Close
Logo
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected 
Invert selection
Deselect all
Deselect all
 Click here to refresh results
 Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
CpG methylation profiling in lung cancer cell lines, tumors and non-tumors
(USC Thesis Other) 

CpG methylation profiling in lung cancer cell lines, tumors and non-tumors

doctype icon
play button
PDF
 Download
 Share
 Open document
 Flip pages
 More
 Summarize
 Download a page range
 Download transcript
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
CpG METHYLATION PROFILING IN LUNG CANCER CELL LINES, TUMORS
AND NON-TUMORS
by

Nikhil Chopra



________________________________________________________________________    



A Thesis Presented to the  
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the  
Requirements for the Degree
MASTER OF SCIENCE
(BIOCHEMISTRY AND MOLECULAR BIOLOGY)


May 2009





Copyright 2009                                                                                             Nikhil Chopra


ii

DEDICATION
This dissertation is dedicated to my loving parents for raising me to what I am today, for
believing and having deep faith in everything I do, for always supporting me and
boosting my confidence. I also dedicate this to my brother, Kapil and my sis-in-law,
Priya for their unconditional support and love.  












iii

ACKNOWLEDGEMENTS
From the starting stages to the final draft, I owe an immense debt of gratitude to my
supervisor Dr. Ite Laird-Offringa. This thesis would have never finished without her
continuous support. I am thankful for her continuous patience in teaching me and guiding
me. She is always there for all help. I respect her as great mentor and who knows well
how to maintain a perfect balance in life.
I want to acknowledge Meleeneh Kazarian and Janice Galler for teaching the basics of
many techniques on top of their own loads of work. If I ever did a PhD, my routine would
be combination of theirs. I want to thank Paul Anglim for all the mini mistakes he had to
correct over and over in editing my protocols and for helping me prepare for my defense.
I want to specially thank Toshi Hinoue for his continuous advice on my research work,
for being a great friend and for answering all my “quick questions”. I want to thank
Suhaida Selemat for always pushing me to write thesis and helping me in learning JMP
and doing many of the statistical analyses. I want to thank Pohan, Candace, David and
Daniel for being great lab-mates to work in especially in late hours of the day. I want to
thank my lab-members and Laird lab members for being such a fun lab to work with and
from whom I learnt a lot of technical, cultural and worldly stuffs.  
I want to thank Dr. Tokes for teaching me and encouraging me throughout my MS. I still
remember him mentioning me how much we have changed and learned over the span of
two years. I am grateful to him for accepting me in the Biochemistry program.
iv

I want to thank Dr. Allen Yang for being on my committee and for being a helpful
neighboring lab. I am grateful to him for providing me insights for different tests and for
providing me DNA methylation data.
Last but not the least, I want to thank my family, my dear friends and my room-mates
Hari and Sumant for who have made these 3 years of fun and study, a memorable one.












v

TABLE OF CONTENTS                                                                                             Page  
DEDICATION ii
ACKNOWLEDGEMENTS                                                                                        

iii
LIST OF TABLES                                                                                                     vi
LIST OF FIGURES                                                                                                   vii

ABSTRACT    

vii
CHAPTER 1: INTRODUCTION        
                                                                     
(1) Lung Cancer                                                                                                            
(2) Types of Lung Cancer                                                                                            
(3) DNA Methylation
(4) Studying DNA Methylation in Lung Cancer                                                                                                                                                        
1

1
1
2
4

CHAPTER 2: MATERIALS AND METHODS          
                                             
(1) Cell Lines and DNA Isolation                                                                                
(2) Illumina Goldengate Methylation Assay                                                                
(3) Statistical Analysis
(4) Panther Analysis for Molecular Functions and Pathways                                                                                                                                

8

8
8
10
11

CHAPTER 3: RESULTS AND DISCUSSION    
                                                     
(1)  Clustering of Similar Samples in a Tight Group and Separation of Tumor          
and Non-tumors
(2)  Cell Lines Represent Tumor-specific Hypermethylation                                      
(3)  At least 10 Genes Represent Cell Line-specific Hypermethylation                      
(4)  Hypomethylation in Cell Lines                                                                              
(5)  Molecular Function of Genes Hypermethylated or Hypomethylated in                
Cancer Cell Lines
(6)  Differential Methylation of WNT Pathway Genes in Cell Lines                          

12

13

18
23
35
42

45

BIBLIOGRAPHY   52
 
                                                                                     
                                                                                                           

vi

LIST OF TABLES                                                                                                 Page
Table 2.1 Table 2.1 Adenocarcinoma Cell Lines                                                      8                
Table 3.1 Significantly Hypermethylated Loci in Adeno Tumors                           20
Table 3.2 Significantly Hypermethylated Loci in Cell Lines (CL)                          25
Table 3.3 Significantly Hypermethylated Genes at all Studied Loci in                   28
Cell lines  

Table 3.4 Loci Hypermethylated in Cell lines with Other Locus                             34  
Hypermethylated in Tumors
Table 3.5 Significantly Hypomethylated Loci in Cell line                                       37
Table 3.6 Genes Significantly Hypomethylated at all Studied Loci                        38    
in Cell Lines
Table 3.7 Molecular Function of Genes                                                                   44
Table 3.8 Differentially Methylated Genes in WNT Pathway                                 48










vii

LIST OF FIGURES                                                                                                 Page
Fig. 1.1 Stage of Cancer Diagnosis                                                                            3
Fig 1.2: Modification of Knudson Two Hit Hypothesis                                             5
Fig. 3.1 Hierarchical clustering of studied samples                                                   15
Fig. 3.2 Two-way hierarchical clustering                                                                  16
Fig 3.3 Flowchart for identification of loci hypermethylated in tumor and               19  
their comparison with cell lines

Fig. 3.4 Loci Significantly Hypermethylated in Cell lines                                         29
Fig. 3.5 Genes Hypomethylated in Cell lines at all Their Studied CpG Loci            39
Fig 3.6 WNT Signaling Pathway (Adapted from KEGG)                                         49












viii

ABSTRACT
Cell lines are widely used for studying DNA methylation in cancer. Their main
applications are in the development of biomarkers, in pharmacological studies and for
studying the effect of demethylating agents and for studying the effect of methylation on
biological pathways. Studies have compared the methylation status in cell lines to
primary tumors in various cancers. Conflicting conclusions have been made as to whether
cell lines represent the methylation observed in tumors.
In this study I have used the Illumina Goldengate assay to compare the methylation status
of 1506 CpG loci (807 genes) in adenocarcinoma cell lines, adeno and squamous frozen
tumors and non-tumor lung samples. Hierarchical clustering was performed to study the
methylation profile in each of these samples and to study the degree of similarity between
adenocarcinoma cell lines and frozen tumors. Further performing parametric tests, I
observed that cell lines do exhibit cancer-specific hypermethylation, closely resembling
that of frozen tumors. On performing Wilcoxon rank sum test on tumors and cell lines at
all loci and applying additional filters on loci that were significantly different, I observed
that 10 genes were significantly hypermethylated and 5 genes were significantly
hypomethylated in cell lines as compared to adeno tumors at more than one locus of the
gene. Hence, cell lines may show differences in degree of methylation for few genes.  
Based on the assumption that this cell line specific hypomethylation might be associated
with expression of genes and cell line-specific hypermethylation to gene silencing, I tried
to deduce their effect on biological pathways. I conclude that cell lines would be
ix

appropriate models for studying hypermethylation and for development of biomarkers but
might not be appropriate models for studying WNT pathway.  















1

CHAPTER 1 INTRODUCTION  
Lung Cancer remains the leading cause of cancer deaths among men and women in the
United States and worldwide. It is the second most common death only next to heart
disease in the United States (14).
It is estimated that in 2008 there were 215,020 new cases of lung cancer in the United
States and 166,280 deaths (14). Lung cancer is one of the leading two cancers that still
shows an increase in death rate from 37.61 per 100,000 in 1991 to 40.87 per 100,000 in
2004 and is responsible for the increased portion of deaths due to cancer among females
(14).
As shown in Fig 1.1, most lung cancer cases are diagnosed at a late stage in comparison
to other cancers like breast and prostate. Only 16% of lung cancer cases are diagnosed
while the cancer is still localized and highly curable (14). The majority of the cases are
diagnosed at regional and distant stages at which the tumor is unresectable and options
are limited to suboptimal therapeutics. Hence, there is an urgent need to develop
techniques that allow identification of lung cancer at an early stage.
Lung Cancer
Based on histology and type of treatment, lung cancer is broadly classified into two
types: Small Cell Lung Cancer (SCLC) and Non Small Cell Lung Cancer (NSCLC).
SCLC comprises about10-15% of all lung cancer cases and has been primarily related to
Types of Lung Cancer
2

smoking. NSCLC occurs in both smokers and never smokers and comprises about 85 %
of all lung cancer cases (34). NSCLC is further classified as Adenocarcinoma, Squamous
Cell Carcinoma, Large Cell Carcinoma and others. Squamous Cell Carcinoma and Small
Cell Lung Cancer arise centrally whereas Adenocarcinoma and Large Cell Carcinoma
arise peripherally (15). Adenocarcinoma comprises about 30-40 % of all lung cancer
cases and is more common among females and non-smokers (30). Hence, even though
there has been a decrease in the number of smokers it is necessary to have diagnostic
technique for early identification of adenocarcinoma. DNA methylation holds a great
promise for developing biomarkers for early diagnosis of lung cancer (2). Studies have
shown that DNA methylation might play a very important role in progression of cancer
(10, 34).    
In mammalian cells, DNA methylation occurs at carbon-5 position of cytosine in CpG
dinucleotides. In normal cells, it is required for proper mammalian development, X-
chromosome inactivation in females and for genomic imprinting to ensure monoallelic
expression. Repetitive and parasitic sequences are also hypermethylated to prevent
transcription and chromosomal instability due to translocation. Many genes requiring
tissue-specific expression are also regulated by hypermethylation (9). In general,
globally, tumors show a lower level of DNA methylation than their normal counterparts.
The loss of methylation is mainly due to hypomethylation of repetitive DNA sequences
like Alu and L1 and general loss of DNA methylation at non-clustered CpGs (2). This  
DNA Methylation
3

 
Fig. 1.1 Stage of Cancer Diagnosis (14)
4

hypomethylation is thought to cause genomic instability and increased movement of
transposons which can cause insertions and deletions and hence affect healthy genes.
Along with global reduction in DNA methylation, tumors also show a local increase in
DNA methylation, observed at CpG rich regions of promoters, called CpG islands.
Methylation of cytosine in the promoter region leads to recruitment of various proteins
like methyl-binding proteins and histone deacetylases. As a result of binding of these
proteins, DNA is not available for binding to transcription factors and hence gene
expression is inhibited. Various tumor suppressor genes in different cancers have been
identified whose expression is inhibited due to DNA hypermethylation (24). Fig 1.2
represents the modification of Knudson two hit hypothesis where one or both of the
alleles might be knocked down by DNA hypermethylation (18).
A large number of studies have now established that different cancers have distinct genes
that are deregulated by DNA hypermethylation (8, 11). Analysis of these genes provides
an opportunity to study the function of DNA hypermethylation in cancer and also
potentially test these genes as biomarkers for early detection of the disease.
Once it was established that aberrant patterns in DNA Methylation may play an important
role in development and progression of cancer, many studies have been undertaken to
understand these differences and explore their applicability for tumor identification and
development of new therapeutic applications. Scientists have used both primary tumors  
Studying DNA Methylation in Lung Cancer

5


Fig 1.2: Modification of Knudson Two Hit Hypothesis (18)  
and cell lines for their experiments to understand methylation, to study the effect of
demethylating agents and to develop biomarkers for early detection of cancer. Primary
tumors obtained after surgery and autopsy would seem to be the most ideal samples to
study. However, tumor samples provide a very limited supply of DNA. In addition, often
the tumor tissue is contaminated with neighboring non-tumor tissue such as stroma, white
blood cells and vasculature. Another disadvantage is that tumor tissue cannot be
propagated and therefore cannot be used to test for response to therapeutic drugs.
Cell lines offer the advantage of a virtually limitless supply of DNA, homogeneity of
tissue and the possibility of culturing in vitro allowing the testing of drugs. They are  
6

easily transportable and can be grown to generate a large amount of DNA. Cell lines can
be assumed to provide consistent DNA samples over time. They also provide easy access
to many different labs to experiment on the same material. This allows a more
appropriate comparison of data between different studies done using the same or different
techniques. Hence, any observed changes could be attributed to differences between the
techniques and not because of differences in samples. Also experiments done on the same
cell line would lead to increased knowledge of characteristics, which could be very useful
to a researcher for designing his/her future experiments and drawing conclusions. This is
of great importance as many studies in cancer have now shown that tumors are
heterogeneous and even two tumors of the same type are not totally identical (34).  
Though cell lines provide the aforesaid advantages, researchers have observed aberrant
and excessive DNA hypermethylation in cell lines in comparison to tumor tissue (28, 30).
The degree of difference in methylation has been observed to be inconsistent for different
cancers (25). Most researchers have studied the methylation status of a few genes. As a
result, conflicting conclusions have been drawn about whether cell lines make
appropriate models for studying DNA methylation (20-22, 25, 28, 30, 31). Most studies
agree that cell lines represent the tumor specific hypermethylation; however the
methylation status of other not so cancer-specific genes and the difference in degree of
methylation in the cancer-specific genes is a matter of debate (20-22, 25, 28, 30, 31).
With increased of cell lines and increased application of DNA methylation for
identification and therapeutic applications it becomes necessary to understand how
accurately a particular cell line represents the general population of tumors. It is
7

necessary for a researcher to differentiate between the changes induced by cell culture
and those as a result of cancer development to design the experiments and accurately
comprehend their results.
The aim of this study is to try to answer the old controversy whether cell lines are good
models to study DNA methylation. We therefore studied the differences in methylation
profile in adenocarcinoma cell lines, tumors and non-tumor samples. Using non-
parametric tests we have identified genes that tend to become differentially methylated in
cell lines as compared to tumors and non-tumors.  
In conclusion, based on our analysis we assume that adenocarcinoma cell lines are very
similar to adenocarcinoma tumors and would be appropriate models for most studies with
respect to DNA methylation. Some biological pathways like Wnt pathway may be altered
in these cell lines. However, additional studies on more samples on gene expression and
DNA methylation at all CpG loci of promoters of the differentially methylated genes
need to be done before the difference in biological roles can be concluded.






8

(1)
CHAPTER 2 MATERIALS AND METHODS

All the six cell lines (H358, H1395, H2347, H1975, A549 and H2228) had been bought
previously from ATCC (Manassas, VA 20108) and stored under nitrogen at -196
o
C. The
cell lines were grown in media as per instructions from ATCC. Sex, smoking status,
cancer stage and KRAS, EGFR and p-53 mutations information is provided from
previous studies in Table 2.1. DNA was extracted using proteinase K digestion and
purified as per standard phenol chloroform method (19).
Cell Lines and DNA Isolation
Cell line
Table 2.1 Adenocarcinoma Cell Lines
Sex
Smoking
Status Cancer stage-type
KRAS
mut EGFR mut p-53
H1975 F NS Adeno WT
L858R,
T790M  
H358 M   BAC WT WT deletion
H2228 F NS Adeno WT    
A549 M   Lung Carcinoma Mut WT WT
H2347 F NS Adeno stage 1   WT  
H1395 F S Adeno Stage 2 WT WT  


(2)
All the samples were bisulphite converted before running them on Goldengate platform
(Epigenome Centre, USC). In bisulfite treatment unmethylated cytosines are deaminated
and converted to uracil however methylated cytosines are not. Uracil has similar
hybridization properties as thymine. Hence when followed by PCR or any other method
that can distinguish between cytosine and thymine we can detect methylation.    
Illumina Goldengate Methylation Assay
9

The Goldengate assay is a high throughput single nucleotide polymorphism genotyping
system and has been adapted to detect methylation using bisulfite converted DNA.
The assay allows measurement of 1506 targeted CpG sites from 807 genes looking
specifically in promoter or first exon regions. For each CpG site, four probes have been
designed: two allele specific oligos (ASO) and two locus specific oligos (LSO). Each
ASO - LSO oligo pair corresponds to either the methylated and unmethylated state of the
CpG site.
The methylation status of an interrogated CpG site is determined by calculating the
intensity of methylated (M) and unmethylated (U) alleles. After measurement of
fluorescence intensities from methylated and unmethylated probes, background intensity
computed from a set of negative controls is subtracted from each analytical data point.
Hence, we get the corrected intensities, M and U for methylated and unmethylated
probes, which can be positive negative or zero based on the measured values. The degree
of methylation is calculated by choosing the maximum of 0 and corrected intensities.
Hence, the degree of methylation, β is calculated using the formula:
Methylation Value (β) =    

              Max (M, 0) _________  
                                            [Max (M, 0) +U (M, 0) + 100]  
β provides a continuous measure of level of DNA methylation ranging from 0 in case of
no significant detectable methylation to 1 for completely methylated DNA sample. A
constant bias of 100 is added to denominator to regularize β when both U and M are
small. Usually, values of intensities are in tens of thousands.
10

(3)
All the statistical calculations and the hierarchal clustering were done using the software
JMP 7.0 (SAS Campus Drive, Building S, Cary, NC, 27513). First all the probes from the
X chromosome were removed as some samples were from males and some from females
in each group. Five other probes that failed on most samples were removed as well.
The next step was to calculate the principal components of analysis (PCA). This was
done under analyze menu and multivariate methods sub menu. This helps in better
visualization of the genes in a cluster. Subsequently, hierarchical clustering was done
using Ward’s function with sorting based on PCA.
Identification of loci hypermethylated in tumors was done using Wilcoxon rank sum test
on JMP. It was performed on all 1412 CpG loci between adeno tumors and non tumors.
Benjamini Hochberg correction was applied and all probes with a p-value lesser than 0.05
were selected. Using excel, a selection filter was used to identify  loci whose average
methylation of adeno tumor samples in comparison to average methylation of non-tumor
samples was at least 2 fold higher and difference in methylation was greater at least by a
value of 0.17. Wilcoxon rank sum test was applied on these selected loci for differences
between adeno tumors and cell lines.  
Statistical Analysis
Identification of genes excessively hypermethylated in cell lines was done by using
Wilcoxon rank sum test on the1412 loci between adeno tumors and cell lines. Benjamini
Hochenberg correction was applied and all probes with a p-value lesser than 0.05 were
selected. Using excel, a selection filter was used to identify  loci whose average
methylation of cell lines in comparison to average methylation of adeno tumor samples
11

was at least 2 fold higher and difference in methylation was greater at least by a value of
0.17. Identification of hypomethylated loci was done using the loci that were significantly  

different and whose average methylation in tumors was at least 2 fold higher in tumors as
compared to cell lines and at least greater by a value of 0.17 in adenocarcinoma tumors.  

(4)
Panther stands for Protein ANalysis THrough Evolutionary Relationships. It is a free
online database and classifies genes by their functions using published scientific evidence
and even predicts functions based on evolutionary relationships.
A “Batch Id search” was done using Panther’s gene menu. The gene set was input for all
Illumina genes, hypermethylated genes and hypomethylated genes which were found to
be statistically significant and met the filtering criteria. Panther tools were used to
classify genes based on molecular functions and role in Pathways. Pathways that showed
6 or more genes from hypermethylated and hypomethylated cluster were selected for
further analysis.
Genecards was used to identify the function of different genes in the pathway. Diagrams
of Wnt signaling pathway was obtained from database KEGG. UCSC genome browser
was used to identify the location of the CpG islands foe the studied gene and its CpG
locus.  

Panther Analysis For Molecular Functions And Pathways

12

The aim of this project is to compare the DNA methylation profile between non-tumor
lung samples, lung tumor samples and lung cancer cell lines which are used as a model
for studying lung cancer. We specifically studied the differences in DNA methylation
status in adenocarcinoma cell lines and lung adenocarcinoma frozen tumors to answer the
question whether cell lines are appropriate models for studying DNA methylation. Hence,
DNA methylation status was evaluated on 5 lung adenocarcinoma cell lines, 1 bronchio-
alveolar carcinoma cell line, 5 paired lung adenocarcinoma frozen tumors and adjacent
non-tumors, 10 paired lung squamous tumors and adjacent non-tumors and 6 cadaver-
derived lung samples. The adeno tumor samples were obtained from Dr. Janice Galler,
squamous tumor samples from Dr. Paul Anglim and cadaver-derived lung samples from
Hyang Ming and Dr. Allen Yang (Norris Cancer Centre, USC).
X-linked CpG loci were removed from the study as some samples were from males and
others from females. This was necessary as CpG loci on X-linked genes are methylated in
one of the two X chromosomes in females. Hence a clustering using them would have
been strongly influenced based on sex. In addition, 5 loci for which probes had failed in
many samples were removed. This reduced the number of studied CpG loci to 1412.


CHAPTER 3 RESULTS AND DISCUSSION

13

In order to gain an overview of methylation profiling and to study the degree of similarity
and dissimilarity between the samples, a two-way hierarchical clustering was done using
Wards function on software JMP 7.0.
This also helped us to study and observe how closely different samples are related to each
other. Figure 3.1 shows only the clustering of samples. Figure 3.2 shows the two-way
hierarchical clustering result with samples and studied CpG loci. The samples include
non-tumor cadaver-derived lungs (Nor Cad), adeno adjacent non-tumor (Adj AD),
squamous adjacent non-tumor (Adj SQ), adeno frozen tumors (AD T), squamous frozen
tumors (SQ T) and cancer cell lines (A549 and those with prefix H). Most of the samples
of a particular type (tumor/non-tumor/adjacent non-tumor) clustered together. This
confirms the hypothesis that DNA methylation status can be used to differentiate between
non-tumor and tumor lung.  
In Fig. 3.1, sections A-G have been marked based on how tightly different samples
clustered together in a group. The designation is random, not based on any mathematical
calculations and made to simplify discussion.
3.1 Clustering of Similar Samples in a Tight Group and Separation of Tumor and Non-
tumors
As shown in section A, five of the six non-tumor cadaver lung samples clustered together
showing a strong correlation and confirming that they are indeed similar. Along with
them clustered an adjacent non-tumor sample showing a distinctive lower methylation
when compared with other non adjacent samples. One of the non-tumor cadaver lung
14

sample (Case no. 5) was diseased with AIDS and clustered separately in a group with
adjacent non-tumor lungs as shown in section D. This non-clustering could have been
because of presence of some hypomethylated sites specific to this sample. This consisted
of 24 loci which were hypomethylated only in this sample.  
As show in section B, four of the five adeno adjacent non-tumor lungs clustered tightly
together but separately from squamous non-tumors. As shown in section C, nine of the
ten squamous adjacent non-tumors clustered tightly together in a group showing a
different methylation profile from adeno non-tumors and cadaver lungs. This could be
due to reason that adeno tissues in lungs have a specific methylation pattern and different
from squamous tissues. This might also be due to different smoking exposure. The
squamous tissue in adjacent non-tumors would be exposed to more smoke, causing a
heavy methylation in proximal tissues and therefore in squamous adjacent samples.
Clustering of tumors and cell lines is shown in sections E-G. As shown in sections E and
F, similarly to squamous adjacent non-tumor samples, nine out of the ten squamous
tumor samples clustered together. This supports the idea that squamous tumors have a
methylation profile distinct from adeno tumors and so it may be possible to have a
different set of biomarkers specific to each tumor subtype.
Adeno tumors clustered most diversely of all samples. As shown in section E, three out
of the five adeno tumors clustered together. As shown in section G, two others clustered
together and were more closely related to adeno cell lines. These two tumors showed a  
15



Nor Cad - Cadaver non-tumor lung
Adj AD- Adeno adjacent non-tumor
Adj SQ-Squamous adjacent non-tumor
AD T- Adeno frozen tumor
SQ T- Squamous frozen tumor
A549- Adeno Cell line
H- Prefix for other cell lines





Fig. 3.1 Hierarchical clustering of studied
samples. Section A-G are marked based
on similarity of different samples in a
group and is not based on any numerical
calculations





















16




Fig. 3.2 Two-way hierarchical clustering. Red box represents the cluster for loci that tend
to be hypomethylated in cell lines.


17

higher degree of methylation at some loci as compared to other adeno tumors and showed
methylation more similar to adeno cell lines.
A reason for lower correlation observed among tumors in comparison to non-tumors
could be that most of the cancer-related genes in tumors are secondary, meaning that they
are important but not necessarily required for tumorogenesis. There can be an alternative
secondary gene whose deregulation by hypermethylation or mutation would suffice for
the pathway alteration. Hence, distortion in a pathway might be brought by deregulation
of different genes. However, the net effect or alteration in the pathway remains the same.  
As shown in section G, all the adenocarcinoma cell lines clustered together. However, the
cell lines were connected by greatest distance (longest bars) on the dendogram scale. This
represents that degree of similarity among different cell lines was least. The correlation
between the same samples within a sub-cluster was higher in cadaver derived non-tumors
and adjacent non-tumor samples. The correlation between tumor samples was
intermediate and could be due to reasons explained above.
The lesser correlation observed in cell lines could be explained by many theories. First
cell lines are grown in vitro and in a different environment than tumors. Many of the
pathways are not required for growth in vitro and hence the genes involved in these
pathways might have been shut down by methylation over time. Secondly, cell lines are
homogenous and may represent a clone derived from few cells. Thus, the methylation
patterns of different cell lines might reflect divergent groups of cells. It is also possible
that some of the studied loci might have changed their methylation status over different
18

passages. However this change may or may not have caused a change in expression of
genes or in pathways.
In order to further address the question, if cell lines are appropriate models for studying
DNA methylation in lung cancer, I performed a statistical analysis to study if cell lines
represent DNA hypermethylation similar to one observed in tumors. As shown in
flowchart Fig. 3.3, I performed Wilcoxon rank sum test between adeno tumors and non-
tumor samples. I further applied Benjamini Hochberg correction to get corrected p-values
and found 334 statistically significant CpG loci. A selection was made for loci for which
the average methylation of adeno tumor samples (β
ADT
) in comparison to average
methylation of non-tumor samples (β
NT
) is at least 2 fold higher and the difference in
methylation is greater at least by a value of 0.17 (3). As a result, I identified 81 loci
tabulated in Table 3.1. I further compared these 81 loci between adenocarcinoma cell
lines and adenocarcinoma tumors by applying Wilcoxon rank sum test and Benjamini
Hochberg correction. 15 loci were observed to be significantly different between cell
lines and tumors. All of these 15 loci had a higher degree of methylation in cell lines in
comparison to tumors. This shows that cell lines represent the hypermethylation observed
in adenocarcinoma tumors. On further interrogating these 15 significant loci, I observed 5
loci had a CpG locus representing the genes ASCL2, CDH13, PENK, TWIST1 and WT1
and highlighted in yellow in Table 3.1. These genes were observed to have another
studied CpG locus (highlighted in green) that was significantly different between tumors  

3.2 Cell Lines Represent Tumor-specific Hypermethylation
19

Wilcoxon Rank test to identify loci
between ADT and NT which are
statistically significant (p -value < 0.05)
after Benjamini Hochberg correction
15 statistically significant loci
1412 studied loci
334 statistically significant loci
81 hypermethylated loci
identified in tumors
Selection of loci with
β
avg ADT
/β
avg NT  
> = 2
β
avg ADT
- β
avg NT
> =  0.17
Wilcoxon rank sum test ADT vs. CL
66 statistically non-significant loci
5 loci with a different
CpG locus of same gene
in 66 non-sig. loci
10 loci representing genes  
not common to genes from
the set of 66 non-
significant  loci
Wilcoxon Rank test to identify loci
between ADT and CL which are
statistically significant (p value <
0.05) after Benjamini Hochberg
correction




Fig 3.3 Flowchart for identification of loci hypermethylated in tumor and their
comparison with cell lines














20

a
Loci highlighted in yellow are significantly different between tumors and cell lines and
correspond to same genes as highlighted in green.
Table 3.1 Significantly Hypermethylated Loci in Adeno Tumors (ADT)

8l significantly
hypermethylated loci in
ADT
a
Mean
NT
(β
NT
)
Mean
ADT
(β
ADT
)
Mean
CL
(β
CL
)
p-value
ADT vs CL
after
correction
p-value
ADT vs.
NT after
correction
MFAP4_P197_F 0.32 0.71 0.94 0.009 0.002
MMP2_P303_R 0.04 0.27 0.87 0.010 0.046
GAS7_E148_F 0.16 0.36 0.88 0.010 0.034
HOXA11_P698_F 0.12 0.70 0.94 0.030 0.002
TPEF_seq_44_S88_R 0.08 0.54 0.85 0.030 0.002
ASCL2_P360_F 0.14 0.55 0.90 0.031 0.024
FRZB_E186_R 0.15 0.48 0.88 0.031 0.004
HOXA9_E252_R 0.13 0.58 0.96 0.031 0.007
HOXA9_P303_F 0.11 0.35 0.77 0.032 0.015
PENK_P447_R 0.05 0.44 0.79 0.032 0.002
WT1_P853_F 0.03 0.47 0.84 0.032 0.002
HOXA9_P1141_R 0.17 0.62 0.90 0.041 0.015
MOS_E60_R 0.10 0.66 0.89 0.046 0.002
CDH13_E102_F 0.04 0.51 0.84 0.049 0.002
TWIST1_P44_R 0.01 0.35 0.75 0.050 0.002
MYH11_P22_F 0.03 0.26 0.67 0.071 0.014
HOXC6_P456_R 0.14 0.33 0.66 0.093 0.040
ISL1_E87_R 0.06 0.23 0.67 0.095 0.020
MYOD1_E156_F 0.06 0.57 0.85 0.095 0.002
WT1_E32_F 0.01 0.36 0.80 0.119 0.002
SOX17_P287_R 0.14 0.63 0.86 0.144 0.002
ASCL2_E76_R 0.11 0.46 0.72 0.155 0.017
HTR1B_E232_R 0.11 0.43 0.68 0.157 0.020
P2RX7_P119_R 0.08 0.31 0.70 0.158 0.002
HS3ST2_P171_F 0.08 0.64 0.86 0.162 0.002
ISL1_P554_F 0.04 0.36 0.72 0.163 0.008
SOX17_P303_F 0.09 0.46 0.77 0.164 0.003
SLC22A3_E122_R 0.04 0.23 0.05 0.199 0.019
CDH13_P88_F 0.24 0.58 0.82 0.201 0.003

   
21

Table 3.1, Continued

    HS3ST2_E145_R 0.11 0.82 0.92 0.248 0.002
TAL1_P594_F 0.13 0.63 0.87 0.202 0.029
GALR1_E52_F 0.02 0.28 0.60 0.257 0.018
HS3ST2_P546_F 0.09 0.52 0.71 0.271 0.002
RARA_P1076_R 0.25 0.56 0.79 0.275 0.010
FGF2_P229_F 0.11 0.57 0.76 0.334 0.005
NEU1_P745_F 0.02 0.27 0.06 0.336 0.006
SOX1_P1018_R 0.01 0.41 0.73 0.337 0.002
DCC_P177_F 0.03 0.23 0.47 0.341 0.017
DLK1_E227_R 0.06 0.50 0.82 0.342 0.002
TWIST1_P355_R 0.09 0.47 0.71 0.345 0.002
DBC1_P351_R 0.04 0.37 0.66 0.429 0.006
DES_E228_R 0.06 0.27 0.61 0.430 0.046
TWIST1_E117_R 0.01 0.60 0.76 0.440 0.001
DCC_E53_R 0.21 0.51 0.72 0.445 0.012
CSPG2_E38_F 0.05 0.27 0.17 0.540 0.033
AGTR1_P41_F 0.01 0.29 0.64 0.661 0.029
CALCA_E174_R 0.15 0.48 0.66 0.676 0.015
NPR2_P618_F 0.08 0.35 0.61 0.685 0.024
GABRB3_E42_F 0.04 0.22 0.56 0.697 0.046
GAS7_E148_F 0.02 0.32 0.64 0.698 0.002
IGF2AS_P203_F 0.30 0.68 0.87 0.856 0.029
SFRP1_E398_R 0.04 0.25 0.61 0.864 0.027
SPARC_P195_F 0.13 0.56 0.71 0.886 0.003
NGFR_E328_F 0.09 0.26 0.47 1.043 0.020
CHGA_E52_F 0.03 0.43 0.69 1.064 0.002
NTRK3_E131_F 0.00 0.22 0.55 1.072 0.013
ADCYAP1_P398_F 0.03 0.37 0.64 1.081 0.005
ISL1_P379_F 0.07 0.41 0.62 1.107 0.016
ONECUT2_E96_F 0.07 0.30 0.25 1.287 0.033
RET_seq_54_S260_F 0.01 0.26 0.55 1.359 0.002
IGF2_E134_R 0.06 0.28 0.32 1.761 0.036
MDR1_seq_42_S300_R 0.00 0.33 0.61 1.777 0.001
MMP2_E21_R 0.04 0.25 0.49 1.838 0.008
PENK_E26_F 0.13 0.50 0.62 1.861 0.002
     
22

Table 3.1, Continued
   

   
MMP2_E21_R 0.04 0.25 0.49 1.838 0.008
SFRP1_P157_F 0.03 0.33 0.63 1.788 0.005
FLT3_E326_R 0.04 0.56 0.71 2.959 0.002
TERT_E20_F 0.02 0.19 0.13 2.995 0.017
HGF_E102_R 0.19 0.57 0.64 3.076 0.002
TUSC3_E29_R 0.07 0.38 0.33 4.443 0.003
MYOD1_P50_F 0.08 0.27 0.46 4.634 0.002
NEFL_E23_R 0.08 0.55 0.47 4.995 0.002
NEFL_P209_R 0.15 0.76 0.49 5.022 0.002
TUSC3_P85_R 0.05 0.26 0.26 5.135 0.002
CCNA1_E7_F 0.06 0.23 0.19 6.674 0.019
ASCL1_P747_F 0.24 0.49 0.50 9.002 0.002
ADTP10A_P147_F 0.12 0.43 0.49 9.076 0.004
FZD9_E458_F 0.25 0.59 0.56 21.460 0.010
SOX1_P294_F 0.08 0.58 0.53 23.376 0.002
IGFBP3_P423_R 0.05 0.25 0.33 26.186 0.016
DCC_P471_R 0.05 0.23 0.33 74.316 0.002
EYA4_E277_F 0.06 0.60 0.60 88.250 0.002
























23

and non-tumors and had a statistically insignificant difference in methylation between
cell lines and tumors and hence were classified in the 66 non-significant loci mentioned
above. This shows that cell lines represent almost all the hypermethylation observed in
tumors, though the degree of methylation may be higher at some CpG loci in cell lines. It
needs to be studied in addition whether there is a biological effect of the observed higher
degree of methylation at these 15 loci and the differences in gene expression of the
corresponding genes.
In conclusion, adenocarcinoma cell lines maintained all the characteristic
hypermethylation observed in adenocarcinoma tumors. The 15 loci that showed a
significant difference had a higher degree of methylation in cell lines as compared to
tumors which may or may not have a biological significance. This supports the idea that
cell lines would be appropriate models for studying DNA methylation in lung cancer.
In order to further answer the old controversy of differential methylation observed in cell
lines, I performed a Wilcoxon rank sum test on all the studied 1412 CpG loci between
adenocarcinoma tumors and adenocarcinoma cell lines. I applied Benjamini Hochberg
correction and found 261 statistically significant loci. In order to identify the loci that are
significantly hypermethylated in cell lines, I selected genes for which mean methylation
in cell lines (β
CL
) was higher than the mean methylation in adeno tumors (β
ADT
) at least
by a value of 0.17 and a methylation fold ratio greater at least by a factor of 2. Seventy
one loci encoding 55 genes were observed to be significantly hypermethylated in cell
3.3 At least 10 Genes Represent Cell Line-Specific Hypermethylation
24

lines in comparison to adenocarcinoma tumors and non-tumor samples. Table 3.2
summarizes the loci and the results.  
Ten of these 55 genes were observed to be hypermethylated at both the studied CpG loci.
The loci corresponding to these genes are shown in Table 3.3 and scatter plots of degree
of methylation of different samples of these loci are shown in Fig 3.4. Since 8 of these 10
genes (except KCKN4 and WNT10B) were hypomethylated in tumors and non-tumor
samples, it is possible that hypermethylation of these genes may provide an in vitro
growth advantage. It is also possible that these genes are non-essential in cell lines as
well as tumors but are more likely to be tolerated in vivo than in vitro.
On further interrogating these 55 genes, I observed 11 genes were significantly
hypermethylated in adenocarcinoma tumors and cell lines at another CpG locus of the
same gene. These genes and the corresponding loci are shown in Table 3.4. Hence, cell
lines may gain significant hypermethylation at an additional CpG locus of genes observed
to be hypermethylated in tumors. It is possible that tumors may gain hypermethylation at
these loci at later stages and hence this significant hypermethylated loci observed in cell
lines may be a representative of tumors. Also, another possible reason for these observed
differences may be limited number of adeno tumor samples used in this study.  
Thirty two of these 55 genes showed variant hypermethylation in cell lines. At least two
CpG loci were studied for each of these 32 genes, however only one of the two loci was
observed to be significantly hypermethylated in cell lines. Since, none of these loci had
significant loss of hypermethylation in adeno tumors as compared to non-tumors, these  
25


Table 3.2 Significantly Hypermethylated Loci in Cell Lines (CL)
71 Significantly
Hypermethylated Loci
p-value
ADT vs. CL
after
correction
Mean
Adeno
tumors
(β
ADT
)
Mean Cell
lines (β
CL
)
Ratio
CL/A
DT
(β
CL
/
β
ADT
)
Differenc
e CL-
ADT
(β
CL
–
β
ADT
)
ROR2_P317_R 0.007 0.004 0.267 66.67 0.26
GSTP1_seq_38_S153_R 0.007 0.002 0.230 115.00 0.23
LIF_E208_F 0.007 0.004 0.492 122.92 0.49
MYH11_P236_R 0.007 0.014 0.557 39.76 0.54
MYCN_P464_R 0.008 0.006 0.180 30.00 0.17
CCKBR_P361_R 0.008 0.006 0.235 39.17 0.23
IGFBP7_P371_F 0.008 0.004 0.335 83.75 0.33
MGMT_P281_F 0.009 0.010 0.252 25.17 0.24
WNT10B_P823_R 0.009 0.388 0.933 2.41 0.55
CCKBR_P480_F 0.009 0.014 0.425 30.36 0.41
HOXA11_P92_R 0.009 0.044 0.330 7.50 0.29
IHH_P529_F 0.009 0.028 0.287 10.24 0.26
KIT_P405_F 0.009 0.056 0.530 9.46 0.47
PDGFRB_P343_F 0.009 0.010 0.372 37.17 0.36
KCNK4_P171_R 0.009 0.392 0.808 2.06 0.42
MFAP4_P10_R 0.010 0.364 0.780 2.14 0.42
MMP2_P303_R 0.010 0.266 0.870 3.27 0.60
SEZ6L_P299_F 0.010 0.048 0.613 12.78 0.57
STAT5A_E42_F 0.010 0.348 0.922 2.65 0.57
ARHGDIB_P148_R 0.010 0.266 0.700 2.63 0.43
GRB10_P496_R 0.010 0.294 0.767 2.61 0.47
HCK_P46_R 0.010 0.020 0.525 26.25 0.51
IGFBP2_P306_F 0.010 0.074 0.653 8.83 0.58
IGFBP2_P353_R 0.010 0.200 0.633 3.17 0.43
KIT_P367_R 0.010 0.050 0.685 13.70 0.64
PDGFRB_P273_F 0.010 0.238 0.743 3.12 0.51
SPARC_E50_R 0.010 0.146 0.485 3.32 0.34
WNT1_P79_R 0.010 0.136 0.537 3.95 0.40
GSTP1_P74_F 0.012 0.038 0.312 8.20 0.27
NTSR1_E109_F 0.012 0.014 0.408 29.17 0.39
     
26



Table 3.2, Continued


 
 
FGF8_E183_F 0.013 0.018 0.267 14.81 0.25
MT1A_E13_R 0.013 0.082 0.548 6.69 0.47
PROK2_E0_F 0.015 0.008 0.223 27.92 0.22
HCK_P858_F 0.017 0.308 0.815 2.65 0.51
PAX6_P50_R 0.017 0.082 0.678 8.27 0.60
WNT10B_P993_F 0.017 0.046 0.413 8.99 0.37
FGF8_P473_F 0.018 0.044 0.585 13.30 0.54
HOXA11_E35_F 0.018 0.240 0.848 3.53 0.61
NGFR_P355_F 0.018 0.332 0.762 2.29 0.43
MYCN_E77_R 0.022 0.024 0.315 13.13 0.29
RBP1_E158_F 0.022 0.008 0.310 38.75 0.30
FRZB_P406_F 0.022 0.018 0.405 22.50 0.39
ESR2_E66_F 0.022 0.010 0.313 31.33 0.30
EPHA8_P256_F 0.023 0.072 0.448 6.23 0.38
IRAK3_E130_F 0.023 0.022 0.510 23.18 0.49
KCNK4_E3_F 0.023 0.404 0.848 2.10 0.44
GDF10_P95_R 0.023 0.130 0.590 4.54 0.46
MME_P388_F 0.026 0.086 0.555 6.45 0.47
PURA_P928_R 0.026 0.036 0.332 9.21 0.30
CFTR_P115_F 0.028 0.018 0.253 14.07 0.24
PTHLH_E251_F 0.029 0.408 0.828 2.03 0.42
MCAM_P169_R 0.030 0.022 0.228 10.38 0.21
FGF3_E198_R 0.031 0.032 0.515 16.09 0.48
GDF10_E39_F 0.031 0.234 0.533 2.28 0.30
JAK3_E64_F 0.031 0.068 0.627 9.22 0.56
PYCARD_P393_F 0.031 0.266 0.568 2.14 0.30
HOXA9_P303_F 0.032 0.354 0.767 2.17 0.41
HTR1B_P107_F 0.032 0.208 0.625 3.00 0.42
HTR1B_P222_F 0.032 0.198 0.798 4.03 0.60
IGF2_P1036_R 0.032 0.050 0.228 4.57 0.18
MMP2_P197_F 0.032 0.128 0.675 5.27 0.55
PTCH2_P568_R 0.032 0.328 0.690 2.10 0.36
RBP1_P426_R 0.032 0.132 0.583 4.42 0.45
     
27

Table 3.2, Continued  
 
 
 
 
 
SMO_P455_R 0.032 0.140 0.740 5.29 0.60
TNF_P158_F 0.032 0.134 0.473 3.53 0.34
ADAMTS12_P250_R 0.040 0.020 0.323 16.17 0.30
POMC_P400_R 0.040 0.384 0.798 2.08 0.41
TAL1_E122_F 0.040 0.098 0.372 3.79 0.27
EPO_E244_R 0.041 0.040 0.607 15.17 0.57
HOXA5_P1324_F 0.041 0.256 0.715 2.79 0.46
SEZ6L_P249_F 0.041 0.038 0.542 14.25 0.50

































28

10 significantly
hypermethylated genes
Table 3.3 Significantly Hypermethylated Genes at all Studied Loci in Cell lines

Mean
non-
tumors
(β
NT
)
Mean
Adeno
tumors
(β
ADT
)
Mean
Cell
lines
(β
CL
)
Difference
CL-ADT
(β
CL
–β
ADT
)
p value
ADT vs. CL
after
correction
CCKBR_P361_R 0.01 0.01 0.24 0.23 0.008
CCKBR_P480_F 0.00 0.01 0.43 0.41 0.009
FGF8_E183_F 0.01 0.02 0.27 0.25 0.013
FGF8_P473_F 0.02 0.04 0.59 0.54 0.018
GDF10_E39_F 0.14 0.23 0.53 0.30 0.031
GDF10_P95_R 0.10 0.13 0.59 0.46 0.023
HCK_P46_R 0.06 0.02 0.53 0.51 0.010
HCK_P858_F 0.08 0.31 0.82 0.51 0.017
IGFBP2_P306_F 0.03 0.07 0.65 0.58 0.010
IGFBP2_P353_R 0.11 0.20 0.63 0.43 0.010
KCNK4_E3_F 0.32 0.40 0.85 0.44 0.023
KCNK4_P171_R 0.41 0.39 0.81 0.42 0.009
KIT_P367_R 0.08 0.05 0.69 0.64 0.010
KIT_P405_F 0.04 0.06 0.53 0.47 0.009
MYCN_E77_R 0.04 0.02 0.32 0.29 0.022
MYCN_P464_R 0.01 0.01 0.18 0.17 0.008
SEZ6L_P249_F 0.01 0.04 0.54 0.50 0.041
SEZ6L_P299_F 0.02 0.05 0.61 0.57 0.010
WNT10B_P823_R 0.43 0.39 0.93 0.55 0.009
WNT10B_P993_F 0.06 0.05 0.41 0.37 0.017













29

CCKBR_P361_R                                                        CCKBR_P480_F  
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
CCKBR_P361_R
ADT CL NT
Type
                                 
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
CCKBR_P480_F
ADT CL NT
Type






FGF8_E183_F                                                             FGF8_P473_F  
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
FGF8_E183_F
ADT CL NT
Type
                                 
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
FGF8_P473_F
ADT CL NT
Type





Fig. 3.4 Loci Significantly Hypermethylated in Cell lines.The standard deviation and
error bars are represented by the red box.


30

Fig. 3.4, Continued

GDF10_E39_F                                                             GDF10_P95_R  
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
GDF10_E39_F
ADT CL NT
Type
                                 
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
GDF10_P95_R
ADT CL NT
Type






HCK_P46_R                                                                HCK_P858_F  
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
HCK_P46_R
ADT CL NT
Type
                                 
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
HCK_P858_F
ADT CL NT
Type







31

Fig. 3.4, Continued

IGFBP2_P306_F                                                          IGFBP2_P353_R  
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
IGFBP2_P306_F
ADT CL NT
Type
                                 
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
IGFBP2_P353_R
ADT CL NT
Type






KCNK4_E3_F                                                             KCNK4_P171_R  
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
KCNK4_E3_F
ADT CL NT
Type
                                 
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
KCNK4_P171_R
ADT CL NT
Type







32

Fig. 3.4, Continued

KIT_P367_R                                                                KIT_P405_F  
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
KIT_P367_R
ADT CL NT
Type
                                 
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
KIT_P405_F
ADT CL NT
Type






MYCN_E77_R                                                            MYCN_P464_R  
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
MYCN_E77_R
ADT CL NT
Type
                                 
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
MYCN_P464_R
ADT CL NT
Type







33

Fig. 3.4, Continued

SEZ6L_P249_F                                                           SEZ6L_P299_F  
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
SEZ6L_P249_F
ADT CL NT
Type
                                 
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
SEZ6L_P299_F
ADT CL NT
Type






WNT10B_P823_R                                                      WNT10B_P993_F  
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
WNT10B_P823_R
ADT CL NT
Type
                                 
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
WNT10B_P993_F
ADT CL NT
Type






34

81 loci hyper. in
ADT over NT
highlighted
a
Table 3.4 Loci Hypermethylated in Cell lines with Other Locus Hypermethylated in
Tumors
a: Loci highlighted in green are significantly hypermethylated in adeno tumors over non-
tumors;  b: Loci highlighted in orange are significantly hypermethylated in cell lines as
compared to adeno tumors.

71 loci hyper.
CL over ADT
highlighted
b
Mean
NT
(β
NT
)
Mean
Adeno
tumors
(β
ADT
)
Mean
Cell
lines
(β
CL
)
p-value
ADT vs.
NT after
correction
p value
ADT vs
CL after
correction
FRZB_E186_R

0.15 0.48 0.88 0.004 0.031

FRZB_P406_F 0.02 0.02 0.41 0.892 0.022

HOXA11_E35_F 0.10 0.24 0.85 0.314 0.018
HOXA11_P698_F

0.12 0.70 0.94 0.002 0.030

HOXA11_P92_R 0.04 0.04 0.33 1.623 0.009
HOXA9_P1141_R

0.17 0.62 0.90 0.015 0.041
HOXA9_P303_F HOXA9_P303_F 0.11 0.35 0.77 0.015 0.032
HTR1B_E232_R

0.11 0.43 0.68 0.020 0.157

HTR1B_P107_F 0.19 0.21 0.63 3.999 0.032

HTR1B_P222_F 0.04 0.20 0.80 0.005 0.032
IGF2_E134_R

0.06 0.28 0.32 0.036 1.761

IGF2_P1036_R 0.03 0.05 0.23 0.397 0.032
 
0.05 0.21 0.39 1.118 0.545

MFAP4_P10_R 0.19 0.36 0.78 0.020 0.010
MFAP4_P197_F

0.32 0.71 0.94 0.002 0.009
MMP2_E21_R

0.04 0.25 0.49 0.008 1.838

MMP2_P197_F 0.05 0.13 0.68 0.551 0.032
MMP2_P303_R MMP2_P303_R 0.04 0.27 0.87 0.046 0.010
MYH11_P22_F

0.03 0.26 0.67 0.014 0.071

MYH11_P236_R 0.00 0.01 0.56 2.105 0.007
NGFR_E328_F

0.09 0.26 0.47 0.020 1.043

NGFR_P355_F 0.22 0.33 0.76 1.735 0.018

SPARC_E50_R 0.10 0.15 0.49 0.156 0.010
SPARC_P195_F

0.13 0.56 0.71 0.003 0.886

TAL1_E122_F 0.02 0.10 0.37 0.009 0.040
TAL1_P594_F

0.13 0.63 0.87 0.029 0.202


35

loci were identified as result of differential hypermethylation in cell lines. Since, most of
the loci corresponding to these 32 genes belonged to the same CpG island, this significant
hypermethylation observed in cell lines is unexpected and may or may not have a
biological significance. The rest 2 genes had only one locus studied on the platform and
hence the difference observed in them may or may not be significant.
This significant hypermethylation may be due to following reasons. Cell lines are more
homogenous since they are clones derived from few cells. As a result they show a more
homogenous methylation pattern. In contrast, frozen tumors samples may be
contaminated by neighboring non-tumor cells. This might result in reduction of observed
methylation in tumors. This hypothesis is supported by the observation that adeno tumor
sample 3 and adeno tumor sample 5 were more similar to cell lines. This could be
because they were more homogenous and had a higher number of cancerous cells as
compared to other adeno frozen samples.
In addition to differential hypermethylation observed in cell lines, a cluster of genes were
observed to be hypomethylated in cell lines as compared to tumors and as marked in Fig
3.2. These loci were hypermethylated in tumors and non-tumor samples.
3.4 Hypomethylation in Cell Lines
As mentioned earlier, Wilcoxon rank sum test was performed on all the studied 1412
CpG loci between adenocarcinoma tumors and adenocarcinoma cell lines and 261
statistically significant loci were found after correcting for FDR. In order to identify loci
that were significantly hypomethylated in cell lines, I applied filters for selecting the loci
for which the mean methylation in cell lines (β
CL
) was lower than the mean methylation
36

in adeno tumors (β
ADT
) by at least a value of 0.17 and a methylation fold ratio smaller at
least by a factor of 2. As a result, 46 significant loci were identified encoding 37 genes.
All these 46 loci are tabulated in Table 3.5. Five of these 37 genes were observed to be
hypomethylated at both of their studied CpG loci in cell lines. These loci are tabulated in
Table 3.6 and the scatter plots of these loci are shown in Fig. 3.5. It is possible that these
genes may be providing an in vitro growth advantage. As expression of these genes was
not studied, we cannot conclude whether the observed hypomethylation had a biological
significance, but it raises concerns on the use of these cell lines for the studies involving
these 5 genes directly or indirectly.
Twenty three of these 37 genes showed variable hypomethylation. At least two CpG loci
were studied for each of these 23 genes, however only one of the two loci was observed
to lose significant methylation in cell lines. Also, even though the loss of methylation was
significant the mean degree of methylation value (β
CL
) was around 0.5 for many genes.  
Hence, it is unlikely that there may be change in expression of these genes leading to an
alteration in cell lines as compared to tumors. The rest 9 genes had only one studied CpG
locus on the platform. Hence, additional studies at more CpG loci of these genes needs to
be done before it can be concluded that they are indeed hypomethylated in cell lines.
In order to further study the role of these genes and determine if the hypomethylation
observed in cell lines is a random phenomenon, we classified them according to their
molecular functions. The classification is shown in Table 3.3.  




37

46 loci
hypomethylated in
cell lines  
Table 3.5 Significantly Hypomethylated Loci in Cell lines

Mean
non-
tumors
(β
NT
)
Mean
Adeno
tumors
(β
ADT
)
Mean Cell
lines (β
CL
)
Difference
ADT-CL
(β
ADT
-β
CL
)
p-value
ADT vs.
CL after
correction
FRK_P36_F 0.66 0.46 0.03 -0.43 0.007
MST1R_P87_R 0.64 0.56 0.03 -0.53 0.007
SFN_E118_F 0.99 0.79 0.04 -0.75 0.007
SPDEF_P6_R 0.47 0.39 0.04 -0.35 0.007
FRK_P258_F 0.85 0.61 0.04 -0.57 0.009
NDN_P1110_F 0.96 0.77 0.04 -0.73 0.009
S100A2_P1186_F 0.89 0.64 0.05 -0.59 0.009
EPHA2_P340_R 0.20 0.28 0.05 -0.23 0.009
IFNG_E293_F 0.85 0.60 0.09 -0.51 0.009
PRSS1_E45_R 0.82 0.65 0.05 -0.61 0.009
CASP8_E474_F 0.37 0.38 0.07 -0.31 0.009
CD1A_P6_F 0.96 0.85 0.25 -0.60 0.009
EPHA2_P203_F 0.29 0.47 0.06 -0.41 0.009
GRB7_E71_R 0.19 0.31 0.05 -0.27 0.009
GRB7_P160_R 0.49 0.37 0.04 -0.32 0.009
PRSS1_P1249_R 0.79 0.56 0.05 -0.51 0.010
SPDEF_E116_R 0.33 0.23 0.06 -0.17 0.010
TGFBI_P173_F 0.16 0.36 0.07 -0.29 0.010
WNT8B_E487_F 0.91 0.87 0.25 -0.62 0.010
ZIM3_P718_R 0.98 0.95 0.44 -0.51 0.010
ZP3_P220_F 0.87 0.78 0.10 -0.69 0.010
DLC1_P695_F 0.79 0.58 0.09 -0.48 0.010
KIAA0125_E29_F 0.92 0.87 0.23 -0.64 0.010
MC2R_P1025_F 0.52 0.52 0.10 -0.42 0.010
NOS3_P38_F 0.92 0.61 0.12 -0.49 0.010
PI3_E107_F 0.89 0.67 0.12 -0.55 0.010
PI3_P274_R 0.92 0.88 0.30 -0.59 0.010
ABCC2_P88_F 0.99 0.97 0.46 -0.51 0.017
MEST_P4_F 0.41 0.24 0.06 -0.18 0.017
MSH3_E3_F 0.89 0.81 0.35 -0.47 0.017
MST1R_E42_R 0.49 0.43 0.08 -0.36 0.017
     
38


Table 3.5, Continued
   
     
SFTPA1_E340_R 0.99 0.76 0.20 -0.56 0.017
ACVR1_P983_F 0.87 0.87 0.42 -0.45 0.018
KRT1_P798_R 0.88 0.72 0.29 -0.43 0.018
CPA4_E20_F 0.65 0.25 0.04 -0.21 0.020
PLAT_P80_F 0.08 0.21 0.03 -0.17 0.022
SGCE_P250_R 0.92 0.95 0.47 -0.49 0.022
SEPT9_P58_R 0.70 0.59 0.13 -0.46 0.028
CXCL9_E268_R 0.89 0.59 0.10 -0.48 0.031
ITK_P114_F 0.91 0.70 0.18 -0.52 0.031
USP29_E274_F 0.99 0.92 0.42 -0.51 0.031
IFNG_P188_F 0.87 0.75 0.18 -0.57 0.032
MAP3K8_P1036_F 0.55 0.53 0.16 -0.37 0.032
MBD2_P233_F 0.71 0.49 0.14 -0.34 0.032
S100A2_E36_R 0.42 0.28 0.08 -0.20 0.032
TFF1_P180_R 0.92 0.66 0.17 -0.50 0.032


5 genes with all loci
hypomethylated  
Table 3.6 Genes Significantly Hypomethylated at all Studied Loci in Cell Lines

Mean
non-
tumors
(β
NT
)
Mean
Adeno
tumors
(β
ADT
)
Mean Cell
lines (β
CL
)
Difference
ADT-NT
(β
NT
–β
AT
)
p value
ADT vs
CL after
correction
EPHA2_P203_F 0.32 0.47 0.06 0.15 0.009
EPHA2_P340_R 0.22 0.28 0.05 0.06 0.009
FRK_P258_F 0.72 0.61 0.04 -0.11 0.009
FRK_P36_F 0.56 0.46 0.03 -0.10 0.007
GRB7_E71_R 0.20 0.31 0.05 0.12 0.009
GRB7_P160_R 0.40 0.37 0.04 -0.04 0.009
S100A2_E36_R 0.46 0.28 0.08 -0.18 0.032
S100A2_P1186_F 0.86 0.64 0.05 -0.22 0.009
SPDEF_E116_R 0.38 0.23 0.06 -0.15 0.010
SPDEF_P6_R 0.47 0.39 0.04 -0.08 0.007


39

     EPHA2_P203_F                                                    EPHA2_P340_R  
   
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
FRK_P258_F
ADT CL NT
Type
                             
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
EPHA2_P203_F
ADT CL NT
Type
       

   
   
  FRK_P258_F                                                            FRK_P36_F                                                            
 
0
0.1
0.2
0.3
0.4
0.5
0.6
EPHA2_P340_R
ADT CL NT
Type
                               
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
FRK_P36_F
ADT CL NT
Type

                                                 




Fig. 3.5 Genes Hypomethylated in Cell lines at all Their Studied CpG Loci  




40


Fig. 3.5, Continued
   
  GRB7_E71_R                                                          GRB7_P160_R  
 
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
GRB7_P160_R
ADT CL NT
Type
                               
0
0.1
0.2
0.3
0.4
0.5
GRB7_E71_R
ADT CL NT
Type
       





S100A2_E36_R                                                              S100A2_P1186_F  
0
0.1
0.2
0.3
0.4
0.5
0.6
S100A2_E36_R
ADT CL NT
Type
                                     
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
S100A2_P1186_F
ADT CL NT
Type






41



Fig. 3.5, Continued
 
SPDEF_E116_R                                                        SPDEF_P6_R  
0
0.1
0.2
0.3
0.4
0.5
SPDEF_E116_R
ADT CL NT
Type
                                 
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
SPDEF_P6_R
ADT CL NT
Type
       











42

To understand the role of these differentially hypomethylated and hypermethylated genes
in cancer cell lines in comparison to tumors, they were classified based on molecular
function using the database Panther. Results are shown in Table 3.7.  
The classification of all the genes on the Illumina Goldengate platform is also provided.
Based on the classification of these genes, the expected number of genes in the group of
hypermethylated and hypomethylated genes is calculated for each molecular function.
As shown in the Table 3.7, the fraction of hypomethylated genes in most of the gene
function categories is similar the fraction of genes with those functions present on the
GoldenGate platform. This indicates that hypomethylation of genes in cell lines may be a
random phenomenon. Only the group of transcription factors, which contained 115
(10.3%) genes on the platform and were observed to be specifically hypermethylated in
tumors and cell lines, showed a wide deviation. The transcription factor group showed
only two genes undergoing hypomethylation while 11 genes were observed to be
hypermethylated in cell lines and tumors as compared to non-tumors.
3.5 Molecular Function of Genes Hypermethylated or Hypomethylated in Cell Lines
Another difference among the group of hypermethylated and hypomethylated genes is
that several classes of genes encoding for proteins like cell junction proteins, immunity
proteins, ligases, membrane proteins, phosphatases and synthases did not show any gene
in the hypermethylation group and had genes in hypomethylation group similar to the
expected number. This supports the hypothesis that hypomethylation of genes is random
though hypermethylation of genes is more specific to genes encoding proteins for
particular functions.  
43

The number of genes is almost all functional categories were similar between genes
hyper-methylated in tumors and cell lines and those significantly hypermethylated in cell
lines. One reason is that many genes were same in the two groups. A particular CpG
locus for a gene was observed to be methylated in tumors and another CpG locus of the
same gene was observed to gain hypermethylation at an additional CpG locus of the same
game in cell lines in comparison to tumors. It may also be possible that the genes
significantly hypermethylated in cell lines correspond to pathways which have been
already altered. Hence, there is no effect of this additional hypermethylation. Rather these
genes may have acquired methylation over the passage of time because the protein was
not required once the pathway was altered.  
Since the expression of these genes was not studied, it is challenging to deduce the
biological effect of the observed changes in DNA methylation. However it might be
worth considering the effect of hypomethylation assuming expression and
hypermethylation assuming complete inactivation of gene expression. Assuming this I
have tried to study the biological impact on WNT pathway and conclude if cell lines
might act in a different manner in comparison to tumors.






44


Table 3.7 Molecular Function of Genes
 Hyper-
methylated
in tumors
(60)  
Hyper-
methylated
in cell lines
(55)
Hypo-
methylated in
cell lines (37)
All Genes  
on
platform
(759)
Gene Function
Obse
rved
Expe
cted
Obse
rved
Expe
cted
Obse
rved
Expect
ed Studied
Cell adhesion
molecule  1 2.5 1 2.3 2 1.6 32
Cell junction protein   0 0.6 0 0.6 0 0.4 8
Chaperone  0 0.2 0 0.1 1 0.1 2
Cytoskeletal protein  2 2.2 1 2.0 3 1.4 28
Immunity proteins  0 2.4 0 2.2 7 1.5 30
Extracellular matrix  1 2.4 3 2.2 1 1.5 30
Hydrolase  2 2.4  0 2.2 2 1.5 30
Ion channel  1 0.9 2 0.8  0 0.5 11
Kinase  12 8.9 7 8.1 6 5.5 112
Ligase   0 0.6  0 0.6  0 0.4 8
Lyase  1 0.2  0 0.1  0 0.1 2
Membrane proteins   0 0.6  0 0.6  0 0.4 8
Misc. function  2 2.7 3 2.5 3 1.7 34
Unclassified  3 5.5 3 5.1 1 3.4 70
Nucleic acid binding  9 8.8 11 8.0 5 5.4 111
Oxidoreductase  2 1.7 2 1.5 1 1.0 21
Phosphatase   0 1.0  0 0.9  0 0.6 13
Protease  2 2.8 5 2.6 7 1.8 36
Receptor  20 11.0 13 10.1 10 6.8 139
Select calcium
binding protein  1 0.8 0  0.7 3 0.5 10
Select regulatory
molecule  2 4.6 3 4.2 3 2.8 58
Signaling molecule  17 10.4 15 9.6 5 6.4 132
Synthase   0 0.4  0 0.4 1 0.2 5
Transcription factor  11 9.1 11 8.3 2 5.6 115
Carrier protein   0 0.7 1 0.7  0 0.4 9
Transferase   0 2.4 2 2.2 1 1.5 30
Transporter  2 1.4  0 1.3 1 0.9 18

45

The WNT signaling pathway controls various biological processes including cell
differentiation, cell proliferation, cell movement, polarity and cell maintenance. In
normal cells, WNT binding to Frizzled receptor leads to receptor dimerization, activation
and subsequently triggering intracellular cascades. There are at least three intracellular
cascades known: the canonical β-catenin dependent and two non-canonical β-catenin
independent.  In canonical pathway, the receptor further leads to activation of
Dishevelled (Dvl) proteins. In normal cells GSK-3β, AXIN and APC are present in the
cytoplasm. In absence of WNT activation, these proteins form an active complex which
phosphorylates β-catenin and marks it for degradation by proteasomes. Once Dvl proteins
are activated, they phosphorylate GSK-3β, inhibiting it to form a complex. Inhibition of
GSK-3β leads to β-catenin accumulation. Subsequently, β-catenin translocates to the
nucleus and causes transcriptional activation of various genes including metalloproteases
(MMP2, MMP3, MMP7, MMP9), cyclinD1, c-myc etc.  
A wide range of WNT genes are expressed in primary lung tissue and cell lines especially
during lung development. These genes include WNT 1, 2, 3, 4, 5A, 7A, 7B, 10B and 11
(33). The WNT’s activating the canonical pathway including WNT 1, WNT 2 and WNT
7 seem to play an important role in formation of peripheral airways of the lungs (37). The
role of non-canonical pathway is not clear.  
3.6 Differential Methylation of WNT Signaling Pathway Genes in Cell Lines  
Activation of the WNT pathway occurs in different cancers by different mechanisms. In
colorectal cancer, activation of the WNT pathway is frequently observed as a result of
mutations in APC or β-catenin (26). Mutation in either of these proteins leads to a
WNT Pathway in Lung Cancer
46

reduction in the degradation of β-catenin and to its accumulation in the cytoplasm. In
contrast, lung cancers exhibit a low frequency of mutation in APC or β-catenin. However,
similar to other cancers accumulation of β-catenin and activation of canonical pathway
has been observed in lung cancer (38). Some studies have reported mutation in CBP and
AXIN1 in lung cancer (17, 23, 26). One other study has shown the higher expression of
WNT1 and WNT2 in non small cell lung cancer cell lines and primary tumors (35, 36).
Interestingly, the cell line A549 showed higher expression of WNT 2 but not WNT1. In
addition, DNA methylation has also been observed to provide a mechanism for
deregulation of this pathway. A study made on WNT antagonists like APC, WIF-1,
RUNX3, DKK1, DKK3, LKB1 and SFRP1 observed that many of these genes have
hypermethylated promoters and that correlated with their expression in lung cancer cell
lines and primary lung tumors (39). Hence clearly, there are various mechanisms by
which WNT canonical pathway has been observed to be activated in lung cancer.    
On the Goldengate platform, we studied the methylation status of WNT 1, 2, 2B 5A, 8B
and 10B. Two CpG loci were studied for each of these genes. One probe of WNT 2 failed
to work and was omitted from the study. Both probes of WNT 10B were observed to be
significantly hypermethylated only in cell lines. Only one probe of WNT 1 and the only
probe of WNT 2 were observed to be hypermethylated in cell lines. However, the
difference was statistically significant only for WNT1. The other probe of WNT 1 and
other WNT genes were observed to be unmethylated in all the samples.  
Fig 3.6 shows the WNT pathway and Table 3.8 shows the other relevant WNT pathway
genes studied on the platform that were methylated or differentially methylated in cell
47

lines. FRZB, HOXC6, HOXA5, FZD9, SFRP1 and CDH13 were observed to be
hypermethylated in adeno tumors and adeno cell lines.  
In adeno tumors, it is likely that reduced expression as a result of hypermethylation of
WNT antagonists: FRZB (Frizzled related protein) and SFRP1 (Soluble frizzled related
protein) may provide a mechanism for canonical pathway activation. Since canonical
WNT genes like WNT1, WNT2 were observed to be unmethylated in tumors, it is likely
they were expressed and would have caused activation or over-activation of the pathway.  
WNT 7A binds to FZD9 (Frizzled homolog 9). A study has shown that there is reduced
expression of WNT 7A (33) in primary lung samples and supports cell proliferation, it is
possible that FZD9 hypermethylation may provide another mechanism of inhibiting
activation by WNT 7A.
However, in cell lines SFRP1, FRZB, FZD9, WNT1, WNT 2 and WNT10B were
observed to be significantly hypermethylated. Hence, similar to tumors hypermethylation
of SFRP1, FRZB, and FZD9 would support canonical pathway activation. However,
since the WNT genes WNT1and WNT2 are hypermethylated it is likely they may not be
expressed and hence will be unavailable for canonical pathway activation. In such a case,
activation of WNT pathway may not occur or it may be caused by other WNT genes. It is
unlikely that other WNT genes would cause the activation of pathway and the
downstream genes in the same way. Hence, WNT pathway may be altered in cell lines.
Since the expression of these genes was not studied, it cannot be concluded.  
Interestingly, genes WNT1 and WNT10B are located head to head on chromosome
12q13. Both probes of WNT10B were observed to be hypermethylated in cell lines and  

48


Table 3.8 Differentially Methylated Genes in WNT Pathway

Pathway Gene Name  and Symbol Methylation Status  
Cell lines Tumors

WNT signaling

Axin 1; AXIN1

Hypo

Hyper

WNT signaling

Frizzled-related protein; FRZB

Hyper

Hyper

WNT signaling

Wingless-type MMTV integration site
family, WNT10B

Hyper

Hypo

WNT signaling

CREB binding protein ;CREBBP

Hypo

Hyper

WNT signaling

Cadherin 13, H-cadherin ; CDH13

Hyper

Hyper

WNT signaling

Wingless-type MMTV integration site  
family, WNT1

Hyper

Hypo

WNT signaling Secreted frizzled-related protein 1, SFRP1

Hyper

Hyper

WNT signaling Homeobox C6, HOXC6

Hyper

Hyper

WNT signaling Homeobox A5, HOXA5

Hyper

Hyper

WNT signaling Frizzled homolog 9,FRZ9

Hyper

Hyper






49


Fig 3.6 WNT Signaling Pathway (Adapted from KEGG)
50

only one probe of WNT1 was observed to be hypermethylated in cell lines. This probe of
WNT 1 is located adjacent to WNT 10B probes. Hence, it is possible that WNT10B was
hypermethylated and that spread of methylation led to hypermethylation of WNT1.  
In conclusion, the WNT pathway can be activated by different mechanisms. Mutation of
different genes such as APC, β-catenin and APC has been observed frequently in
colorectal cancer. Since mutations in these genes are rare in lung cancer it is possible that
alteration of different genes or by different mechanisms is responsible for WNT
canonical pathway activation in lung cancer. In our study, we observed hypermethylation
of SFRP1, FRZB and FZD9 in adeno tumors and cell lines. In addition we observed
hypermethylation of WNT1 and WNT 10B in lung cancer cell lines. The
hypermethylation of WNT1 and WNT2 was unexpected as they are pathway activators
and have been observed to be highly expressed in lung tumors. It is possible that a
difference observed in our cell lines is due to availability of variety of mechanisms
available for WNT activation. A broader study linking hypermethylation of different
WNT genes along with their expression need to be done before it is concluded that WTN
pathway shows a deviation in lung cancer cell lines.
In summary, cell lines make an extremely useful tool for a researcher. The ease of
management, availability of unlimited supply of DNA and RNA, homogeneity of tissue
and reduced chances of contamination from adjacent non-tumor tissue makes them great
material to use. Also, they provide the advantage and possibility of cross-lab and cross-
platform studies with the same material. This is of considerable importance as tumors
have been found to be heterogeneous. So, deducing various characteristics of different  
51

cell lines would provide us with better information of the different mechanisms by which
pathways are regulated in lung cancer. However, it is necessary that methylation in cell
lines is similar to one observed in tumors.  
Some previous studies had looked at very few genes and concluded that the difference in
methylation between cell lines and tumors is of statistical significance and hence cell
lines may not be representative of tumors (8, 11). However, other studies following them
made a conflicting conclusion (25, 30, 31). Hence, to answer this controversy we
compared the methylation status at 1412 CpG loci encoding 759 genes in cell lines,
tumors and non-tumors. This study supports the hypothesis that cell lines retain the
cancer-specific hypermethylation events and suggest cell lines make a great model for
studying DNA methylation. However, cell lines do exhibit differential methylation at
some studied loci. Whether there is an effect of this additional hypermethylation on
biological pathways is still debatable and might show a deviation for WNT pathway.
Studies on the difference in expression between cell lines and tumors of these
differentially methylated genes need to be done to reach to a conclusion. We also
observed hypomethylation at some loci exclusively in cell lines. Though the
hypomethylation of genes seemed to be random event, it might provide a selective
advantage for growth in vitro. These hypomethylation and excess hypermethylation may
have altered the biological pathways in cell lines and hence raises concerns for using the
cell lines for studying these pathways. A broader study of these pathways with studying
the expression of these genes would explain the reason and the effect of the differential
methylation in cell lines.

52

1. Anglim PP, Galler JS, Koss MN, Hagen JA, Turla S, Campan M, Weisenberger DJ,
Laird PW, Siegmund KD, Laird-Offringa IA Identification of a panel of sensitive and
specific DNA methylation markers for squamous cell lung cancer. Mol Cancer. 2008
Jul 10; 7:62
BIBLIOGRAPHY

2. Belinsky SA. Gene-promoter hypermethylation as a biomarker in lung cancer. Nat
Rev. Cancer. 2004 Sep;4(9):707-17.  

3. Bibikova M, Lin Z, Zhou L, Chudin E, Garcia EW, Wu B, Doucet D, Thomas NJ,
Wang Y, Vollmer E, Goldmann T, Seifart C, Jiang W, Barker DL, Chee MS, Floros
J, Fan JB High-throughput DNA methylation profiling using universal bead arrays.
Genome Res. 2006 Mar; 16(3):383-93.

4. Brambilla E, Travis WD, Colby TV, The new World Health organization
classification of lung tumors. Eur Respir J 2001; 18:1059-68

5. Carracedo A, Pandolfi PP. The PTEN-PI3K pathway: of feedbacks and cross-talks.
Oncogene. 2008 Sep 18;27(41):5527-41.PMID: 18794886

6. Chhajed P,Shibuya K, Hoshino H et al. A comparison of video and autofluoroscence
bronchoscopy in patients at high risk of lung cancer.

7. Colby TV, Wistuba II, Gazdar A. Precursors to pulmonary neoplasia. Adv Anat
Pathol 1998; 5(4): 205-15

8. Costello JF, Frühwald MC, Smiraglia DJ, et al. Aberrant CpG-island methylation has
non-random and tumour-type-specific patterns. Nat Genet 2000; 24:132-138.

9. Dubey Sarita, Powewl Charles A. Update in Lung Cancer 2007 American Journal of
Respiratory and Critical Care Medicine Vol 177. pp. 941-946, © 2008 American
Thoracic Society doi: 10.1164/rccm.200801-107 UP  

10. Esteller M., Epigenetics in Cancer : N Engl J Med. 2008 Mar 13;358(11):1148-59

11. Esteller M, Corn PG, Baylin SB, Herman JG. A gene hypermethylation profile of
human cancer. Cancer Res 2001;61:3225-3229

12. Huang CL, Liu D, Ishikawa S, Nakashima T, Nakashima N, Yokomise H, Kadota K,
Ueno M Wnt1 overexpression promotes tumour progression in non-small cell lung
cancer. Eur J Cancer. 2008 Nov; 44 (17):2680-8. Epub 2008 Sep 13. PMID:
18790633
53


13. Imada K, Leonard WJ. The Jak-STAT pathway. Mol Immunol. 2000 Jan-Feb;37(1-
2):1-11. Review PMID: 10781830
14. Jemal A, Siegel R, Ward E, Hao Y, Xu J, Murray T, Thun MJ. Cancer statistics,
2008. CA Cancer J Clin. 2008 Mar-Apr;58 (2):71-96. Epub 2008 Feb 20. PMID:
18287387 http://caonline.amcancersoc.org/cgi/content/full/58/2/71
15. Kawakami M, Kawakami K, Stepensky VA, Maki RA, Robin H, Muller W, Husain
SR, Puri RK. Interleukin 4 receptor on human lung cancer: a molecular target for
cytotoxin therapy. Clin Cancer Res. 2002 Nov;8 (11):3503-11. PMID: 12429641

16.  Kishimoto M, Kohno T, Okudela K, Otsuka A, Sasaki H, Tanabe C, Sakiyama T,
Hirama C, Kitabayashi I, Minna JD, Takenoshita S, Yokota J. Mutations and
deletions of the CBP gene in human lung cancer. Clin Cancer Res. 2005 Jan 15;11(2
Pt 1):512-9. PMID: 15701835

17. Kotsinas A, Evangelou K, Zacharatos P, Kittas C, Gorgoulis VG Proliferation, but
not apoptosis, is associated with distinct beta-catenin expression patterns in non-
small-cell lung carcinomas: relationship with adenomatous polyposis coli and G(1)-to
S-phase cell-cycle regulators. Am J Pathol. 2002 Nov;161 (5):1619-34.  

18. Laird PW, Jones PA et al. Cancer epigenetics comes of age Nat Genet, 1999.
9988266

19. Laird PW, Zijderveld A, Linders K, Rudnicki MA, Jaenisch R, Berns A. Simplified
mammalian DNA isolation procedure. Nucleic Acids Res. 1991 Aug 11;19 (15):4293.
No abstract available. PMID: 1870982

20. Lewin J, Plum A, Hildmann T, Rujan T, Eckhardt F, Liebenberg V, Lofton-Day C,
Wasserkort R Comparative DNA methylation analysis in normal and tumour tissues
and in cancer cell lines using differential methylation hybridisation. Int J Biochem
Cell Biol. 2007;39(7-8):1539-50. Epub 2007 Mar 15.

21. Lind GE, Thorstensen L, Løvig T, Meling GI, Hamelin R, Rognum TO, Esteller M,
Lothe RA A CpG island hypermethylation profile of primary colorectal carcinomas
and colon cancer cell lines. Mol Cancer. 2004 Oct 11;3:28

22. Marchevsky AM, Tsou JA, Laird-Offringa IA Classification of individual lung cancer
cell lines based on DNA methylation markers: use of linear discriminant analysis and
artificial neural networks. J Mol Diagn. 2004 Feb;6(1):28-36

23. Mazieres J, He B, You L, Xu Z, Jablons DM. Wnt signaling in lung cancer. Cancer
Lett. 2005 May 10;222 (1):1-10. PMID: 15837535
54



24. Merlo A, Herman JG, Mao L, Lee DJ, Gabrielson E, Burger PC, Baylin SB,
Sidransky D. 5' CpG island methylation is associated with transcriptional silencing of
the tumour suppressor p16/CDKN2/MTS1 in human cancers. Nat Med. 1995 Jul;1
(7):686-9.

25. Paz MF, Fraga MF, Avila S, Guo M, Pollan M, Herman JG, Esteller M A systematic
profile of DNA methylation in human cancer cell lines. Cancer Res. 2003 Mar 1;63
(5):1114-21.

26. Polakis Paul, The Many ways of Wnt in cancer. Curr. Opinion in Genetics and
Development 2007, 17:45-51

27. Pongracz JE, Stockley RA Wnt signalling in lung development and diseases. Respir
Res. 2006 Jan 26;7:15. Review. PMID: 16438732

28. Smiraglia DJ, Rush LJ, Frühwald MC, Dai Z, Held WA, Costello JF, Lang JC, Eng
C, Li B, Wright FA, Caligiuri MA, Plass C Excessive CpG island hypermethylation
in cancer cell lines versus primary human malignancies. Hum Mol Genet. 2001 Jun
15;10 (13):1413-9.

29. Tsou JA, Galler JS, Siegmund KD, Laird PW, Turla S, Cozen W, Hagen JA, Koss
MN, Laird-Offringa IA. Identification of a panel of sensitive and specific DNA
methylation markers for lung adenocarcinoma. Mol Cancer. 2007 Oct 29;6:70

30. Ueki T, Walter KM, Skinner H, Jaffee E, Hruban RH, Goggins M. Aberrant CpG
island methylation in cancer cell lines arises in the primary cancers from which they
were derived. Oncogene. 2002 Mar 27;21(13):2114-7.

31. Virtanen C, Ishikawa Y, Honjoh D, Kimura M, Shimane M, Miyoshi T, Nomura H,
Jones MH. Integrated classification of lung tumors and cell lines by expression
profiling. Proc Natl Acad Sci U S A. 2002 Sep 17;99 (19):12357-62.

32. Vosshenrich CA, Di Santo JP. Interleukin signaling. Curr Biol. 2002 Nov
19;12(22):R760-3

33. Winn RA, Marek L, Han SY, Rodriguez K, Rodriguez N, Hammond M, Van Scoyk
M, Acosta H, Mirus J, Barry N, Bren-Mattison Y, Van Raay TJ, Nemenoff RA,
Heasley LE Restoration of Wnt-7a expression reverses non-small cell lung cancer
cellular transformation through frizzled-9-mediated growth inhibition and promotion
of cell differentiation. J Biol Chem. 2005 May 20;280 (20):19625-34. Epub 2005 Feb
10. PMID: 15705594
55


34. Wood LD, Parsons DW, Jones S, Lin J, Sjöblom T, Leary RJ, Shen D, Boca SM,
Barber T, Ptak J, Silliman N, Szabo S, Dezso Z, Ustyanksky V, Nikolskaya T,
Nikolsky Y, Karchin R, Wilson PA, Kaminker JS, Zhang Z, Croshaw R, Willis J,
Dawson D, Shipitsin M, Willson JK, Sukumar S, Polyak K, Park BH, Pethiyagoda
CL, Pant PV, Ballinger DG, Sparks AB, Hartigan J, Smith DR, Suh E, Papadopoulos
N, Buckhaults P, Markowitz SD, Parmigiani G, Kinzler KW, Velculescu VE,
Vogelstein B. The genomic landscapes of human breast and colorectal cancers
Science. 2007 Nov 16; 318 (5853):1108-13. Epub 2007 Oct 11

35. You L, He B, Xu Z, Uematsu K, Mazieres J, Mikami I, Reguart N, Moody TW,
Kitajewski J, McCormick F, Jablons DM Inhibition of Wnt-2-mediated signaling
induces programmed cell death in non-small-cell lung cancer cells. Oncogene. 2004
Aug 12;23 (36):6170-4. PMID: 15208662 
Abstract (if available)
Abstract Cell lines are widely used for studying DNA methylation in cancer. Their main applications are in the development of biomarkers, in pharmacological studies and for studying the effect of demethylating agents and for studying the effect of methylation on biological pathways. Studies have compared the methylation status in cell lines to primary tumors in various cancers. Conflicting conclusions have been made as to whether cell lines represent the methylation observed in tumors. 
Linked assets
University of Southern California Dissertations and Theses
doctype icon
University of Southern California Dissertations and Theses 
Conceptually similar
DNA methylation markers for blood-based detection of small cell lung cancer in mouse models
PDF
DNA methylation markers for blood-based detection of small cell lung cancer in mouse models 
DNA methylation as a biomarker in human reproductive health and disease
PDF
DNA methylation as a biomarker in human reproductive health and disease 
The kinetic study of engineered MBD domain interactions with methylated DNA: insight into binding of methylated DNA by MBD2b
PDF
The kinetic study of engineered MBD domain interactions with methylated DNA: insight into binding of methylated DNA by MBD2b 
Development of DNA methylation based biomarkers for the early detection of squamous cell lung cancer
PDF
Development of DNA methylation based biomarkers for the early detection of squamous cell lung cancer 
The expression of human carboxylesterases in normal tissues and cancer cell-lines
PDF
The expression of human carboxylesterases in normal tissues and cancer cell-lines 
CpG poor promoter SULT1C2 regulated by DNA methylation and is induced by cigarette smoke condensate in lung cell lines
PDF
CpG poor promoter SULT1C2 regulated by DNA methylation and is induced by cigarette smoke condensate in lung cell lines 
DNA methylation inhibitors and epigenetic regulation of microRNA expression
PDF
DNA methylation inhibitors and epigenetic regulation of microRNA expression 
DNA methylation changes in the development of lung adenocarcinoma
PDF
DNA methylation changes in the development of lung adenocarcinoma 
Investigating the function and epigenetic regulation of ABCA3, a novel LUAD tumor suppressor gene
PDF
Investigating the function and epigenetic regulation of ABCA3, a novel LUAD tumor suppressor gene 
Development of DNA methylation biomarkers as an early detection tool for human lung adenocarcinoma
PDF
Development of DNA methylation biomarkers as an early detection tool for human lung adenocarcinoma 
Identification of DNA methylation markers in diffuse large B-cell lymphoma
PDF
Identification of DNA methylation markers in diffuse large B-cell lymphoma 
Ancestral inference and cancer stem cell dynamics in colorectal tumors
PDF
Ancestral inference and cancer stem cell dynamics in colorectal tumors 
Differential methylation analysis of colon tissues
PDF
Differential methylation analysis of colon tissues 
The role of DNA methylation in early detection and progression of pancreatic cancer
PDF
The role of DNA methylation in early detection and progression of pancreatic cancer 
Epigenetic regulation of non CPG island gene promoters
PDF
Epigenetic regulation of non CPG island gene promoters 
RNA methylation in cancer plasticity and drug resistance
PDF
RNA methylation in cancer plasticity and drug resistance 
Investigating the role of SASH1 gene located on chromosome 6 in ovarian cancer
PDF
Investigating the role of SASH1 gene located on chromosome 6 in ovarian cancer 
Personalized breast cancer management using 5-fluorouracil and related biomarkers: a molecular biology and business perspective
PDF
Personalized breast cancer management using 5-fluorouracil and related biomarkers: a molecular biology and business perspective 
Epigenetic mechanisms driving bladder cancer
PDF
Epigenetic mechanisms driving bladder cancer 
Immune signature of murine solid tumor models
PDF
Immune signature of murine solid tumor models 
Asset Metadata
Creator Chopra, Nikhil (author) 
Core Title CpG methylation profiling in lung cancer cell lines, tumors and non-tumors 
School Keck School of Medicine 
Degree Master of Science 
Degree Program Biochemistry and Molecular Biology 
Degree Conferral Date 2009-05 
Publication Date 04/28/2009 
Defense Date 03/26/2009 
Publisher University of Southern California (original), University of Southern California. Libraries (digital) 
Tag adenocarcinomas,cancer,cancer biomarkers,cell lines,CpG,DNA methylation,epigenetics,lung cancer,lungs,methylation,OAI-PMH Harvest 
Language English
Contributor Electronically uploaded by the author (provenance) 
Advisor Offringa, Ite Laird (committee chair), Tokes, Zoltan A. (committee member), Yang, Allen S. R. (committee member) 
Creator Email chopranik@gmail.com,nikhilch@usc.edu 
Permanent Link (DOI) https://doi.org/10.25549/usctheses-m2136
Unique identifier UC1166227 
Identifier etd-Chopra-2858 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-221753 (legacy record id),usctheses-m2136 (legacy record id) 
Legacy Identifier etd-Chopra-2858.pdf 
Dmrecord 221753 
Document Type Thesis 
Rights Chopra, Nikhil 
Type texts
Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection) 
Repository Name Libraries, University of Southern California
Repository Location Los Angeles, California
Repository Email cisadmin@lib.usc.edu
Tags
adenocarcinomas
cancer
cancer biomarkers
cell lines
CpG
DNA methylation
epigenetics
lung cancer
lungs
methylation