דוקטור לפילוסופיה Doctor of Philosophy

עבודת גמר )תזה( לתואר Thesis for the degree דוקטור לפילוסופיה Doctor of Philosophy

מוגשת למועצה המדעית של Submitted to the Scientific Council of the מכון ויצמן למדע Weizmann Institute of Science רחובות, ישראל Rehovot, Israel

מאת By גיורא וולפרט Volpert Giora

תאור תפקידו של סרמיד סינטאז 2 בבריאות ובמחלות Delineating the role of ceramide synthase 2 in health and disease

מנחה: :Advisor פרופ' טוני פוטרמן Prof. Tony Futerman

ניסן התשע"ז April 2017 Acknowledgements

As I am approaching the end of my Ph.D., I would like to thank all those who made this journey professional, educational and fun as it was.

First I would like to thank my advisor Prof. Tony Futerman for introducing me to the intriguing world of sphingolipids. For knowing when to let me do science my way and when to give me a “kick in the rear-end”. For the guidance throughout endless meetings, brain storming and Skype calls. For the model of organization which I am striving to adopt, and for the opportunity to be part of an enthusiastic, supporting and open group of people that made this period very rewarding and satisfying.

I would like to specially thank to Dr. Andy Klein for all of our philosophical talks about scientific and non-scientific matters, for being the perfect collaborator to produce our movies and for just being a friend when I needed it.

I want to express my deep appreciation to Dr. Yael Pewzner-Jung and Tammar Joseph for the willingness to help anyone at any time, for the unconditional assistance, technical support and wise advice.

I would like to thank Dr. Elad Laviad, Dr. Rotem Tidhar, Dr. Oshrit Ben-David, Dr. Natalia Santos Frreira, Iris D Zelnik, Eden Rosenfeld-Gur, Dr. Einat Vitner (Vintage), Dr. Hila Zigdon (HilaZ), Chen Yaacobi (Cheng), Shani Blumenreich (Sheng), Ayelet Vardi (Ajelet), Lital Sahar (Oligo) and Inbal Polin (Polish). Thank you for being a great research partners, co- workers and friends.

To all other students in the present and the past, Dr. Modi Ali, Dr. Marton Magyeri, Dr. Ashish Saroha, Dr. Tamar Farfel-Becker, Dr. Woo-Jae Park, Dr. Soo Min Cho and Jiyoon Kim (Leah) for endless lunch breaks, discussions and research assistance.

To our best collaborator, Dr. Shifra Ben-Dor, for giving me the opportunity to work with you and learn from you how one should work professionally, creatively and quickly.

To my mother and late father who raised me to be curious and hard working. To my grand mothers and to my brother who try to understand what exactly I am doing in the lab.

Thank you to my two lovely daughters Daphne and Abigail for letting me know what’s really important in life.

And last but not least to my wife and life companion Yael (Yuli, KofT) Hollender-Volpert. Thank you for your love, support, encouragement and understanding. Thank you for making me a better person, I could not have finished this degree without you. This dissertation is dedicated to you.

2 Table of Contents 1. Abbreviations used...... 4 2. Abstract and main findings ...... 5 3. Introduction ...... 8 3.1. The lipid composition of eukaryotic membranes ...... 8 3.2. SL biosynthesis pathway ...... 9 3.3. Ceramide ...... 11 3.4. Ceramide as a structural unit of complex sphingolipids ...... 11 3.5. The (Dihydro)ceramide synthases (CerS) family ...... 12 3.6. CerS2 ...... 13 3.7. CerS transgenic mice ...... 14 3.8. Endocytosis – more than one way to get inside a cell ...... 15 3.9. CerS in health and disease ...... 17 4. Research objectives ...... 18 5. Materials and methods ...... 19 5.1. Mouse models ...... 19 5.2. Astrocyte culture ...... 19 5.3. Immunohistochemistry ...... 19 5.4. RNA extraction and polymerase chain reaction ...... 20 5.5. CerS assays ...... 21 5.6. Lipid analysis ...... 21 5.7. Proliferation and cell death ...... 21 5.8. Western blotting ...... 21 5.9. Uptake of fluorescently-labeled ligands ...... 22 5.10. Electron microscopy ...... 22 5.11. Transcription factor binding site analysis ...... 23 5.12. Chromatin immune precipitation...... 23 5.13. Reactive oxygen species ...... 23 5.14. Mitochondrial complex IV activity ...... 23 5.15. Confocal microscopy ...... 23 5.16. BM chimeras ...... 24 5.17. SL turnover ...... 24 6. Results ...... 25 6.1. Characterization of primary cultured astrocytes from a CerS2 null mouse ...... 25 6.2. Analyzing the effect of altered SL composition on endocytic pathways ...... 28 6.3. Characterization of CerS1/CerS2 double knockout mice ...... 36 6.4. Exploring the hyper-sensitivity of CerS2 null mice to DSS induced colitis ...... 39 6.5. The involvement of CerS dimerization in glioblastoma...... 41 6.6. Other collaborative projects ...... 42 7. Discussion...... 44 8. List of publications from PhD work ...... 50 9. References ...... 51 10. Declaration...... 62

3 1. Abbreviations used

ACBP Acyl-CoA binding protein BM Bone marrow CCP/CCV Clathrin coated pit/clathrin coated vesicle CerS Ceramide synthase CHC Clathrin heavy chain ChIP Chromatin immune-precipitation CME Clathrin-mediated endocytosis DSS Dextran sulfate sodium ER Endoplasmic reticulum FASN Fatty acid synthase GBM Glioblastoma GFAP Glial fibrillary acidic protein GLTP Glycolipid transfer protein Hsc70 Heat shock cognate 70 HSF1 Heat shock factor 1 KO/ dKO Knock out/double knock out LacCer Lactosylceramide LCB Long chain base LDL Low density lipoprotein NAC N-acetyl cysteine PAH Pulmonary arterial hypertension PBS Phosphate buffer PM Plasma membrane ROS Reactive oxygen species Sa/SaP/S1P Sphinganine /sphinganine-1-phosphate/sphingosine-1-phosphate Sp1 Specificity protein 1 SLs/ VLC-SLs Sphingolipids /very long chain sphingolipids SM/ SMase Sphingomyelin/sphingomyelinase Tf Transferrin TNFR1 Tumor necrosis factor receptor 1

4 2. Abstract and main findings Ceramides with defined acyl chain lengths are involved in a number of human diseases such like multiple sclerosis and several types of cancer, yet little is known about the specific roles of CerS in general and CerS2 in particular in human disease. Therefore, in my PhD research, I strove to understand the role of CerS2 in health and disease. Initially I explored the effect of the lack of very-long chain sphingolipids on cellular trafficking. I generated primary cultured astrocytes from wild type and CerS2 null mouse brains and used several fluorescently-labeled ligands to evaluate the kinetics of two major endocytic pathways: clathrin-mediated endocytosis (CME) and clathrin-independent caveolin-mediated endocytosis. CerS2 null astrocytes displayed slower kinetics towards ligands internalized by CME while no change was detected in the uptake rate of ligands internalized by caveolin- mediated endocytosis. When examining the mechanism of clathrin-mediated endocytosis in CerS2 null astrocytes, I found that Hsc70, a key protein in the machinery of clathrin uncoating, is down-regulated. I also measured the binding and protein levels of Sp1, the transcription factor responsible for Hsc70 expression, both of which were significantly reduced. Further analysis revealed that due to elevation in reactive oxygen species (ROS) in CerS2 null astrocytes, Sp1 levels, which are sensitive to oxidative stress, are reduced. Finally, I showed that I could mimic the slower uptake of CerS2 null astrocytes by addition of ROS to wild type cells and that anti-oxidant treatment of CerS2 null astrocytes rescues all phenotypes. This suggests that at least some of the pathologies of CerS2 null mice can be due to elevated oxidative stress. In collaboration with a fellow lab member, I characterized a CerS1/CerS2 double knockout mouse. This is the first example of a live mouse lacking two of the six CerS family members. These mice are born and live for 3-4 weeks, despite the fact that they only produce a small amount of C18-sphingolipids and fail to produce very-long chain sphingolipids. While the CerS1/CerS2 double knock out mice show no phenotype in their peripheral organs (liver, kidney, intestine, spleen, muscle), they do develop severe brain inflammation within 20 days. In conclusion, CerS1/CerS2 double knockout mice present an exacerbated version of the sum of the pathologies of their CerS1 knockout and CerS2 knockout parents. I also explored the hyper-sensitivity of CerS2 null mice to DSS-induced colitis. I asked whether the cause for the colitis is due to hyper activity of the CerS2 null immune system or due to perturbed barrier function of the CerS2 null intestinal epithelia (i.e. tight junction). Therefore, I generated four groups of bone marrow chimeric mice: wild type mice with either wild or the CerS2 null immune system, and CerS2 null mice with either wild type

5 or the CerS2 null immune system. These mice were administered with DSS and hyper- sensitivity was determined by their weight loss, bloody diarrhea, stool consistency and colon length. Interestingly, we found that the hyper-sensitivity of CerS2 null mice is not due to the immune system. We hypothesize that the underlying cause is related to perturbations in the CerS2 null intestinal barrier, which have yet to be uncovered. In collaboration with another lab, I checked the involvement of CerS dimerization in glioblastoma. I demonstrated that Bcl2L13 knockdown elevates the activity of both CerS2 and CerS6 after curcumin treatment, and that there is a negative correlation between Bcl2L13 mRNA levels and CerS2/CerS6 activity in human glioblastoma samples. Altogether, I found that CerS in general and CerS2 in particular have a role in the normal and healthy function of many processes. Additionally, elimination of CerS2 activity (genetically, chemically or biologically) leads to the pathogenesis of various cellular abnormalities and diseases such as oxidative stress, brain inflammation, colitis and cancer.

6 תקציר:

סרמידים בעלי שרשראות באורכים מסויימים מעורבים במספר מחלות אנושיות כגון טרשת נפוצה ומספר סוגים של סרטן, אם כי מעט מאד ידוע על תפקידם של סרמיד סינטאזות (CerS) בכלל וסרמיד סינטאז CerS2) 2) בפרט במחלות שכאלו. על כן, במהלך הדוקטורט שלי חתרתי להבין את תפקידו של CerS2 הן בבריאות והן במחלות.ראשית חקרתי את ההשפעה של חוסר בספינגוליפידים בעלי שרשראות ארוכות מאד על תעבורה תוך תאית. גידלתי אסטרוציטים ראשוניים שהופקו ממוחות של עכברי זן הבר ומעכברים ללא CerS2 והשתמשתי במספר ליגנדים פלורוצנטיים על מנת לאמוד את הקינטיקה של שני מנגנוני אנדוציטוזה: אנדוציטוזה המתווכת על ידי קלאטרין(CME), ואנדוציטוזה ללא תיווך קלאטרין בתיווך קוואולין. אסטרוציטים חסרי CerS2 הציגו קינטיקה איטית יותר של ליגנדים הנכנסים באמצעות CME בעוד שהקינטיקה של אנדוציטוזה בתיווך קוואולין נותרה ללא שינוי. לאחר בחינת המנגנון של CME באסטרוציטים חסרי CerS2 גיליתי כי Hsc70, חלבון מרכזי במנגנון ה- CME האחראי על פירוק כיסוי הקלאטרין, מבוטא הרבה פחות מאשר באסטרוציטים מזן הבר. בנוסף מדדתי את הקישור של Sp1, החלבון האחראי על שעתוק Hsc70, לפרומוטור של Hsc70- ומצאתי שהקישור נמוך בצורה משמעותית יחסית לזן הבר. מעבר לכך, מצאתי כי בשל עלייה משמעותית ברמות הרדיקלים החמצניים החופשיים )ROS( , רמות החלבון Sp1 אשר ידוע כרגיש ל-ROS היו נמוכות בצורה משמעותית. לבסוף, הראיתי שניתן לחקות את השינוי בקינטיקה של ה-CME ע"י הוספה של ROS חיצוני לתאי זן הבר בעוד שהוספה של נוגדי חימצון לתאים חסרי CerS2 מחזירה את את כל התופעות המתוארות לעיל בחזרה למצב הנורמאלי. לפיכך ניתן להסיק כי לפחות חלק מהתופעות המתוארות ניתנות להסבר על ידי עליה בעקה החימצונית של התאים.

בשיתוף פעולה עם חברה למעבדה, איפיינתי את עכברי ה- Cer1/ CerS2 דאבל נוקאאוט. זהו התיעוד הראשון של עכברים אשר חסרים שניים מתוך ששת הגנים ממשפחת ה-CerS . העכברים הללו נולדים וחיים 3-4 שבועות למרות העובדה שהם מייצרים כמות מזערית של ספינגוליפים באורך 18 פחמנים וכן לא מייצרים כלל ספינגוליפידים בעלי שרשראות ארוכות מאד. למרות שעכברי הדאבל נוקאאוט מציגים נורמאלי באיברים החיוניים )כבד, טחול, שריר, מעי וכליה( המוחות של העכברים הללו מציגים דלקת חריפה בתוך 20 יום. לסיכום עכברי דאבל נוקאאוט מציגים גרסא חריפה יותר של סכום הפתולוגיות של הוריהם עכברים חסרי הCerS2- וחסרי ה- .CerS1

בנוסף חקרתי את רגישות היתר של עכברים חסרי הCerS2- לדלקת המעי הגס הנוצרת בתיווך DSS. שאלתי האם מקור הדלקת הוא בפעילות יתר של מערכת החיסון או שמא בתפקוד לקוי של מחסום המעי בעכברים חסרי ה- CerS2. לפיכך יצרתי ארבע קבוצות של עכברי כימרה של מערכת החיסון. עכברי זן בר ועכברים חסרי CerS2 בעלי מערכת חיסון של זן הבר וכן עכברי זן הבר ועכברים חסרי CerS2 בעלי מערכת חיסון חסרת CerS2. עכברים אלו היו חשופים ל-DSS ורגישות היתר שלהם נמדדה באמצעות קצב הירידה במשקל, אורך המעי הגס, מצב הצבירה של הצואה וכן הימצאות דם צואתי. להפתעתי גיליתי כי מערכת החיסון לא משפיעה על רגישות היתר של עכברים חסרי CerS2 ל- DSS. לפיכך אני משער כי הסיבה לרגישות היתר בעכברים חסרי CerS2 נעוצה בליקוי במחסום המעי אשר מאפייניו עדיין לא פוענחו.

בשיתוף פעולה עם מעבדה נוספת, בחנתי את מידת מעורבתה של דימריזציה של CerS בהתפתחות גליובלסטומה. אני הראיתי כי הפחתה גנטית )נוקדאון( של החלבון Bcl2L13 מעלה את הפעילות של CerS2 ושל CerS6 לאחר טיפול עם כורכומין, וכן שיש קשר הפוך בין כמות תעתיק ה-RNA של Bcl2L13 לבין פעילות CerS2 או CerS6 בביופסיות של גליובלסטומה מנבדקים אנושיים.

לסיכום, מצאתי כי ל-CerS בכלל ולCerS2- בפרט יש משמעות בפעילות התקינה של מספר תפקודים בסיסיים וכן שהפחתה של CerS2 )באמצעים גנטיים, כימיים או ביולוגיים( מובילה למספר פתולוגיות כגון עקה חימצונית, דלקת במוח ובמעי הגס וכן לסרטן.

7 3. Introduction 3.1. The lipid composition of eukaryotic membranes A eukaryotic membrane bilayer consists of three major classes of lipids: sterols, glycerolipids and sphingolipds (SLs), which are distinguished by their chemical backbone1 (Fig. 1). The most abundant building block in eukaryotic membranes are glycerolipids2. They are distinguished by a glycerol backbone attached to fatty acids that differ in length and saturation, via an ester bond in the sn-1 and sn-2 positions. A third fatty acid can be attached in the sn-3 position to give rise to a triglyceride or can be esterified to a phosphate and a head group through a phospho-ester bond to generate other glycerolipids such as phosphatidylcholine, phosphatidylserine, phosphatidylethanolamine etc. The different combinations of the fatty acids and different head groups are responsible for an enormous number of glycerolipid species in membranes. Sterols (mainly cholesterol in mammals) are comprised of three cyclohexane rings (A, B and C rings) fused together with a cyclopentane ring (D ring) and an 8-carbon side chain, giving rise to a planar and rigid cholestane structure. The molecule becomes amphipathic with the addition of a hydroxyl group on ring A. Due to their hydrophobic nature and ability to associate with other lipids, sterols reduce the fluidity of the membrane and the rate of lateral diffusion within the membrane3. As is the case for all membrane lipids, SLs are amphipathic molecules having both hydrophobic and hydrophilic regions. SLs contain a single fatty acid varying in chain length, degree of saturation, and degree of hydroxylation, which is usually attached via amide bond to an 18-carbon amino alcohol backbone known as the sphingoid base or long chain base (LCB)4. SLs are distinguished by either the LCB, the fatty acid composition attached to the LCBs, or by the different head groups attached to the primary hydroxyl such as sugars or phosphate5,6. In the past, SLs were considered as inert structural membrane components, however recent studies show that SLs act as bioactive molecules playing a role in a wide variety of cellular processes such as proliferation, apoptosis, cell migration and intercellular interaction4,7.

Fig. 1. Representative lipids of the three major lipids classes. Phosphatidylcholine as a representative of glycerolipids, sphingomyelin as a representative of sphingolipids and cholesterol as a representative of sterols.

3.2. SL biosynthesis pathway The biochemical pathways, and most of the enzymes in the SL synthesis pathway have been previously identified6. A central role in this pathway is played by ceramide, which is the basic building block of all SLs, and is used for the synthesis of more complex SLs, by addition of different head groups. Moreover, ceramide is an important second messenger that plays a pivotal role in regulating vital biological processes such as vesicular transport, generation of endosomes and apoptosis8. The two biochemical pathways of ceramide generation are i, The salvage pathway, by which ceramide is generated by the breakdown of sphingomyelin (SM)9 and ii, the de novo SL biosynthetic pathway (Fig. 2), that occurs on the cytosolic side of the endoplasmic reticulum (ER)10. The pathway begins by condensation of palmitoyl CoA and serine, by serine palmitoyl transferase forming 3-ketosphinganine, which by subsequent reduction generates sphinganine(Sa), via 3-ketosphinganine reductase. This step is followed by the acylation of Sa via sphinganine N-acyl transferase, also known as (dihydro)ceramide synthase (CerS), to form dihydroceramide. Dihydroceramide is reduced to ceramide via dihydroceramide desaturase, which forms a trans double bond at the 4-5 position (Fig. 3)11.

Fig. 2. The SL biosynthesis pathway. SLs and their precursors are presented in purple; the enzymes participating in the pathway are in black; dihydroceramide synthase, which is the main focus of this study, is marked with a box12.

Fig. 3. Schematic representation of ceramide synthesis.

10 3.3. Ceramide Ceramide consists a LCB and a fatty acid. The composition of the fatty acids in ceramide may vary in chain length (typically 14 to 36 carbon atoms) and degree of saturation (usually being saturated or mono-unsaturated)13 and may contain a hydroxyl group on C2 position. Furthermore, the composition of the LCB varies as well. The predominant LCB in mammals is sphingosine, which is formed by desaturation of Sa and contains 18 carbons with a double bond in the 4-5 position, whereas Sa, the product of de novo synthesis, lacks this double bond. Other LCBs have different numbers of carbons (14-22), hydroxyl groups, more double bonds and branching with a methyl group on carbon 914. The use of recently developed mass spectrometry techniques has enabled characterization of the sphingolipidome, deciphering many new LCB structures, which are sometimes rare and limited to certain organisms or cell types15,16. As ceramide differ in its composition of either the LCB or fatty acid, the huge combinatorial possibilities of these two molecules gives rise to a vast number of ceramide species. As ceramide is the precursor for other complex SLs, the combinatorial property of ceramide, combined with the different head groups, produces an immense number of existing SLs. Additionally, due to their biophysical properties, ceramide molecules tend to segregate within the membrane together with saturated glycerophospholipids and cholesterol, forming ceramide-enriched domains17,18. These domains assemble platforms that enable the amplification of signaling pathways by reorganizing the membrane and by clustering specific signaling components, such as receptors and downstream effectors. These events ensure efficient transmission of extracellular signals19. Ceramide is involved in various cellular events; however, many questions remain open about the mechanisms by which the target proteins interact with ceramide and which ceramide species are involved in modulating the different signaling pathways.

3.4. Ceramide as a structural unit of complex sphingolipids Several enzymes in the SL metabolic pathway use ceramide as a substrate to generate more complex SLs. Ceramide kinase phosphorylates ceramide to generate ceramide-1-phosphate either in the plasma membrane (PM), the ER or the Golgi apparatus20,21. Ceramide is delivered from the ER to the Golgi apparatus by vesicular transport11, where it is glycosylated by glucosylceramide synthase at the cytosolic surface of the Golgi to produce glucosylceramide (GlcCer) (Fig. 2); it is also metabolized by galactosylceramide synthase at the lumenal side of the ER to produce galactosylceramide (GalCer). Both GalCer and GlcCer

11 can be further metabolized to complex SLs by addition of different sugar molecules to their head group6. Additionally, SM synthase forms SM by transferring a phosphorylcholine head group from phosphatidylcholine to ceramide; this reaction occurs at the lumenal surface of the Golgi. Ceramide is delivered to the Golgi apparatus by a specialized ceramide transfer protein, CERT, for SM synthesis22. SM is very abundant in most cells and its hydrolysis by the different sphingomyelinases (SMases) leads to a rapid elevation in ceramide levels. In the catabolic pathway, ceramide is deacylated to sphingosine and a free fatty acid by three different ceramidases; neutral ceramidase in the PM, acid ceramidase in the lysosome or alkaline ceramidase in the ER or the Golgi23-25. Subsequently, sphingosine can become a substrate for CerS in the salvage pathway or phosphorylated by sphingosine kinase generating sphingosine-1-phosphate (S1P)26. The ultimate step in SL degradation is catalyzed by sphingosine-1-phosphate lyase, which breaks down S1P into ethanolamine phosphate and hexadecanal27.

3.5. The (Dihydro)ceramide synthases (CerS) family The first evidence for the molecular identification of CerS was obtained from yeast when two proteins, Lag1 and Lac1, were shown to be necessary for synthesis of the very long chain C26-ceramides found in yeast28,29. Purification of these two proteins led to identification of another subunit, Lip1, which is a single pass trans-membrane protein located in the ER, and is essential for the enzymatic activity of yeast CerS in vivo and in vitro30. However, no Lip1 homolog was found in a bioinformatics analysis in mammalian databases. Additionally, overexpression of CerS5 in a triple mutant yeast strain Δlag1Δlac1ΔLip131 resulted in the synthesis of C16-ceramides, and survival of the mutant yeast cells in the absence of their endogenous C26-ceramides. Moreover, the fact that not all yeast strains contain the Lip1 protein (Ben-Dor et. al. unpublished data) along with biochemical studies from our lab that show that mammalian CerS5 is a bona fide ceramide synthase, that does not require any additional subunits for its activity32, suggest that the function of Lip1 with Lag1/Lac1 is not prerequisite for ceramide synthesis. Lag1 was originally identified in a screen for proteins affecting life span of Saccharomyces cerevisia33 and deficiency of the LAG1 gene led to an increase in the yeast life span. Bioinformatics search for homologs of the Lag1 protein in a human genomic database revealed six Lag1 human homologues, named LASS (Longevity Assurance) genes, due to their similarities to the yeast genes34. About a decade ago, the LASS proteins were renamed as ‘ceramide synthases’ (CerS1-6), to reflect their genuine

12 function in cells35, and in the past few years all mammalian CerS family members were cloned and knocked out individually in transgenic mice36-41. The first evidence for identification of mammalian CerS was obtained when CerS1 was over-expressed in cells, leading to an increase in C18-ceramide synthesis5. Later, it was demonstrated that overexpression of each CerS led to an increase in a specific subset of ceramides containing a unique fatty acid composition. Thus, CerS1 uses mostly C18-CoA5, CerS4 uses C18- and C20-CoAs42, CerS5 and CerS6 use mostly C16-CoA42,43, and CerS3 uses very long chain acyl CoAs (C26 and higher)44. CerS2 can utilize a wider range of fatty acyl CoAs, from C20 to C26, but does not use C16- or C18-CoAs45 (Table 1). The reason for multiple CerS genes in yeast, mammals and other species is not known but ceramides containing specific fatty acids have important roles in cell physiology35. The different ceramide molecules have distinct biophysical properties, enabling specialization and involvement in many cellular events such as homeostasis and signaling. Thus, the regulation of the six CerS, which determines the ceramide acyl chain composition produced in specific cells, is extremely important. Not surprisingly, it was demonstrated that the properties and regulation of the six CerS is different46. The six human CerS paralogs are located on different chromosomes, except for CerS1 and CerS4 which are located on the same chromosome, but with a distant chromosomal location (Table 1).

Table 1. Ceramide Synthases. Acyl-CoA specificity, chromosomal localization and main phenotypes in KO mice12.

3.6. CerS2 CerS2 is the most ubiquitously expressed of all CerS, it has the broadest distribution and is the only CerS to display genomic features characteristic of a ‘housekeeping’ gene42,45. Furthermore, CerS2 expression levels are the highest in the liver and kidney45. In addition, our lab demonstrated that CerS2 is inhibited in vitro by S1P, another bioactive SL, via a noncompetitive mechanism45. CerS2 is inhibited via interaction of S1P with two residues that are

13 part of S1P receptor-like motif found specifically in CerS245. All CerS except CerS1 are phosphorylated at their C-terminus region and the phosphorylation state of CerS regulates its activity47,48. CerS2 activity is also controlled by dimerization. It has been shown that CerS2 co-immunoprecipitates with CerS5 and CerS6 and that dimerization with an active CerS5 increases CerS2 activity. Moreover, acute treatment with curcumin facilitates CerS2 (and CerS5) dimerization49. Furthermore, CerS2 (as are all other CerS except CerS1) is glycosylated43 but the glycosylation state of CerS does not influence its activity (Zelnik et. al. unpublished data). Nevertheless, CerS glycosylation might have implications for understanding the topology of CerS in the context of the ER.

3.7. CerS transgenic mice In vivo models of CerS deficiencies provide major information about the physiological roles of ceramide species with specific acyl chain lengths and greatly enhances our understanding of SL biology especially with regard to their role as bioactive molecules. In the past few years, the study of CerS biology has progressed rapidly, largely due to the generation of CerS null mice. In all cases, the resulting phenotypes appear to be directly related to the loss of the specific ceramide species and down-stream SLs. Moreover, the possibility that some effects are due to loss of interactions among the CerS proteins themselves cannot be excluded. The lack of C18-SLs in the CerS1 KO mice causes progressive ataxia and loss of cerebellar Purkinje cells and lipofuscin accumulation50,36, while ablation of CerS3, which makes ultra- long chain SLs in mice, is lethal, presumably due to skin permeability37. A CerS4 KO mouse demonstrated the importance of CerS4 for synthesis of functionally normal sebum and hair follicle homeostasis38. Interestingly, CerS5 KO mice showed improved glucose maintenance, reduced adipose tissue inflammation after high fat diet and overall better systemic health39, while CerS6 KO mice have habituation disabilities and impaired moto-neuronal function40 (Table 1). Our lab has generated a CerS2 null mouse from embryonic stem cells harboring a gene trap51 retroviral vector insertion in the first intron of the CerS2 gene52. This mouse is devoid of SLs containing very-long (C22-C24) acyl chains as predicted from the acyl chain specificity of CerS241,45. CerS2 null mice exhibit a myriad of phenotypes. They do not survive beyond 16 months, they are born smaller and weigh 20-30% less than their wild type (WT) littermates41. CerS2 null mice develop chronic and progressive liver abnormalities such as nodules of regeneration, high levels of liver enzymes (ALT, AST, ALP), hypoglycemia and high turnover of hepatocytes with increased proliferation and apoptosis41. In addition,

14 CerS2 null mice show some neurological abnormalities and adrenal pathologies53,54. High throughput analysis of RNA expression in the liver revealed up-regulation of genes associated with cell cycle regulation, protein transport, cell-cell interactions and apoptosis, and down-regulation of genes associated with lipid and steroid metabolism, adipocyte signaling and amino acid metabolism. Several papers published by our lab describe the role of very long acyl chain SLs in liver homeostasis41,55-57 (Table 1).

3.8. Endocytosis – more than one way to get inside a cell Endocytosis is a basic cellular process that is used by cells to internalize a variety of molecules. In general, any endocytic pathway requires a mechanism of selection at the cell surface. Next, the PM must be induced to bud and pinch off. Finally, mechanisms for targeting these budding vesicles to their destination and for their subsequent fusion to this target membrane are essential. Thus far, a number of entry pathways into cells have been identified. They vary in the cargoes they transport and in the protein machinery that facilitates the endocytosis process. Clathrin-mediated endocytosis is the main and the best characterized mode by which cells internalize surface cargo proteins. In this process, the cargo molecules bind to complementary transmembrane receptors, at sites of the plasma membrane where they are destined to be internalized. These sites are enriched with PI4,5P2, which serve as nucleation points for the nascent clathrin coated pits (CCPs). A large variety of adaptor protein (AP-2, Dub2, ARH etc.) and accessory proteins are responsible for the deformation of the membrane and subsequent accumulation of the receptors into the maturing pit. At this stage clathrin is recruited and its polymerization stabilizes the deformation of the attached membrane as a curved lattice, helping to bring the membranes surrounding the neck into close apposition. The membrane scission protein dynamin is a large GTPase, which upon GTP hydrolysis, mediates the fission of the vesicle from the plasma membrane58, irreversibly releasing the clathrin coated vesicle (CCV) into the interior of the cell. The released vesicle is then rapidly shed of its clathrin coat in an ATP-dependent reaction carried out by the molecular chaperone, Hsc7059-61. This reaction also requires a co-chaperone, either auxilin-1, which is expressed selectively in the brain, or the ubiquitously expressed auxilin-2 (aka GAK)62. However, not all molecules undergo endocytosis using clathrin. Caveolae-mediated endocytosis is the major clathrin-independent endocytic pathway63. Caveolae are small ~80nm flask-shaped PM invaginations marked with the presence of one of the caveolin family members. Recent studies revealed some additional clathrin- and caveolin- independent

15 endocytic pathways, namely the RhoA pathway, the CDC42 pathway and the ARF6 pathway. These pathways are yet to be fully understood64. Although all endocytosis pathways share the basic aforementioned principles, each pathway has a unique set of attributes ranging from type of cargo, vesicle size, primary carriers, structural proteins and acceptor compartments (Fig.4)63. Using the CerS2 null mouse, our lab has published a manuscript that shows impaired receptor internalization65 upon depletion of very long acyl chain SLs (VLC-SLs). However, to date the role of VLC- SLs in intra- or extra-cellular trafficking has not been explored in any detail.

Fig. 4. Pathways of entry into cells.

Large particles can be taken up by phagocytosis, whereas fluid uptake occurs by macropinocytosis. Compared with the other endocytic pathways, the size of the vesicles formed by phagocytosis and macropinocytosis is much larger. Numerous cargoes can be endocytosed by mechanisms that are independent of the coat protein, clathrin, and the fission GTPase, dynamin. Most internalized cargoes are delivered to the early endosome via vesicular (clathrin- or caveolin-coated vesicles) or tubular intermediates (known as clathrin- and dynamin-independent carriers (CLICs)) that are derived from the plasma membrane. Some pathways may first traffic to intermediate compartments, such as the caveosome or glycosylphosphatidylinositol-anchored protein-enriched early endosomal compartments (GEEC), en route to the early endosome64.

16 3.9. CerS in health and disease Ceramides with defined acyl chain lengths are involved in a number of human diseases, yet little is known about the specific roles of CerS in human disease. Some of the pathologies that have been observed were due to specific mutations in the coding sequence such as the Trp15 mutation in CerS366 that causes Autosomal Recessive Congenital Ichthyosis and the Cers2 Glu115 mutation that is associated with increased rate of albuminuria in diabetic patients67 and with rhegmatogenous retinal detachment68. In other pathologies, changes in CerS activity or expression levels have been linked as an indirect down-stream response to other metabolic alterations. For instance, hepatic insulin resistance was shown to be mediated by the increased activity of CerS1 and CerS669 and also in CerS2 null mice56. Another example is the elevated levels of CerS6 in astrocytes and monocytes in the spinal cord of the experimental autoimmune encephalomyelitis model of multiple sclerosis70. Additionally, it was shown that regulation of CerS6 influenced the release of nitric oxide and TNFα, suggesting a role for CerS6 in the progression of multiple sclerosis71. Finally, CerS were implicated in the pathogenesis of several types of cancer. CerS1 and C18-ceramide levels were shown to be important in the pathogenesis of head and neck squamous cell carcinoma. Cancerous cells displayed lower levels of C18-ceramide whereas transfection with CerS1 and the subsequent elevation in C18-ceramide levels inhibited cell growth72 and enhanced the growth-inhibitory effect of doxorubicin by caspase-3 activation73. In breast cancer, CerS2 and CerS6 expression levels were elevated74 as well as their products C16-, C24:0- and C24:1- ceramides75. Moreover, elevated CerS2 levels increased cell proliferation76, while low levels of CerS2 were associated with poor prognosis in breast cancer patients77. Finally, CerS6 expression was implicated in the development of colon cancer by triggering tumor necrosis factor-related apoptosis-inducing ligand sensitivity78. Over expression of CerS6 enhanced CD95 activation and subsequent cell death while CerS6 knockdown suppressed CD95 activation79, supporting the hypothesis of negative correlation between CerS6 and colon cancer. Despite the accumulating evidence, further research is needed in order to unveil the involvement of CerS in pathologies as well as normal cellular processes.

17 4. Research objectives The working hypothesis was that since CerS2 has characteristics of a house keeping gene, it will have an effect on multiple cellular processes and pathologies. Thus, by using genetic and chemical manipulations of CerS2 I will be able to identify and gain better understanding of the role of CerS2 and its products, very long chain SLs, in these processes and pathologies. The following points were addressed:

1. Characterization of primary cultured astrocytes from CerS2 null mice

2. Analyzing the effect of altered SL composition on endocytosis

3. Characterizing CerS1/CerS2 double knockout mice

4. Exploring the hypersensitivity of CerS2 null mice to DSS induced colitis

5. The involvement of CerS dimerization in glioblastoma

18 5. Materials and methods

5.1. Mouse models CerS2 null mice were generated as described41 and CerS1 null mice were obtained from Dr. Zhao at Jackson laboratories50. All mice were maintained in the Experimental Animal Center of the Weizmann Institute. The Institutional Animal Care and Use Committee of the Weizmann Institute of Science approved all experimental protocols.

5.2. Astrocyte culture Cortical astrocytes from 2-4 day old WT and CerS2 null mice41 were isolated as described80. Astrocytes were maintained at 95% confluency for 7-10 days, removed from the dishes by gentle shaking (2 h, 200 rpm on an orbital shaker), replated and grown to 95% confluency for up to 23 days. Cells were cultured in Dulbecco’s modified eagle’s medium (DMEM, GibcoTM) containing 10% fetal calf serum (GibcoTM), 0.1% Glutamax (GibcoTM), 0.1% Mito+ serum extender (Corning inc.), and 2% penicillin/streptomycin.

5.3. Immunohistochemistry Astrocytes were cultured on 12 or 13 mm glass coverslips to a confluence of ~70%. Cells were fixed by incubating with 3% paraformaldehyde in phosphate buffered saline (PBS) at room temperature for 15 min, rinsed three times with PBS for 4 min and blocked with 4% bovine serum albumin for 90 min. Cells were then either co-incubated for 1 h with anti- GFAP (for astrocytes, 1:1,000, Bio-Rad, Hercules, California, USA) and anti-CD68 (for microglia, 1:1,000) antibodies, or with an anti-clathrin heavy chain (1:50, Cell Signalliing, Bioke, Leiden, the Netherlands) antibody. Cells were washed three times with PBS for 4 min and incubated with a secondary fluorescently-labeled antibody (anti-rat, anti-mouse or anti rabbit respectively; 1:250, Jackson Immunoresearch Labs, West Grove, Pennsylvania, USA) for 60 min. Cells were rinsed three times with PBS for 4 min, incubated for 10 min with Hoechst 33342 (1:5,000, Molecular probes, Eugene, Oregon, USA), and then rinsed three times with PBS for 4 min. Glass coverslips were mounted on slides using Gel Mount (Molecular Probes, Eugene, Oregon, USA) and analyzed by fluorescence microscopy using an Olympus IX 81 FluoView 1000 microscope (Olympus, Tokyo, Japan).

19 5.4. RNA extraction and polymerase chain reaction Total RNA was isolated using the RNeasy mini kit (Qiagen, Hilden, Germany) according to manufacturer’s instructions. cDNA synthesis was performed using a Verso cDNA kit (Thermo Scientific, Waltham, Massachusetts, USA). qPCR was performed as described81. The relative amounts of mRNA were calculated from the cyclic threshold values using hypoxanthine guanine phosphoribosyl transferase (HPRT) or TATA binding protein (TBP) for normalization. Primers are listed in Table 2.

Table 2. Primers used in the thesis. F, forward; R, reverse.

Gene Primers CerS1 TaqMan ™(Mm00433562_m1) CerS2 TaqMan ™(Mm00504086_m1) CerS3 TaqMan ™(Mm03990709_m1) CerS4 TaqMan ™(Mm01212479_m1) CerS5 TaqMan ™(Mm00510996_g1) CerS6 TaqMan ™(Mm00556165_m1) HPRT TaqMan ™(Mm00446968_m1) Auxilin-1 (dnaj6C) TaqMan ™(Mm01265598_m1) Hsc70 F: 5’-CTGCTGCTATTGCTTACGGC-3’ R: 5’-TCAAAAGTGCCACCTCCCAA-3’ Fatty acid synthase F: 5’-CAAGTGTCCACCAACAAGCG-3’ R: 5’-GGAGCGCAGGATAGACTCAC-3’ Glycolipid transfer protein F: 5’-CTGCCGCCCTTCTTTGATTG-3’ R: 5’-AGGGTCTTGAACTTGGCTGG-3’ GFAP F: 5’-TAGTCCAACCCGTTCCTCCA-3’ R: 5’-CCAGTTGTCGACTAGGACCG-3’ Toll like receptor 4 F: 5’-ACCTGGCTGGTTTACACGTC-3’ R: 5’-CTGCCAGAGACATTGCAGAA-3’ TBP F: 5’-TGCTGTTGGTGATTGTTGGT-3’ R: 5’-CTGGCTTGTGTGGGAAAGAT-3’ Hsc70 ChIP -68 F: 5’-ACCTAGGCGAGCGTTCTG-3’ R: 5’-ACGACGAGACCACACAAATG-3’ Hsc70 ChIP -174 F: 5’-CCGAACGCTGCTCTCATTG-3’ R: 5’-AACGCTCGCCTAGGTCCC-3’

20 5.5. CerS assays Cells were homogenized in 20 mM HEPES-KOH, pH 7.2, 25 mM KCl, 250 mM sucrose and

2 mM MgCl2 containing a protease inhibitor cocktail (Sigma, St. Louise, Missouri, USA). Protein content was determined using the BCA reagent (Pierce, Rockford, Illinois, USA). Homogenates were incubated with 15 μM NBD-sphinganine82 (Avanti Polar Lipids, Alabaster, Alabama, USA), 20 μM defatted bovine serum albumin (Sigma, St. Louis, Missouri, USA), and 50 μM C16–CoA (for CerS5/6), C18- CoA (for CerS1/4), C20-CoA (for CerS4) and C22-CoA or C24-CoA (for CerS2) (Avanti Polar Lipids, Alabaster, Alabama, USA) for 10-40 min at 37°C. Lipids were extracted and separated by thin layer chromatography using chloroform/methanol/2M NH4OH (40:10:1, v/v/v) as the developing solvent. NBD-labeled lipids were visualized using a Typhoon 9410 variable mode imager and quantified by ImageQuantTL (GE Healthcare, Chalfont St Giles, UK).

5.6. Lipid analysis Astrocytes from WT and CerS2 null mice or brain homogenates from WT, CerS2 null, CerS1 null and CerS1/CerS2 double knock out mice were sent to the lab of Professor Alfred Merrill (Atlanta, USA) and SL levels analyzed by LC-ESI MS/MS using a PE-Sciex API 3000 triple quadrupole mass spectrometer and an ABI 4000 quadrupole-linear ion trap mass spectrometer 83,84.

5.7. Proliferation and cell death Astrocytes were counted and plated at the indicated amounts in 96-well plates. After 3 days the media was collected and analyzed for cell death using CytoTox-ONE™ Homogeneous Membrane Integrity Assay (Promega, Madison, WN, USA). The cells were analyzed in parallel for cell proliferation using XTT cell proliferation kit (Biological Industries, Beit Ha- emek, Israel).

5.8. Western blotting Western blotting85 was performed with the following antibodies, diluted in PBS and 0.1% Tween 20 and 0.5% sodium azide: rabbit anti-phospho S6K (1:1000), rabbit anti-total S6K (1:1000), mouse anti-phospho Akt, rabbit anti-total Akt, rabbit anti-clathrin heavy chain (1:1000) and rabbit anti-cleaved caspase 3 (1:500, Cell Signalliing, Bioke, Leiden, the Netherlands), rabbit anti-LDL receptor (1:1,000), mouse anti-Hsc70 (1:1,000), rabbit anti- Sp1 (1:1,000), rabbit anti-3-nitrotyrosine (1:1,000) obtained from Abcam (Cambridge, UK), rabbit anti-Tf receptor (1:1,000 Santa Cruz, Huissen, The Netherlands), rat anti-Mac2

21 (1:1000, Cedarlane labs, Burlington, Canada), rabbit anti-GFAP (1:1000, Bio-Rad, Hercules,CA,USA), rabbit anti-calbndin (1:1000, Chemicon, Japan) and mouse anti-GAPDH (1:5,000, Milipore, Billerica, USA). Densitometry was performed using ImageQuant software (Amersham Biosciences, Little Chalfont, UK).

5.9. Uptake of fluorescently-labeled ligands WT and CerS2 null astrocytes were cultured in 96-well plates and grown overnight. Cells were incubated for various times with fluorescently labeled ligands (Dil-LDL (1:100); BODIPY-LacCer (1:100); Life Technologies, Carlsbad, California, USA; Cy5-Tf (1:100), Jackson Immunoresearch Labs, West Grove, Pennsylvania, USA). Cells were washed three times with PBS to remove unbound ligand, and incubated for 4 min with 500 mM NaCl in 200 mM acetic acid, pH 3.5 to remove any Dil-LDL or Cy5-Tf bound to the cell surface86. For BODIPY-LacCer, back exchange was performed by incubating cells with 5% defatted bovine serum albumin for 10 min. In some cases, cells were incubated with 100 μM H2O2 and 100 μM NAC for 4 h prior to addition of the fluorescent ligand. Internalized fluorescent ligand was quantified using a plate reader at the following wavelengths: Dil-LDL; excitation 528 nm, emission, 590 nm; BODIPY-LacCer, excitation 485 nm, emission, 528nm; Cy5- transferrin, excitation 630 nm, emission 680 nm.

5.10. Electron microscopy Cultured astrocytes were washed 3 times in Kosnovsky fixative (2% glutaraldehyde, 3% PFA, 3% sucrose in 0.1M cacodylate buffer) and fixed in the same fixative for 150 min at room temperature, then washed 4 times in 0.1M cacodylate buffer and kept at 4°C until further use. Osmification was made in 1% OsO4, 0.5% K2Cr2O7, 0.5% K4[Fe(CN)6]3, 0.1M cacodylate buffer for 60 min at room temperature. Cells were then washed with 0.1M cacodylate buffer four times for 5 min and three times with DDW for 2 min. Cells were impregnated with 2% uranyl acetate ION-DDW for 30 min. Following dehydration, EPON was introduced gradually into the cultures using increasing concentrations (30%, 50%, 70%, for 120 min each and 100% overnight and additional 100% three times for 120 min the next day). Cells were then transformed into molds in EPON and incubated at 60°C for 3 days. The EPON blocks were then roughly cut into small pieces and re-assembled. Finally, 70-100nm ultra thin slices were cut from each block, mounted onto grids and analyzed in a CM-12 Philips electron microscope; images were captured with BioCam CCD camera.

22 5.11. Transcription factor binding site analysis Transcription factor binding site analysis was performed using the Genomatix Genome Analyzer (GGA) MatInspector program87, Matrix Library Version 9.0, searching with the V$SP1F matrix.

5.12. Chromatin immune precipitation ChIP was performed as described88. Cells were fixed in 0.1 volumes of formaldehyde (11%) followed by 0.06 volumes of 2.5 M glycine. Immunoprecipitation was performed using 3 μg of antibody in 0.1% Brij-35. DNA was purified using a QIAGEN PCR purification kit (QIAGEN Industries, Venlo, Netherlands) followed by real-time qRT-PCR.

5.13. Reactive oxygen species ROS were measured using chloro-methyl 6-carboxy-2’,7’-dichlorodihydrofluorescein diacetate (cmH2DCF-DA) (Invitrogen, Carlsbad, California, USA). WT and CerS2 null astrocytes were incubated with 100 μM cmH2DCF-DA or PBS for 30 min at 37°C and then washed 3 times with PBS. Conversion of cmH2DCF-DA to cmDCF was determined using an Eclipse iCyt flow cytometer equipped with 488 nm solid state air cooled lasers, with 25 mW on the flow cell and with standard filter set-up; cmDCF was measured in the green channel with an excitation of 488 nm and emission of 525 ± 50 nm.

5.14. Mitochondrial complex IV activity Activities of mitochondrial respiratory chain complex IV and citrate synthase, a mitochondrial marker, were determined in cell homogenates as previously described89. Measurements were performed using a doublebeam spectrophotometer (UVIKON 930, Secomam, France).

5.15. Confocal microscopy Confocal microscopy was performed using an Olympus IX 81 FluoView 1000 microscope and a UPLSAPO 60x objective. Images were processed and analyzed using FV-1000 software (Olympus, Tokyo, Japan).

23 5.16. BM chimeras Mouse BM chimeras were created from WT/CerS2 null hosts, which were transplanted with BM from WT/CerS2 null donors. To this end, WT/CerS2 null mice were sub-lethally irradiated with 1 dose of 10Gy and their immune system rescued via BM transplantation from

WT/CerS2 null mice, respectively. Donors were killed by CO2 asphyxiation, and the femur and tibia dissected for BM isolation as described previously90. Cells were re-suspended at a concentration of 5 × 106 BM cells/150 μl; this was administered via the lateral tail vein to recipient mice 3–4 h after irradiation.

5.17. SL turnover HEK cells were cultured to 90% confluence in 6-well plates. Cells were pulsed with 20 μM stable isotope-labeled fatty acid (either d916:0, d9C18:0, d4C24:0 or d7C24:1) in defatted BSAcomplex for 12h. Cells were chased with media containing 20 μM unlabeled fatty acid for 0, 1 ,3, 6, 9, 12 and 24 h. Cells were collected, sonicated, lyophilized and sent for MS analysis using sciex QTRAP5500 at Zora Bioscience in Finland.

24 6. Results 6.1. Characterization of primary cultured astrocytes from a CerS2 null mouse

The first challenge towards the characterization of primary astrocytes was to isolate them from other cell types (e.g. neurons, microglia). Using vigorous mechanical shaking and taking advantage of the adherence properties of astrocytes, I successfully purified and cultured astrocytes (Fig. 5A). Next, I determined the expression and activity levels of all CerS. As expected, CerS2 expression levels in CerS2 null astrocytes were under the limit of detection (Fig. 5B). I did not find elevation of any other CerS mRNAs. Using a method developed in our lab10,82 I quantified CerS activity using several acyl-CoAs (Fig. 5C). Each acyl-CoA corresponds specifically to a distinct CerS activity. C16-CoA corresponds to CerS5 and CerS6 activity, C18-CoA corresponds to CerS1 activity, C20-C22-CoAs correspond to CerS4 activity and C22- C24:1-CoAs correspond to CerS2 activity. Although some C22-CoA activity was detected, we attribute this finding to the ability of CerS4 to N-acylate sphinganine with C22-CoA. The lack of C24:1-CoA activity in CerS2 null astrocytes supports this statement.

Fig. 5. Isolation of primary CerS2 null astrocytes (A) Astrocytes were purified and co-stained with anti-GFAP (green) and anti-CD68 antibodies (red). Cell nuclei were stained with Hoechst 33348 (blue). (B) mRNA levels of all CerS genes. Values are means ± s.e.m, N≥3. ***, P<0.001. (C) CerS activity towards different acyl-CoAs (C16 - C24). Values are means ± s.e.m, N≥4 ***, P<0.001.

25 ESI-MS/MS analysis, performed by Dr. Al Merrill, showed the same SL pattern as CerS2 null liver41. The KO cells contained only trace amounts of very long acyl chain ceramides compared with WT (Fig. 6A). Thus, levels of C22-, C24:0-, and C24:1-SLs were reduced by 2- ,4- ,42-fold respectively in the CerS2 null astrocytes. Levels of C26:1- and C26:0-SLs were relatively low in the WT, and only a small but significant reduction was observed in CerS2 null astrocytes. As described previously, levels of LC-SLs (i.e. C14-, C16-, C18:0- and C20- ceramide) were increased (Fig. 6A), and this increase compensated for the loss of C22– C24-ceramides such that total ceramide levels (i.e. the sum of individual ceramide species containing all acyl chains) were essentially unaltered (Fig. 6B). Thus, CerS2 null mice are not depleted in total ceramide levels but rather show a change in their acyl chain composition.

Fig. 6. Sphingolipid composition of CerS2 null astrocytes Mass spectrometry of (A) ceramide, (B) hexosylceramide and (C) SM species. Values are means ± s.e.m, N=3 for CerS2 null, N=2 for WT. *, P<0.05; **, P<0.01; ***, P<0.001. (D) Total levels of ceramide, hexosylceramide and SM were assessed by mass spectrometry. Values are means ± s.e.m, N=3 for CerS2, N=2 for WT. (E) Summary of the results presented in A-D. Average values are in pmol/mg protein.

(AU)

death

fluorescence Mitochondrialmarker Cell

Cells plated Cells plated Fig. 7. Proliferation and cell death of CerS2 null astrocytes CerS2 null and WT cells were plated at the indicated amounts in 96-well plates and were allowed to proliferate for 72 h. (A) Cell proliferation was measured using XTT (a mitochondrial marker equivalent to cell number) and (B) cell death was measured using LDH (a marker for cell death). Values are average of 3 separate experiments ± SEM. *, P<0.05; ***, P<0.001

I have noticed that CerS2 null astrocytes grew more slowly than WT cells. Therefore, I measured proliferation and cell death using XTT and LDH respectively. It is clear that CerS2 null astrocytes do not die more rapidly than WT cells, rather they proliferate slightly more slowly when they reach high confluence (Fig.7). Ceramide in general91 and C16-ceramide56 in particular, was shown to inhibit Akt phosphorylation. Since the phosphorylation state of Akt can cause, among other things, changes in cellular proliferation92, I measured the levels of total and phospho-Akt in CerS2 null astrocytes. CerS2 null astrocytes exhibit reduced phosphorylation of Akt as well as its down stream target S6 kinase (Fig. 8) which is in line with a previous study from our lab showing that CerS2 null mouse livers exhibit reduced levels of phosphorylated Akt56. This effect might be due to elevated levels of C16-ceramide as overexpression of CerS5 in hepatocytes showed similar reduction in Akt phosphorylation56 upon insulin stimulation.

Fig. 8. Akt signaling in CerS2 null astrocytes Western blot of phosphorylated S6 kinase, total S6 kinase, phosphorylated Akt and total Akt in WT and CerS2 null astrocytes. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a loading control. Results are from a typical experiment repeated at least 3 times with similar results.

27 6.2. Analyzing the effect of altered SL composition on endocytic pathways

For several years, our lab has generated results regarding trafficking perturbations due to lack of VLC-SLs, such as inhibition of tumor necrosis factor receptor 1 (TNFR1) internalization upon LPS treatment of hepatocytes65. To examine this issue in astrocytes, I used fluorescently labeled LDL to investigate the internalization pathway in CerS2 null astrocytes. I found that CerS2 null astrocytes uptake LDL to a lower extent than their WT counterparts as early as 5 min of uptake (Fig. 9A). LDL is a known cargo of clathrin-mediated endocytosis93,94, and I next determined whether other endocytic pathways have the same phenotype as CME. When using fluorescently labeled transferrin (Tf), another well-documented CME cargo95,96, I observed slower uptake kinetics similar to LDL (Fig. 9B,D). In contrast, fluorescently- labeled lactosylceramide (LacCer), a cargo of the clathrin-independent, caveolin-mediated endocytic pathway97, showed no difference in endocytosis between wild type and CerS2 null astrocytes (Fig. 9C).

Fig. 9. Kinetics of uptake of different endocytic ligands in CerS2 null astrocytes

Cells were incubated with (A) fluorescently labeled LDL (dil-LDL), (B) transferrin (Cy5- transferrin) or (C) LacCer (BODIPY-LacCer) for the indicated times. Values are means ± s.e.m, N=3. *, P<0.05; **, P<0.01. (D) Kinetic parameters of the uptake of fluorescently labeled ligands as calculated by non-linear regression using Prism.

28 Since LDL and transferrin receptor levels were unaltered (Fig. 10A), I concluded that CerS2 deficiency specifically modulates clathrin-mediated endocytosis. The effect of CerS2 deficiency on CME mechanism was corroborated by the 3.5-fold elevation of clathrin heavy chain (CHC), a hallmark constituent of CME and by the presence of clathrin heavy chain aggregates visualized by light microscopy (Fig. 10A, B). Nevertheless, electron microscopy revealed no differences in shape or size of clathrin- coated pits and vesicles (Fig. 10C).

Fig. 10. Characterization of clathrin heavy chain and coated pits. (A) Western blot of clathrin heavy chain, LDL receptor and Tf receptor. GAPDH was used as a loading control. Results are from a typical experiment repeated at least 3 times with similar results. Values are means ± s.e.m, **, P<0.01. (B) The distribution of clathrin heavy chain using confocal microscopy. Clathrin heavy chain staining (red), nuclei (blue) white arrows indicate typical clathrin heavy chain aggregates in CerS2 null astrocytes. The image is representative of three independent experiments. Scale bar, 10 μ m. (C) The morphology of clathrin coated pits/vesicles using electron microscopy. The image is representative of two independent experiments. Scale bar = 100nm.

29 Yim et. al98 showed that auxilin KO mice exhibit reduced rates of endocytosis and elevated clathrin levels. As previously mentioned, auxilin and heat shock chaperon70 (Hsc70) use ATP to shed the clathrin coat from the clathrin-coated vesicle. Therefore, I measured auxilin and Hsc70 mRNA levels using qPCR. Although auxilin mRNA levels remained unchanged, Hsc70 mRNA levels and protein levels were significantly reduced in CerS2 null astrocytes (Fig. 11A, B).

Fig. 11. The clathrin coat shedding mechanism (A) mRNA levels of auxilin-1 and Hsc70 in WT and CerS2 null astrocytes. N=2, values are means ± s.e.m. *, P<0.05. (B) Western blot of Hsc70 in WT and CerS2 null astrocytes. GAPDH was used as a loading control. Results are from a typical experiment repeated at least 4 times with similar results. Values are means ± s.e.m, **, P<0.01.

To assess whether lower levels of Hsc70 are critical for CME, I used KNK437, an inhibitor that reduces the transcription of some of the heat shock proteins, including Hsc7099. A 4 h treatment with KNK437 reduced the transcription of Hsc70 in WT cells to the same levels as CerS2 null cells, whereas levels of Hsc70 in CerS2 null astrocytes did not respond to KNK437 treatment (Fig. 12A). Next I measured the effect of KNK437 treatment on the uptake of fluorescently labeled LDL. KNK437 treatment reduced the uptake of LDL in WT astrocytes significantly to the same extent as CerS2 null cells (Fig. 12B).

Fig. 12. Effects of KNK437 on Hsc70 and LDL uptake (A) mRNA levels of Hsc70 in WT and CerS2 null astrocytes with and without KNK437 treatment. N=3, values are means ± s.e.m. *, P<0.05; **, P<0.01. (B) WT and CerS2 null Cells were incubated with fluorescently labeled LDL (dil-LDL) for the indicated times with or without KNK437 (100μM). Values are means ± s.e.m, N=3. *, P<0.05; **, P<0.01.

Since Hsc70 levels were reduced, I sought a transcription factor that can modulate Hsc70 transcription. The heat shock factor-1 (HSF-1) and specificity protein-1 (Sp1) transcription factors are known to regulate Hsc70 levels100,101. Western blot analysis revealed a significant reduction of Sp1 levels in CerS2 null astrocytes, whereas HSF- 1 protein levels remained unchanged (Fig. 13A). The reduction in Sp1 levels was not due to changes in its transcription (Fig. 13B). Chromatin immuneprecipitation (ChIP) confirmed a previous observation100 that the motif in position -68 is the binding site of Sp1 (Fig. 13C) whereas in position -174 the sp1 binding was under the detection limit for both WT and CerS2 null astrocytes. Moreover, Sp1 binding to the -68 motif was reduced significantly in CerS2 null astrocytes (Fig. 13D). Fatty acid synthase (FASN) and glycolipid transfer protein (GLTP) are known Sp1 targets102,103 and their mRNA levels in CerS2 null astrocytes were significantly reduced. TLR4 and GFAP were used as controls as their promoter regions do not contain Sp1 motifs (Fig. 13E). These results further implicate the involvement of Sp1 in the CerS2 null phenotype.

Fig. 13. The Sp1 transcription factor (A) Western blot of Sp1 and HSF-1 in WT and CerS2 null astrocytes. GAPDH was used as a loading control. Results are from a typical experiment repeated at least 4 times with similar results. Values are means ± s.e.m, *, P<0.05. (B) mRNA levels of Sp1 in WT and CerS2 null astrocytes. N=4, values are means ± s.e.m. *, P<0.05. (C) Schematic presentation of the promoter region of the mouse Hsc70 (Hspa8) gene. The first exon is shown as a gray box, the transcription start site (TSS) is indicated by a black arrow, and two Sp1 interaction motifs at positions -68 and -174 are shown in light gray. (Scheme is drawn to scale, mouse genome version GRCm38, chr9:40,801,066-40,801,348). (D) Chromatin IP analysis of the binding of Sp1 transcription factor to its predicted binding site in the Hsc70 promoter region (position - 68). HA antibody was used as a binding control for Sp1 binding. N=3, values are means ± s.e.m. *, P<0.05. (E) mRNA levels of fatty acid synthase (FASN) and glycolipid transfer protein (GLTP), two known targets of Sp1 and TLR4 and GFAP which are predicted to be Sp1 non-targets, in WT and CerS2 null astrocytes. N=3, values are means ± s.e.m. *, P<0.05.

32 Sp1 is sensitive to its O-glycosylation (GlcNAc) state inasmuch as Sp1 hypo-glycosylation leads to its proteasomal degradation104. ROS have been shown previously to induce proteasome mediated reduction of Sp1 levels through reduction of Sp1 O-glycosylation105, whereas N-acetyl cysteine (NAC) has been shown to mitigate this effect106. Thus, I next measured ROS levels in CerS2 null astrocytes, which were elevated ∼fourfold (Fig. 14A), along with an increase in levels of the oxidative stress marker, 3-nitrotyrosine (Fig. 14B). ROS overproduction in CerS2 null astrocytes is likely to be due to reduced activity of mitochondrial respiratory chain complex IV (also known as COX) (Fig. 14C), similar to that observed in the liver from CerS2 null mice107.

Fig. 14. Oxidative stress in CerS2 null astrocytes (A) FACS analysis of ROS levels in WT (green) and CerS2 null (red) astrocytes using cmH2DCF-DA. Results are from a typical experiment repeated 3 times. Individual samples contained at least 30,000 cells in each sample. Values are means ± s.e.m. ***, P<0.001. (B) Western blot of 3-nitrotyrosine (3-NTT) in WT and CerS2 null astrocytes. GAPDH was used as a loading control. Results are from a typical experiment repeated at least 4 times with similar results. Values are means ± s.e.m, **, P<0.01. (C) Activity of mitochondrial complex IV normalized to that of citrate synthase in WT and CerS2 null astrocytes. N=4, Values are means ± s.d, ***, P<0.001.

Next, I used H2O2 to mimic oxidative stress in wild type cells and NAC to rescue CerS2 null cells. A time course showed that 4 h treatment with 100 µM H2O2 to wild type cells reduced LDL and transferrin uptake whereas 4 h treatment with100 µM NAC in CerS2 null cells elevated uptake of LDL and transferrin to similar levels as in WT cells (Fig. 15A, B). Nonetheless, Hsc70 and Sp1 protein levels exhibited reduction in response to oxidative stress and elevation following anti-oxidant treatment (Fig. 15C). Notably, neither LDL receptor nor transferrin receptor protein levels had changed (Fig. 15C).

Fig. 15. Oxidative state of the cell modulates Hsc70, Sp1 and CME

(A,B) WT and CerS2 null astrocytes were incubated with and without H2O2 (100 μM, 4 h) or N-acetyl cysteine (100 μM, 4 h). Subsequently, cells were incubated with fluorescently labeled (Dil) LDL for the indicated times. Images were obtained using confocal microscope. Scale bar, 10 μm. (C,D) WT and CerS2 null astrocytes were incubated with and without H2O2 (100 μM, 4 h) or N-acetyl cysteine (100 μM, 4 h). Subsequently, cells were incubated with fluorescently labeled (Cy5) transferrin. Values are means ± s.e.m, N=3. NS, P>0.05; *, P<0.05; **, P<0.01. (E) Left panel: Western blot of Sp1, Hsc70, CD71 (transferrin receptor) and LDL receptor in WT and CerS2 null astrocytes. GAPDH was used as a loading control. Results are from a typical experiment repeated 3 times with similar results. Right panel: Quantification of protein levels. Values are means ± s.e.m, *, P<0.05; **,P<0.01.

34 Fig. 16 shows a schematic presentation of the proposed mechanism by which endocytosis is altered in CerS2 null astrocytes. Ceramide levels are altered due to CerS2 knock out leading to elevated oxidative stress. Sp1 expression is reduced due to elevated oxidative stress causing reduced binding of Sp1 to its target Hsc70 promoter. Hsc70 down regulation leads to reduced clathrin mediated endocytosis and elevated clathrin levels.

Fig. 16. Putative scheme of the effect of altering the SL acyl chain length on CME (A) Schematic presentation of the regulation of CME by ceramide through an oxidative stress- Sp1-Hsc70 axis. (B) Illustration of the proposed mechanism for the reduced CME. The upper panel illustrates a WT astrocyte with normal CME. The lower panel illustrates a CerS2 null astrocyte (1) that generates excess long chain ceramides due to its inability to generate very-long chain ceramides (2). This elevation in C16-ceramide causes the mitochondria to generate significantly more reactive oxygen species (3). Elevated ROS levels reduce the expression of the Sp1 transcription factor (4), thus reducing the expression of its downstream target Hsc70 (5). Lower levels of Hsc70 are available to participate in the shedding of the clathrin coat from clathrin-coated vesicles (6). This results in clathrin heavy chain aggregates (7), which are not able to participate in another round of endocytosis (8). External modulators of cellular oxidative state can increase or inhibit the mechanism (9).

35 6.3. Characterization of CerS1/CerS2 double knockout mice We have generated a double knock out mouse for both CerS150 and CerS241. Although all of the individual CerS knockout mice were generated in the past years36-40, this is the first mouse harboring a mutation in two CerS proteins. Dr. Ben-David, a former graduate student, and I collaboratively aimed to characterize this mouse and gain better knowledge of the importance of these enzymes in the SL biosynthetic pathway. CerS1/2 double knockout mice (dKO) have a life span of about 4 weeks. They are born smaller than their littermates, they suffer from ataxia starting at day 10, and by the end of week 3-4, they can barely move. All the major organs (i.e. liver, kidney, spleen, intestine, muscle and lung) were analyzed at ~3 weeks of age without a clear pathology with the exception of the brain. CerS in vitro activity assays were performed on brain homogenates to assess the reduction in activity in the mutant mice. When analyzing activity with C18-CoA as a substrate (CerS1 activity), a significant reduction was observed in the CerS1 mutant mouse as well as in the dKO mouse (Fig. 17A). Moreover, activity with C24:1-CoA as a substrate (CerS2 activity) showed a significant reduction in the CerS2 null mouse as well as in the dKO mouse (Fig. 17B), while activity with C16-CoA and C20-CoA (CerS4, CerS5 and CerS6) did not change between the different mice strains (data not shown).

Fig. 17. CerS activity of dKO brain (A) CerS activity towards C18-CoA. Values are means ± s.e.m, N=3 **, P<0.01. (B) CerS activity towards C22-CoA. Values are means ± s.e.m, N=3 *, P<0.05.

36 ESI-MS/MS analysis, performed by Dr. Al Merrill revealed that dKO mice have reduced very-long chain SLs (C22-C24- ceramide and hexosylceramide) to the same extent as CerS2 null. Moreover, dKO mice show a marked reduction in C18-hexosylceramide comparable to the CerS1 point mutation mice. This reduction was not observed in C18-ceramide species, probably due to CerS4 activity (Fig. 18 A,B). In addition, dKO mice display elevation in both ceramide species (C16 and C20), thus maintaining unaltered total ceramide levels (Fig. 18C). dKO mice accumulate sphinganine and sphingaine-1-phosphate (SaP) (6-fold and 56-fold respectively (Fig. 18D).

Fig. 18. The sphingolipidome of dKO brain Mass spectrometry analysis of acyl chain length species of ceramide (A), hexosylceramide (B) total ceramide levels (C) and long chain bases (D). Values are means ± s.e.m, N=3. *, P<0.05; **, P<0.01; (E) Summary of the results presented in A-D. Average values are in pmol/mg protein.

When Dr. Ben David analyzed the brain of these mice by hematoxylin and eosin staining, she observed that the dKO brains have both the phenotype expected from each of the two single CerS ablations, only more severe. The vacuolization observed in CerS2 null mice at the age of 3 months was present at the age of 3 weeks. Additionally, the Purkinje cell loss detected in CerS1 point mutation adult mice was observed in the dKO mice at the age of 4 weeks. Moreover, dKO mice show microglial activation similar to both single mutations. CerS1 mutants show gray matter activation of microglia, while CerS2 null mice have high levels of activated microglia in white matter regions. I verified the Purkinje cell loss by western blot analysis of calbindin (a Purkinje cell marker) and microglial activation by western blot analysis of Mac2 (a microglial marker). I also detected high levels of astrogliosis using GFAP (a glial marker) and high levels of apoptosis by caspase 8 cleavage (Fig. 19). Indeed, dKO brains had lower levels of calbindin to the same extent of CerS1 point mutation and significantly higher levels of GFAP, Mac2 and cleaved caspase 8 (Fig. 19).

Fig. 19. Inflammation and apoptosis in dKO brain Western blot of Mac2, GFAP, calbindin and caspase 8 in WT, CerS1 mutant, CerS2 null and CerS1/2 dKO mouse brains. GAPDH was used as a loading control. Results are from a typical experiment repeated at least 3 times with similar results. Values are means ± s.e.m, *, P<0.05; **, P<0.01; ***, P<0.001.

38 6.4. Exploring the hyper-sensitivity of CerS2 null mice to DSS induced colitis Colitis is an idiopathic disease of the gut that can lead to colon cancer. Elevated levels of the bioactive lipid S1P were shown to cause colitis and colon cancer. Moreover, ulcerative colitis patients display increased levels of TNF α. Since CerS2 null mice exhibit elevated levels of S1P in their serum (unpublished data) and increased TNF alpha signaling108, we tested the response of CerS2 null mice to chemically-induced colitis. Dextran sodium sulfate (DSS) is a well-described model of chemically-induced colitis109 and is used to investigate the mechanisms underlying inflammatory bowel diseases like ulcerative colitis and Crohn’s diseas110. In this model, mice are given DSS in their drinking water after which they develop colitis within 7 days of treatment. The colitis manifests in weight loss, soft stool, bloody diarrhea and shortened colon. The initial observation made by our collaborator, Dr. Park from IOWA university South Korea, was that CerS2 null mice are much more sensitive and show aggravated colitis symptoms in response to DSS. Since colitis involves both the immune system and the intestine epithelium, I tested which of these cells is the culprit for the hyper- sensitivity of CerS2 null mice to DSS. Initially, I generated mixed bone marrow chimeras, where I irradiated WT and CerS2 null mice and transplanted with a 1:1 mixture of WT and CerS2 null bone marrow (BM) cells. Three months after the transplantation, mice were treated with DSS, weighed daily and sacrificed after 7 days of treatment. CerS2 null mice transplanted with mixed BM lost significantly more weight than their WT transplanted with mixed BM littermates (Fig. 20). Fig. 20. Changes in body weight after DSS in mixed BM chimeras Body weight loss of mixed bone marrow transplanted WT and CerS2 null mice during 7 days of DSS treatment. N=5. Values are means ± s.e.m, *, P<0.05; **, P<0.01; ***, P<0.001.

Fig. 21. Clinical characterization of effect of DSS on mixed BM chimeras Colon length and clinical score of mixed bone marrow transplanted WT and CerS2 null mice after 7 days of DSS treatment. N=5. Values are means ± s.e.m, **, P<0.01; ***, P<0.001.

39 Moreover, the clinical score (composed of weight loss, stool consistency and bloody diarrhea) of CerS2 null mice transplanted with mixed BM was 4-fold higher than their WT counterparts, whereas the colon length of CerS2 null mice transplanted with mixed BM was shorter by 2 cm on average than WT controls (Fig. 21). Next, I tested whether an interaction between a CerS2 null immune cell and the CerS2 null epithelium causes the CerS2 null mice to be more sensitive to DSS. Therefore, I repeated the BM transplantation experiment but this time transplanted WT BM to WT and CerS2 null mice and CerS2 null BM to WT and CerS2 null mice. After three months of recuperation the four groups of mice were treated with DSS as described above. Interestingly, CerS2 null mice lost more weight than WT mice regardless of the genotype of the transplanted BM (Fig.22).

Fig. 22. Changes in body weight after DSS in full BM chimeras Body weight of WT/CerS2 null bone marrow transplanted WT and CerS2 null mice during 7 days of DSS treatment. N=5-8. Values are means ± s.e.m, *, P<0.05; ***, P<0.001.

Examining the colon length and clinical score of these mice showed the same pattern. CerS2 null mice displayed a higher clinical score and shorter colon regardless of the genotype of the transplanted BM (Fig. 23). Based on these results, I conclude that the cause of the CerS2 null mice hypersensitivity to DSS induced colitis is not due to defects in the CerS2 null immune system, rather it is due to defects in the barrier function of the CerS2 null epithelium in the colon.

Fig. 23. Clinical characterization of DSS effect in full BM chimeras Colon length and clinical score of WT/CerS2 null bone marrow transplanted WT and CerS2 null mice, after 7 days of DSS treatment. N=5-8. Values are means ± s.e.m, **, P<0.01; ***, P<0.001.

40 6.5. The involvement of CerS dimerization in glioblastoma I participated in a fruitful collaboration with Prof. Stegh111 in which we showed that Bcl2L13 is overexpressed in glioblastoma (GBM) and other malignancies and interacts with CerS2 and 6 rendering them unable to dimerize with other CerS molecles. Bcl-2 and other members of the Bcl-2 protein family play pivotal and well-defined roles in several cancers112. For example, Bcl2L12 was identified as an oncoprotein with frequent mRNA and protein upregulation in GBM tumors. While it has been well established that CerS-mediated ceramide synthesis is an integral part of mitochondria-controlled intrinsic apoptosis signaling, and an important factor regulating tumorigenesis, specific mechanisms of CerS regulation during apoptosis are not well understood. We showed that Bcl2L13 binds CerS2 and CerS6, effectively blocking the formation of the pro-apoptotic CerS2-CerS6 complex, thus, contributing to the GBM resistance to chemotherapy. This paper establishes the Bcl2L13- CerS axis as a therapeutic intervention target. I performed CerS activity assay on control and Bcl2L13 knock down cells treated with curcumin, a CerS2 dimerization agent49, for the indicated times. Both CerS2 and CerS6 activity was increased in the Bcl2L13 knock down cells after curcumin treatment, thus implying that Bcl2L13 is an inhibitor of CerS2-6 dimerization and consequently a regulator of CerS activity (Fig. 24A,B). In addition, I showed that there is a negative correlation between levels of Bcl2L13 mRNA and CerS2 or CerS6 activity in human GBM biopsies (Fig. 24C,D).

Fig. 24. Effects of Bcl2L13 knock down on CerS2 and CerS6 activity in vitro and in vivo Effect of Bcl2L13 knock down on CerS2 (A) and CerS6 (B) enzymatic activity was determined by treating SF767 cells with curcumin (8 μM) for the indicated periods of time before measuring CerS2 and CerS6 activity. Data are means ± SD, *P < 0.05. Correlation between Bcl2L13 mRNA levels and CerS2 (C) or CerS6 activity (D) in human GBM tumor samples.

41 6.6. Other collaborative projects During my Ph.D. I was involved in several collaborations, which were not yet developed into a complete story. Dr. Rotem Tidhar, a former post doc in my lab, showed that by replacing eleven amino acids in CerS5 with their corresponding CerS2 amino acids, CerS5 specificity toward acyl-CoA was altered inasmuch as CerS5(299-309 CerS2) can utilize C24:1-CoA as a substrate to form ceramide. I was able to show that replacing the same region in CerS4 with the same eleven amino acids from CerS2 allowed CerS4(291-301 CerS2) to use C24:1-CoA as well (Fig. 25A). Moreover, replacing this region in CerS4 with the CerS5 corresponding amino acids allowed CerS4(291-301 CerS5) to use C16-CoA (Fig. 25B). Notably, replacing CerS5 with the corresponding region of CerS4 reduced, yet did not abrogate, the ability of CerS5(299-309 CerS4) to use C16-CoA (Fig. 25B). These results suggest that the acyl-CoA specificity site in CerS resides, although not entirely, within these eleven amino acids.

Fig. 25. Eleven amino-acid residues are important for CerS acyl-CoA specificity. (A) CerS4(291-301 CerS2) with C24:1-CoA. (B) CerS5(299-309 CerS4) and CerS4(291-301 CerS5) with C16-CoA. Results are mean values ± S.D, N = 3. **, P<0.05; **, P<0.01

42 Additionaly, I collaborated with Zora Bioscience (Finland) to look at the turnover of SLs. Iris D Zelnik, a Ph.D. student in my lab, and I treated HEK cells with stable isotope labeled fatty acids (obtained from Zora Bioscience) in a pulse chase experiment (as described in the methods). The most exciting finding from this experiment was that C24:1-ceramides are turned over very quickly with a T1/2 of ~5h, whereas C16:0-, C18:0- and C24:0-ceramide levels remained constant throughout the 24h pulse-chase experiment (Fig 26).

Fig. 26. Turnover rate of ceramide with different acyl-CoAs. Cells were pulsed with either C16:0 ( ), C18:0 ( ), C24:0 ( ) and C24:1 ( ) stable isotope labeled fatty acids, and chased for the indicated amount of time. Generation and turnover of ceramide containing different acyl-CoA length was analyzed by MS. (A) C16:0- d18:1, (B) C18:0-d18:1, (C) C24:0-d18:1 and (D) C24:1-d18:1. Results are mean values ± S.D, N=3.

43 7. Discussion In my PhD, I set out to examine the role of CerS2 in the development of various pathologies and the impact of CerS2 deficiency on processes that rely on membrane structure. Since CerS2 deficiency causes significant changes to the biophysical properties of the membrane113, it was expected that trafficking, a cellular process that highly depends on membrane properties, might be affected. Correspondingly, it was shown previously that down regulation of schlank, the Drosophila CerS homolog, led to endosomal trafficking defects114 yet the precise mechanism connecting defective trafficking and SL alterations is not clear. In mammals, our lab has previously observed a number of changes in receptor activation and/or internalization. For instance, altering the SL acyl chain composition abrogated insulin receptor phosphorylation56, and palmitic acid internalization by CerS2 null hepatocytes57. These effects are directly caused by changes in membrane biophysical properties, with the inability of insulin receptor or CD36/FAT (respectively) to translocate into detergent-resistant membranes prior to internalization. Likewise, altering the SL acyl chain composition also affects intracellular trafficking of connexin 32115, again due to altered biophysical properties. Finally, the mechanism by which tumor necrosis factor α receptor-1 internalization was inhibited in CerS2 null hepatocytes was not established65. Since SLs can reside in lipid rafts116, platforms through which caveolae are formed and thereby uptake ligands through caveolin-mediated endocytosis63, we expected to see reduced uptake of cargo internalized by this mechanism. Although we observed significant changes in the SL profile of CerS2 null astrocytes, similar to those observed in CerS2 null liver and whole brain53,56, caveolin-mediated endocytosis was unaltered, presumably since total ceramide levels did not change in CerS2 null astrocytes. This is not the case in CME.

The complex process of CME involves a myriad of proteins that help generate the force needed for vesicle invagination, fission and recycling; deficiency of some causes clear uptake reduction of CME cargo. For instance the BAR-domain proteins, epsins and endophilins, induce and stabilize membrane curvature, and their knockout cause prominent CME defects117,118. Similarly, inhibition of dynamin, the protein responsible for the contraction of the nascent vesicle’s neck and its subsequent fission, causes massive reduction in endocytosis119. In contrast, modulation of CME in CerS2 null astrocytes is milder and resembles the lower severity of CME reduction seen in auxilin-1 knockout98 or Hsc70 knockdown120 which shed the clathrin coat from the vesicles. Alongside CME reduction, auxilin-1 knockout mice exhibit clathrin self-assembly into empty cages and elevated CHC

44 levels121. Moreover, it was shown that Hsc70 prevents CHC aggregation by chaperoning it in the cytosol following clathrin uncoating122 and that low levels of Hsc70 cause protein aggregation that inhibit CME120. Hsc70 is regulated by Sp1 transcription factor100 that can be subjected to numerous posttranslational modifications including acetylation, phosphorylation and glycosylation and its levels are sensitive to the oxidative state of the cell105. Our lab had previously shown107 that ROS generation in CerS2 null liver is caused by direct inhibition of complex IV activity by C16:0-ceramide and sphinganine, but not C24:0 or C24:1-ceramides. The specificity of the effect of C16-ceramide is further strengthened by the lack of effect of C16:0-C18:1-diacylglycerol. Both of these lipids have similar hydrophobic properties, but the lack of effect of the latter on complex IV demonstrates a specific mode of interaction between C16:0-ceramide and complex IV. However, the mechanism by which C16:0- ceramide accesses complex IV in living mitochondria is currently unknown. Moreover, the effects of oxidative stress on endocytosis were not extensively studied and the mechanisms governing these effects are not yet understood. Nevertheless, some clues of how oxidative stress is implicated in CME can be found. For example, oxidative stress can induce the translocation of Hsc70 to the nucleus123, or reduce transferrin binding sites at the cell surface124. Additionally, oxidative stress was found to inhibit LDL and transferrin internalization125,126. Remarkably, antioxidants such as NAC can reverse phenotypes associated with oxidative stress including Sp1 reduced protein levels106 and reduced transferrin uptake127, similar to my observations. Whether other pathologies seen in CerS2 null mice such as encephalopathy53, insulin resistance56 or pheochromocytoma54 might be nullified by chronic treatment with antioxidant such as NAC or vitamin E is yet to be determined. Moreover, this project highlights the need to pay attention to the direct effects of SLs on membrane biophysical properties but more importantly on the indirect effects caused by activation of down-stream pathways such as oxidative stress (Fig. 27)128. For example, CerS2 null mice were shown to be more sensitive to DSS induced colitis by our collaborator, Dr. Park. Oxidative stress and changes in ceramide levels were both implicated in the pathogenesis of hyper-sensitivity to DSS colitis129,130. Since CerS2 deficiency was shown to cause high levels of inflammation108, I attempted to understand whether the immune system or the epithelium barrier is the culprit for the CerS2 null hyper- sensitivity to DSS. In order to delineate between the effect of the epithelium and the immune system, I generated a series of chimeric mice, which were depleted of their immune system by irradiation and were transplanted with bone marrow pluripotent immune cells of specific genotypes. Initially, it seemed that CerS2 null mice hyper-sensitivity to DSS colitis was

45 independent of the immune system and was determined only by the lack of CerS2 in the epithelium of the colon. But advances in a different project in our lab suggested the opposite. In this project, Dr. Ashish Saroha showed that CerS2 null mice lack iNKT cells. Moreover, the amount of iNKT cells in CerS2 null bone marrow chimeras was significantly lower than the WT chimeras regardless of the genotype of the injected bone marrow. The fact that iNKT cells have a role in hyper sensitivity to DSS induced colitis 131 fits with my data and suggests that the hyper-sensitivity of CerS2 null mice to DSS is not only due to a dysfunctional epithelium barrier but also due to the lack of iNKT cells.

Fig. 27. Putative scheme of the effect of altering the SL acyl chain length on CME. Altering the SL acyl chain length (red) can either lead to changes in membrane biophysical properties or to activation of downstream pathways (blue). In the current study (black), we demonstrate the activation of one such pathway, which eventually leads to a reduced rate of CME128.

Until recently, mice deficient of each CerS were generated worldwide. My colleague, Dr. Ben-David, crossed a mouse with a catalytically inactive CerS1 mutation with CerS2 null mice. These dKO mice have the pathological manifestations of both single mutant mice. The vacuoles seen in the CerS2 null mice and the Purkinje neuronal loss seen in the CerS1 mutant mice are both present in the dKO and they are found at a younger age than in both single mutant mice. Due to the fact that CerS1/2 dKO mice do not reach adulthood, we could not assess the behavioral implications of such deficiency as seen both in CerS2 null mice and in human heterologous CerS2 deletion such like epilepsy and light sensitivity53,132. Interestingly other major organs did not display any apparent pathology suggesting that the brain is the most sensitive organ towards CerS1 and CerS2 abrogation. This effect reveals important information about the regulatory mechanisms of the different CerS, which have yet to be discovered in full. On the one hand, each CerS has an independent regulatory mechanism, as

46 loss of two enzymes does not have a synergistic effect. On the other hand, other family members can still compensate, to some extent, for the loss of two enzymes. In addition, all mouse models used in this project (i.e. CerS1, CerS2 and CerS1/2 knock out mice) displayed accumulation of long chain bases, which serve as substrate for all CerSs. Thus, we cannot exclude that the pathologies observed are not due to changes in the acyl chain profile of ceramide species rather they are due to accumulation of long chain bases as a recent study suggested133. Other important information gained by this study is the role (or lack of it) of CerS in embryonic development. Since CerS1/2 dKO mice are born without apparent developmental defects it suggests that CerS1 and CerS2 are not crucial for proper embryonic development.

Glioma tumor grade negatively correlates with overall ceramide levels, suggesting that ceramide metabolism plays a role in the pathogenesis of glioma134. Recently our lab showed that dimeric complex formation regulates CerS activity as the substrate binding to 49 one monomer allosterically reduces the Km of the second monomer for the substrate . This was corroborated by the biphasic kinetics curve of CerS3 toward acyl-CoA135, that suggests more than one acyl-CoA binding site for each Ceramide synthesis reaction. This is particularly important for CerS2 and CerS3, which have a low-level activity as monomers but are the main CerSs responsible for the generation of VLC-ceramides45. In collaboration with Dr. Stegh, we showed that in glioblastoma, Bcl2L13, an atypical member of Bcl-2 family, acts as an antiapoptotic protein by binding to CerS2 (and CerS6) thus inhibiting CerS dimerization and subsequent ceramide formation. This is an example of how investigating CerS structure-function can lead to potential therapeutics. Altogether, I found that CerS in general and CerS2 in particular have a role in the normal and healthy function of cellular processes. Additionally, elimination of CerS2 activity (genetically, chemically or biologically) leads to the pathogenesis of various abnormalities and diseases such as oxidative stress, brain inflammation, colitis and cancer.

Future perspectives: In my Ph.D I have broadened our understanding of the role of CerS2 in health and disease. It has also set the ground for many potential directions that can be pursued. For example, since CerS2 null hepatocytes and astrocytes share several commonalities but also display some differences, it is possible to use the newly available CRISPR tool to generate different CerS2 null cell lines and characterize the impact of the lack of CerS2 in different tissues or cell

47 types. Since it has been established that CerS2 null mice show differences in EEG53 and display behavioral changes132 I believe that generating CerS2 null neurons would be beneficial to understand the role of CerS2 in the appropriate neuronal function. Moreover, using CerS2 null astrocytes and neurons in mixed culture can shed light on the role of CerS2 in the neuron-astrocyte crosstalk136 that is important for neurotransmission, metabolite homeostasis and synapse modulation137. In addition, CerS2 null astrocytes have been shown to accumulate an unidentified storage material in vivo53. Understanding the type of material/ lipid could highlight the relationship between CerS and other enzymes in the SL pathway. Such like it was shown that CerS activity is essential for the beneficial effect on SK1 inhibition in PAH138. Oxidative stress plays a key role in myriad of pathologies such as pheochromocytoma54, insulin resistance139, epilepsy 140 and lung injury induced by Pseudomonas infection141, all of which are pathologies observed in CerS2 null mice. The fact that reduced uptake of LDL and transferrin (another result of oxidative stress123, 124) can be reversed by anti-oxidant treatment suggests that at least some of the other oxidative stress induced pathologies could be rescued as well. Since there is no 3D structure of any of the CerS, it is very important to get as much structure-function information using other methods. For example we now know that the LCB binding site is very promiscuous and allows the binding (and subsequent N-acylation) of various molecules such as fumonisin B1142, Jaspin B (Cingolani et. al. unpublished data), NBD-So, NBD-Sa82, NBD-phyto-So143 and FTY720144. On the contrary, the acyl-CoA binding site of CerS is much more specific inasmuch as it can differentiate between the length of acyl-CoAs that differ only by two carbon atoms (CerS1 can use C18-CoA but not C16- or C20-CoA). Our lab has showed that the acyl-CoA specificity of CerS resides within 11 amino acids. Although it is clear that these 11 amino acids have a role in acyl-CoA specificity it is still not clear if their mutation gives rise to a gain or loss of function. That is whether mutation of this region results in a genuine change of specificity or is it loss of specificity that allows the mutated CerS to use all acyl-CoAs. Thus, the characterization of 11 amino acid region that can change the specificity of CerS can give new insights in the search for specific CerS inhibitors (which do not exist at the moment). The discovery that C24:1- ceramide is far less stable than other ceramide molecules is astonishing. This is the first time anyone has shown the half-life of any ceramide species. Since C24:1 is the product of the only CerS which is ubiquitously expressed, CerS242,45, it is not clear at the moment what this means, but it may suggest that CerS2 has to generate C24:1-ceramide constantly, and that any

48 interference in CerS2 constant activity might result in an acute lack of C24:1-ceramides which in turn may effect pathologies such as ischemic heart disease145. Since these are only preliminary experiments, it is crucial to gain better knowledge of the turnover of other SLs, which may explain other pathologies and anomalies. Finally, I was interested in the acute regulation of CerS activity. For example, I have shown that CerS2 (and CerS6) activity is inhibited by interaction with Bcl2L13 rendering it inaccessible to form dimers. I also strove to understand the importance of CerS phosphorylation on CerS activity, which was published by another group52. Although it is now known that CerS is phosphorylated47 and that this phosphorylation activates CerS, little is known on the mechanisms and signaling pathways leading to the phosphorylation and subsequent activation of CerS. Moreover, there is hardly any data to suggest which kinase(s) are responsible for CerS phosphorylation146,147. Therefore, I suggest to generate a phospho- CerS antibody that will serve as a tool to measure CerS phosphorylation in response to various stimuli, and will also allow to determine which kinase(s) and signaling pathways are involved in the activation of CerS by phosphorylation. Altogether, despite what I have learnt during my Ph.D., there is still much more to learn about the role of CerS2 and its regulation in health and disease.

49 8. List of publications from PhD work Jensen SA, Calvert AE, Volpert G, Kouri FM, Hurley LA, Luciano JP, Wu Y, Chalastanis A, Futerman AH, Stegh AH. Bcl2L13 is a ceramide synthase inhibitor in glioblastoma. (2014) Proc Natl Acad Sci U S A. 2014 Apr 15; 111(15): 5682-7

MacRitchie N, Volpert G, Al Washih M, Watson DG, Futerman AH, Kennedy S, Pyne S, Pyne NJ. Effect of the sphingosine kinase 1 selective inhibitor, PF-543 on arterial and cardiac remodelling in a hypoxic model of pulmonary arterial hypertension. (2016) Cell Signal. Aug; 28(8): 946-55.

Zigdon, H., Meshcheriakova, A., Farfel-Becker, T., Volpert, G., Sabanay, H. and Futerman, A.H. Altered lysosome distribution is an early neuropathological event in neurological forms of Gaucher disease. (2017) FEBS lett. 2017 Mar;591(5):774-783.

Volpert, G., Ben-Dor, S., Tarcic, O., Duan, J., Merrill, A.H. Jr., Pewzner-Jung, Y. and Futerman, A.H. Oxidative stress elicited by modifying the ceramide acyl chain length reduces the rate of clathrin-mediated endocytosis. (2017) J cell sci. 2017 Apr 15;130(8):1486-1493.

Ferreira, N.S., Engelsby, H., Pedersen, D.N., Samuel, K.L., Volpert, G., Merrill, A.H., Jr, Færgeman, N.J and Futerman, A.H. Regulation of very-long acyl chain ceramide synthesis by acyl-CoA binding protein. (2017) J Biol Chem. 2017 Mar 19. pii: jbc.M117.785345.

50 9. References 1. Fahy, E. et al. A comprehensive classification system for lipids. J. Lipid Res. 46, 839–861 (2005). 2. van Meer, G., Voelker, D. R. & Feigenson, G. W. Membrane lipids: where they are and how they behave. Nat. Rev. Mol. Cell Biol. 9, 112–124 (2008). 3. Rubenstein, J. L., Smith, B. A. & McConnell, H. M. Lateral diffusion in binary mixtures of cholesterol and phosphatidylcholines. Proc. Natl. Acad. Sci. U.S.A. 76, 15–18 (1979). 4. Futerman, A. H. & Hannun, Y. A. The complex life of simple sphingolipids. EMBO Rep. 5, 777–782 (2004). 5. Venkataraman, K. et al. Upstream of growth and differentiation factor 1 (uog1), a mammalian homolog of the yeast longevity assurance gene 1 (LAG1), regulates N- stearoyl-sphinganine (C18-(dihydro)ceramide) synthesis in a fumonisin B1- independent manner in mammalian cells. J. Biol. Chem. 277, 35642–35649 (2002). 6. Lahiri, S. & Futerman, A. H. The metabolism and function of sphingolipids and glycosphingolipids. Cell. Mol. Life Sci. 64, 2270–2284 (2007). 7. Hannun, Y. A. & Obeid, L. M. Principles of bioactive lipid signalling: lessons from sphingolipids. Nat. Rev. Mol. Cell Biol. 9, 139–150 (2008). 8. van Blitterswijk, W. J., van der Luit, A. H., Veldman, R. J., Verheij, M. & Borst, J. Ceramide: second messenger or modulator of membrane structure and dynamics? Biochem. J. 369, 199–211 (2003). 9. Clarke, C. J. et al. The extended family of neutral sphingomyelinases. Biochemistry 45, 11247–11256 (2006). 10. Hirschberg, K., Rodger, J. & Futerman, A. H. The long-chain sphingoid base of sphingolipids is acylated at the cytosolic surface of the endoplasmic reticulum in rat liver. Biochem. J. 290 ( Pt 3), 751–757 (1993). 11. Futerman, A. H. & Riezman, H. The ins and outs of sphingolipid synthesis. Trends Cell Biol. 15, 312–318 (2005). 12. Tidhar, R. & Futerman, A. H. The complexity of sphingolipid biosynthesis in the endoplasmic reticulum. Biochim. Biophys. Acta 1833, 2511–2518 (2013). 13. Mullen, T. D., Hannun, Y. A. & Obeid, L. M. Ceramide synthases at the centre of sphingolipid metabolism and biology. Biochem. J. 441, 789–802 (2012). 14. Merrill, A. H. De novo sphingolipid biosynthesis: a necessary, but dangerous, pathway. J. Biol. Chem. 277, 25843–25846 (2002).

51 15. Sonnino, S. & Chigorno, V. Ganglioside molecular species containing C18- and C20-sphingosine in mammalian nervous tissues and neuronal cell cultures. Biochim. Biophys. Acta 1469, 63–77 (2000). 16. Valsecchi, M. et al. Sphingolipidomics of A2780 human ovarian carcinoma cells treated with synthetic retinoids. J. Lipid Res. 51, 1832–1840 (2010). 17. Bollinger, C. R., Teichgräber, V. & Gulbins, E. Ceramide-enriched membrane domains. Biochim. Biophys. Acta 1746, 284–294 (2005). 18. Klemm, R. W. et al. Segregation of sphingolipids and sterols during formation of secretory vesicles at the trans-Golgi network. J. Cell Biol. 185, 601–612 (2009). 19. Gulbins, E. & Grassmé, H. Ceramide and cell death receptor clustering. Biochim. Biophys. Acta 1585, 139–145 (2002). 20. Van Overloop, H., Gijsbers, S. & Van Veldhoven, P. P. Further characterization of mammalian ceramide kinase: substrate delivery and (stereo)specificity, tissue distribution, and subcellular localization studies. J. Lipid Res. 47, 268–283 (2006). 21. Carré, A. et al. Ceramide kinase targeting and activity determined by its N-terminal pleckstrin homology domain. Biochem. Biophys. Res. Commun. 324, 1215–1219 (2004). 22. Hanada, K. [The molecular machinery CERT for intracellular trafficking of ceramide]. Seikagaku 76, 562–570 (2004). 23. Xu, R. et al. Golgi alkaline ceramidase regulates cell proliferation and survival by controlling levels of sphingosine and S1P. FASEB J. 20, 1813–1825 (2006). 24. Hwang, Y.-H., Tani, M., Nakagawa, T., Okino, N. & Ito, M. Subcellular localization of human neutral ceramidase expressed in HEK293 cells. Biochem. Biophys. Res. Commun. 331, 37–42 (2005). 25. Bernardo, K. et al. Purification, characterization, and biosynthesis of human acid ceramidase. J. Biol. Chem. 270, 11098–11102 (1995). 26. Liu, H., Chakravarty, D., Maceyka, M., Milstien, S. & Spiegel, S. Sphingosine kinases: a novel family of lipid kinases. Prog. Nucleic Acid Res. Mol. Biol. 71, 493– 511 (2002). 27. Van Veldhoven, P. P. Sphingosine-1-phosphate lyase. Meth. Enzymol. 311, 244–254 (2000). 28. Schorling, S., Vallée, B., Barz, W. P., Riezman, H. & Oesterhelt, D. Lag1p and Lac1p are essential for the Acyl-CoA-dependent ceramide synthase reaction in Saccharomyces cerevisae. Mol. Biol. Cell 12, 3417–3427 (2001).

52 29. Guillas, I. et al. C26-CoA-dependent ceramide synthesis of Saccharomyces cerevisiae is operated by Lag1p and Lac1p. EMBO J. 20, 2655–2665 (2001). 30. Vallée, B. & Riezman, H. Lip1p: a novel subunit of acyl-CoA ceramide synthase. EMBO J. 24, 730–741 (2005). 31. Cerantola, V. et al. Yeast sphingolipids do not need to contain very long chain fatty acids. Biochem. J. 401, 205–216 (2007). 32. Lahiri, S. & Futerman, A. H. LASS5 is a bona fide dihydroceramide synthase that selectively utilizes palmitoyl-CoA as acyl donor. J. Biol. Chem. 280, 33735–33738 (2005). 33. D'mello, N. P. et al. Cloning and characterization of LAG1, a longevity-assurance gene in yeast. J. Biol. Chem. 269, 15451–15459 (1994). 34. Jiang, J. C., Kirchman, P. A., Zagulski, M., Hunt, J. & Jazwinski, S. M. Homologs of the yeast longevity gene LAG1 in Caenorhabditis elegans and human. Genome Res. 8, 1259–1272 (1998). 35. Pewzner-Jung, Y., Ben-Dor, S. & Futerman, A. H. When do Lasses (longevity assurance genes) become CerS (ceramide synthases)?: Insights into the regulation of ceramide synthesis. J. Biol. Chem. 281, 25001–25005 (2006). 36. Ginkel, C. et al. Ablation of neuronal ceramide synthase 1 in mice decreases ganglioside levels and expression of myelin-associated glycoprotein in oligodendrocytes. J. Biol. Chem. 287, 41888–41902 (2012). 37. Jennemann, R. et al. Loss of ceramide synthase 3 causes lethal skin barrier disruption. Hum. Mol. Genet. 21, 586–608 (2012). 38. Ebel, P. et al. Ceramide synthase 4 deficiency in mice causes lipid alterations in sebum and results in alopecia. Biochem. J. 461, 147–158 (2014). 39. Gosejacob, D. et al. Ceramide Synthase 5 Is Essential to Maintain C16:0-Ceramide Pools and Contributes to the Development of Diet-induced Obesity. J. Biol. Chem. 291, 6989–7003 (2016). 40. Ebel, P. et al. Inactivation of ceramide synthase 6 in mice results in an altered sphingolipid metabolism and behavioral abnormalities. J. Biol. Chem. 288, 21433– 21447 (2013). 41. Pewzner-Jung, Y. et al. A critical role for ceramide synthase 2 in liver homeostasis: I. alterations in lipid metabolic pathways. J. Biol. Chem. 285, 10902–10910 (2010). 42. Riebeling, C., Allegood, J. C., Wang, E., Merrill, A. H. & Futerman, A. H. Two mammalian longevity assurance gene (LAG1) family members, trh1 and trh4,

53 regulate dihydroceramide synthesis using different fatty acyl-CoA donors. J. Biol. Chem. 278, 43452–43459 (2003). 43. Mizutani, Y., Kihara, A. & Igarashi, Y. Mammalian Lass6 and its related family members regulate synthesis of specific ceramides. Biochem. J. 390, 263–271 (2005). 44. Mizutani, Y., Kihara, A. & Igarashi, Y. LASS3 (longevity assurance homologue 3) is a mainly testis-specific (dihydro)ceramide synthase with relatively broad substrate specificity. Biochem. J. 398, 531–538 (2006). 45. Laviad, E. L. et al. Characterization of ceramide synthase 2: tissue distribution, substrate specificity, and inhibition by sphingosine 1-phosphate. J. Biol. Chem. 283, 5677–5684 (2008). 46. Levy, M. & Futerman, A. H. Mammalian ceramide synthases. IUBMB Life 62, 347– 356 (2010). 47. Olsen, J. V. et al. Global, in vivo, and site-specific phosphorylation dynamics in signaling networks. Cell 127, 635–648 (2006). 48. Sassa, T., Hirayama, T. & Kihara, A. Enzyme Activities of the Ceramide Synthases CERS2-6 Are Regulated by Phosphorylation in the C-terminal Region. J. Biol. Chem. 291, 7477–7487 (2016). 49. Laviad, E. L., Kelly, S., Merrill, A. H. & Futerman, A. H. Modulation of ceramide synthase activity via dimerization. J. Biol. Chem. 287, 21025–21033 (2012). 50. Zhao, L. et al. A deficiency of ceramide biosynthesis causes cerebellar purkinje cell neurodegeneration and lipofuscin accumulation. PLoS Genet. 7, e1002063 (2011). 51. Bengtsson, H. et al. Generation and characterization of a Gdf1 conditional null allele. Genesis 46, 368–372 (2008). 52. Friedrich, G. & Soriano, P. Insertional mutagenesis by retroviruses and promoter traps in embryonic stem cells. Meth. Enzymol. 225, 681–701 (1993). 53. Ben-David, O. et al. Encephalopathy caused by ablation of very long acyl chain ceramide synthesis may be largely due to reduced galactosylceramide levels. J. Biol. Chem. 286, 30022–30033 (2011). 54. Park, W.-J. et al. Development of pheochromocytoma in ceramide synthase 2 null mice. Endocr. Relat. Cancer ERC–15–0058 (2015). doi:10.1530/ERC-15-0058 55. Pewzner-Jung, Y. et al. A critical role for ceramide synthase 2 in liver homeostasis: II. insights into molecular changes leading to hepatopathy. J. Biol. Chem. 285, 10911–10923 (2010). 56. Park, J.-W. et al. Ablation of very long acyl chain sphingolipids causes hepatic

54 insulin resistance in mice due to altered detergent-resistant membranes. Hepatology 57, 525–532 (2013). 57. Park, W.-J. et al. Hepatic fatty acid uptake is regulated by the sphingolipid acyl chain length. Biochim. Biophys. Acta 1841, 1754–1766 (2014). 58. Praefcke, G. J. K. & McMahon, H. T. The dynamin superfamily: universal membrane tubulation and fission molecules? Nat. Rev. Mol. Cell Biol. 5, 133–147 (2004). 59. Schlossman, D. M., Schmid, S. L., Braell, W. A. & Rothman, J. E. An enzyme that removes clathrin coats: purification of an uncoating ATPase. J. Cell Biol. 99, 723– 733 (1984). 60. Chappell, T. G. et al. Uncoating ATPase is a member of the 70 kilodalton family of stress proteins. Cell 45, 3–13 (1986). 61. Sousa, R. & Lafer, E. M. Keep the traffic moving: mechanism of the Hsp70 motor. Traffic 7, 1596–1603 (2006). 62. eisenberg, E. & Greene, L. E. Multiple roles of auxilin and hsc70 in clathrin- mediated endocytosis. Traffic 8, 640–646 (2007). 63. Mayor, S. & Pagano, R. E. Pathways of clathrin-independent endocytosis. Nat. Rev. Mol. Cell Biol. 8, 603–612 (2007). 64. Mayor, S., Parton, R. G. & Donaldson, J. G. Clathrin-independent pathways of endocytosis. Cold Spring Harb Perspect Biol 6, a016758–a016758 (2014). 65. Ali, M. et al. Altering the sphingolipid acyl chain composition prevents LPS/GLN- mediated hepatic failure in mice by disrupting TNFR1 internalization. Cell Death Dis 4, e929 (2013). 66. Eckl, K.-M. et al. Impaired epidermal ceramide synthesis causes autosomal recessive congenital ichthyosis and reveals the importance of ceramide acyl chain length. J. Invest. Dermatol. 133, 2202–2211 (2013). 67. Shiffman, D. et al. A gene variant in CERS2 is associated with rate of increase in albuminuria in patients with diabetes from ONTARGET and TRANSCEND. PLoS ONE 9, e106631 (2014). 68. Kirin, M. et al. Genome-wide association study identifies genetic risk underlying primary rhegmatogenous retinal detachment. Hum. Mol. Genet. 22, 3174–3185 (2013). 69. Cinar, R. et al. Hepatic cannabinoid-1 receptors mediate diet-induced insulin resistance by increasing de novo synthesis of long-chain ceramides. Hepatology 59,

55 143–153 (2014). 70. Eberle, M. et. al. exacerbation of experimental autoimmune encephalomyelitis in ceramide synthase 6 knockout mice is associated with enhanced activation/migration of neutrophils. Immunol Cell Biol. 93, 825-836 (2015). 71. Schiffmann, S. et al. Ceramide synthase 6 plays a critical role in the development of experimental autoimmune encephalomyelitis. J. Immunol. 188, 5723–5733 (2012). 72. Koybasi, S. et al. Defects in cell growth regulation by C18:0-ceramide and longevity assurance gene 1 in human head and neck squamous cell carcinomas. J. Biol. Chem. 279, 44311–44319 (2004). 73. Senkal, C. E. et al. Role of human longevity assurance gene 1 and C18-ceramide in chemotherapy-induced cell death in human head and neck squamous cell carcinomas. Mol. Cancer Ther. 6, 712–722 (2007). 74. Erez-Roman, R., Pienik, R. & Futerman, A. H. Increased ceramide synthase 2 and 6 mRNA levels in breast cancer tissues and correlation with sphingosine kinase expression. Biochem. Biophys. Res. Commun. 391, 219–223 (2010). 75. Schiffmann, S. et al. Ceramide synthases and ceramide levels are increased in breast cancer tissue. Carcinogenesis 30, 745–752 (2009). 76. Hartmann, D. et al. Long chain ceramides and very long chain ceramides have opposite effects on human breast and colon cancer cell growth. Int. J. Biochem. Cell Biol. 44, 620–628 (2012). 77. Fan, S. et al. LASS2 enhances chemosensitivity of breast cancer by counteracting acidic tumor microenvironment through inhibiting activity of V-ATPase proton pump. Oncogene 32, 1682–1690 (2013). 78. Voelkel-Johnson, C., Hannun, Y. A. & El-Zawahry, A. Resistance to TRAIL is associated with defects in ceramide signaling that can be overcome by exogenous C6-ceramide without requiring down-regulation of cellular FLICE inhibitory protein. Mol. Cancer Ther. 4, 1320–1327 (2005). 79. Walker, T. et al. Sorafenib and vorinostat kill colon cancer cells by CD95-dependent and -independent mechanisms. Mol. Pharmacol. 76, 342–355 (2009). 80. Schildge, S., Bohrer, C., Beck, K. & Schachtrup, C. Isolation and culture of mouse cortical astrocytes. J Vis Exp e50079–e50079 (2013). doi:10.3791/50079 81. Vitner, E. B. et al. Induction of the type I interferon response in neurological forms of Gaucher disease. J Neuroinflammation 13, 104 (2016). 82. Tidhar, R., Sims, K., Rosenfeld-Gur, E., Shaw, W. & Futerman, A. H. A Rapid

56 Ceramide Synthase Activity using NBD-sphinganine and Solid Phase Extraction. J. Lipid Res. jlr.D052001 (2014). doi:10.1194/jlr.D052001 83. Shaner, R. L. et al. Quantitative analysis of sphingolipids for lipidomics using triple quadrupole and quadrupole linear ion trap mass spectrometers. J. Lipid Res. 50, 1692–1707 (2009). 84. Sullards, M. C., Liu, Y., Chen, Y. & Merrill, A. H. Analysis of mammalian sphingolipids by liquid chromatography tandem mass spectrometry (LC-MS/MS) and tissue imaging mass spectrometry (TIMS). Biochim. Biophys. Acta 1811, 838– 853 (2011). 85. Zigdon, H. et al. Identification of a biomarker in cerebrospinal fluid for neuronopathic forms of Gaucher disease. PLoS ONE 10, e0120194 (2015). 86. Megias, L. et al. Endocytosis and transcytosis in growing astrocytes in primary culture. Possible implications in neural development. Int. J. Dev. Biol. 44, 209–221 (2000). 87. Cartharius, K. et al. MatInspector and beyond: promoter analysis based on transcription factor binding sites. Bioinformatics 21, 2933–2942 (2005). 88. Tarcic, O. et al. RNF20 Links Histone H2B Ubiquitylation with Inflammation and Inflammation-Associated Cancer. Cell Rep 14, 1462–1476 (2016). 89. Saada, A. et. al. Evaluation of enzymatic assays and compounds affecting ATP production in mitochondrial respiratory chain complex I deficiency. Anal Biochem. 355, 66-72 (2004). 90. Chinnery, H. R. et al. Turnover of bone marrow-derived cells in the irradiated mouse cornea. Immunology 125, 541–548 (2008). 91. Stratford, S., Hoehn, K. L., Liu, F. & Summers, S. A. Regulation of insulin action by ceramide: dual mechanisms linking ceramide accumulation to the inhibition of Akt/protein kinase B. J. Biol. Chem. 279, 36608–36615 (2004). 92. Liu, Q., Turner, K. M., Alfred Yung, W. K., Chen, K. & Zhang, W. Role of AKT signaling in DNA repair and clinical response to cancer therapy. Neuro-oncology 16, 1313–1323 (2014). 93. Boucrot, E., Saffarian, S., Zhang, R. & Kirchhausen, T. Roles of AP-2 in clathrin- mediated endocytosis. PLoS ONE 5, e10597 (2010). 94. Maurer, M. E. & Cooper, J. A. The adaptor protein Dab2 sorts LDL receptors into coated pits independently of AP-2 and ARH. J. Cell. Sci. 119, 4235–4246 (2006). 95. Le Roy, C. & Wrana, J. L. Clathrin- and non-clathrin-mediated endocytic regulation

57 of cell signalling. Nat. Rev. Mol. Cell Biol. 6, 112–126 (2005). 96. Motley, A., Bright, N. A., Seaman, M. N. J. & Robinson, M. S. Clathrin-mediated endocytosis in AP-2-depleted cells. J. Cell Biol. 162, 909–918 (2003). 97. Singh, R. D., Marks, D. L. & Pagano, R. E. Using fluorescent sphingolipid analogs to study intracellular lipid trafficking. Curr Protoc Cell Biol Chapter 24, Unit 24.1– 24.1.19 (2007). 98. Yim, Y.-I. et al. Endocytosis and clathrin-uncoating defects at synapses of auxilin knockout mice. Proc. Natl. Acad. Sci. U.S.A. 107, 4412–4417 (2010). 99. Koishi, M. et al. The effects of KNK437, a novel inhibitor of heat shock protein synthesis, on the acquisition of thermotolerance in a murine transplantable tumor in vivo. Clin. Cancer Res. 7, 215–219 (2001). 100. Wilke, N., Sganga, M. W., Gayer, G. G., Hsieh, K. P. & Miles, M. F. Characterization of promoter elements mediating ethanol regulation of hsc70 gene transcription. J. Pharmacol. Exp. Ther. 292, 173–180 (2000). 101. Yasuhara, K. et al. Absence of heat shock transcription factor 1 retards the regrowth of atrophied soleus muscle in mice. J. Appl. Physiol. 111, 1142–1149 (2011). 102. Fukuda, H., Noguchi, T. & Iritani, N. Transcriptional regulation of fatty acid synthase gene and ATP citrate-lyase gene by Sp1 and Sp3 in rat hepatocytes(1). FEBS Lett. 464, 113–117 (1999). 103. Zou, X. et al. Human glycolipid transfer protein gene (GLTP) expression is regulated by Sp1 and Sp3: involvement of the bioactive sphingolipid ceramide. J. Biol. Chem. 286, 1301–1311 (2011). 104. Han, I., Kudlow, K. J. Reduced O glycosylation of Sp1 is associated with increased proteasome susceptibility. Mol Cell Biol. 17, 2550-2558 (1997). 105. Kim, S. Y., Kang, H. T., Han, J. A. & Park, S. C. The transcription factor Sp1 is responsible for aging-dependent altered nucleocytoplasmic trafficking. Aging Cell 11, 1102–1109 (2012). 106. Hsin, I.-L. et al. N-acetyl cysteine mitigates curcumin-mediated telomerase inhibition through rescuing of Sp1 reduction in A549 cells. Mutat. Res. 688, 72–77 (2010). 107. Zigdon, H. et al. Ablation of ceramide synthase 2 causes chronic oxidative stress due to disruption of the mitochondrial respiratory chain. J. Biol. Chem. 288, 4947–4956 (2013). 108. Ali, M., Saroha, A., Pewzner-Jung, Y. & Futerman, A. H. LPS-mediated septic shock

58 is augmented in ceramide synthase 2 null mice due to elevated activity of TNFα- converting enzyme. FEBS Lett. 589, 2213–2217 (2015). 109. Okayasu, I. et al. A novel method in the induction of reliable experimental acute and chronic ulcerative colitis in mice. Gastroenterology 98, 694–702 (1990). 110. Chassaing, B., Aitken, J. D., Malleshappa, M. & Vijay-Kumar, M. Dextran sulfate sodium (DSS)-induced colitis in mice. Curr Protoc Immunol 104, Unit 15.25.– 15.25.14 (2014). 111. Jensen, S. A. et al. Bcl2L13 is a ceramide synthase inhibitor in glioblastoma. Proc. Natl. Acad. Sci. U.S.A. 111, 5682–5687 (2014). 112. Davids, M. S. & Letai, A. Targeting the B-cell lymphoma/leukemia 2 family in cancer. J. Clin. Oncol. 30, 3127–3135 (2012). 113. Silva, L. C. et al. Ablation of ceramide synthase 2 strongly affects biophysical properties of membranes. J. Lipid Res. 53, 430–436 (2012). 114. Bauer, R. et al. Schlank, a member of the ceramide synthase family controls growth and body fat in Drosophila. EMBO J. 28, 3706–3716 (2009). 115. Park, W.-J. et al. Protection of a ceramide synthase 2 null mouse from drug-induced liver injury: role of gap junction dysfunction and connexin 32 mislocalization. J. Biol. Chem. 288, 30904–30916 (2013). 116. Simons, K. & van Meer, G. Lipid sorting in epithelial cells. Biochemistry 27, 6197– 6202 (1988). 117. Messa, M. et al. Epsin deficiency impairs endocytosis by stalling the actin-dependent invagination of endocytic clathrin-coated pits. Elife 3, e03311 (2014). 118. Pechstein, A. et al. Vesicle uncoating regulated by SH3-SH3 domain-mediated complex formation between endophilin and intersectin at synapses. EMBO Rep. 16, 232–239 (2015). 119. Marina-García, N. et al. Clathrin- and dynamin-dependent endocytic pathway regulates muramyl dipeptide internalization and NOD2 activation. J. Immunol. 182, 4321–4327 (2009). 120. Yu, A. et al. Protein aggregation can inhibit clathrin-mediated endocytosis by chaperone competition. Proc. Natl. Acad. Sci. U.S.A. 111, E1481–90 (2014). 121. Hirst, J. et al. Auxilin depletion causes self-assembly of clathrin into membraneless cages in vivo. Traffic 9, 1354–1371 (2008). 122. Jiang, R., Gao, B., Prasad, K., Greene, L. E. & Eisenberg, E. Hsc70 chaperones clathrin and primes it to interact with vesicle membranes. J. Biol. Chem. 275, 8439–

59 8447 (2000). 123. Dastoor, Z. & Dreyer, J. Nuclear translocation and aggregate formation of heat shock cognate protein 70 (Hsc70) in oxidative stress and apoptosis. J. Cell. Sci. 113 ( Pt 16), 2845–2854 (2000). 124. Cheng, J. & Vieira, A. Oxidative stress disrupts internalization and endocytic trafficking of transferrin in a human malignant keratinocyte line. Cell Biochem. Biophys. 45, 177–184 (2006). 125. Parry, S. N., Ellis, N., Li, Z., Maitz, P. & Witting, P. K. Myoglobin induces oxidative stress and decreases endocytosis and monolayer permissiveness in cultured kidney epithelial cells without affecting viability. Kidney Blood Press. Res. 31, 16–28 (2008). 126. Poumay, Y. & Dupal, M. F. R. Incubation of endothelial cells in a superoxide‐ generating system: Impaired low‐ density lipoprotein receptor‐ mediated endocytosis. Journal of Cellular Physiology 136, 1–8 (1988). 127. Malorni, W., Testa, U., Rainaldi, G., Tritarelli, E. & Peschle, C. Oxidative stress leads to a rapid alteration of transferrin receptor intravesicular trafficking. Exp. Cell Res. 241, 102–116 (1998). 128. Volpert, G. et. al. Oxidative stress elicited by the ceramide acyl chain length reduces the rate of clathrin-mediated endocytosis. J Cell Sci. 130, 1486-1493 (2017) 129. Erichsen, K. et al. Effects of ferrous sulphate and non-ionic iron-polymaltose complex on markers of oxidative tissue damage in patients with inflammatory bowel disease. Aliment. Pharmacol. Ther. 22, 831–838 (2005). 130. Bauer, J. et al. Lipid alterations in experimental murine colitis: role of ceramide and imipramine for matrix metalloproteinase-1 expression. PLoS ONE 4, e7197 (2009). 131. Yoshioka, K. et al. Role of natural killer T cells in the mouse colitis-associated colon cancer model. Scand. J. Immunol. 75, 16–26 (2012). 132. Mosbech, M.-B. et al. Reduced ceramide synthase 2 activity causes progressive myoclonic epilepsy. Ann Clin Transl Neurol 1, 88–98 (2014). 133. Spassieva, S. D. et al. Ectopic expression of ceramide synthase 2 in neurons suppresses neurodegeneration induced by ceramide synthase 1 deficiency. Proc. Natl. Acad. Sci. U.S.A. 113, 5928–5933 (2016). 134. Riboni, L. et al. Ceramide levels are inversely associated with malignant progression of human glial tumors. Glia 39, 105–113 (2002).

60 135. Ferreira, N. S. et. al. Regulation of very-long acyl chain ceramide synthesis by Acyl- CoA binding protein. J Biol Chem. pii: jbc.M117.785345 (2017).

136. Colombo, E. et al. Fingolimod may support neuroprotection via blockade of astrocyte nitric oxide. Ann. Neurol. 76, 325–337 (2014). 137. Ricci, G., Volpi, L., Pasquali, L., Petrozzi, L. & Siciliano, G. Astrocyte-neuron interactions in neurological disorders. J Biol Phys 35, 317–336 (2009). 138. MacRitchie, N. et al. Effect of the sphingosine kinase 1 selective inhibitor, PF-543 on arterial and cardiac remodelling in a hypoxic model of pulmonary arterial hypertension. Cell. Signal. 28, 946–955 (2016). 139. Henriksen, E. J., Diamond-Stanic, M. K. & Marchionne, E. M. Oxidative stress and the etiology of insulin resistance and type 2 diabetes. Free Radic. Biol. Med. 51, 993–999 (2011). 140. Shin, E.-J. et al. Role of oxidative stress in epileptic seizures. Neurochem. Int. 59, 122–137 (2011). 141. Suntres, Z. E., Omri, A. & Shek, P. N. Pseudomonas aeruginosa-induced lung injury: role of oxidative stress. Microb. Pathog. 32, 27–34 (2002). 142. Harrer, H., Laviad, E. L., Humpf, H. U. & Futerman, A. H. Identification of N-acyl- fumonisin B1 as new cytotoxic metabolites of fumonisin mycotoxins. Mol Nutr Food Res 57, 516–522 (2013). 143. Gault, C. R., Obeid, L. M. & Hannun, Y. A. An overview of sphingolipid metabolism: from synthesis to breakdown. Adv. Exp. Med. Biol. 688, 1–23 (2010). 144. Lahiri, S. et al. Ceramide synthesis is modulated by the sphingosine analog FTY720 via a mixture of uncompetitive and noncompetitive inhibition in an Acyl-CoA chain length-dependent manner. J. Biol. Chem. 284, 16090–16098 (2009). 145. Argraves, K. M. et al. S1P, dihydro-S1P and C24:1-ceramide levels in the HDL- containing fraction of serum inversely correlate with occurrence of ischemic heart disease. Lipids Health Dis 10, 70 (2011). 146. Muir, A., Ramachandran, S., Roelants, F. M., Timmons, G. & Thorner, J. TORC2- dependent protein kinase Ypk1 phosphorylates ceramide synthase to stimulate synthesis of complex sphingolipids. Elife 3, (2014). 147. Sridevi, P. et al. Ceramide synthase 1 is regulated by proteasomal mediated turnover. Biochim. Biophys. Acta 1793, 1218–1227 (2009).

61 10. Declaration

I hereby declare that I am the sole author of this thesis, which summarizes my independent work.

The MS analysis of the fatty acid composition of ceramide, in Fig. 6 and Fig.18 was performed in the laboratory of Professor Alfred H. Merrill (Department of Biology, Georgia Institute of Technology, Atlanta, USA).

Prediction of Sp1 binding sites in the Hsc70 promoter in Fig. 13 was performed with Dr. Shifra Ben-Dor (Bioinformatics unit, Weizmann Institute).

Chromatin immuno-precipitation assay of the binding of Sp1 to Hsc70 promoter in Fig.13 was performed by Dr. Ohad Tarcic from the laboratory of Moshe Oren (Department of Molecular Cell Biology, Weizmann Institute).

Mitochondrial complex IV and citrate synthase activities in Fig. 14 were measured by Dr. Ann Saada from Hadassah- Hebrew University

The MS analysis of the fatty acid composition of ceramide, in Fig. 26 was performed by Zora BioSience (Finland)

My contribution to each of the collaborations is outlined throughout the text.