The design and synthesis of inhibitors of dethiobiotin synthetase as potential herbicides

Article abstract of DOI:10.1002/(SICI)1096-9063(199903)55:3<236::AID-PS888>3.0.CO;2-0

Dethiobiotin synthetase (DTBS; E.C. 6.6.6.6), the penultimate enzyme in the biosynthesis of the essential vitamin biotin, is a new potential target for novel herbicides. Inhibitors were designed based on mechanistic and structural information. The in-vitro activities of these potential inhibitors versus the bacterial enzyme are reported here. Mimics of 7,8- diaminopelargonic acid (DAPA) or the DAPA carbamate reaction intermediate were substrates or partial substrates for the enzyme. Synergistic binding with ATP was noted with compounds which contained an amino functionality. NMR studies and X-ray structures confirmed that the inhibitors could be phosphorylated by the enzyme. Several series of potential inhibitors were designed to take advantage of this partial substrate activity by generating potentially more tightly bound phosphorylated inhibitors in situ. Structure- activity relationships for these series based on both substrate and inhibitory activity are described herein. An X-ray structure for one of these inhibitors is also discussed. Although considerable potential for inhibitors of this type was demonstrated, none of the compounds reported showed sufficient herbicidal activity to be a commercial proposition.

Full text of DOI:10.1002/(SICI)1096-9063(199903)55:3<236::AID-PS888>3.0.CO;2-0

Pesticide Science
Pestic Sci 55:236–247 (1999)
The design and synthesis of inhibitors of
dethiobiotin synthetase as potential herbicides¹
Alan R Rendina,1* Wendy S Taylor,1* Katharine Gibs on,2 George Lorimerº
Dennis Rayner,1 Bruce Lockett,1 Kevin Kranis ,1 Barry Wexler,1 Dana
Marcovici-Mizrahi,¸ Ayelet Nudelman,3 Abraham Nudelman,3 Eileen Mars ilii,1
Hongji Chi,1 Zdis law Wawrzak,2 Jos eph Calabres e,2 Weijun Huang,4 Jia Jia,4
Gunter Schneider,4 Ylva Lindqvis t4 and Guang Yang1
1 E.I. Du Pont de Nemours and Co, Stine-Has kell Res earch Center, Newark, DE 19714, USA
2 E.I. Du Pont de Nemours and Co, Experimental Station, Wilmington, DE 19880, USA
3 Bar Ilan Univers ity, Ramat Gan 52900, Is rael
4 Karolins ka Ins titute, Stockholm 17177, Sweden
Abstract: Dethiobiotin synthetase (DTBS; E.C. 6.6.6.6), the penultimate enzyme in the biosynthesis of
the essential vitamin biotin, is a new potential target for novel herbicides. Inhibitors were designed
based on mechanistic and structural information. The in-vitro activities of these potential inhibitors
versus the bacterial enzyme are reported here. Mimics of 7,8-diaminopelargonic acid (DAPA) or the
DAPA carbamate reaction intermediate were substrates or partial substrates for the enzyme. Syn-
ergistic binding with ATP was noted with compounds which contained an amino functionality. NMR
studies and X-ray structures conürmed that the inhibitors could be phosphorylated by the enzyme.
Several series of potential inhibitors were designed to take advantage of this partial substrate activity
by generating potentially more tightly bound phosphorylated inhibitors in situ. Structure–activity
relationships for these series based on both substrate and inhibitory activity are described herein. An
X-ray structure for one of these inhibitors is also discussed. Although considerable potential for
inhibitors of this type was demonstrated, none of the compounds reported showed sufficient herbi-
cidal activity to be a commercial proposition.
( 1999 Society of Chemical Industry
Keywords: diamine; diacid; phosphinate; phosphonate; phosphate; diaminopelargonic acid; DAPA; DAPA
carbamate; dethiobiotin synthetase; DTBS ; biotin biosynthesis; herbicide; inhibition
1
INTRODUCTION
commercial herbicides such as sethoxydim and quiz-
alofop.3 This enzyme is also a key branch point in
plant metabolism because the product, malonyl-
CoA, is used for biosynthesis of ýavonoids, stilbe-
noids, naphthoquinones and malonyl aminocyclo-
propane carboxylic acid, to name a few. Thus,
depletion of biotin in plants caused by inhibition of
one of the biosynthetic enzymes should result in
multiple pleiotropic eþects similar to those observed
on direct inhibition of acetyl-CoA carboxylase.
The biotin biosynthetic pathway is understood in
detail in microorganisms,4 but is relatively poorly
understood in plants. However, there is increasing
evidence that plants synthesize biotin using the same
route as that in Escherichia coli (Mig.) Cast. &
Biotin plays a key role in carboxylation reactions
which are critical to microorganisms, animals and
plants.1,2 Thus, the biotin biosynthetic pathway may
provide new targets for herbicide design. Biotin is an
essential, water-soluble vitamin found in nearly all
living cells. Microorganisms, plants and a few fungi
make their own biotin, while animals require trace
amounts in their diets and most fungi obtain biotin
from the environment. Biotin is a covalently attached
cofactor that functions as a carbon dioxide carrier in
carboxylation, transcarboxylation and decarboxyl-
ation reactions.1,2 Acetyl-CoA carboxylase, the rate-
limiting step in lipid biosynthesis in plants, is a
biotin-dependent enzyme and a known target for
* Corres pondence to: Alan R Rendina/Wendy S Taylor, E.I. Du
Pont de Nemours and Co, Stine-Has kell Res earch Center, Newark,
DE 19714, USA
2–7 Augus t 1998.
º Pres ent addres s : Univers ity of Maryland, College Park, MD
20742, USA
E-mail: alan.r.rendina/wendy.s .taylor=us a.dupont.com
¹ Bas ed on pos ter pres entations at the 9th International Congres s
of Pes ticide Chemis try, organis ed by the International Union of
Pure and Applied Chemis try (IUPAC), and held in London, UK,
¸ Pres ent addres s : Is rael Ins titute for Biological Res earch, Nes s
Ziona 74100, Is rael
(Received 6 July 1998; revis ed vers ion received 1 October 1998;
accepted 19 October 1998)
( 1999 Society of Chemical Industry. Pestic Sci 0031-613X/99/$17.50
236

Inhibitors of dethiobiotin synthase as potential herbicides
Table 1. Subs trate activity of DAPA analogs (s tructures s hown in
Fig 3)a
Chalm.5h9 In addition, two biotin auxotrophic
mutants have been identiüed in Arabidopsis thaliana
Heynh.7h9 The bio1 auxotroph has been shown to be
defective in 7,8-diaminopelargonic acid (DAPA)
synthase8 and the bio2 auxotroph is defective in
biotin synthase.9 Both are embryo-defective mutants
which fail to grow in the absence of biotin or the
appropriate biotin precursor. These studies also indi-
cate that the enzymes of the biotin biosynthetic
pathway may be potential new targets for the dis-
covery of environmentally friendly, low use-rate,
broad-spectrum herbicides.
Compound Rel. V
(%)
K
(kM) Rel. V
/K (%)
max M
max
m
1
2
3
4
5
6
7
8
9
100
32
1.3
14
100
3
10
74
2000
1900
8600
È
0.006
0.05
216
0.033
0.0064
0.006
È
È
6.4
1470
1.5b
È
15.4
Dethiobiotin synthetase (DTBS ; E.C. 6.6.6.6)
catalyzes the penultimate step in biotin synthesis, the
formation of the ureido ring of DTB from (7R),
150
13
a Relative V
and V
/K values may be inaccurate for the
weaker s ubs trates s ince K for ATP and CO may be higher and
max
max
M
M
may not be s aturating under the as s ay conditions (eg K for CO
2
(8S)-DAPA, CO and ATP (Fig 1).10 E. coli DTBS
2
M
at 10 mM 3 is 0.7 mM, compared to 0.004 mM with DAPA as the
2
has been overexpressed11h14 and the reaction
s ubs trate).
mechanism10,13h20
and
X-ray
crystal
b Highes t level of 8 was 4 mM ; no K determined.
M
structure11,12,14,15,17,18,20 have been studied and
described in earlier papers. The goal of the work
presented here was to use the mechanistic informa-
tion to design inhibitors of the enzyme and to test
the inhibitors for herbicidal activity. In the absence
of a practical source of the plant enzyme (no plant
sequences have been reported to date, and we have
been able to detect only very small amounts of activ-
ity in plant tissues using a highly sensitive radioche-
design strategies focused on the preparation of
mimics of the substrate and of the DAPA carbamate.
Subsequent inhibitor design focused on the prep-
aration of DAPA carbamate mimics which could
become phosphorylated by the enzyme in situ to gen-
erate a potentially more tightly bound inhibitor.
Direct mimics of the phosphorylated DAPA car-
bamate and postulated tetrahedral intermediate (Fig
1) are not covered in this discussion and were delib-
erately avoided in initial design strategies due to
potentially poor uptake and translocation properties
of such highly charged molecules and to potential
problems arising from competition with high intra-
cellular concentrations of ATP. Preliminary reports
of this work have been presented elsewhere.21,22
mical assay ; Rendina AR
&
Gibson KJ,
unpublished), we used the bacterial enzyme as a
model.
At pH 7.5 substrate K values are in the micro-
M
molar range, being 0.4, 0.3 and 4 lM for ATP,
DAPA and CO , respectively. In contrast, K values
2
i
are greater for the products, ie [1mM for dethiobio-
tin and inorganic phosphate (Pi) and 14lM for
ADP.13,21 DAPA and ATP appear to bind in syn-
ergistic fashion, whereas the concentration of CO
2 EXPERIMENTAL METHODS
2
does not aþect the apparent K of DAPA or
M
Compounds 3 and 10–15 (Tables 1–3) were obtained
from Aldrich or Sigma. DAPA was prepared as
described previously.13 Synthetic methods for other
compounds are outlined at the appropriate point in
Section 3 (See Figs 2, 5, 7–12). Details and spectro-
scopic data for several of the DAPA and DAPA car-
bamate analogs including those labeled with 13C
ATP.13,21 Random addition of substrates21 was con-
ürmed by pulse-chase experiments which showed
that both preformed E*DAPA*CO 13 and
2
E*MgATP (Gibson KJ, unpublished) complexes
were committed to form products. Since the K
values for the reactants were low, initial inhibitor
M
Figure 1. Reaction and propos ed
mechanis m of DTBS.
Pestic Sci 55:236–247 (1999)
237

AR Rendina et al
HO C(CH ) CO H
2 n
2
2
Compound
n
K (low ATP)
K (high ATP)
K
Rel V
Rel V/K
(%)
i
i
M
(kM)
(kM)
(kM)
(%)
10
11
12
13
14
15
10
9
8
7
È
È
È
90
210
240
107
90
1.1
11
0.01
0.06
0.2
8
22
19
7
16
64
1.0
Table 2. Inhibition and s ubs trate
activity of uns ubs tituted diacid
analogs of DAPA carbamate
6
70
548
È
590
7300
12
0.03
0.001
5
1300
6.9
have been published previously.13,23h25 Spectro-
scopic evidence for the remaining compounds pre-
pared for this study is consistent with the structures
shown. Methods for obtaining [13C] and [31P]NMR
spectra are described elsewhere.13 DTBS-catalyzed
reactions for a typical NMR experiment used to
detect new products with alternate substrates con-
viously.15 Details of the crystallography will be
published elsewhere.
Cloning, expression, puriücation and assays of E.
coli DTBS were conducted according to published
procedures.13,14 Coupled spectrophotometric assays
for either ADP or Pi were the primary tools used for
evaluation of substrate activity and inhibition. For
substrate activity, varying amounts of each com-
pound were evaluated at pH 7.5 in solutions contain-
ing MgCl (5mM), ATP (3 mM) and NaHCO
tained HEPES buþer, (100mM; pH 7.8), MgCl
(7 mM), ATP (3 mM), phosphoenolpyruvate (2mM),
2
compound (2mM), pyruvate kinase (5 units; Sigma),
2
3
DTBS (27lM) and water containing D O (25% by
(15 mM). The bicarbonate was required for diamines
and monoamines, but not for substrate activity of
carbamate mimics where acid functionality replaced
the carbamate carboxyl group. A radiochemical assay
2
volume). X-ray crystallographic conditions used to
solve the structure of compound 38 and MnATP
with DTBS are similar to those described pre-
Table 3. Inhibition and s ubs trate activity of N7-DAPA carbamate mimics
Compound
X
n
K (low ATP)
K (high ATP)
K
Rel V
Rel V/K
(%)
i
i
M
(kM)
(kM)
(kM)
(%)
16
17
18
19
20
21
22
23
24
25
NH½
3
Keto
0
1
1
1
2
2
2
3
3
3
70
60
55
1890
3.3
È
È
2620
900
2000
160
4
80
22
3.8
255
9
0.18
0.014
0.032
83
OH
15
NH½
3
Keto
21
84
940
43
0.01
0.48
1.2
OH
50
16
34
0.6
7
NH½
3
Keto
124
76
È
È
È
È
38
È
3040
È
0.003
OH
100
770
NH½
650
19
0.04
3
Figure 2. Synthes is of
diamino acids .
238
Pestic Sci 55:236–247 (1999)

Inhibitors of dethiobiotin synthase as potential herbicides
for üxation of 14CO into acid-stable material was
pyrimidine were not active as substrates (data for
these commercially available compounds are not
shown in Fig 3 or Table 1). In all cases shown in
Table 1, except compounds 3, 7 and 8, cyclization
was observed to form the ureido ring based on the
2
used to detect very weak diamine substrates.13
Maximal velocities (V ) and V /K values are
max
max
M
reported as relative percentages of the activity
observed with the natural, chiral substrate, (7R),
(8S)-DAPA. K values were calculated from I
release of P and the formation of acid-stable pro-
i
50
i
determinations conducted at two diþerent ATP
ducts in the presence of 14CO . The partial sub-
2
levels assuming competition with DAPA and a K
strate activity of the monoamines, 3, 7 and 8,
M
M
for DAPA of 1.2lM (calc K \ I
MK /(K
depended on the presence of CO and was limited to
i
50
M
2
] [DAPA])N. The low ATP K was determined at
compounds with a six-carbon spacer between the
i
pH 7.5 in solutions containing DAPA (2lM), ATP
amine and the acid. This partial substrate activity of
the monoamines was conürmed using NMR analysis
of the enzymatic reaction of compound 3 with [12C]
(1 lM), MgCl (1 mM) and NaHCO (3mM). The
2
3
high ATP K was determined at pH 7.5 in DAPA
i
(2 lM), ATP (3 mM), MgCl (5 mM) and NaHCO
or [13C] NaHCO . The [13C] and [31P]NMR
2
3
3
(15 mM). Competition versus DAPA for several
inhibitors was conürmed by initial velocity kinetic
analysis (data not shown). The phytotoxicity of the
compounds was determined by rating the eþects
visually after germination relative to untreated con-
trols, of doses of 0, 1, 10 and 100 mg litre~1 applied
to the agar prior to germination of Arabidopsis tha-
liana seedlings.26,27
spectra showed the appearance of a new 13C reso-
nance (156.6 ppm, J \ 4.7Hz) and a new 31P reso-
nance (2.8ppm, J \ 4.9Hz) which corresponded to
the phosphorylated carbamate of monoamine 3 (Fig
4). The lack of substrate activity for 8-aminooctanoic
acid and 6-aminohexanoic acid (data not shown)
compared to 7-aminoheptanoic acid (compound 3), is
also consistent with the observations on monoamines
(see above) and with earlier publications on the
enzyme’s regioselectivity for the N7-DAPA car-
bamate.13
3
RESULTS AND DISCUSSION
3.1 Mimics of DAPA
A variety of diamino acids and monoamino acids was
prepared (Fig. 2) and assayed for substrate activity.
The structure–activity relationships of these com-
pounds (shown in Fig 3 and Table 1) provided infor-
mation that was used for the subsequent design of
inhibitors. The optimal chain length for substrate
activity (relative V /K ) for the structures in Fig 3
3.2 Inhibitors based on mimics of the DAPA
carbamate
A series of simple, commercially available com-
pounds was assayed as examples of compounds
which mimic the N7-DAPA carbamate, but did not
have a substituent in the position equivalent to the
8-amine or 9-methyl groups of DAPA (Table 2).
The optimal chain length for substrate activity was
9, the same number of atoms as between the two
acids of the N7-DAPA carbamate. The optimal
chain length for inhibition at subsaturating ATP
levels is 11, suggesting that one of the acids overlaps
the c-phosphate site of ATP and picks up extra
binding interactions. Unlike the natural substrate,
these compounds were not synergistic with ATP, but
rather were antagonistic. Observation of the enzy-
matic reaction with compound 13, azelaic acid, using
[31P]NMR showed that these diacids are partial sub-
strates for DTBS. A new 31P resonance was apparent
at [3.07 ppm (Pi had a resonance at 1.22 ppm). The
products, acyl-phosphate and MgADP, are released
by the enzyme. The acyl-phosphate products can
also be converted to Fe3`-hydroxamates in the pres-
ence of ferric chloride and hydroxylamine (data not
shown). Release of inorganic phosphate with these
compounds was not detected using a coupled assay
for Pi.
max
M
was the same as is found in the natural substrate 1;
reducing or increasing the chain length decreased the
relative V/K . The decreased substrate activity of
M
compound 9 (about 10-fold in V/K ) demonstrates
M
the importance of the 9-methyl group, consistent
with the small hydrophobic pocket surrounding this
group as seen in the X-ray crystallographic models
for DAPA bound to the active site.15 Both cis- and
trans-cyclohexanediamine were also found to be very
weak substrates (relative V/K \ 0.001%), while
M
2,3-diaminopropionic acid, 3,4-diaminobenzoic acid,
1,2-phenylenediamine and 4,5-diamino-6-hydroxy-
These straight-chain analogs were all poor inhibi-
tors of the enzyme, indicating that the substituents
on the C8 of DAPA have important binding contacts
consistent with the crystallographic studies of the
active-site interactions of DAPA and the N7-DAPA
carbamate with the enzyme.15 The methylene groups
can be replaced by double bonds in this series (eg
Figure 3. Structures of compounds reported in Table 1.
Pestic Sci 55:236–247 (1999)
239

AR Rendina et al
Figure 4. NMR s pectra of the enzymatic product produced from compound 3. Top s pectrum: 7-aminoheptanoic acid with unlabeled
bicarbonate; middle and bottom s pectra: 7-aminoheptanoic acid with [13C-]bicarbonate.
trans traumatic acid is a weak partial substrate; data
not shown), but no improved inhibition is obtained.
The binding pocket for the alkyl side-chain of DAPA
can accommodate an aromatic ring in place of two to
strate activity (data and structures not shown). These
results indicate that two acid functional groups are
needed to mimic the N7-DAPA carbamate and are
consistent with the strong H-bonding network
around C1 of DAPA observed in the crystal struc-
tures.15
A series of compounds were prepared as shown in
Fig 5 which more closely mimicked the N7-DAPA
carbamate by incorporating an extra side-chain in the
C7 position of the diacid. The substrate and inhibi-
tory activities of these compounds are summarized in
four of the methylene groups, but the K values are
M
high, the relative V/K values are very low com-
M
pared to other substrates, and the inhibition is weak
at high ATP levels (data not shown). Compounds
with n \ 7 and where one of the carboxylic acids was
replaced with an alcohol, aldehyde, ester or lactone
moiety were very weak inhibitors and had no sub-
Figure 5. Synthes is of N7-DAPA
carbamate analogs .
240
Pestic Sci 55:236–247 (1999)

Inhibitors of dethiobiotin synthase as potential herbicides
Table 4. Subs trate and inhibitory activity of N7-DAPA carbamate mimics (non-cylizable)
Compound
X
K (low ATP)
k (high ATP)
K
Rel V
Rel V/K
(%)
i
i
M
(kM)
(kM)
(kM)
(%)
26
27
28
29
30
31
32
33
34
Cyclobutyl
Methyl
3.5
14.4
10.5
9
23
1250
270
32
495
250
195
120
1100
45
3.3
29
0.13
0.08
0.07
0.05
0.05
0.01
0.004
0.001
0.8
Propyl
13
Is opropyyl
2-Propenyl
Phenyl
150
7.2
5
9
120
18
2800
49
9.4
4.8
3.5
84
2-Pyrrole
91
3-Pyridyl
70
6300
140
2420
140
CH(CH )NHC(\NH ½)NH
2
350
3
2
Table 3. (K values were difficult to determine for
these compounds due to the high substrate activity,
ie rapid release of ADP. The radiochemical assay was
used in place of the ADP-coupled assay for many of
these compounds, especially at high ATP.)
all four diastereomers of the substrate, although the
rates varied. The highest rate was observed for one
of the erythro isomers; the lowest rate was for the
remaining erythro isomer. This indicates that DTBS
prefers the stereochemistry of the natural substrate
and does not exhibit a strong preference at the 8-
position, but does show a preference at the 7-
position. The 31P and 13C experiments for
13C-labeled compound 17 where X \ keto and n \ 1
show that partial substrate activity occurs to release
an acyl-phosphate (31P resonance at [2.94 ppm,
J \ 8.6 Hz and 13C resonance at 173.2 ppm,
J \ 8.5Hz) and MgADP, but not Pi; a cyclized
product is not formed in this case. Compound 16
also does not appear to cyclize, as ADP is released
but not Pi.
At the optimal chain length the inhibitors shown
in Table 3 were better inhibitors than the corre-
sponding straight-chain analogs in Table 2, but the
high substrate activity reduced their efficacy since
the cyclized products are readily released from the
enzyme. The main advantage of the compounds with
substituents at the 7-position (ie the X group shown
in Table 3) was that they began to show synergistic
behavior with ATP, especially when n \ 0 or 1.
Some additional analogs of the compounds shown
in Table 3 were prepared where X was a non-
nucleophilic functionality. These analogs, prepared
using known methods, could not undergo an
enzyme-catalyzed cyclization. It was hoped that
these compounds would remain tightly bound to the
enzyme after phosphorylation in the active site.
These compounds, shown in Table 4, were only
partial substrates for DTBS since NMR con-
ürmed that they become phosphorylated, but do not
cyclize or release Pi. For example, 13C-labeled com-
pound 26 in the presence of DTBS and ATP pro-
i
The optimal substrate activity was observed for
compound 19 when n \ 1 and X \ NH `. Longer
3
derivatives (n \ 2 and 3) showed weaker substrate
activity and weaker inhibition at both high and low
ATP. NMR experiments with 13C-labeled com-
pound 19 showed that this compound rapidly
cyclized in the enzymatic reaction to form the corre-
sponding lactam (Fig 6). The new peaks observed in
the experiment corresponded to the spectra obtained
from the synthetically prepared lactams.13 No new
31P signals were observed nor was Pi detected by
coupled assay. Interestingly, [31P] and [13C]NMR
experiments with unlabeled and 13C-labeled 18, and
concomitant release of ADP and Pi, showed that
compound also to give a cyclized product (the
lactone shown in Fig 6). The cyclization of either
13C-labeled compound shown in Fig 6 proceeded on
duced
a
new 31P resonance at [3.17 ppm
(J \ 8.9Hz) and a new 13C resonance at 175.7 ppm
(J \ 8.9Hz) corresponding to the acyl-phosphate.
Figure 6. Subs trate activity of two carboxylic acid mimics of the
N7 DAPA carbamate.
Pestic Sci 55:236–247 (1999)
241

AR Rendina et al
Other functional groups such as aldehydes, acids,
diols, hydroxamates and oximes could also replace
the N8 amine and the C8 methyl group of DAPA but
were weakly bound (data and structures not shown).
As observed with other series of carbamate mimics,
synergy with ATP (in these cases stronger inhibition
at high ATP) appears to depend on the presence of a
cation or H-bond donor in the 8-position (see the
pyrrole of compound 32 and the guanidinium group
of compound 34). Unfortunately, despite the synergy
with ATP, these compounds are still relatively weak
inhibitors and the phosphorylated products are
rapidly released. When the H-bond donor group is
an OH or an amino group both synergy and potency
occur, but the cyclized products are also rapidly rel-
eased resulting in substrate activity rather than
potent enzyme-activated inhibition. None of the
compounds discussed in this section was herbicidal
below 10mg litre~1 in the agar-based assay with A.
thaliana, nor did they aþect germination.
the N7-DAPA carbamate and varied the chain
length, charges, and the side chain. Since the N–P
bond is relatively weak, cyclization of amino-
substituted analogs should not occur as readily as in
the carboxylic acid series. The syntheses of the
amino, keto and hydroxyl phosphinate and phos-
phonate analogs are shown in Fig 7. The phosphinic
acid, 36, and phosphonates 37 and 39 were mixtures
of two pairs of diastereomers, whereas the phos-
phinic acid 38 was a mixture of three pairs of
diasteriomers. One of the diastereomers of com-
pound 38 was separated from the remaining mixture
of diastereomers. The data for this single isomer are
presented in Table 5, but the activity of the mixture
of diastereomers was much the same as that of the
single diastereomer. The stereochemistry of the
single diastereomer was not determined, although it
could be deduced from the crystal structure (see Fig
14).
The syntheses of the corresponding amino,
hydroxy and alkyl substituted phosphates and phos-
phonates are shown in Figs 8–10. Chiral (L)-alanine
was used in the syntheses shown in Fig 8. Racemic
lactic acid was used in the synthesis in Fig 9.
(Compound 43 was also prepared from (S)-lactic
acid. The diastereomeric mixture obtained from (S)-
lactic acid was less potent than the mixture of four
diasteriomers obtained from racemic lactic acid.)
Compounds 40-45 were mixtures of diastereomers.
Compounds 40–43 tended to be slightly labile under
acidic conditions and were stored as the sodium salts.
Synthetic routes to extended and shortened versions
of these inhibitors are shown in Figs 11 and 12.
Compounds 47, 48, 50, 53 and 54 were mixtures of
two diastereomers. Compound 51 was a mixture of
three diastereomers.
3.3 Phosphorus-based mimics of DAPA carbamate
An approach to inhibitor design that has been
used successfully for glutamine synthetase,28h32
D-alanine : D-alanine
ligase,33,34
glutathione
synthetase35 and c-glutamylcysteine synthetase36 is
to prepare a substrate mimic which can be phos-
phorylated by the enzyme in situ to generate a more
tightly bound inhibitor. The carboxylic acid sub-
strate mimics discussed previously did become phos-
phorylated but did not generate a more tightly bound
inhibitor. Analogs with a nucleophilic group on
C8 underwent a cyclization reaction. Analogs which
could not cyclize were simply released from the
enzyme. Another series of inhibitors were prepared
which were more closely modeled on phos-
phinothricin, the known inhibitor of glutamine syn-
thetase, which becomes more tightly bound after it is
phosphorylated in situ. This series used a phos-
phorus acid as a mimic for the carbamate group on
A number of interesting comparisons can be made
with these phosphorus-based inhibitors (see Table
5). The carbamate binding site in DTBS is in close
proximity to the binding site for the c-phosphate of
Figure 7. Synthes is of amino, keto
and hydroxyl phos phinic and
phos phonic analogs .
242
Pestic Sci 55:236–247 (1999)

Inhibitors of dethiobiotin synthase as potential herbicides
Compound
K (low ATP)
K (high ATP)
Rel V
Rel V/K
(%)
i
i
(kM)
(kM)
(%)
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
31
380
1.8
121
10.9
26
È
8500
25
0.01
0.002
0.03
È
3.6
2.9
0.8
28.7
0.4
È
45
55
2040
175
0.008
È
È
È
È
È
È
È
77
24
56
110
0.5
0.8
25
1400
1300
6850
1300
3500
1900
È
1
È
1
5.4
1.7
23
1
1.2
0.3
1
\0.001
\0.001
\0.001
È
1.6
7.6
71
296
0.4
11.2
È
190
59
1700
2500
1320
2400
0.002
È
0.001
È
36
2.7
Table 5. In-vitro comparis on of
3.1
1.0
phos phorus -bas ed inhibitors
ATP. Most of the phosphorus-based inhibitors did
where ATP normally binds. Only the two analogs 38
and 41 bound synergistically with ATP. These com-
pounds both have phosphorus monoanions and both
have an amino group on the side chain. In the pre-
not bind synergistically with ATP (K at low ATP
i
\K at high ATP), suggesting that the phosphorus-
i
containing group was often overlapping the position
Figure 8. Synthes is of
amino-s ubs tituted phos phonate
and phos phate analogs .
Figure 9. Synthes is of the
hydroxyl-s ubs tituted
phos phate 43.
Pestic Sci 55:236–247 (1999)
243

AR Rendina et al
Figure 10. Synthes is of
alkyl-s ubs tituted phos phates and
phos phonates .
vious series of analogs, the amino group was also
critical for synergistic binding. A comparison of 44
(amine replaced with a methyl group) with 41 shows
that 44 is [50 ] less tightly bound at high ATP. A
comparison of 41 with 40 and of 38 with 37 shows
that the compounds with phosphorus dianions do not
bind synergistically with ATP in contrast to the
compounds with phosphorus monoanions.
The analogs with an oxygen in the position that
corresponds to the N7 position of DAPA were pre-
pared in an attempt to pick up additional binding
interactions with Ser-41 (see Fig 14). (Analogs with
N–P bonds were not prepared due to their lability.)
Mutagenesis studies later showed that Ser-41 has
very little eþect on the binding or turnover of
DAPA, despite being within H-bonding distance of
N7 of DAPA.14 Analysis of the phosphorus-
containing inhibitors shows that an increase in
potency can occur, but not always, when there is an
oxygen in this position. In the case of the phos-
phonate 39 and the phosphate 43, the phosphate
binds more tightly. However, in the case of the phos-
phonate 37 and the phosphate 40, the phosphate
binds less tightly.
Although 38 and 41 bound synergistically with
ATP, they had very poor substrate activity, indicat-
ing that the c-phosphate was not being transferred
efficiently to the inhibitor. The lack of substrate
activity could be a desirable attribute if it were
accompanied by tighter binding over time and slow
release of the ADP and phosphorylated inhibitor
products. However, these compounds exhibited no
time-dependent increase in potency and were rela-
tively weak inhibitors (lM). Shortened analogs,
50–54 (Fig 12), were prepared in an attempt to
decrease the repulsion between the inhibitor and
ATP and thereby increase the phosphorylation of the
inhibitor. Unfortunately, the shortened versions did
Figure 11. Synthes is of extended
phos phonate analogs .
244
Pestic Sci 55:236–247 (1999)

Inhibitors of dethiobiotin synthase as potential herbicides
Figure 12. Synthes is of
s hortened analogs .
not bind synergistically with ATP; in fact antago-
nism was observed. One possible explanation is that
the position of the C8 substituent may dictate where
the phosphorus-containing moiety is situated in the
active site. For comparison, three longer analogs
were prepared (47–49) (Fig. 11). Not unexpectedly
posed on the crystal structure of DTBS-ADP-phos-
phorylated 38. The positions of the c-phosphate of
ACP and of the transferred phosphate are very close
to each other and very little movement is required to
transfer the phosphate to 38. Very little change is
observed in the orientation of the amino acid side-
chains near the active site. DAPA carbamate and the
phosphorylated inhibitor are stabilized by inter-
actions with Lys-37 and Ser-41 at the amino end,
and by an interaction with Tyr-187 and a network of
H-bonds to tightly bound water molecules at the C-1
end. The transferred phosphorus also has inter-
actions with Thr-11, Lys-15 and two metal ions
similar to the interactions observed for the inert
ATP analog, ACP, and to the phosphorylated sub-
strate intermediate.15,20 The 8-amino group in
DAPA and the inhibitors is within hydrogen
bonding distance of the c-phosphate of ATP. No
other major interactions with the amine are
observed. The amine is very close to the surface of
the enzyme and exposed to water. The exposed
nature of the active site may account for the lack of
potent binding of the phosphorylated inhibitors.
Another reason for either the lack of potency or
time-dependent behavior is that, unlike the other
four enzymes where this inhibitor strategy was suc-
cessful, the nucleophilic amino group that attacks the
acyl-phosphate (or carbamyl-phosphate) interme-
diate is intramolecular for DTBS versus intermolec-
ular for the other enzymes. The intramolecular
reaction catalyzed by DTBS (expected to be more
facile chemically) does not require the amine-
they showed 150- to [1000-fold increases in K at
i
the higher level of ATP. In fact, these longer deriv-
atives could be considered to be mimics of the phos-
phorylated carbamate.
Although compound 38 was a very poor substrate
for the enzyme, when the compound was put with
the enzyme under crystallization conditions (mM
levels of enzyme and ligands) in the presence of
ATP, transfer of phosphate occurred (Fig 13) and
the crystal structure was found to contain the phos-
phorylated product shown in Fig 14. Not sur-
prisingly, the crystal structure of phosphorylated 38
is remarkably similar to that of the actual phos-
phorylated substrate intermediate recently reported
using kinetic crystallography techniques.20 In Fig 14
the ATP analog, ACP [adenylyl (b,c-methylene)
diphosphonate], was extracted from the crystal
structure of DTBS-ACP-19 (thin line) and superim-
Figure 13. Reaction of 38 with ATP under X-ray crys tallization
conditions .
Pestic Sci 55:236–247 (1999)
245

AR Rendina et al
Figure 14. X-ray s tructure of phos phorylated 38 s uperimpos ed with the E-DAPA-CO ÉACP s tructure (thin line). Thr-11 and the two Mn2½
2
atoms found in the s tructure of phos phorylated 38 near the trans ferred phos phate are not s hown for purpos es of clarity.
containing second substrate to be bound tightly and
positioned properly, as is required and observed for
the other four enzymes.
remains an open question. Future plans include con-
tinuation of the search for the plant equivalents to
the bacterial enzymes, mostly through the eþorts of
public and private sector genomic sequencing pro-
jects. Once the plant genes and enzymes are
obtained, both screening and design approaches to
inhibitor discovery can be pursued more conüdently ;
eg knowledge acquired from the bacterial model can
be used to design the next generation of inhibitors
that exploit binding pockets unique to the plant
enzyme.
4
CONCLUSIONS
Amino acids and diacids were good mimics of the
DAPA substrate and of the DAPA carbamate inter-
mediate. Low micromolar levels of inhibition were
observed. Dianionic phosphorus-based inhibitors
which mimicked the DAPA carbamate were gener-
ally less potent than monoanionic phosphorus-based
inhibitors. The amine in the C-8 position was critical
for synergistic binding with ATP in both the
phosphorus-containing inhibitors and the carboxylic
inhibitors.
ACKNOWLEDGEMENTS
We thank colleagues at Dupont who have partici-
pated in the dethiobiotin synthetase project, espe-
cially Dennis Flint for getting us all started and for
many helpful discussions, Vern Wittenbach and Pete
Teaney for all the phytotoxicity studies with A. tha-
liana, and Forrest Chumley, Russ Bellina, Mark
Nelson and John Pierce for constructive criticism
and project support.
The design strategy used for glutamine synthetase
(and other enzymes) inhibitors was not successful in
this case, perhaps due to the solvent accessibility of
the DTBS active site. The carboxylic acid inhibitors
which became phosphorylated in situ did not remain
bound to the active site. The phosphorus-containing
inhibitors were very poor partial substrates for
DTBS and the amount of phosphorylated inhibitor
that was generated was very slight. Unfortunately,
none of the inhibitors discussed was a time-
dependent inhibitor, nor did any show signiücant
herbicidal activity against A. thaliana below 10mg
litre~1.
REFERENCES
1 Dakshinamurti K and Bhagavan NH (eds), Biotin. Ann NY
Acad Sci, 447:1–441 (1985).
2 Knowles JR, The mechanism of biotin-dependent enzymes.
Annu Rev Biochem 58:195–221 (1989).
Whether DTBS and the other enzymes in biotin
biosynthesis are good targets for herbicide discovery
3 Rendina AR, Craig-Kennard AC, Beaudoin JD and Breen
MK, Inhibition of acetyl-coenzyme A carboxylase by two
246
Pestic Sci 55:236–247 (1999)

Inhibitors of dethiobiotin synthase as potential herbicides
classes of grass-selective herbicides. J Agric Food Chem
38:1282–1287 (1990) and references therein.
4 Eisenberg M, Biosynthesis of biotin and lipoic acid, in Escheri-
chia coli and Salmonella typhimurium – Cellular and Molecu-
lar Biology, ed by Neidhardt F, American Society for
Microbiology, New York, NY, pp 544–550 (1987).
20 Kack H, Gibson KJ, Lindqvist Y and Schneider G, Snapshot
of a phosphorylated substrate intermediate by kinetic crys-
tallography. Proc Natl Acad Sci USA 95:5495–5500 (1998).
21 Rendina A, Taylor W, Rayner D, Lockett B, Gibson K,
Nudelman A, Marsilii E, Chi H and Yang G, Kinetic
mechanism and alternate substrates for Escherichia coli
dethiobiotin synthetase (DTBS). The FASEB J 11:A1319
(1997).
22 Taylor W, Rendina A, Rayner D, Lockett B, Kranis K, Mar-
silii E, Chi H, Wawrzak Z, Calabrese J, Huang W, Jia J,
Schneider G and Lindqvist Y, Inhibition of Escherichia coli
dethiobiotin synthetase by stable phosphorus analogs of the
carbamate intermediate. The FASEB J 11:A1319 (1997).
23 Maracovici D, Synthesis of biotin, biotin vitamers and analogs.
PhD Thesis, Bar Ilan University, Ramat Gan, Israel (1995).
24 Nudelman A, Biotin, Approaches to total synthesis and inhibi-
tors of biotin biosynthesis. PhD Thesis, Bar Ilan University,
Ramat Gan, Israel (1997).
5 Baldet P, Gerbling H, Axiotis S and Douce R, Biotin biosyn-
thesis in plants: identiücation of intermediates. Eur
Biochem 217:479–485 (1993).
J
6 Patton DA, Johnson M and Ward ER, Biotin synthase from
Arabidopsis thaliana. Plant Physiol 112:371–378 (1996).
7 Shellhammer J and Meinke D, Arrested embryos from the bio1
auxotroph of Arabidopsis thaliana contain reduced levels of
biotin. Plant Physiol 110:1021–1028 (1990).
8 Patton DA, Volrath S and Ward ER, Complementation of the
bio1 Arabidopsis biotin auxotroph with a bacterial biotin bio-
synthetic gene. Mol Gen Genet 251:261–266 (1996).
9 Patton DA, Schetter AL, Franzmann LH, Nelson K, Ward
ER and Meinke DW, An embryo-defective mutant of Ara-
bidopsis disrupted in the ünal step of biotin synthesis. Plant
Physiol 116:935–946 (1998).
25 Nudelman Ay, Nudelman Ab, Marcovici-Mizrahi D and Flint
D, The synthesis of the vitamers of biotin. Bioorganic Chem
(in press).
10 Krell K and Eisenberg MA, The puriücation and properties of
dethiobiotin synthetase. J Biol Chem 245:6558–6566 (1970).
11 Huang W, Lindqvist Y, Schneider G, Gibson KJ, Flint D and
Lorimer G, Crystal structure of an ATP-dependent carbox-
ylase, dethiobiotin synthetase, at 1.65 A resolution. Structure
2:407–414 (1994).
12 Alexeev D, Bury SM, Boys CWG, Turner MA, Sawyer L,
Ramsey AJ, Baxter HC and Baxter RL, Sequence and crys-
tallization of Escherichia coli dethiobiotin synthetase, the
26 Somerville, CR and Ogren WL, Isolation of photorespiration
mutants in Arabidopsis thaliana, in Methods in Chloroplast
Molecular Biology, ed by Edelman M, Hallick RB and Chua
N-H, Elsevier Biomedical, Amsterdam. pp 129–138 (1982).
27 Henderson SA, O’Conner J, Rendina AR, Savage GP and
Simpson GW, The synthesis and biological activity of ‘crip-
pled biotin’. Aust J Chem 48:1907–1916 (1995).
28 Lea PJ and Ridley SM, Glutamine synthetase and its inhibi-
tion, in Herbicide and Plant Metabolism, ed. by Dodge AD,
Soc Exp Biol Seminar Series 38, Cambridge University
Press, Cambridge. pp 137–170 (1989) and references therein.
29 Johnson CR, Boettcher BR, Cherpeck RE and Dolson MG,
Design and synthesis of potent inhibitors of glutamine syn-
thetase: 1. Cyclic analogs of phosphinothricin. Bioorg Chem
18:154–159 (1990).
penultimate enzyme of biotin biosynthesis.
J Mol Biol
235:774–776 (1994).
13 Gibson KJ, Lorimer GH, Rendina AR, Taylor WS, Cohen G,
Gatenby AA, Payne WG, Roe DC, Lockett BA, Nudelman
A, Marcovici D, Nachum A, Wexler BA, Marsilii EL,
Turner IM, Howe LD, Kalbach CE and Chi H, Dethiobio-
tin synthetase: the carbonylation of 7,8-diaminononanoic
acid proceeds regiospeciücally via the N7-carbamate. Bio-
chemistry 34:10976–10984 (1995).
30 Logusch EW, Walker DM, McDonald JF and Franz JE, Inhi-
bition of plant glutamine synthetases by substituted phos-
phinothricins. Plant Physiol 95:1057–1062 (1991).
14 Yang G, Sandalova T, Lohman K, Lindqvist Y and Rendina
AR, Active site mutants of Escherichia coli dethiobiotin syn-
thetase: eþects of mutations on enzyme catalytic and struc-
tural properties. Biochemistry 36:4751–4760 (1997).
31 Abell LM and Villafranca JJ, Investigation of the mechanism
of phosphinothricin inactivation of Escherichia coli glutamine
synthetase using rapid quench kinetic techniques. Biochem-
istry 30:6135–6141 (1991).
15 Huang W, Jia J, Gibson KJ, Taylor WS, Rendina AR, Schnei-
der G and Lindqvist Y, Mechanism of an ATP-dependent
carboxylase, dethiobiotin synthetase, based on crystallo-
graphic studies of complexes with substrates and a reaction
intermediate. Biochemistry 34:10985–10995 (1995).
16 Baxter RL and Baxter HC, Mechanism of Escherichia coli
dethiobiotin synthetase – closure of the ureido ring of
dethiobiotin involves formation of a carbamic phosphate
mixed anhydride. J Chem Soc Chem. Commun 759–760
(1994).
17 Alexeev D, Baxter RL and Sawyer L, Mechanistic implica-
tions and family relationships from the structure of dethio-
biotin synthetase. Structure 15:1061–1072 (1994).
18 Alexeev D, Baxter RL, Smekal O and Sawyer L, Substrate
binding and carboxylation by dethiobiotin synthetase-a
kinetic and X-ray study. Structure 3:1207–1215 (1995).
19 Gibson KJ, Isolation and chemistry of the mixed anhydride
intermediate in the reaction catalyzed by dethiobiotin syn-
thetase. Biochemistry 36:8474–8478 (1997).
32 Liaw S-H and Eisenberg D, Structural model for the reaction
mechanism of glutamine synthetase, based on üve crystal
structures of enzyme-substrate complexes. Biochemistry
33:675–681 (1994).
33 Duncan K and Walsh CT, ATP-dependent inactivation and
slow binding inhibition of Salmonella typhimurium D-
alanine : D-alanine ligase (ADP) by (aminoalkyl)phosphinate
and aminophosphonate analogues of D-alanine. Biochemistry
27:3709–3714 (1988).
34 Fan C, Moews PC, Walsh CT and Knox JR, Vancomycin
resistance: structure of D-alanine : D-alanine ligase at 2.3 A
resolution. Science 266:439–443 (1994).
35 Hiratake J, Kato H and Oda J, Mechanism-based inactivation
of glutathione synthetase by phosphinic acid transition-state
analogue. J Am Chem Soc 116:12059–12060 (1994).
36 Katoh M, Hiratake J, Kato H and Oda J, Mechanism-based
inactivation of E. coli c-glutamylcysteine synthetase by phos-
phinic acid- and sulfoximine-based transition-state ana-
logues. Bioorg Med Chem Lett 6:1437–1442 (1996).
Pestic Sci 55:236–247 (1999)
247