Connection of transport and sensing by UhpC, the sensor
for external glucose-6-phosphate in
Escherichia coli
Christian Schwo¨ ppe
1
, Herbert H. Winkler
2
and H. Ekkehard Neuhaus
1
1
Pflanzenphysiologie, Universita
¨
t Kaiserslautern, Kaiserslautern, Germany;
2
Department of Microbiology and Immunology,
College of Medicine, University of South Alabama, Mobile, AL, USA
UhpC is a membrane-bound sensor protein in Escherichia
coli required for recognizing external glucose-6-phosphate
(Glc6P) and induction of the transport protein UhpT.
Recently, it was shown that UhpC is also able to transport
Glc6P. In this study we investigated whether these transport
and sensing activities are obligatorily coupled in UhpC. We
expressed a His-UhpC protein in a UhpC-deficient E. coli
strain and verified that this construct does not alter the basic
biochemical properties of the Glc6P sensor system. The
effects of arginine replacements, mutations of the central
loop, and introduction of a salt bridge in UhpC on transport
and sensing were compared. The exchanges R46C, R266C
and R149C moderately affected transport by UhpC but
strongly decreased the sensing ability. This suggested that the
affinity for Glc6P as a transported substrate is uncoupled in

function as an inducer [1]. In addition to UhpT, the genomic
locus uhp encodes UhpB, UhpA and UhpC [2,3]. After
recognition of extracellular Glc6P by the constitutively
expressed sensor UhpC, this protein most likely interacts
with the membrane-bound UhpB and stimulates its kinase
activity. Finally, a phosphate group is transferred to UhpA,
a soluble transcription activator that governs the expression
of the uhpT gene [4].
The sensor membrane protein and the transport protein
are homologous molecules sharing about 32% identity [2]
and both are members of the Major Facilitator Superfamily
[5–7]. One postulates that the primordial unregulated gene
that encoded the transport protein was duplicated and then
modified to gain sensor function and lose transport
function.
Strikingly, in Chlamydia pneumoniae the system which
transports hexose phosphates [8] is structurally more similar
to UhpC than to UhpT. Besides, no genes for sensing or
regulation (uhp elements) have been identified in this
species [9]. For an obligate intracellular bacterium such as
Chlamydia there was probably no driving force for the
establishment of a sensor/regulatory system as Glc6P was
always present in the host cell cytosol ready to be
transported. Previous experiments by others led to the
conclusion that UhpC is unlikely to transport Glc6P [2] but
recent analysis demonstrated that UhpC from E. coli can
act not only as a sensor but also as a carrier that facilitates a
Glc6P/P
i
antiport mode of transport [8]. The transport

external Glc6P by UhpC. One might postulate that the
two activities are obligatorily linked because the transport
of a few molecules of substrate changes the conformation
of UhpC and that this change is required for UhpC to
initiate the transcription of the uhpT gene. Alternatively,
the binding of Glc6P could change the conformation of
UhpC with no requirement for translocation of substrate.
We compared the effects of site-specific mutations of the
UhpC protein on both the transport and sensing
functions of this molecule. We mutated arginine residues
as these are known to be involved in binding of anions
to proteins [18] and as some of these are conserved and
essential for function in proteins homologous to UhpC
[19]. In addition, we introduced an intrahelical salt bridge
into UhpC, a bridge identified as necessary for UhpT
function [20], but that is absent in UhpC [21]. Finally, we
changed three residues that are conserved in proteins
similar to UhpC [19] and that are located in the central
hydrophilic loop between transmembrane domains 6 and
7, a loop that was shown to be essential for UhpT
activity but was thought to be less important for sensing
[22]. The very low level of transport by UhpC precluded
doing these experiments with just the chromosomal copy
of uhpC, thus UhpC had to be over-expressed from a
plasmid-borne gene. This changed the ratios of the uhp
operon products, so extrapolation to the normal E. coli
situation with all uhp genes in an operon may not be
valid and such extrapolation was not our goal. However,
we were able to clearly separate the sensing and
transport activities of UhpC membrane protein.

modified pET16b constructs was carried out according to
standard protocols.
Determination of transport activities of the over-
expressed UhpC mutants was carried out using the
UhpT-deficient E. coli strain BL21(DE3) (uhpT::Tn1000).
Overnight cultures were diluted 100-fold into YT medium
plus antibiotics and grown at 37 °C to a turbidity (D
578
)
of 0.5. After induction of T7-RNA polymerase activity
by the addition of isopropyl thio-b-
D
-galactoside (IPTG)
(final concentration 0.012%), cells were grown for a
further 90 min, collected by centrifugation, resuspended
in Mops buffer solution (50 m
M
,pH7.5)andstoredon
ice until use.
Sensing activities of the UhpC mutants were determined
by using the UhpC-deficient E. coli strain BL21(DE3)
(uhpC::Tn1000). When the turbidity of the growing culture
reached 0.5, Glc6P was added and the cells were grown for
additional 15 min. The maximal Glc6P concentration
added during the induction period was 400 l
M
because at
higher concentrations catabolite repression occurs. The
induction of uhpT was analysed by uptake of [
14

14
C]Glc6P transport was stopped by
transfer of the cells to membrane filters (25 mm diameter,
0.45 lm pore size; Pall Life Science, Dreieich, Germany)
prewetted with Mops buffer solution and under vacuum.
After washing with ice-cold buffer solution the filters
were placed in vials containing scintillation cocktail (Quick-
safe A, Zinsser Analytic, Frankfurt/Main, Germany).
The radioactivity was quantified in a Canberra-Packard
Tricarb-2500 counter. The kinetic constants of transport
were estimated using the method of Hanes. All data
represent means of at least three independent experi-
ments. The standard deviation was always less than 9%
of the given mean. The background activity of IPTG-
induced E. coli cells harbouring the empty vector plasmid
pET16b has always been subtracted [8]. Protein content
of E. coli samples was quantified using Coomassie
brilliant blue [24].
Ó FEBS 2003 Bacterial hexose-phosphate transport protein (Eur. J. Biochem. 270) 1451
Cytoplasmic membrane preparation and Western blot
analysis
Site-directed mutations of a membrane protein can influ-
ence the efficiency of integration into the native membrane
and thus influence the apparent transport activity. The
efficiency of protein incorporation into the E. coli
cell membrane was quantified by Western blot analysis
[8,25]. For Western blot analysis E. coli BL21(DE3)
(uhpT::Tn1000) (harbouring the corresponding pET16b
construct) described above for transport assays was used.
Cytoplasmic membrane preparations were carried out

compared the K
(induction)
observed with chromosomally
encoded UhpC and with plasmid-encoded UhpC with a
histidine tag. In both systems increasing concentrations of
external Glc6P induced the Glc6P uptake system (UhpT)
with a K
(induction)
of 3.8 l
M
(Fig. 1A,B) which is close to the
concentration dependence of induction observed by others
[27]. Thus, the over-expression of the wild-type UhpC with a
histidine tag did not alter the concentration of Glc6P
required for half-maximal induction of UhpT.
Site-directed mutations of conserved arginine residues
Arginine residues in proteins are excellent candidates for the
binding of negatively charged substrates like Glc6P [18].
Maloney and coworkers showed that two of 14 arginine
residues in UhpT are critical for its function [19]. To identify
conserved arginine residues in UhpC we aligned several
UhpC- and UhpT-like proteins including the Glc6P trans-
porter from C. pneumoniae (HPTcp [8]); that exhibits a
higher degree of structural identity to the E. coli UhpC
protein than to UhpT [9]. UhpC and UhpT proteins have
been taken from the genomes of E. coli, Salmonella enterica,
Pasteurella multocida, Yersina pestis and Vibrio cholerae. It
should be emphasized that the function of UhpC in Yersinia
is doubtful because Y. pestis contains Uhp A, B, and C but
lacks UhpT (RefSeq: NC003143; GenBank: NC003143). In

C]Glc6P.Insets:The
Hanes analysis (only hyperbolic parts) revealed in both cases an
apparent K
(induction)
of 3.8 l
M
and a V
max(induction)
of 160 nmolÆmg
protein
)1
Æh
)1
.
1452 C. Schwo
¨
ppe et al. (Eur. J. Biochem. 270) Ó FEBS 2003
Fig. 2. Multiple alignment of UhpC- and UhpT-related protein amino acid sequences. The UhpC proteins share 86.9% (S. enterica) to 56.7%
(V. cholerae), the HPT protein from C. pneumoniae shares 45.3%, and the UhpT proteins share 32.6% (E. coli) to 30.0% (V. cholerae)identityto
the E. coli UhpC protein (for details see text). The multiple alignment was performed using
CLUSTALW
(default settings). The asterisks indicate the
positions of the mutated amino acids of the E. coli UhpC protein.
Ó FEBS 2003 Bacterial hexose-phosphate transport protein (Eur. J. Biochem. 270) 1453
these effects. We constructed 11 mutants of E. coli UhpC in
which we exchanged single arginine residues and attempted
to classify the effects into the above categories based on
changes in the K
M
or V

m
determinations that are independent of the amount
of protein, the effect of these mutations on the trans-
location pathway required that V
max
and the relative
insertion of UhpC into the membrane be measured. This
composite value is shown as Ôspecific activityÕ in Table 1.
These five mutants ranged from a fivefold decrease to a
3.6-fold increase with respect to wild-type activity.
Similarly, six mutations at five positions (R318A, R152K,
R437C, R46C, R266C, R318K) caused the same modest
decrease in the affinity for Glc6P in the transport aspect
of UhpC. Again, the effect of these five mutations on
K
(induction)
was remarkably variable. While R437C changed
only fourfold with respect to the wild-type, the K
(induction)
of
R266C increased 245-fold, three mutants (R318A, R318K,
R152K) became constitutive, and the K
(induction)
of R46C
became so high (low affinity) that it was not measurable.
The specific activities measured ranged from a 0.7-fold
decrease to a sixfold increase with respect to wild-type
activity (Table 1). Interestingly, Maloney and coworkers
showed that R46 is critical for transport function of UhpT
[19], but the major effect of the R46C mutation in UhpC

Æh
)1
)
Membrane
incorporation (% wild-type)
Specific activity
(nmolÆmg
)1
Æh
)1
) K
(induction)
(l
M
)
His-UhpC 63 110 100 110 3.8
Arginine mutants
R46C 135 105 22 480 n.m.
R149C 30 154 39 395 2646
R152C 23 30 13 231 0.86
R152A 46 9 37 24 3.2
R152K 82 61 9 678 Constitutive
a
R204C 33 31 32 97 7.3
R266C 145 75 93 81 932
R318C 45 145 81 179 Constitutive
a
R318A 73 152 63 241 Constitutive
a
R318K 150 106 67 158 Constitutive

with insertional mutations in the central hydrophilic loop
between TM6 and TM7 led to the assumption that this
domain, in contrast with the corresponding domain in
UhpT, is not of major importance for the UhpC phenotype
[22]. However, the conservation of the amino acid sequence
of the central hydrophilic loop in UhpC proteins is
remarkable (Fig. 2). Therefore, to investigate whether single
conserved amino acid residues in the central hydrophilic
loop are critical for sensing and/or transport by UhpC we
mutated three conserved residues in this region (Fig. 2):
G213 (that is conserved in all proteins aligned); H222 (that
only appears in UhpC proteins with a complete uhp locus-
E. coli, S. enterica and P. multocida); and D223 (that
appears in all UhpC-like proteins including HPTcp, but
not in the UhpT proteins).
The G213V exchange altered both the transport and the
sensor aspect of the mutated UhpC protein. No transport
activity was measurable with this UhpC protein and its
presence resulted in near wild-type UhpT activity that was
constitutively expressed in the absence of Glc6P during
induction (Table 1, and Fig. 3). Interestingly, the charge-
reversal mutation D223K lost both activities and was
unable to either transport Glc6P or sense Glc6P in the
medium. The H222Q mutant, like the other two loop
mutants, was unable to transport Glc6P. However, most
significantly, this mutant showed intact sensing activity and
it responded to an even lower concentration of Glc6P in the
medium than the wild-type as illustrated by the sevenfold
lower K
(induction)

Although most of the residues mutated in UhpC are highly
conserved in UhpC and UhpT proteins (Fig. 2), and in the
case of R46 and R266 had been shown to be critical for
transport in UhpT [19], 10 of 11 mutations had modest
Fig. 4. Determination of the K
(induction)
of the UhpT-inducing system in
UhpC-deficient E. coli cells BL21(DE3) (uhpC::Tn1000) harbouring the
pET16b construct which encodes the mutated UhpC-H222Q protein.
The cells were induced for 15 min with given Glc6P concentrations.
For quantification of uptake cells were incubated for 1 min with 10 l
M
[
14
C]Glc6P. Inset: The Hanes analysis revealed an apparent K
(induction)
of 0.53 l
M
and a V
max(induction)
of 150 nmolÆmg protein
)1
Æh
)1
.
Fig. 3. Complementation of the UhpC-deficient E. coli strain BL21
(DE3)(uhpC::Tn1000) with the pET16b constructs encoding UhpC
mutants R152C/A/K, R318C/A/K or G213V. The corresponding cul-
tures were either grown with (+) or without (–) 100 l
M

less than 10% of wild-
type but could be fully induced. However, because insertion
of six of the 11 mutated UhpC proteins into the cell
membrane was less than 50% of the insertion in the wild-
type, the calculated transport activity per membrane-
inserted molecule was up to sixfold more than wild-type
and was very low only in the case of R152A (Table 1).
Although the analysis of site-directed mutants of UhpT led
to the hypothesis that arginine residues R46 and R275
(corresponding to 46 and R266 in UhpC, Fig. 2) were
involved in the binding of the transport substrate Glc6P
[19], our observations demonstrate that it is not valid to
transfer data about single amino acid residues critical for
transport by UhpT to UhpC.
In contrast to the modest effects on transport, the
effects of the arginine mutations on induction were large
and varied. The concentration of Glc6P that gave 50%
induction of UhpT increased from 3.8 l
M
to 932 l
M
in
R266C, to 2646 l
M
in R149C, and was so high in R46C
that it could not be measured. This suggests that the
affinity for Glc6P as a transported substrate is uncoupled
in a UhpC molecule from its affinity for Glc6P as an
inducer; this is seen most dramatically in R149C in which
the affinity for Glc6P as the transport substrate increased

exchange) increased the specific transport activity of UhpC
about 14 times in accordance with previous findings
indicating the importance of this salt bridge for transport
by UhpT [25] (Table 1). However, UhpC with this salt
bridge was unable to sense exogenous Glc6P and induce
UhpT. Curiously, this is essentially the same phenotype seen
in the arginine mutant R46C where we removed, rather than
introduced, a residue that was essential to transport by
UhpT. Again, this suggests that in a UhpC molecule the
affinity for Glc6P as a transported substrate is not related to
its affinity for Glc6P as an inducer. It is worth mentioning
that removal of this salt bridge from UhpT does not confer
signalling activity to this transporter when expressed in a
UhpC-deficient strain (data not shown). Thus, removal of
this salt bridge from UhpC after gene duplication appears
necessary to allow sensing activity, but was not sufficient to
create a sensor.
Function of amino acid residues located in the central
loop of UhpC
The alignment reveals that UhpC-like proteins exhibit
a number of highly conserved residues located in the
predicted central hydrophilic loop that are different in
UhpT-like proteins (Fig. 2). Previous observations had
suggested that the large central hydrophilic loop of UhpC
might not be important for exhibiting the Uhp phenotype
[22]. However, the reciprocal exchange D223K abolished
both transport and sensing and the mutant G213V is
constitutive and lacks the ability to transport Glc6P
(Table 1). The mutant H222Q also lacks transport activity
but remarkably possesses an increased affinity for sensing

1456 C. Schwo
¨
ppe et al. (Eur. J. Biochem. 270) Ó FEBS 2003
5. Marger, M.D. & Saier, M.H. (1993) A major superfamily of
transmembrane facilitators that catalyse uniport, symport and
antiport. Trends Biol. Sci. 18, 13–20.
6. O
¨
zcan,S.,Dover,J.,Rosenwald,A.G.,Wo
¨
lfl, S. & Johnston, M.
(1996) Two glucose transporters in Saccharomyces cerevisiae
are glucose sensors that generate a signal for induction of gene
expression. Proc. Natl. Acad. Sci. USA 93, 1–5.
7. Lalonde,S.,Boles,E.,Hellmann,H.,Barker,L.,Patrick,J.W.,
Frommer, W.B. & Ward, J.M. (1999) The dual fucntion of sugar
carriers: transport and sugar sensing. Plant Cell 11, 707–726.
8. Schwo
¨
ppe, C., Winkler, H.H. & Neuhaus, H.E. (2002) Properties
of the glucose 6-phosphate transporter from Chlamydia pneumo-
niae (HPTcp) and the glucose 6-phosphate sensor from Escherichia
coli (UhpC). J. Bacteriol. 184, 2108–2115.
9. Stephens,R.S.,Kalman,S.,Lammel,C.,Fan,J.,Marathe,R.,
Aravind,L.,Mitchell,W.,Olinger,L.,Tatusol,R.L.,Zhao,Q.,
Koonin, E.V. & Davis, R.W. (1998) Genome sequence of an
obligate intracellular pathogen of humans: Chlamydia tracho-
matis. Science 282, 754–759.
10. Postma, P.W., Lengeler, J.W. & Jacobson, G.R. (1993) Phos-
phoenolpyruvate: carbohydrate phosphotransferase systems of

Arabidopsis sucrose transporter, is expressed in cells adjacent to
the vascular tissue and in a carpel cell layer. Plant J. 24, 869–882.
18. Riordan, J.F. (1979) Arginyl residues and anion binding in pro-
teins. Mol. Cell. Biochem. 26, 71–92.
19. Fann,M.,Davies,A.H.,Varadhachary,A.,Kuroda,T.,Sevier,
C., Tsuchiya, T. & Maloney, P.C. (1998) Identification of two
essential arginine residues in UhpT, the sugar phosphate anti-
porter of Escherichia coli. J. Memb. Biol. 164, 187–195.
20. Hall, J.A., Fann, M.C. & Maloney, P.C. (1999) Altered substrate
selectivity in a mutant of an intrahelical salt bridge in UhpT, the
sugar phosphate carrier of Eschericheria coli. J. Biol. Chem. 274,
6148–6153.
21. Friedrich, M.J. & Kadner, R.J. (1987) Nucleotide sequence of the
uhp region of Escherichia coli. J. Bacteriol. 169, 3556–3563.
22. Island, M.D. & Kadner, R.J. (1993) Interplay between the mem-
brane-associated UhpB and UhpC regulatory proteins. J. Bac-
teriol. 175, 5028–5034.
23. Sambrook, J., Fritsch, E.F. & Maniatis, T. (1989) Molecular
Cloning: A Laboratory Manual, Vol. 3, 2nd edn. Cold Spring
Harbor Laboratory Press, Cold Spring Harbor, New York, USA.
24. Bradford, M.M. (1976) A rapid and sensitive method for the
quantification of microgram quantities of protein utilizing the
principle of protein-dye binding. Anal. Biochem. 72, 248–254.
25. Hall, J.A. & Maloney, P.C. (2001) Transmembrane segment 11 of
UhpT, the sugar phosphate carrier of Escherichia coli,isanalpha-
helix that carries determinants of substrate selectivity. J. Biol.
Chem. 276, 25107–25113.
26. Alexeyev, M.F. & Winkler, H.H. (1999) Membrane topology of
the Rickettsia prowazekii ATP/ADP translocase revealed by novel
dual pho-lac reporters. J. Mol. Biol. 285, 1503–1513.