1.5.. .9...

...?Mmumm

.

 

 

x... of . $40,.
... ‘qlit (.5...
31".)5...‘ v3.
. .1. Ii’ulit .Ql
hunt-HI.-

0g
«...:

3... ,5
.2.... 1...
uvv‘oomrtk

 

. :2 . E 5.! 33‘ axigjlh': , < 3‘ ‘Vvl
4H, .... , x..- ..w... it ,
t.

.... ..v... . #4....» ..

 

.33.”... a”. : 3......” . x at...

22-»)

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the
dissertation entitled

INTEGRATION OF PHYSIOLOGICAL, METABOLIC AND
GENETIC INFORMATION TO IDENTIFY THE MECHANISMS
OF FREE FATTY ACID TOXICITY TO HEPATOMA CELLS

Ph.D.

presented by

SHIREESH SRIVASTAVA

has been accepted towards fulfillment
of the requirements for the

degree in Chemical Engineering and
Materials Science

 

 

Major Professor’s Signature
/ [Si/0 ?/
l ,

Date

 

MS U is an Affirmative Action/Equal Opportunity Institution

 

.-.—.—.-.—.-.-i-.-._.—.aun-

- —----4-—a-o-.-l-t-l-l-I-o-I- ---

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/CiRC/DaIeDue.indd-p.1

INTEGRATING PHYSIOLOGICAL, METABOLIC AND
GENETIC INFORMATION TO IDENTIFY THE
MECHANISMS OF FREE FATTY ACID TOXICITY TO
HEPATOMA CELLS

By

SHIREESH SRIVASTAVA

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY

Department of Chemical Engineering and Materials Science

2007

 

ABSTRACT
INTEGRATING PHYSIOLOGICAL, METABOLIC AND GENETIC INFORMATION
TO IDENTIFY THE MECHANISMS OF FREE FATTY ACID TOXICITY TO
HEPATOMA CELLS
By
SHIREESH SRIVASTAVA ..

The development of novel high-throughput techniques has made it possible to identify the
global metabolic and genetic responses of the cells under a variety of conditions. The
challenge is to efficiently analyze and relate the multi-source high-throughput data to
generate novel information about cellular responses. The contributions of this thesis is the
characterization of physiological, global metabolic and genetic alterations caused by
different types of free fatty acids (FFAs) and tumor necrosis factor alpha (TNF-a) in the
context of lipotoxicity, and the development of systems-biology fi'ameworks to integrate
multi-source information.

In the first part of the research, the toxicity of different types of free fatty acids was
investigated. Among the different types of FFAs, only saturated FFA (palmitate) was
toxic to the cells. Interactive effects of palmitate and TNF-a were observed on the
toxicity. The reactive oxygen species (ROS) mediating the palmitate-cytotoxicity were
identified to be hydrogen peroxide and hydroxyl radicals. In the next part, the differences
in the metabolism of the cells in response to different types of FF A were investigated by
applying metabolic flux analysis (MFA). It was identified that palmitate reduced the
synthesis of glutathione by reducing cysteine uptake, caused by a reduction in the levels

of cystine transporter xCT. Supplementing cysteine to the cells reduced the toxicity

significantly. The global genetic changes caused by exposure to FFAs and TNF—a were

identified by microarray analyses. Two novel methods were developed to integrate
genomic and metabolic/ physiological information. In the first method, the genes that
regulate multiple cellular responses were identified by the dual response genetic
algorithm partial least squares (GA/PLS) analysis. This method identified that NADH
dehydrogenase and mitogen activated protein kinases (MAPKs) regulate cytotoxicity and
TG accumulation. These predictions were experimentally verified. Finally, a novel
Bayesian regression mixture model-based framework was developed to incorporate the
gene ontology (GO) information to select the genes that could discriminate between the
subpopulations. It was identified that incorporation of prior information improved the
classification of the genes. Important roles of lysosomal ATPases and adenylate cyclase
in mediating the toxicity were identified by this analysis, which were experimentally
validated. Thus, in this research, the multiple alterations responsible for mediating
palmitate-toxicity to hepatoma cells were identified and novel frameworks were

developed to integrate multi-source high-throughput information.

© COPYRIGHT

Shireesh Srivastava

2007

 

DEDICATED TO MY FAMILY

ACKNOWLEDGEMENTS

This research would not have been possible without the support and help I received from
so many people over the past five and a half years. It is likely that I may not explicitly
name everyone who by his/ her actions made it possible for me to come this far. Yet, I

solemnly remember their contributions to my research and my life.

The first person without whose constant patience, support and encouragement, I would
have never completed my research, is my advisor Dr. Christina Chan. Her appreciation of
my efforts, prompt and critical evaluation of my work, coupled with the intellectual
freedom she provided, helped in my development as a researcher. I can only hope to
follow the myriad of things I learned from her. I am also extremely grateful to every
member of my Ph.D. advisory committee for taking time to participate in the committee.
Their interest in my work, coupled with critical evaluation and support, vastly improved
the quality of the research. My special thanks to Prof. Donald Jump for providing novel
insights into my research, as well as for providing the cells when we lost ours to
contamination and personally showing me how to do thin layer chromatography. I would
also like to acknowledge the lmowledge I gained by my interaction with Prof. Mark
Worden, especially during the Advanced Biochemical Engineering class. I am confident
that what I learnt in that class would continue to help me in the future, as it did in the
current research. Prof. Dale, too was always supportive of my research. I would like to
thank him for always taking time from his busy schedule to attend the committee
meetings, asking pertinent questions and providing useful feedback. I am also grateful to

Dr. S.P. Walton (Pat) for always being appreciative of my work and efforts and providing

vi

useful suggestions. I would also like to thank Dr. Rong Jin of Computer Science

department for interesting discussions on microarray data analysis.

The unwavering support and confidence of my family and friends made it possible for me
to walk through the thick and thins of research. I am thankful to my parents for the
hardships they bore to help me study whatever and wherever I wanted. My wife, Sheenu,
had many contributions in making this research possible. I am thankful to her for all the
technical, emotional (and dietary) support she provided. My brother, Aasheesh, was a

constant source of knowledge, evaluation and inspiration.

I am very grateful to my friends and lab-members for their support and insightful
discussions. My special thanks to Anish, Srivatsan, Sachin, Hemant, Joe, Xuerie, Zheng
and Krishnan. Although it is difficult to name everyone here, I thank each of my labmates

and fi'iends for their encouragement.

Last, but not the least, I thank the Department of Chemical Engineering and Materials
Science, the College of Engineering, the office of international student and scholars,
council of graduate studies (COGS) and the graduate school for the financial support in
the form of Food, Nutrition and Chronic Disease Fellowship and travel fellowships. Their

contributions helped me present my research at the national and international levels.

vii

 

TABLE OF CONTENTS

 

 

 

LIST OF TABLES IX
LIST OF FIGURES X
LIST OF ACRONYMS XIII
CHAPTER 1. INTRODUCTION 1

 

CHAPTER 2. IDENTIFICATION OF THE REACTIVE OXYGEN SPECIES CAUSING THE
CYTOTOXICITY OF PALMITATE TO HEPATOMA CELLS AND THE RELATIONSHIP OF THE

 

 

 

 

 

 

 

 

 

 

CYTOTOXICITY TO MITOCHONDRIAL PERMEABILITY TRANSITION 10
2.1 INTRODUCTION 10
2.2 MATERIALS AND METHODS ............................................................................................................... 12
2.3 RESULTS ....................................... 18
2.5 DISCUSSION ............................... 33

CHAPTER 3. APPLICATION OF METABOLIC FLUX ANALYSIS TO IDENTIFY ALTERED

FLUXES ASSOCIATED WITH THE TOXICITY OF PALMITATE 41
3.1 INTRODUCTION ............................................. 41
3.2 MATERIALS AND METHODS ............................................................................................................... 43
3.3 RESULTS - 55
3.4 DISCUSSION 80

 

CHAPTER 4. IDENTIFICATION OF GENES THAT REGULATE MULTIPLE CELLULAR
PROCESSES] RESPONSES IN THE CONTEXT OF LIPOTOXICITY TO HEPATOMA CELLS .86

 

 

 

 

 

 

 

 

 

 

 

4.1 INTRODUCTION ............................................. ..86
4.2 MATERIALS AND METHODS .................................................................... . ..... 88
4.3 RESULTS - ..... .93
4.4 DISCUSSION ............................................ 109
CHAPTER 5. INTEGRATING THE ONTOLOGY INFORMATION AND EXPRESSION DATA TO
IDENTIFY GENES THAT REGULATE METABOLIC AND PHENOTYPIC FUNCTIONS ........... 116
5.1 INTRODUCTION 116
5.2 MATERIALS AND METHODS ............................................................................................................. 121
5.3 RESULTS AND DISCUSSION ..................................................................... 129
5.4 DISCUSSION ...................................................................................................................................... 137
CHAPTER 6. CONCLUSIONS AND FUTURE DIRECTIONS 142
6.1 CONCLUSIONS .. ................................................ 142
6.2 FUTURE DIRECTIONS ....................................................................................................................... 143
APPENDIX 146
1. TABLE OF METABOLIC FLUXES FOR DAY 1 ........................................................................................ 146
2. VALUES OF THE METABOLIC FLUXES FOR DAY 2 .............................................................................. 149
3. VALUES OF THE METABOLIC FLUXEs FOR DAY 3 ................... . ....... 152
4. MATLAB PROGRAM USED FOR FLUX CALCULATIONS ..................................................................... 155
5. MATLAB PROGRAM USED FOR GROSS MEASUREMENT IDENTIFICATION ACCORDING To WANG
AND STEPHANOPOULOS ......................................................................................................................... 157
6. MATLAD PROGRAM USED FOR THE PERTURBATION ANALYSES ..................................................... 161
BIBLIOGRAPHY 163

 

viii

LIST OF TABLES

Table 3.1 List of the metabolic reactions included in the MFA model .................... 47
Table 3.2 List of intracellular metabolites ...................................................... 50
Table 3.3 List of measured extracellular fluxes ............................................... 52
Table 3.4 Ratio of change in the intracellular to extracellular lactate levels. .............. 59

Table 3.5 Ranked list of Pearson’s correlation coefficient of the metabolic fluxes with
cytotoxicity ........................................................................................... 61

Table 4.1. The cytotoxicity of various treatments ............................................... 92

Table 4.2. Top genes identified for TG accumulation. GA/PLS analysis was conducted to
identify genes related to intracellular TG accumulation. ...................................... 93

Table 4.3. Effect of various treatments on the expression of selected genes based on the
microarray data ..................................................................................... 111

Table 4.4. KEGG pathway represented by the genes identified for TG, LDH and (TG +
LDH) ................................................................................................. 112

Table 4.5. Enriched GO categories identified by the GOTM analyses .................... 113
Table 4.6. Top genes identified for cytotoxicity (LDH release) by GA/PLS analysis. . ...97
Table 4.7. Top genes selected for TG+LDH by dual response GA/PLS .................... 101

Table 4.8. Effect of inhibition of NADH dehydrogenase on the cytotoxicity of palmitate
and TG accumulation. ........................................................................... 115

Table 5 .1 The ranks of the three genes, adenylate cyclase 9 (AC9) and lysosomal
ATPases (ATP6VOAl and ATP6V1G1) as a fimction of Sc and Tau ..................... 131

ix

LIST OF FIGURES
Figure 2.1a Toxicity of different types of fatty acids .......................................... 18
Figure 2.1b. Concentration-dependence of saturated fatty acid toxicity .................... 19

Figure 2.2a Increased ceramide synthesis in response to palmitate, and inhibition by
Fumonisin B1 (FB1)21

Figure 2.2b. Effect of inhibition of ceramide synthesis on the toxicity of palmitate ........ 22

Figure 2.33 Levels of reactive oxygen species (ROS) in the cells following exposure to

palmitate .............................................................................................. 23
Figure 2.3b Release of hydrogen peroxide into the medium following palmitate exposure.
Figure 2.3c Effects of scavengers of H202 and hydroxyl radical on the oxidized status of
the cells ................................................................................................ 25
Figure 2. 3d Effects of scavengers of H202 and hydroxyl radical on palmitate-
toxicity... ... ... 26
Figure 2.4a Mitochondrial potential after exposure to fatty acids or hydrogen peroxide for
24 h .................................................................................................... 28
Figure 2.4b Effect of hydroxyl radical scavengers and the MPT inhibitor on the
mitochondrial potential ............................................................................. 29
Figure 2.4c Effect of the inhibition of MPT on the cytotoxicity .............................. 29

Figure 2.5a Effects of inhibiting mitochondrial complexes I and III on mitochondrial
potential loss ......................................................................................... 32

Figure 2.5b Effects of inhibiting mitochondrial complexes I and III on the cytotoxicity of

palmitate .............................................................................................. 33
Figure 3.1 a. Cytotoxicity of various treatments .................................................. 55
Figure 3.1b Activation of caspase-3 by palmitate ............................................. 56

Figure 3.2 Measured flux through the synthesis of triglycerides in response to FF As ..... 56
Figure 3.3 a. Profile of intracellular lactate levels in response to various treatments... . .58

Figure 3.3 b. Profile of extracellular lactate levels in response to various treatments ...... 58

Figure 3.4 Results of MFA analysis, depicting major intracellular fluxes on day 1 ....... 81
Figure 3.5 Results of MFA analysis, depicting major intracellular fluxes on day 2 ......... 82
Figure 3.6 Results of MFA analysis, depicting major intracellular fluxes on day 3........83

Figure 3.7a Measured flux of release of total ketone bodies (acetoacetate + beta-

hydroxybutyrate) by the cells ...................................................................... 63
Figure 3.7b Effect of inhibition of fatty acid beta-oxidation on the toxicity of saturated

F F A .................................................................................................... 64
Figure 3.8a Calculated flux through ceramide synthesis ....................................... 65
Figure 3.8b Intracellular C-16 ceramide levels in response to palmitate. . . . . . . . . . . . ...66

Figure 3.8c Results of in silico perturbation studies for the effect of serine or palmitate on
the flux of de novo ceramide synthesis ............................................................ 66

Figure 3.8d Effect of inhibition of ceramide synthesis on the levels of C-16 ceramide in
the cells ............................................................................................... 67

Figure 3.8c Effect of inhibition of ceramide synthesis on the cytotoxicity of palmitate...67
Figure 3.9a Calculated relative fluxes of GSH synthesis, as predicted by MFA ........... 70

Figure 3.9b Measured relative levels of total intracellular GSH. Cells were treated with
palmitate or oleate for 3 days. .................................................................... 70

Figure 3.90 Results of in silico perturbation studies on the effect of increasing uptake of
amino acids on GSH synthesis ................................................................... 71

Figure 3.9d Effect of the treatments on the expression of gamma-glutamyl cysteine

synthetase (GCS) heavy subunit. ............................................................... 72
Figure 396 Effect of the treatments on the expression of xCT (a subunit of cystine
transporter) .......................................................................................... 73
Figure 3.9f Effect of NAC supplementation on the cytotoxicity of palmitate . . . . .....74

Figure 3.10a Effect of NAC supplementation on GCL levels in palmitate-treated
cells ................................................................................................... 76

Figure 3.10b Effect of NAC supplementation on GCL levels in control medium. . . .77

xi

Figure 4.1 Intracellular accumulation of TG in response to various FF As and TNF-or

......................................................................................................... 91
Figure 4.2a. Effect of palmitate on the transcript levels of SCD, measured by microarray
and RT-PCR ......................................................................................... 94
Figure 4.2b. Effect of palmitate and oleate on the protein levels of stearoyl-CoA
desaturase (SCD) .................................................................................... 95
Figure 4.3a. Effect of various treatments on the mitochondrial potential ................... 99
Figure 4.3b. Activation of caspase-3 by palmitate treatment .................................. 99

Figure 4.4a. Effect of inhibition of various MAPKS on the cytotoxicity of palmitate. . . 105

Figure 4.4b. Effect of inhibition of various MAPKS on the TG accumulation in response

to palmitate .......................................................................................... 105
Figure 5.1a Cytotoxicity of the treatments ..................................................... 126
Figure 5.1b Profile of acetoacetate release by various treatments ........................... 127

Figure 5.2 Variation of the discriminating ability of the mixture model with the

smoothing constant (t) and the weight to GO (A) .............................................. 128
Figure 5.3a Discrimination of the cytotoxic conditions from the non-toxic conditions for

it = 0.5, t = 3 ........................................................................................ 129
Figure 5.3b Discrimination of the ketogenesis for A = 0.5, t = 3 ............................ 129

Figure 5 .4 Effect of inhibitor of lysosomal ATPase, bafilomycin (Bfl) on the toxicity of
palmitate ............................................................................................. 132

Figure 5.5 Effect of increasing cAMP levels on the caspase-3 activation by
palmitate ............................................................................................. 133

xii

8.

9.

10.

11.

12.

l3.

14.

15.

16.

17.

18.

19.

20.

21.

LIST OF ACRONYMS

1400W - N-(3- (Aminomethyl)benzyl)acetamidine dihydrochloride
2-MA — 2 Mercaptoacetate

AC —- Adenylate Cylcase

AG — Aminoguanidine hydrochloride

ALT — Alanine Aminotransferase

AM-HZ-DCF DA — 6-carboxy-2',7'-dichlorodihydrofluorescein diacetate,
di(acetoxymethyl ester

ANOVA — Analysis Of Variance

ANT — Adenine Nucleotide Translocator
AST — Aspartate Aminotransferase

BCA — Bicinchoninic Acid

Bfl — Bafilomycin

BSA — Bovine Serum Albumin

C/EBP — CCAAT Enhancer Binding Protein
cAMP — Cyclic Adenosine Monophosphate
CASP6 — Caspase 6

CAT - Catalase

CIC4 — Chloride Channel 4

c-PTIO — 2-(4-carboxyphenyl)—4,5-dihydro-4,4,5 ,5-tetramethyl- 1 H-imidazolyl- 1 -
oxy-3-oxide

CSA — Cyclosporine A
Cu-DIPS — Copper(II)2(3,5-diisopropylsalicylate)4

CYP1A1 — Cytochrome P450 1A1

xiii

22. CYP2E1 — Cytochrome P450 2E1

23. CYP46 — Cytochrome P450, subfamily 46 (cholesterol 24-hydroxylase)
24. Cpr - Cyclophilin D

25. DAG — Diacylglycerol

26. DHCR24 — 24-Dehydrocholesterol Reductase

27. DMEM — Dulbecco's Modified Eagle's Medium

28. DMSO — Dimethyl sulfoxide

29. DMU —— Dimethyl Urea

30. ERK — Extracellular signal Regulated Kinase

31. EST —- Expressed Sequence Tag

32. ETFl — Eukaryotic Termination Factor 1

33. Etrnx — Etomoxir

34. FBl - Fumonisin Bl

35. FBS —- Fetal Bovine Serum

36. FCCP — p-trifluoromethoxy Carbonyl Cyanide Phenylhydrazone
37. FDA — F isher’s Discriminant Analysis

38. FFA — Free Fatty Acid

39. GA/PLS — Genetic Algorithm coupled Partial Least Squares
40. GAPDH — Glyceraldehyde phosphate dehydrogenase

41. GCS -— Glutamyl Cysteine Synthase

42. GO — Gene Ontology

43. GOTM - Gene Ontology Tree Machine

44. GSH — Glutathione

xiv

45. HPLC — High Performance Liquid Chromatography
46. IBMX — 3-Isobutyl-l-Methylxanthine

47. IDF — Independent Document Frequency

48. IMGO — Integrative Mixture Model Gene Ontology
49. INPPSA — Inositol Polyphosphate Phosphatase 5 A
50. INPPLl — Inositol Polyphosphate Phosphatase-like l
51. 1P3 —- Inositol Triphosphate

52. INK — c-Jun N-terminal Kinase

53. KEGG —- Kyoto Encyclopedia of Genes and Genome
54. LCAT — Lecithin Cholesterol Acyl Transferase

55. LDH — Lactate Dehydrogenase

56. MAPK — Mitogen Activated Protein Kinases

57. MCD — Methionine Choline Deficient

58. MFA - Metabolic Flux Analysis

59. MnTBAP —- Manganese(III) tetrakis(4-benzoic acid)porphyrin chloride
60. MPP -— Mitochondrial Permeability Pore

61. MPT -— Mitochondrial Permeability Transition

62. NAC - N-Acetyl L-cysteine

63. NaOH — Sodium Hydroxide

64. NASH — Non-Alcoholic Steatohepatitis

65. NDUFABI — NADH dehydrogenase (ubiquinone) l, alpha/beta subcomplex, l
66. NFIX - Nuclear Factor IX

67. PBS — Phosphate Buffered Saline

XV

68. PCA — Principal Component Analysis

69. PDTC — Pyrrolidone dithiocarbamate

70. PKC — Protein Kinase C

71. PLS — Partial Least Squares

72. POBN — alpha-(4-Pyridyl- 1 -Oxide)-N-tert-Butyl Nitrone
73. PP2A — Protein Phosphatase 2A

74. PPZB — Protein Phosphatase 2B

75. PPAR-a — Peroxisome Proliferator Activated Receptor alpha
76. PTPNl -— Protein Tyrosine Phosphatase N1

77. ROS -— Reactive Oxygen Species

78. RT-PCR — Reverse Transcriptase Polymerase Chain Reaction
79. SFA — Saturated Fatty Acid

80. SHIP2 — SH2-containing Inositol 5'-Phosphatase 2

81. SOD — Superoxide Dismutase

82. SREBP — Sterol Receptor Element Binding Protein

83. TBS-T — Tris buffered saline containing 0.1% Tween
84. TG —— Triglyceride

85. tGSH — Total Glutathione

86. TNF-or — Tumor Necrosis Factor Alpha

87. UFA — Unsaturated Fatty Acid

88. VLDL — Very Low Density Lipoprotein

89. XIAP — X-linked Inhibitor of Apoptosis Protein

xvi

CHAPTER 1. INTRODUCTION

1.1 Obesity and the effects of free fatty acids

Obesity is on the rise and with it the incidences of obesity-associated disorders such as
insulin resistance, type 2 diabetes, cardiovascular disorders and lipotoxicity. Not only
adults, but even children are becoming obese at an increasing rate, and many obesity-
related disorders also are developing in young patients (Chia and Boston, 2006). While it
may be most prudent to target obesity itself which is the underlying cause of many
metabolic disorders, this would mean significant changes in our lifestyles, something that
will take many years, if not decades, to come to fruition. For now, treatment strategies
have to be devised to effectively combat the scourge of disorders associated with obesity.
This means understanding how obesity affects different organs of the body, and also how

the organs interact to produce the disease phenotype.

Elevated circulating free fatty acid levels have emerged as a prime factor which mediates
development of various disorders related to obesity (Felber and Golay, 2002, Scheen,
2000 and Boden 1997). The two sources of fatty acids in the body are diet (eating fat-rich
food) (Astrup, 2005), and lipolysis (DeFronzo, 2004). Previous studies have indicated

that both these sources contribute to the increased FFA-levels in obese people.

Prolonged exposure to elevated F FA levels causes many changes in the body. Numerous

studies have been conducted to identify the effects of FFAs on different type of cells and

organelles. These studies have provided essential information on how FFAs regulate
individual genes, enzymes or pathways, and have formed the basis to test various
hypotheses. Most of these changes are due to the interactive effects of the various organs
of the body. For example, increased exposure to fatty acids leads to accumulation of fat
in the adipose tissue, leading to increase in the size of the tissue. This increase in the size
of the tissue is not only due to an increase in the number of cells, but due to increase in
the size of individual cells. The enlarged adipocytes release many factors (adipokines)
such as leptin, interleukins and TNF-alpha, which affect the function of hepatocytes and
muscle cells (Skurk et al., 2006). Similarly fatty acids can compete with glucose for
oxidation and cause insulin resistance in muscle cells (Randle, 1964). This leads to a
smaller rate of reduction of the levels of glucose and thus longer and greater release of
insulin, as seen in the early stages of type 2 diabetes. This causes increased lipid
synthesis and storage, which would increase adipose tissue size and increase adipokine
secretion. Thus, there may be a vicious cycle of events that lead to the development of
disorders at the whole body level. Exposure to elevated FFAs can also lead to multiple
alterations in the function of hepatocytes, including cell death as discussed in section 1.2

below.

Exposure to fatty acids also alters cellular physiology by multiple means. Many fatty acid
metabolites also act as signaling molecules, which not only control cellular physiology
directly by altering the activity of many enzymes, but can also cause changes in the
expression of many enzymes by activating transcription factors. Examples of such

molecules include diacylglycerol (DAG) and ceramide. Elevated levels of DAG are

known to activate protein kinase C (PKC). PKCs are signaling kinases which alter many
downstream processes. There are three types of PKCs: classical, novel and atypical. The
classical and novel forms of PKCs are activated in response to elevated DAG levels,
while the atypical forms respond to other lipid metabolites, such as ceramide. Activation
of PKCs alters multiple processes such as growth, differentiation and even cell death
(Musashi et al, 2000). Ceramide is another lipid which may affect signaling pathways and
alter cellular physiology (Ruvolo, 2003). Among the signaling enzymes activated by
ceramide is protein phosphatase 2A (PP2A). PP2A affects multiples cellular processes,
including dephosphorylation of Bcl-2, an important protein involved in maintaining
mitochondrial integrity and preventing cell death (apoptosis) (Chan and Yu, 2004). Fatty
acids can also alter the functioning of cells by altering the enzyme make-up of the cells
by regulating transcription of genes. Fatty acids, especially the unsaturated fatty acids,
have been shown to be ligands for transcription factors such as peroxisome proliferator
activated receptor alpha (PPAR-a) and sterol receptor element binding protein (SREBP-
lc), which have been shown to regulate multiple genes involved in lipid oxidation and

synthesis (Jump et a1, 2005).

1.2 Role of liver and the effects of FFAs on hepatic metabolism

Liver is the central organ in the metabolism of fatty acids by the body. It is the organ that
takes up the fat coming from the intestines as chylomicrons, and packages it into very
low density lipoprotein (VLDL) which then goes into the bloodstream. Liver is among
the most metabolically active organs, and produces other useful metabolites from fats,

such as ketone bodies, cholesterols and hormones. Liver is also the organ that degrades

insulin, thereby controlling the level of this hormone. Given the central and important
role played by this organ in fatty acid metabolism, it is a necessary first-step to identify
the changes by the fatty acids on the liver itself. Hepatocytes are the primary cell type
found in livers and are the primary contributors to F FA metabolism in the liver.
Therefore, identifying the changes in the metabolism of the hepatocytes on exposure to
elevated FF A levels would provide important information on the development of hepatic
disorders observed in chronic obesity, and perhaps provide insight into the changes that

occur in other organs as well.

Elevated FFAs significantly alter the fimction of hepatocytes. Fatty acids cause greater
synthesis and secretion of cytokines by hepatocytes and other cells of liver (F eldstein et
al., 2004). They inhibit insulin suppression of gluconeogenesis (formation of glucose by
the breakdown of amino acids) and glycogenolysis (breakdown of glycogen to form
glucose) in hepatocytes (Boden, 2004). They cause accumulation of lipids in the liver
(steatosis). While steatosis is benign by itself, it sensitizes hepatocytes to oxidative injury
(Farrell and Latter, 2006). In many individuals, chronic exposure to elevated FFAs leads
to development of a hepatic disease called non-alcoholic steatohepatitis (NASH), which
is characterized by extensive hepatocyte cell death. One study estimated that as much as
1-2% of the US population suffers from NASH. Even though the specific events leading
to the development of NASH from steatosis are unclear, studies suggest thatNASH
develops by ‘two hits’ (Day and James, 1998). In the first hit, there is increased
accumulation of lipids inside the cells which sensitizes the cells to lipid peroxidation and

oxidative injury. In the second hit, an oxidative insult causes hepatocyte cell death,

causing alterations in hepatic function and increased levels of aspartate transaminase
(AST) and alanine transaminase (ALT) in the serum. AST and ALT are (relatively)
hepato-specific enzymes, whose levels increase on hepatic injury. While oxidative stress
is lmown to play an important role in the development of NASH, the causes of increased
oxidative stress are not clear. It is likely that FFAs themselves cause oxidative stress in
hepatocytes as they do in other cell types, though it had not been shown previously.
Similarly, the type of reactive oxygen species (ROS) produced on exposure to FFAs, as
well as the differences in the types of FFAs in causing hepatotoxicity are not clear. It is
now known that it is not just the amount of fat intake, but also its type, which determines
the development of metabolic disorders. The differences in the types of fat arise due to
their fatty acid composition. There are three major types of free fatty acids in the diet and
plasma: saturated fatty acids (e.g. palmitate), monounsaturated fatty acid (oleate) and
polyunsaturated fatty acid (e.g. linoleate). Many previous studies have indicated that the
different types of fatty acids differ not only in their molecular structures, but also in their
metabolic, genetic and physiological effects on the cells. However, the information on the
systems level changes effected by them is lacking. Therefore, it is important to identify
how different fatty acids affect cells at the level of physiological function, metabolism
and gene expression, and how these effects could be related to the development of

metabolic disorders.

While it may be best to study the responses of FFA exposure in primary human liver
cells, their accessibility is relatively poor and the cost of experimentation very high. The

other option is to use primary hepatocytes from animals (usually rodents) or human cell

lines. While primary rodent hepatocytes are a close mimic to in vivo hepatocytes in
function, the experiments with these cells are limited to shorter durations because of their
limited lifespan upon culturing. Furthermore, it is much easier to prepare stable
transformed cells lines for further analyses than it is to manipulate primary cells. Human
cell lines offer the advantages of retaining most of the major biological functions of
hepatocytes, ease of long-term culture and possibility of creating stable transformed cells
to study the long term effects of a target gene manipulation or the interactive effects of
the target protein with other factors, for example, by creating HepGZ cell line
overexpressing cytochrome P450 2E1 (CYP2E1), the role of this protein in ethanol
cytotoxicity as well as arachidonic toxicity has been studied (Caro and Cederbaum, 2006
and Cederbaum, 2006). Considering these advantages, human hepatoma cell line
HepG2/C3A was employed in the present study. HepG2 cells have been employed in
many previous studies on fatty acid, lipid and lipoprotein metabolism. Studies have
shown that the plasma concentrations of FFAS in obese and diabetic patients are about
0.7 mM. Therefore, this concentration of FFAS was employed in the study. Though most
experiments were conducted at this concentration, some experiments on the physiological
effects of saturated FFAS were also conducted at a lower concentration of 0.4 mM. No

qualitative difference was observed upon reduction of the fatty acid concentration.

1.3 Systems biology approach to identify global changes in cellular

make-up and function

Cells can be considered as a network of three large interacting networks: metabolic,

signaling and genetic, with significant cross-talk between the nodes (members) of these

 

II—_

networks. It is this highly interconnected nature of cellular networks that makes cells
extremely resilient to various insults, and maintains cellular homeostasis. A change in the
activity of a particular node usually causes a feedback response from the cells to maintain
cellular homeostasis. These responses can be in the form of altered activity of the
enzymes of the pathway, regulated through a change in the expression or allosteric

control of the enzymes.

For comprehensive evaluation of cellular responses, global metabolic, genomic and
proteomic responses should be measured. The development of high throughput
techniques such as microarray analysis, as well as novel metabolite-detection techniques
such as mass spectrometry etc. have made it possible to identify hundreds of fluxes or
thousands of genes in relatively small times. The challenge is to (i) efficiently analyze the
high-throughput information to identify altered cellular processes and (ii) integrate the
multi-source information to generate comprehensive picture of cellular responses and
identify the underlying genetic and metabolic changes which regulate desired processes.
A lot of effort is being invested to meet these challenges, because the framework has
potential to revolutionize the way medical investigations are done and to considerably

reduce the time it takes to develop treatments against diseases.

1.4 Aims of the current study

The present study is an effort to understand the systems-level effects of different types of
free fatty acids on the physiology, metabolism and gene-transcription of human hepatoma

cells. The overall approach taken was to study the metabolic and genetic effects, using

 

high throughput techniques that provide system-level information on the effects of
perturbations. Novel statistical models were employed to integrate the information
generated and relate to the physiological function of interest. Not only was systems-level
information on the metabolic and genetic effects of the different fatty acids generated, but
also the interactions between the multiple levels were incorporated, thus providing a

comprehensive snapshot of the perturbations caused by these treatments.

1.5 Organization of the thesis

Because of the multi-level (physiological, metabolic, genetic and integrative) studies
conducted in the current study, each chapter in this thesis is based on one of the aspects

studied.

In chapter 2, the results and observations of the physiological responses of the cells to
different FFAS are presented. It was observed that saturated FFA (palmitate) was the only
toxic fatty acid among the different FFAS. Exposure to saturated FFA caused a significant
increase in intracellular ROS levels, which played an important role in mediating the
toxicity of this fatty acid and led to significant reduction in the mitochondrial potential of
the cells, due to mitochondrial permeability transition (MPT). It was also identified that

the MPT does not play a role in the toxicity of this FF A.

In chapter 3, the global metabolic changes caused by exposure to FFAS were studied
using metabolic flux analysis (MFA). Exposure to FF As led to accumulation of TG inside

the cells. Exposure to palmitate caused lower TG accumulation, but greater ketone body

release and de novo ceramide synthesis by the cells as compared to the unsaturated F FAs.
However, none of these alterations were associated with toxicity. On the other hand,
increasing the TG accumulation was found to significantly reduce palmitate-toxicity.
Exposure to palmitate also reduced the synthesis of glutathione by reducing the levels of
cystine transporter and this alteration was found to play an important role in the toxicity

of this fatty acid.

In chapter 4, the genomic changes caused by exposure to FFAs and TNF-a were studied.
In addition, a framework is presented which can help identify genes that regulate multiple
cellular processes. Such genes may serve as better targets to regulate the development of
complex diseases. Application of the framework to current data identified that NADH
dehydrogenase as well as MAPKs regulate both TG accumulation as well as cytotoxicity.

These predictions were experimentally verified.

In chapter 5, a novel framework is presented to incorporate gene ontology information in
the process to select genes pertinent to function(s) of interest, using mixture models to
differentiate the sub-populations. Genes identified by this framework, which separated
the toxic fi'om non-toxic conditions, are presented and literature as well as experimental

evidence is presented to discuss the role of the identified genes in lipotoxicity.

Finally, in chapter 6, the conclusions that can be made based on the current study, as well

as certain predictions and the future directions are presented.

CHAPTER 2. IDENTIFICATION OF THE REACTIVE
OXYGEN SPECIES CAUSING THE CYTOTOXICITY OF
PALMITATE TO HEPATOMA CELLS AND THE
RELATIONSHIP OF THE CYTOTOXICITY TO
MITOCHONDRIAL PERMEABILITY TRANSITION

2.1 Introduction

Elevated plasma free fatty acid (F FA) concentrations are associated with the development

of many hepatic disorders. FFAS increase the production of glucose (Lam Tony,
Carpentier et a1. 2003), the accumulation of triglycerides (Tanaka and Aoyama 2001),
and inhibit insulin clearance by hepatocytes (Stroemblad and Bjoemtorp 1986).
Chronically elevated fatty acid levels are associated with the development of non-
alcoholic steatohepatitis (NASH), characterized by extensive hepatocyte cell death.
Saturated F FAs cause cell death in a variety of cells including pancreatic beta-cells (El-
Assaad, Buteau et al. 2003), cardiomyocytes (Kong and Rabkin 2000), and hepatocytes
(Dashti, Feng et a1. 2000; Ji, Zhang et al. 2005). According to the studies conducted in
non-hepatic cells, such as beta-cells and myocytes, the potential mechanisms of cell death
by these fatty acids involve increased ceramide production (Lu, Mu et a1. 2003), reduced
mitochondrial potential (Kong and Rabkin 2000) and increased production of reactive
OXYgen species (ROS) (Listenberger, Ory et al. 2001; Yamagishi, Okamoto et al. 2002).
Studies in hepatocytes or hepatoma cells have suggested that lysosomal permeabilization
(Wei, Wang et al. 2006), release of cathepsin B (Feldstein, Wemeburg et al. 2004), and
reduction of mitochondrial Bcl-2/Bax ratio (J i, Zhang et al. 2005) play important roles in

cell death caused by palmitate.

10

Different molecular species of ROS have been shown to mediate fatty acid toxicity in
different types of cells, e.g. superoxide anion is known to mediate fatty acid toxicity in
CHO cells (Listenberger, Ory et al. 2001), while nitric oxide mediates the cell death in
pancreatic beta cell line INS-1 (Maestre, Jordan et al. 2003; Okuyama, Fujiwara et a1.
2003). However, the type of ROS which mediate the palmitate-toxicity to hepatoma cells
has not yet been elucidated. Herein, the results of our investigation towards identifying
the ROS molecular species involved in the fatty acid-induced cytotoxicity of hepatoma

cells are reported.

Membrane permeability transition (MPT) is considered an important step in cell death
(Armstrong 2006). MPT leads to release of pro-apoptotic molecules such as cytochrome
c, apoptosis inducing factor (AIF) and endonuclease G from the mitochondria (Bemardi,
Broekemeier et a1. 1994). Cell death in response to many stimuli can be abrogated by
inhibiting MPT, e.g., inhibiting MPT has been shown to reduce ROS formation and cell
death of hepatocytes in response to microcystin treatment (Ding, Shen et a1. 2000) and
the apoptosis of HepG2 cells in response to hypochlorous acid (Whiteman, Rose et a1.
2005). Saturated fatty acids can cause MPT (Petronilli, Cola et al. 1993), which mediates
the cell death by palmitate in cardiomyocytes (Kong and Rabkin 2000). However, it is
not clear whether MPT plays a similar role in controlling fatty acid-induced toxicity of

hepatocytes.

ll

The objectives of this study were to investigate the mechanisms of fatty acid toxicity to
hepatoma cells, identify the molecular species of ROS which mediate(s) the toxicity and
study the inter-relationships between ROS production, MPT and cytotoxicity when these
cells were exposed to FFAS at an elevated physiological concentration of 0.7 mM
(Skowronski, Hollenbeck et al. 1991; Boden, Chen et al. 2001; Mensink, Blaak et al.
2001; Woerle, Popa et al. 2002). The results suggest that saturated fatty acids can cause
lipotoxicity through increased production of hydrogen peroxide and *OH radicals. Even
though MPT was observed in the cells treated with the saturated fatty acid, the toxicity is

not mediated through MPT.

2.2 Materials and Methods

2.2.1 Cell Culture. 1 million HepG2/C3A cells (American Type Culture Collection,
Manassas, VA) were seeded into each well of a 6-well plate. The cells were kept in 2 ml
of the medium containing Dulbecco’s modified Eagle’s medium (DMEM, Invitrogen,
Carlsbad, CA) supplemented with 10% fetal bovine serum (American Type Culture
Collection) and 2% Penicillin-streptomycin (Invitrogen). Cells were incubated at 37°C
and in 10% C02 atmosphere. After the cells reached confluency, the medium was
replaced with 2 ml of the desired medium (Control, BSA, 0.7 mM FFA medium).
Palmitate, oleate and linoleate were used as the representative fatty acids for the
saturated, monounsaturated and polyunsaturated type of fatty acids. These fatty acids are
the most abundant fatty acids of their class in the plasma, e.g. palmitate is the most
abundant saturated fatty acid in the plasma. Plasma FFA levels in the obese and type 2

diabetic patients have been reported to be about 0.7 mM in many studies (Skowronski,

12

Hollenbeck et al. 1991; Boden, Chen et al. 2001; Mensink, Blaak et al. 2001; Woerle,
Popa et al. 2002). Therefore, this concentration of FFAS was employed in the present

study. Media were changed after every 24 h.

2.2.2 Determination of cytotoxicity. Cytotoxicity was measured by measuring the
release of lactate dehydrogenase (LDH) into the medium using the cytotoxicity detection
kit (Roche Applied Science, Indianapolis, IN). LDH is a cytosolic enzyme which is
released when the membrane integrity of the cells has been compromised. In this assay,
the NADH generated during the oxidation of lactate by LDH is utilized as a co-factor by
the enzyme diaphorase to reduce tetrazolium salt (yellow) to formazan dye (red). To
conduct the assay, the supernatant media was collected after desired exposure times (e. g.
24 and 48 h) and the LDH activity in the medium was determined (called LDH-medium).
The cells were washed with phosphate buffered saline (PBS), and lyzed with 1% triton-
X-100 in PBS for 24 h at 37 °C. The lysate was collected, vortexed for 15 seconds and
centrifuged at 5000 g for 3 minutes. LDH activity in the lysate was determined (called
LDH-Triton). The LDH released into the medium was normalized to total LDH, as
shown in the equation below

LDH (medium) x
LDH (medium) + LDH(triton)

 

%LDHrelease =

2.2.3 Inhibitor experiments. Fumonisin Bl (FBI) (Sigma-Aldrich, St. Louis, MO) and
cyclosporine A (CSA) (Calbiochem, San Diego, CA) were employed for inhibiting

ceramide synthase and MPT, respectively. 2 mM stock solution of FBl was prepared in

13

PBS, and a 20mM stock solution of cyclosporine A was prepared in DMSO. The cells

were treated with palmitate in the presence of 2—100 uM FBl , and 0.5-5 uM CSA.

2.2.4. Measurement of ceramide. The cells were treated with 0.7 mM palmitate in the
presence or absence of 20 uM FBI (inhibitor of de novo ceramide synthesis). After the
treatments, the cells were washed with PBS, trypsinized and pelleted down. Lipids were
extracted by the method of Bligh and Dyer (Bligh and Dyer 1959). Briefly, the pellets
were washed once with PBS, resuspended in 3 ml chloroformzmethanol solution (1:2
v/v), and lyzed by sonication in this solution for 30 seconds. 0.8 ml water and 0.5 ml
methanol were then added, and the solution was kept at room temperature for 15 minutes.
Cell debris was pelleted by centrifugation at 5000 g for 5 min. The supernatant was
decanted into a new tube and 1 ml each of chloroform and water was added. The solution
was briefly vortexed and the phases were allowed to separate for 15 minutes. The tubes
were then centrifuged at 3000 g for 5 minutes. The aqueous layer was removed, and 1 m1
of the chloroform solution was transferred to a 1.5 ml microcentrifuge tube, and dried in
a speedvac. The lipids were resuspended in 80 ul chloroform. The levels of ceramide
were then measured using liquid chromatography-tandem mass spectroscopy (LC-MS),
according to a previously published method (Yamaguchi, Miyashita et al. 2004). Protein
levels in the aqueous phase were measured using BCA assay. Ceramide results from LC-

MS measurements were normalized to the total protein.

2.2.5. Measurement of ROS. Two different methods were employed to estimate relative

ROS levels, as discussed below. In the first method, 6-carboxy-2',7'-

14

dichlorodihydrofluorescein diacetate, di(acetoxymethyl ester) (C-2938, Molecular
Probes, Eugene, OR) dye was used to measure the increased oxidized status of the cells
in response to palmitate treatment. Stock solution of the dye was prepared in DMSO, and
diluted to 5 uM in DMEM. Confluent cells were treated with palmitate for different time
periods. Measurement of fluorescence was performed by flow cytometry according to a

previously published protocol (Listenberger et al. 2001).

As a separate measurement, the hydrogen peroxide released into the medium was
measured using a colorimetric hydrogen peroxide assay (Assay Designs, Ann Arbor, MI).
This assay is based on the principle of reaction of coordinated iron with peroxide and
xylenol orange in acidic solution to produce a purple-colored complex, whose color is
proportional to the concentration of hydrogen peroxide in the sample. Confluent cells
were exposed to either HepG2 medium or to palmitate for desired periods. Media were
replaced every 24h. After the treatment periods, media were collected and the hydrogen
peroxide levels in the media were analyzed immediately, according to the manufacturer’s
instructions. Blanks were generated by incubating the media without cells and subtracting
the absorbance of the blanks from that of the samples. The estimated concentrations were
multiplied by the volume of the culture to obtain the total amount, which were then
normalized to the total protein levels. For estimation of the protein, the cells were washed
with PBS and lyzed with 1% triton-X-IOO to estimate protein levels by the bicinchoninic
acid (BCA) assay (Pierce Chemical, Rockford, IL). Hydrogen peroxide levels were then

normalized to cellular protein levels.

15

2.2.6 Radical Scavengers and spin trap. 1, 3 Dimethyl urea (DMU) (MP Biomedical)
and dimethyl sulfoxide (DMSO) (Sigma Aldrich) were employed to scavenge hydroxyl
radicals. Solutions of these scavengers were prepared in the palmitate medium at 5-50
mM for DMU and 0.2-3% (v/v) for DMSO. The spin trap alpha-(4—pyridyl-l-oxide)-N-
tert—butyl nitrone (POBN, 20-100 mM) from Sigma Aldrich was dissolved in palmitate
medium. Catalase (CAT, 3000 U/ml) was used to scavenge hydrogen peroxide. It was
dissolved in the palmitate medium. Cu(II)2(3,5-diisopropylsalicylate)4 (Cu-DIPS) and
Mn(III) tetrakis(4-benzoic acid)porphyrin chloride (MnTBAP) are the cell permeable
mimics of cytosolic (Cu-Zn) and mitochondrial (Mn) superoxide dismutase (SOD),
respectively (Laurent, Nicco et al. 2004). These SOD-mimics were obtained from Sigma
Aldrich, and utilized as scavengers for the superoxide anion. Stock solution of Cu-DIPS
(5 mM) was prepared in DMSO and used in the concentration range 1-10 uM. Stock
solution was prepared by dissolving 50 mM MnTBAP in 0.1 M NaOH, followed by
addition of equal volume of 0.1 M HCl. MnTBAP was used in the concentration range
10-100 uM. Pyrrolidone dithiocarbamate (PDTC) and 4,5-dihydroxy-l,3-benzene-
disulfonic acid (Tiron) were obtained from TCI chemicals, and employed as scavengers
of superoxide anion (Listenberger, Ory et al. 2001). These scavengers were dissolved in
the medium at concentrations of 025-10 mM and 2-40 mM, respectively. 2-(4-
carboxyphenyl)-4,5-dihydro-4,4,5,5-tetramethyl- 1 H-imidazolyl-l -oxy-3-oxide (c-PTIO),
procured from Axxora Chemicals, was used to scavenge nitric oxide (NO) (Dave,
Farrance et al. 1997). 200 mM stock solution of carboxy-PTIO (cPTIO) was prepared in
water and used at final concentration of 20-200 uM. Aminoguanidine hydrochloride (AG)

and N-(3- (Aminomethyl)benzyl)acetamidine dihydrochloride (1400W), both from

16

Axxora Chemicals, were employed to inhibit nitric oxide synthesis (Dave, F arrance et al.
1997). AG was dissolved in PBS at a concentration of l M and used at 0.1-1 mM final
concentration. A 100 mM stock solution of 1400 W was prepared in water, and used in

the concentration range 50-500 uM.

2.2.7 Dose-dependence studies. For all the inhibitors and scavengers, dose-dependence
studies were performed. That is, the cells were exposed to palmitate in the presence of
various concentrations of these scavengers/ inhibitors and the LDH release was
measured. The following treatment ranges were studied- FBl (2-50 uM), catalase (500-
3000 U/ml), DMU (2-50 mM), DMSO (0.5-2% v/v), CuDIPS (0.5-10 uM), MnTBAP
(25-200 uM), Tiron (2-20 uM), PDTC (0.5-10 uM), cPTIO (25-200 uM), AG (0.1-1
mM), 1400W (SO-500 uM), CsA (0.5-5 uM), POBN (20-100 mM), Rotenone (100-2000
nM), antimycin A (SO-1000 nM). Results are presented for concentrations which were

most effective in reducing LDH.

2.2.8 Measurement of the mitochondrial potential. Mitochondrial potential was
measured using JC-l dye (Calbiochem). Potential-sensitive uptake of this dye leads to its
accumulation and aggregate formation in normal, active mitochondria. Aggregates of the
dye (mitochondria) fluoresce red, while the dilute solution (cytoplasmic) fluoresces
green. The ratio of red to green fluorescence was used to estimate the mitochondrial
potential (Cossarizza, Ceccarelli et al. 1996). A 5 mM stock solution of the dye was
prepared in DMSO. After treating the cells for desired times, the media were aspirated,

and the cells were exposed to 5 uM JC—l dye in DMEM. Cells were incubated in the dye-

l7

containing medium for 30 min. Dye was aspirated and the cells washed twice with 1 ml
PBS. 1 ml PBS was added to each well and the plates were bottom-read in a fluorescence
reader (Molecular Probes, Eugene, OR) at excitation of 488 nm and emissions of 527 nm
(green) and 590nm (red). p-trifluoromethoxy carbonyl cyanide phenyl hydrazone (F CCP)
is an established mitochondrial uncoupler and abolishes the mitochondrial potential. The
mitochondrial uncoupler FCCP (40 uM) was used as a control for totally lost membrane

potential.

2.2.9 Statistical Analysis. Treatments were analyzed with analysis of variance
(ANOVA) followed by Bonferroni’s post-hoc comparison to locate specific differences

(Turner and Thayer 2001 ). Analyses were performed using S-plus software.

2.3 Results

2.3.1 Cytotoxicity of different types of fatty acids. To evaluate the toxicity of various
fatty acids, the HepG2 cells were exposed to 0.7 mM palmitate, oleate or linoleate- the
three primary species of saturated, unsaturated and polyunsaturated fatty acids in the
plasma, for two days. Media were changed after 24h of exposure, and the LDH release
into the medium was measured. Among the fatty acids studied, only palmitate was found
to be toxic to the cells and resulted in a significantly higher LDH release on both days
(Figure 2.1a). The toxicity of palmitate was concentration dependent above 0.2 mM
(Figure 2.1b). When exposed to free fatty acids, the cells metabolized about 0.2 mM of
the different free fatty acids (not shown). This suggests that the amount of fatty acid

present in excess to the metabolic capacity causes the toxicity, but this toxicity is specific

18

to saturated fatty acid, because the unsaturated fatty acids that were not fully metabolized

were non-toxic.

 

45
40-
35+

 

 

 

 

LDH released (%)

 

 

 

Control Palm Ole Lin
Fatty acid

Figure 2.1a Toxicity of different types of fatty acids. Confluent HepGZ cells were exposed to different
types of fatty acids (0.7mM, complexed to 4% w/v BSA) for 2 days, with medium-change everyday. LDH
released on each day was measured and normalized to total LDH. Total LDH is defined as the sum of the
LDH released and the intracellular LDH. The intracellular LDH values were obtained by lyzing the cells
with 1% triton (see methods section). Data expressed as averages of nine samples +/- s.d. from three

independent experiments. *, significantly higher than control, p< 0.01.

19

.5
0|

 

[£639 23

8 8
l I

 
  
 

N“
GO

LDH Released (°/o)
N
O

 

 

 

o I I I I
0 0.2 0.4 0.6 0.8

Palmitate Concentration (mM)

Figure 2.1b Concentration-dependence of saturated fatty acid toxicity. The LDH released after
exposure to different concentrations of palmitate for two days was measured. Data expressed as averages of

nine samples +/- s.d. from three independent experiments. *, significantly higher than control, p< 0.01 .

2.3.2 Ceramide metabolism. De nova ceramide synthesis has been shown to be a major
factor in palmitate-induced toxicity in a variety of cell types (Lu, Mu et al. 2003). Among
the primary mechanisms of ceramide-induced toxicity is increased ROS production
(Gomez, Mendoza-Milla et al. 1996; France-Lanard, Brugg et al. 1997; Andrieu-Abadie,
Gouaze et al. 2001). To investigate the role of de nova ceramide synthesis in the toxicity
of palmitate to HepG2 cells, we studied the effect of inhibiting ceramide synthesis on the
cytotoxicity. Fumonisin B1 (FBI) inhibits de nova ceramide synthesis by inhibiting
ceramide synthase, a key enzyme of the pathway. The cells were treated with palmitate in
the presence of 2-50 uM F Bl (Van Der Westhuizen, Shephard et al. 1998). Similar
concentrations of this inhibitor have been shown to inhibit de nova ceramide synthesis in

HepG2 cells (Van, Van Der Wouden et al. 2004). The cells exposed to palmitate had high

20

levels of C-16 ceramide (Figure 2.2a) whereas the cells treated with 4% BSA (control)
had negligible levels (not shown). Treatment with 20 uM FBI reduced the ceramide
levels significantly (42 and 72 % reductions on day l and day 2, respectively, Figure
2.2a). This suggested that the de nova ceramide pathway was responsible for the
increased ceramide synthesis. However, the reduction in ceramide levels was not
associated with a reduction in the toxicity (Figure 2.2b). Similarly, higher concentrations
of the inhibitor (up to 50 uM) were unable to reduce the toxicity. FBI is a specific
inhibitor of ceramide synthase (Y u, Yoo et al. 2004; Basnakian, Ueda et al. 2005), the
final enzyme of the de nava ceramide synthesis pathway. Therefore, the lack of reduction
in the toxicity in response to FBI, in spite of a reduction in ceramide levels, suggests that
the de nova synthesized ceramide does not play an important role in the toxicity. These
results are in agreement with other studies on the toxicity of saturated fatty acids to
hepatocytes (Feldstein, Wemeburg et al. 2006; Wei, Wang et al. 2006). These studies
indicate that saturated fatty acid may cause the toxicity to hepatocytes by mechanisms
other than ceramide generation, such as lysosomal permeabilization (F eldstein,
Wemeburg et al. 2006), BR stress (Wei, Wang et al. 2006), reductions in the
mitochondrial Bch/Bax ratio (Ji, Zhang et al. 2005) and activating JNK pathway (Malhi,

Bronk et al. 2006).

21

 

35 7
‘VlDay1l
30 d EIDayZ

C-16 Ceramides (pmollmg protein)

 

 

 

Palm FB1
Treatment

Figure 2.2a Increased ceramide synthesis in response to palmitate and inhibition by Fumonisin Bl
(FBl). Cells were exposed for two days to 0.7 mM palmitate containing vehicle (PBS), or 20uM fumonisin
Bl (FBI). The media were changed everyday. (a) Effect on C-16 ceramide levels. After treatments, the
lipids were extracted. The levels of C-16 ceramide in the cells were measured using LC-MS, and

normalized to the cellular protein. *, significantly lower than palmitate, p< 0.01.

 

LDH released (%)
N N u on It 48
c or c an o u-

—I
0|
l

10—

 

 

 

Palm FB1
Treatment

Figure 2.2b Effect of inhibition of ceramide synthesis on the toxicity of palmitate. The cells were
treated with palmitate and FBI for two days, and the LDH release was measured as described in materials
and methods section. Data presented as mean +/- s.d. for nine samples from three independent experiments.

22

2.3.3 Role of ROS in the cytotoxicity of palmitate. ROS has been shown to be a major
cause of palmitate-induced cytotoxicity in CH0 and microvascular endothelial cells
(Listenberger, Ory et al. 2001; Yamagishi, Okamoto et al. 2002). To investigate the role
of ROS in palmitate-cytotoxicity in HepG2 cells, a two-pronged approach was employed
- (a) the oxidized status of the cells was estimated using 6—carboxy-2',7'—
dichlorodihydrofluorescein diacetate, di(acetoxymethyl ester) (AM-HZ-DCF DA) dye and
the release of hydrogen peroxide by the cells was measured, (b) the effects of species-
specific ROS scavengers on the LDH release in response to palmitate were measured.
Exposure to palmitate, but not to oleate or linoleate, led to a time-dependent increase in

intracellular oxidized status (Figure 2.3a).

 

 

 

   

 

 

 

4.5
IPalm
4 - * IOle
o ClLln
g 3.5 - *
3
e 3' at
8
E 2.5 I * ‘
8 2~
°‘ *
g 1'5I
it
a 1 “
I:
0.5- ' ' I I
o -
4h 12h 24h

 

Time (h)

Figure 2.3a Levels of reactive oxygen species (ROS) in the cells following exposure to palmitate. Cells
were exposed to 0.7 mM palmitate for different times. Medium was changed after 24h of exposure.
Treatment with palmitate was followed by exposure to 5 uM AM-DCF DA dye for 30 minutes at 37°C, and
staining with 2 ug/ml propidium iodide for 10 minutes. Fluorescence of viable (propidium iodide negative)
cells was measured using flow-cytometry with excitation at 488 nm and emission at 530 nm. Relative

fluorescence values +/- s.d. of six samples from two different experiments are presented. ‘* ’ denotes
significantly different, compared to control, p < 0.01.

23

A measure of ROS production by the cells is the release of hydrogen peroxide (Liu,
Fiskum et al. 2002). The hydrogen peroxide released by the cells in response to palmitate
was measured by a colorimetric assay as described in the methods section. Exposing cells
to palmitate, led to a time-dependent increase in hydrogen peroxide release into the
medium (Figure 2.3b). The profile of hydrogen peroxide release mirrored the observed
toxicity, being lower on day I and increasing significantly on day 2, suggesting that

hydrogen peroxide may be related to the observed toxicity.

 

Hydrogen Peroxide Released
(n moll mg protein)
N w -h 0| 0! \I
O O O O O D

d
O
I

 

 

 

0

CM 12 24 36 48
Hours of Exposure

Figure 2.3b Release of hydrogen peroxide into the medium following palmitate exposure. Cells were
treated with 0.7 mM palmitate for various times. The hydrogen peroxide released into the medium was
measured and normalized to total cellular protein. Data presented as mean +/- s.d. for nine samples from

three independent experiments. ‘ * ’ denotes significantly different, compared to control, p < 0.01.

hydrogen peroxide can form *OH radical, a much more reactive and damaging ROS
species, in the presence of divalent cations such as Fe”. To evaluate the role of hydrogen

peroxide and *OH radical in palmitate toxicity, the cells were treated with palmitate in

24

the presence of catalase (3000U/ml) and *OH radical scavengers l, 3 dimethyl urea
(DMU, 25mM) and dimethyl sulfoxide (DMSO, 1% v/v), and the spin trap alpha-(4-
pyridyl-1-oxide)-N-tert-butyl nitrone (POBN, 70 mM). All these scavengers reduced the
cellular ROS levels (Figure 2.3c), as well as the palmitate-toxicity (Figure 2.3d)
significantly. DMU and DMSO are two specific scavengers of hydroxyl radicals
(Cederbaum, Dicker et al. 1977; Sandler and Andersson 1982; Bruck, Aeed et al. 1999;
Zegura, Sedrnak et al. 2003). Dimethyl thiourea (DMTU) is another commonly used *OH
scavenger, but was not employed in the current study because it had been shown to be
toxic to rats (Beehler, Ely et al. 1994). The same study showed that DMU was not toxic,

and therefore, DMU was used in the present study.

 

2.5

24

1.5 ~

Relative ROS levels

 

 

 

Treatment

Figure 2.3c Effects of scavengers of hydrogen peroxide and hydroxyl radical on the oxidized status of
the cells. The cells were treated with pahnitate in the presence of 3000 U/ml catalase (scavenger for

hydrogen peroxide), or scavengers of *OH radical- dimethyl urea (25 m M), dimethyl sulfoxide (DMSO,
1% v/v), or the spin trap alpha-(4-pyridyl-l-oxide)-N-tert-butyl nitrone (POBN, 70 mM) for 24h, followed
by treatment with 5 uM AM-DCF DA for 30 min in DMEM. Fluorescence of the cells was measured using
a rrricroplate reader with excitation at 488 nm and emission at 530 nm. Relative fluorescence values +/- s.d.

of six samples from two independent experiments are presented. ‘* ’ denotes significantly higher,
compared to control, p < 0.01.

25

45
40 J
35 -

 

N 09
0| O
l 1

N
O

LDH released (%)
a

 

 

 

—I
O 01 O
l l 1

Palm Cat DMU DMSO POBN
Treatment

Figure 2.3d Effects of scavengers of hydrogen peroxide and hydroxyl radical on palmitate-toxicity.
The cells were Heated as in Figure 2.3(c) for 2 days. LDH released on day two is presented. Data presented

as mean +/- s.d. for nine samples from three independent experiments. ‘*’ denotes significantly lower,
compared to palmitate, p < 0.01.

Studies with other cell types have also indicated that superoxide or NO may mediate the
toxicity of palmitate (Listenberger, Ory et al. 2001; Maestre, Jordan et al. 2003;
Okuyama, Fujiwara et al. 2003). To evaluate the role of superoxide anion in the toxicity,
cells were either pretreated for 2h or co-treated with palmitate in the presence of cell-
permeable SOD mimics CuDIPS (0.5-10 uM) and MnTBAP (25-200 uM), or the
superoxide—specific scavengers 4,5-dihydroxy-1,3-benzene-disulfonic acid (Tiron, 2-20
uM) and pyrrolidone dithiocarbamate (PDTC, 0.5-10 uM). However, none of these
treatments produced any significant reduction in the toxicity of palmitate, suggesting that
the superoxide anion is not the primary toxicity causing species. To evaluate the role of

NO in the toxicity, the effect of NO scavengers (cPTIO, 25-200 uM and AG, 0.1-1 mM)

26

or iNOS inhibitor 1400W (50-500 uM) on the palmitate toxicity was measured. Exposure
to palnritate in the presence of these chemicals did not reduce the cytotoxicity, indicating
that NO is not involved in the toxicity of palmitate. The efficacy of hydrogen peroxide
and *OH scavengers, but not of superoxide or nitric oxide scavengers or inhibitors,
indicated that the former ROS are the likely mediators of the palmitate-toxicity in HepG2

cells.

2.3.4 Exposure to palmitate is associated with the loss of mitochondrial potential.
Cytotoxicity is usually associated with the loss of mitochondrial membrane potential. A
significant reduction in the mitochondrial potential was observed when the cells were
treated with palmitate, but not when they were treated with unsaturated fatty acids
(Figure 2.4a). The potential after exposure to palmitate was higher than that observed on
treating the cells with 40 uM FCCP (positive control), suggesting a partial loss of
potential upon palmitate-exposure. To test the involvement of ROS in the observed
reduction of the mitochondrial potential in response to palmitate, the cells were exposed
to 100 uM hydrogen peroxide in the absence of palmitate. This treatment also
significantly reduced the mitochondrial potential (Figure 2.4a), suggesting that the
oxidative stress in response to palmitate may play a role in the loss of mitochondrial

potential.

27

1.4

 

[-5517

1.2 -

Relative Mitochondrial Potential

 

 

 

Control Ole Lin Palm H202 FCCP
Fatty Acid

Figure 2.4a Mitochondrial potential after exposure to fatty acids or hydrogen peroxide for 24 h. Cells
were exposed tolOO uM hydrogen peroxide, to various free fatty acids (0.7 m M) or to 40 uM FCCP for
24h. The cells were then incubated in 5 uM JC-l dye in serum-free medium for 30 min at 37°C, after which
the cells were washed with PBS. The fluorescence was measured in a microplate fluorimeter, with
excitation at 488 nm and emissions at 530 (green) and 590 nm (red). Data presented as mean +/- s.d. for

nine samples from three independent experiments. *, significantly different from control, p < 0.01 relative
to control.

MPT has been suggested as a cause of reduced mitochondrial potential and cell death in
response to saturated fatty acid in beta cells (Kong and Rabkin 2000). To evaluate the
involvement of MPT in the loss of mitochondrial potential and toxicity by the palmitate-
treatment, MPT was inhibited by treating the cells with 0.5-5 uM CSA in the presence of
palmitate. This treatment restored the mitochondrial potential on both days (Figure 2.4b,
data shown for 2 uM). However, the improvement in the mitochondrial potential was not
associated with a reduction in the toxicity (Figure 2.4c), indicating that the toxicity was

not dependent on the loss of mitochondrial potential. Similarly, the other concentrations

28

of the inhibitor (0.5-5 uM) were also unable to significantly reduce the toxicity (not

shown).

1.2

 

 

 

Relative Mitochondrial Potentia

 

 

 

Treatment

Figure 2.4b Effect of hydroxyl radical scavengers and the MPT inhibitor on the mitochondrial
potential. Cells were treated with 0.7 mM palmitate in the presence of hydroxyl radical scavengers DMSO
(1% v/v), DMU (25 mM), or MPT inhibitor CsA (2 uM) for 24h and the mitochondrial potential was
measured, as in ‘a’. Data presented as mean +/- s.d. for nine samples from three independent experiments.

‘ * ’ denotes significantly different from palmitate, p < 0.01.

29

 

 

 

 

Treatments

Figure 2.4c Effect of MPT inhibitor cyclosporine A on the toxicity of palmitate. Cells were treated for
2 days with palmitate in the presence of 2 uM CsA. LDH release was measured and normalized to total
LDH. Data shown is mean +/- s.d. for nine samples from three independent experiments.

In agreement with these observations, treating with palmitate in the presence of DMU,
DMSO, catalase and POBN (all of which had reduced the LDH release significantly) did
not affect the reduction in mitochondrial potential by palmitate (Figure 2.4b). Thus, the
treatment with palmitate leads to MPT, but the toxicity of this fatty acid does not depend

on MPT.

2.3.5 Role of ROS in MPT: effect of mitochondrial complex inhibition. ROS and
MPT can have a two-way interaction, i.e., some studies have reported that MPT has been
associated with increased production of ROS through mitochondrial uncoupling, while
ROS can cause oxidation of the thiol-groups on cyclophilin D (Cyp-D) and adenine

nucleotide translocator (ANT), causing MPT. Mitochondrial complexes I and III are

30

known to be the primary sites of ROS production (Chen, Vazquez et al. 2003; Nohl, Gille
et al. 2005; Grivennikova and Vinogradov 2006). The observation of loss of
mitochondrial potential also suggested alteration of mitochondrial structure and function.
In order to evaluate the role of ROS production at these mitochondrial complexes on the
MPT and the LDH release caused by palmitate, the effects of inhibiting mitochondrial
complexes I and III on these processes were studied. Cells were pretreated for 30 minutes
with various concentrations of rotenone and antimycin A, irreversible inhibitors of
complex I and III, respectively. Inhibiting these complexes partially improved the
mitochondrial potential (Figure 2.53) on day l and reduced the LDH release (Figure 2.5b)
on both days. The partial recovery on day 1 and no effect on day 2 suggested that
mechanisms other than ROS generation at complexes I and III are playing roles in the
MPT caused by palmitate. Furthermore, co-treatment of the cells with 100 nM rotenone
(0.35 pmol/ mg protein) or 50 nM antimycin A (0.17 pmol/ mg protein), following the
pretreatment of the cells with 500 nM rotenone (1.8 nmole/ mg protein) or 200 nM
antimycin A (0.7 nmol/ mg protein), did not protect against the potential loss. These
result show that the ROS production by complexes 1 and III mediates, partly, the early
potential loss and the LDH release in response to palmitate-exposure, but mechanisms
other than ROS production at these complexes are also involved in the MPT induced by
palmitate. The effect of these inhibitors on the LDH release in the absence of palmitate
was also evaluated. Pretreating the cells for 30 min with 500 nM rotenone (1.8 nmole/ mg
protein) and 200 nM antimycin A (0.7 nmol/ mg protein), followed by treatment with
control medium, caused a slight increase in toxicity. Nevertheless, even with 1 uM

rotenone and 500 nM antimycin A, the LDH release was less than 2% (not shown), much

31

less than that observed with palmitate and the inhibitors. This indicated that the
remaining toxicity in the palmitate samples in the presence of these inhibitors is due to

some mechanism other than ROS production at these sites (complexes) in response to

 

palmitate.
1.2 e — A“
* * I Day 1
1 a * * E! Day 2 ‘
0.8 * *

Relative Mito Potential
O
c»

 

 

 

 

Treatment

Figure 2.5a Effects of inhibiting mitochondrial complexes 1 and 111 on mitochondrial potential loss.
Cells were pretreated with different concentrations of rotenone or antimycin A for 30 minutes, followed by
the treatment with 0.7 mM palmitate. At the end of 24h exposure, the media were removed and cells treated
with the same concentrations of rotenone and antimycin A for 30 min, followed by exposure to palmitate
for another 24h. Mitochondrial potential was measured after the 24 and 48 h of exposure to palmitate, using
JC-1 dye, as explained in the materials and methods. Data presented as mean +/- s.d. for nine samples from

three inde endent experiments. ‘*’ denotes significantly different, com ared to almitate, < 0.01.
P P P P

32

 

IDay 1
40 - Day 2

 

 

 

LDH released (%)

 

 

 

 

 

 

 

Ctrl Palm Rote- Rote- Rote- AA- AA-
0.2uM 0.5uM 1.0uM 0.1uM 0.2uM

Treatment

Figure 2.5b Effects of inhibiting mitochondrial complexes I and III on the cytotoxicity of palmitate.
Cells were pretreated with different concentrations of rotenone (Rote) or antimycin A (AA) for 30 minutes,
followed by treatment with 0.7 mM palmitate. After 24h of exposure, cells were treated again with the
inhibitors for 30 min, and then exposed to palmitate for another 24h. LDH release after 24 and 48h were

measured and normalized to the total LDH. ‘ * ’ denotes significantly different, compared to palmitate, p <
0.01.

2.5 Discussion

Free fatty acids are known to play important roles in the development of many hepatic
complications. Previous studies have shown that saturated fatty acids are important
mediators of lipotoxicity. Studies with other cells have implicated de novo ceramide
synthesis (Listenberger, Ory et al. 2001; Yamagishi, Okamoto et al. 2002) in the
palmitate-induced cell death. Palmitate is a substrate of ceramide synthesis and has been
shown to increase de novo synthesis of ceramide (Listenberger, Ory et al. 2001). The
inhibitor experiments conducted in this study suggest that the palmitate-toxicity to

hepatoma cells is not mediated by de novo ceramide synthesis. These results are in

33

agreement with another recent study (Wei, Wang et al. 2006), where the authors reported
that the cell death caused by saturated fatty acids in murine hepatoma cells is independent
of the de novo ceramide generation. It is likely that other mechanisms of cell death such
as ER stress (Wei, Wang et al. 2006), lysosomal permeabilization (Feldstein, Wemeburg
et al. 2006), JNK activation (Malhi, Bronk et al. 2006) or reduction in mitochondrial
Bch/Bax ratio (Ji, Zhang et al. 2005) play important roles in the toxicity of the saturated

fatty acid to HepG2 cells, as observed in the studies with primary hepatocytes.

Another important mediator of cell death by saturated fatty acids is the increased
production of ROS. Mitochondrial complexes I and III are the primary sites of ROS
generation by the cells. Inhibiting these complexes reduced the cytotoxicity significantly,
suggesting that these complexes are an important source of ROS in the palmitate-toxicity
to hepatoma cells. The complex I and III inhibitors employed, rotenone and antimycin A,
can themselves cause cell death through increased ROS production and MPT at higher
concentrations or prolonged exposures (N iknahad, Khan et al. 1995; Isenberg and
Klaunig 2000). Therefore, a dose-response study on the toxicity with these compounds in
the presence of palmitate was conducted to identify the optimal treatment concentrations
and times. In agreement with previous studies, an increase in the cytotoxicity was
observed when the cells were treated at higher concentrations or longer durations.
However, for the optimal treatment conditions shown (pretreatment for 30 min with 0.5
uM rotenone or 0.2 uM antimycin A), a maximum reduction in the cytotoxicity (55 i 10
% decrease in LDH release for the rotenone pretreatment compared to palmitate

treatment alone) was observed. The cytotoxicity was not completely prevented on

34

treating with mitochondrial complex inhibitors. This suggests that mechanisms other than

ROS production in the mitochondria may also contribute to the toxicity of palmitate.

Treatment with palmitate can lead to increased generation of NO in iNOS dependent
(Shimabukuro, Ohneda et al. 1997) and independent (Maestre, Jordan et al. 2003) ways.
While increased NO generation has been suggested to play a role in palmitate toxicity to
a beta-cell line INS-1 (Maestre, Jordan et al. 2003), studies in pancreatic islets have
indicated no involvement of NO in response to elevated fatty acid exposure (Cnop,
Hannaert et al. 2001). Conflicting roles of NO in mediating the cell death in hepatocytes
have been reported. While NO can cause cell death by itself (D'Ambrosio, Gibson-
D'Ambrosio et a1. 2001), it has been shown to protect against TNF-a-induced cell death
(Kim, de Vera et al. 1997; Kim, Kim et al. 2000). It is very likely that NO is non-toxic
(and perhaps cytoprotective) at low concentrations, but becomes toxic beyond a certain
threshold concentration, which depends on the cell-type. In the present study, either NO
did not play a major role in the toxicity or the NO produced in response to palmitate does

not reach toxic levels in these cells.

Hepatocytes are known to have high levels and activities of SOD. Therefore, the
superoxide anions generated can be effectively neutralized to hydrogen peroxide. This
may explain, in part, why the SOD mimics were not effective in reducing the cell death.
The SOD mimics will not reduce the production of hydrogen peroxide - the cytotoxicity
mediating species. Hydrogen peroxide is freely diffusible across membranes and thus

may cause oxidative stress in neighboring cells as well. Extracellular neutralization of

35

hydrogen peroxide using a high molecular weight catalase reduced palmitate-toxicity
significantly, indicating that hydrogen peroxide is the likely carrier of oxidative stress to
nearby cells. An increased concentration of hydrogen peroxide was observed in the
medium in response to palmitate-exposure, but not in the control medium. However, this
increase is not due to an accumulation of hydrogen peroxide, as hydrogen peroxide is not
expected to be stable for hours. While in the absence of cells, hydrogen peroxide is stable
for at least 30 min (Dringen and Hamprecht 1997), the typical half—life in the presence of
cells is a few minutes (Barnard and Matalon 1992; Dringen and Hamprecht 1997).
Therefore, these values likely indicate an increase in cellular secretion rather than
accumulation of hydrogen peroxide. Another variable which would control the net rate of
release by the cells is the level of peroxide-metabolizing enzymes, such as catalase and
glutathione peroxidase. Fatty acids have. been shown to alter the expression of these
enzymes. For example, a high fat diet rich in saturated fatty acid has been shown to
reduce the activities of glutathione peroxidase (GPx) and catalase (CAT), while
increasing the expression of SOD (Yuan, Kitts et al. 1998; Lu and Chiang 2001). Thus
the saturated fat may lead to reduced catabolism of the generated hydrogen peroxide, thus

increasing the net release.

One of the important mediators of hydrogen peroxide toxicity is the generation of *OH in
the presence of divalent cations by the Fenton and Haber-Weiss reactions. *OH is an
extremely reactive and damaging radical and can oxidize any lipid, protein or DNA

molecule in its vicinity. Treating the cells with palmitate in the presence of *OH-

36

scavengers significantly reduced the toxicity, indicating an involvement of this radical in

the cytotoxicity of palmitate.

Mitochondria are the organelles centrally associated with many fundamental cellular
processes, such as energy-generation. They are also the reservoirs of many apoptogenic
factors, such as cytochrome c, apoptosis inducing factor and endonuclease G, etc. A loss
of structural and functional integrity of these organelles can lead to cell death. Transport
of chemicals and proteins to and from these organelles is strictly regulated. The
concentration difference of solutes and the resulting potential gradient across
mitochondrial inner membrane is crucial for mitochondrial structural and fimctional
integrity. One of the key regulators of transport of low molecular weight solutes across
the inner mitochondrial membrane is the mitochondrial permeability pore (MPP)
complex. MPP is a protein complex found at the junctions of the inner and outer
membranes, which controls the influx of small molecular weight compounds into the
mitochondria. MPT refers to the opening of this pore, a process which brings in solutes of
up to 1.5 kDa inside the mitochondria, resulting in the loss of mitochondrial potential,
swelling of the mitochondria, a reduction in the ATP generation and production of ROS
(Bemardi, Broekemeier et a1. 1994). The swelling leads to the rupture of the outer
membrane, releasing intermembrane apoptogenic factors such as cytochrome c, apoptosis
inducing factor and endonuclease G (Weiss, Korge et al. 2003). MPT also causes loss of
anti-oxidants such as glutathione from the mitochondria, making them even more
susceptible to ROS-induced damage (Kroemer, Dallaporta et al. 1998). Two types of

MPT have been reported- classical, which is inhibited by CSA, and non-classical, which

37

is not inhibited by CsA. According to in vitro studies, employing isolated mitochondria

or liposomes, palmitate can cause non-classical MPT (NC-MPT) by forming channels in

the mitochondrial membrane by complexing with Ca2+ (Agafonov, Gritsenko et a1.

2003), or by interacting with and altering mitochondrial membrane permeability
(Bemardi, Penzo et al. 2002). A complete prevention of mitochondrial potential loss upon
treatment with CsA suggested that the loss of mitochondrial potential on exposure to
palmitate is due to classical MPT. However, the inhibition of MPT had no effect on the
cytotoxicity indicating that the cytotoxicity was not dependent on MPT. CsA may also
inhibit protein phosphatase ZB (PP2B/ calcineurin). Inhibition of PP2B is also associated
with a reduction in apoptosis (Su and Schneider 1997). Among the various organs, the
liver isoforrn of PP2B is least active and is resistant to CsA inhibition (Wojtczak and
Schonfeld 1993). Thus, it was presumed that the effects of CSA are not likely due to

calcineurin inhibition.

Palmitate can also indirectly cause MPT by increasing the generation of ROS. ROS can
oxidize the thiol groups on cyclophilin-D (Cpr) and adenine nucleotide translocase
(ANT) (Halestrap, Woodfield et al. 1997), two important proteins of MPP, causing MPT.
Elevated ROS generation also leads to lipid peroxidation, which can also cause MPT
(Gogvadze, Walter et al. 2003). Mitochondrial complexes I and III are the primary sites
of ROS generation. Inhibiting these complexes reduced the loss of potential on day l,
albeit incompletely, but had no effect on the mitochondrial potential loss on the second
day. These results suggest that (a) generation of ROS at complexes I and III plays an

important role in the initial potential loss caused by these fatty acids, and (b) other

38

mechanisms are involved in the MPT by fatty acids. Fatty acids may cause MPT by many
mechanisms other than mitochondrial ROS generation. It is likely that the production of
hydrogen peroxide due to oxidation of fatty acids in peroxisomes may, over time, cause
MPT. Fatty acids can also cause MPT independently of ROS. Fatty acids are
mitochondrial uncouplers, and can reduce mitochondrial potential to below the ‘gating
potential’ (mitochondrial potential below which MPT occurs), causing MPT (Schonfeld
and Bohnensack 1997). They can also cause MPT by binding to and causing structural
changes in the ANT (Wieckowski and Wojtczak 1998). Recently, it was demonstrated

that fatty acids can cause endoplasmic reticulum (ER) stress (Wei, Wang et al. 2006). BR

. . . 2+ . .
rs an important reservorr of cellular Ca . ER stress would increase cytosolic and

mitochondrial Ca2+ levels (Berridge 2002; Tiwari, Kumar et al. 2006), which can cause

MPT by itself, or increase the sensitivity of the MPT to oxidative stress (Halestrap,
Connem et al. 11997). Palmitate can reduce ATP synthesis due to its uncoupling action
(Kadenbach 2003), and reduce cardiolipin synthesis (Ostrander, Sparagna et al. 2001).
Because MPT is sensitive to both ATP levels and mitochondrial membrane structure,

these effects of palmitate would make the mitochondria more susceptible to MPT.

As both cytotoxicity and MPT were observed in the palmitate-treated cells, and were
significantly reduced by the reduction in ROS generation, it might be tempting to
speculate that the same ROS species is mediating both these effects. However, the
inefficacy of the *OH radical scavengers (which reduced the toxicity significantly) on
MPT suggests that these radicals are not involved in causing the MPT in response to

palmitate.

39

In summary, hydrogen peroxide and *OH radical, but not de novo ceramide synthesis, are
the primary mediators of the cytotoxicity of palmitate in hepatoma cells. ROS generation
by the mitochondrial complexes I and III is responsible for mediating the toxicity as well
as, in part, the early potential loss in response to palmitate. Thus, while MPT and
cytotoxicity are related, in that the exposure to palmitate is associated with both MPT and

cytotoxicity, the cytotoxicity is not mediated by MPT.

40

CHAPTER 3. APPLICATION OF METABOLIC FLUX
ANALYSIS TO IDENTIFY ALTERED FLUXES
ASSOCIATED WITH'THE TOXICITY OF PALMITATE

3.1 Introduction

Hepatocytes play a central role in whole-body lipid metabolism and FFAs have been
shown to play an important role in the development of such hepatic disorders as steatosis
and non-alcoholic steatohepatitis (NASH) (Farrell and Larter, 2006). Steatosis refers to a
condition where there is excess accumulation of lipids inside the cells, leading to altered
cellular function. While benign by itself, steatosis primes the cells to injury by oxidative
insults (Angulo and Lindor, 2002). NASH refers to a condition of severe inflammation
and necrosis of the liver, and can be fatal. With the increased incidences of obesity, the
reported cases of NASH are on the rise. It is suggested that about 30% of the population
in the US suffers from steatosis (Browning et al., 2004). Of those, 20% of the patients
develop NASH (McCullough, 2006). One of the important markers of NASH is extensive
hepatocyte cell death and increased serum AST/ALT levels, an indicator of hepatocyte
injury. FFAS have emerged as one of the primary factors in the development of NASH
(Scheen and Luyckx, 2002). However, the mechanisms behind FFA toxicity remain
unclear. Thus, it is important to elucidate the alterations in overall cellular metabolism in

order to understand the development of this disease and devise effective treatments.

Previous studies have identified that the saturated F F As are much more toxic to
hepatocytes and hepatoma cells than the mono and poly-unsaturated FFAS (Srivastava

and Chan, 2007, Malhi et al., 2006 and Wei et al., 2006). It has also been shown that

41

metabolism of these fatty acids is essential for their cytotoxic response (Shimabukuro et
al., 1998). Therefore, we hypothesized that the differences in the metabolism of the
saturated and unsaturated FF As contribute to their differences in cytotoxicity. We studied
the global changes in the metabolism of these cells in response to FFAS by utilizing
Metabolic Flux Analysis (MFA). MFA has previously been applied to understand the
hyperrnetabolism of hepatocytes following burn injury (Lee et al., 2003), to optimize
hepatocyte functions for bio-artificial liver applications by identifying the changes in
hepatocyte metabolism on exposure of plasma (Chan et al., 2003a) and the effects of
supplementing the media with hormones and amino acids (Chan et al., 2003b). MFA has

also been applied to identify the metabolic effects of hypoxia-inducible factor (HIP-10L),

to identify if HIF-la can be a potential target in cancer treatment (Kim and Forbes,

2006).

In the current study, we measured various extracellular metabolic fluxes and applied
MFA to identify the changes in the intracellular metabolism in response to the saturated
and unsaturated FF As for a period of three days. It was observed that the saturated FFA
was oxidized to a greater extent and was channeled into TG to a lesser extent than the
unsaturated FFA oleate. Studies with other cell types have identified that de novo
ceramide synthesis in response to saturated F F A may play an important role in causing
cell death. The flux through the de novo ceramide synthesis was indeed higher in the
palmitate-treated cells, but inhibiting this pathway did not produce any change in the
cytotoxicity. MFA also identified that the flux through the synthesis of glutathione (GSH)

was significantly reduced in cells treated with palmitate but increased in cells treated with

42

oleate, which was experimentally verified. The levels of cysteine transporter xCT, but not
that of glutamyl cysteine synthase (GCS), were reduced by palmitate. Supplementation of
the cells with N-Acetyl L-cysteine (NAC), which increases the GSH synthesis by
providing cysteine independently of the xCT activity, reduced the toxicity of palmitate.
Thus, the metabolic alterations caused by saturated F FAs were identified by MFA and

served as targets to regulate the FFA toxicity, which were experimentally validated.

3.2 Materials and Methods

3.2.1 Materials

Glucose, lactate and free glycerol kits were purchased from Sigma. Enzymatic
triglyceride and beta-hydroxybutyrate assay kits were purchased from Stanbio
Laboratories. Free Fatty Acid Half Micro kit was purchased from Roche Biochemicals.
AccQTag amino acid analysis HPLC column, AccQTag solvent and the sample

preparation reagents were purchased from Waters.

3.2.2 Measurements of Extracellular Fluxes

Extracellular fluxes of various metabolites were measured by calculating the changes in
the concentration of the metabolite in the medium after a specified duration of exposure
(typically 24 h). The linearity of some fluxes over this interval was verified. The
concentrations of glucose, lactate, FFA, glycerol and beta-hydroxybutyrate were
measured by enzymatic kits based on manufacturers’ instructions. The concentration of
acetoacetate in the media was measured by an enzymatic fluorimetric assay (Olsen 1971).

The accumulation of intracellular triglycerides was measured by lyzing the cells with 1%

43

triton in PBS and measuring the concentration of triglycerides in the supernatant by
enzymatic triglyceride assay kit from Stanbio, according to manufacturer’s instructions.
This method measures the amount of glycerol released due to the enzymatic hydrolysis of
the TG. To correct for the free glycerol in the cells, the concentration of free glycerol in
the cells were measured by free glycerol assay from Sigma, and subtracted from the
glycerol values of the TG assay, to give actual triglyceride values. Concentrations of Asp,
Glu, Gly, NH3, Arg, Thr, Ala, Pro, Tyr, Val, Met, Om, Lys, Ile, Leu and Phe were
measured by the AccQTag amino acid analysis method (Waters). The concentrations of
Ser, Asn, Gln and His were measured by a modification of the AccQTag method. Cystine
concentration in the media and supematants was measured using HPLC according to a
previously published protocol (F iskerstrand et al, 1993). Briefly, the protein were
precipitated using sulfosalicylic acid (final concentration 5% by weight), the cystine was
reduced to cysteine using sodium borohydride and derivatized using monobromobimane
(MBB). 20 ul of the samples were then injected onto a Hypersil-ODS silica column
equilibrated with 30 mM ammonium nitrate and 40 mM ammoinium fonnate buffer, pH
3.5. Cysteine was separated by a gradient of acetonitrile (0—10.5%) in 11 min. All the
measured fluxes were normalized to total protein in the cell extract, measured with the

bicinchoninic Acid (BCA) method (Pierce Chemicals).

3.2.3 Metabolic Flux Analysis
MFA is essentially a mass-balance on various intracellular metabolites under the
assumption of pseudo-steady state. Under this assumption, the differential equations

representing the changes in the concentrations of the various metabolites may be

44

represented by the algebraic summation of the fluxes associated with a particular
metabolite. Thus, various such balances for all the intracellular metabolites under
consideration may be represented in matrix notation as

S*v = 0

where, S is the stoichiometric matrix and v is the vector of metabolic fluxes. If m of the

fluxes are measured, then it is possible to partition the matrix S into two submatrices Sm
and Su such that

S.,,"‘vm + S.,"‘vu = 0

Or. vu = (S..* S..’)"* Sm*vm

Thus, it is possible to obtain the vector of unknown fluxes vu from knowledge of the

stoichiometry and a list of measured fluxes vm.

Assumptions

The assumptions employed in the current study with respect to the MFA model are stated

below:

i. Pseudo-steady state was assumed. MFA requires that the intracellular metabolites
be under the condition of pseudo-steady state, that is there is no significant
accumulation of any metabolite. While the exposure to FFAS would change the
concentrations of some of the intracellular metabolites, this change was assumed to
be fairly small compared to the overall flux of the metabolite. Experiments were

conducted to verify this assumption by measuring the profile of intracellular

45

ii.

iii.

iv.

metabolite levels and comparing them with the extracellular metabolic fluxes (See
section 3 of the Results).

The fluxes of metabolites are linear over the 24 h period. Fluxes of the uptake or
release of metabolites were assumed to be linear over the 24h period, such that the
net change in the extracellular concentration of the metabolite after 24h duration
represents the flux of that metabolite. To test this assumption, we measured the
release of lactate over a 24 h period and found that the flux was linear for the 24 h
period (See section 3 in Results).

The secretion of triglycerides is much lesser than its accumulation. It was assumed
that the rate of secretion of triglycerides in the form of lipoproteins was much
smaller than the intracellular accumulation of triglycerides. This assumption is valid
based on studies with HepG2 cells which identified that these cells have active
synthesis but impaired secretion of lipids (Gibbons et al., 1994). Additionally, we
also measured the levels of extracellular triglycerides in response to various
treatments and observed that the levels of extracellular TG were much lesser than
that of the intracellular TG (not shown). It may be noted that intracellular TG was
considered as an “external” flux, that is, it was assumed that the TG synthesized is
rapidly sequestered into lipid droplets. This is valid because TG is very
hydrophobic and forms lipid droplets inside cells (if not exported).

The metabolites are uniformly distributed inside the cell. This assumption implied
that a single MFA model could be applied to perform metabolite balances and

separate models were not needed for the various compartments. Many previous

46

studies have found that this is a fair assumption for mammalian cells (Chan et al,
2003 and Kim and Forbes, 2006).

v. There is no formation of cysteine from methionine. HepG2 cells cannot synthesize
cysteine from methionine due to the lack of S-adenosyl-methionine synthetase
activity (Lu and Huang, 1994). Therefore, this equation was not included in the
model.

vi. Either fatty acid synthesis or its oxidation dominates: Formation of a stable system
of equation requires that none of the equations be linear combinations of any other
equation(s). Meeting this requirement meant that only one of the bidirectional
pathways be considered. Therefore, it was assumed that, in FFA-treated cells, the
exogenous FF A supply was much greater than the de novo fatty acid synthesis, and
the latter may be neglected. Opposite assumption was made for the BSA control

cells, i.e. fatty acid oxidation was neglected

Based on these assumptions, a MFA model consisting of 84 or 85 cellular metabolic
reactions (for control and F FA-treated cells, respectively) involved in the metabolism of
glucose, protein and lipids (including the synthesis of sphingo- and phospho- lipids) was
developed (Table 3.1). The model consisted of 66 intracellular metabolites for control
and palmitate-treated cells and 67 for oleate-treated cells (Table 3.2). In order to solve the
system of equations, 28 metabolic fluxes were measured (Table 3.3). This yielded an
over-determined system of equations which were solved by least-squares fit using the

Moore-Penrose pseudo-inverse calculation.

47

Table 3.1 List of the metabolic reactions included in the MFA model

 

Flux

# Equation

Glycolysis

G+ATP -> G.6-P

G-6-P -> F-6-P

F-6-P +ATP -> Gcheraldehyde-B—P + DHAP
DHAP -> Glyceraldehyde-3P
Glyceraldehyde-3—P -> 3-PGA

3PGA —-> PEP + NADH (+ 2ATP)

PEP -> Pyr

Pyruvate + NADH —-> (Lactate)
Pyruvate -> Acetyl-CoA + NADH +C02

OQNmUI-hODN-l

TCA cycle
10 AcetyI-CoA + 0AA -> Citrate
11 Citrate -> a-KetoGIutarate + NADPH +C02
12 a-Ketoglutarate —> SuccinyI-CoA + NADH + 002
13 Succinyl-CoA -> Fumarate + FADH2 + ATP
14 Fumarate -> 0AA + NADH

Pentose-phosphate pathway
15 G-6—P -> 12 NADPH + 6 002
16 002 out

Ketone Body production
17 2 AcetyI-COA -> AcetoacetyI-CoA
18 Acetoacetyl-CoA -> Acetoacetate
19 Acac out
20 Acetoacetate + NADH -> (B-OH butyrate)

Oxygen uptake and Oxldatlve Phosphorylatlon
21 02 (In)
22 NADH + 0.5 02 -> 2.5 ATP
23 FADH2 + 0.5 02 -> 2 ATP

FFA synthesls and oxidation
24 Glycerol + ATP --> Glycerol-3-P

25 Glycerol-3-P -> Glyceraldehyde-S-P + NADH
*
26 FA-CoA -> 8(9) AcetyI-CoA + 14(16) NADH
a
26 8 Acetyl-CoA + 14 NADPH -> Palm-CoA

27 FA—CoA + DAG -> TG
77 2 FA-CoA + Glycerol-3-P -> Phosphatidate

78 Phosphatidate -> DAG

85* FA-CoA (In)

48

Table 3.1 continued

Amino acid metabolism

28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
56
58
59
60
61
62
63
54
34

Ser (In)

Ser --> NH3 + Pyr

Gln In

His -> Glu + NH4

Asp (In)

Glu (In)

Gly (In)

Gly -> 2 002 + NH3 + NADH + THF + ATP

NH4 (In)

Arg In

Thr -> Pyr + 002 + NH4 + 2 NADH + FADH2

Ala (In)

Glu + Pyr —> Ala + aKG

Pro In

Cys In

Tyr (In)

Tyr + aKG + 2 02 --> Glu + 002 + Acetoacetate + Fumarate
Val + aKG ——> Glu + 002 + 2NADH + FADH2 + Succ-CoA
0m IN

Lys + 2 aKG + NADPH -> 2Glu + Acetoacetyl-CoA + 2002 + 4 NADH + FADH2
He + aKG --> Glu + Succ-CoA + Acetyl-CoA + NADH + FADH2
Leu + aKG --> Glu + AcetyI-CoA + Acetoacetate + 002 + NADH + FADH2
Phe + 02 -> Tyr

Glu + Cys + Gly -> GSH

Gln -> Glu + NH4

Asp + NH4 -> Asn

Thr -> Pyr + 002 + NH4 + 2 NADH + FADH2

Val + aKG -> Glu + 002 + 2NADH + FADH2 + Succ-CoA

Lys + 2 aKG + NADPH —> 2Glu + Acetoacetyl-CoA + 2002 + 4 NADH + FADH2
He + aKG --> Glu + Succ-CoA + AcetyI-CoA + NADH + FADH2
Leu + aKG --> Glu + HMG-CoA + NADH + FADH2

Phe + 02 —> Tyr

Glu -> a-KG + NH3 + NADPH

Urea Cycle

52
53
54
55
57

H003- + NH4 + 0m + 2 ATP -> Citrulline
Citrulline + Asp + ATP -> Fumarate + Arg
Arg -> 0m + Urea

Urea Out

0m + a-KG + 0.5 NADPH + 0.5 NADH --> Pro

Synthesis of Cholesterol and Cholesteryl ester

67
68
69
70
71
72

Acetoacetyl-CoA + Acetyl-CoA --> HMG—CoA
HMG-CoA + 2 NADPH (+ 3ATP) -> IPP

2 IPP --> Geranyl-PP

Geranyl-PP + IPP -> FamesyI-PP

2 FamesyI-PP + 0.5 NADPH + 0.5 NADH --> Squalene
Squalene + 02 + NADPH --> Lanosterol

49

73
75

Table 3.1 continued

Lanosterol + 10.5 NADPH + 4.5 NADH + 10 02 --> Chol + 3 002
Chol + FA-CoA -> Cholesterol Ester

Sphingolipld Metabolism

65
66

Ser + 1 Palm-00A + 1 FA-CoA + NADPH -> Ceramide + 002 + FADH2
Ceramide + PhosphatidyICholine -> Sphingomyelin

Phosphollpid synthesis

77
78
79
80
81
82
83

Phosphatidate -> CDP-DAG

CDP—DAG + Ser -> PhosphatidyISerine
PhosphatidyISerine --> PhosphatidylEthanolamine + 002
CDP-DAG + G-3-P --> PhosphatidylGlycerol

2 PG -> Cardiolipin + Glycerol

DAG + GDP-Choline -> PhosphatidyICholine

DAG + GDP-Ethanolamine -> PhosphatidylEthanolamine

 

* Reactions 26 included only in FF A treatment models.

a Reaction 26a included only in model for control cells.

50

Table 3.2 List of intracellular metabolites

 

Metabolite

 

(DmNODOI-h-wN-I»

 

 

glucose-6-P
fructose-6-P

glyceraldehyde—S—P

3-PGA
DHAP

Phosphoenolpyruvate

pyruvate
002
acetyI-CoA
oxaloacetate
citrate
a-ketoglutarate
succinyI-CoA
fumarte
acetoacetate
acetoacetyI-CoA
B-OH

02
GcheroI-3-P
Palm-00A
NADH
NADPH
FADH2

Ser

Cys

NH4

Glu

Gln

Val

Ile

Leu

Arg

0m

Ala

Gly

Tyr

Thr

Lys

Phe

Pro

His

Asp

Urea
Citrulline
Ceramide
Sphingomyelin
HMG-CoA

 

51

 

48
49
50
51
52
53

55
56
57
58
59
60
61
62
63

65

 

Table 3.2 continued

lsopentenyI-PP
Geranyl-PP
FamesyI-PP
Squalene

Lanosterol
Cholesterol
Cholesterol Ester
Asn

Phosphatidate

DAG

CDP-DAG
PhosphatidyISerine
PhosphatidylEthanolamine
PhosphatidyIGlyceroI
Cardiolipin
CDP-Choline
Phosphatidleholine
GDP-Ethanolamine

 

 

52

Table 3.3 List of measured extracellular fluxes

 

 

Flux # Metabolite
1 Glucose
8 Lactate

19 AcAc
20 B-Ohbutyrate
21 02 In
24 Glycerol
27 TG
28 Ser
30 Gln
31 His
32 Asp
33 Glu
34 Gly
36 NH3
37 Arg
38 Thr
39 Ala
41 Pro
42 Cys
43 Tyr
45 Val
46 0m
47 Lys
48 lie
49 Leu
50 Phe

35 FFA (in)*

 

 

 

 

* Flux 85 measured only for the FF A treatments

3.2.4 Correlation analysis
Correlation analysis was conducted to identify the relationship between the calculated
metabolic fluxes and toxicity. Pearson’s correlation coefficient was calculated between

fluxes from the MFA and cytotoxicity.

53

3.2.5 Perturbation analyses

To determine the effect of an extracellular flux on the intracellular flux of interest,
perturbation studies were conducted in silico. The values of a particular extracellular flux
were changed while keeping all other fluxes constant and the resulting intracellular fluxes

were calculated by MFA.

3.2.6 Western Blotting

The expression of the heavy subunit of glutamyl-cysteine synthase (GCS-H) and xCT (a
subunit of system Xc- which transports cystine into the cell) were measured using
western blotting. After desired treatments, the cells were washed twice with PBS and
scraped off in 200 uL CellLytic M buffer (Sigma). The cells were sonicated in this buffer
and centrifuged at 10,000 g for 5 minutes at 4 °C. The supematants were mixed with
sample loading buffer (5X from Cell Signaling Technologies) containing the reducing
agent and heated at 95 °C for 10 min, followed by cooling on ice for 2 minutes and short
centrifugation to collect the condensate. Equal amounts of protein (150 ug total) were
loaded on a 10% Tris-HCl gel (Bio-Rad), as was 10 ul of the prestained protein ladder
(Bio-Rad) and electrophoresed at 110 V for 95 min. Samples were then blotted onto
nitrocellulose membrane at 110 V for 60 min, followed by blocking of the membrane in
5% fat-free milk solution in Tris buffered saline containing 0.1% Tween (TBS-T) for 1 h
at RT. Samples were then incubated overnight in presence of the primary antibody for
GCS-H or xCT (both from Abcam Inc. and used at a dilution of 1:1000) in the blocking
buffer. After the incubations, the membranes were washed three times for 5 minutes each

with TBS-T, followed by exposure to secondary antibody (1:2000 dilution in TBS-T of

54

goat anti-rat IgG from Pierce Biotechnology) for 1 h at RT. Blots were then washed
thrice for 5 minutes each with TBS-T and developed using Supersignal West Femto
substrate (Pierce Biotechnology) and imaged using a Bio-Rad imager. The
chemilunriniscence was measured for 10 minutes. Following the imaging, the blots were
washed thrice with TBS-T and exposed to anti-beta-actin antibody (1:2000 in TBS-T)
overnight, then washed thrice and exposed to anti-mouse IgG for 1h in TBS—T (1:5000
dilution). Blots were washed again and imaged by measuring the chemiluminiscence for

l min.

3.3 Results

3.3.1 Cytotoxicity of F FAs

In vivo measurements of hepatotoxicity require measurements of activities of liver-
specific enzymes, such as aspartate transaminase (AST) and alanine transaminase (ALT)
in the serum. Many times, a ratio of the activities of ALT/AST is also calculated.
Hepatotoxicity is usually associated ‘with serum AST/ALT ratio of greater than 1.
However, patients suffering from NASH may have AST/ALT ratio of less than 1 (Kojima
et al., 2005). In vitro, with a single cell type, the cytotoxicity may be more easily and
equally measured by calculating the % lactate dehydrogenase (LDH) activity released, as
has been performed in many previous studies (Li et a1, 2005 and Wang et al., 2002).
Therefore, in the current study, the cytotoxicity of the FF As was measured as % of lactate
dehydrogenase (LDH) released. The saturated FFA, palmitate was found to be toxic and
its toxicity increased significantly on the second day of exposure and remained elevated

on the third day, while the unsaturated FFA was not toxic (Figure 3.1a). In addition,

55

exposure to palmitate led to a significant increase in caspase-3 activity (Figure 3.1b).
Because the cytotoxicity of palmitate did not change much from day 2 to day 3, most of
the experimental verifications were conducted until day 2. For some of the treatments,
experiments were conducted until day 3 and it was found that the treatments effective on

day 2 were almost equally effective on day 3.

3.3.2 Accumulation of TO

The intracellular accumulation of TG by the cells was measured. The cells exposed to
FFAS accumulated significant amount of TG, which were visible under the microscope as
intracellular lipid droplets (not shown). Accumulation of lipids inside the liver cells is
called steatosis and is considered the first step in the progression to NASH. The
intracellular TG levels were quantified by enzymatic assay. Cells exposed to unsaturated

FFAs accumulated greater amount of TG than saturated FFAS (Figure 3.2).

56

45

 

I
IDay2‘

I I
35 ,ElDay3

in ’1
4o - I 3’
30 -
25 ~
20

15

LDH released (%)

 

 

 

_L -. ._I_l
BSA Palm Ole

 

 

 

Figure 3.1a Cytotoxicity of various treatments. Confluent HepG2 cells were treated with 0.7 mM
palmitate or oleate complexed to 4% BSA. Control cells were treated with just 4% BSA. Media were
changed every 24b. The cytotoxicity was measured as % LDH released after each 24h exposure, as
explained in the materials and methods. Data presented as average +/- s.d. of three independent

experiments. *, significantly higher than control, p<0.01.

 

Relative Caspase Activity

 

 

 

BSA Ole Palm

Figure 3.1b Activation of caspase—3 by palmitate. Confluent HepGZ cells were treated with 0.7 mM
palmitate or oleate complexed to 4% BSA. Control cells were treated with just 4% BSA. Media were
changed every 24h. Caspase-3 activity was measured after every 24 h of exposure. Data presented as

average +/- s.d. of three independent experiments. *, significantly higher than control, p<0.01.

57

 

P
o
a:

 

 

 

 

 

A s
3 7.6a 1' " * *
'o g Y . *
E _ lDay2‘
'53 0.05 p Dgya‘ *
e *
°- *
en 0.04 -
E
E
a 0.03 -
x
3
2 0.02 ~
(0
o
c
'5 oo1 -
> .
In
2

o _

BSA Palm Ole

Figure 3.2 Measured flux through the synthesis of triglycerides in response to FFAS. Confluent HepG2
cells were treated with 0.7 mM palmitate or oleate complexed to 4% BSA. Control cells were treated with
just 4% BSA. Media were changed every 24h. The cells were washed thrice with PBS after the desired time
of treatment and lyzed with 1% triton-X400 in PBS. The fluxes of triglyceride synthesis were measured

for each day. Data presented as mean +/- s.d. of n=9 from 3 independent experiments. *, significantly
higher than control, p<0.01. #, significantly higher than palmitate treatment, p<0.01.

3.3.3 Verification of the pseudo-steady state and linearity of fluxes over 24 h

To verify that the pseudo-steady state (PSS) assumption is valid, the intracellular and
extracellular concentrations of lactate were measured after 4, 8, 12 and 24h of exposure
to different media. Lactate was chosen because it was the metabolite with the highest
flux, and therefore, offered two advantages: (a) because of its high synthesis, it is likely
to be the most concentrated metabolite, making it easier to detect with higher accuracy,
and (b) because the changes in the levels of lactate were the highest of any metabolite,
lactate may provide the most stringent test of the PSS assumption. As shown in Figures
3.3a, the intracellular concentrations of lactate increased with time in response to FFA

exposure. Among the two FFAS, palmitate led to higher intracellular levels of lactate. The

58

extracellular levels of lactate also increased with time, and were not significantly
different across the treatments (Figure 3.3b). The extracellular release was approximately
linear over the 24h period across different treatments. Similar intra- and extra- cellular
profiles were observed on day 2 (not shown). The ratio of changes in the intracellular to
extracellular lactate levels are shown in Table 3.2. Since the intracellular changes were
about a thousand fold smaller than the extracellular change, it indicates that the

accumulation of lactate is negligible and therefore, the PSS hypothesis is valid.

 

 

 

 

25
—l- BSA
+ Palm
20 - +Ole

 

J
m
I

A
O
I

L
v

 

 

 

I
I

Intracellular Lactate (uglmg protein)

 

 

 

O
-I

Time (h)

Figure 3.3a Profile of intracellular lactate levels in response to various treatments. The cells were
treated with 0.7 mM palmitate and oleate complexed to FFAs for various times as indicated, following
which the media were collected, cells were trypsinized, washed with thrice with PBS and lyzed in
perchloric acid. The lactate levels in the cell-extracts were measured by enzymatic assays. Data presented
as mean +/- s.d. of n=9 from 3 independent experiments.

59

6000 *

 

 

 

 

 

§ “373.9%
g 5000 ~ .
a I
2’ 4000 ~
BI
3
1: 3000 -
O
a
8
E 2000 .
3
g 1000 —
3
0 . .
4 e 13 24
Time (h)

Figure 3.3b Profile of extracellular lactate levels in response to various treatments. The cells were
treated with 0.7 mM palmitate and oleate complexed to FFAS for various times as indicated, following
which the media were collected. The lactate levels in the media were measured by enzymatic assays. Data
presented as mean +/- s.d. of n=9 from 3 independent experiments.

Table 3.4 Ratio of change in the intracellular to extracellular lactate levels. The
maximum changes in the levels of intra and extra-cellular fluxes for lactate were
calculated and their ratios calculated.

 

 

 

Intra Extra Ratio (Intra/Extra)
[BSA 0.88 4603.2 1.91 e-4
Ealm 7.24 4869.2 1.49 e-3
le 3.60 4333.8 8.31 e-4

 

 

3.3.4 MFA analysis

The results of the MFA analysis for days 1, 2 and 3 are displayed in Figures 3.4, 3.5 and
3.6, respectively (Figures available at the end of the chapter). The figures show the major
fluxes, for a complete list of the fluxes, see appendix 1-3. Exposure to FFAS led to an
increase in intracellular TG accumulation (measured flux) and therefore, the reactions

leading to TG formation, such as phosphatidate and diacylglycerol synthesis, were also

60

increased. Exposure to palmitate led to reduced uptake of glycerol and increased release
of ketone bodies as compared with oleate or BSA. No changes in glucose metabolism
were observed in response to the various treatments. Cells exposed to palmitate also had
altered amino acid metabolism. They had, in general, reduced His, Gly, Arg, Cys, Tyr
and Val uptake, but had greater uptake of Ser, Thr, Ile, Leu and Phe than controls (BSA-
treated cells). Among these, Phe, Val, Tyr, Thr, Ile, His, Arg and Len are essential amino
acids while Gly and Cys are required for glutathione synthesis. Palmitate-treated cells
also secreted less Ala as compared to controls. Most of these differences in amino acid
uptake/ release were also observed in palmitate-treated cells when compared to oleate-
treated cells. However, both the palmitate and oleate exposure reduced Arg uptake as
compared to the controls. Some of the fluxes altered by exposure to palmitate were
chosen for further analysis, based on their strong correlation to the toxicity or known
involvement in toxicity, as discussed below. The results of the experiments to evaluate

the roles played by these metabolites in the palmitate cytotoxicity are also presented.

3.3.4.1 Correlation analysis

In order to gain insights into the fluxes which might contribute to the cytotoxicity,
correlation analysis of the calculated fluxes to cytotoxicity was conducted. Table 3.5 lists
the results of the correlation analysis in decreasing order of the correlation coefficients. It
was observed that reactions involved in ketone body production (fluxes 18-20) had the
highest positive correlation coefficient to the cytotoxicity. Similarly, de novo ceramide
synthesis also had high positive correlation coefficient to toxicity. These results

suggested a relation between cytotoxicity and these fluxes, which was experimentally

61

investigated (see section below). Ala release (flux 39) as well as serine uptake (flux 28)

also had high positive correlation coefficients.

Uptake of cysteine (flux 42) had the greatest negative coefficient, followed by uptake and
metabolism of lysine and valine. These results suggested that supplementation of cysteine
could reduce cytotoxicity, which was experimentally validated (see section below).
Lysine and valine are essential fatty acids. Exposure to palmitate reduced the uptake of
these. Essential fatty acids have been shown to affect cell growth (Gardner and Lane,
1993). Certain studies have also shown protective effects against cell death by

supplementing with these amino acids (Hong-Ping and Bao—Shan, 1999).

62

Table 3.5 Ranked list of Pearson’s correlation coefficient of the metabolic fluxes
with cytotoxicity

 

 

 

 

 

 

Corr.
Flux # Equation Coeff

18 Acetoacetyl-CoA -> Acetoacetate 0.98
20 Acetoacetate + NADH -> (B-OH butyrate) 0.97
19 Acac out 0.96
65 Ser + 2 Palm-00A + NADPH --> 002 +Ceramlde + FADH2 0.87
66 Ceramide + Phosphatidlehollne -> Sphingomyelin 0.82
39 Ala (In) 0.80
17 2 AcetyI-COA -> AcetoacetyI-CoA 0.80
38 Thr (In) 0.70
84 Glu + 2 NADPH -> Pro 0.65
28 Ser (In) 0.65
56 Gln --> Glu + NH4 0.57
82 DAG + COP-Choline -> PhosphatidlehoIine 0.55
36 NH4 (In) 0.53
26 Palm-00A -> 8 AcetyI-CoA + 14 NADH 0.50
15 G-6-P -> 12 NADPH + 6 002 0.49
64 Phe + 02 -> Tyr 0.41
30 Gln In 0.39
59 Thr -> Pyr + 002 + NH4 + 2 NADH + FADH2 0.36
62 Ile + aKG —> Glu + Succ-CoA + Acetyl-CoA + NADH + FADH2 0.32
63 Leu + aKG -> Glu + HMG-CoA + NADH + FADH2 0.31
49 Leu (In) 0.28
50 Phe (In) 0.25
77 Phosphatidate -> CDP-DAG 0.25
52 H003- + NH4 + 0m + 2 ATP -> Citrulline 0.21
78 CDP-DAG + Ser -> PhosphatidylSerine 0.20
79 PhosphatidylSerine -> PhosphatidylEthanolamine + 002 0.19
53 Citrulline + Asp + ATP --> Fumarate + Arg 0.16
48 Ile (IN) 0.1 1
81 2 PG -> Cardiolipin + Glycerol 0.07
80 CDP-DAG + G-3-P -> PhosphatidylGcheroI 0.07

9 Pyruvate -> AcetyI-CoA + NADH +002 0.04
73 Lanosterol + 10.5 NADPH + 4.5 NADH + 10 02 --> Chol + 3 002 0.03
72 Squalene + 02 + NADPH —> Lanosterol 0.03
71 2 FamesyI-PP + 0.5 NADPH + 0.5 NADH --> Squalene 0.03
70 Geranyl-PP + IPP -> FamesyI-PP 0.03
68 HMG-CoA + 2 NADPH (+ 3ATP) -> IPP 0.03
69 2 IPP -> GeranyI-PP 0.03
75 2 Palm-00A + GcheroI-3-P -> Phosphatidate -0.02
74 Cholesterol Ester -> Chol + FA-CoA -0.04
67 Acetoacetyl-CoA + Acetyl-CoA -> HMG-CoA -0.05
33 Glu (In) -0.13
83 DAG + CDP-Ethanolamine -> PhosphatidylEthanolamine -0.15
27 Palm-00A + DAG --> TG -0.16
29 Ser --> NH3 + Pyr -0.18
76 Phosphatidate -—> DAG -0.19

63

 

 

 

 

 

Table 3.5 continued

22 NADH + 0.5 02 -> 2.5 ATP -0.22
16 002 out 024
44 Tyr + aKG + 2 02 --> Glu + 002 + Acetoacetate + Fumarate -0.28
41 Pro In -0.36
21 02 (In) -0.40
54 Arg --> 0m + Urea -0.45
55 Urea Out 045

1 G+ATP -> G-6-P -0.49

8 Pyruvate + NADH -> (Lactate) -0.50
58 Asp + NH4 -> Asn -0.50
25 Gcherol-3-P -> GcheraIdehyde-3—P + NADH -0.50
57 0m + a-KG + 0.5 NADPH + 0.5 NADH -> Pro -0.52
34 Gly (In) -0.54
37 Arg In -0.54
51 Glu + Cys + Gly -> GSH 0.54
11 Citrate --> a-KetoGlutarate + NADPH +002 -0.55
12 a-Ketoglutarate -> SuccinyI—CoA + NADH + 002 -0.55
35 Glycine --> 2 002 + NADH + NH3 -0.56
10 AcetyI-CoA + 0AA -> Citrate -0.57
46 0m IN 057

4 DHAP ~> Glyceraldehyde-B-P -0.57

3 F-6-P +ATP -> Gcheraldehyde-B-P + DHAP -0.58

2 Go6-P -> F-6-P -0.58
14 Fumarate -> 0AA + NADH -0.58
13 SuccinyI-CoA -> Fumarate + FADH2 + ATP -0.60

7 PEP -> Pyr -0.61

6 3PGA -> PEP + NADH (+ 2ATP) -0.61

5 Gcheraldehyde-3-P --> 3—PGA -0.61
23 FADH2 + 0.5 02 -> 2 ATP -0.64
43 Tyr (In) -0.66
32 Asp (In) —0.68

Lys + 2 aKG + NADPH -> ZGIu + AcetoacetyI-CoA + 2002 + 4 NADH +
61 FADH2 -0.70
24 Glycerol + ATP ---> Gcherol-3-P -0.73
31 His -> Glu + NH4 -0.75
60 Val + aKG —> Glu + 002 + 2NADH + FADH2 + Succ-CoA -0.81
40 Glu + Pyr --> Ala + aKG -0.83
45 Val (In) -0.86
47 Lys (In) -0.89
42 Cys In -0.91
3.3.4.2 Fatty acid oxidation

Exposure to palmitate led to a greater release of ketone bodies than the controls or oleate

(Figure 3.73). The profile of beta-oxidation resembled that of the cytotoxicity, suggesting

64

a relationship between the two. In fact, the fluxes of ketone body production had the
highest Pearson’s correlation coefficient to the cytotoxicity (see Table 3.5). In order to
test a causative role of increased oxidation on cytotoxicity, we tested the effect of
inhibitors of beta-oxidation on cytotoxicity. Etomoxir (Etrnx) and 2-mercaptoacetate (2-
MA) were employed to inhibit mitochondrial oxidation. Etomoxir has been shown to
completely inhibit beta-oxidation of palmitate in HepG2 cells (Minnich, 2001). Multiple
concentrations of these inhibitors (20, 50 and 100 uM were tested for their ability to
reduce the cytotoxicity. However, inhibiting FFA oxidation did not reduce the
cytotoxicity (Figure 3.7b). This indicated that the increased beta-oxidation of palmitate,

despite its strong correlation, was not responsible for the cytotoxicity of the F FA.

 

 

 

 

2.5
lDay1
* * IDay 2
2,0 - 1' D Day 3

 

1.5 .

 

 

1.0 -

Total Ketone Body Released
(u molel mg protein] day)

0.5 -

 

 

 

 

 

 

 

j

BSA IPaIm Ole

0.0 -

Figure 3.7a Measured flux of release of total ketone bodies (acetoacetate + beta-hydroxybutyrate) by
the cells. Confluent HepG2 cells were treated with 0.7 mM palmitate or oleate complexed to 4% BSA.
Media were replaced daily. The fluxes of acetoacetate and beta-hydroxybutyrate release were measured for

each day. Data presented as mean +/- s.d. of n=9 from 3 independent experiments. *, significantly higher
than control, p<0.01.

65

 

45

I

40 . liPaIrriitateii
IEth I
IIIZ-MA J

0 O)
0 0|
1

  

l

l

..L
0'! O U!
l_

 

LDH released (%)
N N

 

 

 

 

 

1 2 3
Days

Figure 3.7b Effect of inhibition of fatty acid beta-oxidation on the toxicity of saturated FFA. Cells
were treated with 0.7 mM palrrritate in the presence of 100 uM etomoxir (Etrnx) or 100 uM 2-
mercaptoacetate (2—MA) for 3 days. Media were changed daily and the cytotoxicity was measured as LDH
release. Data presented as mean +/- sd of n=9 from three independent experiments.

3.3.4.3 De nova ceramide synthesis

Because de novo ceramide synthesis has been suggested as an important player in the
cytotoxicity of saturated FFAS, the fluxes of de novo ceramide synthesis were calculated
in the various treatments. Cells treated with saturated FFA had increased flux through de
novo ceramide synthesis (Figure 3.8 a), as well as higher intracellular levels of ceramide
(Figure 3.8 b). To identify the limiting metabolite in ceranride synthesis, perturbation
analysis of serine and palmitate levels (the two metabolites required for de novo synthesis
of ceramide) on ceramide synthesis was performed. This analysis revealed that ceramide
synthesis in palmitate-treated cells was more responsive to changes in the uptake of
serine than palmitate (Figure 3.8 c). This suggests serine as a potential regulator of
ceramide synthesis in palmitate-treated cells. Exposure to palmitate also led to higher

uptake of serine. Thus, it is likely that the increased synthesis of de novo ceramide is due

66

to increased availability of both substrates in palmitate-treated cells. Correlation analysis
identified that de novo ceramide synthesis also had high positive correlation to the
cytotoxicity (Table 3.5). In order to test the role of increased de novo ceramide synthesis
on the cytotoxicity, we tested the effect of an inhibitor of de novo ceramide synthesis,
firmonisin Bl (FBI), on the cytotoxicity. The inhibitor reduced the accumulation of
ceramide significantly (Figure 3.8 d), but did not reduce the cytotoxicity of the FFA
(Figure 3.8 e). These results suggested that de novo ceramide synthesis does not play a

significant role in the palmitate-induced toxicity of HepG2 cells.

0.14

 

 

IDay1

0.12 - I Day 2
CI Day 3

 

 

 

0.1 -

I

0.06 -

 

0.04 -

0.02 -

Flux of de novo ceramide synthesis

 

 

 

 

 

 

 

 

BSA Palm Ole

Figure 3.8a Calculated flux through ceramide synthesis. The flux through de novo ceramide synthesis as
predicted by MFA is presented.

67

90

 

N
O
l

50-

40 -

30-

Intracellular ceramide levels
(p moll mg protein)
N
O

'r

 

 

 

 

BSA Palm

Figure 3.8b Intracellular C-16 ceramide levels in response to palmitate. The cells were treated with 0.7
mM palmitate for 48 h. Intracellular ceramide levels were measured using LC/MS. Data presented as mean
+/- s.d. of n=6 from two independent experiments.

 

25
20 e
15

10 -

Change In De Novo Ceramide Synthesis

 

 

 

 

Change in metabolite uptake

Figure 3.8c Results of in silica perturbation studies for the effect of serine or palmitate on the flux of
de nova ceramide synthesis. The fluxes of uptake of palmitate or serine were altered while keeping all the
other measurements same and the changes in the fluxes of de novo ceramide synthesis were calculated.
Data presented as relative change in the flux of de novo ceramide synthesis with respect to relative change
in the flux of metabolite.

68

 

N 03 # g 65 N O O
O O O O O O O
l l I l l

046 ceramide level (p moll mg protein)
3

 

   

 

o _
Palm Palm+FB1
Figure 3.8d Effect of inhibition of ceramide synthesis on the levels of C-16 ceramide in the cells. Cells

were cultured for 48 h in 0.7 mM palmitate in the presence or absence of 20 uM Fumonisin Bl (FBI). The
intracellular lipids were extracted and the levels of C-16 ceramide measured by LC-MS. Data presented as

mean +/- s.d. of n = 6 from 2 independent experiments. *, significantly smaller than palmitate.

 

LDH released (%)

 

 

   

Palm FB1

Figure 3.8c Effect of inhibition of ceramide synthesis on the cytotoxicity of palmitate. Cells were
treated with palmitate in presence of 20 uM FBI and the cytotoxicity after 48 h of treatment was measured.
Data presented as mean +/- 541 of n=9 from 3 independent experiments.

3.3.4.4 Glutathione synthesis

Oxidative stress is considered an important factor in the development of NASH. We have
previously identified an important role of oxidative stress in the toxicity of palmitate
(Srivastava and Chan, 2007). GSH is the primary anti-oxidant molecule of the cell and
protects cells against oxidative insults. Among the different FFAS, palmitate treatment
was consistently associated with reduced flux through GSH synthesis, while oleate had
higher flux (Figure 3.9a). In order to test the predicted reduction in the GSH synthesis,
the total GSH (GSH + GSSG) levels in the cells were measured. The total GSH (tGSH)
levels were measured because tGSH levels are resistant to oxidative stress (Glosli et al.,
2002) and thus, will not be affected by the oxidative stress caused by palmitate. Exposure
to palmitate was associated with reduced levels of tGSH, while exposure to oleate had
increased tGSH levels (Figure 3.9b). A reduction in the synthesis of GSH would render
the cells more susceptible to oxidative stress. To understand the mechanism of reduction
in GSH synthesis, in silico perturbation studies of the constituent amino acids in GSH
synthesis was performed in the palmitate-treated cells. The perturbation study identified
that the GSH synthesis was limited primarily by the supply of cysteine. It was also
observed that exposure to palmitate led to reduced cysteine uptake (Figures 3.5-3.7),
suggesting it may the limiting amino acid. To further identify what limits the GSH
synthesis, the expressions of the rate-limiting enzyme of GSH synthesis and the cystine
transporter were measured. Glutamyl—cysteine synthase (GCS) is the first and the rate-
limiting enzyme of GSH synthesis and catalyzes the synthesis of glutamate and cysteine.
Another important regulator of GSH synthesis is the availability of cysteine, the primary

source of which is cystine in the medium. Cystine is taken up through a transporter, the

70

system Xc— (Anderson and Meister, 1987), which is composed of two proteins- xCT and
4F2hc/CD98. Western blot analyses revealed that the expression of GCS was increased
(Figure 3.9d), but that of xCT was decreased (Figure 3%) significantly in palmitate-
treated cells. Both these proteins were altered in a time-dependent manner, i.e., the
change was greater on day 2 than on day 1. These results identified that the reduced
cystine uptake in the palmitate-treated cells was due to reduced levels of the transporter
and not of GCS. These studies, in conjunction with the perturbation and correlation
studies, indicated that increasing cysteine supply would increase GSH synthesis and thus,
may reduce cytotoxicity, while supplementing glycine would not significantly affect
GSH levels or cytotoxicity. To test these predictions, we measured the cytotoxicity of
palmitate in the presence of N-Acetyl L-cysteine (NAC) and glycine. NAC
supplementation has been shown to increase the cellular GSH levels in HepG2 cells
(Gruetter et al., 1985). Uptake of NAC is independent of the levels of system Xc- (Sato,
2005). The supplementation of NAC significantly reduced the cytotoxicity of the FFA
(Figure 3.9f), while supplementation of glycine had no effect (not shown). Thus, the

predictions of MFA and perturbation studies were experimentally verified.

71

6
en

 

i 05y?
I Day 2
Elpay 3

I

Figure 3.9a Calculated relative fluxes of GSH synthesis, as predicted by MFA.

A
:h

 

..s
N

—I
O
l

P
or

Relative GSH synthesis flux flux
9 o
a- on

P
N
I

 

 

 

 

 

P
o
l

 

 

 

I

Palm Ole

 

Hay?
1.4 « lDavZI *
DDay 3i *

0.8 -
0.6 .

0.4 4

Relative Total GSH levels

0.2 4

 

 

 

 

 

 

o 2

BSA Palm Ole
Figure 3.9b Measured relative levels of total intracellular GSH. Cells were treated with palmitate or
oleate for 3 days. At the end of each day, cells were harvested and the total GSH (tGSH) in the cells were

measured colorimetrically. The data was corrected to the cellular protein levels, measured using BCA
assay. Data presented as mean +/- s.d. of three independent experiments.

72

 

.3
0'!

A
h
l

—b
N
I

.3
O

 

 

Change In GSH synthesis flux
0
en

 

 

 

 

 

 

0.6 -
0.4 —
. 1' .6”
0.2 ~ —>K— Glu
0.0 , . I .
-0.5 0.0 0.5 1.0 1.5 2.0

Change in Amino Acid Flux

Figure 3.9c Results of in silico perturbation studies on the effect of increasing uptake of amino acids
on GSH synthesis. The fluxes of Glutamate, Glycine or Cysteine were varied and the effect on the flux of
GSH synthesis was calculated by the MFA model.

73

 

DAY 1 DAY 2

 

BSA Palm Ole BSA Palm Ole

 

 

 

 

 

(GCS)

 

 

 

 

 

(Beta-Actin)

 

Quantification of western blot intensities

 

* E our]

Relative GCS levels

 

 

 

BSA Palm Ole

Figure 3.9d Effect of the treatments on the expression of gamma-glutamyl cysteine synthetase (GCS)
heavy subunit. Cells were exposed to BSA, palmitate or oleate for 24 and 48 h, following which cells were
lyzed and the expression of GCS were measured using Western Blot. The details of the Western Blot
analyses are provided in the methods section. Beta-actin was employed as the loading control. The
intensities of different blots were analyzed using available software. The relative intensities of three
different blots are presented as mean +/- s.d.

74

 

DAY 1

DAY 2

 

BSA Palm Ole

 

 

BSA Palm

Ole

 

 

 

 

 

 

Quantification of western blot intensities

1.2

 

12

0.8 ~

0.6

0.4 <

Relative xCT level

0.2 ~

 

o _
BSA

Figure 3.9c Effect of the treatments on the expression of xCT (a subunit of cystine transporter). Cells
were exposed to BSA, palmitate or oleate for 24 and 48 h, following which cells were lyzed and the
expression of xCT were measured using Western Blot. The details of the Western Blot analyses are
provided in the methods section. Beta-actin was employed as the loading control. The intensities of
different blots were analyzed using available sofiware. The relative intensities of three different blots are

presented as mean +/- s.d.

IDay1
IDay2‘

Palm

75

 

Ole

 

 

a

 

LDH released (%)
3 at 8 8 8 8 8

0|
1

 

 

   

o _
Palm NAC

Figure 3.9f Effect of NAC supplementation on the cytotoxicity of palmitate. The cells were treated with
0.7 mM palmitate in the presence or absence of 1 mM NAC for two days and the LDH released after 48h of

treatments were measured Data presented as mean +/- s.d. of n=9 from 3 independent experiments. *,
significantly higher than control, p<0.01.

3.3.4.5 Effect of NAC supplementation on the levels of GCS

Many previous studies have suggested that oxidative stress or cysteine limitation
increases the levels of GCS (Lu, 1999), and our study shows that exposure to palmitate
produces both oxidative stress and cysteine limitation. Therefore, it is likely that the
increase in GCS levels in palmitate-treated cells was due to reduced cysteine availability
and/ or oxidative stress. NAC supplementation increases the intracellular cysteine
(Anderson, 1987) and GSH levels (Gruetter et al., 1985) and reduces oxidative stress
(Quanungo et al., 2004). Acute (24 h) NAC supplementation increased the expression of
GCS in both palmitate and control media (Figures 3.10 a and b), but a synergistic
increase in GCS levels was observed in the palmitate—treated cells (Figure 3.10a).

However, chronic exposure to NAC reduced the levels of GCS (day 2) in the palmitate-

76

containing medium, while no significant changes were observed in the control medium.
This suggests that chronic exposure to palmitate increases GCS levels by reducing
cysteine levels and! or by causing oxidative stress. The increase in GCS levels by NAC
supplementation in the control medium are in agreement with that of Lee et al. (2006). In
addition, we also identify an increase in the levels of GCS in the presence of palmitate

and NAC.

77

 

 

DAY 1 DAY 2

 

NAC
(mM) 0 1 5 10 0 1 5 10

 

 

 

(Beta-actin)

Quantification of western blot intensities

 

3.5

3.0 —

2.5 4

Relative GCS levels
5‘. 2.

d
O
I

9
0|
I

 

 

 

Palm NAC- 1mM NAC- 5mM NAC- 10mM

Figure 3.10s Effect of NAC supplementation on GCS levels in palmitate-treated cells. Cells were
treated with palmitate in the presence of increasing amounts of NAC (0, l, 5 and 10 mM) for 48 h in the
presence of 0.7 mM palmitate. Media were changed every 24 h. At the end of desired exposure times, the
expression of GCS was measured using western blots.

78

 

DAY 1 DAY 2

 

NAC
(mM) 0 1 5 10 0 1 5 10

 

 

 

 

l” “‘“-Actln)

 

 

Quantification of western blot intensities

.3371 *
‘ I Day 2

1.4 ~

1.8

 

1.

1.2 .

1.0V

0.8 -

0.6 -

Relative GCS expression

0.4 -

0.2

 

 

 

0.0 5

H62 NAC- 1mM NAC- 5mM NAC- 10mM

Figure 3.1% Effect of NAC supplementation on GCS levels in control medium. Cells were treated with
increasing amounts of NAC (0, l, 5 and 10 mM) for 48 h in control medium. Media were changed every 24
h. At the end of desired exposure times, the expression of GCS was measured using western blots.

79

3.4 Discussion

While it is known that altered metabolism and generation of ROS play important roles in
the development of hepatic lipotoxicity, the underlying global metabolic responses are
not clearly known. MFA of human hepatoma cells in response to various FFAS identified
changes in the intracellular fluxes in response to the different FFAS and helped to
differentiate between the potential mechanisms involved in the cytotoxicity of the
saturated F FAs as compared to the unsaturated FF As. Additionally, MFA also identified
these external fluxes which regulate the internal fluxes of interest, thus suggesting

possible supplementation methods to combat lipotoxicity.

We have previously identified that exposure to palmitate causes significant oxidative
stress and this oxidative stress plays an important role in the toxicity of this fatty acid.
GSH is the primary anti-oxidant thiol. In this paper we find that palmitate also reduces
the synthesis of GSH by reducing cysteine uptake. The underlying mechanism was
identified to be reduced levels in the cysteine transporter xCT. These results indicate a
multi-pronged effect of palmitate on the cellular redox homeostasis which leads to the
cytotoxicity of this fatty acid. The underlying mechanism in the reduction of xCT
expression by palmitate is unknown, but may provide important insights into the
oxidative stress caused by exposure to this FFA. It was also identified that the GSH
synthesis was not limited by the levels of GCS, the first and rate-limiting enzyme of GSH
synthesis. In fact, the expression of GCS was found to be increased in the palmitate-
treated cells, in a time-dependent fashion. However, the observation that NAC

supplementation did not reduce the acute increase in GCS expression in response to

80

palmitate suggests that the acute increase by palmitate was not mediated by oxidative
stress or cysteine limitation, but perhaps through the activation of certain transcription
factors involved in regulating GCS expression, e.g. Nrf2 (Wild, 1999). In fact, activation
of Nrf2 has been shown to protect HepG2 cells against the toxicity of CYP2E1 and
arachidonic acid (Gong and Cederbaum, 2006). The reduction in GCS levels in palmitate-
treated cells by chronic (48 h) treatment of NAC suggests that the chronic increase in
GCS by palmitate (which is much greater than the acute increase in response to
palmitate) was, indeed, mediated through a reduction in cysteine supply and / or

oxidative stress.

Another important result is the lack of involvement of de novo ceramide synthesis in the
toxicity of palmitate. While the de novo ceramide synthesis and ceramide accumulation
were increased in the palmitate-treated cells, inhibiting this pathway did not reduce the
FF A toxicity. In fact, studies by other groups have also indicated that saturated FFA-
toxicity to hepatocytes is independent of de novo ceramide synthesis (Wei et al., 2006
and Feldstein et al., 2006). In our study, the ceramide accumulation was in picomoles/mg
protein, while the flux of synthesis and metabolism of ceramide were in nanomoles/ mg
protein. This indicates that these cells in particular, and probably hepatocytes in general,
can effectively metabolize ceramide so that the accumulations do not reach cytotoxic

levels.

Our results demonstrate the applicability of MFA in identifying the altered metabolic

processes under simulated conditions of metabolic disorders. The results show

81

differential metabolism of the saturated vs. unsaturated FFAS, which results in the toxic
effect of the former and suggest potential treatments to reduce the FFA toxicity. MFA
identifies more than one cellular process to be altered. This information help narrow
down targets for further study, as was illustrated here. Though the treatments did not
reduce the cytotoxicity to basal levels, it is likely that there is some non-specific
cytotoxicity of the FFA due to its hydrophobic effects. Among the different FFAS,
palmitate has the highest hydrophobicity. Therefore, palmitate may cause the disruption

of the plasma membrane (Oliver and Opie, 1994) leading to the cytotoxicity.

The MFA study indicated the complex metabolic alterations induced by FF As were
caused in part by changes in the levels of proteins (e.g. xCT), leading to altered
physiological response (cytotoxicity). Exposure to F FAs would also alter various other
cellular processes which may not cause a significant change in metabolism, but still
initiate cell death, for example, the activities of certain signaling pathways may be
regulated by palmitoylation (Resh, 2006) as well as affect the transcription of genes
involved in multiple cellular processes (Swagell et al., 2005). Though, these processes
were not investigated in this study, a more complete understanding of the development of
lipotoxicity requires generation and integration of such multi-source information.
Application of novel high throughput techniques coupled with advanced modeling would
help in achieving this goal. Such analysis would provide important insights into the
underlying changes, as well as novel information into the inter-relationships among the
changes, thus providing more effective alternatives to regulate the deleterious effects of

F FAs.

82

Figure 3.4 Results of MFA analysis, depicting major intracellular fluxes on day 1

glucose 9'Y°°'°' Q75 Sphingomyelin 0 00 Phospholipids
23.40 047 '
o

   
 
  

0.01

 

0.06
0.01
-O. 18

-0.36
0.00
0.06 Phosphatidate

2g -o.1o
0.01

0.03 -0. 1O

  

O
a 0°C
99.0
832! sea 3;;
2

0-79 0.28
0.17
TCA 901
9.29
Urea 3.52 Cycle 828 Iii)
8.70 848

-0.16 0.39

 

Regular font (1st flux) = BSA
Bold Font (2nd flux) = Palm
Italicized Font (3rd flux) = Ole

 

 

 

 

83

Figure 3.5 Results of MFA analysis, depicting major intracellular fluxes on day 2

glucose 071 Sphingomyelin Phospholipids
27.94 0.44 0'0

2
0.04

    
 
   

27.00 0.07
27.19 0.. ‘0-02 0.17
lactate 0.37 . 0-3’ CDP-
Ceramide DAG
51.20 53.39
50.11 5‘ 59
50.54 °

55.05

 
     
  
 
 

2.12
2.36 0.77
1.84 1.16

1.27
Cholesterol 1

0.04
0.04
0. 03

\

TCA 1024
10.56 .
m 7.53 Cycle 7.50
1 8.56 8.28
0-26 Succ-CoA
5.0. -
om w
-0.05 0.25
0.03 0.01 0.29

3:33 Urea Cycle m

   

 

Regular font (1st flux) = BSA
Bold Font (2nd flux) = Palm
Italicized Font (3rd flux) = Ole

 

 

 

 

84

 

Figure 3.6 Results of MFA analysis, depicting major intracellular fluxes on day 3

glycerol 0.60 Sphingomyelin Phospholiplds
0.01
27.4 0.49
3 0.04

26.73 0.63 0.07
0.79 0.17

0.5 G cerol-3-P
- 0.66 'y 0'31 33‘;

      
  
 
  

 

 

Phosphatidate

-0.07
-0.07
0.03 0.42

  
 
    
   
      
  

Cholesterol

0.03
0.03
0.08

m

  

TCA

9.00 8.85

3.37 Cycle 3.56

9.84 9.60
0.18 w
0.03 -

-0.02

 

Regular font (1st flux) = BSA
Bold Font (2nd flux) = Palm
Italicized Font (3rd flux) = Ole

 

 

 

85

CHAPTER 4. IDENTIFICATION OF GENES THAT
REGULATE MULTIPLE CELLULAR PROCESSES]
RESPONSES IN THE CONTEXT OF LIPOTOXICITY TO
HEPATOMA CELLS

4.1 Introduction

Many human diseases result from alterations in multiple cellular processes.
Therefore, targeting an individual process or response may not be sufficient to combat the
progression of such diseases. For efficient treatment of such diseases, it is imperative to
identify those cellular targets which control multiple functions of interest. For example,
prolonged exposure of cells to elevated levels of FFAS can alter multiple cellular
processes such as reactive oxygen species (ROS) production (Srivastava and Chan, 2007
and Listenberger et al., 2001), mitochondrial (J i et al. 2005), endoplasmic reticulum (Wei
et al., 2006 and Wang et al., 2006) and lysosomal dysfunction (Feldstein et al., 2006) and
activation of a variety of signaling pathways such as protein kinase C (PKC) (N ewsholme
et al., 2007) and caspases (Fernanda et al., 2006). The resultant effect of these
disturbances is lipotoxicity, that is, cell death due to exposure to elevated lipid levels.

Many recent studies have identified that saturated FF As cause greater lipotoxicity
than unsaturated F FAs. It has been suggested that channeling the saturated fatty acids to
TG reduces their cytotoxicity (Listenberger et al, 2003). Therefore, genes that regulate
TG synthesis from saturated FFAS may be useful targets to combat the lipotoxic action of
saturated FAs. We hypothesized that those genes which affect lipid synthesis as well as
lipotoxicity may serve as better targets than those regulating TG synthesis alone. To test

this hypothesis, we studied the cytotoxicity and the accumulation of TG in human

86

hepatoma cells (HepGZ) in response to various types of FFAS and concentrations of
TNF-a. The global gene expressions were obtained by cDNA microarray analyses. A
modified genetic algorithm coupled partial least squares (GA/PLS) analysis (Li and
Chan, 2004) was applied to identify the genes that regulate toxicity and lipid
accumulation, which were selected as potential targets for further functional analysis and
experimental testing.

As a proof of concept, we appliedthe GA/PLS framework to identify the genes
associated with TG accumulation. Many genes with known roles in lipid metabolism (e.g.
stearoyl-CoA desaturase (SCD) and inositol polyphosphate phosphatase-like 1 (INPPL1))
were identified by GA/PLS analysis for T0 accumulation, attesting to the applicability of
this approach. Then, the dual-response GA/PLS analysis was applied to identify the genes
that regulate both TG accumulation and cytotoxicity. In addition to genes with known
roles in the regulation of lipid metabolism and cell physiology, this analysis identified
NADH dehydrogenases and MAPKS as important regulators of both lipid accumulation
and cell death. While NADH dehydrogenases and MAPKS have been shown to play roles
in cell death caused by other insults, this analysis suggested that they regulate not only
cell death in response to fatty acid exposure, but also affect lipid accumulation. The
predicted roles of NADH dehydrogenases and MAPKS were verified experimentally by
employing inhibitors specific to them. Inhibiting NADH dehydrogenase and JNK
significantly reduced the cytotoxicity and increased the intracellular TG accumulation,
while inhibiting ERK reduced the toxicity without affecting the lipid accumulation.

This study demonstrates the applicability of the dual-response GA/PLS analysis to

identify genes which affect diverse cellular processes, and incorporating more

87

information provides better targets to control cellular responses. It must be pointed out
that although the method was applied to identify genes involved in regulating two
responses in the current study, the method is capable of handling more than two

responses for multi-factorial diseases.

4.2 Materials and Methods

4.2.1 Cytotoxicity measurement

The cytotoxicity of the treatments was measured as the fraction of lactate dehydrogenase
(LDH) released into the medium as described in chapter 2. For the experiments with
inhibition of NADH dehydrogenases, the cells were pre-treated with 0.2-1.0 pM rotenone
for 30 min, followed by exposure to 0.7 mM palmitate for 24h without any inhibitor. The
LDH released after the 24h exposure was measured and normalized to the triton-lyzed

values, as described above.

For experiments with the MAPK-inhibitors, stock solutions of these inhibitors, were
prepared in DMSO and diluted to desired concentrations in palmitate medium. Cells were
treated with the palmitate medium containing the desired concentrations of the MAPK
inhibitors. Under no condition was the concentration of the resulting DMSO greater than

0.2% v/v. At these concentrations, DMSO does not affect the cytotoxicity of the FFA.

4.2.2 Measurement of intracellular TG levels
The cells were pretreated for 30 minutes with 0-1 0M rotenone, and then exposed to 0.7

mM palmitate or oleate. In the experiments with MAPK inhibitors, cells were treated

88

with the desired concentrations of the inhibitor dissolved in the FFA medium. After
exposure to the FF As, the cells were washed thrice with PBS and lyzed with 1% triton in
PBS. The lysates were centrifuged at 10000 g for 2 min, and the concentration of TG in
the supernatant was measured by an enzymatic assay kit from Stanbio, according to
manufacturer’s instructions. This method measures the amount of glycerol released by
the enzymatic hydrolysis of the TG. To correct for the free glycerol in the cells, the
concentration of free glycerol in the cells were measured by free glycerol assay from
Sigma, and subtracted from the glycerol values of the TG assay, to give actual TG values.
These values were normalized to total protein in the extract, measured with the

bicinchoninic acid (BCA) method (Pierce Chemicals).

4.2.3 Measurement of caspase—3 activity

Caspase-3 activity was measured using a commercially available colorimetric assay
(Biovision) according to manufacturer’s instructions. The cells were treated with the
control medium or 0.7 mM palmitate for 24 h, washed once with PBS and lyzed using the
available lysis buffer. The activity in the lysate was measured and normalized to the total

protein, measured by bicinchoninic acid (BCA) assay (Pierce Chemicals).

4.2.4 Microarray analyses

Cells were cultured in 10 cm tissue culture plates until confluence and then exposed to
different treatments for 24h. RNA was isolated with the Trizol reagent. The expression
profiles of about 19500 genes were obtained with the cDNA microarrays. The microarray

analyses were conducted at the Van Andel Institute, Grand Rapids, MI. The protocols are

89

available online at (http://microarray.vai.org/protocols/). There were two biological
replicates for each condition and each replicate was measured with the Cy3 and CyS dyes

(i.e. there were two technical replicates for each biological replicate).

4.2.5 RT-PCR

Total RNA was extracted from cells with an RNeasy mini kit (Qiagen) and depleted of
contaminating DNA with RNase-free DNase (Qiagen). Equal amounts of total RNA (1
pg) were reverse-transcribed using an iScript cDNA synthesis kit (Bio-RAD). The first-
strand cDNA was used as a template. The primers used for quantitative RT-PCR analyses
of human SCD (5'-TGAGAAACTGGTGATGTI‘CCA-3' and 5'-
CCACAGCATATCGCAAGAAA-3') and human GAPDH (5'-
AACTTTGGTATCGTGGAAGGA-3' and 5'-CAGTAGAGGCAGGGATGATGT-3')
were synthesized by Operon Biotechnologies, Inc.. RT-PCR was performed in ZS-ul
reactions using 1/10 of the cDNA obtained from the reverse transcription, 0.2 uM each
primer, 1 X SYBR green supermix from Bio-RAD, and an annealing temperature of 60
°C for 40 cycles. Each sample was assayed in three independent RT reactions and
triplicate reactions each and normalized to GAPDH expression. Negative controls
included the absence of enzyme in the RT reaction and absence of template during PCR.
The cycle threshold (CT) values corresponding to the PCR cycle number at which
fluorescence emission in real time reaches a threshold above the base-line emission were

determined using MyIQTM Real-Time PCR Detection System.

90

4.2.6 GA/PLS analysis

Metabolic functions are regulated in part by the enzymes catalyzing the reactions, which
in turn are determined partly by their gene expression levels. Therefore, we hypothesized
that the metabolic functions can be predicted from the expression level of a subset of
genes that are associated with the metabolic function. We approximated the relation
between the metabolic function and expression level of this subset of genes with a log-

linear model:

 

 

Met(treatment) _ I’ll (Gene(treatment)i C(i)
Met(control) i=1 Gene(control), (2)

where Met(treatment) and Met(control) are the metabolic fimction for the treated and

control cultures, respectively; Gene(treatment),- and Gene(control)i are the expression

level of gene i for the treated and control cultures, respectively.

Denoting Y as

Met(treatment)
2 Met(control) ’and

 

log

Xi as

Gene(treatment) .-

 

lo
g2( Gene(c0ntrol)i

The equation (2) is transformed to:

Y = Z C(i)X,.

(3)

91

Since DNA microarray data are typically measured with respect to a reference level, we
applied a log-linear model, which works well when the data are presented as relative
levels. Furthermore, a log-linear model allowed some of the nonlinear relationships
between metabolic function and gene expression to be captured. Log-linear models have
been applied to approximate nonlinear processes in biochemical systems [Ni 1996a, Ni

1996b]. In this study the coefficients C(i) in equation (3) were determined by PLS

analysis and the genes, Genei, were selected by GA/PLS as described in reference [Li and

Chan, 2004]. Briefly, multiple subsets of genes that provide a good prediction of the
metabolic functions with equation (3) were selected and the frequency of each gene that
appeared in these subsets was counted. Those genes with high frequencies were deemed
as important genes to the metabolic function. In equation (3) Y is a matrix of multiple
metabolites, therefore the genes that were important to multiple metabolites (Y) were
selected with the modified GA/PLS. Multiple metabolites in Y were assumed to be
independent of each other. The genes were selected so that the overall prediction
accuracy for all the metabolites, e. g. [Y1, Y2] was maximized, therefore, genes of modest
importance to Y1 and Y2 in single metabolite analysis might be identified to be of high
importance to [Y], Y2} in multiple metabolite analysis. Because the GA/PLS method
does not guarantee a global minimum, the method was run multiple times and the
frequency with which each gene was selected was counted. A total of 14 runs were
performed. The genes that were selected with a frequency of 8 or greater were selected as
important. The greater frequency of selection indicated a greater significance of that gene

in regulating the function.

92

4.3 Results

4.3.1 Lipid accumulation and cytotoxicity

Cells exposed to FFAS accumulated TG, which was quantified using an enzymatic assay.
The intracellular accumulation of TG was higher in the cells exposed to unsaturated
FFAs- oleate and linoleate, as compared to those exposed to saturated FFA- palmitate
(Figure 4.1). No significant effect of TNF-oc treatment on TG accumulation was observed

(not shown), in agreement with others (Bruce and Dyck, 2004).

 

Intracellular TG (ugl mg protein)
8

 

 

H62 BSA Palm Ole Lin

Figure 4.1 Intracellular accumulation of TC in response to various FFAs and TNF—a. The cells were
exposed to various FFA and TNF-a. for 24 h, after which they were washed with PBS and lyzed in 1%
triton-X-IOO in PBS. The levels of TG in the lysates were analyzed using an enzymatic assay and
normalized to the protein levels. Results presented as mean +/- s.d of n=9 fi'om 3 independent experiments.

*, significantly higher than control (p<0.01).'

The cytotoxicity was measured after 24h of treatment as the fraction of the total LDH

activity released. The saturated fatty acid, palmitate, was found to be toxic to these cells

93

and TNF-01 exacerbated this toxicity, while the unsaturated fatty acids with or without

TNF-01 co-supplementation did not have a toxic effect (Table 4.1).

 

 

Medium TNF (nglml) LDH Released (%)
Mean SD
H62 0 1.12 0.68
HG2 20 1.64 1.06
HG2 100 1.58 0.69
BSA 0 1.52 0.85
BSA 20 2.23 1.42
BSA 100 1.95 0.92
Palm 0 4.67 * 1.22
Palm 20 8.87 * 2.3
Palm 100 9.32 * 2.13
Oleate 0 1.34 0.8
Oleate 20 1 .35 0.48
Oleate 100 1.43 0.35
Linoleate 0 1 .23 0.46
Linoleate 20 1 .82 0.39
Linoleate 100 1.25 0.1

Table 4.1. The cytotoxicity of various treatments. Cells were treated for 24 h with 0.7 mM FFA in the
presence or absence of 20 or 100 ng/ml TNF-01. LDH released was measured and normalized to total LDH.

Data presented as mean +/- s.d. of n = 9 from three independent experiments. *, significantly higher than
control, p < 0.01.

4.3.2 Microarray analyses

To identify the genomic response of the cells to the treatments, cDNA microarray
analyses were performed. Two-way ANOVA followed by removal of ESTs identified
that 830 genes were affected at a significance level of p < 0.01 by either FFAS, or TNF-01
or their interaction. Further analyses were conducted using these differentially affected
genes to identify pathways and processes related to lipid accumulation and cytotoxicity.

Since FFAS were the primary effectors of both lipid accumulation and cytotoxicity while

94

TNF-01. had no significant effect on either the lipid accumulation or the toxicity by itself,

the genes affected only by FFAS are discussed below.

4.3.3 Identification of pathways important in lipid accumulation and cytotoxicity
In order to identify the genes associated with intracellular TG accumulation, cytotoxicity
or both, we conducted single and dual-response GA/PLS analysis. The pathways enriched

in the selected genes were identified by Gene Ontology (GO)-based analyses using the
Gene Ontology Tree Machine (GOTM) (Zhang et al., 2004) and F atiGOJr (Al Shahrour et
al., 2005) online tools. While the GOTM used all the genes for the analysis and identified
the significantly enriched processes by employing the hypergeometric test, FatiGOJr

provided information on the KEGG pathways as well as the transcription factors whose
target genes were enriched in the list. Since the two analyses provided complementary

information, both were employed.

4.3.3.1 Genes relevant to TG accumulation
The GA/PLS analysis identified 88 genes among the 830 differentially expressed genes
as relevant to TG accumulation. Genes with the highest frequencies are listed in Table

4.2.

95

Table 4.2. Top genes identified for TG accumulation. GA/PLS analysis was conducted to identify genes
related to intracellular TG accumulation. The genes with highest frequencies are presented.

 

 

Accession
Number Gene Name Freq

AA457700 (gC) stearoyl-CoA desaturase (delta-9-desaturase) (SCD) 12

AA279072 (gK) inositol polyphosphate phosphatase-like 1 (INPPL1) 11

T81764 (90) cell division cycle 27 (CDC27) 11
(9C) protein phosphatase 1, regulatory (inhibitor) subunit 12C

AA464195 (PPP1 R120) 11
(gN) steroid-5-alpha-reductase, alpha polypeptide 1 (3-oxo-5 alpha-

H16833 steroid delta 4-dehydrogenase alpha 1) (SRD5A1) 11

R33650 (9M) inner mitochondrial membrane peptidase 2 like (IMMP2L) 11

 

Many genes with known roles in the regulation of lipid metabolism were identified. For
example, SCD was identified as the most important gene in regulating TG accumulation.
SCD catalyzes the unsaturation of stearoyl-CoA, the higher derivative of palmitate, to
yield oleoyl-CoA. The activity of SCD has been shown to regulate hepatic TG synthesis
in response to a lipogenic diet rich in saturated FFAS (Miyazaki et al., 2001). SCD
increases TG synthesis by increasing the amounts of unsaturated FFA, a preferred
substrate for monoacylglycerol acyltransferase (MGAT), the enzyme which catalyzes the
synthesis of diacylglycerol (DAG) (Coleman and Lee, 2004). The microarray analyses
indicated that exposure to palmitate reduced the transcript levels of SCD, which was also
verified by RT-PCR (Figure 4.2a). Other FFAS also reduced the expression of SCD
(Table 4.3). However, the protein levels of SCD were significantly reduced only in the
palmitate-treated cells (Figure 4.2b). Reduced transcript levels, but no change in protein
levels in the unsaturated FFAS, suggests that these FFAS also reduced the rate of
degradation of SCD so that the net protein levels did not change significantly. Reduced
SCD levels in palmitate would hamper the channeling of this FFA by reducing the
amounts of unsaturated F FA, the preferred substrate for MGAT and thus, TG synthesis.

Reduced channeling of palmitate to TG may exacerbate the toxicity of this fatty acid.

96

Increasing the levels of SCD has been shown to reduce the cytotoxicity of palmitate by

increasing TG synthesis (Listenberger et al., 2003).

SCD

 

,, i ,iiMicrbarra-yj
rlBTflRL r

     

0.8

0.67 77777 L,

0.4 1

Relative Expression

0.2 ~

 

 

BSA Palm

Figure 4.2a Effect of palmitate on the transcript levels of SCD, measured by microarray and RT-
PCR. Cells were exposed to 0.7 mM palmitate for 24 h, controls were exposed to 4% BSA only. The
transcript levels of SCD were measured by RT-PCR. RT-PCR conducted by Xuerui Yang.

Ole Palm Ctrl

 

(sc0)

(Beta-Actin)

Figure 4.2b Effect of palmitate and oleate on the protein levels of stearoyl-CoA desaturase (SCD).
Cells were exposed to 0.7 mM palmitate for 24 h, controls were exposed to 4% BSA only. The protein
levels of SCD were measured by western blotting. Experiment conducted by Zheng Li and Xuerui Yang.

Another important regulator of TG synthesis identified by the GA/PLS analysis was

INPPLl. The product of INPPLI is the phosphatase SH2-containing inositol 5'-

97

phosphatase 2 (SHIP2), which is known to inhibit insulin signaling by hydrolyzing the
inositol triphosphate (1P3). SHIP2 is considered an important target in controlling
obesity, type 2 diabetes and insulin resistance (Sasaoka et al., 2006). It has been shown
that INPPLl-null mice are highly resistant to dietary obesity (Sleeman et al., 2005) and
liver-specific expression of a dominant-negative SHIP2 improves glucose metabolism
and insulin resistance in the db/db mice (Fukui et al., 2005). Palmitate upregulated the
expression level of INPPLI (Table 4.3) and thus is expected to cause a greater inhibition
of insulin signaling than the other FFAS. Indeed, insulin resistance in muscle cells has

been shown to correlate with intracellular levels of saturated FFAS (Manco et al., 2000).

FatiGO+ analysis of the selected genes indicated that the KEGG pathways of insulin
signaling, MAPK signaling, phosphatidylinositol metabolism and apoptosis (Table 4.4)
were related to TG accumulation. The importance of insulin signaling in lipid metabolism
and the role of inositol metabolism in regulating insulin signaling were discussed above.
The analysis also suggested an important role of MAPK in TG accumulation, which was
verified experimentally, see section below. The identification of the apoptosis pathway
suggested a relation between TG accumulation and cell death. It is well known that
excessive intracellular lipid accumulation alters cellular physiology and is considered the
first step of lipotoxicity (Weinberg, 2006). However, despite greater TG accumulation,
the unsaturated FFAS were not toxic. Efficient TG synthesis in response to these fatty
acids prevented their acute toxicity. However, chronic exposure to elevated levels of the

unsaturated FFAS is cytotoxic (Dymkowska et al, 2006).

98

GOTM analysis of the 88 genes selected for TG accumulation vs. the initial list of 830
differentially expressed genes identified that the genes in the GO category of ‘regulation
of cell growth’ were significantly enriched (Table 4.5), suggesting a relationship between
intracellular lipid accumulation and cell growth, which has been previously reported

(Tietge et al., 2001).

4.3.3.2 Genes relevant to cytotoxicity

Of the initial 830 genes, GA/PLS identified 115 as important for LDH release. Table 4.6
lists the genes with the highest frequencies, see additional file 2 for the complete list.
Among the genes with highest frequencies to LDH release were those involved in ROS
generation and scavenging, e.g., cytochrome P450 subfamily 1 (CYP1A1) and glutamate
cysteine ligase (GCL), respectively. Our previous study had identified an importantrole
of ROS generation in palmitate toxicity (Srivastava and Chan, 2007). In addition, genes
involved in the regulation of cell death such as caspase 6 (CASP6) and eukaryotic

translation termination factor 1 (ETFI) were also identified (not shown).

99

Table 4.6 Top genes identified for cytotoxicity (LDH release) by GA/PLS analysis. GA/PLS analysis
was conducted to identify genes related to cytotoxicity. The genes with highest frequencies are presented.

 

 

Accession
Number Gene Name Freq

(gC) cytochrome P450, subfamily l (aromatic compound-inducible),

AA418907 polypeptide 1 (CYP1A1) 12
(90) isocitrate dehydrogenase 2 (NADP+), mitochondrial (IDH2), nuclear

AA679907 gene encoding mitochondrial protein 12

AA877618 (gC) fatty acid amide hydrolase (FAAH) 12

R32848 (90) S100 calcium binding protein P (S1OOP) 12

H93482 (g) glutamate-cysteine ligase, catalytic subunit 11

AA406269 (9) nuclear factor l/X (CCAAT-binding transcription factor) (NFIX) 11

H09819 (gC) phosphomevalonate kinase (PMVK) 11
(90) pleckstrin homology domain containing, family A (phosphoinositide

N95011 binding specific) member 1 (PLEKHA1) 11

 

F atiGO+ analysis of the genes selected by GA/PLS to be important in regulating LDH

release identified that the KEGG pathways of MAPK signaling, oxidative
phosphorylation and insulin signaling had the highest number of genes, suggesting that
alterations in these pathways were significantly associated with LDH release. The cellular

components categories of ‘membrane’ and ‘mitochondrion’ had the highest number of

members (20 and 11, respectively, out of the 63 genes used by FatiGO+), suggesting an

important role of membrane integrity and mitochondria in the release of LDH (see
additional file 2). LDH is a cytosolic enzyme that is released when the cellular membrane
integrity is compromised. We also observed a significant reduction in the mitochondrial
potential (Figure 4.3a) in the palmitate-treated cells. Loss of mitochondrial integrity
would lead to the release of cytochrome c, causing caspase activation. In support of this,
increased caspase-3 activation was observed in the cells treated with palmitate (Figure
4.3b). Thus, these experimental observations corroborated with the genes identified to be

involved in regulating mitochondrial structure and function.

100

1.6
1.4 -
1.2 —

1 -
0.8 -
0.6 —
0.4 -
0.2 -

o _
rum—.8 “9’90 01 (‘9’86 6 “9'90 6 ”0’90 0 “90¢

 

 

 

Relative Mitochondrial Potential

 

Medium —’ Control BSA Palmitate Oleate Llnoleate
Treatment

Figure 4.3a Effect of various treatments on the mitochondrial potential. Cells were treated with 0.7
mM of the indicated FFAs in the absence or presence of indicated concentrations of TNF~01 (in ng/ml).
After 24 h of treatments, the mitochondrial potential was measured using JC-l dye, and represented as
relative ratio of red to green fluorescence. Data presented as mean +/- s.d. of n = 9 from 3 independent
experiments.

 

‘IDay1

8 I IDay2

Relative Caspase Activity
N 00 «b 0! a) N

.L
l

 

 

   

O
1

BSA Ole Palm

Figure 4.3b Activation of caspase-3 by palmitate treatment. Cells were treated with 0.7 mM of the
indicated FFAS. After 24 h of treatments, cells were lyzed and the caspase-3 activity was measured using a
fluorirnetric assay. Data presented as mean +/- s.d. of n = 9 from 3 independent experiments.

101

The GOTM analysis indicated that genes belonging to the category ‘Tumor necrosis
factor receptor binding’ were significantly enriched in the selected list of genes. This
suggests that a number of genes selected for the LDH-release were affected by TNF-01.
Indeed, the LDH release was affected by palmitate and the interaction of palmitate and

TNF-01 (Table 4.1). The genes altered in the selected list were altered by FFAs or TNF-01

alone, or by their interaction. To identify the genes affected by the FFAS only, two-way
ANOVA was performed. Genes related to lipid metabolism, such as lecithin cholesterol
acyl transferase (LCAT), 24—dehydrocholesterol reductase (DHCR24) (Table 4.3) were
affected by FFAS. DHCR24 is involved in cholesterol biosynthesis, while LCAT is
involved in the esterification of the cholesterol. Exposure to palmitate reduced the
expression of LCAT while increased that of DHCR24 (Table 4.3). These results suggest
an increased synthesis, but reduced esterification of cholesterol in the palmitate cultures.
Cholesterol is transported from liver in the form of cholesterol esters. Reduced
esterification of cholesterol would reduce secretion of lipoproteins. In agreement with
this, diets rich in saturated FF As have been shown to have reduced lipoprotein secretion
as compared to monounsaturated or polyunsaturated FFAs (Cheema and Agellon, 1999).
The effect of palmitate on these genes also suggests increased accumulation of

cholesterol in the cells, which itself may cause cytotoxicity (J iminez-Lopez et al., 2006).

4.3.3.3 Dual-response GA/PLS to identify genes relevant to both TG accumulation
and cytotoxicity
In order to find better targets to regulate the toxicity of saturated FFAS, it is important to

identify genes that affect lipid accumulation as well as toxicity. To identify such genes

102

which affect both responses, the dual-response GA/PLS (LDH + TG) was applied. This
analysis identified 92 genes which may regulate both TG accumulation and cytotoxicity

(see Table 4.7 for the genes with the highest frequencies).

Table 4.7 Top genes selected for TG+LDH by dual response GA/PLS. Modified GA/PLS analysis was
performed to identify the genes which regulate both TG accumulation and LDH release. Genes with the
highest frequencies are presented.

 

 

Accession
Number Name Freq
(gN) NADH dehydrogenase (ubiquinone) 1, alpha/beta subcomplex, 1,
AA447569 8kDa (NDUFAB1) 12
R50953 (gC) mitogen-activated protein kinase kinase kinase kinase 2 (MAP4K2) 12
H23978 (gW) general transcription factor IIB (GTFZB) 12
T58773 (gN) inositol polyphosphate-Sphosphatase, 40kDa (INPP5A) 1 1
AA464580 (gC) acetyl-Coenzyme A carboxylase alpha 11
AA431988 (gC) fatty acid amide hydrolase (FAAH) 11
AA463931 (gC) inositol 1,3,4-triphosphate 5/6 kinase (ITPK1) 11
N22904 (gC) 3—phosphoinositide dependent protein kinase-1 (PDPK1) 11
AA025112 (gC) BCL2/adenovirus E1B 19kDa interacting protein 3-like 11
AA019459 (gC) protein tyrosine kinase 9 (PTK9) 11
AA064638 (9|) ubiquitin specific protease 7 (herpes virus-associated) (USP7) 11

 

NADH dehydrogenase (ubiquinone) l, alpha/beta subcomplex, l, 8kDa (NDUFAB1) and
mitogen-activated protein kinase kinase kinase kinase 2 (MAP4K2) were identified to
have the highest frequency among all the genes in the group. In fact, NADH
dehydrogenases formed the most prominent group of genes in the list, with 7 different
NADH dehydrogenases selected. In accordance with this, the GOTM analysis also
identified that the GO category of ‘oxidoreductase activity, acting on NADH or NADPH,
quinone or similar compound as acceptor’ was significantly enriched (p < 0.01) in the
selected list. These results indicated an important role of NADH dehydrogenases in

regulating both cytotoxicity and TG accumulation.

103

Another gene that had the highest frequency was the mitogen-activated protein kinase
kinase kinase kinase 2 (MAP4K2). The FatiGOJr analysis of the selected genes identified
that the maximum number of genes belonged to the KEGG pathway of MAPK signaling
(Table 4.4). In fact, all three GA/PLS analyses identified the KEGG pathway of MAPK
and insulin signaling to be among the most important in regulating lipid accumulation
and cell death. MAPKS are known to regulate diverse cellular processes. For example,
JNK (Simbula et al. 2006, Lee and Bae 2006) and p38 (Kim et al 2006) play important
roles in mediating the cytotoxicity of various stimuli. JNK has also been shown to
regulate TG accumulation in liver (Schattenberg et al., 2006). It is well-known that
insulin signaling increases FFA esterification and reduces cytotoxic stresses, in part, by
activating the cytoprotective protein kinase B (PKB) (Mitsui et al., 2001). PKB
phosphorylates and inactivates apoptotic proteins such as procaspase-9 and BAD (Bcl-
2/Bcl-XL antagonist causing cell death), while activating anti-apoptotic proteins such as
X-linked inhibitor of apoptosis protein (XIAP) (Dickson and Rhodes 2004). Thus,
reduced insulin signaling would reduce lipid accumulation and increase cytotoxicity.
Among the different F F As, exposure to palmitate altered many genes involved in insulin
signaling. Thus, it is likely that palmitate affected insulin signaling to a greater extent

than the other FFAs (Chavez et al., 2003).

4.3.4 Experimental validation of the identified gene-groups
4.3.4.1 NADH dehydrogenases
The dual-response analysis identified that NADH dehydrogenase regulate both TG

accumulation and cytotoxicity. Therefore, the roles of this enzyme in these processes

104

were investigated. NADH dehydrogenases form the mitochondrial complex I, which is
involved not only in the synthesis of ATP, but is also an important site of ROS
generation. ROS generation has been shown to be one of the important mechanisms of
palmitate-toxicity (Srivastava and Chan, 2007). Though a role of ROS generation at
complex I has been suggested in the cell death caused by sulforaphane (Singh et al,
2005), transforming growth factor (Herrera et al., 2004) and deoxycholate (Payne et al.,
2005), a similar role of complex I in lipotoxicity or triglyceride accumulation has not
been indicated. We tested the effects of inhibiting NADH dehydrogenases on the
cytotoxicity and lipid accumulation in response to palmitate. Inhibiting these enzymes
with rotenone reduced the cytotoxicity of palmitate significantly (as shown in Figure 2.5b
and reproduced in Table 4.8 to facilitate its comparision with TG). In addition to its effect
on cytotoxicity, NADH dehydrogenase also increased the accumulation of triglycerides
in response to palrrritate (Table 4.8). Inhibiting NADH dehydrogenase also increased
slightly the lipid accumulation in response to oleate (not shown). These results suggest
that the NADH dehydrogenases may indeed regulate both lipid accumulation and

cytotoxicity in response to palmitate.

Table 4.8 Effect of inhibition of NADH dehydrogenase on the cytotoxicity and TG accumulation in
response to palmitate. Cells were pretreated with the indicated concentrations of rotenone for 30 minutes
and then exposed to 0.7 mM palmitate for 24 h. The cells were then treated again for 30 minutes with
rotenone and put again in 0.7 mM palmitate. The cytotoxicity as well as TG accumulation after 24 and 48 h

was measured. Data presented as mean +/- s.d. of n = 9 from 3 independent experiments. *, significantly
less than the no inhibitor condition, p < 0.01. #, significantly more than the no inhibitor condition, p < 0.01.

 

 

 

 

LDH release LDH release TG (Day 1) TG (Day 2)
Concentration (DAY fl (DAY 2) u m rotein u Im rotein
None 4.5 +/- 0.3 35.7 +/- 3.4 69.4 +/- 16.1 173.8 +/- 16.7
250 nM 3.4 +/- 0.3 * 24.9 +/- 1.4 * 63.3 +/- 14.2 191.1 +/- 12.1
500 nM 2.7 +/- 0.2 * 14.8 +/- 3.2 * 74.3 +/- 12.0 197.8 +/- 11.3
1000 nM 3.4 +/- 0.2 * 20.0 +/- 4.6 * 104.6 +/- 7.8 # 204.9 +/- 8.5 #

 

 

 

105

4.3.4.2 MAP kinases

All three cases (TG, LDH and TG+LDH) identified genes belonging to the KEGG
pathway of MAPK signaling. MAPKS play an important role in cellular signaling
pathways and effect diverse processes. There are three primary branches of the MAPK
signaling network, viz., the c—Jun N-terrninal kinase (JNK), the extracellular signal
regulated kinase (ERK) and the p38 kinase. The roles of the different branches of MAPK
signaling network were evaluated by studying the effect of inhibitors specific to each.
Inhibiting ERK and JNK reduced the toxicity of palmitate significantly (Figure 4.4a),
while p38 had no effect. Inhibiting JNK, but not ERK or p38, also increased the lipid
accumulation in response to palrrritate (Figure 4.4b). Thus, important roles of MAPKS in
regulating the lipotoxicity as well as lipid accumulation, as predicted by the integrative
GA/PLS, were corroborated experimentally. Among the different MAPK pathways, only
JNK has been shown to play an important role in the development of NASH
(Schattenberg et al., 2006). This study found that mice with deletion of JNK gene were
resistant to development of NASH when fed methionine-choline deficient (MCD) diet.
Very low steatosis was observed in these animals and that could partly suggest lower
steatohepatitis. Our results are similar to theirs only in terms of cell death, but different
from their lipid accumulation results. However, no previous study has shown important

role of ERK in lipotoxicity, as shown in the current study.

106

 

 

IDay 1

 

 

 

N
0|

LDH released (%)
N
O

 

 

 

Palm P38 JNK ERK

Figure 4.4a Effect of inhibition of various MAPKs on the cytotoxicity of palmitate. Cells were treated
for 48 hours with palmitate in the presence of 2 uM SB203580 (to inhibit p38 pathway), 75 uM SP-600125
(to inhibit IN K pathway) or 50 uM U-0126 (to inhibit ERK pathway). Media were changed daily. The LDH

released into the medium after 24 and 48 h of treatments was measured and normalized to total LDH. *,
significantly lower than palmitate (p<0.01).

400

 

 

IDay1
IDay 2

 

 

 

‘3’ §

250 -

200 -

3‘

100 -

Intracellular TG (ugl mg protein)

”I
O
L

 

 

 

Palm p38 JNK ERK

Figure 4.4b Effect of inhibition of various MAPKs on the TG accumulation in response to palmitate.
Intracellular levels of TG in the cell lysates were measured after desired time of exposure. Data presented

as mean +/- s.d. of 3 independent experiments. *, significantly higher than palmitate (p<0.01).

107

4.3.5 Genes affected by exposure to palmitate

Exposure to palrrritate significantly increased the expression of INPPSA, PTPNI,
INPPLl, NFIX, DHCR24, 881-3 and CASP6 (Table 4.3). CASP6 is a caspase which can
activate caspase-3, the effector caspase which lyses and activates downstream proteins
causing apoptosis. It is likely that increased activity of CASP6 plays a role in the caspase-
3 activity observed in palmitate-treated cells. The effect of palmitate on the expression of
INPPSA, PTPNI, INPPLI and NFIX was different than the effect of the other FFAS on
these genes. These genes have been shown to play important roles in a variety of
physiological processes. The product of INPPSA is a type I inositol 1,4,5-trisphosphate
(InsP3) S-phosphatase, which antagonizes insulin-signaling by hydrolyzing Ins(1,4,5)P3.
The hydrolysis of 1P3 mobilizes intracellular calcium and acts as a second messenger

mediating the cells response to various stimuli. Exposure to palmitate has been shown to

cause increases in intracellular CaH (Remizov et al. 2003). Hydrolysis of 1P3 would also

counteract the insulin signal transduction. Thus increased levels of INPPSA would cause
multiple deleterious effects on cellular function. PTPNI codes for protein tyrosine
phosphatase PTP-lB. PTP-IB has also been shown to counter insulin signal transduction
by dephosphorylating phosphotyrosine residues in the insulin signaling molecules, and
has been suggested as a risk factor for the development of type 2 diabetes (Kennedy and
Ramachandran 2000, Wagman and Nuss 2001). PTPNl and INPPLI were increased only
in the cells treated with palmitate, indicating that the saturated fatty acids may be
important mediators of insulin resistance. Indeed, insulin resistance in skeletal muscles is
known to correlate directly with the saturated FFA levels (Manco et al., 2000). Our

results suggest that a similar relationship between saturated FFA and insulin resistance

108

may exist in hepatocytes as well. Thus, the genes affected by palmitate suggest a greater
inhibition of insulin signaling in the palmitate-treated cells as compared to the other

FFAS.

4.4 Discussion

Hepatocytes are central to whole-body lipid metabolism. Exposure of hepatocytes to
elevated FFAS and TNF-0L is associated with altered lipid metabolism and cell death
(lipotoxicity). However, our study has identified that saturated F FA are the primary
effectors of hepatic lipotoxicity and that the effect of TNF-or are secondary to the
saturated FFAS. FFA and TNF-01 exert their physiological and metabolic effects, in part,
by altering the expression of many enzymes of lipid metabolism and signal transduction.
No previous study had investigated the genomic effects of combined exposure to these
factors on hepatocytes or hepatoma cells. In addition to identifying the genomic effects of
these treatments, we identified the genes and processes involved in simultaneous
regulation of both lipid accumulation and FFA-toxicity. These analyses identified novel
targets to combat lipotoxicity at multiple levels, some of which were experimentally
validated. A comparison of the genes selected for the (TG+LDH) case with those selected
for TG or LDH individually showed that most (53 out of 92) of the genes selected by the
dual—response analysis were novel to this list. Among the genes selected only in the dual-
response analysis were many NADH dehydrogenases genes, such as NADH
dehydrogenase (ubiquinone) l, alpha/beta subcomplex, l, 8kDa (NDUFAB1), NADH
dehydrogenase (ubiquinone) 1 beta subcomplex, l, 7kDa (NDUFBI), NADH

dehydrogenase (ubiquinone) 1 alpha subcomplex, 7, 14.5kDa (NDUFA7), NADH

109

dehydrogenase (ubiquinone) flavoprotein 2, 24kDa (NDUFV2), NADH dehydrogenase
(ubiquinone) flavoprotein l, SlkDa (NDUFVI). In fact, out of the 6 NADH
dehydrogenase genes that were identified by the dual-response analysis, only 1 (NADH
dehydrogenase (ubiquinone) 1 beta subcomplex, 4, lSkDa) was found for TG analysis,
and none was found in the LDH analysis. As our experiments showed (in this chapter and
in chapter 2), NADH dehydrogenases played an important role in regulating TG
accumulation as well as toxicity. No previous study has shown the role of NADH
dehydrogenases in regulating both lipotoxicity and lipid accumulation. Because NADH
dehydrogenases are known to be involved in oxidative phosphorylation and the
generation of energy as well as ROS (Starkov and F iskum 2003), their role in mediating
lipotoxicity may not be very surprising. However, the GA/PLS analysis and the
experiments identified that these enzymes also play an important role in regulating TG
synthesis. One of the potential ways by which inhibiting NADH dehydrogenases could
increase TG accumulation may be through the reduction of ATP levels. ATP has been
shown to inhibit the incorporation of FFAS into TGs (Hansen et al., 1984 a and b).
Inhibition of NADH dehydrogenases would reduce ATP levels, thus, could increase TG
accumulation. The MAPKS, JNK and ERK, were found to play important roles in the
cytotoxicity of F F A. Inhibiting JNK also increased lipid accumulation, while ERK had no
effect on lipid accumulation. Thus, inhibiting IN K may have reduced the cytotoxicity, in
part, through increased TG synthesis. However, even though we agree with the
conclusion of others on the importance of JNK in obesity (Karin and Gallagher 2005),
insulin resistance (Hirosumi et al., 2002), type 2 diabetes (Bennett et al., 2003,) and

steatohepatitis (Schattenberg et al., 2006), our results suggest an increase in lipid

110

accumulation in response to JNK inhibition, which is contrary to a decrease of obesity
and intrahepatic lipid accumulation in the JNK-knockout mice observed in these studies.
Although the reasons behind the differences in observation are not clear, it is likely that
the differences are due to the different models (rodents in all these studies vs. cell culture
in ours) or the treatments (MCD diet in the study of Schattenberg et al. vs. elevated F FA
levels in ours). In this context, it is important to identify the mechanisms by which INK
affect lipid metabolism in vivo as well as on different cell types, before preceding with
therapeutic applications of these inhibitors. Among the potential ways JNK regulates
lipid metabolism is by inhibiting insulin signaling through phosphorylation of insulin
receptor at serine residues (Hirosumi et al. 2002) and by regulating the activity of SREBP
(Chang et al. 2005). It is also likely that the difference in observation on inhibition of
JNK between this study and others is because of very low levels of insulin (coming only
from the serum in the medium) in our study vs. higher (in vivo system) in the others. In
designing treatment strategies for lipotoxicity, it would be useful to identify targets that
reduce cell death without affecting lipid accumulation, because excessive lipid
accumulation can itself alter cellular physiology and make cells prone to lipid
peroxidation (Weinberg 2006). Our study identified ERK as another potential target to
reduce the cytotoxicity without affecting lipid accumulation. While a role for ERK has
been shown in various alterations caused by F FAs, such as the pro-inflammatory effects
of FFAS (Hennig et al., 2006), TNF-01 expression (Jove et al., 2006), NF-KB expression
(Ajuwon and Spurlock, 2005), apoptosis by conjugated linoleic acid (Miglietta et al.,
2006) and lSd-PGJ2 (Hashimoto et al., 2006), ERK’s role in palmitate induced

lipotoxicity has not been previously shown. The significant reduction in palmitate-

111

toxicity by inhibiting ERK without affecting lipid accumulation identified it as a
potentially important target for lipotoxicity.

Finally, the dual-response GA/PLS methodology was able to integrate the metabolic,
physiologic and genetic information to identify genes relevant to multiple metabolic
functions. While we have applied this analysis to only two responses of interest, the
analysis is amenable to multiple quantitative responses. To incorporate larger number of
metabolites, latent variables can be extracted from the metabolite data with methods such
as principal component analysis (PCA), partial least squares (PLS) (Griffin 2005) and
Fisher Discriminant Analysis (FDA) [Chan 2003]. PCA and PLS can extract latent
variables that account for most of the variance in the data while FDA extracts latent
variables that have the largest discriminating capability. The number of latent variables
can be determined based upon the variance captured by the latent variables. For example,
if we want to capture 90% of the variance in the raw data, we will include enough latent
variables to extract 90% (or more) of the variance. For complex diseases which affect
multiple cellular processes, such analysis could be a useful tool to identify targets which
can simultaneously regulate multiple processes. It is likely that the incorporation of
multiple measurements would provide better targets to regulate cellular responses and the

development of complex diseases.

112

m:

 

 

003.800 003.800 0000800 30392
0035.80 n... 0.00...
888:. 02 02 .9343. 0083
N 000:0. 0005.. 000:0. 00050. 50.08 00.02.00-803... .09
.138: 3. 038.8 02.50
00008:. 02 0038:. 0.00 .:0:0> 8.80:0: 000003
.0090. 0000.08 050.08 030.090.00.080 .0 0000000 .29
003.800 003800 02 ... .Fuhm. F 8.00. :o..05::0. :o..0.0:0.. 5.900030 .29 000000.38.
00008:. 0038:. 00008:. ... .200. 0 05.390 050.28 .0 80008030 .09 m .N 59?.
0000800 0000.800 0000.800 .. .00.... .F 80:88 5.5000030 H.2.... 0:50 506.085.. .09 mNmNE
00008:. 02 0038:. $3010. 000.0300: .90.00.0...00:0>:00.vw .09 000033..
02 0000800 003800 .../0.0... 0008.39.33 88.388.55.00. .9 00009.
02 0000800 00008:. * mowoov<<
.502. .500. 88889. 2.25.280. x: 880. .028: a.
...0.. 8. 00.00.00 00:00
0000800 0000800 0000800 .. .3150. N 0.0:. 0008.330 08.50.20 ..29 mowomv<<
02 02 888... 20562; 09.2. .80.258;90-20588on 9.8:. .29 05005
00008:. 00008:. 02 m 55.530 0000 50:0 0:0. 0008.38.32.53 05.00 .09 3039?.
0000800 0000.800 oz 0:50 0003.80.00 < 083.300.3000 .09 800032.
0000800 0000800 00008:. 5.. .3002: F 0570000800.... 0.0—8008200 .0..0o5 .v.9 «Bong
0000800 0000.800 003800 .3 .000. 0005000000200. 0005.030 <00-_>o.00.0 .09 8:032
0h 8. 0080.00 00:00
0.3.0:... 0.3.0 0.02500 3602 %
:0.00000<

 

 

 

 

 

0..."... =0 .3 3.0.050 00.00....0 000w .._. .m .03.... .050 0... 5.? 0000800 00 .8530 0.5.8.00 E 3808...... 00.00....0 0:0w .

030:0 80053.0 0: .0... 00.00.00. 02 .500 8:38.:— 0... :0 0002. 83» 00.00.00 ..0 3.000008 0... :0 3:08.00... 30.03 ..0 .00....0 m0 030,—.

113

v:

 

 

 

 

00.0 0 6.05.. 0:223.
mud m .6350... no.0 0 0.00809.
05.0:90 .:>>
00... v .0358 .00 0 50:80.05
0588.0 E5200 0.08008 .9005
00.0 0 800388008 0...... m 0350.. 00.0 0 £0.00 05.0:90
02.035 05.080 5.30:. 8500530008020
0... 0 .0358 00.0 0 8.082000% ...: 0 .0358
05.0.5.0 {<2 02.0206 0503.0 0.0.22
00.... n .0358 00.. 0 .0358 00.0. 0 .0358
05.0:90 5.00:. 05.0:90 v.n.<.2 0588.0 580:.
00:00 00:00 8350.. 00:00 00:00 8350.. 0:. .0 0:82 00:00 00:00
.08. .0 a 0:. .0 0E02 .08. .0 ... .08. .0 0
.0 ... .0 .\. .0 .\. 8350.. 2.. .o 2:02
.10.. + 0.. :3 0.

 

 

 

 

.00.. + 05. 0.3 :00 .05 3. 00:05... 880 2.. 3 03:323.. 5:02. 0000. 3. 0......

 

0000800 0000800 0000.800 ._. .908. m 000:... 0500;. 50.08.09 003530.

0000800 02 00008:. ... moooom
..zn...h_. . 09.. 88009-8: 000.0888 05003. 50.08 .09

0000800 0000800 02 * .mxn.<v.n.<.2. .3003
m 000:... 50.08 00.02.00-008... 50.08 00.0>..00-:0mo..E .09

0000.800 0000800 0000.800 ... .08....50. 000:8 000.032
0008.0:0... 08.50.08 .00... 0008.0:0...-w-0:o.:.0.:_m .09
.. 88000: 5.30:. .09

 

 

 

 

 

114

m:

.0000 05005003093 05 0. ”50.8000 5:530...— " .— 0:0 GEOV 030.. "m .0000” ._0 3. 0000005 "m .005” ..0 ._.. 00:00:: "O

 

 

 

 

 

 

 

 

P 000:5 0000000500 0:05.056 03.0
0ov_v~ .N 50.2090:
8:05.653 00050000300 10.02 mm:
09.5 .— 50.9090:
8:05.653 00050000300 19.2 mm:
09.9 .v .x0_0E000:m 0.05 m 05.0:90
F 8:05.653 00050000300 Io<z oi... 050.006 :0 5000:0000 E8
000:. .w 08.080003 0.00 91.00800: .>__E0t00:0 w 05.0:90
r 8:05.653 00050000200 19.2 no? 8:09: 00.00“. 050.00: 8:5. 35 0:528 :0 80000000 $00
09.0 .P 08.080053 0.0000550 .F C ._00E0E 0550:8030 m 50.00 050:5 .200.
8:05.653 0005020300 10.02 005. 02.5 050.0 50.0.2082 030 5305 005-5505 003
09.0.3 K .x0_0E000:0 0:50 F 50.00 F 50.0.0 050:5 00.00“.
F 8:05.653 000.000.0200 19.2 5:. 02050000090000: “.2... 5:: 5306 05.5505 003” “Hwy—”w
a. a. o. 05
0802 0:00 0802 0:00 2:02 0:00 :_ 00:00
$8.0":unmanmummdumcmo 09880
00 0:000:50 :0__E_0 :0 0:05.6 .1092 oFo.on0uom.mumnmm.onwumno wooduanménmnmmdnmnvno 2000.00
:0 19.2 :0 0550 $52.00 000530000050 650:5 5.0000: .903 050.60: 0080... 5:506 :00 :0 520.0001 00:0:cm
In... ... 0... In... 0...

 

 

 

 

83...... Show 2: 3 8:05... 85033 00 3.0.2.0 0... ~35

115

CHAPTER 5. INTEGRATING THE ONTOLOGY
INFORMATION AND EXPRESSION DATA TO IDENTIFY
GENES THAT REGULATE METABOLIC AND
PHENOTYPIC FUNCTIONS

5.1 Introduction

The development of microarray technology has made it possible to procure information
on the genomic responses of the cells to virtually any condition. The challenge now is to
efficiently identify the genes that regulate the desired metabolic/ phenotypic response of
the cells to identify better targets to regulate cellular responses. A substantial amount of
efi°ort has been invested to develop techniques that efficiently analyze gene expression
data. However, current methods to integrate gene-phenotype or gene-metabolite
information do not actively incorporate the functional information of genes. We
hypothesized that active incorporation of functional information of the genes would help

identify better targets to regulate cellular responses.

Many techniques have been deveIOped to identify the genes which can identify the
subpopulations and discriminate between categories, e.g. cancerous vs. non-cancerous
tissue from a set of microarray samples. The simpler ‘filtering’ techniques include the
Wilcoxon's ranksum test (Troyanskaya et al., 2002), Fisher Discriminant Analysis (FDA)
(Hwang et al, 2003), partial least squares (Tan et al., 2004), GA- (Liu et al., 2005) and
principal components- (Tan et al, 2005) based classification and clustering (Guo et al.,
2006 and Chen et al., 2006) etc. The filtering techniques separate the subgroups by

maximizing the ratio of between-group to within-group variance. Genes are ranked

116

according to their correlations to a phenotype and the genes with the highest ranks are
considered discriminatory. However, these techniques suffer from various drawbacks,
e.g., they require the features (genes) to be independent and uncorrelated, which
microarray data is obviously not. Techniques which combine multiple filter methods
include genetic algorithm coupled partial least squares (GA/PLS) (Li and Chan, 2003)
Therefore, sophisticated ‘wrapper’ techniques have been developed, which employ a
trained learning machine to identify a subset of genes and evaluate the relevance of the
genes in the subset to a phenotype. Examples of wrapper techniques include support
vector machines (SVM) (Brown et al., 2000). The wrapper methods are considered better
than the filter methods because they can incorporate the inter-correlation of genes and
can also determine the optimal number of variables. However, these methods are
conceptually difficult and often require significant computational power. A third set of
techniques are also being developed which combine the wrapper and the filter techniques
(e.g. the kernel Fisher discriminant analysis, KFDA) (Cho and Lee et al., 2004) or multi-
layer perceptrons (Wang et al., 2006). The typical procedure is to identify the
discriminating genes and then apply gene ontology (GO) to identify selected/
discriminating GO processes. However, these techniques rely primarily on the microarray
data for information generation and the GO analysis is essentially passive. While
conceptually straight-forward, these purely data-based models have two major
shortcomings. One, they strongly depend on the quality of the microarray data and can
miss genes with altered expressions if the data is noisy. The microarray data is inherently
noisy, increasing the probability of false-negatives. Two, data-based models do not

incorporate the vast amount of information already available on the function of the genes.

117

Usually, the fimctional information of the genes is employed only in the post-processing
of the selected genes. In order to circumvent these issues, alternative analysis methods
are being developed in which prior information on the genes are being incorporated (Pan,
2005, Morrison et al., 2005, Le Philip et al., 2004). Many of these studies support the
hypothesis that incorporation of prior knowledge either improves the classification
efficiency or identify more relevant biological processes.

In the knowledge-based processes the association of a pre-defined gene ontology (GO)
category to a phenotype is statistically evaluated, identifying the discriminating cellular
processes. One such approach is gene set enrichment analysis (GSEA) (Subramanian et
al., 2005, Beisvag et al, 2006 and Draghici et al., 2003). Even though such GO-based
techniques identify the gene sets having significant association with a phenotype, they
usually identify a subset of a G0 category as potential contributors. Individual genes
which may be significantly altered and have important association with the phenotype
may not be selected if the gene-group they belong to is not enriched. Even among the
subsets that are identified as important, the identification of a few (one or two) targets can
be very subjective. Another major drawback of these approaches is the tedious manual
preparation of predefined data sets for the analysis. While data sets are available for
Affymetrix chip for some species, for other platforms, such as the custom cDNA
microarrays, one would need to manually define the gene sets, which can delay
procurement of downstream information or introduce errors. Therefore, such techniques
are useful when the expressions of many genes of an important, causative pathway

change. For other conditions where there is change of only a few, rate-controlling genes

118

of the pathway, such approaches may not be as informative. Therefore, methods that

combine both gene expression as well as the domain knowledge are needed.

Here, we present a new method which integrates the functional information of the genes
with their expression data (X) to identify the genes that regulate a specified quantifiable
cellular response (Y). One of the major advantages of this method is the rapid
assimilation and the ease of update of the functional information of the genes. Another
advantage is the parallel identification of genes relevant to multiple cellular responses
making it a more efficient high-throughput analysis. The ability of this method to handle
potentially thousands of genes makes it possible to utilize less stringent pre-processing
constraints, reducing the risk of false negatives. The method also incorporates the
variance of the microarray data to better identify average expressions. Even with good
quality controls, microarray data are typically noisy and have large variances.
Traditionally, averages of the log-ratios are utilized in the analysis. However, because of
large variance and small number of replicates used to calculate the averages, the average
expression values may not be accurate representation of the true expressions. Therefore,
in the current study, the gene expression data were approximated by a multivariate
Gaussian distribution and the regression weights were estimated by minimizing the
regression error that was averaged over the Gaussian distribution. The regression mixture
models employed in the analysis offer the ability to accurately capture processes with
multiple sub-populations, for example, the clinical data obtained for different age-groups
or gender, or in vitro experiments where cells are exposed to multiple treatments, etc. The

main idea behind the regression mixture model is to first cluster the experimental

119

conditions into a desired number of groups, and build a different regression model for
each group of conditions that relates the gene expression data (X) to the cell response
(Y). The clustering of experimental conditions is based on their regression weights, for
example, two experimental conditions will be grouped into the same clusters if they share
similar regression weights. However, the regression weights of each experimental
condition also depend on the clustering results because a regression model can be built
only for a group of experimental conditions. Hence, the technical challenge of regression
mixture model lies in resolving this dilemma. In this paper, the Expectation

Maximization (EM) algorithm was applied to effectively resolve this problem.

As a model application, we applied the framework to identify the genes that differentiate
the toxicity of the saturated FFA (SFA), palmitate, and TNF-a, from the non-toxicity of
the unsaturated FFAS (UFAs). A mixture model with 2 sub-groups was able to separate
the toxic conditions from the non-toxic ones. Incorporation of prior knowledge improved
the separation achieved by the mixture models. The analysis identified important genes
that separated the cytotoxic and non-toxic responses. The roles of these proteins in
regulating the cytotoxicity of the SPA are discussed. Thus, incorporation of prior

knowledge helped to identify targets that regulate the phenotypes of interest.

120

5.2 Materials and Methods

5.2.1 Bayesian Regression Model

Let X = (x19x29-"9xm) denote the gene expression data of m different

Xk = (x19 X2 9 ' ' ' 9 xn) represents the expression

experimental conditions, where

data of n genes under the kth condition. Let y = (y1 9 yz 9 - - - 9 ym) denote the cellular

responses for the m different conditions. By assuming that the conditional probability

T 2
Pr(y I X 9 W) follows the Gaussian distribution N (X W,0' 1) , the regression

error could be computed as
m T 2 T 2
err(w) = 202,. —X,. w) = "y—X w"2 (1)
i=1

where, W 2 (W1, W2 :- - - a W") are the regression weight assigned to the n genes. The

optimal solution for w that minimizes the above regression error is
T —l
w = (XX ) Xy (2)
Now, consider multiple replicates of the gene expression data under each experimental

_ 1 2 r
condition. Let Xk 9xk9°°°9 Xk denote the r replicates of the gene expression data

under the kth condition. We can approximate the distribution of gene expression data

under the kth condition, i.e.,Pr(Xk), by a Gaussian distribution N (EkaSk) where

Xk and Sk are calculated as follows:

121

——z<)<)

r i=1

Then, instead of regressing the cellular response to the averaged gene expression data

Xk , the Bayesian regression model will search for the regression weights w that

minimizes the following expected regression error:

i=1

(err(w)) = if: “y, — XiTW)2 Pr(x,.)dx,. = ”y —)_(TW“: +WT (2m Sk)w (4)

where, X = (X1, X2 9 ma Xm) . The optimal solution for Bayesian regression model

is

w=(XYT+S)—l)?y (5)

m
where, S = 21%] Sk . Comparing the above expression for w to the expression in (2),

we see that the primary difference between the two expressions is that (5) incorporates
covariance matrix S into its denominator. The introduction of S will assign smaller
regression weights to the genes whose expression data exhibit large variance compared to

the genes with small variance.

5.2.2 Mixture Model
In the regression mixture model, we don’t assume that all the experimental conditions
share the same regression weights w. Instead, we assume that there are underlying K

(K <m) different sets of regression weights, and each experimental condition will

122

choose the most suitable set of regression weights. More precisely, the conditional

probability Pr(y | X ,w) is modeled as
m m K T 2

log Pr(y I X) = ZlogPr(y,- IX.) = 210g ZpkNO‘. W00 ) (6)
i=1 i=1 k=l

In order to incorporate the variance in the gene expression data, similar to the Bayesian

regression model, we compute the expected of log Pr(y | X) that is averaged over the

distribution of the gene expression data X. More specifically, we have the expected

log Pr(y I X) computed as

(log Pr(y | X )) = i Ilog[2qu(xika,0'2))Pr(xi)dxi (7)

k=1

Where qk is the prior of choosing the kth regression model. To facilitate the

computation, we follow the idea of variational method by introducing a variational
distribution ¢i,k = Pr(k I yi, Xi), i: lamama k :13°°°9K and approximating the

above log-likelihood expression as follows

123

 

 

 

20'2
(8)
= iZfi. k log qk ——1—log(2fl0'2)—
i=1 k=l ’ 2
—r 2 T
(M — Xi wk) + wk Siwk
20'2

Thus, we have the following equations for computing ¢i,k , q k , W k , and 0'2

‘ 2

(yi ——X-ika) + wisiwk
¢i,k 0C qk exp _ 2
20'
qk : ‘1' 2m: ¢i k
’ (9)

124

We firrther improve the robustness of the model by the introduction of prior for the

-1
regression weights Pr(wk) as a Gaussian distribution N (09 ’1 ) . Then, instead

of just maximizing the log-likelihood, we will maximize the logarithm of the posterior

probability, i.e.,

log Pr(W | X, y) = log Pr(y | X, W) + log Pr(W) (10)
where, W = (W1: W2 3 "'9 W K) . The updating equations in (9) are almost

unchanged except that the equation for W k is changed to the following equation

wk 2 (Z:1¢i,k [iii-1T +Si]+’u)_l (Z:l¢i,kyi-ii ) (11)

Evidently, the introduction of the uninforrnative prior is in general to reduce the
magnitude of the regression weights. As a result, the small weights will become smaller

and even zeros, which will result in sparse solution for W.

5.2.3 Incorporation of GO information in the similarity matrix

To incorporate the GO information into the regression model, we first represent each
gene by the set of GO codes that are associated with the gene. We further expand the GO
profile of each gene by including the parent nodes of each associated GO code. We then
compute the similarity between two genes based on the overlap between their GO
profiles. Since some GO codes may be more important than others, we adopt the term

frequency independent document frequency (TF.IDF) weighting scheme of information

125

retrieval and weight each GO code by the IDF factor when computing the gene similarity.

More specifically, the IDF factor for a G0 code g is computed as

N+O.5

(12)
N (s)

where N is the total number of genes and Mg) is the number of genes whose profile

H)F(g) = log

include the GO code g. We denote the pairwise gene similarity by the matrix

T = [721' ]nxn where element 7: j represents the similarity between the ith gene and

the jth gene.

In order to regulate the regression weights using the GO information, we assume that two
genes are likely to share similar weights when they share a large number of common GO
functions. In other words, two genes tend to have similar regression weights if they share
large similarity in their GO profiles. Based on this assumption, we can construct the
following energy function to measure the consistency between the assigned regression

weights w and the gene similarity T
n 2 T
l(w,T)= 27;]. (wi—wj) =w Lw (13)
i,j=l
where L is the graph Laplacian of similarity matrix T. Evidently, the smaller the l(w,T)

is, the more consistent the regression weight w is to the gene similarity T. We can then
incorporate the gene similarity T into the regression model as a Bayesian prior for

regression weights W, i.e.,

126

Pr(W) 0C exp *1: W3 — 7: 71,].(w, " WI )2) (14)
i=1

i,j=l

|= exp(—wT [in + rL]w)

Note that in the above, in addition to the prior for the gene similarity T, we also include
the uninforrnative prior through the factor 2.1 . The updating equations for the integrated

Bayesian regression mixture model are almost identical to the ones in (10) except that the

equation for W k is changed to the following equation

w, = (2;), [3,323 + s,]+,u + 11.)" (2:1)”; ) (15)

5.2.4 Formation of the Integrative Mixture GO (IMGO) model

The two key parameters to our mixture model are I. and 1.". Parameter A is related to the
uninforrnative prior, and its role is to reduce the variance in the regression weights. Given
that we have a large number of genes (~700) and relatively small number of experimental
conditions (~14), there will be infinite number of ways to regress the cell response that
are equally valid. The introduction of uninforrnative prior ’11 will allow us to distinguish
among the regression models that have the same regression error. In particular, by
increasing the value of the parameter k, we require the regression model to favor the
regression model where only a small number of genes are assigned large weights and

most genes are assigned very small weights even zero weights. The parameter r is related

to the Bayesian prior for GO information. The larger the ‘L', the more we require the

127

regression weights to be consistent with the GO information of genes. Apparently, we
need to choose the appropriate value for T such that the regression weights can be

consistent with both the GO information and the cell response.

5.2.5 Discriminating ability of the IMGO

The ability of the IMGO to discriminate between the different phenotypes (e. g., cytotoxic
vs. non-toxic conditions or high ketogenic vs. low ketogenic conditions) was measured as
the difference in the probabilities of the two models for the palmitate without TNF-or (P-
0) condition. This condition was chosen because it defined a threshold value for the toxic
phenotype and showed maximum variation with the parameters. Therefore, an ideal
condition is one where the probability of one model is unity and the other zero. Thus, the
separation would have a range of 0 to 1, with 0 meaning both model are equally good and
1 meaning one of the models is superior over the other in modeling the condition of

choice.

5.2.6 Identification of discriminating genes

In order to identify the genes which contribute the greatest to the difference in the two
models (sub-populations), a concept similar to that used in the filter methods was
employed. The absolute differences of the weights for the two models were calculated
and the genes were ranked in decreasing order. The genes which showed greatest values
of the absolute weight differences were identified to have contributed the most in

separating the two models.

128

5.3 Results and Discussion

5.3.1 Cytotoxicity and ketone body release of the treatments

The cytotoxicity of various treatments is shown in Figure 5.1a. Among the different
F FAs studied, the saturated FFA palmitate caused the greatest toxicity. TNF-or co-
supplementation exacerbated the toxicity of the SFA, but TNF-or had no cytotoxic effect
by itself or in the presence of UFAs. In addition to the cytotoxic effects of the fatty acids,
the oxidation of FFAS was measured as the release of acetoacetate into the medium. It
was observed that palmitate was oxidized to a greater extent than the other FFA
treatments (Figure 5.1b) and TNF-or had no significant effect on the release of

acetoacetate.

10 i ‘

 

9.4
8~ *
7-
54
5i
4.

LDH Released (%)

3.
2.
11

 

 

 

0 W‘ T ' I I I I fi' I I I I I

TNF-a-DZO 100 0 20 100 0 20 100 0 20 100 0 20 100

 

 

HepGZ BSA Palmitate Oleate Linoleate

Figure 5.1a Cytotoxicity of the treatments. Confluent HepGZ cells were treated for 24 h with 0.7 mM of
the indicated FFA complexed to 4% (w/v) BSA, in presence or absence of TNF-or (O, 20 or 100 ng/ml). The
cytotoxicity was measured as the % LDH released, as defined in the methods section. Data presented as

mean :1: s.d. of three independent experiments. *, significant FFA effect, p<0.01, #, significant TNF-or
effect, p < 0.01.

129

. 2.00
1.80 -
1.60 -

 

g

 

1.20 ~ 0 " v
1.00 ~
0.80 4 l’
0.60 - '/“
0.40 ~ '
0.20 - ._ ‘L

J.

0000 T I 77 fi I I I I I I I I I
TNF-a-P 20 100 0 20 100 0 20 100 0 20 100 0 20 100

HepGZ BSA Palmitate Oleate Linoleate

J
I

 

 

 

 

 

 

 

Acetoacetate released (umollmg protein)

 

 

Figure 5.1b Profile of acetoacetate release by various treatments. The acetoacetate production in
response to various treatments. Confluent HepGZ cells were treated for 24 h with 0.7 mM of the indicated
FFA complexed to 4% (w/v) BSA, in presence or absence of TNF-or (0, 20 or 100 nyml). Acetoacetate

was measured by enzymatic assay. Data presented as mean :I: s.d. of three independent experiments. *,

significant F FA effect, p<0.01, #, significant TNF-or effect, p < 0.01.

5.3.2 Integrative Mixture Model Gene Ontology (IMGO) analysis

The genes which were affected by the treatments (FFA or TNF-or or their interaction)
were identified by two-way ANOVA analysis. In order to identify the genes which
contribute to the cytotoxicity of palmitate and TNF-or from the other treatments, the
IMGO analysis was performed on the differentially expressed genes. To achieve
separation of palmitate (toxic and ketogenic) conditions from the rest (non-toxic and non-
ketogenic), a two-population mixed model was employed. The discrimination between
the populations was calculated as the difference in the probablilities of the two models.
Because the discrimination by the model is sensitive to the two parameters 3. and r, the

effect of these parameters on the ability of the mixture model to separate the cytotoxic

130

and non-toxic conditions was studied. The results of this analysis are shown in Figure
5.2. For very small 7. (of 0.1), the separation increased with t. For higher 2., the
separation displayed maxima with ‘c = 3. This indicated that incorporation of prior
knowledge (1:) improves the discriminating ability of the IMGO, but as the uninforrnative
prior (A) is increased, only a small incorporation of ‘E is needed to improve the separation.
For higher values of 2., further increase of r weakens the discrimination of toxic vs. non-
toxic conditions. The optirnal values of 7. and r vary, depending upon the sub-populations
as well as the responses studied. Therefore, optimal values of these parameters can not be

identified a priori.

Separation

 

Figure 5.2 Variation of the discriminating ability of the mixture model with the smoothing constant
(7») and the weight to G0 (I). The probability of the model for the cytotoxic condition is plotted as a
frmction of the two parameters A and 1:.

131

 

1.2

Separation
9 o 9
b C! on .3

.°
N
I

 

 

 

 

o ‘ f I r I I I I I I

TNF-a-> 20 100 0 20 100 0 20 100 0 20 100 0 20 100
HepGZ BSA Palmitate Oleate Linoleate

 

 

 

Figure 5.3a Discrimination of the cytotoxic conditions from the non-toxic conditions for A = 0.5, t =
3. A two sub-population model was fit for cytotoxicity and the ability of model to distinguish the cytotoxic
(high LDH release) conditions was tested.

 

Separation
.° .° .° P P P P
0 «lb 0| G N on D d
1 J I i l I i

 

 

 

 

 

0.24
0.1 .
k# — J. _ k —/
o I I rI I f I . I I . - I - I
TNF-a-> 20 100 0 20 100 0 20 100 0 20 100 0 20 100
HepGZ BSA Palmitate Oleate Linoleate

Figure 5.3b Discrimination of the ketogenesis for A = 0.5, 1: = 3. A two sub-population model was fit for
cytotoxicity and the ability of model to distinguish the ketogenic (high acetoacetate release) conditions was
tested.

132

As a preliminary screen of IMGO, its discriminating ability to differentiate the cytotoxic
and metabolically distinct subpopulations was studied. The IMGO analysis was indeed
able to separate the cytotoxic conditions from the non-toxic ones. Figure 5.33 shows the
separation of the cytotoxic conditions for the parameter values of it = 0.3 and T = 3 (the
optimum value of r for I. = 0.3). The IMGO analysis was also able to distinguish the
conditions which had higher acetoacetate release, namely the palmitate-treated samples,
from the others (Figure 5.3b). It must be emphasized that the two responses of LDH and
acetoacetate release were simultaneously analyzed in a single run of the program. In fact,
multiple metabolic and phenotypic responses can be analyzed simultaneously using this

procedure, making it a truly parallel analysis method.

5.3.3 Identification of the genes responsible for separating the toxic vs. non-toxic
phenotype

In order to identify the genes that could discriminate between the toxic and non-toxic
phenotypes, a two-population mixture model was fit and the genes with the greatest
differences in their weights for the two models were identified. Detailed results of this
analysis are available in additional file 1. This analysis identified that lysosomal ATPase
V0 subunit a isoform 1 (ATP6VOA1, accession number NM_005177) and adenylate
cyclase 9 (AC9, accession number H64281) were the genes which had the greatest
differences in their weights for the two models. Another lysosomal ATPase, the 13kDa,
V1 subunit G isoform 1 (ATP6V1G1, accession number AA608567) also had very high
difference in weights for the two models. These genes also had the greatest individual

weights for the second mixture model (which fit the toxic phenotypes better), fiirther

133

emphasizing the important role of these proteins in the F FA-toxicity. The location of
these genes in the ranked list for greatest difference in weights for the two models is
shown in Table 5.1. While the two lysosomal ATPases had consistently higher ranks
irrespective of the separation achieved in the two mixture models (defined as probability
for the model which fit the cytotoxic condition), AC9 had very low rank when the
separation was poor (about 0.5), but its rank became very high (= 2) for higher separation

(0.6 or higher).

 

 

 

 

 

 

 

 

 

 

 

2. 1: Probability Rank (AC9) Rank (ATP6VOA1 ) Rank(ATP6V1G1)
0,1 0 0.5 N/A N/A N/A
3 0.5454 3 1 4
7 0.6696 2 1 4
10 0.9917 2 1 4
0.3 0 0.5 453 2 5
3 0.8991 2 1 4
7 0.8857 2 1 4
10 0.8186 2 1 5
0.5 0 0.5 455 2 5
3 0.9604 2 1 4
7 0.7963 2 1 5
10 0.7479 2 1 5
OJ 0 0.5 460 2 5
3 0.8835 2 1 4
7 0.7508 2 1 5
10 0.6914 2 1 5

 

 

Table 5.1 The ranks of the three genes, adenylate cyclase 9 (AC9) and lysosomal ATPases
(ATP6VOA1 and ATP6V1G1) as a function of 71. and I. Two subpopulation mixture model was fit to the

data using various values of the parameters )1. and 1:. The probability is defined as the probability of
associated with the model that fit the cytotoxic conditions better. The absolute differences of weights
assigned for a gene by the two models were ranked in decreasing order. The location of the three genes in

the ranked list was identified and is shown. All the genes had same weights for the two models when A =
0.1 and 1: =0, thus the differences were all 0 and ranking was not possible.

134

5.3.4 Experimental validation of lysosomal ATPase genes

In order to investigate the role of lysosomal ATPases in the cytotoxicity of palmitate, the
effect of Bafilomycin A1 (Bfl, an inhibitor specific for lysosomal ATPases) on the
cytotoxicity was studied. Treating cells with palmitate in the presence of 5 nM Bfl
significantly increased the toxicity of this fatty acid (Figure 5.4). Similar concentration of
Bfl was not toxic by itself. These results indicated that the functioning of lysosomal

ATPases is affected by palmitate and may mediate the toxicity of this FFA.

 

16

145

127

107

LDH released (%)
ee

 

 

       

o .
Ctrl Ctrl + Bfl Palm Palm + Bfl

Figure 5.4 Effect of inhibitor of lysosomal ATPase, bafilomycin (Bfl) on the toxicity of palmitate.
Cells were treated with 0.7 mM palmitate or control medium in the presence of 5 nM Bfl for 24 h and the
LDH release was measured. Data presented as mean +/- s.d. of six samples from two independent
experiments.

5.3.5 Experimental validation of adenylate cyclase
Caspase-3 is an effector/ terminal caspase, which mediates the changes leading to
apoptosis. Cells exposed to palmitate had about 7-fold higher greater caspase-3 activities

than controls. In order to identify the role of AC in the cytotoxicity of the FFAS, we

135

tested the effect of increasing the cAMP levels of the cells on the activation of caspase-3
by palmitate. Treating the cells with palmitate in the presence of IBMX and forskolin
(chemicals known to increase cAMP levels) further increased the caspase-3 activation in
response to palmitate (Figure 3b), but did not affect caspase-3 activation in controls (not
shown). In fact, after 24 h of exposure to palmitate + forskolin and IBMX, the caspase
activity was about twice than that observed in palmitate alone. This indicated that
elevated cAMP levels could worsen the apoptotic effects of palmitate, corroborating with

the model predictions of the importance of AC9 in the cytotoxicity of palmitate.

 

2.5

 

I Palm
I PL"?! CAMP

-| .0 N
0 0| 0
l l 1

Relative Caspase Activity

9
on
I

 

 

   

0.0 —
Day

Figure 5.5 Effect of increasing cAMP levels on the caspase-3 activation by palmitate. Cells were
treated with 0.7 mM palmitate or control medium in the presence of 5 nM Bfl for 48 h. Media were
replaced daily. The relative caspase-3 activity was measured. Data presented as mean +/- s.d. of nine
samples from three independent experiments.

136

5.4 Discussion

Incorporation of the functional information of genes in the microarray analysis is a new
and active area of research. However, most of the currently available methods utilize the
prior functional (GO) information to generate pre-defined sets of genes (Subramanian et
al., 2005 and Beisvag et al., 2006). The underlying assumption in these analyses is that a
phenotype is altered by concerted change in the expression of many genes of a G0-
category (Subramanian et al., 2005). The approach presented here is conceptually
somewhat similar to these approaches, but with major differences: (a) The similarity
matrix employed in the analysis is based on the overlap in the GO profiles among every
pair of genes in the data-set. Therefore, there are no strict gene sets and the possibility of
interaction/ coregulation of any pair of genes is incorporated as well as weighted, i.e.,
genes with greater overlap (even if they lie within a same GO-group) have greater
coefficient in the similarity matrix and vice versa. (b) The contribution of individual gene
(or maximum a pair of genes) in discriminating between the populations can be
identified. Many times the goal is to identify a few discriminating genes that could be
firrther tested. This technique offers the possibility of objectively identifying such genes
by suggesting the topmost discriminatory genes. (c) One can control the contribution of
the GO information in the model. In our study, the contribution of the similarity matrix
(GO information) is weighted by the factor I, which can be varied. This is in contrast to
other methods where the GO information takes precedence over the data and the
subsequent processing. The advantage of controlling the contribution of GO information
is best shown by Figure 3, where it is shown that even though GO improves the

discrimination, but for better discrimination, the contribution of GO must be regulated. It

/ 137

is likely that at higher values of 9., the constraints imposed by higher I adversely affect
the discrimination. In addition to these advantages, the model offers possibility of
discriminating multiple subpopulations, though this needs to be further tested and

validated.

The genes selected for the F FA-toxicity identified an important role of lysosomes in the
toxicity of palmitate to the HepGZ cells. Lysosomal ATPases regulate the import of H+
ions and regulate the firnction of these organelles. These ATPases have two domains, the
cytosolic V1 and the transmembrane V0. The V1 domain contains the ATP catalytic site.
ATP6VOA1 encodes one of three A subunit proteins and the encoded protein is
associated with clathrin-coated vesicles, while ATP6V1G1 encodes for one of three V1
domain G subunit proteins. The selection of these genes indicates an alteration in the
functioning of lysosomes. Indeed, a recent study has identified an important role of
lysosomal permeabilization in the toxicity of palmitate to HepG2 cells (Feldstein et al.,
2006). This study showed that in hepatocytes and HepG2 cells, exposure to palmitate
leads to translocation of Bax (a pro-apoptotic Bcl-2 family protein) to lysosomes, which
leads to the rupture of lysosomes. The exact mechanism of how Bax causes

permeabilization of lysosomes is not clear.

Another important target identified by the analysis is adenylate cyclase 9 (AC9). AC’s
are membrane bound enzymes that catalyses the formation of cyclic AMP from ATP and
are regulated by a family of G protein-coupled receptors, protein kinases, and calcium.

AC9 is stimulated by beta-adrenergic receptor activation but is insensitive to forskolin,

138

calcium, and somatostatin. It has previously been identified that the activity of adenylate
cyclase is regulated by palmitoylation (Mollner 1995), as is the activity of G-proteins
(Linder et al. 1993 and Degtyarev et al. 1993). It is likely that the elevated palmitate
levels lead to palmitoylation and greater activity of AC9, causing increased intracellular
cAMP levels, which in turn, would activate other signaling pathways by activating c-
AMP—dependent protein kinase (PKA). PKA would then phosphorylate and activate
membrane channels and contractile proteins, as well as affect transcription of various
genes by phosphorylating transcription factors, like C/EBP. Activation of AC and PKA
has been suggested to mediate cell death in response to stimuli such as ethanol (Maas et
al. 2005), morphine (Lim et a1. 2005) and pathogens (Hewlett et al. 2006 and Voth et al.
2005). On the contrary, cytoprotective role of this pathway has been indicated in anoikis
(Joseph et al. 2005), spinal-injury (Chen and Tzeng 2005), CD-34+ cytotoxicity

(N egretto 2006) and myocardial infarction (Takahashi et al. 2006, etc.).

Another gene that had consistently high weight was nuclear factor of kappa light
polypeptide gene enhancer in B-cells inhibitor, alpha (NF KBIA). NFKBIA encodes for
the inhibitor of NF-KB (IKB-alpha). The activity of NF-KB is regulated through IKB-
alpha. NF-KB is an important transcription factor that regulates the expression of a
variety of genes, such as those involved in the regenerative (Taub, 1996), anti-apoptotic
(Schuchmann and Galle, 2002 and Bradham et al., 1998) and anti-oxidant (Huang et al.,
2001) response of the cells. In addition, the rapid activation of NF-KB has been suggested
to protect the hepatocytes and hepatoma cells from the cytotoxic action of TNF-or

(Bradham et al., 1998). Based on these roles of NF-KBIA, it should have a positive

139

association with the cytotoxicity. Indeed, the regression coefficient of this gene to

cytotoxicity was positive.

Other genes which are known to play important role in cellular physiology were also
identified to be very important in relation to the FFA toxicity. It was identified that
NADH dehydrogenase (ubiquinone) Fe-S protein 1 (NDUFSI) was the most important
protein, especially for conditions with small separation of the models (i.e., both model
had very similar weights and so were almost equally good). Our previous study has
identified the important role that NADH dehydrogenases play in regulating the FF A-
toxicity (Srivastava and Chan, 2007). Another gene that was identified as important for
certain conditions was cytochrome P450, subfamily 46 (cholesterol 24-hydroxylase)
(CYP46). CYP46 is involved in the synthesis of cholesterol metabolite 24S-
hydroxycholesterol. Alterations in cholesterol metabolism are associated with various
disorders, and may even lead to cell death (Tabas 2002 and Cutler et al. 2002). Some
conditions also identified chloride channel 4 (CLCN4 / CIC4). CIC4 exhibits chloride-
coupled proton transport activity and transport a significant amount of protons across the
plasma membrane, even against the electrochemical gradient, when activated by positive
voltages (Picollo and Pusch 2005). A redox-regulation of its activity has been suggested
(Mazzanti et al. 2005). However, the role played by this gene in the FFA-toxicity is

currently not clear.

Thus, the IMGO analysis identified proteins that play important role in the toxicity of

FFAS. While some of these proteins have already been shown to play important roles in

140

the toxicity of FFAS, others have been indicated to play important roles in cellular
physiology and thus could contribute to the FFA. Our results indicate the IMGO is a
useful approach to incorporate ontology information in the selection process to identify

important genes that regulate certain firnctions of interest.

141

CHAPTER 6. CONCLUSIONS AND FUTURE DIRECTIONS

6.1 Conclusions

The objectives of this research were to investigate the physiological, metabolic and
genomic alterations caused by FFA, more specifically saturated FFA, which were
associated with its toxicity. This study has revealed complex changes in multiple cellular
levels on exposure to saturated F FA. It was shown that the toxicity was caused by
excessive production of reactive oxygen species (ROS), specially hydrogen peroxide and
hydroxyl radicals. It was also shown that mitochondrial complexes I and III were the
primary sources of the cytotoxic ROS. This information may help in the development of
novel treatment strategies which employ scavengers specific to these radicals, or
inhibitors that reduce ROS generation at the mitochondrial complexes. Such scavengers/
inhibitors may be more effective in protecting liver against the cytotoxic action of FFAS.
The saturated FFAS not only lead to greater ROS generation, but also reduce cellular
antioxidant ability by reducing the levels of cystine transporter, which leads to reduced
synthesis of GSH, the primary anti-oxidant molecule. Thus, the saturated FFA causes
deleterious alterations at multiple levels of cellular redox balance, which perhaps
amplifies its cytotoxic potential. In addition to ROS, various signaling pathways play
important roles in the alterations caused by palmitate. In particular, the roles of JNK and

ERK in the toxicity of these fatty acids were identified.

Chronic exposure to FFAS is known to be associated with development of insulin
resistance as well as lipotoxicity. This study suggests that saturated FFAs cause much

greater lipotoxicity than the unsaturated FFAS. Additionally, the gene-expression data

142

suggests that the genes with known functions in inhibiting insulin signaling, e. g.,
INPPLl, are upregulated only in the saturated FFA. Thus, it is likely that saturated F FAs
also inhibit insulin signaling to a greater extent than the unsaturated FF As in hepatocytes,

as has been shown in other types of cells.

The studies in this thesis have indicated that integration of multi-source high-throughput
information may help identify novel targets that could be more potent in regulating the
progression of metabolic disorders. A case in point is of glutamate-cysteine synthetase/
ligase (GCS/L). The GA/PLS analysis for TG-only had identified GCS as an important
gene. However, GCS was not selected by the dual-response analysis, suggesting that it
was not important to regulate TG and LDH. In fact, in chapter 2 it was identified that
GCS was elevated in palmitate-treated cells and thus, would not be a limitation for the
glutathione synthesis. Therefore, overexpression of GCS is unlikely to improve
palmitate-toxicity. The integrative analyses could help quell such false-positives and
suggest more robust targets than individual analyses. Finally, it was shown that
incorporation of the prior (gene ontology) information of genes could lead to better
discrimination among the sub-populations and can help identify targets with important

roles in the underlying cellular processes.

6.2 Future Directions
The studies presented in this thesis have provided information on multiple alterations by
FFAS and identified some which play important role in lipotoxicity. Greater insights may

be achieved by performing some of the following studies.

143

 

6.2.1 In vivo studies

The ‘acid-tests’ of the findings of this study would be the application of the knowledge
learned in an in vivo system. For example, the effects of specific hydroxyl radical or
peroxide scavengers, or the effects of partial inhibition of complexes I and HI on the
development of lipotoxicity in an animal model may be tested and compared to those of
superoxide-specific scavengers. Similarly, the effects of ERK inhibitors or ERK-null
rodents on a high saturated fat diet should be investigated. Because ERK and JNK can
also regulate the expression of multiple genes, the roles played by these in regulating the
responses of interest could identify novel mechanisms of alterations by F FAs. The
mechanisms behind greater TG accumulation in response to the JNK inhibitor requires
further investigation. Because JNK has been touted as a potent anti-obesity target, such
studies are important to differentiate the whole-body-level and individual-cell-level

effects of this protein.

6.2.2 Insulin signaling in response to different F FAs

While some amount of literature is available on the effect of different types of FFAS on
insulin signaling in beta cells and myocytes, similar studies in hepatocytes are lacking.
The effects of the different types of FFAS on insulin signaling in hepatocytes should be
studied to identify if there are differences in the effects of different types of FFAS on
insulin signaling. Such studies would decipher the association between lipotoxicity and

insulin signaling. If the saturated FFAs emerge as having greater deleterious effect on

144

insulin signaling as well, they could become the primary culprits among the F FAs which

cause various metabolic disorders associated with obesity.

6.2.3 Application of the frameworks developed to other diseases/ data sets

The integrative frameworks developed in chapters 4 and 5 should be applied and tested in
other diseases and compared to available methods. Such analyses would help to
generalize the validity of the approaches as well as to find robust targets to regulate a

cellular response of interest.

I believe that the information, data and frameworks generated in the current thesis would

help advance the field of obesity and lipotoxicity research and have the potential to be

applied to other diseases.

145

 

APPENDIX

1. Table of metabolic fluxes for day 1

 

 

 

 

146

 

 

 

Rxn
No. Equation BSA Palm Ole
1 G+ATP -> G—6-P 27.58 1.82 27.84 1.36 26.61 0.77
2 G-6-P -> F-6-P 28.40 1.40 28.07 1.30 26.99 0.75
F—6-P +ATP -> Gcheraldehyde-3—P +
3 DHAP 28.40 1.40 28.07 1.30 26.99 0.75
4 DHAP -> Glyceraldehyde-3-P 28.40 1.40 28.07 1.30 27.00 0.75
5 Glyceraldehyde-3-P --> 3-PGA 57.83 2.71 56.64 2.51 54.80 1.46
6 3PGA -> PEP + NADH (+ 2ATP) 57.83 2.71 56.64 2.51 54.80 1.46
7 PEP -> Pyr 57.83 2.71 56.64 2.51 54.80 1.46
8 Pyruvate + NADH -> (Lactate) 53.13 3.43 52.81 4.50 50.89 2.81
9 Pyruvate -> Acetyl-CoA + NADH +C02 3.09 3.40 3.07 4.69 2.06 2.90
10 Acetyl-CoA + 0AA -> Citrate 8.67 2.54 8.03 1.79 7.92 2.24
Citrate —> a-KetoGlutarate + NADPH
1 1 +002 8.55 2.54 7.94 1.79 7.74 2.24
a-Ketoglutarate -> SuccinyI-CoA +
12 NADH + 002 9.01 2.57 8.28 1.79 8.48 2.24
Succinyl-CoA -> Fumarate + FADH2 +
13 ATP 9.29 2.54 8.52 1.77 8.70 2.23
14 Fumarate -> 0AA + NADH 8.80 2.54 8.11 1.79 8.11 2.24
15 G-6-P -> 12 NADPH + 6 CO2 -0.81 0.77 -0.23 0.19 -0.38 0.15
16 COZ out 16.35 4.69 17.98 5.03 16.72 4.71
17 2 AcetyI-COA -> AcetoacetyI-CoA -1.03 0.84 -0.39 0.69 -0.68 0.37
18 Acetoacetyl-CoA -> Acetoacetate 0.17 0.21 0.29 0.15 0.29 0.10
19 Acac out 0.05 0.04 0.09 0.03 0.07 0.02
Acetoacetate + NADH -> (B-OH
20 butyrate) 0.03 0.01 0.13 0.05 0.08 0.02
21 02 (In) 20.75 3.29 22.15 3.79 21.83 5.38
22 NADH + 0.5 02 --> 2.5 ATP 32.87 5.35 34.76 6.28 35.47 8.86
23 FADH2 + 0.5 02 -> 2 ATP 10.28 2.54 9.61 1.80 9.74 2.24
24 Glycerol + ATP --> Gcherol-3—P 0.75 0.06 0.47 0.05 0.80 0.06
Glycerol-3-P --> Gcheraldehyde-3—P +
25 NADH 1.03 0.28 0.49 0.29 0.81 0.20
26 Palm-CoA --> 8 AcetyI-CoA + 14 NADH -0.32 0.50 0.42 0.52 0.39 0.32
27 Palm-CoA + DAG -> TG 0.00 0.00 0.03 0.01 0.04 0.02
28 Ser (In) 0.50 0.12 0.50 0.18 0.48 0.05
29 Ser -> NH3 + Pyr 0.27 0.41 0.48 0.28 -0.01 0.11
30 Gln In 2.43 0.30 2.23 0.10 2.48 0.11
31 His --> Glu + NH4 0.42 0.09 0.29 0.08 0.48 0.07
32 Asp (In) -0.10 0.05 -0.11 0.02 0.00 0.00
33 Glu (In) -0.83 0.43 -0.68 0.29 -0.63 0.09
34 Gly (In) 0.09 0.29 -0.17 0.20 0.72 0.07
35 Glycine -> 2 002 + NADH + NH3 -0.40 0.29 -0.52 0.20 0.06 0.08
36 NH4 (In) -2.54 0.82 -2.53 0.54 -2.47 0.21
37 Arg In 0.55 0.09 0.39 0.08 0.35 0.03
38 Thr (In) 0.12 0.21 0.19 0.14 0.18 0.07
39 Ala (In) -1.89 0.82 -1.31 0.45 -1.85 0.28

 

 

40
41
42
43

45
46
47
48
49
50
51

52

53

55
56

57
58

59
60

61
62
63

65
66

67
68
69
70

71
72

73
74

75
76

 

Glu + Pyr --> Ala + aKG

Pro In

Cys In

Tyr (In)

Tyr + aKG + 2 02 --> Glu + 002 +
Acetoacetate + Fumarate

Val (In)

Om IN

Lys (In)

lle (IN)

Leu (In)

Phe (In)

Glu + Cys + Gly -> GSH

HCO3— + NH4 + Om + 2 ATP ->
Citrulline

Citrulline + Asp + ATP -> Fumarate +
Arg

Arg --> Om + Urea

Urea Out

Gln --> Glu + NH4

Om + a-KG + 0.5 NADPH + 0.5 NADH —
> Pro

Asp + NH4 -> Asn

Thr -> Pyr + COZ + NH4 + 2 NADH +
FADH2

Val + aKG —> Glu + CO2 + 2NADH +
FADH2 + Succ-CoA

Lys + 2 aKG + NADPH -> 2Glu +
AcetoacetyI-CoA + 2002 + 4 NADH +
FADH2

Ile + aKG -> Glu + Succ-CoA + Acetyl-
CoA + NADH + FADH2

Leu + aKG -> Glu + HMG-CoA + NADH
+ FADH2

Phe + 02 --> Tyr

Ser + 2 Palm-CoA + NADPH --> C02
+Ceramide + FADH2

Ceramide + Phosphatidleholine -->
Sphingomyelin

AcetoacetyI-CoA + Acetyl-CoA -> HMG—
CoA

HMG-CoA + 2 NADPH (+ 3ATP) -> IPP
2 IPP -> Geranyl-PP

Geranyl-PP + IPP -> Famesyl-PP

2 FamesyI-PP + 0.5 NADPH + 0.5 NADI-I
-> Squalene

Squalene + 02 + NADPH --> Lanosterol
Lanosterol + 10.5 NADPH + 4.5 NADH +
10 02 --> Chol + 3 C02

Cholesterol Ester -> Chol + FA-CoA

2 Palm-CoA + Glycerol-3-P ->
Phosphatidate

Phosphatidate —> DAG

147

 

1.94
-0.42
0.19
0.08

-0.08
0.55
-0.03
0.33
0.20
0.30
0.18
0.43

-0.18

-0.29
0.22
0.22
2.20

0.31
0.12

0.06
0.37

0.23
0.03

0.25
0.01

0.00
0.01

-0.96
-0.71
-0.24
-0.23

-0.11
-0.08

-0.06
0.04

-0.18
-0.10

0.75
0.16
0.01
0.17

0.20
0.32
0.03
0.33
0.20
0.17
0.05
0.10

0.21

0.13
0.16
0.16
0.27

0.27
0.06

0.21
0.30

0.27
0.20

0.18
0.09

0.08
0.04

0.72
0.68
0.23
0.22

0.10
0.08

0.06
0.03

0.20
0.11

 

1.41
—0.51
0.15
-0.07

-0.06
0.25
-0.05
0.34
0.45
0.55
0.37
0.35

-0.16

-0.27
0.01
0.01
2.04

0.02
0.08

0.18
0.06

0.13
0.26

0.45
0.19

0.06
0.00

-0.55
-0.10
-0.03
-0.03

-0.02
-0.01

-0.01
0.01

0.00
0.01

0.41
0.11
0.02
0.11

0.14
0.24
0.02
0.24
0.18
0.16
0.07
0.07

0.14

0.09
0.12
0.12
0.10

0.18
0.04

0.14

0.23

0.18
0.17

0.16
0.08

0.01
0.03

0.62
0.59
0.20
0.19

0.09
0.07

0.05
0.03

0.21
0.08

 

1.92
0.03
0.17
0.16

-0.14
0.45
0.04
0.31
0.48
0.41
0.22
0.54

-0.07

-0.26
0.02
0.02
2.10

0.06
0.19

0.06
0.19

0.16
0.22

0.34
-0.04

0.05
0.02

-0.80
-0.46
-0.16
-0.15

-0.07
-0.06

-0.04
0.02

0.06
-0.10

0.25
0.05
0.02
0.09

0.09
0.16
0.01
0.09
0.09
0.08
0.03
0.04

0.06
0.04
0.05
0.05
0.10

0.09
0.02

0.07
0.14

0.08
0.08

0.08
0.04

0.11
0.01
0.34
0.32
0.11
0.11

0.05
0.04

0.03
0.02

0.14
0.04

 

 

77 Phosphatidate --> CDP-DAG -0.06 0.17 -0.01 0.16 0.18 0.11

78 CDP-DAG + Ser -> PhosphatidylSerine 0.12 0.22 0.01 0.15 0.31 0.07
PhosphatidylSerine ->

79 PhosphatidyIEthanolamine + 002 0.09 0.15 0.01 0.10 0.21 0.05
CDP-DAG + G-3—P -->

80 PhosphatidylGcherol -0.15 0.14 -0.02 0.13 -0.1 1 0.08

81 2 PG --> Cardiolipin + Glycerol -0.06 0.06 -0.01 0.05 -0.04 0.03
DAG + GDP-Choline ->

82 PhosphatidyICholine -0.01 0.02 0.00 0.02 0.00 0.01
DAG + CDP—Ethanolamine ->

83 PhosphatidyIEthanolamine -0.06 0.09 -0.01 0.06 -0.12 0.03

84 Glu + 2 NADPH --> Pro 0.28 0.35 0.67 0.23 0.17 0.11

85 Palm (In) ND. ND. 0.44 0.11 0.57 0.03

 

 

 

 

 

148

 

2. Values of the metabolic fluxes for day 2

 

 

 

 

149

 

 

 

Rxn
No. Equation BSA Palm Ole
1 G+ATP --> G-6-P 27.10 1.13 26.68 2.27 26.86 2.75
2 G-6-P -> F-6-P 27.94 0.98 26.99 2.16 27.19 2.63
F-6-P +ATP --> Glyceraldehyde-B-P +
3 DHAP 27.94 0.98 27.00 2.16 27.19 2.63
4 DHAP -> Glyceraldehyde-3-P 27.94 0.98 27.00 2.16 27.19 2.63
5 GcheraIdehyde-3-P -> 3-PGA 56.89 2.00 54.59 4.15 55.05 5.01
6 3PGA -> PEP + NADH (+ 2ATP) 56.89 2.00 54.59 4.15 55.05 5.01
7 PEP -> Pyr 56.89 2.00 54.60 4.15 55.05 5.01
8 Pyruvate + NADH —> (Lactate) 51.20 3.47 50.11 5.33 50.64 5.51
9 Pyruvate -> AcetyI-CoA + NADH +C02 4.23 2.85 3.20 6.29 2.74 7.01
10 AcetyI-CoA + OAA -> Citrate 9.99 2.44 7.03 2.78 7.83 2.00
Citrate -> a-KetoGIutarate + NADPH
11 +002 9.87 2.44 6.92 2.77 7.66 2.00
a-Ketoglutarate -> SuccinyI-CoA +
12 NADH + C02 10.24 2.47 7.50 2.97 8.28 2.02
SuccinyI-CoA -> Fumarate + FADH2 +
13 ATP 10.56 2.44 7.58 2.86 8.56 2.01
14 Fumarate -> OAA + NADH 10.11 2.44 7.14 2.79 7.99 2.00
15 G-6-P -> 12 NADPH + 6 C02 -0.83 0.66 -0.31 0.30 -0.32 0.23
16 C02 out 19.56 4.30 15.52 7.12 16.82 7.00
17 2 Acetyl-COA -> AcetoacetyI-CoA -1.01 0.69 0.26 1.11 -0.74 0.85
18 AcetoacetyI-CoA -> Acetoacetate 0.27 0.20 1.57 0.40 0.38 0.43
19 Acac out 0.06 0.04 0.68 0.11 0.14 0.02
Acetoacetate + NADH -> (B-OH
20 butyrate) 0.05 0.02 0.79 0.22 0.04 0.04
21 02 (In) 23.78 3.69 19.48 5.89 20.91 4.15
22 NADH + 0.5 02 -> 2.5 ATP 38.24 5.92 32.10 9.86 34.26 6.99
23 FADH2 + 0.5 02 --> 2 ATP 11.59 2.45 8.51 2.87 9.63 2.04
24 Glycerol + ATP --> Glycerol-3-P 0.71 0.08 0.44 0.06 0.66 0.05
GcheroI-3-P -> GcheraIdehyde-3-P +
25 NADH 1.01 0.24 0.60 0.41 0.67 0.40
26 Palm-CoA -> 8 AcetyI-CoA + 14 NADH -0.32 0.42 0.37 0.72 0.28 0.69
27 Palm-CoA + DAG --> TG 0.00 0.00 0.04 0.01 0.07 0.02
28 Ser (In) 0.45 0.11 0.52 0.61 0.55 0.08
29 Ser -> NH3 + Pyr 0.16 0.33 0.04 0.71 0.02 0.29
30 Gln In 2.65 0.40 2.68 1.32 2.64 0.34
31 His --> Glu + NH4 0.35 0.07 0.30 0.14 0.43 0.01
32 Asp (In) 0.00 0.01 -0.11 0.02 0.00 0.01
33 Glu (In) -0.73 0.22 -0.78 0.55 —0.59 0.32
34 Gly (In) 0.08 0.17 -0.15 0.22 0.11 0.13
35 Glycine -> 2 C02 + NADH + NH3 -0.46 0.20 -0.63 0.40 -0.56 0.18
36 NH4 (In) -2.31 0.61 -2.03 0.71 -1.87 0.64
37 Arg In 0.50 0.16 0.25 0.11 0.29 0.13
38 Thr (In) 0.13 0.17 0.17 0.57 0.15 0.12
39 Ala (In) -1.61 0.68 -1.34 0.73 -1.64 0.63
40 Glu + Pyr --> Ala + aKG 1.67 0.63 1.40 0.67 1.72 0.58
41 Pro In -0.66 0.31 -0.53 0.41 -0.30 0.36

 

 

 

42
43

45
46
47
48
49
50
51

52

53

55

57
58

59

60

61
62
63

65
66

67
68
69
70

71
72

73
74

75
76
77
78

 

Cys In

Tyr (In)

Tyr + aKG + 2 02 --> Glu + C02 +
Acetoacetate + Fumarate

Val (In)

Om IN

Lys (In)

Ile (IN)

Leu (In)

Phe (In)

Glu + Cys + Gly --> GSH

HCO3- + NH4 + Om + 2 ATP ->
Citrulline

Citrulline + Asp + ATP -> Fumarate +
Arg

Arg -> Om + Urea

Urea Out

Gln —> GIu + NH4

Om + a-KG + 0.5 NADPH + 0.5 NADH -—
> Pro

Asp + NH4 --> Asn

Thr -> Pyr + 002 + NH4 + 2 NADH +
FADH2

Val + aKG --> Glu + COZ + 2NADH +
FADH2 + Succ-CoA

Lys + 2 aKG + NADPH -> 2G|u +
AcetoacetyI-CoA + 2002 + 4 NADH +
FADH2

Ile + aKG -> Glu + Succ-CoA + Acetyl-
CoA + NADH + FADH2

Leu + aKG --> Glu + HMG-CoA + NADH
+ FADH2

Phe + 02 —-> Tyr

Ser + 2 PaIm-CoA + NADPH -> C02
+Ceramide + FADH2

Ceramide + PhosphatidlehoIine ->
Sphingomyelin

Acetoacetyl-CoA + AcetyI-CoA --> HMG-
CoA

HMG-CoA + 2 NADPH (+ 3ATP) --> IPP
2 IPP --> Geranyl-PP

GeranyI-PP + IPP -> Famesyl-PP

2 FamesyI-PP + 0.5 NADPH + 0.5
NADH -> Squalene

Squalene + 02 + NADPH -> Lanosterol
Lanosterol + 10.5 NADPH + 4.5 NADH +
10 02 -> Chol + 3 COZ

Cholesterol Ester --> Chol + FA-CoA

2 Palm-CoA + Glycerol-3-P -->
Phosphatidate

Phosphatidate —> DAG

Phosphatidate -> CDP-DAG

CDP-DAG + Ser --> PhosphatidylSerine

150

 

0.20
0.07

-0.16
0.53
0.00
0.36
0.29
0.31
0.14
0.46

-0.03

-0.17
0.26
0.26
2.38

0.21
0.12

0.06

0.34

0.22
0.10

_ 0.25
-0.04

0.06
0.02

-1.06
-0.81
-0.27
-0.26

-0.12
-0.10

-0.07
0.04

-0.19
-0.11
-0.05

0.15

0.00
0.12

0.19
0.27
0.02
0.29
0.24
0.24
0.11
0.09

0.17

0.10
0.19
0.19
0.34

0.26
0.05

0.17

0.25

0.24
0.23

0.23
0.12

0.09
0.03

0.61
0.56
0.19
0.18

0.09
0.07

0.05
0.03

0.17
0.09
0.14
0.17

 

0.11
-0.04

-0.10
0.13
-0.03
0.23
0.39
0.48
0.26
0.38

-0.06

-0.23
-0.04
-0.04

2.41

-0.05
0.11

0.07

-0.03

0.13
0.22

0.43
0.10

0.12
0.04

-1.18
-0.75
-0.25
-0.24

-0.11
-0.09

-0.06
0.04

-0.05
-0.09
0.07
0.26

0.03
0.22

0.31
0.22
0.03
0.24
0.18
0.18
0.10
0.22

0.24

0.14
0.21
0.21
1.13

0.31
0.09

0.53
0.23

0.22
0.21

0.19
0.15

0.18
0.07

1.09
1.07
0.36
0.35

0.16
0.13

0.09
0.06

0.29
0.17
0.25
0.36

 

0.17
0.14

-0.20
0.52
0.02
0.35
0.41
0.42
0.14
0.54

0.01

-0.20
0.01
0.01
2.27

-0.05
0.17

0.02
0.28

0.19
0.17

0.35
-0.10

0.06
0.02

-0.92
-0.57
-0.19
-0.18

-0.09
-0.07

-0.05
0.03

0.06
-0.09
0.18
0.34

0.02
0.49

0.43
0.16
0.01
0.20
0.13
0.12
0.06
0.09

0.16

0.11
0.18
0.18
0.30

0.22
0.05

0.13

0.16

0.18
0.14

0.13
0.09

0.04
0.02

0.74
0.71
0.24
0.23

0.11
0.08

0.06
0.04

0.30
0.10
0.24
0.19

 

 

79

80
81

82
83

85

 

PhosphatidylSerine ->
PhosphatidylEthanolamine + C02
CDP-DAG + G-3—P ->
PhosphatidyIGlycerol

2 PG -> Cardiolipin + Glycerol
DAG + CDP-Choline ->
PhosphatidyICholine

DAG + CDP-Ethanolamine ->
PhosphatidylEthanolamine
Glu + 2 NADPH -> Pro

Palm (In)

 

0.11

-0.17
-0.07

-0.01

-0.07
0.64
N.D.

0.12

0.12
0.05

0.02

0.07
0.39
MB.

 

0.18

-0.16
-0.06

0.01

-0.11
0.75
0.49

0.25

0.23
0.09

0.03

0.14
0.53
0.10

 

0.23

-0.13
-0.05

0.00

-0.13
0.59
0.56

0.13

0.17
0.07

0.02

0.07
0.39
0.03

 

151

 

 

3. Values of the metabolic fluxes for day 3

 

 

 

Rxn
No. Equation BSA Palm Ole
1 G+ATP -> G-6-P 26.81 1.18 26.40 6.06 28.18 1.25
2 G-6-P -> F-6-P 27.43 0.95 26.73 5.74 28.79 1.21
F—6-P +ATP -> GcheraIdehyde—3-P +
3 DHAP 27.43 0.95 26.73 5.74 28.79 1.21
4 DHAP -> Glyceraldehyde-B-P 27.44 0.95 26.73 5.74 28.80 1.21
5 Glyceraldehyde-3P -> 3-PGA 55.67 1.87 54.06 10.89 58.26 2.32
6 3PGA -> PEP + NADH (+ 2ATP) 55.67 1.87 54.06 10.89 58.26 2.32
7 PEP -> Pyr 55.67 1.87 54.06 10.89 58.26 2.32
8 Pyruvate + NADH —-> (Lactate) 49.58 2.68 48.47 7.49 49.89 2.81
9 Pyruvate -> AcetyI-CoA + NADH +C02 4.60 2.42 4.78 12.66 6.64 3.45
10 AcetyI-CoA + OAA -—> Citrate 8.41 2.21 8.05 3.17 9.16 2.07
Citrate —> a-KetoGlutarate + NADPH
11 +002 8.29 2.20 7.95 3.17 8.98 2.07
a-Ketoglutarate -> SuccinyI-CoA +
12 NADH + COZ 8.85 2.23 8.56 3.17 9.60 2.07
SuccinyI-CoA -> Fumarate + FADH2 +
13 ATP 9.00 2.21 8.67 3.16 9.84 2.07
14 Fumarate --> OAA + NADH 8.53 2.21 8.14 3.18 9.34 2.07
15 G-6-P -> 12 NADPH + 6 C02 -0.62 0.57 -0.32 0.41 -0.60 0.17
16 CO2 out 18.56 3.81 19.02 12.01 21.87 4.70
17 2 Acetyl-COA -> Acetoacetyl-CoA —0.75 0.65 0.15 1.51 -1.89 0.62
18 AcetoacetyI-CoA -> Acetoacetate 0.23 0.10 1.19 0.47 0.21 0.28
19 Acac out 0.04 0.01 0.45 0.39 0.03 0.07
Acetoacetate + NADH -> (B-OH
20 butyrate) 0.05 0.02 0.50 0.14 0.09 0.07
21 02 (In) 21.50 3.45 22.05 5.67 21.83 4.83
22 NADH + 0.5 02 -> 2.5 ATP 34.81 5.52 36.68 9.40 36.16 8.01
23 FADH2 + 0.5 02 -> 2 ATP 9.85 2.19 9.53 3.15 10.89 2.09
24 Glycerol + ATP --> Gcherol-3-P 0.60 0.04 0.49 0.41 0.63 0.19
GcheroI-3-P -> Glyceraldehydec3-P +
25 NADH 0.79 0.21 0.59 0.83 0.66 0.27
26 Palm-CoA --> 8 Acetyl-CoA + 14 NADH -0.20 0.37 0.31 1.36 -0.38 0.37
27 Palm-CoA + DAG -—> T6 0.00 0.00 0.04 0.01 0.64 0.12
28 Ser (In) 0.45 0.11 0.59 0.32 0.49 0.06
29 Ser -> NH3 + Pyr 0.26 0.40 0.14 0.41 0.08 0.24
30 Gln In 2.34 0.47 2.63 0.37 2.66 0.13
31 His -> Glu + NH4 0.41 0.08 0.24 0.13 0.39 0.05
32 Asp (In) -0.01 0.01 -0.11 0.02 0.00 0.01
33 Glu (In) -0.73 0.24 -0.63 0.43 —0.59 0.28
34 Gly (In) 0.05 0.07 -0.16 0.22 0.58 0.10
35 Glycine --> 2 C02 + NADH + NH3 -0.39 0.19 -0.63 0.24 —0.11 0.14
36 NH4 (In) -2.48 0.81 -2.08 0.60 -2.33 0.55
37 Arg In 0.41 0.11 0.32 0.09 0.32 0.06
38 Thr (In) 0.15 0.18 0.24 0.38 0.14 0.10
39 Ala (In) -1.83 0.26 -1.01 0.58 -1.75 0.62
40 Glu + Pyr —> Na + aKG 1.85 0.24 1.10 0.53 1.84 0.57
41 Pro In -0.46 0.07 -0.51 0.20 -0.16 0.17

 

152

 

 

 

 

 

 

42
43

45
46
47
48
49
50
51

52
53

55

57
58

59

60

61
62
63

65
66

67
68
69
70

71
72

73
74

75
76
77
78

 

Cys In

Tyr (In)

Tyr + aKG + 2 02 -> Glu + 002 +
Acetoacetate + Fumarate

Val (In)

Om IN

Lys (In)

Ile (IN)

Leu (In)

Phe (In)

Glu + Cys + Gly —> GSH

HCO3- + NH4 + Om + 2 ATP ->
Citrulline

Citrulline + Asp + ATP --> Fumarate +
Arg

Arg -> Om + Urea

Urea Out

Gln -> GIu + NH4

Om + a-KG + 0.5 NADPH + 0.5 NADH —
> Pro

Asp + NH4 -> Asn

Thr -> Pyr + C02 + NH4 + 2 NADH +
FADH2

Val + aKG -> Glu + C02 + 2NADH +
FADH2 + Succ-CoA

Lys + 2 aKG + NADPH -> 2Glu +
Acetoacetyl-CoA + 2C02 + 4 NADH +
FADH2

Ile + aKG -> Glu + Succ-CoA + Acetyl-
CoA + NADH + FADH2

Leu + aKG -> Glu + HMG-CoA + NADH
+ FADH2

Phe + 02 —-> Tyr

Ser + 2 Palm-CoA + NADPH --> 002
+Ceramide + FADH2

Ceramide + Phosphatidleholine ->
Sphingomyelin

AcetoacetyI-CoA + AcetyI-CoA -> HMG-
CoA

HMG-CoA + 2 NADPH (+ 3ATP) -> IPP
2 IPP -> GeranyI-PP

GeranyI-PP + IPP -> FamesyI-PP

2 FamesyI-PP + 0.5 NADPH + 0.5 NADH
-> Squalene

Squalene + 02 + NADPH --> Lanosterol
Lanosterol + 10.5 NADPH + 4.5 NADH +
10 02 -> Chol + 3 C02

Cholesterol Ester —> Chol + FA—CoA

2 Palm-CoA + Glycerol-3-P ->
Phosphatidate

Phosphatidate -> DAG

Phosphatidate -> CDP-DAG

CDP-DAG + Ser -> PhosphatidylSerine

153

 

0.19
0.06

-0.14
0.41
0.01
0.34
0.15
0.17
0.09
0.39

-0.13

-0.21
0.18
0.18
2.14

0.30
0.12

0.10

0.27

0.30
0.00

0.15
-0.06

0.04
0.01

-0.68
-0.53
-0.18
-0.17

-0.08
-0.06

-0.04
0.03

-0.12
-0.07
-0.03

0.10

0.01
0.04

0.10
0.01
0.00
0.09
0.06
0.05
0.03
0.10

0.14

0.08
0.14
0.14
0.40

0.14
0.03

0.18
0.04

0.09
0.07

0.06
0.05

0.10
0.04

0.61
0.59
0.20
0.19

0.09
0.07

0.05
0.03

0.15
0.10
0.13
0.21

 

0.09
-0.05

-0.23
0.21
-0.04
0.27
0.35
0.39
0.18
0.37

—0.02

-0.20
0.03
0.03
2.35

-0.08
0.09

0.15
0.03

0.10
0.17

0.30
0.00

0.11
0.04

-0.94
-0.63
-0.21
-0.20

-0.10
-0.07

-0.05
0.03

-0.02
-0.07
0.08
0.24

0.03
0.20

0.22
0.24
0.03
0.24
0.18
0.16
0.07
0.10

0.17

0.11
0.15
0.15
0.32

0.21
0.05

0.35

0.23

0.20
0.18

0.17
0.10

0.12
0.05

1.41
1.38
0.46
0.45

0.21
0.16

0.12
0.07

0.56
0.15
0.40
0.21

 

0.15
0.24

-0.09
0.47
0.02
0.37
0.48
0.48
0.19
0.55

-0.01

-0.23
0.00
0.00
2.26

-0.07
0.18

0.01

0.20

0.19
0.21

0.40
-0.07

0.03
0.00

-1.90
-1.50
-0.50
-0.49

-0.23
-0.18

-0.13
0.08

0.18
0.42
-0.17
0.25

0.00
0.30

0.27
0.16
0.01
0.10
0.11
0.10
0.04
0.06

0.14

0.09
0.11
0.11
0.12

0.17
0.04

0.11

0.15

0.13
0.11

0.11
0.07

0.03
0.01

0.51
0.49
0.16
0.16

0.08
0.06

0.04
0.03

0.16
0.12
0.17
0.15

 

 

PhosphatidylSerine -->

79 PhosphatidylEthanolamine + C02 0.08 0.15 0.17 0.14 0.19 0.10
CDP-DAG + G—3—P -->

80 PhosphatidyIGlycerol -0.1 1 0.13 -0.13 0.29 -0.35 0.11

81 2 PG -> Cardiolipin + Glycerol -0.04 0.05 -0.05 0.12 -0.14 0.05
DAG + CDP-Choline ->

82 PhosphatidyIChoIine 0.00 0.01 0.01 0.05 -0.03 0.01
DAG + CDP-Ethanolamine -->

83 PhosphatidylEthanolamine -0.05 0.08 -0.10 0.08 -0.13 0.06

84 Glu + 2 NADPH -> Pro 0.31 0.17 0.77 0.30 0.50 0.25

85 Palm (In) ND. ND. 0.48 0.15 0.58 0.03

 

 

 

 

 

154

 

4. Matlab Program Used for Flux Calculations

function mfa(S,M)

% S: stoichiometric matrix

% M: measurement matrix

% where M has the following format:

% column 1: Index vector specifying measured reactions.
°/o column 2 to k: measurements

% column k + 1: measurement average
% column k + 2: measurement sd
clear;

S = Ioad('StoichO|e.txt');

M_in = Ioad('MerleD2.txt');

°/o Get data size
[data_rows,data_co|umns] = size(M_in);
num_groups = data_columns-3;

O/
I u

% Identifying Measurement Errors
0L

M = MsmtErr3(S,M_in);
tot_eqns = Iength(S);

list = M(:,1);

Gc = S;

Gc(:,list) = fl;

disp('The condition no. of Go matrix is');

disp(cond(Gc));

CN_GC = cond(Gc);

if CN_GC > 10"6
disp('condition number too high, QUITTING !!I');
return;

end

Gm = S(:,Iist);

output = fl;
fori= 1:(num_groups+1)
vm = M(:,i+1);
temp = mppi(Gc,Gm,vm,S,M);
output = [output,temp(:,2)];
end;
sd = M(:,data_columns);
var_vm = power(sd,2):
Gc_pi = inv(Gc.'*Gc)*Gc.';
B = -Gc_pi*Gm;
[B_row,B_coI] = size(B):
for i = 1:B_row
coeff = B(i,:);
coeff_2 = power(coeff,2);
var_vc(i) = coeff_2*var_vm;
end;
var_vc = var_vc.';

155

var_top = [compidx(S,M),var_vc];
var_bottom = [Iist,var_vm];

var = [var_top;var_bottom];

var = sortrows(var,1);

sd = power(var(:,2),0.5):

output = [output,sd];

%write the output matrix
output

% Save?

°/o save = input('Save? yes or no.\n>','s');

% while (strcmp(save,'yes') | strcmp(save,'no')) == 0

% save = input('lnvalid response. Enter yes or no.\n>','s');
% end;

% if strcmp(save,'no') == 0

% [fname, pathname] = uiputfile(’*.txt','Save Fit Results As');
% cd(pathname);

% %save fname output;

% mwrite(output,fname);

% disp('Fit results saved.')

"/0 else

% disp(’Fit results not saved.')

% end;

156

5. Matlab Program Used for Gross Measurement Identification according to Wang
and Stephanopoulos

function [MsmtAfterDeIn] = MsmtErr3(S,M)
%clear;

list = M(:,1);

n_msr = length(|ist); % ---------- Total number of measurements-----
[row_S, oolm_S] = size(S);

not_msr = co|m_S - n_msr;

x = M(:,2);

%xu = x;

%
"/0 Partition the S matrix (bring to the 1st columns those fluxes which are NOT measured)
OI.
Gc = S;
Gc(:,list) = [|;
%size(Gc);
Gm = S(:,|ist);
G = [Gc Gm];

0/_
% Eliminate the unmeasured columns and obtain the 'A' matrix
0].
G_red = rrefSS(G, n_msr);
%size(G__red)
for i = (not_msr+1) : row_S
forj = (not_msr+1) : coIm__S
A(i-not_msr,j-not_msr) = G_red(i,j);
end
end;
[nrow_A, ncol_A] = size(A);

min_msr = ncol_A-nrow_A; % -------- Min # of msrmt of just determination -----

 

°/o
% Obtain the Psi, Epsilon, Phi matrices

0/_
I u

Err = M(:,3);
Xbar = M(:,2);

Psi = diag(Err.*Err);
Eps = -A*Xbar,

Phi = A*Psi*A’;
Inv_Phi = pinv(Phi);

h = Eps'*lnv_Phi*Eps;

0/.
IU

% Calculate Chi-Square Parameter and compare h with it
OI.

n_fluxEq = length(S);
chi_sq_param = chi2inv(0.9, n_fluxEq-n_msr);

157

°/o
%

disp('Chi_Sq_param = '):
disp(chi__sq_param);

Msmt4Deln = [1;

if h > chi_sq_param

disp('lll Error in Measurements: Adjusting values !I!');

0/
°/oSTEP 1 : Initialize
OI.
XC = 1];

A0 = I];
Psi_c = fl;
xu = x;

Au = A;
Psi_u = Psi;

OI.
% Run the deletion multiple times until h < ChiSqr, or i = n_msr-1
0/
fori = 1:n_msr-1

[h_a, test _pass, GrossErr] = DeIMsmt2(nrow_A, Au, Ac, xu, xc, Psi_u, Psi_c, Eps, Inv_Phi,

n_msr, min_msr,i);

%
°/o

%

if test __pass ==
Msmt4Deln = [Msmt4DeIn, GrossErr];
index = D;
for row = 1:n_msr
forj = 1:Iength(Msmt4DeIn)
if (Xbar(row) == Msmt4Deln(j))
index = findex,row];
end
end
end
disp('The following measurements have gross errors');
disp(M(index,:));
M(index,:) = I];
MsmtAfterDeln = M;
disp('Success III, Errors reduced');
return
else
%---------Find the Index of msmt whose deletion produces the
%min value of h_a
I_ha = length(h_a);
for j = 1:I_ha
if h_a(j) == min(h_a)
disp('DeIeting the fell measurement produces least h');
diSp(XU(I));
Msmt4Deln = [Msmt4DeIn, xu(j)];
xc = [xc, xu(j)];
XUG) = I]:
Ac = [Ac Au(:,j)];
AU(I.I) = I];
Au(j,:) = I]:

158

Psi_c = diag(xc.*xc);

 

 

 

Psi_u(j,:) = fl;
Psi_u(:,j) = fl;
end °/o-----------End of If h_a == mmin(h_a)
end % End of forj = 1:I_ha
end °/o End of If-Else
end % End of For Looping
else % That is. if h <= chi2param

 

disp('Good Measurements--CaIculating Fluxes ..... ');
MsmtAfterDeln = M;
end 94. End of If h_a > chi2param

 

function [A] = rrefSS(A, n_msr, tol)
% A modification of the RREF command/ program in Matlab

[m,n] = size(A);

% Does it appear that elements of A are ratios of small integers?
[num, den] = rat(A);
rats = isequal(A,nudeen);

% Compute the default tolerance if none was provided.
if (nargin < 3), tol = max(m,n)*eps(c|ass(A))*norm(A,'inf‘); end

°/o Loop over the entire matrix.
i = 1;
j = 1:
ib = I]:
while (i <= m) && 0 <= n_msr) °/o Just change the n here if you want to eliminate
<n colms
"/0 Find value and index of largest element in the remainder of column j.
[p.k] = max(abs(A(i:m,j))); k = k+i-1;
if (p <= tol)
°/o The column is negligible, zero it out.
A(i:m,j) = zeros(m-i+1,1);
I =1 + 1:
else
% Remember column index
lb = [lb 1];
% Swap i-th and k-th rows.
Alli 10.11") = A(lk i].j:n);
% Divide the pivot row by the pivot element.
A(i,j:n) = A(i,j:n)/A(i,j);
% Subtract multiples of the pivot row from all the other rows.
for k = [1 :i-1 i+1:m]
A(k,j:n) = A(k,j:n) - A(k,j)*A(i,j:n);
end
i = i + 1;
i =1 + 1:
end
end

159

% Return "rational" numbers if appropriate.
if rats

[num,den] = rat(A);

A=num./den;
end

160

6. Matlab Program Used for the Perturbation Analyses
function mfa_Perturb_SS(S,M)

% S: stoichiometric matrix

% M: measurement matrix

% where M has the following format:

% column 1: Index vector specifying measured reactions.
% column 2 to k: measurements

% column k + 1: measurement average

% column k + 2: measurement sd

clear,

8 = load('StoichPa|m.txt');
M_in = load('MsrPalmD1.txt');

% Get data size
[data_rows,data_co|umns] = size(M_in);
num_groups = data_columns-B;

01’.
IU

% Identifying Measurement Errors
OI.

M = MsmtErr3(S,M_in):
tot_eqns = Iength(S):

list = M(:,1);

Gc = S;

Gc(:,list) = l];

disp('The condition no. of Go matrix is');

disp(cond(Gc));

CN_GC = cond(Gc);

if CN_GC > 10"6
disp('condition number too high, QUITTING I!!’);
return;

end

Gm = S(:,|ist);

Varbl = M(27,2); %Change here for the variable
FluxVaIs = I];
for Glu = -2*Varb|:0.5*VarbI:2*VarbI
output = fl;
var_vc = fl;
M(27,2) = Glu; % Change the index of M corresp to msmt
for i = 1:(num_groups+1)
vm = M(:,i+1);
temp = mppi(Gc,Gm,vm,S,M);
output = [output,temp(:,2)];
end;
sd = M(:,data_columns);
var_vm = power(sd,2);
Gc J1 = inv(Gc.'*Gc)*Gc.';
B = -Gc_pi*Gm;

161

[B_row,B_coI] = size(B);
for i = 1:B__row
amfi=B¢m
coeff_2 = power(coeff,2);
var_vc(i) = coeff_2*var_vm;
end;
var_vc = var_vc';
var_top = [compidx(S,M), var_vc];
var_bottom = [Iist,var_vm];
var = [var_top;var_bottom];
var = sortrows(var,1);
sd = power(var(:,2),0.5);
output = [output,sd];
FquVaIs = [FluxVals, output];
end

°/o-------Change the file name Acc to the Amino Acid- ---------
dImwrite('EffectPaImVarnD1.txt', FluxVaIs, 'delimiter‘, '\t‘);

162

10.

ll.

12.

BIBLIOGRAPHY

Agafonov A, Gritsenko E, Belosludtsev K, Kovalev A, Gateau-Roesch O, Saris NE,
Mironova GD. A permeability transition in liposomes induced by the formation of
Ca2+/palmitic acid complexes. Biochim Biophys Acta. 2003 Jan 31;]609(2):153-
60.

Ahuja HS, Szanto A, Nagy L, and Davies PJ. The retinoid X receptor and its
ligands: versatile regulators of metabolic function, cell differentiation and cell
death. J Biol Regul Homeost Agents. 2003 Jan-Mar;17(l):29-45.

Ajuwon KM, Spurlock ME. Palmitate activates the NF-kappaB transcription factor
and induces IL-6 and TNFalpha expression in 3T3-L1 adipocytes. J Nutr. 2005
Aug;l35(8):1841-6.

Al-Shahrour F, Minguez P, Vaquerizas JM, Conde L and Dopazo J. Babelomics: a
suite of web-tools for ftmctional annotation and analysis of group of genes in high-
throughput experiments, Nucleic Acids Research 2005; 33(Web Server issue):
W460-W464.

Anderson ME and Meister A. Intracellular delivery of cysteine. Methods Enzymol.
1987;143: 313-326.

Andrieu-Abadie N, Gouaze V, Salvayre R, Levade T. Ceramide in apoptosis
signaling: relationship with oxidative stress. Free Radic Biol Med. 2001 Sep
15;31(6):717-28.

Angulo P, Lindor KD. Non-alcoholic fatty liver disease. J Gastroenterol Hepatol.
2002 Feb;17 Supple186-90.

Armstrong J S. Mitochondrial membrane permeabilization: the sine qua non for cell
death. Bioessays. 2006 Mar;28(3):253-60.

Astrup A. The role of dietary fat in obesity. Semin Vasc Med. 2005 Feb;5(1):40-7.

Bandyopadhyay S, Zhan R, Wang Y, Pai SK, Hirota S, Hosobe S, Takano Y, Saito
K, Furuta E, Iiizumi M, Mohinta S, Watabe M, Chalfant C, and Watabe K.
Mechanism of apoptosis induced by the inhibition of fatty acid synthase in breast
cancer cells. Cancer Res. 2006 Jun 1;66(11):5934-40.

Barnard, M. L. and S. Matalon. Mechanisms of extracellular reactive oxygen
species injury to the pulmonary microvasculature. J Appl Physiol. 1992 May; 72(5):
1724-9.

Basnakian AG, Ueda N, Hong X, Galitovsky VE, Yin X, Shah SV. Ceramide
synthase is essential for endonuclease-mediated death of renal tubular epithelial

163

l3.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

cells induced by hypoxia-reoxygenation. Am J Physiol Renal Physiol. 2005
Feb;288(2):F308-14..

Beehler CJ, Ely ME, Rutledge KS, Simchuk ML, Reiss OK, Shanley PF, Repine
JE. Toxic effects of dimethylthiourea in rats. J Lab Clin Med. 1994 Jan;123(1):73-
80.

Beisvag V, Junge FK, Bergum H, Jolsum L, Lydersen S, Gunther CC, Ramampiaro
H, Langaas M, Sandvik AK, Laegreid A. GeneToolsuapplication for functional
annotation and statistical hypothesis testing. BMC Bioinformatics. 2006 Oct
24;7:470.

Bennett BL, Satoh Y, and Lewis A]. JNK: a new therapeutic target for diabetes.
Curr Opin Pharmacol. 2003 Aug;3(4):420-5.

Bemardi P, Penzo D, Wojtczak L. Mitochondrial energy dissipation by fatty acids.
Mechanisms and implications for cell death. Vitam Horm. 2002;65:97-126.

Bemardi P, Broekemeier KM, Pfeiffer DR. Recent progress on regulation of the
mitochondrial permeability transition pore; a cyclosporin-sensitive pore in the inner
mitochondrial membrane. J Bioenerg Biomembr. 1994 Oct;26(5):509-l7.

Berridge MI. The endoplasmic reticulum: a multifunctional signaling organelle.
Cell Calcium. 2002 Nov-Dec;32(5-6):235-49.

Bligh, E. G. and W. J. Dyer. A rapid method of total lipid extraction and
purification. Can J Biochem Physiol 1959 Aug;37(8): 911-7.

Boden G, She P, Mozzoli M, Cheung P, Gumireddy K, Reddy P, Xiang X, Luo Z,
and Ruderman N. Free fatty acids produce insulin resistance and activate the
proinflammatory nuclear factor-kappaB pathway in rat liver. Diabetes. 2005
Dec;54(12):3458-65.

Boden G. Role of fatty acids in the pathogenesis of insulin resistance and NIDDM.
Diabetes. 1997 Jan;46(1):3-10.

Boden G, Chen X, Capulong E, Mozzoli M. Effects of free fatty acids on
gluconeogenesis and autoregulation of glucose production in type 2 diabetes.
Diabetes. 2001 Apr;50(4):810-6.

Bradham CA, Plumpe J, Manns MP, Brenner DA, Trautwein C. Mechanisms of
hepatic toxicity. 1. TNF-induced liver injury. Am J Physiol. 1998 Sep;275(3 Pt
1):G387-92.

Brown M.P.S., Grundy W.N., Lin D., Cristianini N., Sugnet C., Furey T.S., Ares
M., Jr., Haussler D. Knowledge-based analysis of microarray gene expression data
using support vector machines. Proc. Natl. Acad. Sci. USA 2000 97(1):262-267.

164

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

Browning JD, Szczepaniak LS, Dobbins R, Nuremberg P, Horton JD, Cohen JC, et
al. Prevalence of hepatic steatosis in an urban population in the United States:
impact of ethnicity. Hepatology 2004; 40: 1387-1395.

Brownsey RW, Boone AN, Elliott JE, Kulpa JE, and Lee WM. Regulation of
acetyl-CoA carboxylase. Biochem Soc Trans. 2006 Apr;34(Pt 2):223-7.

Bruce CR, Dyck DJ. Cytokine regulation of skeletal muscle fatty acid metabolism:
effect of interleukin-6 and tumor necrosis factor—alpha. Am J Physiol Endocrinol
Metab. 2004 Oct;287(4):E61 6-21 .

Bruck R, Aeed H, Shirin H, Matas Z, Zaidel L, Avni Y, Halpem Z. The hydroxyl
radical scavengers dimethylsulfoxide and dimethylthiourea protect rats against
thioacetamide-induced fulminant hepatic failure. J Hepatol. 1999 Jul;3l(l):27-38.

Brun T, Assimacopoulos-Jeannet F, Corkey BE, and Prentki M. Long-chain fatty
acids inhibit acetyl-CoA carboxylase gene expression in the pancreatic beta-cell
line INS-1. Diabetes. 1997 Mar;46(3):393-400.

Brusselmans K, De Schrijver E, Verhoeven G, and Swinnen IV. RNA interference-
mediated silencing of the acetyl-CoA-carboxylase-alpha gene induces grth
inhibition and apoptosis of prostate cancer cells. Cancer Res. 2005 Aug
1;65(15):6719-25.

Cai LQ, Imperato-McGinley J, and Zhu YS. Regulation of prostate Salpha-
reductase-2 gene expression and prostate weight by dietary fat and caloric intake in
the rat. Prostate. 2006 May 15;66(7):738-48.

Cantley LC. The phosphoinositide 3-kinase pathway. Science. 2002 May
31;296(5573):1655-7.

Cederbaum AI, Dicker E, Rubin E, Cohen G. The effect of dimethylsulfoxide and
other hydroxyl radical scavengers on the oxidation of ethanol by rat liver
microsomes. Biochem Biophys Res Commun. 1977 Oct 24;78(4):1254-62.

Chajes V, Cambot M, Moreau K, Lenoir GM, and Joulin V. Acetyl-CoA
carboxylase alpha is essential to breast cancer cell survival. Cancer Res. 2006 May
15;66(10):5287-94.

Chan C, Berthiaume F, Lee K, Yarmush ML. Metabolic flux analysis of cultured
hepatocytes exposed to plasma. Biotechnol Bioeng. 2003 Jan 5;81(1):33-49.

Chan C, Berthiaume F, Lee K, Yarmush ML. Metabolic flux analysis of hepatocyte

function in hormone- and amino acid-supplemented plasma. Metab Eng. 2003
Jan;5(l):1-15.

165

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

Chan C, Hwang D, Stephanopoulos GN, Yarmush ML, Stephanopoulos G.
Application of multivariate analysis to optimize fimction of cultured hepatocytes.
Biotechnol Prog. 2003 Mar-Apr;l9(2):580-98.

Chan SL, Yu VC. Proteins of the bcl-2 family in apoptosis signalling: fiom
mechanistic insights to therapeutic opportunities. Clin Exp Pharmacol Physiol.
2004 Mar;3 1(3):] 19-28.

Chang Y, Wang J, Lu X, Thewke DP, Mason RJ. KGF induces lipogenic genes
through a PI3K and JNK/SREBP-l pathway in H292 cells. J Lipid Res. 2005
Dec;46(12):2624-35.

Cheema SK and Agellon LB. Metabolism of cholesterol is altered in the liver of
C3H mice fed fats enriched with different C-l8 fatty acids. J Nutr. 1999
Sep; 129(9): 171 8-24.

Chen T, Guo L, Zhang L, Shi L, Fang H, Sun Y, Fuscoe JC, Mei N. Gene
Expression Profiles Distinguish the Carcinogenic Effects of Aristolochic Acid in
Target (Kidney) and Non-target (Liver) Tissues in Rats. BMC Bioinformatics. 2006
Sep 26;7 Suppl 2:820

Chen WH, Tzeng SF. Pituitary adenylate cyclase-activating polypeptide prevents
cell death in the spinal cord with traumatic injury. Neurosci Lett. 2005 Aug 12-
l9;384(l-2):l 17-2].

Chen Q, Vazquez EJ, Moghaddas S, Hoppel CL, Lesnefsky EJ. Production of
reactive oxygen species by mitochondria: central role of complex III. J Biol Chem.
2003 Sep 19;278(38):36027-31.

Chia DJ, Boston BA. Childhood obesity and the metabolic syndrome. Adv Pediatr.
2006;53:23-53.

Cho JH, Lee D, Park JH, Lee IB. Gene selection and classification from microarray
data using kernel machine. FEBS Lett. 2004 Jul 30;571(1-3):93-8.

Cnop M, Hannaert JC, Hoorens A, Eizirik DL, Pipeleers DG. Inverse relationship
between cytotoxicity of free fatty acids in pancreatic islet cells and cellular
triglyceride accumulation. Diabetes. 2001 Aug;50(8): 1771-7.

Coleman RA, Lee DP. Enzymes of triacylglycerol synthesis and their regulation.
Prog Lipid Res. 2004 Mar;43(2):134-76.

Cossarizza A, Ceccarelli D, Masini A. Functional heterogeneity of an isolated
mitochondria] population revealed by cytofluorometric analysis at the single
organelle level. Exp Cell Res. 1996 Jan 10;222(1):84-94.

Cousin SP, Hug] SR, Wrede CE, Kajio H, Myers MG Jr, and Rhodes CJ. Free fatty
acid-induced inhibition of glucose and insulin-like growth factor I-induced

166

 

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

deoxyribonucleic acid synthesis in the pancreatic beta-cell line IN S-l.
Endocrinology. 2001 Jan;142(l):229-40.

Cutler RG, Pedersen WA, Camandola S, Rothstein JD, Mattson MP. Evidence that
accumulation of ceramides and cholesterol esters mediates oxidative stress-induced
death of motor neurons in amyotrophic lateral sclerosis. Ann Neurol. 2002

Oct;52(4):448-57.

D'Ambrosio SM, Gibson-D'Ambrosio RE, Brady T, Oberyszyn AS, Robertson FM.
Mechanisms of nitric oxide-induced cytotoxicity in normal human hepatocytes.
Environ Mol Mutagen 2001;37:46-54.

Dashti N, Feng Q, Franklin F A. Long-term effects of cis and trans monounsaturated
(18:1) and saturated (16:0) fatty acids on the synthesis and secretion of
apolipoprotein A-1- and apolipoprotein B-containing lipoproteins in HepG2 cells.
Journal of Lipid Research 2000;41:1980-1990.

Dave S, Farrance DP, Whitehead SA. Evidence that nitric oxide inhibits
steroidogenesis in cultured rat granulosa cells. Clin Sci (Lond) 1997;92:277-284.

Day CP, James OF. Steatohepatitis: a tale of two "hits"? Gastroenterology. 1998
Apr; 1 14(4):842-5.

DeFronzo RA. Dysfunctional fat cells, lipotoxicity and type 2 diabetes. Int J Clin
Pract Suppl. 2004 Oct;(143):9-21.

Degtyarev MY, Spiegel AM, Jones TL. The G protein alpha 8 subunit incorporates
[3H]palmitic acid and mutation of cysteine-3 prevents this modification.
Biochemistry. 1993 Aug 17;32(32):8057-61.

Devireddy LR, Teodoro JG, Richard FA, and Green MR. Induction of apoptosis by
a secreted lipocalin that is transcriptionally regulated by IL-3 deprivation. Science.
2001 Aug 3;293(5531):829-34.

Dickson LM, and Rhodes CJ. Pancreatic beta-cell growth and survival in the onset
of type 2 diabetes: a role for protein kinase B in the Akt? Am J Physiol Endocrinol
Metab. 2004 Aug;287(2):E192-8.

Ding W-X, Shen H-M, Ong C-N. Critical role of reactive oxygen species and
mitochondrial permeability transition in microcystin-induced rapid apoptosis in rat
hepatocytes. Hepatology (Philadelphia) 2000;32:547-555.

Draghici S, Khatri P, Bhavsar P, Shah A, Krawetz SA, Tainsky MA. Onto-Tools,
the toolkit of the modern biologist: Onto-Express, Onto-Compare, Onto-Design and
Onto-Translate. Nucleic Acids Res. 2003;31:3775-3781.

167

6].

62.

63.

64.

65.

66.

67.

68.

69.

70.

71.

72.

73.

Dringen R, Hamprecht B. Involvement of glutathione peroxidase and catalase in the
disposal of exogenous hydrogen peroxide by cultured astroglial cells. Brain Res
1997;759:67-75.

Dymkowska D, Szczepanowska J, Wieckowski MR, Wojtczak L. Short-term and
long-term effects of fatty acids in rat hepatoma AS-3OD cells: the way to apoptosis.
Biochim Biophys Acta. 2006 Feb;1763(2):152-63.

Edelstein LC, Lagos L, Simmons M, Tirumalai H, and Gelinas C. NF-kappa B-
dependent assembly of an enhanceosome-like complex on the promoter region of
apoptosis inhibitor Bfl-l/Al. Mol Cell Biol. 2003 Apr;23(8):2749-6l.

El-Assaad W, Buteau J, Peyot M-L, Nolan C, Roduit R, Hardy S, Joly E, et al.
Saturated fatty acids synergize with elevated glucose to cause pancreatic b-cell
death. Endocrinology 2003;144:4154-4163.

Farrell GC, Larter CZ. Nonalcoholic fatty liver disease: from steatosis to cirrhosis.
Hepatology. 2006 Feb;43(2 Suppl 1):S99-S112.

Felber JP, Golay A. Pathways from obesity to diabetes. Int J Obes Relat Metab
Disord. 2002 Sep;26 Supp12:S39-45.

F eldstein AE, Wemeburg NW, Canbay A, Guicciardi ME, Bronk SF, Rydzewski R,
Burgart L], et a]. Free fatty acids promote hepatic lipotoxicity by stimulating TNF-
alpha expression via a lysosomal pathway. Hepatology 2004;40:185-194.

F eldstein AE, Wemeburg NW, Li Z, Bronk SF, Gores GJ. Bax inhibition protects
against free fatty acid-induced lysosomal permeabilization. Am J Physiol
Gastrointest Liver Physiol. 2006 Jun;290(6):G1339-46.

Fernanda Cury-Boaventura M, Cristine Kanunfre C, Gorjao R, Martins de Lima T,
Curi R. Mechanisms involved in Jurkat cell death induced by oleic and linoleic
acids. Clin Nutr. 2006 Dec;25(6): 1004-14.

France-Lanord V, Brugg B, Michel PP, Agid Y, Ruberg M. Mitochondrial free
radical signal in ceramide-dependent apoptosis: a putative mechanism for neuronal
death in Parkinson's disease. J Neurochem 1997;69:1612-1621.

Fujita N and Tsuruo T. Survival-signaling pathway as a promising target for cancer
chemotherapy. Cancer Chemother Pharmacol. 2003 Jul;52 Suppl 1:S24-8.

Fukui K, Wada T, Kagawa S, Nagira K, Ikubo M, Ishihara H, Kobayashi M,
Sasaoka T. Impact of the liver-specific expression of SHIP2 (SHZ-containing
inositol 5'-phosphatase 2) on insulin signaling and glucose metabolism in mice.
Diabetes. 2005 Jul;54(7):1958-67.

Gardner DK, Lane M. Amino acids and ammonium regulate mouse embryo
development in culture. Biol Reprod. 1993 F eb;48(2):377-85.

168

 

74.

75.

76.

77.

78.

79.

80.

81.

82.

83.

84.

Gibbons GF, Khurana R, Odwell A, Seelaender MC. Lipid balance in HepG2 cells:
active synthesis and impaired mobilization. J Lipid Res. 1994 Oct;35(10): 1 801-8.

Glosli H, Gudbrandsen OA, Mullen AJ, Halvorsen B, Rost TH, Wergedahl H,
Prydz H, Aukrust P, and Berge RK. Down-regulated expression of PPARalpha
target genes, reduced fatty acid oxidation and altered fatty acid composition in the
liver of mice transgenic for hTNFalpha. Biochim Biophys Acta. 2005 Jun
1;]734(3):235-46.

Glosli H, Tronstad KJ, Wergedal H, Muller F, Svardal A, Aukrust P, Berge RK,
Prydz H. Human TNF-alpha in transgenic mice induces differential changes in
redox status and glutathione-regulating enzymes. FASEB J. 2002 Sep;16(l l):l450—
2.

Gogvadze V, Walter PB, Ames EN. The role of Fe2+-induced lipid peroxidation in
the initiation of the mitochondrial permeability transition. Arch Biochem Biophys
2003;414:255-260.

Gomez E0, Mendoza-Milla C, Ibarra-Sanchez MJ, Ventura-Gallegos JL, Zentella
A. Ceramide reproduces late appearance of oxidative stress during TNF-mediated
cell death in L929 cells. Biochem Biophys Res Commun 1996;228:505-509.

Gong P, Cederbaum AI. Transcription factor Ner protects HepG2 cells against
CYP2E1 plus arachidonic acid-dependent toxicity. J Biol Chem. 2006 May
26;281(21):14573-9.

Greutter CA and Lemke SM (1985) Dissociation of cysteine and glutathione levels
from nitroglycerin-induced relaxation. Eur J Pharmacol 111: 85-95.

Griffin JL, Bonney SA, Mann C, Hebbachi AM, Gibbons GF, Nicholson JK,
Shoulders CC, Scott J. An integrated reverse fimctional genomic and metabolic
approach to understanding orotic acid-induced fatty liver. Physiol Genomics. 2004
Apr 13;17(2):140-9. Grivennikova VG, Vinogradov AD. Generation of superoxide
by the mitochondrial Complex 1. Biochim Biophys Acta 2006.

Guo L, Zhang L, Sun Y, Muskhelishvili L, Blann E, Dial S, Shi L, Schroth G,
Dragan YP. Differences in hepatotoxicity and gene expression profiles by anti-
diabetic PPAR gamma agonists on rat primary hepatocytes and human HepG2 cells.
Mol Divers. 2006 Aug;lO(3):349-60.

Halestrap AP, Connem CP, Griffiths EJ, Kerr PM. Cyclosporin A binding to
mitochondrial cyclophilin inhibits the permeability transition pore and protects
hearts from ischaemia/reperfusion injury. Mol Cell Biochem 1997;174:167-172.

Halestrap AP, Woodfield KY, Connem CP. Oxidative stress, thiol reagents, and
membrane potential modulate the mitochondrial permeability transition by affecting

nucleotide binding to the adenine nucleotide translocase. J Biol Chem
1997;272:3346-3354.

169

 

85.

86.

87.

88.

89.

90.

91.

92.

93.

94.

95.

96.

97.

Hamacher-Brady A, Brady NR, Logue SE, Sayen MR, J inno M, Kirshenbaum LA,
Gottlieb RA, and Gustafsson AB. Response to myocardial ischemia/reperfusion
injury involves Bnip3 and autophagy. Cell Death Differ. 2006 Apr 28.

Hansen HO, Grunnet I, and Knudsen J. Triacylglycerol synthesis in goat mammary
gland. Factors influencing the esterification of fatty acids synthesized de novo.
Biochem J. 1984 Jun l;220(2):521-7.

Hansen HO, Grunnet I, and Knudsen J. Triacylglycerol synthesis in goat mammary
gland. The effect of ATP, Mg2+ and glycerol 3-phosphate on the esterification of
fatty acids synthesized de novo. Biochem J. 1984 Jun 1;220(2):513-9.

Hashimoto K, Farrow BJ, Evers BM. Activation and role of MAP kinases in 15d-
PGJ2-induced apoptosis in the human pancreatic cancer cell line MIA PaCa-2.
Pancreas. 2004 Mar;28(2): 153-9.

Hennig B, Lei W, Arzuaga X, Ghosh DD, Saraswathi V, Toborek M. Linoleic acid
induces proinflammatory events in vascular endothelial cells via activation of
PI3K/Akt and ERKl/2 signaling. J Nutr Biochem. 2006 Nov;l7(1l):766-72.

Hewlett EL, Donato GM, Gray MC. Macrophage cytotoxicity produced by
adenylate cyclase toxin from Bordetella pertussis: more than just making cyclic
AMP! Mol Microbiol. 2006 Jan;59(2):447-59.

Higa M, Shimabukuro M, Shimajiri Y, Takasu N, Shinjyo T, and Inaba T. Protein
kinase B/Akt signalling is required for palmitate-induced beta-cell lipotoxicity.
Diabetes Obes Metab. 2006 Mar;8(2):228-33.

Hirosumi J, Tuncman G, Chang L, Gorgun CZ, Uysal KT, Maeda K, Karin M, and
Hotamisligil GS. A central role for JNK in obesity and insulin resistance. Nature.
2002 Nov 21 ;420(6913):333-6.

Hong-Ping G, Bao-Shan KU. Neuroprotective effect of L-lysine
monohydrochloride on acute iterative anoxia in rats with quantitative analysis of
electrocorticogram. Life Sci. 1999;65(2):PL19-25.

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene&cmd=Retrieve&dopt=full
_report&list_uids=1 1 83

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene&cmd=Retrieve&dopt=full
_report&list_uids=535

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene&cmd=Retrieve&dopt=full
_report&list_uids=9550

Huang ZZ, Chen C, Zeng Z, Yang H, Oh J, Chen L, Lu SC. Mechanism and

significance of increased glutathione level in human hepatocellular carcinoma and
liver regeneration. FASEB J. 2001 Jan;15(1):19-21.

170

 

 

98.

99.

100.

101.

102.

103.

104.

105.

106.

107.

108.

109.

Hwang D, Alevizos I, Schmitt WA, Misra J, Ohyama H, Todd R, Mahadevappa M,
Warrington JA, Stephanopoulos G, Wong DT, Stephanopoulos G. Genomic
dissection for characterization of cancerous oral epithelium tissues using
transcription profiling. Oral Oncol. 2003 Apr;39(3):259-68.

Ichihara K, and Tanaka C. Some properties of progesterone 5 alpha-reductase
solubilized fiom rat liver microsomes. Biochem Int. 1987 Nov;15(5):1005-11.

Imazu T, Shimizu S, Tagami S, Matsushima M, Nakamura Y, Miki T, Okuyama A,
and Tsujimoto Y. Bcl-2/E1B 19 kDa-interacting protein 3-like protein (Bnip3L)
interacts with bcl-2/Bcl-xL and induces apoptosis by altering mitochondrial
membrane permeability. Oncogene. 1999 Aug 12;l8(32):4523-9.

Isenberg J S, Klaunig JE. Role of the mitochondrial membrane permeability
transition (MPT) in rotenone-induced apoptosis in liver cells. Toxicol Sci
2000;53:340-351.

Issandou M, Guillard R, Boullay AB, Linhart V, Lopez-Perez E. Up-regulation of
low-density lipoprotein receptor in human hepatocytes is induced by sequestration

of free cholesterol in the endosomal/lysosomal compartment. Biochem Pharmacol.
2004 Jun 15;67(12):2281-9.

Ji J, Zhang L, Wang P, Mu YM, Zhu XY, Wu YY, Yu H, Zhang B, Chen SM, and
Sun XZ. Saturated free fatty acid, palmitic acid, induces apoptosis in fetal
hepatocytes in culture. Exp Toxicol Pathol. 2005 Apr;56(6):369-76.

Jimenez-Lopez JM, Carrasco MP, Marco C, and Segovia JL.
Hexadecylphosphocholine disrupts cholesterol homeostasis and induces the
accumulation of free cholesterol in HepG2 tumour cells. Biochem Pharmacol. 2006
Apr l4;71(8):1114—21.

Joffe BI, Goldberg RB,.Sefiel HC, Distiller LA. Insulin, glucose and triglyceride
relationships in obese African subjects. Am J Clin Nutr. 1975 Jun;28(6):616-20.

Joseph R, Yazer E, Hanakawa Y, Stadnyk AW. Prostaglandins and activation of
AC/cAMP prevents anoikis in IEC-18. Apoptosis. 2005 Dec;10(6): 1221-33.

Jove M, Planavila A, Sanchez RM, Merlos M, Laguna JC, Vazquez-Carrem M.
Palmitate induces tumor necrosis factor-alpha expression in C2C12 skeletal muscle
cells by a mechanism involving protein kinase C and nuclear factor-kappaB
activation. Endocrinology. 2006 Jan; 147(1):552-6l .

Kadenbach B. Intrinsic and extrinsic uncoupling of oxidative phosphorylation.
Biochim Biophys Acta 2003;1604z77-94.

Karin M and Gallagher E. From JNK to pay dirt: jun kinases, their biochemistry,
physiology and clinical importance. IUBMB Life. 2005 Apr-May;57(4-5):283-95.

171

 

110.

111.

112.

113.

114.

115.

116.

117.

118.

119.

120.

121.

122.

Kennedy BP and Ramachandran C. Protein tyrosine phosphatase-IE in diabetes.
Biochem Pharmacol. 2000 Oct 1;60(7):877-83.

Kim B], Forbes NS. Flux analysis shows that hypoxia-inducible-factor—l-alpha
minimally affects intracellular metabolism in tumor spheroids. Biotechnol Bioeng.
2006 Sep 28;

Kim BJ, Ryu SW, Song BJ. JNK- and p38 kinase-mediated phosphorylation of Bax
leads to its activation and mitochondrial translocation and to apoptosis of human
hepatoma HepG2 cells. J Biol Chem. 2006 Jul 28;281(30):21256-65.

Kim Y-M, de Vera ME, Watkins SC, Billiar TR. Nitric oxide protects cultured rat
hepatocytes from tumor necrosis factor-a-induced apoptosis by inducing heat shock
protein 70 expression. Journal of Biological Chemistry 1997;272:1402-1411.

Kim Y-M, Kim T-H, Chung H-T, Talanian RV, Yin X-M, Billiar TR. Nitric oxide
prevents tumor necrosis factor a-induced rat hepatocyte apoptosis by the

interruption of mitochondrial apoptotic signaling through S-nitrosylation of
caspase-8. Hepatology (Philadelphia) 2000;32:770-778.

Kisseleva MV, Cao L, and Majerus PW. Phosphoinositide-specific inositol
polyphosphate 5-phosphatase IV inhibits Akt/protein kinase B phosphorylation and
leads to apoptotic cell death. J Biol Chem. 2002 Feb 22;277(8):6266-72.

Kojima H, Sakurai S, Uemura M, Takekawa T, Morimoto H, Tamagawa Y, Fukui
H. Difference and similarity between non-alcoholic steatohepatitis and alcoholic
liver disease. Alcohol Clin Exp Res. 2005 Dec;29(12 Suppl):259S-63S.

Kong JY, Rabkin SW. Palrnitate-induced apoptosis in cardiomyocytes is mediated
through alterations in mitochondria: prevention by cyclosporin A. Biochimica et
Biophysica Acta 2000;1485245-55.

Kroemer G, Dallaporta B, Resche-Rigon M. The mitochondrial death/life regulator
in apoptosis and necrosis. Annu Rev Physiol 1998;60:619-642.

Laurent A, Nicco C, Tran Van Nhieu J, Borderie D, Chereau C, Conti F, Jaffray P,
et al. Pivotal role of superoxide anion and beneficial effect of antioxidant molecules
in murine steatohepatitis. Hepatology 2004;39:1277-1285.

Le Phillip P, Bah] A, Ungar LH. Using prior knowledge to improve genetic network
reconstruction from microarray data. In Silico Biol. 2004;4(3):335-53.

Lee H, Kang J, and Stipanuk MH. Differential regulation of glutamate-cysteine
ligase subunit expression and increased holoenzyme formation in response to
cysteine deprivation. Biochem J. 2006 January 1; 393(Pt 1): 181-190.

Lee JS, Pinnamaneni SK, Eo SJ, Cho IH, Pyo JH, Kim CK, Sinclair AJ, Febbraio
MA, Watt MJ. Saturated, but not n-6 polyunsaturated, fatty acids induce insulin

172

 

123.

124.

125.

126.

127.

128.

129.

130.

131.

132.

133.

134.

resistance: role of intramuscular accumulation of lipid metabolites. J Appl Physiol.
2006 May; 100(5): 1467-74.

Lee K, Berthiaume F, Stephanopoulos GN, Yarmush ML. Profiling of dynamic
changes in hyperrnetabolic livers. Biotechnol Bioeng. 2003 Aug 20;83(4):400-15.

Lee M, Bae MA. Docosahexaenoic acid induces apoptosis in CYP2El-containing
HepG2 cells by activating the c-Jun N-terrninal protein kinase related mitochondrial
damage. J Nutr Biochem. 2006 Sep 7.

Leo C, Horn LC, and Hockel M. Hypoxia and expression of the proapoptotic
regulator BNIP3 in cervical cancer. Int J Gynecol Cancer. 2006 May-
Jun;16(3):1314-20.

Li J, Li LJ, Chao HC, Yang Q, Liu XL, Sheng JF, Yu HY, Huang JR. Isolation and
short term cultivation of swine hepatocytes for bioartificial liver support system.
Hepatobiliary Pancreat Dis Int. 2005 May;4(2):249-53.

Li Z, and Chan C. Integrating gene expression and metabolic profiles. J Biol Chem.
2004 Jun 25;279(26):27124—37.

Liang T and Liao S. Inhibition of steroid 5 alpha-reductase by specific aliphatic
unsaturated fatty acids. Biochem J. 1992 Jul 15;285 (Pt 2):557-62.

Lim G, Wang S, Lim JA, Mao J. Activity of adenylyl cyclase and protein kinase A
contributes to morphine-induced spinal apoptosis. Neurosci Lett. 2005 Dec
2;389(2): 104-8.

Linder et al., Proc. Natl. Acad. Sci. USA 90 (1993) 3675-3679.

Listenberger LL, Han X, Lewis SE, Cases S, Farese RV Jr, Ory DS, and Schaffer
J E. Triglyceride accumulation protects against fatty acid-induced lipotoxicity. Proc
Natl Acad Sci U S A. 2003 Mar l8;100(6):3077-82.

Listenberger LL, Ory DS, Schaffer JE. Palmitate-induced apoptosis can occur
through a ceramide-independent pathway. J Biol Chem. 2001 May
4;276(18):14890-5.

Littler DR, Assaad NN, Harrop SJ, Brown LJ, Pankhurst GJ, Luciani P, Aguilar MI,
Mazzanti M, Berryman MA, Breit SN, Curmi PM. Crystal structure of the soluble
form of the redox-regulated chloride ion channel protein CLIC4. FEBS J. 2005
Oct;272(19):4996—5007.

Liu J], Cutler G, Li W, Pan Z, Peng S, Hoey T, Chen L, Ling XB. Multiclass cancer
classification and biomarker discovery using GA-based algorithms. Bioinformatics.
2005 Jun 1;21(1 1):2691-7.

173

135.

136.

137.

138.

139.

140.

141.

142.

143.

144.

145.

146.

Liu Y, F iskum G, Schubert D. Generation of reactive oxygen species by the
mitochondrial electron transport chain. J Neurochem 2002;80:780-787.

Lu SC, Huang HY. Comparison of sulfiir amino acid utilization for GSH synthesis
between HepGZ cells and cultured rat hepatocytes. Biochem Pharmacol. 1994 Mar
2;47(5):859-69.

Lu SC. Regulation of hepatic glutathione synthesis: current concepts and
controversies. FASEB J. 1999 Jul;13(10):]169-83.

Lu YF, Chiang CF. Effect of dietary cholesterol and fat levels on lipid peroxidation
and the activities of antioxidant enzymes in rats. Int J Vitam Nutr Res 2001;71:339-
346.

Lu Z-H, Mu Y-M, Wang B-A, Li X-L, Lu J-M, Li J-Y, Pan C-Y, et a]. Saturated
free fatty acids, palmitic acid and stearic acid, induce apoptosis by stimulation of
ceramide generation in rat testicular Leydig cell. Biochemical and Biophysical
Research Communications 2003;303:1002-1007.

Maas JW Jr, Indacochea RA, Muglia LM, Tran TT, Vogt SK, West T, Benz A,
Shute AA, Holtzman DM, Mennerick S, Olney JW, Muglia LJ. Calcium-stimulated

adenylyl cyclases modulate ethanol-induced neurodegeneration in the neonatal
brain. J Neurosci. 2005 Mar 2;25(9):2376—85.

Maestre I, Jordan J, Calvo S, Reig JA, Cena V, Soria B, Prentki M, et al.
Mitochondrial dysfunction is involved in apoptosis induced by serum withdrawal
and fatty acids in the b-cell line INS-1. Endocrinology 2003;144:335-345.

Malhi H, Bronk SF, Wemeburg NW, Gores GJ. Free fatty acids induce JNK-
dependent hepatocyte lipoapoptosis. J Biol Chem. 2006 Apr 28;281(17):12093-101.

Manco M, Mingrone G, Greco AV, Capristo E, Gniuli D, De Gaetano A, and
Gasbarrini G. Insulin resistance directly correlates with increased saturated fatty
acids in skeletal muscle triglycerides. Metabolism. 2000 Feb;49(2):220-4.

McCullough AJ. Pathophysiology of nonalcoholic steatohepatitis. J Clin
Gastroenterol. 2006 Mar;40(3 Suppl 1):S17-29.

Menendez JA, Mehmi 1, Atlas E, Colomer R, and Lupu R. Novel signaling
molecules implicated in tumor-associated fatty acid synthase-dependent breast
cancer cell proliferation and survival: Role of exogenous dietary fatty acids, p53-
p21WAF1/CIP1, ERK1/2 MAPK, p27KlPl, BRCA1, and NF-kappaB. Int J Oncol.
2004 Mar;24(3):591-608.

Mensink M, Blaak EE, van Baak MA, Wagenmakers AJ, Saris WH. Plasma free
Fatty Acid uptake and oxidation are already diminished in subjects at high risk for
developing type 2 diabetes. Diabetes 2001;50:2548-2554.

174

147.

148.

149.

150.

151.

152.

153.

154.

155.

156.

157.

158.

Miglietta A, Bozzo F, Bocca C, Gabriel L, Trombetta A, Belotti S, Canuto RA.
Conjugated linoleic acid induces apoptosis in MDA-MB-231 breast cancer cells
through ERK/MAPK signalling and mitochondrial pathway. Cancer Lett. 2006 Mar
28;234(2):149-57.

Minnich A, Tian N, Byan L, Bilder G. A potent PPARalpha agonist stimulates
mitochondrial fatty acid beta-oxidation in liver and skeletal muscle. Am J Physiol
Endocrinol Metab. 2001 Feb;280(2):E270-9.

Mitsui H, Takuwa N, Maruyama T, Maekawa H, Hirayama M, Sawatari T,
Hashimoto N, Takuwa Y, Kimura S. The MEKl-ERK map kinase pathway and the
PI 3-kinase-Akt pathway independently mediate anti-apoptotic signals in HepG2
liver cancer cells. Int J Cancer. 2001 Apr 1;92(l):55-62.

Miyazaki M, Kim YC, and Ntambi JM. A lipogenic diet in mice with a disruption
of the stearoyl-CoA desaturase 1 gene reveals a stringent requirement of

endogenous monounsaturated fatty acids for triglyceride synthesis. J Lipid Res.
2001 Jul;42(7): 1018-24.

Miyazaki M, Kim YC, Gray-Keller MP, Attie AD, and Ntambi JM. The
biosynthesis of hepatic cholesterol esters and triglycerides is impaired in mice with

a disruption of the gene for stearoyl-CoA desaturase l. J Biol Chem. 2000 Sep
29;275(39):30132-8.

Mollner S, Beck K, Pfeuffer T. Acylation of adenylyl cyclase catalyst is important
for enzymic activity. FEBS Lett. 1995 Sep 11;371(3):241-4.

Morrison JL, Breitling R, Higham DJ, Gilbert DR. GeneRank: using search engine
technology for the analysis of microarray experiments. BMC Bioinformatics. 2005
Sep 21;6:233.

Musashi M, Ota S, Shiroshita N. The role of protein kinase C isofonns in cell
proliferation and apoptosis. Int J Hematol. 2000 Ju1;72(1):12-9.

Negrotto S, Pacienza N, D'Atri LP, Pozner RG, Malaver E, Torres 0, Lazzari MA,
Gomez RM, Schattner M. Activation of cyclic AMP pathway prevents CD34(+)
cell apoptosis. Exp Hematol. 2006 Oct;34(10):1420-8.

Newsholme P, Keane D, Welters HJ, Morgan NG. Life and death decisions of the
pancreatic beta-cell: the role of fatty acids. Clin Sci (Lond). 2007 Jan;112(1):27-42.

Nguyen GH, French R, Radhakrishna H. Protein kinase A inhibits lysophosphatidic
acid induction of serum response factor via alterations in the actin cytoskeleton.
Cell Signal. 2004 Oct;16(10):1141-51.

Ni TC and Savageau MA. Application of biochemical systems theory to
metabolism in human red blood cells. Signal propagation and accuracy of
representation. Journal of biological chemistry 1996;271:7927-41.

175

159.

160.

161.

162.

163.

164.

165.

166.

167.

168.

169.

170.

Ni TC and Savageau MA. Model assessment and refinement using strategies from
biochemical systems theory: application to metabolism in human red blood cells.
Journal of theoretical biology 1996;179:329-68.

Niknahad H, Khan S, O'Brien PJ. Hepatocyte injury resulting from the inhibition of
mitochondrial respiration at low oxygen concentrations involves reductive stress
and oxygen activation. Chem Biol Interact 1995;98:27-44.

Nohl H, Gille L, Staniek K. Intracellular generation of reactive oxygen species by
mitochondria. Biochem Pharmacol 2005;69:719-723.

Okuyama R, Fujiwara T, Ohsumi J. High glucose potentiates palmitate-induced
NO-mediated cytotoxicity through generation of superoxide in clonal b-cell HIT-
T15. FEBS Letters 2003;545:219-223.

Oliver MF, Opie LH. Effects of glucose and fatty acids on myocardial ischaemia
and arrhythmias. Lancet. 1994;343:155-158.

Olsen C. An enzymatic fluorimetric rnicromethod for the determination of
acetoacetate, -hydroxybutyrate, pyruvate and lactate. Clin Chim Acta. 1971
Jul;33(2):293-300.

Ono H, Shimano H, Katagiri H, Yahagi N, Sakoda H, Onishi Y, Anai M, Ogihara
T, Fujishiro M, Viana AY, Fukushima Y, Abe M, Shojima N, Kikuchi M, Yamada
N, Oka Y, and Asano T. Hepatic Akt activation induces marked hypoglycemia,
hepatomegaly, and hypertriglyceridemia with sterol regulatory element binding
protein involvement. Diabetes. 2003 Dec;52(12):2905-13.

Ostrander DB, Sparagna GC, Amoscato AA, McMillin JB, Dowhan W. Decreased
cardiolipin synthesis corresponds with cytochrome c release in palmitate-induced
cardiomyocyte apoptosis. Journal of Biological Chemistry 2001 ;276:38061-3 8067.

Pan W. Incorporating biological information as a prior in an empirical bayes
approach to analyzing microarray data. Stat Appl Genet Mol Biol.
2005;4:Article12.

Petronilli V, Cola C, Massari S, Colonna R, Bemardi P. Physiological effectors
modify voltage sensing by the cyclosporin A-sensitive permeability transition pore
of mitochondria. J Biol Chem 1993;268:21939-21945.

Picollo A, Pusch M. Chloride/proton antiporter activity of mammalian CLC
proteins ClC-4 and ClC-5. Nature. 2005 Jul 21;436(7049):420-3.

Qanungo S, Wang M, and Nieminen AL. N-Acetyl-L-cysteine Enhances Apoptosis

through Inhibition of Nuclear Factor-{kappa}B in Hypoxic Murine Embryonic
Fibroblasts. J. Biol. Chem., Vol. 279, Issue 48, 50455-50464, November 26, 2004.

176

 

171.

172.

173.

174.

175.

176.

177.

178.

179.

180.

181.

182.

183.

Randle PJ. The interrelationships of hormones, fatty acid and glucose in the
provision of energy. Postgrad Med J. 1964 Aug;40:457-63.

Raynaud JP, Cousse H, and Martin PM. Inhibition of type 1 and type 2 5alpha-

reductase activity by free fatty acids, active ingredients of Permixon. J Steroid
Biochem Mol Biol. 2002 Oct;82(2-3):233-9.

Remizov O, Jakubov R, Dufer M, Krippeit Drews P, Drews G, Waring M, Brabant
G, Wienbergen A, Rustenbeck I, Schofl C. Palmitate-induced Ca2+-signaling in
pancreatic beta-cells. Mol Cell Endocrinol. 2003 Dec 30;212(1-2):1-9.

Resh MD. Palmitoylation of ligands, receptors, and intracellular signaling
molecules. Sci STKE. 2006 Oct 31;2006(359):re14.

Roden M. Muscle triglycerides and mitochondrial function: possible mechanisms
for the development of type 2 diabetes. Int J Obes (Lond). 2005 Sep;29 Suppl
2:8111-5.

Ruvolo PP. Intracellular signal transduction pathways activated by ceramide and its
metabolites. Pharmacol Res. 2003 May;47(5):383-92.

Rys-Sikora KB and Gill DL. Fatty acid-mediated calcium sequestration within
intracellular calcium pools. J Biol Chem. 1998 Dec 4;273(49):32627-35.

Sandler S, Andersson A. The partial protective effect of the hydroxyl radical
scavenger dimethyl urea on streptozotocin-induced diabetes in the mouse in vivo
and in vitro. Diabetologia 1982;23:374-378.

Sanyal AJ, Campbell-Sargent C, Mirshahi F, Rizzo WB, Contos MJ, Sterling RK,
Luketic VA, Shiffinan ML, and Clore JN. Nonalcoholic steatohepatitis: association

of insulin resistance and mitochondrial abnormalities. Gastroenterology. 2001
Apr; 120(5): 1 1 83-92.

Sasaoka T, Wada T, and Tsuneki H. Lipid phosphatases as a possible therapeutic
target in cases of type 2 diabetes and obesity. Pharmacol Ther. 2006 Jul 12

Sato H, Shiiya A, Kimata M, Maebara K, Tamba M, Sakakura Y, Makino N,
Sugiyama F, Yagami K, Moriguchi T, Takahashi S, Bannai S. Redox imbalance in

cystine/glutamate transporter-deficient mice. J Biol Chem. 2005 Nov
11;280(45):37423-9.

Schattenberg JM, Singh R, Wang Y, Leflcowitch JH, Rigoli RM, Scherer PE, Czaja
MJ. JNK] but not JNK2 promotes the development of steatohepatitis in mice.
Hepatology. 2006 Jan;43(1):163-72.

Scheen AJ, Luyckx FH. Obesity and liver disease. Best Pract Res Clin Endocrinol
Metab. 2002 Dec;16(4):703-16.

177

 

184.

185.

186.

187.

188.

189.

190.

191.

192.

193.

194.

195.

Schonfeld P, Bohnensack R. Fatty acid-promoted mitochondrial permeability
transition by membrane depolarization and binding to the ADP/ATP carrier. FEBS
Lett 1997;420:167-170.

Schuchmann M, Galle PR. Dead or alive -- NF-kappaB, the guardian which tips the
balance. J Hepatol. 2002 Jun;36(6):827-8.

Shi C8 and Kehrl JH. Tumor necrosis factor (TNF)-induced germinal center kinase-
related (GCKR) and stress-activated protein kinase (SAPK) activation depends
upon the E2/E3 complex chl3-Uele/TNF receptor-associated factor 2
(TRAF2). J Biol Chem. 2003 Apr 25;278(17):15429-34.

Shimabukuro M, Ohneda M, Lee Y, Unger RH. Role of nitric oxide in obesity-
induced beta cell disease. J Clin Invest 1997;100:290-295.

Shimabukuro M, Zhou YT, Levi M, Unger RH. Fatty acid-induced beta cell
apoptosis: a link between obesity and diabetes.Proc Natl Acad Sci U S A. 1998 Mar
3;95(5):2498-502.

Simbula G, Columbano A, Ledda-Columbano GM, Sanna L, Deidda M, Diana A,
Pibiri M. Increased ROS generation and p53 activation in alpha-lipoic acid-induced
apoptosis of hepatoma cells. Apoptosis. 2006 Nov 21

Skowronski R, Hollenbeck CB, Varasteh BB, Chen YD, Reaven GM. Regulation of
non-esterified fatty acid and glycerol concentration by insulin in normal individuals
and patients with type 2 diabetes. Diabet Med 1991;8z330-333.

Skurk T, Alberti-Huber C, Herder C, Hauner H. Relationship between adipocyte
size and adipokine expression and secretion. J Clin Endocrinol Metab. 2006.

Sleeman MW, Wortley KE, Lai KM, Gowen LC, Kintner J, Kline W0, Garcia K,
Stitt TN, and Yancopoulos GD, Wiegand SJ, and Glass DJ. Absence of the lipid
phosphatase SHIP2 confers resistance to dietary obesity. Nat Med. 2005
Feb;11(2):199-205.

Srivastava S and Chan C. Hydrogen peroxide and hydroxyl radicals mediate
palmitate toxicity to human hepatoma cells: relation to mitochondrial permeability
transition. Free Rad Res. 2007 Jan; 41(1):38-49.

Starkov AA, Fiskum G. Regulation of brain mitochondrial H202 production by
membrane potential and NAD(P)H redox state. J Neurochem. 2003
Sep;86(5):l 101-7.

Strange K, Denton J, and Nehrke K. Ste20—type kinases: evolutionarily conserved

regulators of ion transport and cell volume. Physiology (Bethesda). 2006
Feb;21:61-8.

178

 

196.

197.

198.

199.

200.

201.

202.

203.

204.

205.

206.

207.

Stroemblad G, Bjoemtorp P. Reduced hepatic insulin clearance in rats with dietary-
induced obesity. Metabolism, Clinical and Experimental 1986;35:323-327.

Su F, Schneider RJ. Hepatitis B virus HBx protein sensitizes cells to apoptotic
killing by tumor necrosis factor alpha. Proc Natl Acad Sci U S A 1997;94:8744-
8749.

Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA,
Paulovich A, Pomeroy SL, Golub TR, Lander E8, Mesirov JP. Gene set enrichment

analysis: a knowledge-based approach for interpreting genome-wide expression
profiles. Proc Natl Acad Sci U S A. 2005 Oct 25;102(43):15545-50.

Swagell CD, Henly DC, Morris CP. Expression analysis of a human hepatic cell
line in response to palmitate. Biochem Biophys Res Commun. 2005 Mar
11;328(2):432-41.

Tabas 1. Consequences of cellular cholesterol accumulation: basic concepts and
physiological implications. J Clin Invest. 2002 Oct;110(7):905-11.

Takahashi T, Tang T, Lai NC, Roth DM, Rebolledo B, Saito M, Lew WY, Clopton
P, Hammond HK. Increased cardiac adenylyl cyclase expression is associated with
increased survival after myocardial infarction. Circulation. 2006 Aug 1;114(5):388-
96.

Tan Y, Shi L, Tong W, Hwang GT, Wang C. Multi-class tumor classification by
discriminant partial least squares using microarray gene expression data and
assessment of classification models. Comput Biol Chem. 2004 Jul;28(3):235-44.

Tan Y, Shi L, Tong W, Wang C. Multi-class cancer classification by total principal
component regression (TPCR) using microarray gene expression data. Nucleic
Acids Res. 2005 Jan 7;33(1):56—65.

Tanaka N, Aoyama T. Mechanism of the onset and development of hepatic steatosis
and PPAR. Cell (Tokyo, Japan) 2001;33:289-292.

Taub R. Liver regeneration 4: transcriptional control of liver regeneration. FASEB
J. 1996 Mar;10(4):413-27.

Tietge UJ, Sun G, Czamecki 8, Yu Q, Lohse P, Du H, Grabowski GA, Glick JM,
Rader DJ. Phenotypic correction of lipid storage and growth arrest in wolman

disease fibroblasts by gene transfer of lysosomal acid lipase. Hum Gene Ther. 2001
Feb 10;12(3):279-89.

Tiwari M, Kumar A, Sinha RA, Shrivastava A, Balapure AK, Sharma R, Bajpai
VK, et a1. Mechanism of 4-HPR induced apoptosis in glioma cells: evidences
suggesting role of mitochondrial-mediated pathway and endoplasmic reticulum
stress. Carcinogenesis 2006.

179

 

208.

209.

210.

211.

212.

213.

214.

215.

216.

217.

218.

219.

Troyanskaya OG, Garber ME, Brown PO, Botstein D, Altman RB. Nonparametric
methods for identifying differentially expressed genes in microarray data.
Bioinformatics. 2002 Nov;18(1 1): 1454-61.

Tuttle RL, Gill NS, Pugh W, Lee JP, Koeberlein B, Furth EE, Polonsky K8, Naji A,
and Birnbaum MJ. Regulation of pancreatic beta-cell grth and survival by the
serine/threonine protein kinase AktI/PKBalpha. Nat Med. 2001 Oct;7(10):1133-7.

Ueki K, Kondo T, Tseng YH, Kahn CR. Central role of suppressors of cytokine
signaling proteins in hepatic steatosis, insulin resistance, and the metabolic
syndrome in the mouse. Proc Natl Acad Sci U S A. 2004 Jul 13;101(28): 10422-7.

Van Der Westhuizen L, Shephard GS, Snyrnan SD, Abel S, Swanevelder S,
Gelderblom WCA. Inhibition of sphingolipid biosynthesis in rat primary hepatocyte

cultures by fumonisin BI and other structurally related compounds. Food and
Chemical Toxicology 1998;36:497-503.

Van ISC, Van Der Wouden JM, Liebisch G, Schrnitz G, Hoekstra D. Polarized
membrane traffic and cell polarity development is dependent on dihydroceramide
synthase-regulated sphinganine turnover. Mol Biol Cell 2004;15:4115-4124.

Verbsky J and Majerus PW. Increased levels of inositol hexakisphosphate (InsP6)
protect HEK293 cells from tumor necrosis factor (alpha)- and Fas-induced
apoptosis. J Biol Chem. 2005 Aug 12;280(32):29263-8.

Verbsky JW, Chang 8C, Wilson MP, Mochizuki Y, and Majerus PW. The pathway
for the production of inositol hexakisphosphate in human cells. J Biol Chem. 2005
Jan 21;280(3):1911-20.

Voth DE, Hamm EE, Nguyen LG, Tucker AE, Salles II, Ortiz-Leduc W, Ballard
JD. Bacillus anthracis oedema toxin as a cause of tissue necrosis and cell type-
specific cytotoxicity. Cell Microbiol. 2005 Aug;7(8):1139-49.

Vugmeyster Y, Borodovsky A, Maurice MM, Maehr R, Furrnan MH, and Ploegh
HL. The ubiquitin-proteasome pathway in thymocyte apoptosis: caspase-dependent
processing of the deubiquitinating enzyme USP7 (I-IAUSP). Mol Immunol. 2002
Nov;39(7-8):431-41.

Wagman AS and Nuss JM. Current therapies and emerging targets for the treatment
of diabetes. Curr Pharm Des. 2001 Apr;7(6):417-50.

Wang A, Xia T, Ran P, Chen X, Nuessler AK. Qualitative study of three cell
culture methods. J Huazhong Univ Sci Technolog Med Sci. 2002;22(4):288-91.

Wang D and Sul HS. (1998) Insulin stimulation of the fatty acid synthase promoter
is mediated by the phosphatidylinositol 3-kinase pathway: involvement of protein
kinase B/Akt. J Biol Chem. 1998 Sep 25;273(39):25420—6.

180

 

220.

221.

222.

223.

224.

225.

226.

227.

228.

229.

230.

23].

Wang D, Wei Y, Pagliassotti MJ. Saturated fatty acids promote endoplasmic

reticulum stress and liver injury in rats with hepatic steatosis. Endocrinology. 2006
Feb; 147(2):943-5 1.

Wang Z, Wang Y, Xuan J, Dong Y, Bakay M, Feng Y, Clarke R, Hoffman EP.
Optimized multilayer perceptrons for molecular classification and diagnosis using
genomic data. Bioinformatics. 2006 Mar 15;22(6):755-61.

Wei Y, Wang D, Topczewski F, Pagliassotti MJ. Saturated fatty acids induce
endoplasmic reticulum stress and apoptosis independently of ceramide in liver cells.
Am J Physiol Endocrinol Metab. 2006 Aug;29l(2):B275-81.

Weinberg JM. Lipotoxicity. Kidney Int. 2006 Nov;70(9):1560-6.

Weiss JN, Korge P, Honda HM, Ping P. Role of the mitochondrial permeability
transition in myocardial disease. Circ Res 2003;93:292-301.

Whiteman M, Rose P, Siau JL, Cheung N8, Tan G8, Halliwell B, Armstrong JS.
Hypochlorous acid-mediated mitochondrial dysfimction and apoptosis in human

hepatoma HepG2 and human fetal liver cells: role of mitochondrial permeability
transition. Free Radic Biol Med 2005;38:1571-1584.

Wieckowski MR, Wojtczak L. Fatty acid-induced uncoupling of oxidative
phosphorylation is partly due to opening of the mitochondrial permeability
transition pore. FEBS Lett 1998;423:339-342.

Wild AC, Moinova HR, and Mulcahy RT. Regulation of gamma -Glutamylcysteine
Synthetase Subunit Gene Expression by the Transcription Factor Ner. J Biol
Chem, Vol. 274, Issue 47, 33627-33636, November 19, 1999.

Woerle HJ, POpa E, Dostou J, Welle 8, Gerich J, Meyer C. Exogenous insulin
replacement in type 2 diabetes reverses excessive hepatic glucose release, but not
excessive renal glucose release and impaired free fatty acid clearance. Metabolism
2002;51:1494-1500.

Wojtczak L, Schonfeld P. Effect of fatty acids on energy coupling processes in
mitochondria. Biochim Biophys Acta 1993;] 183 :41 -57.

Yamagishi 8-i, Okamoto T, Amano S, Inagaki Y, Koga K, Koga M, Choei H, et al.
Palmitate-induced apoptosis of microvascular endothelial cells and pericytes.
Molecular Medicine (Baltimore, MD, United States) 2002;82179-184.

Yamaguchi M, Miyashita Y, Kumagai Y, Kojo 8. Change in liver and plasma

ceramides during D-galactosamine-induced acute hepatic injury by LC-MS/MS.
Bioorg Med Chem Lett 2004;14:4061-4064.

181

 

 

232.

233.

234.

Yu MU, Yoo JM, Lee YS, Lee YM, Hong JT, Oh KW, Song 8, et al. Altered de
novo sphingolipid biosynthesis is involved in the serum deprivation-induced cell
death in LLC-PKI cells. J Toxicol Environ Health A 2004;67:2085-2094.

Yuan YV, Kitts DD, Godin DV. Variations in dietary fat and cholesterol intakes
modify antioxidant status of SHR and WKY rats. J Nutr 1998;128:1620-1630.

Zegura B, Sedmak B, Filipic M. Microcystin-LR induces oxidative DNA damage in
human hepatoma cell line HepG2. Toxicon 2003;41:41-48.

182

 

 

IIIIIIIIIIIIIIIIIIIIIIIIIIIIII

ll|lllllllIllllllllllllllllllllllllllllllll