Acetylenic linkers in lead compounds: A study of the stability of the propargyl-linked antifolates

Article abstract of DOI:10.1124/dmd.112.046870

Propargyl-linked antifolates that target dihydrofolate reductase are potent inhibitors of several species of pathogenic bacteria and fungi. This novel class of antifolates possesses a relatively uncommon acetylenic linker designed to span a narrow passage in the enzyme active site and join two larger functional domains. Because the use of alkyne functionality in drug molecules is limited, it was important to evaluate some key physicochemical properties of these molecules and specifically to assess the overall stability of the acetylene. Herein, we report studies on four compounds from our lead series that vary specifically in the environment of the alkyne. We show that the compounds are soluble, chemically stable in water, as well as simulated gastric and intestinal fluids with half-lives of approximately 30 min after incubation with mouse liver microsomes. Their primary in vitro route of metabolism involves oxidative transformations of pendant functionality with little direct alteration of the alkyne. Identification of several major metabolites indicated the formation of N-oxides; the rate of formation of these oxides was highly influenced by branching substitutions around the propargyl linker. On the basis of the lessons of these metabolic studies, a more advanced inhibitor was designed, synthesized, and shown to have increased (t_1/2 = 65 min) metabolic stability while maintaining potent enzyme inhibition. Copyright

Full text of DOI:10.1124/dmd.112.046870

Supplemental material to this article can be found at:
http://dmd.aspetjournals.org/content/suppl/2012/07/18/dmd.112.046870.DC1.html
1521-009X/12/4010-2002–2008$25.00
DRUG METABOLISM AND DISPOSITION
Copyright © 2012 by The American Society for Pharmacology and Experimental Therapeutics
DMD 40:2002–2008, 2012
Vol. 40, No. 10
46870/3796554
Acetylenic Linkers in Lead Compounds: A Study of the Stability
□
S
of the Propargyl-Linked Antifolates
Wangda Zhou, Kishore Viswanathan, Dennis Hill, Amy C. Anderson, and Dennis L. Wright
Department of Pharmaceutical Sciences, University of Connecticut, Storrs, Connecticut
Received May 22, 2012; accepted July 18, 2012
ABSTRACT:
Propargyl-linked antifolates that target dihydrofolate reductase ble in water, as well as simulated gastric and intestinal fluids with
are potent inhibitors of several species of pathogenic bacteria and half-lives of approximately 30 min after incubation with mouse liver
fungi. This novel class of antifolates possesses a relatively uncom- microsomes. Their primary in vitro route of metabolism involves
mon acetylenic linker designed to span a narrow passage in the oxidative transformations of pendant functionality with little direct
enzyme active site and join two larger functional domains. Be- alteration of the alkyne. Identification of several major metabolites
cause the use of alkyne functionality in drug molecules is limited,
it was important to evaluate some key physicochemical properties oxides was highly influenced by branching substitutions around
of these molecules and specifically to assess the overall stability of the propargyl linker. On the basis of the lessons of these metabolic
indicated the formation of N-oxides; the rate of formation of these
the acetylene. Herein, we report studies on four compounds from studies, a more advanced inhibitor was designed, synthesized, and
our lead series that vary specifically in the environment of the shown to have increased (t1/2 ؍ 65 min) metabolic stability while
alkyne. We show that the compounds are soluble, chemically sta- maintaining potent enzyme inhibition.
Introduction
IC50 values of 42 and 0.5 nM, respectively. The biphenyl-based
compounds also inhibit the growth of methicillin-resistant S. aureus as
well as C. glabrata in culture with minimal inhibitory concentration
(MIC) values of 5.76 and 3 ␮g/ml, respectively (Liu et al., 2008;
Viswanathan et al., 2012).
Because the lead series incorporating a biphenyl substructure
substantially increased the hydrophobicity of the compounds, we
designed and synthesized a series of propargyl-linked antifolates
with nitrogenous heterocyclic moieties (compounds 2–8) predicted
to increase solubility (Viswanathan et al., 2012). Several of these
heterocyclic-bearing analogs show superior activity against S. au-
reus DHFR with IC50 values of 19 nM and MIC values of 0.09
␮g/ml.
Over the past several years, we have focused on the development of
a novel class of antifolates designed to inhibit dihydrofolate reductase
(DHFR) from pathogenic organisms. These compounds are charac-
terized by a conserved diaminopyrimidine moiety linked through a
propargylic spacer to a variable hydrophobic domain (see Fig. 1)
(Pelphrey et al., 2007; Bolstad et al., 2008; Liu et al., 2008, 2009;
Paulsen et al., 2009). Crystal structures of DHFR from Staphylococ-
cus aureus (Frey et al., 2009, 2010b; Viswanathan et al., 2012),
Candida glabrata, Candida albicans (Liu et al., 2008, 2009; Paulsen
et al., 2011), and Bacillus anthracis (Beierlein et al., 2008) reveal that
the propargyl linker occupies a narrow space bridging two critical
pockets in the enzyme, one of which binds the diaminopyrimidine
group and the other that is primarily hydrophobic. Using these crystal
structures, several compounds with this generalized scaffold and a
To further investigate the potential of the propargyl-linked antifo-
lates as drug leads, we present here an investigation into the physi-
biphenyl moiety (example compound 1 in Fig. 1) were designed, cochemical properties of key propargyl-linked antifolates. The acety-
synthesized, and shown to be potent leads against a variety of pro-
karyotic and eukaryotic pathogens. For example, a 3,5-dimethyl de-
rivative of compound 1 inhibits S. aureus and C. glabrata DHFR with
lenic linker in these antifolates is relatively rare in drug molecules.
However, the appearance of acetylenic groups in drugs is increasing,
given the frequency of use and ease of the Sonogashira coupling
reaction (King and Yasuda, 2004) that provides for a facile coupling
of terminal alkynes to various aromatic systems. In fact, an ethynyl
group is an important feature of ponatinib, a new breakpoint cluster
region-Abelson kinase protein kinase inhibitor (O’Hare et al., 2009;
Huang et al., 2010), where it reduces bulk near the resistance-confer-
ring T315I mutation. Efavirenz, a nonnucleoside reverse transcriptase
inhibitor (Lindberg et al., 2002), uses an alkyne linker to extend
through a passage composed of Leu100, Tyr181, and Tyr188. There-
fore, an investigation into the stability of compounds with a propar-
This work was supported by the National Institutes of Health National Institute
of Allergy and Infectious Diseases Extramural Activities [Grants AI073375,
AI065143]; and the National Institutes of Health National Institute of General
Medical Sciences [Grant GM067542].
Article, publication date, and citation information can be found at
http://dmd.aspetjournals.org.
http://dx.doi.org/10.1124/dmd.112.046870.
□S The online version of this article (available at http://dmd.aspetjournals.org)
contains supplemental material.
ABBREVIATIONS: DHFR, dihydrofolate reductase; MIC, minimal inhibition concentration; HPLC, high-performance liquid chromatography;
DMSO, dimethyl sulfoxide; ACN, acetonitrile; HPMC, hydroxypropylmethylcellulose; AUC, area under the curve; MLM, mouse liver microsomes;
LC-MS/MS, liquid chromatography/tandem mass spectrometry; CID, collision-induced dissociation; THF, tetrahydrofuran.
2002

STABILITY AND THE PROPARGYL-LINKED ANTIFOLATES
2003
NADPH regenerating system at 37°C. The final concentrations of inhibitor,
microsomes, NADPϩ, glucose 6-phosphate, glucose-6-phosphate dehydroge-
nase, and MgCl2 were 5 ␮g/ml, 0.5 mg/ml, 1.3 mM, 3.3 mM, 0.5 U/ml, and 3.3
mM, respectively. DMSO and compounds without MLM were used as nega-
tive controls. The reactions were initiated by the addition of microsomes and
quenched by the addition of an equal volume of ice-cold ACN after 60 or 120
min. The samples were subjected to centrifugation at 15,000 rpm for 10 min,
and the supernatant phase was purified using Oasis hydrophilic-lipophilic-
balanced sample extraction cartridges (Waters, Milford, MA) following the
manufacturer’s protocol. Methanol (1 ml) was used to elute the sample, which
was further diluted 2-fold with water. One hundred microliters of the diluted
sample was injected on to the HPLC reversed-phase column. To compare the
metabolic stability of different compounds, the percentage of the parent com-
pound remaining after incubation as compared with the amount before incu-
bation was calculated. The in vitro metabolic half-life was calculated using a
single time point approach following first-order kinetics as described previ-
ously (Kerns and Di, 2008).
To identify metabolites, compounds 2, 3, 7, 8, and 10 (at 10 ␮g/ml) were
incubated with MLM. Negative controls included 1) using the same volume of
DMSO or 2) quenching the reaction with ACN immediately after adding
microsomes. Both samples and negative controls were incubated for 2 h, and
the reaction was terminated with an equal volume of ice-cold ACN. After
removal of protein precipitates by centrifugation, samples and negative con-
trols were purified as described in the paragraph above. The concentration of
compounds before and after metabolic incubation was calculated by standard
curves obtained with each compound. Standard curves were performed mul-
tiple times and showed intraday and interday CV of less than 5%.
Liquid Chromatography/Tandem Mass Spectrometry Analysis and
Metabolite Identification. An Agilent 1100 HPLC system fitted with a
Zorbax-C18 (3 ␮m, 1.0 ϫ 150 mm) (Agilent Technologies, Santa Clara, CA)
was used for separation. Mobile phase A, consisting of 0.01% heptafluorobu-
tyric acid in water, and mobile phase B, consisting of 0.01% heptafluorobutyric
acid in ACN, were used for a linear gradient elution as follows: 0 to 100% B
in 17 min and isocratic hold at 100% B for 5 min. The flow rate was 75 ␮l/min,
and the injection volume was 8 ␮l. Mass spectrometric detection was per-
formed on a Micromass Q-Tof-2 mass spectrometer from Waters, equipped
with an electrospray ionization source. The mass spectrometer was operated in
the positive ionization mode and was calibrated with Glu1-fibrinopeptide B on
each experiment day. Capillary voltage and cone voltage were set at 3000 and
20 V, respectively. Source and desolvation temperatures were 100 and 150°C.
Nitrogen was used as a desolvation gas at a flow rate of 450 ml/h. Full-scan
time-of-flight spectra were first acquired for parent compounds and metabolites in
the mass spectrometry mode. Subsequently, collision-induced dissociation (CID)
fragmentation spectra were obtained on the isolated protonated molecular ion of
parent and metabolites in the tandem mass spectrometry mode. Argon was used as
the collision gas, and the collision energy was optimized for each compound in the
range of 20 to 40 eV. Accurate mass was calculated using fragments of the
coanalyzed calibration standard lisinopril as described previously (Hill et al.,
2008). The distribution of metabolites was calculated by dividing the AUC of the
individual metabolite by the total AUC for all metabolites within the same
injection.
FIG. 1. Propargyl-linked lead compounds. Top, propargyl-linked antifolates and
their IC50 and MIC values against S. aureus DHFR and methicillin-resistant S.
aureus. Bottom, compound 1 bound to C. glabrata DHFR (from Protein Data Bank
identifier 3EEJ).
gylic linker would not only be informative for the development of this
lead series but may also be applied more broadly.
Materials and Methods
General High-Performance Liquid Chromatography Analysis. A Shi-
madzu Prominence 20 high-performance liquid chromatography (HPLC) in-
strument (Shimadzu, Kyoto, Japan) fitted with a Luna 5 ␮m C18(2) 100 Å
column (5 ␮M, 4.6 ϫ 250 mm; Phenomenex, Torrance, CA) and a UV diode
array detection at 254 nm was used to quantify compounds.
Kinetic Solubility Assay. Compounds were initially dissolved as 20 or 40
mM dimethyl sulfoxide (DMSO) solutions and diluted in filtered water in the
presence or absence of 200 ␮g/ml methylcellulose (METHOCEL A4M; Dow
Corning, Midland, MI). All samples were centrifuged for 10 min at 15,000
rpm, incubated at room temperature for 30 min, and analyzed by reversed-
phase HPLC. The mobile phase consisted of 50% acetonitrile (ACN) and 50%
potassium phosphate buffer (50 mM, pH 7.0), using an isocratic flow rate of
1.5 ml/min. Solubility was determined as the maximal concentration for which
absorption is linearly related to the log of the concentration (Kerns and Di,
2008).
Synthesis of N-Oxides. To a solution of compound 3 (15 mg, 0.04 mmol) in
tetrahydrofuran (THF) at 0°C, m-chloroperoxybenzoic acid (7 mg, 0.04 mmol) was
added in one portion. The reaction mixture was stirred for 30 min, and the reaction was
quenched by adding sodium metabisulfite. The reaction mixture was filtered, and
potassium carbonate was added in excess to quench the acid. The organic layer was
then dried over MgSO4, filtered, and concentrated. The mixture of N-oxides was then
separated from starting material by reversed-phase chromatography using amino-
capped silica and dichloromethane as an eluent.
Solution Stability Assay. Compounds at 5 ␮g/ml were incubated with
water, simulated gastric fluid and simulated intestinal fluid at 37°C for 24 h.
All solutions contained 200 ␮g/ml hydroxypropylmethylcellulose (HPMC)
A4M. Samples were collected before and after incubation. Assays were ter-
minated by adding an equal volume of ice-cold ACN and centrifuging at
15,000 rpm for 10 min. One hundred microliters of supernatant was injected on
Isolation of Pyridine-N-Oxide. An Agilent 1200 HPLC system fitted with
an Eclipse-C18 (5 ␮m, 4.6 ϫ 150 mm; Agilent Technologies) column was
used for separation. The mobile phase consisted of 60% MeOH and 40% H2O,
the reversed-phase column of the HPLC; solution stability was calculated using an isocratic flow rate of 1.0 ml/min. The required pyridine-N-oxide
using a ratio of area under the curve (AUC) (after incubation)/AUC (before separated from the mixture with a retention time of 5.01 min.
incubation).
Metabolic Stability Assay. Compounds were incubated with mouse liver the rate of NADPH consumption at 340 nm over 5 min. Reactions were
microsomes (MLM; BD Biosciences, San Jose, CA) in 0.1 M potassium performed with 50 mM KCl, 10 mM 2-mercaptoethanol, 0.5 mM EDTA, and
phosphate buffer (pH 7.4) and 200 ␮g/ml HPMC A4M in the presence of an 1 mg/ml bovine serum albumin. Concentrations of cofactor (100 ␮M NADPH)
Enzyme Inhibition. Enzyme activity assays were performed by monitoring

2004
ZHOU ET AL.
simulated gastric fluid (pH 1.2), and water. HPLC analysis reveals
TABLE 1
Solubility values for compounds 1 through 6 in 0.02% HPMC and water
that there is no change in AUC or peak shape after 24 h in any of the
fluids, suggesting that the compounds are stable under all tested
conditions.
Compound Identifier
cLogD7.4
Solubility
␮g/ml
All four compounds were evaluated for phase I metabolic stability
by incubating with MLM in the presence of an NADPH-regenerating
system for 1 h. The percentage of compound remaining after incuba-
tion relative to the initial concentration was calculated (Table 2).
Compounds 2 and 3 that differ in substitution at C6 both show
moderate and similar stability, with values of 33.7 and 29.2% remain-
ing, respectively, after 1 h. However, substituting the methyl group at
1
2
3
4
5
6
3.89
2.39
3.44
1.85
2.90
1.11
20
40
40
Ͼ80
Ͼ80
60
and substrate (1 mM dihydrofolate) were used with limiting concentrations of the propargyl position with a hydrogen (compound 7) significantly
pure enzyme. The assay was performed in triplicate.
reduces the metabolic stability whereby only a trace of the parent
compound remains after 1 h. Compound 8, with a hydrogen at the
propargylic position and a methoxy group at the 2Ј position instead of
Results
Design of the Heterocyclic Propargyl-Linked Antifolates. The the 3Ј position, exhibits metabolic stability similar to that of com-
crystal structures of C. glabrata DHFR and S. aureus DHFR bound to pounds 2 and 3. Taken together, it is clear that the metabolic stability
biphenyl-propargyl-linked antifolates (Frey et al., 2009, 2010a; Liu et of these compounds is highly sensitive to substitution at propargylic
al., 2009; Paulsen et al., 2011) guided the placement of solubility- position as well as the 2Ј and 3Ј positions on the adjacent phenyl ring.
enhancing functionality. In particular, these structures reveal that al-
The metabolites of compounds 2, 3, 7 and 8 were identified and
though substitutions at the meta and para positions of the aryl ring interact characterized using LC-MS/MS. Analysis of the crude microsomal
productively with the enzyme, the region is largely solvent exposed (Fig. incubation of compounds 2, 3, 7, and 8 showed the presence of
1) and could tolerate the substitution of more polar atoms.
metabolic products at masses corresponding to M-14, M ϩ 16, and
A comparison of cLogD values for compounds with a phenyl ring M ϩ 32 (where M is monoisotopic molecular weight). In all four
at the distal position (compound 1) with values for compounds that metabolites, the M-14 product was easily assigned as the result of
replace the phenyl ring with heterocycles containing endocyclic ni- oxidative O-demethylation. Assignment of the M ϩ 16 products was
trogen or oxygen (compounds 2–6) (Table 1) predicts that the pres- more complicated because a variety of potential oxidation sites exist
ence of heteroatoms in the distal ring will lower the hydrophobicity of in these molecules including unsubstituted aromatic systems, the
the compounds. Several analogs, including compounds 2 to 6, were activated propargylic position, C6 benzylic positions, the acetylene
synthesized to evaluate this hypothesis (Viswanathan et al., 2012). linker, and the endocyclic basic nitrogens of the pyrimidine and
Kinetic solubility data for compounds 2 to 6 were determined (Table pyridine systems. Using both analysis of the mass spectrum and
1) to experimentally gauge the impact of these substitutions. HPMC chemical synthesis, attempts were made to assign the structures of the
(0.02%) was added to the aqueous solutions to prevent precipitates major products of mono-oxidation.
from forming after standing overnight.
One of the more interesting features of the metabolic studies was
All of the heterocyclic substitutions correlated with increased sol- the greatly reduced half-life for compound 7 relative to compound 3
ubility of the compounds. Compounds with a pyridyl ring system such that differs only by branching at the activated propargylic position. It
as 2 and 3 are very potent inhibitors of DHFR, specifically from S. appeared likely that a facile oxidation of the unsubstituted propargylic
aureus, with IC50 values of 26 and 19 nM, respectively (Fig. 1) position in 7 to the secondary alcohol would occur more rapidly than
(Viswanathan et al., 2012). In addition to 2 and 3, pyridyl compounds the corresponding oxidation of 3 to the tertiary alcohol and would
7 and 8 were also found to be potent inhibitors of S. aureus DHFR, rationalize the difference in half-lives for these compounds. Synthesis
with IC50 values of 12 and 21 nM, respectively, and MIC values of of an authentic sample of the racemic alcohol was straightforward
0.04 and 0.02 ␮g/ml, respectively.
(Fig. 2) and afforded the necessary compound (i.e., 9) for comparison.
To investigate the chemical and metabolic stability of the propar- However, compound 9 did not match any of the mono-oxidation
gyl-linked antifolates, compounds 2, 3, 7, and 8 were chosen for products observed in the microsomal incubation, thus indicating that
analysis. These four compounds have varying substitution patterns at the direct oxidation of the propargylic position does not appear to be
the C6 position of the pyrimidine (R1 in Fig. 1; 2 versus 3, 7–8), the a primary route of metabolism for these compounds.
propargylic position (RP in Fig. 1; 2, 3 versus 7, 8), and the 2Ј and 3Ј
positions (R2Ј and R3Ј in Fig. 1; 2–3, 7 versus 8) that provide different assigning sites of transformation (Table 3). Three principle metabo-
chemical environments for the alkyne. lites arise from compound 8, one of which is the product of O-dem-
Metabolites produced from compound 8 proved especially useful in
Compounds 2, 3, 7, and 8 were first examined for chemical stability ethylation (M8-2, 19.7%), two arising from mono-oxidation [74.2
after 24 h of incubation at 37°C in simulated intestinal fluid (pH 7.4), (M8-1) and 6.3% (M8-3)] and a minor amount of a product arising
TABLE 2
Half-life of compounds and distribution of metabolites
Percentage Remaining
after 1 h
Oxidation on Pyrimidinyl
Substructure (ϩ16)
Compound
Half-Life
Pyridyl N-Oxidation (ϩ16)
Demethylation (Ϫ14)
Bis-Oxidation (ϩ32)
min
%
2
3
7
8
33.7
29.2
0
38
34
Ͻ10
31
59.9% (M2-1)
45.8% (M3-1)
70.5% (M7-1)
74.2% (M8-1)
15.9%a (M2-2)
28.0% (M3-2)
6.3% (M7-2)
19.7% (M8-2)
13.8% (M2-3)
25.8% (M3-3)Ͻ5% (M3-4)
5.3% (M7-3)
10.3
10.1
15.1
1.7
26.5
6.3% (M8-3)
a Percentages based on total AUC.

STABILITY AND THE PROPARGYL-LINKED ANTIFOLATES
2005
FIG. 2. Synthesis of compound 9. a, Pd
(PPh3)2Cl2, pyridine-4-boronic acid, Cs2CO3,
dioxane, 80°C, 85%. b, ethynyl-MgBr, THF,
0°C, 88%. c, Pd(PPh3)2Cl2, CuI, Et3N, dimeth-
ylformamide, 70%.
from two consecutive oxidations (1.7%). The CID spectrum of the M3-2 and the unassigned oxidation product M3-3 eluted with
parent compound 8 showed that major fragments were formed retention times shorter than those of the parent. This difference in
through cleavage of either the a–b bond or the b–c bond, producing retention time is likely due to the loss of one of the basic nitrogens,
two key ions F1 and F2, respectively (Fig. 3). The ion F1 at m/z ϭ an effect that would be expected to be more pronounced at the low
198.1 was assigned as a benzylic cation, whereas ion F2 at m/z ϭ pH of the mobile phase.
175.1 was assigned as the propargylic cation. The appearance of either
To further validate the identity of these types of metabolites, we
the m/z ϭ 198.1 fragment or the m/z ϭ 175.1 fragment in the CID attempted to prepare authentic standards of the N-oxidation products
spectrum of metabolite M8-1 or M8-3 was used to assign the site of through chemical oxidation. It was found that controlled oxidation of
oxidation to either the F1 or the F2 domains of the inhibitor. The 3 with m-chloroperoxybenzoic acid at low temperature generated a
major metabolite M8-1 shows the presence of the pyrimidinyl F2 mixture of three unique N-oxides that were assigned by nuclear
fragment, thus indicating that oxidation had occurred on the 4-phe- magnetic resonance analysis as the two regioisomeric pyrimidine-N-
nylpyridyl subunit (Fig. 4). The spectrum of the minor metabolite oxides and the pyridine-N-oxide. Assignment of the pyridine-N-oxide
M8-3 showed a strong fragment at m/z ϭ 198.1, signifying that the was straightforward because of a pronounced change in the chemical
biaryl domain was unchanged in the metabolite and that oxidation had shifts of the protons adjacent to the endocyclic nitrogen from 8.65 to
occurred on the pyrimidine heterocycle. Although this mode of frag- 8.25 ppm (Fig. 5). In the pyrimidine-N-oxides, no significant analo-
mentation was observed for the parent compounds (2, 3, or 7), these gous shift was observed. Careful separation of the pyridine-N-oxide
diagnostic ions did not appear in the CID spectra of the metabolites. from the two pyrimidine-N-oxides was possible using HPLC; LC-
For oxidation occurring on both the pyrimidine or 4-phenylpyridine MS/MS analysis of the mixture of N-oxides indicated that two of the
substructures of the molecule, both C- and N-oxidation products are three chemical oxidation products exactly matched two of the metab-
possible, leading to pyrimidine-N-oxides [as observed with trim- olites formed during microsomal incubation. This analysis allowed us
ethoprim (Sigel et al., 1973)], pyridine-N-oxides, as well as various to unambiguously assign the structure of M3-1 as the pyridine-N-
phenols and alcohols. A reported analysis of pyridine-N-oxides de- oxide using both retention time and CID spectra (Fig. 4). In addition,
rived from the metabolic transformation of desloratadine (Ramana- one of the two isomeric pyrimidine N-oxides was also matched to the
than, 2011) showed that two diagnostic fragments were produced by minor metabolite M3-4 (less than 5%). The remaining mono-
a loss of 17 and 18 mass units from the parent ion. Because similar oxidation product (M3-3; 25.8%) has not been unambiguously as-
diagnostic peaks were evident in the CID spectra of M3-1 and M3-4 signed, although the fragmentation pattern clearly indicates that oxi-
(see supplemental data) and are clearly not related by isotopic effects dation had occurred on the fragment containing the pyrimidine and is
(estimated to be 22.8%), as assessed by integration of the two peaks, likely the product of benzylic oxidation at C6 of the pyrimidine. The
the formation of N-oxides as a major metabolic pathway seemed chemical validation experiments provided a valuable correlation to the
probable. In comparison, the spectra of M3-3 show two peaks with assignments based on tandem mass spectrometry showing that M3-1
loss of 17 and 18 mass units, but because the integration is close and M3-4 possessed the diagnostic fragments with loss of 17 and 18
to the expected isotopic effect (23%), the presence of an N-oxide mass units, respectively.
cannot be unambiguously assigned. Moreover, the peaks associ-
Using retention time and fragmentation patterns, the remaining
ated with the N-oxides M3-1 and M3-4 eluted with retention time metabolites from compounds 2, 7, and 8 were assigned on the basis of
longer than that of the parent, whereas the demethylation product analogy to the studies with compound 3. Specifically, the primary
TABLE 3
Predicted mass, diagnostic ions and assignment of metabolites for compounds 3, 8, and 10
Compound Identifier
Predicted M ϩ H
Diagnostic Daughter Ions
Notes
3
374.1976
390.1925
360.1819
390.1925
390.1925
360.1819
376.1768
346.1663
376.1768
374.1976
390.1925
360.1819
390.1925
390.1925
390.1925
175.1; 212.1;
373.1888;
175.1; 198.1;
213.1141; 372.1836;
373.1914;
175.1; 198.1;
175.1; 214.1;
175.1; 184.1;
198.1;
175.1008; 197.0862;
175.0974 228.1016; 373.1907;
175.1021; 184.0792;
174.0863; 213.1113; 372.1768;
175.0955; 228.1010; 373.1862;
212.1076; 372.1859;
Parent
M3-1b
M3-2
M3-3b
M3-4b
8
Pyridine-N-oxidea
Demethylation
Alternative oxidation on pyrimidinyl substructure
Pyrimidinyl N-oxidea
Parent
Pyridine-N-oxide
Demethylation
M8-1
M8-2
M8-3
10b
Alternative oxidation on pyrimidinyl substructure
Parent
M10-1b
M10-2b
M10-3b
M10-4b
M10-5b
Pyridine-N-oxidea
Demethylation
Alternative oxidation on pyrimidinyl substructure
Pyrimidinyl N-oxidea
Alternative oxidation on pyrimidinyl substructure
a Structures confirmed by synthesized standards.
b Analyte masses were accurate to 10 ppm.

2006
ZHOU ET AL.
FIG. 3. Diagnostic fragments derived from
ionization of compound 8.
metabolite for all three compounds was assigned as the pyridine-N- the acetylene bridge, at either the propargylic or the C2Ј position,
oxide because it showed a longer retention time than and fragmenta- slows the rate of metabolism at the distal pyridine nitrogen likely by
tion pattern similar to that of M3-1. For all three compounds, the altering the presentation of the nitrogen to the cytochrome active site.
percentage of pyrimidine-N-oxide appeared to be minimal. Metabo- This hypothesis is supported by the observation of significant confor-
lites M2-3, M7-3, and M8-3 are likely to be products of oxidation on mational differences of compounds with varying substitutions at the
the pyrimidine substructure, such as C6 side-chain oxidation or hy- propargylic and 2Ј-phenyl positions in crystal structures with the
droxylamine formation.
target DHFR enzyme (Viswanathan et al., 2012).
Half-life and product distribution analysis of compounds 2, 3, 7,
It was anticipated that incorporation of both of these design ele-
and 8 (Table 2) suggested that remote substituents both at the propar- ments may be useful in extending the overall half-life of the lead
gylic position and the aromatic phenyl ring could influence metabo- series assuming that key biological activity could be retained. A
lism at more distal sites in the molecule. For example, comparison of hybrid inhibitor, compound 10, was predicted to bind the active site of
the metabolic profiles of compounds 3 and 7 shows that incorporation DHFR because a model of its interactions with S. aureus DHFR
of branching at the propargyl position (3) increases the half-life and shows conservation of the hydrogen bonds between the pyrimidine
decreases the relative abundance of the pyridyl-N-oxide while increas- and active site residues as well as hydrophobic interactions between
ing the degree of O-demethylation. Likewise, as noted previously, the C6 ethyl and propargyl group with Val31, Leu28, Leu54, Ile51,
comparison of compounds 7 and 8 suggests that the placement of a 2Ј and Phe92. The 2Ј methoxy group may form van der Waals interac-
methoxy substituent increases the half-life, although product distribu- tions with Met42 or Thr46. Compound 10 was synthesized according
tion is largely unaffected. These effects suggest that substitution near to the method described below (Fig. 6) to test these hypotheses.
FIG. 4. Major metabolites derived from
compounds 2, 3, 7, 8, and 10.

STABILITY AND THE PROPARGYL-LINKED ANTIFOLATES
2007
FIG. 5. Proton nuclear magnetic resonance spectra of compound 3 and the associated pyridine-N-oxide. Note the pronounced shift in the pyridyl protons (circled) upon
N-oxide formation.
Incubation of compound 10 with MLM for 1 h shows that the M10-3, and M10-5) resulted from oxidation. Once again, the presence
half-life doubles (t1/2 ϭ 65 min) relative to the initial lead series. of a 2Ј-methoxy donor promoted fragmentation along the central
LC-MS/MS analysis of the crude reaction mixture revealed that five propargylic linker to generate diagnostic fragments that were useful in
major metabolites were formed during microsomal incubation (Fig. 4; assigning sites of oxidation to the pyrimidinyl or 4-phenylpyridinyl
Table 3). One of these metabolites M10-2 (20.3%) was again easily portions of the molecule. Likewise, chemical oxidation of compound
determined to be the product of O-demethylation, whereas the mass 10 was instrumental in assigning N-oxidation products. On the basis
spectrum indicated that the other four products (M10-1, M10-2, of the analysis, the pyridine N-oxide remained the predominant me-
FIG. 6. Synthesis of compound 10. a, MeI,
K2CO3, dimethylformamide. b, MeMgBr,
THF. c, MnO2, CH2Cl2, 75% over three
steps. d, Pd(PPh3)2Cl2, Cs2CO3, dioxane,
80°C, 85%. e, Ph3P ϭ CHOMe, THF. f,
Hg(OAc)2, KI, THF/H2O, 74% over two
steps. g, dimethyl(1-diazo-2 oxopropyl)
phosphonate, K2CO3, MeOH, 80%. h,
Pd(PPh3)2Cl2, ethyl pyrimidine, CuI, Et3N,
dimethylformamide, 76%.

2008
ZHOU ET AL.
Conducted experiments: Zhou, Viswanathan, and Hill.
Performed data analysis: Zhou, Viswanathan, Anderson, and Wright.
Wrote or contributed to the writing of the manuscript: Zhou, Viswanathan,
Anderson, and Wright.
tabolite. The chemically matched pyridine-N-oxide was present at
54.3%, whereas a matched pyrimidine-N-oxide was present at 7.2%.
There were two additional mono-oxidation products localized to the
pyrimidinyl fragment (8.7 and 9.4%) that have been tentatively as-
signed to the C6 position of the pyrimidine.
References
Structural analysis of the DHFR protein bound to compound 3
clearly shows that the pyridyl nitrogen does not directly contact the
enzyme and in fact is extended toward the solvent interface, suggest-
ing that the N-oxide product may still retain enzyme inhibition activ-
ity. A sample from the chemical oxidation of compound 10 was
purified by HPLC, and the weight was determined by liquid chroma-
tography/mass spectrometry using a known quantity of compound 10
as a standard. To evaluate residual activity in the primary metabolite,
the purified sample was included in enzyme inhibition assays with C.
glabrata DHFR. We were excited to find that C. glabrata DHFR
inhibition assays with the purified pyridine-N-oxide M10-1 show that
the compound retains potency (IC50 ϭ 45 nM), suggesting the for-
mation of an active metabolite.
Beierlein JM, Frey KM, Bolstad DB, Pelphrey PM, Joska TM, Smith AE, Priestley ND, Wright
DL, and Anderson AC (2008) Synthetic and crystallographic studies of a new inhibitor series
targeting Bacillus anthracis dihydrofolate reductase. J Med Chem 51:7532–7540.
Bolstad DB, Bolstad ES, Frey KM, Wright DL, and Anderson AC (2008) Structure-based
approach to the development of potent and selective inhibitors of dihydrofolate reductase from
Cryptosporidium. J Med Chem 51:6839–6852.
Frey KM, Georgiev I, Donald BR, and Anderson AC (2010a) Predicting resistance mutations
using protein design algorithms. Proc Natl Acad Sci U S A 107:13707–13712.
Frey KM, Liu J, Lombardo MN, Bolstad DB, Wright DL, and Anderson AC (2009) Crystal
structures of wild-type and mutant methicillin-resistant Staphylococcus aureus dihydrofolate
reductase reveal an alternate conformation of NADPH that may be linked to trimethoprim
resistance. J Mol Biol 387:1298–1308.
Frey KM, Lombardo MN, Wright DL, and Anderson AC (2010b) Towards the understanding of
resistance mechanisms in clinically isolated trimethoprim-resistant, methicillin-resistant
Staphylococcus aureus dihydrofolate reductase. J Struct Biol 170:93–97.
Hill DW, Kertesz TM, Fontaine D, Friedman R, and Grant DF (2008) Mass spectral metabo-
nomics beyond elemental formula: chemical database querying by matching experimental with
computational fragmentation spectra. Anal Chem 80:5574–5582.
Huang WS, Metcalf CA, Sundaramoorthi R, Wang Y, Zou D, Thomas RM, Zhu X, Cai L, Wen
D, Liu S, et al. (2010) Discovery of 3-[2-(imidazo[1,2-b]pyridazin-3-yl)ethynyl]-4-methyl-N-
{4-[(4-methylpiperazin-1-yl)methyl]-3-(trifluoromethyl)phenyl}benzamide (AP24534), a po-
tent, orally active pan-inhibitor of breakpoint cluster region-abelson (BCR-ABL) kinase
including the T315I gatekeeper mutant. J Med Chem 53:4701–4719.
Discussion
As we have improved the potency of the propargyl-linked antifo-
lates against several bacterial and fungal species, it has become
increasingly important to consider their properties as drug leads. In
this study, we show that compounds with heterocyclic functionality
intended to bind the hydrophobic pocket have favorable solubility for
in vivo studies and are more soluble than the previous biphenyl-based
leads. Three propargyl-linked antifolates with pyridyl ring systems
were found to be chemically stable in water as well as simulated
intestinal and gastric fluids. Four promising propargyl-linked antifo-
lates that differ in the chemical environment of the alkyne were
evaluated for metabolic stability. Fragmentation analysis and the use
of chemically matched compounds provide confirmation that the
primary route of metabolism involves the formation of N-oxide prod-
ucts and that the propargyl linker appears to be stable. Results show
that branching at the propargylic position and substitution of the
aromatic ring affect metabolic stability; compounds that possess a
methyl at the propargylic position show longer half-lives compared
with otherwise similar compounds with a hydrogen at this position.
On the basis of these remote effects, a new active inhibitor was
designed and shows improved half-life.
Kerns EH and Di L (2008) Drug-like Properties: Concepts, Structure, Design and Methods,
Academic Press, Burlington.
King AO and Yasuda N (2004) Palladium-catalyzed cross-coupling reactions in the synthesis of
pharmaceuticals, in Topics Organometallic Chemistry vol 6, pp 205–245, Springer-Verlag,
Heidelberg.
Lindberg J, Sigurdsson S, Lo¨wgren S, Andersson HO, Sahlberg C, Nore´en R, Fridborg K, Zhang
H, and Unge T (2002) Structural basis for the inhibitory efficacy of efavirenz (DMP-266),
MSC194 and PNU142721 towards the HIV-1 RT K103N mutant. Eur J Biochem 269:1670–
1677.
Liu J, Bolstad DB, Smith AE, Priestley ND, Wright DL, and Anderson AC (2008) Structure-
guided development of efficacious antifungal agents targeting Candida glabrata dihydrofolate
reductase. Chem Biol 15:990–996.
Liu J, Bolstad DB, Smith AE, Priestley ND, Wright DL, and Anderson AC (2009) Probing the
active site of Candida glabrata dihydrofolate reductase with high resolution crystal structures
and the synthesis of new inhibitors. Chem Biol Drug Des 73:62–74.
O’Hare T, Shakespeare WC, Zhu X, Eide CA, Rivera VM, Wang F, Adrian LT, Zhou T, Huang
WS, Xu Q, et al. (2009) AP24534, a pan-BCR-ABL inhibitor for chronic myeloid leukemia,
potently inhibits the T315I mutant and overcomes mutation-based resistance. Cancer Cell
16:401–412.
Paulsen JL, Bendel SD, and Anderson AC (2011) Crystal structures of Candida albicans
dihydrofolate reductase bound to propargyl-linked antifolates reveal the flexibility of active
site loop residues critical for ligand potency and selectivity. Chem Biol Drug Des 78:505–512.
Paulsen JL, Liu J, Bolstad DB, Smith AE, Priestley ND, Wright DL, and Anderson AC (2009)
In vitro biological activity and structural analysis of 2,4-diamino-5-(2Ј-arylpropargyl)pyrimi-
dine inhibitors of Candida albicans. Bioorg Med Chem 17:4866–4872.
Pelphrey PM, Popov VM, Joska TM, Beierlein JM, Bolstad ES, Fillingham YA, Wright DL, and
Anderson AC (2007) Highly efficient ligands for dihydrofolate reductase from Cryptospo-
ridium hominis and Toxoplasma gondii inspired by structural analysis. J Med Chem 50:940–
950.
Ramanathan R (2011) Mass Spectrometry in Drug Metabolism and Pharmacokinetics, John
Wiley & Sons, Hoboken.
This new class of antifolates uses a somewhat uncommon acety-
lenic linker to bridge the pyrimidine head group and hydrophobic
biaryl domains. The unique geometry and steric environment of the
alkyne are central to the efficacy of the inhibitors and are not easily
mimicked by other functionality. As such, determining the stability of
this unsaturated unit is critical for further development of these
compounds. These studies indicate that the linkage is chemically
stable and is not a primary site of metabolic transformation.
Sigel CW, Grace ME, and Nichol CA (1973) Metabolism of trimethoprim in man and measure-
ment of a new metabolite: a new fluorescence assay. J Infect Dis 128:S580–S583.
Viswanathan K, Frey KM, Scocchera EW, Martin BD, Swain PW, Alverson JB, Priestley ND,
Anderson AC, and Wright DL (2012) Toward new therapeutics for skin and soft tissue
infections: propargyl-linked antifolates are potent inhibitors of MRSA and Streptococcus
pyogenes. PLoS ONE 7:e29434.
Address correspondence to: Dennis L. Wright, Department of Pharmaceuti-
cal Sciences, 69 N. Eagleville Rd., University of Connecticut, Storrs, CT 06269.
E-mail: dennis.wright@uconn.edu
Authorship Contributions
Participated in research design: Zhou, Viswanathan, Anderson, and
Wright.