Báo cáo khoa học: Elements of the C-terminal t peptide of acetylcholinesterase that determine amphiphilicity, homomeric and heteromeric associations, secretion and degradation docx

Elements of the C-terminal t peptide of acetylcholinesterase that
determine amphiphilicity, homomeric and heteromeric associations,
secretion and degradation
Ste
´
phanie Belbeoc’h, Cinzia Falasca, Jacqueline Leroy, Annick Ayon, Jean Massoulie
´
and Suzanne Bon
Laboratoire de Neurobiologie Cellulaire et Mole
´
culaire, CNRS UMR 8544, Ecole Normale Supe
´
rieure, Paris, France
The C-terminal t peptide (40 residues) of vertebrate acetyl-
cholinesterase (AChE) T subunits possesses a series of seven
conserved aromatic residues and forms an amphiphilic
a-helix; it allows the formation of homo-oligomers (mono-
mers, dimers and tetramers) and heteromeric associations
with the anchoring proteins, ColQ and PRiMA, which
contain a proline-rich motif (PRAD). We analyzed the
influence of mutations in the t peptide of Torpedo AChE
T
on
oligomerization and secretion. Charged residues influenced
the distribution of homo-oligomers but had little effect on
the heteromeric association with Q
N
, a PRAD-containing
N-terminal fragment of ColQ. The formation of homo-
tetramers and Q
N

-linked tetramers required a central core of
four aromatic residues and a peptide segment extending to
residue 31; the last nine residues (32–40) were not necessary,
although the formation of disulfide bonds by cysteine C37
stabilized T
4
and T
4
–Q
N
tetramers. The last two residues of
the t peptide (EL) induced a partial intracellular retention;
replacement of the C-terminal CAEL tetrapeptide by KDEL
did not prevent tetramerization and heteromeric association
with Q
N
, indicating that these associations take place in
the endoplasmic reticulum. Mutations that disorganize the
a-helical structure of the t peptide were found to enhance
degradation. Co-expression with Q
N
generally increased
secretion, mostly as T
4
–Q
N
complexes, but reduced it for
some mutants. Thus, mutations in this small, autonomous
interaction domain bring information on the features that
determine oligomeric associations of AChE

T
subunits and
the choice between secretion and degradation.
Keywords: acetylcholinesterase; degradation; disulfide
bonds; oligomerization; secretion.
In vertebrates, the acetylcholinesterase (AChE) gene gener-
ates several types of catalytic subunits through alternative
splicing in the 3¢ region of the transcripts [1–3]. These
subunits possess the same common catalytic domain,
followed by distinct C-terminal peptides (r, h and t),
characterizing the AChE
R
,AChE
H
and AChE
T
variants
[4–6]. In mammals, AChE
R
subunits seem to be expressed
mostly during embryogenesis and in the brain after stress
[7,8]; they correspond to a soluble, monomeric enzyme
species. AChE
H
subunits possess one or two cysteines and a
GPI-addition signal in their C-terminal peptide: they
generate GPI-anchored, disulfide-linked dimers, which
represent a major fraction of AChE in Torpedo electric
organs and muscles, and are expressed on the surface of
blood cells in mammals [9–12]. AChE

T
subunits are
expressed in muscles and in the nervous system of higher
vertebrates and therefore represent the functional cholinest-
erase species in the cholinergic system [3,13,14].
The C-terminal t peptide confers several characteristic
properties to AChE
T
subunits, allowing them to form a
series of homo-oligomers (monomers, dimers, tetramers and
higher oligomers) when expressed in transfected COS cells
[13,15]; some of these molecules are amphiphilic, i.e. interact
with detergent micelles [16,17]. AChE
T
subunits also form
hetero-oligomers with the collagen, ColQ, or with the
transmembrane protein, PRiMA [18,19]; in mammals,
these structural proteins anchor the major functional species
of cholinesterases in the basal lamina of the neuromuscular
junction and in neuronal cell membranes, respectively [20,21].
In the collagen-tailed and hydrophobic-tailed forms, four
catalytic AChE subunits are associated, through their C-
terminal t peptides, with proline-rich attachment domains
(PRAD) localized in the N-terminal regions of ColQ or
PRiMA [19,22,23].
The t peptide of AChE consists of 40 residues, with a series
of seven strictly conserved aromatic residues, including three
evenly spaced tryptophans, as well as acidic and basic
residues that are conserved or semiconserved in most
vertebrates [5]. This peptide is necessary for the amphiphilic

properties which characterize AChE
T
subunits and some
of their oligomers (T
a
1
,T
a
2
,T
a
4
), for the formation of
nonamphiphilic homotetramers (T
na
4
), as well as for the
Correspondence to S. Bon, Laboratoire de Neurobiologie Cellulaire et
Mole
´
culaire, CNRS UMR 8544, Ecole Normale Supe
´
rieure,
46 rue d’Ulm, 75005 Paris, France.
Fax: + 33 1 44 32 38 87, Tel.: + 33 1 44 32 38 91,
E-mail:
Abbreviations: AChE, acetylcholinesterase; AChE
H
, AChE subunit of
type H; AChE

R
, AChE subunit of type R; AChE
T
,AChEsubunit
of type T; ERAD, endoplasmic reticulum associated degradation;
PRAD, proline-rich attachment domain; r, h, t, alternative C-terminal
peptides of AChE; WAT, tryptophan (W) amphiphilic tetramerization
domain.
Enzymes: acetylcholinesterase (E.C. 3.1.1.7).
(Received 8 January 2004, revised 20 February 2004,
accepted 24 February 2004)
Eur. J. Biochem. 271, 1476–1487 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04052.x
heteromeric association of AChE
T
subunits with Q
N
,an
N-terminal fragment of collagen ColQ that contains a
proline-rich motif, thus producing T
4
–Q
N
complexes [23,24].
The t peptide constitutes an autonomous interaction
domain, called the WAT [tryptophan (W) amphiphilic
tetramerization] domain, because it can associate with a
PRAD, even in the absence of the catalytic domain;
moreover, addition of a t peptide at the C-terminus of
foreign proteins, green fluorescent protein and alkaline
phosphatase, endowed them with amphiphilic properties

and enabled them to form PRAD-associated tetramers [25].
We also found that the simultaneous presence of the t
peptide and of mutations at the interface of AChE dimers –
the Ôfour helix bundleÕ [26] – prevented the secretion of
AChE
T
subunits [27]. We recently showed that the t
peptide induces intracellular degradation through the
endoplasmic reticulum associated degradation (ERAD)/
proteasome pathway, to different extents, depending on the
protein to which it is attached, and that aromatic residues
are necessary for this effect [28].
Recent spectroscopic studies showed that the t peptide is
organized as an amphiphilic a helix, in which aromatic
residues form a hydrophobic sector [29,30]. In addition,
an analysis of intercatenary disulfide bonds in the T
4
–Q
N
complex also demonstrated that the four t peptides are
parallel and oriented in the same direction, opposite to that
of the PRAD [30]. The structure of a complex formed
between synthetic peptides (four t peptides with one PRAD)
confirmed this orientation (M. Harel et al., manuscript in
preparation).
In the present study, we mutated aromatic and charged
residues, suppressed the C-terminal cysteine or introduced
cysteines at other positions, and deleted more or less extended
C-terminal segments of the t peptide, in Torpedo AChE
T

subunits, to determine the structural basis of the character-
istic properties that the t peptide confers to AChE
T
subunits.
Materials and methods
AChE constructs and site-directed mutagenesis
Mutagenesis was performed according to the method of
Kunkel et al. [31]. cDNAs encoding wild-type and mutated
Torpedo AChE
T
, as well as the previously described
Torpedo Q
N
protein [23], intact or deleted of its PRAD
motif (residues 70–86), were inserted into the pEFBos
vector. Throughout this article, the residues of the t peptide
are numbered from 1 to 40, corresponding to positions
536–575 in the Torpedo AChE
T
subunit, so that the Torpedo
mutants are indicated by the modified residues, e.g. W17P.
Transfection of COS cells
COS cells were transfected by the DEAE-dextran method,
as described previously [24], using 4 lg of DNA encoding
the AChE catalytic subunit and 4 lg of DNA encoding Q
N
or PRAD-deleted Q
N
, per 100 mm dish. Because Torpedo
AChE folds into its active conformation at 27 °C, but not at

37 °C, the cells were incubated for 2 days at 37 °Cafter
transfection, then transferred to 27 °C and maintained at
this temperature for 3–4 days, in a medium containing 10%
Nuserum (Inotech, Dottikon, Switzerland), which had been
pretreated with 10
)6
M
soman to inactivate serum cholin-
esterases.
To analyze its heteromeric interaction with an associated
structural protein, AChE
T
was coexpressed with Q
N
[23].
By using Q
N
, rather than full-length ColQ, we avoid the
complexity caused by the formation of the triple helical
collagen and by the low salt aggregation of collagen-
tailed AChE forms [32]. We added a flag epitope
(DYKDDDDK) at the C-terminus of Q
N
,sothat
complexes containing this protein could be characterized
with the anti-flag immunoglobulin, M2 (Kodak), as
described previously [24]. The effect of Q
N
on the level of
cellular and secreted activity was analysed by comparing

the coexpression of AChE
T
with full-length Q
N
andwitha
PRAD-deleted Q
N
, to compensate for competition
between the two transfected vectors.
Cell extracts
The cells were extracted at 20 °C with TMg buffer (1%
Triton X-100, 50 m
M
Tris/HCl, pH 7.5, 10 m
M
MgCl
2
),
and then centrifuged at 10 000 g for 30 min. Media were
also centrifuged at 10 000 g for 30 min to remove cell debris
before analysis.
Enzyme assays
AChE activity was determined according to the colorimetric
method of Ellman et al. [33] at room temperature. As
the monomeric Torpedo AChE forms produced by some
mutants were inactivated by 5,5¢-dithiobis(2-nitrobenzoic
acid) [34], the enzyme samples were incubated for variable
periods of time, depending on their activity, with a reaction
medium containing acetylthiocholine iodide in phosphate
buffer, pH 7; 5,5¢-dithiobis(2-nitrobenzoic acid) was then

added and the absorbance at 414 nm was determined using
a Labsystems (Helsinki, Finland) Multiskan RC automatic
plate reader. Alkaline phosphatase and b-galactosidase
from Escherichia coli were assayed with the chromo-
genic substrates p-nitrophenyl phosphate and o-nitrophenyl
galactoside, respectively.
Sedimentation and electrophoretic analyses
Centrifugation was performed in 5–20% sucrose gradients
(50 m
M
Tris/HCl, pH 7.5, 50 m
M
MgCl
2
, either in the
presence of 0.2% Brij-97 or in the presence of 0.2% Triton
X-100) in a Beckman SW41 rotor, at 36 000 r.p.m., for 18 h
at 6 °C. The gradients contained E. coli b-galactosidase
(16 S) and alkaline phosphatase (6.1 S) as internal sedi-
mentation standards [24]. Amphiphilic molecules generally
sediment faster in the presence of Triton X-100 than in
the presence of Brij-97, providing an indication of their
amphiphilic character.
Electrophoresis in nondenaturating polyacrylamide gels
was performed as described by Bon et al.[16],andAChE
activity was revealed by the histochemical method of
Karnovsky & Roots [35]. In charge shift electrophoresis,
the electrophoretic migration of amphiphilic molecules was
accelerated in the presence of sodium deoxycholate, when
compared to migration in the presence of the neutral

detergent, Triton X-100, alone. As an index of the degree of
Ó FEBS 2004 Oligomerization and secretion of the AChE t peptide (Eur. J. Biochem. 271) 1477
amphiphilicity, we used the ratio between migration in the
presence of DOC to migration in Triton X-100 alone, after
normalizing these migrations to that of a nonamphiphilic
species, the wild-type tetramers T
na
4
or T
4
–Q
N
.
Both sedimentation and nondenaturing electrophoresis
provide semiquantitative information on the interaction of
AChE molecules with micelles, and are generally in
complete agreement. However, in the present study, we
found that some mutations in the t peptide perturb
amphiphilic interactions in such a way that sedimentation
became essentially identical in the presence of Triton X-100
and Brij-97, while charge shift electrophoresis still showed a
marked influence of the detergent: this was the case for
dimers of aromatic mutants such as W17H or W17A. In
addition, the T
4
–Q
N
complexes formed by mutants W17F
and W17A showed an unusual retardation in sedimentation
in the presence of Triton X-100, compared with Brij-97.

Results
Analyses of AChE activity and molecular forms
Figure 1 shows the sequence of the t peptide of Torpedo
AChE
T
subunits, and schematically illustrates its proposed
a helical structure, its association with the PRAD of
ColQ, and the various oligomers of AChE
T
subunits that
result from its interactions.
We analyzed how mutations in the t peptide affect the
levels of cellular and secreted activity of Torpedo AChE in
transfected COS cells. The activities were normalized to
those obtained for wild-type AChE
T
in parallel transfections.
Immunofluorescence of the protein produced at early stages
after transfection indicated that all mutants were expressed
in a similar manner. After 2 days at 27 °C, a temperature
which allows the correct folding of active Torpedo AChE
(see the Materials and methods), the level of cellular activity
reached a plateau and the rate of secretion remained
constant. Maximal secretion was obtained for a truncated
mutant (I3C/stop4), which retained only the first two
residues of the t peptide, followed by a cysteine at position
3; this cysteine allowed the formation of dimers, which
lacked the aromatic residues and were therefore nonamphi-
philic. The secretion of active wild-type AChE
T

subunits
was less than 10% of the truncated mutant, showing that a
large fraction is degraded intracellularly [27,28].
The molecular forms of AChE were identified by
electrophoresis in nondenaturing polyacrylamide gels and
their amphiphilic character was evaluated by charge shift
electrophoresis in the presence or absence of sodium
deoxycholate [16]. As Torpedo AChE
T
monomers are
rapidly inactivated under the conditions of electrophoretic
migration, the distribution of AChE molecular forms was
analyzed by sedimentation in sucrose gradients.
To analyse the capacity of Torpedo AChE
T
subunits to
associate with a PRAD, we coexpressed them with protein
Q
N
(Fig. 1D). This Q
N
protein organizes wild-type AChE
T
subunits into tetramers (T
4
–Q
N
) that are nonamphiphilic
and efficiently secreted [24], reaching  40% of the secretion
observed with the truncated I3C/stop4 mutant. The forma-

tion of Q
N
-linked oligomers therefore rescued an important
fraction of the wild-type catalytic Torpedo AChE
T
subunits
from intracellular degradation.
Fig. 1. Structure of the t peptide and oligomeric associations of
acetylcholinesterase type T subunits (AChE
T
). (A) primary sequence
of the t peptide from Torpedo AChE
T
subunits. The residues of the
t peptide, encoded by an alternatively spliced 3¢ exon, are numbered
from 1 to 40 and correspond to residues 536–575 of the mature
Torpedo AChE
T
subunit; cysteine C37, which is responsible for
intercatenary disulfide bonds, is circled. (B) Side view of the t
peptide, with its 1–32 segment organized as an a helix. The con-
served aromatic residues are located in the upper sector of the helix.
(C) Wheel representation of the entire t peptide, putatively organ-
ized as an a helix. Aromatic residues, shown in shaded circles, are
located in the upper sector; charged residues are in double circles
(white for basic residues, grey for acidic residues) and possible salt
bridges are marked by hatched bars; cysteine C37 is in a double,
grey circle; arrowheads indicate residues that have been mutated to
cysteines. (D) Primary sequence of the proline-rich attachment
domain (PRAD) motif from Torpedo ColQ. The PRAD residues

are shown in bold text (from cysteines 70 and 71 to phenylalanine
86), and a few adjacent residues are shown in non-bold text.
(E) Schematic representation of a complex between four t peptides
and a PRAD. The N- and C-terminal extremities (indicated N and
C) and arrows show the orientations of the t peptides (black zig-
zags) running opposite to the PRAD (grey line); cysteines are
indicated by circles, joined by lines representing disulfide bonds.
(F) Major types of homomeric and heteromeric associations
analyzed in this study: T
a
1
,T
a
2
and T
a
4
, amphiphilic monomer,
dimer and tetramer of AChE
T
subunits; T
na
4
, nonamphiphilic tetr-
amer; T
4
–Q
N
, tetramer associated with the N-terminal Q
N

fragment
of ColQ, containing the PRAD motif. The schemes of heteromeric
complexes are derived from recent studies (M. Harel, H. Dvir,
S. Bon, W. Q. Liu, M. Vidal, C. Garbay, J. L. Sussman,
J. Massoulie
´
& I. Silman, unpublished results) [30].
1478 S. Belbeoc’h et al. (Eur. J. Biochem. 271) Ó FEBS 2004
Mutation of charged residues of the t peptide
The t peptide contains seven acidic (D, E) and eight basic
(H, K, R) residues, which may form intracatenary salt
bridges in the helical conformation (Fig. 1C) and perhaps
intercatenary salt bridges in oligomeric assemblies; we
mutated these residues to alanines, individually or in groups.
Mutations D4A/E5A, E7A/R8A, K11A, E13A, R16A,
K25A, or D29A did not markedly modify the levels of
cellular and secreted activities. However, other mutations
had a stronger effect, as shown in Fig. 2A. Both cellular and
secreted activities were increased by the point mutation,
E1A, but decreased by replacement of the first four acidic
residues (E1, D4, E5, E7) by alanines. Mutation H15A
enhanced the efficiency of secretion, because it decreased the
cellular activity but increased the secreted activity.
Likethewild-typeAChE
T
subunits, all mutants produced
amphiphilic dimers (T
a
2
) and nonamphiphilic tetramers

(T
na
4
). However, their proportions varied, as illustrated by
the sedimentation profiles of four mutants (Fig. 2B). These
profiles did not change with time after transfection. They
characterize each mutant and are not simply related to the
intracellular concentration of the enzyme, as shown by the
fact that the D4A/E5A and R16A mutants produced appro-
ximately the same levels of cellular activity with the same
proportions of molecular forms as the wild type, but differed
in the activity and molecular forms of the secreted enzyme.
Conversely, the secreted enzyme was quantitatively and
qualitatively similar for mutants K11A and D29A, although
the patterns of cellular molecular forms were different.
In all cases, coexpression with Q
N
increased the level of
secretion and produced T
4
–Q
N
complexes, as for the wild
type.
Mutation of aromatic residues
The three tryptophans (W10, W17, W24) and Y31 were
mutatedtoalanines,andallsevenaromaticresidueswere
mutated to prolines. These mutations had little effect on the
cellular activity; the secreted activity was reduced by about
half by most mutations, but significantly increased by Y31A

(data not shown) and Y31P (Fig. 3A). The major molecular
Fig. 2. Mutations of charged residues in the t peptide. (A) Acetylcho-
linesterase (AChE) activities in cell extracts and secreted into the cul-
ture medium are shown for the wild type and four mutants. Grey bars
and hatched bars correspond to the AChE activities of mutants
expressed without or with Q
N
, respectively (Materials and methods);
the activities are normalized to those obtained for the wild type (100%)
both in the cell extracts and in the medium; the standard errors were
obtained from five independent experiments. For other individual
mutations (K11A, E13A, R16A, K25A and D29A), the cellular
activities ranged from 77% to 100%, and the secreted activities
between 68% and 136%. (B) Sedimentation patterns of cellular and
secreted AChE, in sucrose gradients containing 0.2% Triton X-100.
The shaded areas, as well as the total areas under the sedimentation
profiles, are proportional to the relative activities of the mutants, so
that the surface of each peak represents the actual activity of the
corresponding molecular form: monomers (T
1
), dimers (T
2
)and
tetramers (T
4
).
Fig. 3. Mutations of aromatic residues. (A) Secreted activities, nor-
malized to that of the wild type, for mutants of aromatic residues to
prolines: the bars represent secreted activities obtained when acetyl-
cholinesterase type T subunits (AChE

T
) were expressed without Q
N
(greybars)andwithQ
N
(hatched bars); the indicated values are the
means of at least three independent experiments. The cellular activities
ranged from 86 to 114% without Q
N
andfrom65to117%withQ
N
.(B)
Amphiphilic character of dimers produced by each mutant, indicated by
charge shift electrophoresis. RDOC/TX is the ratio of electrophoretic
migrations in the presence of Triton X-100 with sodium deoxycholate
and Triton X-100 alone, normalized to those of a nonamphiphilic
species (T
na
4
); the indicated values represent the means of three to six
independent experiments. (C) Existence of Q
N
-linked dimers with mu-
tant W10P: electrophoretic patterns, in the presence of Triton X-100
and sodium deoxycholate; the third lane shows that a fraction of dimers
and tetramers was retarded by the M2 antibody (x), indicating that they
were associated with the Q
N
-flag protein. Note that the coexpression
with Q

N
increased the secretion of dimers as well as tetramers. (D)
AChE molecular forms secreted by mutants of aromatic residues to
prolines, expressed with and without Q
N
; electrophoretic analysis in the
presence of Triton X-100 and deoxycholate. s,AChE
T
dimers; d,
Q
N
-linked dimers; h, tetramers; j,T
4
–Q
N
complexes (i.e. tetramer
associated with the N-terminal Q
N
fragment of ColQ, containing the
proline-rich attachment domain motif); the origin of migration is shown
by a thin line.
Ó FEBS 2004 Oligomerization and secretion of the AChE t peptide (Eur. J. Biochem. 271) 1479
forms produced by these mutants were T
2
dimers, as
illustrated in electrophoretic patterns (Fig. 3D); the pro-
duction of tetramers was strongly reduced or abolished,
again with the exception of Y31 mutants. The amphiphilic
character of T
2

dimers was retained when individual
aromatic residues were replaced with alanines, but it was
reduced when the central aromatic residues were mutated to
prolines, in a position-dependent manner (Fig. 3B).
The production of T
4
–Q
N
complexes, resulting in an
increased secretion, was not affected by mutation of Y31,
but was reduced by mutations of W10 and F28, and was
essentially suppressed by mutations of F14, W17, Y20
and W24, either to alanine (not shown) or to proline
(Fig. 3A,D). As illustrated in Fig. 3C, we found that when
the W10P mutant was coexpressed with the flagged Q
N
protein, the anti-flag M2 immunoglobulin reacted with a
fraction of dimers as well as with tetramers, indicating the
presence of T
2
–Q
N
complexes, in addition to T
4
–Q
N
complexes, in the culture medium.
We replaced the central tryptophan (W17) with a
hydrophobic aromatic residue (F), an aliphatic aromatic
residue (L), a heterocyclic residue (H), as well as a proline

(P) and an alanine (A). As shown in Fig. 4, these mutations
did not strongly modify the cellular activity, which remained
within the range of 83–127% of the wild type, but reduced
or suppressed the formation of homotetramers; the amphi-
philic character of the resulting dimers was similar to that of
the wild type with F, L or A, it was significantly reduced
with H and it was essentially abolished with P.
Figure 4 also shows that there was no interaction with
Q
N
in the case of W17L and W17P, a very small production
of T
4
–Q
N
in the case of W17A, and a significant production
of this complex in the case of W17F and W17H. For these
two mutants, coexpression with Q
N
increased the level of
secreted activity to about 60% of that obtained in the wild
type; in addition to the T
4
–Q
N
complex, this coexpression
markedly increased the secretion of dimers, particularly for
W17H, but unlike those formed with the W10P mutant,
these dimers did not seem to be associated with Q
N

,as
they did not react with the M2 antibody. Whereas the
sedimentation of the wild-type T
4
–Q
N
complex was abso-
lutely unaffected by the presence of Triton X-100 or Brij-97
in the gradient, the sedimentation of T
4
–Q
N
complexes
formed with W17F and W17A was reproducibly retarded in
the presence of Triton X-100 (compared to Brij-97),
showing an opposite effect to that normally observed for
amphiphilic enzyme species, such as T
a
1
,T
a
2
or T
a
4
(see
Fig. 6A); the electrophoretic migration of these complexes
was also slower than that of the wild-type complex. This
may reveal an interaction with Triton X-100 micelles, but
not with Brij-97 micelles, perhaps because of an unusual

exposure of the aromatic groups in these complexes.
It is noteworthy that, in contrast to the wild-type T
na
4
and
T
4
–Q
N
, the tetramers formed with the W17F, W17H or
W17A mutants were only observed in the medium, but not
in the cell extract. This indicates a significant difference in
the cellular trafficking of the wild-type and mutant
complexes.
Perturbation of the helical organization of an aromatic
cluster
To perturb the a helical organization of the aromatic-rich
segment of the t peptide, we deleted residues T12 and M21,
located, respectively, in its N-terminal region and near its
centre (Fig. 1A). Mutation M21W introduced an additional
aromatic residue, which might create a steric disturbance in
oligomers or in heteromeric complexes with Q
N
.
These mutations had moderate effects on the level of
cellular activity, and decreased secretion to  50% of the
wild type. The three mutants produced mostly amphiphilic
T
a
2

dimers in the cell extracts; in the case of M21D,the
medium only contained T
na
4
tetramers (Fig. 5A), in contrast
to the wild type, in which these molecular forms are present
both in the cell extracts and in the medium.
Figure 5A also shows that coexpression with Q
N
increased secretion for T12D and M21W (to about 35%
and 50% of the wild type, respectively), but not for M21D;
T
4
–Q
N
complexes of T12D and M21W were characterized
in the medium by reaction with the M2 antibody, but were
undetectable or barely detectable in the cell extracts, in
contrast to the wild-type T
4
–Q
N
complex.
Effect of a cysteine at various positions in the t peptide
The formation of intercatenary disulfide bonds between
wild-type AChE
T
subunits depends on the free cysteine
residue located near the C-terminus of the t peptide, C37.
Mutation of this cysteine to a serine reduced both cellular

and secreted activities; it suppressed the formation of dimers
and reduced cellular and secreted tetramers (Fig. 6A); in the
presence of Q
N
, the secretion of T
4
–Q
N
complexes was
reduced to  75% of that of the wild type. Thus, the
presence of an intercatenary disulfide bond appears to be
necessary for dimerization, but not for tetramerization,
particularly in the presence of Q
N
.
To determine whether cysteines at other positions could
allow dimerization and further oligomerization, we
replaced residues I3, A6, T12, S19, M21, M22 or H34
with a cysteine, with or without mutation of C37 (C37S).
Fig. 4. Molecular forms produced by W17 mutants; interaction with
Q
N
. Sedimentation patterns of cellular and secreted molecular forms;
the areas under the profiles are proportional to the corresponding
activities; the top of the wild-type T
4
–Q
N
(tetramer associated with the
N-terminal Q

N
fragment of ColQ, containing the proline-rich attach-
ment domain motif) peak exceeds the frame and is shifted downwards.
Molecular forms expressed without Q
N
(––s––) and with Q
N
(- - -j )
were analyzed in the presence of Triton X-100; sedimentation was also
performed in the presence of Brij-97 (Bj) for molecular forms secreted by
mutants W17F, W17H and W17A (ÆÆÆÆÆÆ). Note an unusual retardation
by Triton X-100 (Tx) for W17F and W17A.
1480 S. Belbeoc’h et al. (Eur. J. Biochem. 271) Ó FEBS 2004
The relative levels of cellular and secreted activities, as
well as the distribution of molecular forms, are illustrated
in Fig. 6A,B. Unlike C37S, none of these mutants
produced monomers without dimers; therefore, when
two cysteines were present, they were not engaged in an
intracatenary disulfide bond, but could form intercatenary
bonds in dimers.
Mutation I3C (with or without C37S) considerably
increased the cellular activity, mostly as amphiphilic dimers;
secreted activity was also increased, but to a much lesser
degree. The presence of the N-terminal cysteine thus
appears to facilitate dimerization and to reduce degrada-
tion. This mutation also increased the cellular activity
obtained with the W17P mutant, without restoring its
capacity to interact with Q
N
.

Mutation A6C somewhat decreased the cellular activity,
but strongly increased secretion. The A6C mutant mainly
produced a nonamphiphilic 14 S species, possibly corres-
ponding to octamers. This unusual oligomer dissociated
during storage, particularly in the presence of detergent
(Triton X-100), transiently producing amphiphilic tetramers
(T
a
4
) (which are not usually observed in the wild type) and
amphiphilic dimers (T
a
2
). When C37 was absent (A6C/
C37S), the 14 S species was observed in the cell extracts, but
seemed to be less stable, being almost entirely converted to
T
a
2
dimers in the medium.
The T12C mutation introduced a cysteine in the
N-terminal part of the aromatic-rich segment (not shown).
Fig. 6. Oligomeric forms obtained with cysteines at different positions in
the t peptide. (A) Sedimentation patterns of cellular and secreted
molecular forms in gradients containing Triton X-100 (––) and Brij-97
(- - -); the shaded areas and the areas under the profiles are propor-
tional to the corresponding activities. Note that dimers containing
cysteines at position 21 did not sediment faster with Triton X-100 than
with Brij-97, in contrast to the amphiphilic dimers (wild type, or with
cysteines at positions 3, 6 or 34). Mutants A6C and A6C/C37S pro-

duced a 14 S species that was progressively dissociated into amphi-
philic tetramers (T
a
4
) and ultimately amphiphilic dimers (T
a
2
), as
shown for cell extracts in an upper profile. (B) Interaction of cysteine
mutants with Q
N
, as shown by electrophoretic analysis of secreted
molecular forms, in the presence of Triton X-100 and sodium deoxy-
cholate(comparewithFig.5A).NotethatmutantM21C(with
cysteine C37) produced complexes with Q
N
, retarded by M2, whereas
mutant M21C/C37S (without cysteine C37) did not.
Fig. 5. Mutations in the aromatic-rich region; mutations of methionines
M21 and M22. (A) Electrophoretic patterns of cellular and secreted
molecular forms, in the presence of Triton X-100, with and without
deoxycholate (DOC); complexes with Q
N
that were retarded by M2
are indicated by x; the symbols are as in Fig. 4. The secretion of
mutant M21D was not increased by coexpression with Q
N
, and no
complex reacting with M2 was detected. In the case of T12D and
M21W, T

4
–Q
N
(tetramer associated with the N-terminal Q
N
fragment
of ColQ, containing the proline-rich attachment domain motif) com-
plexes were secreted, but undetectable or barely detectable in cell
extracts. (B) Cellular and secreted activities (represented as in Fig. 2A)
for mutants containing cysteine C37 or not (C37S). Coexpression with
Q
N
increased the secretion of mutants of methionine M22 and mutants
of methionine M21 which possessed cysteine C37, but not of mutants
M21A/C37S and M21S/C37S (data not shown), lacking both M21
and C37.
Ó FEBS 2004 Oligomerization and secretion of the AChE t peptide (Eur. J. Biochem. 271) 1481
In the absence of cysteine C37, this allowed the formation
of amphiphilic dimers, which were secreted together with
nonamphiphilic tetramers and a 14 S species. This species
was, in fact, predominant in the secreted enzyme and
appeared to be much more stable than that formed with
A6C, as it was not dissociated after secretion. In the
presence of cysteine C37, the T12C mutant produced mainly
nonamphiphilic tetramers, which represented the only
secreted form. This suggests that tetramers may be stabilized
when disulfide bonds were formed at the two positions,
12 and 37.
A cysteine at position 19, in the aromatic-rich segment
but opposite to the aromatic cluster, had very different

effects, depending on the presence of cysteine C37. Without
C37, mutant S19C/C37S produced very low levels of
cellular or secreted activity. In contrast, mutant S19C
(containing two cysteines at positions 19 and 37) showed a
high level of secretion, mostly as nonamphiphilic tetramers,
as observed for T12C.
Mutations M21C and M22C, with or without cysteine
C37, had little effect on cellular activity (compared to the
wild-type and C37S mutant, respectively), but increased
secretion to various degrees. M21C and M21C/C37S
secreted both tetramers and nonamphiphilic dimers, while
M22C and M22C/C37S secreted mostly tetramers. The
dimers formed with a cysteine at position 21 appeared
nonamphiphilic, suggesting that the aromatic clusters may
be masked by an intercatenary disulfide bond in the
aromatic-rich segment.
Finally, the mutants containing a cysteine at position 34
(H34C, H34C/C37S) behaved essentially like the wild type,
suggesting that the C-terminal segment of the t peptide is
flexible.
Figure 6B shows that the various cysteine mutants
formed T
4
–Q
N
complexes (reacting with the anti-flag M2
immunoglobulin), except M21C/C37S. Thus, mutation
M21C suppressed the heteromeric complex when cysteine
C37 was absent, but not when it was present: this illustrates
the importance of the C-terminal cysteine for the assembly

and/or stabilization of the T
4
–Q
N
complex, in agreement
with the formation of intercaternary disulfide bonds
between the t peptide and Q
N
cysteines.
Coexpression with Q
N
generally increased secretion when
T
4
–Q
N
complexes were produced, although this effect was
marginal or absent for mutants that showed a high level of
secretion without Q
N
(A6C, A6C/C37S, S19C). However,
coexpression induced a decrease, of  40%, in the secretion
of mutant M21C/C37S, for which complexes could not be
detected (not shown). This suggests that Q
N
did interact
with the mutant AChE
T
subunits, but induced their
degradation rather than the assembly of a stable, secretable

hetero-oligomer.
The role of methionine 21
The fact that M21C/C37S did not associate with Q
N
,
whereas M22C/C37S formed a T
4
–Q
N
complex, may be
related to the orientation of the two adjacent methionines
relative to the aromatic cluster, and to a possible structural
role of methionine M21: this residue is conserved in
vertebrate t peptides, while M22 is replaced with other
residues in some species. To examine this possibility, we
mutated M21 and M22 to alanines or serines (with or
without C37S). Figure 5B shows that coexpression with Q
N
increased the level of secretion, indicating the formation of
T
4
–Q
N
complexes, for all mutants except M21A/C37S (and
M21S/C37S). Thus, the presence of a methionine at position
22 is dispensable, but a methionine at position 21 contri-
butes to the stability of the complex, especially when
cysteine 37 is absent.
The C-terminal region of the t peptide: a retention motif?
The last four residues of the t peptide, CAEL, contain the

cysteine involved in intercatenary disulfide bonds and also
resemble the classical ER-retention signal, KDEL. To
determine whether its presence might induce a partial
retention of AChE
T
subunits, we introduced various
mutations in this motif (Fig. 7A).
It should first be noted that mutation C37S (where the
cysteine was removed) did not increase secretion, but rather
decreased both cellular and secreted activities; this effect
may result from the fact that suppression of the cysteine
prevented dimerization and reduced the level of secreted
tetramers, as discussed above.
Fig. 7. Effects of the C-terminal cysteine (C37), of C-terminal segments
and of a KDEL motif on acetylcholinesterase (AChE) molecular forms.
(A) Sedimentation patterns of cellular (upper row) and secreted (lower
row) enzyme, in gradients containing Triton X-100. The shaded areas
and the areas under the sedimentation profiles are proportional to the
activities. In mutant T
KDEL
, the C-terminal tetrapeptide (CAEL) was
replaced with the canonical ER retention motif (KDEL). All mutants
lacking a cysteine produced and secreted amphiphilic monomers with
variable proportions of nonamphiphilic tetramers; these tetramers
were secreted at a higher level for T
KDEL
than for C37S (SAEL).
(B) Cellular and secreted activities obtained with and without Q
N
.The

cellular activities decreased with the extent of C-terminal deletions;
coexpression with Q
N
increased secretion to a level comparable to that
of the wild type for mutants C37S, T
KDEL
, stop34 and stop32, but had
essentially no effect for stop29.
1482 S. Belbeoc’h et al. (Eur. J. Biochem. 271) Ó FEBS 2004
Dimerization was also suppressed when the CAEL motif
was replaced with KDEL (T
KDEL
) because this mutation
removed the cysteine. The presence of a C-terminal KDEL
tetrapeptide increased the level of cellular activity by about
threefold relative to C37S, mostly corresponding to T
a
1
;this
increase in cellular enzyme appeared to facilitate tetra-
merization, as tetramers (T
na
4
) were secreted at a higher
levelthanwithC37S(Fig.7A).
We also deleted the last two residues, EL (stop39): the
mutant in which the C-terminal motif was reduced to CA
produced dimers and secreted 1.7-times more activity than
the wild type. More extensive deletions, which removed the
cysteine, suppressed dimerization, so that monomers were

predominant in the cells and in the medium, and the levels
of activity were reduced in both compartments; the secretion
of homotetramers was reduced in stop34, compared to
C37S, and tetramerization was abolished in stop32.
Like C37S, the T
KDEL
, stop39, stop34 and stop32
mutants formed T
4
–Q
N
complexes, so that their secretion
was increased in the presence of Q
N
(Fig. 7B). In contrast,
the shorter mutant, stop29, showed no interaction with Q
N
,
as indicated by the fact that coexpression did not affect
either the secreted activity or the molecular forms, charac-
terized by sedimentation. To determine whether the differ-
ence between stop32 and stop29 could be ascribed to one of
the three residues D29, Q30 or Y31, we mutated each of
them to alanine: the level of T
4
–Q
N
complexes was similar to
that of stop32 for Y31A/stop32 and Q30A/stop32, but
considerably reduced for D29A/stop32, suggesting a specific

influence of residue D29; the strong effect observed by
mutation of this charged residue contrasts with the result
obtained when it was mutated in the full-length t peptide
(see above).
Progressive deletions from the C-terminus
of the I3C mutant
To analyze the effect of C-terminal deletions without
preventing dimerization, we used the I3C/C37S mutant,
which produced mostly amphiphilic dimers (Fig. 6A) and
formed T
4
–Q
N
complexes when coexpressed with Q
N
(Fig. 8A). As in the wild type, the replacement of the
C-terminal tetrapeptide by KDEL increased the cellular
activity (mainly T
a
2
) and reduced its secretion, but did not
abolish association with Q
N
, producing T
4
–Q
N
complexes
which were secreted. Deletion of the last nine residues
(I3C/stop32) did not abolish association with Q

N
, but
deletion of the last 12 residues (I3C/stop29) suppressed it
completely. Thus, the presence of a cysteine at position 3
did not modify the requirement of residues 29–31 for
interaction with Q
N
.
The effect of deletions on the cellular and secreted
activities is illustrated in Fig. 8B. The cellular activity
remained approximately constant for all deletions, about
50% of the value observed with the full-length t peptide.
Removal of the last two residues (EL) increased secretion, as
in the case of the wild type. More extensive deletions in the
C-terminal region reduced secretion, compared to that of
I3C/C37S/stop39, but deletions within the aromatic region
progressively increased it, reaching a plateau when all aro-
matic residues were removed. The dimers were amphiphilic
when they contained at least 29 residues of the t peptide
(stop29 and longer), but not if they contained 24 residues
or fewer (stop24 and shorter), i.e. when they lacked some of
the core aromatic residues.
Fig. 8. Effects of C-terminal segments and of a KDEL motif, in
mutants containing an N-terminal cysteine (I3C). (A) Interaction with
Q
N
, indicated by electrophoretic patterns of cellular (top) and secreted
(bottom) molecular forms obtained with and without Q
N
,inthe

presence of Triton X-100 and sodium deoxycholate. Note that mu-
tants I3C/C37S, I3C/KDEL and I3C/stop37 produced homomeric
T
2
a
dimers (s), homomeric T
na
4
tetramers (h)andT
4
–Q
N
(tetramer
associated with the N-terminal Q
N
fragment of ColQ, containing the
proline-rich attachment domain motif) complexes (j), whereas I3C/
stop29 did not. Homomeric tetramers of the I3C/KDEL mutant
appeared to be partially retained intracellularly. (B) Effect of
C-terminal deletions on cellular and secreted activities; progressive
deletions were made from the C-terminus of mutants containing an
N-terminal cysteine (I3C) which allows an efficient dimerization; in
the two longer mutants, the C-terminal cysteine was replaced with a
serine (C37S); mutated residues are underlined in the sequence, shown
along the horizontal axis. Cellular (upper frame) and secreted (lower
frame) activities, expressed as percentage of the wild type, are shown
as a function of the remaining length of the t peptide. Asterisks
indicate mutants that produced amphiphilic dimers; mutants stop34
and shorter produced nonamphiphilic dimers.
Ó FEBS 2004 Oligomerization and secretion of the AChE t peptide (Eur. J. Biochem. 271) 1483

Discussion
The t peptide: an elongated amphiphilic a helix
with a cluster of aromatic residues
Previous studies suggested that the amphiphilic properties
of the t peptide reflect the formation of a cluster of aromatic
residues in its a helical conformation [30]. The present
mutations confirm the role of aromatic residues, but show
that they differ considerably in their importance.
Amphiphilicity was not affected by mutations of charged
residues, or by deletions of the C-terminal region which
removed up to 11 residues, i.e. maintained all the aromatic
residues, except Y31. The amphiphilic character was
reduced to various extents when the central residues (F14,
W17, Y20, W24, F28) were replaced with prolines, but was
not affected by mutation of the first and last residues (W10,
Y31). Replacement of the most critical residue, W17, by
other, different, amino acids (F, L, A, H, P) showed that
amphiphilicity was indifferent to their aromatic nature, that
it did not strongly depend on their hydrophobicity, but was
very sensitive to the a helical structure. Thus, the amphi-
philic properties of the t peptide appear to depend
predominantly on the spatial organization of a cluster of
hydrophobic residues.
In agreement with a previous study [30], we obtained no
evidence that two cysteines, introduced at several positions
in the N- and C-terminal regions of the mutated t peptides,
could form an intracaternary disulfide bond. The present
results thus confirm that the t peptide forms an elongated
amphiphilic helix, rather than folding back on itself as a
hairpin in which the N- and C-terminal ends might be joined

by a disulfide bond.
Homomeric associations of AChE
T
subunits
In contrast to the wild-type AChE
T
subunits, the C37S
mutant did not form stable dimers, showing that an
intercaternary disulfide bond is necessary. The position of
this bond appeared very flexible, as dimers were formed
when cysteines were introduced at various positions along
the t peptide (with or without cysteine C37): this did not
seem to depend on the orientation of the residue relative
to the helical axis, although some positions produced
higher proportions of dimers than others. However,
although most dimers were amphiphilic, those formed in
the presence of a cysteine at position 21 were nonamphi-
philic, indicating that, in this case, the hydrophobic
patches occluded each other because of the formation of a
disulfide bond joining the aromatic clusters near their
centers.
Depending on the position of an added cysteine, AChE
T
subunits could form predominantly dimers or tetramers – or
even higher oligomers sedimenting at 14 S, possibly octa-
mers. This shows that the interactions between the t peptides
may present different geometries. Thus, the t peptides can
form different homomeric assemblies, which may be
stabilized by intercaternary disulfide bonds and are influ-
enced by the positions of these linkages.

In contrast to dimers, tetramers can be formed in the
absence of cysteine in the t peptide [27,36], although at a
lower level than for the wild type. Tetramers are generally
nonamphiphilic (T
na
4
), but some tetramers may also be
amphiphilic (T
a
4
), particularly those resulting from the
dissociation of the nonamphiphilic 14 S species. Thus,
aromatic clusters may be either occluded or at least partly
exposed in tetrameric assemblies, indicating that they
correspond to distinct quaternary organizations.
Heteromeric associations with the PRAD-containing
protein, Q
N
The major physiological role of the t peptide is clearly to
allow the functional localization of AChE
T
tetramers
through their association with PRAD-containing proteins,
ColQ and PRiMA. In the present study, we focused our
attention on the formation of quaternary associations with
an N-terminal fragment of ColQ, Q
N
. This protein assem-
bles with wild-type AChE
T

subunits to form Q
N
-linked
tetramers (T
4
–Q
N
), which are nonamphiphilic. Previous
studies showed that in these hetero-oligomers, two catalytic
subunits are disulfide-linked with Q
N
, while the other two
are disulfide-linked together. However, in the absence of
cysteine C37, this association still occurs, indicating that it
does not require the formation of intercaternary disulfide
bonds.
The complex was formed when the t peptide carried an
additional cysteine at positions 3, 6, 12, 19, 22 or 34, with or
without the original cysteine C37. It was not formed with a
cysteine at position 21, except when C37 was present: a
cysteine instead of a methionine at position 21 therefore
appears unfavorable. This may be partly because of the
formation of nonamphiphilic disulfide-linked dimers in
which the aromatic clusters are not available for interaction
with the proline-rich domain of Q
N
(PRAD). However, the
mutation of methionine 21 to an alanine or a serine also
weakened the formation of the complex, which again
required the presence of cysteine C37: mutants M21A/C37S

and M21S/C37S did not associate with Q
N
. In contrast,
similar mutations of methionine M22 did not prevent the
formation of T
4
–Q
N
complexes. This demonstrates that the
complex was stabilized by disulfide bonds through cysteine
37, and by the presence of methonine 21, in agreement with
strong interactions of this methionine with the PRAD in a
complex of isolated peptides (WAT)
4
PRAD (M. Harel
et al., manuscript in preparation). Nevertheless, the fact that
M21 could be replaced with a tryptophan suggests that the
complex can accommodate the steric constraint owing to a
more bulky residue.
The role of aromatic residues in the formation of the Q
N
-
linked complex has been established previously [4,28,37].
Using deletions of single residues, we show here that the
structure of the cluster is crucial for this quaternary
interaction: it was abolished by deletion of residue 21, in
the middle of the aromatic-rich segment. The fact that
deletion of residue 12, near the N-terminal end of the
aromatic cluster, had no such effect, indicates that the
orientation of the aromatic cluster relative to the catalytic

domain is not crucial. This is consistent with the notion of a
flexible junction between the catalytic domain and the
amphiphilic helical region of the t peptide [30]. In fact,
addition of a variable number of residues between the
catalytic domain and the t peptide did not prevent
association with Q
N
(N. Morel & S. Bon, unpublished).
1484 S. Belbeoc’h et al. (Eur. J. Biochem. 271) Ó FEBS 2004
In the present study, we assessed the importance of
aromatic residues, individually, by point mutations. We
observed that mutation of the central residues (F14, W17,
Y20, W24) to proline or alanine had a much stronger effect
than mutation of W10 and F28, and that Y31 had no effect.
Although not identical, the impacts of these mutations were
similar to those observed on the amphiphilic character of
AChE
T
subunits. However, mutations of W17 to different
amino acids clearly dissociated the two properties, as
mutation W17L maintained the amphiphilic character,
but totally abolished the association with Q
N
, like mutation
W17P. This association was reduced when W17 was
replaced with a phenylalanine or a histidine, and even more
strongly when it was replaced with an alanine, emphasizing
theimportanceofanaromaticside-chain.
With some mutants, we obtained evidence that Q
N

induced the formation of AChE
T
dimers: coexpression with
Q
N
strongly increased the secretion of dimers that were not
covalently associated with Q
N
(as in the case of W17F and
W17H), or were at least partially disulfide-linked with it (as
in the case of W10P). Such dimers may represent an
intermediate stage in the assembly of Q
N
-linked tetramers,
or result from the dissociation of unstable tetramers.
Q
N
-linked tetramers are usually nonamphiphilic, indica-
ting that the clusters of aromatic residues are occluded when
the t peptides are associated with the PRAD, in agreement
with their strong involvement in these quaternary interac-
tions and with the crystallographic structure of a complex of
synthetic peptides (M. Harel, H. Dvir, S. Bon, W. Q. Liu,
M.Vidal,C.Garbay,J.L.Sussman,J.Massoulie
´
&
I. Silman, unpublished results). However, the Q
N
-linked
tetramers, found with the W17F and W17A mutants, showed

some interaction with detergents, suggesting that they were
less compact than the wild-type complexes. In fact, the
formation of Q
N
-linked complexes in the presence of cysteines
within the t peptides reveals a considerable flexibility, because
disulfide bonds between these residues do not seem to be
compatible with the distances between pairs of homologous
residues in a complex of wild-type synthetic peptides [30].
The heteromeric assocation with Q
N
was not suppressed
by removal of the last nine residues (following Y31),
showing that it depends primarily on the aromatic-rich
segment and does not require the C-terminal part of the
t peptide. However, removal of three additional residues
(D29, Q30, Y31) abolished the interaction, and point
mutations showed that this was mostly caused by the
deletion of D29: although mutations of charged residues in
the full-length t peptide had little effect on association with
Q
N
, this suggests that a salt bridge contributed significantly
to its stability when the complex was weakened by a
C-terminal deletion.
Similarity between nonamphiphilic homomeric
and Q
N
-linked tetramers
We observed that the formation of homomeric T

na
4
tetramers was suppressed by all mutations which affected
the heterometic complex, T
4
–Q
N
, suggesting that both
quaternary assemblies depend on the same interactions and
possess a similar organization, in agreement with the fact
that they are both nonamphiphilic. In fact, except for
M21C/C37S, mutations that affected Q
N
-linked tetramers
appeared to reduce homomeric tetramers more severely,
indicating that tetramers are generally stabilized by the
presence of a PRAD. In T
4
–Q
N
complexes, the four a helical
t peptides are organized as a super-helix, forming a hollow
tube lined by aromatic side-chains, which is occupied by the
PRAD. This central space is unlikely to be filled with water
molecules or to remain empty in T
na
4
tetramers; it may be
reduced by a change in the pitch of the super-helix.
Homo- and hetero-oligomerization occur in the ER,

subcellular trafficking
The presence of an ER-retention motif (KDEL) at the
C-terminus of the t peptide blocked secretion, as expected,
but did not prevent dimerization when a cysteine was
introduced at position 3, in the N-terminal region of the
t peptide. The KDEL motif actually increased the formation
of homotetramers (T
na
4
)intheT
KDEL
mutant, compared to
the C37S mutant which also lacked the C-terminal cysteine
and was terminated by the tetrapeptide, SAEL. Similarly, a
C-terminal KDEL did not block the formation of hetero-
meric complexes T
4
–Q
N
; furthermore, the KDEL motif
was sterically masked in the complexes, as the complexes
were efficiently secreted. The fact that retention of AChE
T
subunits in the ER did not prevent homomeric or hetero-
meric associations indicates that they occur in this com-
partment.
We found that the last two residues of the t peptide, EL,
which are also present in the ER-retention tetrapeptide,
KDEL, exert a weaker, but significant, retention effect, as
their deletion increased secretion. It is possible that these

residues help to retain isolated AChE
T
subunits in the ER,
and thus facilitate their physiological association with the
anchoring proteins, ColQ and PRiMA.
Structural differences certainly explain that complexes
formed with the T12D andsomearomaticmutantsfollowed
a different cellular trafficking than the wild type: these
mutant complexes were secreted, but not detectable in
cellular extracts, showing that they did not accumulate in
the secretory compartment, but were either rapidly secreted
or degraded. We made a similar observation for homomeric
tetramers in the case of mutant M21D.
Oligomerization, secretion and degradation
In steady-state cultures, we may assume that all mutants
were produced at the same rate, as they only differ in the
short C-terminal t peptide, so that the rates of secretion of
different mutants represent the difference between the
common rate of synthesis and the rate of degradation. In
a previous study, we established that the presence of the
t peptide induces a partial degradation of AChE
T
subunits
through the ERAD pathway and that this effect depends on
the presence of aromatic residues [28]. Occlusion of these
residues may explain that oligomerization generally facili-
tates secretion: for example, the relative proportions of
monomers, dimers and tetramers in cellular extracts and in
the medium indicate that the secretion of wild-type AChE
T

subunits increases with their degree of oligomerization. This
explains why the introduction of cysteines at certain
positions increased secretion, by facilitating the formation
of dimers (I3C), tetramers (M22C) or 14 S oligomers (A6C).
Ó FEBS 2004 Oligomerization and secretion of the AChE t peptide (Eur. J. Biochem. 271) 1485
Conversely, suppression of cysteine C37 (C37S or deletions)
reduced secretion.
Similarly, coexpression with Q
N
generally increased
secretion by recruiting AChE
T
subunits into T
4
–Q
N
hetero-
oligomers. However, we observed an opposite effect in the
case of a few mutants (S19C, M21C/C37S): this implies
that the interaction of AChE
T
subunits with Q
N
produced
distorted complexes that were degraded more actively than
without Q
N
. Thus, oligomerization does not always protect
against degradation, but may actually increase it.
We found that extensive deletions of the aromatic-rich

segment increased secretion, indicating a reduced degrada-
tion, in agreement with a determining role of aromatic
residues in this process [4,28,37]. However, the present
analysis of point mutations within this segment shows
that degradation is not simply determined by the number
of aromatic residues: degradation was increased by a large
variety of mutations, including addition of an aromatic
residue (M21W), replacement with aromatic or other
residues (e.g. W17 to F, L, H, A or P; M21 to A or S) or
deletions (T12D,M21D). This suggests that degradation was
increased by disorganization of the aromatic cluster of the
wild-type t peptide. Thus, the structure of the wild-type
t peptide limits degradation.
Variations in the secretion level showed that degradation
was also affected by mutations in the N-terminal and
C-terminal region flanking the aromatic-rich segment.
Secretion was increased by mutation E1A and decreased
by mutation Ô4AÕ in the N-terminal region. It was increased
by deletion of the last two C-terminal residues (EL), and
decreased by mutation of the cysteine (C37S) or by deletion
of four to 11 residues from the C-terminus. When secretion
was suppressed by addition of a C-terminal ER-retention
signal (KDEL), changes in the level of cellular activity
reflected variations in degradation: an analysis of such
mutants confirmed that deletion in the C-terminal segment
increased degradation (stop32-KDEL, stop29-KDEL), rel-
ative to a mutant in which only two residues were mutated
to introduce the KDEL tetrapeptide (T
KDEL
). The fact that

deletions of C-terminal segments increased degradation may
result from a disorganization of the a-helical structure.
Mutations of charged residues showed that the level of
degradation is not simply correlated with the capacity of the
t peptide for oligomerization. Using progressive deletions in
a mutant that possessed an N-terminal cysteine at position
3, allowing an efficient dimerization, we found that the
targeting of AChE
T
subunits towards degradation or
secretion depends, in a complex manner, on different
regions of the t peptide. All of these results show that
various elements and features of the t peptide contribute to
determine the cellular fate of the protein and deserve further
investigations. It is highly probable that processes similar to
those demonstrated here, in COS cells, operate in muscle
and nerve cells that normally express functional AChE, but
this will have to be established.
Acknowledgements
We thank Ms Claudine Schmitt for technical assistance and M. Noe
¨
l
Perrier for critical reading of the manuscript. This work was
supported by grants from the Centre National de la Recherche
Scientifique, the Association Franc¸ aise contre les Myopathies, the
Direction des Forces et de la Prospective, and the European
Community; S. Belbeoc’h was the recipient of a PhD grant from
the French Ministry of Research.
References
1. Sikorav, J.L., Duval, N., Anselmet, A., Bon, S., Krejci, E., Legay,

C., Osterlund, M., Reimund, B. & Massoulie
´
, J. (1988) Complex
alternative splicing of acetylcholinesterase transcripts in Torpedo
electric organ; primary structure of the precursor of the glycolipid-
anchored dimeric form. EMBO J. 7, 2983–2993.
2. Li, Y., Camp, S., Rachinsky, T.L., Getman, D. & Taylor, P.
(1991) Gene structure of mammalian acetylcholinesterase. Alter-
native exons dictate tissue-specific expression. J. Biol. Chem. 266,
23083–23090.
3. Li, Y., Camp, S. & Taylor, P. (1993) Tissue-specific expression and
alternative mRNA processing of the mammalian acetyl-
cholinesterase gene. J. Biol. Chem. 268, 5790–5797.
4. Massoulie
´
, J., Pezzementi, L., Bon, S., Krejci, E. & Vallette, F.M.
(1993) Molecular and cellular biology of cholinesterases. Prog.
Neurobiol. 41, 31–91.
5. Massoulie
´
,J.,Anselmet,A.,Bon,S.,Krejci,E.,Legay,C.,Morel,
N. & Simon, S. (1998) Acetylcholinesterase: C-terminal domains,
molecular forms and functional localization. J. Physiol. (Paris)
92, 183–190.
6. Massoulie
´
, J. (2002) The origin of the molecular diversity and
functional anchoring of cholinesterases. Neurosignals 11, 130–143.
7. Legay, C., Bon, S. & Massoulie
´

, J. (1993) Expression of a cDNA
encoding the glycolipid-anchored form of rat acetylcholinesterase.
FEBS Lett. 315, 163–166.
8. Kaufer, D., Friedman, A., Seidman, S. & Soreq, H. (1998) Acute
stress facilitates long-lasting changes in cholinergic gene expres-
sion. Nature 393, 373–377.
9. Bon, S. (1982) Molecular forms of acetylcholinesterase in devel-
oping Torpedo embryos. Neurochem. Int. 4, 577–585.
10. Futerman, A.H., Low, M.G., Michaelson, D.M. & Silman, I.
(1985) Solubilization of membrane-bound acetylcholinesterase by
a phosphatidylinositol-specific phospholipase C. J. Neurochem.
45, 1487–1494.
11. Coussen, F., Bonnerot, C. & Massoulie
´
, J. (1995) Stable expres-
sion of acetylcholinesterase and associated collagenic subunits in
transfected RBL cell lines: production of GPI-anchored dimers
and collagen-tailed forms. Eur. J. Cell Biol. 67, 254–260.
12. Coussen, F., Ayon, A., Le Goff, A., Leroy, J., Massoulie
´
,J.&
Bon, S. (2001) Addition of a glycophosphatidylinositol to acetyl-
cholinesterase; processing, degradation, and secretion. J. Biol.
Chem. 276, 27881–27892.
13. Legay, C., Bon, S., Vernier, P., Coussen, F. & Massoulie
´
, J. (1993)
Cloning and expression of a rat acetylcholinesterase subunit:
generation of multiple molecular forms and complementarity with
a Torpedo collagenic subunit. J. Neurochem. 60, 337–346.

14. Krejci, E., Legay, C., Thomine, S., Sketelj, J. & Massoulie
´
,J.
(1999) Differences in expression of acetylcholinesterase and
collagen Q control the distribution and oligomerization of the
collagen-tailed forms in fast and slow muscles. J. Neurosci. 19,
10672–10679.
15. Duval, N., Massoulie
´
, J. & Bon, S. (1992) H and T subunits of
acetylcholinesterase from Torpedo, expressed in COS cells, gen-
erate all types of globular forms. J. Cell Biol. 118, 641–653.
16. Bon, S., Toutant, J.P., Me
´
flah, K. & Massoulie
´
, J. (1988)
Amphiphilic and nonamphiphilic forms of Torpedo cholines-
terases. II. Electrophoretic variants and phosphatidylinositol
phospholipase C-sensitive and -insensitive forms. J. Neurochem.
51, 786–794.
17. Bon, S., Rosenberry, T.L. & Massoulie
´
, J. (1991) Amphiphilic,
glycophosphatidylinositol-specific phospholipase C (PI-PLC)-
1486 S. Belbeoc’h et al. (Eur. J. Biochem. 271) Ó FEBS 2004
insensitive monomers and dimers of acetylcholinesterase. Cell.
Mol. Neurobiol. 11, 157–172.
18. Krejci, E., Thomine, S., Boschetti, N., Legay, C., Sketelj, J. &
Massoulie

´
, J. (1997) The mammalian gene of acetylcholinesterase-
associated collagen. J. Biol. Chem. 272, 22840–22847.
19. Perrier, A.L., Massoulie
´
,J.&Krejci,E.(2002)PRiMA,the
membrane anchor of acetylcholinesterase in brain. Neuron 33,
275–285.
20. Fernandez, H.L., Moreno, R.D. & Inestrosa, N.C. (1996)
Tetrameric (G4) acetylcholinesterase: structure, localization, and
physiological regulation. J. Neurochem. 66, 1335–1346.
21. Feng, G., Krejci, E., Molgo, J., Cunningham, J.M., Massoulie
´
,J.
& Sanes, J.R. (1999) Genetic analysis of collagen Q: roles
in acetylcholinesterase and butyrylcholinesterase assembly and in
synaptic structure and function. J. Cell Biol. 144, 1349–1360.
22. Duval, N., Krejci, E., Grassi, J., Coussen, F., Massoulie
´
,J.&Bon,
S. (1992) Molecular architecture of acetylcholinesterase collagen-
tailed forms; construction of a glycolipid-tailed tetramer. EMBO
J. 11, 3255–3261.
23. Bon, S., Coussen, F. & Massoulie
´
, J. (1997) Quaternary associa-
tions of acetylcholinesterase; II. the polyproline attachment
domain of the collagen tail. J. Biol. Chem. 272, 3016–3021.
24. Bon, S. & Massoulie
´

, J. (1997) Quaternary associations of
acetylcholinesterase. I. Oligomeric associations of T subunits with
and without the amino-terminal domain of the collagen tail.
J. Biol. Chem. 272, 3007–3015.
25. Simon, S., Krejci, E. & Massoulie
´
, J. (1998) A four-to-one asso-
ciation between peptide motifs: four C-terminal domains from
cholinesterase assemble with one proline-rich attachment domain
(PRAD) in the secretory pathway. EMBO J. 17, 6178–6187.
26. Sussman, J.L., Harel, M., Frolow, F., Oefner, C., Goldman, A.,
Toker, L. & Silman, I. (1991) Atomic structure of acetyl-
cholinesterase from Torpedo californica: a prototypic acetylcho-
line-binding protein. Science 253, 872–879.
27. Morel, N., Leroy, J., Ayon, A., Massoulie
´
,J.&Bon,S.(2001)
Acetylcholinesterase H and T dimers are associated through the
same contact; mutations at this interface interfere with the
C-terminal T peptide, inducing degradation rather than secretion.
J. Biol. Chem. 276, 37379–37389.
28. Belbeoc’h, S., Massoulie
´
, J. & Bon, S. (2003) The C-terminal T
peptide of acetylcholinesterase enhances degradation of unass-
embled active subunits through the ERAD pathway. EMBO J. 22,
3536–3545.
29. Cottingham, M.G., Voskuil, J.L.A. & Vaux, D.J.T. (2003) The
intact human acetylcholinesterase C-terminal oligomerization
domain is alpha-helical in situ and in isolation, but a shorter

fragment forms beta-sheet-rich amyloid fibrils and protofibrillar
oligomers. Biochemistry 42, 10863–10873.
30. Bon,S.,Dufourcq,J.,Leroy,J.,Cornut,I.&Massoulie
´
, J. (2004)
The C-terminal t peptide of acetylcholinesterase forms an alpha
helix that supports homomeric and heteromeric interactions. Eur.
J. Biochem. 271, 33–47.
31. Kunkel, T.A., Roberts, J.D. & Zakour, R.A. (1987) Rapid and
efficient site-specific mutagenesis without phenotypic selection.
Methods Enzymol. 154, 367–382.
32. Bon,S.,Ayon,A.,Leroy,J.&Massoulie
´
, J. (2003) Trimerization
domain of the collagen tail of acetylcholinesterase. Neurochem.
Res. 28, 523–535.
33. Ellman, G.L., Courtney, K.D., Andres, V. & Featherstone, R.M.
(1961) A new and rapid colorimetric determination of acetyl-
cholinesterase activity. Biochem. Pharmacol. 7, 88–95.
34. Morel, N., Bon, S., Greenblatt, H.M., Van Belle, D., Wodak, S.J.,
Sussman, J.L., Massoulie
´
, J. & Silman, I. (1999) Effect of muta-
tions within the peripheral anionic site on the stability of acetyl-
cholinesterase. Mol. Pharmacol. 55, 982–992.
35. Karnovsky, M.J. & Roots, L. (1964) A direct-coloring thiocholine
method for cholinesterase. J. Histochem. Cytochem. 12, 219–222.
36. Velan, B., Grosfeld, H., Kronman, C., Leitner, M., Gozes, Y.,
Lazar, A., Flashner, Y., Marcus, D., Cohen, S. & Shafferman, A.
(1991) The effect of elimination of intersubunit disulfide bonds

on the activity, assembly, and secretion of recombinant
human acetylcholinesterase. Expression of acetylcholinesterase
Cys-580fiAla mutant. J. Biol. Chem. 266, 23977–23984.
37. Blong, R.M., Bedows, E. & Lockridge, O. (1997) Tetramerization
domain of human butyrylcholinesterase is at the C-terminus.
Biochem. J. 327, 747–757.
Ó FEBS 2004 Oligomerization and secretion of the AChE t peptide (Eur. J. Biochem. 271) 1487