Identifying Lysophosphatidic Acid Acyltransferase β (LPAAT-β) as the Target of a Nanomolar Angiogenesis Inhibitor from a Phenotypic Screen Using the Polypharmacology Browser PPB2

Article abstract of DOI:10.1002/cmdc.201800554

By screening a focused library of kinase inhibitor analogues in a phenotypic co-culture assay for angiogenesis inhibition, we identified an aminotriazine that acts as a cytostatic nanomolar inhibitor. However, this aminotriazine was found to be completely inactive in a whole-kinome profiling assay. To decipher its mechanism of action, we used the online target prediction tool PPB2 (http://ppb2.gdb.tools), which suggested lysophosphatidic acid acyltransferase β (LPAAT-β) as a possible target for this aminotriazine as well as several analogues identified by structure–activity relationship profiling. LPAAT-β inhibition (IC₅₀ ≈15 nm) was confirmed in a biochemical assay and by its effects on cell proliferation in comparison with a known LPAAT-β inhibitor. These experiments illustrate the value of target-prediction tools to guide target identification for phenotypic screening hits and significantly expand the rather limited pharmacology of LPAAT-β inhibitors.

Full text of DOI:10.1002/cmdc.201800554

DOI: 10.1002/cmdc.201800554
Full Papers
Identifying Lysophosphatidic Acid Acyltransferase b
(LPAAT-b) as the Target of a Nanomolar Angiogenesis
Inhibitor from a Phenotypic Screen Using the
Polypharmacology Browser PPB2
Marion Poirier,[a] Mahendra Awale,[a] Matthias A. Roelli,[b] Guy T. Giuffredi,[a] Lars Ruddigkeit,[a]
Lasse Evensen,[c] Amandine Stooss,[b] Serafina Calarco,[b] James B. Lorens,[c] Roch-
Philippe Charles,*[b] and Jean-Louis Reymond*[a]
By screening a focused library of kinase inhibitor analogues in
a phenotypic co-culture assay for angiogenesis inhibition, we
identified an aminotriazine that acts as a cytostatic nanomolar
inhibitor. However, this aminotriazine was found to be com-
pletely inactive in a whole-kinome profiling assay. To decipher
its mechanism of action, we used the online target prediction
tool PPB2 (http://ppb2.gdb.tools), which suggested lysophos-
phatidic acid acyltransferase b (LPAAT-b) as a possible target
for this aminotriazine as well as several analogues identified by
structure–activity relationship profiling. LPAAT-b inhibition (IC50
ꢀ15 nm) was confirmed in a biochemical assay and by its ef-
fects on cell proliferation in comparison with a known LPAAT-b
inhibitor. These experiments illustrate the value of target-pre-
diction tools to guide target identification for phenotypic
screening hits and significantly expand the rather limited phar-
macology of LPAAT-b inhibitors.
Introduction
The availability of large databases of compound-target associa-
tions such as ChEMBL[1] has allowed the development of com-
putational tools to predict the target of any given screening
hit based on its structural similarity to know bioactive com-
pounds.[2] Herein we report the application of this approach to
a nanomolar hit identified in a phenotypic screen for angio-
genesis inhibition.
search for new angiogenesis inhibitors we assembled a small li-
brary of kinase inhibitor analogues and tested their activity in
an angiogenesis co-culture assay.[9] The assay revealed that
aminotriazine 1 exerted a nanomolar inhibition of angiogene-
sis (Figure 1). We observed a similar activity pattern with fur-
Angiogenesis, that is, the formation of new blood vessels, is
a key biological process during embryogenesis[3] and wound
healing.[4] Angiogenesis inhibition is an important target to ad-
dress diabetic retinopathy[5] and cancer,[6] in particular by inter-
fering with the vascular endothelial growth factor receptor 2
(VEGFR-2),[7] but also with other receptor tyrosine kinases such
as Axl for which several inhibitors have been investigated.[8] To
Figure 1. Structure of aminotriazine 1.
[a] M. Poirier, Dr. M. Awale, Dr. G. T. Giuffredi, Dr. L. Ruddigkeit,
Prof. Dr. J.-L. Reymond
ther triazines identified in a structure–activity relationship
(SAR) study. However whole kinome profiling did not reveal
any significant kinase inhibition by aminotriazine 1, excluding
inhibition of kinases such as VEGFR-2 or Axl as the mechanism
of action. We therefore turned our attention to computational
target predictions using PPB2 (Polypharmacology Browser 2.0),
which predicts possible targets for any small molecule based
on ChEMBL data (http://ppb2.gdb.tools).[10] Besides several kin-
ases, PPB2 suggested that lysophosphatidic acid acyl transfer-
ase b (LPAAT-b) might be the target of our triazines due to a
similarity to a known LPAAT-b inhibitor. Target assignment was
confirmed in a biochemical assay and was also consistent with
the effect on whole cells in comparison with a known LPAAT-b
inhibitor.
Department of Chemistry and Biochemistry, National Center of Competence
in Research NCCR TransCure, University of Bern, Freiestrasse 3, 3012 Bern
(Switzerland)
E-mail: jean-louis.reymond@dcb.unibe.ch
[b] Dr. M. A. Roelli, A. Stooss, S. Calarco, Prof. Dr. R.-P. Charles
Institute of Biochemistry and Molecular Medicine, National Center of Com-
petence in Research NCCR TransCure, University of Bern, Bꢀhlstrasse 28,
3000 Bern 9 (Switzerland)
E-mail: roch-philippe.charles@ibmm.unibe.ch
[c] Dr. L. Evensen, Prof. Dr. J. B. Lorens
Department of Biomedicine, Centre for Cancer Biomarkers (CCBIO), Univer-
sity of Bergen, Jonas Lies vei 91, 5009 Bergen (Norway)
Supporting information and the ORCID identification number(s) for the
author(s) of this article can be found under:
https://doi.org/10.1002/cmdc.201800554.
ChemMedChem 2018, 13, 1 – 14
1
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Results and Discussion
Table 1. Cytotoxicity of triazines on HeLa cells.
Compd[a]
R1
R2
IC50 [nm][b]
Discovery of an angiogenesis inhibitor with potent cytotox-
icity but unknown target
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Et
Cl
OMe
iPr
nPr
nHex
Ph
F
Br
CF3
NHAc
Et
Et
OMe
F
Cl
Br
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Et
nBu
Et
Et
Et
Et
111Æ12
252Æ7
190Æ3
814Æ20
To search for new angiogenesis inhibitors, we selected 201
compounds from commercial catalogs due to their 3D-shape
and pharmacophore similarity to four known receptor tyrosine
kinase inhibitors (Supporting Information Figure S1a) using our
previously reported algorithm xLOS,[11] an approach which had
served us well previously to discover Aurora A kinase inhibi-
tors.[12] We then screened our focused library for angiogenesis
inhibition activity in a co-culture assay in which kinase inhibi-
tors are detected because they disrupt cellular network forma-
tion.[9] This phenotypic screen pointed to aminotriazine 1 as a
particularly potent compound (Figure 1). Triazine 1 disrupted
cellular network formation in the low nanomolar range, and in-
duced strong underdevelopment of zebrafishes at 500 nm
under conditions where receptor tyrosine kinase inhibitors
show an effect on neovascularization in the mm range after
24 h (Figure S1b–c).[13]
>10000
>10000
>10000
3060Æ110
264Æ5
263Æ6
2920Æ60
51Æ1
780Æ30
12.4Æ0.2
4550Æ940
140Æ8
69Æ13
CF3
Et
110Æ56
[a] See Scheme 1. [b] HeLa cells (10000 cells per well) were incubated
with various concentrations of compounds for 96 h at 378C and 5% CO2.
Cell viability was determined after addition of WST-8 working solution
and phenazine ethosulfate. Final absorbance was measured at 450 nm.
Results were normalized to control values; data are the meanÆSD (stan-
dard deviation) from triplicate measurements.
Our focused library contained 33 triazines related to 1. The
screening data with these triazines indicated that activity re-
quired a methyl substituent on the furan ring and a para sub-
stituent on the aniline ring (Table S1). To further elucidate the
SAR of this compound series, we synthesized seventeen addi-
tional triazines (2–18) with this substitution pattern. Following
known procedures,[14] condensation of p-substituted anilines
with N-cyanoguanidine gave phenyl biguanide intermediates
19–29, which were condensed with methyl chloroacetate to
form chloromethyl triazine intermediates 30–40. These inter-
mediates were finally coupled with o-hydroxy acetophenone
and analogues to form the final products (Scheme 1). Activity
screening for cytotoxicity on HeLa cells revealed three addi-
tional strongly cytotoxic triazines bearing an ethyl substituent
on the furan ring and an ethyl (12), methoxy (14) or chloro
(16) substituent on the aniline ring (Table 1).
The selection of test compounds by similarity to known tyro-
sine kinase inhibitors and the structural resemblance of hit 1
and analogues to triazine type kinase inhibitors[15] suggested
that angiogenesis inhibition and cytotoxicity might result from
inhibition of one or more kinases as originally expected. To
check this hypothesis whole kinome profiling was performed
with 1 and the most potent analogue 14 at 10 mm, which
should reveal any possible interaction with kinases. However,
both compounds turned out almost completely inactive. The
original triazine 1 only gave a weak inhibition signal with JAK1,
PIP5K2C and TYK2. In the case of 14, three kinases gave sub-
micromolar IC50 values, namely JAK1, EPHB6 and GAK. Given
the high intracellular ATP concentration however these values
were far too weak to explain the observed nanomolar activity
on cells (Table 2, Figures S2 and S3).
Target prediction with PPB2
To search for additional possible targets besides kinases that
might explain the observed nanomolar biological effects, we
performed target prediction with our recently reported target
prediction tool PPB2,[10] which proposes possible targets for
any drug-like compound by similarity to compounds with
known activities from the ChEMBL database based on three
different molecular fingerprints, namely MQN (Molecular Quan-
tum Numbers, representing composition),[16] Xfp (Atom Cate-
gory Extended Atom Pair fingerprint, representing molecular
shape and pharmacophores),[17] and ECfp4 (Extended connec-
tivity fingerprint up to 4 bonds, representing detailed substruc-
tures).[18] PPB2 offers eight search options: 1–3) nearest-neigh-
bor (NN) searches with each of the three fingerprints; 4–6) NN
searches combined with a na¨ıve Bayes machine learning
Scheme 1. Synthesis of triazines 1–18: a) N-cyanoguanidine, HCl, 2-BuOH,
1008C, 24 h; b) ClCH2CO2Me, NaOMe, MeOH, 258C, 2 h, then HCl, H2O, 258C,
15 h; c) K2CO3, DMF, 1558C, 2 h.
&
ChemMedChem 2018, 13, 1 – 14
www.chemmedchem.org
2
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
We analyzed the predicted targets for the original triazine 1
as well as for its analogues 12, 14 and 16, which covered three
different substituents on the para position of the aniline ring
as well as the methyl and ethyl substitution of the furan ring.
In each case we used each of the eight search methods avail-
able in PPB2 and analyzed the top predicted target in terms of
the most similar ChEMBL compound for this target as given by
PPB2, which was in general representative of the compound
class associated with this target (Table 3, Figure 2).
Table 2. Kinome scan with triazines 1 and 14.[a]
Compd
Kinase
Percent of
control at
10 mm
K
D [nm]
1
JAK1 (JH2 domain pseudopocket)
PIP5K2C
4.5
16
31
n.d.
n.d.
n.d.
TYK2 (JH2 domain pseudopocket)
14
JAK1 (pseudokinase)
4.2
5.3
6
380
860
3100
EPHB6
PIKFYVE
GRK7
TAK1
IRAK3
GAK
In the two MQN driven target prediction options available in
PPB2 all target associated molecules were substantially differ-
ent from our triazines except for 44 associated with the adeno-
sine receptor 2a (A2aR). However, the o- and m-substitution of
the aniline ring in 44 contradicted our observation that angio-
genesis inhibition and cytotoxicity by our triazines only oc-
curred with p-substituted anilines.
6
>60000
>60000
3500
7.2
12
15
15
490
1800
TYK2 (JH2 domain pseudokinase)
[a] The kinome scan was performed for binding of the ATP binding pock-
ets on 442 kinases by DiscoveRx (San Diego, CA, USA); the test com-
pound concentration was 10 mm.
The two Xfp-based predictions in PPB2 by contrast were
indeed based on triazines resembling our query molecules, in
particular compound 43 indicating lysophosphatidic acid acyl
transferase b (LPAAT-b/AGPAT2) as target and featuring a p-
bromo-substituted aniline, as well as its m-substituted ana-
logue 52. Although the benzofuran ring was absent in both
model based on ECfp4 (NN+NB); 7–8) na¨ıve Bayes (NB) ma-
chine learning or deep neural network (DNN) based on ECfp4.
Table 3. Target predictions with PPB2.[a]
Method/Compd
1
12
14
16
NNMQN +NBECfp4
NNXfp +NBECfp4
NNECfp4 +NBECfp4
NNMQN
NNXfp
NNECfp4
A2aR (42)
A2aR (44)
VEGFR-2 (45)
Tyrosine-protein kinase BTK (41)
LPAAT-b (43)
A2aR (48)
A2aR (44)
LPAAT-b (52)
VEGFR-2 (54)
Tyrosine-protein kinase SRC (50)
VEGFR-2 (51)
JAK2 (53)
A2aR (44)
LPAAT-b (43)
A2aR (44)
Vanilloid receptor (56)
LPAAT-b (43)
A2aR (44)
LPAAT-b (43)
VEGFR-2 (49)
5-HT2c (47)
LPAAT-b (43)
A2aR (44)
A2aR (44)
NBECfp4
DNNECfp4
PI3-kinase p110-a subunit (46)
A2aR (44)
PI3-kinase p110-a subunit (46)
A2aR (44)
Galanin receptor 2 (55)
A2aR (44)
LPAAT-b (57)
LPAAT-b (57)
[a] For each structure and search method the first target predicted is listed, and the most similar ChEMBL molecule to the query in the target-associated
list is given (see structures in Figure 2).
Figure 2. Structures of target-associated ChEMBL compounds selected by PPB2 as the closest analogues of triazines 1, 12, 14, and 16 (Table 3).
ChemMedChem 2018, 13, 1 – 14
www.chemmedchem.org
3
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
compounds and replaced by a substituted benzene ring, the
overall pharmacophoric features of 43 and 52 strongly resem-
bled those of our triazines.
Finally, the four ECfp4 driven predictions of PPB2 pointed to
molecules containing substructures common to the queries
but which lacked an overall agreement in pharmacophore,
making the predicted targets rather unlikely. Furthermore, the
indication of VEGFR-2 (45, 49, 54), JAK2 kinase (53) and PIP3-
kinase (46) as possible targets, while compatible with the ob-
served biological effects, contradicted the very weak kinase in-
hibition by our triazines observed in the kinome scan. One no-
table exception in the ECfp4 prediction was compound 57 fea-
turing a close analogue of 43 and indicating again LPAAT-b.
Figure 3. Activity of triazines 1, 16 and 43 on LPAAT-b by quantifying the re-
lease of free CoA-SH in the presence of DTNB monitored by the change in
absorbance at 405 nm over 3 min at room temperature (200 mm of 18:1-
CoA, 200 mm sn-1-18:1 lysoPA, 500 mm DTNB, 25 mm HEPES·HCl, pH 7.5,
1 mm EDTA, 100 mm NaCl and various concentrations of inhibitors).
SD=standard deviation from triplicate measurements.
Experimental validation
Among the various targets predicted by PPB2 we first tested
possible inhibition of VEGFR-2 due to the strong association of
this target with angiogenesis biology, using triazines 1 and 16
as test cases. These compounds did not have any significant
effect on VEGFR-2 when tested at 10 mm, excluding VEGFR-2
inhibition as a possible explanation for the observed cellular
effects. The compounds also showed no significant inhibition
on the adenosine receptor A2a, showing that association with
triazine 44 did not yield a valid prediction. Several additional
targets were experimentally tested that were indicated by
PPB2 upon inspection of the five top targets predicted for 1
and 16 in PPB2, giving no significant inhibition (Table S2 and
S3). The experimental invalidation of these off-targets left
LPAAT-b as the most probable target of our triazines.
We were further interested in comparing the effect of our
most effective compound 14 with the known inhibitor 43 on
the cell cycle of MEF and SK-OV-3 cells at concentrations that
inhibited growth in the proliferation assay. Strikingly, both
compounds triggered a G2 arrest in both cell lines (Figure 4d).
Due to the shorter cell cycle duration of MEF cells, the ob-
served effect was more dramatic in this cell line. Both com-
pounds had very similar effects on the cell cycle, suggesting
that they engage in a similar mechanism. Considering that the
reference compound 43 has been shown to inhibit ERK and
AKT pathways in several cancer models,[20,21] we also observed
the effect of the two compounds on the activity of the RAF-
MEK-ERK and the PI3’K-AKT pathways by measuring levels of
ERK and AKT phosphorylation, respectively (Figure 4e). As con-
trol compounds, we used PD-325901 (a potent MEK inhibitor)
and GDC-0941 (a pan-class I PI3’K inhibitor) to target the RAF-
MEK-ERK and the PI3’K-AKT pathways, respectively. Whereas
we could not observe an effect on ERK phosphorylation at the
used doses, AKT phosphorylation was significantly reduced by
25% with both compounds. These data replicate results from
earlier publications using 43,[20,21] and further solidifies our hy-
pothesis that our triazines (e.g. 14) and 43 act on the same
pathway. The assignment of LPAAT-b as their target, supported
by our biochemical assay discussed above, is also consistent
with the results of LPAAT-b deletion by siRNA on cells.[22]
The source literature associated with ChEMBL compounds
43, 52 and 57 as LPAAT-b inhibitors reported that these com-
pounds were cytotoxic at levels similar to our triazines.[19] We
synthesized a pure sample of 43 following the published pro-
cedure, and indeed observed strong cytotoxicity in cells to the
reported level. To confirm that LPAAT-b was a target of our tria-
zines, we measured their activity on LPAAT-b directly by ex-
pressing this protein in oocytes and measuring its catalytic ac-
tivity by quantifying the free coenzyme A (CoA-SH) released by
catalyzed acyl transfer from palmitoleyl-CoA colorimetrically
using 5,5’-dithiobis(2-nitrobenzoic acid) (DTNB).[20] The mea-
surement was performed with triazines 1 and 16 and gave IC50
values in the low nanomolar range compatible with their ob-
served effects on cells, and even stronger than the LPAAT-b in-
hibition measured with the reference inhibitor 43, which was
similar to the reported values for this compound (Figure 3).
To gain a closer insight into the biological effects of our tria-
zines, we evaluated cell proliferation of our most potent tria-
zines 1, 12, 14 and 16 in comparison with the known LPAAT-b
inhibitor 43 by subjecting MEF (noncancerous cells) and SK-
OV-3 (ovarian cancer cells), as well as CHO and HEK-293 cells
to increasing concentrations of the respective compounds. The
expression of LPAAT-b was confirmed on the mRNA level by re-
verse transcription PCR (Figure 4a) as well as on the protein
level by western blot staining (Figure 4b). Proliferation in all
cell lines was inhibited at similar levels (Figure 4c, Table 4).
Comparing PPB2 with other target prediction tools
As a comparison we tested whether LPAAT-b might also be as-
signed as the target of our triazines by a selection of available
public target prediction websites. We compared the prediction
of PPB2 with those of PASS,[23] SEA,[24] HitPick,[25] Modlab
Spider,[26] SwissTargetPrediction,[27] SuperPred,[28] and PPB
(Table 5 and Table S4).[29] The differences in predictions reflect a
combination of the search method and the list of targets con-
sidered by each tool. For PPB2 LPPAT-b consistently ranked
first for all four test compounds with methods using our atom-
pair based pharmacophore fingerprint Xfp. Results were also
excellent with PASS, one of the oldest online target prediction
&
ChemMedChem 2018, 13, 1 – 14
www.chemmedchem.org
4
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Figure 4. a) Reverse transcription PCR. The gel demonstrates that all the used cell lines contain RNA transcripts for LPAAT-b. GAPDH was used for control. The
positive control is a pool of cDNA coming from all the cell lines used. The negative control being a RT-PCR without cDNA. b) Western blot confirming the ex-
pression of LPAAT-b protein in all cell lines used. b-Actin was used as a loading control. c) CHO, MEF, SK-OV-3 and HEK-293 cells, respectively were subjected
to increasing doses of 1, 12, 14, 16 and 43. Dose response was measured by crystal violet nuclei staining. IC50 ÆSD values are listed in Table 4. d) Effect of 43
and 14 on the cell cycle given as percentage of cells in G1, S or G2 phases. Doses of drugs were selected to be growth inhibitory. For MEF cells: 43: 260 nm;
14: 200 nm. For SK-OV-3 cells: 43: 260 nm; 14: 70 nm. e) Effect of 43 and 14 on the activity of the MAPK pathway (ERK phosphorylation) and the PI3’K-AKT
pathway (AKT phosphorylation) of SK-OV-3 cells. The same doses were used as in panel (d).
ChemMedChem 2018, 13, 1 – 14
www.chemmedchem.org
5
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Table 4. Cell growth inhibition.
Table 6. Structural similarity and ranking of LPAAT-b inhibitor 43 and
A2aR inhibitor 44 relative to triazine 1.
Cells
IC50 [nm][a]
1
12
14
16
43
Method
43
Similarity[a]
44
Similarity[a]
Rank[b]
Rank[b]
CHO
HEK-293
MEF
63Æ9
25Æ7
48Æ19
79Æ43
32Æ5
19Æ5
20Æ4
58Æ18
23Æ5
11Æ1
23Æ3
18Æ5
102Æ32
51Æ10
94Æ25
37Æ15
42 Æ23
92Æ83
ECfp4
FCfp4
Sfp
Xfp
MQN
0.39
0.28
0.47
167
24
4
199
17
0.55
0.70
0.84
355
11.00
1
1
1
34
71
62Æ11
SK-OV-3
179Æ139
1
[a] CHO, MEF, SK-OV-3 and HEK-293 cells were subjected to increasing
doses of compounds 1, 12, 14, 16 and 43. Dose response was measured
by crystal violet nuclei staining, data from Figure 4c. Values are the
meanÆSD (standard deviation) from triplicate measurements.
7378
Maccs[c]
0.64
0.79
0.78
1235
1250
1235
0.75
0.86
0.85
54
51
54
RDkitfp[c]
Morgan[c]
0.47
0.65
0.64
14
8
14
0.83
0.91
0.90
1
1
1
tools, which indicated LPAAT-b at the first rank (14) or second
rank (1, 12, 16). PASS predicts targets with a Bayesian probabil-
istic model built using the MNA (Multilevel Neighborhood of
Atoms) fingerprint and its own curated database of bioactive
compounds.[30] The SEA method gave LPAAT-b at rank 1 with
16, but the target appeared lower in the list for the other com-
pounds. The target was also found with SwissTargetPrediction
and PPB, however much lower in the list. The other methods
tested did not identify LPAAT-b although all but Modlab Spider
contain compound 43 and LPAAT-b as target.
0.37
0.54
0.53
6
6
6
0.56
0.72
0.71
1
1
1
[a] Similarity to 1 as Tanimoto coefficient (1=identity), City-block distance
(0=identity, for Xfp and MQN), Cosine (1=identity, 2nd line of Maccs,
RDkitfp and Morgan), or Dice (1=identity, 3rd line of Maccs, RDkitfp and
Morgan). [b] Rank in ChEMBL subset of PPB2 sorted by similarity to 1.
[c] Maccs, RDkitfp, and Morgan are standard molecular fingerprint from
the RDkit cheminformatics software.
This analysis shows that identifying LPAAT-b as the target of
triazine 1 could have been more difficult or even impossible
depending on which tool we would have used. It should be
noted that the similarity of triazine 1 to the known LPAAT-b in-
hibitor 43 is lower than to the known A2aR inhibitor 44 with
most fingerprints except Xfp, and that many other molecules
with high similarity to 1 that are not known as LPAAT-b inhibi-
tors are listed in ChEMBL. As a consequence, compound 43 in-
dicating the correct target only appears quite late in a sorted
list of ChEMBL molecules, irrespective of the similarity metrics
used (Table 6). In that respect it is not surprising that many of
the online target prediction tools tested would not have cor-
rectly assigned LPAAT-b as the target of triazine 1.
The difficulty of target prediction is further illustrated by vis-
ualizing the relationship between LPAAT-b inhibitors 1–18 and
ChEMBL compounds annotated with LPAAT-b, A2aR or VEGFR-
2 in form of a similarity map[31] of the Xfp-chemical space
using WebMolCS (Figure 5).[32] This visualization illustrates that
the correct target assignment was made possible mostly by
the presence of LPAAT-b inhibitor 43 in the target prediction
database, while compounds annotated with either A2aR or
VEGFR-2 were on average more closely related to triazines 1–
18 than other known LPAAT-b inhibitors.
Table 5. Rank of LPAAT-b prediction by selected online target prediction tools.[a]
Tool
Website
1
12
14
16
PPB2 (NN)[b]
PPB2 (NN+NB)[b]
PPB2 (ML)[c]
PASS[d]
SEA
HitPick[e]
Modlab Spider
SwissTargetPrediction[f]
SuperPred[g]
PPB[h]
http://ppb2.gdb.tools
http://ppb2.gdb.tools
http://ppb2.gdb.tools
http://www.way2drug.com/passtargets/
http://sea.bkslab.org/
http://mips.helmholtz-muenchen.de/proj/hitpick
http://modlabcadd.ethz.ch/software/spider
http://www.swisstargetprediction.ch
http://prediction.charite.de
2/1/–
–/2/–
2/–
2
4
–
–
10
–
3/1/–
–/1/–
4/–
2
6
–
–
–
–
9/7/3
–/1/–
–/11
1
8
–
–
8
–
2/1/20
8/1/–
1/1
2
1
–
–
6
–
http://www.gdb.unibe.ch
33
34
36
31
[a] The rank of LPAAT-b in the target list output is given; (–): target not found. [b] Nearest-neighbor search with ECfp4/Xfp/MQN, alone (NN) or combined
with an ECfp4-based na¨ıve Bayes model with the 2000 NN (NN+NB). [c] Machine learning (ML) models: na¨ıve Bayes (NB) or deep neural network (DNN)
built from the entire database using ECfp4 as fingerprint. [d] PASS gives histamine release inhibitor as rank 1 or 2. [e] HitPick gives adenosine receptor
A2aR in all four cases. [f] SwissTargetPrediction gives microtubule-associated protein tau as rank 1 and adenosine receptor A2aR as rank 2. [g] SuperPred
does not give any target. [h] Obtained with scanning 40 targets per method. PPB incorporates organisms and all non-human proteins and has a much
longer target list than PPB2. See Supporting Information Table S4 for details of targets.
&
ChemMedChem 2018, 13, 1 – 14
www.chemmedchem.org
6
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
the positive mode recorded on a Thermo Scientific LTQ Orbitrap
XL. 1H and 13C NMR spectra were measured on a Bruker Avance
300 spectrometer (at 300 MHz and 75 MHz, respectively) or on a
Bruker Avance II 400 spectrometer (at 400 MHz and 101 MHz, re-
spectively). 1H and 13C chemical shifts are quoted relative to sol-
vent signals, and resonance multiplicities are reported as s (singlet),
d (doublet), t (triplet), q (quartet), p (pentet), and m (multiplet);
br=broad peak. Compound purities were assessed by analytical
reversed phase HPLC (RP-HPLC) at a detection wavelength of
214 nm. The purity of tested compounds was >95%, unless other-
wise stated (Table S3). Analytical RP-HPLC was performed on a
Dionex Ultimate 3000 RSLC System (DAD-3000 RS Photodiode
Array Detector) using a Dionex Acclaim RSLC 120 column (C18, 3.0ꢁ
50 mm, particle size 2.2 mm, 120 ꢂ pore size) and a flow rate of
1.2 mLminÀ1. Data were recorded and processed with Dionex
Chromeleon Management System (version 6.8) and Xcalibur (ver-
sion 2.2, Thermo Scientific). Eluents for analytical RP-HPLC were as
follows: (A) milliQ-deionized water containing 0.05% TFA, and (D)
HPLC-grade acetonitrile/milliQ-deionized water (9:1) containing
0.05% TFA. Conditions for analytical RP-HPLC were as follows: in
2.2 min from 100% A to 100% D, then staying on 100% D (meth-
od A), in 2.2 min from 50% A/50% D to 100% D, then staying on
100% D (method B), in 7.5 min from 100% A to 100% D, then stay-
ing on 100% D (method C). Melting point were recorded on a
Bꢃchi B-545 apparatus and are uncorrected.
Figure 5. 3D-similarity map of triazines 1–18 (red spheres) and nearest
neighbors of 1 from ChEMBL annotated as LPAAT-b (cyan), A2aR (blue) or
VEGFR-2 (yellow) inhibitors. The interactive color-coded 3D map is available
at http://gdbtools.unibe.ch:8080/webMolCS/yourSIM.html?jo-
bID=1534320217879&fp=Xfp.
Conclusions
Procedure A. A mixture of aniline (1 equiv) and N-cyanoguanidine
(1.1 equiv) in HCl (concd) and iPrOH or 2-BuOH was stirred at
1058C for 24 h. After allowing the mixture to cool to room temper-
ature, the residue was washed with iPrOH and dried in vacuo to
afford the biguanide hydrochloride salts. The procedure was used
for biguanides 19–29 (see Supporting Information).
Screening of a focused library of kinase inhibitor analogues in
a phenotypic assay for angiogenesis revealed triazine 1 as a
nanomolar inhibitor, however this compound showed no sig-
nificant kinase inhibitory activity in a whole kinome profiling
assay. To identify its target we used our target prediction tool
PPB2 with triazine 1 and the three potent analogues 12, 14
and 16 identified by SAR. In each case we performed all the
possible search methods, and visually inspected the ChEMBL
molecules associated with the most probable targets. Predic-
tions of A2aR and VEGFR-2, which were not convincing due to
a poor overall similarity between our triazines and the corre-
sponding ChEMBL molecules 44 and 49, were experimentally
disproved. PPB2 also predicted that our triazines might inhibit
LPAAT-b due to their overall similarity to the related triazine
43. Indeed, a direct biochemical assay confirmed that 1 and 16
inhibited LPAAT-b in the same range as the reference inhibitor
43. The effects on cell proliferation, cell cycling and signaling
were also consistent with LPAAT-b inhibition. These experi-
ments illustrate the value of target prediction tools such as
PPB2 to guide target identification for phenotypic screening
hits, and significantly expand the rather limited pharmacology
of LPAAT-b inhibitors.
Procedure B. Finely powdered biguanide salt (1 equiv) was added
to a solution of methyl chloroacetate (1 equiv) and sodium meth-
oxide (1.2 equiv, 0.5m) in MeOH at 08C. The reaction was stirred
for 3 h at room temperature before the addition of HCl (88 mL
conc HCl in 250 mL H2O per mmol). After stirring for 12–18 h at
room temperature, water was added and the precipitate filtered
off and dried in vacuo. The procedure was used for triazines 30–40
(see Supporting Information).
Procedure C. A mixture of triazine (1 equiv), hydroxyphenone
(1 equiv) and anhydrous K2CO3 in dimethylformamide (1m) was
heated at 1558C for 3 h. After cooling, cold water was added and
the precipitate filtered, washed with water and dried in vacuo.
Where necessary, purification by silica gel chromatography deliv-
ered the benzofuran product as a white solid. The procedure was
used for triazines 1–18.
6-(3-Methyl-2-benzofuranyl)-N2-(4-ethylphenyl)-1,3,5-triazine-
2,4-diamine (1). Following general procedure C, using triazine 30
(132 mg, 1.0 equiv, 0.5 mmol) and 2-hydroxyacetophenone (56 mL,
1.0 equiv, 0.5 mmol), benzofuran 1 (118 mg, 0.34 mmol, 68%) was
isolated as a white solid after purification by column chromatogra-
phy (gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-UPLC: tR =
Experimental Section
1
Chemistry
2.23 min (method A); mp: 220–2218C; H NMR (300 MHz, MeOD):
d=7.69 (1H, d, J=7.7 Hz, Hfuran), 7.60 (2H, brd, J=8.4 Hz, Haniline),
7.54 (1H, d, J=8.2 Hz, Hfuran), 7.44 (1H, td, J=8.2, 1.2 Hz, Hfuran),
7.31 (1H,td, J=7.7, 0.9 Hz, Hfuran), 7.17 (2H, d, J=8.4 Hz, Haniline),
2.75 (3H, s, CCH3), 2.63 (2H, q, J=7.5 Hz, CH2CH3), 1.24 ppm (3H, t,
J=7.5 Hz, CH2CH3); 13C NMR (75 MHz, [D6]DMSO): d=166.6, 165.1,
164.1 (CN2), 153.3, 146.9, 137.6, 137.3, 129.8, 127.6, 126.6, 122.9,
120.9, 120.8, 120.4, 111.3 (Carom/CHarom), 27.6 (CH2CH3), 15.8
General remarks. All reagents were purchased from commercial
sources and were used without further purification. Flash chroma-
tography purifications were performed with silica Gel 60 (Fluka,
0.040–0.063 nm, 230–400 mesh ASTM). Low-resolution mass spec-
tra were obtained by electron spray ionization (ESI-MS) in the posi-
tive mode on a Thermo Scientific LCQ Fleet. High-resolution mass
spectra were obtained by electron spray ionization (HR-ESI-MS) in
ChemMedChem 2018, 13, 1 – 14
www.chemmedchem.org
7
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
(CH2CH3), 9.8 ppm (CCH3); HRMS calculated for C20H20N5O: m/z
346.1662, m/z found 346.1654.
[D6]DMSO): d=166.7, 165.1, 164.2 (CN2), 153.3, 146.9, 137.4, 136.0,
129.8, 128.2, 126.6, 122.9, 120.9, 120.8, 120.3, 111.3 (Carom/CHarom),
36.7 (CCH2CH2CH3), 24.2 (CH2CH2CH3), 13.6 (CH2CH2CH3), 9.8 ppm
(CCH3); HRMS calculated for C21H22ON5: m/z 360.1819, m/z found
360.1816.
6-(3-Methyl-2-benzofuranyl)-N2-(4-chlorophenyl)-1,3,5-triazine-
2,4-diamine (2). Following general procedure C, using triazine 31
(68 mg, 1.0 equiv, 0.25 mmol) and 2-hydroxyacetophenone (30 mL,
1.0 equiv, 0.25 mmol), triazine 2 (33 mg, 0.12 mmol, 49%) was iso-
lated as a white solid after purification by column chromatography
(gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-UPLC: tR =
4.53 min (method C); mp: 265–2668C; 1H NMR (300 MHz,
[D6]DMSO): d=9.82 (1H, brs, NHAr), 7.91 (2H, d, J=8.7 Hz, Haniline),
7.77 (1H, d, J=7.6 Hz, Hfuran), 7.64 (1H, d, J=8.1 Hz, Hfuran), 7.47
6-(3-Methyl-2-benzofuranyl)-N2-(4-hexylphenyl)-1,3,5-triazine-
2,4-diamine (6). Following general procedure C, using triazine 35
(80 mg, 1.0 equiv, 0.25 mmol) and 2-hydroxyacetophenone (30 mL,
1.0 equiv, 0.25 mmol), triazine 6 (76 mg, 0.19 mmol, 76%) was iso-
lated as a white solid after purification by column chromatography
(gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-UPLC: tR =
1.90 min (method B); mp: 157–1588C; 1H NMR (300 MHz,
[D6]DMSO): d=9.55 (1H, brs, NHAr), 7.76 (1H, ddd, J=7.8, 1.3,
0.7 Hz, Hfuran), 7.72 (2H, d, J=8.1 Hz, Haniline), 7.64 (1H, td, J=8.2,
0.7 Hz, Hfuran), 7.46 (1H, ddd, J=8.2, 7.2, 1.2 Hz, Hfuran), 7.35 (1H,
ddd, J=7.7, 7.2, 0.8 Hz, Hfuran), 7.26 (1H, brs, NH), 7.12 (2H, d, J=
8.5 Hz, Haniline), 2.75 (3H, s, CCH3), 2.52 (2H, t, J=7.0 Hz, CH2Ar),
1.65–1.50 (2H, m, CH2), 1.37–1.24 (6H, m, 3ꢁCH2), 0.88 ppm (3H, t,
J=6.6 Hz, CH2CH3); 13C NMR (75 MHz, [D6]DMSO): d=166.7, 165.1,
164.1 (CN2), 153.3, 146.9, 137.7, 136.2, 129.8, 128.1, 126.6, 122.9,
120.9, 120.8, 120.3, 111.3 (Carom/CHarom), 34.5, 31.1, 31.1, 28.3, 22.1
(CH2), 13.9 (CH2CH3), 9.8 ppm (CCH3); HRMS calculated for
C24H28ON5: m/z 402.2288, m/z found 402.2279.
(1H, td, J=7.9, 1.0 Hz, Hfuran), 7.42–7.20 (5H, m, 2ꢁHaniline, 1ꢁHfuran
,
2ꢁNH2), 2.76 ppm (3H, s, CCH3); 13C NMR (75 MHz, [D6]DMSO): d=
166.6, 165.1, 164.1 (CN2), 153.4, 146.7, 138.8, 129.7, 128.2, 126.8,
125.7, 122.9, 121.5, 121.3, 120.8, 111.4 (Carom/CHarom), 9.9 ppm
(CCH3furan); HRMS calculated for C18H15ON535Cl: m/z 352.0960, m/z
found 352.0957.
6-(3-Methyl-2-benzofuranyl)-N2-(4-methoxyphenyl)-1,3,5-tria-
zine-2,4-diamine (3). Following general procedure C, using triazine
32 (66 mg, 1.0 equiv, 0.25 mmol) and 2-hydroxyacetophenone
(30 mL, 1.0 equiv, 0.25 mmol), triazine 3 (49 mg, 0.12 mmol, 46%)
was isolated as a white solid after purification by column chroma-
tography (gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-
UPLC: tR =2.02 min (method A); mp: 228–2298C; 1H NMR
(300 MHz, [D6]DMSO): d=9.49 (1H, brs, NHAr), 7.76 (1H, d, J=
7.5 Hz, Hfuran) 7.71 (2H, d, J=9.1 Hz, Haniline), 7.64 (1H, d, J=8.2 Hz,
6-(3-Methyl-2-benzofuranyl)-N2-(4-biphenyl)-1,3,5-triazine-2,4-di-
amine (7). Following general procedure C, using triazine 36
(78 mg, 1.0 equiv, 0.25 mmol) and 2-hydroxyacetophenone (30 mL,
1.0 equiv, 0.25 mmol), triazine 7 (42 mg, 0.11 mmol, 43%) was iso-
lated as a white solid after purification by column chromatography
(gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-UPLC: tR =
2.38 min (method A); mp: 253–2548C; 1H NMR (300 MHz,
[D6]DMSO): d=9.79 (1H, brs, NHAr), 7.97 (2H, d, J=8.4 Hz, Haniline),
7.78 (1H, d, J=7.6 Hz, Hfuran), 7.72–7.60 (5H, m, Haniline/Hfuran), 7.52–
7.43 (3H, m, Haniline/Hfuran), 7.41–7.31 (5H, m, Haniline/Hfuran/1ꢁNH),
7.25 (1H, brs, NH), 2.79 ppm (3H, s, CH3); 13C NMR (75 MHz,
[D6]DMSO): d=166.6, 165.1, 164.1 (CN2), 153.4, 146.8, 139.9, 139.3,
133.8, 129.8, 128.9, 126.8, 126.7, 126.6, 126.2, 122.9, 121.2, 120.8,
120.4, 111.4 (Carom/CHarom), 9.9 ppm (CCH3); HRMS calculated for
C24H20ON5: m/z 394.1662, m/z found 394.1653.
H
furan), 7.46 (1H, ddd, J=8.4, 7.2, 1.4 Hz, Hfuran), 7.35 (1H, ddd, J=
7.7, 6.7, 1.0 Hz, Hfuran), 7.22 (1H, brs, NH), 7.10 (1H, brs, NH), 6.90
(2H, d, J=9.1 Hz, Haniline), 3.76 (3H, s, OCH3), 2.75 ppm (3H, s,
CCH3); 13C NMR (75 MHz, [D6]DMSO): d=166.7, 165.0, 164.1 (CN2),
154.8, 153.3, 146.9, 132.7, 129.8, 126.6, 122.9, 122.0, 120.8, 120.8,
113.6, 111.3 (Carom/CHarom), 55.2 (OCH3), 9.8 ppm (CCH3); HRMS calcu-
lated for C19H18O2N5: m/z 348.1452, m/z found 348.1455.
6-(3-Methyl-2-benzofuranyl)-N2-(4-isopropylphenyl)-1,3,5-tria-
zine-2,4-diamine (4). Following general procedure C, using triazine
33 (69 mg, 1.0 equiv, 0.25 mmol) and 2-hydroxyacetophenone
(30 mL, 1.0 equiv, 0.25 mmol), triazine 4 (51 mg, 0.14 mmol, 57%)
was isolated as a white solid after purification by column chroma-
tography (gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-
UPLC: tR =4.80 min (method C); mp: 211–2228C; 1H NMR
(300 MHz, [D6]DMSO): d=9.57 (1H, brs, NHAr), 7.78–7.69 (3H, m,
2ꢁHaniline, 1ꢁHfuran), 7.63 (1H, d, J=8.1 Hz, Hfuran), 7.46 (1H, ddd, J=
8.1, 7.2, 1.0 Hz, Hfuran), 7.34 (1H, t, J=7.6 Hz, Hfuran), 7.31–7.09 (4H,
m, 2ꢁHaniline, 2ꢁNH2), 2.87 (1H, quit., J=6.9 Hz, CH(CH3)2), 2.76
(3H, s, CCH3), 1.21 ppm (6H, d, J=6.9 Hz, CH(CH3)2); 13C NMR
(75 MHz, [D6]DMSO): d=166.6, 165.0, 164.1 (CN2), 153.3, 146.8,
142.3, 137.4, 129.8, 126.6, 126.1, 122.9, 121.0, 120.8, 120.5, 111.3
(Carom/CHarom), 32.8 (CH(CH3)2), 24.0 (CH(CH3)2), 9.8 ppm (CCH3);
HRMS calculated for C21H22ON5: m/z 360.1819, m/z found 360.1816.
6-(3-Methyl-2-benzofuranyl)-N2-(4-fluorophenyl)-1,3,5-triazine-
2,4-diamine (8).[14c] Following general procedure C, using triazine
37 (63 mg, 1.0 equiv, 0.25 mmol) and 2-hydroxyacetophenone
(30 mL, 1.0 equiv, 0.25 mmol), triazine 8 (53 mg, 0.16 mmol, 63%)
was isolated as a white solid after purification by column chroma-
tography (gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-
UPLC: tR =4.16 min (method C); mp: 209–2108C; 1H NMR
(300 MHz, [D6]DMSO): d=9.72 (1H, brs, NHAr), 7.87 (2H, brs, Haniline),
7.77 (1H, d, J=7.7 Hz, Hfuran), 7.65 (1H, d, J=8.2 Hz, Hfuran), 7.48
(1H, t, J=7.5 Hz, Hfuran), 7.35 (2H, t, J=7.3 Hz, 2ꢁNH2, Hfuran), 7.17
(3H, t, J=8.3 Hz, 2ꢁHaniline, 1ꢁNH2), 2.77 ppm (3H, s, CCH3);
13C NMR (75 MHz, [D6]DMSO): d=166.7, 165.2, 164.1 (CN2), 157.6 (d,
J=239 Hz), 153.4, 146.8, 136.1, 129.8, 126.7, 122.9, 121.9 (brs),
121.1, 120.8, 114.8 (d J=23 Hz), 111.4 (CHfuran), 9.8 ppm (CCH3furan);
HRMS calculated for C18H15ON5F: m/z 336.1255, m/z found
336.1251.
6-(3-Methyl-2-benzofuranyl)-N2-(4-propylphenyl)-1,3,5-triazine-
2,4-diamine (5). Following general procedure C, using triazine 34
(69 mg, 1.0 equiv, 0.25 mmol) and 2-hydroxyacetophenone (30 mL,
1.0 equiv, 0.25 mmol), triazine 5 (53 mg, 0.14 mmol, 57%) was iso-
lated as a white solid after purification by column chromatography
(gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-UPLC: tR =
4.87 min (method C); mp: 175–1858C; 1H NMR (300 MHz,
[D6]DMSO): d=9.58 (1H, brs, NHAr), 7.85–7.69 (3H, m, 2ꢁHaniline, 1ꢁ
6-(3-Methyl-2-benzofuranyl)-N2-(4-bromophenyl)-1,3,5-triazine-
2,4-diamine (9). Following general procedure C, using triazine 38
(79 mg, 1.0 equiv, 0.25 mmol) and 2-hydroxyacetophenone (30 ml,
1.0 equiv, 0.25 mmol), triazine 9 (72 mg, 0.18 mmol, 73%) was iso-
lated as a white solid after purification by column chromatography
(gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-UPLC: tR =
2.31 min (method A); mp: 243–2448C; 1H NMR (300 MHz,
H
furan), 7.65 (1H, d, J=8.0 Hz, Hfuran), 7.47 (1H, t, J=7.8 Hz, Hfuran),
7.35 (1H, t, J=7.2 Hz, Hfuran), 7.32–6.98 (4H, m, 2ꢁHaniline, 2ꢁNH2),
2.77 (3H, s, CCH3), 2.57–2.51 (2H, m, CCH2), 1.68–1.51 (2H, m,
CH2CH3), 0.92 ppm (3H, t, J=7.0 Hz, CH2CH3); 13C NMR (75 MHz,
&
ChemMedChem 2018, 13, 1 – 14
www.chemmedchem.org
8
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
[D6]DMSO): d=9.82 (1H, brs, NHAr), 7.86 (2H, d, J=8.8 Hz, Haniline),
7.76 (1H, d, J=7.6 Hz, Hfuran), 7.64 (1H, d, J=8.1 Hz, Hfuran), 7.52–
7.43 (3H, m, 2ꢁHaniline, 1ꢁHfuran), 7.41–7.22 (3H, m, 1ꢁHfuran, 2ꢁ
NH2), 2.76 ppm (3H, s, CCH3); 13C NMR (75 MHz, [D6]DMSO): d=
166.6, 165.1, 164.1 (CN2), 153.4, 146.7, 139.3, 131.1, 129.7, 126.7,
122.9, 121.9, 121.2, 120.8, 113.7, 111.4 (Carom/CHarom), 9.9 ppm
(CCH3); HRMS calculated for C18H15ON579Br: m/z 396.0454, m/z
found 396.0456.
(66 mg, 1.0 equiv, 0.25 mmol) and 2-hydroxybutyrophenone
(45 mg, 1.0 equiv, 0.25 mmol), triazine 13 (65 mg, 0.17 mmol, 67%)
was isolated as a white solid after purification by column chroma-
tography (gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-
UPLC: tR =5.32 min (method C); mp: 171–1728C; 1H NMR
(300 MHz, [D6]DMSO): d=9.51 (1H, brs, NHAr), 7.76 (1H, d, J
ꢀ7.8 Hz, Hfuran), 7.68 (1H, brs, Haniline), 7.63 (1H, d, J ꢀ8.4 Hz, Hfuran),
7.45 (1H, d, J ꢀ7.4 Hz, Hfuran), 7.33 (1H, d, J ꢀ7.4 Hz, Hfuran), 7.15
(4H, brd, J ꢀ7.4 Hz, Haniline, NH2), 3.34 (2H, CCH2furan, overlapping
with solvent peak), 2.58 (2H, q, J=7.5 Hz, CH2CH3aniline), 1.64 (2H,
brt, J=7.4 Hz, CCH2CH2furan), 1.46–1.27 (2H, m, CH2CH3furan), 1.19
(3H, t, J=7.5 Hz, CH2CH3aniline) 0.88 ppm (3H, t, J=7.0 Hz, CH3furan);
13C NMR (75 MHz, [D6]DMSO): d=166.7, 165.1, 164.3 (CN2), 153.4,
146.8, 137.8, 137.3, 129.2, 127.6, 126.5, 125.6, 122.9, 120.9, 120.9,
111.4 (Carom/CHarom), 31.8 (CCH2CH2), 27.6 (CCH2CH3), 23.2 (CCH2CH2),
21.9 (CH2CH2CH3), 15.8 (CCH2CH3), 13.9 ppm (CH2CH2CH3); HRMS
calculated for C23H26N5O: m/z 388.2132, m/z found 388.2131.
6-(3-Methyl-2-benzofuranyl)-N2-(4-trifluoromethylphenyl)-1,3,5-
triazine-2,4-diamine (10). Following general procedure C, using
triazine 39 (76 mg, 1.0 equiv, 0.25 mmol) and 2-hydroxyacetophe-
none (30 mL, 1.0 equiv, 0.25 mmol), triazine 10 (40 mg, 0.10 mmol,
42%) was isolated as a white solid after purification by column
chromatography (gradient, hexane/ethyl acetate, 50:50 to 15:85).
RP-UPLC: tR =1.33 min (method B); mp: 228–2298C; 1H NMR
(300 MHz, [D6]DMSO): d=10.07 (1H, brs, NHAr), 8.10 (2H, d, J=
8.7 Hz, Haniline), 7.78 (1H, dd, J=7.6, 1.2 Hz, Hfuran), 7.66 (3H, brd, J=
9.0 Hz, 2ꢁHaniline 1ꢁHfuran), 7.48 (2H, ddd, J=8.4, 7.1, 1.2 Hz, 1ꢁ
6-(3-Ethyl-2-benzofuranyl)-N2-(4-methoxyphenyl)-1,3,5-triazine-
2,4-diamine (14). Following general procedure C, using triazine 32
(66 mg, 1.0 equiv, 0.25 mmol) and 2-hydroxypropriophenone
(34 mL, 1.0 equiv, 0.25 mmol), triazine 14 (47 mg, 0.13 mmol, 52%)
was isolated as a white solid after purification by column chroma-
tography (gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-
UPLC: tR =2.06 min (method A); mp: 183–1848C; 1H NMR
(300 MHz, [D6]DMSO): d=9.51 (1H, brs, NHAr), 7.79 (1H, d, J=
7.9 Hz, Hfuran), 7.75–7.58 (3H, m, 2ꢁHaniline, 1ꢁNH3+), 7.46 (1H, td,
J=7.2, 1.2 Hz, Hfuran), 7.34 (1H, t, J=7.5 Hz, Hfuran), 7.22 (2H, brs, 2ꢁ
NH3+), 6.91 (2H, d, J=8.9 Hz, Haniline), 3.76 (3H, s, OCH3), 3.33 (2H,
brs, CH2CH3), 1.25 ppm (3H, t, J=7.0 Hz, CH2CH3); 13C NMR
(75 MHz, [D6]DMSO): d=166.4, 164.6, 164.1 (CN2), 155.0. 153.5.
146.1. 132.5. 128.8. 127.3, 126.6, 122.9, 122.2, 120.9, 113.6, 111.5
(Carom/CHarom), 55.2 (OCH3), 17.1 (CH2CH3), 14.5 ppm (CH2CH3); HRMS
calculated for C21H22ON5: m/z 362.1585, m/z found 362.1597.
H
furan, 1ꢁNH), 7.36 (1H, ddd, J=8.0, 7.4, 0.9 Hz, Hfuran), 7.33 (1H,
brs, NH), 2.78 ppm (3H, s, CH3); 13C NMR (75 MHz, [D6]DMSO): d=
166.6, 165.2, 164.2 (CN2), 153.4. 146.6, 143.6, 129.7, 126.4, 125.5 (q,
J=4 Hz), 124.6 (q, J=271 Hz), 123.0, 121.6 (q, J=32 Hz), 121.5,
120.9, 119.6 (Carom/CHarom), 9.9 ppm (CCH3); HRMS calculated for
C19H15ON5F3: m/z 386.1223, m/z found 386.1214.
6-(3-Methyl-2-benzofuranyl)-N2-(4-acetaminophenyl)-1,3,5-tria-
zine-2,4-diamine (11). Following general procedure C, using tria-
zine 40 (73 mg, 1.0 equiv, 0.25 mmol) and 2-hydroxyacetophenone
(30 mL, 1.0 equiv, 0.25 mmol), triazine 11 (79 mg, 0.20 mmol, 81%)
was isolated as a white solid after purification by column chroma-
tography (gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-
UPLC: tR =1.73 min (method A); mp: 274–2758C; 1H NMR
(300 MHz, [D6]DMSO): d=9.89 (1H, brs, NHCO), 9.60 (1H, brs, NHAr),
7.82–7.69 (3H, m, 2ꢁHaniline, 1ꢁHfuran), 7.64 (1H, d, J=8.2 Hz, Hfuran),
7.54 (2H, d, J=8.7 Hz, Haniline), 7.47 (1H, t, J=7.7 Hz, Hfuran), 7.35
(1H, t, J=7.4 Hz, Hfuran), 7.28 (1H, brs, 1ꢁNH2), 7.19 (1H, brs, 1ꢁ
NH2), 2.76 (3H, s, COCH3), 2.06 ppm (3H, s, CCH3); 13C NMR
(75 MHz, [D6]DMSO): d=167.9 (CO), 166.7, 165.1, 164.1 (CN2), 153.3,
146.9, 135.0, 134.1, 129.8, 126.6, 122.9, 120.9, 120.8, 120.6, 119.2,
111.3 (CHfuran), 23.9 (COCH3), 9.9 ppm (CCH3furan); HRMS calculated
for C20H19O2N6: m/z 375.1564, m/z found 375.1560.
6-(3-Ethyl-2-benzofuranyl)-N2-(4-fluorophenyl)-1,3,5-triazine-2,4-
diamine (15). Following general procedure C, using triazine 37
(63 mg, 1.0 equiv, 0.25 mmol) and 2-hydroxypropriophenone
(34 mL, 1.0 equiv, 0.25 mmol), triazine 15 (59 mg, 0.17 mmol, 68%)
was isolated as a white solid after purification by column chroma-
tography (gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-
UPLC: tR =4.38 min (method C); mp: 285–1868C; 1H NMR
(300 MHz, [D6]DMSO): d=9.67 (1H, brs, NHAr), 7.83 (1H, brs, 1ꢁ
NH3+), 7.79 (2H, d, J=7.6 Hz, Haniline), 7.64 (1H, d, J=8.2 Hz, Hfuran),
7.46 (1H, ddd, J=8.2, 7.4, 1.1 Hz, Hfuran), 7.33 (2H, td, J=7.6, 0.6 Hz,
1ꢁHfuran, 1ꢁNH3+) 7.16 (3H, t, J=8.8 Hz, 1ꢁNH3+, 2ꢁHaniline), 3.33
(2H, q, J=7.4 Hz, CH2CH3), 1.27 ppm (3H, t, J=7.4 Hz, CH2CH3);
13C NMR (75 MHz, [D6]DMSO): d=166.7, 165.0, 164.2 (CN2), 157.8 (d,
J=240 Hz), 153.5, 146.2, 136.0, 128.8, 127.2, 126.6, 122.9, 122.1
(brs), 120.8, 114.9 (d, J=22 Hz), 111.5 (Carom/CHarom), 17.1 (CH2CH3),
14.5 ppm (CH2CH3); HRMS calculated for C19H17ON5F: m/z 350.1412,
m/z found 350.1406.
6-(3-Ethyl-2-benzofuranyl)-N2-(4-ethylphenyl)-1,3,5-triazine-2,4-
diamine (12). Following general procedure C, using triazine 30
(66 mg, 1.0 equiv, 0.25 mmol) and 2-hydroxypropriophenone
(34 mL, 1.0 equiv, 0.25 mmol), triazine 12 (48 mg, 0.13 mmol, 53%)
was isolated as a white solid after purification by column chroma-
tography (gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-
UPLC: tR =1.24 min (method B); mp: 187–1888C; 1H NMR
(300 MHz, [D6]DMSO): d=9.52 (1H, brs, NHAr), 7.80 (1H, d, J=
7.9 Hz, Hfuran), 7.70 (2H, d, J=8.5 Hz, Haniline), 7.64 (1H, d, J=7.9 Hz,
H
furan), 7.46 (1H, ddd, J=7.9, 7.5, 1.1 Hz, Hfuran), 7.34 (1H, ddd, J=
7.9, 7.5, 0.6 Hz, Hfuran), 7.26 (1H, brs, NH), 7.16 (2H, d, J=8.5 Hz,
aniline), 7.12 (1H, brs, NH), 3.35 (2H, q, J=7.6 Hz, CH2CH3furan), 2.59
6-(3-Ethyl-2-benzofuranyl)-N2-(4-chlorophenyl)-1,3,5-triazine-2,4-
diamine (16). Following general procedure C, using triazine 31
(68 mg, 1.0 equiv, 0.25 mmol) and 2-hydroxypropriophenone
(34 mL, 1.0 equiv, 0.25 mmol), triazine 16 (74 mg, 0.20 mmol, 81%)
was isolated as a white solid after purification by column chroma-
tography (gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-
UPLC: tR =2.27 min (method A); mp: 237–2388C; 1H NMR
(300 MHz, [D6]DMSO): d=9.80 (1H, brs, NHAr), 7.89 (2H, d, J=
7.3 Hz, Haniline), 7.80 (1H, d, J=7.7 Hz, Hfuran), 7.66 (1H, d, J=8.1 Hz,
H
(2H, q, J=7.5 Hz, CH2CH3aniline), 1.27 (3H, t, J=7.5 Hz, CH2CH3furan),
1.20 ppm (3H, t, J=7.5 Hz, CH2CH3aniline); 13C NMR (75 MHz,
[D6]DMSO): d=166.7, 165.0, 164.3 (CN2), 153.5, 146.3, 137.7, 137.3,
128.8, 127.6, 127.1, 126.6, 122.9, 120.8, 120.6, 111.5 (Carom/CHarom),
27.6 (CH2CH3aniline), 17.1 (CH2CH3furan), 15.8 (CH2CH3aniline), 14.5 ppm
(CH2CH3furan); HRMS calculated for C21H22ON5: m/z 360.1819, m/z
found 360.1816.
H
furan), 7.47 (1H, t, J=7.3 Hz, Hfuran) 7.46–7.16 (5H, m, 2ꢁHaniline, 2ꢁ
6-(3-Butyl-2-benzofuranyl)-N2-(4-ethylphenyl)-1,3,5-triazine-2,4-
diamine (13). Following general procedure C, using triazine 30
NH2, 1ꢁHfuran), 3.36 (2H, q, J=7.1 Hz, CH2CH3), 1.29 ppm (3H, t, J=
7.1 Hz, CH2CH3); 13C NMR (75 MHz, [D6]DMSO): d=166.7, 165.1,
ChemMedChem 2018, 13, 1 – 14
www.chemmedchem.org
9
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
164.2 (CN2), 153.3, 146.2, 138.8, 128.8, 128.2, 127.3, 126.6, 125.8,
122.9, 121.6, 120.9, 111.5 (Carom/CHarom), 17.1 (CH2CH3), 14.5 ppm
(CH2CH3); HRMS calculated for C19H17ON535Cl: m/z 366.1116, m/z
found 366.1109.
Cloning of LPAAT into the pMJB08 expression vector. Optimized
for Eukaryotes DNA from LPAAT-beta (UniProtKB entry: O15120,
GeneWiki: AGPAT2) was produced by GeneArtꢄ and was cloned
from a carrier construct of GeneArtꢄ into the pMJB08[35] expression
vector without tags. The vectors were digested with the restriction
enzymes BamHI and HindIII, and ligated into the pMJB08 vector.
The DNA construct was verified by sequencing.[36] Expression of re-
combinant LPAAT-b using this construct results in a protein with-
out tags. For more details on pMJB08 see Bergeron et al.[35] The cal-
culated molecular masses of recombinant LPAAT-b based on the
amino acid sequence are ꢀ30.9 kDa (full-length).
Protein expression in Xenopus laevis oocytes. Expression of
LPAAT-b was performed as previously described for other mamma-
lian and human transporters by Bergeron et al.[35] Ovarian tissue
was surgically removed from the frogs and treated with 3 mgmLÀ1
of collagenase A (Roche) for 2 h in Modified Barth’s Medium (MBM:
88 mm NaCl, 1 mm KCl, 2.4 mm NaHCO3, 0.82 mm MgSO4, 0.66 mm
NaNO3, 0.75 mm CaCl2, 10 mm HEPES-NaOH (pH 7.4)) but without
CaCl2 and supplemented with P/S antibiotics (GIBCOꢅ Penicillin-
Steptomycin liquid, Invitrogen). This buffer is isoosmolar relative to
the X. laevis plasma (ꢀ200 mOsm). Subsequently, defolliculated
stage V-VI oocytes were isolated, transferred into plates containing
MBM medium supplemented with P/S antibiotics and maintained
at 188C prior to injection. Oocytes were injected with 40 ng of
cDNA-derived cRNA (in vitro transcription with the mMESSAGE
mMACHINEꢄ T7 kit, Ambion) and maintained for 3 days at 188C in
MBM supplemented with P/S antibiotics.
6-(3-Ethyl-2-benzofuranyl)-N2-(4-bromophenyl)-1,3,5-triazine-2,4-
diamine (17). Following general procedure C, using triazine 38
(79 mg, 1.0 equiv, 0.25 mmol) and 2-hydroxyisoproprione (34 mL,
1.0 equiv, 0.25 mmol), triazine 17 (94 mg, 0.23 mmol, 92%) was iso-
lated as a white solid after purification by column chromatography
(gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-UPLC: tR =
2.36 min (method A); mp: 237–2388C; 1H NMR (300 MHz,
[D6]DMSO): d=9.80 (1H, brs, NHAr), 7.84 (2H, d, J=8.4 Hz, Haniline),
7.79 (1H, d, J=7.8 Hz, Hfuran), 7.65 (1H, d, J=8.2 Hz, Hfuran), 7.53–
7.17 (6H, m, 2ꢁHaniline, 2ꢁNH2, 2ꢁHfuran), 3.35 (2H, q, J=7.6 Hz,
CH2CH3), 1.28 ppm (3H, t, J=7.6 Hz, CH2CH3); 13C NMR (75 MHz,
[D6]DMSO): d=166.7, 165.1, 164.2 (CN2), 153.5, 146.2, 139.3, 131.1,
128.8, 127.4, 126.6, 122.9, 122.0, 120.9, 113.8, 111.5 (Carom/CHarom),
17.1 (CH2CH3), 14.5 ppm (CH2CH3); HRMS calculated for
C19H17ON579Br: m/z 410.0611, m/z found 410.0603.
6-(3-Ethyl-2-benzofuranyl)-N2-(4-trifluoromethylphenyl)-1,3,5-tri-
azine-2,4-diamine (18). Following general procedure C, using tria-
zine 39 (38 mg, 1.0 equiv, 0.125 mmol) and 2-hydroxyisoproprione
(17 mL, 1.0 equiv, 0.125 mmol), triazine 18 (29 mg, 0.07 mmol, 58%)
was isolated as a white solid after purification by column chroma-
tography (gradient, hexane/ethyl acetate, 50:50 to 15:85). RP-
UPLC: tR =5.13 min (method C); mp: 180–1818C; 1H NMR
(300 MHz, CDCl3): d=7.79 (2H, d, J=8.5 Hz, Haniline), 7.70 (1H, d, J=
7.9 Hz, Hfuran), 7.67–7.59 (3H, m, 2ꢁHaniline, 1ꢁHfuran), 7.49–7.41 (2H,
m, 1ꢁHfuran, 1ꢁNHAr), 7.32 (1H, td, J=7.4. 1.9 Hz, Hfuran), 5.51 (2H,
brs, 2ꢁNH2), 3.36 (2H, q, J=7.4 Hz, CH2CH3), 1.38 ppm (3H, t, J=
7.4 Hz, CH2CH3); 13C NMR (75 MHz, CDCl3): d=166.9, 166.4, 164.4
(CN2), 154.6, 145.6, 141.5, 130.0, 129.3, 127.0, 125.2 (q, J=223 Hz),
124.0, 122.9 (q, J=39 Hz), 120.8, 119.9, 112.2 (Carom/CHarom), 29.7
(CH2CH3), 17.9 ppm (CH2CH3); HRMS calculated for C20H17ON5F3: m/z
400.1380, m/z found 400.1373.
Isolation of egg yolk-depleted LPAAT-b from oocytes. 1000 oo-
cytes were homogenized in 10 mL of ice-cold 25 mm HEPES·HCl
(pH 7.5), 1 mm EDTA, 100 mm NaCl supplemented with the Com-
pleteꢄ protease inhibitor cocktail (Roche). To remove cell debris,
nuclei and pigments, the homogenate was subjected to a low spin
centrifugation (1500ꢁg; 2 min; 48C) and the supernatant was
saved for enzyme assay.
LPAAT-b enzymatic activity. LPAAT-b activity was measured by for-
mation of PA (200 mm of 18:1-CoA, 200 mm sn-1-18:1 lysoPA,
500 mm DTNB, 25 mm HEPES·HCl, pH 7.5, 1 mm EDTA, 100 mm NaCl
and various concentrations of inhibitors) during a 3-min incubation
period at room temperature in flat-bottom, MicroWell 96-well poly-
styrene plates (TPP, Trasadingen, Switzerland). Product formation
was monitored by the change in absorbance at 405 nm over 3 min
on a Spectra Max 250 Microplate Reader from Molecular Devices.
IC50 values were calculated using a nonlinear regression of the soft-
ware GraphPad Prism.
Biology
EC-vSMC co-culture angiogenesis assay. Co-cultures were seeded
in 96-well plates with 12000 human umbilical cord endothelial
cells (HUVEC) and 50000 pulmonary artery-derived vascular
smooth muscle cells (PA-vSMC) per well. Cells were allowed to
attach for 3 h (network formation) or 72 h (disintegration of
formed network) before compound addition. 100 mL of EGM-2 was
pipetted to each well of another 96-well plate. 1 mL of compound
solution was added and mixed with the medium. These were care-
fully transferred to the pre-seeded co-cultures. The co-cultured
cells were incubated with the compound solution (200 mL per well)
for 72 h at 5% CO2 and 378C. Imaging was performed with a BD
Pathway 855 Bioimager with a 10x objective. Images were acquired
as 2ꢁ2 montages with a GFP LP515 filter with 0.11 s exposure
time. Human Iliac Vein Endothelial Cells (HIVEC) was trans-ducted
with a GFP-construct and showed stable expression of GFP. Image
analysis was performed with Attovisoin v1.6.2 using the tube for-
mation algorithm.[33]
Kinome scan. Kinase profiling was performed for binding of the
ATP binding pockets on 442 kinases by DiscoveRx in San Diego.
The tested concentration was 10 mm in DMSO.[37]
Evaluation of IC50. MEF, SK-OV-3 (AKR-225; LuBioscience), CHO and
HEK-293 cells were cultivated in Dulbecco’s modified Eagle’s
medium (DMEM) (D5796-6X500ML; Sigma) complemented with
10% fetal calf serum (FCS) (A3160801; Life Technologies), 2 mm l-
glutamine (25030123; LuBioscience), MEM non-essential amino
acids (NEAA) 1:100 (5–13K00-H; Bioconcept), Penicillin 100 UmLÀ1
and Streptomycin 0.1 mgmLÀ1 (P0781-100ML; Sigma–Aldrich). To
assure HEK-293 cell adherence to the wells, plates were coated
with poly-d-lysine hydrobromide from (P6407-5MG; Sigma–Al-
drich). We incubated plates with 50 mL per well of an 0.1 mg/mL
solution of polylysine diluted in sterile water. Then plates were
washed two times with 100 mL of sterile water and allowed to dry
for 10 min before seeding the cells. All other cell types were
seeded without coating. Cells were seeded in 96-Well plates at a
Cell culture (Table 1). HeLa cells were obtained from the European
Collection of Animal Cell Cultures or the American Type Culture
Collection and were cultured according to the supplier’s instruc-
tions. Cells were seeded in 96-well plates and incubated with in-
hibitors for 96 h.[34] Cell viability was assessed using Dojindo Cell
Counting Kit-8 to measure WST-8 (2-(2-methoxy-4-nitrophenyl)-3-
(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium) conversion.
&
ChemMedChem 2018, 13, 1 – 14
www.chemmedchem.org
10
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
density of 10% in 100 mL medium, treated with decreasing doses
of 1, 12, 14, 16 and 43 and fixed with neutrally buffered 10% for-
malin solution (HT501128; Sigma–Aldrich) at the time the DMSO
(D4540-1L; Sigma–Aldrich) controls reached a confluency of 100%.
Fixation was carried out as follows: Without removing the 100 mL
of cell culture medium, 100 mL of the formalin solution was added
to each well. Fixation was allowed to continue for 10 min at room
temperature (RT). After fixation, the plates were washed 2 times
with 100 mL PBS (137 mm NaCl (S9888; Sigma–Aldrich); 2.7 mm KCl
(P3911; Sigma–Aldrich); 18 mm KH2PO4 (P9791; Sigma–Aldrich);
100 mm Na2HPO4 (S5136; Sigma–Aldrich)). After washing, cells
were stained with crystal violet. To that end, 100 mL of a crystal
violet solution containing 0.2% crystal violet (C3886; Sigma–Al-
drich) and 2% EtOH (G0152; Dr. Grogg Chemie AG) in H2O. Fixed
cells were stained for 10 min at RT. After staining, the plates were
washed by plunging in a water bath, until all the surplus ink was
washed out. Then, plates were dried overnight. The dried crystal
violet in the plates was solubilized with 100 mL 1% SDS (05030;
Sigma–Aldrich) solution by shaking gently for 30 min at RT and the
optical density at 500 nm (OD550) was measured. Values are dis-
played as the ratio of the OD550 of the treated cells to the OD550 of
the respective DMSO control.
926–32210). Blots were scanned using a Li-Cor ODYSSEY Sa fluores-
cent western blot scanner and quantified with the ODYSSEY Image
Studio software. Ratios in the graph are depicted as the ratio of
the brightness of the phosphorylated band (pERK; pAKT) band to
the total (tERK; pan-AKT) band, normalized to control conditions.
Cell-cycle analysis. To investigate the effect of the compounds on
the cell cycle MEF and SK-OV-3 cells were seeded in 10 cm dishes
and left to proliferate for 2, respectively 3 days until they reached a
confluency of 80% before the experiment was started. Then Cells
were treated with either DMSO, 14 or 43 for 24 h. MEF cells were
treated with 260 nm 43 and 200 nm 14, SK-OV-3 cells were treated
with 260 nm 43 and 70 nm 14. These doses of drugs were selected
to be fully growth inhibitory. After treatment, the supernatant was
removed and centrifuged for 5 min at 300ꢁg to pellet detached
cells, carefully discard supernatant. Cells on plates were detached
with 3 mL trypsin-EDTA 5% (25300054; Thermo Fisher) for 5 min at
378C. The detached cells in trypsin were taken up in 7 mL com-
plete DMEM medium and added to the pellet of the detached
cells. Cells were centrifuged for 5 min at 300ꢁg at room tempera-
ture and resuspended in 10 mL PBS for counting. After counting,
cells were pelleted once more to the above conditions and resus-
pended in PBS for washing. After washing, cells were pelleted for
5 min at 300ꢁg and 48C, then resuspended in ice cold PBS at a
concentration of 3.33ꢁ106 cellsmLÀ1 (or 0.3 mL of PBS per 106
cells). For fixation, À208C cold EtOH is added to the cells drop
wise while vortexing, to reach a final concentration of 1ꢁ106
cellsmLÀ1 (or 0.7 mL EtOH per 106 cells). Cells were then stored at
48C overnight, after they were kept at À208C before analysis. For
analysis, 1 mL of the fixed cells are centrifuged for 10 min at
1000ꢁg and 48C. After centrifugation, the supernatant is decanted
and the pellet washed with 1 mL of ice cold PBS. The cells were
pelleted by centrifugation for 10 min at 1000ꢁg and 48C. Then
pellets were resuspended in 250 mL ice cold PBS and RNase A
(R4642; Sigma–Aldrich) was added at a final concentration of 0.2-0-
5 mgmLÀ1 and the solution was incubated for 1 h at 378C for RNA
digestion. For staining of DNA propidium iodide (P4864; Sigma–Al-
drich) was added at a final concentration of 40 mgmLÀ1 incubated
for 10 min at room temperature in the dark. Samples were scanned
with a Becton Dickinson FACScan FACS machine. Numbers are de-
picted as percentage of total counts of a certain cell cycle state.
Western blots. To evaluate the effect of 43 and 14 on the activity
of the PI3’K-AKT and the RAF-MEK-ERK pathways, levels of AKT
phosphorylation (pAKT) and ERK phosphorylation (pERK) were
measured. To that end, SK-OV-3 cells were seeded at subconfluent
densities in complete DMEM (see above) in 10 cm cell culture petri
dishes. When cells had grown to 70–80% confluency 2–3 days
after seeding, they were starved over night with DMEM containing
0.5% FBS. After the overnight starvation, cells were treated with
the corresponding drugs for 1 h. Drug concentrations were accord-
ingly: PD-325901: 100 nm (Abmole Bioscience); GDC-0941, 1 mm
(Abmole Bioscience); 43: 260 nm; 14: 70 nm. Concentrations of 14
and 43 were chosen to be growth inhibitory. For protein isolation,
cells were washed twice with ice cold PBS, recovered by scraping
and pelleted by centrifugation at 2300ꢁg 5 min at 48C. Then, pel-
lets were lysed in RIPA buffer (20 mm TRIS base (T1503; Sigma–Al-
drich) pH 8; 150 mm NaCl; 1% Triton-X100 (T8787; Sigma–Aldrich);
0.1% SDS; 0.5% deoxycholate Na+ (D6750; Sigma–Aldrich)) com-
plemented with Haltꢅ Inhibitor Cocktail 1:100 (11834111; Pierce,
ThermoFischer Scientific). For lysing, pellets were resuspended by
pipetting up and down, followed by short vortexing and incuba-
tion at 48C for 30 min. Lysates were cleared by 30 min of centrifu-
gation at full speed and 48C. Lysates were mixed with 5ꢁ sample
buffer (0.294m sucrose (84097; Sigma–Aldrich), 2% SDS, 1 mm
EDTA (E9884; Sigma–Aldrich), 60 mm TRIS pH 8.8, 0.05% bromo-
phenol blue (114391; Sigma–Aldrich), and 26 mm dithiothreitol
(43819; Sigma–Aldrich)). For denaturation, samples were heated at
958C for 5 min. For western blotting, protein extracts were either
loaded separately (for quantifications) or pooled (for Figure pic-
tures) for each treatment condition. Proteins were run on TGX pre-
cast 4–20% gels (456–1096; BioRad, Switzerland), transferred onto
Trans-Blot transfer pack nitrocellulose (170–4158; BioRad, Switzer-
land). Western blots were probed with the following antibodies
and concentrations: Primary antibodies: ERK1/2 (1:5000, Cell Sig-
naling, 9107); pERK1/2 (1:2000, Cell Signaling, 4370); pan-AKT
(1:2000, Cell Signaling, 2920); pAKT-Ser473 (1:2000, Cell Signaling,
4060); actin (1:4000), Sigma, A5316); LPAAT-b (1:500, Biorbyt,
orb5859). Secondary antibodies, Li-Cor Biosciences (Bad Homburg
Germany): IRDye 680RD Goat anti-Mouse IgG (H + L) (1:10000,
926–68071), IRDye 800CW Goat anti-Rabbit IgG (H+L) (1:10000,
926–32210), IRDye 680RD Goat anti Rabbit IgG (H+L) (1:10000
926–68071); IRDye 800CW Goat anti mouse IgG (H+L) (1:10000
Quantitative PCR. After reaching a confluency of around 80–90%,
cells were washed two times with cold PBS and detached from cul-
ture flasks by scraping. For cell pelleting, they were centrifuged at
48C for 5 min at 1000ꢁg and pellets were snap frozen and stored
at À808C for later RNA isolation. Total RNA extraction was per-
formed using standard a phenol-chloroform protocol. Phenol was
substituted by QIAzol Lysis Reagent from Qiagen and the chloro-
form by chloroform/isoamyl alcohol mixture (25668-100ML;
Sigma–Aldrich). RNA was precipitated with isopropanol
(1096341000; Merck) and the obtained pellet was washed with
75% ethanol (diluted absolute alcohol, 02860-4X2.5 L; Sigma–Al-
drich). The pure RNA pellet was resuspended in 70 mL of treated
water with DEPC (D5758-50ML; Sigma–Aldrich). Reverse transcrip-
tion was performed in two steps using 500 ng of total RNA. First a
14 mL mix containing, 0.5 mL of Oligo(dT)12–18 Primer (0.5 mg/mL,
18418-012; Invitrogen), 1 mL of dNTPS mix (10 mm solution from
set of dATP, dCTP, dGTP, dTTP 25 mmol each, U1420; Promega),
500 ng of total RNA and DEPC-treated water was heated for 5 min
at 658C. Second step, we added 6 mL mix containing DEPC-treated
water, 40 U of reverse transcriptase (SuperScriptꢅ II Reverse Tran-
scriptase, 10000 U, 18064-014; Invitrogen), 4 mL of FSB (5ꢁ first-
strand buffer, supplied with the enzyme, Invitrogen) and 1 mL of
ChemMedChem 2018, 13, 1 – 14
www.chemmedchem.org
11
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
[8] a) J. P. O’Bryan, R. A. Frye, P. C. Cogswell, A. Neubauer, B. Kitch, C.
Prokop, R. Espinosa, M. M. Le Beau, H. S. Earp, E. T. Liu, Mol. Cell. Biol.
1991, 11, 5016–5031; b) R. M. Linger, A. K. Keating, H. S. Earp, D. K.
Graham, Expert Opin. Ther. Targets 2010, 14, 1073–1090; c) J. P. Eder,
G. I. Shapiro, L. J. Appleman, A. X. Zhu, D. Miles, H. Keer, B. Cancilla, F.
Chu, S. Hitchcock-Bryan, L. Sherman, S. McCallum, E. I. Heath, S. A.
Boerner, P. M. LoRusso, Clin. Cancer Res. 2010, 16, 3507–3516; d) S. J.
Holland, A. Pan, C. Franci, Y. Hu, B. Chang, W. Li, M. Duan, A. Torneros, J.
Yu, T. J. Heckrodt, J. Zhang, P. Ding, A. Apatira, J. Chua, R. Brandt, P.
Pine, D. Goff, R. Singh, D. G. Payan, Y. Hitoshi, Cancer Res. 2010, 70,
1544–1554; e) A. K. Ghosh, C. Secreto, J. Boysen, T. Sassoon, T. D. Shana-
felt, D. Mukhopadhyay, N. E. Kay, Blood 2011, 117, 1928–1937; f) C. M.
Gay, K. Balaji, L. A. Byers, Br. J. Cancer 2017, 116, 415–423.
[9] a) G. A. Truskey, Int. J. High Throughput Screen. 2010, 2010, 171–181;
b) L. Evensen, W. Link, J. B. Lorens, Curr. Pharm. Des. 2010, 16, 3958–
3963.
DTT (100 mm DTT, supplied with the enzyme, Invitrogen). We incu-
bated tubes for 1 h at 428C for cDNA synthesis. The reaction was
inactivated by incubation at 708C for 15 min. Real-time qPCR was
performed using the GoTaqꢄ qPCR Master Mix (A6002; Promega).
A single tube contained 10 mL of GoTaqꢄ qPCR Master Mix, 5 mL of
DEPC-treated water, 1 mL of primers mix (10 mm of forward and re-
verse primers) and 4 mL of cDNA product (diluted 5 times with
DEPC-treated water). For control, GAPDH was used. Following pri-
mers were used: forward mouse Agpat2: CTCAAAGTGTGGATC-
TACCCAG, reverse mouse Agpat2: GCACTTGTACCTTGATTGTTCC
(size product: 190 bp), forward mouse Gapdh: TGCCCCCATGTTTGT-
GATG, reverse mouse Gapdh: TGTGGTCATGAGCCCTTCC (size prod-
uct: 151 bp), forward human AGPAT2: GTGGGCCTCATCATGTACCTC,
reverse human AGPAT2: CGATGGGCACGTTCTCCC (product size:
118 bp), forward human GAPDH: AATCCCATCACCATCTTCCA, re-
verse human GAPDH: TGGACTCCACGACGTACTCA (product size:
82 bp). Mouse primers were used on MEF and CHO cell lines and
human primers were used on SK-OV-3 and HEK cell lines.
[10] M. Awale, J. L. Reymond, ChemRXiv 2018, DOI: https://doi.org/
10.26434/chemrxiv.6895646.v1.
[11] C. Simonin, M. Awale, M. Brand, R. van Deursen, J. Schwartz, M. Fine, G.
Kovacs, P. Hꢇfliger, G. Gyimesi, A. Sithampari, R. P. Charles, M. Hediger,
J. L. Reymond, Angew. Chem. Int. Ed. 2015, 54, 14748–14752; Angew.
Chem. 2015, 127, 14961–14965.
Acknowledgements
[12] F. Kilchmann, M. J. Marcaida, S. Kotak, T. Schick, S. D. Boss, M. Awale, P.
Gçnczy, J.-L. Reymond, J. Med. Chem. 2016, 59, 7188–7211.
[13] M. Radi, L. Evensen, E. Dreassi, C. Zamperini, M. Caporicci, F. Falchi, F.
Musumeci, S. Schenone, J. B. Lorens, M. Botta, Bioorg. Med. Chem. Lett.
2012, 22, 5579–5583.
[14] a) P. Mamalis, J. Green, D. J. Outred, M. Rix, J. Chem. Soc. 1962, 3915–
3926; b) T. Vilaivan, N. Saesaengseerung, D. Jarprung, S. Kamchonwong-
paisan, W. Sirawaraporn, Y. Yuthavong, Bioorg. Med. Chem. 2003, 11,
217–224; c) K. S. Shikhaliev, M. M. Liberman, D. V. Kryl’skii, Russ. Chem.
Bull. 2004, 53, 2876–2877; d) S. Mayer, D. M. Daigle, E. D. Brown, J.
Khatri, M. G. Organ, J. Comb. Chem. 2004, 6, 776–782; e) A. Maarouf, A.
Farahat, K. Selim, H. Eisa, Med. Chem. Res. 2012, 21, 703–710.
[15] M. K. Gilson, T. Liu, M. Baitaluk, G. Nicola, L. Hwang, J. Chong, Nucleic
Acids Res. 2016, 44, D1045–D1053.
This work was supported financially by the University of Bern, the
Swiss National Science Foundation, and NCCR TransCure. We
thank ChemAxon Pvt. Ltd. for providing free academic and web
licenses for their products, and Prof. Dr. Jꢀrg Gertsch for assis-
tance in off-target evaluation and helpful discussions.
Conflict of interest
The authors declare no conflict of interest.
[16] K. T. Nguyen, L. C. Blum, R. van Deursen, J.-L. Reymond, ChemMedChem
2009, 4, 1803–1805.
Keywords: angiogenesis
screening · polypharmacology · target prediction
· kinase inhibitors · phenotypic
[17] M. Awale, J. L. Reymond, J. Chem. Inf. Model. 2014, 54, 1892–1897.
[18] D. Rogers, M. Hahn, J. Chem. Inf. Model. 2010, 50, 742–754.
[19] a) T. Hideshima, D. Chauhan, T. Hayashi, K. Podar, M. Akiyama, C. Mit-
siades, M. I. N, B. Gong, L. Bonham, P. de Vries, N. Munshi, P. G. Richard-
son, J. W. Singer, K. C. Anderson, Cancer Res. 2003, 63, 8428–8436; b) F.
Hong, D. Hollenback, J. W. Singer, P. Klein, Bioorg. Med. Chem. Lett.
2005, 15, 4703–4707; c) J. M. Pagel, C. Laugen, L. Bonham, R. C. Hack-
man, D. M. Hockenbery, R. Bhatt, D. Hollenback, H. Carew, J. W. Singer,
O. W. Press, Clin. Cancer Res. 2005, 11, 4857–4866; d) R. Bhatt, B. Gong,
F. Hong, S. A. Jenkins, J. P. Klein, A. M. Kumar, J. Tulinsky (Cell Therapeu-
tics Inc.), Int. PCT Pub. No. WO2003037346 A1, 2003.
[20] M. Coon, A. Ball, J. Pound, S. Ap, D. Hollenback, T. White, J. Tulinsky, L.
Bonham, D. K. Morrison, R. Finney, J. W. Singer, Mol. Cancer Ther. 2003,
2, 1067–1078.
[21] P. La Rosee, T. Jia, S. Demehri, N. Hartel, P. de Vries, L. Bonham, D. Hol-
lenback, J. W. Singer, J. V. Melo, B. J. Druker, M. W. Deininger, Clin.
Cancer Res. 2006, 12, 6540–6546.
[1] A. P. Bento, A. Gaulton, A. Hersey, L. J. Bellis, J. Chambers, M. Davies,
F. A. Krꢃger, Y. Light, L. Mak, S. McGlinchey, M. Nowotka, G. Papadatos,
R. Santos, J. P. Overington, Nucleic Acids Res. 2014, 42, D1083–D1090.
[2] a) J. L. Jenkins, A. Bender, J. W. Davies, Drug Discovery Today Technol.
2006, 3, 413–421; b) A. Koutsoukas, B. Simms, J. Kirchmair, P. J. Bond,
A. V. Whitmore, S. Zimmer, M. P. Young, J. L. Jenkins, M. Glick, R. C. Glen,
A. Bender, J. Proteomics 2011, 74, 2554–2574; c) J. Lee, M. Bogyo, Curr.
Opin. Chem. Biol. 2013, 17, 118–126; d) A. Cereto-Massaguꢆ, M. J.
Ojeda, C. Valls, M. Mulero, G. Pujadas, S. Garcia-Vallve, Methods 2015,
71, 98–103; e) A. Lavecchia, C. Cerchia, Drug Discovery Today 2016, 21,
288–298; f) X. Liu, H. A. Baarsma, C. H. Thiam, C. Montrone, B. Brauner,
G. Fobo, J.-S. Heier, S. Duscha, M. Kçnigshoff, V. Angeli, A. Ruepp, M.
Campillos, Cell Chem. Biol. 2016, 23, 1302–1313; g) P. Schneider, G.
Schneider, Angew. Chem. Int. Ed. 2017, 56, 11520–11524; Angew. Chem.
2017, 129, 11678–11682; h) J. J. Irwin, G. Gaskins, T. Sterling, M. M. My-
singer, M. J. Keiser, J. Chem. Inf. Model. 2018, 58, 148–164.
[22] M. A. Blaskovich, V. Yendluri, H. R. Lawrence, N. J. Lawrence, S. M. Sebti,
G. M. Springett, PLoS One 2013, 8, e78632.
[23] A. Lagunin, A. Stepanchikova, D. Filimonov, V. Poroikov, Bioinformatics
2000, 16, 747–748.
[3] T. Kume, Histol. Histopathol. 2010, 25, 637–646.
[4] J. M. Reinke, H. Sorg, Eur. Surg. Res. 2012, 49, 35–43.
[24] M. J. Keiser, B. L. Roth, B. N. Armbruster, P. Ernsberger, J. J. Irwin, B. K.
Shoichet, Nat. Biotechnol. 2007, 25, 197–206.
[25] L. Wang, C. Ma, P. Wipf, H. Liu, W. Su, X.-Q. Xie, AAPS J. 2013, 15, 395–
406.
[5] a) Y. Alvarez, M. Cederlund, D. Cottell, B. Bill, S. Ekker, J. Torres-Vazquez,
B. Weinstein, D. Hyde, T. Vihtelic, B. Kennedy, BMC Dev. Biol. 2007, 7,
114; b) F. Bandello, R. Lattanzio, I. Zucchiatti, C. Turco, Acta Diabetol.
2013, 50, 1–20.
[26] D. Reker, T. Rodrigues, P. Schneider, G. Schneider, Proc. Natl. Acad. Sci.
USA 2014, 111, 4067–4072.
[27] D. Gfeller, A. Grosdidier, M. Wirth, A. Daina, O. Michielin, V. Zoete, Nucle-
ic Acids Res. 2014, 42, W32–W38.
[6] a) S. Yoshida, H. Amano, I. Hayashi, H. Kitasato, M. Kamata, M. Inukai, H.
Yoshimura, M. Majima, Lab. Invest. 2003, 83, 1385–1394; b) W. H.
Huang, M. C. Chang, K. S. Tsai, M. C. Hung, H. L. Chen, S. C. Hung, Onco-
gene 2013, 32, 4343–4354.
[28] J. Nickel, B.-O. Gohlke, J. Erehman, P. Banerjee, W. W. Rong, A. Goede, M.
Dunkel, R. Preissner, Nucleic Acids Res. 2014, 42, W26–W31.
[29] M. Awale, J. L. Reymond, J. Cheminf. 2017, 9, 11.
[7] a) K. Holmes, O. L. Roberts, A. M. Thomas, M. J. Cross, Cell. Signalling
2007, 19, 2003–2012; b) L. Claesson-Welsh, M. Welsh, J. Intern. Med.
2013, 273, 114–127.
&
ChemMedChem 2018, 13, 1 – 14
www.chemmedchem.org
12
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
[30] D. Filimonov, V. Poroikov, Y. Borodina, T. Gloriozova, J. Chem. Inf.
Comput. Sci. 1999, 39, 666–670.
[35] M. J. Bergeron, R. Boggavarapu, M. Meury, Z. Ucurum, L. Caron, P. Isenr-
ing, M. A. Hediger, D. Fotiadis, PLoS One 2011, 6, e21901.
[36] R. Boggavarapu, J.-M. Jeckelmann, D. Harder, P. Schneider, Z. Ucurum,
M. Hediger, D. Fotiadis, PLoS One 2013, 8, e76427.
[31] a) J. L. Medina-Franco, G. M. Maggiora, M. A. Giulianotti, C. Pinilla, R. A.
Houghten, Chem. Biol. Drug Des. 2007, 70, 393–412; b) J. L. Medina-
Franco, R. Aguayo-Ortiz, Mol. Inf. 2013, 32, 942–953; c) M. Awale, J. L.
Reymond, J. Chem. Inf. Model. 2015, 55, 1509–1516.
[32] M. Awale, D. Probst, J. L. Reymond, J. Chem. Inf. Model. 2017, 57, 643–
649.
[37] M. W. Karaman, S. Herrgard, D. K. Treiber, P. Gallant, C. E. Atteridge, B. T.
Campbell, K. W. Chan, P. Ciceri, M. I. Davis, P. T. Edeen, R. Faraoni, M.
Floyd, J. P. Hunt, D. J. Lockhart, Z. V. Milanov, M. J. Morrison, G. Pallares,
H. K. Patel, S. Pritchard, L. M. Wodicka, P. P. Zarrinkar, Nat. Biotechnol.
2008, 26, 127–132.
[33] L. Evensen, D. R. Micklem, W. Link, J. B. Lorens, Cytometry A 2010, 77A,
41–51.
[34] E. A. Harrington, D. Bebbington, J. Moore, R. K. Rasmussen, A. O. Ajose-
Adeogun, T. Nakayama, J. A. Graham, C. Demur, T. Hercend, A. Diu-Her-
cend, M. Su, J. M. C. Golec, K. M. Miller, Nat. Med. 2004, 10, 262–267.
Manuscript received: August 16, 2018
Accepted manuscript online: && &&, 0000
Version of record online: && &&, 0000
ChemMedChem 2018, 13, 1 – 14
www.chemmedchem.org
13
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

FULL PAPERS
M. Poirier, M. Awale, M. A. Roelli,
G. T. Giuffredi, L. Ruddigkeit, L. Evensen,
A. Stooss, S. Calarco, J. B. Lorens,
R.-P. Charles,* J.-L. Reymond*
Pharmacology expanded! We identi-
fied an aminotriazine that acts as a cyto-
static nanomolar inhibitor by screening
a focused library of kinase inhibitor ana-
logues in a phenotypic co-culture assay
for angiogenesis inhibition. To decipher
its mechanism of action, we used the
online target prediction tool PPB2
&& – &&
Identifying Lysophosphatidic Acid
Acyltransferase b (LPAAT-b) as the
Target of a Nanomolar Angiogenesis
Inhibitor from a Phenotypic Screen
Using the Polypharmacology Browser
PPB2
(http://ppb2.gdb.tools), which suggest-
ed lysophosphatidic acid acyltransferase
b (LPAAT-b) as a possible target for this
aminotriazine along with several ana-
logues identified by SAR profiling.
&
ChemMedChem 2018, 13, 1 – 14
www.chemmedchem.org
14
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!