Thesis Harish Srinivasan

Thesis Harish Srinivasan
Dissertation
Submitted to the
Combined Faculties for the Natural Sciences and for Mathematics
of the Ruperto-Carola University of Heidelberg, Germany
for the degree of
Doctor of Natural Science
presented by
Master Sci. Harish Srinivasan
Born in Dharmapuri, Tamilnadu, India
Oral examination: 21.01.2014
Antibody Microarray as a proteomic tool for effective diagnosis and
prediction of prognosis in cancer
This work was carried out in Division of Functional Genome Analysis
at the German Cancer Research Centre (DKFZ)
Head of division: Dr. Jörg Hoheisel
Referees:
Prof. Dr. Herbert Steinbeisser
PD. Dr. Renate Voit
Thesis Declaration
I hereby declare that I have written the submitted dissertation myself and in this process, I have
used no other sources or materials that those listed in the references.
Place and date:
…………………………………
Harish Srinivasan.
Dedicated to the
Thousands of Tamil lives
Lost during
The last stages of
War in Sri Lanka.
Acknowledgements
Acknowledgements
I am grateful to;
-
My parents who are my best pals and teachers in my life and my brother who has
been a constant support at home while I am away here in Germany.
-
Dr. Jörg Hoheisel for providing me with this opportunity to work in his group and his
valuable inputs and support throughout my stay in his group.
-
Prof. Dr. Herbert Steinbeisser (Human genetics ins., HD) and PD. Dr. Renate Voit
(DKFZ), members of my advisory committee for their guidance and advice and PD.
Dr. Ralf Bischoff (DKFZ) and PD. Dr. Karin Müller-Decker (DKFZ) for their
participation in my oral examination.
-
Christoph Schröder, Mohamed Alhamdani and Stefanie Kutschmann for introducing
me to antibody microarrays and for their discussions and help in initial experiments.
Martin Sill and Christoph Schröder for most of the statistical analysis and
bioinformatics.
-
All the collaborators who have been kind to provide invaluable patient materials.
-
Steffen Klein and Roland Weiss for their extreme support in laboratory work.
-
Sandeep Kumar Botla, Pedro Simonini, Andrea Bauer, Mohanachary Amaravadi,
Aseel Marzoq, Pouria Jandaghi, Syafrizayanti, Christian Betzen, Damjana Kastelic
and Smiths Lueong for their valuable support and discussions in my work.
-
Anke Mahler, Marie Leroy-Schell, Melanie Bier and Sandra Widder for their
administrative support.
-
Laureen Sander and Christian Betzen for their help in translation of the summary in
the thesis to German.
-
All other current and past members of B070 (my group) for their friendship and
friendly atmosphere.
-
My Indian friends, Vijayan, Siva, Namas, Gopal, Haran, Naga and Deepitha for their
wonderful support and time.
-
Last but not the least, DAAD for funding my PhD and my fiancée Deepa for her
support and understanding in the last phases of my work here.
v
Table of contents
Table of contents
Thesis declaration………………………………………………………………………………...iii
Acknowledgement……………………………………………………………………………...…v
Table of contents………………………………………………………………………………….vi
Abbreviations……………………………………………………………………………………...x
Summary…………………………………………………………………………...…………….. 1
Zusammenfassung……………………………………………………………………………..…..2
1
Introduction…………………………………………………………………………………... 4
1.1 Cancer……………………………………………………………….......…………….….... 4
1.1.1 Bladder cancer……………………………………………………………...…………... 6
1.1.1.1 Urothelial cell bladder cancer………………………………...…………………….. 6
1.1.1.2 Prognostic assessment of non-muscle invasive bladder cancer……………………...7
1.1.2 Gastric cancer……………………………………………………………………….……8
1.1.2.1 Gastric adenocarcinoma grading and characteristics……………………………..….9
1.1.2.2 Biomarkers and treatment in gastric adenocarcinoma……………………………...10
1.2 Antibody microarrays in cancer……………………………………………………………12
1.3 Aim………………………......…………………………………………...………………. 14
2
Material and Methods…………………………………………………………………....…..15
2.1 Materials…………………………………………………………………...…………...… 15
2.1.1 Instruments…………………………………………………………………………….. 15
2.1.2 Chemical reagents, enzymes and general materials………………………....………… 16
2.1.3 Cell culture………………………………………………..……………...……………. 18
2.1.4 Dyes and Kits…………………………………………....…………………………….. 18
vi
Table of contents
2.1.5 Antibodies and synthetic RNAs…………………………………………..…….………18
2.1.6 Cell lines………………………………………………….…………………………….19
2.1.7 Buffers and Solutions……………………………………………………..…………… 19
2.2 Methods………………………………………………………………….…………………21
2.2.1 Selection and collection of samples for the study………………...…………………….21
2.2.1.1 Bladder cancer samples……………………………………………….…………….21
2.2.1.2 Gastric cancer samples……………………………………...………………………21
2.2.2 General methods in molecular biology…………………………………………………22
2.2.2.1 Protein isolation…………………………………………………...………………..22
2.2.2.2 Lipid removal from protein samples………………………..………………………23
2.2.2.3 Protein quantification by BCA assay……………………………………………… 23
2.2.2.4 Western blot analysis………………………………...……………………………..23
2.2.3 Methods in cell culture and related assays………………………….…………………..25
2.2.3.1 Routine maintenance of cells……………………….………………………..……..25
2.2.3.2 Transfection of cells………………………………….………..……………………25
2.2.3.3 Cell viability using Sulphorhodamine B (SRB)…………………………………….26
2.2.4 Antibody microarray………………………………………………..…………………..26
2.2.4.1 Selection of antibodies and generation of microarray …………….……………….26
2.2.4.2 Labeling of protein samples…………………………………………….…………..27
2.2.4.3 Blocking of slides and Incubation of samples………………….…………………..28
2.2.4.4 Scanning and signal detection of slides…………………………………………….28
vii
Table of contents
2.2.4.5 Software used in image, data analysis and interpretation……………..……………30
2.2.4.6 Quantification of signal intensities from slides………………………..…..……… 31
2.2.5 Bioinformatics and statistical methods………………………………...…...…………. 32
2.2.5.1 Data analysis by LIMMA bioconductors………………..………………………….32
2.2.5.2 Methods used in sample classification……………………..……………………….32
2.2.5.3 Data analysis by Chipster package…………………………………………...…… 33
2.2.5.4 Interaction and pathway analysis……………………..…………………………….34
3
Results……………………………………………………….……..….......…………...…… 35
3.1 Recurrence prediction in non-muscle invasive bladder cancer…………...………………..35
3.1.1 Hierarchical clustering of samples to study incubation batch effect………..………….35
3.1.2 Differential exp. of proteins between normal and non-muscle invasive bladder tumors37
3.1.3 Differential exp. of proteins between recurrent and non-recurrent
non-muscle invasive bladder tumors………………….……………………………… 38
3.1.4 Accurate prediction of recurrence in non-muscle invasive bladder cancer by
protein signature………………………………………………………………………. 42
3.1.5 Functional annotation of highly significantly abundant proteins……………….…….. 45
3.2 Protein biomarker identification in gastric adenocarcinoma…………………………….. 49
3.2.1 Hierarchical clustering of samples after normalization………………….…………… 49
3.2.2 Differential expression of proteins between normal gastric tissues and
gastric adenocarcinoma tissues…………………………….…………………………. 51
3.2.2.1 Differential expression of proteins- training set of samples…………..……………51
viii
Table of contents
3.2.2.2 Differential expression of proteins- test set of samples………………….…...…… 53
4
Discussion………………………………………………………………………...………… 57
4.1 Cancer biomarkers identification by antibody microarrays………………………………. 57
4.2 Non-muscle invasive bladder cancer……………………………………………...……… 59
4.2.1 Differential exp. of proteins between recurrent and non-recurrent
non-muscle invasive bladder tumors………………………………………..………… 60
4.2.2 Apoptotic proteins and recurrent non-muscle invasive bladder cancers…....……….... 62
4.2.3 Functional annotations by pathway analysis and interactive studies………..………… 63
4.2.4 Recurrence prediction in non-muscle invasive bladder cancers…………….....……… 65
4.3 Biomarker identification for gastric adenocarcinoma by antibody microarrays…………. 66
4.3.1 Differentially regulated proteins in gastric adenocarcinoma……………….…………. 67
5
Conclusions…………………………………………………………………………………. 73
6
References…………………………………………………………………….…………….. 74
7
Appendix………………………………………………………………………..........………87
7.1 Supplementary table S1…………………………………………...........…...……………. 87
7.2 Supplementary table S2……………………………………………...........……………… 88
7.3 Supplementary figure S1…………………………………..........……………….…………96
7.4 Supplementary figure S2………………………………............……………….…………..97
7.5 Publications based on the thesis………………………….............………………….……..98
ix
Abbreviations
Abbreviations
APC
Adenomatous Polyposis Coli protein
APS
Ammonium peroxydisulfate
AUC
Area Under the Curve
BCL2
Apoptosis regulator BCL-2
BSA
Bovine Serum Albumin
CASP3
Caspase-3
CCD
Charge-Coupled Device
CCL18
C-C motif Chemokine 18
CDKN1A
Cyclin-Dependent Kinase Inhibitor 1A
DKFZ
German Cancer Research Centre
DMEM
Dulbecco’s Modified Eagle’s Medium
DMSO
Dimethyl Sulfoxide
DNA
De-oxy Ribonucleic acid
EDTA
Ethylenediaminetetraacetic acid
EGFR
Epidermal Growth Factor Receptor
ERBB2
Proto-oncogene c-erbB2
FBS
Fetal Bovine Serum
HCl
Hydrochloric acid
HPV
Human Palliloma Virus
IHC
Immuno-histochemistry
JUN
proto-oncogene c-JUN
KEGG
Kyoto Encyclopedia of Genes and Genomes
LIMMA
Linear Models for Microarray Data
x
Abbreviations
LMNA
Lamin A
LYAM1
L-Selectin
mRNA
MessengerRNA
OCLN
Occludin
PBS
Phosphate-buffered Saline
RNA
Ribonucleic acid
ROC
Receiver Operating Characteristic
rpm
rotations per minute
RPMI1640
Rosewell Park Memorial Institute Medium
S100A9
Calgranulin B
SD
Standard Deviation
SDS
Sodium dodecyl sulfate
siRNA
small interfering RNA
TBS
Tris-buffered Saline
TCA
Trichloro acetic acid
TGFB
Transforming Growth Factor β
TIA1
Nucleolysin TIA1 isoform-p40
TMA
Tissue Microarrays
TP53
Tumor Protein p53
UV
Ultra-Violet
v/v
volume per volume
VEGFA
Vascular Endothelial Growth Factor A
w/v
weight per volume
WHO
World Health Organization
WMA
World Medical Association
xi
Summary
Summary
Effective prediction of diagnosis and prognosis in cancer is an important step for selection of a
suitable treatment regimen. In this dissertation, the importance of using antibody microarray for
effective diagnosis and prognosis of cancer was studied using bladder and gastric cancers.
In the first part of study, establishment of a protein signature to predict recurrence of non-muscle
invasive bladder cancer was aimed at. Antibodies against cancer-related proteins were spotted
and proteins from recurrent and non-recurrent non-muscle invasive bladder cancer tissues
incubated. The protein profiles of the samples were analyzed for statistical significance and
differential expression of proteins among the cancer groups. After a series of analysis using
bioinformatic tools, a 20 protein-signature predicting recurrence of non-muscle invasive bladder
cancer was identified along with important molecular mechanisms underlying recurrence.
High grade gastric adenocarcinomas are often lethal with metastasis and frequent recurrence.
The second study concentrated on more personalized cancer medicine by direct comparison of
healthy controls and gastric adenocarcinoma tissues from the same patient. Antibody microarray
was used to study the protein profiles of gastric cancer by incubating protein samples from
healthy controls in tandem with the cancer protein from the same patient. Statistically and
clinically significant proteins were identified including a 16 protein signature for betterment of
individual-based cancer treatment regimen. Identified biomarkers included known therapeutic
targets such as VEGFA, S100A9 and newly identified markers like OCLN and TIA1.
The analyses on two cancer types revealed two different protein signatures with high specificity
and sensitivity. Moreover, our findings were clinically relevant and superior to many other
approved available methods for diagnosis and prognosis.
1
Summary
Zusammenfassung
Die effektive Diagnose und Vorhesage der Prognose einer Krebserkrankung sind wichtige
Schritte für die Auswahl eines geeigneten Behandlungsschemas. In dieser Dissertation wurde die
Bedeutung von Antikörper-Microarrays für die effektive Diagnose und Prognose von
Krebserkrankungen am Beispiel von Blasen- und Magenkrebs untersucht.
Das Ziel des ersten Teils der Arbeit war die Etablierung einer Proteinsignatur zur Vorhersage
eines Rezidivs bei nicht-muskulärem invasivem Blasenkrebs. Antikörper gegen krebsrelevante
Proteine wurden auf Microarrays aufgebracht und mit Proteinen aus Tumorgeweben von
rezidivierendem und nicht-rezidivierendem, nicht-muskulärem invasivem Blasenkrebs inkubiert.
Die Proteinprofile der Proben wurden auf statistische signifikante Unterschiede geprüft und die
differentielle Expression der Proteine zwischen den unterschiedlichen Krebsarten analysiert. Mit
Hilfe von bioinformatischen Auswertungsmethoden konnte eine Proteinsignatur bestehend aus
20 Proteinen zur Vorhersage eines Rezidivs bei nicht-muskulärem invasivem Blasenkrebs sowie
wichtige, dem Rezidiv zugrundeliegende molekulare Mechanismen identifiziert werden.
Magen-Adenokarzinome
in
fortgeschrittenen
Stadien
verlaufen
häufig
tödlich,
mit
Metastasierung und hoher Rezidivrate. Der zweite Teil der Arbeit konzentrierte sich stärker auf
personalisierte Krebsmedizin. Hierzu wurden Kontrollen aus gesundem Gewebe und
Gewebeproben aus dem Magen-Adenokarzinom eines Patienten direkt miteinander verglichen.
Unter Verwendung eines Antikörper-Microarrays wurden die Proteinprofile von Magenkrebs
untersucht, indem Proteinproben aus den gesunden Kontrollproben und Tumorproben des
gleichen Patienten zusammen inkubiert wurden. Statistisch und klinisch signifikant
unterschiedlich experimente Proteine einschließlich einer Proteinsignatur aus 16 Proteinen zur
2
Summary
Verbesserung eines individualisierten Krebs-Behandlungsschemas wurden identifiziert. Die
identifizierten Biomarker umfassten sowohl bekannte therapeutische Targets wie VEGFA und
S100A9 als auch neu entdeckte Marker wie OCLN und TIA1.
Die Analyse zweier Krebsarten ergab zwei unterschiedliche Proteinsignaturen mit hoher
Spezifität und Sensitivität. Darüber hinaus waren unsere Ergebnisse klinisch relevant und vielen
anderen anerkannten und verfügbaren Methoden zur Diagnose und Prognose überlegen.
3
Introduction
1 Introduction
1.1 Cancer
Cancer is a non-communicable disease with high complexity and heterogeneity. It is the second
leading cause of death after cardio-vascular diseases (Lozano et al., 2012). 13% of all death
occurrences in world are caused by cancers. Approximately, five hundred out of a hundred
thousand people are at high risk of cancer occurrence in their life time (WHO, 2009). Cancer
nomenclature depends on the part of the body they originate from. All types of cancers are
clinically distinct to each other but they all develop from uncontrolled cell growth resulting in
formation of tissue mass of unstructured cells called tumor. Cancer is classified into two general
types based on their aggressiveness in growth. Benign tumors are those that grow without
invading other adjacent tissues. Tumors that invade nearby tissues and often organs are called
malignant tumors. Malignant tumors often infiltrate into the blood stream, travel to other distant
areas from their organ of origin and form colonies of tumors there. This process is called
metastasis and this potential of tumors best describes the aggressiveness of the corresponding
cancer (Weinberg, 2007). The metastatic cancers more lethal that they account of over 90% of all
cancer related deaths (WHO, 2009).
Cancer is a disease of malfunctioning cells because the native tissue is both structurally and
functionally disorganized in such a way that the immune responses are often evaded or
suppressed. A transformation of a single cell within an organism can lead to formation of a
tumor. This transformation can be facilitated by cancer causing agents like free radicals, toxins,
UV exposure and viruses for example human papilloma virus (HPV). Most of these agents cause
functional mutations in DNA (Weinberg, 2007). Mutations are known to be spontaneous but
4
Introduction
with high rarity in human body. Various safeguard mechanisms facilitate human body to protect
it from serious cellular defects and mutations. Mechanisms like DNA damage repair, cell cycle
arrest, deregulation of cellular energetics are normally active in cells. For example, DNA repair
system removes a part of mutated DNA that might happen during the DNA replication process.
Cell division process, highly controlled normally is arrested when the cells undergo severe
damage due to mutations or stress. In certain cases when a fundamental damage happen to DNA
or cell itself that cannot be repaired, then cells undergo a process of programmed cell death also
called as apoptosis (Alberts B, 2007).
Most types of cancers arise when the above mentioned mechanisms do not work properly,
especially under constant dysregulation of the genes involved in these key processes. Based on
their activities, these self-protective genes are classified into two types: oncogenes and tumor
suppressor genes. Growth-promoting genes that contribute to carcinogenesis by favoring the
tumor cells to evade various growth control signaling mechanisms are called oncogenes. Tumor
suppressor genes, often down-regulated in cancers are generally necessary for cells to maintain
the important growth control barriers and balancing proliferating ability and death of the cells
(Weinberg, 2007). Apart from these genes, several other mechanisms like angiogenesis,
promoting inflammation and self signaling for proliferation of cells also facilitate cancer
development (Hanahan and Weinberg, 2011). All the above mentioned processes account for the
multistep process of carcinogenesis which makes treating the disease a challenge for modern
biologists (Hanahan and Weinberg, 2011).
5
Introduction
1.1.1 Bladder cancer
Bladder cancer is the fourth most common cancer in men and ninth most common cancer in
women. It accounts for more than 3% of total cancer related deaths in a year (Siegel et al., 2012).
According to estimations by world health organization (WHO), hundred and fifty thousand
people die every year of bladder cancer and it belongs to highly monitored post-operative
cancers (Jacobs et al., 2010; WHO, 2009). Though the cancer is less lethal than many other
major cancer types, the cost of treatment and periodical surveillance of the patients are rather
expensive and often result in disappointing results (Sanchez-Carbayo and Cordon-Cardo, 2007).
Cancers in bladder are clinically classified into two major types based on the types of cells
cancer develop from, urothelial cell carcinomas which constitute of 95% of malignant bladder
cancers and squamous cell carcinoma that constitute 5% of bladder cancers (Luis et al., 2007).
1.1.1.1 Urothelial cell bladder cancer
Pathologically, urothelial cell cancers are classified into superficial muscle or non-muscle
invasive cancers and muscle invasive cancers based on their invasiveness. About 75% to 85%
newly diagnosed bladder cancers are non-muscle invasive cancers (grades Ta and T1) and 15%
to 25% are muscle invasive bladder cancers (grades T2 and T3) (figure 1). Even after an initially
successful treatment by complete surgical resection of the tumors, 60% to 70% of the tumors will
recur and 10% to 30% will progress to become muscle invasive cancers. Most of the muscle
invasive cancers are also metastatic (Jacobs et al., 2010). Most of the non-muscle invasive
tumors recur within five year after the surgical resection of the primary tumors. Regular
surveillance cystoscopy and urine cytology is employed every three months during the first two
years after resection, at longer intervals over the next two years, and annually thereafter (Montie
et al., 2009). The lifetime surveillance costs right from pre-operative phases ranges from US$
6
Introduction
99,000 to US$ 121,000 (Avritscher et al., 2006). Above the financial burden, the physical and
psychological stress, the patients are under is more important. Patients live in a state of
uncertainty and are confronted with the possibility of yet another cancer diagnosis every three
months of continuous surveillance (Jacobs et al., 2010; Luis et al., 2007).
Figure 1. Anatomy of bladder- grading of urothelial cell carcinoma based on their depth of
invasion (grades Ta to T3b). Ta tumors do not invade basal muscle or lamina propria and T1 do
not invade muscle (Adapted from department of urology at Michigan urology centre, University
of Michigan).
1.1.1.2 Prognostic assessment of non-muscle invasive bladder cancer
Most prognostic assessments of non-muscle invasive bladder cancer usually employ several
features of tumors like multifocality, clinical grading, size and stage (Dubosq et al., 2012). Many
immuno-histochemical markers have been proposed but in clinical settings, the practice of these
markers has very less impact (Matsushita et al., 2011). Molecules like proliferation activators,
proto-oncogenes and caveolins at mRNA level are also being used with not much correlation to
the recurrence of the cancer (Dubosq et al., 2012; Mitra et al., 2009). Epigenetical modifications
7
Introduction
in promoter sequences are being implemented as predictors of tumor progression (Jeong et al.,
2012). Lower levels of galectin-8 in tumor cells compared to healthy controls are known to
predict recurrence of non-muscle invasive cancer but without great specificity (Kramer et al.,
2011). Mutations in TP53 are also studied as predictors of high risk for recurrence (Andreasson
et al., 2008; Dexlin et al., 2008). Apoptosis regulators like CASP3, BCL2 and TP53 in
combination with other genes predict the outcome of bladder cancer in non-muscle invasive
patients (Korkolopoulou et al., 2000). However, no dependable biomarker or a set of markers
presently exist to predict recurrence with clinically significant accuracy in non-muscle invasive
bladder cancer (Babjuk et al., 2008; Babjuk et al., 2011; Dubosq et al., 2012). Thus, biomarkers
for effective prediction of recurrence in non-muscle invasive bladder cancer can dramatically
reduce the burden caused by the disease on patients and extremely high financial need for health
care.
1.1.2 Gastric cancer
Gastric cancer is second most common in mortality among cancers worldwide, mostly occurring
in developing nations (Dicken et al., 2005). According to WHO estimations, gastric cancer leads
to seven hundred and thirty seven thousand deaths every year which is 10% of overall cancer
related deaths worldwide (Dicken et al., 2005; WHO, 2009). The mortality rate has remained
unchanged over the past three decades despite advancement of surgical techniques and
therapeutics. Gastric cancers are generally highly incident in patients over forty five years old,
hence the mortality rate is quite high despite curative therapy most commonly involving surgical
resection of whole stomach or a part of it. The treatment strategy and response are quite poor
such that over all five year survival rates rage about 10% to 25% (Takeno et al., 2008).
8
Introduction
Besides the high number of incidence and mortality, gastric cancer is a heterogeneous cancer in
terms of clinical and pathological classification, genomic and proteomic background of the
tumor mass and clinical outcomes (Zheng et al., 2004). Gastric cancer types are named after their
cells they originate from. Most of the diagnosed gastric cancers are adenocarcinomas (95%),
other types include squamous cell carcinoma, neuroendocrine tumors, stromal tumors and gastric
lymphoma (Dicken et al., 2005; Zheng et al., 2004).
1.1.2.1 Gastric adenocarcinoma grading and characteristics
Gastric adenocarcinoma arising from secretory epithelia in the inner lining of gastric mucosa is
the most aggressive form of gastric cancers. Based on the invasiveness of the tumor and
differentiation of the cells, adenocarcinoma is clinically and pathologically graded into four
types (figure 2) (Dicken et al., 2005). Most of the adenocarcinomas diagnosed belong to
advanced stages (T3/T4) of the cancer accompanied by lymph node metastasis. The association
of numerous lymph nodes morphologically with stomach tissues enables irresistible metastasis of
tumor cells in advanced stages of the cancer (Senapati et al., 2008). Furthermore, advanced
stages of gastric adenocarcinoma recur at relative higher rates of 40% to 70% after the curative
surgical resection of primary tumor. Routine surveillance by means of endoscopy followed by
adjuvant chemotherapy after primary resection leads to a median survival of nineteen months. In
case of limited surveillance it drops to just eight months. Advanced stages of carcinoma without
curative resection of tumors have a median survival rate of merely five months (Dicken et al.,
2005). Such low survival rates combined lymph node metastasis make treatment options not so
much responsive and in turn lead to increased mortality (Dicken et al., 2005; Pietrantonio et al.,
2013; Zheng et al., 2004).
9
Introduction
Figure 2. Anatomy of stomach- staging of gastric adenocarcinoma based on their extent of
invasiveness (grades Ta to T4). Ta tumors originate at the inner lining and slowly protrude out
invading other layers of stomach. T4 tumors are highly metastatic (Adapted from cancer research
UK-homepage for gastric cancers).
1.1.2.2 Biomarkers and treatment in gastric adenocarcinoma
Gastric adenocarcniomas are often diagnosed merely by the localization of the tumor cells and
their invasiveness. There are very few molecular biomarkers available to grade them according
to their invasiveness (Pietrantonio et al., 2013). Diagnosis based on the transcript levels of p53
and mutations in APC genes predicted gastric cancer but failed to differentiate the type of tumor
(Zheng et al., 2004). Growth factor receptors like EGFR, c-met and ERBB2 have been used as
predictors of later stages of malignancy but with much instability. CCL18 combined with certain
T-cell receptors are employed as prognostic indicators in lower stages of gastric cancer (Leung et
10
Introduction
al., 2004). Epithelial-mesenchymal transition marker proteins like E-cadherin are also used as
predictors for higher stages of gastric adenocarcinomas. Certain cytokines are used as predictors
of prognosis in gastric cancers associated with Helicobacter pylori infection but they do not have
much significance in prediction of other types of adenocarcinoma (Ellmark et al., 2006).
Many promising molecular biomarkers like S100A9, VEGFA and mucin family of proteins are
also tested as potential therapeutic targets. Despite recent developments in gene sequencing and
molecular diagnostics, many of these biomarkers are inconsistent in predicting a unanimous
treatment regimen. Few treatment strategies have been introduced but molecular complexity and
drug resisting capability of the cancers make the treatment of adenocarcinomas nearly
impossible. Nevertheless, many biomarkers are being evaluated by various clinical trials to
identify individual-based criteria and establish customized personal therapeutic approaches
(Pietrantonio et al., 2013). The best treatment regimen available so far for higher stages of gastric
cancer is curative surgical resection of the tumor followed by adjuvant chemotherapy with
oxaliplatin and capecitabine. These two agents are reportedly highly efficient combined with less
toxic effects and side-effects (Cunningham et al., 2010). Unfortunately, clinical trials on different
biomarkers are experiencing critical ignorance such that most fail rather than considered as
potential targets. Thus, introducing new therapeutic targets and predictive markers along with
immuno-based screening tools will enable developing patient-specific chemotherapy in turn
optimal drug efficiency and minimal adverse effects.
11
Introduction
1.2 Antibody microarrays in cancer
Omic technologies are excellent tools for both diagnostic and prognostic biomarker identification
in cancers, a disease with high molecular and metabolic complexity. DNA microarrays are
widely employed for initial screening studies on biomarkers at transcript levels in a wide variety
of diseases. Inconsistency in validation of biomarkers identified by various genomic technologies
led to the emergence of proteomic applications in biomarker discovery. Antibody-based
technologies are extremely advantageous in biomarker discovery as they use biological end
product which indulge in vital cellular activities (Michaud et al., 2003). NHS–based protein
labeling methods provide a platform for large scale analysis of antibodies. Optimizing various
factors including sensitivity and specificity for targeting analytes in low quantities brought in a
new method of proteomic analysis (Wingren et al., 2007). Thus antibody microarrays are
relatively a new tool for an analysis of protein abundance in parallel and highly multiplexed
manner (Brennan et al., 2010) (figure 3).
Antibody microarrays are extremely different tissue microarrays where the interpretation and
validation of enriched targets are often subjected to differing opinions among individuals.
Another advantage antibody array has over tissue microarray is the use of non-invasive body
fluids like urine, saliva and plasma (Alhamdani et al., 2009). Solid supports like glass slides
coated with chemical binders are used to immobilize antibodies on the glass surface. Cancerrelated antibodies are used for the production of antibody arrays in large scale (Kusnezow et al.,
2003). The number of targets used in the arrays varied from mere hundred to as high as thousand
in fact with higher specificity and sensitivity (Sanchez-Carbayo et al., 2006; Schroder et al.,
2010). Studies involving protein extracts from tissues and cells are rarely done using antibody
microarrays due to technical aspects (Hoheisel et al., 2013). However, optimized protocols and
12
Introduction
detailed assessment are being established which permit antibody microarray with high accuracy
and reproducibility for effective cancer diagnosis and prognosis (Alhamdani et al., 2010;
Schroder et al., 2010). Recently, higher sensitivity detecting single molecules using antibody
microarrays have been reported (Schmidt et al., 2011). Thus, antibody microarray being an
immuno-based assay, can bridge the gap between effective diagnosis and personalized cancer
medicine.
Figure 3. Proteomic technologies being employed for effective cancer diagnosis and
personalized cancer medicine. (Adapted from Brennan et al 2010 Nature reviews cancer).
13
Introduction
1.3 Aim
In this dissertation, potentiality in using antibody microarray as a tool for effective cancer
diagnosis and prediction of prognosis is studied.
The first part of the study is aimed at predicting recurrence of non-muscle invasive bladder
cancer. For this purpose, twenty five non-muscle invasive bladder cancer samples with a
minimum of five year follow-up time after the initial surgical resection were analyzed using
antibody microarray. After the analysis, the identified protein signature would be helpful for the
prediction of recurrence in non-muscle invasive bladder cancer and to establish a signature-based
treatment regimen for bladder cancer patients.
The second part of the study is aimed at identifying more personalized biomarkers for effective
diagnosis and treatment of gastric adenocarcinomas. Thirty cancer tissues and healthy control
tissues from the same patient were analyzed using antibody microarray to identify a significant
protein profile. Identified protein profile will provide valuable insights on tumor development in
gastric adenocarcinoma. The identified protein profile will not only help in an individual-based
treatment regimen for gastric cancer treatment but also will identify more therapeutically
important targets.
14
Materials and Methods
2 Material and Methods
2.1 Materials
2.1.1 Instruments
Instrument
Manufacturer
12-channel-pipette Biohit proline
Biohit, Helsinki, Finland
96-well flat-bottom block
Qiagen, Hilden, Germany
96-well reaction plates
Steinbrenner, Wiesenbach, Germany
6-well culture plates
Steinbrenner, Wiesenbach, Germany
24-well culture plates
Steinbrenner, Wiesenbach, Germany
Automatic developing machine
Amersham, Freiburg, Germany
Beckmann GS-6KR centrifuge
Beckmann, Wiesloch, Germany
Centrifuge 5810 R
Eppendorf, Hamburg, Germany
Centrifuge 580 R
Eppendorf, Hamburg, Germany
Cell culture incubator
Koettermann, Haenigsen, Germany
Cell culture microscope
Carl Zeiss, Jena, Germany
Cell viability analyzer
Beckmann, Wiesloch, Germany
Dismembrator
B.Braun Biotech, Melsungen, Germany
Dry block heating system
Grant instruments, Cambridge, UK
Electrophoresis power supply
E-C apparatus corporation, USA
Epoxysilane-coated slides Nexterion-E
Schott, Jena, Germany
Infinite m200 multimode reader
Tecan Grp Ltd, Maennedorf, Switzerland
MicroGrid II Array-Roboter
Biorobotics, Cambridge, UK
Mini-protein electrophoresis system
Bio-Rad Laboratories, Munich, Germany
Power supply E835
Hoefer, CA, USA
QuadriPERM plates
Vivascience, Hannover, Germany
ScanArray 4000XL
Perkin Elmer, Massachusetts, USA
Sigma 2k15 centrifuge
M&S laborgeraete gmbh, Wiesloch, Germany
15
Materials and Methods
SlideBooster hybridization station
Advalytix, Munich, Germany
SMP3B stealth pins
Telechem, CA, USA
TE70 PWR semidry transfer unit
Amersham, San Fransisco, USA
Vortex
Scientific industries Genie-2, New York,USA
Water bath SW22
Julabo Labortechnik, Seelbach, Germany
2.1.2 Chemical reagents, enzymes and general materials
Reagent
Supplier
2-Mercaptoethanol
Roche Diagnostics, Mannheim, Germany
6-Aminocaproic acid
Sigma-Aldrich, Munich, Germany
12-Maltoside
Sigma-Aldrich, Munich, Germany
Acetic acid
Mallinckrodt Baker, Greisheim, Germany
Acrylamide (30%w/v)/Bisacrylamide
(29/10.8%)
Ammonium peroxydisulfate (APS)
Bio-Rad Laboratories, Munich, Germany
ASB-14
Sigma-Aldrich, Munich, Germany
Bovine serum albumin (BSA)
Sigma-Aldrich, Munich, Germany
Bromphenol blue
Sigma-Aldrich, Munich, Germany
Chloroform
Merck, Darmstadt, Germany
CleanasciteTM lipid removal agent
Biotech support group, New Jersey, USA
Dimethylsulfoxide (DMSO)
Sigma-Aldrich, Munich, Germany
ECL hyperfilm
GE Healthcare Europe, Freiburg, Germany
Ethylenediaminetetraacetic acid (EDTA)
Merck, Darmstadt, Germany
Ethanol
Merck, Darmstadt, Germany
Glycerin
Roth, Karlsruhe, Germany
Glycine
Roth, Karlsruhe, Germany
Hydrogen chloride (HCl)
Merck, Darmstadt, Germany
Isopropanol (2-propanol)
Mallinckrodt Baker, Greisheim, Germany
Magnesium chloride
Merck, Darmstadt, Germany
Sigma-Aldrich, Munich, Germany
16
Materials and Methods
Membran Protran NCBA85
Whatman gmbh, Hassel, Germany
Methanol
Merck, Darmstadt, Germany
Milk powder (Skimmed)
Sigma-Aldrich, Munich, Germany
Na2HPO4
Merck, Darmstadt, Germany
NaH2PO4
Merck, Darmstadt, Germany
NaOH
Merck, Darmstadt, Germany
Natriumacetat
Merck, Darmstadt, Germany
Natriumazide
Merck, Darmstadt, Germany
Natriumchloride
Merck, Darmstadt, Germany
Natriumcitrate
Merck, Darmstadt, Germany
Na-cholate
Sigma-Aldrich, Munich, Germany
NP-40 substitute
Sigma-Aldrich, Munich, Germany
Nuclease free water
Ambion, Austin, USA
PMSF
Sigma-Aldrich, Munich, Germany
Potassium Chloride
Sigma-Aldrich, Munich, Germany
Potassium dihydrogen phosphate
Sigma-Aldrich, Munich, Germany
Sodium Carbonate
Sigma-Aldrich, Munich, Germany
Sodium dodecyl sulfate (SDS)
Sigma-Aldrich, Munich, Germany
Spectra multicolor broad range protein ladder
Thermo Scientific, Rockford, USA
Sulphorhodamine B (SRB)
Sigma-Aldrich, Munich, Germany
T-PER tissue protein extraction reagent
Thermo Scientific, Rockford, USA
TEMED
Bio-Rad Laboratories, Munich, Germany
Trichloro acetic acid (TCA)
Fisher Chemicals, Reading, UK
Tris-Base
Sigma-Aldrich, Munich, Germany
Tris-HCl
Sigma-Aldrich, Munich, Germany
Triton-X100
Sigma-Aldrich, Munich, Germany
Tween-20
Sigma-Aldrich, Munich, Germany
Tween-80
Sigma-Aldrich, Munich, Germany
17
Materials and Methods
2.1.3 Cell culture
Reagent
Supplier
DMEM
Gibco/Invitrogen, Karlsruhe, Germany
DMEM/F12 (Ham)
Gibco/Invitrogen, Karlsruhe, Germany
Fetal Bovine Serum (FBS)
Gibco/Invitrogen, Karlsruhe, Germany
L-Glutamine
Gibco/Invitrogen, Karlsruhe, Germany
Phosphate buffered saline (PBS)
Gibco/Invitrogen, Karlsruhe, Germany
Penicillin 1000u/ml-Streptomycin 100µg/ml
Gibco/Invitrogen, Karlsruhe, Germany
RPMI
Gibco/Invitrogen, Karlsruhe, Germany
Trypsin/EDTA solution
Gibco/Invitrogen, Karlsruhe, Germany
2.1.4 Dyes and Kits
Item
Supplier
The blocking solution
CANDOR Biosci. gmbh, Wangen, Germany
Dy-549-NHS
Dyomics, Dresden, Germany
Dy-649-NHS
Dyomics, Dresden, Germany
ECL prime western blot detection kit
GE Healthcare Europe, Freiburg, Germany
Novagen BCA protein assay kit
Merck, Darmstadt, Germany
siPORTTM NeoFXTM Reverse transfection kit
Ambion, Austin, USA
2.1.5 Antibodies and synthetic RNAs
Item
Supplier
Dilutions
Anti.TIA-1
Santa Cruz, Biotech. Inc., Texas, USA
1:1000
Anti.S100A9
Santa Cruz, Biotech. Inc., Texas, USA
1:500
Sec. antibody-Goat
Santa Cruz, Biotech. Inc., Texas, USA
1:5000
GAPDH-HRP conjugated
Sigma-Aldrich, Munich, Germany
1:25000
TIA1 siRNA-1
Qiagen, Hilden, Germany (SI00133098)
18
Materials and Methods
TIA1 siRNA-2
Qiagen, Hilden, Germany (SI00133105)
TIA1 siRNA-3
Qiagen, Hilden, Germany (SI00133112)
TIA1 siRNA-5
Qiagen, Hilden, Germany (SI00301917)
TIA1 siRNA SMARTpool
Dharmacon, Rockford, USA (M-013042-02-005)
Allstars Negative control siRNA
Qiagen, Hilden, Germany (1027280)
2.1.6 Cell lines
Cell lines
Cellular origin
Tumorigenic
Source
AGS
Stomach
Yes
Cell lines service (300408)
HGC-27
Stomach
Yes
Cell lines service (300436)
MKN-45
Stomach
Yes
Dr. Christiane Dinsart, DKFZ
2.1.7 Buffers and Solutions
Solutions and Buffers
Components
APS
10% APS w/v in ddH2O
10X PBS
137 mM NaCl
27 mM KCl
100 mM NaH2PO4
17 mM KH2PO2
dissolved in ddH2O
PBST
10X PBS with 0.1% (v/v) Tween 20
PBSTT
10X PBS with 0.1% (v/v) Tween 20 and
0.1% (v/v) Triton-X 100
10X TBS
500 mM Tris.HCl
1500 mM NaCl
dissolve in ddH2O and pH 7.6
TBST
10X TBS with 0.1% (v/v) Tween 20
10% SDS
10% (w/v) SDS dissolve in ddH2O
19
Materials and Methods
Stacking gel buffer
1.5 M Tris.HCl
dissolve in ddH2O and pH 6.8
Resolving gel buffer
1.5 M Tris.HCl
dissolve in ddH2O and pH 8.8
10X SDS gel Tank buffer
50 mM Tris-Base
400 mM glycine
10% (w/v) SDS
dissolve in ddH2O
Lysis buffer for protein isolation
500 µl of NP-40
1000 µl of Na-Cholate
1000 µl of ASB-14
1000 µl of 12-maltoside
2000 µl of glycerol (99%)
1000 µl of sodium carbonate
167 µl of magnesium chloride
500 µl of EDTA.2Na
50 µl of PMSF
100 µl of protease and phosphotase inhibitor
0.4 µl of Benzonase (100U/µl)
2683 µl of ddH2O
Western blot anode buffer I
30 mM Tris-Base
20% (v/v) methanol dissolve in ddH2O
Western blot anode buffer II
5 mM Tris-Base
20% (v/v) methanol
dissolve in ddH2O
Western blot cathode buffer
20 mM 6-aminocaproic acid
20% (v/v) methanol
dissolve in ddH2O
Milk-blocking buffer
5% milk powder in 1XTBST
Milk blocking buffer for microarray
10% milk powder in 1XPBSTT
20
Materials and Methods
2.2 Methods
2.2.1 Selection and collection of samples for the study
2.2.1.1 Bladder cancer samples
Primary tumors were resurrected from patients with low-grade (stages Ta and T1) non-muscleinvasive bladder cancer at the Department of Urology and Pathology of Henri Mondor Hospital
in Paris, France. All subjects were informed with a written consent and the same was obtained
from all subjects. The analysis was approved by the local ethics committee and all the
experiments conformed to the principles set out in the WMA Declaration of Helsinki. Only
samples were considered with a patient follow-up of at least five years post the surgical resection
of the primary tumour. Nineteen patients experienced recurrence within the five years of followup while six patients did not. In addition, three normal bladder tissue samples were available as
controls. The age and gender of all the samples studied are equally distributed and none of the
patients did receive any cancer-related therapy before sampling. The tissue samples were snapfrozen in liquid nitrogen soon after resection to protect the tissue and cellular components. Then,
the tissue samples were pulverized using dismembrator under deep freeze conditions (-50ºC) and
protein isolation was carried out.
2.2.1.2 Gastric cancer samples
Primary tumors from patients diagnosed with high-grade and stage invasive gastric
adenocarcinoma along with a part of the normal tissue surrounding the tumors were resurrected
at the Department of Gastroenterology, University medical centre, Ljubljana, Slovenia. All
subjects were informed with a written consent and the same was obtained from all subjects. The
analysis was approved by the local ethics committee and all the experiments conformed to the
21
Materials and Methods
principles set out in the WMA Declaration of Helsinki. Twenty five pairs (normal and tumour) of
tissue samples were collected from twenty five patients and the samples were snap-frozen in
liquid nitrogen until further use to protect the tissue and cellular components. Then, the tissue
samples were pulverized using dismembrator under deep freeze conditions (-50ºC) and protein
isolation was carried out.
2.2.2 General methods in molecular biology
2.2.2.1 Protein isolation
Homemade lysis buffer was used for the isolation of whole protein from tissues and cells.
From Tissues
Pulverized tissue was collected in eppendorf tubes and ice cold lysis buffer was added at a
volume of 10 µl/1 µg of tissue. A disposable syringe with needle was used to break and tear the
tissue and the solution was aspirated and suspended continuously. The eppendorf tubes were
shaken gently for 10 minutes on ice and centrifuged for 5 minutes at 13,000 rpm in a table top
centrifuge at 4ºC. The supernatant was collected in a fresh eppendorf tube and stored at -20ºC.
From Cells
Medium from the 6-well plates was aspirated and 60 µl of lysis buffer was added to each well of
the plate. A cell scraper was used to spread the lysis buffer over the cells and to detach them.
Lysates were transferred to eppendorf tubes and the tubes were gently shaken on ice for 10
minutes. Then the tubes were centrifuged for 5 minutes at 13,000 rpm in a table top centrifuge at
4ºC. The supernatant was collected in a fresh eppendorf tube and stored at -20ºC.
22
Materials and Methods
2.2.2.2 Lipid removal from protein samples
Protein samples were often contaminated with lipids that made quantification and labeling of the
samples difficult. CleanasciteTM was added to the samples to remove lipids. Cleanascite solution
was added to the sample at a ratio of 1:2 and incubated for an hour at 4ºC with constant shaking
at a very low speed. After centrifugation at 13,000 rpm for 5 minutes, the supernatant free from
lipids was collected in a fresh eppendorf tube and the concentration was measured by BCA
assay. Then the tubes were stored at -20ºC. The precipitate containing lipids was discarded.
2.2.2.3 Protein quantification by BCA assay
Total protein from both tissue samples and cells were quantified using Novagen BCA protein
assay kit. This method involves the colorimetric detection of cuprous cation using bicinchoninic
acid (BCA) which is formed by the reduction of Cu+2 to Cu+1 by the protein in an alkaline
medium. BCA reagent A and reagent B were mixed at a ratio of 50:1 to make the BCA working
reagent. 10 µl of BSA dissolved in lysis buffer and samples were pipette into the wells of a fresh
96-well microtiter plate. 200 µl of the BCA working reagent was added to each well and the
plate was incubated at 37ºC for 30 minutes. The absorbance was then measured at 562 nm and a
standard curve of absorbance versus standard proteins (in µg) was prepared to determine the
concentration of the samples.
2.2.2.4 Western blot analysis
10% SDS gels were used for resolving protein. 10% and 5% acrylamide/bisarcylamide were used
respectively for resolving and stacking part of the gel. 0.06% (w/v) ammoniumpersulphate and
0.1% (v/v) N, N, N’, N’ – tetramethylethylenediamine (TEMED) were used to induce the
polymerization of the gel. 10 µg of protein with rotiload loading dye were boiled together for 5
minutes and loaded into the respective slots in the gel. A prestained- protein ladder was also
23
Materials and Methods
loaded referring to the molecular weight. Electrophoresis of the gel was carried out for 90
minutes at 135 V and 500 mA in 1X SDS-gel tank buffer. The transfer of polypeptides from the
gel to a nitrocellulose membrane was carried out by TE70 PWR semidry transfer apparatus. A
sandwich model was prepared by soaking Whatman filter papers in anode buffers I, II and
cathode buffer. The membrane was activated in anode buffer II.
The stacking part of the gel was cut and the sandwich was assembled with the filter papers,
membrane and the gel after which the semidry electrophoretic transfer was carried out for 60
minutes at 35 V and 500 mA. To detect the transferred protein, after the transfer the membrane
was blocked for 1h at room temperature with the milk blocking buffer. After blocking, the
membrane was incubated with the diluted primary antibody over night at 4ºC. After incubation,
the membrane was washed 3 times with 1XTBST and was incubated with secondary antibody
conjugated with horse radish peroxidase for 1h at room temperature. Then, the membrane was
washed for 3 times with 1XTBST and protein was detected by enhanced chemiluminescence
(ECL) using the ECL prime western blot detection kit. The ECL substrate was prepared
according to the manufacturer’s instructions and incubated on the membrane for 1 minute and
the solution was drained off. Now, the membrane was kept on a clean plate inside the LAS
Fujifilm 5000 machine and images were captured using a CCD camera on exposing the
membrane gradually to the X- rays. Similarly, the procedure was repeated for the detection of the
house keeping protein in the same membrane.
24
Materials and Methods
2.2.3 Methods in cell culture and related assays
2.2.3.1 Routine maintenance of cells
Cell lines were maintained at 37ºC and 5% CO2. MKN-45 cells were maintained in RPMI with
4.5 g/L D-glucose supplemented with 10% fetal bovine serum (FBS), 1000 u/ml penicillin and
100 µg/ml streptomycin and 2 mM glutamine. AGS cells were maintained in DMEM with 4.5
g/L D-glucose supplemented with 10% fetal bovine serum (FBS), 1000 u/ml penicillin and 100
µg/ml streptomycin and 2 mM glutamine. HGC-27 cells were maintained in DMEM/F12 (ham)
with 4.5 g/L D-glucose supplemented with 5% fetal bovine serum (FBS), 1000 u/ml penicillin
and 100 µg/ml streptomycin and 2 mM glutamine. All cells were passaged every 2 to 3 days and
sub cultured in fresh medium.
2.2.3.2 Transfection of cells
All the RNA transfections in this study were carried out in 6-well or 96-well plates using
siPORTTM NeoFXTM (Ambion) reagent. Reverse transfection by means of siPORTTM NeoFXTM
involves simultaneous tranfecting and plating of cells. siPORTTM NeoFXTM transfection agent
and the RNA molecules are mixed and distributed on the culture plates over which the cells are
overlaid. The final transfection volume in a 6-well plate is 2.5 ml of medium containing 2 x 105
cells per well and in a 96-well plate is 100 µl of medium containing 5 x 103 cells per well. As the
transfection complexes are stable in presence of serum, no change of medium or other
precautionary measures taken in case of traditional transfections methods are needed. The final
concentration of the RNA molecules transfected ranges from 5 nM to 50 nM. After this
procedure, the plates were maintained at 37ºC and 5% CO2.
25
Materials and Methods
2.2.3.3 Cell viability using Sulphorhodamine B (SRB)
After 48 hours of incubation of cells at 37ºC, the plates were taken out and medium is discarded.
10% (w/v) TCA was added to the wells and the plates were incubated at 4ºC for 2 hours to fix
the cells. TCA is then discarded and the plates were rinsed with water and dried at 37ºC for 20
minutes. 0.05% (w/v) of sulphorhodamine B (SRB) reagent was added to the wells and the plates
were incubated for 30 minutes at room temperature in dark. The plates were washed for 3 to 4
times with 1% (v/v) acetic acid to remove SRB reagent and then dried for 20 minutes at 37ºC.
100 mM Tris was added to the plates and the plates were shaken for 10 minutes after which, the
absorbance was measured at 570 nm from the stained cells and at 650 nm from blank after
which, the results were tabulated for calculating the percentage of viable cells after transfection.
2.2.4 Antibody microarray
2.2.4.1 Selection of antibodies and generation of microarray
In-house developed antibody microarray had 813 antibodies directed against 724 cancer related
proteins. The antibodies were selected based on previous studies of transcription profiling
involving many cancer entities and strong literature search on cancer related proteins. The list of
antibodies is found in annexure A. The antibodies were printed on epoxysilane-coated NexitronE slides using the contact printer MicroGrid-2 and SMP3B pins at a humidity of 40% to 45%.
The buffer composed of 0.1 M carbonate buffer of pH 8.5 with 0.01% tween-20, 0.05% sodium
azide, 0.5% dextran, 5 mM magnesium chloride and 1 mg/ml of antibody. Streptavidin
molecules labeled with Cy3 or Cy5 were spotted as dye controls and the slides were allowed to
quilibrate at humidity of 40% to 45% overnight and then stored at 4ºC in dry, dark conditions
until further use.
26
Materials and Methods
Figure 4. overview of production of antibody microarrays. Steps a to f constitute the proteome
profiling of cancers using antibody microarrays.
2.2.4.2 Labeling of protein samples
Extracted and estimated protein samples were labeled at a dye/protein molar ratio of 1:18,
assuming that the average weight of protein is 60 KDa. The NHS-esters of dye paris DY-549
(Cy3) and DY-649 (Cy5) were used to label the protein samples. The protein samples were all
adjusted to a concentration of 1 mg/ml and labeling was carried on in the dark in 0.1 M
27
Materials and Methods
carbonate buffer, pH 8.5 at 4ºC for 2 hrs. Later, the unreacted dye was quenched by adding 10%
glycine in the dark at 4ºC for 30 minutes. The non-incorporated and inactivated dye molecules
were removed from the samples by using zeba spin-desalting columns with a cutoff of 3.5 KDa
and a buffer change to PBS according to the manufacturer’s recommendations. After dye
removal, the labeled samples were stored at -20ºC until further use.
2.2.4.3 Blocking of slides and Incubation of samples
The slides were taken out and allowed to settle at room temperature for about 10 minutes. Then
they were washed with 1X PBST in continuous agitation at room temperature for 4 times at
regular time interval. Then the slides were blocked with 1 ml of the blocking solution from
candor buffer on the slide booster instrument for 4 hrs at room temperature. The blocking
solution was then removed and 10 µl of each of the labeled samples (Cy3 and Cy5) were added
to 580 µl of the blocking solution from candor buffer such that the total incubation volume of
600 µl kept ready for incubation. 600 µl of incubation solution was added to each slide and the
incubation was carried on the slide booster instrument for 16 hrs at room temperature. After the
incubation, the slides were washed with 1X PBST on the slide booster and then washed 4 times
in continuous agitation at room temperature. The slides were quickly air dried using a strong
clean air blower.
2.2.4.4 Scanning and signal detection of slides
Slides were scanned and the fluorescence of Cy3 and Cy5 were measured using ScanArray
4000XL instrument. The slides were scanned at a resolution of 10 µm, maintaining constant laser
power and photomultiplier (PMT). The excitation of Cy5 was achieved by a laser emission at a
wavelength of 635 nm while Cy3 excitation was achieved by a laser emission at a wavelength of
532 nm. Two separate images were made for each dye and were then quantified.
28
Materials and Methods
Figure 5. Flowchart of the experimental procedure to analyse non-muscle invasive bladder
cancer samples.
29
Materials and Methods
Figure 6. Flowchart of the experimental procedure to analyse gastric cancer samples.
2.2.4.5 Software used in image, data analysis and interpretation
Software
Manufacturer
GenePix Pro 6.0
Axon Instruments, CA, USA
LIMMA Package R-bioconductors
Fred Hutchinson cancer research centre, USA
Chipster package (v1.4.6)
CSC-IT centre for science, Espoo, Finland
STRING 9.1 database for known and predicted protein-protein interaction.
DAVID bioinformatics resources 6.7- functional annotation and microarray analysis
KEGG (Kyoto Encyclopedia of Genes and Genomes) Kanehisa Labs, Kyoto, Japan
30
Materials and Methods
2.2.4.6 Quantification of signal intensities from slides
Image analysis was carried out using GenePix Pro 6.0 software where the images were uploaded
into the software and overlaid one over the other. The spots in the merged image were analyzed
primarily where by the yellow spots represented equal expression of the target protein in both
channels and red, green colors representing the differentially expressed target proteins in either
of the channels. Further quantification was done by .gal files which were placed over the merged
image spots. The gal files were made as a source to carry information on the spots and
corresponding protein targets. Thus the signal intensity from a spot in the image is transferred as
information on a target protein. Similarly the raw signal intensities form each spot were
calculated and stored as a valued table (.gpr files). Apart from the intensities of the spots the gal
files also transfer the information on intensity of the background, number of pixels standard
deviation and median etc. of both spot and background intensities.
Figure 7. (a) A dual color picture of an antibody microarray slide (only a part) incubated with
the bladder cancer samples. (b) spots showing the differential expression (colors yellow and
green)
31
Materials and Methods
2.2.5 Bioinformatics and statistical methods
2.2.5.1 Data analysis by LIMMA bioconductors
The tables with values from spot intensities (.gpr files) were analyzed using the LIMMA package
of R-bioconductors (Smyth et al., 2005). The resulting data were normalized by the application
of an invariant Loess method developed and described by the usefulness of the available
reference samples (Sill et al., 2010). The sample pools and the individual samples were analyzed
using a one-factorial linear model fitted with LIMMA which resulted in a t-test based on
moderated statistics. Then the resulting p-values were adjusted for multiple testing by controlled
discovery of false proteins according to the Benjamini and Hochberg methods (Smyth, 2004).
The differential abundance of proteins in the samples were presented in log2 fold changes where
highly abundant proteins had log2 values above 0(positive values) and less abundant proteins
had values below 0(negative values). Adjusted p-values were also presented in log form and the
proteins with p-values less than 0.05 were considered significantly abundant between two test
classes. LIMMA bioconductors was also used for further analysis on the differentially abundant
proteins which resulted in volcano plots, box plots and dot plots. Bladder cancer samples and
gastric cancer samples (test set only) were analyzed using LIMMA bioconductors.
2.2.5.2 Methods used in sample classification
After normalization of the raw data, an unsupervised hierarchical cluster of samples was created
to check the various technical and handling artifacts in the data. Then the highly significantly
differentially abundant proteins from the LIMMA analysis were used to build a random forest
classifier (Breiman, 2001). The samples were analyzed by building different test and training sets
by the use of leave-one-out algorithm in a reiterated outer-loop. This was performed by the
application of framework of R called comprehensive package for supervised classification
32
Materials and Methods
(CMA) (Slawski et al., 2008). After a serious of leave-one-out loops, the classification results
were plot in appropriate ROC-curve and AUC plots.
2.2.5.3 Data analysis by Chipster package
A training set of Gastric cancer samples were analyzed using Chipster software package (v1.4.6.
CSC, Finland). After image analysis using GenePix pro 6.0, the tables with values from spot
intensities (.gpr) files were loaded in the Chipster software and analyzed. The intensity ratios
were generated using the median signal intensities with local background of each spot in both red
(DY-649) and green (DY-549) channels. By the application of a Loess method with a
background correction offset [0, 50] for the normexp method, the intensity ratios were
normalized . An unsupervised hierarchical cluster of samples was made with the normalized data
to study and neglect (if any) technical and handling artifacts. The significance in differential
abundance of proteins was tabulated by the application of empirical Bayes test with the p-values
adjusted as per Benjamini-Hochberg method (Smyth, 2004). The empirical Bayes make use of a
moderated t-statistic in which posterior residual standard deviations were applied rather than of
ordinary standard deviations, which gave a far more stable inference when the number of arrays
were small (Smyth, 2004). The proteins with an adjusted p-value of 0.05 and less than that were
considered significantly differentially abundant in the samples. An unsupervised hierarchical
clustering coupled with a boot-strap method was followed to get an optimized signature to
classify the normal and tumor groups. The resulting list consisting signature proteins were then
tabulated with their log fold change and adjusted p-values respectively. The total number of
samples analyzed and classified based on the signature of proteins were then tabulated.
33
Materials and Methods
2.2.5.4 Interaction and pathway analysis
The functions and interactions between the proteins were studied and evaluated using STRING
9.1, open-source software (Szklarczyk et al., 2011). The list of significantly differentially
abundant proteins were loaded into the software online and results reported interactions among
the proteins were evaluated and compared to the pattern of abundance of the proteins in the
sample sets. The functional significance of the various interactions were also evaluated and
compared to the abundance pattern of the proteins. The desired results were then tabulated and
separate figures interpreting the interactions were made. The list of significantly differentially
abundant proteins and their pattern of abundance was loaded into KEGG pathway (Kyoto
Encyclopedia of Genes and Genomes) and the resulting pathways were evaluated (Kanehisa et
al., 2012). The resulting pathways were also used as a reference for detailed analysis of the
abundant proteins.
34
Results
3 Results
3.1 Recurrence prediction in non-muscle invasive bladder cancer
3.1.1 Hierarchical clustering of samples to study incubation batch effect
In the present study, the protein samples from non-muscle invasive bladder cancer were
compared for the identification of a protein signature to predict recurrence. The samples were
analyzed using antibody microarrays constructed as mentioned in sections 2.2.4(1, 2 and 3) in
methods. To also study the possible technical artifacts that may result, four samples were
repeatedly incubated in different batches. As the incubations of the samples on the antibody
microarrays were carried out on different time points, an unsupervised clustering of the samples
was done after normalization of the data. In the hierarchical cluster, the samples were found
equally distributed irrespective of the batch they were incubated on the arrays (figure 8). Most of
the duplicates for example samples 1868, 4131 and sometimes even the triplicates for example
samples 677-1, 2552 were clustered together in the hierarchical cluster (figure 8) referring to the
presence of no technical artifacts such as handling, day of incubation, position of the slides in
slidebooster and incubation of samples on different arrays itself.
35
Results
Figure 8. Hierarchical cluster analysis of the sample set. The Euclidean distance and the average linkage method were used. For
each sample, the identifier and the incubation batch are given. The incubation batches do not cluster. Repeated samples,
however, are located next or very close to each other. Samples whose names start with the letter V represent healthy bladder and
cluster next to each other.
36
Results
3.1.2 Differential expression of proteins between normal bladder and nonmuscle invasive bladder tumors
Protein samples isolated from three normal bladder tissues (healthy), labeled and incubated in
duplicates was compared with protein samples from twenty five non-muscle invasive bladder
tumor tissues to check the differential abundance of proteins. Overall, sixteen proteins (adjusted
p-value less than 0.05) were found to be significantly differentially abundant between tumor and
normal tissues.
Cancer versus healthy controls
Figure 9. Volcano plot summarising expression differences between samples from cancer and
healthy patients. The Log-fold change and respective significance level as an adjusted p-value
are given. The black dots above the red line indicate proteins with significantly (p<0.05)
different expression.
37
Results
A volcano plot containing the details of the differentially abundant proteins is shown in figure 9.
Among the significantly abundant proteins, Tumor necrosis factor receptor superfamily member
10A (TNFRSF10A), Cyclin-dependent kinase inhibitor 1C (CDKN1C) and Lamina-associated
polypeptide 2, Thymopoietin isoform alpha (LAP2A) were significantly very highly abundant in
tumor samples with log2 fold change and adjusted p-value of (0.54, 0.0002), (0.32, 0.007) and
(0.30, 0.0005) respectively. Similarly proteins, Interleukin-4 (IL4), Thyroglobulin (THYG) and
Interleukin-2 (IL2) were significantly less abundant in tumor samples with log2 fold change and
adjusted p-value of (-2.15, 2.87e-13), (-1.87, 2.91e-33) and (-1.58, 2.61e-24) respectively. Two
different antibodies representing TNFRSF10A were found significantly more abundant in the
tumor samples which explain the high specificity and robustness of the antibody microarray. A
table with the log2 fold changes and adjusted p-values of all the twenty significantly
differentially abundant proteins is shown in the supplementary section (table S1).
3.1.3 Differential expression of proteins between recurrent and non-recurrent
non-muscle invasive bladder tumors
To understand the proteins involved in recurrence of non-muscle invasive bladder cancers,
twenty five tumor samples were analyzed with complex antibody microarrays. All the samples
considered for analysis belonged to low grade and stage (Ta and T1) of tumor. Those patient
samples only with a follow-up period of five years after the resection of primary tumors were
considered for analysis. Tumors that recurred within five years post primary resection were
considered recurrent group while those that did not recur within five years were considered to be
cured of bladder cancer. Among the twenty five samples used in the analysis, nineteen samples
had recurred within five years and six samples did not recur and were considered as a group of
non-recurrent tumors or cured. Protein samples were isolated from all the patient tissues along
38
Results
with three normal bladder tissues and were labeled in two fluorescent dyes Cy3 (DY-549) and
Cy5 (DY-649). A common reference pool was made by pooling equal amount of protein from all
the samples for normalization of the data and incubations of the individual samples were
performed in presence of the reference pool for the same purpose. All the samples were
incubated on the antibody microarray consisting eight hundred and thirteen antibodies directed
against seven hundred and twenty four cancer related proteins. Incubations of random samples
were performed in different days and time but without any technical and handling variations
(figure 8).
Among the seven hundred and twenty four antibodies detected in the array, two hundred and
fifty five proteins were significantly differentially abundant among the recurrent and nonrecurrent group (table S2). A hundred and five proteins among the significantly differentially
abundant were more abundant in recurrent tumors while remaining hundred and fifty were found
lesser in abundance in the group. A hundred and two proteins among the two hundred and fifty
five were very highly significantly differentially abundant with an adjusted p-value less than
0.003. The results from the whole analysis were exhibited as a volcano plot most of the proteins
named with the log fold change and adjusted p-values (figure 10). The individual protein
expression variations were studied to assess the significance of findings from individual patients.
A box plot was made for some of the proteins and is represented in figure 11. Lamin-A (LMNA)
and transcription factor AP-1 (JUN) were the most abundant proteins in recurrent tumors while
L-Selectin (LYAM1) and Cyclin-Dependent Kinase Inhibitor 1A (CDKN1A) were found with
very less abundance in recurrent tumors (figure 11).
39
Results
Recurrence versus no-recurrence
Figure 10. Volcano plot of differentially abundant proteins. Horizontally, the degree of
expression variation is shown; the vertical axis indicates the significance level. The black dots
represent the proteins that were analysed. The red line stands for an adjusted p-value of 0.003.
All proteins located above this threshold are highly significantly informative about the
recurrence status of a patient. Some particular proteins are named.
40
Results
Figure 11. Distribution of protein expression in all patient samples. Data are shown for proteins
LMNA and JUN (up-regulated in non-recurrent tumours) as well as LYAM1 and CDN1A
(down-regulated in non-recurrent tumours).
41
Results
3.1.4 Accurate prediction of recurrence in non-muscle invasive bladder cancer
by protein signature
A multivariate analysis was performed to assess the prognostic capabilities of the highly
significantly differentially abundant proteins. A rule for accurate classification of the samples
was established based on random forest method (Breiman, 2001) combined with the twenty
proteins that were most significantly differentially abundant. The training set was classified with
100% sensitivity and specificity. Leave-one-out cross validations were performed to assess the
efficient transferability of the classification to other test sets. The classification results for the
different test sets in the cross validation steps were summarized as a receiver operating
characteristics (ROC) curve (figure 12). The miss-classification rate for the cross-validation was
20% (SD 0.08). Also the area under the curve (AUC) value of 0.91 exhibits the extremely high
accuracy of discriminating between the patients with and with-out recurrence (figure 12). That
corresponds to the overall sensitivity of 80% at 100% specificity. Twenty proteins that were
highly significantly differentially abundant based on which the samples were classified is shown
in table 1 along with their corresponding log fold change and adjusted p-value. A strong study on
already published peer reviews connecting these twenty abundant proteins is also shown in the
table 1.
42
Results
Figure 12. Receiver operating characteristic (ROC) curves resulting from random forest
classification. Discrimination within the trainings set was 100% (dotted line). Stringent leaveone-out cross validation yielded an AUC value of 0.91 (black line).
43
Results
Protein Uniprot
entry name
Uniprot
accession
Adjusted
p-value
Log fold
change
LMNA_HUMAN
P02545
3.3e-09
0.72
YBOX1_HUMAN
P67809
1.9e-08
0.52
JUN_HUMAN
P05412
1.3e-07
0.50
AKT3_HUMAN
Q9Y243
2.5e-07
0.48
YETS2_HUMAN
Q9ULM3
5.2e-06
CADH1_HUMAN
P12830
TIA1_HUMAN
Reported studies
RC
C
RBC
BC
RC
BC
RC
(Konstantakou et al., 2009;
Stravopodis et al., 2009)
(To et al., 2010)
(Gluz et al., 2009)
(Mitra et al., 2009)
(Ling et al., 2011)
(Ouyang et al., 2008)
(Ching and Hansel, 2010)
(Bonin et al., 2008)
0.32
-
No cancer related reference
1.2e-05
0.33
RBC
(Gallagher et al., 2008; Negraes
et al., 2008)
P31483
2.1e-07
-0.39
C
(Alvaro et al., 2005)
SMAD3_HUMAN
P84022
2.1e-07
-0.58
C
(Penuelas et al., 2009; Poncelet
and Schnaper, 2001)
PABP1_HUMAN
P11940
2.1e-07
-0.36
C
(van Duin et al., 2005)
CDN1A_HUMAN
P38936
3.4e-07
-0.52
RBC
BC
(Shariat et al., 2008)
(Shariat et al., 2007)
LYAM1_HUMAN
P14151
9.6e-07
-0.51
C
(St Hill, 2011)
AKTIP_HUMAN
Q9H8T0
5.4e-06
-0.34
C
(Cinghu et al., 2011)
PRI1_HUMAN
P49642
1.0e-05
-0.35
C
(Yotov et al., 1999)
HSP7C_HUMAN
P11142
1.0e-05
-0.34
-
No cancer related reference
RSSA_HUMAN
P08865
1.1e-05
-0.35
C
(Qiu et al., 2008)
GRM1A_HUMAN
Q96CP6
1.1e-05
0.28
C
TPA_HUMAN
P00750
1.1e-05
0.23
BC
ZBT17_HUMAN
Q13105
1.1e-05
-0.54
C
LAMP2_HUMAN
P13473
1.1e-05
0.26
C
JUN_HUMAN
P05412
1.3e-05
0.48
RBC
BC
RC
BC
(Martino et al., 2012)
(Speyer et al., 2012)
(Louhelainen et al., 2006)
(Knowles et al., 1993)
(Ikegaki et al., 2007)
(Iraci et al., 2011)
(Lee et al., 2012)
(Tung et al., 2010)
(Mitra et al., 2009)
(Ling et al., 2011)
(Ouyang et al., 2008)
Table 1. List of the twenty proteins with the most significant expression variations between
recurrent and non-recurrent tumours. Log fold changes in expression and the related adjusted pvalues are shown. Also, literature is listed in which the protein or respective gene was reported in
connection with bladder cancer (BC), recurrence of bladder cancer (RBC), recurrence of other
cancer forms (RC), or cancer overall (C).
44
Results
3.1.5 Functional annotation of highly significantly abundant proteins
The possible functional aspects between recurrence and non-recurrence of the samples were
studied based on the functional studies on the significantly abundant proteins using a web-based
analysis platform DAVID (Huang da et al., 2009). About 56% proteins among the highly
abundant in recurrent tumors were expressed inside the cells while 61% of the proteins among
those that were less abundant were secreted by the cells or expressed in the extracellular space.
Web-based pathway analysis program KEGG (Kanehisa et al., 2012) showed a strong
suppression in the TGF-beta signaling pathway in the recurrent cancer when the significant
proteins were tested using the program (figure 13). The expression of important factors like
INFG, TNFA, TGFB and THBS1 was less in recurrent samples while inhibitor of MAPK3
(known as ERK1) was highly abundant and again Mothers against Decapentaplegic Homolog
proteins like SMAD1, 2 and SMAD3 were significantly less abundant in recurrent samples.
Transcription factor SP1, an important downstream signaling protein was also less abundant in
recurrent samples. Along with TGFB pathway, proteins involved in apoptosis were also strong
regulated. Among those more abundant proteins in recurrent samples were effectors Caspase 3
(CASP3) and 9 (CASP9), both in their active form and the regulator Bcl2 associated –X (BAX)
(table S2). Interestingly, the Trail-receptor which could active all the above proteins was also
high in abundance in recurrent samples. On the other hand, FAS was less abundant in recurrent
samples. Similarly some of the transcription factors previously studied in carcinogenesis and
more widely in other diseases were also differentially abundant in recurrent samples. Variations
in regulation of most of the cancer related proteins in cancer pathways were colored and given as
a supplementary (figure S1, S2).
45
Results
Figure 13. KEGG analysis of the TGF-beta signalling pathway. Affected proteins are labelled in green or red
if their expression was lower or higher, respectively, in recurrent rather than non-recurrent cancer.
46
Results
The pathways in signaling and cancer from KEGG analysis most often highlighted those proteins
involved in apoptosis and cell proliferation. This fact was also identified by the studies on gene
ontology as functions of many abundant proteins in recurrent samples were representing both
pro-apoptotic and anti-proliferative effects. In addition to pathways and analysis of proteins
based on their ontology, important proteins like transcription factors, cell cycle regulators, those
among the abundant proteins were also analyzed by web-based protein-protein interaction
software STRING 9.1 (Szklarczyk et al., 2011). Among the highly abundant proteins in recurrent
tumors, transcription factor AP-1 (JUN), Nuclear factor NF-kappa B1 (NFκB1), DNA Topo
isomerase-2 alpha (TOP2A), Krueppel-like factor (KLF5) and Cadherin-1 (CDH1) were
interacting among each other by protein-protein binding and activation of expression (figure 14).
The less abundant proteins in recurrent tumors like Mothers against Decapentaplegic Homolog-3
(SMAD3), Matrix Metalloproteinase-1 (MMP1), Tissue Inhibitor of Metalloproteinases 1
(TIMP1), Vascular Endothelial Growth Factor A (VEGFA) and transcription factor SP1 were
reported to interact among each other by activation, inhibition and binding (figure 14). Among
the interactions, JUN a highly abundant protein was studied earlier to inhibit TIMP1, a less
abundant protein in recurrent tumors. Similarly, many such interactions were reported from the
analysis and studied vastly through peer reviewed publications.
47
Results
Figure 14. Protein interactions of strongly regulated proteins. Analysis of the protein expression
data with the pathway analysis software STRING 9.1 revealed several protein interactions.
Proteins coloured in red were higher expressed in recurrent than non-recurrent tumours; greencoloured proteins exhibited lower expression.
48
Results
3.2 Protein biomarker identification in Gastric adenocarcinoma
3.2.1 Hierarchical clustering of samples after normalization
In the current study, the protein samples from normal gastric tissues and high stage and grade
gastric adenocarcinoma tissues were compared for the identification of protein biomarkers for
effective diagnosis and prognosis. The samples were analyzed using antibody microarrays
constructed as mentioned in sections 2.2.4(1, 2 and 3) in methods. To also study the consistency
of the expression, all samples were incubated in duplicates on the arrays. The samples were
incubated on the antibody microarrays in a single batch to minimize the technical artifacts that
may arise. An unsupervised clustering of the samples was done after normalization of the data.
In the hierarchical cluster, the samples were found clustered according to the type of the tissue
(figure 15). All the duplicates N,DN and T,DT were clustered together in the hierarchical cluster
(figure 15) referring to the presence of no technical artifacts such as handling, position of the
slides in slidebooster and incubation of samples on different arrays itself. These results suggested
that the clustering and segregation of normal and tumor samples were only based on the
differential abundance of proteins in them and not any other artifacts. After normalization the
data were analyzed for differential abundance of proteins.
49
Results
Figure 15. Hierarchical cluster analysis of the sample set. The Euclidean distance and the
average linkage method were used. The samples names are given, normal (N, DN) and tumor (T,
DT). The incubation was done on the same day. The duplicates of the samples were clustered
apart from the cluster groups of normal and tumor groups.
50
Results
3.2.2 Differential expression of proteins between normal gastric tissues and
gastric adenocarcinoma tissues
To identify protein biomarkers for effective diagnosis of gastric adenocarcinoma, twenty five
surgically removed tumor tissue samples along with adjacent normal tissue samples from
patients diagnosed with gastric adenocarcinoma were analyzed with complex antibody
microarrays. All the samples considered for analysis belonged to high grade and stage (T3 and
T4) of tumor. Those patient samples only without prior treatment for gastric cancer were
considered for analysis. Among the twenty five pair (normal and tumor) of tissues used in the
analysis, ten pairs were considered as training set and fifteen pairs were considered as test set.
Protein samples were isolated from all the tumor and normal tissues and were labeled in two
fluorescent dyes Cy3 (DY-549) and Cy5 (DY-649). For incubations of the training set of
samples, a common reference pool was made by pooling equal amount of protein from all the
samples for normalization of the data and incubations of the individual samples were performed
in presence of the reference pool for the same purpose. The test sets were incubated separately
by swapping the dyes labeled and nature of the samples and vice versa. All the samples were
incubated on the antibody microarray consisting eight hundred and thirteen antibodies directed
against seven hundred and twenty four cancer related proteins.
3.2.2.1 Differential expression of proteins- training set of samples
Among the seven hundred and twenty four antibodies detected in the array, eight proteins were
significantly differentially abundant between normal and tumor samples (table 2). Three proteins
were significantly differentially more abundant in tumor samples while remaining five proteins
were found lesser in abundance. These proteins were very highly significantly differentially
abundant with an adjusted p-value less than 0.04 (table 2). A heat-map featuring the eight highly
51
Results
differentially abundant proteins is represented in figure 16. Nucleolysin-TIA1 (TIA1) was found
highly significantly differentially more abundant in the tumor samples with a log fold change of
1.19 and an adjusted p-value of 0.04. While, Mucin-6 (MUC6) was found very less in abundance
in tumor samples with a very high significance (log fold change -2.45 and adjusted p-value 9e06). Gamma-Enolase (ENOG), Epidermal growth factor (EGF) and Folate receptor alpha
(FOLR1) were also among the highly regulated proteins (figure 16).
Protein Uniprot
entry name
Uniprot
accession
Adjusted pvalue
Log fold
change
EGF_HUMAN
P01133
0.04
-0.30
ENOG_HUMAN
P09104
0.02
0.28
IRS2_HUMAN
Q9Y4H2
0.03
-0.26
MPIP2_HUMAN
P30305
0.04
-0.24
FOLR1_HUMAN
P15328
0.04
0.53
GBRB1_HUMAN
P18505
0.04
-0.28
MUC6_HUMAN
Q6W4X9
9e-06
-2.45
TIA1_HUMAN
P31483
0.04
1.19
Table 2. List of the eight proteins with the most significant abundance between normal and
tumor samples in the training set. Log fold changes in expression and the related adjusted pvalues are shown.
52
Results
Figure 16. Heat map of proteins with significant differential abundance in normal and tumor
samples (also in duplicates). Highly abundant proteins were represented in red while proteins
with less abundance were represented in blue. Only proteins with an adjusted p-value less than
0.05 were represented in the heat map. The sample set here is considered to be the training set.
3.2.2.2 Differential expression of proteins- test set of samples
Incubations of fifteen pairs (normal and tumor) samples were carried out by dye swap method.
Normal sample labeled with Cy3 from a patient is incubated with tumor sample labeled with Cy5
from the same patient. Alternatively, duplicates from the same samples were incubated in the
opposite manner by labeling tumor samples in Cy3 and normal samples in Cy5. Similar steps
were followed for incubations of all the thirty samples. Among the seven hundred and twenty
four proteins detected in the array, ninety one proteins were found significantly differentially
abundant between normal and tumor samples, out of forty two proteins were more abundant in
tumor samples and forty nine proteins were less abundant in them. About sixteen proteins were
highly significantly regulated with an adjusted p-value less than 0.03. Many differentially
abundant proteins among the training set samples were also found differentially abundant in the
test set samples. Among the most significantly abundant proteins, Calgranulin B (S100A9),
Insulin-like growth factor-binding protein 7 (IGFBP7) and Calponin-2 (CNN2) were found more
abundant in tumor samples while Mucin 6 (MUC6) was less in abundance in tumor samples.
53
Results
Uniprot
accession
Adjusted pvalue
Log fold
change
P29972
0.002
1.74
CD27_HUMAN
P26842
0.02
3.45
CNN2_HUMAN
Q99439
1.4e-07
1.84
ENOG_HUMAN
P09104
0.03
1.04
IBP7_HUMAN
Q16270
6.13e-06
2.02
IFNG_HUMAN
P01579
0.004
1.74
IL10_HUMAN
P22301
0.002
2.77
ITA5_HUMAN
P08648
0.002
2.21
OCLN_HUMAN
Q16625
0.03
0.57
S10A9_HUMAN
P06702
2.31e-05
2.01
TFPI2_HUMAN
P48307
0.01
2.51
TIA1_HUMAN
P31483
0.001
0.67
VEGFA_HUMAN
P15692
0.002
2.88
CKS2_HUMAN
P33552
0.004
-1.83
DKK3_HUMAN
Q9UBP4
0.03
-0.85
MUC6_HUMAN
Q6W4X9
1.88e-07
-3.71
Protein Uniprot
entry name
AQP_HUMAN
Table 3. List of the top sixteen proteins with the most significant abundance between normal and
tumor samples in the test set. Log fold changes in expression and the related adjusted p-values
are shown. Three proteins (colored green), ENOG, TIA1 and MUC6 were found significantly
differentially abundant in both the training and test sets of samples.
Three proteins found highly regulated in the training set samples were also found among the
highly differentially abundant proteins in tumor samples of the test set. ENOG and TIA1were
found more abundant among the highly regulated proteins in the test set samples while MUC6
was found less abundant in them. Most significantly differentially abundant proteins were listed
54
Results
along with their log fold change and adjusted p-values in table 3. The results from the whole
analysis were exhibited as a volcano plot most of the proteins named with the log fold change
and adjusted p-values (figure 17).
Cancer versus healthy controls
Figure 17. Volcano plot of differentially abundant proteins. Horizontally, the degree of
expression variation is shown; the vertical axis indicates the significance level. The black dots
represent the proteins that were analysed. The pink dot represents TIA1 which was also found
differentially abundant in the training set. Some other particular proteins are named.
55
Results
The individual protein expression variations were studied to assess the significance of findings
from individual patients. To test differential abundance of proteins and specificity of their
regulation in normal and tumor tissues, a western blot analysis was carried out. Two protein
pools were made by random selection of normal and tumor samples. These two pools of normal
(Normal 1 and 2) and tumor protein (Tumor1 and 2) were resolved in 10% and 15% SDS-poly
acrylamide gels and transferred to nitrocellulose membranes as mentioned early in methods
section 2.2.2.4. Antibodies against human TIA1 and S100A9 were incubated on the membranes
to detect the expression of respective proteins in normal and tumor samples. Antibody against
human glyceraldehyde-3-phosphate dehyrogenase (GAPDH) was used as a loading control. The
differential abundance of these proteins is exhibited visually in figure 18a and 18b.
a)
b)
Figure 18. Immunoblot analysis of expression of proteins TIA1 and S100A9. Normal 1 and 2
are pools of proteins made by pooling random normal samples. Tumor 1 and 2 are pools of
proteins made by pooling random tumor samples. Anti-GAPDH conjugated with horse-radish
peroxidase against human is used as a loading control. a) TIA1 found more abundant in tumor
samples as compared to normal. b) S100A9 was also found in more abundance in tumor samples
as compared to normal.
56
Discussion
4 Discussion
4.1 Cancer biomarkers identification by antibody microarrays
Cancer as a disease is too difficult to manage as it is life-threatening when it affects many vital
organs in humans. There is a need for early and non-invasive diagnosis of the disease in one
hand and to minimize the patient’s psychological and financial stress on the other. An effective
diagnosis will provide ways to not only reduce the false positive cases but also the strategies for
effective therapy (Hartwell et al., 2006). Gold standard diagnostic methods using both genomic
and proteomic methods were studied and developed extensively in the past decade. Proteomics
as it deals with proteins, the biological end-product and acting-product in a complex system
provided further more details in the mechanics of disease occurrence and recurrence (Alhamdani
et al., 2009). Two-dimensional gel electrophoresis and Mass-spectrometry were two proteomic
methods, extensively used with vast limitations on sensitivity and reproducibility and
optimization.
The use of complex affinity protein-array technologies improved the limitations with other
proteomic methods and increased the relevance of underlying molecular mechanisms such as
protein-protein interactions, binding and activation of expression etc. (Alhamdani et al., 2009).
These systems not only improve the specificity, sensitivity and robustness but also operate with
minute probe and sample requirements mostly in picolitre and femtolitre (Alhamdani et al.,
2009). Like DNA microarrays, the antibody microarrays were also printed on a solid support
(often microscopic slides) coated with chemical binders on the surface (Wingren et al., 2007).
Antibody microarrays being a novel technology have evolved from very less number of probes
being used (between twenty and two hundred previously) to large numbers (eight hundred in our
57
Discussion
case) (Schroder et al., 2010; Wingren et al., 2007). Experimental procedures like sample
preparation, labeling and incubation procedures were optimized regularly (Alhamdani et al.,
2009; Alhamdani et al., 2010). The selection of antibodies is based on the previous results from
other high-throughput techniques with a vast coverage of previous studies on cancer biomarkers
identification. Cancers often studied are pancreatic, breast and colon in comparison to their
respective healthy controls. Eight hundred and ten antibodies raised against seven hundred and
forty one cancer-related proteins are spotted on glass slides coated with epoxysaline. Six hundred
and sixty eight affinity-purified polyclonal antibodies from rabbit were provided by Eurogentec
and other hundred and forty two were purchased commercially and provided by collaborating
partners. All these antibodies are immobilized on the slide surface in duplicates for consistency
of the results. 10 µg of antibody is used for producing 1000 microarrays which is much lesser
than the amount spent on various ELISA experiments. In case of two dimensional gel
electrophoresis, samples are denatured and iso-electrically separated for an effective analysis.
Furthermore, samples are depleted in mass-spectrometry experiments. Samples incubated on the
antibody microarrays are maintained in their native state (Alhamdani et al., 2009; Schroder et al.,
2010). All these advantages provide more native and specific at the same time sensitive analysis
of samples (Alhamdani et al., 2009; Schroder et al., 2010).
Above all, as mentioned earlier, the whole method deals with proteins, the biological endproduct. Therefore, it seems that antibody microarrays can be implemented in effective
biomarker identification for diagnosis and therapy of various cancers at proteome level. In
reference to the above claims, two cancers (bladder and gastric) were analyzed using above
developed antibody microarray.
58
Discussion
4.2 Non-muscle invasive bladder cancer
Non-muscle invasive bladder cancer, being a low grade, low stage tumor with high probability
of recurrence is very important to study as the prognosis and prediction of recurrence can have
several clinical consequences. The current study uses complex antibody microarrays to look out
for molecular variations to predict recurrence at protein level since they are responsible for most
of the cellular activities and thus representing the best functionality relevant differences. Many of
the earlier studies used body fluids from both low and high grade tumors for an effective
diagnosis and prognosis of the disease in patients with bladder cancers. Some of them even
implemented proteomic methods. Antibody microarrays were implemented to profile serum
samples from bladder cancer patients (Sanchez-Carbayo et al., 2006). The study revealed not
only the differentiation of healthy and cancer patients but also a four protein signature to separate
patients into high and low risk groups. The study used mere hundred and fifty antibodies on the
array which were previously studied in bladder cancers. A mass-spectrometry based study was
done by analyzing the urine samples of patients diagnosed with muscle invasive bladder cancer
(Schiffer et al., 2009). The analysis could predict muscle invasive bladder cancer with a
sensitivity of 92% and specificity of 68%.
Most of the proteomic-based studies in bladder cancers were concerned with both low and high
grade cancers, while we focus only on the low grade and low stage cancers. Another interesting
aspect of our study is the follow-up time. Although a five-year follow-up limited the sample size,
we insisted on it as the recurrence history of most of the studied non-muscle invasive bladder
tumors were within five years of the follow up. The proteome analysis is carried on the proteins
from the tissue extracts of primary tumors diagnosed with non-muscle invasive disease since
patient prognosis depends strongly on the likelihood of local recurrence. As regular transurethral
59
Discussion
resections of tissues are performed upon during initial diagnosis of the cancer, enough patient
material was available for the whole analysis.
4.2.1 Differential expression of proteins between recurrent and non-recurrent
non-muscle invasive bladder tumors
The current study on non-muscle invasive bladder cancers with long term follow-up on
recurrence identifies two hundred and fifty five proteins (figure 10) with significant differential
abundance which is surprising as most of the target proteins were selected on the basis of studies
from different cancers in comparison to their related healthy tissue. High number of proteins
regulated between two tumor types is very interesting to study along with their functional
annotation. These very high numbers of variations in proteins levels indicate massive differences
in the biology of recurrent and non-recurrent tumors. As proteins prominently studied in bladder
cancers are also differentially regulated, the results are easy to compare with the available
literature and to identify new biomarkers along with the already studied ones (table 1). More
importantly, many such proteins those are significantly differentially more abundant in recurrent
tumors are known key players in vital functions in oncogenesis. At the same time, many
abundant proteins were either studied vastly in high grade bladder cancers or other cancers as
potential therapeutic targets. These previous studies strongly encouraged carrying on further
bioinformatics analysis on pathways involved, interactions and disease related modifications.
Proteins, JUN, TOP2A and NFκB1 were often studied in various types of cancers. The top
twenty significantly differentially abundant proteins were tabulated with some of their
respectively studied literature (table 1).
60
Discussion
Among the significantly highly abundant proteins in recurrent tumors, JUN, a well known
transcription factor and a proto-oncogene was extensively studied previously. Elevated levels of
JUN had been identified as a prognostic marker for high risk and recurrence of prostate cancer
(Ouyang et al., 2008). Few studies in transcriptional level in different cancers also identified
JUN as a prognostic factor for high risk and disease progression. A transcriptional analysis to
predict survival and recurrence in bladder cancer samples identified a four-transcript signature
including JUN. The analysis also predicted bad outcome in correlation with high abundance of
JUN (Mitra et al., 2009). The knockdown of phospholipase Cε affected the expression of JUN, in
turn negatively regulating the proliferation of BIU-87 bladder cancer cells (Ling et al., 2011).
Among the significantly less abundant proteins in recurrent tumors, SP1, a transcription factor
that binds to gc-rich motifs with high affinity to regulate the expression of vital genes involved in
cell growth, apoptosis and differentiation was well studied in high grade cancer including muscle
invasive bladder cancer. SP1 was identified as a potential therapeutic target in high grade high
stage muscle invasive bladder cancers. Small drug molecules targeting specificity proteins (SP1,
SP2 and SP3) decreased their expression, inhibited the cell proliferation and tumor cell growth in
muscle invasive bladder cancers (Chadalapaka et al., 2010; Chadalapaka et al., 2008; Jutooru et
al., 2010). It is quite fascinating that the abundance of two transcription factors already studied in
bladder cancers appeared differently in the current study. The studies from literature correlate
with their abundance in recurrent non-muscle invasive tumors. From the above studies, one can
hypothese that different mechanisms by transcription factors JUN and SP1 account for
recurrence and progression of non-muscle invasive bladder respectively. The current study any
way is not detailing the functional aspects of the differentially abundant proteins.
61
Discussion
4.2.2 Apoptotic proteins and recurrent non-muscle invasive bladder cancers
Interestingly, pro-apoptotic proteins like active or cleaved CASP3, CASP9 and BAX are found
more abundant in the recurrent non-muscle invasive bladder cancers (figure 10). CASP3 levels
were known to be low in high grade muscle invasive bladder cancers (Karam et al., 2007). In
other cancer entities, CASP3 over-expression was associated with recurrence and high cancer
mortality rates (Huang et al., 2011; Jonges et al., 2001; Konstantinidou et al., 2007). CASP3, a
cysteine-aspartic acid protease that activates a cascade of pro-apoptotic proteins and in turn
apoptosis, is also known for its non-apoptotic roles such as differentiation, dedifferentiation and
activation of immune cells (Fujita et al., 2008; Kennedy et al., 1999; Szymczyk et al., 2006).
Active CASP3 had been reported as a prognostic marker in colorectal cancers with combined
activation and over- expression of CD57 (Jonges et al., 2001). The increase in recurrence of
apoptotic tumors after radio or chemotherapy was illustrated by the ability of active CASP3 to
repopulate tumor cells through prostaglandin E2 (Huang et al., 2011). Apoptotic tumor cells
leads to an increase in the levels of arachidonic acid in the extracellular space through CASP3mediated calcium-dependent phospholipase A2 (iPLA2) activation. This activates of
prostaglandin E2 through prostaglandin synthase H (PGH2), a downstream product of
arachidonic acid and is known to be a key regulator of tumor growth. Apart from the above
observations, Huang et al could also correlate higher levels of active CASP3 with high
occurrence of tumor recurrence in head and neck squamous cell carcinoma and breast carcinoma
which also supports our observations. The current study identifies both active form of CASP3
and CASP9 along with PGH2 more in abundance in the recurrent non-muscle invasive tumors.
As the pattern of abundance correlate with the above observations, it can be proposed that a
62
Discussion
similar mechanism of active CASP3 apoptosis-mediated tumor cell repopulation via PGH2 is
seen also in recurrence of non-muscle invasive bladder cancers.
4.2.3 Functional annotations by pathway analysis and interactive studies
Apart from the apoptotic-regulatory proteins, a detailed functional annotation on the significantly
differentially abundant proteins reveals a strong inhibition in the TGFβ signaling pathway in the
recurrent non-muscle invasive tumors. The activators of the signaling pathway like IFNγ, TNFα,
TGFβ and SMAD family of proteins along with transcription factor SP1 are found less abundant
in recurrent tumor samples, while on the other hand MAPK3 ( also known as ERK1) is high in
abundance (figure 13). Similar repression of TGFβ/SMAD signaling were previously observed
in different cancer entities and the repression was strongly associated with either a high risk for
recurrence of the disease or a poor outcome (Alazzouzi et al., 2005; Xie et al., 2003). Further
detailed analysis using online bioinformatics tools like KEGG, DAVIDGO and STRING 9.1
provide more insights on protein interactions and activation or repression of expression (figure
14). Among the highly significantly more abundant proteins in recurrent tumors, the interactions
among JUN, NFκB1, KLF5, TOP2A and CDH1 are extensively reported in studies on
oncogenesis and other disease conditions. TOP2A, a nuclear enzyme involved in processes such
as chromosome decondensation, chromatid separation and relief of torsional stress that occurs
during transcription and replication of DNA was observed to interact directly with JUN which
increases the decatenation activity of TOP2A on DNA (Kroll et al., 1993). JUN was also
observed to interact with NFκB1, a transcriptional regulator that is activated by various intraand extra-cellular stimuli like cytokines, oxidant-free radicals and viral products. When
activated, NFκB1 is translocated into the nucleus to stimulate the expression of genes involved in
a wide variety of biological functions. The interaction between JUN and NFκB1 in endothelial
63
Discussion
cells was exhibited to regulate the expression of vascular cell adhesion molecule-1 (VCAM1).
This activation is an important feature of the initial steps of pathogenesis in atherosclerosis
(Ahmad et al., 1998). KLF5, a transcription factor that binds to gc box promoter elements,
localized to the nucleus binding the EGF response element was observed interacting with
inflammatory response element NFκB1. Platelet derived growth factor-A (PDGFA) gene
expression was exhibited to be activated by the interaction of p50 subunit of NFκB1 and KLF5.
The interaction forms a protein complex on the binding site of KLF5 on the promoter chains of
PDGFA (Aizawa et al., 2004). CDH1, a calcium dependent cell-cell adhesion glycoprotein
whose loss of function is associated with cell proliferation and invasion is inhibited by NFκB1.
NFκB1 is observed to suppress the expression of CDH1and induce the expression of vimentin, a
mesenchymal specific gene. This induction leading to the epithelial to mesechymal transition of
mammary cells is exhibited in breast cancers (Chua et al., 2007).
Significantly less abundant proteins in recurrent non-muscle invasive tumors are observed
interacting among themselves, especially proteins such as SP1, SMAD3, VEGFA and TIMP1 are
often reported in previous studies on cancer entities (figure 14). VEGFA, a growth factor
activating angiogenesis and endothelial cell proliferation interacted with SP1 in pancreatic
adenocarcinoma cells through IRS2, a downstream molecule of IGF 1 receptor signaling
pathway. This activation of expression of VEGFA by SP1 causes pancreatic cancer cell
proliferation (Neid et al., 2004). Induction of VEGFA expression by SP1 when associated with
P38 kinase was also observed in cardiomyocytes (Lin et al., 2011). TIMP1 expression was
effected by SP1 in human embryonic kidney cells which leads to uncontrolled proliferation of
them (Lee et al., 2004). Combined over-expression of SMAD3 and SP1 in glomerular cell
cancers induced the promoter activity of α2 (I) collagen gene (COL1A2) that lead to the
64
Discussion
activation of TGFβ signaling pathway (Poncelet and Schnaper, 2001). These interactions
reported and observed in various disease conditions explain their biological significance and
vitality in not only onceogenesis but also general growth regulating mechanisms. It can be easily
conceivable that interactions among both highly abundant and less abundant proteins in recurrent
non-muscle invasive tumors are directly or indirectly responsible for disease recurrence or
progression.
4.2.4 Recurrence prediction in non-muscle invasive bladder cancers
The current study reveals a 20 protein signature to predict the recurrence of cancer with very
high accuracy (figure 12). The available methods for prediction of recurrence appear to be very
inferior to the high accuracy in sensitivity and specificity of this 20 protein signature. The protein
signature could predict recurrence with 80% sensitivity and 100% specificity as compared to the
available approved NMP22 assay for bladder cancer diagnosis (Grossman et al., 2005) which
exhibits a sensitivity of merely 55.7% and a specificity of 85.7%. Though, the number of
samples analyzed appears to be very less for such a claim, the five year follow-up time appears
clinically significant. The sample limitations can also be rectified by some non-invasive analysis
on body fluids as most of the proteins in the signature are known to be secreted. The goldstandard diagnostic methods in clinical settings including immuno-histo chemistry are all
immuno-based assays. Antibody microarray, as itself is an immuno-based assay has the great
potential to be used in clinical situations. Furthermore, these marker proteins can be translated
easily to standard immuno-based assay platforms that are routine in use in clinical settings. On a
broader outlook, such a test for prediction of recurrence can adjust the treatment regimen and
rigidity of surveillance, especially in bladder cancers where patients were followed up for years
65
Discussion
using cytoscopy methods. This can not only reduce surveillance costs but also improve the
patients’ outcome substantially.
4.3 Biomarker identification for gastric adenocarcinoma by
antibody microarrays
Far too many studies emphasize the role of different regulators of cell cycle, cell growth,
proliferation, angiogenesis and apoptosis in gastric cancer formation and evaluate their potential
as a valuable therapeutic target. Various methods are being employed to study the molecular
basis of gastric adenocarcinoma, but only few evolve till the treatment strategies. Despite recent
developments in high-throughput methods like signature based molecular diagnostics, mass
spectrometry (MS) based proteomic methods and next generation sequencing, availability of
suitable biomarkers at protein levels often, is disappointing. Either occurrence of high false
positive cases or controversial published data on the role of molecular markers restricts, not only
the implementation of many markers but also fails to address the response due to inter-individual
variability (Pietrantonio et al., 2013). DNA microarray, as a powerful high-throughput technique
only identifies novel genes which as biomarkers, fail at protein level when conventional
immuno-based assays such as immuno-histochemistry (IHC) and ELISA were used (Zheng et al.,
2004). Advances in large-scale gene expression profiling combined with network analysis
identified several biomarkers without much clinical relevance on the stage dependent expression
of the genes. Moreover, these markers do not predict the progression and patient outcome which
is again important for an optimal treatment strategy (Takeno et al., 2008). Apart from DNA and
RNA microarrays, tissue microarrays are good tools for protein level biomarker identification
with much limitations in number of molecules studied (Senapati et al., 2008). Individual analysis
66
Discussion
on cytotoxic molecules predicts acute gastric mucosal lesions (Suzuki et al., 2003). Similar
analysis on other molecules involved in key biological processes, showed their significance as
prognostic indicators in gastric adenocarcinomas but with much inconsistence disallowing them
from being recognized as standard biomarkers (Kodama et al., 2008; Leung et al., 2004). So far,
only antibody microarray based study on gastric adenocarcinomas deals with cancers associated
with Helicobacter pylori infection. The study composed of mere hundred and twenty seven
antibodies against immunoregulatory antigens. Apart from identifying plenty of previously
reported proteins, the analysis also identified protein signatures associated with tumors and
bacterial infection (Ellmark et al., 2006).
We used complex antibody microarrays with eight hundred and ten antibodies described earlier
(section 4.1) to analyze high grade, high stage (most even had lymph node metastasis) gastric
adenocarcinoma samples. To get more insight on a personalized treatment strategy point of view,
gastric adenocarcinoma tissues along with healthy tissue controls from the same patients are
analyzed. The analysis comparing twenty five healthy controls with same number of gastric
adenocarcinomas (both training and test sets) reveals several interesting candidates.
4.3.1 Differentially regulated proteins in gastric adenocarcinoma
Two different analyses on different sample sets identify ninety eight significantly differentially
abundant proteins between cancer and healthy controls (table 2, 3 and figure 16, 17). Along with
a lot of newly observed proteins, few proteins already described in gastric oncogenesis are
identified from the analysis. Interestingly, five proteins are differentially abundant in both the
training and test set of samples among which three are in the most significantly abundant
proteins (table 3). Strikingly, proteins involved among highly invasive cancers including gastric
adenocarcinomas are also present among the significantly abundant proteins. These results
67
Discussion
triggered our interest for further analyses on the proteins with literatures. Proteins like AQP1,
CNN2, TIA1, OCLN, S100A9, VEGFA and DKK3 best described oncogenesis.
AQP1
Aquaporin 1 (AQP1), a water channel protein facilitates water flux across cell membranes is
found highly significantly more abundant in gastric adenocarcinoma tissues. Belonging to a
family of small integral membrane proteins, they physically resemble channel proteins and
highly abundant in erythrocytes and renal tubes. Also found in abundance in epithelial and
endothelial cells, they are associated with cell migration, metastasis and angiogenesis. It is
previously reported to be over-expressed in various human malignancies (Hu and Verkman,
2006). AQP1 is proposed as a biomarker of early diagnosis of renal cancer on the analysis of
urine samples from patients with renal cancer (Morrissey et al., 2010). AQP1 in protein level is
a poor prognostic factor in basal-like breast carcinomas as occurrence of death in patients
suffering from breast cancers directly correlated with the high abundance of AQP1 (Otterbach et
al., 2010). To our interest, this protein is not previously reported in gastric adenocarcinomas.
CNN2
Calponin 2 (CNN2) plays a role in smooth muscle contraction and cell-cell adhesion. It is found
highly significantly more abundant in gastric cancer tissues. CNN2 is known to bind to actin and
other channel proteins; for example, a calcium channel protein called calmodulin and involves in
the structural organization of actin filaments. Other known functions of this protein include
wound healing and positive regulation of cell proliferation. CNN2 being highly abundant in
rectal cancers is proposed as a diagnostic biomarker at transcript level (Choi et al., 2011). CNN2
is only reported in transcript levels in gastric cancers.
68
Discussion
TIA1
Nucelolysin TIA1 (TIA1), also called as T-cell intracellular antigen is an mRNA binding protein
is expressed when cells undergo specific stress. They involve in both transcriptional and posttranscriptional gene expression in eukaryotic cells. They activate nucleolysis against cells that
target cytotoxic lymphocytes. Over expression of TIA1 under various stress conditions is studied
in different human cells (Dinh et al., 2013; Gottschald et al., 2010). An important function of
TIA1 is to regulate alternative pre-mRNA splicing of approximately 15% of the cassette human
exons. High lethality rates in mice lacking TIA1 also shows their importance in regulation of
important genes. Interestingly, TIA1 is not studied well in gastric cancers except inconsistent
reports on abundance in protein level but in cancers like, colon, ovary and lymphoma TIA1 is
found high in abundance (Izquierdo et al., 2011).
OCLN
Occludin (OCLN) found high in abundance in gastric adenocarcinoma is a protein known for its
localization in tight junctions of both epithelial and endothelial cells. Over-expressing OCLN in
cells lacking tight junctions induces cell adhesion. Mutations involved in OCLN gene leads to a
rare neurological disorder called pseudo-TORCH syndrome. Belonging to the claudin family of
proteins, OCLN is associated with epithelial-mesenchymal transition in colorectal cancers, which
is one of the initial mechanisms in oncogenesis. It if found that OCLN is regulated by
transcription factor AP4 which is a strong prognostic factor in colorectal cancers (Jackstadt et al.,
2013). In skin cancers, it is observed that OCLN is involved in Calcium ion-dependent
homeostasis and other important cellular processes in oncogenesis like apoptosis, adhesion and
differentiation (Rachow et al., 2013). One of the earlier studies on the expression of OCLN in
esophageal, colon and gastric tissues found OCLN’s localization in the tight junctions of the
69
Discussion
polarized epithelial cells. Further investigation with immuno-histochemical methods exhibited
the expression of two tight junction proteins, OCLN and Zonula occludens (ZO1) in the highly
differentiated tumor epithelia of colon and gastric cancers (Kimura et al., 1997). Gastric cancer
cells in response to epidermal growth factor (EGF) treatment translocated OCLN from cytoplasm
to the tight junctions which explain the role of it in cell-cell adhesion (Yoshida et al., 2005).
From these results we suggest that OCLN is involved by similar mechanisms of oncogenesis in
gastric adeonocarcinomas.
S100A9
S100A9 also known as calgranulin-B belongs to S100 family of calcium binding proteins
localized mostly in the cytoplasm and nucleus. S100A9 is often studied in cancers for its
properties other than calcium binding like apoptosis, cell cycle progression and differentiation. It
also plays a role as a pro-inflammatory mediator in acute chronic inflammation. S100A9 is
highly abundant in gastric cancer cells in the transcript level (El-Rifai et al., 2002). Apart from
gastric cancers S100A9 is reported to be highly abundant at protein level in ovarian, hepatocellular and breast cancer. Interestingly, S100A9 is highly abundant in both malignant epithelial
and stromal cells in high grade gastric adenocarcinoma leading to their detection in plasma
samples of gastric cancer patients (Wang et al., 2013). It is not so surprising that S100A9 triggers
cascade of signaling molecules including p38 mitogen-activated protein kinase and NFκB
leading to the migration of gastric cancer cells and in turn
adenocarcinoma cells (Kwon et al., 2013).
70
invasiveness of the gastric
Discussion
VEGFA
Vascular endothelial growth factor A (VEGFA) is a protein with all great potential functions
involved in oncogenesis such as enhancing endothelial cell proliferation, promoting cell
migration and inhibition of apoptotic signals. It is not only a growth factor with great abilities but
also an active protein in vasculogenesis, endothelial cell growth, angiogenesis and promotes
permeability of blood vessels. Such a protein is often a therapeutic target in many cancers and
other diseases also. High abundance of VEGFA combined with SP1 at protein levels is directly
correlated with the grade and stage of gastric adenocarcinomas. This co-expression resulted in
poor prognosis of the disease (Yao et al., 2004). Metastasis and migration abilities of gastric
adenocarcinoma cells are inhibited strongly when integrin alpha v beta 6 (IAVB6) function is
lost. VEGFA is known to activate the integrin molecule IAVB6 which strongly suggests the
therapeutic importance of drugs inhibiting VEGFA (Zhao et al., 2010). Another interesting study
involving protein kinase D2 (PKD2) and VEGFA in gastric cancer cells finds the importance of
the role of VEGFA in endothelial cell proliferation and migration. Invivo knockdown of VEGFA
and PKD2 in mice results in inhibition of angiogenesis and tumor growth (Azoitei et al., 2010).
Strikingly, VEGFA is very highly abundant in our analysis on gastric adenocarcinoma samples.
DKK3
Dickkopf related-protein 3 (DKK3) is found less abundant in the gastric adenocarcinomas. Often
referred as one of the vital tumor suppressor genes, DKK3 is often missing in various cancer
cells. Wnt signaling which is widely studied in biology of cancers can be modulated by varying
concentrations of other soluble including DKK3. DKK3 at protein level is very less in abundance
in gastric and colon cancer cells (Byun et al., 2005). More detailed molecular analysis of gastric
cancers illustrated epigenetic down-regulation leading to the inactivation in the translation of
71
Discussion
DKK3 genes in gastric cancers (Sato et al., 2007). More recently, plasmid-mediated functional
resurrection of DKK3 in gastric cancer cells decreased the aggressiveness and migration of the
cells indicating the possible tumor suppressor role of DKK3 in gastric cancer. Moreover, DKK3
expression in gastric cancers negatively correlated with the tumor size, metastasis and grade of
the tumor indicating the prognostic significance of it in gastric cancer (Xu et al., 2012).
From the above exhibited significantly abundant proteins in gastric adenocarcinomas, we suggest
that similar mechanisms of oncogenesis may be the causes of cancer in sets of samples in
analyses. The stability and consistency of some of the markers can be a boon to available
molecular diagnostics as all of them are studied at the protein level. A signature of top sixteen
proteins from table 3 will help not only to predict prognosis but also to establish individual-based
treatment strategy as the comparisons here are made matching the healthy controls and the
gastric adenocarcinoma tissues after surgical resection from the same patient.
72
Conclusions
5 Conclusions
The current study presented here focused on the application of complex antibody microarrays as
a diagnostic and prognostic tool in human cancers. In the first part of the study, protein profiles
of non-muscle invasive bladder cancer samples were studied and a signature of 20 proteins was
identified to predict the recurrence of the cancer. Apart from the prediction of recurrence with
high specificity and sensitivity, the important molecular mechanisms that may lead to recurrence
of cancer were also identified. The results from this analysis will prove to be efficient in
predicting the recurrence of cancer by the application of the protein signature in large sets of
non-muscle bladder cancer samples with appropriate follow-up time. Moreover, this effective
prediction will aid the clinicians to develop an optimal strategy to follow-up and treat the
patients.
In the second part of the study, analysis of gastric adenocarcinoma samples by two different
methods of incubations on the antibody microarrays identified a set of 16 proteins highly
differentially regulated between healthy controls and cancer tissues. Proteins identified included
many validated and well studied therapeutic targets which are under clinical trials. A strong
literature search on the identified proteins provided more insights on the molecular mechanisms
of oncogenesis and metastasis in gastric tissues. The results from this analysis will help
clinicians to identify patients with higher risk of metastasis and death occurrence. Validation of
the protein profile in large sets of tumors will favor individual-based treatment regimen for
gastric cancer. More functional validations on the identified targets will help in the development
of therapeutics to treat gastric adenocarcinoma.
73
References
6. References
Ahmad, M., Theofanidis, P., and Medford, R.M. (1998). Role of activating protein-1 in the
regulation of the vascular cell adhesion molecule-1 gene expression by tumor necrosis factoralpha. The Journal of biological chemistry 273, 4616-4621.
Aizawa, K., Suzuki, T., Kada, N., Ishihara, A., Kawai-Kowase, K., Matsumura, T., Sasaki, K.,
Munemasa, Y., Manabe, I., Kurabayashi, M., et al. (2004). Regulation of platelet-derived growth
factor-A chain by Kruppel-like factor 5: new pathway of cooperative activation with nuclear
factor-kappaB. The Journal of biological chemistry 279, 70-76.
Alazzouzi, H., Alhopuro, P., Salovaara, R., Sammalkorpi, H., Jarvinen, H., Mecklin, J.P.,
Hemminki, A., Schwartz, S., Jr., Aaltonen, L.A., and Arango, D. (2005). SMAD4 as a prognostic
marker in colorectal cancer. Clinical cancer research : an official journal of the American
Association for Cancer Research 11, 2606-2611.
Alberts B, J.A., Lewis J, Raff M, Roberts K, Walter P (2007). Molecular biology of the cell.
New York and London: Garland Science.
Alhamdani, M.S., Schroder, C., and Hoheisel, J.D. (2009). Oncoproteomic profiling with
antibody microarrays. Genome medicine 1, 68.
Alhamdani, M.S., Schroder, C., Werner, J., Giese, N., Bauer, A., and Hoheisel, J.D. (2010).
Single-step procedure for the isolation of proteins at near-native conditions from mammalian
tissue for proteomic analysis on antibody microarrays. Journal of proteome research 9, 963-971.
Alvaro, T., Lejeune, M., Salvado, M.T., Bosch, R., Garcia, J.F., Jaen, J., Banham, A.H.,
Roncador, G., Montalban, C., and Piris, M.A. (2005). Outcome in Hodgkin's lymphoma can be
predicted from the presence of accompanying cytotoxic and regulatory T cells. Clinical cancer
research : an official journal of the American Association for Cancer Research 11, 1467-1473.
Andreasson, U., Ek, S., Merz, H., Rosenquist, R., Andersen, N., Jerkeman, M., Dictor, M., and
Borrebaeck, C.A. (2008). B cell lymphomas express CX3CR1 a non-B cell lineage adhesion
molecule. Cancer letters 259, 138-145.
Avritscher, E.B., Cooksley, C.D., Grossman, H.B., Sabichi, A.L., Hamblin, L., Dinney, C.P., and
Elting, L.S. (2006). Clinical model of lifetime cost of treating bladder cancer and associated
complications. Urology 68, 549-553.
74
References
Azoitei, N., Pusapati, G.V., Kleger, A., Moller, P., Kufer, R., Genze, F., Wagner, M., van Lint,
J., Carmeliet, P., Adler, G., et al. (2010). Protein kinase D2 is a crucial regulator of tumour cellendothelial cell communication in gastrointestinal tumours. Gut 59, 1316-1330.
Babjuk, M., Oosterlinck, W., Sylvester, R., Kaasinen, E., Bohle, A., Palou-Redorta, J., and
European Association of, U. (2008). EAU guidelines on non-muscle-invasive urothelial
carcinoma of the bladder. European urology 54, 303-314.
Babjuk, M., Oosterlinck, W., Sylvester, R., Kaasinen, E., Bohle, A., Palou-Redorta, J., Roupret,
M., and European Association of, U. (2011). EAU guidelines on non-muscle-invasive urothelial
carcinoma of the bladder, the 2011 update. European urology 59, 997-1008.
Bonin, S., Brunetti, D., Benedetti, E., Dotti, I., Gorji, N., and Stanta, G. (2008). Molecular
characterisation of breast cancer patients at high and low recurrence risk. Virchows Archiv : an
international journal of pathology 452, 241-250.
Breiman (2001). Random Forests. Machine Learning.
Brennan, D.J., O'Connor, D.P., Rexhepaj, E., Ponten, F., and Gallagher, W.M. (2010). Antibodybased proteomics: fast-tracking molecular diagnostics in oncology. Nature reviews Cancer 10,
605-617.
Byun, T., Karimi, M., Marsh, J.L., Milovanovic, T., Lin, F., and Holcombe, R.F. (2005).
Expression of secreted Wnt antagonists in gastrointestinal tissues: potential role in stem cell
homeostasis. Journal of clinical pathology 58, 515-519.
Chadalapaka, G., Jutooru, I., Burghardt, R., and Safe, S. (2010). Drugs that target specificity
proteins downregulate epidermal growth factor receptor in bladder cancer cells. Molecular
cancer research : MCR 8, 739-750.
Chadalapaka, G., Jutooru, I., Chintharlapalli, S., Papineni, S., Smith, R., 3rd, Li, X., and Safe, S.
(2008). Curcumin decreases specificity protein expression in bladder cancer cells. Cancer
research 68, 5345-5354.
Ching, C.B., and Hansel, D.E. (2010). Expanding therapeutic targets in bladder cancer: the
PI3K/Akt/mTOR pathway. Laboratory investigation; a journal of technical methods and
pathology 90, 1406-1414.
75
References
Choi, S.Y., Jang, J.H., and Kim, K.R. (2011). Analysis of differentially expressed genes in
human rectal carcinoma using suppression subtractive hybridization. Clinical and experimental
medicine 11, 219-226.
Chua, H.L., Bhat-Nakshatri, P., Clare, S.E., Morimiya, A., Badve, S., and Nakshatri, H. (2007).
NF-kappaB represses E-cadherin expression and enhances epithelial to mesenchymal transition
of mammary epithelial cells: potential involvement of ZEB-1 and ZEB-2. Oncogene 26, 711724.
Cinghu, S., Anandharaj, A., Lee, H.C., Yu, J.R., and Park, W.Y. (2011). FTS (fused toes
homolog) a novel oncoprotein involved in uterine cervical carcinogenesis and a potential
diagnostic marker for cervical cancer. Journal of cellular physiology 226, 1564-1572.
Cunningham, D., Okines, A.F., and Ashley, S. (2010). Capecitabine and oxaliplatin for advanced
esophagogastric cancer. The New England journal of medicine 362, 858-859.
Dexlin, L., Ingvarsson, J., Frendeus, B., Borrebaeck, C.A., and Wingren, C. (2008). Design of
recombinant antibody microarrays for cell surface membrane proteomics. Journal of proteome
research 7, 319-327.
Dicken, B.J., Bigam, D.L., Cass, C., Mackey, J.R., Joy, A.A., and Hamilton, S.M. (2005).
Gastric adenocarcinoma: review and considerations for future directions. Annals of surgery 241,
27-39.
Dinh, P.X., Beura, L.K., Das, P.B., Panda, D., Das, A., and Pattnaik, A.K. (2013). Induction of
stress granule-like structures in vesicular stomatitis virus-infected cells. Journal of virology 87,
372-383.
Dubosq, F., Ploussard, G., Soliman, H., Turpin, E., Latil, A., Desgrandchamps, F., de The, H.,
and Mongiat-Artus, P. (2012). Identification of a three-gene expression signature of early
recurrence in non-muscle-invasive urothelial cell carcinoma of the bladder. Urologic oncology
30, 833-840.
El-Rifai, W., Moskaluk, C.A., Abdrabbo, M.K., Harper, J., Yoshida, C., Riggins, G.J., Frierson,
H.F., Jr., and Powell, S.M. (2002). Gastric cancers overexpress S100A calcium-binding proteins.
Cancer research 62, 6823-6826.
Ellmark, P., Ingvarsson, J., Carlsson, A., Lundin, B.S., Wingren, C., and Borrebaeck, C.A.
(2006). Identification of protein expression signatures associated with Helicobacter pylori
76
References
infection and gastric adenocarcinoma using recombinant antibody microarrays. Molecular &
cellular proteomics : MCP 5, 1638-1646.
Fujita, J., Crane, A.M., Souza, M.K., Dejosez, M., Kyba, M., Flavell, R.A., Thomson, J.A., and
Zwaka, T.P. (2008). Caspase activity mediates the differentiation of embryonic stem cells. Cell
stem cell 2, 595-601.
Gallagher, E.M., O'Shea, D.M., Fitzpatrick, P., Harrison, M., Gilmartin, B., Watson, J.A.,
Clarke, T., Leonard, M.O., McGoldrick, A., Meehan, M., et al. (2008). Recurrence of urothelial
carcinoma of the bladder: a role for insulin-like growth factor-II loss of imprinting and
cytoplasmic E-cadherin immunolocalization. Clinical cancer research : an official journal of the
American Association for Cancer Research 14, 6829-6838.
Gluz, O., Mengele, K., Schmitt, M., Kates, R., Diallo-Danebrock, R., Neff, F., Royer, H.D.,
Eckstein, N., Mohrmann, S., Ting, E., et al. (2009). Y-box-binding protein YB-1 identifies highrisk patients with primary breast cancer benefiting from rapidly cycled tandem high-dose
adjuvant chemotherapy. Journal of clinical oncology : official journal of the American Society of
Clinical Oncology 27, 6144-6151.
Gottschald, O.R., Malec, V., Krasteva, G., Hasan, D., Kamlah, F., Herold, S., Rose, F., Seeger,
W., and Hanze, J. (2010). TIAR and TIA-1 mRNA-binding proteins co-aggregate under
conditions of rapid oxygen decline and extreme hypoxia and suppress the HIF-1alpha pathway.
Journal of molecular cell biology 2, 345-356.
Grossman, H.B., Messing, E., Soloway, M., Tomera, K., Katz, G., Berger, Y., and Shen, Y.
(2005). Detection of bladder cancer using a point-of-care proteomic assay. JAMA : the journal of
the American Medical Association 293, 810-816.
Hanahan, D., and Weinberg, R.A. (2011). Hallmarks of cancer: the next generation. Cell 144,
646-674.
Hartwell, L., Mankoff, D., Paulovich, A., Ramsey, S., and Swisher, E. (2006). Cancer
biomarkers: a systems approach. Nature biotechnology 24, 905-908.
Hoheisel, J.D., Alhamdani, M.S., and Schroder, C. (2013). Affinity-based microarrays for
proteomic analysis of cancer tissues. Proteomics Clinical applications 7, 8-15.
Hu, J., and Verkman, A.S. (2006). Increased migration and metastatic potential of tumor cells
expressing aquaporin water channels. FASEB journal : official publication of the Federation of
American Societies for Experimental Biology 20, 1892-1894.
77
References
Huang da, W., Sherman, B.T., and Lempicki, R.A. (2009). Systematic and integrative analysis of
large gene lists using DAVID bioinformatics resources. Nature protocols 4, 44-57.
Huang, Q., Li, F., Liu, X., Li, W., Shi, W., Liu, F.F., O'Sullivan, B., He, Z., Peng, Y., Tan, A.C.,
et al. (2011). Caspase 3-mediated stimulation of tumor cell repopulation during cancer
radiotherapy. Nature medicine 17, 860-866.
Ikegaki, N., Gotoh, T., Kung, B., Riceberg, J.S., Kim, D.Y., Zhao, H., Rappaport, E.F., Hicks,
S.L., Seeger, R.C., and Tang, X.X. (2007). De novo identification of MIZ-1 (ZBTB17) encoding
a MYC-interacting zinc-finger protein as a new favorable neuroblastoma gene. Clinical cancer
research : an official journal of the American Association for Cancer Research 13, 6001-6009.
Iraci, N., Diolaiti, D., Papa, A., Porro, A., Valli, E., Gherardi, S., Herold, S., Eilers, M.,
Bernardoni, R., Della Valle, G., et al. (2011). A SP1/MIZ1/MYCN repression complex recruits
HDAC1 at the TRKA and p75NTR promoters and affects neuroblastoma malignancy by
inhibiting the cell response to NGF. Cancer research 71, 404-412.
Izquierdo, J.M., Alcalde, J., Carrascoso, I., Reyes, R., and Ludena, M.D. (2011). Knockdown of
T-cell intracellular antigens triggers cell proliferation, invasion and tumour growth. The
Biochemical journal 435, 337-344.
Jackstadt, R., Roh, S., Neumann, J., Jung, P., Hoffmann, R., Horst, D., Berens, C., Bornkamm,
G.W., Kirchner, T., Menssen, A., et al. (2013). AP4 is a mediator of epithelial-mesenchymal
transition and metastasis in colorectal cancer. The Journal of experimental medicine 210, 13311350.
Jacobs, B.L., Lee, C.T., and Montie, J.E. (2010). Bladder cancer in 2010: how far have we
come? CA: a cancer journal for clinicians 60, 244-272.
Jeong, P., Min, B.D., Ha, Y.S., Song, P.H., Kim, I.Y., Ryu, K.H., Kim, J.H., Yun, S.J., and Kim,
W.J. (2012). RUNX3 methylation in normal surrounding urothelium of patients with nonmuscle-invasive bladder cancer: potential role in the prediction of tumor progression. European
journal of surgical oncology : the journal of the European Society of Surgical Oncology and the
British Association of Surgical Oncology 38, 1095-1100.
Jonges, L.E., Nagelkerke, J.F., Ensink, N.G., van der Velde, E.A., Tollenaar, R.A., Fleuren, G.J.,
van de Velde, C.J., Morreau, H., and Kuppen, P.J. (2001). Caspase-3 activity as a prognostic
factor in colorectal carcinoma. Laboratory investigation; a journal of technical methods and
pathology 81, 681-688.
78
References
Jutooru, I., Chadalapaka, G., Sreevalsan, S., Lei, P., Barhoumi, R., Burghardt, R., and Safe, S.
(2010). Arsenic trioxide downregulates specificity protein (Sp) transcription factors and inhibits
bladder cancer cell and tumor growth. Experimental cell research 316, 2174-2188.
Kanehisa, M., Goto, S., Sato, Y., Furumichi, M., and Tanabe, M. (2012). KEGG for integration
and interpretation of large-scale molecular data sets. Nucleic acids research 40, D109-114.
Karam, J.A., Lotan, Y., Karakiewicz, P.I., Ashfaq, R., Sagalowsky, A.I., Roehrborn, C.G., and
Shariat, S.F. (2007). Use of combined apoptosis biomarkers for prediction of bladder cancer
recurrence and mortality after radical cystectomy. The lancet oncology 8, 128-136.
Kennedy, N.J., Kataoka, T., Tschopp, J., and Budd, R.C. (1999). Caspase activation is required
for T cell proliferation. The Journal of experimental medicine 190, 1891-1896.
Kimura, Y., Shiozaki, H., Hirao, M., Maeno, Y., Doki, Y., Inoue, M., Monden, T., AndoAkatsuka, Y., Furuse, M., Tsukita, S., et al. (1997). Expression of occludin, tight-junctionassociated protein, in human digestive tract. The American journal of pathology 151, 45-54.
Knowles, M.A., Shaw, M.E., and Proctor, A.J. (1993). Deletion mapping of chromosome 8 in
cancers of the urinary bladder using restriction fragment length polymorphisms and
microsatellite polymorphisms. Oncogene 8, 1357-1364.
Kodama, M., Kitadai, Y., Tanaka, M., Kuwai, T., Tanaka, S., Oue, N., Yasui, W., and Chayama,
K. (2008). Vascular endothelial growth factor C stimulates progression of human gastric cancer
via both autocrine and paracrine mechanisms. Clinical cancer research : an official journal of the
American Association for Cancer Research 14, 7205-7214.
Konstantakou, E.G., Voutsinas, G.E., Karkoulis, P.K., Aravantinos, G., Margaritis, L.H., and
Stravopodis, D.J. (2009). Human bladder cancer cells undergo cisplatin-induced apoptosis that is
associated with p53-dependent and p53-independent responses. International journal of oncology
35, 401-416.
Konstantinidou, A.E., Givalos, N., Gakiopoulou, H., Korkolopoulou, P., Kotsiakis, X., Boviatsis,
E., Agrogiannis, G., Mahera, H., and Patsouris, E. (2007). Caspase-3 immunohistochemical
expression is a marker of apoptosis, increased grade and early recurrence in intracranial
meningiomas. Apoptosis : an international journal on programmed cell death 12, 695-705.
Korkolopoulou, P., Konstantinidou, A.E., Christodoulou, P., Patsouris, E., Thomas-Tsangli, E.,
Kapralos, P., and Davaris, P. (2000). Apoptosis in bladder carcinomas detected with monoclonal
79
References
antibody to single-stranded DNA: relation to cell cycle regulators and survival. Urology 56, 516520.
Kramer, M.W., Waalkes, S., Serth, J., Hennenlotter, J., Tezval, H., Stenzl, A., Kuczyk, M.A.,
and Merseburger, A.S. (2011). Decreased galectin-8 is a strong marker for recurrence in
urothelial carcinoma of the bladder. Urologia internationalis 87, 143-150.
Kroll, D.J., Sullivan, D.M., Gutierrez-Hartmann, A., and Hoeffler, J.P. (1993). Modification of
DNA topoisomerase II activity via direct interactions with the cyclic adenosine-3',5'monophosphate response element-binding protein and related transcription factors. Molecular
endocrinology 7, 305-318.
Kusnezow, W., Jacob, A., Walijew, A., Diehl, F., and Hoheisel, J.D. (2003). Antibody
microarrays: an evaluation of production parameters. Proteomics 3, 254-264.
Kwon, C.H., Moon, H.J., Park, H.J., Choi, J.H., and Park do, Y. (2013). S100A8 and S100A9
promotes invasion and migration through p38 mitogen-activated protein kinase-dependent NFkappaB activation in gastric cancer cells. Molecules and cells 35, 226-234.
Lee, A.J., Roylance, R., Sander, J., Gorman, P., Endesfelder, D., Kschischo, M., Jones, N.P.,
East, P., Nicke, B., Spassieva, S., et al. (2012). CERT depletion predicts chemotherapy benefit
and mediates cytotoxic and polyploid-specific cancer cell death through autophagy induction.
The Journal of pathology 226, 482-494.
Lee, M., Song, S.U., Ryu, J.K., and Suh, J.K. (2004). Sp1-dependent regulation of the tissue
inhibitor of metalloproteinases-1 promoter. Journal of cellular biochemistry 91, 1260-1268.
Leung, S.Y., Yuen, S.T., Chu, K.M., Mathy, J.A., Li, R., Chan, A.S., Law, S., Wong, J., Chen,
X., and So, S. (2004). Expression profiling identifies chemokine (C-C motif) ligand 18 as an
independent prognostic indicator in gastric cancer. Gastroenterology 127, 457-469.
Lin, H.H., Lai, S.C., and Chau, L.Y. (2011). Heme oxygenase-1/carbon monoxide induces
vascular endothelial growth factor expression via p38 kinase-dependent activation of Sp1. The
Journal of biological chemistry 286, 3829-3838.
Ling, Y., Chunli, L., Xiaohou, W., and Qiaoling, Z. (2011). Involvement of the
PLCepsilon/PKCalpha pathway in human BIU-87 bladder cancer cell proliferation. Cell biology
international 35, 1031-1036.
80
References
Louhelainen, J.P., Hurst, C.D., Pitt, E., Nishiyama, H., Pickett, H.A., and Knowles, M.A. (2006).
DBC1 re-expression alters the expression of multiple components of the plasminogen pathway.
Oncogene 25, 2409-2419.
Lozano, R., Naghavi, M., Foreman, K., Lim, S., Shibuya, K., Aboyans, V., Abraham, J., Adair,
T., Aggarwal, R., Ahn, S.Y., et al. (2012). Global and regional mortality from 235 causes of
death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden of
Disease Study 2010. Lancet 380, 2095-2128.
Luis, N.M., Lopez-Knowles, E., and Real, F.X. (2007). Molecular biology of bladder cancer.
Clinical & translational oncology : official publication of the Federation of Spanish Oncology
Societies and of the National Cancer Institute of Mexico 9, 5-12.
Martino, J.J., Wall, B.A., Mastrantoni, E., Wilimczyk, B.J., La Cava, S.N., Degenhardt, K.,
White, E., and Chen, S. (2012). Metabotropic glutamate receptor 1 (Grm1) is an oncogene in
epithelial cells. Oncogene.
Matsushita, K., Cha, E.K., Matsumoto, K., Baba, S., Chromecki, T.F., Fajkovic, H., Sun, M.,
Karakiewicz, P.I., Scherr, D.S., and Shariat, S.F. (2011). Immunohistochemical biomarkers for
bladder cancer prognosis. International journal of urology : official journal of the Japanese
Urological Association 18, 616-629.
Michaud, G.A., Salcius, M., Zhou, F., Bangham, R., Bonin, J., Guo, H., Snyder, M., Predki, P.F.,
and Schweitzer, B.I. (2003). Analyzing antibody specificity with whole proteome microarrays.
Nature biotechnology 21, 1509-1512.
Mitra, A.P., Pagliarulo, V., Yang, D., Waldman, F.M., Datar, R.H., Skinner, D.G., Groshen, S.,
and Cote, R.J. (2009). Generation of a concise gene panel for outcome prediction in urinary
bladder cancer. Journal of clinical oncology : official journal of the American Society of Clinical
Oncology 27, 3929-3937.
Montie, J.E., Clark, P.E., Eisenberger, M.A., El-Galley, R., Greenberg, R.E., Herr, H.W., Hudes,
G.R., Kuban, D.A., Kuzel, T.M., Lange, P.H., et al. (2009). Bladder cancer. Journal of the
National Comprehensive Cancer Network : JNCCN 7, 8-39.
Morrissey, J.J., London, A.N., Luo, J., and Kharasch, E.D. (2010). Urinary biomarkers for the
early diagnosis of kidney cancer. Mayo Clinic proceedings Mayo Clinic 85, 413-421.
81
References
Negraes, P.D., Favaro, F.P., Camargo, J.L., Oliveira, M.L., Goldberg, J., Rainho, C.A., and
Salvadori, D.M. (2008). DNA methylation patterns in bladder cancer and washing cell
sediments: a perspective for tumor recurrence detection. BMC cancer 8, 238.
Neid, M., Datta, K., Stephan, S., Khanna, I., Pal, S., Shaw, L., White, M., and Mukhopadhyay,
D. (2004). Role of insulin receptor substrates and protein kinase C-zeta in vascular permeability
factor/vascular endothelial growth factor expression in pancreatic cancer cells. The Journal of
biological chemistry 279, 3941-3948.
Otterbach, F., Callies, R., Adamzik, M., Kimmig, R., Siffert, W., Schmid, K.W., and Bankfalvi,
A. (2010). Aquaporin 1 (AQP1) expression is a novel characteristic feature of a particularly
aggressive subgroup of basal-like breast carcinomas. Breast cancer research and treatment 120,
67-76.
Ouyang, X., Jessen, W.J., Al-Ahmadie, H., Serio, A.M., Lin, Y., Shih, W.J., Reuter, V.E.,
Scardino, P.T., Shen, M.M., Aronow, B.J., et al. (2008). Activator protein-1 transcription factors
are associated with progression and recurrence of prostate cancer. Cancer research 68, 21322144.
Penuelas, S., Anido, J., Prieto-Sanchez, R.M., Folch, G., Barba, I., Cuartas, I., Garcia-Dorado,
D., Poca, M.A., Sahuquillo, J., Baselga, J., et al. (2009). TGF-beta increases glioma-initiating
cell self-renewal through the induction of LIF in human glioblastoma. Cancer cell 15, 315-327.
Pietrantonio, F., De Braud, F., Da Prat, V., Perrone, F., Pierotti, M.A., Gariboldi, M., Fanetti, G.,
Biondani, P., Pellegrinelli, A., Bossi, I., et al. (2013). A review on biomarkers for prediction of
treatment outcome in gastric cancer. Anticancer research 33, 1257-1266.
Poncelet, A.C., and Schnaper, H.W. (2001). Sp1 and Smad proteins cooperate to mediate
transforming growth factor-beta 1-induced alpha 2(I) collagen expression in human glomerular
mesangial cells. The Journal of biological chemistry 276, 6983-6992.
Qiu, J., Choi, G., Li, L., Wang, H., Pitteri, S.J., Pereira-Faca, S.R., Krasnoselsky, A.L.,
Randolph, T.W., Omenn, G.S., Edelstein, C., et al. (2008). Occurrence of autoantibodies to
annexin I, 14-3-3 theta and LAMR1 in prediagnostic lung cancer sera. Journal of clinical
oncology : official journal of the American Society of Clinical Oncology 26, 5060-5066.
Rachow, S., Zorn-Kruppa, M., Ohnemus, U., Kirschner, N., Vidal-y-Sy, S., von den Driesch, P.,
Bornchen, C., Eberle, J., Mildner, M., Vettorazzi, E., et al. (2013). Occludin is involved in
adhesion, apoptosis, differentiation and Ca2+-homeostasis of human keratinocytes: implications
for tumorigenesis. PloS one 8, e55116.
82
References
Sanchez-Carbayo, M., and Cordon-Cardo, C. (2007). Molecular alterations associated with
bladder cancer progression. Seminars in oncology 34, 75-84.
Sanchez-Carbayo, M., Socci, N.D., Lozano, J.J., Haab, B.B., and Cordon-Cardo, C. (2006).
Profiling bladder cancer using targeted antibody arrays. The American journal of pathology 168,
93-103.
Sato, H., Suzuki, H., Toyota, M., Nojima, M., Maruyama, R., Sasaki, S., Takagi, H., Sogabe, Y.,
Sasaki, Y., Idogawa, M., et al. (2007). Frequent epigenetic inactivation of DICKKOPF family
genes in human gastrointestinal tumors. Carcinogenesis 28, 2459-2466.
Schiffer, E., Vlahou, A., Petrolekas, A., Stravodimos, K., Tauber, R., Geschwend, J.E., Neuhaus,
J., Stolzenburg, J.U., Conaway, M.R., Mischak, H., et al. (2009). Prediction of muscle-invasive
bladder cancer using urinary proteomics. Clinical cancer research : an official journal of the
American Association for Cancer Research 15, 4935-4943.
Schmidt, R., Jacak, J., Schirwitz, C., Stadler, V., Michel, G., Marme, N., Schutz, G.J., Hoheisel,
J.D., and Knemeyer, J.P. (2011). Single-molecule detection on a protein-array assay platform for
the exposure of a tuberculosis antigen. Journal of proteome research 10, 1316-1322.
Schroder, C., Jacob, A., Tonack, S., Radon, T.P., Sill, M., Zucknick, M., Ruffer, S., Costello, E.,
Neoptolemos, J.P., Crnogorac-Jurcevic, T., et al. (2010). Dual-color proteomic profiling of
complex samples with a microarray of 810 cancer-related antibodies. Molecular & cellular
proteomics : MCP 9, 1271-1280.
Senapati, S., Chaturvedi, P., Sharma, P., Venkatraman, G., Meza, J.L., El-Rifai, W., Roy, H.K.,
and Batra, S.K. (2008). Deregulation of MUC4 in gastric adenocarcinoma: potential
pathobiological implication in poorly differentiated non-signet ring cell type gastric cancer.
British journal of cancer 99, 949-956.
Shariat, S.F., Karakiewicz, P.I., Ashfaq, R., Lerner, S.P., Palapattu, G.S., Cote, R.J., Sagalowsky,
A.I., and Lotan, Y. (2008). Multiple biomarkers improve prediction of bladder cancer recurrence
and mortality in patients undergoing cystectomy. Cancer 112, 315-325.
Shariat, S.F., Zlotta, A.R., Ashfaq, R., Sagalowsky, A.I., and Lotan, Y. (2007). Cooperative
effect of cell-cycle regulators expression on bladder cancer development and biologic
aggressiveness. Modern pathology : an official journal of the United States and Canadian
Academy of Pathology, Inc 20, 445-459.
83
References
Siegel, R., Naishadham, D., and Jemal, A. (2012). Cancer statistics, 2012. CA: a cancer journal
for clinicians 62, 10-29.
Sill, M., Schroder, C., Hoheisel, J.D., Benner, A., and Zucknick, M. (2010). Assessment and
optimisation of normalisation methods for dual-colour antibody microarrays. BMC
bioinformatics 11, 556.
Slawski, M., Daumer, M., and Boulesteix, A.L. (2008). CMA: a comprehensive Bioconductor
package for supervised classification with high dimensional data. BMC bioinformatics 9, 439.
Smyth, G.K. (2004). Linear models and empirical bayes methods for assessing differential
expression in microarray experiments. Statistical applications in genetics and molecular biology
3, Article3.
Smyth, G.K., Michaud, J., and Scott, H.S. (2005). Use of within-array replicate spots for
assessing differential expression in microarray experiments. Bioinformatics 21, 2067-2075.
Speyer, C.L., Smith, J.S., Banda, M., DeVries, J.A., Mekani, T., and Gorski, D.H. (2012).
Metabotropic glutamate receptor-1: a potential therapeutic target for the treatment of breast
cancer. Breast cancer research and treatment 132, 565-573.
St Hill, C.A. (2011). Interactions between endothelial selectins and cancer cells regulate
metastasis. Frontiers in bioscience : a journal and virtual library 17, 3233-3251.
Stravopodis, D.J., Karkoulis, P.K., Konstantakou, E.G., Melachroinou, S., Lampidonis, A.D.,
Anastasiou, D., Kachrilas, S., Messini-Nikolaki, N., Papassideri, I.S., Aravantinos, G., et al.
(2009). Grade-dependent effects on cell cycle progression and apoptosis in response to
doxorubicin in human bladder cancer cell lines. International journal of oncology 34, 137-160.
Suzuki, T., Ito, M., Hayasaki, N., Ishihara, A., Ando, T., Ina, K., and Kusugami, K. (2003).
Cytotoxic molecules expressed by intraepithelial lymphocytes may be involved in the
pathogenesis of acute gastric mucosal lesions. Journal of gastroenterology 38, 216-221.
Szklarczyk, D., Franceschini, A., Kuhn, M., Simonovic, M., Roth, A., Minguez, P., Doerks, T.,
Stark, M., Muller, J., Bork, P., et al. (2011). The STRING database in 2011: functional
interaction networks of proteins, globally integrated and scored. Nucleic acids research 39,
D561-568.
Szymczyk, K.H., Freeman, T.A., Adams, C.S., Srinivas, V., and Steinbeck, M.J. (2006). Active
caspase-3 is required for osteoclast differentiation. Journal of cellular physiology 209, 836-844.
84
References
Takeno, A., Takemasa, I., Doki, Y., Yamasaki, M., Miyata, H., Takiguchi, S., Fujiwara, Y.,
Matsubara, K., and Monden, M. (2008). Integrative approach for differentially overexpressed
genes in gastric cancer by combining large-scale gene expression profiling and network analysis.
British journal of cancer 99, 1307-1315.
To, K., Fotovati, A., Reipas, K.M., Law, J.H., Hu, K., Wang, J., Astanehe, A., Davies, A.H., Lee,
L., Stratford, A.L., et al. (2010). Y-box binding protein-1 induces the expression of CD44 and
CD49f leading to enhanced self-renewal, mammosphere growth, and drug resistance. Cancer
research 70, 2840-2851.
Tung, J.N., Tsao, T.Y., Tai, C.J., Yeh, K.T., Cheng, Y.W., and Jiang, M.C. (2010). Distribution
of lysosome-associated membrane proteins-1 and -2, and cathepsin D in eosinophilic granular
bodies: possible relationship to cyst development in pilocytic astrocytomas. The Journal of
international medical research 38, 1354-1364.
van Duin, M., van Marion, R., Vissers, K., Watson, J.E., van Weerden, W.M., Schröder, F.H.,
Hop, W.C., van der Kwast, T.H., Collins, C., and van Dekken, H. (2005). High-resolution array
comparative genomic hybridization of chromosome arm 8q: evaluation of genetic progression
markers for prostate cancer. Genes, chromosomes & cancer 44, 438-449.
Wang, L., Chang, E.W., Wong, S.C., Ong, S.M., Chong, D.Q., and Ling, K.L. (2013). Increased
myeloid-derived suppressor cells in gastric cancer correlate with cancer stage and plasma
S100A8/A9 proinflammatory proteins. Journal of immunology 190, 794-804.
Weinberg, R.A. (2007). The Biology of Cancer. New York and London: Garland Science.
WHO (2009). WHO facts sheets N 297. cancer.
Wingren, C., Ingvarsson, J., Dexlin, L., Szul, D., and Borrebaeck, C.A. (2007). Design of
recombinant antibody microarrays for complex proteome analysis: choice of sample labeling-tag
and solid support. Proteomics 7, 3055-3065.
Xie, W., Rimm, D.L., Lin, Y., Shih, W.J., and Reiss, M. (2003). Loss of Smad signaling in
human colorectal cancer is associated with advanced disease and poor prognosis. Cancer journal
9, 302-312.
Xu, X.Y., Xia, P., Yu, M., Nie, X.C., Yang, X., Xing, Y.N., Liu, Y.P., Takano, Y., and Zheng,
H.C. (2012). The roles of REIC gene and its encoding product in gastric carcinoma. Cell cycle
11, 1414-1431.
85
References
Yao, J.C., Wang, L., Wei, D., Gong, W., Hassan, M., Wu, T.T., Mansfield, P., Ajani, J., and Xie,
K. (2004). Association between expression of transcription factor Sp1 and increased vascular
endothelial growth factor expression, advanced stage, and poor survival in patients with resected
gastric cancer. Clinical cancer research : an official journal of the American Association for
Cancer Research 10, 4109-4117.
Yoshida, K., Kanaoka, S., Takai, T., Uezato, T., Miura, N., Kajimura, M., and Hishida, A.
(2005). EGF rapidly translocates tight junction proteins from the cytoplasm to the cell-cell
contact via protein kinase C activation in TMK-1 gastric cancer cells. Experimental cell research
309, 397-409.
Yotov, W.V., Hamel, H., Rivard, G.E., Champagne, M.A., Russo, P.A., Leclerc, J.M., Bernstein,
M.L., and Levy, E. (1999). Amplifications of DNA primase 1 (PRIM1) in human osteosarcoma.
Genes, chromosomes & cancer 26, 62-69.
Zhao, R., Liu, X.Q., Wu, X.P., Liu, Y.F., Zhang, Z.Y., Yang, G.Y., Guo, S., Niu, J., Wang, J.Y.,
and Xu, K.S. (2010). Vascular endothelial growth factor (VEGF) enhances gastric carcinoma
invasiveness via integrin alpha(v)beta6. Cancer letters 287, 150-156.
Zheng, L., Wang, L., Ajani, J., and Xie, K. (2004). Molecular basis of gastric cancer
development and progression. Gastric cancer : official journal of the International Gastric Cancer
Association and the Japanese Gastric Cancer Association 7, 61-77.
86
Appendix
7. Appendix
7.1 Supplementary table S1
Supplementary table S1. List of the sixteen proteins with the significant expression variations
between healthy controls and tumours. Log fold changes in expression and the related adjusted
p-values are shown.
Sl.No.
Uniprot
Entryname
Uniprot
adjusted
accession P Value
1
THYG_HUMAN
P01266
log
Fold
change
8.65E-30
-1.86
2
IL2_HUMAN
P60568
1.90E-22
-1.58
3
PTEN_HUMAN
P60484
1.20E-15
-1.54
4
IL4_HUMAN
P05112
7.06E-13
-2.14
5
IFNG_HUMAN
P01579
6.67E-12
-1.16
6
VEGFA_HUMAN
P15692
1.68E-09
-0.93
7
FUS_HUMAN
P35637
3.25E-09
-0.66
8
LAP2A_HUMAN
P42166
0.002
0.30
9
TR10A_HUMAN
O00220
0.004
0.53
10
TNR21_HUMAN
O75509
0.008
-0.52
11
ALBU_HUMAN
P02768
0.010
0.89
12
ICAM1_HUMAN
P05362
0.012
0.26
13
TR10A_HUMAN
O00220
0.015
0.36
14
CDN1C_HUMAN
P49918
0.045
0.30
15
IL6_HUMAN
P05231
0.046
-0.20
16
ETS1_HUMAN
P14921
0.046
-0.39
87
Appendix
7.2 Supplementary table S2
Supplementary table S2. List of the two hundred and fifty five proteins with the significant
expression variations between recurrent and non-recurrent non-muscle invasive bladder cancer.
Log fold changes in expression and the related adjusted p-values are shown.
Sl.No. Uniprot
Uniprot
adjusted
log Fold
Entryname
accession
P Value
change
1
LMNA_HUMAN
P02545
3.32E-09
0.72
2
YBOX1_HUMAN
P67809
1.87E-08
0.52
3
JUN_HUMAN
P05412
1.27E-07
0.50
4
TIA1_HUMAN
P31483
2.13E-07
-0.39
5
SMAD3_HUMAN
P84022
2.13E-07
-0.59
6
PABP1_HUMAN
P11940
2.13E-07
-0.36
7
AKT3_HUMAN
Q9Y243
2.47E-07
0.48
8
CDN1A_HUMAN
P38936
3.38E-07
-0.52
9
LYAM1_HUMAN
P14151
9.62E-07
-0.52
10
YETS2_HUMAN
Q9ULM3
5.28E-06
0.32
11
AKTIP_HUMAN
Q9H8T0
5.38E-06
-0.35
12
PRI1_HUMAN
P49642
1.04E-05
-0.36
13
HSP7C_HUMAN
P11142
1.04E-05
-0.34
14
RSSA_HUMAN
P08865
1.07E-05
-0.35
15
GRM1A_HUMAN
Q96CP6
1.10E-05
0.27
16
TPA_HUMAN
P00750
1.14E-05
0.23
17
ZBT17_HUMAN
Q13105
1.14E-05
-0.55
18
CADH1_HUMAN
P12830
1.15E-05
0.33
19
LAMP2_HUMAN
P13473
1.15E-05
0.26
20
JUN_HUMAN
P05412
1.34E-05
0.48
21
LIFR_HUMAN
P42702
2.17E-05
0.30
22
TOP2A_HUMAN
P11388
2.17E-05
0.48
23
SPS2L_HUMAN
Q9NUQ6
2.20E-05
-0.23
24
UBIQ_HUMAN
P62988
2.47E-05
0.24
25
NFAC4_HUMAN
Q14934
2.47E-05
0.35
26
SF3B3_HUMAN
Q15393
2.47E-05
0.32
88
Appendix
27
2DMB_HUMAN
P28068
3.41E-05
-0.35
28
FAK1_HUMAN
Q05397
3.93E-05
-0.36
29
IFNG_HUMAN
P01579
3.95E-05
-0.46
30
SP1_HUMAN
P08047
4.14E-05
-0.36
31
ACTN1_HUMAN
P12814
4.31E-05
-0.36
32
TIE1_HUMAN
P35590
6.17E-05
-0.30
33
TIMP1_HUMAN
P01033
6.32E-05
-0.33
34
MMP13_HUMAN
P45452
6.32E-05
0.25
35
VTNC_HUMAN
P04004
6.76E-05
-0.51
36
K1C17_HUMAN
Q04695
0.0002
-0.20
37
NFKB1_HUMAN
P19838
0.0002
0.32
38
KLF5_HUMAN
Q13887
0.0002
0.37
39
NAP1_HUMAN
Q9BU70
0.0002
-0.27
40
RL10_HUMAN
P27635
0.0002
-0.28
41
MMP1_HUMAN
P03956
0.0002
-0.26
42
CDKN3_HUMAN
Q16667
0.0002
-0.33
43
CD59_HUMAN
P13987
0.0003
-0.33
44
PO2F2_HUMAN
P09086
0.0003
-0.35
45
MPIP2_HUMAN
P30305
0.0003
-0.28
46
FRAP_HUMAN
P42345
0.0003
-0.27
47
IRS2_HUMAN
Q9Y4H2
0.0003
-0.33
48
B2LA1_HUMAN
Q16548
0.0003
0.24
49
ERBB2_HUMAN
P04626
0.0004
-0.24
50
CASP3_HUMAN
P42574
0.0004
0.40
51
FINC_HUMAN
P02751
0.0004
-0.31
52
LAC_HUMAN
P01842
0.0004
-0.26
53
AURKB_HUMAN
Q96GD4
0.0004
-0.33
54
MPP3_HUMAN
Q13368
0.0004
-0.21
55
CD2A2_HUMAN
Q8N726
0.0004
-0.32
56
SOX9_HUMAN
P48436
0.0004
0.23
57
EPCAM_HUMAN
P16422
0.0004
-0.30
58
TSP3_HUMAN
P49746
0.0004
-0.24
59
O00446_HUMAN
O00446
0.0005
0.25
60
CP3A7_HUMAN
P24462
0.0006
-0.23
89
Appendix
61
THYG_HUMAN
P01266
0.0006
-0.34
62
NMDE3_HUMAN
Q14957
0.0006
-0.27
63
IL15_HUMAN
P40933
0.0006
-0.80
64
AQP1_HUMAN
P29972
0.0006
0.24
65
LAT1_HUMAN
Q01650
0.0006
-0.24
66
GSHB_HUMAN
P48637
0.0006
0.17
67
RPB3_HUMAN
P19387
0.0007
-0.23
68
K1C19_HUMAN
P08727
0.0007
0.21
69
PAK2_HUMAN
Q13177
0.0007
0.20
70
ZN593_HUMAN
O00488
0.0008
0.24
71
MYD88_HUMAN
Q99836
0.0008
0.22
72
IL8_HUMAN
P10145
0.0009
-0.23
73
SEP15_HUMAN
O60613
0.0010
-0.21
74
CUL2_HUMAN
Q13617
0.0010
-0.24
75
TNF13_HUMAN
O75888
0.0010
-0.24
76
EPHB3_HUMAN
P54753
0.001
0.21
77
APBA1_HUMAN
Q02410
0.001
0.25
78
MK10_HUMAN
P53779
0.001
-0.18
79
IL10_HUMAN
P22301
0.001
0.22
80
GDN_HUMAN
P07093
0.001
-0.26
81
HMMR_HUMAN
O75330
0.001
-0.26
82
OLFM4_HUMAN
Q6UX06
0.001
-0.23
83
LYAM1_HUMAN
P14151
0.001
0.21
84
CISY_HUMAN
O75390
0.001
-0.30
85
ID2_HUMAN
Q02363
0.001
-0.28
86
MUTED_HUMAN
Q8TDH9
0.001
-0.29
87
SEPR_HUMAN
Q12884
0.001
-0.25
88
TR10A_HUMAN
O00220
0.002
-0.37
89
K2C8_HUMAN
P05787
0.002
-0.19
90
TNFB_HUMAN
P01374
0.002
0.20
91
ANFB_HUMAN
P16860
0.002
0.19
92
CP1B1_HUMAN
Q16678
0.002
-0.30
93
BRPF3_HUMAN
Q9ULD4
0.002
-0.18
94
GBRB1_HUMAN
P18505
0.002
-0.21
90
Appendix
95
AP4B1_HUMAN
Q9Y6B7
0.002
-0.21
96
SIA7F_HUMAN
Q969X2
0.002
0.18
97
HXC11_HUMAN
O43248
0.002
-0.17
98
PIGC_HUMAN
Q92535
0.002
-0.15
99
TRI22_HUMAN
Q8IYM9
0.002
-0.23
100
OSTP_HUMAN
P10451
0.002
-0.23
101
ZO2_HUMAN
Q9UDY2
0.002
0.19
102
PO2F1_HUMAN
P14859
0.003
-0.23
103
ARI4A_HUMAN
P29374
0.003
0.16
104
KPYM_HUMAN
P14618
0.003
-0.20
105
SHIP1_HUMAN
Q92835
0.003
-0.24
106
K1C18_HUMAN
P05783
0.003
-0.15
107
LYAM1_HUMAN
P14151
0.003
0.20
108
ENOG_HUMAN
P09104
0.004
0.19
109
IFNG_HUMAN
P01579
0.004
-0.24
110
RL29_HUMAN
P47914
0.004
0.24
111
EWS_HUMAN
Q01844
0.004
0.41
112
HSF1_HUMAN
Q00613
0.004
-0.14
113
KLF8_HUMAN
O95600
0.004
0.29
114
TR10B_HUMAN
O14763
0.004
-0.20
115
NFKB2_HUMAN
Q00653
0.004
-0.22
116
NEP_HUMAN
P08473
0.004
-0.19
117
IFM1_HUMAN
P13164
0.005
0.14
118
RPN2_HUMAN
P04844
0.005
0.18
119
TGFA_HUMAN
P01135
0.005
0.18
120
CP3A5_HUMAN
P20815
0.005
0.13
121
CATA_HUMAN
P04040
0.006
-0.47
122
MLH1_HUMAN
P40692
0.006
-0.17
123
MAGA8_HUMAN
P43361
0.006
-0.16
124
P53_HUMAN
P04637
0.006
-0.19
125
IFNA1_HUMAN
P01562
0.006
-0.33
126
KSYK_HUMAN
P43405
0.006
0.18
127
K1C14_HUMAN
P02533
0.007
0.17
128
CCL4_HUMAN
P13236
0.007
-0.23
91
Appendix
129
TR13C_HUMAN
Q96RJ3
0.007
0.26
130
STK19_HUMAN
P49842
0.007
0.15
131
FLOT1_HUMAN
O75955
0.007
0.18
132
FANCC_HUMAN
Q00597
0.007
-0.17
133
Q16590_HUMAN
Q16590
0.007
-0.22
134
EF1G_HUMAN
P26641
0.007
0.14
135
PIR_HUMAN
O00625
0.008
-0.28
136
EDN1_HUMAN
P05305
0.008
-0.18
137
PCNA_HUMAN
P12004
0.008
-0.17
138
S100P_HUMAN
P25815
0.008
-0.15
139
CDC5L_HUMAN
Q99459
0.008
-0.19
140
MUC6_HUMAN
Q6W4X9
0.009
-0.31
141
KLH12_HUMAN
Q53G59
0.009
-0.17
142
DNLI3_HUMAN
P49916
0.009
0.16
143
MAD1_HUMAN
Q05195
0.009
0.18
144
CTRB1_HUMAN
P17538
0.009
-0.14
145
SRGEF_HUMAN
Q9UGK8
0.009
-0.15
146
OSTP_HUMAN
P10451
0.009
-0.15
147
TIMP2_HUMAN
P16035
0.010
-0.22
148
GRPR_HUMAN
P30550
0.010
-0.28
149
VIME_HUMAN
P08670
0.010
-0.21
150
TOP2A_HUMAN
P11388
0.010
-0.23
151
MUC2_HUMAN
Q02817
0.010
0.15
152
MLH3_HUMAN
Q9UHC1
0.010
0.18
153
MOT4_HUMAN
O15427
0.010
-0.19
154
VEGFA_HUMAN
P15692
0.010
-0.29
155
ID1_HUMAN
P41134
0.010
-0.23
156
EI2BA_HUMAN
Q14232
0.011
-0.17
157
HNRPC_HUMAN
P07910
0.011
-0.15
158
CASP9_HUMAN
P55211
0.014
0.30
159
ST14_HUMAN
Q9Y5Y6
0.011
0.18
160
VAS1_HUMAN
Q15904
0.011
-0.19
161
SMAD4_HUMAN
Q13485
0.013
-0.13
162
MTA1_HUMAN
Q13330
0.013
0.14
92
Appendix
163
CAV3_HUMAN
P56539
0.013
0.15
164
PTEN_HUMAN
P60484
0.014
-0.26
165
LUM_HUMAN
P51884
0.014
-0.15
166
TMED2_HUMAN
Q15363
0.014
0.23
167
CD81_HUMAN
P60033
0.014
0.21
168
MD2L1_HUMAN
Q13257
0.014
0.14
169
DC12_HUMAN
Q96FZ2
0.015
0.17
170
ICAM1_HUMAN
P05362
0.015
-0.16
171
TMM54_HUMAN
Q969K7
0.015
-0.14
172
IL1RA_HUMAN
P18510
0.015
0.16
173
TNFA_HUMAN
P01375
0.015
-0.24
174
RBM42_HUMAN
Q9BTD8
0.015
-0.13
175
TNF13_HUMAN
O75888
0.015
0.18
176
RARB_HUMAN
P10826
0.015
-0.14
177
IL6RA_HUMAN
P08887
0.015
0.13
178
TR10A_HUMAN
O00220
0.016
-0.22
179
PHB_HUMAN
P35232
0.017
-0.13
180
CAH2_HUMAN
P00918
0.017
0.15
181
NPT1_HUMAN
Q14916
0.018
-0.19
182
IGF2_HUMAN
P01344
0.018
-0.19
183
MMP1_HUMAN
P03956
0.018
-0.19
184
PSB4_HUMAN
P28070
0.018
-0.14
185
MK03_HUMAN
P27361
0.018
0.20
186
SODC_HUMAN
P00441
0.018
-0.16
187
MCP_HUMAN
P15529
0.019
0.13
188
MTHR_HUMAN
P42898
0.020
-0.12
189
BEX3_HUMAN
Q00994
0.021
0.13
190
GPX4_HUMAN
P36969
0.021
0.18
191
MTA2_HUMAN
O94776
0.021
-0.13
192
IL2_HUMAN
P60568
0.021
-0.23
193
FGF7_HUMAN
P21781
0.021
-0.14
194
ICAM1_HUMAN
P05362
0.022
-0.14
195
1C01_HUMAN
P30499
0.023
0.15
196
HSBP1_HUMAN
O75506
0.024
0.17
93
Appendix
197
TRFE_HUMAN
P02787
0.024
-0.41
198
P5CR1_HUMAN
P32322
0.025
0.16
199
PGH2_HUMAN
P35354
0.025
0.17
200
CATD_HUMAN
P07339
0.027
0.30
201
CCR7_HUMAN
P32248
0.025
0.17
202
SORL_HUMAN
Q92673
0.028
0.15
203
AIFM1_HUMAN
O95831
0.030
0.26
204
CBX3_HUMAN
Q13185
0.029
-0.13
205
L1CAM_HUMAN
P32004
0.031
-0.20
206
BAX_HUMAN
Q07812
0.031
0.16
207
IBP3_HUMAN
P17936
0.031
0.14
208
MK04_HUMAN
P31152
0.031
0.11
209
CFLAR_HUMAN
O15519
0.031
0.15
210
SYSC_HUMAN
P49591
0.031
-0.17
211
IL4_HUMAN
P05112
0.031
-0.14
212
ETS2_HUMAN
P15036
0.031
-0.18
213
ZG16_HUMAN
O60844
0.032
0.11
214
MMP10_HUMAN
P09238
0.032
0.13
215
T4S4_HUMAN
P48230
0.032
0.13
216
P63_HUMAN
Q9H3D4
0.032
0.11
217
BRAF1_HUMAN
P15056
0.032
-0.22
218
RL27A_HUMAN
P46776
0.032
-0.15
219
WNT2B_HUMAN
Q93097
0.033
0.13
220
CAD13_HUMAN
P55290
0.033
-0.14
221
IL1R2_HUMAN
P27930
0.033
0.13
222
HXA9_HUMAN
P31269
0.033
-0.11
223
O14997_HUMAN
O14997
0.034
0.15
224
HERP1_HUMAN
Q15011
0.034
-0.20
225
TIP_HUMAN
Q8TB96
0.034
-0.16
226
NEK2_HUMAN
P51955
0.034
0.12
227
DHRS2_HUMAN
Q13268
0.036
-0.15
228
HMGB1_HUMAN
P09429
0.035
0.14
229
CENPF_HUMAN
P49454
0.035
-0.11
230
PO5F1_HUMAN
Q01860
0.036
0.20
94
Appendix
231
ECHM_HUMAN
P30084
0.036
-0.57
232
TR10B_HUMAN
O14763
0.037
0.22
233
S19A1_HUMAN
P41440
0.036
-0.12
234
FGF2_HUMAN
P09038
0.037
-0.22
235
ABCG2_HUMAN
Q9UNQ0
0.037
-0.12
236
RAB1A_HUMAN
P62820
0.037
0.12
237
TBB5_HUMAN
P07437
0.037
0.15
238
LEUK_HUMAN
P16150
0.039
0.18
239
CKAP2_HUMAN
Q8WWK9
0.039
0.13
240
NUPR1_HUMAN
O60356
0.039
-0.13
241
TNR6_HUMAN
P25445
0.041
-0.29
242
VINEX_HUMAN
O60504
0.040
-0.12
243
PIGT_HUMAN
Q969N2
0.040
0.13
244
CAV2_HUMAN
P51636
0.041
0.18
245
UN93B_HUMAN
Q9H1C4
0.042
-0.16
246
FASTK_HUMAN
Q14296
0.042
0.13
247
VGFR2_HUMAN
P35968
0.042
-0.14
248
SYG_HUMAN
P41250
0.042
0.14
249
TGFB2_HUMAN
P61812
0.042
-0.12
250
IL13_HUMAN
P35225
0.043
0.18
251
S1PR1_HUMAN
P21453
0.043
0.12
252
CCNB1_HUMAN
P14635
0.043
-0.15
253
CD44_HUMAN
P16070
0.043
-0.16
254
ENC1_HUMAN
O14682
0.047
-0.16
255
PLSL_HUMAN
P13796
0.047
-0.11
95
Appendix
7.3 Supplementary figure S1
Supplementary figure S1. Differential abundance of apoptotic-related proteins between
recurrent and non-recurrent non-muscle invasive bladder cancer. Highly abundant proteins were
coloured red and less abundant proteins were coloured green. KEGG pathway analysis software
was used to identify the proteins.
96
Appendix
7.4 Supplementary figure S2
Supplementary figure S2. Differential abundance of proteins between recurrent and nonrecurrent non-muscle invasive bladder cancer in various cancer pathways. Highly abundant
proteins were coloured red and less abundant proteins were coloured green. KEGG pathway
analysis software was used to identify the proteins.
97
Appendix
7.5 Publications based on the thesis
1. Harish Srinivasan, Yves Allory, Martin Sill, Mohamed Saiel Saeed Alhamdani, Francois
Radvanyi, Jörg D. Hoheisel and Christoph Schröder (2013). Prediction of recurrence of non
muscle-invasive bladder cancer by means of a protein signature identified by antibody
microarray analyses. (Submitted- In revision).
2. Harish Srinivasan, Damjana Kastelic, Radovan Komel, Christoph Schröder and Jörg D.
Hoheisel (2013). Immuno-based biomarker identification for effective diagnosis and prediction
of prognosis in gastric adenocarcinomas- an antibody microarray study. (In preparation).
98
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement