Tài liệu Báo cáo Y học: The glucose-specific carrier of the Escherichia coli phosphotransferase system Synthesis of selective inhibitors and inactivation studies pptx

The glucose-specific carrier of the
Escherichia coli
phosphotransferase
system
Synthesis of selective inhibitors and inactivation studies
Luis Fernando Garcı
´
a-Alles, Vera Navdaeva, Simon Haenni and Bernhard Erni
Departement fu
¨
r Chemie und Biochemie, Universita
¨
t Bern, Freiestrasse 3, CH-3012, Bern, Switzerland

Thirteen glucose analogues bearing electrophilic groups
were synthesized (five of them for the first time) and screened
as inhibitors of the glucose transporter (EII
Glc
)ofthe
Escherichia coli phosphoenolpyruvate–sugar phospho-
transferase system (PTS). 2¢,3¢-Epoxypropyl b-
D
-glucopyr-
anoside (3a) is an inhibitor and also a pseudosubstrate. Five
analogues are inhibitors of nonvectorial Glc phosphorylation
by EII

Glc
but not pseudosubstrates. They are selective for
EII
Glc
as demonstrated by comparison with EII
Man
, another
Glc-specific but structurally different transporter. 3a is the
only analogue that inhibits EII
Glc
by binding to the high-
affinity cytoplasmic binding site and also strongly inhibits

sugar uptake mediated by this transporter. The most potent
inhibitor in vitro,methyl6,7-anhydro-
D
,
L
-glycero-a-
D
-
gluco-heptopyranoside (1d), preferentially interacts with the
low-affinity cytoplasmic site but only weakly inhibits Glc
uptake. Binding and/or phosphorylation from the cyto-
plasmicsideofEII

Glc
is more permissive than sugar binding
and/or translocation of substrates via the periplasmic site.
EII
Glc
is rapidly inactivated by the 6-O-bromoacetyl esters of
methyl a-
D
-glucopyranoside (1a)andmethyla-
D
-manno-
pyranoside (1c), methyl 6-deoxy-6-isothiocyanato-a-

D
-
glucopyranoside (1e), b-
D
-glucopyranosyl isothiocyanate
(3c)andb-
D
-glucopyranosyl phenyl isothiocyanate (3d).
Phosphorylation of EII
Glc
protects, indicating that inacti-
vation occurs by alkylation of Cys421. Glc does not protect,

but sensitizes EII
Glc
for inactivation by 1e and 3d,whichis
interpreted as the effect of glucose-induced conformational
changes in the dimeric transporter. Glc also sensitizes EII
Glc
for inactivation by 1a and 1c of uptake by starved cells. This
indicates that Cys421 which is located on the cytoplasmic
domain of EII
Glc
becomes transiently accessible to substrate
analogues on the periplasmic side of the transporter.

Keywords: binding site; carbohydrate chemistry; cysteine;
glucose transporter; irreversible inhibitor.
Escherichia coli has two transporters for glucose, EII
Glc
(IIA
Glc
-IICB
Glc
) [1] and EII
Man
(IIAB
Man

-IIC
Man
-IID
Man
)
[2,3], which mediate uptake concomitant with phosphory-
lation of their substrates. The immediate source of high-
energy phosphate is the phosphoryl carrier protein HPr
which in turn is phosphorylated by phosphoenolpyruvate in
a reaction catalysed by enzyme I (EI). EI and HPr together
with the carbohydrate transporters (enzymes II, EII
sugar

)of
diverse specificity and structure are components of the
bacterial phosphoenolpyruvate–sugar phosphotransferase
system (PTS) [4]. The PTS in addition comprises a number
of proteins that act as allosteric regulators of enzymes and/
or transcription factors.
The PTS transporters are homodimers, as indicated by
cross-linking, ultracentrifugation, gel filtration, interallelic
complementation and cryo-electron crystallography [5–9].
One protomer comprises three (or four) functional units,
IIA, IIB and IIC(IID), which occur either as protein
subunits or as domains in polypeptide chains. IIA and IIB

sequentially transfer phosphoryl groups from HPr to the
transported sugars. IIC contains the major determinants for
sugar recognition and translocation, as inferred from
binding studies [10] and the substrate selectivity of a
chimeric EII
GlcNAc/Glc
[11]. EI, HPr and IIA are phos-
phorylated at His, whereas IIB domains are phosphorylated
at Cys421 in EII
Glc
and at His175 in EII
Man

.EII
Glc
is
specific for Glc, but EII
Man
has a broader substrate
specificity for Glc, Man, and other derivatives of Glc
altered at the C-2 carbon. Both transporters phosphorylate
their hexose substrates at OH-6. In spite of their overlapping
substrate specificity and analogous mechanism of action,
EII
Glc

and EII
Man
do not share amino-acid sequence
similarity, and, as judged by the known X-ray structures
of their cytoplasmic domains, also assume completely
different folds (for a review see [12]). The topology of the
membrane-spanning units IIC
Glc
and IIC
Man
-IID
Man

are
also different, as judged by the characterization of protein
fusions between C-terminally truncated IIC(D) domains
with alkaline phosphatase and b-galactosidase [13,14].
Whereas the sites of EII phosphorylation are known and
easily recognized from the invariant amino-acid sequence
motifs, residues participating in sugar binding have not been
identified. Each protomer has been proposed to have a
sugar-binding site of its own with the two sites in the dimer
being distinguished by their different affinity for the
substrate [15]. Both sites are simultaneously accessible from
the cytoplasmic face. The IICB

Glc
subunits co-operate in so
Correspondence to L. F. Garcı
´
a Alles, Departement fu
¨
rChemieund
Biochemie, Universita
¨
t Bern, Freiestrasse 3, CH-3012 Bern,
Schweiz. Fax: + 41 31 631 48 87, Tel.: + 41 31 631 37 92,
E-mail:

Abbreviations: PTS, phosphoenolpyruvate–sugar phosphotransferase
system; aMGlc, methyl a-
D
-glucopyranoside; 2dGlc, 2-deoxy-
D
-glucose;
IC
50
, half inhibitory concentration; FC, flash chromatography.
(Received 9 June 2002, revised 16 August 2002,
accepted 21 August 2002)
Eur. J. Biochem. 269, 4969–4980 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03197.x

far as phosphoryl transfer from Cys421 on the IIB domain
of one subunit to Glc bound to the IIC domain of the
second subunit is possible [7]. However, whether and how
the substrate-binding sites on the IIC domains reorient
themselves with respect to the extracellular and cytoplasmic
compartment, and how they interact with phosphorylation
domains (IIB), is the objective of continuing research.
With the aim of finding selective irreversible inhibitors of
the Glc-specific transporters EII
Glc
and EII
Man

and of
eventually identifying their substrate-binding sites, 13 glu-
cose analogues have been synthesized (compounds 1a)3d,
Scheme 1). Epoxides, a-halocarbonyls, isothiocynates or
a,b-unsaturated esters were introduced at C-1, where
modifications are expected to be tolerated by EII
Glc
[15],
and at C-6 because of its presumed proximity to the
catalytic residues in the active site of the transporter.
Similarly modified carbohydrates have been used previously
to study sugar-recognizing enzymes, e.g. epoxypropyl

derivatives of N-acetyl-
D
-glucosamine to label lysozyme
[16], maltosyl isothiocyanate to label the human erythrocyte
Glc transporter [17], and N-bromoacetylglucosamine to
label hexokinase [18]. The glucose analogues have been
characterized as pseudosubstrates and as reversible and
irreversible inhibitors of EII
Glc
and EII
Man
. Two assays

were employed: (a) nonvectorial phosphorylation of Glc
(analogues) by solubilized and purified EII and by
EII-containing membrane fractions – this in vitro assay
was used to examine how sugar recognition is effected from
the cytoplasmic side of the transporters (a consequence of
the inside-out orientation of EII in membrane vesicles); (b)
inhibition of Glc uptake by starved cells expressing either
EII
Glc
or EII
Man
. These experiments served to study binding

to EII from the periplasmic space.
MATERIALS AND METHODS
Materials, bacterial strains and proteins
Starting materials for the preparation of compounds 1a)3d,
and other components were purchased from commercial
sources as specified previously [15]. Organic solvents of the
highest purity available were dried following standard
procedures. The membrane transporters were overexpressed
and purified from an Escherichia coli K12 strain
ZSC112LDG(glk manZ DptsG:Cm) [19]. The plasmid
pTSGH11 encodes under the control of Ptac a IICB
Glc

with
a C-terminal hexahistidine tag [7]. The plasmid pJFLPM
encodes the three subunits of the Escherichia coli mannose
transporter under the control of the Ptac promoter [20].
Membranes containing EII
Glc
and EII
Man
, and purified
EII
Glc
, EI, HPr, IIA

Glc
and IIAB
Man
were prepared as
described in [15].
In vitro
phosphotransferase assays
Pyruvate evolution was measured spectrophotometrically in
96-well microtiter plates at 30 °C, using the coupled assay
with
L
-lactate dehydrogenase and NADH. Final assay

concentrations were: 0.5 l
M
EI, 1 l
M
HPr, 15 l
M
IIA
Glc
(or
0.5 l
M
IIAB

Man
) and 0.0013 lLÆlL
)1
membrane extract.
Other conditions were as described in [15]. Typical back-
ground activities of 2 lmolÆmin
)1
were measured in the
absence of sugar. They were subtracted before subsequent
calculations. Half inhibitory concentrations (IC
50
)were

determined by measuring the phosphorylation rate of
0.5 m
MD
-Glc, in the presence of 0–5 m
M
concentrations
of the inhibitors. Glc6P was detected in these experiments
using Glc6P dehydrogenase (1 UÆmL
)1
)and1m
M
NADP

[21].
Fitting of kinetic data using
DYNAFIT
This software is available free of charge at http://www.
biokin.com [22]. The kinetic constants were estimated as
reported [15].
Inactivation of
nonvectorial
phosphorylation
A 10-lL sample of inhibitor in buffer A [50 m
M
Hepes,

pH 7.5, 4 m
M
dithiothreitol, 5 m
M
MgCl
2
,1mgÆmL
)1
BSA, 0.5 mgÆmL
)1
egg yolk lecithin (Sigma)] was preincu-
bated for 10 min at 30 °C,andthenaddedto40–60lL

purified IICB
Glc
in buffer A (final concentration: 10–3 l
M
).
Then 10-lL aliquots were withdrawn at intervals and
dilutedinto290lL buffer A at 4 °C. The diluted IICB
Glc
samples were assayed for in vitro PTS activity using the
D
-Glc6P dehydrogenase assay. The final concentrations in
the activity assay were 0.08–0.04 l

M
IICB
Glc
and 2 m
M
Glc.
In vivo
transport inhibition
Uptake of [
14
C]methyl a-
D

-glucopyranoside ([
14
C]aMGlc)
by starved E. coli K12 ZSC112LDG cells expressing EII
Glc
or of 2-deoxy-
D
-[
14
C]glucose ([
14
C]2dGlc) by starved cells

Scheme 1.
4970 L. F. Garcı
´
a-Alles et al.(Eur. J. Biochem. 269) Ó FEBS 2002
expressing EII
Man
was assayed as described previously [15].
Transport rates were calculated from the amount of
[
14
C]sugar accumulated inside the cells, typically after 5,
15, 25, 40 and 120 s.

Inactivation of uptake by starved cells
E. coli K12 ZSC112LDG cells expressing either EII
Glc
or
EII
Man
were prepared as described previously [15]. To
0.65 mL cell suspension in M9 medium (0.2–0.1 gÆmL
)1
)at
room temperature were added 26 lL of a stock solution of
the irreversible inhibitor (0.5–0.025

M
), with or without Glc
(0.25
M
). Aliquots (0.15 mL) were withdrawn at the indi-
cated time points and diluted into ice-cold M9 medium
(0.85 mL). The cells were collected by centrifugation and
resuspendedin1mLfreshM9medium.Thewashedcells
were then assayed for uptake activity as described above.
Synthesis of compounds 1a-3d
6-O-Bromoacetyl derivatives (1a,c) [23], methyl 6-deoxy-6-
isothiocyanato-a-

D
-glucopyranoside (1e)[24],andb-
D
-
glucopyranosyl isothiocyanate (3c)[25]werepreparedby
the reported procedures. Methyl 6-O-chloroacetyl-a-
D
-
glucopyranoside (1b) was prepared like 1a using chloroace-
tyl chloride instead of bromoacetyl bromide. The yield was
56% after flash chromatography (FC) (ethyl acetate/meth-
anol, 96 : 4, v/v):

1
HNMR(CD
3
OD) d:4.66(1H,d,J ¼
3.7 Hz, H1), 4.47 (1H, dd, J ¼ 11.8, 2.2 Hz), 4.31 (1H, dd,
J ¼ 11.8, 6.3 Hz), 4.24 (2H, s, CH
2
Cl), 3.73 (1H, m), 3.61
(1H, dd, J ¼ 9.5, 8.8 Hz), 3.40 (3H, s. CH
3
O), 3.39 (1H,
overlapped), 3.30 (1H, m).

13
CNMR(CD
3
OD) d:169.1,
101.3, 75.1, 73.5, 71.8, 70.9, 66.3, 55.7, 41.7. MS (ESI) m/z
293 ([M + Na]
+
, 40%). Methyl 2,3-anhydro-a-
D
-allopyr-
anoside (2a) and methyl 3,4-anhydro-a-
D

-galactopyrano-
side (2b) were obtained by desilylation of 1 mmol of the
6-O-[dimethyl-(1,1,2-trimethylpropyl)silyl]-protected forms
[26] with CsF (3 mmol) in dimethylformamide (50 mL) at
110 °C, for 30 min. Evaporation in vacuo of dimethylform-
amide, and FC (ethyl acetate/methanol, 96 : 4, v/v)
afforded 2a in 81% yield [26] and 2b in 48% yield [27].
b-
D
-
G
lucopyranosyl phenyl isothiocyanate (3d)was

purchased (Sigma).
Preparation of methyl 6,7-anhydro-
D
,
L
-glycero-a-
D
-
glucoheptopyranoside (1d)
Methyl 2,3,4-tri-O-benzyl-a-
D
-glucopyranoside (4)was

obtained from methyl a-
D
-glucopyranoside, following con-
ventional sugar transformations [28]. A solution of 4 (2.6 g,
5.6 mmol) in 12 mL dichloromethane was added dropwise
to a suspension of activated powdered 4 A
˚
molecular sieves
(11 g) and pyridinium chlorochromate (5.6 g, 26 mmol) in
dry dichloromethane (80 mL). The resulting mixture was
stirred for 10 min at room temperature, 80 mL hexane was
added, and the mixture was filtered through a pad of silica

gel. Elution with ethyl acetate/hexane (1 : 1, v/v) and
concentration furnished 1.92 g (75%) methyl 2,3,4-tri-
O-benzyl-a-
D
-glucohexodialdo-1,5-pyranoside (5) [29]. The
aldehyde 5 (1 g, 2.1 mmol, in 5 mL of dry tetrahydrofurane)
was slowly added under argon to a flask containing
methyltriphenylphosphonium ylide (3.4 mmol) in 40 mL
dry tetrahydrofurane at )78 °C [30]. After 15 min, the
cooling bath was removed and stirring was continued for
1.5 h. The reaction was stopped at 0 °C by the addition of
10 mL methanol. After concentration, 150 mL diethyl ether

was added. The solution was washed with brine (2 · 50 mL),
and the organic phase dried over MgSO
4
. Concentration
followed by FC (hexane/ethyl acetate 8 : 2, v/v) gave 0.69 g
(70%) methyl 6,7-dideoxy-2,3,4-tri-O-benzyl-a-
D
-gluco-
hept-6-enopyranoside (6) [31]. A solution of compound 6
(0.2 g, 0.43 mmol) in 5 mL dry dichloromethane was stirred
at room temperature with 3-chloroperoxybenzoic acid
(0.5 g, 2.9 mmol) for 15 h. Diethyl ether (50 mL) was added

and the solution was washed with 0.12
M
aqueous Na
2
S
2
O
3
(3 · 20 mL), saturated NaHCO
3
(2 · 20 mL) and brine
(20 mL). The ether phase was dried over MgSO

4
, filtered and
evaporated, yielding 0.18 g of the mixture of epoxides (7),
92%. Compound 7 (0.12 g, 0.25 mmol) in 6 mL 4.4%
formic acid in methanol was added to a suspension of 0.3 g
palladium black in 8 mL 4.4% formic acid in methanol [32].
After 30 min, the catalyst was removed by filtration through
celite, and washing with methanol (2 · 5 mL). After eva-
poration of the solvent, 76 mg (81%) of the diastereomeric
mixture of 1d was recovered.
1
HNMR(CD

3
OD) d:4.85(1H,
m, H1), 3.92–3.64 (3H,m), 3.60–3.30 (3H, m), 3.55 (3H, s.
CH
3
O), 3.05–2.87 (1H, m).
13
CNMR(CD
3
OD) d:101.7,
75.5, 75.4, 74.2, 73.8, 73.6, 73.5, 56.1, 53.7, 53.5, 45.8, 44.8.
MS (ESI) m/z 229 ([M + Na]

+
, 100%).
Synthesis of methyl (6
E
)-6,7-dideoxy-a-
D
-gluco-oct-6-
enopyranosiduronic acid (1g) and its methyl ester (1f)
Methyl 2,3,4-tris-O-(trimethylsilyl)-a-
D
-gluco-hexodialdo-
1,5-pyranoside (9) (0.3 g, 0.73 mmol), prepared as described

in ref [15], was olefinated at room temperature with methyl
triphenylphosphoranylidene acetate (0.33 g, 1 mmol) in
5 mL dry dichloroethane. After 2 h reaction, the solvent
was removed by evaporation, and 10 was purified by FC
(hexane/ethyl acetate, 92 : 8, v/v): 0.2 g, 44% yield. Com-
pound 10 (0.1 g, 0.21 mmol) was desilylated by stirring for
2 h at room temperature with methanol (2 mL) and K
2
CO
3
(2 mg). After evaporation 54 mg 1f (100%) was obtained :
1

HNMR(CD
3
OD) d:7.29(1H,dd,J ¼ 15.8,4.4Hz,H6),
6.31 (1H, dd, J ¼ 15.8,1.8 Hz,H7),4.93(1H,d,J ¼ 4.0 Hz,
H1), 4.13 (1H, ddd, J ¼ 9.9, 4.4, 1.5 Hz, H5), 3.84 (1H, dd,
J ¼ 9.6, 8.8 Hz, H3), 3.61 (1H, dd, J ¼ 9.6, 3.7 Hz, H2), 3.58
(3H, s, CH
3
O),3.52(3H,s,CH
3
O), 3.29 (1H, ddd, J ¼ 9.9,
8.8, 1.1 Hz, H4);

13
CNMR(CD
3
OD) d: 168.8, 147.0, 122.1,
101.8, 75.7, 75.4, 73.7, 71.8, 56.2, 52.4. MS (ESI) m/z 287
([M + K]
+
, 100%). Hydrolysis of 1f (23 mg, 0.09 mmol,
0.1
M
solution in water) was effected at pH 12 (1
M

KOH)
for 2 h. The derivative 1g was obtained as a potassium salt in
quantitative yield:
1
HNMR(D
2
O, pD 7) d: 6.51 (1H, dd, J ¼
15.8, 7.0 Hz, H6), 6.11 (1H, dd, J ¼ 15.8, 1.1 Hz, H7), 4.80
(1H, d, J ¼ 3.7 Hz, H1, overlapped with water signal), 4.14
(1H, dd, J ¼ 9.6, 7.0 Hz, H5), 3.70–3.58 (2H, m), 3.41 (3H, s,
CH
3

O), 3.33 (1H, dd, J ¼ 9.6, 8.8, 1.1 Hz, H4);
13
CNMR
(D
2
O, pD 7) d: 177.2, 140.6, 133.5, 102.1, 75.2, 75.4, 73.8,
73.7, 57.9. MS (ESI) m/z 233 ([M-H]
–
, 100%).
Synthesis of 2¢,3¢-epoxypropyl b-
D
-glucopyranoside (3a)

The allyl b-
D
-glucopyranoside (12) was prepared by
reaction of acetobromoglucose (11) with allyl alcohol
[33].Anice-cooledsolutionof12 (90 mg, 0.23 mmol) in
Ó FEBS 2002 Inhibition of EII
Glc
(Eur. J. Biochem. 269) 4971
2 mL dichloroethane was treated with freshly prepared
dimethyldioxirane in acetone (3 mL,  0.3 mmol) [34].
After 1 h, the ice bath was removed, and dimethyldioxi-
rane (2 mL) was added after 2 and 4 h. The reaction was

continued overnight. Solvent was removed by evapor-
ation, and the resulting epoxypropyl (13)(93mg)was
deacetylated as described [35], to furnish 3a as a 6 : 4
diastereomeric mixture:
1
HNMR(CD
3
OD) d:4.52(0.4H,
d, J ¼ 7.7 Hz, H1), 4.49 (0.6H, d, J ¼ 7.7 Hz, H1), 4.31–
3.68 (4H, m), 3.55–3.38 (5H, m), 2.98 (1H, m, CH
2
epoxide), 2.88–2.81 (1H, m, CH

2
epoxide);
13
CNMR
(CD
3
OD) d: 104.4, 104.3, 77.9, 75.1, 75.0, 71.6, 71.5, 71.4,
71.2, 62.8, 62.7, 52.0, 51.8, 45.2, 45.0. MS (ESI) m/z 259
([M + Na]
+
, 100%) [35].
The diastereomeric mixture 3a was also prepared by

following an alternative route. Tri-O-acetyl-
D
-glucal (15)
(3 g, 10.8 mmol) was refluxed with benzyl chloride (24 mL),
KOH (9.4 g) and toluene (20 mL). After 5 h, the solution
was concentrated in vacuo. Then 150 mL diethyl ether was
added and the solution was washed with water
(2 · 100 mL) and saturated NaHCO
3
(100 mL). The
organic phase was dried over MgSO
4

, filtered, and concen-
trated. The residue was chromatographed (hexane/diethyl
ether, 7 : 3, v/v) giving 2.3 g (52%) of 16. The epoxide 17
was then prepared from 16 by reaction with dimethyldioxi-
rane, as described [36]. The derivative 17 (0.24 g,
0.55 mmol) was treated with 5 mL racemic glycidol at
room temperature. After 2 h reaction, the excess glycidol
was removed under vacuum, and the residue was chroma-
tographed with diethyl ether, giving 0.13 g of 14 [37]. The
epoxide 14 (50 mg, 0.1 mmol) was debenzylated in 30 min
by following the same method as for 1d.23mgofa1:1
diastereomeric mixture of 3a was obtained.

Preparation of chloroacetyl b-
D
-glucopyranoside (3b)
The epoxide 17 (0.2 g, 0.46 mmol) was treated with a
solution of chloroacetic acid (0.11 g, 1.15 mmol) in dry
dichloromethane (10 mL). The mixture was stirred at room
temperature overnight. Evaporation and FC (hexane/
diethyl ether, 1 : 1, v/v) furnished 0.17 g (70%) chloroacetyl
3,4,6-tri-O-benzyl-b-
D
-glucopyranoside (18). Removal of
the benzyl groups, as described for compound 1d (see

above), and FC (ethyl acetate/methanol, 9 : 1, v/v) resulted
in 73 mg (88%) compound 3b:
1
HNMR(CD
3
OD) d:5.73
(1H, d, J ¼ 7.7 Hz, H1), 4.48 (2H, s, CH
2
Cl), 4.03 (1H, dd,
J ¼ 12.1, 2.8 Hz, H6a), 3.86 (1H, dd, J ¼ 12.1, 4.8 Hz,
H6b), 3.63–3.46 (4H, m);
13

CNMR(CD
3
OD) d: 168.1,
96.8, 78.9, 77.8, 73.9, 70.9, 62.2, 41.6; MS (ESI) m/z 279
([M + Na]
+
, 100%).
RESULTS
Synthesis of inhibitors
The synthesis and characterization of compounds 1b, 1d, 1f,
1g and 3b (Scheme 1) is reported for the first time, and the
epoxypropyl derivative 3a is prepared by a new route. All

other compounds of Scheme 1 were synthesized following
described procedures, with modifications as specified in
Materials and Methods. All compounds were characterized
by
1
H-NMR and
13
C-NMR spectroscopy and by electro-
spray MS.
The epoxide 1d was prepared in seven steps from methyl
a-
D

-glucopyranoside (aMGlc, Scheme 2). Conventional
procedures were followed for the synthesis of the C-6
hydroxyl-free analogue 4 [28]: (a) selective protection of the
6-hydroxy group by reaction with trityl chloride, (b)
benzylation of the 2, 3, 4-OH groups, and (c) acid-catalysed
removal of the 6-O-trityl group. Oxidation of the free C-6
hydroxymethylene of 4 to aldehyde with pyridinium chlo-
rochromate, followed by Wittig methylenation at C-6,
epoxidation of the newly created double bond of 6 with
3-chloroperoxybenzoic acid, and removal of the protecting
benzyl groups present in 7 by catalytic transfer hydrogen-
ation led to the epoxide 1d. This compound was obtained as

a C-6 diastereomeric mixture which was used without
further separation.
The a,b-unsaturated methyl ester 1f and its free carboxy-
lic acid 1g were synthesized as depicted in Scheme 3, as
described previously [38] and the modifications which were
recently introduced for the preparation of C-6 aldehyde
derivatives of Glc [15]. The key step is the use of Collins
reagent for the selective oxidation of the primary trimethyl-
silyl ether of the fully silylated monosaccharide 8 to an
aldehyde (step ii). The resulting 2,3,4-tris-trimethylsilylated
derivative 9 was then condensed with methyl triphenyl-
phosphoranylidene acetate to the a,b-unsaturated methyl

ester 10. This reaction produced exclusively the E-isomer, as
judged from the value of the NMR coupling constant
between the protons connected to the double bond
(
3
J
H6-H7
¼ 15.8 Hz). Removal of the trimethylsilyl groups
afforded the methyl ester 1f. This compound was hydro-
lyzed under controlled alkaline conditions to the potassium
salt of the carboxylic acid 1g.
The epoxypropyl and chloroacetyl derivatives 3a and 3b

were prepared as shown in Scheme 4. The epoxypropyl
derivative was synthesized via two routes. (a) The allyl
2,3,4,6-tetra-O-acetyl-b-
D
-glucopyranoside 12 was first
obtained by silver oxide promoted nucleophilic substitution
Scheme 2.
4972 L. F. Garcı
´
a-Alles et al.(Eur. J. Biochem. 269) Ó FEBS 2002
at C-1 of acetobromoglucose (11) and then reacted with
dimethyldioxirane to afford the epoxide 13. This compound

was deacetylated in the presence of catalytic amounts of
methanolic sodium methoxide, giving the mixture of C-1
diastereomers 3a.(b)Tri-O-benzyl-
D
-glucal (16) was con-
verted into the reactive a-epoxide 17 by reaction with
dimethyldioxirane [36]. The oxirane ring was then opened
by treatment with (±)-glycidol to afford the pure b anomer
14. In a similar manner, the pure chloroacetyl b-
D
-gluco-
pyranoside 18 was obtained by reaction between 17 and

chloroacetic acid. Removal of the protecting benzyl groups
of 14 and 18 by catalytic transfer hydrogenation resulted in
compounds 3a and 3b, respectively.
Glucose analogues as pseudosubstrates of EII
Glc
and EII
Man
Compounds 1a)3d were assayed in vitro as substrates of the
two PTS transporters. Phosphoenolpyruvate-dependent
phosphotransferase activity was monitored by coupling
the formation of pyruvate (evolved from phosphoenolpyru-
vate) with its reduction to lactate catalysed by

L
-lactate
dehydrogenase. The C-1 epoxypropyl derivative 3a was the
only one out of the 13 glucose analogues that functioned as
a good substrate of EII
Glc
. The apparent K
m
of EII
Glc
for 3a
is 28 l

M
, which is comparable to the 60 l
M
value
determined in a parallel experiment for Glc (Table 1). V
max
of EII
Glc
for 3a was 14.9 l
M
Æmin
)1

which is only three times
slower than for Glc (40 l
M
Æmin
)1
). Compound 3a is highly
selective for EII
Glc
and is not phosphorylated by EII
Man
.
The C-1 chloroacetyl derivative 3b induced a slow con-

sumption of NADH in the presence of
L
-lactate dehydro-
genase, but also formation of NADPH in the presence of
Glc6P dehydrogenase. This background activity must
therefore be due to phosphorylation of Glc released by
slow hydrolysis of 3b, and not to phosphorylation of intact
3b. The C-1 isothiocyanate (3c) coexists in a 3 : 2 ratio with
the 1,2-cyclic thiocarbamate form [25], neither of which was
a substrate of EII
Glc
or EII

Man
. The bulky C-1 phenyl
isothiocyanate 3d and the epoxides 2a and 2b were not
substrates. This confirms the earlier observation that OH-2,
OH-3 and OH-4 are essential for recognition and that a
distortion of the pyranose ring by the epoxide ring is not
tolerated [15]. Compounds 1a–g are modified at C-6 and
therefore cannot be phosphorylated.
Glucose analogues as reversible inhibitors of EII
Glc
and EII
Man

Compounds 1a)3d were assayed in vitro as inhibitors of Glc
phosphorylation by the two PTS transporters. The concen-
tration of the glucose analogues was varied between 0 and
5m
M
while the substrate,
D
-Glc, was kept constant at
0.5 m
M
. To minimize the effect of potential time-dependent
irreversible inactivation, the assays were started by the

simultaneous addition of Glc and the inhibitor. Phosphoryl-
ation of Glc was measured with the Glc6P dehydrogenase-
coupled assay. Representative data for three compounds are
shown in Fig. 1, and IC
50
of all compounds are listed in
Table 2. Without exception, EII
Glc
was more strongly
inhibited than EII
Man
. The C-6 epoxide 1d was the strongest

inhibitor. It inhibited EII
Glc
with an IC
50
of 0.07 m
M
, but
had almost no effect on EII
Man
. This result was confirmed
Scheme 3.
Scheme 4.

Ó FEBS 2002 Inhibition of EII
Glc
(Eur. J. Biochem. 269) 4973
with [
14
C]aMGlc as substrate and direct detection of
[
14
C]aMGlc6P (results not shown). The second best C-6-
modified analogues, bromoacetyl-Glc (1a), bromoacetyl-
Man (1c), and isothiocyano-Glc (1e)hada10timeshigher
IC

50
than 1d. The epimeric bromoacetyl derivatives 1a and
1c both inhibited EII
Glc
, although EII
Glc
strongly discrimi-
nates between Glc and Man. The chemically less reactive
chloroacetyl-Glc (1b) did not inhibit EII
Glc
. This already
suggests that inhibition by 1a and 1c might be nonspecific

and due to rapid alkylation of Cys421 (see below). Of the
analogues modified at C-1, the epoxide 3a (a pseudosub-
strate) and the chloroacetyl 3b had an IC
50
of 1 m
M
.The
remaining analogues with bulky and rigid substituents had
IC
50
>2.4m
M

or did not inhibit at all.
Inhibition of Glc phosphorylation by the C-1 and C-6
epoxides 3a and 1d, the two most potent analogues, was
examined in more detail. EII
Glc
-dependent Glc phosphoryl-
ation was measured at four different concentrations of 3a
and 1d, and the results were plotted in the Eadie–Hofstee
form (Fig. 2). In the absence of an inhibitor, EII
Glc
displayed biphasic kinetics (Fig. 2, solid symbols), consis-
tent with the presence of two binding sites of different

affinity [15]. Addition of the C-6 epoxide 1d (Fig. 2A) did
not change the biphasic shape. In contrast, addition of the
phosphorylatable C-1 epoxide 3a (Fig. 2B) resulted in a
transition from the biphasic to a monophasic shape of the
curve. The datapoints in Fig. 2 were fitted to the two-active-
site model that was recently introduced to explain kinetic
data collected with several glucose analogues as pseudo-
substrates of EII
Glc
and EII
Man
[15]. A high-affinity

low-turnover site (represented by E1) and a low-affinity
high-turnover site (E2) were proposed to coexist at the
cytoplasmic side of EII
Glc
. According to this model, the
estimated ratio of inhibition constant/substrate dissociation
constant for the pair 1d and Glc (I/S) was K
I2
/K
S2
¼ 0.04 at
the low-affinity site, and K

I1
/K
S1
¼ 5 at the high-affinity site.
The curved shape of the plot and the K
I
/K
S
ratios indicate
that the C-6 epoxide 1d, like the C-6 aldehydes of Glc and
aMGlc [15], preferentially inhibits the low-affinity site of
EII

Glc
. For the phosphorylatable C-1 epoxide 3a (Fig. 2B),
Fig. 1. Inhibition of nonvectorial phosphorylation. Relative rate of Glc
phosphorylation by membranes containing EII
Glc
(solid symbols) and
EII
Man
(open symbols) in the presence of inhibitors 1d (squares), 1e
(triangles) and 3a (circles). The IC
50
values obtained from these and

similar plots are listed in Table 2. [Glc] ¼ 0.5 m
M
. Glc phosphoryla-
tion was detected with the
D
-Glc6P dehydrogenase assay.
Table 1. Kinetic constants of EII
Glc
and EII
Man
for analogues of
D

-glucose. Phosphorylation was measured at 30 °Cwiththe
L
-lactate dehydro-
genase-coupled assay. Kinetic constants were derived from a best fit to a Michaelis–Menten hyperbola. NS, No saturation observed.
Substrate
EII
Glc
EII
Man
V
max
a

(l
M
Æmin
)1
)
K
m
(l
M
)
V
max

/K
m
a
( · 10
3
min
)1
)
V
max
a
(l

M
Æmin
)1
)
K
m
(l
M
)
V
max
/K

m
a
( · 10
3
min
)1
)
Glc 40 ± 2 60 ± 10 700 ± 130 32 ± 2 50 ± 10 680 ± 200
3a 14.9 ± 0.6 28 ± 5 530 ± 110 NS NS 7 ± 2
3b
b
35 ± 4 700 ± 100 50 ± 20 40 ± 10 1100 ± 400 37 ± 20

a
Using 0.0013 lLÆlL
)1
membrane extract. Concentrations of other PTS components are indicated in Materials and methods.
b
Using
freshly purified compound. The reaction is also detected with the
D
-Glc6P dehydrogenase assay.
Table 2. Compounds 1a)3d as inhibitors of
D
-glucose phosphorylation.

Phosphorylation of
D
-Glc (0.5 m
M
) was measured using the
D
-Glc6P
dehydrogenase-coupled assay at 30 °C in the presence of 0–5 m
M
concentrations of the inhibitors. ND, No significant inhibition detec-
ted. Half inhibitory concentrations (IC
50

) are given in m
M
.Valuesin
parentheses determined measuring inhibition of phosphorylation of
[
14
C]a-MGlc (0.5 m
M
).
Inhibitor
IC
50

EII
Glc
EII
Man
1a 0.8 (2) ND
1b >5 ND
1c 0.8 (3) ND
1d 0.07
a
>5
1e 0.9 (2) ND
1f  5

a
ND
1g >5
a
ND
2a ND ND
2b ND ND
3a 1.0
a
ND
3b 1.3
a

ND
3c >5 ND
3d 2.4 ND
a
Confirmed with the radioactivity-based assay.
4974 L. F. Garcı
´
a-Alles et al.(Eur. J. Biochem. 269) Ó FEBS 2002
the corresponding ratios were K
I2
/K
S2

¼ 5andK
I1
/
K
S1
¼ 0.7. These values and the transition of the Eadie–
Hofstee plot from the biphasic to a monophasic shape are
consistent with inhibition by a compound that preferentially
binds to the high-affinity site. The different affinities of the
analogues modified at C-1 and C-6 for the two sites also
explain the observed difference of their IC
50

: low for C-6,
high for the C-1 epoxides. They were determined at 0.5 m
M
Glc, at which concentration the high-turnover site (low
affinity) is saturated and therefore preponderant in catalysis.
Glucose analogues as irreversible inhibitors of EII
Glc
To assay for irreversible inhibition, membrane fractions
containing EII
Glc
were preincubated with the different
compounds 1a)3d at 30 °C. Preliminary experiments

showed that the extent of inactivation depended on the
concentration of dithiothreitol present during the incuba-
tion. For instance, inactivation of EII
Glc
by iodoacetamide
was 50% in the presence of 0.5 m
M
, and almost complete in
the presence of 4 m
M
dithiothreitol (results not shown). For
this reason EII

Glc
-containing membranes were always
preincubated in the presence of 4 m
M
dithiothreitol. Three
conditions were assayed: (a) treatment with the inhibitor
alone; (b) in the presence of a 10 m
M
concentration of a
protective substrate, glucose (+ Glc, Table 3); (c) in the
presence of phosphoenolpyruvate and the soluble PTS
proteins necessary to keep the reactive Cys421 of EII

Glc
in
the phosphorylated state (+ PEP). Aliquots were with-
drawn after different time intervals and assayed for glucose
phosphotransferase activity. Controls without inhibitor
were run in parallel to correct for thermal inactivation,
which in all cases was less than 10% of the activity at time
zero. The corrected data were then fitted to decay curves
from which the inactivation rates (k
inact
) under conditions
(a) to (c) were calculated.

Fig. 2. Reversible inhibition of EII
Glc
. Eadie–Hofstee plots of non-
vectorial phosphorylation of
D
-Glc (0–2 m
M
)bymembranefractions
containing EII
Glc
. Phosphorylation was assayed in the presence of
inhibitors 1d [A, 0 l

M
(squares), 33.3 l
M
(triangles), 100 l
M
(circles)
and 300 l
M
(stars)] and 3a [B, 0 m
M
(squares), 0.33 m
M

(triangles),
1m
M
(circles) and 3 m
M
(stars)]. The lines represent the best global
least-squares fit of the data to a kinetic model of EII with two inde-
pendent enzymatic activities, E1 (high affinity) and E2 (low affinity)
[15]. Binding of the inhibitor to both E1 and E2 was allowed. The
kinetic constants obtained from the best fit are: with 1d (A)
K
S1

¼ 4 l
M
, k
1
¼ 23 min
)1
, K
I1
¼ 19 l
M
, K
S2

¼ 190 l
M
,
k
2
¼ 39 min
)1
, K
I2
¼ 8 l
M
;with3a (B) K

S1
¼ 4 l
M
, k
1
¼ 16 min
)1
,
K
I1
¼ 3 l
M

, K
S2
¼ 140 l
M
, k
2
¼ 32 min
)1
, K
I2
¼ 700 l
M

. K
S1
and
K
S2
are the dissociation constants of E1 and E2 for Glc, K
I1
and K
I2
the
dissociation constants for the inhibitor, and k
1

and k
2
are the turn-
over numbers.
DYNAFIT
was used to fit the experimental data to the
theoretical model and in the subsequent simulations [22].
Table 3. Rates of inactivation of IICB
Glc
. Incubation of purified IICB
Glc
with the indicated concentration (m

M
) of the analogues 1a)3d was carried
outat30°C. Rate constants (min
)1
) were calculated by nonlinear fit to a first-order decay function of the form: y ¼ A exp
(–kinact t)
+residual.
Inhibitor Concn.
k
inact
– + Glc
a

+PEP
b
a-Haloester analogues
1a
c
1 1.51 ± 0.09 1.5 ± 0.3 < 0.001
1b 30 0.12 ± 0.02 0.15 ± 0.02 < 0.001
1c
c
1 0.9 ± 0.1 – < 0.001
3b 4 0.023 ± 0.006 – –
IAcNH

2
1 0.61 ± 0.06 0.54 ± 0.09 < 0.001
BrAcOH 1 0.101 ± 0.004 – < 0.001
Epoxides
1d 60 0.007 ± 0.001 – –
2a 60 < 0.001 – –
2b 60 < 0.001 – –
3a 60 0.0038 ± 0.0004 – –
Isothiocyanates
1e 15 0.35 ± 0.04 0.7 ± 0.1 < 0.001
3c 15 0.54 ± 0.04 0.58 ± 0.06 0.4 ± 0.1
3d 5 0.55 ± 0.09 0.67 ± 0.06 < 0.001

a,b-Unsaturated carboxylic acid derivatives
1f 60 < 0.001 – –
1g 60 < 0.001 – –
a
Incubation in the simultaneous presence of 10 m
M
Glc.
b
Incubation in the presence of 1.5 m
M
phosphoenolpyruvate, 0.5 l
M

E1, 0.5 l
M
HPr, 1 l
M
IIA
Glc
.
c
Rate constants calculated by nonlinear fit to a second-order decay function: y ¼ A/(1 + k
inact
t) + residual.
Ó FEBS 2002 Inhibition of EII

Glc
(Eur. J. Biochem. 269) 4975
Representative examples of the inactivation curves
obtained with 1a, iodoacetamide and bromoacetic acid are
given in Fig. 3, and the results obtained with all compounds
are listed in Table 3. Rates of inactivation were fastest with
the C-6 bromoacetyl analogues 1a and 1c andwiththe
isothiocyanates, more than 10 times slower with the C-6 and
C-l chloroacetyl compounds 1b and 3b, and at least 100
times slower for the epoxides. The presence of 10 m
M
Glc

did not protect against inactivation. To the contrary, the
presence of Glc slightly sensitized EII
Glc
for inactivation by
the isothiocyanates 1e and 3d. On the other hand, the rate of
inactivation by bromoacetyl-Glc (1a) was 15 times faster
than by bromoacetic acid and 2.5 times faster than by the
chemically more reactive iodoacetamide, suggesting some
specificity and selectivity of the glucose analogues for EII
Glc
.
Phosphorylation of EII

Glc
completely protected against
inactivation, indicating that Cys421 is the most, if not the
only, reactive residue. Protection was incomplete in the
presence of the C-1 SCN-Glc (3c) which by its free OH-6, at
the high EII
Glc
concentrations present during the incuba-
tion, can accept a phosphoryl group and thereby deprotect
Cys421.
Inhibition and inactivation of sugar uptake by starved
cells

Analogues 1a)3d were assayed as competitive inhibitors of
[
14
C]sugar uptake by intact cells. The nonmetabolizable
[
14
C]aMG and [
14
C]2dGlc were used as substrates, instead
of [
14
C]glucose. These glucose analogues are selectively

transported via EII
Glc
and EII
Man
, respectively, and conse-
quently further guarantee that uptake is due to the studied
transporter. The reactive analogues and [
14
C]aMG or
[
14
C]2dGlc were added in molar ratio of 10 : 1 and the

uptake of [
14
C]aMGlc via EII
Glc
or of [
14
C]2dGlc via EII
Man
was measured (Fig. 4). The C-1 epoxide 3a was the only
analogue that efficiently blocked EII
Glc
-dependent uptake.

It also slightly reduced the rate of EII
Man
-mediated uptake
of 2dGlc. The C-6 epoxide 1d weakly inhibited uptake by
EII
Glc
only, whereas the other analogues were inactive with
both transporters.
To test for inactivation of transport, starved cells were
preincubated with the C-6 bromoacetyl-Glc 1a and bromo-
acetyl-Man 1c, the C-6 epoxide 1d, the C-1 epoxide 3a,the
C-6 isothiocyanate 1e and the C-1 isothiocyanate 3d,and

then the residual uptake activity was determined (see
Table 4). The C-6 bromoacetyl compounds 1a and 1c
completely blocked uptake by EII
Glc
and EII
Man
.Preincu-
bation with 20 m
M
C-1 phenylisothiocyanate 3d reduced the
uptake rate fivefold and 20-fold, respectively. The other
analogues were less inhibitory.

Inactivation by the bromoacetyl-Glc (1a) and bromoace-
tyl-Man (1c) was examined in more detail. Taking into
account that Glc appeared to sensitize rather than protect
EII
Glc
for inactivation in vitro (see above), cells were
preincubated for 2 and 5 min with and without inhibitor
in the absence and presence of 10 m
M
Glc. With short
incubation times, cells expressing EII
Glc

were inactivated
slightly faster by the glucose analogue 1a than by the
Fig. 3. Irreversible inhibition of EII
Glc
. A membrane preparation
containing EII
Glc
was preincubated with 1 m
M
inhibitor 1a (circles),
iodoacetamide (stars) and bromoacetic acid (tickmarks) in the presence
of 4 m

M
dithiothreitol at 30 °C. Aliquots were withdrawn after the
indicated incubation time, 30-fold diluted into cold buffer, and residual
PTS activity was then measured in a standard phosphotransferase
assay. In the lowest part of the figure are presented the residuals of the
fit to exponential (upper panels), second-order (central panels) and
biphasic (lower panels) decay curves for inactivation by 1a (circles)and
iodoacetamide (stars).
Fig. 4. Inhibition of sugar uptake by starved cells. EII
Glc
-dependent
uptake of [

14
C]aMGlc (0.1 m
M
,blackbars),andEII
Man
-dependent
uptake of [
14
C]2dGlc (0.1 m
M
, grey bars) in the presence of the
indicated inhibitors (1 m

M
). DPTS, Background uptake by a strain
lacking both EII
Glc
and EII
Man
. 100% uptake corresponds to
25 nmolÆmin
)1
Æmg
)1
dry weight of cells expressing EII

Glc
,and
90 nmolÆmin
)1
Æmg
)1
cells expressing EII
Man
.
4976 L. F. Garcı
´
a-Alles et al.(Eur. J. Biochem. 269) Ó FEBS 2002

mannose epimer 1c (Fig. 5A) whereas the opposite was true
for EII
Man
(Fig. 5B). This indicates that EII
Glc
and EII
Man
are, to some extent, selectively inactivated by their cognate
substrate analogues. As observed above, the presence of Glc
did not protect, but to the contrary sensitized, EII
Glc
for

inactivation (Fig. 5A, grey bars). In the presence of Glc, the
rates of EII
Glc
inactivation by 1a and 1c increased 18-fold
and 27-fold, respectively. This effect of Glc is specific for
EII
Glc
and the C-6 bromoacetyl sugars. It was not observed
with the C-1 isothiocyanate 3d and the C-6 epoxide 1d
(results not shown), nor with EII
Man
(Fig. 5B).

Antibacterial activity
Initially, PTS-specific toxic sugars can be considered as
potential antibiotics. For that reason, and in view of the
results presented above, the analogues 1a)3d were screened
as antibacterial agents towards E. coli cells expressing either
EII
Glc
or EII
Man
. Cell growth in mineral medium supple-
mented with glucose was monitored spectrophotometrically
(550 nm), in the presence of variable concentrations of the

glucose analogues. The PTS specificity of these compounds
was assessed in two ways: (a) also using the background
E. coli strain lacking both transporters, and (b) studying
growth with glycerol as carbon source, instead of glucose.
Thus, cell growth was prevented or delayed by 1a and 1c
(> 0.04 m
M
concentration required), 1d (> 4 m
M
), 1e
(> 0.2 m
M

), 3c and 3d (> 0.8 m
M
). However, none of the
analogues showed PTS-mediated antibacterial activity. All
kinds of cells, expressing the PTS transporters or not, were
inhibited to the same extent (not shown). Moreover, the
results were independent of whether glucose or glycerol were
added as carbon source.
DISCUSSION
Thirteen glucose analogues with a-haloester, isothiocyanate,
epoxide and a,b-unsaturated ester functions at positions C-1
and C-6 were synthesized and characterized as pseudosub-

strates, reversible and irreversible inhibitors of EII
Glc
and
EII
Man
. The C-1 epoxide analogue 3a was the only efficient
pseudosubstrate of EII
Glc
in vitro, and the only reversible
inhibitor of sugar uptake by starved cells. The C-6
isothiocyanate 1e and epoxide 1d and the C-1 epoxide 3a
and chloroacetate 3b were selective reversible inhibitors of

nonvectorial phosphorylation by EII
Glc
.TheC-6bromo-
acetylglucose and bromoacetylmannose derivatives 1a,c
irreversibly blocked in vitro phosphorylation and uptake
by starved cells. The isothiocyanates only blocked in vitro
phosphorylation by EII
Glc
in membrane preparations, but
not uptake. The C-6 bromoacetyl derivatives and isothio-
cyanates presumably reacted with the active-site residue
Cys421. This cysteine transfers the phosphoryl group from

the IIA
Glc
subunit to the OH-6 of the substrate in a double-
displacement reaction [39]. It is highly exposed at the edge of
a split a/b sheet [40], and from this position rapidly
quenches the reactive analogues. It was expected that this
residue would react, but not that it would be the only
reactive one. It is noteworthy that (a) the rate of inactivation
by bromoacetyl-Glc is 2.5 times faster than by the chemi-
cally more reactive but unspecific iodoacetamide, (b) EII
Glc
is completely protected against inactivation if Cys421 is

phosphorylated or converted into a disulfide before expo-
sure to the alkylating analogues (results not shown), and (c)
inactivation of EII
Glc
is accelerated in the presence of Glc
(see below). Although the dominant reactivity of Cys421
compromised the labelling of other active-site residues, the
glucose analogues nevertheless provided new, and con-
firmed recent, insight into (a) the kinetic properties [15], (b)
the selectivity, and (c) the conformational coupling of the
EII
Glc

active sites.
Fig. 5. Glucose-sensitized inactivation of [
14
C]sugar uptake by starved
cells. Cells expressing EII
Glc
(A) or EII
Man
(B) were incubated for
2 min with and without inhibitors (10 m
M
) in the absence (black bars)

and presence (grey bars) of 10 m
MD
-Glc. Cells were washed to remove
excess inhibitor and Glc and assayed for uptake activity as described in
Materials and methods and in the legend to Fig. 4.
Table 4. Inactivation of sugar uptake by starved cells by compounds
1a-3d. Cells were treated with the indicated concentrations (m
M
)of
inhibitor for 60 min at room temperature. The rate of accumulation of
radioactive sugar in the pretreated cells was then measured. Uptake
rates are in nmolÆmin

)1
Æmg
)1
dry weight of cells. EII
Glc
was measured
using [
14
C]aMGlc (0.1 m
M
,6400d.p.m.Ænmol
)1

). EII
Man
was meas-
ured using [
14
C]2dGlc (0.1 m
M
, 5600 d.p.m.Ænmol
)1
).
Inhibitor Concn
Uptake rate

EII
Glc
EII
Man
– – 23 ± 10 41 ± 17
1a 1 < 0.02 < 0.02
1c 1 < 0.02 < 0.02
1d 510±4–
20 4 ± 3 19 ± 3
1e 5 16±2 14±1
20 6 ± 4 8 ± 2
3a 517±9–

20 8 ± 5 17 ± 4
3d 5 20 ± 6 6.4 ± 0.8
20 5 ± 1 2 ± 1
Ó FEBS 2002 Inhibition of EII
Glc
(Eur. J. Biochem. 269) 4977
(a) EII
Glc
,EII
Man
and the mannitol transporter, EII
Mtl

display biphasic phosphorylation kinetics towards their
natural substrates, indicating that activity in vitro is the
sum of contributions from two independent sites (Fig. 6),
one of high affinity and low turnover, the second one of low
affinity and high turnover [15,41]. There exist, however,
pseudosubstrates, for which EII displays Michaelis–Menten-
like kinetics. 3-Deoxy-3-fluoro-
D
-glucose (3FGlc), for
instance, preferentially if not exclusively binds to the low-
affinity site of EII
Glc

, as deduced from the same K
i
measured
for a C-6 aldehyde analogue of Glc as inhibitor of Glc and
3FGlc phosphorylation by EII
Glc
[15]. The newly synthesized
C-1 epoxide 3a is the first and so far only analogue that
preferentially binds to the Glc high-affinity site of EII
Glc
.The
K

m
and V
max
of EII
Glc
for 3a are lower than for Glc, and 3a is
a strong competitive inhibitor of uptake.
(b) The high-affinity periplasmic site and the low-affinity
cytoplasmic site of the transporter recognize different
features of the substrate. Of six analogues that reversibly
inhibited EII
Glc

atthecytoplasmicsite(inmembrane
preparations), only one, 3a, also inhibited uptake by intact
cells (Fig. 6). A comparison of structure and reactivity
between the six analogues suggests that inhibitors of uptake
that bind to the periplasmic site of the protein must have a
free OH-6, whereas inhibitors of phosphorylation that bind
to the cytoplasmic site may or may not have one. Thus, the
C-1 epoxide 3a with a free OH-6 was a potent inhibitor of
uptake, whereas the most potent inhibitor of nonvectorial
phosphorylation, the C-6 epoxide 1d, was a comparatively
weak inhibitor of uptake. Like 1d, two glucose-6-aldehyde
analogues have recently been shown to display a similar

preference for the low-affinity site [15].
(c) Substrate protection is commonly used to confirm the
specificity of an active-site labelling reaction. Addition of
Glc, however, did not protect but sensitized EII
Glc
for
inactivation (Table 3). This could indicate that binding of
Glc to one site increases the reactivity of a second site. Also
pointing in this direction is the second-order or biphasic
shape of the inactivation curve of EII
Glc
by 1a (see residuals

in Fig. 3). This may indicate that binding of a first molecule
of 1a to the EII
Glc
dimer does not inactivate, but increases
the reactivity of, Cys421 towards a second molecule.
Alternatively, biphasic inactivation by 1a (with two inacti-
vation rates differing by a factor of 10) may originate from
the different accessibility of the two cysteines of the EII
Glc
dimer for the glucose analogues. For comparison, inactiva-
tion by iodoacetamide fits better to an exponential function
(Fig. 3) as expected of a small nonspecific reagent with

equal access to both Cys. Whatever the cause, sensitization
by Glc cannot be the (trivial) effect of Glc-induced
dephosphorylation/deprotection of Cys421, because EII
Glc
in membrane preparations is already dephosphorylated [6],
as indicated by the complete inactivation induced by
iodoacetamide.
The bromoacetyl derivatives 1a and 1c also inactivated
Glc uptake by starved cells. Being modified at OH-6, 1a and
1c were neither substrates nor competitive inhibitors of
uptake (see Fig. 4). That they nevertheless inactivated EII
Glc

suggests that the reactive Cys421 must be directly accessible
from the periplasmic side of the membrane and that
accessibility is increased in the presence of Glc. Because this
same effect was not observed with EII
Man
-expressing cells,
nonspecific effects on essential PTS components other that
EII
Glc
can be excluded. What cannot be excluded is that
dephosphorylation and/or catalytic turnover of EII
Glc

,
rather than binding of Glc, enhanced the reactivity of
Cys421. As Cys421 is the only invariant cysteine in
homologous transporters and also the only essential cysteine
for IICB
Glc
activity [39], it must be the reactive one and
accessible from the periplasm. Our results confirm experi-
ments of Robillard et al. [42], who demonstrated that
EII
Glc
-dependent uptake can be inactivated by membrane-

impermeable thiol reagents, and, on the basis of this,
concluded that a reactive thiol group must be accessible
from the periplasmic side.
In conclusion, chemically reactive glucose analogues
turned out to be instrumental in the characterization of
EII
Glc
as a dimeric transport protein with two mutually
interacting binding sites containing an active-site cysteine
that is accessible from both faces of the membrane. The
nature of the structural rearrangement for this alternating
accessibility is now being examined with heterodimers

between variants with mutations in the different domains.
ACKNOWLEDGEMENTS
This study was supported by grant 3100-063420 from the Swiss
National Science Foundation.
REFERENCES
1. Bouma, C.L., Meadow, N.D., Stover, E.W. & Roseman, S. (1987)
II-BGlc, a glucose receptor of the bacterial phosphotransferase
system: molecular cloning of ptsG and purification of the receptor
from an overproducing strain of Escherichia coli. Proc. Natl. Acad.
Sci. USA 84, 930–934.
2.Erni,B.,Zanolari,B.&Kocher,H.P.(1987)Themannose
permease of Escherichia coli consists of three different proteins.

Amino acid sequence and function in sugar transport, sugar
phosphorylation, and penetration of phage lambda DNA. J. Biol.
Chem. 262, 5238–5247.
Fig. 6. Proposed model for the IICB
Glc
dimer of EII
Glc
. IICB
Glc
con-
sists of a membrane-spanning C domain (grey) and the cytoplasmic
IIB domain (black). IICB

Glc
is phosphorylated at Cys421 by the sol-
uble IIA
Glc
subunit. It is proposed that two nonvectorial phosphory-
lation sites are present at the cytoplasmic side of the transporter [15].
The affinities of these two sites for glucose are very different. Pseudo-
substrates such as 3FGlc, or inhibitors such as the epoxide 1d would
interact preferentially with the glucose low-affinity site. To the con-
trary, the C-1 epoxypropyl analogue 3a might react in the high-affinity
site. The inactivation data presented here, and in a previous study [42],
indicate that Cys421 is accessible to reactive, membrane-impermeable

reagents, such as analogues 1a,b, from the periplasmic side.
4978 L. F. Garcı
´
a-Alles et al.(Eur. J. Biochem. 269) Ó FEBS 2002
3. Williams, N., Fox, D.K., Shea, C. & Roseman, S. (1986) Pel, the
protein that permits lambda DNA penetration of Escherichia coli,
is encoded by a gene in ptsM and is required for mannose utili-
zation by the phosphotransferase system. Proc. Natl. Acad. Sci.
USA 83, 8934–8938.
4. Postma, P.W., Lengeler, J.W. & Jacobson, G.R. (1993) Phos-
phoenolpyruvate:carbohydrate phosphotransferase systems of
bacteria. Microbiol. Rev. 57, 543–594.

5. Boer, H., ten Hoeve-Duurkens, R.H., Schuurman-Wolters, G.K.,
Dijkstra, A. & Robillard, G.T. (1994) Expression, purification,
and kinetic characterization of the mannitol transport domain of
the phosphoenolpyruvate-dependent mannitol phosphotransfer-
ase system of Escherichia coli. Kinetic evidence that the E. coli
mannitol transport protein is a functional dimer. J. Biol. Chem.
269, 17863–17871.
6. Erni, B. (1986) Glucose-specific permease of the bacterial phos-
photransferase system: phosphorylation and oligomeric structure
of the glucose-specific IIGlc-IIIGlc complex of Salmonella
typhimurium. Biochemistry 25, 305–312.
7. Lanz, R. & Erni, B. (1998) The glucose transporter of the

Escherichia coli phosphotransferase system. Mutant analysis of the
invariant arginines, histidines, and domain linker. J. Biol. Chem.
273, 12239–12243.
8. van Montfort, B.A., Schuurman-Wolters, G.K., Duurkens, R.H.,
Mensen, R., Poolman, B. & Robillard, G.T. (2001) Cysteine cross-
linking defines part of the dimer and B/C domain interface of the
Escherichia coli mannitol permease. J. Biol. Chem. 276, 12756–
12763.
9. Koning, R.I., Keegstra, W., Oostergetel, G.T., Schuurman-
Wolters, G., Robillard, G.T. & Brisson, A. (1999) The 5 A
projection structure of the transmembrane domain of the
mannitol transporter enzyme II. J. Mol. Biol. 287, 845–851.

10. Lolkema, J.S., Dijkstra, D.S.R.H., T.H D. & Robillard, G.T.
(1990) The membrane-bound domain of the phosphotransferase
enzyme IImtl of Escherichia coli constitutes a mannitol trans-
locating unit. Biochemistry 29, 10659–10663.
11. Hummel, U., Nuoffer, C., Zanolari, B. & Erni, B. (1992) A
functional protein hybrid between the glucose transporter and the
N-acetylglucosamine transporter of Escherichia coli. Protein Sci. 1,
356–362.
12. Robillard, G.T. & Broos, J. (1999) Structure/function studies on
the bacterial carbohydrate transporters, enzymes II, of the phos-
phoenolpyruvate-dependent phosphotransferase system. Biochim.
Biophys. Acta 1422, 73–104.

13. Buhr, A. & Erni, B. (1993) Membrane topology of the
glucose transporter of Escherichia coli. J. Biol. Chem. 268,
11599–11603.
14. Huber, F. & Erni, B. (1996) Membrane topology of the mannose
transporter of Escherichia coli K12. Eur. J. Biochem. 239, 810–817.
15. Garcia-Alles, L.F., Zahn, A. & Erni, B. (2002) Sugar recognition
by the glucose and mannose permeases of Escherichia coli.Steady-
state kinetics and inhibition studies. Biochemistry 41, 10077–
10086.
16. Thomas, E.W., McKelvy, J.F. & Sharon, N. (1969) Specific and
irreversible inhibition of lysozyme by 2¢,3¢-epoxypropyl beta-
glycosides of N-acetyl-

D
-glucosamine oligomers. Nature (London)
222, 485–486.
17. Mullins, R.E. & Langdon, R.G. (1980) Maltosyl isothiocyanate:
an affinity label for the glucose transporter of the human
erythrocyte membrane. 2. Identification of the transporter.
Biochemistry 19, 1205–1212.
18. Connolly, B.A. & Trayer, I.P. (1979) Affinity labelling of rat-
muscle hexokinase type II by a glucose-derived alkylating agent.
Eur. J. Biochem. 93, 375–383.
19. Buhr, A., Fluekiger, K. & Erni, B. (1994) The glucose transporter
of Escherichia coli. Overexpression, purification, and character-

ization of functional domains. J. Biol. Chem. 269, 23437–23443.
20. Esquinas-Rychen, M. & Erni, B. (2001) Facilitation of bacter-
iophage lambda DNA injection by inner membrane proteins of
the bacterial phosphoenolpyruvate: carbohydrate phospho-
transferase system (PTS). J. Mol. Microbiol. Biotechnol. 3,
361–370.
21. Garcia-Alles, L.F., Flukiger, K., Hewel, J., Gutknecht, R., Siebold,
C.,Schurch,S.&Erni,B.(2002)Mechanism-basedinhibitionof
enzyme I of the Escherichia coli phosphotransferase system:
Cys502 is an essential residue. J. Biol. Chem. 277, 6934–6942.
22. Kuzmic, P. (1996) Program DynaFit for the analysis of enzyme
kinetic data: application to HIV proteinase. Anal. Biochem. 237,

260–273.
23. Ziegler, T., Kovac, P. & Glaudemans, C.P.J. (1989) Selective
bromoacetylation of alkyl hexopyranosides: a facile preparation of
intermediates for the synthesis of (1 fi 6)-linked oligosaccharides.
Carbohydr. Res. 194, 185–198.
24. Garcia Fernandez, J.M., OrtiZ. Mellet, C. & Fuentes, J. (1993)
Chiral 2-thioxotetrahydro-1,3-O, N-heterocycles from carb-
ohydrates. 2. Stereocontrolled synthesis of oxazolidine pseudo-
C-nucleosides and bicyclic oxazine-2-thiones. J. Org. Chem. 58,
5192–5199.
25. Maya, I., Lopez, O., Fernandez-Bolanos, J.G., Robina, I. &
Fuentes, J. (2001) A practical one-pot synthesis of O-unprotected

glycosyl thioureas. Tetrahedron Lett. 42, 5413–5416.
26. Rehnberg, N. & Magnusson, G. (1990) Chiral aldehydes by ring
contraction of pento- and hexopyranoside epoxides. J. Org. Chem.
55, 5467–5476.
27. Ogawa, S., Iwasawa, Y., Toyokuni, T. & Suami, T. (1985)
Synthesis of pseudooligosaccharidic glycosidase inhibitors. Part 1.
Synthesis of adiposin-1 and related compounds. Carbohydr. Res.
141, 29–40.
28. Chang, C.W., Chen, X.H. & Liu, H.W. (1998) CDP-6-deoxy-6,6-
difluoro-
D
-glucose: a mechanism-based inhibitor for CDP-D-

glucose 4,6-dehydratase. J. Am. Chem. Soc. 120, 9698–9699.
29. Hashimoto, H., Asano, K., Fujii, F. & Yoshimura, J. (1982)
Synthesis of destomic and epidestomic acid, and their C-6 epimers.
Carbohydr. Res. 104, 87–104.
30. Johns, B.A., Pan, Y.T., Elbein, A.D. & Johnson, C.R. (1997)
Synthesis and biological evaluation of aza-C-disaccharides:
(1 fi 6) (1 fi 4), and (1 fi 1)-linked sugar mimics. J. Am. Chem.
Soc. 119, 4856–4865.
31. Hung, S.C. & Wong, C.H. (1996) Synthesis of glycosyl chlorides
with acid-labile protecting groups. Tetrahedron Lett. 37, 4903–
4906.
32. EIAmin, B., Anantharamaiah, G.M., Royer, G.P. & Means, G.E.

(1979) Removal of benzyl-type protecting groups from peptides by
catalytic transfer hydrogenation with formic acid. J. Org. Chem.
44, 3442–3444.
33. Talley, E.A., Vale, M.D. & Yanovsky, E. (1945) Allyl ethers of
carbohydrates. III. Ethers of glucose and galactose. J. Am. Chem.
Soc. 67, 2037–2039.
34. Adam, W., Bialas, J. & Hadjiarapoglou, L. (1991) A convenient
preparation of acetone solutions of dimethyldioxirane. Chem. Ber.
124, 2377.
35. Rodriguez, E.B. & Stick, R.V. (1990) The synthesis of active-site
directed inhibitors of some b-glucan hydrolases. Aust. J. Chem. 43,
665–679.

36. Halcomb, R.L. & Danishefsky, S.J. (1989) On the direct epoxi-
dation of glycals: application of a reiterative strategy for the
synthesis of b-linked oligosaccharides. J. Am. Chem. Soc. 111,
6661–6666.
37. Bellucci,G.,Catelani,G.,Chiappe,C.,D’andrea,F.&Grigo,G.
(1997) An efficient stereoselective synthesis of enantiomerically
pure mono- and di-O-hexadecyl-b-
D
-glucosylglycerol ethers by
epoxidation of an allyl b-
D
-glucopyranoside asymmetrically

induced by the glucide moiety. Tetrahedron: Asymmetry 8,
765–773.
Ó FEBS 2002 Inhibition of EII
Glc
(Eur. J. Biochem. 269) 4979
38. Mahrwald, R., Theil, F., Schick, H., Schwarz, S., Palme, H.J. &
Weber, G. (1986) The oxidation of primary trimethylsilyl ethers to
aldehydes: a selective conversion of a primary hydroxy group into
an aldehyde group in the presence of a secondary hydroxy group.
J. Prakt. Chem. 328, 777–783.
39. Nuoffer,C.,Zanolari,B.&Erni,B.(1988)Glucosepermeaseof
Escherichia coli. The effect of cysteine to serine mutations on the

function, stability, and regulation of transport and phosphoryla-
tion. J. Biol. Chem. 263, 6647–6655.
40. Eberstadt, M., Grdadolnik, S.G., Gemmecker, G., Kessler, H.,
Buhr, A. & Erni, B. (1996) Solution structure of the IIB domain of
the glucose transporter of Escherichia coli. Biochemistry 35,
11286–11292.
41. Lolkema, J.S., ten Hoeve-Duurkens, R.H. & Robillard, G.T.
(1993) Steady state kinetics of mannitol phosphorylation catalyzed
by enzyme IImtl of the Escherichia coli phosphoenolpyruvate-
dependent phosphotransferase system. J. Biol. Chem. 268, 17844–
17849.
42. Robillard, G.T. & Beechey, R.B. (1986) Evidence for the existence

of a channel in the glucose-specific carrier EIIGlc of the Salmo-
nella typhimurium phosphoenolpyruvate-dependent phospho-
transferase system. Biochemistry 25, 1346–1354.
SUPPLEMENTARY MATERIAL
The following material is available from ck
well-science.com/products/journals/suppmat/EJB/
EJB3197/ EJB3197sm.htm
Table S1.
1
H,
13
C NMR and mass spectrometry (ESI) data

for the synthetic intermediates 5, 7, 10 and 18.
4980 L. F. Garcı
´
a-Alles et al.(Eur. J. Biochem. 269) Ó FEBS 2002