Bisubstrate analogue inhibitors of 6-hydroxymethyl-7,8-dihydropterin pyrophosphokinase: Synthesis and biochemical and crystallographic studies

Article abstract of DOI:10.1021/jm0004493

6-Hydroxymethyl-7,8-dihydropterin pyrophosphokinase (HPPK) catalyzes the transfer of pyrophosphate from ATP to 6-hydroxymethyl-7,8-dihydropterin (HP), leading to the biosynthesis of folate cofactors. Like other enzymes in the folate pathway, HPPK is an ideal target for the development of antimicrobial agents because the enzyme is essential for microorganisms but is absent from human and animals. Three bisubstrate analogues have been synthesized for HPPK and characterized by biochemical and X-ray crystallographic analyses. All three bisubstrate analogues consist of a pterin, an adenosine moiety, and a link composed of 2-4 phosphoryl groups. P¹-(6-Hydroxymethylpterin)-P²-(5′-adenosyl) diphosphate (HP₂A, 5) shows little affinity and inhibitory activity for E. coli HPPK. P¹-(6-Hydroxymethylpterin)-P³-(5′-adenosyl) triphosphate (HP₃A, 6) shows moderate affinity and inhibitory activity with K_d = 4.25 μM in the presence of Mg²⁺ and IC₅₀ = 1.27 μM. P¹-(6-Hydroxymethylpterin)-P⁴-(5′-adenosyl) tetraphosphate (HP₄A, 7) shows the highest affinity and inhibitory activity with K_d = 0.47 μM in the presence of Mg²⁺ and IC₅₀ = 0.44 μM. The affinity of MgHP₄A for HPPK is ～116 and 76 times higher than that of MgADP and 6-hydroxymethylpterin, respectively. The crystal structure of HPPK in complex with 7 (HPPK·MgHP₄A) has been determined at 1.85 A? resolution with a crystallographic R factor of 0.185. The crystal structure shows that 7 occupies both HP- and ATP-binding sites and induces significant conformational changes in HPPK. The biochemical and structural studies of the bisubstrate analogues indicate that the bisubstrate analogue approach can produce more potent inhibitors for HPPK and the minimum length of the link for a bisubstrate analogue is ～7 A?.

Full text of DOI:10.1021/jm0004493

1364
J . Med. Chem. 2001, 44, 1364-1371
Bisu bstr a te An a logu e In h ibitor s of 6-Hyd r oxym eth yl-7,8-d ih yd r op ter in
P yr op h osp h ok in a se: Syn th esis a n d Bioch em ica l a n d Cr ysta llogr a p h ic Stu d ies†
Genbin Shi,§,# J aroslaw Blaszczyk,|,# Xinhua J i,*,| and Honggao Yan*,§
Department of Biochemistry and Molecular Biology, Michigan State University, East Lansing, Michigan 48824, and
Macromolecular Crystallography Laboratory, National Cancer Institute, Frederick, Maryland 21702
Received October 20, 2000
6-Hydroxymethyl-7,8-dihydropterin pyrophosphokinase (HPPK) catalyzes the transfer of
pyrophosphate from ATP to 6-hydroxymethyl-7,8-dihydropterin (HP), leading to the biosynthesis
of folate cofactors. Like other enzymes in the folate pathway, HPPK is an ideal target for the
development of antimicrobial agents because the enzyme is essential for microorganisms but
is absent from human and animals. Three bisubstrate analogues have been synthesized for
HPPK and characterized by biochemical and X-ray crystallographic analyses. All three
bisubstrate analogues consist of a pterin, an adenosine moiety, and a link composed of 2-4
phosphoryl groups. P1-(6-Hydroxymethylpterin)-P2-(5′-adenosyl)diphosphate (HP2A, 5) shows
little affinity and inhibitory activity for E. coli HPPK. P1-(6-Hydroxymethylpterin)-P3-(5′-
adenosyl)triphosphate (HP3A, 6) shows moderate affinity and inhibitory activity with Kd )
4.25 µM in the presence of Mg2+ and IC50 ) 1.27 µM. P1-(6-Hydroxymethylpterin)-P4-(5′-
adenosyl)tetraphosphate (HP4A, 7) shows the highest affinity and inhibitory activity with Kd
) 0.47 µM in the presence of Mg2+ and IC50 ) 0.44 µM. The affinity of MgHP4A for HPPK is
∼116 and 76 times higher than that of MgADP and 6-hydroxymethylpterin, respectively. The
crystal structure of HPPK in complex with 7 (HPPK‚MgHP4A) has been determined at 1.85 Å
resolution with a crystallographic R factor of 0.185. The crystal structure shows that 7 occupies
both HP- and ATP-binding sites and induces significant conformational changes in HPPK.
The biochemical and structural studies of the bisubstrate analogues indicate that the bisubstrate
analogue approach can produce more potent inhibitors for HPPK and the minimum length of
the link for a bisubstrate analogue is ∼7 Å.
Ch a r t 1. Reaction Catalyzed by HPPK and Chemical
Structures of the Bisubstrate Analogues and Key
Precursorsa
In tr od u ction
Rapidly increasing antibiotic resistance in recent
years has rendered the current antibiotics ineffective
for treating many microbial infections, resulting in a
worldwide health care crisis.1-5 According to a 1999
WHO report,6 infectious diseases are the leading causes
of death and the main causes of premature death in the
world. The crisis is further aggravated by the fact that
most new antibiotics are chemical modifications of the
basic structures of existing antimicrobial agents against
old targets and thus less effective against widespread
antibiotic resistance. Therefore, new targets for develop-
ing novel antimicrobial agents are urgently needed for
combating the antibiotic crisis.
6-Hydroxymethyl-7,8-dihydropterin pyrophosphoki-
nase (HPPK) catalyzes the transfer of pyrophosphate
from ATP to 6-hydroxymethyl-7,8-dihydropterin (HP)
(Chart 1), the first reaction in the folate biosynthetic
pathway.7 Folate cofactors are essential for life.8 Mam-
mals have an active transport system for deriving
folates from their diet. In contrast, most microorganisms
must synthesize folates de novo because they lack the
a
1, 6-hydroxymethyl-7,8-dihydropterin; 2, 6-hydroxymethyl-7,8-
dihydropterin pyrophosphate; 3, 6-hydroxymethylpterin; 4, 6-hy-
droxymethylpterin monophosphate (HP2A); 5, P1-(6-hydroxymeth-
ylpterin)-P2-(5′-adenosyl)diphosphate (HP3A); 6, P1-(6-hydroxy-
methylpterin)-P3-(5′-adenosyl)triphosphate; 7, P1-(6-hydroxymeth-
ylpterin)-P4-(5′-adenosyl)tetraphosphate (HP4A).
† The coordinates and structure factors have been deposited with
the Protein Data Bank: accession code 1ex8.
* To whom correspondence should be addressed. Tel: (517) 353-
8786 (H.Y.); (301) 846-5035 (X.J .). Fax: (517) 353-9334 (H.Y.); (301)
846-6073 (X.J .). E-mail: yanh@pilot.msu.edu (H.Y.); jix@ncifcrf.gov
(X.J .).
§ Michigan State University.
active transport system. Therefore, the folate biosyn-
thetic pathway is an ideal target for developing anti-
| National Cancer Institute.
#
These authors contributed equally to this work.
10.1021/jm0004493 CCC: $20.00 © 2001 American Chemical Society
Published on Web 03/24/2001

Bisubstrate Analogue Inhibitors of HPPK
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 9 1365
microbial agents. Inhibitors of dihydropteroate syn-
thase9 and dihydrofolate reductase,10 the second and
fourth enzymes in the folate pathway, are currently
used as antibiotics for treating many infectious diseases.
HPPK represents a new target in a biosynthetic path-
way that has been proven effective for the development
of antimicrobial agents.
Because of its small size and high thermal stabil-
ity,11,12 HPPK is also an excellent model enzyme for
studying the mechanisms of enzymatic pyrophosphoryl
transfer. Although the mechanisms of many kinases
that catalyze monophosphoryl transfer have been ex-
tensively characterized, little is known about the mech-
anisms of pyrophosphokinases.13
To seek an effective approach to make potent inhibi-
tors for HPPK and probe the mechanism of the enzyme,
we have synthesized three bisubstrate analogues (Chart
1). All three bisubstrate analogues consist of a pterin,
an adenosine moiety, and a link composed of 2-4
phosphoryl groups. Biochemical analysis of the bisub-
strate analogues has shown that P1-(6-hydroxymeth-
ylpterin)-P4-(5′-adenosyl)tetraphosphate (HP4A, 7) is the
most potent inhibitor of HPPK. The crystal structure
of HPPK in complex with 7 (HPPK‚MgHP4A) has been
determined at 1.85 Å resolution with a crystallographic
R factor of 0.185.
Resu lts
Syn th esis. The bisubstrate analogues were synthe-
sized by the method of Hoard and Ott.14 Our initial
attempt was to activate ATP with 1,1′-carbonyldiimid-
azole and then mix it in situ with 6-hydroxymethyl-
pterin (3), but this was not successful. 6-Hydroxymeth-
ylpterin monophosphate (HPP, 4) was then synthesized
by the method of Ho et al.15 The compound was
activated with 1,1′-carbonyldiimidazole and treated in
situ with AMP, ADP, or ATP. The yields for HPP (4),
HP2A (5), HP3A (6), and HP4A (7) were ca. 65%, 31%,
35%, and 16%, respectively. The proton NMR spectrum
of 4 showed two methylene protons as a doublet from
splitting by phosphate. The phosphate chemical shift
(2.84 ppm) of HPP was very similar to that (3.13 ppm)
of AMP. The 31P NMR spectra of 5-7 showed charac-
teristic peaks at approximately -11 and -22 ppm
corresponding to the terminal and middle phosphates,
respectively. The 31P chemical shifts were very similar
to those reported for P1-(5-adenosyl)-Pn-(5-thymidyl)-
polyphosphate.16,17
F igu r e 1. (a) Inhibition of HPPK by HP4A. The solid line was
obtained by nonlinear least-squares fit of the experimental
data. The details of the assay are described in the Experimen-
tal Section. (b) Fluometric titration of HP4A with HPPK. The
concentration of the HPPK stock solution was 190 µM. The
solid line was obtained by nonlinear least-squares fit of the
data as described in the Experimental Section.
Ta ble 1. Summary of the IC50 and Kd Values Obtained by
Inhibition and Binding Studies
HP2A
HP3A
HP4A
Kd (µM) (without Mg2+
)
nd
nd
>100
17.4 ( 1.5
4.25 ( 0.28
1.27 ( 0.3
1.93 ( 0.05
0.47 ( 0.04
0.44 ( 0.16
Kd (µM) (with Mg2+
IC50 (µM)
)
Bioch em ical Stu dies. Representative titration curves
for the inhibition and binding studies are shown in
Figure 1a,b, respectively. The IC50 and Kd values
obtained by nonlinear least-squares analysis of the data
are summarized in Table 1. Among the three bisubstrate
analogues, 7 was the most potent inhibitor of HPPK
with IC50 ) 0.44 µM. The IC50 of 6 was about 3 times of
that of 7. At a concentration of 100 µM, 5 showed little
inhibition of HPPK activity. The Kd values measured
in the presence of Mg2+ were very similar to the
corresponding IC50 values. In the absence of Mg2+, the
Kd values increased by about a factor of 4.
which occupies both the HP- and ATP-binding sites, has
well-defined electron density (Figure 2b).
The overall structure of the protein is similar to that
observed in the HPPK‚HP‚MgAMPCPP complex.18 The
CR trace of the two structures shows some differences
only in loop 2, loop 3, and a region near the C terminus.
Loop 2 consists of residues 44-50, most of which are
involved in HP binding. Loop 3 consists of residues 80-
95, most of which are involved in the binding of ATP. A
main-chain displacement of ∼1.8 Å is observed for
residue Asp49 in loop 2 and Lys85 in loop 3. In HPPK‚
MgHP4A, loop 2 moves away from the active center and
loop 3 moves further toward the active center when
compared with those in the ternary complex. Accord-
ingly, some side chains on the loops have rather differ-
Cr ysta l Str u ctu r e of HP P K‚MgHP 4A. The 1.85 Å
crystal structure of HPPK‚MgHP4A (Figure 2a) contains
1 HPPK molecule, 1 HP4A molecule, 1 Mg2+ ion, 1 Cl1-
ion, and 167 water molecules. The HP4A molecule,

1366 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 9
Shi et al.
except one hydrogen bond between N8 of pterin and
HPPK. In HPPK‚HP‚MgAMPCPP, N8 of HP is hydrogen-
bonded to the carbonyl oxygen of Leu45. In HPPK‚
MgHP4A, however, this hydrogen bond does not exist
(Figure 3b). Because the pterin moiety of HP4A is
oxidized, N8 cannot serve as a hydrogen bond donor.
Consequently, the carbonyl group of Leu45 flips by
∼180° probably due to the unfavorable interaction
between the lone pairs of electrons of the nitrogen and
oxygen atoms. The C-O bond of the hydroxymethyl
group of the pterin moiety rotates by ∼65° probably due
to the positioning of the phosphate chain.
The Mg2+ coordination is rather different between the
two structures. Two Mg2+ ions are found in HPPK‚HP‚
MgAMPCPP,18 one coordinated with the R- and â-
phosphates and the other with the â- and γ-phosphate
groups (Figure 3c). Both Mg2+ ions are coordinated with
the carboxyl groups of Asp95 and Asp97. Only one Mg2+
ion is found in HPPK‚MgHP4A; one oxygen atom from
each of the four phosphate groups and two water
molecules satisfy the octahedral coordination of the
Mg2+ ion (this work, Figure 3d). The two water mol-
ecules are hydrogen-bonded to the carboxyl group of
Asp97. Asp95 forms salt bridges with both Arg82 and
Arg92.
The conformation of the phosphate chain is also
different between the two structures, and so their
interactions with the protein are different as well
(Figure 3c,d). When superimposed, the R-phosphorus of
HP4A is close to the counterpart of ATP (0.81 Å), the â-
and γ-phosphorus atoms of HP4A are close to the
â-phosphorus of ATP (2.89 and 2.30 Å, respectively), and
the δ-phosphorus of HP4A is close to the γ-phosphorus
of ATP (1.61 Å). In HPPK‚MgHP4A (this work), Arg82
interacts with the R-phosphate of HP4A and Arg92
forms hydrogen bonds with γ-phosphate (Figure 3d),
whereas in HPPK‚HP‚MgAMPCPP,18 Arg92 interacts
with the R-phosphate of AMPCPP and Arg82 is not
involved in phosphate binding at all (Figure 3c). The
side chain of Arg88 in the ternary complex interacts
with the oxygen bridge between the â- and γ-phosphates
via a water molecule, whereas in HPPK‚HP4A (this
work) it interacts with a â-phosphate oxygen (Figure
3d). Other residues that show differences in substrate/
inhibitor binding include Trp89, His115, Tyr116, and
Arg121 as shown in Figure 3c,d.
F igu r e 2. (a) Overall structure of the HPPK‚MgHP4A com-
plex. Secondary structure elements and three flexible loops
are labeled according to their order of appearance in the
polypeptide chain. (b) Final 2Fo - Fc electron density of the
inhibitor and Mg2+ ion contoured at the level of 1.3σ. The figure
was generated with BOBSCRIPT.40
Ta ble 2. Strongest Intermolecular Interactions of the
C-Terminal Region in the Crystal Lattice of HPPK‚MgHP4A
(this work) and the Ternary Complex HPPK‚HP‚MgAMPCPP18
HPPK‚MgHP4A
(Å) HPPK‚HP‚MgAMPCPP (Å)
Glu141 OE1-Arg75 NH2 2.90 Asp153 OD1-Arg75 NH2 2.93
Glu141 OE2-Arg75 NH1 2.86 Asp153 OD2-Arg75 NH1 2.87
Asp153 OD1-Arg110 NH1 2.50 His148 ND1-Glu30 OE1 2.58
Asp153 OD2-Glu67 OE1 2.52 Gln145 NE2-Gln80 O
Arg150 NH2-Asn71 O 2.75
2.91
Discu ssion
ent conformations. The largest main-chain atom dis-
placement is observed for Arg150 that is located in the
region between R4-helix and the C-terminal loop (Figure
2a). The conformational differences are most likely due
to the different intermolecular interactions in the two
crystals (Table 2). Five hydrogen bonds and 11 van der
Waals contacts (e 3.50 Å) are found between this region
and symmetry-related molecules in HPPK‚MgHP4A
(this work), whereas 4 hydrogen bonds and 7 short
contacts are found for HPPK‚HP‚MgAMPCPP.18 None
of the intermolecular hydrogen bonds in this region is
the same between the two structures.
The pterin and adenosine moieties of HP4A can be
superimposed well with the counterparts in HPPK‚HP‚
MgAMPCPP.18 The adenosine moiety binds to the
protein with the same hydrogen bond pattern in both
structures (Figure 3a). The hydrogen bond pattern
between the pterin moiety and protein is also the same
Bisu bstr a te An a logu e In h ibitor Design . One of
the most effective approaches to enhance the specificity
and potency in enzyme inhibition is to design multi-
substrate analogue inhibitors.19,20 This approach has
been widely used for designing inhibitors for kinases.
A number of potent bisubstrate analogue inhibitors have
been made by this approach.20,21 Classical examples
include P1,P5-bis(5′-adenosyl)pentaphosphate (AP5A), a
potent inhibitor of adenylate kinase,22 and P1-(5-adeno-
syl)-P5-(5′-thymidyl)pentaphosphate (AP5T), a potent
inhibitor of thymidylate kinase.23 However, not all
multisubstrate analogues are strong inhibitors. For
example, the bisubstrate analogues of hexokinase P1-
(5-adenosyl)-P3-(6-glucosyl)triphosphate and P1-(5-ad-
enosyl)-P4-(6-glucosyl)tetraphosphate are only weak
inhibitors of hexokinase with affinities for the enzyme
much lower than either substrate ATP or glucose.24

F igu r e 3. Stereoviews of the active-site architecture in HPPK‚MgHP4A (orange, this work) in alignment with that of HPPK‚HP‚MgAMPCPP,18 showing the (a) adenosine-binding
site and (b) HP-binding site; (c) binding of the three phosphate groups of AMPCPP in HPPK‚HP‚MgAMPCPP and (d) binding of the tetraphosphate bridge in HPPK‚HP4A. Note the
presence of 2 Mg2+ ions in the ternary complex and 1 Mg2+ ion (orange) in HPPK‚MgHP4A. The figure was generated with BOBSCRIPT.40

1368 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 9
Shi et al.
The pyrophosphoryl-transfer reaction catalyzed by
HPPK is different from the phosphoryl-transfer reac-
tions catalyzed by kinases in that the substitution
reaction takes place at â-phosphate. One might expect
to lose the interactions between the γ-phosphate of ATP
and HPPK when making bisubstrate analogues. Whether
the bisubstrate analogue approach works for pyrophos-
phokinase is not known.
charges carried by the polyphosphate links. To improve
the bioavailability, we will replace the link with neutral
ones in designing the next generation of bisubstrate
inhibitors. Other improvements may include modifica-
tion of the pterin moiety of the inhibitors. The crystal
structure of HPPK‚MgHP4A reported here will be very
valuable for designing the next generation of bisubstrate
inhibitors for HPPK.
Mech a n istic Im p lica tion s. It was surprising to find
that the side chains of several strictly conserved resi-
dues, including those of Gln50, Arg82, and Arg92, point
away from the active center when the structure of apo-
HPPK was determined,26 suggesting that significant
conformational changes must occur upon binding of the
substrates. Indeed, the binding of HP and AMPCPP
induces significant conformational changes in the three
loops that contain these conserved residues.18 Similar
induced conformational changes occur when 7 binds to
HPPK. Loop 2 moves toward the active center by as
much as 6.8 Å and loop 3 by as much as 9.2 Å. Loops 2
and 3 are involved in binding of HP and ATP, respec-
tively. All the arginine residues (Arg82, Arg84, Arg88,
and Arg92) on loop 3 move toward the active center and
interact with the tetraphosphate. The guanidinium
groups of Arg82, Arg84, Arg88, and Arg92 move by as
much as 15.0, 15.4, 17.3, and 10.9 Å, respectively. The
side-chain amide of Gln50 on loop 2 moves toward the
active center by 6.6 Å. As a result, it forms a hydrogen
bond with the side-chain amide of Asn10 on loop 1 and
the carbonyls of Trp89 on loop 3 and Pro47 on loop2. It
is apparent that these parts of the active center are
assembled only after the binding of the substrates.
The secondary structure and folding of the N-terminal
two-thirds polypeptide chain of HPPK is similar to that
of the corresponding part of NDP kinase.18,27 When the
structures of the two enzymes are superimposed, the
phosphate-binding regions overlap but the nucleoside-
binding regions are in opposite directions. The struc-
tural similarity raises the important question whether
the two enzymes follow a similar kinetic mechanism.
NDP kinase is unique among kinases because it has
only one substrate-binding site per subunit and follows
a ping-pong mechanism with a phosphohistidine co-
valent intermediate.28 However, HPPK has two sub-
strate-binding sites and forms a ternary complex with
both substrates.27 The crystal structure of HPPK‚
MgHP4A clearly shows that the bisubstrate analogue
occupies both substrate-binding sites. Furthermore, the
substitution of the conserved histidine residue His115
in E. coli HPPK causes a slight increase in kcat rather
than abolishing the catalytic activity (Gong et al.,
unpublished data), ruling out the possibility of the
formation of a covalent intermediate involving His115.
The results suggest that the HPPK-catalyzed reaction
follows a sequential mechanism involving the formation
of a ternary complex with both substrates bound.
To test whether the bisubstrate analogue approach
is effective for making potent inhibitors of HPPK, we
have synthesized three bisubstrate analogues (5-7).
Since the pyrophosphoryl-transfer reaction catalyzed by
HPPK takes place at â-phosphate, MgADP and 6-hy-
droxymethylpterin are the appropriate standards for
comparing the affinities of the bisubstrate analogues for
HPPK. Both compounds are competitive inhibitors of
the enzyme. The Kd values of MgADP25 and 6-hy-
droxymethylpterin (Shi et al., unpublished data) are 55
and 36 µM, respectively. Since HP2A shows little inhibi-
tory activity at 100 µM concentration, it must have a
much lower affinity for HPPK than MgADP and HP.
The Kd of MgHP3A is ∼4 µM, indicating that the affinity
of MgHP3A for HPPK is ∼12 and 7 times higher than
that of MgADP and 6-hydroxymethylpterin. The affinity
of MgHP4A for HPPK is ∼116 and 76 times higher than
that of MgADP and 6-hydroxymethylpterin, respec-
tively. The results demonstrate that the bisubstrate
analogue approach can produce more potent inhibitors
for HPPK.
Why is the affinity of HP2A (5) for HPPK much lower
than that of 6-hydroxymethylpterin or MgADP? Most
likely it is due to the fact that the diphosphate link is
not long enough to connect pterin and adenosine in their
respective binding pockets in the enzyme. As revealed
by the crystal structure of HPPK‚HP‚MgAMPCPP,18 the
distance between the hydroxyl oxygen of the pterin
moiety and the 5′-hydroxyl oxygen of the adenosine
moiety is 7.2 Å. As shown by the crystal structures of
both HPPK‚MgHP4A (this work) and HPPK‚HP‚Mg-
AMPCPP,18 both the pterin and the adenosine moieties
are well-fixed in the active center of HPPK by many
intermolecular hydrogen bonds. Furthermore, the pterin
and adenosine binding pockets are situated on the core
of the protein and cannot move toward each other
without a major steric clash. Accordingly, the minimum
length for the link of a bisubstrate analogue is likely to
be ∼7 Å, and a minimum of three phosphates is needed
to connect the pterin and adenosine moieties. With a
diphosphate link, probably only one end of 5 (either the
pterin moiety or the adenosine moiety) is bound to
HPPK leaving the other end dangling in solution.
The next question, then, is why 7 has a much higher
affinity for HPPK than 6 if a triphosphate link is long
enough to connect the pterin and adenosine moieties.
This question is addressed by comparing the crystal
structure of HPPK‚MgHP4A with that of HPPK‚HP‚
MgAMPCPP (Figure 3c,d).18 As shown in Figure 3c, the
γ-phosphate of AMPCPP forms five hydrogen bonds
with HPPK. With a triphosphate link, 6 may lose all
these electrostatic interactions. With an extra phos-
phate, 7 is able to keep two of the five electrostatic
interactions (Figure 3d).
Exp er im en ta l Section
Ma ter ia ls. Folic acid, 1,1′-carbonyldiimidazole, and all
nucleotides were purchased from Sigma. Sodium borohydride
(98%), pyrophosphoric acid (tech.), n-tributylamine (99%),
sodium perchlorate hydrate (99%), phosphoric acid (99%) and
other general organic reagents were purchased from Aldrich.
48% HBr was purchased from Spectrum. Bromine was pur-
chased from Baker. DEAE-cellulose was purchased from
Although 6 and 7 are potent inhibitors of HPPK, they
may have poor bioavailability because of many negative

Bisubstrate Analogue Inhibitors of HPPK
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 9 1369
Whatman. TLC sheets silica gel 60 F254 were purchased from
Riedel-de Haen. TLC sheets PEI-cellulose F were purchased
from EM Science.
For HP4A, 1H NMR: 4.30-4.53 (m, 3H), 4.22 (m, 2H), 5.04 (d,
3J ) 7.33 Hz, 2H), 5.88 (d, 3J ) 5.13 Hz, 1H), 8.04 (s, 1H),
8.19 (s, 1H), 8.61 (s, 1H). 31P NMR: -11.02 (broad, 2P), -22.48
(broad, 2P). MALDI-MS (ATT): m/z ) 760.729; 761.01 for
Syn th esis of 6-Hyd r oxym eth ylp ter in (3). 6-Hydroxy-
methylpterin was synthesized according to the method of
Thijssen.29 Briefly, 6-formylpterin was obtained by heating a
solution of folic acid in 40% hydrogen bromide containing
excess bromine at 100 °C for 2 h followed by washing with
water and extraction with hot acetone. 6-Hydroxymethylpterin
was obtained by reduction of 6-formylpterin with sodium
borohydride. The UV and NMR spectra of the 6-hydroxy-
methylpterin were identical to those of the same compound
purchased from Sigma. For kinetic assays, the compound was
reduced to 6-hydroxymethyl-7,8-dihydropterin by sodium
dithionite as previously described.30
C
17H22N10O17P4, [M - H]-.
NMR a n d MS An a lyses. All NMR spectra were recorded
at 25 °C on a Varian INOVA 300 spectrometer. The samples
1
were dissolved in D2O. H chemical shifts were referenced to
the residual water resonance (δ ) 4.63 ppm). 31P chemical
shifts were referenced to external 85% phosphoric acid. FAB-
MS spectra were recorded on a J EOL J MS-HX 110 spectrom-
eter and 3-nitrobenzyl alcohol (NBA) was used as matrix.
MALDI-MS spectra were recorded in the negative mode on a
Perspective Biosystem Voyager DE spectrometer and 6-aza-
2-thiothymine (ATT) was used as matrix.
En zym e Assa y. E. coli HPPK was purified as previously
described.31 The inhibition assay was performed in a microtube
containing the following components in 100 mM Tris-HCl, pH
8.3: 0.1 mM 6-hydroxymethyl-7,8-dihydropterin, 1.5 mM ATP,
8 mM MgCl2, 5 mM DTT, 4.4 nM HPPK, a trace amount of
[R-32P]ATP, and various amounts of an inhibitor (Hp2A, Hp3A,
or Hp4A). The total volume was 200 µL. The reaction was
initiated by the addition of ATP at 30 °C. After 30 min, 50
µL of the reaction mixture was transferred to a new microtube
containing 50 µL 120 mM EDTA and mixed on a vortex. An
aliquot of the reaction mixture was then spotted onto a PEI-
cellulose plate. The plate was developed with 0.3 M KH2PO4.
The radioactivity of the plate was measured by a Phosphor-
Imager system (Molecular Dynamics Storm 820). IC50 was
obtained by fitting the data to a logistic equation by nonlinear
least-squares regression32 using the software Origin from
Microcal.
F lu or om etr ic Titr a tion . Protein and ligand stock solu-
tions were made in 100 mM Tris-HCl, pH 8.3, and their
concentrations were determined spectrophotometrically using
the following extinction coefficients: 21 600 M-1 cm-1 at 280
nm for HPPK and 7 000 M-1 cm-1 at 366 nm for the bisub-
strate inhibitors. A 2-mL diluted inhibitor solution in a
fluorometric cuvette was titrated with the protein stock
solution. The temperature was regulated at 25 °C by a
circulating water bath. Fluorescence was measured on a Spex
FluoroMax-2 fluorometer. The excitation wavelength and slit
were 364 and 3 nm, respectively, and the emission wavelength
and slit were 450 and 5 nm, respectively. A few HPPK
preparations showed some fluorescence at the excitation and
emission wavelengths. For these HPPK preparations, a control
experiment, in which a 2-mL buffer solution was titrated with
the protein solution, was performed. The control data was
subtracted from the titration data. The corrected titration data
was then analyzed by nonlinear least-squares regression using
the software Origin and the equation:
Syn th esis of 6-Hyd r oxym eth ylp ter in Mon op h osp h a te
(4). The method for the synthesis of 6-hydroxymethylpterin
pyrophosphate15 was adapted for the synthesis of 6-hydroxy-
methylpterin monophosphate. Briefly, phosphoric acid was
first dried under vacuum for 2 days and then melted and mixed
slowly with 6-hydroxymethylpterin at 60-65 °C in a glass-
stoppered flask wrapped with aluminum foil. The mixture was
stirred and heated at 60-65 °C for 4 h. The product was
adsorbed to active charcoal, extracted with 3 N ammonium
hydroxide and purified by chromatography on DEAE-cellulose.
3
1H NMR: δ 4.94 (d, J ) 7.3 Hz, 2H), 8.74 (s, 1H). 31P NMR:
δ 2.84 (s, 1P). FAB-MS (NBA): m/z ) 272.04; 272.03 for
C7H8N5O5P, [M - H]-.
Syn th esis of Bisu bstr a te An a logu es 5-7. The bisub-
strate analogues were synthesized by coupling adenosine
nucleotides to 6-hydroxymethylpterin monophosphate previ-
ously synthesized and activated with 1,1′-carbonyldiimid-
azole.14,16 Pyridinium form of Dowex 50W-X8 was obtained by
allowing the resin to stand with 50% aqueous pyridine.
Tributylammonium salts (1 mmol) of AMP, ADP, ATP, and
6-hydroxymethylpterin monophosphate were prepared by first
converting their sodium salts to pyridinium salts by chroma-
tography on a Dowex 50W-X8 column. The column was eluted
with 50% aqueous methanol. The eluent was evaporated under
reduced pressure to 5-10 mL, and then 1 mol equiv of
tributylamine/mol of phosphorus and 10-15 mL methanol
were added. After 30 min of stirring, the solution was
concentrated under vacuum. The residue was dried by re-
peated addition and evaporation of anhydrous pyridine fol-
lowed by two 10-mL portions of N,N-dimethylformamide
(DMF).
To synthesize the bisubstrate analogues, the anhydrous
tributylammonium salt of 6-hydroxymethylpterin monophos-
phate (0.1 mmol) and 1,1′-carbonyldiimidazole (0.5 mmol) both
dissolved in 1 mL dry DMF were mixed. After 2 h of stirring
at room temperature, 0.7 mmol methanol was added and 30
min later, the anhydrous tributylammonium salt of one of the
adenosine nucleotides (0.25 mmol) in a minimal volume of dry
DMF was added. After additional 24 h of stirring, the reaction
mixture was evaporated to dryness, dissolved in a minimal
volume of water and loaded on to a DEAE-cellulose column.
The column was eluted with a linear gradient (0-0.8 M) of
triethylammonium bicarbonate, pH 7.5. Appropriate fractions
were evaporated under vacuum; ethanol was added and
evaporated again to remove triethylammonium bicarbonate.
The residue was dissolved in a minimal volume of methanol,
and 7 mmol sodium perchlorate in 50 mL acetone was added.
The precipitated sodium salt was collected by centrifugation
and washed with acetone.
Fobs ) ꢀfLt +
(ꢀ - ꢀ )(L + E + K - (L + E + K )2 - 4E L )
x
b
f
t
t
d
t
t
d
t
t
2
where Fobs is the observed fluorescence, ꢀf and ꢀb are the
fluorescence coefficients of the ligand in the free and protein-
bound states, respectively, Lt is the total concentration of the
ligand, and Et is the total concentration of HPPK. Lt and Et
were varied during the titration process according to the
following expressions:
The results of NMR and MS analyses of the bisubstrate
analogues are summarized as follows. For HP2A, 1H NMR:
4.15-4.53 (m, 5H), 4.95 (d, 3J ) 4.4 Hz, 2H), 5.95 (d, 3J )
5.13 Hz, 1H), 8.04 (s, 1H), 8.12 (s, 1H), 8.39 (s, 1H). 31P NMR:
-11.04, (broad, 2P). MALDI-MS (ATT): m/z ) 601.589; 601.08
for C17H20N10O11P2, [M - H]-. For HP3A, 1H NMR: 4.20-4.52
(m, 5H), 4.94 (d, J ) 6.6 Hz, 2H), 5.95 (d, J ) 4.4 Hz, 1H),
8.00 (s, 1H), 8.17 (s, 1H), 8.41 (s, 1H). 31P NMR: -10.8, (broad,
1P), -10.94 (broad, 1P), -22.33 (broad, 1P). MALDI-MS
(ATT): m/z ) 680.275; 681.05 for C17H21N10O14P3, [M - H]-.
L0V0
Lt )
V0 + ∆V
E0∆V
Et )
V0 + ∆V
3
3
where E0 is the concentration of the HPPK stock solution, L0
is the initial concentration of the ligand, V0 is the initial

1370 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 9
Shi et al.
Ta ble 3. Summary of Crystallographic Data Collection,
Structure Solution, and Refinement for HPPK‚MgHP4A
final R factor and Rfree were 0.185 and 0.217, respectively, for
all data with their intensities equal to or above 2σ (Table 3).
The final electron density did not show more interpretable
features. The root-mean-square (rms) deviations for bond
lengths and angle distances were 0.006 and 0.021 Å, respec-
tively (Table 3). The structure has been assessed using
PROCHECK,39 showing that 91.2% of the residues are in the
most favored regions of the Ramachandran plot and no residue
falls into disallowed or generously allowed regions (see Table
3).
crystal shape
block
crystal dimensions (mm)
0.1 × 0.1 × 0.2
89.5 (77.6)
0.154 (0.338)
5.9 (2.1)
9368/12025
480/614
5668
overall (last shell)a completeness (%)
b
overall (last shell) Rsym
overall (last shell) I/σ(I)
no. reflns used for refinement: I2σ(I)/all
no. reflns used for Rfree: I2σ(I)/all
no. least-squares parameters
no. residues/(non-H) atoms
no. heterogen atoms
158/1267
50
Ack n ow led gm en t. We thank Dr. Nicholas V. Cozzi
for advice on IC50 measurements. This work was sup-
ported in part by NIH Grant GM51901 awarded to H.Y.
no. water oxygen atoms
167
R
free: I2σ(I)/all
0.217/0.242
0.185/0.212
0.006
R factor: I2σ(I)/all
rms deviations from ideal (Å): bond lengths
rms deviations from ideal (Å): angle distances 0.021
Su p p or t in g In for m a t ion Ava ila b le: Stereodiagrams
showing the structural alignments between HPPK‚MgHP4A
(this work) and HPPK‚HP‚MgAMPCPP18 and between HPPK‚
MgHP4A (this work) and apo-HPPK.26 This material is avail-
able free of charge via the Internet at http://pubs.acs.org.
estimated coordinate error (Å)
Ramachandran statistics: residues in
most favored regions (%)
0.18
91.2
Ramachandran statistics: residues in
additional allowed regions (%)
8.8
a
b
Resolution range 1.92-1.85 Å. Rsym ) Σ|I - I |/ΣI. Friedel
Refer en ces
pairs were merged.
(1) Cohen, M. L. Epideminology of drug resistance: implications for
a post-antimicrobial era. Science 1992, 257, 1050-1055.
(2) Neu, H. C. The crisis in antibiotic resistance. Science 1992, 257,
1064-1072.
volume of the titration, and ∆V is the total volume of the added
HPPK solution.
(3) Kunin, C. M. Resistance to antimicrobial drugs - a worldwide
calamity. Ann. Intern. Med. 1993, 118, 557-561.
(4) Levy, S. B. Antimicrobial resistance: A global perspective. Adv.
Exp. Med. Biol. 1995, 390, 1-15.
Cr ysta lliza tion of HP P K‚MgHP 4A. The crystals were
grown at room temperature (293 K) in hanging drops. The
protein stock solution contained 10 mg/mL HPPK, 35 mM
HP4A, 50 mM MgCl2 and 10 M Tris-HCl (pH 8.0); the reservoir
solution consisted of 28% (w/v) PEG 4000, 0.18 M ammonium
acetate, 0.15 M sodium acetate (pH 5.5), and 10% (v/v) glycerol.
The drops were composed with equal volumes of protein and
reservoir solutions. The crystals appeared within 3 days, and
after 1 week they were ready for X-ray data collection (Table
3).
(5) Murray, B. E. Antibiotic resistance. Adv. Intern. Med. 1997, 42,
339-367.
(6) World Health Organization Report on Infectious Diseases: Re-
moving Obstacles to Healthy Development; World Health Orga-
nization: Geneva, 1999.
(7) Shiota, T. Biosynthesis of folate from pterin precursors. Chem-
istry and Biochemistry of Folates; J ohn Wiley & Sons: New York,
1984; pp 121-134.
(8) Blakley, R. L.; Benkovic, S. J . Chemistry and Biochemistry of
Folates. Folates and Pterins; J ohn Wiley & Sons: New York,
1984.
(9) Hughes, D. T. D. Sulphonamides. Antibiotic and Chemotherapy,
7 ed.; Churchill Livingstone: New York, 1997; pp 460-468.
(10) Hughes, D. T. D. Diaminopyrimidines. Antibiotic and Chemo-
therapy, 7 ed.; Churchill Livingstone: New York, 1997; pp 346-
356.
(11) Talarico, T. L.; Dev, I. K.; Dallas, W. S.; Ferone, R.; Ray, P. H.
Purification and partial characterization of 7,8-dihydro-6-hy-
droxymethylpterin-pyrophosphokinase and 7,8-dihydropteroate
synthase from Escherichia coli MC4100. J . Bacteriol. 1991, 173,
7029-7032.
Da ta Collection , Str u ctu r e Solu tion , a n d Refin em en t.
The diffraction data were collected from a single crystal at a
cryogenic temperature (100 K; Oxford cryosystem), using a
Rigaku rotating anode operated at 50 kV and 100 mA with a
MAR345 image plate system. The crystal diffracted to 1.85 Å
resolution. Data processing was carried out with the HKL2000
program suite.33 The data set consisted of 12 732 unique
reflections from a total of 39 784 measurements and was 89.5%
complete. The internal R factor between symmetry-related
reflections was 0.15 (Table 3). The complex crystallized in the
orthorhombic space group P212121 with 1 HPPK‚HP4A complex
and 1 Mg2+ ion in the asymmetric unit. The unit-cell dimen-
sions were a ) 38.053 Å, b ) 61.857 Å, c ) 67.106 Å, and R )
â ) γ ) 90°. The solvent content of the crystals was 39.8%,
and the Matthews coefficient (Vm)34 was 2.1. The crystal
structure was solved by molecular replacement with the
program suite AmoRe.35 The search model was the crystal
structure of apo-HPPK26 without solvent molecules. The
solution from molecular replacement after rotational and
translational search consisted of one molecule and had a
correlation coefficient of 0.59 and a crystallographic R factor
of 0.38. Structure refinement involved 4 rounds of the simu-
lated annealing procedure of CNS36 and 14 rounds of least-
squares refinement using SHELXL-9737 for X-ray data in the
resolution range 40.0-1.85 Å. Model building during the initial
stage of refinement involved the adjustment of loop conforma-
tion and side chains according to electron density. Further
model building was focused on the positioning of the HP4A
molecule according to difference electron density above 3σ, and
the addition of solvent molecules was done at a later stage.
All model building and adjustments were done using the O
graphics package.38 The final model of the structure consisted
of all nonhydrogen atoms of the polypeptide chain (158
residues), 1 HP4A molecule, 1 Mg2+ ion, 1 Cl1- ion, and 167
water molecules. All protein atoms, inhibitor, ions and solvent
molecules had evident electron densities above 1σ on the 2Fo
- Fc map, except for the side chains of Arg84, Lys85, and
Glu87, which are located at the surface of the molecule. The
(12) Talarico, T. L.; Ray, P. H.; Dev, I. K.; Merrill, B. M.; Dallas, W.
S. Cloning, sequence analysis, and overexpression of Escherichia
coli folK, the gene coding for 7,8-dihydro-6-hydroxymethylpterin-
pyrophosphokinase. J . Bacteriol. 1992, 174, 5971-5977.
(13) Switzer, R. L. Phosphoribosylpyrophosphate synthetase and
related pyrophosphokinases. The Enzymes, 3rd ed.; Academic
Press: New York, 1974; pp 607-629.
(14) Hoard, D. E.; Ott, D. G. Conversion of mono- and oligodeoxy-
ribonucleotides to 5′-triphosphates. J . Am. Chem. Soc. 1965, 87,
1785-1788.
(15) Ho, R. I.; Corman, L.; Foye, W. O. Synthesis and biological
evaluation of 2-amino-4-hydroxy-6-hydroxymethylpteridine py-
rophosphate. J . Pharm. Sci. 1974, 1472-1476.
(16) Bone, R.; Cheng, Y. C.; Wolfenden, R. Inhibition of thymidine
kinase by P1-(adenosine-5′)-P5-(thymidine-5′)-pentaphosphate. J .
Biol. Chem. 1986, 261, 5731-5735.
(17) Davies, L. C.; Stock, J . A.; Barrie, S. E.; Orr, R. M.; Harrap, K.
R. Dinucleotide analogues as inhibitors of thymidine kinase,
thymidylate kinase, and ribonucleotide reductase. J . Med. Chem.
1988, 31, 1305-1308.
(18) Blaszczyk, J .; Shi, G.; Yan, H.; J i, X. Catalytic center assembly
of 6-hydroxymethyl-7,8-dihydropterin pyrophosphokinase as
revealed by the crystal structure of a ternary complex at 1.25 Å
resolution. Structure 2000, 8, 1049-1058.
(19) Broom, A. D. Rational Design of Enzyme-Inhibitors - Multi-
substrate Analogue Inhibitors. J . Med. Chem. 1989, 32, 2-7.
(20) Radzicka, A.; Wolfenden, R. Transition state and multisubstrate
analogue inhibitors. Methods Enzymol. 1995, 249, 284-312.
(21) Liu, M.; Haddad, J .; Azucena, E.; Kotra, L. P.; Kirzhner et al.
Tethered bisubstrate derivatives as probes for mechanism and
as inhibitors of aminoglycoside 3′-phosphotransferase. J . Org.
Chem. 2000, 65, 7422-7431.

Bisubstrate Analogue Inhibitors of HPPK
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 9 1371
(22) Lienhard, G. E.; Secemski, I. I. P1,P5-di(adenosine-5′)pentaphos-
phate, a potent multisubstrate inhibitor of adenylate kinase. J .
Biol. Chem. 1973, 248, 1121-1123.
(31) Shi, G.; Gao, J .; Yan, H. 1H, 13C and 15N resonance assignments
of Escherichia coli 6-hydroxymethyl-7,8-dihydropterin pyrophos-
phokinase and its complex with MgAMPPCP. J . Biomol. NMR
1999, 14, 189-190.
(32) Graeser, D.; Neubig, R. R. Methods for the study of receptor/G-
protein. Signal Transduction; IRL Press: Oxford, 1992; pp 1-30.
(33) Otwinowski, Z.; Minor, W. Processing of X-ray diffraction data
collected in oscillation mode. Methods Enzymol. 1997, 276, 307-
326.
(34) Matthews, B. W. Solvent content of protein crystals. J . Mol. Biol.
1968, 33, 491-497.
(35) Navaza, J . An automated package for molecular replacement.
Acta Crystallogr. 1994, A50, 157-163.
(36) Bru¨nger, A. T.; Adams, P. D.; Clore, G. M.; DeLano, W. L.; Gros,
P. et al. Crystallography and NMR system (CNS): A new
software system for macromolecular structure determination.
Acta Crystallogr. 1998, D54, 905-921.
(37) Sheldrick, G. M.; Schneider, T. R. SHELXL: high-resolution
refinement. Methods Enzymol. 1997, 277, 319-343.
(38) J ones, T. A.; Zou, J . Y.; Cowan, S. W.; Kjeldgaard, M. Improved
methods for building protein models in electron density maps
and the location of errors in these models. Acta Crystallogr. 1991,
A47, 110-119.
(39) Laskowski, R. A.; MacArthur, M. W.; Moss, D. S.; Thornton, J .
M. PROCHECK: a program to check the stereochemical quality
of protein structures. J . Appl. Crystallogr. 1993, 26, 283-291.
(40) Esnouf, R. M. An extensively modified version of MolScript that
includes greatly enhanced coloring capabilities. J . Mol. Graph
Model. 1997, 15, 132-134.
(23) Bone, R.; Cheng, Y. C.; Wolfenden, R. Inhibition of adenosine
and thymidylate kinases by bisubstrate analogues. J . Biol.
Chem. 1986, 261, 16410-16413.
(24) Danenberg, P. V.; Danenberg, K. D. Inhibition of hexokinase by
multisubstrate analogues. Biochim. Biophys. Acta 1977, 480,
351-356.
(25) Shi, G.; Gong, Y.; Savchenko, A.; Zeikus, J . G.; Xiao, B. et al.
Dissecting the nucleotide binding properties of Escherichia coli
6-hydroxymethyl-7,8-dihydropterin pyrophosphokinase with fluo-
rescent 3′(2)′-o-anthraniloyladenosine 5′-triphosphate. Biochim.
Biophys. Acta 2000, 1478, 289-299.
(26) Xiao, B.; Shi, G.; Chen, X.; Yan, H.; J i, X. Crystal Structure of
6-Hydroxymethyl-7,8-dihydropterin Pyrophosphokinase, a Po-
tential Target for Development of Novel Antimicrobial Agents.
Structure 1999, 7, 489-496.
(27) Stammers, D. K.; Achari, A.; Somers, D. O.; Bryant, P. K.;
Rosemond, J . et al. 2.0 Å X-ray structure of the ternary complex
of 7,8-dihydro-6-hydroxymethylpterinpyrophosphokinase from
Escherichia coli with ATP and a substrate analogue. FEBS Lett.
1999, 456, 49-53.
2-
(28) Hengge, A. C. Transfer of the PO3 group. Comprehensive
Biological Catalysis; Academic Press: San Diego, 1998; pp 517-
542.
(29) Thijjssen, H. H. W. A simple method for preparing 2-amino-4-
hydroxy-6-formylpteridine, a precursor of the pteridine substrate
of dihydropteroate biosynthesis. Anal. Biochem. 1973, 54, 609-
611.
(30) Scrimgeour, K. G. Methods for reduction, stabilization, and
analyses of folates. Methods Enzymol. 1980, 66, 517-523.
J M0004493