ORIGINAL RESEARCH ARTICLE
published: 29 October 2013
A comprehensive promoter landscape identifies a novel
promoter for CD133 in restricted tissues, cancers, and
Ramakrishna Sompallae1†, Oliver Hofmann1,2†, Christopher A. Maher3, Craig Gedye4,
Andreas Behren4, Morana Vitezic5,6, Carsten O. Daub5,7, Sylvie Devalle8, Otavia L. Caballero9,
Piero Carninci5,7, Yoshihide Hayashizaki5,10, Elizabeth R. Lawlor11, Jonathan Cebon4and
1Department of Biostatistics, Harvard School of Public Health, Boston, MA, USA
2Harvard Stem Cell Institute, Faculty of Arts and Sciences, Holyoke Center, Cambridge, MA, USA
3The Genome Institute, Washington University School of Medicine, St. Louis, MO, USA
4Ludwig Institute for Cancer Research, Heidelberg, VIC, Australia
5RIKEN Omics Science Center, RIKEN Yokohama Institute, Kanagawa, Japan
6Department of Cell and Molecular Biology, Karolinska Institute, Stockholm, Sweden
7RIKEN Center for Life Science Technologies, Division of Genomic Technologies, Yokohama, Kanagawa, Japan
8Ludwig Institute for Cancer Research Ltd., New York Branch at Memorial Sloan-Kettering Cancer Center, New York, NY, USA
9Ludwig Collaborative Laboratory, Department of Neurosurgery, Johns Hopkins University, Baltimore, MD, USA
10RIKEN Preventive Medicine and Diagnosis Innovation Program, Wako, Saitama, Japan
11Departments of Pediatrics and Pathology, University of Michigan, Ann Arbor, MI, USA
David M. Thomas, Peter MacCallum
Cancer Centre, Australia
Xiu-bao Chang, Mayo Clinic Arizona,
Paola Parrella, IRCCS Casa Sollievo
Della Sofferenza, Italy
Winston Hide, Department of
Biostatistics, Harvard School of
Public Health, 655 Huntington Ave.,
Boston, MA 02115, USA
†These authors have contributed
equally to this work.
PROM1 is the gene encoding prominin-1 or CD133, an important cell surface marker for
the isolation of both normal and cancer stem cells. PROM1 transcripts initiate at a range
of transcription start sites (TSS) associated with distinct tissue and cancer expression
profiles. Using high resolution Cap Analysis of Gene Expression (CAGE) sequencing we
characterize TSS utilization across a broad range of normal and developmental tissues.
We identify a novel proximal promoter (P6) within CD133+melanoma cell lines and
stem cells. Additional exon array sampling finds P6 to be active in populations enriched
for mesenchyme, neural stem cells and within CD133+enriched Ewing sarcomas. The
P6 promoter is enriched with respect to previously characterized PROM1 promoters for
a HMGI/Y (HMGA1) family transcription factor binding site motif and exhibits different
epigenetic modifications relative to the canonical promoter region of PROM1.
Keywords: PROM1 protein, human, AC133 antigen, transcription start site, promoter regions, genetic, melanoma,
cancer stem cells
Surface markers play an important role in the purification of
stem and progenitor cells. CD133, (PROM1) is a transmem-
brane protein (Corbeil et al., 2001; Shmelkov et al., 2005) that
is widely used as a cell-surface marker for stem cell and can-
cer stem cell populations (Bussolati et al., 2005; Lee et al., 2005;
Tirino et al., 2008; Zhang et al., 2008). Originally identified in
hematopoetic progenitor cells by the AC133 antibody (Miraglia
et al., 1997; Yin et al., 1997; Fargeas et al., 2003) CD133 dis-
plays restricted expression in both adult human (Shmelkov et al.,
2004; Florek et al., 2005) and adult mouse tissues (Mizrak et al.,
2007). CD133+cells have also been broadly identified in non-
hematopoetic tissues during differentiation in vitro and in vivo
(Bussolati et al., 2005; Lee et al., 2005; Snippert et al., 2009)
and have been isolated from brain and other cancers that pos-
sess stem-cell properties. For some tumor types (e.g., brain, liver,
RIKEN Omics Science Center ceased to exist as of April 1st, 2013, due to
to be more tumorigenic than CD133−cell populations in
xenograft assays (Singh et al., 2004; Jiang et al., 2010; Tang et al.,
2011; von Levetzow et al., 2011). However, in colon cancers
CD133+and CD133−populations have been found to be equally
capable of tumor initiation in xenografts (Shmelkov et al., 2008),
and both cell fractions have substantial tumor initiating activ-
ity in melanoma, lung, and ovarian cancer (Meng et al., 2009;
Shackleton, 2010; Stewart et al., 2011), making CD133 a contro-
versial marker for cancer stem cells (LaBarge and Bissell, 2008).
toma differentiation has been described (Takenobu et al., 2010),
further complicating the understanding of its role and value as a
suitable surface marker.
To date, using sampling across limited types of tissues and
cancers, five TATA-less promoters (P1–P5) have been identified
in the 5?upstream region of PROM1. These promoters differen-
tially regulate expression of PROM1 in adult tissues and cancer
cell types (Shmelkov et al., 2004). The most distally located pro-
moter, P5, is present at 46kb from the start codon followed by
P1, P2, P3, and P4 promoters which are present at a distance
October 2013 | Volume 4 | Article 209 | 1
Sompallae et al. A novel promoter of CD133
of −10, −8, −7.8, and −6kb, respectively (Figure1). Promoters
P1 and P2 drive PROM1 transcription in kidney, liver, pan-
creas, placenta, lung, spleen, and colon, but can also exhibit
tissue restriction with P1 activity in small intestine and prostate
whereas P2 is active in brain and ovary (Table1); P3 is rarely
active and has only been reported in skeletal muscle, P4 and P5
activity appears to be restricted to testis (Shmelkov et al., 2004).
Further characterization has shown that P1, P2, and P3 promot-
ers contain stretches of CpG islands under epigenetic regulation
(Pleshkan et al., 2008; Tabu et al., 2008; Pellacani et al., 2011)
under transcriptional control of Sp1 and Myc (Gopisetty et al.,
2012). Collectively these findings suggest that PROM1 expression
is tightly regulated in adult tissues through the choice of specific
promoters across different cell types. However, additional rela-
tionships between choice of promoter, regulatory elements, and
expression restriction in normal and malignant tissues have yet to
As cancer cells can acquire the properties of stem cells, and
contain a stem-like population marked with CD133, a compre-
hensive understanding of the differential utilization of PROM1
promoters that regulate the expression of CD133 may illus-
trate the differences in its expression within populations of cells
with stem-like phenotypes. In order to more broadly determine
aspects of PROM1 regulation and to identify key regulatory
FIGURE 1 | Summary of PROM1 promoter architecture and
cross-platform activity. Upper panel: CD133 promoter structure with
known promoter regions P1–P5 (black rectangles), novel proximal
promoter P6 (white rectangle), known exons A–E (gray rectangles), and
CpG islands (∼). Lower panel: Promoter activity (colored rectangles,
number of studies expressing that category) in normal tissues, cancer,
and developmental systems/stem cells as captured by different platforms.
Orange: known promoter activities reported in the literature based on
single gene studies. Blue: CAGE and RNA seq studies. Green: Affymetrix
exon array exome expression. Promoters P1–P3 are widely expressed in
the literature and using CAGE and RNA-seq assay. P6 is not yet reported
in the literature.
Table 1 | PROM1 alternative promoters and regulation.
PromoterGenomic location Distance from
TissueTranscription factors References
16085637 – 16087537
−10008 Fetal liver, liver, kidney,
pancreas, placenta, lung,
spleen, colon, small intestine,
Glioblastoma, lung cancer
OCT4, SOX2 (during
Shmelkov et al., 2004; Iida et al.,
2011; Gopisetty et al., 2012
16085338 – 16085704
−8175Brain, ovary, kidney, liver,
pancreas, placenta, lung,
spleen and colon
Sp1/Myc Shmelkov et al., 2004; Tabu et al.,
2008; Gopisetty et al., 2012
16084913 – 16085337
−7808 Skeletal muscle Sp1/MycShmelkov et al., 2004; Gopisetty
et al., 2012
P4 16082163 – 16083762
−6233TestisNot studied Shmelkov et al., 2004; Tabu et al.,
P516122394 – 16123893
−46364TestisETSShmelkov et al., 2004; Tabu et al.,
P616077254 – 16077627
−98 Cancer, stem cells, and retinaHMG–
aContains CpG island.
bHypomethylated in glioblastomas.
cStart codon at location 16077529 in Exon 2 (16077741 – 16077309).
Frontiers in Genetics | Cancer Genetics
October 2013 | Volume 4 | Article 209 | 2
Sompallae et al.A novel promoter of CD133
elements associated with PROM1 expression in cancer and stem
cells, we have used Cap Analysis of Gene Expression [CAGE,
(Kodzius et al., 2006)] to perform an exhaustive assessment of
the landscape of the PROM1 upstream promoter region. CAGE
precisely defines the location of transcription start sites (TSS)
by sequencing from the 5?end of capped, full-length mRNA.
In addition to TSS identification, CAGE can measure transcript
abundance, allowing comparison of promoter activity between
To gain additional insight into PROM1 promoter activity in
the context of cancer cells with stem-like properties we have per-
formed high-coverage CAGE sequencing of five melanoma cell
lines directly derived from patient biopsies (TableS2), sorted by
CD133+into a small minority of cells from the total CD133−
population (Gedye et al., 2009). The promoter activity in these
from 72 tissues and cell types, including 13 normal tissues, 25
cancer tissues, and 34 developmental states (TableS1).
We identify a previously unknown promoter that shows differ-
ential expression and regulation of PROM1 mRNA in restricted
tissues, stem-like cells within cancer cell lines and stem cells.
MATERIALS AND METHODS
To isolate melanoma cells from fresh human melanoma explants,
freshly excised human melanoma specimens were inspected
by pathologists and fragments removed for cell line estab-
lishment without disturbing surgical margins. The melanoma
cell lines were derived from metastatic melanoma tissue and
used before passage 10. Description of the cell lines and
associated gene-expression data have been reported previously
(Behren et al., 2013). Patient consent was collected and ethical
approval for the use of the cell lines has been granted by the
Austin Health Human Research Ethics Committee (HREC). Cell
lines were cultured in our standard media (“RF10”) compris-
ing RPMI 1640 supplemented with 2mM Glutamax®, 25mM
HEPES, 50µM 2-mercaptoethanol (Hamburger and Salmon,
1977), 100U/mL penicillin, 100µg/mL streptomycin (all from
Invitrogen, Mulgrave, Australia) plus 10% fetal calf serum
(FCS; from CSL, Melbourne, Australia). Tissue fragments were
mechanically dissociated and passed through a cell strainer,
remaining fragments were subjected to enzymatic digestion
in a collagenase/DNAse/serum-free digestion media mixture
overnight at 37◦C and single cell suspension plated out the next
morning. Once established the cell lines were HLA-typed by the
Red Cross in Melbourne to ensure the match with donor tis-
sue and were tested for mycoplasma contamination. Harvested
cells were washed, counted and plated into 96 well round bottom
plates at 104–105per well. After pelleting by centrifuge the cells
were washed once with PBS cells and blocked in 50µL PBS/10%
normal human serum for 10min. The plate was gently vortexed
to resuspend cells and 1µL of AC133-PE antibody (Miltenyi
Biotec, Bergisch Gladbach, Germany) was added to each well
prior to incubation at 4◦C for 15min. Cells were washed and
resuspended and immediately analyzed on a FACSCalibur flow
cytometer (Becton Dickson, San Jose, CA). An anti-CD4-PE anti-
body was used at same concentration as Isotype control. The five
different melanoma cell cultures derived from biopsy specimens
of patients with malignant melanoma were evaluated for CD133
CELL SEPARATION AND RNA EXTRACTION
in the cell separation MACS buffer prepared according to the
manufacturer’s instructions (PBS pH 7.4 with 0.5% BSA and
0.5mM EDTA). MACS columns were refrigerated for at least 1h
prior to use. Positive selection of cells was performed using LS
columns followed by depletion with LD columns. 107cells were
resuspended in 80µL MACS buffer; 20µL FcR blocking reagent
+20µl of directly conjugated CD133 beads added, mixed well,
and incubated for 30min at 4◦C. Labeled cells were washed,
resuspended, and applied to the column. To increase purity
and were then applied to LD columns as “pre-depleted” cells. The
a second LS column to increase enrichment. After separation all
fractions were stained as described and purity of subpopulation
measured by flow cytometry.
RNA was extracted from 107purified CD133+or CD133−
(Molecular Resarch Center, Inc., Cincinnati, OH, USA). Briefly,
cells were homogenized in Trireagent, RNA collected in the
aqueous phase after addition of chloroform and precipitated
by isopropanol addition. RNA was quality checked by gel
electrophoresis and quantified using a nanodrop.
GENE EXPRESSION TRANSCRIPT ANALYSIS
CAGE was performed as described previously (Kodzius et al.,
2006; Kawaji et al., 2009). Total RNA extracted from CD133+
and CD133−melanoma cells was used to synthesize the cDNA.
RNA and cDNA pools were treated with RNAse I to cleave all
ssRNA, leaving only full length cDNA/RNA hybrids for capture
with biotin-streptavidin interactions in an cDNA/RNA hybrid
enrichment process called as cap-trapping. In this process full
length cDNAs are then ligated with specific linker oligos con-
taining MmeI restriction sites and the second strand cDNA is
synthesized. Double-stranded cDNAs are digested with MmeI
creating ∼20nt of cDNA sequence attached to a 5?linker. After
ligation of the second linker XmaJI to MmeI-cleaved 3?ends of
cDNA, fragments are subjected PCR amplification and restric-
tion site digestion to obtain CAGE sequencing tags (see Kodzius
et al., 2006, for details). The resulting CAGE tags were then
concatenated and cloned into pZEr0-2 plasmids (Invitrogen)
for sequencing. Sequence reads were extracted, filtered and
aligned to the hg18 genome build using Nexalign (Lassmann,
the methods described in (Kawaji et al., 2009). TSS in the
upstream PROM1 gene region based on CAGE were identified
from clustered sequence reads using HPeak (Qin et al., 2010) and
mapped to known PROM1 promoters extracted from GenBank
gene records (AY275524, AY438641, AY438640), resulting in the
confirmation of known promoters P1–4 and identification of a
October 2013 | Volume 4 | Article 209 | 3
Sompallae et al. A novel promoter of CD133
to in silico TSS predictions in the oPOSSUM (Ho Sui et al., 2007)
and SwitchGear [UCSC Genome Browser track, (Karolchik et al.,
RAPID AMPLIFICATION OF cDNA ENDS (5?-RACE)
To confirm the novel PROM1 TSS, 5?-RACE PCR was
performed according to the manufacturer’s protocol (invit-
rogen). RNA was prepared from a CD133+
melanoma cell line by ligating the RNA with a 5?RACE
UUG-CUG-GCU-UUG-AUG-AAA-3?); and a single-stranded
cDNA was generated. Two CD133 specific anti-sense primers
were chosen from exon2 for a nested PCR to enhance speci-
ficity and to obtain a sufficient amplification product. Primers
were designed using Primer3 software (Rozen and Skaletsky,
2000), and checked for uniqueness by querying against the
human genome using BLAST (Altschul et al., 1990) (Table2).
Amplification of 5?-RACE cDNA was performed with nested
reverse primers of PROM1 and adapter specific primers with 1µl
of the first-strand cDNA reaction. Amplified products were sep-
arated on an agarose gel and visualized by ethidium bromide
5?RACE products were cloned into the pcDNA3.1 TA
cloning vector and transformed into bacteria. The clones of
each transformation were subjected to colony PCR and the
sequencing of inserts was carried out with RACE adapter
primers and specific reverse primers. Sanger sequence products
of RACE PCR amplified fragments were separated and aligned
to the human genome using BLAT (Kent, 2002) together with
CAGE mapped TSS to confirm unique mapping to the target
Human embryonic stem cell-derived neural crest stem cells
(hESC-NCSC), human adult bone marrow-derived mes-
enchymal stem cells as well as hESC-NCSC differentiated for
6 weeks in vitro together with three independent CD133-
FACS-sorted cell populations from STA-ET-8.2 Ewing sarcoma
cells (TableS3) were profiled by Affymetrix Human Exon 1.0
(HuEx) arrays as previously described (Jiang et al., 2010).
HuEx arrays generated from four primary Ewing sarcomas
(tumor RNA graciously provided by Tissue Biorepositories at
Children’s Hospital Los Angeles and the Children’s Oncology
Group) were also included for analysis. HuEx data for addi-
tional adult tissues was obtained from an Affymetrix tissue
data/exon_array_data.affx). HuEx data was RMA normalized
using BioConductor (package affy) and probe intensities for
to identify differences in TSS utilization.
To characterize regulatory motifs of the novel PROM1 promoter
P6; an additional set of proximal promoters (−300/+100bp) was
selected from a total of 149 TSS found to be co-upregulated
with P6 in at least four out of five CD133+of the melanoma
cell lines. The 149 TSS were selected based upon a signifi-
cant difference in CAGE peaks in LM-MEL14/34/42/47/62 from
five different patients, CD133+over CD133−as determined by
HPeak,(p ≤ 0.05,TableS4).TheTSSsetwastestedfornucleotide
motif enrichment using MEME (Bailey et al., 2010) (motif width
4–21 nucleotides, both strands, any number of repetitions, p-
value ≤ 0.05) and compared to a random background distri-
bution of 10,000 CAGE-based proximal promoters taken from
the FANTOM4 collection (Kawaji et al., 2009). Significant motifs
were tested for overlap with the JASPAR 2009 Core Transcription
Factor collection (Sandelin et al., 2004) using TomTom (Bailey
et al., 2009). Additional experimentally determined transcription
factor binding sites from the ENCODE TF ChIP-seq collec-
tion (ENCODE Project Consortium, 2004) were retrieved from
the UCSC Genome Browser (hg18, update 2010-06-24, track
EPIGENETIC CHANGES AND RNA-seq
CpG island information and differential methylation in the
Encyclopedia of DNA Elements (ENCODE) project were
retrieved from the UCSC Genome Browser’s summary track
(Ernst and Kellis, 2010).
EXISTING PROMOTER LANDSCAPE OF PROM1 AND A NOVEL
PROMOTER IN MELANOMA CELL LINES
To determine PROM1 promoter utilization across diverse tissues
melanoma cell lines and an additional 72 samples grouped as
cancer, normal adult tissues and developmental stages from the
public FANTOM4 data set (TableS1). Our CAGE analysis con-
firmed known promoters and identified a novel, sixth promoter
(P6) close to the translational start codon (AUG), strongly upreg-
ulated in CD133+melanoma cells and with weak expression in
normal colon and small intestine libraries (Figure2A). We deter-
mined that TSS marked by CAGE tags from cancer, normal adult
tissues, and developmental stages were consistent with previous
reports showing P1–P3 to be widely expressed canonical pro-
moters. CAGE tags revealed utilization of all three promoters in
normal tissues, whereas promoter utilization cancer samples are
Table 2 | Primers used for validating CD133 novel TSS.
LocationForward primer (5?-3?)
(specific to RACE adapter)
Reverse primer (5?-3?)
(specific to CD133)
Frontiers in Genetics | Cancer Genetics
October 2013 | Volume 4 | Article 209 | 4
Sompallae et al.A novel promoter of CD133
FIGURE 2 | PROM1 regulatory landscape. (A) Previously described PROM1
TSS (Promoter 1–3, 4) supported by Capped Analysis of Gene Expression
(CAGE) assays from the FANTOM3 collection 57 containing 72 experimental
samples of which 13 from normal tissues, 25 from different cancers, and 34
are from developmental states (Table S3). Non-pathogenic tissue CAGE tags
are distributed evenly whereas those obtained from tumors are biased toward
promoter 1 and 2; transcription in embryonic tissues is biased toward promoter
2. Promoter 6 is supported by CAGE tags obtained from melanoma cell lines
and normal colon, small intestine, and rectum. TSS for promoters 1–4 and
promoter 6 are supported by predictive models from the SwitchGear and the
oPOSSUM database. Exon array probes used to characterize differential exon
usage (see Figure 4) are highlighted in red. (B) Epigenetic changes in PROM1
promoters. Data from ENCODE tier 1 human cell lines (H1-hESC: embryonic
stem cell line, HepG2: liver carcinoma, GM12878: lymphoblastoid, HUVEC:
umbilical vein endothelial cells, NHKE: kidney epithelium, K562: myelogenous
leukemia). Promoters 1–3 are supported by active (weak) promoters in
H1-hESC (HepG2) cells respectively, with stronger expression in H1-hESC as
indicated by matching RNA-seq. PROM1 expression from all promoters is
polycomb-repressed in GM12878, HUVEC, NHKE, and K562 resulting in no
discernible expression. Expression in H1-hESC cells is enhanced at promoter
6, indicating an independent TSS unrelated to transcription driven by canonical
promoters P1–3. The known CpG-island (REF) associated with the canonical
promoters is unmethylated in both H1-hESC and HepG2 and methylated in
GM12878 and K562, further repressing expression in those cells.
(C) Regulatory potential. A strong regulatory potential is detected around the
proximal promoter and promoter 1–3 (ESPERR, REF), with confirmed binding
sites for transcription factors POUF2, STATS1, NFKB, and others in the
ENCODE tier 1 samples (1: H1-hESC; G: GM12878; g: GM12891/GM12892/
GM15510/GM18505/GM18526/GM18951/GM19099/GM19193; H: HeLa-S3;
K:K562; L:HepG2). Binding sites for transcription factor found to be enriched
in the proximal promoters (+300/−100 nucleotides) of 149 TSS found to be
co-activated in CD133+-melanoma cells classify the PROM1 promoters into
SP1-rich (promoters 1–3, in agreement with the CpG-island) or HMG-IY-rich
(promoter 6, promoter 4).
These promoter profiles are consistent with previous reports
(Table1). The CAGE tissue panel includes one low coverage
testis library with insufficient CAGE tags to support a previously
reported testis-specific P4 TSS (Shmelkov et al., 2004).
VALIDATION OF NOVEL PROMOTER
To confirm the initiation of transcripts at P6 in CD133+sorted
cells from the melanoma cell lines we used 5?-RACE PCR. mRNA
isolated from CD133+cells derived from the melanoma cell line
October 2013 | Volume 4 | Article 209 | 5
Sompallae et al. A novel promoter of CD133
LM-MEL-34 was used to amplify the 5?end of the PROM1 tran-
script with the help of a pair of forward primers targeted to
a RACE adapter and the other pair of reverse primers specif-
ically targeted to 5?exon of PROM1 (Table2, Figure3A). Gel
electrophoresis of amplified products from these cells showed
a stronger band with the expected size of 62bp (Figure3B).
There was no sign of non-specific product in the negative con-
trol. Further, the amplified product was cloned into pcDNA3.1
plasmid for sequencing. The sequenced insert region was then
mapped to the PROM1 promoter facilitating the TSS identi-
fication and the approximate location of promoter elements
(Figure3C). 5?RACE PCR efficiently detected the initiation RNA
transcripts at P6.
CROSS PLATFORM EVIDENCE FOR A NOVEL P6 PROMOTER
The novel P6 promoter is supported by in silico TSS predic-
tions from the oPOSSUM and SwitchGear databases. In order
to further explore its utilization we studied human exon array
expression data from mesenchymal stem cells (bone marrow
MSC), neural crest stem cells (undifferentiated NCSC), Ewing
sarcoma (primary tumors and sorted cells, STA-ET-8.2), and an
Affymetrix panel of 12 adult tissues. Affymetrix Human Exon 1.0
(HuEx) arrays have an average of four probes per known exon,
and seven probes cover the 5?region of PROM1, four of which
showed significant expression, one matching the P1–P2 location
and three cover the novel P6 promoter region. Using expression
information from these probes we classified promoter utilization
across different cell types, testing for stronger expression (higher
probe intensity) at P6 than at upstream promoters P1–P2. We
found increased expression of probes at the P6 promoter com-
pared to P1–P2 in CD133+Ewing sarcoma cells, mesenchymal
stem cells, and undifferentiated NCSC, all of which express high
levels of CD133. In contrast, CD133−sorted Ewing sarcoma cells,
CD133−mesenchymal stem cells, and differentiated neural crest
stem cells show no significant difference in intensities between
probes covering P1–P2 and P6 (Figure4). We found higher
expression at P6 in four tissues (colon, pancreas, kidney, and
testis), with inconsistent replicate patterns or no discernible dif-
ference between P1–P2 and P6 in the other tissues (FigureS1).
NOVEL PROMOTER REGULATION
TRANSCRIPTION FACTOR BINDING SITE (TFBS) MOTIF ENRICHMENT
We performed an enrichment analysis of proximal promot-
ers from 149 TSS in genes in the CD133 enriched CD133+
melanoma cell lines which were found to be co-upregulated with
P6 in at least four of five CD133+cell lines (see methods). We
identified three significant motifs characterized as binding sites
for AZF/HMG-I/Y, Sp1, and Klf4, two of which (AZF/HMG-I/Y
and Sp1) are present in the P6 core promoter region (Figure2C)
and are evolutionarily conserved between human, chimp, and
EPIGENETIC MODIFICATIONS OF PROM1 PROMOTER OBSERVED IN
Epigenetic modifications are key factors for regulation of gene
transcription. Since HMG-I/Y has a role in regulating chromatin
structure we explored epigenetic modifications of the PROM1
promoter landscape in the publicly available ENCODE consor-
cells and cell lines (Gopisetty et al., 2012). Differential methyla-
tion of the same region observed in ENCODE cell lines indicate
an unmethylated CpG island in ENCODE cell lines H1-hESC
(embryonic) and HepG2 (liver) and methylation of the same
CpG island in K562 (blood, leukemic) and GM12878 (blood,
lymphoblastoid), in agreement with their RNA-seq expression
status in the same ENCODE cell lines (Figure2B). The novel
P6 promoter region does not overlap known CpG islands; in
particular, RNA-seq data indicates independent transcription
from P6 in H1-hESC. We explored histone modifications as an
FIGURE 3 | Validation of novel PROM1 TSS using 5?RACE PCR.
(A) Schema illustrating primer design for PROM1 TSS validation. PROM1
TSS is amplified using a pair of outer and inner primer sets targeted
specifically to 5?Rapid amplification of cDNA ends (RACE) adapter and
PROM1 sequences. (B) The 5?end regions of PROM1 mRNA were
amplified from CD133+and CD133−populations of LM-Mel-34 cell lines.
Gel electrophoresis of amplified products shows difference in expression
level in CD133+and CD133−populations. Amplified 5?ends of PROM1
RNA were cloned in to plasmid pcDNA3.1 and then sequenced. (C) UCSC
genome browser view of sequenced 5?end region of PROM1 mapped to
the promoter. The tracks shown here illustrates CAGE captured TSS
regions from CD133+melanoma cell lines and the RACE PCR identified 5?
end of the PROM1 transcript. The other tracks show outer and inner RACE
PCR primers specific to PROM1.
Frontiers in Genetics | Cancer Genetics
October 2013 | Volume 4 | Article 209 | 6
Sompallae et al.A novel promoter of CD133
FIGURE 4 | PROM1 promoter activity using exon arrays. Transcript wide
expression pattern of PROM1 measured by Affymetrix exon arrays with
specific probes targeting exonic regions. Y-axis: normalized signal intensity
(expression level). X-axis: genomic coordinates of the transcript. Expression
of 5?probes in CD133+neural crest stem cells (blue), Ewing sarcoma (red),
and CD133−neural crest stem cells (black), mesenchymal stem cells (gray),
and Ewing sarcoma (green). Thick black bars: Probe regions overlapping
CAGE defined TSS.
alternative regulatory mechanism using summary information
generated by ChromHMM (Figure2B, FigureS1), an algorithm
that characterizes chromatin states by integrating multiple ChIP-
seq histone modification data sets (Ernst et al., 2011). Observed
histone changes support an enhancer upstream of P6 as well
as transcriptional activity around P1–3 and P6, whereas P1–3
appears poised or repressed in all other ENCODE cell lines rep-
resented, mostly due to H3K27me3 and H3K36me3 silencing. In
summary, an ensemble of histone methylation marks, RNA pol II
binding sites and sequence conservation observed in the PROM1
promoter region support the likelihood of transcript initiation at
the P6 promoter.
Although CD133 is widely used as a stem cell marker, its signifi-
cance and relationship to cancer cells with stem-like properties is
controversial (Wu and Wu, 2009; Campos et al., 2011). Previous
studies have established five alternative promoters (P1–P5) which
drive CD133 expression in normal tissues and cancer cell lines.
This transcriptional complexity raises questions in relation to dif-
ferential regulation of the alternative promoters, a genetic feature
which has recently been widely reported (Davuluri, 2008; Pal
et al., 2011). To provide a coherent overview of PROM1 pro-
moter choice and regulation of expression of CD133 in disease
of TSS activity using genome wide assay of transcription
Using transcriptional initiation events from a representative
panel of 72 developmental, cancer, and normal CAGE (De Hoon
and Hayashizaki, 2008; Hoskins et al., 2010; Kurosawa et al.,
2010) libraries we have characterized PROM1 promoter utiliza-
tion, confirming the activity of four out of five known promoters
(P1–P4) and one novel alternate promoter (P6). P1 and P3 are
consistently utilized in normal tissues, cancer, and development
while P2 is active in developmental samples. A novel, proxi-
mal promoter P6 was identified in our high-resolution CAGE
assays of a CD133+subpopulation derived from melanoma
cell lines and independently confirmed using 5?-RACE PCR.
Initiation from P6 results in a significantly shorter 5?untranslated
EXPRESSION OF P6
The discovery of P6 by genome wide CAGE assay is supported by
existing in silico predictions and by comparison with exon array
probes overlapping regions of PROM1 TSS. PROM1 transcripts
initiate at P6 in cancer tissues, CD133+melanoma cells, adult
tissues, and stem cell enriched populations, including CD133+
stem cell behavior has been established (Jiang et al., 2010).
TRANSCRIPTIONAL REGULATION OF P6
Comparison of 149 core promoters of TSS found to be con-
sistently co-expressed with the novel PROM1 P6 promoter in
at least four out of five CD133+melanoma cell lines identified
October 2013 | Volume 4 | Article 209 | 7
Sompallae et al. A novel promoter of CD133
motif enrichment for Sp1 binding sites present in all PROM1
promoters, in agreement with their importance in the CpG
islands located around P1–3 (Gopisetty et al., 2012). A second
enriched motif, HMGI/Y, was found in the P6 promoter and
also the testis-specific promoter P4. HMG family proteins are
ubiquitously expressed nuclear proteins which regulate transcrip-
tion and chromatin structure (Reeves and Beckerbauer, 2001)
and have role in differentiation, tumor progression, and malig-
nancy (Wisniewski and Schwanbeck, 2000) by controlling genes
involved in tumor initiation, invasion, cell proliferation, and
angiogenesis (Reeves et al., 2001). HMGI/Y (HMG1A) is usu-
ally expressed at low levels in adult tissues, but found at high
expression levels in embryonic and neoplastic tissues (Chiappetta
et al., 1996), its aberrant expression has been associated with
tumorigenesis (Tkachenko et al., 1997) and high expression is
a requirement for the production of CXC ligand 1, a major
effector of tumor growth(Amiri et al., 2006). Both isoforms
(HMGI and HMGY) are expressed in neuroblastic tumors,
with higher levels in less differentiated tumor (Giannini et al.,
2000). In gliomas, HMGI/Y expression correlates with malig-
nancy, proliferation, and invasion (Pang et al., 2011). High levels
of HMGI/Y are found in more aggressive tumors and corre-
late with poor prognosis and are associated with a stem-like
state (Shah and Resar, 2012). In addition, ENCODE ChIP-seq
data indicates binding of two transcription factors (POUF2 and
NF-kb) known to interact with HMGI/Y (Reeves et al., 2001)
immediately downstream of P6 (POUF2) and in an upstream
enhancer (NF-kb). Both Sp1 and HMGI/Y are expressed in
CD133+melanoma cell lines. We did not find them to be dif-
ferentially expressed when compared to CD133-depleted cells,
although HMGI/Y undergoes extensive post-translational mod-
ifications which influence its binding properties (Bianchi and
Given the role of HMGI/Y in modifying chromatin structure we
explored epigenetic changes of the upstream PROM1 region in
ENCODE cell line data. Based on ChIP-qPCR analysis, methyla-
tion is thought to affect CD133 expression only in cell lines but
not in primary tissues (Pellacani et al., 2011), although methy-
lation of P2 is thought to be tissue specific (Pleshkan et al.,
2008). As expected, the CpG island close to P1–P3 was found
to be differentially methylated between different ENCODE cell
lines, with h-ESC and hepG2 being free of methylation and
In addition, P1–3 were found to be polycomb-repressed in all
surveyed cell line types with the exception of h1-ESC and, to a
h1-ESC showed signs of transcriptional transition between P1–3
and P6 which in combination with the active enhancer region
upstream of P6 might explain increased transcriptional activity
found in h1-ESC RNA-seq data around P6.
By combining comprehensive bioinformatics analysis of genome-
wide exon array and exhaustively and consistently sequenced
CAGE samples across a broad range of cell and tissue types and
a series of melanoma cell lines, it has been possible to reveal a
strong association between a specific new promoter and clono-
genic CD133+cells. Together, these findings provide evidence
of multiple regulatory events contributing to the diversity of
PROM1 expression and indicate a potential role for HMGI/Y
in combination with epigenetic changes to initiate transcription
from P6 in less differentiated cells or stem cells, resulting in an
upregulation of CD133. This study provides one of the few links
between expression of a stem cell marker and its likely regulation.
Ramakrishna Sompallae: Collection and/or assembly of data,
Data analysis and interpretation, Manuscript writing; Oliver
Hofmann: Collection and/or assembly of data, Data analysis and
design, Provision of study material or patients, Manuscript writ-
ing; Andreas Behren: Study design and conception, provision of
study material or patients, Manuscript writing; Morana Vitezic:
Data analysis and interpretation; Sylvie Devalle: Collection
and/or assembly of data; Otavia L. Caballero: Collection and/or
assembly of data, Elizabeth R. Lawlor: Provision of study material
or patients, Collection and/or assembly of data, Manuscript writ-
ing, Final approval of manuscript; Jonathan Cebon: Provision of
of manuscript; Winston Hide: Conception and design, Financial
support, Data analysis and interpretation, Manuscript writing,
Final approval of manuscript.
This research was supported by a Melanoma Research Alliance
Team Science Award, in part by Operational Infrastructure
Support Program Funding of the Victorian State Government
and a research grant for RIKEN OSC from MEXT to Yoshihide
Hayashizaki. MA received post graduate scholarship support
from Cancer Council Victoria, University of Melbourne, and
the Natural Sciences and Engineering Research Council of
Canada. Jonathan Cabon was supported by NHMRC Practitioner
Fellowships (487905). Andreas Behren is the recipient of
a Postdoctoral Fellowship from the Cure Cancer Australia
The Supplementary Material for this article can be found online
Figure S1 | PROM1 promoter activity using exon arrays in additional tissue
panels. Transcript wide expression pattern of PROM1 measured by
Affymetrix exon arrays with specific probes targeting exonic regions. Left
panel: colon, pancreas, testis, and kidney expressing P6. Right panel:
spleen, prostate, muscle, and thyroid expressing P1-P2.
Table S1 | Tissues surveyed for PROM1 promoter activity.
Table S2 | Melanoma cell cultures used in the study and the patient
disease stage at collection.
Table S3 | Exon arrays used in the study.
Table S4 | CD133hi_set_4_5.
Frontiers in Genetics | Cancer Genetics
October 2013 | Volume 4 | Article 209 | 8
Sompallae et al.A novel promoter of CD133
Altschul, S. F., Gish, W., Miller, W.,
Myers, E. W., and Lipman, D. J.
(1990). Basic local alignment search
tool. J. Mol. Biol. 215, 403–410. doi:
Amiri, K. I., Ha, H. C., Smulson,
of CXC ligand
tion in melanoma cell lines by
Bailey, T. L., Bodén, M., Buske, F. A.,
Frith, M., Grant, C. E., Clementi, L.,
et al. (2009). MEME SUITE: tools
for motif discovery and searching.
Nucleic Acids Res. 37, W202–W208.
Bailey, T. L., Bodén, M., Whitington,
T., and Machanick, P. (2010). The
value of position-specific priors
in motif discovery using MEME.
BMC Bioinformatics 11:179. doi:
Behren, A., Anaka, M., Lo, P.-H., Vella,
L. J., Davis, I. D., Catimel, J.,
et al. (2013). The Ludwig insti-
tute for cancer research Melbourne
melanoma cell line panel. Pigment
Cell Melanoma Res. 26, 597–600.
Bianchi, M. E., and Agresti, A. (2005).
HMG proteins: dynamic players in
gene regulation and differentiation.
C., Buttiglieri, S., Deregibus, M.
C., Cantino, D., et al. (2005).
cells from adult human kidney.
Am. J. Pathol. 166, 545–555. doi:
P. H., Eckstein, V., Unterberg,
A., Mairbaurl, H., et al. (2011).
Expression and regulation of AC133
1974–1986. doi: 10.1002/glia.21239
Visconti, R., Fedele, M., Battista, S.,
Trapasso, F., et al. (1996). High level
expression of the HMGI (Y) gene
Oncogene 13, 2439–2446.
W. B. (2001). Rat prominin, like
its mouse and
logues, is a pentaspan membrane
Res. Commun. 285, 939–944. doi:
Davuluri, R. (2008). The functional
consequences of alternative pro-
moter use in mammalian genomes.
Trends Genet. 24, 167–177. doi:
De Hoon, M., and Hayashizaki, Y.
(2008). Deep cap analysis gene
wide identification of promoters,
Biotechniques 44, 627–628, 630, 632.
ENCODE Project Consortium. (2004).
Discovery and characterization of
chromatin states for systematic
annotation of the human genome.
Nat. Biotechnol. 28, 817–825. doi:
Ernst, J., Kheradpour, P., Mikkelsen,
T. S., Shoresh, N., Ward, L. D.,
Mapping and analysis of chromatin
state dynamics in nine human
cell types. Nature 473, 43–49. doi:
Huttner, W. B. (2003). AC133
prominin-2, etc.: prominin fam-
ily gene products in need of a
rational nomenclature. Stem Cells
(Dayton, Ohio) 21, 506–508. doi:
Florek, M., Haase, M., Marzesco, A. M.,
Freund, D., Ehninger, G., Huttner,
W. B., et al. (2005). Prominin-
1/CD133, a neural and hematopoi-
etic stem cell marker, is expressed
in adult human differentiated cells
and certain types of kidney can-
cer. Cell Tissue Res. 319, 15–26. doi:
Gedye, C., Quirk, J., Browning, J.,
Svobodová, S., John, T., Sluka,
antigens can be immunological
melanoma cells. Cancer Immunol.
Immunother. 58, 1635–1646. doi:
B., Cerignoli, F., et al. (2000).
gene products in human neu-
Gopisetty, G., Xu, J., Sampath, D.,
Colman, H., and Puduvalli, V.
K. (2012). Epigenetic regulation
of CD133/PROM1 expression in
glioma stem cells by Sp1/Myc and
promoter methylation. Oncogene
32, 3119–3129. doi: 10.1038/onc.
Hamburger, A. W., and Salmon, S. E.
(1977). Primary bioassay of human
tumor stem cells. Science (New York,
NY) 197, 461–463. doi: 10.1126/sci-
Ho Sui, S. J., Fulton, D. L., Arenillas,
D. J., Kwon, A. T., and Wasserman,
W. W. (2007). oPOSSUM: inte-
grated tools for analysis of reg-
ulatory motif over-representation.
Nucleic Acids Res. 35, W245–W252.
Hoskins, R. A., Landolin, J. M.,
ture in Drosophila melanogaster.
Genome Res. 21, 182–192. doi:
Iida, H., Suzuki, M., Goitsuka, R., and
Ueno, H. (2011). Hypoxia induces
CD133 expression in human lung
cancer cells by up-regulation of
OCT3/4 and SOX2. Int. J. Oncol. 40,
71–79. doi: 10.3892/ijo.2011.1207
Jiang, X., Gwye, Y., Russell, D., Cao,
C., Douglas, D., Hung, L., et al.
cells. BMC Cancer 10:116. doi:
browser database: 2008 update.
Nucleic Acids Res. 36, D773–D779.
Kawaji, H., Severin, J., Lizio, M.,
from mammalian transcriptional
landscape to its dynamic regula-
tion. Genome Biol. 10, R40. doi:
Kent, W. J. (2002). BLAT–the BLAST-
like alignment tool. Genome Res. 12,
656–664. doi: 10.1101/gr.229202
Kodzius, R., Kojima, M., Nishiyori,
H., Nakamura, M., Fukuda, S.,
Tagami, M., et al. (2006). CAGE:
cap analysis of gene expression.
Nat. Methods 3, 211–222. doi:
(Clifton, NJ) 687, 147–163. doi:
LaBarge, M. A., and Bissell, M. J.
(2008). Is CD133 a marker of
metastatic colon cancer stem cells?
J. Clin. Invest. 118, 2021–2024. doi:
Lee, A., Kessler, J. D., Read, T. A.,
Kaiser, C., Corbeil, D., Huttner, W.
B., et al. (2005). Isolation of neural
stem cells from the postnatal cere-
bellum. Nat. Neurosci. 8, 723–729.
Meng, X., Li, M., Wang, X., Wang,
ulations of A549 and H446 cells
Cancer Sci. 100, 1040–1046. doi:
Miraglia, S., Godfrey, W., Yin, A. H.,
Atkins, K., Warnke, R., Holden,
J. T., et al. (1997). A novel five-
stem cell antigen: isolation, charac-
terization, and molecular cloning.
Blood 90, 5013–5021.
Mizrak, D., Brittan, M., and Alison,
M. (2007). CD133: molecule of the
moment. J. Pathol. 214, 3–9. doi:
V., Showe, L. C., et al. (2011).
Alternative transcription exceeds
Pang, B., Fan, H., Zhang, I. Y.,
et al. (2011). HMGA1 expres-
sion in human gliomas and its
correlation with tumor prolifer-
ation, invasion and angiogenesis.
J. Neurooncol. 106, 543–549. doi:
Pellacani, D., Packer, R. J., Frame,
F. M., Oldridge, E. E., Berry, P.
A., Labarthe, M. C., et al. (2011).
Regulation of the stem cell marker
CD133 is independent of promoter
hypermethylation in human epithe-
lial differentiation and cancer. Mol.
Cancer 10, 94. doi: 10.1186/1476-
Pleshkan, V. V., Vinogradova, T. V., and
Sverdlov, E. D. (2008). Methylation
of the prominin 1 TATA-less main
promoters and tissue specificity of
their transcript content. Biochim.
Biophys. Acta 1779, 599–605. doi:
Qin, Z. S., Yu, J., Shen, J., Maher,
C. A., Hu, M., Kalyana-Sundaram,
S., et al. (2010). HPeak: an HMM-
based algorithm for defining read-
enriched regions in ChIP-seq data.
BMC Bioinformatics 11:369. doi:
Reeves, R., and Beckerbauer, L. (2001).
HMGI/Y proteins: flexible regula-
tors of transcription and chromatin
structure. Biochim. Biophys. Acta
1519, 13–29. doi: 10.1016/S0167-
Reeves, R., Edberg, D. D., and Li, Y.
(2001). Architectural transcription
October 2013 | Volume 4 | Article 209 | 9
Sompallae et al. A novel promoter of CD133 Download full-text
factor HMGI(Y) promotes tumor
progression and mesenchymal tran-
sition of human epithelial cells.
Mol. Cell Biol. 21, 575–594. doi:
Rozen, S., and Skaletsky, H. (2000).
Primer3 on the WWW for general
users and for biologist program-
mers. Methods Mol. Biol. 132,
Sandelin, A., Alkema, W., Engström, P.,
Wasserman, W. W., and Lenhard,
B. (2004). JASPAR: an open-access
scription factor binding profiles.
Nucleic Acids Res. 32, D91–D94. doi:
cells: similar and different. Sem.
Shah, S. N., and Resar, L. M. (2012).
High mobility group A1 and can-
cer: potential biomarker and thera-
peutic target. Histol. Histopathol. 27,
Lyden, D., and Rafii, S. (2005).
Int. J. Biochem. Cell Biol. 37,
Shmelkov, S. V., Butler, J. M., Hooper,
A. T., Hormigo, A., Kushner, J.,
Milde, T., et al. (2008). CD133
expression is not restricted to
stem cells, and both CD133+ and
cer cells initiate tumors. J. Clin.
Shmelkov, S. V., Jun, L., St Clair, R.,
A., Usenko, J. K., et al. (2004).
encodes stem cell surface protein
AC133. Blood 103, 2055–2061. doi:
Singh, S. K., Hawkins, C., Clarke,
I. D., Squire, J. A., Bayani, J.,
Hide, T., et al. (2004). Identification
of human brain tumour initiat-
ing cells. Nature 432, 396–401. doi:
Snippert, H. J., van Es, J. H., van
den Born, M.,
Stange, D. E., Barker, N., et al.
(2009). Prominin-1/CD133 marks
tors in mouse small intestine.
Gastroenterology 136, 2187–2194 e1.
Stewart, J. M., Shaw, P. A., Gedye,
B. G., and Ailles, L. E. (2011).
tumor-initiating cells. Proc. Natl.
Acad. Sci. U.S.A. 108, 6468–6473.
Wang, L., Aoyanagi, E., Kohsaka,
hypomethylation regulates CD133
Tabu, K., Kimura, T., Sasai, K., Wang,
L., Bizen, N., Nishihara, H., et al.
(2010). Analysis of an alternative
human CD133 promoter reveals the
implication of Ras/ERK pathway
in tumor stem-like hallmarks. Mol.
Cancer 9, 39. doi: 10.1186/1476-
Yamaguchi, Y., Ohira, M., et al.
(2010). CD133 suppresses neu-
tion. Oncogene 30, 97–105. doi:
Tang, K. H., Ma, S., Lee, T. K., Chan,
Y. P., Kwan, P. S., Tong, C. M.,
et al. (2011).
tumor angiogenesis, growth, and
self-renewal through Neurotensin/
Tirino, V., Desiderio, V., d’Aquino,
G., Graziano, A., et al. (2008).
of CD133+ cancer stem cells in
human solid tumours. PLoS ONE
Tkachenko, A., Ashar, H. R., Meloni,
A. M., Sandberg, A. A., and Chada,
K. K. (1997). Misexpression of
disrupted HMGI architectural fac-
tors activates alternative pathways
of tumorigenesis. Cancer Res. 57,
von Levetzow, C., Jiang, X., Gwye, Y.,
von Levetzow, G., Hung, L., Cooper,
A., et al. (2011). Modeling initiation
of ewing sarcoma in human neural
crest cells. PLoS ONE 6:e19305. doi:
Wisniewski, J. R., and Schwanbeck, R.
(2000). High mobility group I/Y:
multifunctional chromosomal pro-
teins causally involved in tumor
Wu, Y., and Wu, P. Y. (2009). CD133
as a marker for cancer stem cells:
Cells Dev. 18, 1127–1134. doi:
Yin, A. H., Miraglia, S., Zanjani, E.
D., Almeida-Porada, G., Ogawa,
M., Leary, A. G., et al. (1997).
AC133, a novel marker for human
hematopoietic stem and progenitor
cells. Blood 90, 5002–5012.
Zhang, M., Song, T., Yang, L., Chen,
R., Wu, L., Yang, Z., et al. (2008).
Nestin and CD133: valuable stem
cell-specific markers for determin-
ing clinical outcome of glioma
patients. J. Exp. Clin. Cancer Res. 27,
85. doi: 10.1186/1756-9966-27-85
Conflict of Interest Statement: The
was conducted in the absence of any
commercial or financial relationships
that could be construed as a potential
conflict of interest.
Received: 08 July 2013; accepted: 30
September 2013; published online: 29
Citation: Sompallae R, Hofmann O,
Maher CA, Gedye C, Behren A, Vitezic
M, Daub CO, Devalle S, Caballero OL,
Carninci P, Hayashizaki Y, Lawlor ER,
Cebon J and Hide W (2013) A com-
prehensive promoter landscape identi-
fies a novel promoter for CD133 in
restricted tissues, cancers, and stem cells.
Front. Genet. 4:209. doi: 10.3389/fgene.
This article was submitted to Cancer
Genetics, a section of the journal
Frontiers in Genetics.
Carninci, Hayashizaki, Lawlor, Cebon
and Hide. This is an open-access article
distributed under the terms of the
Creative Commons Attribution License
(CC BY). The use, distribution or repro-
duction in other forums is permitted,
provided the original author(s) or licen-
sor are credited and that the original
publication in this journal is cited, in
accordance with accepted academic prac-
tice. No use, distribution or reproduction
is permitted which does not comply with
Frontiers in Genetics | Cancer Genetics
October 2013 | Volume 4 | Article 209 | 10