Lead-discovery of bis-aromatic alkynes: A novel class of herbicides

Article abstract of DOI:10.2533/chimia.2008.23

The search for new active molecules with novel modes of action and desirable physical properties is an ongoing endeavour.^[1,2] This publication describes the follow-up chemistry of a biological hit discovered in the screening system of Novartis

Full text of DOI:10.2533/chimia.2008.23

HOT TOPICS
23
doi:10.2533/chimia.2008.23
CHIMIA 2008, 62, No. 1/2
Chimia 62 (2008) 23–28
© Schweizerische Chemische Gesellschaft
ISSN 0009–4293
Lead-Discovery of bis-Aromatic Alkynes:
A Novel Class of Herbicides
Jutta Glocka , James Allena , Thierry Nidermana , Hans Ulrich Haasa , Renold Cholleta , Martin Eberleb,
Peter Renolda , William Lutza , Jürg Ehrlerc , Marian Valentinid, Markus Waltia , Evelyne Siegera ,
Thomas Vettigera , Hans Brunnera , Julie E. Penzottie , Peter D. J. Grootenhuisf, Steven Bondyg,
Daniel D. Comerh, Soan Chengh, Arthur Steigera , Martin Zellera , Grit Laueb, Adrian Friedmanna ,
Olivier Jacoba , Mafalda Ninaa , Hans-Jürg Widmera , Klaus Kreuzi , and G. Wayne Craig*a
Recipients of the Sandmeyer Prize 2007 of the Swiss Chemical Society
Abstract: The search for new active molecules with novel modes of action and desirable physical properties is
an ongoing endeavour.[1,2] This publication describes the follow-up chemistry of a biological hit discovered in the
screening system of Novartis Crop Protection, the legacy agrochemical parent of Syngenta Crop Protection. This
chemistry was optimized through classical synthetic methods and automated parallel synthesis with coverage of
important physical properties such as lipophilicity or Clog P and solubility. Preliminary biological activity from the
greenhouse and field data with symptomology is presented.
Keywords: Agrochemicals · Alkyne template · Combinatorial chemistry · Discovery · Sonogashira coupling
1. Introduction
O
O
There is an abundance of alkyne-contain-
O
ing natural products which exhibit diverse
O
forms of biological activity (Fig. 1).[3]
O
OR
Their structural elements can be found in
n R
n
H
AcO
the spiroketals 1 and 2, salicylate 3, and
3
4a, 4b R = 2-furan n = 1
5a, 5b R = -Me n = 2
1a, 1b R = Ac E/Z n = 2
2a, 2b R = H E/Z n = 2
aromatic alkynes 4a,b. Many such com-
pounds derive from fatty acid biosynthe-
sis and are reported to show fungicidal,[4a]
insecticidal,[4b] and algicidal activity.[4c]
Fig. 1.
Cross-indication high-throughput screen- donor−acceptor-group relationship while
ing (HTS) of the fungicide analog 6 (Fig. 2) further exploring chemical space around
elicited a weak herbicidal response at 500 the original lead compound (Fig. 2).
g/Ha on Stellaria but was inactive on Nas- Essential in this ‘discovery process’was
turtium, Agrostis and Digitaria species. The the desire to investigate modifications that
symptoms were unknown and the chemis- would examine
try was attractive because it would be a i) physical property space
new class of agrochemical compounds.[5] ii) the intrinsic elements required for activ-
Therefore a combinatorial approach in col-
ity and
*Correspondence: Dr. G. W. Craiga
Tel.: + 41 61 323 2708
Fax: + 41 61 323 8726
laboration with CombiChem Inc.[6a] was es- iii) similarity to chemistry in the published
tablished to study the scope of this chemical
class.
domain.
E-mail: cralion@freesurf.ch
Our strategy was to use high-speed syn-
thesis (HSS) and classical synthetic meth-
ods to generate structure−activity relation-
ships (SAR).
aSyngenta Crop Protection Münchwilen A.G.
Schaffhauserstr., CH-4332 Stein
bNovartis Pharma A.G., CH-4002 Basel
cBachem A.G., CH-4416 Bubendorf
dSynkola A.G., SK-842 15 Bratislava, Slovak Republic
e5731 16th Ave NE, Seattle, WA 98105, USA
fVertex Pharmaceuticals, 11010 Torreyana Road,
San Diego CA 92121, USA
2. CombiChem Library Generation
The CombiChem approach used itera-
tion and integration with automated synthe- 3. Computational Methods
gGilead Sciences Inc., 333 Lakeside Dr., Foster City,
CA 94404, USA
sis and purification to refine computational
models and discover new compounds.[6a]
Three rounds of the iterative discovery
hTorrey Pines Therapeutics, Inc., 11085 N.
Torrey Pines Rd, Suite 300, La Jolla, CA 92037, USA
iBASF A.G., Ludwigshafen, Germany
In this application, the iterative discovery strategy were completed (Fig. 2). In each
strategy sought to maintain an optimal iteration, active and inactive data were

HOT TOPICS
24
CHIMIA 2008, 62, No. 1/2
pounds which contained 100 different phar-
macophores. Biological screening at No-
The Discovery Engine TM
vartis led to 15 compounds selected for fur-
O
CH3
O
Ensemble Hypotheses
Shape / Feature
ther profile screening (Table 1). Improved
activity was identified for compounds with
a preference for C(4)-hydrogen or C(4)-cy-
Cl
N
6
Pharmacophore
ano group in ring A or with a C(3)-halogen
or C(3)-acetonitrilo group in ring B (Fig. 5).
The herbicidal activity of the unsubstituted
analog 14 was confirmed in field tests and
*
Prior Information
Virtual Library
Design
ꢀ Lead Optimization
ꢀ Lead Evolution
Hypotheses
analyze data
search
became the lead compound for this proj-
ect.
+++>++>+
data analysis
Analysis of Table 1 showed that re-
Data
Lead
placement of the C(4)-Cl with a C(3)-ace-
tonitrilo in ring B (1 vs 4) or replacement of
the C(4)-acetonitrilo group in ring A with a
screen
synthesize
C(4)-cyano group (1 vs 6) resulted in 100%
herbicidal control of the broad-leaf plants.
Compounds
**
Simultaneous change at both these posi-
tions gave a slightly less active analogue 2
on broad-leafs. Furthermore, the herbicidal
activity is diminished in the absence of the
Universal Informer Library TM
*ex NCP: HTS-results / patent info / literature
**not applied in current collaboration
Fig. 2. Discovery
process. Increasing
biological activity
ꢀ
Lead Generation
C(2)-methoxy group (4 vs 5). An additional
C(4)-Cl to the C(3)-Cl in ring-B maintains
analyzed to derive an ensemble model
composed of three-dimensional pharma-
cophores associated with herbicidal activ-
the pharmacophore space described by the
model. With each iteration of design, syn-
thesis and screening, the pharmacophore
space of the ensemble model was evolved
to encompass newly discovered active and
herbicidal effects (3 vs 8). The C(3)-posi-
tion in ring B is judged the most responsive
ity. These models were used to design new
position overall.
chemical libraries to explore and validate
inactive compounds. The iterative Combi-
Chem approach to ensemble modeling has
been described previously.[7−10] The phar-
5. Exploration of Structure
macophore descriptor space was composed
of 2-, 3-, and 4-point pharmacophores de-
fined by a choice out of six feature types
(hydrogen-bond acceptors and donors, hy-
drophobes, negative and positive ionizable
groups, and aromatic groups) and their cor-
responding 3D distances (Fig. 3).[6b]
The general synthetic route is shown
a)
in Scheme 1.[15] Alkylation of substituted
phenols 15 with propargylic bromide 16 in
step i consistently gave 75−85% yields of
intermediate phenolic ethers 17. The So-
nogashira coupling[16] was optimally per-
formed in piperidine as solvent.[17] These
new variations confirmed the weak herbi-
cidal symptoms in subsequent biological
screening.
4. Biology
b)
The biological response and symptom-
5.1. Structure Comparisons in the
ology are shown in Figs 4a and 4b.
With parallel synthesis, CombiChem
synthesized in three cycles 1201 new com-
Patent Literature
We postulated the same mode-of-action
(MoA)fortheSankyochemistry[18] basedon
Biological response
a)
Biological symptoms
b)
c)
Fig. 3. Computation of molecule 6, (see Table
1) aligned to three examples of four-point
pharmacophores associated with herbicidal
effects. Each sphere represents
a chemical
feature point in the pharmacophore: hydrogen
bond acceptors (pink), aromatic groups (yellow)
and hydrophobic groups (brown).
Fig. 4. a) Stellaria, chickweed: the mature plant showed symptoms of desiccation and curling of the
leaves. b) Vicia fabia, fi eld beans: circled area in photo. The symptomology of discoloration in Vicia
fabia is also observed.

HOT TOPICS
25
CHIMIA 2008, 62, No. 1/2
CH3
O
R2
Br
O
R1
O
O
I
O
16
i)
18
ii)
3
1
5
R2
O
O
R2
3
5
OH
1
R1
R1
R1
14 R1 = -H 20 R1 = -NH2
15
17
Fig.5.Genericstructure,
see Tables 1 and 2
Ring A
Bridge
Ring B
19 R1 = -F 21 R1 = -CHO
R2 = -CH2CN
i) K2CO3 , acetone, reflux, 2-12 h or K2CO3 , DMF, rt, 2-4 h
ii) 1-5mol% Pd(TPP)4, Cu (I) iodide, piperidine, rt, 2-4 h,
Table 1. Names in black = Broad leafs; names in green = grasses. Ratings
of 3 = 90% activity; 2 = 50% activity; 1 = 10% activity. Application at 500 g/
Ha. Asterisk denotes the generic structure. Fig. 5 without a 2-OCH3 group.
More active values in red.
Scheme 1.
Substitutents
High-Throughput Screen (HTS)
Sankyo
Hybrids
O
O
O
R 1
R 2
O
Alkyne Project
O
N
H
O
N
H
O
X
Cl
23a, b X = Cl or -CH2CN
22a
O
N
O
O
O
N
N
O
1
3
3
3
2
3
3
3
3
0
2
2
0
0
0
0
0
1
0
2
3
3
2
3
3
2
3
0
0
0
0
0
0
0
0
0
O
N
1 3.91
2 2.68
3 3.97
4 2.36
5 2.62
6 3.97
7 4.12
8 4.56
9 2.94
4-CH2CN
4-CN
4-Cl
3-CH2CN
3-Cl
-
-
-
-
O
N
O
X
7
N
H
O
0
0
0
0
0
0
0
0
1
1
0
0
0
0
0
1
3
2
3
2
3
3
2
3
3
3
3
1
3
3
3
3
22b
24a, b X = Cl or -CH2CN
4-CN
4-CH2CN 3-CH2CN
4-CH2CN* 3-CH2CN
Scheme 2.
a)
4-CN
4-CN
4-CN
-H
4-Cl
3-Br
b)
3,4-diCl
3-CH 2CN
O
H
N
O
Test Species. The following plant species were used in primary and
secondary screening. Broad-leaf weeds tested included chickweed
(Stellaria media); pigweed (Amaranthus retroflexus); watercress
(Nasturtium); black nightshade (Solanum nigrum); lambsquarters
(Chenopodium album). Grasses tested: crabgrass (Digitaria sanguinalis);
bluegrass (Poa annua); barnyard grass (Echinochloa crus-galli); giant
foxtail (Setaria faberi); bentgrass (Agrostis).
N
O
O
Fig. 6. a) Isoxaben; b) lead 14
a similar biological spectrum, symptomol-
ogy and structural elements (Scheme 2).
The active Sankyo analogs 22a,b showed
that the bridge and ring B were exchange-
R 3
R 3
O
O
i)
Cl
R 1
+
Cl
R 1
OH
O
Tsl
O
able elements. Therefore the appropriate
R 1 = NHCOR
29a
R 1 = NHCOR
R 3 = -H
25
27
hybrids of Sankyo’s chemistry[18] (red, yel-
low in Scheme 2) and Novartis’ chemistry
(blue) were synthesized. But the inactivity
of compounds 23 and 24 implicated dif-
ferent requirements for biological activity
and probably a different MoA.
N
N
N
N
N
NH
N
NH
R 1
=
i) K2CO3, DMF, 500C
R 1
=
N
N
N
N
N
N
N
N
28a
28b
28c
26a
26b
26c
Scheme 3.
In the course of systematic investi-
gations, the C(4)-F analog 19 was syn-
5.2. Structural Overlap Comparison
5.3. Chemical Modifications of Ring A
thesized. This compound showed a ten-
fold-rate increase in herbicidal activity.
Similarly fluorine was introduced at all
Although the MoA was not known at
this stage of the project some observed
symptomatic similarities to the known
The HSS group synthesized a number
of aniline derivatives 27 (Scheme 3. R1 =
-NHCOR) in order to explore the physical
property space to improve uptake or trans-
port properties. Although the lipophilicity
or Clog P varied from 1.2 to 6.8 most of
these simple anilides 27 did not show im-
proved biological activity.
Cyano-mimicssuchaspyrazole26a, tet-
razole 26b or imidazole 26c intermediates
positions in ring A of lead 14. The stron-
commercial broad-spectrum pre-emer-
gence herbicide, isoxaben (Fig. 6a) were
ger C(4)-F bond effect on physical proper-
ties may improve uptake-transport to the
observed. This herbicide induces blacken-
target site.[19] The ring A analogues with
ing of the leaves and swelling of roots in
sugarbeets, a broad-leaf crop. Furthermore,
the structural shape of isoxaben was com-
patible with our alkyne system (Fig. 6b).
This led to introduction of different bridge
types or replacement of either benzene ring
with a heterocycle.
fluorine in positions C(3), C(5) and C(6)
maintained herbicidal symptoms but were
weaker herbicides. Moreover the C(4)-F
may also improve the biological fit of the
inhibitor at the active site.
(Scheme 3) were synthesized by literature
methods utilizing C(1)-O-methyl-deprotec-
tion with thiolate salts.[20] This strategy pro-

HOT TOPICS
26
CHIMIA 2008, 62, No. 1/2
oxygen were detrimental to the herbi-
cidal activity.
Cl
N
N
O
HO
O
Cl
Restricted rotation of ring A or ring B
by a tethered connection to the bridge i.e.
benzofuranyl or 1,4-benzodioxinyl 39a,b
respectively (Scheme 5) resulted in weaker
analogues than the lead 14.
30
i)
B
B
A
N
A
O
B
N
N
N
Cl
A
32a
32b
N
=
O
O
N
32a-d
29
N
N
i) K2 CO3 , DMF, propargyl alcohol,
60 °C
5.5. Chemical Modifications of Ring B
32c
32d
3-Iodophenylacetonitrile (40d) was pro-
duced by free-radical bromination and cya-
nation of the 3-iodobenzylbromide (40b).
Scheme 4. Heterocyclic ring A
Phase-transfer alkylation[5d] of the arylce-
tonitrile 40d gave 40f, 40g whereas the lead
14 failed to react under the same conditions.
Sodiumhydridealkylationofarylacetonitrile
40d gave cyclopropane 40e in good yield
(Scheme 6). Extension of the cyano group in
40c by LDA-alkylation of 3-iodobenzylbro-
mide (40b) in acetonitrile gave the desired
building block 40c. Subsequent cross-cou-
pling gave the corresponding analogues but
most of these modifications resulted in loss
CH3
O
O
O
CH3
OH
O
O
i)
ii) 36, iii)
Br
R1
O
O
+
O
O
R1
R1
R1
34a R1 = -H, -Me
34b R1 = -Me, -Me
33
47
39a
O
CH3
O
CH3
iv), v)
O
O
O
O
O
O
P
N+
38
O
R1
39a
R1
R1
of herbicidal activity. However the pyridine
ring B analogue 41 did show significant her-
bicidal effects comparable to the lead mol-
H
R1
O
39b
N
38
37a
i) K2CO3, DMF, 800C, 8 h, ii) DiBal, Toluene, -600C, CH2Cl2 iii) Swern
Tethered systems
oxidation iv) Ohira Reagent 38, -600C v) K2CO3, MeOH, 00C
ecule 14 (Table 2).
Palladium coupling was the most ef-
ficient with iodobenzene partners which
Scheme 5. Alternative synthesis of alkyne bridge
readilycross-coupledatroomtemperature.
Cross-coupling with bromobenzenes on
the other hand gave variable results and of-
ten required microwave conditions, 150 °C
N
for 15 min. to give appreciable product.
Attempts to utilize benzene triflate esters
in the cross-coupling were unsuccessful
except for meta-substituted benzenes.
I
N
N
I
Br
CH3
iv)
v)
40 c
40 d
40 e
i)
iii)
ii)
I
I
H
CH3
3CH
3CH
N
N
Thiophene and pyridine iodides 42, 43
respectively coupled efficiently. The pyri-
dine analogues 42a, 42b showed moderate
40 b
40 a
I
I
I
+
40 f
40 g
herbicidalactivityandalthoughthiophene,
i) NBS, AIBN, dibenzoylperoxide ii) n-butyl) 4NBr, KCN, NaOH aq, CH2Cl2, iii) LDA, acetonitrile,
- 600C iv) BrCH2CH2Br, NaH, DMF, rt, 12 h v) n-butyl) 4NBr, CH3-I, NaOH aq, CH2Cl2
considered an isostere for benzene[23]
gave the inactive analogue, 43. The rela-
tive position of the thiophene acetonitrile
group may be critical. More significantly,
the pyridine analogues encouraged us to
look at more lipophilic functionality such
as oxime-ethers rather than basic polar
Scheme 6. Ring B variations
duced three isosteric replacements for the ever gave undesired elimination products heterocycles to improve the herbicidal
activity. The synthesis of oxime ethers
An alternative approach was undertaken 44, 45, 46a, 46b was accessible from
cyano group but all analogues 28a−c were (Scheme 3, R3 = -CH3).
biologically inactive.
A number of heterocyclic analogues whereby alkylation of catechol 33 with ethyl the benzaldehyde analog 21 (Scheme 2).
were synthesized to evaluate the scope of α-bromoacetate34a, reductionofresulting Their biological results (Table 2) showed
variations for ring A. These analogues were
ester 35 followed by Swern oxidation gave a significantly improved broad-leaf weed
obtained by alkylation of propargyl alcohol the phenoxyacetaldehyde 37a (Scheme 5, spectrum.
30 with halogenated pyridines 29a, pyra- R1 , R1 = -H, -CH3). Subsequent Wittig-
zine 29b and pyrimidines 29c−d,[21] to yield reaction with Ohira reagent 38 produced
the corresponding ethers 30 for cross-cou- the alkyne 39a in excellent yield.[22] Fi- 6. Lead Optimization
pling with 3-bromophenylacetonitrile 31a nally, cross-coupling afforded the desired
(Scheme 4).
bridged-alkylated analogue 14a which 6.1. Field Results
was biologically weaker than the lead 14.
The field screening is the most critical
5.4. Chemical Modifications of the
Isosteric replacement of the triple bond phase, since here the herbicidal activity
Bridge
with
must withstand variations in physical pa-
rameters such as humidity, temperature and
soil variations which can cause the com-
In order to obtain alkyl modifications of i) a phenyl ring
the bridge, the molecule was built up from ii) extension with another triple bond or
ring B. Activation of the propargylic alcohol iii) incorporation of carbon, nitrogen or pound to be poorly transported to the plant
as a tosylate or mesylate (29a, 29b) how- sulfur atoms for the propargylic C(1)- or where chemical instability decreases its

HOT TOPICS
27
CHIMIA 2008, 62, No. 1/2
Table 2. Secondary screen of broad-leaf plants. Ratings of 1 = 90% activity; 3 = 70% activity; 9 =
inactive. Application at 250 g/Ha. Asterisk denotes the generic structure. Fig. 5 contains a pyridine Ring
B. Note that Table 2 ratings are reversed compared to those in Table 1. More active values in red.
gave a flat SAR response. Biological at-
tenuation appears to be restricted to sim-
ple functionality and that some specific
substitutents appear essential, such as a
C(2)-methoxy group (ring A) and a C(3)-
Substitutents
Broad-Leaf Weeds
acetonitrilo group (ring B) for biological
activity.
This successful lead-discovery project
was subsequently promoted to optimiza-
tion status. Although this lead develop-
R1
R2
ment did not succeed in bridging the gap
to another structure class, the potential for
broad-spectrum development in crops still
4
3
4
4
3
2
3
3
9
4
9
9
9
9
9
4
9
5
7
5
5
9
6
4
4
1
4
5
1
4
4
1
14
19
41
42
43
-H
4-F
4-F
-H
3-CH2CN
3-CH2CN
3-Cl-4*
1
3
4
3
1
2
1
1
2
3
3
3
3
3
3
3
3
6
7
2
7
7
8
9
3
2
7
3
3
3
4
3
2
1
3
6
1
2
2
2
2
2
4
9
3
2
3
4
3
2
6
5
4
9
5
4
4
3
8
6
4
4
5
3
2
2
2
3
2
2
1
1
remains. The interest in broad-leaf control
in crops such as cereals and overall im-
provement of the herbicidal activity remain
the two primary objectives in its optimiza-
tion project status for this novel chemistry
and unknown MoA.
3-Me2CCN
3-Me
Acknowledgments
4 -CH=N
-OAllyl
We are indebted to Syngenta FT-NMR
specialist, Dr. Tammo Winkler and LC-MS
expert, Dr. Andreas Stämpfli, currently at
Hoffmann-La Roche, for valuable technical
assistance. Also we thank our colleagues Drs.
Kurt Nebel and Juergen Schaetzer for their
professional support in the transfer of this
project to optimization development and Dr.
Anita Schnyder at Solvias AG, for important
44a 4 -CH=N
-OMe
44b 4 -CH=N
-OMe
3-Cl
3-CH2CN
3-CH2CN
45
4 -CH=N
-OEt
probe-experiments for large-scale applications
of the Sonogashira chemistry. Finally, we thank
our experienced chemistry practikanten, Fabien
Dufour at University of Metz, and Olivier
Sonntag at Nottingham University for their
diligent synthetic contributions.
Fig. 7. Compound
19 applied at 500 g/
Ha. Test crops: (from
left to right) wheat,
durum wheat, barley.
Test species: Capsella,
Chenopodium, Emex,
Gallium, Kochia,
Lamium, Matricaria,
Papaver, Polygonum,
Stellaria, Veronica hed.,
and Viola.
Received: November 25, 2007
[1] S. F. Donovan in ACS Symposium
Series, 948, ‘Synthesis and Chemistry of
Agrochemicals VII’, American Chemical
Society, 2007, pp.7−22.
[2] K. Tietjen,
M. Drewes, K. Stenzel,
Combinatorial Chemistry and High-
Throughput Screening 2000, 8, 589.
[3] a) H. Boehm, D. Banner, S. Bendels, M.
Kansy, B. Kuhn, K. Mueller, U. Obst-
Sander, M. Stahl, Chem. Bio. Chem.
2004, 5, 637; b) P. L. Christensen, Recent
semble refined by the CONAN algorithm
Research DevelopmentsinPhytochemistry
1998, 2, 227.
overall availability (Fig. 7). The C(4)-F
gave a first generation set of analogues.
Subsequent combinatorial methodology
using standard Sonogashira conditions
produced a new set of molecules that re-
sulted in a number of improved biological
actives. In parallel, we focused on their
structural similarity to known chemistry in
the patent literature in order to clarify its
MoA. In addition to the herbicidal effects,
the novel discoloration symptoms in Vicia
fabia were used to focus on what structural
changes might be altered to delineate the
scope of this new class of herbicide. After
two further rounds of iterations, new mol-
analog 19 showed greater than 90% con-
trol of broad-leaf weeds without any phy-
totoxic effects on grass-like crops such as
wheat or barley. Moreover, additional sedge
and Echinochloa activity give this herbicide
potential in rice applications.
[4] a) N. H. Oberlies, L. L. Rogers, J. M.
Martin, J. L. McLaughlin, J. Nat. Prod.
1998, 61, 781; b) H. Y. Li, S. Matsunaga,
N. Fusetani, J. Nat. Prod. 1994, 57, 1464;
c) V. U. Ahmad, N. Alam, J. Nat. Prod.
1995, 58, 1426; c) L. Hagmann, F. Jüttner,
Tetrahedron Lett. 1996, 37, 6539.
[5] a) K. Hirai, A. Uchida, R. Ohno, ‘Major
Synthetic Routes for Modern Herbicide
Classes and Agrochemical Characteristics’,
in ‘Herbicide Classes in Development’,
Eds. P. Boeger, K. Wakabayashi, K.
Hirai, Springer-Verlag, Berlin, Germany,
2002, pp.179−289; b) F. Cederbaum,
A. De Mesmaeker, A. Jeanguenat, H-J.
7. Conclusion
A major challenge in the discovery of
new chemistry lies in the progression from
one structure class to another with im-
provement of desired biological activity.
ecules showed hints toward grass selectivi-
The initial HTS screening hit compound 6,
Kempf, C. Lamberth, A. Schnyder, M.
Zeller, R. Zeun, Chimia 2003, 57, 680;
c) G. W. Craig, M. Eberle, B. Irminger,
A. Schückenböhmer, Y. Laime, P. Müller,
Heterocycles 2007, 71, 1967.
ty. Functional groups such as oxime-ethers
showed hints of herbicidal effects but was
weak overall. The biological symptoms in-
43−45 showed a trend toward a broadened
biological spectrum. However it was ob-
served that many structural modifications
dicate a novel MoA. Lead evolution using
an established computational model en-

HOT TOPICS
28
CHIMIA 2008, 62, No. 1/2
[6] a) CombiChem Inc., Molecular Design
Group, 1804 Embarcadero Road, Suite
201, Palo Alto, 94303, USA; b) A mo-
lecular ‘signature’ was created for each
molecule by generating a full conforma-
tional model using CONAN,[11] and then
mapping the presence or absence of all 2-,
3-, and 4-point pharmacophores that are
present in a molecule’s conformers into
scaffolds were evaluated using the round
three model and a statistical sampling
approach for virtual library searching.[13]
Bondy, D. D. Comer, J. E. Penzotti, P. D.
J. Grootenhuis, WO Patent Publication
No. WO2003013247, 2003.
This sampling scheme estimates the rates
[16] a) K. Sonogashira, Y. Tohda, N. Hagihara,
that a virtual library will produce com-
pounds matching a defined property such
as model-matching compounds. Model
matching compounds were defined as
compounds which matched 40 or more
Tetrahedron Lett. 1975, 16, 4467; b) K.
Sonogashira in ‘Metal-Catalyzed Cross
Coupling Reactions’, Eds. F. Diederich,
P. J. Stang, Wiley-VCH, New York, 1998,
Chap. 5.
a single bit string. Thus, each bit in this
of the 100 pharmacophores in the round [17] R. Chinchilla, C. Najera, Chem. Rev. 2007,
three ensemble model. Three templates 107, 874.
were selected for larger designed librari- [18] M. Shimizu, T. Hiyama, Angew. Chem.,
es based on their predicted rates of gene- Int. Ed. 2005, 44, 214.
rating model-matching compounds, the [19] a) H. Sano, H. Komai, J. Kadotani, S.
success of test chemistries and chemical Morimoto, JP Patent Publication No.
similarity to other chosen templates. For JP11147866A, 1999; b) S. Sugai, H.
each of these three templates, an informa-
string corresponds to a particular pharma-
cophore that the molecule is able to pre-
sent to a receptor (bit = 1) or not (bit =
0). Flexible, feature-rich molecules may
be able to present thousands of different
pharmacophores. To select the subset of
all pharmacophore hypotheses potentially
associated with herbicidal activity, the sig-
natures of all compounds (active and inac-
tive) were systematically analyzed in the
context of their associated activity data.
Komai, N. Kudo, K. Sato, T. Honma,
J. Kadotani, K. Koi, M. Ito, EP Patent
Publication No. EP795550 A1, 1997.
tive design strategy based on maximizing
Shannon entropy was used to design com-
binatorial matrix libraries with maximal [20] A. K. Chakraborti, L. Sharma, M. K.
information about the ensemble.[14] For
Nayak, J. Org. Chem. 2002, 67, 6406.
each template library design, monomer [21] C. Lamberth, Heterocycles 2006, 68, 561.
analogs of active compounds were forced [22] a) S. Ohira, Synth. Commun. 1989, 19,
into the design to address specific SAR 561; b) F. Eymery, B. Iorga, P. Savignac,
questions. A second set of monomers were Synthesis 2000, 185; c) Z. Wang, S.
computational selected using informative Campagna, G. Xu, M. E. Pierce, J. M.
active and inactive compounds. The top design from a subset of monomers pre- Fortunak, P. N. Confalone, Tetrahedron
valent in model-matching virtual library Lett. 2000, 21, 4007.
compounds. Additional monomers were [23] C. G. Wermuth, J. de la Fontaine,
computationally selected to further explo- ‘Molecular Variations Based on Isosteric
re the chemical space of each template. Replacements’, in ‘Practice of Medicinal
ted for each of the three rounds. The fi rst The result of this design method is a selec- Chemistry’, Ed. C. G. Wermuth, 2nd
The hypotheses were then ranked based
on their ability to discriminate between
actives and inactives across the entire data
set. The ranking criterion used was ‘infor-
mation content’,[12] a function of both the
100 pharmacophores were used to analy-
ze virtual chemical libraries and to select
compounds for the next round of synthe-
sis and screening. This process was repea-
design cycle began with one compound
in the propargyl template shown above
exhibiting weak herbicidal activity, fi ve
chemically related inactive compounds
and a set of chemically diverse inactives.
Compounds libraries were designed to
further explore the SAR around the weak
tion of molecules whose combination of
on-bits (accessible pharmacophores) and
off-bits (inaccessible pharmacophores) in
their molecular signatures systematically
investigates the pharmacophore hypothe-
ses combined with compounds selected to
address specific SAR questions.
edition, Elsevier, London, UK, 2003,
pp.189−214.
lead compound. The designs included
chemists’ selections and computationally
designed compounds selected for mat-
[7] E. K. Bradley, P. Beroza, J. E. Penzotti, P.
D. J. Grootenhuis, D. C. Spellmeyer, J. L.
Miller, J. Med. Chem. 2000, 43, 2770.
ching pharmacophores from an ensemb- [8] V. Molteni, J. Penzotti, D. M. Wilson,
le model derived from the initial weakly
active lead compounds. A total of 268
compounds were synthesized and scree-
ned yielding 15 new active compounds. A
A. P. Termin, L. Mao, C. M. Crane, F.
Hassman, T. Wang, H. Wong, K. J. Miller,
S. Grossman, P. D. J. Grootenhuis, J. Med.
Chem. 2004, 47, 2426.
second round of design and synthesis in- [9] J. E. Penzotti, M. L. Lamb, E. Evensen,
corporating information from the new ac-
tives and inactives from the fi rst round re-
P. D. J. Grootenhuis, J. Med. Chem. 2002,
45, 1737.
sulted in 18 additional active compounds [10] A. Smellie, R. Stanton, R. Henne, S. Teig,
out of 302 total compounds synthesized J. Comput. Chem. 2003, 24, 10.
and screened. By the third round, 372 [11] J. K. Lanctot, S. Putta, C. Lemmen, J.
compounds were used for the training set
including 16 compounds with the high-
Greene, J. Chem. Inf. Comput. Sci. 2003,
43, 2163.
est herbicidal activity rating of 3 against [12] P. Beroza, E. K. Bradley, J. E.
Stellaria and 16 additional compounds Eksterowicz, R. Feinstein, J. Greene, P. D.
with weaker activity against Stellaria. The J. Grootenhuis, R. M. Henne, J. Mount,
W. A. Shirley, A. Smellie, R. V. Stanton,
D. C. Spellmeyer, J. Mol. Graph. Model.
2000, 18, 335.
pharmacophores associated with herbici- [13] J. L. Miller, E. K. Bradley, S. L. Teig, J.
dal activity included three chemical fea- Chem. Inf. Comput. Sci. 2003, 43, 47.
ture types: hydrogen-bond acceptors, hyd- [14] E. K. Bradley, J. L. Miller, E. Saiah, P. D.
inactives now included compounds from
the propargyl ether template in addition
to other diverse compounds. The top 100
rophobes and aromatic groups (Scheme 2
shows three examples). Ninety-nine of the
J. Grootenhuis, J. Med. Chem. 2003, 46,
4360.
top 100 pharmacophores were four-point [15] a) G. W. Craig, M. Eberle, M. Zeller,
pharmacophores with the most commonly
observed feature combinations: Acceptor-
S. S. Bondy, D. D. Comer, S. Cheng,
J. E. Penzotti, P. D. J. Grootenhuis,
J. Ehrler, WO Patent Publication No.
WO2001055066, 2001; b) J. Schaetzer,
M. Eberle, J. Wenger, S. Berteina-Raboin,
K. Nebel, A. Stoller, R. G. Hall, S. S.
Acceptor-Hydrophobe-Aromatic (n
=
41) and Acceptor-Acceptor-Hydrophobe-
Hydrophobe (n = 39). More than 4 mil-
lion compounds on 11 different chemical