EFFECTS OF ACSVL3 KNOCKOUT ON LIPID AND GLUCOSE METABOLISM IN MALIGNANT GLIOMA CELLS
By
Elizabeth Anne Kolar
A dissertation submitted to The Johns Hopkins University in conformity with the
requirements of the degree of Doctor of Philosophy
Baltimore, Maryland
March 2016
ABSTRACT
Gliomas are the largest category of primary central nervous system tumors.
Glioblastoma multiforme (GBM) is a World Health Organization Grade IV glioma that
comprises 70% of all gliomas. Prognosis is very poor once diagnosed, and current
treatments cannot prolong survival after relapse. Very long-chain acyl-CoA synthetase 3
(ACSVL3) is overexpressed in malignant glioma, and depleting ACSVL3 in GBM cells
(e.g. U87MG) diminishes their tumorigenic properties and affects signaling through receptor tyrosine kinases. An ACSVL3-deficient knockout (KO) U87MG cell line was generated to study how ACSVL3 contributes to the malignant properties of glioma cells.
Acyl-CoA synthetase enzyme activity was measured with long- and very long-chain fatty acids: palmitic acid (C16:0), stearic acid (C18:0), behenic acid (C22:0), and lignoceric acid (C24:0). There were significant decreases in the activation of stearic and behenic acids, while the activation of palmitic acid and lignoceric acid did not change. Ceramide synthesis assays and liquid chromatography/tandem mass spectrometry (LC/MS-MS) analysis revealed a decrease in C18:0 and C22:0 ceramides, reinforcing the enzyme activity assay results. LC/MS-MS analysis also revealed a decrease in sphingosine 1- phosphate, an important signaling molecule that affects growth and proliferation.
Proteomic analysis showed lower protein levels for enzymes involved in ceramide synthesis in the ACSVL3 KO line. Fluorescent microscopy and thin layer chromatography analyses show that ACSVL3 deficiency affects lipid rafts and ganglioside synthesis. Proteomic analysis also predicted changes in glycolysis and the tricarboxylic acid cycle (TCA) when ACSVL3 is depleted in U87MG cells. Glycolytic
ii enzymes were higher while TCA enzymes were lower in ACSVL3 KO cells.
Immunofluoresence using an antibody that detects Tom20, a mitochondrial outer membrane marker, revealed differences in mitochondrial morphology in the ACSVL3
KO cells when compared to the U87MG cells. From these studies, we conclude that
ACSVL3 is important for the synthesis of structural and signaling sphingolipids that contribute to the growth and proliferation of the GBM cells, and that this enzyme likely contributes to mitochondrial-involved carbohydrate metabolism.
Thesis Advisor: Paul A. Watkins, MD, PhD
Thesis Reader: Dan Raben, PhD
iii ACKNOWLEDGMENTS
There were a lot of people who helped me accomplish this work, whether it was helping with experiments in the lab or encouraging me from home, and I would like to acknowledge those who have made this possible.
I came to Hopkins with very specific ideas of what type of research I wanted to do, but as with many things in life, I was led in a very different direction. I never saw myself studying lipid biology, but I found my way to Paul’s lab, and I am very glad that I did. Paul, more than anyone else, knows of the struggles we have had in the lab.
Through his encouragement and expertise, I learned so much about science and about myself. I will always be grateful to Paul for all the opportunities and all that he has done.
Chapter 1 of this thesis would not have been possible without the collaborative spirit of our lab, and I will always be appreciative of the former and current lab members of the Watkins lab – Zhengtong Pei, Cicely Exeter, Xiaohai Shi, Haiyan Yang, Yanqiu
Liu, Xiaoli Ye, and Emily Clay. We got through the struggle with dead/dying/reverting cells together, solved the problem, and characterized the knockout cells as a lab. Above all, I don’t know if I would have made it through graduate school without everyone’s discussions, opinions, hard work, and above all, the laughter and fun they provided every day.
Thank you also to my thesis committee – Dan Raben, Will Wong, and Greg
Riggins. Their ideas and knowledge helped shape this thesis into what it is.
I will always cherish the amazing friends I found here at Hopkins. We came from all over the country – and sometimes from all over the world! – and graduate school
iv brought us together. I want to especially acknowledge Marcus Seldin, Michael
Multhaup, and Steven Wang for their support and friendship through the years. We have
been through a lot, whether it was changing dissertation topics or trying to survive the
city with as little damage as possible, we did it together.
I want to thank all of my aunts, uncles, and cousins for their support and love. I love you, and thank you for sticking with me through it all. You have all been my rock through this journey.
This is for my grandparents, Nicholas and Mary Kolar and Andrew and Patricia
Szoke, who mean everything to me. They were always quick with a prayer or novena, and I knew that I always, always had their support. I dedicate this to them, especially to my grandmother, Patricia, who will not be here to celebrate. I know she is always watching over me. I hope I have made her proud.
A huge thank you goes to Basil Hussain, without whom I don’t know if this would have been possible. We made it through graduate school together, and there is finally the light at the end of the tunnel. You were a pillar of support through the good and the bad, and I am excited to start our next journey. I love you.
Finally, I want to thank my parents and my brother Stephen, who always believed in me. They supported every crazy dream I ever had. Because of them, I realized I could achieve whatever I wanted, especially this. This degree was probably the hardest thing I ever had to do, and even if they didn’t always understand what I was doing and why I was so upset, they were always ones I could lean on. Thank you for everything.
v TABLE OF CONTENTS
Abstract ii
Acknowledgments iv
Table of Contents vi
List of Figures vii
List of Tables x
Introduction 1
Chapter 1
Introduction 24
Materials and Methods 26
Results 36
Discussion 52
Chapter 2
Introduction 57
Methods and Materials 61
Results 69
Discussion 92
Chapter 3
Introduction 100
Methods and Materials 104
Results 108
Discussion 130
References 133
Curriculum Vitae 144
vi LIST OF FIGURES
Figure 1. Lipid rafts are membrane microdomains enriched with cholesterol,
sphingolipids, and proteins………………………………………………………..5
Figure 2. A simplified receptor tyrosine kinase signaling cascade………………………9
Figure 3. The Warburg Effect in proliferating tissues increases lactate through aerobic
glycolysis………………………………………………………………………...12
Figure 4. Fatty acid activation reaction………………………………………………....16
Figure 5. Once activated, fatty acyl-CoA molecules can be used by cells in different
ways……………………………………………………………………………...17
Figure 6. ACS enzymes can be grouped by substrate specificity in a phylogenetic
study……………………………………………………………………………..18
Figure 7. Immunohistochemistry shows that ACSVL3 is overexpressed in glioma……21
Figure 8. ZFNs create a genomic ACSVL3 KO by deleting a 210 bp region…………..39
Figure 9. Proteomic analysis and qPCR show a decrease in ACSVL3 relative to the
U87MG cell line………………………………………………………………....40
Figure 10. ACSVL3 KO cells have an altered morphology and grow significantly
slower than the U87MG cells……………………………………………………42
Figure 11. ACSVL3 KO cells grow smaller subcutaneous tumors in nude mice………44
Figure 12. Total activation of long- and very long-chain fatty acids is lower in
ACSVL3 KO cells……………………………………………………………….49
Figure 13. The majority of ACSVL3 KO cells are in S-phase………………………….51
vii Figure 14. Total cholesterol is not different between U87MG cells and
ACSVL3 KO cells………………………………………………………………71
Figure 15. Lipid synthesis is not affected by depletion of ACSVL3 in U87MG cells…73
Figure 16. Ceramide synthesis decreases in ACSVL3 KO cells when the
substrate is C18:0……………………………………………………………….76
Figure 17. Ceramdies with acyl chains C18:0-C22:0 are decreased in ACSVL3
KO cells…………………………………………………………………………77
Figure 18. Sphingosine and S1P levels decrease when cells are depleted of
ACSVL3………………………………………………………………………..79
Figure 19. Fluorescent microscopy reveals an increase in lipid raft staining in
ACSVL3 KO cells………………………………………………………………85
Figure 20. Total gangliosides decrease in ACSVL3 KO cells, but GM1 ganglioside
increases compared to U87MG cells……………………………………………87
Figure 21. ACSVL3 KO cells do not increase lipid synthesis when incubated
with EGF………………………………………………………………………..90
Figure 22. Decreasing activation of fatty acids destined for sphingolipid synthesis
leads to aberrant signaling from receptor tyrosine kinases……………………..99
Figure 23. Glucose uptake in ACSVL3 KO cells relative to U87MG cells…………..110
Figure 24. Glycolytic pathway metabolites do not significantly change despite
a change in enzymes……………………………………………………………114
Figure 25. Pyruvate levels do not change with an ACSVL3 KO……………………...115
Figure 26. LDH enzyme levels increase in an ACSVL3 KO, but cellular lactate
levels are unchanged……………………………………………………………116
viii Figure 27. Oxidation of glucose is significantly downregulated in the
ACSVL3 KO……………………………………………………………………118
Figure 28. TCA Cycle intermediates are not changed overall in the ACSVL3 KO
cells……………………………………………………………………………..122
Figure 29. Mitochondrial morphology is different between the U87MG cell line and
the ACSVL3 KO cell line………………………………………………………128
ix LIST OF TABLES
Table 1. The relative abundance of the long- and very long-chain acyl-CoA
synthetases in U87MG cells and ACSVL3 KO cells…………………………….47
Table 2. Enyzmes involved in ceramide synthesis are decreased in the
ACSVL3 KO cells……………………………………………………………….82
Table 3. Glycolytic enzyme levels are increased in ACSVL3 KO cells………………113
Table 4. Levels of TCA cycle enzymes are decreased in ACSVL3 KO cells………....121
Table 5. The complexes of the Electron Transport Chain are lower in
the ACSVL3 KO cells………………………………………………………….124
Table 6. ATP synthase subunits levels are decreased in the ACSVL3 KO cells……...125
Table 7. Proteins involved with mitochondrial dynamics show differences in the
ACSVL3 KO cells……………………………………………………………..129
x INTRODUCTION
Glioblastoma Multiforme
Gliomas are the largest category of primary central nervous system (CNS) tumors, as they account for 78% of all primary intrinsic malignant CNS tumors (CBTRUS, 2006).
These tumors, which mainly occur in adults, are characterized based on histologic similarity to mature glial cells. The World Health Organization (WHO) grades gliomas by pathologic evaluation of the tumor. Low grade gliomas (WHO grade II) are well- differentiated and tend to exhibit benign tendencies and the patient usually has a better prognosis. High grade gliomas (WHO grades III and IV) are undifferentiated, are malignant, and carry a worse prognosis for patients. Diffuse tumors, such as astrocytomas (WHO grades II-IV), oligodendrogliomas (WHO grades II and III), and oligoastrocytomas (WHO grades II-IV), are characterized by their infiltrative properties.
Glioblastoma multiforme (GBM), the highest grade of primary brain tumor
(WHO grade IV), is the most common and fatal type. GBM is usually diagnosed without previous lower-grade tumors, and this is known as a primary or de novo GBM. It comprises 70% of all gliomas. GBM tumors grow quickly and are difficult to treat due to the cell type heterogeneity (Mineo et al., 2007). Prognosis is very poor once diagnosed.
Radiation and chemotherapy with the alkylating agent temozolomide increases survival slightly, however, most of these patients experience tumor recurrence (Sathornsumetee et al., 2007), and none of the current treatments can effectively prolong survival after
1 relapse (Weller et al., 2014). Therefore, new treatment strategies and new therapeutic targets are needed.
The U87MG cell line is a long established cell culture line in GBM research. The cells were isolated from a 44 year old Caucasian male with grade IV GBM. There have been over 1,700 publications using the U87MG cell line. It is hypodiploid with 44 chromosomes occurring in 48% of the cells with higher ploidy occurring at a rate of
5.9%. Even with the large number of chromosomal aberrations, the U87MG line has been relatively stable for years, and has not significantly changed based on genotyping by microarray (Clark et al., 2010). This cell line lacks the PTEN protein, a tumor suppressor that negatively regulates the Akt/Protein kinase B (PKB) signaling pathway. U87MG cells can successfully be used to grow subcutaneous and intracranial tumors in mice in order to study GBM in an in vivo environment.
Lipid Metabolism in Cancer
The mechanisms that integrate cell signaling and metabolism are, for the most part, conserved between normal cells and cancer cells. However, tumor cells differ through uncontrolled cell division and have a greater requirement for energy for rapid proliferation (Cantor and Sabatini, 2012). These cells often exhibit mutations so that these pathways are constitutive and allow the cells to maintain a high biosynthetic capacity (DeBerardinis et al., 2008). Over 50 years ago, it was noted that neoplastic tissues are able to synthesize lipids in a manner similar to embryonic tissue (Medes et al.,
1953). Cancer cells need increased lipid metabolism in order to meet the requirements for rapid proliferation. These lipids serve a variety of functions including a source of
2 building blocks of cellular membranes, and are required for cell signaling cascades that
encourage cellular growth and proliferation. Lipids also provide cells with a pool of acyl
groups that can be used for post-transcriptional modifications of proteins.
Cells, both normal and cancerous, require lipids from either an exogenous source, or via de novo biosynthetic pathways. The de novo synthesis of lipids is regulated by the sterol regulatory element-binding proteins (SREBPs) (Horton, 2002; Eberle, et al., 1997).
SREBPs are helix-loop-helix leucine zipper transcription factors that are translated as 125 kDa precursors that reside in the endoplasmic reticulum membrane and are bound by
SREBP cleavage-activating protein (SCAP). Once the proteins are processed through two rounds of proteolysis, the transcription factors can enter the nucleus and turn on lipid metabolism genes, such as acetyl-CoA carboxylase (ACC) and fatty acid synthase
(FASN). ACC catalyzes the rate-limiting step of de novo fatty acid (FA) synthesis, by converting acetyl-CoA to malonyl-CoA. Acetyl and malonyl groups can then be condensed by FASN to form a 4-carbon β-ketoacid, with release of CO2. Sequential
reduction, dehydration, and reduction steps, also catalyzed by FASN, convert the β-
ketoacid into a 4-carbon FA. Repeated rounds of condensation, reduction, dehydration, and reduction ultimately produce the 16-carbon long saturated FA, palmitic acid.
Kuhajda and colleagues (1996) demonstrated that a prognostic marker they had identified in breast cancer was, in fact, FASN. Since then, multiple studies have shown that tumor cells reactivate de novo lipid synthesis. Inhibition of ACC has been shown to induce growth arrest and apoptosis in prostate cancer cells (Gray-Bablin, et al., 1997).
Cholesterol is an essential component of cellular membranes. It regulates the fluidity of the lipid bilayer and also forms microdomains known as lipid rafts that
3 coordinate the activation of signaling pathways (Lingwood and Simons, 2010).
Cholesterol is also the precursor of bile acids and steroid hormones (Simons and Ikonen,
2000). Because of its importance, cholesterol’s synthesis and distribution are highly
regulated, with complex mechanisms overseeing intracellular synthesis, absorption, and
removal. Lipid rafts (Figure 1) serve as platforms for cell surface receptors, including
receptor tyrosine kinases (RTK) and G-protein coupled receptors (GPCR). These receptors can signal through a number of different intracellular pathways. One well known pathway is RTK signaling through Akt, a serine-threonine protein kinase that mediates cell survival and growth. It was recently shown in a prostate tumor model that a
reduction of membrane cholesterol levels caused the rearrangement of lipid rafts with
strong effects on Akt (Pommier et al., 2010).
4
Glycoprotein Integral protein
Glycosphingolipid
Phospholipid Cholesterol Peripheral protein
Figure 1. Lipid rafts are membrane microdomains enriched with cholesterol,
sphingolipids, and proteins. Lipid rafts are considered distinct microdomains in lipid bilayers that function as signaling platforms. These regions are distinct from other regions of bilayers due to their higher cholesterol and sphingolipid content, and are thought to be more “detergent-resistant”. Receptor tyrosine kinases are believed to mainly signal from these regions, and signaling can be influenced by different lipid content.
5 Regulation of Lipid Metabolism by Receptor Tyrosine Kinases
Receptor tyrosine kinases (RTKs, Figure 2) are important to cells for receiving
information from their environment and translating that into signaling cascades to
regulate metabolism, gene expression, cell growth, and cell cycle control (Blume-Jensen
and Hunter, 2001; Ullrich and Schlessinger, 1990). Humans are known to have 58
different RTKs. All RTKs consist of a single transmembrane domain that separates the
intracellular tyrosine kinase region from the extracellular portion (Ullrich and
Schlessinger, 1990). Ligand binding to the extracellular domain leads to conformational
changes that induce and stabilize receptor dimerization leading to kinase activity and
autophosphorylation of tyrosine residues (Greenfield et al., 1989; Ullrich and
Schlessinger, 1990; Heldin 1995). The mechanism of activation and key components of
the intracellular signaling pathways that are triggered through ligand binding are highly
conserved in evolution from C. elegans to humans, consistent with the roles they play.
Cancer cells have a wide range of mutations and deregulation of genes that lead to changes in cellular structure and function that contribute to the malignancy. Six hallmark traits of cancer have been described: self-sufficiency in growth signals, insensitivity to growth inhibitory signals, evasion of apoptosis, limitless replicative potential, induction of sustained angiogenesis, and invasion and metastasis (Hanahan and Weinberg, 2000).
Many of these traits involve RTKs and their signaling pathways. In fact, overactivation of tumor RTKs and their downstream signaling pathways is closely associated with malignant progression and poor patient survival (Birchmeier et al., 2003; Wong et al.,
1987; Simmons et al., 2001; Abounader and Laterra, 2005).
6 Two RTKs that have been implicated in the pathology of glioblastoma, as well as
other cancers, are the epidermal growth factor receptor (EGFR) and the hepatocyte
growth factor (HGF) receptor known as c-MET. EGFR overexpression is a significant
trait of GBMs, but it is rare in low-grade gliomas, suggesting a role for aberrant EGFR
signaling in the pathogenesis of GBM. RTK signal transduction continues mainly
through the protein kinase Akt/Protein Kinase B pathway or the MAPK pathway to
ultimately increase proliferation, angiogenesis, metastasis, and decreased apoptosis
(Mosesson et al., 2004). The tyrosine kinase activity may be dysregulated by gene
mutations, an increase in gene copy number, and/or EGFR protein overexpression
(Ciardiello and Tortora, 2008). Because of its role as a central regulator of biological
processes in glioma, as well as possibly contributing to resistance to apoptotic stimuli and
certain types of chemotherapy, it has attracted attention as a therapeutic target. It has
been shown that EGFR signaling through Akt promotes SREBP-1 cleavage and translocation to the nucleus to increase fatty acid concentration in GBM cells such as the
U87MG cell line (Guo et al., 2009). The same study analyzed tissue from 140 patients with primary GBMs, and demonstrated that this phenomenon found in U87MG cells was also seen in tissue, demonstrating that EGFR signaling is important to the pathology of glioblastoma.
HGF and c-MET are expressed in human gliomas with expression levels
correlating with tumor grade (Koochekpour et al., 1997; Moriyama et al., 1998; Rosen et
al., 1996). When HGF or c-MET is inhibited, in vivo tumor formation and growth is
inhibited Abounader et al., 1999, 2002). Upon binding of HGF to c-MET, the receptor
undergoes autophosphorylation and activates kinase activity, with signal transduction
7 occurring through the PI3K-Akt-mTORC pathway and the ERK/MAPK pathway. This can induce proliferative and anti-apoptotic responses in various cell types, and signaling through ERK/MAPK regulates cell proliferation, junctional competence, and motility. c-
MET mutations and overexpression have been described in a number of cancers (Ma et al., 2003; Danilkovitch-Miagkova and Zbar, 2002). Most of these mutations occur at the cytoplasmic activation loop domain.
8
Lipid Raft RTK GF
Cell membrane
Akt/PKB
Angiogenesis Protein Synthesis
Proliferation Cell Survival Lipid Synthesis
Figure 2. A simplified receptor tyrosine kinase signaling cascade. Receptor tyrosine kinases (RTKs) are proteins within the cell membrane that receive signals, usually in the form of growth factors (GF), from the extracellular environment. A number of different cascades exist, but the one most commonly upregulated in cancer is the Akt/Protein
Kinase B (PKB) signaling cascade. This is a signal for a number of different processes to occur, such as proliferation, cell survival, and lipid synthesis, all important processes for cancer cells.
9 Glucose Metabolism and Cancer
Otto Warburg (1927) made the initial observation that cancer cells were highly fermentative, and he hypothesized that this was due to some metabolic injury (Warburg,
1930). The Warburg effect describes a tumor cell’s ability to completely oxidize glucose, even in the presence of oxygen. Generally, under these conditions, cells prefer the more energetically favorable process of oxidative phosphorylation to produce ATP. Cancer cells generally have increased glucose consumption and lactate production (Figure 3).
Glycolytic enzymes are usually found to be overexpressed in cancer cells (Kim and
Dang, 2005). Even though ATP generation through aerobic glycolysis is less efficient, it facilitates uptake and incorporation of glycolytic intermediates into nucleotides, amino
acids, and lipids. This allows the cell to meet the biosynthetic demands required by
highly proliferative cancer cells (DeBerardinis et al., 2008; Heiden et al., 2009; Kroemer
and Pouyssegur, 2008; Menendez and Lupu, 2008). With more recent research into the
Warburg effect, it has been shown that this can occur, even without the defects in
oxidative metabolism (Moreno-Sanchez et al., 2007). However, the pathogenic role of
increased glycolysis in cancer cells remains controversial (Vander Heiden et al., 2009),
and discovering the true metabolism of these cells would be a significant step in
identifying more effective drugs and therapeutic targets.
Because the Warburg effect was so widely accepted, the status of mitochondria in
cancer cells was generally disregarded until recently. What has become increasingly
clear that aerobic glycolysis in cancer is a result of a more complex metabolic
rearrangement in which the mitochondria play an important role. All of these different
pathways – glycolysis, the tricarboxylic acid cycle, beta-oxidation, and anabolic
10 metabolism – adapt to respond to the uncontrolled proliferation characteristic of cancer cells by providing the cell with the energy and molecules necessary for synthesis of macromolecules required for survival (Moreno-Sanchez et al., 2009).
11
Figure 3. The Warburg Effect in proliferating tissues increases lactate through aerobic glycolysis. Cells from normal tissue (green), use glucose as a source of energy.
Under aerobic conditions, the cells produce pyruvate through glycolysis. Pyruvate can then enter the mitochondria to power the tricarboxylic acid (TCA) cycle to produce ATP efficiently through oxidative phosphorylation. Cancer cells undergo glycolysis, but produce lactate, even under aerobic conditions, with very little oxidative phosphorylation.
Figure from Goncalves et al., 2015.
12 Lipid Metabolism and Acyl-CoA Synthetases
Lipids are important macromolecules that play a variety of functions within and
outside the cell. FAs are characterized as a carboxylic acid with a long aliphatic chain,
and can be either saturated or unsaturated. They are incorporated into phospholipids that
are used to create the cell’s membrane and the membranes of the various organelles.
Phospholipids can be used as signaling molecules either intracellular or between
neighboring cells. FAs are also constituents of sphingolipids and glycolipids found in
cell membranes, and of lipid storage molecules such as triacylglycerol and cholesterol
esters. Free FAs are important to a cell because these can be oxidized by mitochondria
for energy production through beta-oxidation. Very long-chain FAs, which can be toxic to cells if present in excess, can be chain-shortened via beta-oxidation in peroxisomes; products of this process can then be further degraded by mitochondrial beta-oxidation.
Acyl-CoA synthetases (ACSs) are a family of enzymes that catalyze the thioesterification of fatty acids to Coenzyme A. This is a fundamental metabolic process that is conserved from Archaea to man (Watkins, 1997) that occurs in a two-step reaction.
First, the enzyme catalyzes the ATP-dependent formation an acyl-AMP conjugate with
the release of inorganic pyrophosphate. In the second step, the enzyme catalyzes the
formation of a thioester bond between Coenzyme A and the acyl-AMP to form the
activated fatty acyl-CoA with the release of AMP (Figure 4). Formation of acyl-CoA allows relatively unreactive fatty acids to participate in a number of metabolic pathways, such as signaling, lipid protein acylation, as well as for utilization in structural and storage lipid biosynthesis (Watkins, 2013) (Figure 5). More than 50 years ago, it was recognized that activation of short-, medium-, and long-chain fatty acids required
13 different enzymes. It is now known that there are 26 ACSs found in the human genome.
Phylogenetic analysis of their amino acid sequences has allowed classification by their
FA substrate chain-length specificity (Figure 6). Their specificity ranges from short-
(ACSS) to medium- (ACSM) to long- (ACSL) to very-long- (ACSVL) chain fatty acids.
An additional subfamily of ACSs homologous to the defective enzyme in the Drosophila melanogaster “bubblegum” mutant (ACSBG) and four “orphan” ACSs (ASCF1-4) complete the family.
The ACSVL family, also known as Fatty Acid Transport Proteins (FATPs), is conserved in many species. There are six proteins found in humans and rodents (Hirsch, et al., 1998; Watkins, et al., 1999; Steinberg, et al., 2000), eight in Zebrafish (Danio rerio), six in fruit fly (Drosophila melanogaster), and two in the roundworm
(Caenorhabditis elegans) genomes (Watkins, et al., 2007). The six human ACSVL proteins share 37-59% identity, but are only 17-25% identical to the other human ACSs.
All enzymatically active ACSs have two conserved domains in the amino acid sequence.
One is a 10-residue AMP-binding domain, known as Motif I, which is conserved in many enzymes whose reaction mechanism involves adenylation with release of inorganic pyrophosphate. The other is a 35-residue domain (Motif II) containing what was originally proposed as the fatty acid-binding “signature motif” (Watkins et al., 2007).
Both of these domains are highly conserved from archaea to humans. Sequence homology within Motif II has been used to assign the ACSs to subfamilies. Both the
ACSL enzymes and the ACSVL family are capable of activating fatty acids containing
16-18 carbons, but only the ACSVL family has the ability to activate fatty acyl substrates
14 that are 22 carbons or longer. Each ACSVL has a unique role in lipid metabolism based on tissue expression patterns, subcellular location, and substrate preferences.
15
Figure 4. Fatty acid activation reaction. The ACSs catalyze the thioesterification of Coenzyme A to a fatty acyl chain. The reaction is a two-step process. The first step requires the enzyme to bind the fatty acid and an ATP to form an acyl-AMP molecule and releases inorganic pyrophosphate. The second step forms a thioester bond between the Coenzyme A and the acyl chain to form the activated fatty acyl-CA, with the AMP being released.
16
Figure 5. Once activated, fatty acyl-CoA molecules can be used by cells in different ways. After activation, a fatty acyl-CoA can have one of many different fates. These lipids can be used to provide energy for the cell through oxidation, they can be used in the synthesis of more complex lipids, or for the modification of proteins.
17
Figure 6. ACS enzymes can be grouped by substrate specificity in a phylogenetic study. Phylogenetic analysis was performed using sequences from Motif II, the fatty acid binding motif. When grouped, the enzymes formed subfamilies based on substrate chain length. ACSS – short chain; ACSM – medium chain; ACSL – long chain; ACSVL – very long chain; ACSBG – “bubblegum” family. The ACSF enzymes are “orphan” enzymes that do not group with the other ACS subfamilies.
18 Very-Long Chain Acyl-CoA Synthetase 3 (ACSVL3) and Cancer
ACSVL3 (a.k.a SLC27A3 or FATP3) is one of the most recently characterized members of the ACS family (Pei et al., 2004). Although ACSVL family members are also known as FATPs, ACSVL3 is one of three in this family that does not transport fatty acids (Pei et al., 2004), the other two being ACSVL2 and ACSVL6 (Watkins, 2008).
Mouse ACSVL3 mRNA was found primarily in steroidogenic tissues – the adrenal, testis, and ovary – and to a lesser extent the brain, lung, and kidney (Pei, et al., 2004).
High levels of ACSVL3 mRNA can be detected in embryonic mouse brain (E12); however, levels rapidly decreased and by one month of age, the mRNA was barely detectable. Adult mouse brain sections show low levels of ACSVL3 protein by immunohistochemistry in cortical neurons, hippocampal neurons, and cerebellar Pukinje cells, but was not detected in glial cells (Pei, et al., 2004). ACSVL3 can be found in punctate vesicles via immunofluorescence in U87MG cells and mouse MA-10 Leydig cells and do not co-stain with markers for mitochondria or peroxisomes.
ACSVL3 has been found to be important in many solid malignancies (Pei et al.,
2009). Immunohistochemistry experiments using antibodies towards ACSVL3 have shown robust overexpression in different types of lung cancer (Pei, et al., 2013) and glioblastoma (Pei et al., 2009). In a tissue array of 79 gliomas of varying degrees,
ACSVL3 was overexpressed in all (Figure 7). In contrast to the adult immunohistochemistry, this protein is primarily expressed in neurons in normal human brain. The acyl-CoA synthetase ACSBG1, which is also primarily found in neurons, was used as a negative control to show that the overexpression of ACSVL3 in gliomas was specific. RNA interference (RNAi) targeting ACSVL3 in both lung cancer cell lines and
19 the U87 glioma cell line has been shown to slow cell proliferation and decrease the
number of colonies formed in soft agar (Pei et al., 2013).
In addition to our results, it has been shown that there is a higher frequency of genomic alteration at 1q21.3, which contains the SLC27A3 (ACSVL3) gene, in undifferentiated pleomorphic sarcomas (UPS) and leiomyosarcomas (LMS) (Silveira et al., 2013.). ACSVL3 mRNA was shown to be increased by qPCR, and they show that gains in 1q21.3 were an independent prognostic marker of shorter survival in LMS.
Mouse studies with A549 lung cancer cell xenografts showed decreased tumor volume when treated with fenretinide, a synthetic retinoid (Durante et al., 2014). ACSVL3 was
not shown to be a target of the compound, but by qPCR they show a reduction in
ACSVL3 levels in the treated tumors. Studies by our lab have also demonstrated that
ACSVL3 expression is enhanced with increased signaling through EGFR and c-MET
(Pei et al., 2009), and reducing ACSVL3 expression by RNAi was associated with aberrant signaling through the RTK-Akt axis. These data suggest that ACSVL3 plays an important role in the tumorigenicity of cancer cells.
20
Figure 7. Immunohistochemistry shows that ACSVL3 is overexpressed in glioma.
Pei et al. (2009) performed immunohistochemistry against ACSVL3 on an array of 79 tissues from patients with varying grades of glioma. Every grade exhibited an overexpression of ACSVL3 when compared to normal brain tissue. Another ACS,
ACSBG1, was used as a control on the same tissues, and was not found to be overexpressed in glioma.
21 Gaps in Knowledge
While we have shown that ACSVL3 is upregulated in various types of cancer, we
do not understand the consequences of this upregulation. The long chain acyl-CoA synthetases have been studied at great length in relation to disease states. Many papers investigating ACSL1 and ACSL4 have shown these enzymes are upregulated in a number of different cancers, suggesting that ACS activity plays an important role in cancer cell survival (Cao et al., 2000; Cao et al., 2001; Gassler et al., 2005). The ACS inhibitor
triacsin c was identified as an agent that shows selective cytotoxicity to malignant cancer
cells (Mashima et al., 2005), suggesting the importance of ACSs to tumor cell survival.
Very little research has been done to try to resolve the importance of ACSVL3 in cancer
cells. Knockdown of ACSVL3 using shRNA shows that while depleting the cells of the
enzyme slows down their growth, it is not required by the cells to grow.
The specific role of ACSVL3 in cellular lipid metabolism has not yet been
established is the current focus of our lab and the main subject of this Thesis. The
enzyme has a broad range of specificity between long- and very-long-chain fatty acyl-
CoAs (Pei et al., 2004) which can be used in a variety of different pathways. In addition to ACSVL3 knock-down via RNAi, I will also demonstrate that ACSVL3 is important by knocking out (KO) the gene using zinc-finger nucleases. The data presented in this
Thesis show that ACSVL3 KO U87 glioma cells exhibit the same phenotype as the
knock-down cells. The ACSVL3 KO cells exhibit slow growth, and when used to grow
subcutaneous xenografts, the injection sites either grow very small tumors compared to
the U87MG cells or they do not grow at all. Importantly, this study demonstrates for the
first time that ACSVL3 may play an important role in regulating sphingolipid
22 metabolism. We also observe differences in mitochondrial metabolism and morphology when cells are depleted of ACSVL3 that could also influence metabolic processes within the cells.
23 CHAPTER 1
Introduction
Glioblastoma multiforme (GBM) has a very poor prognosis with median survival
time of less than fifteen months after diagnosis. Recent studies indicate improved
survival with radiation and Temozolomide chemotherapy, but even then, survival is just
over one year. Numerous GBM cell lines have been established and studied over the
years. One of the most prevalent lines, the U87MG cell line, is a long established line
isolated from a GBM tumor in a 44-year old patient. It has been used in over 1,700
publications.
ACSVL3, a very-long chain acyl-CoA synthetase and our target gene, has been
shown to be overexpressed in U87MG cells, by Western blotting, and in glioma tissue
arrays, by immunohistochemistry. Our lab has done previous work using RNA
interference (RNAi) to knockdown (KD) ACSVL3 in the U87MG cells. We used two
different stable KD constructs, known as “KD3” and “KD4”, which were shRNA
sequences that targeted different regions of the mRNA (KD3:
5’CACGGCTCGCGGCGCTTTA-3’ targeted bp 394-412; KD4:
5’CGTCTATGGAGTCACTGTG- 3’ targeted bp 1861-1879). The RNAi experiments
showed that a decrease in ACSVL3 in U87MG cells results in slower cell growth,
decreases colony number in soft agar, and leads to significantly smaller xenografts in
nude mice (Pei et al., 2009). However, these stable KD cell lines proved to be sub-
optimal for investigation of the role of ACSVL3 in cancer: 1) The use of hygromycin to
keep selective pressure on the stable lines began to have adverse effects on cell
24 metabolism, and 2) the cells were found to revert after a several passages. To try to overcome these issues, we then turned to newer technology available at that time, and made knockout (KO) ACSVL3 U87MG cell lines using zinc finger nuclease (ZFN) technology.
ZFNs are fusions of the FokI restriction endonuclease with custom-designed
Cys2-His2 zinc-finger proteins (Kim and Chandrasegaran, 1994). These nucleases induce sequence-specific DNA double-strand breaks (DSBs) that can be repaired by the error-prone nonhomologous end joining (NHEJ) to yield small mutations within the targeted genomic sequence. This technology has been shown to be robust and specific within different cell types and model organisms.
After producing genomic ACSVL3 KOs, I sought to characterize them to determine whether they mimicked the properties initially observed in the RNAi KD cell lines. Here, we show that the KOs behave as predicted, in the same way as the KDs: the
ACSVL3 KO cells grew significantly slower than the wild type U87MG cells, produce smaller subcutaneous tumors, and have lower total ACS activity. We have also begun looking as changes in cell cycle as a possible mechanism underlying the slower growth of the KO cells, and preliminary results suggest that a very large percentage of the population is stalling in S phase, while the majority of the U87MG cells are found to be in G0/G1.
25 Methods and Materials
Materials and Media
The U87MG parental cell line was obtained from American Type Culture
Collection (HTB-14). For culturing and maintenance of U87MG cells and the ACSVL3
KO cell line, the cells were grown in MEM +10% FBS + 1% of the following – Sodium
Bicarbonate, Sodium Pyruvate, Non-Essential Amino Acids, and HEPES – all from
Corning Cellgro. Cells were maintained at 37C in a 5% CO2 atmosphere. Cells were
harvested for experiments (unless stated otherwise) using trypsin, then washed twice with
phosphate-buffered saline. The resulting cell pellets were resuspended in a buffer of 0.25
M sucrose, 1 mM Tris-HCl, pH 7.5, and 1 mM EDTA (STE). Protein concentrations
were measured using the assay as outlined by Lowry et al, (1951) or using the Pierce 660
nm Protein Assay (Thermo Fisher Scientific). All oligonucleotide primers used for
analysis were from IDT. [1-14C]Palmitic acid (C16:0), [1-14C]lignoceric acid (C24:0), [1-
14C]stearic acid (C18:0), and [1-14C]docosanoic acid (C22:0) were purchased from
Moravek.
Western blotting
TBS-T for Western blotting consisted of 0.15M NaCl, 0.05% Tween-20, and
0.01M Tris, pH 7.4. For sodium dodecyl sulfate-polyacrylamide gel electrophoresis
(SDS-PAGE), 5x sample loading buffer was prepared to yield a final 1x concentration of
2% SDS, 50 mM Tris, 0.2 mg/mL bromophenol blue, 5% 2-mercaptoethanol, 12.5 mM
EDTA, and 10% glycerol. Blocking and antibody dilution buffer for Western blotting
26 was 10% (w/v) dry milk in TBS-T. Protein samples were made by diluting to 2 ug/uL in
STE plus loading buffer, and boiled for 5 min. Bis-Tris acrylamide gels were prepared
by the method of Laemmli (1970) and contained 8% acrylamide/bis-acrylamide (Sigma
A3699) and 0.375M Tris-HCl; polymerization was facilitated by the addition of 0.067%
ammonium persulfate and 0.05% TEMED. 30 ug of protein per lane was loaded, and
proteins were separated at 95V. Proteins were transferred to PVDF for 3 hours in 1x
Tris-Glycine buffer with 20% MeOH. PVDF membranes were blocked with 10% milk in
TBS-T for one hour, washed with TBS-T and then incubated in anti-ACSVL3 antibody overnight (1:400 dilution). ACSVL3 antibody was raised in rabbits against the C- terminal 175 amino acids and affinity purified as previously described (Pei et al., 2006).
After incubation with primary antibody, the membrane was washed for 5 min with TBS-
T x3, after which it was incubated for one hour with goat anti-rabbit secondary antibody conjugated to horseradish peroxidase (Santa Cruz Biotechnology). The membrane was again washed TBS-T as above. Proteins were detected by chemiluminescence using the
SuperSignal West Pico reagent (Thermo).
Generation of the ACSVL3 KO in U87MG cells
The U87MG ACSVL3 KO cell line was produced using the zinc finger nuclease
(ZFN) method, using reagents designed and supplied by Sigma-Aldrich (CompoZr
Knockout Zinc Finger Nucleases). Two plasmids were obtained; pZFN1 (5’-
TGTGCTTTCCCACCCTTCTA-3’) targets the forward genomic DNA strand just upstream of Exon 2, and pZFN2 (5’-AGGTGAGGAGACTGGGAGT-3’) targets the reverse strand just downstream of Exon 2. Plasmids were co-transfected into U87MG
27 cells using FuGENE (Promega). After 48 hours, cells were harveste and counted using a
hemocytometer. Cells were plated into 96-well plates at an average density of 0.9 cells/well to produce clones generated from single cells. Clones were selected, expanded, and tested for ACSVL3 knockout by Western blot using affinity-purified ACSVL3 antibody as described above. Both genomic DNA and cDNA were prepared from clones and used for polymerase chain reaction- (PCR-) mediated amplification of the ZFN- targeted region.
Growth Curves
Cells were counted using a hemocytometer. U87MG and ACSVL3 KO cells were plated onto 6-well plates at 5,000 cells per well in U87MG media. Cells from triplicate wells were harvested and counted every other day from day 3 through day 11.
Counts were then averaged and the mean and standard deviation were calculated for each time point. Statistics were performed using a two-tailed Student’s t-test.
Acyl-CoA Synthetase Assay
Activation of fatty acids to their Coenzyme A derivatives was measured by utilizing radiolabeled palmitic acid (C16:0), stearic acid (C18:0), behenic acid
(docosanoic acid, C22:0), and lignoceric acid (C24:0). To yield a final assay concentration of 20 µM, working solutions of radiolabeled fatty acids were made in benzene such that 50 µL contains 1 nmol of [1-14C]fatty acid and 4 nmol unlabeled fatty
acid. For each sample, duplicate 50 uL aliquots of working solution were dried under
nitrogen in 13x100 mm glass tubes and fatty acids were solubilized by adding 50 µl of α-
28 cyclodextrin (10 mg/ml in 10 mM Tris-HCl, pH 8.0) and sonicating in a water-bath sonicator (Branson) for 5 min at 37C. After an additional incubation for 30 minutes in a
37C water bath with gentle shaking, respective amounts of protein (depending on the fatty acid being assayed) in 50 µL STE buffer were added to the tubes. For palmitic acid,
15 ug protein was used; for stearic acid, 30 ug protein was used; and for behenic acid and lignoceric acid, 60 ug protein was assayed. To initiate the reaction, 150 uL of a mixture containing (final concentrations) 10 mM ATP, 1 mM MgCl2, 0.2 mM CoA, and 0.2 mM dithiothreitol was added to the tubes, and the tubes were briefly vortexed. Blanks contained STE buffer alone. The samples were then incubated for 20 min in a 37C water bath with shaking. Reactions were stopped by addition of 1.25 mL modified Dole’s solution (Isopropanol: Heptane: 2N H2SO4 – 40: 10: 1).
The tubes were allowed to cool for at least 20 min at room temperature before separation of fatty acid substrate and acyl-CoA product by the method of Dole (1956).
Samples were centrifuged at room temperature for 10 min at 2,000 rpm. The supernatants were transferred to new 13x100 mm glass tubes. 0.75 mL heptane and 0.5 mL deionized water were added to each tube. The tubes were vortexed and once the two phases settled, the upper phase was discarded by aspiration. Another 0.75 mL heptane was added, tubes were vortexed, and the upper phase was aspirated; this process was repeated one additional time. After a final wash with 0.75 mL heptane and a 1 min centrifugation to fully separate the phases, the upper phase was aspirated and the lower phase was transferred to small scintillation vials. After addition of 5 mL Budget Solve
(RPI), radioactivity (disintegrations per minute (dpm)) in each sample was determined by liquid scintillation counting.
29 Results from duplicate samples were averaged, and appropriate blanks subtracted.
An aliquot of working solution was counted to calculate the specific activity (dpm/nmol)
for each substrate. Using the known protein concentrations for each sample and the
calculated specific activity, results were ultimately expressed as nmol/20 min/mg protein.
Standard deviation was determined and p-values were calculated using a two-tailed t-test.
Xenograft Studies
All animal protocols were approved by the Johns Hopkins Animal Care and Use
Committee. In vivo tumorigenesis of U87MG and ACSVL3 KO cells was assessed in 4-
to 6-week old female mice as previously described (Laterra et al., 1997; Kim et al.,
2006). NIH III Xid/Beige/Nude mice (National Cancer Institute, Frederick, MD) were injected with 4 x 106 cells in 0.1 mL of PBS in the dorsal areas. Tumor size was
measured using calipers, and volume was estimated by the following formula: volume =
(length x width2) / 2. Mice that were injected with U87MG cells were sacrificed 15 days
after injection, and those that were injected with ACSVL3 KO cells were sacrificed on
day 19. The xenografts were harvested and then weighed.
Proteomic Profiling of U87MG and the ACSVL3 KO cells
Sample preparation. Cells for proteomic profiling were cultured as biological
replicates, harvested by gentle trypsinization, and transported to the laboratory of Dr.
Akhilesh Pandey (Johns Hopkins University School of Medicine). Further preparation
and sample analysis was performed by Pandey lab personnel as follows. 4% SDS lysis
buffer (4% SDS, 100mM Tris pH 8.0) was added to the cell pellet and was heated at 95C
30 for five minutes to destroy protease activity. After heating, tubes were kept at room temperature. Samples were sonicated using a probe sonicator operating in a continuous mode at 20 Hz output for three 15 sec cycles. Samples were heated at 95̊C for another 10 minutes to ensure complete solubility, followed by centrifugation at 17,000 g for 15 minutes. The cleared lysate was collected and protein was measured using the
Bicinchoninic acid (BCA) assay (Pierce, Waltham, MA). Samples were stored at -80C until further analysis.
Protein digestion and tandem mass tag (TMT) labeling. 200 ug of protein lysate was reduced with 10mM DTT and incubated at 60C for 20 minutes. A 30 kDa molecular weight cut-off filter (Millipore, USA) was used to remove SDS and to exchange the buffer with 8M Urea in 50mM triethylammonium bicarbonate (TEAB) buffer by centrifugation at 14,000 x g for 15 minutes at room temperature. Reduced proteins were alkylated by adding 20mM iodoacetic acid and incubated in the dark at room temperature for another 30 minutes. Buffer exchange with urea was continued for another 4 cycles, after which the urea was depleted using 2 cycles of 50mM TEAB buffer alone. The final retentate was reconstituted in 200 ul of 50mM TEAB buffer and protein was measured using the BCA assay. Equal amounts of protein from each sample were digested with trypsin (Promega) in 50mM TEAB for 16h at 37C at an enzyme to substrate ratio of 1:50.
The digested peptides were purified using Sep-Pak® C18 cartridges (Waters, USA), lyophilized, and subjected to tandem mass tag (TMT) labeling according to the manufacturer’s instructions (Thermo Fisher, Pierce Scientific). After TMT labeling, samples were pooled, vacuum dried, and fractionated by high pH reversed-phase liquid chromatography (bRPLC) on an Agilent 1100 LC system. A total of 96 fractions were
31 collected and concatenated to 24 fractions. These fractions were then vacuum dried and
were subjected to LC-MS/MS analysis
LC-MS/MS analysis of enriched peptides. A total of 12 bRPLC fractions were
analyzed on an Orbitrap Elite mass spectrometer (Thermo Electron, Bremen, Germany)
interfaced with Easy-nLC II nanoflow liquid chromatography system (Thermo
Scientific, Odense, Denmark). The peptide digests were reconstituted in 0.1% formic acid
and loaded onto a trap column (75 µm x 2 cm) packed with Magic C18 AQ (Michrom
Bioresources, Inc., Auburn, CA, USA). Peptides were resolved on an analytical column
(75 µm x 30 cm) at a flow rate of 300 nL/min using a linear gradient of 10-35% solvent B
(0.1% formic acid in 95% acetonitrile) over 90 min. The total run time including sample loading and column reconditioning was 120 min. The Orbitrap Elite mass spectrometer was operated in a data dependent mode at High-High Orbitrap mode. The full scans in the range of m/z 350-1800 was used with 1.6 m/z isolation width and were measured using an Orbitrap mass analyzer at a mass resolution of 120,000 at m/z 400. The top 15 precursor ions were selected and were fragmented using normalized higher energy collisional dissociation (HCD) at 37% and measured using Orbitrap mass analyzer at a resolution of 30,000 at m/z 400. The ion filling times and AGC targets were set at 100 ms and 1X106 for full scans and 5X104 and 200ms for MS2 and MS3 scans respectively.
Dynamic exclusion was enabled and repeat duration was set for 45 seconds. Isobaric mass tag m/z 230.14 (TMT) was excluded for MS3 mode. Internal calibration was carried out using lock mass option (m/z 445.1200025) from ambient air.
Mass spectrometry data analysis. A total of 24 LC-MS/MS raw files that were acquired as technical runs were used for mass spectrometry data analysis for the peptide
32 identification and quantification. The raw data was processed through the Proteome
Discoverer 2.1.0.81 software suite to generate peak list files for the database searches.
Data was searched against the human Refseq 73 protein database. Combined Sequest and
Mascot search algorithms were used for the peptide identification and reporter ion quantifier node was enabled for peptide quantification. A dual processing workflow and consensus workflows were created with the following parameters. a) minimum and maximum precursor were selected as 350 Da and 8000 Da respectively; b) trypsin is selected as protease and a maximum of 2 missed cleavages were allowed; c) precursor and fragment mass tolerance were set as 20 ppm and 0.1 Da respectively; d) oxidation of methionine residue as static modification and carbamidomethylation of cysteine, peptide
N-terminus and lysine side chain as TMT reported tag were selected as variable modifications. Percolator node was used for calculating p values of identified PSMs and peptides for statistical significance. 1% protein level and peptide level FDR was used in the consensus workflow. Peptide quantification was carried out using reporter ion quantifier and isolation interference cut-off was set as 30% in order to account for the interference from co-eluting peptides. Only peptides that fell below this criterion were considered for peptide quantification.
Real-time PCR and Conventional PCR of ACSVL3 cDNA
RNA was extracted from U87MG and ACSVL3 KO cells using TRIzol (Thermo
Fisher Scientific). RNA concentrations were measured using a NanoDrop spectrophotometer, and a total of 3 ug of RNA per sample was used to synthesize cDNA using SuperScript III Reverse Transcriptase (Thermo Fisher Scientific) as described in
33 the manual. Reactions included 7.5 uL of SYBR Green Master Mix (Thermo Fisher
Scientific), 1 uL each of Forward and Reverse primers, 5 uL cDNA, and 0.5 uL dH2O.
The primers used for qPCR are as follows – Fwd: 5’-CAATGCCAGGGCTTCTATC-3’;
Rev: 5’-GGGCGAGGTAGATCACAT-3’. Primers were designed for RPS5 cDNA as
the normalizer – Fwd: 5’-AAAGCTCAGTGTCCCATTG-3’; Rev: 5’-
CAGGTGTATGATCTCGAAGG-3’. The reactions ran for 35 cycles in a Bio-Rad
CFX96 Real Time PCR Detection System with a melting temperature of 95̊C for 30
seconds, an annealing temperature of 55̊C for 15 seconds, and extension at 72̊C time of
15 seconds for each cycle.
Conventional PCR of ACSVL3 cDNA from U87MG and ACSVL3 KO cells was
performed in a T100 Thermal Cycler (Bio-Rad). Samples amplified over 35 cycles using
Phusion High Fidelity Polymerase (New England Biosciences). For each cycle the
following temperatures and times were used: initial melting occurred at 98̊C for 30 sec;
melting 98̊C for 8 sec; primers annealed at 58̊C for 20 sec; extension at 72̊C for 18 sec.
Final extension was at 72̊C for 8 min. ACSVL3 ZFN deletion forward primer sequence;
5’-CCTGCTGGAATTAGCGATT-3’. ACSVL3 ZFN deletion reverse primer sequence;
5’ –TGCCAGAGGTGAAGATGTA-3’
Cell Cycle Assays
Cells were harvested from 10 cm dishes using trypsin. Cells were counted using a
hemocytometer. A total of ~1 x 106 cells were used for each sample. Samples were labeled and processed as described by the Muse Cell Cycle Assay kit (Millipore). The cells were transferred to a 15 mL conical tube and were centrifuged at 300 x g for 5 min.
The supernatant was removed without disturbing the pellet. 1 mL of PBS per 1 x 106
34 cells was added and the cells were mixed well by pipetting several times. Samples were
centrifuged at 300 x g for 5 min. The supernatant was discarded, leaving ~50 uL per 1 x
106 cells. The pellet was resuspended in the PBS by pipetting several times. The cells
were resuspended drop-wise into a tube containing 1 mL of ice cold 70% ethanol while
vortexting at medium speed. The tubes were capped and kept at -20̊C for 72 hours before staining.
To stain with the Muse Cell Cycle Reagent (with propidium iodide), 200 uL of the ethanol-fixed cells were transferred to a 1.5 mL microcentrifuge tube. The samples were centrifuged at 300 x g for 5 min at room temperature. The supernatant was then discarded. The cell pellet was resuspended in 250 uL PBS, and this was centrifuged again at 300 x g for 5 min at room tempertature. The supernatant was removed. The cells were resuspended in 200 uL of the Muse Cell Cycle Reagent and incubated for 30 min, protected from light. Gating was adjusted once the first sample was run. The Muse was set to acquire 10,000 events per sample.
35 Results
Genomic Knockout (KO) of ACSVL3 in U87MG Cells Using Zinc Finger Nucleases
(ZFNs)
ACSVL3 is overexpressed in malignant gliomas, as well as lung and prostate
tumors (Pei et al., 2009, 2013) (Z Pei and PA Watkins, unpublished). Using the U87MG
glioma cell line as a model system, our lab previously showed that depleting ACSVL3 by
RNA interference decreased the malignant properties of these cells, both in culture and
when used to produce xenografts in mice. To extend these observations, and to
investigate the mechanisms underlying ACSVL3’s role in cancer, zinc-finger nuclease
(ZFN) reagents were obtained from Sigma-Aldrich and used to create a genomic KO of
ACSVL3 in U87MG cells. Our lab also sent the parental U87MG line to the Genetic
Resources Core Facility (Johns Hopkins School of Medicine, Institute of Genetic
Medicine) to authenticate our cell line by short tandem repeats, and found that the cells were indeed U87MG cells.
To characterize ACSVL3 KO cells, we wanted to 1) identify the mutation site within the gene and 2) make sure that it was a KO at both the genetic and protein level.
Genomic DNA was extracted from both the U87MG cells and the newly made KO cells, and was sequenced using primers upstream and downstream of the predicted KO site.
Sequencing showed that in the KO cells, there was an in-frame 210 bp deletion of Exon
2. More importantly, more than half of the nucleotides (19 out of 30) which encode
Motif I, the AMP binding domain of ACSVL3, were deleted (Figure 8A). Previous studies of Motif I of bacterial (Escherichia coli) and simple eukaryotic (Sacchromyces
36 cerevisiae) ACSs showed that mutation of any of the residues within this conserved
sequence abolishes or severely reduces catalytic activity (Weimer et al., 2002; Zou et al.,
2002).
In addition to sequencing, we also performed PCR of the deletion region using primers that were within the deletion region, with a predicted product size of 103 bp.
RNA was extracted from wild-type and KO cells and used to synthesize cDNA that was used as template for the PCR. The product was run on a 1% agarose 1x TBE gel and visualized using ethidium bromide. The results show that the region was indeed deleted in the proposed ACSVL3 KO, and that a specific band was amplified in the U87MG wild-type of the correct predicted size (Figure 8B).
To see if there was less ACSVL3 protein as a result of the ZFN deletion, we performed Western Blotting with a custom antibody that recognizes an epitope within the
C-terminal 175 amino acids of wild-type ACSVL3. 30 ug of protein was loaded into each well of an 8% polyacrylamide gel. The protein was transferred to a PVDF membrane and was probed with rabbit ACSVL3 antibody overnight (1:400 in 10% milk
+ TBS-T). After incubation with anti-rabbit secondary antibody and exposure to film, we find a very strong band in the U87MG cells of the correct molecular weight (~78 kDa)
(Figure 8C). Surprisingly, we found a weaker band around the same size in the ACSVL3
KO cell line. Another potential KO candidate, clone 27, was also included on the membrane to see if it was also an ACSVL3 KO cell line. This potential KO line had a 9 bp deletion that also abolished part of Motif I. While we find a weaker putative ACSVL3 band in this candidate cell line, we later discovered through sequencing that this was a heterozygous cell line, and was excluded from further investigation.
37 A proteomics study of the ACSVL3 KO and the U87MG cells was performed in
collaboration with Akhilesh Pandey, Ph.D. at Johns Hopkins University. First, we
wanted to see if the proteomics data agreed with our Western Blotting data showing that
there was lower ACSVL3 protein. Only two peptides, both N-terminal to the ZFN deletion site, were detected, and the data does show that there is a decrease in ACSVL3 protein levels in the KO compared to the U87MG cells (Figure 9A). However, like the
Western blotting result, we still do see peptides derived from the ACSVL3 protein. We believe that we may be seeing read-through of the transcript due to the in-frame deletion, and the lower levels of protein could be because the deletion makes the protein unstable.
Real-time PCR was also performed using SYBR green and primers specific for ACSVL3 cDNA to see if there was also a lower level of transcripts. A total of 5 samples per cell line were assayed, and there is a much lower level of ACSVL3 transcripts in the
ACSVL3 KO cell line (Figure 9B), which can also contribute to the lower levels of protein. Taken together, these data indicate that the ACSVL3 KO cell line can be used to investigate the metabolic role of this enzyme in cancer.
38
Figure 8. ZFNs create a genomic ACSVL3 KO by deleting a 210 bp region. (A)
Sequencing of both the genomic DNA and the mRNA (by synthesizing cDNA) of the
putative ACSVL3 KO cell line revealed an in-frame 210 bp region that was deleted in both the DNA and the mRNA correlating with most of Exon 2, including most of Motif I.
(B PCR amplification using primers specific for the deletion region show that DNA is
amplified in the U87MG cells, and not the ACSVL3 KO line). (C) Western blotting
shows a significant decrease in protein levels of ACSVL3 in the KO line and another
potential KO line, “#27”. While we detect a small amount of protein, it is believed to be
catalytically dead.
39
Figure 9. Proteomic analysis and qPCR show a decrease in ACSVL3 relative to the
U87MG cell line. (A) Proteomic analysis shows a 50% decrease in ACSVL3 peptide
levels in the ACSVL3 KO line. We believe that these peptides are indicative of read- through of transcript that is being produced, and that the product is catalytically dead. A total of two peptides were found during the analysis that were both N-terminal of the
ZFN deletion site. (B) qPCR revealed a marked decrease in ACSVL3 transcripts compared to the U87MG line.
40 ACSVL3 KO in U87MG Cells Causes a Change in Morphology and a Slower
Growth Rate
During the characterization of the ACSVL3 KO cells, I observed that the cell
morphology was different when compared to the U87MG cell line (Figure 10A). When
ACSVL3 is absent from U87MG cells, the cells take on a flatter morphology, and appear
take up much more surface area.
To determine whether the ACSVL3 KO cells exhibit the same growth defect that
was found in the stable KD cell lines, I performed a growth curve. The ACSVL3 KO cell
line and the U87MG cell line were both seeded at 5,000 cells per well in a 6-well cell culture dish in regular U87 culture media. Starting three days after seeding, triplicate wells of cells were harvested and counted every other day for 11 days (Figure 10B). The growth rate of ACSVL3 KO cells was significantly slower than U87MG cells. On day
11, the rate for KO cells was 12% of U87MG cells, a statistically significant difference
(p<0.005). This result was in agreement with that previously seen in ACSVL3 KD cells when compared to the U87MG cells (Pei, et al., 2009).
41
Figure 10. ACSVL3 KO cells have an altered morphology and grow significantly slower than the U87MG cells. (A) Light microscopy reveals differences in morphology between the U87MG cells (left) and the ACSVL3 KO cells (right). The
ACSVL3 KO cells appear larger, flatter, and take up more surface area. (B) 5,000 cells were seeded in each well on Day 0. Cells were counted every other day starting from
Day 3. When the assay ended on Day 11, the ACSVL3 KO cells showed significantly slower growth (p = 0.0003) and grew at rate that was 12% of the U87MG growth rate.
42 ACSVL3 KO Xenografts are Significantly Smaller than U87MG Xenografts
Previous studies demonstrated that the ACSVL3 KD cell lines had a gross effect
on glioma cell tumorigenicity in vivo (Pei et al, 2009). The KD cell lines exhibited
slower tumor growth rates over time and fewer tumors formed at injection sites when
compare to the U87MG cells. To test if the newly derived ACSVL3 KO cells grow in the
same manner as the ACSVL3 KD cells in vivo, nude mice were subcutaneously injected with 4 x 106 U87MG or ACSVL3 KO cells. For each cell type, 5 mice were injected in
both flanks. After one week of growth, the tumors were measured with calipers every
other day for width and length (Figure 11C). In mice bearing U87MG xenografts, tumors
became quite large (Figure 11A, B), necessitating sacrifice on day 15 for humane
reasons; mice injected with KO cells were sacrificed on day 19. Xenografts formed at all
10 sites in mice injected with U87MG cells. In contrast, tumors failed to form at 3 out of
10 sites injected with KO cells. After xenograft harvesting, the tumors were weighed.
The U87MG xenografts weighed significantly more than the ACSVL3 KO xenografts
(Figure 11D). These results corroborate the results that were obtained with the stable
ACSVL3 KD cell lines, and demonstrate that ACSVL3 is important for in vivo tumor
growth.
43
Figure 11. ACSVL3 KO cells grow smaller subcutaneous tumors in nude mice.
Cells of both U87MG and ACSVL3 were injected into nude mice (4 x 106 cells/injection site). (A) Subcutaneous tumors in the nude mice on Day 15 after injection. The
ACSVL3 tumors (top) are much smaller than the U87MG tumors (bottom). (B) Tumors of U87MG cells were harvested 15 days after injection (left) and the ACSVL3 KO tumors (right) were harvested on Day 19. Representative tumors are shown. (C) Tumor size was measured using calipers throughout the study, with the U87MG tumors growing significantly larger than the KO. (D) After harvesting, the tumors were weighed. The
U87MG tumors weighed about 6x more than the ACSVL3 KO tumors.
44 The Acyl-CoA Synthetase Profile of the ACSVL3 KO line
We also analyzed the proteomics data generated by the Pandey lab to see if there
was an overall change to the acyl-CoA synthetase profile of the ACSVL3 KO line. In
total, we were able to detect 7 long- and very-long chain ACSs – ACSL3, ACSVL4
(SLC27A4), ACSL1, ACSVL5 (SLC27A1), ACSL4, ACSVL3 (SLC27A3), and ACSL5
(Table 1). In both of these cell lines, we find that ACSL3 was the most abundant ACS,
and the value found in the U87MG cells was set at 1.00 for comparison of the other
ACSs, and between the U87MG cells and the ACSVL3 KO cells. ACSL3 is slightly
increased in the ACSVL3 KO line (Table 1). This has been observed in other cancers;
microarray analysis of melanoma cells shows that there is an upregulation of ACSL3
(Sumantran et al., 2015), however since the upregulation is occurring in the KO cells
with the growth defect, we do not think that this particular enzyme is important to
glioma. We do see slight decreases in the other long- and very long-chain ACSs, except
for ACSL5. ACSL5 has been shown to be a pro-apoptotic gene and is elevated with
DHX32, a gene implicated in proliferation and cell growth. It was depleted in colorectal cancer cells (Lin et al., 2015). Because it did not change with the depletion of ACSVL3 in U87MG cells, it suggests that changes we see in growth and tumorigenicity are because of changes in ACSVL3, and not because of other ACSs that have been implicated in cancer.
We also analyzed proteomic data from primary astrocytes, obtained in a separate experiment, from the Pandey lab. While we were not able to directly compare these data to data obtained with U87MG and KO cells, we used them here as a reference.
Interestingly, normal astrocytes expressed the same long- and very long-chain ACSs
45 found in U87MG and KO cells. Again, the major ACS was ACSL3, and this was set to
1.00 and we compared the long- and very long-chain ACSs to that. We found that
ACSVL3 was expressed at an extremely low level (0.04 relative to ACSL3) in normal astrocytes, compared to 0.38 relative to ACSL3 in U87MG cells. The low level in normal astrocytes also correlates with the IHC data for normal brain from the glioma tissue array shown in Figure 7 (Pei et al., 2009). ACSL5 was found to be below the level of detection in the primary astrocytes.
46
Table 1. The relative abundance of the long- and very long-chain acyl-CoA synthetases in U87MG cells and ACSVL3 KO cells. Proteomic analysis was performed on both cell lines. Proteomic data from primary astrocytes from a separate experiment was used as a reference. ACSL3 was the most abundant ACS found in
U87MG cells, and was set at 1.00. All other ACS values are relative to ACSL3. We see slight decreases in most of the ACSs in the ACSVL3 KO cells, except for ACSL3, which is increased.
47 ACSVL3 KO Cells Have Lower Activation of Long and Very-Long Chain Fatty
Acids
To investigate the effects lower ACSVL3 levels have on the activation of fatty
acids, we looked at the acyl-CoA synthetase enzyme activity using several long and very-
long chain fatty acid substrates. The activation of palmitate (C16:0), stearic acid (C18:0),
behenic acid (C22:0), and lignoceric acid (C24:0) was measured by incubating cell
suspensions with solubilized [1-14C]-labeled fatty acid and cofactors, including
Coenzyme A. The fatty acid substrates were separated from the acyl-CoA products using
Dole’s method, and radioactivity (dpm) in the latter was measured per sample using a
liquid scintillation counter; results were expressed as nmol/20 min/mg protein for final
analysis. When measuring enzyme activity using radiolabeled C16:0 and C24:0,
substrates that the lab had used before as representative long- and very long-chain fatty acids, there is a slight decrease in the KO cells, however this was not statistically significant (Figure 12A, D). The enzyme activity assay was then repeated with C18:0 and C22:0, and with these substrates we see a very significant decrease when ACSVL3 is depleted from cells (Figure 12B, C). The relevance of the significant decreases in activation of C18:0 and C22:0 vs. C16:0 and C24:0 will be discussed further in Chapter
2.
48
Figure 12. Total Activation of long- and very long-chain fatty acids is lower in
ACSVL3 KO cells. A total of 4 samples were assayed for each [1-14C]-labeled fatty
acid. C16:0 palmitic acid (A) and C18:0 stearic acid (B) are considered long-chain fatty acids, with C22:0 behenic acid (C) and C24:0 lignoceric acid (D) are very long-chain fatty acids. There was a slight decrease in the activation of C16:0 and C24:0, but the difference was not statistically significant. Significant decreases (p<0.001) were observed when C18:0 and C22:0 was used as substrates, with lower activation in the
ACSVL3 KO cells.
49 U87MG Cells and ACSVL3 KO Cells Proceed Through the Cell Cycle Differently
We have been exploring the factors responsible for the growth phenotype of the
ACSVL3 KO cells. There could be several reasons for the differences in growth we have observed, including changes in apoptosis, autophagy, cell cycle, or metabolism. Several indicies of apoptosis or autophagy were measured. Those studies concluded that the
ACSVL3 KO cell line did not have increased rates of either apoptosis or autophagy when compared to the U87MG cells (Watkins et al., unpublished). To assess if there were differences in cell cycle between our two cell lines that can explain the growth phenotype, I used the Muse Cell Cycle Assay (Millipore) This assay relies on the difference in binding and increasing fluorescence of propidium iodide to the DNA and can differentiate between the different phases of the cell cycle. The majority of U87MG cells appear to be in the G0/G1 phase, with an average of 71.1±3.8% (mean ± SD) of the population in that phase. Only 7.3±0.5% of the U87MG cells appeared to be in S phase, and 9.3±2.0% were in G2/M. The ACSVL3 KO cells exhibited a very different profile.
A very small percentage of the cells – 0.3% – were in G0/G1. The majority of the KO cells were found to be in S-phase, at 66.2±0.9%, and 9.3±0.2% were in G2/M (Figure
13). Further studies will be necessary to understand this result. We cannot conclude if the ACSVL3 KO cells are proceeding through S-phase slowly, or if there is a block at a checkpoint that is preventing a large population of the cells to enter into G2.
50
Figure 13. The majority of ACSVL3 KO cells are in S-phase. The Muse Cell Cycle
Assay was used, which relies on propidium iodide, to determine cell cycle status of the population of both U87MG and ACSVL3 KO cell lines. A total of 10,000 events were recorded. We found that the majority of the U87MG cells are in G0/G1 (71%), with a small percentage (7.3%) being in S phase. Surprisingly, the vast majority of the
ACSVL3 KO cells were in S phase (66%), with hardly any being in G0/G1 (0.3%).
51 Discussion
Acyl-CoA synthetases play a central role in fatty acid metabolism by “activating”
the fatty acids with the addition of a Coenzyme A. These activated fatty acyl-CoAs can be utilized by the cell in a number of metabolic pathways. They can be used as an energy source via β-oxidation in the mitochondria, provide substrates for synthesis of phospholipids and sphingolipids, or can be used to acylate proteins. Because of the varying lengths of fatty acids and the diverse nature of their metabolic fates, the human genome contains 26 different acyl-CoA synthetases, specific for varying substrate chain lengths.
Our lab has found that one of the very long-chain acyl-CoA synthetases,
ACSVL3, is overexpressed in a number of different cancers, including WHO grade IV glioblastoma multiforme, through a tissue array of 79 patient tissues of varying grades of glioma. To investigate the importance of ACSVL3 to tumorigenesis in glioma, the lab made a stable knockdown (KD) construct in the U87MG cell line. This cell line demonstrated that acyl-CoA synthetase activity was lower, cell growth was slower, and xenografts in nude mice were smaller when compared to the parental U87MG line. We wanted to further investigate the effects of glioma cells lacking ACSVL3, and created a genomic knockout (KO) cell line using zinc-finger nucleases (ZFNs).
The deletion region created by the ZFNs was characterized by genomic DNA sequencing. We determined that the deletion was 210 bp of genomic DNA that deleted most of Exon 2, which included most of Motif I, the AMP-binding site. It was also determined that the deletion was also in-frame. This site is highly conserved and is
52 critical for enzyme activity, as mutagenesis of any part of the motif abolishes enzymatic
activity. We were able to confirm the sequencing results through PCR using primers
specific for the deletion region. We were able to amplify product in the U87MG cell line
while we were not able to detect a band in the ACSVL3 KO line. We wanted to see if we
could detect any protein via Western Blotting using a custom antibody raised against the
C-terminal region of ACSVL3. We found that ACSVL3 is present in U87MG cells, but
that the protein levels are nearly zero in the ACSVL3 KO band. While some ACSVL3
protein is present, we believe that it is enzymatically dead.
Characterization of the ACSVL3 KO cell line shows that it shares similar
characteristics as the ACSVL3 KD line, making the KO line a viable model to further
study the mechanistic importance of ACSVL3 in glioma. When glioma cells are depleted
of ACSVL3, as proven by DNA and protein analysis, we find that cell growth is
significantly slower over an 11 day period. To show that it was an ACSVL3-specific
growth defect, Pei, et al. also used shRNA to deplete cells two other acyl-CoA synthetases, which are both expressed in U87MG cells. FATP4, another very long-chain acyl-CoA synthetase (Watkins, 2008), and ACSF2, which activates medium-chain fatty acids (Watkins, et al., 2007) were depleted in U87MG cells, and growth rate was measured in the same manner. They showed that knockdown of either FATP4 or ACSF2 did not affect growth over the same time period as ACSVL3 affected growth of U87MG cells.
We wanted to explore the growth defect further to determine the cause of the slower growth observed in the ACSVL3 KO. Previous unpublished work from our lab looked at potential differences in apoptosis or autophagy as possible explanations for this
53 difference we see between our two cell lines. The study concluded that neither pathway
could explain the growth differences. The other possibility was a difference in cell cycle
progression when ACSVL3 is absent in U87MG cells. We used the Muse Cell Cycle
Assay kit from Millipore to assess if there were any differences. This approach uses
propidium iodide, a DNA intercalating molecule, to detect DNA content differences. We
found that while most of the U87MG population was in G0/G1, the vast majority of the
ACSVL3 KO cells were in S phase. A typical rapidly proliferating human cell stays in S
phase for about 8 hours, G2 for about 4 hours, M/cytokinesis for about 1 hour, and the new cells enter G1 for about 11 hours (Cooper, 2009). We do not have enough information to know if the ACSVL3 KO cells are arresting in S phase, or if they are proceeding through S phase slower than the U87MG cells. This will be investigated
further to understand the effects that ACSVL3 has on the cell cycle.
We also show that the ACSVL3 KO cell line either forms very small
subcutaneous xenografts, or do not form tumors at all at the injection site. The
xenografts that do form are significantly smaller in size and lower in weight compared to those formed using U87MG cells. This characterization, along with the similar effects of the ACSVL3 siRNA KD lines, also demonstrates that ACSVL3 is not necessary for growth of wild type cells, but it is essential for the malignant properties exhibited by the
U87MG cells.
After we had determined that the ACSVL3 KO cell line affected the growth properties and tumorigenic properties of the parental U87MG cells, we wanted to look at the ACS profile of the two cell lines. Proteomic analysis of the U87MG line and our
ACSVL3 KO line show that there were 7 out of 11 long- and very-long chain ACSs
54 identified. We also obtained data from primary astrocytes from another experiment to
use as a reference, but not a direct comparison. These data were useful in determining
whether deleting functional ACSVL3 protein would affect the expression of the other
ACSs found in U87MG cells. We found that ACSL3 actually increased in the ACSVL3
KO cells, but another ACS known to be important in cancer, ASCL5, was found to be in
the same abundance across the two different cell lines.
At this point, we do not know if there are off target-effects and if this is the cause of any decreases we do see in the KO line in the other ACSs found. It could be due to the decrease in products of ACSVL3 or of the protein itself that may be affecting the expression of the other enzymes. Not much is known about the transcriptional and translational regulation of the acyl-CoA synthetases. Some data has been published suggesting that the SREBPs are involved in transcription, in particular for ACSL1 (Singh et al., 2016). The literature also shows PPARs are involved in regulating transcription of
ACSL4 (Kan et al., 2015).
We also tested the overall activation of long- and very long-chain fatty acids in
the U87MG cells and the ACSVL3 KO cells. We found an expected decrease in KO cell
enzyme activity with each of the [1-14C]labeled fatty acids we used. There was a slight
decrease in activation when we used C16:0 and C24:0 fatty acids, which are commonly
used by our lab as representative long- and very long-chain fatty acids. We also used
C18:0 and C22:0, and surprisingly, we found greater, statistically significant decreases in the ACSVL3 KO cells with both fatty acids. The long- and very long-chain ACSs have great substrate overlap in vitro, and so we expected to see decreases in KO cells with all of these fatty acids. However, we believe the significant decreases in C18:0 and C22:0
55 in the ACSVL3 KO cells indicate a greater specificity for these fatty acids over C16:0
and C24:0. This proved to be an important consideration when choosing substrates for
experiments that will be described in Chapter 2.
Now that we have established a viable way to study the effects of ACSVL3 in a
glioma cell model, we can ask why the cells behave the way they do when the cells are
depleted of ACSVL3. The physiological purpose of ACSVL3 overexpression has not
been resolved. The products of ACSVL3 are long- and very long-chain fatty acyl-CoAs, which can be used by the cell in a variety of different reactions, including β-oxidation for energy production, protein acylation, and incorporation into complex lipids. Research into the importance of the ACSs in regards to cancer, particularly the long- and very long-chain acyl-CoA synthetases is relatively new, and even then, most of the publications focus on the ACSL family of enzymes. Our lab was the first to show that
ACSVL3 is overexpressed in a number of different cancers, and to describe its role in glioma stem cell maintenance and tumorigenicity (Sun et al., 2014). It is also interesting to note that there is a correlation between ACSVL3 and Akt signaling in U87MG cells.
When ACSVL3 is depleted by siRNA in U87MG cells, Akt signaling also decreases, and that it in part is due to aberrant receptor tyrosine kinase (RTK) signaling. Using the new
ACSVL3 KO cells, we hope to identify the precise biochemistry by which ACSVL3 supports the malignant phenotype, including how it contributes to RTK signaling through
Akt.
56 Chapter 2
Introduction
Sphingolipids represent the second largest class of membrane lipids, and thousands of species have been identified. This class of lipids regulates a wide range of functions within a cell and is responsible for, or at least play a part in, pathways related to proliferation, differentiation, migration, and apoptosis (Hannun and Obeid, 2008). The basic structure of a sphingolipid contains a sphingoid base, a fatty amino alcohol, typically 18 carbons long. De novo synthesis of the base begins in the endoplasmic reticulum with the condensation of a palmitoyl-CoA and serine, a reaction that is catalyzed by the serine palmitoyltransferase enzymes (SPTs), to create dihydrosphingosine. This is then amino-acylated with a long- to very-long chain acyl-
CoA to form different species of dihydroceramide. This reaction is catalyzed by one of six different ceramide synthase enzymes. Dihydroceramide is desaturated to form ceramide, a biologically active molecule that can also be used as a precursor to more complex sphingolipids. Ceramide can also be deacylated by ceramidases to yield sphingosine. Sphingosine is another bioactive molecule, and this can be phosphorylated by sphingosine kinases to produce sphingosine 1-phosphate (S1P), another signaling molecule. This can also be dephosphorylated back to sphingosine, and sphingosine can be re-acylated back to ceramide. It is this interconversion of these three metabolites that form the basis of a regulatory “sphingolipid rheostat” hypothesis described by Spiegel and colleagues (Cuvillier et al., 1996).
57 Many studies performed over the last few decades have increased our understanding of bioactive lipids, and ceramide in particular. Ceramide, the most fundamental sphingolipid, has been found to be a tumor suppressor, promoting apoptotic and autophagic pathways, and inhibiting cell growth. Many enzymes responsible for the synthesis of ceramide are often found to be altered in cancer, resulting in a decrease of ceramide levels (Morad and Cabot, 2013). Six ceramide synthases (CerS1-6) have been discovered, and each enzyme has specificity for the synthesis of different ceramide species for different biological functions (Saddoughi and Ogretmen, 2013).
In contrast to ceramide, S1P is a potent proliferative, prosurvival, and promigratory factor (Maceyka et al., 2012; Rosen et al., 2009). Many growth factors, such as epidermal growth factor (EGF), as well as certain cytokines activate sphingosine kinase (SK1) acutely, resulting in higher levels of S1P (Hait et al., 2006). Once S1P has been generated, it can be transported from the inner leaflet of the plasma membrane to the exterior, where it can bind to specific receptors, S1P receptors1-5 (S1PRs). These receptors initiate signaling for motility, proliferation, migration, and survival, through typical G-protein coupled receptor signaling, including through Akt (Taha et al., 2004).
Gangliosides are another type of lipid important in cancer biology and signaling.
Gangliosides are glycosylated sphingolipids found on the outer leaflet of the plasma membrane, with their ceramide backbone inside the membrane and their carbohydrate head groups facing the extracellular environment. Gangliosides are important to many normal physiological processes, including cell adhesion and cell signaling (Hakomori,
2002), as well as many pathological ones, such as those related to cancer (Hakomori,
58 1996). Tumor-associated gangliosides play key roles in cancer cell invasion and metastasis, and this has been an active field of cancer research in the past few decades.
Cell membranes are structurally heterogeneous and composed of different domains with unique properties. These domains make it possible for disordered and ordered phases to coexist in the same membrane (Mesquite et al., 2000; Wang, et al.,
2000; and Wang and Silvius, 2000). Lipid rafts are one type of domain found in the cell membrane, and the hypothesis that these existed was proposed by Kai Simons and Elina
Ikonen in 1997 (Simons and Ikonen, 1997). Biochemical techniques have been developed to try to isolate lipid rafts by exploiting their detergent-resistant property in order to study their composition. Studies have shown that these tightly packed regions have high concentrations of sphingolipids, which include gangliosides, saturated fatty acids, and cholesterol.
Results presented here show that a decrease in ACSVL3 in U87MG cells causes a decrease in sphingolipids, but not cholesterol. We believe that the ceramides found in the parental U87MG cells provides a pool for the synthesis of other sphingolipids, including gangliosides and sphingosine 1-phosphate, therefore increasing signaling from lipid rafts, and perhaps signaling directly through the lipids internally, without increasing cholesterol levels in the cell. Depleting cells of ACSVL3 decreases ceramide levels affecting downstream lipids important in signaling and lipid raft structure. We observe a decrease in signaling through Akt in the ACSVL3 KO cells when incubated with EGF. U87MG cells have more C18:0 and C22:0 ceramide content than the ACSVL3 KO cells, as these species are decreased when ACSVL3 is absent, while C16:0 and C24:0 ceramides do not change. This , in addition to the acyl-CoA synthetase activity assays presented in
59 Chapter 1, supports the notion that the acyl chain-length preference of ACSVL3 is physiologically relevant.
60 Methods and Materials
General materials and media
U87MG and ACSVL3 KO cells were grown in standard U87 media. Ceramide
standards for thin layer chromatography were purchased from Matreya (Pleasant Gap,
PA). [1-14C]fatty acids were purchased from Moravek (Brea, CA) and unlabeled fatty
acids were purchased from Matreya. The internal standards for sphingosine and S1P
analysis, heptadecasphing-4-enine-1-phosphate and heptadecasphinganine-1-phosphate, were a gift from Ann Moser (Kennedy Krieger Institute). Ganglioside standards solution and equipment for ganglioside visualization via thin layer chromatography were kind gifts from Dr. Ronald Schnaar. Thin layer chromatography (TLC) plates used for ganglioside analysis were purchased from EMD Millipore by the Schnaar lab (Cat #
105635). Epidermal growth factor (EGF) was obtained from Sigma-Aldrich. [1-
14C]acetate was purchased from Moravek. TLC plates for the phospholipid and neutral
lipid synthesis analysis was purchased from Analtech, Inc (Cat # 01011).
Neutral and Phospholipid Synthesis Assay and Visualization
Cells were allowed to grow to confluence in 10 cm dishes. A total of 2 uCi of [1-
14 o C]acetate was added to each plate and allowed to incubate at 37 C in a 5% CO2
environment for 30 min. After incubation, the cells were harvested using trypsin and
were washed twice with PBS. The cell pellets were resuspended in STE buffer. Protein
concentrations were measured using the Lowry protein assay, using bovine serum
albumin (BSA) as the standard. A portion of cell suspension containing a total of 400 ug
61 protein was extracted with chloroform and methanol using the method of Folch et al.
(1957). Briefly, cell pellets in 1x PBS in a volume of 250 uL were transferred to a
16x125 glass tube with Teflon-lined caps. 5 mL of 2:1 chloroform:MeOH + 5mM HCl
and 1 mL of water was added to the cell suspension. The samples were mixed by
vortexing and then centrifuged for 2 min at 2,000 rpm in a table top centrifuge
(Beckman). After centrifugation, the aqueous upper phase was discarded. Folch’s
theoretical upper phase (3:48:47 chloroform:MeOH:H2O) was added to each tube. The
tubes were mixed by vortexing and centrifuged for 7 min at 2,000 rpm. The upper phase
was discarded and the wash was repeated until the white interphase was gone. After
removal of the upper (aqueous) phase, the lower organic phase containing extracted lipids
was evaporated to dryness under a stream of N2. Dried lipid samples were dissolved in
100 uL chloroform and a volume (20 uL) equal to 80 ug protein was spotted per lane on
the TLC plate in duplicate. Neutral lipids were separated via TLC using the solvent
system Hexane: Ethyl Ether: Glacial Acetic Acid (80:20:1). For separation of
phospholipids, the solvent system was Chloroform: Ethanol: dH2O: Triethylamine
(30:35:7:35). The following standards were used for analysis of neutral lipids and
phospholipids: cholesteryl ester was purchased from Sigma-Aldrich (Cat. #C6072); cholesterol was purchased from Matreya (#1006); tripalmitin, 1,3-dipalmitin, 1,2- dipalmitin, and monopalmitin were purchased from Nu-Chek Prep (Elysian, MN);
phospholipid standards (phosphatidic acid, phosphatidylinositol,
phosphatidylethanolamine, phosphatidylserine, phosphatidylcholine) were purchased
from Doosan Serdary Research Laboratories (Toronto, ON, Canada).
62 Ceramide Synthesis Assay
For assay of fatty acid incorporation into ceramide, “working solutions” of [1-
14C]fatty acids were prepared in benzene such that 50 uL contained ~1 nmol of radio- labeled fatty acid plus 4 nmol unlabeled fatty acid, for a total of 5 nmol. Working solutions of C16:0, C18:0, C22:0, and C24:0 were prepared. For assays, duplicate 50 uL aliquots of each fatty acid were dried down with nitrogen gas in a 13x100 mm glass tube.
The fatty acid was then solubilized by adding 50 uL of alpha-cyclodextrin (10 mg/ml in
10 mM Tris, pH 8.0) and sonicating in a water bath sonicator (Branson) for 2 minutes.
The fatty acid in alpha-cyclodextrin was further incubated in a 37C water bath for 30 minutes with gentle shaking. A reaction mix was made so that, when combined with solubilized fatty acid and cell suspension, it contained (final concentrations) : 40 mM
Tris, pH 8.0; 10 mM ATP, 10 mM MgCl2; 0.2 mM Coenzyme A, 0.2 mM dithiothreitol;
1 mM Serine; 0.2 mM NADPH. The pH was adjusted to 7.5 with 1N KOH. A cell suspension volume that was equivalent to 250 ug protein was added to the appropriate glass tubes and STE was added to adjust the total volume to 50 uL. Tubes that served as the blank received 50 uL STE only. 150 uL of mix solution was added to each tube, and the samples were then incubated in a 37C water bath for 2 hours with gentle shaking.
The reaction was stopped with 3.75 mL of 1:1 chloroform:methanol. Samples were extracted by the method of Folch et al. as described above for neutral and phoshpholipid synthesis. The final upper phase was discarded and the lower phase was dried under nitrogen gas. The resulting lipid pellet was resuspended in 50 uL 1:1 hexane:ethanol. 20 uL was spotted onto a TLC plate that was washed with 1:1 chloroform:methanol and dried before use. To resolve lipids on the plate, a solvent system of 94:1:5
63 chloroform:MeOH:glacial acetic acid was used, and the solvent front was allowed to run resolve until the front reached ~1.5cm from the top.
Ceramide extraction for LC-MS/MS Analysis
Cells were pelleted in a 1.5 microcentrifuge tube after harvesting and resuspended in 200 uL of dH2O. The cell suspensions were sonicated using a probe sonicator
(Missonix) three times, 5 seconds each time while on ice and a minute in between pulses.
600 uL of methanol, containing the internal standards (C12:0 ceramide, Avanti Polar
Lipids), was added. The samples were then vortexed. 800 uL of chloroform was then added. Samples were vortexed for 5 seconds three times. Tubes were centrifuged at
14,000 rpm for 5 minutes to separate organic and aqueous phases cleanly. The lower organic phase was collected and transferred to a separate glass vial and analyzed via LC-
MS/MS.
Sphingosine 1-phosphate and Sphingosine extraction for LC-MS/MS Analysis
The extraction method for sphingosine and S1P was adapted from Shaner, et al.,
(2009). The cells were washed twice with 1x PBS, aspirating as much of the liquid as possible after the last wash. Cells were scraped from the cell culture dishes in the residual PBS (around 0.2 mL). Protein concentrations were measured using the Lowry assay (Lowry et al., 1951). 400 ug protein was transferred to a 13x100mm borosilicate tube with a Teflon-lined cap for each sample. To each of the samples the following was added: 0.5 mL MeOH, 0.25 mL chloroform, and the internal standard cocktail for a total of 1 ng of each standard per sample (C17:0 sphingosine 1-phosphate, C17:0 sphinganine
64 1-phosphate, and C17:0 sphingosine – Avanti Polar Lipids). The samples were dispersed using a water bath sonicator at room temperature for 30 seconds. The mixture was incubated at 48C overnight in a heating block. This affords optimal extraction of sphingolipids due to their high phase transition temperatures. After cooling back to room temperature, 75 uL of 1M KOH in MeOH was added to each tube and after brief sonication, was incubated in a shaking water bath for 2 hr at 37C to cleave potentially interfering glycerolipids. The samples were again cooled to room temperature. 3-6 uL of
glacial acetic acid were added to bring the extract to a neutral pH. This was tested using
pH paper. The tubes were centrifuged to remove the insoluble residue and the
supernatant was collected. The samples were purified by centrifugation through a Spin-X
filter (Corning) and were analyzed by LC-MS/MS by Ann Moser in the Peroxisomal
Diseases Laboratory (Kennedy Krieger Institute).
Lipid Raft visualization
Plasma membrane lipid rafts between the U87MG cells and the ACSVL3 KO cells were compared using the Vybrant Alex Fluor 555 Lipid Raft Labeling Kit
(ThermoFisher). Cells were grown on 10-12 mm diameter circular glass coverslips in 6- well plates until confluent. Cells were labeled as described in the manual provided in the kit. The cells were labeled with 1 ug/mL Alexa 555-conjugated cholera toxin-B in
U87MG media for 10 min at 4C. After this, the cells were washed 5 times with ice-cold
1x PBS. To crosslink the CT-B-labeled lipid rafts, the cells were then incubated in a
1:200 dilution of rabbit anti-CT-B antibody in U87MG media for 15 min at 4C. Cells were again washed 5 times with ice-cold 1x PBS. Cells were fixed with 3.7%
65 paraformaldehyde in 1x PBS with magnesium and calcium, and then coverslips were
mounted on glass slides using mount solution (1.0 mg/ml phenylene diamine in 0.1M
Tris, pH 8.7 and 90% anhydrous glycerol) and sealed with clear fingernail polish to
prevent fading of the fluorescent signal. Raft-associated fluorescence was visualized
using a Zeiss Axioplan microscope and Axiovision software; digital images were
captured using identical exposure times.
Ganglioside extraction and TLC analysis
Once the U87MG and the ACSVL3 KO cells were confluent, media was aspirated
and the cells were washed with PBS twice. Cells were harvested in 500 uL deionized
H2O using a cell scraper. Another 500 uL of H2O was added to the plates and scraped to
retrieve remaining cells. The cells were then pelleted and weighed in pre-weighed 1.5 mL microcentrifuge tubes, and 4 volumes of deionized water was added (4 mL water/g cells).
The volume to acquire equal cell pellet weight for each sample was calculated and then appropriate volumes were transferred to a 16 x 125 mm glass tube. The total aqueous volume was calculated (volume of dH2O + 80% weight of tissue). To the samples, 2.67 volumes of room temperature methanol was added and the tubes were vortexed. 1.33 volumes of chloroform were added, and then the samples were vortexed again. The
samples were centrifuged for 15 min at room temperature and 2,000 rpm. The resulting
supernatant was then transferred to a new tube with a graduated pipet in order to note
how much of the volume was recovered. 0.173 volumes of dH2O were added and the
samples were vortexed. The samples were centrifuged for another 15 minutes at 2000
rpm. The upper phase was transferred to a new tube using a Pasteur pipet. During the
66 centrifugation, a tC18 Sep-Pak on a glass syringe was pre-washed with 3 mL of the
following in order – CHCl3:MeOH:H2O (2:43:55), MeOH:H2O (1:1), MeOH,
MeOH:H2O (1:1), and CHCl3:MeOH:H2O (2:43:55). The upper phase after
centrifugation was loaded onto the pre-washed Sep-Pak. The Sep-Pak was then washed with 3 mL of the following in order - CHCl3:MeOH:H2O (2:43:55) and MeOH:H2O
(1:1). Gangliosides were eluted with 3 mL of MeOH and collected in a new glass tube with a pointed bottom. The samples were dried down under nitrogen gas, and the sides of the tube were washed down with MeOH to collect any gangliosides that were on the walls on the tube, and then subsequently dried with nitrogen gas.
TLC plates were prepared by allowing the plate to dry in a 125C oven for 10 min.
Samples were dissolved in 10 uL of MeOH and 5 uL were spotted on to the TLC per lane using a glass syringe. Gangliosides were resolved using a solvent system consisting of
CHCl3:MeOH:0.25% aqueous KCl (60:35:8). Samples were allowed to resolve until the solvent front reached 2 cm from the top of the plate. The plate was allowed to dry for 10 min before spraying with resorcinol (0.3% resorcinol, .0031% CuSO4*5H2O, 30% HCl).
The plate was covered with a glass plate and clipped together in order to retain the resorcinol on the plate, and was then placed in an oven at 125C for 20 min. Gangliosides appear as light purple bands, which were scanned using a Microtek ScanMaker 8700.
Densitrometric analysis was performed using ImageJ (National Institutes of Health, http://imagej.nih.gov/ij/download.html ).
67 Incubation with EGF and Lipid Synthesis with [1-14C]acetate
Cells were grown and processed in the same way as previously described when analyzing neutral and phospholipid synthesis in U87MG and ACSVL3 KO cells. Before incubation for 30 min with [1-14C]acetate, EGF at a final concentration of 0.2ug/mL was added to the media and the cells were allowed to incubate for a total of two hours. For the last 30 minutes of EGF incubation, a total of 2 uCi of [1-14C]acetate was added to each plate.
68 Results
ACSVL3 Depletion Does Not Affect Cholesterol Levels or Synthesis
Initial efforts to characterize cholesterol homeostasis in the ACSVL3 stable
knockdown cell lines suggested that depleting cells of ACSVL3 decreased both synthesis
and total levels of cholesterol (Z. Pei and PA Watkins, unpublished). Therefore, I wanted
to verify and extend this observation in the ACSVL3 KO cells. Also, since cholesterol is
enriched in lipid rafts, I hypothesized that a decrease in cholesterol was mechanistically
involved in the decrease in RTK signaling observed in the ACSVL3 knockdown cells
when compared to the wild type U87MG cells.
The Amplex Red Cholesterol assay kit by Molecular Probes was used to quantify
total cholesterol levels per mg of cellular protein. Both wild type and KO cell lines were allowed to grow in 10cm dishes until confluent and then harvested using trypsin. As shown in Figure 14, total cholesterol levels were not changed when the ACSVL3 KO cells were compared to the U87MG cells. The method used allows measurement of both total and unesterified cholesterol, and the esterified cholesterol fraction can be calculated
by subtraction. However, in both U87MG and ACSVL3 KO cells, cholesterol esters
accounted for <1% of total cholesterol (data not shown). Thus, the majority of the cellular
cholesterol in both cell types was unesterified sterol, and there were no differences
between wild type and KO.
To investigate whether extracellular cholesterol present in the culture medium
affected cellular cholesterol levels, both cell lines were grown in U87 growth media
containing delipidated fetal bovine serum for 72 hours and then harvested with trypsin.
If any differences were observed, this could mean that there is a difference in cholesterol
69 uptake when ACSVL3 is depleted. The total cholesterol was measured using the Amplex
Red Cholesterol assay kit, and resulting cholesterol amount was normalized to protein
used in the assay, as above. Incubation with delipidated serum resulted in a decrease in overall cellular cholesterol when compared to the cells grown in regular media, but again, no difference was observed between the ACSVL3 KO cells when compared to the
U87MG cells (Figure 14).
70
Figure 14. Total cholesterol is not different between U87MG cells and ACSVL3 KO cells. The Amplex Red Cholesterol Assay was used to determine if cholesterol levels were different between the U87MG cells and the ACSVL3 KO cells. We did not observe a difference between the two cell lines when grown in regular (“reg”) U87MG growth media. Cells were also grown in media made with delipidated FBS (“DL”) to see if differences could be seen without an exogenous source of lipids. While the cholesterol levels decreased slightly, there still were no differences between the U87MG cells and the KO cells.
71 ACSVL3 Depletion Does Not Affect de novo Synthesis of Neutral Lipids or
Phospholipids
ACSVL3 depletion in glioma cells or lung cancer cells by siRNA (Pei et al. 2009;
2013) or in the ACSVL3 KO cells (Figure 10B) resulted in a significant decrease in
growth rate. One hypothesis to explain this phenomenon is that ACSVL3 is necessary to
supply bulk lipids for the high growth rate in cancer cells by de novo synthesis. This was
investigated by [1-14C]acetate incorporation studies. The TLC system used to resolve
cholesterol and cholesterol esters also separates triglycerides, diglycerides, and free FA.
There was no significant difference in synthesis of these neutral lipids when ACSVL3
was absent (Figure 15). To investigate de novo synthesis of phospholipids, cells were incubated with [1-14C]acetate and extracted lipids were separated by TLC using a
different solvent system. There were also no significant differences in de novo synthesis of phosphatidic acid, phosphatidylcholine, phosphatidylethanolamine, phosphatidylserine, or phoshphatidylinositol when U87MG cells lacked ACSVL3. These observations suggest that ACSVL3 in U87MG cells does not affect growth rates by upregulating de novo lipid synthesisStudies to assess whether ACSVL3 KO affected incorporation of [1-14C]FA into neutral lipids and phospholipids were also performed in
the lab; in agreement with my findings with [1-14C]acetate, incorporation of C16:0 or
C24:0 into lipids was not significantly different in U87MG and KO cells (X Shi and PA
Watkins, unpublished). In view of these data, I explored other hypotheses that could
explain the importance of ACSVL3 in malignancy.
72
Figure 15. Lipid synthesis is not affected by depletion of ACSVL3 in U87MG cells.
Lipids from cells incubated with [1-14C]acetate were separated using thin layer chromatography. Left, neutral lipids. Chol, cholesterol; FFA, free fatty acids; TAG, triacylglycerol; ChE, cholesterol esters. Diacylglycerols were observed as faint bands running between Chol and FFA. Right, phospholipids. PA, phosphatidic acid; PE, phosphatidylethanolamine; PI, phosphatidylinositol; PC, phosphatidylcholine; PS, phosphatidyl serine. There were no significant differences in de novo lipid synthesis in the ACSVL3 KO cells when compared to the U87MG cells.
73 Ceramide Synthesis and Concentrations are Decreased in ACSVL3 KO Cells
Ceramides provide the basic structural precursors for biosynthesis of
sphingolipids that are important in signaling, as well as being important in supporting
signaling via receptor tyrosine kinases and G-protein coupled receptors found in lipid rafts. In addition, ceramides themselves have been shown to be a key player in intracellular signaling and are involved in apoptosis, proliferation, cell growth, and differentiation (Pettus et al., 2002). It has also been shown that long-chain and very-long chain ceramides have differing effects on cellular pathways. Because we observed a decrease in RTK signaling when ACSVL3 was depleted in ACSVL3 KD cells (Pei et al.,
2009) and in preliminary studies with the ACSVL3 KO cells when compared to the
U87MG cells, I wanted to determine whether the differences were due to ceramide and sphingolipid content and synthesis.
Ceramide synthesis was tested in both the U87MG cells and the ACSVL3 KO cells. I initially assayed synthesis using radiolabeled palmitic acid (C16:0) and lignoceric acid (C24:0), representative long and very-long chain fatty acids. The activation of these lipids was shown to be slightly lower in ACSVL3 KO cells in acyl-CoA synthetase activity assays, suggesting that incorporation of C16:0 and C24:0 into ceramides would also be lower. Cells were incubated with [1-14C]palmitate or [1-14C]lignocerate for 2 hrs;
lipids were extracted and labeled lipids were separated on a TLC plate and visualized
(Figure 16). No differences were observed when we compared the U87MG cells to the
ACSVL3 KO cells. As noted in Chapter 1, KO of ACSVL3 had a greater effect on
activation of 18- and 22-carbon FAs than on 16- and 24-carbon FAs (Figure 12).
Therefore, I investigated ceramide synthesis using 14C-labeled stearic acid (C18:0) and
74 behenic acid (C22:0) to see if perhaps there was a similar chain length preference.
When stearic acid and behenic acid were used as substrates, ACSVL3 KO cells
synthesized less ceramide when using C18:0 than the U87MG cells (Figure 16). No
change was seen in ceramide synthesis when C22:0 was the substrate.
To determine whether the decreased ceramide synthesis reflected decreased levels
of ceramides containing stearic or behenic acids, we extracted cellular lipids and looked
at ceramide levels by acyl chain length using liquid chromatography-tandem mass spectrometry (LC-MS/MS). The results of this analysis reinforce the ceramide synthesis data. I saw a significant decrease in ceramides that contain C18:0, C20:0, and C22:0 acyl chains in the ACSVL3 KO cells, while ceramides that contain C16:0 and C24:0 were not affected. The data suggests that there is a preference for ceramides that contain stearic acid, arachidic acid, and behenic acid in U87MG cells, and that the levels of these ceramide species decrease when cells are depleted of ACSVL3 (Figure 17). It also suggests that ACSVL3 has a fatty acyl-chain preference. ACSVL3, like many of the
ACSs, shows a broad range of activity, being able to activate long- to very long-chain fatty acids, however this is the first time we see any type of ACSVL3 preference for particular fatty acids reflected in their incorporation into downstream lipid species.
75
Figure 16. Ceramide synthesis decreases in ACSVL3 KO cells when the substrate is
C18:0. Ceramide synthesis was tested in the U87MG and ACSVL3 KO cells. Cells were incubated with either C16:0, C18:0, C22:0, or C24:0. Ceramides were extracted, and duplicate samples (a, b) were resolved by TLC. Changes in synthesis were not observed when C16:0, C22:0, or C24:0 were used as substrates. A significant decrease was seen when C18:0 was used.
76
Figure 17. Ceramides with acyl chains C18:0-C22:0 are decreased in ACSVL3 KO
cells. Ceramides were extracted from U87MG and ACSVL3 KO cells and quantitated by
LC-MS/MS as described in Methods. Analyte levels were normalized by the amount of internal standard (IS) recovered. The levels of C16:0 and C24:0 ceramides were the same in the U87MG cells and the ACSVL3 KO cells. Differences were found in the ceramides with C18:0-C22:0 acyl chains, with the KO cell line showing a decrease compared to the U87MG cells.
77 Sphingosine and Sphingosine 1-Phosphate Levels are Also Decreased in the
ACSVL3 KO Glioma cells
Ceramidase, or N-acylsphingosine deacylase, releases a fatty acyl chain from
ceramide to produce sphingosine. I therefore measured the sphingosine and S1P levels in
both the U87MG and ACSVL3 KO cells to see if reduced ceramide levels led to a
decrease in these bioactive molecules as well. Sphingosine and S1P were extracted using
the method published by Shaner, et al (2009) and then measured by LC-MS/MS. Our results show that the overall levels of both sphingosine and S1P are decreased in the
ACSVL3 KO cells when compared to the U87MG cells, with S1P showing a statistically significant decrease (Figure 18).
78
Figure 18. Sphingosine and S1P levels decrease when cells are depleted of ACSVL3.
Sphingosine and S1P levels were quantified using LC-MS/MS to see if there was also a decrease in sphingosine levels. (A) A decrease in sphingosine was observed in the
ACSVL3 KO cells, but was not statistically significant (p = 0.06). (B) S1P levels were also measured and quantified. Here, there was a significant decrease in the ACSVL3 KO cells (p = 0.03).
79 Proteomic Analysis Shows an Overall Decrease in Proteins Important in Ceramide
Synthesis
To determine whether decreased ceramide levels could be due to lower expression
of enzymes involved in its synthesis, U87MG and ACSVL3 cells were collected for
proteomic analysis. Proteins important in ceramide synthesis were identified within the
dataset (Table 2). The three different serine palmitoyltransferase enzymes (SPTLC1-3), which catalyze the first, and rate-limiting, step in synthesis, were identified, and all showed a decrease in the ACSVL3 KO cells relative to the U87MG cells. The product of this reaction is the sphingoid base, which is then used to make ceramide and other sphingolipids. The next set of enzymes that shows a decrease is the ceramide synthase family. Three of the six ceramide synthase (CerS) proteins were identified in the analysis
– CerS2, CerS5, and CerS6. These showed an overall decrease as well. Because of the decrease found in the first step, it suggests that there is less starting material in the
ACSVL3 KO cells to produce subsequent products, such as ceramide and sphingosine.
Even though this would indicate an overall decrease in ceramides, we only find the decrease in ceramides containing C18:0-C22:0, suggesting an important role for
ACSVL3 in ceramide synthesis.
More importantly for ceramides, because the CerS enzymes identified are lower in the KO cells, there is less ceramide formation, in agreement with the ceramide synthesis assay result, showing that there is less ceramide synthesis. The main CerS known to be responsible for the synthesis of C18:0 to C22:0 ceramides in pancreatic β- cells is CerS4 (Veret et al., 2011). Half of the CerS enzymes were not detected in the proteomic analysis, but they cannot be ruled out as not present and not contributing to the
80 overall ceramide synthesis of the cells. The overall results suggest that there is a decrease in ceramide synthesis machinery when ACSVL3 is depleted in U87MG cells.
81
Table 2. Enyzmes involved in ceramide synthesis are decreased in ACSVL3 KO
cells. The proteomic data was analyzed for the presence of enzymes involved in
ceramide synthesis. Three serine palmitoyltransferases (SPTLC1-3) were detected and the levels of all three were all decreased in the ACSVL3 KO cells. Three of six ceramide synthases could be found in the data set (Ceramide synthases 2,5, and 6), and all were decreased in ACSVL3 KO cells.
82 Lipid Rafts Are Increased in ACSVL3 KO Cells Compared to the U87MG Cells.
Our lab had previously shown (Pei et al., 2009) that when ACSVL3 is depleted
using shRNA, signaling through c-Met, the hepatocyte growth factor (HGF) receptor, is
disrupted. In ACSVL3 KD cells, phospho-Akt (Ser473) was shown to be slightly
elevated after a short time after incubation with HGF, but then total and phospho-Akt
levels decreased to nearly undetectable levels between 1 and 2 hours after treatment. In
contrast, both total and phospho-Akt remained high for at least 6 hours of HGF
incubation with wild type U87MG cells. Similar results were obtained when the KD
cells and U87MG cells were incubated with epidermal growth factor (EGF). Lipid rafts
are thought to be important signaling platforms in the plasma membrane, and these
microdomains are enriched in sphingolipids, such as sphingomyelin and
glycosphingolipids like gangliosides. Even though we found that cholesterol is
unchanged in the ACSVL3 KO cells, we decided to visualize lipid rafts because of the
changes seen in ceramide synthesis.
To assess lipid raft differences between the ACSVL3 KO cell line and the wild-
type U87MG cell line, the Vybrant Lipid Raft Labeling Kit from Molecular Probes was
used. This method employs cholera toxin subunit B (CT-B) conjugated to an Alexa Fluor
555 fluorescent dye. CT-B is known to bind to ganglioside GM1, a commonly used marker for lipid rafts (Simons and Ikonen, 1997), and when viewed with a fluorescent microscope, can be used to indicate differences in lipid rafts between different cell types.
We had hypothesized that there would be more lipid rafts in the U87MG cells than in the
ACSVL3 KO cells, due to the aberrations we saw in signaling with the KD cells.
83 However, we found that GM1 staining was actually increased in the ACSVL3 KO when compared to the U87MG cells (Figure 19).
84
Figure 19. Fluorescent microscopy reveals an increase in lipid raft staining in
ACSVL3 KO cells. The Vybrant Lipid Raft kit was used to detect differences in lipid rafts between the U87MG cell line and the ACSVL3 KO cell line. Lipid raft staining was increased in ACSVL3 KO cells, contrary to the hypothesis. CT-B binds to ganglioside
GM1 in the lipid rafts, and this may be more indicative of ganglioside composition than overall lipid raft prevalence.
85 Overall Ganglioside Levels Decrease in the Absence of ACSVL3
Because of the unexpected finding of increased ganglioside GM1 in ACSVL3 KO cells, we wanted to see if ganglioside levels extracted from U87MG cells and ACSVL3
KO cells reflected the Vybrant Lipid Raft staining. Ceramides provide the base lipid for gangliosides, and results presented above showed that there was a decrease in ceramide synthesis and overall concentration in the ACSVL3 KO cells. Gangliosides were extracted and analyzed by TLC as described by Ronald Schnaar and colleague (1994).
Ganglioside standards were run on the same TLC, and were a gift from Dr. Schnaar. The
TLC shows that overall ganglioside levels are decreased in ACSVL3 KO cells (Figure
20). The major gangliosides that were found on the TLC were GM1 and GM3. GM1, as was demonstrated in the Vybrant lipid raft images, shows a decrease in the ACSVL3 KO cells. In contrast, GM3 was found to be decreased in ACSVL3 KO cells. Both gangliosides are implicated in signaling, either enhancing (GM1) or inhibiting (GM3)
RTKs. The overall ganglioside result is consistent with previous results showing that
C18:0 to C22:0 ceramides were lower in the ACSVL3 KO cells. We believe that the decrease in ceramide precursors results in a decrease in gangliosides. This aberration could contribute to the differences in signaling through Akt that we observed in the
ACSVL3 KD cells (Pei et al., 2009), and what we are seeing with the KO line
(unpublished).
86
Figure 20. Total gangliosides decrease in ACSVL3 KO cells, but GM1 ganglioside increases compared to U87MG cells. Gangliosides were extracted from U87MG cells and ACSVL3 KO cells, and were separated by TLC. The major gangliosides that were found were GM3 and GM1. GM3 amounts were decreased in the KO cells compared to the parental U87MG cells while GM1 was increased. This finding was in agreement with the Vybrant Lipid Raft staining, where KO cells had increased GM1 in the cell membrane.
87 EGF Signaling Upregulates Lipid Synthesis in the U87MG Cells, While Not
Affecting Synthesis in the ACSVL3 KO Cells
Because of the effects we see in the ACSVL3 KO cells pertaining to cell sphingolipid composition and the differences we see in signaling between the two cell lines (unpublished data), we wanted to see if incubation with a growth factor would affect lipid synthesis. Growth factors, such as EGF, signal through RTKs that are found in lipid rafts, and a perturbation in raft composition may affect signaling. Both ACSVL3 KD
(Pei et al. 2009) and ACSVL3 KO cells (unpublished data) showed aberrant signaling through Akt. As previously mentioned, U87MG cells and ACSVL3 KO cells incubated with radiolabeled acetate do not exhibit any differences in lipid synthesis. To explore if signaling has an effect on lipid synthesis, we incubated cells with EGF for 1 hour. [1-
14C]acetate was added to the media and allowed to incubate for the last 30 min. Cells were also incubated with [1-14C]acetate for 30 min without EGF. Lipids were extracted and then separated on TLC plates (Figure 21). First, despite prior findings in Figure 15 where there were no differences in acetate incorporation into cholesterol between the
ACSVL3 KO cells and the U87MG cells, here we do find a difference in synthesis without EGF. U87MG cells exhibited higher cholesterol synthesis. When the cells are incubated for 1 hour with EGF, we find that overall lipid synthesis increases in both cell lines, but the only difference we see is in cholesterol synthesis, where synthesis is greatly upregulated in the U87MG cells. We also looked at phospholipid synthesis, and find that there is an overall increase in synthesis in response to EGF in both cell lines, but the synthesis at both steady-state (no EGF) and with growth factor are equal in the ACSVL3
KO cell line and the U87MG cell line. One potential explanation for the difference
88 between results shown in Figure 15 vs. those shown here for unstimulated cholesterol synthesis is that different lots of FBS were used to maintain the cell lines, and these lots potentially have varying profiles of growth factors that could affect cellular metabolism.
Given that both cell types were grown in the same serum, this would suggest that there are differences in the lipid synthesis, especially in the synthesis of cholesterol, that we didn’t see previously.
89
Figure 21. ACSVL3 KO cells do not increase lipid synthesis when incubated with
EGF. Both U87MG and ACSVL3 KO cell lines were incubated with EGF + [1-
14C]acetate to observe how lipid synthesis responds to incubation with growth factor.
Cells that were not exposed to EGF but incubated with labeled acetate were included to look at unstimulated levels of lipid synthesis. (A) Neutral lipids. Cells that were not incubated with EGF do not show differences in synthesis of free fatty acids (FFA), triacylglycerol (TAG), or cholesterol esters (ChE). Cholesterol (Chol) synthesis was lower in ACSVL3 KO cells. There was an overall increase in lipid synthesis when both cell lines were incubated with EGF. Again, FFA, TAG and ChE synthesis were unaffected; however, the differences in cholesterol synthesis were more pronounced than in cells incubated without EGF. (B) Phospholipids. The synthesis of phospholipids including phosphatidic acid (PA), phosphatidylethanolamine (PE), phosphatidylinositol
90 (PI), and phosphatidylcholine (PC) not affected by ACSVL3 KO, either in the presence or absence of EGF. Like neutral lipid synthesis, there was an overall increase in phospholipid synthesis when cells were treated with EGF.
91 Discussion
The goal of this study was to understand whether ACSVL3, a very long-chain
acyl-CoA synthetase, is important in regulating the lipid components of lipid rafts in
U87MG cells. In previous studies with U87MG cells where ACSVL3 was depleted via
shRNA, it was shown that in the absence of ACSVL3, decreased signaling through Akt occurred when the cells were incubated in the presence of either hepatocyte growth factor
(HGF) or epidermal growth factor (EGF) (Pei et al., 2009). We hypothesized that signaling defects may be caused by a difference in lipid profiles of rafts in which the receptor tyrosine kinases can be found, as well as differences in important bioactive sphingolipid molecules, such as sphingosine 1-phosphate (S1P) when ACSVL3 is absent.
The first thing examined was the cholesterol levels of the U87MG cells versus the
ACSVL3 KO cells. Early unpublished experiments with ACSVL3 KD cells suggested that total cholesterol levels decreased when ACSVL3 levels were lowered, and could be a cause of the aberrant signaling through the receptor tyrosine kinases c-MET, the receptor for HGF, and the EGF receptor. Lipid rafts are transient plasma membrane domains enriched in cholesterol that form stable platforms for receptors and signaling molecules.
However, in this study, I show using KO cells that ACSVL3 levels do not impact total cholesterol levels. Cholesterol levels were affected by amount of sterol in the surrounding media (regular FBS vs. delipidated FBS), but this change was equal in the
U87MG cells and the ACSVL3 KO cells. I conclude that any differences in signaling seen in the KO cells is not due to the cholesterol levels, as these are not different when compared to the U87MG cells.
92 Interestingly, while we did not find a difference in cholesterol content, we did
find differences between the two cell lines when assaying for cholesterol synthesis,
especially in the presence of EGF. The U87MG and ACSVL3 KO cell lines were
incubated with [1-14C]acetate, with and without EGF. While EGF did increase the
overall synthesis of both neutral and phospholipids, there was a clear difference in
cholesterol synthesis between U87MG cells and ACSVL3 KO cells. We hypothesize that
we did not see this difference in previous experiments when looking at the steady-state
synthesis of lipids is because of differences in lots of FBS. FBS is rich in lipids and
growth factors, and this could have affected previous studies. More work will have to
been done in the area of cholesterol synthesis to elucidate any role ACSVL3 may have in the pathway.
To see if there were any differences in lipid rafts between the U87MG cells and the ACSVL3 KO cells, I visualized lipid raft domains of the plasma membranes. Using the Vybrant Lipid Raft kit, which labels ganglioside GM1 commonly found to be enriched in lipid rafts, I found an increase in labeling in the ACSVL3 KO cells, contrary to what we hypothesized. To explore this further, I extracted gangliosides from both cell lines, separate them by TLC, and visualized them. Gangliosides GM3 and GM1 appear to be the most abundant. Although there was an overall decrease in gangliosides in the
ACSVL3 KO cells, GM3 was decreased in the ACSVL3 KO cells, while GM1 was increased, which agreed with and explained the Vybrant Lipid Raft labeling. While it has been documented in the literature that GM3 inhibits EGFR signaling (Hanai et al., 1988), and consequently inhibits cell growth (Bremer et al., 2986), we find that even though
GM3 is found in higher quantities in the U87MG cells, they still grow at a much faster
93 rate than the ACSVL3 KO cells. Similarly, GM1 levels are reverse of what would be
expected. Further study will be needed to fully understand the correlation between
ganglioside levels and RTK signaling in U87MG cells.
Because the difference is not cholesterol-dependent, as demonstrated by the total cholesterol levels in the U87MG and the ACSVL3 KO cells, we wondered if this was due to differences in sphingolipid synthesis and content, something not previously described in the ACSVL3 KD cells. Sphingolipids are present in all eukaryotic cells and contribute to membrane biology and signaling events that influence both cell behavior and function
(Holthuis et al., 2001; Merrill and Sandhoff, 2002). Biosynthesis of the sphingosine backbone begins in the endoplasmic reticulum with the condensation of serine and palmitoyl-CoA, followed by a reduction to dihydrosphingosine. Dihydrosphingosine is oxidized by FAD to produce sphingosine. Sphingosine can also be made from ceramide via ceramidase, which cleaves the fatty acyl chain from ceramide, leaving the sphingosine backbone.. Once created, the backbone can be phosphorylated to produce sphingosine 1-phosphate, a biologically active molecule or acylated by CerS enzymes to create ceramide, which is in itself an important biological molecule, or go on to create more complex sphingolipids, like gangliosides.
Proteomics data shows that there is a decrease in all three SPTLC enzymes in
ACSVL3 KO cells. These enzymes are the first step in sphingosine synthesis, catalyzing the condensation of serine and palmitoyl-CoA. In published RNAi studies targeting
SPTLC1, lack of this enzyme was found to affect cell growth in U87MG cells, with slower cell growth found over six days (Bernhart et al., 2015). The same study also
94 shows a decrease in ceramide levels when SPTLC1 is knocked down, which is similar to
what is observed in the ACSVL3 KO cells.
The study described here demonstrates that when ACSVL3 is absent in U87MG
cells, the levels of S1P decrease significantly. Recent studies indicate that S1P is a potent
mitogenic factor for several glioblastoma cell lines, stimulating proliferation in U373MG
cells through a G-coupled receptor and activating several different signaling pathways
(Brocklyn et al., 2003). Evidence suggests that S1P and its receptors are central players
that regulate glioblastoma growth, migration, and invasion (Kim et al., 2009), and
exogenously added S1P promotes glioma growth and enhances invasiveness (Brocklyn et al., 2002; Brocklyn et al., 2003; Annabi et al., 2009; Young et al., 2009). Sphingosine kinase (SphK), the enzyme that phosphorylates sphingosine to produce S1P, has been shown to be regulated by a number of extracellular ligands, including EGF, and this has been shown to activate both isoforms of SphK – SphK1 and SphK2 (Paugh et al., 2008;
Hait et al., 2005). Stimulation by EGF has also been shown to increase cellular invasion, and this occurs through SphK/S1P dependent activation of the ezrin, radixin, and moesin
(ERM) family of proteins, which are involved in cytoskeletal reorganization, cell division, migration, and invasion (Gandy et al., 2013). While this study does not demonstrate that SphK1 is higher in U87MG cells when compared to the ACSVL3 KO cells, the higher levels of S1P suggest that the regulation of one or both of these enzymes is altered in U87MG cells, and when ACSVL3 is depleted, the activity of SphK1/2 is also decreased, therefore suggesting an important role in the growth and proliferation of the cells.
95 Our data also suggests that ACSVL3 has a very important role in activating fatty
acids that go on to synthesize ceramides, specifically those that contain C18:0. This data is in agreement with our acyl-CoA synthetase activity assay data, showing that while
ACSVL3 has a very broad affect – C16:0 through at least C24:0 – there is significantly less activation of C18:0 and C22:0. It is interesting that in published data (Koybasi et al.,
2004), C18:0 ceramide was found to be increased with an overexpression of a gene known as LASS1 (Longevity Assurance), and lead to an inhibition of cell proliferation in human head and neck squamous cell carcinomas. There is evidence that counters this in breast cancer – ceramide levels were measured in both “normal” breast tissue and malignant breast tissue, and it was found that C18:0 ceramide was significantly higher in the malignant tissue (Schiffmann et al., 2009).
The ceramide synthase enzymes, CerS1-6, catalyze the N-acylation of the dihydrosphingosine backbone to produce ceramide (Stiban et al., 2010). Each CerS has been recognized to have specificity towards certain fatty acyl chain lengths. CerS1 has been shown to use C18-acyl-CoA (Venkataraman et al., 2002), CerS2 uses C22- or C24-
acyl-CoAs (Lavlad et al., 2008), CerS4 has specificity towards using C20-acyl-CoA, and
CerS5 and CerS uses C14- and C16-acyl-CoA (Riebeling et al., 2008; Mizutani et al.,
2003). While we do not find CerS1 in our proteomic analysis, we do find a decrease in
CerS2, as well as a decrease in CerS5 and CerS6. This suggests that there could be a
general decrease in the levels of CerS enzymes, affecting overall synthesis of ceramides.
While it is interesting to note that C16:0 or C24:0 ceramides are not decreased when the
CerS enzymes responsible for their synthesis are decreased, our data could suggest that it
96 is due to the significant decrease in activation of C18:0 and C22:0 by ACSVL3 that leads to the decrease of the corresponding ceramides.
ACSVL3 is important to U87MG cell proliferation, signaling, and the ability for the cells to form subcutaneous and orthotopic xenografts in nude mice. We asked why this particular acyl-CoA synthetase – whose only known function is to synthesize acyl-
CoAs – is important to the malignant properties of the cells. Results presented here suggest that this may be due to a decrease in sphingolipid synthesis. We find a significant decrease in S1P levels in the ACSVL3 KO, which is a biologically active molecule. Sphingosine, the backbone of sphingolipids, is also decreased, but to a lesser extent. Along with the lower levels of sphingosine, an ACSVL3 KO cell also sees decreased C18:0 and C22:0 ceramide content and synthesis. Lower levels of C18:0 ceramide are interesting because it is noted in the literature as to having a negative effect on head and neck squamous cell carcinoma tissue. Higher C18:0 ceramide levels were
50% lower in tumor tissue when compared to noncancerous tissue (Koybasi et al., 2004).
We believe that the lowered levels of sphingolipids and S1P in the ACSVL3 KO are due to a lack of starting materials (Figure 22). There is less sphingosine present in the
ACSVL3 KO cells, meaning that there is less material for signaling through S1P receptors and less starting material for ceramide synthesis. Because we see evidence of
ACSVL3 having greater specificity towards C18:0 and C22:0 activation, there is less acyl-CoA to be incorporated into ceramide. If there is less ceramide available, there is less incorporated into higher sphingolipids that are involved in signaling. With less sphingolipids at the plasma membrane supporting growth factor receptors, it could
97 explain the decrease in signaling through Akt, therefore making the ACSVL3 KO
U87MG cells less tumorigenic.
98
Figure 22. Decreasing activation of fatty acids destined for sphingolipid synthesis leads to aberrant signaling from receptor tyrosine kinases. Results presented in this chapter show that a decrease in the activation of fatty acids in the ACSVL3 KO cells leads to a decrease in synthesis of ceramides, S1P, and gangliosides, leading to a decrease in signaling from RTKs in the cell membrane localized to lipid rafts, or through S1P signaling. This may account for the decrease in cell proliferation, growth, and lipid synthesis in the ACSVL3 KO cells when compared to the U87MG cell line.
99 CHAPTER 3
Introduction
When a cell has made the switch from being a normal cell to becoming a cancer
cell, the cell is obligated to switch from a normal metabolic state to a state that favors
survival and proliferation (Ramanujan, 2015). One of the first identified and common
biochemical characteristics of cancer cells is aberrant glucose metabolism. Glucose
serves as a main source of energy in the form of ATP, and it can supply the precursors
and cofactors required for anabolic pathways f (Vander Heiden et al., 2009).
Under normal oxygen conditions, non-cancerous cells rely on mitochondrial
oxidative phosphorylation to generate ATP and switch to anaerobic glycolysis under
hypoxic conditions. Otto Warburg’s observation, that tumor cells took up very large
amounts of glucose and rapidly used it to produce lactate under aerobic conditions, is still
the most fundamental observation in tumor metabolism (Warburg, 1925; Warburg, 1956).
This was termed the Warburg Effect in the 1970s, and studies demonstrated that a
number of oncogenes are associated with the switch from oxidative phosphorylation to
glycolysis, such as myc, NF-kB, EGF, and Akt. With this increase in aerobic glycolysis,
higher levels of lactate are produced. High levels of lactate have been correlated with the
likelihood of metastases, tumor recurrence, and poor patient survival in a number of solid
tumors (Walenta, et al., 2016). Local acidosis and elevated lactate with or without
hypoxia characterizes many different tumors, with higher levels of acidosis associated
with increased invasive properties (Gillies, et al., 2008).
100 Warburg initially hypothesized that cancer cells develop a defect in mitochondria
that leads to impaired aerobic respiration, and thus rely on aerobic glycolysis (Warburg,
1956). However, it was found that the presence of aerobic glycolysis was not indicative
of impaired oxidative phosphorylation, and recent studies found that defects of
mitochondrial oxidative phosphorylation are not common in spontaneous tumors,
(Kroemer and Pouyssegur, 2008; Koppenol et al., 2011; Griguer et al., 2005). Still, the
role of mitochondria in tumor cells is often reduced to a provider of biosynthetic
intermediates for cellular proliferation (DeBerardinis, et al., 2008) which are now being
recognized as essential mediators of tumorigenesis. Consistent with this, for example,
breast and melanoma models indicate that metastasizing cancer cells have increased
levels of oxidative phosphorylation (LeBleu, et al., 2014).
The tricarboxylic acid (TCA) cycle plays a central role in the metabolism of
sugars, lipids, and amino acids (Scheffler, 2008). Pyruvate enters the mitochondria
through a heterodimeric complex composed of mitochondrial pyruvate carrier 1 and 2
(MPC1/2). From there, pyruvate can enter the TCA cycle (also known as the citric acid
cycle) to produce ATP, CO2, and the electron transport chain (ETC) cofactors, NADH
and FADH2. Specific cancer-associated mutations have been reported in nuclear-
encoded mitochondrial enzymes of the TCA cycle, including fumarate hydratase (FH)
(Xiao et al., 2012), succinate dehydrogenase (SDH) (Xiao et al., 2012), and isocitrate
dehydrogenase (IDH) (Dang et al., 2009).
Mitochondria also find themselves at a crucial intersection with lipid metabolism.
It is becoming clear that fatty acid oxidation by itself can promote metastasis of cancer cells. Inhibition of β-oxidation has been suggested as a potential therapy to slow tumor
101 growth (Samudio et al., 2010; Tirado-Velez et al., 2012), and several studies show that mitochondrial β-oxidation of free fatty acids is functional in cancer cells. This fatty acid metabolism may drive oxidative phosphorylation-dependent ATP production for cell
proliferation (Rodriguez-Enriquez et al., 2015). However, the data is somewhat
inconsistent with other studies showing some of the mRNA coding for components of the
β-oxidation machinery was significantly lower in colorectal carcinoma when compared to
normal colon tissue (Birkenkamp-Demtroder et al., 2002). Another study proposed that
mitochondrial β-oxidation is non-functional in several carcinoma cell lines but this
conclusion was based upon the high level of malonyl-CoA detected in the samples
(DeBerardinis et al., 2006).
Mitochondria are dynamic networks that are regulated and maintained by fusion
and fission (Westermann, 2010). Generally, it is thought that quiescent cells tend to have
mitochondria that exist as a meshwork of interconnected tubes while energy-producing
mitochondria are distributed through fission to regions of the cell with high energy
demands (Sanchez-Madrid and Serrador, 2009). Alterations to these dynamics are
associated with abnormal cell function which can leads to many human diseases (Liesa et
al., 2009), including cancer. Recent studies showed that a lung cancer cell line exhibited
excess mitochondrial fission and a defect in fusion that was important in cell cycle
progression (Rehman et al., 2012). Breast cancer cells that have impaired mitochondrial
fusion have inhibited migration and invasion while fission, which is regulated by the
protein Drp1 (Zhao et al, 2013), facilitates these processes..
In the study presented here, we show that a knockout of ACSVL3 results in
changes in glucose metabolism and the TCA cycle. ACSVL3 KO cell lines display an
102 increase in glycolytic enzymes and a decrease in TCA cycle enzymes protein levels.
Even with these large changes in enzyme levels, however, we do not see significant changes to metabolic intermediates of either pathway based on metabolomic analyses.
We also find that despite this decrease in TCA enzymes, in vivo glucose oxidation does decrease in the ACSVL3 KO cells as measured by the conversion of glucose to CO2.
While not an exhaustive study, we observe differences in mitochondrial morphology and proteins that suggests that ACSVL3 is important to mitochondrial function and dynamics.
103 Methods and Materials
Glucose Uptake Assay using 3H-2-Deoxyglucose
Both U87MG and ACSVL3 KO cells were cultured in duplicate 12-well culture plates until confluent. KRH buffer (0.5% BSA, 25mM HEPES, 120mM NaCl, 5mM
KCl, 1.2mM MgSO4, 1.3mM CaCl2, 1.3mM KH2PO4), serum-free U87 culture medium,
0.8% Triton, and 1x Dulbecco’s phosphate buffer saline solution (DPBS, with calcium and magnesium) were kept at 37C. The labeling buffer was made by adding 1 uL of 1 uCi/uL 3H-2-Deoxyglucose (a gift from Dr. William Wong) per 1 mL KRH buffer and mixed. Half of the labeling medium was kept at 37C (experimental plate) and the other half placed on ice (control plate). On the day of assay, cells were washed twice with warm 1x DPBS. 1 mL of warm serum-free U87 culture medium was added to each well, and the cells were placed in a CO2 incubator. After a 1 hour incubation, the cells were washed again twice with warm 1x DPBS. To each well, 1 mL of warm KRH buffer was added, and the cells were incubated for 20 min at 37C. After the 20 min incubation with
KRH buffer, the control plate was placed on ice. The KRH buffer was aspirated and 0.5 mL of warm labeling solution were added to each well of the experimental plate, and ice- cold labeling buffer added to the control plate on ice. The experimental wells were incubated at 37C for 10 min. The labeling buffer was aspirated, and the plates were put on a bed of ice to stop the reaction. Both experimental and control wells were washed twice with cold 1x DPBS. 240 uL of warm 0.8% Triton was added to the wells and were placed in a water bath sonicator for 10 min at 37C. Cell lysates were counted (150 uL) in a scintillation counter. All counts were normalized by protein.
104
Immunofluorescence
Immunofluorescence (IF) and Western blotting analysis was done using rabbit
anti-Tom20 antibody, a gift from Dr. Hiromi Sesaki. IF was performed by culturing
U87G and ACSVL3 KO cells on glass coverslips until nearly confluent. The coverslips were then washed twice with 1x PBS and fixed with 4% formaldehyde in 1x PBS for 20 min at room temperature. The fixation buffer was aspirated, and the coverslips were washed with 1x PBS 3x for 5 min each wash. Cells were permeabilized with 0.1% Triton
X-100 in 1x PBS for 8 min at room temperature. After removal of the Triton X-100, the coverslips were again washed with 1x PBS as above. Anti-Tom20 antibody was diluted to a final concentration of 1:400 in 0.5% BSA in 1x PBS. 20 uL of antibody was pipetted onto parafilm and the coverslips were gently placed upside down on the antibody dilution. Coverslips were incubated overnight at 4̊C in a humidified chamber.
Coverslips were washed the next day and then incubated in anti-rabbit secondary antibody conjugated to Cy3 (1:250) for 1 hour at room temperature. The coverslips were washed 3x with 1x PBS, and then mounted onto slides. Fluorescent images were captured using a Zeiss AxioImager fluorescent microscope with ApoTome attachment, and analyzed using Adobe PhotoShop CS5.
Metabolomic analysis of the U87MG cells and ACSVL3 KO cells
Both the U87MG cells and the ACSVL3 KO cells were seeded onto 150 mm dishes in U87 growth media. Plates were given to Dr. Anne Le (Dept. of Pathology) for
105 extraction and metabolomics analysis; this was done on a fee-for-service basis. Once the dishes were at least 85% confluent, the dishes were placed on ice. The cells were washed with ice-cold 1x PBS by adding 4 mL PBS to the dish and tilting the dish back and forth.
The 1x PBS was aspirated, and the cells were washed a second time the same way. After the PBS was aspirated for the last time, the cells were frozen by placing the dishes on dry ice. A 2.5:1 MeOH:H2O mix was made and was kept on ice until needed. 525 uL of the
MeOH:H2O mix was added to the 150 mm dish and tilted back and forth to cover the whole surface. The dish was placed on ice. Cells were scraped off of the 150 mm dish and cells+MeOH:H2O were collected into a 2 mL Eppendorf tube on ice. Another 525
uL of the MeOH:H2O mix was added to the dishes and any sample remaining on dish was
collected the same way and added to the 2 mL Eppendorf tube. The samples were
sonicated 3-5 times for 1 second each time while keeping the tube on ice. 375 uL
chloroform was added to the tubes while on ice for a ratio of 1:2:0.8
chloroform:MeOH:H2O. Tubes were mixed by vortexing, and then 375 uL of water was
added to the tubes. Tubes were mixed again, and another 175 uL chloroform was added
while the samples were kept on ice. The samples were mixed a final time and were
centrifuged for 30 min at 3,000 x g. After centrifugation, the samples separate into 3
layers – top (polar compounds), middle (protein), and bottom (lipids). The top layer of
each sample was transferred to appropriate tubes, volume was reduced using a SpeedVac,
and finally samples were lyophilized. The middle layer was used for protein
quantification, and was dried down using a SpeedVac. The bottom layer was saved and
dried for future lipidomic studies. Samples were analyzed using Q-TOF LC/MS with
data analysis performed with Qual and MPP software.
106
Oxidation of Glucose to Carbon Dioxide
U87MG and ACSVL3 KO cells were cultured in T-25 flasks until confluent. On
the day of the assay, the media was aspirated and the cells were washed 2x with pre-
warmed (37C) 1x PBS. One set of flasks was harvested with trypsin for protein. The
cells that were used in the assay, as well as a set of flasks used as a “blank” (flasks
without cells), were incubated with pre-warmed (37C) serum-free U87 media for 1 hour in a CO2 incubator. The media was aspirated, and then washed with 1x PBS twice. The
flasks were incubated with 5 mL of pre-warmed KRH buffer for 20 min in the 37C warm
room. During the incubation, uniformly 14C-labeled D-glucose (Moravek Biochemicals)
was added to pre-warmed KRH buffer for a final concentration of 5mM. The KRH
buffer was aspirated from the flasks, and 5 mL of labeled KRH buffer was added. The
flasks were capped with rubber stoppers fitted with hanging wells containing Whatman
glass microfiber filter paper that had been wetted with 20 uL of 1.0M KOH to trap the
14 CO2 produced. The flasks were incubated for 1 hour in the 37C warm room with gentle
shaking. The reactions were terminated by injection of 250uL of 60% perchloric acid
through the rubber stopper, and the flasks were kept at 4C overnight with gentle shaking
14 to trap the CO2. The filter paper was transferred to scintillation vials and 10 mL of
BudgetSolve counting solution was added. 14C in the vials was assayed by scintillation
counting. Dpms and protein concentrations were used to express results as nmol/20
min/mg protein.
107 Results
Glucose Transporters Glut1 and Glut3 Are Higher in U87MG cells, But Glucose
Uptake Remains the Same
Because U87MG cells lacking ACSVL3 have more normal growth properties compared to wild-type U87MG cells, we were interested to see if glucose metabolism was different in the two cell types. Glucose and lipid metabolism are interconnected through a number of different pathways, particularly in how the cell produces energy for growth and proliferation. We asked if glucose uptake would be affected by an ACSVL3
KO. We first looked at our proteomic data to see if we detected any glucose transporters.
Our analysis did detect two – Glut1 and Glut 3. These facilitative glucose transporters are overexpressed in different types of solid tumors consistent with their potential role to supply the cells with glucose needed for ATP generation through the Warburg effect
(Macheda et al., 2005; Amann and Hellerbrand, 2009; Shim et al., 2013; Ramani et al,
2013). Glut1 and Glut3 were both lower in our ACSVL3 KO cells, half of what was detected in the U87MG cells (Figure 22A).
To determine whether the decrease in the glucose transporters caused a decrease in glucose uptake in the KO cells, we performed a glucose uptake assay by incubating both cell lines with [3H]2-deoxyglucose. This deoxy sugar is taken up by glucose transporters but is not metabolized, thus providing an estimate of transport alone.
Following uptake, cells were lysed in 0.8% Triton and counted in a scintillation counter.
Final calculations were performed to reflect nmol/20min/mg protein. We found that even though the glucose transporter levels are decreased in our ACSVL3 KO cells, glucose
108 uptake was not affected (Figure 22B). One possible reason for this is because of the low
KM of the transporters (KM= ~1mM) which is well below the serum glucose concentration ~3-5mM) indicating these levels are not likely limiting.
109
Figure 23. Glucose uptake in ACSVL3 KO cells relative to U87MG cells. (A)
Proteomic analysis of the cell lines detected two of the glucose transporters: Glut1 and
Glut3. Both of these facilitative transporters were lower (0.5) in the ACSVL3 cells
compared to the U87MG cells. (B) Because the two detected glucose transporters were
lower, we wanted to see if glucose uptake was affected. Cells were incubated in a
labeling buffer that included 3H-2-deoxyglucose and samples were counted for radioactivity. There were no significant differences between the two cell lines (n = 5).
110 Glycolytic Enzymes Are Upregulated in ACSVL3-Depleted Cells Without
Significantly Affecting Metabolic Intermediates
Proteomic analysis of U87MG cells and ACSVL3 KO cells revealed a general increase in enzymes involved in glycolysis in the KO cells compared to the U87MG cells. Other than the two different glucose transporters, Glut1 and Glut3, which were both decreased by half in the KO cells, every other glycolytic enzyme was upregulated by at least 1.4-fold in the KO cells (Table 3). This came as a surprise, considering the KO cells’ less tumorigenic characteristics we didn’t expect an increase in glycolytic enzymes.
Importantly, however, although the glycolytic enzymes are more abundant in the KO cells, we cannot definitively conclude that there is an increase in glycolytic activity. To gain further insight, we used a metabolomics approach and investigated the levels of metabolic intermediates of glycolysis to establish whether or not glucose metabolism might actually be altered, and perhaps upregulated in the ACSVL3 KO cells.
Untargeted metabolomic analysis was performed on our U87MG cells and the
ACSVL3 KO cells in collaboration with the laboratory of Ann Le, Ph.D. (Johns Hopkins
University School of Medicine). Most intermediates of the glycolytic pathway were
detected, and levels in the U87MG cells and the ACSVL3 KO cells were compared
(Figure 23). Levels of glucose-6-phosphate, the product of the hexokinase enzymes, did not show an appreciable difference between the two cell lines (Figure 23A). There was also no significant change found in glyceraldehyde 3-phosphate levels (Figure 23B), but there was a significant decrease in glycerate 3-phosphate in the ACSVL3 KO cells
(Figure 23C). This would suggest that there is possibly less ATP being made even with an increase in expression of phosphoglycerate kinase. In the final ATP-generating step
111 of glycolysis, where phosphoenolpyruvate is converted to pyruvate, we do not detect any significant changes in the metabolites (Figure 23Dand E), even with an increase in enzyme levels.
The final product of glycolysis, pyruvate, was detected in the metabolomic analysis and was found at levels were not significantly different in the ACSVL3 KO cells
(Figure 24). Although both lactate dehydrogenase enzymes (LDH-A and LDH-B) were higher in KO cells, cellular lactate levels were not significantly higher (Figure 25).
However, other investigators in the lab measured lactate secreted into the culture medium after a one hour incubation and found that KO cells secreted ~1.3-fold more of this metabolite than U87MG cells (Y Liu and PA Watkins, unpublished). This may partly explain the observed upregulation of glycolytic enzymes without significant intracellular elevations in levels of pathway intermediates. Lactate levels are generally thought to promote a more tumorigenic phenotype. In contrast, cellular plus secreted lactate was higher in the less tumorigenic ACSVL3 KO cells. We conclude that the higher growth and proliferation levels seen in U87MG cells are not due to increased lactate production.
112
Table 3. Glycolytic enzyme levels are increased in ACSVL3 KO cells. Proteomic analysis reveals an increase in the expression of glycolytic enzymes in the ACSVL3 KO cells when compared to the U87MG cells. Some of these increases do not reach 2-fold that of what is found in U87MG (Hexokinase 1/2), however there are several large changes that are more than 4-fold higher than the parental cell line.
113
Figure 24. Glycolytic pathway metabolites do not significantly change despite a
change in enzymes. Glycolytic metabolites were extracted from U87MG cells and
ACSVL3 cells and were measured using Q-TOF LC-MS (n = 6). Data was normalized to protein in each sample. (A) Glucose 6-phosphate, (B) glyceraldehyde 3-phosphate, (C)
fructose 1,6-bisphosphate, (D) glycerate 3-phosphate, and (E) phosphoenol pyruvate
were all detected in the metabolomics study. The only metabolite that exhibited a
significant change was glycerate 3-phosphate (p = 0.01).
114
Figure 25. Pyruvate levels do not change with an ACSVL3 KO. Despite the increase in pyruvate kinase from the proteomic data, we do not see a change in pyruvate levels between the U87MG cells and the ACSVL3 KO cells.
115
Figure 26. LDH enzyme levels increase in an ACSVL3 KO, but cellular lactate levels are unchanged. (A) Proteomic analysis revealed that both LDH-A and LDH-B were higher in the ACSVL3 KO cells relative to the U87MG cells. (B) Metabolomic analysis showed that cellular lactate levels were not significantly different between
U87MG and ACSVL3 KO cells.
116 Glucose Oxidation to CO2 is Downregulated in ACSVL3 KO Cells
Glycolysis is a cytosolic process that intersects with processes that occur within
the mitochondria. Pyruvate, the end product of glycolysis, enters TCA cycle in the
mitochondria to produce CO2 and electron donors for the electron transport chain. A
glucose oxidation assay, which measures glucose conversion to CO2, was conducted to
determine whether this pathway was affected by lack of ACSVL3. We incubated both
U87MG and KO cell lines with uniformly labeled 14C- D-glucose. The reaction was stopped with 60% perchloric acid to release the CO2 from the media to be captured on filter paper with 1M KOH. We find that the U87MG cells converted significantly more glucose into CO2 than did the ACSVL3 KO cells do (Figure #26). This observation
suggests that depletion of ACSVL3 affects TCA cycle activity, flux through the electron
transport chain, or both.
117
Figure 27. Oxidation of glucose is significantly downregulated in the ACSVL3 KO.
Cells of both lines were treated with uniformly labeled 14C-D-glucose to look at how much glucose is converted to CO2. The U87MG cells converted significantly more glucose to CO2 (2.65 nmol/20 min/mg protein) than the ACSVL3 KO cells (0.39 nmol/20 min/mg protein). N = 4; p = 0.006.
118 Levels of the Citric Acid Cycle Enzymes are Lower in ACSVL3 KO Cells Without
Changing the Metabolite Profile
We next wanted to see was if there were any differences between the U87MG
cells and the ACSVL3 KO cells in regards to the tricarboxylic acid cycle (TCA) proteins
and metabolites. While we saw large changes in the glycolytic enzymes in the ACSVL3
KO cells, we wondered if the TCA enzymes would follow the same trend. We were able
to identify all of the enzymes involved, and we find that there is an overall decrease in
levels in the ACSVL3 KO cells (Table 4), in contrast to the glycolytic enzymes which
were elevated. Pyruvate dehydrogenase E1 is the first component enzyme of the
pyruvate dehydrogenase complex (PDC) and catalyzes the conversion of pyruvate to
lipoic acid. This is considered the rate-limiting step of the complex. We find that this enzyme is reduced by half in the ACSVL3 KO cells, with the other enzyme components of the PDC also reduced to some degree. The only enzyme involved in the TCA cycle that does not seem to decrease is citrate synthase, which catalyzes the synthesis of citrate from oxaloacetate and acetyl-CoA which is needed for lipid synthesis.
The metabolomic data was then analyzed for TCA cycle intermediate levels.
Seven of the eight metabolic intermediates were identified in the analysis (Figure 27).
Isocitrate was not individually detected due to the similarities between it and its precursor, citrate. Like the data for glycolysis, even with a decrease in TCA enzymes, we do not see significant changes for most of the TCA intermediates. While we may be seeing trends towards higher levels of certain intermediates in the ACSVL3 KO cells, like citrate and isocitrate, we cannot conclude that these changes are significant.
Metabolomic analysis detected a significant increase (p = 0.017) in α-ketoglutarate in the
119 ACSVL3 KO cells. This is the product of the isocitrate dehydrogenases (IDH). We find that the mitochondrial isoforms of this enzyme are lower, however the cytosolic isoform is increased in the ACSVL3 KO cells. This potentially confounds interpretation of the metabolomics because we cannot say with certainty that these levels accurately reflect the
TCA cycle of the cells. The other increase in the ACSVL3 KO cells we detect is with succinate (p = 0.002), but we do not see a difference in the next intermediate of the TCA cycle, fumarate, between the two cell lines. This could be due to a decrease in succinate dehydrogenase, but this will have to be explored in further detail. Even though there were not large changes in TCA cycle intermediates in the ACSVL3 KO cells, decreased
TCA enzyme levels may help explain the decreased glucose oxidation to CO2 seen in these cells.
120
Table 4. Levels of TCA cycle enzymes are decreased in ACSVL3 KO cells.
Proteomic analysis identified the enzymes of the TCA cycle in the U87MG cells and the
ACSVL3 KO cells. The TCA cycle enzymes, except for citrate synthase, were found to be decreased. Citrate synthase was found to be equal between the two cell lines.
121
Figure 28. TCA Cycle intermediates are not changed overall in the ACSVL3 KO cells. Metabolomic analysis detected six TCA cycle intermediates in both the KO cells and the U87MG cells – (A) Citrate, (B) α-Ketoglutarate, (C) Succinate, (D) Fumarate, (E)
Malate, and (F) Oxaloacetate. Significant increases are found in α-ketoglutarate (p =
0.017) and succinate levels (p =0.002). These measurements are total cellular levels, and may not reflect only mitochondrial metabolism. Overall, we do not see increases in TCA cycle intermediates.
122 Electron Transport Chain Complexes and ATP Synthase Subunits are Lower in the
ACSVL3 KO Cells
When we observed the decrease in the enzymes involved in the TCA cycle, we asked if the enzymes of processes downstream, e.g. the electron transport chain (ETC) and oxidative phosphorylation, would be affected as well. Levels of ETC complexes I,
II, III, and IV were all, in fact, decreased in the ACSVL3 KO cells (Table 5). Most of the subunits of ATP synthase were also decreased (Table 6). These observations, along with decreased levels of TCA cycle enzymes in ACSVL3 KO cells, are in agreement with the lower rate of glucose oxidation to CO2. Further studies will be needed to clarify the mechanism(s) that underlie changes in mitochondrial function caused by overexpression of ACSVL3 in U87MG cells.
123
Table 5. The complexes of the Electron Transport Chain are lower in the ACSVL3
KO cells. Proteomics identified most subunits of each of the four complexes of the ETC and analysis shows that expression of most, with the exception of a subunit of Complex I and one in Complex IV, is downregulated in the ACSVL3 KO cells.
124
Table 6. ATP synthase subunits levels are decreased in the ACSVL3 KO cells.
Proteomic analysis also reveals that there is a decrease in ATP synthase subunits which may lead to energetic differences between the ACSVL3 KO cells and the U87MG cells.
125 Changes in Mitochondrial Morphology Occur When U87MG Cells Are ACSVL3-
Deficient
Studies of glucose oxidation, along with proteomic and metabolomics analysis of
TCA cycle and the ETC, suggested that depletion of ACSVL3 affects mitochondrial
function. We therefore asked if this was accompanied by differences in mitochondrial
morphology. Mitochondrial morphology varies across different cell types, and the
regulation of fusion and fission responds rapidly to metabolic cues (Benard et al., 2007).
We performed immunoflurorescence (IF) using a rabbit antibody against Tom20, a mitochondrial outer membrane transporter, and found that there were striking differences in mitochondrial morphology in the U87MG and ACSVL3 KO cell lines. The mitochondria in the U87MG cell line appear more fragmented, while the mitochondria in the ACSVL3 KO cells are more fused and form reticular structures (Figure 28).
We referred to the proteomic data to see if there were any differences in proteins that are involved in mitochondrial fission and fusion that would explain these morphological differences (Table 7). These two functions have been shown to play important roles in cancer. Rehman et al. (2012) showed in human lung cancer cell lines that the cells exhibited excess mitochondrial fission. While we do see decreases in proteins that are involved in fusion in the ACSVL3 KO cells, it is the decrease that we see with the protein Mtfp1 that more likely accounts for the differences in morphology we are seeing in the mitochondria. Mtfp1 is a protein important in mitochondrial fission and when cells are depleted of only this protein, fission cannot occur (Tondera et al.,
2005).
126 Overall, results presented in this chapter suggest that ACSVL3 indirectly affects both glucose homeostasis and mitochondrial function and morphology, along with its more direct effects on sphingolipid metabolism. One possible explanation under consideration is an effect of ACSVL3 on mitochondrial membrane lipid composition.
Future experimentation will address this hypothesis.
127
Figure 29. Mitochondrial morphology is different between the U87MG cell line and the ACSVL3 KO line. Fluorescent microscopy using an antibody against Tom20, an outer mitochondrial membrane translocase, found morphological differences between the
U87MG cells and the ACSVL3 KO cells. The U87MG cell mitochondria appear more punctate while the mitochondria of the ACSVL3 KO cells are appear to be more filamentous.
128
Table 7. Proteins involved with mitochondrial dynamics show differences in the
ACSVL3 KO cells. Proteins involved in mitochondrial fusion and fission were identified in the proteomics study. Fusion proteins are decreased while most fission proteins are increased. Mtfp1, an important fission protein, is the only one decreased, and may indicate a failure for the ACSVL3 KO mitochondria to undergo fission.
129 Discussion
Upregulated expression of ACSVL3 in U87MG cells alters lipid metabolism by
increasing sphingolipid synthesis as well as increasing cholesterol levels when incubated with
growth factor relative to cells that are deficient in ACSVL3 enzyme, as shown in the previous
chapter, as we expected. Results presented in this chapter show that ACSVL3 also affects
glucose metabolism. While this was somewhat unexpected, regulatory links between glucose and
lipid metabolism are important control mechanisms for overall cellular energy homeostasis. For
example, the TCA cycle is an important crossroad between glucose and lipid metabolism.
Pyruvate, the end product of glycolysis, is metabolized by the pyruvate dehydrogenase complex
and then enters the TCA cycle as acetyl-CoA. Acetyl-CoA can be completely oxidized to CO2 for energy production, or can be used as building blocks for fatty acid and cholesterol synthesis.
More complex regulatory mechanisms linking glucose and lipid metabolism are also being discovered. A recent study showed that glucose can regulate lipid metabolism through N- glycosylation of the protein SREBP-cleavage activating protein (SCAP) to activate the SREBP transcription factors that induce lipid metabolism genes such as FASN and the LDL receptor
(Cheng et al, 2015). Aberrant glucose metabolism in cancer has been an extensive subject of study since Otto Warburg made his observation that cancer cells have increased glycolysis, even in the presence of oxygen. Findings presented in this chapter show that ACSVL3 KO in the
U87MG cell line affects glucose metabolism and mitochondrial function.
It is interesting that we found marked differences in two pathways that are linked through pyruvate – we found an increase in glycolytic enzymes and a decrease in enzymes of the TCA cycle in the ACSVL3 KO cells. However, even with these changes in enzyme levels, we found that the levels of the metabolic intermediates did not reflect these changes. Even with the decrease in the TCA cycle enzymes in ACSVL3 KO cells when compared to the U87MG cells,
130 we found some of the intermediates – such as succinate, α-ketoglutarate, and citrate – were higher in the KO cells. We looked at all of the different isozymes of the proteins important in the TCA cycle, and proteomic analysis detected both cytoplasmic and mitochondrial forms. For example, all three IDH isoforms were found, with the cytoplasmic isoforms more abundant in the ACSVL3
KO cells compared to the parental U87MG cells. The same trend was found for the malate dehydrogenase isoforms. This is important when considering the metabolomics data of the two different cell lines. While the levels of all of the mitochondrial enzymes are lower in the
ACSVL3 KO cells, the cytoplasmic isoforms could be contributing to 1) the overall level of the particular metabolite since analysis was a whole cell analysis, and 2) the metabolites could be shuttled into the mitochondria and fuel the production of the subsequent metabolites in the pathway.
The results of proteomics and metabolomics studies, in conjunction with follow-up experiments, indicated that there might be changes in mitochondrial structure when cells were depleted of ACSVL3. When we compared the cytoplasmic enzymatic isoforms to the mitochondrial isoforms of some proteins, we saw that there was an increase in the cytoplasmic isoforms in the ACSVL3 KO cells. Since there weren’t significant changes in metabolomics, we speculated that these differences originated from the mitochondria themselves. Mitochondrial morphology, particularly the dynamics of the mitochondria, has been shown to play an important role in cancer metastasis. A recent study showed that human lung cancer cell lines exhibited excess mitochondrial fission and impaired mitochondrial fusion (Rehman et al., 2012). In our study with the ACSVL3 KO cell line we find that the KO cells have network-like mitochondria, whereas the U87MG cells have more punctate mitochondria. We examined our proteomic data to see if we could find changes in expression of proteins that were important in fission and fusion, as our KO cells might have a defect in fission. We found that surprisingly, fusion proteins are low in the ACSVL3 KO cells, but fission proteins are higher, except for Mtfp1, also known as Mtp18.
This protein was found to be essential for mitochondrial fission, as loss of function (by RNAi)
131 resulted in highly fused mitochondria, even with an overexpression of another fission protein,
Fis1 (Tondera et al., 2005). This decrease in Mtp18 may be sufficient to prevent fission and keep
mitochondria in a highly fused form, despite the overexpression of other fission proteins.
Of course, since ACSVL3 is an enzyme responsible for the activation of fatty acids, we cannot ignore the possibility that what we see with mitochondrial morphology and the differences in mitochondrial enzymes is due to differences in mitochondrial membrane lipid components.
Cardiolipin is an important phospholipid of the mitochondrial inner membrane. It is unusual in that it is a diphosphatidylglycerol consisting of three glycerol backbones and four acyl chains, resulting in a very specific role in mitochondrial function. It has been implicated in the formation of enzyme complexes of the electron transport chain and in the mitochondrial apoptotic pathway.
However, when phospholipid synthesis from [1-14C]acetate was assessed in U87MG and
ACSVL3 KO cells, cardiolipin synthesis was below the level of detection on thin layer
chromatography plates. Further studies will be needed to resolve this issue.
This is the first time our lab has associated ACSVL3 with changes in mitochondria.
Mitochondria have been at the center of cancer metabolism for nearly a century. We have found
interesting data that raises only more questions as to the function of ACSVL3 in cancer cells.
This is not an exhaustive study as to why ACSVL3 is important in mitochondrial dynamics and
metabolism, but it opens up our lab to new questions relating ACSVL3 to the overall importance
of cancer metabolism, and more specifically, the importance of our enzyme to mitochondria. To
try to further answer the question of the importance of ACSVL3, studies with isolated
mitochondria will need to be performed to reduce any background from similar biological
processes occurring in other cellular compartments. However, we do believe that this has laid the
foundation to show that ACSVL3 is important in gliomas and other cancers.
132 REFERENCES
Abounader, R., Ranganathan, S., Lal, B., Fielding, K., Book, A., Dietz, H., Burger, P., and Laterra, J. (1999). Reversion of human glioblastoma malignancy by U1 small nuclear RNA/ribozyme targeting of scatter factor/hepatocyte growth factor and c-met expression. J Natl Cancer Inst 91, 1548-56.
Abounader, R., Lal, B., Luddy, C., Koe, G., Davidson, B., Rosen, E.M., and Laterra, J. (2002). In vivo targeting of SF/HGF and c-met expression via U1snRNA/ribozymes inhibits glioma growth and angiogenesis and promotes apoptosis. FASEB J 16, 108-10.
Abounader, R. and Laterra, J. (2005). Scatter factor/hepatocyte growth factor in brain tumor growth and angiogenesis. Neuro Oncol 7, 436-51.
Amann, T. and Hellerbrand, C. (2009). GLUT1 as a therapeutic target in hepatocellular carcinoma. Expert Opin Ther Targets 13, 1411-27.
Annabi, B., Lachambre, M.P., Plouffe, K., Sartelet, H., and Beliveau, R. (2009). Modulation of invasive properties of CD133+ glioblastoma stem cells: a role for MT1- MMP in bioactive lysophospholipid signaling. Mol Carcinog 48, 910-9.
Bernhart, E., Damm, S., Wintersperger, A., Nusshold, C., Brunner, A.M., Plastira, I., Rechberger, G., Reicher, H., Wadsack, C., Zimmer, A., Malle, E., and Sattler, W. (2015). Interference with distinct steps of sphingolipid synthesis and signaling attenuates proliferation of U87MG glioma cells. Biochem Pharmacol 96, 119-30.
Benard, G., Bellance, N., James, D., Parrone, P., Fernandez, H., Letellier, T., and Rossignol, R. (2007). Mitochondrial bioenergetics and structural network organization. J. Cell Sci. 120, 838-848.
Birchmeier, C., Birchmeier, W., Gherardi, E., and Vande Woude, G.F. (2003). Met, metastasis, motility and more. Nat Rev Mol Cell Biol 4, 915-25.
Birkenkamp-Demtroder, K., Christensen, L.L., Olesen, S.H., Frederiksen, C.M., Laiho, P., Aaltonen, L.A., Laurberg, S., Sorensen, F.B., Hagemann, R., and ORntoft, T.F. (2002). Gene expression in colorectal cancer. Cancer Res 62, 4352-63.
Blume-Jensen, P. and Hunter, T. (2001). Oncogenic kinase signaling. Nature 411, 355- 65.
Bremer, E.G., Schlessinger, J., and Hakomori, S. (1986). Ganglioside-mediated modulation of cell growth. Specific effects of GM3 on tyrosine phosphorylation of the epidermal growth factor receptor. J Biol Chem 261, 2434-40.
133 Cao, Y., Pearman, A.T., Zimmerman, G.A., McIntyre, T.M., Prescott, S.M. (2000). Intracellular unesterified arachidonic acid signals apoptosis. Proc Natl Acad Sci USA 97, 11280-5.
Cao, Y., Dave, K.B., Doan, T.P., and Prescott, S.M. (2001). Fatty acid CoA ligase 4 is up-regulated in colon adenocarcinoma. Cancer Res 61, 8429-34.
Central Brain Tumor Registry of the United States. (2006).
Cheng, C., Ru, P., Geng, F., Liu, J., Yoo, J.Y., Wu, X., Cheng, X., Euthine, V., Hu, P., Guo, J.Y., Lefai, E., Kaur, B., Nohturfft, A., Ma, J., Chakravarti, A., and Guo, D. (2015). Glucose-mediated N-glycosylation of SCAP is essential for SREBP-1 activation and tumor growth. Cancer Cell 28, 569-81.
Ciardiello, F. and Tortora, G. (2008). EGFR antagonists in cancer treatment. N Engl J Med 358, 1160-74.
Clark, M.J., Homer, N., O’Connor, B.D., Chen, Z., Eskin, A., Lee, H., Merriman, B., and Nelson, S.F. (2010). U87MG decoded: the genomic sequence of a cytogenetically aberrant human cancer cell line. Plos Genetics DOI: 10.1371/journal.pgen.1000832
Cooper, G.M. (2000). The Eukaryotic Cell Cycle. The Cell: A Molecular Approach, 2nd edition. Sunderland (MA): Sinauer Associates.
Cuvillier, O., Pirianov, G., Kleurser, B., Vanek, P.G., Coso, O.A., Gutkind, S., and Spiegel, S. (1996). Suppression of ceramide-mediated programmed cell death by sphingosine 1-phosphate. Nature 381, 800-803.
Dang, L., White, D.W., Gross, S., Bennett, B.D., Bittinger, M.A., Driggers, E.M., Fantin, V.R., Jang, H.G., Jin, S., Keenan, M.C., Marks, K.M., Prins R.M., Ward, P.S., Yen, K.E., Liau, L.M., Rabinowitz, J.D., Cantley, L.C., Thompson, C.B., Vander-Heiden, M.G., and Su, S.M. (2009). Cancer-associated IDH1 mutations produce 2-hydroxyglutarate. Nature 462, 739-744.
Danilkovitch-Miagkova, A. and Zbar, B. (2002). Dysregulation of Met receptor tyrosine kinase activity in invasive tumors. J Clin Invest 109, 863-7.
DeBerardinis, R.J., Lum, J.J., and Thompson, C.B. (2006). Phosphatidylinositol 3- kinase-dependent modulation of carnitine palmitoyltransferase 1A expression regulates lipid metabolism during hematopoietic cell growth. J Biol Chem 281, 37372-80.
DeBerardinis, R.J., Lum, J.J., Hatzivassiliou, G., and Thompson, C.B. (2008). The biology of cancer: metabolic reprogramming fuels cell growth and proliferation. Cell Metab 7, 11-20.
134 Dole, V.P., (1956). A relation between non-esterified fatty acids in plasma and the metabolism of glucose. J Clin Invest 35, 150-4.
Durante, S., Orienti, I., Teti, G., Salvatore, V., Focaroli, S., Tesei, A., Pignatta, S., and Falconi, M. (2014). Anti-tumor activity of fenretinide complexed with human serum albumin in lung cancer xenograft mouse model. Oncotarget 5, 4811-4820.
Eberle, D., Hegarty, B., Bossard, P., Ferre, P., and Foufelle, F. (2004). SREBP transcriptions factors: master regulators of lipid homeostasis. Biochimie 86, 839-48.
Folch, J., Lees, M., and Sloane-Stanley, G.H. (1957). A simple method for the isolation and purification of total lipides from animal tissues. J Biol Chem
Gandy, KA and Obeid, L.M. (2013). Regulation of the sphingosine kinase/sphingosine 1-phosphate pathway. Handb Exp Pharmacol 216, 275-303.
Gassler, N., Herr, I., Schneider, A., Penzel, R., Langbein, L., Schirmacher, P., and Kopitz, J. (2005). Impaired expression of acyl-CoA synthetase 5 in sporadic colorectal adenocarcinomas. J Pathol 207, 295-300.
Gillies, R.J. and Gatenby, R.A. (2007). Adaptive landscapes and emergent phenotypes: why do cancers have high glycolysis? J Bioenerg Biomembr 39, 251-257.
Griguer, C.E., Oliva, C.R., and Gillespie, G.Y. (2005). Glucose metabolism heterogeneity in human and mouse malignant glioma cell lines. J Neuro-Onc. 74, 123- 133.
Guo, D., Hildebrandt, I.J., Prins, R.M., Soto, H., Mazzotta, M.M., Dang, J., Czernin, J., Shyy, J.Y.J., Watson, A.D., Phelps, M., Radu, C.G., Cloughesy, T.F., and Mischel P.S. (2009). The AMPK agonist AICAR inhibits the growth of EGFRvIII-expressing glioblastomas by inhibiting lipogenesis. Proc Natl Acad Sci 106, 12932-12937.
Hait, N.C., Sarkar, S., Le Stunff, H., Mikami, A., Maceyka, M., Milstien, S., and Spiegel, S. (2005). Role of sphingosine kinase 2 in cell migration toward epidermal growth factor. J Biol Chem, 280, 29462-9.
Hait, N.C., Oskeritzian, C.A., Paugh, S.W., Milstien, S., and Spiegel, S. (2006). Sphingosine kinases, sphingosine 1-phosphate, apoptosis and disease. Biochim Biophys Acta 1758, 2016-26.
Hakomori, S. (1996). Tumor malignancy defined by aberrant glycosylation and sphingo(glyco)lipid metabolism. Cancer Res 56, 5309-18.
Hakomori, S.I. (2002). The glycosynapse. Proc Natl Acad Sci USA 99, 225-32.
Hanahan, D. and Weinberg, R.A. (2000). The hallmarks of cancer. Cell 100, 57-70.
135
Hanai, N., Dohi, T., Nores, G.A., Hakomori, S. (1988). A novel ganglioside, de-N- acetyl-GM3 (II3NeuNH2LacCer), acting as a strong promoter for epidermal growth factor receptor kinase and as a stimulator for cell growth. J Biol Chem 263, 6296-301.
Heldin, C.H. (1995). Dimerization of cell surface receptors in signal transduction. Cell 80, 213-23.
Hirsch, D., Stahl, A, and Lodish, H.F. (1998). A family of fatty acid transporters conserved from mycobacterium to man. Proc Natl Acad Sci USA 95, 8625-9.
Holthuis, J.C., Pomorski, T., Raggers, R.J., Sprong, H., and Van Meer, G. (2001). The organizing potential of sphingolipids in intracellular membrane transport. Physiol Rev 81, 1689-723.
Horton, J.D., Goldstein, J.L., and Brown, M.S. (2002). SREBPs; activators of the complete program of cholesterol and fatty acid synthesis in the liver. J Clin Invest 109, 1125-31.
Kan, C.F., Singh, A.B., Dong, B., Shende, V.R., and Liu, J. (2015). PPARδ activation induces hepatic long-chain acyl-CoA synthetase 4 expression in vivio and in vitro. Biochim Biophys Acta 1851, 577-87.
Kim, Y.G. and Chandrasegara, S. (1994). Chimeric restriction endonuclease. Proc Natl Acad Sci USA 91, 883-7.
Kim, J.W. and Dang, C.V. (2005). Multifaceted roles of glycolytic enzymes. Trends Biochem Sci 30, 142-50.
Koochekpour, S., Jeffers, M., Rulong, S., Taylor, G., Klineberg, E., Hudson, E.A., Resau, J.H., and VandeWould, G.F. (1997). Met and hepatocyte growth factor/scatter factor expression in human gliomas. Cancer Res 57, 5391-8.
Koppenol, W.H., Bounds, P.L., and Dang, C.V. (2011). Otto Warburg’s contributions to current concepts of cancer metabolism. Nat Rev Cancer 11, 325-337.
Koybasi, S., Senkal C.E., Sundararaj, K., Spassieva, S., Bielawski, J., Osta, W., Day, T.A., Jiang, J.C., Jazwinski, S.M., Hannun, Y.A., Obeid, L.M, and Ogretmen, B. (2004). 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, 44311-9.
Kroemer, G. and Pouyssegur, J. (2008). Tumor cell metabolism: cancer’s Achilles’ heel. Cancer Cell 13, 472-482.
136 Kuhajda, F.P., Jenner, K., Wood, F.W., Henninger, R.A., Jacobs, L.B., Dick, J.D., and Pasternack, G.R. (1994). Fatty acid synthesis: a potential selective target for anti- neoplastic therapy. Proc Natl Acad Sci USA 91, 6379-6383.
Laemmli, U.K. (1970). Cleaveage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680-5.
Laterra, J., Rosen, E., Nam, M., Ranganathan, S., Fielding, K., and Johnston, P. (1997). Scatter factor/hepatocyte growth factor expression enhances human glioblastoma tumorigenicity and growth. Biochm Biophys Res Commun 235, 743-7.
Laviad, E.L., Albee, L., Pankova-Kholmyansky, I., Epstein, S., Park, H., Merrill, A.H. Jr., and Futerman, A.H. (2008). Characterization of ceramide synthase 2: tissue distribution, substrate specificity, and inhibition by sphingosine 1-phosphate. J Biol Chem 283, 5677-84.
LeBleu, V.S., O’Connell, J.T., Gonzalez-Herrera, K.N., Wikman, H., Pantel, K., Haigis, M.C., de Carvalho, F.M., Damascena, A., Domingos-Dhinen, L.T., Rocha, R.M., Asara, J.M., and Kalluri, R. (2014). PGC-1α mediates mitochondrial biogenesis and oxidative phosphorylation in cancer cells to promote metastasis. Nat Cell Biol 16, 992-1003.
Liesa, M., Palacin, M., and Zorzano, A., (2009). Mitochondrial dynamics in mammalian health and disease. Physiol Rev 89, 799-845.
Lin, H., Liu, W., Fang, Z., Liang, X., Li, J., Bai, Y., Lin, L., You, H., Pei, Y., Wang, F., and Zhang Z.Y. (2015). Overexpression of DHX32 contributes to the growth and metastasis of colorectal cancer. Sci Rep 5, 9247.
Lingwood, D. and Simons, K. (2010). Lipid rafts as a membrane-organizing principle. Science 327, 46-50.
Lowry, O.H., Rosebrough, N.J., Farr, A.L., and Randall, R.J. (1951). Protein measurement with the Folin phenol reagent. J Biol Chem 193, 265-75.
Ma, P.C., Kijima, T., Maulik, G., Fox, E.A., Sattler, M., Griffin, J.D., Johnson, B.E., and Salgia, R. (2003). c-MET mutational analysis in small cell lung cancer: novel juxtamembrane domain mutations regulating cytoskeletal functions. Cancer Res 63, 6272-81.
Maceyka, M., Harikumar, K.B., Milstien, S., and Spiegel, S. (2012). Sphingosine 1- phosphate signaling and its role in disease. Trends Cell Biol 22, 50-60.
Macheda, M.L., Rogers, S., and Best, J.D. (2005). Molecular and cellular regulation of glucose transporter (GLUT) proteins in cancer. J Cell Physiol 202, 654-62.
137 Mashima, T., Oh-hara, T., Sato, S., Mochizuki, M., Sugimoto, Y., Yamazaki, K., Hamada, J., Tada, M., Moriuchi, T., Ishikawa, Y., Kato, Y., Tomoda, H., Yamori, T., and Tsuruo, T. (2005). P53-defective tumors with a functional apoptosome-mediated pathway: a new therapeutic target. J Natl Cancer Inst 97, 765-77.
Medes, G., Thomas, A., and Weinhouse, S. (1953). Metabolism of neoplastic tissue. IV. A study of lipid synthesis in neoplastic tissue slices in vitro. Cancer Res 13, 27-29.
Merrill, A.H. and Sandhoff, K. (2002). Sphingolipids: metabolism and cell signaling. D.E. Vance, J.E. Vance (Eds), Biochemistry of Lipids, Lipoproteins and Membranes, Elsevier, Amsterdam, The Netherlands, 373-407.
Mineo, J.F., Bordron, A., Baroncini, M., Ramirez, C., Maurage, C.A., Blond, S., and Dam-Hieu, P. (2007). Prognosis factors of survival time in patients with glioblastoma multiforme: a multivariate analysis of 340 patients. Acta Neurochir 149: 245-253.
Mizutani, Y., Kihara, A., Chiba, H., Tojo, H., and Igarashi, Y. (2008). 2-hydroxy- ceramide synthesis by ceramide synthase family: enzymatic basis for the preference of FA chain length. J Lipid Res 49, 2356-2364.
Morad, S.A. and Cabot, M.C. (2013). Ceramide-orchestrated signaling in cancer cells. Nat Rev Cancer 13, 51-65.
Moreno-Sanchez, R., Rodriguez-Enriquez, S., Marin-Hernandez, A., and Saavedra, E. (2007). Energy metabolism in tumor cells. FEBS J 274, 1393-1418.
Moreno-Sanchez, R., Rodriguez-Enriquez, S., Saavedra, E., Marin-Hernandez, A., and Gallardo-Perez, J.C. (2009). The bioenergetics of cancer: is glycolysis the main ATP supplier in all tumor cells? Biofactors 35, 209-25.
Moriyama, T., Kataoka, H., Hamasuna, R., Yokogami, K., Uehara, H., Kawano, H., Goya, T., Tsubouchi, H., Koono, M., and Wakisaka, S. (1998). Up-regulation of vascular endothelial growth factor induced by hepatocyte growth factor/scatter factor stimulation in human glioma cells. Biochem Biophys Res Commun 249, 73-7.
Mosesson, Y. and Yarden, Y. (2004). Oncogenic growth factor receptors: implications for signal transduction therapy. Semin Cancer Biol 14, 262-270.
Paugh, S.W., Paugh, B.S., Rahman, M., Kapitnov, D., Almenara, J.A., Kordula, T., Milstien, S., Adams, J.K., Zipkin, R.E., Grant, S., and Spiegel, S. (2008). A selective sphingosine kinase 1 inhibitor integrates multiple molecular therapeutic targets in human leukemia. Blood 112, 1382-91.
Pei, Z., Oey, N.A., Zuidervaart, M.M., Jia, Z., Li, Y., Steinberg, S.J., Smith, K.D., and Watkins, P.A. (2003). The acyl-CoA synthetase “bubblegum” (lipidosin): further
138 characterization and role in neuronal fatty acid beta oxidation. J Biol Chem 278, 47070- 8.
Pei, Z., Fraisl, P., Berger, J., Jia, Z., Forss-Petter, S., Watkins, P.A. (2004). Mouse very long-chain Acyl-CoA synthetase 3/fatty acid transport protein 3 catalyzes fatty acid activation but not fatty acid transport in MA-10 cells. J Biol Chem 279, 54454-62.
Pei, Z., Jia, Z., and Watkins, P.A. (2006). The second member of the human and murince bubblegum family is a testis- and brainstem-specific acyl-CoA synthetase. J Biol Chem 281, 6632-41.
Pei, Z., Sun, P., Huang, P., Lal, B., Laterra, J., and Watkins, P.A. (2009). Lipid metabolism enzyme ACSVL3 supports glioblastoma stem cell maintenance and tumorigenicity. Cancer Res 69, 9175- 9182.
Pei, Z., Fraisl, P., Shi, X., Gabrielson, E., Forss-Petter, S., Berger, J., and Watkins, P.A. (2013). Very long-chain acyl-CoA synthetase 3: overexpression and growth dependence in lung cancer. Plos One 8, e69392.
Pettus, B.J., Chalfant, C.E., and Hannun, Y.A. (2002). Ceramide in apoptosis: an overview and current perspectives. Biochim Biophys Acta 1585, 114-25.
Pommier, A.J., Alves, G., Viennois, E., Bernard, S., Communal, Y., Sion, B., Marceau, G., Damon, C., Mouzat, K., Caira, F., Baron, S., and Lobaccaro, J.M. (2010). Liver X receptor activation downregulates AKT survival signaling in lipid rafts and induces apoptosis of prostate cancer cells. Oncogene 29, 2712-23.
Ramanujan, V.K. (2015). Metabolic plasticity in cancer cells: reconnecting mitochondrial function to cancer control. J Cell Sci Ther 6, 211. doi:10.4172/2157- 7013.1000211.
Rehman, J., Zhang, H.J., Toth, P.T., Zhang, Y., marsboom, G., Hong, Z., Salgia, R., Husain, A.N., Wietholt, C., and Archer, S.L. (2012). Inhibition of mitochondrial fission prevents cell cycle progression in lung cancer. FASEB J 26, 2175-86.
Riebeling, C., Allegood, J.C., Wang, E., Merrill, A.H. Jr., and Futerman, A.H. (2003). Two mammalian longevity assurance gene (LAG1) family members, trh1 and trh4, regulate dihydroceramide synthesis using different fatty acyl-CoA donors. J Biol Chem 278, 43452-59.
Rodriguez-Enriquez, S., Hernandez-Esquivel, L., Marin-Hernandez, A., El Hafidi, M., Gallard- Perez, J.C., Hernandez-Resendiz, I., Rodriguez-Zavala, J.S., Pacheco-Velazquez, S.C., and Moreno-Sanchex, R. (2015). Mitochondrial free fatty acid β-oxidation supports oxidative phosphorylation and proliferation in cancer cells. Int J Biochem Cell Biol 65, 209-221.
Rosen, E.M., Laterra, J., Joseph, A., Jin, L., Fuchs, A., Way, D., Witte, M., Weinand, M., and Goldberg, I.D. (1996). Scatter factor expression and regulation in human glial tumors. Int J Cancer 67, 248-255.
139 Rosen, H., Gonzalez-Cabrera, P.J., Sanna, M.G., and Brown, S. (2009). Sphingosine 1- phosphate receptor signaling. Annu Rev Biochem 78, 743-68.
Saddoughi, S.A. and Ogretmen, B. (2013). Diverse functions of ceramide in cancer cell death and proliferation. Adv Cancer Res 117, 37-58.
Samudio, I., Harmancey, R., Fiegl, M., Kantarjian, H., Konopleva, M., Korchin, B., Kaluarachchi, K., Bornmann, W., Duvvuri, S., Taegtmeyer, H., and Andreeff, M. (2010). Pharmacologic inhibition of ffatty acid oxidation sensitizes human leukemia cells to apoptosis induction. J Clin Invest 120, 142-56.
Sanchez-Madrid, F. and Serrador, J.M. (2009). Bringing up the rear: defining the roles of the uropod. Nat Rev Mol Cell Biol 10, 353-359.
Sathornsumetee, S., Reardon, D.A., Desjardins, A., Quinn, J.A., Vredenburgh, J.J., Rich, J.N. (2007). Molecularly targeted therapy for malignant glioma. Cancer 110, 13-24.
Scheffler, I.E. (2008). Mitochondria. John Wiley & Sons, Inc, 2nd Edition. Schiffmann, S., Sandner, J., Birod, K., Wobst, I., Angioni, C., Ruckhaberle, E., Kaufmann, M., Ackermann, H., Lotsch, J., Schmidt, H., Geisslinger, G., and Grosch, S. (2009). Ceramide synthases and ceramide levels are increased in breast cancer tissue. Carcinogenesis 30, 745-52.
Schnaar, R.L. and Needham, L.K. (1994). Thin-layer chromatography of glycosphingolipids. Methods Enzymol 230, 371-389.
Shaner, R.L., Allegood, J.C., Park, H., Wang, E., Kelly, S., Haynes, C.A., Sullards, M.C., Merrill, A.H. Jr. (2009). Quantitative analysis of sphingolipids for lipidomics using triple quadrupole and quadrupole linear ion trap mass spectrometers. J Lipid Res 50, 1692-707.
Silveira, S.M., Villacis, R.A.R., Marchi, F.A, Barros Filho, M.C., Drigo, C.S., Neto, CS., Lopes, A., Werneck da Cunha, I., and Rogatto, S.R. (2013). Genomic signatures predict poor outcome in undifferentiated pleomorphic sarcomas and leiomyosarcomas. Plos One
Simons, K. and Ikonen, E. (1997). Functional rafts in cell membranes. Nature 387, 569- 72.
Simons, K. and Ikonen, E. (2000). How cells handle cholesterol. Science 290, 1721-6.
Steinberg, S.J., Morgenthaler, J., Heinzer, A.K., Smith, K.D., and Watkins, P.A. (2000). Very long-chain acyl-CoA synthetases. Human “bubblegum” represents a new family of proteins capable of activating very long-chain fatty acids. J Biol Chem 275, 35162-9.
Stiban, J., Tidhar, R., and Futerman, A.H. (2010). Ceramide synthases: roles in cell physiology and signaling. Adv Exp Med Biol 688, 60-71.
140 Sumantran, V.N., Mishra, P., and Sudhakar, N. (2015). Microarray analysis of differentially expressed genes regulating lipid metabolism during melanoma progression. Indian J Biochem Biophys 52, 125-31.
Sun, P., Xia, S., Lal, B., Shi, X., Yang, K.S., Watkins, P.A., and Laterra, J. (2014). Lipid metabolism enzyme ACSVL3 supports glioblastoma stem cell maintenance and tumorigenicity. BMC Cancer 14, doi: 10.1186/1471-2407-14-401
Taha, T.A., Argraves, K.M., and Obeid, L.M. (2004). Sphingosine 1-phosphate receptors: receptor specificity versus functional redundancy. Biochim Biophys Acta 1682, 48-55.
Tirado-Velez, J.M., Joumady, I., Saez-Benito, A., Cozar-Castellano, I., and Perdomo, G. (2012). Inhibition of fatty acid metabolism reduces human myeloma cells proliferation. Plos One 7, e46484. doi:10.1371/journal.pone.0046484 Tondera, D., Czauderna, F., Paulick, K., Schwarzer, R., Kaufmann, J., and Santel, A. (2005). The mitochondrial protein MTP18 contributes to mitochondrial fission in mammalian cells. J Cell Sci 118, 3049-59.
Ullrich, A. and Schlessinger, J. (1990). Signal transduction by receptors with tyrosine kinase activity. Cell 61, 203-12.
Van Brocklyn, J., Letterle, C, Snyder, P., and Prior, T. (2002). Sphingosine 1-phosphate stimulates human glioma cell proliferation through Gi-coupled receptors: role of ERK MAP kinase and phosphatidylinositol 3-kinase beta. Cancer Letters 181, 195-204.
Van Brocklyn, J.R., Young, N., and Roof, R. (2003). Sphingosine 1-phosphate stimulates motility and invasiveness of human glioblastoma multiforme cells. Cancer Letters 199, 53-60.
Vander Heiden, M.G., Cantley, L.C., and Thompson, C.B. (2009). Understanding the Warburg effect: the metabolic requirements of cell proliferation. Science 324, 1029-33.
Venktaraman, K., Riebeling, C., Bodennec, J., Riezman, H., Allegood, J.C., Sullards, M.C., Merrill, A.H. Jr, and Futerman, A.H. (2002). Upstream of growth and idifferentiation 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, 35642-9.
Veret, J., Coant, N., Berdyshev, E.V., Skobeleva, A., Therville, N., Bailbe, D., Gorshkova, I., Natarajan, V., Portha, B., and Le Stunff, H. (2011). Ceramide synthase 4 and de novo production of ceramides with specific N-acyl chain lengths are involved in glucolipotoxicity-induced apoptosis of INS-1 β-cells. Biochem J 438, 177-89.
141 Walenta, S., Schroeder, T., and Mueller-Klieser, W. (2016). Lactate in solid malignant tumors: potential basis of a metabolic classification in clinical oncology. Curr Med Chem 23, 2195-2204.
Wang, T.Y. and Silvius, J.R. (2000). Different sphingolipids show differential partitioning into sphingolipid/cholesterol-rich domains in lipid bilayers. Biophys J 79, 1478-89.
Wang, T.Y., Leventis, R., and Silvius, J.R. (2000). Fluorescence-based evaluation of the partitioning of lipids and lapidated peptides into liquid-ordered lipid microdomains: a model for molecular partitioning into “lipid rafts”. Biophys J 79, 919-33.
Warburg, O., Wind, F., and Negelein, E. (1927). The metabolism of tumors in the body. J Gen Physiol 8, 519-30.
Warburg, O. (1930). The Metabolism of Tumors. Constable and Company, Ltd. London. 327 pp.
Warburg, O. (1956). On the origin of cancer cells. Science 123, 309-14.
Watkins, P.A., Pevsner, J., and Steinberg, S.J. (1999). Human very long-chain acyl-CoA synthetase and two human homologs: initial characterization and relationship to fatty acid transport protein. Prostaglandins Leukot Essent Fatty Acids 60, 323-8.
Watkins, P.A. (2007). Fatty acid activation. Prog Lipid Res 36, 55-83.
Watkins, P.A., Maiguel, D., Jia, Z., and Pevsner, J. (2007). Evidence for 26 distinct acyl-Coenzyme A synthetase genes in the human genome. J Lipid Res 48, 2736-50.
Watkins, P.A. (2008). Very long-chain acyl-CoA synthetases. J. Biol Chem 283, 1773- 1777.
Weller, M., van den Bent, M., Hopkins, K., Tonn, J.C., Stupp, R., Falini, A., Cohen- Jonathan-Moyal, E., Frappaz, D., Henriksson, R., Balana, C., Chinot, O., Ram, Z., Reifenberger, G., Soffietti, R., Wick, W., European Association for Neuro-Oncology (EANO) Task Force on malignant Glioma. (2014). EANO guidelines for the diagnosis and treatment of anaplastic gliomas and glioblastoma. Lancet Oncol 15, e395-403.
Westermann, B. (2010). Mitochondrial fusion and fission in cell life and death. Nat Rev Mol Cell Biol 11, 872-884.
Xiao, M., Yang, H., Xu, W., Ma, S., Lin, H., Zhu, H., Liu, L., Liu, Y., Yang, C., Xu, Y., Zhao, S., Ye, D., Xiong, Y., Guan, K.L. (2012). Inhibition of α-KG-dependent histone and DNA demethylases by fumarate and succinate that are accumulated in mutations of FH and SDH tumor suppressors. Genes Dev 26, 1326-1338.
142 Young, N., Pearl, D.K., and Van Brocklyn, J.R. (2009). Sphingosine 1-phosphate regulates glioblastoma cell invasiveness through the urokinase plasminogen activator system and CCN1/Cyr61. Mol Cancer Res 7, 23-32.
Zhao, J., Zhang, J., Yu, M., Xie, Y., Huang, Y., Wolff, D.W., Abel, P.W., and Tu, Y. (2013). Mitochondrial dynamics regulates migration and invasion of breast cancer cells. Oncogene 32, 4814-24.
143 CURRICULUM VITAE FOR Ph.D. CANDIDATES
THE JOHNS HOPKINS UNIVERSITY SCHOOL OF MEDICINE
ELIZABETH ANNE KOLAR MARCH 18, 2016
15 Charles Plaza Apt 1105 Baltimore, MD 21201 [email protected] 410-207-8320
EDUCATIONAL EXPERIENCE
Ph.D. expected 2016 BCMB Program Johns Hopkins School of Medicine Mentor: Paul A. Watkins
B.S. 2007 Cellular and Molecular Biology Salisbury University Minor: Chemistry
OTHER PROFESSIONAL EXPERIENCE Thesis Research 2010-2016 Lab of Paul Watkins Johns Hopkins SOM
Identified novel role for one of the very long chain acyl-CoA synthetases (ACSVL3) that is upregulated in malignant glioma in U87 cells. An ACSVL3 knockout line using zinc finger nuclease technology in the parental U87 cells results in a decrease in cell growth, smaller subcutaneous tumors, decreased fatty acid activation and sphingolipid synthesis. We also find that this enzyme affects mitochondrial morphology and metabolism.
Post-Baccalaureate research 2007-2008 Lab of Michael Lichten National Cancer Institute
Examined the role of enzymes that are important in the resolution of joint molecules during meiotic DNA recombination in Sacchromyces cerevisiae.
Undergraduate Research 2004-2007 Lab of Kimberly Hunter Salisbury University
Investigated the effects of ploidy level on nordihydroguaiaretic acid (NDGA) in the leaves of the creosote bush, Larrea tridentata. Also examined microsatellite regions of the Larrea genome as a new way to examine the phylogenetic history of the genus.
FELLOWSHIPS
F31 Pre-Doctoral Fellowship 2011-2014 National Institutes of Health, NINDS
144 AWARDS AND HONORS
Graduate Student Travel Award 2012 ASBMB
Was awarded a travel award in the amount of $1,000 from the American Society of Biochemistry and Molecular Biology to travel to the annual conference, Experimental Biology, in San Diego, CA.
Young Botanist of the Year 2007 Botanical Society of America
Recognized nationally for contributions made to the field of Botany through undergraduate research.
PUBLICATIONS, PEER REVIEWED
DeMuyt A, Jessop L, Kolar E, Sourirajan A, Chen J, Dayani Y, Lichten M. (2012) BLM helicase ortholog Sgs1 is a central regulator of meiotic recombination intermediate metabolism. Molecular Cell 46(1):43-53. PMCID: PMC3328772
Jordan P, Copsey A, Newnham L, Kolar E, Lichten M, Hoffmann E. (2009) Ipl1/Aurora B kinase coordinates synaptonemal complex disassembly with cell cycle progression and crossover formation in budding yeast meiosis. Genes and Development 23(18):2237-51. PMCID: PMC2751982
Gradolatto A, Smart SK, Byrum S, Blair LP, Rogers RS, Kolar EA, Lavender H, Larson SK, Aitchison JD, Taverna SD, Tackett AJ. (2009) A noncanonical bromodomain in the AAA ATPase protein Yta7 directs chromosomal positioning and barrier chromatin activity. Molecular and Cellular Biology 29(17):4604-11. PMCID: PMC2725702
PRESENTATIONS
Kolar EA, Richard JP, Pei Z, Laterra J, and Watkins PA. (2012) Importance of very long chain acyl-CoA synthetase 3 (ACSVL3) in cholesterol homeostasis and lipid raft signaling in U87 glioma cells. Experimental Biology 2012, San Diego, CA, April 21-25, 2012 (presenter).
Watkins PA, Pei Z, Kolar EA, Clay EM, Shi X, Laterra J. (2012) Lipid metabolism alterations in U87 glioma cells deficient in very long-chain acyl-CoA synthetase 3 are associated with a less malignant phenotype. Experimental Biology 2012, San Diego, CA, April 21-25, 2012.
Kolar EA, Burnett EN, Miller KR, Hunter R, Hunter K. (2007) Quantification of nordihydroguaiaretic acid (NDGA) in the three ploidy levels of Larrea tridentata. Annual Meeting of the Botanical Society of America, Botany and Plant Biology 2007, Chicago, IL, July 7-12, 2007 (Speaker).
145 Kolar EA, Wiley M, Burnett AL, Parke K, Hunter R, Hunter K. (2005) Development of microsatellite markers in Larrea (Zygophyllaceae): A new way to investigate the evolutionary history of Larrea. Annual Meeting of the Botanical Society of America, Botany 2005, Austin, TX, August 12-15, 2005 (presenter).
SERVICE AND LEADERSHIP
Editor-In-Chief, Hopkins Biotech Network, Johns Hopkins University 2015-present
Recruited and coordinated volunteer writers for the bimonthly newsletter, The Transcript (~2,000 subscribers). Also responsible for reviewing, editing, and the creation of content on the HBN website.
Program Director, Connect-A-Kid (national) 2015-present
Organizes programming for Connect-A-Kid, a national 501(c)(3) organization that mentors young international adoptees and their families. Coordinates 10 teams across the country. Facilitates mentor training and family support. Advises the National Leadership team and Executive Board on upcoming programming for organization. Actively recruits mentors and families for current and new teams.
Team Lead and Mentor, Baltimore Team, Connect-A-Kid 2015-present
Organized the launch of the new Baltimore Connect-A-Kid team. Actively recruited mentors and families. Facilitated communication between the team and families about monthly outings with the adoptees.
Executive Board Member, Adoption Links, DC 2015-present
Worked with three other board members to apply for 501(c)(3) status and incorporation of organization. Communicates with over 400 members via social media. Organize monthly outings for adult adoptees to socialize and share experiences with adoption.
146