Discovery of novel allosteric non-bisphosphonate inhibitors of farnesyl pyrophosphate synthase by integrated lead finding

Article abstract of DOI:10.1002/cmdc.201500338

Farnesyl pyrophosphate synthase (FPPS) is an established target for the treatment of bone diseases, but also shows promise as an anticancer and anti-infective drug target. Currently available anti-FPPS drugs are active-site-directed bisphosphonate inhibitors, the peculiar pharmacological profile of which is inadequate for therapeutic indications beyond bone diseases. The recent discovery of an allosteric binding site has paved the way toward the development of novel non-bisphosphonate FPPS inhibitors with broader therapeutic potential, notably as immunomodulators in oncology. Herein we report the discovery, by an integrated lead finding approach, of two new chemical classes of allosteric FPPS inhibitors that belong to the salicylic acid and quinoline chemotypes. We present their synthesis, biochemical and cellular activities, structure-activity relationships, and provide X-ray structures of several representative FPPS complexes. These novel allosteric FPPS inhibitors are devoid of any affinity for bone mineral and could serve as leads to evaluate their potential in none-bone diseases. Beyond the bone: Farnesyl pyrophosphate synthase (FPPS) is an important target for osteoporosis and bone metastases, and holds promise for a number of non-bone diseases, including cancer, parasitic infections, progeria, and Alzheimer's disease. Herein we describe two novel chemotypes of allosteric FPPS inhibitors. These are useful leads to evaluate the therapeutic potential of FPPS inhibitors for these new indications.

Full text of DOI:10.1002/cmdc.201500338

DOI: 10.1002/cmdc.201500338
Full Papers
Very Important Paper
Discovery of Novel Allosteric Non-Bisphosphonate
Inhibitors of Farnesyl Pyrophosphate Synthase by
Integrated Lead Finding
Andreas L. Marzinzik,*[a] RenØ Amstutz,[a, b] Guido Bold,[a] Emmanuelle Bourgier,[a]
Simona Cotesta,[a] J. Fraser Glickman,[a, c] Marjo Gçtte,[a] Christelle Henry,[a] Sylvie Lehmann,[a]
J. Constanze D. Hartwieg,[a] Silvio Ofner,[a] Xavier PellØ,[a] Thomas P. Roddy,[a, d] Jean-
Michel Rondeau,[a] FrØdØric Stauffer,[a] Steven J. Stout,[a, e] Armin Widmer,[a]
Johann Zimmermann,[a, f] Thomas Zoller,[a] and Wolfgang Jahnke*[a]
Farnesyl pyrophosphate synthase (FPPS) is an established
target for the treatment of bone diseases, but also shows
promise as an anticancer and anti-infective drug target. Cur-
rently available anti-FPPS drugs are active-site-directed bi-
sphosphonate inhibitors, the peculiar pharmacological profile
of which is inadequate for therapeutic indications beyond
bone diseases. The recent discovery of an allosteric binding
site has paved the way toward the development of novel non-
bisphosphonate FPPS inhibitors with broader therapeutic po-
tential, notably as immunomodulators in oncology. Herein we
report the discovery, by an integrated lead finding approach,
of two new chemical classes of allosteric FPPS inhibitors that
belong to the salicylic acid and quinoline chemotypes. We
present their synthesis, biochemical and cellular activities,
structure–activity relationships, and provide X-ray structures of
several representative FPPS complexes. These novel allosteric
FPPS inhibitors are devoid of any affinity for bone mineral and
could serve as leads to evaluate their potential in none-bone
diseases.
Introduction
Farnesyl pyrophosphate synthase (FPPS) is the main molecular
target of nitrogen-containing bisphosphonates (N-BPs), one of
the most transformative class of drugs of the past 25 years[1]
that had a huge impact on the treatment of bone diseases
such as osteoporosis, Paget’s disease of bone, and bone meta-
stasis.[2–4] FPPS is a key node of the mevalonate pathway and
provides the isoprenoid precursors geranyl- (GPP) and farnesyl-
(FPP) pyrophosphate necessary for the synthesis of cholesterol,
dolichol, and ubiquinone, and for post-translational prenylation
of small GTPases such as Ras, Rac, and Rho. Bisphosphonate
drugs such as zoledronic acid (ZOL) 1 (Scheme 1) strongly bind
to bone and are then selectively taken up into bone-resorbing
osteoclasts by fluid-phase endocytosis, subsequently inducing
apoptosis and thereby inhibiting bone resorption.
[a] Dr. A. L. Marzinzik, Dr. R. Amstutz, Dr. G. Bold, E. Bourgier, Dr. S. Cotesta,
Dr. J. F. Glickman, Dr. M. Gçtte, C. Henry, S. Lehmann, Dr. J. C. D. Hartwieg,
Dr. S. Ofner, X. PellØ, Dr. T. P. Roddy, Dr. J.-M. Rondeau, Dr. F. Stauffer,
S. J. Stout, A. Widmer, Dr. J. Zimmermann, Dr. T. Zoller, Dr. W. Jahnke
Novartis Institutes for BioMedical Research
Basel, 4002 (Switzerland)
E-mail: wolfgang.jahnke@novartis.com
andreas.marzinzik@novartis.com
[b] Dr. R. Amstutz
Current address: Conim AG, Oberwiler Kirchweg 4c, 6300 Zug (Switzerland)
[c] Dr. J. F. Glickman
Current address: High Throughput and Spectroscopy Resource Center
Rockefeller University, New York, NY 10065 (USA)
[d] Dr. T. P. Roddy
Current address: Agios, Cambridge, MA 02139-4169 (USA)
[e] S. J. Stout
Current address: Merck Research Laboratories
126 E. Lincoln Avenue, Rahway, NJ, 07065 (USA)
[f] Dr. J. Zimmermann
Current address: Polyphor
Hegenheimermattweg 125, 4123 Allschwil (Switzerland)
Scheme 1. Structures of zoledronic acid 1 and selected allosteric FPPS inhibi-
tors 2 and 4, discovered by fragment-based screening.[18] Compound 3 rep-
resents a series of new FPPS inhibitors that consist of a salicylic acid frag-
ment substituted at C4.
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201500338.
ChemMedChem 2015, 10, 1884 – 1891
1884
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Intriguingly, N-BPs have been found to exhibit a surprising
variety of other pharmacological activities that include in-
creased longevity in a mouse model of progeria,[5] as well as
antiviral[6] and antiparasitic[7] effects. In the clinic, antitumor ef-
fects have been observed in breast cancer[8] and multiple mye-
loma[9] patients. The exact mechanisms underlying the antitu-
mor activity are not yet fully understood. They may include
apoptosis of tumor cells due to increased demand for prenyla-
tion in rapidly proliferating cells, and activation of gd Tcells by
accumulating and presenting phospho-antigens such as iso-
pentenyl pyrophosphate (IPP), one of the FPPS substrates.[10,11]
Both mechanisms involve FPPS inhibition, therefore pointing
to FPPS as a potential novel drug target for direct antitumor
therapy.
Herein we describe two additional chemical lead series of al-
losteric FPPS inhibitors that were discovered during the course
of our integrated lead finding efforts. These two series, salicylic
acid and quinoline derivatives, are fivefold more potent in
vitro than our first lead series represented by the benzoindole
derivative 2, and are equally potent to the best bisphospho-
nates. They are again devoid of significant affinity toward bone
mineral, and therefore represent interesting lead molecules for
the treatment of non-bone-related diseases as anticancer or
anti-infective agents.
In an accompanying report, we describe how such lead mol-
ecules can also be optimized for bone indications by fine-
tuning their affinity toward bone mineral through covalent at-
tachment of a suitable bone-affinity tag, while simultaneously
retaining desirable properties such as oral bioavailability.[22]
N-BPs have unusual pharmacokinetic (PK) properties charac-
terized by rapid and strong binding to bone mineral, associat-
ed with low plasma drug concentration soon after administra-
tion, low permeability into non-endocytic cells, and low oral Results and Discussion
bioavailability.[2] While this peculiar PK profile is a blessing for
Discovery of the salicylic acid class of allosteric FPPS
inhibitors
the treatment of bone diseases and contributes to the excep-
tional safety and potency of this class of drugs, it severely
compromises the efficacy of N-BPs as antitumor drugs, for
which high cell permeability and high plasma levels over an
extended period of time are required.
The fragment screen by NMR and fragment optimization work
were performed so rapidly that compound 2 had already been
discovered[18] by the time the high-throughput screen was run.
The HTS was based on a scintillation proximity assay (SPA) that
Crystallographic analyses of human FPPS complexes with
marketed N-BPs[12,13] have revealed how these drugs mimic
one of the FPPS substrates, dimethylallyl pyrophosphate
(DMAPP), and explained why early attempts to move away
from the bisphosphonate chemotype were largely unsuccess-
ful.[14] Currently marketed N-BPs are very small in size but are
highly efficient inhibitors. The geminal bisphosphonate moiety
of these drugs fills the site normally occupied by the pyro-
phosphate group of DMAPP, and plays a major role in recogni-
tion and binding through strong electrostatic interactions with
a trinuclear Mg2+ center and three basic side chains: Arg112,
Lys200, and Lys257. As all bisphosphonate oxygen atoms make
important contributions to these interactions, modifying or re-
placing the bisphosphonate moiety by a less polar isostere
long seemed virtually impossible, leaving the central carbon
atom as the only available position for chemical elaboration of
novel inhibitors. In particular, N-BPs with extended, lipophilic
side chains exploiting the GPP/FPP pocket have been investi-
gated, with the goal to expand the clinical utility of FPPS inhib-
itors.[15–17]
3
monitored the incorporation of H-labeled IPP into [3H]farnesyl
pyrophosphate ([3H]FPP) by capturing the latter onto a phos-
phatidylserine-coated scintillating microtiter plate (“flash-
plate”).[23] In total, 850000 compounds were screened at a con-
centration of 10 mm. The best non-bisphosphonate hits had
IC50 values in the single-digit to double-digit micromolar range.
Of particular interest was compound 3a (Scheme 1, R=H),
a salicylic acid derivative with an IC50 value of 54 mm in the SPA
identified by a similarity search from a salicylic acid fragment
identified by NMR. Although the activity of 3a was close to
the detection limit, we noticed that this hit shared some sub-
structures (naphthyl and benzyloxy moiety, carboxylic acid
functionality) with other NMR fragment hits or optimized allo-
steric inhibitors, such as 2 and 4 (Scheme 1). As this compound
was also deemed chemically attractive, it was selected for
follow-up validation and characterization studies.
FPPS was co-crystallized with 3a, and the X-ray structure of
the complex was determined. Crystallographic analysis con-
firmed that this hit was a genuine allosteric inhibitor of FPPS,
binding to the same allosteric pocket as the previously identi-
fied NMR fragment hits, with the enzyme in the open state
and the flexible C-terminal tail disordered (Figure 1A).[12] This
allosteric pocket is mainly lined by a-helices aC, aG, aH, and aJ
and also involves Tyr10 from helix aA as well as Lys57 from the
loop connecting helices aB and aC. The naphthyl group binds
to the hydrophobic base of the pocket, making close contacts
to Tyr10, Phe206, Phe239, Leu344, Thr63, and to the alkyl por-
tion of the side chains of Lys347, Arg60, and Asn59 (Figure 1B).
The side chain amide of Asn59 is in a parallel orientation with
respect to the naphthyl group and contributes amide–p stack-
ing interactions. The salicylic acid ring forms a face-to-edge ar-
omatic–aromatic interaction with Phe239, while one carboxylic
With the same goal in mind, we initiated a drug discovery
project aimed at novel FPPS inhibitors based on an integrated
lead finding approach combining fragment-based screening
(FBS), high-throughput screening (HTS), in silico screening,
SAR-by-inventory, biophysical validation of hits, and hit-to-lead
chemistry. The fragment-based screen rapidly identified the
first allosteric non-bisphosphonate inhibitors of FPPS, and
quickly led to the benzoindole derivative
2 (Scheme 1),
a novel, potent (IC50 =80 nm) FPPS inhibitor devoid of avidity
toward bone mineral.[18] This part of our work has already been
described in detail elsewhere.[18] Building on these results,
other research groups have recently reported further examples
of non-bisphosphonate allosteric FPPS inhibitors.[19–21]
ChemMedChem 2015, 10, 1884 – 1891
www.chemmedchem.org
1885
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Table 1. Inhibition of human FPPS by salicylic acid derivatives.
Compd
R
IC50 [mm][a]
3a
H
6.8
Figure 1. Crystal structure of the binary complex of FPPS with the allosteric
inhibitor 3a. A) Overall view of one FPPS subunit showing the location of
the allosteric site (green stick) in relation to the DMAPP/ZOL (red) and IPP
(yellow) binding sites. Note the different conformational states adopted by
the enzyme in the binary complex with 3a (open state, blue ribbon) and in
the ternary complex with ZOL and IPP[12] (closed state, gray ribbon).
B) Close-up view of the complex with 3a (green stick) with protein side
chains within 4.0 Š distance shown in blue stick.
3c
3d
3e
3 f
20
0.85
0.25
0.058
oxygen atom and the ether oxygen engage in short hydrogen
bonding contacts with the side chain amide of Asn59. In addi-
tion, the carboxylic group makes strong electrostatic interac-
tions with Arg60 (Figure 1B). The side chain amino group of
Lys57 forms a cation–p interaction with the aromatic ring of
the salicylic acid group.
3g
0.038
To guide the optimization of 3a, a virtual library of 17000
salicylic acid analogues was created, and high-throughput
docking simulations into FPPS were performed using the crys-
tal structure of the complex with 3a as a template. The virtual
library allowed substitutions at the salicylic acid and naphthyl
rings, and for a and b linkage of the naphthyl ring. Analysis of
the results of the docking simulations indicated that the a-
linked naphthyl ring fit the allosteric pocket best. Regarding
the salicylic acid ring, the position para to the carboxylic acid
seemed to offer the best opportunities for improvement.
Therefore, we decided to prepare the key intermediate 3b
with a bromine at this position that could be used as a starting
point for the parallel synthesis of derivatives by Suzuki cou-
pling or Buchwald–Hartwig amination (Scheme 2). The result-
ing derivatives were tested in an LC–MS-based biochemical
assay (see Supporting Information), and their IC50 values were
determined (Table 1). In this assay, 3a had an IC50 value of
6.8 mm. Gratifyingly, these efforts resulted in several improved
FPPS inhibitors with IC50 values in the double-digit nanomolar
range. Notably, 4-substituted phenyl groups were particularly
3h
3i
0.049
0.080
3j
3k
3l
1.23
0.19
0.27
3m
3n
3o
0.11
0.017
0.14
[a] Determined by LC–MS assay; all values are the average of at least
three independent measurements.
Scheme 2. Synthesis of the salicylic acid library. Reagents and conditions:
a) 1. boronic acid, Pd(PPh3)2Cl2, Na2CO3, DME/EtOH/H2O, MW (1108C), 10 min,
2. LiOH, MeOH/THF, MW (1208C), 12 min.
ChemMedChem 2015, 10, 1884 – 1891
www.chemmedchem.org
1886
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
efficient in enhancing binding affinity, with the 4-acetamido-
phenyl derivative 3n showing a 400-fold improvement over
the parent compound 3a. The activity of the 6-methoxy-
naphthyl derivative 3g was similar to that of 3n. The N-meth-
ylindole derivative 3l was selected for isothermal titration calo-
rimetry (ITC) analysis. Enthalpically driven binding was ob-
served (À8.03 kcalmolÀ1) with a favorable entropic contribu-
tion (+3.10 kcalmolÀ1 KÀ1) and an estimated KD value of
280 nm, well in line with the IC50 value in the LC–MS assay.
However, these salicylic acid derivatives were, at best, only
weakly active in a cellular plasma membrane translocation
assay.[24] With an IC50 value of 20 mm, compound 3i was found
to be among the most active compounds in this cellular assay.
The poor cellular permeability of this compound series, result-
ing from the carboxylic acid functionality, probably precluded
a higher cellular potency.
Discovery of the quinoline class of allosteric FPPS inhibitors
A third lead series emerged from our integrated lead finding
efforts. Compound 5, an HTS hit with an IC50 value of 7.1 mm in
the LC–MS assay, also attracted our attention because of its
fragment-like character (Mr =287 Da), its relatively good ligand
efficiency (LE: 0.32), and its structural features which suggest-
ed, in light of our previous results, that this compound might
be an allosteric inhibitor as well. A crystal structure of 5 in
complex with FPPS could be obtained, indeed revealing clear
electron density for 5 bound to the allosteric site. Remarkably,
the mode of binding of 5 was highly similar to that of 3a, as
demonstrated by a structural overlay (Figure 2A). The naphthyl
groups of the two ligands superimpose with each other almost
exactly. The indole moiety of 5 made an angle of 63.58 with re-
spect to its naphthyl group, placing the carboxylic acid in the
same location as its counterpart in compound 3a. The two car-
boxylic groups showed a distinct orientation, but one of their
oxygen atoms, the one closest to Arg60, virtually superim-
posed in the two complexes (Figure 2A). In spite of the short
connection to the naphthyl group, the indole moiety of 5, like
the salicylic acid ring of 3a, formed favorable face-to-edge aro-
matic–aromatic interactions with Phe239. Thus, compound 5
appeared as an attractive, more compact, rigid analogue of 3a.
Interestingly, 5 was not able to form the set of strong hydro-
gen bonding interactions with Asn59 that were a recurring fea-
ture of many FPPS complexes with allosteric inhibitors ob-
served thus far, including compound 3a in particular. The car-
boxylic acid group of 5 was rotated away from this residue.
Furthermore, 5 did not have any counterpart for the ether
oxygen atom of 3a. The C3 atom rather than the indole nitro-
gen was in close contact (3.2 Š) with Asn59, as a consequence
of the attachment of the naphthyl group at position 4, rather
than 7, of the indole six-membered ring. Moreover, the indole
nitrogen atom did not seem to form any direct polar contacts
with FPPS. Hence, compound 5 also showed clear potential for
affinity improvement, as its indole-carboxylate/naphthyl scaf-
fold seemed suboptimal.
Figure 2. Crystal structures of FPPS in complex with A) compound 5 (green
carbon atoms), also showing a structural overlay with compound 3a
(orange); B) compound 6, observed in two alternate conformations (green
and orange models); C) compound 8a (green). Protein side chains within
4.0 Š distance of the allosteric inhibitor are shown in blue stick. See text for
details.
folds. Replacing the naphthyl group with an indole (compound
6, Table 2) slightly weakened the activity to 13.9 mm. Instead,
retaining the naphthyl group and replacing the indole carbox-
ylate with a quinoline carboxylate with the naphthyl attached
at position 8 of the quinoline system led to a clear improve-
ment in potency (compound 8a, IC50 =1.2 mm), whereas replac-
ing the naphthyl group in 8a with an indole to form com-
pound 7 again decreased the activity to 20 mm (Table 2). The
crystal structure of the FPPS complex with 6 confirmed that an
indole moiety was not a snug fit into the base of the allosteric
pocket, and therefore not a suitable replacement for the naph-
thyl group. The observed electron density indicated the pres-
In a first step, we therefore looked for a better combination
of ring systems using the corresponding non-decorated scaf-
ChemMedChem 2015, 10, 1884 – 1891
www.chemmedchem.org
1887
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
pled under palladium catalysis to the biaryl systems 14 and 13.
Transfer hydrogenation with ammonium formiate and palladi-
um on charcoal yielded the corresponding acids 7 and 8a.
A variety of quinoline derivatives were then synthesized.
Scheme 5 illustrates the synthesis of the 6-amino-8-(naphtha-
len-1-yl)quinoline-2-carboxylic acid derivatives 8c, 8d, and 8 f.
Nitration of 8-bromoquinoline-2(1H)-one[25] in a boiling mixture
of fuming nitric and acetic acids occurs preferably at the 6-po-
sition.[26] However, in our hands the reaction was dangerously
exothermic under the described conditions. By exchanging the
acetic acid for trifluoroacetic acid this nitration could be safely
executed at or below room temperature. Suzuki coupling of
the crude nitro derivative 15 with naphthyl boronate then
gave 16. Chlorination with POCl3 in the presence of N,N-dime-
thylaniline and Et4NCl[27] yielded 17. Palladium-catalyzed car-
bonylation of 17 was accompanied by reduction of the nitro
group, yielding a mixture of products from which the 6-amino-
2-ethoxycarbonyl derivative 18 was isolated in moderate yield.
Additionally, 19 and 20 were isolated as side products. Saponi-
fication finally gave the corresponding acids 8c, 8d, and 8 f.
The 6-aryl-8-(naphthalen-1-yl)quinoline-2-carboxylic acid de-
rivatives 8b and 8e were synthesized as depicted in
Scheme 6: Sandmeyer-type conditions transformed 18 into the
iodide 21, which was cross-coupled with aryl-3-boronic acids
to give 22 or 23. Saponification or acidic hydrolysis and cleav-
age of the Boc protecting group finally lead to 8b or 8e, re-
spectively.
Table 2. Biaryl allosteric inhibitors of human FPPS.
Compd
IC50 [mm][a]
5
6
7.1
13.9
7
20.0
8a
1.2
[a] Determined by LC–MS assay; all values are the average of at least
three independent measurements.
SAR of quinoline derivatives
ence of two alternate binding modes, corresponding to oppo-
site orientations of the buried indole group (Figure 2B). In one
orientation, the indole nitrogen atom was within hydrogen
bonding distance of Thr63 Og, while in the other orientation it
appeared to interact with Ser205 O. In contrast, the crystal
structure of the FPPS complex with the quinoline derivative 8a
confirmed that an 8-substituted quinoline 2-carboxylic acid
group was a better match to the binding site than the original
4-substituted indole 2-carboxylic acid of 5 (Figure 2C). In par-
ticular, the hydrogen bonding interactions involving Asn59
were largely restored in this complex. The quinoline
nitrogen was engaged in a short (2.6 Š) hydrogen
bonding contact to Asn59 Nd2, and one carboxylic
acid oxygen atom appeared to form a bifurcated hy-
drogen bond with the main chain nitrogen (2.9 Š) as
well as with the side chain amide (3.1 Š) of Asn59.
The analysis of this X-ray structure also suggested
that suitable substituents at positions 5 or 6 of the
quinoline group may increase binding affinity.
The inhibitory potencies of all quinoline derivatives were evalu-
ated by LC–MS biochemical assay (Table 3). As with the salicylic
acid series, low double-digit nanomolar activity against human
FPPS was achieved with the most potent compounds, 8c–f. By
comparison, zoledronic acid showed an IC50 value of 150 nm
under the same experimental conditions (without pre-incuba-
tion). X-ray analysis of the FPPS complex with 8e confirmed
a conserved mode of binding with respect to the parent com-
pound 8a, with additional contacts involving the pyrrole sub-
Synthesis of biaryl and quinoline derivatives
The synthesis of compounds 5 and 6 is shown in
Scheme 3, and the synthesis of 7 and 8a is shown in
Scheme 4. Esterification of 8-hydroxyquinoline-2-car-
boxylic acid under Mitsunobu conditions gave the
Scheme 3. Synthesis of 5 and 6. Reagents and conditions: a) Na2CO3, [(C6H5)3P]4Pd, EtOH,
3 h, reflux 80%; b) Na2CO3, [(C6H5)3P]4Pd, dioxane, 12 h, 908C, 44%; c) NaOH, MeOH, 2 h,
benzyl ester 10. The 8-hydroxy functionality was con-
verted into the boron pinacolate 12, which was cou- 608C 48%.
ChemMedChem 2015, 10, 1884 – 1891
www.chemmedchem.org
1888
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
stituent. However, also with this
series, cellular activities were
found to be in the double-digit
micromolar range, presumably
due to poor permeability. At-
tempts to replace the carboxylic
acid by a tetrazole met with lim-
ited success in the cellular assay.
Conclusions
FPPS is a highly validated drug
target with an armamentarium
of safe and highly efficient ap-
proved drugs. However, all the
currently available anti-FPPS
drugs belong to the same class,
the so-called nitrogen-containing
bisphosphonates (N-BPs). The
unique pharmacokinetic proper-
ties of this class of drugs, domi-
nated by its avidity to bone min-
eral, has largely restricted its use
to bone related indications, in-
cluding metastatic bone diseas-
es. Nevertheless, evidence has
been accumulating over the past
decade that FPPS inhibitors have
Scheme 4. Synthesis of 7 and 8a. Reagents and conditions: a) benzyl alcohol, P(C6H5)3, diethyl azodicarboxylate,
THF, 08C (88%); b) (F3CSO2)2O, pyridine, CH2Cl2/dioxane, À788C!RT (~quant.); c) bis(pinacolato)diboron, KOAc,
molecular sieves (4 Š), [1,1’-bis(diphenylphosphino)ferrocene]palladium(II) chloride, DMF, 808C; d) 4-bromoindole,
Cs2CO3, [(C6H5)3P]4Pd, DMF, 808C (50%, two steps); e) ammonium formiate, Pd/C, MeOH, 658C (83%); f) 1-bromo-
naphthalene, K2CO3, [(C6H5)3P]4Pd, toluene, 908C (56%, two steps); g) ammonium formiate, Pd/C, MeOH, 658C
(89%).
a
much broader potential as
cancer
agents.[28–32]
chemotherapeutic
To fully realize this potential,
however, novel non-bisphospho-
nate FPPS inhibitors are needed.
This objective seemed elusive
for a long time, as the first X-ray
analyses had confirmed that N-
BPs were substrate mimetics and
had indicated that the design of
non-bisphosphonate substrate
analogues as a rational approach
toward new FPPS inhibitors was
doomed. The recent discovery of
allosteric FPPS inhibitors, howev-
er, has revolutionized the field
and opened new avenues
toward the development of
novel classes of anti-FPPS drugs.
This article complements the
available chemical matter for al-
losteric inhibition of FPPS by
two new lead series. These new
classes were discovered by an
integrated lead finding ap-
proach, which was essential for
the rapid identification and char-
acterization of promising, but
Scheme 5. Synthesis of 8c, 8d, and 8 f. Reagents and conditions: a) fuming HNO3, TFA, 08C!RT; b) K2CO3,
[(C6H5)3P]2PdCl2, DMF/H2O, 80–928C (57%, two steps); c) POCl3, Et4NCl, N,N-dimethylaniline, CH3CN, ›fl (78%);
d) CO, [(C6H5)3P]2PdCl2, Et3N, EtOH, 110 8C; e) LiOH, H2O/dioxane (31–63%).
ChemMedChem 2015, 10, 1884 – 1891
www.chemmedchem.org
1889
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
comprise compounds that are
10- to 20-fold more potent than
the initial lead series. These com-
pounds are exemplified by 3 f
and 3n for the salicylic acid
series, and by 8d and 8 f for the
quinoline series.
Notably, the in vitro potencies
of these allosteric inhibitors are
in the same range or even
higher than those of the most
potent N-BPs, even if the latter
are pre-incubated with FPPS
before the start of the reaction,
which increases their activities
by a factor of 10–100.[33] More-
over, both classes of FPPS inhibi-
tors lack the bisphosphonate
functionality and are thus radi-
cally distinct from the N-BPs. The
bisphosphonate moiety is the
“bone hook” that targets these
drugs to bone, thus compromis-
Scheme 6. Synthesis of 8b and 8e. Reagents and conditions: a) 1. NaNO2, HCl, H2O, À158C, 2. KI, 08C!RT (28%);
b) thiophen-3-boronic acid; Na2CO3, [(C6H5)3P]2PdCl2, DMF, 1008C (81%); c) LiOH, H2O/dioxane (83%); d) [1-(tert-bu-
toxycarbonyl)-1H-pyrrol-2-yl]boronic acid, Na2CO3, [(C6H5)3P]2PdCl2, DMF, 1008C (69%); e) 2m HCl in THF/H2O (1:1),
508C (38%).
ing their broader utility for non-skeletal indications. Allosteric
inhibitors of FPPS, in contrast, are not substrate mimetics and
therefore do not require a bisphosphonate group for binding
affinity. With this major limitation of the N-BP class circumvent-
ed, and also in view of their relatively small size, there is still
ample room for further chemical elaboration of these allosteric
inhibitors, and, in particular, for improving their permeability
and pharmacokinetic properties. So far, the cellular activity of
our allosteric inhibitors in the membrane translocation assay
was disappointing, with the best IC50 values in the double-digit
micromolar range. Nevertheless, this is a first step forward in
comparison with N-BP drugs, such as zoledronic acid.
Table 3. Structures and FPPS inhibitory activities of analogues 8.
Compd
R1
H
R2
H
IC50 [mm][a]
8a
1.2
8b
H
0.10
The allosteric binding site of FPPS is relatively basic owing
to its vicinity to the IPP binding site, the nearby basic C-termi-
nal tail of the enzyme with the sequence motif 350KRRK353, and
the presence of several positively charged residues—Arg60,
Lys57, and Lys347—within the pocket itself. Not too surprising-
ly, our best allosteric inhibitors feature at least one carboxylic
acid functionality, which is absolutely critical for binding affini-
ty. While carboxylic acid groups do not necessarily preclude
cell permeability and are found in a wide variety of common
drugs, including in particular NSAIDs (aspirin, ibuprofen) and
cholesterol-lowering agents (Lipitor, Lescol, Crestor), further
medicinal chemistry efforts are clearly required to develop one
of these three series of allosteric FPPS inhibitors into a useful
drug. A number of potential bioisosteres of carboxylic acid
have been described,[34] and prodrug approaches for carboxylic
acids are also well documented.[35] Pursuing all available op-
tions appears to be a very worthwhile endeavor. An allosteric
FPPS drug should prove useful for a broad range of important
indications with significant medical unmet needs. In oncology,
the direct antitumor effects, synergistic effects with existing
chemotherapies, as well as indirect immunostimulatory antitu-
mor effects observed with zoledronic acid and other N-BPs
8c
8d
NH2
NH2
H
OEt
0.069
0.024
8e
8 f
H
H
0.077
0.037
[a] Determined by LC–MS assay; all values are the average of at least
three independent measurements.
weak hits from a high-throughput screen. In particular, the
knowledge we had gained from our initial fragment-based
work greatly helped us focus on a few compounds that exhib-
ited the expected pharmacophore for putative new allosteric
site binders.
One of the best inhibitors of our previous benzoindole lead
series, compound 2, had an IC50 value of 200 nm in the scintil-
lation proximity assay.[18] In our new LC–MS-based assay, which
was used throughout the current study, compound 2 showed
an IC50 value of 350 nm. The two new classes reported herein
ChemMedChem 2015, 10, 1884 – 1891
www.chemmedchem.org
1890
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[10] K. Thompson, M. J. Rogers, J. Bone Miner. Res. 2004, 19, 278–288.
[11] R. E. Hewitt, A. Lissina, A. E. Green, E. S. Slay, D. A. Price, A. K. Sewell,
Clin. Exp. Immunol. 2005, 139, 101–111.
[12] J.-M. Rondeau, F. Bitsch, E. Bourgier, M. Geiser, R. Hemmig, M. Kroemer,
S. Lehmann, P. Ramage, S. Rieffel, A. Strauss, J. R. Green, W. Jahnke,
ChemMedChem 2006, 1, 267–273.
[13] K. L. Kavanagh, K. Guo, J. E. Dunford, X. Wu, S. Knapp, F. H. Ebetino,
M. J. Rogers, R. G. Russell, U. Oppermann, Proc. Natl. Acad. Sci. USA
2006, 103, 7829–7834.
[14] M. S. Marma, Z. Xia, C. Stewart, F. Coxon, J. E. Dunford, R. Baron, B. A.
Kashemirov, F. H. Ebetino, J. T. Triffitt, R. G. G. Russell, C. E. McKenna, J.
Med. Chem. 2007, 50, 5967–5975.
[15] Y. Zhang, R. Cao, F. Yin, M. P. Hudock, R.-T. Guo, K. Krysiak, S. Mukherjee,
Y.-G. Gao, H. Robinson, Y. Song, J. H. No, K. Bergan, A. Leon, L. Cass, A.
Goddard, T.-K. Chang, F.-Y. Lin, E. Van Beek, S. Papapoulos, A. H.-J. Wang,
T. Kubo, M. Ochi, D. Mukkamala, E. Oldfield, J. Am. Chem. Soc. 2009,
131, 5153–5162.
[16] Y. Zhang, R. Cao, F. Yin, F.-Y. Lin, H. Wang, K. Krysiak, J.-H. No, D. Mukka-
mala, K. Houlihan, J. Li, C. T. Morita, E. Oldfield, Angew. Chem. Int. Ed.
2010, 49, 1136–1138; Angew. Chem. 2010, 122, 1154–1156.
[17] Y.-S. Lin, J. Park, J. W. De Schutter, X. F. Huang, A. M. Berghuis, M. Sebag,
Y. S. Tsantrizos, J. Med. Chem. 2012, 55, 3201–3215.
hold great promise for a next-generation allosteric FPPS drug
with a superior pharmacokinetic profile. In addition, such
a drug may prove useful as a novel cholesterol-lowering agent
and whenever excessive lipid production causes disease, either
by excessive prenylation of target proteins or overproduction
of downstream products by the mevalonate pathway. As one
example, the involvement of this pathway in human prema-
ture aging has raised the possibility of developing FPPS inhibi-
tors for the treatment of devastating human progeria syn-
dromes. Last but not least, the allosteric FPPS inhibitors report-
ed herein may also serve as lead molecules for the develop-
ment of new anti-parasitic agents, notably for the treatment of
Chagas disease, malaria, and leishmaniasis, which affect several
million people worldwide. For all of these indications, bi-
sphosphonates have shown activity as tool compounds, again
demonstrating the value of FPPS as a therapeutic target, but
they are hardly suitable as drugs. While it is still unclear if the
allosteric pocket of FPPS has any natural ligand or function, ex-
ploiting the Achilles’ heel of this important therapeutic target
will remain a fascinating scientific and medical endeavor for
years to come.
[18] W. Jahnke, J.-M. Rondeau, S. Cotesta, A. Marzinzik, X. PellØ, M. Geiser, A.
Strauss, M. Gçtte, F. Bitsch, R. Hemmig, C. Henry, S. Lehmann, J. F. Glick-
man, T. P. Roddy, S. J. Stout, J. R. Green, Nat. Chem. Biol. 2010, 6, 660–
666.
[19] a) S. Lindert, W. Zhu, Y.-L. Liu, R. Pang, E. Oldfield, J. A. McCammon,
Chem. Biol. Drug Des. 2013, 81, 742–748; b) Y.-L. Liu, R. Cao, Y. Wang, E.
Oldfield, ACS Med. Chem. Lett. 2015, 6, 349–354.
Acknowledgements
[20] a) J. W. Schmidberger, R. Schnell, G. Schneider, Acta Crystallogr. Sect. D
2015, 71, 721; b) J. Liu, W. Liu, H. Ge, J. Gao, Q. He, L. Su, J. Xu, L.-q. Gu,
Z.-s. Huang, D. Li, Biochim. Biophys. Acta 2014, 1840, 1051–1062.
[21] J. W. De Schutter, J. Park, C. Y. Leung, P. Gormley, Y-S Lin, Z. Hu, A. M.
Berghuis, J. Poirier, Y. S. Tsantrizos, J. Med. Chem. 2014, 57, 5764–5776.
[22] W. Jahnke, G. Bold, A. Marzinzik, S. Ofner, X. PellØ, S. Cotesta, E. Bourgi-
er, S. Lehmann, C. Henry, R. Hemmig, F. Stauffer, J. C. D. Hartwieg, J. R.
Green, J.-M. Rondeau, Angew. Chem. Int. Ed. 2015, DOI: 10.1002/
anie.201507064; Angew. Chem. 2015, DOI: 10.1002/ange.201507064.
[23] J. F. Glickman, A. Schmid, Assay Drug Dev. Technol. 2007, 5, 205–214.
[24] M. Simonen, Y. Ibig-Rehm, G. Hofmann, J. Zimmermann, G. Albrecht, M.
Magnier, V. Heidinger, D. Gabriel, J. Biomol. Screening 2008, 13, 456–
467.
The authors are grateful to Peggy Lefeuvre for performing the ITC
analysis of compound 3l, to Dr. Kazuhide Konishi for the synthe-
sis of intermediate 3b, to Christian Zwingelstein for the synthesis
of compound 6, and to Werner Beck, Ursula Dürler, Mario Madçr-
in, and Christian Ragot for the synthesis of compounds 7 and
8a–f. We thank Martin Geiser, Sebastien Rieffel, AndrØ Strauss,
and RenØ Hemmig for the cloning, expression, and purification of
human FPPS. The coordinates of the crystal structures presented
herein have been deposited with the RCSB Protein Data Bank
(www.rcsb.org, PDB codes 5DIQ (FPPS+3a), 5DJP (FPPS+5),
5DJR (FPPS+6), 5DGN (FPPS+8a), 5DJV (FPPS+8e).
[25] F. Cottet, M. Marull, O. Lefebvre, M. Schlosser, Eur. J. Org. Chem. 2003,
1559–1568.
[26] G. Leclerc, G. Marciniak, N. Decker, J. Schwartz, J. Med. Chem. 1986, 29,
2433–2438.
[27] H. Harada, O. Asano, T. Kawata, T. Inoue, T. Horizoe, N. Yasuda, K.
Nagata, M. Murakami, J. Nagaoka, S. Kobayashi, I. Tanaka, S. Abe, Bioorg.
Med. Chem. 2001, 9, 2709–2726.
[28] J. R. Berenson, Curr. Opin. Support Palliat. Care 2011, 5, 233–240.
[29] G. Morgan, A. Lipton, Semin. Oncol. 2010, 37, S30–40.
[30] H. K. Koul, S. Koul, R. B. Meacham, Prostate Cancer Prostatic Dis. 2012,
15, 111 –119.
[31] D. Santini, V. Virzi, M. E. Fratto, F. Bertoldo, R. Sabbatini, R. Berardi, N.
Calipari, D. Ottaviani, T. Ibrahim, Curr. Cancer Drug Targets 2010, 10, 46–
54.
Keywords: allosteric inhibitors · antitumor agents · anti-
infective drug discovery
immunotherapy
· FPPS · immunomodulation ·
[1] A. S. Kesselheim, J. Avorn, Nat. Rev. Drug Discovery 2013, 12, 425–443.
[2] R. G. G. Russell, N. B. Watts, F. H. Ebetino, M. J. Rogers, Osteoporosis Int.
2008, 19, 733–759.
[3] M. J. Rogers, J. C. Crockett, F. P. Coxon, J. Monkkonen, Bone 2011, 49,
34–41.
[4] F. P. Coxon, K. Thompson, M. J. Rogers, Curr. Opin. Pharmacol. 2006, 6,
307–312.
[32] M. Gnant, P. ClØzardin, Cancer Treat. Rev. 2012, 38, 407–415.
[33] J. E. Dunford, A. A. Kwaasi, M. J. Rogers, B. L. Barnett, F. H. Ebetino,
R. G. G. Russell, U. Oppermann, K. L. Kavanagh, J. Med. Chem. 2008, 51,
2187–2195.
[34] C. Ballatore, D. M. Huryn, A. B. Smith III, ChemMedChem 2013, 8, 385–
395.
[35] H. Maag in Prodrugs: Challenges and Rewards (Eds.: V. J. Stella, R. T.
Borchardt, M. J. Hageman, R. Oliyai, H. Maag, J. Tilley), Springer, New
York, 2007, pp. 3–30.
[5] I. Varela, S. Pereira, A. P. Ugalde, C. L. Navarro, M. F. Suµrez, P. Cau, J. Ca-
diÇanos, F. G. Osorio, N. Foray, J. Cobo, F. de Carlos, N. LØvy, J. M. P.
Freije, C. López-Otín, Nat. Med. 2008, 14, 767–772.
[6] X. Wang, E. R. Hinson, P. Cresswell, Cell Host Microbe 2007, 2, 96–105.
[7] R. Docampo, S. N. Moreno, Curr. Drug Targets Infect. Disord. 2001, 1, 51–
61.
[8] M. Gnant, B. Mlineritsch, H. Stoeger, G. Luschin-Ebengreuth, D. Heck, C.
Menzel, R. Jakesz, M. Seifert, M. Hubalek, G. Pristauz, T. Bauernhofer, H.
Eidtmann, W. Eiermann, G. Steger, W. Kwasny, P. Dubsky, G. Hochreiner,
E. P. Forsthuber, C. Fesl, R. Greil, et al., N. Engl. J. Med. 2009, 360, 679–
691.
Received: July 30, 2015
[9] G. J. Morgan, Lancet 2010, 376, 1989–1999.
Published online on September 18, 2015
ChemMedChem 2015, 10, 1884 – 1891
www.chemmedchem.org
1891
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim