Essential Hydrophilic Carboxyl-terminal Regions Including Cysteine Residues of the Yeast Stretch-activated Calcium-permeable Channel Mid1*

The yeast Saccharomyces cerevisiae MID1 gene encodes a stretch-activated Ca2+-permeable nonselective cation channel composed of 548 amino acid residues. A physiological role of the Mid1 channel is known to maintain the viability of yeast cells exposed to mating pheromone, but its structural basis remains to be clarified. To solve this problem, we identified the mutation sites of mid1 mutant alleles generated by in vivo ethyl methanesulfonate mutagenesis and found that two mid1 alleles have nonsense mutations at the codon for Trp441, generating a truncated Mid1 protein lacking two-thirds of the intracellular carboxyl-terminal region from Asn389 to Thr548. In vitro random mutagenesis with hydroxylamine also showed that the carboxyl-terminal region is essential. To identify the functional portion of the carboxyl-terminal region in detail, we performed a progressive carboxyl-terminal truncation followed by functional analyses and found that the truncated protein produced from the mid1 allele bearing the amber mutation at the codon for Phe522 (F522Am) complemented the mating pheromone-induced death phenotype of themid1 mutant and increased its Ca2+ uptake activity to a wild-type level, whereas N521Am did not. This result indicates that the carboxyl-terminal domain spanning from Asn389 to Asn521 is required for Mid1 function. Interestingly, this domain is cysteine-rich, and alanine-scanning mutagenesis revealed that seven out of 10 cysteine residues are unexchangeable. These results clearly indicate that the carboxyl-terminal domain including the cysteine residues is important for Mid1 function.

From the ‡Department of Biology, Tokyo Gakugei University, 4-1-1 Nukui kita-machi, Koganei-shi, Tokyo 184-8501, Japan and §CREST, Japan Science and Technology Corporation, 4-1-8 Honcho, Kawaguchi-shi, Saitama 332-0012, Japan The yeast Saccharomyces cerevisiae MID1 gene encodes a stretch-activated Ca 2؉ -permeable nonselective cation channel composed of 548 amino acid residues. A physiological role of the Mid1 channel is known to maintain the viability of yeast cells exposed to mating pheromone, but its structural basis remains to be clarified. To solve this problem, we identified the mutation sites of mid1 mutant alleles generated by in vivo ethyl methanesulfonate mutagenesis and found that two mid1 alleles have nonsense mutations at the codon for Trp 441 , generating a truncated Mid1 protein lacking two-thirds of the intracellular carboxyl-terminal region from Asn 389 to Thr 548 . In vitro random mutagenesis with hydroxylamine also showed that the carboxyl-terminal region is essential. To identify the functional portion of the carboxyl-terminal region in detail, we performed a progressive carboxyl-terminal truncation followed by functional analyses and found that the truncated protein produced from the mid1 allele bearing the amber mutation at the codon for Phe 522 (F522Am) complemented the mating pheromone-induced death phenotype of the mid1 mutant and increased its Ca 2؉ uptake activity to a wild-type level, whereas N521Am did not. This result indicates that the carboxyl-terminal domain spanning from Asn 389 to Asn 521 is required for Mid1 function. Interestingly, this domain is cysteine-rich, and alaninescanning mutagenesis revealed that seven out of 10 cysteine residues are unexchangeable. These results clearly indicate that the carboxyl-terminal domain including the cysteine residues is important for Mid1 function.
The molecular mechanisms by which mechanical signals direct biological responses remain a frontier in the field of signal transduction. Electrophysiological studies have indicated that mechanotransduction can be mediated by ion channels that open or close in response to mechanical stimuli (1)(2)(3)(4)(5). Such channels play essential roles in a wide variety of activities including cell volume control, development, morphogenesis, and neuronal signaling underlying touch, hearing, and balance. However, eukaryotic mechanosensitive ion channels have not been cloned until recently, and thus little is understood of their structures and functions at the molecular level. Recently, genes encoding eukaryotic mechanosensitive channels or their candidates have been found in some species, including MID1 in budding yeast (6,7), MEC4 in the nematode (8,9), SIC in the rat (10), and NOMPC in the fly (11). A MID1 homologue in fission yeast, ehs1 ϩ /yam8 ϩ , has been reported (12,13).
Ca 2ϩ signaling constitutes an important backbone pathway, which is essential for a wide variety of cellular events. In Saccharomyces cerevisiae, Ca 2ϩ has essential roles in the mating process induced by the mating pheromone, ␣-factor (14). This process is divided into early and late stages, and Ca 2ϩ seems to be required for the late stage only (15). In the early stage of this process, the mating pheromone binds to its cell surface receptor coupled with a heterotrimeric GTP-binding protein that activates a mitogen-activated protein kinase pathway, and the cells are eventually arrested in the G 1 phase of the cell cycle (16). In the late stage, at about 30 min after the binding of the mating pheromone, the cells differentiate into morphologically distinct cells having a mating projection, called shmoos, in which the cell wall and plasma membrane should be rearranged to support the polarized growth. This change accompanies Ca 2ϩ influx necessary for the viability of shmoos (14). When incubated in Ca 2ϩ -deficient medium, wildtype cells can grow normally in the absence of the mating pheromone but die after differentiation into shmoos if the mating pheromone is added (14).
Mutants showing the mid (mating pheromone-induced death) phenotype have been isolated, one of which is the mid1 mutant (17). This mutant dies in response to the mating pheromone even in Ca 2ϩ -containing media because of a deficiency of Ca 2ϩ influx and is rescued when high concentrations of CaCl 2 are supplemented. Other yeast mutants that show the mid phenotype have been identified. Those include mutants deficient in calmodulin (18), Ca 2ϩ /calmodulin-dependent protein kinases (18), calcineurin (18,19), and a homologue of the ␣ 1 subunit of mammalian, voltage-gated Ca 2ϩ channels (20,21).
The MID1 gene encodes a stretch-activated Ca 2ϩ -permeable channel composed of 548 amino acid residues and has four hydrophobic segments named H1-H4 and 16 putative N-linked glycosylation sites (17). Although the Mid1 polypeptide has no overall sequence similarity to known ion channels, the H2 and H4 segments are partially similar to the transmembrane seg-ments of known ion channels (17,22). The region downstream of the H4 segment, which is the carboxyl-terminal region, is hydrophilic, considerably cysteine-rich, and located in the cytoplasm (see Fig. 1). Because little is known about structurefunction relationships in eukaryotic stretch-activated channels including the Mid1 channel, we performed mutational analyses on the identification of essential domains necessary for Mid1 function by taking advantage of the ease of molecular genetic approaches in yeast. Here, we report that the carboxyl-terminal region has a positively regulatory domain for Mid1 function. In addition, we have identified seven functionally important cysteine residues located in the carboxyl-terminal region.

EXPERIMENTAL PROCEDURES
Yeast Strains and Media-The yeast strains used in this study are listed in Table I. Rich media and a synthetic medium, SD, were prepared as described by Sherman et al. (23). Because SD medium contains 680.2 M CaCl 2 and 0.8 M calcium pantothenate, to make the Ca 2ϩdeficient medium SD-Ca, CaCl 2 was omitted, and calcium pantothenate was replaced by sodium pantothenate. SD.Ca100 medium was prepared by adding 100 M CaCl 2 to SD-Ca medium. LB and 2ϫ YT media were prepared by the method of Sambrook et al. (24).
Isolation of mid1 Alleles by the Gap Repair Method-To isolate mid1 alleles, the gap repair method (25) was employed. The plasmid YCpMID1-21 (17) containing the wild-type MID1 gene and the selection marker LEU2 was digested with restriction enzymes, HindIII and SnaBI, and digested plasmids without MID1 were introduced into seven mid1 mutants by a lithium acetate method (26). The transformants were plated onto SD-leucine plates, and then the gap-repaired plasmids were rescued from Leu ϩ transformants by the method of Holm et al. (27). The recovered plasmids were propagated in Escherichia coli, purified, and subjected to DNA sequencing (28) with a Thermo Sequenase fluorescent labeled primer cycle sequencing kit with 7-deaza-dGTP (Amersham Biosciences) and a DNA sequencer, DSQ 2000L (Shimadzu, Kyoto, Japan).
In Vitro Random Mutagenesis with Hydroxylamine-The YCpMID1-23 plasmid (17) containing the wild-type MID1 gene was mutagenized with hydroxylamine (HA) 1 with a mutation frequency of 0.9% by the method of Adams et al. (29). The HA solution contained 0.35 g of HA-HCl and 0.09 g of NaOH in 5 ml of H 2 O. Ten micrograms of YCpMID1-23 was added to 500 l of the HA solution sterilized through a filter (Millex-GV, 0.22 m, Millipore Corp., Bedford, MA), and the mixture was incubated for 20 h at 30°C. On the following day, 10 l of 5 M NaCl and 50 l of 1 mg/ml bovine serum albumin were added to the mixture to stop the reaction, and the plasmid DNA was precipitated by ethanol. The mutagenized and purified plasmids were introduced into the E. coli strain XL1-Blue for amplification. The amplified plasmids were introduced into the mid1-1 mutant, and the viability of the transformants was examined by the methylene blue plate method (see below). The plasmids incapable of complementing the mid1-1 mutation were treated with BamHI and XhoI to cut out the DNA fragment containing the MID1 coding region as well as the 5Ј-and 3Ј-flanking regions (150 bp upstream of the initiation codon and 858 bp downstream of the termination codon). The DNA fragments were inserted into the vector portion of unmutagenized YCpMID1-23 treated with BamHI and XhoI and tested again for complementing activity. This procedure eliminates possible mutations in the vector. The plasmids finally identified were subjected to DNA sequence analysis (28).
Construction of mid1 Mutants by Site-directed Mutagenesis-Sitedirected mutagenesis of the MID1 gene in the plasmid YCpMID1-23 (17) was performed with the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) according to the method described by the manufacturer. Mutagenic primers (Table II) were synthesized to order by Sawady Technology Co., Ltd. (Tokyo, Japan).
Determination of mid1 Complementation Activity by Methylene Blue Methods-For screening mutants bearing a nonfunctional mid1 gene after in vitro random mutagenesis with HA, we employed a methylene blue plate (MBP) method (17). This method is based on the fact that viable cells, but not inviable cells, reduce methylene blue to give the colorless leukomethylene blue (30,31). Thus, the viable cells remain white, and inviable cells stain blue. Cells of the mid1 mutant transformed by HA-treated YCpMID1-23 (LEU2) were plated onto SD plates lacking leucine to give about 1000 colonies/plate and incubated for 2-3 days at 30°C. The Leu ϩ transformants on the plates received 0.1 ml of 1 mM ␣-factor and 7 ml of SD medium containing 0.5% agar and 0.01% methylene blue and were incubated for several days. At this stage, the colonies mainly with dead cells become blue but still have viable cells in parts, whereas colonies formed by cells transformed with intact YCpMID1-23 remain white. The blue colonies were selected, suspended in 0.1 ml of SD medium, streaked onto an SD plate, and incubated for 2 days. To avoid selecting cells contaminated from other colonies at the step of pouring 0.5% agar described above, six colonies from each single blue colony were then subjected to quantitative viability assay using the methylene blue liquid (MBL) method (14).
In the MBL method, 0.1 ml of exponentially growing culture in SD.Ca100 medium or that treated with ␣-factor for 4 or 8 h was mixed with an equal volume of 0.01% methylene blue, 2% sodium citrate solution, and the number of viable white cells and inviable blue cells was counted under a differential interference-contrast microscope 1 The abbreviation used is: HA, hydroxylamine; MBL, methylene blue liquid; GST, glutathione S-transferase. (Olympus, Tokyo, Japan). The viability was expressed as the number of viable white cells as a percentage of the total number of cells. Colonies that produced about 30% viable culture 8 h after receiving ␣-factor were selected for further analyses.
Preparation of Anti-Mid1 Antibodies and Immunoblot Analysis-Rabbit polyclonal antibodies to the Mid1 protein were raised against the glutathione S-transferase-Mid1 (GST-Mid1) fusion protein. To obtain the GST-Mid1 protein, we constructed pGEX-6P-2-MID1, an E. coli plasmid that expresses it under the control of the tac promoter. Fusion of Mid1 with GST was necessary, because Mid1 was toxic to the cells when expressed alone. pGEX-6P-2 (Amersham Biosciences) was digested with BamHI and ligated with a linker DNA containing the NsiI site (5Ј-pGATCCATGCATCTGCAGATGCATG-3Ј, where the NsiI sites are underlined) to create the pGEX-6P-2-N vector. pGEX-6P-2-MID1 was constructed by inserting the 1.7-kb NsiI-NsiI fragment that contains the entire MID1 open reading frame from pBSMID1-N2 into the NsiI site of pGEX-6P-2-N. The E. coli strain JM109 was transformed with pGEX-6P-2-MID1, and the transformant was incubated to the stationary phase in 2ϫ YT medium at 37°C. The culture was diluted to one-tenth by 2ϫ YT medium and incubated for 1 h. Then the culture received 1 mM isopropyl-1-thio-␤-D-galactopyranoside and was incubated for 3 h at 37°C to induce the expression of GST-Mid1. The cells were harvested and disrupted by sonication. The protein extracts were fractionated by centrifuging at 20,000 ϫ g, and the pellets were washed twice with PBS containing 2% Triton X-100 (Bio-Rad). Because the GST-Mid1 fusion protein was present in the inclusion body, 7 M urea solution made in 50 mM Tris-HCl (pH 8.5), 10 mM dithiothreitol, and a protease inhibitor mixture (catalog no. P8849; Sigma) was used to solubilize it. The solubilized materials were mixed with an equal volume of 2ϫ SDS-sample buffer and separated on 7.5% SDS-polyacrylamide gel. After staining with Coomassie Brilliant Blue, the desired gel strip containing the 97-kDa GST-Mid1 protein was collected, and it was electrophoretically purified by a model 422 electroeluter (Bio-Rad). Protein concentration was adjusted to 1 mg/ml or higher by using an ultrafiltration filter, Microcon YM-50 (Millipore Corp.). Preparation of rabbit anti-Mid1 antibodies was ordered from Asahi Techno Glass, Inc. (Chiba, Japan). The antibodies specifically recognized the Mid1 protein, as revealed by immunoblot analysis on yeast whole cell proteins, and were used at a dilution 1:10,000. Immunoblot analysis was performed as described previously (17).
Measurement of Ca 2ϩ Accumulation-Ca 2ϩ accumulation in the cells was measured according to the method described by Iida et al. (14) except that SD.Ca100 medium was used instead of SD or SD-Ca medium. Two-hybrid Assay-The two-hybrid assay was used to examine the possible protein-protein interactions as described previously (33,34). Combinations of control and fusion plasmids were transformed into the PJ69-4A strain (34). Transformants were streaked onto agar plates containing synthetic medium without histidine or adenine. The transformants were also examined for ␤-galactosidase activity by the filter method (36) and the quantitative liquid method (37).

Analysis of mid1 Alleles Induced by EMS in Vivo Reveals the Importance of the H2 and Carboxyl-terminal Domains-To
identify the amino acids required for the function of Mid1, we isolated and analyzed the seven EMS-induced mid1 alleles that we reported previously (17). The mid1 alleles were recovered by the gap repair method as described under "Experimental Procedures," and the gap-repaired mid1 alleles that did not complement the mid phenotype (see above) of the corresponding mid1 mutants were selected and then subjected to DNA sequencing and computational polypeptide analysis. As shown in Fig. 2, the mid1-1 allele product had two mutations, Y254H (TAC to CAC) and W441Op (TGG to TGA). Analysis of the single mutations generated by site-directed mutagenesis showed that the Y254H mutation was a silent mutation and that the W441Op mutation was responsible for the mid phenotype associated with the mid1-1 mutation (data not shown). The mid1-2 and mid1-5 allele products had a common mutation, Y254H (TAC to CAC), indicating that the two alleles have no loss-of-function mutation in the coding region. The mid1-3 allele product had G104D (GGC to GAC) and Y254H (TAC to CAC) mutations. Again, analysis of the single mutations generated by site-directed mutagenesis showed that only the G104D mutation was responsible for the mid phenotype associated with the mid1-3 mutation (data not shown). It is of interest that Gly 104 is located in the hydrophobic segment H2. The mid1-4 allele product had Y254H (TAC to CAC) and W441Am (TGG to TAG) mutations. Note that this allele product is the same as the mid1-1 allele product with different stop codons. The mid1-6 allele had a deletion of a nucleotide in the codon for Met 109 , resulting in a frameshift mutation producing a protein lacking about four-fifths of the Mid1 polypeptide after Gln 108 with two extra amino acid residues, Cys and Pro. The mid1-7 allele product had the C498Y (TGC to TAC) mutation only. These results indicate that Gly 104 , Cys 498 , and the carboxyl-terminal region (Trp 441 -Thr 548 ) are required for the function of Mid1.
Analysis of mid1 Alleles Induced by Hydroxylamine in Vitro Reveals the Importance of the Carboxyl-terminal Region-To further identify the amino acids or regions required for the function of Mid1, we analyzed hydroxylamine-induced mid1 alleles. The plasmid YCpMID1-23 (17) was mutagenized with hydroxylamine in vitro with a mutation frequency of 0.9% according to Adams et al. (29), and the mutagenized plasmid was examined for the ability to complement the mid1-1 mutation by the MBP method followed by the MBL method as described under "Experimental Procedures." Thirty-seven plas- Mid1, we performed a progressive carboxyl-terminal truncation analysis. The MID1 gene was mutagenized by site-directed mutagenesis to introduce a stop codon, generating progressive carboxyl-terminal truncation of Mid1 to Trp 441 . The YCpbased, low copy plasmids bearing the mutagenized MID1 genes were examined for their ability to complement the mid1-1 mutation that has been intensively studied previously (17). Cells of the mid1-1 mutant having each one of the plasmids were incubated with ␣-factor for 8 h, and the viability was determined by the MBL method. Fig. 3 shows that W441Op, V445Op, C450Op, Y457Am, K463Op, Y481Am, V495Op, F502Am, and L510Am did not complement the mid1-1 mutation, whereas T526Op and V541Op did. The mutant Mid1 proteins that have no complementing ability were present at a level comparable with the wild-type protein, as revealed by immunoblot analysis (Fig. 4A), indicating that the inability is not due to instability of the mutant proteins. These results indicate that the region between Thr 526 and the carboxyl terminus (Thr 548 ) is dispensable and suggests that there is a boundary defining the shortest functional Mid1 between Leu 510 and Thr 526 . It is to be noted that the carboxyl-terminal region between the H4 segment and Thr 526 has interesting features (Fig. 1B), such as cysteine-rich regions including Cys 498 , which has been found to be indispensable (Fig. 2), a casein kinase 2 phosphorylation motif ( 511 TTED 514 , where a putative phosphorylation site is underlined), a sheet-turn-sheet structure ( 512 TEDL-LYQSYNFYMDT 526 , where amino acid residues underlined contribute to the turn), and a helix-loop-helix structure (Phe 408 -Val 445 ) similar to the EF-hand structure (see "Discussion").
To examine the possible contribution of the putative casein kinase 2 phosphorylation site to Mid1 function, we further carried out mutational analysis. The results showed that the mutant alleles, in which the putative phosphorylation site Thr 511 was substituted with Ala or Val (T511A and T511V), complemented the mid phenotype of the mid1-1 mutant (Fig.  5A), indicating that phosphorylation of this site is not a factor for Mid1 function.
We then examined the possible involvement of the sheetturn-sheet structure in Mid1 function by replacing Tyr 520 with Ala or Gly to disrupt this structure. As shown in Fig. 5A, both the Y520A and Y520G mutant alleles did not complement the mid1-1 mutation, suggesting that this structure may be required for Mid1 function. In contrast, carboxyl-terminal truncation into this structure up to Phe 522 did not affect the complementing ability of the truncated alleles (Fig. 5B).
Finally, this carboxyl-terminal truncation experiment defined the carboxyl-terminal boundary for the shortest functional Mid1. The F522Am allele, but not the N521Am allele, complemented the mid phenotype (Fig. 5B), indicating that the site between Asn 521 and Phe 522 is the boundary. In other words, the Met 1 -Asn 521 polypeptide is the shortest functional Mid1 in terms of carboxyl-terminal truncation.
Complementing Ability Correlates with Ca 2ϩ Accumulation Activity-Because another phenotype of the mid1 mutants has a low activity in Ca 2ϩ uptake (17), we measured the Ca 2ϩ accumulation in cells having mid1 mutations at or near the boundary as described above (i.e. at or near Asn 521 and Phe 522 ). Fig. 6 shows that L510Am, Y520A, Y520G, Y520Am, and N521Am, which do not complement the mid1 mutation, had low activities in accumulating Ca 2ϩ , comparable with or close to a mid1-1 mutant level, while F522Am and T526Op, which complement the mid phenotype, accumulated Ca 2ϩ normally. This result indicates that the complementing ability of the mid phenotype correlates with the activity of Ca 2ϩ accumulation.
Seven Cysteine Residues Are Required for Mid1 Function-The C498Y mutation found in the mid1-7 allele (Fig. 2) implies that some cysteine residues in the carboxyl-terminal region are important. We therefore exchanged each one of the cysteine residues in that region for alanine residues and examined the ability of each mutant allele to complement the mid1-1 mutation. As shown in Fig. 7, C417A, C431A, C434A, and C498A mutants did not have the complementing ability just like a negative control (the mid1-1 mutant itself), whereas C443A, C450A, and C487A had full complementing ability. C491A, C506A, and C531A had intermediate complementation ability. Thus, seven out of 10 cysteine residues are necessary for Mid1 function. Immnoblot analysis showed that the level of the mutant Mid1 proteins is essentially the same as that of the wildtype proteins (Fig. 4B). DISCUSSION We have identified the carboxyl-terminal region and amino acid residues important for the function of the Mid1 channel by conventional mutagenesis with EMS, in vitro random mutagenesis with hydroxylamine, and site-directed in vitro mutagenesis. A previous study has indicated that the carboxylterminal region after the H4 domain is hydrophilic, cysteine-rich, and intracellular (17). Thus, the region has been postulated to be a regulatory region for this channel. Our present study has clearly shown that this region is essential for Mid1 function. This region is indeed required for the complementing ability and Ca 2ϩ uptake activity, and the essential carboxyl-terminal region excludes the distal portion from Phe 522 to Thr 548 .
The essential carboxyl-terminal region has at least three motifs, including a putative casein kinase 2 phosphorylation motif from Thr 511 to Asp 514 , an EF-hand-like structure from FIG. 2. Diagram of mid1 alleles. The four hydrophobic segments were indicated by gray boxes. The mid1-1 and mid1-4 alleles produce proteins with 440 amino acid residues, lacking most of the carboxyl terminus. The Y254H mutation is a silent mutation. The mid1-2 and mid1-5 alleles produce a protein having the silent mutation in the coding region. The mid1-3 allele produces a protein with the G104D mutation in the hydrophobic segment H2 as well as the silent mutation. The mid1-6 allele has a frameshift mutation, producing a protein with only 110 amino acid residues. The mid1-7 allele produces a protein having the C498Y mutation. Stop codons, TAG and TGA, are called amber (Am) and opal (Op) mutations, respectively. The asterisks indicate silent mutations. Amino acids are abbreviated as follows: Y, tyrosine; H, histidine; G, glycine; D, aspartic acid; C, cysteine.
Phe 408 to Val 445 , and a sheet-turn-sheet motif from Thr 512 to Thr 526 .
As mentioned before, the putative casein kinase 2 phosphorylation motif is not required for Mid1 function (Fig. 5A). We have found by the 45 Ca 2ϩ overlay technique that the EF-handlike structure (Phe 408 -Val 445 ) does not bind 45 Ca 2ϩ under the conditions under which calmodulin, a positive control, binds it (data not shown), suggesting that this structure is not a high affinity Ca 2ϩ -binding site. This is probably because the structure has crucial amino acid deletions and substitutions. However, this structure could still be important, because Cys 417 , Cys 431 , and Cys 434 in the E-or F-like helix structure are essential for Mid1 function (Fig. 7). In addition, it is known that the distorted EF-hand can function as a protein-protein interaction site (38 -40). Thus, the EF-hand-like structure of the Mid1 channel could contribute to interact with other proteins or with Mid1 protein itself.
To examine the possible self-interaction between the Mid1 proteins via the carboxyl-terminal regions, we performed the Gal4-based two-hybrid assay on the following three combinations (i.e. interactions between (a) the carboxyl termini, (b) the carboxyl terminus and the central domain (H2-H4), and (c) the H2-H4 domains (for details, see "Experimental Procedures")). The results were all negative (data not shown). Therefore, we still do not know unambiguously if the carboxyl-terminal region as well as the central region between H2 and H4 contributes to a possible self-interaction.
This study has revealed the importance of cysteine residues in the essential carboxyl-terminal region. There are 10 cysteine residues in this region, and they are all conserved in the Mid1 homologue of a fission yeast Schizosaccharomyces pombe pro-  (46), was used as a loading standard. A, a series of progressive carboxylterminal truncations; B, a series of alanine-scanning mutagenesis on the cysteine-rich regions.
FIG. 5. Complementing ability of Mid1 proteins having a mutation in putative functional motifs. A, cells expressing a mutant Mid1 protein having a mutation in a putative casein kinase 2 phosphorylation site (T511A and T511V) or that in a putative sheet-turn-sheet structure (Y520A and Y520G) were incubated with 6 M ␣-factor for 8 h, and the viability of the cells was examined by the MBL method. Note that Tyr 520 is in the turn structure. B, mutant Mid1 proteins that have a nonsense mutation in the putative sheet-turn-sheet structure were expressed in the mid1-1 mutant, and the viability of the cells was examined as described above. tein, Ehs1/Yam8 (12,13 . Indeed, as discussed above, the essential cysteine residues Cys 417 , Cys 431 , and Cys 434 are in the EF-hand-like structure. The latter group of cysteine residues, Cys 491 , Cys 498 , Cys 506 , and Cys 531 , may contribute to a functional unit. When the partially inactive mutations C491A and C531A were combined in a cell, the resultant double mutant became fully inactive, comparable with the mid1-1 mutant (data not shown). This suggests that Cys 491 and Cys 531 take part in a common function.
Cysteine-rich regions are known to be present in many ion channels, such as DEG/ENaC channels including the candidate mechanosensitive ion channel MEC-4 of Caenorhabditis elegans (41,42) and ATP-activated cation channels (43), and are thought to form disulfide bonds or to serve for protein-protein interactions to modify channel activity. Although the role of these cysteine-rich regions has not been fully elucidated, it is postulated that the disulfide bonds of ATP-activated ion channels may stabilize a ligand-binding pocket (43). The cGMPgated potassium channel contains a cysteine-rich region that is highly homologous to integrin ␤5 (44), implying a possible protein-protein interaction. Thus, the cysteine-rich regions of the Mid1 carboxyl-terminal region could regulate its channel activity through the interaction with cellular components.
Our examination on the possible involvement of the sheetturn-sheet structure (Thr 512 -Thr 526 ) in Mid1 function provided a complicated result. When the turn was disrupted by replacing Tyr 520 with Ala or Gly, both the resultant Y520A and Y520G alleles did not complement the mid1-1 mutation (Fig. 5A). This suggests that the sheet-turn-sheet structure may be important for Mid1 function. However, carboxyl-terminal truncation into this structure up to Phe 522 , which resulted in a loss of one-third of the structure, did not injure Mid1 function (Fig. 5B). These results imply two possibilities; the sheet-turn-sheet structure is required for Mid1 function when the carboxyl-terminal region is intact. Alternatively, this structure is not responsible for Mid1 function, and the Y520A and Y520G mutations resulted in a conformational change leading to the inactivation of Mid1. In this case, the carboxyl-terminal region downstream of this structure might have some inhibitory role, because the truncated alleles, such as F522Am, are normally functional.
We have also identified the important amino acid residue, Gly 104 , in the hydrophobic segment H2. The G104D mutation in the mid1-3 mutant renders yeast cells defective in maintaining cell viability and in Ca 2ϩ uptake activity (17). Glycine residues in the transmembrane segment of ion channels are known to be responsible for the channel activity. For example, a mutation in the conserved glycine residue in the first transmembrane segment of the ␤ 2 subunit of the ␥-aminobutyric acid type A (GABA A ) receptor, a ligand-gated Cl Ϫ channel, eliminates agonist-induced enhancement of GABA current (45). The glycine residue in the Mid1 H2 segment could be responsible for channel gating.
Our present study with molecular genetic approaches should provide clues to understand the structure-function relationship of eukaryotic stretch-activated channels with further analyses by electrophysiological methods. Complementing ability of Mid1 proteins having a mutation in cysteine residues. Ten cysteine residues were substituted with alanine residues. The viability of the cells expressing a mutant Mid1 protein was determined as described in the legend to Fig. 3.