The multiple physiological roles of the lipid ceramide in the yeast Saccharomyces cerevisiae

(1)

Thesis

Reference

The multiple physiological roles of the lipid ceramide in the yeast Saccharomyces cerevisiae

EPSTEIN, Sharon

Abstract

Les Céramides sont des composants lipidiques cellulaires essentiels. Ils représentent la colonne vertébrale de tous les sphingolipides. Il a été observé qu'ils jouent un rôle important en tant que molécules «signal» lors de différentes transmissions de signal. Les Céramides sont composés d'une base sphingoïde attachée à un acide gras via une liaison amide. Tous les Eucaryotes supérieurs possèdent ce lipide et lors des vingt dernières années les enzymes qui catalysent sa synthèse ont été identifiées dans une grande variété d'organismes. Le but de ce travail de thèse est d'étudier les fonctions physiologiques des Céramides dans la levure en tant qu'organisme modèle. Comment la longueur de l'acide gras incorporé aux Céramides et par conséquence aux sphingolipides complexes influence des processus physiologiques?

Quelles autres voies de transduction du signal sont influencées par les niveaux de Céramide dans d'autres organismes modèles tels que le Nématode? En premier lieu nous avons cherché à comprendre les effets de l'accumulation de sphingolipides à courtes chaines lipidiques [...]

EPSTEIN, Sharon. The multiple physiological roles of the lipid ceramide in the yeast Saccharomyces cerevisiae . Thèse de doctorat : Univ. Genève, 2012, no. Sc. 4406

URN : urn:nbn:ch:unige-188513

DOI : 10.13097/archive-ouverte/unige:18851

Available at:

http://archive-ouverte.unige.ch/unige:18851

Disclaimer: layout of this document may differ from the published version.

1 / 1

(2)

UNIVERSITÉ DE GENÈVE

Section de chimie et biochimie FACULTÉ DES SCIENCES

Département de biochimie Professeur Howard Riezman

Les différentes fonctions physiologiques des ceramides chez la levure Saccharomyces cerevisiae

THÈSE

présentée à la Faculté des sciences de l’Université de Genève pour obtenir le grade de Docteur ès sciences, mention biochimie

par

Sharon EPSTEIN

de Allemagne

Thèse N

^o

4406

GENÈVE Atelier Repromail

2012

(3)

(4)

Remerciements (acknowledgements, agradecimentos) :

Cette thèse n’aurait pu aboutir sans la présence de mon précieux jury de thèse composé de Manuel Muniz, de Reika Watanabe et de Howard Riezman. Merci encore d’avoir accepté de faire partie de mon jury de thèse.

Merci encore à Howard Riezman de m’avoir accueilli dans son laboratoire.

Merci aussi aux actuels et anciens membres du laboratoire pour leur précieuse aide, surtout mon trois « mentors » : Cleiton De Souza, Guillaume Castillon et Thomas Hannich.

Merci aux membres du département et aux membres du secrétariat.

Je voudrais présenter mes remerciements aux personnes avec qui j’ai travaillé

dans le passé, et qui ont fortement contribué à ma formation.

(5)

The Plan

Résumé en Français………7

I-‐Introduction………..………….………..10

I.A-‐ Ceramides….………..……….…….…………..……….11

I.A.1-‐CerS, the mammalian ceramide synthases……….…..………12

I.A.2-‐Hyl genes, the worm ceramide synthase……….…..……14

I.A.3-‐-‐-‐Sphingolipid chain length……….…..………15

I.B-‐ Review: Sphingolipid signaling in yeast: potential implications for understanding disease (Epstein and Riezman, 2011)………..…..………….17

I.C-‐-‐-‐UPR and the p24 complex.………..……….31

I.C.1-‐-‐-‐UPR in yeast………..…..………....31

I.C.2-‐-‐-‐UPR in mammalian cells……….…..………..32

I.C.3. The p24 complex in yeast and mammalian cells………..32

II-‐Thesis project………..………35

III-‐Results……….………37

III.A-‐ An essential function of sphingolipids in cytokinesis (Epstein et al., submitted manuscript)……….……...………..…….….………38

III.B-‐ Activation of the unfolded protein response pathway causes ceramide accumulation in yeast and INS-‐1E insulinoma cells (Epstein et al., submitted manuscript)………..………..…….……..……….74

III.C-‐ Loss of ceramide synthase 3 causes lethal skin barrier disruption

(Jennemann et al., 2011)………..………..…………..…………..100

(6)

III.D-‐ The yeast p24 complex regulates GPI-‐anchored protein transport and quality control by monitoring anchor remodeling (Castillon et al., 2011)………125

III.E-‐ Protection of C. elegans from anoxia by HYL-‐2 ceramide synthase (Menuz et al., 2009)………..………140 IV-‐Discussion, Outlook and Conclusion………..………..………175 V-‐References………..……..………183

(7)

Le résumé

Les Céramides sont des composants lipidiques cellulaires essentiels. Ils représentent la colonne vertébrale de tous les sphingolipides. Il a été observé qu’ils jouent un rôle important en tant que molécules «signal» lors de différentes transmissions de signal. Les Céramides sont composés d’une base sphingoïde attachée à un acide gras via une liaison amide. Tous les Eucaryotes supérieurs possèdent ce lipide et lors des vingt dernières années les enzymes qui catalysent sa synthèse ont été identifiées dans une grande variété d’organismes.

Le but de ce travail de thèse est d’étudier les fonctions physiologiques des Céramides dans la levure en tant qu’organisme modèle. Comment la longueur de l’acide gras incorporé aux Céramides et par conséquence aux sphingolipides complexes influence des processus physiologiques? Quelles autres voies de transduction du signal sont influencées par les niveaux de Céramide dans d’autres organismes modèles tels que le Nématode?

En premier lieu nous avons cherché à comprendre les effets de l’accumulation de

sphingolipides à courtes chaines lipidiques résultant de l’expression dans la levure de la

céramide-‐synthase de plante. L’introduction de la céramide-‐synthase de plante engendre la

production de ces Céramides à courtes chaines lipidiques à des niveaux mille fois supérieurs

au niveau normalement trouvé dans la levure. Cependant cela n’est pas toxique pour la

levure. Nous avons alors pu démontrer qu’il est possible de supprimer le gène essentiel

AUR1 dans cette souche de levure et ainsi obtenu une souche de Levure incapable de

synthétiser des sphingolipides complexes. Cette nouvelle souche présente un défaut de

cytocinèse et accumule des dropelettes lipidiques.

(8)

Dans un autre travail, nous avons collaboré à la caractérisation des acides gras utilisés préférentiellement par deux des céramide-‐synthases de mammifères. Nous avons utilisé la levure comme système d’expression de gènes hétérologues et avons pu montrer que la céramide-‐synthase de mammifères CerS3, et non CerS2, incorpore des acides gras à longue chaine dans les Céramides.

Nous avons également montré que les Nématodes déficitaires pour une céramide synthase spécifique (HYL-‐2) sont très sensibles à l’anoxie. Après optimisation d’un test enzymatique in vitro, nous avons pu observer que HYL-‐2 utilise préférentiellement des acides gras à chaine longue.

Durant l’étude des céramide-‐synthases nous avons pu comprendre un phénotype vieux de plus de 10 ans. Dans la levure, la suppression du gène EMP24 sauve un des phénotypes de la souche déficiente pour l’activité de la serine palmitoyl transférase encodée par LCB1. La mutation du gène LCB1 (lcb1-‐100) conduit à un défaut de synthèse des sphingolipides qui sont essentiels à la bonne association à la membrane et localisation des protéines à ancrage GPI (Glycosyl Phosphatidyl Inositol). Nous savons par ailleurs que la suppression du gène EMP24 induit l’activation de l’UPR (Unfolded Protein Response). Nous avons alors décidé d’établir s’il y a un lien entre le défaut de niveaux de sphingolipides et le stress du Réticulum Endoplasmique (RE).

Dans la deuxième partie de la thèse, nous avons élucidé le mécanisme de sauvetage

phénotypique de la souche lcb1-‐100 par l’absence du gène EMP24. La suppression du gène

EMP24 élève les niveaux de céramides et restaure partiellement la viabilité de la souche

lcb1-‐100 à 35°C. Ce sauvetage phénotypique peut être reproduit par l’addition de DTT et

donc par l’activation de l’UPR. La suppression du gène HAC1 empêche le sauvetage

(9)

phénotypique de lcb1-‐100 par l’addition de DTT. L’induction de la synthèse de céramide suite à un stress du RE peut être reproduite dans les cellules Insulinoma (INS-‐1). Dans les cellules INS-‐1 un stress du RE induit la surproduction de céramide dont l’acide gras contient 16 carbones (C16). De manière concomitante les niveaux d’ARN messagers de CerS6 (la céramide-‐synthase spécifique pour la synthèse des céramides à C16 chaine) sont augmentés.

Cette étude révèle un lien nouveau entre les céramides et l’UPR, lien conservé chez la levure

et chez les cellules de mammifères.

(10)

I-‐Introduction

(11)

I.A-‐Ceramides

Ceramides are important cellular lipid components. They are the backbone for all sphingolipids and have been suggested to play important roles as second messengers in various signaling pathways. Ceramides are composed of an sphingoid base attached through an amide bond to a fatty acid (fig. 1). This lipid is found in all higher eukaryotes and in the past 20 years the enzymes that catalyze its formation have been identified in a variety of organisms. The discovery of the LAG1 and LAC1 genes in yeast as bona fide ceramide synthases was the first step towards the identification of homologous genes in other organisms(Guillas et al., 2001; Schorling et al., 2001). The importance of ceramides in yeast and the entire sphingolipid pathway is discussed in part IB, which has been published as a review in Frontiers in Biosciences. Therefore, in this first part of the introduction I will focus on what is known about the ceramide synthases in other organisms. After the identification of the LAG1 gene in yeast a bioinformatics approach showed homologs in mammalian systems (6 genes) and C. elegans (3 genes)(Jiang et al., 1998). In plants, one homolog was found in tomato plants and named Asc1(Brandwagt et al., 2000) while in cotton plants, two homologs were identified (Yongmei Qin, personal communication). In the yeast Pichia pastoris two orthologs of Lag1p have been found and one of them was shown to be a bona fide ceramide synthase, with a preference for short chained fatty acids (in this case 16-‐18 carbons) (Ternes et al.). In Arabidopsis thaliana three orthologs were identified and characterized, by knockdown and overexpression and also showed to be each, specific

and part of the TLC domain; the relevance of these short isoforms is currently unknown but may imply similar mechanisms of transcriptional regulation. Fig. 3 shows the genomic organization and protein isoforms of one of theLASSgenes, namely LASS2. Recently, a splice variant ofLASS5has been shown to be expressed in lymphoma and other tumor cells and may be involved in tumor recognition by the immune system (29).

The Lass genes appear to encode multi-transmembrane (TM) spanning proteins. The exact number of TM domains, and their topology, has not been resolved experimentally (9, 20, 25), although a recent study suggested that the yeast proteins, Lag1 and Lac1, contain eight putative TM domains with the N and C termini of the proteins facing the cytoplasm (30). The subcellular location of the CerS proteins (at least those for which experimental data is available) is the ER (27, 31, 32), consistent with earlier observations (11, 12) and similar to the location of Tram proteins (22).

The yeast genes, Lag1 and Lac1, act in an obligate complex with an additional protein, Lip1 (33), an integral ER membrane protein with one predicted TM domain. The Lip1 regions required for CerS activity may be in the membrane or in the lumen of the ER.

Mammalian homologs of Lip1 have not been found in data base searches. The activity of mammalian Lass proteins might conceiv- ably be regulated by other TLC family members (34).

The first evidence for specific functional roles of mammalian Lassgenes was obtained upon overexpression of LASS1 (for- merly known as UOG1), which resulted in a selective increase in C18-ceramide in mammalian cells (31). LASS4 (TRH1) and LASS5 (TRH4) were subsequently shown to selectively utilize C18/20 and C16 acyl-CoAs, respectively (32), LASS6 to produce shorter acyl chain ceramides (C14 and C16) (27), and LASS3 to produce C18- and C24-ceramides (35), although the surprisingly high levels of C18-ceramide synthesis are at vari-

ance with other analyses.⁵Verification that mammalian LASS proteins arebona fideceramide synthases, rather than regulators of endogenous ceramide synthases, was obtained when purified LASS5 was shown to possess CerS activity (36).

Together with the CerS activity of the purified Lag1-Lac1-Lip1 complex in yeast (33), this supports the concept that LASS proteins are genuine ceramide synthases, with each mammalian protein utilizing a relatively restricted subset of fatty acyl-CoAs.

It is assumed that the six known mammalian LASS proteins account for the synthesis of all known ceramides, but the pos- sibility cannot be excluded that some other proteins, such as ceramidases (37), may also contribute to the synthesis of ceramides with restricted fatty acid composition.

Roles of Ceramides Containing Distinct Fatty Acids The reason that mammals (and other species such as plants (38)) have multiple CerSgenes, whereas most of the other enzymes in the SL biosynthetic pathway exist in only one or two isoforms (9), is not known but implies an important role for ceramides containing specific fatty acids in cell physiology.

Support for this notion has been provided by the development of new and more sensitive analytical techniques, particularly mass spectrometry (8, 39, 40), enabling analysis of the fatty acid composition of ceramides using relatively small amounts of tis- sue or cells. Mass spectrometric analysis has revealed that spe-

5I. Pankova-Kholmyansky, S. Epstein, E. Wang, J. C. Allegood, S. Kelly, A. H.

Merrill, Jr., and A. H. Futerman, unpublished observations.

FIGURE 2.Phylogenetic tree of the 16 human TLC domain-containing pro- teins. Sequences were taken from Swiss-Prot with the exception of H17C473.1, which does not have a full-length mRNA in human and is based on a gene model from Ecgene and which closely matches the mouse cDNA available for the gene. Alignment was performed using ClustalW (version 1.82). One hundred data sets were created by Seqboot in the Phylip package (version 3.65). The trees were built with Proml (maximum likelihood, Phylip package), and a consensus tree was constructed by Consense (Phylip). A tree with the same topology was obtained using Neighbor Joining (in ClustalW) with 1000 bootstrap values. The tree wascoloredon the branches with bootstrap values of 1000. Thenon-coloredbranches had insignificant bootstrap values.

FIGURE 1.Structure of ceramide.Ceramide consists of a sphingoid long chain base (shown inblack), normally sphingosine, dihydrosphingosine (sphinganine), or 4-hydroxysphinganine (phytosphingosine), to which a fatty acid (shown inblue) is attached via an amide bond at C-2. The sphingoid base in the figure is sphingosine (which differs from sphinganine inasmuch as it con- tains atrans4 –5 double bond). Naturally occurring ceramide exists in the D-erythroconformation (2S,3R). The fatty acid in the figure is palmitic acid, one of the major fatty acids found in ceramide, but ceramides contain a wide spectrum of fatty acids.

MINIREVIEW: Regulation of Ceramide Synthesis

25002 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 281 • NUMBER 35 •SEPTEMBER 1, 2006

at Bibliotheque Faculte Medecine Geneve, on November 21, 2011www.jbc.orgDownloaded from

Fig. 1 – Structure of a ceramide containing a sphingosine (in black) and a C16 fatty acid (in blue) (Pewzner-‐Jung et al., 2006).

(12)

towards different fatty acids(Ternes et al., 2011).

The ceramide synthase enzymes are transmembrane proteins located in the ER. In yeast it has been shown that they span the membrane eight times and have both C and N terminus facing the cytoplasm(Kageyama-‐Yahara and Riezman, 2006). They belong to a large family of proteins that share a domain of approximately 200 amino acids and is called Tram-‐Lag-‐CLN8 (TLC) after the first proteins in which they were identified(Yu et al., 2006). However all enzymes shown to date to be bona fide ceramide synthases have an even smaller domain, of 52 amino acids, called the lag motif that is not found in other members of the TLC family (Venkataraman and Futerman, 2002). Inside this domain many residues have been shown to be essential for catalysis, including two histidines, which may be essential for substrate binding(Yu et al., 2006). Recently a minimal region of 150 amino acids has been identified as essential for CerS specificity towards acyl CoAs(Tidhar et al., 2011).

I.A.1-‐CerS, the mammalian ceramide synthases

As mentioned before, six homologs containing the lag1p motif were found in mammalian cells. This proteins were first called Lass 1-‐6 (for longevity assurance) but have since been renamed CerS 1-‐6(Pewzner-‐Jung et al., 2006). It is interesting to notice that contrary to yeast, where both Lag1p and Lac1p seem to have a preference for C26 acyl-‐

COAs, it was soon discovered, through protein overexpression, that the mammalian

homologs seem to be specialized in producing different kinds of ceramides, having a

preference towards fatty acids with specific chain lengths (fig. 2). The first one to be

characterized, CerS1 (then called Lass1), was shown to preferentially produce C18

ceramides(Venkataraman et al., 2002). Shortly afterwards CerS4 was shown to utilize C18

and to some extent C20; and CerS5, C16(Riebeling et al., 2003). CerS6, which is very closely

(13)

related to CerS5, also utilizes C16 and, in vitro, C14(Mizutani et al., 2005). CerS3 was the last one to be characterized and the first reports seemed to indicate that this enzyme had no specific preference, being able to utilize any fatty acid between 18 and 24 carbons(Mizutani et al., 2006). In section III.C however we present work that has been recently published in which our own results suggest that CerS3 actually has a preference for longer chained fatty acids (over 26 carbons) in vivo(Jennemann et al., 2011). To date, the only of these enzyme purified, CerS5, does not require additional components to catalyze the reaction (Lahiri and Futerman, 2005)., in contrast to yeast, where Lag1p and Lac1p require the co-‐factor Lip1p for their activity(Vallee and Riezman, 2005).

Further contributing to the importance of acyl COA length specificity of these enzymes is the fact that after characterization of their kinetics, the Km value towards the sphingoid base part of the ceramide producing reaction was very similar to all CerS enzymes(Lahiri et al., 2007). This would imply that the difference in activity is due to their

Fig. 2. The roles of CerS in synthesizing ceramides with different acyl chain lengths

Ceramides can differ in their acyl chain length, as shown in the figure, as well as in their in their degree of saturation and -hydroxylation (7). Sphinganine is show in blue, the acyl chain in yellow, and the CerS that synthesizes each ceramide is shown in red.

Levy and Futerman Page 16

IUBMB Life. Author manuscript; available in PMC 2011 May 1.

NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript

Fig. 2 – Each of the mammalian ceramide synthases (CerS in red) has a different specificity towards the fatty acid chain length (Levy and Futerman, 2010).

(14)

differential preference and affinity towards the acyl-‐COA part of the reaction. Also when the sphingosine analog FTY720 (a immunosuppressant drug) was tested for modulation of ceramide synthesis it was shown to have different inhibitory properties depending on the acyl-‐COA chain being used for the assay (Lahiri et al., 2009).

I.A.2-‐Hyl genes, the worm ceramide synthases

After the discovery of the LAG1 gene in yeast as a longevity assurance gene, database searches were conducted looking for homologs in diverse organisms. In C. elegans, the presence of three genes containing a lag1p motif was discovered. Two of these genes were named hyl-‐1 and -‐2 (which stands for Homolog of Yeast Longevity genes) and one named lagr-‐1 (for LAG related gene)(Tedesco et al., 2008). Another gene distantly related and containing a tram motif was found, but since it lacks a lag1p motif it is not thought to be a bona fide ceramide synthase and it has more amino acid differences towards Lag1 (fig. 3).

Deletion of only the hyl-‐1 gene however showed no effects on life span and only under very specific conditions did the inhibition of hyl-‐2 and lagr-‐1 by siRNA lead to shorter life span(Tedesco et al., 2008).

Both hyl-‐1 and hyl-‐2 however have been shown to be bona fide ceramide synthases since they are capable of replacing the yeast ceramide synthases(Menuz et al.,

with apvalue of 0.034. We used one negative control

vector for these experiments, pUC18, and the positive RNAi control vector,daf-2.

In many cases, RNAi can be more efficacious for longevity extension at lower levels of interference (Ventura and Rea2007), so we reduced the amount of the pJW9 RNAi vector to 10% and 1% by diluting the pJW9 RNAi strain with the empty vector strain pUC18.

Neither of these conditions gave a long-life phenotype (data not shown), indicating that the undiluted RNAi vector pJW9 was the best choice for inducing longevity. We also looked at the effect of growing wild- type worms for more than one generation on pJW9. In these experiments, increased life span was only seen with first generation RNAi worms and not with worms grown for two or three generations on pJW9.

hyl-1and other genes

hyl-1has three homologs (Table3), and it may be that more than one is involved in ceramide synthesis and specification of longevity. Therefore, we looked at these and a few other genes implicated in the aging process using RNAi mediated survival assays (Table4).

N2 animals treated with RNAi againsthyl-2,lagr-1, ortram-1did not have increased longevity, compared to empty vector. In fact, each of these vectors produced a decrease in mean life span, which reached significantpvalues forlagr-1with a 10% decrease. In these experiments,daf-2andage-1served as long- lived controls and showed an increase in longevity of 77% and 56% withpvalues <0.00001.

RT-PCR analysis

Given the results obtained with RNAi mediated survivals, we decided to look at gene expression

levels using RT-PCR. RT-PCR analysis showed significant changes in gene expression forhyl-1and its homologs (Table5). In the deletion strain TJ1091, levels ofhyl-1were undetectable but, surprisingly, we saw more than a 50% increase in levels of hyl-2 mRNA. We also observed decreased levels forlagr-1 (0.76-fold), and increased levels oftram-1of 1.29- fold. All of these changes were significant with p values of 0.004 or less, but did not lead to increased life span in the deletion mutant. By contrast, treatment with the pJW9 RNAi vector resulted in a significant decrease in both hyl-1and hyl-2, compared to the control vector pUC18. N2 worms on pJW9 hadhyl-1 and hyl-2 levels of 0.7 and 0.68, respectively, compared to the pUC18 vector (p=0.0002 and 0.025), but there were no significant changes in lagr-1ortram-1expression levels. This combination of expression level changes is apparently sufficient to produce a long-lived response.

hyl-1and thedaf-16pathway

We wanted to look at the interaction ofhyl-1with the daf-16insulin/IGF-1 like (IIL) signaling pathway and

0 20 40 60 80 100

0 10 20 30 40 50

day

% survival

Fig. 2 Combined survival data for pJW9 at 20°C. Wild-type animals were put onto various RNAi strains and followed for survival. Strains weredaf-2(♦), gfp (

⋄

^{), ht (}

△

), ev (×), pJW9 (

▪

), and pUC18 (□)

b

Fig. 1 (continued)

48 AGE (2008) 30:43–52

Fig.3 -‐ Phylogenetic tree of Lag1p homologs in C.

elegans (Tedesco et al., 2008)

(15)

2009).

I.A.3-‐Sphingolipid chain length

I.A.3.1-‐The mammalian paradigm

The discovery that higher eukaryotes have multiple ceramide synthases with different fatty acids specificities begs the question as to their functions? It is reasonable to assume that ceramides with different acyl chains and their downstream products, sphingolipids, will have different physiological roles and/or that these proteins will be differentially regulated.

Some evidence has been uncovered in recent years as to the specific roles of CerS genes. CerS1 was the first of the CerSs to be shown to play a potential role in cancer cells, when C18 ceramide was shown to be down regulated in neck squamous cell carcinoma(Koybasi et al., 2004). Overexpression of CerS1 was also shown to render cells more sensitive to a number of chemotherapeutic drugs and to translocate to the Golgi complex in order to decrease its activity(Min et al., 2007; Sridevi et al., 2009). CerS2 and CerS6 have been implicated in two cancer studies in which total ceramides levels and mRNA levels for these two genes were elevated(Erez-‐Roman et al., 2010; Schiffmann et al., 2009).

Regulation by other lipids has only been shown so far for CerS2, which is down regulated by sphingosine-‐1-‐Kinase(Laviad et al., 2008). CerS4 expression was shown to be elevated in the brain of mouse models for Alzheimer disease(Wang et al., 2008). Recently, evidence has surfaced for an involvement of ceramide synthases in the activation of UPR response (Spassieva et al., 2009). More on UPR will be discussed in the third part of this introduction.

The down regulation of CerS2 affected ceramide homeostasis leading to an increase in C16

(16)

ceramide levels, probably resulting from up regulation of CerS5 and Cers6 mRNAs. It also led to a series of physiological responses, including induction of UPR(Spassieva et al., 2009).

Another aspect of variable length sphingolipids is the different properties they might confer to membrane domains. An antibody raised specifically against mixtures of C16-‐

ceramide and cholesterol has shown the presence of organized lipid domains with such specific constitution in a very small subset of organelles(Goldschmidt-‐Arzi et al., 2011). Also, an in vitro assay using phosphatidylcholine membrane models showed that the size of ceramides has an impact on the morphology and size of gel domains and that only very long ceramides are capable of forming tubular structures(Pinto et al., 2011).

I.A.3.2-‐The worm examples

As will be described in section III.E the worm’s ceramide synthases have different preferences towards the fatty acid moieties utilized in the formation of ceramides. These differences lead to either survival or hypersensitivity to anoxia. However that is not the only example of how the modification of the sphingolipids chain length results in physiological perturbances in C. elegans. It had already been shown that the down regulation of both hyl-‐

1 and hyl-‐2 simultaneously by siRNA leads to increased life span (Tedesco et al., 2008).,

although the double deletion is lethal (Vincent Menuz, personal communication). However

deletion of either gene on its own showed no difference in longevity(Tedesco et al., 2008). It

was also shown that although deletion of the Lag homologs does not influence somatic

apoptosis, deletion of either hyl-‐1 or lagr-‐1 prevented ionizing radiation-‐induced

apoptosis(Deng et al., 2008). This phenotype was rescued by the exogenous addition of C16

ceramides, indicating that this lipid is an essential component of the apoptosis signaling

cascade(Deng et al., 2008).

(17)

I.B-‐ Review: Sphingolipid signaling in yeast: potential implications for

understanding disease

(18)

Sphingolipid signaling in yeast; potential implications for understanding disease Sharon Epstein and Howard Riezman

NCCR Chemical Biology, Department of Biochemistry, 30 quai Ernest Ansermet, University of Geneva, CH-1211 Geneva 4, Switzerland

TABLE OF CONTENTS

1. Abstract 2. Introduction

3. Sphingoid base, the upstream physiological regulators 4. Dihydroceramides

5. Downstream of dihydroceramides: complex sphingolipids 6. Systems approach

7. Conclusion 8. Acknowledgment 9. References 1. ABSTRACT

Sphingolipids are essential components of membranes and important for cellular integrity. The main focus of research in the past years has been to demonstrate their role as second messengers. The yeast Saccharomyces cerevisiae is an excellent model for the study of sphingolipids, because the first steps of this metabolic pathway are highly conserved among fungal, plant and the animal kingdoms. The yeast model is a valuable system for the understanding of pathways and development of tools that will help to better understand and intervene into the molecular mechanisms controlling health and disease.Different classes of sphingolipids have been shown to act in different pathways. Sphingoid bases were shown to be involved in protection against a series of stresses such as heat shock, osmotic stress and low pH. Ceramides have been shown to be involved in G1 arrest, heat shock response and more recently as a target of the TORC 2. Complex sphingolipids are essential for cell wall integrity and proper localization of GPI anchored proteins.

2. INTRODUCTION

The structural role of lipids in the function of cellular membranes has been appreciated for a long time.

However, it was not until the late 1990’s and early 2000’s that sphingolipids were also recognized to have a role as signaling molecules in different physiological pathways (1, 2). Since then much evidence, both in yeast and mammals, has been obtained to demonstrate roles as second messengers for different classes of sphingolipids. The yeast Saccharomyces cerevisiae is an excellent model for the study of sphingolipids, because the first steps of this metabolic pathway are highly conserved across fungal, plant and the animal kingdoms (3). The enzymes involved in the early part of this pathway were all identified with the help of yeast genetics. Up to the stage of synthesis of ceramide there are only minor differences with relation to chain length, hydroxylation and saturation of the chains in the synthesis of sphingolipids. Whereas in mammalian cells the steps after the synthesis of ceramide lead to the formation of an enormous variety of complex sphingolipids, with a multitude of different head groups, in S.

cerevisiae only three types of complex sphingolipids are produced: IPC, MIPC and M(IP)₂C. This diminished complexity, although it cannot perfectly mimic the mammalian diversity, provides a much easier model system to manipulate, that can, nonetheless, give us important insights into the function of complex sphingolipids. For a more complex fungal model one could turn to Pichia pastoris which has both inositolphosphorylceramides and glucosylceramides(4).

The sphingolipid metabolic route in yeast starts with the serine palmitoyltransferase, composed of three subunits, Lcb1p, Lcb2p and Tsc3p (Figure 1), which synthesizes 3-ketosphinganine and which have mammalian homologs. This enzyme is essential for cell viability in yeast, but a deletion mutant can be grown by supplementation with sphinganine or 4-hydroxysphinganine. A temperature sensitive allele has been isolated (5) that has allowed the study of the role of sphingolipid biosynthesis in many processes. This allele has allowed researchers to conditionally inactivate Lcb1p, which stops the de novo synthesis of sphingolipids and modulates its pathway. Many studies in signaling link the sphingoid bases produced after this step and before the synthesis of ceramide to diverse physiological processes leading to the hypothesis that the two yeast sphingoid bases (PHS and DHS) and their phosphorylated forms are crucial as second messengers, as will be discussed bellow.

The next crucial step in sphingolipid biosynthesis is the condensation of sphingoid bases and fatty acids to form dihydroceramide. The dihydroceramide synthase genes were first discovered in yeast as two redundant homologs called Lag1p and Lac1p(6, 7). They were soon proved to be required for dihydroceramide synthase activity and orthologs were found in most species(8). The deletion of either of the dihydroceramide synthases alone is not enough to suppress the ceramide production, has no clear growth phenotype, but expands life span of yeast cells up to 50%(9). The deletion of both genes can be lethal in certain yeast strain backgrounds(6, 10), but in others the strain is viable and makes some still not fully identified lipids. The Lac1p and Lag1p proteins were purified together with another essential subunit, Lip1p, and the complex showed dihydroceramide synthase activity in vitro demonstrating that the genes encode a bona fide ceramide synthase (11). No clear homologs of the LIP1

(19)

subunit have been found in higher eukaryotes and some of these mammalian homologs of LAG1 do not require LIP1 for activity when expressed in yeast(12).

In yeast, the endogenous dihydroceramide synthases enzymes are only capable of utilizing very long chain fatty acids (with a strong preference for C26, and to a much smaller extent C24) and this has an effect on downstream complex sphingolipids that all have very long chains. Mammalian cells have six orthologs that have been named CerS 1 through 6. Various studies have shown that each of CerS enyzmes has different fatty acid specificity and the variations in the length of the fatty acyl chains have different physiological consequences(12- 17). One of these proteins has been purified and shown to be a bona fide dihydroceramide synthase that, unlike its yeast counterpart, does not require any other subunits for functionality(13). The difference in chain length can therefore be studied in yeast by the replacement of the endogenous dihydroceramide synthases by the expression of the mammalian homologs. This has been done successfully for all 6 genes(18), but little has been shown in terms of physiological effects for the moment.

At this stage two hydroxylations can be added to modify the ceramides. SUR2 is the gene responsible for the hydroxylation of the sphingoid base and SCS7 for the fatty acid(19). Another important gene is ISC1 that encodes a yeast homolog of the mammalian sphingomyelinase(20), which. hydolyzes complex sphingolipids producing ceramide and releasing the head group, thereby recycling complex lipids. After the synthesis of dihydroceramide, the pathways in yeast and mammalian cells diverge. While mammals produce a wide variety of complex sphingolipids, starting with sphingomyelin and glucosylceramide, yeast has only three complex sphingolipids formed in a sequential manner. The first step is the addition of an inositol phosphate using phosphatidylinositol as a donor by an enzyme called Aur1p, forming an inositol phosphoryceramide (IPC).

Deletion of AUR1 is lethal(21) and its activity can be blocked by the antifungal Aureobasidin A. Currently there is much discussion in the literature if this lethality is the consequence of accumulation of dihydroceramide or the lack of complex sphingolipids. Some mutants allow cells to grow in the presence of AbA, notably strains defective in dihydroceramide synthase (8) and the elongation of fatty acids, which also reduces dihydroceramide synthesis, suggesting that the very long chain ceramide is toxic(22). There is also evidence that complex sphingolipids can be required for growth because another group found that even the lag1 lac1 double mutant, which is resistant to AbA since it does not accumulate ceramide, became inviable after being grown consecutive times in the presence of AbA(23). Recently another subunit essential for IPC synthesis has been discovered, Kei1p. It is thought that this protein is essential for Aur1p transport from the ER to Golgi and without it cells show decreased levels of IPC(24).

The next step in sphingolipid biosynthesis in yeast is the addition of a mannose from GDP-mannose, forming mannose inositol phosphorylceramide (MIPC). This reaction is catalyzed by either Sur1p (Csg1p) or Csh1p, the catalytic subunits, that together with Csg2p, the regulatory subunit, (25, 26), produces the reaction.

Deletion of SUR1 and CSH1 render the cells unable to produce MIPC and have reduced levels of M(IP)2C, but this is not lethal, indicating that complex mannosylated sphingolipids are not essential for the survival of yeast cells(25).

The last step in sphingolipid biosynthesis in yeast is the addition of a second inositol phosphate from phophatidylinositol to MIPC forming M(IP)2C(27) and is carried out by Ipt1p. A deletion mutant of IPT1 is viable indicating that the final inositol phosphorylation is not required for viability.

3. SPHINGOID BASE, THE UPSTREAM PHYSIOLOGICAL REGULATORS

In yeast there are two main forms of sphingoid bases, sphinganine (called dihydrosphingosine, DHS) and its hydroxylated form 4-hydroxysphinganine(called phytosphingosine, PHS). Both are present in very low amounts in yeast cells and can either be converted into dihydroceramide or phosphorylated to form DHS1-P and PHS1-P respectively. The first experiments with these molecules were done in mammalian cells but the discovery of the related genes in yeast provided a simpler and more effective way to study the intracellular roles of sphingoid bases.

The first clues in yeast for a signaling role for sphingoid bases was the discovery that cells defective in their synthesis were viable but unable to survive a series of stresses such as heat shock, osmotic stress and low pH (28).

Two genes encoding sphingoid base kinases, Lcb4p and Lcb5p, have been identified in yeast and the corresponding mutants have been useful to probe the function of sphingoid bases and their phosphorylated derivatives(29). These, together with dpl1 (sphinganine 1 phosphate lyase) and other mutants in the pathway were used to probe the role of long chain bases. The accumulation of sphingoid base phosphates proved to inhibit cell growth(30). The levels of accumulation in this study were however much higher (>70 fold) than the ones observed under heat shock conditions (5 to 8 fold). This indicates that total amounts and the duration of their elevation are essential to define if sphingoid bases accumulation will lead to cell proliferation or arrest(30).

Heat shock stress provided the first experiments where sphingoid bases were shown to be second messengers. PHS1-P and DHS1-P were shown to exist in very small amounts that increased transiently during heat shock, in a mechanism essential to help the cells survive the temperature change(31). The deletion of the sphingosine phosphate lyase (DPL1) or of one of the phosphatases (LCB3), thus impeding the catabolism of the phosphorylated lipids, led to an increase in heat shock resistance and a better survival rate than the WT strain, while the down regulation of Lcb1p (serine palmitoyl tranferase) lead to cell death via the inability of the cells to transiently increase their levels of PHS1-P and DHS1-P(31). One of the mechanisms of action of resistance involves the induction of heat shock proteins (HSPs) which help to disaggregate and refold misfolded or aggregated proteins. When the HSPs are unsuccessful another process increased by heat shock, ubiquitin- dependent degradation of misfolded proteins takes over. The lcb1-100 mutant fails to induce heat shock proteins upon heat shock, but the strain was able to survive the heat shock if ubiquitin was overexpressed (32) suggesting that it is not the loss of activity of proteins through heat shock that is most critical under these conditions, but the

(20)

accumulation of misfolded or aggregated proteins. These findings could be relevant to diseases associated with protein misfolding and aggregation. Another role of sphingoid bases during heat shock is in protein translation control. The increase in sphingoid bases seems to be required for translation initiation of the heat shock proteins(33), which are translated during heat shock while other mRNAs are not. Furthermore, the deletion of the lyase gene, DPL1, leads to the accumulation of sphinganine-1-phosphate, which under certain conditions has been demonstrated to cause a block of cell division and a failure to recruit cells to G1 phase (34).

The above cases are proposed functions of phosphorylated forms of sphingoid bases, but several physiological functions in yeast seem to depend upon sphingoid bases. The absence of sphingoid bases leads to a defect in the internalization step of endocytosis and proper actin organization, which can be rescued by the addition of external bases(35). The blockage of sphingoid base phosphorylation does not seem to play a role in this process, and this study in yeast was the first to propose a role for sphingoid bases as active molecules instead of their phosphorylated forms. The mechanism of action was later found to be mediated by the Pkh kinases, homologs of the PDK kinases in animals, and the overexpression of the Pkh kinases led to a rescue of the endocytic defect(36). In vitro experiments showed that even nanomolar amounts of sphingoid bases were capable of activating the kinases and downstream effectors were identified(36).

Sphingoid bases have also been linked to calcium influx in both yeast and mammalian cells. In yeast the role of the two kinases (Lcb4p and Lcb5p) was demonstrated to be essential when exogenous sphingosine was added to stimulate the calcium signaling pathway. The absence of the kinases rendered the addition of the sphingosine innocuous, while the activity of the lyase and phosphatase (Dpl1p and Lcb3p) inhibited the activation of the pathway(37).

Very recently yeast has been used to investigate Parkinson disease associated toxicity of alpha- synuclein. This protein has been linked to Parkinson’s and is thought to be involved in other neurodegenerative diseases. They expressed this protein and a subset of its mutants in yeast defective in sphingolipid metabolism and looked for increased toxicity. The study was done only on single mutants which probably prevented the finding of additional interactions with the sphingolipid pathway, but one subset of genes that did emerge from the screens were the ELO1, FEN1 and SUR4. These 3 genes are fatty acid elongases that work in sequence to produce the very long chain fatty acids used for dihydroceramide synthesis. All three mutants showed increased sensitivity to the wt alpha-sync as well as to two of the three variants tested, as well as less viability of old cells(38). The cause of this sensitivity remains elusive. The elo mutants lower amounts of sphingolipids, due to the specificity of the dihydroceramide synthases, which have a low affinity towards shorter (<26 C) fatty acyl CoAs. The approach using yeast is a way to simplify the study of a very complex problem and the insights gained in yeast might be applicable to neuronal cells, which would make it useful for the development of drugs and markers for neurodegenerative malignancies.

Yeast has also been used as a model system to investigate the metabolism and possible targets of the immunosuppressant FTY720, a sphingolipid analog (39) that has recently been approved to treat multiple sclerosis(40). In vertebrates the mechanism of action that seems to be important for FTY720 involve its phosphorylation by sphingosine kinase 2 and action as a sphingosine-1P mimic(41). It has also been shown to inhibit sphingosine kinase 1(42). In yeast the effects do not seem to depend upon FTY720 phosphorylation suggesting that the molecule might have other effects, mimicking sphingoid bases, that need to be understood, including effects on the ubiquitin pathway, trafficking of amino acid permeases, and on transcriptional profiles(39, 43).

FTY720 has also been reported to inhibit the sphingosine phosphate lyase(44). It is not known if this inhibition plays a physiological role in the mechanism of action of FTY720. Structure-function relationships in the yeast homolog of the enzyme, Dpl1p, have been studied(45)and provide information about the localization and function of the enzyme. Furthermore, recent studies have determined the 3-D structure of a related enzyme from bacteria by X-ray crystallography (46), which has allowed the modeling of the structure of the eukaryotic enzyme.

Information about the active site of the enzyme and its structure should allow the design of novel inhibitors and perhaps other modulators of the lyase activity. A specific inhibitor should allow the dissection of the role of this inhibition in physiological processes.

4. DIHYDROCERAMIDES

Sphingoid bases and their phosphorylated forms are not the only intermediates in the sphingolipid biosynthesis pathway with signaling functions. Dihydroceramides themselves have been shown to be involved in G1 arrest, heat shock response and more recently as a target of the TORC 2 (target of rapamycin complex 2) pathway(47-50).

The first insight into how ceramides could act as second messenger in yeast came in 1993 (51) where the investigators showed that small amounts of soluble ceramides inhibited cell growth in yeast. The treated cells had an activated phosphatase, that could be inhibited by okadaic acid, making it a class 2A ceramide activated phosphatase (CAPP)(51). Nickels and Broach extended this study showing that ceramide can activate CAPP, whose catalytic subunit is encoded by SIT4 and regulatory subunits by CDC55 and TPD3. Activation of CAPP leads to G1 arrest(49) . It was also shown that this pathway could be counteracting a cAMP-dependent protein kinase of the RAS pathway.

Ceramide has also been implicated in heat stress response. Although much of the research in this area has focused on the transient increase in sphingoid bases (see above), the heat stress also generates a more durable elevation of ceramides. This elevation is the result of de novo synthesis, because the addition of australifungin, a dihydroceramide synthase inhibitor, was shown to block the increase(50). This finding differed from what was

(21)

postulated previously about ceramide generation under stress conditions in vertebrates, where most of the ceramides produced come from the degradation of complex sphingolipids. This finding illustrated the importance of the de novo pathway and encouraged the study of dihydroceramide synthesis as a possible candidate source for signaling molecules.

More recently the interaction between sphingolipid metabolism and the TOR pathway has been demonstrated. The TOR kinase, which was first identified as the target of rapamycin(52) and has been shown to regulate cell growth and metabolism. It forms two complexes, TORC1 and TORC2, of which only the former is sensitive to rapamycin(53). The kinase gene is conserved through evolution in eukaryotes and its study in mammalian cells has associated it with several diseases such as cancer, cardiovascular, autoimmunity and metabolic disorders. Many excellent reviews exist on the subject(54-56), one of which discusses the relationship of TORC with lipid synthesis, specifically its control of lipogenesis. In yeast the TORC2 complex clearly has an influence on sphingolipid metabolism, however, the precise mechanism is still unclear. The most direct experiments involve the investigation of the function of AVO3, which encodes a subunit of the TORC2 complex. It was shown that incubation of a temperature sensitive avo3-30 mutant at nonpermissive temperature led to a reduction in ceramide levels and an increase in phosphorylated sphingoid bases. This mutant has a slow growth phenotype suggesting that the lack of ceramide and complex sphingolipids led to cell cycle arrest or cell death(47).

The precise mechanism of this regulation is unclear, however, they showed a genetic interaction with the calcineurin pathway. The calcineurin pathway has previously been shown to interact with TORC2 (57) and with another set of homologous proteins, Slm1p and Slm2p(58). Furthermore, the Slm proteins have been implicated in regulation of sphingolipids(59) and the actin cytoskeleton, another function of TORC2(60). More recently, the plekstrin homology domain of the Slm proteins, which are required for actin organization and bind phosphoinositides, has been shown also to bind sphingolipids(61). It will be very interesting to see to what extent this regulation can be reproduced in vertebrates as there are no obvious Slm1p homologs.

In a systematic synthetic interaction screen with a thermosensitive mutant in the phosphatidylinositol transfer protein (Sec14p) implicated in the secretory pathway and Golgi function, a strong interactor was the snare protein Tlg2(48), which functions in membrane trafficking associated with the Golgi complex and endosomes(62- 63). The combination of the sec14 and tlg2 mutations affected the TOR signaling pathway, the unfolded protein response (UPR) pathway in the endoplasmic reticulum and caused the accumulation of ceramides perhaps the cause of the UPR. The proposed mechanism is that the double defect in trafficking around the Golgi compartment causes an increased catabolism of complex sphingolipids (IPC, MIPC and MIP₂C) that would elevate the pools of ceramide(48). This elevation would in turn affect a ceramide activated phosphatase in a similar mechanism to that proposed above.

Apart from signaling functions ceramides are also important in the intracellular trafficking of GPI- anchored proteins. The anchors of most GPI(glycosylphosphatidylinositol)-anchored protein in S. cerevisiae are remodeled from a diacylglycerol structure to a ceramide structure in the endoplasmic reticulum, with some contribution from later compartments(64). In a screen for inhibitors of GPI-anchored protein biogenesis, a potent inhibitor of serine-palmitoyltransferase, myriocin, was found(65). The synthesis of ceramide is critical for GPI- anchored protein transport because only stereoisomers of sphinganine that can be incorporated into ceramide can restore transport when serine palmitoyltransferase is blocked(66). GPI-anchor remodeling is required for ER exit(67) acting at the step of concentration into ER exit sites(68). In mammalian cells ceramide synthesis is not required for GPI-anchored protein transport(69), however, the process of remodeling is required for ER exit(70), and the mechanism of transport seems to be conserved although the organization of the pathway is somewhat different with respect to the sites and nature of the latter remodeling steps. Defects in GPI biosynthesis can lead to diseases, such as paroxysmal nocturnal hemoglobinuria(71).

Ceramides can also be modified by hydroxylation of either the sphingoid base or the fatty acid moiety.

SCS7, the gene that introduces a hydroxyl group to position 2 of fatty acids has been shown to be important for resistance to the drug PM02734, a novel synthetic antitumor drug. Its mode of action is the induction of rapid necrotic cell death in yeast. The deletion of SCS7 renders cells more resistant to necrosis and these results have been validated in mammalian cells, where SCS7 has a homolog, FA2H(72).

Another pathway for the formation of ceramide is the degradation of complex sphingolipids. In mammalian cells this function is carried out by sphingomyelinases and in yeast by a single gene, ISC1, which is capable of cleaving the headgroups of different complex sphingolipids. It has been shown that the mammalian sphingomyelinase 2 is capable of rescuing the yeast ISC1 deletion (20). This deletion strain also showed cell cycle defects, being blocked at the G2/M phase, when treated with methyl methanasulfonate or hydroxyurea(73) and having a lower life span with death by apoptosis when treated with hydrogen peroxidase(74). Curiously, the protein encoded by ISC1, which is normally localized in the endoplasmic reticulum has been located in the mitochondria following glucose depletion treatment or late in the growth phase (73). These results suggest that the higher levels of ceramide seen in apoptotic cells might come from the degradation of complex sphingolipids rather then from the de novo synthesis and this mechanism is conserved in yeast. Furthermore ISC1 has been shown to be involved in other stress response pathways like the halotolerance against Na+ and Li+ ions(75).

5. DOWNSTREAM OF DIHYDROCERAMIDES: COMPLEX SPHINGOLIPIDS

Although there is much data on the role of the products of earlier steps of sphingolipid biosynthesis in physiological functions, much less is known about the roles played by complex sphingolipids. The synthesis of complex sphingolipids is simpler in S. cerevisiae than in higher eukaryotes. Mammalian cells produce sphingomyelin and glucosylceramides, the latter being transformed into a series of different glycolipids. S.

The multiple physiological roles of the lipid ceramide in the yeast Saccharomyces cerevisiae

Thesis

Reference