Báo cáo khoa học: Crystal structure of the second PDZ domain of SAP97 in complex with a GluR-A C-terminal peptide ppt

Crystal structure of the second PDZ domain of SAP97
in complex with a GluR-A C-terminal peptide
Ingemar von Ossowski
1
, Esko Oksanen
2
, Lotta von Ossowski
1
, Chunlin Cai
1
, Maria Sundberg
1
,
Adrian Goldman
2,3
and Kari Keina
¨
nen
1
1 Department of Biological and Environmental Sciences (Division of Biochemistry), University of Helsinki, Finland
2 Institute of Biotechnology, University of Helsinki, Finland
3 Neuroscience Center, University of Helsinki, Finland
Selective insertion ⁄ removal of GluR-A (GluR1) sub-
unit, containing a-amino-5-methyl-3-hydroxy-4-isoxa-
zole propionic acid (AMPA) receptors to⁄ from the
postsynaptic membrane, is a key early event in some
experimental models of activity-dependent regulation
of synaptic strength [1–5]. Rapid insertion of GluR-A
to synaptic membrane is dependent on its cytoplasmic
C-terminal domain of 80 amino acid residues [6].
Keywords

crystal structure; GluR1; GluR-A; PDZ
domain; SAP97
Correspondence
K. Keina
¨
nen, Department of Biological and
Environmental Sciences (Division of
Biochemistry), Viikinkaari 5, P.O.Box56,
FI-00014 University of Helsinki, Helsinki,
Finland
Fax: +358 9191 59068
Tel: +358 9191 59606
E-mail: kari.keinanen@helsinki.fi
(Received 25 August 2006, revised 28
September 2006, accepted 2 October 2006)
doi:10.1111/j.1742-4658.2006.05521.x
Synaptic targeting of GluR-A subunit-containing glutamate receptors
involves an interaction with synapse-associated protein 97 (SAP97). The
C-terminus of GluR-A, which contains a class I PDZ ligand motif
(-x-Ser ⁄ Thr-x-/-COOH where / is an aliphatic amino acid) associates
preferentially with the second PDZ domain of SAP97 (SAP97
PDZ2
). To
understand the structural basis of this interaction, we have determined the
crystal structures of wild-type and a SAP97
PDZ2
variant in complex with
an 18-mer C-terminal peptide (residues 890–907) of GluR-A and of two
variant PDZ2 domains in unliganded state at 1.8–2.44 A
˚

resolutions.
SAP97
PDZ2
folds to a compact globular domain comprising six b-strands
and two a-helices, a typical architecture for PDZ domains. In the structure
of the peptide complex, only the last four C-terminal residues of the GluR-
A are visible, and align as an antiparallel b-strand in the binding groove of
SAP97
PDZ2
. The free carboxylate group and the aliphatic side chain of the
C-terminal leucine (Leu907), and the hydroxyl group of Thr905 of the
GluR-A peptide are engaged in essential class I PDZ interactions. Compar-
ison between the free and complexed structures reveals conformational
changes which take place upon peptide binding. The bA)bB loop moves
away from the C-terminal end of aB leading to a slight opening of the
binding groove, which may better accommodate the peptide ligand. The
two conformational states are stabilized by alternative hydrogen bond and
coulombic interactions of Lys324 in bA)bB loop with Asp396 or Thr394
in bB. Results of in vitro binding and immunoprecipitation experiments
using a PDZ motif-destroying L907A mutation as well as the insertion of
an extra alanine residue between the C-terminal Leu907 and the stop codon
are also consistent with a ‘classical’ type I PDZ interaction between SAP97
and GluR-A C-terminus.
Abbreviations
AMPA, a-amino-5-methyl-3-hydroxy-4-isoxazole propionic acid; GluR-A, ionotropic glutamate receptor subunit A; GST, glutathione
S-transferase; Maguk, membrane-associated guanylate kinase homolog; PDZ, postsynaptic density )95 ⁄ Discs large ⁄ zona occludens-1;
PSD-93, postsynaptic density )93; PSD-95, postsynaptic density )95; SAP97, synapse-associated protein 97; SAP102, synapse-associated
protein 102.
FEBS Journal 273 (2006) 5219–5229 ª 2006 The Authors Journal compilation ª 2006 FEBS 5219
The GluR-A C-terminus contains a class I PDZ ligand

motif which is necessary for the stimulated synaptic
incorporation of GluR-A subunit in an acute fashion
[7–9]. Interestingly, it was recently reported that trans-
genic mice expressing GluR-A variant lacking seven
C-terminal residues display apparently normal synaptic
plasticity and basal GluR-A localization [10], suggest-
ing developmental plasticity and ⁄ or existence of multi-
ple parallel pathways in GluR-A transport. To date,
four PDZ domain-containing proteins have been
reported to associate with GluR-A via its C-terminus:
SAP97 [11], mLin-10 [12], Shank3 [13], and RIL [14],
although the binding of RIL to GluR-A is not PDZ
domain-mediated. From these candidates, the multi-
domain scaffolding protein SAP97 emerges as a strong
candidate to subserve synaptic delivery of GluR-A
AMPA receptors. Overexpression of SAP97 drives
GluR-A to synapses and increases AMPA receptor
mediated synaptic currents [15,16], and concomitantly
occludes long-term potentiation [16]. In a converse
manner, RNAi block of SAP97 expression in cultured
neurons inhibits expression of surface GluR-A and
AMPA receptor mediated synaptic currents [16].
SAP97 may also play a role in the endocytosis of
GluR-A AMPA receptors, based on identification of a
ternary complex between SAP97, GluR-A and myosin
VI, a minus-end directed actin motor [17,18].
Detailed information on the molecular mechanism of
SAP97–GluR-A interaction would help us to under-
stand its physiological roles and regulation. In glu-
tathione S-transferase (GST) pulldown assay, GluR-A

C-terminus binds to the second PDZ domain of SAP97
(SAP97
PDZ2
). The binding is dependent on the PDZ
binding motif, but is also strongly affected by sequences
upstream of the C-terminus of GluR-A [19,20], includ-
ing formation of a disulfide-linked complex between
synthetic GluR-A C-terminal peptide and SAP97
PDZ2
[19]. In an attempt to understand the structural basis of
GluR-A–SAP97 interaction, we have determined the
crystal structure of SAP97
PDZ2
in the presence and
absence of a GluR-A C-terminal 18-mer peptide ligand.
The structure reveals an archetypical class I PDZ inter-
action that involves the last four residues of GluR-A
with no apparent contribution by other residues.
Results
Binding of GluR-A C-terminal domain to SAP97
in cultured cells
To complement and extend our earlier GST pulldown
analysis of GluR-A–SAP97 PDZ interaction, we first
examined the interaction of the C-terminal domain
(CTD; residues 827–907) of GluR-A with SAP97 under
cellular conditions. We created three different GFP-
tagged GluR-A CTDs: the wild-type with the C-term-
inal residues Ala904-Thr905-Gly906-Leu907 and two
point-mutated versions, a L907A substitution which
eliminates the class I PDZ binding motif, and ‘XA908’

in which an extra alanine residue was inserted between
the C-terminal Leu907 and the stop codon. These were
cotransfected with myc-tagged SAP97 in HEK293 cells.
All proteins were expressed at roughly equal levels as
indicated by the intensities of anti-GFP and anti-myc
immunoblots, but only the wild-type CTD associated
with myc-tagged SAP97 (Fig. 1A). This result is in
agreement with our earlier in vitro binding analysis [20],
which confirms the importance of the canonical class I
PDZ binding motif. As SAP97 is endogenously present
in HEK293 cells, we also analyzed anti-SAP97 immuno-
precipitates of cells transfected only with GFP-GluR-
A
CTD
expression plasmids. Consistent with the above
results, only the wild-type CTD associated with the
native SAP97 (Fig. 1B).
In vitro binding of GluR-A C-terminus to SAP97
PDZ domains
To further examine the canonical class I PDZ binding
motif of the GluR-A C-terminus, we used a microplate
A
Lysate
Blot: anti-GFP
37
25
Lysate
Blot: anti-myc
100
150

IP: anti-GFP
Blot: anti-myc
100
150
wt L907A XA908
myc-SAP97 + GFP-GluRA
CTD
kDa
B
wt L907A XA908
25
37
37
25
Lysate
IP: Anti-SAP97
N
GFP-GluRA
CTD
Blot: anti-GFP
kDa
Fig. 1. Coimmunoprecipitation of GluR-A CTD with SAP97. HEK293
cells were transfected for expression of wild-type (‘wt’) GFP-GluR-
A CTD, and mutated CTDs L907A and ‘XA908’ with (A) or without
(B) N-terminally myc-tagged SAP97. The cells were subjected to
immunoprecipitation and immunoblotting by using anti-myc or anti-
GFP as indicated. Molecular size markers are indicated on the left.
IP, immunoprecipitation.
SAP97 PDZ2–GluR-A peptide complex I. von Ossowski et al.
5220 FEBS Journal 273 (2006) 5219–5229 ª 2006 The Authors Journal compilation ª 2006 FEBS

binding assay to analyze the interaction between puri-
fied SAP97 PDZ proteins. Purified His-tagged PDZ1,
PDZ2, and PDZ3 domains of SAP97 were adsorbed on
microwells, followed by blocking of nonspecific protein
adsorption sites by an excess of BSA. The binding of
wild-type or mutated GluR-A C-terminal 11-mer pep-
tides (GST-A
11
) produced as GST fusions was deter-
mined by using anti-GST-horseradish peroxidase
conjugate as an enzymatic marker (Fig. 2A). Short
C-terminal peptides were used instead of ‘full-length’
CTDs because of the higher stability of the former in
bacterial expression. GST-A
11
bound consistently stron-
gest to the PDZ2 domain, whereas the L907A mutation
led to significantly decreased binding (Fig. 2B). More-
over, GST-A
11
showed somewhat weaker, but detect-
able PDZ motif-dependent binding to the first but not
to the third PDZ domain of SAP97 (Fig. 2B). PDZ
motif-independent binding of GST fusions was consis-
tently higher to PDZ2 coated wells than to PDZ1 and
PDZ3 wells. Control experiments indicate that both ele-
vated background adsorption of the anti-GST-peroxi-
dase conjugate to SAP97
PDZ2
and nonspecific formation

of disulfide links between GST proteins and SAP97 con-
tribute to the higher background (results not shown).
Crystal structure of SAP97
PDZ2
In order to obtain detailed structural information on
the binding, we set up crystallization screens for
purified PDZ1 and PDZ2 domains and PDZ1-3 segment
of SAP97. Attempts to crystallize SAP97
PDZ1
or
SAP97
PDZ1-3
were unsuccessful, but crystals were
obtained from SAP97
PDZ2
both alone and complexed
with GluR-A C-terminal 18-mer peptide (A
18
). During
initial purification of SAP97
PDZ2
, we noticed the occa-
sional formation of disulfide-linked dimers due to the
presence of a single cysteine residue at position 378. The
corresponding residue in the homologous PDZ2 domain
of PSD-95 is Gly219 (Fig. 3), which is located on the
protein surface 15 A
˚
from the binding pocket in the
solution NMR structure of PSD-95

PDZ2
[21]. To avoid
problems arising from the potential heterogeneity of
dimeric and monomeric PDZ domains and from disul-
fide-linked PDZ domain–peptide ligand complexes
observed previously in in vitro peptide binding studies
[19], we expressed and purified two variants in which
Cys378 was replaced either by glycine (C378G) or
the isosteric serine residue (C378S). While crystals of
wild-type SAP97
PDZ2
never diffracted in the absence
of the GluR-A
18
peptide, well-diffracting crystals,
belonging to space group P2
1
, were obtained from the
C378S (resolution 1.8 A
˚
) and C378G (2.44 A
˚
) variants
A
B
Fig. 2. Binding of GluR-A wild-type and mutated GluR-A C-termini to
SAP97 PDZ domain-coated microwells. (A) Schematic outline of the
assay. The immobilized PDZ domains and GST fusion protein con-
taining 11-residue C-terminal segment from GluR-A are illustrated.
(B) Binding of wild-type and L907A variant GST-GluR-A fusion pep-

tides to immobilized PDZ1, PDZ2, and PDZ3 domains of SAP97.
Fig. 3. Amino acid sequence similarities of PDZ domain proteins. Multiple sequence alignment of PDZ domains from SAP97, PSD-95,
PSD93, SAP102, Shank3, and mLin-10 proteins. The secondary structural elements of SAP97 PDZ2 are included at the top of the sequence
alignment. The residues (Gly328-Leu329-Gly330-Phe331) of the carboxylate binding loop and the critical histidine residue (His384) in aB are
highlighted by green and blue, respectively. The single cysteine residue (Cys378) in SAP97 is indicated by red color. The numbers refer to
amino acid sequence of the corresponding full-length proteins. Identity and similarity of the aligned residues are indicated by asterisks, semi-
colons (high similarity), and dots (low similarity), respectively.
I. von Ossowski et al. SAP97 PDZ2–GluR-A peptide complex
FEBS Journal 273 (2006) 5219–5229 ª 2006 The Authors Journal compilation ª 2006 FEBS 5221
(Tables 1 and 2). The structures were solved by molecu-
lar replacement (see Experimental procedures). There
are two PDZ domains in the asymmetric unit, each with
a typical PDZ domain folding topology with six
b-strands (bAtobF) and two a-helices (aA and aB)
(Fig. 4). The structure of the C378G variant of
SAP97
PDZ2
was practically identical to that of C378S
variant, except for the mutated residue, and because of
its better resolution, only the structure of the C378S var-
iant will be discussed below. The first two N-terminal
residues are not visible in either of the two chains, and
11 (chain A) or 13 (chain B) C-terminal residues includ-
ing the hexahistidinyl tags also appear unstructured.
The binding site region of chain B contains additional
electron density, modeled tentatively as two histidines,
packed into the binding groove between bB and aB,
which may arise from binding of the C-terminus of a
neighboring PDZ domain in the crystal. We therefore
used chain A, which does not contain anything in the

ligand binding site, to compare with the peptide-bound
structure. The serine and glycine residues, used to
replace Cys378 in the native PDZ domain in the C378S
and C378G variants are located on the outer surface
and the serine side chain pointing outwardly into solu-
tion (Fig. 4).
Comparison to closely related PDZ domain
structures
The solution structure of the second PDZ domain of
PSD-95 has been determined by NMR [21], and, quite
recently, coordinates of the PDZ2 domain of SAP102
crystallized in the absence of ligand have been depos-
ited in the PDB (access code: 2FE5). As these domains
Table 1. Data collection and processing statistics.
Data set
SAP97
PDZ2
wt
+ GluR-A
18
SAP97
PDZ2
C378G
+ GluR-A
18
SAP97
PDZ2
C378G SAP97
PDZ2
C378S

Space group P2
1
P2
1
P2
1
P2
1
Unit cell parameters
a 34.8 A
˚
34.4 A
˚
32.2 A
˚
32.3 A
˚
b 54.4 A
˚
54.3 A
˚
52.1 A
˚
52.4 A
˚
c 55.1 A
˚
54.6 A
˚
52.6 A

˚
52.9 A
˚
a 90° 90° 90° 90°
b 94.0° 84.0° 102.7° 102.7°
c 90° 90° 90° 90°
Wavelength (A
˚
) 0.934 0.934 1.519 0.931
Resolution (A
˚
)
a
20–2.35 (2.41–2.35) 34.1–2.2 (2.33–2.2) 20–2.44 (2.59–2.44) 20–1.8 (1.91–1.8)
Completeness (%)
a
97.5 (96.6) 99.7 (99.3) 96.6 (90.8) 97.7 (97.3)
Redundancy
a
7.5 (7.6) 5.0 (5.0) 7.4 (7.2) 2.5 (2.5)
R
merge
a,b
0.081 (0.517) 0.054 (0.40) 0.025 (0.08) 0.055 (0.40)
I ⁄ r
a
16.8 (4.1) 16.6 (3.5) 37.2 (17.8) 14.1 (2.8)
a
Values for the highest resolution shell are given in parentheses.
b

R
merge
¼ S
hkl
|I(hkl) – <I(hkl)>| ⁄S
hkl
I(hkl) where <I(hkl)> is the mean of the
symmetry equivalent reflections of I(hkl).
Table 2. Refinement statistics.
Data set
SAP97
PDZ2
wt
+ GluR-A
18
SAP97
PDZ2
C378G
+ GluR-A
18
SAP97
PDZ2
C378G SAP97
PDZ2
C378S
Total unique reflections ⁄ test set
(% size of the test set) 8005 ⁄ 422
(5)
9658 ⁄ 509
(5)

5908 ⁄ 311
(5)
14964 ⁄ 788
(5)
R
work
⁄ R
free
a,b
0.218 ⁄ 0.295
(0.319 ⁄ 0.350)
0.242 ⁄ 0.326
(0.350 ⁄ 0.443)
0.180 ⁄ 0.295
(0.209 ⁄ 0.380)
0.188 ⁄ 0.254
(0.245 ⁄ 0.374)
Bond rmsd from ideality (A
˚
) 0.017 0.017 0.015 0.012
Angle rmsd from ideality (°) 1.78 1.75 1.56 1.97
Residues in favored Ramachandran plot regions (%) 91.0 92.8 94.5 97.2
Residues in allowed Ramachandran plot regions (%) 98.9 98.3 98.9 100
a
Values for the highest resolution shell are given in parentheses.
b
R
work
¼ S
hkl

(|F
o
(hkl)| - |F
c
(hkl)|) ⁄S
hkl
|F
o
(hkl)| for F(hkl) not belonging to the
test set; R
free
¼ S
hkl
(|F
o
(hkl)| ) |F
c
(hkl)|) ⁄S
hkl
|F
o
(hkl)| for F(hkl) in the test set.
SAP97 PDZ2–GluR-A peptide complex I. von Ossowski et al.
5222 FEBS Journal 273 (2006) 5219–5229 ª 2006 The Authors Journal compilation ª 2006 FEBS
show 90% sequence similarity (80–89% identity) to
SAP97
PDZ2
, it was interesting to compare the three
unliganded PDZ domain structures. Figure 5 shows a
superposition of SAP97

PDZ2
C378S structure with the
NMR structure of PSD-95
PDZ2
and the crystal struc-
ture of SAP102. The backbone structures are highly
similar, except for some of the surface loops. The loop
bA-bB is packed closer to the C-terminus of aBin
SAP97 than in PSD-95 or SAP102. In addition, bB-bC
loop is extended outwards to the solution in SAP97
and SAP102, whereas it bends towards the C-terminus
in PSD-95.
Ligand binding interactions of SAP97
PDZ2
Well-diffracting crystals belonging to the same space
group as the free PDZ domains were obtained for the
wild-type and C378G variant of SAP97
PDZ2
cocrystal-
lized with a synthetic peptide representing the 18
C-terminal residues of GluR-A (A
18
). The two struc-
tures were solved and turned out to be highly similar
to each other (Table 3). Only the last four C-terminal
residues of A
18
peptide are visible in the complex as an
extended b-strand sandwiched between aB and bB
and antiparallel to the b-strand (Fig. 6A). The term-

inal carboxylate group of Leu907 forms hydrogen
bonds with main chain amide nitrogens of Leu329,
Gly330, and Phe331. The aliphatic side chain of the
C-terminal leucine is accommodated in a hydrophobic
pocket formed by the apolar side chains of Phe331,
Ile333, and Leu391 (Fig. 6B). The threonine at posi-
tion P
)2
forms an essential hydrogen bond with the
His384 at the N-terminal end of aB. These results are
in agreement with our previous mutagenesis study
which indicated that the ability of point-mutated
GluR-A C-terminal domains to bind to PDZ1-3 seg-
ment of SAP97 is not sensitive to residue changes at
positions P
)1
and P
)3
, while P
)2
and the C-terminal P
0
are critical [20]. The side chain of Cys378 points out
into solution in the complex of wild-type SAP97 and
αA
C
N
C378S
βA
βF

βC
βB
βD
αB
Fig. 4. Ribbon diagram of the SAP97 PDZ2 C378S variant structure.
Helices and strands are named according to Doyle et al. [22].
Ser378 is shown in sticks.
α A
αA
αB
αB
βF
βF
βA βA
βD
βD
βB
βB
βC
βC
Loop βA-βB Loop βA-βB
Loop βB-βC
Loop βB-βC
Fig. 5. Stereoscopic representation of
superposed backbone structures for
SAP97
PDZ2
C378S (blue), SAP102
PDZ2
(orange; 2FE5), and PSD-95

PDZ2
(magenta;
1QLC) [21]. Secondary structural elements
and loops are labeled accordingly.
I. von Ossowski et al. SAP97 PDZ2–GluR-A peptide complex
FEBS Journal 273 (2006) 5219–5229 ª 2006 The Authors Journal compilation ª 2006 FEBS 5223
A
18
peptide, consistent with a potential reactivity in
disulfide bond formation.
Conformational differences between the free
and peptide bound SAP97
PDZ2
Superposition of the a-carbon traces of the complex
structure and apo PDZ domain (C378S) reveals differ-
ences that may arise from ligand-induced conforma-
tional changes in the PDZ domain (Fig. 7A). The
major differences can be seen in areas which are close
to the C-terminus of the bound peptide: in the unli-
ganded PDZ domain, the bA-bB loop is closer to the
C-terminal end of the aB than in the complex struc-
ture, suggesting that peptide binding is accompanied
by slight opening of the structure, also including slight
alterations in the aB-bF loop. The peptide binding
cavity in the uncomplexed PDZ domain is partially
filled by five water molecules that occupy similar posi-
tions to each of the polar atoms, both main chain and
side chain, in the bound peptide. In addition, there is a
water molecule occupying the hydrophobic pocket for
the terminal leucine. Moreover, there is an intriguing

switch involving Lys324 in the bA-bB loop. In the
unliganded structure, it forms a tight ion-pair (2.6 A
˚
)
with Asp396 between aB and bF, while in the liganded
structure, Asp396 moves to the outside and Lys324
forms hydrogen-bonds to Thr394 Oc2, Thr394 O and
Ser395 O (Fig. 7B). Unlike the interactions already
described in Doyle et al. [22] (see Discussion), such
Table 3. Superpositions of PDZ domains.
No. of aligned residues ⁄ Ca rmsd (A
˚
) ⁄
Sequence identity (%) Stationary Stationary II
Moving SAP97
PDZ2
C378S (chain A) SAP97
PDZ2
+ GluR-A
18
(chain A)
SAP97
PDZ2
C378G + GluR-A
18
(chain A) 90 ⁄ 0.69 ⁄ 98 90 ⁄ 0.33 ⁄ 98
SAP102
PDZ2
(2FE5) 89 ⁄ 0.73 ⁄ 80 90 ⁄ 0.71 ⁄ 80
PSD95

PDZ3
(1BFE) 83 ⁄ 0.99 ⁄ 37 83 ⁄ 1.03 ⁄ 37
PSD95
PDZ3
+ CRIPT peptide (1BE9) 83 ⁄ 1.14 ⁄ 37 83 ⁄ 1.03 ⁄ 37
AB
Fig. 6. Ligand binding interactions between GluR-A C-terminal (A
18
) peptide and SAP97
PDZ2
. (A) A Fo-Fc omit electron density map of the
last four C-terminal residues of the GluR-A
18
peptide contoured at 3.0 r and the residues Gly328-Leu329-Gly330-Phe331-Ser332-Ile333 of
the carboxylate binding loop. The peptide backbone is shown in green and the carboxylate binding loop in grey. Hydrogen bonds between
GluR-A C-terminus and the carboxylate binding loop are indicated as dotted lines. (B) Schematic presentation produced by
LIGPLOT of the
polar and hydrophobic interactions between the C-terminal Leu907 of GluR-A
18
peptide and the PDZ domain. Carbon, nitrogen and oxygen
atoms are shown as black, blue and red spheres, respectively. Hydrogen bond are indicated as dotted lines (including distances in A
˚
) and
hydrophobic interactions as arcs with radial spokes.
SAP97 PDZ2–GluR-A peptide complex I. von Ossowski et al.
5224 FEBS Journal 273 (2006) 5219–5229 ª 2006 The Authors Journal compilation ª 2006 FEBS
bond movements could be the switch between the two
conformations for facilitating better accommodation of
the peptide in the PDZ binding groove. Recently, it
was reported [23] that residues within the bB-bC loop

of ZO1-PDZ1 rearrange in the liganded structure to
produce a deep pocket not present in the unliganded
form, for binding of the residue at P
)6
. However, in
our study, no conformational change in the extended
bB-bC loop was observed (Fig. 7A), and consequently
this region is less likely to contribute to ligand recogni-
tion in SAP97–GluR-A interaction.
Discussion
Leonard et al. [11] originally identified an interaction
between GluR-A and SAP97, and showed that the
extreme C-terminus of GluR-A binds to the PDZ1-3
segment of SAP97 in vitro. This PDZ interaction is
probably solely responsible for the association, because
GluR-A subunit lacking seven C-terminal residues
expressed in transgenic mice did not form an immuno-
precipitable complex with SAP97 unlike its wild-type
counterpart [10]. In vitro binding of SAP97
PDZ1-3
to
GluR-A C-terminal domain showed an absolute
requirement for a large hydrophobic side chain at the
C-terminus and a side chain hydroxyl group at the P
)2
position, in agreement with a typical class I PDZ inter-
action [20]. Of the three PDZ domains in SAP97, only
PDZ2 bound to the GluR-A C-terminus in our earlier
GST pulldown assay [20]. However, in our present
study, the use of microplate assay with immobilized

PDZ domains to analyze the interaction revealed that
the PDZ1 domain is also active, albeit less so than
PDZ2. The ability of SAP97
PDZ1
to bind to GluR-A is
in agreement with some earlier findings [24,25] and
with reports indicating that the PDZ1 and PDZ2
domains of PSD-95 family membrane-associated gua-
nylate kinase homologs (Maguks) display similar pre-
ferences for their peptide ligands, and differ in this
respect from PDZ3 [26].
The crystal structure of SAP97
PDZ2
with bound
GluR-A peptide, determined in the present study, is
consistent with the archetypical features of class I PDZ
domains. The interaction with bound peptide occurs at
positions P
0
(C-terminus) and P
)2
, but not at P
)1
and
P
)3
other than through main chain hydrogen bonds
with bB. In most class I PDZ domain-ligand pairs,
additional interactions engaging the amino acid residues
at P

)1
and P
)3
of the peptide in multiple polar and non-
polar contacts with the PDZ domain impart increased
affinity and sequence specificity, important for physio-
logical interactions. The absence of such supplementary
interactions from the GluR-A–SAP97
PDZ2
complex
may not be surprising considering that in GluR-A, ala-
nine and glycine residues occupy at P
)3
and P
)1
, which
is unusual for a class I PDZ ligand. These two residues
do not show any unusual structural features in the pep-
tide complex, and can be mutated to glycine ⁄ alanine
without affecting binding in GST pulldown assay [20].
Therefore, it seems probable that additional interac-
tions or mechanisms are required to provide further
specificity and affinity.
Previously, we had demonstrated with both in vitro
peptide binding assays and NMR spectroscopy that
under nonreducing conditions a disulfide bond may
form between the Cys893 of a GluR-A
18
peptide and
the Cys378 of SAP97

PDZ2
domain, and consequently
we suggested this covalent coupling could be the
basis for a redox-regulated binding mechanism [19].
Formation of a disulfide bond would be consistent
with the finding that in GST pulldown experiments,
A
B
Fig. 7. Structural changes in PDZ domain upon peptide binding. (A)
Superposition of the wild-type SAP97
PDZ2
(green) in the peptide
complex with the unliganded structure of the C378S variant (blue),
highlighting the potential movements of the bA-bB loop and aB
upon peptide binding. (B) Conformational switch in the bA-bB car-
boxylate binding loop between apo (blue) and liganded (green)
structures. Relevant amino acids and secondary structural elements
are labeled.
I. von Ossowski et al. SAP97 PDZ2–GluR-A peptide complex
FEBS Journal 273 (2006) 5219–5229 ª 2006 The Authors Journal compilation ª 2006 FEBS 5225
replacement of a tripeptide sequence Ser897-Ser898-
Gly899 in GluR-A C-terminal domain by three alanine
residues leads to loss of binding, possibly due to
decreased flexibility of the C-terminus [20]. However,
in our crystal structure of wild-type SAP97
PDZ2
cocrys-
tallized with the GluR-A
18
peptide, there was no dis-

cernible electron density for bound peptide beyond the
last four residues of the PDZ-motif and therefore
the structure does not provide an explanation for the
above-mentioned contributions by upstream sequence
elements. Whether this is due to insufficient length of
the peptide can definitely be answered only through
determination of the larger structures including the
entire C-terminal domain. Our attempts to form crys-
tals of SAP97
PDZ2
in complex with the GluR-A
CTD
have so far remained unsuccessful due to the poor
solubility and stability of purified full-length GluR-
A
CTD
protein.
Notwithstanding the remaining questions concerning
the structural basis for the exquisite specificity of the
interaction between SAP97 and GluR-A, the present
crystal structures provide interesting new findings. Of
the PSD-95 family Maguk proteins, only the structure
of PSD-95
PDZ3
has been solved in the presence of a
peptide ligand [22]. In addition, nonliganded structures
have been solved for the third PDZ domain of human
SAP97 [27], and for the second PDZ domain of
SAP102 (2FE5; Structural Genomics Consortium,
unpublished results). In the PSD)95

PDZ3
peptide com-
plex structure, peptide binding was accompanied by
only minor conformational changes limited to a few
amino acid side chains [22]. In their structure, the
authors also saw a small shift between liganded and
nonliganded forms, but ascribed it to crystal contacts, a
viewpoint we do not entirely share. While it is true that
there are some crystal contacts at the bottom of the
ligand-binding pocket, there is a correlation between
binding of ligand, changes in cell dimensions, and the
opening and closing of the pocket; whenever the ligand
is bound, the b-angle, for instance, is 84–93°, and
whenever it is not, the b-angle is 103° (Table 1). It
seems clear that the rearrangement of the interactions
around Lys324 provides two different but stable con-
formations that correlate with ligand being bound or
not. In contrast, in the structure of PSD-95 PDZ3 pep-
tide ligand complex, the residue equivalent to Lys324,
Arg318, binds the terminal carboxylate through a water
molecule. As described above, the interactions we see
are quite different and may help explain some of the
differences in binding properties between PSD-95
PDZ3
and SAP97
PDZ2
. Furthermore, although it may be a
coincidence, we find it interesting that both the PSD-95
and our structure make crystal contacts in the same
region (at the bottom of the peptide-binding groove)

even though the space group is different and the
sequence identity is low in this region. We wonder if
this may indicate some of the regions of allosteric speci-
ficity, and the possibility that other proteins interact
with this region as well. Direct structural evidence exists
for an analogous situation for non-PDZ protein inter-
action domains in PSD-95: an intramolecular interac-
tion between a ‘hook’ domain and guanylate kinase
domain keeps the molecule in a closed state [28,29].
In conclusion, our crystallographic analysis shows a
canonical class I interaction between SAP97
PDZ2
and
GluR-A
18
peptide. The discrepancy between the
relative scarcity of interactions in the complex struc-
ture and the high affinity and specificity evident from
in vitro binding experiments [19,20] suggests the
existence of additional, but as yet unclear, mechanisms
in this interaction. An unexpected and novel feature of
SAP97
PDZ2
structure is the conformational shift in
bA-bB loop between the free and liganded structures.
Experimental procedures
Bacterial expression and protein purification
C-terminally His-tagged SAP97
PDZ2
(residues 315–410 in

SwissProt Q911D0) was expressed in Escherichia coli
BL21(DE3)pLysS and purified by immobilized metal chela-
tion affinity chromatography on Ni
2+
-charged NTA-Agar-
ose (Qiagen, Hilden, Germany) as described previously
[19,20]. SAP97
PDZ2
C378S and C378G mutations were gen-
erated by PCR, and the mutated proteins were expressed
and purified as the wild-type domain. His-tagged PDZ1
(residues 213–313), PDZ3 (residues 459–450) and PDZ1-3
(213–450) segment were expressed and purified as above. All
PDZ constructs contained an eight amino acid residue seg-
ment SRHHHHHH at their C-termini. GST fusion proteins
of wild-type GluR-A C-terminal 11-mer peptide (GST-A
11
)
has been described previously [20]. GST fusion of C-term-
inal 11-mer containing a L907A mutation was produced by
PCR mutagenesis. GST fusion proteins were expressed in
E. coli BL21 and purified by GSH-Sepharose chromatogra-
phy (Amersham Biotech, Uppsala, Sweden) [20].
Mammalian cell expression
Expression plasmids encoding GFP-tagged GluR-A
CTD
(resi-
dues 827–907 in rat GluR-A; M36418) and GluR-A
CTD
var-

iants L907A and ‘XA908’ (carrying an extra alanine residue
after C-terminal Leu907) were generated by subcloning
appropriate PCR fragments between EcoRI and HindIII sites
of a derivative of pEGFP-C1 (Clontech, Palo Alto, CA,
USA), which harbored a hexahistidinyl tag linker
(HHHHHHEF) between GFP and the C-terminal insert.
SAP97 PDZ2–GluR-A peptide complex I. von Ossowski et al.
5226 FEBS Journal 273 (2006) 5219–5229 ª 2006 The Authors Journal compilation ª 2006 FEBS
HEK293 cells were grown in Dulbecco’s modified Eagle’s
medium supplemented with 2 mm glutamine, 10% (v ⁄ v) fetal
calf serum, penicillin (50 units ⁄ mL), and streptomycin (50
units ⁄ mL) at 37 °C under 5% CO
2
. Cells were transfected
with wild-type and point-mutated GFP-GluR-A
CTD
fusion
plasmids either alone or together with expression plasmid
encoding myc-tagged SAP97 [30] by using calcium phosphate
coprecipitation, and harvested 36–48 h after the transfection
and used for immunoprecipitations as described below.
Immunoprecipitations
Detergent extracts were prepared from transfected HEK293
cells by homogenizing the cells in 10 volumes of 50 mm
Tris ⁄ HCl, pH 8.0, 140 mm NaCl, 1 mm EDTA, 1 mm
Na
3
VO
4
,1mm NaF, and 1 mm phenylmethylsulfonyl fluor-

ide (TNE buffer). Triton X-100 was added to a final con-
centration of 1% (w ⁄ v), and the suspension was mixed at
4 °C for 2 h, followed by centrifugation at 20 000 g for
15 min. Aliquots (1-mL) of supernatant were incubated
with anti-GFP IgG (2 lg; Abcam, Cambridge, UK) or
SAP97
N
antiserum (2 lL) [20] overnight at 4 °C. Immuno-
complexes were bound to GammaBind G-Sepharose
(Amersham Biosciences; 20 lL) for 2 h under gentle rota-
tion. Sepharose beads were collected by brief centrifugation,
washed three times with TNE, and twice with NaCl ⁄ P
i
, and
finally suspended in 20 lL of SDS sample buffer, heated at
95 °C for 5 min and analyzed by SDS ⁄ PAGE and immuno-
blotting with anti-GFP or anti-myc antibodies [30].
Microplate binding assay
In vitro binding of wild-type and mutated GluR-A C-ter-
mini to SAP97 PDZ domains was analyzed by using a plate
binding assay. Briefly, microtiter plates were coated with
purified PDZ1, PDZ2, and PDZ3 domains (0.5 lm in
NaCl ⁄ P
i
; overnight at 4 °C), blocked with bovine serum
albumin, and incubated with GST fusion proteins
(5 lgÆmL
)1
in NaCl ⁄ P
i

) for 5–10 min at room temperature,
conditions determined to yield the highest signal-to-back-
ground ratio. The wells were washed three times with
NaCl ⁄ P
i
⁄ 0.5% Tween 20 and incubated with anti-GST-
IgG–horseradish peroxidase conjugate for 1 h at room tem-
perature. After further washes, peroxidase substrate was
added, and the absorbance values at 450 nm were recorded
after 30–60 min from substrate addition. Binding of GST
fusion proteins to bovine serum albumin-coated control
wells, generally representing <15% of total, was subtracted
from the absorbance values as nonspecific binding.
Protein crystallization
SAP97 PDZ2 domain proteins were crystallized at room
temperature using the sitting drop vapor diffusion method
equilibrated against a reservoir solution of 30% polyethy-
lene glycol (PEG 6000), 0.1 m Tris ⁄ HCl, pH 8.8, and 0.2 m
sodium acetate. Crystals of the C378G and C378S variant
proteins (1.3 and 0.85 mm in 10 mm Bis-Tris pH 6.5,
respectively) were grown in a mixture of 1 lL protein and
1 lL reservoir solution. Both the wild-type (2.3 mm) and
the C378G variant (1.3 mm) proteins were cocrystallized
with the GluRA
18
peptide (SIPCMSHSSGMPLGATGL)
by mixing 1 lL of protein and 1 lL of peptide (5 mm in
10 mm Bis-Tris pH 6.5) with 2 lL of the reservoir solution.
All crystals grew as rod clusters within 1–3 days.
Crystallographic structure determination

The data were collected at the ESRF (European Synchro-
tron Radiation Facility) beamlines ID29 for a 1.3 A
˚
reso-
lution data set of C378G, ID14-1 for wild-type and
C378G with peptide and ID14-3 for C378S. The data for
C378G variant were collected at EMBL ⁄ DESY beamline
BW7A. All data were collected at 100 K and processed
with the program xds [31] (Table 1). Attempts to use the
NMR structures of the PDZ1 [32] and PDZ2 [21]
domains of PSD-95 with 53% and 89% sequence iden-
tity, respectively, to SAP97
PDZ2
as search models yielded
no useful solutions for molecular replacement, probably
due to the rather high rmsd ⁄ Ca (Table 3). A molecular
replacement solution could be obtained for the high reso-
lution data set of the C378G variant using the crystal
structure of PDZ3 domain of SAP97 ⁄ hDlg, PDB-ID
1PDR [27] as a model. The molecular replacement was
performed with the program molrep [33] in the CCP4
suite [34]. Automatic model building and refinement was
carried out with the program arp ⁄ warp [35]. The model
thus obtained could not be refined satisfactorily by man-
ual model building and refinement, but using one mole-
cule from partly refined structure as a molecular
replacement model, the other data sets could be phased
by molecular replacement, although no solution was
obtained with 1PDR as a model. The program refmac5
was used for refinement and the programs o [36] and

coot [37] for manual model building. Structure validation
was done with the molprobity web server [38]. The
superpositions were carried out using the SSM algorithm
[39]. Figures were prepared with pymol [40] and ligplot
[41].
Protein data bank accession numbers
X-ray structural data and the derived atomic coordinates
have been deposited with the Protein Data Bank (http://
www.rcsb.org/pdb) under the accession codes: 2G2L (wild-
type SAP97 PDZ2 with GluR-A
18
peptide), 2AWX (C378S
variant), 2AWW (C378G variant with GluR-A
18
peptide),
and 2AWU (C378G variant).
I. von Ossowski et al. SAP97 PDZ2–GluR-A peptide complex
FEBS Journal 273 (2006) 5219–5229 ª 2006 The Authors Journal compilation ª 2006 FEBS 5227
Acknowledgements
We acknowledge the European Synchrotron Radiation
Facility MX beamlines and The EMBL BW7A beam-
line at the DORIS storage ring, DESY, Hamburg for
provision of synchrotron radiation facilities and we
would like to thank the beamline personnel for assis-
tance in data collection. We thank D. Bansfield for
technical assistance. This work was supported by
grants from the Academy of Finland to AG (1105157),
and to KK (202892). AG is a member of
Biocentrum Helsinki and gratefully acknowledges its
support.

References
1 Shi SH, Hayashi Y, Petralia RS, Zaman SH, Wenthold
RJ, Svoboda K & Malinow R (1999) Rapid spine deliv-
ery and redistribution of AMPA receptors after synaptic
NMDA receptor activation. Science 284, 1811–1816.
2 Zamanillo D, Sprengel R, Hvalby O, Jensen V, Burna-
shev N, Rozov A, Kaiser KM, Koster HJ, Borchardt T,
Worley P, et al. (1999) Importance of AMPA receptors
for hippocampal synaptic plasticity but not for spatial
learning. Science 284, 1805–1811.
3 Lissin DV, Carroll RC, Nicoll RA, Malenka RC & von
Zastrow M (1999) Rapid, activation-induced redistribu-
tion of ionotropic glutamate receptors in cultured hip-
pocampal neurons. J Neurosci 19, 1263–1272.
4 Bredt DS & Nicoll RA (2003) AMPA receptor traffick-
ing at excitatory synapses. Neuron 40, 361–379.
5 Malinow R & Malenka RC (2002) AMPA receptor traf-
ficking and synaptic plasticity. Annu Rev Neurosci 25,
103–126.
6 Passafaro M, Piech V & Sheng M (2001) Subunit-speci-
fic temporal and spatial patterns of AMPA receptor
exocytosis in hippocampal neurons. Nat Neurosci 4,
917–926.
7 Hayashi Y, Shi SH, Esteban JA, Piccini A, Poncer JC
& Malinow R (2000) Driving AMPA receptors into
synapses by LTP and CaMKII: requirement for GluR1
and PDZ domain interaction. Science 287, 2262–2267.
8 Shi S, Hayashi Y, Esteban JA & Malinow R (2001)
Subunit-specific rules governing AMPA receptor traf-
ficking to synapses in hippocampal pyramidal neurons.

Cell 105, 331–343.
9 Piccini A & Malinow R (2002) Critical postsynaptic
density 95 ⁄ disc large ⁄ zonula occludens)1 interactions
by glutamate receptor 1 (GluR1) and GluR2 required at
different subcellular sites. J Neurosci 22, 5387–5392.
10 Kim CH, Takamiya K, Petralia RS, Sattler R, Yu S,
Zhou W, Kalb R, Wenthold R & Huganir R (2005) Per-
sistent hippocampal CA1 LTP in mice lacking the
C-terminal PDZ ligand of GluR1. Nat Neurosci 8, 985–
987.
11 Leonard AS, Davare MA, Horne MC, Garner CC &
Hell JW (1998) SAP97 is associated with the alpha-
amino-3-hydroxy-5-methylisoxazole-4-propionic acid
receptor GluR1 subunit. J Biol Chem 273, 19518–19524.
12 Stricker NL & Huganir RL (2003) The PDZ domains
of mLin-10 regulate its trans-Golgi network targeting
and the surface expression of AMPA receptors. Neuro-
pharmacology 45, 837–848.
13 Uchino S, Wada H, Honda S, Nakamura Y, Ondo Y,
Uchiyama T, Tsutsumi M, Suzuki E, Hirasawa T &
Kohsaka S (2006) Direct interaction of post-synaptic
density-95 ⁄ Dlg ⁄ ZO-1 domain-containing synaptic mole-
cule Shank3 with GluR1 alpha-amino-3-hydroxy-5-
methyl-4-isoxazole propionic acid receptor. J Neurochem
97, 1203–1214.
14 Schulz TW, Nakagawa T, Licznerski P, Pawlak V,
Kolleker A, Rozov A, Kim J, Dittgen T, Kohr G,
Sheng M, et al. (2004) Actin ⁄ alpha-actinin-dependent
transport of AMPA receptors in dendritic spines: role
of the PDZ-LIM protein RIL. J Neurosci 24, 8584–

8594.
15 Rumbaugh G, Sia GM, Garner CC & Huganir RL
(2003) Synapse-associated protein-97 isoform-specific
regulation of surface AMPA receptors and synaptic
function in cultured neurons. J Neurosci
23, 4567–4576.
16 Nakagawa T, Futai K, Lashuel HA, Lo I, Okamoto K,
Walz T, Hayashi Y & Sheng M (2004) Quaternary
structure, protein dynamics, and synaptic function of
SAP97 controlled by L27 domain interactions. Neuron
44, 453–467.
17 Wu H, Nash JE, Zamorano P & Garner CC (2002)
Interaction of SAP97 with minus-end-directed actin
motor myosin VI. Implications for AMPA receptor traf-
ficking. J Biol Chem 277, 30928–30934.
18 Osterweil E, Wells DG & Mooseker MS (2005) A role
for myosin VI in postsynaptic structure and glutamate
receptor endocytosis. J Cell Biol 168, 329–338.
19 von Ossowski L, Tossavainen H, von Ossowski I, Cai
C, Aitio O, Fredriksson K, Permi P, Annila A & Keina
¨
-
nen K (2006) Peptide binding and NMR analysis of the
interaction between SAP97 PDZ2 and GluR-A: poten-
tial involvement of a disulfide bond. Biochemistry 45,
5567–5575.
20 Cai C, Coleman SK, Niemi K & Keina
¨
nen K (2002)
Selective binding of synapse-associated protein 97 to

GluR-A alpha-amino-5-hydroxy-3-methyl-4-isoxazole
propionate receptor subunit is determined by a novel
sequence motif. J Biol Chem 277, 31484–31490.
21 Tochio H, Hung F, Li M, Bredt DS & Zhang M (2000)
Solution structure and backbone dynamics of the sec-
ond PDZ domain of postsynaptic density-95. J Mol Biol
295, 225–237.
22 Doyle DA, Lee A, Lewis J, Kim E, Sheng M &
MacKinnon R (1996) Crystal structures of a complexed
and peptide-free membrane protein-binding domain:
SAP97 PDZ2–GluR-A peptide complex I. von Ossowski et al.
5228 FEBS Journal 273 (2006) 5219–5229 ª 2006 The Authors Journal compilation ª 2006 FEBS
molecular basis of peptide recognition by PDZ. Cell 85,
1067–1076.
23 Appleton BA, Zhang Y, Wu P, Yin JP, Hunziker W,
Skelton NJ, Sidhu SS & Wiesmann C (2006) Compara-
tive structural analysis of the Erbin PDZ domain and
the first PDZ domain of ZO-1. Insights into determi-
nants of PDZ domain specificity. J Biol Chem 281,
22312–22320.
24 Mehta S, Wu H, Garner CC & Marshall J (2001) Mole-
cular mechanisms regulating the differential association
of kainate receptor subunits with SAP90 ⁄ PSD-95 and
SAP97. J Biol Chem 276, 16092–16099.
25 Gardoni F, Mauceri D, Fiorentini C, Bellone C, Missale
C, Cattabeni F & Di Luca M (2003) CaMKII-depen-
dent phosphorylation regulates SAP97 ⁄ NR2A interac-
tion. J Biol Chem 278, 44745–44752.
26 Lim IA, Hall DD & Hell JW (2002) Selectivity and pro-
miscuity of the first and second PDZ domains of PSD-

95 and synapse-associated protein 102. J Biol Chem 277,
21697–21711.
27 Morais Cabral JH, Petosa C, Sutcliffe MJ, Raza S,
Byron O, Poy F, Marfatia SM, Chishti AH & Lidding-
ton RC (1996) Crystal structure of a PDZ domain.
Nature 382, 649–652.
28 McGee AW, Dakoji SR, Olsen O, Bredt DS, Lim WA
& Prehoda KE (2001) Structure of the SH3-guanylate
kinase module from PSD-95 suggests a mechanism for
regulated assembly of MAGUK scaffolding proteins.
Mol Cell 8, 1291–1301.
29 Tavares GA, Panepucci EH & Brunger AT (2001)
Structural characterization of the intramolecular interac-
tion between the SH3 and guanylate kinase domains of
PSD-95. Mol Cell 8, 1313–1325.
30 Cai C, Li H, Rivera C & Keina
¨
nen K (2005) Interaction
between SAP97 and PSD-95, two Maguk proteins
involved in synaptic trafficking of AMPA receptors.
J Biol Chem 281, 4267–4273.
31 Kabsch W (1993) Automatic processing of rotation dif-
fraction data from crystals of initially unknown symme-
try and cell constants. J Appl Cryst 26, 795–800.
32 Long JF, Tochio H, Wang P, Fan JS, Sala C, Nietham-
mer M, Sheng M & Zhang M (2003) Supramodular
structure and synergistic target binding of the N-term-
inal tandem PDZ domains of PSD-95. J Mol Biol 327,
203–214.
33 Vagin A & Teplyakov A (1997) MOLREP: an auto-

mated program for molecular replacement. J Appl Cryst
30, 1022–1025.
34 Collaborative Computational Project Number 4 (1994)
The CCP4 suite: programs for protein crystallography.
Acta Cryst D 50, 760–763.
35 Perrakis A, Morris R & Lamzin VS (1999) Automated
protein model building combined with iterative structure
refinement. Nat Struct Biol 6, 458–463.
36 Jones TA, Zou JY, Cowan SW & Kjeldgaard M (1991)
Improved methods for building protein models in elec-
tron density maps and the location of errors in these
models. Acta Cryst A 47, 110–119.
37 Emsley P & Cowtan K (2004) Coot: model-building
tools for molecular graphics. Acta Cryst D 60, 2126–
2132.
38 Lovell SC, Davis IW, Arendall WB 3rd, de Bakker PI,
Word JM, Prisant MG, Richardson JS & Richardson
DC (2003) Structure validation by Calpha geometry:
phi,psi and Cbeta deviation. Proteins 50, 437–450.
39 Krissinel E & Henrick K (2004) Secondary-structure
matching (SSM), a new tool for fast protein structure
alignment in three dimensions. Acta Cryst D 60, 2256–
2268.
40 DeLano WL (2002) The PyMOL Molecular Graphics
System. DeLano Scientific, San Carlos, CA, USA.
41 Wallace AC, Laskowski RA & Thornton JM (1995)
LIGPLOT: a program to generate schematic diagrams
of protein–ligand interactions. Protein Eng 8, 127–134.
I. von Ossowski et al. SAP97 PDZ2–GluR-A peptide complex
FEBS Journal 273 (2006) 5219–5229 ª 2006 The Authors Journal compilation ª 2006 FEBS 5229