Activation and localization of inositol phosphosphingolipid phospholipase C, Isc1p, to the mitochondria during growth of Saccharomyces cerevisiae.

Sphingomyelinases (SMases) generate ceramides, which are known to regulate cell cycle and growth. Only one enzyme that belongs to the extended family of SMases is present in S. cerevisiae, Isc1p; however, little is known about its regulation or physiologic function. Deletion of ISC1 in S. cerevisiae resulted in a growth defect, and the slow growth phenotype was rescued by plasmid-borne expression of Isc1, confirming its role in growth. The levels of phytoceramide exhibited an Isc1p-dependent increase of approximately 4-fold after 24 h of growth. In addition, the specific activity of Isc1p was significantly elevated (>3-fold) between the early logarithmic and the late logarithmic/start of stationary phases of growth. The activation of the enzyme was not associated with increased levels of the protein, indicating that the mechanism is independent of transcription/translation. Interestingly, this activation was lost upon delipidation of the enzyme, raising the possibility of regulation by associated lipids. Confocal microscopy revealed that the enzyme was predominantly in the ER during early growth but became associated with mitochondria in late logarithmic growth. These results were also supported by differential centrifugation and isolation of mitochondria and further confirmed in mitochondria purified using sucrose gradients at the different stages of growth. These results reveal that the activity and localization of Isc1p are regulated in a growth-dependent manner. A novel mechanism for activation of Isc1p through localization to mitochondria is proposed. The results also suggest a role for Isc1p-generated ceramides in optimal regulation of growth.

In mammalian cells, the breakdown of sphingomyelin (SM) 1 results in the formation of ceramide, a bioactive lipid that plays roles in regulating cell responses to a variety of extracellular signals through the modulation of the activity of downstream effectors, which in turn control basic cellular functions such as cell growth, cell cycle arrest, apoptosis, and senescence (1)(2)(3)(4)(5)(6).
Sphingolipids are also important regulatory molecules in yeast, where they are known to be required for viability (7), for optimal life span (8), and for the regulation of responses of yeast cells to stress (1,9). Indeed, recent studies have disclosed important roles for yeast sphingoid bases in the regulation of cell cycle arrest (10), degradation of nutrient permeases (11), and endocytosis (12). Yeast ceramides have also been shown to regulate cell cycle progression, possibly through activation of trimeric protein phosphatases composed of Tpd3, Cdc55, and Sit4 (13), analogous to the role of ceramide and ceramideactivated protein phosphatases in regulating mammalian cell cycle and apoptotic responses (14,15).
Sphingomyelinases (SMases) hydrolyze the phosphodiester linkage of SM, producing ceramide and phosphorylcholine. Such activity was found to exist in organisms ranging from bacteria to mammals, including S. cerevisiae, and SMases and SMase-like enzymes have been recently cloned from mice, humans, Saccharomyces cerevisiae, and Schizosaccharomyces pombe (16 -19).
yer019w was recently identified as the S. cerevisiae gene that encodes for an inositol phosphosphingolipid phospholipase C, named ISC1 (18). Comparison of recent reports on Isc1p (18,20) and human neutral SMase 2 (nSMase 2) (17,21) shows that these enzymes share many common features. They both (a) hydrolyze SM; (b) require Mg 2ϩ for optimal activity; (c) demonstrate a neutral pH optimum; (d) are activated by anionic phospholipids; (e) present ϳ30% identity; and (f) contain a newly discovered domain that is conserved in the entire family of SMases, the P-loop-like domain (20), which appears to be important for substrate binding and/or catalysis.
Indeed, Isc1p has emerged as the leading enzyme in the study of the mechanisms of this emerging family of sphingophospholipid phosphodiesterases. A recent study demonstrated that the second transmembrane domain and the C terminus of Isc1p are required for binding of Isc1p to anionic phospholipids, especially CL, PG, and PS, phospholipids that are potent and essential activators of the enzyme in vitro (22). It was also shown that the C terminus interacts with the remainder of the enzyme, and it was proposed that this interaction plays a critical role in enzyme function through a novel tethering mechanism of enzyme activation by lipid cofactors (22).
Since mammalian SMases play roles in cell growth and the molecular mechanisms for the regulation of these enzymes are not well understood, we became interested in determining whether Isc1p plays a role in cellular growth of S. cerevisiae, in which case the regulation of Isc1p may become an important model. Although deletion of ISC1 rendered viable cells, these cells have growth defects, and a recent study showed that Css1p, the S. pombe homologue of Isc1p, is required for viability, that it regulates the coordination of cell wall formation, and that its function can be complemented by Isc1p (19).
Here we report that Isc1p is required for optimal growth of S. cerevisiae and that the enzyme is activated during growth. Overexpression of ISC1 under the control of a heterologous promoter and analysis of microsomal fractions suggested that this increase in specific activity was not due to transcriptional/ translational activation. Interestingly and instead, data from confocal microscopy in combination with biochemical experiments using isolated and purified mitochondria demonstrated that Isc1p translocates/localizes to the mitochondria during the postdiauxic phase of growth. Therefore, these results show that Isc1p is regulated (activated and translocated) during growth and suggest a novel mechanism for the activation of Isc1p through the in vivo translocation/localization to the mitochondria during the late logarithmic and postdiauxic phases of S. cerevisiae growth.

EXPERIMENTAL PROCEDURES
Materials-Monoclonal mouse anti-FLAG M2 antibody was obtained from Sigma. Monoclonal mouse anti-porin and mouse anti-dolichol phosphate mannose synthase (Dpm1) were obtained from Molecular Probes, Inc. (Eugene, OR), and polyclonal goat anti-Kar2p was from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Goat anti-mouse and donkey anti-goat horseradish peroxidase-conjugated antibodies and donkey anti-rabbit rhodamine-conjugated antibody were acquired from Jackson Immunoresearch Laboratories, Inc. (West Grove, PA). Anti-GFP antibody was obtained from Clontech. [choline-methyl-14 C]SM was synthesized in the Lipidomics Core Facility at the Medical University of South Carolina as described (23). All lipids were purchased from Avanti Polar Lipids Inc. (Alabaster, AL). All other reagents were purchased from Sigma. Yeast extract and peptone were from Difco. Synthetic minimal medium (SD), SD/Gal, and Ura dropout supplement were purchased from BD Bioscience PharMingen/Clontech.
Yeast Strains, Media, and Culture Conditions-The yeast deletion mutant strain JK9 -3d␣/⌬ISC1 (⌬isc1 cells) (MAT␣ trp1 leu2-3 his4 ura3 ade2 rme1 ISC1::G418) (18) was used in this study, and other strains were derived from it. The strain expressing FLAG-tagged ISC1 under the control of the GAL1 promoter was previously described (18). Wild type and ⌬isc1 were grown in YPD or SD 2% glucose. Strains carrying pYes2 or pRS406 constructs were grown in SD 2% glucose and ura dropout supplement. Single colonies were inoculated in SD-ura medium and incubated at 30°C in a shaker incubator at 250 rpm. Exponentially growing cultures were induced with galactose overnight, and the induced exponentially growing cultures were washed and diluted with fresh media to a density of 0.1 A 600 units and incubated for the indicated times at 30°C. Growth curves were determined by measuring the A 600 at different time points.
Construction of Yeast Strains-Sequences of primer pairs used in this study are shown in Table I. For the cDNA coding ISC1-GFPuv, the ISC1 coding sequence without the stop codon was amplified from FLAG-ISC1 cDNA (18) by PCR (forward primer, 5Ј-CGG GGT ACC ATG TAC AAC AGA AAA GAC AGA GAT-3Ј; reverse primer, 5Ј-CCC GAA TTC TTT CTC GCT CAA GAA AGT TTG C-3Ј) and then cloned into the KpnI and EcoRI sites of pYES2-GFPuv (24) in frame with the GFPuv coding sequence. To express a FLAG-ISC1 construct under the endogenous promoter of ISC1, a 0.5-kb fragment containing the 5Јflanking region of the ISC1 gene was amplified from the yeast genomic DNA (forward primer, 5Ј-GCT AGC GCT AAG GTC GAC TGC CGT  CTA GAT AAC TCA-3Ј; reverse primer, 5Ј-GGT ACC ATA TCT GCT  TTT TTT CCC TTT TTA CGC GG-3Ј), sequenced, and cloned into the XbaI and KpnI sites of pRS416, Yeast Centromere Plasmid Vector (Stratagene). The FLAG-tagged ISC1 coding sequence was amplified from the yeast genomic DNA (forward primer, 5Ј-GGT ACC ATG GAC TAC AAG GAC GAC GAT GAT AAG-3Ј; reverse primer, 5Ј-CCC GAA TTC TTT CTC GCT CAA GAA AGT TTG C-3Ј), sequenced, and cloned into the KpnI and EcoRI sites of pRS416.
Preparation of Cellular Lysates-Yeast cells were suspended in buffer containing 25 mM Tris-Cl (pH 7.4), 5 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 4 g/ml each chymostatin, leupeptin, antipain, and pepstatin A. Spheroplasts were prepared according to previous reports (25,26). Cells were pretreated with 0.1 M Tris-SO 4 , pH 9.4, containing 10 mM dithiothreitol for 10 min at 30°C. After centrifugation, cells were washed and suspended in 1.2 M sorbitol, 20 mM potassium phosphate, and Zymolase 100T was added for 50 min. The spheroplasts thus formed were harvested by centrifugation at 1,500 ϫ g for 5 min at room temperature and broken by 15 strokes in a Dounce homogenizer.
Assay of Isc1p Activity-Since untransformed ⌬isc1 cells present no sphingomyelinase activity and because the activity of partially purified enzyme from Isc1p-overexpressing cells was unstable, as described previously (22), unless otherwise stated, cell lysates were used to determine Isc1p activity as described with modifications (18). Briefly, cell lysates were incubated in 100 l of buffer containing 100 mM Tris (pH 7.5), 5 mM MgCl 2 , 5 mM dithiothreitol, 0.1% Triton X-100, 10 nmol (6.7 mol %) of PS, 10 nmol (6.7 mol %) of unlabeled SM, and 100,000 dpm of [choline-methyl- 14 C]SM at 30°C for 30 min. After the incubation, 1.0 ml of chloroform, 0.5 ml of methanol, and 0.2 ml of water were added according to the method of Folch et al. (27), and the radioactivity in a portion (400 l) of the upper phase was mixed with Safety Solve (Research Products International) for liquid scintillation counting.
Protein Determination, SDS-PAGE, and Western Blotting-Five micrograms of total protein (for detection using anti-FLAG antibody in overexpressor extracts), 30 g of total protein (for anti-FLAG detection in single copy expression), or 10 g unless otherwise specified, were resuspended in reducing buffer, resolved on 10% SDS-PAGE gels, and transferred to nitrocellulose. Membranes were blocked, and immunoblotting was performed as previously described (28). Protein concentration was determined using Bio-Rad protein assay reagent.
Isolation of Subcellular Membranes-To obtain microsomal fractions, cell lysates were centrifuged at 100,000 ϫ g for 1 h to obtain the microsomal and cytosolic fractions. Microsomal membranes were delipidated by incubation in lysis buffer in the presence of 1% Triton X-100 and incubated for 1 h at 4°C. The suspensions were centrifuged at 100,000 ϫ g for 90 min, and aliquots from the supernatants were used for enzymatic determinations under the conditions described above for the Isc1p activity assay. For differential centrifugation experiments, cell lysates were centrifuged at 13,000 ϫ g for 20 min to obtain the P13 pellets. The supernatants were transferred and centrifuged at 100,000 ϫ g for 1 h to obtain the P100 pellets. Isolation of mitochondria was based on published procedures (26,29,30) that reported that microsomes associated with mitochondrial surface can be removed in part by a decrease of the pH value of the isolation buffer (25). Spheroplasts were prepared in buffer A (20 mM potassium phosphate, pH 7.4, 1.2 M sorbitol) and resuspended in buffer B containing 10 mM K ϩ -MES, pH 6.0, 0.6 M sorbitol, 0.5 mM EDTA. After the addition of 5 mM phenylmethylsulfonyl fluoride, the cellular homogenate was prepared, and crude mitochondrial preparations were obtained as described by Glick and Pon (29).
Isolation of Purified Mitochondria-Crude mitochondria were isolated as described above, suspended in buffer C (0.6 M sorbitol, 20 mM K ϩ -MES, pH 6.0), and layered on top of a density gradient composed of 30 -60% sucrose constructed in 1-ml steps differing in 3.3% sucrose increments in buffer C. The gradient was centrifuged at 100,000 ϫ g for 3 h at 4°C using a SW41 rotor. Pure mitochondria were harvested from the lower third of the gradient as previously described (30,31) and then suspended in buffer C.  GAA TTC TTT CTC GCT CAA GAA AGT TTG C  Isc1EndoProm  F: GCT AGC GCT AAG GTC GAC TGC CGT CTA GAT AAC TCA  0.5-kb promoter region  R: GGT ACC ATA TCT GCT TTT TTT CCC TTT TTA CGC GG  FLAG-Isc1 F: GGT ACC ATG GAC TAC AAG GAC GAC GAT GAT AAG Entire coding sequence R: CCC GAA TTC TTT CTC GCT CAA GAA AGT TTG C Measurements of Mass Levels of Phytoceramide-Cells were harvested and washed with water, and lipids were extracted following the method of Bligh and Dyer (32). The chloroform organic phase was divided into aliquots, dried down, and processed for inorganic phosphorous determination or phytoceramide measurements using the Escherichia coli diacylglycerol kinase assay. Phytoceramide was quantitated using external standards and normalized for phosphorous content (33).
Confocal Microscopy-Cells were examined under a confocal laserscanning microscope (Olympus I X 70) with a Plan Apo ϫ 60 oil objective (numerical aperture 1.4), and images were captured using PerkinElmer Ultraview software that was set at the spinning disc mode.
Cell Fixation and Fluorescence Staining-Mitochondria were stained as described (34). In brief, cells were collected by centrifugation and gently suspended in growth medium containing 500 nM dye MitoTracker TM Red CM-H 2 XROS (Molecular Probes), incubated for 30 min in the dark at 30°C, and washed three times with phosphate-buffered saline. Cells were fixed for 2 h with 3.7% formaldehyde in the medium and washed three times with phosphate-buffered saline, and spheroplasts were obtained as described above. For immunofluorescence, spheroplasts were loaded on a glass slide precoated with poly-L-lysine. Washing, blocking, and antibody incubations were performed in a humid chamber as described (35), and the cells were covered with a coverslip for microscopical visualization. The excitation wavelength for GFP was 488 nm, and 525/550-nm emission was detected. The excitation wavelength for MitoTracker Red was 568, and the emission detection was Ͼ590 nm. No optical cross-talk between the emission channels was observed. For quantification of the colocalization of GFP-Isc1 with intracellular markers, the number of cells showing colocalization was expressed as a percentage of the total number of evaluable cells. Double-labeling experiments were performed using MitoTracker Red and Alexa Fluor 488-conjugated secondary antibodies (Molecular Probes) as described above. To invert the fluorophores, MitoTracker Green FM-(Molecular Probes) and Alexa Fluor 633-conjugated antibodies (Molecular Probes) were also used.

RESULTS
Deletion of ISC1 (⌬isc1) caused cells to grow slowly. The defect in growth was readily detectable at late log and stationary phases when compared with the parental wild type strain (Fig. 1). The optical density (OD) values for ⌬isc1 strain (ko) grown for 72 h were comparable with the values for the 24-h wild type cultures, and, at most, ⌬isc1 reached ϳ70% of the wild type stationary phase growth. The ⌬isc1 slow growth phenotype was rescued by plasmid-borne expression of ISC1 ( Fig. 1), confirming that the slow growth was indeed caused by loss of ISC1. Thus, Isc1p is required for optimal cellular growth.
The above results suggested the hypothesis that Isc1p enzyme activity may be regulated differentially during different phases of yeast growth, and two approaches were undertaken in order to determine whether Isc1p is selectively activated. First, the levels of phytoceramide were measured in extracts prepared from wild type or ⌬isc1 cells, grown for 4, 12, 24, and 48 h after dilution of the culture to 0.1 A 600 units, as described under "Experimental Procedures." Lipids were extracted, and the levels of phytoceramide were analyzed by the diacylglycerol kinase assay. Both wild type and ⌬isc1 showed an initial decrease (ϳ50%) in the levels of phytoceramide during the first 10 h of growth. This reduction was followed by a ϳ4-fold increase of the total level of phytoceramide in the wild type cells after 24 h of growth (five A 600 units). In contrast, the phytoceramide content in the deletion strain did not increase over this time period (Fig. 2). Therefore, these results demonstrate that the later production of phytoceramide is dependent on the expression of Isc1p, suggesting that Isc1p may be preferentially activated in the different phases of growth.
Second and based on the above, we investigated whether Isc1p showed changes in its activity during different growth phases. To evaluate this, the specific activity of Isc1p was measured in cellular extracts prepared from wild type or ⌬isc1 cells obtained at different times during the growth phases. As expected, no activity was observed in the lysates from the knock-out strain at any time during growth, thus ruling out a temporally dependent induction of another gene product with an enzymatic function similar to that of Isc1p. In contrast, the basal specific activity of native Isc1p in vitro was markedly elevated with the phase of growth, showing a significant increase of ϳ3-fold from 4 to 24 h (Fig. 3a), which was sustained up to 96 h of culture, although at a somewhat reduced level (2-3-fold). These data suggest that Isc1p activity is regulated during the logarithmic and postdiauxic phases of growth.
Next, we investigated whether the activation of Isc1p during growth was due to transcriptional or other regulatory events. For this purpose and to monitor the expression of ISC1, a 5Ј-FLAG-tagged version of ISC1 under the control of its endogenous promoter was constructed, cloned into a single copy plasmid, and transformed into the background of the ISC1 deletion strain. An increase of specific activity of FLAG-Isc1 at 24 versus 4 h was again found (Fig. 3b), similar to that observed with the wild type strain, indicating that the FLAG-tagged version of Isc1p is also active and regulated over time similar to the endogenous Isc1. The 4-and 24-h time points were chosen for future experiments, since they evidenced increases of specific activity in extracts from all strains expressing ISC1.
Next, and in order to determine whether the increase in activity was due to the synthesis of more protein, microsomal fractions were prepared from cells expressing FLAG-Isc1 cultured for 4 or 24 h, and immunoblotting was performed. A specific ϳ55-kDa band (predicted size for FLAG-Isc1) was detected with the anti-FLAG antibody in the fractions from both time points (Fig. 3b). Notably, immunoblotting analysis, per- formed on equal loading of total protein, did not reveal an increase in the expression levels of Isc1p at 24 h compared with 4 h, and actually a decrease was reproducibly observed. Thus, it is concluded that the increase in specific activity of Isc1p is not due to an increase in the relative expression levels of Isc1p. Since these results argue against a transcriptional/translational mechanism of Isc1p regulation, other mechanisms of activation were investigated.
In order to perform additional biochemical experiments, a higher amount of Isc1 protein was required. For this purpose, FLAG-Isc1 was overexpressed under the control of a promoter inducible by galactose, in the background of ISC1 deletion strain. This approach also allowed further investigation to determine whether regulation of Isc1p activity was at the level of transcription, since these studies employed a heterologous promoter. Initially, we determined whether this strain exhibited the same features that the wild type did. Again, an increase (ϳ4 -5-fold) of specific activity at 24 versus 4 h was reproducibly obtained (Fig. 3c). Importantly, these results demonstrate that the increase in specific activity is promoter-independent and validate the use of this construct to investigate the basis of the activation of Isc1p during cell growth. Also, immunoblotting did not reveal an increase in the expression levels of Isc1p ϳ55-kDa band at 24 h compared with 4 h (Fig. 3d, upper  panel). To control for the effect of the galactose promoter, we monitored in parallel the expression levels of another FLAGtagged protein, mammalian nSMase 2, also expressed under the control of galactose-inducible promoter. Expression pattern analysis revealed a slightly stronger band of the expected size (ϳ78 kDa) at 4 compared with 24 h (Fig 3d, middle panel), indicating that indeed, expression of FLAG proteins under the GAL1 promoter do not increase during the time frame of analysis. Also, immunoblotting for an endogenous yeast protein, Kar2p, showed no increase in its expression (Fig. 3d, lower  panel). Thus, these results provide further evidence against transcriptional/translational activation of Isc1p.
Next, specific activity was measured in microsomes obtained from cells overexpressing FLAG-Isc1 at 4 or 24 h, in parallel to total lysates as a control. Microsomes showed only 1.3-fold of activation and did not parallel the activation (ϳ4 -5-fold) found in total lysates (Table II). These results suggested the involvement of a regulator (loss of activator or gain of inhibitor) in the microsomal fractions containing Isc1p as opposed to total lysates. Importantly, these results clearly negate significant regulation of Isc1p activity at the transcriptional or translational level as the cause for the enzyme activation, and they even argue against most post-translational covalent modifications, since there was no change in the chromatographic behavior of Isc1p on SDS gel electrophoresis and since the activation was not retained upon further purification of the enzyme. promoter, and activity was measured as described above, except that 0.5 g of total protein was assayed. d, 5 g of protein were subjected to immunoblotting using mouse anti-FLAG or goat anti-Kar2p antibodies. The results shown are from one experiment, representative of three experiments.
FIG. 3. Activation of Isc1p during growth. a, activation of endogenous Isc1p. Wild type (wt) and ⌬isc1 (ko) cells were grown for the indicated times, and cell lysates were prepared as described under "Experimental Procedures." Activity was determined using 30 g of total protein. Results are averages Ϯ S.D. of three different experiments. b, activation of FLAG-Isc1 (lower panel). FLAG-Isc1 was expressed under the control of the endogenous ISC1 promoter, and activity was measured as described above. Protein (30 g) was subjected to immunoblotting using anti-FLAG antibody (upper panel). The results shown are from one experiment representative of three separate experiments. c, FLAG-Isc1 was overexpressed under the control of galactose In studies on mammalian nSMase, it was noted that the enzyme was tightly associated with activating lipids (21) and that delipidation of microsomes rendered the enzyme totally dependent on anionic phospholipids (such as PS). Since Isc1p shares with nSMase a strict dependence on anionic phospholipids for activation (18), we investigated the effects of delipidation (36) on the observed activation of Isc1p. For these studies, microsomal fractions obtained from cells grown for 4 or 24 h were incubated in the presence of 1% Triton X-100 and centrifuged, and aliquots from the supernatants obtained were assayed for Isc1p in the presence of constant amounts of detergent and activating phosopholipids. The results showed clearly that following delipidation of microsomes, the activation was completely lost (Table II).
The above results suggested the presence of activating lipids associated with Isc1p in the lysates of late stage cells and not in early growth phase. This raised the possibility that the environment of Isc1p during late logarithmic growth phase may be different from early cells. Therefore, it became important to determine the intracellular localization of Isc1p in different phases of growth. To define the in vivo localization of Isc1p, a construct was engineered in which GFP was fused to the N terminus of Isc1p. This chimera was biochemically similar to the native and FLAG-Isc1 enzymes, as revealed by enzyme activity, cofactor binding, and phospholipid dependence in the mixed micelle activity assay (data not shown). Control cells overexpressing GFP presented a diffuse cytosolic pattern of localization, and this pattern did not change up to 24 h (Fig. 4a,  upper left panels). The phase and phase optic/GFP merged images are shown in the right panels. In cells expressing GFP-Isc1p, discrete and "eyelash-shaped" segments were primarily observed at the 4-h growth stage (Fig. 4a, third panel). On the other hand, a different shape with the structure of "tubular branched networks" was mainly observed for GFP-Isc1p in 24-h cultured cells (Fig. 4a, bottom left). The patterns of GFP-Isc1 were indistinguishable whether the cells were fixed or not (data not shown). These tubular branched network structures are well established in yeast microscopy as mitochondria (37)(38)(39) and frequently contain multiple branching points, thus forming the network, which is partially acquired in a confocal image. In contrast, the eyelash segment structures lack branching points and seem to correspond to the smaller and brighter circle of the yeast ER. Segment and tubular network GFP-Isc1p structures coexisted during growth; however, their frequencies varied significantly at 4 and 24 h. These two Isc1pcontaining structures were counted in 100 cells, and the segment/tubular network ratio was calculated for each time point (Fig. 4a). A marked decrease in this ratio from 4.5 to 0.4 was found between the 4-and 24-h time points, respectively. These results suggested a major change in the subcellular localization of GFP-Isc1p during the early and late phases of growth.
Although these distinct patterns for ER and mitochondria are widely accepted, it was still important to define this localization more accurately, especially given the change in localization. Therefore, the intracellular distribution of GFP-Isc1 was visualized in colocalization studies in which fluorescent dye staining or indirect immunofluorescence were followed by confocal microscopy. To test whether Isc1p localized to the ER, cells from 4-and 24-h cultures were fixed and subjected to immunofluorescence. A resident ER protein, Kar2p, was used as a marker for this compartment. No differences were observed in the pattern of Kar2p in GFP-Isc1-overexpressing and control cells at 4 or 24 h (Fig. 4, b and d), demonstrating that the ER does not change shape during growth and indicating lack of effect of Isc1p on ER morphology. At 4 h, colocalization between Kar2p staining and GFP-Isc1p was observed, and quantification of the merged fluorescence showed 82% colocalization. At 24 h, a close but distinct separation of the Kar2p and Isc1p compartments was observed, and at a much lower (23%) frequency. This partial colocalization and the reduction in the percentage of association with ER seem to indicate that Isc1p localized to another compartment close to ER such as mitochondria at 24 h, as suggested by the morphology of Isc1 localization.
Therefore, co-localization experiments for GFP-Isc1 and mitochondria were undertaken next. GFP-Isc1p was visualized in 4-and 24-h cultures stained with the mitochondrial directed dye mitotracker. No co-localization between Isc1p and mitochondria was observed at 4 h (Fig. 4c, upper panels). Notably, significant co-localization (71%) between Isc1p and mitochondria was evident at 24 h (Fig. 4c, bottom panels), demonstrating that the previously observed GFP branched networks corresponded to mitochondria. Double-labeling experiments showed that Kar2p and mitotracker labeled two distinct compartments (ER and mitochondria, respectively) (Fig. 4d), which was also confirmed by inverting the fluorophores as described under "Experimental Procedures" (data not shown). In addition, zones of proximity between these two compartments were ob-served, as previously reported (34). These results indicate a change in the localization of Isc1p, suggesting that during the logarithmic phase of growth the enzyme colocalizes with the ER and that it may move to the mitochondria at 24 h of growth, also remaining close to the ER. Thus, taken together, the morphology studies and the co-localization results demonstrate significant differences in the localization of GFP-Isc1p between the early and late phases of growth.
Biochemical fractionation studies were employed next to provide an independent confirmation for a change in localization of Isc1p during the different growth phases. Total lysates were prepared from cells overexpressing FLAG-Isc1 cultured for 4 and 24 h and then subjected to differential centrifugation to obtain a 13,000 ϫ g mitochondria-rich fraction (P13) and a 100,000 ϫ g microsome-rich fraction (P100). At both time points, Isc1p activity in each fraction was determined and plotted as a percentage of total activity present in the unfractionated homogenate from the same time point (Fig. 5a). Interestingly, the distribution of Isc1p was significantly and dramatically changed during these phases of growth. About 75% of the activity present at 4 h precipitated with the P100 fraction. In contrast, the majority of the activity (Ͼ90%) present in the 24-h extracts fractionated into the P13 pellet, consistent with a change in localization from ER to mitochondria (26,40).
Immunoblot analysis provided another confirmation for the change in the localization of Isc1p (Fig. 5b). When equal amounts of protein were loaded, the analysis of the fractions (P13 and P100) obtained from the 4-h total lysate indicated that Isc1p distributed primarily to the P100 at the logarithmic phase, in agreement with the activity and the colocalization results. Instead, at 24 h, most of the FLAG-Isc1 fractionated to the P13 fraction, also paralleling the activity and microscopical observations. Altogether, these data demonstrate a significant change in the intracellular localization of Isc1p during growth of S. cerevisiae.
Next, crude mitochondrial fractions were prepared from 4and 24-h cultures as previously described (29) and analyzed by immunoblotting for Isc1p and other mitochondrial and ERresident proteins. As with the previous result, some Isc1p was detected in the P13 fraction at 4 h of growth, but this expression was lost in the further purified mitochondrial fraction (Fig.  6). Also, both porin (mitochondrial protein) and DpmIp (ER protein) were detected in the P13 crude fraction, but only porin became enriched in the mitochondrial fraction. Thus, the partial sedimentation of Isc1p in the P13 fraction at 4 h probably resembles the general sedimentation behavior of ER proteins (such as Dpm1), which is in agreement with previous observations that show partial sedimentation of ER proteins in the P13 fraction (41). Importantly, and in stark contrast, whereas Isc1p FIG. 5. Biochemical fractionation analyses of Isc1p during growth. a, homogenates from cells expressing FLAG-Isc1 grown for 4 or 24 h were obtained from spheroplasts and subjected to differential centrifugation as described under "Experimental Procedures." Isc1p activity was determined in each fraction using 0.5 g of total protein and plotted as the percentage of total activity in the unfractionated homogenate for that time point. Results are averages Ϯ S.D. of three different experiments. Results are representative of at least three independent experiments. b, homogenates (Homo) and fractionated cellular lysates (P13 and P100) obtained from 4-or 24-h cultures as described for a were subjected to immunoblotting using 5 g of protein and anti-FLAG antibody. The results shown are from one experiment representative of three experiments.
FIG. 6. Isc1p is enriched in mitochondria. Cellular lysates were obtained from spheroplasts prepared at 4 and 24 h and subjected to fractionation to obtain the P13 fractions as described in the legend to Fig. 5 and to further isolate mitochondrial fractions (Mito) as described under "Experimental Procedures." Immunoblotting of the P13 and mitochondrial fractions was performed using anti-FLAG (upper panels), anti-DpmI (middle panels), or anti-porin (lower panels) antibodies. Left panels, 4-h fractions; right panels, 24-h fractions. The results shown are from one experiment representative of two experiments. was almost absent from mitochondria obtained at 4 h, the mitochondria isolated at 24 h were enriched in Isc1p (Fig. 6). These results provide further evidence for a change in localization of Isc1p from ER to mitochondria.
Finally, pure mitochondria were obtained by further purification of crude mitochondria using sucrose gradients as described (30,31).
Step gradients from 30 -60% sucrose were prepared to analyze crude mitochondria from 4 and 24 h. Porin was detected in the fractions from the lower third of the tube as previously reported (30,31). Evaluation of Isc1p showed that the porin-containing mitochondrial fractions from the 4-h growth sample did not show the presence of FLAG-Isc1 (Fig. 7). On the other hand, the mitochondrial fractions obtained at 24 h revealed that the Isc1p peak co-sedimented with the peak fractions of porin (Fig. 7). These results provide further evidence for mitochondrial relocalization of Isc1p at 24 h.

DISCUSSION
The results from this study demonstrate that Isc1p is required for optimal growth of S. cerevisiae and that the enzyme regulates the production of phytoceramide during the late logarithmic/early stationary phase. The data also show that Isc1p is activated during growth in a post-transcriptional and posttranslational manner. Importantly, the results reveal that Isc1p changes its intracellular localization during growth such that it localizes to the mitochondria in the postdiauxic phase of growth, raising the possibility of a novel mechanism of activation.
The results from the current study show that Isc1p is responsible for the generation of phytoceramide during the late log phase of growth, such that the deletion of the enzyme abrogated the increase in yeast ceramide levels in the late phase of growth. Importantly, the results do not distinguish whether Isc1p is directly responsible for this increase in ceramide (i.e. that the bulk of this ceramide derives from hydrolysis of yeast complex inositolphosphoceramides) or whether this is an indirect effect (e.g. the ceramide derived from direct action of Isc1p regulates other pathways of ceramide metabolism). Indeed, the regulation of ceramide levels appears to be a complex process involving the action of many enzymes that either generate ceramide such as the de novo pathway, ceramide synthases (Lag1 and Lac1), and Isc1p, all of which would contribute to an increase in ceramide, or of enzymes that use it as a substrate, such as the action of ceramidases and Ipc1, which would attenuate the levels of ceramide (reviewed in Ref. 1).
Moreover, analysis of the changes in the phytoceramide levels shows that they do not strictly correlate with the determination of the specific activity of Isc1p during growth. Whereas ceramide levels decreased over the first 10 h of growth in both the wild type and in the Isc1 deletion strain, the activity studies showed an increase in Isc1 activity starting at 4 h and later. It should be noted that the lipid measurements reflect changes that occur in vivo and therefore are contributed to by multiple pathways that modulate the levels of ceramide. On the other hand, the specific activity data were determined in vitro assays of only one enzyme (Isc1p) that regulates ceramide generation. Thus, the results illustrate the complexities of ceramide metabolism in vivo. Importantly, the results support the following conclusions: (a) that Isc1 is activated, (b) that Isc1 does not contribute to the early decrease in ceramide levels, and (c) that Isc1 does contribute to the later ceramide formation in vivo.
The results from this study indicate that the reduction of the levels of phytoceramide in the ⌬isc1 strain may play a key role in the impairment to optimal cellular growth. At present, how ceramide functions to regulate yeast growth is not understood. Previous studies have disclosed a ceramide-activated phosphatase in yeast, composed of the catalytic subunit Sit4 and the regulatory subunits CDC55 and Tpd3 (13). In mammals, ceramides directly activate multiple targets including protein phosphatases and protein kinases (1). Obviously, much work is required to decipher these pathways.
The major conclusion from this study relates to the subcellular localization of Isc1p. Four lines of evidence support the change in the intracellular localization of Isc1 during growth. In relation to the fractionation studies, it is worth noting that whereas most of the mitochondria are known to sediment with the P13 fraction, ER has a heterogeneous sedimentation profile, and ER proteins are found in P13, P30, P40, and P100 fractions, with ϳ50 -60% of the ER in the 13,000 ϫ g spin (41). The ER marker Dpm1p has been shown to be present both in the P13 and the P100 fractions (42). Importantly, most of these studies have been performed using cultures entering the stationary phase, so care needs to be taken when making conclusions about organelle properties during the early log phase of growth. The differential centrifugation data (Fig. 5) revealed that at 4 h most of Isc1p was present in the P100 fraction (mostly ER) and some in the P13 (ER or mitochondria); however, the confocal data did not show colocalization of Isc1p with mitochondria at this time point, and thus the data suggest that the enzyme may be primarily in the ER rather than mitochondria during early log growth. This was further supported by further purification of the P13 fraction as shown in Fig. 6, where Isc1p became depleted from the mitochondrial fraction at 4 h. In contrast, it became evident that during the postdiauxic shift, Isc1p localized to mitochondria based on the microscopical and the biochemical observations. Importantly, this change in subcellular localization of Isc1p may provide a mechanism for activation of the enzyme. There was no evidence of change in the levels of the protein or any major post-translational modification as judged by Western blotting; however, more subtle changes such as phosphorylation without change in mobility on SDS-PAGE cannot be ruled out. Nevertheless, such stable post-translational modifications cannot be sufficient for activation, since activation was lost upon delipidation. Therefore, it is tempting to speculate that the enzyme becomes activated once it localizes to the mitochon- Step gradients from 30 -60% sucrose were prepared to analyze pure mitochondria from 4 h (left) and 24 h (right). Porin was detected by immunoblot in the fractions from the lower third of the tube (see lane numbers). Membranes were stripped and reimmunoblotted using anti-FLAG antibodies. The results shown are from one experiment representative of two experiments. dria. Indeed, the enzyme is nearly totally dependent on anionic phospholipids, especially PG, PS, and CL for activity in vitro (18), and we have previously proposed a novel "tether-and-pull" in vitro model for activation of Isc1p whereby the enzyme requires interaction of its carboxyl terminus with anionic phospholipids (22). This results in tethering the carboxyl terminus to the membrane, which in turn pulls the catalytic site close to its substrates in membranes. Notably, mitochondria are rich in PG and CL, which then may act as "activators" of Isc1p. According to this hypothesis, Isc1p, during early log phase, is located in a compartment (in the ER or close to the ER such as mitochondrial associated membranes (MAMs)) where concentrations of lipid activators/cofactors are low. Following the diauxic shift, the enzyme localizes to an environment rich in anionic phospholipids, namely the mitochondria.
The localization of Isc1p to mitochondria raises questions as to localization of its substrates. In yeast, the major sphingolipids inositol phosphoceramide/mannosyl inositol phosphorylceramide/mannosyl diinositol phosphorylceramides function as substrates for Isc1p, and they have been suggested to be enriched at the plasma membrane; however, pools of these three lipids were also detected in mitochondria (25), raising the possibility that Isc1p may act directly on the inositol phosphoceramide/mannosyl inositol phosphorylceramide/mannosyl diinositol phosphorylceramides in mitochondria following the diauxic shift. In mammalian cells, SM is enriched in the plasma membrane. However, two pieces of evidence favor the additional presence of a ceramide-generating mitochondrial lipid pool. Mammalian nSMase 2 was enriched in the heavy membrane fraction. 2 In addition, using compartment-specific targeting of SMase, Birbes et al. (43) demonstrated that only when SMase was targeted to the mitochondria did cells undergo apoptosis. Therefore, enzyme translocation and generation of ceramide at the mitochondria might be additional common features of yeast and mammalian nSMases.
The present findings raise the question of whether Isc1p associated with mitochondria at an early step in mitochondrial biogenesis or if the enzyme locates to the mitochondria once this organelle is mature and fully active and, in this case, if it happens through a vesicle-mediated transfer or transfer via regions of membrane continuity between organelles. Transport from ER to mitochondria has been studied both from the phospholipid and from the protein point of view. MAMs have been defined as a subfraction of the endoplasmic reticulum that is associated with the mitochondria (30). In the case of transport of PS to the mitochondria, it was shown that the association between the ER and mitochondria in the MAM fraction facilitates interorganelle phospholipid transport, required for decarboxylation of PS into phosphatidylethanolamine (34); however, the molecular mechanisms are not fully understood. Recent work identified an essential product of a yeast gene (MET30) that complements the lack of PS transport from MAM to mitochondria and the growth defect of the previously characterized pstA1-1 mutant. Interestingly, MET30 plays a role in ubiquitination, since it encodes a subunit of the (suppressor of rinetochore protein 1 cullin, F-box) ubiquitin ligase complex that is involved in substrate recognition (31).
Similarly, protein transport to mitochondria is only partly understood, and novel signal sequences for mitochondria localization continue to appear (31,44,45). Interestingly, it was recently reported that mitochondrial outer membrane proteins, which have a transmembrane domain near the C terminus and a N-terminal cytosolic moiety, post-translationally select their target membrane and that C-terminal charged residues play an important role (46). C-terminal substitution with one or two cationic amino acids, as opposed to an anionic amino acid, resulted in mitochondrial outer membrane-targeted localization versus promiscuous localization, through a transport process that did not proceed through the complete ER-Golgi pathway (46). Interestingly, Isc1p has two transmembrane domains followed by a soluble C-terminal domain, which contains 5 cationic residues. This raises the possibility that Isc1p may locate in the mitochondria through a similar process. More intriguingly, the finding that this domain in Isc1p interacts with anionic phospholipids, and especially the mitochondrial lipids PG and CL, raises the possibility that these lipids may play a key role in the mechanism of activation and/or localization of Isc1 to mitochondria.
The data on Isc1p presented here represent the first evidence for the in vivo mitochondrial localization of a yeast enzyme of the sphingolipid metabolism. In this regard, the growth phasedependent localization of Isc1p to the mitochondria may be an important ceramide-generating process, which in turn may have a role in signaling in yeast that contributes to the regulation of growth. In addition, the regulated mitochondrial localization of Isc1p may also support a key role for the generation of mitochondrial ceramide in growth. Finally, because of its regulated localization, Isc1p may serve as a model to study the molecular mechanisms of regulation for the mammalian members of the family of neutral SMases as well as to understand the effects of the compartmentalized generation of ceramides. Further investigation is required to define the mechanisms that regulate the localization of Isc1p, its activation, and the mechanisms by which the generated ceramides regulate cell growth.