Báo cáo khoa học: Identification of a domain in the a-subunit of the oxaloacetate decarboxylase Na+ pump that accomplishes complex formation with the c-subunit pot

Identification of a domain in the a-subunit of the
oxaloacetate decarboxylase Na
+
pump that accomplishes
complex formation with the c-subunit
Pius Dahinden, Klaas M. Pos* and Peter Dimroth
Institute of Microbiology ETH Zu
¨
rich, ETH Ho
¨
nggerberg, Zu
¨
rich, Switzerland
Oxaloacetate decarboxylase is a member of the sodium
ion transport decarboxylase (NaT-DC) enzyme family
which also includes methylmalonyl-CoA decarboxy-
lase, malonate decarboxylase, and glutaconyl-CoA
decarboxylase [1–3]. These enzymes are found exclu-
sively in anaerobic bacteria. They convert the free
energy of a specific decarboxylation reaction into an
electrochemical gradient of Na
+
ions which plays a
profound role in the energy metabolism of these bac-
teria [4].
Oxaloacetate decarboxylase is a membrane-bound
enzyme complex composed of subunits a, b and c with
molecular masses of approximately 63–65, 40–45, and
9–10 kDa, respectively. The a-subunit is located peri-
pheral to the membrane. It contains the carboxyl-
transferase domain in the N-terminal part and the

biotin-binding domain in the C-terminal part [5]. The
b-subunit is an integral membrane protein with nine
membrane-spanning a-helices and a fragment inserting
into the membrane but not traversing it [6]. The c-sub-
unit contacts the b-subunit with its N-terminal a-heli-
cal region and the a-subunit with its hydrophilic
C-terminal domain. One histidine residue of the histi-
dine triplet near the C terminus of c is specifically
required for complex formation with the a-subunit.
The c-subunit therefore plays an important role in the
in the assembly of the a ⁄ b ⁄ c-complex [7,8].
Vibrio cholerae contains two oxaloacetate decarboxy-
lase-encoding gene clusters, termed oad-1 and oad-2.
The flanking regions of oad-1 do not code for enzymes
of a specific metabolic pathway in which the oxaloace-
tate decarboxylase could participate. In contrast, the
Keywords
association domain; flexible linker peptide;
oxaloacetate decarboxylase; protein–protein
interaction; sodium ion transport
decarboxylase
Correspondence
P. Dimroth, Institute of Microbiology ETH
Zu
¨
rich, ETH Ho
¨
nggerberg, Wolfgang-Pauli-
Strasse 10, CH-8093 Zu
¨

rich, Switzerland
E-mail:
*Present address
Institute of Physiology, University of Zu
¨
rich,
Winterthurerstrasse 190, CH-8057 Zu
¨
rich,
Switzerland
(Received 18 October 2004, revised
1 December 2004, accepted 10 December
2004)
doi:10.1111/j.1742-4658.2004.04524.x
The oxaloacetate decarboxylase Na
+
pumps OAD-1 and OAD-2 of Vibrio
cholerae are composed of a peripheral a-subunit associated with two integ-
ral membrane-bound subunits (b and c). The a-subunit contains the carb-
oxyltransferase domain in its N-terminal portion and the biotin-binding
domain in its C-terminal portion. The c-subunit plays a profound role in
the assembly of the complex. It interacts with the b-subunit through its
N-terminal membrane-spanning region and with the a-subunit through its
hydrophilic C-terminal domain. The biochemical and structural require-
ments for the latter interaction were analysed with OAD-2 expression
clones for subunit a-2 and the C-terminal domain of c-2, termed c¢-2. If
the two proteins were synthesized together in Escherichia coli they formed
a complex that was stable at neutral pH and dissociated at pH<5.0. An
internal stretch of 40 amino acids of a-2 was identified by deletion muta-
genesis to be essential for the binding with c¢-2. This portion of the a-sub-

unit is connected via flexible linker peptides to the carboxyltransferase
domain at its N terminus and to the biotin-binding domain at its C termi-
nus. Results of site-directed mutagenesis indicated that a conserved tyrosine
(491) and threonine 494 of this peptide contributed significantly to the sta-
bility of the complex with c¢-2. This peptide therefore represents a newly
identified, separate domain of the a-subunit and has been called the ‘asso-
ciation domain’.
846 FEBS Journal 272 (2005) 846–855 ª 2005 FEBS
oad-2 genes are part of the citrate fermentation operon
and accordingly, the oad-2 genes are expressed during
anaerobic growth of V. cholerae on citrate (data not
shown).
The catalytic cycle starts with the transfer of the
carboxyl group from oxaloacetate to the prosthetic
biotin group. The carboxyltransfer reaction is catalysed
at low rates by the a-subunit alone and with about
1000 times higher rates by the a ⁄ c-complex [8]. This
rate increase has been attributed to polarizing the car-
bonyl oxygen bond of oxaloacetate by the Zn
2+
metal
ion on the c-subunit which is therefore part of the
carboxyltransferase active site [8]. In the next step the
carboxybiotin switches from the carboxyltransferase
site to the decarboxylase site on the b-subunit. Two
Na
+
ions pass through the cytoplasmic access channel
contributed in part by the highly conserved helix VIII
and bind to specific sites in the middle of the

membrane [9–12]. According to a mechanistic model,
binding of the second Na
+
ion to the Y229- and S382-
including site abstracts the phenolic proton from
Y229. The proton is thought to move through the
channel to the carboxybiotin where it catalyses the
decarboxylation of this acid-labile compound [11]. This
event triggers a conformational change by which the
cytoplasmic channel closes and the periplasmic channel
opens. The two Na
+
ions then diffuse into the peri-
plasmic reservoir and a proton diffuses from the peri-
plasm to Y229 where it restores the phenolic hydroxyl
group [11,13]. Overall, the decarboxylation of one
oxaloacetate leads to the transport of two Na
+
ions
into the periplasm and the consumption of a periplas-
mically derived proton [14,15].
This sophisticated machinery requires specific flexi-
ble segments for the mechanical movements side by
side with segments that guarantee the structural integ-
rity of the three-subunit complex. A remarkable region
is the extended proline ⁄ alanine linker between the two
domains of the a-subunit of oxaloacetate decarboxy-
lase from Klebsiella pneumoniae. Such an extended
linker peptide is not apparent, however, in the two
oxaloacetate decarboxylases (OAD-1 or OAD-2) of

V. cholerae, and the corresponding segments of the
OADs known so far differ widely in the linker region
and the flanking sequences on both sides. Nevertheless,
all of these segments contain numerous proline, alan-
ine and serine residues that probably contribute the
flexibility necessary for catalysis. Interestingly, also the
segments of the cytosolic domains of the c-subunits
differ widely among species. As a interacts with c sup-
posedly via amino acids in its C-terminal part, these
variable regions between the carboxyltransferase
domain and the biotin-binding domain might consti-
tute a specific interacting interface. Here we probed
the interacting parts between subunits a and c by dele-
tion and site-specific mutagenesis with the OAD-2 of
V. cholerae. The binding domain on a was identified as
a stretch of 40 amino acids (480–520) that is flanked at
its N terminus by the carboxyltransferase domain and
at its C terminus by the biotin-binding domain. This
portion of the a-subunit has been termed the associ-
ation domain. Particularly important amino acids in
this domain for complex stability were Y491 and
T494.
Results
Complex formation between the a- and
c-subunits and dissociation at acidic pH
A detailed analysis of the interaction of the C-terminal
domain of the c-subunit with the a-subunit was per-
formed using the recombinantly synthesized sub-
units ⁄ domains of the OAD-2 from V. cholerae.As
shown in Fig. 1 the c-subunits of the OAD-2 and

OAD-1 isoforms harbour the Zn
2+
-binding motif pre-
viously identified in the c-subunit of the OAD from
K. pneumoniae. The histidine triplet near the C terminus
of the c-subunit is also conserved in c-2 of V. cholerae.
For practical reasons the membrane part of c-2 was
substituted by a peptide of 10 histidine residues. The
resulting soluble protein (c¢-2) was synthesized together
with the a-subunit (a-2) in Escherichia coli. These two
Fig. 1. Domain structure and catalytic zinc binding motif of the
oxaloacetate decarboxylase c -subunits from K. pneumoniae and
V. cholerae. The sequences of the c-subunits of the oxaloacetate
decarboxylases from K. pneumoniae and V. cholerae are compared.
They have a common domain structure: a short periplasmic seg-
ment (amino acids 1–11) is followed by a transmembrane segment
(amino acids 12–32, indicated by the rectangular box) to which the
cytosolic domain is linked by a flexible linker peptide rich in proline
and alanine residues. Characteristic for subunit c-2 from V. cholerae
and subunit c from K. pneumoniae is a histidine triplet at the C-ter-
minal end. In the Klebsiella c-subunit D62, H77 and H82 are
involved in Zn
2+
binding (indicated by arrows) and H78 is essential
for binding of the a -subunit [8]. Amino acid residues supposed to
be involved in Zn
2+
binding by the Vibrio c-subunits are D62, H77
and H79 of c-1 and E71, H81 and H83 of c-2 (indicated by arrows).
P. Dahinden et al. Association domain of oxaloacetate decarboxylase

FEBS Journal 272 (2005) 846–855 ª 2005 FEBS 847
proteins assembled within the E. coli cells to a stable
a-2 ⁄ c¢-2-complex as both subunits are purified together
by Ni–NTA or monomeric avidin–Sepharose chroma-
tography which specifically bind the His
10
tag on c¢-2
or the biotin group on a-2, respectively. The stability
of the complex was investigated after binding the
a-2 ⁄ c¢-2-complex to a monomeric avidin–Sepharose
column. The complex was stable during washing with
buffer at neutral pH. However, with citrate buffer of
pH < 5.0 the complex dissociated and only a-2 was
retained on the avidin column (Fig. 2). The separated
subunits reassociated at pH > 5.0 to a stable complex
that was retained on a Ni–NTA agarose column
(Fig. 2). Therefore, amino acid residues which become
protonated at pH < 5.0 seem to be involved in the
binding of a-2 to c-2. The most likely candidates are
the histidine residues or the glutamate residue at the
C-terminal end of c-2 (Fig. 1).
Effect of point mutations in the C-terminal
domain of c¢-2 on complex stability
To investigate whether one of the histidine residues of
the histidine triplet at the C terminus of c-2 is import-
ant for the interaction with a-2, each histidine was
mutated individually to alanine. To analyse the
complex formation between a-2 and the c¢-2 mutants,
the C-terminal part of a-2 (a-2-C) was synthesized
together with c¢-2 and mutants thereof in E. coli. The

construct of a-2-C covers the 151 C-terminal residues
of a-2 and has an N-terminal extension of the four
residues MTVD. The construct contains the C-terminal
biotin-binding domain and upstream segments of a-2.
As expected, the entire a-2-C protein formed a strong
complex with the wild-type c¢-2 protein. In the
mutants c¢-2-H82A and c¢-2-H83A binding of a-2-C
was not affected, as shown by the copurification of the
c¢-2 mutant proteins with a-2-C by affinity chromato-
graphy on avidin-Sepharose (Fig. 3). The c¢-2-H81A
mutant protein, however, was only copurified in sub-
stoichiometric amounts with a-2-C. This indicates that
the histidine at position 81 of the c-subunit contributes
to the stability of the a ⁄ c complex. It was unclear,
however, which residues of a-2 participate in the inter-
action. To answer this question, the complex stability
was analysed with various deletion mutants of a-2.
Complex formation between a-2 deletion
mutants and c¢-2
To elucidate the binding domain for c¢-2 on a-2, a
number of C-terminal deletion mutants of a-2 were
generated. The a-2 deletion mutants were then synthes-
ized together with c¢-2 in E. coli and cell extracts were
subjected to Ni–NTA affinity chromatography to iso-
late c¢-2 via its His
10
tag. The competence of a-2 dele-
tion mutants for complex formation with c¢ -2 could
thus easily be assessed by the copurification of both
proteins. Deletions of a-2 with up to 80 amino acid

residues from the C terminus were copurified with c¢-2
showing that this part of the protein is not involved in
Fig. 2.
10
Dissociation and reassociation of c¢-2 and a-2. The dissoci-
ation of the proteins that were coexpressed in E. coli and purified
by Ni
2+
–NTA affinity chromatography was achieved by binding the
protein to avidin–Sepharose and washing with buffer of pH < 5.
The a-2 subunit still bound to the avidin–Sepharose was then
eluted with biotin. Reassociation was analysed by combining the
dissociated c¢-2 with the eluted a-2 at pH 8.0. Dissociation and
reassociation was analysed by SDS ⁄ PAGE. Two micrograms of pro-
tein were loaded on each lane and the gel was stained with silver.
M, Bio-Rad
11
broad molecular mass standard (Bio-Rad Laboratories
AG, Reinach, Switzerland); 1, a-2 ⁄ c¢-2 complex purified by Ni
2+
–
NTA; 2, wash fraction pH 6.0 of purified a-2 ⁄ c¢-2 applied to avidin-
Sepharose; 3, wash fraction pH 5.0; 4, wash fraction pH 4.0; 5,
wash fraction pH 8.0; 6, elution fraction (pH 8.0); 7, flow-through
fraction of reassociated a-2 ⁄ c¢-2 applied to Ni
2+
–NTA agarose; 8,
wash fraction; 9, elution fraction.
Fig. 3. Complex formation of c¢-2 point mutants with a -2-C. The
proteins were coexpressed in E. coli and complex formation was

analysed by SDS ⁄ PAGE following affinity chromatography on avi-
din–Sepharose. Two micrograms of protein were loaded on each
lane and the gel was stained with silver. 1, a-2-C ⁄ c¢-H81A; 2, a-2-
C ⁄ c¢-H82A; 3, a-2-C ⁄ c¢-H83A; 4, wild-type a-2-C ⁄ c¢-2; M, Bio-Rad
broad molecular mass standard.
Association domain of oxaloacetate decarboxylase P. Dahinden et al.
848 FEBS Journal 272 (2005) 846–855 ª 2005 FEBS
the complex formation. Deletion mutants of a-2 lack-
ing 100 or more C-terminal amino acid residues,
however, were not copurified with c¢-2 (Fig. 4A),
indicating that amino acids 80–100 from the C termi-
nus comprise part of the binding domain. This part of
the protein is upstream of the putative linker segment
connecting the C-terminal biotin-binding domain with
the residual parts of a-2 (Fig. 5).
For further analyses of the domain of a-2 which is
crucial for complex formation with c-2, the C-terminal
part of a-2 (a-2-C) was synthesized together with c¢-2 in
E. coli. As was shown above, the entire a-2-C protein
formed a strong complex with the c¢-2 domain. This
situation did not change if up to 20 amino acid residues
were deleted from the N terminus of a-2-C. However,
the binding affinity between a-2-C and c¢-2 was reduced
if 30 amino acids were deleted and was abolished com-
pletely after deleting 40 or more amino acid residues
(Fig. 4B). From these results we conclude that a domain
of approximately 40 amino acids of a-2-C is essential
for the binding of the a- to the c-subunit (Fig. 5).
In separate experiments it was shown that the isola-
ted N- and C-terminal portions of the a-subunit did not

associate to form a complex. For this purpose a-2-N
(residues 1–453) and a-2-C (residues 449–599) were syn-
thesized separately in E. coli and incubated together
before applying the mixture to a monomeric avidin–
Sepharose column. Only a-2-C was retained and speci-
fically eluted with biotin (data not shown).
Effect of point mutations in the binding domain
of a-2-C on the binding to c¢-2
To elucidate whether single amino acids in the binding
domain of a-2-C were particularly important for the
formation of a stable complex with c¢-2, a series of
conservative point mutations were constructed on the
AB
Fig. 4. Complex formation of a-2-C deletion mutants with c¢-2. The
proteins were coexpressed in E. coli and complex formation was
analysed by SDS ⁄ PAGE following affinity chromatography on
Ni–NTA agarose. Two micrograms of protein were loaded on each
lane and the gel was stained with silver. (A) D20–D120, deletions of
20–120 amino acids from the C terminus of a-2. (B) D90–D140,
remaining 90–120 amino acids after deletion of amino acids at the
N terminus of a-2-C. M, Bio-Rad broad molecular mass standard.
Fig. 5. Alignment of the C-terminal sequences of subunit a from K. pneumoniae and subunit a-2 from V. cholerae. The sequences shown
include the biotin-binding domains (light grey) with the biotin-binding lysine residue 35 residues before the C terminus, the newly discovered
association domains (dark grey) and upstream sequences. The association domains are highly conserved in the central portion but deviate
significantly within their distal parts. The central portion of 20 amino acids of the association domain includes Y491 and T494 which were
shown by site-specific mutagenesis to have a major impact on the stability of the a-2 ⁄ c-2 complex. The association domains are flanked on
both sides by linker peptides (black bars above and below sequence) containing an accumulation of proline and alanine residues. These linker
peptides were predicted by two independent programs,
PSIPRED [26,27] and PSA [28,29]. A particularly extended linker peptide is present in
the downstream region of the association domain of K. pneumoniae. The point mutations which have been introduced into a-2 of V. cholerae

are shown in the lines labeled ‘mutants 1’ and ‘mutants 2’. The deletions introduced are marked by D20–D140. The corresponding numbers
indicate deletions from the C terminus of a-2 or the number of remaining amino acids of a-2-C.
P. Dahinden et al. Association domain of oxaloacetate decarboxylase
FEBS Journal 272 (2005) 846–855 ª 2005 FEBS 849
plasmid synthesizing a-2-C and c¢-2 (see Fig. 5,
‘mutants 1’). The complex stability of mutant proteins
was then assessed by avidin–Sepharose affinity chroma-
tography. From 16 mutants 13 had no significant effect
on complex stability, but in mutants Y491F, T494V
and D509N the stability of the complex was affected
(Fig. 6A). To further analyse the significance of these
residues for complex formation each one was individu-
ally exchanged to alanine (see Fig. 5, ‘mutants 2’). In
the mutants a-2-C-Y491A and a-2-C-T494A complex
formation with c ¢-2 was completely impaired. In
contrast, the stability of the complex between the a-2-C-
D509A mutant and c¢-2 was similar to that of the wild-
type (Fig. 6B). The complex with the Y491F ⁄ D509N
double mutation was less stable than that with the
Y491F mutation but significantly more stable than that
with the Y491A mutation (data not shown).
The c¢-2 protein could only be isolated from E. coli
expression hosts that also synthesized a-2 or a-2-C
indicating that c¢-2 is degraded in the cells if it is not
present as a complex with a-2. This observation can
conveniently be used to assess complex formation
in vivo with some of the mutants described above. No
c¢-2 could be detected upon coexpression with the a-2-
C-Y491A or a-2-C-T494A mutant confirming the
importance of Y491 and T494 for proper binding to

c¢-2. In contrast, c¢-2 was not degraded upon coexpres-
sion with the a-2-C mutants Y491F, T494V and
D509N. These results indicate complex formation
in vivo between c¢-2 and the a-2-C mutants in spite of
the fact that these complexes were not strong enough
to survive washing with buffer on a monomeric avidin
affinity column. The c¢-2 that was eluted in the wash-
ing step was competent to form a native-like complex
with a-2-C, as both proteins were copurified with a
Ni–NTA agarose column which specifically binds the
His-tag of c¢-2 (Fig. 7A). Complete degradation of c¢-2
was also observed if it was synthesized in E. coli
together with the deletion mutant a-2-Del120. How-
ever, if wild-type a-2-C was also synthesized by these
cells, c¢-2 was not degraded and could be isolated as a
stable complex with a-2-C (Fig. 7B).
Discussion
Upstream of the biotin domain of the a-subunit is a
linker peptide which in case of the K. pneumoniae
enzyme, with which most of the fundamental biochem-
istry has been explored, contains mostly proline and
alanine residues. Peptides with this composition are
known to be very flexible and such a flexible region
seems to be required for moving the carboxybiotin
from the carboxyltransferase site on the N-terminal
domain of the a-subunit to the decarboxylase site on
the b-subunit. Regions of high flexibility within a pro-
tein are disadvantageous, however, for structural stud-
ies because they may prevent the protein from
adopting a uniform conformation which is the pre-

requisite for crystallization. In this context we investi-
gated the genome sequences of various oxaloacetate
decarboxylase containing organisms and found that
V. cholerae contains the genes for two different oxalo-
acetate decarboxylases, named OAD-1 and OAD-2,
which both lack the extended proline ⁄ alanine linker in
AB
Fig. 7. Complex formation of c¢-2 with a-2-C mutants. c¢-2 was coex-
pressed with a-2-C-mutants in E. coli. One of these proteins was
subsequently purified via its specific tag and complex formation with
the other one analysed by SDS ⁄ PAGE with 2 lg protein and silver
staining. (A) The a-2-C-Y491F ⁄ c¢-2-complex was bound to monomer-
ic avidin–Sepharose. The majority of c¢-2 was washed off the
column, and the fraction eluted with biotin contained mainly a-2-C-
Y491F (1). The wash fractions containing c¢-2 were incubated with
wild-type a-2-C and subjected to Ni–NTA chromatography, resulting
in the purification of a stable a-2-C ⁄ c¢-2-complex (2). (B) If a-2-
Del120 ⁄ c¢-2 was expressed in E. coli, c¢-2 was degraded. To demon-
strate the expression of c¢-2 wild-type a-2-C was coexpressed
together with the other two proteins and the extract chromato-
graphed by a Ni–NTA column. a-2-Del120 and wild-type a-2-C were
found in the flow-through (FT) and wash (W) fractions and a stable
a-2-C ⁄ c¢-2-complex was eluted with imidazole (lane E). M, Marker
proteins.
A
B
Fig. 6. Complex formation of c¢-2 point mutants with a-2-C. The
proteins were coexpressed in E. coli and complex formation was
analysed by SDS ⁄ PAGE following affinity chromatography on avidin–
Sepharose. Two micrograms of protein were loaded on each lane

and the gel was stained with silver. The c¢-2 subunit bands are
shown. (A) The indicated conservative point mutants were loaded.
(B) The indicated alanine mutants were loaded. wt, Wild-type protein.
Association domain of oxaloacetate decarboxylase P. Dahinden et al.
850 FEBS Journal 272 (2005) 846–855 ª 2005 FEBS
the a-subunit. We reasoned therefore that these
enzymes might be more suitable for structural studies.
Preliminary experiments with OAD-2 from V. cholerae
indicated improved stability properties as compared to
the OAD from K. pneumoniae, and we therefore deci-
ded to perform further studies with this enzyme.
Here, we have identified a domain of 40 amino acid
residues within the C-terminal portion of 151 amino
acids of the a-subunit which is responsible for the for-
mation of a stable complex with the c-subunit and thus
is essential for the assembly of the a ⁄ b ⁄ c-complex. This
assembly domain is located just upstream of the puta-
tive linker peptide that forms the connection to the bio-
tin-binding domain (Fig. 5). The linker peptide of
OAD-2 from V. cholerae contains three proline and five
alanine residues within a stretch of 18 amino acid resi-
dues (514–531) while the OAD from K. pneumoniae has
seven proline and 15 alanine residues within the 27
amino acids forming the linker peptide (502–528).
Upstream of the assembly domain of OAD-2 there is a
stretch of 25 amino acid residues (455–479) containing
five proline and five alanine residues which could serve
as another flexible region within the protein. The cor-
responding flexible region of the OAD from K. pneu-
moniae comprises residues 450–476 and contains five

proline and nine alanine residues. The central part of
the association domain of the OAD from K. pneumo-
niae (residues 489–509, Fig. 5) is reasonably well con-
served. Interestingly, this part of the association
domain contains all three residues which were shown
by site-directed mutagenesis to contribute significantly
to the stability of the complex. Two of the residues
(Y491 and D509) are conserved in the K. pneumoniae
OAD and T494 is exchanged by a glutamate. These
results establish a three-domain-structure for the a-sub-
unit consisting of the N-terminal carboxyltransferase
domain and the C-terminal biotin-binding domain
which are connected by the association domain sand-
wiched by a flexible linker peptide on both sides.
Astonishingly, the mutant a-2-C-D509A affected
complex formation with c¢-2 only marginally although
the mutant a-2-C-D509N had a major destabilizing
impact on complex formation. As the mutation D509N
is much more conservative than the mutation D509A
the mutation D509A probably causes a conformational
change in a-2-C resulting in a rearrangement of the
binding surface which in turn allows another residue to
take over the role of D509. Different acidic residues are
not far from D509 which could alternatively take over
its role. The mutation D509N on the other hand is very
conservative and therefore supports the assumption
that the negative charge of this residue is of importance
for the complex formation with c-2.
The dissociation and association of the OAD-com-
plex from K. pneumoniae was shown previously to be

pH dependent following titration curves with inflection
points at pH 6.5 which suggested that a histidine plays
an important role in the assembly of the enzyme [16].
According to this model, the enzyme could assemble
with the crucial histidine in the neutral form and
would dissociate if the histidine becomes protonated.
In accordance with this hypothesis it was found by
mutagenesis of the OAD from K. pneumoniae that H78
of the c-subunit plays a crucial role in the formation
of the a ⁄ c-complex. H78 is part of a cluster of four
histidine residues near the C terminus of the c-subunit
of which H77 and H82 together with D62 were shown
to be ligands for Zn
2+
binding [8]. A similar histidine
cluster also exists at the C terminus of the c-subunit of
the OAD-2 from V. cholerae, suggesting that two of
the histidines are Zn
2+
ligands, whereas one may be
involved in complex formation (Fig. 3). This role of a
histidine is compatible with the observation that the
a-2 ⁄ c¢-2-complex is stable at neutral pH but dissociates
at pH < 5.0. By mutagenesis of H81 of c¢-2 to alanine
but not by mutagenesis of the other histidines of the
cluster, the stability of the a-2 ⁄ c¢-2-complex was signi-
ficantly affected, indicating that H81 probably is
involved in the interaction between a and c.
We would like to emphasize that the results presen-
ted here not only provide new structural information

but may also reveal a dynamic aspect of the enzyme’s
function. It is now clear that the a-subunit binds the
c-subunit with a distinct association domain which is
flanked on both sides with proline- and alanine-rich
linker peptides. These linker peptides may allow hinge
movements of the association domain against the carb-
oxyltransferase and the biotin domain. The dynamics
of conformational motions within the catalytic cycle of
the enzyme probably also includes the motion of the
entire soluble part of the enzyme against the mem-
brane anchor of subunits c and b because a pro-
line ⁄ alanine linker peptide also connects the membrane
segment and the soluble domain of the c-subunit [7]. A
concerted action of these hinge movements may be
required to move the prosthetic biotin or carboxybio-
tin group back and forth between the carboxyltrans-
ferase and decarboxylase catalytic sites on subunits a
and b, respectively, as the enzyme operates.
Experimental procedures
Strains and growth conditions
For general cloning purposes E. coli DH5a (Bethesda
Research Laboratories, Gaithersburg, MD, USA)
1
was used.
P. Dahinden et al. Association domain of oxaloacetate decarboxylase
FEBS Journal 272 (2005) 846–855 ª 2005 FEBS 851
For overexpression of protein the strain E. coli RNE41(DE3)
(a gift from B Miroux
2
, CNRS-CEREMOD, Meudon,

France) was used. Strains were routinely grown at 37 °C and
180 r.p.m. in baffled Erlenmeyer flasks containing 200 mL to
2 L Luria–Bertani medium containing 10 gÆL
)1
NaCl. The
medium was inoculated with 1% of an overnight culture
and incubated at 37 °C and 180 r.p.m. At an attenuance
3
at
600 nm (D
600
)of 0.7 the cultures were cooled on ice,
and expression was induced by the addition of isopropyl
thio-b-d-galactside (100 lm). The cells were harvested after
incubation for additional 3–4 h at 30 °C and 180 r.p.m.
Recombinant DNA techniques and sequencing
Genomic DNA was prepared by the CTAB method
according to Ausubel et al. [17]. Extraction of plasmid
DNA, restriction enzyme digestions, DNA ligations, and
transformation of E. coli with plasmids were carried out
by standard methods [17,18]. PCRs were performed with
an air thermo-cycler (Idaho Technology
4
, Salt Lake City,
UT; model 1605) using Pfu polymerase. Oligonucleotides
used for mutagenesis were custom-synthesized by Micro-
synth (Balgach, Switzerland). All inserts derived from
PCR as well as ligation sites were checked by DNA
sequencing according to the dideoxynucleotide chain-
termination method [19] by Microsynth. In the case of

site-directed mutagenesis by PCR whole plasmids were
amplified, but only the sequence of the genes to be over-
expressed was verified by sequencing.
Construction of expression plasmids
The primers used for site-directed mutagenesis are listed in
the Supplementary material. Genomic DNA prepared from
V. cholerae O395-N1 [20] served as template for the amplifi-
cation of the oad-1 and oad-2 genes by PCR [21]. For the
expression of oadA-2 with an N- or C-terminal His tag
oadA-2 was amplified from pET24-VcoadGAB-2 harbour-
ing the oad-2 genes, with the oligonucleotide primers Vco-
adA2_for and VcoadA2_rev containing an NdeIoranXhoI
site, respectively. The PCR product was ligated directly
with pKS vector restricted with EcoRV. Positive clones
selected by blue ⁄ white screening were restricted with NdeI
and XhoI, and the obtained oadA-2 fragment was ligated
with accordingly restricted pET16b or pET24b vector to
give pET16-VcoadA-2 or pET24-VcoadA-2, respectively
(Fig. 2A).
The N-terminal carboxyltransferase domain and the
C-terminal biotin-binding domain of oadA-2 were amplified
from pET24-VcoadGAB-2 with the oligonucleotide primers
VcoadA2-NT_NdeI and VcoadA2-NT_XhoI or VcoadA2-
CT_NdeI and VcoadA2-CT_Xho I, respectively. The accord-
ingly restricted PCR products were ligated with vector
pET16b restricted with the same enzymes to give pET16-
VcoadA-2-N or pET16-VcoadA-2-C, respectively, for the
expression of the two VcOadA-2 domains with an N-ter-
minal His tag (Fig. 2B).
To examine complex formation between c-2 and a-2 a

construct was made for the coexpression of the C-terminal,
cytosolic domain of c-2 (c¢-2, C-terminal 59 amino acids)
with a-2 or C-terminal deletion mutants of a-2. PCR prod-
ucts comprising the appropriate DNA fragments were
obtained by using the oligonucleotide primers summarized
in the Supplementary material. The oligonucleotide primers
VcoadG(CT)A2_for and VcoadG(CT)A2_rev were used to
amplify an oadG¢A-2 fragment from the vector pET24-Vco-
adGAB-2. This fragment and the vector pET16b were
restricted with NdeI and XhoI and ligated to give the vector
pET16-VcoadG¢A-2. From this clone oadG¢A-2-DelNN
fragments were amplified with the primer Vco-
adG(CT)A2_for and one of the primers with the suffix
‘_DelNN’, where ‘NN’ is substituted with the number of
amino acids missing in the corresponding gene product.
The obtained fragments and vector pET16b were restricted
with NdeI and XhoI and ligated to give the plasmids
pET16-VcoadG¢A-2-DelNN (Fig. 2C).
To get deletion mutants from the N-terminal end of
the C-terminal part of a-2, a construct for the coexpres-
sion of c¢-2 and the 151 C-terminal amino acids of a-2
was made. The oligonucleotide primers used to generate
appropriate DNA fragments are summarized in the Sup-
plementary material. In a first PCR run the primers
VcOG¢_to_A-CT_fo and VcOG¢_to_A-CT_mre containing
an XagIoraSalI site, respectively, were used to amplify
the oadG¢ fragment and the oligonucleotide primers
VcOG¢_to_A-CT_mfo and VcOG¢_to_A-CT_re containing
a SalIoraHindIII site, respectively, were used to
amplify the oadA-2-C fragment. The two PCR products

of the first run were used as templates in a second PCR
together with the oligonucleotide primers VcOG¢_to_
A-CT_fo and VcOG¢_to_A-CT_re to give the fragment
oadG¢A-2-C, which was restricted with XagI and HindIII
and ligated with pET16-VcoadG¢A-2 restricted with the
same enzymes to give the vector pET16-VcoadG¢A-2-C.
From this clone oadG¢A-2-C-DelNN fragments were
amplified with the oligonucleotide primer VcOG¢-A-2-
CT_re containing one Pvu I site and one of the primers
with the suffix ‘_DelNN’, where ‘NN’ is substituted with
the number of amino acids missing in the corresponding
gene product. The fragments were restricted with SalI
and PvuI and ligated with the plasmid pET16-VcoadG¢A-
2-C restricted with the same enzymes to give the plasmids
pET16-VcoadG¢A-2-C-DelNN (Fig. 2D).
To examine the competition of a-2-C-D120 with a-2 or
a-2-C for binding to c¢-2 the oadA-2 and oadA-2-C genes
were cloned into the vector pET124b. The vector was con-
structed with the p15A instead of the ColE1 origin of repli-
cation [22] and was therefore compatible with the vectors
pET16b and pET24b. The oadA-2 gene was amplified
from pET24-VcoadGAB-2 with the oligonucleotide primers
Association domain of oxaloacetate decarboxylase P. Dahinden et al.
852 FEBS Journal 272 (2005) 846–855 ª 2005 FEBS
VcoadA-2_for and VcoadG(CT)A2_rev. The PCR product
was restricted with NdeI and XhoI and ligated into
pET124b restricted with the same enzymes to give pET124-
VcoadA-2 for the expression of OadA-2 without tag. To
get pET124-VcoadA-2-C for the expression of a-2-C with-
out tag the vector pET16-VcoadA-2-C was restricted with

NdeI and XhoI and the VcoadA-2-C fragment was ligated
with pET124b restricted with the same enzymes.
Site directed mutagenesis by PCR
Site directed mutagenesis was performed essentially as
described by Fisher and Pei [23]. Each reaction contained
in 50 lL 20 pmol of one of the complementary oligonu-
cleotide primer pairs summarized in the Supplementary
material and 30 ng of the plasmid pET16-VcoadG¢A-2-C
as template. DNA was amplified by 12 cycles 15 s at
95 °C, 15 s at 56 °C, and 13 min 30 s at 68 °C with Pfu
polymerase. Before the first cycle the DNA was dena-
tured for 2 min at 95 °C, and after the last cycle the
samples were incubated for an additional 8 min at 68 °C.
After cooling to 4 °C the PCR products were treated
with DpnI for 1 h at 37 °C to cut the parental DNA
strand by adding the enzyme directly to the PCR mix-
ture. After heat inactivation for 10 min at 65 °C5lLof
the digested PCR samples were used to transform E. coli
DH5a.
Preparation of cytosolic fraction and membranes
For the preparation of cell extracts, cells obtained from
expression cultures were resuspended in 7 mLÆg
)1
of cells
(wet weight) of a suitable buffer. After addition of
0.2 mm diisopropylfluorophosphate (final concentration)
and approximately 50 lg DNase I, the cells were disrup-
ted by three passages through a French pressure cell at
110 MPa. Intact cells and cell debris were removed by
centrifugation (30 min at 8000 g), and the cell-free super-

natant was subjected to ultracentrifugation (1 h at
200 000 g) to separate the cytosolic fraction and the
membrane fraction.
Purification VcOadA-2, VcOadA-2-C, and
VcOadG¢A-2 and its derivatives by monomeric
avidin-Sepharose affinity chromatography
The plasmids containing the corresponding genes were
transferred into and expressed in E. coli RNE41(DE3). The
cells obtained from expression cultures were resuspended in
buffer A (50 mm Tris ⁄ HCl pH 8.0, 250 mm NaCl) contain-
ing 1 mm MgK
2
EDTA. The cytosolic fraction was pre-
pared as described above and applied to a monomeric
avidin–Sepharose column, which was washed with 7 bed
volumes of buffer A. Biotinylated protein was finally eluted
with 1 bed volume of buffer A containing 5 mm (+)-d-
biotin.
Purification of VcOadG¢A-2 and its derivatives
by Ni
2+
–NTA chromatography
The plasmids containing the corresponding genes were
transferred into and expressed in E. coli RNE41(DE3). The
cells obtained from expression cultures were resuspended in
HisBind buffer (20 mm Tris ⁄ HCl pH 8.0, 500 mm NaCl)
containing 10 mm imidazole, the cytosolic fraction prepared
as described above and then applied to a Ni
2+
–NTA-

agarose column (2 mL bed volume, Qiagen AG, Basel,
Switzerland)
6
, pre-equilibrated with the same buffer. The
column was washed with 10 bed volumes HisBind buffer
containing 20 mm imidazole and with 8 bed volumes
HisBind buffer containing 25 mm imidazole. Finally, the
bound protein was eluted with 4 bed volumes of HisBind
buffer containing 150 mm imidazole.
Dissociation of VcOadG¢A-2 complex
The plasmid pET16b-VcOadG¢A-2 was transferred into and
expressed in E. coli RNE41(DE3). The a-2 ⁄ c¢-2 complex
expressed by these cells was purified by Ni–NTA chroma-
tography. The obtained elution fraction was applied to an
avidin–Sepharose column which was subsequently washed
with two bed volumes citrate buffer pH 6.0 (20 mm Na-
citrate, 250 mm NaCl), 2 bed volumes citrate buffer pH 5.0,
two bed volumes citrate buffer pH 4.0 and two bed vol-
umes Tris buffer pH 8.0 (50 mm Tris ⁄ HCl pH 8.0, 250 mm
NaCl). The wash fractions with citrate buffer pH 5.0 and
4.0 and Tris buffer pH 8.0 each were collected in two bed
volumes of 250 mm Tris ⁄ HCl pH 8.0, 250 mm NaCl. a-2
was eluted with two bed volumes Tris buffer containing
5mm (+)-d-biotin. The neutralized wash fractions and the
elution fraction were combined and incubated overnight at
4 °C. To prevent unspecific binding NaCl and imidazole
were added to a final concentration of 500 mm and 10 mm,
respectively, before applying the sample to a Ni–NTA–
agarose column (2 mL bed volume, Qiagen), pre-equili-
brated with buffer A. The column was washed with four

bed volumes HisBind buffer (20 mm Tris ⁄ HCl pH 8.0,
500 mm NaCl) containing 20 mm imidazole and with four
bed volumes HisBind buffer containing 25 mm imidazole.
Finally, the bound protein was eluted with four bed vol-
umes of HisBind buffer containing 150 mm imidazole.
Protein detection methods
Protein concentration was determined by the BCA method
(Pierce, Lausanne, Switzerland)
7
using BSA as standard.
SDS ⁄ PAGE was performed as described
8
[24]. Gels were
stained with Coomassie Brilliant Blue R 250 or with silver [25].
P. Dahinden et al. Association domain of oxaloacetate decarboxylase
FEBS Journal 272 (2005) 846–855 ª 2005 FEBS 853
Secondary structure prediction
Two different programs were used for the prediction of sec-
ondary structure elements such as flexible regions: psipred
[26,27] and psa [28,29].
Acknowledgements
This work was supported by Swiss National Science
Foundation. We thank Dr Miroux for the gift of the
strain E. coli RNE41(DE3).
References
1 Buckel W (2001) Sodium ion-translocating decarboxy-
lases. Biochim Biophys Acta 1505, 15–27.
2 Dimroth P (1997) Primary sodium ion translocating
enzymes. Biochim Biophys Acta 1318, 11–51.
3 Dimroth P

9
(2004) Molecular basis for bacterial growth
on citrate or malonate. Module 3.4.6, In EcoSal –
Escherichia coli and Salmonella: Cellular and Molecular
Biology [Online] />ASMPress, Washington, D.C.
4 Dimroth P & Schink B (1998) Energy conservation in
the decarboxylation of dicarboxylic acids by fermenting
bacteria. Arch Microbiol 170, 69–77.
5 Schwarz E, Oesterhelt D, Reinke H, Beyreuther K &
Dimroth P (1988) The sodium ion translocating oxalo-
acetate decarboxylase of Klebsiella pneumoniae.
Sequence of the biotin-containing a-subunit and rela-
tionship to other biotin-containing enzymes. J Biol
Chem 263, 9640–9645.
6 Jockel P, Di Berardino M & Dimroth P (1999) Mem-
brane topology of the b-subunit of the oxaloacetate
decarboxylase Na
+
pump from Klebsiella pneumoniae.
Biochemistry 38, 13461–13472.
7 Laussermair E, Schwarz E, Oesterhelt D, Reinke H,
Beyreuther K & Dimroth P (1989) The sodium ion
translocating oxaloacetate decarboxylase of Klebsiella
pneumoniae. Sequence of the integral membrane-bound
subunits b and c. J Biol Chem 264, 14710–14715.
8 Schmid M, Wild MR, Dahinden P & Dimroth P (2002)
Subunit c of the oxaloacetate decarboxylase Na
+
pump:
interaction with other subunits ⁄ domains of the complex

and binding site for the Zn
2+
metal ion. Biochemistry
41, 1285–1292.
9 Jockel P, Schmid M, Choinowski T & Dimroth P
(2000) Essential role of tyrosine 229 of the oxalo-
acetate decarboxylase b-subunit in the energy coupling
mechanism of the Na
+
pump. Biochemistry 39, 4320–
4326.
10 Jockel P, Schmid M, Steuber J & Dimroth P (2000) A
molecular coupling mechanism for the oxaloacetate
decarboxylase Na
+
pump as inferred from mutational
analysis. Biochemistry 39, 2307–2315.
11 Schmid M, Vorburger T, Pos KM & Dimroth P (2002)
Role of conserved residues within helices IV and VIII of
the oxaloacetate decarboxylase b subunit in the energy
coupling mechanism of the Na
+
pump. Eur J Biochem
269, 2997–3004.
12 Wild MR, Pos KM & Dimroth P (2003) Site-directed
sulfhydryl labeling of the oxaloacetate decarboxylase
Na
+
pump of Klebsiella pneumoniae: helix VIII compri-
ses a portion of the sodium ion channel. Biochemistry

42, 11615–11624.
13 Dimroth P, Jockel P & Schmid M (2001) Coupling
mechanism of the oxaloacetate decarboxylase Na
+
pump. Biochim Biophys Acta 1505, 1–14.
14 Di Berardino M & Dimroth P (1996) Aspartate 203 of
the oxaloacetate decarboxylase b-subunit catalyses both
the chemical and vectorial reaction of the Na
+
pump.
EMBO J 15, 1842–1849.
15 Dimroth P & Thomer A (1993) On the mechanism of
sodium ion translocation by oxaloacetate decarboxy-
lase of Klebsiella pneumoniae. Biochemistry 32, 1734–
1739.
16 Dimroth P & Thomer A (1988) Dissociation of the
sodium-ion-translocating oxaloacetate decarboxylase of
Klebsiella pneumoniae and reconstitution of the active
complex from the isolated subunits. Eur J Biochem 175,
175–180.
17 Ausubel FM, Brent R, Kingston RE, Moore DD, Seid-
man JG, Smith JA & Struhl K (1999) Short Protocols in
Molecular Biology, 4th edn. John Wiley & Sons, Inc.,
New York.
18 Sambrook J, Fritsch EF & Maniatis T (1989) Molecular
Cloning: a Laboratory Manual, 2nd edn, Cold Spring
Harbor Laboratory Press, Cold Spring Harbor.
19 Sanger F, Nicklen S & Coulson AR (1977) DNA
sequencing with chain-terminating inhibitors. Proc Natl
Acad Sci USA 74, 5463–5467.

20 Mekalanos JJ, Swartz DJ, Pearson GD, Harford N,
Groyne F & de Wilde M (1983) Cholera toxin genes:
nucleotide sequence, deletion analysis and vaccine devel-
opment. Nature 306, 551–557.
21 Dahinden P, Auchli Y, Granjon T, Taralczak M, Wild
M & Dimroth P. Oxaloacetate decarboxylase of Vibrio
cholerae: purification, characterization and expression of
the genes in Escherichia coli. Arch Microbiol in press.
22 Schneider K, Dimroth P & Bott M (2000) Biosynthesis
of the prosthetic group of citrate lyase. Biochemistry 39,
9438–9450.
23 Fisher CL & Pei GK (1997) Modification of a PCR-
based site-directed mutagenesis method. Biotechniques
23, 574.
24 Scha
¨
gger H & von Jagow G (1987) Tricine-sodium
dodecyl sulfate-polyacrylamide gel electrophoresis for
the separation of proteins in the range from 1 to 100
kDa. Anal Biochem 166, 368–379.
Association domain of oxaloacetate decarboxylase P. Dahinden et al.
854 FEBS Journal 272 (2005) 846–855 ª 2005 FEBS
25 Wray W, Boulikas T, Wray VP & Hancock R (1981)
Silver staining of proteins in polyacrylamide gels. Anal
Biochem 118, 197–203.
26 McGuffin LJ, Bryson K & Jones DT (2000) The
PSIPRED protein structure prediction server. Bioinfor-
matics 16, 404–405.
27 Jones DT (1999) Protein secondary structure prediction
based on position-specific scoring matrices. J Mol Biol

292, 195–202.
28 Stultz CM, White JV & Smith TF (1993) Structural
analysis based on state-space modeling. Protein Sci 2,
305–314.
29 Stultz CM, Nambudripad R, Lathrop RH & White JV
(1997) Predicting protein structure with probabilistic
models. In Protein Structural Biology in Bio-Medical
Research (N Allewell & C Woodward, eds), pp. 447–
506. JAI Press, Greenwich.
Supplementary material
The following material is available from http://www.
blackwellpublishing.com/products/journals/sup pmat/
EJB/EJB4524/EJB4524sm.htm
Table S1. Primers used for construction of the expres-
sion plasmids and for mutagenesis of c and a and
figure
12
depicting the constructs obtained.
Fig. S1. Construction of expression vectors to examine
complex formation between c-2 and a-2.
P. Dahinden et al. Association domain of oxaloacetate decarboxylase
FEBS Journal 272 (2005) 846–855 ª 2005 FEBS 855