Bisphosphonate Inhibitors of Mammalian Glycolytic Aldolase

Article abstract of DOI:10.1021/acs.jmedchem.8b01000

The glycolytic enzyme aldolase is an emerging drug target in diseases such as cancer and protozoan infections which are dependent on a hyperglycolytic phenotype to synthesize adenosine 5′-triphosphate and metabolic precursors for biomass production. To da

Full text of DOI:10.1021/acs.jmedchem.8b01000

Subscriber access provided by CHAPMAN UNIV
Article
Bisphosphonate inhibitors of mammalian glycolytic aldolase
Paul Heron, Marta Abellán-Flos, Laurent Salmon, and Jurgen Sygusch
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b01000 • Publication Date (Web): 12 Nov 2018
Downloaded from http://pubs.acs.org on November 13, 2018
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 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Bisphosphonate inhibitors of mammalian glycolytic
aldolase
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
§
†
†
§
Paul W. Heron , Marta Abellán-Flos , Laurent Salmon , Jurgen Sygusch*
*§ Département de biochimie et médecine moléculaire, Université de Montréal. CP 6128,
Succursale Centre-Ville, Montréal, Québec, Canada H3C 3J7
†
Equipe de Chimie Bioorganique et Bioinorganique, Institut de Chimie Moléculaire et des
Matériaux d’Orsay (ICMMO), Univ Paris-Saclay, Univ Paris-Sud, CNRS UMR8182, LabEx
LERMIT, rue du doyen Georges Poitou, F-91405 Orsay, France
ACS Paragon Plus Environment
1

Journal of Medicinal Chemistry
Page 2 of 59
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
The glycolytic enzyme aldolase is an emerging drug target in diseases such as cancer and
protozoan infections which are dependent on a hyperglycolytic phenotype to synthesize ATP and
metabolic precursors for biomass production. To date, structural information for the enzyme in
complex with phosphate-derived inhibitors has been lacking. Thus, we determined the crystal
structure of mammalian aldolase in complex with naphthalene 2,6-bisphosphate (1) that served as
a template for the design of bisphosphonate-based inhibitors, namely 2-phosphate-naphthalene 6-
bisphosphonate (2), 2-naphthol 6-bisphosphonate (3), and 1-phosphate-benzene 4-bisphosphonate
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). All inhibitors targeted the active site and the most promising lead, 2, exhibited slow-binding
inhibition with an overall inhibition constant of ~38 nM. Compound 2 inhibited proliferation of
HeLa cancer cells while HEK293 cells expressing a normal phenotype were not inhibited. The
crystal structures delineated the essential features of high-affinity phosphate-derived inhibitors and
provide a template for the development of inhibitors with prophylaxis potential.
ACS Paragon Plus Environment
2

Page 3 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
INTRODUCTION
Glucose metabolism in cancer cells is characterized by a marked increase in both glucose uptake
and aerobic glycolysis, the fermentation of glucose into lactate in the presence of oxygen.
Enhanced glucose uptake in cancer cells is now exploited clinically for diagnostic purposes with
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
8
positron emission tomography (PET) where radiolabeled glucose ( fluoro-deoxyglucose) is used
1
as a tracer to assay for accumulation in tumors. The latter effect of increased fermentation has
been known for several decades, and was first proposed by German physiologist and Nobel
laureate, Otto Warburg, who, in the 1920s showed that tumor cells produce higher levels of lactate
2
compared to normal tissues, even in the presence of oxygen. Hyperglycolysis not only produces
ATP at a faster rate than oxidative phosphorylation, but also provides metabolic precursors for
biomass production (e.g. nucleotide, amino acid, and lipid biosynthesis), aiding in the rapid
3
proliferation of cancer cells. The term “Warburg effect” was coined and ongoing conceptual and
empirical advances over several decades have led to the acceptance of altered metabolism as one
4
the hallmarks of cancer. Tumor glycolysis is actively studied as a potential target for cancer
therapy, however viable clinical leads have proven elusive.5
A leading candidate is fructose-1,6-bisphosphate (FBP) aldolase (EC 4.1.2.13), a central enzyme
in glycolysis responsible for the reversible aldol reaction of fructose-1,6-bisphosphate to
glyceraldehyde-3-phosphate and dihydroxyacetone phosphate. Glycolytic aldolase is a Class I
enzyme, found in eukaryotes and higher plants, and characterized by the formation of a protonated
imine (Schiff base) with an active site lysine. Aldolase expression levels were pronounced when
investigated in several malignant cell lines such as human lung squamous carcinomas,6–8
hepatocellular carcinomas,9,10 pancreatic and colorectal cancers. Aldolase is already a target in
11
12
parasitic organisms such as Trypanosoma brucei (causative agent for African sleeping sickness)
ACS Paragon Plus Environment
3

Journal of Medicinal Chemistry
Page 4 of 59
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
for which glycolysis is the sole source of ATP production.13,14 Earlier work in the literature and
from our laboratory has focused on the development of class I aldolase inhibitors in such
organisms.14,15
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
Among these, phosphorylated substrate analogues of fructose-1,6-bisphosphate aldolases
possessing an aromatic moiety were of considerable interest due to their ability to strongly interact
with the active site. First intended for their interest in probing the nature of the active site, which
1
6
was assumed to be hydrophobic based on residues surrounding the reactive lysine, these aromatic
derivatives were among the most powerful competitive inhibitors of the rabbit muscle aldolase-
catalyzed reaction. Notably, these aromatic time-dependent reversible inhibitors (slow-binding)
1
7
capable of targeting active site lysine residues, namely 2-hydroxybenzaldehyde-4-phosphate and
18
1
-hydroxy-2-napthaldehyde-6-phosphate, inactivated aldolase and implicated a slow reversible
Schiff base mechanism. A naphthalenic derivative, naphthalene 2,6-bisphosphate (NA-P ) was
2
1
8
shown to be a potent competitive inhibitor with K = 0.28 μM. The mode of inhibition by which
i
these phosphorylated aromatic derivatives inhibited aldolase was established by enzyme kinetics,
UV/visible difference spectroscopy and site-directed mutagenesis,17 yet these studies did not
provide detailed structural insight, with the exception of computational modeling studies involving
1
-hydroxy-2-napthaldehyde-6-phosphate.18
Different types of bis-phosphonate derivatives were previously reported to act as highly effective
enzyme inhibitors. In the case of zinc-metallo enzymes such as matrix metalloproteinases and
carbonic anhydrase, arylamino-bisphosphonates were used as chelating agents of the active site
zinc cofactor, as well as mimics of the tetrahedral high-energy intermediate involved in the
1
9
catalyzed reaction. Other bisphosphonates derivatives were also reported as potent inhibitors of
2
+
20–22
the Mg -dependent farnesyl pyrophosphate synthase
and geranylgeranyl pyrophosphate
ACS Paragon Plus Environment
4

Page 5 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
synthase,20,23,24 acting both as metal binding agents and pyrophosphate mimics. In addition,
methylene bisphosphonate inhibitors were successfully employed as inhibitors of HIV reverse
transcriptase phosphorolytic activity.25
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
We sought to elucidate the interaction between several phosphorylated aromatic derivatives and
the aldolase active site to serve as a template for designing potent high-affinity phosphorylated
inhibitors. Here, using the previously known competitive inhibitor NA-P (1) as template, we
2
describe the binding of aromatic bisphosphonate substrate analogues of aldolase, namely 2-
phosphate-naphthalene 6-bisphosphonate (PNAB) (2), 2-naphthol 6-bisphosphonate (NAB) (3),
and 1-phosphate-benzene 4-bisphosphonate (PBB) (4) (Chart 1). The mode of binding was
determined by a combination of x-ray crystallography and enzyme inhibition kinetics. The high-
resolution crystal structures of these phosphorylated aromatic inhibitors in complex with aldolase
are described and should serve as a basis for the design of potent lead compounds.
Chart 1. Bisphosphonate Aromatic Analogues Based on FBP Aldolase Inhibitor 1a
OLi
O P OLi
O
ONa
ONa
OLi
OLi
O
P
P
O
P
P
NaO
NaO
LiO
LiO
O
O
O
O
1
3
2
4
OLi
O P OLi
OLi
O P OLi
OLi
OLi
P
O
P
P
OLi
LiO
LiO
OLi
O
O
HO
O
a
Fructose-1,6-bisphosphate (FBP) aldolase inhibitors: naphthalene 2,6-bisphosphate (1, sodium
salt, NA-P2),16,18 2-phosphate-naphthalene 6-bisphosphonate (2, lithium salt, PNAB), 2-naphthol
ACS Paragon Plus Environment
5

Journal of Medicinal Chemistry
Page 6 of 59
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
6
-bisphosphonate (3, lithium salt, NAB), and 1-phosphate-benzene 4-bisphosphonate (4, lithium
salt, PBB).
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
ACS Paragon Plus Environment
6

Page 7 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
RESULTS AND DISCUSSION
Structure of ALDOA-1 (NA-P ) complex. To investigate the binding mode of phosphorylated
2
aromatic compounds, rabbit muscle aldolase (ALDOA) crystals were soaked in crystallization
buffer containing compound 1. The rabbit isoform is routinely used for its ease of purification,
crystallization, and is a good model candidate for the human isoform because of its exceptionally
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
26
high homology (98% sequence identity; 100% sequence homology). The final crystal structure
was determined to 1.97 Å resolution (Rwork = 12.87%; R = 16.65%) and shows a tetramer in the
free
asymmetric unit of space group P2 , consistent with aldolase structures previously reported for
1
these crystallization conditions27 (see Experimental Section for details). Data collection and
refinement statistics for all crystal structures are shown in Table 4 in the Experimental Section. A
tetramer in the asymmetric unit cell gives the opportunity to make four independent observations,
and, expected from previous work, that the compound should bind in a 1:1 ratio of [inhibitor:
subunit].16
The overall topology of the active site shown in Figure 1A highlights the basic electrostatic
potential surface of the catalytic pocket nestled inside a TIM-barrel. The electron density
surrounding the bound inhibitor (Figure 1B) was unambiguous and allowed for confident modeling
into the active site of each tetramer subunit. The phosphate binding loci of 1 are homologous with
the P - and P -phosphate binding site of the substrate FBP (shown in Figure S1). A feature of the
1
6
FBP P -binding site was the electrostatic interaction implicating Lys107, described in high-
6
2
7
28
resolution structures and deduced from differential protection experiments. Attachment by 1
involves the same electrostatic interaction with Lys107 at the P -site as was found for the substrate,
6
and is consistent with the finding that a Lys107 variant (K107M) exhibits a reduced affinity (higher
1
8
K ) for 1.
i
ACS Paragon Plus Environment
7

Journal of Medicinal Chemistry
Page 8 of 59
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. Crystal structure of the ALDOA-1 complex showing syn and anti conformers of
compound 1. (A) Electrostatic surface potentials of rabbit muscle aldolase (ALDOA) calculated
using the Adaptive Poisson-Boltzmann Solver (APBS) software package with contour field of
1
2
± 5 kT/e. Acidic residues are colored red, basic residues are colored blue. (B) Difference
electron density calculated from a 1.97 Å simulated annealing F - F omit map encompassing
o
c
compound 1 and contoured at 3.0σ. Syn and anti- conformations of 1 are shown. (C) Hydrogen-
bonding network of 1 with active site residues and water molecules. Notable interactions are
those formed by compound 1 phosphate moieties in the P - and P - phosphate loci of FBP (P :
1
6
1
vicinal to Arg303; P : vicinal to Ser35).
6
The hydrogen-bonding network of this binding locus is similar to that observed in the Schiff
ACS Paragon Plus Environment
8

Page 9 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
27
base complex formed with FBP [PDB id: 1ZAI]. Notably, Lys107 interacts with two phosphate
oxyanions (contacts of 3.1 Å and 2.9 Å) of 1 and, similarly to FBP, Ser35 and Ser38 hydroxyl
groups contact the phosphate oxygens. However, the phosphate moiety of 1 is displaced with
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
respect to the FBP P -phosphate by 1 Å beyond the center of the active site cleft, consistent with
6
the longer intramolecular distance between phosphate moieties of 9.8 Å in 1 compared to 8.9 Å in
FBP. The flanking helical region (residues 34-65) undergoes a displacement upon binding by 1,
that is smaller by ~1 Å compared to FBP binding and fixes the bound phosphate at the P -site
6
(
Figure S1). Displacement by this flanking helix has been noted upon active site binding,29,27,30
and is responsible for narrowing the active site, functioning as a clamp to bind the FBP P6-
phosphate and, more generally, capable of accommodating binding by phosphate moieties from
phosphorylated analogues of varying sizes.
A surprising structural feature in the ALDOA-1 complex was the conformational mobility at the
P -locus shown by 1 binding, not observed with FBP or DHAP. Upon P -phosphate binding by
1
1
aldolase substrates, FBP or DHAP, Arg303 side chain typically undergoes a conformational
change which enables it to grasp and immobilize the incoming P -phosphate, by forming a strong
1
bidentate hydrogen-bond. In compound 1, an unexpected binding mode was observed consistent
with two configurations of the phosphate oxyanion at the P -site (Figure 1B): a major anti
1
conformation with respect to the orientation of the trans-annular phosphates across the aromatic
core of 1; and a minor syn conformation. Only the syn conformer forms the anticipated salt bridge
with Arg303. Population of the minor species corresponded to a refined occupancy of 0.35 ± 0.02
(
average of four subunits) compared to the anti conformer whose occupancy was 0.65 ± 0.02
Figure 1C - yellow).
(
The observation of two distinct binding modes for the ALDOA-1 complex was used as a
ACS Paragon Plus Environment
9

Journal of Medicinal Chemistry
Page 10 of 59
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
template to design a series of novel bisphosphonate analogues, with the premise that a
bisphosphonate moiety would be capable of mimicking both syn and anti phosphate positions
observed in the ALDOA-1 complex. A bisphosphonate moiety, which structurally resembles
pyrophosphate, is a compelling proposition for therapeutic applications as substitution of the
central oxygen atom (pyrophosphate) for carbon (bisphosphonate) procures added stability 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
3
1
resistance to thermal, chemical, and enzymatic degradation. Bisphosphonates are a class of drugs
3
2
that are primarily used for intervention in osteoporosis but are also being investigated because of
their inhibitory activity against several carbonic anhydrase isozymes, some of which are
overexpressed in hypoxic tumors.19,33 Here, we sought to exploit the stability of bisphosphonates
by substituting the phosphate group of 1 for a bisphosphonate, generating compound 2 (PNAB)
(Scheme 1). Further, to optimize targeting to the compound 1 binding site, additional topologies
were sampled. Notably compounds 3 (NAB) and 4 (PBB) were synthesized either by permutation
of the central naphthalene core to a benzene core (4) or by elimination of the second phosphate
group (3) (Scheme 1).
Scheme 1. Synthesis of inhibitors 2 4a
a) synthesis of inhibitor 2
OEt
O
O P OEt
OEt
H
a
P
b
OEt
O
HO
HO
5
OEt
O P OEt
OLi
O P OLi
OEt
OLi
O
P
P
c
O
P
P
EtO
EtO
OEt
LiO
LiO
OLi
O
O
O
O
6
2
b) synthesis of inhibitor 3
ACS Paragon Plus Environment
1
0

Page 11 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
OLi
O P OLi
OLi
c
P
5
OLi
O
HO
3
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
c) synthesis of inhibitor 4
OEt
O P OEt
O
OEt
H
a
P
b
OEt
O
HO
HO
7
OEt
O P OEt
OLi
O P OLi
OEt
OLi
O
EtO
P
P
c
O
P
P
OEt
LiO
LiO
OLi
O
O
EtO
O
O
8
4
a
34,35
Reagents and conditions: (a) (EtO) P(O)H, Na, rt, 48 h, 5 80%, 7
99%; (b): (EtO) P(O)Cl,
2
2
1
,4-diazabicyclo[2.2.2]octane, THF, rt, 48 h, 6 58%, 8 80%; (c): (i) TMSBr, 2,4,6-collidine,
CH Cl , rt, 16 h, (ii) LiOH, H O, 80 °C, 16 h, pH 10, 2 99%, 3 91%, 4 85%.
2
2
2
Inhibitors 2, 3, and 4 were prepared as represented in Scheme 1. Following reported
procedures,34,35 6-hydroxy-2-naphthaldehyde and 4-hydroxybenzaldehyde were treated with a
large excess of sodium diethylphosphite generated in situ to provide the respective
bisphosphonates 5 and 734,35 in a single step in 80% and 99% yields, respectively. The free
hydroxyls of bisphosphonates 5 and 7 were phosphorylated to yield 6 and 8, respectively, by
reaction with diethyl chlorophosphate in the presence of 1,4-diazabicyclo[2.2.2]octane (DABCO)
which serves both as catalyst and proton scavenger.36 Finally, compounds 6, 5, and 8 were
hydrolyzed under modified McKenna conditions37 by sequential reactions with
bromotrimethylsilane and 2,4,6-collidine, followed by basic treatment with 2 N LiOH, affording
ACS Paragon Plus Environment
1
1

Journal of Medicinal Chemistry
Page 12 of 59
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
inhibitors PNAB (2), NAB (3), and PBB (4) in 99%, 91%, and 85% yields, respectively. All
inhibitors and intermediates were characterized by HRMS and H, 13C, and 31P NMR
spectroscopies.
1
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
High-resolution crystal structures of bisphosphonate analogues in complex with ALDOA.
Native ALDOA crystals were soaked for 30 min in a solution consisting of mother liquor and a
given bisphosphonate compound at final concentrations of 1 mM, 5 mM, and 20 mM for
compounds 2, 3, and 4 respectively. Data collection and refinement statistics for the three
structures are shown in Table 4 in the Experimental Section. The statistics and final resolution are
indicative of the soaking conditions for each bisphosphonate: the higher concentrations used for
crystal soaking with 3 and 4 to ensure adequate active site occupancy slightly impacted crystal
quality, manifested by lower resolutions (2.20 Å and 2.29 Å respectively) and higher mosaicities
of their intensity profiles, compared to the 1.58 Å resolution obtained for compound 2. A control
dataset for the native crystal (not shown), diffracted to 1.63 Å. All three bisphosphonates targeted
the active site yet displayed surprisingly different modes of binding.
ALDOA-2 complex structure. Electron density delineating compound 2 (Figure 2A) was
unambiguous and allowed for confident determination of its binding mode. The average occupancy
of 2 in the four subunits was 0.78 ± 0.04. Soaking times of 2 with native crystals were limited due
to degradation in crystal quality at longer soaking times. Comparison of the ALDOA-2 complex
with the ALDOA-1 complex reveals an indistinguishable superposition of the inhibitors, notably
at the P -site (shown in Figure 3A). In the ALDOA-1 complex (Figure 1C), the phosphate forms
1
5
hydrogen bonds in either the syn or anti conformations whereas in the ALDOA-2 complex, the
bisphosphonate moiety of 2 engages in three additional hydrogen bonds with active site residues
forming a network of 8 hydrogen bonds (Figure 2) (not including water molecules).
ACS Paragon Plus Environment
1
2

Page 13 of 59
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. Crystal structure of the ALDOA-2 complex. (A) Difference electron density
calculated from a 1.58 Å simulated annealing F - F omit map encompassing compound 2 and
o
c
contoured at 3.0σ. (B) Close-up of hydrogen-bonding contacts between bisphosphonate moiety
and surrounding backbone atoms, side-chain atoms, and water molecules at the P -phosphate
1
locus. (C) Extensive hydrogen-bonding network (dashed lines) between 2 and active site
residues and water molecules is illustrated.
Inspection of the P -binding site is informative for comprehending the differences in affinity and
1
binding kinetics in compounds 1 and 2. Although ALDOA-1 and ALDOA-2 complexes appear
isostructural, a notable difference was observed in the hydrogen-bonding pattern with regards to
active site Ser271 (Figure 3A), where the residue which interacted with 1 no longer interacted
ACS Paragon Plus Environment
1
3

Journal of Medicinal Chemistry
Page 14 of 59
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
directly with 2 due to active site deformation by 2. Furthermore, attachment by both complexes
enabled determination of the protonation state of the phosphate and bisphosphonate moieties for
binding at the P locus that was based upon their interactions with surrounding backbone and side-
1
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
chain atoms. In the syn conformation of the ALDOA-1 complex (Figure 3B), the Gly272 backbone
carbonyl engages in a hydrogen-bond with phosphate O atom that is consistent with protonation
8
of the phosphate O oxyanion. Upon syn-anti rotational isomerization of the phosphate, the
8
proximity of O to Ser300 carbonyl implies hydrogen bond formation and also consistent with
8
protonation of the phosphate O oxyanion in the anti conformation. This hydrogen-bonding pattern
8
implies mono-protonation of the bound P phosphate oxyanion. In the ALDOA-2 complex, Ser271
1
shifts to the rear of the active site (by 1.7 Å compared to ALDOA-1) to accommodate bisphosphate
binding and is replaced by W1, which donates two hydrogen bonds to the bisphosphonate moiety
(Figure 3C – green dashes). W1 also accepts hydrogen bonds from surrounding active site atoms
(Gly302 N and Ser271 O ). For this binding geometry, O interacts with Ser300 carbonyl while
H
H
17
O interacts with Gly272 carbonyl indicating binding by 2 is consistent with protonation of each
2
1
phosphonate moiety (i.e. dibasic bisphosphonate). Displacement of Ser271 with respect to the
native enzyme reflects active plasticity in accommodating binding by compounds 1 and 2.
Prior to synthesis of compound 2, the naphthalenic derivative 1 represented the most potent
reversible competitive inhibitor of the rabbit muscle aldolase-catalyzed reactions. The design of
compound 1 was based on two chemical features useful for enhancing active site binding.
Protection experiments had shown that the active site binding embodied two distinct phosphate
binding sites: a higher affinity site – the P -site; and a lower affinity site – the P -site making
1
6
binding by compounds with suitable spanning phosphate groups to both sites highly specific.28,38
A second distinguishing feature of potent aldolase inhibitors (0.1 – 1.0 μM range) were those
ACS Paragon Plus Environment
1
4

Page 15 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
16
possessing aromatic moieties. The rigid structure imposed by the aromatic rings compared to a
flexible compound such as hexanediol 1,6-bisphosphate (K = 25 µM) suggests that a smaller
i
configurational entropy loss upon active site binding by a rigid aromatic compound may underpin the
3
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
observed differences in binding affinity to aldolase. The additional feature of a bisphosphonate moiety
in compound 2 took advantage of the cryptic phosphate binding site at the P site to enhance
1
binding affinity. Modifications were made of the bisphosphonate derivative 2 that were designed
to explore additional conformational space in the aldolase active site. First, the synergy between
the two phosphate loci was probed by replacing the phosphate group with a hydroxyl group (3),
also providing an indication of the significance of the bisphosphonate moiety towards potency.
Then, to assess the sufficiency of the naphthyl core in spanning the phosphate loci, a shorter
aromatic core was introduced - a phenyl group (4).
ALDOA-3 complex structure. The crystal structures of ALDOA-3 and ALDOA-4 are shown
in Figure 4. The binding mode of 3 and 4 are noticeably different than 2. First, their bisphosphonate
moieties do not bind at the P -site. The bisphosphonate in the ALDOA-3 structure (Figure 4A)
1
points towards the outside of the active site cavity and forms hydrogen bonds with the side chains
of surrounding residues (Ser45, Lys311, Arg303, Arg42). Further, the naphthyl moiety of 3 is
conjugated to Arg42 via a cation-π interaction, a highly stabilizing interaction.40,41 The distance
between the guanidinium group and the naphthalene ring in all four subunits is 3.48 ± 0.10 Å,
consistent with reported cation-π interactions for arginine.40,42 The interaction is further stabilized
by a salt bridge that forms between Arg42 and the bisphosphonate oxygens. Also, Arg303 side-
chain carbon atoms align on the distal sides of 3, producing a stabilizing hydrophobic environment
for the naphthyl moiety. The distal O group of 3 is anchored by a hydrogen bond with the Glu34
H
carboxyl.
ACS Paragon Plus Environment
1
5

Journal of Medicinal Chemistry
Page 16 of 59
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. Superposition of the ALDOA-1 and ALDOA-2 complexes and interactions at the P1-
site. (A) The structures of ALDOA complexed with 1 (green) and 2 (magenta) were superposed
in PyMol to illustrate the structural similarities of their binding modes (r.m.s.d. = 0.21 Å for
alignment of C atoms in subunit A). One notable difference was observed for Ser271, shown
α
in the inset. (B) Hydrogen-bonding network (grey dashes) of ALDOA-1 complex at the P1-
binding site is illustrated for the syn and anti conformers. The syn-anti rotational isomerization
is depicted by a green arc. The proximity of the Gly272 carbonyl to the phosphate O of syn-1
8
(2.89 ± 0.14 Å) implies protonation of the phosphate O and indicates proton donation by
8
Ser271 O (red dashes) to O (syn: 2.80 ± 0.08 Å; anti: 2.65 ± 0.20 Å). Further, proximity of
H
8
anti-1 phosphate O to the Ser300 carbonyl (3.00 ± 0.10 Å) is consistent with protonation of
8
O . (C) In the ALDOA-2 complex (magenta), the vicinal Gly272 and Ser300 backbone
8
carbonyls also indicate protonation of the partnered oxygens on the bisphosphonate (O – 2.76
2
1
±
0.06 Å; O – 2.82 ± 0.01 Å, respectively). Ser271 side-chain shifts to the rear, making room
17
for a water molecule (W1) that donates hydrogen bonds to O (3.22 ± 0.04 Å) and O (2.31 ±
17
21
0
.13 Å) (green dashes), and accepts hydrogen bonds from Ser271 side-chain O and Gly302
H
backbone NH.
ACS Paragon Plus Environment
1
6

Page 17 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
A second binding site for 3 was identified near the P -locus but electron density was only visible
6
for the bisphosphonate group (seen in Figure 4A), consistent with purity assessments showing no
indication of free bisphosphonate, and indicative of positional disorder of the naphthyl group at
the second binding site.
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
The hydrophobic pocket used by 3 is a recognized binding locus for aldolase-binding partners,
including the C-terminal peptide of the actin nucleation-promoting factor WASP and the LC4
domain of SNX9, a key regulator of endocytosis in cellular cargo transportation, which both
intercalate a tryptophan indole ring between Arg42 and Arg303.43,44 A novel naphthyl phosphate-
based inhibitor of aldolase (NASEP) identified from the WASP study was competitive for active
site binding and had a K of 0.1 mM. Comparison of the ALDOA-3 complex with the ALDOA-
i
o
NASEP complex (PDB id: 2OT1) (illustrated in Figure S2) revealed a difference of ~81 in the
angle of insertion of the naphthyl group with respect to the primary axis of the NASEP naphthyl
ring. The stabilizing cation-π interaction is also observed in the ALDOA-NASEP complex and
appears to be a driving force for binding at this locus but is insufficient alone as native ALDOA
crystals soaked in saturating tryptophan in the crystallization buffer (6 mM L-tryptophan) did not
4
3
reveal evidence of tryptophan active site binding. The ancillary hydrogen bonds formed between
inhibitor oxygens and basic active site residues are therefore required to promote active site
binding at this locus.
ALDOA-4 complex structure. The ALDOA-4 complex (Figure 4B) revealed a novel mode of
binding different from the previous bisphosphonate compounds. The bisphosphonate moiety of 4
is isostructural with the secondary binding site identified in the ALDOA-3 complex. This site is
located near the P -binding locus: one phosphonate group of 4 contacts several active site residues
6
(Ser35, Ser38 and Lys107); the second phosphonate is stabilized by hydrogen-bonding to Lys146.
ACS Paragon Plus Environment
1
7

Journal of Medicinal Chemistry
Page 18 of 59
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 phosphate group forms an electrostatic interaction with Arg303. To accommodate the binding
mode of 4 and avoid clashing with the benzene group, Arg42 folds into the hydrophobic pocket
described above for 3.
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
ACS Paragon Plus Environment
1
8

Page 19 of 59
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 4. Crystal structure of ALDOA in complex with compounds 3 and 4. (A) Difference
electron density calculated from a 2.20 Å 2mF – DF Feature Enhanced Map (FEM) showing
o
c
compound 3 and contoured at 1.0σ. Noteworthy is the cation-π interaction involving Arg42 and
the naphthalene moiety of compound 3 and the hydrophobic interaction with the side-chain of
Arg303, shown on the right-panel. (B) Difference electron density calculated from a 2.29 Å
2
mF – DF FEM comprising compound 4 and contoured at 1.0σ. Compound 4 binding mode
o c
is different from other bisphosphonates. The hydrogen-bonding networks shows the
bisphosphonate moiety contacting the P -phosphate binding locus involving Ser35 and Lys107.
6
ACS Paragon Plus Environment
1
9

Journal of Medicinal Chemistry
Page 20 of 59
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
Inhibition kinetics of the bisphosphonate inhibitors. The potency of the bisphosphonate
analogues 2, 3 and 4 were investigated by inhibition kinetics using previous buffer conditions (see
Experimental Section). Potency of 2 was evaluated using IC50 assays, summarized in Table 1
Indication that 2 exhibited slow-binding behavior arose from IC50 experiments with different pre-
incubation periods of the enzyme with 2 in solution. An IC50 value of 0.67 ± 0.07 μM was
measured with no pre-incubation period which decreased 10-fold to an IC50 of 0.060 ± 0.008 μM
with a pre-incubation of 90 min. To determine the nature of the slow-binding behavior by 2, time-
dependent inhibition by 2 was assessed using the method first described by Morrison and Walsh
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
5
(
see Experimental Section for details). Briefly, in the presence of classical competitive inhibitors,
product formation remains linear prior to substrate depletion. With slow-binding inhibitors,
inhibition exhibits a two phase relationship: an initial burst in product formation (v ) corresponding
o
to an apparent first-order loss (k ) of activity followed by a slower steady state rate in product
a
formation (v ) where progress curves becomes asymptotic to the v . Equations 2 – 4 described in
s
s
the Experimental Section were used to extract inhibition constants and are reported in Table 2
direct and linearized plots are reported in Figure S3). An overall inhibition constant K * of 38 ± 2
(
i
nM was determined. The off-rate constant of the slow-binding event (k ) was 6-fold slower than
-4
the on-rate (k ) and predicts a turnover of inhibited complex (EI*) consistent with the long pre-
+
4
incubation periods required to observe full inhibition in IC50 assays. The K for compound 2 was
i
2
80 ± 10 nM which is virtually identical to the K of compound 1 of 280 ± 30 nM, coherent with
i
the isostructural binding mode of compounds 1 and 2. Binding of 1 and 2 at the P -site is quasi-
6
isostructural, indicating that binding features at the P -site likely account for the difference in
1
overall affinity. The 8 hydrogen bonds for 2 at the P -site with surrounding residues compared to
1
5 hydrogen bonds for 1 affords a greater enthalpic binding energy for 2, yet the entropic penalty
ACS Paragon Plus Environment
2
0

Page 21 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
for binding of 1 is smaller due to the syn-anti rotational isomerization at the P -site. However,
1
binding of 2 provokes the expulsion of one additional highly ordered water molecule (W1)
compared to 1 which is entropically favorable, and suggesting a higher overall affinity of 2 for
aldolase.
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
Table 1. IC50 values for aldolase inhibition by 2 measured for varying pre-incubation periods.
Pre-incubation
period (min)
IC50 (µM, pH 7.5) a
0
0.67 ± 0.07
0.17 ± 0.02
1
0
30
90
0.13 ± 0.01
0.060 ± 0.008
0.068 ± 0.004
1
80
a
Measured in Tris-Acetate buffer (pH 7.5) with 10 μM FBP
using activity assay described in Experimental Section.
Table 2. Kinetic parameters describing the slow-binding inhibition of aldolase with 2.
Parameter
(U/mg)
Value a
V
m
11.7 ± 0.2
K
m
(μM)
2.7 ± 0.3
K
i
(μM)
0.28 ± 0.01
0.010 ± 0.002
0.064 ± 0.002
0.038 ± 0.002
-1
k
-4 (min )
-1
k
+4 (min )
* (μM)
K
i
a
Values were calculated using the slow-binding model
described in the Experimental Section. Direct and linearized
Dixon plots are shown in Figure S3.
Dual competitive inhibition of aldolase by 3. Compound 3 did not show any changes in IC50
with different incubation times and was analyzed in terms of competitive inhibition as kinetic
model. Given the presence of a second binding site by 3 in the active site, we applied a semi-
generalized formulation for single-enzyme multiple inhibition using two reversible linear
inhibitors, as previously reported and described by equation 5 (details in Experimental Section).46
The inhibition constants for the two binding sites yielded values of 13 ± 3 μM and 9 ± 1 μM
ACS Paragon Plus Environment
2
1

Journal of Medicinal Chemistry
Page 22 of 59
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
(
reported in Table 3; inhibition kinetics are shown in Figure S4). Additional enthalpic
contributions afforded by 3 extra hydrogen bonds in the ALDOA-3 complex (7 hydrogen bonds)
compared to ALDOA-NASEP (4 hydrogen bonds) is consistent with tighter binding of 3 compared
to the similar NASEP inhibitor (K = 0.1 mM). Further, the naphthol O group has an important
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
i
H
role in binding affinity as it anchors the naphthalene to Glu-34 (Figure 4A), a contact not formed
in the ALDOA-NASEP complex (Figure S2).
Competitive inhibition of aldolase by 4. Compound 4 showed no evidence for slow binding
behavior and exhibited classical competitive inhibition consistent with the observation in the
ALDOA-4 complex showing active site tethering at the P -phosphate site. The inhibition constant
6
has a value of 42 ± 5 μM (Table 3; inhibition kinetics are illustrated in Figure S5).
Table 3. Inhibition mode and constants for inhibitors 1 – 4.
Inh
1
Mode
K
i
(μM)
K
i
* (μM)
Competitive a
Slow-binding b
Dual Compet. c
Competitive d
0.280 ± 0.030
0.280 ± 0.010
13 ± 3
-
2
0.038 ± 0.002
3
9 ± 1
-
4
42 ± 5
a
Previously reported18;
b
Calculated using slow-binding
* is the overall
inhibition constant; Calculated using dual linear competitive
inhibition described in Experimental Section; Here, K and K
inhibition described in Experimental Section; K
i
c
i
i
*
represent the inhibition constants for the two binding sites in the
d
aldolase active site; Calculated using classical competitive
inhibition kinetics: direct and double-reciprocal plots are shown
in Figure S3 – Figure S5. Inh, Inhibitor.
Slow-binding inhibition mechanism. Compounds 1 and 2 provided an ideal opportunity to
investigate the structural features responsible for their different modes of inhibition. Both
compounds occupy the same binding sites yet differ in that 2 displays slow binding inhibition
while 1 does not. Earlier work had shown that substitution of a phosphate group on 1 with an
aldehyde (1,6-dihydroxy-2-naphthaldehyde - HNA-P) results in Schiff base formation between
18
Lys107 and HNA-P. It was suggested that the rate of covalent bond formation was responsible
ACS Paragon Plus Environment
2
2

Page 23 of 59
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 5. Active site dynamics upon aldolase-binding of 1 and 2. ALDOA-1 (blue), ALDOA-
2
(red) and native ALDOA (grey) (PDB id: 1ZAH) were aligned using ProSMART, identifying
two regions with significant conformational differences (the flanking helix and the flexible
active site loop). The inset depicts a close-up of the flexible loop, comparing the differences
between all three structures. In ALDOA-2, the loop undergoes significant movement of
residues 272-275 (r.m.s.d. of 0.66 ± 0.03 Å) towards the active site cleft, affording a hydrogen
bond between Gly272 and the bisphosphonate not present in ALDOA-1, where the loop
undergoes minimal displacement with respect to ALDOA (r.m.s.d. of 0.23 ± 0.02 Å). For
illustration purposes, the side-chains were omitted from the loop.
for the slow binding reaction. However, the slow-binding mechanism of aldolase bisphosphonate
inhibitors does not entail formation of a covalent adduct. In order to identify hot spots in the
structure that may explain the slow-binding origin of 2, the ALDOA-1 and ALDOA-2 complexes
were aligned to native ALDOA (PDB id: 1ZAH) using the ProSMART algorithm.47
Overall global root mean square deviations (r.m.s.d. based on Cα atoms) were comparable
(ALDOA-1-native: 0.31 ± 0.01 Å; ALDOA-2-native: 0.23 ± 0.02 Å) and localized to two main
regions: 1) the active site flanking helix comprising residues 34-45; and 2) a loop comprising
residues 269-275 (Figure 5). The conformational changes induced in the flanking helix upon FBP
binding upon active site binding have not been associated with a slow binding mechanism in case
ACS Paragon Plus Environment
2
3

Journal of Medicinal Chemistry
Page 24 of 59
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
27
of inhibitors such as the tagatose-bisphosphate diastereoisomer (K = 120 ± 10 μM). The
i
conformational changes for the loop comprising residues 269-275 were 0.46 ± 0.02 Å in ALDOA-
2 compared to ALDOA-1 of 0.31 ± 0.01 Å (r.m.s.d. for residues 268-276 from alignment of Cα
atoms) with respect to the native structure ALDOA. The local r.m.s.d. of the same loop for the
FBP bound structure (PDB id: 1ZAI) was 0.31 ± 0.02 Å. The local r.m.s.d. becomes even greater
for ALDOA-2 using only residues 272-275 that exhibit the greatest movement (ALDOA-2 vs
native: 0.66 ± 0.03 Å and ALDOA-1 vs native: 0.23 ± 0.02 Å). We speculate that movement by
this loop to accommodate the bisphosphonate moiety in 2 may be the structural basis for the slow-
binding inhibition. Indeed, shown in Figure 5, the greatest backbone movement incurred by
residues 272-275 in ALDOA-2 (Figure 5 – red) implicated the displacement of Gly272 carbonyl
towards the bisphosphate moiety, producing a strong hydrogen-bond (2.76 ± 0.06 Å). This
movement is absent in ALDOA-1 and where side chain displacements were limited to a single
residue, Ser-271. We also considered the contribution of the protonation state of the
bisphosphonate in the slow binding inhibition.
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
Ionization state of bisphosphonates. Bisphosphonates have four ionizable species (monobasic,
pKa1=1.3; dibasic, pKa2=3.2; tribasic, pKa3=6.9; tetrabasic, pKa4~11).48,49 The protonation state
of 2 was inferred from the hydrogen bonding network of the ALDOA-2 complex (shown in Figure
3
C) and corresponded to a mono-protonation state for each phosphonate group which is consistent
with a dibasic species as the dominant protonated state between pH 3.2 – 6.9, according to the
pKa values listed above. At pH 7.5, the conditions used for the inhibition kinetics implies that both
dibasic and tribasic species of 2 are present in solution and only the dibasic species would have
the protonation state competent for binding. Preferential binding by the dibasic species implies
even tighter binding by 2 when the pH dependence of this species is considered. At pH 7.5, the
ACS Paragon Plus Environment
2
4

Page 25 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
population of the dibasic species represents ~20% of the ionizable species and binding constants
when corrected for this population correspond to apparent K and K * values of ~56 nM and ~7.6
i
i
nM, respectively.
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
Minimal requirements of high-affinity binders. Fundamental to all known high-affinity
aldolase binders is the inclusion of phosphate oxyanions that afford interactions with the highly
basic active site, which includes 5 lysine and 3 arginine residues. The apparent promiscuity of the
compound binding modes in the active site may in part be due to the positively charged landscape
making up the active site that is not very heterogeneous in terms of charge distribution (see Figure
1
A) and hence would not enforce strongly similar binding modes on the bisphosphonate
derivatives. Comparison of the structural and kinetic data does however provide guiding principles
that can be applied to the next generation of high-affinity binders. First, the weaker binding
bisphosphonates compounds 3 and 4 compared to compound 2 clearly demonstrated the
importance of the P -site for high affinity binding. Substitution of a phosphate in 2 for a hydroxyl
6
in 3 had a dramatic effect on the binding mode and inhibition constant, consistent with the shorter
span of compound 3 to simultaneously contact both P - and P -sites. This was corroborated by 4,
1
6
whose benzene core had a shorter span than the naphthalene core of 2 (~2.2 Å shorter). The
minimum span required to contact the P - and P -sites is ~9 Å, which is possible with 2 (9.7 Å)
1
6
however, 4 has a span of ~7.5 Å, insufficient to contact both phosphate binding loci. Attachment
to not only the P -binding locus but also to the P -binding locus is synergistic and is an important
1
6
consideration in the design of high affinity phosphorylated binders.
A similar trend that inhibitors capable of simultaneously satisfy binding to both phosphate
binding loci exhibit tighter binding can be drawn from a study of linear bisphosphorylated
5
0
inhibitors used as structural analogs of FBP for probing the active site of aldolase. K values for
i
ACS Paragon Plus Environment
2
5

Journal of Medicinal Chemistry
Page 26 of 59
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
propanediol-, butanediol-, pentanediol-, and hexanediol-bisphosphate inhibitors were determined
to be 120, 55, 29, and 25 µM respectively. Modelling inhibitors into the active site, using the P1-
and P -binding locus of FBP as guides, indeed found the phosphate moieties of pentane- and
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
hexanediol-bisphosphate could readily satisfy binding to both loci without steric collisions
5
0,51
whereas the propanediol- and butanediol-bisphosphate are unable to span the P – P distance.
1
6
Inhibition of aldolase in vivo by a bisphosphonate inhibitor. Finally, we performed in vivo
inhibition assays in order to test cellular proliferation in the presence of the most potent
bisphosphonate inhibitor, 2. The results of the dose response assay shown in Figure 6 indicate that
increased inhibitor concentration progressively inhibits proliferation of HeLa cells (Figure 6A),
yet surprisingly with no effect on cellular proliferation at any tested concentration in HEK293 cells
(Figure 6B). Furthermore, there was no growth inhibition of HeLa cells with the least potent
bisphosphonate, 4, under identical incubation conditions (Figure 6C). The decrease of ~50% after
72 h of incubation by 2 at 50 µM (p ≤ 0.01) is consistent with an EC50 ~50 μM for 2. Incubation
of cell growth ~ 30% by 72 h (p ≤ 0.01) at the lower concentration of 10 μM of compound 2
supports an EC50 estimate of 50 μM.
The charged nature of phosphate groups greatly impedes endogenous phosphate-derivatives
from escaping the cell, and renders exogenous phosphate-derivatives highly impermeable to entry
5
2
across the membrane. Indeed, 2 displayed nano-molar inhibition in vitro, while only exhibiting
inhibition of cellular proliferation in the micro-molar range suggesting that the membrane
permeability of 2 may be an issue given the presence of one phosphate group and two phosphonate
moieties on 2. Surprisingly, phosphorylated compounds such as FBP at high doses have been
shown to display a protective effect in inflammation,53,54 and exert their therapeutic effect by
diffusion through membrane bilayers in a dose-dependent manner with no loss in cell viability and
ACS Paragon Plus Environment
2
6

Page 27 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
a transport efficiency of 5% for 20 mM extracellular dose of FBP.55 From this transport
efficiency, the corrected K for 2 would predict an EC50 of  5 μM in the dose response shown in
i
Figure 6A. The presence of an additional charged phosphate moiety on 2 compared to FBP would
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
likely further reduce transport efficiency which is not inconsistent with the observed EC50 ~50
μM.
ACS Paragon Plus Environment
2
7

Journal of Medicinal Chemistry
Page 28 of 59
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 6. Cell growth inhibition assay. (A) Graph shows cell count as a function of inhibitor
concentration with HeLa cells. Cells were incubated for periods of 24, 48, and 72 h with
inhibitor. Cell count was determined by Trypan blue exclusion and hemocytometer. The same
is shown in (B) for HEK293 cells. (C) The least potent bisphosphonate, compound 4, served as
a control treatment for HeLa cells. * p ≤ 0.05; ** p ≤ 0.01 (Student’s two-tailed t-test). Data are
from at least three independent experiments (standard deviations are shown as error bars).
ACS Paragon Plus Environment
2
8

Page 29 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Cellular targeting of glycolytic aldolase by 2 is thus not inconsistent with the proliferation data.
Lessor dependence on glycolytic flux in normal phenotypes would afford selectivity as inhibition
of glycolytic flux in hyperglycolytic phenotypes by 2 such as in cancer cells or protozoan
infections would be detrimental to their proliferation. The fact that the dose response assay for 4,
shown in Figure 6C, does not reduce cell viability (Figure 6B), and that 4 inhibits aldolase with
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
50-fold lower efficacy compared to 2 indicating a very large EC50 suggests that the aldolase
inhibition of cellular proliferation may have selectivity by inhibiting proliferation of cancerous
phenotypes without effect on normal ones. In hyperglycolytic cells where glycolysis is the main
source of energy and biomass production (Warburg effect), targeting glycolysis via aldolase
inhibition has been shown to be a viable strategy for inhibiting cellular proliferation.56
Upregulation of ALDOA has recently been identified as an oncogene in a subpopulation of patients
1
1
with highly metastatic pancreatic cancer and an overall increase in aldolase gene transcripts was
observed for AML (acute myeloid leukemia) cells upon stromal interaction.57
Implications for Drug Design. From this study, functional groups, including phosphate,
bisphosphonate, and naphthyl moieties of bisphosphonate inhibitors are important constituents for
tight binding and represent a template for the design of future inhibitors. The highly basic nature
of the aldolase active site is amenable to binding by phosphate oxyanions and drives the binding
of the tested bisphosphonates to a number of different basic loci. In order to target binding to
specific loci (i.e. P & P ) and afford the greatest potency, the following essential design features
1
6
should be taken into consideration: 1) ability by compound phosphates to bind both P - and P -
1
6
loci; and 2) retention of a rigid scaffold (naphthyl moiety) to reduce the entropic cost of binding.
Using this template, we can envisage the addition of functional groups based on the proximity to
surrounding residues to yield a library of high-affinity inhibitors that can serve as leads for future
ACS Paragon Plus Environment
2
9