Abstract
Saccharomyces cerevisiae cells contain two homologues of the mammalian t-SNARE protein SNAP-25, encoded by the SEC9 and SPO20 genes. Although both gene products participate in post-Golgi vesicle fusion events, they cannot substitute for one another; Sec9p is active primarily in vegetative cells while Spo20p functions only during sporulation. We have investigated the basis for the developmental stage-specific differences in the function of these two proteins. Localization of the other plasma membrane SNARE subunits, Ssop and Sncp, in sporulating cells suggests that these proteins act in conjunction with Spo20p in the formation of the prospore membrane. In vitro binding studies demonstrate that, like Sec9p, Spo20p binds specifically to the t-SNARE Sso1p and, once bound to Sso1p, can complex with the v-SNARE Snc2p. Therefore, Sec9p and Spo20p interact with the same binding partners, but developmental conditions appear to favor the assembly of complexes with Spo20p in sporulating cells. Analysis of chimeric Sec9p/Spo20p molecules indicates that regions in both the SNAP-25 domain and the unique N terminus of Spo20p are required for activity during sporulation. Additionally, the N terminus of Spo20p is inhibitory in vegetative cells. Deletion studies indicate that activation and inhibition are separable functions of the Spo20p N terminus. Our results reveal an additional layer of regulation of the SNARE complex, which is necessary only in sporulating cells.
RANSPORT through the secretory pathway in eukaryotic cells requires an ordered series of vesicular budding and fusion events to move proteins and lipids between membrane-bound compartments. Fusion between vesicles and target membranes must be strictly regulated so that a given vesicle fuses only with the appropriate compartment. This specificity is achieved, at least in part, by the use of compartment-specific, soluble NSF attachment protein receptor (SNARE) complexes at each step in the secretory pathway (Pelham 1999). SNARE complexes are formed by the oligomerization of a membrane protein of the vesicle (v-SNARE) with a similar membrane protein on the vesicle's target membrane (t-SNARE). Specific interactions between a v-SNARE and its cognate t-SNARE are thought to provide much of the specificity for vesicle fusion (Sollneret al. 1993; Rothman and Warren 1994; Sogaardet al. 1994). Formation of the SNARE complex is critical for the fusion of a vesicle with the membrane and, in fact, the SNARE complex appears to be the core fusion machinery (Weberet al. 1998). The crystal structure of one SNARE complex has been solved (Suttonet al. 1998). This complex, which consists of the v-SNARE synaptobrevin and the t-SNARE subunits syntaxin (an integral membrane protein) and SNAP-25 (a peripheral membrane protein), is required for the fusion of synaptic vesicles in neurotransmitter release. In the assembled complex, the three proteins form a heterotrimer in which one helix from synaptobrevin, one helix from syntaxin, and two helices from SNAP-25 intertwine in a parallel arrangement to form a four-helix bundle (Suttonet al. 1998).
In the budding yeast Saccharomyces cerevisiae, homologues of neuronal syntaxin, synaptobrevin, and SNAP-25 are required for the fusion of post-Golgi secretory vesicles with the plasma membrane (Brennwaldet al. 1994). The synaptobrevin homologue, Sncp, is encoded by the redundant genes SNC1 and SNC2 (Protopopovet al. 1993). The syntaxin homologue Ssop is similarly encoded by a pair of redundant genes, SSO1 and SSO2 (Aaltoet al. 1993). Finally, the product of the SEC9 gene is the S. cerevisiae homologue of SNAP-25 (Brennwaldet al. 1994). These three proteins constitute the v- and t-SNAREs for secretory vesicle fusion in yeast and assemble into a four-helix bundle in a similar all-parallel arrangement to that found in the neuronal SNARE complex (Katzet al. 1998). The products of at least 10 other genes, the late-acting SEC genes, are also required for secretory vesicle fusion in S. cerevisiae (Novicket al. 1981). These genes may act upstream of the Ssop/Sncp/Sec9p SNARE in the pathway of secretory vesicle fusion (Guoet al. 1999).
In addition to their function in exocytosis in vegetatively growing yeast cells, many of the late-acting SEC genes are required for the formation of spores (Neiman 1998). In response to starvation, MATa/MATα diploid cells of S. cerevisiae enter a developmental program in which they undergo meiosis and differentiate into haploid spores. The process of sporulation involves the de novo generation of intracellular membranes, called prospore membranes, that encapsulate each of the daughter nuclei, forming daughter cells (Moens 1971; Moens and Rapport 1971; Byers 1981). These prospore membranes are created by the fusion of secretory vesicles within the cytoplasm (Neiman 1998). Several of the late-acting SEC genes that regulate formation of the Sncp/Ssop/Sec9p SNARE are required for secretory vesicle fusion with the prospore membrane. Interestingly, SEC9 itself is not required for this vesicle fusion event. This lack of requirement for SEC9 is due to the sporulation-specific induction of a second SNAP-25 homologue, termed SPO20 (Neiman 1998).
Sec9p and Spo20p have distinct functions, though there is evidence for some functional overlap between the two gene products during sporulation. In a sec9 mutant, sporulation is normal, but in a spo20Δ mutant, the prospore membranes form abnormally and fail to encapsulate the daughter nuclei. Redundancy between SEC9 and SPO20 is indicated by the observation that in a double mutant strain no prospore membranes are formed (Neiman 1998). However, the amount of functional overlap between the two gene products is limited. Overexpression of SEC9 in a spo20Δ mutant cannot rescue the sporulation defect. Conversely, though SPO20 obviates a requirement for SEC9 during sporulation, expression of SPO20 in vegetative cells cannot rescue the lethality of a sec9 mutation. Thus, the ability of each protein to function is limited to a specific developmental stage of the S. cerevisiae life cycle.
In this article, we examine the basis for the developmental stage-specific functions of Sec9p and Spo20p. In vitro binding studies and localization of the Sso and Snc proteins in sporulating cells suggest that Spo20p assembles into SNARE complexes with the same proteins as Sec9p. We also describe the use of chimeric molecules to define domains of Sec9p and Spo20p required for function during specific developmental stages.
MATERIALS AND METHODS
Strains and media: Unless otherwise noted, standard media and genetic methods were used (Roseet al. 1990). To construct strain AN147, a haploid segregant from strain MCY3259 (Tuet al. 1996) was crossed with strain S2683 (Neiman 1998). This diploid was sporulated and dissected to generate strains AN117-4B (MATα his3 ura3 trp1-hisG leu2 arg4-NspI lys2 hoΔ::LYS2 rme1::LEU2) and AN117-16D (MATa his3 ura3 trp1-hisG leu2 lys2 hoΔ::LYS2). These two strains were mated to give strain AN120. To create the spo20 deletion strain AN147, the oligonucleotides ANO134 (5′-TTT TTT TAT TAC TTT AGG TTT TTC CGT TTG TGA ATA GCC ATT TTT AGA TAT ATA CCC GGG CTG CAG GAA TTC) and ANO135 (5′-TTA GAA ACT ATT TAT TCA ATA TAT TTA TAC ACG ATA TTT TGT GTG TAT AAC AGA GTC GAG GGT ATC GAT AAG) were used to amplify the Schizosaccharomyces pombe his5+ gene (Wachet al. 1997). The PCR product was transformed into AN117-4B, introducing a precise deletion of the SPO20 gene and generating strain AN1052. AN1052 was then crossed to AN117-16D, dissected, and segregants were crossed to create AN147 (MATa/MATα his3/his3 ura3/ura3 trp1-hisG/trp1-hisG leu2/leu2 lys2/lys2 arg4-NspI/ARG4 rme1::LEU2/RME1 hoΔ:: LYS2/hoΔLYS2 spo20Δ::his5+/spo20Δ::his5+). BY41 (MATa sec9-4 ura3-52) is a derivative of NY57 from P. Novick's strain collection.
Sporulation assays: Strains to be tested were grown over-night in YPD medium. Cells (1.5 ml) were then pelleted, washed once in 1 ml 2% KOAc, and then resuspended in 10 ml 2% KOAc. Sporulation was assayed after 2 days of incubation in 2% KOAc by observation in the light microscope and by ether test. Ether tests were performed as follows: 5 μl of sporulating cells were spotted onto a YPD plate. The plate was inverted over a paper filter soaked with 2 ml of ethyl ether for 45 min and then removed and incubated at 30° for 24 hr before being photographed.
Immunofluorescence, antibodies, and Western blots: Immunofluorescence was performed as described (Neiman 1998). Affinity-purified anti-Sec9-NT, anti-Spo20p, anti-Ssop, and anti-Sncp antibodies were used at a 1:5 dilution (final concentration 10–40 μg/ml). The secondary antibody was goat anti-rabbit coupled to Cy3 (Cappel Laboratories, Malvern, PA) or coupled to Alexa 488 (Molecular Probes, Eugene, OR). Spr3p antibodies (Fareset al. 1996) were provided by J. Pringle (University of North Carolina, Chapel Hill). Affinity-purified Ssop and Sncp antibodies were prepared as described (Rossiet al. 1997).
Rabbit antisera were raised against three recombinant proteins containing (1) the first 150 amino acids of Sec9p (anti-Sec9NT), (2) the SNAP-25-like domain of Sec9p (anti-Sec9CT), or (3) the C-terminal 123 amino acids of Spo20p (anti-Spo20) fused to GST. Recombinant proteins were purified on glutathione agarose and eluted by boiling in boiling buffer (0.1% SDS, 10 mm Tris pH 7.5, 150 mm NaCl); amounts of protein were estimated by BCA protein assay (Pierce Chemical, Rockford, IL) as well as by SDS-PAGE and staining on Coomassie brilliant blue R (Sigma, St. Louis). Rabbit antisera production against each protein was carried out by Cocalico Biologicals (Reamstown, PA). For Western blotting, α-Sec9NT antibody was used at a 1:1000 dilution; α-Sec9CT antibody was affinity purified as described (Brennwaldet al. 1994) and used at a 1:200 dilution.
Samples were prepared for Western blotting as follows: cultures were grown to the midlog phase, 3 OD units of cells were harvested, washed in 10/10 buffer (10 mm Tris pH 7.5, 10 mm NaN3), and pellets were resuspended in 150 μl of 10/10 buffer. Glass beads (425–600 μm, acid-washed; Sigma) were added to 3/4 volume, and samples were vortexed and diluted in 450 μl of sample buffer. Samples were then run on SDS-PAGE and blotted to a nitrocellulose filter. After incubation with primary antibody, the filters were incubated with 125I-labeled protein A and the bands quantitated on a phosphoimager. To correct for variation in total protein between extracts, the level of each chimera was first normalized by comparison to the level of Sso1p in the same extract and that value was then expressed as a percentage of the level of the wild-type Sec9 protein.
Plasmids: For expression of the various chimeric molecules, the genes were cloned into one of two URA3 integrating vectors, containing either the SEC9 or SPO20 promoter. These expression vectors were constructed by cloning the upstream regions of the SEC9 gene [nucleotides (nt) −550 to −4] or the SPO20 gene (nt −750 to −4) into the polylinker of the plasmid pRS306 (Sikorski and Hieter 1989). The SEC9 upstream region was amplified using oligonucleotides ANO128 (5′-CCT TGA GGT ACC GGA TCC ATT ACG TAA TAA GTG) and ANO129 (5′-CCT TGA CTC GAG GTC AAT GGT GTA TTA TTC CG). The PCR fragment was digested with Asp718 and XhoI and then cloned into similarly digested pRS306 to create pRS306-SEC9pr. The SPO20 promoter was similarly amplified and subcloned using the oligonucleotides ANO130 (5′-CCT TGA GGT ACC AAG TCT AGG CGC TTT CAA CC) and ANO131 (5′-CCT TGA CTC GAG AAA AAT GGC TAT TCA CAA AC) to create pR306-SPO20pr. Both promoter regions were completely sequenced to ensure that no nucleotide changes were introduced by the PCR (data not shown).
The GST-Sec22 and GST-Sed5 expression plasmids contained full cytoplasmic domains of the proteins (Lupashin and Waters 1997) and were generously provided by G. Waters. GST-Vam3 cytoplasmic domain construct was a gift from T. Sato and S. Emr and GST-Pep12 (amino acids 76–273) was a gift from C. Burd.
Construction of chimeras: Chimeric molecules were constructed by overlapping PCR (Hortonet al. 1989; Yon and Fried 1989). As an example, to construct the chimera encoding a fusion of the Sec9p N terminus to the Spo20p C-terminal region (SPPP), the 5′ end of the Sec9p gene was first amplified with the Pwo polymerase (Boehringer Mannheim, Indianapolis) using the primers Sec9-+15 and S9NT/S20H1 (see Table 1 for primer sequences). The conditions used for all PCR reactions were 40 pmol each primer, 100 ng template, and 250 μm of each dNTP in a final volume of 100 μl. Amplification was for 25 cycles of 1 min 94°, 1 min 52°, and 2 min 72°. The S9NT/S20H1 primer has homology to the SEC9 N-terminal region at its 3′ end and SPO20 helix 1 at its 5′ end. The product of this amplification was purified using a QIAquick column (QIAGEN, Chatsworth, CA) and one-twentieth of the product was then included in a second PCR reaction using SPO20 as a template and the SEC9-+15 and SPO20-DS primers. These outside primers introduce an XhoI site three nucleotides 5′ of the start codon and a SacI site ~200 bp 3′ of the stop codon of the chimeric gene. The PCR product was purified, digested with XhoI and SacI, and cloned into similarly digested pRS306-SPO20pr or pRS306-SEC9pr. All the chimeras were constructed in a similar fashion, varying the primers and the templates used (see Table 1).
Chimeras with deletions of the Sec9p and Spo20p N termini were constructed as outlined in Table 1, using oligonucleotides ANO163 or ANO164 to remove amino acids 3–410 of Sec9p or amino acids 3–150 of Spo20p. For shorter deletions of SPO20 or the PSPS chimera, the oligonucleotides ANO175 (5′-CTT GTT CTC GAG ATA AGT GGG TCG CAT CAC AGT CGC CAC), ANO176 (5′-CTT GTT CTC GAG ATA ATG GGG CGC TAT GTA CTT ATT TCC), ANO168 (5′-CTT GTT CTG GAG ATA ATG GGG GAC AAT TGT TCA GGA AGC), or ANO165 (5′-CTT GTT CTC GAG ATA ATG GGG GTC TCT TAT GAG GTG CCC G) were used. These primers remove, respectively, amino acids 3–10, 3–29, 3–51, or 3–95 of Spo20p. The deletion genes were constructed by using oligonucleotides in a PCR reaction with the SPO20-DS or SEC9-DS oligonucleotide and using either the SPO20 gene or the PSPS chimera, respectively, as templates. Again, these PCR products were digested with XhoI and SacI and cloned into pRS306-SEC9pr or pRS306-SPO20pr for transformation and assay.
·In vitro· ·binding assay:· SEC9 and SPO20 sequences were placed under control of the T7 promoter by PCR. The primers T7SEC9 (5′-TAA TAC GAC TCA CTA TAG GGA GAC CAC ATG GAG GAG GCT CGC CAG CA) and S9-DS (5′-GCG TCA AGC TTG GGA TCC CGA AGG TAT TCT TTC AAT TCA C) were used to amplify SEC9. The resulting PCR product encodes amino acids 414–651 of Sec9p under control of the T7 promoter. Similarly, the primers T7SPO20 (5′-TAA TAC GAC TCA CTA TAG GGA GAC CAC ATG GAG AAT GTC CAG CCT GA) and Spo20-DS (5′-CAG TAG GAG CTC CGA TGT TTG AAT GCA C) were used to amplify a region encoding amino acids 161–397 of Spo20p and place them under T7 control. To in vitro translate constructs containing the N-terminal domains of SEC9 and SPO20, the following primers were used: S20-FLT7 (5′-TAA TAC GAC TCA CTA TAG GGA GAC CAC ATG GGG TTC AGA AAA ATA CTT GC) and S9-FLT7 (5′-TAA TAC GAC TCA CTA TAG GGA GAC CAC ATG GGA TTA AAG AAA TTT TTT AAG A).
The amplified DNA was added directly to a reticulocyte-coupled in vitro transcription-translation system (Promega, Madison, WI) in the presence of [35S]methionine and the resulting labeled protein products were used in the binding assay. Binding reactions were performed as described (Rossiet al. 1997). In all cases, 4 μl of the in vitro-translated proteins were added to the 100-μl binding reaction containing a final concentration of 1 μm of immobilized fusion proteins. For assays of ternary complex formation, 3 μm of the soluble Ssop1p was added to the binding reaction. Band volume was quantitated using the ImageQuant program (Molecular Dynamics, Sunnyvale, CA).
RESULTS
Spo20p and Sec9p localize to the prospore membrane: One possible explanation for the failure of SEC9 to rescue the spo20Δ sporulation defect is that Spo20p contains a specific signal promoting localization to the prospore membrane that Sec9p lacks. To test this possibility, the localization of both Sec9p and Spo20p in sporulating cells was examined (Figure 1). As observed previously for Sec9 in vegetatively growing cells, no signal was obtained when the proteins were expressed at endogenous, single-copy levels (Brennwaldet al. 1994). Likewise, no signal was detected for single-copy, endogenous Spo20. To obtain detectable amounts of these proteins, both genes were introduced on high-copy plasmids. Although both proteins display a diffuse cytosolic background staining, they nonetheless show significant localization to the prospore membrane. The similar localization of Spo20p and Sec9p in sporulating cells suggests that different subcellular localization of the two proteins is not likely to be the basis for their functional specificity.
Ssop and Sncp localize to the prospore membrane: A second possible determinant for the functional differences between SEC9 and SPO20 could be interactions with distinct partner SNARE molecules. The t-SNARE Ssop and the v-SNARE Sncp are known to form complexes with Sec9p that are required for vesicle fusion with the plasma membrane in vegetative cells (Brennwaldet al. 1994). If these proteins also act in concert with Spo20p in sporulating cells, their localization would be expected to change from the plasma membrane to the prospore membrane. To test this prediction, the localization of Ssop and Sncp in sporulating cells was examined. (Our antibodies do not distinguish between Sso1p and Sso2p or Snc1p and Snc2p; for simplicity, therefore, we refer to the pairs of proteins as Ssop and Sncp rather than Sso1/2p and Snc1/2p.) Ssop is localized to the prospore membrane in sporulating cells (Figure 2A). This result is consistent with the possibility that Ssop plays a role in vesicle fusion with the prospore membrane and is similar to the relocalization seen for other cell surface proteins (Neiman 1998). The Snc protein is also localized to the prospore membrane in sporulating cells (Figure 2E). Localization of the v-SNARE Sncp to this compartment is likely a consequence of the fusion of secretory vesicles with the prospore membrane and is consistent with a role for Snc proteins in vesicle fusion during prospore membrane formation.
Construction of chimeras: primers and templates
Localization of Spo20p and Sec9p in sporulating cells. Strain AN120 was transformed with high copy plasmids carrying either SPO20 (A and B) or SEC9 (C and D). These strains were then used for indirect immunofluorescence using antibodies against Spo20p (A) or the Sec9p N terminus (C) and counterstained with 4′,6-diamidino-2-phenylindole (DAPI) to visualize DNA as described in materials and methods.
·Sncp localization is disrupted in· ·spo20·Δ ·mutants:· In spo20Δ mutant cells, prospore membranes form abnormally such that daughter nuclei are not packaged, leading to the occurrence of anucleate spore bodies (Neiman 1998). To assess the role of Ssop and Sncp in the development of these abnormal prospore membranes, the localization of each of these proteins was examined in spo20Δ mutant cells. Sso protein localized to the prospore membrane in spo20Δ cells, as it did in wild-type cells (Figure 2C). However, although prospore membrane staining was seen with the anti-Ssop antibodies, quantitative analysis of the number of prospore membranes seen per cell revealed a defect in the spo20Δ mutant (Table 2). While 100% of the sporulating wild-type cells stained with the anti-Ssop antibody exhibited four prospore membranes, only 32% of the spo20Δ cells exhibited such a pattern. In fact, 24% of the spo20Δ cells exhibited no Ssop staining. This result could be due to either a failure of Ssop to localize to prospore membranes in spo20Δ cells or a failure of the spo20Δ mutant to form four prospore membranes per cell. We favor the latter possibility because a similar distribution of prospore membranes per cell was seen when the spo20Δ mutant strain was stained with antibodies to Spr3p (Fareset al. 1996), a cytosolic protein previously shown to localize to prospore membranes (data not shown). Thus, one result evident from this analysis is that the effect of spo20Δ on prospore membrane formation is more severe than originally suggested by analysis in the electron microscope (Neiman 1998).
Localization of Ssop and Sncp during sporulation in wild-type and spo20Δ cells. Strain AN120 (SOP20) or AN147 (spo20Δ) was sporulated, subjected to indirect immunofluorescence using anti-Ssop or -Sncp antibodies, and counterstained with DAPI to visualize DNA as described in materials and methods.
Distribution of Ssop and Sncp in proteins sporulating cells
An even more striking result was seen when localization of Sncp was examined (Figure 2G). Staining of prospore membranes with the anti-Snc antibodies is reduced, though not entirely lost, in spo20Δ mutant cells. As shown in Table 2, only 8% of the cells exhibit staining of four prospore membranes, and nearly 60% of the spo20Δ cells show no prospore membrane localization of the Snc protein. In these cells, Sncp displays a punctate distribution throughout the cytoplasm. The Ssop staining reveals that far more prospore membranes are being formed in the spo20Δ cells than are seen with the anti-Snc antibodies. Therefore, the Snc protein is failing to efficiently localize to many of the prospore membranes that form in spo20Δ. Because localization of Sncp to the prospore membrane is a consequence of secretory vesicle fusion, this result suggests that Sncp-dependent vesicles are not fusing efficiently in the spo20Δ mutant.
·Spo20p can bind to Ssop and Sncp· ·in vitro··:· The immunofluorescence data suggest that Spo20p, like Sec9p, may act in conjunction with Ssop and Sncp. This would presumably involve Spo20p forming heterotrimeric SNARE complexes with Ssop and Sncp in sporulating cells, similar to that formed between Sec9p and Ssop and Sncp in vegetative cells. To test this idea directly, we assessed the binding of Spo20p to Ssop and Sncp in vitro, using a binding assay described previously for Sec9p (Rossiet al. 1997). The SNAP-25 domains of Sec9p (amino acids 414–651) and Spo20p (amino acids 161–397) were produced in an in vitro translation system and tested for their ability to bind to various GST-fusion proteins. The ability of Sec9p or Spo20p to associate with the individual SNARE proteins or combinations of SNARE proteins was assessed by quantitation of the percentage of the total Sec9p or Spo20p that remained bound to the GST-fusion SNAREs on glutathione-agarose beads following extensive washing with binding buffer.
As observed previously (Rossiet al. 1997), Sec9p binds GST-Sso1p in vitro to form a binary complex (Figure 3A, top, lanes 5 and 6). Sec9p did not bind directly to GST-Snc2p. However, if soluble Sso1p (not fused to GST) is included in the binding reaction, either as the entire cytoplasmic domain (Sso1p1-265) or just the C-terminal helical region (Sso1p193-265), a ternary complex forms between Sec9p, Sso1p, and GST-Snc2p as shown by the precipitation of Sec9p with GST-Snc2p (Figure 3B, top, lanes 3–6). Recent reports have shown that SNARE molecules that function in different subcellular compartments readily associate in vitro, leading to the suggestion that there is little intrinsic specificity to SNARE-SNARE interactions (Fasshaueret al. 1999; Yanget al. 1999). By contrast, under our conditions, formation of the binary complex between Sso1p and Sec9p is specific in that Sec9p did not bind a GST fusion to the Golgi t-SNARE Sed5p (Hardwick and Pelham 1992), the endosomal t-SNARE Pep12p (Bechereret al. 1996), or the vacuolar t-SNARE Vam3p (Wadaet al. 1997; Figure 3A, top, lanes 7–12). A lack of specificity is seen in formation of the ternary complex, however, as Sec9p/Sso1p will bind to the ER-Golgi v-SNARE Sec22p (Dascheret al. 1991; Newmanet al. 1992) as readily as to Snc2p (Figure 3B, top, lanes 5–8). Therefore, while we see a high degree of specificity in binary t-SNARE formation between Sec9p and Sso1p, we also see noncognate, ternary SNARE complex formation with Sec22p in place of Snc1p.
Spo20p displayed nearly identical behavior as Sec9p in the binding assay (Figure 3, bottom). It readily formed binary complexes with GST-Sso1p but did not bind to GST-Snc2p. When untagged Sso1p was included in the binding reaction, ternary complexes between Spo20p, Sso1p, and GST-Snc2p were formed. Spo20p displayed specificity for Sso1p as it did not form complexes with Sed5p or Vam3p. The one marked difference between Spo20p and Sec9p is that Spo20p readily bound to Pep12p (Figure 3A, bottom, lanes 9 and 10), though Sec9p did not. The significance of this Spo20p-Pep12p interaction is unclear. Nonetheless, these results demonstrate that Spo20p and Sec9p are very similar in forming binary and ternary SNARE complexes with Ssop and Sncp. Further, taken with the immunofluorescence data (Figures 1 and 2), these results strongly suggest that during sporulation Spo20p binds to, and functions in conjunction with, the same SNARE proteins that Sec9p interacts with during exocytosis in vegetative cells.
Sec9p helical domains are required for function in vegetative cells: The Sec9p and Spo20p proteins share ~40% amino acid identity in their SNAP-25 domains (Neiman 1998) and both interact with the Sso and Snc proteins. However, as noted above, they are not interchangeable; function of Sec9p is largely limited to vegetative cells and Spo20p cannot replace Sec9p when ectopically expressed in vegetative cells (Neiman 1998). To gain insight into the structural basis of the developmental stage specificity of these two SNARE proteins, a series of chimeric molecules were constructed to define domains of Sec9p or Spo20p that confer stage-specific function.
Both proteins were divided into four domains (Figure 4): the nonconserved N-terminal regions, the first helix of the SNAP-25 domain, the interhelical region, and the second helix. Swaps were made at conserved amino acids near the ends of the helices, as predicted by structure prediction programs and alignment of these proteins with SNAP-25. Subsequent to this work, the crystal structure of the mammalian synaptic SNARE complex was published (Suttonet al. 1998). The swaps that we have constructed fall close to the boundaries predicted by comparison of the ends of the helices found in the crystal structure (Suttonet al. 1998). The swapped helical regions include six and eight amino acids, respectively, on the C-terminal and N-terminal side of helix 1 and eight amino acids to the N-terminal side of helix 2 that are not found in the crystal structure. These short stretches may not, therefore, participate directly in oligomer formation with Sncp and Ssop. Nonetheless, the crystal structure and more detailed sequence alignments (Weimbset al. 1997) indicate that the entire helical domains have been switched. After construction of the chimeras by overlapping PCR, all of the chimeric genes were cloned into integrating vectors under control of the SEC9 or the SPO20 promoter. The SEC9 promoter-driven genes were then transformed into strain BY41 (sec9-4) to assess vegetative function, and the SPO20 promoter-driven genes were transformed into AN147 (spo20Δ::his5/spo20Δ::his5) to assess function in sporulating cells. First, the effects in vegetative cells of replacing any individual domain of Sec9p with the corresponding region of Spo20p were examined (Table 3). For brevity, each domain in a chimeric gene is designated “S” if the sequence comes from SEC9 or “P” if the sequence comes from SPO20. The first letter indicates the N-terminal domain, the second letter helix 1, the third letter the interhelical domain, and the fourth letter helix 2. As controls, the SEC9 and SPO20 genes were amplified by PCR using the same flanking primers used for chimera construction (Table 1) and cloned into the pRS306-SEC9pr expression vector. When integrated in this expression vector, the SEC9 gene complements the sec9-4 defect (Table 3). Replacement of the N-terminal domain or the first helix of Sec9p (PSSS or SPSS) abolished Sec9p activity. Substitution of the second helix (SSSP) led to a protein with reduced function as indicated by weak growth at 37°. By contrast, replacement of the interhelical region of Sec9p with that from Spo20p (SSPS) had no obvious effect on activity of the protein. None of the chimeras containing more than one domain of Spo20p could rescue sec9-4 (Table 3).
Spo20p binds Sso1p and Snc2p. (A) Formation of binary complexes. (Top) 35S-labeled Sec9p was mixed with the indicated GST fusion proteins and precipitated on glutathione agarose as described in materials and methods. The percentage of input Sec9p found in the pellet fraction, as determined on a phosphorimager, is given. (Bottom) Identical experiments using 35S-labeled Spo20p. (B) Ternary complex formation. (Top) 35S-labeled Sec9p and soluble, recombinant Sso1p were mixed with GST-Snc2 or GST-Sec22 and precipitated as in A. (Bottom) Identical experiments using 35S-labeled Spo20p.
Alignment of Sec9 and Spo20 showing structure of the chimeras. Overline shows positions of the N-terminal, helix 1, interhelical, and helix 2 domains. Residues at which the swaps were constructed are shown in boldface. They are E421 (Sec9)/E166 (Spo20), A505/A250, and R569/R311. Underline denotes the regions of Sec9p and Spo20p that correspond to those found in the crystal structure of the neuronal SNARE complex (Suttonet al. 1998).
Rescue of· ·sec9-4· ·by Sec9p/Spo20p chimeras·
Expression of the chimeric proteins. Extracts of strain BY41 (sec9-4) expressing the indicated genes were prepared as described in materials and methods. (A) Western blot using antibodies directed at the N terminus (lanes 1–9) or C terminus (lanes 10 and 11) of Sec9p. The band present at the position of Sec9p in the vector lane (lane 1) and the PSSS lane (lane 11) is the protein produced by the sec9-4 allele present in the strain background. The band marked by the arrow (lane 11) is the PSSS protein, which is smaller than the others because it contains the shorter Spo20p N terminus. (B) Relative levels of each of the chimeric proteins, normalized to Sec9p (100) as described in materials and methods.
To ensure that the various chimeric genes were expressed, antibodies raised against the N-terminal or C-terminal regions of Sec9p were used to measure the level of each protein as described in materials and methods (Figure 5). All of the chimeric proteins were expressed at detectable levels and all but two (SPPS and SSPP) were within threefold of the level of the wild-type protein. We cannot rule out the possibility, particularly for SPPS and SSPP, that lower expression levels are responsible for the lack of function of these chimeras. However, this seems unlikely to be the major reason for the differences between the chimeras as the nonfunctional chimeras SPSP, SPPP, and PSSS are all expressed at higher levels than the partially functional SSSP.
These results suggest that the N terminus and the first helix are critical, and the second helix important, for Sec9p function. The requirement for the N terminus is somewhat surprising given previous results that deletion of the N terminus of Sec9p has no effect on rescue of sec9-4 (Brennwaldet al. 1994). To examine the N-terminal requirement more closely, chimeras were constructed in which the N terminus was removed (Table 4). Consistent with previous reports, we found that the Sec9p SNAP-25 region alone (xSSS) is sufficient to complement sec9-4. Again, replacement of the interhelical region with that from Spo20p (xSPS) had no effect. In the absence of the Sec9p N terminus, all other combinations, including replacement of the second helix, led to loss of function.
Taken together, these experiments indicate that only the two helices of Sec9p are required for the function of this t-SNARE in vegetative cells. Removal of the N terminus, replacement of the interhelical region, or both, has no effect. Therefore, the observation that replacement of the Sec9p N terminus with that of Spo20p leads to inactivation raises the possibility that the N terminus of Spo20p is inhibiting the function of the SNAP-25 domain in vegetative cells.
Rescue of· ·sec9-4· ·by chimeras lacking an N-terminal domain·
Both the N- and C-terminal domains of Spo20p contribute to sporulation-specific function: We next examined the ability of the various chimeras to rescue the sporulation defect of a spo20Δ mutant (Figure 6). Sporulation was assessed quantitatively by counting the percentage of cells that form asci in the light microscope and qualitatively by using an ether test (Rockmillet al. 1991). Ether kills vegetative cells but not spores (Dawes and Hardie 1974). Therefore, the amount of growth after ether treatment is an indicator of the ability to produce viable spores. As controls, the SPO20 and SEC9 genes were amplified by PCR and cloned into the pRS306-SPO20-promoter expression vector. When integrated in this plasmid, the SPO20 gene efficiently rescued the sporulation defect of spo20Δ (Figure 6). Replacement of the first helix of Spo20p with the corresponding region of Sec9p (PSPP) had no effect, while substitution of the second helix (PPPS) had only a modest effect. In fact, replacement of both helices (PSPS) produced a protein capable of rescuing the spo20Δ sporulation defect as well as wild-type protein. In contrast, substitution of the interhelical domain (PPSP) resulted in a 30- to 40-fold reduction in sporulation efficiency. Even more strikingly, replacement of the N terminus of Spo20p with that of Sec9p (SPPP) produced a nonfunctional protein (Figure 6). This effect is not due to inhibition by the Sec9p N-terminal domain, as seen for the Spo20p N terminus, because complete deletion of the Spo20p N terminus (data not shown) or a smaller deletion within the N terminus (Figure 7) also inactivates the protein. The requirement for the N-terminal domain is strong but not absolute. Though no asci were visible in the light microscope, some spores must still have been produced by the strain carrying the Sec9p N-terminal/Spo20 N-terminal chimera (SPPP) as evidenced by the appearance of a few ether-resistant papillae (Figure 6). These results indicate that the helical domains, which are the domains critical for Sec9p activity in vegetative cells, do not contribute to the sporulation specificity of Spo20p. On the contrary, precisely opposite to what we observe for Sec9p (Table 3), it is the N-terminal and interhelical regions that are important for Spo20p function.
Rescue of spo20Δ by Sec9p/Spo20p chimeras. Integrating plasmids carrying the indicated gene under control of the SPO20 promoter were integrated into strain AN147 (spo20Δ::his5+/spo20Δ::his5+). Cultures were sporulated and then analyzed in the light microscope to determine the percentage sporulation (minimum 500 cells counted per culture). As a second assay of sporulation, ether tests were performed on sporulated cultures as described in materials and methods.
We do not have antibodies that would allow us to examine the levels of expression for all the chimeras used in the experiment in Figure 6. Therefore, we cannot rule out the possibility that some of the differences in function are due to differences in protein stability. However, given that all of the chimeras were detectably expressed in the sec9-4 strains (Figure 5) and the consistency of the data, it seems unlikely that stability is a major reason for the differing phenotypes observed.
A region of the N terminus of Spo20p is required in sporulation and inhibits SNAP-25 function in vegetative cells: One interesting aspect of these data is that a chimera containing the helices of Sec9p and the interhelical domain of Spo20p can be functional in either vegetative or sporulating cells. When the SNAP-25 domain is in this configuration, the N terminus determines the developmental stage specificity of the protein. If there is no N terminus, or the Sec9p N terminus is present (xSPS or SSPS), then the protein will rescue sec9-4 but not spo20Δ (Tables 2 and 3; data not shown). However, if the N terminus from Spo20p is fused to this SNAP-25 domain (PSPS), then the protein rescues spo20Δ, but not sec9-4 (Figure 6; data not shown). Thus, while the Spo20p N terminus is required for function in sporulating cells, it actively inhibits function in vegetative cells.
Deletion analysis of the Spo20p N-terminal domain. Rescue of the spo20Δ phenotype was assayed as described in the legend to Figure 6. The various PSPS constructs were assayed for rescue of sec9-4 as described in the legend to Table 3. In contrast to the PSPS construct assayed in Figure 6, the PSPS constructs used in this experiment are all expressed under control of the SEC9 promoter.
To determine what regions contribute to the inhibitory or activating functions of the Spo20p N terminus, a series of deletions in the N-terminal domain were constructed. These truncations were made either in the context of the C-terminal chimera containing the Sec9p helices and the Spo20p interhelical region (PSPS) or in the context of the native Spo20p. The former genes were transformed into strain BY41 (sec9-4) to assay for inhibitory activity and the latter into AN147 (spo20Δ) to assay for the activation function of the N terminus.
Deletion of amino acids 3–10, 3–29, or 3–51 within the N-terminal region of Spo20p had no significant effect on the ability of the protein to rescue the spo20Δ sporulation defect (Figure 7). Further deletions of amino acids 3–95, however, inactivate the protein. These observations indicate that residues between amino acids 51 and 95 are necessary for the function of Spo20p.
When the same set of N-terminal deletions was assayed for inhibitory function in vegetative cells (in the context of the PSPS chimera), a slightly different pattern emerged. The degree of inhibition exerted by the Spo20p N terminus in this context was somewhat variable between experiments. However, a general pattern of greater rescue of sec9-4 temperature sensitivity with larger deletions was found in all experiments (Figure 7; A. M. Neiman and L. Katz, unpublished observations). The full-length PSPS chimera rescued sec9-4 poorly or not at all. Deletion of amino acids 3–10 or 3–28 resulted in somewhat better rescue of sec9-4. Restoration of growth to wild-type levels, however, required deletion of amino acids 3–51 or 3–95 (Figure 7). Thus, the inhibitory and activating functions of the Spo20p N terminus are separable. Residues in the first 51 amino acids are required for the inhibitory function, whereas these residues are dispensable for activation.
Finally, these same deletions in the PSPS chimera, under the control of the SEC9 promoter, were introduced into AN147 and assessed for their ability to rescue spo20Δ. The PSPS chimera rescues the spo20Δ defect very poorly when under control of the SEC9 promoter (Figure 7) as compared to its efficacy when expressed from the SPO20 promoter (Figure 6). This difference may be due to lower expression from the SEC9 promoter, as compared to the SPO20 promoter, in sporulating cells. Deletion of amino acids 3–10 or 3–28 has little effect on rescue; however, deletion to amino acid 51 restores sporulation to near wild-type levels. Further deletion to amino acid 95 eliminates this efficient rescue (Figure 7). These observations support the idea that sequences necessary for the function of Spo20p in sporulating cells lie between amino acids 51 and 95. Also, they indicate that the inhibitory function of the Spo20p N terminus is present in sporulating cells as well as in vegetative cells and can be eliminated by removal of sequences between amino acids 29 and 51.
DISCUSSION
In this article, we seek to understand the molecular basis of the developmental stage-specific function of the two yeast SNAP-25 homologues, Sec9p and Spo20p. One way for this difference to arise is if Sec9p and Spo20p are components of entirely distinct SNARE complexes. An important question, therefore, is whether these two proteins interact with the same v- and t-SNARE partners. Recent work has demonstrated that in strain backgrounds where the SNC genes are not essential for viability, the SNC genes are still required for sporulation, as would be predicted if these gene products work with Spo20p to mediate vesicle fusion to the prospore membrane (Davidet al. 1998). Spo20p interaction with Ssop and Sncp was examined using two different assays: the binding of the Snc and Sso proteins directly to Spo20p in vitro and the localization of the Sso and Snc proteins in sporulating cells. Spo20p formed binary complexes with Sso1p and ternary complexes with Sso1p and Snc2p essentially identically to Sec9p. Furthermore, Spo20p, Ssop, and Sncp are all localized to prospore membranes in sporulating cells (Figures 1 and 2). This localization for Ssop and Sncp is expected if they participate in fusion of secretory vesicles with the prospore membrane. Thus, the binding of Ssop and Sncp to Spo20pand their localization to the prospore membrane strongly implicate these proteins in the fusion of secretory vesicles to the prospore membrane.
In contrast to their co-localization in wild-type cells, the patterns of Ssop and Sncp staining differed during sporulation in the spo20Δ mutant strain. Ssop continues to localize strongly to the abnormal prospore membranes that form in spo20Δ mutants, whereas localization of Sncp to the prospore membrane is reduced. Because localization of Sncp to the prospore membrane serves as a marker for fusion of secretory vesicles with the prospore membrane, this observation suggests that Sncp-containing vesicles do not fuse as efficiently with the prospore membrane in a spo20Δ mutant. A number of other proteins that localize to the prospore membrane in wild-type cells, for example, Gas1p (Neiman 1998) and Spo14p (Rudgeet al. 1998), fail to reach this compartment in spo20Δ mutants (A. M. Neiman, unpublished observations; J. Engebrecht, personal communication), which is consistent with these proteins being cargo molecules of Sncp-dependent vesicles.
The failure of Sncp to localize to the prospore membrane in spo20Δ has several interesting implications. First, because formation of the residual prospore membranes found in a spo20Δ mutant requires Sec9p, the reduced localization of Sncp to this compartment suggests that during sporulation Sncp may not efficiently oligomerize with a t-SNARE containing Sec9p. Thus, the developmental stage specificity of Spo20p and Sec9p may be explained, in part, by a change in v-SNARE affinity. Regulation of Sncp so that it preferentially forms oligomers with Sec9p containing t-SNAREs in vegetative cells and Spo20p containing t-SNAREs in sporulating cells may explain the failure of SEC9 and SPO20 to substitute for one another. A number of mechanisms could account for such a switch in v-SNARE preference. One simple possibility is that Snc proteins are developmentally modified and the differently modified forms have different affinities for the alternative t-SNARE complexes. It should be noted, however, that no such specificity was obvious in our in vitro binding experiments. Alternatively, sporulation- or vegetative-specific accessory proteins might affect v-SNARE preference by regulating assembly of the oligomer. Our analysis of the Spo20p N-terminal domain (see below) suggests the existence of such a sporulation-specific factor.
The results presented here also suggest an alternative interpretation of the cytological phenotype of spo20Δ mutants. In previous work we have shown that in spo20Δ mutants prospore membranes dissociate prematurely from nuclei, leading to the formation of anucleate spores (Neiman 1998). On the basis of this observation, Spo20p was proposed to play some role in anchoring the prospore membrane to the nucleus. However, the altered localization of Sncp in spo20Δ cells strongly suggests that the primary defect in spo20Δ is in vesicular fusion with the prospore membrane. The anchoring defect, therefore, may be a secondary consequence of either a slower overall growth rate of the prospore membrane or the failure to transport some other protein critical for membrane anchorage.
Finally, the continued localization of Ssop to the prospore membrane in spo20Δ cells indicates that, unlike Sncp, Ssop localization is SPO20 independent. Ssop is the only cell surface marker we have found whose localization to the prospore membrane is unaffected by disruption of SPO20. Recently, Gerst and colleagues (Davidet al. 1998) provided evidence for a Sncp-independent pathway of vesicular fusion with the plasma membrane in vegetative cells. This pathway, however, requires the function of both Ssop and Sec9p. It is possible that delivery of Ssop to the prospore membrane is through this Sncp-independent pathway and is thus insensitive to the loss of SPO20.
A function for the interhelical region revealed by chimeric SNAP-25 domains: Sec9p and Spo20p interact with the same binding partners and yet they function at distinct stages of development. Because the proteins are homologous, we took the approach of constructing chimeric molecules to try to define specific regions of the proteins that contribute to function in specific developmental conditions. Several patterns emerge from our analysis of Spo20/Sec9 chimeras. First, with regard to the helical domains, various levels of function are seen with different combinations of helices. The general pattern in both vegetative and sporulating cells is that pairing of the helices from the same protein works very well, the combination of helix 1 from Sec9p and helix 2 from Spo20p works reasonably well, and the combination of helix 1 from Spo20p and helix 2 from Sec9p works relatively poorly. The poorer function of the chimeras with helices from both proteins may reflect poor interactions between the helices themselves, which are expected to contact each other as well as Ssop and Sncp in the oligomer (Suttonet al. 1998). The only exception to this general pattern is when both helices come from Spo20p; in this case, the protein can function only in sporulating cells. This observation is somewhat surprising given that the helices are not actually required for function in sporulation. One explanation for this result would be that the Spo20 helices preferentially oligomerize with a sporulation-specific form of the Sncp proteins.
The second, more striking, pattern in the chimera data is how different the domain requirements are for Sec9p and Spo20p function. Maintaining just the two helices of Sec9p is sufficient for function in vegetative cells. Complete removal of the N terminus, substitution of the interhelical domain with that from Spo20p, or both does not seem to significantly affect the ability of the protein to rescue sec9-4. By contrast, Spo20p is largely insensitive to swapping of the helices, but requires both the N-terminal and interhelical domains for efficient rescue of spo20Δ (Figure 6).
It is not known what role the interhelical domain plays in the function of SNAP-25 proteins. One function for the interhelical domain is suggested by the fact that in SNAP-25 this region contains cysteine residues that are palmitoylated (Hesset al. 1992; Lane and Liu 1997) and contribute to the membrane association of SNAP-25 (Lane and Liu 1997; Kotichaet al. 1999). Though no cysteines are present in the interhelical domains of either Sec9p or Spo20p, it is possible that these domains similarly contribute to the localization of the proteins. However, the observation that both Spo20p and Sec9p can localize to the prospore membrane in sporulating cells (Figure 1) argues against this domain conferring a differential localization of the two proteins. Alternatively, the interhelical domain of SNAP-25 has been shown to be required for the formation of multimeric SNARE complexes (Poirieret al. 1998). Conceivably, the interhelical domain of Spo20p could mediate the formation of a particular arrangement of SNARE complexes required for efficient fusion with the prospore membrane.
The Spo20 N terminus is required for function in sporulating cells: The most obvious difference between Sec9p and Spo20p is in the function of their N-terminal domains. While the Sec9p N terminus is dispensable, the Spo20p N terminus is not only necessary in sporulating cells, it inhibits the function of an associated SNAP-25 domain in vegetative cells.
Our deletion analysis suggests that these inhibitory and activating activities require separable regions of the Spo20p N terminus. Further, the phenotypes of the various N-terminal deletions in the context of the PSPS chimera indicate that, though inhibition by the N terminus is active in both sporulating and vegetative cells, activation by the N terminus is required only in sporulation. This observation is consistent with these activities requiring separate regions of Spo20p.
A possible mechanism of inhibition by the N terminus is steric hindrance by an interaction between this part of the protein and the SNAP-25 domain. A similar cis inhibition has been seen in the Sso1 protein (Nicholsonet al. 1998). However, in in vitro binding assays using full length Spo20p or the PSSS chimera, the presence of the Spo20 N-terminal domain did not significantly impair the ability of these proteins to form complexes with Sso1p and Snc2p (L. Katz and P. J. Brennwald, unpublished observations). Alternatively, this region may serve as a binding site for some other factor. If this region does function as a binding site, then the putative inhibitory factor must be present in both vegetative and sporulating cells. In contrast, the ability of the PSPS chimera to rescue spo20Δ but not sec9-4 suggests that the N terminus can provide its activating function only in sporulating cells. One simple model to explain our observations is that the N-terminal domain of Spo20p inhibits the function of an associated SNAP-25 domain and binding of a sporulation-specific factor to an adjacent region of the N terminus relieves this inhibition. Note, however, that the activating region appears to play a direct, positive role in addition to overcoming the inhibitory role of the N terminus. This is indicated by the reduced ability of the Δ3-95PSPS chimera to rescue spo20Δ compared to the Δ3-51PSPS chimera, even though both gene products rescue sec9-4.
Regulation of SNARE formation during sporulation: Whatever the mechanism by which the N-terminal region of Spo20p exerts its negative and positive effects on vesicular fusion, our results clearly indicate an additional layer of regulation of SNARE function in sporulating cells. In principle, this regulation could reflect a novel role for Spo20p in the generation or transport of secretory vesicles to the prospore membrane. However, the fact that prospore membranes form in the spo20Δ mutant at the proper locations in the cell argues against such models. The most likely explanation, therefore, is that either the assembly of the SNARE complex or the ability of the assembled complex to promote fusion is regulated.
The reason for this additional control on SNARE function during the process of spore formation is not yet clear. We suggest two possibilities. First, regulation of SNARE assembly during prospore membrane formation could play a role in enhancing the fidelity of secretory vesicle delivery. For instance, if a necessary regulatory factor were specifically localized to the prospore membrane compartment, stray secretory vesicles might be prevented from fusing with the plasma membrane. Alternatively, such a regulatory factor could play a role in maintaining even and coordinated growth of the four different membranes forming simultaneously. Differentiation between these, and other, possibilities will require the identification and analysis of the putative regulatory factors.
Acknowledgments
The authors are grateful to N. Hollingsworth and R. Sternglanz for comments on the manuscript; G. Waters, C. Burd, T. Sato, and S. Emr for plasmids; P. Novick and J. Pringle for strains and antibodies; N. Hollingsworth, N. Dean, and R. Sternglanz for helpful discussions; and J. Engebrecht for communication of results prior to publication. A.M.N. wishes to thank R. Sternglanz for material support throughout the course of this project. This work was supported by National Institutes of Health grant GM28220 to R. Sternglanz and by grants from the Mathers Charitable Foundation, the Pew Scholars in Biomedical Sciences Program, and the National Institutes of Health (GM54712) to P.J.B.
Footnotes
-
Communicating editor: M. D. Rose
- Received November 9, 1999.
- Accepted April 26, 2000.
- Copyright © 2000 by the Genetics Society of America
