The Yeast Magmas Ortholog Pam16 Has an Essential Function in Fermentative Growth That Involves Sphingolipid Metabolism

Magmas is a growth factor responsive gene encoding an essential mitochondrial protein in mammalian cells. Pam16, the Magmas ortholog in Saccharomyces cerevisiae, is a component of the presequence translocase-associated motor. A temperature-sensitive allele (pam16-I61N) was used to query an array of non-essential gene-deletion strains for synthetic genetic interactions. The pam16-I61N mutation at ambient temperature caused synthetic lethal or sick phenotypes with genes involved in lipid metabolism, perixosome synthesis, histone deacetylation and mitochondrial protein import. The gene deletion array was also screened for suppressors of the pam16-I61N growth defect to identify compensatory pathways. Five suppressor genes were identified (SUR4, ISC1, IPT1, SKN1, and FEN1) and all are involved in sphingolipid metabolism. pam16-I61N cells cultured in glucose at non-permissive temperatures resulted in rapid growth inhibition and G1 cell cycle arrest, but cell viability was maintained. Altered mitochondria morphology, reduced peroxisome induction in glycerol/ethanol and oleate, and changes in the levels of several sphingolipids including C18 alpha-hydroxy-phytoceramide, were also observed in the temperature sensitive strain. Deletion of SUR4, the strongest suppressor, reversed the temperature sensitive fermentative growth defect, the morphological changes and the elevated levels of C18 alpha-hydroxy phytoceramide in pam16-I61N. Deletion of the other four suppressor genes had similar effects on C18 alpha-hydroxy-phytoceramide levels and restored proliferation to the pam16-I61N strain. In addition, pam16-I61N inhibited respiratory growth, likely by reducing cardiolipin, which is essential for mitochondrial function. Our results suggest that the pleiotropic effects caused by impaired Pam16/Magmas function are mediated in part by changes in lipid metabolism.


Introduction
Mitochondrial associated granulocyte macrophage signaling molecule (Magmas) is a nuclear encoded protein found in the mitochondrial matrix of mammalian cells. It was identified as a transcript that is rapidly induced in the multipotential myeloid cell line PGMD1 by granulocyte-macrophage colony-stimulating factor (GM-CSF) [1]. Magmas is highly conserved [2] and is essential for survival in eukaryotic cells [3,4,5]. In addition to the mitochondrial targeting domain [1] a comparison of Magmas homologs from 52 species distributed among animals, plants and fungi revealed three conserved sequence motifs [2]. Kingdom specific amino acid signatures, the potential targeting of the protein to locations other than the mitochondria, and the presence of multiple isoforms in higher eukaryotes suggest that Magmas may function in multiple contexts [2].
Magmas is variably expressed and protein levels are developmentally regulated in mammalian cells [1,6]. During murine embryogenesis, the highest levels of Magmas are observed in heart, notochord, choroid plexus, cervical ganglion, nasal mucosa and liver. Adult tissues with high Magmas expression include muscle, pancreas, intestinal mucosa, and testes [6].
Magmas expression also differs in normal and neoplastic prostate [7]. The protein is barely detectable in normal prostate glands but increased amounts are observed in some higher-grade tumors. The increased Magmas in malignant cells results from higher amounts of Magmas/cell and not from changes in the number of mitochondria/cell [7]. Investigations in breast cancer correlate increased Magmas expression to poor outcome [8,9], while studies in pituitary adenoma and Burkitt's lymphoma cells suggest that increased Magmas levels protect against apoptosis [10,11]. These data suggest that Magmas has an important role in human cancers.
Human Magmas and its yeast ortholog, Pam 16, are 42% identical and exhibit functional conservation. Expression of human Magmas in yeast fully complements the lethal phenotype of PAM16 null cells [12,13] and validates the use of yeast as a model system to study the fundamental properties of the human protein. Pam16 is a regulatory component of the presequence translocase-associated motor (PAM), which is responsible for the transport of proteins into the mitochondrial matrix [14,15,16]. Pam16 binding to Pam18 results in recruitment and stabilization of Pam18 to the TIM23 translocase complex, located in the inner mitochondrial membrane. These interactions stimulate the ATPase activity of Ssc1 (the yeast homolog of Hsp70) which produces the driving force responsible for the translocation of proteins into the mitochondrial matrix [12,13,16,17,18,19,20].
To better understand Magmas function we conducted genetic and functional studies on PAM16. A temperature sensitive PAM16 mutation (I61N) was used to identify synthetic sick and lethal (SSL) genetic interactions of PAM16. In addition to genetic interactions that support the role of Pam16 in mitochondrial protein import, novel interactions of PAM16 were identified suggesting its involvement in other functions including lipid metabolism and transcription. Genetic screening also identified gene deletions that suppressed the pam16-I61N ts proliferation defect. SUR4, encoding a fatty acid elongase involved in sphingolipid biosynthesis [21,22], was the most effective suppressor. Strains containing pam16-I61N with or without a deletion of SUR4 were additionally characterized by proliferation studies, cell cycle analysis, morphology, and sphingolipid profiles. Our results suggest that the deleterious effects of impaired PAM16/Magmas activity are mediated in part through changes in sphingolipid metabolites.

A Temperature-sensitive PAM16 Mutant Exhibits Defects in Fermentation and Respiration
Microscopic examination of spores dissected from PAM16 heterozygous null asci revealed two normal colonies and two single cells (data not shown) demonstrating that haploid null spores are unable to undergo a single cell division and confirming that PAM16 is essential for proliferation. Thus to characterize the functional roles of Magmas, conditional alleles of PAM16 were necessary. Temperature sensitive alleles were created by random mutagenesis and screened for their inability to grow at elevated temperatures. Four unique mutant strains that displayed impaired growth at 37uC on glucose and on glycerol/ethanol media (YPGE) contained only a single amino acid substitution. The strain that grew best at 30uC but that did not grow at 37uC had a change at amino acid 61 from isoleucine to asparagine (I61N). This amino acid is identical in Pam16 and Magmas and is located within a sequence of high homology (EX 3 IL) conserved across fungi, plants and animals [2]. Pam16-I61N cells incubated for 7 days on YEPD at 37uC remained fully viable and when cultured at 30uC the same numbers of colonies were obtained as on duplicate plates grown at 30uC for 3 days (data not shown).
The temperature dependent growth of the pam16-I61N strain on fermentable (glucose) or non-fermentable (glycerol/ethanol) carbon sources is shown in Figure S1. The pam16-I61N strain exhibited a strong ts phenotype on both media although the temperature required to completely inhibit growth was lower on glycerol/ethanol. Since the pam16-I61N mutation affects fermentative growth in addition to respiratory metabolism, it appears that Pam16 may affect metabolic pathways outside of the mitochondrial matrix.

Synthetic Genetic Interactions
The pam16-I61N mutation was introduced into the Y7092 strain (Table 1) to serve as a query strain for SGA analysis. A control strain (Y8835) was screened in parallel and pam16-I61N xxxD double mutant colonies were identified that failed to grow as well as pam16-I61N at 30uC and 32uC. pam16-I61N xxxD colonies were also incubated at the restrictive temperature of 34uC to identify gene-deletions that suppress the loss of PAM16 function. Sixty candidate synthetic genetic interactions were initially identified using automated image analysis and visual inspection. Forty-six of these were verified as synthetic sick or lethal (SSL) interactions by random spore analysis ( Table 2).
Fourteen of the 46 SSL partners had a petite phenotype in the original gene-deletion array but were more severely growth impaired in the double mutant (Table 2). Petite strains do not have respiration competent mitochondria. Since only 227 of the 4285 gene-deletion array strains have a ''respiratory growth absent'' phenotype [23], the 5.7-fold enrichment for this phenotype in our screen (p,0.01) supports the role of PAM16/ Magmas in specific interactions with genes involved in the regulation of oxidative metabolism [13,24]. Figure S2 shows an interaction network of all the PAM16 SSL genes depicted as nodes colored according to gene ontology (GO) biological process designations [25] and edges representing known physical or genetic interactions. When the genes were categorized according to biological function, four major groups were apparent: mitochondrial lipid metabolism, histone modification, mitochondrial protein import, and peroxisome biogenesis and glycolysis ( Figure 1). These groups served as the basis for further analysis of Pam16 function.

Mitochondrial Lipid Metabolism
The functional group containing the largest number of interacting genes established a synthetic relationship between PAM16 and 12 genes involved in mitochondrial lipid metabolism ( Figure 1, blue nodes and Table 2). Ten of these genes encode proteins that are located in the mitochondrial matrix (Table 2) and are involved in mitochondrial specific fatty acid synthesis and the lipoylation of proteins [26,27,28]. The two other genes, CRD1 and TAZ1, are essential for the biosynthesis of cardiolipin in the mitochondrial inner membrane. Cardiolipin is important for maintaining mitochondrial membrane potential [29], vacuolar homeostasis [30,31] and stabilizing respiratory chain complexes [32].

Histone Modification
The second most numerous SSL gene set includes 7 genes involved in histone modification ( Figure 1, green nodes and Table 2). Five of these genes are involved in the Rpd3L histone deacetylase complex, which regulates transcription at distinct promoter regions [33,34]. Rpd3L negatively regulates genes encoding mitochondrial proteins and proteins involved in meiosis and metabolism. The positively regulated genes are involved in heat shock, osmotic stress and fermentative growth [35,36]. All the genes in the histone modification group have hundreds of physical and genetic interactions (Saccharomyces Genome Database) besides those shown in the pam16-I61N SSL partner set ( Figure  S2, inner circle). These seven genetic interactions associated with histone modification have not been previously linked to PAM16.

Mitochondrial Protein Import
A third functional cluster consists of genes involved in mitochondrial protein import. Pam16 is a constituent of the presequence translocase-associated motor (PAM) complex. Six genes from the mitochondrial membrane protein import system resulted in significantly impaired growth with pam16-I61N ( Figure 1, red nodes and Table 2). MMM1, TOM37, and TOM70 are mitochondrial outer membrane proteins involved in protein import and additionally have a role in phospholipid biosynthesis and calcium exchange [37,38,39]. PHB2, YME1, and FMP18 are associated with the mitochondrial inner membrane. Phb2 [40] is a subunit of the prohibitin complex, a chaperone for newly synthesized proteins [41]. YME1 codes for the catalytic subunit of the mitochondrial inner membrane i-AAA protease complex, which degrades misfolded proteins in the mitochondrial inner membrane and matrix [42,43]. Of the six genes in this group only Fmp18 (Pam17) has been reported to physically interact with Pam16 [44]. Twenty-two of the 46 SSL partners of pam16-I61N were not common to any of the genetic or physical interactions previously identified for the other genes/proteins involved in mitochondrial protein import (data not shown) [23]. This supports our findings that Pam16 is involved in several biological functions.

Peroxisome Biogenesis and Glycolysis
Negative synthetic interactions were also observed with genes encoding integral membrane proteins of the peroxisome, an organelle responsible for production of fatty acids and energy intermediates during degradation of complex lipids (Figure 1, yellow nodes and Table 2). PEX29 and PEX30 are genes required  for regulating peroxisome size, number and distribution [45,46]. PFK2 encodes phosphofructokinase a major glycolytic enzyme.

Genetic Suppressors of pam16-I61N are Involved in Sphingolipid Metabolism
Five gene deletions (SUR4, ISC1, IPT1, SKN1, and FEN1) were identified as suppressors of the temperature-sensitive phenotype of pam16-I61N at 34uC. All of these genes are involved in sphingolipid metabolism. SUR4 encodes an elongase that synthesizes very long chain 20-26-carbon fatty acids from C18-CoA primers and is involved in sphingolipid biosynthesis [21]. The Fen1 elongase has a similar role, with a maximum fatty acid chain length of 24 carbons [22]. Isc1 (inositol phosphosphingolipid phospholipase C) hydrolyzes complex phosphorylceramides to produce alpha-hydroxy-phytoceramides and polar headgroups [47,48]. Deletion of ISC1 causes defects in mitochondrial function and failure to up-regulate genes involved in aerobic metabolism after the diauxic shift [49,50]. Ipt1 (inositolphosphotransferase) and Skn1 (a putative glucosyltransferase) are both involved in the synthesis of mannose-(inositol-P)2-ceramide (M(IP)2C), the most abundant complex sphingolipid in yeast [51,52]. Skn1 and Ipt1 can complement each other in M(IP) 2 C synthesis under limited nutrients whereas skn1D ipt1D strains show increased autophagy and DNA fragmentation [53]. Deletion of FEN1, IPT1 or SKN1 suppressed the 34uC ts phenotype of pam16-I61N although not as well as deletion of ISC1 or SUR4. Of the five suppressor strains, the pam16-I61N sur4D strain showed the strongest growth phenotype and was chosen for further analysis. Notably, the suppressor phenotype requires a partially functional pam16 allele since it cannot suppress the lethality of a PAM16 deletion (data not shown).

Synthetic Genetic Interactions of a pam16-I61N sur4D Double Mutant Query Strain
To understand the functional relationship between sur4D suppression of pam16-I61N and the SSL interactions with pam16-I61N we screened the deletion array using a pam16-I61N sur4D double mutant query strain ( Table 2). Nineteen of 46 SSL partners of pam16-I61N retained their SSL phenotype while the remaining 27 SSL interactions were suppressed by the deletion of SUR4. The sur4D suppression phenotypes can be divided along functional lines. No SSL interactions in the histone modification group were suppressed. Most SSL interactions with mitochondrial protein import genes were also not suppressed. However, two genes in this group (TOM37 and MMM1) associated with the outer mitochondrial membrane were no longer SSL with pam16-I61N when SUR4 was deleted. Similarly, 9 of 12 SSL interactions of pam16-I61N with the mitochondrial lipid metabolism genes were suppressed by deletion of SUR4. The genes in this group that remained SSL were CRD1, PDB1 and RPM2.
Growth Characterization of Wild Type, pam16-I61N, sur4D and the pam16-I61N sur4D Strains The growth of wild type (wt), pam16-I61N, sur4D and pam16-I61N sur4D strains was evaluated at the three temperatures used in the SGA screens. Growth curves in YEPD media show the typical rapid initial rise in cell density (characteristic of fermentative growth) until the depletion of glucose results in the switch to respiratory metabolism (diauxic shift) and slower growth prior to entry into stationary phase [54,55] (Figure 2A-C). The fermentative doubling time for the strains at each temperature is shown in Figure 2D. The wt, pam16-I61N, sur4D and the pam16-I61N sur4D strains all grew at 30uC in glucose although the growth rate of the mutant strains was slower than wt cells.
The reduced ability of pam16-I61N to grow on glucose at 32uC and inability to grow at 34uC was partially corrected to wt levels by the deletion of SUR4 ( Figure 2B-D, and Figure 3, top four rows). This data is in agreement with the SGA screen, which showed that proliferation of pam16I61N in glucose containing media was restored at 34uC by deletion of SUR4. The sur4D suppression of the pam16-I61N growth defect in media containing glucose similarly occurs in rho 0 strains, which lack functional  Figure S3). The rho 0 results suggest that Pam16 provides an essential activity during fermentative growth that is independent of its role in the presequence translocase and other mitochondrial functions.
No growth was observed in pam16-I61N after the diauxic shift at 32uC and 34uC ( Figure 2B,C; blue line). At 30uC the pam16-I61N grew slower than the wt strain and sur4D strains. In contrast to all the other strains pam16-I61N sur4D was unable to proliferate at 30uC following the diauxic shift ( Figure 2A; red line). This demonstrates that deletion of SUR4 increases the severity of the respiratory growth defect in the pam16-I61N strain.
In addition to very long chain fatty acids, the synthesis of sphingolipids requires a long chain base (LCB) and a polar head group. LCB synthesis is normally required for viability [56] and thus the potential functional relationship between LCBs and pam16-I61N was not assessed by screening the deletion array. We therefore examined whether sur4D suppression of pam16-I61N could be phenocopied by pharmacological reduction of LCB synthesis using myriocin, an inhibitor of serine palmitoyltransferase [57]. Myriocin at 400 ng/ml slowed the growth of all strains consistent with the global inhibition of LCB synthesis ( Figure 3). However, neither this dose of the inhibitor nor a lower concentration (200 ng/ml, which did not substantially affect growth of the wt strain) suppressed the pam16-I61N (ts) phenotype.

Cell Cycle Analysis of Strains Containing pam16-I61N and sur4D
The cell cycle profile of wt, pam16-I61N, sur4D and pam16-I61N sur4D strains was assessed by flow cytometry at 30uC and 34uC to determine whether the cells arrested at a specific phase ( Figure 4). During normal exponential growth at 30uC in YEPD ( Figure 4, row 1), cell cycle analysis showed that 29% of wt cells were in G1 compared to 42% of pam16-I61N cells. The sur4D and the pam16-I61N sur4D strains had fewer cells in G1 than the wt and pam16-I61N strain.
When shifted to the non-permissive temperature of 34uC ( Figure 4, row 2) wt cells continued to replicate with a similar cell cycle profile at 6 h and 24 h with slightly more cells accumulating in G1 than at 30uC. In contrast, the majority of pam16-I61N cells arrested in G1. This G1 arrest phenotype was quantitatively complete in cells shifted to 37uC (data not shown). The G1 arrest in pam16-I61N was relieved in the pam16-I61N sur4D strain ( Figure 4, row 3), and correlates with the improved fermentative growth of the double mutant strain at 34uC ( Figure 2C, D, Figure 3).

Strains Containing pam16-I61N and sur4D have Altered Morphological Features
Wild type, pam16-I61N, sur4D, and pam16-I61N sur4D cells grown at 32uC were examined by transmission electron microscopy ( Figure 5 and Figure S4). Cell walls, cell membranes and the nucleus were similar in all four strains. In contrast, the pam16-I61N mutant had fewer and smaller mitochondria than wt cells ( Figure 5A, B; horizontal arrows). In addition, the pam16-I61N strain contained large vesicles with no electron dense material in the center and thickened walls (small angled arrows). These structures were less apparent in the wt strain.
The mitochondria of the sur4D strain ( Figure 5C) appeared similar to those in the wt strain. The sur4D strain also contained an additional feature, a large number of highly electron dense vesicles (vertical arrows). pam16-I61N sur4D cells ( Figure 5D) had improved mitochondrial morphology and contained smaller vesicles compared to pam16-I61N cells. Similar to sur4D cells, the pam16-I61N sur4D strain had numerous other vesicles containing highly Strains Containing pam16-I61N and sur4D have Altered Numbers of Peroxisomes Pex29 and Pex30 are peroxisomal membrane proteins that control peroxisome size, number, and distribution. The absence of either of these genes reduced pam16-I61N viability ( Figure 1 and Table 2). The electron microscopy of the two strains deleted in SUR4 showed abundant electron dense vesicles, which could be peroxisomes ( Figure 5C, D and Figure S4C, D). To determine if these vesicles were peroxisomes, we performed live cell imaging using a Pot1-GFP fusion protein. Pot1 is specific marker for mature functional peroxisomes.
Fluorescence microscopy showed that the wt strain grown in glucose had few, small peroxisomes (0.6 peroxisomes/cell). The fluorescence intensity of Pot1-GFP was weak and consistent with the low protein expression observed on western blot ( Figure 6A, F). Growing wt cells for 4 h in glycerol/ethanol or oleic acid resulted in an increase in peroxisome number (3.7-4.1 peroxisomes/cell), increased organelle size, and increased fluorescence intensity as expected ( Figure 6A and E). These changes were not seen in pam16-I61N. The pam16-I61N strain had approximately twice as many peroxisomes as wt cells in glucose but exhibited no increase in peroxisome number or fluorescence intensity following culture in glycerol/ethanol or oleic acid ( Figure 6B, E).
The sur4D and the pam16-I61N sur4D strains had 4.1 and 3.2 fold, respectively, more peroxisomes than the wt strain grown in glucose ( Figure 6C-E). In glycerol/ethanol or oleic acid containing media, the number of fluorescent structures increased in sur4D (2.2-2.4 fold) and the pam16-I61N sur4D (1.9 fold in glycerol/ethanol, 1.3 fold in oleic acid). The intensity of Pot1-GFP in the sur4D cells was higher than those in pam16-I61N sur4D cells. Quantitation of Pot1-GFP levels by western blot confirmed that the two strains with pam16-I61N had reduced Pot1 expression ( Figure 6F).
Increased peroxisome production in Saccharomyces cerevisiae occurs in response to changes in carbon source and involves multiple signaling pathways leading to transcriptional activation of genes encoding peroxisomal proteins (reviewed in [59,60]). Pot1 expression is repressed in glucose, derepressed in glycerol/ethanol and induced in oleic acid containing media [61]. Although sur4D and pam16-I61N sur4D cells grow well in glucose at 32uC (Figure 3), they have a higher baseline peroxisome content ( Figure 6) suggesting partial derepression of peroxisomal protein gene transcription. Since SUR4 deletion results in increased peroxisome content in glycerol/ethanol or oleate these strains have overcome the block in peroxisome induction characteristic of the pam16-I61N strain.

Sphingolipid Composition is Altered in the pam16-I61N Strain
Sphingolipids consist of a long-chain base amide linked to a long chain fatty acid. They are critical for plasma membrane structure,   lipid raft formation, and play important roles in numerous signaling pathways. The observation that the five suppressors of pam16-I61N are involved in sphingolipid metabolism suggested that there could be changes in sphingolipid levels that were biologically important. To investigate this possibility we measured the levels of these molecules in each of the four strains by quantitative mass spectrometry.
Log phase wt, pam16-I61N, sur4D and pam16-I61N sur4D strains cultured in YEPD (30uC) were then incubated for 6 h at 34uC, a temperature at which pam16-I61N cells lose the ability to proliferate, accumulate in G1, and remain fully viable. pam16-I61N sur4D cells grow at 34uC but with doubling times that are 1.76 longer than wt cells ( Figure 2C, D). The levels of the predominant sphingoid bases, dihydrosphingosine and phytosphingosine, as well as their phosphorylated derivatives were the same for wt and pam16-I61N cells (Table S1). The highly abundant C22-26 alpha-hydroxy-phytoceramides and the unsaturated alpha-hydroxy-phytoceramides were also similar in these two strains. The initial steps in de novo sphingolipid biosynthesis therefore appear to be unaffected by the pam16-I61N mutation.
In contrast, the amounts of several other sphingolipids were different in pam16-I61N cells compared to wt cells. C14 and C16 phytoceramide levels were elevated in pam16-I61N, while C20-C26 phytoceramides were 3-4 fold less abundant in the pam16-I61N strain. This suggests that the reduced levels of long chain phytoceramides were offset by an increase in shorter carbon chain phytoceramides. The levels of alpha-hydroxy-phytoceramides with chain lengths from C16 to C20 were increased in the pam16-I61N mutant compared to wt (Table S1 and Figure 7). Most striking was the 14 fold higher levels of C18 alpha-hydroxy-phytoceramide (C18aHP) in the pam16-I61N strain compared to wt (p = 0.043) (Figure 7).
Next we compared whether the deletion of SUR4 affected sphingolipid levels in the pam16-I61N strain. pam16-I61N sur4D resulted in increased levels of many sphingolipids species compared to pam16-I61N including the sphingolipid bases dihydrosphingosine and phytosphingosine (Table S1). The phosphorylated forms of these bases were even more elevated than the unphosphorylated forms. pam16-I61N sur4D also exhibited increases in C20:1 and C22 alpha-hydroxy-phytoceramide and low levels of the very long chain C24 and C26 molecules as expected with a SUR4 deletion. Many of these changes in sphingolipids were also observed in the sur4D strain. Importantly, deletion of SUR4, which by itself had no effect on the level of C18 alpha hydroxy phytoceramides, suppressed the elevated levels of this metabolite in the pam16-I61N mutant to essentially the wt level (Table S1 and Figure 7). C16aHP levels showed a similar trend in these strains but were not statistically significant.
The dramatic effect of sur4D in suppressing the 14-fold increase in C18 alpha hydroxy phytoceramides in the pam16-I61N strain suggested that a similar effect might underlie the suppressor phenotypes of the other sphingolipid biosynthetic genes. To test this possibility, we constructed suppressor strains by deleting ISC1, IPT1, SKN1 and FEN1 in the pam16-I61N strain and performed sphingolipid profiling as before. C18aHP was 7.5 fold higher in the pam16-I61N strain compared to wt ( Figure S7) and except for fen1D, the weakest suppressor, the double mutant strains had C18aHP levels similar to wt cells. These data identify C18aHP as an important molecule that may underlie the effects we observe in fermentative growth in the pam16-I61N strain.
Cardiolipin Levels are Decreased in pam16-I61N and pam16-I61N sur4D Since phospholipids regulate Isc1 activity [62] and CRD1 and TAZ1 are SSL with PAM16 ( Figure 1, Figure S2, and Table 2) we measured the major phospholipid content of wt, pam16-I61N, sur4D and pam16-I61N sur4D strains by thin layer chromatography ( Figure  S6). No differences in phosphatidic acid, phosphatidylserine, phosphatidylcholine, phosphatidylethanolamine or phosphatidylinositol were detected suggesting that they did not contribute to the growth and morphology characteristics observed in the pam16-I61N strain at non-permissive temperatures. However, cardiolipin levels in pam16-I61N were 50% of the wt strain (p,0.05) and were further reduced in pam16-I61N sur4D to 14% of wt (p,0.01), n = 4. Since cardiolipin levels were further reduced in pam16-I61N by deletion of SUR4 cardiolipin is not involved in the fermentative growth defect. Decreased cardiolipin levels have been shown to impair oxidative phosphorylation [29] suggesting that the loss of respiratory growth Figure 6. Effects of pam16-I61N and sur4D on peroxisome formation in glucose, glycerol/ethanol and oleic acid containing media. Live-cell fluorescence microscopy (top frames) and the corresponding differential interference contrast (DIC) image (bottom frames) of wt (A), pam16-I61N (B), sur4D (C), and pam16-I61N sur4D (D) cells containing GFP-tagged Pot1 to identify peroxisomes. Cultures grown in synthetic media at 30uC with glucose were then grown at 32uC for an additional 3 h with glucose, glycerol/ethanol or oleic acid as the carbon source. Quantitation of peroxisomes/cell (E) for each strain grown on the indicated carbon source. Western blot (F) of Pot1-GFP (M r = 75 kD) and Pgk1 (M r = 45 kD) in wt, pam16-I61N, sur4D and pam16-I61N sur4D whole cell lysates (40 mg protein/lane). Pgk1 was used to demonstrate that Pot1 is specifically induced by media with glycerol/ethanol and oleic acid carbon sources. Numbers indicate the quantitation of band volumes for Pot1-GFP (610 25 ) and Pgk1 (610 26 ) in pixels. BD indicates that the band was below the level of detection. doi:10.1371/journal.pone.0039428.g006 Figure 7. C18 alpha-hydroxy-phytoceramide levels are elevated in the pam16-I61N strain compared to wt and are similar to wt in pam16-I61N sur4D. Overnight yeast cultures of wt, pam16-I61N, sur4D, and pam16-I61N sur4D strains grown in YEPD at 30uC were diluted into early log phase and grown for 6 h at 34uC. The mean level 6 SEM (pmole/nmole phosphate) of C18 alpha-hydroxy-phytoceramide was determined by mass spectrometry. C18 alpha-hydroxy-phytoceramide was 12-15 fold higher in the pam16-I61N strain compared to wt, *p = 0.043, n = 3. The sur4D and pam16-I61N sur4D strains had C18 alpha-hydroxy-phytoceramide levels that were no different than wt cells. doi:10.1371/journal.pone.0039428.g007

PAM16 Genetic Interactions and Function
To identify biological pathways affected by Magmas, conditional mutants were generated in the yeast ortholog PAM16. The pam16-I61N mutant strain, which was unable to proliferate on glucose at 34uC, was used to identify negative synthetic interactions and to determine the effects of impaired Pam16 function. We found that impaired Pam16 affects growth, cell morphology, mitochondrial protein import, oxidative metabolism, fermentation, lipid/sphingolipid metabolism, peroxisome biosynthesis, and histone modification.

Suppressors of pam16-I61N Participate in Sphingolipid Biosynthesis
To detect additional genes functionally connected to PAM16 we screened the gene-deletion array to find second site suppressors of pam16-I61N. Five genes (SUR4, ISC1, IPT1, SKN1 and FEN1) were identified that when deleted rescued the fermentative but not the respiratory proliferation defect of the pam16-I61N strain at 34uC. All of these suppressors are involved in sphingolipid biosynthesis ( Figure 8B) indicating an important role for sphingolipids in the pam16-I61N phenotype [47,63,64,65]. The lack of pam16-I61N suppression with myriocin treatment suggests that the reversal of the pam16-I61N fermentative growth phenotype by deletion of SUR4, ISC1, IPT1, SKN1 or FEN1 is specific.

sur4D Rescues a Subset of pam16-I61N SSL Interactions
To determine which pam16-I61N SSL interactions could be compensated by sur4D, the gene-deletion array was queried with the double mutant (Table 2). Synthetic lethal genes found in the mitochondrial lipid metabolism and Pfk2, a glycolytic enzyme, were corrected by the absence of SUR4 demonstrating interdependencies between these pathways ( Figure 1 and Table 2). In contrast, deletions of genes involved in the Rpd3L complex and the two PEX genes remained synthetic lethal with pam16-I61N sur4D. Thus, compromised Rpd3L function or compromised peroxisome biogenesis impedes growth in Pam16 deficient cells by mechanisms that are distinct from Pfk2 and the mitochondrial lipid metabolism genes described above.
The deletion of SUR4 results in increased peroxisome biosynthesis and the restoration of fermentative growth in the pam16-I61N strain. However, deletion of SUR4 is unable to suppress the slow growth phenotype of pam16-I61N pex29D or pam16-I61N pex30D. This may reflect a requirement of PEX29 and PEX30 for the efficient production of peroxisomes, which cannot be bypassed by deletion of SUR4. Increased peroxisome number in pam16-I61N sur4D cells may be necessary for maintaining energy production via beta-oxidation and/or for reducing detrimental reactive oxygen species that could result from impaired electron transfer in the oxidative phosphorylation complexes [66].

Biological Features of pam16-I61N and pam16-I61N sur4D
After establishing the importance of the pam16-I61N mutation and the sur4D suppressor we evaluated the biological effects of these two mutations in vivo. Significant morphological changes in the pam16-I61N strain included small mitochondria and enlarged clear-centered vesicles. Rudimentary mitochondria are a known consequence of defective mitochondrial protein import [67] corroborating the importance of Pam16 in this function. The vesicle accumulation seen in pam16-I61N is similar to that described in several mutants strains with defects in endoplasmic reticulum to Golgi trafficking of phospholipids and may reflect altered sphingolipid metabolism in the endoplasmic reticulum [68]. These large vesicles are similar to those that occur during the unfolded protein response [69,70,71]. Another feature of pam16-I61N cells is their failure to induce peroxisomes when grown on oleate or glycerol/ethanol containing media. This differs from wt strains which increase peroxisome biogenesis and fatty acid oxidation when grown on these carbon sources [72].
The absence of Sur4 reversed many of the biological effects seen in the pam16-I61N strain. pam16-I61N sur4D cells had larger mitochondria, and fewer large clear vesicles than pam16-I61N. The improvement in mitochondrial morphology observed with the deletion of SUR4 in pam16-I61N may result from the normalization of Pam16 function or from the reduced activity of specific inhibitory sphingolipids (for example, C18aHP). Significantly, pam16-I61N sur4D recovered the ability to increase peroxisome number and size when grown in YPGE or oleic acid. The recovery of perixosome synthesis in the double mutant correlates with the resumption of fermentative growth. Alleviation of the unfolded protein response requires lipid biosynthesis in the ER [69] and precursor molecules required for lipid biosynthesis could be supplied by peroxisomal degradation of very long chain fatty acids (VLCFA). Thus recovery from G1 arrest may be due to increased availability of lipid intermediates and of substrates for energy production.
Elevated C18 Alpha-hydroxy-phytoceramide Levels are Observed with pam16-I61N Sphingolipids are incorporated into plasma and organelle membranes, lipid rafts and cytoplasmic protein complexes. The sphingolipid content of membranes is known to affect many cellular processes including endocytosis, exocytosis and actin cytoskeleton dynamics, calcium signaling, transcription and translation, cell cycle control, differentiation, stress resistance and aging (reviewed in [47,63]). In addition to their role in membrane biology, sphingolipids and their metabolites can act as intracellular signaling molecules to coordinate nutrition and energy production through sterol metabolism [73], serine utilization, and phospholipid metabolism [74].
The pam16-I61N strain had altered levels of many sphingolipid metabolites compared to wt cells (Table S1). Some of the changes in sphingolipid content in pam16-I61N cells were corrected towards wt levels by the deletion of SUR4. The most dramatic effect occurred for C18aHP, which had 14 fold higher levels in pam16-peroxisome biogenesis and energy metabolism. Defects in Pam16 result in rudimentary mitochondria and impaired fermentation and oxidative phosphorylation. The increase in C18aHP likely results from the Isc1 mediated breakdown of complex sphingolipids. The inability of pam16-I61N to induce peroxisomes results in decreased beta-oxidation of VLCFA. (B) pam16-I61N sur4D strain. Negative genetic interactions leading to a SSL phenotype are indicated in blue. Deletion of any one of five genes (red bold font) resulted in a block in the sphingolipid biosynthesis pathway (red bar) suppressing the fermentative growth defect of pam16-I61N. The pam16-I61N sur4D stain has morphologically restored mitochondria, increased numbers of inducible peroxisomes and wt levels of C18 alpha-hydroxy-phytoceramide. Metabolic intermediates and complex sphingolipids synthesized in the endoplasmic reticulum and Golgi are shown in bold font. Green arrows indicate changes up (increased) or down (reduced). Abbreviations: DHS: dihydrosphingosine; DHSP: dihydrosphingosine-1-phosphate; PHS: phytosphingosine; PHSP: phytosphingosine-1-phosphate; VLCFA: very long chain fatty acid. doi:10.1371/journal.pone.0039428.g008 I61N than in wt strains and were restored to wt levels in pam16-I61N sur4D (Figure 7 and Figure 8A and B). Since the deletion of SUR4 also restored fermentative proliferation and most of the morphological changes resulting from pam16-I61N, C18aHP could be an important mediator of the effects of this mutation. Studies in Neurospora crassa also demonstrate the importance of C18aHP. In this organism heat stress and carbohydrate deprivation specifically induced C18aHP, which resulted in cell death [75].

Potential Mechanisms by which pam16-I61N Increases C18aHP Levels During Fermentative Growth
It is unclear whether the dramatic increase in C18aHP levels is a cause or a consequence of the fermentative growth arrest of pam16-I61N. However all five suppressor strains reduced the pam16-I61N induced elevation of C18aHP ( Figure S7) suggesting that increased levels of C18aHP are responsible for the impaired fermentative growth. The elevated levels of C18aHP observed in the pam16-I61N strain could result from either increased synthesis or reduced degradation.
Since Pam16 is required for mitochondrial presequence translocase activity and C18aHP is enriched in mitochondrial membranes [50], it is reasonable to consider that pam16-I61N would result in increased accumulation of C18aHP by reducing its mitochondrial utilization or degradation. This would require that C18aHP enter the mitochondria through a translocase dependent carrier protein. Alternatively pam16-I61N could affect C18aHP degradation pathways instead of its mitochondrial uptake. However neither of these possibilities is consistent with the 5 deletion gene suppressor results.
Instead our data is most consistent with the hypothesis that elevated C18aHP in pam16-I61N is predominantly a result of its increased synthesis from complex sphingolipids and not from the de novo synthesis pathway (Figure 8). Deletion of SCS7, a gene required for de novo synthesis of alpha-hydroxy-phytoceramides [76], did not suppress the pam16-I61N slow growth phenotype (data not shown). This implies that the accumulation of C18aHP results from increased degradation of complex sphingolipids and the lower levels in suppressed strains results from reduced degradation of complex sphingolipids.
Deletion of ISC1 results in decreased cellular C18aHP levels [50]. The most abundant Isc1 substrates in wt strains are those sphingolipids with C26 very long chain fatty acids (VLCFA) that are synthesized by Sur4, and the most abundant Isc1 product is C26 alpha-hydroxy-phytoceramide [50]. The pam16-I61N sur4D strain may have less C18aHP because C26 VLCFAs are not made in sur4D strains and short chain ceramides are used by the cell in place of ceramides that contain VLCFAs.
Our data suggests that pam16-I61N induced elevation of C18aHP is independent of the known mitochondrial activities of Pam16. The growth arrest phenotype of pam16-I61N and its reversal in pam16-I61N sur4D still occurs in the respective rho 0 strains lacking functional mitochondria. Pam16 may regulate production of C18aHP by Isc1. The quantitative distribution of Isc1 in microenvironments within the cytoplasm is uncertain although much of its activity occurs in the endoplasmic reticulum during early growth and in the mitochondria after the diauxic shift [49,50,77]. Pam16 may regulate Isc1 production of C18aHPs that are typically enriched in the mitochondrial membrane [50]. However the Isc1 preferred substrates, C26 ceramides, are located primarily in the plasma membrane [47]. Neutral sphingomyelinase 2 (nSMase2) the mammalian homolog of Isc1 localizes to the plasma membrane and can functionally replace the yeast gene [78]. It is possible that Isc1 may degrade ceramides at the plasma membrane during fermentative growth. Since Magmas is present in extra mitochondrial locations in the cytoplasm [1], there could be direct interaction between Magmas and SMase2 and their respective counterparts in yeast in the cytoplasm.

The Inhibitory Effects of Pam16-I61N on Respiratory Growth are Independent of its Effect on Fermentative Growth
The Pam16 mutant strain was unable to sustain respiratory growth at the non-permissive temperature ( Figure S1 and Figure 2). Unlike fermentative growth, deletion of genes involved in sphingolipid synthesis (SUR4, ISC1, IPT1, SKN1 and FEN1) did not correct the respiratory growth defect (Figure 2 and data not shown). pam16-I61N resulted in reduced cardiolipin levels and mitochondrial dysfunction consistent with the severe compromises in mitochondrial morphology (Figure 5 and S5). Decreased levels of cardiolipin have been shown to specifically impair respiratory super complex formation and cytochrome oxidase activity [32,79,80].
Studies on Blp the Drosophila homolog of Pam16, are consistent with the findings observed in Saccharomyces cerevisiae during respiratory growth. Drosophila S2 cells depleted of Blp by RNAi caused mitochondrial membrane depolarization, decreased ATP levels, increased ROS, cell cycle arrest and uncoupling of oxidative phosphorylation [66]. Blp depletion preferentially inhibited cytochrome oxidase activity, which preceded the inhibition of protein import into the mitochondrial matrix. Similarly, a Blp inhibitor (SMMI: small molecule Magmas inhibitor) [81], reduced mitochondrial membrane potential and cytochrome oxidase activity prior to detectable effects on the presequence translocase.

Magmas in Human Malignancy
Our findings in yeast suggest that Magmas may be involved in human malignancy. Unlike their normal counterparts malignant cells often use aerobic glycolysis for energy requirements. Aerobic glycolysis results in alterations in the levels of many cellular metabolites some of which are involved in the pathogenesis of cancer (reviewed in [82,83]). Since Pam16 has significant effects on both respiration and fermentation it is likely that Magmas contributes to the metabolic alterations leading to the initiation and maintenance of the malignant phenotype.

Yeast Strains and Media
Strains used in this study are listed in Table 1. rho 0 derivatives were isolated from parental strains by two rounds of growth to saturation in minimal medium containing 2% glucose plus 25 mg/ ml ethidium bromide [84], followed by isolation of single colonies on YEPD and testing for no growth on YPGE. Suppressor strains were generated in a pam16-I61N mutant background by one-step replacement of each suppressor gene by a hygromycin resistance cassette and selection on YEPD plates with 200 mg/ml hygromycin B (Invitrogen, Carlsbad, CA). Yeast were grown in Yeast Extract and Peptone plus adenine medium with 2% glucose (YEPD) or 2% glycerol and ethanol (YPGE) on agar plates or in liquid culture. Synthetic genetic array experiments used yeast nitrogen base (Becton, Dickinson and Co. Sparks, MD) with added specific amino acid drop out selection and nourseothricin (cloNAT; WERNER BioAgents, Germany), canavanine, thialysine, and G418 (Sigma, St. Louis, MO) for serial selection procedures. In the experiments involving peroxisome analysis the growth medium was yeast nitrogen base plus amino acids and 2% glucose, glycerol and ethanol, or 0.25% oleic acid and 0.3% Brij35. YEPD supplemented with myriocin (Sigma) at the indicated concentrations was used in the serine palmitoyltransferase inhibition experiments.

PAM16 Mutations
PAM16 ts mutants were made by error-prone PCR using yeast strain W303 [85]. PAM16 cDNA encoding amino acids S41 through D142 was subcloned into pCR-TOPO (Invitrogen) and used as the template for mutagenesis. The error-prone PCR products were subcloned into wt PAM16 in pRS315 and transformed into E. coli XL-1B (Agilent, Santa Clara, CA). Bacterial colonies were scraped from plates and plasmid DNA was extracted. W303a haploid yeast containing pam16D::HIS3 and a wt copy of PAM16 on pRS316 was transformed with the mutant plasmid library. After eviction of the plasmid by growth at 25uC on YEPD/FOA [86], surviving colonies were plated on YEPD or YPGE at 18uC, 30uC, and 37uC to identify ts conditional mutants. The plasmid DNA was sequenced to determine the mutation responsible for the conditional phenotype.

Synthetic Genetic Array (SGA)
Query strains were derived from Y7092 (Table 1) with a nourseothricin resistant gene located adjacent to the chromosomal pam16-I61N mutant or replacing non-essential genes. The pam16-I61N sur4D double mutant was derived from the pam16-I61N query by replacing the SUR4 coding region with URA3. The array of deletion strains has been described by Tong and Boone [87]. Automated construction of double mutants was achieved by crossing the set of viable gene-deletion mutants with the chosen query strain. Deletion strains were arrayed on 7 plates at a density of 1536 colonies per plate with duplicates of each gene-deletion. Each SGA screen was performed twice using a Singer RoToR HDA robot (Singer Instruments, Somerset, UK) for all pinning procedures. Plates were photographed 30 h after the final pinning step. The growth of double mutant colonies obtained using a control strain (Y8835) or the PAM16 mutant strain (pam16-I61N) were compared by visual inspection and by computer-based image analysis (ColonyImager software) [88]. All synthetic genetic interactions were validated by random spore analysis [89].

Proliferation Assay
Log phase cultures of selected strains were counted by hemocytometer, diluted to 5610 5 cells/ml, and grown in YEPD in a microtiter plate shaken at high speed at the indicated temperature for 48 h in a Bioscreen C MBR (Growth Curves USA, Piscataway, NJ). Growth curves were derived from OD 600 nm readings of triplicate culture samples measured every 30 minutes. Standard deviations were less than 0.1% of each mean value. Doubling times were calculated from the slope of the natural log OD 600 nm vs. time at cell concentrations between OD 0.2-0.6. Two independent experiments were performed.

Flow Cytometry
Cell cycle and cell size were analyzed by fluorescence activated cell sorting on a LSRII analyzer (Becton Dickinson, Franklin Lakes, NJ) at the Albert Einstein College of Medicine Flow Cytometry Core Facility. Log phase cultures of yeast strains grown in YEPD at 30uC were switched to 34uC for the times indicated, washed in buffer and fixed in 70% ethanol/H 2 O. Fixed cells were digested with RNAse A and proteinase K (Sigma) and stained with Sytox green (Invitrogen). Cells were sonicated for 5 seconds at 30% power with a Sonic Dismembrator Model 500 (GE Healthcare, Piscataway, NJ) prior to analysis. Quantitation of percent cells in cell cycle stages was determined using FloJo analysis software (BD Biosciences, San Jose, CA). Data shown is representative of 3 independent experiments.

Electron Microscopy
Cultures of wt, pam16-I61N, sur4D and pam16-I61N sur4D grown overnight in YEPD at 30uC were split into prewarmed 32uC YEPD and grown for 4-6 h until OD 600 nm reached 0.500. 9.5 ml of each culture was fixed (2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer) and postfixed with 1% osmium tetroxide, followed by 2% uranyl acetate, dehydrated through a graded series of ethanol and embedded in LX112 resin (LADD Research Industries, Burlington, VT). Sections were cut on a Reichert Ultracut UCT, stained with uranyl acetate followed by lead citrate and viewed on a JEOL 1200EX transmission electron microscope at 80 kv at the Analytic Imaging Facility at the Albert Einstein College of Medicine (AIF). At least 100 cells of each strain were examined. Mitochondria and peroxisomal number and cell and mitochondrial cross sectional areas were measured for every cell contained in two low power fields for each strain using ImageJ 1.40 g software (National Institutes of Health, USA). Differences between strains were determined by comparing the total organelle area divided by total cell area. Two independent experiments were performed.

Live Cell Fluorescence Microscopy
Yeast strains carrying C-terminal GFP-tagged Pot1 to identify peroxisomes were grown overnight in synthetic media supplemented with 2% dextrose. After dilution to an OD 600 nm of ,0.3 (to obtain log phase cultures) in synthetic media supplemented with 2% dextrose, 2% glycerol and ethanol, or with 0.25% oleic acid and 0.3% Brig 35 (Sigma), cultures were grown for an additional 3 h at 32uC. Yeast strains with a plasmid encoding a mitochondrial leader sequence-GFP fusion protein (pYX232mtgfp) [58] were grown overnight in YEPD at 30uC, split into prewarmed 32uC YEPD and grown for 4 h. Cells were harvested prior to live-cell fluorescence microscopy by centrifugation at 3000 rpm for 5 minutes and resuspended in their respective media. Live-cell fluorescence of the strains was monitored using a fluorescence microscope (Olympus BX61) at the AIF with a 606 numerical aperture 1.46 objective (PlanApo). Fluorescence or differential interference contrast (DIC) images were captured with a cooled CCD camera (Cooke Sensicam QE, Cooke Corporation, Romulus, MI) using IPLab 4.0 software (BD Biosciences). Images were analyzed with ImageJ software 1.42q (National Institutes of Health, USA). In the Pot1-GFP experiments the number of peroxisomes per cell was the average obtained from at least 100 cells. Two independent experiments were performed.

Western Blot
Forty micrograms of total cellular protein from each yeast strain was solubilized in Laemmli sample buffer and the proteins were separated by SDS-PAGE. The proteins were transferred to a nylon membrane (Westran S, Whatman Inc., Sanford ME) in a TRIS/glycine-20% methanol solution (100 volt-hour) at 14uC. Membranes were rinsed, blocked, and then incubated with a monoclonal antibody to GFP (Roche Applied Science, Indianapolis, IN). After washing, the membrane was immersed for 1 h with peroxidase-labeled secondary antibody, washed four times with buffer containing 50 mM TRIS at pH 7.5/150 mM NaCl/0.1% Tween 20 (TBST), for 15 minutes each, and developed in luminol substrate (Immobilon, Millipore, Billerica, MA). The resulting image was captured with the ImageQuant LAS4000 biomolecular imager and the signal intensities of specific bands determined using ImageQuant TL7.0 software (both from GE Healthcare Life Sciences, Piscataway, NJ) and expressed as band volumes. Blots were stripped and reprobed with a monoclonal antibody to yeast Pgk1 (Invitrogen) Pgk1 is a protein induced by glucose [90].

Sphingolipid Analysis
Yeast strains were grown in liquid culture in YEPD overnight at 30uC, diluted to 1610 6 cells/ml, and incubated for 6 h at 34uC in a shaking water bath. Cultures were washed twice in cold water and snap frozen in liquid nitrogen. Lipids were quantitatively analyzed at the Lipidomics Analytical Unit of the Medical University of South Carolina (http://www.musc.edu/ BCMB/lipidomics/index.shtml) using multicomponent LC/MS analysis or simultaneous ESI/MS/MS analysis on a Thermo Finnigan TSQ 7000 triple quadrupole mass spectrometer operating in a multiple reaction monitoring (MRM) positive ionization mode as described [91]. The quantity of each sphingolipid was normalized to organic phosphate determination. The results shown in Table S1 and Figure 7 are the mean values from three separate experiments. Sphingolipid content is expressed as pmole sphingolipid/nmole phosphate. Values were compared using a one-way ANOVA followed by Dunnett's multiple comparison test.

Phospholipid Analysis
Relative phospholipid levels were determined according to the method of Vaden et al. [92]. Yeast strains were grown in YEPD at 32uC containing 10 mCi 32 P i ml 21 at an OD 600 nm of 0.025 and incubated to an OD of 0.800. Cells were washed in H 2 O once and digested with zymolase to form spheroplasts. Phospholipids were extracted from the spheroplasts and dissolved in 2:1 chloroform/ methanol. Samples were applied to Whatman LK5 silica gel TLC plates (GE Healthcare) and separated in chloroform/ethanol/ water/triethylamine 30/35/7/35 v/v for 2 h, dried, and run again in the same solvent system for another 2 h. The signal intensity of each band was captured on a phosphoimager screen, and was analyzed using a Fujifilm FLA-7000 Phosphoimager and ImageQuant software (GE Healthcare). Four independent experiments were performed. Figure S1 Growth characteristics of yeast strains. Five fold serial dilutions of the wt query strain, and pam16-I61N, were spotted on YEPD or YPGE plates. pam16-I61N displayed temperature sensitive growth inhibition on medium containing glucose or glycerol/ethanol at 32uC and growth arrest at 34uC and 37uC. (TIF) Figure S2 Synthetic lethal partners of Pam16-I61N. Schematic of genetic network interactions of SSL partners of Pam-I61N. 46 deletion strains were synthetic sick or lethal when paired with mutant Pam16. The synthetic lethal partners were assigned by the Osprey software program [93] into 10 of 17 broad gene ontology (GO) process categories; including 20 in Cell Organization and Biogenesis (purple), 14 in Metabolism (blue), 2 in Unknown Function (gray), 2 in Protein Transport (hot pink), and 1 each in the Stress Response (flesh), Cell Cycle (green), DNA Metabolism (beige), Protein Amino Acid Phosphorylation (brown), and Transcription (turquoise). The schematic shows the genes with their GO category as a colored circle with the lines connecting them designating a physical or genetic interaction. The genes were arranged alphabetically in two concentric circles with the inner circle containing the genes with the most interactions with other members of the set and those in the outer circle having fewer interactions. Edge lines between nodes are colored to represent the experimental system used to determine the association. These are synthetic lethality (light green), affinity capture (navy blue), dosage rescue (orange), two-hybrid (aqua) phenotypic enhancement (pea green), reconstituted complex (rust), biochemical activity (dark blue). Two genes, YNL198c and YML090W, were open reading frames in the SGD. YNL198c was deleted from our SSL set as redundant because it overlaps the opposite DNA strand coding for GCR2. YML090W was replaced by RPM2, an essential gene on the opposite DNA strand whose transcription is inhibited but not destroyed by the YML090W deletion (Table 2). (TIF) Figure S3 Strains lacking functional mitochondria have the same temperature sensitive effects on growth as their parental strains. Five fold serial dilutions of two independent rho 0 strains derived from wt, pam16-I61N, sur4D, and pam16-I61N sur4D were spotted on YEPD plates at the temperatures indicated. pam16-I61N temperature sensitive growth inhibition at 34uC and 37uC and suppression in pam16-I61N sur4D occurred in cells lacking functional mitochondria suggesting that these effects may be mediated elsewhere in the cell. n = 2. (TIF) Figure S4 Effects of pam16-I61N and sur4D on yeast morphology. Transmission electron microscopy of yeast strains wt (A), pam16-I61N mutant strain (B), sur4D (C), and pam16-I61N sur4D (D) are shown in a low power field to demonstrate that the cells depicted in Figure 5 are representative of the entire population. (5,0006 magnification; bar = 1 um). (TIF) Figure S5 Effects of pam16-I61N and sur4D on mitochondrial morphology. Live-cell fluorescence microscopy of wt (A), pam16-I61N (B), sur4D (C) and pam16-I61N sur4D (D) containing mitochondrial leader-GFP to display mitochondrial morphology at the semi-permissive temperature (32uC) (left panels) and the corresponding differential interference contrast image are shown in the right panels. The fragmented mitochondria morphology observed in pam16-I61N was restored to the wt elongated reticular mitochondrial morphology in pam16-I61N sur4D.