Functional characterization of the maltose ATP-binding-cassette
transporter of
Salmonella typhimurium
by means of monoclonal
antibodies directed against the MalK subunit
Anke Stein
1
, Martina Seifert
2
, Rudolf Volkmer-Engert
2
,Jo¨ rg Siepelmeyer
3
, Knut Jahreis
3
and Erwin Schneider
1
1
Humboldt Universita
¨t zu Berlin, Institut fu
¨r Biologie, Berlin, Germany;
2
Humboldt Universita
¨t zu Berlin, Institut fu
¨r Medizinische
Immunologie, Berlin;
3
Universita
¨t Osnabru
¨ck, Fachbereich Biologie/Chemie, Germany
The maltose ATP-binding cassette transporter of Salmonella
typhimurium is composed of a membrane-associated com-
plex (MalFGK
2
) and a periplasmic receptor (MalE). In
addition to its role in transport, the complex acts as a
repressor of maltose-regulated gene expression and is subject
to inhibition in the process of inducer exclusion. These
activities are thought to be mediated by interactions of the
ATPase subunit, MalK, with the transcriptional activator,
MalT, and nonphosphorylated enzyme IIA of the glucose
phosphotransferase system, respectively. To gain further
insight in protein regions that are critical for these functions,
we have generated nine MalK-specific monoclonal anti-
bodies. These bind to four nonoverlapping linear epitopes:
60-LFig-63 (5B5), 113-RVNQVAEVLQL-123 (represented
by 4H12), 309-GHETQI-314 (2F9) and 352-LFREDG
SACR-361 (represented by 4B3). All mAbs recognize their
epitopes in soluble MalK and in the MalFGK
2
complex with
K
d
values ranging from 10
)6
to 10
)8
M
. ATP reduced the
affinity of the mAbs for soluble MalK, indicating a confor-
mational change that renders the epitopes less accessible.
4H12 and 5B5 inhibit the ATPase activity of MalK and the
MalE/maltose-stimulated ATPase activity of proteolipo-
somes, while their Fab fragments displayed no significant
effect. The results suggest a similar solvent-exposed position
of helix 3 in the MalK dimer and in the intact complex and
might argue against a direct role in the catalytic process. 4B3
and 2F9 exhibit reduced binding to the MalFGK
2
complex
in the presence of MalT and enzyme IIA
Glc
, respectively,
thereby providing the first direct evidence for the C-terminal
domain of MalK being the site of interaction with the reg-
ulatory proteins.
Keywords: ABC transporter; MalFGK
2
; enzyme IIA
Glc
;
MalT; monoclonal antibodies.
The family of ATP-binding-cassette (ABC) transport sys-
tems comprises an extremely diverse class of membrane
proteins that couple the energy of ATP hydrolysis to the
translocation of solutes across biological membranes [1,2].
A prototype ABC transporter is composed of four
entities: two membrane-integral domains, which presuma-
bly constitute a translocation pore, and two ATPase
domains (also referred to as ABC subunits/domains), that
provide the energy for the transport process. The ABC
domains are characterized by a set of canonical Walker A
and B motifs, required for nucleotide binding and by a
unique signature sequence (LSGGQ motif) of still unknown
function [3]. The crystal structures of several prokaryotic
ABC domains have been solved in recent years that agree
largely on the overall folds. Accordingly, the structures can
be subdivided in an F
1
-type ATP-binding domain, encom-
passing both Walker sites, a specific a-helical subdomain,
containing the LSGGQ motif and a specific antiparallel-b-
subdomain [4–7].
The binding protein-dependent maltose/maltodextrin
transporter of enterobacteria, such as Escherichia coli and
Salmonella typhimurium, is a well-characterized model
system for studying the mechanism of action of the ABC
transport family [8]. Based on computational analysis, it
belongs to a subclass of ABC importers designated CUT1
(carbohydrate uptake transporter) [9] or OSP (oligosaccha-
rides and polyols) [10], respectively. Members of this
subclass transport a variety of di- and oligosaccharides,
glycerol phosphate and polyols and are recognized by their
common subunit composition (two individual membrane-
spanning subunits and two copies of a single ABC protein)
and by an extension of approximately a hundred amino acid
residues at the C-terminus of the ABC protein [11].
The maltose transporter of E. coli/S. typhimurium is
composed of the periplasmic maltose binding protein, MalE,
and of the membrane-associated complex, MalFGK
2
,con-
sisting of one copy each of the hydrophobic subunits MalF
and MalG and two copies of the nucleotide-binding subunit
MalK [12]. Crystals of Salmonella MalK are available [13]
but their structure could not be solved yet. However, the
tertiary structure of a MalK homolog, isolated from the
Correspondence to E. Schneider, Humboldt Universita
¨t zu Berlin,
Mathematisch-Naturwissenschaftliche Fakulta
¨tI,
Institut fu
¨r Biologie, Bakterienphysiologie, Chausseestr. 117,
D-10115 Berlin, Germany.
Tel.: + 49 (0)30 2093 8121, Fax: + 49 (0)30 2093 8126,
E-mail: erwin.schneider@rz.hu-berlin.de
Abbreviations: IF-medium, Iscove’s DMEM/NUT MIX F12.
(Received 27 March 2002, revised 6 June 2002, accepted 8 July 2002)
Eur. J. Biochem. 269, 4074–4085 (2002) FEBS 2002 doi:10.1046/j.1432-1033.2002.03099.x
hyperthermophilic archaeon Thermococcus litoralis,was
recently determined [5]. Two molecules are present per
asymmetric unit that contact each other through the ATPase
domains with the C-terminal domains attached at opposite
poles. Based on these data, a 3D model of the E. coli MalK
protein was recently presented [14].
Enterobacterial MalK can be purified in fairly large
amounts [15] and displays a spontaneous ATPase activity
that is insensitive to inhibition by vanadate, a typical
inhibitor of ABC transporters [16]. The purified MalFGK
2
complex, when incorporated into liposomes, also exhibits a
low intrinsic ATPase activity that, however, is stimulated
severalfold in the presence of substrate-loaded MalE and is
vanadate-sensitive [12,17–19].
According to a current transport model, the presence of
substrate in the medium is thought to be signalled by
liganded MalE via interaction with externally exposed
peptide loops of MalF and MalG [20]. As a consequence,
conformational changes of the latter are transmitted to the
MalK subunits which, in turn, become activated. Hydro-
lysis of ATP would then trigger subsequent conformational
changes that eventually lead to the translocation of the
substrate molecule. Recent findings suggested that these
steps occur rather simultaneously [21].
Interaction of MalK with the hydrophobic subunits
involves contact of residues in the helical subdomain
with conserved cytoplasmic loops (EAA motifs) in MalF
and MalG [22–24]. This view, based on suppressor
mutational analyses and cross-linking studies, is largely
consistent with the recently solved crystal structure of
MsbA, an ABC transporter mediating the export of
the lipid A component of the E. coli outer membrane
[25].
Besides acting as an import system for maltose/malto-
dextrins, the MalFGK
2
complex is involved in the regula-
tion of genes belonging to the maltose regulon [8]. In the
absence of substrate, the idle transporter is thought to
interact with the positive transcriptional regulator, MalT,
via the MalK subunits, thereby preventing MalT from
binding to its target sequences upstream of maltose-
regulated promotors. When the transporter becomes
engaged in translocating maltose across the membrane,
MalT is released and transcription of maltose-regulated
genes can occur [26].
In addition, the maltose transporter is subject to inhibi-
tion by binding of dephosphorylated enzyme IIA of the
glucose transporter (phosphoenolpyruvate phosphotrans-
ferase system) to the MalK subunits in a process called
inducer exclusion in the context of global carbon regulation
in enteric bacteria [27].
Both regulatory activities of MalK are largely mediated
by the C-terminal domain of the protein [5,28–30].
Obviously, specific protein–protein interactions within
the MalFGK
2
complex as well as between the transporter
and regulatory proteins are crucial for its role in intact cells.
However, in the absence of tertiary structural information
on the complete transporter, these interactions are still
poorly understood at the molecular level. Here, we describe
the use of monoclonal antibodies raised against the MalK
subunit as tools to gain further insights in the structural
basis of transporter functions.
EXPERIMENTAL PROCEDURES
Preparative procedures
MalK [15], MalFGK
2
[18], and MalE [31] were purified
as described. MalE/maltose-loaded proteoliposomes con-
taining the MalFGK
2
complex were prepared by a
detergent dilution procedure as published elsewhere
[18,32].
Enzyme IIA
Glc
was purified from the cytosolic fraction of
E. coli strain BL21 D(pts43crr::kan
R
) harbouring plasmid
pCRL13 (crr on pET23A) [33] by Ni-NTA affinity chro-
matography.
Crude extract containing MalT was prepared according
to [34] from E. coli strain JM109 (Stratagene), carrying
plasmid pAS8 (malT
E.c.
on pSE380, p
trc
,amp
R
) (this study).
For competitive inhibition ELISA, N-terminally his-tagged
MalT was partially purified from strain JM109, harbouring
plasmid pAS9 (malT
E.c.
on pQE9, p
T5
,amp
R
)byNi-NTA
chromatography.
Preparation of mAbs
Ten-week-old-femaleBalb/cmicewereimmunizedintra-
peritoneally either with native MalK or with an
N-terminal fragment (encompassing residues 1–179 [30])
(100 lg each), dissolved in NaCl/P
i
[35]. On day 12, 25
and 62 the animals were boosted with 50 lgofprotein
each. The final boost was given 4 days prior to the fusion.
For hybridoma production spleen cells were isolated and
fused with myeloma cells SP2/0 as described [36] using
poly(ethylene glycol) 1500 as fusion agent. Selection of
hybridoma cells was performed in hypoxanthine, aminop-
terin and thymidine selection medium supplement. Grow-
ing hybridomas were screened by ELISA using MalK as
bound antigen. Selected hybrid cell lines were cloned at
least three times by limiting dilution. Cloned hybridoma
cells were maintained in 20% IF-medium, supplemented
with 70% fetal bovine serum and 10% dimethylsulfoxide
for several days at )80 C and subsequently stored in
liquid nitrogen. For the production of mAbs, cells were
grown in IF or RPMI 1690 medium (Bichrom KG,
Berlin) in 2 L culture flasks.
mAbs were purified by loading concentrated culture
supernatant on a Protein G-Sepharose 4 fast flow matrix
equilibrated with 20 m
M
sodium phosphate buffer, pH 7.
After washing off unbound material, mAbs were eluted with
0.1
M
glycine/HCl, pH 2.7, and immediately dialyzed
against NaCl/P
i
overnight at 4 C.
Isolation of Fab fragments
One millilitre of mAbs (1–3 mg) were mixed with 0.5 mL of
papain-agarose beads in 20 m
M
phosphate buffer, pH 7.5,
supplemented with 20 m
ML
-cysteine and 1 m
M
EDTA, and
incubated overnight at 37 C. Subsequently, Fab fragments
were separated from uncleaved mAbs by incubating the
mixture with protein A–Sepharose for 1 h at 4 C.
Unbound material (Fab fragments) was collected, dialyzed
overnight against 3 L of 50 m
M
Tris/HCl, pH 7.5, and
stored at 4 C until use.
FEBS 2002 Protein–protein interactions of MalFGK
2
(Eur. J. Biochem. 269) 4075
Determination of isotypes
Isotopes of the mAbs were determined by using Roche’s
ISO STRIP-mouse isotyping kit according to the manufac-
turer’s instructions.
Peptide synthesis on cellulose membranes – SPOT
synthesis
Cellulose-bound peptide libraries were automatically pre-
pared on Whatman 50 paper (Whatman, Maidstone, UK)
according to standard SPOT synthesis protocols [37] using a
SPOT synthesizer (Abimed GmbH, Langenfeld, Germany)
as described elsewhere [38–41]. The sequence files were
generated with the software
DIGEN
(Jerini AG, Berlin,
Germany). The peptides were derived from S. typhimurium
MalK. Libraries consisting of 10meric peptides (overlapping
by 9 amino acids) and peptide-substitutional analyses were
synthesized. All peptides are C-terminally attached to
cellulose via a (a-Ala)
2
spacer.
Epitope mapping
The screening of cellulose-bound peptides followed a
protocol published elsewhere [39,40]. Peptide libraries
were incubated with mAbs overnight at 4 C in blocking
buffer (10% blocking reagent, Roche, in TNT, 10%
sucrose) and binding was detected with peroxidase-
conjugated goat anti(mouse IgG) antibody on hyperfilm
(Amersham/Pharmacia, Braunschweig, Germany) using the
Western Blot Chemiluminescence Reagent Plus System of
NEN (Boston, MA, USA).
Peptide synthesis
Peptides, structurally derived from the epitopes identified
after screening of the peptide libraries as described above
were synthesized on solid phase (50 lmol scale) on Tentagel
SRam resin (Rapp Polymere, Tu
¨bingen, Germany) by using
PyBOP activation and a standard Fmoc-chemistry-based
protocol of an AMS 422 Peptide Synthesizer (Abimed,
Langenfeld, Germany). Side-chain protections of amino
acids are as follows: Glu, Asp (OtBu); Ser, Thr, Tyr,
Trp (tBu); His, Lys (Boc); Asn, Gln (Trt); Arg (Pbf).
Trifluoracetic acid /phenole/triisopropylsilane/H
2
O(9.4:
0.1 : 0.3 : 0.2) was used for resin cleavage and side-chain
deblocking. The crude peptides were purified to homogen-
eity by RP-HPLC using the linear solvent gradient 5–60% B
in A for 30 min, with A ¼0.05% trifluoracetic acid in
water, and B ¼0.05% trifluoracetic acid in acetonitrile.
The HPLC had the UV detector at 214 nm, a Vydac C
18
column of 20 ·250 mm, and a flow rate 10 mLÆmin
)1
.The
MS were performed on a matrix-assisted laser desorption
ionization-time of flight mass spectrometer (Laser Bench-
TopII, Applied Biosystems). The purity of the product was
characterized by analytical HPLC.
ELISA
For ELISA, microtiter plates were coated with purified
MalK (2.5 pmol) diluted in 100 m
M
sodium carbonate
buffer, pH 9.6, and incubated overnight at 4 C. Remaining
binding sites were blocked with 2% BSA in NaCl/P
i
150 m
M
NaCl, 3 m
M
KCl, 8 m
M
Na
2
HPO
4
·2H
2
O, 1 m
M
KH
2
PO
4
) for 2 h at room temperature. Subsequently, the
wells were incubated overnight at 4 C with mAb diluted in
2% BSA in NaCl/P
i
/Tween (NaCl/P
i
containing 0.5%
Tween 80). Incubation with the second antibody (HRP-
conjugated goat anti(mouse IgG); 5 ·10
)2
-fold dilution]
occurred for 2 h at room temperature. After each step the
excess protein was removed by fourfold washing with NaCl/
P
i
/Tween. Antibody binding was detected by adding 100 lL
of 62.5 lgmL
)1
3,3¢,5,4¢-tetramethylbenzidine, 0.0026%
(v/v) H
2
O
2
in 0.1
M
sodium acetate/0.1
M
citric acid, pH 6.
After 10 min at room temperature,the reaction was stopped
by addition of 100 lLH
2
SO
4
and the color development
was measured at 450 nm.
Binding constants of mAbs were measured by com-
petitive inhibition ELISA according to [42]. Suitable
concentrations of mAbs were first determined by adding
various amounts of antibody to microtiter wells, coated
with various amounts of MalK, under experimental
conditions identical to those used in binding experiments
(ELISA).
The competitive inhibition ELISA was essentially carried
out as described above. In order to allow competition
between bound antigen (MalK) and free antigen (MalK,
MalFGK
2
-containing proteoliposomes, synthetic peptides)
the mAbs and an equal volume of free antigen in different
concentrations were incubated overnight at 4 C in the wells
coated with MalK (2.5 pmol). The mAbs were used in the
following concentrations: 2F9 and 4B3, 2 ·10
)9
M
;4H12,
6.5 ·10
)9
M
;5B5,8·10
)10
M
.
Analytical methods
Hydrolysis of ATP was assayed in microtiter plates
essentially as described in [43]. Protein was assayed using
the BCA kit from Bio-Rad. SDS/PAGE and immunoblot
analyses were performed as described in [44].
RESULTS
Monoclonal antibodies recognize epitopes
in the N-terminal (ATPase) domain and in the C-terminal
domain of MalK, respectively
Monoclonal antibodies were prepared against the MalK
subunit of the S. typhimurium maltose ABC transporter
using purified, nondenatured MalK or an N-terminal MalK
fragment (MalKN1, encompassing residues 1–179 [30]), as
antigen for immunization. Nine individual hybridoma cell
lines producing antibodies of the Ig subclass IgG1 were
obtained and immunoblot analyses revealed a specific
reaction with the corresponding antigen in each case when
purified MalK or MalFGK
2
complex were separated by
SDS/PAGE.
Immunoblots using truncated MalK proteins [30]
suggested that five mAbs, obtained with MalKN1,
recognize epitopes in the N-terminal (ATPase) domain
while the remaining four mAbs, obtained with intact
MalK, bind to the C-terminal (regulatory) domain. For
precise determination of the epitopes overlapping deca-
peptides corresponding to the entire MalK sequence were
synthesized on cellulose membranes by SPOT-synthesis
[37–41]. The results for the binding analyses of the
4076 A. Stein et al.(Eur. J. Biochem. 269)FEBS 2002
different mAbs are shown in Fig. 2. Four peptide epitopes
were identified. One mAb (5B5) recognizes the peptide
53-ETITSGDLTRM-67, located close to the Walker A
motif, four (4H12, 6E6, 3A12, 4D8) bind to
111-NQRVNQVAEVLQL-123, located within the helical
subdomain (helices 2–4, Fig. 1A), three (2F9, 1D8, 2G4)
Fig. 1. Location of epitopes in the amino acid sequence of S. typhimurium MalK (A) and in the modelled 3D structure of E. coli MalK (B). (A) The
Walker A and B motifs and the ABC signature are highlighted in yellow. The epitopes recognized by the mAbs are highlighted in red. Residues that
when mutated render E. coli MalK insensitive to inducer exclusion are underlined while residues that cause a regulatory phenotype when mutated
are doubly underlined [14]. a-Helices and b-strands that have been identified in the structure of T. litoralis MalK [5] are indicated above the
sequences as broken and dotted lines, respectively. Please note that the primary structures of S. typhimurium (acc. no X54292) and E. coli MalK
(acc. no. J01648) differ only by 16 amino acid changes and by the lack of the dipeptide PM in S. typhimurium MalK after residue L258.
Furthermore, A320 (underlined) corresponds to S322 in E. coli MalK. (B) Stereo representation of the model of monomeric E. coli MalK [14]. The
epitopes recognized by the mAbs are indicated in red. The figure was drawn with RasMol 2.6 (http://www.umass.edu/microbio/rasmol) using the
coordinates provided by W. Welte (Universita
¨tKonstanz).
FEBS 2002 Protein–protein interactions of MalFGK
2
(Eur. J. Biochem. 269) 4077
recognize 304-VVEQLGHETQIHIQIP-319 and one (4B3)
binds to 352-LFREDGSACR-361, both located in the
C-terminal domain (Fig. 1A and B). The fact that in each
case strong signals with successive overlapping peptides
were obtained argues in favour of linear rather than
discontinuous epitopes. Only mAbs 5B5, 4H12, 2F9, and
4B3 were further characterized.
In order to identify those amino acid residues that are
indispensable for binding within each epitope substitu-
tional analyses of the peptides were performed. In these
experiments every position was substituted one-at-a-time
by all other genetically encoded amino acids. Thus, all
possible single site substitution analogs were synthesized
and screened. Discrete substitution patterns were
identified (Fig. 3) and the results are summarized in
Table 1.
In the case of 4H12, four residues at the N-terminus
(N111–V114) are not essential for binding. However, the
third and fourth position of the peptide are nonetheless
required as revealed by an additional analyses using
peptides that varied in length at the N- or C-terminal end
or both (not shown). Thus, the minimum epitope encom-
passes residues R113 to L123. Furthermore, E119 and Q122
can be replaced by various amino acids without loss of
binding.
Binding of mAb 5B5 is strongly dependent on the
residues L60–G63 which are either indispensable or can be
substituted only by chemically related amino acids
(Fig. 3B). This result was confirmed by length analysis
(not shown).
Similarly, the data clearly revealed that the peptide G309-
I314 is absolutely essential for binding of mAb 2F9
(Fig. 3C). The observation that 4B3 bound only to one
spot (out of 150) in the peptide scan (Fig. 2D) already
suggested that the peptide L352–R361 would be the
minimum epitope. This notion was basically confirmed by
mutational analyses (Fig. 3D) and by the failure of the mAb
to recognize peptides lacking residues at either the N- or
C-terminus (not shown). Interestingly, substitution of
several residues, in particular L352C, R354N, E355M,
A359I/V and C360F/L, resulted in significantly increased
binding of 4B3.
ATP affects binding of mAbs to soluble MalK
but not to MalFGK
2
-containing proteoliposomes
The affinities of the mAbs for their respective antigens were
determined by competitive inhibition ELISA according to
Friguet et al. (1987) [42], using soluble MalK, proteolipo-
somes containing the MalFGK
2
complex or synthetic
soluble peptides as free antigen. The resulting dissociation
constants are summarized in Table 2.
All mAbs have largely similar affinities for their respective
epitopes in both MalK and the MalFGK
2
complex with K
d
values ranging from 0.1 l
M
(4H12) to 10 l
M
(4B3). This
finding is not only consistent with the surface-exposed
localization of the epitopes in the tertiary structure of MalK
[5] (Fig. 1B) but also suggests that complex assembly is not
accompanied by a significant change in accessibility. None-
theless, the use of synthetic peptides as free antigen resulted
Fig. 2. Binding of mAbs to MalK-derived peptide scans (10-mers). The MalK fragments given below were scanned with cellulose-bound peptides
shifted by one amino acid. The numbers of spots in each row and the total number of rows are indicated above and at the right-hand side of each
blot, respectively. Blots were incubated with mAbs and developed as described in Experimental procedures. (A) mAbs 4H12, 3A12, 4D8, 6E6:
fragment G104–L134, elongated at the N-terminal end by the tripeptide QAA (42 spots in total); peptide sequences read as follows: row 1/spot 1,
empty; 1/2, QAAG104AKKEVM-110; 1/3, AAG104AKKEVMN-111; 1/4, AG104AKKEVMNQ-112 and so forth. (B) 5B5: fragment G51–F98,
elongated at the C-terminal end by the dipeptide RP (41 spots in total); peptide sequences read as follows: row 1/spot 1, 51-GLETITSGDL60; 1/2,
52-LETITSGDLF-61; 1/3, 53-ETITSGDLFI-62 and so forth. (C) 2F9, 1D8, 2G4: fragment R211-V369 (150 spots in total); peptide sequences read
as follows: row 1/spot 1, 211-RVAQVGKPLE220; 1/2, 212-VAQVGKPLEL221; 1/3, 213-AQVGKPLELY222 and so forth. D. 4B3: fragment
R211-V369 (150 spots in total); peptide sequences read as in C. See Fig. 1 A for sequence information.
4078 A. Stein et al.(Eur. J. Biochem. 269)FEBS 2002