Assembly of Acetylcholinesterase Tetramers by Peptidic Motifs from the Proline-rich Membrane Anchor, PRiMA

The membrane-bound form of acetylcholinesterase (AChE) constitutes the major component of this enzyme in the mammalian brain. These molecules are hetero-oligomers, composed of four AChE catalytic subunits of type T (AChET), associated with a transmembrane protein of type 1, called PRiMA (proline-rich membrane anchor). PRiMA consists of a signal peptide, an extracellular domain that contains a proline-rich motif (14 prolines with an intervening leucine, P4LP10), a transmembrane domain, and a cytoplasmic domain. Expression of AChET subunits in transfected COS cells with a truncated PRiMA, without its transmembrane and cytoplasmic domains (Pstp54 mutant), produced secreted heteromeric complexes (T4-Pstp54), instead of membrane-bound tetramers. In this study, we used a series of deletions and point mutations to analyze the interaction between the extracellular domain of PRiMA and AChET subunits. We confirmed the importance of the polyproline stretches and defined a peptidic motif (RP4LP10RL), which induces the assembly and secretion of a heteromeric complex with four AChET subunits, nearly as efficiently as the entire extracellular domain of PRiMA. It is noteworthy that deletion of the N-terminal segment preceding the prolines had little effect. Interestingly, short PRiMA mutants, truncated within the proline-rich motif, reduced both cellular and secreted AChE activity, suggesting that their interaction with AChET subunits induces their intracellular degradation.

In the nervous tissue and muscles of mammals, acetylcholinesterase (AChE, 2 EC 3.1.1.7) controls cholinergic transmission by rapidly hydrolyzing the neurotransmitter acetylcholine after its release from nerve terminals. The functional localization of AChE depends on the association of its T splice variant with structural proteins (1)(2)(3)(4). Thus, the physiologically active AChE species correspond essentially to the collagen-tailed forms at the neuromuscular junctions and to membrane-bound tetramers in the brain.
The AChE T splice variant is characterized by its 40-residue C-terminal peptide (t peptide), which contains a C-terminal cysteine and seven aromatic residues, including three evenly spaced tryptophans, and can be organized as an amphiphilic ␣-helix (5). This peptide behaves as an autonomous interaction domain (the tryptophan (W) amphiphilic tetramerization domain (WAT)) (6); it allows oligomerization of AChE T subunits into homomeric dimers (T 2 ) and tetramers (T 4 ), as well as heteromeric associations of tetramers with anchoring proteins (7)(8)(9).
In the collagen-tailed forms, AChE T tetramers are associated with a specific collagen, called ColQ (7,10). This interaction has been extensively studied; it is based on a tight interaction between four t peptides (6) and a proline-rich motif, called PRAD ("proline-rich attachment domain"), located in the N-terminal noncollagenous region of ColQ (11). Synthetic t and PRAD peptides (40 and 15 residues, respectively) spontaneously form a complex, the structure of which has been determined by crystallography; four ␣-helical t peptides form a staggered coiled-coil around the PRAD, organized as an elongated polyproline II helix (12). All aromatic residues are oriented toward the interior of this compact cylindrical complex, and the tryptophans are apposed to the rings of the proline residues. From this interaction, it was possible to deduce models for the quaternary organization of four AChE T subunits linked to a ColQ chain (12,13).
The existence of an N-glycosylated 20-kDa hydrophobic protein, associated with membrane-bound AChE tetramers, was originally discovered in 1987 by Gennari et al. (14) and by Inestrosa et al. (15); this membrane anchor was more recently cloned and called PRiMA ("proline-rich membrane anchor"), because it contains a proline-rich motif, like ColQ (16). This suggests that AChE T subunits may associate with PRiMA and with ColQ in a similar manner. However, there are significant differences in the numbers of prolines (8 in ColQ and 14 in PRiMA) and in the numbers and positions of cysteines that could form intercatenary disulfide bonds with a cysteine * This work was supported by grants from the CNRS, the Association Française Contre les Myopathies, the Direction des Forces et de la Prospective, and the European Community. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom correspondence should be addressed. Tel.: 33-1-44-32-38-91; Fax: 33-1-44-32-38-87; E-mail: jean.massoulie@biologie.ens.fr. 2 The abbreviations used are: AChE, acetylcholinesterase; AChE T , splice variant T; PriMA, proline-rich membrane anchor; ColQ, collagen Q; ERAD, endoplasmic reticulum-associated degradation; PRAD, proline-rich attachment domain; P stpn , truncated PRiMA mutant, terminating with a stop codon at position n; ⌬ N -X-P stpn , PRiMA mutant with a deletion of the N-terminal fragment, starting with residue X before the prolines and terminating with a stop codon at position n; WAT, tryptophan (W) amphiphilic tetramerization domain.
located near the C terminus of each of the four t peptides. In addition, PRiMA contains a putative N-glycosylation site between the polyproline stretches and the transmembrane domain (17). Alternative splicing produces two PRiMA variants, which differ in their C-terminal domains as follows: the intracellular domains of the major variant (PRiMA I) and of the minor variant (PRiMA II) contain 40 and 11 residues, respectively (18). The mode of association between PRiMA and AChE T subunits is physiologically important because the resulting PRiMA-anchored AChE tetramers (16,18) represent the major enzyme species in the brain (18), and their level is regulated by exercise in muscles (19).
We have undertaken an analysis of the association of AChE T with PRiMA. In this study, we mostly used truncated PRiMA mutants containing only the extracellular domain or fragments of this domain, but not the transmembrane and cytoplasmic domains, thus producing soluble heteromeric complexes with AChE T subunits. In previous studies, we have shown that a significant fraction of AChE T subunits is degraded intracellularly, through the ERAD process ("endoplasmic reticulumassociated degradation") (20,21), and that this is mostly induced by exposed aromatic residues (21,22), in agreement with the fact that the formation of a complex in which these residues are occluded may reduce their degradation and increase their secretion. In this study, we examine how co-expression with PRiMA mutants affects the trafficking, degradation, and secretion of AChE T subunits. We show that some truncated mutants of PRiMA act as degradation inducers, assembling AChE T subunits into complexes that are degraded intracellularly, indicating that they fail to pass the quality control of the secretory pathway. By using deletions and mutations, we analyze the influence of residues flanking the polyproline stretches and also of the leucine located between the prolines, and we define a peptidic motif (RP 4 LP 10 RL), which is nearly as efficient as the complete extracellular domain of PRiMA for recruitment of AChE T subunits into secreted heteromeric complexes.

EXPERIMENTAL PROCEDURES
Vectors and Site-directed Mutagenesis-The AChE T subunits of rat AChE and intact or mutated mouse PRiMA were expressed by inserting the corresponding cDNAs into pEF-Bos vectors (23). Throughout this study, the numbering of PRiMA residues corresponds to the mature protein (Fig. 1A); the extracellular domain corresponds to residues 1-53. Mutagenesis was performed by the method of Kunkel et al. (24), as described previously (25). For deletions, we used mutagenic oligonucleotides of about 30 nucleotides containing 15 nucleotides on each side of the deleted fragment. Truncated mutants are indicated by the position of stop codons and by the modified residues; for example, R36E-P stp37 indicates a mutant in which a stop codon was introduced at position 37 and arginine 36 was replaced by a glutamic acid. Fig. 1B shows the structure of mutants used in this study. We introduced a FLAG epitope (DYKDE) after the cleavage site of the signal peptide in PRiMA, so that it was recognized by the anti-FLAG monoclonal antibody M1 (Sigma). We also used an N-terminal fragment of Torpedo ColQ, with or without its PRAD domain (Fig. 1C).
Transfection in COS Cells-Plasmids were transfected in COS cells with the DEAE-dextran method, as described previously (9), using 2 g of vector DNA encoding the catalytic subunit AChE T and various amounts of vector DNA encoding PRiMA mutants, as specified, per 60-mm dish. For comparison, we also used an N-terminal fragment of Torpedo marmorata ColQ (Q stp69 ). In each series of transfections, we completed the amount of vector encoding PRiMA mutants with a vector encoding a noninteracting protein, the N-terminal domain of ColQ from which the PRAD interaction motif was deleted (⌬(28 -44)-Q stp69 ) (11), so that the total amount of vector remained constant, to avoid changes in the synthetic capacity of the cells. After transfection, COS cells were incubated at 37°C, in a medium containing 10% NUserum (Inotech, Dottikon, Switzerland), which had been pretreated with 10 Ϫ5 M soman to inactivate serum cholinesterases. The medium and the cells were collected after 3-4 days.
Analysis of AChE Recovery after Irreversible Inhibition-The AChE activity of transfected cells (3 or 4 days after transfection) was irreversibly inhibited by incubation with the membranepermeant inhibitor soman (pinacolyl methylphosphonofluoridate) at 5 ϫ 10 Ϫ7 M for 30 min, as described previously (22). After extensive washing, the recovery of AChE activity was determined by collecting cells at various times in fresh culture medium at 37°C. Secretion of newly synthesized AChE only resumed after about 150 -180 min, and during that period the recovery of cellular activity reflected a balance between neosynthesis and intracellular degradation.
Cell Extracts-Intracellular and membrane-bound AChE was extracted for 15 min at 20°C in a TMg buffer (1% Triton X-100, 50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 ) containing 25 mM benzamidine, followed by centrifugation for 10 min at 13,000 rpm at 4°C. The culture medium containing the secreted enzyme was also centrifuged at 13,000 rpm for 10 min to remove cell debris before analysis.
Enzyme Assays-AChE activity was determined with the colorimetric method of Ellman et al. (26) at room temperature. The reaction was monitored at 414 nm with a Labsystems Multiskan RC automatic plate reader (Helsinki, Finland); the optical density was recorded at 20-s intervals over a period of 10 min. Alkaline phosphatase and ␤-galactosidase from Escherichia coli were assayed with the chromogenic substrates p-nitrophenyl phosphate and o-nitrophenyl galactoside, respectively.
Sedimentation and Electrophoresis Analyses-Centrifugation in 5-20% sucrose gradients (50 mM Tris-HCl, pH 7.5, 20 mM MgCl 2 , in the presence of 1% Brij-96 or 0.2% Triton X-100) was performed in a Beckman SW41 rotor, at 36,000 rpm, for 17 h 30 min at 6°C. Approximately 40 fractions were collected, and AChE activity was measured with the Ellman colorimetric assay, allowing the determination of the sedimentation coefficients of the different molecular forms, and of their relative activities. The gradients contained E. coli ␤-galactosidase (16 S) and alkaline phosphatase (6.1 S) as internal sedimentation standards. Amphiphilic molecules are characterized by the fact that they interact with detergent micelles; they generally sediment slower in the presence of Brij-96 than of Triton X-100.
Electrophoresis in nondenaturing polyacrylamide gels was performed as described by Bon et al. (27). The gels contained 0.25% Triton X-100 with or without 0.05% deoxycholate; electrophoresis was performed in a refrigerated apparatus under 40 V/cm for 2 h. Enzymatic activity was revealed by the histochemical method of Karnovsky and Roots (28). This method allows a rapid qualitative comparison of up to 20 samples in a single gel. In charge shift electrophoresis, the electrophoretic migration of amphiphilic molecules was accelerated in the presence of sodium deoxycholate, when compared with migration in the presence of the neutral detergent Triton X-100.
Effect of Synthetic Peptides on Oligomerization of AChE T Subunits-Synthetic peptides corresponding to the interaction motif of PRiMA (RP 4 LP 10 and RP 4 LP 10 RL) were synthesized by the Merrifield solid phase method in an Applied Biosystems 431A automated peptide synthesizer, with small scale 9-fluorenylmethoxycarbonyl (Fmoc) chemistry (29). Protected amino acids were purchased from Applied Biosystems (Foster City, CA) (30). Crude peptides were purified by reverse phase high pressure liquid chromatography on a Vydac C18 column (5 m, 250 ϫ 10 mm 2 ), using appropriate acetonitrile gradients containing 0.1% trifluoroacetic acid.
A 10 Ϫ2 M solution of peptide in Tris-HCl, pH 8 (1 M), was added with fresh medium to transfected COS cells expressing only AChE T subunits, at a final concentration of 10 Ϫ4 M. Peptides were also added to cell homogenates, at a final concentration of 10 Ϫ4 M, and incubated overnight at 20 or 37°C. In some samples, a synthetic t peptide (WAT) was added at 2 ϫ 10 Ϫ4 M. The oligomeric state of AChE was then analyzed by nondenaturing electrophoresis, after dilution 2-fold in buffer containing 1% Triton X-100.

RESULTS
PRiMA Mutants- Fig. 1A shows the primary sequence of mouse PRiMA and the structure of mutants used in this study. PRiMA consists of a signal peptide (residues Ϫ35 to Ϫ1), an extracellular N-terminal domain (residues 1-53) that includes four cysteines, a proline-rich motif and a putative N-glycosylation site, a transmembrane domain (residues 54 -78), and two possible cytoplasmic C-terminal domains that define PRiMA I and PRiMA II. We studied the interaction of AChE T subunits with the extracellular domain of PRiMA, using mutants lacking the transmembrane and cytoplasmic domains (Fig. 1B). First, we introduced stop codons at various positions (n), producing proteins P stpn . Second, we deleted most of the N-terminal region preceding the prolines, in mutants designed as ⌬ N ; to maintain a correct signal peptide cleavage site, we conserved the first or the last residue of this region (Glu-1 and Arg-20, respectively). We also assessed the possible influence of electrostatic effects, by mutating the residue preceding the prolines by other charged or uncharged residues (Asp, Lys, or Ala). Finally, we analyzed the role of the leucine located within the polyproline region (Leu-25), by replacing it with a proline, and by adding a leucine at position 31, with or without the original In the extracellular domain, the putative N-glycosylation site is indicated by an asterisk and the cysteines are underlined. B, structure of mutants. All mutants possess the signal peptide (not shown) and fragments of the extracellular domain but not the transmembrane and cytoplasmic domains. They are designated by the position of introduced stop codons (e.g. P stp54 contains the entire extracellular domain, residues 1-53). The deletion of the region preceding the polyproline stretches is indicated by ⌬. In these mutants, a single residue precedes the prolines, so as to maintain a proper cleavage site after the signal peptide. This residue is indicated, e.g. arginine (R) in ⌬ N -R-P stp54 . Mutants in the proline-rich region are indicated by the distribution of prolines and leucine residues (e.g. P 4 LP 10 in the case of the wild type). The interaction motif (RP 4 LP 10 RL) corresponds to mutant ⌬ N -R-P stp38 . C, structure of the N-terminal fragment of Torpedo ColQ (Q stp69 ) (the signal peptide is not shown). The underlined fragment corresponds to the PRAD interaction domain. It was deleted in ⌬[28 -44]Q stp69 . FEBRUARY 9, 2007 • VOLUME 282 • NUMBER 6 JOURNAL OF BIOLOGICAL CHEMISTRY 3489 leucine. Thus the original motif P 4 LP 10 was changed to P 15 , P 10 LP 4 , and P 4 LP 5 LP 4 .

Assembly of AChE Tetramers by PRiMA
Formation and Secretion of Soluble Complexes with Truncated PRiMA Mutants, Lacking the Transmembrane Domain-PRiMA and its different mutants were expressed in COS cells together with AChE T subunits (also called T subunits). To evaluate the formation of heteromeric complexes, we analyzed the total AChE activity of the cellular extracts and of the medium, as well as their composition in molecular forms, using both sedimentation and nondenaturing electrophoresis. It should be noted that all molecular forms of AChE, which differ in their degree of oligomerization, with or without associated proteins, possess the same catalytic activity per active site, so that the level of AChE activity reflects the number of AChE T subunits, independently of the composition in molecular forms. Fig. 2 illustrates the molecular forms obtained by expressing AChE T subunits alone, with full-length PRiMA, and with its extracellular fragment, P stp54 . As a control, we used an N-terminal fragment of ColQ that contains the PRAD (residues 28 -44), without the collagenous and trimerization domains (Q stp69 ) (Fig. 1C). When expressed alone or in the presence of a noninteracting fragment of ColQ from which the proline-rich motif was deleted (⌬(28 -44)-Q stp69 ), T subunits produced mostly amphiphilic monomers (T 1 ) and dimers (T 2 ) (27), with a small proportion of nonamphiphilic tetramers sedimenting at 10.5 S, which are thought to be homomeric (T 4 ), i.e. with no associated endogenous noncatalytic component ( Fig. 2A). Monomers and dimers are not readily resolved in the sedimentation profiles, and we therefore quantified their sum (T 1 ϩ T 2 ). The proportion of T 4 tetramers is markedly higher in the medium than in the cell extract, indicating that they are secreted more efficiently than monomers and dimers, as observed previously (9).
Co-expression of T subunits with full-length PRiMA induced the formation of membrane-bound T 4 -PRiMA complexes, expressed at the cell surface as demonstrated by immunofluorescence (not shown). These heteromeric complexes were recovered in the cell extract after solubilization in the presence of 1% Triton X-100; unlike homomeric tetramers (T4), which sediment at 10.5 S regardless of the presence or absence of detergents, the PRiMA-associated tetramers are amphiphilic, and their sedimentation is influenced by detergents: they sediment at 9.8 S in the presence of Triton X-100 and 9 S in the presence of Brij-96. This was also shown by the fact that they FIGURE 2. Molecular forms of AChE produced and secreted by transient transfection of COS cells with AChE T expressed alone, with full-length PRiMA, and with truncated PRiMA, reduced to its extracellular domain (P stp54 ). A, representative sedimentation profiles, obtained from gradients containing 1% Brij-96, normalized to the same total AChE activity, indicating the relative proportions of monomers (T 1 ), dimers (T 2 ), and tetramers, associated or not with PRiMA or P stp54 . Profiles corresponding to cell extracts are shown in the top panel, and to the medium in the middle panel. E, AChE T expressed alone; f, AChE T expressed with PRiMA; OE, AChE T expressed with P stp54 . In general, the relative proportion in the medium, compared with cell extracts, increases with the degree of oligomerization, indicating that secretion is facilitated by oligomerization. Soluble nonamphiphilic tetramers (T 4 ) and heteromeric complexes (T 4 -P stp54 ) sediment around 10.5 S, whereas amphiphilic T 4 -PRiMA complexes, containing the transmembrane domain of PRiMA, are retarded by their interaction with the detergent (Brij-96) and sediment around 9 S. B, migration of AChE forms from cell extracts and culture media in nondenaturing electrophoresis. In the profiles corresponding to cotransfected cells (AChE T with PRiMA and AChE T with the extracellular domain of PRiMA, P stp54 ), the simple arrows indicate nonamphiphilic AChE tetrameric forms; the arrow with an asterisk indicates the amphiphilic tetramer associated with a complete PRiMA protein. c indicates cellular extract; m indicates medium. migrated more slowly than T 4 tetramers in nondenaturing electrophoresis (Fig. 2B). In addition, the medium contained an increased level of soluble molecules sedimenting at 10.5 S, suggesting that they correspond to nonamphiphilic tetramers. When AChE T subunits were co-expressed with an N-flagged PRiMA, these secreted tetrameric molecules were recognized by an anti-FLAG antibody (M1), as shown by retardation of their migration in nondenaturing electrophoresis (Fig. 3), demonstrating that they contain an N-terminal fragment of PRiMA that might be produced by proteolysis, either intracellularly or at the cell surface.
Co-expression of AChE T subunits with the P stp54 mutant, corresponding to the extracellular domain of PRiMA, induced the formation of a soluble AChE form, sedimenting at 10.5 S, and migrating as an AChE tetramer in nondenaturing electrophoresis (Fig. 2, A and B). This form, which could not be distinguished from homomeric T 4 tetramers by sedimentation, corresponds to AChE T tetramers associated with the P stp54 protein (T 4 -P stp54 ). The fact that the ratio of T 4 -P stp54 to T 1 ϩ T 2 is higher in the medium than in the cell extract shows that they are preferentially secreted, as observed for homomeric tetramers T 4 .
The formation of soluble T 4 -P stp54 complexes demonstrates that the amphiphilic character of T 4 -PRiMA complexes is due to the transmembrane domain of PRiMA and confirms that AChE T subunits interact with a peptidic motif located in the N-terminal extracellular region of PRiMA, in agreement with a previous study (16).
Effect of Progressive C-terminal Deletions in the Extracellular Domain of PRiMA on the Recruitment of AChE T Subunits-We examined the interaction between AChE T subunits and PRiMA mutants in which stop codons were introduced upstream of position 54: P stp46 , P stp43 , P stp39 , P stp38 , P stp37 , P stp36 , P stp33 , and P stp31 (see Fig. 1B). We studied the AChE activity and molecular forms produced in COS cells expressing a constant amount of AChE T subunits (2 g of vector DNA per 60-mm culture dish), together with varying amounts of each mutant. The proportion of AChE T tetramers increased in the cells and in the medium with the amount of associated noncatalytic protein, as illustrated for P stp33 , P stp54 , and Q stp69 in Fig. 4. This proportion was found to plateau at similar values for P stp54 and Q stp69 but at a much lower level for P stp33 . Fig. 5 illustrates the effect of the P stpn mutants on the cellular activity and on the rate of secretion, with a fixed amount of plasmid encoding the P stpn mutants (2 g of DNA encoding AChE T and PRiMA mutant, each, per 60-mm culture dish). The level of cellular activity  transfected with a fixed amount of DNA encoding AChE T subunits (2 g per 60-mm culture dish) and the indicated amount of DNA encoding PRiMA mutants, or the N-terminal region of ColQ (Q stp69 ), complemented with DNA encoding a noninteracting protein, so that the total amount of vector was kept constant to avoid modifications in the protein synthesis capacity of the cells. The proportion of tetramers, associated or not with P stpn proteins (T 4 ϩ T 4 -P stpn ) was determined from sedimentation profiles, as in Fig. 2. The figure illustrates the results obtained for P stp31 and P stp54 mutants, producing high or low proportions, respectively, of T 4 -P stpn complexes. The curves obtained with all other mutants were very similar in shape, differing mostly by the maximal levels of tetramers; co-transfection with 2 g of DNA/dish encoding the various mutants was used in most subsequent experiments.

Assembly of AChE Tetramers by PRiMA
varied very little when AChE T subunits were co-transfected with the various PRiMA mutants. Secretion increased by about 30% when AChE T subunits were co-transfected with Q stp69 , indicating that they were partially rescued from intracellular degradation by the formation of T 4 -Q stp69 complexes (not shown), but reduced by about 50% when they were co-transfected with P stp31 . The secreted activity gradually increased with the longer mutants P stp33 and P stp36 , reaching the approximate value obtained with AChE T alone for longer constructs.
The level of secreted T 1 ϩ T 2 was markedly reduced when AChE T subunits were co-transfected with any of the truncated P stpn mutants, as well as with Q stp69 , indicating that they interacted with all mutants, even the shorter ones, such as P stp33 or P stp31 , which produced only minimal levels of cellular or secreted T 4 -P stpn form. The variations observed in the level of secreted activity appeared correlated with the proportion of heteromeric complex in the medium, as shown in Fig. 5C. Remarkably, this correlation includes all the truncated PRiMA mutants analyzed in this study, and also Q stp69 . Fig. 5D shows that the secretion of T 4 -P stpn complexes appears proportional to their cellular activity, except for T 4 -P stp36 , which was more efficiently secreted, and for mutants containing Ala, Asp, or Glu residues preceding the proline-rich segment, which are less efficiently secreted.
The fact that AChE activity was reduced, both in the cells and in the medium, when AChE T subunits were co-expressed with short truncated PRiMA mutants such as P stp31 suggests that they were more degraded than when expressed alone. This was verified by following the initial rate of recovery of AChE activity, during the 2 h after irreversible inhibition, i.e. before secretion of the newly synthesized enzyme; this rate represents the balance between neosynthesis and intracellular degradation. The rate of neosynthesis must be identical when AChE T subunits are expressed with a noninteracting protein and with different PRiMA mutants. We found that the rates of recovery varied in the order AChE T ϩ ⌬(28 -44)-Q stp69 Ͼ AChE T ϩ P stp46 Ͼ AChE T ϩ P stp33 , indicating that P stp33 , and to a lesser degree P stp46 , induced some degradation of newly synthesized active AChE T subunits.
The ratio of secreted to the cellular activity of each molecular form of AChE can be considered as an index of its secretability; Fig. 6 shows that this ratio remained essentially constant for monomers and dimers (T 1 ϩ T 2 ), but varied markedly for AChE tetramers associated with P stpn proteins. This indicates that the secretion of the complexes depended on the length of the PRiMA fragment, especially between P stp31 and P stp36 , confirming that complexes formed with the shorter PRiMA mutants were not efficiently secreted. In addition, the fact that this ratio varied in a nonmonotonic manner as a function of the length of the mutants from 36 to 54 suggests that the C-terminal residues of the P stpn proteins may either facilitate or reduce secretion.
The P stp38 mutant seemed to produce a maximum of secreted heteromeric complexes, whereas P stp36 appeared opti- FIGURE 5. Effect of the PRiMA mutants on cellular and secreted AChE activity and molecular forms. COS cells were co-transfected with 2 g of DNA encoding AChE T and 2 g of DNA encoding PRiMA mutants. A, sequence and numbering of the extracellular domain of PRiMA; the cysteines are underlined and the putative N-glycosylation site is indicated by an asterisk. B, the cellular and secreted AChE activities corresponding to the different mutants were normalized, for each experiment, to those of AChE T subunits expressed alone, taken as 100%. The proportions of the molecular forms were determined from sedimentation profiles. The data corresponding to C-terminal truncated mutants 1-9 are the means of 4 -7 independent transfection experiments, Ϯ S.E.; those for other mutants correspond to a representative experiment. The lightly hatched boxes correspond to the activity of monomers and dimers (T 1 ϩ T 2 ), and the darkly hatched boxes to tetramers (homomeric tetramers T 4 and heteromeric T 4 -P stpn complexes). If homomeric T 4 tetramers are present in the same proportion to T 1 and T 2 , as in the case of AChE T subunits alone, they represent a minor species and can be generally neglected. In mutants 16 -18, the deletion of the N-terminal region is associated with mutations in the proline-rich segment, as indicated in parentheses. Note that a mutant reduced to the RP 4 LP 10 RL motif (expressed as mutant ⌬ N -R-P stp38 ) induced the production and secretion of essentially the same amount of complex as the complete extracellular domain (P stp54 mutant). C, correlation between the total secreted AChE activity and the proportion of tetramers in the medium. The numbers identify the mutants, as indicated in the left column of B. D, correlation between the percentages of T 4 complexes in secreted and cellular AChE. FIGURE 6. Ratio of secreted to cellular activity of AChE forms. The secreted activity reflects the rate of secretion, whereas the cellular activity, 3-4 days after transfection, approximately corresponds to a steady state value; therefore, these activities are not comparable, but their ratio may be considered as an index of the efficiency of secretion for each AChE form. These ratios were arbitrarily normalized to 100 for AChE tetramers produced in co-transfection with the complete N-terminal domain of PRiMA, P stp54 . For monomers and dimers (T 1 ϩ T 2 ), the ratio remained approximately constant, in agreement with the fact that their structure did not change. FEBRUARY 9, 2007 • VOLUME 282 • NUMBER 6 mal for the secretability of such complexes. The differences observed between P stp36 , P stp37 , P stp38 , and P stp39 show that the residues located immediately downstream of the prolines (RLL) influence the formation and the secretion of the T 4 -P stpn complexes. However, replacement of Arg by an alanine or a glutamic acid in mutant P stp37 (R36A-P stp37 and R36E-P stp37 ) or replacement of RLL by alanines in mutant (RLL/AAA-P stp39 ) had essentially no effect on recruitment of AChE T subunits, indicating that the charge of these residues is not important.

Assembly of AChE Tetramers by PRiMA
It is interesting that P stp46 and P stp54 produced very similar results, because P stp54 possesses a putative N-glycosylation site and a cysteine that are removed in P stp46 , suggesting that these two elements do not interfere, positively or negatively, with the interaction between PRiMA and AChE T subunits in transfected COS cells. The lack of influence of these elements was confirmed by mutating Cys-48 to a serine in P stp54 and by suppressing the glycosylation site by replacing Thr-46 by an alanine; these mutations did not affect the production and secretion of T 4 -P stpn complexes (not shown).
Influence of the Region Preceding the Polyproline Stretches-To assess the possible role of the peptidic region that precedes the polyproline stretches of PRiMA, we deleted it in P stp46 and P stp54 . To maintain a proper cleavage site between the signal peptide and the mature protein, the prolines stretches were preceded by an arginine (as in the wild type) or by other charged or neutral residues (Lys, Glu, Asp, and Ala), as shown in Fig. 1B.
With an arginine preceding the prolines, we found that deletion of the N-terminal fragment (residues 1-19) from mutants P stp46 or P stp54 (⌬ N -R-P stp46 and ⌬ N -R-P stp54 ) slightly reduced the amount of cellular and secreted complexes (Fig. 5B). The fact that the two deleted mutants produced the same effect confirms that residues 45-53 do not influence the assembly or the trafficking of the complexes.
We then examined a possible effect of a charged or neutral residue preceding the prolines, in mutants ⌬ N -R-P stp46 , ⌬ N -K-P stp46 , ⌬ N -A-P stp46 , ⌬ N -D-P stp46 , and ⌬ N -E-P stp46 (Fig. 5B and Fig. 6). We found that the amount of secreted AChE tetramers is higher with a neutral or acidic residue but that their secretability appears lower, as indicated by the ratio of secreted to cellular activities, suggesting that this residue influences the efficiency of secretion (Fig. 5D).
It was surprising that the secretion of AChE complexes was lower with a positively charged residue upstream of the prolines, because this residue is an arginine in the wild type PRiMA. However, when we replaced this arginine by a glutamic acid in P stp46 , we found no modification of the yield of T 4 -P stp46 complexes, indicating that this charge effect is negligible when the N-terminal segment (residues 1-19) is present (not shown).
Mutations in the Proline-rich Region-PRiMA contains two groups of successive prolines, separated by a leucine residue (P 4 LP 10 ). To determine whether this leucine is required for the interaction with AChE T subunits, we mutated it to a proline, starting from the ⌬ N -E-P stp46 mutant. The production and secretion of AChE tetramers remained unchanged with the resulting mutant, noted ⌬ N -(P 15 )-P stp46 . However, the production of complexes was significantly reduced by displacement of the leucine to a symmetrical position in the proline stretches (P 10 LP 4 ) and even more by addition of a second leucine (P 4 LP 5 LP 4 ), as shown in Fig. 5B. These mutations had little effect on the secretability of the complexes (Fig. 6). Thus, the recruitment of AChE T subunits into heteromeric T 4 -P stp46 complexes is weakened by an interruption of the proline stretches by a leucine, and this suggests that a succession of 10 prolines is sufficient to ensure a maximal efficiency.
A Proline-rich Peptidic Motif, Sufficient for Recruitment of AChE T Subunits-In the preceding sections, we showed that it is possible to maintain an efficient recruitment of AChE T subunits into T 4 -P stpn hetero-oligomers after deletion of the N-terminal region of PRiMA preceding Arg-20 and of the C-terminal region following Leu-37. Therefore, the 20 -37 peptidic motif, which contains the polyproline stretches (RP 4 LP 10 RL), may be sufficient for this interaction. To verify this conclusion, we combined N-terminal and C-terminal deletions in the ⌬ N -R-P stp38 mutant. This mutant was co-expressed with AChE T subunits, and compared with P stp54 , ⌬ N -R-P stp38 , and P stp38 , all expressed at 2 g of vector/dish (Fig. 5). All mutants induced the production of T 4 -P stpn , confirming the effects observed with separate deletions of the N-terminal and C-terminal regions. Thus, the 18-residue peptide 20 -37, derived from PRiMA, possesses the capacity to assemble four AChE T subunits into stable T 4 -P stpn complexes in a manner similar to that of the full-length extracellular domain of PRiMA (P stp54 ).
It is noteworthy that deletion of the N-terminal region, in ⌬ N -R-P stp54 and ⌬ N -R-P stp38 , increased the ratio of secreted to cellular T 4 -P stpn by about 25%, compared with the corresponding mutants P stp54 and P stp38 , suggesting that secretion was facilitated by the N-terminal deletion. Deletion of the C-terminal region, in P stp38 and ⌬ N -R-P stp38 , compared with P stp54 and ⌬ N -R-P stp54 , also appeared to increase secretability.
Assembly of AChE T Tetramers by Synthetic Peptides-We synthesized the PRiMA-derived peptides RP 4 LP 10 and RP 4 LP 10 RL, to determine whether they could induce the assembly of AChE T monomers and dimers into tetramers. In a first series of experiments, the peptides were added at a final concentration of 10 Ϫ4 M to COS cells expressing only AChE T subunits, during the transfection process and/or in the culture medium. The cells and the medium were analyzed by nondenaturing electrophoresis after 3 days, as indicated above. This showed a clear increase in the proportion of AChE T tetramers, most probably associated with the peptides (Fig. 7A). This result indicates that the peptides were able to penetrate the biosynthetic compartments of the living cells.
In a second series of experiments, homogenates of COS cells expressing only AChE T subunits were prepared without detergent and incubated overnight at 37°C with the peptides (10 Ϫ4 M). As shown in Fig. 7B, both peptides induced the assembly of tetramers, under these acellular conditions. When a synthetic t peptide (WAT), at 2 ϫ 10 Ϫ4 M, was added to the cell homogenate together with the PRiMA-derived peptides, it reduced markedly the recruitment of AChE T monomers and dimers into tetramers, indicating a direct interaction between these peptides.

DISCUSSION
The N-terminal Extracellular Domain of PRiMA Assembles AChE T Subunits into Soluble T 4 -P stpn Complexes-The assembly of AChE T tetramers with the transmembrane protein PRiMA produces membrane-bound complexes, both in the nervous system (14 -18, 31) and in muscles (19). In addition, the availability of PRiMA seems to limit the stabilization as well as the functional localization of AChE in the mammalian brain (18). It is therefore important to analyze the interaction between PRiMA and AChE T subunits. In this study, we attempted to define the peptidic motif of PRiMA that is responsible for its quaternary association with AChE T subunits, by using deletions upstream, downstream, and within the prolinerich segment.
We mostly used truncated PRiMA mutants, lacking the transmembrane and cytoplasmic domains (P stpn with a stop codon at position n), which were co-expressed with AChE T subunits in transiently transfected COS cells, producing soluble complexes (T 4 -P stpn ), which were secreted instead of remaining attached to the cell membrane.
Truncated PRiMA Mutants Recruit AChE T Subunits into Heteromeric Complexes, Following Either Secretion or Degradation Pathways-We have shown in a previous study that a significant fraction of AChE T subunits is degraded intracellularly through the ERAD process, whereas another fraction is secreted (20,21). Therefore, the level of intracellular AChE activity reflects a balance between the rate of synthesis and the combined rates of degradation and secretion.
When COS cells were transfected with a fixed amount of plasmid encoding AChE T subunits, together with various amounts of plasmid encoding PRiMA mutants, a variable frac-tion of the subunits was integrated in T 4 -P stpn complexes, both in the cells and in the medium. It was surprising to find that although the proportion of T 4 -P stpn complexes varied widely, it appeared to plateau for similar amounts of plasmid DNA, suggesting that all mutants presented a similar apparent "affinity" for AChE T subunits. This paradox can be resolved by assuming that even when the observed proportion of complexes was very low, mutants such as P stp31 did in fact interact with AChE T subunits, as indicated by a marked reduction in the secretion of T 1 ϩT 2 , but produced rapidly degraded complexes. With longer mutants, the level of secreted T 4 -P stpn complex progressively increased. Thus, the variations observed in the amount of secreted complex appear to reflect a competition between secretion and degradation. Depending on its organization, each type of T 4 -P stpn complex would be oriented with different probabilities toward the secretion pathway or the degradation pathway. This suggests that the rate of secretion depends critically on the structure of the complex.
It is a classical notion that proteins destined to function as a multisubunit complex may be retained in the endoplasmic reticulum by the quality control system of the cell, unless they are appropriately associated with their partners (32). This can be rationalized by the fact that they expose hydrophobic surfaces in their unassembled state but not in the complex. This seems to be the case for Torpedo AChE T subunits, which are secreted very little when expressed alone in COS cells but can be secreted when co-expressed with an N-terminal fragment of ColQ (Q stp69 ) (5). We also observed here that rat AChE T subunits can be rescued by Q stp69 from degradation through ERAD (20). This is consistent with the fact that their physiological function depends on association with an anchoring protein, ColQ or PRiMA.
The fact that short truncated PRiMA mutants increase the intracellular degradation of AChE T subunits is more unusual. However, Ulloa-Aguirre et al. (33) recently suggested that some proteins induce an inappropriate conformation of normally externalized molecules and thus target them toward degradation by the cellular quality control and proposed to call this hypothetical process "protein shipwrecking" (33). Here we present evidence that some PRiMA mutants may behave in this manner, and that there is in fact a continuous spectrum between mutants that induce degradation and those that rescue AChE T subunits from ERAD and promote their secretion as heteromeric complexes.
Because the assembly of a complex between proteins occludes interaction surfaces, rather than exposing hydrophobic zones, how could the association with short PRiMA mutants such as P stp31 induce the degradation of AChE T subunits? Although we have presently no experimental evidence, we may propose a hypothetical explanation. We have demonstrated previously that aromatic (but not hydrophobic) residues of the C-terminal t peptide of AChE play a crucial role to induce the degradation of AChE T subunits by ERAD (22). It is possible that a group of aromatic residues remains exposed in the complexes formed with short PRiMA mutants, engaging their translocation into the cytoplasm and degradation by proteasomes, whereas these residues are masked in complexes formed FIGURE 7. Synthetic PRiMA peptides induce oligomerization of AChE T subunits in vitro. A, COS cells were transfected with AChE T , and the synthetic peptides RP 4 LP 10 and RP 4 LP 10 RL were added with fresh culture medium after 2 days. The cells were extracted 24 h later and analyzed by nondenaturing electrophoresis, as well as the medium. B, peptides were added at a concentration of 10 Ϫ4 M to an homogenate of COS cells expressing only AChE T subunits, and incubated overnight at 37°C. In some samples, a synthetic t peptide (WAT) was added at 2 10 Ϫ4 M, as indicated. The homogenate was then analyzed by nondenaturing electrophoresis in the presence of detergents (see "Experimental Procedures"). The peptides induced the formation of stable AChE T tetramers (indicated by arrows); this effect was reduced by addition of the t peptide (WAT).
with longer mutants, as demonstrated by crystallography in a complex of four t peptides with the PRAD of ColQ (12).
Definition of a Sufficient Interaction Motif (PRAD) in PRiMA-The efficiency of secretion of AChE oligomers increases with their degree of oligomerization (9). The secretability of heteromeric T 4 -P stpn complexes, indicated by the ratio of secreted to cellular activity, was variable; it is noteworthy that it was increased by deletion of peptidic regions located upstream or downstream of the polyproline stretches. A maximum was obtained by removing all residues following the prolines (P stp36 ); conversely, it was reduced for mutants in which the N-terminal region was deleted (⌬ N ) and the proline-rich segment was preceded by residues Ala, Glu, or Asp, as opposed to basic residues Arg or Lys.
Most of the N-terminal and C-terminal regions flanking the polyproline stretches could be deleted without suppressing the capacity of P mutants to engage AChE T subunits into T 4 -P stpn complexes. This capacity in fact appeared maximal when most of the C-terminal region was deleted, leaving only two residues (RL) after the prolines, in P stp73 . However, it is not certain that these residues play a specific role in the association with t peptides, because replacement of the three residues following the prolines by alanines had essentially no effect (in RLL/AAA-P stp39 versus P stp39 ). Even though we have no direct evidence that P stp54 is N-glycosylated under our experimental conditions, we found that the presence of a putative N-glycosylation site, located at position 44, did not influence the association of PRiMA mutants with AChE T subunits. Similarly, the cysteine located at position 48 did not affect the formation of T 4 -P stpn complexes.
The PRAD of ColQ is normally disulfide-linked to two of the t peptides, while the other two are disulfide-linked together (5). However, we found previously that the formation of heteromeric complexes with the N-terminal region of ColQ could occur without the two adjacent cysteines preceding the PRAD, indicating that disulfide bonds between ColQ and two AChE T subunits are not required (5,11). The present results suggest a similar conclusion in the case of PRiMA, because the formation of complexes was only slightly reduced (20% or less) by deletion of most of the N-terminal region, which contains four cysteines, located at positions 6, 13, 17, and 19. This does not rule out the possibility that some of these cysteines may be involved in intercatenary disulfide bonds between PRiMA and AChE T subunits, as shown in membrane-anchored AChE tetramers from bovine brain (14,15,34), and thus stabilize the complex.
PRiMA contains two polyproline stretches, separated by a leucine (P 4 LP 10 ). We found that this leucine is not required for the interaction and could be replaced by a proline, resulting in a single suite of 15 prolines (P 15 ). The secretion of T 4 -P stpn complexes was decreased by displacement of the leucine (P 10 LP 4 ) or interruption of the proline stretches by a second leucine (P 4 LP 5 LP 4 ), and it was almost totally abolished when 5 prolines were deleted in P stp31 (P 4 LP 5 ). Thus, the formation of secretable complexes requires the presence of a minimal number of uninterrupted prolines.
Our deletion experiments converged on the idea that the 20 -37 segment of PRiMA (RP 4 LP 10 RL) may be able to recruit AChE T subunits nearly as efficiently as the entire extracellular domain of PRiMA, and this was confirmed by expressing this 18-residue peptide as mutant ⌬ N -R-P stp38 . In fact, addition of the synthetic peptides RP 4 LP 10 and RP 4 LP 10 RL to COS cells expressing only AChE T subunits induced the recruitment of monomers and dimers into stable tetramers. Moreover, we obtained the same result by incubating a cell homogenate with the peptides, and we observed that the assembly of AChE T tetramers was partially blocked by addition of a synthetic t peptide (WAT). This proves that the PRiMA-derived and t peptides spontaneously interact in vitro, as shown previously for a ColQderived PRAD peptide (12), and opens the way to the formation, crystallization, and structural analysis of the core complex that forms the basis of the association between PRiMA and AChE T . It is noteworthy that the PRAD of ColQ is very efficient with only 8 prolines (P 3 MFP 5 ). This raises the intriguing possibility that the complexes of t peptides with the proline-rich motifs of ColQ and PRiMA might significantly differ in their quaternary organization. Using short PRiMA-derived peptides, it may also be possible to shed some light on the structural differences between complexes that pass or fail the quality control in the secretory pathway.