Nonracemic Antifolates Stereoselectively Recruit Alternate Cofactors and Overcome Resistance in S. aureus

Article abstract of DOI:10.1021/jacs.5b01442

While antifolates such as Bactrim (trimethoprim-sulfamethoxazole; TMP-SMX) continue to play an important role in treating community-acquired methicillin-resistant Staphylococcus aureus (CA-MRSA), resistance-conferring mutations, specifically F98Y of dihydrofolate reductase (DHFR), have arisen and compromise continued use. In an attempt to extend the lifetime of this important class, we have developed a class of propargyl-linked antifolates (PLAs) that exhibit potent inhibition of the enzyme and bacterial strains. Probing the role of the configuration at the single propargylic stereocenter in these inhibitors required us to develop a new approach to nonracemic 3-aryl-1-butyne building blocks by the pairwise use of asymmetric conjugate addition and aldehyde dehydration protocols. Using this new route, a series of nonracemic PLA inhibitors was prepared and shown to possess potent enzyme inhibition (IC₅₀ values <50 nM), antibacterial effects (several with MIC values <1 μg/mL) and to form stable ternary complexes with both wild-type and resistant mutants. Unexpectedly, crystal structures of a pair of individual enantiomers in the wild-type DHFR revealed that the single change in configuration of the stereocenter drove the selection of an alternative NADPH cofactor, with the minor α-anomer appearing with R-27. Remarkably, this cofactor switching becomes much more prevalent when the F98Y mutation is present. The observation of cofactor site plasticity leads to a postulate for the structural basis of TMP resistance in DHFR and also suggests design strategies that can be used to target these resistant enzymes. (Chemical Equation Presented).

Full text of DOI:10.1021/jacs.5b01442

Article
pubs.acs.org/JACS
Nonracemic Antifolates Stereoselectively Recruit Alternate Cofactors
and Overcome Resistance in S. aureus
‡
‡
Santosh Keshipeddy, Stephanie M. Reeve, Amy C. Anderson,* and Dennis L. Wright*
Department of Pharmaceutical Sciences, University of Connecticut, 69 North Eagleville Road, Storrs, Connecticut 06269, United
States
*
S Supporting Information
ABSTRACT: While antifolates such as Bactrim (trimethoprim-
sulfamethoxazole; TMP-SMX) continue to play an important role
in treating community-acquired methicillin-resistant Staphylococ-
cus aureus (CA-MRSA), resistance-conferring mutations, specif-
ically F98Y of dihydrofolate reductase (DHFR), have arisen and
compromise continued use. In an attempt to extend the lifetime
of this important class, we have developed a class of propargyl-
linked antifolates (PLAs) that exhibit potent inhibition of the
enzyme and bacterial strains. Probing the role of the configuration
at the single propargylic stereocenter in these inhibitors required
us to develop a new approach to nonracemic 3-aryl-1-butyne
building blocks by the pairwise use of asymmetric conjugate
addition and aldehyde dehydration protocols. Using this new route, a series of nonracemic PLA inhibitors was prepared and
shown to possess potent enzyme inhibition (IC values <50 nM), antibacterial effects (several with MIC values <1 μg/mL) and
5
0
to form stable ternary complexes with both wild-type and resistant mutants. Unexpectedly, crystal structures of a pair of
individual enantiomers in the wild-type DHFR revealed that the single change in configuration of the stereocenter drove the
selection of an alternative NADPH cofactor, with the minor α-anomer appearing with R-27. Remarkably, this cofactor switching
becomes much more prevalent when the F98Y mutation is present. The observation of cofactor site plasticity leads to a postulate
for the structural basis of TMP resistance in DHFR and also suggests design strategies that can be used to target these resistant
enzymes.
INTRODUCTION
DHFR is the pivotal mutation clinically observed to confer high
levels of resistance to trimethoprim, primarily resulting in a
■
5
Among the many classes of therapeutic agents, antibiotics differ
in that their long-term utility is often compromised by the
emergence of bacterial resistance, whereby previously effica-
cious agents lose their ability to effectively combat the current
slate of pathogenic strains. Resistance can arise either through
the acquisition of specific mechanisms (target mutation, efflux,
or drug modifying enzymes) or the emergence of new strains
that fall outside of a compound’s traditional spectrum of
coverage. However, medicinal chemistry efforts are often
successful at overcoming these resistance mechanisms through
rational modification of the antibiotic structure, leading to
successive generations of agents with improved activity. Often,
these efforts require innovations in synthetic chemistry to
efficiently access significant new derivatives.
Perhaps one of the most common organisms that displays
extensive resistance profiles is Staphylococcus aureus, specifically
the methicillin-resistant strains of S. aureus (MRSA). In
community-acquired strains of MRSA, trimethoprim-sulfame-
thoxazole (TMP-SMX, Bactrim) is first-line therapy, targeting
the essential enzymes dihydrofolate reductase (DHFR) and
change in entropy of ligand binding and a loss of synergy or
binding cooperativity between trimethoprim and the NADPH
7
cofactor. Interestingly, the binding cooperativity between
cofactor and the DHF substrate is not significant compared
8
to TMP. New generations of antifolates that effectively target
the mutated forms of DHFR will be critical for prolonging the
utility of this class of antibiotics.
We have been focused on the development of next-
generation antifolates that can target both the wild-type and
predominant TMP-resistant strains. Using a structure-based
approach, we have developed an advanced lead series of
inhibitors that displays low nanomolar inhibition of the wild-
type DHFR and potent activity against a range of MRSA strains
(
MIC values 0.04−0.72 μg/mL) and other important Gram-
9−12
positive pathogens.
This compound class is characterized
by a unique propargylic linker between the polar diaminopyr-
imidine headgroup and a hydrophobic biaryl domain as
exemplified in Figure 1. Alkyne functionality is unique and
structurally distinct from other unsaturated units in that the
1
,2
dihydropteroate synthase (DHPS), respectively. However,
resistance to Bactrim now accounts for a significant proportion
Received: February 13, 2015
3
−6
of the circulating strains.
The F98Y point mutation in
©
XXXX American Chemical Society
A
DOI: 10.1021/jacs.5b01442
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Enzymatic Inhibition Assays. IC50 values were determined by
enzyme inhibition assays, which were performed in triplicate by
monitoring the rate of NADPH oxidation by DHFR via absorbance at
3
40 nm. The reaction was performed at room temperature in buffer
containing 20 mM TES, 50 mM KCl, 0.5 mM EDTA, 10 mM β-
mercaptoethanol, and 1 mg/mL BSA using 0.1 mM NADPH and 2
μg/mL enzyme. One microliter of inhibitor at 0.001−1 μM in DMSO
was added to enzyme/NADPH mixture and allowed to incubate for 5
min before the addition of 0.1 mM DHF in 50 mM TES, pH 7.0. The
mixtures had a final DMSO concentration of 1.8%.
Figure 1. Trimethoprim (TMP) and a potent PLA.
linear, cylindrical nature of the group allows it to fit through
very narrow passages in a binding site such as in the case of
Minimum Inhibitory Concentrations with Efflux Inhibitors.
1
3,14
ponatinib binding the mutant form of Bcr-Abl.
Our work
Minimum inhibitory concentrations were determined as described
1
7
previously, containing either (+)-verapamil (Sigma-Aldrich, stored at
has shown that this group is important for achieving an optimal
1
7
2
0 mg/mL in water) at 100 μg/mL, reserpine (Sigma-Aldrich, stored
fit to the active site and in conferring potency against TMP-
12
15
at 4 mg/mL in DMSO) at 20 μg/mL or thioridazine (Sigma-Aldrich,
stored at 20 mg/mL in DMSO) at 12.5 μg/mL, final DMSO
concentrations of solutions at 0.05%, 0.11% and 0.6%, respectively
Inhibitor stocks kept at 20 mg/mL in DMSO, compounds tested at 10
μg/mL. The MIC was defined as the lowest concentration of inhibitor
to visually inhibit growth. Growth was monitored at A600 after 18 h of
incubation at 37 °C. MIC values were colorimetrically confirmed using
Presto Blue (Life Technologies).
resistant species of DHFR; furthermore, it is both chemically
16
and metabolically stable.
Our prior studies with racemic mixtures of propargyl-linked
antifolates (PLAs) show that this class of compounds maintains
good inhibitory activity against F98Y mutants of S. aureus
1
0
DHFR. Structural studies with a series of PLAs showed that
the branched substituents from the propargylic position are
proximal to the cofactor binding site and may provide
compensatory interactions with NADPH as well as providing
conformational control of the biaryl ring system in both wild-
type and mutant enzymes. Therefore, investigating the role of
the stereogenicity of the propargyl center in governing activity
against the mutant enzymes became a priority. These molecules
possess a challenging stereogenic center containing both
acetylenic and aryl substituents; this is an uncommon
arrangement with very limited synthetic access. Herein, we
describe an efficient asymmetric route to these compounds that
was used to prepare a variety of enantiopure PLAs. Excitingly,
many of these compounds are the most potent inhibitors to
date of both the wild-type and F98Y forms of S. aureus DHFR.
Interestingly, individual crystal structures of the S. aureus
DHFR enzyme bound to members of an enantiomeric pair
show that the stereogenic center drives not only the
conformation of the biaryl system and interactions with the
enzyme, but also drives the placement of the cofactor, NADPH,
revealing plasticity in the cofactor binding site. The detailed
investigation of the bound structure, inhibition profile and
antibacterial activity of the pairs of enantiomers presented here
allowed us to refine a model of antifolate resistance in DHFR
that postulates a role for plasticity in the cofactor site, enhanced
by the predominant mutant binding an alternative cofactor that
greatly attenuates inhibitor potency.
Minimum Inhibitory Concentrations with Thymidine. Mini-
mum inhibitory concentrations were determined in Isosensitest Broth
Oxoid) containing thymidine (Sigma-Aldrich) at 10 μg/mL. The
MIC was defined as the lowest concentration of inhibitor to visually
inhibit growth. Growth was monitored at A600 after 18 h of incubation
at 37 °C. MICs were colorimetrically confirmed using Presto Blue
Life Technologies).
On-Site Dissociation Experiments. The steady-state NADPH
(
(
turnover rate was determined by performing enzyme inhibition assays
following 18 h of 4 °C incubation with 2 μg/mL purified DHFR, 100
μM NADPH, and 50 μM inhibitor in buffer containing 20 mM TES,
5
0 mM KCl, 0.5 mM EDTA, 10 mM β-mercaptoethanol, and 1 mg/
mL BSA. The reaction was initiated with 0.1 mM DHF and monitored
at an absorbance of 340 nm. The steady-state NADPH rate was
extrapolated from data by converting the slope of A340 vs time to
change in NADPH concentration using a molar extinction coefficient
3
−1
−1
of 6.2 × 10 L mol cm .
Crystal Structure Determination. Sa(WT)DHFR was cocrystal-
lized with NADPH and R-27 and S-27 using the hanging drop
vaporization method. Purified protein (17 and 21 mg/mL,
respectively) was incubated with 2 mM NADPH (Sigma-Aldrich)
and 1 mM inhibitor in DMSO for 2h on ice. Equal volumes of the
protein/cofactor solutions were mixed with an optimized crystal-
lization solution containing 13% PEG 10 000, 0.1 M sodium acetate,
.1 M MES pH 6.0−6.25, and 0.5% γ-butyrolactone. When stored at 4
C, conditions typically yielded crystals within 7 days. Crystals were
frozen in cryo-protectant buffer containing 25% glycerol. High-
resolution data were collected on the X25A Beamline at Brookhaven
National Laboratories for Sa(WT)DHFR/NADPH/R-27 and at the
Rigaku HighFlux HomeLab Protein Crystallography X-ray system at
the University of Connecticut for Sa(WT)/NADPH/S-27.
0
°
EXPERIMENTAL METHODS
■
Evaluation of Antibacterial Activity. Minimum inhibitory
concentrations were determined according to Clinical and Laboratory
Standards Institute guideline’s standard microdilution broth assay
Data for Sa(WT)DHFR/NADPH/R-27 and S-27 were indexed and
scaled using HKL2000 or d*TREK, respectively. Phaser was used to
identify molecular replacement solutions for the structures of both
5
using a final inoculum of 5 × 10 CFU/mL in Isosensitest Broth
1
1
structures using PDB/3F0Q as a probe. The programs Coot and
Phenix were used for structure refinement until acceptable R and
(
4
Oxoid). MICs were determined using S. aureus quality control strain
10
3300 (ATCC) and a lab generated F98Y mutant strain. The MIC
work
1
8,19
Rfree were achieved.
Structural geometry was evaluated via
was defined as the lowest concentration of inhibitor to visually inhibit
growth. Growth was monitored at A600 after 18 h of incubation at 37
2
0
Procheck and Ramachandran plots.
°
C. MICs were colorimetrically confirmed using Presto Blue (Life
Technologies).
Enzyme Expression and Purification. Procedures for cloning
RESULTS AND DISCUSSION
Traditionally, acetylenic moieties have been far less frequently
employed in pharmaceutical products than alkyl or alkenyl
■
the Sa(WT)DHFR construct in pET41a(+) have been previously
10
reported. The recombinant Sa (WT) enzyme was overexpressed in
E. coli BL21 (DE3) (Invitrogen) cells and purified using nickel affinity
chromatography (5Prime). Protein was desalted using a PD-10
column (GE Healthcare) into buffer containing 20 mM Tris pH 7.0,
2
−
21,22
linkers but in recent years have increased in prevalence.
One of the main drivers of this increase is advances in
preparative methodology that allow for the mild introduction of
acetylenic units into complex molecules, most notably those
relying on palladium-mediated coupling reactions. However,
0% glycerol, 0.1 mM EDTA, and 2 mM DTT. Aliquots were stored at
80 °C.
B
DOI: 10.1021/jacs.5b01442
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
methods for the preparation of complex alkynes, specifically
nonracemic versions, have been less thoroughly developed and
limits access to diverse alkyne building blocks. In earlier
inhibitor generations, we had established the asymmetric
cinnamates has remained particularly challenging, and many
systems deliver the methyl addition products in poor yields or
29,30
25−28
with low enantioselectivity.
Feringa and co-workers
recently evaluated copper-catalyzed methyl additions to
cinnamate thioesters in the presence of chiral phosphine
ligands. The Feringa study suggested that application to the
PLA building blocks would be limited, as the presence of
required alkoxy groups on the cinnamate aryl ring was shown to
dramatically decrease the reactivity of the cinnamate, leading to
poor reaction yields. However, we were intrigued by their
observation that an electron-withdrawing chlorine atom was
beneficial and speculated that delaying installation of the C-ring
until after the methyl addition would permit the precursor
bromine to balance the electron donating methoxy groups and
permit an efficient addition to take place. Initial studies showed
that the presence of the deactivating bromide on the cinnamate
provided compensatory electronic effects, leading to better
yields of the product relative to the nonbrominated congeners.
Further improvement to the reaction was made by an elevation
in temperature to −55 °C and an increase in catalyst loading.
Utilizing these optimized conditions, it was possible to secure
good isolated yields and excellent enantioselectivities of the
addition products even with two electron-donating alkoxy
groups present on the aromatic ring (see 13 and 15 in Table 1).
Through this process, we were able to prepare five
enantiomeric pairs (11−15 R/S) with differential electron
donating substitution on the aromatic moiety. The subsequent
cross-coupling reaction to the aryl bromide presented some
concern owing to the known propensity of thioesters to engage
directly in palladium-catalyzed processes. Pleasingly, palladium-
promoted Suzuki reaction between the thio-cinnamates (11−
23
propargylic center through an Evan’s asymmetric alkylation.
Subsequently, Fu disclosed an elegant asymmetric catalytic aryl
24
propargylation approach to these inhibitors. However, the
presence of basic heterocycles in the third-generation inhibitors
proved problematic and the earlier routes were plagued both by
ready epimerization when the center was adjacent to electron-
withdrawing functionality and by facile isomerization to the
conjugated allene. Herein, we report a versatile route to these
nonracemic 3-aryl-1-butynes that should prove generally
valuable in the synthesis of sensitive pharmaceutical com-
pounds.
It was envisioned that the application of an asymmetric
conjugate addition strategy to suitably deactivated styrene
derivative 5 could be a good solution to this problem, provided
that a relatively straightforward conversion of the acceptor
functionality in 4 to a terminal alkyne 3 without allene
formation would be possible. This route insulates the benzylic
stereocenter from the necessary electron-withdrawing carbonyl
functionality and thus significantly decreases the potential for
racemization (Scheme 1).
Scheme 1. Retrosynthetic Analysis of Non-Racemic
Propargyl-Linked Antifolates
15 R/S) and heteroaryl boronic acid proceeded exclusively at
the bromide to deliver the biaryl derivatives in very good yields
(
Scheme 2).
With an optimized conjugate addition/Suzuki coupling
protocol in place, the second key process in the route was
the conversion of thiocarbonyl group to the corresponding
terminal acetylene without racemization of the single stereo-
center or isomerization to the allene. This transformation was
accomplished smoothly in two steps by initial conversion to the
aldehyde (16−20 R/S) through catalytic hydrogenation
followed by application of an underutilized direct aldehyde
While there have been several reports on effective
asymmetric conjugate addition reactions of carbon nucleophiles
to electron poor olefins, the addition of methyl onto
Table 1. Asymmetric Conjugate Addition of Methyl Group onto Cinnamic Acid Thioesters
b
c
b
c
product entry
R1
R2
R3
% yield with L1
ee with L1
% yield with ent-L1
ee with ent-L1
d
e
e
11
12
13
14
15
OMe
H
H
H
H
75
84
72
52
80
95 (S)
77
80
71
55
82
92 (R)
OMe
OMe
H
93 (S)
97 (S)
90 (S)
90 (S)
93 (R)
95 (R)
94 (R)
H
OMe
OMe
OCH2
H
d
H
OCH2
94 (R)
a
b
Conditions: 6−10 (1 equiv), MeMgBr (3.3 equiv), CuBr·SMe (9 mol %), L1 or ent-L1 (10 mol %) in MTBE at −55 °C, 18 h. Isolated yield.
2
c
d
Determined by chiral HPLC. ee also determined after asymmetric conjugate addition reaction. For entries 11−15 ee was determined after the
e
formation of alkynes 21−25. Absolute configuration assigned based on Feringa’s work and crystal structures presented here.
C
DOI: 10.1021/jacs.5b01442
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 2. Synthesis of Non-Racemic Propargyl-Linked
Antifolates
31
dehydration reaction. Exposure of our highly functionalized
aldehydes to nonaflyl fluoride and a strong phosphazane base
resulted in a smooth conversion to the terminal alkynes (21−
Figure 2. Crystal structures of wild-type S. aureus DHFR bound to
cofactor and inhibitors: (a) compound S-27 (teal) and β-NADPH
(salmon) or (b) compound R-27 (purple) and α-NADPH (orange).
Panel c shows a superposition of the two structures.
25 R/S) with only trace amounts of allene being observed.
Importantly, determination of the enantiomeric excess for the
derivatives revealed that no epimerization of the propargylic
methyl group had taken place under the strongly basic
conditions of the reactions. Final Sonogashira coupling of
these alkynes with 6-ethyl-5-iodo-2,4-diaminopyrimidine as
rather than directly contacting Ala 7 (Figure 2b). Despite this
displacement, the pyrophosphate tail and adenine nucleotide
are ultimately placed in the identical binding sites. We had first
observed this altered binding mode for NADPH in structures of
mutant (F98Y) SaDHFR bound to earlier generation
23
previously described, delivered the enantiomerically pure
inhibitor series for biological study (Scheme 2). Importantly, it
was also possible to effectively scale-up these routes to gram-
scale level and allowed the direct preparation of 900 mg of
inhibitor S-26 for ongoing in vivo studies.
10
propargyl-linked antifolates. In fact, the structure of an earlier
generation PLA (UCP115A; PDB ID 3FQF) in complex with
the wild-type enzyme binds β-NADPH with 100% occupancy
but shows an exclusive preference for the displaced form of
NADPH with the F98Y mutant enzyme. In general, there is
attenuated Sa(F98Y) enzyme inhibition with increasing
occupancy of the displaced NADPH observed in the crystal
Structural Studies. Crystal structures of SaDHFR/
NADPH/R-27 and SaDHFR/NADPH/S-27 reveal unexpected
and significant differences in the binding mode of the two
enantiomers with the enzyme. After collection of diffraction
data (2.69 and 2.16 Å, respectively; figures showing electron
density are found in SI Table S1 and Figures S1 and S2), the
structures were determined by molecular replacement with
1
0
structure. Overlaying the wild-type Sa/NADPH/R-27
structure reported here with PDB ID:3FQF reveals complete
superposition, including two of the key water molecules
bridging the carboxamide (see SI Figure S3). These changes
in NADPH were initially attributed to a change in
conformation of β-NADPH; however, further analysis has
revealed that the conformational change is actually a
consequence of a change in configuration at the anomeric
carbon of the ribose ring to produce the α-form. Although
NADPH is known to exist in an equilibrium between the two
32
probe molecule PDB ID 3F0Q. Both crystals belong to space
group P6 22 and possess equivalent crystal contacts, none of
1
which are near the active sites. The structure of Sa/NADPH/S-
27 reported here is identical to the previously reported
structure of S. aureus DHFR/NADPH/rac-27, suggesting that
the ternary complex with S-27 is the more thermodynamically
stable complex relative to that formed with R-27. In the
structure of Sa/NADPH/S-27, the enzyme binds the extended,
β-form of NADPH and there are a number of key van der
Waals contacts between the inhibitor and the nicotinamide ring
of the cofactor involving each of the four carbons of the unique
propargyl linker that characterizes this compound series. In
addition to the expected interactions between the diaminopyr-
imidine and residues Asp 27, Val 31, and Leu 28 there are
contacts between Leu 28, Ile 50, Met 42, and Leu 54 and the
hydrophobic 4-arylpyridine moiety that favor a coplanar
arrangement between the two joined aromatic rings. The
acetylenic linker is critical for penetrating the narrow space near
Phe92 (Figure 2a).
33
anomers, with a distribution of approximately 1.5% α-form, it
is rarely observed in crystal structures with any NADPH-
dependent reductase. Furthermore, while β-NADPH is the
common cofactor used in DHFR catalysis, α-NADPH can be
an effective reducing agent for the structurally distinct R67
34
DHFR and has recently been shown to be a natural substrate
for renalase, which catalyzes both an anomerization and
+
33
oxidation reaction to regenerate β-NADP .
Here, wild-type Sa/NADPH/R-27 binds α-NADPH at 100%
occupancy, where α-NADPH is primarily stabilized by a
projection of the methyl group of the R-enantiomer into a
pocket formed by Phe 92 and Thr 46. In this position, Asn 18
forms a hydrogen bond with the ribose hydroxyl from α-
NADPH. Overall, the cofactor change to α-NADPH diminishes
the interactions between the inhibitor and the cofactor.
Specifically, only the propargylic carbon forms a van der
Waals interaction with α-NADPH, whereas all four carbon
Surprisingly, the structure of Sa/NADPH/R-27 reveals a
major change at the NADPH cofactor binding site, whereby the
nicotinamide and ribose rings are displaced (∼3.2 Å) relative to
the standard β-form of NADPH and the displaced carboxamide
forms three water-mediated hydrogen bonds to the protein
D
DOI: 10.1021/jacs.5b01442
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Table 2. Biological Activity of the Propargyl-Linked Antifolates
NADPH
rate
(μM/min)
NADPH rate
F98Y
(μM/min)
MICwild-type
S. aureus
(μg/mL)
c
d
c
Sa IC50
(nM)
SaIC50
Sa(F98Y)IC50
(nM)
MICSa(F98Y)
(μg/mL)
a
inhibitor
R1
R2
R3
(nM)
R-26
S-26
R-27
S-27
R-28
S-28
R-29
S-29
R-30
S-30
TMP
OMe
OMe
H
H
H
H
H
H
H
19 ± 1
11 ± 0.7
15 ± 0.7
18 ± 2
46 ± 4
74 ± 6
202 ± 20
35 ± 6
46 ± 6
35 ± 2
23 ± 3
464 ± 27
37 ± 5
69 ± 5
155 ± 5
26 ± 3
98 ± 8
458 ± 37
211 ± 12
78 ± 4
45 ± 6
77 ± 4
3503 ± 100
261 ± 16
510 ± 50
111 ± 6
267 ± 3
2048 ± 91
12,714 ± 700
2030 ± 60
1179 ± 19
176 ± 14
−4.571
−0.454
−0.919
−0.965
−0.850
−2.712
−19.944
−0.428
−1.241
−0.359
−0.505
−10.129
−1.514
−4.959
−0.975
1.25
10
1.25
5
0.078
0.3125
0.0391
1.25
OMe
OMe
OMe
OMe
H
H
0.625
5
e
H
OMe
OMe
OMe
OMe
OCH2
OCH2
ND
e
H
ND
2.5
20
H
−9.588
−2.701
−6.515
−1.45
1.25
40
H
H
0.3125
0.3125
0.078
0.3125
20
H
OCH2
OCH2
2.5
1.25
10
H
1700 ± 19
−6.178
a
b
Note the change in the descriptor assignment (i.e., S) for the inhibitor 26 because of the 2′-OMe substitution. Dissociation experiments were
performed in duplicate and MIC, IC50 experiments were performed in triplicate. Average NADPH rates for no-inhibitor controls were −138.3 μM/
c
d
min for Sa DHFR and −103.53 μM/min for Sa F98Y DHFR. Measured with 5 min preincubation of enzyme and inhibitor. Measured with no
e
preincubation of enzyme and inhibitor. ND: not determined.
atoms in the linker form interactions with β-NADPH.
Independent of the NADPH switch, Ile 50 undergoes a second
significant conformational change in order to accommodate a
change in enantiomeric preference. Both a 1.5 Å displacement
of the protein backbone and a change in the side chain rotamer
of Ile 50 are necessary to accommodate the R-configuration of
those that did not have a preincubation step, we attempted to
detect differences in the capacity of enantiomers to quickly
form a stable, inhibited ternary complex. The results show an
increase in the observed IC50 value with the magnitude of
increase depending on the specific pairs. For example, the
dioxalane derivatives, R/S-30, display only marginal increases in
IC value while the 2′-methoxy derivative, R-26, shows a more
substantial increase in observed IC50 (24.4-fold) as well as a
more significant difference between the two enantiomers (3.4-
vs 24.4-fold). Overall, these experiments suggest that some
inhibitors require the preincubation step in order to form a high
affinity complex; a conformational change in the enzyme may
be occurring during this incubation.
27. Therefore, when a racemic mixture of the ligand is present,
50
the protein would need to undergo a significant conformational
change to bind the opposite enantiomer.
Biological Studies. Evaluation of the ten enantiopure
inhibitors revealed that all of the compounds maintain strong
inhibition of the wild type bacterial dihydrofolate reductase
with the majority of inhibitors having IC50 values less than 50
nM (Table 2). The compounds also display significant
antibacterial activity against a strain of MRSA (ATCC
Dissociation of the complex was determined for five pairs of
enantiomers. Dissociation rates were determined by following
37
43300), with the most active congeners showing an 8-fold
experiments outlined for neuraminidase, whereby various
inhibitors were incubated with the enzyme for a period of time
followed by the addition of substrate and measurement of the
reaction. Here, each inhibitor was incubated with DHFR and
NADPH for an 18-h period to form the ternary complex. After
the incubation period, dihydrofolate was added to the complex
and catalysis, measured by the formation of the oxidized
cofactor, was monitored over a 1-h time period. With the
exception of R/S-27, for which the rates were similar, these
studies show that there is a differential rate of dissolution of the
inhibited complex depending on the enantiomer present (Table
2 and SI Figure S4).
increase in activity relative to trimethoprim. In contrast to
TMP, the PLAs were designed to form alternative and
additional contacts with residues in the active site produced
by the stereoselective steering of the biaryl domain by the
branching at the propargylic position and to become less
dependent on the cofactor for binding cooperativity toward the
enzyme (Figure 2). Accordingly, inhibitors such as S-26, S-27,
and S-30 are much more potent against the F98Y mutant
enzyme than TMP. The PLAs exhibit a loss in activity of only
3
.3−12-fold as compared to TMP that loses 74-fold against the
mutant enzyme. The PLAs also retain very good antibacterial
activity (MIC values 0.625−1.25 μg/mL).
Overall, the target inhibition profile experiments show that
several inhibitors, including S-26, R-28, S-28, R-30, and S-30,
appear to form an inhibited complex immediately upon mixing.
Furthermore, several inhibitors, including S-26, R-27, S-27, R-
28, S-29, and S-30 have a relatively slow rate of dissolution of
that ternary complex. As effective antibacterial agents must
quickly form an effective and long-lasting inhibitory target
complex, these results are promising, especially for compounds
S-26 and S-30 that possess both of these qualities.
As antibacterial activity is also highly dependent on the
degree and persistence of the blockade of the essential targeted
pathway, we investigated the target inhibition profile of the
enantiomers with DHFR. Previous studies have shown the
importance of conformational changes in the catalytic cycle of
3
5,36
DHFR,
including the presence of two isoforms of the apo
enzyme. Ligands may possess modulated affinity to the
isoforms, but eventually stabilize a single conformation of the
3
5
enzyme. By comparing enzyme inhibition values using the
standard incubation of enzyme and inhibitor (5 min) with
We then employed dissociation experiments in order to
evaluate the dissolution of the ternary complexes with
E
DOI: 10.1021/jacs.5b01442
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Sa(F98Y) DHFR, NADPH and inhibitor. We found that the
complex composed of Sa(F98Y)DHFR/NADPH/TMP was
the least stable of all complexes evaluated compared to its wild-
type counterpart (Table 2 and SI Figure S5) with a dissolution
rate of −6.18 μM/min, 12.4-fold greater than the complex of
wild-type enzyme and TMP. Interestingly, when the complexes
between Sa(F98Y) and the enantiomers were evaluated, many
are quite stable. For example, S-26, S-27, S-29, and S-30 have
very similar rates of dissolution with the mutant enzyme relative
to those with the wild-type enzyme.
Interestingly, there is a noticeable difference in the
antibacterial activity between the two antipodes in most of
the derivatives. For example, while the S-enantiomer of 26
shows a modest 1.7-fold increase in enzyme inhibitory potency
relative to the R-stereoisomer, the antibacterial activity of the S-
isomer is 16-fold greater. This same trend is also seen in
compound 30.
SUMMARY
■
In this study, the pairwise use of asymmetric conjugate addition
reactions and a direct aldehyde dehydration process was pivotal
for the preparation of several enantiomeric pairs of propargyl-
linked antifolates. Biological studies show that the configuration
of this chiral center provides essential conformational control of
the biaryl moiety as well as projects functionality specifically
proximal to the cofactor. These interactions are critical to
maintaining potency against the enzyme with the F98Y point
mutation as well as providing superior potency against the wild-
type enzyme. Accordingly, several of the compounds
demonstrated significant levels of activity against both the
wild-type MRSA and mutants harboring the key F98Y
mutation. A noteworthy observation is that two enantiomers,
containing only a single stereogenic center, drove a unique
change in the adjacent NADPH binding site, leading to the
binding of a minor component of the natural cofactor.
Remarkably, these molecules produce an identical structural
effect as the key clinical F98Y mutation, thus proving invaluable
in refining our understanding of the structural basis of a
resistance mutation on the plasticity of the cofactor binding
site. The molecules also reveal a strategy for overcoming the
primary mechanism for antifolate resistance in MRSA.
Bacterial membrane permeability, selective efflux, possible
second target activity, and differences in target inhibition profile
are some of the most likely determinants of the differential
antibacterial activity. Although differences between enantiomers
with regard to bacterial membrane permeability cannot be
completely ruled out, this is less likely as the major descriptors
of permeability such as partition coefficient, molecular weight,
or topological polar surface area are identical for both
enantiomers. Moreover, the inherent flexibility about the single
stereogenic center limits overall differences in the shape of the
molecules. In order to evaluate the role of efflux transporters,
we measured antibacterial activity in the presence of reserpine,
verapamil and thioridazine, three known efflux pump inhibitors
that have been shown to inhibit the activity of the majority of
As observed in the two crystal structures reported here, the
wild-type enzyme possesses greater plasticity in the cofactor
binding site than may be initially anticipated, permitting the
binding of the minor anomer of the cofactor. It is not known
whether α-NADPH actively reduces dihydrofolate. Previous
experiments show that it is not active using E. coli DHFR or
34
DHFR from L1210 cells; however, it is active with R67
34
40
41
DHFR, glutathione reductase and old yellow enzyme.
This plasticity in the cofactor binding site is reminiscent of the
structures with the F98Y mutation where the mutation is
responsible for altering the population distribution of the β-
NADPH-bound complexes. There is a conserved organized
water network observed in the wild-type enzyme that becomes
substantially more energetically favorable upon introduction of
the tyrosine mutation, whereby the phenolic hydroxyl group
interacts strongly with this network. Upon binding of β-
NADPH to either wild-type or the F98Y mutant, this network
is displaced. In the background of the F98Y mutation,
complexes with simply the water network (apo enzyme) or,
as observed with the PLAs, the water network engaged with α-
NADPH, are likely to be more prevalent than in the wild-type
enzyme (see Figure 3 for a representative cartoon).
3
8,39
pumps in S. aureus, including NorA, MepA, and SepA.
A
decrease in the MIC value of at least 8-fold in the presence of
the pump inhibitors is accepted as evidence of significant efflux
transport activity. As MIC values in the presence of these efflux
pump inhibitors do not change significantly (SI Table S2), it is
unlikely that the different antibiotic potencies of the
enantiomers are related to selective efflux. Next, we examined
the potential that the enantiomers have differential affinity to a
possible second target in the bacteria that would lead to
increased antibacterial activity. As the PLAs are inhibitors of the
folic acid biosynthetic pathway that is primarily responsible for
the production of deoxythymidine monophosphate, rescue
experiments were conducted to measure MIC values in the
presence of the end-product, thymidine. With 10 μg/mL
thymidine, MIC values rose to 5−10 μg/mL (see SI Table S3),
indicating on-target activity. As the pairs of enantiomers possess
the same MIC values in the presence of thymidine, it is unlikely
that the enantiomers have differential affinity for a second
target outside the pyrimidine biosynthesis pathway. Notably,
however, these experiments do not necessarily determine
whether the compounds coinhibit thymidylate synthase or
enzymes upstream of DHFR since the end-product thymidine
captures multiple enzymes on the pathway. Overall, there are
many factors that play significant roles in determining
antibacterial activity. While we have attempted to investigate
the roles of a few important variables here, we recognize the
complexity of the bacteria and stipulate that a number of forces
may be at play to determine the differential activity between the
enantiomers.
Interestingly, the crystal structure of TMP bound to
Sa(F98Y) shows bound β-NADPH and almost completely
superimposes with the structure of the wild-type enzyme bound
to TMP and β-NADPH, making it difficult to explain the 74-
fold loss in enzyme inhibition. However, while this structure of
Sa(F98Y)/β-NADPH/TMP may be thermodynamically stable
in the crystal, there is significant evidence that the formation of
the intermediate binary complexes are less readily formed.
Isothermal calorimetry experiments reveal attenuated binding
for both TMP and β-NADPH in the formation of binary
7
complexes; additionally, dissociation experiments shown here
reveal a very fast dissolution of the complex between Sa(F98Y)
and TMP. Although studies on the catalytic cycle of DHFR
have described a classical binding order where NADPH binds
35
7,8
prior to DHF, various solution experiments have shown
that the binding of TMP and NADPH can occur in any order,
ultimately leading to a stable ternary complex. Additionally,
F
DOI: 10.1021/jacs.5b01442
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
of resistance in which F98Y hinders cofactor binding and
obstructs the formation of the ternary complex required for
stability. Further analysis indicates that the PLAs evade this
mechanism by effectively binding and disabling both the wild-
type and mutant enzymes. The alteration of the cofactor
binding site by a key active site mutation represents a highly
unusual mechanism for drug resistance.
Figure 3. Proposed Mechanism of F98Y-Mediated Resistance. DHFR
can exist in multiple states, bound to the major (a) and minor (c)
anomers of NADPH as well as the solvated form (b). As the binary
complex with β-NADPH (a) is believed to predominate in the wild-
type enzyme, TMP, or the PLAs can bind, form a stable complex, and
exert a strong antibiotic effect. However, the F98Y mutation alters the
distribution between these three states, leading to an increase in (b)
and/or (c) at the expense of (a), effectively protecting a portion of the
enzyme from forming a stable ternary complex with TMP and allowing
bacterial growth. The binding modes of the PLA chemotype allow
these inhibitors to evade this mechanism of resistance by effectively
competing with DHF in all three states, as PLA inhibitors bind
regardless of the cofactor status.
ASSOCIATED CONTENT
Supporting Information
■
*
S
Crystallographic data collection and refinement statistics,
experimental procedures, spectral data, and biological data are
provided. The crystallographic coordinates are deposited in the
Protein Data Bank (PDB codes 4XEC and 4TU5). The
Supporting Information is available free of charge on the ACS
Publications website at DOI: 10.1021/jacs.5b01442.
AUTHOR INFORMATION
Corresponding Authors
amy.anderson@uconn.edu
dennis.wright@uconn.edu
■
*
*
several structures of binary complexes of DHFR indicate that
mimics of DHF and various inhibitors can bind the apo
Author Contributions
These authors contributed equally.
Notes
‡
4
2−45
enzyme.
While TMP can bind the enzyme without β-
NADPH, it is only when β-NADPH binds that a stable ternary
The authors declare no competing financial interest.
8
complex is formed. As dihydrofolate does not bind
8
cooperatively with cofactor, it likely forms complexes with
ACKNOWLEDGMENTS
We gratefully acknowledge grant support from the NIH
R01AI111957) to D.L.W. and A.C.A. Data for this study
■
(
either anomer or the apo enzyme, eventually undergoing
catalysis when productive complexes with β-NADPH arise.
Therefore, TMP resistance is mediated by an increase in
Sa(F98Y) species lacking bound β-NADPH and hence,
lowering affinity for TMP, yet reserving a pool of enzyme
capable of binding and ultimately reducing dihydrofolate. As we
show here that the PLAs can bind to complexes with both α-
NADPH and β-NADPH, it is expected that they could form
inhibitory complexes irrespective of the cofactor. As PLAs do
not appear to depend on the presence of β-NADPH for
binding, it is anticipated that they will also bind effectively to
the apo enzyme. The combination of these possible binding
events leads to a more effective blockade of the folate pathway
in both wild-type and F98Y enzymes. The ability of PLAs to
form ternary complexes with α-NADPH raises the intriguing
possibility that the enzyme is simultaneously bound to two
inhibitory molecules and that sequential replacement of both
would be required before catalytic activity could be restored.
In conclusion, we have demonstrated the importance of
pursuing enantiomerically pure PLAs as the configuration
drives major structural, biochemical, and antibacterial effects
and aids in the translation to studies in vivo. We have
confirmed that this center is critical for activity and that the
preferred configuration is a parameter that is dependent on the
overall composition of the inhibitor. Analysis of ten
enantiomerically pure PLAs identified an exemplary inhibitor,
S-27, which is potent at both the enzymatic and cellular levels
against wild-type and mutant F98Y DHFR. The efficiency of
the synthetic strategy described herein allows us to prepare
compound for ongoing in vivo efficacy analysis. Interestingly,
the crystal structures of one pair of enantiomers reveal that the
least active enantiomer complex bound an alternative cofactor,
α-NADPH. Concurrently, it was observed that the least active
enantiomer of the pair biochemically mirrors the effect of the
F98Y mutation in DHFR that leads to trimethoprim resistance.
These observations lead to a refined model of the mechanism
were measured at beamline X25 of the National Synchrotron
Light Source with the help of Dr. Annie Heroux. We also thank
Pablo Gainza and Bruce Donald for discussion of α-NADPH.
REFERENCES
■
(
1) Frei, C.; Miller, M.; Lewis, J.; Lawson, K.; Hunter, J.;
Oramasionwu, C.; Talbert, R. J. Am. Board Fam. Med. 2010, 23, 714.
(
2) Nathwani, D.; Morgan, M.; Masterton, R.; Dryden, M.; Cookson,
B.; French, G.; Lewis, D. J. Antimicrob. Chemother. 2008, 61, 976.
3) Klevens, M.; Edwards, J.; Tenover, F.; McDonald, L. C.; Horan,
T.; Gaynes, R. NNISS Clin. Infect. Dis. 2006, 42, 389.
4) Wisplinghoff, H.; Bischoff, T.; Tallent, S.; Seifert, H.; Wenzel, R.;
(
(
Edmond, M. Clin. Infect. Dis. 2004, 39, 309.
(5) Dale, G.; Broger, C.; D’Arcy, A.; Hartman, P.; DeHoogt, R.;
Jolidon, S.; Kompis, I.; Labhardt, A.; Langen, H.; Locher, H.; Page, M.;
Stuber, D.; Then, R.; Wipf, B.; Oefner, C. J. Mol. Biol. 1997, 266, 23.
(
6) Houvinen, P.; Sundstrom, L.; Swedberg, G.; Skold, O. Antimicrob.
Agents Chemother. 1995, 39, 279.
7) Heaslet, H.; Harris, M.; Fahnoe, K.; Sarver, R.; Putz, H.; Chang,
J.; Subramanyam, C.; Barreiro, G.; Miller, J. R. Proteins 2009, 76, 706.
8) Feeney, J.; Birdsall, B.; Kovalevskaya, N. V.; Smurnyy, Y. D.;
Navarro Peran, E. M.; Polshakov, V. I. Biochemistry 2011, 50, 3609.
9) Viswanathan, K.; Frey, K.; Scocchera, E.; Martin, B.; Swain, P.;
(
(
(
Alverson, J.; Priestley, N.; Anderson, A.; Wright, D. PLoS One 2012, 7,
e29434.
(10) Frey, K.; Liu, J.; Lombardo, M.; Bolstad, D.; Wright, D.;
Anderson, A. J. Mol. Biol. 2009, 387, 1298.
(
11) Frey, K.; Lombardo, M.; Wright, D.; Anderson, A. J. Struc. Biol.
010, 170, 93.
12) Frey, K.; Viswanathan, K.; Wright, D.; Anderson, A. Antimicrob.
Agents Chemother. 2012, 56, 3556.
13) Huang, W.-S.; Metcalf, C. A.; Sundaramoorthi, R.; Wang, Y.;
2
(
(
Zou, D.; Thomas, R. M.; Zhu, X.; Cai, L.; Wen, D.; Liu, S.; Romero, J.;
Qi, J.; Chen, I.; Banda, G.; Lentini, S. P.; Das, S.; Xu, Q.; Keats, J.;
Wang, F.; Wardwell, S.; Ning, Y.; Snodgrass, J. T.; Broudy, M. I.;
Russian, K.; Zhou, T.; Commodore, L.; Narasimhan, N. I.;
G
DOI: 10.1021/jacs.5b01442
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Mohemmad, Q. K.; Iuliucci, J.; Rivera, V. M.; Dalgarno, D. C.; Sawyer,
T. K.; Clackson, T.; Shakespeare, W. C. J. Med. Chem. 2012, 53, 4701.
(
14) O’Hare, T.; Shakespeare, W. C.; Zhu, X.; Eide, C.; Rivera, V. M.;
Wang, F.; Adrian, L.; Zhou, T.; Huang, W.-S.; Xu, Q.; Metcalf, C. A.;
Tyner, J.; Loriaux, M.; Corbin, A.; Wardwell, S.; Ning, Y.; Keats, J.;
Wang, Y.; Sundaramoorthi, R.; Thomas, R. M.; Zhou, D.; Snodgrass, J.
T.; Commodore, L.; Sawyer, T. K.; Dalgarno, D. C.; Deininger, M.;
Druker, B.; Clackson, T. Cancer Cell 2009, 16, 401.
(
15) Zhou, W.; Scocchera, E.; Wright, D.; Anderson, A. Med. Chem.
Comm. 2013, 4, 908.
16) Zhou, W.; Viswanathan, K.; Hill, D.; Anderson, A.; Wright, D.
Drug Metab. Distrib. 2012, 40, 2002.
17) Aeschlimann, J. R.; Dresser, L. D.; Kaatz, G. W.; Rybak, M. J.
Antimicrob. Agents Chemother. 1999, 43, 335.
18) Adams, P.; Afonine, P.; Bunkoczi, G.; Chen, V.; Davis, I.;
(
(
(
́
Echools, N.; Headd, J.; Hung, L.-W.; Kapral, G.; Grosse-Kunstleve, R.;
McCoy, A.; Moriarty, N.; Oeffner, R.; Read, R.; Richardson, D.;
Richardson, J.; Terwilliger, T.; Zwart, P. Acta Crystallogr. 2010, D66,
213.
(
(
19) Emsley, P.; Cowtan, K. Acta Crystallogr. 2004, D60, 2126.
20) Laskowski, R.; MacArthur, M.; Moss, D.; Thornton, J. J. Appl.
Crystallogr. 1993, 26, 283.
(21) King, A.; Yasuda, N. Top. Organomet. Chem. 2004, 6, 205.
(22) Chinchilla, R.; Najera, C. Chem. Rev. 2014, 114, 1783.
(23) Bolstad, D.; Bolstad, E.; Frey, K.; Wright, D.; Anderson, A. J.
Med. Chem. 2008, 51, 6839.
(
24) Smith, S.; Fu, G. J. Am. Chem. Soc. 2008, 130, 12645.
(
25) Lopez, F.; Harutyunyan, S. R.; Meetsma, A.; Minnaard, A. J.;
Feringa, B. L. Angew. Chem., Int. Ed. 2005, 44, 2752.
26) Palais, L.; Babel, L.; Quintard, A.; Belot, S.; Alexakis, A. Org.
Lett. 2010, 12, 1988.
(
(
(
27) Wang, S.; Loh, T. Chem. Commun. 2010, 46, 8694.
28) Wang, S. Y.; Lum, T. K.; Ji, S. J.; Loh, T. P. Adv. Synth. Catal.
2
008, 350, 673.
29) Des Mazery, R.; Pullez, M.; Lopez, F.; Harutyunyan, S. R.;
Minnaard, A. J.; Feringa, B. L. J. Am. Chem. Soc. 2005, 127, 9966.
30) Ruiz, B. M.; Geurts, K.; Fernandez-Ibanez, M. Å.; Horst, B. T.;
Minnaard, A. J.; Feringa, B. L. Org. Lett. 2007, 9, 5123.
31) Lyapkalo, I. M.; Vogel, M. A. K.; Boltukhina, E. V.; Vavrik, J.
Synlett. 2009, 558.
32) Frey, K.; Georgiev, I.; Donald, B.; Anderson, A. Proc. Natl. Acad.
Sci., U. S. A. 2010, 107, 13707.
33) Beaupre, B.; Carmichael, B.; Hoag, M.; Shah, D.; Moran, G. J.
Am. Chem. Soc. 2013, 135, 13980.
34) Smith, S.; Burchall, J. Proc. Natl. Acad. Sci., U. S. A. 1983, 80,
619.
35) Schnell, J. R.; Dyson, H. J.; Wright, P. E. Annu. Rev. Biophys.
Biomol. Struct. 2004, 33, 119.
36) Boehr, D. D.; McElheny, D.; Dyson, H. J.; Wright, P. E. Science
006, 313, 1638.
37) Bantia, S.; Arnold, C.; Parker, C.; Upshaw, R.; Chand, P.
Antiviral Res. 2006, 69, 39.
38) Couto, I.; Costa, S. S.; Viveiros, M.; Martins, M.; Amaral, L. J.
Antimicrob. Chemother. 2008, 62, 504.
(
(
(
(
(
(
4
(
(
2
(
(
(
39) Thanacoody, H. K. R. Br. J. Clin. Pharmacol. 2007, 64, 566.
(
40) Klemm, A.; Steiner, T.; Flotgen, U.; Cumme, G. A.; Horn, A.
̈
Methods Enzymol. 1997, 280, 171.
(
(
41) Massey, V.; Schopfer, L. M. J. Biol. Chem. 1986, 261, 1215.
42) Bourne, C. R.; Bunce, R. A.; Bourne, P. C.; Berlin, K. D.;
Barrow, E. W.; Barrow, W. W. Antimicrob. Agents Chemother. 2009, 53,
065.
43) Bennett, B.; Langan, P.; Coates, L.; Mustyakimov, M.;
3
(
Schoenborn, B.; Howell, E. E.; Dealwis, C. Proc. Natl. Acad. Sci., U.
S. A. 2006, 103, 18493.
(
44) Lee, H.; Reyes, V. M.; Kraut, J. Biochemistry 1996, 35, 7012.
(
45) Osborne, M. J.; Schnell, J.; Benkovic, S. J.; Dyson, H. J.; Wright,
P. E. Biochemistry 2001, 40, 9846.
H
DOI: 10.1021/jacs.5b01442
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX