β-Triketone inhibitors of plant p-hydroxyphenylpyruvate dioxygenase: Modeling and comparative molecular field analysis of their interactions

Article abstract of DOI:10.1021/jf9005593

p-Hydroxyphenylpyruvate dioxygenase (HPPD) is the target site of β-triketone herbicides in current use. Nineteen β-triketones and analogues, including the naturally occurring leptospermone and grandiflorone, were synthesized and tested as inhibitors of pu

Full text of DOI:10.1021/jf9005593

5194 J. Agric. Food Chem. 2009, 57, 5194–5200
DOI:10.1021/jf9005593
β-Triketone Inhibitors of Plant p-Hydroxyphenylpyruvate
Dioxygenase: Modeling and Comparative Molecular Field
Analysis of Their Interactions
‡
F
RANCK E. DAYAN,*,† NIDHI
S
INGH,‡ CHRISTOPHER R. M
C
C
URDY
#
,
COLETTE A. GODFREY
,
§,# LESLEY
§,#
L
ARSEN
,
§,# REX T. WEAVERS
§,#
,
J
OHN W. VAN
K
LINK
,
AND
NIGEL B. PERRY
†Natural Products Utilization Research Unit, Agricultural Research Service, U.S. Department of
Agriculture, P.O. Box 8048, University, Mississippi 38677, ‡Department of Medicinal Chemistry,
University of Mississippi, University, Mississippi 38677, §New Zealand Institute for Plant and Food
Research Ltd. and #Department of Chemistry, University of Otago, P.O. Box 56, Dunedin, New Zealand
p-Hydroxyphenylpyruvate dioxygenase (HPPD) is the target site of β-triketone herbicides in current
use. Nineteen β-triketones and analogues, including the naturally occurring leptospermone and
grandiflorone, were synthesized and tested as inhibitors of purified Arabidopsis thaliana HPPD. The
most active compound was a β-triketone with a C9 alkyl side chain, not reported as natural, which
inhibited HPPD with an I50 of 19 ( 1 nM. This is significantly more active than sulcotrione, which had
an I50 of 250 ( 21 nM in this assay system. The most active naturally occurring β-triketone was
grandiflorone, which had an I50 of 750 ( 70 nM. This compound is of potential interest as a natural
herbicide because it can be extracted with good yield and purity from some Leptospermum shrubs.
Analogues without the 1,3-diketone group needed to interact with Fe2+ at the HPPD active site were
inactive (I50s > 50 μM), as were analogues with prenyl or ethyl groups on the triketone ring.
Modeling of the binding of the triketones to HPPD, three-dimensional QSAR analysis using CoMFA
(comparative molecular field analysis), and evaluation of the hydrophobic contribution with HINT
(hydropathic interactions) provided a structural basis to describe the ligand/receptor interactions.
KEYWORDS: Natural products; triketones; phytotoxins; herbicides; mode of action; p-hydroxyphe-
nylpyruvate dioxygenase; structure-activity relationships; essential oils
INTRODUCTION
tembotrione (5-11). This class of herbicides was inspired by
the natural β-triketone phytotoxin leptospermone (4, Figure 1)
produced by the bottlebrush plant (Callistemon spp.) (7, 12, 13).
β-Triketones are also found in other Australasian woody plants
in the family Myrtaceae, for example, Leptospermum, Eucalyptus,
and Corymbia (14-17). β-Triketones, and essential oils rich in
these compounds, have a variety of biological activities (18):
antifungal and antimicrobial (19-22), antiviral (21, 23), insecti-
cidal and molluscicidal (21), and herbicidal (13, 24, 25).
p-Hydroxyphenylpyruvate dioxygenase (HPPD; EC 1.13.11.27,
EC 1.14.2.2) is a nonheme, iron II containing, R-keto acid-
dependent enzyme that catalyzes the formation of homogentisic
acid (HGA). This mechanistically complex reaction involves
the oxidative decarboxylation of the 2-oxoacid side chain of
4-hydroxyphenylpyruvate (4-HPP), the subsequent hydroxylation
of the aromatic ring, and a 1,2 (ortho) rearrangement of the
carboxymethyl group (1-3).
The structure-activity relationship (SAR) of synthetic
β-triketones against HPPD has been extensively pursued, leading
to the design of several successful commercial herbicides
[e.g., sulcotrione (2-[2-chloro-4-methanesulfonylbenzoyl]cyclohex-
ane-1,3-dione) and mesotrione (2-[4-methylsulfonyl-2-nitroben-
zoyl]cyclohexane-1,3-dione)] (5-9). We recently reported
that natural β-triketones leptospermone 4 and grandiflorone 6
(Table 1), from the essential oil of manuka (Leptospermum
scoparium), inhibited HPPD (25 ). However, in contrast to their
synthetic counterparts, these compounds were competitive, rever-
sible inhibitors (25 ). A series of β-triketones has been synthesized
and tested for their antimicrobial activity (19 ), and we now report
the activity of 15 of these compounds, plus four new analogues,
against HPPD. Modeling of their interaction with the catalytic site
of HPPD and three-dimensional quantitative structure-activity
In plants, HGA is a key precursor in the biosynthesis of
tocochromanols (tocopherols and tocotrienols) and prenylqui-
nones. The prenylquinone plastoquinone is an essential cofactor
for phytoene desaturase (4 ). Inhibition of HPPD causes photo-
dynamic bleaching of the foliage because the reduction of
plastoquinone levels hampers phytoene desaturase activity (5 ).
Under high light intensity, the reduced pool of carotenoids within
the thylakoid membranes no longer quenches the excess energy,
destabilizing the photosynthetic apparatus, which leads to a rapid
degradation of chlorophylls.
HPPD is the molecular target site for β-triketone herbicides
in current use, for example, sulcotrione, mesotrione, and
*Author to whom correspondence should be addressed [telephone
(662) 915 1039; fax (662) 915 1035; e-mail fdayan@olemiss.edu].
pubs.acs.org/JAFC
Published on Web 05/12/2009
© 2009 American Chemical Society

Article
J. Agric. Food Chem., Vol. 57, No. 12, 2009 5195
Table 1. In Vitro Inhibition of p-Hydroxyphenylpyruvate Dioxygenase by
β-Triketones and Selected Calculated Properties
natural product I50a (nM (
name 1 SE)
Clog HINT log volumed
binding
energye
3
(A )
compd
pI50 Pb
Pc
˚
1
2
>50000 <4.3 1.13 0.25
44500 ( 4.35 1.97 1.34
4000
201.8
234.3
-42.1
-55.9
flavesone
3
4
5
18900 ( 4.72 2.87 1.81
1800
270.2
254.8
252.7
-34.1
-27.9
-51.8
leptospermone
11800 ( 4.93 2.58 1.88
1000
isoleptospermone 14300 ( 4.84 2.49 1.88
1200
6
grandiflorone
papuanone
750 ( 70 6.13 3.22 2.35
960 ( 100 6.02 3.24 2.42
>50000 <4.3 3.16 2.77
170 ( 20 6.77 4.31 3.73
410 ( 50 6.39 4.78 4.43
19 ( 1 7.72 5.36 4.58
250 ( 50 6.60 6.42 5.66
310 ( 10 6.51 8.53 7.82
2270 ( 280 5.64 5.81 5.21
>50000 <4.3 2.86 2.05
>50000 <4.3 4.79 3.89
>50000 <4.3 7.97 7.82
>50000 <4.3 8.53 7.26
>50000 <4.3 11.85 11.58
284.9
269.1
275.8
310.9
325.4
336.1
369.4
435.8
376.1
258.5
303.5
438.9
494.5
639.3
-35.0
-19.1
-60.4
-32.0
-85.1
-24.5
-36.6
-12.6
-26.1
-5.2
7
8
9
10
11
12
13
14
15
16
17
18
19
Figure 1. Structures of the β-triketones and related compounds tested in
this study.
relationship analysis with comparative molecular field analysis
(CoMFA) were used to explain the activity of these natural
herbicides and analogues.
-3.9
>106
>106
>106
MATERIALS AND METHODS
General Experimental Procedures. All solvents were distilled before
use and were removed by rotary evaporation at temperatures up to 35 °C.
Merck silica gel 60, 200-400 mesh, 40-63 μm, was used for silica gel flash
chromatography. TLC was carried out using Merck DC-plastikfolien
Kieselgel 60 F254, visualized with a UV lamp. High-resolution mass, UV,
and IR spectra were recorded on Kratos MS-80 (EI) or Bruker micro
Q-TOF (ESI), Shimadzu UV 240, and Perkin-Elmer 1600 FTIR instru-
ments, respectively. NMR spectra, at 25 °C, were recorded at 300 MHz for
1H and at 75 MHz for 13C on a Varian INOVA 300 spectrometer.
Chemical shifts are given in parts per million on the δ scale referenced
to the solvent peaks CHCl3 at 7.25 ppm and CDCl3 at 77.0 ppm.
Syntheses. All of the compounds were prepared as described
previously (19), except for 10 and 14-16, which were synthesized as
described below. Structural assignments were made by comparison with
previously synthesized compounds (19) and/or additional 2D NMR
experiments to confirm connectivities.
sulcotrione
250 ( 21 6.60 0.92 0.20
243.3
a I50 = concentration required for 50% inhibition HPPD ( standard error.
b Calculated with ChemBioDraw Ultra, version 11.1. c Calculated with Sybyl/HINT,
see ref (33). d Total volume calculated using fast Connolly algorithm. Volume of
cyclic triketone moieties: 1-9 and 11-13, 184.7; 10, 199.2; 14, 244.7; 18, 254.2; 17
3
and 19, 438.3 A . e Energy of ligand/receptor interaction in kcal/mol calculated by
˚
Sybyl/Flexidock.
1.71 (s, H-500), 1.78 (s, H-400), 3.04 (t, J = 7 Hz, H-20), 3.31 (d, J = 7 Hz, H-100),
5.21 (t, J = 7 Hz, H-200), 6.15 (s, H-3), 9.34 (s, 4-OH), 12.19 (s, 6-OH),
14.29 (s, 2-OH). To this monoprenylated 2-cyclohexylethyl phloroglucinol
20 (200 mg, 0.6 mmol) in NaOMe (380 mg of Na in 5 mL of MeOH) was
added MeI (5 mL). The reaction mixture was heated at reflux for 2 h,
acidified with HCl (3 M), and extracted with EtOAc. Separation, drying,
and removal of organic solvent gave an orange residue. Column chroma-
tography on silica gel using CH2Cl2 gave the methylated product 14 as a
colorless gum (80 mg, 0.21 mmol, 36%): IR (film) νmax 2980, 2925, 2852,
1719, 1672, 1560, 1450, 1379, 1034 cm-1; UV (MeOH) λmax (log ε) 278,
(4.0), 230 (3.8) nm; HR-ESIMS m/z 375.2525 [M + H]+ (375.2530 calcd
for C23H35O4). NMR spectra of 14 showed two tautomers (approximate
ratio 3:2): 1H NMR (CDCl3, 300 MHz) δ 1.24 (m, H-40), 1.30 + 1.29 +
1.26 (m, 3-Me), 1.44 + 1.40 + 1.39, (m, 5-Me), 1.49 + 1.17 (m, H-60 +H-80),
1.49 (m, H-30), 1.60 + 1.56 (m, H-400) + 1.52 + 1.47 (m, H-500), 1.66 +
0.93 (m, H-50 + H-90), 1.71 (m, H-70), 2.52 (m, H-100), 2.98 (m, H-20),
4.76 (m, H-200), 18.34 (s, 2-OH); 13C NMR (CDCl3, 75 MHz) major
tautomer δ 17.8 (C-500), 20.3 (3/5-Me), 22.3 (3/5-Me), 25.8 (C-400), 26.1
(3/5-Me), 26.2 (C-60 + C-80), 26.5 (C-70), 32.4 (C-30), 33.1 (C-50 + C-90),
37.1 (C-40), 37.5 (C-20), 38.9 (C-100), 51.9 (C-3), 57.2 (C-5), 110.8 (C-1),
118.0 (C-200), 137.1 (C-300), 196.5 (C-6), 197.7 (C-2), 205.1 (C-10),
210.2 (C-4); minor tautomer δ 17.9 (C-500), 20.8 (3/5-Me), 22.2 (3/5-Me),
25.9 (C-400), 26.1 (3/5-Me), 26.2 (C-60 +C-80), 26.5 (C-70), 32.4 (C-30), 33.1
(C-50 + C-90), 37.0 (C-40), 37.5 (C-20), 38.0 (C-100), 51.4 (C-3), 56.1 (C-5),
110.1 (C-1), 117.5 (C-200), 135.8 (C-300), 196.2 (C-6), 199.1 (C-2), 205.0
(C-10), 209.8 (C-4).
Compound 10. β-Triketone 9 (100 mg, 0.31 mmol) and silver oxide
(50 mg, 0.22 mmol) were refluxed for 5 h with MeI (2.5 mL). The MeI was
left to evaporate and the residue taken up in CH2Cl2 and filtered through
Celite. Column chromatography with CH2Cl2/EtOAc (1:0 to 5:1) gave the
O-methylated product 10 as an amorphous white solid (50 mg, 0.15 mmol,
48%): IR (film) νmax 2982, 2925, 2852, 1723, 1672, 1558, 1472, 1449, 1382,
1049 cm-1; UV (MeOH) λmax (log ε) 278 (4.2) nm; HREIMS (70 eV) m/z
(rel int) 334.2132 [M]+ (2%, 334.2144 calcd for C20H30O4), 319 (30), 223
(100), 191 (44), 168 (70); 1H NMR (CDCl3, 300 MHz) δ 0.89 (2H, m,
H-50 + H-90), 1.16 (2H, m, H-60 + H-80), 1.24 (1H, m, H-40), 1.35 (6H, s,
5-Me), 1.40 (6H, s, 3-Me), 1.53 (2H, m, H-30), 1.60 (2H, m, H-60 + H-80),
1.62 (2H, m, H-70), 1.68 (2H, m, H-50 + H-90), 2.73 (2H, t, J = 7 Hz, H-20),
4.77 (3H, s, 6-O-Me); 13C NMR (CDCl3, 75 MHz) δ 23.9 (3-Me), 25.3
(5-Me), 26.2 (C-60 + C-80), 26.5 (C-70), 33.1 (C-50 + C-90), 31.4 (C-30),
37.2 (C-40), 42.9 (C-20), 49.6 (C-3), 55.6 (C-5), 61.3 (6-O-Me), 117.5 (C-1),
174.8 (C-6), 198.7 (C-2), 205.5 (C-10), 211.1 (C-4).
Compound 14. To the 2-cyclohexylethyl phloroglucinol described
previously (19) (500 mg, 1.9 mmol) in acetone (5 mL) was added
K2CO3 (240 mg, 1.74 mmol) and prenyl bromide (0.2 mL, 1.9 mmol).
The mixture was heated at reflux for 4 h, acidified with HCl (3 M) and
extracted with EtOAc. The solvent was then removed to give a yellow
residue. Column chromatography on silica gel with CH2Cl2/EtOAc gave
the monoprenylated product phloroglucinol 20 as a yellow gum (200 mg,
0.6 mmol, 32%): HR-ESIMS m/z 333.2038 [M + H]+ (333.2060 calcd for
C20H29O4); 1H NMR (CDCl3, 300 MHz) δ 1.03 + 1.82 (m, H-50 + H-90),
1.21 (m, H-40), 1.29 + 1.70 (m, H-60 + H-80), 1.56 (m, H-30), 1.74 (m, H-70),
Compound 15. H2O2 (30%, 1.0 mL) was added to leptospermone 4
(100 mg, 0.38 mmol) suspended in dioxane (1.5 mL) and stirred at room
temperature for 24 h. The reaction mixture was diluted with H2O and
extractedwith CH2Cl2. Semipreparative HPLC (see below) gaveunreacted
leptospermone and the new compound 15 as an amorphous white solid
(22 mg, 0.05 mmol, 14%, 0.39 mmol): IR (film) νmax 2961, 2875, 1767,
1721, 1615, 1470, 1385, 1181, 1094, 1034 cm-1; UV (MeOH) λmax (log ε)
262 (4.0) nm; HREIMS (70 eV) m/z (rel int) 282.1466 [M]+ (trace,

5196 J. Agric. Food Chem., Vol. 57, No. 12, 2009
Dayan et al.
282.1467 calcd for C15H22O5), 198 (10), 140 (11), 85 (100), 57 (100);
1H NMR (CDCl3, 300 MHz) δ 1.03 (6H, d, J = 7 Hz, H-40 + H-50), 1.40
(12H, br s, 3-Me + 5-Me), 2.16 (1H, m, J = 7 Hz, H-30), 2.42 (2H, d, J =
7 Hz, H-20), 6.04 (1H, s, H-1); 13C NMR (CDCl3, 75 MHz) δ 22.2 (C-40 +
C-50), 24.3 (3-Me + 5-Me), 25.9 (C-30), 42.6 (C-20), 56.3 (C-3 + C-5)
79.9 (C-1), 170.0 (C-10), 198.7 (C-2 + C-6), 210.8 (C-4); NMR signals
assigned from HMBC spectrum.
The active site on the A. thaliana HPPD was defined as a spherical
˚
region of 7 A radius centered from the β-triketone ligands. This sphere
encompasses all residues known to be involved in inhibitor binding (30).
The default Sybyl/Flexidock parameters were used. Of the several possible
conformations obtained for each ligand, the conformation with the lowest
binding energy to At-HPPD was selected for the CoMFA study.
CoMFA and Prediction of the Inhibitory Activity. CoMFA studies
require that the molecules to be analyzed be aligned according to a
suitable conformational template, which is assumed to be a “bioactive”
conformation (31). Therefore, all of the active molecules were “relaxed”
within the binding domain of HPPD using MMF94 (32 ). The structures
were then aligned along carbons 2, 4, and 6 of their triketone backboneand
their 10 atom (a carbonyl carbon for all the structure, except for 15 and 16)
(Figure 1). All analyses were performed using the default lattice parameters
Semipreparative HPLC was carried out using a Waters system com-
prising a 717 autosampler, a 600 controller, and a 2487 programmable
multiwavelength detector, at 20 °C with a C18 column (Phenomenex Luna
˚
ODS(3) 5 μm 100 A 250 ꢀ 10 mm) with a 2 ꢀ 4 mm C18 guard column.
Peaks were monitored at 210 and 254 nm. The mobile phase was initially
50:50 0.1% formic acid in H2O/0.1% formic acid in MeCN with a linear
gradient over 10 min to 0.1% formic acid in MeCN, with a flow rate of
5 mL min-1, giving compound 15 with a retention time of 7.5 min.
Compound 16. Hydroxylamine hydrochloride (30 mg) and sodium
acetate (45 mg, 0.54 mmol) were added to a solution of β-triketone 9 (19 )
(50 mg, 0.16 mmol) in H2O (1 mL) and ethanol (2 mL). The reaction mixture
was heated at reflux for 18 h. Column chromatography on silica gel with
CH2Cl2 gave the isoxazole 16 as a colorless gum (48 mg, 0.15 mmol, 97%):
˚
for CoMFA consisting of a three-dimensional grid with a width of 2 A.
CoMFA descriptors were calculated using a sp3 carbon probe atom with a
˚
van der Waals radius of 1.5 A and a charge of +1.0 to generate steric field
energies and electrostatic fields with a distance-dependent dielectric at each
lattice point. The Sybyl default energy cutoff of 30 kcal/mol was used. The
CoMFA steric and electrostatic fields generated were scaled by the
CoMFA standard method in Sybyl.
IR (film) νmax 2981, 2925, 2852, 1724, 1683, 1608, 1474, 1455, 1047 cm-1
;
UV (MeOH) λmax (log ε) 230 (3.9) nm; HR-ESIMS m/z 318.2053 [M + H]+
(318.2064 calcd for C19H27NO3); 1H NMR (CDCl3, 300 MHz) δ 0.97
(2H, m, H-50 + H-90), 1.19 (2H, m, H-60 + H-80), 1.26 (1H, m, H-40), 1.39
(6H, s, 5-Me), 1.56 (6H, s, 3-Me), 1.59 (2H, overlapped m, H-30), 1.60 (2H,
m, H-60 + H-80), 1.62 (2H, m, H-70), 1.74 (2H, m, H-50 + H-90), 2.91 (2H, br
t, J = 7 Hz, H-20); 13C NMR (CDCl3, 75 MHz) δ 23.2 (3-Me), 24.0 (C-20),
25.4(5-Me), 26.2 (C-60 + C-80), 26.6 (C-70), 32.9 (C-50 + C-90), 34.7 (C-30),
37.5 (C-40), 45.8 (C-3), 57.3 (C-5), 111.4 (C-1), 162.3 (C-2), 181.8 (C-10),
192.4 (C-6), 211.4 (C-4).
HINT (Hydropathic Interactions) Analysis. The catalytic and
substrate binding domain of A. thaliana HPPD was extracted from the
˚
entire protein by selecting all of the amino acids within a 6 A radius around
13, the longest β-triketone present in the data set. The partition coefficient
of this domain was calculated using all atoms (including essential hydro-
gens) and following the partition dictionary for the amino acids available
in the HINT (33) module of Sybyl. The polar/hydropathic map was then
˚
derived by identifying atoms interacting within a 6 A radius around 13.
HINT logP calculations can be imported in Sybyl to derive hydropathic
surface maps useful to visualize the non-covalent hydrophobic interactions
of ligands and the binding domain of proteins. The contour map of the
interaction between the binding domain and 13 was displayed with polar
regions in red and hydrophobic regions in white. The HINT logP values
for all compounds tested in this study were also calculated.
Expression of HPPD and Enzymatic Assays. Recombinant HPPD
from Arabidopsis thaliana was overexpressed in Escherichia coli and
purified by immobilized metal affinity chromatography. Enzyme activity
was measured as described before (25). The HPLC system used to measure
enzyme activity was composed of a Waters Corp. system (Milford, MA)
that included a model 600E pump, a model 717 autosampler, a Millenium
2010 controller, and a model 996 photodiode detector equipped with a
7.8 mm ꢀ 100 mm X-Terra C18 (5 μm) reversed phase column. Solvent A
was 0.1% (v/v) trifluoroacetic acid in ddH2O, and solvent B was 0.08%
(v/v) trifluoroacetic acid in 80% (v/v) HPLC-grade acetonitrile/ddH2O.
The solvent system consisted of a linear gradient beginning at 0% (100%
A) to 70% B from 0 to 2 min, 70-100% B from 2 to 4 min, 100% B from
4 to 6 min, 100-0% B from 6 to 7 min, and 0% B from 7 to 8 min. The flow
rate was 3 mL min-1, and the injection volume was 100 μL. HGA was
detected by the UV absorbance at 288 nm (26). A calibration curve
was established by injecting various concentrations of HGA. Data from
dose-response experiments were analyzed using the dose-response curve
module (27) using R version 2.2.1 (28). Mean I50 values and standard
deviations, obtained using the untransformed data, are given in Table 1.
The synthetic HPPD inhibitor sulcotrione was included as a positive
control (25).
RESULTS AND DISCUSSION
The β-triketone natural products 2 and 4-7 and structural
analogues 1, 3, 8, 9, 11-13, and 17-19 (Figure 1) were synthesized
for an earlier study examining the relationships of their structures to
their antibacterial activity (19 ). These compounds contain a wide
range of hydrophobic alkyl side chains attached to C-10, and either
tetramethyl, tetraethyl, or tetraprenyl substituents on the triketone
ring (Figure 1). For the current work on the herbicidal activity of
triketones, the new compounds 10 and 14-16 were synthesized and
characterized. Compound 10, with the β-triketone moiety modified
but still containing a 1,3-diketone moiety, was prepared by simple
methylation of 9. NOESY NMR of 10 showed correlations between
the methoxyl protons and the C-30-methyl protons and the H-20 and
H-30 protons on the flexible ethylene bridge.
The 1,3-diketone moiety was removed in compounds 15 and 16.
The oxidation of leptospermone 4 to give 15 was based on the
reported reactions of another natural β-triketone, hyperforin (34 ).
Several peroxidic reagents were tested to induce oxidative rearran-
gement, and it was found that suspending leptospermone 4 in
dioxane and treating it with a high concentration of hydrogen
peroxide gave a low yield (16%) of 15. 2D NMR confirmed the
proposed ester structure 15, in particular an HMBC correlation
between H-1 and C-10. The presence of H-1 showed that the most
stable form of ester 15 had nonconjugated C-2 and C-6 ketone
groups, rather than the conjugated ketone-enol found in the
β-triketones (16 ). Isoxazole 16 was obtained in good yield by
treating β-triketone 9 with hydroxylamine, as per the method of
Du Bois et al. (35 ) with spectroscopic data supporting the
isoxazole structure. Compound 14, with a single prenyl group on
the ring, was prepared via monoprenyl phloroglucinol 20, which
was then permethylated.
Docking of the β-Triketones to HPPD. Molecular modeling was
performed using Sybyl v.7.1 software from Tripos Inc. (St. Louis, MO) on
a Silicon Graphics Octane 2 workstation, equipped with two parallel
R12000 processors. The homology model of A. thaliana designed in
our previous research (25 ) was used. A subsequent minimization was
performed on the transferred ligands with the Tripos force field, as
implemented in the Biopolymer module of Sybyl. The root-mean-square
(rms) differences between our model and the recently published crystal
structure of A. thaliana HPPD (29) (pdb: 1TG5) was 0.885, indicating the
high similarity between the two structures.
The initial structures of β-triketone analogues were derived from the
coordinates of leptospermone obtained previously (25). Each structure
was minimized for 1000 steps each of steepest descent, followed by
conjugate gradient, and finally by the BFGS method to a gradient of
˚
0.001 kcal/mol/A or less. The charges were added to the molecules using
the Sybyl Gasteiger-Huckel algorithm. These analogues were docked to
the active site of At-HPPD using the Flexidock (Genetic algorithm-based
flexible docking) routine. FlexiDock explores the conformational and
orientational space that defines possible interactions between the ligand
and its binding site.

Article
J. Agric. Food Chem., Vol. 57, No. 12, 2009 5197
were inactive, whereas most of the tetramethyl compounds were
active (Table 1). The monoprenyl-substituted 14 was active (I50
=
2,270 nM), but was 10 times less potent than the tetramethyl
analogue 9. Compound 14 has a chiral center at C-5 but was
synthesized as a racemic mixture of enantiomers, which occurs as
a 3:2 mixture of tautomers. The moderate inhibitory activity of
this mixture indicated that one or more of these configurations
interacted with the catalytic site of HPPD. Analysis of the binding
domain, however, revealed that the space at the proximity of the
Fe atom is limited, with the exception of a channel toward the
bottom of the domain. Therefore, the bioactive form of 14 was
presumed to be its C-5 carbon with an R absolute configuration
(i.e., with the prenyl group β-oriented). Additionally, the binding
domain could accommodate only the tautomeric form with the
C-5 prenyl side chain pointing down. Therefore, this configura-
tion of 14 was included in the data set used for the structure-
activity relationship analysis.
The side chain on the β-triketones (R1, Figure 1) had a major
influence on the activity. For example, the presence of a simple
methyl group (1) or cyclohexyl group (8) yielded inactive com-
pounds, whereas the presence of a nonyl side chain (11) resulted in
the most active compound in this study (I50 = 19 nM, Table 1).
Other structure-activity work performed during the develop-
ment of synthetic β-triketone herbicides also reported that
compounds with short aliphatic groups on C-10 had little to no
activity (24).
Figure 2. Inhibition ofHPPD by triketones2 (4), 4 (3), 5 (O), 6 (0), 7 (]),
12 (ꢀ), 3 (2), 9 (1), 13 (b), 11 (9), 10 ([), and 14 (`). Each data
point represents the mean of three independent experiments ( 1 SD.
Compounds 1, 8, and 15-19 are not shown because their I50 values were
>50 μM.
In the previous work on the antibacterial structure-activity
relationships (SAR) of β-triketones, there was a simple relation-
ship between antibacterial potency and the calculated water-
octanol partition coefficient (Clog P) (19). In that study, 12 with
the undecyl side chain was the most active compound in the
sample set. Compounds with higher or lower Clog P values were
progressively less active (19). Previous works on the activity of
natural products against plant HPPD also suggested that side-
chain lipophilicity was also important in this biological
system (25, 39). However, Clog P values for the β-triketones in
the present study showed that there was no simple relationship
between this parameter and activity against HPPD (Table 1). For
example, compound 8 with cyclohexyl ring attached to C-10 has a
Clog P similar to that of grandiflorone 6, but it was at least
60 times less active against HPPD. HINT log P values (40)
correlated closely with Clog P values (Table 1) and did not
provide a better understanding of the contribution of log P to
the activity of these compounds.
Therefore, 3D modeling techniques of the ligand/receptor
interactions and CoMFA were used. The CoMFA method
evaluates variations in steric and electronic field potentials of
structural analogues that have been overlaid on a three-dimen-
sional lattice and correlates these surfaces with biological activity
using partial least-squares (PLS) analysis (31). The mathematical
models obtained can be used to predict the activity of an analogue
not included (leave-one-out) inthe originaldata set orasa toolfor
visualizing SAR surfaces maps in three dimensions. CoMFA
results include an r2 value, which indicates how much of the data
set’s variation is accounted for by the model, and a cross-
validated r2 (q2), which indicates how well biological activity is
predicted for each compound by the other analogues in the data
set (31). Experimentally, any model with a q2 > 0.5 has a strong
and statistically valid predictive power (31).
These triketones and related structures (Figure 1) were tested
for their HPPD inhibitory activity as previously described (25).
Dose-response curves against overexpressed and purified A.
thaliana HPPD were obtained for all of the compounds (Figure 2),
and I50 values were calculated using a four-parameter logistic
function (Table 1). The results obtained with the leptospermone 4
and grandiflorone 6 synthesized in the laboratory (Table 1) were
within experimental error of the results earlier reported for
these compounds isolated from L. scoparium essential oil (25).
Grandiflorone 6 was the most active of the naturally occurring
β-triketones (I50 = 750 nM), which was closely followed by
papuanone 7 (I50 = 960 nM). These levels of activity were close to
that of the synthetic herbicide sulcotrione (I50 = 250 nM).
Grandiflorone 6 is present at low levels in some essential oils
distilled from leaves of L. scoparium (14) and is the major
component in essential oil from leaves of L. morrisonii (17).
Accumulation of toxic secondary metabolites provides protection
against herbivory and may also participate in allelopathy. In fact,
the allelopathic properties of Callistemon citrinus were traced to
the β-triketones it produced (24), which ultimately led to the
discovery and development of the β-triketone synthetic
herbicides (7 ).
The mechanism that enables Leptospermum species to be
protected from these potent natural herbicides is not well under-
stood. However, leaves of Leptospermum and other Myrtaceae
species possess schizogenous cavities (36). These specialized
structures, along with glandular trichomes and other glands,
are often the site of synthesis and/or serve as a repository for
bioactive natural products (37), thereby compartmentalizing
these toxins and protecting the rest of the plant from autotoxic
effects (38).
The synthetic compounds with no 1,3-diketone moiety, that is,
the ester 15and isoxazole 16, were inactive(I50 > 50μM; Table 1).
This is consistent with the known requirement for the
1,3-diketone moiety to interact with the Fe2+ in the active site
of HPPD. The O-methylated compound 10 retained this moiety
and remained active (I50 = 410 nM), although to a lower degree
than the parent β-triketone 9 (Table 1). Compounds 17-19 with
bulky tetraprenyl or tetraethyl substituents on the triketone ring
Structures were built and minimized in Sybyl and docked into a
homology model of A. thaliana HPPD (At-HPPD) that had been
developed previously(25). The coordinates of the protein-ligand
interactions between the keto groups of the cocrystallized ligand
and the Fe2+ ion in the catalytic domain NTBC were used to
pre-position the compounds used this study. The coordinates of

5198 J. Agric. Food Chem., Vol. 57, No. 12, 2009
Dayan et al.
Figure 4. CoMFA maps of the β-triketones: (A) steric map where green
areas indicate increase in steric bulk is favored and yellow areas indicate
increaseinstericbulkisnotfavored; (B) electrostaticmap whereblue areas
indicate positive charge and H-bond donors are favored and negative
charge and H-bond acceptors are not favored. Red areas indicate negative
charge and H-bond acceptor are favored and positive charge and H-bond
donors are not favored.
Figure 3. Observed and CoMFA predicted activities of β-triketones
against HPPD. Cross-validated partial least-squares (PLS) analysis had
an r2 = 0.96, and the ability of the model to predict activity is q2 = 0.63.
The activities of sulcotrione S and 8, which were not used to derive the
CoMFA model, are shown in large type.
NTBC complexed with HPPD (30) were used instead of those
of the pyrazole present in the A. thaliana HPPD crystal
structure (29) because of the greater similarity between the
triketones used in this study and NTBC than between the
triketones and the pyrazole (see Figure S1 in the Supporting
Information). Once pre-positioned in the catalytic site, the
hydrophobic side chains were permitted to relax within the
boundary of the binding domain to estimate their bioactive
conformations and relative binding energies (Table 1).
The structures were then aligned on their common atoms
within the triketone ring and subjected to CoMFA, which
resulted in a strong linear relationship between the measured
and predicted inhibitory activity, with an r2 of 0.96 (Figure 3). The
q2 value of 0.63 suggests that the model can predict the activity of
structurally related triketones. The standard error of regression
(RSE) was near the expected noise level, indicating that the
model was not overspecified. The CoMFA model indicates that
66% of the relationship was contributed by the steric energies of
interaction, and 34% was due to electrostatic energies.
Compounds with large tetraethyl or tetraprenyl groups on the
β-triketone ring (17-19) were not active because of steric hin-
drance near the catalytic site. The volume of the tetramethyl cyclic
3
˚
triketone moiety of compounds 1-12 (184.7 A ) is sufficiently
Figure 5. Binding mode of (A) leptospermone (4) and (B) 11 within the
substrate binding domain of HPPD.
3
˚
small to fit within the pocket, whereas the volumes of 17 (254.2 A )
3
˚
and of 18 and 19 (438.3 A ) prohibit binding (Table 1). The
inability of larger β-triketones to fit within this space was reflected
by their exceedingly high calculated binding energies (Table 1).
CoMFA maps identified that the presence of bulky substituents on
the ring structure had a negative impact on the activity of the
compounds (yellow surfaces in Figure 4A). This is consistent with
the observation that the volume of the binding domain can
accommodate the β-triketones from the tetramethyl homologous
series, but it is not large enough for triketones decorated with
bulkier side chains (17-19).
On the other hand, the calculated binding energies of 15 and 16
suggest that the compounds could fit within the substrate binding
domain (Table 1), but were not active due to the absence of the
1,3-diketone moiety (Figure 1) that is necessary to form coordi-
nated octahedral complexes with the Fe2+ ion and three strictly
conserved active site residues in the catalytic site (25, 41).
The CoMFA map identifies this region as important for electro-
static interactions (see blue and red surfaces surrounding the
1,3-diketone moiety in Figure 4B).
binding domain regardless of the size or length of the R1 side
chain (Figure 5). The presence of a cyclohexyl ring (or benzyl ring)
does not preclude activity, as observed with 3, 6, and 9, but its
proximity to the C-10 of the triketone ring has a strong effect on
activity. The structure-activity relationship between 8, 3, 6, 10,
and 9 clearly shows that increasing the distance between a side-
chain ring and C-10 improves activity (Table 1).
The I50 value of 8 could not be calculated from the dose-
response curve because the highest concentration tested (100 μM)
provided only 35% inhibition of HPPD. However, the CoMFA
model predicted an I50 of 69 μM for 8, which is a reasonable
approximation of its low inhibitory activity against HPPD
(see estimated position of 8 in Figure 3). The CoMFA map
identifies the region proximal to the C-10 of the triketone ring as
particularly sensitive to steric hindrance (yellow area in
Figure 4A). At the other end of the activity range, the CoMFA
model predicted the I50 of sulcotrione to be 7 nM, whereas its
measured activity I50 against HPPD was 250 nM (see position of
S in Figure 3). This overestimation of the activity of sulcotrione
may be due to the fact that none of the structures used to derive
The relatively low binding energies of the remaining β-trike-
tones (Table 1) suggest that they fit well within the substrate

Article
J. Agric. Food Chem., Vol. 57, No. 12, 2009 5199
LITERATURE CITED
(1) Que, L.; Ho, R. Y. N. Dioxygen activation by enzymes with
mononuclear non-heme iron active sites. Chem. Rev. 1996, 96,
2607–2624.
(2) Crouch, N. P.; Adlington, R. M.; Baldwin, J. E.; Lee, M.-H.;
MacKinnnon, C. H. A mechanistic rationalisation for the substrate
specificity of recombinant mammalian 4-hydroxyphenylpyruvate
dioxygenase. Tetrahedron 1997, 53, 6993–7010.
(3) Pascal, R. A.; Oliver, M. A.; Chen, Y. C. J. Alternate substrates
and inhibitors of bacterial 4-hydroxyphenylpyruvate dioxygenase.
Biochemistry 1985, 24, 3158–3165.
(4) Norris, S. R.; Barrette, T. R.; DellaPenna, D. Genetic dissection of
carotenoid synthesis in Arabidopsis defines plastoquinone as an
essential component of phytoene desaturation. Plant Cell 1995, 7,
2139–2149.
(5) Pallett, K. E.; Little, J. P.; Sheekey, M.; Veerasekaran, P. The mode
of action of isoxaflutole I. Physiological effects, metabolism, and
selectivity. Pestic. Biochem. Physiol. 1998, 62, 113–124.
(6) Schulz, A.; Ort, O.; Beyer, P.; Kleinig, H. SC-0051, a 2-benzoyl-
cyclohexane-1,3-dione bleaching herbicide, is a potent inhibitor of
the enzyme p-hydroxyphenylpyruvate dioxygenase. FEBS Lett.
1993, 318, 162–166.
(7) Lee, D. L.; Prisbylla, M. P.; Cromartie, T. H.; Dagarin, D. P.;
Howard, S. W.; Provan, W. M.; Ellis, M. K.; Fraser, T.; Mutter,
L. C. The discovery and structural requirements of inhibitors of
p-hydroxyphenylpyruvate dioxygenase. Weed Sci. 1997, 45,
601–609.
(8) Viviani, E.; Little, J. P.; Pallett, K. E. The mode of action of
isoxaflutole II. Characterization of the inhibition of carrot 4-hydro-
xyphenylpyruvate dioxygenase by the diketonitrile derivative of
isoxaflutole. Pestic. Biochem. Physiol. 1998, 62, 125–134.
(9) Schultz, T. W.; Bearden, A. P. Structure-toxicity relationships for
selected naphthoquinones to Tetrahymena pyriformis. Bull. Environ.
Contam. Toxicol. 1998, 61, 405–410.
(10) Senseman, S. A. Herbicide Handbook, 9th ed.; Weed Science Society
of America: Lawrence, KS, 2007; p 458.
(11) Williams, M. M.; Pataky, J. K. Genetic basis of sensitivity in sweet
corn to tembotrione. Weed Sci. 2008, 56, 364–370.
(12) Lee, D. L.; Knudsen, C. G.; Michaely, W. J.; Chin, H.-L.; Nguyen,
N. H.; Carter, C. G.; Cromartie, T. H.; Lake, B. H.; Shribbs, J. M.;
Fraser, T. The structure-activity relationships of the triketone class
of HPPD herbicides. Pestic. Sci. 1998, 54, 377–384.
(13) Gray, R. A.; Tseng, C. K.; Rusay, R. J. 1-Hydroxy-2-(alkylketo)-
4,4,6,6-tetramethyl cyclohexen-3,5-dione herbicides. U.S. Patent
4,227,919, 1980.
Figure 6. HINT polar/hydrophobic surface map of the catalytic site and
binding domain of HPPD. Red surfaces represent polar areas and white
surfaces represent hydrophobic areas contributed by the amino acids lining
the substrate binding domain. Note the highly hydrophobic region
surrounding the aliphatic tail of the inhibitor (13).
the CoMFA model had the chloro and methane sufonyl
groups built into this synthetic herbicide. The presence of strong
electron-withdrawing groups para on the phenyl ring of synthetic
herbicidal compounds increased activity by 40-fold (12, 42).
The biological activities 12 and 13 were lower than that of 11,
which has a shorter and less lipophilic side chain. Analysis of the
relationship between log P and activity could not account for this,
but the three-dimensional CoMFA approach successfully pre-
dicted the activity of these compounds (Figure 3). The steric
CoMFA map identified this by the presence of a yellow area
toward the end of the hydrophobic tail (Figure 4A).
Finally, the substrate binding domain of HPPD consists of a
lipophilic region lined with Met314, Pro315, Pro318, Phe360,
Phe371, Phe403, Leu405, and Phe407 (Figure 6). This domain
favors the binding of ligands with lipophilic moieties. For this
reason, triketones with long hydrophobic side chains tend to have
higher inhibitory activity than those with short side chains. This is
consistent with previous observations that the biological activity
of both natural and synthetic β-triketones is greatly affected by
the nature of any side chain attached to the triketones (24, 25, 43).
Synthetic triketone-based HPPD inhibitors were discovered
following a study on the phytotoxic properties of leptospermone
4 (24). The modeling ofβ-triketoneinteractionswithplant HPPD
reported above has yielded important information to further
understand the basis for inhibition by these natural product
derived herbicides. Although the physicochemical properties of
the β-triketones used in this study are suboptimal when compared
to those of the synthetic products commercially available, work is
underway to determine the in vivo activity of these molecules and
the potential use of the natural β-triketones as herbicides in
organic agriculture.
(14) Douglas, M. H.; van Klink, J. W.; Smallfield, B. M.; Perry, N. B.;
Anderson, R. E.; Johnstone, P.; Weavers, R. T. Essential oils
from New Zealand manuka: triketone and other chemotypes of
Leptospermum scoparium. Phytochemistry 2004, 65, 1255–1264.
(15) Hellyer, R. O. The occurrence of β-triketones in the steam-volatile
oils of some myrtaceous australian plants. Aust. J. Chem 1968, 21,
2825–2828.
(16) van Klink, J. W.; Brophy, N. B.; Perry, N. B.; Weavers, R. T.
Triketones from myrtaceae: isoleptospermone from Leptospermum
scoparium and papuanone from Corymbia dallachiana. J. Nat. Prod.
1999, 62, 487–489.
ABBREVIATIONS USED
(17) Brophy, J. J.; Goldsack, R. J.; Bean, A. R.; Forster, P. I.; Lepschi, B.
J. Leaf essential oils of the genus Leptospermum (Myrtaceae) in
eastern Australia. Part 6. Leptospermum polygalifolium and allies.
Flavour Fragrance J. 2000, 15, 271–277.
HPPD, p-hydroxyphenylpyruvate dioxygenase; 4-HPP, 4-hydro-
xyphenylpyruvate; HGA, homogentisic acid; NTBC, 2-[2-nitro-
4-(trifluoromethyl)benzoyl]-1,3-cyclohexanedione.
(18) Rubinov, D. B.; Rubinova, I. L.; Akhrem, A. A. 2-Acylcycloalkane-
1,3-diones. Chem. Nat. Compd. 1995, 31, 537–559.
ACKNOWLEDGMENT
(19) van Klink, J. W.; Larsen, L.; Perry, N. B.; Weavers, R. T.; Cook, G.
M.; Bremer, P. J.; MacKenzie, A. D.; Kirika, T. Triketones active
against antibiotic-resistant bacteria: synthesis, structure-activity
relationships, and mode of action. Bioorg. Med. Chem. 2005, 13,
6651–6662.
We thank E. Burgess and S. B. Watson for their technical
assistance, B. Clark and I. Stewart for MS, and M. Thomas for
^
assistance with NMR. We are also grateful to Rhone-Poulenc for
the generous gift of the recombinant HPPD.
(20) Christoph, F.; Kaulfers, P.-M.; Stahl-Biskup, E. A comparative
study of the in vitro antimicrobial activity of tea tree oils with
special reference to the activity of β-triketones. Planta Med. 2000,
66, 556–560.
Supporting Information Available: Structures of leptosper-
mone, NTBC, and DAS645. This material is available free of
charge via the Internet at http://pubs.acs.org.

5200 J. Agric. Food Chem., Vol. 57, No. 12, 2009
Dayan et al.
(21) Spooner-Hart, R. N.; Basta, A. H. Pesticidal compositions contain-
ing β-diketones and β-triketones from essential oils. Pct. Int. Appl.
Coden PIXXD2 WO 2002089587, AI 20021114, CAN 137:347894,
AN 2002:868663, 2002; 88 pp.
(32) Halgren, T. A. Maximally diagonal force constants in dependent
angle-bending coordinates. II. Implications for the design of empiri-
cal force fields. J. Am. Chem. Soc. 1990, 112, 4723–4728.
(33) Kellogg, G. E.; Semus, S. F.; Abraham, D. J. HINT: a new method of
empirical hydrophobic field calculation for CoMFA. J. Comput.-
Aid. Mol. Des. 1991, 5, 545–552.
(34) Verotta, L.; Lovaglio, E.; Sterner, O.; Appendino, G.; Bombardelli,
E. Oxidative fragmentation of the bridged β-triketone core of
hyperforin. Eur. J. Org. Chem. 2004, 6, 1193–1197.
(35) DuBois, G. E.; Crosby, G. A.; Stephenson, R. A. Dihydrochalcone
sweeteners;a study of the atypical temporal phenomena. J. Med.
Chem. 1981, 24, 408–428.
(36) List, S.; Brown, P. H.; Walsh, K. B. Functional anatomy of the oil
glands of Melaleuca alternifolia (Myrtaceae). Aust. J. Bot. 1995, 43,
629–641.
(37) Dayan, F. E.; Duke, S. O. Trichomes and root hairs: natural
pesticide factories. Pestic. Outlook 2003, 4, 175–178.
(38) Sirikantaramas, S.; Yamakazi, M.; Saito, K. Mechanisms of
resistance to self-produced toxic secondary metabolites in plants.
Phytochem. Rev. 2008, 7, 467–477.
(39) Meazza, G.; Scheffler, B. E.; Tellez, M. R.; Rimando, A. M.;
Nanayakkara, N. P. D.; Khan, I. A.; Abourashed, E. A.; Romagni,
J. G.; Duke, S. O.; Dayan, F. E. The inhibitory activity of
natural products on plant p-hydroxyphenylpyruvate dioxygenase.
Phytochemistry 2002, 59, 281–288.
(40) Kellogg, G. E.; Abraham, D. J. Hydrophobicity: is LogPo/w more
than the sum of its parts?. Eur. J. Med. Chem. 2000, 35, 651–661.
(41) Kakidani, H.; Hirai, K. Three-dimensional modeling of plant
4-hydroxyphenylpyruvate dioxygenase, a molecular target of trike-
tone-type herbicides. J. Pestic. Sci. 2003, 28, 409–415.
(22) Christoph, F.; Kubeczka, K. H.; Stahl-Biskup, E. The composition
of commercial manuka oils from New Zealand. J. Essent. Oil Res.
1999, 11, 705–710.
(23) Reichling, J.; Koch, C.; Stahl-Biskup, E.; Sojka, C.; Schnitzler, P.
Virucidal activity of a β-triketone-rich essential oil of Leptospermum
scoparium (manuka oil) against HSV-1 and HSV-2 in cell culture.
Planta Med. 2005, 71, 1123–1127.
(24) Knudsen, C. G.; Lee, D. L.; Michaely, W. J.; Chin, H.-L.; Nguyen,
N. H.; Rusay, R. J.; Cromartie, T. H.; Gray, R.; Lake, B. H.; Fraser,
T. E. M.; Cartwright, D. Discovery of the triketone class of HPPD
inhibiting herbicides and their relationship to naturally occurring
β-triketones. In Allelopathy in Ecological Agriculture and Forestry;
Narwal, S. S., Ed.; Kluwer Academic Publishers: Dordrecht, The
Netherlands, 2000; pp 101-111.
(25) Dayan, F. E.; Duke, S. O.; Sauldubois, A.; Singh, N.; McCurdy, C.;
Cantrell, C. L. p-Hydroxyphenylpyruvate dioxygenase is a herbici-
dal target site for β-triketones from Leptospermum scoparium.
Phytochemistry 2007, 68, 2004–2014.
(26) Garcia, I.; Rodgers, M.; Pepin, R.; Hsieh, T.-Z.; Matringe, M.
Characterization and subcellular compartmentation of recombinant
4-hydroxyphenylpyruvate dioxygenase from Arabidopsis in trans-
genic tobacco. Plant Physiol. 1999, 119, 1507–1516.
(27) Ritz, C.; Streibig, J. C. Bioassay analysis using R. J. Statist. Soft
2005, 12, 1–22.
(28) R-Development-Core-Team. R: A Language and Environment for
Statistical Computing, 2.2.1; R Foundation for Statistical Comput-
ing: Vienna, Austria, 2005.
(42) Ellis, M. K.; Whitfield, A. C.; Gowans, L. A.; Auton, T. R.; Provan,
W. M.; Lock, E. A.; Lee, D. L.; Smith, L. L. Characterization of the
interaction of 2-[2-nitro-4-(trifluoromethyl)benzoyl]-4,4,6,6-tetra-
methylcyclohexane-1,3,5-trione with rat hepatic 4-hydroxyphenyl-
pyruvate dioxygenase. Chem. Res. Toxicol. 1996, 9, 24–27.
(43) Huang, M.; Yang, D.-Y.; Shang, Z.; Zou, J.; Yu, Q. 3D-QSAR
studies on 4-hydroxyphenylpyruvate dioxygenase inhibitors by
comparative molecular field analysis (CoMFA). Bioorg. Med. Chem.
Lett. 2002, 12, 2271–2275.
(29) Yang, C.; Pflugrath, J. W.; Camper, D. L.; Foster, M. L.; Pernich, D.
J.; Walsh, T. A. Structural basis for herbicidal inhibitor selectivity
revealed by comparison of crystal structures of plant and mamma-
lian 4-hydroxyphenylpyruvate dioxygenases. Biochemistry 2004, 43,
10414–10423.
(30) Brownlee, J. M.; Johnson-Winters, K.; Harrison, D. H.; Moran, G.
R. Structure of the ferrous form of (4-hydroxyphenyl)pyruvate
dioxygenase from Streptomyces avermitilis in complex with the
therapeutic herbicide, NTBC. Biochemistry 2004, 43, 6370–6377.
(31) Cramer, R. D. I.; Patterson, D. E.; Bunce, J. D. Comparative
molecular field analysis (CoMFA). Effect of shape of binding of
steroids to carrier proteins. J. Am. Chem. Soc. 1988, 110, 5959–5967.
Received September 12, 2008. Revised manuscript received April 21,
2009. This research was supported in part by the New Zealand
Foundation for Research, Science and Technology.