The effects of hyperglycemia on glucose metabolism
in epithelial ovarian cancer
by
Lisa Danielle Kellenberger
A Thesis
presented to
The University of Guelph
In partial fulfillment of the requirements
for the degree of
Doctor of Philosophy
in
Biomedical Science
Guelph, Ontario, Canada
© Lisa Kellenberger, February, 2016
ABSTRACT
THE EFFECTS OF HYPERGLYCEMIA ON GLUCOSE METABOLISM
IN EPITHELIAL OVARIAN CANCER
Lisa Danielle Kellenberger
University of Guelph, 2016
Advisor:
Dr. Jim Petrik
Glucose metabolism in cancer cells is uniquely adapted to maximize the production of
both ATP and the precursors needed to support cell proliferation and tissue growth.
Glycolysis, an anaerobic process normally limited by the presence of oxygen, proceeds
at an elevated rate in both aerobic and anaerobic conditions. Glycolysis consumes large
quantities of glucose while producing relatively small amounts of energy. The
inefficiency of this process is not detrimental to the cell when glucose is abundant.
However, evidence suggests that in cancer the normal concentration of circulating
glucose does not meet the energy demands of the tumour and is therefore a limiting
factor in cancer cell metabolism. Hyperglycemic conditions such as diabetes are
becoming common comorbidities in cancer patients and are associated with increased
risk and poorer prognosis of epithelial ovarian cancer (EOC). We hypothesise that
elevated blood glucose permits tumour metabolism to function at maximal capacity
thereby facilitating tumour growth and metastasis.
The goal of this thesis was to examine the metabolic response and glucose transport
kinetics of EOC in hyperglycemic environments. By inducing ovarian tumour growth in
mouse models of Type 1 and Type 2 diabetes, we found that accelerated disease
progression occurs in a glucose concentration-dependent manner.
In addition, for the first time we show that EOC cells express not only the passive
glucose transporters (GLUTs), but also a class of active glucose transporters, the
sodium-glucose symporters (SGLTs). Glucose uptake into EOC cells in hyperglycemia
is mediated by this sodium-dependent glucose transport. Surprisingly, SGLT2 appears
to be a tumour suppressor: knockdown of SGLT2 in ovarian cancer cells increases the
relative risk of death in tumour-bearing mice by more than 50 times.
PET scans measuring accumulation of the glucose analogue fluoro-deoxyglucose
(FDG) are a vital component of EOC treatment. However, FDG is a poor substrate for
SGLTs, suggesting that current imaging fails to detect a large fraction of glucose
transport into tumours. The use of SGLT-specific glucose analogues, particularly in
patients with metabolic dysfunction, presents an opportunity to enhance the power of
PET imaging and provide a more complete picture of tumour metabolism which may
have therapeutic benefits.
Acknowledgements
Dr. Jim Petrik, Esteemed Professor, who always puts things in perspective. Thank you.
It’s been righteous.
Dr. Alison Holloway, a mentor, advocate, and voice of reason for me since I was 17.
My advisory committee: Dr. Jon Lamarre, Dr. Roger Moorehead and Dr. Lindsay
Robinson.
Maka Natsvlishvili, an expert and a wonderful teacher. Everyone at the Central Animal
Facility for their help, support and genuine interest in these projects.
Michelle Ross, who always listened and commiserated.
Dr. Jim Greenaway, the ultimate scientist, cynic, and skeptic.
Amanda Kerr, who shared the science happy-dance. Sam Russell, Jackie Dynes, and
Funk Nights. Simone ten Kortenaar, the dancer, whistler, and optimist. Kata Osz, an
artist, technician and general badass. Josh Antunes, for the techno and the drifting.
Dr. Jenny Bruin, who trusted me with her work.
Dr. Amy McPhedran, who always did more than her job.
Mom, Dad, Nonnie and Grandpa for their unconditional support. Mark and Kevin who
have been friends, foes, defenders, and great little brothers.
Kelsey, who promised to read my “book”.
Labmates, officemates, roommates.
The pool, the road, the gym.
The Science Gods (sometimes).
Generous funding from a CIHR Vanier Canada Graduate Scholarship.
Those damned mice.
iv
Declaration of work performed
I declare that with the exception of the items listed below, all work reported in this thesis
was performed by me.
All tumour induction surgeries were performed with the assistance of Dr. Jim Petrik. Dr.
Jim Petrik and Dr. Jim Greenaway assisted with some of the tissue collection. Insulin
staining on pancreas sections in Figure 2A was done by Michelle Ross and serum
insulin quantification in Figure 8A was done by Nicole DeLong in the lab of Dr. Alison
Holloway at McMaster University. Intravenous injections of radiolabelled glucose
analogues (Chapter 3) were performed by Jennifer Randall of the University of Guelph
Central Animal Facility. Statistical analyses of mouse data in Chapter 3 were performed
with the help of Dr. Cate Dewey. Kata Osz stained the tumour tissue slides shown in
Figure 21B and C. The human ovarian tumour tissue microarray (Chapter 3) was
imaged at the Advanced Optical Microscopy Facility (AOMF) at the University Health
Network in Toronto.
v
TABLE OF CONTENTS
Declaration of work performed .................................................................................... v
LIST OF TABLES ...........................................................................................................ix
LIST OF FIGURES.......................................................................................................... x
LIST OF ABBREVIATIONS .......................................................................................... xiii
CHAPTER 1: INTRODUCTION AND LITERATURE REVIEW ....................................... 1
Ovarian cancer ........................................................................................................... 1
Histological classifications ........................................................................................ 1
Screening and prevention ......................................................................................... 2
Tumour development ................................................................................................ 3
Diagnosis and clinical staging ................................................................................... 4
Treatment and disease management ....................................................................... 6
Current challenges.................................................................................................... 6
Glucose metabolism .................................................................................................. 7
Glycolysis ................................................................................................................. 7
Cellular Respiration .................................................................................................. 8
The citric acid cycle ............................................................................................... 8
The electron transport chain and oxidative phosphorylation.................................. 8
Anaerobic glycolysis (fermentation) .......................................................................... 9
Glucose metabolism in cancer ................................................................................. 9
The Warburg Effect............................................................................................... 10
Mitochondria ........................................................................................................ 11
Hypoxia ............................................................................................................... 12
Tumour acidification ............................................................................................ 13
Speed of metabolism ........................................................................................... 13
vi
Metabolic intermediates....................................................................................... 14
Primary genetic changes ..................................................................................... 15
Hexokinase 2 (HK2) ............................................................................................ 16
High interstitial fluid pressure .............................................................................. 16
Metabolic flexibility............................................................................................... 17
Contributions of the tumour microenvironment ................................................ 18
The Crabtree Effect ............................................................................................. 18
Glucose transport .................................................................................................... 19
GLUTs .................................................................................................................... 19
SGLTs .................................................................................................................... 21
Positron emission tomography (PET) .................................................................... 24
Principle .................................................................................................................. 25
Limitations .............................................................................................................. 25
Use in gynecological cancers ................................................................................. 27
RATIONALE ................................................................................................................. 28
Hypothesis and objectives ...................................................................................... 28
CHAPTER 2: SYSTEMIC HYPERGLYCEMIA AND EOC PROGRESSION ................ 29
Introduction .............................................................................................................. 29
Materials and Methods ............................................................................................ 31
Results ...................................................................................................................... 39
Discussion ................................................................................................................ 58
CHAPTER 3: GLUCOSE TRANSPORT IN HYPERGLYCEMIC EOC ......................... 65
Introduction .............................................................................................................. 65
Materials and Methods ............................................................................................ 67
Discussion .............................................................................................................. 106
vii
CHAPTER 4: ANTIDIABETIC EFFECTS OF METFORMIN AS EOC THERAPY ...... 114
Introduction ............................................................................................................ 114
Materials and Methods .......................................................................................... 117
Results .................................................................................................................... 123
Discussion .............................................................................................................. 143
CHAPTER 5: GENERAL DISCUSSION ..................................................................... 149
Overview ................................................................................................................. 149
The metabolic syndrome, diabetes, and EOC ..................................................... 151
Coordinated metabolism in the tumour microenvironment ............................... 154
Intersection of tumour metabolism and systemic metabolism.......................... 155
Glucose transport and diagnostic potential ........................................................ 157
Study Limitations ................................................................................................... 158
Conclusions ........................................................................................................... 160
REFERENCES ............................................................................................................ 161
Appendix I: Supplementary figures ......................................................................... 184
Appendix II: Source of supplies and materials ....................................................... 188
Appendix III: Recipes for solutions ......................................................................... 192
Appendix IV: RT-PCR primer sets............................................................................ 196
viii
LIST OF TABLES
Table 1: Prevalence of ovarian tumours by histology ...................................................... 2
Table 2: Frequency of EOC diagnoses by FIGO stage ................................................... 5
Table 3: SGLT protein expression in human ovarian cancer ......................................... 81
Table 4: Cox proportional hazard regression model for ovarian cancer survival ......... 102
Table 5: Median survival times of mice with SGLT2 KD tumours ................................ 103
ix
LIST OF FIGURES
Figure 1: SGLT secondary active symporters ............................................................... 22
Figure 2: Streptozotocin treatment models Type 1 diabetes ......................................... 41
Figure 3: Type 1 diabetes promotes tumour growth in proportion to blood glucose ...... 42
Figure 4: Chronic hyperglycemia limits ovarian tumour growth ..................................... 44
Figure 5: Acute and chronic hyperglycemia decrease survival in mice with EOC ......... 45
Figure 6: Tumour induction improves glucose tolerance in chronically diabetic mice ... 47
Figure 7: High blood glucose in Type 2 diabetic mice predicts larger tumours .............. 50
Figure 8: Tumours improve insulin sensitivity in Type 2 diabetic mice .......................... 51
Figure 9: EOC cells conditioned to hyperglycemic media accelerate scratch wound
closing .......................................................................................................... 54
Figure 10: Cells conditioned to hyperglycemia are sensitive to glucose deprivation ..... 55
Figure 11: ID8-25 cells, but not ID8-6 cells, increase glucose consumption in
hyperglycemia. ............................................................................................. 56
Figure 12: In high glucose media, cells conditioned to hyperglycemia have lower
metabolic viability than cells conditioned to physiological glucose ............... 57
Figure 13: GLUT and SGLT transporters are expressed in ID8 cells ............................ 79
Figure 14: SGLT2 is overexpressed in human serous ovarian cancer .......................... 80
Figure 15: SGLTs affect glucose transport in hyperglycemia in cells conditioned to
normal physiological glucose concentrations (ID8-6) ................................... 83
Figure 16: SGLTs affect glucose transport in hyperglycemia in cells conditioned to high
glucose concentrations (ID8-25)................................................................... 84
Figure 17: SGLT inhibition does not affect cell metabolic viability ................................. 86
x
Figure 18: Phlorizin and phloretin inhibit scratch wound healing in ID8-6 cells ............. 88
Figure 19: SGLT glucose transporters limit cell invasion in ID8-6 cells ......................... 90
Figure 20: Knockdown of SGLT2 in ID8 cells changes proliferative behaviour in culture
..................................................................................................................... 92
Figure 21: GLUT1 protein expression is higher the tumours of diabetic mice than in
those of WT mice ......................................................................................... 94
Figure 22: Both GLUT and SGLT transporters are responsible for functional glucose
uptake into tumours ...................................................................................... 96
Figure 23: SGLT2 KD impairs glucose uptake through active transport, but total glucose
uptake is maintained through compensation by GLUTs ............................... 97
Figure 24: Expression of GLUT1 is highly upregulated in tumours with reduced SGLT2
expression at 45 days PTI ............................................................................ 99
Figure 25: SGLT2 knockdown leads to significantly larger primary tumours at death . 101
Figure 26: SGLT2 KD in tumours significantly decreases overall survival................... 102
Figure 27: Secondary disease is much more severe in mice with SGLT2 KD tumours
................................................................................................................... 104
Figure 28: Poor overall survival is associated with severe secondary disease,
regardless of primary tumour weight, in mice bearing SGLT2 KD tumours 105
Figure 29: Cellular effects of metformin....................................................................... 116
Figure 30: Metformin inhibits EOC cell viability in high glucose .................................. 125
Figure 31: Metformin reduces cell number in glucose-deprived cultures .................... 127
Figure 32: Metformin increases cellular glucose consumption .................................... 129
xi
Figure 33: Metformin-mediated increase in glucose consumption is independent of
glucose transporter expression .................................................................. 130
Figure 34: Small tumours normalize blood glucose in diabetic mice ........................... 133
Figure 35: Daily metformin treatment does not affect EOC outcomes in WT mice ...... 135
Figure 36: Metformin treatment in WT mice with tumours impairs glucose tolerance over
time. ........................................................................................................... 136
Figure 37: Daily metformin treatment does not affect EOC outcomes in Type 2 diabetic
mice ............................................................................................................ 139
Figure 38: The presence of a tumour or treatment with metformin both improve glucose
tolerance in diabetic mice ........................................................................... 140
Figure 39: Metformin treatment does not reduce the weight of established tumours .. 142
Figure 40: EOC metabolism is a product of cell genotype, the tumour microenvironment,
and systemic glucose conditions ................................................................ 151
Figure 41: Positive SGLT1 staining on human ovarian tumour tissue ......................... 184
Figure 42: Dapagliflozin treatment does not affect tumour weight ............................... 186
xii
LIST OF ABBREVIATIONS
ABAM
antibiotic/antimycotic
AGE-RAGE advanced glycation end product receptor complex
AMG
14C-α-methyl-d-glucopyranoside
ANOVA
analysis of variance
ATP
adenosine triphosphate
BRCA
breast cancer
CB
cytochalasin B
DAB
3,3’-diaminobenzidine
DAPI
4’,6-diamidino-2-phenylindole
DMEM
Dulbecco’s Modified Eagle Medium
DNA
deoxyribonucleic acid
EOC
epithelial ovarian cancer
ETC
electron transport chain
FBG
fasting blood glucose
FBS
fetal bovine serum
FDG
fluoro-deoxyglucose
FIGO
International Federation of Gynecology and Obstetrics
G6P
glucose-6-phosphate
GLOX
glucose oxidase
GLUT
facilitative glucose transporter
IPGTT
intraperitoneal glucose tolerance test
HIF-1α
hypoxia inducible factor alpha
HK
hexokinase
HRP
horseradish peroxidase
ID8
mouse surface epithelial ovarian cancer cells
IGF
insulin-like growth factor
IHC
immunohistochemistry
LDH
lactate dehydrogenase
KD
knockdown
xiii
KO
genetic knockout
MTT
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide
MRI
magnetic resonance imaging
OXPHOS
oxidative phosphorylation
PBS
phosphate buffered saline
PET
positron emission tomography
PFK
phosphofructokinase
PK
pyruvate kinase
PPP
pentose phosphate pathway
PT
phloretin
PTI
post tumour induction
PZ
phlorizin
RIPA
radioimmunoprecipitation assay
RT-PCR
polymerase chain reaction
SDS PAGE sodium dodecyl sulfate polyacrylamide gel electrophoresis
SEER
Surveillance Epidemiology and End Results (National Cancer Institute)
SGLT
sodium-glucose transporter
STIC
serous tubal intraepithelial carcinoma
STZ
streptozotocin
T1DM
Type one diabetes mellitus
T2DM
Type two diabetes mellitus
TBST
Tris-buffered saline with tween20
TCA
the citric acid cycle
TZD
thiazolidinedione
VDAC
voltage-dependent anion channel
WB
Western blot
WHO
World Health Organization
WST-1
water-soluble tetrazolium salt 1
WT
wild type
xiv
CHAPTER 1: INTRODUCTION AND LITERATURE REVIEW
Ovarian cancer
Currently there are approximately 17,000 Canadian women living with ovarian cancer.
Each year nearly 2,800 new cases are diagnosed and the disease claims the lives of
1,750 women, making it the most deadly gynecologic cancer and the fifth leading cause
of all cancer deaths in women.1,2 Ovarian cancer most commonly affects postmenopausal women (median age at diagnosis is 62) and both incidence and mortality
increase with age. In the general population the lifetime risk of developing ovarian
cancer is 1.4% (1 in 71).2 BRCA gene mutation carriers are at a significantly increased
risk (BRCA1: 30-60%, BRCA2: 15-30%)3 and cases in these women appear ten years
earlier on average.4
Histological classifications
Ovarian tumours can have stromal, germinal, or epithelial origins. Epithelial tumours are
by far the most prevalent and make up close to 90% of ovarian cancers. The World
Health Organization (WHO) has identified eight histologic subtypes of epithelial ovarian
cancer (EOC): serous, endometrioid, mucinous, clear cell, transitional cell, squamous
cell, mixed epithelial and undifferentiated.5 Table 1 indicates the prevalence of ovarian
tumours based on histopathology.6
1
Table 1: Prevalence of ovarian tumours by histology
Origin*
Tumour
Malignancy*
Histology
Benign 60%
Borderline 10%
High-grade serous 70%
Epithelial 60%
Low-grade serous <5%
Malignant 30%
85-90%
Mucinous 3%
Endometrioid 10%
Clear cell 5-10%
Germ cell 30%
Malignant
<5%
Sex-cord
8%
stromal
Malignant
<5%
* Remaining percentages of disease have undefined pathologies
It is becoming clear that the histopathological subtypes of ovarian cancer are also
associated with particular mutations.7 One of the most significant advances in the past
decade has been the development of more descriptive classifications of ovarian cancer
subtypes that also include cells of origin and genetic signatures (reviewed in 6).
Screening and prevention
Disease screening and prevention have been extremely challenging in EOC. Up to 90%
of EOCs do not have an identified genetic component and there are no easily
identifiable precursor lesions. Traditionally, the most successful clinical tools have been
2
transvaginal ultrasound and identification of elevated levels of the serum marker CA125, however neither is sensitive or specific enough for screening.7 In women at high
risk for the disease, primarily BRCA mutation carriers, risk-reducing salpingooophorectomy is the most protective strategy.8
Tumour development
For many years, epithelial ovarian cancer was suspected to arise from the continual
disruption of the ovarian surface epithelium during ovulation. According to this
“incessant ovulation” hypothesis, repeated post-ovulatory follicle repair9–12 and its
associated inflammation13 enhance the possibility of mutagenesis. Continual follicle
repair also increases the frequency of cortical inclusion cysts which are invaginations of
the surface epithelium that expose naïve epithelial cells to the hormone-rich, thus
potentially tumourigenic, ovarian stroma.14 This hypothesis supports many of the clinical
observations related to disease risk factors: in general, circumstances that increase the
number of ovulatory events are associated with increased risk of ovarian cancer, such
as nulliparity, history of pelvic inflammatory disease and endometriosis. Conversely,
fewer ovulatory events are generally associated with decreased risk of the disease,
including oral contraceptive use and multiple pregnancies.15
The understanding of ovarian tumour development has been complicated by the fact
that “epithelial ovarian cancer” has historically encompassed a histologically diverse
group of pelvic tumours.16 As a way to reconcile this diversity of tumours Kurman and
others used an integrated analysis of clinical, pathological and molecular genetics to
3
propose a two-pathway model that divides ovarian cancers into Type I, low-grade and
borderline lesions; and Type II, high-grade aggressive tumours.17 Characterization of
these pathways continues to improve as new genomic and microenvironmental data
become available.9
The past decade has seen a substantial paradigm shift in the understanding of the
cellular origins of EOC. Despite the fact that tumours were thought to arise from the
ovarian epithelium, both tubal ligation and hysterectomy were found to provide a 67%
reduction in the risk of developing ovarian cancer.15 In 2001, Piek and others noticed
that the fallopian tubes of women carrying BRCA mutations often had dysplastic regions
and occult cancers18 and shortly after, Medeiros and others implicated the fimbriae of
the fallopian tubes as sites development of early serous tumours.19 In one study 71% of
cases diagnosed as “serous ovarian carcinoma” showed lesions in the tubal epithelium
known as serous tubal intraepithelial carcinomas (STICs).20 It is now accepted that
many serous tumours originally considered as ovarian in origin actually arise from the
fimbriated ends of the fallopian tubes.21 Current mouse models provide direct evidence
that metastatic serous “ovarian” cancer can develop from primary fallopian tube tumours
that grow to the engulf the ovary.22
Diagnosis and clinical staging
Diagnosis of ovarian cancer is complicated by the fact that its symptoms are ambiguous
and often resemble gastrointestinal problems.23,24 The most significant barrier to
improving ovarian cancer outcomes is the inability to effectively identify the disease in
4
its early stages. The five-year survival rate drops precipitously once the disease has
spread beyond the ovary, yet the majority of cases are not detected until after this point.
At diagnosis, EOC is most commonly characterized as Stage III or IV using the
International Federation of Gynecology and Obstetrics (FIGO) staging system (Table
2).25 The most current FIGO staging criteria reflect the importance of histological
type/grading at the time of staging and the identification of STICs at the primary site. 26
Table 2: Frequency of EOC diagnoses by FIGO stage
FIGO Stage
Description
Percent
at diagnosis*
5-year
survival
I
Confined to ovary/fallopian
tube and peritoneal fluid
15%
90%
II
Extension/metastasis to
other pelvic organs or
peritoneum
19%
70%
III
Spread along both pelvic
and abdominal peritoneal
surfaces
IV
39%
60%
Distant metastases
17%
* Remaining percentages of disease have undefined stages
The spread of EOC is facilitated by the circulation of peritoneal fluid that carries
particularly aggressive cells from the primary tumour to other pelvic and abdominal
surfaces.10,27 Close to 40% of EOC patients present with malignant abdominal ascites
that plays a critical part in chemoresistant and recurrent disease.28
5
Treatment and disease management
Despite the fact that EOCs are so histologically diverse, the same treatment regime is
generally used on tumours of all subtypes: cytoreductive surgery followed by several
cycles of carboplatin and paclitaxel chemotherapy.29 The goal of the initial surgery is
histopathological diagnosis, staging, and maximum removal of tumour tissue. 29
Minimum residual disease is associated with longer survival30 and it has been
suggested that a low volume of disease after surgery may be a more important
prognostic factor than an initial low stage.17,31 Recommended surgery may include total
hysterectomy, bilateral salpingo-oophorectomy, tumour debulking, and omentectomy.32
Because ovarian cancer usually presents as peritoneal disease, surgery often examines
structures adjacent to the ovaries including the fallopian tubes.33 Following surgery and
chemotherapy, the disease is monitored using medical imaging and/or serum markers
such as CA-125.7 Most women who succumb to the disease die of malignant bowel
obstruction from widespread metastatic disease and limited mortality is associated with
the primary ovarian tumour.29
Current challenges
Despite advances in treatment over the past forty years, death rates from ovarian
cancer have not changed appreciably between 2001 and 2012.25 In addition to late
detection and disease spread within the peritoneal cavity, treatment challenges include
drug resistance and cancer recurrence even after initial response to treatment. On
average, advanced disease recurs within 18 months of treatment.29 Patients are given
platinum-based chemotherapies until their tumours become resistant, at which point
6
doxorubicin or topotecan is recommended.34 The median survival of patients with
platinum-resistant, recurrent disease is about three years, and just one year from the
onset of platinum resistance.35 According to a recent international consortium, improving
patient outcomes will come from using distinct histological and genetic subtypes to:
stratify clinical trial design; identify new therapeutic targets; develop better experimental
models; and design tools for prevention and early detection.36
Glucose metabolism
In order to survive and replicate, the cells in ovarian tumours rely on the oxidation of
glucose as their primary source of ATP energy, using metabolic processes that can
occur both aerobically (oxidative phosphorylation, oxphos) and anaerobically (glucose
fermentation).
Glycolysis
The anaerobic process of glycolysis in the cell cytoplasm initiates both oxphos and
fermentation by metabolising a glucose molecule into two 3-carbon pyruvate molecules
and two molecules of ATP. The net result of glycolysis can be represented as the
following equation:
Glucose + 2 NAD+ + 2 ADP + 2 Pi  2 NADH + 2 pyruvate + 2 ATP + 2 H2O + 4 H+
The rate of pyruvate production is controlled at the levels of the regulatory enzymes
hexokinase (HK), phosphofructokinase (PFK), and pyruvate kinase (PK) in reactions
that are virtually irreversible and are allosterically regulated by their products. While the
7
efficiency of ATP formation in glycolysis is only about 3%, the glycolytic reactions also
supply the cell with various anabolic precursors. For example, in addition to continuing
through the glycolytic cycle the first product of glycolysis, glucose-6-phosphate (G6P),
can be further metabolised by G6P dehydrogenase to enter the pentose phosphate
pathway (PPP) and synthesize ribose-5-phosphate required for nucleotide
biosynthesis.37,38
Cellular Respiration
The citric acid cycle
When oxygen is present, pyruvate (the end product of glycolysis) is converted to acetyl
coenzyme A which is further metabolized through the enzymes of the citric acid cycle
(TCA), a series of reactions that take place in the mitochondria.
The net reaction of the TCA can be summarized as:
2 acetyl-CoA + 6 H2O + 2 ADP  4 CO2 + 16 H+ + 2 CoA + 2 ATP + 2 FADH2 + 6 NADH
While the TCA itself only produces two molecules of ATP, there are four dehydrogenase
enzymes in the cycle that reduce the electron carriers NAD+ to NADH and FAD+ to
FADH2. For every two electrons subsequently released from these carriers, up to three
molecules of ATP can be synthesized through the electron transport chain (ETC).
The electron transport chain and oxidative phosphorylation
Oxidative phosphorylation (oxphos) refers to the generation of ATP energy in the
mitochondria through the electron transport chain and the ATP synthetase enzyme. The
8
ETC is a series of four protein complexes located in the mitochondrial membrane,
where each consecutive protein has a higher affinity for electrons. As electrons are
released from the carriers NADH and FADH2 and move down the chain, the resulting
release of energy is used to create a proton gradient across the mitochondrial
membranes. When H+ ions are pumped from the inner mitochondrion to the outer
chamber of mitochondrion, they create a strong negative electrical potential in inner
mitochondrial matrix. Hydrogen atoms flow with this concentration gradient through ATP
synthetase to form ATP from ADP. The newly synthesized ATP diffuses through the
mitochondrial membranes to cytoplasm. In aerobic conditions, cells can produce up to
36 molecules of ATP per molecule of glucose from glycolysis and cellular respiration.
Anaerobic glycolysis (fermentation)
In the absence of oxygen, cells undergo lactic acid fermentation to regenerate the NAD+
required for glycolysis.
Glucose2 lactate + 2 H+
When oxygen becomes available again, this reaction is reversed and pyruvate + NADH
+ H+ are regenerated.
Glucose metabolism in cancer
Normal cell metabolism is under the influence of the “Pasteur effect” where oxygen
inhibits glycolysis through phosphofructokinase (PFK) regulation to facilitate the
mitochondrial oxidation of glucose. However, glucose metabolism in cancer cells does
9
not adhere to these same patterns of regulation. In 1924 the German researcher Otto
Warburg made a seminal observation of cancer cell metabolism: tumours produce
significantly more lactate than normal tissue, indicating an abnormally active glycolytic
pathway.39 The glycolytic rate in tumour cells can be more than 30 times higher than in
normal cells.40 Curiously, Warburg’s tumours appeared to have uncoupled glycolysis
from oxygen levels to maintain their highly glycolytic phenotype even in the presence of
adequate oxygen, thus demonstrating “aerobic glycolysis.”41 Although not all cancerous
cells have the same abnormalities in metabolism,42 the Warburg phenotype is so well
preserved across diverse malignancies that it is now considered an independent
hallmark of cancer.43
The Warburg Effect
Understanding the significance of the Warburg phenotype remains a challenge and it is
still not clear whether the effect is a “cause, correlate or facilitator” of a progressing
malignancy.44 Warburg originally postulated that the abnormal metabolism of cancer
cells is a result of defective mitochondria, writing that “the respiration of all cancer cells
is damaged”41 – despite the fact that many investigators, including Warburg himself,
found that cancer tissues maintain normal levels of respiration in addition to generating
high levels of lactic acid.39,45–47 Warburg believed that impaired respiration was the
ultimate cause of cancer, thus only cancer cells would show this property, even though
significant evidence showed that aerobic glycolysis is common to many rapidly
proliferating cell populations.48 Rather than damage to respiration, the Warburg effect
more accurately reflects an impairment of glycolytic control.49
10
The diversity of factors associated with the development of the Warburg effect suggests
that there is no single cause or consequence of abnormal metabolism in tumours.
Parallel, reciprocal changes to genetic and environmental factors drive the evolution of
the glycolytic switch.50,51 The structure and function of cancer cell mitochondria, the
hypoxic nature of tumours, lactate-driven acidification of the tumour microenvironment,
the capacity for metabolic turnover, production of biosynthetic intermediates, primary
genetic changes, and the interstitial fluid pressure of tumours are all associated with the
metabolic profile of cancer.
Mitochondria
Both structural and functional impairment of mitochondria are significant factors in the
altered metabolism of cancer cells.52,53 Although damages to cellular respiration are not
as complete or as universal as Warburg predicted, there is evidence that mitochondria
in some highly glycolytic tumour cells have an impaired ability to transport electrons
from pyruvate to molecular oxygen.54,55 While the mitochondria are not inactive, they
may act at a low capacity56 and cancerous cells often have fewer organelles and
reduced mitochondrial DNA (mtDNA) content.57 There are mutations and copy number
changes to the mtDNA of many cancers53 in addition to nuclear DNA mutations in
proteins involved in respiration.49 The function of TCA proteins located in the
mitochondria including succinate dehydrogenase, fumarate dehydrogenase, and
isocitrate dehydrogenase are also impaired in some cancers.49,58–60 Furthermore,
inhibition of the respiratory enzyme citrate synthase can activate an epithelialmesenchymal transition that increases metastasis and proliferation.61 Taken together,
11
mitochondrial dysfunction leads to a more highly glycolytic phenotype that has been
directly linked to tumourigenicity, while increasing oxidative phosphorylation can reduce
the growth of cancer in vivo and in vitro.62
Hypoxia
The Warburg Effect is often thought to arise as an adaptation to environmental
pressures. In this case, cancer cells are not believed to be inherently glycolytic, but
rather that the phenotype of aerobic glycolysis is secondary to intermittent hypoxia.63,64
Oxygen levels in the tumour microenvironment are significantly reduced as a result of
pathologic tumour vasculature. Solid tumours inefficiently and chaotically develop new
blood vessels to support their growing mass and consequently develop large areas of
poorly perfused tissue.43 In these areas a steep oxygen gradient develops between the
vasculature and the diffusion limit of oxygen, exposing many tumour cells to hypoxic
conditions.65 A cell with an increased capacity for glycolytic metabolism would have a
survival advantage in these transient hypoxic conditions. In support of this hypothesis,
short-term hypoxia has been found to lead to the irreversible oncogenic transformation
of embryonic mouse cells by promoting aerobic glycolysis that remains and supports
growth even once normal oxygen pressure is restored.41
The cellular response to low oxygen is mediated primarily through the hypoxia-inducible
factor (HIF) signaling cascade which regulates processes including glucose metabolism,
angiogenesis, apoptosis, proliferation and survival and pH regulation. Gregg Semenza’s
group in particular has made significant contributions to the understanding of HIF’s
12
effects on metabolism.66 Oncogenic stabilization of HIF1α induces the glycolytic
proteins HK2, PFK-1, aldolase, enolase, and PKM2.40,66 Thus, hypoxia leads to the
Warburg phenotype because glycolytic products build up and exceed the capacity of
aerobic respiration, the production of lactate acid increases as Warburg observed.
Tumour acidification
A number of studies have shown that the lactate produced by increased glycolysis
contributes to the aggressiveness of tumours. In order to prevent cellular acidification,
lactate is shuttled out of the cell leading to the acidification of the tumour
microenvironment.40 Acidity has been shown to promote proliferation,63 signal67 and
facilitate68 tumour angiogenesis, suppress the immune response to a developing
tumour,69 and increase mutation rate70,71 and resistance to chemotherapy.72 The pH of
tumour tissue is around 0.5 units lower than normal, so H+ ions from the tumour diffuse
and cause damage to surrounding normal tissue, thus facilitating invasion.73–75
Mathematical modeling shows that tumour acid production alone can explain both
benign (a result of autotoxicity) and invasive tumour growth76, though the Warburg effect
is not necessarily associated with metastatic potential.77
Speed of metabolism
Cancer cells produce about 10% more total ATP than normal cells to support their rapid
proliferation.49 When tissues are rapidly consuming energy, most of it is regenerated by
the high turnover of anaerobic glycolysis.38 Although this is a relatively inefficient
13
process, if the cell is not limited by glucose availability aerobic glycolysis can supply
ATP up to 100 times faster than oxphos.40 The production of lactate is necessary to
sustain a high rate of glycolysis. The coenzyme NAD+ is required for the GAPDH
reaction in the fifth glycolytic reaction. In oxphos NAD+ is regenerated by the electron
transport chain, a process that is too slow to sustain cancer cell metabolism. As a
consequence, tumour cells must regenerate NAD+ by converting pyruvate into lactate
through lactate dehydrogenase (LDH).
Metabolic intermediates
In order to proliferate, the cell not only requires ATP energy, but also sufficient
macromolecule biosynthesis and maintenance of redox balance.51 High rates of
glycolysis have been shown to increase the production of biosynthetic building blocks
including proteins, phospholipids, fat and nucleic acids.78,79 The ability of cells to
produce these precursors allows them to become autonomous for their ability to take up
nutrients, and thus reduce their dependence on the extracellular environment. Pyruvate
kinase activity in glycolysis appears to be central to the production of anabolic
precursors. Blocking the PKM2 isoform in cancer cells from converting
phosphoenolpyruvate to pyruvate causes accumulation of upstream products which are
then available substrates for the PPP and thus cell precursors for biosynthesis.80 Other
secreted molecules from glucose metabolism, like lactate and alanine, can be used as
precursors for hepatic gluconeogenesis which may provide further fuel for the tumour.81
14
Primary genetic changes
While the Warburg effect may arise secondary to factors in the tumour
microenvironment, there is also substantial evidence to suggest that it is a result of
mutations in non-metabolic tumour suppressors or oncogenes.50,82,83 As early as the
1960s, it was noticed that the treatment of cells with an oncolytic virus increases
glucose transport beyond what is expected for rapid growth alone, suggesting a switch
to a less efficient metabolic process.84,85 More recently, a study showed that a glycolytic
phenotype emerged from progressive transformation of primary human fibroblast cells
with different oncogenes.86 The p53 tumour suppressor, which is mutated in 60-80% of
all EOC cases,3 stimulates respiration and suppresses glycolysis. Loss of p53 function
induces a Warburg phenotype through elevated activity of the glycolytic enzyme
phosphofructokinase49 and enhanced flux through the pentose phosphate pathway. 87,88
Several papers have implicated the Akt kinase pathway in cancer cells as the main
driver of Warburg effects as Akt oncogenes activate HK2 and PFK1/2, and promote
GLUT recruitment to the plasma membrane.89,90 Expression of the M2 fetal isoform of
pyruvate kinase (PK) rather than the PKM1 splice variant also plays an important role in
the glycolytic switch and results in tumorigenesis.91 Most of the research on cancer cell
metabolism and the Warburg effect has been done in solid tumours where the
microenvironment is thought to largely mediate the Warburg effect. However, the effect
also appears in hematologic malignancies: acute leukemia is dependent on glycolysis
for survival and abnormally high glucose uptake visualized by FdG-PET is used to
detect and stage Hodgkin’s disease, non-Hodgkin’s lymphomas and multiple
myeloma.92 Because the “metabolic switch” can occur independently of tumour stromal
15
tissue, it may be an inherent property of malignant cells rather than exclusively a
consequence of the microenvironment.
Hexokinase 2 (HK2)
Hexokinase 2 catalyses the formation of glucose to G6P, the first step of metabolism
through both PPP and glycolysis. HK2 has emerged as an essential mediator of the
Warburg Effect and, as the slowest step in the glycolytic pathway in some cancers,93 it
may be the most important step in determining glycolytic flux.94–96 HK2 expression has
been correlated with poor overall survival and its inhibition can restore oxidative
metabolism and reduce proliferation.97 The Pederson group has contributed much of the
research on the role of HK2 in cancer metabolism and describe hexokinase 2 as a
“facilitator and gatekeeper of malignancy.” 98,99 HK2 can translocate between the
mitochondria and cytosol in response to glucose. When it is bound to the mitochondrial
membrane by the voltage dependent anion channel (VDAC), it promotes glycolysis by
escaping allosteric inhibition by G6P.100 Pederson has proposed that the Warburg effect
results from overexpressed HK2, VDAC binding HK2 to the mitochondria, and
upregulation of the HK2 gene.99
High interstitial fluid pressure
Tumours have been described as “biological pressure pumps” due to their high
interstitial fluid pressure and microvascular pressure.101,102 This characteristic provides a
major challenge to the delivery of therapeutic agents and immune cells, while
16
simultaneously helping to propel potentially metastatic cells out of the tumour. 103 The
Van’t Hoff equation predicts that CO2 from the aerobic metabolism of glucose lowers
intracellular osmotic pressure, while lactate produced from glycolysis increases
intracellular osmotic pressure. A “Warburg ratio” (lactate:CO2) of greater than 1.0
indicates that there is relatively more glycolysis than oxphos, and osmotic pressure will
rise, driving outward currents of interstitial fluid.103 Therefore the high interstitial fluid
pressure common to solid tumours may be a direct consequence of tumour metabolism.
Metabolic flexibility
Although tumours have been described as “glucose addicted” they are in fact quite
flexible in their fuel use and metabolize amino acids, fatty acids,104 lactate,51 ketones105
and particularly glutamine.106,107 Moreover, they are not committed to glycolysis. The
Warburg effect is often misunderstood as a preference for glycolysis over respiration
despite the fact that even in the most aggressive tumour cells tested, Warburg found
almost an equal contribution of fermentation and respiration.41 In some tumour cells
mitochondria are responsible for most of the ATP production.108 This flexibility means
that for cancer cells to “starve” both respiration and fermentation must be compromised.
In normal cells, impairing only one pathway is sufficient to kill.109 Cancer cells display
different balances of glycolysis, oxidative phosphorylation and glutaminolysis depending
on glucose availability and other environmental conditions.110–113
17
Contributions of the tumour microenvironment
In solid tumours, intercellular interactions are crucial for cancer progression. Stromal
cells including fibroblasts, immune cells and vascular endothelial cells make up the
heterogeneous tumour tissue and contribute to tumour growth and resistance to
therapeutics.114 Intratumoral heterogeneity of metabolism is emerging as an essential
determinant of carcinogenesis.115 Michael Lisanti’s group has described a “reverse
Warburg effect” in which cancer-associated fibroblasts (CAFs) are metabolically
reprogrammed to perform aerobic glycolysis. This transformation allows the fibroblasts
to enrich the microenvironment with lactate, pyruvate and ketone bodies which epithelial
cancer cells then take up and metabolise through oxidative phosphorylation to
synthesize high volumes of ATP.105,116,117 To facilitate this, cancer cells also induce
increased expression of the H+-coupled lactate export protein monocarboxylate
transporter 4 (MCT4) in stromal cells.117,118 Others have shown that a metabolic
symbiosis exists between cancer cells in oxygenated and hypoxic environments within
the tumour. The expression of the MCT1 importer in oxidative cancer cells allows them
to use lactate exported by glycolytic cancer cells as a primary fuel.119,120
The Crabtree Effect
Herbert Crabtree was a contemporary of Warburg who described a local, reversible
adaptation in tumours where high levels of glucose suppress oxygen consumption and
respiration and accelerate glycolysis. Crabtree confirmed Warburg’s observation that
tumours showed abnormally high levels of glycolysis and oxygen is ineffective at
18
reducing glycolysis. However, he also observed that glycolytic activity in vitro limited
respiration on average 12% in glucose vs non-glucose solution.121 Recent studies in
Barrett’s esophagus confirm that the Crabtree effect is important to tumour development
and is related advancement of pre-malignant lesions overt cancer.122
Glucose transport
A primary consequence of the Warburg Effect is that cells have abnormally high glucose
requirements. Glucose is a large, hydrophilic molecule that requires specific
transmembrane proteins to cross the cell’s lipid bilayer, and so the expression of
glucose transporters is a key mediator of aerobic glycolysis.123 There are two classes of
transporters: facilitative glucose transporters (GLUTs) and sodium/glucose
cotransporters (SGLTs).
GLUTs
The thirteen identified mammalian GLUT isoforms are grouped into three classes based
on sequence similarities. Class I GLUTS (1,2,3,4) have been well characterized and are
the most important in human disease.123 The GLUT proteins rely on a downhill
concentration gradient to move glucose and other hexoses across the lipid bilayer. The
magnitude and direction of glucose transport is dependent on both the number of
glucose transporters and the steepness of the gradient.123 The expression of transporter
isoforms is highly tissue specific. GLUT1 is expressed in erythrocytes but is also widely
distributed in tissues throughout the body. GLUT2 is specific to the liver, pancreas and
19
small intestine; GLUT3 to the brain; and GLUT4 to insulin-responsive tissues including
skeletal muscle, cardiac and adipose tissue.
GLUT expression is integral to many metabolic and energy-sensing pathways including
HIF, c-myc, Akt and AMPK. A pool of GLUTs is stored in vesicles in the cytoplasm and
are translocated to the cell membrane as needed, most notably through insulin signaling
pathways. GLUT transport works through a model of alternate conformation. Glucose
binds to sites either on inside or outside of cell and this binding changes the protein
conformation to open to the other side of membrane.38
GLUTs in cancer. GLUTs are essential to cancer metabolism (reviewed in
124)
and their
function is well documented in the literature. Experiments that oncogenically transform
fibroblasts show that oncogenic transformation itself may trigger GLUT1 activity,40 one
of the main determinants of the Warburg effect.125
GLUT1 is highly expressed in ovarian cancer, where tumour status (benign, borderline,
or malignant) is correlated with the level of GLUT1 expression.126,127 Almost all invasive
epithelial carcinomas are positive for GLUT1, independent of stage, grade, or
histological subtype.126,128 Expression of GLUT-1 correlates with tumour proliferation
and microvessel density in EOC. In addition, patients with rapidly proliferating advanced
stage tumours overexpressing GLUT-1 have a lesser chance for optimal
cytoreduction.127 Antibodies to GLUT1 decrease proliferation, induce apoptosis in
nonsmall cell lung cancer and breast cancer cell lines, and appear to synergize with a
20
number of chemotherapeutics to enhance their apoptotic effects.79 Likewise, antisense
GLUT1 mRNA inhibits tumour growth.129
SGLTs
The sodium/glucose cotransporters work against the glucose concentration gradient
and are required to establish and maintain higher intracellular glucose
concentrations.130 The main actions of the SGLTs are in glucose reabsorption from the
proximal tubule of the kidney (primarily SGLT2) and the brush border of the small
intestine (primarily SGLT1). In the enterocyte, after being pumped into the cell, glucose
flows with its concentration gradient through GLUT2 on the basolateral side of the
enterocyte, thus moving glucose back into the blood stream (Figure 1).130 SGLTs are
also expressed in the heart, brain, liver, thyroid and skeletal muscle.131
21
Figure 1: SGLT secondary active symporters
SGLT in pathology. While there is a large body of research on the role of GLUTs in
cancer, the contributions of the second class of glucose transporters is virtually
unknown. Most of the research on SGLTs comes from outside of the cancer literature.
In the last decade, the action of SGLT2 has garnered attention as a therapeutic target in
Type II diabetes (reviewed in 132) as it increases renal glucosuria and adds to the effects
of other antidiabetic drugs to improve glucose tolerance.133 SGLTs also play essential
roles in mediating ischemic stress in the brain,134 protecting against the LPS-induced
22
apoptosis implicated in inflammatory bowel diseases (IBD) and bacterial enteritis,135 and
defending against cardiac disease.136
Isoforms and structure. SGLTs belong to the 12-member SLC5 gene family of hexose
transporters. SGLT1 and SGLT2 import sodium and glucose with 2:1 and 1:1
stoichiometry respectively and differ in the substrates they transport. SGLT1 is a high
affinity transporter that moves both glucose and galactose while SGLT2, specific for
glucose, transports at a high capacity but has a glucose affinity that is about four times
lower. SGLT3 appears to act as a glucose sensor rather than a functional transporter;
glucose binds similarly and produces an inward current of Na+, but no glucose
molecules are transported across the cell membrane.137–140
Regulation. SGLT1 upregulation is mediated by the cAMP-PKA signaling pathway and
takes several days to act.141 Over minutes, activation of cAMP pathway results in
recruitment of an intracellular reserve pool of transporters to the plasma membrane
where the number of cotransporters in the plasma membrane regulated by endo- and
exo- cytosis of vesicles.130 In cardiac tissue, SGLT1 expression and function are
regulated by insulin and leptin.136 SGLT expression and translocation to the cell
membrane is also regulated through AMPK activation.142–144 Protein-protein interaction
with extracellular domain of EGFR stabilizes SGLT-1 to prevent proteasomal
degradation.145 Because SGLT1 expression is dependent on EGFR expression, it has
been suggested that EGFR has a role in maintaining basal intracellular glucose level by
this mechanism.40,145
23
Mechanism of action. Using the electrochemical gradient maintained by the Na+/K+
ATPase pump, SGLTs are secondary active symporters that transport glucose into the
cell coupled with Na+. Glucose/sodium symport is followed by anions and water to
maintain electroneutral isotonic fluid absorption.146 In addition to Na+ driven cotransport,
glucose can enter the cell with protons although the affinity for sugar becomes orders of
magnitude lower. The SGLT protein transports with a 6-step model of internalization.
External Na+ binds, increasing transporter’s affinity for glucose. Glucose then binds,
creating a conformational change that expose Na+ and glucose to the inside of the cell.
Glucose and then sodium are released and the protein returns to its original
conformation. Transport is reversible depending on the concentration gradient of Na+,
but it is asymmetrical and favours the internalization of glucose.146
Actions other than glucose transport
SGLTs can act as uniporters to move sodium into the cell in the absence of glucose131
or glucose transport.147 In certain protein conformations, SGLT1 is also a water
cotransporter with an osmotic permeability similar to AQP0130,148,149 which has
therapeutic applications in oral rehydration therapy.150 These glucose-independent roles
may contribute to cancer cell function, particularly in maintaining osmotic balance and
cell membrane potential, both of which are altered in cancer cells.151
Positron emission tomography (PET)
Both glucose metabolism and transporter expression in cancer cells are clinically useful.
Positron emission tomography (PET) is an imaging technique that has become standard
24
in cancer diagnostics. PET takes advantage of the cancer cell’s metabolic requirements
by visualizing tumours based on their abnormally high uptake of glucose which may be
20-30 times higher than in normal cells.40 However, this imaging is not consistently
useful across a wide variety of tumours. In current practice only glucose taken up
through the GLUT transport channels is detectable130, suggesting that there is potential
to increase the strength of this imaging if SGLT-mediated transport is also visualized.
Principle
Patients are fasted overnight and given an intravenous injection of the radiolabelled
glucose analogue 18F-2-deoxy-D-glucose (FDG). FDG is taken up into cells through
GLUT transporters with similar kinetics as glucose and is phosphorylated by hexokinase
(HK) to FDG-6-phosphate, but no further.152 This property allows its accumulation in
tissue as a function of time and can be used to calculate transport and HK activity.130
Gamma rays emitted from the fluorine molecule are used to visualize the accumulation
of glucose in tissues on a PET scan.153 Tissues with high glucose transport and
hexokinase activity, like tumour cells, appear brighter on the scan. This technique is
usually combined with other imaging techniques like computerized tomography (CT) or
magnetic resonance imaging (MRI), to achieve better spatial resolution that improves
the power and utility of the technique.149,154 FDG-PET can be used as a quantitative
measurement using standardized uptake value (SUV) or ratio (SUR; uptake per dose).
Limitations
1. Tumour biology. Not all tumours have the same level of metabolic activity and up to
30% of tumours are FDG-PET-negative.155 Even within tumours up to a quarter of
25
FDG in a tumour can be from non-tumour tissue.156 Furthermore, while PET is good
at differentiating between live and dead cells, it cannot distinguish live cells with
lowered metabolism (quiescent cells).155 The technique requires that FDG
accumulates in tissues, so it is not useful in malignancies in tissues where dephosphorylation of FDG-6-phosphate allows glucose to diffuse out of cell.157 FDG
uptake is also at least partially dependent on blood flow which is often severely
compromised in tumours.158 In addition, the phase of menstrual cycle also affects
level of uptake in ovaries.159
2. Signal interference. Although patients are fasted to decrease circulating insulin
before a scan, insulin levels can affect FDG distribution, as can insulin-sensitizing
drugs such metformin.160 Chemotherapy or its associated inflammation can result in
false positive FDG uptake.161,162 Hyperglycemia can affect FDG uptake as normal
glucose provides greater competition for labelled glucose to enter the tumour.163 The
timing of imaging can have an effect as tissues have varying FDG uptake on
delayed images which can alter background levels of uptake.164 High glucose uptake
in pelvic organs, particularly the bladder, can obscure tumour metabolism in these
tissues.159
3. Sensitivity and resolution. The resolution of PET, even when combined with other
imaging modalities like CT, is poor below 0.8 cm3 and thus too low to detect early
tumours.165 It is especially poor at finding new, previously unknown metastases.166
26
There are numerous causes of false-negatives and false-positives that can make it
difficult to interpret scans.167
Use in gynecological cancers
Despite these limitations, FDG uptake is increased in both ovarian cell lines168 and
ovarian tumours,169 and FDG-PET reportedly has high specificity for ovarian lesions
larger than 5mm.159 Tumour morphology, background activity, limited spatial resolution
have been implicated where FDG-PET has performed poorly in clinical use.168
A 2009 report from Cancer Care Ontario concluded that PET adds limited value to
diagnosis of pelvic masses and does not recommend its use in the diagnosis of ovarian
cancer170,171, though in some cases the SUVmax of FDG-PET/CT can be useful for
differentiating benign or borderline from malignant tumours.170,172 FDG-PET can help
clinicians decide for or against surgery based on whether there is a localized mass or
widespread disease171 and it can help reduce the need for second-look surgeries
following first-line treatment.161 It is most useful in monitoring the response to treatment,
residual/refractory disease, and recurrence.161,166,173,174 In combination with rising serum
CA-125, FDG-PET/CT can also be sensitive and specific for early diagnosis of relapse
in otherwise asymptomatic patients.175
27
RATIONALE
Most tumours undergo a “glycolytic shift” as they become more aggressive. In contrast
to glucose metabolism through oxidative phosphorylation, glycolysis provides the cell
with rapid but inefficient ATP production. As a result, tumours consume exponentially
more glucose than normal tissue. Since circulating glucose is the primary source of fuel
for transformed cells, systemic metabolic control may be an important mediator of
epithelial ovarian cancer (EOC). As the prevalence of diabetes continues to rise, it will
be important to understand the effects of long-term comorbidities like hyperglycemia.
Hypothesis and objectives
We hypothesize that chronic hyperglycemia will accelerate the growth of ovarian
tumours by providing an abundance of fuel needed for rapid proliferation, and that
tumour metabolism may be mitigated by normalization of systemic blood glucose. The
following objectives were designed to test this hypothesis:
Objective 1. Determine the effects of hyperglycemia on EOC metabolism, growth and
survival in vitro and in a diabetic mouse model of EOC.
Objective 2. Examine the mechanisms of glucose uptake in physiologic and
hyperglycemic environments.
Objective 3. Assess the treatment potential of systemic glucose control to prolong
disease-free survival in vivo.
28
CHAPTER 2: SYSTEMIC HYPERGLYCEMIA AND EOC PROGRESSION
Introduction
Epithelial ovarian cancer (EOC) is the deadliest gynecologic cancer2 and despite
advances in treatment over the past thirty years, the overall survival rate remains below
40%.9 In the majority of cases, a diagnosis is not made until after the disease has
spread beyond the ovary and is much more challenging to treat. Although first-line
surgery and platinum-based chemotherapy can be effective, most late-stage cancers
recur within 16 to 18 months.29,176 There are limited options for second-line therapy and
most cancers will eventually become chemo-resistant.7
In addition to these challenges, comorbid diabetes affects 8-18% of cancer patients.177
Women with type 2 diabetes (T2DM) are at an increased risk of developing ovarian
cancer178 and EOC patients with pre-existing diabetes have a much poorer overall
survival rate than non-diabetic patients.179–182 T2DM is characterized by hyperglycemia
and hyperinsulinemia and is often related to obesity, which carries its own risk factors
including high body mass index (BMI), sex steroid metabolism, and inflammation.183,184
Despite substantial epidemiological evidence, the underlying mechanisms of the
cancer-diabetes association are still poorly understood.
Insulin increases the circulating, local and bioavailable growth factors including insulinlike growth factor 1, leading many to believe that hyperinsulinemia is the primary
influence on cancer risk and progression in type 2 diabetics.181,185 The role of
29
hyperglycemia has been largely overlooked, despite its significant pro-tumorigenic
potential.186
Like most malignancies, EOC cells metabolize considerably more glucose than their
normal counterparts.187 Abnormally elevated glycolytic activity, even in aerobic
conditions, is characteristic of the glucose metabolism in most cancer cells,
accompanied by the conversion of pyruvate to lactate in large amounts (the Warburg
Effect).41,43 Evidence suggests that the quantity of glucose required to maintain this
metabolic phenotype can exceed the normal glucose content of blood.188 The rate of
glucose uptake also increases with increasing extracellular glucose,123 suggesting that
its supply is a rate-limiting factor for metabolism in cancerous cells.189
Increased glucose uptake has been shown to coincide with the transition from premalignant lesions to invasive cancer and has been linked with overall tumour
aggressiveness.63,190 We hypothesized that the hyperglycemic conditions of diabetes
can remove growth restrictions imposed by limited glucose availability and lead to more
aggressive progression of epithelial ovarian cancer. In this study we investigated the in
vivo and in vitro effects of hyperglycemia on epithelial ovarian cancer cell growth.
30
Materials and Methods
Animals
Wildtype (WT) C57bl/6 mice were purchased from Charles River Laboratories
(Wilmington, MA, USA), and Akt2 null C57bl/6 mice (https://www.jax.org/strain/006966)
were purchased from Jackson Laboratories (Bar Harbour, ME, USA). An Akt2 null
mouse colony was established from breeder mice and animals were genotyped prior to
beginning experiments. Animals were housed at the Central Animal Facility at the
University of Guelph and all experiments were conducted according to Canadian
Council on Animal Care guidelines. Mice were kept on a 12h light:dark cycle and had
free access to food and water. For all experiments, adult mice were between 12 and 24
weeks old at the time of surgery.
Mouse model of epithelial ovarian cancer
Spontaneously transformed murine epithelial cells from C57Bl/6 mice (ID8) were diluted
in 5µl phosphate buffered saline (PBS) and injected under the ovarian bursa of WT or
Akt2 null mice. This orthotopic, syngeneic mouse model of EOC closely replicates
ovarian serous adenocarcinoma in women. By 90 days post-tumour induction, mice
develop large primary tumours, secondary peritoneal lesions, and significant abdominal
ascites.10 In survival experiments, mice were sacrificed when they became moribund
due to the accumulation of ascites fluid.
31
Type 1 Diabetes Mellitus (T1DM)
Streptozotocin ([2-deoxy-2-(3-(methyl-3-nitrosoureido)-D- glucopyranose]; STZ) is a
pancreatic β-cell toxin commonly used to model T1DM.191,192 Mice were treated with a
standard streptozotocin dosing schedule (Animal Models of Diabetic Complications
Consortium; AMDCC) two weeks prior to tumour induction surgery. Briefly, STZ (Sigma,
Oakville, ON) was dissolved in 50mM sodium citrate buffer (pH 4.5) made up fresh
daily. Mice received 200µl IP injections of 50mg/kg STZ or vehicle control for six
consecutive days. Blood glucose was measured ten days following the last injection and
mice were considered hyperglycemic if non-fasted glucose was above 300mg/dl
(16.7mM).
Type 2 Diabetes Mellitus (T2DM)
C57BL/6 mice homozygous for a mutant Akt2 allele have poor glucose tolerance and
reduced insulin sensitivity that result in hyperglycemia and hyperinsulinemia but are not
obese.193,194 Obesity often accompanies T2DM and is independently associated with
diabetes and cancer so this model allowed us to isolate the contributions of high
glucose and high insulin. Mice were genotyped using the following primers designed by
Jackson Laboratories: Akt2 wildtype reverse: TGCACAATCTGTCTTCATGCCAC; Akt2
common forward: ACCAACCCCCTTTCAGCACTTG; Akt2 mutant reverse:
TACACTTCATTCTCAGTATTGTTTTGC. Products are 110bp (wildtype), 277bp
(mutant), or 110bp and 277bp (heterozygote).
32
Food Consumption
Food consumption at three weeks post tumour induction (PTI) was calculated by taking
the initial weight of dry food and subtracting food weight after five days. This value was
divided by the number of mice in the cage and the number of days in the measurement
period to determine food consumed per mouse per day.
Intraperitoneal glucose tolerance tests
Mice were fasted for four hours prior to intraperitoneal glucose tolerance tests (IPGTT).
After recording baseline blood glucose, mice were given an intraperitoneal injection of
glucose (1 mg/g body weight in saline solution at a volume of 10µl/g body weight).
Blood glucose measurements were taken at 15, 30, 60, and 120 minutes after glucose
challenge using a handheld glucose monitor (Freestyle Lite, Abbott Laboratories).
Values that exceeded the measurement range of the glucose monitor were considered
27.8mM (maximum reading). Insulin levels were measured in trunk blood from fasted
animals at sacrifice using an ultra-sensitive mouse insulin ELISA kit (Crystal Chem,
Downers Grove, IL). Data are presented as the average integrated area under the curve
(iAUC) ± SEM where iAUC is the area under the curve relative to fasting blood glucose
(FBG) at time 0 (just before glucose injection). iAUC was calculated using the
trapezoidal method. The glucose:insulin ratio was used as an indication of insulin
sensitivity.195
33
Tissue collection and preparation
Mice were fasted for 4-6 hours prior to sacrifice by cervical dislocation. Following
euthanasia, trunk blood was collected and fasting blood glucose was measured
immediately using a handheld glucose monitor. The remaining blood was allowed to clot
then centrifuged to obtain serum samples and stored at -80oC. Ovarian tumours were
removed, weighed, and divided into sections for subsequent analysis. Sections were
either fixed in 10% neutral buffered formalin (Fisher Scientific, Whitby, ON) for wax
embedding and tissue sectioning, or flash frozen in liquid nitrogen and stored at -80oC
for protein extraction. Secondary disease was evaluated at time of sacrifice by
aspirating ascites and by visual assessment of the number of secondary lesions. Both
ascites volume and secondary lesions were scored qualitatively on a relative 4-point
scale with 0 indicating absence of secondary disease and 3 indicating the most
widespread disease. Mice with scores of 0 and 1 were considered “low” and scores of 2
and 3 were considered “high”.
Immunohistochemistry
In order to confirm the effects of streptozotocin, immunohistochemical detection of
insulin was performed on 5µm paraffin-embedded sections of pancreatic tissue from
STZ-treated mice. Sections were deparaffinized in xylene and rehydrated in graded
alcohol solutions. Endogenous peroxidase activity was quenched using 1% (vol/vol)
hydrogen peroxide and antigen retrieval was performed by immersing slides in 10mM
citrate buffer at 90°C for 12 minutes. Tissues were blocked in 5% (wt/vol) bovine serum
34
albumin in PBS for 10 minutes and slides were then incubated overnight with antiinsulin primary antibody (Cell Signal Technologies, Danvers, MA) diluted in 0.01M PBS
(pH 7.5) containing 2% (wt/vol) BSA and 0.01% (wt/vol) sodium azide at 4°C in a
humidity chamber. All subsequent incubations were performed at room temperature.
Anti-rabbit biotinylated secondary antibody (Sigma, Oakville, ON) was diluted in the
same buffer and incubated for 2 hours. Tissues were then washed in PBS and
incubated with avidin and biotinylated horseradish peroxidase (ExtrAvidin, Sigma,
Oakville, ON) for 1 hour. Antibodies were visualized using 3’ 3’-diaminobenzidine
tetrahydrochloride (DAB) (Sigma, Oakville, ON) and tissue was counterstained with
Carazzi’s Hematoxylin, dehydrated, and mounted with Permount (Fisher Scientific,
Whitby, ON).
Cell Culture
Mouse-derived epithelial ovarian cancer cells (ID8; kindly donated by Drs. Paul
Terranova and Kathy Roby, Kansas State University, Kansas, USA), human ovarian
cancer cells (SKOV-3; ATCC, Manassas, VA, USA), and human normal ovarian surface
epithelium (NOSE; generously donated by Dr. Jinsong Liu, MD Anderson Cancer
Center, Houston, TX) were maintained in DMEM supplemented with 10% fetal bovine
serum (FBS), 1% antibiotic/antimycotic (Gibco BRL, Burlington, ON) and 2% Lglutamine (Life Technologies, Burlington ON). All experimental treatments were
performed in serum-free DMEM.
35
ID8 and SKOV3 cells are typically cultured in DMEM that contains supra-physiological
amounts of glucose (25mM). We refer to these “diabetic” cells as ID8-25 and SKOV325. To test the effects of hyperglycemia, we created a parallel line of “normal” cells
chronically cultured in physiological glucose (6mM) that we named ID8-6 and SKOV3-6.
The -6 lines were created by passaging the -25 cells in gradually reduced
concentrations of glucose over a period of three weeks. These lines were then
maintained in 6mM glucose. To account for any differences caused by higher passage
numbers during the glucose reduction, ID8-6 and ID8-25 cells were passage-matched in
all experiments.
Viability assay
The MTT assay measures cellular reduction of a tetrazolium salt to a coloured formazan
product in an NADH- and NADPH-dependent manner and reflects primarily glycolytic
activity.196 Cells were seeded at a density of 1500 cells/well (ID8), 3000 cells/well
(SKOV3), or 4000 cells/well (NOSE) in triplicate in 96 well plates and allowed to attach
overnight. Cells were then serum starved for 6 hours and treated with different
concentrations of glucose for 48 hours. Following treatment, cells were incubated with
MTT (5mg/ml; Sigma, Oakville, ON) for one hour at 37°C and lysed with MTT lysis
buffer overnight at 37°C. Absorbance was quantified in a microplate reader at 570nm.
36
Scratch wound assay
To examine the motility of cells in different concentrations of glucose, ID8-6 and ID8-25
cells were grown to confluency in 24 well plates and serum starved for 6 hours. A plastic
pipette tip was used to remove an area of cells down the centre of the dish. Wells were
gently washed with PBS to remove detached cells before treatment. Phase contrast
images were taken at three marked points along the scratch at t=0 and t=24 hours.
Wound sizes were determined using ImageScope software and are presented as the
fraction of the original scratch covered by cells.
Glucose consumption and cell number
Glucose concentration in the cell media was measured using the glucose oxidase
method and reagents from Pointe Scientific (Canton MI, USA). After 24 hours of
treatment, glucose concentration in the sample wells was subtracted from the glucose
concentration in blank wells to determine the amount of glucose consumed. Relative
cell number was approximated by the absorbance of the crystal violet nuclear stain; to
account for any differences in the rate of cell proliferation, glucose consumption was
divided by relative crystal violet absorbance (590nm) to determine per-cell glucose
uptake.
37
Statistical Analysis
Data were analyzed using GraphPad Prism statistical analysis software. Results are
reported as mean ± standard error of the mean (SEM). Significance threshold was set
at 0.05 and p values are listed in the figure legends. Comparisons between means were
measured by two-way ANOVA followed by a Bonferroni correction, or by unpaired twotailed Student’s t-test. The relationship between tumour weight and FBG or insulin was
considered significant if the slope of a linear regression was significantly different from
zero. Overall survival data are plotted on a Kaplan-Meier survival curve and differences
were determined by a log-rank (Mantel-Cox) test.
38
Results
Mice with higher systemic glucose develop more severe disease in a model of
epithelial ovarian cancer.
Fasting blood glucose is normally maintained within a narrow physiological range
through a highly dependent relationship with insulin. To determine the insulinindependent effects of glucose on tumour growth, we used the beta-cell toxin
streptozotocin (STZ) to inhibit pancreatic insulin production and model Type 1 diabetes
(T1DM). Without the production of insulin, Type 1 diabetic mice are unable to import
glucose into cells and thus have significantly elevated fasting blood glucose (18.81 ±
1.93 mM vs 6.41 ± 0.27 mM in controls). As in T1DM, STZ-treated animals also have
lower body weights despite consuming more food than their untreated counterparts
(Figure 2C).
We treated mice with STZ two weeks prior to inducing ovarian tumours by sub-bursal
injection of syngeneic malignant cells (ID8). Blood glucose levels remained significantly
elevated until the end of the study at 90 days post-tumour induction (PTI) (Figure 3A).
Although diabetic patients are reported to have higher operative morbidity,197 we did not
find that STZ-treated mice suffered any adverse events related to hyperglycemia either
at surgery or during subsequent incision healing.
Mice were sacrificed at 30, 60 and 90 days after tumour initiation. STZ-treated animals
developed significantly larger tumours than control mice as early as 30 days PTI (Figure
3B) and tumour size was proportional to fasting blood glucose (Figure 3C).
Furthermore, STZ-treated hyperglycemic mice had a greater burden of secondary
39
disease than normoglycemic mice as measured by the number of secondary tumour
lesions disseminated throughout the peritoneal cavity (Figure 3E). The accumulation of
abdominal ascites (Figure 3D), another indicator of secondary disease, was unchanged
with STZ treatment.
40
Figure 2: Streptozotocin treatment models Type 1 diabetes
A) Representative insulin staining (brown) of pancreas sections from mice four weeks
after STZ treatment (two weeks post-tumour induction). Arrows in STZ image indicate
pancreatic islets. 100x magnification. B) Fasting blood glucose prior to tumour-induction
surgery. Bars are mean + SEM, n=8. ***p<0.001 vs mice without STZ treatment
(control). C) Food consumption at 3 weeks post tumour induction (PTI). Bars are mean
+ SEM, n≥3. ***p<0.0001 vs mice without STZ treatment (control). D) Body weights of
control and STZ treated animals. Values are mean ± SEM, n=4-16. *p<0.05.
41
Figure 3: Type 1 diabetes promotes tumour growth in proportion to blood glucose
A) Fasting blood glucose in tumour-bearing mice with or without STZ-induced
hyperglycemia. Values are mean ± SEM, n=4-16. *p<0.05. B) Mean tumour weights at
30, 60 and 90 days post tumour induction (PTI). Bars are mean + SEM n≥4. *p<0.05. C)
Relationship between fasting blood glucose and tumour weight at 60 days PTI. Each
point represents a single animal. Open circles represent control mice, closed circles
represent STZ-treated mice. n≥6. Slope is significantly non-zero by linear regression
(p<0.0001). D) Qualitatively assessed ascites volume and E) secondary lesions. Values
are proportion of mice with relatively “high” or “low” ascites volume or number of
secondary lesions. n=4.
42
Chronic glucose deficiency in mice interferes with ovarian tumour growth.
Type 1 Diabetes, formerly known as juvenile-onset diabetes, is commonly diagnosed in
youth and adolescents and leads to glucose-related tissue damage in patients as young
as eight years old.198 In addition to the acute hyperglycemia modeled by STZ treatment
two weeks prior to tumour induction, we wanted to assess the impact of type 1 diabetic
tissue damage on tumour development. To investigate the effects of long-term (chronic)
hyperglycemia, we treated mice with STZ 15 weeks before introducing ovarian tumours.
These mice had significantly higher fasting blood glucose (10.58 ± 0.59 mM vs 5.03 ±
0.29 mM in control mice) at surgery and the hyperglycemia was maintained until mice
became moribund (Figure 4D). Six mice were excluded from the study due to
complications of diabetes.
Surprisingly, we found that at death, hyperglycemic mice had smaller tumours (Figure
4A) than normoglycemic mice, in addition to more ascites (Figure 4B) and fewer
secondary lesions (Figure 4C). Furthermore, in four of the nine STZ-treated mice that
developed ascites, the abdominal fluid was pale and watery (not shown) unlike the
hemorrhagic ascites typical to this mouse model. In contrast to the effects of short-term
STZ treatment, there was an inverse relationship between fasting blood glucose at
surgery and tumour weight at death (Figure 4E). Fasting blood glucose at death was
unrelated to tumour weight (Figure 4F). Despite the development of smaller primary
tumours and fewer secondary lesions in these mice, both long-term and short-term
hyperglycemia led to decreased survival times compared to control mice (Figure 5).
43
Figure 4: Chronic hyperglycemia limits ovarian tumour growth
Mice were treated with STZ or vehicle control 15 weeks prior to tumour induction
surgery to model chronic Type 1 diabetes. A) Tumours were removed and weighed
when mice became moribund. Values are mean + SEM, n≥9. ***p<0.001 vs control.
Secondary disease was evaluated at time of death by B) ascites volume and C) number
of secondary lesions. Values are proportion of mice with “high” or “low” scores. n≥9. D)
Fasting blood glucose at time of surgery and when mice became moribund. Bars are
mean + SEM, n=9 control, n=19 STZ. Different letters indicate significant (p<0.05)
differences between bars. E) Relationship between fasting blood glucose at tumour
induction surgery and tumour weight at death. n=7 control, n=15 STZ. Each point
represents a single animal. Slope is significantly non-zero by linear regression
(p<0.001). F) Relationship between fasting blood glucose and tumour weight at death.
n=9 control, n=10 STZ. Slope is not significantly different from zero.
44
Percent survival
100
Acute control
Acute STZ
Chronic control
Chronic STZ
50
0
0
20
40
60
80
100
Days PTI
Figure 5: Acute and chronic hyperglycemia decrease survival in mice with EOC
Kaplan-Meier survival curve of acutely (dashed lines) and chronically (solid lines)
diabetic mice (Type 1 diabetes). There is no statistical difference in survival between
mice in acute or chronic groups. STZ-treated mice had a significantly poorer overall
survival than control (nondiabetic) mice. Log-rank (Mantel-Cox) between STZ and
controls, p<0.05. n≥8.
45
The development of ovarian tumours improves impaired glucose tolerance in
chronically hyperglycemic mice.
In addition to elevated fasting blood glucose, diabetes is associated with glucose
intolerance, or an impaired ability to clear glucose from the blood in response to a
glucose challenge. To better characterize the effects of systemic hyperglycemia on
EOC, we performed intraperitoneal glucose tolerance tests (IPGTT). After a period of
fasting, mice were injected with 1 mg/g body weight glucose in saline. Blood glucose
was then measured at 15, 30, 60 and 120 minutes after injection and the integrated
area under the plot of blood glucose was calculated in relation to the baseline glucose
level (integrated area under the curve, iAUC). One week after tumour induction, STZtreated mice had a significantly higher iAUC of blood glucose (Figure 6A and B) than
normoglycemic mice, indicating poorer glycemic control. Interestingly, by 55 days PTI
STZ-treated mice showed significant improvements in glucose tolerance that had
reduced iAUC values to normal (Figure 6C).
46
Figure 6: Tumour induction improves glucose tolerance in chronically diabetic mice
A) and B) Blood glucose is higher in STZ-treated mice (closed circles) than in control
mice (closed circles) at all timepoints (p<0.05) following an intraperitoneal glucose
challenge. Values are mean ± SEM, n≥9. C) Integrated area under the curve (iAUC) of
glucose tolerance graphs in A) and B). Different letters indicate significant differences
between bars (p<0.05).
47
Mice with Type 2 diabetes grow larger tumours and die earlier in a model of
epithelial ovarian cancer.
T2DM is a prevalent health concern that has been correlated with many cancers. 199 In
contrast to T1DM which is a condition of insulin deficiency, Type 2 diabetes mellitus
(T2DM) is characterized by an impaired insulin response that promotes a compensatory
increase in production and leads to hyperinsulinemia. In the STZ model we
demonstrated the glucose alone can affect tumour size, but others have predicted that
the hyperinsulinemia, rather than hyperglycemia, drives cancer development. 181,185 To
investigate the combined effects of insulin and glucose on EOC tumour growth, we used
a non-obese Akt2-/- (KO) mouse model of hyperglycemia (Figure 7C) and
hyperinsulinemia (Figure 8A). Sixty days after initiating ovarian tumours, KO mice had
larger tumours than Akt2+/+ (WT) mice, and mice heterozygous for Akt (+/-) had an
intermediate tumour weight (Figure 7A). A survival study also indicated more aggressive
cancer associated with T2DM as KO mice died significantly earlier than WT mice
(Figure 7B). Interestingly, growing tumours were associated with a drop in insulin
concentration in the blood (Figure 8C) and a lower glucose:insulin ratio which indicated
better insulin sensitivity.
Higher fasting blood glucose predicts larger tumours in Akt2 KO mice.
There was a significant positive correlation between the level of fasting blood glucose
(FBG) at 30 days post tumour induction (PTI) and tumour weight at 90 days PTI (Figure
7C). Blood glucose levels at the time of tumour collection had no correlation to tumour
weight (Figure 7D). Most surprisingly, we found that insulin levels were not associated
48
with tumour weight as many have predicted (Figure 8C). Furthermore, in contrast to the
normal relationship between insulin and glucose, we found no correlation of insulin with
fasting blood glucose at any experimental timepoint (Figure 8D).
49
Figure 7: High blood glucose in Type 2 diabetic mice predicts larger tumours
Ovarian tumours were induced in wildtype mice (WT) and mice homozygous (Akt2-/-) or
heterozygous (Akt2+/-) for Akt2. A) Mean tumour weight at 30, 60 and 90 days post
tumour induction (PTI). Bars are mean + SEM, n≥5. *p<0.05. B) Percent survival of
wildtype (solid line) and Akt2-/- (dashed line) tumour bearing mice. Time to death is
significantly shorter in Akt2-/- mice than in WT controls. p=0.0193, n=6. C) Mean fasting
blood glucose (FBG) measured at tumour induction surgery. Bars are mean + SEM,
n≥62. ***p<0.001. D) Relationship between FBG at 30 days PTI and tumour weight at
90 days PTI. Slope is significantly non-zero by linear regression (p=0.0093). E)
Relationship between FBG at 90 days and tumour weight at 90 days PTI. Linear
regression slope not significantly different from zero (p=0.067). Each point represents a
single animal. n=7 WT, n=6 Akt2-/-.
50
Figure 8: Tumours improve insulin sensitivity in Type 2 diabetic mice
A) Fasting blood insulin and B) glucose:insulin ratio at 30, 60 and 90 days post tumour
induction (PTI). Bars are mean + SEM, n≥6. Bars sharing the same letter are not
significantly different (p<0.05). C) Relationship between tumour weight and insulin
concentration at 90 days PTI and D) relationship between fasting blood glucose and
insulin concentration at 90 days PTI. Slope is not significantly different from zero. Each
point represents a single animal. n≥5.
51
Cells conditioned to high glucose media have altered glucose-dependent growth
kinetics.
In normal homeostatic conditions, fasting blood glucose in humans falls in the range of
3.9 - 5.5mM, while random (unfasted) samples are typically 4.4 - 6.6mM and under
7.8mM two hours after a meal.200 Many cancerous cell lines are cultured in media with
supra-physiological concentrations of glucose, including ID8 cells which are normally
grown in high glucose (25mM) DMEM. From these “diabetic” or “hyperglycemic” cells,
we derived a “normoglycemic” cell line, ID8-6, by reducing glucose concentrations in the
media over a period of three weeks and maintaining the cells in DMEM with 6mM
glucose. The chronic culture conditions of ID8 cells significantly altered their growth
kinetics. Although ID8-25 cells were more successful at closing a scratch wound than
ID8-6 cells (Figure 9) they also appeared to be more sensitive to glucose deprivation
(Figure 10) based on measurements of cell number.
We predicted that environmental glucose was a rate limiting factor in glucose
metabolism and that when more glucose was available, cells would respond by
increasing glucose uptake. We found that although this was true of ID8-25 cells, ID8-6
cells did not show the same response (Figure 11). Based on our hypothesis, we
predicted that cells would increase uptake in direct proportion to the total glucose
available. However, we found that all cells consumed proportionally less glucose in
hyperglycemic conditions (Figure 11A).
One of the hallmarks of cancer cells that drives their glucose consumption is the rate of
metabolism through the glycolytic and oxidative phosphorylation pathways. We
52
measured the metabolic viability of cells using the MTT assay and again found
differences between cells cultured in 6mM and those cultured in 25mM glucose. While
ID8-6 cell viability increased in a glucose-dependent fashion from 0mM to 25mM, ID825 cell viability plateaued at physiological glucose levels (Figure 12A). To evaluate
whether these results were specific to our murine cell line, we established normal- and
high- glucose variants of the human ovarian cancer cell line SKOV3 (SKOV3-6 and
SKOV3-25). As in ID8 cells, SKOV3-6 cells were more metabolically active than
SKOV3-25 cells in hyperglycemia (Figure 12B).
53
Figure 9: EOC cells conditioned to hyperglycemic media accelerate scratch wound
closing
A) Images were analyzed to determine the fraction of the scratches covered by cells
after 24 hours. B) Representative images of cells after 24 hours at 10x magnification.
Dashed lines indicate the width of the initial scratch. Bars are mean + SEM, n=3.
**p<0.01, ***p<0.001.
54
Cells relative
to normoglycemic control
2.5
ID8-6
ID8-25
2.0
1.5
1.0
b
*
*
a
a
a
0.5
0.0
2mM
6mM
16.7mM
25mM
Glucose in media
Figure 10: Cells conditioned to hyperglycemia are sensitive to glucose deprivation
Cells were cultured in different glucose concentrations for 24 hours. Crystal violet
absorbance at 590 nm was used as a cell number proxy. Bars are mean + SEM relative
to normoglycemic (6mM glucose) conditions. Different letters represent significant
differences between ID8-25 cells (black bars) by one-way ANOVA (p<0.05). ID8-6 cells
(white bars) are not significantly different across glucose concentrations. Brackets
indicate differences between cell lines at the same glucose concentration. ID8-6 n=4;
ID8-25 n=9 *p<0.05.
55
Figure 11: ID8-25 cells, but not ID8-6 cells, increase glucose consumption in
hyperglycemia.
A) Fraction of glucose consumed relative to the initial glucose concentration in the cell
media. B) Total glucose consumed in mg/dL. C) Relative glucose consumption per cell.
Bars are mean + SEM, n≥5. Different letters represent significant differences between
bars (p<0.05).
56
Figure 12: In high glucose media, cells conditioned to hyperglycemia have lower
metabolic viability than cells conditioned to physiological glucose
MTT absorbance was used to determine the metabolic viability of cells. A) Viability of
mouse epithelial ovarian cells acclimatized to physiological (ID8-6, white bars) or
hyperglycemic (ID8-25, black bars) conditions. ID8-6, n≥8; ID8-25, n=13. B) Viability of
human ovarian cancer cells acclimatized to physiological (SKOV3-6, white bars) or
hyperglycemic (SKOV3-25, black bars) conditions. SKOV3-6, n=4; SKOV3-25, n=11.
Values are mean + SEM. Stars indicate differences vs 6mM (normoglycemic control).
Brackets indicate differences between cell lines at a single glucose concentration.
*p<0.05, **p<0.01, ***p<0.001.
57
Discussion
Type 2 diabetes (T2DM) is a large and growing health burden worldwide. In population
studies it is consistently associated with increased incidence and mortality in multiple
cancers.199 T2DM is characterized by insulin insensitivity which leads to a combination
of hyperglycemia and hyperinsulinemia. In this study we showed that high systemic
glucose, with or without insulin, accelerates ovarian tumour progression.
Tumour growth and survival in Type 1 diabetes
Insulin is a powerful mitogen and pro-tumorigenic factor and is generally considered a
driving force for cancer development in diabetics.201,202 Type 1 diabetes is characterized
by hyperglycemia but a deficiency in insulin which makes it a good model for isolating
the effects of high blood glucose from hyperinsulinemia.
Type 1 diabetes (T1DM) is an autoimmune disorder that results in the destruction of
insulin-producing beta cells in the pancreas. We induced T1DM in vivo by using the
beta-cell toxin streptozotocin (STZ). Two weeks after STZ treatment, we injected
transformed epithelial ovarian (ID8) cells under the ovarian bursa in c57bl/6 mice in an
orthotopic, syngeneic model of EOC developed in our lab. In this model, tumour
development at 30 days post-tumour induction (PTI) models Stage I human disease and
EOC progression at 60 and 90 days PTI are representative of Stages II and III. We
found that as early as 30 days PTI, STZ-treated mice had significantly larger tumours
than normoglycemic controls. By 60 days, STZ-treated mice showed more aggressive
58
secondary disease. Moreover, we found that the degree of dysglycemia directly
influenced the size of the primary tumour (Figure 3C). This is in agreement with the
glucose concentration-dependent increase in proliferation that has been shown in
cancer cells in vitro203 and human studies that have shown that there are proportional
increases in the risk of overall cancer mortality with concurrent increases in plasma
glucose.204 There are currently no other studies that have used a T1DM model of
diabetes to evaluate cancer progression.
Based on these intriguing results, we anticipate that a short-term STZ model will be a
powerful tool for investigating tumour cell metabolism. However, such an acute and
short-term hyperglycemia is probably not representative of a human diabetic condition.
In reality, diabetes is a chronic disease that causes significant damage to normal cells,
particularly those of the endothelium. It is reasonable to expect that these changes will
impact the development of a tumour, which relies on interactions with its local
environment. In order to more accurately model the physiological effects of chronic
hyperglycemia, we also treated mice with STZ 15 weeks prior to tumour induction.
These mice had milder, but still significant, hyperglycemia (10.58 ± 0.59 mM) at tumour
induction surgery than the mice with “acute” diabetes (18.81 ± 1.93 mM). We suspect
that this difference reflects a partial recovery of the beta-cell population in the time
between STZ treatment and tumour induction.205 Chronically elevated blood glucose
interfered with the growth of ovarian tumours, which were significantly smaller than
those in normoglycemic mice (Figure 4A). Because cancer cells rely on the recruitment
of other cell types to populate a tumour and facilitate its growth, these unanticipated
results could be a reflection of irreversible cellular damage caused by chronic
59
hyperglycemia which prevented somatic cells from interacting with cancer cells. Cancer
associated fibroblasts (CAF) in particular make up a large portion of tumour mass and
are significant contributors to cancer progression.206 Diabetes, however, is associated
with low numbers of fibroblasts and fibroblasts with impaired motility.207 If local
fibroblasts have sustained cellular damage due to hyperglycemia, it is possible that
cancer cells will be unable to efficiently recruit them into tumours. Similarly, vascular
damage in diabetes may inhibit the tumour’s ability to recruit new vessels which would
limit tumour growth and metastasis.43,208
In addition to metabolic influences of blood glucose on tumour growth, the pathological
stress of diabetes itself contributes to long-term mortality in cancer patients,209 affecting
overall survival even if disease-free survival is not significantly worse.179 Adding to this
evidence, we found that despite the differences in tumour size, STZ treatments at 2 and
15 weeks prior to surgery both similarly decreased survival compared to nondiabetic
mice. Comorbid hyperglycemia can accelerate time to death in cancer patients, and
may do so independently of primary tumour size. The discrepancy between survival and
tumour size in the T1DM STZ mice may also reflect differences in secondary disease
that our scoring system does not take into account, in particular immune response and
the tumour-promoting effects of chronic systemic inflammation.210 The ascites in the 15week STZ animals was less hemorrhagic than that in the control animals (data not
shown), which might indicate differences in composition or malignant potential which
could affect tumour spread within the peritoneal cavity.
60
Tumour growth and survival in Type 2 diabetes
In women, the hyperinsulinemia and hyperglycemia of T2DM is often associated with
obesity. Models of diet-induced T2DM have shown increased primary tumour growth
and more metastatic disease in orthotopic ovarian211 and breast201 cancers. However,
obesity can independently contribute to tumour growth. To separate the effects of
obesity from the influence of insulin and glucose in T2DM, we used non-obese Akt2
knockout mice on a c57bl/6 background that have an impaired ability to translocate
glucose transport proteins to the cell surface in response to insulin. As in the short-term
STZ-treated animals, we found that these mice grew larger primary tumours in
proportion to higher fasting blood glucose and died earlier than their wildtype
counterparts (Figure 7A and B).
Systemic glucose
Tumours have been shown to act as glucose sinks, inducing host hypoglycemia in a
tumour mass-dependent fashion,212,213 particularly in the areas closest to the
tumour.212,214 In some cancer patients with epithelial-derived malignancies, this property
of tumours leads a condition called non-islet cell tumour-induced hypoglycemia
(NICTH).215 In addition to testing whether circulating glucose can affect tumour growth,
we were interested in the reciprocal effect of the tumour on systemic glucose. We
showed that the growth of ovarian tumours improved glycemic control in T2DM by both
normalizing the fasting insulin level and by improving insulin sensitivity (Figure 8).
Intraperitoneal glucose tolerance tests in STZ-treated mice demonstrated that growing
61
tumours also improved glucose tolerance in T1DM (Figure 6), suggesting that the
tumour-mediated reduction in hyperglycemia is not entirely insulin-mediated.
Surprisingly, we found that many of our mice had substantial abdominal fat (data not
shown), despite the fact that Akt2 KO results in an 80-90% reduction of fat depots.194
We interpret this as further evidence that tumours caused significant disturbances in
global metabolic maintenance.
In vitro
ID8 cells are normally maintained in high glucose DMEM media (25mM glucose, ID825). In order to compare EOC cell growth in normal and hyperglycemic conditions, we
created a parallel line of ID8 cells that were passaged in progressively lower
concentrations of glucose and eventually maintained at a physiological level (6mM, ID86). These cells allowed us to model the effects of both glucose addition and glucose
deprivation. ID8-6 and ID8-25 cells showed markedly different responses to
environmental glucose. Subsequently, another group showed that breast cancer cells in
glucose concentrations above 5mM have significantly higher rates of proliferation,
clonogenicity, motility, upregulation/activation of pro-oncogenic signaling, and
reductions in apoptosis.216
In order to measure metabolic viability, we used the MTT assay which is sensitive to
NADH produced primarily through glycolysis. Although MTT activity is generally thought
to reflect mitochondrial succinate activity, only a small proportion of tetrazolium
reduction occurs this way.196
62
Although ID8-25 cells consumed more glucose in response to substrate availability, the
MTT assay showed no increase glycolysis compared to control (Figure 12). This
suggests that glycolytic metabolism may be saturated at normal glucose concentrations.
However, because the glucose uptake increases, the saturation point must occur
downstream of hexokinase (HK) which is necessary to keep glucose from diffusing back
out of the cell. An increase in glucose uptake with no change in glycolysis could suggest
less efficient metabolism. This observation could be an indication of the Crabtree Effect,
in which high glucose saturates enzymes responsible for oxidative phosphorylation.121 It
may also be reflective of a defect in oxidative phosphorylation in the ID8-25 cells that
could necessitate an increase in glycolysis, representative of the Warburg Effect. Both
the Crabtree and Warburg effects are associated with increased malignancy.217
Conversely, ID8-6 cells did not increase glucose uptake in response to substrate
availability, but they did show a relative increase in glycolytic flux in 25mM glucose
compared to 6mM glucose. These cells may have limited capacity to upregulate glucose
transporters, or HK could potentially be saturated in these conditions and prevent
further glucose uptake. However, the increase in glycolytic activity indicates that
enzymes lower down the glycolytic cascade (phosphofructokinase, pyruvate kinase) are
not saturated at normal glucose concentrations and are therefore able to increase
energy production. Although we did not see higher glucose consumption over 24 hours,
it is possible that over a longer period this increased glycolytic activity would result in
increased glucose uptake.
As mouse epithelial ovarian cells become more glycolytic they also become malignant
and aggressive218 which would suggest that in ID8-6 cells, a high glucose environment
63
would be associated with more aggressive disease than a lower glucose environment.
Glucose starvation, which we modeled in ID8-25 cells in 6mM glucose, has been shown
to induce a caspase-independent cell death similar to autophagy.219 Consequently, we
would expect that ID8-25 cells in 6mM would be susceptible to death. We found that
there were fewer ID8-25 cells in total in conditions of glucose deprivation (Figure 10),
but environmental glucose did not influence their ability to close a scratch wound
(Figure 9). Neither of these assays shows what processes are responsible for changes
in cell number, so it is unclear whether these data indicate differences in cell
proliferation or cell death. Also, although treating ID8-25 cells in 6mM glucose
represents glucose deprivation, the cells are not in hypoglycemic media and so may not
be under enough metabolic stress to induce cell death.
Both ovarian cancer and T2DM are insidious diseases that are often diagnosed well
after pathological changes have occurred.25,220 In patients presenting with both
illnesses, the combined systemic metabolic defects in diabetes and the cellular
metabolic transformations in ovarian cancer cells have had a long window of interaction.
The in vitro results from this study show that cancer cells developed in hyperglycemic
environments are metabolically distinct from those in normoglycemic environments.
These differences could indicate that certain cancer treatments will have different
outcomes in diabetic compared to nondiabetic patients.
Hyperglycemia is a potentially modifiable risk factor for poor overall survival in women
with EOC and the results of this study support the possibility that anti-hyperglycemic
drugs would be a beneficial adjuvant cancer therapy.
64
CHAPTER 3: GLUCOSE TRANSPORT IN HYPERGLYCEMIC EOC
Introduction
Epithelial ovarian cancer (EOC) is the fifth leading cause of cancer death among
women.1,2 EOC is a challenge both to detect and to treat, resulting in a survival rate that
has been virtually unchanged in the last 30 years.
In large population studies, T2DM has been associated with the risk and prognosis of
many different cancers.199 Previous studies in our lab have shown that ovarian tumour
growth is more aggressive in diabetic environments independent of insulin levels
(Chapter 2). Hyperglycemia also increases the ability of EOC cells to take up glucose,
but the expression, regulation and uptake kinetics of glucose transporters in these
conditions are unknown.
Glucose primarily enters cells by facilitated diffusion through glucose transporters
(GLUTs). The activation of GLUT genes is one of the earliest events in oncogenesis 221
and high GLUT1 expression is often associated with poor prognosis and more
aggressive disease.222 Almost all invasive epithelial carcinomas are positive for GLUT1,
independent of stage, grade, or histological subtype.126,127 As a passive transporter,
GLUT activity is dependent on the direction and magnitude of the glucose concentration
gradient across the cell membrane. Glucose can also enter the cell against its
concentration gradient through a class of sodium-dependent secondary active
symporters (SGLTs). The most well studied isoforms, SGLT1 and SGLT2, are normally
expressed in the small intestine and kidney proximal tubule respectively. Few papers
65
have looked at SGLT expression in malignancies and only one published study has
demonstrated functional glucose uptake through the SGLTs in tumours.223
One of the hallmark characteristics of cancer is a glycolytic phenotype that necessitates
high glucose uptake to produce ATP and the biosynthetic precursors required for
proliferation. This elevated glucose uptake has been exploited in clinical imaging. Using
positron emission tomography (PET), the accumulation of the glucose analogue
18fluoro-deoxy-glucose
(FDG) can differentiate tumour tissue from normal tissue. In
EOC, FGD-PET can be useful in monitoring response to treatment but it has poor utility
as a diagnostic tool. FDG is efficiently transported through GLUT transporters, but it is a
poor substrate for SGLTs.224 FDG is the only oncolytic tracer currently in clinical use. If
sodium-glucose transport is functionally important in glucose uptake into ovarian tumour
tissue, the use of an SGLT-transportable glucose analogue could substantially increase
the power of imaging.
Glucose transport through SGLTs has the potential to release cancer cells from the
constraints of extracellular glucose concentration and thus provide a level of selfsufficiency for energy production. The goal of this study was to characterize the
expression and relative contributions of the GLUT and SGLT transporter families in the
glucose uptake and tumour growth of epithelial ovarian cancer, in both normoglycemic
and hyperglycemic environments.
66
Materials and Methods
Reverse transcription PCR
Reverse transcription (RT)-PCR was used to determine the mRNA expression of
glucose transporters SGLT1-3 and GLUT1-4 in ID8-25 cells. Total RNA was extracted
from cell lysates of serum-starved ID8-25 cells using the Qiagen RNA Extraction Kit
(RNeasy Mini kit, Qiagen, Mississauga, ON) and stored at -80°C prior to use.
RNA samples were reverse transcribed to cDNA using in 50µl reactions with 1µl
Oligo(dT) 12-18 primer, 1µl 10mM dNTP, 10µl of each RNA sample (65ºC for 5 minutes
in an MJ Research PTC-200 Thermo Cycler); four parts 5X First-Strand buffer, 2 parts
0.1M DTT (supplied with SuperScript II Reverse Transcriptase kit, Life Technologies
Inc., Burlington, ON) and one part RNasin in 7µl (42ºC for 2 minutes); 1µl of SuperScript
II Reverse Transcriptase. The reverse transcription reaction was then carried out at
42ºC for 50 minutes. Inactivation was induced by incubating the tubes at 70ºC for 15
minutes. Final cDNA products were stored at -20ºC.
Primer sequences were determined by Mus musculus gene sequences using NCBI
Primer-Blast. These primers, generated from Laboratory Services Molecular Biology
Lab (Guelph, ON), were stored at -20ºC. Sequences are provided in Appendix IV.
The cDNA was amplified in 50 µl reactions containing one part 10mM dNTP mix, five
parts 10x buffer and one part 1x Taq (New England BioLabs Ltd., Pickering, ON),
forward and reverse primers, 1µl of cDNA and nuclease-free water. The PCR reaction
95ºC for 30 seconds; appropriate annealing temperature (previously determined) for 30
67
seconds; 72ºC for 1 minute; for 35 cycles. PCR products were then incubated at 72ºC
for 10 minutes and stored at 4ºC until they were separated on a 1% agarose gel and
visualized with ethidium bromide using the FluorChem8800 gel documentation imaging
system.
Immunohistochemistry
In order to determine the protein expression of SGLT transporters in human EOC,
immunohistochemical detection of SGLT1 and SGLT2 was performed on a microarray
that included normal ovarian tissue, serous adenocarcinomas and metastatic tissue
cores (OV208 microarray; US Biomax, Rockville, MD). Sections were deparaffinized in
xylene and rehydrated in graded alcohol solutions. Endogenous peroxidase activity was
quenched using 1% (vol/vol) hydrogen peroxide and antigen retrieval was performed by
immersing slides in 10mM citrate buffer at 90°C for 12 minutes. Tissues were blocked in
5% (wt/vol) bovine serum albumin in PBS for 10 minutes and slides were then
incubated overnight with anti-SGLT1 or -SGLT2 primary antibody (Abcam, Cambridge
Science Park, Cambridge, MA) diluted in 0.01M PBS (pH 7.5) containing 2% (wt/vol)
BSA and 0.01% (wt/vol) sodium azide at 4°C in a humidity chamber. All subsequent
incubations were performed at room temperature. Anti-rabbit biotinylated secondary
antibody (Sigma, Oakville, ON) was diluted in the same buffer and incubated for 2
hours. Tissues were then washed in PBS and incubated with avidin and biotinylated
horseradish peroxidase (ExtrAvidin, Sigma, Oakville, ON) for 1 hour. Antibodies were
visualized using 3’ 3’-diaminobenzidine tetrahydrochloride (DAB) (Sigma, Oakville, ON),
tissue was counterstained with Carazzi’s Hematoxylin then dehydrated and mounted
68
with Permount (Fisher Scientific, Whitby, ON). Tissue cores were considered to have no
SGLT protein expression if total positive pixel counts were less than 2%.
Immunohistochemistry of SGLT2 (Abcam, Cambridge Science Park, Cambridge, MA)
and GLUT1 (Santa Cruz Biotechnologies, Santa Cruz, CA) in mouse tissue sections
was performed identically.
Cell Culture
Mouse-derived epithelial ovarian cancer cells (ID8; kindly donated by Drs. Paul
Terranova and Kathy Roby, Kansas State University, Kansas, USA), human ovarian
cancer cells (SKOV-3; ATCC, Manassas, VA, USA), and human normal ovarian surface
epithelium (NOSE; generously donated by Dr. Jinsong Liu, MD Anderson Cancer
Center, Houston, TX) were maintained in DMEM supplemented with 10% fetal bovine
serum (FBS), 1% antibiotic/antimycotic (Gibco BRL, Burlington, ON) and 2% Lglutamine (Life Technologies, Burlington ON). All experimental treatments were
performed in serum-free DMEM.
ID8 and SKOV3 cells are typically cultured in DMEM that contains supra-physiological
amounts of glucose (25mM). These “diabetic” cells were referred to as ID8-25 and
SKOV3-25. To test the effects of hyperglycemia, we created a parallel line of “normal”
cells chronically cultured in physiological glucose (6mM) that we named ID8-6 and
SKOV3-6. The -6 lines were created by passaging the -25 cells in reducing
concentrations of glucose over a period of three weeks. These lines were then
maintained in 6mM glucose. To account for any differences caused by higher passage
69
numbers during the glucose reduction, ID8-6 and ID8-25 cells were passage-matched in
all experiments.
Drugs
Phlorizin (PZ; Sigma, Oakville, ON) competitively inhibits glucose transport through
sodium/glucose cotransporters (SGLTs). Cytochalasin B (CB; Sigma, Oakville, ON) and
phloretin (PT; Santa Cruz Biotechnologies, Santa Cruz, CA) inhibit glucose transport
through the passive glucose transporters (GLUTs). Cytochalasin B is a more potent
GLUT inhibitor than phloretin, however it has two binding sites, one of which affects cell
cytoskeleton by interfering with actin assembly. Thus, CB was not appropriate for
motility studies and was replaced by PT.225
Viability assay
The WST-1 assay measures extracellular reduction of a tetrazolium salt to a coloured
formazan product dependent on the net rate of NADH reduction. NADH produced in the
mitochondrial TCA cycle is the primary reductant for extracellular WST-1 reduction via
transmembrane electron transport.226 ID8 cells were seeded at a density of 1500
cells/well in triplicate in 96 well plates and allowed to attach overnight. Cells were then
serum starved for 6 hours and treated for 48 hours in serum-free conditions. Following
treatment, cells were incubated with treatment media containing water-soluble
tetrazolium salt (WST)-1 reagent (Roche Applied Science, Laval, QC) at a 1:10 dilution
70
at 37ºC for one hour. Absorbance was quantified 450nm with 630nm correction on a
microplate reader.
Scratch wound assay
To examine the motility of cells in different concentrations of glucose, ID8-6 and ID8-25
cells were grown to confluency in 24 well plates and serum starved for 6 hours. A plastic
pipette tip was used to remove an area of cells down the centre of the dish. Wells were
gently washed with PBS to remove detached cells before treatment. Phase contrast
images were taken at three marked points along the scratch at t=0 and t=24 hours.
Wound sizes were determined using ImageScope software and are presented as the
fraction of the original scratch covered by cells.
Glucose consumption and cell number
Glucose concentration in the cell media was measured using the glucose oxidase
method and reagents from Pointe Scientific (Canton MI, USA). After 24 hours of
treatment, glucose concentration in the sample wells was subtracted from the glucose
concentration in blank wells to determine the amount of glucose consumed. Relative
cell number was approximated by the absorbance of the crystal violet nuclear stain. To
account for any differences in the rate of cell proliferation, glucose consumption was
divided by relative crystal violet absorbance (590nm) to determine a per-cell uptake
comparison.
71
Invasion Assay (Boyden chamber)
In vitro cell invasive/migratory capacity was measured using a Boyden chamber assay
(Biocoat Matrigel invasion chambers; BD Biosciences, Franklin Lakes, NJ). ID8-6 and
ID8-25 cells were grown on a Matrigel-coated upper membrane (10,000 cells per well)
and treated with 100µM phlorizin or DMSO vehicle control, in serum-free DMEM
containing 6mM or 25mM glucose. The lower chamber contained drug-free media with
10% FBS to act as a chemoattractant. Cells were incubated at 37°C and allowed to
migrate for 20 hours. Cells were removed from the upper chamber using cotton swabs.
500µl of 5% glutaraldehyde PBS was added to the lower chamber and incubated for 10
minutes at room temperature. Migrated cells were stained by adding 500µl 0.5%
toluidine blue staining solution in the lower chamber and incubating for 15 minutes.
Manual counting of cells in three fields of view per membrane was used to quantify
number of invaded cells.
shRNA SGLT2 knockdown
ID8-6 and ID8-25 cells were cultured in 12-well plates in complete medium (6mM
glucose or 25mM glucose DMEM containing FBS). At 40% confluence, cells were
treated with 5µg/ml Polybrene (Santa Cruz Biotechnologies, Santa Cruz, CA) and
infected with either: a pool of three shRNA constructs, each encoding SGLT2-specific
19-25 nt (plus hairpin loop) shRNA; or with control shRNA particles (scrambled
sequence) (Santa Cruz Biotechnologies, Santa Cruz, CA) overnight. Polybrene media
was then replaced with normal media and incubated overnight. After 48 hours, cells
72
were selected with puromycin dihydrochloride (Santa Cruz Biotechnologies, Santa Cruz,
CA) for 12 weeks. Clonal knockdown colonies were grown from single puromycinresistant cells. Western blot analysis was used to evaluate SGLT2 protein knockdown.
Animals
Wildtype c57bl/6 mice were purchased from Charles River Laboratories (Wilmington,
MA, USA), c57bl/6 Akt2 null mice (https://www.jax.org/strain/006966) were purchased
from Jackson Laboratories (Bar Harbour, ME, USA). An Akt2 null mouse colony was
established from breeder mice and animals were genotyped prior to beginning
experiments. Animals were housed at the Central Animal Facility at the University of
Guelph and all experiments were conducted according to Canadian Council on Animal
Care guidelines. Mice were kept on a 12h light:dark cycle and had free access to food
and water. For all experiments, adult mice were between 12 and 24 weeks old at time of
surgery.
Mouse model of epithelial ovarian cancer
Spontaneously transformed murine epithelial cells from C57Bl/6 mice (ID8) were diluted
in 5µl phosphate buffered saline (PBS) and injected under the ovarian bursa of WT or
Akt2 null mice. This orthotopic, syngeneic mouse model of EOC closely replicates
ovarian serous adenocarcinoma in women. By 90 days post-tumour induction, mice
develop large primary tumours, secondary peritoneal lesions, and significant abdominal
ascites.10 In survival experiments, mice were sacrificed when they became moribund
due to the accumulation of ascites fluid.
73
Type 2 Diabetes Mellitus (T2DM)
C57BL/6 mice homozygous for a mutant Akt2 allele are non-obese animals with poor
glucose tolerance and reduced insulin sensitivity that result in hyperglycemia and
hyperinsulinemia.193,194 Mice were genotyped using the following primers designed by
Jackson Laboratories: Akt2 wildtype reverse: TGCACAATCTGTCTTCATGCCAC; Akt2
common forward: ACCAACCCCCTTTCAGCACTTG; Akt2 mutant reverse:
TACACTTCATTCTCAGTATTGTTTTGC. Products are 110bp (wildtype), 277bp
(mutant), or 110bp and 277bp (heterozygote).
Intraperitoneal glucose tolerance tests
Mice were fasted for four hours prior to intraperitoneal glucose tolerance tests (IPGTT).
After recording baseline blood glucose, mice were given an intraperitoneal injection of
glucose (1 mg/g body weight in saline solution at a volume of 10µl/g body weight).
Blood glucose measurements were taken at 15, 30, 60, and 120 minutes after glucose
challenge using a handheld glucose monitor (Freestyle Lite, Abbott Laboratories, Abbott
Park, IL). Values that exceeded the measurement range of the glucose monitor were
considered 27.8mM (maximum reading).
Plasma for insulin quantification was collected in heparinized capillary tubes during the
IPGTT or from trunk blood from fasted animals at sacrifice using an ultra-sensitive
mouse insulin ELISA kit (Crystal Chem, Downers Grove, IL). Data are presented as the
average integrated area under the curve (iAUC) ± SEM where iAUC is the area under
the curve relative to fasting blood glucose (FBG) at time 0 (immediately before glucose
74
injection). iAUC was calculated using the trapezoidal method. The glucose:insulin ratio
was used as an indication of insulin sensitivity.195
Tissue collection and preparation
Mice were fasted for 6 hours prior to sacrifice by cervical dislocation. Following
euthanasia, trunk blood was collected and fasting blood glucose was measured
immediately using a handheld glucometer. The remaining blood was allowed to clot
then centrifuged to obtain serum samples and stored at -80oC. Ovarian tumours were
removed, weighed, and divided into sections for subsequent analysis. Sections were
either fixed in 10% neutral buffered formalin (Fisher Scientific, Whitby, ON) for wax
embedding and tissue sectioning or flash frozen in liquid nitrogen and stored at -80oC
for protein extraction. Secondary disease was evaluated at time of sacrifice by ascites
volume and number of secondary lesions. Mice were scored on a 4-point scale with 0
indicating absence of secondary disease. Mice with scores of 0 and 1 were considered
“low” and scores of 2 and 3 were considered “high”.
Radioisotope uptake
Alpha-methyl-glucopyranoside (AMG) and 2-Deoxy-D-glucose (2-DG) are glucose
analogues that are preferentially transported by SGLTs and GLUTs respectively.224
Transporter-specific glucose uptake was determined using [3H] 2-DG and [14C] AMG
(Moravek Biochemicals, Brea, CA). Mice were given a single tail vein injection of 25µCi
75
of [3H]2-DG and 5µCi of [14C]AMG 1.5 hours before sacrifice. Volumes of blood and
ascites and sections of tumour, skeletal muscle, liver and ovary were collected. The
samples were incubated at 55°C in ScintiGest solubilisation fluid (Fisher Scientific,
Whitby, ON) until the tissue dissolved. Thirty percent H2O2 was added to decolorize
samples. An aliquot of scintillation cocktail (CytoScint; MP Biomedicals, Santa Ana, CA)
was added to each sample and the total radioactive disintegrations per minute (dpm)
were determined using a Beckman LS 6500 liquid scintillation counter (Beckman
Coulter, Fullerton, CA).
Western Blots
Protein was collected from tumour tissue or SGLT2 knockdown cells using RIPA lysis
buffer and quantified using the DC Protein Quantification kit (Bio-Rad, Mississauga,
ON). Total protein lysates (20-40 µg) were used for immunoblot analysis. Lysates were
separated by SDS-PAGE on 10% gels. Proteins were electro-transferred onto
nitrocellulose membranes (Amersham, Piscataway, NJ) then blocked for 1 hour at room
temperature in 5% (wt/vol) skim milk in TBS-T (TBS with 1%(vol/vol) Tween 20).
Primary antibodies (SGLT1 (1:500, Abcam, Cambridge Science Park, Cambridge, MA),
SGLT2 (1:500, Abcam, Cambridge Science Park, Cambridge, MA), GLUT1 (1:200,
Santa Cruz) and β-actin (1:8000, Cell Signal Technologies, Danvers, MA) were diluted
in TBS-T skim milk and incubated overnight at 4°C on a rocking platform. After washing
with TBS-T blots were incubated with peroxidase-conjugated anti-rabbit secondary
antibody (1:1000 dilution; Cell Signal Technologies, Danvers, MA) or anti-mouse IgG
(1:1000 dilution, Cell Signal Technologies, Danvers, MA) for one hour at room
76
temperature on a rocking platform. Reactive protein was detected with ECL
chemiluminescence (Perkin Elmer, Waltham, MA). Membranes were imaged on the BioRad ChemiDoc XRS+ and densitometry was performed using Bio-Rad Image Lab 5.1.
Protein values were expressed as a ratio with β-actin.
Statistical Analysis
In vivo data were analyzed using Statistix 7 analytical software. Results are reported as
mean ± standard error of the mean (SEM). Significance threshold was set at 0.05 and p
values are listed in the figure legends. Tumour weight at survival was analyzed using a
one-way ANOVA followed by Tukey’s post-hoc. The significance of metastastic
secondary disease was measured using a Wilcox Rank Sum test for metastasis by
knockdown (ID8-6 and genotype were not significant in this model). Differences in
ascites volume by group were significant (chi squared approximation p=0.0025) but
there were no differences in pairwise comparisons. Differences in ascites volume were
therefore determined using a Kruskal-Wallis one-way nonparametric ANOVA by
knockdown. Overall survival data are plotted on a Kaplan-Meier survival curve and
differences were determined by a proportional hazards regression. All other data were
analyzed using GraphPad Prism statistical analysis software. Comparisons between
means were measured by two-way ANOVA followed by a Bonferroni correction, or by
unpaired two-tailed Student’s t-test.
77
Results
Both passive and active glucose transporters are expressed in EOC
Using reverse transcription PCR we showed the expression of both passive (GLUT1,
GLUT3) and active (SGLT2) glucose transporters in ID8 murine epithelial ovarian
cancer (EOC) cells (Figure 13). This is the first demonstration that sodium-glucose
transporters are expressed in ovarian cancer.
To further characterize the expression of SGLT transporters, we immunostained human
tissue microarrays containing tissue samples from serous ovarian adenocarcinomas,
metastases, and normal ovarian tissue (Figure 14). Using Imagescope software to
quantify protein expression, we found that both SGLT1 and SGLT2 were present in
normal ovaries at low levels (3.1 ± 0.2 % and 23.8 ± 2.8 % respectively). In tumours,
SGLT1 was only expressed in about a third of samples, in which staining density was
only about 16%. SGLT2, however, was expressed in all samples with high density (67.5
± 2.9 % positive pixels). Both isoforms were overexpressed in tumours, though
expression in metastases was no different than in normal tissue. Expression did not
differ by stage, but low grade tumours were associated with higher SGLT expression.
Staining was diffuse throughout the cytoplasm and there was no clearly defined
expression at the cell membrane. Positive staining for SGLT1 is shown in the
supplementary figures (Figure 41, Appendix I).
78
Figure 13: GLUT and SGLT transporters are expressed in ID8 cells
Reverse transcription PCR for glucose transporters in mouse epithelial ovarian cancer
cells (ID8-25). A) Secondary active sodium glucose transporter 2 (SGLT-2) expressed.
Positive controls: mouse kidney. B) Facilitative glucose transporters (GLUT) GLUT-1
and GLUT-3 are expressed. Positive controls: GLUT-2, mouse liver and GLUT-4,
mouse skeletal muscle. NTC, no-template control.
79
Figure 14: SGLT2 is overexpressed in human serous ovarian cancer
Immunostaining for SGLT1 and SGLT2 on tissue microarrays of human serous
epithelial ovarian cancers. Staining was quantified using Imagescope software to
determine the number of positive and negative pixels within a tissue section. A)
Representative tissue cores of normal and malignant (FIGO Stage 3c, Grade 3) ovarian
80
tissue. Pixels intensity was categorized as negative, weak positive, positive, or strong
positive. Normal tissue (n=26), carcinomas (n=48), metastases (n=18). B) Staining by
tissue type C) Staining of serous adenocarcinomas by FIGO stage D) Staining of serous
adenocarcinomas by histological grade *p<0.05, **p<0.01, ***p<0.001 between total
positive pixel counts.
Table 3: SGLT protein expression in human ovarian cancer
SGLT1
Tissue expression
SGLT2
Positive pixel density
Mean ± SEM (%)
N
Percent
Normal
18
100.0
3.1
±
Tumours
15
31.3
15.8
Grade 2
9
56.3
Grade 3
6
6
Tissue expression
Positive pixel density
Mean ± SEM (%)
N
Percent
0.2
16
100.0
23.8
±
3.8
±
4.0
48
100.0
67.5
±
2.9
20.9
±
5.9
16
100.0
78.4
±
3.6
18.8
8.3
±
2.4
32
100.0
62.1
±
3.7
15
100.0
68.4
±
3.9
28.6
19.4
±
8.4
6
100.0
48.2
±
11.9
Stage 1
Stage 2
Stage 3
8
33.3
13.1
±
4.3
20
100.0
69.0
±
4.5
Metastases
7
38.9
3.7
±
0.6
18
100.0
37.2
±
5.3
81
GLUT transporters are active in physiological glucose concentrations (6mM) and
SGLT transporters are active in hyperglycemic conditions (25mM)
In order to evaluate the relative contributions of GLUT and SGLT transporters, we used
class-specific glucose transport inhibitors. Phlorizin (PZ) selectively inhibits glucose
transport through the SGLTs, while phloretin (PT) and cytochalasin B (CB) inhibitor only
GLUT-mediated glucose transport. Based on the mRNA expression in ID8 cells, we can
infer that in ID8 cells, PZ is acting on SGLT2 and PT/CB are acting on GLUT1 and
GLUT3.
ID8 cells were differentiated into two subtypes through serial passaging. ID8-6 cells are
conditioned to normal physiological concentrations of glucose (6mM) and ID8-25 cells
are conditioned to supra-physiological glucose (25mM). In both cell lines, PZ inhibited
glucose consumption in high-glucose media compared to cells treated with a vehicle
control (Figures 15 and 16). PZ had no significant inhibitory effect on glucose uptake in
normal glucose concentrations. PT, on the other hand, significantly reduced glucose
uptake in 6mM but not 25mM glucose.
In ID8-25 cells the combination treatment with PZ and CB did not have additive
inhibitory effects on glucose uptake in either normal or high glucose (Figure 16).
Regardless of transporter inhibition, these cells consumed more glucose in high-glucose
media than in physiological glucose, even though glucose was plentiful. Even in 6mM
glucose media, the total consumption represented less than a third of the total glucose
available.
82
Figure 15: SGLTs affect glucose transport in hyperglycemia in cells conditioned to
normal physiological glucose concentrations (ID8-6)
Glucose consumption by ID8-6 cells in normoglycemic and hyperglycemic culture
conditions. Cells were treated with 50µM PT or 200µM PZ in 0.01% DMSO. A) Total
glucose consumed in mg/dL. B) Fraction of glucose consumed relative to the initial
glucose concentration. C) Relative glucose consumption per cell. Bars are mean +
SEM, n≥3. Bars sharing the same letter are not significantly different (p<0.05).
83
Figure 16: SGLTs affect glucose transport in hyperglycemia in cells conditioned to high
glucose concentrations (ID8-25)
A) Total glucose consumed in mg/dL. B) Fraction of glucose consumed relative to the
initial glucose concentration. C) Relative glucose consumption per cell. Drugs were
dissolved in 0.01% DMSO. Combination treatment was 50µM PZ and 0.5µM CB. Bars
are mean + SEM, n≥3. Different letters indicate significant differences between bars
(p<0.05)
84
Glucose concentration does not alter the effect of glucose inhibitors on viability.
The metabolic viability of cells was measured using the WST-1 assay (Figure 17).
Phlorizin (PZ, SGLT inhibitor) had no effect on cell viability, while phloretin (PT, GLUT
inhibitor) potently decreased viability in a concentration-dependent manner in both ID86 and ID8-25 cells. The combination of PZ and CB on ID8-25 cells had no effect on cell
viability.
85
Figure 17: SGLT inhibition does not affect cell metabolic viability
A) and B) Viability of mouse epithelial ovarian cells acclimatized to physiological
conditions. ID8-6, n≥3; C)-E) Viability of mouse epithelial ovarian cells acclimatized to
hyperglycemic conditions. Combination treatment (PZ+CB) is 50uM PZ and 0.5uM CB.
ID8-25, n=3. *p<0.05, **p<0.01, ***p<0.001. Bars are mean + SEM. Different letters, or
brackets with stars, indicate significant differences between bars (p<0.05)
86
Inhibition of the SGLTs does not affect wound healing in hyperglycemia
In “non-diabetic” ID8-6 cells, the GLUT inhibitor phloretin (PT) inhibited wound healing a
scratch assay in both normal- and high-glucose media (Figure 18). The SGLT inhibitor
phlorizin (PZ), however, only inhibited wound healing in 6mM glucose. The combination
of phlorizin and phloretin did not have an additive effect on the fraction of the scratch
wound that was filled in over 24 hours.
87
Figure 18: Phlorizin and phloretin inhibit scratch wound healing in ID8-6 cells
Images were analyzed to determine the fraction of the scratches covered by cells after
24 hours. Representative images of cells after 24 hours at 4x magnification. Dashed
lines indicate the width of the initial scratch. A) ID8-6 cells in 6mM glucose. B) ID8-6
cells in 25mM glucose. Combo is 100µM PZ + 100µM PT. Bars are mean + SEM, n=3.
**p<0.01, ***p<0.001.
88
Phlorizin increases the invasive capacity of ID8-6 but not ID8-25 cells
Cells conditioned to high glucose (ID8-25) were more invasive than ID8-6 cells
regardless of available glucose (Figure 19A, a vs b). However, SGLT channel inhibition
with PZ equalized the number of migrated cells across all experimental groups by
increasing ID8-6 cell invasiveness and decreasing (6mM) or maintaining (25mM) ID8-25
invasiveness (Figure 19A, black bars). Relative to cells treated with DMSO vehicle
control, ID8-6 cells treated with PZ were significantly more invasive than ID8-25 cells
(Figure 19B).
89
Figure 19: SGLT glucose transporters limit cell invasion in ID8-6 cells
Cells were cultured in the upper chamber of trans-wells coated with Matrigel and
allowed to migrate towards the lower chamber using FBS as a chemoattractant. Purple
staining indicates cells that have migrated through the Matrigel membrane. Bars are
mean + SEM. Different letters indicate significant differences within DMSO control
treatments (white bars) p<0.05. Phlorizin treatments (black bars) were not significantly
90
different from one another. (p<0.05). Stars indicate significant differences between
DMSO and PZ treatments within each cell line/glucose condition. *p<0.05, ***p<0.001.
A) Number of cells migrated through Matrigel, with representative images. B) Migration
of PZ-treated cells relative to DMSO control.
Knockdown of SGLT2 opposes contact-dependent growth inhibition
The SGLT2, the sodium-glucose transporter isoform present in our cells, was stably
knocked down using shRNA (Figure 20A). Cells with lowered SGLT2 expression had an
altered pattern of growth in culture where individual cells appeared to migrate away
from a group of cells, rather than multiplying to expand an existing group of cells (Figure
20B). KD cells were also resistant to contact-dependent growth inhibition and continued
to grow on top of a fully confluent monolayer of cells, forming three-dimensional foci
with cords of stacked cells projecting outward (Figure 20C).
91
Figure 20: Knockdown of SGLT2 in ID8 cells changes proliferative behaviour in culture
SGLT2 expression in ID8-6 and ID8-25 cells was knocked down using shRNA
packaged in lentiviral particles. A) Representative Western blots showing protein
expression of SGLT2 in cells treated with empty vectors (left) and shRNA constructs
(KD, right). Protein knockdown was evaluated by densitometry. ID8-6 cells had an
apparently complete SGLT2 knockdown by this technique and ID8-25 cells protein
expression was reduced by 65%. B) Representative images of cell migration in cultures
of KD clones at 4x magnification. C) Representative images of cell foci formed at high
confluency by KD clones.
92
GLUT1 protein expression appears greater in diabetic mice than in WT mice
In both tumours with and without SGLT2 expression (ID8-25), GLUT1 expression
appears qualitatively greater in Akt2 KO mice than in normoglycemic WT mice (Figure
21A and B). In mice whose ovaries were injected with SGLT2 KD cells, SGLT2
expression did not differ by genotype, but in WT mice appeared considerably higher
than GLUT1 expression (Figure 21C).
93
Figure 21: GLUT1 protein expression is higher the tumours of diabetic mice than in
those of WT mice
Representative images of tumour tissue stained by immunohistochemistry. A) GLUT1
expression in from Akt2 KO mice appears higher than in WT mice. B) In mice whose
tumours were initiated with ID8-25 SGLT2 knockdown cells, GLUT1 expression appears
higher in Akt2 KO than in WT mice. C) SGLT2 expression in mice with SGLT2
knockdown cells does not appear to differ between genotypes. (Not quantified).
94
Glucose is transported into orthotopic ovarian tumours by both GLUT and SGLT.
In order to determine the functional expression of GLUT and SGLT transporters in
murine EOC tumours, we injected radioisotope-tagged glucose analogues into the tail
veins of wildtype (WT) and Akt2 -/- (KO) mice four months PTI. Alpha-methylglucopyranoside (14C-AMG) is specifically transported through the SGLT transporter
family and deoxyglucose (3H-DG) is specifically transported through the GLUT
transporter family. Both GLUT and SGLT transporter families contribute to glucose
uptake in tumours in approximately equal amounts (Figure 22). All tissues took up the
same total glucose per gram (data not shown). Total glucose uptake per gram was
unchanged in SGLT2 KD cells and a significant increase in the amount of GLUTmediated transport completely compensated for the loss of SGLT2 expression (Figure
23).
95
Figure 22: Both GLUT and SGLT transporters are responsible for functional glucose
uptake into tumours
Bars show the relative uptake of each isotope. Stars indicate significant differences in
the fractions of 3H-DG (transported by GLUTs) and 14C-AMG (transported by SGLTs).
Different letters indicate differences in uptake between tissues. (p<0.05). N=4.
96
Figure 23: SGLT2 KD impairs glucose uptake through active transport, but total glucose
uptake is maintained through compensation by GLUTs
A) A 3H-DG : 14C-AMG ratio above 1.0 indicates that there was more GLUT-mediated
glucose transport than SGLT-mediated transport. There were no significant differences
between control WT and Akt2 KO ratios in SGLT2 control mice. B) Total glucose uptake
per gram of tissue (sum of disintegrations per minute of 3H-DG + 14C-AMG divided by
tissue weight). Control n=4; KD n=1. Bars represent mean + SEM (control) or values
from the single animal with an SGLT2 KD tumour.
97
45 days after tumour induction, GLUT1 protein expression is upregulated in
diabetic mice with SGLT2 KD tumours.
At 45 days PTI we found no significant differences in the weights of tumours in any
experimental condition (Figure 24A). Western blot analysis showed that in Akt2 KO
mice, GLUT1 expression was upregulated beyond accurate quantification. Tumours
from mice that were injected with SGLT2 KD ID8 cells had the same SGLT2 protein
expression as those injected with control ID8 cells (Figure 24B and C).
98
Figure 24: Expression of GLUT1 is highly upregulated in tumours with reduced SGLT2
expression at 45 days PTI
Transformed mouse ovarian epithelial cells (ID8-25) were injected under the ovarian
bursa of wildtype (WT) and Akt2-/- (KO) mice. Injected cells had normal (control) or
knocked down (SGLT2 KD) expression of SGLT2. A) Mean tumour weight at 45 days
99
post tumour induction (PTI). Bars are mean + SEM n=4. Differences in tumour weights
were non-significant. p=0.065 between control and SGLT2 KD in WT mice. B) Protein
expression of glucose transporters in 45-day tumours. Each lane represents protein
from an individual animal. C) Densitometry of Western blots relative to β-actin loading
control. Bars are mean + SEM n≥3. ***p<0.001.
SGLT2 KD increases tumour weight and decreases overall survival
SGLT knockdown significantly increased tumour weight at the time of death in a cell
line-dependent manner (Figure 25). Tumours initiated with ID8-6 cells were associated
with decreased tumour weight after controlling for the variation due to the SGLT2 KD. In
a univariate analysis, however, cell type (ID8-6 / ID8-25) had no effect. Knockdown of
SGLT2 in ID8-25 cells increased tumour weight in WT mice, while SGLT2 KD in ID8-6
cells caused significantly larger tumours in the diabetic Akt2 KO mice.
Mice whose tumours had knocked-down SGLT2 expression had a relative risk of dying
of nearly 25 times. When considered in a multivariate model, after controlling for
variation in ID8-6 and WT the relative risk of dying was more than 57 times higher in
mice with SGLT2-deficient tumours (Figure 26). ID8-6 cells or a WT genotype also
increased the relative risk of death by 5 and 3.5 times respectively, although their
effects were non-significant if SGLT2 KD mice were removed from the model.
100
Figure 25: SGLT2 knockdown leads to significantly larger primary tumours at death
Tumours were removed and weighed when mice became moribund. Representative
images of A) wildtype mice and B) Akt2-/- mice and tumours. (ID8-25 cells, SGLT2
control and SGLT2 KD.) Bars represent mean + SEM. Different letters represent
significant differences between all groups (p<0.05). n=5 per group. SGLT2 KD
(p<0.0001) increases tumour weight. ID8-6 decreases (p=0.0364) tumour weight after
controlling for the variation due to KD, but without KD its effects are not significant.
101
Control, WT
SGLT2 KD, WT
Percent survival
100
Control, KO
SGLT2 KD, KO
50
0
0
50
100
150
Days PTI
Figure 26: SGLT2 KD in tumours significantly decreases overall survival
Solid lines indicate tumours developed from ID8-25 cells, dashed lines indicate those
from ID8-6 cells. The experiment was terminated at 126 days and mice that were
sacrificed at 126 days were considered to have lived. N=5 per group. Survival of ID8-25
control mice overlaps survival of ID8-6 control mice.
Table 4: Cox proportional hazard regression model for ovarian cancer survival
Variable
RR
p
95% CI
lower
95% CI
upper
SGLT2 knockdown
24.48
0.0019
-
-
Univariate
SGLT2 knockdown
57.26
0.0002
79.628
103.3
Multivariate
ID8-6 (vs ID8-25)
5.29
0.0053
66.946
95.498
Multivariate
Akt2 WT (vs KO)
3.45
0.0319
74.61
103.79
Multivariate
102
Table 5: Median survival times of mice with SGLT2 KD tumours
Experimental Group
Median survival (days PTI)
WT
KD
ID8-6
69
Akt2 KO
KD
ID8-6
86
WT
KD
ID8-25
111
WT
Control
ID8-25
112
Akt2 KO
KD
ID8-25
115.5
Akt2 KO
Control
ID8-6
126
Akt2 KO
Control
ID8-25
126
WT
Control
ID8-6
126
Tumours deficient in SGLT2 have more severe secondary disease
The number and dispersal of secondary lesions plus the volume of ascites in tumourbearing mice constitute our assessment of secondary disease. Knockdown mice had
higher rates of metastasis (p>0.00001), but differences were not significant by ID8-6 or
genotype (Figure 27). The severity of secondary disease is associated with overall
survival. Widespread metastatic lesions are more consistently associated with morbidity
than tumour size (Figure 28).
103
Figure 27: Secondary disease is much more severe in mice with SGLT2 KD tumours
A) Representative images of advanced peritoneal disease. Secondary lesions (B, C)
were scored on a 4-point scale with 0 indicating absence of secondary disease and 4
indicating more than 10 lesions spread throughout the peritoneal cavity. Tumours with
SGLT2 knockdown have higher rates of metastasis (p>0.00001). Neither cell type (ID86/ID8-25) nor genotype (WT/Akt2-/-) significantly affected number of secondary lesions.
104
4
400
300
3
200
2
100
1
Secondary score
Tumour weight (mg)
Primary tumours
ID8-6 KO
ID8-25 KO
ID8-6 WT
ID8-25 WT
Secondary lesions
0
0
60
75
90
105
120
135
Days PTI
Figure 28: Poor overall survival is associated with severe secondary disease,
regardless of primary tumour weight, in mice bearing SGLT2 KD tumours
Data represent the cohort of mice (WT or Akt2 KO) injected with ID8-6 or ID8-25 SGLT2
KD cells. Coloured circles represent primary tumour weights at death. The secondary
lesion scores for animals that died on a particular day are marked by blue squares.
Shorter survival times were generally associated with small primary tumours but high
secondary scores which indicate multiple metastatic lesions spread throughout the
peritoneal cavity. In mice that survived longer, there appears to be more variability in
both tumour weight and secondary scores.
105
Discussion
Glucose uptake in epithelial ovarian tumours occurs through the coordinated
actions of active (SGLT) and passive (GLUT) transporters
Glucose can enter cells through two classes of transporters. The GLUT proteins
facilitate transmembrane diffusion and are upregulated in most cancers.227 The sodiumglucose transporters (SGLT) can move glucose against its concentration gradient by
using the inward sodium concentration gradient maintained by the Na+/K+ pump. Here
we demonstrated for the first time that members of both transporter families are
expressed in epithelial ovarian cancer (EOC). ID8 murine EOC cells express the mRNA
of SGLT2, GLUT1 and GLUT3, all of which are high capacity glucose transporters.
GLUT1 and GLUT3 have previously been reported in EOC, however in our subsequent
experiments we decided to focus on the contributions of the GLUT1 isoform because of
its stronger association with tumour development.126
Very few data exist on the expression of the SGLTs in human cancers. Where it has
been examined, SGLT isoforms vary among tumour types: preliminary studies have
shown that SGLT1 is expressed in breast145,228, prostate145,228, head and neck229,
colon230, laryngeal231, pancreatic232, oral233, and ovarian cancers234 while SGLT2
expression has been noted in lung,235 pancreatic223, and prostate223 cancers. We used a
tissue microarray of serous ovarian adenocarcinomas from women to examine the
protein expression of SGLT1 and SGLT2. SGLT1 protein was expressed at low levels in
about 30% of serous adenocarcinoma samples, while SGLT2 was strongly expressed in
all samples (Figure 14). Neither isoform was correlated with tumour stage, but lower
106
expression was associated with more poorly differentiated tumours. Subsequent to
these experiments, another group investigated SGLT expression in ovarian cancer.
Contrary to our findings, this large study found that SGLT1 was overexpressed and was
an independent biomarker for poor prognosis. The reason for the differences in SGLT1
expression in the two sets of samples is puzzling considering that the histochemical
methodology and antibody used were the same.234 It is unclear whether or not they
looked for SGLT2 expression.
In order to determine the relative contributions of each class of transporters, we used
class-specific inhibitors of GLUTs and SGLTs. Inhibition of the GLUT transporters with
cytochalasin B or phloretin reduced glucose uptake in cells cultured in a physiological
glucose concentration (6mM), but was ineffective in hyperglycemic conditions (25mM
glucose). Conversely, inhibition of the SGLT transporters with phlorizin had greater
effects in high glucose than in normal glucose. Rather surprisingly, the combination of
SGLT and GLUT inhibitors did not have an additive effect on glucose uptake. We
hypothesize that this is a result of their reciprocal effects in high and low glucose:
because the GLUT transporters had low activity in high glucose, competitively inhibiting
them in these conditions would have had little additional effect on overall glucose
consumption. The same would be true of inhibiting SGLT transporters physiological
concentrations of glucose.
In vivo, GLUT-mediated uptake compensated for a lack of SGLT2. Mice whose tumours
had reduced expression of SGLT2 took up relatively less of the SGLT-specific glucose
analogue 14C-AMG than those with SGLT2. However, compensation from the GLUT
transporters maintained similar total glucose uptake (Figure 23). Expression of the two
107
families of transporters appears to be inversely associated. In colon cancer cells SGLT1
is downregulated following oncogenic transformation while GLUT1 is upregulated.230
Similarly, in renal epithelial cells hypoxia increases GLUT1 expression but concurrently
decreases SGLT1 and SGLT2.236 The influence of oxygen levels on SGLT expression
suggests that they may be regulated by HIF-1α expression, whose pro-tumourigenic
effects could explain the increased SGLT expression after oncogenic transformation.
In tumours from diabetic mice at 45 days PTI, we saw significant upregulation of GLUT1
protein in the SGLT2 KD group which matched the pattern of radioactive glucose uptake
observed at 120 days (Figure 24). However, despite confirmation of SGLT2 knockdown
in the tumour-initiating cells, we did not see a decrease in SGLT2 protein expression in
tumours at 45 days by Western blot. In WT animals, tumours initiated with SGLT2
deficient cells did not show changes in the expression of either SGLT2 or GLUT1,
although again there was an increase functional activity of GLUT transporters at 120
days PTI. In addition, although GLUT1 appeared to be significantly upregulated in
SGLT2 KD cells by Western blot at 45 days PTI, immunohistochemical analysis showed
the opposite pattern in tumour samples taken at morbidity (Figure 21). This may be due
to the relatively small size of these tumours, as well as the fact that they were taken at
animal death. We were unable to collect enough sample to stain for SGLT2 expression
in tumours initiated with SGLT2-competent cells. These results indicate that transporter
expression is not directly associated with uptake kinetics, and suggest that SGLT2
expression in non-cancerous cells in the tumour may obscure the knockdown that was
clearly seen in isolated cells in vitro.
108
The active sodium-glucose symporter SGLT2 is protective against cancer death,
primary tumour growth, and metastatic secondary disease in an orthotopic
mouse model of EOC
The uptake and phosphorylation of glucose are the most important steps in determining
glycolytic flux.94–96 Hypothesising that inhibited SGLT transport would slow tumour
growth, we stably knocked down the expression of the SGLT2 isoform in ID8 cells using
shRNA. Although SGLT1 has the capacity to take up glucose when SGLT2 activity is
compromised,237 Western blot analysis suggested that this isoform had limited
expression in the ID8-cell tumours. Although there was efficient uptake of the lentiviral
particles containing the shRNA against SGLT2, achieving knockdown was very difficult
particularly in ID8-25 (cells conditioned to hyperglycemia). If SGLT2 is more important in
cells in high glucose (ID8-25), then KD cells would be less likely to survive even if they
are puromycin resistant. Because we selected single cells that appeared most active in
the presence of puromycin, it is possible that we selected for cells that had lesser
knockdown due to the random integration of the lentiviral construct into the genome of
the cell.
The results of this study strongly opposed our hypothesis and rather than improving the
outcome of EOC in a mouse model, SGLT2 KD actually increased the relative risk of
dying by 57 times. We can only speculate on the mechanism behind these surprising
results. Interestingly, SGLTs appear to be protective in many cases: increasing overall
survival231 or disease free survival232, and higher expression in better differentiated
tissues.229,230 We expected that the ability for SGLT2 to take up glucose independently
of its extracellular concentration would lead to more aggressive disease. However, it
109
could instead have relieved some of the selective pressure for mutations or non-genetic
actions that would facilitate metastasis. If SGLT2 KD provides a strong impetus for cells
to move away from the primary tumour to gain access to resources in a more abundant
and less competitive environment, we would expect to see more cell motility and a
greater spread of secondary disease. Results from the mouse model (Figure 27), cell
invasion assay (Figure 19) and the movement of individual cells in culture (Figure 20B)
all seem to support this premise. Secondary disease is a better predictor of death than
primary tumour weight. The impact of SGLT KD in EOC may also be related to its role
as an ion transporter. Transport through SGLTs can be driven by protons as well as
sodium.130 The dysregulated metabolism of cancer cells results in high extracellular H+
concentrations that may favour SGLT-mediated glucose uptake.
There is an extensive body of evidence indicating that GLUT1 overexpression is
associated with aggressive disease. Overexpression of GLUT1 in EOC predicts
response to chemotherapy and shorter disease-free survival238 and expression
progressively increases with stage and grade suggesting it may be related to malignant
transformation.239,240 These data indicate that SGLT2 knockdown may be promoting
EOC growth indirectly by increasing GLUT1, which would suggest that SGLT somehow
limits GLUT upregulation.
In our experiments GLUT1 compensation did not increase overall glucose uptake, but
simply maintained the levels seen in tumours with SGLT2-competent cells. However,
we saw profound effects on EOC outcome, so the contribution of SGLT2 is not
necessarily a glucose-related phenomenon. A similar situation has been noted in
neurons, where SGLT1 expression appears to be important even in cells that have an
110
abundance of GLUT3 and more than sufficient glucose uptake capacity.130 The glucosesensing or water- and ion- transporting roles of SGLT could be the significant effectors
preventing oncogenesis.
In the mouse model, the level of systemic glucose played a limited role in the impact of
SGLT2 KD on cancer progression. Neither genotype nor cell type had significant effects
on their own, which highlights the dominant effect of SGLT2 KD, but they did influence
survival in multivariate analyses. After controlling for SGLT2 KD and cell type,
normoglycemic (wildtype) mice were more 3.5 times more likely to die than
hyperglycemic and hyperinsulinemic (Akt2 KO) mice. This could be an indication that
diabetic conditions support tumour growth, particularly when glucose transport is
dysregulated. Mice whose tumours were induced by injection of ID8-6 cells were more
than five times more likely to die than those with ID8-25 tumours after controlling for
SGLT2 KD and genotype. Part of the effect of ID8-6 over ID8-25 could be attributed to
the difference in amount of SGLT knockdown (Figure 20A).
The studies presented here show compelling data that SGLT2 acts as a tumour
suppressor. Knockdown of SGLT2 in tumour cells increased tumour size, severity of
secondary disease, and substantially reduced survival times. In vitro observations
complemented this finding and collectively indicated a protective role for SGLT2 in cell
proliferation and migration, particularly in hyperglycemic conditions. Taken together,
SGLT2 appears to be most powerful on cell motility/invasion. There are several
examples of things that have limited effect on primary tumour growth but that promote
migration and increase the formation of metastases, including L-lactate.105
111
The overexpression of SGLT2 in the tissue microarray and the ability of its inhibition to
promote tumour growth seem at odds with one another. However, protein expression
does not necessarily reflect membrane expression or the kinetics of glucose uptake. As
yet there have been no functional studies of SGLT-mediated glucose uptake in human
ovarian tumours. Moreover, the ability of GLUTs to compensate for reductions in SGLTmediated glucose uptake make these results hard to interpret.
Based on these findings, we would expect the use of SGLT2 inhibitors to be associated
with an increase cancer or cancer-related death. Safety studies of SGLT2-inhibiting
diabetes drugs have not shown prohibitively high incidences of cancer, although it is
acknowledged that these data are not conclusive.241
FDG may underrepresent the total amount of glucose uptake into cancer cells
The high accumulation of the glucose analogue FDG is used to differentiate highly
glycolytic tumour tissue from normal tissue. Currently, FDG is the only FDA approved
oncolytic tracer and more than 90% of oncological PET imaging uses FDG. 161 However,
FDG is a poor substrate for SGLTs. Because FDG is not transported through the SGLT
transporters, this suggests that tumours fulfilling metabolic needs through SGLTs will
have weaker PET results that may contribute to a false-negative diagnosis. Moreover,
the results of this study suggest that low or falling SGLT transport may indicate more
aggressive disease.
112
GLUTs and SGLTs appear to coordinate glucose transport into cancer cells and have
differential importance in normal and high glucose environments. GLUT was able to
compensate for the decreased 14C-AMG uptake in SGLT KD tumours by raising its
uptake of 3H-DG. It would be interesting to see if SGLT transport could compensate in
the same way if the GLUTs were compromised.
Conclusions
GLUT1 expression is historically considered the most important glucose transporter in
cancer cells and is correlated with many markers of advanced disease. For the first
time, we show that a second class of transporter, SGLT, is not only expressed, but also
makes significant functional contributions to EOC progression.
113
CHAPTER 4: ANTIDIABETIC EFFECTS OF METFORMIN AS EOC
THERAPY
Introduction
Metformin is a commonly prescribed and inexpensive first-line diabetes treatment that
has shown promise as a novel cancer therapeutic. In epithelial ovarian cancer (EOC),
however, its antineoplastic effects are inconsistent across different study populations
and are confounded by the underlying interaction between cancer and diabetes.242
Some studies have shown that metformin decreases cancer risk in patients with
T2DM,243,244 while others have found no association with the risk of ovarian cancer.245
Similarly, some studies have shown that metformin improves the survival for ovarian
cancer patients with concurrent diabetes181,246,247 while others have found that
metformin use among diabetic EOC patients does not affect progression-free survival or
overall survival.180,248
Metformin lowers glucose in an insulin-independent manner by inhibiting hepatic
gluconeogenesis, increasing peripheral glucose uptake in skeletal muscle and
decreasing renal and intestinal glucose reabsorption (reviewed in 249). These effects are
primarily mediated through the AMPK energy sensing pathway and its downstream
inhibition of mTOR.250,251 Increases in AMPK activation in response to a high AMP:ATP
ratio signal an energy deficit and promote a catabolic state through mechanisms such
as autophagy and the inhibition of anabolic proteins. Pre-clinical studies have shown
that the AMPK pathway also mediates anti-proliferative effect of metformin in EOC.252
Metformin has a further range of effects on cancer cells through AMPK-independent
114
mechanisms that inhibit mTOR signaling253 and cell proliferation, and promote apoptosis
and cell cycle arrest.254,255 Figure 29 illustrates some of metformin’s cellular effects.
Metformin appears to be particularly toxic to p53 negative cells in colon cancer256 and
can selectively kill cancer stem cells in breast cancer.257 However, the greatest potential
of metformin as a cancer treatment may be its synergistic effect with various
chemotherapeutics.258 Metformin potentiates the effects of doxorubicin,257
paclitaxel259,260 and cisplatin261–264 and can resensitize previously chemo-resistant cells.
Authors have postulated that the mechanisms behind this cooperation include the
additive affects of chemotherapy on activation of AMPK259,262, the ability of metformin to
selectively target cancer stem cells that are not affected by chemotherapy alone 257,
further cytotoxicity261, or though inhibition of the MAPK/ERK pathway, leading to
apoptosis.264 Combination with chemotherapy may allow therapeutic effects to be
reached at much lower concentrations than required by chemotherapy as a single
agent.265 The NIH clinical trials database lists three phase II and III trials recruiting to
look at ovarian cancer and metformin + paclitaxel and/or carboplatin.
Although metformin is an antidiabetic drug and diabetes is related to poor cancer
outcomes,199 very little research has examined metformin’s antineoplastic effects
through this mechanism. We hypothesized that despite metformin’s direct cellular
effects on EOC, its primary action in diabetic cancer patients is through the reduction of
systemic fuel availability. The goals of this study were to investigate both the effects of
metformin on diabetic vs nondiabetic populations in a mouse model of EOC, and the
effects of hyperglycemia on metformin action in vitro.
115
Figure 29: Cellular effects of metformin
116
Materials and Methods
Cell Culture
Mouse-derived epithelial ovarian cancer cells (ID8; kindly donated by Drs. Paul
Terranova and Kathy Roby, Kansas State University, Kansas, USA) were maintained in
DMEM supplemented with 10% fetal bovine serum (FBS), 1% antibiotic/antimycotic
(Gibco BRL, Burlington, ON) and 2% L-glutamine (Life Technologies, Burlington ON).
All experimental treatments were performed in serum-free DMEM.
Drugs
Phlorizin (PZ; Sigma, Oakville, ON) competitively inhibits glucose transport through
sodium/glucose cotransporters (SGLTs). Cytochalasin B (CB; Sigma, Oakville, ON)
inhibits glucose transport through the passive glucose transporters (GLUTs). Metformin
hydrochloride was purchased from Sigma (Oakville, ON).
Viability assay
The MTT assay measures cellular reduction of a tetrazolium salt to a coloured formazan
product in an NADH- and NADPH-dependent manner and reflects primarily glycolytic
activity. Cells were seeded at a density of 1500 cells/well in triplicate in 96 well plates
and allowed to attach overnight. Cells were then serum starved for 6 hours and treated
for 48 h in serum-free conditions. Following treatment, cells were incubated with MTT
117
(5mg/ml; Sigma, Oakville, ON) for one hour at 37°C and lysed with MTT lysis buffer
overnight at 37°C. Absorbance was quantified in a microplate reader at 570nm.
Scratch wound assay
To examine the motility of cells in different concentrations of glucose, ID8-6 and ID8-25
cells were grown to confluency in 24 well plates and serum starved for 6 hours. A plastic
pipette tip was used to remove an area of cells down the centre of the dish. Wells were
gently washed with PBS to remove detached cells before treatment. Phase contrast
images were taken at three marked points along the scratch at t=0 and t=24 hours.
Wound sizes were determined using ImageScope software and are presented as the
fraction of the original scratch covered by cells.
Glucose consumption and cell number
Glucose concentration in the cell media was measured using the glucose oxidase
method and reagents from Pointe Scientific (Canton MI, USA). After 24 hours of
treatment, glucose concentration in the sample wells was subtracted from the glucose
concentration in blank wells to determine the amount of glucose consumed. Relative
cell number was approximated by the absorbance of the crystal violet nuclear stain. To
account for any differences in the rate of cell proliferation, glucose consumption was
divided by relative crystal violet absorbance (590nm) to determine a per-cell uptake
comparison.
118
Animals
Wildtype c57bl/6 mice were purchased from Charles River Laboratories (Wilmington,
MA, USA), and c57bl/6 Akt2 null mice (https://www.jax.org/strain/006966) were
purchased from Jackson Laboratories (Bar Harbour, ME, USA). An Akt2 null mouse
colony was established from breeder mice and animals were genotyped prior to
beginning experiments. Animals were housed at the Central Animal Facility at the
University of Guelph and all experiments were conducted according to Canadian
Council on Animal Care guidelines. Mice were kept on a 12h light:dark cycle and had
free access to food and water. For all experiments, adult mice were between 12 and 24
weeks old at time of surgery. Metformin hydrochloride (Sigma, Oakville, ON) was
dissolved in sterile PBS and administered as a daily IP injection at either 50mg/kg/day
or 150mg/kg/day.
Mouse model of epithelial ovarian cancer
Spontaneously transformed murine epithelial cells from C57Bl/6 mice (ID8) were diluted
in 5µl phosphate buffered saline (PBS) and injected under the ovarian bursa of WT or
Akt2 null mice. This orthotopic, syngeneic mouse model of EOC closely replicates
ovarian serous adenocarcinoma in women. By 90 days post-tumour induction, mice
develop large primary tumours, secondary peritoneal lesions, and significant abdominal
ascites.10 In survival experiments, mice were sacrificed when they became moribund
due to the accumulation of ascites fluid.
119
Type 2 Diabetes Mellitus (T2DM)
C57BL/6 mice homozygous for a mutant Akt2 allele have poor glucose tolerance and
reduced insulin sensitivity that result in hyperglycemia and hyperinsulinemia.193,194
Obesity often accompanies T2DM and is independently associated with diabetes and
cancer. Akt2-/- mice are not obese so this model isolates the contributions of high
glucose and high insulin. Mice were genotyped using the following primers designed by
Jackson Laboratories: Akt2 wildtype reverse: TGCACAATCTGTCTTCATGCCAC; Akt2
common forward: ACCAACCCCCTTTCAGCACTTG; Akt2 mutant reverse:
TACACTTCATTCTCAGTATTGTTTTGC. Products are 110bp (wildtype), 277bp
(mutant), or 110bp and 277bp (heterozygote).
Food Consumption
Weekly food consumption was calculated by taking the initial weight of dry food and
subtracting food weight after 5 days. This value was divided by the number of mice in
the cage and the number of days in the measurement period to determine food
consumed per mouse per day.
Intraperitoneal glucose tolerance tests
Mice were fasted for four hours prior to intraperitoneal glucose tolerance tests (IPGTT).
After recording baseline blood glucose, mice were given an intraperitoneal injection of
glucose (1 mg/g body weight in saline solution at a volume of 10µl/g body weight).
Blood glucose measurements were taken at 15, 30, 60, and 120 minutes after glucose
120
challenge using a handheld glucose monitor (Freestyle Lite, Abbott Laboratories, Abbott
Park, IL). Values that exceeded the measurement range of the glucose monitor were
considered 27.8mM (maximum reading possible). Data are presented as the average
integrated area under the curve (iAUC) ± SEM. The iAUC is the area under the curve
relative to FBG at time 0 and was calculated using the trapezoidal method.
Tissue collection and preparation
Mice were fasted for 6 hours prior to sacrifice by cervical dislocation. Following
euthanasia, trunk blood was collected and fasting blood glucose was measured
immediately using a handheld glucometer. The remaining blood was allowed to clot
then centrifuged to obtain serum samples and stored at -80oC. Ovarian tumours were
removed, weighed, and divided into sections for subsequent analysis. Sections were
either fixed in 10% neutral buffered formalin (Fisher Scientific, Whitby, ON) for wax
embedding and tissue sectioning or flash frozen in liquid nitrogen and stored at -80oC
for protein extraction. Secondary disease was evaluated at time of sacrifice by ascites
volume and number of secondary lesions. Mice were scored on a 4-point scale with 0
indicating absence of secondary disease. Mice with scores of 0 and 1 were considered
“low” and scores of 2 and 3 were considered “high”.
Statistical Analysis
Data were analyzed using GraphPad Prism statistical analysis software. Results are
reported as mean ± standard error of the mean (SEM). Significance threshold was set
121
at 0.05 and p values are listed in the figure legends. Comparisons between means were
measured by two-way ANOVA followed by a Bonferroni correction, or by unpaired twotailed Student’s t-test, unless otherwise noted. Overall survival data are plotted on a
Kaplan-Meier survival curve and differences were determined by a log-rank (MantelCox) test.
122
Results
In vitro
Metformin affects systemic metabolism in diabetic patients but has little risk of
hypoglycemia in nondiabetics. To determine if there were also differential effects at a
cellular level in EOC cells, we tested the effects of metformin in physiological glucose
concentrations (6mM) and hyperglycemic concentrations (25mM). The MTT assay was
used to measure the metabolic viability of ID8, SKOV3, and normal human ovarian
epithelium (NOSE) treated with metformin. The concentration of glucose did not affect
the metabolic response to metformin (Figure 29A).
In previous experiments we have found that the chronic culture conditions of cells can
affect their response to treatment media with different concentrations of glucose. We
compared the effects of metformin in 6mM glucose media on cells chronically cultured
in normoglycemic conditions (-6 cell lines) and those cultured in diabetic conditions (-25
cell lines). The metabolic viability of cells acclimatized to hyperglycemic conditions was
significantly more sensitive to the effects of metformin. ID8-6 cells showed no significant
response to metformin until 80mM and SKOV3-6 were unaffected until 40mM.
Interestingly, the metabolic response of SKOV3-6 cells was identical to that of normal
ovarian epithelium (Figure 29B).
In ovarian cancer cells acclimatized to hyperglycemia (ID8-25, SKOV3-25; Figure 31,
part A) there were dose-responsive declines in viability that were significant after 5mM
and 2.5mM metformin respectively. These findings are in line with other studies which
have also found no direct toxicity (necrosis, apoptosis) in this dose range indicating that
123
it is at a therapeutic level.266 The viability of NOSE cells was only reduced at metformin
concentrations greater than 40mM.
The MTT assay measures NADH formed primarily through glycolysis compared to
oxidative phosphorylation. These results imply that the glycolytic flux in “diabetic” cancer
cells is more sensitive to inhibition by metformin.
124
Figure 30: Metformin inhibits EOC cell viability in high glucose
MTT absorbance was used to determine cell viability. A) Viability of mouse epithelial
ovarian cancer, human ovarian cancer, and normal human ovarian epithelial cells (ID8,
SKOV3, NOSE) in media containing physiological glucose (6mM) or high glucose
(25mM). B) Viability of cells chronically cultured in physiological (red, 6mM glucose) or
hyperglycemic (blue, 25mM glucose) conditions. Left: murine, right: human. Black
dashed line indicates cells derived from normal epithelial and cultured in physiological
glucose. n=3. Values are mean ± SEM. Stars indicate differences between cell lines at a
single metformin concentration. *p<0.05, **p<0.01, ***p<0.001.
125
Although metformin had a greater effect on metabolic viability in high glucose, cell
number appeared to be more affected in conditions of glucose deprivation (Figure 30A).
Consistent with this, the ability of metformin-treated cells to heal a scratch wound was
impaired in low but not high glucose (Figure 30B).
126
Figure 31: Metformin reduces cell number in glucose-deprived cultures
A) ID8-25 cells were cultured in different glucose concentrations for 48 hours with or
without 50mM metformin. Crystal violet absorbance at 590 nm was used as an indicator
of cell number. Bars are mean + SEM. Stars indicate significant differences from crystal
violet absorbance in 6mM glucose. Brackets indicate differences between treatments at
the same glucose concentration. n=2; **p<0.01, ***p<0.001. B) Scratch wounds were
made by clearing an area of cells with a pipette tip. Images were analyzed to determine
the fraction of the scratches covered by cells after 48 hours of treatment with 5mM
metformin. Bars are mean + SEM, n=4. **p<0.01, ***p<0.001.
127
Metformin significantly increased cellular glucose consumption regardless of
environmental glucose availability (Figure 31). To determine whether glucose uptake
capacity contributed to this increase in glucose uptake, we used glucose transport
inhibitors to both the passive (GLUT; cytochalasin B) and active (SGLT; phlorizin)
glucose transporters. In low glucose, total glucose uptake was still enhanced compared
to control in the presence of glucose transport inhibitors. In high glucose, GLUT
inhibition abrogated this effect and glucose uptake was unchanged with cytochalasin B
treatment (Figure 32A). On a per-cell basis the increases in glucose consumption were
reduced, however a statistical trend remained (p<0.09) (Figure 32B).
128
Figure 32: Metformin increases cellular glucose consumption
Cells were treated with 5mM metformin in media containing normal (6mM) and high
(25mM) concentrations of glucose. A) Total glucose consumed in mg/dL B) Relative
glucose consumption per cell. C) Fraction of glucose consumed relative to the initial
glucose concentration in the cell media. Bars are mean + SEM, n≥5. Different letters
represent significant differences between bars (p<0.05).
129
Figure 33: Metformin-mediated increase in glucose consumption is independent of
glucose transporter expression
Cells were treated with 5mM metformin dissolved directly into culture medium, 50µM
phlorizin (PZ in 0.01% DMSO; SGLT inhibitor), or 0.5µM cytochalasin B (CB in 0.01%
DMSO; GLUT inhibitor) in media containing normal and high concentrations of glucose.
A) Total glucose consumed in mg/dL. B) Relative glucose consumption per cell. Bars
are mean + SEM, n≥4. *p<0.05, **p<0.01, ***p<0.001 between metformin and vehicle
control.
130
In previous work we have shown that EOC progresses faster in both type 1 and type 2
diabetic mice than in nondiabetic animals. We predicted that normalizing blood glucose
in diabetic animals would improve EOC outcome by reducing the amount of systemic
glucose available for the tumour to consume. Metformin has also been shown to have
beneficial effects in normoglycemic preclinical models through direct effects on cellular
signalling pathways. In order to assess the effects of metformin on EOC tumour growth
we used induced tumours in both wildtype (WT) mice and a genetic model of non-obese
mice with a T2DM hyperglycemic, hyperinsulinemic phenotype (Akt2 KO).
Three in vivo studies were performed using a mouse model of ovarian cancer
developed in our lab. Briefly, syngeneic ID8 cells were injected orthotopically beneath
the ovarian bursa of C57bl/6 mice. These cells develop into disease that closely
replicates human serous ovarian cancer. Tumour development over 30, 60 and 90 days
is representative of EOC Stages 1, 2 and 3 respectively.
In population studies, the use of metformin has been correlated with a lower risk of
developing cancer. The first study was designed to evaluate the ability of chronic
metformin administration to prevent tumour growth. WT and Akt2 KO mice received
daily IP injections of either metformin or PBS control at 50 mg/kg body weight between
two weeks after surgery until 90 days PTI. Metformin had no effect on fasting blood
glucose in WT mice. In KO mice, although FBG dropped significantly over time posttumour induction, the effects of metformin and PBS were identical (Figure 33B). At 90
days PTI, the weight of tumours in the metformin treated mice was no different than
those in PBS treated mice within each genotype (Figure 33A). In contrast to previous
experiments which showed that Akt2 KO mice grew significantly larger tumours than
131
WT mice 90 days PTI (Figure 7A), a one-way ANOVA of control and metformin in both
genotypes (comparison between Figure 33As) in this trial showed no significant
differences in tumour weight. Tumour initiation in this trial was weak and the lack of a
genotype-related effect was probably due to the very small size of these tumours.
132
Figure 34: Small tumours normalize blood glucose in diabetic mice
Tumours in wildtype and diabetic (Akt2 -/-) mice were treated with daily IP injections of
50mg/kg/day metformin (2 weeks PTI to 60 days PTI) then 150mg/kg/day (60 until 90
days PTI). A) Metformin treatment did not affect tumour. B) Fasting blood glucose. Bars
are mean + SEM, n=5. *p<0.05, **p<0.01, ***p<0.001.
133
The human dose is of metformin is approximately 35mg/kg/day orally.267 In our
experiments, the original dose of 50 mg/kg/day IP was chosen based on literature at the
time which used this dose in chronic administration.268 However, because this dose did
not affect the blood glucose in the Akt2 KO mice as we had expected, we increased the
daily dose of metformin to 150 mg/kg/day in a second study.263 In this second study, we
again began metformin treatment at two weeks PTI in order to evaluate the effects of
metformin on tumour growth prevention. Mice were sacrificed when they became
moribund. We found no differences in the length of survival or tumour weight at survival
in WT mice treated with either metformin or PBS control (Figure 34A and B).
Qualitatively there was a slight reduction in the number of secondary lesions, but our
scoring system for evaluating secondary disease did not allow us to determine statistical
significance from these data (Figure 34C).
We measured food consumption to determine whether the drug was affecting total
glucose intake and found that metformin had no effect (Figure 35A). In PBS treated
mice, the presence of the developing disease resulted in significant fluctuations in
fasting blood glucose. Metformin appeared to protect against this variability (Figure
35B). We performed IPGTTs in mice with and without tumours to determine whether
metformin alone could affect glucose tolerance. Metformin had no effect on the glucose
tolerance of mice with tumours, but surprisingly it significantly worsened glucose
tolerance in mice without tumours (Figure 35C and D).
134
Figure 35: Daily metformin treatment does not affect EOC outcomes in WT mice
A) Tumours were removed and weighed when mice became moribund. Values are
mean + SEM, control: n=9, metformin: n=6. There was no significant difference in
tumour weight at death between metformin- or control-treated mice. B) Kaplan-Meier
survival curve of WT mice with ovarian cancer that received 150mg/kg daily metformin
in PBS IP treatment from two weeks PTI until death. There is no statistical difference in
survival between groups by log-rank (Mantel-Cox) test. C) Secondary lesions and
ascites volume. Values are proportion of mice with “high” or “low” scores.
135
Figure 36: Metformin treatment in WT mice with tumours impairs glucose tolerance over
time.
WT mice bearing tumours were treated with daily metformin injections (150mg/kg)
beginning two weeks PTI. A) Food consumption at 2, 7 and 10 weeks PTI (0, 5 and 8
weeks of metformin treatment). Bars are mean + SEM, n=5. B) Fasting blood glucose
136
between surgery (0 weeks) and death (†). Bars are mean + SEM, n=5; *p<0.05 vs
baseline. Comparison of FBG from baseline is by two-way ANOVA with Dunnett’s posthoc test. C) Results of intraperitoneal glucose tolerance tests (IPGTT). ***p<0.001 vs
control at a specific time point. D) Integrated area under the curve (iAUC) of IPGTTs.
Glucose tolerance was impaired at 90 days PTI vs 60 days PTI in metformin-treated
mice with tumours. *p<0.05. Control: n=9, metformin: n=6.
137
In diabetic mice (Akt2 KO), there were no differences in tumour weight, length of
survival or severity of secondary disease (Figure 36). Mice in this trial died earlier than
expected (Figure 36B), likely due to tumour initiation with high passage cells that
accelerated disease progression. Food consumption decreased over time as EOC
progressed, but there were no differences between PBS and metformin treatments
(Figure 37A). In IPGTTs, metformin had no effect on glucose tolerance regardless of the
presence of a tumour. However, in all Akt2 KO groups, glucose tolerance improved with
time and the addition of metformin did not add to these effects (Figure 37D). Mice
without tumours that were treated with metformin showed the same improvement in
glucose tolerance as untreated mice with tumours. As in the wildtype mice, although
there were clear differences in glucose tolerance between groups, FBG was more
variable (Figure 37B).
138
Figure 37: Daily metformin treatment does not affect EOC outcomes in Type 2 diabetic
mice
Tumours were removed and weighed when mice became moribund. Values are mean +
SEM, control: n=14. There was no significant difference in tumour weight at death
between metformin- or control-treated mice. B) Kaplan-Meier survival curve of WT mice
with ovarian cancer that received 150mg/kg daily metformin in PBS IP treatment from
two weeks PTI until death. There is no statistical difference in survival between groups
by log-rank (Mantel-Cox) test. C) Secondary lesions and ascites volume. Values are
proportion of mice with “high” or “low” scores.
139
Figure 38: The presence of a tumour or treatment with metformin both improve glucose
tolerance in diabetic mice
Akt2 KO mice bearing tumours were treated with daily metformin injections (150mg/kg)
beginning two weeks PTI. A) Food consumption at 2, 4 and 7 weeks PTI (0, 2 and 5
weeks of metformin treatment). Bars are mean + SEM, n=5. B) Fasting blood glucose
140
between 3 weeks PTI (1 week of metformin treatment) and 8 weeks PTI (six weeks of
metformin treatment. Bars are mean + SEM, n=5; Brackets indicate significant
differences between bars. *p<0.05. C) Results of intraperitoneal glucose tolerance tests
(IPGTT). D) Integrated area under the curve (iAUC) of IPGTTs. In Akt2 KO diabetic
mice, glucose tolerance between 30 and 60 days PTI was improved by the presence of
tumours. Mice that underwent a sham surgery showed improved glucose tolerance with
metformin treatment *p<0.05. Control: n=9, metformin: n=6.
141
In the third study, we investigated whether metformin administered to WT mice with
established tumours (from 60d PTI to 90d PTI), could have an effect on disease
progression/regression. There were no differences in tumour size.
Figure 39: Metformin treatment does not reduce the weight of established tumours
Wildtype mice were allowed to grow tumours for 60 days before 30 days of daily
treatment with 150mg/kg/day metformin IP. Bars are mean tumour weight at 90 days
PTI + SEM, n=5.
142
Discussion
Preclinical evidence suggests that the glucose-lowering drug metformin is effective in
limiting EOC growth in vitro. The OCT1 transporter is necessary for metformin entry into
cells and is expressed in the human ovarian cancer cells lines SKOV3 and OVCAR3, as
well as in malignant cells in ovarian tumours.269 Metformin can inhibit proliferation,
migration, invasion, and adhesion in EOC monolayer culture as well as inhibit the
formation of tumour spheres, induce cell cycle arrest and increase apoptosis.260–263,270
We have previously shown that both Type 1 and Type 2 diabetes significantly
accelerate the growth of ovarian tumours in a mouse model of EOC. The goal of this
study was to determine if EOC progression can be prevented or reversed by lowering
blood glucose. Metformin is a widely prescribed antidiabetic drug which has recently
been shown to improve both the risk and outcome of a number of cancers. By inducing
ovarian cancer in both WT and diabetic (Akt2 KO) animals, we hoped to identify any
protective effects of metformin through its action in lowering systemic hyperglycemia as
well direct its effects on EOC cells.
Cellular effects in different glucose levels
The effects of metformin appear to be dependent on the level of environmental glucose.
Metformin reduced cell number in restricted glucose but not hyperglycemia (Figure 30A)
and it inhibited wound healing in normal but not high glucose (Figure 30B). These
findings are consistent with other studies that have shown high glucose impairs or even
reverses the anti-cancer effects of metformin.271,272 When treated with metformin, cells
143
in 5mM glucose show gene expression changes in pro-tumorigenic pathways; this effect
is abolished cells in high glucose (10mM or above).216 This is perhaps not surprising
because many of metformin’s effects are dependent on AMPK activation, which also
responds to the AMP:ATP ratio. A low AMP:ATP ratio indicates high energy availability
and would oppose the effects of metformin, meaning that the same drug dose in high
glucose would be less effective at promoting AMPK-mediated anti-neoplastic effects.
Several studies have suggested that increased metformin efficacy in low glucose is due
partly to downregulation of the glycolytic enzyme pyruvate kinase M2 (PKM2) in these
conditions.271,273,274
The effects of metformin on a systemic level are mostly concentrated in the liver and
skeletal muscle where AMPK activation increases glucose uptake and decreases
gluconeogenesis to reduce blood glucose. In vivo, we evaluated metformin’s effect on
tumour weight and overall survival. Although metformin has direct effects on cells, we
anticipated that when given systemically in hyperglycemic mice, metformin’s anti-cancer
effects would be mediated primarily through its ability to reduce blood glucose.
In our model, metformin was unable to oppose tumour growth in either normal or
hyperglycemic mice. The doses of metformin that we used (50mg/kg/day and
150mg/kg/day) did not reduce blood glucose and so we cannot draw conclusions on the
effect of lowering glucose on the progression of EOC. In a pilot study, we administered
up to 250 mg/kg/day IP in an effort to reduce blood glucose, however within four days of
this dosing three mice had died and we reduced the dose to 150mg/kg/day. However,
the lack of effect of metformin on tumour growth in normoglycemic mice suggests that it
has limited influence directly on cancer cells. This finding is in disagreement with a
144
number of studies of EOC that demonstrate metformin can reduce primary tumour
growth and proliferation, and the formation of metastases.260,262,263,266 Although the
metformin doses in these studies were similar to what we used, these studies were all
performed with subcutaneous xenografts in immunocompromised mice. Both the
ovarian microenvironment10 and immune-related inflammation43 significantly impact
tumour biology, which could explain why metformin treatments were ineffective in our
orthotopic model.
Litchfield and others used a similar ID8 orthotopic model and found that metformin
inhibited tumour growth in normoglycemic mice but not diet-induced hyperglycemic
mice.272 In their study, metformin treatment began three weeks prior to tumour induction
and the tumours that developed were much smaller than what we typically see (~40mg
in untreated normoglycemic mice and ~90mg in untreated hyperglycemic mice vs
~150mg and ~350mg in our mice). Taken together with our findings, these data suggest
that metformin’s effects against EOC are primarily preventative rather than reductive.
These same conclusions have been drawn in colorectal cancer where metformin has
been found to limit tumour initiation but have no effects on existing tumours.275
Alterations in angiogenesis may provide the mechanism for this limited tumour growth.
Metformin can decrease VEGF and/or microvessel density262,276 and in a breast cancer
model it has been shown to reduce tumour angiogenesis, but not affect tumour
volume.270
145
Metformin and reprogramming of cellular metabolism
Several studies contend that metformin's inhibitory effects on cancer progression are
cancer cell autonomous and AMPK-independent, instead acting through inhibition of
mitochondrial enzymes to decrease cellular respiration and promote a compensatory
increase in glucose uptake and glycolytic flux.277–280 We found that although short-term
changes in environmental glucose did not affect viability with metformin treatment
(Figure 29A), chronic culture in high glucose was associated with reduced viability, i.e. a
decrease in glycolytic flux (Figure 29B). A highly glycolytic phenotype is generally
related to more aggressively malignant cells, but forcing glycolysis also increases the
tumour’s demand for glucose and can make cells acutely sensitive to glucose
deprivation. The sensitivity of cells to metformin is dependent on their ability to cope
with energetic stress.280 In the presence of glucose, metformin can suppress the overall
efficiency of metabolism to inhibit the proliferation of human colon and thyroid
cancers.271 When glucose becomes a limiting factor for metabolism, this metformininduced metabolic inefficiency induces cell death.271,277,279 The higher rate of glucose
consumption we demonstrated with metformin treatment (Figure 31B) may suggest that
metformin increased glycolytic activity.
Dosing
Drug repositioning often means using different concentrations to elicit different
physiological responses. In diabetes, the effective dose for metformin has maximal
pharmacological effect on tissue-specific metabolism in the liver and skeletal muscle,
but minimal systemic toxicity. Using metformin in cancer treatment may require a
146
different dose and different dosing schedule that induces both pharmacological and
toxic effects.281 For example, metformin at a concentration that effectively upregulates
pAMPK may have no influence on decreasing mammary cancer severity in nondiabetic
rats and mice.282 There are also large differences in the drug concentrations that are
effective over different time periods. One group refers to the “metformin paradox” where
long term metformin treatment over weeks or months requires only micromolar
concentrations of metformin to elicit anti-cancer effects, whereas short term (12-48
hour) experiments need higher drug concentrations far beyond the human therapeutic
dose of 15-17 µM.216 It is possible that rather than having a dose-responsive effect
against cancer, there is a threshold concentration for drug effect that 150mg/kg/day did
not reach which could partially explain the lack of effect that we saw in our in vivo
models.
Conclusion
This study does not support a role for single-agent metformin treatment of EOC in either
physiological or diabetic environments. We were unable to evaluate its effect on tumour
growth through reduction in blood glucose, however others have shown that in diabetic
animals, combined treatment of both anti-tumour and anti-hyperglycemia therapies give
the best tumour-reductive outcome.213 Our in vitro findings support reports that high
environmental glucose inhibits metformin’s anti-cancer effects. Taken together with our
previous work that showed markedly worse EOC outcomes in diabetic mice, we
conclude that glycemic control is an important consideration in EOC management. In
diabetic mice the lower cell-level efficacy of metformin in hyperglycemia could be offset
147
by the fact that it reduces this hyperglycemia through actions on the liver and skeletal
muscle. If there is a role for metformin in the treatment of EOC, the variability of its
effects in preclinical models has shown that it will be essential to identify the subset of
patients likely to respond, which could include evaluation of systemic glucose
homeostasis, concurrent therapies, OCT1 transporter expression283 and oncogenic
mutations.284
148
CHAPTER 5: GENERAL DISCUSSION
Overview
Glucose is an important energy substrate, survival factor, and pro-angiogenic molecule.
This thesis examines the influence of high glucose availability on the growth of epithelial
ovarian tumours, and attempts to the regulation of systemic glucose can be exploited
therapeutically in cancer treatment.
Since beginning this work, a number of significant discoveries have altered our
understanding of cancer cell metabolism and tumour metabolism. In 2011, Hanahan
and Weinberg updated their seminal “Hallmarks of Cancer” to include “reprogramming
energy metabolism.”43 Hexokinase52,98,99 and the PKM2 fetal isoform of pyruvate
kinase91 have emerged as key drivers of the Warburg effect. The sodium glucose
transporter 1 (SGLT1) was found to be expressed in cancer cells and physically
stabilized by EGFR, suggesting a role for EGFR in glucose balance.145,228 Perhaps most
significantly, the meaning of abnormal “cancer metabolism” has expanded to include
both cellular and intercellular characteristics. A metabolic symbiosis has been
discovered between oxygenated and hypoxic cancer cells as they share lactate to
maximize energy production.119 In addition, observation of a “Reverse Warburg” effect
demonstrated an essential role for metabolically reprogrammed cancer-associated
fibroblasts.116 Disrupting the unique energy balance of both cancer cells and tumours
have direct anti-cancer consequences.
Advances in the efficacy of EOC treatment have come from stratifying patients by
disease type on histological and genetic levels. However, neither of these takes into
149
account the functional bioenergetic heterogeneity inherent at the levels of the cell,
tumour and body. As the results presented here demonstrate, tumours with the same
histologic and genetic characteristics can behave differently in different glucose
conditions.
As studies move toward the development of clinical applications, research will need to
consider another level of complexity—interaction with the systemic environment. The
results of this thesis demonstrate pro-tumorigenic roles for both long-term and shortterm exposure to systemic hyperglycemia and implicate common metabolic
comorbidities of EOC, particularly diabetes, as significant challenges to address. Can
systemic metabolic dysfunction mask cancer diagnosis or monitoring, particularly using
FDG-PET? Can it increase cancer aggressiveness and lead to poorer prognoses by
providing an unlimited fuel supply? Do standard treatments for diabetes, like insulin or
lactate infusions, have detrimental effects by enhancing tumour growth or interfering
with cancer treatments?
The key findings of this thesis are that:
1. Hyperglycemia promotes EOC independent of insulin;
2. Tumour development ameliorates dysfunctional systemic metabolism;
3. Knockdown of the active glucose transporter SGLT2 promotes EOC spread;
4. Secondary disease predicts death independent of primary tumour size; and
5. Anti-neoplastic effects of metformin are highly glucose-specific.
150
Figure 40: EOC metabolism is a product of cell genotype, the tumour microenvironment,
and systemic glucose conditions
The metabolic syndrome, diabetes, and EOC
The metabolic syndrome is defined by the pathologic triad of hyperglycemia and
hyperinsulinemia (T2DM), and obesity. It is a significant and growing population health
concern with as much as one third of the population affected. The syndrome has health
151
effects on multiple organs and is related to various vascular and microvascular
diseases.198 Its effect as both a risk factor and poor prognostic factor in cancers is
undeniable but poorly understood.199 In some studies, ovarian cancer has been linked
to the metabolic syndrome and each of its three conditions independently. 178,179
The effects of mitogenic insulin signalling are thought to be the primary drivers of this
association. However, insulin-independent mechanisms may also contribute, particularly
through energy-sensing pathways and glucotoxic damage. Toxicities of excess glucose
could benefit tumour progression by facilitating cellular and genetic damage through the
development of an acidic environment, transient hypoxia, oxidative stress, and glycation
including the development of advanced glycation end products and their receptors
(AGE/RAGE).186,285 Taken together hyperglycemia provides ongoing genomic instability,
one of the hallmarks of cancer.43
Dysregulated systemic metabolism can also affect cancer cell metabolism. The aerobic
glycolysis in many cancer cells is associated with more aggressive malignancies in
vitro.48,121 Highly glycolytic cells demonstrate the Crabtree effect where glucose induces
a further shift away from oxidative energy production. Diabetes, particularly T1DM, is
associated with a “thrifty” genotype where cells are genetically or epigenetically
reprogrammed take up more glucose.286 It is possible that this is a risk factor for cellular
transformation or a promotional influence on tumour development.
The metabolic syndrome is primarily a consequence of lifestyle and thus understanding
its connections to cancer can potentially indicate modifiable risk factors. In ovarian
cancer, where the etiology is still poorly understood and treatment of advanced disease
152
is still largely ineffective over the long term, this would be a significant finding.
Moreover, researchers have suggested that the effects of diabetes can be extrapolated
to all cancer patients as the associated risks may occur within the normoglycemic and
insulinemic ranges.184
Diabetes treatment and cancer
Treatments for diabetes and cancer can interact in detrimental ways. Antihyperglycemic treatments inducing elevated plasma insulin (sulfonylureas) seem to
increase cancer risk but insulin-sensitizers (metformin, thiazolidinediones) seem to
reduce cancer risk.242,287,288 Some groups have proposed that the increased risk of
cancer in diabetics is not attributable to insulin itself, but rather to the effects of insulin
insensitivity on the dysregulation of AMPK signalling or the action of insulin-like growth
factor-1 (IGF-1)-cholesterol synthesis pathway.202
On the other hand, patients with diabetes tend to receive less aggressive cancer
treatment.289 Diabetes is associated with higher infection rates177 and compared with
their nondiabetic counterparts, cancer patients with pre-existing diabetes are
approximately 50% more likely to die after surgery.290 This high operative morbidity may
limit the number of diabetic patients having comprehensive surgical staging179 and
explain why diabetic patients are more often diagnosed with higher tumour stages. 289
153
Coordinated metabolism in the tumour microenvironment
Transformed cells are one part of a complex and heterogeneous tumour
microenvironment that plays an essential role in cancer progression.114 Our lab has
shown that the interaction of transformed ovarian epithelial cells with the ovarian stroma
is crucial to the development of highly aggressive tumours in the mouse model
described in this thesis.10 The microenvironment, particularly its oxygen levels, also
influences metabolism.121 While the Warburg effect describes genetic adaptations of
tumour cells, the Crabtree effect (inhibition of oxphos and promotion of glycolysis in high
glucose) describes reversible, short term adaptations to changes in energy and oxygen
balance which may occur at protein or epigenetic levels. In the 1960s, the Crabtree
effect was considered characteristic of cancer cells and part of the malignant
transformation.291 Although the Crabtree effect is not often mentioned in contemporary
literature, in addition to the Warburg effect it is important to consider this as part of the
microenvironment-metabolism interaction.217,292
The reverse Warburg effect demonstrates a central role of reprogrammed cancerassociated fibroblasts (CAFs) in tumour metabolism.116 Similarly, lactate is shared in a
metabolic symbiosis between oxic and hypoxic cancer cells within a tumour. 119 The in
vitro studies in this thesis show that increased environmental glucose can influence
cells to increase their glucose uptake. However, in vivo tumours are poorly perfused
tissues. Interestingly, the reverse Warburg effect shows that tumours can enrich their
own microenvironment and thus gain some autonomy from systemic circulation and the
need for tumour vascularization or angiogenesis.105 This could have implications for the
effectiveness of antiangiogenic therapies or chemotherapy delivery to the tumour. In
154
hyperglycemic conditions, tumours are also able to increase glucose uptake by
promoting glucose transfer across the capillary walls into the IF, which occurs without
the need for glucose transport proteins.214 Glucose can accumulate within the tumour
tissue to potentially create a locally hyperglycemic environment, which could explain the
ability of passive GLUTs to compensate for lack of active SGLT2-mediated glucose
transport (Chapter 3).
Intersection of tumour metabolism and systemic metabolism
Just as cellular growth occurs within the tumour microenvironment, tumour growth
occurs within the larger ecosystem of the host which varies in its own systemic
metabolism. In this thesis we hypothesized that high blood glucose levels would
promote tumour growth by providing an abundance of energetic fuel. However, the
relationship between systemic and tumour glucose control is reciprocal and cancer itself
can drive changes in systemic metabolism. We saw that tumours in our animal model
actually improved the diabetes of both STZ and Akt2 KO mice. In T1DM mice lacking
insulin, the presence of a tumour improved glucose tolerance (Figure 6). In
hyperinsulinemic Akt2 KO mice, tumours improved insulin sensitivity (Figure 8A and B).
In fact, tumours appeared to be as effective as metformin at normalizing glucose
tolerance in T2DM mice (Figure 37D).
The high glucose demands of tumour tissue might suggest that it acts as a glucose sink
and improvements in systemic glucose would be due to a large tumour mass. However,
the tumours in the chronically diabetic STZ mice with improved glucose tolerance were
155
smaller than tumours in control mice (Figure 4A). This could indicate that the tumour is
influencing systemic glucose indirectly through secreted factors in the blood.
Hypoglycemia has been noted in many cancer patients and is thought to be due to a
combination of cachexia, renal and hepatic impairment, excessive glucose consumption
or decreased glucose production293 which can be resolved with removal of the
tumour.294 However, the body is able to compensate for much of the tumour’s effect, in
some cases overcompensating,295–297 by increasing glycogenolysis and
gluconeogenesis from non-carbohydrate sources212,298 This compensation at the site of
the liver could lead to relative hyperglycemia in the hepatic veins and might contribute to
the fact that the liver is such a common site of metastasis (Figure 27A). Similarly, we
saw that the intestines were a major site of metastasis. Aside from being well perfused,
metastatic cells in this location may co-opt dietary glucose as it is absorbed.
High blood glucose, or an associated hyperglycemic peritoneal fluid, could promote the
survival of circulating tumour cells with metastatic potential. If high blood glucose is
cancer-promoting, then it might be expected that haematological malignancies would be
most acutely accelerated in diabetic conditions, however there is no published data
looking at these effects.
Diabetes has similar influences on the body as cancer: cachexia, heightened
gluconeogenesis concomitant with an increase in liver mass, and dysregulation of
insulin and glucagon levels. At a cellular level diabetes induces similar host
compensation: impaired oxphos leads to increased glucose tolerance, reduced fat
mass, and increased insulin sensitivity, which oppose the progression of the metabolic
156
syndrome.299 As a comorbidity with EOC, diabetes may accelerate these cancer-driven
changes in the host ecosystem.
Glucose transport and diagnostic potential
These studies demonstrated that glucose enters cancer cells through both GLUT and
SGLT. It is possible that PET scans using FDG are only accounting for a portion of
glucose uptake, underestimating the number and metabolic activity of malignant lesions.
A first generation SGLT-specific tracer, alpha-methyl-4-deoxy-4-[18F]fluoro-Dglucopyranoside (Me4FDG) has been developed.130 Because SGLT-mediated glucose
transport is reliant on the extracellular sodium concentration, using multi-modal uptake
imaging combining FDG + Me4FDG + Na+ could add power to PET scanning in
diagnosis and treatment monitoring.300–303
Implications for EOC treatment
The best available treatment tools against ovarian cancer are surgery304, imaging and
monitoring blood levels of CA-125305–309, and platinum-based chemotherapies310, all of
which are confounded by hyperglycemia.
In addition to the dysregulation of systemic metabolism caused by the tumour itself,
many standard cancer treatments induce hyperglycemia. Total parenteral nutrition, tube
feeding, steroids and glucocorticoids are often used in oncology and have been shown
to result in hyperglycemia.177,311 Intravenous solutions that contain lactate (Ringer’s,
157
Hartmann’s) are also often given to cancer patients and, given the tumour’s ability to
use lactate as a fuel, may in fact be detrimental.105 Even in previously diabetic patients,
hyperglycemic treatment varies widely. In many cases, comprehensive glucose
management may not be included as part of an overall treatment plan for the cancer,312
partly because it can be particularly difficult to both aggressively monitor blood glucose
and maintain quality of life during cancer treatment.313 Since blood glucose can often be
managed effectively through diet, nutrition could become an important part of EOC
treatment. Specific low-carbohydrate diets have been suggested as a means to starve
cancer cells of glucose, and studies using ketogenic diets have shown potential anticancer benefits.314 However, ketones have recently been shown to actually fuel cancer
in part because of their contributions to the “reverse Warburg” effect. The ability of
tumours to use ketones as fuel has even been suggested to partly explain the diabetescancer connection.105 Taken together, any modifications to systemic fuel availability
should be taken into consideration as potentially confounding factors when treating
ovarian cancer patients.
Study Limitations
One of the most significant limitations of cancer metabolic research in vitro is that cell
culture conditions are hyperglycemic and hyperoxic,64 both of which directly oppose the
usual conditions of the tumour microenvironment. The phenotypic differences between
ID8-6 and ID8-25 cells clearly demonstrated that high glucose medium influences cell
robustness and metabolism.
158
Through the use of both Type 1 and Type 2 diabetic animal models, we were able to
separate some of the effects of insulin from those of glucose. However, the reductive
approach we took did not consider other contributors to glucose regulation, particularly
other hormonal factors.315 Furthermore, we did not consider the effects of inflammation,
immunity and angiogenesis which are inextricably linked to both metabolism and tumour
growth.
Limitations of animal models
The Akt2 KO model was originally used because of its non-obese type 2 diabetic
phenotype, however it appears that under some conditions, Akt2 KO can also influence
tumour development, potentially confounding our results.316 In addition, we were unable
to reduce systemic blood glucose with metformin in these mice and therefore could not
directly test the hypothesis that normalizing blood glucose could reduce tumour growth
in diabetic environments. This lack of impact on blood glucose could be due to opposing
effects of the Akt2 KO and metformin. Akt2 plays an important role in the exocytosis of
intracellular GLUT4 to the cell membrane to facilitate insulin-dependent glucose
uptake.317 Eliminating Akt2 expression in mice results in insulin insensitivity and
consequently hyperglycemia. On the other hand, metformin achieves most of its
antidiabetic effects by increasing glucose uptake into skeletal muscle. Metformin
decreases the endocytosis of GLUT4 transporters from the cell membrane back into the
cytosol.318 The inactivation of Akt2-mediated GLUT4 trafficking antagonizes metformin’s
ability to maintain transporter expression and increase glucose consumption in cells,
which would limit its ability to reduce blood glucose in Akt2 KO mice.
159
Another limitation related to in vivo work was the difficulty in characterizing secondary
disease, which we found to be an essential measure of morbidity. In chronically diabetic
mice, we saw significant reductions in overall survival compared to normoglycemic
animals together with significantly smaller tumours which suggested a large contribution
of secondary disease. In addition, mice with SGLT2-deficient tumours presented with
many metastatic lesions that were strongly associated with poor survival. In women
metastases lead to death because of their physical affect vital organs, but the
secondary lesions in our model were too small to have this same influence. “Secondary
disease” includes humoral and intraperitoneal factors that are not accounted for in our
scoring method.
Conclusions
Collectively, the data presented in this thesis indicate that altered tumour metabolism is
much broader than the Warburg effect. The effects of systemic metabolic comorbidities
such as diabetes can confound EOC diagnosis, treatment and overall survival through
fundamental changes in tumour biology.209 Antidiabetic pharmaceuticals have great
potential as adjuvants to cancer treatment and because altered glucose homeostasis is
common in cancer patients, these therapies have wide applicability to normoglycemic
patients. Moreover, the identification of SGLT2 as a powerful tumour suppressor opens
exciting possibilities for improved screening and drug targeting.
160
REFERENCES
1.
Navaneelan, T. Trends in the incidence and mortality of female reproductive
system cancers. Statistics Canada, catalogue no. 82–624–X (2015).
2.
Canadian Cancer Society. Canadian Cancer Statistics Special topic : Predictions
of the future burden of cancer in Canada. 1–151 (2015).
3.
Bast, R. C., Hennessy, B. & Mills, G. B. The biology of ovarian cancer: new
opportunities for translation. Nat. Rev. Cancer 9, 415–28 (2009).
4.
Boyd, J. & Rubin, S. C. Hereditary ovarian cancer: molecular genetics and clinical
implications. Gynecol. Oncol. 64, 196–206 (1997).
5.
Kurman, R. J., Carcangiu, M. & Herrington, C. World Health Organisation
Classification of Tumours of the Female Reproductive Organs. International
Agency for Research on Cancer 4th ed., (2014).
6.
Devouassoux-Shisheboran, M. & Genestie, C. Pathobiology of ovarian
carcinomas. Chin. J. Cancer 34, 50–5 (2015).
7.
Karst, A. M. & Drapkin, R. Ovarian cancer pathogenesis: a model in evolution. J.
Oncol. 2010, 932371 (2010).
8.
Kauff, N. D. et al. Risk-reducing salpingo-oophorectomy in women with a BRCA1
or BRCA2 mutation. N. Engl. J. Med. 346, 1609–15 (2002).
9.
Landen, C. N., Birrer, M. J. & Sood, A. K. Early events in the pathogenesis of
epithelial ovarian cancer. J. Clin. Oncol. 26, 995–1005 (2008).
10.
Greenaway, J., Moorehead, R., Shaw, P. & Petrik, J. Epithelial-stromal interaction
increases cell proliferation, survival and tumorigenicity in a mouse model of
human epithelial ovarian cancer. Gynecol. Oncol. 108, 385–94 (2008).
11.
Risch, H. A. Hormonal etiology of epithelial ovarian cancer, with a hypothesis
concerning the role of androgens and progesterone. J. Natl. Cancer Inst. 90,
1774–86 (1998).
12.
Choi, J.-H., Wong, A. S. T., Huang, H.-F. & Leung, P. C. K. Gonadotropins and
ovarian cancer. Endocr. Rev. 28, 440–61 (2007).
13.
Ness, R. B. & Cottreau, C. Possible role of ovarian epithelial inflammation in
ovarian cancer. J. Natl. Cancer Inst. 91, 1459–67 (1999).
14.
Auersperg, N., Wong, a S., Choi, K. C., Kang, S. K. & Leung, P. C. Ovarian
surface epithelium: biology, endocrinology, and pathology. Endocr. Rev. 22, 255–
88 (2001).
161
15.
Holschneider, C. H. & Berek, J. S. Ovarian cancer: epidemiology, biology, and
prognostic factors. Semin. Surg. Oncol. 19, 3–10 (2000).
16.
Dubeau, L. & Drapkin, R. Coming into focus: the nonovarian origins of ovarian
cancer. Ann. Oncol. 24, viii28–viii35 (2013).
17.
Kurman, R. J. & Shih, I.-M. Pathogenesis of ovarian cancer: lessons from
morphology and molecular biology and their clinical implications. Int. J. Gynecol.
Pathol. 27, 151–60 (2008).
18.
Piek, J. M. et al. Dysplastic changes in prophylactically removed Fallopian tubes
of women predisposed to developing ovarian cancer. J. Pathol. 195, 451–6
(2001).
19.
Medeiros, F. et al. The tubal fimbria is a preferred site for early adenocarcinoma
in women with familial ovarian cancer syndrome. Am. J. Surg. Pathol. 30, 230–6
(2006).
20.
Kindelberger, D. W. et al. Intraepithelial Carcinoma of the Fimbria and Pelvic
Serous Carcinoma: Evidence for a Causal Relationship. Am. J. Surg. Pathol. 31,
161–169 (2007).
21.
Dubeau, L. The cell of origin of ovarian epithelial tumours. Lancet Oncol. 9, 1191–
1197 (2008).
22.
Kim, J. et al. High-grade serous ovarian cancer arises from fallopian tube in a
mouse model. Proceedings of the National Academy of Sciences 109, 3921–3926
(2012).
23.
Bankhead, C. R. et al. Identifying symptoms of ovarian cancer: a qualitative and
quantitative study. BJOG 115, 1008–14 (2008).
24.
Goff, B. A., Mandel, L. S., Melancon, C. H. & Muntz, H. G. Frequency of
symptoms of ovarian cancer in women presenting to primary care clinics. JAMA
291, 2705–12 (2004).
25.
National Cancer Institute. Surveillance Epidemiology and End Results (SEER).
(2012).
26.
Prat, J. FIGO Committee on Gynecologic Oncology. Staging classification for
cancer of the ovary, fallopian tube, and peritoneum. Int. J. Gynecol. Obstet. 124,
1–5 (2014).
27.
Hynninen, J. et al. FDG PET/CT in staging of advanced epithelial ovarian cancer:
Frequency of supradiaphragmatic lymph node metastasis challenges the
traditional pattern of disease spread. Gynecol. Oncol. 126, 64–68 (2012).
28.
Ahmed, N. & Stenvers, K. L. Getting to know ovarian cancer ascites: opportunities
162
for targeted therapy-based translational research. Front. Oncol. 3, 256 (2013).
29.
Jayson, G. C., Kohn, E. C., Kitchener, H. C. & Ledermann, J. A. Ovarian cancer.
Lancet 384, 1376–1388 (2014).
30.
du Bois, A. et al. Role of surgical outcome as prognostic factor in advanced
epithelial ovarian cancer: a combined exploratory analysis of 3 prospectively
randomized phase 3 multicenter trials: by the Arbeitsgemeinschaft
Gynaekologische Onkologie Studiengruppe Ovarialkarzin. Cancer 115, 1234–44
(2009).
31.
Kurman, R. J., Visvanathan, K., Roden, R., Wu, T. C. & Shih, I.-M. Early detection
and treatment of ovarian cancer: shifting from early stage to minimal volume of
disease based on a new model of carcinogenesis. Am. J. Obstet. Gynecol. 198,
351–6 (2008).
32.
King, M.-C., Marks, J. H. & Mandell, J. B. Breast and ovarian cancer risks due to
inherited mutations in BRCA1 and BRCA2. Science 302, 643–6 (2003).
33.
Kyriazi, S., Kaye, S. B. & deSouza, N. M. Imaging ovarian cancer and peritoneal
metastases--current and emerging techniques. Nat. Rev. Clin. Oncol. 7, 381–93
(2010).
34.
Gordon, A. N. et al. Recurrent epithelial ovarian carcinoma: a randomized phase
III study of pegylated liposomal doxorubicin versus topotecan. J. Clin. Oncol. 19,
3312–22 (2001).
35.
Griffiths, R. W. et al. Outcomes After Multiple Lines of Chemotherapy for
Platinum-Resistant Epithelial Cancers of the Ovary, Peritoneum, and Fallopian
Tube. Int. J. Gynecol. Cancer 21, 58–65 (2011).
36.
Vaughan, S. et al. Rethinking ovarian cancer: recommendations for improving
outcomes. Nat. Rev. Cancer 11, 719–25 (2011).
37.
Guyton, A. C. & Hall, J. E. Guyton and Hall Textbook of Medical Physiology.
(Saunders Elsevier, 2011).
38.
Voet, D. & Voet, J. G. Biochemistry. (John Wiley & Sons, Inc, 2004).
39.
Warburg, O., K, P. & E, N. Ueber den stoffwechsel der tumoren. Biochem Z 152,
319–344 (1924).
40.
Ganapathy, V., Thangaraju, M. & Prasad, P. D. Nutrient transporters in cancer:
Relevance to Warburg hypothesis and beyond. Pharmacol. Ther. 121, 29–40
(2009).
41.
Warburg, O. On the Origin of Cancer Cells. Science (80-. ). 123, 309–314 (1956).
163
42.
Funes, J. M. et al. Transformation of human mesenchymal stem cells increases
their dependency on oxidative phosphorylation for energy production. Proc. Natl.
Acad. Sci. U. S. A. 104, 6223–8 (2007).
43.
Hanahan, D. & Weinberg, R. A. Hallmarks of cancer: the next generation. Cell
144, 646–74 (2011).
44.
Kim, J. & Dang, C. V. Cancer’s molecular sweet tooth and the Warburg effect.
Cancer Res. 66, 8927–30 (2006).
45.
Chance, B. & Hess, B. On the control of metabolism in ascites tumor cell
suspensions. Ann. N. Y. Acad. Sci. 63, 1008–16 (1956).
46.
Chance, B. & Castor, L. N. Some Patterns of the Respiratory Pigments of Ascites
Tumors of Mice. Science 116, 200–2 (1952).
47.
Weinhouse, S. On respiratory impairment in cancer cells. Science 124, 267–9
(1956).
48.
Greiner, E. F., Guppy, M. & Brand, K. Glucose is essential for proliferation and the
glycolytic enzyme induction that provokes a transition to glycolytic energy
production. J. Biol. Chem. 269, 31484–90 (1994).
49.
Koppenol, W. H., Bounds, P. L. & Dang, C. V. Otto Warburg’s contributions to
current concepts of cancer metabolism. Nat. Rev. Cancer 11, 325–337 (2011).
50.
Kroemer, G. & Pouyssegur, J. Tumor cell metabolism: cancer’s Achilles' heel.
Cancer Cell 13, 472–82 (2008).
51.
Cairns, R. A., Harris, I. S. & Mak, T. W. Regulation of cancer cell metabolism.
Nature reviews. Cancer 11, 85–95 (2011).
52.
Pedersen, P. L. The cancer cell’s ‘power plants’ as promising therapeutic targets:
an overview. J. Bioenerg. Biomembr. 39, 1–12 (2007).
53.
Modica-Napolitano, J. S., Kulawiec, M. & Singh, K. K. Mitochondria and human
cancer. Curr. Mol. Med. 7, 121–31 (2007).
54.
Eboli, M. L., Paradies, G., Galeotti, T. & Papa, S. Pyruvate transport in tumourcell mitochondria. Biochim. Biophys. Acta 460, 183–7 (1977).
55.
Pelicano, H. et al. Mitochondrial respiration defects in cancer cells cause
activation of Akt survival pathway through a redox-mediated mechanism. J. Cell
Biol. 175, 913–23 (2006).
56.
Bellance, N. et al. Bioenergetics of lung tumors: Alteration of mitochondrial
biogenesis and respiratory capacity. Int. J. Biochem. Cell Biol. 41, 2566–2577
(2009).
164
57.
Desouki, M. M., Kulawiec, M., Bansal, S., Das, G. M. & Singh, K. K. Cross talk
between mitochondria and superoxide generating NADPH oxidase in breast and
ovarian tumors. Cancer Biol. Ther. 4, 1367–73 (2005).
58.
Pollard, P. J. Accumulation of Krebs cycle intermediates and over-expression of
HIF1 in tumours which result from germline FH and SDH mutations. Hum. Mol.
Genet. 14, 2231–2239 (2005).
59.
Dang, L. et al. Cancer-associated IDH1 mutations produce 2-hydroxyglutarate.
Nature 465, 966 (2010).
60.
Gottlieb, E. & Tomlinson, I. P. M. Mitochondrial tumour suppressors: a genetic
and biochemical update. Nat. Rev. Cancer 5, 857–66 (2005).
61.
Lin, C.-C. et al. Loss of the respiratory enzyme citrate synthase directly links the
Warburg effect to tumor malignancy. Sci. Rep. 2, 785 (2012).
62.
Schulz, T. J. et al. Induction of oxidative metabolism by mitochondrial frataxin
inhibits cancer growth: Otto Warburg revisited. J. Biol. Chem. 281, 977–81 (2006).
63.
Gatenby, R. a & Gillies, R. J. Why do cancers have high aerobic glycolysis? Nat.
Rev. Cancer 4, 891–9 (2004).
64.
Zu, X. L. & Guppy, M. Cancer metabolism: facts, fantasy, and fiction☆. Biochem.
Biophys. Res. Commun. 313, 459–465 (2004).
65.
Hockel, M. & Vaupel, P. Tumor Hypoxia: Definitions and Current Clinical, Biologic,
and Molecular Aspects. JNCI J. Natl. Cancer Inst. 93, 266–276 (2001).
66.
Semenza, G. L. Targeting HIF-1 for cancer therapy. Nat. Rev. Cancer 3, 721–732
(2003).
67.
Goodwin, M. L., Gladden, L. B., Nijsten, M. W. N. & Jones, K. B. Lactate and
Cancer: Revisiting the Warburg Effect in an Era of Lactate Shuttling. Front. Nutr.
1, 2014–2016 (2015).
68.
Walenta, S. & Mueller-Klieser, W. F. Lactate: mirror and motor of tumor
malignancy. Semin. Radiat. Oncol. 14, 267–74 (2004).
69.
Choi, S. Y. C., Collins, C. C., Gout, P. W. & Wang, Y. Cancer-generated lactic
acid: a regulatory, immunosuppressive metabolite? J. Pathol. 230, 350–5 (2013).
70.
Morita, T., Nagaki, T., Fukuda, I. & Okumura, K. Clastogenicity of low pH to
various cultured mammalian cells. Mutat. Res. 268, 297–305 (1992).
71.
Yuan, J., Narayanan, L., Rockwell, S. & Glazer, P. M. Diminished DNA repair and
elevated mutagenesis in mammalian cells exposed to hypoxia and low pH.
Cancer Res. 60, 4372–6 (2000).
165
72.
Raghunand, N., Mahoney, B., van Sluis, R., Baggett, B. & Gillies, R. J. Acute
metabolic alkalosis enhances response of C3H mouse mammary tumors to the
weak base mitoxantrone. Neoplasia 3, 227–35
73.
Martínez-Zaguilán, R. et al. Acidic pH enhances the invasive behavior of human
melanoma cells. Clin. Exp. Metastasis 14, 176–86 (1996).
74.
Gatenby, R. A. & Gawlinski, E. T. The glycolytic phenotype in carcinogenesis and
tumor invasion: insights through mathematical models. Cancer Res. 63, 3847–54
(2003).
75.
Gatenby, R. A. & Gawlinski, E. T. A reaction-diffusion model of cancer invasion.
Cancer Res. 56, 5745–53 (1996).
76.
Smallbone, K., Gavaghan, D. J., Gatenby, R. A. & Maini, P. K. The role of acidity
in solid tumour growth and invasion. J. Theor. Biol. 235, 476–84 (2005).
77.
Xu, H. N. et al. Is Higher Lactate an Indicator of Tumor Metastatic Risk? A Pilot
MRS Study Using Hyperpolarized (13)C-Pyruvate. Acad. Radiol. 21, 223–31
(2014).
78.
Arora, K. K. & Pedersen, P. L. Functional significance of mitochondrial bound
hexokinase in tumor cell metabolism. Evidence for preferential phosphorylation of
glucose by intramitochondrially generated ATP. J. Biol. Chem. 263, 17422–8
(1988).
79.
Rastogi, S., Banerjee, S., Chellappan, S. & Simon, G. R. Glut-1 antibodies induce
growth arrest and apoptosis in human cancer cell lines. Cancer Lett. 257, 244–
251 (2007).
80.
Friedrich, M. J. Researchers aim to stop tumor growth by shutting off cancer’s fuel
supply. JAMA 303, 1021–2 (2010).
81.
Deberardinis, R. J., Sayed, N., Ditsworth, D. & Thompson, C. B. Brick by brick:
metabolism and tumor cell growth. Curr. Opin. Genet. Dev. 18, 54–61 (2008).
82.
Jang, M., Kim, S. S. & Lee, J. Cancer cell metabolism: implications for therapeutic
targets. Exp. Mol. Med. 45, e45 (2013).
83.
Ward, P. S. & Thompson, C. B. Metabolic reprogramming: a cancer hallmark
even warburg did not anticipate. Cancer Cell 21, 297–308 (2012).
84.
Venuta, S. & Rubin, H. Effects of glucose starvation on normal and rous sarcoma
virus-transformed chick cells. J. Natl. Cancer Inst. 54, 395–400 (1975).
85.
Hatanaka, M., Huebner, R. J. & Gilden, R. V. Alterations in the characteristics of
sugar uptake by mouse cells transformed by murine sarcoma viruses. J. Natl.
Cancer Inst. 43, 1091–6 (1969).
166
86.
Ramanathan, A., Wang, C. & Schreiber, S. L. Perturbational profiling of a cell-line
model of tumorigenesis by using metabolic measurements. Proc. Natl. Acad. Sci.
U. S. A. 102, 5992–7 (2005).
87.
Jiang, P. et al. p53 regulates biosynthesis through direct inactivation of glucose-6phosphate dehydrogenase. Nat. Cell Biol. 13, 310–6 (2011).
88.
Goldstein, I. et al. P53 Promotes the Expression of Gluconeogenesis-Related
Genes and Enhances Hepatic Glucose Production. Cancer Metab. 1, 9 (2013).
89.
Elstrom, R. L. et al. Akt stimulates aerobic glycolysis in cancer cells. Cancer Res.
64, 3892–9 (2004).
90.
Robey, R. B. & Hay, N. Is Akt the ‘Warburg kinase’?-Akt-energy metabolism
interactions and oncogenesis. Semin. Cancer Biol. 19, 25–31 (2009).
91.
Christofk, H. R. et al. The M2 splice isoform of pyruvate kinase is important for
cancer metabolism and tumour growth. Nature 452, 230–3 (2008).
92.
Shanmugam, M., McBrayer, S. K. & Rosen, S. T. Targeting the Warburg effect in
hematological malignancies: from PET to therapy. Curr. Opin. Oncol. 21, 531–6
(2009).
93.
Schulz, J. et al. Regulation of anaerobic glycolysis in Ehrlich ascites tumour cells.
Acta Biol. Med. Ger. 36, 1379–91 (1977).
94.
Mathupala, S. P., Rempel, A. & Pedersen, P. L. Aberrant glycolytic metabolism of
cancer cells: a remarkable coordination of genetic, transcriptional, posttranslational, and mutational events that lead to a critical role for type II
hexokinase. J. Bioenerg. Biomembr. 29, 339–43 (1997).
95.
Rivenzon-Segal, D., Boldin-Adamsky, S., Seger, D., Seger, R. & Degani, H.
Glycolysis and glucose transporter 1 as markers of response to hormonal therapy
in breast cancer. Int. J. Cancer 107, 177–82 (2003).
96.
Artemov, D., Bhujwalla, Z. M., Pilatus, U. & Glickson, J. D. Two-compartment
model for determination of glycolytic rates of solid tumors by in vivo 13C NMR
spectroscopy. NMR Biomed. 11, 395–404 (1998).
97.
Wolf, A. et al. Hexokinase 2 is a key mediator of aerobic glycolysis and promotes
tumor growth in human glioblastoma multiforme. J. Exp. Med. 208, 313–26
(2011).
98.
Mathupala, S. P., Ko, Y. H. & Pedersen, P. L. Hexokinase-2 bound to
mitochondria: cancer’s stygian link to the ‘Warburg Effect’ and a pivotal target for
effective therapy. Semin. Cancer Biol. 19, 17–24 (2009).
99.
Pedersen, P. L. Warburg, me and Hexokinase 2: Multiple discoveries of key
167
molecular events underlying one of cancers’ most common phenotypes, the
‘Warburg Effect’, i.e., elevated glycolysis in the presence of oxygen. J. Bioenerg.
Biomembr. 39, 211–22 (2007).
100. John, S., Weiss, J. N. & Ribalet, B. Subcellular localization of hexokinases I and II
directs the metabolic fate of glucose. PLoS One 6, e17674 (2011).
101. Netti, P. A., Baxter, L. T., Boucher, Y., Skalak, R. & Jain, R. K. Time-dependent
behavior of interstitial fluid pressure in solid tumors: implications for drug delivery.
Cancer Res. 55, 5451–8 (1995).
102. Boucher, Y. & Jain, R. K. Microvascular pressure is the principal driving force for
interstitial hypertension in solid tumors: implications for vascular collapse. Cancer
Res. 52, 5110–4 (1992).
103. Rutz, H. P. A biophysical basis of enhanced interstitial fluid pressure in tumors.
Med. Hypotheses 53, 526–529 (1999).
104. Samudio, I. et al. Pharmacologic inhibition of fatty acid oxidation sensitizes human
leukemia cells to apoptosis induction. J. Clin. Invest. 120, 142–56 (2010).
105. Bonuccelli, G. et al. Ketones and lactate ‘fuel’ tumor growth and metastasis:
Evidence that epithelial cancer cells use oxidative mitochondrial metabolism. Cell
Cycle 9, 3506–3514 (2010).
106. Wise, D. R. & Thompson, C. B. Glutamine addiction: a new therapeutic target in
cancer. Trends Biochem. Sci. 35, 427–33 (2010).
107. Buzzai, M. et al. The glucose dependence of Akt-transformed cells can be
reversed by pharmacologic activation of fatty acid beta-oxidation. Oncogene 24,
4165–73 (2005).
108. Rodríguez-Enríquez, S. et al. Oxidative phosphorylation is impaired by prolonged
hypoxia in breast and possibly in cervix carcinoma. Int. J. Biochem. Cell Biol. 42,
1744–1751 (2010).
109. Warburg, O., Wind, F. & Negelein, E. THE METABOLISM OF TUMORS IN THE
BODY. J. Gen. Physiol. 8, 519–30 (1927).
110. Jose, C., Bellance, N. & Rossignol, R. Choosing between glycolysis and oxidative
phosphorylation: A tumor’s dilemma? Biochim. Biophys. Acta 1807, 552–61
(2011).
111. Anderson, A. S., Roberts, P. C., Frisard, M. I., Hulver, M. W. & Schmelz, E. M.
Ovarian tumor-initiating cells display a flexible metabolism. Exp. Cell Res. 328,
44–57 (2014).
112. Griguer, C. E., Oliva, C. R. & Gillespie, G. Y. Glucose metabolism heterogeneity
168
in human and mouse malignant glioma cell lines. J. Neurooncol. 74, 123–33
(2005).
113. Berridge, M. V, Herst, P. M. & Tan, A. S. Metabolic flexibility and cell hierarchy in
metastatic cancer. Mitochondrion 10, 584–8 (2010).
114. Bissell, M. J. & Radisky, D. Putting tumours in context. Nat. Rev. Cancer 1, 46–54
(2001).
115. Yoshida, G. J. Metabolic reprogramming: the emerging concept and associated
therapeutic strategies. J. Exp. Clin. Cancer Res. 34, 111 (2015).
116. Pavlides, S. et al. The reverse Warburg effect: Aerobic glycolysis in cancer
associated fibroblasts and the tumor stroma. Cell Cycle 8, 3984–4001 (2009).
117. Martinez-Outschoorn, U. E., Lisanti, M. P. & Sotgia, F. Catabolic CancerAssociated Fibroblasts (CAFs) Transfer Energy and Biomass to Anabolic Cancer
Cells, Fueling Tumor Growth. Semin. Cancer Biol. (2014).
118. Fiaschi, T. et al. Reciprocal metabolic reprogramming through lactate shuttle
coordinately influences tumor-stroma interplay. Cancer Res. 72, 5130–40 (2012).
119. Sonveaux, P. et al. Targeting lactate-fueled respiration selectively kills hypoxic
tumor cells in mice. J. Clin. Invest. 118, 3930–42 (2008).
120. Semenza, G. L. Tumor metabolism: cancer cells give and take lactate. J. Clin.
Invest. 118, 3835–7 (2008).
121. Crabtree, H. G. Observations on the carbohydrate metabolism of tumours.
Biochem. J. 23, 536–45 (1929).
122. Suchorolski, M. T., Paulson, T. G., Sanchez, C. A., Hockenbery, D. & Reid, B. J.
Warburg and Crabtree effects in premalignant Barrett’s esophagus cell lines with
active mitochondria. PLoS One 8, e56884 (2013).
123. Bouché, C., Serdy, S., Kahn, C. R. & Goldfine, A. B. The cellular fate of glucose
and its relevance in type 2 diabetes. Endocr. Rev. 25, 807–830 (2004).
124. Szablewski, L. Expression of glucose transporters in cancers. Biochim. Biophys.
Acta 1835, 164–9 (2013).
125. Yamamoto, T. et al. Over-expression of facilitative glucose transporter genes in
human cancer. Biochem. Biophys. Res. Commun. 170, 223–30 (1990).
126. Rudlowski, C. et al. GLUT1 mRNA and protein expression in ovarian borderline
tumors and cancer. Oncology 66, 404–410 (2004).
127. Semaan, A. et al. Expression of GLUT-1 in epithelial ovarian carcinoma:
169
correlation with tumor cell proliferation, angiogenesis, survival and ability to
predict optimal cytoreduction. Gynecol. Oncol. 121, 181–6 (2011).
128. Kalir, T. et al. Immunohistochemical staining of GLUT1 in benign, borderline, and
malignant ovarian epithelia. Cancer 94, 1078–1082 (2002).
129. Noguchi, Y. et al. Suppression of facilitative glucose transporter 1 mRNA can
suppress tumor growth. Cancer Lett. 154, 175–82 (2000).
130. Wright, E. M., Loo, D. D. F. & Hirayama, B. A. Biology of human sodium glucose
transporters. Physiol. Rev. 91, 733–794 (2011).
131. Wright, E. M., Loo, D. D. F., Hirayama, B. a & Turk, E. Surprising versatility of
Na+-glucose cotransporters: SLC5. Physiology (Bethesda). 19, 370–376 (2004).
132. Narendran, P. & Saeed, M. Dapagliflozin for the treatment of type 2 diabetes: a
review of the literature. Drug Des. Devel. Ther. 8, 2493 (2014).
133. Tahara, A., Takasu, T., Yokono, M., Imamura, M. & Kurosaki, E. Effects of the
combination of SGLT2 selective inhibitor ipragliflozin and various antidiabetic
drugs in type 2 diabetic mice. Arch. Pharm. Res. (2015). doi:10.1007/s12272-0150621-8
134. Yu, A. S. et al. Regional distribution of SGLT activity in rat brain in vivo. Am. J.
Physiol. Cell Physiol. 304, C240–7 (2013).
135. Yu, L. C. H., Flynn, A. N., Turner, J. R. & Buret, A. G. SGLT-1-mediated glucose
uptake protects intestinal epithelial cells against LPS-induced apoptosis and
barrier defects: a novel cellular rescue mechanism? FASEB J. 19, 1822–35
(2005).
136. Banerjee, S. K., McGaffin, K. R., Pastor-Soler, N. M. & Ahmad, F. SGLT1 is a
novel cardiac glucose transporter that is perturbed in disease states. Cardiovasc.
Res. 84, 111–118 (2009).
137. Barcelona, S., Menegaz, D. & Diez-Sampedro, A. Mouse SGLT3a generates
proton-activated currents but does not transport sugar. AJP Cell Physiol. 302,
C1073–C1082 (2012).
138. Balon, T. W. SGLT and GLUT: are they teammates? Focus on ‘Mouse SGLT3a
generates proton-activated currents but does not transport sugar’. Am. J. Physiol.
Cell Physiol. 302, C1071–2 (2012).
139. Diez-Sampedro, A. et al. A glucose sensor hiding in a family of transporters. Proc.
Natl. Acad. Sci. 100, 11753–11758 (2003).
140. Bianchi, L. & Díez-Sampedro, A. A single amino acid change converts the sugar
sensor SGLT3 into a sugar transporter. PLoS One 5, (2010).
170
141. Wright, E. M., Hirsch, J. R., Loo, D. D. & Zampighi, G. a. Regulation of
Na+/glucose cotransporters. J. Exp. Biol. 200, 287–93 (1997).
142. Sakar, Y. et al. Metformin-induced regulation of the intestinal d-glucose
transporters. J. Physiol. Pharmacol. 61, 301–307 (2010).
143. Sopjani, M. et al. Regulation of Na+-coupled glucose carrier SGLT1 by AMPactivated protein kinase. Mol. Membr. Biol. 27, 137–44 (2010).
144. Yamazaki, Y., Ogihara, S., Harada, S. & Tokuyama, S. Activation of cerebral
sodium-glucose transporter type 1 function mediated by post-ischemic
hyperglycemia exacerbates the development of cerebral ischemia. Neuroscience
310, 674–85 (2015).
145. Weihua, Z. et al. Survival of cancer cells is maintained by EGFR independent of
its kinase activity. Cancer Cell 13, 385–93 (2008).
146. Wright, E. M., Hirayama, B. A. & Loo, D. F. Active sugar transport in health and
disease. J. Intern. Med. 261, 32–43 (2007).
147. Kothinti, R. K., Blodgett, A. B., North, P. E., Roman, R. J. & Tabatabai, N. M. A
novel SGLT is expressed in the human kidney. Eur. J. Pharmacol. 690, 77–83
(2012).
148. Sasseville, L. J., Cuervo, J. E., Lapointe, J.-Y. & Noskov, S. Y. The structural
pathway for water permeation through sodium-glucose cotransporters. Biophys. J.
101, 1887–95 (2011).
149. Panayotova-Heiermann, M. & Wright, E. M. Mapping the urea channel through
the rabbit Na(+)-glucose cotransporter SGLT1. J. Physiol. 535, 419–25 (2001).
150. Hirschhorn, N. & Greenough, W. B. Progress in oral rehydration therapy. Sci. Am.
264, 50–6 (1991).
151. Yang, M. & Brackenbury, W. J. Membrane potential and cancer progression.
Front. Physiol. 4, 185 (2013).
152. Blodgett, T. M., Meltzer, C. C. & Townsend, D. W. PET/CT: form and function.
Radiology 242, 360–85 (2007).
153. Pimlott, S. L. & Sutherland, A. Molecular tracers for the PET and SPECT imaging
of disease. Chem. Soc. Rev. 40, 149–162 (2011).
154. Kitajima, K., Ebina, Y. & Sugimura, K. Present and future role of FDG-PET/CT
imaging in the management of gynecologic malignancies. Jpn J Radiol 32, 313–
323 (2014).
155. Gambhir, S. S. Molecular imaging of cancer with positron emission tomography.
171
Nat. Rev. Cancer 2, 683–93 (2002).
156. Kubota, R. et al. Intratumoral distribution of fluorine-18-fluorodeoxyglucose in
vivo: high accumulation in macrophages and granulation tissues studied by
microautoradiography. J. Nucl. Med. 33, 1972–80 (1992).
157. Som, P. et al. A fluorinated glucose analog, 2-fluoro-2-deoxy-D-glucose (F-18):
nontoxic tracer for rapid tumor detection. J. Nucl. Med. 21, 670–675 (1980).
158. Smith, T. a. The rate-limiting step for tumor [18F]fluoro-2-deoxy-D-glucose (FDG)
incorporation. Nucl. Med. Biol. 28, 1–4 (2001).
159. Prakash, P., Cronin, C. G. & Blake, M. A. Role of PET/CT in Ovarian Cancer.
American Roentgen Ray Society W464–W470 (2009). at
<http://www.ajronline.org/doi/pdf/10.2214/AJR.09.3843>
160. Steenkamp, D. W., McDonnell, M. E. & Meibom, S. Metformin may be associated
with false-negative cancer detection in the gastrointestinal tract on PET/CT.
Endocr. Pract. 20, 1079–83 (2014).
161. Zhu, A., Lee, D. & Shim, H. Metabolic positron emission tomography imaging in
cancer detection and therapy response. Semin. Oncol. 38, 55–69 (2011).
162. Ryu, J. S., Choi, N. C., Fischman, A. J., Lynch, T. J. & Mathisen, D. J. FDG-PET
in staging and restaging non-small cell lung cancer after neoadjuvant
chemoradiotherapy: correlation with histopathology. Lung Cancer 35, 179–87
(2002).
163. Alavi, A. et al. Positron emission tomography imaging in nonmalignant thoracic
disorders. Semin. Nucl. Med. 32, 293–321 (2002).
164. Cheng, G. et al. Dynamic changes of FDG uptake and clearance in normal
tissues. Mol. Imaging Biol. 15, 345–52 (2013).
165. Kostakoglu, L., Agress, H. & Goldsmith, S. J. Clinical role of FDG PET in
evaluation of cancer patients. Radiographics 23, 315–40; quiz 533 (2003).
166. Gaeta, C. M., Sher, A. C., Kohan, A., Rubbert, C. & Avril, N. Recurrent and
metastatic breast cancer PET, PET/CT, PET/MRI: FDG and new biomarkers. Q J
Nucl Med Mol Imaging (2013).
167. Chang, J. M. et al. False positive and false negative FDG-PET scans in various
thoracic diseases. Korean J. Radiol. 7, 57–69
168. Lutz, a M., Ray, P., Willmann, J. K., Drescher, C. & Gambhir, S. S. 2-deoxy-2-[F18]fluoro-D-glucose accumulation in ovarian carcinoma cell lines. Mol. Imaging
Biol. 9, 260–6 (2007).
172
169. Rohren, E. M., Turkington, T. G. & Coleman, R. E. Clinical applications of PET in
oncology. Radiology 231, 305–32 (2004).
170. Schwarz, J. K., Grigsby, P. W., Dehdashti, F. & Delbeke, D. The role of 18F-FDG
PET in assessing therapy response in cancer of the cervix and ovaries. J. Nucl.
Med. 50 Suppl 1, 64S–73S (2009).
171. Prefontaine, M. & Walker-Dilks, C. PET Recommendation Report 7 IN REVIEW
PET Imaging in Ovarian Cancer PET Imaging in Ovarian Cancer :
Recommendations. (2009).
172. Tanizaki, Y. et al. Diagnostic Value of Preoperative SUVmax on FDG-PET/CT for
the Detection of Ovarian Cancer. Int. J. Gynecol. Cancer (2014).
173. Son, H. et al. Role of FDG PET/CT in staging of recurrent ovarian cancer.
Radiographics 31, 569–83 (2011).
174. Gouhar, G. K., Siam, S., Sadek, S. M. & Ahmed, R. A. Prospective assessment of
18F-FDG PET/CT in detection of recurrent ovarian cancer. Egypt. J. Radiol. Nucl.
Med. 44, 913–922 (2013).
175. Ghosh, J. et al. Role of FDG PET-CT in asymptomatic epithelial ovarian cancer
with rising serum CA-125: a pilot study. Natl. Med. J. India 26, 327–31 (2013).
176. Berek, J. S., Crum, C. & Friedlander, M. Cancer of the ovary, fallopian tube, and
peritoneum. Int. J. Gynecol. Obstet. 119, S118–S129 (2012).
177. Psarakis, H. M. Clinical Challenges in Caring for Patients With Diabetes and
Cancer. Diabetes Spectr. 19, 157–162 (2006).
178. Lee, J.-Y. et al. Diabetes mellitus and ovarian cancer risk: a systematic review
and meta-analysis of observational studies. Int. J. Gynecol. Cancer 23, 402–12
(2013).
179. Bakhru, A., Buckanovich, R. J. & Griggs, J. J. The impact of diabetes on survival
in women with ovarian cancer. Gynecol. Oncol. 121, 106–11 (2011).
180. Shah, M. M. et al. Diabetes mellitus and ovarian cancer: More complex than just
increasing risk. Gynecol. Oncol. 135, 273–277 (2014).
181. Gallagher, E. J. & LeRoith, D. Diabetes, antihyperglycemic medications and
cancer risk: smoke or fire? Curr. Opin. Endocrinol. Diabetes. Obes. 20, 485–94
(2013).
182. Lamkin, D. M. et al. Glucose as a prognostic factor in ovarian carcinoma. Cancer
115, 1021–7 (2009).
183. Leitzmann, M. F. et al. Body mass index and risk of ovarian cancer. Cancer 115,
173
812–22 (2009).
184. Grote, V. A., Becker, S. & Kaaks, R. Diabetes mellitus type 2 - an independent
risk factor for cancer? Exp. Clin. Endocrinol. diabetes Off. J. Ger. Soc. Endocrinol.
Ger. Diabetes Assoc. 118, 4–8 (2010).
185. Giovannucci, E. et al. Diabetes and cancer: a consensus report. in Diabetes care
33, 1674–1685 (2010).
186. Kellenberger, L. D. et al. The role of dysregulated glucose metabolism in epithelial
ovarian cancer. J. Oncol. 2010, 514310 (2010).
187. Zhang, X. D. et al. Effect of 2-deoxy-D-glucose on various malignant cell lines in
vitro. Anticancer Res. 26, 3561–6
188. Tayek, J. A. A review of cancer cachexia and abnormal glucose metabolism in
humans with cancer. J. Am. Coll. Nutr. 11, 445–56 (1992).
189. Rodríguez-Enríquez, S., Marín-Hernández, A., Gallardo-Pérez, J. C. & MorenoSánchez, R. Kinetics of transport and phosphorylation of glucose in cancer cells.
J. Cell. Physiol. 221, 552–559 (2009).
190. Younes, M., Lechago, L. V, Somoano, J. R., Mosharaf, M. & Lechago, J. Wide
expression of the human erythrocyte glucose transporter Glut1 in human cancers.
Cancer Res. 56, 1164–7 (1996).
191. Van Belle, T. L. L., Taylor, P. & von Herrath, M. G. G. Mouse Models for Type 1
Diabetes. Drug Discov. Today Dis. Model. 6, 41–45 (2009).
192. King, A. J. F. The use of animal models in diabetes research. Br. J. Pharmacol.
166, 877–894 (2012).
193. Dummler, B. et al. Life with a single isoform of Akt: mice lacking Akt2 and Akt3
are viable but display impaired glucose homeostasis and growth deficiencies. Mol.
Cell. Biol. 26, 8042–51 (2006).
194. Garofalo, R. S. et al. Severe diabetes, age-dependent loss of adipose tissue, and
mild growth deficiency in mice lacking Akt2/PKB beta. J. Clin. Invest. 112, 197–
208 (2003).
195. Vuguin, P., Saenger, P. & Dimartino-Nardi, J. Fasting Glucose Insulin Ratio: A
Useful Measure of Insulin Resistance in Girls with Premature Adrenarche. J. Clin.
Endocrinol. Metab. 86, 4618–4621 (2001).
196. Berridge, M. V, Tan, A. N. S., Mccoy, K. D. & Wang, R. U. I. The Biochemical and
Cellular Basis of Cell Proliferation Assays That Use Tetrazolium Salts. 4–9
(1996).
174
197. Bower, W. F. et al. Overt diabetes mellitus adversely affects surgical outcomes of
noncardiovascular patients. Surgery 147, 670–5 (2010).
198. White, N. H. Long-term Outcomes in Youths with Diabetes Mellitus. Pediatr. Clin.
North Am. 62, 889–909 (2015).
199. Zelenko, Z. & Gallagher, E. J. Diabetes and cancer. Endocrinology and
Metabolism Clinics of North America 43, 167–185 (2014).
200. American Diabetes Association. Standards of medical care in diabetes. Diabetes
Care 37, (Suppl 1): S14–S80 (2014).
201. Phoenix, K. N., Vumbaca, F., Fox, M. M., Evans, R. & Claffey, K. P. Dietary
energy availability affects primary and metastatic breast cancer and metformin
efficacy. Breast Cancer Res. Treat. 123, 333–44 (2010).
202. Yang, X.-L. & Chan, J. C. Diabetes, insulin and cancer risk. World J. Diabetes 3,
60–4 (2012).
203. Han, L. et al. High Glucose Promotes Pancreatic Cancer Cell Proliferation via the
Induction of EGF Expression and Transactivation of EGFR. PLoS One 6, e27074
(2011).
204. Parekh, N., Lin, Y., Hayes, R. B., Albu, J. B. & Lu-Yao, G. L. Longitudinal
associations of blood markers of insulin and glucose metabolism and cancer
mortality in the third National Health and Nutrition Examination Survey. Cancer
Causes Control 21, 631–42 (2010).
205. Plesner, A., Ten Holder, J. T. & Verchere, C. B. Islet remodeling in female mice
with spontaneous autoimmune and streptozotocin-induced diabetes. PLoS One 9,
(2014).
206. Karagiannis, G. S. et al. Cancer-associated fibroblasts drive the progression of
metastasis through both paracrine and mechanical pressure on cancer tissue.
Mol. Cancer Res. 10, 1403–18 (2012).
207. Lerman, O. Z., Galiano, R. D., Armour, M., Levine, J. P. & Gurtner, G. C. Cellular
dysfunction in the diabetic fibroblast: impairment in migration, vascular endothelial
growth factor production, and response to hypoxia. Am. J. Pathol. 162, 303–12
(2003).
208. Nerlich, A. G., Hagedorn, H. G., Böheim, M. & Schleicher, E. D. Patients with
diabetes-induced microangiopathy show a reduced frequency of carcinomas. In
Vivo 12, 667–70
209. Renehan, A. G. et al. Diabetes and cancer (2): Evaluating the impact of diabetes
on mortality In patients with cancer. Diabetologia 55, 1619–1632 (2012).
175
210. Kerr, A. D. The Role of Chronic , Systemic Inflammation in the Progression of
Epithelial Ovarian Cancer. (2012).
211. Al-Wahab, Z. et al. Metformin prevents aggressive ovarian cancer growth driven
by high-energy diet: similarity with calorie restriction. Oncotarget 6, 10908–10923
(2015).
212. Shapot, V. S. & Blinov, V. a. Blood glucose levels and gluconeogenesis in
animals bearing transplantable tumors. Cancer Res. 34, 1827–32 (1974).
213. Pavelić, K. et al. Growth and treatment of Ehrlich tumor in mice with alloxaninduced diabetes. Cancer Res. 39, 1807–13 (1979).
214. Gullino, P. M., Grantham, F. H. & Courtney, a H. Glucose consumption by
transplanted tumors in vivo. Cancer Res. 27, 1031–40 (1967).
215. de Groot, J. W. B. et al. Non-islet cell tumour-induced hypoglycaemia: a review of
the literature including two new cases. Endocr. Relat. Cancer 14, 979–93 (2007).
216. Wahdan-Alaswad, R. et al. Glucose promotes breast cancer aggression and
reduces metformin efficacy. Cell Cycle 12, 3759–69 (2013).
217. Diaz-Ruiz, R., Rigoulet, M. & Devin, A. The Warburg and Crabtree effects: On the
origin of cancer cell energy metabolism and of yeast glucose repression. Biochim.
Biophys. Acta 1807, 568–76 (2011).
218. Anderson, A. S. et al. Metabolic changes during ovarian cancer progression as
targets for sphingosine treatment. Exp. Cell Res. 319, 1431–42 (2013).
219. Kueck, A. et al. Resveratrol inhibits glucose metabolism in human ovarian cancer
cells. Gynecol. Oncol. 107, 450–7 (2007).
220. Harris, M. I., Klein, R., Welborn, T. A. & Knuiman, M. W. Onset of NIDDM occurs
at least 4-7 yr before clinical diagnosis. Diabetes Care 15, 815–9 (1992).
221. Shibata, K. et al. Placental leucine aminopeptidase (P-LAP) and glucose
transporter 4 (GLUT4) expression in benign, borderline, and malignant ovarian
epithelia. Gynecol. Oncol. 98, 11–8 (2005).
222. Cho, H., Lee, Y. S., Kim, J., Chung, J.-Y. & Kim, J.-H. Overexpression of Glucose
Transporter-1 (GLUT-1) Predicts Poor Prognosis in Epithelial Ovarian Cancer.
Cancer Invest. 31, 607–15 (2013).
223. Scafoglio, C. et al. Functional expression of sodium-glucose transporters in
cancer. Proc. Natl. Acad. Sci. 112, E4111–E4119 (2015).
224. Bormans, G. M. et al. Synthesis and biologic evaluation of (11)c-methyl-dglucoside, a tracer of the sodium-dependent glucose transporters. J. Nucl. Med.
176
44, 1075–1081 (2003).
225. Lin, S., Lin, D. C. & Flanagan, M. D. Specificity of the effects of cytochalasin B on
transport and motile processes. Proc. Natl. Acad. Sci. U. S. A. 75, 329–33 (1978).
226. Berridge, M. V, Herst, P. M. & Tan, A. S. Tetrazolium dyes as tools in cell biology:
new insights into their cellular reduction. Biotechnol. Annu. Rev. 11, 127–52
(2005).
227. Macheda, M. L., Rogers, S. & Best, J. D. Molecular and cellular regulation of
glucose transporter (GLUT) proteins in cancer. J. Cell. Physiol. 202, 654–662
(2005).
228. Blessing, A. et al. Sodium/Glucose Co-transporter 1 Expression Increases in
Human Diseased Prostate. J. Cancer Sci. Ther. 4(9), 302–312 (2012).
229. Helmke, B. M. et al. Expression of SGLT-1 in preneoplastic and neoplastic lesions
of the head and neck. Oral Oncol. 40, 28–35 (2004).
230. Baron-Delage, S. et al. Deregulation of hexose transporter expression in Caco-2
cells by ras and polyoma middle T oncogenes. Am. J. Physiol. 270, G314–23
(1996).
231. de Miguel-Luken, M.-J. et al. MAP17 (PDZKIP1) as a novel prognostic biomarker
for laryngeal cancer. Oncotarget 6, 12625–36 (2015).
232. Casneuf, V. F. et al. Expression of SGLT1, Bcl-2 and p53 in primary pancreatic
cancer related to survival. Cancer Invest. 26, 852–859 (2008).
233. Hanabata, Y., Nakajima, Y., Morita, K.-I., Kayamori, K. & Omura, K. Coexpression
of SGLT1 and EGFR is associated with tumor differentiation in oral squamous cell
carcinoma. Odontology 100, 1–8 (2011).
234. Lai, B. et al. Overexpression of SGLT1 is correlated with tumor development and
poor prognosis of ovarian carcinoma. Arch. Gynecol. Obstet. 285, 1455–61
(2012).
235. Ishikawa, N., Oguri, T., Isobe, T., Fujitaka, K. & Kohno, N. SGLT gene expression
in primary lung cancers and their metastatic lesions. Jpn. J. Cancer Res. 92, 874–
9 (2001).
236. Zapata-Morales, J. R., Galicia-Cruz, O. G., Franco, M. & Morales, F. M. Hypoxiainducible factor-1α(HIF-1α) protein diminishes sodium glucose transport 1
(SGLT1) and SGLT2 protein expression In renal epithelial tubular cells (LLC-PK1)
under hypoxia. J. Biol. Chem. 289, 346–357 (2014).
237. Vallon, V. et al. Knockout of Na-glucose transporter SGLT2 attenuates
hyperglycemia and glomerular hyperfiltration but not kidney growth or injury in
177
diabetes mellitus. Am. J. Physiol. Renal Physiol. 304, F156–67 (2013).
238. Cantuaria, G. et al. GLUT-1 expression in ovarian carcinoma: association with
survival and response to chemotherapy. Cancer 92, 1144–50 (2001).
239. Cantuaria, G. et al. Expression of GLUT-1 glucose transporter in borderline and
malignant epithelial tumors of the ovary. Gynecol. Oncol. 79, 33–7 (2000).
240. Shin, S. J. et al. Ciglitazone enhances ovarian cancer cell death via inhibition of
glucose transporter-1. Eur. J. Pharmacol. 743, 17–23 (2014).
241. Lin, H.-W. & Tseng, C.-H. A Review on the Relationship between SGLT2
Inhibitors and Cancer. Int. J. Endocrinol. 2014, 719578 (2014).
242. Bowker, S. L., Yasui, Y., Veugelers, P. & Johnson, J. a. Glucose-lowering agents
and cancer mortality rates in type 2 diabetes: assessing effects of time-varying
exposure. Diabetologia 53, 1631–7 (2010).
243. Evans, J. M. M., Donnelly, L. a, Emslie-Smith, A. M., Alessi, D. R. & Morris, A. D.
Metformin and reduced risk of cancer in diabetic patients. BMJ 330, 1304–5
(2005).
244. Dilokthornsakul, P. et al. The effects of metformin on ovarian cancer: a systematic
review. Int. J. Gynecol. Cancer 23, 1544–51 (2013).
245. Franciosi, M. et al. Metformin therapy and risk of cancer in patients with type 2
diabetes: systematic review. PLoS One 8, e71583 (2013).
246. Zhang, Z.-J. & Li, S. The prognostic value of metformin for cancer patients with
concurrent diabetes: a systematic review and meta-analysis. Diabetes. Obes.
Metab. 16(8), 707–710 (2014).
247. Kumar, S. et al. Metformin intake is associated with better survival in ovarian
cancer: a case-control study. Cancer 119, 555–62 (2013).
248. Romero, I. L. et al. Relationship of type II diabetes and metformin use to ovarian
cancer progression, survival, and chemosensitivity. Obstet. Gynecol. 119, 61–7
(2012).
249. Hur, K. Y. & Lee, M.-S. New mechanisms of metformin action: Focusing on
mitochondria and the gut. J. Diabetes Investig. 6, 600–9 (2015).
250. Zakikhani, M., Dowling, R., Fantus, I. G., Sonenberg, N. & Pollak, M. Metformin is
an AMP kinase-dependent growth inhibitor for breast cancer cells. Cancer Res.
66, 10269–73 (2006).
251. Dowling, R. J. O., Zakikhani, M., Fantus, I. G., Pollak, M. & Sonenberg, N.
Metformin inhibits mammalian target of rapamycin-dependent translation initiation
178
in breast cancer cells. Cancer Res. 67, 10804–12 (2007).
252. Rattan, R., Giri, S., Hartmann, L. C. & Shridhar, V. Metformin attenuates ovarian
cancer cell growth in an AMP-kinase dispensable manner. J. Cell. Mol. Med. 15,
166–78 (2011).
253. Kalender, A. et al. Metformin, independent of AMPK, inhibits mTORC1 in a rag
GTPase-dependent manner. Cell Metab. 11, 390–401 (2010).
254. Ben Sahra, I. et al. The antidiabetic drug metformin exerts an antitumoral effect in
vitro and in vivo through a decrease of cyclin D1 level. Oncogene 27, 3576–86
(2008).
255. Yasmeen, A. et al. Induction of apoptosis by metformin in epithelial ovarian
cancer: involvement of the Bcl-2 family proteins. Gynecol. Oncol. 121, 492–8
(2011).
256. Buzzai, M. et al. Systemic treatment with the antidiabetic drug metformin
selectively impairs p53-deficient tumor cell growth. Cancer Res. 67, 6745–52
(2007).
257. Hirsch, H. a., Iliopoulos, D., Tsichlis, P. N. & Struhl, K. Metformin selectively
targets cancer stem cells, and acts together with chemotherapy to block tumor
growth and prolong remission. Cancer Res. 69, 8832–8833 (2009).
258. Patel, S., Kumar, L. & Singh, N. Metformin and epithelial ovarian cancer
therapeutics. Cell. Oncol. (Dordr). 38, 365–75 (2015).
259. Rocha, G. Z. et al. Metformin Amplifies Chemotherapy-Induced AMPK Activation
and Antitumoral Growth. Clin. Cancer Res. 17, 3993–4005 (2011).
260. Lengyel, E. et al. Metformin inhibits ovarian cancer growth and increases
sensitivity to paclitaxel in mouse models. Am. J. Obstet. Gynecol. 212, 479.e1–
479.e10 (2015).
261. Gotlieb, W. H. et al. In vitro metformin anti-neoplastic activity in epithelial ovarian
cancer. Gynecol. Oncol. 110, 246–50 (2008).
262. Rattan, R., Graham, R. P., Maguire, J. L., Giri, S. & Shridhar, V. Metformin
suppresses ovarian cancer growth and metastasis with enhancement of cisplatin
cytotoxicity in vivo. Neoplasia 13, 483–491 (2011).
263. Shank, J. J. et al. Metformin targets ovarian cancer stem cells in vitro and in vivo.
Gynecol. Oncol. 127, 390–7 (2012).
264. Xie, Y. et al. Metformin combined with p38 MAPK inhibitor improves cisplatin
sensitivity in cisplatin‑resistant ovarian cancer. Mol. Med. Rep. 10, 2346–50
(2014).
179
265. Iliopoulos, D., Hirsch, H. A. & Struhl, K. Metformin decreases the dose of
chemotherapy for prolonging tumor remission in mouse xenografts involving
multiple cancer cell types. Cancer Res. 71, 3196–201 (2011).
266. Wu, B. et al. Metformin inhibits the development and metastasis of ovarian
cancer. Oncol. Rep. 28, 903–8 (2012).
267. Kinaan, M., Ding, H. & Triggle, C. R. Metformin: An Old Drug for the Treatment of
Diabetes but a New Drug for the Protection of the Endothelium. Med. Princ. Pract.
24, 401–15 (2015).
268. Hawley, S. A., Gadalla, A. E., Olsen, G. S. & Hardie, D. G. The antidiabetic drug
metformin activates the AMP-activated protein kinase cascade via an adenine
nucleotide-independent mechanism. Diabetes 51, 2420–5 (2002).
269. Segal, E. D. et al. Relevance of the OCT1 transporter to the antineoplastic effect
of biguanides. Biochem. Biophys. Res. Commun. 414, 694–9 (2011).
270. Gao, S. et al. Attenuating tumour angiogenesis: a preventive role of metformin
against breast cancer. Biomed Res. Int. 2015, 592523 (2015).
271. Bikas, A. D. et al. Glucose-deprivation increases thyroid cancer cell sensitivity to
metformin. Endocr. Relat. Cancer 22, 919–32 (2015).
272. Litchfield, L. M. et al. Hyperglycemia-induced metabolic compensation inhibits
metformin sensitivity in ovarian cancer. Oncotarget 6, 23548–23560 (2015).
273. Silvestri, A. et al. Metformin Induces Apoptosis and Downregulates Pyruvate
Kinase M2 in Breast Cancer Cells Only When Grown in Nutrient-Poor Conditions.
PLoS One 10, e0136250 (2015).
274. Birsoy, K. et al. Metabolic determinants of cancer cell sensitivity to glucose
limitation and biguanides. Nature 508, 108–112 (2014).
275. Sui, X. et al. Use of metformin alone is not associated with survival outcomes of
colorectal cancer cell but AMPK activator AICAR sensitizes anticancer effect of 5fluorouracil through AMPK activation. PLoS One 9, e97781 (2014).
276. Yilmaz, B. et al. Metformin regresses endometriotic implants in rats by improving
implant levels of superoxide dismutase, vascular endothelial growth factor, tissue
inhibitor of metalloproteinase-2, and matrix metalloproteinase-9. Am. J. Obstet.
Gynecol. 202, 368.e1–8 (2010).
277. Scotland, S. et al. Mitochondrial energetic and AKT status mediate metabolic
effects and apoptosis of metformin in human leukemic cells. Leukemia 27, 2129–
38 (2013).
278. Dykens, J. a et al. Biguanide-induced mitochondrial dysfunction yields increased
180
lactate production and cytotoxicity of aerobically-poised HepG2 cells and human
hepatocytes in vitro. Toxicol. Appl. Pharmacol. 233, 203–10 (2008).
279. Wheaton, W. W. et al. Metformin inhibits mitochondrial complex I of cancer cells
to reduce tumorigenesis. Elife 3, e02242 (2014).
280. Andrzejewski, S., Gravel, S.-P., Pollak, M. & St-Pierre, J. Metformin directly acts
on mitochondria to alter cellular bioenergetics. Cancer Metab. 2, 12 (2014).
281. Menendez, J. A. et al. Oncobiguanides: Paracelsus’ law and nonconventional
routes for administering diabetobiguanides for cancer treatment. Oncotarget 5,
2344–8 (2014).
282. Thompson, M. D. et al. Lack of effect of metformin on mammary carcinogenesis in
nondiabetic rat and mouse models. Cancer Prev. Res. (Phila). 8, 231–9 (2015).
283. Shu, Y. et al. Effect of genetic variation in the organic cation transporter 1 (OCT1)
on metformin action. J. Clin. … 117, 1422–31 (2007).
284. Ma, Y. et al. K‑ras gene mutation as a predictor of cancer cell responsiveness to
metformin. Mol. Med. Rep. 8, 763–768 (2013).
285. Aleksandrovski, Y. A. Molecular mechanisms of the cross-impact of pathological
processes in combined diabetes and cancer. Research and clinical aspects.
Biochem. Biokhimiia 67, 1329–1346 (2002).
286. Ross, S. a & Milner, J. a. Epigenetic modulation and cancer: effect of metabolic
syndrome? Am. J. Clin. Nutr. 86, s872–7 (2007).
287. Simon, D. & Balkau, B. Diabetes mellitus, hyperglycaemia and cancer. Diabetes
Metab. 36, 182–191 (2010).
288. Bodmer, M., Becker, C., Meier, C., Jick, S. S. & Meier, C. R. Use of metformin
and the risk of ovarian cancer: a case-control analysis. Gynecol. Oncol. 123, 200–
4 (2011).
289. Van De Poll-Franse, L. V et al. Less aggressive treatment and worse overall
survival in cancer patients with diabetes: a large population based analysis. Int. J.
cancer 120, 1986–1992 (2007).
290. Barone, B. B. et al. Postoperative mortality in cancer patients with preexisting
diabetes: systematic review and meta-analysis. Diabetes Care 33, 931–9 (2010).
291. Rogozan, I. O. N. Relations Between Tumor and Metastasis. 978–984 (1965).
292. Vadlakonda, L., Dash, A., Pasupuleti, M., Anil Kumar, K. & Reddanna, P. Did we
get pasteur, warburg, and crabtree on a right note? Front. Oncol. 3, 186 (2013).
181
293. Singh, R. et al. Non-hyperinsulinemic hypoglycemia in a patient with a
gastrointestinal stromal tumor. Eur. J. Intern. Med. 17, 127–9 (2006).
294. Marks, L. J., Steinke, J., Podolsky, S. & Egdahl, R. H. Hypoglycemia associated
with neoplasia. Ann. N. Y. Acad. Sci. 230, 147–60 (1974).
295. Bishop, J. S. & Marks, P. A. Studies on carbohydrate metabolism in patients with
neoplastic disease. II. Response to insulin administration. J. Clin. Invest. 38, 668–
72 (1959).
296. Jasani, B., Donaldson, L. J., Ratcliffe, J. G. & Sokhi, G. S. Mechanism of impaired
glucose tolerance in patients with neoplasia. Br. J. Cancer 38, 287–92 (1978).
297. Norton, J. A., Maher, M., Wesley, R., White, D. & Brennan, M. F. Glucose
intolerance in sarcoma patients. Cancer 54, 3022–3027 (1984).
298. Denton, I. C., Kerlan, R. A. & McGraw, R. Brain tumor with hyperosmolar
hyperglycemic nonketotic diabetic coma. JAMA 218, 256–7 (1971).
299. Pospisilik, J. A. et al. Targeted deletion of AIF decreases mitochondrial oxidative
phosphorylation and protects from obesity and diabetes. Cell 131, 476–91 (2007).
300. Jacobs, M. A. et al. Multiparametric magnetic resonance imaging, spectroscopy
and multinuclear (23Na) imaging monitoring of preoperative chemotherapy for
locally advanced breast cancer. Acad. Radiol. 17, 1477–85 (2010).
301. Ouwerkerk, R. Sodium magnetic resonance imaging: from research to clinical
use. J. Am. Coll. Radiol. 4, 739–41 (2007).
302. Ouwerkerk, R. et al. Elevated tissue sodium concentration in malignant breast
lesions detected with non-invasive 23Na MRI. Breast Cancer Res. Treat. 106,
151–160 (2007).
303. Jacobs, M. A. et al. Monitoring of neoadjuvant chemotherapy using
multiparametric, 23Na sodium MR, and multimodality (PET/CT/MRI) imaging in
locally advanced breast cancer. Breast Cancer Res. Treat. 128, 119–126 (2011).
304. Ferriss, J. S. et al. Does significant medical comorbidity negate the benefit of upfront cytoreduction in advanced ovarian cancer? Int. J. Gynecol. Cancer 22, 762–
9 (2012).
305. Menzin, A. W., Kobrin, S., Pollak, E., Goodman, D. B. & Rubin, S. C. The effect of
renal function on serum levels of CA 125. Gynecol. Oncol. 58, 375–7 (1995).
306. Turgutalp, K. et al. Serum levels of cancer biomarkers in diabetic and non-diabetic
proteinuric patients: a preliminary study. Clin. Chem. Lab. Med. 51, 889–95
(2013).
182
307. Esteghamati, A. et al. The inverse relation of CA-125 to diabetes, metabolic
syndrome, and associated clinical variables. Metab. Syndr. Relat. Disord. 11,
256–61 (2013).
308. Breborowicz, A., Breborowicz, M., Pyda, M., Połubinska, A. & Oreopoulos, D.
Limitations of CA125 as an index of peritoneal mesothelial cell mass. Nephron.
Clin. Pract. 100, c46–51 (2005).
309. Joo, N.-S., Kim, K.-N. & Kim, K. S. Serum CA125 concentration has inverse
correlation with metabolic syndrome. J. Korean Med. Sci. 26, 1328–32 (2011).
310. da Silva Faria, M. C. et al. Effect of diabetes on biodistribution, nephrotoxicity and
antitumor activity of cisplatin in mice. Chem. Biol. Interact. 229, 119–31 (2015).
311. Harris, D. et al. Glucocorticoid-induced hyperglycemia is prevalent and
unpredictable for patients undergoing cancer therapy: an observational cohort
study. Current Oncology 20, e532–e538 (2013).
312. Balding, L., Philips, T. M. & Fleming, J. S. The management of diabetes in cancer
patients receiving palliative care in a hospital setting. Palliat. Med. 24, S129
(2010).
313. Poulson, J. The management of diabetes in patients with advanced cancer. J.
Pain Symptom Manage. 13, 339–46 (1997).
314. Stanley, I. a, Ribeiro, S. M., Giménez-Cassina, A., Norberg, E. & Danial, N. N.
Changing appetites: the adaptive advantages of fuel choice. Trends Cell Biol. 24,
118–27 (2014).
315. Aronoff, S. L., Berkowitz, K., Shreiner, B. & Want, L. Glucose Metabolism and
Regulation: Beyond Insulin and Glucagon. Diabetes Spectr. 17, 183–190 (2004).
316. Linnerth-Petrik, N. M., Santry, L. A., Petrik, J. J. & Wootton, S. K. Opposing
functions of Akt isoforms in lung tumor initiation and progression. PLoS One 9,
e94595 (2014).
317. Cho, H. et al. Insulin resistance and a diabetes mellitus-like syndrome in mice
lacking the protein kinase Akt2 (PKB beta). Science 292, 1728–31 (2001).
318. Yang, J. & Holman, G. D. Long-term metformin treatment stimulates
cardiomyocyte glucose transport through an AMP-activated protein kinasedependent reduction in GLUT4 endocytosis. Endocrinology 147, 2728–36 (2006).
319. Cefalu, W. T. Paradoxical insights into whole body metabolic adaptations
following SGLT2 inhibition. J. Clin. Invest. 124, 485–7 (2014).
320. Merovci, A. et al. Dapagliflozin improves muscle insulin sensitivity but enhances
endogenous glucose production. J. Clin. Invest. 124, 509–14 (2014).
183
Appendix I: Supplementary figures
Figure 41: Positive SGLT1 staining on human ovarian tumour tissue
184
Dapagliflozin treatment does not affect the weight of established tumours
We hypothesized that reducing systemic blood glucose in diabetic mice would moderate
the growth of epithelial ovarian tumours by limiting fuel availability. We were unable to
reduce fasting blood glucose in metformin-treated mice (Chapter 4), so we attempted to
do so using the SGLT2 inhibitor dapagliflozin, a member of the thiazolidinedione (TZD)
class of drugs that is based on the structure of phlorizin (PZ). TZDs have recently been
developed as a promising new treatment in Type 2 diabetes (T2DM): SGLT2 increases
glucose reuptake into the renal circulation, and blocking this reuptake significantly
reduces blood glucose in an insulin-independent manner.
Dapagliflozin (Cayman Chemical, Ann Arbor, MI) was formulated in polyethylene
glycol/water/ethanol (45:45:10, v/v/v) at a concentration of 5 mg/ml. Mice received a
daily dose of 1 mg/kg dapagliflozin by oral gavage for 45 days (80 days PTI to 125 days
PTI).
185
Figure 42: Dapagliflozin treatment does not affect tumour weight
A) Fasting blood glucose after two weeks (PTI day 95) and six weeks (PTI day 125) of
dapagliflozin treatment. B) Tumour weights from mice sacrificed at PTI day 125. Bars
are mean + SEM. Brackets with stars indicate differences between bars by two-way
ANOVA with Bonferroni corrections. N=5 mice per group. *p<0.05, **p<0.01.
186
In mice 125 days post-tumour induction, daily dapagliflozin treatment for approximately
six weeks (45 days) did not result in a reduction in blood glucose. After two weeks of
treatment, WT animals had a paradoxical increase in blood glucose that may represent
a systemic compensation for dapagliflozin-induced glucosuria (Figure 41A).
Dapagliflozin has been associated with an increase in endogenous glucose production
that is independent of the decrease in plasma glucose concentration319 and can reduce
the glucose-lowering effects by half.320 When mice were sacrificed at 125 days PTI, the
differences in FBG were abrogated. It is possible that by this point in EOC development,
tumours had already altered glucose metabolism (Figures 6 and 8), reducing the
efficacy of dapagliflozin to reduce systemic glucose.
Dapagliflozin treatment was also unable to affect tumour weight (Figure 41B), although
the small size of the tumours may have masked any statistical differences. These data
indicate that dapagliflozin cannot reduce the weight of established tumours, however
they do not address whether or not glucose reduction at tumour initiation can limit
tumour growth.
187
Appendix II: Source of supplies and materials
[14C] α-methyl-D-glucopyranoside
Moravek Biochemicals, Brea, CA
[3H] 2-deoxy-D-glucose
Moravek Biochemicals, Brea, CA
Acrylamide
Bio-Rad Laboratories, Mississauga, ON
Agarose
Life Technologies Inc., Burlington, ON
Ammonium persulfate (APS)
Fisher Scientific, Whitby, ON
Antibiotic-antimycotic (ABAM)
Life Technologies Inc., Burlington, ON
Anti-GLUT1 antibody
Santa Cruz Biotechnologies, Santa
Cruz, CA
Anti-mouse secondary antibody
Cell Signal Technologies, Danvers, MA
Anti-rabbit secondary antibody
Cell Signal Technologies, Danvers, MA
Anti-SGLT1 antibody
Abcam, Cambridge, MA
Anti-SGLT2 antibody
Abcam, Cambridge, MA
Anti-β-actin antibody
Cell Signal Technologies, Danvers, MA
Aprotinin
Sigma, Oakville, ON
Biocoat Matrigel invasion chambers
BD Biosciences, Franklin Lakes, NJ
Biotinylated anti-rabbit secondary antibody
Sigma, Oakville, ON
Bovine serum albumin (BSA)
Sigma, Oakville, ON
Buffered formalin (10%)
Fisher Scientific, Whitby, ON
Citric acid
Sigma, Oakville, ON
Corning cryovials
Fisher Scientific, Whitby, ON
Costar culture dishes
Corning Life Sciences, Lowell, MA
Crystal Violet
Sigma, Oakville, ON
Cytochalasin B
Sigma, Oakville, ON
CytoScint liquid scintillation cocktail
MP Biomedicals, Santa Ana, CA
Dapagliflozin
Cayman Chemical, Ann Arbor, MI
DC Protein Quantification kit
Bio-Rad Laboratories, Mississauga, ON
Deoxyribonucleoside triphosphate (dNTP)
Life Technologies Inc., Burlington, ON
Diaminobenzidine tetrahydrochloride (DAB)
Sigma, Oakville, ON
Dithiothreitol (DTT)
Life Technologies Inc., Burlington, ON
188
DMSO
Sigma, Oakville, ON
DNA ladder (100bp)
Life Technologies Inc., Burlington, ON
Dulbecco's Modified Eagle Medium (DMEM)
Life Technologies Inc., Burlington, ON
Enhanced chemiluminescence (ECL)
Perkin Elmer, Waltham, MA
Ethanol
Greenfield Ethanol Inc., Brampton, ON
Ethidium Bromide
Bio-Rad Laboratories, Mississauga, ON
Ethylenediaminetetraacetic acid (EDTA)
Fisher Scientific, Whitby, ON
Extravidin peroxidase
Sigma, Oakville, ON
Fetal bovine serum (FBS)
Life Technologies Inc., Burlington, ON
Glass coverslips
Fisher Scientific, Whitby, ON
Glucose
Fisher Scientific, Whitby, ON
Glycerol
Fisher Scientific, Whitby, ON
Hematoxylin
Fisher Scientific, Whitby, ON
Hydrogen peroxide
Sigma, Oakville, ON
Leupeptin
Sigma, Oakville, ON
L-glutamine
Life Technologies Inc., Burlington, ON
Metformin hydrochloride
Sigma, Oakville, ON
Methanol
Fisher Scientific, Whitby, ON
MTT
Sigma, Oakville, ON
Nitrocellulose membranes
Amersham, Piscataway, NJ
Nuclease-free water
Applied Biosystems, Foster City, CA
Pepstatin A
Sigma, Oakville, ON
Permount
Fisher Scientific, Whitby, ON
Phenylmethanesulfony fluoride (PMSF)
Roche Applied Science, Laval, QC
Phloretin
Santa Cruz Biotechnologies, Santa
Cruz, CA
Phlorizin
Sigma, Oakville, ON
Phosphate buffered saline
Life Technologies Inc., Burlington, ON
Polybrene
Santa Cruz Biotechnologies, Santa
Cruz, CA
Potassium chloride
Fisher Scientific, Whitby, ON
189
Potassium iodide
Fisher Scientific, Whitby, ON
Potassium phosphate monobasic
Fisher Scientific, Whitby, ON
puromycin dihydrochloride
Santa Cruz Biotechnologies, Santa
Cruz, CA
Rnasin
Fisher Scientific, Whitby, ON
Rneasy Mini Kit
Qiagen, Toronto, ON
ScintiGest tissue solubilizer
Fisher Scientific, Whitby, ON
SGLT2 shRNA lentiviral particles
Santa Cruz Biotechnologies, Santa
Cruz, CA
Skim milk powder
Fisher Scientific, Whitby, ON
SKOV-3 cell line
ATCC, Manassas, VA
Sodium azide
Fisher Scientific, Whitby, ON
Sodium chloride
Fisher Scientific, Whitby, ON
Sodium citrate dihydrate
Fisher Scientific, Whitby, ON
Sodium dodecyl sulfate (SDS)
Sigma, Oakville, ON
Sodium fluoride (NaF)
Fisher Scientific, Whitby, ON
Sodium orthovanadate (NaV)
Sigma, Oakville, ON
Streptozotocin
Sigma, Oakville, ON
Sucrose
Sigma, Oakville, ON
Superfrost Plus glass slides
Fisher Scientific, Whitby, ON
SuperScript II Reverse Transcriptase kit
Life Technologies Inc., Burlington, ON
Taq DNA polymerase
New England BioLabs, Pickering, ON
Tetramethylethylenediamine (TEMED)
Sigma, Oakville, ON
Tissue microarray (human ovarian cancer)
US Biomax, Rockville, MD
Tris base
Fisher Scientific, Whitby, ON
Tris Hydrochloride
Fisher Scientific, Whitby, ON
Triton X-100
Sigma, Oakville, ON
Trypsin-EDTA (10x)
Invitrogen (Life Technologies)
Turbo DNA-free kit
Applied Biosystems, Foster City, CA
Tween-20
Fisher Scientific, Whitby, ON
Vacutainer Serum Separator Tube
Fisher Scientific, Whitby, ON
190
Western Blot apparatus
Bio-Rad
WST-1
Roche Applied Science, Laval, QC
Xylene
Fisher Scientific, Whitby, ON
191
Appendix III: Recipes for solutions
50 mM sodium citrate buffer (pH 4.5 for streptozotocin)
Sodium citrate
0.441 g in
RO H2O
up to 15mL
20x Phosphate Buffered Saline (Stock PBS for IHC)
Sodium chloride
16.0g or
Potassium chloride
0.4g
Sodium potassium dibasic anhydrous
2.3g
Potassium phosphate monobasic (pH 7.4)
0.4g
Immunohistochemistry Blocking Solution
5% BSA
10g
Sodium azide
0.2g
1x PBS
200mL
Citrate Buffer
Stock A – Citrate Buffer
Citric Acid
2.1g
RO H2O
100mL
Stock B – Citrate Buffer
Sodium citrate dihydrate
14.7g
RO H2O
500mL
192
Working Citrate Buffer (pH 6.0)
Stock A
18mL
Stock B
82mL
RO H2O
up to 1L
Sodium Citrate with Tween 20 (pH 6.0)
Sodium citrate dihydrate
2.94g
RO H2O
up to 1L
Tween 20
0.5mL
Carazzi’s Hematoxylin
Hematoxylin
0.25g
Glycerol
50mL
Aluminum potassium sulfate (APS)
12.5g
Potassium iodide
0.05g
RO H2O
200mL
RIPA Lysis Buffer
10mM Tris HCl
0.788g
RO H2O (pH adjusted to 7.6)
up to 495mL
5mM Ethylenediaminetetraacetic acid (EDTA)
0.7306g
50mM Sodium chloride
1.461g
30mM Tetrasodium pyrophosphate
3.988g
1% Triton X-100
5mL
193
Protease inhibitors per mL RIPA (added fresh before each use)
Aprotinin (2.5mg/mL)
2µL
Phenylmethanesulfonyl fluoride (PMSF; 0.871g/mL)
20µL
Sodium Orthovanadate (NaV; 0.1mM)
2µL
Sodium fluoride (NaF; 50mM)
50µL
Pepstatin A (1mg/mL)
1µL
Leupeptin (2mg/mL)
1µL
Reducing Buffer (3x) (Western blot)
10% SDS
2mL
Glycerol
1mL
1M Tris HCl
0.5mL
pH adjusted to 6.8
Bromeophenol Blue
10mg
RO H2O
up to 9mL
DTT
1:7 ratio
5x Running Buffer (Tris-Glycine Buffer; Western blot)
Tris base
15.1g
Glycine
72.1g
10% SDS
10mL
RO H2O
up to 1L
Transfer Buffer (Western blot)
Tris base
7.0g
Glycine
3.02g
Methanol (added fresh before use)
200mL
RO H2O
up to 1L
194
10x Tris buffered saline (TBS; pH 7.6)
Tris base
24.2g
Sodium chloride
80.0g
RO H2O
up to 1L
Tween 20
1 ml
50x Tris-acetate-EDTA (TAE) Buffer (PCR)
Tris base
121g
Glacial acetic acid
28.6g
0.5M EDTA
50mL
RO H2O
up to 500mL
195
Appendix IV: RT-PCR primer sets
Primer set and
accession number
Annealing
temperature
Product
size (bp)
55°C
191
55°C
156
SGLT3
Forward TGAGAGACCCAACAGATGAAGA
(slc5a3)
nm_017391 Reverse TCAAGAAGCCAGCCATTAGAG
55°C
165
GLUT1
Forward
(slc2a1)
nm_011400 Reverse
CAGATGATGCGGGAGAAGAA
60°C
236
GLUT2
Forward
(slc2a2)
nm_031197 Reverse
TCCTGGTCTTCACCCTGTTT
60°C
189
60°C
209
60°C
171
SGLT1
Forward
(slc5a1)
nm_019810 Reverse
SGLT2
Forward
(slc5a2)
nm_133254 Reverse
Sequence (3’-5’)
TCTTCGTCATCAGCGTCATC
GGGGGCTTCTGTGTCCATTTC
ATGATGACCGTGGCTGTGT
TTCCTGCCCTGTTCCTTTTC
ACAACAAACAGCGACACCAC
GGTCGGTTCCTCGGTTTTAG
GLUT3
Forward
(slc2a3)
nm_011401 Reverse
GAACAAAAGACCCCCTCCTC
GLUT4
Forward
(slc2a4)
nm_009204 Reverse
CCTGCTTGGCTTCTTCATCT
AACCTCTCTCCAGCACCAAA
TTTGCCCCTCAGTCATTCTC
196