Supplementary Materials and Methods
GISTIC algorithm
In cBioPortal, the Genomic Identification of Significant Targets in Cancer (GISTIC) algorithm
was used to determine the copy number status of each gene in each sample [1,2]. The GISTIC
takes segmented copy number ratios as input, separates chromosome arm-level events from focal
events, and then performs two tests: (i) it identifies significantly amplified/deleted chromosome
arms; and (ii) it identifies regions that are significantly focally amplified or deleted. Each
aberration is assigned a G-score that considers the amplitude of the aberration as well as the
frequency of its occurrence across samples. False-discovery rate q-values are then calculated for
the aberrant regions, and regions with q-values below a user-defined threshold are considered
significant. The putative copy number level is obtained by applying both low- and high-level
thresholds to the gene copy levels of all the samples. The entries with value +/- 2 exceed the
high-level thresholds for amplifications/deletions, and those with +/- 1 exceed the low-level
thresholds but not the high-level thresholds.
Multivariate survival analysis
A Cox proportional hazard model was used for multivariate analysis, and hazard ratio (HR) was
calculated according to the cut-off value of a 95% confidence interval (CI). For 468 TCGA
breast cancer samples, factors included in the multivariate analysis model were age at diagnosis,
ER status (positive vs. negative), PR status (positive vs. negative), HER2 status (positive vs.
negative), tumor size (>20 mm vs. ≤20mm), lymph node status (positive vs. negative), metastasis
status (positive vs. negative), and PAM50 subtype (basal vs. non-basal). To analyze the
relationships between HMT mRNA expression and overall patient survival in breast cancer,
samples were divided into low (n=234) and high (n=234) groups based on mRNA expression Zscores [RNA-Seq V2 RSEM (RNA-Seq by Expectation-Maximization)] of each KDM.
Multivariate survival analysis was conducted using the Cox regression function (“coxph”) in the
R statistical programming language.
References
1.
2.
Mermel CH, Schumacher SE, Hill B, Meyerson ML, Beroukhim R, Getz G. GISTIC2.0
facilitates sensitive and confident localization of the targets of focal somatic copy-number
alteration in human cancers. Genome Biol 2011;12(4):R41.
Beroukhim R, Getz G, Nghiemphu L et al. Assessing the significance of chromosomal
aberrations in cancer: methodology and application to glioma. Proc Natl Acad Sci U S A
2007;104(50):20007-20012.
Supplementary Figure Legends
Figure S1. Expression level (Y-axis) versus copy number alteration (X-axis) of KDM2A, PHF2,
and NO66 in 976 TCGA breast cancer samples. Amp= high-level amplification; Gain=low-level
gain; Hetloss=heterozygous deletion; and Homdel= homozygous deletion.
Figure S2. Schematic structure and functional domains of KDM2A long and short isoforms.
Figure S3. Heatmap of segmented copy number data at the KDM2A region of 46 breast cancer
cell lines based on the CGH database on the UCSC Cancer Genomics Browser. The genomic
structure of both the long and short isoforms of KDM2A was based on the NCBI database
(Bottom).
Figure S4. (A) Breast cancer subtypes and mRNA expression Z-scores of 16 TCGA breast
cancer samples with KDM2A short isoform amplification based on the UCSC Cancer Browser
database. (B) The top 10 samples with the amplification of KDM2A short isoform clearly show
higher expression of exons 13-21 that code for the short isoform, compared to exons 1-12, which
code for the long isoform based on the estimated exon-level transcription (RPKM: reads per
kilobase of transcript per million reads mapped).
Figure S5. KDM2A knockdown inhibited growth of T47D breast cancer cells. (A) Top, black
and white images, and bottom, TurboGFP images of T47D cells after viral infection with control
shRNA and KDM2A shRNA #1 and #3. (B) Bar graph shows relative cell growth after knocking
down KDM2A in T47D breast cancer cells (p<0.05). Data are expressed as mean ± SD.
Figure S6. KDM2A knockdown had no impact on cell growth in SUM52 cells without KDM2A
amplification. (A) Knockdown of KDM2A in SUM52 cells with two different shRNAs was
confirmed by qRT-PCR. The primers detect both isoforms of KDM2A. (B) TurboGFP images of
SUM52 cells after viral infection with control shRNA and KDM2A shRNA #1 and #3.
Figure S7. Effects of inhibiting KDM2A on anchorage-independent growth of (A) ZR75-1 and
(B) T47D breast cancer cells. ZR75-1 and T47D cells transduced with a nonsilencing control
shRNA or with KDM2A shRNAs were used for soft agar assays. Cells were fed 3 times/week
for 3-4 weeks, stained with 500 µg/ml p-iodonitrotetrazolium violet (Sigma-Aldrich, St. Louis,
MO, USA) overnight, photographed, and counted with an automated mammalian cell colony
counter (Oxford Optronix GELCOUNT, Oxford, United Kingdom). Bar chart shows data from
three independent experiments, mean ± SD (*p<0.05, **p<0.01). (C) KDM2A knockdown
inhibited the invasive capacity of ZR75-1 cells. (C) shRNA knockdown of KDM2A inhibited the
invasive capacity of ZR75-1 cells.
Figure S8. MTT assays showed that knockdown of the KDM2A-S isoform, but not the
KDM2A-L isoform, inhibited T47D breast cancer cell growth. (A) qRT-PCR detection of
KDM2A-S isoform in T47D cells at day 5 after siRNA transfection. si=siRNA. (B) MTT assays
showed that knockdown of the KDM2A-S isoform, but not the KDM2A-L isoform, inhibited
T47D breast cancer cell growth.
Figure S9. (A) MTT assays showed that knockdown of the KDM2A-S with two specific siRNAs
had only a minor effect on MCF10A cell growth. (B) Representative images of MCF10A cells
after treatment with control siRNA and two KDM2A-S siRNAs at day 5.
Figure S10. Knockdown of the KDM2A-S isoform, but not the KDM2A-L isoform, inhibited the
invasive capacity of ZR75-1 cells.
Figure S11. ChIP-Seq binding peaks for KDM2A at the braf and fabp5 genomic loci in mouse
embryonic stem cells. The data were obtained from Cistrome
(http://cistrome.org/Cistrome/Cistrome_Project.html).