Inactivation of Glucosamine-6-Phosphate Synthase by N3-Oxoacyl Derivatives of L-2,3-Diaminopropanoic Acid

Article abstract of DOI:10.1002/cbic.201100587

N³-Oxoacyl derivatives of L-2,3-diaminopropanoic acid 1-4, containing either an epoxide group or a conjugated double bond system, inactivate Saccharomyces cerevisiae glucosamine-6-phosphate (GlcN-6-P) synthase in a time- and concentration dependent manner. The results of kinetics studies on inactivation suggested a biphasic course, with formation of the enzyme-ligand complex preceding irreversible modification of the enzyme. The examined compounds differed markedly in their affinity to the enzyme active site. Inhibitors containing a phenyl ketone moiety bound much more strongly than their methyl ketone counterparts. The molecular mechanism of enzyme inactivation by phenyl ketone compounds 1 and 3 was elucidated by using a stepwise approach with 2D NMR, MS and UV-visible spectroscopy. A substituted thiazine derivative was identified as the final product of a model reaction between an epoxide compound, 1, and L-cysteine ethyl ester (CEE); and the respective cyclic product, found as a result of reaction between 1 and CGIF tetrapeptide, was identical to the N-terminal fragment of GlcN-6-P synthase. On the other hand, the reaction of a double-bond-containing compound, 3, with CEE, CGIF and GlcN-6-P synthase led to the formation of a C-S bond, without any further conversion or rearrangement. Molecular mechanisms of the reactions studied are proposed.

Full text of DOI:10.1002/cbic.201100587

DOI: 10.1002/cbic.201100587
Inactivation of Glucosamine-6-Phosphate Synthase by N3-
Oxoacyl Derivatives of l-2,3-Diaminopropanoic Acid
[a]
[b]
[b]
[c]
´
Robert Je˛drzejczak, Marek Wojciechowski, Ryszard Andruszkiewicz, Paweł Sowinski,
Agata Kot-Wasik,[d] and Sławomir Milewski*[b]
N3-Oxoacyl derivatives of l-2,3-diaminopropanoic acid 1–4,
containing either an epoxide group or a conjugated double
bond system, inactivate Saccharomyces cerevisiae glucosamine-
6-phosphate (GlcN-6-P) synthase in a time- and concentration
dependent manner. The results of kinetics studies on inactiva-
tion suggested a biphasic course, with formation of the
enzyme–ligand complex preceding irreversible modification of
the enzyme. The examined compounds differed markedly in
their affinity to the enzyme active site. Inhibitors containing a
phenyl ketone moiety bound much more strongly than their
methyl ketone counterparts. The molecular mechanism of
enzyme inactivation by phenyl ketone compounds 1 and 3
was elucidated by using a stepwise approach with 2D NMR,
MS and UV–visible spectroscopy. A substituted thiazine deriva-
tive was identified as the final product of a model reaction be-
tween an epoxide compound, 1, and l-cysteine ethyl ester
(CEE); and the respective cyclic product, found as a result of re-
action between 1 and CGIF tetrapeptide, was identical to the
N-terminal fragment of GlcN-6-P synthase. On the other hand,
the reaction of a double-bond-containing compound, 3, with
CEE, CGIF and GlcN-6-P synthase led to the formation of a CꢀS
bond, without any further conversion or rearrangement. Mo-
lecular mechanisms of the reactions studied are proposed.
Introduction
l-Glutamine is a substrate for several enzymes known as ami-
dotransferases; these catalyse the transfer of a g-amide nitro-
gen from l-Gln to various acceptor molecules.[1] One of the
amidotransferases, glucosamine-6-phosphate synthase (GlcN-6-
P synthase), which uses d-fructose-6-phosphate (Fru-6-P) as an
acceptor substrate, is the subject of interest as a potential
target for antifungal chemotherapy and for possible pharmaco-
logical intervention in insulin-independent diabetes mellitus.[2,3]
Thus the molecular properties of the enzyme have been exten-
sively studied, including determination of the crystal structure
of its bacterial version (as an intact protein and in complexes
with several substrates, substrate analogues, products and
inhibitors).[4] The results of these studies showed that E. coli
GlcN-6-P synthase is a dimer of two identical subunits, each
composed of two functional domains: the glutamine amide-
hydrolysing domain (GAH) and the sugar phosphate-isomeris-
ing domain (ISOM), connected by an intramolecular channel.
The catalytic mechanism is complex and involves: 1) Fru-6-P
binding at ISOM, which triggers GAH ordering and assembly of
the interdomain channel; 2) l-Gln binding at GAH, hydrolysis
of its amide and ammonia transfer to ISOM through the chan-
nel; and 3) formation of the fructoseimine intermediate and its
subsequent isomerisation to the final product, GlcN-6-P.[5]
Much less structural data is available for eukaryotic versions of
GlcN-6-P synthase, as so far only the crystal structures of the
ISOM domains of Candida albicans and human enzymes have
been determined,[6,7] but there is little doubt that the mecha-
nism of catalysis in eukaryotic and prokaryotic GlcN-6-P syn-
thases is exactly the same.
It is believed that inhibitors of GlcN-6-P synthase have the
potential to become new drugs. Several such compounds (sub-
strate analogues, transition-state-analogue inhibitors and het-
erocyclic compounds that inhibit enzyme dimerisation) have
been described.[4,8] These include a number of glutamine ana-
logues, N3-acyl derivatives of l-2,3-diaminopropanoic acid that
inhibit/inactivate GlcN-6-P synthase. N3-(4-methoxyfumaroyl)-l-
2,3-diaminopropanoic acid (FMDP) is one of the strongest in-
hibitors of GlcN-6-P synthase known to date. It inactivates the
enzyme in a time- and concentration-dependent manner by
forming a covalent bond with the catalytic Cys1 residue at the
enzyme active site.[9] The molecular mechanism of GlcN-6-P in-
activation by FMDP, established by Badet and co-workers, in-
volves a Michael-type addition of Cys1 to the fumaroyl double-
[a] Dr. R. Je˛drzejczak
Present address: Midwest Center for Structural Genomics
and Structural Biology Center, Biosciences, Argonne National Laboratory
Building 202, 9700 South Cass Avenue, Argonne, IL 60439 (USA)
[b] Dr. M. Wojciechowski, Dr. R. Andruszkiewicz, Dr. S. Milewski
Department of Pharmaceutical Technology and Biochemistry
´
Gdansk University of Technology
´
11/12 Narutowicza Str., 80-233 Gdansk (Poland)
E-mail: slamilew@pg.gda.pl
´
[c] Dr. P. Sowinski
´
Intercollegiate NMR Laboratory, Gdansk University of Technology
´
11/12 Narutowicza Str., 80-233 Gdansk (Poland)
[d] Dr. A. Kot-Wasik
´
Department of Analytical Chemistry, Gdansk University of Technology
´
11/12 Narutowicza Str., 80-233 Gdansk (Poland)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.201100587.
ChemBioChem 2012, 13, 85 – 96
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
85

S. Milewski et al.
bond, followed by two consecutive rearrangements, thus lead-
ing finally to the formation of a substituted 1,4-thiazin-3-one
ring.[10]
During the course of our search for novel inhibitors of
fungal GlcN-6-P synthases, several derivatives of N3-oxoacyl-l-
2,3-diaminopropanoic acid were synthesised and found to be
effective inhibitors of the target enzyme.[11] Now, evidence for
the molecular mechanism of GlcN-6-P synthase inactivation by
these compounds is presented. The biological model used in
this study was the Saccharomyces cerevisiae enzyme, overpro-
duced in host cells, purified to near homogeneity and charac-
terised for its basic molecular and catalytic properties.
Results
Isolation and characterisation of yeast GlcN-6-P synthase
When grown in YPD medium, the YRS23-3 transformant, which
contains the original GFA1 gene under the control of a PGK1
promoter, reproducibly expressed S. cerevisiae GlcN-6-P syn-
thase at 5–7% of total cytoplasmic protein, as revealed by den-
sitometric SDS-PAGE analysis. The enzyme was purified to at
least 98% homogeneity with 42% final yield, by using the six-
step procedure described in the Experimental Section. The Km
of the pure enzyme was 0.74 mm for l-Gln, 0.39 mm for d-Fru-
6-P, and kcat was 35.5 sꢀ1. Pure GlcN-6-P synthase was chroma-
tofocused on a MonoP HR 5/5 fast protein liquid chromatogra-
phy (FPLC) column with a pH 6 to 4 gradient. The enzyme was
partially denatured during column development, but the activi-
ty profile (not shown) enabled estimation of an isoelectric
point of 5.35ꢁ0.05. This value is lower than the 6.01 calculat-
ed for the known amino acid composition of the protein, but a
similar phenomenon was previously found for the Escherichia
coli and C. albicans counterparts of the yeast GlcN-6-P syn-
thase.[12,13] The molecular weight (MW) of the native protein
was determined by size exclusion chromatography at 320ꢁ
2 kDa, while the MW of the Gfa1p subunit determined by SDS-
PAGE was 80ꢁ0.5 kDa. The latter value is in agreement with
the theoretical MW (79.916 kDa) calculated from the known
amino acid sequence. It is most likely that native yeast GlcN-6-
P synthase is a homotetramer (i.e. of identical subunits), as was
previously shown for some other eukaryotic versions of this
enzyme.[4] The pure enzyme was inhibited by uridine diphos-
phate N-acetylglucosamine (UDP-GlcNAc) with IC50 =0.52 mm.
The inhibition was non-competitive with respect to l-Gln and
uncompetitive with respect to d-Fru-6-P, thus suggesting that
the UDP-GlcNAc binding site is not the enzyme active centre
and indicating preferential binding of the inhibitor to the
enzyme:Fru-6-P complex. It is highly probable that the UDP-
GlcNAc-binding site is located in the S. cerevisiae enzyme in a
similar manner as was shown for the truncated version of C. al-
bicans Gfa1p.[6] Amino acid residues constituting the centre in
C. albicans Gfa1p are completely conserved in the yeast
enzyme, as shown in Figure 1. Yeast GlcN-6-P synthase was
chosen as a biological model for the studies on enzyme inacti-
vation because of its appropriate N-terminal amino sequence:
a chymotrypsin cleavage site is present close to the N termi-
86
www.chembiochem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2012, 13, 85 – 96

Inactivation of GlcN-6-P Synthase by Glutamine Analogues
nus, at the carboxyl side of Phe4, which facilitates identifica-
tion of any products of Cys1 covalent modification in a chymo-
tryptic digest. The properties of the enzyme, purified to near
homogeneity, closely resemble those reported previously for
other fungal version of this protein, including its homotetra-
meric structure and inhibition by UDP-GlcNAc. Although the
crystal structure of the yeast enzyme is not known, there is
little doubt that the constitution and 3D configuration of the
active centres are very well conserved, so any conclusions
drawn from the present study should be of a general character
and valid for all versions of this enzyme.
concentration [I] is described by Equation (1):
1=kapp ¼ ð1=k2Þ þ ðKinact=k2Þ ꢃ ð1=½IꢂÞ
ð1Þ
where k2 =ln2/T, Kinact =k /k1, kapp is the apparent inactivation
ꢀ1
velocity constant at the given inhibitor concentration, and T is
the minimum inactivation half-time at infinite inactivator con-
centration. The IC50 values for inhibition and the kinetics pa-
rameters of inactivation are summarised in Table 1.
Table 1. Parameters of inhibition and inactivation of S. cerevisiae GlcN-6-P
synthase by N3-oxoacyl derivatives of l-2,3-diaminopropanoic acid.
Inactivation of GlcN-6-P synthase by l-2,3-diaminopropanoic
acid derivatives
Compound
Inhibition
IC50
[mm]
Inactivation
Kinact
[mm]
T
k2
k2/Kinact
[mꢀ1 sꢀ1
]
[min]
[minꢀ1
]
N3-Oxoacyl derivatives of l-2,3-diaminopropanoic acid 1–4
shown in Scheme 1 are structural analogues of the previously
reported specific inhibitors of GlcN-6-P synthase (in that they
contain either fumaroyl or epoxysuccinoyl fragments in the N3-
acyl portion of the molecule) including the most efficient
FMDP, 5.[14,15]
1
2
3
4
660ꢁ50
7350ꢁ250
270ꢁ30
450
(1250)
19400
1.1
0.630
23.3
(7.71)
0.48
(0.63)
157
(43.8)
0.51
(0.95)
(1.2)
1.25
(1.2)
1.3
(0.578)
0.554
(0.578)
0.533
(15200)
56.5
(195)
15200
(8700)
(1.35)
1.5
(0.513)
0.462
5600ꢁ200
(1.4)
(0.495)
Parameters in parentheses of inactivation determined in the presence of
10 mm Fru-6-P.
Whereas all the tested compounds demonstrated similar
reactivity towards the enzyme (as reflected by the k2 and T
values), they differed markedly in their affinity to the enzyme
active site (as revealed by the Kinact values). Compounds con-
taining the phenyl ketone terminal moiety (1 and 3) bound
much more strongly than their counterparts containing a ter-
minal methyl ketone (2 and 4). On the other hand, by analy-
sing the effect of epoxide/double bond substitution in the 1–3
and 2–4 pairs, one can observe a slightly higher reactivity (k2)
but lower affinity (Kinact) for the epoxide compounds. Generally,
compounds containing the double bond (3 and 4) were slight-
ly stronger inhibitors than their respective epoxide counter-
parts (1 and 2) in terms of the IC50 values. Interestingly
enough, progressive yellowing of the reaction mixtures was
observed during reaction of the enzyme with 1, while no
colour changes were noted in the reaction mixtures containing
2, 3 or 4. When inactivation of the enzyme was performed in
the presence of d-Fru-6-P, Kinact values for 1 and 3 were sub-
stantially higher, while those for 2 and 4 were slightly lower.
Scheme 1. Structures of N3-oxoacyl derivatives of l-2,3-diaminopropanoic
acid and FMDP.
The novel compounds 1–4 were tested as inhibitors and in-
activators of yeast GlcN-6-P synthase. The inhibition assay was
followed under standard conditions, with saturating concentra-
tions of both substrates, to determine the IC50 values. Enzyme
inactivation was performed in the absence of l-Gln and in the
presence or absence of d-Fru-6-P. Under these conditions all
tested compounds inactivated the enzyme in a time- and con-
centration-dependent manner. The determined inactivation
rate constants demonstrated a hyperbolic dependence on ini-
tial inactivator concentration, thus suggesting two-step inacti-
vation, with formation of a reversible complex preceding cova-
lent modification of the enzyme, according to:
Model reactions of 1 or 3 with cysteine ethyl ester (CEE) or
CGIF tetrapeptide
S. cerevisiae GlcN-6-P synthase, like its homologues from other
sources, contains the essential N-terminal Cys residue that was
previously identified as a catalytic nucleophile at the GAH
active centre of the bacterial enzyme. In the yeast enzyme,
Cys1 is followed consecutively by Gly, Ile and Phe residues (see
Figure 1), with the carboxyl side of Phe4 being the first specific
cleavage site for chymotrypsin. Therefore, CEE and the CGIF
k
þ1
k2
ƒ!
E þ I
½E : Iꢂ
E ꢀ I
ƒ!
ƒ
k
ꢀ1
where E is the free enzyme, I is an inactivator, [E:I] is the rever-
sible complex and E–I is the inactive, covalently modified
enzyme. Kinetic parameters of inactivation were determined
by assuming that the relationship between kapp and inactivator
ChemBioChem 2012, 13, 85 – 96
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
87

S. Milewski et al.
tetrapeptide were chosen as the most appropriate low-molec-
ular-weight substrates for the model reactions, aimed at iden-
tification of possible products.
A progressive disappearance of peaks corresponding to sub-
strates, and the appearance of the main product peak derived
from a substance with (M+z)/z=392.1, are evident. The prod-
uct was isolated and analysed by NMR. Analysis of these data
resulted in an unequivocal assignment of the structure shown
in Scheme 2, from which MW =391 can be calculated. This
The reaction between CEE and 1 in stoichiometric amounts
was performed under pH-controlled conditions (pH 5.0, 7.0
and 8.0), under argon and at room temperature. Reaction
progress was monitored by determination of the free thiol
content with 5,5’-dithiobis-(2-nitrobenzoic acid) (DTNB). Con-
sumption of CEE at pH 5.0 was very slow; on the other hand,
disappearance of the CEE thiol at pH 7.0 and 8.0 was rapid:
the determined second order velocity constants were 25.0 and
71.5mꢀ1 ꢀ1, respectively. The reaction mixture, entirely colour-
s
less at the beginning of reaction, progressively turned yellow
in a similar way as was observed for the enzyme inactivation
by 1, thus indicating formation of chromophore product(s).
Samples of the reaction mixtures were collected at time inter-
vals and analysed by reversed-phase FPLC (RP-FPLC). The re-
sults of these analyses are shown in Figure 2.
Scheme 2. Structure of the final product of the 1:CEE model reaction, de-
duced from the NMR experiments.
value is in accordance with the MW of the product giving the
[M+H]+ =392.1 signal in MS detection of the RP-FPLC analysis.
As the MW of 1 is 278 Da and that of CEE is 149 Da, MWproduct
is 36 Da lower than the sum of MW1 and MWCEE (278+149=
427). The UV–visible spectrum of this compound is obviously
different from that of 1 (Figure 3A). The most striking differ-
ence is the presence in the product spectrum of an additional
band in the 320–360 nm region, with lmax at 330 nm.
Under the same conditions, CEE reacted rapidly with 3 at
pH 7.0 and 9.0, to give a single product with (M+z)/z=412.2
(Figure 4). By assuming single protonation, the MW of this prod-
uct (411 Da) is equal to the sum of MW3 and MWCEE (262 and
149 Da, respectively). The reaction mixture remained colourless
during the entire reaction, and the overall shape of UV–visible
spectrum of the isolated product was almost identical to that
of 3 (Figure 3C), thus suggesting no changes in the chromo-
phore part of the substrate upon reaction with CEE. The prod-
uct was isolated and subjected to NMR analysis. The data ob-
tained were consistent with the structure of a product of a
Michael-type addition of the CEE thiol to the C5 atom of 3 (7
in Scheme 4, below).
Figure 2. RP-FPLC analysis of components of the model 1:CEE reaction mix-
ture. A) sample taken after 2 min; B) sample taken after 10 min. Positions of
peaks corresponding to 1 (c) and CEE (– · –) are provided for reference.
Inset: MS ESI spectra of substances indicated by the arrows. RI=relative in-
tensity.
Figure 3. The UV/Vis spectra of substrates and products of the model reactions. A) 1:CEE; B) 1:CGIF; C) 3:CEE; D) 3:CGIF. AU=arbitrary units; Prod=product.
88
www.chembiochem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2012, 13, 85 – 96

Inactivation of GlcN-6-P Synthase by Glutamine Analogues
Two other components of peak 1 in Figure 5A are expected
to have MW of 716 and 680 Da, as signals of their [M+H]+ and
[M+2H]2+ ions are present in the MS spectrum. The relative
content of component 716 in the reaction mixture decreased
with time, while that of component 680 increased. Peak 2, the
intensity of which decreased with time, is derived from a
single component with expected MW =698 or 699 Da. One
may thus conclude that signals of decreasing intensity derived
from intermediates, and that these were converted eventually
into the final product of 680 Da. The MW of this product is 18
and 36 Da lower than those of the two intermediates, while
the MW of the largest putative intermediate is equal to the
sum of MW1 and MWCGIF. The UV–visible spectrum of the isolat-
ed final product ( Figure 3B) is very similar to that of the prod-
uct of the reaction of 1 with CEE, thus suggesting a close simi-
larity of chromophore regions of both compounds. This as-
sumption was also confirmed by the progressive yellowing of
the reaction mixture containing CGIF and 1.
The overall shape of the UV–visible spectrum of the product
of the reaction between 3 and CGIF was almost identical to
those of the other substrates (Figure 3D), and a single peak,
different from those of the substrates and corresponding to
the substance with (M+z)/z=700.1 and 351.0, was found by
the RP-FPLC/MS analysis (details not shown). This finding indi-
cates that the course of reaction between the CGIF tetrapep-
tide and 3, and the chemical nature of the single product
formed in this reaction are the same as for the reaction of 3
with CEE.
Figure 4. RP-FPLC analysis of components of the model 3:CEE reaction mix-
ture. The positions of peaks corresponding to 3 (c) and CEE (– · –) are
provided for reference. Inset: MS ESI spectrum of the product indicated by
the arrow. RI=relative intensity.
The reaction of 1 with the CGIF tetrapeptide was slower
than that with CEE (the second order velocity constant of the
reaction at pH 7.0 was 1.25mꢀ1 ꢀ1), and RP-FPLC/MS analysis
s
of the components of the reaction mixture revealed more spe-
cies. In the chromatograms shown in Figure 5 one can distin-
MALDI-TOF analysis of the reaction between 1 or 3 with
GlcN-6-P synthase
Pure GlcN-6-P synthase was treated with 1 or 3 under condi-
tions ensuring complete enzyme inactivation. Components of
the reaction mixture were separated by electrophoresis, and
protein was digested in-gel with chymotrypsin. The digests
were subjected to MALDI-TOF analysis, and the obtained spec-
tra were compared to those of the chymotryptic digest of the
enzyme. Spectra of the chymotryptic digests of the native and
1-inactivated enzymes and their difference spectrum are
shown in Figure 6. The signal with the largest intensity, present
only in the B) spectrum, has (M+z)/z=681.2, identical to that
of the final product of the model reaction between 1 and
CGIF. On the other hand, the 439.1 signal, present exclusively
in the A) spectrum, seems to be derived from the CGIF tetra-
peptide.
Figure 5. RP-FPLC analysis of components of the model 1:CGIF reaction mix-
ture. A) sample taken after 5 min; B) sample taken after 20 min. The posi-
tions of peaks corresponding to 1 (c) and CGIF (– · –) are provided for ref-
erence. Inset: MS ESI spectra of substances indicated by the arrows. RI=rela-
tive intensity.
The 439.1 signal present in the MALDI-TOF spectrum of the
chymotryptic digest of native GlcN-6-P synthase was absent in
the spectrum of the chymotryptic digest of the 3-inactivated
enzyme (data not shown). On the other hand, the 700.1 signal
was found exclusively in the spectrum of the chymotryptic
digest of the 3-inactivated enzyme, although its intensity was
low.
guish two additional peaks (one of them quite complex), differ-
ent from those of the substrates. The intensities of these peaks
changed with time, thus indicating reaction progress and con-
version of intermediates into the final product. Peak 1 in chro-
matogram 5 A has three components. One, with (M+z)/z=
439.1 and seemingly derived from CGIF (MW =438 Da), is
absent from the respective peak 1 in the chromatogram
shown in Figure 5B).
Both difference spectra contained the inactivated enzyme-
specific signal 539.1 of unknown origin (see Figure 6).
ChemBioChem 2012, 13, 85 – 96
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
89

S. Milewski et al.
almost certainly valid for the
purpose of studies on inhibitor
binding and inactivation of
yeast GlcN-6-P synthase, be-
cause all residues identified as
crucial for substrate binding
and catalysis in the bacterial
enzyme are perfectly conserved
in other versions of this
enzyme, including that of S. cer-
evisiae (see Figure 1). In particu-
lar, the molecular gate in the
yeast enzyme is undoubtedly
the Trp87 residue.
Docking experiments re-
vealed that both 1 and 3 bind
at the active centre of GAH in a
very consistent way. Clustering
analysis of the resulting ligand
conformations showed that
only five qualitatively different
solutions could be identified,
and that there was only one
dominant cluster of ligand con-
formations (containing 35 mem-
Figure 6. MALDI-TOF analysis of the chymotryptic digests. Difference spectrum obtained by superposition of the
m/z 200–800 regions of spectra A) native GlcN-6-P synthase and B) 1-modified enzyme. Peaks derived from the
spectrum A) are shown as dashed lines and these from spectrum B) as solid lines. Signals present in both spectra
and signals of relative intensity lower than 10% are not shown.
Molecular modelling of inhibitor binding and its conversion
bers). At the same time, the representative conformation of
this cluster represented the minimum energy geometry of the
complex. The second cluster contained only ten solutions, with
a mean binding energy only about 0.4 kcalmolꢀ1 higher than
that of the first. The differences between the particular confor-
mations concerned only the phenyl ketone moiety of the
ligand, while the positions of its diaminopropanoic skeleton
and the amide bond closely resembled locations of the
respective parts of the bound natural substrate, glutamine
(Figure 7). The a-carboxy and a-amino groups of the ligands
interacted with a specific set of the active-site amino acids,
namely Arg73, Asp123, His86, Thr76, His77 and Gly99, in exact-
ly the same way as the natural substrate.[17] Moreover, the
amide moiety of the diaminopropanoic acid derivatives was
able to form hydrogen bonds to the main chain carbonyl
by the GAH domain of GlcN-6-P synthase
The molecular modelling of 1 and 3 docked at the active site
of the GAH domain of the E. coli GlcN-6-P synthase (PDB ID:
1GMS) has shed some light on the geometry of the complexes
and on the nature of the interactions between inhibitors and
amino acid residues within the active site. The conformations
of the crucial amino acid residues at the active site of this tem-
plate are almost identical to those in the complete “empty”
enzyme (PDB ID: 1JXA),[16] including that of the “molecular
gate”, Trp74, which is in the open conformation. The most im-
portant difference is the conformation of the catalytic Cys1 res-
idue, which is locked in the inactive conformation in 1GMS
while in 1JXA it is present in the active conformation. The
conclusions drawn from the results of these experiments are
Figure 7. Stereoview comparison of the geometries of 1 and 3 complexed with the glutamine binding site of GlcN-6-P synthase. Interactions of ligands with
Arg73, Trp74, Gly99 and Asp123 residues are indicated as black springs. Ligands are represented as thick sticks, and residues of the binding site within 4 ꢁ of
the ligands are represented as thin sticks. Trp74 is in the open conformation, and Cys1 is in the inactive conformation.
90
www.chembiochem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2012, 13, 85 – 96

Inactivation of GlcN-6-P Synthase by Glutamine Analogues
oxygen of Gly99 and the main chain amide nitrogen of Trp74.
All these interactions locked the respective parts of these in-
hibitors very firmly inside the binding pocket. The most nota-
ble difference between the 1:GAH and BADP:GAH complexes
was the presence in the former of an additional hydrogen
bond acceptor (the oxirane oxygen), absent in the three mole-
cule. As shown in Figure 7, in the 1:GAH complex, this atom
participates in the interaction with the ꢀNHꢀ group of the
Trp74 backbone. In the complex formed with 3, the average
distance between the backbone oxygen of Gly99 and the
hydrogen bond donor (the ꢀNHꢀ nitrogen of the ligand) is
2.85 ꢁ, while the average distance between the backbone
ꢀNHꢀ group nitrogen of Trp74 and the amide oxygen of the
ligand is 3.1 ꢁ. The analogous distances in the 1:GAH complex,
with the oxirane oxygen as the hydrogen bond acceptor, are
2.95 and 2.85 ꢁ, respectively.
The switch of the Cys1 residue to the active conformation,
known to occur upon Fru-6-P binding,[20] brings its sulfhydryl
group close to C6 of the 1 oxirane ring (4 ꢁ) and brings its a-
amino group to a position suitable for Schiff base formation
with the carbonyl group of this inhibitor. It is well known that
in the catalytic mechanism of GlcN-6-P synthase, the a-amino
group of Cys1 participates in deprotonation of its own thiol
moiety.[5,21] It seems very likely that when 1 is bound at the
GAH active site instead of glutamine, the proton captured by
the a-amino group of Cys1 can be subsequently transferred to
the oxirane oxygen, thus facilitating ring opening upon nucle-
ophilic attack of the cysteine thiolate at C6. However, the
proton might also be derived from any other active site resi-
due, as suggested by Hollenhorst et al. in their recent studies
on the mechanism of GlcN-6-P synthase inactivation by enea-
mide and epoxyamide inhibitors.[22] The nucleophilic a-amino
group of Cys1 might then easily form the Schiff base with the
keto group of the inhibitor, thus leading to closure of the six-
membered ring. It is not clear whether there are any factors
promoting the subsequent water elimination to give the final
structure (corresponding to that formed in the model reaction
of 1 with CEE), however, they do not seem necessary. More
importantly, the substituted ring system involving a-nitrogen,
two carbon atoms and the sulfur of Cys1 is well accommodat-
ed at the GAH active centre, as shown in Figure 8. It seems,
therefore, that formation of the final cyclic product of enzyme
inactivation by 1 does not require initial binding or any major
rearrangements of the participating molecules (ligand or bind-
ing site residues) except for the switch of Cys1 to the active
conformation.
The docking experiments revealed that for inhibitors that
contain the phenyl ketone moiety, that is, 1 and 3, there is an-
other site of specific interaction with the enzyme active centre
(not present in 2 and 4). In the bound conformation, the
phenyl group of both 1 and 3 is in a position where it can par-
ticipate in p-stacking interactions with the indole ring of
Trp74. The most abundant cluster and the lowest energy con-
formation of the 1:GAH and 3:GAH complexes clearly show
that the planes of the phenyl ring of both inhibitors and the
indole ring of Trp74 face each other at an average distance of
3.5 ꢁ, with their geometric centres located against each other.
Such a relative orientation of the aromatic rings is considered
optimal for p-stacking interactions.[18] Obviously these interac-
tions are missing in the complexes of GAH with 2 and 4, and
this is consistent with the much lower affinity of the methyl
ketone inhibitors to the enzyme, in comparison to their phenyl
ketone counterparts (Table 1). Moreover, analysis of the geo-
metries grouped in the less abundant clusters, as well as the
results of preliminary molecular dynamics simulations (data
not shown) suggest that the phenyl ketone moiety might also
participate in favourable cation–p interactions with the Arg26
residue, and that this might also contribute to the stronger
binding of 1 and 3, in comparison with 2 and 4. This interac-
tion is only possible when the Trp74 residue is in the open
conformation with its indole ring pointing toward the accept-
or’s binding site. When the receptor used for docking has its
Trp74 residue switched to the closed conformation (intersect-
ing the ammonia channel), the most abundant conformation
of the docked ligand has a completely different geometry of
the phenyl ketone moiety. It points away from the binding
site, thereby making any p–p interactions with the Trp74
indole ring impossible. This observation explains why the
stacking interactions of 1 and 3 with Trp74 must be lost when
the ISOM domain of GlcN-6-P synthase is occupied by Fru-6-P,
as binding of this substrate to the enzyme triggers closure of
the molecular gate, due to the rotation of the indole ring by
about 758.[19] This conformational change is the most likely ex-
planation for the substantially higher Kinact values determined
for inactivation of GlcN-6-P synthase by 1 and 3, in the pres-
ence of 10 mm Fru-6-P (Table 1).
Discussion
Glucosamine-6-phosphate synthase belongs to the class II sub-
family of glutamine-dependent amidotransferases. Like the
other members of this subfamily, the enzyme contains an N-
terminal Cys1 residue that provides the reactive thiol nucleo-
phile to participate in glutamine hydrolysis in the initial stage
of GlcN-6-P-synthase-catalysed formation of glucosamine-6-
phosphate. A number of reactive electrophilic glutamine ana-
logues reacting with Cys1 of GlcN-6-P synthase (and conse-
quently inactivating the enzyme) are known, including a-diazo-
ketones, a-haloketones, g-dimethylsulfonium salts, N3-fumaroyl
derivatives of l-2,3-diaminopropanoic acid and its epoxy deriv-
atives.[4] So far, a detailed molecular mechanism of enzyme
inactivation has been determined only for FMDP, where a
Michael-type nucleophilic addition to the conjugated double
bond system was, under certain conditions, followed by intra-
molecular cyclisation by involving the a-amino group of a sub-
stituted Cys1.[10] In this paper, the results of studies on S. cerevi-
siae GlcN-6-P synthase inactivation by FMDP analogues bearing
the N3-oxoacyl functionality instead of the N3-fumaroyl are pre-
sented. Two out of the four studied compounds contained the
phenyl ketone substituents (1 and 3), while the remaining two
had methyl ketone counterparts (2 and 4). The electrophilic
centre was an activated double bond in 3 and 4 and an epox-
ide group in 1 and 2. All four compounds inactivated yeast
ChemBioChem 2012, 13, 85 – 96
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
91

S. Milewski et al.
Figure 8. Stereoview of the geometry of the putative cyclic product of the inactivation of GlcN-6-P synthase by 1. Top: general view at the GAH domain cova-
lently modified by 1. Below: binding site (enlarged). The protein is presented as a cartoon model, with the N-terminal b-strand in front (dark) and the Cys1
residue incorporated into the cyclic product of inactivation drawn as sticks.
GlcN-6-P synthase in time- and concentration-depen-
dent manner. Yellowing of the reaction mixture
during enzyme inactivation by 1 suggested forma-
tion of a novel chromophore system. On the other
hand, substantial difference in enzyme inhibitory
potential between phenyl ketone and methyl ketone
compounds, and contradictory effects of Fru-6-P
presence on enzyme inactivation, indicated the pos-
sibility of any undetermined specific enzyme–inhibi-
tor interactions, triggered upon Fru-6-P binding. In
order to determine the molecular mechanisms of
enzyme inactivation, including the above-mentioned
phenomena, we used a stepwise approach by involv-
ing studies on model reactions with low molecular-
weight mimics of the enzyme catalytic residue by
means of various spectroscopic techniques.
Reaction of 1 with CEE was rapid and led to the
formation of a single product that was isolated and
characterised by 2D NMR, MS-ESI and UV–visible
spectroscopy. The structure deduced from these
analyses, shown in Scheme 2, is the only one consis-
tent with all experimental data. A possible mecha-
nism of the reaction between 1 and CEE is presented
in Scheme 3 and consists of the four main steps. Ini-
tially, the thiolane group of CEE attacks the C6 atom
of 1, to open the oxirane ring and form the SꢀC
bond, and thus gives the putative intermediate (a).
In the second step, the a-amino group of CEE at-
Scheme 3. Hypothetical mechanism of reaction between 1 and CEE.
92
www.chembiochem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2012, 13, 85 – 96

Inactivation of GlcN-6-P Synthase by Glutamine Analogues
tacks C7 of the ketone carbonyl to form a Schiff base and
closes the six-membered ring. A water molecule is eliminated
during this condensation, and the putative intermediate (b) is
formed. Subsequently, elimination of another water molecule
(a reaction that is characteristic for b-hydroxy acids and its
derivatives and noted also in some biological systems without
participation of any enzymatic catalysis)[23] gives rise to the
C5=C6 double bond. This reaction seems to be favoured by
the presence of electron-withdrawing substituents at C6 that
should enhance acidity of the proton there and thus stimulate
its removal as the first step of b-elimination. Formation of a
conjugated double system is another factor stimulating this
reaction step. Finally, the conjugated p electron system is iso-
merised to the (probably more stable) intra-ring system in 6
that is the final product of reaction between 1 and CEE, as
identified by NMR.
Cys1 thiol at C6 of the inhibitor molecule, which is a prerequi-
site for the subsequent formation of the thiazine ring system.
This attack should be additionally facilitated by the preceding
protonation of the epoxide oxygen. The proton might be de-
rived from a water molecule, with participation of the N-termi-
nal a-amino group of Cys1, thus serving as a general base cat-
alyst, as has been similarly proposed to take place during l-Gln
hydrolysis when catalysed by GlcN-6-P synthase.[21] Such facili-
tated protonation and stabilisation of the final product (and
probably also the intermediates) by the hydrogen bond net-
work seem to be the reason for the faster reaction of GlcN-6-P
synthase with 1, in comparison to the model reaction of CGIF
with this compound.
The mechanism of reactions of 3 with CEE, CGIF and GlcN-6-
P synthase seems to be much different from that of its epoxide
analogue 1. For 3, there is little doubt that all three reactions
stopped after the initial Michael-type addition of the cysteinyl
thiol nucleophile to the conjugated enone system. This is clear-
ly suggested by the results of RP-FPLC/MS (model reactions:
single products, MW of the products equal to the sum of sub-
strate MWs; enzyme inactivation: the 700.1 signal substituting
the 439.1 signal in MALDI-TOF spectrum of chymotryptic di-
gests) and UV–visible analysis (shape of the product spectrum
similar to those of the substrates, no yellowing of the reaction
mixture). However, the target of the initial nucleophilic attack
(C5 or C6) is not clear, but in our opinion (and consistently
Obviously, in 1 there are two possible electrophilic sites for
nucleophilic attack: the C5 and C6 carbons. Preference for C6
over C5 results from at least two factors: 1) a quite significant
difference in the electrostatic potential (ESP), ꢀ0.229 at C5 and
+0.090 at C6, thus favouring attack at the latter; 2) if the CꢀS
bond was formed with C5, a seven-membered ring would be
formed at the second stage. Such a system is much less stable
than a six-membered ring and thus unlikely to be formed.
Therefore it seems that, in this case, subsequent formation of
the stable ring system is an important factor stimulating direc-
tion of the initial nucleophilic attack. However, it is worth men-
tioning that there are numerous examples of epoxide-contain-
ing inhibitors of other cysteinyl enzymes, including E-64 (a
well-known inhibitor of cysteinyl proteases), that demonstrate
significant regiospecific preference of the nucleophilic attack;
this was also evident in model reactions with low molecular-
weight thiols.[24]
1
with the H NMR data) this should be the C5 atom. Both atoms
bear negative EPS values (ꢀ0.244 at C5 and ꢀ0.155 at C6), so
there is no clear electrostatic preference. On the other hand, if
assuming attack of the thiol group of CEE, CGIF and Cys1 of
the enzyme on C6, subsequent formation of the six-membered
ring (Scheme 4) would seem unavoidable, while the formation
of a Schiff base (theoretically possible after attack at C5),
would result in an unstable seven-membered ring. However,
the possibility of formation of any cyclic products was exclud-
ed by the results of spectroscopic analyses.
Intermediates corresponding to (a) and (b) in Scheme 3
were identified in the reaction mixture of CGIF and 1. This re-
action was much slower, and thus allowed capturing of inter-
mediates and obtaining their MS-ESI spectra. Although we
were not able to isolate the pure final product of this reaction
and determine its structure by NMR, the MW determined by
MS-ESI corresponding to (MW1 +MWCGIF)ꢀ(2ꢂ18), and the fact
that the UV–visible spectrum of this compound and that of
the product of reaction between 1 and CEE are apparently
identical seem to provide strong evidence that the substituted
ring systems formed in both reactions are identical.
Therefore attack at C6 and subsequent formation of a stable
six-membered ring is unlikely to occur. One may thus conclude
that attack at C5 results in formation of 7 as the final product,
without any further conversion/rearrangement. Notably, the
same direction of the initial nucleophilic attack was previously
found in reactions of FMDP 5 with l-cysteine or with E. coli
GlcN-6-P synthase.[10]
One of the most surprising findings was the unfavourable
effect of Fru-6-P on GlcN-6-P inactivation by 1 and 3, as all pre-
vious reports suggested that binding of this substrate to GlcN-
6-P synthase facilitates enzyme inactivation by reactive gluta-
mine analogues. This effect, originally found with inactivation
of the E. coli GlcN-6-P synthase by 5[25] was later explained as a
consequence of the ordering of the GAH active centre upon
Fru-6-P binding at ISOM.[19] This explanation seems valid for
inactivation of the S. cerevisiae enzyme by 2 and 4, but in the
case of the reaction involving 1 and 3 it is apparently prevent-
ed due to interactions between the aromatic ring of the inhibi-
tors and the indole ring of Trp87.
Results of MALDI-TOF analysis of components of the chymo-
tryptic digests obtained upon treatment of native and 1-inacti-
vated GlcN-6-P synthase, especially the 681.2 signal substitut-
ing the 439.1 signal, clearly suggest that reaction of 1 with the
enzyme leads to the formation of the same derivative of the
CGIF tetrapeptide that was detected in the model reaction
1:CGIF. The possibility of formation of a thiazine ring is further
supported by the identical colour changes noted in both re-
action mixtures. Further support arises from the results of the
molecular modelling shown in Figure 8, which confirm the
possibility of accommodation of the ring system at the active
site of the GAH domain. On the other hand, docking of 1 at
the active site of GAH revealed the possibility of attack by
The mechanism of GlcN-6-P synthase inactivation by 1 and 3
is clearly different from that found previously by Kucharczyk
ChemBioChem 2012, 13, 85 – 96
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
93

S. Milewski et al.
Conclusions
N3-Oxoacyl derivatives of l-2,3-diaminopropanoic acid react
with CEE, CGIF tetrapeptide and GlcN-6-P synthase. Interac-
tions of the phenyl rings present in some of these compounds
with the indole ring of the tryptophan residue, which consti-
tutes the “molecular gate” at the enzyme active centre, strong-
ly influence inhibitor binding. The bound inhibitors react with
the N-terminal catalytic Cys1 residue in the same way as the
low-molecular-weight mimics of the enzyme N terminus (i.e.,
CEE and CGIF). Formation of the CꢀS bond upon attack of the
cysteine thiolane at the electrophilic centre of an inhibitor (ep-
oxide or conjugated carbon–carbon double bond) is the first
step of these reactions. The presence of an appropriately posi-
tioned keto functionality in the compounds studied affords an
opportunity for subsequent cyclisation, as a result of the Schiff
base formation upon reaction with the a-amino group of a
cysteinyl substrate. Substantial differences in the target of the
initial nucleophilic attack and further rearrangements of an in-
termediate were found for the structurally related compounds,
1 and 3. Attack at C6, followed by formation of the substituted
1,4-thiazine-3-en derivative are elements of reactions involving
compound 1, which contains an epoxide functionality. On the
other hand, 3, which contains a conjugated double bond, is
attacked at C5, and the product of a simple Michael-type addi-
tion does not appear to undergo any further rearrangement.
Experimental Section
Chemicals: l-2,3-diaminopropanoic acid derivatives 1–4 were syn-
thesised as described previously.[11] The CGIF tetrapeptide was syn-
thesised manually by a solid-phase peptide synthesis method by
using the Fmoc/But strategy. Other chemicals, including CEE, were
from Sigma.
Scheme 4. Alternative mechanism of reaction between 1 and CEE.
et al. for 5 with the bacterial enzyme.[25] The only similarity is
the fact that both mechanisms are valid for model systems
(l-cysteine or its ester, N-terminal oligopeptide) and for the
whole enzyme. Probably, this is because Cys1 is the only cata-
lytic residue at the active centre of GAH, and therefore the
whole enzyme might react with glutamine analogues like the
large, a-carboxyl substituted cysteine derivative. All the other
aspects of these mechanisms (sequence of the steps, structure
of the intermediates, final products) are different and seem to
be specific for particular types of inactivator structure. In the
reaction involving 5, the presence of a methyl ester group is
crucial for the first-step cyclisation, thus leading to the forma-
tion of the substituted succinimide intermediate, promoted
under denaturing conditions.[26] However, the initial cyclisation
depicted in Scheme 1 is possible because of the properly posi-
tioned keto functionality in 1, thereby giving rise to the Schiff
base formation. Therefore, the reaction of 1 with GlcN-6-P syn-
thase represents an entirely novel mechanism of enzyme in-
activation by a glutamine analogue, with the formation of an
exceptional, specific, chromophoric ring system as the final
product.
Plasmids, yeast and bacterial strains and culture conditions: S.
cerevisiae BJ1991 (MATa, pep4–3, prb1, ura3, leu2, trp1) was provid-
ed by I. Purvis (Glaxo Group Research, Greenwood, UK). E. coli
DH5aF9 (Life Technologies, Carlsbad, CA) was used for plasmid se-
lection and amplification. The YEpGW42 plasmid (8.7 kb), carrying
the S. cerevisiae GFA1 gene on a 3.5 kb EcoRI fragment inserted
into YEp352,[26] was a gift from W. Tanner (Regensburg, Germany).
YEpMA91 was a yeast shuttle vector carrying the LEU2 marker and
the promoter and terminator from PKG1 separated by a BglII
site.[27] The YRS-23–3 strain overproducing GlcN-6-P synthase was
obtained by transformation of S. cerevisiae BJ 1991 cells with the
YEpRS23–3 plasmid, containing the GFA1 gene under control of
the PGK1 promoter, and based on the YEpMA91 vector. Detailed
protocols for plasmid construction and yeast transformation are
those described previously for the preparation of S. cerevisiae cells
overproducing C. albicans GlcN-6-P synthase.[13] Yeast cells were
grown in YPD medium (2% glucose, 2% Bactopeptone, 1% yeast
extract).
Purification of the enzyme
I. Preparation of crude extract. YRS-23–3 cells (10 g wet weight)
from an overnight culture in YPD were harvested by centrifugation
(5000g, 10 min) and washed with buffer A (potassium phosphate
(20 mm, pH 7), EDTA (1 mm)). Cells were mixed (1:1 v/v) with buf-
fer B (potassium phosphate (20 mm, pH 7), EDTA (1 mm), dithio-
94
www.chembiochem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2012, 13, 85 – 96

Inactivation of GlcN-6-P Synthase by Glutamine Analogues
threitol (DTT, 1 mm)) and then disrupted by using a French press.
Cell debris was spun down (15000g, 20 min), and the supernatant
was saved as a crude extract.
Samples (0.5 mL) were collected, and pH and GlcN-6-P synthase ac-
tivity were measured.
Determination of kinetics of inactivation of GlcN-6-P synthase
by glutamine analogues: Incubation mixtures containing of GlcN-
6-P synthase (5 mg), potassium phosphate (50 mm, pH 7.0), EDTA
(1 mm), inactivators at various concentrations and Fru-6-P (10 mm)
if necessary, in a total volume of 1 mL were incubated at 258C. To
follow the inactivation of the enzyme, aliquots (200 mL) were with-
drawn from the mixtures, applied to the tops of mini-columns
packed with gel slurry (1 mL; Sephadex G-25 equilibrated with po-
tassium buffer (50 mm, pH 7.0)) and centrifuged (500g, 1 min, 48C).
Under these conditions the unbound inhibitor was separated from
the enzyme, and protein was recovered in clean test tubes. Appro-
priate eluent aliquots were used for the determination of the resid-
ual enzyme activity.
II. Protamine treatment. A solution containing 1% protamine sulfate
in buffer B was added dropwise to the crude extract (1 mL per
140 mg of protein present in the crude extract) and stirred moder-
ately. The precipitated solid was removed by centrifugation
(15000g, 20 min) and the supernatant was saved.
III. Ammonium sulfate precipitation. Ammonium sulfate solution in
buffer B (80% saturation) was added dropwise to the protamine
sulfate supernatant, and stirred gently until 55% saturation with
respect to ammonium sulfate was reached. The obtained suspen-
sion was centrifuged (15000g, 20 min), the supernatant was dis-
carded and precipitate was dissolved in buffer B (10 mL).
IV. Polyethylene glycol precipitation. A solution containing 50% poly-
ethylene glycol (MW 6000–7500, 2.5 mL) was added dropwise to
the gently stirred solution from the previous step. The obtained
suspension was centrifuged (15000g, 20 min), the supernatant was
discarded and the precipitate was dissolved in a minimal amount
of buffer C (composition as for B, supplemented by Fru-6-P
(1 mm)).
Reaction of CEE with glutamine analogues or CGIF tetrapeptide:
Equimolar amounts of CEE or CGIF tetrapeptide with 1 or 3
(1 mmol each) were dissolved in oxygen-free potassium phosphate
buffer (10 mL, 5 mm, pH 5.0, 7.0 or 8.0), and the mixtures were
kept at 258C under argon. Samples of the reaction mixture
(0.2 mL) were collected at various time intervals. Aliquots (50 mL)
were taken for immediate determination of free thiol content, and
components present in the remaining aliquots (150 mL) were sepa-
rated by RP-FPLC on an RPC column. Elution was with methanol/
water, and detection was at 212 nm. Fractions containing compo-
nents were pooled, evaporated to remove methanol, frozen and
lyophilised. Final products of reactions between CEE and com-
pounds 1 and 3, (i.e., 6 and 7, respectively) were isolated and ana-
lysed by NMR. The NMR spectra of product 6 from HMBC and
HSQC experiments are provided in the Supporting Information.
V. Ion-exchange chromatography. A solution from the previous step
was loaded onto a Resource Q FPLC column equilibrated with buf-
fer D (Tris-HCl (25 mm, pH 7.5), EDTA (1 mm), DTT (1 mm), Fru-6-P
(1 mm)). The column was washed with buffer D (5 mL) and elution
was performed with a linear KCl gradient (0–0.5m) in buffer D at
1.0 mLminꢀ1. Active fractions were pooled and concentrated by ul-
trafiltration with a Centricon 10 device.
1
VI. Size-exclusion chromatography. The pooled concentrated active
fraction from the previous step was loaded on a Superdex 200 HR
10/30 column equilibrated with buffer D containing NaCl (0.15m).
Protein was eluted with the same buffer (0.5 mLminꢀ1). Active frac-
tions were pooled. Steps I–IV were run at 48C, and steps V and VI
were run at room temperature.
Data for product 7: H NMR (300 MHz, D2O): d=7.68–7.45 (m, 5H;
Ph), 4.29 (q, J=4.8 Hz, 2H; CH3CH2O), 4.16 (t, J=11.1 Hz, 1H;
NHC(O)CHSCH2), 3.98 (t, J=5.5 Hz 1H; SCH2(NH2)CHCOOEt), 3.91 (t,
J=8.3 Hz, 1H; HOOC(NH2)CHCH2NH), 3.65–3.54 (dd, J=6.5 Hz, 2H;
HOOC(NH2)CHCH2NH), 3.28 (d, J=6.7 Hz, 2H; SCHCH2C(O)Ph), 3.18–
3.11 (dd, J=12.3 Hz, 2H; SCH2CH(NH2)COOEt), 1.18 (t, J=6.6 Hz,
3H; CH3CH2O).
Determination of GlcN-6-P synthase activity: A standard incuba-
tion mixture consisted of Fru-6-P (10 mm), l-glutamine (10 mm),
EDTA (1 mm), DTT (1 mm) and potassium phosphate (50 mm,
pH 7.0), with appropriately diluted enzyme preparation and inhibi-
tors where necessary. Final concentration of the pure GlcN-6-P syn-
thase was 0.5–1.0 mgmLꢀ1. The reaction was started by adding the
enzyme, then incubated at 378C for 30 min and terminated by
boiling for 1 min. The concentration of GlcN-6-P produced by the
enzyme was determined by a modified Elson-Morgan procedure,[28]
and this increased linearly for at least 60 min. One unit of specific
activity was defined as an amount of enzyme that catalysed the
Determination of free thiol content: Aliquots (50 mL) were collect-
ed from the reaction mixtures and combined with potassium phos-
phate (900 mL, pH 7.0) and DTNB (50 mL, 3 mm). Mixtures were kept
for 5 min at room temperature, and then the absorption at 412 nm
was measured.
Inactivation of GlcN-6-P synthase and preparation of chymotryp-
tic/tryptic digests: Homogenous GlcN-6-P synthase (2 mg,
~6 nmol) dissolved in potassium phosphate buffer (20 mm, pH 7.0,
4 mL) was incubated at 258C under argon with 1 or 3 (1 mm) or
alone. Samples (20 mL) were collected and subjected to SDS-PAGE
electrophoresis. Native or inactivated protein present in the
sample were digested by sequencing-grade chymotrypsin or by
trypsin by using the in-gel proteolysis procedure of Shevchenko
et al.[30]
formation of 1 mmol GlcN-6-P minꢀ1 mg proteinꢀ1
.
Molecular weight determination: Gel filtration was performed on
a Superdex 200 HR 10/30, and eluted at 0.5 mLminꢀ1 with potassi-
um phosphate (25 mm, pH 6.8) containing NaCl (0.15m), DTT
(1 mm) and EDTA (1 mm). Protein elution was followed at 280 nm,
and GlcN-6-P synthase activity was measured colorimetrically in
0.5 mL samples. Discontinuous SDS-PAGE was performed by the
method of Laemmli,[29] with a 5% stacking gel and a 7.5% separat-
ing gel.
Molecular modelling: The structure of the receptor for docking
calculations was built on the basis of the PDB file 1GMS (complex
of the GAH domain of E. coli glucosamine-6-phosphate synthase
with g-glutamyl hydroxamate).[17] As the X-ray-derived protein
structure lacks all hydrogens, the hydrogen atoms bound to aro-
matic fragments and heteroatoms were added for the simulations
with the pdb2gmx tool included in the Gromacs package
(http:www.gromacs.org), and the complete structure was energy
minimised by using the gromos 43a2 forcefield.[31] The resulting
“minimised structure” was then used for the subsequent flexible
Determination of an isoelectric point: Chromatofocusing was per-
formed on a Mono P HR 5/5 column. The purified GlcN-6-P syn-
thase (2 mg) was dissolved in Bis·Tris·HCl, (25 mm, pH 6.3) as a start-
ing buffer, and a pH 6–4 gradient was generated during the elution
with Polybuffer 74 solution (20 mL, diluted 1:10 in water, pH 4).
ChemBioChem 2012, 13, 85 – 96
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
95

S. Milewski et al.
docking simulations with the AutoDock 4.2 suite of programs.[32]
AutoDock atom types and Gasteiger partial charges were assigned
to all atoms by means of the accompanying AutoDockTools
Python script prepare_receptor4. The grid box (60ꢂ60ꢂ60 grid-
points), embracing the entire binding site, was calculated by the
autogrid4 tool with default spacing of 0.375 ꢁ. The structures of li-
gands 1 and 3, as well as the structure of the putative cyclic prod-
uct of enzyme covalent inactivation, were built by using the In-
sightII molecular modelling environment from Accelrys.[33] The li-
gands were prepared for docking simulations by AutoDockTools
script prepare_ligand4. Ligand atom types were determined at this
stage; Gasteiger partial charges were calculated and assigned to all
atoms, and suitable single bonds were marked as flexible. The La-
marckian genetic algorithm (LGA) was used as the search protocol.
The actual docking simulations were carried out with 50 independ-
ent runs and initial populations of 150 solutions. A maximum of
25 million energy function evaluations and 27000 generations
were set to achieve convergence and avoid premature search-pro-
cedure termination. The rates of crossover and mutations were set
to 0.8 and 0.02, respectively. After the final docking, the resulting
ligand conformations were grouped with an RMSD clustering toler-
ance of 1.8 ꢁ for analysis. Atomic charge distribution in ligands
was calculated by means of the GAMESS package[34] at an HF/6–
31G** level to obtain the ESP charges.
[7] Y. Nakaishi, M. Bando, H. Shimizu, K. Watanabe, F. Goto, H. Tsuge, K.
Kondo, M. Komatsu, FEBS Lett. 2009, 583, 163–167.
[8] N. Floquet, C. Richez, P. Durand, B. Maigret, B. Badet, M.-A. Badet-Deni-
sot, Bioorg. Med. Chem. Lett. 2007, 17, 1966–1970.
[9] S. Milewski, H. Chmara, R. Andruszkiewicz, E. Borowski, Biochim. Biophys.
Acta Protein Struct. Mol. Enzymol. 1985, 828, 247–254.
[10] N. Kucharczyk, M.-A. Denisot, F. Le Goffic, B. Badet, Biochemistry 1990,
29, 3668–3676.
[11] R. Andruszkiewicz, R. Je˛drzejczak, T. Zieniawa, M. Wojciechowski, E. Bor-
owski, J. Enzyme Inhib. 2000, 15, 429–441.
[12] B. Badet, P. Vermoote, P. Y. Haumont, F. Lederer, F. Le Goffic, Biochemistry
1987, 26, 1940–1948.
[13] S. Milewski, D. Kuszczak, R. Je˛drzejczak, R. J. Smith, A. J. P. Brown, G.
Gooday, J. Biol. Chem. 1999, 274, 4000–4008.
[14] R. Andruszkiewicz, H. Chmara, S. Milewski, E. Borowski, Int. J. Peptide
Protein Res. 1986, 27, 449–453.
[15] R. Andruszkiewicz, S. Milewski, E. Borowski, J. Enzyme Inhib. 1995, 9,
123–133.
[16] A. Teplyakov, G. Obmolova, B. Badet, M.-A. Badet-Denisot, J. Mol. Biol.
2001, 313, 1093–1102.
[17] M. N. Isupov, G. Obmolova, S. Butterworth, M.-A. Badet-Denisot, B.
Badet, I. Polikarpov, J. A. Littlechild, A. Teplyakov, Structure 1996, 4,
801–810.
ˇ
[18] E. C. Lee, D. Kim, P. Jurecka, P. Tarakeshwar, P. Hobza, K. S. Kim, J. Phys.
Chem. A 2007, 111, 3446–3457.
[19] S. Mouilleron, M.-A. Badet-Denisot, B. Gollinelli-Pimpaneau, J. Biol.
Chem. 2006, 281, 4404–4412.
[20] S. Mouilleron, M.-A. Badet-Denisot, B. Golinelli-Pimpaneau, J. Mol. Biol.
2008, 377, 1174–1185.
[21] A. Teplyakov, C. Leriche, G. Obmolova, B. Badet, M.-A. Badet-Denisot,
Nat. Prod. Rep. 2002, 19, 60–69.
[22] M. A. Hollenhorst, I. Ntai, B. Badet, N. L. Kelleher, C. T. Walsh, Biochem-
istry 2011, 50, 3859–3861.
[23] C. Machacek-Pitsch, P. Rauschenbach, H. Simon, Biol. Chem. Hoppe-
Seyler 1985, 366, 1057–1062.
Other methods: Protein concentration was assayed by the Brad-
ford procedure[35] with bovine serum albumin as a standard. The
MS ESI spectra of substrates and products of the model reactions
were taken with the Agilent 1100 LC MS system equipped with the
qadrupole detector. MALDI-TOF of the peptide mixtures present in
chymotryptic/tryptic digests was performed in an Applied Biosys-
tem Voyager-DE STR. NMR spectra were taken by using Varianplus
300 MHz and 500 MHz instruments.
[24] Y. Yabe, D. Guillaume, D. H. Rich, J. Am. Chem. Soc. 1988, 110, 4043–
4044.
[25] B. Badet, P. Vermoote, F. Le Goffic, Biochemistry 1988, 27, 2282–2287.
[26] G. Watzele, W. Tanner, J. Biol. Chem. 1989, 264, 8753–8758.
[27] S. M. Kingsman, D. Cousens, C. A. Stanway, A. Chambers, M. Wilson, A. J.
Kingsman, Methods Enzymol. 1990, 185, 329–341.
[28] M. Kenig, E. Vandamme, E. P. Abraham, J. Gen. Microbiol. 1976, 94, 46–
54.
Acknowledgements
Financial support of these studies by the Polish Ministry for Sci-
ence and Higher Education is gratefully acknowledged.
[29] U. K. Laemmli, Nature 1970, 227, 680–685.
[30] A. Shevchenko, M. Wilm, O. Vorm, M. Mann, Anal. Chem. 1996, 68, 850–
858.
Keywords: enzyme inhibitors
modeling · reaction mechanisms
·
epoxides
·
molecular
[31] a) B. Hess, C. Kutzner, D. van der Spoel, E. Lindahl, J. Chem. Theory
Comput. 2008, 4, 435–447; b) H. J. C. Berendsen, D. van der Spoel, R.
van Drunen, Comput. Physics Commun. 1995, 91, 43–56.
[32] G. M. Morris, R. Huey, W. Lindstrom, M. F. Sanner, R. K. Belew, D. S. Good-
sell, A. J. Olson, J. Comput. Chem. 2009, 30, 2785–2791.
[33] Insight II; Accelrys Inc., San Diego, CA.
[34] M. W. Schmidt, K. K. Baldridge, J. A. Boatz, S. T. Elbert, M. S. Gordon, J. H.
Jensen, S. Koseki, N. Matsunaga, K. A. Nguyen, S. Su, T. L. Windus, M.
Dupuis, J. A. Montgomery, Jr., J. Comput. Chem. 1993, 14, 1347–1363.
[35] M. M. Bradford, Anal. Biochem. 1976, 72, 248–254.
[1] F. Massiꢃre, M.-A. Badet-Denisot, Cell. Mol. Life Sci. 1998, 54, 205–222.
[2] E. Borowski, Il Farmaco 2000, 55, 206–208.
[3] K.-C. Chou, J. Proteome Res. 2004, 3, 1284–1288.
[4] For reviews, see: a) S. Milewski, Biochim. Biophys. Acta Protein Struct.
Mol. Enzymol. 2002, 1597, 173–192; b) P. Durand, B. Golinelli-Pimpa-
neau, S. Mouilleron, B. Badet, M.-A. Badet-Denisot, Arch. Biochem. Bio-
phys. 2008, 474, 302–317.
[5] S. Mouilleron, M.-A. Badet-Denisot, B. Badet, B. Golinelli-Pimpaneau,
Arch. Biochem. Biophys. 2011, 505, 1–12.
[6] J. Raczynska, J. Olchowy, P. V. Konariev, D. I. Svergun, S. Milewski, W.
Rypniewski, J. Mol. Biol. 2007, 372, 672–678.
Received: September 15, 2011
Published online on November 28, 2011
96
www.chembiochem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2012, 13, 85 – 96