University of Southern California Dissertations and Theses

Transposable element suppression in basal-like breast cancer

(USC Thesis Other)

Transposable element suppression in basal-like breast cancer

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Biometrics 64, 1–21 DOI: 123456
January 2019
Transposable Element Suppression in Basal-like Breast Cancer
Anthony Colombo
Keck School of Medicine of University of Southern California, Los Angeles, California, 90033, USA.
Department of Preventive Medicine, Biostatistics.
University of Southern California
Requirement of the Master of Science in Biostatistics Degree.
email: anthonycolombo60@gmail.com
Abstract: The Cancer Genome Atlas (TCGA) described many large-scale transcriptomes in cancer, but did not
examine transposable elements. In order to further understand transposable elements in breast-cancer, we analyzed
the transcriptome of 484 subjects using TCGA, and compared the repeatomes from each subtype classification.
DespiteBasal-likehypo-methylation,transposableelementshadsuppressedexpression,whereasLuminaltypesshowed
increased levels.
Key words: Basal-like Breast Cancer; Interferon Response; Transposable Elements.
ThispaperisarequirementfortheMasterofScienceinBiostatisticsUniversityofSouthernCalifornia
May, 2019
Transposable Element Suppression in Basal-like Breast Cancer 1
CONTENTS
1 Introduction 3
2 Results 7
2.1 TCGA TE Phenotype Profile: TE Heterogeneity in Breast Cancer Subtypes 7
2.2 Down-Regulation of Basal-like and HER2-enriched Transposable Elements 7
2.3 AE-TE in BRCA Shows Distinct Basal-like Clustering 10
2.4 TCGA Checkpoint Analysis of Bulk Primary Breast Tissues 10
2.5 TCGA Inflammatory Signature in Primary Breast Samples 11
3 Discussion 11
4 Methods 13
4.1 Data Acquisition 13
4.2 Data Processing 14
4.3 Data Normalization 14
4.4 Principal Component Analysis 14
4.5 Altered Expression and Differential Expression 15
4.6 Antiviral Pathway Analysis 15
5 Figure Legends 16
5.1 Figure 1 16
5.2 Figure 2 16
5.3 Figure 3 17
5.4 Figure 4 17
5.5 Figure 5 17
6 Supplementary Materials 17
6.1 Supplementary Table 1 17
6.2 Supplementary Table 2 18
2 Biometrics, January 2019
6.3 Supplementary Table 3 18
6.4 Supplementary Table 4 18

Transposable Element Suppression in Basal-like Breast Cancer 3
Acknowledgements
The author thanks Dr. Giridharan Ramsingh for his brilliance and identification of the
biological model, Dr. Julie Lang for her aid in building the breast cancer model, and Dr.
Pushpinder Kaur for her helpfulness and support.
1. Introduction
Repetitive elements, or transposable elements (TE), are non-coding regions of the genome
which consist of greater than 50% of the genome. TE are known as ‘jumping genes’, or
retrotransposableelements,duetotheirabilitytoreinsertthemselvesbackintothegenome,
whereas coding genes form messenger RNA (mRNA) and exit the nucleus purposed for
translation. TE elements can insert themselves autonomously into the genome using a ‘cut-
and-paste’mechanism,whileotherTEsrequiretranscriptionfactorsandarenonautonomous.
Two classes of retrotransposons are long terminal repeat (LTR) retrotransposons and non-
LTR retrotransposons
[12][14]
. The LTR retrotransposons have structural similarity to that of
the retrovirus with the exception of the function env gene
[14]
. LTRs are transcribed by RNA
Pol II to form proteins that create chains of virus-like particles and re-insert themselves into
the host genome
[19]
. The common types of LTR retrotransposons include several endogenous
retrovirus biotypes ERVK, ERV1, and ERVL.
Thenon-LTRretrotransposonsconsistofLINEsthatincludeactiveLINE-1 (L1)elements
and inactive LINEs (L2). The human genome consists of 21% of autonomous non-LTR
retrotransposons, and approximately 17% of the genome consists of L1 elements
[12]
. LINEs
contain two open reading frames (ORF) and are autonomous non-LTRs because they can
be encoded by reverse transcriptase and transcribed by RNA Polymerase II. The non-
LTR retrotransposons also include short interspersed nuclear elements (SINEs). SINEs are
non-autonomous because they do not contain an ORF. Their transcription is dependent
on the reverse transcriptase recruitment from a separate element and are transcribed by
4 Biometrics, January 2019
RNA Polymerase III. In contrast to pluripotent stem cells, L1 retrotransposons are rarely
expressed in differentiated stem cells
[10]
. L1s are frequently transcribed in human embryonic
stemcells(hESCs),buttheL1 genomicre-integrationisnotdetectableatthisstage.Rather,
L1 re-insertions occur during embryogenesis
[19]
. Genomic integration of L1 transposable
elements may have influenced genetic inheritance and played an evolutionary role in the
development of primate lineages
[13]
.
Recentstudiesindicatedthattransposableelementexpressioninleukemicstemcells(LSC)
were uniformly suppressed
[8]
, whereas in leukemic blasts TE were significantly activated
across all TE biotypes. Additionally, blasts had up-regulation of interferon response in
comparison to LSCs which suggests immune-mediated cancer cell killing is likely to target
leukemic blasts, and ignore LSC predecessors. In myelodysplasia (MDS) subjects with high-
risk for AML disease progression were observed to have repressed transposable element
expressionandun-activatedinterferonresponse,whereaslow-riskpatientsrevealedactivated
TEandType-1interferonsignaling
[8]
.Thesemodelssuggestamechanismofdiseaseprogres-
sion becausehigh-riskMDSandLSCcellsescapeimmunedetectionthroughTEsuppression
and subsequent silencing of interferon immune signaling.
TherelationshipbetweenTEactivationandinterferonimmune-mediatedclearancehavere-
centlybeenexploredinsenescencemodels.Recentreportsexaminedthetransposableelement
signature comparing senescent and active proliferating cells using immortalized Li-Fraumeni
fibroblast cell lines
[5]
. Li-Fraumeni syndrome (LFS) is caused by germline p53 mutations,
andischaracterizedbyahereditarypredispositiontonumerouscancers.Colombo,Elias,and
Ramsingh
[5]
observed in LFS senescent cells that TE mRNA were uniformly up-regulated
compared to active non-senescent cells. Interestingly, the TE activation in senescent cells
corresponded with activated anti-viral pathway defense and increased interferon stimulated
gene mRNA. Additionally, senescence induction through DNA damaging agents such as
Transposable Element Suppression in Basal-like Breast Cancer 5
Adriamycin, 5-Azacytidine, and oxidative stress had a similar TE response compared to
natural, replicative, senescence. Senescence revealed correspondence between TE signaling
and pattern recognition sensors such as RIG-1, IFNB1, suggesting a beneficial role of TEs
and immune surveillance. However, the active proliferating cells, likely responsible for the
cancerous development in LFS, indicated immune escape by negative regulation of TE.
Transposable elements are regulated by methylation, and their expression can be sup-
pressed by compact chromatin arrangements. DNA methylation inhibitors (DNMTis) have
beenshowntoactivatetransposableelementswhichleadtoanti-viraldefenseandsubsequent
interferon response
[4]
. Hypomethylating agents in low-dose, such as 5-Azacytidine, have
significant anti-tumor effects that are mediated by the activation of MDA5 through a viral
recognition (dsRNA) response
[17]
. Interferon stimulation by TE activation from hypomethy-
lating agents may have important anti-tumor treatment potential in other cancers.
In order to identify therapeutic targets in other diseases, studies of the relationship be-
tween genomic instability, DNA loss of methylation, and TE induction must be explored
[1]
.
Non-coding elements have been studied in colorectal cancer, but the TE transcriptome
profile has not been adequately described in other large scale cancer databases. The Cancer
Genome Atlas breast cancer study
[9]
(TCGA-BRCA) described in detail 484 breast-cancer
transcriptomes and their mutational profile, but did not examine transposable elements. Us-
ing TCGA, Colombo, Triche Jr., and Ramsingh described the adult acute myeloid leukemia
(TCGA-AML) TE transcriptome profile, and survival prognosis in TCGA-AML, pediatric
(TARGET-AML), and relapsed AML
[7]
. In adult AML(TCGA-AML) TP53 mutations cor-
responded with significant down-regulation of ERV1 biotype
[7]
. Based on these findings in
AML we hypothesize that among the TCGA-BRCA Basal-like subjects, ERV1s should have
down-regulation compared to the other breast cancer subtypes because they had the largest
proportion (84%) of TP53 mutations.
6 Biometrics, January 2019
The TCGA-BRCA study classified primary breast cancer tissues into 4 major subtypes:
Basal-like, LuminalA, LuminalB, and HER2-enriched. The breast cancer subtypes are iden-
tifiedusinggeneexpressionorimmunohistochemistrypathologicalstaining(IHC).Basal-like
breast cancer is known as triple-negative breast cancer because the cancer is negative for
hormone, estrogen and progesterone receptors. LuminalA breast cancer subtype is estrogen-
receptorpositiveandhumanepidermalgrowthfactorreceptor2(HER2)negative.LuminalA
cancer has lower levels of cancer proliferation indicated by low levels of Ki67 protein and
are usually favorable prognosis. In contrast, LuminalB subtype is estrogen-receptor positive
and can be either HER2 positive or negative. LuminalB breast cancer is known to have
increased Ki67 and worse survival prognosis. A TCGA-BRCA study indicated that Basal-
like subtype showed significant hypomethylation
[9]
, whereas Luminal A/B subtypes showed
significant hypermethylation. The survival prognosis of the main breast cancer subtypes and
the relationship with methylation gain/loss are widely studied, but the role of transposable
elements in this relationship is not well understood.
DNA methylation at the transposon promoter sites regulates the retrotransposition of
many TE and may indicate a relationship between cancer risk and L1 loss of DNA methy-
lation
[2]
. The non-LTR non-autonomous family may also have an epigenetic role in cancer
becauseSINEssuchasSVAcontainabundantCpGsandarehighlycontrolledbymethylation
in human tissues and human embryonic stem cells
[18]
. Given that TCGA-BRCA Basal-like
subjects showed significant hypomethylation, we hypothesize that Basal-like cancer should
have increased L1 and SVA expression compared to Luminal A/B subtypes. Our study
used the TCGA-BRCA database to characterize TE in the breast cancer transcriptome and
describe the correspondence between genomic instability, DNA methylation loss, and TE
dysregulation.
Transposable Element Suppression in Basal-like Breast Cancer 7
2. Results
2.1 TCGA TE Phenotype Profile: TE Heterogeneity in Breast Cancer Subtypes
Principalcomponentanalysis(PCA)ofthenormalizedcounts-per-million(CPM)expression
of the complete transposable element transcriptome (746 canonical TE transcripts) in 484
BRCA subjects (TCGA) identified the patient variability of TE expression (Figure 1A). We
obtained each patient’s breast cancer subtype from the Genomic Data Commons (GDC),
allowing for comparisons of each subtype. The variation explained by the first two principal
componentswas95.4%(PC1=94.9%,PC2=0.43%,PC3=0.18%)andthereforePC1andPC2
wereused.Wethencomparedeachsubtype’sprincipalcomponent(PC)coordinatedistances.
The Euclidean distances of the patient PCs from Basal patients to the LuminalA subjects
identified Basal like subjects were significantly further from LuminalA (Euclidean distance
of centroids =0.075, wald t-test 2 sided multiple comparison adjusted p-value=0.0095).
Comparing the Euclidean distances of Basal subjects to LuminalB subjects (Euclidean dis-
tance of centroids =0.071, wald t-test 2 sided multiple comparison adjusted p-value=0.013).
Comparing the Euclidean distances of Basal-like subtype to HER2-enriched subjects did not
identifysignificantdifferencesoftheircentroids(Euclideandistanceofcentroids=0.038,wald
t-test 2 sided multiple comparison adjusted p-value=0.12). The patients classified as Basal-
like(N=107) showed furthest separation from LuminalA (N=196), and LuminalB (N=122)
patients,butdidnothavesignificantseparationfromHER2-enriched(N=50).Thisindicates
thattheBasal-liketransposableelement(TE)transcriptomeismoreheterogeneous,whereas
the Luminal A/B subtypes show more TE homogeneity.
2.2 Down-Regulation of Basal-like and HER2-enriched Transposable Elements
We constructed a broad multivariate linear model that included independent variables such
as breast cancer subtype classification (LuminalB (N=122), HER2-enriched (N=50), Lumi-
nalA(N=196),andBasal-like(N=107)),mutationalhotspots(suchasATM(N=50),RB1(N=7),
8 Biometrics, January 2019
GATA3(N=23),PIK3CA(N=62),FGFR2(N=6),MAP3K1(N=17),BRCA1(N=6),AKT1(N=10),
and TP53(N=83)), age (above the median age of 58.05 years, N=242), race (white/non-
white, N=341), and immunohistochemistry clinical categorization such as TNBC status
(N=64), ER-/PR-/HER2+ (N=22), ER+/PR+/HER2- (N=146). The dependent variables
were the TE transcript expression.
The multivariate model identified 76 TE transcripts with altered expression (AE-TE) (i.e
with a minimum absolute log-FC>0.5, and hierarchical test: Bonferroni-Hochberg (BH) ad-
justedFDR<0.05).ThesignificantAE-TEtranscriptbiotypesincluded37ERV1,10LINE-1,
7 ERVL, 5 ERVK, 5 hAT, 2 ERV3, 2 SVA, 1 TcMar, 1 Satellite, 1 LTR Retrotransposon,
1 SAT, 1 Endogenous Retrovirus, 1 snRNA, 1 PiggyBac, and 1 centromeric (Figure 1B,
Figure 1D). Interestingly, the Basal-like phenotype had significant up-regulation of SVA TE
biotype,andmarginalup-regulationofERVK.Furthermore,ERV1 andERVLLTRbiotypes
showeddown-regulation.HER2-enriched,LuminalA/Bsubtypesshowedup-regulationofL1
non-LTRs,ERVK,andLTRretrotransposons,anddown-regulationofhAT class(Figure1D).
The Basal-like AE-TE estimates were adjusted for potential confounders such as age, race,
and1%prevalentgeneticmutations.TheAE-TEpredictedbyBasal-likeandHER2-enriched
subtypes yielded an overall marginal negative median average dysregulation rate (absolute
median< 1).Incontrast,LuminalA/Bsubtypeshadapositivemedianaveragedysregulation
rate (Figure 1E). The multivariate model identified 15 AE-TE in the Basal-like subtype, 12
of which were uniquely dysregulated in the Basal-like subjects (Figure 1C, Supplementary
Table 1). The 12 Basal-like AE-TE included 5 ERV1 (HERVK3I, LTR12E, LTR25, MER4E,
MER84I), 2 ERVL (MLT1C, MSTC), 2 SVA (SVA A, SVA F), 1 hAT (CHARLIE9),
1 ERVK (LTR5 Hs), and 1 centromeric (GSAT) (Supplementary Table 1). Compared to
LuminalA, Basal-like patientshad uniqueup-regulation ofthebiotypefamily SVA(adjusted
averagelogFC:1.93,pairwisetest:BHadjustedFDR<0.05)(Table1).TheBasal-likeERVK
Transposable Element Suppression in Basal-like Breast Cancer 9
biotypeclassidentifiedtranscriptsLTR5,LTR5 Hs asup-regulated(adjustedaveragelogFC:
0.403,pairwisetest:BHadj.FDR < 0.05).Thedown-regulatedtranscriptsintheERV1bio-
types were HERVK3I, LTR12E, LTR25, MER84I, and MER87B) (adjusted average logFC:
-0.365,pairwisetest:BHadjFDR <0.05).ERVLdown-regulationofTEtranscriptsincluded
MLT1C and MSTC (adjusted average logFC: -0.366, pairwise test, BH adj FDR<0.05).
Interestingly, Basal-like subtype did not have any LINE1 dysregulation.
The multivariate linear model identified 17 AE-TE in HER2-enriched subjects (hierar-
chical test: BH adj. FDR<0.05, min. absolute logFC>0.5)(Figure 1C). Of the 17 AE-TE,
7 were uniquely dysregulated in HER2-enriched subjects including 4 ERV1 (HERVK11DI,
MER57E3, MER65I, MER9), 1 ERVK (LTR14C), 1 PiggyBac (MER75B), and 1 TcMar
(TIGGER5.A). Compared to Basal-like subtype, HER2-enriched patients had up-regulated
L1(avglogFC=0.898,pairwisetesttranscripts:L1HS,L1PA2),andERV1(avglogFC=0.0314,
pairwisetesttranscripts:LTR31,HERVK11DI,HERVS71,MER65I,MER52C,MER61F,MER65C)
(Table 2).
The multivariate linear model indicated LuminalB subtype had 6 unique AE-TE that
included 3 L1 (L1PA3, L1PA5, L1PA6), 1 Satellite (BSRa), 1 ERV1 (MER69A), and 1
ERVL (THE1A). Compared to Basal-like subtype, LuminalB yielded LINE1 up-regulation
(avg logFC: 0.686, pairwise test: BH adj. FDR<0.05)(Figure 1D, Table 3)).
LuminalAsubtypehad4uniqueAE-TEwhichincluded3ERV1(HERVFH21I,HERVL66I,
MER39B), and 1 ERVK (LTR14). Compared to Basal-like subtype, LuminalA yielded
LINE1 up-regulation (avg logFC: 0.567, pairwise test: BH adj. FDR<0.05) and up-regulated
ERV1 (avg. logFC:0.706, pairwise test: BH adj. FDR < 0.05) (Figure 1D, Table 4)).
In summary, compared to Basal-like subtype, HER2-enriched, LuminalA and LuminalB
subtypes showed up-regulation of L1 expression. Basal-like subtype indicated no significant
10 Biometrics, January 2019
alterations of L1 transcripts. Additionally, Basal-like breast cancer revealed unique up-
regulation of SVA TE, and marginal down-regulation of ERV1 transcripts.
2.3 AE-TE in BRCA Shows Distinct Basal-like Clustering
The 76 AE-TE (absolute minimum log fold change>0.5, hierarchical test: BH-adjusted
FDR<0.05) identified from the multivariate linear regression model revealed a distinct clus-
tering of patient expression profile of Basal-like subjects (Figure 2). Although the Basal-
like covariate was ranked 4th in the magnitude of TE dysregulation (Figure 1B), the AE-
TE expression profile revealed a distinct TE clustering of Basal-like subjects. Interestingly
LuminalB subtype had the highest magnitude of absolute TE dysregulation; however the
correspondingexpressionprofileofthe76AE-TEdidnotidentifyauniquepatientclustering
by the Prediction Analysis Microarray 50 genes (PAM50) nor immunohistochemistry clini-
cal classification (IHC) (Figure 1A). The heatmap of the normalized transcript-per-million
(TPM) expression indicates that SVA F and SVA A had the highest differential expression.
The LINE1 and ERV1 down-regulation in Basal-like subjects appears marginal.
2.4 TCGA Checkpoint Analysis of Bulk Primary Breast Tissues
Wehypothesizedthatthedown-regulationofTEsignalsinBasal-likebreastcancermaybea
resultofimmuno-suppressiveenvironment,andhenceexaminedtheimmunecheckpointgenes
of the primary bulk breast cancer tissues in TCGA-BRCA patients. We examined immune
checkpointgeneexpressioninabroadmultivariatelinearmodel,andalsoperformedpairwise
testing comparing Basal-like subtype to the others (BH adj.p.val< 0.05). LuminalA showed
themoststatisticallysignificantdown-regulationofCD40,CD86,HAVCR2,CTLA4,TIGIT,
and BTLA (absolute minimum logFC>0.5,pairwise test, BH adjusted FDR<0.05)(Table 5).
Basal-like immune-related pathway activation corresponded to significant up-regulation of
CEACAM1,andLAG3 (absoluteminimumlogFC>0.5,pairwisetestBHadjustedFDR<0.05),
and marginal up-regulation of TIGIT and PDL1(Table 5, Table 6, Table 7). The immune
Transposable Element Suppression in Basal-like Breast Cancer 11
checkpoint gene expression analysis suggests that Basal-like subtype has increased immuno-
suppressive environment compared to the other subtypes.
2.5 TCGA Inflammatory Signature in Primary Breast Samples
Inflammation and breast cancer has been widely studied, and as a positive control, our
model indicated increased inflammatory cytokines/chemokines gene expression. We exam-
ined TCGA primary breast cancer samples chemokine/cytokine ligand expression across
subtypes. There were 57 chemokines/cytokines significantly altered in expression (absolute
minimum logFC>0.5, nested F test: BH adjusted FDR<0.05) across the subtypes. The
Basal-like subtype showed 11 down-regulated, and 14 up-regulated genes that included
CXCL16, CCL7, CCL8, CCL1, CCL20, CXCL3, CXCL5, CXCL1, CXCL10, CXCL11, and
CXCL8 (Supplementary Table 4). The LuminalA subtype showed 38 cytokines/chemokines
significantly uniformly down-regulated and 0 up-regulated. LuminalB subtype showed down
regulation of 30 cytokines/chemokines, with only CCL7 showing up-regulation. The HER2-
enriched subtype showed 9 up-regulated, and 17 down-regulated genes (Figure 4). Inflam-
mation has been reported in Basal-like breast cancer, and our model validates Basal-like
increased expression of pro-inflammatory cytokines/chemokines ligands in primary tissues.
3. Discussion
Previous comprehensive molecular studies of breast cancers (TCGA)
[9]
indicated that Lu-
minalB subtype showed hypermethylation phenotype and 32% TP53 mutations, whereas
PAM50 Basal-like subtype had the lowest levels of DNA methylation and 84% TP53 muta-
tions. The HER2-enriched subtype had a modest methylation phenotype, with 75% TP53
mutations. In hypomethylated Basal-like subtype, we observed a unique up-regulation of
SINE-Variant Alu transposable element transcript expression, but did not observed signifi-
cantalterationsinexpressionofLINEsnorAlu.WeexpectedtheBasal-likehypomethylation
12 Biometrics, January 2019
to correspond with an increase of TE retrotransposition; however TE biotype retrotrans-
position may occur at different times during tumorigenesis. Hence, for Basal-like cancer,
LTR retrotransposition of ERV1 and ERVL may be an early cancer event suggesting that
suppression in primary breast tissue may indicate previous activation.
TEsuppressioncouldalsoindicatethatprioractivationresultedinaviraldefenseresponse
and/or TE induced genomic instability leading to cancer cell killing. In the development of
Basal-likebreastcancer,ERVKandnon-LTRclassSVAmayhavedelayedretrotransposition.
This may suggest that in Basal-like cancer development, cell clearance may involve ERV1 or
LINEs at early stages. Since SVA and ERVK were observed as active in primary tissue, the
tumorcellsmayhavedevelopedtolerancetotheseTE,indicatingthatBasaldiseaseprogres-
sion requires the depletion of ERV1/L1s but is tolerable to SVA/ERVK. Future prospective
studies are important for elucidating important hypo-methylation and retrotransposition
time points of each TE class.
Prospectivebreast-cancerstudiesfocusedonnon-LTR(AluandLINE-1)epigeneticchanges
[15]
during cancer development suggest that LINE-1 elements are hypomethylated at initial
disease stages. Park et.al observed that Alu methylation levels did not significantly differ
comparing normal breast tissues, atypical ductal hyperplasia, ductal carcinoma in situ, and
inflammatory breast cancer. The lack of hypomethylation of Alu could explain why our
study did not identify Alu elements as variably expressed. Furthermore, Park et.al showed
LINE-1 de-methylation in HER2-enriched subtype, but only at the earliest stages of disease.
This could support our observation showing HER2-enriched with increased LINE-1, but
furtherinvestigationsareneededtounderstandtherelationshipbetweenLuminalA/BLINE-
1 activation and its hypermethylation status.
The immune checkpoint gene expression coupled with an inflammatory signature suggests
that Basal-like breast cancer has an immuno-suppressive environment. This may support
Transposable Element Suppression in Basal-like Breast Cancer 13
a future hypothesis that Basal TE suppression results from a previous immune attack at
an early cancer stage, and checkpoint regulation in later periods. It is widely known that
triple-negative breast cancer subjects respond to checkpoint inhibitors, but future studies
investigating potential roles of TE in immune checkpoint therapy are warranted. Although
weobservedmarginalLINE1andERV1suppressioninBasal-likebreastcancer,theantiviral
gene signature suggests that, similar to the LFS model, Basal-like cells with activated TE
could have been arrested. Future studies which interfere with APOBEC defense response
would elucidate the role in antiviral mechanisms and TE induction that could identify
improved therapies.
4. Methods
4.1 Data Acquisition
The data were acquired with dbGap access and downloaded FASTQs were obtained from
the Genomic Data Commons (GDC) TCGA-BRCA study. The metadata were obtained
from the GDC portal. Dr. Julie Lang provided dbGAP access, and the author downloaded
and processed all files. The total sample size was 484 subjects. The patient metadata had
age as a continuous variable, but we dichotomized age recorded at diagnosis (1 : above the
medianageof58.05,and0:otherwise)inordertosimplifytheinterpretationofthemultivari-
ate model. The model covariates included race (white/non-white), age at diagnosis (above
median age 58.05 years/ below median), basal-like invasive ductal carcinoma (IDC) classifi-
cation, HER2-enriched IDC classification, LuminalB classification, LuminalA classification,
ER+PR+HER2- Immunohistochemistry (IHC) classification, ER-PR-HER2+ IHC classifi-
cation, triple negative breast cancer classification, and patient mutational hotspot identi-
fication from whole exome sequencing including the following mutations: PIK3CA, NF1,
PIK3CB, RPTOR, AKT1, FBXW7, PIK3R1, GRB7, MAP3K1, KRAS, EGFR, MAP2K1,
14 Biometrics, January 2019
JAK2,ERBB2,ERBB3,CCND3,CDK6,TP53,MDM2,RB1,MDM4,NOTCH4,NOTCH1,
TBX3, MET, CXCR4, FGFR2, FGFR1, ATM, PALB2, BRCA1, BRCA2, GATA3, IL4,
TGFB1, FOXA1, ESR1, AR, ESR2, and PGR (Table 8).
4.2 Data Processing
The RNA-seq reads were quantified using Kallisto
[3]
. The data were containerized using
Arkas
[6]
, and the annotations for the transcripts were obtained through Arkas
[6]
.
4.3 Data Normalization
The individual transcripts were collapsed into their corresponding gene ID. We removed
low counted transcripts to control for low read count biases. The minimum read count
threshold was 16 for inclusion in this analyses. Genes with fewer counts were discarded. The
subsequentdatawerenormalizedusinglimma/voom
[16]
.Theheteroscedasticnormalizeddata
were used in Empirical Bayesian linear models to identify differentially expressed genes/TE
using limma.
4.4 Principal Component Analysis
Using 747 transposable element transcripts, the principal components of patients were iden-
tified. Using the GDC metadata, we computed the centroids of each breast cancer subtype
defined as the average PC1 location, and the average PC2 location. We then calculated the
Euclidean distance comparing the Basal centroid to each other subtypes. The measure of
separation used the Euclidean distance defined as
• d(basal,reference):
q
(PC1
basal
−PC1
reference
)
2
+(PC2
basal
−PC2
reference
)
2
The null hypothesis is defined that the d(basal,reference)=0, and the alternative is that
the d(basal,reference) 6= 0. The Wald t-test was computed t=
d(basal,reference)−0
√
SE(d(basal,reference))
. Using
the t-value, the p-value was computed and multiplied by 3 to correct for multiple pairwise
comparisons.
Transposable Element Suppression in Basal-like Breast Cancer 15
4.5 Altered Expression and Differential Expression
The multivariate altered-expression elements were identified using the hierarchical testing
method with alpha threshold of 0.05 and a multiple test correction (Benjamini-Hochberg)
adjusted p-value threshold of 0.05. In addition to the adjusted p-value thresholds, we in-
cluded a threshold on the absolute log-fold-change in order to filter marginal genes/TE. The
minimum absolute log fold-change threshold was set to 0.50.
ForthedetectionofdifferentialTE/geneexpression,weconductedpairwisecomparisonsof
Basal-like to LuminalA subtypes (Table 1), HER2-enriched (Table 2), and LuminalB (Table
3) using Benjamini-Hochberg adjusted p-value threshold of 0.05. We reported the AE-TE
differential expression only.
The ‘median average dysregulation rate’ used the median logFC of all the AE-TE cor-
responding to each PAM50 subtype. The logFC represents an average dysregulation rate,
hence the median of the dysregulation rates defines median average dysregulation.
Forasupervisedmultivariateanalysisofaltered-expressedinflammatorychemokines/cytokines
andimmunecheckpointgenes,anested-Ftestwasconducted(BHadjustedalphalevel0.05).
Fortheadjustedestimatesinregardtoimmunecheckpointgenes,pairwisecomparisonswere
conducted on Basal-like subtype to the others (BH adj.p-value< 0.05).(Table 5, Table 6,
Table 7).
4.6 Antiviral Pathway Analysis
The antiviral pathways included gene sets related to antiviral defense defined as APOBEC,
Helicases (DDX/DHX), RNA interference genes, and post-transcriptional regulators of TE.
These gene sets were identified by Goodier
[11]
. The pathway analysis was performed using
Quantitative Set Analysis Gene Enrichment (QuSAGE)
[20]
. QuSAGE reports gene-sets with
Benjamini-Hochberg adjusted p-value threshold of 0.05. Furthermore, QuSAGE adjusts for
variance inflation factor of the correlated genes within a pathway set.
16 Biometrics, January 2019
5. Figure Legends
5.1 Figure 1
Transposable Element Expression Summary of Invasive Ductal Carcinoma of
Primary Breast Cancer Tissue (TCGA): A) Principal component analysis of normal-
ized TE transcript expression from 484 subjects. Each color corresponds to the molecular
subtype classification. X-axis are the coordinates of the first component. Y-axis is second
PC component coordinates. B) Multivariate linear model which includes classification, and
prevalentmutations,andpatientclinicalfeatures.Thex-axiscorrespondstothefeaturesused
in the multivariate linear model. The y-axis corresponds to the number of TE transcripts
significantly altered in expression (AE-TE) per feature. The color indicates up/down dys-
regulation. C) Venn diagram depicting overlaps of dysregulated TE transcripts predicted by
classification.D)ThebiotypefamilysummaryofaveragedysregulationofAE-TEtranscripts
identified in B. The x-axis indicates features used in the multivariate model. The y-axis
denote the TE transcript biotypes. E) The boxplot summary of expected dysregulation rate
corresponding to significant AE-TE identified from the multivariate model in B. The x-axis
indicates the subtypes. The y-axis indicates the log2 fold-change of the AE-TE.
5.2 Figure 2
TCGA Expression of the AE-TE Identified from the Multivariate Linear Model:
The y-axis indicates TE transcripts identified from the multivariate model (Figure 1B)
defined as significant AE-TE. Adjacent to each TE transcripts indicates the transcript
biotypefamily,andgenebiotypefamily.Thex-axisarethe484patientsamplesfromTCGA.
The top-most colors indicate patient molecular subtype. The bottom-most colors indicate
TNBC, Estrogen receptor status, Progesterone receptor status. The expression transcripts
per million (TPM) values are log2 transformed.
Transposable Element Suppression in Basal-like Breast Cancer 17
5.3 Figure 3
TCGA Breast Cancer Patient Immune Checkpoint Altered Expression: The x-
axisareTCGApatientsamples,they-axisdenotestheimmunecheckpointgenessignificantly
alteredinexpressionidentifiedbythebroadmultivariatemodel(BHadjustedp.value<0.05).
5.4 Figure 4
Cytokine Gene Dysregulation in TCGA Breast Cancer Patient Samples: The
x-axis denotes patient TCGA samples. The y-axis indicates cytokine AEG. The top-most
annotation bar indicates the patient subtype classification.
5.5 Figure 5
TCGAAPOBECAntiviralGeneActivityUsingTCGA-BRCAsamples,weperformed
a gene-set enrichment analysis (QuSAGE, FDR<0.05) comparing Basal-like samples to non-
Basal-like subtypes. The x-axis denotes the APOBEC genes, the y-axis denotes the log2 fold
change, each point contains a 95% confidence interval.
6. Supplementary Materials
6.1 Supplementary Table 1
TCGA Transposable Elements Significantly Altered in Expression Identified
from Multiple Linear Regression.
Using multiple linear regression previously described, the multivariate model identified
AE-TE using hierarchical test with Benjamini-Hochberg filtering procedure adjusted p.value
threshold< 0.05, and absolute minimum fold-change of 0.50.
• 1.A: The rows are the 12 unique AE-TE predicted for Basal-like subtype. The logFC and
BH.adj.P.val are from the multiple regression model.
18 Biometrics, January 2019
• 1.B: The 7 unique AE-TE predicted by HER2-enriched subtype. The logFC and BH
adj.p.value are from the multiple regression model.
• 1.C: The 4 unique AE-TE predicted by LuminalA subtype. Labeling similar to ST.1.
• 1.D: The 6 unique AE-TE predicted by LuminalB subtype. Labeling similar to ST.1.
6.2 Supplementary Table 2
TCGA Immune Checkpoint Gene Dysregulation.
We identified immune checkpoing genes with altered expression using parameters absolute
min logFC>0.5, and BH adjusted p.value threshold of 0.05 (nested-F test) in multiple
regression. The rows correspond to the significant altered checkpoint genes, and the columns
denote the multiple regression testing results (+1 up-regulated, -1 down-regulated).
6.3 Supplementary Table 3
TCGA Antiviral Pathway Enrichment Activity.
Goodier
[11]
describedgenesinvolvedaslinesofdefenseagainsttransposableelementswhich
included Helicases, RNA interference genes, and APOBEC family. We tested each pathway
for significant enrichment in Basal - like samples compared to non-Basal-like samples. The
rows correspond to the pathway statistics for each antiviral gene set
[20]
.
6.4 Supplementary Table 4
TCGA Cytokines/Chemokines Statistically Significantly Altered in Expression.
The rows correspond to cytokines/chemokines that are altered in at least one predictive
features using parameters absolute minimum fold change>0.05, nested-F test coupled with
Benjamini-Hochberg adjusted p.value threshold<0.05 (q.value). The columns correspond to
the features used in the multiple regression.
Transposable Element Suppression in Basal-like Breast Cancer 19
References
1. AnwarSL,WulaningsihW,LehmannU.TransposableElementsinHumanCancer:Causes
and Consequences of Deregulation. (2017). Int J Mol Sci.18(5).
2. Baba, Y., Watanabe, M., Murata, A., Shigaki, H., Miyake, K., Ishimoto, T.,... Baba, H.
(2014). LINE-1 hypomethylation, DNA copy number alterations, and CDK6 amplification
in esophageal squamous cell carcinoma. Clin Cancer Res, 20(5), 1114–1124.
3. Bray, N. L., Pimentel, H., Melsted, P., & Pachter, L. (2016). Near-optimal probabilistic
RNA-seq quantification. Nat Biotechnol, 34(5), 525–527.
4. Chiappinelli, K. B., Strissel, P. L., Desrichard, A., Li, H., Henke, C., Akman, B., . . .
Strick, R. (2015). Inhibiting DNA Methylation Causes an Interferon Response in Cancer
via dsRNA Including Endogenous Retroviruses. Cell, 162(5), 974–986.
5. Colombo, A. R., Elias, H. K., & Ramsingh, G. (2018). Senescence induction universally
activates transposable element expression. Cell Cycle, 17(14), 1846–1857.
6. Colombo, A. R., J. Triche T, J., & Ramsingh, G. (2017). Arkas: Rapid reproducible
RNAseq analysis. F1000Res, 6, 586.
7. Colombo,A.R.,Triche,T.,Jr.,&Ramsingh,G.(2018).TransposableElementExpression
in Acute Myeloid Leukemia Transcriptome and Prognosis. Sci Rep, 8(1), 16449.
8. Colombo, A. R., Zubair, A., Thiagarajan, D., Nuzhdin, S., Triche, T. J., & Ramsingh,
G. (2017). Suppression of Transposable Elements in Leukemic Stem Cells. Sci Rep, 7(1),
7029.
9. Comprehensivemolecularportraitsofhumanbreasttumours.(2012).Nature,490(7418),
61–70.
10. Estecio, M. R., Gharibyan, V., Shen, L., Ibrahim, A. E., Doshi, K., He, R., . . . Issa,
J. P. (2007). LINE-1 hypomethylation in cancer is highly variable and inversely correlated
with microsatellite instability. PLoS One, 2(5), e399.
20 Biometrics, January 2019
11. Goodier, J. L. (2016). Restricting retrotransposons: a review. Mob DNA, 7, 16.
12. Goodier, J. L., Ostertag, E. M., Engleka, K. A., Seleme, M. C., & Kazazian, H. H., Jr.
(2004).ApotentialroleforthenucleolusinL1retrotransposition. Hum Mol Genet,13(10),
1041–1048.
13. Lee, H. E., Ayarpadikannan, S., & Kim, H. S. (2015). Role of transposable elements
in genomic rearrangement, evolution, gene regulation and epigenetics in primates. Genes
Genet Syst, 90(5), 245–257.
14. Ostertag, E. M., & Kazazian, H. H., Jr. (2001). Biology of mammalian L1 retrotrans-
posons. Annu Rev Genet, 35, 501–538.
15. Park, S. Y., Seo, A. N., Jung, H. Y., Gwak, J. M., Jung, N., Cho, N. Y., & Kang, G.
H. (2014). Alu and LINE-1 hypomethylation is associated with HER2 enriched subtype of
breast cancer. PLoS One, 9(6), e100429.
16. Ritchie, M. E., Phipson, B., Wu, D., Hu, Y., Law, C. W., Shi, W., & Smyth, G. K.
(2015). limma powers differential expression analyses for RNA-sequencing and microarray
studies. Nucleic Acids Res, 43(7), e47.
17. Roulois,D.,LooYau,H.,Singhania,R.,Wang,Y.,Danesh,A.,Shen,S.Y.,DeCarvalho,
D.D.(2015).DNA-DemethylatingAgentsTargetColorectalCancerCellsbyInducingViral
Mimicry by Endogenous Transcripts. Cell, 162(5), 961–973.
18. Strichman-Almashanu,L.Z.,Lee,R.S.,Onyango,P.O.,Perlman,E.,Flam,F.,Frieman,
M. B., & Feinberg, A. P. (2002). A genome-wide screen for normally methylated human
CpG islands that can identify novel imprinted genes. Genome Res, 12(4), 543–554.
19. Tanaka, Y., Chung, L., & Park, I. H. (2012). Impact of retrotransposons in pluripotent
stem cells. Mol Cells, 34(6), 509–516.
20. Yaari, G., Bolen, C. R., Thakar, J., & Kleinstein, S. H. (2013). Quantitative set analysis
for gene expression: a method to quantify gene set differential expression including gene-
Transposable Element Suppression in Basal-like Breast Cancer 21
gene correlations. Nucleic Acids Res, 41(18), e170.
22 Biometrics, January 2019

TABLE 1

Adjusted estimates for AE-TE of Basal-like compared to LuminalA subtypes. (Adj. p-value
denotes the Benjamini-Hochberg adjusted p-value)

LogFC Lower 95%CI Upper 95%CI Adj. p-value(BH) TE biotype

SVA F 2.22759 1.86476 2.59043 0.00000 SVA
SVA A 1.63071 1.30736 1.95406 0.00000 SVA
GSAT 2.11677 1.58978 2.64376 0.00000 centromeric
HERVK3I -0.88134 -1.16742 -0.59525 0.00000 ERV1
LTR5

Hs 0.63051 0.41163 0.84939 0.00000 ERVK

LTR25 0.96792 0.58662 1.34922 0.00002 ERV1
MER87B -0.83251 -1.23562 -0.42939 0.00058 ERV1
MER84I -1.02130 -1.52991 -0.51268 0.00084 ERV1
MLT1C -0.36118 -0.56518 -0.15717 0.00376 ERVL
CHARLIE9 0.68850 0.21340 1.16361 0.02315 hAT
MSTC -0.37149 -0.63493 -0.10806 0.02810 ERVL
LTR12E -0.28910 -0.51077 -0.06742 0.04580 ERV1

Transposable Element Suppression in Basal-like Breast Cancer 23

TABLE 2

(BH) Adjusted estimates for AE-TE of HER2-enriched compared to Basal-like subjects.

LogFC Lower 95%CI Upper 95%CI Adj.p-value TE biotype

LTR31 -2.65813 -3.26905 -2.04722 0.00000 ERV1
HERVK11DI 2.00530 1.41703 2.59357 0.00000 ERV1
HUERS-P3 1.02431 0.68071 1.36791 0.00000 LTR Retrotransposon
HERVS71 1.33157 0.83612 1.82701 0.00001 ERV1
L1PA2 0.82671 0.51379 1.13963 0.00002 L1
LTR14C 1.16398 0.67444 1.65352 0.00015 ERVK
L1HS 0.96989 0.52767 1.41211 0.00062 L1
MER65I 1.60498 0.84271 2.36725 0.00119 ERV1
MER52C -0.62500 -0.92377 -0.32622 0.00124 ERV1
MER61F -0.69394 -1.13836 -0.24951 0.02571 ERV1
MER65C -0.74494 -1.23186 -0.25803 0.02965 ERV1

24 Biometrics, January 2019

TABLE 3

(BH) Adjusted estimates for AE-TE of LuminalB compared to Basal-like subjects.

LogFC Lower 95%CI Upper 95%CI Adj. p-value TE biotype

PRIMA4 I 1.92639 1.57576 2.27703 0.00000 Endogenous Retrovirus
LTR31 -2.59471 -3.12998 -2.05944 0.00000 ERV1
MER52C -0.91886 -1.18064 -0.65708 0.00000 ERV1
HERVKC4 1.92385 1.36377 2.48393 0.00000 ERVK
L1PA5 0.75004 0.52761 0.97246 0.00000 L1
HUERS-P3 0.97810 0.67705 1.27916 0.00000 LTR Retrotransposon
LTR18B -1.49332 -1.95451 -1.03214 0.00000 ERVL
LTR1C1 -1.54226 -2.06269 -1.02183 0.00000 ERV1
L1PA6 0.59229 0.38373 0.80085 0.00000 L1
LTR4 1.51456 0.98042 2.04870 0.00000 ERV1
L1PA2 0.68894 0.41477 0.96311 0.00002 L1
THE1A 0.36762 0.21946 0.51577 0.00002 ERVL
L1HS 0.86684 0.47938 1.25430 0.00017 L1
MER61F -0.85587 -1.24526 -0.46647 0.00023 ERV1
HERVS71 0.93265 0.49855 1.36675 0.00032 ERV1
L1PA3 0.53275 0.28188 0.78363 0.00038 L1
MER65C -0.79156 -1.21819 -0.36494 0.00227 ERV1
MER69A -0.67173 -1.05050 -0.29295 0.00368 ERV1
BSRa -0.44011 -0.78421 -0.09601 0.04898 Satellite

Transposable Element Suppression in Basal-like Breast Cancer 25

TABLE 4

(BH) Adjusted estimates for AE-TE of LuminalA compared to Basal-like subjects

LogFC Lower 95%CI Upper 95%CI Adj. p-value TE biotype

PRIMA4 I 2.10911 1.76447 2.45376 0.00000 Endogenous Retrovirus
MER39B 1.51257 1.19850 1.82663 0.00000 ERV1
HERVKC4 2.45703 1.90652 3.00754 0.00000 ERVK
HERVL66I 2.15301 1.65969 2.64634 0.00000 ERV1
HERVFH21I 1.40855 1.05838 1.75871 0.00000 ERV1
MER52C -0.98367 -1.24097 -0.72636 0.00000 ERV1
HUERS-P3 0.97936 0.68345 1.27527 0.00000 LTR Retrotransposon
LTR18B -1.45698 -1.91029 -1.00368 0.00000 ERVL
LTR4 1.64976 1.12475 2.17478 0.00000 ERV1
LTR14 0.88811 0.58707 1.18915 0.00000 ERVK
LTR1C1 -1.50679 -2.01833 -0.99525 0.00000 ERV1
L1PA2 0.56668 0.29719 0.83616 0.00044 L1

26 Biometrics, January 2019

TABLE 5

(BH) Adjusted estimates for checkpoint genes altered in Basal-like expression compared to LuminalA subjects

LogFC Lower 95%CI Upper 95%CI Adj. p-value TE biotype

ENSG00000089692 1.77586 1.28660 2.26512 0.00000 LAG3
ENSG00000163599 1.50240 0.96470 2.04009 0.00000 CTLA4
ENSG00000181847 1.12445 0.55193 1.69697 0.00074 TIGIT
ENSG00000121594 0.73310 0.32912 1.13707 0.00201 CD80
ENSG00000120217 0.65972 0.27241 1.04703 0.00406 CD274/PDL1
ENSG00000101017 0.54635 0.19247 0.90024 0.01019 CD40
ENSG00000079385 0.62161 0.11747 1.12575 0.04814 CEACAM1

Transposable Element Suppression in Basal-like Breast Cancer 27

TABLE 6

(BH) Adjusted estimates for checkpoint genes altered in Basal-like expression compared to LuminalB subjects

LogFC Lower 95%CI Upper 95%CI Adj. p-value Symbol

ENSG00000089692 1.02646 0.52869 1.52423 0.00040 LAG3
ENSG00000079385 1.03485 0.52195 1.54776 0.00055 CEACAM1
ENSG00000101017 0.65232 0.29229 1.01236 0.00218 CD40
ENSG00000163599 0.83814 0.29110 1.38519 0.01116 CTLA4
ENSG00000181847 0.74171 0.15924 1.32418 0.03915 TIGIT

28 Biometrics, January 2019

TABLE 7

(BH) Adjusted estimates for checkpoint genes altered in Basal-like expression compared to HER2-enriched subjects.

LogFC Lower 95%CI Upper 95%CI Adj. p-value Symbol

ENSG00000089692 0.81692 0.24881 1.38503 0.03083 LAG3

Transposable Element Suppression in Basal-like Breast Cancer 29
TABLE 8
TCGA-BRCA Clinical Summary

Demographical Covariates N (%)

Caucasian Race 341 70.45455
Age at diagnosis above median 58.05 years 242 50.00000

TCGA-BRCA Subtypes

LuminalA 196 40.49587
ER+/PR+/HER2- 146 30.16529
LuminalB 122 25.20661
Basal-like 107 22.10744
TNBC 64 13.22314
HER2-enriched 50 10.33058
ER-/PR-/HER2+ 22 4.54545

1% Prevalent Mutations

TP53 83 17.14876
PIK3CA 62 12.80992
GATA3 23 4.75207
MAP3K1 17 3.51240
AKT1 10 2.06612
RB1 7 1.44628
FGFR2 6 1.23967
ATM 6 1.23967
BRCA1 6 1.23967

Figure 1

A

B 5

C

D
Figure 2

TCGA AE-TE Expression

Figure 3

Figure 4

Figure 5

Transposable Elements Suppression in Basal-like Breast Cancer

Anthony R. Colombo, Pushpinder Bains, Akil Merchant, Julie
Lang*, and Giridharan Ramsingh*

Supplementary Information

1.A

Supplementary Table 1: TCGA Transposable elements significantly altered in expression identified from
multiple regression

AE-TE Basal-like AE-TE Hiearchical Test Result biotype logFC BH.adj.P.Val
CHARLIE9 1 hAT 1.176039954 0.00E+00
GSAT 1 centromeric 1.550586984 1.89E-289
HERVK3I -1 ERV1 -0.982870914 3.54E-19
LTR12E -1 ERV1 -0.710220582 6.79E-16
LTR25 1 ERV1 1.141477619 2.29E-170
LTR5_HS 1 ERVK 0.577674993 0
MER4E -1 ERV1 -0.568385667 0.00468845
MER84I -1 ERV1 -1.503690214 1.90E-224
MLT1C -1 ERVL -0.676605817 6.78E-150
MSTC -1 ERVL -0.870412821 1.37E-68
SVA_A 1 SVA 1.088204333 3.91E-100
SVA_F 1 SVA 1.513812516 6.05E-231

1.A: AE-TE predicted by Basal-like status: The 12 unique AE-TE predicted by Basal subtype Figure 1. The hierarchical
test results used minimum absolute logFC of 0.5, with BH adjusted p.value of 0.05 (+1 up-regulated, -1 down-regulated).

1.B

1.C

HER2 AETE AE-TE Hierarchical Test results biotype logFC BH.adj.P.Val
HERVK11DI 1 ERV1 1.610098903 1.10E-73
LTR14C 1 ERVK 1.477257703 7.96E-230
MER57E3 1 ERV1 1.315189522 0
MER65I 1 ERV1 2.022762067 4.73E-158
MER75B -1 PiggyBac -1.19756879 7.92E-197
MER9 -1 ERV1 -0.529157805 3.65E-58
TIGGER5_A -1 TcMar -1.338826752 9.00E-171

1.B: AE-TE predicted by HER2-enriched status: The 7 unique AE-TE predicted by

HER2 subtype Figure 1. Similar to ST.1 labeling.

LuminalA AE-TE AE-TE Hierarchical Test results biotype logFC BH.adj.P.Val

HERVFH21I 1 ERV1 1.13286221 5.71E-81
HERVL66I 1 ERV1 1.67431056 3.48E-27
LTR14 1 ERVK 0.85837478 6.38E-79

MER39B 1 ERV1 0.95128568 7.43E-67

1.C: AE-TE predicted by LuminalA status: The 4 unique AE-TE predicted by Luminal A status in
Figure 1. labeling similar to ST.1.

1.D
LuminalB AE-TE AE-TE Hierarchical Test results biotype logFC BH.adj.P.Val
BSRA -1 Satellite -0.542937 0

L1PA3 1 L1 0.52388104 0

L1PA5 1 L1 0.64952393 0

L1PA6 1 L1 0.55291379 0

MER69A -1 ERV1 -0.7782669 0

THE1A 1 ERVL 0.59462284 1.84E-37

1.D: AE-TE predicted by LuminalB status: The 6 unique AE-TE predicted by

Luminal B status in Figure 1. labeling similar to ST.1.

Supplementary Table 2

Supplementary Table 3

age_at_
diagn
HER2.e
nriche
ER+/
PR+/
HER
s
ymbol
s race os is
Bas
al.like d
Lumi
nalB
Lumin
alA 2-
ER-/ PR-/
HER2+
TN
BC
PIK3C
A AKT1
MAP
3K1 TP53 RB1
FG
FR2 ATM BRCA1
G
ATA
3
CEAC
AM1 0 0 1 0 -1 0 0 0 0 0 0 -1 0 1 1 1 1 1
LAG 3 0 0 1 0 0 -1 0 0 0 0 0 -1 1 1 1 1 0 0
CD40 0 0 0 -1 -1 -1 0 0 0 0 0 0 0 0 0 0 0 0
CD86 0 0 0 0 -1 -1 0 0 0 0 0 0 0 0 0 0 0 0
CD27
4 0 0 0 0 0 -1 0 0 0 0 1 0 0 0 1 0 0 0
CD80 0 0 0 0 0 -1 0 0 0 0 1 0 0 0 1 0 0 0
HAVC
R2 0 0 0 0 0 -1 0 0 0 0 0 0 0 0 0 0 0 0
CTLA
4 0 0 0 0 -1 -1 0 0 0 0 1 0 1 0 0 1 0 0
TIG IT 0 0 0 0 -1 -1 0 0 0 0 1 0 1 0 0 0 0 0
BTLA 0 0 -1 0 -1 -1 0 0 0 0 0 0 1 0 0 0 0 0
pathway.name log.fold.change p.Value FDR

DExD_DDX_Helicases 0.116652534 1.02E-09 6.12E-09

RNA_Interference 0.218470465 5.79E-07 1.74E-06

Post_Transcriptional -0.127272352 6.11E-06 1.20E-05

Helicases_DHX_DDX 0.083175185 8.03E-06 1.20E-05

APOBEC_Family 0.322471823 2.97E-05 3.56E-05

DExH_DHX_Helicases 0.001574146 0.95192638 0.95192638

Abstract (if available)

Abstract The Cancer Genome Atlas (TCGA) described many large-scale transcriptomes in cancer, but did not examine transposable elements. In order to further understand transposable elements in breast-cancer, we analyzed the transcriptome of 484 subjects using TCGA, and compared the repeatomes from each subtype classification. Despite Basal-like hypo-methylation, transposable elements had suppressed expression, whereas Luminal types showed increased levels.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Identifying prognostic gene mutations in colorectal cancer with random forest survival analysis

PDF

Disparities in colorectal cancer survival among Latinos in California

PDF

Racial and ethnic disparities in delays of surgical treatment for breast cancer

PDF

Mammographic density and different breast cancer subtypes defined by prognostic biomarkers

PDF

Effect of glutamate excitotoxicity on multiple sclerosis-related fatigue

PDF

Instability of heart rate and rating of perceived exertion during high-intensity interval training in breast cancer patients undergoing anthracycline chemotherapy

PDF

The effects of tobacco exposure on hormone levels and breast cancer risk among young women

PDF

Spatial modeling of non-tailpipe emissions and its association with children's lung function

PDF

Air pollution and breast cancer survival in California teachers: using address histories and individual-level data

PDF

Using multi-angle imaging spectroradiometer aerosol mixture properties and meteorology for PM₂.₅ assessment in Iran

PDF

Molecular classification of breast cancer specimens from tissue morphology

PDF

Predictive factors of breast cancer survival: a population-based study

PDF

Validation of the Children’s International Mucositis Evaluation Scale (ChIMES) in pediatric cancer and SCT

PDF

Differential methylation analysis of colon tissues

PDF

A comparison of three different sources of data in assessing the adolescent and young adults cancer survivors

PDF

Genomic, transcriptomic and immunologic landscapes of human cancers

PDF

Forecasting traffic volume using machine learning and kriging methods

PDF

Surgical aortic arch intervention at the time of extended ascending aortic replacement is associated with increased mortality

PDF

Comparisons of four commonly used methods in GWAS to detect gene-environment interactions

PDF

Identification of differentially connected gene expression subnetworks in asthma symptom

Asset Metadata

Creator Colombo, Anthony Richard (author)

Core Title Transposable element suppression in basal-like breast cancer

School Keck School of Medicine

Degree Master of Science

Degree Program Biostatistics

Publication Date 05/08/2019

Defense Date 05/08/2019

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag basal-like breast cancer,interferon response,OAI-PMH Harvest,transposable elements

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Franklin, Meredith (committee member), Gauderman, James (committee member), Mack, Wendy (committee member)

Creator Email acolombo@usc.edu,anthonycolombo60@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-169283

Unique identifier UC11660088

Identifier etd-ColomboAnt-7426.pdf (filename),usctheses-c89-169283 (legacy record id)

Legacy Identifier etd-ColomboAnt-7426.pdf

Dmrecord 169283

Document Type Thesis

Format application/pdf (imt)

Rights Colombo, Anthony Richard

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA