Pharmacophore Mapping of Thienopyrimidine-Based Monophosphonate (ThP-MP) Inhibitors of the Human Farnesyl Pyrophosphate Synthase

Article abstract of DOI:10.1021/acs.jmedchem.6b01888

The human farnesyl pyrophosphate synthase (hFPPS), a key regulatory enzyme in the mevalonate pathway, catalyzes the biosynthesis of the C-15 isoprenoid farnesyl pyrophosphate (FPP). FPP plays a crucial role in the post-translational prenylation of small GTPases that perform a plethora of cellular functions. Although hFPPS is a well-established therapeutic target for lytic bone diseases, the currently available bisphosphonate drugs exhibit poor cellular uptake and distribution into nonskeletal tissues. Recent drug discovery efforts have focused primarily on allosteric inhibition of hFPPS and the discovery of non-bisphosphonate drugs for potentially treating nonskeletal diseases. Hit-to-lead optimization of a new series of thienopyrimidine-based monosphosphonates (ThP-MPs) led to the identification of analogs with nanomolar potency in inhibiting hFPPS. Their interactions with the allosteric pocket of the enzyme were characterized by crystallography, and the results provide further insight into the pharmacophore requirements for allosteric inhibition.

Full text of DOI:10.1021/acs.jmedchem.6b01888

Subscriber access provided by University of Newcastle, Australia
Article
Pharmacophore Mapping of Thienopyrimidine-Based Monophosphonate
(ThP-MP) Inhibitors of the Human Farnesyl Pyrophosphate Synthase
Jaeok Park, Chun Yuen Leung, Alexios N. Matralis, Cyrus M. Lacbay, Michail Tsakos,
Guillermo Fernandez De Troconiz, Albert M. Berghuis, and Youla S. Tsantrizos
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.6b01888 • Publication Date (Web): 16 Feb 2017
Downloaded from http://pubs.acs.org on February 18, 2017
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 64
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Pharmacophore mapping of thienopyrimidine-based monophosphonate (ThP-MP)
inhibitors of the human farnesyl pyrophosphate synthase
1
1
12
13
†
#
‡#
‡
‡
‡
Jaeok Park, Chun Yuen Leung, Alexios N. Matralis, Cyrus M. Lacbay, Michail Tsakos,
14
15
16
Guillermo Fernandez De Troconiz, Albert M. Berghuis,†‡ Youla S. Tsantrizos*‡†‡
‡
17
18
19
2
2
2
23
‡Department of Chemistry, McGill University, 801 Sherbrooke Street West, Montreal, QC,
Canada H3A 0B8
24
25
26
27
†Department of Biochemistry, McGill University, 3649 Promenade Sir William Osler, Montreal,
QC, Canada H3G 0B1
28
29
30
31
3
‡Groupe de Recherche Axé sur la Structure des Protéines, McGill University, 3649 Promenade Sir
William Osler, Montreal, QC, Canada
3
34
35
36
37
38
39
40
41
42
4
4
45
46
47
48
49
50
#
J. Park and C. Y. Leung contributed equally to the work described in this manuscript
51
52
53
5
5
56
57
58
59
60
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 64
1
2
3
4
5
6
7
8
9
ABSTRACT
The human farnesyl pyrophosphate synthase (hFPPS), a key regulatory enzyme in the
mevalonate pathway, catalyzes the biosynthesis of the C-15 isoprenoid farnesyl pyrophosphate
1
1
(FPP). FPP plays a crucial role in the post-translational prenylation of small GTPases that
12
13
14
15
16
17
18
19
20
2
2
23
24
25
26
27
perform a plethora of cellular functions. Although hFPPS is a well-established therapeutic target
for lytic bone diseases, the currently available bisphosphonate drugs exhibit poor cellular uptake
and distribution into non-skeletal tissues. Recent drug discovery efforts have focused primarily
on allosteric inhibition of hFPPS and the discovery of non-bisphosphonate drugs for potentially
treating non-skeletal diseases. Hit-to-lead optimization of a new series of thienopyrimidine-
based monosphosphonates (ThP-MPs) led to the identification of analogs with nanomolar
potency in inhibiting hFPPS. Their interactions with the allosteric pocket of the enzyme were
characterized by crystallography and the results provide further insight into the pharmacophore
requirements for allosteric inhibition.
28
29
30
31
3
3
34
35
36
37
38
39
40
41
42
4
4
45
46
47
48
49
50
51
52
53
5
5
56
57
58
59
60
2
ACS Paragon Plus Environment

Page 3 of 64
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
INTRODUCTION
Human farnesyl pyrophosphate synthase (hFPPS) is the gate-keeper for the biosynthesis
of all mammalian isoprenoids (Fig. 1). The catalytic product of hFPPS, farnesyl pyrophosphate
1
1
12
13
(FPP), is the precursor of many human metabolites, including geranylgeranyl pyrophosphate
14
15
16
(GGPP). FPP and GGPP are essential for the post-translational prenylation of all small GTPase
1
18
19
20
1
proteins (Fig. 1), which play a crucial role in a plethora of cellular biological functions,
including cell signaling, proliferation, and synaptic plasticity.2
2
2
23
3
Although hFPPS has been a therapeutic target of interest for over 30 years, the highly
24
25
26
27
28
29
30
charged nature of its active site cavity has prevented the development of active site inhibitors
with good drug-like molecules (in the classical sense). Currently, nitrogen-containing
bisphosphonates (N-BPs) are the only clinically validated drugs that target hFPPS (e.g. 1, 2). N-
31
3
4
BPs bind avidly to bone, inhibiting osteoclasts, but suffer from poor oral bioavailability, rapid
3
3
35
clearance from the systemic circulation, and almost negligible distribution to non-skeletal
3
37
5
tissues. Consequently, these drugs do not allow clear validation of hFPPS as a therapeutic target
38
39
40
41
for treating non-skeletal diseases, such as cancer. Nonetheless, biochemical studies have shown
that inhibition of hFPPS can directly downregulate the activity of mutated H-Ras, K-Ras and N-
42
4
4
6
Ras proteins that are major drivers in oncogenesis. We have also previously shown that active
45
4
site inhibitors of hFPPS induce apoptosis and inhibit phosphorylation of ERK in multiple
47
48
49
7
myeloma cells (ERK is a signaling protein downstream from Ras). Additionally, clinical
50
51
52
53
evidence suggests that inhibition of hFPPS with the most potent of the N-BP drug, zoledronic
acid (1), improves survival of cancer patients via mechanism(s) unrelated to the skeletal effects
5
5
56
8
of this drug. For example, a randomized clinical trial (with >1,500 patients) has shown that
57
58
treatment with standard care chemotherapy plus zoledronic acid (1) increases the progression-
59
60
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 64
1
2
3
4
5
6
7
8
9
free survival of multiple myeloma patients by a statistically significant period as compared to
8
clodronate. Although clodronate is also a bisphosphonate compound, it is not a potent inhibitor
of hFFPS.
1
1
Inhibition of hFPPS impacts both the down-stream and the up-stream events in the
12
1
14
1
9
mevalonate pathway, inducing cell apoptosis and activation of the human innate immune
10
response. Inhibition of hFPPS also blocks the biosynthesis of GGPP, an indirect consequence
16
17
18
19
20
2
2
23
24
25
26
27
28
29
30
31
3
3
34
35
36
37
38
39
40
of intracellular depletion of the FPP substrate required by the human geranylgeranyl
pyrophosphate synthase (hGGPPS;Fig. 1). Not surprisingly, treatment with zoledronic acid (1)
was found to cause intracellular decrease in Rap1A geranylgeranylation and activation,
providing an additional (or alternative) mechanism that could account for the in vivo
antimyeloma effects observed with this drug.8,11 Intracellular depletion of GGPP is also expected
to have a significant effect on the prenylation cascade from GGPP → RhoA-cdc42, which leads
to activation of GSK3-β and upregulation of phosphorylated tau (P-Tau) protein in the brain. P-
Tau accumulation is associated with the physiology of tauopathies, such as frontotemporal
dementia, neuronal damage and the progression of the Alzheimer’s-associated tangle formation
in neurons.12,17
41
42
4
4
45
46
47
48
49
50
51
52
53
5
5
56
57
58
59
60
4
ACS Paragon Plus Environment

Page 5 of 64
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
12
13
14
15
16
17
18
19
20
2
2
23
24
25
26
27
28
29
30
31
3
3
34
35
36
37
38
Figure 1: Biosynthesis of FPP and GGPP and their role in protein prenylation; key enzymes are
highlighted in blue
39
40
41
4
4
4
45
46
47
48
49
The presumed potential value of hFPPS as a therapeutic target for non-skeletal diseases has
fueled efforts towards the identification of non-bisphosphonates inhibitors of this target. Jahnke and
co-workers at Novartis were the first to employ fragment-based screening by NMR and X-ray
crystallography to identify a catalytically relevant allosteric pocket, near the IPP substrate binding
50
5
52
13
site of the hFPPS active site. Recently, we explored the biological role of this allosteric pocket
53
5
5
56
and showed that FPP product can bind to this allosteric pocket, locking the enzyme in an inactive
conformation and consequently, providing a feed-back mechanism for controlling the intracellular
57
58
59
60
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 64
1
2
3
4
5
6
7
8
9
levels of isoprenoid biosynthesis in vivo.14 To date, numerous structurally diverse non-
bisphosphonate inhibitors have been reported that bind to the hFPPS allosteric pocket with high
affinity; these include analogs 3, 4 and 5.13,15 In vitro potencies in the low nanomolar range were
observed with some of these compounds (e.g. 4) using an hFPPS inhibition assay based on
1
1
12
13
14
LC/MS/MS.13,15 Efforts by other research groups have also contributed to this field, although
16
1
16
some of the compounds reported do not actually bind inside the hFPPS allosteric pocket.16a
17
18
19
20
2
2
23
24
25
26
27
28
In the course of our own investigations, we reported the discovery of thienopyrimidine-based
bisphosphonates (ThP-BPs; e.g. compound 6) that exhibit a promiscuous binding mode and interact
with both the active site and the allosteric pocket of hFPPS under physiologically relevant
conditions.17 Structural re-modeling of this class of compounds allowed us to probe the
pharmacophore elements that discriminate between binding in the allosteric pocket of hFPPS as
2
30
18
opposed to the allylic sub-pocket of its active site cavity. Truncation of one phosphonate moiety
31
3
3
34
35
36
37
38
39
40
41
42
4
4
45
46
47
48
49
50
51
52
53
5
led to the identification of a new chemotype of hFPPS inhibitors (e.g. compound 9) that binds
exclusively in the allosteric pocket and engage in similar interactions with the protein as those
observed with inhibitors 3-5.13,15 In this report, we describe our efforts towards the optimization of
our initial hit and the development of a structure-activity relationship (SAR) model that could guide
further the optimization of thienopyrimidine-based inhibitors. Herein, we describe the identification
of analogs with nanomolar potency and characterization of the protein-inhibitor interactions of
structurally diverse analogs with hFPPS. These studies provide further insight into the essential
minimal pharmacophore that is required for allosteric inhibition of hFPPS and point towards new
directions in the design of non-phosphonate thienopyrimidine-based allosteric inhibitors targeting
this enzyme.
5
56
57
58
59
60
6
ACS Paragon Plus Environment

Page 7 of 64
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
12
13
14
15
16
17
18
19
20
2
2
23
24
25
26
27
28
29
CHEMISTRY
30
31
3
3
34
35
36
Potency optimization and structure-activity relationship studies of the thienopyrimidine
monophosphonate inhibitors of hFPPS were initiated with the synthesis of compound libraries
focusing mainly on (a) the linker connecting the lipophilic thienopyrimidine core to the charged
37
3
39
40
41
42
4
phosphate moiety and (b) the substitutions at R
from these libraries are shown in Figures 2 and 3 (analogs 7 to 48). It should be noted that although
all analogs with a Cα substituent at R were initially synthesized (as part of the parallel library
4
and R
6
of the scaffold. Representative examples
4
4
4
46
47
48
49
50
51
52
synthesis) in racemic form, the role of the Cα substitution was also evaluated with enantiomerically
enriched compounds, such as the (R)- and (S)-13, which were synthesized in 98% and 92% ee,
respectively. In addition to the compound libraries, and in order to better understand the binding
contributions of the presumed essential negatively charged moiety,13,15,16b,17 we also synthesized
analogs 49 and 50 as individual compounds. For brevity, the experimental procedures for the
53
5
5
56
57
58
59
60
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 64
1
2
3
4
5
6
7
8
9
preparation of (R)- and (S)-13 analogs and compounds 49 and 50 are summarized in the Supporting
Information.
The synthesis of ThP-BP analogs was initiated from thieno[2,3-d]pyrimidin-4(3H)-ones
1
1
12
13
14
15
16
17
18
19
20
2
2
23
24
25
26
51a or 51b), prepared as previously reported (Scheme 1).17,19,20,21,22 In general, the R
were installed via Pd-catalyzed Suzuki cross-coupling reactions with the C-6 bromide, as
previously reported.17,19,20 Installation of the R
moiety (in library synthesis mode) turned out
to be more challenging. We attempted to directly couple the thienopyrimidinones 51 with
various α-aminophosphonate esters using the phosphonium-mediated S Ar protocol reported by
Wan and co-workers;23 however, in our hands this approach was unsuccessful using our
thienopyrimidine scaffold. In contrast, S Ar displacement of the C-4 chloride of intermediates
(
6
moieties
4
N
N
27
2
29
30
31
3
3
34
35
54 or 55 was possible, but the conversion was often slow, requiring high temperatures and
leading to modest or poor yields, particularly with sterically hindered amines having an electron
withdrawing groups at Cα. We also contemplated the conversion of the starting
thienopyrimidinones (51 or 53) directly to the corresponding 4-fluoropyrimidine, in order to
36
37
38
improve the outcome of the subsequent S Ar reaction. In the past, Ritter and co-workers reported
N
39
40
41
42
the conversion of quinazolin-4(3H)-one to the 4-fluoroquinazoline using 1,3-bis(2,6-
diisopropylphenyl)-2,2-difluoro-2,3-dihydro-1H-imidazole as the fluorinating agent in 34%
4
4
45
46
47
48
49
50
2
4
yield. However, in addition to the modest yields, the cost of this reagent ($375.00 CAD/10 mL
of 0.1M solution), discouraged us from using this methodology. The alternative approach of
preparing intermediate 56 via the triflate (instead of the chloride 55) and then fluorinating under
51
52
53
5
5
56
57
2
5
very mild reaction conditions, as reported by Buchwald and co-workers was also considered.
Ultimately, given the high thermal and chemical stability of the thienopyrimidine scaffold, we
chose to prepare the fluoride intermediate 56, by treating 55 with KF in DMSO, as reported by
58
59
60
8
ACS Paragon Plus Environment

Page 9 of 64
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
26
Shestopalov, Gakh and co-workers; this approach proved to be the most efficient and atom-
economical path to the fluoride 56 and in most cases, provided excellent yields (≥85%). The
subsequent S Ar displacement of the C-4 fluoride with various α-aminophosphonate esters in
N
1
1
DMSO under basic conditions gave the diethyl esters of compounds 7-48 in moderate to
12
1
14
excellent isolated yields (45-90%). Finally, the diethyl monosphosphonate esters of compounds
15
16
7
-48 were first treated with TMSBr and the trimethylsilyl esters generated were then cleaved
17
18
19
20
with methanol to give the final allosteric inhibitors, thienopyrimidine-based phosphonic acids 7-
48 (Fig. 2 and 3).
2
2
23
24
25
26
27
28
29
30
31
3
3
34
35
36
37
38
39
40
41
42
4
4
45
46
Scheme 1: Parallel synthesis of a permutation library of hFPPS allosteric inhibitors; fragments
o
f1-f32 were used for R
1
and R
6
(Fig. 4). Reagents and conditions: (a) Br
2
, AcOH, 80 C, 1.5 h
47
48
o
(98%); (b) 5 mol% Pd(PPh
3
)
4
, aryl boronic acid or boronate ester, dioxane:H
2
O (1:1), 110 C,
4
50
51
52
o
o
1
2 h (60-80%); (c) POCl
3
, 95 C, 5 h (50%); (d) KF.2H
2
O, DMSO, 120 C, MW, 15 min (85%);
o
(
e) diethyl α-aminophosphonate ester 60, Et
3
N, DMSO, 100 C, 3 h (47-87%); (f) TMSBr then
53
5
MeOH, 50% to quantitative.
5
56
57
58
59
60
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 64
1
2
3
4
5
6
7
8
9
The synthesis of racemic diethyl α-aminophosphonate esters with general structure 60
(
Scheme 2) was previously achieved using the Kabachnick-Fields three-component condensation
of aldehydes, amines and diethyl phosphite (Scheme 2I).27,28 In our hands, this method was
1
1
reasonably successful (40-90% yield) with simple aryl aldehydes (e.g. benzaldehyde-type
12
1
14
15
16
17
18
19
2
9
reagents), particularly in the presence of catalytic amounts of zirconium(IV), but led to low yields
and numerous side products with more complex and benzylic-type of aldehydes. Given these results
and the limited availability of commercial aldehydes, the direct Cα alkylation of diethyl
(aminomethyl)phosphonate appeared more desirable and was explored (Scheme 2-II). We
20
2
2
23
30
examined in parallel the deprotonation at Cα of imine 59 and dibenzylaminomethyl phosphonate
24
25
3
1
5
8 and subsequent alkylation with an alkyl halide. The latter approach was found to be generally
26
27
28
more reliable in providing higher yields of the desired products with many structurally diverse
2
30
31
3
3
34
35
36
32
33
halides (26-95%). The synthesis of the β-aminophosphonate building blocks 61 and 62 , leading
to compounds 14-16 (Fig. 2) was achieved as previously described (with only minor modifications)
and summarized briefly in Scheme 2-III and -IV, respectively (experimental details are provided in
the Supporting Information).
37
38
39
40
41
42
4
4
45
46
47
48
49
50
51
52
53
5
5
56
57
58
59
60
1
0
ACS Paragon Plus Environment

Page 11 of 64
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
12
13
14
15
16
17
18
19
20
2
2
23
24
25
26
27
28
29
3
31
3
3
34
35
Scheme 2: Synthesis of diethyl α-aminophosphonate esters; R selected from fragments f1-f32
1
o
(Fig. 1). Reagents and conditions: (a) ZrOCl
2
.H
2
O, neat, 80 C (40-90%); (b) H
2
, Pd/C, MeOH;
o
o
(c) THF, 50 C, 24h (63%); (d) LDA, R
1
Br, THF, -78 C to RT, 2 h (25-95%); (e) i.
o
CH
3
PO(OEt)
2
, n-BuLi, THF, -78 C to RT, ii. NaBH
4
, AcOH, RT (60%); (f) allyl bromide, n-
, dioxane:H O, RT,
, reflux (89%); (j)
o
BuLi, THF, -78 C to reflux (75%); (g) i. 4-methyl-N-morpholine oxide, OsO
4
2
36
37
38
ii. NaIO
4
, RT (92%); (h) H
2
O
2
, SeO
2
, THF, reflux (70%); (i) EtOH, H
2
SO
4
39
40
NH OH, RT (97%; (k) HCO
4
2
H, CH
3
CN:H O, RT (83%).
2
41
42
4
4
45
46
47
48
49
50
51
52
53
5
5
56
57
58
59
60
1
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 64
1
2
3
4
5
6
7
8
9
1
1
12
13
14
15
16
17
18
19
20
2
2
23
24
25
26
27
28
29
30
31
3
3
34
35
36
37
38
39
40
41
Figure 2: Permutation library of hFPPS allosteric inhibitors; the structure of all the fragments
42
4
used (i.e. f1 to f32) are shown in Fig. 4.
4
45
46
47
48
49
50
51
52
53
5
5
56
57
58
59
60
1
2
ACS Paragon Plus Environment

Page 13 of 64
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
12
13
14
15
16
17
18
19
20
2
2
23
24
2
Figure 3: Permutation library of hFPPS allosteric inhibitors; the structure of all the fragments
26
27
used (i.e. f1 to f32) are shown in Fig. 4.
28
29
30
31
3
3
34
35
36
37
38
39
40
41
42
4
4
45
46
47
48
49
50
51
52
53
5
5
56
57
Figure 4: Structures of fragments f1 to f32 used in the synthesis of the compound libraries (Fig. 2
and 3)
58
59
60
1
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 64
1
2
3
4
5
6
7
8
9
The synthesis of (2-aminoacetyl)phosphoramidic acid inhibitors, such as compounds 29 and
3
4
35
3
0 (Fig. 2) is outlined in Scheme 3. The amides of L-phenylalanine and a-phenylglycine were
synthesized according to literature procedures, and their reaction with the thienopyrimidine scaffold
was achieved via S Ar to give the primary amides 63 and 65. These intermediates were
1
1
N
12
1
14
15
16
17
18
19
20
2
2
23
24
25
subsequently treated with tetrabenzyl pyrophosphate in the presence of t-BuOLi to afford the
respective phosphoroamidates. The benzyl protecting group could be removed with TMSBr fairly
fast (2h compared to 3-4 days for removal of the ethyl groups) and its use eliminated concerns over
any potential hydrolysis of the P-N bond. As indicated earlier, analogs (R)- and (S)-13, 49 and 50,
were not synthesized as part of the libraries (Fig. 2 and 3); the synthesis of these compounds is
summarized in the Supporting Information.
26
27
28
29
30
31
3
3
34
35
36
37
38
39
4
41
42
Scheme 3: Synthesis of (2-aminoacetyl)phosphoramidic acid analogs (Fig 2). Reagents and
o
conditions: (a) NH
4
OH, EDC, HOBt, DMF/CH
2
Cl
2
, 0 C to RT, 60-70%; (b) TFA, CH
2
Cl
2
, RT,
4
4
o
quant.; (c) intermediate 55a, DIPEA, dioxane, 100 C, 75-85%; (d) [PO(OBn)
]
2 2
O, t-BuOLi, THF,
4
46
o
Ar, RT, 45-52%; (e) i. TMSBr, CH
2
Cl
2
, 0 C to RT then ii. MeOH, RT, 64-95%.
47
48
49
50
51
52
53
5
5
56
57
58
59
60
1
4
ACS Paragon Plus Environment

Page 15 of 64
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
12
13
14
15
16
17
18
19
20
RESULTS AND DISCUSSION
2
2
23
24
25
26
27
Dysregulation of the mevalonate pathway and upregulation of isoprenoid biosynthesis has
been implicated in numerous human diseases, including metabolic disorders (e.g.
hypercholesterolemia and osteoporosis), cancers that are mostly prevalent in aging populations (e.g.
28
2
3
6
37,38
multiple myeloma, breast and prostate cancers), and neurodegeneration.
Consequently, the
30
31
3
3
34
35
36
37
38
discovery of therapeutic agents that can selectively block the catalytic function of key enzymes in
this pathway are promising areas of research. Amongst the biological targets of interest is the human
FPPS, the gate-keeper of isoprenoid biosynthesis and consequently, prenylation/activation of all
small GTPases. Until recently, the only known inhibitors of this enzyme were the bisphosphonate
active site inhibitors, such as the N-BP drugs zoledronic acid (1)39 and risedronic acid (2).40
39
40
41
42
4
4
45
46
47
48
49
1
3
However, the discovery of a catalytically relevant allosteric pocket, located near the active site of
the enzyme, opened a new area of drug discovery leading towards the identification of non-
bisphosphonate inhibitors, with potentially superior drug-like properties and consequently,
therapeutic value as compared to the N-BPs.
50
51
52
53
5
5
56
57
58
A number of non-bisphosphonate allosteric inhibitors of hFFPs (e.g. compounds 3-5), with
good potency in vitro, have been reported and their binding interactions with the allosteric pocket
have been characterized by crystallography;13,15 unfortunately, none of these compounds have
59
60
1
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 64
1
2
3
4
5
6
7
8
9
1
1
12
13
shown promising cell-based potency. In our own search for new structural classes of hFPPS
allosteric inhibitors, we identified thienopyrimidine-based bisphosphonates (ThP-BPs; e.g.
compound 6) that exhibit a highly unusual binding mode (for bisphosphonate-type compounds) in
that they can bind to both the active site and the allosteric pocket of hFPPS under physiological
conditions. We investigated this dual binding mode by DSF, NMR and ITC and finally confirmed
14
1
16
2+
17
our results by co-crystallizing compound 6 with hFPPS in the presence and absence of Mg ions.
17
18
19
20
2
+
As expected, in the presence of Mg ions, compound 6 was driven to bind in the allylic subpocket
of the active site (refered to as the hFPPS/6/Pi(act); PDB code: 4JVJ). In contrast, co-crystallization
2
2
23
2+
of 6 in the absence of Mg ions revealed binding exclusively in the allosteric pocket of the enzyme
24
25
(
refered to as the hFPPS/6/Pi(allo) complex; PDB code: 4LPG; Fig. 5a). Removal of one phosphonate
26
2
28
moiety led to the corresponding monophosphonate analog 9, which was shown to selectively bind
29
30
in the allosteric pocket of hFPPS (Fig 5b; PDB code: 4LPH).
31
3
3
34
35
36
37
38
39
40
41
42
4
4
45
46
47
48
49
50
51
52
53
Figure 5: Co-crystal structures of hFPPS and ligand complexes. (a) Structure of the
hFPPS/6/Pi(allo) complex (PDB ID: 4LPG). Polar and stacking interactions are indicated with
dashes. An inorganic phosphate ion is bound in the IPP subpocket of hFPPS; its interactions
with the protein are omitted for simplicity. The side chain of K57 could not be fully modeled
due to disorder. (b) Structure of the hFPPS/9/Pi complex (PDB ID: 4LPH). Represented residues
are the same as in panel (a).
5
5
56
57
58
59
60
1
6
ACS Paragon Plus Environment

Page 17 of 64
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
12
13
14
15
16
17
18
19
20
2
2
23
24
25
26
27
28
29
30
31
3
Our crystallographic data also confirmed that the interactions between 9 and the protein surface
within the allosteric pocket were very similar to those previously observed in the hFPPS/6/Pi(allo)
complex (Fig 5a). In both cases, the predominant inhibitor/protein interactions involve π-stacking
of the 6-phenylthieno[2,3-d]pyrimidin-4-amine core with F206 and F239; in fact, the p-tolyl
substituent of 9 was buried in the allosteric pocket, engaging in a “sandwich” of π-stacking
interactions with the side chains of N59 and F206. At the time, we presumed that the low potency
of inhibitor 9 was mainly due to the fact that its phosphonate moiety formed only water-mediated
interaction with the guanidinium side chain of R60, whereas the carboxylic acid moieties of the
more potent inhibitors 3 and 4 engages in direct interactions with the side chains of K57 and
R60.13,15 To gain more insight into the importance of each key interaction between the ThP-MP
inhibitors and the allosteric pocket, we generated permutation libraries of structurally diverse
analogs, starting from the “naked” thienopyrimidine monophosphonate core (i.e. analog 7 ) and the
3
3
35
building blocks shown in Figure 4 as the variable substituents at the Cα/Cβ of R
and at R (Fig. 2
4
6
36
37
and 3).
38
39
Initial biological screening of all compounds was carried out using a previously described
40
41
42
4
4
45
46
47
48
49
50
51
1
9
in vitro M2 hFPPS inhibition assay at a fixed concentration of 10 μM. A full dose-response
inhibition curve (IC50) was determined only with compounds exhibiting more than 80% inhibition
at 10 μM (IC50 average values for assays run in triplicate or more are shown in Table 1). For the
purpose of our investigations, the IC50 values shown in Table 1, were fairly consistent with the SAR
suggested by the preliminary screening data shown in Figures 6, 7 and 8.
52
5
5
5
56
It is noteworthy that compound 7 was virtually inactive in our in vitro assay at
concentrations up to 10 μM and only slightly active even at 100 μM (Table 1; Fig. 6). However, a
57
58
sharp increase in potency was observed when the R of 7 was substituted with a phenyl or a p-tolyl
6
59
60
1
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 64
1
2
3
4
5
6
7
8
9
1
1
12
13
moiety (i.e. analogs 7 vs 8 vs 9; Fig. 6a), confirming the importance of the π-stacking interactions
between this group and residues F206 and N59 observed previously (Fig. 5). Having achieved a
reasonable potency in inhibiting hFPPS, we turned our attention to probing the impact of the
phosphonate pharmacophore. Initially, we replaced the α-aminophosphonate moiety of 9 with the
corresponding ethylphosphonic acid and glycine amino acid (i.e. analogs 10 and 26, respectively;
Fig. 2) and observed a strong correlation between loss in potency and decrease in the acidity of the
14
15
16
17
18
19
20
1
7
negatively-charged pharmacophore (i.e. analogs 9 vs 10 or 26; Fig. 6). Similarly, capping of one
hydroxyl group of the phosphonate as the ethyl ester led to a significant loss of potency (i.e. analog
2
2
23
9
vs 18; Fig 6), once again, suggesting that a phosphonate dianion may be essential for binding. In
24
25
26
27
28
29
30
31
3
3
34
35
addition, since interactions between the negatively charged pharmacophore of 9 and the
guanidinium side chain of R60 were previously shown to be water mediated (Fig. 5), we presumed
that introducing modifications to the linker that would allow similar binding of the thienopyrimidine
core and direct interactions of the phosphonate with R60 would lead to improvement in the potency
of our inhibitors. Several modifications were attempted, including linker extension (i.e. analog 9 vs
36
37
38
39
40
41
42
1
1; Fig 6b), partial rigidification of the extended linker via substitution at both the α- and β-carbons
of a (2-aminoethyl)phosphonic acid moiety [e.g. analog 12 vs 14 (Fig. 6c) and 17 vs 15 or 16 (Fig.
d)] and replacement of the α-aminophosphonate moiety with a (2-aminoacetyl)phosphoramidic
6
4
4
45
46
47
48
49
acid (e.g. analog 29 and 30; Fig. 6e). Contrary to our expectations, these extended linkers
connecting the thienopyrimidine core with the phosphonate moiety led to compounds that were
either equipotent or slightly less potent than the derivatives having an α-aminophosphonate moiety
50
51
52
53
5
5
56
at R
derivative 13 (IC50 = 1.4 μM) and the corresponding (2-aminoacetyl)phosphoramidic acid analog
9 (IC50 = 3.7 μM) was minimal.
4
. For example, the difference in potency between the α-amino-(2-phenylethyl)phosphonic acid
2
57
58
59
60
1
8
ACS Paragon Plus Environment

Page 19 of 64
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
12
13
14
15
16
17
18
19
20
2
2
23
24
25
26
27
28
29
30
31
3
3
34
35
Since the enzyme-bound conformation of inhibitor 9 (Fig. 5b) placed the phosphonate moiety
folded over the C-5 position of the thienopyrimidine core, we then assumed that connecting the
phosphonate moiety to the C-5 would allow direct interaction with R60. Based on this assumption
and preliminary in silico docking studies, we designed analogs such as 49 or 50 that were expected
to produce the desired outcome. Unfortunately, both of these compounds were found to be
significantly less active than anticipated (IC50 values of ≥100μM; Table 1), in spite of the fact that
the phosphonate moiety indeed interacts directly with the guanidinium ion of R60 at least for analog
49 (confirmed with the co-crystal structure of hFPPS/49; Fig. 9a). A likely reason is that the
methylene linker between the C-5 of the scaffold and the phosphonate clashed with the protein
surface and consequently, the aromatic core is pushed out of its preferred binding mode for strong
π-staking interactions. Additionally, unfavorable torsional stress was observed in the bound
compound 49 (i.e. loss of the co-planarity between the two aromatic rings; Fig. 9a). Given these
observations, it is reasonable to assume that the same effects are (at least in part) responsible for
the poor potency of the cyclic phostone 50.
36
37
We then turned our attention to the SAR at the Cα position of the R α-aminophosphonic acid
4
38
3
40
41
42
4
4
45
46
47
48
49
moiety. Initially, we observed some minor improvements in potency when a phenyl or benzyl
moiety was attached to the Cα as compared to the unsubstituted hit 9. For example, the racemic
benzyl-substituted derivative 13 was approximately 4-fold more potent than the unsubstituted
compound 9, whereas the racemic phenyl-substituted analog 19 was only 2-fold more potent than
the corresponding unsubstituted derivative 17 (Table 1). Analogs such as 17, having an additional
50
51
52
53
5
5
56
meta-halo substituent on the R
hit 9 (Table 1). These preliminary results suggested that further exploration at both the Cα position
of the C-4 α-aminophosphonate moiety and at R was justified.
6
p-tolyl group, appeared to be slightly more potent than the initial
6
57
58
59
60
1
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 64
1
2
3
4
5
6
7
8
9
1
1
12
13
14
15
16
17
18
19
20
2
2
23
24
25
26
27
28
29
30
31
3
3
3
35
36
st
Figure 6: Screening data for select compounds from the 1 library of hFPPS allosteric inhibitors
at a fixed concentration of 10 μM (average of % inhibition from triplicate test runs); preliminary
SAR at R
4
and R
6
of the thienopyrimidine core (fragment A to I are as shown in Fig. 2)
37
38
39
40
41
42
4
4
45
46
47
48
49
50
51
52
53
5
5
56
57
58
59
60
2
0
ACS Paragon Plus Environment

Page 21 of 64
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
12
13
14
15
16
1
18
19
20
2
2
23
24
25
2
27
28
29
30
31
3
3
34
3
36
37
Table 1: Inhibition data for key compounds
hFPPS
hFPPS
Compound IC50 (μM)
Compound IC50 (μM)
3α
7
0.85
>100
5.0
4.8
1.4
3.3
1.6
2.3
8.8
1.6
8.4
4.6
10
2.6
3.7
28
29
33
36
37
38
42α
43
44
47α
48
49
9
3.0
12
2.8
1
1
1
3
7
9
1.6
1.1
0.86
1.5
20
2
2
2
1
2
4
1.0
0.87
4.6
25
~100
2
7
38
39
40
41
α
Average IC50 values from three determinations; average IC50 values from six determinations;
4
4
(standard deviation ≤10%).
4
45
46
47
48
49
5
51
52
53
5
5
56
5
58
A library of compounds with variables substituents at R
having a phenyl or benzyl substituent at the Cα position of R
synthesized, examples are shown in Figure 7 (R = f1-f21). In general, R
a large meta or para moiety (e.g. OCH , CF , N(CH
contrast, when the p-tolyl moiety (i.e. R = f2) was further substituted with a 3-fluoro or 3-chloro
6
of the thienopyrimidine scaffold
4
(i.e. R
4
= D or E; Fig. 2) was
6
6
phenyl derivatives with
3
3
3
) ) were unfavorable (data not shown). In
2
6
59
60
2
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 64
1
2
3
4
5
6
7
8
9
(
i.e. f9 and f10, respectively) or 2-methoxy-5-fluoro/chloro (i.e. f20 and 21, respectively) a modest
improvement in potency was observed. Confirming our previous observations (i.e. Fig. 6), a very
modest improvement in potency was usually observed when a phenyl or benzyl substituent was
1
1
12
13
14
15
16
17
introduced at the Cα position of the R
the IC50 values of the most potent compounds confirmed a ~2-4-fold improvement in potency when
the Cα position of R was substituted with a benzyl group (e.g. IC50 values of analogs 9 and 13 are
.0 µM and 1.4 µM, respectively; Table 1). Since all analogs were initially synthesized as racemic
4
moiety (i.e. scaffolds II and III vs I; Fig. 7). For example,
4
5
18
19
2
2
2
23
24
25
26
27
28
29
30
31
mixtures, these results seemed promising for further optimization; initially, we presumed that only
one enantiomer was responsible for all the activity observed. Analogs (R)- and (S)-13 were
synthesized in 98% and 92% ee, respectively, and their IC50 values determined to be 4 μM and 1
μM, respectively; this small difference in potency between the two enantiomers hinted to a
potentially inconsistent binding mode for these molecules. Nonetheless, a small library of racemic
compounds with structural diversity at the Cα of the α-aminophosphonic acid was investigated in
3
3
34
35
36
37
38
3
40
41
42
4
4
45
4
47
48
49
combination with fragment f2, f10, f20 and f21 at the R
representative examples from this library are shown in Figure 8. In general, both aromatic and alkyl
groups were well tolerated (e.g. analogs 13 with R =f22 was found to be equipotent to analog 37
with R =f24 (Fig. 8; Table 1), suggesting some van der Waals interactions between the R moiety
and the protein. Significant loss in potency was observed when R was a basic moiety, such as
pyridinyl derivatives (e.g. R = f29, f30 or f31) or a disubstituted alkyl derivatives (e.g. f32; Fig.
). From this cluster of compounds, the most potent hFPPS inhibitors identified were those having
a combinations of fragments f23 at the R position and either f10 or f20 at R (i.e. analogs 42 and
7, respectively; Fig. 8; Table 1). The IC50 values of compounds 42 (IC50 = 0.86 µM) and 47 (IC50
6
position (i.e. sub-scaffold IV; Fig. 8);
1
1
1
1
1
8
50
51
52
53
1
6
4
5
5
56
57
58
59
60
2
2
ACS Paragon Plus Environment

Page 23 of 64
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
=
0.87 µM) were found to be in the same potency range as allosteric inhibitor 3 (IC50 = 0.85 µM),
when tested in parallel in our in vitro M2 hFPPS inhibition assay (Table 1).19
1
1
12
13
14
15
16
17
18
19
20
2
2
23
24
25
26
27
28
29
30
31
3
3
34
35
3
nd
Figure 7: Screening inhibition data for select compounds from the 2 library of hFPPS allosteric
inhibitors at a fixed concentration of 10 μM (average of % inhibition from triplicate test runs); the
37
38
39
40
R
(
6
position of the (aminomethyl)- phosphonic acid (I), (amino(phenyl)methyl)phosphonic acid
II) and an (1-amino-2-phenylethyl)phosphonic acid (III) analogs were substituted with
fragments f1 to f21 (Fig. 4).
41
42
4
4
45
46
47
48
49
50
51
52
53
5
5
56
57
58
59
60
2
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 64
1
2
3
4
5
6
7
8
9
1
1
12
13
14
15
16
17
18
19
2
2
2
rd
Figure 8: Screening inhibition data for select compounds from the 3 library of hFPPS allosteric
inhibitors at a fixed concentration of 10 μM (average of % inhibition from triplicate test runs); x-
axis and z-axis indicate R
1
and R fragments, respectively, from Fig. 4.
6
23
24
25
26
27
28
29
30
31
3
3
34
35
36
37
38
39
40
41
42
Mindful of the conformational plasticity of this enzyme, which permits the binding of many
structurally diverse compounds at its allosteric pocket (e.g. 3, 4, 5 and 9), several representative
molecules, including analogs 29, 33, 38, 42 and 49, were co-crystallized with hFPPS in order to
identify a consensus binding mode. All compounds were found to bind at the allosteric site of
hFPPS. However, three compounds (analogs 29, 38, and 42) were found to bind in an “upside
down” orientation with respect to the binding of our initial hit 9, in which the sulfur atom of the
thienopyrimidine scaffold is facing the active site cavity instead of the C-terminal helix (Fig. 9c,d,e
vs Fig. 5b). In addition to the sandwich π-stacking interactions of the thienopyrimidine core, the
exocyclic nitrogen of the analogs 29 and 38 formed a H-bond with the carbonyl oxygen of K347
4
4
45
46
47
48
49
(Fig. 9c,d). The substituents (2-aminoacetyl)phosphoramidate and α-aminophosphonate also made
50
5
52
an electrostatic interaction with K347 and a water-mediated H-bond with N59 (Fig. 9c,d).
53
5
Compound 42 did not form any of these additional interactions; however, its fluorobenzyl moiety
5
56
at R
1
(i.e. the Cα of R ) made a new stacking interaction with F239, tucked tightly in the small
4
57
58
59
60
2
4
ACS Paragon Plus Environment

Page 25 of 64
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
depression formed by F239, Q242, D243, and I348 (Fig. 9e; SI Fig. S3). It is interesting that
compound 33, unlike the three analogs above, assumed the orientation observed with the original
hit 9 (Fig. 9b vs Fig. 5). Comparison with the other structures indicates that the dihydrobenzofuran
1
1
12
1
14
15
16
derivative 33, with a bulkier and less flexible R
of α and α in the upside down binding mode (Fig. 9b; SI Fig. S3).
Interestingly, when we superimposed all of our co-crystal structures, we found that the
1
moiety, likely causes steric clashes with residues
H
J
17
18
19
20
2
2
23
24
25
26
27
28
29
30
31
3
3
34
35
36
37
38
39
40
41
42
4
4
thienopyrimidine cores of the inhibitors generally occupy the same 3D space within the allosteric
binding site, regardless of the binding orientation of each molecule (i.e. “up” or “down”). This
finding highlights the importance of the π-stacking interactions provided by the scaffold for the
binding of these compounds. Notably, the thienopyrimidine of compound 42 aligns slightly off
from those of the rest of the inhibitors (Fig. 9f) and makes the closest contact with F239 of the
protein (Fig. 9e). Since 42 differs from 38 only by a single chloride atom, its distinct binding pose
is attributed to this substitution. A close examination reveals that the m-chloro tolyl group fits more
tightly in the allosteric pocket and in a slightly tilted angle that pushes the thienopyrimidine core
towards the active site cavity (Fig. 9g). This binding pose not only allows stronger interaction
between the thienopyrimidine core and F239 but also enables the aforementioned burial of the Cα
fluorobenzyl moiety in the protein. The monophosphonate of 42 in this binding pose is too far from
K347 to interact with this residue.
45
46
47
48
49
50
51
52
53
5
5
56
57
58
59
60
2
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 64
1
2
3
4
5
6
7
8
9
1
1
12
13
14
15
16
17
18
19
20
2
2
23
24
25
26
27
28
29
30
31
3
3
34
35
36
37
38
39
40
41
42
Figure 9. Binding poses of ThP-MP inhibitors at the allosteric site of hFPPS. Inhibitors 49, 33, 29,
3
8, and 42 are shown in (a), (b), (c), (d), and (e), respectively. The red and green spheres represent
water molecules and a chloride ion, respectively. (f) Superposition of the new ThP-MPs and the
4
4
45
46
initial hit, compound 9 (cyan). The thienopyrimidine of 42 is marked with a black arrow. (g)
Binding of the R tolyl groups of 38 (left panel) and 42 (right). The inhibitors are shown as space-
6
filling models, and the protein surfaces as meshes.
47
48
49
50
51
52
53
5
5
56
Based on all previous co-crystal structures of inhibitors bound to the hFPPS allosteric
pocket,13,15,17 we expected the polar interactions between the monophosphonate group and the side
57
58
59
60
2
6
ACS Paragon Plus Environment

Page 27 of 64
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
12
13
14
15
16
17
18
19
20
2
chain(s) of N59 and/or R60 to be crucial for inhibitor/enzyme molecular recognition. Consequently,
the upside down binding mode of analogs 29, 38 and 42 was surprising, particularly since the
binding orientation of the ThP-MPs (i.e. “up” or “down”) does not seem to significantly affect their
potency (e.g. IC50 values of analog 33 and 42 are 3.0 and 0.86 μM, respectively, Table 1). More
importantly, our results indicate that, given the right combination of substituent around the
thienopyrimidine core, the phosphonate-mediated interactions are entirely dispensable; the
monophosphonate of compound 42 when bound to hFPPS is exposed to bulk solvent without
making any contact with the protein (Fig. 9e; SI Fig. S3e). The loss in the binding energy seems to
2
23
be fully compensated by the non-polar interactions provided by the m-chloro substituent on the R
6
24
25
p-tolyl group and the fluorobenzyl substituent at the Cα of the R in compound 42, which shows
4
26
2
28
29
30
31
3
3
34
35
36
37
38
39
40
the greatest potency amongst all of our ThP-MP compounds and is equipotent to the allosteric
inhibitor 3 (when tested in parallel in our M2 in vitro assay; Table 1). In contrast, compound 49 is
approximately 100-fold less potent than 42 (Table 1), in spite of the fact that its phosphonate moiety
makes direct interactions with both N59 and R60 (Fig. 9a). Therefore, analogs such as 42 may
suggest a new direction towards the identification of potent hFPPS allosteric inhibitors that do not
require a highly charged moiety.
41
42
4
4
45
4
47
CONCLUSIONS:
48
49
50
51
52
53
5
Over the last few years, there has been an on-going debate in the literature as to whether N-
BP inhibitors of hFPPS possess antitumor therapeutic properties, in addition to their undisputed
skeletal benefits. This debate has been fueled by somewhat controversial (or unclear) clinical data
5
56
8
41
from multiple myeloma and breast cancer patients treated with standard chemotherapy plus
57
58
59
60
2
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 64
1
2
3
4
5
6
7
8
9
1
1
12
13
14
15
16
17
18
19
20
2
2
23
24
25
26
27
28
29
30
31
3
3
34
35
36
37
zoledronic acid (1). However, given the extremely poor soft-tissue distribution and low systemic
exposure of N-BP drugs, it is unlikely that a better understanding of the role of hFPPS in
oncogenesis can be gained from in vivo studies using these compounds. Herein, we disclose the
synthesis of compound libraries used to conduct hit-to-lead optimization of a new series of
thienopyrimidine-based allosteric inhibitors of hFPPS. These studies led to the identification of
hFPPS allosteric inhibitors with nanomolar potency, such as compounds 42 and 47. X-ray
crystallography was used to identify the molecular recognition elements dictating the
inhibitor/protein interactions. These studies strongly suggest that the 6-tolylthienopyrimidine
scaffold drives the binding by providing a number of consistent π-staking interactions with residues
F239, F206 and N59. Interestingly, in the co-crystal structure of our most potent inhibitor, analog
42, the phosphonate moiety was not involved in any interactions with the protein. This observation
may appear to contradict our previous SAR data, which indicated that the phosphonate
pharmacophore was essential for binding. However, it suggests that it may be possible to design
allosteric inhibitors of hFPPS that binding in a different orientation from that of the original
thienopyrimidine hits (i.e. analogs 6 and 9; Fig. 5), yet maintain the same strong π-stacking
interactions observed with residues F239, F206 and N59 (in addition to other interactions), but
lacking the highly charged phosphonate moiety. Such investigations are currently in progress and
if successful, they may provide the molecular tools needed to clearly validate (or disqualify) hFPPS
as a therapeutic target for non-skeletal diseases, such as cancer.
38
39
40
41
42
4
4
45
46
47
48
49
50
51
52
53
5
5
56
57
58
59
60
2
8
ACS Paragon Plus Environment

Page 29 of 64
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
12
13
EXPERIMENTAL
General Procedures for Characterization of Compounds:
The phosphonate ester precursor of inhibitors 7 - 48 were purified by normal phase flash column
chromatography using a CombiFlash instrument on silica gel and a solvent gradient from 5%
EtOAc in hexanes to 100% EtOAc and then to 20% MeOH in EtOAc, unless otherwise indicated.
Only phosphonate esters with homogeneity ≥95% were processed further though the hydrolysis
step to the final ThP-MP phosphonic acids. The purity and homogeneity of all final inhibitors to
14
15
16
17
18
19
2
2
1
31
≥95% was confirmed by NMR ( H, and P) and reversed-phase HPLC. HPLC Analysis was
2
23
performed using a Waters ALLIANCE® instrument (e2695 with 2489 UV detector and 3100 mass
24
25
1
13
31
spectrometer). Key compounds and intermediates were fully characterized by H, C and P NMR,
26
27
28
1
and HRMS. Chemical shifts () are reported in ppm relative to the internal deuterated solvent ( H,
29
30
31
3
13C) or external H
PO
( 0.00 31P), unless indicated otherwise. The NMR spectra of all final
or in D O (with ~2%-5% ND OD or as the
3
4
inhibitors were acquired either in DMSO-d
6
2
4
3
34
35
36
37
38
39
40
corresponding di-sodium salt). The high-resolution MS spectra were recorded using electrospray
ionization (ESI+/-) and Fourier transform ion cyclotron resonance mass analyzer (FTMS).
Method (homogeneity analysis using a Waters Atlantis T3 C18 5 µm column):
41
42
4
4
45
46
47
48
49
50
51
52
Solvent A: H
Solvent B: CH
Mobile phase: linear gradient from 95%A and 5%B to 5%A and 95%B in 13 min, then 2 min at
00% B
Flow rate: 1 mL/min
2
O, 0.1% formic acid
3
CN, 0.1% formic acid
1
53
5
5
56
57
58
59
60
2
9
ACS Paragon Plus Environment