Discovery of Lipophilic Bisphosphonates That Target Bacterial Cell Wall and Quinone Biosynthesis

Article abstract of DOI:10.1021/acs.jmedchem.8b01878

We report that alkyl-substituted bisphosphonates have activity against Bacillus anthracis Sterne (0.40 μg/mL), Mycobacterium smegmatis (1.4 μg/mL), Bacillus subtilis (1.0 μg/mL), and Staphylococcus aureus (13 μg/mL). In many cases, there is no effect of serum binding, as well as low activity against a human embryonic kidney cell line. Targeting of isoprenoid biosynthesis is involved with 74 having IC₅₀ values of ～100 nM against heptaprenyl diphosphate synthase and 200 nM against farnesyl diphosphate synthase. B. subtilis growth inhibition was rescued by addition of farnesyl diphosphate, menaquinone-4 (MK-4), or undecaprenyl phosphate (UP), and the combination of MK-4 and UP resulted in a 25× increase in ED₅₀, indicating targeting of both quinone and cell wall biosynthesis. Clostridioides difficile was inhibited by 74, and since this organism does not synthesize quinones, cell wall biosynthesis is the likely target. We also solved three X-ray structures of inhibitors bound to octaprenyl diphosphate and/or undecaprenyl diphosphate synthases.

Full text of DOI:10.1021/acs.jmedchem.8b01878

Subscriber access provided by McMaster University Library
Article
Discovery of Lipophilic Bisphosphonates that
Target Bacterial Cell Wall and Quinone Biosynthesis
satish malwal, Lu Chen, Hunter Hicks, Fiona Qu, Weidong Liu, Alli Shillo, Wen
Xuan Law, Jianan Zhang, Neal Chandnani, Xu Han, Yingying Zheng, Chun-Chi
Chen, Rey-Ting Guo, Ahmed AbdelKhalek, Mohamed N. Seleem, and Eric Oldfield
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b01878 • Publication Date (Web): 07 Feb 2019
Downloaded from http://pubs.acs.org on February 8, 2019
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 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 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.
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 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Discovery of Lipophilic Bisphosphonates that Target Bacterial Cell Wall and
Quinone Biosynthesis
Satish R. Malwal *, Lu Chen , Hunter Hicks , Fiona Qu , Weidong Liu3,4, Alli Shillo1
1
,
2,
1,
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
5
1
3
3
Wen Xuan Law , Jianan Zhang , Neal Chandnani , Xu Han , Yingying Zheng , Chun-
4
3,4
6
6,7
Chi Chen , Rey-Ting Guo , Ahmed AbdelKhalek , Mohamed N. Seleem , and Eric
1
,
Oldfield *
1Department of Chemistry, University of Illinois at Urbana-Champaign, Urbana, IL 61801, USA
2Department of Biochemistry, University of Illinois at Urbana-Champaign, Urbana, IL 61801,
USA
3Industrial Enzymes National Engineering Laboratory, Tianjin Institute of Industrial
Biotechnology, Chinese Academy of Sciences, Tianjin 200208, China
4State Key Laboratory of Biocatalysis and Enzyme Engineering, Hubei Collaborative Innovation
Center for Green Transformation of Bio-resources, Hubei Engineering Research Center for Bio-
enzyme Catalysis, Hubei Key Laboratory of Industrial Biotechnology, School of Life Sciences,
Hubei University, Wuhan 430062, China
5School of Molecular and Cellular Biology, University of Illinois at Urbana-Champaign, Urbana,
IL 61801, USA
6Department of Comparative Pathobiology, College of Veterinary Medicine, Purdue University,
West Lafayette, IN 47907, USA
7Purdue Institute of Inflammation, Immunology, and Infectious Disease, West Lafayette, IN 47907,
USA
Correspondence to: satishm@illinois.edu (S.R.M); eoldfiel@illinois.edu (E. O.)
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 58
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Abstract
We report that alkyl-subsituted bisphosphonates have activity against Bacillus anthracis Sterne
0.40 µg/mL), Mycobacterium smegmatis (1.4 µg/mL), Bacillus subtilis (1.0 µg/mL) and
(
Staphylococcus aureus (13 µg/mL). In many cases, there is no effect of serum binding, as well as
low activity against a human embryonic kidney cell line. Targeting of isoprenoid biosynthesis is
involved with74 having IC50 values of ~100 nM against heptaprenyl diphosphate synthase and
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
00 nM against a farnesyl diphosphate synthase. Bacillus subtilis growth inhibition was rescued
by addition of farnesyl diphosphate, menaquinone-4 (MK-4) or undecaprenyl diphosphate (UP)
and the combination of MK-4 plus UP resulted in a 25x increase in ED , indicating targeting of
50
both quinone and cell wall biosynthesis. Clostridioides difficile was inhibited by 74 and since this
organism does not synthesize quinones, cell wall biosynthesis is the likely target. We also solved
three X-ray structures of inhibitors bound to octaprenyl diphosphate and/or undecaprenyl
diphosphate synthases.
2
ACS Paragon Plus Environment

Page 3 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Introduction
There is a need for new antibiotics because of the rise in drug resistance and in our group
we have been investigating enzymes involved in isoprenoid biosynthesis as potential anti-infective
drug targets.1-2 These enzymes include farnesyl diphosphate synthase (FPPS) and undecaprenyl
diphosphate synthase (UPPS), both of which are involved in bacterial cell wall biosynthesis, and
we discovered several inhibitors,2-3 one of which was active in vivo in a mouse model of
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
4
Staphylococcus aureus infection. We also recently discovered several inhibitors of S. aureus
heptaprenyl diphosphate synthase (SaHepPPS), essential for the formation of menaquinone,
required for electron transport (ET) and hence, ATP synthesis, in many bacteria and as with UPPS,
this enzyme is not produced by humans. Here, we sought to find new compounds that might target
octaprenyl diphosphate synthase (EcOPPS) or heptaprenyl diphosphate synthase (SaHepPPs),
inhibiting quinone biosynthesis. A simplified version of the enzymes involved in quinone as well
as cell wall biosynthesis in many bacteria is shown in Figure 1a, together with the sites of action
of several antibiotics, and in Figure 1b we show the chemical structures of selected
substrates/products in Staphylococcus aureus.
Isoprenoid biosynthesis begins with the formation of two C -isoprenoid diphosphates,
5
isopentenyl diphosphate (1, IPP) and dimethylallyl diphosphate (2, DMAPP) which are produced
by either the mevalonate (MEV) pathway, as in S. aureus, or by the so-called non-mevalonate or
deoxyxylulose-5-phosphate (DXP) pathway, the latter being far more common than the MEV
pathway. These two isoprenoids react to form (C ) geranyl diphosphate and then, via addition of
10
a second IPP, farnesyl diphosphate (FPP, 3), both reactions being catalyzed by the enzyme
farnesyl diphosphate synthase (FPPS). FPP is then elongated in further reactions with IPP to
produce long chain (typically all-trans C , C , C ) diphosphates (e.g. 4) that then react with e.g.
3
0
35
40
1
,4-dihydroxy-2-naphthoic acid (DHNA) to form quinone precursors, Figure 1b. In some bacteria
(
e.g. E. coli), both menaquinones as well as ubiquinones are formed. FPP is also converted to very
long-chain cis-isoprenoids such as undecaprenyl diphosphate (UPP, 5), in
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 58
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1. Schematic illustration of some of the enzymes involved in cell wall and quinone
biosynthesis in many bacteria, together with chemical structures of substrates and
intermediates and sites of action of some inhibitors (red) and “rescue” agents (blue). (a)
Enzymes, products, inhibitors and rescue agents. UP = undecaprenyl phosphate; MK-4 =
menaquinone-4. (b) Chemical structures of selected enzyme substrates and products
discussed in the Text. HepPPS (heptaprenyl diphosphate synthase) is a heterodimeric
enzyme. Some bacteria such as E. coli use the homodimeric octaprenyl diphosphate
synthase (OPPS), and also produce ubiquinones (not shown). DXP = the 1-deoxy-D-
xylulose 5-phosphate pathway, found in most bacteria; MEV = the mevalonate pathway,
found in e.g. S. aureus. In some bacteria, e.g. Listeria monocytogenes, both the DXP and
MEV pathways are present.
4
ACS Paragon Plus Environment

Page 5 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
reactions catalyzed by undecaprenyl diphosphate synthase (UPPS), with UPP then being
converted to undecaprenyl (mono)phosphate (UP) by undecaprenyl diphosphate phosphatase
(
UPPP), essential for cell wall peptidoglycan biosynthesis.
5
OPPS is an essential gene in E. coli and catalyzes the formation of the C isoprenoid
4
0
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
octaprenyl diphosphate (OPP) from farnesyl diphosphate (FPP) and five molecules of isopentenyl
6
diphosphate (IPP). Its structure is known and is highly -helical, resembling that of other trans-
prenyltransferases, such as FPPS.7-8 Since OPPS is absent in humans, this makes it a potential
anti-bacterial drug target. There are, however, no reports of bacterial OPPS inhibitors. In S.
aureus, the corresponding long-chain prenyl transferase is the C35 heptaprenyl diphosphate
4
synthase, SaHepPPS. This has a somewhat unusual heterodimeric structure consisting of a large,
"
catalytic" domain (SaHepPPS-2) and a smaller, regulatory domain (SaHepPPS-1), and both are
essential for activity. Similar heterodimeric structures are found in other bacteria—such as
9
Micrococcus luteus, Bacillus spp. and Enterococcus spp. and the structures of the catalytic sites
in OPPS, HepPPS as well as FPPS are quite similar. In this work, we first sought to find inhibitors
of OPPS and HepPPS, active in cells. Then, we extended this work to better understand inhibitor
mechanisms of action, in addition to solving several structures of interest.
In order to find new, long-chain prenyl transferase inhibitors we first screened a library of
previously-reported compounds including bisphosphonates, benzoic, salicylic, anthranilic and
diketoacids, for OPPS inhibition, since these classes of compound were previously shown to
2
inhibit prenyltransferases , and some have anti-bacterial activity. We then screened a subset of
compounds for bacterial cell growth inhibition (against S. aureus, B. subtilis, B. anthracis Sterne,
Mycobacterium smegmatis, E. coli, Acinetobacter baumannii, Klebsiella pneumoniae,
Pseudomonas aeruginosa) as well as the fungus, Candida albicans, and used the results obtained
to guide the synthesis of 24 new compounds. We then investigated the mechanism of action of
some of the most potent compounds in cells using enzyme inhibition assays, as well as bacterial
cell growth inhibition “rescues” by putative enzyme end-products, followed by measurements of
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 58
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
the effects of serum binding on antibacterial activity, and toxicity to a mammalian cell line. We
also determined activity against Clostridioides difficile, a bacterium that does not use quinones
(
or hemes) in ATP biosynthesis, again to help clarify mechanisms of action. Finally, we solved
several X-ray crystal structures, to determine inhibitor binding modes. Overall, the results are of
general interest since we have discovered several potent multi-target antibacterials with low
serum binding and low toxicity against a human cell line that act by inhibiting both quinone as
well as cell wall biosynthesis.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Results and Discussion
Initial Screening Results. We first investigated the inhibition of E. coli octaprenyl
diphosphate synthase (EcOPPS) since in previous work we found that this protein expressed well
and was more stable than the corresponding enzyme from S. aureus, SaHepPPS, and it has a
better resolved crystal structure. We screened a diverse range of potential inhibitors (6-67):
bisphosphonates, phosphonates, benzoic/anthranilic acids, and amines/diamines/diamidines,
classes of compounds that in earlier work1-4, 10 we found inhibited prenyltransferases such as
3
10
4
2
FPPS, geranylgeranyl diphosphate synthase (GGPPS), HepPPS, UPPS, MenA (1,4-dihydroxy-
4
1
2
-naphthoate polyprenyltransferase ) and dehydrosqualene synthase. About one third of the
compounds (Supporting Information, Figure S1), had essentially no activity (IC >1 mM) against
50
EcOPPS, but there were several compounds having low nM activity. All of the most potent hits
were bisphosphonates, the most active species (6, Figure 2) having an IC value of ~6 nM. The
50
lipophilic benzoic/anthranilic/salicylic acids were less active, the most potent species having an
IC ~5 µM, Figure S1.
50
We then tested 6, together with 5 other compounds (33, 68-71; Figure 2 and Table S1),
against S. aureus, B. subtilis, B. anthracis Sterne, M. smegmatis, E. coli, A. baumannii, K.
pneumoniae, P. aeruginosa as well as C. albicans, Table S1. Three compounds were lipophilic
6
ACS Paragon Plus Environment

Page 7 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 2. Structures of some compounds that inhibit isoprenoid biosynthesis enzymes
such as FPPS, GGPPS, UPPS, and UPPP that are discussed in the Text.
1
1
bisphosphonates (33, 68, 69), and two were benzoic acids (70, 71), originally designed as
phosphate/diphosphate isosteres and known to inhibit other isoprenoid biosynthesis enzymes
including the cis-prenyltransferase UPPS,11-12 the trans- prenyltransferase GGPPS, and prenyl
12
1
1
11
diphosphate phosphatase, UPPP. Plus, e.g. 71 has potent anti-bacterial activity. There was
little-to-no activity (ED50 > 50 µg/mL) against the gram-negative bacteria or C. albicans for all
compounds, but there was activity in the ~3-8 µg/mL range for some compounds against S.
aureus, M. smegmatis, B. subtilis or B. anthracis Sterne, Table S1. The lack of activity against the
gram-negative bacteria was unexpected because in previous work13 we found that lipophilic
bisphosphonates such as 72 (Figure 2) had quite potent (~2 µg/mL) activity against the same
gram-negative bacteria tested here, but had low activity against the gram-positive bacteria S.
aureus (30 µg/mL) and B. subtilis (> 100 µg/mL). However, with 6, we see the opposite trend,
and one possibility is that the presence of an aromatic group “distal” to the bisphosphonate
1
4
backbone (seen also with 73, Figure 2) is required for transport into gram-negative bacteria.
What is also interesting about the bisphosphonate results is that the patterns of OPPS
inhibition are similar to those we find with the shorter (C ) prenyl synthase FPPS, as opposed to
1
5
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 58
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
the longer (C ) chain synthase, geranylgeranyl diphosphate synthase (GGPPS). For example, 24
2
0
(
zoledronate, Figure 2) is a <1 µM inhibitor of OPPS as well as of human FPPS,15-16 but only a very
weak (IC50 ~100 µM) inhibitor of human GGPPS.12 For potent FPPS inhibition, we proposed
previously17-19 that there was a requirement for either a cationic or protonatable group close to the
bisphosphonate backbone for activity, mimicking a reactive intermediate in FPPS catalysis. There
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
20
was, however, no requirement for such a cationic feature for GGPPS inhibition. For example,
6
, containing a cationic charge center, has an IC50 = 100 nM for FPPS inhibition and an IC50
=
2
80 nM for GGPPS inhibition, while 36 (Figure 2), which lacks this feature, has an IC = 550 µM
50
2
0
for FPPS inhibition but an IC = 590 nM for GGPPS inhibition. Here, we find that 36 has an
50
IC = 11 µM against OPPS, while as noted above, 6 is far more potent, as discussed more below.
50
Synthesis and testing of novel bisphosphonates. Based on the results described above, we
reasoned that it would be of interest to synthesize a series of analogs of 6 in which we varied: 1)
the nature of the aliphatic side-chain connected to the aryl group; 2) the position of the aliphatic
side-chain on the ring; 3) the nature of the side-chain linker (O or CH ) between the aliphatic and
2
aryl group and 4), the nature of the linker (NH, NHCH , CH ) to the bisphosphonate backbone
2
2
with in most cases, a protonatable/cationic feature being included. We thus synthesized the 24
compounds (74-97) shown in Figure 3 and tested each of them for B. subtilis, B. anthracis Sterne,
S. aureus and M. smegmatis growth inhibition. ED values (in µg/mL) and computed clogP and
50
logD7.4 values are shown in Table 1, and for convenience are shown together with their
corresponding chemical structures in Table S2. There are several points of interest. First, we see
that there is activity (in the ~0.6-2 µg/mL range) against B. anthracis Sterne, M. smegmatis and
B. subtilis, though less so with S. aureus (~13 µg/mL). Second, with B. subtilis (where there is a
large range in activity) we see (Figures S2) that there is a parabolic dependence between cell
8
ACS Paragon Plus Environment

Page 9 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3. Structures of bisphosphonates synthesized. Compounds are rank-ordered by
activity against B. subtilis from most active (74, top-left) to least active (97, bottom). Most
of the active compounds contain meta-substituted pyridinium rings and a medium-size
side-chain (cyan). Short or long chain substituents (red) are inactive.
growth inhibition (log ED , µg/mL) values and clogP with a Pearson correlation coefficient r =
1
0
50
0
.81. The most active compounds have clogP ~-2 while some of the least active compounds are
either much more polar (91, 93) and will have very poor cell penetration, or are far more
hydrophobic (97, with a C side-chain) and may get “trapped” in cell membranes. What is clear,
2
0
however, is that—as can be seen in Table S2—all of the most active compounds have fixed (+1)
charges on the aromatic group. That is, they are either pyridiniums or imidazoliums. This is of
1
9
interest since in previous work on bisphosphonate inhibition of FPPS we found, using solid-state
NMR, that bisphosphonates with basic side-chains (such as aminopyridines) bound to FPPS with
their side-chains protonated, suggesting the possibility that both FPPS as well as a very long chain
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 58
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 1. Bacterial cell growth inhibition by bisphosphonates together with clogP and logD7.4
values.a
Cpd#
Bs ED50
Ba ED50
Sa ED50
Ms ED50
HEK 293
clogP logD7.4
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
g/mL (M) g/mL (M)
g/mL (M)
g/mL (M)
g/mL (M)
7
7
7
4
5
6
1.0 (2.4)
1.7 (4.1)
2.1 (4.8)
2.2 (5.2)
2.4 (5.7)
2.8 (6.4)
3.1 (7.4)
3.6 (8.2)
4.6 (12)
5.5 (12)
5.7 (14)
2.3 (5.4)
1.8 (4.4)
0.6 (1.4)
2.1 (5.0)
0.8 (1.9)
0.4 (0.9)
2.2 (5.2)
2.4 (5.5)
4.3 (11)
14 (33)
26 (63)
13 (30)
13 (31)
13 (31)
18 (41)
21 (50)
51 (116)
34 (86)
30 (67)
28 (66)
5.9 (14)
1.4 (3.4)
6.1 (14)
5.9 (14)
5.5 (13)
6.2 (14)
7.2 (17)
10 (23)
15 (38)
262 (620)
181 (440)
96 (220)
-2.25
-2.08
-0.50
-1.14
-5.82
-5.65
-4.07
-4.71
-6.07
-5.09
-4.71
-6.25
-6.07
-3.65
-5.82
7
7
194 (460)
186 (440)
224 (510)
295 (700)
480 (1100)
170 (420)
22 (48)
7
8
9
-2.50
-1.52
-1.14
7
8
0
8
1
-2.46
-2.50
-0.08
-2.25
82
83
84
2.2 (4.9)
2.5 (5.9)
7.2 (16)
150 (350)
420 (1000)
8
5
6
7.0 (16)
1.1 (2.5)
31 (71)
70 (160)
440 (1000)
-1.52
-5.08
8
8.3 (19)
21 (48)
5.0 (12)
3.2 (7.3)
23 (54)
35 (80)
150 (350)
75 (170)
ND
-0.1
-2.36
-3.10
8
7
530 (1200)
-0.51
1
0
ACS Paragon Plus Environment

Page 11 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Cpd#
Bs ED50
Ba ED50
Sa ED50
Ms ED50
HEK 293
clogP logD7.4

g/mL (M) g/mL (M)
g/mL (M)
g/mL (M)
g/mL (M)
8
8
21 (49)
26 (60)
36 (73)
41 (120)
75 (170)
97 (340)
7.4 (17)
3.0 (6.9)
5.0 (10)
8.6 (24)
4.2 (9.6)
2.3 (8.0)
11 (28)
36 (84)
32 (73)
150 (350)
93 (210)
ND
480 (1100)
ND
0.48
-0.24
-0.27
-4.24
-0.52
-3.73
-0.36
1.20
-2.12
-2.00
-3.48
-7.81
-2.24
-6.35
-2.96
-1.63
-2.31
-1.85
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
89
90
91
22 (45)
91 (180)
17 (48)
65 (180)
ND
92
93
94
95
96
60 (140)
230 (800)
56 (140)
69 (140)
52 (120)
25 (44)
200 (460)
290 (1000)
96 (240)
79 (160)
ND
ND
100 (250)
110 (230)
180 (420)
260 (460)
ND
2.5 (5.2)
11 (26)
630 (1300)
ND
100 (240)
210 (370)
0.77
9
7
7.7 (14)
ND
1.72
a
Abbreviations used: Bs=B. subtilis; Ba=B. anthracis Sterne; Sa=S. aureus; Ms = Mycobacterium
smegmatis; HEK293 = human embryonic kidney cell line # 293; clogP = the computed logarithm
of the octanol/water partition coefficient; logD7.4 = the logarithm of the computed octanol/water
partition coefficient at pH = 7.4.
1
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 58
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 4. Typical dose-response curves for SaHepPPS inhibition. The most potent inhibitors also
have potent activity (~10-20 nM) against EcOPPS (Figure S3), but were not active E. coli or other
gram-negative bacteria. Results shown represent three pooled data sets taken on different days,
fit to single dose-response curves.
1
2
ACS Paragon Plus Environment

Page 13 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
trans-prenyl diphosphate transferases (such as HepPPS) might be being targeted, which would
lead to inhibition of both quinone as well as cell wall biosynthesis. We thus next sought to
determine which enzymes might actually be being targeted. To do this we used an expressed
SaHepPPS (a hetero-dimer found in Bacilli) and for comparison, EcOPPS (a homo-dimer).
Typical dose-response results for SaHepPPS are shown in Figure 4. What is clear from Figure 4 is
that two of the most active compounds are 6 and 74, compounds that also have potent activity
against B. anthracis Sterne, B. subtilis as well as M. smegmatis. A key feature for potent activity
against HepPPS as well as these bacteria is the presence of a long-chain m-substituted pyridinium
group. The amino-pyridine bisphosphonates as well as the aryl-alkyl imidazolium
bisphosphonates have lower activity against HepPPS, consistent with the cell growth inhibition
assay results. We also found very potent activity of some compounds against EcOPPS, typical
results are shown in Figure S3. For example, 74 had a 12 nM IC and 6 a 20 nM IC value, Figure
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
50
50
S3. These compounds thus inhibit both SaHepPPS and EcOPPS—but only have activity against
the gram-positive bacteria.
Figure 5. Cartoon illustration of the similarity between a putative FPP transition state/reactive
intermediate (top) and a potent SaHepPPS inhibitor, 74 (bottom). Note that this mechanistic
proposal would only apply to “long-chain” (~C , C , C ) trans prenyltransferases (which use
3
0
35
40
FPP as a substrate) and not to short-chain prenyl transferases, such as FPPS, since FPP is the
product and presumably would have only weak binding to FPPS. However, FPP is also known (in
1
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 58
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
human FPPS) to bind to the allosteric (i.e. non-catalytic) FPPS site and acts as an FPPS inhibitor,
and it is possible that FPP-analogs may also bind in this way.
The most active HepPPS, OPPS as well as cell growth inhibitors have very similar overall lengths,
corresponding to ~ 13±1 contiguous “heavy” atoms appended to the bisphosphonate C1 backbone
carbon. This length corresponds closely to the length of the FPP substrate of HepPPS (and OPPS),
as illustrated in Figure 5, suggesting that the most potent inhibitors target a long-chain
prenyltransferase, in cells. However, it also seemed possible that FPPS could be inhibited by 74.
This might seem a surprising suggestion since FPP is the product of FPPS and not a substrate,
and would presumably be too big to bind to the GPP substrate-binding site. However, in recent
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
1
work, Park et al. have shown that FPP binds to the allosteric site in FPPS discovered by Jahnke
22
et al. , suggesting that (long-chain) FPP isosteres could also bind to this site and inhibit FPPS.
We were unable to obtain active SaFPPS (from either of two commercial entities) to test this
hypothesis, so we next tested for human FPPS inhibition by 74, finding an IC =230 nM, which
50
is remarkably similar to that found against HepPPS (100 nM), as well as the 100 nM for 6 against
2
0
HsFPPS. This IC value for FPPS inhibition by 74 is also very similar to that we obtain for the
50
potent FPPS inhibitor zoledronate (24, IC = 250 nM) in the same assay. In summary then, we
50
find that 74 is a potent inhibitor of HepPPS as well as FPPS, suggesting that its activity in bacteria
could be due to inhibition of both quinone as well as cell wall biosynthesis.
Mechanisms of Action: Quinones, Cell Walls, and Clostridioides difficile. Based on
the results discussed above, we hypothesized that 74 (chosen over e.g. 75, since 74 exhibited low
human cell toxicity) targets both HepPPS as well as FPPS. We tested this hypothesis in five ways.
First, if HepPPS inhibition (or that of the corresponding enzymes in Bacilli spp.) is important,
then it should be possible to effect a rescue (or partial rescue, if multiple targets are involved) by
growing cells in the presence of a menaquinone (MK-4), an approach that we and others have
1
4
ACS Paragon Plus Environment

Page 15 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
used previously.23-24 Second, if FPPS inhibition is important, it should be possible to effect a
growth inhibition rescue by growing cells in the presence of FPP or undecaprenyl phosphate (UP).
This is because FPPS inhibition is expected to block both quinone as well as cell wall biosynthesis
(
Figure 1), so UP should partially rescue cell growth inhibition, as seen for example with the
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
inhibition of UPPS by clomiphene.25 An FPP rescue should also be quite potent though more
difficult to assess given its much less hydrophobic nature and hence, decreased ability to enter
cells. Third, we hypothesized that a combination MK-4 plus UP rescue might be particularly
effective. Fourth, we hypothesized that if FPPS inhibition is important, it should be possible to
inhibit the growth of bacteria that lack quinones (and HepPPS) because cell wall biosynthesis
would still be targeted. Here, pathogens such as C. difficile are ideal for testing this hypothesis
since they lack all quinone biosynthesis machinery. Moreover, they also lack heme biosynthesis
enzymes (which involve heme prenylation), another potential but heretofore unexplored area of
prenylsynthase inhibition. Fifth, we hypothesized that an “upstream” target in the DXP pathway
might be involved, so we used the IPP/DMAPP precursor 1-hydroxy-2-methyl-but-2-enyl 4-
diphosphate (HMBPP) to test this hypothesis. While this would not rule out inhibition of IspH,
we previously reported that IspH is not inhibited by bisphosphonates.26 Overall, there are,
therefore, 10 sets of “rescue experiments” to perform: 1) HMBPP; 2) FPP; 3) UP; 4) MK-4; 5)
HMBPP + FPP; 6) HMBPP + UP; 7) HMBPP + MK-4; 8) FPP + UP; 9) FPP + MK-4, and 10) UP
+
MK-4 and we carried each of these out using a 50 µM “rescue agent” concentration. This
concentration is of course somewhat arbitrary but was chosen based on previous work with
farnesol (200 M) in gram-negative bacteria;13 MK-4 (at 10, 100 and 1000 M) in M.
tuberculosis24 and FPP (at ~ 30 M) in Trypanosoma brucei rhodesiense.27 So 50 M for all
agents seemed reasonable and indeed, rescue effects (without toxicity) was generally observed.
We also carried out additional experiments with decaprenyl phosphate (DP), but the results were
the same as with UP (data not shown). We chose MK-4 (rather than MK-7) for most experiments
since it is more soluble, but the results with MK-7 were the same (data not shown).
1
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 58
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
We first added menaquinone (MK-4) to B. subtilis cells and measured growth in the
presence of different concentrations of 74 to determine if this quinone “rescued” cells from the
growth-inhibitory effects of 74. Results are shown in Figure 6a,d and indicate that cell growth
inhibition is indeed rescued by MK-4, the ED value for cell growth inhibition by 74 increasing
50
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
by a factor of ~5. This quinone-rescue effect is consistent with 74 inhibiting quinone biosynthesis
via HepPPS and/or FPPS. In a second experiment, we added undecaprenyl
Figure 6. Effects of MK-4, UP, HMBPP and FPP as well as pairwise combinations on B. subtilis
growth inhibition by 74. (a) Effects of MK-4, UP and MK-4 plus UP on growth inhibition by 74.
(
b) Effects of HMBPP, FPP with or without MK-4 or UP as well as HMBPP+FPP on 74 inhibition
of B. subtilis cell growth. All rescue agents were at 50 µM. (c) Schematic illustration of the x-fold
rescues by compounds or pairs of compounds on B. subtilis growth inhibition by 74. The largest
effect (25x) is found with UP (50 µM) + MK-4 (50 µM). Cell growth inhibition assays were carried
out in duplicate. The largest effects (~>9x) are seen with MK-4 or UP combinations and are
colored cyan. d) Summary of IC values from data in a) and b).
50
1
6
ACS Paragon Plus Environment

Page 17 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
phosphate which is also expected to (partially) rescue cells from FPPS inhibition. There was, once
again, a large (~7x) increase in the ED value for cell growth inhibition, consistent with FPPS as
50
a second target. Also, addition of MK-4 plus UP resulted in a ~25x increase in IC , Figure 6a.
50
That is, 74 is a multi-target inhibitor that inhibits both HepPPS as well as FPPS. We also found
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
that addition of FPP (at 50 µM) had, as expected, a significant effect on growth inhibition with a
3
.5x increase in ED , Figures 6 b-d. This increase is, perhaps, smaller than might be expected for
50
a compound that should rescue both cell wall as well as quinone biosynthesis. However, both MK-
and UP are for more hydrophobic than FPP and are likely to concentrate in the cell membrane
4
which is where they actually function (in electron transfer or in peptidoglycan biosynthesis) while
the more polar FPP needs to enter into cells as well as be processed (with HepPPS and UPPS),
and its efficacy is evidently less than with MK-4 or UP. The combinations FPP+UP and FPP+MK-4
had larger effects (8.9x and 11x, respectively) than either agent alone, though again smaller than
when using UP+MK-4 (~25x). It is also possible that there might be other targets for 74 such as
UPPS or UPPP, or a target upstream of FPPS. However, we found no inhibition 0f UPPS, and
UPPP inhibition was weak, ~66 M. Upstream inhibition also seems less likely since there was
only a ~2x increase in ED50 with HMBPP, Figures 6b-d. Interestingly, HMBPP potentiated the
rescue effect of MK-4 (5.1x → 8.9x) and FPP (3.5x → 6.4x), probably due to a “law of mass action”
effect (due to increased substrate concentration). However, with UP, the rescue effect was
diminished (7.2x → 4.2x). Reasons are not known but might be related to decreased UP uptake
into cells if transporters are involved.
Another target possibility for 74 is might be MenA, but this seems unlikely since in
4
previous work on S. aureus inhibition by analogs of 75 we found a good correlation (r = 0.91)
between SaHepPPS inhibition and cell growth inhibition, but only weak (~10 µM) MenA
4
inhibition . Nevertheless, to further test the hypothesis of FPPS+HepPPS targeting, it appeared
to be of interest to investigate the extent of growth inhibition of a bacterium that does not utilize
1
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 58
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
quinones in electron transfer/ATP biosynthesis, since then there would be no HepPPS (or MenA)
to inhibit. One such organism is C. difficile which utilizes a complex of ferredoxins and
flavoproteins to generate a Z gradient that drives ATP synthesis via a Na -ATP synthase.28-29
The organism lacks long-chain prenyl synthases such as HepPPS or OPPS, as well as downstream
quinone biosynthesis enzymes like MenA or UbiA, and has just a single trans-prenyltransferase,
FPPS, which is used in cell wall biosynthesis. Moreover, C. difficile lacks heme O synthase, the
prenyltransferase28 that farnesylates heme B in many bacteria, which is then oxidized to heme
+
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
0
A. We thus tested a series of known prenyl synthase inhibitors (33, 68-71, 74, 92, 95, 96, 98,
99) as well as two putative pro-drug forms (100, 101) against two strains of C. difficile (ATCC
4
3255 and ATCC 1870). Compound 100 is putative pro-drug form of 71, a potent anti-bacterial
11
targeting UPPS and UPPP , but these lipophilic benzoic acids bind to serum albumin and we
reasoned that a prodrug form might have lower serum binding and perhaps, better cell
penetration. Compound 101 is Piv-ester prodrug form of zoledronate (24), a very potent FPPS

inhibitor, in which one phosphonate PO group is masked and again we reasoned that such a
compound might have better cell penetration than 24. Results are shown in Table S3. The
compound with the most potent activity against both strains was 74 with an MIC of 6.8 µg/mL
and since all the other potential targets discussed above are absent, this strongly indicates that
FPPS is the major target for 74 in C. difficile. So, when all of the results described above are
considered, we conclude that HepPPS and FPPS are major targets for lipophilic bisphosphonates
such as 74 in the pathogens B. anthracis Sterne and S. aureus , and in B. subtilis, while only
FPPS is targeted in C. difficile.
As to the bisphosphonate features that contribute to activity (or inactivity) in gram-
positive and gram-negative bacteria: it appears that for activity against gram-negative bacteria,
FPPS is a major target, as reflected in the ~100-200 nM activity of 72 against e.g. EcFPPS and P.
aeruginosa FPPS , and by increased EC values in cells that overexpress FPPS.13 Such
1
3
50
compounds (e.g. 72, 73) have an aryl-alkyl-aryl-bisphosphonate motif, but poor activity against
1
8
ACS Paragon Plus Environment

Page 19 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
13
gram-positive bacteria. We suggest this is due at least in part to an uptake/efflux effect since
based on structure/sequence similarities, there are no obvious differences between e.g. EcFPPS
and SaFPPS X-ray structures. With the alkyl-aryl bisphosphonates such as 6 and 74, the opposite
trend is seen. That is, these compounds inhibit gram-positive but not gram-negative bacteria, and
there is good evidence for inhibition of HepPPS as well as FPPS. The primary difference between
the two inhibitor types is, then, that the new compounds lack terminal aryl groups, which appear
to be required for gram-negative activity.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Serum binding and human cell toxicity. Next, we sought to determine to what extent the
antibacterial effects of some of the more active bisphosphonates might be compromised by
binding to serum proteins, as well as whether there might be potent human cell toxicity. We
investigated five new compounds (74, 76, 77, 80 and 84), as well as one compound reported
1
3
previously (72) that had activity in bacterial (gram-negative) cell growth. The effects of bovine
serum on B. subtilis cell growth are shown in Table 3 and Figure S4, and effects on a human
embryonic kidney cell line (HEK293) are shown in Table 3 and Figure S5. The six compounds
investigated have four distinct structural motifs, illustrated for clarity (together with ED values)
50
in Table 3. First, 74 and 77 have reduced geranyl (tetrahydrogeranyl) groups linked via O (74) or
CH (77) to a meta-pyridinium bisphosphonate. Second, 84 and 80 have the same substituents
2
but attached para to the bisphosphonate moiety. Third, 76 has an n-undecyl group (lacking any
methyl substituents) located at the meta position. Fourth, 72 contains the quite different aryl-
alkyl-aryl motif (Table 3) and is of interest since, as noted above, it is a potent inhibitor of gram-
negative bacterial cell growth.
1
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 58
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 3. Effects of serum binding on B. subtilis growth inhibition, and toxicity to HEK293 cells.
Cpd
#
B. subtilis
HEK293
Structure
Substituent type
-
1
(µg mL )
ED50 (µM)
-
Serum>
10%
Serum
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
24
hrs
96
hrs
+
m-tetrahydrogeranyl-O
1.0→1.7
620
460
76
7
4
m-tetrahydrogeranyl-C
p-tetrahydrogeranyl-O
p-tetrahydrogeranyl-C
2.4→1.7
5.7→28
2.3→78
140
7
7
>100
800
84
0
700
220
>1000
7.3
8
0
76
m-n-undecyl
2.2→2.0
26→38
phenyl-alkyl-aryl
600
13
72
As can be seen in Table 3 and Figure S4, the addition of (4% or 10%) fetal bovine serum
had little effect on bacterial cell growth with 74. Also, we see that toxicity in the HEK293 cell line
is extremely low after a 24 hr incubation (620 µM) and is 76 µM at 96 hours, Figure S5. The results
with the methylene analog (77) are similar (serum no effect; HEK293, 460 µM at 24 hrs, 140 µM
2
0
ACS Paragon Plus Environment

Page 21 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
at 96 hours). Results with the para analogs (84, 80) are different. With both compounds we find
that there are quite large increases (4.9x, 34x) in ED at 10% serum and with both compounds
50
there is almost no inhibition of HEK293 cell growth (IC50 values ~1 mM, Table 3). This could
mean that 84, 80 simply bind tightly to serum in the B. subtilis assay, and to serum in the
HEK293 cell growth inhibition assay, blocking bisphosphonate activity. Alternatively, it might be
that 84/80 are simply weak inhibitors of human FPPS (the bisphosphonate target in human cells)
since as noted previously, 74 is a good FPPS inhibitor (IC50 ~230 nM, in our assay—about the
same strength as the commercial drug zoledronate, 24). We thus tested 84 against an expressed
HsFPPS finding an IC = 100 µM. It thus appears that the lack of activity of 84 (and by analogy,
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
50
80) against human cells is not due to uptake/serum binding but rather, to the inefficiency of
para-substituted bisphosphonates in inhibiting isoprenoid (FPP) biosynthesis.
With the meta n-alkyl substituted (n-undecyloxy) compound 76, there is good activity
against B. subtilis, and serum binding has no effect on bacterial cell growth (Table 3). However,
76 is a very potent inhibitor of HEK293 cell growth (Table 3)—at least at 96 hours (the ED50
decreases from 220 µM at 24 hours to 7.3 µM at 96 hours, Table 3), making it of less interest as
an antibacterial lead. We also investigated the aryl-alkyl-aryl species 72 which we previously
1
3
found was a good inhibitor of gram-negative bacterial cell growth. In the B. subtilis assay there
was only a small effect of 10% serum (a 50% increase in ED ), but we found strong toxicity (13
50
µM) against HEK293 cells, at least after a 96 hour incubation, Table 3. Thus, of all of the
compounds investigated, 74 has, overall, the most potent activity as well as low serum binding
and low toxicity.
In the future, it will be of interest to probe in more detail the molecular basis for the
toxicity of compounds such as 76 since this might facilitate the design of better antibacterials. For
example, is the decreased toxicity of 74 due to the presence of the two methyl groups? Are there
other isoprenyl-like side-chain substituents that have better antibacterial, plasma binding and
toxicity profiles? For example, are unsaturated side-chains, such as geranyl (versus saturated
2
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 58
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
species) more promising? Interestingly, we find that while the tetrahydrogeranyl meta-
substituted species 102 (Figure 2) is a good (1.9 µg/mL) B. subtilis growth inhibitor, the
(
unsaturated) geranyl analog 103 is not (ED ~9.2 µg/mL), indicating that side-chain flexibility
50
is desirable for activity, facilitating perhaps, binding to multiple targets.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Crystallographic investigations. There are currently no reported X-ray structures of very
long-chain trans-prenyl diphosphate synthases such as SaHepPPS (or EcOPPS) with bound
inhibitors. The only exception is, arguably, that of the S-thiolo analog of FPP (FSPP) bound to
EcOPPS (PDB ID code 3WJN), since FSPP is frequently used as a non-reactive FPP substrate-
analog. We thus sought to investigate how inhibitors might bind to HepPPS and OPPS, but we
were unable to obtain structures with any the most potent inhibitors. We did, however, obtain
structures of the bisphosphonate 69 (PDB ID code 5ZLF) as well as the salicylate 70 (PDB ID
code 5ZE6), a diphosphate isostere, bound to EcOPPS, and of 70 bound to EcUPPS (PDB ID code
5ZHE), an important antibacterial target. Full data acquisition and refinement details are given
in Table S4, and in the Experimental Section. Activity against EcOPPS was weak, ~330 M for 69
and ~450 M for 70, but the structures are still of interest since they do represent the first
structures of any very long-chain bacterial prenyltransferase with bound inhibitors. Plus, both 69
and 70 are 1-2 M HepPPS inhibitors (Figure 4).
EcOPPS has structural similarity to FPPSs (both bacterial and human) as well as GGPPS
and SaHepPPS, where the C root mean square deviation (rmsd) between EcOPPS (PDB ID code
3
WJN) and SaHepPPS (PDB ID code 5H9D) is 1.61 Å over 280 aligned residues. The C rmsd for
EcOPPS with 69 bound (PDB ID code 5ZLF) is 2.38 Å over 244 residues versus Saccharomyces
cerevisae GGPPS with 69 bound (ScGGPPS; PDB ID code 2E93). In GGPPS, there are four sites
to which inhibitor/substrate/product groups can bind: the allylic substrate
diphosphate/bisphosphonate inhibitor site a; the allylic substrate side-chain site b; the
2
2
ACS Paragon Plus Environment

Page 23 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
homoallylic diphosphate-binding site c, and a hydrophobic product (GGPP) sidechain site d.
Ligands can thus bind as follows: ab; ad; bc or cd. With 69 bound to GGPPS, Figures 7a,b two
1
2
molecules bind, one to site ab and one to site cd. We show in Figures 7a,b alignments of the new
EcOPPS/69 structure (yellow; PDB ID code 5ZLF) with that of ScGGPPS containing two 69
bisphosphonate ligands (blue; PDB ID code 2E93). The single bisphosphonate inhibitor in the
OPPS structure binds to the same allylic binding site ab as does 69 in ScGGPPS and has a very
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
+
similar conformation, Figure 7b. As expected, the bisphosphonate binds to the two Mg that are
coordinated by two aspartates in the first aspartate-rich domain. The third Mg2+ seen in some
GGPPS and essentially all FPPS structures is absent due, perhaps, to pH/crystallization
conditions. There is no occupancy of the second 69 binding site that is found in ScGGPPS. The
ab site in EcOPPS corresponds to the FPP-binding site found in the X-ray structure of EcOPPS
with FSPP, the S-thiolo analog of FPP, Figure 7c, and as can be seen in Figure 7d, the FSPP (blue)
and 69 (yellow) inhibitors basically overlap each other. And while we do not yet have the structure
of 74 bound to either EcOPPS or SaHepPPS, the 69 ligand-bound EcOPPS structure is
remarkably similar to that of the lipophilic bisphosphonate 104 (Figure 2), bound to the
bifunctional Plasmodium vivax farnesyl/geranylgeranyl diphosphate synthase (PvGGPPS; PBD
ID code 3RBM), Figures 8e,f, strongly suggesting that 74 will bind in a similar manner in
SaHepPPS and EcOPPS.
2
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 58
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 7. Structure of 69 bound to EcOPPS and a comparison with S. cerevisae GGPPS and P.
vivax F/GGPPPS ligand-bound structures. (a) Structure superimposition EcOPPS•69 (PDB ID
code 5ZLF) with ScGGPPS•69 (PDB ID code 2E93) blue = OPPS; yellow = GGPPS. (b)
Superimposition of 69 ligands shown in a) pink = GGPPS; green = OPPS. (c) Superimposition of
EcOPPS•69 with EcOPPS•FSPP (PDB ID code 3WJN). Pink = OPPS; cyan = FSPP. (d)
Superimposition of 69 (yellow) and FSPP (cyan) ligands from (c). (e) Superimposition of
EcOPPS•69 (orange) with P. vivax—F/GGPPS•105 (color; PDB ID code 3RBM). (f)
Superimposition of 69 (brown) and 104 (blue) ligands from (e). 69 binds to just the allylic site
2
+
(
ab) in EcOPPS) and coordinates to 1 Mg .
2
4
ACS Paragon Plus Environment

Page 25 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 8. Structure of 70 and FSPP bound to EcOPPS. (a) Superimposition of EcOPPS•70 (blue;
Chain B; PDB ID code 5ZE6) with EcOPPS•FSPP (yellow; PDB ID code 3WJN). (b) Illustration of
70 penetrating the dimer interface in EcOPPS•70. The ligand (cyan) is only present in Chain B
and is close to the monomer surface, but is buried in the dimer interface.
The second inhibitor of interest is 70, a lipophilic salicylate. While not a bisphosphonate,
the salicylates are bisphosphonate/diphosphate isosteres and inhibit prenyl diphosphate
synthases with 70 having been shown to inhibit both (gram-positive) E. faecalis and E. faecium
31
cell growth with MIC values of 4 µg/mL and here, we found activity against SaHepPPS with an
IC of 1.5 M (Figure 4), and a 19 M IC for S. aureus cell growth inhibition. With this lipophilic
50
50
salicylate, we again find that the inhibitor binds solely to the OPPS allylic binding site, as shown
in Figure 8a (PDB ID code 5ZE6), and it closely aligns with FSPP, the allylic substrate-like analog,
Figure 8a. So, the bisphosphonate and salicylate inhibitors bind to the allylic substrate-binding
site, with both bisphosphonate and salicylate groups being isosteres for diphosphate. With 70,
the inhibitor appears to bind with the terminal part of its side-chain in a highly solvent-exposed
positionif just the monomer structure is viewed, Figure 8a. However, the side-chain is actually
located in the OPPS dimer-interface and it partially penetrates the second subunit in the dimer,
2
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 58
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Figure 8b, so is protected from solvent exposure. These results in and of themselves do not explain
why 70 is a good (~1.5 µM) SaHepPPS inhibitor but a poor EcOPPS inhibitor. They do, however,
suggest the possibility that the difference in activity might be due to 70 binding to the regulatory
subunit in SaHepPPS. This is because in earlier work using atomic force microscopy, Suzuki et
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
2
4
al. discovered that FPP binds to the regulatory domain in B. subtilis HepPPS, and in our group
we found that at high concentrations, FPP inhibits SaHepPPS. It thus appears that binding to the
essential for activity) regulatory subunit in both BsHepPPS as well as SaHepPPS by FPP-isosteres
(
such as 70 could play a role in their inhibitory activity.
In addition to the OPPS/70 structure, we were able to obtain the structure of 70 bound to
UPPS (Table 3 and PDB ID code 5ZHE). Surprisingly, unlike the two structures discussed above
in which the inhibitors bind to substrate-binding sites, as can be seen in Figure 9a, 70 binds into
the hydrophobic center of the protein and not to the more polar, substrate binding sites. This is
33
(
approximately) the binding pose adopted by two other inhibitors, a pyrazole , Figure 9a,b, and
25
clomiphene, Figure 9c,d . This preference for a hydrophobic binding site by 70 is also seen when
comparing the 70 and 69 UPPS structure (PDB 1D code 2E98) where in UPPS there are four 69
ligands. The salicylate overlaps the hydrophobic 69 side-chains in sites 1 and 4 with the long alkyl
side-chain located throughout the “center” of the protein, Figures 9e,f. What is of interest about
the 70/UPPS structure is that the ligand pose is very different to that we reported previously with
1
05 which has a 2-carbon shorter side-chain in which two 105 ligands (not one) bind to UPPS, as
shown (in slate) in Figures 9g,h. Apparently, the longer (C alkyloxy) side-chain in 70 does not
14
permit binding of two molecules to UPPS, while the shorter (C ) species 105 does enable such
1
2
binding, due to decreased steric clashes.
2
6
ACS Paragon Plus Environment

Page 27 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 9. Structure of 70 bound to EcUPPS together with comparisons with pyrazole,
clomiphene, 69 and 105 bound structures. (a) Structure of EcUPPS•70 (pink; PDB ID code 5ZHE)
superimposed on a UPPS•pyrazole [N-(3-amino-3-oxopropyl)-5-(benzo[b]thiophen-5-yl)-1-
benzyl-N-(4-isopropoxybenzyl)-1H-pyrazol e-4-carboxamide] (gold; PDB ID code 5KH5)
structure (from S. pneumoniae). (b) Zoomed-in view of the ligand-binding region in (a) 70 in
cyan. (c) Superimposition of EcUPPS•70 (blue) with EcUPPS•clomiphene (not all clomiphene
atoms were resolved; green; PDB ID code 5CGJ). (d) Zoomed-in view of (c) 70 in yellow. (e)
2
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 58
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Superimposition of EcUPPS•70 (blue) with EcUPPS•69 (cyan; PDB ID code 2E98). (f) Zoomed-
in view of (e) 70 in pink. (g) Superimposition of EcUPPS•70 (green) with EcUPPS•105 (pink;
PDB ID code 3SH0). (h) Zoomed-in view of (g).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
As to future structural work: The structure of the EcOPPS/69 complex is of interest since
6
9 is a potent (~250 nM) UPPS inhibitor that we find here has activity against S. aureus (ED =
50
1
9 µM), so future studies of HepPPS inhibition (IC = 1.1 M, Figure 4) with additional analogs
50
will be of interest since it may be possible to develop dual UPPS/HepPPS inhibitors. Also, of
course, it will be of great interest to see whether the long chain bisphosphonates bind to the
allosteric site in HsFPPS, of interest in the context of developing FPPS/GGPPS inhibitors
34
35
targeting prenylation in tumor cells, and as vaccine adjuvants. Plus, whether the allosteric site
is in fact present in bacterial FPPSs is worth further investigation.
Conclusions and Future Prospects. The results we have presented here are of interest for
several reasons. First, after screening a library of known prenyltransferase inhibitors for activity
against the long chain prenyltransferase octaprenyl diphosphate synthase (OPPS), we tested
several compounds against a variety of gram-positive and gram-negative bacteria, as well as the
fungus C. albicans. We found promising activity against the gram-positive bacteria, so we next
synthesized and tested 24 analogs of the most potent hit (in both enzyme and cell growth
inhibition assays). We found several compounds that inhibited OPPS as well as its analog,
HepPPS, that also inhibited gram-positive (but not gram-negative) bacterial cell growth in the 
1
-10 µg/mL range. The generally most potent hit or lead was 74, a tetrahydrogeranyl-substituted
pyridinium bisphosphonate. Second, we found that the activity of 74 (and close analogs) was not
inhibited by human serum, plus, there was low toxicity against the human embryonic kidney cell
line, HEK293 (~0.5 mM after 24 hrs, ~70 µM at 96 hours). The potent activity against gram-
positive bacteria was in sharp contrast to that we reported earlier with bisphosphonates such as
2
8
ACS Paragon Plus Environment

Page 29 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
7
2 and 73 which only inhibited gram-negative bacterial cell growth, perhaps due to differences in
bacterial influx/efflux. Third, we investigated the mechanism of action of 74. Compound 74
inhibited both EcOPPS as well as SaHepPPS with ~10-100 nM IC values, as well as a (human)
50
FPPS, at ~200 nM. These results suggested the possibility that 74 (and related compounds) might
act by inhibiting both short (FPPS) as well as long-chain (HepPPS) isoprenoid diphosphate
synthase enzymes, in bacteria. To test this hypothesis we used a “rescue” approach in which we
added downstream products FPP, menaquinone-4 (MK-4) and undecaprenyl diphosphate (UP)
to cells, grown in the presence of 74. There were large increases in the ED50 for cell growth
inhibition by MK-4 and UP and when added together, the ED50 for cell (B. subtilis) growth
inhibition increased by a factor of 25. Addition of FPP itself also had an effect, ~3.5x, since it
“rescues” FPPS inhibition. Addition of the metabolite HMBPP (E-1-hydroxy-2-methyl-but-2-enyl
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
-diphosphate), which is upstream of FPP, also had a small-effect and potentiated the rescue
effects of MK-4 and UP. One obvious explanation for these observations is that addition of the
very hydrophobic species MK-4 and UP is extremely effective in rescuing cells from inhibition of
their biosynthesis simply because being so hydrophobic (MK-4, clogP = 8.48; UP, clogP = 17.32),
they rapidly bind to bacterial cell membranes—which is where they normally function. On the
other hand, while addition of FPP might be expected to help rescue both quinone as well as cell
wall biosynthesis, it is not particularly effective because not only does it have to be transported
into the cytoplasm, it also has to be processed by UPPS and HepPPS, and HepPPS is actually
inhibited by e.g. 74. Fourth, we solved the structures of three prenyl-transferase inhibitor
complexes. The structure of the bisphosphonate 69 bound to OPPS showed that it bound to the
(allylic) substrate-binding site much better, as did the salicylate inhibitor 70. However, 70 was a
surprisingly bad (several hundred-fold) HepPPS inhibitor than an OPPS inhibitor and we propose
that the salicylate (70), a diphosphate isostere, binds to the regulatory subunit because in
previous work we found that FPP inhibits HepPP product formation at high concentrations (Kd
~
35 µM) in SaHepPPS, and other workers have shown that FPP binds to the regulatory subunit,
2
9
ACS Paragon Plus Environment