Fragment-based screening by X-ray crystallography, MS and isothermal titration calorimetry to identify PNMT (phenylethanolamine N-methyltransferase) inhibitors

Article abstract of DOI:10.1042/BJ20100651

CNS (central nervous system) adrenaline (epinephrine) is implicated in a wide range of physiological and pathological conditions. PNMT (phenylethanolamine N-methyltransferase) catalyses the final step in the biosynthesis of adrenaline, the conversion of n

Full text of DOI:10.1042/BJ20100651

www.biochemj.org
Biochem. J. (2010) 431, 51–61 (Printed in Great Britain) doi:10.1042/BJ20100651
51
Fragment-based screening by X-ray crystallography, MS and isothermal
titration calorimetry to identify PNMT (phenylethanolamine
N-methyltransferase) inhibitors
Nyssa DRINKWATER*, Hoan VU†, Kimberly M. LOVELL‡, Kevin R. CRISCIONE‡, Brett M. COLLINS*, Thomas E. PRISINZANO‡,
Sally-Ann POULSEN†, Michael J. MCLEISH§, Gary L. GRUNEWALD‡ and Jennifer L. MARTIN*1
*University of Queensland, Institute for Molecular Bioscience, Division of Chemistry and Structural Biology, Brisbane, Queensland 4072, Australia, †Griffith University, Eskitis Institute,
Brisbane, Queensland 4111, Australia, ‡Department of Medicinal Chemistry, University of Kansas, Lawrence, KS 66045-7582, U.S.A., and §Department of Chemistry and Chemical
Biology, Indiana University–Purdue University Indianapolis (IUPUI), 402 N. Blackford St, LD 326D, IN 46202, U.S.A.
CNS (central nervous system) adrenaline (epinephrine) is
implicated in a wide range of physiological and pathological
conditions. PNMT (phenylethanolamine N-methyltransferase)
catalyses the final step in the biosynthesis of adrenaline, the
conversion of noradrenaline (norepinephrine) to adrenaline by
methylation. To help elucidate the role of CNS adrenaline, and to
develop potential drug leads, potent, selective and CNS-active
inhibitors are required. The fragment screening approach has
advantages over other lead discovery methods including high
hit rates, more efficient hits and the ability to sample chemical
diversity more easily. In the present study we applied fragment-
based screening approaches to the enzyme PNMT. We used
crystallography as the primary screen and identified 12 hits from
a small commercial library of 384 drug-like fragments. The hits
include nine chemicals with two fused rings and three single-ring
chemical systems. Eight of the hits come from three chemical
classes: benzimidazoles (a known class of PNMT inhibitor),
purines and quinolines. Nine of the hits have measurable binding
affinities (∼5–700 μM) as determined by isothermal titration
calorimetry and all nine have ligand efficiencies of 0.39 kcal/mol
per heavy atom or better (1 kcal ≈ 4.184 kJ). We synthesized
five elaborated benzimidazole compounds and characterized their
binding to PNMT, showing for the first time how this class of
inhibitors interact with the noradrenaline-binding site. Finally, we
performed a pilot study with PNMT for fragment-based screening
by MS showing that this approach could be used as a fast and
efficient first-pass screening method prior to characterization of
binding mode and affinity of hits.
Key words: benzimidazole, catecholamine, drug discovery,
enzyme inhibition, fragment-based screening, phenylethanol-
amine N-methyltransferase (PNMT).
FBS-X (FBS by X-ray crystallography) provides the greatest
level of information for the elaboration of hit compounds [8].
However, costly infrastructure for crystallization, crystal handling
and data collection is required for screening hundreds of crystals
in a high-throughput manner. Although many different FBS
methods are now used, there have been few reports comparing hits
identified against the same specific target by orthogonal methods.
NMR is frequently employed as a first-pass screen before detailed
structural characterization of hits by X-ray crystallography, and
MS has the potential to be applied in a similar manner [9].
Comparative FBS-X and FBS-MS campaigns have not been
reported, as far as we are aware, yet there are potentially significant
advantages to a combined MS/X-ray screening approach due to
their complementary attributes.
Pharmaceutical companies have developed and applied
FBS-X with success [10]. An ideal target for FBS-X has an
accessible active site, can be crystallized in the apo form and
requires only minor conformational changes in the crystal form
to accommodate ligand binding. However, many drug targets
do not conform to these requirements, including the human
enzyme PNMT (phenylethanolamine N-methyltransferase; EC
INTRODUCTION
The identification of suitable starting compounds for development
into drugs is a major challenge for drug discovery and
development programmes. FBS (fragment-based screening) has
been developed as a method for the rapid discovery and
subsequent elaboration of hits into quality lead compounds [1,2].
This approach is based on the principle of screening simple
low-molecular-mass compounds that make a few high-quality
interactions with the target site. Resulting hits tend to bind to
the targets efficiently yet with low affinity and, with a practically
designed library, are readily tractable for follow up chemistry [3].
These fragment hits are used as ‘anchors’ that can be elaborated
to fill additional regions of the target site. This results in potent
compounds with minimal chemical complexity that can then be
developed into drugs. A major challenge in establishing FBS has
been the development of sensitive methods to detect the weak
binding of hits (with typically milli- to micro-molar binding
affinities) [4]. Methods currently used include ITC (isothermal
titration calorimetry), NMR, surface plasmon resonance, MS and
X-ray crystallography [1,5–7].
Abbreviations used: AdoHcy, S-adenosyl-L-homocysteine; AdoMet, S-adenosyl-L-methionine; CCD, charge-coupled-device; CNS, central nervous
system; ESI, electrospray ionization, FBS, fragment-based screening, FBS-X, FBS by X-ray crystallography; FTMS, Fourier transform MS; HA, heavy atom;
HRMS, high-resolution MS; ITC, isothermal titration calorimetry; LE, ligand-binding efficiency; m.p., melting point; MSD, Merck Sharp and Dohme Research
Laboratories; PNMT, phenylethanolamine N-methyltransferase; hPNMT, human PNMT.
1
To whom correspondence should be addressed (email j.martin@imb.uq.edu.au).
The co-ordinates and structure factors for all 17 crystal structures have been deposited with the PDB under accession codes 3KPJ, 3KPU, 3KPV, 3KPW,
3KPY, 3KQM, 3KQS, 3KQT, 3KQV, 3KQW, 3KQO, 3KQP, 3KQQ, 3KQY, 3KR0, 3KR1 and 3KR2.
ꢀ
c
ꢀ
c
The Authors Journal compilation 2010 Biochemical Society

52
N. Drinkwater and others
benzimidazole class. Our pilot study showed that the orthogonal
screening method of FBS-MS could be used as an even more
efficient first step in an FBS campaign.
EXPERIMENTAL
Materials
The FBS library was purchased from ActiveSight (a Rigaku
company), and comprised 384 compounds with a mean molecular
mass of 142 Da and of predominantly rigid low-complexity
(see the Supplementary Table Library at http://www.BiochemJ.
org/bj/431/bj4310051add.htm); all compounds were provided
as 200 mM stock solutions in 100% DMSO and as cocktails
of four fragments each at 50 mM in 100% DMSO. All the
compounds are also available for purchase directly from Sigma–
Aldrich. For chemical syntheses, all reagents and solvents used
were reagent grade and of the highest purity commercially
available. 4-Methoxybenzene-1,2-diamine was sourced from Alfa
Aesar, all other compounds and reagents were sourced from
Sigma–Aldrich.
Crystallization of hPNMT
Figure 1 PNMT catalytic reaction
C-terminally His6-tagged hPNMT was expressed and purified
as described previously [18]. The protein was concentrated to
50 mg/ml and mixed with AdoHcy (to a final concentration of
2 mM) taking the final concentration of protein to 40 mg/ml or
1.25 mM. The protein mixture was crystallized by hanging drop
vapour diffusion using QIAGEN EasyXtal Tool crystallization
trays for ease of soaking. Drops comprised 1.5 μl of protein
and 1.5 μl of precipitant equilibrated over 400 μl of precipitant
solution {4–8% PEG [poly(ethylene glycol)] 6000, 0.25 M LiCl
and 0.1 M sodium cacodylate, pH 5.5–6.0}.
(A) The reaction catalysed by PNMT involves transfer of
a methyl group from
AdoMet to noradrenaline, forming adrenaline and AdoHcy. (B) The crystal structure of
hPNMT–AdoHcy–SK&F29661, shown in dark grey and light grey space-filling representation
respectively, are AdoHcy and SK&F29661; the Cα trace of hPNMT is shown predominantly in
light grey, with the regions covering the AdoHcy- and SK&F29661-binding sites shown in dark
grey. The boxed region shows a close-up of the AdoHcy- and SK&F29661-binding sites.
2.1.1.28; 30.7 kDa), which catalyses the final step in adrenaline
(epinephrine) biosynthesis, the methylation of noradrenaline
(norepinephrine) to form adrenaline (Figure 1A). PNMT
has a potential role in a wide range of disease-relevant
processes including the central control of blood pressure [11],
pituitary hormone secretion [12], ethanol intoxication [13],
Parkinson’s disease [14] and the neurodegeneration observed
in Alzheimer’s disease [15]. Structures of recombinant hPNMT
(human PNMT) in complex with inhibitors and substrates (e.g.
PDB code 1HNN) show that hPNMT has an enclosed active site
(Figure 1B) suggesting that enzyme conformational changes may
be needed for ligands to bind [16]. Apo crystals are not readily
available because we were unable to crystallize the enzyme in the
absence of AdoHcy (S-adenosyl-L-homocysteine), which binds to
the cofactor-binding site. Furthermore, when crystals are grown
in the absence of an inhibitor or substrate at the noradrenaline-
binding site, this site is occupied by a phosphate molecule [17].
However, we were encouraged to attempt FBS-X because we
solved the crystal structure of hPNMT (PDB code 3HCD) in
complex with the physiological substrate R-noradrenaline by
using rapid soaking methods on hPNMT–AdoHcy–PO4 crystals
[17]. We therefore proposed to apply this same soaking method
for FBS-X screening of hPNMT, with the aim of identifying novel
chemical frameworks for inhibitor design. We also performed an
FBS-MS pilot study to determine whether this approach could be
used as a more efficient first-pass screen prior to crystallographic
analysis of selected hits.
FBS by X-ray crystallography
Library screening was conducted blind, in that we made no
assumptions about the contents of the library, or what compounds
might be expected to bind PNMT. Prior to screening, control
experiments were performed using known hPNMT inhibitors to
confirm that ligands could be soaked into hPNMT–AdoHcy–
PO4 crystals and that the solvents used for fragment screening
and cryoprotection did not interfere with binding. Crystals
(∼0.25 mm in each dimension) were transferred into 1 μl of
soaking solution. Crystal soak solutions were prepared by adding
1 μl of fragment cocktail to 9 μl of stabilizing solution, taking
the final concentrations to 6.6% PEG 6000, 0.3 M LiCl, 0.1 M
sodium cacodylate, pH 5.7, 4× 5 mM fragments and 10% (v/v)
DMSO. Crystals were soaked for 15 min.
Crystals were cryoprotected by transferring directly from the
soak solution into mother liquor supplemented with 25% ethylene
glycol for 10–15 s before being flash-cooled. Data for hPNMT–
AdoHcy–PO4 were measured from a crystal transferred directly
into the cryoprotectant solution without undergoing a soak step.
X-ray diffraction data from hPNMT crystals were measured using
the UQ ROCX Diffraction Facility or the Australian Synchrotron.
At the UQ ROCX Diffraction Facility, crystals were irradiated
with X-rays of wavelength 1.542 Å (1 Å = 0.1 nm) generated
from a Rigaku FR-E Superbright copper rotating anode generator
operating at 45 kV, 45 mA and data were measured using
one of two alternate methods: (i) crystals were flash-cooled
by plunging into liquid nitrogen and placed on the camera
using a Rigaku ACTORTM crystal mounting robot, X-rays were
focused with Osmic Vari-Max HF optics and diffraction data
Overall, we showed that FBS-X can be successfully imple-
mented within an academic environment. Our crystallographic
screening of the non-ideal enzyme target hPNMT identified 12
hits, including three benzimidazoles (a known class of PNMT
inhibitors), and nine of the hits could be characterized by ITC.
Five compounds were synthesized to further characterize the
ꢀ
c
ꢀ
c
The Authors Journal compilation 2010 Biochemical Society

Fragment-based screening to identify PNMT inhibitors
53
measured on a Rigaku Saturn 944 CCD (charge-coupled-device)
detector; or (ii) crystals were flash-cooled on the camera in
the gaseous nitrogen stream at 100K, X-rays were focused
with Osmic HiRes2 Confocal Max-FluxTM optics and diffraction
data measured with a Rigaku R-AXIS IV++ imaging plate area
detector. Data collected at UQ ROCX Diffraction Facility were
processed using Crystal Clear (Rigaku). Data collected at the
Australian Synchrotron were measured at 100K with X-rays of
wavelength 0.95667 Å on beamline 3BM1 [19] using a MAR
165 CCD detector, and processed using HKL2000 [20]. Crystals
diffracting to better than 2.7 Å resolution were used for data
collection.
Phasing and automatic structure solution were carried out
using MIfit (ActiveSight). The structures were solved by
molecular replacement using the structure of hPNMT–AdoHcy–
SK&F29661 (PDB code 1HNN), with the inhibitor and
waters removed, as the starting model. Active-site density was
individually evaluated and assessed for the presence or absence
of a ligand. If density indicating a bound fragment was present,
ligands were modelled manually using COOT [21] and co-
ordinates and refinement parameters were generated by PRODRG
[22] and PHENIX [23]. Further refinement was carried out using
PHENIX [23]. Cross-validation was done by R-free analysis from
5% of the reflections set aside from refinement.
Where density was present at the active site but did
not unambiguously identify the bound ligand from the four
compounds in the cocktail, a deconvolution step was performed.
This involved collecting data from crystals soaked in solutions
containing the individual compounds (rather than the cocktail
of four compounds). In these cases, the final concentration of
the compounds in the soak solution was increased to 20 mM
rather than the 5 mM concentration used in the cocktail screen.
The higher concentration generally resulted in improved density.
Deconvolution also allowed identification of the binding of more
than one fragment per cocktail.
FBS by MS
MS was performed on an APEX® III 4.7 Tesla FTICR mass
spectrometer (Bruker Daltonics) fitted with an ApolloTM ESI
(electrospray ionization) source operated in positive ion mode.
XMASS NT V6.1.2 MS software on a PC platform was used
for data acquisition. Broadband excitation was used to analyse
a mass range from m/z 50–6000 and each spectrum was an
average of 64 transients (scans) with 512000 data points acquired
in low-resolution mode with an acquisition time of approx.
4 min/sample. Samples were infused into the ESI source at
2 μl/min. Nitrogen was used as both the drying gas (125◦C) and
nebulizing gas. Relevant parameters include a 104-fold pressure
reduction between source and analyser regions with an ESI
source pressure (10−7 kPa) and high-vacuum analyser region
pressure (6 × 10−11 kPa). Agilent ESI tuning mix (Santa Clara)
was used for an external three-point calibration. The hexapole ion
accumulation time was 3 s.
To confirm correct ESI-MS parameter adjustment, para-
meters were optimized using
a known hPNMT ligand
(7-aminosulfonyl-3-ethyl-1,2,3,4-tetrahydroisoquinoline; Ki of
1.8 μM) [24]. Fragments analysed (Supplementary Figure
S1 at http://www.BiochemJ.org/bj/431/bj4310051add.htm) were
selected to include a range of FBS-X-positive and -negative hits.
A total of 12 fragments were provided with no indication as
to which compounds were hits by FBS-X. A stock solution of
31.5 μM hPNMT was prepared in 10 mM ammonium acetate
(pH 7); stock solutions of AdoHcy and each of the 12 fragments
(representing three of the FBS-X cocktails) were prepared in
methanol. Aliquots of all solutions were used immediately or
stored at −30◦C. Samples for MS analysis were prepared as
follows: 3 μM hPNMT, 0 or 30 μM AdoHcy and 300 μM
(100 equiv.) fragment in a total volume of 100 μl. Samples were
incubated at room temperature (25◦C) for 30 min prior to MS
analysis. A total of 24 MS datasets were acquired to analyse the
12 fragments (i.e. with and without added AdoHcy cofactor).
The protein consumption per 100 μl sample was 0.01 mg,
however, this sample volume was prepared for convenience and
could be reduced as the consumed sample volume was 8 μl per
MS acquisition (4 min acquisition at 2 μl/min).
Structure determination of hPNMT in complex with elaborated
benzimidazoles
If a hPNMT–fragment or hPNMT–AdoHcy– fragment non-
covalent complex is formed in solution then the complex
was observed in the ESI mass spectrum. The mass difference
between the peaks for the unbound hPNMT and the hPNMT–
fragment (ꢀm/z) can be multiplied by the charge state to give
directly the molecular mass of the binding fragment i.e. Mr
fragment = ꢀm/z × z to provide confirmation of the identity of the
hit fragment. Complexes with AdoHcy were similarly treated to
identify and confirm the fragment hits. ESI-FT (Fourier transform)
MS analysis of a solution containing 3 μM hPNMT yielded the
ESI positive ion mass spectrum shown in Supplementary Figure
S2(A) at http://www.BiochemJ.org/bj/431/bj4310051add.htm.
Peaks corresponding to the 13+ to 10+ charge states of hPNMT
were observed. The ESI-FTMS spectrum of 3 μM hPNMT with
added 30 μM AdoHcy yielded the ESI positive ion mass spectrum
shown in Supplementary Figure S2(B) and, as an example, the
ESI-FTMS mass spectrum of a hPNMT–AdoHcy–hit (compound
6) solution is shown in Supplementary Figure S2(C). In this
spectrum each charge state consists of a peak corresponding
to hPNMT–AdoHcy and a peak corresponding to the hPNMT–
AdoHcy–fragment non-covalent complex. Actual mass values
of the peaks observed in Supplementary Figure S2(C) and
calculation of the mass of the fragment hit with compound 6
are given in Supplementary Table S1 at http://www.BiochemJ.
org/bj/431/bj4310051add.htm.
Compounds 13–17 were designed and synthesized (see below
for details of synthesis) to further characterize the binding
of benzimidazoles to hPNMT. The crystal structures of these
compounds in complex with hPNMT were determined as follows.
Crystals of hPNMT–AdoHcy–PO4 were soaked in a solution
of the synthesized compounds as described above for fragment
screening. In this case, compounds were solubilized at 100 mM
in 50% DMSO/water and used to prepare soak solutions
of final concentrations 6.6% PEG 6000, 0.3 M LiCl, 0.1 M
sodium cacodylate, pH 5.7, 5–10 mM compounds and 2.5–5%
DMSO.
Crystals were cryoprotected, diffraction data collected and
processed, and structures solved and refined at the UQ ROCX
diffraction facility using method (ii) described above, with the
exception that phasing was carried out using PHENIX [23]
and structures were solved by difference Fourier methods. The
procedure in this case was to model and refine the structure of
the protein first, followed by addition of water molecules, then
AdoHcy and finally the ligand. The criteria used to include a water
molecule were: the presence of 2Fo−Fc density at 1σ and Fo−Fc
density at 3σ and at least one possible hydrogen bond within
3.2 Å. Cross-validation was performed as described above. All of
the crystal structure electron density maps revealed density indic-
ative of AdoHcy at the cofactor-binding site and density consistent
with a benzimidazole bound at the noradrenaline-binding site.
ꢀ
c
ꢀ
c
The Authors Journal compilation 2010 Biochemical Society

54
N. Drinkwater and others
bromide. The solution was refluxed at 100◦C for 16 h. The
reaction was washed with diethyl ether and evaporated under
reduced pressure to yield the crude compound 13 · HBr as a brown
solid. Trituration from 2-propanol/hexane yielded a viscous brown
liquid that was dried in vacuo to yield 13 · HBr as a brown solid
(0.24 g, 66%), m.p. >250◦C: 1H-NMR (500 MHz, MeOD) δ 7.07
(t, J = 8.1, 1H), 6.84 (d, J = 7.5, 1H), 6.70 (dd, J = 0.6, 8.1, 1H);
13C-NMR (126 MHz, MeOD) δ 151.8, 145.0, 132.7, 125.9, 119.7,
110.8, 103.6; HRMS (m/z): [M+H] calculated for C7H7N3OBr,
227.9772, found 227.9775.
ITC
Compounds found to bind hPNMT by FBS-X or FBS-MS were
subsequently analysed by ITC. ITC was carried out using a high-
sensitivity MicroCal (GE Healthcare) iTC200 system at 25◦C.
Purified protein at a concentration of 200 μM plus 600 μM
AdoMet (S-adenosyl-L-methionine) in 20 mM Tris/HCl, pH 7.2,
1 mM EDTA and 4.3% DMSO were loaded into the sample
cell (Vcell, reaction cell volume of ≈ 200 μl). DMSO in the
sample was calibrated to match the residual DMSO after fragment
dilutions. The fragment hit at 1.5–3 mM in the same buffer was
then titrated into the sample cell with 20 × 2.5 μl injections. The
heat released was measured and integrated using MicroCal Origin
7.0 software. The association constant (Ka; 1/Kd), enthalpy (ꢀH)
and stoichiometry (N) were calculated using a single-site binding
model. Measurements were repeated 2–4 times for each ligand
and errors calculated as the S.D. from replicate experiments. In
each case the protein concentration was fixed as a known quantity
and the apparent ligand concentration was adjusted slightly during
curve fitting so that the binding stoichiometry was approx. 1 (i.e.
an assumption of 1:1 binding was made).
The energies of interaction were calculated from Ka values
using the Gibbs free energy equation: ꢀG = −RT(ln Ka), where
R is the gas constant (1.9872 kcal/mol; 1 kcal ≈ 4.184 kJ), T is the
temperature of the experiment (298 K) and Ka is the association
constant for the compound. LE (ligand-binding efficiency) values
were calculated according to LE = −ꢀG/HA, where ꢀG is the
free energy of binding (kcal/mol) and HA is the number of heavy
atoms (non-hydrogen atoms) in the compound.
2-Amino-5(6)-chloro-7(4)-hydroxybenzimidazole · hydrobromide (compound
17 · HBr)
Reduction [26] of 4-chloro-2-methoxy-6-nitroaniline [27]
(compound 19) was performed by dissolving compound 19
(0.30 g, 1.8 mmol) in methanol (40 ml) and adding 30 mg of
10% palladium on carbon. The mixture was stirred under
hydrogen for 16 h. The palladium was removed by filtration over
Celite®. The Celite® was washed with methanol and ethyl acetate.
The combined organic washes were concentrated under reduced
pressure to a viscous brown liquid. Column chromatography
(silica gel, ethyl acetate/hexanes of 3:7) yielded 5-chloro-3-
methoxybenzene-1,2-diamine (compound 20) as a brown solid
(0.18 g, 58%) that was used for the next synthetic step without
further purification: m.p., 97–99◦C; 1H-NMR (500 MHz, CDCl3)
δ 6.39 (dd, J = 2.1, 4.4, 2H), 3.82 (s, 3H), 3.42 (s, 4H); 13C-
NMR (126 MHz, C2HCl3) δ 149.0, 136.4, 124.5, 122.0, 109.7,
103.3, 56.1; HRMS (m/z): [M+H] calculated for C7H10ClN2O,
173.0482; found 173.0373.
Diamine compound 20 (0.18 g, 1.0 mmol) was dissolved
in acetonitrile (20 ml) and water (5 ml) and cooled to 0◦C.
Cyanogen bromide in acetonitrile (5M, 0.20 ml, 1.0 mmol)
was added dropwise over 5 min. The reaction mixture was
warmed to room temperature and stirred overnight. The solvents
were evaporated under reduced pressure to give 2-amino-5(6)-
chloro-7(4)-methoxybenzimidazole · hydrobromide (compound
21 · HBr) as a brown solid (0.30 g), which was used for the next
Compound synthesis
Benzimidazoles (compounds 13–17, see Figure 5) were
synthesized using procedures published previously with minor
modifications as indicated (see the Supplementary Experimental
section at http://www.BiochemJ.org/bj/431/bj4310051add.htm).
Generally, an appropriately substituted nitroaniline was reduced
by palladium-catalysed hydrogenation to the corresponding
phenylenediamine. The phenylenediamine derivative was con-
densed with cyanogen bromide to form the desired 2-aminobenzi-
midazole as the hydrobromide salt. Methoxy-substituted 2-
aminobenzimidazoles were converted to their corresponding
phenol derivatives using 48% hydrogen bromide. Compounds
13 and 17 have not been reported previously and details of
their synthesis are reported below. Compounds 14–16 are known
compounds and the details of our syntheses of them are given in
the Supplementary Experimental section.
The m.p. (melting point) was determined in open capillary
tubes on a Thomas–Hoover melting point apparatus calibrated
with known compounds, but are uncorrected. Proton (1H-NMR)
and carbon (13C-NMR) NMR spectra were taken on a Bruker
DRX-500 spectrophotometer. Proton chemical shifts are reported
in p.p.m. relative to TMS (tetramethylsilane; 0.00 p.p.m.) or to
methanol-d4 (3.31 p.p.m.). Carbon chemical shifts are reported
in p.p.m. relative to C2HCl3 (77.2 p.p.m.) or methanol-d4
(49.1 p.p.m.). HRMS (high-resolution MS) was performed on
a Ribermag R 10-10 mass spectrophotometer. Hexane refers to
the mixture of hexane isomers (boiling point of 68–70◦C).
1
synthetic step without further purification, m.p., 68–70◦C: H-
NMR (500 MHz, MeOD) δ 7.01 (d, J = 1.6, 1H), 6.94 (d, J = 1.6,
1H), 3.98 (s, 3H); 13C-NMR (126 MHz, MeOD) δ 152.6, 147.5,
132.8, 131.2, 119.4, 107.9, 105.6, 57.1; HRMS (m/z): [M+H]
calculated for C8H9ClN3O, 198.0434; found 198.0383.
Ether cleavage of compound 21 · HBr was carried out using the
procedure given for the synthesis of compound 13 · HBr above,
yielding compound 17 · HBr as a brown solid (0.10 g, 36%),
1
m.p. >250◦C: H-NMR (500 MHz, MeOD) δ 6.88 (d, J = 1.7,
1H), 6.72 (d, J = 1.7, 1H); 13C-NMR (126 MHz, MeOD) δ 152.3,
145.5, 133.1, 130.7, 118.8, 111.2, 104.0; HRMS (m/z): [M+H]
calculated for C7H7N3OCl, 184.0278; found 184.0266.
RESULTS
FBS-X
For the FBS-X experiments, we used crystals of hPNMT grown
in the absence of inhibitor or ligand for the noradrenaline-
binding site. The crystal structure resulting from such crystals
reveals a phosphate molecule bound in the noradrenaline-binding
site (Figure 2); the phosphate apparently co-purifies with the
enzyme during an affinity purification step. The phosphate forms
favourable interactions with the side chains of Lys57 and Asn39.
For a hit to be detected in the FBS-X experiments, one or
more of the four fragments of the cocktail must displace the
2-Amino-4(7)-hydroxybenzimidazole · hydrobromide (compound 13 · HBr)
Ether-mediated cleavage [25] of 2-amino-4(7)-methoxy-
benzimidazole [25] (compound 18) was performed by dissolving
compound 18 (0.40 g, 2.4 mmol) in 20 ml of 48% hydrogen
ꢀ
c
ꢀ
c
The Authors Journal compilation 2010 Biochemical Society

Fragment-based screening to identify PNMT inhibitors
55
included as Supplementary Table S2. The structures of the 12
hits and their binding mode to hPNMT are shown in Figure 3.
Analysis of fragment hits
The 12 fragment hits all bind the noradrenaline-binding pocket of
hPNMT, the same site as other structurally characterized hPNMT
inhibitors. Furthermore, all 12 fragments bind within the same
region of the pocket (Figure 4A). All 12 fragment hits contain
either or both a five- or six-membered aromatic ring, sandwiched
between the side chains of Phe182 and Asn39. In this position, the
rings form π–π stacking interactions with the side chain of Phe182
.
The importance of this interaction is evident in the structure of
6-chloro-oxindole (compound 4) bound to hPNMT (Figure 3).
Unlike other ligand-bound hPNMT structures, there are no direct
hydrogen-bond interactions formed between compound 4 and
hPNMT. In addition to the π–π stacking interactions between
the benzene ring and Phe182, the only other interaction that might
contribute to binding of compound 4 is a hydrophobic interaction
Figure 2 hPNMT noradrenaline-binding site
The hPNMT target binding site for inhibitors is shown for the enzyme crystallized in the absence
of inhibitor/substrate. This crystal form was used for FBS-X screening. The phosphate molecule
in the noradrenaline site must be displaced by a soaked fragment for a hit to be detected.
between the chlorine atom and the aromatic ring of Tyr222
.
Table 1 Statistics for FBS-X screening on hPNMT
Three chemical frameworks are present in more than one
fragment hit: (i) the benzimidazole core (compounds 6, 7, and 9);
(ii) the purine core (compounds 2 and 10); and (iii) the quinoline
core (compounds 1, 3 and 11). Despite different substituents, the
benzimidazole-based fragments all bind in the same orientation
within the active site. In contrast, the binding modes of the
two purine-based fragments and the three quinoline fragments
vary. This variation allows different functional groups to make
favourable interactions (e.g. see compounds 2 and 10, in Figure 3)
demonstrating that the presence of substituents can alter the
binding pose of a core fragment. It should also be noted that
the library used for FBS-X screening contains a further ten
benzimidazole compounds, none of which were found to bind the
noradrenaline site. Similarly, the screening library also contained
other quinolines and purines that did not bind to the target
site under the conditions used for FBS-X. The observation that
different substituents affect binding of compounds to PNMT
provides further validation for the specificity of hits identified
by FBS-X and may be useful for structure–activity relationship
analysis in the development of more potent inhibitors.
Superposition of the fragment hits demonstrates how opposite
ends of the noradrenaline pocket bind different chemical
functionalities (Figures 4A and 4B). Hydrophilic nitrogen and
oxygen atoms of hit compounds are clustered at one end of the
pocket, forming direct and water-mediated interactions with
the active-site residues Glu185, Glu219, Asp267 and the main chain of
Phe182. In contrast, carbon and chlorine atoms of hit compounds
are clustered at a hydrophobic region of the pocket consisting
of the side chains of Val53, Met258, Val269 and Val272. These
same amino acid residues are involved in binding characterized
previously between inhibitors and substrates [28–30].
The hit rate is calculated on the basis of 12 hits from 376 screened compounds.
Parameter
Results
Number of compounds in library/number screened
Number of datasets collected in first round screen
Number of datasets collected in deconvolution steps
Average dataset resolution
Average Rwork/Rfree
Number of hits
384/376
94
51
2.5 A
˚
0.232/0.298
12
Hit rate
3.2%
˚
Average resolution of hit structures
Average Rwork/Rfree of hit structures
2.4 A
0.255/0.281
phosphate during the 15 min soaking step. Data collection and
refinement statistics for a representative crystal structure of
hPNMT–AdoHcy–PO4 are included in Supplementary Table S2
at http://www.BiochemJ.org/bj/431/bj4310051add.htm.
Of the 96 cocktails containing four compounds (384
compounds in total) in the library, X-ray diffraction data were
collected for 94 cocktails (Table 1). Two of the 96 cocktails caused
crystal cracking during the soak, preventing high-resolution
diffraction data measurement. Owing to the high-throughput
nature of the screen, these two cocktails were not considered
further. Evaluation of the eight compounds in these two cocktails
will be investigated in another study. A total of 16 of the 94
cocktail soaks gave electron density suggestive of hits, i.e. binding
of a chemical to the noradrenaline-binding site. The chemical
identity of two of the hits was obvious from the initial X-ray
data. An additional 51 datasets were measured to deconvolute
hits where it was not clear which compound of the cocktail
had bound or to measure higher resolution data for confirmed
hits. During this process, four of the initially identified hits
were found to be false positives. In summary, a total of 12
compounds (out of a possible 376) were identified as binding
in the target site of hPNMT, giving a hit rate of 3.2% for the
FBS-X process. The 2Fo−Fc maps showing the ligand density
for these 12 hits are provided as Supplementary Figure S3 at
http://www.BiochemJ.org/bj/431/bj4310051add.htm. Summary
statistics for the X-ray data are given in Table 1 and data collection
and refinement statistics for the 12 hit crystal structures are
In some cases, we found unfavourable contacts between the
fragments and hPNMT residues. Many unfavourable interactions
resulted from clashes with the hydroxy group of Tyr35, which is
close to the hydrophobic residues Val53, Met258, Val269 and Val272
.
We also observed cases where the electron density for all or part
of the fragment was weak. For example, density for the flexible
chain of formanilide (compound 8) is weak. In two other cases,
1-aminoisoquinoline (compound 3) and 2-hydroxynicotinic acid
(compound 12), the fragments adopt different binding poses in
the two hPNMT molecules in the asymmetric unit. This may be
due to a lower affinity of these ligands compared with other hits.
Finally, for 4-bromo-1H-imidazole (compound 5), it was difficult
to determine from the electron density alone the correct position
ꢀ
c
ꢀ
c
The Authors Journal compilation 2010 Biochemical Society

56
N. Drinkwater and others
Figure 3 Summary of the results of FBS-X targeting hPNMT
The 12 fragment hits were identified, and the chemical structures and binding poses of each are shown. Hydrogen bond interactions are shown as black and hydrophobic interactions are shown as
+
orange broken lines. Binding affinities determined by ITC (Kd in μM S.D.) and calculated LE values (in kcal/mol per HA) are given for each hit.
−
of the ring nitrogen atoms because the compound fits the density
equally well in two orientations, rotated 180◦ around the long
axis. Compound 5 was modelled in one orientation on the basis
of predicted favourable interactions (see Figure 3).
that these regions favour ligand interactions and would probably
be amenable to interactions with elaborated inhibitors.
Water molecules in the active site are important because
they can indicate potential areas for favourable elaboration of
fragments. We found that many water molecules are conserved
across the 12 enzyme–fragment structures (Figure 4C). There
are five major water-binding sites, four of which are located in
the hydrophilic region of the binding pocket described above.
These four sites, show the greatest level of conservation across
the 12 structures, and mediate interactions with Glu185, Glu219 and
Asp267. The presence of defined water-binding pockets suggests
FBS by MS
MS has high sensitivity and a wide dynamic range and can be
used to detect non-covalent complexes with a Kd in the range
from 10 nM to 1 mM [9]. We were interested in using MS as
a first-pass FBS method, so we performed a pilot study on
hPNMT to ascertain whether MS could identify the same hits
as FBS-X. For proof-of-concept, an ESI-FTMS analysis was
conducted on a subset of 12 of the 384 fragments from the library.
ꢀ
c
ꢀ
c
The Authors Journal compilation 2010 Biochemical Society

Fragment-based screening to identify PNMT inhibitors
57
constants (Kd) for hPNMT, and thermodynamic parameters
of binding. All of the nine fragment hits detected by ITC
demonstrated favourable enthalpic contributions (Figure 3 and
Table 2). Of the 12 FBS-X hits, the measured binding affinities
for nine fragments ranged from approx. 5 to 700 μM. Of these,
two fragments (compounds 1 and 11) show low-affinity binding
(below 200 μM) that results in sub-optimal Wiseman c-values
(<1) for accurate affinity determination by ITC and this is
reflected in the relatively large errors [31]. Nonetheless the
data do allow prioritization of these compounds for follow-
up. Binding of the remaining three FBS-X hits (compounds 4,
8 and 12) could not be detected by ITC. The two additional
hits identified by ESI-FTMS were also not detected by ITC.
Furthermore, whereas the binding of 4-quinolinol (compound
1) was detected in ITC experiments, and a Kd of 0.7 mM
was calculated from the ITC measurements, the signal was
weak. The two tightest binding fragments with Kd values
of approx. 5 μM (2-aminobenzimidazole, compound 6, and
2-amino-1-methybenzimidazole, compound 7), are both based
on the benzimidazole core. As discussed above, benzimidazoles
have been shown previously to inhibit PNMT [32]. The third
fragment hit with a benzimidazole core (5-chlorobenzimidazole,
compound 9), had a 20-fold lower binding affinity (Kd of 97 μM),
suggesting that direct and water-mediated hydrogen bonds to the
2-amino groups of compounds 6 and 7 contribute considerably
to ligand affinity for hPNMT. The dissociation constants were
used to calculate their LE values (Figure 3 and Table 2). The LE
values ranged from 0.39 kcal/mol per HA for 4-quinolinol to
0.86 kcal/mol per HA for 4-bromo-1H-imidazole (compound
5). The two tightest binding benzimidazole-based fragments,
compounds 6 and 7, had high LEs of ∼0.7 kcal/mol per HA.
Elaborated benzimidazole compounds
Three of the identified hits (compounds 6, 7 and 9) contain a
benzimidazole ring system. The electron density for these three
hits was excellent, and ITC measurements confirmed that all
three bind to hPNMT, two of them with very high LE values.
Benzimidazoles have been investigated previously as inhibitors
of PNMT, although the earlier work was performed on the
bovine enzyme [32]. To further characterize the interaction of
benzimidazole-based fragments with the human enzyme, a series
of compounds (compounds 13–17; Figure 5) closely related to
the FBS-X benzimidazole hits were synthesized. The 2-amino
group was retained in this series because in compounds 6 and
7 it forms direct and water-mediated hydrogen bonds to the
enzyme. The binding of compounds 13–17 to hPNMT was
evaluated by ITC and X-ray crystallography. The results of these
analyses are shown in Figure 5 and data collection and refinement
statistics for the structures are included as Supplementary
Table S2. ITC curves are shown in Supplementary Figure S4
at http://www.BiochemJ.org/bj/431/bj4310051add.htm, and the
ITC-derived thermodynamic parameters of the 12 fragment hits
and five elaborated compounds are given in Table 2.
Figure 4 Solvent accessible surface of the hPNMT active site showing the
overlaid structures of the 12 FBS-X hits
(A and B) Binding positions of the FBS-X hits in the hPNMT target site, shown in two different
orientations. The 12 fragment hits are shown superimposed in space-filling representation
(oxygen atoms in red, nitrogen atoms in blue, carbon atoms in white and chlorine atoms
in green). The positions of residues discussed in the text are indicated, although for clarity
the structures of these residues are not shown. (C) Water molecules are bound in five major
sub-sites represented by green, blue, purple, orange or red spheres. Fragment hits are shown
in transparent space filling representation. This panel is shown in the same orientation as (B).
The 12 fragments were intentionally chosen to include several
FBS-X hits. ESI-FTMS screening was conducted blind in that the
researcher collecting and analysing the data did not know which
of the 12 compounds were FBS-X hits. Of the 12 fragments tested,
FBS-X identified four fragments as hits (compounds 1, 6, 11 and
12; Figure 3), three of which were detected by ESI-FTMS on
first inspection of the data (compounds 1, 6 and 12). The fourth
FBS-X hit, compound 11, was detected upon re-evaluation of the
ESI-FTMS spectra, although it was identified as a weak complex.
In addition, ESI-FTMS identified two other hits (Supplementary
Figure S3); these two fragments were not identified by FBS-X.
Hydroxy-substituted 2-aminobenzimidazole compounds
(compounds 13 and 14)
Superposition of the benzimidazole-based fragments with the
fragment hit 4-quinolinol (compound 1), suggested that a
hydroxy group added to the 7- or 6- position of compound 6
would interact with Lys57 and potentially with Tyr40. To test
this, compounds 13 and 14 were synthesized and their binding
characterized (Figure 5 and Table 2). The crystal structures of
ITC evaluation of fragment hits
To confirm and characterize further the 14 fragment hits (12
identified by FBS-X and the two identified by ESI-FTMS),
and to prioritize the fragment hits for follow up chemistry
and optimization, ITC was used to determine their dissociation
ꢀ
c
ꢀ
c
The Authors Journal compilation 2010 Biochemical Society

58
N. Drinkwater and others
Table 2 Thermodynamic parameters determined by ITC
Errors are S.D., calculated from averaging results for the replicates.
Compound
Number of replicates used
Stoichiometry (N)
Wiseman c-value
ꢀG (kcal/mol)
ꢀH (kcal/mol)
−TꢀS (kcal/mol)
Kd (μM)
LE (kcal/mol per HA)
+
+
+
+
1
2
3
4
5
6
7
8
2
2
4
–
2
3
2
–
2
2
2
–
3
3
3
3
3
0.8
1.3
0.9
–
1.0
1.0
1.0
–
1.2
0.9
0.9
–
1.0
0.9
1.0
1.0
0.9
0.3
1.1
14.3
–
1.2
31.7
43.5
–
2.1
1.4
0.5
–
13.3
10.0
111
27
−4.34 0.28
−11.12 4.76
6.78 5.0
3.83 6.21
690 220
180 90
0.39
0.52
0.60
–
0.86
0.71
0.66
–
0.55
0.53
0.42
–
0.60
0.58
0.72
0.64
0.55
−
−
−
−
+
+
+
+
−5.20 0.47
−9.03 5.74
−
−
−
−
+
+
+
+
−6.63 0.09
−11.60 1.13
4.97 1.22
14.0 2.0
−
−
−
−
–
–
–
–
+
+
+
+
−5.15 0.09
−10.19 1.52
5.04 1.43
170 16
−
−
−
−
+
+
+
+
−7.10 0.01
−12.37 0.73
5.26 0.72
6.30 0.02
−
−
−
−
+
+
+
+
−7.30 0.14
−11.73 1.59
4.43 1.45
4.60 0.69
−
−
−
−
–
–
–
–
+
+
+
+
9
−5.47 0.06
−8.56 2.15
3.09 2.21
97 5.9
−
−
−
−
+
+
+
+
10
11
12
13
14
15
16
17
−5.25 0.17
−11.54 3.62
6.29 3.79
140 27
−
−
−
−
+
+
+
+
−4.66 0.10
−8.69 1.33
4.02 1.43
380 44
−
−
−
−
–
–
–
–
+
+
+
+
−6.58 0.05
−7.20 0.54
0.62 0.52
15 1.0
−
−
−
−
+
+
+
+
−6.40 0.09
−7.57 0.23
1.16 0.15
20 2.5
−
−
−
−
+
+
+
+
−7.96 0.01
−11.70 1.60
−3.73 1.52
1.8 0.3
−
−
−
−
+
+
+
+
−6.99 0.01
−7.35 0.65
−0.36 0.70
7.2 0.9
−
−
−
−
+
+
+
+
14.3
−6.62 0.04
−10.13 1.25
3.51 1.27
14 0.8
−
−
−
−
hPNMT–AdoHcy–compound 13 and hPNMT–AdoHcy–
compound 14 showed that both these elaborated compounds
bound in a flipped binding mode compared with the predicted
binding mode, i.e. instead of forming an interaction with Lys57,
the hydroxy group is oriented away from Lys57. In hPNMT–
AdoHcy–compound 13 the hydroxy group makes no favourable
interactions, and sits in a hydrophobic pocket comprising the side
chains of Met258 and Val269. This same position is occupied by
the chloro substituent of compound 9 when it forms a complex
with hPNMT (Figure 3). In addition, the benzimidazole core
of compound 13 is shifted away from the hydrophobic pocket
compared with the original hit fragment, compound 6. This
movement results in an unfavourable interaction between the
six-membered aromatic ring and Lys57. These changes result
in a reduced enthalpic binding contribution of compound 13
compared with compound 6; this is balanced by a smaller entropy
loss upon binding with overall a small net decrease in the free
energy of binding. This is reflected in a decreased binding affinity
and LE of the elaborated compound compared with the original
hit (compound 13, Kd of 15 μM and LE of 0.60 kcal/mol per
HA; compound 6, Kd of 6.3 μM and LE of 0.71 kcal/mol
per HA), indicating that the hydroxy modification to compound
6 is not favourable for binding.
substituent of compound 6 and the smaller fluoro substituent at
the 5-position (Figure 5). Both compounds 15 and 16 bound as
predicted and the 2-chloro derivative compound 15 demonstrated
the most favourable binding affinity (Kd of 1.8 μM and LE
0.72 kcal/mol per HA) of this series. The binding affinity of
the fluoro-substituted compound 16 (Kd of 7.2 μM), was similar
to the parent compound 6, although the enthalpic and entropic
contributions were slightly altered, with a less favourable enthalpy
but more favourable entropy.
Hydroxy- and chloro-di-substituted 2-aminobenzimidazole
(compound 17)
Di-substitution of the 2-aminobenzimidazole core with the
5-chloro and 7-hydroxy substituents in compound 17, did
not significantly improve affinity (Kd of 14 μM) compared
with the hydroxy substitution alone (compound 13, Kd of
15 μM). As a result, the LE was reduced compared with the
original, unsubstituted fragment compound 6, and the separately
substituted compounds 13 and 15. Despite extensive trials we
were unable to measure X-ray diffraction data for hPNMT–
AdoHcy–compound 17. Crystals soaked in this compound
cracked during cryoprotection or gave very weak diffraction,
indicating that binding of the compound induces crystal disorder
or conformational changes to accommodate compound 17. As
no improvement in binding affinity or enthalpic contribution was
observed, we did not investigate compound 17 further.
Similarly, the hydroxy group of compound 14 is not favourable
for binding to hPNMT. In the crystal structure of the hPNMT
complex, the hydroxy group is unfavourably close to Met258
and waters are displaced from the hydrogen-bonding network.
These changes result in a reduced enthalpic contribution, a
reduced binding affinity and a reduced LE compared with the
original hit, compound 6 (compound 14, Kd of 20 μM and LE of
0.58 kcal/mol/HA).
DISCUSSION
Using FBS-X we found that 12 compounds from a library of 374
compounds bound to the noradrenaline-binding site of hPNMT
by displacing phosphate from the crystal structure. Our hit rate
of 3.2% is similar to that of other FBS-X campaigns [33–38],
despite the fact that our target crystal structure was not in the apo
form. We also investigated the use of high-sensitivity ESI-FTMS,
which has been used previously to screen natural product and
combinatorial chemistry libraries against other enzymes [39,40].
For our proof-of-concept study, we used a subset of the fragment
library. ESI-FTMS identified the same four hits found by FBS-X
for this subset. FBS-MS data acquisition took only 4 min, so an
Chloro- and fluoro-substituted 2-aminobenzimidazole compounds
(compounds 15 and 16)
Superposition of the fragment hit compounds 6 and 9 bound to
hPNMT indicated that a hydrophobic group at the 5-position
of the benzimidazole system of compound 6 might contribute
additional favourable interactions. To investigate this further, two
more compounds were designed: compound 15, combining the
2-amino substituent of compound 6 and the 5-chloro substituent
of compound 9; and compound 16 combining the 2-amino
ꢀ
c
ꢀ
c
The Authors Journal compilation 2010 Biochemical Society

Fragment-based screening to identify PNMT inhibitors
59
X-ray crystallography. This could in theory reduce the number
of crystal structures to be measured from ∼150 to ∼20, thereby
saving time, human resources and protein, and making the
technique accessible to many more academic laboratories.
Using ITC we determined the dissociation constants of nine of
the 12 FBS-X fragment hits. The results showed that the fragments
bind with affinities in the high to low micromolar range and all
nine hits for which affinity data were measured had LE values
ꢀ 0.4 kcal/mol per HA. Generally, LE values above 0.4 kcal/mol
per HA are desirable through the process of optimization [38,41].
Many of our hits had considerably higher LE values, emphasizing
their high quality.
Several of the FBS-X hits are benzimidazoles, a known
class of PNMT inhibitor. In 1970, researchers at the MSD
(Merck Sharp and Dohme Research Laboratories) showed that
benzimidazoles (including 2-aminobenzimidazole, compound 6,
and 5-chlorobenzimidazole, compound 9) inhibit bovine PNMT.
Thus at 28 μg/ml compound 6 inhibits binding of 0.57 mM DL-
normetanephrine to bovine PNMT by 30% and compound 9
inhibits binding by 57% [32]. Our FBS-X results identified that
these compounds also interact with human PNMT and show for
the first time how they bind in the noradrenaline pocket. The MSD
study also showed that unsubstituted benzimidazole is an inhibitor
of bovine PNMT (28% inhibition under the conditions described
above) [32]. Benzimidazole is one of the 376 compounds in the
library used in the present study, but was not identified as a hit
by FBS-X. This finding could be explained by one or more of the
following: (i) benzimidazole does not inhibit PNMT by binding
the noradrenaline-binding site; (ii) the noradrenaline-binding site
of the human and bovine enzymes differ sufficiently that they
recognize benzimidazole differently; or (iii) the screening method
we employed is less sensitive than the assay used to detect
inhibition in the bovine enzyme.
Several PNMT inhibitors have been shown to be competitive for
noradrenaline binding [42–44], although it has been shown that
a dichloro-substituted benzimidazole PNMT inhibitor was non-
competitive with noradrenaline and uncompetitive with AdoMet
[45]. These early results indicate the possibility of an additional
binding site for benzimidazoles. Our data clearly show that the
three benzimidazole hits and the elaborated compounds bind at
the noradrenaline site. However, the hPNMT crystals used for our
studies have AdoHcy (as a cofactor product) bound at the AdoMet
site. Furthermore, ITC analysis was performed in the presence of
excess AdoHcy. Consequently, if the benzimidazole compounds
bind to the hPNMT AdoMet site this would not be identified under
the conditions of our experiments.
The improved binding affinity of the designed benzimidazole
compound 15 compared with the original compound 6, is
largely due to an increased entropic contribution to binding
(Table 2). Enthalpic optimization is more desirable than entropic
opimization in progressing from hit to lead [46]. Typically,
enthalpically optimized leads more closely mimic the thermo-
dynamic characteristics of natural protein–ligand interactions, are
more soluble and often represent best-in-class compounds in drug
discovery programmes [46,47]. Our results therefore indicate that
although compound 15 shows higher affinity than compound 6 this
improvement may not necessarily correlate with an improved drug
lead.
In summary, we have shown that FBS-X can be implemented
successfully within an academic environment on a non-ideal
protein target. Using FBS-X we identified 12 hits that bind
hPNMT at the noradrenaline site. Our pilot study suggests the
screening process could be streamlined further by performing
FBS-MS first. Three of the FBS-X hits are from a chemical
class of known PNMT inhibitors that provide an unanticipated
Figure 5 Enzyme–ligand structures and binding information of elaborated
benzimidazoles
Chemical structures and binding poses for the elaborated benzimidazole compounds are given
(the binding pose for compound 17 is not provided because crystallographic data could not be
measured for that complex with hPNMT). Hydrogen bonding interactions are shown in black,
hydrophobic interactions in orange and unfavourable interactions in cyan broken lines. Binding
+
affinities determined by ITC (Kd in μM S.D.) and calculated LE values (in kcal/mol per HA)
are also indicated.
−
automated approach could process a 384 compound library
in just over 24 h using 0.4 mg of protein. By comparison, our
successful FBS-X campaign took 2–3 months and used ∼20 mg
of protein. These results suggest that ESI-FTMS could be used as
a filter to generate a short-list for subsequent characterization by
ꢀ
c
ꢀ
c
The Authors Journal compilation 2010 Biochemical Society

60
N. Drinkwater and others
14 Gearhart, D. A., Neafsey, E. J. and Collins, M. A. (2002) Phenylethanolamine
N-methyltransferase has β-carboline 2N-methyltransferase activity: hypothetical
relevance to Parkinson’s disease. Neurochem. Int. 40, 611–620
15 Kennedy, B. P., Bottiglieri, T., Arning, E., Ziegler, M. G., Hansen, L. A. and Masliah, E.
(2004) Elevated S-adenosylhomocysteine in Alzheimer brain: influence on
methyltransferases and cognitive function. J. Neural Transm. 111, 547–567
16 Martin, J. L., Begun, J., McLeish, M. J., Caine, J. M. and Grunewald, G. L. (2001) Getting
the adrenaline going: crystal structure of the adrenaline-synthesizing enzyme PNMT.
Structure 9, 977–985
17 Drinkwater, N., Gee, C. L., Puri, M., Criscione, K. R., McLeish, M. J., Grunewald, G. L.
and Martin, J. L. (2009) Molecular recognition of physiological substrate noradrenaline
by the adrenaline-synthesizing enzyme PNMT and factors influencing its
methyltransferase activity. Biochem. J. 422, 463–471
18 Gee, C. L., Tyndall, J. D., Grunewald, G. L., Wu, Q., McLeish, M. J. and Martin, J. L.
(2005) Mode of binding of methyl acceptor substrates to the adrenaline-synthesizing
enzyme phenylethanolamine N-methyltransferase: implications for catalysis.
Biochemistry 44, 16875–16885
19 McPhillips, T. M., McPhillips, S. E., Chiu, H. J., Cohen, A. E., Deacon, A. M., Ellis, P. J.,
Garman, E., Gonzalez, A., Sauter, N. K., Phizackerley, R. P. et al. (2002) Blu-Ice and the
distributed control system: software for data acquisition and instrument control at
macromolecular crystallography beamlines. J. Synchrotron Radiat. 9, 401–406
20 Otwinowski, Z. and Minor, W. (1997) Processing of X-ray diffraction data collected in
oscillation mode. Methods Enzymol. 276, p.307–326
21 Emsley, P. and Cowtan, K. (2004) COOT: model-building tools for molecular graphics.
Acta Crystallogr. Sect. D Biol. Crystallogr. 60, 2126–2132
22 Schuettelkopf, A. W. and van Aalten, D. M. F. (2004) PRODRG – a tool for high-
throughput crystallography of protein–ligand complexes. Acta Crystallogr. Sect. D Biol.
Crystallogr. 60, 1355–1363
internal control for the screen. Overall, we structurally and
thermodynamically characterized the binding of 17 small mol-
ecules to hPNMT, including the first such characterization of the
benzimidazole class of compounds bound to the noradrenaline-
binding site.
AUTHOR CONTRIBUTION
Nyssa Drinkwater and Jennifer Martin conceived the idea. Nyssa Drinkwater performed the
majority of the research and Jennifer Martin co-ordinated the study. Hoan Vu and Sally-
Ann Poulsen designed and performed the MS experiments. Kimberly Lovell, Thomas
Prisinzano, Kevin Criscione and Gary Grunewald designed and supervized compound
synthesis and contributed to compound characterization. Michael McLeish provided the
plasmid for the enzyme and contributed to data interpretation. Brett Collins assisted in
measurement and analysis of ITC data. Nyssa Drinkwater prepared the first draft of the
paper. All authors contributed to discussions on the study and provided intellectual input
to the manuscript.
ACKNOWLEDGEMENTS
We are grateful for access to the UQ ROCX Diffraction Facility and thank Karl Byriel and
Gordon King for their assistance. We thank staff at the MX1 beamline of the Australian
Synchrotron for their invaluable advice. The MS data could not have been measured
without access to the Eskitis Institute FTMS facility. We thank Dr F. Anthony Romero for
helpful preliminary discussions and Duncan McRee for advice and information regarding
the ActiveSight library.
23 Adams, P. D., Afonine, P. V., Bunkoczi, G., Chen, V. B., Davis, I. W., Echols, N., Headd,
J. J., Hung, L. W., Kapral, G. J., Grosse-Kunstleve, R. W. et al. (2010) PHENIX: a
comprehensive Python-based system for macromolecular structure solution. Acta
Crystallogr. Sect. D Biol. Crystallogr. 66, 213–221
FUNDING
24 Grunewald, G. L., Romero, F. A., Chieu, A. D., Fincham, K. J., Bhat, S. R. and Criscione,
K. R. (2005) Exploring the active site of phenylethanolamine N-methyltransferase:
3-alkyl-7-substituted-1,2,3,4-tetrahydroisoquinoline inhibitors. Bioorg. Med. Chem. 13,
1261–1273
25 Morningstar, M. L., Roth, T., Farnsworth, D. W., Smith, M. K., Watson, K., Buckheit, R. W.,
Das, K., Zhang, W. Y., Arnold, E., Julias, J. G. et al. (2007) Synthesis, biological activity,
and crystal structure of potent nonnucleoside inhibitors of HIV-1 reverse transcriptase
that retain activity against mutant forms of the enzyme. J. Med. Chem. 50, 4003–4015
26 Galan, A. A., Chen, J., Du, H., Forsyth, T., Huynh, T. P., Johnson, H. W. B., Kearney, P.,
Leahy, J. W., Lee, M. S., Mann, G. et al. (2008) Preparation of 1H-imidazole-4,5-
dicarboxamides as JAK-2 modulators. European Patent Foundation no.
WO2008042282-A2
27 Parmee, E. R., Kim, R. M., Rouse, E. A., Schmidt, D. R., Sinz, C. J. and Chang, J. (2005)
Preparation of cyclic guanidines as glucagon receptor antagonists for the treatment of
type 2 diabetes. PCT Int. Appl., no. WO 2005065680-A1 20050721
28 Gee, C. L., Drinkwater, N., Tyndall, J. D. A., Grunewald, G. L., Wu, Q., McLeish, M. J. and
Martin, J. L. (2007) Enzyme adaptation to inhibitor binding: a cryptic binding site in
phenylethanolamine N-methyltransferase. J. Med. Chem. 50, 4845–4853
29 McMillan, F. M., Archbold, J., McLeish, M. J., Caine, J. M., Criscione, K. R., Grunewald,
G. L. and Martin, J. L. (2004) Molecular recognition of sub-micromolar inhibitors by the
epinephrine-synthesizing enzyme phenylethanolamine N-methyltransferase. J. Med.
Chem. 47, 37–44
This work was supported by an University of Queensland Graduate School Research
Travel Grant and an Australian Postgraduate Award (to N.D.); an Australian Research
Council Discovery Project award [grant number DP0664564 (to J.L.M.)]; an Australian
National Health and Medical Research Council Biomedical RD Wright Career Development
Award (to B.M.C.) and Senior Research Fellowship [grant number 455829 (to J.L.M.)]; an
Australian Research Council Linkage Equipment and Infrastructure Award [grant numbers
LE0668382 (to J.L.M.), LE237908 (to S.-A.P.)]; and by the National Institutes of Health
[grant number NIH HL034193 (to G.L.G.)].
REFERENCES
1
2
3
4
5
6
7
Hajduk, P. J. and Greer, J. (2007) A decade of fragment-based drug design: strategic
advances and lessons learned. Nat. Rev. Drug Discov. 6, 211–219
Murray, C. W. and Rees, D. C. (2009) The rise of fragment-based drug discovery. Nat.
Chem. 1, 187–192
Congreve, M., Carr, R., Murray, C. and Jhoti, H. (2003) A rule of three for fragment-based
lead discovery? Drug Discov. Today 8, 876–877
Rees, D. C., Congreve, M., Murray, C. W. and Carr, R. (2004) Fragment-based lead
discovery. Nat. Rev. Drug Discov. 3, 660–672
Erlanson, D. A. (2006) Fragment-based lead discovery: a chemical update. Curr. Opin.
Biotech. 17, 643–652
Neumann, T., Junker, H. D., Schmidt, K. and Sekul, R. (2007) SPR-based fragment
screening: advantages and applications. Curr. Top. Med. Chem. 7, 1630–1642
Ciulli, A., Williams, G., Smith, A. G., Blundell, T. L. and Abell, C. (2006) Probing hot
spots at protein-ligand-binding sites: a fragment-based approach using biophysical
methods. J. Med. Chem. 49, 4992–5000
30 Wu, Q., Gee, C. L., Lin, F., Tyndall, J. D., Martin, J. L., Grunewald, G. L. and McLeish,
M. J. (2005) Structural, mutagenic, and kinetic analysis of the binding of substrates and
inhibitors of human phenylethanolamine N-methyltransferase. J. Med. Chem. 48,
7243–7252
31 Turnbull, W. B. and Daranas, A. H. (2003) On the value of c: can low affinity systems be
studied by isothermal titration calorimetry? J. Am. Chem. Soc. 125, 14859–14866
32 Mandel, L. R., Porter, C. C., Kuehl, F. A., Jensen, N. P., Schmitt, S. M., Windholz, T. B.,
Beattie, T. R., Carty, J. A., Christen, B. G. and Shen, T. Y. (1970) Inhibition of adrenal
phenethanolamine N-methyltransferase by substituted benzimidazoles. J. Med. Chem.
13, 1043–1047
8
9
Orita, M., Warizaya, M., Amano, Y., Ohno, K. and Niimi, T. (2009) Advances in
fragment-based drug discovery platforms. Exp. Opin. Drug Discov. 4, 1125–1144
Hofstadler, S. A. and Sannes-Lowery, K. A. (2006) Applications of ESI-MS in drug
discovery: interrogation of noncovalent complexes. Nat. Rev. Drug Discov. 5, 585–595
10 Schulz, M. N. and Hubbard, R. E. (2009) Recent progress in fragment-based lead
discovery. Curr. Opin. Pharmacol. 9, 615–621
33 Antonysamy, S. S., Aubol, B., Blaney, J., Browner, M. F., Giannetti, A. M., Harris, S. F.,
Hebert, N., Hendle, J., Hopkins, S., Jefferson, E. et al. (2008) Fragment-based discovery
of hepatitis C virus NS5b RNA polymerase inhibitors. Bioorg. Med. Chem. Lett. 18,
2990–2995
34 Bosch, J., Robien, M. A., Mehlin, C., Boni, E., Riechers, A., Buckner, F. S., Van Voorhis,
W. C., Myler, P. J., Worthey, E. A., DeTitta, G. et al. (2006) Using fragment cocktail
crystallography to assist inhibitor design of Trypanosoma brucei nucleoside
2-deoxyribosyltransferase. J. Med. Chem. 49, 5939–5946
11 Fuller, R. W. (1982) Pharmacology of brain epinephrine neurons. Ann. Rev. Pharmacol.
Toxicol. 22, 31–55
12 Crowley, W. R., Terry, L. C. and Johnson, M. D. (1982) Evidence for the involvement of
central epinephrine systems in the regulation of luteinizing-hormone, prolactin, and
growth-hormone release in female rats. Endocrinology 110, 1102–1107
13 Mefford, I. N., Lister, R. G., Ota, M. and Linnoila, M. (1990) Antagonism of ethanol
intoxication in rats by inhibitors of phenylethanolamine N-methyltransferase. Alcohol
Clin. Exper. Res. 14, 53–57
ꢀ
c
ꢀ
c
The Authors Journal compilation 2010 Biochemical Society

Fragment-based screening to identify PNMT inhibitors
61
35 Lesuisse, D., Lange, G., Deprez, P., Benard, D., Schoot, B., Delettre, G., Marquette, J. P.,
Broto, P., Jean-Baptiste, V., Bichet, P. et al. (2002) SAR and X-ray. A new approach
combining fragment-based screening and rational drug design: application to the
discovery of nanomolar inhibitors of Src SH2. J. Med. Chem. 45, 2379–2387
36 Nienaber, V. L., Richardson, P. L., Klighofer, V., Bouska, J. J., Giranda, V. L. and Greer, J.
(2000) Discovering novel ligands for macromolecules using X-ray crystallographic
screening. Nat. Biotech. 18, 1105–1108
37 Schuffenhauer, A., Ruedisser, S., Marzinzik, A. L., Jahnke, W., Blommers, M., Selzer, P.
and Jacoby, E. (2005) Library design for fragment based screening. Curr. Top. Med.
Chem. 5, 751–762
38 Wyatt, P. G., Woodhead, A. J., Berdini, V., Boulstridge, J. A., Carr, M. G., Cross, D. M.,
Davis, D. J., Devine, L. A., Early, T. R., Feltell, R. E. et al. (2008) Identification of
N-(4-piperidinyl)-4-(2,6-dichlorobenzoylamino)-1H-pyrazole-3-carboxamide (AT7519),
a novel cyclin dependent kinase inhibitor using fragment-based X-ray crystallography
and structure based drug design. J. Med. Chem. 51, 4986–4999
41 Bembenek, S. D., Tounge, B. A. and Reynolds, C. H. (2009) Ligand efficiency and
fragment-based drug discovery. Drug Discov. Today 14, 278–283
42 Fuller, R. W., Hemrick-Luecke, S. K. and Perry, K. W. (1983) Inhibition of brain
epinephrine synthesis by 3,4-dichlorophenylethanolamine, a competitive
substrate for norepinephrine N-methyltransferase. Biochem. Pharmacol. 32,
215–220
43 Fuller, R. W., Roush, B. W., Snoddy, H. D., Day, W. A. and Molloy, B. B. (1975)
Norepinephrine N-methyltransferase inhibition by benzamidines, phenylacetamidines,
benzylguanidines, and phenylethylguanidines. J. Med. Chem. 18, 304–307
44 Fuller, R. W., Roush, B. W., Snoddy, H. D. and Molloy, B. B. (1973) Inhibition of
phenylethanolamine N-methyltransferase by benzylamines. 2. In-vivo studies with
2,3-dichloro-α-methylbenzylamine. J. Med. Chem. 16, 106–109
45 Pendleton, R. G., Green, H., Snow, I. B., Wang, J. and Kaiser, C. (1972) Studies on the
mechanism of phenylethanolamine-N-methyl-transferase inhibition by a
dichloro-substituted benzimidazole. Biochem. Pharmacol. 21, 2967–2975
39 Poulsen, S. A. (2006) Direct screening of a dynamic combinatorial library using mass
spectrometry. J. Am. Soc. Mass Spec. 17, 1074–1080
46 Ladbury, J. E., Klebe, G. and Freire, E. (2010) Adding calorimetric data to decision making
in lead discovery: a hot tip. Nat. Rev. Drug Discov. 9, 23–27
40 Vu, H., Pharm, N. B. and Quinn, R. J. (2008) Direct screening of natural product extracts
using mass spectrometry. J. Biomol. Screen. 13, 265–275
47 Freire, E. (2008) Do enthalpy and entropy distinguish first in class from best in class?
Drug Discov. Today 13, 869–874
Received 27 April 2010/29 June 2010; accepted 20 July 2010
Published as BJ Immediate Publication 20 July 2010, doi:10.1042/BJ20100651
ꢀ
c
ꢀ
c
The Authors Journal compilation 2010 Biochemical Society

Biochem. J. (2010) 431, 51–61 (Printed in Great Britain) doi:10.1042/BJ20100651
SUPPLEMENTARY ONLINE DATA
Fragment-based screening by X-ray crystallography, MS and isothermal
titration calorimetry to identify PNMT (phenylethanolamine
N-methyltransferase) inhibitors
Nyssa DRINKWATER*, Hoan VU†, Kimberly M. LOVELL‡, Kevin R. CRISCIONE‡, Brett M. COLLINS*, Thomas E. PRISINZANO‡,
Sally-Ann POULSEN†, Michael J. MCLEISH§, Gary L. GRUNEWALD‡ and Jennifer L. MARTIN*1
*University of Queensland, Institute for Molecular Bioscience, Division of Chemistry and Structural Biology, Brisbane, Queensland 4072, Australia, †Griffith University, Eskitis Institute,
Brisbane, Queensland 4111, Australia, ‡Department of Medicinal Chemistry, University of Kansas, Lawrence, KS 66045-7582, U.S.A., and §Department of Chemistry and Chemical
Biology, Indiana University–Purdue University Indianapolis (IUPUI), 402 N. Blackford St, LD 326D, IN 46202, U.S.A.
1,2-diamine (0.50 g, 3.5 mmol). After stirring overnight, the
cyanogen bromide reaction mixture was evaporated under reduced
pressure to give compound 15 · HBr as a brown solid, which was
recrystallized from 2-propanol/hexanes (0.39 g, 46%), m.p. 241–
1
243◦C: H-NMR (500 MHz, MeOD) δ 7.39 (d, J = 1.9, 1H),
7.34 (d, J = 8.5, 1H), 7.27 (dd, J = 1.9, 8.5, 1H); 13C-NMR (126
MHz, MeOD) δ 153.0, 132.1, 130.3, 130.0, 125.1, 113.6, 112.7;
HRMS (m/z): [M+H] calculated for C7H7N3Cl, 168.0329; found
168.0332.
A sample of (compound 15 · HBr) was dissolved in aqueous
NaOH (1.25 M, 20 ml) and extracted with diethyl ether (3×
15 ml). The combined organic layers were washed with brine
(10 ml) and dried over Na2SO4. The solvent was evaporated under
reduced pressure and the resulting residue was triturated with
hexanes/ethyl acetate (3:1) to afford compound 15 as a yellow
solid: m.p. 165–170◦C (compared with the reported m.p. of 167–
168◦C [3]).
Figure S1 Chemical structures of fragments screened by MS
EXPERIMENTAL
2-Amino-5(6)-fluorobenzimidazole hydrobromide (compound 16 · HBr)
Chemical synthesis details for compounds 14, 15 and 16
Benzimidazole compound 16 · HBr was synthesized according to
the procedure described previously [4] in two steps from 4-fluoro-
2-nitroaniline (1.1 g, 7.0 mmol) with the following modifications:
(i) the reduction step was carried out at atmospheric pressure;
2-Amino-5(6)-hydroxybenzimidazole hydrobromide (compound 14 · HBr)
Benzimidazole compound 14 · HBr was synthesized from 4-
methoxybenzene-1,2-diamine in two steps [1,2] (93% overall
yield), m.p. 207–212◦C: 1H-NMR (500 MHz, MeOD) δ 7.16 (t,
J = 7.3, 1H), 6.78 (d, J = 2.3, 1H), 6.72 (dd, = 2.3, 8.6, 1H); 13C-
NMR (126 MHz, MeOD) δ 156.1, 152.1, 132.0, 124.0, 113.0,
112.7, 99.3; HRMS (m/z): [M+H] calculated for C7H7N3OBr,
227.9772; found 227.9783.
and (ii) after stirring overnight, the cyanogen bromide reaction
mixture was evaporated under reduced pressure to give a brown
oil. Trituration with 2-propanol/diethyl ether yielded compound
16 · HBr as a viscous brown liquid that was dried in vacuo to a
brown solid (0.11 g, 6.6% overall yield), m.p. 212–220◦C; H-
1
NMR (500 MHz, MeOD) δ 7.36 (dd, J = 4.3, 8.8, 1H), 7.16
(dd, J = 2.4, 8.4, 1H), 7.04 (ddd, J = 2.5, 8.8, 9.7, 1H); 13C-
NMR (126 MHz, MeOD) δ 161.2 (d, J = 239.4), 153.2 (s), 131.8
(d, J = 12.6), 127.5 (s), 113.4 (d, J = 10.1), 112.0 (d, J = 25.2),
100.3 (d, J = 30.2); HRMS (m/z): [M+H] calculated for C7H7N3F,
152.0624; found 152.0627.
2-Amino-5(6)-chlorobenzimidazole hydrobromide (compound 15 · HBr)
Benzimidazole (compound 15 · HBr) was synthesized according
to the procedure described previously [3] from 4-chlorobenzene-
1
To whom correspondence should be addressed (email j.martin@imb.uq.edu.au).
The co-ordinates and structure factors for all 17 crystal structures have been deposited with the PDB under codes 3KPJ, 3KPU, 3KPV, 3KPW, 3KPY,
3KQM, 3KQS, 3KQT, 3KQV, 3KQW, 3KQO, 3KQP, 3KQQ, 3KQY, 3KR0, 3KR1 and 3KR2.
ꢀ
c
ꢀ
c
The Authors Journal compilation 2010 Biochemical Society

N. Drinkwater and others
Figure S3 Electron density (2Fo−Fc shown at 1σ) for ligands bound to
hPNMT
(A–P) represent density for compounds 1–17.
Figure S2 ESI-FTMS screening
(A) ESI-FTMS positive ion mass spectrum of hPNMT (3 μM) in 10 mM ammonium acetate
solution. Peaks corresponding to hPNMT in complex with AdoHcy are also observed indicating
a small amount of AdoHcy co-purifies with the enzyme. (B) ESI-FTMS positive ion mass
spectrum of a mixture of hPNMT (3 μM) and AdoHcy (30 μM) in 10 mM ammonium acetate
and 5% methanol. (C) ESI-FTMS positive ion mass spectrum of a mixture of hPNMT (3 μM),
AdoHcy (30 μM) and compound 6 (300 μM) in 10 mM ammonium acetate and 5% methanol.
(D) Structures of the two additional fragments detected by ESI-FTMS, but not detected by FBS-X
or ITC.
ꢀ
c
ꢀ
c
The Authors Journal compilation 2010 Biochemical Society

Fragment-based screening to identify PNMT inhibitors
Figure S4 ITC curves for fragment hits and derivatized benzimidazoles binding to hPNMT
Table S1 Mass/charge (m/z) values for hPNMT and hPNMT bound ligands (AdoHcy and compound 6) identified from screening of compound 6 by MS
The molecular mass of the fragment of compound 6 is calculated according to: Mr fragment = Dm/z × z = [m/z (hPNMT–AdoHcy–fragment)−m/z (hPNMT–AdoHcy)]×z
Charge state (z)
m/z (hPNMT)
m/z (hPNMT–AdoHcy)
m/z (hPNMT–AdoHcy–fragment)
Mr fragment
13+
12+
11+
10+
2437.6
2641.0
2880.6
3168.2
2467.1
2673.1
2915.6
3206.9
–
–
2684.0
2927.6
3220.1
130.8
132.0
132.0
ꢀ
c
ꢀ
c
The Authors Journal compilation 2010 Biochemical Society

Table S2 Crystallographic data and refinement statistics for structures of fragment hits in complex with hPNMT–AdoHcy
The data for fragment hit compounds 1–5 were collected in house using the UQ ROCX Diffraction Facility, whereas the data for fragment hit compounds 6–12 were collected at the Australian Synchrotron (3BM1). The completeness is number of measured unique
reflections divided by the number of theoretical reflections, expressed as a percentage. Rmerge = ꢀ|Iobs−Iav|/ꢀIav, over all symmetry-related observations. Rcryst = ꢀ|Fobs−Fcalc|/ꢀ|Fobs|, over all reflections. Rfree is calculated as for Rcryst from 5–10% of the data
excluded from refinement. Values in parentheses are for the highest resolution shell of data.
Fragment hit
Parameter
–
1
2
3
4
5
6
7
8
9
10
11
12
Space group
Unit cell
P43212
P43212
P43212
P43212
P43212
P43212
P43212
P43212
P43212
P43212
P43212
P43212
P43212
˚
a,b (A)
94.4
186.9
90
94.4
188.5
90
94.2
188.8
90
94.3
188.6
90
93.9
188.6
90
93.8
188.7
90
94.7
189.2
90
93.9
188.5
90
94.4
189.4
90
94.4
189.5
90
94.3
189.0
90
93.6
188.8
90
94.0
188.9
90
˚
c (A)
◦
α,β,γ ( )
No. observations
No. unique reflections
Resolution (A)
112323
29790
150000
31980
196450
28540
179390
33323
149936
30511
228299
30146
510457
58278
317982
33746
325385
38736
211596
28974
171507
33145
151324
30322
147430
26826
˚
33.38–2.50
(2.59– 2.50)
3.8 (3.8)
11.7 (2.2)
99.1 (99.9)
5.0 (52.5)
35.03–2.40
(2.49–2.40)
4.7 (3.2)
5.3 (2.2)
93.8 (88.3)
15.1 (43.6)
45.69–2.40
(2.49–2.40)
6.9 (7.1)
10.5 (6.4)
83.8 (54.3)
10.0 (24.6)
37.73–2.40
(2.49–2.40)
5.4 (2.6)
12.1 (1.6)
97.7 (90.1)
7.1 (43.9)
45.56–2.40
(2.49–2.40)
4.9 (3.9)
8.6 (3.8)
90.2 (75.0)
9.6 (27.6)
45.63–2.40
(2.49–2.40)
7.6 (6.1)
6.8 (3.0)
89.3 (68.3)
15.3 (46.2)
50.00–2.00
(2.01–2.00)
8.8 (6.3)
8.7 (1.6)
99.6 (98.5)
7.5 (5.6)
50.00–2.40
(2.49–2.40)
9.4 (8.3)
9.1 (2.7)
99.7 (99.8)
7.7 (54.6)
50.00–2.30
(2.38–2.30)
8.4 (7.8)
12.4 (5.1)
99.6 (99.7)
5.5 (28.1)
50.0–2.50
(2.59–2.50)
7.3 (3.5)
6.5 (0.9)
93.6 (62.5)
14.3 (58.5)
47.13–2.40
(2.49–2.40)
5.2 (3.2)
6.7 (2.1)
97.1 (89.3)
12.3 (43.3)
45.41– 2.40
(2.49–2.40)
5.0 (3.5)
6.2 (2.9)
90.3 (82.9)
13.8 (33.8)
38.40–2.50
(2.59–2.50)
5.5 (4.8)
7.3 (2.6)
89.2 (82.4)
11.7 (42.9)
Redundancy
I/σI
Completeness (%)
Rmerge (%)
Refinement
Rcryst/Rfree (%)
21.1/26.7
22.7/29.1
21.4/27.2
21.2/28.4
20.6/27.1
22.7/29.2
18.8/21.7
18.1/23.4
18.4/23.0
19.4/26.2
21.9/27.9
23.1/30.2
22.5/30.6

Fragment-based screening to identify PNMT inhibitors
Table S3 Crystallographic data and refinement statistics for structures of elaborated benzimidazole compounds 13–16 in complex with hPNMT–AdoHcy
The completeness is number of measured unique reflections divided by the number of theoretical reflections, expressed as a percentage. Rmerge = ꢀ|Iobs−Iav|/ꢀIav, over all symmetry-related
observations. Rcryst = ꢀ|Fobs−Fcalc|/ꢀ|Fobs|, over all reflections. Rfree is calculated as for Rcryst from 5–10% of the data excluded from refinement. Values in parentheses are for the highest
resolution shell of data.
Compound
Parameter
13
14
15
16
Space Group
Unit Cell
P43212
P43212
P43212
P43212
˚
a,b (A)
94.5
188.3
90
94.3
188.5
90
94.1
189.4
90
94.4
188.2
90
˚
c (A)
◦
α,β,γ ( )
No. observations
No. unique reflections
Resolution (A)
Redundancy
I/σI
Completeness (%)
Rmerge (%)
184367
44046
47.26–2.20 (2.28–2.20)
4.19 (4.15)
10.0 (2.1)
99.8 (100.0)
5.6 (50.3)
83731
26367
45.73–2.60 (2.69 - 2.60)
3.18 (3.16)
9.6 (2.3)
97.8 (99.3)
6.5 (37.6)
134165
38506
42.30–2.30 (2.38–2.30)
3.48 (3.50)
10.8 (2.3)
99.8 (99.8)
5.0 (42.9)
148978
38507
45.72–2.30 (2.38–2.30)
3.87 (3.89)
10.0 (2.3)
99.7 (100.0)
6.0 (47.7)
˚
Refinement
Rcryst/Rfree (%)
22.8/26.2
21.7/25.7
21.8/24.9
21.2/24.7
REFERENCES
1
Galan, A. A., Chen, J., Du, H., Forsyth, T., Huynh, T. P., Johnson, H. W. B., Kearney, P.,
Leahy, J. W., Lee, M. S., Mann, G. et al. (2008) Preparation of 1H-imidazole-4,5-
dicarboxamides as JAK-2 modulators. European Patent Foundation no.
WO2008042282-A2
2
Lattanzi, R., Spetea, M., Schullner, F., Rief, S. B., Krassnig, R., Negri, L. and
Schmidhammer, H. (2005) Synthesis and biological evaluation of 14-alkoxymorphinans.
22. Influence of the 14-alkoxy group and the substitution in position 5 in
14-alkoxymorphinan-6-ones on in vitro and in vivo activities. J. Med. Chem. 48,
3372–3378
3
4
Leonard, N. J., Curtin, D. Y. and Beck, K. M. (1947) Sulfonate salts of substituted
benzimidazoles. J. Am. Chem. Soc. 69, 2459–2461
Mandel, L. R., Porter, C. C., Kuehl, F. A., Jensen, N. P., Schmitt, S. M., Windholz, T. B.,
Beattie, T. R., Carty, J. A., Christen, B. G. and Shen, T. Y. (1970) Inhibition of adrenal
phenethanolamine N-methyltransferase by substituted benzimidazoles. J. Med. Chem. 13,
1043–1047
Received 27 April 2010/29 June 2010; accepted 20 July 2010
Published as BJ Immediate Publication 20 July 2010, doi:10.1042/BJ20100651
ꢀ
c
ꢀ
c
The Authors Journal compilation 2010 Biochemical Society