ther investigation in cellular and animal experiments.
Docking studies and structure–activity relationships
To rationalize the binding of the presented compounds, initial
docking experiments using the published SHP2 structures
(PDB IDs: 3B7O and 3O5X) were conducted. Because the
ligand conformation in complex 3O5X shows considerable
strain with regard to the cis-amide bond of the co-crystallized
ligand, re-docking experiments could not reproduce the
bound ligand conformation, which might be explained by crys-
tal packing effects near the binding pocket. Therefore, we de-
cided to crystallize SHP2 protein ourselves, and four different
apoprotein crystal structures were generated for this study
(PDB IDs: 3ZM1, 3ZM0, 3ZM2, and 3ZM3; see Supporting Infor-
mation for details). An ensemble docking protocol was then
performed with all six consequently available SHP2 structures,
meaning that each compound was docked into the catalytic
site of all six available protein crystal structures to sample pos-
sible binding geometries, using the software package GOLD
ver. 5.0.1.[18,2,28–30] The most plausible docking poses in terms of
formed interactions were obtained for the SHP2 structure
3ZM3 and subsequently analyzed in terms of formed 3D phar-
macophores (Figure 1, see Experimental Section for details).
Molecular docking results indicate that the core interaction
motif for all compounds consists of a hydrogen bonding net-
work with the backbone of residues near the catalytic center
(Cys459, Ser460, Ala461, Gly462, Gly464, Arg465 and Gln506),
as observed previously for compound 1 using a homology
model of SHP2 based on the structure of PTP1B.[16] Additional-
ly, the phenyl ring at R2 is embedded between Lys366 and
Thr357 forming hydrophobic contacts, and the nitrophenyl
substitution at R1 shows a p–cation interaction with Arg362.
The newly introduced nitro group attached to the phenyl
moiety at R3 serves as a hydrogen bond acceptor for Thr507
which underlies the logic behind introducing this new group
into the lead structure 1. The central imino bond displays con-
siderable strain, which might be a starting point for further op-
timization. All compounds show geometrically highly conver-
gent conformations preserving the main interaction motif with
the catalytic center (Figure S4, Supporting Information). Varia-
tions in R2 mainly differ with respect to their ability to form
Figure 1. Interactions between the most plausible conformation of com-
pound 25 and SHP2 in a) 2D and b) 3D derived from docking into crystal
structure PDB ID: 3ZM3. Red arrows indicate hydrogen bond acceptors,
green arrows represent donors, blue disks are p–cation interactions, and
yellow spheres show hydrophobic contacts.
a hydrogen bond with Thr507 and to present their lipophilic
moieties to the methyl group of Thr507, as well as to Ala509
and Val428. The nitrophenyl substitution of compound GS-493
forms a geometrically ideal interaction with Thr507, which can
explain its improved biological activity (cf. below). The dichlor-
ophenyl substitution of compound 8 in R2 only slightly shifts
the sulfonyl group, distorting the interactions with the catalytic
center, but at the same time new hydrophobic contacts with
Thr507 are formed. Remarkably, compound 16 allows an ac-
ceptor-like fluorine interaction from the trifluoromethyl group
with the hydroxy group of Thr507, while this is not possible
for the chlorine substitution in 17 and 18, which can explain
the activity difference between these two compounds. The
phenyl ring in compound 11 is in a non-optimal orientation for
hydrophobic contacts to each of the lipophilic residues in this
area, which is also the case for the cyclohexyl residue in com-
pound 9. The benzoxazole substitution in compound 10 is too
short and presents the phenyl ring to the oxygen atom of
Thr507, which can explain the decreased activity of this com-
pound. The sulfonamide at position R1 of compound 6
changes the orientation of the attached phenyl ring and thus
distorts the geometry of the p–cation interaction with Arg362,
explaining its low activity.
ChemMedChem 2015, 10, 815 – 826
www.chemmedchem.org
819
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
The docking studies were performed with the catalytic
domain of human SHP2 comprising residues 248–527. Notably,
all the above-described contacts involve amino acids kept con-
stant in mutants of SHP2 as they are typically found in patients
of Noonan or LEOPARD syndromes. Although not tested in this
work there are good reasons to assume that our series of SHP2
inhibitor compounds will work on naturally occurring mutant
forms of human SHP2 in a similar way.
their potential to revert the scattering phenotype; 25 of the
synthesized compounds were tested in a scatter assay with the
human cells at concentrations between 62 nm and 10 mm.
Quantitative evaluation was conducted as described above.
Only the novel specific SHP2 inhibitors GS-458 and GS-493
were able to reverse the scattering phenotype completely. At
an inhibitor concentration of 10 mm (non-cytotoxic for both
compounds) the percentage of cells with minimum neighbor
distances below the threshold was even higher than with HGF-
negative cells, indicating that the cell colonies were even more
densely packed than those of untreated cells (Figure 2b). For
comparison, the nonspecific inhibitor 1, even at the highest
concentration (10 mm), led only to a partial reversion of the
scattering phenotype. Therefore, the potent SHP2 inhibitors
block the scattering of human cancer cells in a concentration-
dependent manner. Furthermore, the cell images show the be-
havior of HPAF II cells in the presence and absence of HGF/SF
(Figure 2a).
Improved inhibitors block HGF/SF induced scattering of
human HPAF II pancreatic cancer cells
As a test for its cellular activity, the improved specific SHP2 in-
hibitors 24 (GS-458) and 25 (GS-493) as well as the initial com-
pound 1 were investigated in an assay to block the epithelial–
mesenchymal transition of human pancreatic cancer cells. This
so-called “scattering” assay is generally considered as a model
to investigate the motility, invasiveness, and metastatic poten-
tial of cancer cells.[6,31] To identify a suitable human cell line for
this assay, several human cancer cell lines were treated with
hepatocyte growth factor (HGF) also denominated as scatter
factor (SF), the native ligand of the MET receptor.[16,31] Cells of
the human pancreatic cancer cell line HPAF II displayed the ex-
pected phenotype for a successful epithelial–mesenchymal
transition: activation of the MET receptor by addition of HGF/
SF led to the scattering of cell colonies in tissue culture. For
a quantitative analysis of the cell scattering phenotype, a stand-
ardized protocol for use with an automated fluorescence mi-
croscope was developed. The nuclei of HPAF II cells were
stained with Hoechst dye in order to determine the average
proximity of adjacent cell nuclei, recorded as the minimum
neighbor distance (Figure 2). For each well in the 384-well mi-
crotiter plates, nine images were recorded with an autofocus
interval of three fields at 10” magnification in the Hoechst
channel, typically resulting in the analysis of ~1000 cells per
well. Image analysis was conducted using the nuclei of the
cells as primary objects (see the Experimental Section and Sup-
porting Information for details).
Improved SHP2 inhibitors block growth of the human NSCL
cancer cell line LXFA 526L in soft agar and in a nude mouse
xenograft model
As described above, the two novel SHP2 inhibitors GS-458 and
GS-493 efficiently blocked the scattering of HPAF II human
pancreatic cancer cells. This finding encouraged us to further
investigate the cellular effects of the SHP2 inhibitors in cancer
treatment. These further investigations were performed with
a different cell line for two reasons: 1) to strengthen the hy-
pothesis of a more general applicability of our SHP2 inhibitors,
and 2) to directly test for an anti-proliferating effect in an
animal model. For these experiments it was essential to first
identify a human carcinoma cell line whose mobility and prolif-
eration depend critically on the functional activity of the SHP2
protein. Such an SHP2-dependent cell line would furthermore
enable a comparison of the effects of SHP2 inhibitors with
those of SHP2-directed siRNA suppressing SHP2 mRNA transla-
tion. A SHP2-dependent cancer cell line was successfully identi-
fied from a collection of patient-derived human tumor cells
lines provided by Oncotest GmbH (Freiburg, Germany). Using
siRNA targeting the MET receptor, several cell lines from the
collection had been established as being MET dependent.[32]
Selected MET-dependent human tumor cell lines were then
treated with four commercially available SHP2 siRNAs of differ-
ent nucleotide sequence (HP Genome Wide siRNA Hs_PTPN11_
1(A),_5(B),_6(C),_7(D) by Qiagen) at concentrations of 15 and
50 nm. Cells were then lysed and probed for SHP2 expression
by standard immunoblotting procedures against untreated
control samples. In the case of the patient-derived non-small-
cell lung (NSCL) adenocarcinoma cell line LXFA 526L,[32] all four
applied siRNAs led to an efficient suppression of SHP2 expres-
sion (Figure 3a). Therefore, this cell line was selected for fur-
ther testing of our SHP2 inhibitors. In LXFA 526L cells the
strongest down-regulation of SHP2 protein by siRNA to a level
of 37% was found for sequence A (Qiagen HsPTP11_1;
#SI00044002; 5’-GCCG CTCA TGAC TATA CGCTA-3’). This se-
quence was therefore selected as a standard for direct compar-
For assay setup, the average proximity of adjacent cells was
determined with and without incubation of the cells with HGF/
SF overnight. As expected, the addition of HGF/SF led to the
scattered phenotype as compared with non-scattered cell colo-
nies prior to treatment. (Figure 2; Figures S1 and S2). The scat-
tered cells displayed a significantly broader distribution of the
neighbor minimum distances (Figure S3). To quantitatively
evaluate the shift of the HGF-treated cell population to in-
creased minimum neighbor distances, a threshold of 21 mm
was defined, and the percentage of cells displaying minimal
distance values higher than the threshold was calculated. This
evaluation protocol allowed a clear and robust distinction be-
tween non-scattered and scattered phenotypes as well as
a quantitative description of the degree of scattering observed
after treatment with HGF/SF.
Next, the effects of the improved SHP2 inhibitors were inves-
tigated in the scatter assay with HPAF II cells. For this purpose
cells were treated with HGF/SF (to induce the scattering phe-
notype) and with one of the SHP2 inhibitors in order to test
ChemMedChem 2015, 10, 815 – 826
www.chemmedchem.org
820
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 2. Colony dissociation of HGF/SF-stimulated HPAF II cells. a) Addition of HGF/SF induces the dissociation of cell colonies; the addition of SHP2 inhibitor
1 led to partial, and 25 (GS-493) to full inhibition of cell dissociation in a concentration-dependent manner. b) Quantitative evaluation of cell scattering by re-
cording the relative neighbor minimum distance of cells treated with compounds 1, 24 (GS-458), and 25 (GS-493) at five different concentrations and com-
pared to controls with and without HGF/SF. Data shown are from one of three independent sets of experiments. Columns represent the meanÆSD percent-
age of cells of three samples having a minimum neighbor distance greater than the threshold value defined from the positive control experiment (cells stimu-
lated with HGF without SHP2 inhibitor added). Values of the control experiments were determined as the mean of twelve samples. Compounds 24 and 25,
but not 1, led to full reversion of the scattered phenotype (ÀHGF). See the Supporting Information for further explanations of these experiments.
ison with the two new SHP2 inhibitors, GS-458 and GS-493.
First the effects of the SHP2-targeted siRNA A on the anchor-
age-independent colony formation of LXFA 526L cells were in-
vestigated in a soft agar assay and compared with the effects
of the two SHP2 inhibitors (Figure 3b). Anchorage-independ-
ent colony formation has been considered as a model system
for the invasiveness and thus the tumor-promoting properties
of cancer cells.[33] Addition of siRNA A to the cell culture de-
creased the number of colonies from 100% (control) down to
a level of 38%. The two novel SHP2 inhibitors, compounds 24
(GS-458) and 25 (GS-493), decreased colonies to 20 and 30%,
respectively, similar to the effect of siRNA A (Figure 3b). Thus,
the phenotypes obtained by siRNA knock-down of SHP2 trans-
lation and by the novel SHP2 inhibitors were found to be qual-
itatively and quantitatively in the same range. This observation
further supports the concept that the cellular effects of com-
pounds GS-458 and GS-493 are exerted via inhibition of the
functional activity of SHP2.
Consequently, the SHP2-dependent cell line LXFA 526L iden-
tified in the studies above was also used in a murine xenograft
model[34] (see Supporting Information for details). First, dose-
finding studies were conducted in nude mice and indicated no
acute toxicity and no persistent weight loss with compound
24 at 115 mgkgÀ1 dayÀ1 and 25 at 46 mgkgÀ1 dayÀ1. Then, the
efficacy of compound 25 (which showed the best inhibitory
activity toward the SHP2 in vitro; cf. above) to stop tumor
ChemMedChem 2015, 10, 815 – 826
www.chemmedchem.org
821
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 4. Change in relative tumor size values in nude mice during 27 days
of treatment, expressed as median from five test mice. Nude mice (Charles
River, Sulzfeld, Germany) were subcutaneously implanted with fragments of
a tumor derived from the patient lung adenocarcinoma cell line LXFA 526
and grown as mouse xenografts. Mice carrying a mean tumor diameter of
6–8 mm were randomly distributed into two groups and either treated with
once-daily doses of DMSO or of compound 25 dissolved in DMSO
(46 mgkgÀ1 dayÀ1). Tumor volumes were measured twice weekly, and the
median values were calculated. The day of randomization was designated as
day 0, and the tumor size at this day was designated as 100% (absolute
tumor sizes (median): 79.3 mm3 for DMSO group and 89.2 mm3 for GS-493
group). All animal experiments were performed at Oncotest GmbH (Frei-
burg, Germany, study number P227H) according to the guidelines of the
German Animal Welfare Act. Single data points corresponding to the
median values are also indicated.
Figure 3. a) Expression of SHP2 protein as shown by western blot after trans-
fection of the cell line LXFA 526L with four different siRNAs (A–D) directed
against SHP2 at concentrations of 15 nm (panel a, A–D) and of 50 nm
(panel b, A–D). b) Relative changes in soft-agar assay of colony formation
from LXFA 526L human non-small-cell lung adenocarcinoma cells as induced
by treatment with siRNA of sequence A or with compounds 24 (GS-458) and
25 (GS-493), respectively. The soft-agar assay shows that tumor growth is de-
pendent on SHP2, as reduced expression of SHP2 resulted in a significantly
decreased (p<0.01) number of tumor colonies (37%, column b) relative to
controls (columns a and c; both non-transfected tumor cell colonies). The in-
hibitory activities of compounds 24 (GS-458, 40 mm; column d) and 25 (GS-
493, 40 mm; column e) are similar to that observed with siRNA (column b).
Compounds 24 and 25 decreased the number of tumor cell colonies to 20
and 32%, respectively.
Conclusions
Potent and selective inhibitors of SHP2 have been developed
by rational variation of molecular fragments attached to a pyra-
zolone scaffold. The best compound 25 (GS-493) inhibited
SHP2 at nanomolar concentrations in the enzymatic assay and
was 29- and 45-fold more active for SHP2 than for SHP1 and
PTP1B. Compounds 24 and 25 were soluble in aqueous buffer,
and their inhibitory activities were not altered in the presence
of BSA or of surfactant (Tween-20) in the assay buffer. Molecu-
lar modeling of GS-493 binding to a newly recorded crystal
structure of the catalytic domain of SHP2 rationalized the rec-
ognition of the inhibitor by the primary phosphotyrosine
pocket and by secondary binding sites of the protein. As the
active inhibitors bind competitively to the catalytic domain of
wild-type SHP2, they can be expected to inhibit SHP2 mutants
with activating mutations in the regulatory domain as well.
Such activating point mutations in the regulatory domain of
SHP2 are found in JMML, AML, and Noonan disease and do
not affect the structure of the catalytic domain. In cellular
assays compound 25 (GS-493) reverted the endothelial–mesen-
chymal transition induced by hepatocyte growth factor (scatter
factor) HGF/SF in the human pancreatic cancer cell line HPAF II.
For quantification of epithelial–mesenchymal transition, the
minimal neighbor distances of cells were measured in an auto-
mated fluorescence microscopy assay. For comparison, earlier
and less selective inhibitors of SHP2 led only to a partial rever-
sion of the scattering phenotype. In addition, the SHP2 inhibi-
tor compound 25 was found to block anchorage-independent
growth was tested in a nude mouse xenograft model. For this,
two groups of mice (each n=5) were implanted subcutane-
ously with a tumor of LXFA 526L, one group receiving vehicle
(DMSO) only, and the other receiving compound 25 (dissolved
in DMSO) once daily. Tumor volumes (mm3) were measured
twice weekly, and data were collected over 27 days for both
groups (Figure 4). After 27 days the median relative tumor
volume in the control group reached 175% of the initial tumor
size, indicating implantation of an aggressively growing tumor.
In contrast, in the group treated with compound 25 the
median relative tumor volume remained essentially constant
and reached a minimal quotient (treated group)/(control
group) of 42.6% at day 27. The observed inhibition of com-
pound 25 on tumor growth was statistically significant (p=
0.048). This indicates a mechanism of control over the aggres-
sive tumor behavior. We interpret this result as a promising
hint that pharmacological targeting of SHP2 can contribute to
new concepts in tumor therapy.
ChemMedChem 2015, 10, 815 – 826
www.chemmedchem.org
822
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Methyl
3-(4-nitrophenyl)-3-oxo-2-(triphenyl-l5-phosphoranyli-
growth of the human NSCL cancer cell line LXFA 526L with the
same efficacy as the positive control, in which the SHP2 mRNA
transcript was functionally blocked by transfection with SHP2-
specific siRNA. This finding further supports that the inhibition
of SHP2 by compound 25 was responsible for the observed
decrease in anchorage-independent growth of the cancer cells
and is in agreement with earlier results for the closely related
inhibitor 1, which was demonstrated to interfere with SHP2-de-
pendent signaling.[16] Finally, the SHP2-overexpressing cell line
LXFA 526L was used in a murine xenograft model, which dem-
onstrated the inhibitory effect of the new compound on solid
tumor growth in vivo (Figure 4).
dene)propanoate (II): Methyl 2-(triphenylphosphoranylidene)ace-
tate (1 equiv, 1.5 mmol) was dissolved in dry CH2Cl2 (5 mL). A mix-
ture of 4-nitrobenzoic acid (1.2 equiv, 1.8 mmol), 1-(mesitylene-2-
sulfonyl)-3-nitro-1H-1,2,4-triazole (MSNT; 1.2 equiv, 1.8 mmol) as
coupling reagent and lutidine (1.4 equiv, 1.7 mmol) in dry CH2Cl2
(5 mL) was added. The reaction mixture was stirred at room tem-
perature overnight (16 h) and subsequently quenched with 5 mL
of water. The two layers were separated, and the aqueous layer
was extracted with CH2Cl2 (3”10 mL). The combined organic layers
were dried over magnesium sulfate, filtered, and concentrated in
vacuo. The crude product was purified by flash chromatography
over silica gel using EtOAc/hexane as eluent (80:20, v/v) to yield
compound II (610 mg, 84%) as a yellow solid; mp: 206–2088C;
1H NMR (300 MHz, CDCl3): d=8.20 (d, J=10 Hz, 2H, NO2-Ph), 7.78–
7.48 (m, 17H, NO2-Ph, PPh3), 3.10 ppm (s, 3H, Me); 13C NMR
(75 MHz, CDCl3): d=190.8 (d, JCÀP =5 Hz), 167.6 (d, JCÀP =13 Hz),
149.2 (d, JCÀP =9 Hz), 148.0, 133.4, 133.3, 132.3, 128.9, 128.7, 128.6,
125.8, 124.6, 122.8, 49.9 (d, JCÀP =1 Hz), 45.5 ppm; Mass calcd for
C28H22NO5P [M]: 483.45 Da, found [M+H]+: 484.2 m/z.
During all cell-based experiments of this work no acute
signs of cytotoxic effects of the test compounds were ob-
served. In addition, compounds were tested in a systematic
tolerance study to investigate mortality and weight loss in
nude mice (Oncotest, Freiburg, Germany). The experiments
showed that the compounds were well tolerated in mice, as
no mortality and no phenotypic abnormalities were observed
in comparison with the vehicle-treated animals. In summary,
the results for compound 25 (GS-493) strengthen the hypothe-
sis that SHP2 protein can be a relevant target for the inhibition
of cancer cell mobility and invasiveness and that potent inhibi-
tors of SHP2 can be developed for in vivo use. This optimistic
perspective also seems to be in good accordance with the
recent work of Yu et al.,[35] who described the successful evalu-
ation of a heteroaryl-substituted b-mercapto-N-naphtylaceta-
mide as a selective SHP2 inhibitor. Very recently the findings of
this article were further substantiated by a report that GS-493
through SHP2-inhibition induces senescence in mammary
gland cancer in mice.[51]
Methyl 2,2-dihydroxy-3-(4-nitrophenyl)-3-oxopropanoate (III): To
a solution of methyl 3-(4-nitrophenyl)-3-oxo-2-(triphenyl-l5-phos-
phoranylidene)propanoate II (1 equiv, 1.2 mmol) in CH2Cl2, dime-
thyldioxirane (DMDO; 3 equiv, 3.6 mmol) in acetone was added.
The reaction mixture was stirred at room temperature for 1 h until
the solution was decolored, and subsequently concentrated in
vacuo. The crude product was purified by flash chromatography
over silica gel using EtOAc/hexane as eluent (40:60, v/v) to yield
compound III (250 mg, 86%) as a yellow solid; mp: 110–1128C;
1H NMR (300 MHz, [D6]DMSO): d=8.35–8.23 (m, 4H, arom),
3.65 ppm (s, 3H, Me); 13C NMR (75 MHz, [D6]DMSO): d=193.0,
169.8, 149.9, 137.8, 130.9, 123.6, 94.7, 52.5 ppm; Mass calcd for
C10H7NO6 [M], 237.17 Da, found [M+H]+, 238.1 m/z.
4-{(2Z)-2-[1-(2,3-Dihydro-1,4-benzodioxin-6-yl)-3-(4-nitrophenyl)-
5-oxo-1,5-dihydro-4H-pyrazol-4-ylidene]hydrazino}benzenesul-
fonic acid (23): Methyl 2,2-dihydroxy-3-(4-nitrophenyl)-3-oxopropa-
noate III (1 equiv, 0.21 mmol) was dissolved in EtOH (10 mL), and
concentrated HCl (1% in solution) and 4-hydrazinylbenzenesulfonic
acid hemihydrate (1.2 equiv, 0.25 mmol) were added. The reaction
mixture was stirred for 16 h at 858C under reflux. After complete
conversion into the methyl 3-oxo-2-[(4-sulfophenyl)hydrazono]pro-
panoate (analytic via LC–MS) to the crude intermediate 2,3-dihy-
Experimental Section
Chemistry
Materials and general procedures: Unless otherwise noted, sol-
vents and reagents were reagent grade from commercial suppliers
and used without further purification. Dry solvents were purchased
and stored over molecular sieves. Reactions were monitored by
TLC on silica gel 60 F254 plates with UV detection (l 254 nm). The
purity of all compounds was determined by analytical HPLC-MS to
be >95% (254 nm), which was conducted with an Agilent 1100
series HPLC equipped with a diode array detector and coupled
with a single quadrupole mass spectrometer with an ESI source
(Agilent Technologies). For HPLC/ToF–MS measurements, an Agi-
lent 1200 series HPLC equipped with a diode array detector was
used, coupled to an ESI source and Accurate Mass Time-of-Flight
mass spectrometer 6220 from Agilent Technologies. For analytical
separations an Agilent Zorbax Eclipse XDB-C18 Rapid Resolution
HT column was used (50”4.6 mm with 1.8 mm particle size). As elu-
ents, mixtures of water and acetonitrile were used with 0.1%
formic acid. For preparative separations, an 1100 series HPLC (Agi-
lent Technologies) was used, equipped with a multi-wavelength
detector. Mixtures of water and acetonitrile were used as eluents,
with the addition of 0.1% trifluoroacetic acid. 1H and 13C NMR
spectra were obtained on a Bruker Avance 300 MHz spectrometer,
analyzed with Topspin 2.0.a, and were referenced to the solvent
residual peak or tetramethylsilane (TMS).
dro(benzo[1,4]dioxin-6-yl)hydrazine
hydrochloride
(1.2 equiv,
0.25 mmol) was added. The reaction mixture was stirred for anoth-
er 16 h at 858C under reflux. The formed precipitate was separated
by centrifugation and washed three times with EtOH and once
with hexane. Afterward the solid was dried at high vacuum to
yield compound 23 (83.7 mg, 75%) as a brown solid; mp: 281–
2838C; 1H NMR (300 MHz, [D6]DMSO): d=9.90 (s, 1H, NH), 8.47–
8.40 (m, 4H, arom), 7.72–7.51 (m, 5H, arom), 7.01–6.81 (m, 2H,
arom), 4.29 ppm (m, 4H, 2”CH2); 13C NMR (75 MHz, [D6]DMSO): d=
161.0, 153.9, 146.1, 143.4, 140.4, 136.1, 134.6, 131.3, 128.1, 127.2,
126.5, 124.2, 120.6, 118.5, 118.1, 116.2, 112.1, 64.17 ppm;
C23H17N5O8S [M], 523.47 Da, HRMS (ESI): calcd m/z 524.0871 [M+
H]+, found m/z 524.0873 [M+H]+.
4-{(2Z)-2-[1-(1,3-Benzodioxol-5-yl)-3-(4-nitrophenyl)-5-oxo-1,5-di-
hydro-4H-pyrazol-4-ylidene]hydrazino}benzenesulfonic acid (24):
Methyl 2,2-dihydroxy-3-(4-nitrophenyl)-3-oxopropanoate II (1 equiv,
0.21 mmol) was dissolved in EtOH (10 mL), concentrated HCl (1%
in solution) and 4-hydrazinylbenzenesulfonic acid hemihydrate
(1.2 equiv, 0.25 mmol) were added. The reaction mixture was
stirred for 16 h at 858C under reflux. After complete conversion
ChemMedChem 2015, 10, 815 – 826
www.chemmedchem.org
823
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
into the methyl 3-oxo-2-[(4-sulfophenyl)hydrazono]propanoate (an-
alytic via LC–MS) to the crude intermediate 1,3-benzodioxol-5-ylhy-
drazine (1.2 equiv, 0.25 mmol) was added. The reaction mixture
was stirred for another 16 h at 858C under reflux. The formed pre-
cipitate was separated by centrifugation and washed three times
with EtOH and once with hexane. Afterward the solid was dried at
high vacuum to yield compound 24 (53.4 mg, 53%) as a brown
lytic center was added as a protein hydrogen bonding constraint.
All docking solutions were post-minimized in their respective bind-
ing site environment using the MMFF94[45] implementation of Li-
gandScout 3.03.[46,47] Analysis and prioritization of the ensemble of
resulting poses was conducted by starting from the most plausible
conformation of 25, subsequently deriving a structure-based 3D
pharmacophore[48] using LigandScout 3.03[46,47] and using the re-
sulting chemical feature set for prioritization of the docking poses.
Figures were created with LigandScout and rendered by POV-Ray
ver. 3.5.1 (www.povray.org).
1
solid; mp: 301–3038C; HNMR (300 MHz, [D6]DMSO): d=8.48–8.40
(m, 6H, arom), 7.68 (d, arom, J=5 Hz), 7.06 (d, arom, J=9 Hz),
6.09 ppm (s, 2H, CH2); 13C NMR (75 MHz, [D6]DMSO): d=161.3,
157.0, 147.7, 145.3, 143.6, 141.2, 128.3, 127.4, 124.3, 119.6, 116.4,
112.6, 101.8, 101,1 ppm; C22H15N5O8S [M], 509.45 Da, HRMS (ESI):
calcd m/z 510.0714 [M+H]+, found m/z 510.0713 [M+H]+.
Biological methods
Enzymatic DiFMUP assay: An enzymatic 6,8-difluoro-4-methylum-
belliferyl phosphate (DiFMUP) assay was used for the determina-
tion of activities of SHP2, SHP1, and PTP1B. Test compounds were
dissolved in dimethyl sulfoxide (DMSO) at a concentration of 10 or
100 mm, and the assay was carried out at a final DMSO concentra-
tion <1%. The DiFMUP assay buffer contained a final concentra-
tion of 50 mm MOPSO (pH 6.5), 200 mm NaCl, 0.03% Tween-20,
0.1% BSA, 1 mm dithiothreitol (DTT, freshly added prior to each
measurement) and 0.003 ngmLÀ1 SHP2 (final concentration). The
final assay volume was 30 mL. Enzyme and test compound in
buffer solution were incubated for 1 h at RT. The reaction was start-
ed by adding DiFMUP to the following final concentrations: 14 mm
for SHP1, 72 mm for SHP2, and 3 mm for PTP1B. These substrate
concentrations correspond to the experimentally determined KM
values of the enzymes. Measurements were performed on
4-{(2Z)-2-[1,3-Bis(4-nitrophenyl)-5-oxo-1,5-dihydro-4H-pyrazol-4-
yliden]hydrazino}benzenesulfonic acid (25): Methyl 2,2-dihy-
droxy-3-(4-nitrophenyl)-3-oxopropanoate II (1 equiv, 0.21 mmol)
was dissolved in EtOH (10 mL), concentrated HCl (1% in solution)
and 4-hydrazinylbenzenesulfonic acid hemihydrate (1.2 equiv,
0.25 mmol) were added. The reaction mixture was stirred for 16 h
at 858C under reflux. After complete conversion into the methyl 3-
oxo-2-[(4-sulfophenyl)hydrazono]propanoate (analysis via LC–MS)
to the crude intermediate 4-nitrophenylhydrazine (97%, ꢀ30%
water as stabilizer; 1.2 equiv, 0.25 mmol) was added. The reaction
mixture was stirred for another 16 h at 858C under reflux. The
formed precipitate was separated by centrifugation and washed
three times with EtOH and once with hexane. Afterward the solid
was dried at high vacuum to yield compound 25 (99.5 mg, 92%)
1
as an orange solid; mp: 304–3068C; H NMR (300 MHz, [D6]DMSO):
a
Genius Pro Reader (SAFIRE II, instrument serial number:
d=9.19 (s, 1H, NH), 8.58–8.49 (m, 8H, arom), 7.76–7.70 ppm (m,
4H, arom); 13C NMR (75 MHz, [D6]DMSO): d=161.7, 157.3, 148.7,
143.2, 141.6, 140.8, 133.1, 128.3, 127.1, 124.1, 117.8, 116.4 ppm;
C21H14N6O8S [M], 510.44 Da, HRMS (ESI): calcd m/z 511.0667 [M+
H]+, found m/z 511.0669 [M+H]+.
512000014) with the following settings: measurement mode: Fluo-
rescence Top; lex: 360 nm (bandwidth 20 nm); lem: 460 nm (band-
width 20 nm); gain (manual): 60; number of reads: 8; FlashMode:
high sensitivity; integration time: 40 ms; lag time: 0 ms; Z-position
(manual): 13900 mm; number of kinetic cycles: 5; kinetic interval:
135 s; total kinetic run time: 9 min. Measurements were performed
in triplicate. IC50 values were calculated with Prism 5 (for Windows,
Version 5.01, GraphPad Software Inc.).
X-ray crystallography and docking protocols
Crystallization and structure determination of SHP2: Crystals of
the catalytic domain of SHP2 were obtained at 7 mgmLÀ1 by sit-
ting-drop vapor diffusion at 208C. The protein solution (400 nL)
was mixed with reservoir solution (400 nL) and equilibrated against
75 mL 16–22% (w/v) PEG3350 and 0.1–0.2m sodium citrate.
Expression and purification of the catalytic domain of human SHP2:
The catalytic domain of human SHP2 (residues 248–527) was puri-
fied from E. coli Rosetta (DE3). The overexpressed protein was
fused to an N-terminal hexahistidine tag and a tobacco etch virus
(TEV) cleavage site in between. Cells were harvested and re-sus-
pended in 50 mm HEPES (pH 7.5), 500 mm NaCl, 5% (v/v) glycerol,
0.5 mm DTT. After cell lysis, debris was removed by centrifugation
at 50000 g for 30 min at 48C. The supernatant was filtered prior to
affinity purification with Ni-NTA (Qiagen). The slurry was washed
with the lysis buffer supplemented with 20 mm imidazole. Reason-
ably pure protein was eluted with 250 mm imidazole. After cleav-
age of the affinity tag purification was completed by size-exclusion
chromatography with a Superdex 75 26/60 column (GE Healthcare)
equilibrated with 20 mm HEPES (pH 7.5), 250 mm NaCl, and 10 mm
DTT.
X-ray diffraction data were collected at BESSY, BL14.1[36] at 100 K.
Statistics for data collection are listed in Supporting Information
Table S1. Data were processed and scaled with XDS and XSCALE.[37]
Initial phases were obtained by molecular replacement with anoth-
er SHP2 structure as search model[21] using Phaser.[38] Automatic
model building was done with ARP/wARP.[39] Refinement with itera-
tive cycles of manual model building with Coot[40] and refinement
with Refmac[41] resulted in the final models (refinement statistics in
Table S2). During refinement the stereochemical quality of the
models was verified with PROCHECK[42] and WHATCHECK.[43] The
agreement between atomic model and diffraction data was validat-
ed with SFCHECK.[44]
Scatter assay: The human pancreatic adenocarcinoma cell line
HPAF II was purchased from the ATCC (ATCC CRL-1997). Cells were
maintained in a 378C, 5% CO2 incubator and sub-cultured by tryp-
sination. The HPAF II cells were maintained in Dulbecco’s modified
Eagle’s medium (DMEM, Gibco BRL, Gaithersburg, MD, USA) con-
taining 4.5 gLÀ1 glucose and supplemented with 10% (v/v) heat-in-
activated fetal bovine serum (FBS, Gibco BRL), and penicillin/strep-
tomycin (100 IU penicillin, 100 mgmLÀ1 streptomycin).
Ensemble docking and pharmacophore analysis: Docking studies
and protein preparation were performed using GOLD v 5.0.1 by
the Cambridge Crystallographic Data Centre (Cambridge, UK) with
default parameters (GoldScore, 100% search efficiency). PDB IDs:
3B7O,[21] 3O5X,[18] and the four crystal structures presented in this
work were aligned in the coordinate frame of 3O5X for ensemble
docking after removing crystal waters from all structures. The
active site was defined using a 20 Š sphere around position (31.10,
3.51, 6.29), and a hydrogen bond to N1574 of Arg465 in the cata-
Cells were dispensed at a density of 650 cells per well in 45 mL
medium (supplemented with 5% FBS) in black-walled, clear-
ChemMedChem 2015, 10, 815 – 826
www.chemmedchem.org
824
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
bottom 384-well cell-bind microplates (Corning Life Sciences,
Acton, MA, USA) and allowed to settle overnight at 378C in a hu-
midified atmosphere of 5% CO2. Test compounds were dissolved
at fixed concentrations in DMSO and added to the medium, ach-
ieving a final DMSO concentration of 0.05% (v/v). Following 1 h in-
cubation (378C, 5% CO2) recombinant human HGF (provided by
Dr. Ermanno Gherardi, University of Cambridge, UK) was added to
a final concentration of 350 pm. The cells were further incubated
at 378C in an atmosphere of 5% CO2 for at least 20 h. Thereafter
the cells were fixed with 4% paraformaldehyde/PBS for 30 min,
and permeabilized with 0.1% Triton X-100/PBS for 3 min. The
nuclei were stained with Hoechst 33342 (10 mm, Molecular Probes,
Carlsbad, CA, USA) and the F-actin with TRITC-conjugated phalloi-
din (0.2 mm, Sigma) for 45 min at room temperature. An Array-
Scan VTI Reader (Thermo Fisher Scientific Inc.) was used to acquire
images with the filter cube XF93 and an excitation filter 365WB50
for Hoechst 33342 and excitation filter 549DF8 for TRITC-conjugat-
ed phalloidin, with detection at 20” magnification (Figure 2a).
Quantitative analysis for the inhibition of scattering was based on
the proximity of neighboring nuclei (Figure 2b). For each well,
nine images were acquired with an autofocus interval of three
fields at 10” magnification in the Hoechst channel. Analysis was
done in the Bioapplication Morphology Explorer ver. 4 with the nu-
cleus of the cell as primary object. The displayed data (Figure 2)
are from one of three independent sets of experiments. Columns
represent the median percentage of cells of three samples having
a minimum neighbor distance greater than the threshold value de-
fined by the positive control experiment (cells stimulated with HGF
but without SHP2 inhibitor). The threshold value for the control ex-
periments was determined as the median of twelve samples.
colonies within a couple of days which were measured by deter-
mining cell viability using a fluorescent dye. Addition of inhibitors
after cell seeding allowed for the analysis of antitumorigenic ef-
fects. As controls, either siRNA (Qiagen HsPTP11_1; #SI00044002;
5’-GCCG CTCA TGAC TATA CGCTA-3’) interfering with the expres-
sion of SHP2, or the cytostatic agent 5-fluorouracil (5-FU) were
used.
For the soft agar assay the following details were applied: cell
medium: Iscove’s modified Dulbeccos’ medium (IMDM) including
20% (v/v) FCS and 0.01% (w/v) gentamicin; bottom layer: 0.75%
(w/v) Bacto Agar in cell medium; intermediate layer with cells: 4”
104 cells in 0.2 mL cell medium including 0.4% (w/v) Bacto Agar;
top layer with test compound: test compound (3” concentration)
in cell medium; 2-(4-iodophenyl)-3-(4-nitrophenyl)-5-phenyltetrazo-
lium chloride: 1 mgmLÀ1, 100 mL per well; 200 mL per well were
plated as bottom layer, and the cell layer was applied on this layer.
Compounds in DMSO were added to the medium after 24 h (final
concentration: 40 mm). The plates were incubated for up to 20
days at 378C and 7.5% CO2. During this period an in vitro colony
growth of >50 mm in diameter was observed. After the maximum
colony formation, the number of colonies was determined using
an automatic image analysis system (BIOREADER 5000 PRO-XI,
Biosys GmbH); 24 h prior to evaluation, live colonies were stained
using a sterile aqueous solution of 2-(4-iodophenyl)-3-(4-nitrophen-
yl)-5-phenyltetrazolium chloride.[49] The assay was considered to be
valid if the following criteria were met: the control shows ꢀ20 col-
onies each having a diameter of >50 mm; the coefficient of varia-
tion of control wells per plate is ꢁ50%, and the positive control 5-
FU shows a decrease in colony number of <30% relative to the
untreated control.[50]
Cell transfection with siRNA and analysis of SHP2 expression: The
human lung adenocarcinoma cell line LXFA 526L (Oncotest, Frei-
burg, Germany) was maintained in a 378C, 5% CO2 incubator in
RPMI 1640 medium supplemented with 10% (v/v) fetal calf serum
(FCS), and penicillin/streptomycin as above. Four different siRNAs
directed against SHP2 (HP Genome Wide siRNA Hs_PTPN11_1(A),_
5(B),_(C),_(D) by Qiagen) in final concentrations of 15 and 50 nm
were added to the cells. Cells were then transfected by adding
4.5 mL HiPerFect (Qiagen). Following incubation for 24 h cells were
stimulated by adding EGF (10 ngmLÀ1), incubated for another 24 h
and then lysed (lysis buffer, Bio-Rad) for analysis of SHP2 expres-
sion by western blot. As negative control, cells were treated in the
same way but without the siRNA. Laemmli buffer was added to
the protein extracts, samples were denatured at 958C for 5 min
and loaded onto a 10% SDS-polyacrylamide gel. SeeBluePlus2 Pre-
Stained Standard (Invitrogen) was used as a protein size reference.
The gel was run at 300 V in standard Tris/glycine/SDS buffer. The
separated proteins were electroblotted onto a PVDF membrane at
90 mA for 1 h. The membrane was blocked with 3% dry milk in
TBS–Tween-20 buffer. For immunodetection of SHP2, a rabbit poly-
clonal anti-human SHP2 antibody (Abcam #ab17753) and a polyclo-
nal goat anti-rabbit HRP-labeled IgG antibody (Cell Signaling
#7074) were applied. HRP-labeled immune complexes were finally
detected by the ECL western blotting substrate (Amersham
#RPN2106) on X-ray films.
Nude mouse xenograft model: Efficacy testing of compound 25 (GS-
493) in a nude mouse xenograft model was performed at Oncotest
GmbH (Freiburg, Germany; study protocol P227H). All experiments
were conducted according to the guidelines of the German Animal
Welfare Act. Briefly, tumor cells were obtained from xenograft
LXFA 526 of a human lung adenocarcinoma in serial passage in
nude mice. Fragments (4–5 mm diameter) of this tumor were sub-
cutaneously implanted in recipient mice (strain NMRI nu/nu;
Charles River, Sulzfeld, Germany). Animals carrying a tumor of ap-
propriate size (mean tumor diameter: 6–8 mm, minimum tumor di-
ameter: 5 mm) were randomly distributed into two groups of five
each receiving either vehicle (DMSO) or compound 25 dissolved in
DMSO once daily i.p. The daily dose for DMSO was 3 mLkgÀ1 (max-
imum tolerable dose) and for GS-493 was 45.93 mg in 3 mL DMSO
per kg. Mice were weighed twice a week. Tumor volumes were de-
termined by two-dimensional measurements twice daily according
to the formula (a”b2)”0.5, for which a represents the largest and
b the perpendicular tumor diameter. The study was conducted for
27 days, after which the animals were sacrificed. During the study
time no acute toxicity was observed in both animal groups.
Acknowledgements
The authors gratefully acknowledge excellent technical support
from Carola Seyffarth and Andreas Oder (Leibniz-Institut für Mo-
lekulare Pharmakologie, Berlin, Germany). Dr. Ermanno Gherardi
(University of Cambridge, UK) provided recombinant human hep-
atocyte growth factor / scatter factor (HGF/SF) for cellular experi-
ments. The authors also thank Dr. Vera Martos (Leibniz-Institut
für Molekulare Pharmakologie) for helpful discussions, and On-
cotest GmbH (Freiburg, Germany) for kind support. J.R. acknowl-
Soft agar assay (tumor cell colony formation assay): A soft agar
assay as described by Hamburger and Salmon[33] was used to de-
termine the antitumorigenic effects (reduction of cell colony for-
mation) of compounds 24 (GS-458) and 25 (GS-493) on cells isolat-
ed from human lung adenocarcinoma LXFA 526L and harvested
from a mouse xenograft.[33] Principally, cells were seeded in semi-
solid agar medium in 24-well plates and then formed multicellular
ChemMedChem 2015, 10, 815 – 826
www.chemmedchem.org
825
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[23] A. Naing, J. M. Reuben, L. H. Camacho, H. Gao, B. N. Lee, E. N. Cohen, C.
Verschraegen, S. Stephen, J. Aaron, D. Hong, J. Wheler, R. Kurzrock, J.
Cancer 2011, 2, 81–89.
[24] J. Yang, L. Liu, D. He, X. Song, X. Liang, Z. J. Zhao, G. W. Zhou, J. Biol.
Chem. 2003, 278, 6516–6520.
edges support of this research by the Deutsche Forschungsge-
meinschaft (DFG) (FOR806, Ra895/5) and the Berlin School of
Integrative Oncology (BSIO), Germany.
[25] S. Weik, T. Luksch, A. Evers, J. Bçttcher, C. A. Sotriffer, A. Hasilik, H. G.
Lçffler, G. Klebe, J. Rademann, ChemMedChem 2006, 1, 445–457.
[26] H. H. Wasserman, C. M. Baldino, S. J. Coats, J. Org. Chem. 1995, 60,
8231–8235.
Keywords: cancer cell mobility
transitions · metastasis · protein tyrosine phosphatases · SHP2
inhibitors
· epithelial–mesenchymal
[27] W. Adam, Y. Chan, D. Cremer, J. Gauss, D. Scheutzow, M. Schindler, J.
Org. Chem. 1987, 52, 2800–2803.
[28] D. F. McCain, L. Wu, P. Nickel, M. U. Kassack, A. Kreimeyer, A. Gagliardi,
D. C. Collins, Z. Y. Zhang, J. Biol. Chem. 2004, 279, 14713–14725.
[29] G. Jones, P. Willett, R. C. Glen, A. R. Leach, R. Taylor, J. Mol. Biol. 1997,
267, 727–748.
[1] S. De Munter, M. Kçhn, M. Bollen, ACS Chem. Biol. 2013, 8, 36–45.
[2] B. G. Neel, H. Gu, L. Pao, Trends Biochem. Sci. 2003, 28, 284–293.
[3] K. S. Grossmann, M. Rosµrio, C. Birchmeier, W. Birchmeier, Adv. Cancer
Res. 2010, 106, 53–89.
[4] P. Hof, S. Pluskey, S. Dhe-Paganon, M. J. Eck, S. E. Shoelso, Cell 1998, 92,
441–450.
[30] G. Jones, P. Willett, R. C. Glen, J. Mol. Biol. 1995, 245, 43–53.
[31] M. Stoker, J. Cell. Physiol. 1989, 139, 565–569.
[5] D. Barford, B. Neel, Structure 1998, 6, 249–254.
[32] R. Krumbach, J. Schüler, M. Hofmann, T. Giesemann, H. H. Fiebig, T.
Beckers, Eur. J. Cancer 2011, 47, 1231–1243.
[6] M. Rosµrio, W. Birchmeier, Trends Cell Biol. 2003, 13, 328–335.
[7] M. Tartaglia, C. M. Niemeyer, A. Fragale, X. Song, J. Buechner, A. Jung, K.
Hählen, H. Hasle, J. D. Licht, B. D. Gelb, Nat. Genet. 2003, 34, 148–150.
[8] M. Tartaglia, S. Martinelli, G. Cazzaniga, V. Cordeddu, I. Iavarone, M.
Spinelli, C. Palmi, C. Carta, A. Pession, M. Aricó, G. Masera, G. Basso, M.
Sorcini, B. D. Gelb, A. Biondi, Blood 2004, 104, 307–313.
[9] H. Higashi, R. Tsutsumi, S. Muto, T. Sugiyama, T. Azuma, M. Asaka, M. Ha-
takeyama, Science 2002, 295, 683–686.
[33] A. W. Hamburger, S. E. Salmon, Science 1977, 197, 461–463.
[34] S. Grosskopf, J. Rademann, C. Eckert, W. Birchmeier (Max Delbrück
Center for Molecular Medicine (MDC) and Forschungsverbund Berlin
e.V. (FVB), Germany), Int. PCT Pub. No. WO 2012041524, 2011.
[35] B. Yu, W. Liu, W. M. Yu, M. L. Loh, S. Alter, O. Guvench, A. D. MacKerell,
L. D. Tang, C. K. Qu, Mol. Cancer Ther. 2013, 12, 1738–1748.
[36] U. Heinemann, K. Bussow, U. Müller, P. Umbach, Acc. Chem. Res. 2003,
36, 157–163.
[10] T. Meyer-ter-Vehn, A. Covacci, M. Kist, H. L. Pahl, J. Biol. Chem. 2000,
275, 16064–16072.
[37] W. Kabsch, J. Appl. Crystallogr. 1993, 26, 795–800.
[38] A. J. McCoy, R. W. Grosse-Kunstleve, P. A. Adams, M. Winn, L. C. Storoni,
R. J. Read, J. Appl. Crystallogr. 2007, 40, 658–674.
[39] A. Perrakis, M. Harkiolaki, K. S. Wilson, V. S. Lamzin, Acta Crystallogr. Sect.
D 2001, 57, 1445–1450.
[40] P. Emsley, K. Cowtan, Acta Crystallogr. Sect. D 2004, 60, 2126–2132.
[41] G. N. Murshudov, A. A. Vagin, E. J. Dodson, Acta Crystallogr. Sect. D
1997, 53, 240–255.
[42] R. A. Laskowski, M. W. MacArthur, D. S. Moss, J. M. Thornton, J. Appl.
Crystallogr. 1993, 26, 283–291.
[11] M. Tartaglia, B. D. Gelb, Annu. Rev. Genomics Hum. Genet. 2005, 6, 45–
68.
[12] M. Tartaglia, E. L. Mehler, R. Goldberg, G. Zampino, H. G. Brunner, H.
Kremer, I. van der Burgt, A. H. Crosby, A. Ion, S. Jeffrey, K. Kalidas, M. A.
Patton, R. S. Kucherlapti, B. D. Gelb, Nat. Genet. 2001, 29, 465–468.
[13] R. J. Gorlin, R. C. Anderson, J. H. Moller, Birth Defects Orig. Artic. Ser.
1971, 7, 110–115.
[14] M. C. Digilio, G. Pacileo, A. Sarkozy, G. Limongelli, E. Conti, F. Cerrato, B.
Marino, A. Pizzuti, R. Calabró, B. Dallapiccola, Birth Defects Res. Part A
2004, 70, 95–98.
[43] R. W. Hooft, G. Vriend, C. Sander, E. E. Abola, Nature 1996, 381, 272.
[44] A. A. Vaguine, J. Richelle, S. J. Wodak, Acta Crystallogr. Sect. D 1999, 55,
191–205.
[45] T. A. Halgren, J. Comput. Chem. 1999, 20, 730–748.
[46] G. Wolber, T. Langer, J. Chem. Inf. Model. 2005, 45, 160–169.
[47] G. Wolber, A. A. Dornhofer, T. Langer, J. Comput.-Aided Mol. Des. 2006,
20, 773–788.
[48] T. Seidel, G. Ibis, F. Bendix, G. Wolber, Drug Discovery Today Technol.
2010, 7, e221–e228.
[49] M. C. Alley, C. B. Uhl, M. M. Lieber, Life Sci. 1982, 31, 3071–3078.
[50] C. Heidelberger, N. K. Chaudhuri, P. Danneberg, D. Mooren, L. Griesbach,
R. Duschinsky, R. J. Schnitzer, E. Pleven, J. Scheiner, Nature 1957, 179,
663–666.
[15] B. Keren, A. Hadchouel, S. Saba, Y. Sznajer, D. Bonneau, B. Leheup, O.
Boute, D. Gaillard, D. Lacombe, V. Layet, S. Marlin, G. Mortier, A. Toutain,
C. Beylot, C. Baumann, A. Verloes, H. CavØ, J. Med. Genet. 2004, 41, e117.
[16] K. Hellmuth, S. Grosskopf, C. T. Lum, M. Würtele, N. Rçder, J. P. von Kries,
J. Rademann, W. Birchmeier, Proc. Natl. Acad. Sci. USA 2008, 105, 7275–
7280.
[17] P. Huang, J. Ramphal, J. Wei, C. Liang, B. Jallal, G. McMahon, C. Tang,
Bioorg. Med. Chem. 2003, 11, 1835–1849.
[18] L. Chen, S. S. Sung, M. L. Yip, H. R. Lawrence, Y. Ren, W. C. Guida, Mol.
Pharmacol. 2006, 70, 562–570.
[19] X. Zhang, Y. He, S. Liu, Z. Yu, Z. X. Jiang, Z. Yang, Y. Dong, S. C. Nabinger,
L. Wu, A. M. Gunawan, L. Wang, R. J. Chan, Z.-Y. Zhang, J. Med. Chem.
2010, 53, 2482–2493.
[51] L. Lan, J. D. Holland, J. Qi, S. Grosskopf, R. Vogel, B. Gyçrffy, A. Wulf-
Goldenberg, W. Birchmeier, EMBO J. 2015; DOI: 10.15252/
embj.201489004.
[20] R. He, L. F. Zeng, Y. He, S. Zhang, Z.-Y. Zhang, FEBS J. 2013, 280, 731–
750.
[21] A. J. Barr, E. Ugochukwu, W. H. Lee, O. N. F. King, P. Filippakopoulos, I.
Alfano, P. Savitsky, N. A. Burgess-Brown, S. Müller, S. Knapp, Cell 2009,
136, 352–363.
Received: January 15, 2015
[22] M. F. Schmidt, M. R. Groves, J. Rademann, ChemBioChem 2011, 12,
2640–2646.
Published online on April 15, 2015
ChemMedChem 2015, 10, 815 – 826
www.chemmedchem.org
826
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim