Tài liệu Báo cáo khoa học: The distinct nucleotide binding states of the transporter associated with antigen processing (TAP) are regulated by the nonhomologous C-terminal tails of TAP1 and TAP2 ppt

The distinct nucleotide binding states of the transporter associated
with antigen processing (TAP) are regulated by the nonhomologous
C-terminal tails of TAP1 and TAP2
Hicham Bouabe* and Michael R. Knittler
Institute for Genetics, University of Cologne, Germany
The transporter associated with antigen processing (TAP)
delivers peptides into the lumen of the endoplasmic reticu-
lum for binding onto major histocompatibility complex
class I molecules. TAP comprises two polypeptides, TAP1
and TAP2, each with an N-terminal transmembrane domain
and a C-terminal cytosolic nucleotide binding domain
(NBD). The two NBDs have distinct intrinsic nucleotide
binding properties. In the resting state of TAP, the NBD1
has a much higher binding activity for ATP than the NBD2,
while the binding of ADP to the two NBDs is equivalent. To
attribute the different nucleotide binding behaviour of
NBD1 and NBD2 to specific sequences, we generated
chimeric TAP1 and TAP2 polypeptides in which either the
nonhomologous C-terminal tails downstream of the Walker
B motif, or the core NBDs which are enclosed by the con-
served Walker A and B motifs, were reciprocally exchanged.
Our biochemical and functional studies on the different TAP
chimeras show that the distinct nucleotide binding beha-
viour of TAP1 and TAP2 is controlled by the nonhomolo-
gous C-terminal tails of the two TAP chains. In addition, our
data suggest that the C-terminal tail of TAP2 is required for
a functional transporter by regulating ATP binding. Further
experiments indicate that ATP binding to NBD2 is
important because it prevents simultaneous uptake of ATP
by TAP1. We propose that the C-terminal tails of TAP1 and
TAP2 play a crucial regulatory role in the coordination of

nucleotide binding and ATP hydrolysis by TAP.
Keywords: antigen presentation; transporter associated with
antigen processing; endoplasmic reticulum; peptide trans-
port; nucleotide binding domains.
The transporter associated with antigen processing (TAP)
translocates antigenic peptides from the cytosol into the
lumen of the endoplasmic reticulum where the peptides are
loaded onto the major histocompatibility complex (MHC)
class I molecules [1]. Cytotoxic T lymphocytes identify
and eliminate cells harbouring pathogens by monitoring
the peptide–MHC class I complex at the cell surface. TAP-
deficient cell lines show low MHC class I cell surface
expression demonstrating the essential role of TAP for
MHC class I-restricted antigen presentation [1]. TAP
belongs to the ATP binding-cassette (ABC) family of
transporters that use ATP hydrolysis to move a remarkable
variety of substrates across cellular membranes [2]. TAP is
an endoplasmic reticulum membrane protein consisting of
two subunits, TAP1 and TAP2, each of which has an
N-terminal transmembrane domain (TMD) and a
C-terminal cytosolic nucleotide binding domain (NBD).
The four-domain (two TMDs, two NBDs) structure appears
to be general in the ABC-transporters although the chain
composition making up the four domains is variable within
the superfamily. The TMDs are involved in substrate
interaction and translocation whereas the NBDs energize
the transport by ATP hydrolysis. Several conserved sequence
motifs common to the NBDs of all ABC-transporters have
been identified, including the Walker A and B motifs, which
are involved in ATP binding and hydrolysis, the Q- and

D
-
loop, the signature motif and the switch region (Fig. 1A).
Studies on several different ABC transporters [3–9] describe
distinct functional and biochemical properties for the two
NBDs of a single transporter. In the case of TAP we showed,
under experimental conditions not allowing nucleotide
hydrolysis, that TAP1 has a much higher ATP binding
activity than TAP2 [10]. Similar results were reported by
others observations [11–14]. Models of the transport cycle of
TAP were proposed in which the NBDs bind and hydrolyze
nucleotides in an alternating and strongly interdependent
manner [10,12,15]. Reconstitution of purified human TAP
into proteoliposomes has recently allowed the measurement
of the ATPase activity of the transporter [16]. The authors
calculated that a single TAP complex hydrolyses about five
ATP molecules per second to transport two to three peptides,
a rate that is compatible with a requirement for ATP
hydrolysis by both TAP chains for a single transport cycle.
Correspondence to M. R. Knittler, Institute for Genetics, University of
Cologne, Zu
¨
lpicher Strasse 47, 50674 Cologne, Germany.
Fax: + 49 221 4705015, Tel.: + 49 221 470 5292,
E-mail:
Abbreviations: ABC, ATP binding-cassette; CFTR, cystic fibrosis
transmembrane conductance regulator; FACS, fluorescence-activated
cell sorting; MHC, major histocompatibility complex; NBD,
nucleotide binding domain; TAP, transporter associated with
antigen processing; tapasin, TAP-associated glycoprotein;

TMD, transmembrane domain.
*Present address: Max von Pettenkofer-Institut fu
¨
r Hygiene und
Medizinische Mikrobiologie, Mu
¨
nchen, Pettenkofer Str. 9a,
80336 Mu
¨
nchen, Germany.
(Received 18 July 2003, revised 17 September 2003,
accepted 23 September 2003)
Eur. J. Biochem. 270, 4531–4546 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03848.x
The different nucleotide binding behaviours of TAP1
and TAP2 are intrinsic properties of their NBDs [9]. Thus,
the critical sequences responsible must be sought within the
NBDs themselves. The core NBDs of TAP containing the
ATP binding-cassette between the Walker A and B motifs
have an overall sequence homology of about 75%. The
most variable part of the core NBDs, in other ABC trans-
porters as well as TAP, lies within the helical subdomains
Fig. 1. Chimeric TAP variants: sequence exchange and expression in T2 cells. (A)AminoacidalignmentoftheNBDsofratTAP1
a
and rat TAP2
a
.
Sequences were retrieved from the GenBank database (GenBank X57523 and X63854) and aligned using the software
VECTOR NTI
(Informax).
Identical residues are marked by black boxes while grey boxes indicate similar residues. The conserved sequence motifs termed Walker A (WA),

Q-loop, signature motif, Walker B (WB),
D
-loop and switch region (switch) as well as the a
6
-andb
11
-region are indicated on top of the aligned
sequences. The sequences of the core NBDs containing Walker A motif, Q-loop, signature motif, Walker B motif and
D
-loop (residues 506–652 in
TAP1 and residues 494–639 in TAP2) and the C-terminal tails downstream the
D
-loop (residues 653–725 in TAP1and residues 640–703 in TAP2)
that were mutually exchanged between TAP1 and TAP2 are underlined in black. In addition, the amino acid sequence encoded by exon 11
(GenBank AL732652) is underlined by a dashed black line (residues 658–725 in TAP1 and residues 645–703 in TAP2). A vertical line behind the
D
-loop indicates the breakpoint of the truncated TAP2 chain 2DV (after residue 639 in TAP2). The region of the truncated alternative C-terminal
tail in the human splice variant TAP2iso (KTLWKFMIF, in the single amino acid letter code), which is encoded by exon 12, is indicated and
underlined by a grey line. (B) Expression and schematic overview of wild-type and chimeric TAP subunits. T2 transfectants were lysed, separated by
SDS/PAGE and blotted onto nitrocellulose as described (see Materials and methods). Western blots were probed for the different TAP chains by
using antisera D90 (C-term. NBD1), 116/5 (C-term. NBD2) and antibody MAC 394 (core NBD2). A pictorial overview of the wild-type TAP and
the different chimeric TAP subunits termed 1V2, 2V1, 1C2 and 2C1 is shown at the bottom of the analysis. TMDs and NBDs of TAP1 are indicated
in black while the corresponding domains of TAP2 are indicated in grey.
4532 H. Bouabe and M. R. Knittler (Eur. J. Biochem. 270) Ó FEBS 2003
between the Walker A and Walker B motifs. It has been
suggested that this approximately 100 amino acid long
region containing the Q-loop and the signature motif is
mainly involved in interactions with the TMDs rather than
in the catalytic process of the NBDs [17]. Low sequence
homology is also a characteristic feature of the C-terminal

tails directly downstream of the conserved
D
-loop compri-
sing 64 residues in NBD2 and 73 in NBD1 of rat TAP
(Fig. 1A). The overall sequence similarity of these NBD-
segments is below 30%. Structural analysis of the NBD1 of
human TAP showed that the C-terminus is close to the
nucleotide binding site and might play an important role in
modulating the catalytic function [18].
To identify the sequence region that imposes the distinct
nucleotide binding and accordingly the different function-
ality of NBD1 and NBD2, we generated TAP1 and TAP2
chimeras in which either the nonhomologous C-terminal
tails (residues 640–703 in TAP2 and residues 653–725 in
TAP1) or the core NBDs (residues 494–639 in TAP2 and
residues 506–652 in TAP1) were mutually exchanged. For
biochemical and functional characterization, we established
T2 cell lines that stably express either single TAP chains or
different combinations of wild-type and chimeric transpor-
ter subunits. Our findings demonstrate that the distinct
nucleotide binding behaviour of the TAP-NBDs is deter-
mined by the nonhomologous C-terminal tails. A chimeric
NBD2 with the C-terminal tail of NBD1 exhibits the ATP
binding capacity and the function of wild-type NBD1. This
indicates that TAP2 has a catalytically active ATP binding-
cassette, which is functionally regulated by the C-terminal
tail. In accordance with this, we found that truncated TAP2
chains deprived of their C-terminal tails retain the ability to
bind to ADP but cannot mediate the transport function
of TAP. Furthermore, our findings indicate that the

C-terminal control of nucleotide interaction in NBD1 is
morecomplexthaninNBD2.AchimericNBD1withthe
C-terminal tail of NBD2 shows a nucleotide binding
behaviour similar to NBD2 but is defective in exchanging
ADP to ATP. We also provide evidence that ATP binding
in TAP2 prevents simultaneous uptake of ATP by TAP1.
Based on our data, we propose that structural influence
from the C-terminal tails and the conformational cross-talk
between the core NBDs build the mechanistic scaffold for
the alternating catalytic cycle of ATP binding and hydro-
lysis of TAP.
Materials and methods
Cell lines and cell culture
T2 is a human lymphoblastoid cell line that lacks both TAP
genes, and expresses only the HLA-A2 and -B5 class I
molecules [19]. Transfectants of T2 containing rat TAP1
a
and rat TAP2
a
wild-type chains [20] were cultured in
IMDM (Gibco BRL) supplemented with 10% FCS (BIO
Whittaker) and 1 mgÆmL
)1
G418 (PAA, Co
¨
lbe). T2 cells
expressing single TAP chimeras 1N2 or 2N1 (formerly
named 1/2 and 2/1) [9] were cultured in the same medium,
whereas transfectants containing chimeric TAP variants
1–2N1 and 2–1N2 (formerly named 1–2/1 and 2–1/2) [9]

were grown in IMDM supplemented with puromycin
(750 ngÆmL
)1
).
Cloning and expression of chimeric TAP1
and TAP2 chains
The 2.6 and 2.4 kb EcoRI fragments containing full-length
cDNA from rat TAP1
a
and TAP2
a
, respectively [1,21] were
cloned into the multiple cloning site of pBluescript KS
+
(Stratagene). The QuickChange
TM
Site directed mutagenesis
procedure (Stratagene) was used to create a ScaIsiteinTAP1
at position 1904 and in TAP2 at position 1943 (position 1 is
the A of the first AUG). For TAP1 we used the comple-
mentary primers 5¢-GGACGATGCCACCAGTACTCTG
GATGCTGGCAACC-3¢ and 5¢-GGTTGCCAGCATCC
AGAGTACTGGTGGCATCGTCC-3¢ and for TAP2 the
complementary primers 5¢-GGATGAGGCTACCAGTAC
TCTGGACGCCGAGTGCG-3¢ and 5¢-CGCACTCGGC
GTCCAGAGTACTGGTAGCCTCATCC-3¢. All primers
were purchased from ARK/Sigma. The chimeric TAP
construct 1V2 was created by ligation of the 1.6 kb ScaI-
fragment containing the C-terminus of TAP2 to the 3.8 kb
ScaI-fragment containing the TMD and core NBD of TAP1.

Inthecaseof2V1,the1.8kbScaI-fragment containing the
C-terminus of TAP1 was ligated to the 3.8 kb ScaI-fragment
containing the TMD and core NBD of TAP2. To restore the
original amino acid sequence, a further site-directed muta-
genesis was performed using the complementary primers
for variant 1V2 5¢-GGACGATGCCACCAGTGCCCTG
GACGCCGAGTGCG-3¢ and 5¢-CGCACTCGGCGTC
CAGGGCACTGGTGGCATCGTCC-3¢ and comple-
mentary primers 5¢-GGATGAGGCTACCAGTGCCCT
GGATGCTGGCAACC-3¢ and 5¢-GGTTGCCAGCATC
CAGGGCACTGGTAGCCTCATCC-3¢ for 2V1. The
resulting TAP constructs were cloned into the EcoRI site of
pHbApr1neo [22] and sequenced fully in both directions.
Chimera 1V2 encoded residues 1–652 of TAP1 and residues
640–703 of TAP2 and chimera 2V1 encoded residues 1–639
of TAP2 and residues 653–725 of TAP1. TAP variants 1C2
and 2C1 were created by the same site-directed mutagenesis
procedure using the cDNA templates of the chimeric variants
TAP 1/2 (TAP1N2) and TAP 2/1 (TAP2N1) [9]. Chimera
1C2 encoded residues 1–505 and 653–725 of TAP1 and
residues 494–639 of TAP2 and chimera 2C1 encoded residues
1–493 and 640–703 of TAP2 and residues 506–652 of TAP1.
The exchange was performed at the amino acid sequence
positions 647 in TAP1/2 (TAP1N2) and 636 in TAP2/1
(TAP2N1). The C-terminal deletion construct TAP2DVwas
created by introducing a stop codon at position 1919 of
the wild-type TAP2 sequence with site-directed muta-
genesis. Therefore, we used the complementary primers
5¢-GGATGAGGCTACCAGTGC TC TGGACGCCTAG
TGCGAGCAGGC-3¢ and 5¢-GCCTGCTCGCACTAGG

CGTCCAGAGCACTGGTAGCCTCATCC-3¢.AllTAP
constructs were transfected into T2 cells by electroporation
using a Bio-Rad gene pulser at 270 V and 500 lF. After
selection with G418 (1 mgÆmL
)1
) for 4–6 weeks, stable
transfectants were subcloned and screened for TAP chain
expression by Western blotting.
Antibodies
116/5 is a polyclonal rabbit antiserum recognizing the
C-terminus of rat TAP2 chains [20]. D90 is a polyclonal
rabbit antiserum recognizing the C-terminus of rat TAP1
Ó FEBS 2003 Regulation of the nucleotide binding state of TAP (Eur. J. Biochem. 270) 4533
chains [21]. MAC 394 is a monoclonal mouse antibody
(mAb) against rat TAP2
a
[23] derived from immunization
with recombinant His-tagged cytoplasmic domain of rat
TAP2
a
. MAC 394 fails to detect TAP2
u
due to the
polymorphic residues at position 538 and 539 in the core
NBD (M. R. Knittler, unpublished results). 4E is a
conformation-dependent mouse mAb, which recogni-
zes an epitope common to all HLA-B and -C antigens
[24].
Immunoprecipitation and Western blotting
Cells (5 · 10

6
) were washed twice in ice-cold NaCl/P
i
(1.7 m
M
KH
2
PO
4
, 10 m
M
Na
2
HPO
4
, 140 m
M
NaCl,
2.7 m
M
KCl), pH 7.5, prior to solubilization in lysis
buffer [NaCl/P
i
, pH 7.5, containing 1% Triton X-100
(Sigma)]. Immunoprecipitations with anti-rat TAP2 (116/
5) were performed as described previously [23]. Immuno-
precipitates were washed with NaCl/P
i
, 1% Triton
X-100 and eluted with 10 m

M
Tris/HCl pH 8.8 con-
taining 0.5% SDS. Samples were analyzed by
Western blotting treated with specific primary antibody.
Bands were visualized with horseradish peroxidase-
conjugated secondary antibodies (goat anti-rabbit
IgG–HRP) and enhanced chemiluminescence substrate
(Amersham).
Transport assay and peptide cross-linking
Cells (2 · 10
6
) were permeabilized with streptolysin O
(SLO) (2 UÆmL
)1
; Murex). After washing with NaCl/P
i
,
0.5 l
M
radioiodinated peptide S8 (TVDNKTRYR),
10 m
M
ATP, and incubation buffer [50 m
M
Hepes
pH 7.5, 250 m
M
sucrose, 150 m
M
CH

3
COOK, 5 m
M
(CH
3
COO)
2
MgÆ4H
2
O, 1 m
M
dithiothreitol, 1 m
M
Pefabloc
(Boehringer Mannheim), 1.8 lgÆmL
)1
aprotinin (Sigma)]
were added and incubated for 10 min at 37 °C. Following
lysis with 20 m
M
Tris/HCl pH 7.5, 500 m
M
NaCl, 0.1%
Nonidet P-40 (Sigma), transported glycosylated peptides
were isolated with Con A-sepharose (Pharmacia) and
quantitated by gamma counting [25]. For peptide cross-
linking permeabilized cells were incubated with 1 l
M
radioiodinated and HSAB-conjugated peptide S8. Cross-
linking was induced by irradiation with a UV lamp at

254 nm for 5 min on ice. Cells were lysed by adding 1%
Triton X-100 in NaCl/P
i
.
Nucleotide binding assays
The nucleotide binding assay was performed as described
previously [26]. Nucleotide binding experiments were
performedwithN6-coupledATP-,ADP-andAMP-
agarose (Sigma) using Triton X-100 solubilized cell
membranes. For photolabelling of TAP with radiola-
belled 8-azido-ATP, membranes of cells were prepared
and resuspended in 250 m
M
sucrose, 50 m
M
KCl, 2 m
M
MgCl
2
,2m
M
EGTA and 10 m
M
Tris pH 6.8 [26]. Mem-
branes corresponding to 3 · 10
6
cells in a final volume of
100 lL were incubated with 2 l
M
8-azido-ATP[a-

32
P] or
8-azido-ATP[c-
32
P] (ICN Biomedicals) for 5 min at 4 °C.
Cross-linking was induced by irradiation with a UV lamp
at 254 nm for 5 min at 4 °C.
Preparation of microsomal membranes
Microsomes from 10
8
T2 cells expressing chimeric and wild-
type TAP proteins were generated by a sucrose gradient
fractionation [27]. Cells were washed twice with ice cold
NaCl/P
i
, resuspended in 10 mL of 10 m
M
Tris, pH 7.4 with
protease inhibitor cocktail (Complete
TM
Protease Inhibitor,
Roche) and incubated on ice for 10 min. The lysed cells
were then homogenized and centrifuged at 800 g for 5 min
at 4 °C. The resulting supernatants were resuspended in
5mL1.3
M
sucrose buffer [20 m
M
Hepes pH 7.5, 25 m
M

CH
3
COOK, 5 m
M
(CH
3
COO)
2
MgÆ4H
2
O, 1 m
M
dithio-
threitol, protease inhibitor (mix)] and centrifuged again at
800 g at 4 °C for 10 min. The supernatants were then
centrifuged at 68 000 g at 4 °C for 2 h and the membrane
pellets resuspended in 800 lLof0.25
M
sucrose buffer.
Afterwards, 5.6 mL of 2.5
M
sucrose gradient buffer was
added and the suspension overlaid carefully with 2.9 mL of
2
M
and 2.9 mL of 1.3
M
sucrose buffer. About 800 lLof
0.25
M

sucrose buffer was carefully loaded on the top of the
gradient. The sucrose gradient was centrifuged at 100 000 g
for 16 h at 4 °C. The microsomes were collected at the
interface between the 2
M
and 1.3
M
sucrose buffer, diluted
in 20 m
M
Hepes buffer [20 m
M
Hepes (pH 7.5), 25 m
M
CH
3
COOK, 5 m
M
(CH
3
COO)
2
MgÆ4H
2
O, 1 m
M
dithio-
threitol and protease inhibitor cocktail], homogenized and
centrifuged at 68 000 g at 4 °C for 1 h. The microsomal
pellets were resuspended in 200 lL20m

M
Hepes buffer.
Finally, aliquots of 30–50 lL were snap frozen in liquid
nitrogen and stored at –80 °C.
Flow cytometry
Experiments were performed as described previously [26].
Results
The nonhomologous C-terminal tails of the NBDs
control the distinctive nucleotide binding properties
of TAP1 and TAP2
To identify the sequence region within the NBDs that
imposes the different nucleotide binding properties of TAP1
and TAP2, we constructed chimeric TAP chains by
exchanging either the highly homologous core NBDs
(termed C1 and C2) or the less homologous C-terminal
tails downstream of the Walker B motif (termed V1 and V2)
corresponding essentially to exon 11 (Fig. 1A). The cDNAs
for the chimeric TAP chains were stably transfected into
TAP-negative human T2 cells. The expression of chimeric
TAP polypeptides was analyzed in the transfectants by
Western blot (Fig. 1B) with antisera D90 and 116/5, which
recognize the C-terminal 14 amino acids of rat TAP1 [21]
and C-terminal 15 amino acids of rat TAP2 [20], respect-
ively, and the monoclonal antibody MAC 394 [23] which
binds specifically to the core NBD of TAP2 (see Materials
and methods). The chimeric TAP chains termed 2V1 and
1C2 which both have the C-terminal tail of TAP1 and the
core NBD2 were detected both by the antiserum D90
(Fig. 1B, upper) and antibody MAC 394 (Fig. 1B, bottom),
whereas the chimeric TAP chains termed 1V2 and 2C1,

which both have the C-terminal tail of TAP2 were
4534 H. Bouabe and M. R. Knittler (Eur. J. Biochem. 270) Ó FEBS 2003
recognized by the antiserum 116/5 (Fig. 1B, middle). With
the exception of the chimera 2V1, where the expression is
low, all the chimeras were expressed to roughly the same
amount as wild-type TAP chains (Fig. 1B).
Membrane lysates from T2 cell lines expressing the
chimeric TAP chains were incubated with ATP- and ADP-
agarose beads [26]. Bound proteins were eluted and
analyzed by Western blotting. The binding of the chimeric
TAP polypeptides to the nucleotide agaroses was compared
with that of wild-type TAP subunits and of the TAP
chimeras 1N2 and 2N1 with switched NBDs (formerly
named TAP 1/2 and 2/1 [9]) (Fig. 2). The latter chimeras
confirmed that distinct nucleotide binding behaviour is an
inherent property of the NBDs [9]. Thus, TAP1 and 2N1,
both with NBD1, bound efficiently to ATP- as well as to
ADP-agarose whereas TAP2 and 1N2, both with NBD2,
bound only to ADP-agarose (Fig. 2, top and second panels,
left and right). The new TAP chimeric polypeptides 2V1 and
1C2 both bound to ATP- as well as ADP-agarose (Fig. 2,
left column, third and fourth panels from top), whereas the
chimeras 1V2 and 2C1 bound only to ADP-agarose (Fig. 2,
right column, third and fourth panels from top). Thus, the
chimeric NBD consisting of the core NBD2 with the
C-terminal segment of TAP1 confers the nucleotide binding
behaviour of wild-type TAP1 whereas the chimeric NBD1
bearing the core NBD1 and the C-terminal segment of
TAP2 confers the characteristic ADP binding of wild-type
TAP2. Taken together, our data suggest that in the resting

state of the transporter the nonhomologous C-terminal
segments of TAP1 and TAP2, and not the core NBDs,
determine the distinct nucleotide binding properties of the
two polypeptides.
Functional correlation between the C-terminal regulated
nucleotide binding and the transport activity of TAP
From our previous experiments [9] we observed that
chimeric transporter variants with two identical NBDs are
not functional for peptide transport. In contrast to
functional TAP molecules such chimeras have the same
nucleotide binding properties on both polypeptides, either
TAP1-like (ATP and ADP) or TAP2-like (ADP only)
depending on the construct. We asked whether simply
exchanging the C-terminal segment on one chain of such
disabled transporters with two identical NBDs, and thus
modulating the nucleotide binding activity of this chain,
could lead to rescue of the transport function. We therefore
created TAP variants with two identical core NBDs but two
different C-terminal tails by coexpressing either wild-type
TAP1 with 2C1 (TAP variant 1–2C1) or wild-type TAP2
with 1C2 (TAP variant 2–1C2). The expression levels of
these TAP variants were similar to that of wild-type TAP
and of the original nonfunctional TAP variants 1–2N1 and
2–1N2 (Fig. 3A, top) and showed normal subunit assembly
(Fig. 3A, bottom).
We measured the peptide transport function of the new
chimeric transporters in the Neefjes assay [25] using the
iodinated model peptide S8 (TVDNKTRYR, in the single
amino acid letter code). In Fig. 3B, TAP variant 2–1C2
showed a significant recovery in transport activity when

compared to the original variant 2–1N2 with identical
C-terminal segments. The transport efficiency of variant
2–1C2 was 40–55% of that of wild-type TAP. Thus, the
chimeric polypeptide 1C2 seems to acquire not only
the intrinsic nucleotide binding behaviour but also nearly
the full function of wild-type TAP1. In contrast, however,
no peptide transport above background was seen for TAP
variant 1–2C1 (Fig. 3B). The contrasting peptide transport
activities of these reciprocally chimeric TAP molecules were
also reflected in different surface expression levels of mature
MHC class I molecules determined by FACS analysis
(Fig. 3C). Thus, although polypeptide 2C1 adopts the
nucleotide binding behaviour of TAP2 (Fig. 2), the chimeric
chain is not able to express the functional properties of the
wild-type TAP2 polypeptide.
The ability to exchange ADP for ATP is a prerequisite
for the function of the TAP-NBDs
In photo-cross-linking experiments with radioactive
8-azido-ATP, the wild-type TAP1–TAP2 complex shows a
characteristic ratio of ATP binding to the two chains of
about 5 : 1 in favour of TAP1 ([10], Fig. 4A, top panel and
Fig. 4B) reflecting the nucleotide binding capacities of the
two polypeptides in the TAP complex [10,14,28]. A similar
pattern of ATP-labelling was found for the functional TAP
variant 2–1C2 where the labelling efficiency for the 1C2
polypeptide is fourfold higher than for the associated wild-
type TAP2 (Fig. 4A, top panel, and Fig. 4B). In contrast, in
the case of the inactive TAP variant 1–2C1, ATP-cross-
linking was detectable exclusively for the wild-type TAP1
polypeptide (Fig. 4A, top panel). In confirmation of

previous suggestions that the function of the peptide
binding site of TAP is conformationally linked to the
NBDs of both TAP chains, the defective transporter 1–2C1
Fig. 2. Nucleotide binding properties of wild-type and chimeric TAP
subunits. Membrane fractions of T2 transfectants were resuspended in
lysis buffer containing 1% Triton X-100 and incubated with different
nucleotide agaroses. Bound proteins were eluted with SDS-sample
buffer and analyzed in Western blots probed for the C-terminal tail
of TAP1 with antiserum D90 (TAP1, 2N1, 1C2 and 2V1) or the
C-terminal tail of TAP2 with antiserum 116/5 (TAP2, 1N2, 2C1 and
1V2). TAP variants are indicated by pictograms.
Ó FEBS 2003 Regulation of the nucleotide binding state of TAP (Eur. J. Biochem. 270) 4535
was also found to be unable to bind free peptide in a photo-
cross-linking assay, while normal peptide binding was seen
for the functional chimeric transporter 2–1C2 (Fig. 4A,
middle panel). These results together were consistent with
the idea that the chimeric NBD1 in 2C1 is locked in a
conformation that does not allow the binding of ATP
during the transport cycle.
To investigate this, we performed affinity chromato-
graphy with ADP-agarose for the chimeric 1C2 and 2C1
polypeptides as well as the wild-type TAP subunits. The
Fig. 3. Functional properties of different chimeric transporter variants. (A) Expression levels and schematic overview of wild-type and chimeric TAP
transporters (top panel). T2 transfectants were lysed in buffer containing 1% Triton X-100. Lysates were separated by SDS/PAGE and blotted onto
nitrocellulose. Western blot analysis was performed as described in Fig. 1B. TAP variants are indicated by pictograms. Subunit assembly of the
TAP variants 2–1C2 and 1–2C1 (bottom panel). Transfected T2 cells were lysed in 1% Triton X-100 and TAP complexes were immunoprecipitated
with anti-TAP2 serum 116/5. Immunoisolated proteins were separated on an SDS gel and analyzed in Western blots probed for TAP1- (D90) or
TAP2-NBD (116/5). (B) TAP-mediated peptide transport. Transfected and nontransfected T2 cells were permeabilized with streptolysin O and
incubated in transport buffer containing ATP and radioiodinated peptide S8 for 10 min at 37 °C. Bar graphs show the recovered amount of
transported labelled peptides as counts per minute (cpm) and represent the average values of experiments carried out in duplicate. (C) Surface

expression of MHC class I molecules. Cells were incubated with mAb 4E that recognizes HLA-B5 followed by fluorescein isothiocyanate-labelled
secondary antibody. Surface expression of HLA-B5 was detected by flow cytometry (shaded peaks). Mean values of the fluorescence intensity are
indicated. Background staining was determined by incubating only with secondary antibody (nonshaded peaks).
4536 H. Bouabe and M. R. Knittler (Eur. J. Biochem. 270) Ó FEBS 2003
ADP-bound polypeptides were eluted with increasing
concentrations of free MgATP (0–1.0 m
M
) (Fig. 4C). As
can be seen from Fig. 4C, the wild-type TAP1 and TAP2
chains could both be released from the ADP-agarose by
MgATP, and as expected [9,10], much more efficiently in the
case of TAP1 than of TAP2 (Fig. 4C, top left and right).
The functional chimeric chain, 1C2, was as efficiently eluted
by MgATP as was the wild type TAP1 (Fig. 4C, bottom
left), but the nonfunctional chimera, 2C1, could not be
detectably eluted even at the highest MgATP concentration
(Fig. 4C, bottom right). Thus, in contrast to the chimeric
NBD of 1C2 and the wild-type domains, the chimeric NBD
of 2C1 appears to have lost the ability to exchange ADP for
ATP. The specificity of ADP-binding was tested by the
addition of free MgADP. All TAP chains showed a 40–50%
release at 1 m
M
MgADP (data not shown).
TAP2 exerts allosteric control over the nucleotide
binding of TAP1
Current working models propose that ATP binding and
hydrolysis in the TAP-NBDs alternate in a cooperative
fashion [10,12,15]. From Ôvanadate trappingÕ experiments it
has been speculated that during the transport cycle, ATP

binding and hydrolysis in NBD2 are involved in the
regulation of ATP binding by NBD1 [12]. The experimental
procedure used, however, does not directly demonstrate
ATP binding by TAP2 and does not distinguish between
vanadate-trapped TAP molecules that were generated in the
presence and absence of preceding ATP metabolism [29,30].
Our construction of an ATP binding chimeric variant of
NBD2 (Figs 2 and 3) could provide a direct experimental
strategy to investigate whether TAP2 exerts allosteric
control over the nucleotide binding of TAP1. We therefore
established an appropriate TAP variant in T2 cells [T2(1–
2V1)] with wild-type TAP1 and the ATP binding chimera,
2V1 (Fig. 1B). For comparison we also set up the reciprocal
T2 transfectant [T2(2–1V2)], with wild-type TAP2 and the
chimera 1V2, which contains the chimeric NBD1 (Fig. 1B).
T2(1–2V1) cells appear to express lower levels of TAP than
T2(2–1V2) and T2(TAPwt) but show the same balanced
expression (Fig. 5A, left) and assembly (Fig. 5A, right
panel) of both TAP subunits. Photo-cross-linking of
8-azido-ATP was performed on membrane preparations
from both these cell lines and assessed for labelling of TAP
polypeptides as before (Fig. 5B). For variant 1–2V1 we
found a clear ATP cross-link corresponding to the chimera
2V1 but, in contrast to wild-type transporter, essentially no
ATP cross-link to TAP1 (Fig. 5B). ATP binding thus
appears to be interchanged between the two subunits when
compared to the wild-type transporter. Thus, binding of
ATP to the chimeric TAP2 chain seems to interfere with
ATP binding to TAP1, presumably via a conformational
interaction between the two NBDs. No detectable ATP

cross-linking was observed for variant 2–1V2, suggesting
that binding of ADP by variant 1V2 does not shift the
nucleotide binding behaviour of wild-type TAP2 from ADP
to ATP.
We compared the transport activity of TAP variant
1–2V1 and 2–1V2 with that of wild-type TAP (Fig. 5C, left).
As expected, T2(2–1V2) was transport-inactive, however,
peptide translocation was clearly detectable in the T2(1–2V1)
cell line, consistent with the elevated HLA-B5 surface
expression data . The reduced level of peptide translocation
by T2(1–2V1) cells (15–20% of wild-type TAP) may be at
least partially attributed to the reduced TAP expression
noted above (Fig. 5A, left) though there is probably a
residual functional deficit as well. Nevertheless, the finding
that 1–2V1 forms a functional transporter strongly suggests
that indeed the chimeric NBD2 in chimera 2V1 adopts a
conformation reflecting a functional ATP binding state.
The C-terminal tail of the NBD2 is essential for ATP
binding and the catalytic function of TAP
Our results have shown that the C-terminal segment is
directly involved in the functional regulation of nucleotide
binding in rat TAP2. However, Yan et al. have described a
human TAP2 splice variant, named TAP2iso, which lacks
essentially the entire C-terminal tail encoded by the exon 11
but nevertheless forms an active transporter in conjunction
with TAP1 [31]. We therefore asked whether the core NBD
of rat TAP2 might be able to bind nucleotide and retain
catalytic function without the C-terminal tail, by creating a
truncated TAP2 variant (2DV) lacking the C-terminal 64
amino acids. The variant 2DV can be expressed in T2 cells,

showing an apparent molecular weight of about 55 kDa on
SDS gels (Fig. 6A, left), and has similar, ADP-restricted
nucleotide binding activity to that of wild-type TAP2 in a
nucleotide-agarose binding assay (Fig. 6A, right). In con-
trast to wild-type TAP2, however, ADP-agarose bound
2DV could not be detectably released with free MgATP
(Fig. 6B, compare left and right panels) and thus lacks the
normal ability of wild-type TAP2 to allow nucleotide
exchange. Wild-type TAP2 and 2 DV showed a half-
maximal elution from ADP-agarose with 1 m
M
free
MgADP (data not shown). We expressed the truncated
TAP2 chain together with wild-type TAP1 (TAP variant
1–2DV) (Fig. 7A) and tested the subunit assembly (Fig. 7B)
and the activity of this transporter variant in peptide
transport (Fig. 7C). Removal of the C-terminal tail in TAP2
had apparently no influence on the assembly of the two
TAP chains (Fig. 7B) but abolished the translocation of
radiolabelled peptides completely (Fig. 7C). Thus, in the
absence of the C-terminal segment, the core NBD2 alone,
while retaining the competence to bind ADP, cannot
exchange ADP for ATP or support the transport function
of TAP.
Discussion
Previous studies have shown that the two NBDs of ABC-
transporter TAP are not equivalent either in terms of
nucleotide binding or function [10,12,13,32]. The core
NBDs of TAP1 and TAP2, containing the ATP binding-
cassettes with the essential Walker A and B motifs, while not

identical are highly conserved whereas the C-terminal
segments of both TAP subunits, essentially encoded by
exon 11 (Fig. 1A), have low sequence homology to each
other. To investigate whether the distinctive nucleotide
binding behaviour of TAP1 and TAP2 can be attributed to
the sequence differences between the C-terminal tails, or
between the core NBDs, we created chimeric TAP chains by
exchanging one or other of these segments between TAP1
Ó FEBS 2003 Regulation of the nucleotide binding state of TAP (Eur. J. Biochem. 270) 4537
and TAP2 (Fig. 1). We were able to show that in the resting
state the distinctive nucleotide binding behaviours of TAP1
and TAP2 depend directly on the divergent C-terminal tails
(Fig. 2). A chimeric NBD2 with the C-terminal segment of
TAP1 adopts the ATP binding behaviour of wild-type
NBD1 whereas a corresponding chimeric NBD1 shows the
characteristic ADP binding properties of wild-type NBD2
(Fig. 2). Further, the chimeric NBD2 acquires not only the
nucleotide binding behaviour but also the functional
properties of NBD1 (Figs 3 and 5) and can participate in
a functional transporter. This result shows for the first time
that the core NBD of TAP2, normally seen only as an ADP-
binding structure, has indeed a potentially catalytically
active ATP binding-cassette, which must normally be tightly
4538 H. Bouabe and M. R. Knittler (Eur. J. Biochem. 270) Ó FEBS 2003
controlled by the C-terminal tail. Moreover, the function-
ality of variant 2–1C2 (Fig. 3), shows that the two core
NBD2s can form functional interfaces similar to those in
wild-type TAP.
The sequence differences within the ATP binding-cas-
settes of TAP1 and TAP2 make apparently no contribution

to the functional asymmetry of the NBDs, contrary to
previous proposals [12,18]. In the RAD50 homodimer, the
signature motifs are adjacent to the opposing Walker A sites
and the serines of each signature motif form hydrogen
bonds with the c-phosphate of the ATP bound by the
opposing subunit [33]. This kind of molecular bridging is
thought to be generally important to promote subunit
assembly and ATP hydrolysis of NBDs in ABC-transport-
ers. Based on these and other studies Karttunen et al.[12]
proposed that the serine in the canonical signature motif of
TAP1 (LSGGQ) (Fig. 1A) forms a hydrogen bond with the
c-phosphate of ATP bound to TAP2 and is involved in the
stimulation of ATP hydrolysis during the transport cycle.
This residue is an alanine (LAVGQ in rat and LAAGQ in
human) in TAP2 which would disallow such a hydrogen
bonding interaction, thus contributing to the functional
asymmetry of the two chains. Our demonstration that the
2–1C2 transporter is functional, despite having alanine in
the signature motif on both chains, excludes this model. In
line with this, it was shown for the sulfonylurea receptor
(SUR) that the serines in the signature motif are not
required for ATPase activity but seem to be involved in
transducing structural information between the ABC-
transporter domains [34]. The somewhat lowered transport
activity of TAP variant 2–1C2 (Fig. 3) could be due to
alterations in the signature motif-dependent cooperation
between the TAP domains. Recent experiments on mutated
TAP chains in which the serine and the second glycine of the
TAP1 signature motif and the glycine of the TAP2 signature
motif were exchanged by alanine showed that the signature

motifs are required for peptide translocation but not peptide
binding [35]. From our results in Fig. 3 it is suggestive that
the substitution of the glycines rather than substitution of
the serine caused the observed defect in peptide transport.
Results reported for SUR [6,36,37], appear to be highly
relevant to the TAP case. The two splice variants, 2A and
2B, of this tandem ABC-protein differ in their sequence only
for the 42 amino acid-long C-terminal tails of NBD2, which
are encoded by different exons. SUR2B has a much higher
nucleotide binding activity than SUR2A and it is suggested
that this functional difference arises from an interaction
between the C-terminal tails and their respective NBDs [38].
A sequence of about seven amino acids in the b
11
-strand of
the C-terminal tail of NBD2 is necessary and sufficient to
confer the different nucleotide binding and functional
properties of SUR2A and 2B [39]. As the sequence of the
TAP-NBD2 is homologous to the C-terminal NBD of
SUR, regulation of nucleotide binding in NBD2 of TAP
may be based on a similar mechanism. Following the
proposed working model for SUR [39] polar and charged
residues in the middle portion of the b
11
region (Fig. 1A)
may be also critical to control the distinct nucleotide binding
of the NBDs in TAP1 and TAP2. As the residues of the b
11
region with charge differences in the two TAP-NBDs are
located within a distance that allows interaction with the

Walker A motifs over a short distance [18], one possibility
might be that the conformation of the phosphate binding
loop in NBD1 and NBD2 is differently affected by
electrostatic interactions and thereby regulates the distinct
nucleotide binding in the TAP molecule. Thus, it will be of
interest to find out whether sequence differences in the b
11
-
strands (Fig. 1A) of TAP1 and TAP2 are directly involved
in the distinct nucleotide binding and function of the NBDs
of TAP.
It might be also possible that the C-terminal tails control
the ATP binding to the nucleotide binding pocket by other
sequence elements than the b
11
region. Crystal structure
analysis of the human TAP-NBD1 shows that the end part
of the a
6
region points structurally into the nucleotide
binding pocket and is close to the sequences of the Walker A
and B motif [18]. Most interestingly, the a
6
regions in the
NBDs of TAP1 and TAP2 are characterized by differences
in sequence and also in length (Fig. 1A). Thus, the a
6
region
could function as conformational regulator of the
C-terminal tail in controlling the access of ATP to the

nucleotide binding pocket during the peptide transport
cycle. Indeed, our own studies suggest that the a
6
region is
important for the distinct nucleotide binding and function
of the two TAP-NBDs (S. Ehses and M. R. Knittler,
unpublished results) and might have a steric effect for
arranging the critical sequence elements in the nucleotide
binding pocket of the NBDs in TAP1 and TAP2.
The switch region (Fig. 1A) is another defined sequence
element in the C-terminal segment, postulated to sense
c-phosphate binding [40], which could contribute to func-
tional asymmetry [18]. TAP2 contains the consensus
sequence of the switch region whereas TAP1 has a
glutamine in place of the conserved histidine found in most
Fig. 4. Nucleotide- and peptide-binding properties of chimeric TAP
variants. (A) Biochemical characteristics of different TAP variants. To
assess ATP binding capacity of wild-type and chimeric transporters
(top panel) membrane fractions of T2 transfectants were incubated
with 2 l
M
radiolabelled 8-azido-ATP for 5 min at 4 °C. After UV
cross-linking and lysis in 1% Triton X-100, TAP variants were
immunoprecipitated with either anti-TAP1 (D90) or anti-rat TAP2
(116/5) serum and separated on an SDS gel. The peptide binding
activity of the TAP variants (middle panel) was analyzed by substrate
cross-linking. Microsomal fractions were resuspended in binding buf-
fer and incubated with 1 l
M
iodinated and HSAB-conjugated peptide

S8. After cross-linking, cells were lysed and TAP was immunoisolated
with anti-TAP2 or anti-TAP1 serum. Migration behaviour and
amount of TAP chains was controlled by Western blots of the cor-
responding lysates (bottom panel) probed with a mixture of anti-TAP1
and anti-TAP2 serum. TAP variants are indicated by pictograms.
(B) The results of the ATP cross-link experiment were quantified by
phosphoimager. Peak integrals of TAP1- and TAP2-ATP complexes
were plotted in arbitrary units. (C) ADP to ATP exchange in wild-type
and chimeric TAP chains. Membrane fractions of T2 transfectants
expressing single wild-type (TAP1 or TAP2, top) or chimeric TAP
chains (1C2 or 2C1, bottom) were resuspended in lysis buffer and
incubated with ADP-agarose. Bound TAP chains were eluted with
increasing concentrations (0–1.0 m
M
) of MgATP. The nucleotide
matrix and the eluted fractions were analyzed in Western blots probed
for TAP1 (D90) and TAP2 (116/5). Enhanced chemiluminescence
fluorographs were quantified by densitometric scanning and the
obtained peak integrals were plotted in arbitrary units.
Ó FEBS 2003 Regulation of the nucleotide binding state of TAP (Eur. J. Biochem. 270) 4539
vertebrates. However, our own experiments suggest that
sequence differences in this region are not responsible for
the distinctive nucleotide binding of the TAP subunits
(H. Bouabe and M. R. Knittler, unpublished finding).
Functional importance of the C-terminal tail of NBDs
has been discussed for several ABC-transporters [41–44].
Our experiments on the truncated TAP2 variant 2DV
demonstrate that, although the core NBD2 retains the
ability to bind to ADP, the C-terminal tail of NBD2 is
indispensable for the active transport function of TAP

(Figs 6 and 7). In P-glycoprotein, only small C-terminal
deletions of up to 23 amino acids leave a functional
transporter [41], while truncation of the C-terminal 50–60
amino acids in the cystic fibrosis transmembrane conduct-
ance regulator (CFTR) severely impairs the ability of the
NBD2 to bind and/or hydrolyse ATP [45]; similarly, that
the impairment of variant 2DV may primarily involve
regulation of ATP binding. We also found a decreased
Fig. 5. Allosteric cross-talk between core NBDs in the TAP complex. (A) Expression levels of transporter variants 1–2V1 and 2–1V2 (left). T2
transfectants were lysed in buffer containing 1% Triton X-100. Lysates were analyzed in Western blots probed with antiserum D90 (C-term.
NBD1), antiserum 116/5 (C-term. NBD2) and antibody MAC 394 (core NBD2). TAP variants are indicated by pictograms. Subunit assembly of
the TAP variants 1–2V1 and 2–1V2 (right panel). Transfected T2 cells were lysed in 1% Triton X-100. TAP complexes were immunoprecipitated
with antibody MAC 394 and analyzed by Western blots probed for the C-terminal tail of TAP1 with antiserum D90 (TAPwt and 1–2V1) or the C-
terminal tail of TAP2 with antiserum 116/5 (2–1V2). (B) Nucleotide binding properties of wild-type TAP1 and TAP2 when expressed in a
combination with chimeric TAP chains. Membrane fractions of T2 cells expressing wild-type or chimeric transporters were lysed in 50 m
M
Tris/HCl
pH 7.5, 150 m
M
NaCl, 3 m
M
MgCl
2
, 1% Triton X-100 containing 2 l
M
of radiolabelled 8-azido-ATP. After UV cross-linking, TAP variants were
immunoprecipitated with an anti-TAP1 serum and separated by SDS/PAGE. (C) Peptide transport activity by TAP variants 1–2V1 and 2–1V2.
Streptolysin O -permeabilized T2 cells expressing wild-type TAP or chimeric transporter variants were incubated with iodinated reporter peptide S8
at 37 °C for the times indicated, the reactions were stopped by adding cold lysis buffer containing 1% Triton X-100 and peptides were quantitated
by gamma counting (left panel). Peptide supply by wild-type TAP and chimeric transporter variants to HLA-B5 molecules was analyzed by FACS

analysis (right panel) as described in Fig. 3C.
4540 H. Bouabe and M. R. Knittler (Eur. J. Biochem. 270) Ó FEBS 2003
steady state expression level for the TAP2 variant 2DV
compared with wild-type TAP2 chains (Figs 6 and 7). We
were unable to generate transfectants expressing a corres-
ponding C-terminal truncated TAP1 variant 1DV presum-
ably because of extreme instability. Recently, a conserved
hydrophobic patch in the C-terminal tail of NBD2 of
CFTR has been shown to be important for protein stability
[46]. The hydrophobic patch builds the central portion of
the b
11
-strand, which is also thought to be involved in the
regulation of nucleotide binding in SUR [39].
As mentioned above, a recently described human TAP2
splice variant, TAP2iso, lacks the entire sequence for the
C-terminal tail encoded by exon 11 [31]. Instead it contains a
newly identified exon 12 coding for an alternative sequence
of only nine residues (KTLWKFMIF, in the single amino
acid letter code). The translation product of TAP2iso
terminates 14 amino acid residues downstream of the
conserved
D
-loop. The splice variant does not contain the
switch region and the b
11
-strand (Fig. 1A) and thus seems to
have the shortest C-terminal region within the ABC-
transporter family. Interestingly, the TAP2iso product was
reported to be capable of peptide transport, although with

somewhat altered peptide selectivity [31]. As the polypeptide
chain of TAP2DV ends with the
D
-loop region (Fig. 1A),
the differences in transport activity of the TAP2iso product
and TAP2DV might be due to the different length of the
C-terminal truncations and the ability of the exon
12-encoded C-terminal tail to regulate the function of
NBD2. Nevertheless, our findings and the phenotypes of
other C-terminal truncated ABC-transporters [41–43,46]
Fig. 6. Biochemical characteristics of a C-terminal truncated TAP2 chain. (A) Steady state expression of TAP chimera 2DV (left panel). T2
transfectants expressing single TAP2 or TAP2DV were lysed in buffer containing 1% Triton X-100. Lysates were separated by SDS/PAGE and
blotted onto nitrocellulose. Western blots were probed with antibody MAC 394. Nucleotide binding capacity of TAP2DV(rightpanel).Triton
X-100 lysed membrane fractions expressing TAP2 or TAP2DV were incubated with different nucleotide agaroses. Bound proteins were eluted with
SDS sample buffer and analyzed in Western blots probed with MAC 394. (B) ADP to ATP exchange in wild-type and truncated TAP2 chains.
Membrane fractions of T2 transfectants expressing single wild-type (left panel) or C-terminal truncated TAP2 chains (right panel) were resuspended
in lysis buffer and incubated with ADP-agarose. Bound TAP chains were eluted with increasing concentrations (0–1.0 m
M
)offreeMgATP.The
nucleotide matrix and the eluted fractions were analyzed by Western blots probed for rat TAP2 with MAC 394. Enhanced chemiluminescence
fluorographs were quantified as described in Fig. 4C.
Ó FEBS 2003 Regulation of the nucleotide binding state of TAP (Eur. J. Biochem. 270) 4541
suggest that the basis for the in vivo stability and functional
properties of the TAP2iso product should be investi-
gated further. However, it is reasonable to assume that
residues at the extreme C-terminus of the NBDs of TAP2
are not directly involved in the regulation of nucleotide
binding. In the natural human TAP2 allele, TAP2A,a
premature termination signal creates a polypeptide lacking
the C-terminal 17 amino acids of the full-length sequence

[47,48]. In accordance with findings on other ABC-trans-
porters [41,46] this allelic variation at the C-terminus of
TAP2 has apparently no effect on the biochemical and
functional properties of TAP2 [49].
Replacement of the C-terminal tail in wild-type NBD1
with that of wild-type NBD2 leads to a chimeric NBD1,
which is apparently locked in an ADP binding conforma-
tion that blocks peptide translocation of 1–2C1 and 2–1V2
(Figs 4 and 5). These findings and our results on the
truncated TAP2 chain (Fig. 6B) provide evidence that the
correct interplay between the core NBD and C-terminal tail
is absolutely necessary for the functional nucleotide
exchange reaction during the active peptide transport cycle.
As ADP-bound wild-type and chimeric TAP chains showed
in our control experiments similar elution profiles with
increasing concentration of free MgADP (E
50
¼ 1m
M
)(see
Fig. 7. Peptide translocation by TAP requires the C-terminal tail of NBD2. (A) Cellular expression of TAP variant 1–2DV. T2 transfectants
expressing wild-type TAP1 in combination with TAP2DV were lysed with 1% Triton X-100. Cell extracts were separated by SDS/PAGE and
blotted onto nitrocellulose membranes. Western blots were probed with antiserum D90 (C-terminal NBD1), antiserum 116/5 (C-terminal NBD2)
and antibody MAC 394 (core NBD2). (B) Subunit assembly of the TAP variant 1–2DV. Transfected T2 cells were lysed in 1% Triton X-100. TAP
complexes were immunoprecipitated with antibody MAC 394 and analyzed in Western blots probed for the C-terminal tail of TAP1 with antiserum
D90 (top panel) or the core NBD of TAP2 with antibody MAC 394 (bottom panel). (C) Peptide transport activity of mutant and wild-type TAP
variants. Streptolysin O-permeabilized T2 cells expressing wild-type TAP or variant 1–2DV were incubated with iodinated peptide S8 at 37 °Cfor
10 min, and the reaction was stopped by adding cold lysis buffer containing 1% Triton X-100 and quantitated by gamma counting.
Fig. 8. Working model of the peptide translocation by wild-type TAP and chimeric transporters. (A) Based on our results of the different chimeric
TAP variants, we suggest a revised scheme for the transport cycle of wild-type TAP in which the C-terminal tails of NBD1 and NBD2 regulate the

access of nucleotides during the peptide translocation cycle. In the resting state of TAP (Fig. 8A, top), which is characterized by high substrate
affinity, the C-terminal tail of TAP2 induces a conformation in NBD2 that is nonpermissive for ATP binding. The ADP-binding state of TAP2
allows binding of ATP to TAP1. We hypothesize that peptide binding to TAP leads to transient change in the conformation of NBD2 that is
transduced to the C-terminal tail and results in an ÔATP-onÕ state of TAP2. The structural alteration in TAP2 affects ATP binding of TAP1 via
allosteric cross-talk of the core NBDs and induces ATP hydrolysis in NBD1. This step in the catalytic cycle energizes and initiates peptide
translocation and might be controlled also by the C-terminal tail of NBD1. Finally, the cycle is completed and reinitiated (Fig. 8A, bottom) when
ATP hydrolysis in NBD2 allows the recharging of NBD1 with ATP and resets the transporter for a conformation with high substrate affinity. (B)
Proposed working model for the transport cycle of the active TAP variant 2–1C2. The initiation step and the progress of the transport cycle are
similar to the hypothesized peptide translocation model of the wild-type transporter. (C) Cyclic transport process of the functional chimeric
transporter 1–2V1. In the hypothesized starting complex of variant 1–2V1 the chimeric TAP2 chain 2V1 is in an ATP binding state. This ATP
binding by the chimeric TAP2 chain seems critical as it blocks allosterically the simultaneous binding of ATP to TAP1. This resting state of variant
1–2V1 might be related to the postulated transient state of wild-type TAP, which in our view initiates the recovery phase of the transport cycle
(Fig. 8A). Thus, the first hydrolytic event and the nucleotide exchange reaction are required to bring the chimeric transporter variant 1–2V1 into a
transient state in the transport cycle that normally initiates peptide translocation upon substrate binding. Blocked peptide translocation in the TAP
variants 1–2C1 (D) and 2–1V2 (E). From our data we propose that the defect in the ADP to ATP exchange in the chimeric NBD1 of 1–2C1 and 2–
1V2 prevents initiation and progress of the peptide transport cycle in both chimeric TAP variants. A similar defect might also block the function of
the variant TAP1–2DV that contains a C-terminal truncated TAP2 chain. It is likely that the absence of detectable ATP binding in the starting
complex of TAP variant 2–1V2 (E) is also responsible for the defect in peptide transport.
4542 H. Bouabe and M. R. Knittler (Eur. J. Biochem. 270) Ó FEBS 2003
above) the observed differences between wild-type TAP2
and chimera 2C1 in the nucleotide exchange reaction cannot
be explained by an altered ADP binding affinity. In our
view, it seems more likely that the chimeric NBD1 in the
2C1 and 1V2 is in a so called Ôoccluded stateÕ [50] that
retained the normal ADP binding capacity but lost the
conformational properties allowing the access of free
MgATP to the nucleotide binding pocket and accordingly
Ó FEBS 2003 Regulation of the nucleotide binding state of TAP (Eur. J. Biochem. 270) 4543
the nucleotide exchange during the peptide translocation
process.

The nonreciprocal behaviour of the chimeric TAP-NBDs
suggests that the C-terminal regulation of nucleotide
interaction in NBD1 is more complex than in NBD2 and
that there are different requirements for the functional
regulation of the ATP binding-cassettes. Several groups
have postulated that ATP binding and hydrolysis in TAP
proceeds in an alternating way and that the catalytic activity
of the two NBDs is linked cooperatively [9,10,12–14,35].
The functional change of the chimeric NBD2, 2V1, towards
ATP-binding (Fig. 2) allowed us to ascertain directly
whether the nucleotide bound by TAP2 regulates the
nucleotide binding properties of TAP1 (Fig. 5). Indeed in
the transport active TAP variant 1–2V1, ATP binding by
TAP1 is drastically reduced, in accordance with recent
findings on the functional interplay of NBDs in P-
glycoprotein [51]. The resting state of variant 1–2V1 may
mimic the postulated transient state of wild-type TAP,
which in our view initiates the recovery phase of the
transport cycle [10]. The observed allosteric effect seems to
depend on the combination of the core NBD1 with core
NBD2, and not of the C-terminal tails, as previous
experiments on the related TAP variant 1–2N1, which
contains two identical NBD1s, showed equal ATP binding
to both TAP chains [9].
In contrast to variant 1–2V1, no allosteric effect on
nucleotide binding was found for the resting states of wild-
type TAP and variant 2–1V2 (Fig. 5). In both cases
nucleotide binding of the two TAP chains reflect intrinsic
properties of the individual subunits (Figs 2 and 5). Thus,
the conformational cross-talk between NBD1 and NBD2

must not work in a simple symmetrical manner. Whether
and to what extent sequence deviations in the NBD regions
forming the predicted dimer interface [52] are involved in the
nonreciprocal cooperation regulating the specificity of
nucleotide binding is the subject of our current studies on
TAP.
Our present findings are consistent with a model in which
only one of the two TAP-NBDs binds and utilizes ATP at
any given time. We propose a revised scheme for the
transport activity of TAP [10] in which the C-terminal tails
and conformational cross-talk between the NBDs control
the peptide translocation cycle (Fig. 8A). In one half of the
cycle, the C-terminal tail of TAP2 induces a conformation
in NBD2 that is nonpermissive for ATP binding. The
resulting ADP-binding state of TAP2 and the C-terminal
tail of NBD1 allow binding of ATP to TAP1. Peptide
binding leads to a change in the conformation of the TMD,
which is transmitted via the core NBD2 to the C-terminal
tail and results in an ÔATP-onÕ state of NBD2. As ATP
binding in TAP2 affects the binding of ATP by NBD1, the
allosteric cross-talk between core NBD2 and core NBD1
causes ATP hydrolysis in TAP1 and initiates the peptide
transport cycle. After peptide translocation, the cycle is
completed when ATP hydrolysis in NBD2 allows the
recharging of NBD1 with ATP and resets the transporter
for a conformation with high peptide affinity. There are
different views concerning the functional role of ATP
hydrolysis in TAP1 and the Ôstarting pointÕ of the transport
cycle [10,15,53]. Nevertheless, in line with our proposed
working model, recent findings provide evidence that ATP

hydrolysis by TAP1 is indeed necessary for active peptide
translocation [35]. Our studies showed that the chimera
1C2 can replace mainly the function of wild-type TAP1 in
the transporter complex (Fig. 3). Therefore, we propose the
same kind of starting complex and catalytic events in the
transport cycle of the transport active TAP variant 2–1C2
(Fig. 8B). In the case of the functional TAP variant 1–2V1
the results of our nucleotide binding studies suggest that in
the ground state ATP binding by the chimera 2V1 blocks
simultaneous binding of ATP by TAP1 (Fig. 5). On the
basis of our working model, it is tempting to speculate that
the first hydrolytic event and the nucleotide exchange
reaction in this TAP variant is required to bring the
transporter into a state in the transport cycle that normally
initiates peptide translocation upon substrate binding
(Fig. 8C). For the two nonfunctional TAP variants
1–2C1 and 2–1V2 (Fig. 8D,E) the peptide transport is
blocked as one of the two subunits contains a chimeric
NBD1 with the C-terminal tail of TAP2, which seems to be
locked in an ADP binding state (Fig. 4). A similar defect
could also block the transport function of the variant
TAP1–2DV (Figs 6 and 7) that contains a C-terminal
truncated TAP2 chain. The defective ADP to ATP
exchange in the chimeric NBD1 (Fig. 8D,E) and the
truncated NBD2 prevents initiation and progress of the
peptide transport cycle. The malfunction of variant 2–1V2
might be also due to the fact that the resting state, which in
our view represents the starting complex (Fig. 8E), does not
show detectable binding of ATP (Fig. 5).
Structural analysis of TAP2, the heterodimeric TAP

complex or the paired NBDs as well as further muta-
tional analysis on the TAP chains are required to
provide detailed insight into the molecular mechanisms
underlying C-terminal regulation and the conformational
interplay of the two NBDs. Further comparative analysis
is required to identify functional principles in common
between ABC-transporters, just as the recent findings on
the C-terminal variation in the tandem ABC-protein,
SUR, have provided a model to understand our results
on TAP.
Acknowledgements
We thank Professor T. Langer for stimulating discussions and
Professor J. C. Howard for the help in preparing the final text of this
manuscript. We thank Dr L. Guethlein, Dr R. Uthaiah, J. Dramiga
and Dr I. Arnold for critical reading the manuscript. The Deutsche
Forschungsgemeinschaft (DFG-Grant: KN541/1–1) and the Land
Nordrhein-Westfalen through the University of Cologne supported our
work.
References
1. Deverson, E.V., Gow, I.R., Coadwell, W.J., Monaco, J.J., But-
cher, G.W. & Howard, J.C. (1990) MHC class II region encoding
proteins related to the multidrug resistance family of transmem-
brane transporters. Nature 348, 738–741.
2. Higgins, C.F. (1992) ABC transporters: from microorganisms to
man. Ann. Rev. Cell. Biol. 8, 67–113.
3. Carson, M.R., Travis, S.M. & Welsh, M.J. (1995) The two
nucleotide-binding domains of cystic fibrosis transmembrane
conductance regulator (CFTR) have distinct functions in con-
trolling channel activity. J. Biol. Chem. 270, 1711–1717.
4544 H. Bouabe and M. R. Knittler (Eur. J. Biochem. 270) Ó FEBS 2003

4. Hrycyna, C.A., Ramachandra, M., Germann, U.A., Cheng, W.C.,
Pastan, I. & Gottesmann, M.M. (1999) Both ATP sites of human
P-glycoprotein are essential but not symmetric. Biochemistry 38,
13887–13899.
5. Kreimer, D.I., Chai, K.P. & Ferro-Luzzi Ames, G. (2000)
Nonequivalence of the nucleotide-binding subunits of an ABC
transporter, the histidine permease, and conformational changes
inthemembranecomplex.Biochemistry 39, 14183–14195.
6. Matsuo, M., Tanabe, K., Kioka, N., Amachi, T. & Ueda, K.
(2000) Different binding properties and affinities for ATP and
ADP among sulfonylurea receptor subtypes, SUR1, SUR2A, and
SUR2B. J. Biol. Chem. 275, 28757–28763.
7. Proff, C. & Kolling, R. (2001) Functional asymmetry of the two
nucleotide binding domains in the ABC transporter Ste6. Mol.
Gen. Genet. 264, 883–893.
8. Aleksandrov, A.A., Aleksandrov, L. & Riordan, J.R. (2002)
Nucleoside triphosphate pentose ring impact on CFTR gating and
hydrolysis. FEBS Lett. 518, 183–188.
9. Daumke, O. & Knittler, M.R. (2001) Functional asymmetry of the
ATP binding-cassettes of the ABC transporter TAP is determined
by intrinsic properties of the nucleotide binding domains. Eur. J.
Biochem. 268, 4776–4786.
10. Alberts, P., Daumke, O., Deverson, E.D., Howard, J.C. & Knit-
tler, K.R. (2001) Distinct functional properties of the TAP sub-
units coordinate the nucleotide dependent transport cycle. Curr.
Biol. 11, 242–251.
11. Hewitt, E.W., Gupta, S.S. & Lehner, P.J. (2001) The human
cytomegalovirus gene product US6 inhibits ATP binding by TAP.
EMBO J. 20, 387–396.
12. Karttunen, J.T., Lehner, P.J., Gupta, S.S., Hewitt, E.W. &

Cresswell, P. (2001) Distinct functions and cooperative interaction
of the subunits of the transporter associated with antigen proces-
sing (TAP). Proc. Natl Acad. Sci. USA 98, 7431–7436.
13. Saveanu, L., Daniel, S. & van Endert, P.M. (2001) Distinct
functions of the ATP binding cassettes of transporters associated
with antigen processing: a mutational analysis of Walker A and B
sequences. J. Biol. Chem. 276, 22107–22113.
14. Lapinski, P.E., Raghuraman, G. & Raghavan, M. (2003)
Nucleotide interactions with membrane-bound transporter asso-
ciated with antigen processing (TAP) proteins. J. Biol. Chem. 278,
8229–8237.
15. van Endert, P., Saveanu, L., Hewitt, E. & Lehner, P. (2002)
Powering the peptide pump: TAP crosstalk with energetic
nucleotides. Trends Biochem. Sci. 27, 454–461.
16. Gorbulev, S., Abele, R. & Tampe
´
, R. (2001) Allosteric crosstalk
between peptide-binding, transport, and ATP hydrolysis of the
ABC transporter TAP. Proc. Natl Acad. Sci. USA 98, 3732–3737.
17. Schneider, E. & Hunke, S. (1998) ATP binding-cassette (ABC)
transport systems: functional and structural aspects of the ATP-
hydrolyzing subunits/domains. FEMS Microbiol. Rev. 22, 1–20.
18. Gaudet, R. & Wiley, D.C. (2001) Structure of the ABC ATPase
domain of human TAP1, the transporter associated with antigen
processing. EMBO J. 20, 4964–4972.
19. DeMars, R., Chang, C.C., Shaw, S., Reitnauer, P.J. & Sondel,
P.M. (1984) Homozygous deletions that simultaneously eliminate
expressionsofclassIandclassIIantigensofEBV-tansformed
B-lymphoblastoid cells. I. Reduced proliferance responses of
autologous and allogeneic T cells that have decreased expression

ofclassIIantigens.Hum. Immunol. 11, 77–97.
20. Momburg, F., Ortiz-Navarrete, V., Neefjes, J., Goulmy, E., van de
Wal, Y., Spits, H., Powis, S.J., Butcher, G.W., Howard, J.C.,
Walden, P. et al. (1992) Proteasome subunits encoded by the
major histocompatibility complex are not essential for antigen
presentation. Nature 360, 174–177.
21. Powis, S.J., Townsend, A.R., Deverson, E.V., Bastin, J., Butcher,
G.W. & Howard, J.C. (1991) Restoration of antigen presentation
to the mutant cell line RMA-S by an MHC-linked transporter.
Nature 354, 528–531.
22. Gunning, P., Leavitt, J., Muscat, G., Ng, S.Y. & Kedes, L. (1987)
Ahumanb-actin expression vector system directs high-level
accumulation of antisence transcripts. Proc. Natl Acad. Sci. USA
84, 4831–4840.
23. Knittler, M.R., Gu
¨
low, K., Seelig, A. & Howard, J.C. (1998)
MHC class I molecules compete in the endoplasmic reticulum
for access to transporter associated with antigen processing.
J. Immunol. 161, 5967–5977.
24. Yang, S.Y., Morishima, Y., Collins, N.H., Alton, T., Pollak, M.S.,
Yunis, E.J. & Dupont, B. (1984) Comparison of one-dimensional
IEF patterns for serologically detectable HLA-A and -B allotypes.
Immunogenetics 19, 217–231.
25. Neefjes, J.J., Momburg, F. & Ha
¨
mmerling, G.J. (1993) Selective
and ATP-dependent translocation of peptides by the MHC-
encoded transporter. Science 261, 769–771.
26. Knittler, M.R., Alberts, P., Deverson, E.V. & Howard, J.C. (1999)

Nucleotide binding by TAP mediates association with peptide and
release of assembled MHC class I molecules. Curr. Biol. 9,
999–1008.
27. van Endert, P.M., Tampe
´
, R., Meyer, T.H., Tisch, R., Bach, J.F.
& McDevitt, H.O. (1994) A sequential model for peptide binding
and transport by the transporters associated with antigen pro-
cessing. Immunity 1, 491–500.
28. Velarde, G., Ford, R.C., Rosenberg, M.F. & Powis, S.J. (2001)
Three-dimensional structure of transporter associated with anti-
gen processing (TAP) obtained by single particle image analysis.
J. Biol. Chem. 276, 46054–46063.
29. Sauna, Z.E., Smith, M.M., Muller, M. & Ambudkar, S.V.
(2001) Functionally similar vanadate-induced 8-azidoadenosine
5¢-[a-
32
P]diphosphate-trapped transition state intermediates of
human P-glycoprotin are generated in the absence and presence of
ATP hydrolysis. J. Biol. Chem. 276, 21199–21208.
30. Chen, M., Abele, R. & Tampe
´
, R. (2003) Peptides induce ATP
hydrolysis at both subunits of the transporter associated with
antigen processing (TAP). J. Biol. Chem. 278, 29686–29692.
31. Yan, G., Shi, L. & Faustman, D. (1999) Novel splicing of the
human MHC-encoded peptide transporter confers unique prop-
erties. J. Immunol. 162, 852–859.
32. Lapinski, P.E., Neubig, R.R. & Raghavan, M. (2001) Walker A
lysine mutations of TAP1 and TAP2 interfere with peptide

translocation but not peptide binding. J. Biol. Chem. 276, 7526–
7533.
33. Hopfner, K.P., Karcher, A., Shin, D.S., Craig, L., Arthur, L.M.,
Carney, J.P. & Tainer, J.A. (2000) Structural biology of Rad50
ATPase: ATP-driven conformational control in DNA double-
strand break repair and the ABC-ATPase superfamily. Cell 101,
789–800.
34. Matsuo, M., Dabrowski, M., Ueda, K. & Ashcroft, F.M. (2002)
Mutations in the linker domain of NBD2 of SUR inhibit trans-
duction but not nucleotide binding. EMBO J. 21, 4250–4258.
35. Hewitt, E.W. & Lehner, P.J. (2003) The ABC-transporter sig-
nature motif is required for peptide translocation but not peptide
binding by TAP. Eur. J. Immunol. 33, 422–427.
36. Matsuoka, T., Matsushita, K., Katayama, Y., Fujita, A., Inageda,
K., Tanemoto, M., Inanobe, A., Yamashita, S., Matsuzawa, Y. &
Kurachi, Y. (2000) C-terminal tails of sulfonylurea receptors
control ADP-induced activation and diazoxide modulation of
ATP-sensitive K (+) channels. Circ. Res. 87, 873–880.
37. Reimann, F., Gribble, F.M. & Ashcroft, F.M. (2000) Differential
response of K
ATP
channels containing SUR2A or SUR2B sub-
units to nucleotides and pinacidil. Mol. Pharmacol. 58, 1318–1325.
38. Song, D.K. & Ashcroft, F.M. (2001) ATP modulation of ATP-
sensitive potassium channel ATP sensitivity varies with the type of
SUR subunit. J. Biol. Chem. 276, 7143–7149.
Ó FEBS 2003 Regulation of the nucleotide binding state of TAP (Eur. J. Biochem. 270) 4545
39. Matsushita, K., Kinoshita, K., Matsuoka, T., Fujita, A., Fujika-
do, T., Tano, Y., Nakamura, H. & Kurachi, Y. (2002) Intra-
molecular interaction of SUR2 subtypes for intracellular

ADP-induced differential control of K (ATP) channels. Circ. Res.
90, 554–561.
40. Hung, L W., Wang, I.X., Nikaido, K., Liu, P Q., Ferro-Luzzi
Ames, G. & Kim, S H. (1998) Crystal structure of the ATP
binding subunit of an ABC-transporter. Nature 396, 703–707.
41. Currier, S.J., Ueda, K., Willingham, M.C., Pastan, I. & Gottes-
man, M.M. (1989) Deletion and insertion mutants of the multi-
drug transporter. J. Biol. Chem. 264, 14376–14381.
42. Holzinger, A., Maier, E., Stockler-Ipsiroglu, S., Braun, A. &
Roscher, A.A. (1998) Characterization of a noval mutation in
exon 10 of the adrenoleukodystrophy gene. Clin. Genet. 53,
482–487.
43. Haardt, M., Benharouga, M., Lechardeur, D., Kartner, N. &
Lukacs, G.L. (1999) C-terminal truncations destabilize the cystic
fibrosis transmembrane conductance regulator without impairing
its biogenesis. A novel class of mutation. J. Biol. Chem. 274,
21873–21877.
44. Sharma, N., Crane, A., Clement, J.P.T., Gonzalez, G., Babenko,
A.P., Bryan, J. & Aguilar-Bryan, L. (1999) The C-terminus of
SUR1 is required for trafficking of K (ATP) channels. J. Biol.
Chem. 274, 20628–20632.
45. Gentzsch, M., Aleksandrov, A., Aleksandrov, L. & Riordan, J.R.
(2002) Functional analysis of the C-terminal boundary of the
second nucleotide binding domain of the cystic fibrosis trans-
membrane conductance regulator and structural implications.
Biochem. J. 366, 541–548.
46. Gentzsch, M. & Riordan, J.R. (2001) Localization of sequences
within the C-terminal domain of the cystic fibrosis transmembrane
conductance regulator which impact maturation and stability.
J. Biol. Chem. 276, 1291–1298.

47. Powis, S.H., Mockridge, I., Kelly, A., Kerr, L.A., Glynne, R.,
Gileadi, U., Beck, S. & Trowsdale, J. (1992) Polymorphism in a
second ABC transporter gene located within the class II region of
the human major histocompatibility complex. Proc. Natl Acad.
Sci. USA 89, 1463–1467.
48. Colonna, M., Bresnahan, M., Bahram, S., Strominger, J.L. &
Spies, T. (1992) Allelic variants of the human putative peptide
transporter involved in antigen processing. Proc. Natl Acad. Sci.
USA 89, 3932–3936.
49. Daniel, S., Caillat-Zucman, S., Hammer, J., Bach, J.F. & van
Endert, P.M. (1997) Absence of functional relevance of human
transporter associated with antigen processing polymorphism for
peptide selection. J. Immunol. 159, 2350–2357.
50. Kyritsis, C., Gorbulev, S., Hutschenreiter, S., Pawlitschko, K.,
Abele, R. & Tampe
´
, R. (2001) Molecular mechanism and struc-
tural aspects of the transporter associated with antigen processing
inhibition by the cytomegalovirus protein US6. J. Biol. Chem. 276,
48031–48039.
51. Sauna, Z.E., Smith, M.M., Muller, M., Kerr, K.M. & Ambu-
dkar, S.V. (2001) The mechanism of action of multidrug-
resistance-linked P-glycoprotein. J. Bioenerg. Biomembr. 33,
481–491.
52. Locher, K.P., Lee, A.T. & Rees, D.C. (2002) The E. coli BtuCD
structure: a framework for ABC transporter architecture and
mechanism. Science 296, 1091–1098.
53. Arora, S., Lapinski, P.E. & Raghavan, M. (2001) Use of chimeric
proteins to investigate the role of transporter associated with
antigen processing (TAP) structural domains in peptide binding

and translocation. Proc.NatlAcad.Sci.USA98, 7241–7246.
4546 H. Bouabe and M. R. Knittler (Eur. J. Biochem. 270) Ó FEBS 2003