Lys314 is a nucleophile in non-classical reactions of orotidine-5′- monophosphate decarboxylase

Article abstract of DOI:10.1002/chem.200900397

Orotidine-5′-monophosphate decarboxylase (OMPD) catalyzes the decarboxylation of orotidine-5′-monophosphate (OMP) to uridine-5′-monophosphate (UMP) in an extremely proficient manner. The reaction does not require any cofactors and proceeds by an unknown mechanism. In addition to decarboxylation, OMPD is able to catalyze other reactions. We show that several C6-substituted UMP derivatives undergo hydrolysis or substitution reactions that depend on a lysine residue (Lys314) in the OMPD active site. 6-Cyano-UMP is converted to UMP, and UMP derivatives with good leaving groups inhibit OMPD by a suicide mechanism in which Lys314 covalently binds to the substrate. These non-classical reactivities of human OMPD were characterized by cocrystallization and freeze-trapping experiments with wildtype OMPD and two active-site mutants by using substrate and inhibitor nucleotides. The structures show that the C6-substituents are not coplanar with the pyrimidine ring. The extent of this substrate distortion is a function of the substituent geometry. Structurebased mechanisms for the reaction of 6-substituted UMP derivatives are extracted in accordance with results from mutagenesis, mass spectrometry, and OMPD enzyme activity. The Lys314based mechanisms explain the chemodiversity of OMPD, and offer a strategy to design mechanism-based inhibitors that could be used for antineoplastic purposes for example.

Full text of DOI:10.1002/chem.200900397

FULL PAPER
DOI: 10.1002/chem.200900397
Lys314 is a Nucleophile in Non-Classical Reactions of Orotidine-5’-
Monophosphate Decarboxylase
Daniel Heinrich,[a] Ulf Diederichsen,*[a] and Markus Georg Rudolph*[b]
Abstract: Orotidine-5’-monophosphate
decarboxylase (OMPD) catalyzes the
decarboxylation of orotidine-5’-mono-
phosphate (OMP) to uridine-5’-mono-
phosphate (UMP) in an extremely pro-
ficient manner. The reaction does not
require any cofactors and proceeds by
an unknown mechanism. In addition to
decarboxylation, OMPD is able to cat-
alyze other reactions. We show that
several C6-substituted UMP deriva-
tives undergo hydrolysis or substitution
reactions that depend on a lysine resi-
due (Lys314) in the OMPD active site.
6-Cyano-UMP is converted to UMP,
and UMP derivatives with good leaving
groups inhibit OMPD by a suicide
mechanism in which Lys314 covalently
binds to the substrate. These non-clas-
sical reactivities of human OMPD were
characterized by cocrystallization and
freeze-trapping experiments with wild-
type OMPD and two active-site mu-
tants by using substrate and inhibitor
nucleotides. The structures show that
the C6-substituents are not coplanar
with the pyrimidine ring. The extent of
this substrate distortion is a function of
the substituent geometry. Structure-
based mechanisms for the reaction of
6-substituted UMP derivatives are ex-
tracted in accordance with results from
mutagenesis, mass spectrometry, and
OMPD enzyme activity. The Lys314-
based mechanisms explain the chemo-
diversity of OMPD, and offer a strat-
egy to design mechanism-based inhibi-
tors that could be used for antineoplas-
tic purposes for example.
Keywords: decarboxylation
·
en-
zymes · hydrolysis · nucleophilic
substitution · nucleotides
Introduction
plexes of OMPD from various organisms and a large body
of kinetic evidence, a chemical mechanism of OMP decar-
boxylation has not been agreed on.[2] The active form of
OMPD from bacterial, fungal, and mammalian sources is a
homodimeric TIM-barrel without any cofactors or metal
ions. Important residues in the active site of OMPD are two
aspartic acids and a lysine residue (Figure 1). Comparison of
the UMP complexes of OMPD from human Methanobacte-
rium thermoautotrophicum and Plasmodium falciparum
shows that whereas the key active-site residues are con-
served, the architecture of the active site and the conforma-
tion of the catalytic side-chains can differ markedly (Sup-
porting Information, Figure S0). The active-site residues
Asp312 and Lys314 (human OMPD numbering) are of par-
ticular interest and have been implicated in exerting electro-
static stress on the substrate and in functioning as a proton
donor for an intermediate carbanion, respectively.[3,4]
Orotidine-5’-monophosphate decarboxylase (OMPD) cata-
lyzes the last step of de novo pyrimidine nucleotide biosyn-
thesis, the decarboxylation of orotidine-5’-monophosphate
(OMP, 1) to uridine-5’-monophosphate (UMP, 2). OMPD
accelerates the decarboxylation reaction by 1017-fold com-
pared to the uncatalyzed reaction, and thus constitutes one
of the most proficient enzymes known.[1] Despite a plethora
of crystal structures of substrate, product, and inhibitor com-
[a] Dr. D. Heinrich, Prof. Dr. U. Diederichsen
Institut fꢀr Organische und Biomolekulare Chemie
Georg-August-Universitꢁt Gçttingen, Tammannstrasse 2
37077 Gçttingen (Germany)
Fax : (+49)551-39-22944
E-mail: udieder@gwdg.de
Because the progress of cancer relies on the rapid synthe-
sis of DNA,[5,6] inhibition of enzymes of de novo nucleotide
biosynthesis pathways, such as OMPD, is an attractive objec-
tive for the development of antineoplastic drugs. OMPD
shows unusual chemodiversity, that is, it converts non-natu-
ral nucleotides into 2 or other products by substitution, hy-
drolysis, and decarboxylation reactions that have no coun-
[b] Dr. M. G. Rudolph
Department of Molecular Structural Biology
Georg-August-Universitꢁt Gçttingen and
F. Hoffmann-La Roche, Basel (Switzerland)
Fax : (+41)61-267-2189
E-mail: Markus.Rudolph@bio.uni-goettingen.de
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200900397.
Chem. Eur. J. 2009, 15, 6619 – 6625
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
6619

Results and Discussion
Crystal structures of OMPDs in
complex with uridine nucleo-
tides consistently show the syn
conformation of the ligand; this
exposes the C5- and C6-posi-
tions of the nucleobase to
Lys314. Lys314 is a known key
catalytic residue, and ligand
design thus focused on C5 and
C6-modified UMP nucleotides
that could react with this poten-
tial nucleophile (Figure 1).[3a,7,11]
Figure 1. Left: Nucleotides that were investigated as OMPD substrates or inhibitors: OMP (1), UMP (2),
BMP (3), 6-cyano-UMP (4), 6-acetyl-UMP (5), 5-cyano-UMP (6), 6-azido-UMP (7), 6-mercapto-UMP (8). RP
denotes the phosphoribosyl residue. Right: The OMPD active-site cavity is shown in gray with its chemical en-
vironment shown as stick representation and the nucleotide 6-cyano-UMP (4) bound.[3a] Residues in proximity
to the nucleobase are labeled. The catalytic residues Lys314 and Asp312 are highlighted in yellow and Asp317
(blue) is contributed from the second subunit.
Substrate distortion, hydrolysis,
and decarboxylation of 6-
cyano-UMP: For yeast OMPD
6-cyano-UMP (4) was described
as a weakly competitive (Kd =
terpart in vivo. This chemodiversity might be exploited for
drug development. OMPD does not tolerate substantial var-
iations at the 5’-phosphoribofuranose moiety, which is re-
quired both for high-affinity binding and for substrate con-
version.[2a,7] In contrast, OMPD accepts a diverse set of pyri-
midine-modified nucleotides, many of which are high-affini-
ty competitive inhibitors, or even exhibit time-dependent co-
valent inhibition.[5a,8,9] For example, OMPD from
M. thermoautotrophicum engages in suicide inhibition by nu-
cleophilic aromatic substitution of 6-iodo-UMP by the es-
sential Lys72 (Lys314 in the human enzyme).[7] In addition,
M. thermoautotrophicum OMPD catalyzes the conversion of
6-cyano-uridine-5’-monophosphate (6-cyano-UMP, 4) to bar-
bituric acid 5’-monophosphate (BMP, 3).[10] We found that
human OMPD is also covalently inhibited by 6-iodo-UMP,
but report here different reaction products for the conver-
sion of 4.[5a] In humans, 4 is converted to UMP (2) and not
BMP (3); this suggests different reaction mechanisms for
the two enzymes.
To determine the structural basis for this different behav-
ior and to further probe the determinants for the chemodi-
versity of OMPD, we synthesized several 5- and 6-substitut-
ed UMP derivatives for interaction with wild-type (wt)
human OMPD. Two OMPD mutants, Asp312Asn and
Lys314Ala, were generated to abolish enzyme activity and
trap reaction intermediates. Crystal structures of wt-OMPD
that were cocrystallized or briefly soaked with 6-cyano-
UMP provided information on the product UMP and a bent
reaction intermediate, respectively. Structures of the mu-
tants resulted in unexpected products and revealed that
Lys314 exhibits increased nucleophilicity in the Asp312Asn
mutant compared to wt-OMPD. Furthermore, the
Lys314Ala mutant is still active for hydrolysis and decarbox-
ylation as exemplified by a cocrystal structure with 6-cyano-
UMP that resulted in a UMP complex. Understanding the
details of such atypical OMPD reactions might give insight
into the primary chemical mechanism of OMP decarboxyla-
tion.
11 mm) and chemically stable inhibitor.[12] Based on the CN
stretching frequency, the cyano group was proposed to be
slightly bent out of the plane of the pyrimidine ring by ap-
proximately 208.[9] In contrast, 6-cyano-UMP is known to in-
hibit wt-OMPD from M. thermoautotrophicum after its
transformation to BMP 3, which is the most efficient com-
petitive OMPD inhibitor known to date.[8] When human
OMPD was cocrystallized with 6-cyano-UMP, the resulting
crystal structure surprisingly showed the presence of UMP
(2) in the active site (data not shown). Within at most a few
days 6-cyano-UMP (4) was converted to UMP (2) by human
OMPD and not to BMP (3). In both cases, an initial cova-
lent addition of Lys314 can explain the obtained product
(Scheme 1).
In principle, involvement of a lysine residue is not neces-
sary. BMP (3) could be generated directly from 6-cyano-
UMP (4) by substitution with water in the M. thermoauto-
trophicum active site. Likewise, 4 could be hydrolyzed to
OMP and then decarboxylated by human OMPD. However,
the higher nucleophilicity of primary amines compared to
water should favor a pathway involving lysine. Also, the 6-
cyano-uridine moiety is stable in aqueous solution for sever-
al days (unpublished observation). For human OMPD, the
likely route is Lys314-assisted hydrolysis of the nitrile 4 to
the carboxylic acid OMP 1 via covalent intermediates, fol-
lowed by normal decarboxylation. The nitrile is thus a
masked carboxylic acid, similar to the use of nitriles in or-
ganic synthesis. Because hydrolysis of 4 is a prerequisite for
UMP (2) formation, the active site of human OMPD should
allow solvent to enter at later stages during catalysis. The
apparent discrepancy in product distribution between the
M. thermoautotrophicum and human enzymes might be as-
cribed to species differences of the active sites. For instance,
the active-site residues Asn341 and His283 in human
OMPD are alanine and glycine, respectively, in M. ther-
moautotrophicum OMPD.[8] These smaller side-chains might
change the active-site polarity, leading to a different mecha-
6620
www.chemeurj.org
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 6619 – 6625

OMPD Reactivity
FULL PAPER
Scheme 1. Covalent mechanisms by using Lys314 as a nucleophile can explain the formation of BMP and UMP from 6-cyano-UMP.
nism, or allow water to more easily enter the active site
during catalysis and alter the sequence of reactions.
trile nitrogen atom (Figure 2). For reasons of conjugation,
the cyano group in cyano-UMP nucleotides 4 and 6 is ex-
pected to be coplanar with the aromatic pyrimidine ring. By
contrast, in complex with OMPD the 6-cyano group is bent
out of the ring plane by 578, which is much more than antici-
pated from the CN stretching frequency.[9] Similar substrate
distortion has been observed for OMP in the crystal struc-
tures of inactivated human and M. thermoautotrophicum
OMPDs in which the carboxylate groups are also kinked
from the pyrimidine ring planes.[3,8]
To evaluate the hypothesis of 6-cyano-UMP hydrolysis
prior to decarboxylation and to test if Lys314 acts as a nu-
cleophile in this reaction, two strategies were employed.
First, short-time soaks of apo-enzyme crystals with 6-cyano-
UMP were performed to trap an enzyme–substrate complex.
Second, active-site OMPD mutants that were expected to
abolish or reduce activity were generated. We have shown
earlier that the human Asp312Asn OMPD variant is inac-
tive with respect to decarboxylation.[3a] In addition, the
Lys314Ala mutant lacking the nucleophilic group was gener-
ated, thus serving as a negative control. The cocrystal struc-
ture of wt human OMPD with 6-cyano-UMP, obtained by a
25 min soak of the substrate with apo-crystals, revealed
intact nucleotide bound in the active site (Figure 2). A
longer soak of 60 min already showed significant loss of the
cyano group; this puts the timescale of hydrolysis–decarbox-
ylation reaction within hours at ambient temperature. In the
structure, a water molecule occupies the position of the ni-
Quite surprisingly, in the Asp312Asn mutant 6-cyano-
UMP forms a covalent OMPD–substrate complex. Lys314 is
linked to the cyano moiety through an amidine or amide
(Figure 2). MALDI-TOF spectra of a mixture of OMPD
and 6-cyano-UMP confirmed the crystallographic finding of
a covalent complex (Supporting Information, Figure S1–S3).
Despite the current atomic resolution (1.1 ꢃ) of the
Asp312Asn OMPD–6-cyano-UMP complex it cannot be de-
termined whether the structure represents an amidine 11 or
an amide 12 (Scheme 1) due to a lack of electron density for
polar hydrogen atoms around the amidine/amide group.
Burial of a positively charged amidine is disfavored because
the Asp312Asn mutant lacks a compensating negative
charge. An amide would therefore result in a more stable
OMPD complex, but in the absence of more data the ami-
dine was modeled as it is the simpler adduct. The compari-
son of the reaction products from wt-OMPD and the
Asp312Asn mutant raises the intriguing questions as to why
a similar adduct was not obtained with wt-OMPD and
where the water molecules that are required for hydrolysis
come from. Once the amidine 11 is formed in wt-OMPD, it
is converted to UMP within a matter of hours. Therefore,
the active site in wt-OMPD must be accessible to water
molecules from bulk solvent. The observation of a water
molecule close to the C6-position in the OMPD–6-cyano-
UMP product complex indicates that solvent might enter
the OMPD active site during catalysis. Apparently the cova-
lent complex in the Asp312Asn mutant is more stable than
in wt-OMPD. If Asp312 acts as a general base for deproto-
nation of water to attack the amidine or amide bonds
formed with Lys314, then the covalent intermediates were
too unstable to be observed crystallographically in wt-
Figure 2. Cocrystal structures of the reaction products of 6-cyano-UMP
with OMPD. Top left: wt-OMPD leading to substrate distortion. Top
right: wt-OMPD with UMP that was formed from 6-cyano-UMP. Bottom
left: OMPD Asp312Asn reacts with 6-cyano-UMP to form a covalent
amidine complex. Bottom right: OMPD Lys314Ala degrades 6-cyano-
UMP to UMP. Possible hydrogen bonds (<3.2 ꢃ) are shown as dashed
blue lines, and unfavorable contacts are highlighted with dashed orange
lines. Note the water-mediated hydrogen bond network in the Lys314Ala
mutant.
Chem. Eur. J. 2009, 15, 6619 – 6625
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6621

U. Diederichsen, M. G. Rudolph and D. Heinrich
OMPD. By contrast, removal of the general base in the
Asp312Asn mutant would result in a kinetic trap, enabling
long-term population of the covalent intermediate.
After it was established that Lys314 acts as a nucleophile
and water can access the OMPD active site during degrada-
tion of 6-cyano-UMP, the next question was what effects a
Lys314Ala mutation would have on this reaction? Elimina-
tion of the nucleophile might be expected to result in an au-
thentic and stable OMPD–6-cyano-UMP complex. Instead,
the UMP complex was observed, similar to the result with
wt-OMPD (Figure 2). Residual decarboxylase activity in a
M. thermoautotrophicum OMPD mutant lacking the lysine
has been reported.[13] The human Lys314Ala OMPD mutant
also displays residual decarboxylase activity: a 1.1 ꢃ resolu-
tion structure, which was obtained from cocrystallization of
OMP with the K314A mutant, shows UMP only (data not
shown because the outcome is the same as in the right panel
of Figure 2). Apparently the functionality of the lysine resi-
due, regardless of its involvement in decarboxylation or hy-
drolysis reactions, can be replaced by other parts of the
enzyme or water. The residual activity of the Lys314Ala
mutant is sufficient to slowly convert 6-cyano-UMP to UMP,
but must involve a different mechanism than wt-OMPD.
The likely course of action is mediated by water. Removal
of the lysine side-chain leaves a cavity that is filled with
water molecules that establish a hydrogen-bond network
(Figure 2). Again, Asp312 might act as general base that ac-
tivates water to convert the nitrile to the carboxylic acid.
Another candidate for a general base is Asp317 (Figure 1).
In summary, Lys314 is not absolutely required for decarbox-
ylation and hydrolysis reactions of human OMPD. However,
compared to water Lys314 is much more efficient in catalyz-
ing these reactions.
Figure 3. The OMPD mutant Lys314Ala converts 6-azido-UMP to BMP
(middle left). A possible reaction scheme for this transformation is given
at the top. The structure of wt-OMPD in complex with BMP (middle
right) is similar to the Lys314Ala-BMP complex. A covalent adduct of
UMP and Lys314 is seen in the Asp312Asn–6-azido-UMP complex
(bottom left) and also in the wt-OMPD–6-azido-UMP complex (bottom
right).
position adopts the role of the nucleophile with Asp312 and
Asp317 as possible general bases. Both wt and Asp312Asn
OMPD react with 6-azido-UMP (7) to form covalent bonds
with the C6-position of UMP; this directly displaces azide
(Figure 3). Taken together, the studies with 6-azido-UMP
underscore the conclusions that were derived from 6-cyano-
UMP. The azide moiety in 6-azido-UMP constitutes such a
good leaving group that any nucleophile within the confine-
ments of the OMPD protein will displace it. This result is in
agreement with the analogous covalent inhibition of M. ther-
moautotrophicum OMPD by 6-azido-UMP.[7]
Differential reactivity of 6-azido-UMP with OMPD mu-
tants: To exclude singular observations that are peculiar to
6-cyano-UMP, an additional set of crystal structures with an-
other nucleotide was essential. The azido and cyano groups
are roughly isosteric, and both contain an electrophilic
center. 6-Azido-UMP (7) is a covalent (suicide) inhibitor of
Plasmodium falciparum OMPD.[7] 6-Azido-UMP was there-
fore used to address the question whether substrate distor-
tion and reactivities similar to 6-cyano-UMP are also ob-
served with this nucleotide, and whether these reactivities
are dependent on the catalytic Lys314. Cocrystal structures
of 6-azido-UMP were determined with wt-OMPD and the
Asp312Asn and Lys314Ala mutants (Figure 3). The cocrys-
tal structure of 6-azido-UMP and Lys314Ala OMPD re-
vealed BMP as the product. Similar to the case of 6-cyano-
UMP (see above), removal of the nucleophilic lysine is in-
sufficient to arrest chemistry but alters the reaction pathway.
A hydroxyl group is substituted for the azide moiety, and a
water molecule that is present at the position of the former
Lys314 could act as a nucleophile. Comparison of this struc-
ture with the wt-OMPD–BMP complex (Figure 3) revealed
little conformational differences. As discussed for 6-cyano-
UMP, in the absence of Lys314 water located close to this
6-Acetyl-UMP reveals a higher nucleophilicity of Lys314 in
Asp312Asn versus wt-OMPD: From the observed covalent
complexes and water-induced reaction products with the 6-
cyano-UMP and 6-azido-UMP nucleotides, it can be con-
cluded that the OMPD active site has a propensity to allow
for nucleophilic substitutions. The question arises as to
which residues in the binding site might determine the nu-
cleophilicity? To address this issue, 6-acetyl-uridine-5’-
monophosphate (6-acetyl-UMP, 5) was synthesized and re-
acted with wt-OMPD and the Asp312Asn variant. 6-Acetyl-
UMP is isosteric with the substrate OMP, but cannot decom-
pose to UMP. Similar to 6-cyano-UMP and 6-azido-UMP,
the carbonyl group in 6-acetyl-UMP is electrophilic, but not
prone to hydrolysis or substitution. This ligand is, therefore,
suited to detect differences in nucleophilicity without being
engaged in side-reactions such as hydrolysis.
The complex of wt-OMPD with 6-acetyl-UMP revealed
an intact but distorted nucleotide, similar to the distortion
6622
www.chemeurj.org
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 6619 – 6625

OMPD Reactivity
FULL PAPER
of OMP that was observed in the OMPD–Asp312Asn com-
plex.[3] The acetyl group is tilted out of the plane of the pyri-
midine ring, but no covalent interaction with Lys314 is pres-
ent (Figure 4). The carbonyl group interacts through a hy-
or substrate distortions by OMPD were limited to function-
alities at the C6-position of the pyrimidine ring, the regioiso-
mer 5-cyano-UMP (6) was chosen as a probe. 5-Cyano-
UMP is a close homologue to 6-cyano-UMP and should ex-
perience similar forces upon binding to OMPD as 6-cyano-
UMP. However, in the cocrystal structure with wt-OMPD,
5-cyano-UMP is intact and displays canonical geometry
(Figure 5). In stark contrast to 6-cyano-UMP, no substrate
Figure 4. 6-Acetyl-UMP (5) in the active site of wt-OMPD (left) and co-
valently bound to Nz of Lys314 of OMPD Asp312Asn (right).
Figure 5. Comparison of 5-cyano-UMP with 6-cyano-UMP. In case of the
5-cyano moiety no distortion of the substituent is present (left), whereas
the cyano moiety in the wt-OMPD–6-cyano-UMP complex shows a
strong distortion from planarity with the aromatic pyrimidine ring (right
and see Figure 2).
drogen bond with the Lys314 side-chain. By contrast, the
cocrystal structure of the Asp312Asn mutant and 6-acetyl-
UMP clearly shows a covalent bond between the carbonyl
group and the Nz atom of Lys314, which forms an imine
(Figure 4). The geometry of this adduct is very similar to the
amidine that is formed with 6-cyano-UMP (Figure 2). The
difference in reactivity of Lys314 in wt-OMPD and the
Asp312Asn mutant stems from the removal of the negative-
ly charged aspartate and the change in hydrogen-bonding
capabilities by the introduced asparagine. In the wt struc-
ture, Asp312 forms hydrogen bonds to both Lys314 and
Asn317. The direct contact with Lys314 leads to a charged
hydrogen bond with the protonated amino group devoid of
any nucleophilicity in the ground state. Upon mutation to
asparagine, the hydrogen bond to Lys314 is retained but
weakened due to a rotation of the Asn312 carboxamide
group by 908. The loss of a negative charge will reduce the
pKa and increase the nucleophilicity of Lys314, which will
proceed to form a covalent complex.
Why is 6-cyano-UMP but not 6-acetyl-UMP a substrate
for wt-OMPD? Molecular crowding of Lys314 with the C6
substituent would be expected to lead to a fast addition re-
action that is mainly entropy-driven without significant sub-
strate specificity. The OMPD–6-acetyl-UMP structure shows
an antiparallel orientation of the e-amino group of Lys314
and the carbonyl group in 6-acetyl-UMP. Nucleophilic
attack requires approach of Lys314 along the Bꢀrgi–Dunitz
trajectory at an angle of 1078, which is impossible for the
stereochemistry in the structure.[14,15] Such a limitation does
not exist for a nitrile with rotational symmetry around the
electrophile. The hydrogen bond of Lys314 to the carbonyl
group of 6-acetyl-UMP is another possibility for reduced nu-
cleophilicity, but there is no analogous interaction of Lys314
in the OMPD–6-cyano-UMP structure. Whereas the slow re-
action of Lys314 with electrophiles will depend on molecular
crowding in the OMPD active site, additional stereochemi-
cal and electrostatic requirements need to be fulfilled for a
reaction to occur.
distortion is present in 5-cyano-UMP, and the cyano group
is coplanar with the pyrimidine ring. The nitrile displaces
Lys314 by about 2.5 ꢃ and fits well into a hydrophobic
pocket that is lined by Ile368, Met371, and Ile401.[3] Thus, in
contrast to the 6-position, the 5-position of pyrimidine nu-
cleotides is not affected by ligand strain; this indicates that
OMPD exerts ligand strain in a very localized fashion spe-
cifically onto the 6-position.[16]
The question arises whether the nature of the C6 substitu-
À
ent determines the extent of ligand strain. The C1 C6 tor-
sion angle is a measure of the degree of tilting of the C6
substituent from the pyrimidine ring plane. A survey of the
available human OMPD structures complexed to a C6-
modified UMP derivative shows that this torsion angle
seems to correlate with the van der Waals volume and struc-
ture of the C6 substituent. For instance, in the atomic reso-
lution (1.24 ꢃ) wt-OMPD–BMP structure, the C1-C6 tor-
sion angle is only 2.48, whereas this angle increases to 26.08
for the homologous 6-mercapto-UMP (8) (Figure 6 and Sup-
Figure 6. Distortion of substrates in wt (top) and Asp312Asn OMPD
(bottom). Top from left to right: BMP, 6-mercapto-UMP, 6-acetyl-UMP,
6-cyano-UMP. Bottom: 6-Hydroxymethyl-UMP, OMP, 6-acetyl-UMP, 6-
À
cyano-UMP. The view is along the glycosidic bond and the C1 C6 torsion
angle is visible as the deviation from planarity as indicated by the dashed
lines for 6-mercapto-UMP. The distortion in degrees and the difference
in volumes of the undistorted ligands relative to UMP are given.
Regiospecificity and molecular determinants of substrate
distortion: To investigate whether chemical transformations
Chem. Eur. J. 2009, 15, 6619 – 6625
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6623

U. Diederichsen, M. G. Rudolph and D. Heinrich
porting Information Figure S4). Despite comparable vol-
umes, the 6-cyano group is bent twice as strongly (578) as
the mercapto group, possibly due to the longer “lever” and
less polarizability of the nitrile compared to the thiol. The
more flexible hydroxymethyl group, having the same
volume as the cyano and mercapto group, is least distorted.
Finally, comparison of the substituents acetyl and carboxyl
that have similar shapes shows a comparable degree of dis-
tortion (<408), indicating that charge does not influence the
amount of ligand strain. In summary, it appears that ligand
strain is maximal for large, non-polarizable, non-flexible
substituents that extend far away from the pyrimidine ring.
This information could prove useful for inhibitor design. By
introducing a chiral center at C6 that carries a substituent
that is prone to strong deformation in the context of a pyri-
midine, the deformation energy would be gained as binding
affinity.
Straining of small molecules by OMPD requires signifi-
cant energy that might be stored either in the enzyme, or
paid for by low binding affinities of the ligands. The Ki
values of the nucleotides that do not form covalent com-
plexes with wt-OMPD are all in the high mm range
(Table 1). Ligand strain might be the reason for these low
affinities, but UMP, which lacks a 6-substituent, also displays
is indeed observed in the crystal structures. Upon covalent
attachment of Lys314 in the Asp312Asn mutant to either
the acetyl or cyano substituent, ligand strain is released. The
torsion angles are only 5–108 and thus close to the situation
of the stable BMP complex (Figure 6). For M thermoauto-
trophicum OMPD, electrostatic stress exerted by Asp70
(Asp312 in human OMPD) was suggested to prime OMP
for decarboxylation.[3] Due to such electrostatic ground-state
destabilization, OMPD is expected to exhibit affinity for nu-
cleotides with positively charged C6 substituents. However,
binding of 6-methylamino-UMP was found to occur in the
neutral state, which would argue against electrostatic desta-
bilization of the enzyme–substrate complex.[18] Although the
exact mechanism of ligand strain and its relevance for
OMPD catalysis is not known, there are now enough struc-
tural examples of OMPD complexes with strained ligands to
assign ligand strain as a general property of OMPD. In this
light, the removal of the kinked OMP carboxylate group
might constitute an important driving force for decarboxyla-
tion.
Whereas substrate distortion is a genuine property of
OMPD, it is not limited to this enzyme. For instance, b-gly-
cosidases deform their substrates to achieve a pseudoaxial
orientation of the scissile bond.[19] In the crystal structure of
the B. subtilis pyrophosphatase the pyrophosphate analogue
[20]
À
PNP is deformed and the P N bond is extended. Finally,
Table 1. Approximate IC50 values for inhibitors of human OMPD at
258C.
FTIR spectroscopy revealed substantial strain of the lactam
carbonyl group in b-lactamase enzyme–substrate com-
plexes.[21] A common feature of these examples is a saturat-
ed or more flexible substrate compared to the rigid pyrimi-
dine nucleotide of OMPD. Distorting an aromatic ligand as
seen in most OMPD crystal structures that are described
here is expected to require considerably more energy than
distorting a saturated or more flexible ligand, and it will be
interesting to address the molecular mechanism leading to
such unusual distortions.
Inhibitor
IC50 (mM)
Inhibitor
IC50 (mM)
6-mercapto-UMP
6-acetyl-UMP
6-cyano-UMP[a]
5-cyano-UMP
1680Æ540
24Æ13
6-iodo-UMP[b]
25Æ5
36Æ17
–
6-azido-UMP[b]
ca. 200
6-hydroxymethyl-UMP[c]
79Æ36
[a] Inhibition does not follow a simple Michaelis–Menten model. The
IC50 value was estimated as the inhibitor concentration leading to a half-
maximal reaction rate. [b] Pseudo-IC50 values because 6-iodo-UMP and
6-azido-UMP are covalent inhibitors. [c] No complete inhibition was ach-
ieved.
low affinity.[3] It thus appears that ligand distortion is ener-
getically silent. A possible explanation is the engagement of
new productive interactions after straining of the ligand. For
instance, the kinked substituents of 6-mercapto-UMP and 6-
cyano-UMP point into a hydrophobic pocket that is lined by
OMPD residues Phe86, Ile177, and Ile 224. Being more
strongly kinked, the nitrile fits much better into this pocket,
and this might explain its increased affinity compared to the
thiol. This notion is supported by the fact that the KM value
for OMP, which is strained when bound to OMPD, is 200-
fold smaller than the Ki for UMP;[2a] these values indicate
that the strained enzyme–substrate complex is more stable
than the product complex and that the kinked carboxylate
entertains favorable interactions. An additional explanation
why unfavorable distortions of the substrate are tolerated is
their overcompensation by strong energetic contributions
from the ribose-phosphate part of the nucleotide (Circe
effect).[17] It should be expected that the strain energy is re-
leased upon chemical transformation of the ligand, and this
Conclusions
Recent studies on OMPD from human, P. falciparum, and
M. thermoautotrophicum have found that this enzyme in
general can display rather peculiar chemistry that does not
appear to bear any relation to the decarboxylation reac-
tion.[3,8] These addition, substitution, and hydrolysis reac-
tions are orders of magnitude slower than the native decar-
boxylation reaction, yet efficient enough to be readily ob-
served. The chemodiversity is species-dependent because,
for instance, 6-cyano-UMP degradation results in different
products in human (see above) and M. thermoautotrophicum
OMPD.[8] Likewise, the non-natural activities might be ex-
ploited for drug development, with the suicide inhibition of
the malaria parasite P. falciparum OMPD by 6-iodo-UMP
as an example.[3,4] All of these off-pathway activities rely on
a key lysine residue, the functionality of which can only in
part be replaced by water. The existence of such activities
raises the question as to why OMPD harbors a nucleophile
6624
www.chemeurj.org
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 6619 – 6625

OMPD Reactivity
FULL PAPER
Proc. Natl. Acad. Sci. USA 2000, 97, 2005–2010; d) P. Beak, B.
Siegel, J. Am. Chem. Soc. 1976, 98, 3601–3606; e) J. K. Lee, K. N.
Houk, Science 1997, 276, 942–945.
that is apparently not used during catalysis but solely serves
as a proton donor/acceptor? Recent carbon and deuterium
isotope effect studies have provided evidence for a short-
lived carbanion at the C6-position of the pyrimidine ring,
which upon protonation, presumably by the lysine, gives the
product UMP.[22] Apparently, the nucleophilicity of the
lysine is harnessed in the ground state of the wild-type
enzyme, but upon mutation of a nearby aspartic acid to as-
paragine, the nucleophilicity is increased as exemplified by
the covalent linkage to the electrophilic groups of 6-cyano-
UMP and 6-acetyl-UMP. This effect emphasizes the impor-
tance of the intact hydrogen-bonding network between
charged side-chains in the OMPD active site for decarboxy-
lation activity.[31]
[3] N. Wu, Y. Mo J. Gao, E. F. Pai, Proc. Natl. Acad. Sci. USA 2000, 97,
2017–2022.
[4] J. A. Smiley, M. E. Jones, Biochemistry 1992, 31, 12162–12168.
[5] a) J. Wittmann, D. Heinrich, K. Gasow, A. Frey, U. Diederichsen,
M. Rudolph, Structure 2008, 16, 82–92; b) A. M. Bello, E. Poduch,
M. Fujihashi, M. Amani, Y. Li, I. Crandall, R. Hui, P. I. Lee, K. C.
Kain, E. F. Pai, L. P. Kotra, J. Med. Chem. 2007, 50, 915–921.
[6] M. E. Meza-Avina, L. Wei, M. G. Buhendwa, E. Poduch, A. M.
Bello, E. F. Pai, L. P. Kotra, Mini-Rev. Med. Chem. 2008, 8, 239–
247.
[7] T. L. Amyes, J. P. Richard, J. J. Tait, J. Am. Chem. Soc. 2005, 127,
15708–15709.
[8] K. Shostak, M. E. Jones, Biochemistry 1992, 31, 12155–12161.
[9] A. M. Bello, E. Poduch, Y. Liu, L. Wei, I. Crandall, X. Wang, C.
Dyanand, K. C. Kain, E. F. Pai, L. P. Kotra, J. Med. Chem. 2008, 51,
439–448.
[10] M. Fujihashi, A. M. Bello, E. Poduch, L. Wei, S. C. Annedi, E. F.
Pai, L. P. Kotra, J. Am. Chem. Soc. 2005, 127, 15048–15050.
[11] B. P. Callahan, A. F. Bell, P. J. Tonge, R. Wolfenden, Russ. J. Bioorg.
Chem. 2006, 32, 59–65.
[12] B. G. Miller, M. J. Snider, S. A. Short, R. Wolfenden, J. Biol. Chem.
2001, 276, 15174–15176.
[13] N. Wu, W. Gillon, E. F. Pai, Biochemistry 2002, 41, 4002–4011.
[14] J. D. Dunitz, X-ray Analysis and the Structure of Organic Molecules,
Cornell University Press, Ithaca, 1979, pp. 366–372.
[15] J. E. Davies, A. J. Kirby, I. V. Komarov, Helv. Chim. Acta 2003, 86,
1222–1233.
[16] A. Fersht, Structure and Mechanism in Protein Science, Freeman,
New York, 1999, pp. 372–375.
[17] a) W. P. Jencks, Adv. Enzymol. Relat. Areas Mol. Biol. 1975, 43,
219–410; b) B. G. Miller, M. J. Snider, S. A. Short, R. Wolfenden,
Biochemistry 2000, 39, 8813–8818; c) A. Sievers, R. Wolfenden,
Russ. J. Bioorg. Chem. 2005, 31, 45–52.
[18] B. P. Callahan, R. Wolfenden, J. Am. Chem. Soc. 2004, 126, 14698–
14699.
[19] A. Vasella, G. J. Davies, M. Bçhm, Curr. Opin. Chem. Biol. 2002, 6,
619–629.
Experimental Section
Protein purification, inhibitor studies, crystallization, and soaking experi-
ments: OMPD was cloned, produced, purified, and crystallized as de-
scribed.[23] Enzymatic activity, inhibitor studies, and cocrystallization ex-
periments were performed as described in reference [5a]. Soaking experi-
ments were performed for wt-OMPD with BMP and 6-cyano-UMP; and
for Asp312Asn OMPD with 6-azido-UMP, 6-cyano-UMP, and 6-acetyl-
UMP. The final nucleotide concentration in the crystallization drop was
2 mm, and soaking times varied from minutes (6-cyano-UMP) to several
days (all other nucleotides).
Data collection, structure determination, and refinement: Data were col-
lected at 100 K and reduced with the HKL programs (HKL research) or
XDS.[24] The structures were determined by molecular replacement by
using COMO and the best solution for the protein-only coordinates of
previously published OMPD structures.[25,5a] Models were built in COOT
and refined with REFMAC5.[26,27] The data that were derived from the
60 min soak of a wt-OMPD crystal with 6-cyano-UMP is slightly twinned
with a twin operator (l,Àk,h) and a twin fraction of 13%. Twinning in
OMPD was described earlier,[23] and the wt-OMPD–6-cyano-UMP struc-
[20] I. P. Fabrichniy, L. Lehtiç, M. Tammenkoski, A. B. Zyryanov, E. Ok-
sanen, A. A. Baykov, R. Lahti, A. A. Goldman, J. Biol. Chem. 2006,
282, 1422–1431.
[21] M. J. Hokenson, G. A. Cope, E. R. Lewis, K. A. Oberg, A. L. Fink,
Biochemistry 2000, 39, 6538–6545.
[22] a) J. L. van Vleet, L. A. Reinhardt, B. G. Miller, A. Sievers, W. W.
Cleland, Biochemistry 2008, 47, 798–803; b) K. Toth, T. L. Amyes,
B. M. Wood, K. Chan, J. A. Gerlt, J. P. Richard, J. Am. Chem. Soc.
2007, 129, 12946–12947.
[23] J. G. Wittmann, M. G. Rudolph, Acta Crystallogr. Sect. D 2007, 63,
744–749.
ture was refined against intensities by using
a recent version of
REFMAC5 that takes twinning into account. All other structures were
derived from untwinned data. The same set of 5% of reflections was re-
served for R-free cross-validation in all structures of a given crystal
form.[28] Statistics are summarized in Table 1. Figures were created with
Bobscript and rendered with Raster3D,[29,30] or PyMol (www.pymol.org).
Volume calculations were done with Moloc (www.moloc.ch).
A detailed description of chemical syntheses, kinetics, and mass spectro-
metric data is provided in the Supporting Information.
[24] W. Kabsch, J. Appl. Crystallogr. 1988, 21, 916–924.
[25] G. Jogl, X. Tao, Y. Xu, L. Tong, Acta Crystallogr. Sect. D 2001, 57,
1127–1134.
Acknowledgements
[26] P. Emsley, K. Cowtan, Acta Crystallogr. Sect. D 2004, 60, 2126–2132.
[27] CCP4: Acta Crystallogr. Sect. D 1994, 50, 760–763.
[28] A. T. Brꢀnger, Nature 1992, 355, 472–475.
[29] R. M. Esnouf, J. Mol. Graphics 1997, 15, 132–134.
[30] E. A. Merritt, M. E. P. Murphy, Acta Crystallogr. Sect. D 1994, 50,
869–873.
[31] Note added in proof: Comparable results were published recently:
M. Fujihashi, L. Wei, L. P. Kotra, E. F. Pai, J. Mol. Biol. 2009, 387,
1199–1210.
We thank Julia Wittmann for kinetic measurements on 6-cyano-UMP,
Alexandra Frey for the synthesis of 6-acetyl-UMP, Bernhard Schmidt
and Nicole Eiselt for MALDI-TOF mass spectrometry, the colleagues at
the Swiss Light Source for beam time and guidance, and Dagmar Kloster-
meier for helpful suggestions on the manuscript. This work was support-
ed by a grant from the Deutsche Forschungsgemeinschaft.
[1] A. Radzicka, R. Wolfenden, Science 1995, 267, 90–93.
[2] a) B. G. Miller, R. Wolfenden, Annu. Rev. Biochem. 2002, 71, 847–
885; b) R. B. Silverman, M. P. Groziak, J. Am. Chem. Soc. 1982, 104,
6434–6439; c) T. C. Appleby, C. Kinsland, T. P. Begley, S. E. Ealick,
Received: February 13, 2009
Published online: May 26, 2009
Chem. Eur. J. 2009, 15, 6619 – 6625
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6625