Amino Acid Sequence Requirement for Efficient Incorporation of Glycosylphosphatidylinositol-associated Proteins into the Cell Wall of Saccharomyces cerevisiae *

During cell wall biogenesis inSaccharomyces cerevisiae, some glycosylphosphatidylinositol (GPI)-attached proteins are detached from GPI moieties and bound to β-1,6-glucan of the cell wall. The amino acid sequence requirement for the incorporation of GPI-attached proteins into the cell wall was studied by using reporter fusion proteins. Only the short ω-minus region composed of five amino acids, which is located upstream of the ω site for GPI attachment, determined the cellular localization of the GPI-associated proteins. Within the ω-minus region, amino acid residues at the ω-4 or -5 and ω-2 sites were important for the cell wall incorporation. Yap3p, a well characterized GPI-anchored plasma membrane aspartic protease, was localized in the cell wall when the ω-minus region was mutated to sequences containing Val or Ile at the ω-4 or -5 site and Val or Tyr at the ω-2 site.

GPI-associated proteins have been isolated from various organisms from yeasts and protozoa to mammals (16 -18). These proteins are structurally related in that they all contain a signal sequence for secretion in the N terminus and a GPI signal for an attachment to a GPI in the C terminus. The GPI-signal region is composed of three domains: a GPI attachment region comprising , ϩ1, and ϩ2 sites; a spacer of 5-10 amino acids; and a hydrophobic stretch of 10 -15 amino acids. A protein containing the GPI-signal is cleaved at the site, and the resulting carboxyl terminus of the protein becomes covalently bound to a GPI moiety. This reaction occurs in the luminal face of the endoplasmic reticulum and, in yeast, requires GAA1 (19) and GPI8 (20) gene products. The GPI-at-tached proteins are then transported to the cell surface, where they are exposed on the extracytoplasmic face of the plasma membrane. During the transportation from the endoplasmic reticulum to the plasma membrane, the proteins are mannosylated and become GPI-anchored mannoproteins.
In protozoa and mammals, the GPI-anchored mannoproteins remain on the plasma membrane and take biological functions related to cell-cell and cell-environment interactions (18,21,22). In addition to the above functions, some of the GPI-anchored mannoproteins in the yeast are further processed and are incorporated into the cell wall. The incorporation of GPIassociated proteins into the cell wall is thought to occur in two steps: detachment of a GPI moiety from the protein and linking of the GPI-detached protein to ␤-1,6-glucan of the cell wall (3). Our knowledge of this process is very rudimentary. The addition of 30 -40 amino acids from the C terminus of GPI-dependent cell wall proteins to a reporter protein directs incorporation of the chimeric proteins into the cell wall (23)(24)(25). This suggests that the short C-terminal sequences contain all the information necessary for the incorporation of proteins into the cell wall. Therefore, it is possible that there is a gene product that would recognize a sequence or structural motif in the C-terminal region of GPI-associated proteins and direct their final cellular localization. A sequence alignment from our previous study on GPI-dependent cell wall proteins (24) suggested that there is some sequence similarity among the -minus regions, corresponding to regions upstream of the site, of GPI-dependent cell wall proteins identified in the yeast. In this study, we examined, in detail, the amino acid sequences of the -minus regions and determined the sequence requirement for the incorporation of GPI-attached proteins into the cell wall.

EXPERIMENTAL PROCEDURES
Strains and Media-All experiments were carried out in S. cerevisiae strain YPH499 (Mata, his3 leu2 trp1 ura3; Ref. 26). Selective medium supplemented with the appropriated nutrients except uracil (SC-Ura; Ref. 27) was used for plasmid maintenance. Escherichia coli strain DH5 (28) was used for the construction and amplification of plasmids. Bacterial medium (LB broth) was prepared as described by Maniatis et al. (28).
Plasmids and DNA Manipulation-pE␣GALHA (24) containing a reporter gene comprising the signal sequence of yeast invertase, ␣-galactosidase from guar (Cyamopsis tetragonoloba), and a hemagglutinin (HA) epitope was used as a host vector. The reporter gene is transcribed through the PGK1 promoter. A DNA fragment covering the C-terminal amino acids of a GPI-attached protein and about 200 base pairs of its 3Ј-noncoding region was amplified by polymerase chain reaction (PCR) with primers containing a SalI or BamHI restriction site. The resulting fragment was inserted into the SalI and BamHI sites of pE␣GALHA and then used to transform the yeast strain YPH499. Replacement of amino acid residues and construction of chimeric fusion genes were carried out by PCR-based site-directed mutagenesis (29).
A 3ϫ HA epitope sequence was inserted in-frame into the 514position of the Yap3p amino acid sequence (30). A fragment containing the 400 base pairs of the 5Ј-noncoding region and the N-terminal 514 amino acids of Yap3p was amplified by PCR and inserted into the SphI and SalI sites of pEHA, generating pEHAnYAP3. pEHA contains the 3 ϫ HA epitope sequence (31) between the BamHI and SalI sites of the multicloning site of YEplac195 (32). Another fragment containing the C-terminal 55 amino acids of Yap3p and the 460 base pairs of its 3Ј-noncoding region was amplified and inserted into the BamHI and PstI on pEHAnYAP3, generating pEHAYAP3.
Treatment of Membranes with Phosphatidylinositol-specific Phospholipase C (PI-PLC)-Yeast transformant cells grown at 30°C in the SC-Ura medium were harvested by centrifugation and washed with TE buffer (10 mM Tris-HCl, pH 7.6, 1 mM EDTA). Membrane proteins were isolated from the transformant cells and subjected to phase separation with Triton X-114 as described previously (24,33,34). Yeast cells were broken with glass beads in LY buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 5 mM EDTA, 1 mM phenylmethylsulfonyl fluoride) containing protease inhibitor mixture (Complete; Boehringer Mannheim). After removal of cell debris by centrifugation (in a microcentrifuge, 1 min, 2000 rpm, 4°C), a 10% Triton X-114 solution was added to the crude lysate to a final concentration of 1.4%, and the mixture was incubated on ice for 40 min. The solution was centrifuged (in a microcentrifuge, 5 min, 10,000 rpm, 4°C), and the supernatant was transferred to a new tube. After incubation at 33°C for 3 min, the solution was centrifuged (in a microcentrifuge, 20 s, 10,000 rpm, room temperature) and separated into an aqueous phase and a detergent phase. The aqueous phase was removed, and the remaining detergent phase was twice extracted with LY buffer. Membrane proteins in the detergent phase were treated with PI-PLC as described previously (35). Membrane proteins resolved in LY buffer without NaCl were incubated with 0.1 unit of PI-PLC from Bacillus thuringiensis (Oxford GlycoSciences) at 33°C for 30 min and then separated into a detergent phase and an aqueous phase. Proteins in both phases were separately precipitated with trichloroacetic acid with carrier bovine serum albumin and analyzed by SDS-polyacrylamide gel electrophoresis and Western blotting (36). An anti-HA monoclonal antibody (12CA5, Boehringer Mannheim) was used for detection of the fusion proteins on blots.
Treatment of Cell Walls with Laminarinase-Cell walls were isolated from yeast transformant cells as described previously (24,37,38). Yeast cells were broken with glass beads in LY buffer containing 2% SDS, and the cell lysate was centrifuged (in a microcentrifuge, 1 min, 10,000 rpm, room temperature). The resulting pellet was twice treated with a hot 2% SDS-containing LY buffer by heating for 10 min at 95°C each time and washed five times with LM buffer (100 mM sodium acetate, pH 5.5, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride). The washed cell walls were incubated with laminarinase (Sigma L5144) at a final concentration of 0.25 unit/ml twice for 1 h each at 37°C as described previously (24,39). The treated sample was centrifuged (in a microcentrifuge, 2 min, 10,000 rpm, 4°C), and the resulting supernatant was analyzed by SDS-polyacrylamide gel electrophoresis and Western blotting. Detection of the proteins on blots was carried out using the anti-HA monoclonal antibody.
Endoglycosidase H (Endo H) Digestion-Mannoproteins were digested with Endo H as described previously (5). Mannoproteins liberated from the cell wall with the laminarinase treatment were heated at 100°C for 10 min in LM buffer containing 0.1% SDS and 0.5% 2-mercaptoethanol. Recombinant Endo H (Boehringer Mannheim) was added to the solution to a final concentration of 0.3 milliunit/ml with protease inhibitor mixture (Complete; Boehringer Mannheim). The reaction solution was incubated overnight at 37°C and then further incubated at 37°C for a further 3 h after the addition of the same amount of Endo H.

Determination of Sites for GPI Attachment-Potential
sites for GPI attachment are predictable based on the consensus rules described by Nuoffer et al. (33) and by Udenfriend and Kodukula (17). A sequence analysis of putative GPI-associated proteins has suggested that in S. cerevisiae two amino acid residues, Asn and Gly, are preferentially used as site residues (24,40). A mutational analysis of the site residue in Gas1p, a well characterized yeast GPI-anchored plasma membrane protein (41,42), has shown selectivity of the site residue. Substitution of the site residue from Asn, an authentic site residue, with Ser, Gly, Ala, Asp, or Cys slightly decreased the efficiency of GPI attachment, while replacement of Asn with other amino acid residues including Gln, Glu, and Thr resulted in an almost complete loss of the GPI attachment efficiency (33). Based on this selectivity of the site residue, we determined the sites of two GPI-dependent cell wall proteins, Yjr151c (24) and Ydr077w/Sed1p (15), and two GPI-anchored plasma membrane proteins, Yir039c (24) and Ylr120c/Yap3p (43)(44)(45). The C-terminal sequences, shown in Fig. 1A, with mutated amino residues at the putative site were fused to the reporter gene, and the resulting fusion proteins were expressed in yeast cells. GPI anchoring of the fusion proteins was examined by using phase separation of membrane proteins with Triton X-114 and digestion with PI-PLC (Fig. 1B). In the Cterminal sequences of Ydr077w and Ylr120c containing the wild type residue Asn at the putative site, the fusion proteins were detected in the detergent phase and were released into the aqueous phase by the PI-PLC treatment, indicating GPIdependent anchoring in membranes. Mutant fusion proteins with replacement of Asn with Gly or Asp at the site were also extracted in the detergent phase. The amounts of these proteins were comparable with that of the wild type, and they were also PI-PLC-sensitive. On the other hand, mutant fusion proteins with replacement of Asn with Gln, Glu, or Thr at the site were not detected in the detergent phase as much as the wild type. The amounts of these proteins detected were quite low, and most of them were not PI-PLC-sensitive, indicating loss of GPI anchoring efficiency. All these effects on GPI anchoring efficiency suggest the same selectivity of the site residue for GPI anchoring to that previously reported in Gas1p (33). Similar results were also obtained with amino acid changes at the sites for the C-terminal sequences of Yjr151c and Yir039c (data not shown). From these results, it could be concluded that for all four proteins the N residues function as the sites for GPI attachment.
Identification of -Minus Regions Determining Cellular Localization of the Fusion Proteins-Our previous study (24) has shown that the addition of the most C-terminal 40 amino acids of Yjr151c or Ydr077w to a reporter protein directs the resulting fusion proteins into the cell wall and that the addition of the C-terminal amino acids of Yir039c or Ylr120c directs their fusion proteins to the plasma membrane. This indicates that the C-terminal amino acid sequences determine the final cellular localization of GPI-associated proteins. These C-terminal amino acid sequences can be divided into two regions: the -plus and -minus regions with the site as a center (Fig.  1A). The -plus region contains the residues of , ϩ1, and ϩ2 that function as the sequence required for GPI attachment; a spacer sequence of 5-10 amino acids; and a hydrophobic stretch of 10 -15 amino acids. The -minus region is Nterminal to the site. To determine which of the two regions is responsible for determination of cellular localization, chimeric genes between the two regions were made and expressed in yeast cells, and the cellular localization of their corresponding chimeric fusion proteins was examined (Fig. 2). The chimeric proteins of the -minus region from either Yjr151c or Ydr077w and the -plus region from either Yir039c or Ylr120c were detected in the cell wall as much as those of the wild type Yjr151c and Ydr077w (Fig. 2B). They were released only by digestion of the cell wall with glucanase. On the other hand, the chimeric proteins of the -minus region from either Yir039c or Ylr120c and the -plus region from either Yjr151c or Ydr077w were slightly detected in the cell wall (Fig. 2B) Their amounts were as low as those observed in the wild types of Yir039c and Ylr120c. All chimeric proteins were determined to be membrane-associated GPI-associated proteins, because they were released from membrane fractions by the PI-PLC treatment (Fig. 2C). From these results, it could be concluded that only the -minus region is responsible for determination of the final cellular localization of GPI-associated proteins. Short -minus sequences of 15 amino acids were used in these experiments (Figs. 1A and 2A). Therefore, a signal determining the final cellular localization could be located within these short -minus sequences.
Mutational Analysis of -Minus Regions-To identify amino acid sequence required for the cell wall incorporation, we further examined the -minus regions of Yjr151c and Ydr077w. Since the -minus regions of the GPI-dependent cell wall proteins examined are Ser-and/or Thr-rich (24, 40), we replaced each amino acid residue in the sequence with Ser and examined the effects of the mutations on the incorporation of Yjr151c-and Ydr077w-derived fusion proteins into the cell wall (Fig. 3A). All fusion proteins examined were expressed in almost same levels (data not shown, but see Fig. 3C, lanes T). For the C-terminal sequence of Yjr151c, changes only within the sequence VSINT corresponding to the -5 to -1 sites reduced the efficiency for the cell wall incorporation (Fig. 3B, CT22). This reduction was not due to a decrease of GPI-anchoring efficiency, because the amount of the fusion proteins released by the PI-PLC treatment of the membrane fractions was comparable with that of a fusion protein with the wild type sequence (Fig. 3C). Within the sequence VSINT, a single change of Val to Ser at the -5 site severely reduced the incorporation efficiency into the cell wall (Fig. 3B, CT22-M05). Single or double amino acid replacement including the -2 site residue moderately reduced the cell wall incorporation (Fig. 3B, CT22-M07-09). Furthermore, a double replacement at both the -5 and the -2 sites caused the most severe reduction in the cell wall incorporation efficiency (Fig. 3B, CT22-M11). Similarly, for the C-terminal sequence of Ydr077w, amino acid replacements only in the sequence VVINS corresponding to the -5 to -1 sites reduced the efficiency for the cell wall incorporation. Single changes of the -5 or -2 site residue in the sequence showed a partial reduction of the cell wall incorporation efficiency (Fig. 3B, CT26-M04 and -M08), while double replacements of both -5 and -2 site residues caused a severe reduction of cell wall incorporation (Fig. 3B, CT26-M09). From these results, it could be concluded that the signal for the cell wall incorporation would be located in the sequence of the -5 to -1 sites and that within the sequence the -5 and -2 sites are the most important.
Identification of Amino Acid Residues Required for the Cell Wall Incorporation-A sequence alignment of the -minus regions in the GPI-dependent cell wall proteins (24) indicates that residues Ile and Val preferentially appear at the -5 site and residues Tyr and Asn are predominant at the -2 site. We examined the requirement of these amino acid residues for the cell wall incorporation by using a model construct. It contains a stretch of 15 serines followed by the site and the -plus region of Ylr120c/Yap3p, a well characterized GPI-anchored plasma membrane protein (43)(44)(45), fused to a HA-tagged ␣-galactosidase (Fig. 4A, SCT01). As expected, this model fusion protein was mostly localized in membranes as a GPI-anchored protein (data not shown). However, a small amount of the FIG. 1. Determination of sites for GPI attachment. A, amino acid sequences of the C termini of two GPI-dependent cell wall proteins, Yjr151c (24) and Ydr077w/Sed1p (15), and two GPI-anchored plasma membrane proteins, Yir039c (24) and Ylr120c/Yap3p (43)(44)(45). -Minus and -plus regions represent amino acid sequences of N-terminal and C-terminal to putative sites (, marked in boldface type) for GPI attachment, respectively. The C-terminal sequences, designated CT22, CT26, CT04, and CT05, from Yjr151c, Ydr077w, Yir039c, and Ylr120c, respectively, were fused to the reporter gene comprising the signal sequence of yeast invertase, ␣-galactosidase from guar and the HA epitope on a multicopy plasmid, pE␣GALHA (24). B, analysis of membrane-bound forms of site-mutated CT26 and CT05 fusion proteins by phase separation and PI-PLC. Amino acid residues of putative sites were changed from the wild type Asn to Asp, Gln, Glu, Gly, and Thr by using PCR-based mutagenesis. YPH499 cells carrying plasmids with the wild type site residue (Asn) or mutated site residue (Asp, Gln, Glu, Gly, or Thr) were grown in SD-Ura media to a concentration of A 600 ϭ 1.5-2.0. Fusion proteins extracted in detergent phases by phase separation with Triton X-114 were prepared. They were treated (ϩ) or not treated (Ϫ) with PI-PLC and then separated into a detergent phase (D) and an aqueous phase (A). Fusion proteins in each fraction were precipitated with trichloroacetic acid and, with total cell lysates (T), analyzed by Western blotting using anti-HA monoclonal antibody.
fusion protein was detected in the cell wall as a glucanaseextractable protein (Fig. 4B, SCT01). The addition of the -minus sequence containing the -5 to -1 region from Yjr004c/ Ag␣1p, Yjr151c, Ydr077c/Sed1p, Ygr189c, and Yar050w/Flo1p, all GPI-dependent cell wall proteins, significantly increased the amount of fusion proteins incorporated into the cell wall (Fig. 4B, SCT02-06). Their amount of incorporation was various among the five constructs.
We replaced Ser with Val at each residue between the -6 and -1 sites and examined their effects on the cell wall incorporation of the fusion proteins (Fig. 4B, SCT07-12). Only a change at the -2 site increased the incorporation efficiency of these fusion proteins. A similar result was obtained when each Ser was changed to Ile (data not shown). Next, Ser at the -2 site was replaced with Val, Ile, Tyr, Asn, or Lys (Fig. 4B,  SCT11-16). The efficiency of the cell wall incorporation was increased strongly by the replacement with Val or Ile and weakly by the replacement with Tyr or Asn. No increase was observed by the replacement with Lys. Finally, in addition to the change of Ser to Tyr at the -2 site, a change of Ser to Ile at the other sites was introduced (Fig. 4B, SCT14-20). The additional changes at the -5 or -4 site further increased the efficiency of the cell wall incorporation. In SCT01, only 4.1 Ϯ 0.9% of the total fusion protein was detected in the cell wall, while in SCT18 and SCT19, 26.6 Ϯ 4.2 and 24.9 Ϯ 3.7%, respectively, of the total fusion proteins were detected in the cell wall. The efficiency of the cell wall incorporation was enhanced 6-fold by combining the mutation of Ser to Ile at the -5 or -4 site with a mutation of Ser to Tyr at the -2 site. Similar enhancement of the cell wall incorporation was observed by the replacement of Ser with Val at the -5 or -4 site and of Ser with Asn at the -2 site (data not shown). A slight enhancement of the cell wall incorporation effect was observed by a change of Ser to Val at the -4 or -5 site in addition to the change of Ser to Val at the -2 site (data not shown). From all of these results, it could be concluded that combinations of Ile or Val at the -4 or -5 site and Tyr, Asn, or Val at the -2 site act to promote the incorporation of fusion proteins into the cell wall.
Cell Wall Incorporation of Mutated Yap3p GPI-anchored Plasma Membrane Protein-S. cerevisiae Yap3p, an aspartic protease, is a GPI-anchored plasma membrane protein (43)(44)(45). Using this protein, we tested whether the replacement of amino acid residues at the -4 or -5 and -2 sites directs the protein into the cell wall. First, the C-terminal sequence of Yap3p including the -minus region was used (Fig. 5A). The fusion protein with this C-terminal sequence was not detected in the cell wall as a glucanase-extractable protein (Fig. 5B,  CT05). Changes of the -minus region from the wild type TSSKR to VSSVS, SVSVS, ISSYS, or VSSYS resulted in the fusion protein being incorporated into the cell wall (Fig. 5B, CT05-M02-05). These sequences correspond to the -5 to -1 regions of GPI-dependent cell wall proteins. Interestingly, a change from TSSKR to SSSSS also led to the incorporation of the fusion protein in the cell wall, but its amount was very low (Fig. 5B, CT05-M01). Second, the intact Yap3p tagged with HA was used. The Yap3p protein prepared from membrane fractions was detected as a broad band around 70 kDa after digestion with Endo H (Fig. 6B). Yap3p was extracted in a detergent phase of two-phase separation with Triton X-114 and was released from membrane fractions by treatment with PI-PLC (Fig. 6B), indicating that Yap3p is a GPI-anchored membrane protein. Mutant Yap3p proteins examined were also detected as GPI-attached membrane proteins (Fig. 6C). Next, incorporation of the wild type and mutant Yap3p proteins into the cell wall was examined. Yap3p proteins liberated from the cell wall by glucanase were further digested with Endo H, because without the Endo H treatment they formed diffused bands near the origin of gels in SDS-polyacrylamide gel electrophoresis. Compared with the wild type Yap3p, large amounts of three mutant Yap3p proteins were released from the cell wall fractions by the glucanase treatment (Fig. 6D). They contain mutated sequences of either VSSVS, TVSVS, or ISSYS in the -5 to -2 region (Fig. 6A). A mutant Yap3p containing SSSSS in the -5

FIG. 2. Cell wall localization of chimeric proteins.
A, schematic representation of chimeric proteins examined. Chimeric C-terminal sequences were constructed by connecting a -minus region of a gene to a -plus region of another gene. CTX(22:04) is, for example, a chimera between the -minus region of CT22 and the -plus region of CT04. The chimeric C-terminal sequences were then fused to the reporter gene on pE␣GALHA. CT22, CT26, CT04, and CT05 represent the C-terminal sequences of Yjr151c, Ydr077w, Yir039c, and Ylr120c, respectively (see Fig. 1A for their sequences). B, analysis of cell wall-bound forms of chimeric fusion proteins. YPH499 cells carrying plasmids with chimeric fusion genes were grown in SD-Ura media to a concentration of A 600 ϭ 1.5-2.0. Cell wall fractions were prepared and digested with laminarinase. The proteins released from the cell wall were analyzed by Western blotting using anti-HA monoclonal antibody. C, analysis of membranebound forms of chimeric fusion proteins by phase separation and PI-PLC. Results for the two chimeric fusion proteins, CTX(26:05) that was abundantly incorporated into the cell wall and CTX(05:26) that was slightly incorporated into the cell wall, are shown here. See the legend of Fig. 1B for further explanation. to -1 region was slightly detected in the cell wall, and its amount was comparable with that of the wild type (Fig. 6D, 05HA-M01). DISCUSSION We have studied the sequence requirement for incorporation of GPI-attached proteins into the cell wall and reached the conclusion that only the short -minus region composed of five amino acids (called the -minus 5-1 sequence), which is located upstream of the site for GPI attachment, determines the cellular localization of the GPI-associated proteins and that within the -minus 5-1 sequence amino acid residues at the -4 or -5 and -2 sites are important for the determination (called the -4/5 and -2 rule). Yap3p, a well characterized GPI-anchored plasma membrane protein (43)(44)(45), became localized in the cell wall when the -minus 5-1 sequence of Yap3p was mutated to sequences containing Val or Ile at the -4 or -5 site and Val or Tyr at the -2 site. In the previous study (24), we showed some sequence similarity in the -minus regions of 15 GPI-dependent cell wall proteins where either Val or Ile occupies the -5 site and Tyr is predominant at the -2 site. This study demonstrated their importance for the cell wall incorpo- ration. Although a single residue of Val, Ile, Tyr, or Asn at the -2 site has some tendency to direct fusion proteins into the cell wall, the combination of the Val, Ile, Tyr, or Asn at the -2 site with Val or Ile at the -5 site strongly directed the fusion protein into the cell wall. In fact, out of 15 GPI-dependent cell wall proteins identified, seven, two, and three proteins possess combinations of Ile and Tyr, Val and Tyr, and Val and Asn, respectively, at the -5 and -2 sites (24).
The fusion protein containing the wild type sequence of TSSKR as the -minus 5-1 sequence of Yap3p was not localized in the cell wall, while the fusion protein containing the mutated sequence of SSSSS was partially localized in the cell wall (Fig.  5B). The sequence difference of KR in TSSKR and SS in SSSSS is important, because a changing of TSSKR to SSSKR led to no detectable change in the localization of the fusion protein (data not shown). It is possible that the Lys and Arg residues at the -2 and -1 sites may act negatively for the cell wall incorporation. Dibasic residues like this have been identified in the -minus region in some of putative GPI-anchored plasma membrane proteins (24,40,46). These dibasic residues may be a motif acting as a negative signal. A protein containing the dibasic motif may not be processed for incorporation into the cell wall and will then remain on the plasma membrane with a GPI. If this is the case, the action of the -4/5 and -2 rule for the cell wall incorporation would be interpreted as removal of the negative signal by interrupting the dibasic motif. However, FIG. 4. Determination of amino acid residues required for the cell wall incorporation. A, schematic representation of amino acid sequences. The sequence of SCT01 was constructed by replacement of the -minus region of CT05 with a stretch of 15 serines and used as a starter sequence. CT05 represents the C-terminal sequence of Ylr120c/Yap3p that is a GPI-anchored plasma membrane protein. Amino acid residues were mutated by using PCR-based mutagenesis, and only amino acid residues different from that of CT05 are indicated. SCT02, -03, -04, -05, and -06 contain the -5 to -1 sequences of Yjr004c/Ag␣1p (7), Yjr151c (24), Ydr077c/Sed1p (15), Ygr189c (24), and Yar050w/Flo1p (12), respectively, which are all known as GPI-dependent cell wall proteins. The SCT01 and mutant sequences were fused to the reporter gene on pE␣GALHA for analysis. this interpretation would not be enough to explain the positional and additional effects by the -4/5 and -2 rule on the cell wall incorporation. Sequences meeting the rule clearly incorporated the fusion proteins into the cell wall more efficiently than that of the SSSSS sequence. Therefore, although the rule may not be indispensable for the cell wall incorporation, it would be required for efficient incorporation of GPIattached proteins into the cell wall.
The incorporation of GPI-associated proteins into the cell wall has been thought to occur in two steps: detachment of a GPI-moiety from a GPI-attached protein and binding of the GPI-detached protein to ␤-1,6-glucan of the cell wall (3). This protein-bonded ␤-1,6-glucan is covalently associated with ␤-1,3-glucan as well as chitin and forms a large complex involving cell wall components (6,47,48). Since ␤-1,6-glucan assembly is assumed to be intracellularly initiated (49), it is possible that the binding of a GPI-detached protein to ␤-1,6-glucan occurs somewhere in intracellular space during or after the ␤-1,6-glucan synthesis. However, the study using an anti-␤-1,6-glucan antibody refuted this possibility. The antibody reacted with proteins collected from the cell wall, indicating association with ␤-1,6-glucan (4 -6, 23, 50), while it did not react with proteins intracellularly collected (5). Therefore, the removal of the GPI moiety and the linking to ␤-1,6-glucan would occur in the extracellular space. This suggests that GPIdependent cell wall proteins are transported to the plasma membrane as a form of GPI-attached proteins that are associated with membranes by the GPI moiety. As expected, the GPI-attached form of the cell wall proteins was extracted in the detergent phase of Triton X-114 and showed PI-PLC sensitivity. The cell wall incorporation reaction would require an enzymatic activity, and at least the first step of the incorporation Mutations in the -5 to -1 region of Yap3p were constructed by two-step PCR-based mutagenesis. B, analysis of membrane-bound forms of the wild type Yap3p protein by phase separation and PI-PLC. Yap3p extracted in a detergent phase of phase separation with Triton X-114 was treated with (ϩ) and without (Ϫ) PI-PLC and then separated into a detergent (D) and an aqueous phase (A). Proteins in each fraction were treated with (ϩ) and without (Ϫ) Endo H and then precipitated with trichloroacetic acid. The proteins were analyzed by Western blotting using anti-HA antibody. T, total cell lysates. C, analysis of membrane-bound forms of the wild type and mutant Yap3p proteins. The same procedure as described in B was carried out for the mutant Yap3p proteins. Only parts of results are shown here. ϩ and Ϫ represent ϩ, A, ϩ and Ϫ, A, Ϫ, respectively, for treatment with PI-PLC, aqueous/detergent phase in two-phase separation and treatment with Endo H, respectively, as explained in B. D, analysis of cell wall-bound forms of the wild type and mutant Yap3p proteins. The Yap3p proteins released from the cell wall with laminarinase were treated with Endo H and then analyzed by Western blotting using anti-HA monoclonal antibody. reaction, detachment of a GPI moiety, would require proper binding of the enzyme to the GPI-attached protein. The -4/5 and -2 rule may be related to this binding. Mutations in the -minus 5-1 sequence that reduced the cell wall incorporation did not increase the amount of the fusion proteins secreted into the medium (data not shown). This suggests that the low cell wall incorporation observed in these mutants is probably due to reduction of the GPI detachment efficiency caused by insufficient binding between the enzyme and the GPI-attached protein.
A structural analysis of GPI-dependent cell wall proteins has demonstrated that ␤-1,6-glucan is attached to the protein through a GPI remnant. Several lines of evidence indicate that the GPI remnant contains ethanolamine but does not contain glucosamine and phosphatidylinositol, suggesting that ␤-1,6glucan is bound to one of the mannose residues of the GPI moiety (6,37,51). From this structural information, it is possible that the GPI moiety may be removed by cleavage either among mannose residues or between the mannose and myoinositol residues, and then the resulting end is directly bound to ␤-1,6-glucan. Alternatively, as shown in the Gce1p protein (38), the GPI moiety may be lipolytically cleaved by an enzyme like PI-PLC. In this case, trimming of myo-inositol and glucosamine residues may be required before the binding of it to ␤-1,6-glucan, because the enzyme is expected to cleave between phosphatidylinositol and lipid. Identification of a gene/enzyme responsible for the incorporation of GPI-attached proteins into the cell wall is needed for understanding the molecular mechanism for the incorporation, and our current efforts are focused on the identification of the gene/enzyme.