Transcriptome Profiling of a Multiple Recurrent Muscle-Invasive Urothelial Carcinoma of the Bladder by Deep Sequencing
et al. (2014) Transcriptome Profiling of a Multiple Recurrent Muscle-Invasive Urothelial Carcinoma of the Bladder
by Deep Sequencing. PLoS ONE 9(3): e91466. doi:10.1371/journal.pone.0091466
Transcriptome Profiling of a Multiple Recurrent Muscle- Invasive Urothelial Carcinoma of the Bladder by Deep Sequencing
Shufang Zhang 0
Yanxuan Liu 0
Zhenxiang Liu 0
Chong Zhang 0
Hui Cao 0
Yongqing Ye 0
Shunlan Wang 0
Ying'ai Zhang 0
Sifang Xiao 0
Peng Yang 0
Jindong Li 0
Zhiming Bai 0
Georgios Gakis, Eberhard-Karls University, Germany
0 1 Affiliated Haikou Hospital, Xiangya School of Medicine Central South University, Haikou Municipal People's Hospital , Haikou , China , 2 Department of Genetic Disease, the First Affiliated Hospital of Xinxiang Medical University , Xinxiang , China , 3 Department of Shanghai Claison Bio-Technology , Shanghai , China
Urothelial carcinoma of the bladder (UCB) is one of the commonly diagnosed cancers in the world. The UCB has the highest rate of recurrence of any malignancy. A genome-wide screening of transcriptome dysregulation between cancer and normal tissue would provide insight into the molecular basis of UCB recurrence and is a key step to discovering biomarkers for diagnosis and therapeutic targets. Compared with microarray technology, which is commonly used to identify expression level changes, the recently developed RNA-seq technique has the ability to detect other abnormal regulations in the cancer transcriptome, such as alternative splicing. In this study, we performed high-throughput transcriptome sequencing at ,506 coverage on a recurrent muscle-invasive cisplatin-resistance UCB tissue and the adjacent non-tumor tissue. The results revealed cancer-specific differentially expressed genes between the tumor and non-tumor tissue enriched in the cell adhesion molecules, focal adhesion and ECM-receptor interaction pathway. Five dysregulated genes, including CDH1, VEGFA, PTPRF, CLDN7, and MMP2 were confirmed by Real time qPCR in the sequencing samples and the additional eleven samples. Our data revealed that more than three hundred genes showed differential splicing patterns between tumor tissue and non-tumor tissue. Among these genes, we filtered 24 cancer-associated alternative splicing genes with differential exon usage. The findings from RNA-Seq were validated by Real time qPCR for CD44, PDGFA, NUMB, and LPHN2. This study provides a comprehensive survey of the UCB transcriptome, which provides better insight into the complexity of regulatory changes during recurrence and metastasis.
Competing Interests: The authors have declared that no competing interests exist. Yongqing Ye is employed by the Department of Shanghai Claison
BioTechnology, which was involved in helping the authors performthe sample collection. The authors have no other relationship with the Department of Shanghai
Claison Bio-Technology relating to employment, consultancy, patents, products in development or marketed products. This does not alter the authors adherence
to all the PLOS ONE policies on sharing data and materials.
The bladder cancer is the seventh most prevalent type of cancer
worldwide. Global estimates suggest that in 2008, approximately
386,300 new bladder cancer cases were diagnosed and that
150,200 patients succumbed to the disease . As the major
subtype of bladder cancer, urothelial carcinoma of the bladder
(UCB) is the fifth most expensive cancer to treat, accounting for
$3.7 billion in direct costs in 2001 . The costs are high because
most patients survive long term, recurrence is frequent and lifelong
surveillance is required. This disease occurs predominantly in
men, yet it is increasing in incidence among women in a manner
that cannot be entirely explained by increased tobacco use .
Approximately 80% of bladder cancers present as non-muscle
invasive urothelial carcinoma, 70% of them will recur, and 10
20% of them will progress and invade the bladder muscle . Of
the patients initially presenting with muscle-invasive UCB, 50%
will relapse with metastatic disease [5,6]. High-grade
muscleinvasive disease represents a life-threatening condition and
requires timely treatment [7,8].
Prior studies of genomic alterations have revealed that somatic
changes, including point mutations [9,10], DNA rearrangements
(reviewed in ) and copy number variations  , can result
in mutations that drive the development of UCB. As a
consequence of changes in the cancer genome, the reprogramming
of the transcriptome leads to abnormal cellular behavior and thus
directly contributes to cancer progression [13,14]. Studying the
cancer transcriptome not only enables us to fill in the gap between
driver mutations and cancer cell behavior, but also allows us to
identify additional candidate cancer-related mutations and the
molecular basis of gene regulation .
Alternative splicing (AS), the process by which splice sites are
differentially utilized to produce different mRNA isoforms, is a key
component in expanding a relatively limited number of genes into
very complex proteomes in metazoans. Several evidences
suggested that AS changes were associated with cancers [15,16,17]. The
cancer-specific splice variants may potentially be used as
diagnostic, prognostic, and predictive biomarkers as well as
therapeutic targets .
The recent development of massively parallel sequencing
(RNAseq) provides a powerful approach to profile the transcriptome
with greater efficiency and higher resolution . The advantage
of RNA-seq is that this technique makes feasible the study of the
cancer transcriptome complexity, including alternative splicing,
isoform usage, gene fusions and novel transcripts (reviewed in
[20,21]. Despite the prevalence of using RNA-seq to study various
cancer transcriptomes , the deep annotation of UCB gene
expression profiling has not been performed.
In this study, we aimed to thoroughly annotate the
transcriptomes of UCB tissue and adjacent non-tumor tissue from a single
recurrent and cisplatin-resistance patient by RNA-seq. First, we
found several dysregulated genes. Second, we performed the
enrich analysis of Gene Ontology (GO) and pathway analysis of
the dysregulated genes. Third, we investigated the differential
splicing pattern between tumor and non-tumor tissue, and found
out the cancer-associated genes with different exon skipping
events. Finally, to validate our sequencing results, quantitative
real-time PCR (qPCR) was used to confirm the difference of gene
expression and the differential usage of splice variants in the
sequencing patient and eleven additional patients.
Analysis of RNA-Seq data
Two samples UCB tissue (stage II, multiple recurrent and
cisplatin-resistance) and distant non-tumor tissue were collected
from a Chinese male patient. Fig. S1 showed the pathological
diagnostic images of the UCB tissue. All samples were subjected to
massively parallel paired-end cDNA sequencing. In total, we
obtained 32.0 million and 31.4 million read pairs from the UCB
and non-tumor tissue, respectively. We used TopHat to align the
reads to the UCSC (the University of California Santa) reference
human genome Hg19. The uniquely aligned reads for the two
samples ranged from 26.4 million to 28 million pairs. The
proportion of reads that mapped to the Ensembl reference genes
was ,78% for the both samples. The average coverage of our
sequencing depth was approximately 50 times of human
transcriptome (approximately 113 millon bp, based on the total
length of the uniquely annotated exon region in the Ensembl
database). In addition, only ,1% reads were mapped to rRNA,
indicating that our libraries are properly constructed and faithfully
represent the expression of genes with ploy (A). The details of the
mapping results are listed in Table 1.
Analysis of differentially expressed genes
After mapping the RNA-Seq reads to the reference genome
with TopHat, transcripts were assembled and their relative
abundances were calculated using Cufflinks . The Cufflinks
use Cuffdiff algorithm to measure the gene expression and to
identify the differentially expressed genes (DEGs). The normalized
expression level of each gene was measured by Fragments Per
Kilobase of exon per Million fragments mapped (FPKM). By
requiring that the FPKM was greater than one, we detected
14,520 and 14,199 expressed genes in the tumor and non-tumor
samples respectively, which included the majority of the annotated
human reference genes (See Table S1 for details). The global gene
expression profiles of two samples was correlated (Pearson
correlation coefficient R = 0.77) (Fig. S2A). We totally detected
1879 significant DEGs (FDR,0.01, FDR: False Discovery Rate)
between the two samples (Table S1). The volcano plot (Fig. S2B)
and MA-plot (Fig. S2C) of the gene expression profiles show that
the number of up- and down-regulated genes was nearly equal
relative to the q-value and expression level, suggesting that the
significance of the statistics test was not bias toward up- or
downregulated genes and the dysregulated genes is not biased toward
highly or lowly expressed genes.
Function enrichment analysis of differentially expressed
To better understand the function of DEGs, we conducted an
enrichment analysis of Gene Ontology (GO) for the dysregulated
genes. We performed enrichment tests for significantly
dysregulated genes that were detected in the UCB and non-tumor tissue
using online tools from DAVID . In total, the dysregulated
genes in UCB were categorized into 22 GO terms of Biological
Process (Table 2, p,0.05, corrected by Bonferroni correction).
Most of terms were related to immune response, cell adhesion,
response to wounding, extracellular structure organization,
locomotion (chemotaxis and taxis), leukocyte activation, and so on.
A more informative analysis of functional annotation can be
achieved by studying the enrichment of differentially expressed
genes in a particular pathway. We used DAVID  to analyze
which KEGG pathway was enriched with dysregulated genes in
UCB. The pathways enriched with DEGs are listed in Table 3
(FDR,0.05). The cell adhesion molecules (CAMs) pathway was
the most significant pathway (FDR = 2.67E-08). In addition, the
focal adhesion, ECM (extracellular matrix)-receptor interaction
pathway, and some disease pathway were also enriched.
To experimentally confirm the differentially expressed genes
identified by RNA-seq, we performed the validation by
quantitative real-time PCR (qRT-PCR). We chose five candidate genes
Uniquely Mapped Single Reads
Uniquely Mapped Paired Reads
Total Uniquely Mapped Reads
Uniquely Splice Junction Reads
Total Uniquely Mapped length (bp)
#: Sequencing coverage on human transcriptome (approximately 113 million bps which was estimated as the total length of all unique exons according to Ensembl
GO Term in Biological Process
Corrected p value*
#: Fold Enrichment = (number of differentially expressed genes with the GO term/number of differentially expressed genes)/(number of expressed genes with the GO
term/number of expressed genes)
*: p value corrected by method of Bonferroni, and only GO terms of the corrected p value less than 0.05 were shown.
(PTPRF, MMP2, VEGFA, CDH1 and CLDN7) that were
detected differential expression by Cuffdiff (Table S2) and involved
in Bladder cancer pathway, cell adhesion molecules (CAMs)
pathway and focal adhesion pathway. We used GAPDH as an
endogenous control in these reactions. The qRT-PCR results
confirmed that all of these candidate genes expressed differently
between UCB and non-tumor tissue, as shown in Fig. 1.
#: Fold Enrichment = (number of differentially expressed genes in the pathway/number of differentially expressed genes)/(number of expressed genes in the pathway/
number of expressed genes)
*: False Discovery Rate provided by DAVID, only pathways of the FDR less than 0.05 were shown.
Figure 1. The differentially expressed genes detected by RNA-seq are confirmed by qRT-PCR. qRT-PCR was performed for five genes that
are identified as differential expressed genes between UCB and non-tumor tissues. The expression level of each gene was normalized to the level in
non-tumor tissue. A-E: PTPRF, MMP2, VEGFA, CDH1 and CLDN7.
To examine whether these genes were always dysregulated in
the bladder cancer, we performed the qRT-PCR to test the
expression changes for the five genes between the paired cancer
and none-cancer tissue in eleven additional patients (which
including 6 recurrent UCB patients and 5 newly diagnosed).
The result showed that, CDH1, VEGFA, PTPRF and CLDN7
were up-regulated in six cancer samples, and MMP2 was
downregulated in ten cancer samples, suggested that these genes,
especially MMP2, were dysregualted in most UCB samples (Table
S3). And we also found that CDH1, VEGFA, PTPRF were
upregulated in 66.7% (4/6) recurrent patients but only 40% (2/5)
newly diagnosed patients (Fig. 2), suggesting the three genes might
associated with the recurrence of UCB.
Alternative splicing events in bladder cancer
One gene locus can express multiple isoforms by alternative
splicing (AS). The transcript diversity leads to plastic
transcriptional networks in cancer, which are important to generate the
unusual properties of cancer cells [17,24]. We thus perform
genome-wide screening to identify the cancer-restricted alternative
splicing events using software MISO (the Mixture of Isoforms)
. In total, we detected 25,695 and 23,769 alternative splicing
events in the UCB and non-tumor tissue, respectively (Table 4).
These events included seven different patterns: alternative 39splice
sites (A3SS), alternative 59splice sites (A5SS), alternative first exons
(AFE), mutually exclusive exons (MXE), retained introns (RI),
skipped exons (SE) and tandem 39UTRs (TUTR). Half of these
events were exon skipping (Table 4).
We next detected the differential splicing events (DSEs) between
UCB and non-tumor samples using MISO (Table S4 raw). We
found 462 DSEs from 390 unique genes, and more than half of
DSEs belong to skipped exon (Table 5). We defined the genes with
DSEs as differential splicing genes (DSGs). To identify reliable
DSGs associated with cancer, we further filtered the DSEs by a
series steps (Materials and Methods) and obtained 43 reliable
DSEs from 38 unique cancer-associated DSGs (Table S4
cancerassociated). Of these DSEs, 25 events from 24 DSGs belong to
splicing pattern skipped exon (Table 6). As an example, Fig. 3
showed the coverage of reads of PDGFA in the differential exon
usage. The ratio of junction-reads number for the exon inclusion
versus the exon exclusion was obviously higher in the cancer tissue
than that in the non-tumor tissue for both of the two genes. Since
Tumor Non-tumor Tumor Non-tumor
the skipped exon is the most common way to generate protein
products with alternative functions by truncating the functional
domain in mammals , we focus on the analysis of differential
splicing events of skipped exon in the future steps.
To experimentally confirm the skipped-exon DSGs identified by
RNA-seq, the relative expression levels between skipped exons and
their neighboring exon of selected genes were measured in the
UCB and non-tumor sample by quantitative real-time PCR
(qRTPCR). We chose four candidate genes involved in KEGG
pathways, including the CD44, GSK3B, PDGFA and NUMB,
from above 24 differential splicing genes from MISO (the primers
shown in Table S2). We used GAPDH as an endogenous control
in these reactions. The result showed that except for GSK3B,
another three genes, including CD44, PDGFA and NUMB, were
validated (Fig. 4).
We next chose six differential splicing genes (CD44, PDGFA,
NUMB, LPHN2, NIN, FAT1) to perform qRT-PCR validation in
the eleven additional patients used in differentially expressed gene
validation (Table S5). The result showed that CD44 (36%, 4/11),
PDGFA (64%, 7/11), NUMB (64%, 7/11) and LPHN2 (73%, 8/
11) showed exon increased exon inclusion in considerable number
of UCB patients, but few patients showed the increased exon
exclusion in gene NIN (18%, 2/11) and 9% (1/11). We also found
that PDFGA showed increased exon inclusion in 83% (5/6)
recurrent UCB samples, but only 40% (2/5) newly diagnosed
samples (Fig. 5). And CD44 also showed higher proportion of exon
inclusion in the recurrent samples (50%, 3/6) than that newly
diagnosed (20%, 1/5). It suggested that the increased exon
inclusion PDGFA (chr7:540068-540136) and CD44
(chr11:35231512-35231601) might associated with the recurrence
Bioinformatics prediction of gene fusion events
We used two algorithms, deFuse  and TopHat-Fusion ,
to detect gene fusion based on the pair-ends reads in the two
samples. Although various results were generated by deFuse and
TopHat-Fusion (Table S6), however, none reliable fusion
transcript was found by manually checking the reads mapping to the
fusion sequence (Methods).
Our study provides the first comprehensive insight into the
transcriptome of a recurrent, drug-resistant and muscle-invasive
urothelial carcinoma of the bladder with RNA-Seq. In total,
approximately 60 million reads were generated per sample, which
enabled us to quantify the gene expression abundance at a wide
range . The percentage of uniquely reads mapping,
approximated uniform coverage in each gene (Fig. S3) and the number of
expressed genes (FPKM.0) revealed that the data satisfied the
quality standards of the RNA-seq and represented the majority of
the transcriptome. We identified the levels of differentially
expressed genes and alternative splicing patterns associated with
Differentially expressed genes in UCB
In this study, we sampled cancer and distant non-tumor tissue
from a single individual to conduct transcriptome comparisons. To
determine whether our findings were in agreement with previously
reported results, we systematically compared the changes in the
expression of specific UCB-related genes.
We found that the vascular endothelial growth factor A
(VEGFA), a member of the PDGF/VEGF growth factor family
that promotes angiogenesis through nitric oxide synthase, was
A3SS: Alternative 39splice sites 3293
A5SS: Alternative 59splice sites
AFE: Alternative first exons
MXE: Mutually exclusive exons 681
TUTR: Tandem 39UTRs
Pattern of alternative splicing
# of differential splicing events (percentage)
# of unique differential splicing genes (percentage)
significantly up-regulated in UCB in cancer tissue compared to
non-tumor tissue. Our result is coincident with the recent two
studies using microarrays and digital gene expression profile,
which both found the up-regulation of VEGFA in a large number
of UCB patients [29,30], suggesting that VEGFA might be a
commonly over-expressed gene in UCB.
We found that most of matrix metalloproteinases (MMPs),
especially MMP2 and MMP9, is down-regulated in cancer tissues
compared to non-tumor tissue. The MMPs activate basic and
acidic fibroblast growth factors (bFGF and aFGF, respectively),
which in turn restimulate the MMPs to promote endothelial cell
migration . MMPs also stimulate scatter factor (SF), which
stimulates angiogenesis. High levels of MMP-2 and MMP-9 have
location of skipped exon
&: Y, percentage spliced in, denotes the fraction of mRNAs that represent the inclusion isoform; Y1: Y in cancer sample, Y2: Y in non-tumor sample.
*: The diff is provided by the MISO, and indicated the degree of splicing difference between samples. It was in [21, 1]. The positive diff value means that the exon
was skipped more in the non-tumor tissue than that in the cancer tissue, and the negative values means the exon skipped more in the cancer tissue.
$: The bayes factor provided by MISO indicate the significance of the splicing difference. It was in [0, +), and it was greater, then the difference was more significant.
Figure 3. RNA-Seq read mapping to the reference gene PDGFA. A: RNA-Seq read mapping to the UCSC reference genome (hg19) of the gene
PDGFA for UCB and non-tumor tissues in this study. The UCB tracks are shown in red and non-tumor tissue in green. The pink band indicated the
location of skipped exon. B: The detail of junction reads mapping to the skipped exon and its neighboring exons. The Y (percentage spliced in)
indicates the ratio of reads supporting inclusion exon vs. total reads supporting both inclusion and exclusion exon. The Y posterior distributions 
were shown in the right side.
been associated with increasing stage and grade of UCB [32,33],
and MMP2 overexpression can predict poor relapse-free and
disease-specific survival . However, in the recent two studies,
MMP2 was reported under-expression in UCB [29,30].
We also detected some biological markers in the diagnosis of
recurrent bladder cancer was dysregulated in UCB, including
KRT20 (Cytokeratin 20) , BIRC5 (Survivin) [36,37], CDH1
(E-cadherin)  and PSCA (Prostate Stem Cell Antigen-14)
[39,40]. The investigation in DEGs showed that our findings from
RNA-Seq agreed with previous reports.
In addition, several known driver factors that are frequently
mutated in UCB, including ARPC5 (p16)  and FGF2 ,
showed no change in expression in this study, suggesting that the
genetic heterogeneity of UCB or the mutated products might be
deleterious even if the expression level is unaffected.
The bladder cancer is characterized by chemoresistance
although the mechanism is still not entirely known . The
UCB sample used in this study was diagnosed as the cisplatin
resistance. We investigated the expression of the drug-resistant
genes mentioned by Ko berle et al., which listed the genes with
cisplatin-based resistance in bladder cancer . We found that
genes associated with DNA repair and apoptosis pathway were
dysregulated in UCB samples (Table S7), suggesting that the
chemoresistance of this cancer sample might be associated with the
increased DNA repair and suppression of apoptosis.
In this study, the CAMs pathway is the most significant pathway
enriched with DEGs, this result confirmed with the previous report
that the CAMs is common pathway enriched with DEGs in
carcinomas of the bladder, kidney and testis . Aberration of
the CAMs pathway and ECM receptors enables cancer cells to
escape their primary tumor masses, invade adjacent tissues and
colonize elsewhere [45,46]. Additionally, as demonstrated in our
study, frequent deregulation of the cytokine-related pathways as
well as the immune and inflammatory response processes is
another common hallmark of human cancer . For many solid
tumors, cytokines, together with CAMs, play important roles in
the induction of antitumor immune responses and tumor rejection
in the tumor microenvironment where immune and malignant
cells interact . Moreover, recent emerging data suggested that
cancer-related inflammation contributes to the proliferation and
survival of tumor cells and linked this inflammation to the
therapeutic response and prognosis of cancer patients .
Cancer-associated differential exon skipping events
Alternative regulation of gene expression can be achieved by
transcriptional and post-transcriptional regulation. The first class
of dysregulation of UCB at the transcriptional level has been well
studied using microarray technology [50,51,52]. Quantifying the
second class of regulatory change remains challenging despite the
invention of the exon array . RNA-seq technology enables the
simultaneous study of these two different mechanisms
[19,22,54,55]. In this study, we also investigated the second class
of transcriptional dysregulation by analyzing the alternative
splicing in UCB.
We performed the analysis by MISO, a probabilistic framework
to quantitate the expression level of alternatively spliced genes
from RNA-Seq data, and identifies differentially regulated
isoforms or exons across samples . By adopting the more
stringent cut-off and crossing with known cancer-associated genes,
we find 24 highly reliable cancer-associated differential splicing
Some splicing events have been reported to be related to
bladder cancer using exon arrays, including CD44, CLSTN1 and
CTNND1 . CD44 is a transmembrane glycoprotein that
participates in many cellular processes including regulation of cell
division, survival, migration, and adhesion . Its splice variant
CD44E (exon v8-10 expressed) can serve as a prognostic predictor
and indicator of disease extent in patients with urothelial cancer
[57,58,59]. In our study, the variant exon v8, v9 and v10
expressed in the UCB tissue, but not in the non-tumor tissue (Fig.
S4), suggesting that CD44E was cancer-specific. Our result
supported the CD44E as a marker in the bladder cancer diagnose.
Some cancer-associated DSGs in our result were reported in the
other cancers but not reported in the bladder cancer yet, such as
PDGFA, MACF1, ADD3 and NUMB. PDGFA, a member of
platelet-derived growth factor family, have two isoforms
corresponding to a long (PDGFAA) and a short (PDGFAB) form due to
alternative splicing of exon 6 . In this study, the exon 6 was
skipped in non-tumor tissue but not in the cancer tissue (Fig. 3),
which means that the long isoform PDGFAA was mainly
expressed in cancer tissues and the short isoform PDGFAB
expressed in non-tumor tissue. PDGFAA has a basic
carboxyterminal tail encoded by exon 6, attaching it to the extracellular
matrix whereas PDGFAB is freely diffusible in the extracellular
fluid since it lacks this retention motif [61,62]. The role of the basic
extension in PDGFAA and how it makes the long form
functionally different from the short remain unknown. Expression
of the long form of PDGFA was originally identified in tumor cells
[60,63,64], and PDGFAA was cloned from a human glioma cell
line . The different expression of long isoforms of PDGFA was
also reported in gliosarcomas of mouse  and liver cancer of rat
. A recent study showed that the long isoform of PDGFA
overexpressed in the brain abnormalities and glioma-Like lesions
in astrocytic cells in mice, and induced accumulation of immature
cells in the mouse brain . Further investigations are needed to
understand the particular mechanism of the long isoform of
PDGFA in UCB.
The AS events in MAFC1, ADD3 and NUMB were reported in
the two recent researches in non-small cell lung cancer [67,68].
MAFC1 belongs to the plakin family of cytoskeletal linker proteins
which form bridges between different cytoskeletal elements by
specialized modular domains. Using exon array and qPCR
validation, Misquitta-Ali et al. found that the exon number 8
(Ensembl exon id: ENSE00001770152) of MACF1 was expressed
in non-small lung cancer and breast cancer but not in the
pairmatched normal tissues . Our study found this AS change
between the UCB and the paired non-tumor tissue (Fig. S5),
suggesting that the increased exon inclusion of exon 8 of MAFC1
might be common in these cancers. MACF1 has no direct relation
with cancer, it has been reported to function in the Wnt signaling
pathway and to be associated with a complex containing Axin,
beta-catenin, glycogen synthase kinase 3 (GSK3B), and
adenomatous polyposis coli (APC) , which have been linked to
Figure 4. The qRT-PCR validation of differential splicing events detected by RNA-seq. qRT-PCR was performed for four genes that are
identified as differential splicing genes between UCB and non-tumor tissues. The result of qRT-PCR is the relative expression level of the skipped exon
and the neighboring constitutive exon. The expression level of each exon was normalized to the level in non-tumor tissue. The YMISO was the result
of MISO, indicates the ratio of reads supporting inclusion exon vs. total reads supporting both inclusion and exclusion exon. A,D: CD44, PDGFA,
NUMB and GSK3B.
tumorigenesis [70,71]. Based on this, the increased inclusion of the
alternative exon in MACF1 transcripts was proposed to contribute
to altered Wnt signaling in the lung and colon cancers , and
our result expands this supposition to the bladder cancer.
ADD3 (Gamma-adducin) is s a structural constituent of the
spectrin-actin cytoskeleton that contains at least 16 exons, of which
exon number 15 (ENSE00000986819) is a known cassette exon of
96 bp. Langer et al. reported that the long isoform of ADD3 with
inclusion of exon number 15 was specifically expressed in the
nonsmall cell lung cancer, but the cancer-special function of the
isoform is unclear . Our result showed the same AS event
difference between in the UCB and non-tumor tissue (Fig. S6),
suggesting that the long isoform of ADD3 might be also a
cancerspecific transcript in bladder cancer.
NUMB plays a role in the determination of cell fates during
development. The degradation of NUMB is induced in a
proteasome-dependent manner by MDM2, is a membrane-bound
protein that has been shown to associate with EPS15, LNX1, and
NOTCH1. The increased inclusion exon 9 (ENSE00001689532)
of NUMB transcripts is a highly widespread tumor-associated AS
event, which was detected by exon arrays and validated by PCR in
both of two independent laboratories [67,68]. The event was
detected in 37 additional patients with lung, breast and colon
cancer by Misquitta-Ali et al., and in 5 of 6 patients in the
study of Langer et al. . In our study using RNA-Seq, the
inclusion exon 9 of NUMB was also significantly increased in the
UCB tissue (Fig. S7). Functional analysis in lung cancer showed
that tumor-associated increases in NUMB exon 9 inclusion
correlated with reduced levels of NUMB protein expression and
activation of the Notch signaling pathway, an event that has been
linked to tumorigenesis . These findings suggested that the
increased inclusion exon 9 of NUMB is supposed to be a candidate
of marker in the diagnosis of multiple cancers including lung,
breast, colon and bladder cancer.
There are also some differential splicing events which have not
been reported to be associated with tumors, such as UCB
increased exon inclusion of LPHN2 (chr1:82452585-82452713),
EIF4A2 (chr3:186505197-186505373), FAT1
(chr4:187511522187511557), exon exclusion of CD151 (chr11:834458-834591),
and so on. The increased exon inclusion of LPHN2 was validated
in 73% (8/11) UCB patients by qRT-PCR validation. The splicing
events might be a novel alternative splicing changes associated
with bladder cancer. Investigation of these events will help to
understand the mechanism of tumor formation and progress. In
addition, we tried to ask whether the drug-resistance was related to
the differential alternative splicing due to the drug-resistance of
this UCB tissue. We compared the drug-resistant genes listed in
Table S7 with all DSEs we detected (listed in Table S4 raw).
However, none of the drug-resistant genes showed the differential
splicing events, suggesting that the drug-resistance might be
majorly associated with the dysregulation in expression level but
not in alternative splicing.
Materials and Methods
Written informed consent from the patients were obtained, and
this series of studies was reviewed and approved by Institutional
Ethics Committees of Haikou Municipal Peoples Hospital
(Haikou, China). Distant normaltissue of the urinary bladder
and urothelial carcinoma of the bladder (UCB), were obtained
from one 69-year-old Chinese male patient who initially suffered
UCB in May 2010, and recurrently in October, November 2010
and June 2011, respectively. Partial cystectomies were performed
immediately following each detection. Samples used in this study
were collected in the last surgery. H&E (hematoxylin and eosin
stained) slides of frozen UBC tissue with patient-matched frozen
normal tissue were examined by the pathologists of this study to
ensure that the tumor tissues selected had high-density cancer foci
and that the normal tissues were without tumor contamination.
The tumor was consisted of pure transitional epithelium
carcinoma, without any atypical glandular epithelial cells or squamous
epithelial cells. Besides the histology, it was observed that the
tumor invaded muscle (T2), no regional lymph nodes could not be
assessed (N0) and no distant metastasis (M0). According TNM
classification of carcinomas of the urinary bladder, the case should
defined as Stage II by the International agency of research on
Cancer. Tumor chemosensitivity assay reported that the tumor
was resistant to common used cisplatin-based chemotherapy
drugs. Fig. S1 provides the histological image of the cancerous
tissue. The percentage of tumor cells in the UCB tissue was 78%
by counting the relative at a 4006 magnification. The additional
twenty-two paired cancer and non-cancer samples using in
validation were collected from six recurrent and
cisplatinresistance UCB patients (stage II) and five newly diagnosed
UCB patients (stage II). All samples were independently reviewed
by an additional gynecologic pathologist. The treatment histories,
including chemotherapy, of cases that represent recurrence were
shown in Table S8.
Total RNA was extracted from non-tumor and cancerous
bladder tissues with TRIzol according to the manufacturers
protocol (Invitrogen). For mRNA-seq sample preparation, the
Illumina standard kit was used according to the TruSeq RNA
SamplePrep Guide (Illumina). Briefly, 10 mg of total RNA from
each sample was used for polyA mRNA selection using poly T
oligo-conjugated magnetic beads by two rounds of purification,
followed by thermal mRNA fragmentation. The fragmented
mRNA was subjected to cDNA synthesis using reverse
transcriptase (SuperScript II) and random primers. The cDNA was further
converted into double-stranded cDNA, and after end repair
(Klenow fragment, T4 polynucleotide kinase, T4 polymerase and
3-A add process [Klenow exo-fragment]), the product was ligated
to Illumina Truseq adaptors. Size selection was performed using a
2% agarose gel, generating 380-bp cDNA libraries. Finally, the
libraries were enriched using 15 cycles of PCR and purified with
the QIAquick PCR purification kit (Qiagen). The enriched
libraries were diluted with elution buffer to a final concentration
of 10 nM.
Sequencing and quality filtering
Libraries from non-tumor tissue and cancerous bladder tissue
were analyzed at a concentration of 11 pM on a single Genome
Analyzer IIx (GAIIx) lane using 115-bp sequencing. Raw RNA-seq
data were filtered by Fastx-tools (http://hannonlab.cshl.edu/
fastx_toolkit/) according to the following criteria: 1) reads
containing sequencing adaptors were removed; 2) nucleotides
with a quality score lower than 20 were trimmed from the end of
the sequence; 3) reads shorter than 50 were discarded; and 4)
artificial reads were removed. After the filtering pipeline, a total of
21.5G bp of cleaned, paired-end reads were produced.
RNA-seq reads mapping
The clean reads were then aligned with the UCSC H. sapiens
reference genome (build hg19) using TopHat v1.3.1, which
initially removes a portion of the reads based on quality
information accompanying each read and then maps the reads
to the reference genome. The pre-built H. sapiens UCSC hg19
index was downloaded from the TopHat homepage and used as
the reference genome. TopHat allows multiple alignments per
read (up to 20 by default) and a maximum of two mismatches
when mapping the reads to the reference. TopHat builds a
database of potential splice junctions and confirms these by
comparing the previously unmapped reads against the database of
putative junctions. The default parameters for the TopHat method
Transcript abundance estimation
The aligned read files were processed by Cufflinks v1.0.3 ,
which uses the normalized RNA-seq fragment counts to measure
the relative abundances of the transcripts. The unit of
measurement is Fragments Per Kilobase of exon per Million fragments
mapped (FPKM). Confidence intervals for FPKM estimates were
calculated using a Bayesian inference method . The reference
GTF annotation file used in Cufflinks was downloaded from the
Ensembl database (Homo_sapiens.GRCh37.63.gtf ). The
transcript abundance data has been submitted to the GEO
database with accession ID GSE33782.
Detection of differentially expressed gene
The downloaded Ensembl GTF file was passed to Cuffdiff along
with the original alignment (.SAM) files produced by TopHat.
Cuffdiff re-estimates the abundance of the transcripts listed in the
GTF file using alignments from the.SAM file and concurrently
tests for differential expression. Only the comparisons with
q_value less than 0.01 and test status marked as OK in the
Cuffidff output were regarded as showing differential expression.
Functional enrichment analysis of differentially expressed
The Database for Annotation, Visualization and Integrated
Discovery (DAVID) v6.7 is a set of web-based functional
annotation tools . The unique lists of differentially expressed
genes and all the expressed genes (FPKM.0 in any sample) were
submitted to the web interface as the gene list and background,
respectively. The cut-off of the False Discovery Rate (FDR) was set
at 5%, and only the results from the GO FAT and KEGG
pathways were selected as functional annotation categories for this
Detection of differential splicing events
The Mixture of Isoforms (MISO) analysis  was used to
detect differentially regulated exons across samples. The MISO
analysis was performed according to the tools given workflow
using paired-end reads (http://genes.mit.edu/burgelab/miso/
docs/). The reads alignment files (.SAM) produced by TopHat
and the pre-build human genome (Hg19) alternative events
downloaded from the MISO reference manual page (http://
used as the input. To identify highly reliable cancer-associated
DES events, we filtered the DES events by the flowing steps: 1) use
the stringent cuff-offs to filter the result of MISO (the absolute
value of diff .0.3 and bayes factor .1000, the default cut-off of
MISO were 0.2 and 10); 2) keep the genes that are overlapped
with the cancer-associated gene set, which were collected from the
NCBI gene database (searched by oncogene and tumor
suppressor gene) and the Bushman Lab web (http://
Visualization of mapped reads
The mapping results were visualized using the Integrative
Genomics Viewer (IGV) available at http://www.broadinstitute.
org/igv/. Views of other individual genes were generated by
uploading coverage.wig files to the UCSC Genome browser.
Identifying and checking the gene fusions
All the filtered RNA-seq reads were mapped to the reference
transcript sequences that were downloaded from the Ensembl
database (Homo_sapiens.GRCh37.63.cdna.all.fa) using TopHat.
The read pairs mapping to the same transcripts were removed,
and the ends of remaining reads were truncated to maintain the
75-bp length using in-house Perl scripts. These fixed-length reads
were passed to two software packages, deFuse (deFuse-0.4.2) 
and TopHat-Fusion (TopHatFusion-0.1.0) , to find the
candidate gene fusions. The bowtie-index used in the
TopHatFusion was downloaded from the TopHat homepage (H. sapiens
UCSC hg19). The parameters of the TopHat-Fusion used were
obtained from the Getting Started (http://tophat-fusion.
sourceforge.net/tutorial.html) tutorial. The deFuse parameters
were the default settings, as described in the deFuse manual. The
check of fusion transcripts was performed by mapping the reads to
the identified fusion sequences. The count of unique reads
spanned the fusion sites of the sequence should be greater than
5 and the reads was expected to be relatively uniformly distributed
in the fusion sequences.
Differentially expressed gene validation
The differentially expressed genes were validated by Real-Time
Quantitative Polymerase Chain Reaction (RT-qPCR) using a
LightCycler 480 Instrument II (Roche). The PCR volume
included 10 ml sample, 5 ml 26 SYBR Green Master Mix
(TOYOBO), 1 ml cDNA template and 1 pmol/ml of each
oligonucleotide. The RT-qPCR thermal profile was obtained
using the following procedure: 95uC for 1 min, 40 cycles of 95uC
for 10 sec, 60uC for 30 sec and 72uC for 10 sec, followed by 72uC
for 5 min. The program was set to reveal the melting curve of each
amplicon from 60uC to 95uC and obtain a read every 0.5uC. The
primer sequences are listed in Table S2. All the RT-qPCR
reactions were performed in triplicate to capture intra-assay
The expression levels of each target gene in the tested
experimental condition (cancerous bladder tissue) were compared
to the control condition ( non-tumor bladder tissue) according to
Cook et al. . The data were normalized using GAPDH, which
had previously been identified as the best reference gene under
different experimental conditions . In the present analysis,
GAPDH was confirmed to be stable and always showed variability
less than 61 cycle.
Differential splicing events validation
The primers (Table S2) were designed using Primer 5 software
(PREMIER Biosoft International, Palo Alto, Calif.), and The PCR
experiments were performed using a Veriti Thermal Cycler (ABI).
The PCR volume used comprised 10 ml sample, 1 ml 106PCR
buffer, 1 ml cDNA template, 0.2 ml dNTP, 0.2 ml Taq Enzyme
(Genscript), and 0.2 pmol/ml each oligonucleotide. PCR was
performed using the following procedure: 95uC for 1 min, 40
cycles of 95uC for 15 sec, 55uC for 30 sec and 72uC for 15 sec,
followed by 72uC for 5 min. We confirmed the presence of the
fusion gene in cancerous colon tissue. GAPDH was used as the
loading control. The PCR products of the fusion gene were cloned
Figure S6 RNA-Seq read mapping to the reference gene
ADD3. A: RNA-Seq read mapping to the UCSC reference
genome (hg19) of the gene ADD3 for UCB and normal tissues in
this study. The UCB tracks are shown in red and normal tissue in
green. The pink band indicated the location of skipped exon. B:
The detail of junction reads mapping to the skipped exon and its
Figure S7 RNA-Seq read mapping to the reference gene
NUMB. A: RNA-Seq read mapping to the UCSC reference
genome (hg19) of the gene NUMB for UCB and normal tissues in
this study. The UCB tracks are shown in red and normal tissue in
green. The pink band indicated the location of skipped exon. B:
The detail of junction reads mapping to the skipped exon and its
Gene expression and differentially expressed
in the pGEM-T Easy Vector (Promega) and then sequenced with
the T7 primer using a 3730 DNA Analyzer (ABI).
The raw sequencing data has been deposited to the NCBI Short
Read Archive on accession number SRP009386.
Figure S1 Histological image of a
hematoxylin/eosinstained section of the bladder cancer sample (original
magnification 6400) (A) and distant non-tumor
epithelial tissue of the urinary bladder and UCB tissues (B).
Figure S2 Differential expression analysis in the cancer
and normal tissue. A: The scatter plot for global expression
between samples; the Pearson correlation coefficient is shown; B:
Volcano plots for all the genes to reveal the relation between
expression fold-change and q value in DEG detecting. The red
and blue dots indicate that up- and down-regulated DEGs were
significant at q values less than 0.01. C: MA plots for all expressed
genes to reveal the relation between expression level and
foldchange. Each dots stands for one gene in comparison, the dotted
line in grey indicates M = 0. Differentially expressed genes were
plotted in red (up-regulated) and blue (down-regulated).
Figure S3 Homogeneity of reads coverage. The genes of
which FPKM.1 and cDNA length$300 bp were assigned as
three groups according to gene expression (high: the top 25%,
blue; middle: the middle 50%, red; and low: the bottom 25%,
green). All cDNA were divided into 100 bins, the median of reads
number in each bins was shown for each group. A: Reads
coverage in normal tissue; B: Reads coverage in cancer tissue.
Figure S4 RNA-Seq read mapping to the reference gene
CD44. A: RNA-Seq read mapping to the UCSC reference
genome (hg19) of the gene PDGFA for UCB and normal tissues in
this study. The UCB tracks are shown in red and normal tissue in
green. The pink band indicated the location of skipped exon. B:
The detail of junction reads mapping to the skipped exon and its
neighboring exons. The Y (percentage spliced in) indicates the
ratio of reads supporting inclusion exon vs. total reads supporting
both inclusion and exclusion exon. The Y posterior distributions
were shown in the right side.
Figure S5 RNA-Seq read mapping to the reference gene
MACF1. A: RNA-Seq read mapping to the UCSC reference
genome (hg19) of the gene MACF1 for UCB and normal tissues in
this study. The UCB tracks are shown in red and normal tissue in
green. The pink band indicated the location of skipped exon. B:
Conceived and designed the experiments: ZB SZ. Performed the
experiments: ZL CZ SW YZ SX. Analyzed the data: YL HC YY YZ.
Contributed reagents/materials/analysis tools: YY PY JL. Wrote the
paper: SZ ZB.
Differential splicing events.
qRT-PCR valication of six differential splicing
Gene fusions output by deFuse and
1. Jemal A , Siegel R , Ward E , Hao Y , Xu J , et al. ( 2008 ) Cancer statistics , 2008 . CA Cancer J Clin 58 : 71 - 96 .
2. Botteman MF , Pashos CL , Redaelli A , Laskin B , Hauser R ( 2003 ) The health economics of bladder cancer: a comprehensive review of the published literature . Pharmacoeconomics 21 : 1315 - 1330 .
3. Hayne D , Arya M , Quinn MJ , Babb PJ , Beacock CJ , et al. ( 2004 ) Current trends in bladder cancer in England and Wales . J Urol 172 : 1051 - 1055 .
4. Knowles MA ( 2001 ) What we could do now: molecular pathology of bladder cancer . Molecular Pathology 54 : 215 - 221 .
5. Williams SG , Stein JP ( 2004 ) Molecular pathways in bladder cancer . Urological Research 32 : 373 - 385 .
6. Wolff EM , Liang G , Jones PA ( 2005 ) Mechanisms of Disease: genetic and epigenetic alterations that drive bladder cancer . Nat Clin Pract Urol 2 : 502 - 510 .
7. Chang SS , Hassan JM , Cookson MS , Wells N , Smith JA ( 2003 ) Delaying radical cystectomy for muscle invasive bladder cancer results in worse pathological stage . J Urol 170 : 1085 - 1087 .
8. Herr HW , Dotan Z , Donat SM , Bajorin DF ( 2007 ) Defining optimal therapy for muscle invasive bladder cancer . J Urol 177 : 437 - 443 .
9. Reddy EP , Reynolds RK , Santos E , Barbacid M ( 1982 ) A point mutation is responsible for the acquisition of transforming properties by the T24 human bladder carcinoma oncogene . Nature 300 : 149 - 152 .
10. Kompier LC , Lurkin I , van der Aa MNM , van Rhijn BWG , van der Kwast TH , et al. ( 2010 ) FGFR3, HRAS, KRAS, NRAS and PIK3CA Mutations in Bladder Cancer and Their Potential as Biomarkers for Surveillance and Therapy. PLoS ONE 5: e13821 .
11. El-Rifai We , Kamel D , Larramendy ML , Shoman S , Gad Y , et al. ( 2000 ) DNA Copy Number Changes in Schistosoma-Associated and Non-SchistosomaAssociated Bladder Cancer . The American Journal of Pathology 156 : 871 - 878 .
12. Norskov MS , Frikke-Schmidt R , Bojesen SE , Nordestgaard BG , Loft S , et al. ( 2011 ) Copy number variation in glutathione-S-transferase T1 and M1 predicts incidence and 5-year survival from prostate and bladder cancer, and incidence of corpus uteri cancer in the general population . Pharmacogenomics J 11 : 292 - 299 .
13. Wong KM , Hudson TJ , McPherson JD ( 2011 ) Unraveling the genetics of cancer: genome sequencing and beyond . Annu Rev Genomics Hum Genet 12 : 407 - 430 .
14. Cancer Genome Atlas Research Network ( 2011 ) Integrated genomic analyses of ovarian carcinoma . Nature 474 : 609 - 615 .
15. Caceres JF , Kornblihtt AR ( 2002 ) Alternative splicing: multiple control mechanisms and involvement in human disease . Trends Genet 18 : 186 - 193 .
16. Cooper TA , Wan L , Dreyfuss G ( 2009 ) RNA and disease . Cell 136 : 777 - 793 .
17. Venables JP ( 2004 ) Aberrant and alternative splicing in cancer . Cancer Research 64 : 7647 - 7654 .
18. Pajares MJ , Ezponda T , Catena R , Calvo A , Pio R , et al. ( 2007 ) Alternative splicing: an emerging topic in molecular and clinical oncology . Lancet Oncol 8 : 349 - 357 .
19. Metzker ML ( 2010 ) Sequencing technologies - the next generation . Nat Rev Genet 11 : 31 - 46 .
20. Ozsolak F , Milos PM ( 2011 ) RNA sequencing: advances, challenges and opportunities . Nat Rev Genet 12 : 87 - 98 .
21. Wang Z , Gerstein M , Snyder M ( 2009 ) RNA-Seq: a revolutionary tool for transcriptomics . Nat Rev Genet 10 : 57 - 63 .
22. Trapnell C , Williams BA , Pertea G , Mortazavi A , Kwan G , et al. ( 2010 ) Transcript assembly and quantification by RNA-Seq reveals unannotated transcripts and isoform switching during cell differentiation . Nat Biotech 28 : 511 - 515 .
23. Huang da W , Sherman BT , Lempicki RA ( 2009 ) Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources . Nat Protoc 4 : 44 - 57 .
24. David CJ , Manley JL ( 2010 ) Alternative pre-mRNA splicing regulation in cancer: pathways and programs unhinged . Genes Dev 24 : 2343 - 2364 .
25. Katz Y , Wang ET , Airoldi EM , Burge CB ( 2010 ) Analysis and design of RNA sequencing experiments for identifying isoform regulation . Nat Methods 7 : 1009 - 1015 .
26. McPherson A , Hormozdiari F , Zayed A , Giuliany R , Ha G , et al. ( 2011 ) deFuse: An Algorithm for Gene Fusion Discovery in Tumor RNA-Seq Data . PLoS Comput Biol 7 : e1001138 .
27. Kim D , Salzberg SL ( 2011 ) TopHat-Fusion: An algorithm for Discovery of Novel Fusion Transcripts . CSHL Biology of Genomes conference.
28. Mortazavi A , Williams BA , McCue K , Schaeffer L , Wold B ( 2008 ) Mapping and quantifying mammalian transcriptomes by RNA-Seq . Nat Methods 5 : 621 - 628 .
29. Zaravinos A , Lambrou GI , Boulalas I , Delakas D , Spandidos DA ( 2011 ) Identification of Common Differentially Expressed Genes in Urinary Bladder Cancer . PLoS ONE 6 : e18135 .
30. Li X , Chen J , Hu X , Huang Y , Li Z , et al. ( 2011 ) Comparative mRNA and microRNA Expression Profiling of Three Genitourinary Cancers Reveals Common Hallmarks and Cancer-Specific Molecular Events . PLoS ONE 6 : e22570 .
31. Mitra AP , Lin H , Datar RH , Cote RJ ( 2006 ) Molecular biology of bladder cancer: Prognostic and clinical implications . Clinical Genitourinary Cancer 5 : 67 - 77 .
32. Davies B , Waxman J , Wasan H , Abel P , Williams G , et al. ( 1993 ) Levels of matrix metalloproteases in bladder cancer correlate with tumor grade and invasion . Cancer Res 53 : 5365 - 5369 .
33. Gerhards S , Jung K , Koenig F , Daniltchenko D , Hauptmann S , et al. ( 2001 ) Excretion of matrix metalloproteinases 2 and 9 in urine is associated with a high stage and grade of bladder carcinoma . Urology 57 : 675 - 679 .
34. Vasala K , Paakko P, Turpeenniemi-Hujanen T ( 2003 ) Matrix metalloproteinase-2 immunoreactive protein as a prognostic marker inbladder cancer . Urology 62 : 952 - 957 .
35. Marn-Aguilera M , Mengual L , Ribal MJ , Ars E , R os J , et al. ( 2012 ) Utility of Urothelial mRNA Markers in Blood for Staging and Monitoring Bladder Cancer . Urology 79 : 240 .e249- 240 .e215.
36. Smith SD , Wheeler MA , Plescia J , Colberg JW , Weiss RM , et al. ( 2001 ) Urine detection of survivin and diagnosis of bladder cancer . JAMA 285 : 324 - 328 .
37. Swana HS , Grossman D , Anthony JN , Weiss RM , Altieri DC ( 1999 ) Tumor content of the antiapoptosis molecule survivin and recurrence of bladder cancer . N Engl J Med 341 : 452 - 453 .
38. Lipponen PK , Eskelinen MJ ( 1995 ) Reduced expression of E-cadherin is related to invasive disease and frequent recurrence in bladder cancer . Journal of Cancer Research and Clinical Oncology 121 : 303 - 308 .
39. Wu X , Ye Y , Kiemeney LA , Sulem P , Rafnar T , et al. ( 2009 ) Genetic variation in the prostate stem cell antigen gene PSCA confers susceptibility to urinary bladder cancer . Nat Genet 41 : 991 - 995 .
40. Elsamman E , Fukumori T , Kasai T , Nakatsuji H , Nishitani MA , et al. ( 2006 ) Prostate stem cell antigen predicts tumour recurrence in superficial transitional cell carcinoma of the urinary bladder . BJU Int 97 : 1202 - 1207 .
41. Korkolopoulou P , Christodoulou P , Lazaris A , Thomas-Tsagli E , Kapralos P , et al. ( 2001 ) Prognostic implications of aberrations in p16/pRb pathway in urothelial bladder carcinomas: a multivariate analysis including p53 expression and proliferation markers . Eur Urol 39 : 167 - 177 .
42. Bai Y , Mao QQ , Qin J , Zheng XY , Wang YB , et al. ( 2010 ) Resveratrol induces apoptosis and cell cycle arrest of human T24 bladder cancer cells in vitro and inhibits tumor growth in vivo . Cancer Sci 101 : 488 - 493 .
43. Drayton RM , Catto JWF ( 2012 ) Molecular mechanisms of cisplatin resistance in bladder cancer . Expert Review of Anticancer Therapy 12 : 271 - 281 .
44. Koberle B , Piee-Staffa A ( 2012 ) The Molecular Basis of Cisplatin Resistance in Bladder Cancer Cells .
45. Cavallaro U , Christofori G ( 2004 ) Cell adhesion and signalling by cadherins and Ig-CAMs in cancer . Nat Rev Cancer 4 : 118 - 132 .
46. Stetler-Stevenson WG , Aznavoorian S , Liotta LA ( 1993 ) Tumor Cell Interactions with the Extracellular Matrix During Invasion and Metastasis . Annual Review of Cell Biology 9 : 541 - 573 .
47. Yeh HJ , Ruit KG , Wang YX , Parks WC , Snider WD , et al. ( 1991 ) PDGF Achain gene is expressed by mammalian neurons during development and in maturity. Cell 64 : 209 - 216 .
48. Emerich DF , Vasconcellos AV , Elliott RB , Skinner SJ , Borlongan CV ( 2004 ) The choroid plexus: function, pathology and therapeutic potential of its transplantation . Expert Opin Biol Ther 4 : 1191 - 1201 .
49. Crews L , Wyss-Coray T , Masliah E ( 2004 ) Insights into the pathogenesis of hydrocephalus from transgenic and experimental animal models . Brain Pathology 14 : 312 - 316 .
50. Zaravinos A , Lambrou GI , Volanis D , Delakas D , Spandidos DA ( 2011 ) Spotlight on Differentially Expressed Genes in Urinary Bladder Cancer . PLoS ONE 6 : e18255 .
51. Dong L , Bard AJ , Richards WG , Nitz MD , Theodorescu D , et al. ( 2009 ) A gene expression ratio-based diagnostic test for bladder cancer . Advances and Applications in Bioinformatics and Chemistry: 17.
52. Dyrskjt L , Kruhffer M , Thykjaer T , Marcussen N , Jensen JL , et al. ( 2004 ) Gene Expression in the Urinary Bladder . Cancer Research 64 : 4040 - 4048 .
53. Thorsen K , Srensen KD , Brems-Eskildsen AS , Modin C , Gaustadnes M , et al. ( 2008 ) Alternative Splicing in Colon, Bladder, and Prostate Cancer Identified by Exon Array Analysis . Molecular & Cellular Proteomics 7 : 1214 - 1224 .
54. Trapnell C , Pachter L , Salzberg SL ( 2009 ) TopHat: discovering splice junctions with RNA-Seq . Bioinformatics 25 : 1105 - 1111 .
55. Garber M , Grabherr MG , Guttman M , Trapnell C ( 2011 ) Computational methods for transcriptome annotation and quantification using RNA-seq . Nat Meth 8 : 469 - 477 .
56. Gotte M , Yip GW ( 2006 ) Heparanase, Hyaluronan, and CD44 in Cancers: A Breast Carcinoma Perspective . Cancer Research 66 : 10233 - 10237 .
57. Miyake H , Hara II , Arakawa S , Kamidono S ( 1999 ) Utility of Competitive Reverse Transcription-Polymerase Chain Reaction Analysis of Specific CD44 Variant RNA for Detecting Upper Urinary Tract Transitional-Cell Carcinoma . Mol Urol 3 : 365 - 370 .
58. Miyake H , Eto H , Arakawa S , Kamidono S , Hara I ( 2002 ) Over Expression of CD44V8-10 in Urinary Exfoliated Cells as an Independent Prognostic Predictor in Patients with Urothelial Cancer . The Journal of Urology 167 : 1282 - 1287 .
59. Miyake H , Hara I , Kamidono S , Eto H ( 2004 ) Multifocal Transitional Cell Carcinoma of the Bladder and Upper Urinary Tract: Molecular Screening of Clonal Origin by Characterizing CD44 Alternative Splicing Patterns . The Journal of Urology 172 : 1127 - 1129 .
60. Rorsman F , Bywater M , Knott TJ , Scott J , Betsholtz C ( 1988 ) Structural characterization of the human platelet-derived growth factor A-chain cDNA and gene: alternative exon usage predicts two different precursor proteins . Molecular and Cellular Biology 8 : 571 - 577 .
61. Heldin CH , Eriksson U , Ostman A ( 2002 ) New members of the platelet-derived growth factor family of mitogens . Archives of biochemistry and biophysics 398 : 284 - 290 .
62. Heldin CH , Westermark B ( 1999 ) Mechanism of Action and In Vivo Role of Platelet-Derived Growth Factor . Physiological Reviews 79 : 1283 - 1316 .
63. Betsholtz C , Johnsson A , Heldin CH , Westermark B , Lind P , et al. ( 1986 ) cDNA sequence and chromosomal localization of human platelet-derived growth factor A-chain and its expression in tumour cell lines . Nature 320 : 695 - 699 .
64. Collins T , Bonthron DT , Orkin SH ( 1987 ) Alternative RNA splicing affects function of encoded platelet-derived growth factor A chain . Nature 328 : 621 - 624 .
65. Nazarenko I , Hedren A , Sjodin H, Orrego A , Andrae J , et al. ( 2011 ) Brain Abnormalities and Glioma-Like Lesions in Mice Overexpressing the Long Isoform of PDGF-A in Astrocytic Cells . PLoS ONE 6 : e18303 .
66. Cook JL , Giardina JF , Zhang Z , Re RN ( 2002 ) Intracellular Angiotensin II Increases the Long Isoform of PDGF mRNA in Rat Hepatoma Cells . Journal of Molecular and Cellular Cardiology 34 : 1525 - 1537 .
67. Langer W , Sohler F , Leder G , Beckmann G , Seidel H , et al. ( 2010 ) Exon Array Analysis using re-defined probe sets results in reliable identification of alternatively spliced genes in non-small cell lung cancer . BMC Genomics 11 : 676 .
68. Misquitta-Ali CM , Cheng E , O'Hanlon D , Liu N , McGlade CJ , et al. ( 2011 ) Global Profiling and Molecular Characterization of Alternative Splicing Events Misregulated in Lung Cancer . Molecular and Cellular Biology 31 : 138 - 150 .
69. Chen HJ , Lin CM , Lin CS , Perez-Olle R , Leung CL , et al. ( 2006 ) The role of microtubule actin cross-linking factor 1 (MACF1) in the Wnt signaling pathway . Genes Dev 20 : 1933 - 1945 .
70. Saadeddin A , Babaei-Jadidi R , Spencer-Dene B , Nateri AS ( 2009 ) The Links between Transcription, b- catenin/JNK Signaling, and Carcinogenesis. Molecular Cancer Research 7 : 1189 - 1196 .
71. Wang Y ( 2009 ) Wnt/Planar cell polarity signaling: A new paradigm for cancer therapy . Molecular Cancer Therapeutics 8 : 2103 - 2109 .
72. Jiang H , Wong WH ( 2009 ) Statistical inferences for isoform expression in RNASeq . Bioinformatics 25 : 1026 - 1032 .
73. Hubbard TJP , Aken BL , Ayling S , Ballester B , Beal K , et al. ( 2009 ) Ensembl 2009 . Nucleic Acids Research 37 : D690 - D697 .
74. Cook NL , Vink R , Donkin JJ , van den Heuvel C ( 2009 ) Validation of reference genes for normalization of real-time quantitative RT-PCR data in traumatic brain injury . J Neurosci Res 87 : 34 - 41 .
75. Barber RD , Harmer DW , Coleman RA , Clark BJ ( 2005 ) GAPDH as a housekeeping gene: analysis of GAPDH mRNA expression in a panel of 72 human tissues . Physiol Genomics 21 : 389 - 395 .