Coordination Between Fission Yeast Glucan Formation and Growth Requires a Sphingolipase Activity

Corresponding author: Kathleen L. Gould, HHMI and Department of Cell Biology, B2309 MCN, Vanderbilt University School of Medicine, 1161 21st Ave. S., Nashville, TN 37232., kathy.gould{at}mcmail.vanderbilt.edu (E-mail)

Communicating editor: P. RUSSELL

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

css1 mutants display a novel defect in Schizosaccharomyces pombe cell wall formation. The mutant cells are temperature-sensitive and accumulate large deposits of material that stain with calcofluor and aniline blue in their periplasmic space. Biochemical analyses of this material indicate that it consists of - and ß-glucans in the same ratio as found in cell walls of wild-type S. pombe. Strikingly, the glucan deposits in css1 mutant cells do not affect their overall morphology. The cells remain rod shaped, and the thickness of their walls is unaltered. Css1p is an essential protein related to mammalian neutral sphingomyelinase and is responsible for the inositolphosphosphingolipid-phospholipase C activity observed in S. pombe membranes. Furthermore, expression of css1⁺ can compensate for loss of ISC1, the enzyme responsible for this activity in Saccharomyces cerevisiae membranes. Css1p localizes to the entire plasma membrane and secretory pathway; a C-terminal fragment of Css1p, predicted to encode a single membrane-spanning segment, is sufficient to direct membrane localization of the heterologous protein, GFP. Our results predict the existence of an enzyme(s) or process(es) essential for the coordination of S. pombe cell wall formation and division that is, in turn, regulated by a sphingolipid metabolite.

SCHIZOSACCHAROMYCES pombe cells, like other yeast, are surrounded by a rigid cell wall that provides mechanical strength and protection from environmental stresses (CABIB et al. 1997 ). The wall is required for the formation and maintenance of cell shape and is remodeled continuously during the cell cycle to allow cell expansion. Cell wall remodeling also occurs in response to extracellular cues that trigger shape changes, such as nutrient deprivation and exposure to mating factors. During vegetative growth, S. pombe cells are rod shaped and grow by tip elongation, first at the old end and then at both ends. Subsequently, cell wall deposition occurs during septation and the two new ends are thus sealed off (MITCHISON and NURSE 1985 ). Patches of F-actin are concentrated at these areas of growth, and it is a widely held view that they direct wall formation to these sites (MARKS et al. 1987 ; NURSE 1994 ). Electron micrographic studies of regenerating S. pombe protoplasts have substantiated this view by demonstrating an intimate association between F-actin structures and sites of new cell wall formation (OSUMI et al. 1998 ).

The carbohydrate composition of the S. pombe cell wall has been known for some time to consist of -galactomannans; alkali-soluble -(1,3)-, ß-(1,6)-, and ß-(1,3)-linked glucans; and an alkali-insoluble ß-(1,3)-linked glucan with the ß-(1,3)-linked glucans forming the majority of cell wall polysaccharide (BUSH et al. 1974 ; MANNERS and MEYER 1977 ). These components are arranged in distinct layers that can be visualized by electron microscopy (KOPECKA et al. 1995 ; OSUMI et al. 1998 ). There is an outer shell comprised of glycoproteins, a middle layer of amorphous glucan material, and an inner layer of densely woven glucan fibrils. Several S. pombe galactosyltransferases, Golgi enzymes required for the production of cell wall glycoproteins, have been identified and characterized (CHAPPELL and WARREN 1989 ; CHAPPELL et al. 1994 ; YOKO-O et al. 1998 ). Several enzymes also have been identified that are required for the synthesis of - and ß-glucans. There are at least five -glucan synthases of which one, termed Mok1p/Ags1p, has been characterized (HOCHSTENBACH et al. 1998 ; KATAYAMA et al. 1999 ). There is also a family of ß-glucan synthases in S. pombe. One of these, Cps1p, is critical for the formation of the septum (ISHIGURO et al. 1997 ; LE GOFF et al. 1999 ; LIU et al. 1999 ). Another, Bgs2p, is essential for maturation of the spore wall (LIU et al. 2000 ; MARTIN et al. 2000 ). Mok1p/Ags1p and Cps1p are essential for cell wall integrity and viability of vegetatively growing cells. In their absence, the cell wall becomes less well organized, thicker, and more sensitive to degradation, and the cells become rounded, swollen, and eventually lyse unless provided with osmostabilizing medium. The activities of the glucan synthases are regulated in a complex manner by the activities of Rho family GTPases and protein kinase C's (reviewed in ARELLANO et al. 1999 ). Under- or overproduction of these regulators dramatically affects S. pombe morphology and cell wall integrity. Indeed, in such studies, considerable progress was made in identifying and understanding the signaling pathways governing the synthesis of cell wall components. In contrast, virtually nothing is understood in S. pombe about how the various components of the wall are organized into a functional unit subsequent to their synthesis.

Here, we present the isolation and characterization of css1 mutants that display a novel defect in cell wall formation. In these mutants, a large excess of - and ß-glucans accumulates in the periplasmic space without being incorporated into the cell wall and without affecting the rod shape of the cells. We present evidence that Css1p, related to human neutral sphingomyelinase, is an essential membrane protein that functions as an inositolphosphosphingolipid-phospholipase C in S. pombe. Css1p can also substitute for loss of this enzyme activity in Saccharomyces cerevisiae encoded by ISC1 (WELLS et al. 1998 ; SAWAI et al. 2000 ). Many agonists and stresses activate mammalian sphingomyelinases with subsequent hydrolysis of sphingomyelin in the plasma membrane to generate phosphocholine and ceramide, an intracellular signaling molecule (KOLESNICK and KRONKE 1998 ; KRONKE 1999 ; LEVADE and JAFFREZOU 1999 ). The agonists and stresses include cytokines, interleukin 1ß and Fas ligands, chemotherapeutic drugs, environmental stresses, and injury or infections. Ceramide is thought to regulate signaling pathways that control growth arrest, apoptosis, senescence, and immune responses in mammals. Processes regulated by sphingolipases have not been identified in fungi (DICKSON 1998 ). Our results suggest that Css1p hydrolyzes complex sphingolipids to yield ceramide, which regulates a pathway or process essential for the coordination between cell wall formation and growth in S. pombe. This is the first example of a sphingolipase that is essential for fungal cell growth.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Yeast strains, methods, and media:
S. pombe strains used in this study are listed in Table 1. They were grown in yeast extract medium (YE) or minimal medium with appropriate supplements (MORENO et al. 1991 ). Crosses were performed on glutamate medium (minimal medium lacking ammonium chloride and containing 0.01 M glutamate, pH 5.6). Random spore and tetrad analyses were performed as described (MORENO et al. 1991 ). Transformations were performed by electroporation (PRENTICE 1992 ). For regulated expression of css1⁺ by the nmt1 promoter (MAUNDRELL 1993 ), cells were grown either in minimal media lacking thiamine to allow expression or with the addition of 5 µM thiamine to repress expression. Double mutant strains were constructed and identified by tetrad analysis or by random spore analysis. To test for dominance, the css1-3 strain was crossed to a mei1-102 strain (mei1-102 mutants will mate with h^- partners, but the resulting diploids will not sporulate) and stable diploids were obtained using complementation of nutritional markers.

The parental S. cerevisiae strain was diploid JK9-3d (homozygous for leu2-3, 112 ura3-52 rme1 trp1 his4) and MATa/MAT (HEITMAN et al. 1991 ). RCD319 is a MATa haploid derivative of JK9-3d carrying the yer019-1::Kan allele, which has bases -4 to +967 replaced by a lox-kanamycin-lox cassette (GULDENER et al. 1996 ). YPD (SHERMAN 1991 ) and S-leu (SKRZYPEK et al. 1998 ) media were used to culture S. cerevisiae cells from which membranes were made. S. cerevisiae cells were grown in PYED medium (SKRZYPEK et al. 1998 ) to radiolabel sphingolipids.

Plasmids and molecular biological techniques:
All plasmid manipulations and bacterial transformations were by standard techniques (SAMBROOK et al. 1989 ). Essential features of plasmid construction are described. All sequencing of plasmid DNA was performed using Thermo Sequenase (Amersham Life Sciences, Cleveland, OH) according to manufacturer's instructions. Yeast genomic DNA was isolated as described (MORENO et al. 1991 ; HOFFMAN 1993 ).

For expression of css1⁺ under control of the nmt1 promoter, a NdeI site was introduced at the initiating methionine codon by PCR amplification of the open reading frame (ORF) with Pfu polymerase. A BamHI site was also introduced just 3' of the termination codon. The PCR fragment was subcloned into pZERO (Invitrogen, San Diego). The css1⁺ ORF was then removed with XhoI and BamHI and subcloned into the pREPX series of fission yeast expression plasmids (BASI et al. 1993 ; FORSBURG 1993 ; MAUNDRELL 1993 ). For expression in S. cerevisiae, the same fragment was subcloned into the expression vector pRS425GAL1 (MUMBERG et al. 1994 ). By dropping out the internal NdeI fragment of pKG1560 (pREP1GFPcss1), the plasmid pKG1845 was created that fused green fluorescent protein (GFP) to Css1p residues 332–424. GFP fusions to amino acids 332–389 and 332–663 of Css1p were made by amplification of css1⁺ sequences with Pfu polymerase using oligonucleotides 5'-CGTCTTCATATGAGACTTCG-3' and either 5'-CTAGGATCCTTATCTACCAAATAGCAATCC-3' or 5'-CTAGGATCCTTACTTTAACCAAGCGGG-3', respectively. The PCR products were digested with NdeI and BamHI and subcloned into pREP1GFP (pKG1538) to generate pKG1528 and pKG1974. A GFP fusion protein to Css1p residues 1–331 was created by subcloning the css1⁺ NdeI fragment into pKG1538. A GFP fusion protein with Css1p residues 1–389 was created by inserting the css1⁺ NdeI fragment (encoding residues 1–331) into the unique NdeI site of pKG1528 to create pKG2138.

Genomic cloning and DNA sequencing:
The css1-3 strain was transformed with a S. pombe genomic library constructed in the plasmid vector pUR19 (BARBET et al. 1992 ), and three Ura⁺ colonies that could grow at 36° were isolated. Plasmid DNAs were recovered from these colonies and a comparison of the genomic DNA inserts by restriction mapping and Southern blotting revealed that they contained overlapping DNA sequences. The region of these clones responsible for complementation of css1-3 was narrowed by constructing subclones and testing them for their ability to rescue. The DNA sequences of the ends of the smallest rescuing fragment (MluI-MluI) were then determined and compared to sequences within the databases. At one end, the DNA sequence matched a terminal segment of cosmid c32F12 in the S. pombe genome sequencing project database. The other end did not match a sequence within this database but did match a region adjacent to the orp2⁺ gene (LEATHERWOOD et al. 1996 ). However, there was an 800-bp gap between the ends of these DNA fragments. The DNA sequence of this gap region was then determined on both strands using custom-synthesized oligonucleotides that allowed the identification of a single open reading frame.

For integration mapping, the genomic fragment shown in Fig 4A was subcloned into pJK210, an integrating vector that contains the ura4⁺ selectable marker (KEENEY and BOEKE 1994 ). This construct was linearized with NdeI and integrated into the genome of the css1-3 ura4-D18 strain by homologous recombination. In outcrosses of the integrant strain to wild type, the ura4⁺ marker segregated with a css1⁺ phenotype in 14 out of 14 tetrads analyzed.

View larger version (79K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Phenotype of the css1 mutants. css1-2 and css1-3 cells were grown at 25° to midlog phase in YE medium, shifted to 36° for the indicated number of hours, and fixed with formaldehyde. Wild type grown at 25° (A) and css1-2 cells (B and C) were then stained with calcofluor, and css1-3 cells were permeabilized and stained with phalloidin (D and F) and DAPI (E and G). Arrows indicate the regions accumulating the calcofluor-staining material. The percentage of css1-3 (H) and css1-2 (I) cells accumulating calcofluor-stainable material at one end, both ends, in the middle, or both at the tips and in the middle was quantified.

View larger version (42K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Physiological analysis of css1 mutant cells. Wild-type and css1-2 cells were grown at 25° and then shifted to 36° for the indicated number of hours. (A) Cell number was determined in triplicate. (B) To determine the percentage of survival, samples were taken at each time point after shift to 36°, cell number was determined, and the same number of cells was plated in triplicate. The number of colonies to form at 25° after 5 days was then quantified. (C and D) css1-2 cells were synchronized in early G2 phase by centrifugal elutriation and shifted to 36° for the indicated number of hours. The septation index was determined (C) and the positions where calcofluor-stainable material was deposited and accumulated were also determined (D). Abnormal deposits of material were observed in the middle of the cells and also at old tips. (E) The calcofluor-stainable material also binds the ß-glucan-specific dye aniline blue. Wild-type cells were grown at 25° and css1-2 cells were grown at 25° to midlog phase in YE medium, shifted to 36° for 12 hr, and then stained with aniline blue.

View larger version (28K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Composition of the cell wall in the css1 mutant. (A) The relative levels of [14C]glucose radioactivity incorporated into each cell wall polysaccharide in a 4-hr labeling are shown for the wild-type (972h-) and css1-2 (KGY2550) strains grown at different temperatures. Values are the means of five independent experiments with duplicated samples. Standard deviations for the total carbohydrate values are shown. (B) The levels of radioactivity from [14C]glucose incorporated into cell wall polysaccharides from wild-type and css1-2 mutants are shown. Values per milligram of wet weight are means of duplicate cultures. Measurements of individual cell wall fractions were done in triplicate for each culture. Cells were grown for 10 hr at 23° in YE, 1% glucose (1 µCi [U-14C]glucose/ml), and then harvested or shifted to 36° for 12 hr before harvesting. Cell walls were isolated and fractionated as described in MATERIALS AND METHODS. Exact values and standard deviations are shown in Table 2.

View larger version (122K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Cloning of css1+. (A) Genomic organization of css1+, constructs tested for css1-3 rescue, and deletion strategy. The ORF is indicated by the solid box with the direction of transcription indicated by the arrow. X, XbaI; M, MluI; H, HindIII; A, Acc651. (B) Amino acid sequence alignment of Css1p with S. cerevisiae Isc1p. Identical amino acids are indicated by vertical lines. The glutamic acid and histidine residues predicted to be involved in substrate binding and catalysis, respectively, of this family of enzymes are indicated with an arrowhead and a box. The residues mutated in the css1 mutant strains are indicated with asterisks. (C) A hydropathy plot of Css1p with the two potential transmembrane domains near the C terminus. (D) Css1p is a membrane-associated protein. css1myc (lanes 1, 3, 5, 7, 9, and 11) and wild-type (lanes 2, 4, 6, 8, 10, and 12) cells were lysed with no detergent in TEG buffer (lanes 1, 2, 7, and 8), 1% NP-40 in NP-40 buffer (lanes 3, 4, 9, and 10), or in 1% SDS lysis buffer, and boiled immediately (lanes 5, 6, 11, and 12). The lysates were fractionated into soluble (lanes 1–6) and insoluble (lanes 7–12) fractions. Css1p-myc was detected by immunoblotting with 9E10.

For sequencing of css1 mutant alleles, genomic DNA was prepared from the relevant strains and the css1 coding region was amplified by PCR with the following oligonucleotides: css1-F6 (5'-TTGACGGCAATAACTCCG-3') and css1-R6 (5'-TAAATTTATGATTTAGC-3'). The PCR products were sequenced directly using the thermosequenase cycle sequencing kit (Amersham) and customized oligonucleotides.

Gene deletion:
To delete css1⁺ from the S. pombe genome, a HindIII fragment containing the majority of the open reading frame was replaced with the ura4⁺ gene (see Fig 4). A linear fragment from this construct including css1⁺ 5' and 3' flanking sequences was then transformed into a diploid strain with the relevant genotype of css1⁺/css1-2. Stable Ura+ transformants were selected at 25° and then screened for temperature-sensitive growth. Temperature-sensitive diploids that were presumably css1::ura4⁺/css1-2 were screened by PCR and Southern blotting for the appropriate gene replacement.

Epitope tagging of css1⁺:
A genomic version of css1 encoding 13 copies of the myc epitope fused to the C terminus of Css1p was generated by the method described in BAHLER et al. 1998 . Oligonucleotide primers were used to amplify the myc13-kan cassette (BAHLER et al. 1998 ), and the PCR fragment was transformed into diploid cells (KGY1218). The cells were allowed to recover for 12 hr on YE plates and then replica plated onto YE plates containing 100 µg/ml G418 (Geneticin; GIBCO BRL, Grand Island, NY). G418-resistant colonies were screened for homologous recombinants by PCR; the presence of the epitope was confirmed by immunoblotting with the 9E10 antibody. Sporulation and tetrad analysis showed 2:2 segregation of the G418 resistance, confirming that the haploid strain, css1::css1-myc, was viable.

Fluorescence microscopy:
All fluorescence microscopy was performed using a Zeiss axioscope and the appropriate set of filters. Calcofluor was used to detect cell wall material in live or formaldehyde-fixed cells, DNA was visualized using 4',6-diamidino-2-phenylindole (DAPI), and actin was visualized using rhodamine-conjugated phalloidin as described (BALASUBRAMANIAN et al. 1997 ). The cell wall was also visualized by staining formaldehyde-fixed cells with 0.5 mg/ml methyl blue (M-6900; Sigma, St. Louis) in PBS (KIPPERT and LLOYD 1995 ). Methyl blue contains aniline blue that stains specifically ß-(1,3)-linked glucan (SMITH and MCCULLY 1978 ). For immunostaining, cells were fixed with a mixture of formaldehyde and glutaraldehyde (BALASUBRAMANIAN et al. 1997 ) and stained with 9E10 monoclonal antibody followed by Alexa-conjugated goat anti-mouse IgG. GFP-Css1p fusion proteins were visualized in live cells.

Analysis of cell wall polysaccharides:
Two slightly different procedures were used to analyze the cell wall polysaccharide composition. Both are based on the fact that the metabolic conversion of uniformly labeled [¹⁴C]glucose to labeled mannose and galactose occurs rapidly, leading to a homogeneous labeling of every cell wall component. In the first procedure, labeling and fractionation of cell wall polysaccharides was done as described (ARELLANO et al. 1997 ) and is presented in Fig 3A. Briefly, exponentially growing cultures of S. pombe, incubated at 25°, were supplemented with [U-¹⁴C]glucose (1 µCi/ml) and incubated for an additional 4 hr at different temperatures (25°, 28°, 32°, and 37°). Cells were harvested and total glucose incorporation was monitored by measuring the radioactivity of the TCA insoluble material. Mechanical breakage of cells was done using prechilled glass beads added to the cells, and lysis was achieved in a Fast-Prep system FP120 (BIO 101, Vista, CA; Savant, Farmingdale, NY), using two 15-sec intervals at 5.5 speed. Cell walls were pelleted at 1000 x g for 5 min and washed with 5% NaCl three times and with 1 mM EDTA three times. Aliquots (100 µl) of the total wall were incubated with 100 units of Zymolyase 100T (Seikagaku America, Ijamsville, MD) or Quantazyme (Quantum Biotechnologies, Blaine, WA) for 36 hr at 30°. Aliquots without enzyme were included as control. The samples were centrifuged and the supernatant and washed pellet were counted separately. The supernatant from the Zymolyase 100T reaction was considered to be ß-glucan plus galactomannan and the pellet to be -glucan. The supernatant from the Quantazyme reactions was considered to be ß-glucan and the pellet to be -glucan plus galactomannan.

In the second procedure (Fig 3B and Table 2), wild-type and css1-2 cells were grown at 23° in 75 ml of YE 1% glucose, 1 µCi/ml D-[¹⁴C(U)]glucose (300Ci/mmol; American Radiochemical Corporation, St. Louis) until an OD₆₀₀ of 0.3. Cultures were either harvested or shifted to 36° for 12 hr before harvesting. Cells were washed in PBS-10 mM sodium azide, resuspended in PBS containing 1 mM phenylmethylsulfonyl fluoride (PMSF), and subjected to mechanical breakage with glass beads until >95% of the cells had lysed (microscopic observation). The glass beads were allowed to settle and the recovered homogenate, together with three washes, was centrifuged at 1600 x g for 10 min to pellet the cell wall. The supernatant representing the total intracellular material was then separated into a 100,000-g pellet (intracellular organelles) and cytosol. The total radioactivity incorporated into every fraction was determined after total acid hydrolysis by liquid scintillation counting. In wild-type cells at both temperatures, as well as the css1-2 mutant at 23°, 15–20% of the total radioactivity incorporated was found in the cell wall fraction. In the css1-2 mutant grown at 36°, 50–55% of the total radioactivity incorporated was found in the cell wall fraction. The wall fractions were boiled for 5 min, washed 10 times with PBS/1 mM PMSF, and frozen at -70°.

Cell walls were fractionated according to MANNERS and MEYER 1977 with minor modifications. Briefly, the walls were extracted three times with 3% NaOH for 60 min at 75° to obtain, after centrifugation, an alkali-soluble and an alkali-insoluble fraction. The alkali-insoluble fraction contained about one-half of the ß-1,3 glucan, and the rest was in the alkali-soluble fraction. To quantify the ß-1,3 glucan, both fractions were independently treated with ß-1,3 glucanase (Zymolyase 100T; Seikagaku America). This enzyme is not recombinant but we determined that it was completely devoid of ß-1,6 glucanase activity by its inability to hydrolyze purified ß-1,6 glucan (pustulan). The digestion products were separated by P4 (Bio-Rad, Richmond, CA) chromatography and characterized by thin-layer chromatography in Silica Gel G (solvent system: butanol/ethanol/H₂O; 5:3:2. The amount of laminarioligosaccharides (ß-1,3-linked glucose oligosaccharides) detected served to quantify the ß-1,3 glucan. The -1,3 glucan was obtained in the void volume of the P4 chromatography of ß-1,3 glucanase-treated alkalki-soluble fraction. The galactomannan was obtained from the alkali-soluble fraction by selective precipitation (Fehling's reagent). Monosaccharide composition indicated the presence of mannose, galactose, and <5% glucose.

Cell fractionation and immunoblotting:
Protein lysates were prepared by glass bead disruption of S. pombe cells in one of three buffers: TEG (KATAYAMA et al. 1999 ), NP-40, or SDS-lysis (GOULD et al. 1991 ). In the latter buffer, protein lysates were heated immediately following lysis at 96° and diluted with NP-40 buffer. Following removal by filtration of glass beads, soluble and insoluble fractions of each lysate were separated by microcentrifugation at 14,000 rpm for 20 min at 4°. To each supernatant fraction was added ¹/₅ volume of 5x SDS gel sample buffer. Each pellet was resuspended in a volume of 1x SDS gel sample buffer equivalent to that of the total supernatant fractions. Both fractions were then heated to 96° for 5 min. Equal volumes of soluble and insoluble fractions were then resolved by 6–20% SDS-PAGE, transferred to polyvinylidene fluoride membrane, and Css1p-myc was visualized by enhanced chemiluminescence (ECL) subsequent to incubation with the 9E10 antibody (1 µg/ml) and peroxidase-conjugated goat anti-mouse IgG (1:50,000; Jackson Laboratories, West Grove, PA).

Electron microscopy:
For electron microscopy, 10⁶ css1-2 cells were fixed after a 6-hr shift to 36° in 1.5% potassium permanganate. They were then treated with 2% glutaraldehyde and osmium tetroxide, block stained with uranyl acetate, and dehydrated in ethanol series. Cells were embedded in spurr in BEEM capsules, which were polymerized under vacuum at 65°. Sections were cut to 75 µm with a Leica Ultracut, picked onto grids, stained with uranyl acetate and lead citrate, and imaged on a Philips 300 electron microscope at 60 kV.

Phospholipase C assays:
The phospholipase assays were performed as described (WELLS et al. 1998 ) with minor modifications. For each reaction 4 x 10⁴ cpm of [³H]dihydrosphingosine-labeled inositolphosphoceramide (IPC-3), which has a ceramide composed of phytosphingosine and -OH-C26 fatty acid (LESTER and DICKSON 1993 ), and 10 nmol of nonradioactive IPC-3 (both from S. cerevisiae) were dried in a conical glass tube (Kimble no. 73785, 15 ml, screw thread, Teflon-lined cap) under a stream of N₂. The dried lipids were suspended in 270 µl of buffer [final concentrations: 250 mM 2-(N-Morpholino)ethanesulfonic acid monohydrate (pH 6.0), 0.6% n-octyl-ß-glucoside, 5 mM MgCl₂] and treated for 10 min in a sonic water bath. Membrane protein (30 µl, 75–300 µg) was added and the reaction was incubated with intermittent shaking for up to 2 hr at 30°. The reaction was stopped by adding 0.9 ml of 2.27% sodium dodecylsulfate and extracted twice with an equal volume of methyl-tert-butylether. The upper layer was transferred to a glass tube and extracted with 1 ml of water. The upper organic layer containing the [³H]ceramide was transferred to a glass scintillation vial, taken to dryness under a stream of N₂, resuspended in 4.5 ml of UltimaGold scintillation fluid (Packard, Meriden, CT), sonicated in a water bath for 10 min, and counted in a liquid scintillation counter.

Hydrolysis of radiolabeled S. pombe mannose-inositol-phosphoceramide (MIPC) was determined by drying 1.5 x 10⁴ cpm of [³H]inositol-MIPC or 4000 cpm of [³H]dihydrosphingosine-MIPC in a glass screw-capped tube under a stream of N₂, followed by suspension in reaction buffer (final concentrations: 25 mM potassium phosphate, pH 7.0, 2.5 mM dithiothreitol, 5 mM MgCl₂, 0.6% n-octyl-ß-glucoside). The sample was sonicated as described above and warmed to 30°, and the reaction was started by adding 150 µg of membranes (total volume of 150 µl). Reactions were incubated for up to 2 hr and stopped by addition of 15 µl of 0.5 M disodium ethylenediamine acetate, pH 7.1. Samples containing [³H]inositol-MIPC were heated in a boiling water bath for 2 min followed by 2-min treatment in a sonic water bath. Reactions containing [³H]dihydrosphingosine-MIPC were vortexed with organic solvent (5 volumes of sample:16 volumes chloroform:16 volumes methanol) and centrifuged to pellet denatured proteins. In both cases 5–50 µl were chromatographed on a 15 x 15-cm piece of silica gel-impregnated Whatman SG-81 paper treated with ethylenediamine tetraacetate (STEINER and LESTER 1972 ). The chromatogram was developed in chloroform/methanol/4.2 M ammonia (9:7:2, v/v/v), dried, and each lane was cut into 2-cm wide by 1-cm tall pieces. The pieces were placed in 4.5 ml of UltimaGold scintillation fluid and sonicated for 3 min before counting in a liquid scintillation counter. The background control sample lacked membranes.

Membranes were prepared from 50–100 A₆₀₀ units of S. cerevisiae cells grown to an A₆₀₀ of 3–6 in YPD medium at 30°. The cells were centrifuged at room temperature, washed twice with distilled water, and resuspended in 1 ml of ice-cold extraction buffer (50 mM potassium phosphate, pH 7.0, 5 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride). One-half volume of glass beads (0.5-mm diameter) was added, and the sample was chilled on ice and then vortexed eight times for 30 sec with a 30-sec chilling period between vortexing. The sample was centrifuged for 10 min at 4000 x g in a Sorvall SS-34 rotor at 4°. The resulting supernatant fluid was centrifuged for 1 hr at 100,000 x g in a Beckman TLA100 rotor at 4°. The resulting membrane pellet was suspended in extraction buffer by 20 up-and-down strokes in a glass tissue homogenizer. The centrifugation and suspension steps were repeated and the final membrane pellet was suspended in extraction buffer containing 30% glycerol. Aliquots of 50 µl were frozen and stored at -80°. For enzyme assay, the frozen sample was thawed on ice and used immediately. Membranes were prepared in the same way from S. pombe cells grown at 25° in yeast extract medium to an A₆₀₀ of 0.5 (before the temperature shift) and switched to 36° for 2 hr of incubation. Protein concentration was determined by using the Bio-Rad DC protein assay kit with bovine serum albumin as the standard.

Radiolabeled sphingolipids:
S. cerevisiae sphingolipids were labeled with [³H]inositol and purified using a published procedure (WELLS et al. 1998 ). S. pombe sphingolipids were labeled with [³H]inositol or [³H]dihydrosphingosine in a similar manner except that the cells were grown in yeast extract medium. The [³H]inositol-labeled S. pombe MIPC was purified by acid precipitation. Following extraction of lipids from cells using solvent B (WELLS et al. 1998 ), 37.5 µl of Glacial acetic acid was added per milliliter of sample and stored overnight at 4°. The sample was centrifuged at 10,000 rpm in a Sorvall SS-34 rotor for 15 min at 4°. The resulting precipitate was resuspended in 0.5 ml of solvent C (chloroform/methanol/water 16:16:5, v/v/v) by warming to 60° and treating for 5 min in a sonic water bath. The sample was dried under a stream of N₂, resuspended in 0.5 ml of monomethylamine reagent (CLARKE and DAWSON 1981 ), and incubated 30 min at 50° to deacylate glycerolipids. After evaporation to dryness and resuspension in solvent C, the radiopurity of the MIPC was verified by chromatography on a Whatman HP-K thin layer plate (200 µm, 10 x 20 cm) using chloroform/methanol/4.2 M ammonia (9:7:2, v/v/v) as solvent. Radioactivity was located by using a BioScan apparatus. For [³H]inositol-sphingolipids 90% of the counts were in MIPC, 5% were in IPC, and <5% were in two small broad bands of unknown composition near the origin. The same procedures were used to purify [³H]dihydrosphingosine-labeled MIPC, except that further purification was achieved by chromatography on a Whatman HP-K TLC plate. The band of MIPC on the TLC was located (BioScan apparatus), scraped from the plate, loaded into a glass column (6 ml no. 504394; Supelco, Bellefonte, PA), and eluted with solvent C. Radioactive fractions (0.5 ml) were pooled, dried, and resuspended in 0.5 ml of solvent C.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Phenotype of css1 mutants:
In a visual screen for temperature-sensitive lethal mutants (BALASUBRAMANIAN et al. 1998 ), we identified four mutants in which the accumulation of the nonvital dye, phloxin B, was uneven within the cells. On phloxin B-containing plates, portions of these mutant cells remained white while other parts were pink. Pairwise crosses between these mutants indicated that all contained mutations within the same, or closely linked, gene, which we termed css1⁺ (see below). The css1 mutations were recessive; diploids with the relevant genotype css1-3/css1⁺ and css1-2/css1⁺ grew well at 36° (data not shown). We chose two mutants, css1-2 and css1-3, for further phenotypic analyses.

To address why the mutant cells stained unevenly with phloxin B, they were grown in liquid culture, shifted to 36° for various periods of time, and stained with calcofluor to visualize cell wall material (Fig 1B and Fig C). By 6 hr, it was clear that the mutant cells were accumulating material that stained intensely with calcofluor and by 24 hr at nonpermissive temperature, the cells were nearly filled with this material (Fig 1C). Because the cell wall, particularly that within the S. pombe septum, is normally the only structure to stain intensely with calcofluor (Fig 1A), the mutant was designated css1⁺ for can't stop synthesizing cell wall. When cells were stained with phalloidin to visualize F-actin and DAPI to visualize DNA, it became clear that the cytoplasm of css1 mutant cells was being compressed over time at the nonpermissive temperature by increasing amounts of "cell wall" material (Fig 1, D–G). The relative percentages of cells containing material at one end of the cell, both ends of the cells, or in the middle of the cells were quantified in two alleles (Fig 1H and Fig I). The localization of this material suggested that the onset of this phenotype was independent of cell cycle position.

Physiological analysis of css1 mutants:
To determine how long it took css1 mutant cells to arrest division, cell numbers of wild-type and css1-2 mutant cells were quantified after shift to the nonpermissive temperature. Whereas wild-type cells continued to grow exponentially, css1-2 cells divided at most once (Fig 2A). During the same experiment, a portion of css1-2 and wild-type cells was released from 36° to 25° to determine whether the division arrest of css1-2 cells was reversible. Unlike wild-type cells, css1-2 cells lost viability rapidly upon incubation at 36°, indicating that their arrest is terminal (Fig 2B).

Although our initial analysis from asynchronous cultures indicated that there was no specific time in the cell cycle when cell wall material was deposited in css1 mutants, we wanted to determine whether material would be deposited preferentially at one site if the cells were synchronized prior to inactivation of Css1-2p. To examine this, css1-2 cells in the G2 phase of the cell cycle were selected by elutriation and were shifted to 36°. We observed that when the septation index began to rise, 50–60% of the cells failed to proceed beyond that point (Fig 2C) and accumulated calcofluor-stainable material in the medial region (Fig 2D). The remainder of the cells underwent septation and cell division normally and then accumulated material at the old end of the cell where growth had resumed (Fig 2D). These observations indicate that the material is likely to be deposited at the site of active growth when Css1-2p is inactivated.

css1 mutants accumulate - and ß-glucans:
In an effort to determine the identity of the accumulated material, we stained css1 mutant cells with aniline blue. This fluorochrome was reported to bind preferentially to ß-(1,3) glucans (SMITH and MCCULLY 1978 ) and to stain specifically the cell wall and septum of S. pombe (KIPPERT and LLOYD 1995 ). The material in css1 mutants stained well with aniline blue (Fig 2E), indicating that at least a portion of it was comprised of ß-glucan. Treatment of either whole cells or isolated walls with recombinant ß-1,3 glucanase (Quantazyme) completely inhibited aniline blue binding (data not shown).

To identify the material more precisely, cell wall constituents were isolated and characterized after growing the cells in the presence of [U-¹⁴C]glucose. Incorporation of radioactive glucose into the cell wall of css1-2 cells was similar to that of wild type at 25°, slightly higher at 28°, and considerably higher than in the wild type at 32° (from 32 to 45% of total glucose incorporated) or 36° (from 33 to 53%; Fig 3A). A dramatic six- to sevenfold increase in the amount of -1,3 and ß-1,3 glucan was detected in css1-2 mutants grown for 12 hr at 36° (Fig 3B and Table 2). The ratio between ß/-glucan was the same as that found in wild-type S. pombe cell walls, indicating a simultaneous increase in both glucan polymers (Fig 3 and Table 2). On the other hand the amount of galactomannan was not significantly affected, leading to a drastic alteration in the glucan to galactomannan ratio. At 23° both wildtype and css1-2 mutants exhibited 2–3 times more glucan than galactomannan. At 36° we detected almost 10 times more glucan than galactomannan in the css1-2 mutant (Table 2). These changes are solely due to the css1 mutation because the temperature shift per se did not cause major alterations in the cell wall composition of wild-type S. pombe (Fig 3 and Table 2). These results suggest that css1 mutants, when grown at the restrictive temperature, have lost the ability to properly coordinate glucan polymer synthesis with cell growth. Because the specific activity of the (1,3)ß-D-glucan synthase, measured in vitro, was not increased (data not shown), it is likely that feedback regulation to prevent cell wall formation in the absence of further division is lost in css1 mutants. In this scenario, the glucan synthases would continue to produce cell wall material that is not needed.

Gene cloning:
To explore the molecular basis for the css1^- phenotype, as well as to determine the phenotype of a css1 null allele, we cloned the css1⁺ gene by complementation of the css1-3 mutant phenotype. Sequence analysis of the smallest rescuing fragment of the cloned DNAs (MluI-MluI in Fig 4A) revealed the presence of a single ORF that encoded a predicted protein product of 424 amino acids and 48 kD (Fig 4B). A database search with the predicted amino acid sequence of Css1p revealed a hypothetical protein of S. cerevisiae encoded by ORF YER019w (now termed ISC1), which was 37% identical to Css1p (Fig 4B). These yeast proteins shared significant sequence similarity, especially within their N-terminal 160 amino acids, to mammalian neutral sphingomyelinases (TOMIUK et al. 1998 ). Like the mammalian enzymes, Css1p and Isc1p lack a signal sequence at their N terminus, and a hydropathy plot suggests the presence of two adjacent membrane-spanning domains at the Css1p C terminus (Fig 4C). The mammalian enzymes were found to be similar to a large family of Mg²⁺-dependent phosphodiesterases (TOMIUK et al. 1998 ). In fact, subsequent mutagenesis of certain conserved residues predicted to be involved in catalysis abrogates function of the mammalian enzyme in vitro (RODRIGUES-LIMA et al. 2000 ; TOMIUK et al. 2000 ). Css1p contains these glutamic acid and histidine residues, which are predicted to be involved in substrate binding and catalysis, respectively, of this family of enzymes (Fig 4B).

That the rescuing gene encoded css1⁺ and not a high copy suppressor was confirmed by integration mapping. In addition, the css1 mutant alleles were sequenced and found to contain single point mutations that resulted in single amino acid substitutions. In the css1-2 allele, glycine 18 was altered to arginine; in the css1-3 allele, alanine 291 was altered to valine; in the css1-4 allele, arginine 255 was altered to histidine; in the css1-5 allele, glycine 15 was altered to glutamic acid (Fig 4B). Thus, we concluded that the cloned DNA represented the css1⁺ gene.

To determine whether css1⁺ was an essential gene, one copy of the gene was replaced with the ura4⁺ gene in a diploid with the relevant genotype css1⁺ ura4-D18/css1-2 ura4-D18 by the one-step gene deletion method. Ura4+ temperature-sensitive diploids that had the correct gene replacement, as determined by PCR, were sporulated. In each tetrad, only two Ura- temperature-sensitive colonies formed (data not shown), indicating that css1⁺ was indeed an essential gene. In contrast, ISC1 is not essential for S. cerevisiae viability.

Intracellular localization of Css1p:
To study the localization of Css1p, the chromosomal css1⁺ locus was modified to encode a protein with 13 copies of the myc epitope at its C terminus. Since the haploid strain producing Css1p-myc grew with morphology and kinetics indistinguishable from a wild-type strain, the tagged version of Css1p was functional. Because Css1p is predicted to contain two transmembrane domains, we tested whether Css1p was a membrane-associated protein. Css1p-myc and wild-type cells were lysed in buffers containing no detergent, 1% NP-40, or 1% SDS, and soluble and insoluble fractions were separated by centrifugation. Immunoblot analysis of the fractions indicated that detergent was required to release significant amounts of Css1p into the soluble pool (Fig 4D). It also indicated that Css1p released into the supernatant was readily degraded. Indeed, in NP-40 buffer, no soluble full-length protein was detected (Fig 4D, lane 3).

To determine the intracellular localization of Css1p, asynchronously growing css1myc cells were stained with antibodies to the myc epitope (9E10; Fig 5A and Fig B). Specific staining outlined the cell and nuclear peripheries. Css1p-myc was also detected in speckles throughout the cytoplasm and was excluded from nuclei. These staining patterns are consistent with the localization of Css1p to membrane structures, especially the plasma membrane, and to the secretory pathway. To confirm this localization pattern, the css1⁺ ORF was appended in frame to the 3' end of the gene encoding GFP and expressed in wild-type cells under control of the thiamine-repressible nmt1 promoter. Like Css1p-myc, GFP-Css1p was detected primarily around the entire cell periphery and in structures resembling vesicles throughout the cytoplasm (Fig 5C). Weak staining around the nuclei could be observed in some cells.

View larger version (113K): [in this window] [in a new window] [Download PPT slide]   Figure 5. Css1p is localized to membranes. css1myc cells were grown at 32° and stained with (A) 9E10 and (B) DAPI. Wild-type cells expressing (C) GFP-Css1p, (D) GFP-Css1Cp, (E) GFP-Css1p332-424, (F) GFP-Css1p332-389, (G) GFP-Css1p332-363, and (H) GFP-Css1p1-389 were grown to midlog phase in the absence of thiamine for 18 hr. Images of live cells were captured. (I) css1-2 cells expressing GFP-Css1p332-424 were grown to midlog phase at 25° in the absence of thiamine for 14 hr and shifted to 36° for 6 hr. Images of live cells were captured.

To determine whether the predicted transmembrane domains were required for the observed localization pattern, we removed from the GFP fusion the C-terminal sequences encoding the last 92 amino acids that are predicted to contain the transmembrane domains. The resultant protein, GFP-Css1Cp, was present throughout the cytoplasm and excluded from the nucleus. Unlike the full-length protein, it was not detected at the cell periphery, at the division plane, or around the nucleus (Fig 5D). Overproduced GFP-Css1Cp also appeared to form aggregates in some cells. To determine whether the last 92 amino acids (332–424) of Css1p were sufficient for membrane localization, they were fused to GFP. GFP-Css1p332-424 was localized to the cell periphery and to speckles; its staining around the nuclear periphery was much more intense than that observed with the full-length protein (Fig 5E). To determine whether the two putative transmembrane domains of Css1p were sufficient for membrane localization, Css1p residues 332–389 were fused to GFP. Again, we observed localization of this GFP fusion protein to the nuclear and cell peripheries (Fig 5F). Finally, we fused amino acids 332–363 encoding just the first transmembrane domain to GFP and found that this fusion protein localized to the plasma membrane and apparently to membranes surrounding the nucleus (Fig 5G). We conclude from these data that the first transmembrane domain of Css1p is necessary and sufficient for its membrane localization. Consistent with a putative function at the membrane, full-length GFP-Css1p but not GFP-Css1Cp was able to rescue growth of css1-2 (data not shown). GFP-Css1p1-389, which lacks only sequences C-terminal to the transmembrane domains, was able to localize correctly to membranes (Fig 5H) but was unable to rescue growth of css1-2 (data not shown). This observation suggests that this C-terminal extension of Css1p provides an essential function unrelated to membrane targeting.

css1 mutant cells maintained a normal rod shape despite the accumulation of glucans. This is dissimilar to the situation where loss of function or overproduction of many genes involved in cell wall biosynthesis causes changes in cell and cell wall morphology (reviewed in ARELLANO et al. 1999 ). Thus, we wanted to determine whether excess glucans were being deposited in the periplasmic space or in the cytoplasm. For this purpose, we utilized the GFP-Css1p332-424 construct to visualize the position of the plasma membrane relative to the deposits. When GFP-Css1p332-424 was expressed in css1-2 cells at the nonpermissive temperature, the plasma membrane was clearly no longer adjacent to the cell wall but separated from it by the accumulated material (Fig 5I). This situation was also observed by electron microscopy of css1-2 mutant cells (Fig 6). The glucans accumulated between the cell wall and the plasma membrane and did not affect the integrity or appearance of the cell wall. These results are consistent with our observation that css1-2 cells are no more resistant or sensitive to cell wall degrading enzymes than wild-type cells (data not shown).

View larger version (111K): [in this window] [in a new window] [Download PPT slide]   Figure 6. css1 mutants accumulate glucan polymers in the periplasmic space. Electron micrograph of a representative css1-2 mutant cell after a 6-hr shift to 36°. Note the unperturbed appearance of the cell wall.

Css1p is a sphingolipase:
Css1p shows greatest sequence similarity to mammalian neutral sphingomyelinases (RODRIGUES-LIMA et al. 2000 ). Fungi do not contain sphingomyelin; rather they often contain sphingolipids that have inositol phosphate linked to ceramide to form inositolphosphosphingolipids (DICKSON and LESTER 1999 ). We had previously determined that S. pombe contains primarily MIPC and a small amount of IPC (G. B. WELLS and R. L. LESTER, unpublished results). One hypothesis that would explain why css1-2 cells stop growing at 36° is that they have an abnormal amount of sphingolipids due to loss of Css1p activity. We examined this hypothesis by measuring the accumulation of radioactive sphingolipids following a shift of cells to the restrictive temperature. At the time of shifting log phase cells from 25° to 36°, [³H]inositol was added to the culture, which was incubated for 4 hr. Lipids were then extracted and analyzed. We found that wild-type and mutant css1-2 cells contained about the same amount of radiolabeled MIPC per A₆₀₀ unit of cells (data not shown). These data indicate that Css1p is not regulating sphingolipid synthesis or accumulation.

We next determined if Css1p could hydrolyze IPC. To assay for such activity, the css1⁺ gene was overexpressed in S. cerevisiae strain RCD318 using the galactose-inducible GAL1 promoter. The S. cerevisiae homolog of css1, ISC1, is deleted in strain RCD318 and the strain lacks the phospholipase C activity present in wild-type S. cerevisiae cells that hydrolyzes inositolphosphosphingolipids (WELLS et al. 1998 ). For example, membranes prepared from strain RCD318 transformed with an empty vector (RCD318-GAL1) do not hydrolyze [³H]IPC-3 whereas membranes prepared from wild-type JK9-3d S. cerevisiae cells do (Fig 7A). We found that membranes from the Css1p overexpressing strain (RCD-318GAL1css1) hydrolyzed IPC-3 as well as wild-type S. cerevisiae membranes (Fig 7A). These data strongly suggest that Css1p is able to hydrolyze inositolphosphosphingolipids.

View larger version (30K): [in this window] [in a new window] [Download PPT slide]   Figure 7. css1 has inositolphosphosphingolipase-phospholipase C activity. (A) Hydrolysis of IPC-3 by Css1p produced in S. cerevisiae. Membranes prepared from S. cerevisiae parental strain JK9-3d, an isc1 deletion derivative transformed with a galactose-inducible css1+ gene (RCD318-GAL1css1), or the deletion mutant transformed with a vector (RCD318-GAL1) were assayed for hydrolysis of S. cerevisiae [3H]DHS-IPC-3, using the procedure described in MATERIALS AND METHODS. Cells were grown overnight at 30° to stationary phase in S-leu medium containing 2% galactose as the carbon source. Values are the average of two assays ± standard deviation. (B) S. pombe css1-2 cells lack a phospholipase activity that hydrolyzes IPC-3. Membranes were prepared from log phase (A600 of 0.5–0.6) wild-type 972h- and css1-2 mutant cells grown at the permissive temperature (25°) or from cells shifted to the restrictive temperature (36°) and grown for 2 hr. S. cerevisiae [3H]DHSIPC-3 was used as the substrate and the product ceramide values are the average for membranes prepared from two cultures ± standard deviation. (C and D) Phospholipase C activity in S. pombe wild-type cells. Membranes were prepared from log phase (A600 of 0.5–0.6) wild-type 972h- and css1-2 mutant cells grown at the permissive temperature (25°) or from cells shifted to the restrictive temperature (36°) and grown for 2 hr. The substrate was [3H]inositol-labeled MIPC (C) or [3H]dihydrosphingosine-labeled MIPC (D) isolated from strain 972h-. Data are presented as the percentage of substrate converted to radioactive product [MIP (C) or ceramide (D)]. The ratio of substrate to product was determined by chromatography on SG-81 paper as described in MATERIALS AND METHODS.

The results shown in Fig 7A predict that membranes prepared from wild-type S. pombe cells should contain an activity that hydrolyzes IPC-3. This prediction was verified using membranes prepared from cells grown at 25° and from cells shifted to 36° for 2 hr prior to membrane isolation (Fig 7B). We expected that membranes prepared from css1-2 cells grown at the restrictive temperature (2 hr at 36°) should lack this enzyme activity and this is what we found (Fig 7B). Unexpectedly, membranes prepared from css1-2 grown at the permissive temperature (25°) also lacked enzyme activity (Fig 7B). Lack of activity in cells grown at the permissive temperature probably reflects a variant protein that is inactivated during membrane preparation or during the enzyme assay, which contains a low concentration of detergent. Whatever the reason, it is clear from these data that css1-2 cells lack the IPC-3 hydrolase activity present in wild-type cells. These data also establish that css1 encodes the only sphingolipid hydrolytic activity present in S. pombe cells under our assay conditions.

We next verified that the sphingolipid hydrolytic activity in S. pombe cells behaves like a phospholipase type C enzyme. Cleavage of MIPC by a phospholipase C-like enzyme should yield ceramide and MIP as the reaction products. For these experiments we used MIPC prepared from wild-type cells radiolabeled with [³H]inositol to label the polar head group or radiolabeled with [³H]dihydrosphingosine to label ceramide. The radiolabeled product of the reactions was analyzed by paper chromatography using a solvent system in which MIP remained at the origin while ceramide moved with the solvent front (WELLS et al. 1998 ). Cleavage of [³H-inositol]MIPC gave radiolabeled MIP as the product (Fig 7C) and cleavage of [³H-dihydrosphingosine]MIPC gave radiolabeled ceramide as the product (Fig 7D). Thus, the sphingolipid hydrolase in wild-type S. pombe cells behaves like a phospholipase C.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In this study, we presented the characterization of css1 mutants that display a novel defect in cell wall metabolism. css1 mutants arrest division quickly but continue to accumulate large amounts of both - and ß-glucans in their periplasmic space. Because this material is not incorporated into the cell wall, it does not affect their rod shape but does cause compaction of the cell's cytoplasm. The growth arrest and unusual glucan deposition observed in css1 mutants appear to be irreversible lethal events. css1 mutants are defective in sphingophospholipid-phospholipase C activity and, thus, their phenotype points to a connection between sphingolipid turnover and the activity of enzymes important for coordinating glucan synthesis with cell growth.

Using two different polysaccharide labeling and analysis strategies, the same conclusion was reached regarding increased levels of both - and ß-glucans, but not galactomannans, in css1 mutant cells. This result is consistent with the accumulated material staining with both calcofluor and aniline blue, dyes known to bind glucan polymers. Because there are so many genes encoding - and ß-glucan synthases in S. pombe cells, it was not possible to prevent glucan deposition by making appropriate mutant strains. For example, both css1mok1 and css1cps1 double mutant strains still accumulated large amounts of calcofluor-staining material (data not shown). This is not surprising given that in the absence of a single glucan synthase gene, many others are presumably still functional. Typically, there is strict coordination between the synthesis of cell wall components and growth of the cell wall, but it is not known how the synthesis of the different polymers is synchronized. The ratio between - and ß-glucan is maintained in spite of the dramatic polymer accumulation, indicating that the synchrony between - and ß-glucan synthases is not altered in css1 mutant cells. Since the other polysaccharide component of the cell wall, galactomannan, did not accumulate in the css1 mutant, there appears to be a specific defect in the feedback control of - and ß-glucan synthases. Although their activity does not appear to be increased in css1 mutants, as determined by measuring total ß-glucan synthase activity in vitro (data not shown), it appears that the normal feedback control of these enzymes is not operative in css1 mutants. It may be possible to discover the biochemical nature of these control mechanisms by further analysis of css1-interacting genes.

Css1p is most similar to mammalian neutral sphingomyelinases in sequence. These enzymes play roles in apoptosis, senescence, cell cycle arrest, and differentiation. Their exact roles remain to be determined, as does the way in which their activity is regulated (KOLESNICK and KRONKE 1998 ; KRONKE 1999 ; LEVADE and JAFFREZOU 1999 ). The mammalian enzymes have an N-terminal catalytic domain and a C-terminal extension relative to bacterial sphingomyelinases that contains two putative transmembrane domains (RODRIGUES-LIMA et al. 2000 ). Css1p is organized similarly. The mutations in the four css1 mutant alleles that we isolated map to sequences encoding the putative catalytic domain. However, we showed that the region C-terminal to the transmembrane domains and the transmembrane domains are also essential for its function. Why sequences C-terminal to the transmembrane domains are essential is not known. We showed that they are not required for proper Css1p localization; they might influence the activity of the catalytic domain. Given that Css1p behaves as an integral membrane protein and is localized to vesicles and the plasma membrane, it is not surprising that the transmembrane domains are essential for its function. Most likely Css1p must be anchored in membranes to function correctly. The closest known relative of Css1p among the mammalian neutral sphingomyelinases, nSMase1, is thought to be localized predominantly to endoplasmic reticulum (ER) membranes (RODRIGUES-LIMA et al. 2000 ; TOMIUK et al. 2000 ) although sphingomyelinase activity is also abundant in brain plasma membrane preparations (TOMIUK et al. 2000 ). Thus, whether Css1p plays a role in the ER and secretory apparatus as well as at the plasma membrane remains to be determined.

An unusual feature of Css1p is its membrane-targeting domain. Although predicted to contain two membrane-spanning domains likely to form a hairpin loop through the membrane, the first membrane-spanning domain of Css1p was sufficient to direct GFP to S. pombe membranes. Interestingly, a GFP fusion to the first transmembrane domain of mammalian nSNase1 also resulted in the targeting of GFP to the ER of mammalian cells (RODRIGUES-LIMA et al. 2000 ). While unusual, such a membrane insertion strategy is not unique. Several proteins, including synaptobrevin, are known to insert themselves into ER membranes through C-terminal membrane-spanning domains and to be transported through the secretory pathway (reviewed in RAPOPORT et al. 1996 ). Our results suggest that, once directed to the ER by the transmembrane domain, other portions of Css1p facilitate its transport through the secretory pathway since GFP fusions to the C-terminal domain of Css1p were much more concentrated in juxta-nuclear regions than was full-length GFP-Css1p.

S. pombe cells make IPC and MIPC (DICKSON and LESTER 1999 ) but the presence of other sphingolipids, such as sphingomyelin, has not been reported. We showed that Css1p function is clearly necessary for the yeast cells' ability to hydrolyze IPC and MIPC and, at least for MIPC, ceramide is generated in vitro (see Fig 8 for model of Css1p function). We also find that Css1p and its S. cerevisiae homolog, Isc1p, are responsible for this type of enzymatic activity in yeast membrane preparations.

View larger version (14K): [in this window] [in a new window] [Download PPT slide]   Figure 8. Models of Css1p function. Three types of substrates that Css1p might act on along with the products of the reactions are shown. Our data support model A, where the substrates would be IPC and MIPC, but do not exclude models B or C. Cellular processes or enzymes that could be targets for the hydrolysis products are discussed in the text.

The mammalian homolog, nSMaseI, is also able to function as a major sphingolipase of membrane preparations in vitro (FENSOME et al. 2000 ; TOMIUK et al. 2000 ). Unlike the situation in yeast membranes, however, there appear to be additional isozymes since immunodepletion of nSMaseI does not remove all of the neutral sphingomyelinase activity from such membrane preparations. Interestingly, nSMaseI produced in bacteria or insect cells possesses sphingolipase activity comparable to its activity in membrane preparations, arguing that this class of enzyme does not require other subunits for its activity (FENSOME et al. 2000 ). While S. cerevisiae appear not to require sphingolipid hydrolysis since they tolerate deletion of the css1 homolog ISC1, S. pombe cells cannot survive without css1. We did not detect a significant increase in the amount of S. pombe sphingolipids in css1 mutant cells (data not shown) and it may be that there is reduced synthesis of sphingolipids in response to a drop in their hydrolysis.

Since there is no detectable buildup of sphingolipids, the phenotype caused by loss of css1 function seems more likely due to a decrease in a product of sphingolipid hydrolysis, i.e., ceramide and MIP or IP (Fig 8, model A). The role of ceramide in mammalian cells is of great interest because its level increases in response to extracellular ligands and stresses, and it is thought to act as a second messenger (reviewed in DICKSON 1998 ). In S. cerevisiae, there is limited evidence that sphingolipids or their metabolites serve as second messengers except perhaps during stress responses (reviewed in DICKSON and LESTER 1999 ). The data presented here suggest that Css1p hydrolyzes sphingolipids to generate ceramide, which acts as a signal to regulate cell wall formation (Fig 8, model A).

Besides signaling, other functions for ceramides have been identified in S. cerevisiae. There is compelling evidence that ceramide enhances the transport of glycosylphosphatidylinositol (GPI)-anchored proteins through the secretory apparatus (HORVATH et al. 1994 ; SKRZYPEK et al. 1997 ) and can be the lipid moiety of lipid-linked proteins that are major constituents of the cell wall (CONZELMANN et al. 1992 ; reviewed in DICKSON 1998 ). It is possible, therefore, that loss of Css1p function causes a reduction in an unknown ceramide-anchored cell wall protein(s) whose function is to coordinate glucan synthesis and incorporation into the wall with cell growth (Fig 8, model B). Finally, our data do not exclude the possibility that Css1p acts on other lipid phosphodiesters lacking ceramide (Fig 8, model C). Because S. pombe Css1p has an essential function, further analysis of css1-interacting genes will provide a unique opportunity to probe the role of ceramides and other sphingolipid metabolites in yeast cell function.

	ACKNOWLEDGMENTS

We thank Dan McCollum and Mohan Balasubramanian for css1 mutant isolation. This work was supported by National Institutes of Health (NIH) grant GM 59773 to C.A., NIH grant GM41302 to R.C.D, grant IFD97-1570-C02-01 from the Comision Interministerial de Ciencia y Technologia, Spain to P.P., and by the Howard Hughes Medical Institute of which K.L.G. is an associate investigator.

Manuscript received April 3, 2001; Accepted for publication May 3, 2001.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ARELLANO, M., A. DURAN, and P. PEREZ, 1997  Localisation of the Schizosaccharomyces pombe rho1p GTPase and its involvement in the organisation of the actin cytoskeleton. J. Cell Sci. 110:2547-2555[Abstract].

ARELLANO, M., P. M. COLL, and P. PEREZ, 1999  RHO GTPases in the control of cell morphology, cell polarity, and actin localization in fission yeast. Microsc. Res. Tech. 47:51-60[Medline].

BAHLER, J., J. Q. WU, M. S. LONGTINE, N. G. SHAH, and A. MCKENZIE, III et al., 1998  Heterologous modules for efficient and versatile PCR-based gene targeting in Schizosaccharomyces pombe. Yeast 14:943-951[Medline].

BALASUBRAMANIAN, M. K., D. MCCOLLUM, and K. L. GOULD, 1997  Cytokinesis in fission yeast Schizosaccharomyces pombe. Methods Enzymol. 283:494-506[Medline].

BALASUBRAMANIAN, M. K., D. MCCOLLUM, L. CHANG, K. C. WONG, and N. I. NAQVI et al., 1998  Isolation and characterization of new fission yeast cytokinesis mutants. Genetics 149:1265-1275[Abstract/Free Full Text].

BARBET, N., W. J. MURIEL, and A. M. CARR, 1992  Versatile shuttle vectors and genomic libraries for use with Schizosaccharomyces pombe. Gene 114:59-66[Medline].

BASI, G., E. SCHMID, and K. MAUNDRELL, 1993  TATA box mutations in the Schizosaccharomyces pombe nmt1 promoter affect transcription efficiency but not the transcription start point or thiamine repressibility. Gene 123:131-136[Medline].

BUSH, D. A., M. HORISBERGER, I. HORMAN, and P. WURSCH, 1974  The wall structure of Schizosaccharomyces pombe. J. Gen. Microbiol. 81:199-206[Medline].

CABIB, E., T. DRGON, J. DRGONOVA, R. A. FORD, and R. KOLLAR, 1997  The yeast cell wall, a dynamic structure engaged in growth and morphogenesis. Biochem. Soc. Trans. 25:200-204[Medline].

CHAPPELL, T. G. and G. WARREN, 1989  A galactosyltransferase from the fission yeast Schizosaccharomyces pombe. J. Cell Biol. 109:2693-2702[Abstract/Free Full Text].

CHAPPELL, T. G., M. A. HAJIBAGHERI, K. AYSCOUGH, M. PIERCE, and G. WARREN, 1994  Localization of an alpha 1,2 galactosyltransferase activity to the Golgi apparatus of Schizosaccharomyces pombe. Mol. Biol. Cell 5:519-528[Abstract].

CLARKE, N. G. and R. M. DAWSON, 1981  Alkaline O leads to N-transacylation. A new method for the quantitative deacylation of phospholipids. Biochem. J. 195:301-306[Medline].

CONZELMANN, A., A. PUOTI, R. L. LESTER, and C. DESPONDS, 1992  Two different types of lipid moieties are present in glycophosphoinositol-anchored membrane proteins of Saccharomyces cerevisiae. EMBO J. 11:457-466[Medline].

DICKSON, R. C., 1998  Sphingolipid functions in Saccharomyces cerevisiae: comparison to mammals. Annu. Rev. Biochem. 67:27-48[Medline].

DICKSON, R. C. and R. L. LESTER