Design, synthesis, and biochemical evaluation of 1,5,6,7-tetrahydro-6,7- dioxo-9-D-ribitylaminolumazines bearing alkyl phosphate substituents as inhibitors of lumazine synthase and riboflavin synthase

Article abstract of DOI:10.1021/jo051332v

The last two steps in the biosynthesis of riboflavin, an essential metabolite that is involved in electron transport, are catalyzed by lumazine synthase and riboflavin synthase. To obtain structural probes and inhibitors of these two enzymes, two ribityll

Full text of DOI:10.1021/jo051332v

Design, Synthesis, and Biochemical Evaluation of
1,5,6,7-Tetrahydro-6,7-dioxo-9-D-Ribitylaminolumazines Bearing
Alkyl Phosphate Substituents as Inhibitors of Lumazine Synthase
and Riboflavin Synthase
Mark Cushman,*,† Guangyi Jin,† Boris Illarionov,‡ Markus Fischer,‡ Rudolf Ladenstein,§ and
Adelbert Bacher‡
Department of Medicinal Chemistry and Molecular Pharmacology and the Purdue Cancer Center, School
of Pharmacy and Pharmaceutical Sciences, Purdue University, West Lafayette, Indiana 47907, Lehrstuhl
fu¨r Organische Chemie und Biochemie, Technische Universita¨t Mu¨nchen, D-85747 Garching, Germany,
and NOVUM Centre for Structural Biochemistry, Karolinska Institute, S-14157 Huddinge, Sweden
cushman@pharmacy.purdue.edu
Received June 28, 2005
The last two steps in the biosynthesis of riboflavin, an essential metabolite that is involved in
electron transport, are catalyzed by lumazine synthase and riboflavin synthase. To obtain structural
probes and inhibitors of these two enzymes, two ribityllumazinediones bearing alkyl phosphate
substituents were synthesized. The synthesis involved the generation of the ribityl side chain, the
phosphate side chain, and the lumazine system in protected form, followed by the simultaneous
removal of three different types of protecting groups. The products were designed as intermediate
analogue inhibitors of lumazine synthase that would bind to its phosphate-binding site as well as
its lumazine binding site. Both compounds were found to be effective inhibitors of Bacillus subtilis
lumazine synthase as well as Escherichia coli riboflavin synthase. Molecular modeling of the binding
of one of the two compounds provided a structural explanation for how these compounds are able
to effectively inhibit both enzymes. In phosphate-free buffer, the phosphate moieties of the inhibitors
were found to contribute positively to their binding to Mycobacterium tuberculosis lumazine
synthase, resulting in very potent inhibitors with Ki values in the low nanomolar range. The
additional carbonyl in the dioxolumazine system versus the purinetrione system was found to make
a positive contribution to its binding to E. coli riboflavin synthase.
Introduction
mazine synthase substrate 1. In a beautiful example of
natural recycling, the product 1 of the riboflavin syn-
thase-catalyzed reaction is then utilized by lumazine
synthase. The overall stoichiometry involves the con-
sumption of one molecule of the pyrimidinedione 1 and
two molecules of the organophosphate 2 to form one
molecule of riboflavin (4).1-5
The last two steps in the biosynthesis of riboflavin
(Scheme 1) involve the lumazine synthase-catalyzed
reaction of the four-carbon phosphate 2 with the ribity-
laminopyrimidinedione 1 to form 6,7-dimethyl-8-ribityl-
lumazine (3), followed by the riboflavin synthase-
catalyzed dismutation of two molecules of 3 to form one
molecule of riboflavin (4) and one molecule of the lu-
(1) Plaut, G. W. E.; Smith, C. M.; Alworth, W. L. Annu. Rev.
Biochem. 1974, 43, 899-922.
(2) Plaut, G. W. E. In Comprehensive Biochemistry; Florkin, M.,
Stotz, E. H., Eds.; Elsevier: Amsterdam, 1971; Vol. 21, pp 11-45.
(3) Beach, R. L.; Plaut, G. W. E. J. Am. Chem. Soc. 1970, 92, 2913-
2916.
* Corresponding author: phone 765-494-1465, fax 765-494-6790.
† Purdue University.
‡ Technische Universita¨t Mu¨nchen.
§ Karolinska Institute.
10.1021/jo051332v CCC: $30.25 © 2005 American Chemical Society
Published on Web 09/02/2005
8162
J. Org. Chem. 2005, 70, 8162-8170

Lumazine Synthase and Riboflavin Synthase Inhibitors
SCHEME 1
SCHEME 2
The mechanism proposed for the lumazine synthase-
catalyzed reaction has gone through several iterations,
the most recent of which is outlined in Scheme 2.6 The
location of the phosphate moiety in the enzyme complex
involving the hypothetical intermediate 5, relative to the
remainder of the molecule, is roughly approximated as
shown in the structure of the initial carbinolamine
intermediate 5 as displayed in Scheme 2, as evidenced
by the crystal structure of the complex formed between
Saccharomyces cerevisiae lumazine synthase and the
intermediate analogue 117 (Figure 1).8 Rotation of the
whole phosphate-containing side chain toward the ribi-
tylamino moiety would then generate conformer 6, which
could eliminate water to form the cis Schiff base 7.
Phosphate elimination would lead to the enol 8, followed
by tautomerization to the ketone 9. Nucleophilic attack
of the ribitylamino group on the ketone would result in
the carbinolamine 10, which could eliminate water to
form the final product 3.
view, the exact sequence of the events required to form
the final product has not been rigorously established. For
example, phosphate elimination could occur after Schiff
base formation and before the conformational reorgani-
zation of the side chain to favor formation of the six-
membered ring.9
When determined in the presence of fixed substrate 1
concentration and variable substrate 2 concentration, the
(4) Bacher, A.; Eberhardt, S.; Richter, G. In Escherichia coli and
Salmonella: Cellular and Molecular Biology; 2nd ed.; Neidhardt, F.
C., Ed.; ASM Press: Washington, D. C., 1996; p 657-664.
(5) Bacher, A.; Fischer, M.; Kis, K.; Kugelbrey, K.; Mo¨rtl, S.;
Scheuring, J.; Weinkauf, S.; Eberhardt, S.; Schmidt-Ba¨se, K.; Huber,
R.; Ritsert, K.; Cushman, M.; Ladenstein, R. Biochem. Soc. Trans.
1996, 24, 89-94.
(6) Zhang, X.; Meining, W.; Cushman, M.; Haase, I.; Fischer, M.;
Bacher, A.; Ladenstein, R. J. Mol. Biol. 2003, 328, 167-182.
(7) Cushman, M.; Mihalic, J. T.; Kis, K.; Bacher, A. J. Org. Chem.
1999, 64, 3838-3845.
(8) Meining, W.; Mo¨rtl, S.; Fischer, M.; Cushman, M.; Bacher, A.;
Ladenstein, R. J. Mol. Biol. 2000, 299, 181-197.
(9) Volk, R.; Bacher, A. J. Am. Chem. Soc. 1988, 110, 3651-3653.
Although the pathway outlined in Scheme 2 seems
reasonable from a structural and mechanistic point of
J. Org. Chem, Vol. 70, No. 20, 2005 8163

Cushman et al.
FIGURE 1. Crystal structure of phosphonate 11 bound in the active site of S. cerevisiae lumazine synthase.8 The figure is
programmed for walleyed viewing.
phosphonate 11 is a moderately active inhibitor of
Bacillus subtilis lumazine synthase (mixed inhibition, Ki
180 µM, Kis 350 µM).7 When tested under the same
conditions, the purinetrione 12 proved to be more potent,
displaying mixed inhibition with a Ki of 46 µM and a Kis
of 250 µM.10 The combination of the purinetrione ring
system with a C-5 phosphate side chain to form 13
resulted in a less potent inhibitor of B. subtilis lumazine
synthase (Ki 852 µM, Kis 817 µM) when tested in the
presence of variable substrate 2 concentration.11 How-
ever, the substituted purinetrione 13 unexpectedly proved
to be a very potent competitive inhibitor of Mycobacte-
rium tuberculosis lumazine synthase when tested in Tris
buffer (phosphate-free) with a Ki of 4.7 nM!11 The
selectivity of this compound and its phosphate homo-
logues for the M. tuberculosis enzyme as opposed to the
B. subtilis enzyme is truly remarkable. The alkylphos-
phonate side chain does in fact contribute positively to
inhibition of M. tuberculosis lumazine synthase, since the
purinetrione system 12 itself is a Ki 9.1 µM inhibitor of
the enzyme.11 The alkylphosphonate derivative 13 is also
a competitive inhibitor of Escherichia coli riboflavin
synthase with a Ki of 2.44 µM.11
hypothesis that the attachment of alkyl phosphate side
chains on N-5 of the 6,7-dioxo-8-ribityllumazine (14)
system would likely result in promising inhibitors of E.
coli riboflavin synthase as well as M. tuberculosis lu-
mazine synthase. The present paper describes the syn-
thesis and biological testing of the suggested five-carbon
and six-carbon alkyl phosphate derivatives of 14.
The rationale for the potential medical use of riboflavin
synthase inhibitors stems from the fact that certain
Gram-negative pathogenic bacteria and yeasts have been
shown to lack an efficient riboflavin uptake system and
are therefore absolutely dependent on endogenous ribo-
flavin biosynthesis.14-17 In contrast, humans lack ribo-
flavin biosynthesis enzymes and obtain this essential
nutrient entirely from dietary sources. Therefore, inhibi-
tors of riboflavin biosynthesis can be expected to display
selective cytotoxicity for pathogenic microorganisms as
opposed to human cells.
Results and Discussion
The linker from the phosphate to the pyrimidine ring
in the proposed reaction intermediate 5 is four atoms
long, and the results from previous studies on the
purinetrione derivative 13 and its homologues revealed
that the best linkers between phosphate and purinetrione
ring are from three to five atoms long for inhibition of
M. tuberculosis lumazine synthase. It was therefore
logical to choose the alkyl phosphate derivatives of 14
with polymethylene chains containing four and five
carbon atoms.11,18
These results suggest consideration of the 6,7-dihydro-
6,7-dioxo-8-ribityllumazine system 14, which was previ-
ously shown to be an exceedingly potent inhibitor of
baker’s yeast riboflavin synthase (Ki 25 nM)12 and Ashbya
gossypii riboflavin synthase (Ki 9 nM).13 This led to the
(14) Wang, A. I Chuan Hsueh Pao 1992, 19, 362-368.
(15) Oltmanns, O.; Lingens, F. Z. Naturforsch., B: Anorg. Chem.,
Org. Chem., Biochem., Biophys., Biol. 1967, 22b, 751-754.
(16) Logvinenko, E. M.; Shavlovsky, G. M. Mikrobiologiya 1967, 41,
978-979.
(10) Cushman, M.; Yang, D.; Kis, K.; Bacher, A. J. Org. Chem. 2001,
66, 8320-8327.
(11) Cushman, M.; Sambaiah, T.; Jin, G.; Illarionov, B.; Fischer, M.;
Bacher, A. J. Org. Chem. 2004, 69, 601-612.
(12) Al-Hassan, S. S.; Kulick, R. J.; Livingston, D. B.; Suckling, C.
J.; Wood, H. C. S.; Wrigglesworth, R.; Ferone, R. J. Chem. Soc., Perkin
Trans. 1 1980, 2645-2656.
(17) Neuberger, G.; Bacher, A. Biochem. Biophys. Res. Commun.
1985, 127, 175-181.
(18) Morgunova, K.; Meining, W.; Illarionov, B.; Haase, I.; Jin, G.;
Bacher, A.; Cushman, M.; Fischer, M.; Ladenstein, R. Biochemistry
2005, 44, 2746-2758.
(13) Winestock, C. H.; Aogaichi, T.; Plaut, G. W. E. J. Biol. Chem.
1963, 238, 2866-2874.
8164 J. Org. Chem., Vol. 70, No. 20, 2005

Lumazine Synthase and Riboflavin Synthase Inhibitors
SCHEME 3a
TABLE 1. Inhibition Constants versus B. subtilis
Lumazine Synthase, M. tuberculosis Lumazine Synthase,
and E. coli Riboflavin Synthase
riboflavin
synthase
lumazine synthase
parameter
B. subtilisa M. tuberculosisb
E. colic
Compound 14
inhibition mechanism competitive competitive
competitive
2.7 ( 0.2
Ksd (µM)
kcate (min-1
Kif (µM)
6.3 ( 0.5
2.6 ( 0.1
7.8 ( 0.5
250 ( 43
1.1 ( 0.1
1.4 ( 0.1
)
15.0 ( 0.3
0.0062 ( 0.0005
Compound 21
inhibition mechanism mixed
competitive
240 ( 29
mixed
Ks (µM)
7.1 ( 1.0
2.5 ( 0.3
15.0 ( 0.4
9.7 ( 4.1
21 ( 5
kcat (min-1
Ki (µM)
)
2.1 ( 0.1
150 ( 43
400 ( 130
1.2 ( 0.1
0.036 ( 0.004
Kisg (µM)
Compound 22
inhibition mechanism competitive competitive
competitive
2.9 ( 0.4
14 ( 1
Ks (µM)
5.8 ( 0.4
2.7 ( 0.1
27 ( 6
210 ( 28
0.99 ( 0.07
0.012 ( 0.004
kcat (min-1
Ki (µM)
)
0.14 ( 0.02
Compound 23
inhibition mechanism competitive competitive
competitive
2.1 ( 0.2
Ks (µM)
3.2 ( 0.4
3.1 ( 0.1
170 ( 26
63 ( 6
kcat (min-1
Ki (µM)
)
1.4 ( 0.1
17.0 ( 0.4
0.0041 ( 0.0023 330 ( 83
Compound 24
inhibition mechanism mixed
competitive
63 ( 5
competitive
2.0 ( 0.2
Ks (µM)
3.8 ( 0.4
kcat (min-1
Ki (µM)
)
3.15 ( 0.10 1.40 ( 0.04
17.0 ( 0.4
270 ( 85
0.0047 ( 0.0019 2.4 ( 0.2
Kis (µM)
650 ( 160
a Recombinant lumazine synthase from B. subtilis. b Recombi-
nant lumazine synthase from M. tuberculosis. c Recombinant
riboflavin synthase from E. coli. d Ks is the substrate dissociation
e
constant for the equilibrium E + S a ES. kcat is the rate constant
for the process ES f E + P. f Ki is the inhibitor dissociation
a Reagents and conditions: (a) (1) H2, Pd/C, MeOH, 23 °C (3
days); (2) EtOCOCOCl, Et3N, CH2Cl2, 0 °C (10 h); (3) Et3N, EtOH,
reflux (72 h). (b) Diiodoalkane, K2CO3, CH3CN, reflux (10 h). (c)
AgOPO(OBn)2, CH3CN, reflux (12 h). (d) [48% HBr-H2O (2:1)]-
MeOH (1:1), 55-60 °C (3 h).
constant for the process E + I a EI. Kis is the inhibitor
g
dissociation constant for the process ES + I a ESI. Ki values of
less than 1 µM are shown in boldface type.
recombinant E. coli riboflavin synthase, and recombinant
M. tuberculosis lumazine synthase. The inhibition con-
stants and inhibition mechanisms for 21 and 22 are listed
in Table 1. To evaluate the effect of substitution with the
alkyl phosphate chains in these compounds, the dioxolu-
mazine system 14 itself was also tested as an inhibitor
of all three enzymes, and the previously determined data
for the purinetriones 23 and 24 are also listed in Table
1 for comparison.11 Representative Lineweaver-Burk
plots for inhibition of B. subtilis lumazine synthase, M.
tuberculosis lumazine synthase, and E. coli riboflavin
synthase by inhibitor 22 are presented in Figure 2.
The syntheses of the two phosphates 21 and 22 having
four- and five-methylene linker chains between the
phosphate and the lumazinedione ring system are out-
lined in Scheme 3. Since the lumazinedione ring system
14 and the ribityl hydroxyl groups can be alkylated, the
synthesis required the generation of these two moieties
in protected form before the desired alkylation reaction
could be carried out. Starting from the known pyrimidine
15,11 catalytic reduction of the nitro group over palladium
on carbon yielded an unstable amine intermediate that
was reacted immediately with ethyl chlorooxoacetate,
followed by heating the intermediate with triethylamine
in refluxing ethanol, to provide the pteridine system 16.
Reaction of 16 with 1,4-diiodobutane and 1,5-diiodopen-
tane resulted in displacement of one of the iodides from
each diiodo compound to afford the desired alkyl iodides
17 and 18. Displacement reactions on the iodides 17 and
18 with silver dibenzyl phosphate resulted in the alkyl
phosphates 19 and 20. Simultaneous removal of all three
types of protecting groups (TBDMS, methyl, and benzyl)
was accomplished by heating intermediates 19 and 20
with HBr in aqueous methanol at 55-60 °C to afford the
target compounds 21 and 22 in good yields.
The phosphates 21 and 22 were tested as inhibitors of
recombinant B. subtilis lumazine synthase â60 capsids,
Crystal structures are now available for complexes
formed between the substrate analogue 25 and the
J. Org. Chem, Vol. 70, No. 20, 2005 8165

Cushman et al.
tuberculosis lumazine synthase all allow the rational
docking and energy minimization of additional lumazine
synthase inhibitors to create hypothetical structures of
their enzyme complexes. In the present case, a molecular
model was constructed by overlapping the structure of
22 with that of 25 in a 15 Å spherical fragment sur-
rounding the ligand in one of the 60 equivalent active
sites of B. subtilis lumazine synthase. The structure of
25 was then deleted and the energy of the complex was
minimized by use of the MMFF94s force field and
MMFF94 charges, while the ligand and the protein
structure contained within a 6 Å sphere surrounding the
ligand were allowed to be flexible and the remainder of
the protein structure was immobile. The resulting struc-
ture is displayed in Figure 3, which shows the bound
ligand 22 (red) and the surrounding amino acid residues
of the protein that are calculated to be involved in
hydrogen bonding (yellow lines) to the ligand. The
calculated structure of the complex shows the expected
stacking of the lumazinedione ring system of the ligand
with the benzene ring of Phe22. There is extensive
bonding of the phosphate moiety with the side-chain
nitrogens of Arg127, the side-chain hydroxyl of Thr86,
and the backbone nitrogens of Thr86 and Ala85. The
imide fragment of the heterocyclic system has contacts
with the backbone nitrogen of Ala56, the backbone
carbonyl of Thr80, and the side-chain nitrogen of Asn23.
The side-chain nitrogen of Lys135 is calculated to play a
strong role in stabilizing the complex, with possible
hydrogen-bonding contacts with the C-7 carbonyl of the
heterocyclic system of the ligand as well as with the 3′-
and 4′-hydroxyl groups. A water molecule hydrogen-
bonds to the 2′-hydroxyl group of the ribityl moiety, and
the 4′-hydroxyl group hydrogen-bonds to the side-chain
carbonyl of Phe113′. The 5′-hydroxyl group is calculated
to bond to the side-chain hydroxy group of Ser142 and
the carboxyl of Glu58. All of the calculated hydrogen
bonds and distances are displayed in Figure 4.
FIGURE 2. Lineweaver-Burk plots for the inhibition of B.
subtilis lumazine synthase by inhibitor 22 (top panel; inhibitor
concentrations 0, 3.0, 6.0, and 8.6 µM, mechanism competitive),
M. tuberculosis lumazine synthase (center panel; inhibitor
concentrations 0, 0.8, 1.0, and 1.2 µM, mechanism competitive),
and E. coli riboflavin synthase (bottom panel; inhibitor con-
centrations 0, 1.5, and 2.0 µM, mechanism competitive). The
inset in the middle panel shows an expanded view of the
simulation plot in the region 1/S ) 0. The kinetic data were
fitted with a nonlinear regression method by the program
Dynafit.25 Different kinetic models were considered, and the
most likely inhibition mechanism was determined to be
competitive inhibition in each case.
New antibiotics that are active against drug-resistant
M. tuberculosis are needed urgently. Therefore, the
phosphates 21 and 22, as well as the parent compound
14, were tested as inhibitors of M. tuberculosis lumazine
synthase. Both of the phosphates 21 (Ki 0.036 µM) and
22 (Ki 0.012 µM), like 23 and 24, proved to be very potent
inhibitors with Ki values in the low nanomolar range.
Interestingly, the attachment of the phosphate chains to
the lumazinedione 14 caused a dramatic increase in
potency versus M. tuberculosis lumazine synthase (com-
lumazine synthases of B. subtilis and Schizosaccharo-
myces pombe.19,20 The X-ray structures of the intermedi-
ate analogue 11 bound to S. cerevisiae lumazine synthase
(Figure 1),8 the product analogue 26 bound to S. pombe
lumazine synthase,20 and the intermediate analogue
inhibitor 23 and its lower homologue bound to M.
(19) Ritsert, K.; Huber, R.; Turk, D.; Ladenstein, R.; Schmidt-Ba¨se,
K.; Bacher, A. J. Mol. Biol. 1995, 253, 151-167.
(20) Gerhardt, S.; Haase, I.; Steinbacher, S.; Kaiser, J. T.; Cushman,
M.; Bacher, A.; Huber, R.; Fischer, M. J. Mol. Biol. 2002, 318, 1317-
1329.
8166 J. Org. Chem., Vol. 70, No. 20, 2005

Lumazine Synthase and Riboflavin Synthase Inhibitors
FIGURE 3. Hypothetical model for the binding of compound 22 to B. subtilis lumazine synthase. The figure is programmed for
walleyed viewing.
mixed inhibitor with Ki 9.7 µM and Kis 21 µM, while 22
was a competitive inhibitor with Ki 0.14 µM. This
documents a significant effect of chain length on potency,
and it is similar to the effect seen with the purinetriones
23 and 24. However, both of the lumazinedione phos-
phate derivatives 21 and 22 were much less active than
the lumazinedione system 14 itself (Ki 0.0062 µM), so the
attachment of the phosphate side chains to these mol-
ecules causes a significant drop in potency versus E. coli
riboflavin synthase. One of the reasons for originally
conducting this study was to test the hypothesis that the
pyrimidinediones 21 and 22 would be more potent as
riboflavin synthase inhibitors than the purinetriones 23
and 24, and the results indicate that they in fact are more
potent inhibitors. The additional carbonyl group present
in 21 and 22 therefore does have a significantly positive
effect on the riboflavin synthase enzyme inhibitory
activity that is also reflected in the parent compounds
12 (Ki 0.61 µM) and 14 (Ki 0.0062 µM).
The crystal structure of E. coli riboflavin synthase,
which contains three identical subunits, has been pub-
lished along with a hypothetical model of bound substrate
3.21 A crystal structure has also been determined of
monomeric S. pombe riboflavin synthase with bound
substrate analogue 26,22 and an NMR investigation of
the structure of the N-terminal domain of riboflavin
synthase has also appeared.23 The studies have estab-
lished that the active site of riboflavin synthase is derived
from the intermolecular juxtaposition of residues belong-
ing to both of the barrels in the C-terminal and N-
terminal domains of two subunits, with one substrate
molecule bound to each barrel. Thus, structures are
available for the trimeric apoenzyme and a ligand-bound
monomer, and the structures of the active site can be
proposed by combining elements of the two structures.
The structures allow the molecular modeling of the
complex formed between E. coli riboflavin synthase and
the inhibitor 22 by a similar procedure as described for
FIGURE 4. Hydrogen bonds and distances in the calculated
model of the inhibitor 22 bound in the active site of B. subtilis
lumazine synthase.
pare 14 versus 21 and 22). This is in contrast to the
results seen with the B. subtilis enzyme, where decreases
in potency resulted from side-chain attachment when
tested in the presence of variable substrate 1 concentra-
tion. As with the B. subtilis enzyme, the increase in the
side-chain length resulted in a modest increase in potency
versus M. tuberculosis lumazine synthase (compare 21
versus 22). However, both 21 and 22 appeared to be
slightly less active versus M. tuberculosis lumazine
synthase than 23 and 24. All of the compounds tested
proved to be competitive inhibitors of M. tuberculosis
lumazine synthase.
The purinetrione system 12 can be considered to be a
product analogue of lumazine synthase and a substrate
analogue of riboflavin synthase. Accordingly, it was
previously tested as an inhibitor of both enzymes and
found to be a Ki 46 µM inhibitor of B. subtilis lumazine
synthase (with variable 2 concentration) and a 0.61 µM
inhibitor of E. coli riboflavin synthase.10 However, un-
expectedly, the phosphate derivatives 23 and 24 also
proved to be riboflavin synthase inhibitors.11 Accordingly,
both lumazindione derivatives 21 and 22 were tested
versus E. coli riboflavin synthase. Compound 21 was a
(21) Liao, D.-I.; Wawrzak, Z.; Calabrese, J. C.; Viitanen, P. V.;
Jordan, D. B. Structure 2001, 9, 399-408.
(22) Gerhardt, S.; Schott, A.-K.; Kairies, N.; Cushman, M.; Illarionov,
B.; Eisenreich, W.; Bacher, A.; Huber, R.; Steinbacher, S.; Fischer, M.
Structure 2002, 10, 1371-1381.
(23) Truffault, V.; Coles, M.; Diericks, T.; Abelmann, K.; Eberhardt,
S.; Lu¨ttgen, H.; Bacher, A.; Kessler, H. J. Mol. Biol. 2001, 309, 949-
960.
J. Org. Chem, Vol. 70, No. 20, 2005 8167

Cushman et al.
FIGURE 5. Hypothetical model of the binding of two molecules of inhibitor 22 to E. coli riboflavin synthase. The C-barrel is
magenta and the N-barrel is green. The model is programmed for walleyed viewing.
reflux for 72 h. The solvent was removed under reduced
pressure, and the residue was purified by flash chromatogra-
phy (silica gel, 50 g, 230-400 mesh), eluted with hexanes-
ethyl acetate (1:1), to furnish pure 16 (1.55 g, 36%) as a foamy
solid. 1H NMR (300 MHz, CDCl3) δ 8.52 (s, 1 H), 5.05 (dd, J )
9.9 and 13.15 Hz, 1 H), 4.47 (d, J ) 9.9 Hz, 1 H), 4.38 (dd, J
) 2.49 and 13.1 Hz, 1 H), 4.12 (s, 3 H), 4.05 (t, J ) 2.13 Hz, 1
H), 4.01 (s, 3 H), 3.84 (m, 1 H), 3.72 (m, 1 H), 3.60 (dd, J )
6.03 and 9.64 Hz, 1 H), 0.97 (s, 9 H), 0.90 (s, 9 H), 0.89 (s, 9
H), 0.64 (s, 9 H), 0.19 (s, 3 H), 0.14 (s, 3 H), 0.12 (s, 9 H), 0.07
(s, 3 H), 0.06 (s, 3 H), -0.02 (s, 3 H); IR (KBr) 2954, 2957,
1713, 1603, 1472, 1488, 1378, 1254, 1103, 834 cm-1; EIMS m/z
815 (MH+). Anal. Calcd for C37H74N4O4Si8: C, 54.50; H, 9.15;
N, 6.87. Found: C, 54.50; H, 9.16; N, 6.67.
8-(2′,3′,4′,5′-Tetrakis-t-butyldimethylsilyl-D-ribityl)-
5,6,7,8-tetrahydro-5-(4′-iodobutyl)-2,4-dimethoxy-6,7-di-
oxopteridine (17). Compound 16 (300 mg, 0.37 mmol) was
dissolved in dry CH3CN (20 mL). Anhydrous K2CO3 (200 mg,
1.45 mmol) and 1,4-diiodobutane (0.24 mL, 1.84 mmol) were
added, and the mixture was heated at reflux overnight. The
mixture was filtered and the solvent was removed under
reduced pressure. The residue was purified by flash chroma-
tography (SiO2, 230-400 mesh) (elution with hexanes-EtOAc
95:5) to afford pure 17 (234 mg, 63.0%) as colorless oil. 1H NMR
(300 MHz, CDCl3) δ 5.08 (dd, J ) 10.09 and 13.06 Hz, 1 H),
4.42 (d, J ) 9.79 Hz, 1 H), 4.34 (t, J ) 9.41 Hz, 3 H), 4.09 (s,
3 H), 4.07 (d, J ) 13.3 Hz, 1 H), 3.98 (s, 3 H), 3.84 (t, J ) 6.48
Hz, 1 H), 3.69 (dd, J ) 7.23 and 17.95 Hz, 1 H), 3.58 (dd, J )
6.03 and 10.56 Hz, 1 H), 3.21 (t, J ) 6.46 Hz, 2 H), 1.88 (quint,
J ) 7.21 Hz, 2 H), 1.82 (quint, J ) 7.21 Hz, 2 H), 0.94 (s, 9 H),
0.87 (s, 9 H), 0.86 (s, 9 H), 0.62 (s, 9 H), 0.16 (s, 3 H), 0.12 (s,
3 H), 0.09 (s, 6 H), 0.043 (s, 3 H), 0.036 (s, 3 H), -0.055 (s, 3
H), -0.49 (s, 3 H). EIMS m/z 997 (MH+). Anal. Calcd for C41H81-
IN4O8Si4: C, 49.37; H, 8.19; N, 5.62. Found: C, 49.21; H, 8.38;
N, 5.27.
modeling with lumazine synthase, and the result is
displayed in Figure 5. The C-barrel in Figure 5 is colored
magenta and the N-barrel is green. The two ligand mol-
ecules are stacked with their ribityl side chains pointing
in opposite directions, which is consistent with both the
known regiochemistry of the riboflavin synthase-cata-
lyzed reaction and the available crystal structures.3 This
means that the two phosphate side chains must also be
pointing in opposite directions. Overall, the structure
indicates that riboflavin synthase could in fact accommo-
date two molecules of the inhibitor 22 in the active site.
In conclusion, two alkyl phosphate derivatives of the
6,7-dihydro-6,7-dioxo-8-ribityllumazine system were syn-
thesized and found to be inhibitors of B. subtilis and M.
tuberculosis lumazine synthase as well as E. coli ribo-
flavin synthase. The possible binding modes of these
compounds were investigated through molecular model-
ing of complexes formed between compound 22 and both
lumazine synthase and riboflavin synthase. Potential
antibiotics that inhibit both lumazine synthase and
riboflavin synthase would have a potential therapeutic
advantage over those that inhibit only one enzyme
because microorganisms would have to mutate both
enzymes at the same time in order to acquire drug
resistance.
Experimental Section
General. Melting points were determined in capillary tubes
and are uncorrected. Proton nuclear magnetic resonance
spectra (1H NMR) were determined at 300 MHz or at 500 MHz
as noted. Silica gel used for column chromatography was 230-
400 mesh.
8-(2′,3′,4′,5′-Tetrakis-t-butyldimethylsilyl-D-ribityl)-
5,6,7,8-tetrahydro-5-(5′-iodopentyl)-2,4-dimethoxy-6,7-di-
oxopteridine (18). Compound 16 (250 mg, 0.31 mmol) was
dissolved in dry CH3CN (10 mL). Anhydrous K2CO3 (200 mg,
1.45 mmol) and 1,5-diiodopropane (0.24 mL, 1.53 mmol) were
added, and the mixture was heated at reflux overnight. The
mixture was filtered and the solvent was removed under
reduced pressure. The residue was purified by flash chroma-
tography (SiO2, 230-400 mesh) (elution with hexanes-EtOAc
8-(2′,3′,4′,5′-Tetrakis-t-butyldimethylsilyl-D-ribityl)-
5,6,7,8-tetrahydro-2,4-dimethoxy-6,7-dioxopteridine (16).
The nitro compound 15 (4.21 g, 5.32 mmol) was dissolved in
methanol (150 mL), and palladium on charcoal (10%, 500 mg)
was added. The reaction mixture was hydrogenated at room
temperature and 1 atm for 3 days. The catalyst was filtered
off, the filtrate was concentrated to remove the solvent
completely, and the residue was dried well under vacuum.
Triethylamine (2.2 mL, 16 mmol) and dichloromethane (30 mL)
were added and the reaction mixture was cooled to 0 °C. A
solution of ethyl oxalyl chloride (0.6 mL, 5.32 mmol) in
dichloromethane (10 mL) was added dropwise and the mixture
was stirred overnight at RT. The reaction mixture was washed
with water (2 × 20 mL), dried with anhydrous Na2SO4, filtered,
and concentrated under vacuum. The residue was dissolved
in absolute ethanol (150 mL) and TEA (20 mL) and heated at
1
95:5) to afford pure 18 (211 mg, 68.0%) as a colorless oil. H
NMR (500 MHz, CDCl3) δ 5.08 (dd, J ) 10.14 and 13.01 Hz, 1
H), 4.42 (d, J ) 9.78 Hz, 1 H), 4.31 (t, J ) 9.41 Hz, 3 H), 4.08
(s, 3 H), 4.05 (d, J ) 11.0 Hz, 1 H), 3.98 (s, 3 H), 3.84 (t, J )
5.9 Hz, 1 H), 3.69 (dd, J ) 7.23 and 17.95 Hz, 1 H), 3.58 (dd,
J ) 6.06 and 10.56 Hz, 1 H), 3.18 (t, J ) 6.82 Hz, 2 H), 1.85
(quint, J ) 7.21 Hz, 2 H), 1.69 (quint, J ) 7.21 Hz, 2 H), 1.48
(quint, J ) 7.21 Hz, 2 H), 0.94 (s, 9 H), 0.87 (s, 9 H), 0.86 (s,
8168 J. Org. Chem., Vol. 70, No. 20, 2005

Lumazine Synthase and Riboflavin Synthase Inhibitors
9 H), 0.62 (s, 9 H), 0.16 (s, 3 H), 0.12 (s, 3 H), 0.09 (s, 6 H),
0.043 (s, 3 H), 0.036 (s, 3 H), -0.057 (s, 3 H), -0.49 (s, 3 H);
EIMS m/z 1011 (MH+). Anal. Calcd for C42H83IN4O8Si4: C,
49.88; H, 8.27; N, 5.54. Found: C, 49.79; H, 7.85; N, 5.46.
structure of the complex of 5-nitro-6-ribitylamino-2,4-(1H,3H)-
pyrimidinedione (25) and the lumazine synthase of B. subtilis
(1RVV)19 was clipped to include information within a 15
Å-radius of one of the 60 equivalent ligand molecules. The
residues that were clipped in this cut complex were capped
with either neutral amino or carboxyl groups. The structure
of the inhibitor 22 was overlapped with the structure of 5-nitro-
6-ribitylamino-2,4-(1H,3H)pyrimidinedione (25), which was
then deleted. Hydrogen atoms were added to the complex.
MMFF94 charges were loaded, and the energy of the complex
was minimized by the Powell method to a termination gradient
of 0.05 kcal/mol while the MMFF94s force field was employed.
During the minimization of the complex, inhibitor 22 and a 6
Å sphere surrounding it were allowed to remain flexible, while
the remaining portion of the complex was held rigid by use of
the aggregate function. Figure 3 was constructed by displaying
the amino acid residues of the enzyme that are involved in
hydrogen bonding with the inhibitor 22.
Molecular Modeling on Riboflavin Synthase. By use
of Sybyl (Tripos, Inc., version 7.0, 2004), the X-ray crystal
structure of E. coli riboflavin synthase (1I8D) was downloaded
and two molecules of the ligand 22 were docked and oriented
as suggested by the published model of the binding of two
molecules of the substrate 1 in the active site,21 as well as by
the structure of 22 bound to S. pombe riboflavin synthase.22
The C- and N-terminal groups were changed to neutral
carboxylic acid and amino groups, and hydrogens were added
to the protein structure and to the oxygens of the water
molecules. MMFF94 charges were loaded, and the energy of a
6 Å-radius spherical subset including and surrounding the two
ligand molecules was minimized by the Powell method to a
termination gradient of 0.05 kcal/mol while the MMFF94s
force field was employed. During energy minimization, the
remaining protein structure was held rigid by use of the
aggregate function. Figure 5 was constructed by displaying
the amino acid residues in the C- and N-barrels surrounding
the two ligand molecules.
Dibenzyl 4-[8-(2′,3′,4′,5′-Tetrakis-t-butyldimethylsilyl-
D-ribityl)-5,6,7,8-tetrahydro-2,4-dimethoxy-6,7-dioxopterid-
5-yl]butane 1-Phosphate (19). Compound 17 (0.10 g, 0.10
mmol) and silver dibenzyl phosphate (0.06 g, 0.15 mmol) were
heated at reflux in dry CH3CN (5.0 mL) for 10 h. The solution
was then filtered and concentrated under reduced pressure.
The resulting oil was purified by silica gel flash chromatog-
raphy (SiO2, 230-400 mesh), eluted with hexanes-ethyl
acetate (1:1), to furnish the desired compound 19 (0.08 g, 70%)
as a colorless oil. 1H NMR (300 MHz, CDCl3) δ 7.31 (s, 10 H),
5.07 (dd, J ) 10.09 and 13.06 Hz, 1 H), 4.99 (m, 4 H), 4.41 (d,
J ) 9.41 Hz, 1 H), 4.33 (d, J ) 13.15 Hz, 1 H), 4.26 (t, J )
6.75 Hz, 2 H), 3.99 (q, J ) 6.04 Hz, 2 H), 3.98 (s, 3 H), 3.97 (s,
3 H), 3.83 (t, J ) 6.45 Hz, 1 H), 3.68 (t, J ) 8.82 Hz, 1 H), 3.57
(dd, J ) 5.99 and 10.33 Hz, 1 H), 1.67-1.66 (m, 4 H), 0.93 (s,
9 H), 0.87 (s, 9 H), 0.86 (s, 9 H), 0.60 (s, 9 H), 0.15 (s, 3 H),
0.11 (s, 3 H), 0.087 (s, 6 H), 0.038 (s, 3 H), 0.032 (s, 3 H), -0.07
(s, 3 H), -0.52 (s, 3 H); EIMS m/z 1147 (MH+). Anal. Calcd
for C55H95N4O12PSi4: C, 57.56; H, 8.34; N, 4.88. Found: C,
57.33; H, 8.40; N, 5.28.
Dibenzyl 4-[8-(2′,3′,4′,5′-Tetrakis-t-butyldimethylsilyl-
D-ribityl)-5,6,7,8-tetrahydro-2,4-dimethoxy-6,7-dioxopterid-
5-yl]pentane 1-Phosphate (20). Compound 18 (0.21 g, 0.21
mmol) and silver dibenzyl phosphate (0.12 g, 0.31 mmol) were
heated at reflux in dry CH3CN (10 mL) for 10 h. The solution
was then filtered and concentrated under reduced pressure.
The resulting oil was purified by silica gel flash chromatog-
raphy (SiO2, 230-400 mesh), eluted with hexanes-ethyl
acetate (1:1), to furnish the desired compound 20 (0.17 g, 71%)
as a colorless oil. 1H NMR (300 MHz, CDCl3) δ 7.32 (s, 10 H),
5.08 (dd, J ) 10.07 and 13.06 Hz, 1 H), 5.01 (m, 4 H), 4.41 (d,
J ) 9.87 Hz, 1 H), 4.33 (d, J ) 13.17 Hz, 1 H), 4.25 (t, J )
7.72 Hz, 2 H), 4.00 (s, 3 H), 3.97 (q, J ) 6.04 Hz, 2 H), 3.96 (s,
3 H), 3.84 (t, J ) 5.79 Hz, 1 H), 3.69 (t, J ) 8.82 Hz, 1 H), 3.58
(dd, J ) 5.99 and 10.33 Hz, 1 H), 1.65-1.63 (m, 4 H), 1.53 (m,
2 H), 0.94 (s, 9 H), 0.87 (s, 9 H), 0.86 (s, 9 H), 0.61 (s, 9 H),
0.16 (s, 3 H), 0.11 (s, 3 H), 0.092 (s, 6 H), 0.043 (s, 3 H), 0.038
(s, 3 H), -0.061 (s, 3 H), -0.51 (s, 3 H); EIMS m/z 1161 (MH+).
Anal. Calcd for C56H97N4O12PSi4: C, 57.93; H, 8.36; N, 4.82.
Found: C, 57.79; H, 8.06; N, 4.84.
Lumazine Synthase Assay.24 The experiments with B.
subtilis lumazine synthase were performed as follows. Reaction
mixtures contained 100 mM potassium phosphate, pH 7.0, 5
mM ethylenediaminetetraacetic acid (EDTA), 5 mM dithio-
threitol, inhibitor (0-500 µM), 150 µM L-3,4-dihydroxy-2-
butanone 4-phosphate (2), and B. subtilis lumazine synthase
(7 µg, specific activity 12.8 µmol mg-1 h-1) in a total volume
of 1000 µL. The solution was incubated at 37 °C, and the
reaction was started by the addition of a small volume (20 µL)
of 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione (1) to
a final concentration of 5-100 µM. The experiments with M.
tuberculosis lumazine synthase were conducted as follows.
Reaction mixtures contained 50 mM Tris HCl, pH 7.0, 100 mM
NaCl, 5 mM EDTA, 5 mM dithiothreitol, 150 µM L-3,4-
dihydroxy-2-butanone 4-phosphate (2), inhibitor (0-500 µM),
and M. tuberculosis enzyme (12.5 µg, specific activity 4.3 µmol
mg-1 h-1) in a total volume of 1000 µL. The mixtures were
incubated at 37 °C, and the reaction was started by the
addition of a small volume (20 µL) of substrate 1 to a final
concentration of 20-400 µM. The formation of 6,7-dimethyl-
8-ribityllumazine (3) was measured online with a computer-
controlled photometer at 408 nm (ꢀlumazine ) 10 200 M-1cm-1).
The velocity-substrate data were fitted for all inhibitor
concentrations with a nonlinear regression method by use of
the program DynaFit.25 Different inhibition models were
considered for the calculation. Ki and Kis values ( standard
deviations were obtained from the fit under consideration of
the most likely inhibition model.
4-(1,5,6,7-Tetrahydro-6,7-dioxo-8-D-ribityllumazin-5-
yl)butane 1-Phosphate (21). Compound 19 (82 mg, 0.07
mmol) was dissolved in a solution of [48% HBr-H2O (2:1)]-
MeOH (1:1; 5 mL), and the mixture was stirred at 55-60 °C
for 3 h. The solvent was removed in vacuo, the residue was
dissolved in MeOH (1 mL), and ethyl ether (5 mL) was added.
After 24 h in the refrigerator, the precipitate was filtered out
as a white solid, which was dissolved in water (5 mL),
decolorized with active charcoal, and filtered. The colorless
filtrate was lyophilized to furnish 21 (25 mg, 73%) as a white,
highly hygroscopic amorphous solid. 1H NMR (D2O) (300 MHz)
δ 3.41-4.27 (m, 11 H), 1.45 (m, 4 H); 13C NMR (D2O) (75 MHz)
δ 158.7, 156.7, 154.5, 149.9, 137.8, 100.6, 76.4, 73.0, 72.4, 71.6,
70.1, 49.2, 45.8, 26.2, 24.9; ESI-MS (negative ion mode) m/z
481 [(M - H+)-]. Anal. Calcd for C15H23N4O12P‚0.6H2O: C,
36.50; H, 4.91; N, 11.35. Found: C, 36.84; H, 4.71; N, 10.92.
5-(1,5,6,7-Tetrahydro-6,7-dioxo-8-D-ribityllumazin-5-yl-
)pentane 1-Phosphate (22). This compound was prepared
from 20 in 78% yield by the procedure above to afford the
product as a white, highly hygroscopic amorphous solid (38
1
mg, 78%). H NMR (D2O) (300 MHz) δ 3.64-4.58 (m, 11 H),
Riboflavin Synthase Assay.26 Reaction mixtures con-
tained buffer (100 mM potassium phosphate, pH 7.0, 10 mM
EDTA, and 10 mM sodium sulfite), inhibitor (0-300 µM), and
1.75 (m, 4 H), 1.52 (m, 2 H); 13C NMR (D2O) (75 MHz) δ 158.7,
156.8, 154.5, 150.0, 137.9, 100.6, 73.0, 72.4, 70.1, 66.6, 62.8,
47.0, 46.1, 29.6, 27.9, 22.4; ESI-MS m/z 495 [(M - H+)-]. Anal.
Calcd for C16H25N4O12P‚0.5H2O: C, 37.99; H, 5.14; N, 11.08.
Found: C, 37.98; H, 4.85; N, 10.73.
(24) Kis, K.; Bacher, A. J. Biol. Chem. 1995, 270, 16788-16795.
(25) Kuzmic, P. Anal. Biochem. 1996, 237, 260-273.
(26) Eberhardt, S.; Richter, G.; Gimbel, W.; Werner, T.; Bacher, A.
Eur. J. Biochem. 1996, 242, 712-718.
Molecular Modeling on Lumazine Synthase. By use of
Sybyl (Tripos, Inc., version 7.0, 2004), the X-ray crystal
J. Org. Chem, Vol. 70, No. 20, 2005 8169

Cushman et al.
riboflavin synthase (1.2 µg, specific activity 45 µmol mg-1 h-1).
After preincubation, the reactions were started by the addition
of various amounts of 6,7-dimethyl-8-ribityllumazine (3) (2.5-
200 µM) to a total volume of 1000 µL. The formation of
riboflavin (4) was measured online with a computer-controlled
photometer at 470 nm (ꢀriboflavin ) 9600 M-1 cm-1). The
evaluation of data sets was performed in the same manner as
described above.
Acknowledgment. This research was made possible
by NIH Grant GM51469 as well as by support from
the Deutsche Forschungsgemeinschaft, the Fonds der
Chemischen Industrie, and the Hans Fischer Gesell-
schaft e.V.
JO051332V
8170 J. Org. Chem., Vol. 70, No. 20, 2005