Formation of Crystalloid Endoplasmic Reticulum Induced by Expression of Synaptotagmin Lacking the Conserved WH X L Motif in the C Terminus

Synaptotagmin (Syt) is a family of type I membrane proteins that consists of a single transmembrane domain, a spacer domain, two Ca 2 (cid:1) -binding C2 domains, and a short C terminus. We recently showed that deletion of the short C terminus (17 amino acids) of Syt IV prevented the Golgi localization of Syt IV proteins in PC12 cells and induced granular structures of various sizes in the cell body by an unknown mechanism (Fukuda, M., Ibata, K., and Mikoshiba, K. (2001) J. Neu-rochem. 77, 730–740). In this study we showed by electron microscopy that these structures are crystalloid endoplasmic reticulum (ER), analyzed the mechanism of its induction, and demonstrated that: ( a ) mutation or deletion of the evolutionarily conserved WH X L motif in the C terminus of the synaptotagmin family (Syt (cid:2) C) destabilizes the C2B domain structure ( i.e. causes misfolding the C terminus of Syts I–XI is essential for plasma membrane association in PC12 cells and that the WH X L motif of Syt I regulates synaptic vesicle docking to the presynaptic plasma membrane in the squid giant synapse (5). In this study we report a novel role of the WH X L motif of the synaptotagmin family in proper folding of the molecules. We have shown by deletion and mutation analyses that the WH X L motif is essential for correct folding of

Synaptotagmin I (Syt I) 1 is an integral membrane protein of synaptic vesicles and a key molecule that regulates various steps of the synaptic vesicle cycle. It consists of a short intravesicular N terminus, a single transmembrane domain, a spacer domain, a C2A domain, a C2B domain, and a short C terminus (reviewed in Refs. [1][2][3][4]. Each domain of Syt I has been shown to be crucial to expression of its functions and to have a distinct role in the synaptic vesicle cycle. For instance, the short C terminus (WHXL motif) of Syt I is involved in synaptic vesicle docking to the presynaptic plasma membrane, probably by binding to certain plasma membrane receptors, such as neurexins (5). The Ca 2ϩ binding activity of the C2A domain is crucial for synaptic vesicle fusion to the presynaptic plasma membrane, because antibodies to the C2A domain that inhibits Ca 2ϩ /phospholipid or Ca 2ϩ /syntaxin binding to it interfere with the fusion step of secretory vesicles (6 -11). The Syt I C2A domain is also important for neurite outgrowth and the axonal repair process (12)(13)(14)(15). The C2B domain is involved in recycling of synaptic vesicles, probably by binding to the clathrin assembly protein AP-2 (16 -18). The Ca 2ϩ -dependent selfoligomerization activity of the C2B domain has also been suggested to be involved in the fusion step of synaptic vesicles (19 -22), because a Tyr-312 to Asn substitution in the C2B domain of Syt II, corresponding to the Drosophila AD3 mutation, impairs Ca 2ϩ -dependent self-oligomerization activity (20,21), and inositol 1,3,4,5-tetrakisphosphate, a strong blocker of neurotransmitter release, inhibits Ca 2ϩ -dependent self-oligomerization activity of the C2B domain in vitro (20,(23)(24)(25). In addition, the phenotypes of Syt I mutants of Drosophila, Caenorhabditis elegans, and mice strongly support the above multiple functions of Syt I in the synaptic vesicle cycle (26 -32).
Synaptotagmins are now known to form a large family of integral membrane proteins in both vertebrates and invertebrates (1,33,34), and it has been suggested that they regulate vesicular trafficking, because Syts II-XIII basically share the same domain structure with Syt I (i.e. a transmembrane domain, spacer domain, two C2 domains, and short C terminus). One of these domains, the N-terminal region of the C terminus, is highly conserved across phyla as well as different isoforms (Syts I-XIII), and it has been suggested to have functions common to the synaptotagmin family (5,35). In our previous study, we showed that the WHXL motif in the C terminus of Syts I-XI is essential for plasma membrane association in PC12 cells and that the WHXL motif of Syt I regulates synaptic vesicle docking to the presynaptic plasma membrane in the squid giant synapse (5). In this study we report a novel role of the WHXL motif of the synaptotagmin family in proper folding of the molecules. We have shown by deletion and mutation analyses that the WHXL motif is essential for correct folding of the C2B domain (i.e. the WHXL motif is located in the ␤-8 strand of the C2B domain rather than being an independent domain; see Fig. 3A). Expression of synaptotagmin mutants lacking the WHXL motif in cultured cells induced crystalloid endoplasmic reticulum (ER) structures (Refs. 36 and 37, and references therein) due to the abnormal folding of the proteins. Based on these results, we discuss the structural importance of the ␤-8 strand of the C2 domains of synaptotagmins.

EXPERIMENTAL PROCEDURES
Materials-Recombinant Taq DNA polymerase and restriction enzymes were obtained from Toyobo Biochemicals (Tokyo, Japan). Mouse monoclonal antibody (M2) and rabbit polyclonal antibody against FLAG peptide were obtained from Sigma Chemical Co. (St. Louis, MO) and Zymed Laboratories Inc. (San Francisco, CA), respectively. Horseradish peroxidase (HRP)-conjugated anti-T7 tag antibody and anti-T7 tag antibody-conjugated agarose were from Novagen (Madison, WI). HRPconjugated anti-FLAG tag antibody was from Sigma. Anti-TGN38 antibody was from Transduction Laboratories (Lexington, KY). Anti-BiP (immunoglobulin binding protein, also called Grp78, a 78-kDa glucoseregulated protein) antibody was from StressGen Biotechnologies Corp. (Victoria, BC, Canada). All other chemicals were commercial products of reagent grade. Solutions were made up in deionized water prepared with an Elix10 water purification system and Milli-Q Biocel A10 system (Millipore Corp., Bedford, MA).
Electron Microscopy-PC12 cells cultured on collagen-coated plastic coverslips (Sumitomo Bakelite, Tokyo, Japan) were fixed with 2.5% glutaraldehyde in 0.1 M sodium phosphate buffer (pH 7.4) for 2 h. After washing in the buffer, cells were post-fixed in 1% OsO 4 in 0.1 M sodium phosphate buffer (pH 7.4) for 1 h. The cells were then washed in distilled water, incubated with 50% ethanol for 10 min, and stained with 2% uranyl acetate in 70% ethanol for 2 h. The cells were further dehydrated with a graded series of ethanol and embedded in epoxy resin. Ultra-thin sections were doubly stained with uranyl acetate and lead citrate and then examined under a Hitachi H7000 electron microscope.
Endoglycosidase H Digestion of T7-N-Gly-synaptotagmins-PC12 cells (4 ϫ 10 6 cells, the day before transfection) were cultured on 10-cm collagen type I-coated dishes (Becton Dickinson) in Dulbecco's modified Eagle's medium containing 10% horse serum and 10% fetal bovine serum at 37°C under 5% CO 2 . Transfection was achieved by using the LipofectAMINE Plus reagent as described above. Immunoprecipitation of T7-Syts by anti-T7 tag antibody-conjugated agarose was performed as described previously (38,39,45). The beads were resuspended in 30 l of the buffer (50 mM sodium acetate, pH 5.3, 0.1% SDS, 0.5% IGEPAL CA-630 (Sigma Chemical Co.), 0.1 mM phenylmethylsulfonyl fluoride, 10 M leupeptin, and 10 M pepstatin A) and then divided into two microtubes. After denaturation by boiling for 3 min and cooling to 37°C, 5 milliunits of endoglycosidase H (Roche Diagnostics, Mannheim, Germany) was added to one tube, and the mixtures were incubated for 1 h at 37°C (41). Reactions were stopped by adding SDS sample buffer and boiling for 3 min. Proteins were subjected to 10% SDS-PAGE and immunoblotting with HRP-conjugated anti-T7 tag antibody as described previously (38 -40).
Trypsin Digestion of Recombinant C2B Domain of Synaptotagmin IV-Syt IV-C2B⌬423-425 and mutant Syt IV-C2B⌬423-425(AAA) fused to glutathione S-transferase (GST) were prepared by a standard method (24,46), and after incubating 4 g of GST fusion proteins in 10 mM Tris-HCl, pH 7.5, and 100 mM NaCl with 1 ng/l trypsin (Sigma Chemical Co.) for 0 -30 min at 25°C, the reactions were stopped by addition of SDS sample buffer and boiling for 3 min. Samples were then analyzed by 12.5% SDS-PAGE and staining with Coomassie Brilliant Blue R-250.

Induction of Crystalloid ER by Expression of Synaptotagmin
IV Lacking the WHXL Motif-In the course of determining the Golgi localization signal of Syt IV in PC12 cells, we found that the deletion of the short C terminus (amino acid residues 409 -425) prevented the Golgi localization of Syt IV and induced large granular structures of various sizes (Ref. 41 and Fig. 1, A and C). Interestingly, these granular structures in which Syt IV⌬C proteins had accumulated were easily identified on lightfield micrographs (arrowheads in Fig. 1D). Because they were also positive with anti-BiP antibody, a marker for the ER, we suspected that they represented crystalloid ER (36, 37; see also Fig. 1, E-G, arrows). Electron microscopic observation of the cells expressing Syt IV⌬C revealed a unique membranous structure characteristic to "crystalloid ER" (36,37), which is composed of tightly packed smooth ER cisternae (Fig. 3H, arrowheads) or tubules (Fig. 3H, arrows). The crystalloid ER was occasionally continuous to rough ER cisternae (Fig. 3H, small arrows).
To identify the amino acid residues responsible for the induction of crystalloid ER by expression of Syt IV⌬C, we constructed various additional C-terminal deletion mutants ( Fig.  2A). After deletion of more than seven amino acids from the C terminus (FLAG-Syt IV⌬419 -425), we observed mis-sorting of Syt IV mutant proteins and formation of crystalloid ER of various sizes within the cell body (Fig. 2C, ⌬414 -425, ⌬419 -425 in green and TGN38 in red), but deletion of three amino acids did not affect the Golgi localization of Syt IV (Fig. 2C, ⌬423-425, arrowheads). When the conserved WHML (amino acids 419 -422) sequences were mutated to AAMA, expression of FLAG-Syt IV⌬423-425(AAA) induced granule formation similar to that of Syt IV⌬C (Fig. 2C). Thus, the WHXL sequences are in some way related to the formation of crystalloid ER and the Golgi localization of Syt IV. Because deletion of the whole C2B domain (FLAG-Syt IV⌬C2B) attenuated the crystalloid ER formation induced by deletion of the C terminus (⌬C) and Syt IV⌬C2B proteins were localized in the Golgi (Fig. 2C,  arrowhead in center panel), the C2B domain was further systematically deleted from the C terminus, and the putative effector domain of the C2B domain was deleted or mutated to clarify the relationship between the C2B domain and the WHXL sequences (Fig. 2C) (25,47,48). These mutations and deletions, however, failed to restore the Golgi localization of Syt IV⌬C proteins (Fig. 2C, ⌬329 -425, ⌬361-425, ⌬C(341-347), and ⌬C(KA) in green and TGN38 in red).
The Formation of Crystalloid ER Structures by Expression of Other Synaptotagmins Lacking the WHXL Motif-The WHXL motif is highly conserved among the C-terminal type tandem C2 proteins, including the synaptotagmin family, rabphilin-3A, the Doc2 family, and the Slp family (synaptotagmin-like protein) (35,49,50) and was proposed to be an independent domain (4), because the short C terminus of these proteins interacts with polymorphic plasma membrane receptors, neurexins, in vitro (33,49,(51)(52)(53)(54). However, recent demonstration of the three-dimensional structure of the C2B domain of Syt III (55) and rabphilin-3A (56) has provided strong evidence against this notion, and the WHXL motif must be a part of the C2B domain rather than an independent domain. The C2B domain of Syt III is composed of an eight-stranded anti-parallel ␤-sandwich structure consisting of four-stranded ␤-sheets (anti-parallel pairing of ␤-1 and ␤-8, ␤-2 and ␤-5, ␤-3 and ␤-4, and ␤-6 and ␤-7 strands), the same as the Syt I C2A domain (reviewed in Ref. 57; see also Fig. 3A). The major difference between the C2A and C2B domains is that the C2B domain contains an additional ␣-helix region between the ␤-7 and ␤-8 strands (Fig.  3A). It should be noted that this ␣-helix region was previously thought to be a ␤-8 strand of the C2B domain (Fig. 3B, broken arrow) (55,56). Based on these findings, together with the fact that crystalloid ER is induced by expression of malfolded proteins (58), we hypothesized that deletion of a portion of the ␤-8 strand (the WHXL motif) of the C2B domain destabilizes the C2B domain structure (i.e. the C2B domain lacking the WHXL motif cannot fold correctly) and induces the crystalloid ER structures.
If this hypothesis were true, deletion of the ␤-8 strand of the C2B domain of other synaptotagmin isoforms should induce the same crystalloid ER structures, because the WHXL motif is highly conserved in the synaptotagmin family (Fig. 3B). As expected, expression of Syts I-XI⌬C proteins also induced crystalloid ER-like structures in PC12 cells (Fig. 4A, arrowheads and data not shown), although the appearance of the crystalloid ER-like structures differs according to the Syt isoform (5-50% of total cells). Interestingly, expression of an Syt I⌬C mutant that lacks oligomerization activity ( Fig. 4B; Syt I⌬spacer/⌬C) (59) failed to induce crystalloid ER-like structures, despite the absence of the WHXL motif (Fig. 4A, right  upper panel).
Next, we focused on the ␤-1 strand of the C2B domain of Syt IV, because the N-terminal ␤-1 strand and C-terminal ␤-8 Syt IV (A, B) and -Syt IV⌬C (C-H) proteins were expressed in PC12 cells as described under "Experimental Procedures." PC12 cells were fixed, permeabilized, and stained with anti-FLAG rabbit antibody (green in A, C, E, and G) and anti-BiP antibody (red in F and G). B and D are light-field micrographs of A and C, respectively. The inset in D is a superposition of C and D. Note that the granules could be detected by light-field microscopy (arrowheads in D) and that they colocalized with BiP (arrows in E-G). H, electron micrograph of PC12 cells expressing FLAG-Syt IV⌬C proteins. The crystalloid ER is composed of tightly packed smooth ER cisternae (arrowheads in H) or tubule (arrows in H). The crystalloid ER was occasionally continuous to rough ER cisternae (small arrows in H). DCV, dense-core vesicle; RER, rough ER; M, mitochondria. Scale bars in G and H indicate 20 m and 500 nm, respectively. strand pair to form anti-parallel ␤-sheets (Fig. 3A). Deletion of the ␤-1 strand of the C2B domain of Syt IV (C2B-⌬␤1, Fig. 5) or substitution of AAA for the conserved G(E/D)(L/I) sequence of the ␤-1 strand of Syt IV (C2B-␤1A3) induced crystalloid ERlike structures, the same as Syt IV⌬C, suggesting that pairing of the ␤-1 and ␤-8 strands is important for proper folding of the C2B domain. These results, together with the fact that the basic structure of both the C2A and C2B domains is the same (55)(56)(57), prompted to us to examine whether deletion of the ␤-8 strand of the C2A domain would also induce crystalloid ER. As expected, expression of Syt IV(C2A)⌬C proteins induced the same crystalloid ER-like structures (Fig. 5B, top panel).

FIG. 1. Deletion of the short C terminus prevents Golgi localization of synaptotagmin IV and induces crystalloid ER. FLAG-
Incomplete Post-translational Modifications of Syt ⌬C Proteins in PC12 Cells-Vertebrate Syts I and II contain a single N-glycosylation site in the N-terminal intravesicular domain (24,39,41,51,60), and the N-liked sugars are further converted to complex oligosaccharides in the Golgi of PC12 cells (41). Because the dense-core vesicles where Syt I is abundant are thought to be derived from the TGN (trans-Golgi network) in PC12 cells, we decided to investigate whether Syt I⌬C proteins pass through the Golgi and TGN or are only retained in the ER. To do so, we used endoglycosidase H, which cleaves unmodified N-linked sugars but not complex oligosaccharides (41). The N-linked sugars of the wild-type T7-Syt I, -Syt II, -N-Gly-Syt VIII, and -N-Gly-Syt IX in PC12 cells were further modified and were resistant to endoglycosidase H treatment, whereas the mutant proteins (Syt ⌬C) were mostly sensitive to endoglycosidase H (Fig. 6, A and B). By contrast, the N-linked sugars of both the wild-type and mutant proteins were cleaved by N-glycosidase F, which cleaves all N-linked sugars, regardless of modifications (data not shown). These results indicate that the newly synthesized Syt ⌬C proteins had been correctly inserted into the ER membrane and had been retained in the ER without being transported to the Golgi, where complex modification occurs.
Essential Role of the WHXL Motif in the C2B Domain Structure-Finally, we investigated whether the WHXL motif is essential for the proper folding of the C2B domain biochemically by monitoring its sensitivity to proteolysis. If the above hypothesis were correct, mutation of the WHXL motif to AAXA would cause it to fail to form a stable C2 domain structure by disrupting the pairing of the ␤-1 and ␤-8 strands, and the mutant proteins should be more sensitive to proteinase than the wild-type proteins. When the GST-Syt IV-C2B⌬423-425 proteins (including the WHXL motif) were treated with low concentrations of trypsin, they were rapidly cut into two pieces at the fusion site between GST (open arrowheads) and Syt IV-C2B (asterisk in Fig. 7, upper panel). The C2B⌬423-425 proteins (including the WHXL motif) were highly resistant during 30 min of incubation at this concentration of trypsin, whereas the mutant C2B domains were degraded within 10 min. Thus, the WHXL motif was indeed found to be essential for the packed C2B domain structure. DISCUSSION We previously found that deletion of 17 amino acids C-terminal of Syt IV induced granular structures and prevented the Golgi localization of Syt IV (41). In this study we showed by electron microscopy that this granular structure is crystalloid ER (36, 37) and analyzed the mechanism of its induction by expression of Syts I-XI lacking the conserved WHXL motif in the C terminus. Crystalloid ER was found to be induced by The arrows represent eight strands of the C2B domain (␤1-␤8). Three loops are formed at the tip of the ␤-sandwich structure, and two of them are involved in Ca 2ϩ -binding in the C2A domain of Syt I (57). The overall structure of the C2B domain is quite similar to the C2A domain, but an additional ␣-helix is present between the ␤-7 and ␤-8 strands in the C2B domain (55,56). Note that the N-terminal ␤-1 strand and C-terminal ␤-8 strand form a pair to create an anti-parallel ␤-sheet. B, sequence alignment of the ␤-1 and ␤-8 strands (solid arrows) and the ␣-helix (broken line) of the Syts I-XIII C2B domains (33,38,55). Asterisks indicate the conserved WHXL motif (5). The number signs indicate the positions of amino acid substitutions (GEL-to-AAA) in the Syt IV(C2B-␤1A3) mutant (see also Fig. 5A). Note that the ␣-helix region was previously thought to be a ␤-8 strand of the C2B domain by analogy with the C2A domain (␤8Ј, broken arrow at the bottom) (57).

FIG. 4. Induction of crystalloid ER-like structures by C-terminal truncation of the synaptotagmin family.
A, FLAG-Syt I⌬C, -Syt I⌬spacer/⌬C, -Syt VI⌬C, and -Syt XI⌬C were expressed in PC12 cells as described under "Experimental Procedures." PC12 cells were fixed, permeabilized, and stained with anti-FLAG antibody. Note that the crystalloid ER-like structures were induced by expression of Syt I⌬C, Syt VI⌬C, and Syt XI⌬C (arrowheads), but not Syt I⌬spacer/⌬C, which did not exhibit Ca 2ϩ -independent oligomerization activity (59) (see also B). Scale bar indicates 20 m. B, oligomerization activity of Syt I⌬C (or Syt I⌬spacer/⌬C). pEF-T7-Syt I and FLAG-Syt I⌬C (or -Syt I⌬spacer/⌬C) were co-transfected into COS-7 cells. Association between T7-Syt I and FLAG-Syt I⌬C (or -Syt I⌬spacer/⌬C) was evaluated by immunoprecipitation with anti-T7 tag antibody-conjugated agarose (IP) as described previously (38,45).  (Fig. 7). Because anti-parallel pairing of other ␤-sheets occurred between adjacent ␤-3 and ␤-4 or ␤-6 and ␤-7 strands (57), we speculate that mutations or deletions in the ␤-sheets formed by the N-terminal ␤-1 strand and the C-terminal ␤-8 strand add a most striking effect to C2 domain structure. (b) The resulting malfolded proteins are correctly inserted into the ER (showing type I membrane topology) but then are not transported to the Golgi (Fig. 6) and accumulate in the ER. Accumulation of the malfolded proteins in the ER induces the expression of BiP, one of the ER stress proteins, which are known to be induced by accumulation of malfolded proteins and to be present in stable complexes in the malfolded proteins (61,62). Consistent with this, Syt IV⌬C proteins colocalized with BiP in the crystalloid ER (Fig. 1, E-G, arrows). Because the cytoplasmic domain of Syts I-XI lacking the WHXL motif (Syt-cyto⌬C) was expressed in the cytosol, and its expression could not induce crystalloid ER (5,43), malfolded proteins seem to be recognized only in the ER. (c) We previously proposed that crystalloid ER is formed by head-to-head association of malfolded molecules on apposed smooth ER membranes based on the results of an experiment using rat liver microsomal aldehyde dehydrogenase (37). It should be noted that the synaptotagmin family also shows Ca 2ϩ -independent oligomerization activity via around the spacer domain and/or fatty-acylated Cys cluster at the interface between the transmembrane and spacer domains, and disulfide bonding at the extracellular domain (17, 38 -40, 59). Indeed, the expression of Syt I⌬spacer/⌬C proteins, which lack oligomerization activity, could not induce crystalloid ER (Fig.  4). Therefore, it is most likely that the ERs in which Syt ⌬C proteins have accumulated associate with each other to form crystalloid ER as a result of the oligomerizing capacity of the Syt family.
Because the WHXL motif was first proposed to be an independent domain based on sequence comparisons with the C2A domain structure (4), many in vitro binding experiments investigating the C2B domain, including our own, have involved the use of recombinant C2B domains lacking the WHXL motif (i.e. ␤-8 strand), as minimum C2B domains (25,48,63,64,66). Accordingly, some in vitro binding experiments on the C2B domain should be re-evaluated in terms of ligand binding specificity or ligand affinity. Indeed, we found that the C2B domain T7-Syts were expressed in PC12 cells and immunoprecipitated by anti-T7 tag antibody-conjugated agarose as described previously (38,45). Immunoprecipitants were exposed to or not exposed to glycosidase, subjected to 10% SDS-PAGE, and then analyzed by immunoblotting with HRP-conjugated anti-T7 tag antibody. Note that Syts I and II were resistant to endoglycosidase H (broken boxed), but Syts I⌬C and II⌬C were highly sensitive to endoglycosidase H. lacking the WHXL motif showed weaker Ca 2ϩ -dependent oligomerization activity in Syt VII than the wild-type protein (45) and found distinct inositol polyphosphate binding specificity in Syt II (24). Based on these findings, we suggest that differences in the length of the C terminus of the recombinant C2B domain, in addition to the impurity of recombinant proteins (67), are another important factor responsible for the contradictory results obtained in regard to the in vitro binding properties of the C2B domain (48, 65, 68 -70). Krasnov and Enikolopov (35) recently reported that mutation or deletion of the C terminus of Syt II prevented it from being transported to the tips of neurites where dense-core vesicles are abundant in nerve growth factor-differentiated PC12 cells, and they proposed that the WHXL motif is an active signal for targeting to the tips of neurites. However, our finding that the malfolded Syt ⌬C proteins lacking the WHXL motif are retained in the ER provides strong evidence against this idea. We suggest that Syt II⌬C proteins accumulate in the ER, are not transported to the Golgi, and, consequently, are not present in the dense-core vesicles derived from the TGN.
In conclusion, we have demonstrated that the WHXL motif functions as a protein interaction site (e.g. neurexins) as well as being important for proper folding of the C2B domain to stabilize anti-parallel ␤-sheet formation between ␤-1 and ␤-8 strands. We propose that recombinant C2B domains that contain the entire C terminus should be used for in vitro binding experiments to avoid abnormal folding of the C2B domain, which is only evident in the ER of living cells.