Onco a ge 6235
www.impac jou nals.com/onco a ge
www.impac jou nals.com/onco a ge / Onco a ge , Vol. 6, No.8
Recu en SKIL-ac i a ing ea angemen s in ETS-nega i e
p os a e cance
Ma i Annala1,2,*, Ka i Ki inummi1,2,*, Joonas Tuominen1,3, Se da Ka aku 1,3, Ki si
G anbe g1,2, Leena La onen1,3, An i Ylipää1,2, Liisa Sjöblom1,3, Pekka Ruusu uo i1,2,
Ou i Sa amäki1,3, Ki si M. Kaukoniemi1,3, Olli Yli-Ha ja2, Robe L. Vessella4, Teu o
L.J. Tammela5, Wei Zhang6, Tapio Visako pi1,3 and Ma i Nyk e 1,2
1 Ins i u e o Biosciences and Medical Technology - BioMediTech, Uni e si y o Tampe e, Tampe e, Finland
2 Ins i u e o Biosciences and Medical Technology - BioMediTech, Tampe e Uni e si y o Technology, Tampe e, Finland
3 Fimlab Labo a o ies, Tampe e Uni e si y Hospi al, Tampe e, Finland
4 Depa men o U ology, Uni e si y o Washing on, Sea le, WA, USA
5 Depa men o U ology, Tampe e Uni e si y Hospi al and Medical School, Uni e si y o Tampe e, Tampe e, Finland
6 Depa men o Pa hology, Uni e si y o Texas M.D. Ande son Cance Cen e , Hous on, TX, USA
* These au ho s con ibu ed equally o his wo k
Co espondence o: Ma i Nyk e , email: [email p o ec ed]
Co espondence o: Tapio Visako pi, email: [email p o ec ed]
Keywo ds: p os a e cance , sequencing, usion gene, SKIL
Recei ed: Oc obe 27, 2014 Accep ed: Janua y 15, 2015 Published: Janua y 31, 2015
This is an open-access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion License, which pe mi s un es ic ed use,
dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal au ho and sou ce a e c edi ed.
ABSTRACT
P os a e cance is he hi d mos common cause o male cance dea h in
de eloped coun ies, and one o he mos comp ehensi ely cha ac e ized human
cance s. Roughly 60% o p os a e cance s ha bo gene usions ha jux apose ETS-
amily ansc ip ion ac o s wi h and ogen egula ed p omo e s. A second sub ype,
cha ac e ized by SPINK1 o e exp ession, accoun s o 15% o p os a e cance s.
He e we epo he disco e y o a new p os a e cance sub ype cha ac e ized by
ea angemen s jux aposing he SMAD inhibi o SKIL wi h and ogen egula ed
p omo e s, leading o inc eased SKIL exp ession. SKIL usions we e ound in 6 o
540 (1.1%) p os a e cance s and 1 o 27 (3.7%) cell lines and xenog a s. 6 o
7 SKIL-posi i e cance s we e nega i e o ETS o e exp ession, sugges ing mu ual
exclusi i y wi h ETS usions. SKIL knockdown led o g ow h a es in PC-3 and
LNCaP cell line models o p os a e cance , and i s o e exp ession led o inc eased
in asi eness in RWPE-1 cells. The ole o SKIL as a p os a e cance oncogene lends
suppo o ecen s udies on he ole o TGF-β signaling as a a e-limi ing s ep in
p os a e cance p og ession. Ou indings highligh SKIL as an oncogene and po en ial
he apeu ic a ge in 1-2% o p os a e cance s, amoun ing o an es ima ed 10,000
cance diagnoses pe yea wo ldwide.
INTRODUCTION
P os a e cance is diagnosed in o e 900,000
men wo ldwide e e y yea , making i he second mos
common cance among men [1]. The s anda d-o -ca e
o localized p os a e cance is adical p os a ec omy o
adia ion he apy, whe eas ad anced umo s a e ea ed
wi h sys emic he apies ha inhibi and ogen signaling
[2]. Mo e speci ic d ug a ge s and d i e mu a ions ha e
been sough h ough ex ensi e genomic cha ac e iza ion
e o s [3,4]. We now know ha genomic ea angemen
plays a majo ole in he onse o p os a e cance , wi h
60% o umo s ha bo ing ch omosomal ea angemen s
ha jux apose and ogen egula ed p omo e s wi h he ETS
amily ansc ip ion ac o s ERG, ETV1, ETV4 o FLI1
[5,6]. Howe e , he p oduc s o hese usion genes ha e
p o en di icul o a ge wi h small molecule inhibi o s.
Onco a ge 6236
www.impac jou nals.com/onco a ge
Mo e ecen ly, o e exp ession o he ypsin inhibi o
SPINK1 was ound o de ine a second p os a e cance
sub ype mu ually exclusi e wi h ETS o e exp ession [7].
Monoclonal SPINK1 an ibodies ha e shown e icacy in
p eclinical models [8], sugges ing ha SPINK1 inhibi ion
may p o e a bene icial ea men s a egy in he 15% o
p os a e cance s posi i e o SPINK1 o e exp ession.
Recen s udies ha e iden i ied o he al e a ions mu ually
exclusi e wi h ETS usions, including mu a ions in he
SPOP gene [4], and dele ions o he ch oma in emodeling
gene CHD1 [9]. Despi e hese disco e ies, a signi ican
ac ion o p os a e cance s do no ha bo any o he abo e
al e a ions.
In addi ion o ETS usions and associa ed e en s,
genomic cha ac e iza ion s udies ha e iden i ied non-
synonymous mu a ions in TP53, MED12, and PTEN
[3,4,10], and g oss dele ions o he umo supp esso
genes PTEN, RB1 and TP53 [3,4,10]. Mo e ecen ly, a
numbe o s udies ha e highligh ed he ole o a enua ed
TGF-β signaling in p os a e cance p og ession [11–13].
SMAD4, a c i ical componen o he TGF-β signaling
cascade, is inac i a ed in a subse o ad anced p os a e
cance s h ough p omo e hype me hyla ion [14] o
soma ic mu a ion [10], and i s exp ession is educed in
me as a ic p os a e cance [11]. Mouse s udies ha e shown
ha TGF-β signaling inhibi s p og ession o PTEN-null
umo s, and ha SMAD4 dele ion can o e come his
inhibi ion [11]. In panc ea ic adenoca cinoma, biallelic
inac i a ion o SMAD4 is obse ed in 50% o umo s [15].
In his s udy, we pe o med ansc ip ome and low-
co e age whole genome sequencing on 28 un ea ed and
13 cas a ion esis an p os a e cance s, and iden i ied a
new p os a e cance sub ype cha ac e ized by ac i a ing
ea angemen s o he SMAD inhibi o SKIL.
RESULTS
Sample acquisi ion and sequencing
F esh- ozen issue om 12 benign p os a ic
hype plasias (BPH), 28 un ea ed p os a e cance s (PC),
and 13 cas a ion esis an p os a e cance s (CRPC)
was acqui ed om he Tampe e Uni e si y Hospi al
(Tampe e, Finland). All samples con ained a minimum
o 70% cance ous o hype plas ic epi helial cells. PC
samples we e ob ained by adical p os a ec omy and
locally ecu en CRPCs by ansu e h al esec ion o he
p os a e (Supplemen a y Table 1). Mean age a diagnosis
was 60.8 yea s ( ange: 47.4-71.8) and mean PSA a
diagnosis was 10.8 ng/ml ( ange: 3.5-48.1). Lib a ies we e
p epa ed o pai ed-end analysis on he Illumina HiSeq
2000. On a e age, we ob ained 150 million pai ed end
eads pe sample om he low co e age whole genome
sequencing, and 110 million pai ed end eads om he
whole ansc ip ome sequencing (Supplemen a y Table 2).
Disco e y o ecu en SKIL-ac i a ing
ea angemen s
Fusion e en s in ol ing an ETS amily ansc ip ion
ac o o SPINK1 o e exp ession we e iden i ied in 32 o
41 umo s based on ansc ip ome sequencing (Figu e
1a, Supplemen a y Table 3). No SPOP mu a ions we e
iden i ied in ou coho . We iden i ied a no el TMPRSS2-
SKIL usion gene in one CRPC sample (Figu e 1A-
B, Supplemen a y Figu e 1) and alida ed i using
Sange sequencing (Figu e 1B) and luo escence in si u
hyb idiza ion (Figu e 1C). The usion me ged he i s
h ee exons o TMPRSS2 wi h ull leng h SKIL and led
o SKIL o e exp ession due o he and ogen egula ed
TMPRSS2 p omo e (Figu e 1D). SKIL encodes a SKI-
like p o ein ha inhibi s TGF-β signaling by binding o
and dis up ing he he e ome ic SMAD complex [16]. To
sea ch o mo e posi i e cases, we sc eened 76 addi ional
umo s (Supplemen a y Table 1) and 22 LuCaP xenog a s
wi h qRT-PCR, and iden i ied SKIL o e exp ession in
one xenog a and one clinical sample. T ansc ip ome
sequencing o hese samples e ealed a SLC45A3-SKIL
usion in LuCaP-77 and a MIPEP-SKIL usion in he
clinical sample, con i ming SKIL as a ecu en 3’ usion
pa ne in p os a e cance (Figu e 1E-F). Analysis o
ansc ip ome sequencing da a om he Cance Genome
A las (TCGA) p os a e adenoca cinoma p ojec e ealed
addi ional SKIL-ac i a ing ea angemen s in 4 o 423
samples, wi h concomi an SKIL o e exp ession (Figu e
2A-B). In he Taylo e al. da ase , wo ETS-nega i e
samples (PCA0015 and PCA0056) exhibi ed ou lie
o e exp ession o SKIL, bu we e excluded om u he
analysis due o lack o sequencing da a [3]. In e es ingly,
all 5 SKIL-posi i e clinical samples wi h clinical
in o ma ion ( wo TCGA samples lacked clinical da a)
con ained a Gleason g ade 5 componen o ep esen ed
me as a ic p os a e cance , sugges ing ha SKIL-ac i a ing
al e a ions may associa e wi h high-g ade p os a e cance .
None o he se en SKIL ea angemen s dis up ed
he p o ein coding sequence o SKIL, sugges ing ha ull-
leng h SKIL p o ein is necessa y o oncogenic unc ion. 6
o 7 SKIL ea angemen s in ol ed an and ogen egula ed
p omo e , indica ing selec ion owa ds jux aposi ions wi h
highly ac i e p omo e s. 6 o 7 usion posi i e samples
we e nega i e o ETS o e exp ession, sugges ing mu ual
exclusi i y be ween SKIL and ETS ea angemen s (p
= 0.047, Fishe ’s exac es ). Sample TCGA-YL-A8SJ
o e exp essed bo h SKIL and ETV1 (Figu e 2A), al hough
we ound no eads suppo ing an ETV1 ea angemen in
ei he he ansc ip ome o exome sequencing da a o his
sample.
In sample TCGA-HC-7211, he ea angemen
be ween ACPP and SKIL had an unexpec ed s uc u e,
Onco a ge 6237
www.impac jou nals.com/onco a ge
Figu e 1: Recu en SKIL-ac i a ing ea angemen s in p os a e cance . (A) Ma ix showing mu ually exclusi e
o e exp ession o ERG, ETV1, ETV4, SPINK1, and SKIL in a ansc ip ome sequencing coho o 41 p os a e cance s. Red ec angles
indica e o e exp ession, and black inne ec angles indica e usion e en s. (B) S uc u e o he TMPRSS2-SKIL usion gene iden i ied in
sample CRPC_348. Black lines indica e exon-exon junc ions wi h ansc ip ome sequencing e idence. Fusion ansc ip was alida ed
wi h Sange sequencing om cDNA. (C) Fluo escence in si u hyb idiza ion alida es he usion a genomic le el. One example o a usion
posi i e cell is shown. (D) SKIL exp ession in he ansc ip ome sequencing coho o 41 p os a e cance s and 12 BPHs. SKIL is s ongly
o e exp essed in he TMPRSS2-SKIL posi i e sample. (E) SKIL exp ession was measu ed using qRT-PCR in a alida ion coho o 76
p os a ec omy samples. Sample PC_11423 exhibi ed SKIL o e exp ession and was ound o con ain a MIPEP-SKIL ea angemen by
ansc ip ome sequencing. (F) SKIL exp ession was measu ed using qRT-PCR in LuCaP xenog a s and cell line models o p os a e cance .
Xenog a LuCaP-77 was ound o con ain an SLC45A3-SKIL ea angemen by ansc ip ome sequencing.
Onco a ge 6238
www.impac jou nals.com/onco a ge
wi h he i s exon and p omo e o ACPP placed
downs eam o SKIL in an isense o ien a ion (Figu e 2B).
Despi e he non-canonical s uc u e, he ea angemen
led o s ong o e exp ession o ull-leng h SKIL (Figu e
2A), possibly due o ch oma in emodeling induced by
he and ogen egula ed ACPP p omo e . The an isense
p omo e p o oked exp ession o a spliced an isense
ansc ip composed o c yp ic exons loca ed in SKIL
in ons (Supplemen a y Figu e 2). Exp ession o he
an isense ansc ip did no appea o in e e e wi h SKIL
splicing, as sense ansc ip s we e no mally spliced.
Le el o nuclea SKIL p o ein is ele a ed in a
usion posi i e sample
To de e mine whe he usion posi i e clinical
samples also o e exp essed SKIL a he p o ein
le el, we used a monoclonal an ibody o pe o m
immunohis ochemis y on he usion posi i e TURP
sample CRPC_348 and 8 nega i e con ols. The TURP
sample exhibi ed s ong nuclea and modes cy oplasmic
s aining o SKIL, while nega i e con ols showed no
s aining o only weak cy oplasmic s aining o SKIL
(Figu e 3A). We also showed he o e exp ession o
SKIL a he RNA le el in CRPC_348 using RNA in si u
hyb idiza ion (Figu e 3B).
Figu e 2: SKIL-ac i a ing ea angemen s in he TCGA p os a e adenoca cinoma sequencing coho . (A) Ba plo showing
exp ession o ERG, ETV1, ETV4, SPINK1 and SKIL in TCGA samples. Fou samples exhibi ed signi ican SKIL o e exp ession and we e
ound o ha bo SKIL-ac i a ing ea angemen s. (B) S uc u es o he ACPP-SKIL, SLC45A3-SKIL, MIPOL1-SKIL and HMGN2P46-SKIL
ea angemen s. Black lines indica e exon-exon junc ions wi h ansc ip ome sequencing e idence.
Onco a ge 6239
www.impac jou nals.com/onco a ge
SKIL egula es p oli e a ion and in asi eness o
p os a e cance cells
To be e unde s and he biological ole o SKIL in
p os a e cance cells, we knocked SKIL down in PC-3
cells using wo di e en siRNA (Figu e 4A), and obse ed
educed cell g ow h (Figu e 4B), in asi eness (Figu e 4c),
and colony o ma ion (Figu e 4D) ela i e o sc ambled
siRNA. The e ec on cell g ow h was eplica ed in LNCaP
cells (Figu e 4E-F). To show ha SKIL exp ession is uly
and ogen dependen in cells whe e SKIL is used wi h
and ogen egula ed p omo e s, we ex ac ed LuCaP-77
xenog a issue (wi h an SLC45A3-SKIL usion) om
cas a e and non-cas a e mice 1, 3 and 7 days pos -
cas a ion. We hen used qRT-PCR o quan i y SKIL and
KLK3 (PSA) exp ession a each imepoin . Exp ession
alues we e no malized agains TATA-box binding p o ein
(TBP) and compa ed agains non-cas a e con ol mice.
We obse ed a s ong educ ion in he exp ession o bo h
KLK3 and SKIL on day 7, indica ing and ogen egula ed
exp ession (Figu e 4G).
Nex , we c ea ed a SKIL o e exp ession model by
ans ec ing immo alized p os a e epi helial cells (RWPE-
1) wi h a pCI-Neo ec o exp essing hemagglu inin (HA)
agged SKIL [17]. Con ol RWPE-1 cells we e ans ec ed
wi h an emp y pCI-Neo ec o (EV). In compa ison o
con ol cells, SKIL- ans ec ed RWPE-1 cells exhibi ed
highe exp ession o SKIL a bo h RNA and p o ein le els
(Figu e 5A-B), and g ea e in asi e po en ial in a ma igel
in asion assay (n = 3, p = 0.044, unpai ed wo- ailed - es )
(Figu e 5C). SKIL o e exp ession had no e ec on he
g ow h o RWPE-1 cells (da a no shown).
O he genomic al e a ions in he sequencing
coho
A p e ious s udy has shown ha combined
dele ion o SMAD4 and PTEN in mouse p os a es leads
o agg essi e p os a e cance wi h 100% pene ance [11].
The e o e, we se ou o check whe he SKIL- ea anged
cance s ha bo ed concomi an PTEN dele ions. PTEN
exp ession was no abe an ly low in any o he 7 SKIL-
posi i e cases, and only 1 o 4 SKIL-posi i e cance s in
he TCGA coho showed e idence o PTEN dele ion,
sugges ing ha SKIL ea angemen s do no equi e
combined PTEN loss. Genes associa ed wi h ERG
o e exp ession, such as COL2A1, ALOX15, CRISP3,
B3GNT6 and TDRD1, we e no o e exp essed in SKIL-
ea anged umo s.
Figu e 3: Immunohis ochemis y and RNA in si u hyb idiza ion o SKIL in a SKIL- ea anged umo . (A) An i-SKIL
s aining o pa a in-embedded sec ions om he SKIL- ea anged TURP sample CRPC_348 and wo ep esen a i e p os a ec omies
nega i e o SKIL ea angemen showing no s aining o weak cy oplasmic s aining. (B) RNA in si u hyb idiza ion o CRPC_348 and wo
ep esen a i e p os a ec omies nega i e o SKIL ea angemen , wi h p obes a ge ing SKIL mRNA. Nuclei we e s ained wi h hema oxylin.

Onco a ge 6240
www.impac jou nals.com/onco a ge
Exp ession analysis o he TGF-β and BMP
pa hways ( ha bo h con e ge on SMAD4) e ealed
signi ican ly educed exp ession o TGF-β and BMP
ligands in bo h un ea ed and cas a ion esis an p os a e
cance (Figu e 6, Supplemen a y Figu e 3). O he
genomic al e a ions in SKIL-posi i e umo s included
TP53 mu a ion in CRPC_348, hemizygous TP53 dele ion
in TCGA-KK-A8IL and TCGA-YL-A8SJ, hemizygous
PTEN dele ion in TCGA-YL-A8SJ, hemizygous NKX3-
1 dele ion in TCGA-HC-7211 and TCGA-YL-A8SJ, and
an MLL3 ameshi dele ion mu a ion in CRPC_348
(Supplemen a y Table 4).
In addi ion o SKIL ea angemen s, we ound
a ious o he genomic al e a ions in ou sequencing
coho (Figu e 7, Supplemen a y Figu e 4, Supplemen a y
Table 5). Since he o iginal sequencing did no include
pai ed no mal issues, we used a ge ed sequencing
on pai ed blood samples o il e ou ge mline a ian s.
The umo supp esso TP53 was nonsynonymously
mu a ed in 2 PCs and 5 CRPCs, wi h addi ional loss-
o -he e ozygosi y in 3 CRPCs. PTEN was dis up ed by
a s opgain mu a ion in one CRPC sample and dele ed
in eigh o he umo s. The AR nega i e umo PC_6864
ca ied a KRAS p.G12R mu a ion ha is known o cause
cons i u i e KRAS ac i a ion in cance s o he colon,
panc eas and lungs, bu is less common in p os a e cance
[18]. Sample CRPC_489 ha bo ed wo dis inc AKT1
mu a ions, one o which (p.E17K) has been associa ed
wi h dys egula ed issue g ow h in he P o eus synd ome
[19]. We also iden i ied soma ic mu a ions ha al e ed
he o khead domain and he N- e minal ansac i a ion
domain o he AR co ac o FOXA1. Fou samples we e
posi i e o he HOXB13 p.G84E ge mline a ian ha has
been associa ed wi h p os a e cance suscep ibili y [20],
including one homozygous sample. One AR-nega i e
CRPC sample had acqui ed a DOT1L-HES6 usion.
We and o he s ha e shown ha HES6 o e exp ession is
su icien o induce comple ely and ogen independen
g ow h in p os a e cance cells [21,22]. In bo h PC and
CRPC, we iden i ied equen al e a ions in ch oma in
modi ie s including nonsynonymous mu a ions in CHD4,
MLL3, HDAC5, KDM5B and MBD6, and a homozygous
dele ion o KDM6A in one sample (Figu e 7).
DISCUSSION
SKIL (also known as SnoN) is a 684 amino acid
Figu e 4: In i o knockdown expe imen s on SKIL. (A) SKIL exp ession was silenced in PC3 cells using wo siRNAs, esul ing in
(B) educed g ow h (n = 4), (C) in asion (n = 4) and (D) colony o ma ion (n = 2). (E) SKIL exp ession was silenced in LNCaP cells using
wo siRNAs, esul ing in (F) educed g ow h (n = 4). (G) qRT-PCR ime se ies o SKIL and PSA exp ession in cas a e and non-cas a e
mice ca ying LuCaP-77 xenog a s (n = 2). E o ba s, s.e.m. wi h i s -o de e o p opaga ion; *P<0.05; **P<0.01; ***P<0.001, unpai ed
wo- ailed - es .
Onco a ge 6241
www.impac jou nals.com/onco a ge
Figu e 6: Genomic and ansc ip omic changes in he con ex o he TGF-β signaling pa hway. Genes a e shown as boxes
wi h wo hal es: he le hal shows he pe cen age o un ea ed p os a e cance s wi h wo- old up egula ion ( ed) o down egula ion (blue)
ela i e o BPH, and he igh hal shows he same o cas a ion esis an p os a e cance s. A ows indica e in e ac ions be ween p o eins
o genes, he in e ac ion ype is w i en nex o he a ow.
Figu e 5: In i o o e exp ession expe imen s on SKIL. (A) An i-SKIL wes e n blo showing inc eased SKIL p o ein in SKIL-
ans ec ed RWPE-1 cells. MCF-7 cells ans ec ed wi h SKIL o sc ambled siRNA a e used o alida e he band. (B) qRT-PCR quan i ica ion
o SKIL in RWPE-1 cells ans ec ed wi h SKIL o emp y ec o . (C) Ma igen in asion assay on RWPE-1 cells ans ec ed wi h SKIL
o emp y ec o (n = 3). E o ba s, s.e.m. wi h i s -o de e o p opaga ion; *P<0.05; **P<0.01; ***P<0.001, unpai ed wo- ailed - es .
Onco a ge 6242
www.impac jou nals.com/onco a ge
nuclea p o ein ha is ubiqui ously exp essed in human
issues [23,24] and sha es i s domain s uc u e wi h
SKI, a p o ein ha was o iginally disco e ed h ough
i s simila i y wi h he ans o ming componen o he
Sloan-Ke e ing Vi us [25]. O e exp ession o ei he
SKI o SKIL in chicken emb yo ib oblas s is su icien
o induce oncogenic ans o ma ion [26], and SKIL
exp ession is ele a ed in many human cance s, including
cance s o he skin, b eas , colon and blood [27]. The
3q26 locus is ampli ied in se e al cance ypes, and
SKIL (along wi h TLOC1) has been highligh ed as he
mos po en oncogene in his egion [28]. Bo h SKI and
SKIL con ain an 80 aa SAND-like domain ha can bind
wi h he MH2 domain o SMAD4 [29]. SMAD4 (co-
Smad) is an i eplaceable pa o he he e ome ic SMAD
complexes ha ac as downs eam media o s o TGF-β
signaling. These he e ome ic complexes a e o med when
SMAD4 binds wi h one o mo e ecep o SMADs such as
SMAD2 and SMAD3 [29]. The complex hen ansloca es
o he nucleus and ac i a es ansc ip ion o TGF-β
esponsi e genes. Binding o SKI/SKIL wi h he SMAD4
MH2 domain inhibi s his ansc ip ional ac i a ion
by p e en ing SMAD complex o ma ion [29] o by
ec ui men o he nuclea co- ep esso NCOR1 [16]. In
addi ion o i s binding wi h SMAD4, SKIL can also bind
wi h he MH2 domains ound in ecep o SMADs h ough
a domain loca ed close o i s N- e minal [16]. Nei he SKI
no SKIL has been shown o di ec ly bind DNA, despi e
bo h p o eins con aining a Dachshund homology domain
ha sha es ea u es wi h he o khead/winged-helix amily
o DNA binding p o eins [27].
In addi ion o he inhibi o y ole o SKIL on TGF-β
signaling, a ecen s udy has p oposed ha SKIL may play
a ole in egula ing epi helial- o-mesenchymal ansi ion
(EMT) by inducing exp ession o SNAI2 (SLUG), a
mas e egula o o EMT [28]. The same s udy also ound
ha SMAD4 knockdown inc eased he in asi eness
o human mamma y epi helial cells, while SKIL
o e exp ession had no e ec on cell g ow h, in ag eemen
wi h ou indings [28]. Ano he ecen s udy has p oposed
ha SKIL can in e ac wi h and p omo e he ac i i y o
es ogen ecep o α in he nuclei o b eas ca cinoma cells.
The in e ac ion occu s ia wo highly conse ed nuclea
ecep o binding LxxLL-like mo i s in SKIL [30]. This
Figu e 7: Cha ac e iza ion o genomic and ansc ip omic changes in ou p os a e cance sequencing coho . Exp ession
and copy numbe changes a e shown in blue and ed. Poin mu a ions and indels a ec ing p o ein coding sequences a e shown in g een.
Whi e squa es indica e missing da a.
Onco a ge 6243
www.impac jou nals.com/onco a ge
inding is in iguing as i sugges s a po en ial in e ac ion
be ween SKIL and and ogen ecep o , as some LxxLL
mo i s can bind wi h he ligand binding domain o AR
[31].
The SKIL ea angemen s epo ed in his pape
occu ed in bo h un ea ed and cas a ion esis an p os a e
cance s, and in ol ed he 5’ pa ne genes TMPRSS2,
SLC45A3, MIPOL1, ACPP, MIPEP and HMGN2P46. The
and ogen egula ed genes TMPRSS2 and SLC45A3 a e
he mos common 5’ pa ne s in ol ed in ETS usions in
p os a e cance [6]. MIPOL1 is ano he and ogen egula ed
gene ha is in ol ed in MIPOL1-ETS ea angemen s in
p os a e cance s [32]. Exp ession o ACPP is and ogen
egula ed and highly p os a e speci ic among no mal
issue ypes [33]. MIPEP is no gene ally conside ed
and ogen egula ed o highly exp essed in he p os a e, bu
we obse ed high exp ession o bo h MIPEP and SKIL
in he MIPEP-SKIL posi i e sample. HMGN2P46 is a
pseudogene ha is s ongly exp essed in AR-exp essing
p os a e cance s bu no exp essed in AR-nega i e cance s
in ou coho , sugges ing an and ogen egula ed p omo e .
We conclude ha six SKIL ea angemen s in ol ed
an and ogen egula ed p omo e , and one in ol ed an
o he wise highly ac i e p omo e .
The disco e y o SKIL-ac i a ing ea angemen s
in bo h un ea ed and cas a ion esis an p os a e cance s
sugges s ha SKIL ea angemen s may ep esen an
ea ly e en in p os a e umo igenesis. This hypo hesis is
suppo ed by he ac ha SKIL ea angemen s appea o
be mu ually exclusi e wi h ETS usions, which ep esen
a known ea ly e en in p os a e cance p og ession [34].
The complex s uc u e o wo ea angemen s highligh s
he ac ha ea angemen s can a ec gene exp ession
in a clinically signi ican manne wi hou dis up ing
he ansc ibed po ion o a gene. As an example o
his complexi y, one SKIL-ac i a ing ea angemen in
ou s udy jux aposed an ac i e p omo e o a posi ion
downs eam o SKIL in opposi e o ien a ion, and s ill
esul ed in s ongly ele a ed exp ession o ull leng h SKIL
ansc ip .
Based on he coho s s udied in his manusc ip , we
es ima e ha SKIL-ac i a ing ea angemen s a e ound
in 1-2% o diagnosed p os a e cance s. Due o he high
incidence o p os a e cance , his ac ion ansla es o
an es ima ed 10,000 diagnoses and 3,000 dea hs caused
by SKIL-posi i e p os a e cance s pe yea wo ldwide
[1]. Whe he SKIL-posi i e p os a e cance s di e in
hei clinical cou se om o he p os a e cance s emains
o be e alua ed in a la ge s udy. Since 6 o 7 SKIL
ea angemen s in ol ed and ogen egula ed p omo e s,
we expec ha exis ing ea men modali ies based on
and ogen abla ion will be e ec i e a ea ing SKIL-
ea anged cance s. None heless, SKIL p o ides an
in iguing new molecula a ge o pe sonalized he apy,
and highligh s he ole o TGF-β signaling in p os a e
cance p og ession.
MATERIALS AND METHODS
Sequencing coho
F esh- ozen issue specimens om 12 benign
p os a e hype plasias, 28 un ea ed p os a e cance s, and
13 cas a ion esis an p os a e cance s we e acqui ed
om Tampe e Uni e si y Hospi al (Tampe e, Finland).
Un ea ed p os a e cance samples we e ob ained by
adical p os a ec omy and locally ecu en CRPCs by
ansu e h al esec ion o he p os a e. Samples we e
snap- ozen and s o ed in liquid ni ogen. His ological
e alua ion and Gleason g ading we e pe o med by a
pa hologis based on hema oxylin/eosin-s ained slides.
All samples con ained a minimum o 70% cance ous
o hype plas ic cells. The use o clinical ma e ial was
app o ed by he e hical commi ee o he Tampe e
Uni e si y Hospi al and he Na ional Au ho i y o
Medicolegal A ai s. W i en in o med consen was
ob ained om he subjec s.
Valida ion coho
76 addi ional ho monally un ea ed PC
p os a ec omy samples we e acqui ed om he Tampe e
Uni e si y Hospi al (Tampe e, Finland). Samples we e
snap- ozen and s o ed in liquid ni ogen. His ological
e alua ion and Gleason g ading we e pe o med by a
pa hologis based on hema oxylin/eosin-s ained slides.
Samples we e con i med o con ain a minimum o 70%
cance ous o hype plas ic cells by hema oxylin-eosin
s aining. Mean age a diagnosis was 62.1 yea s ( ange:
47.4-71.8), mean PSA a diagnosis was 11.8 ( ange: 3.15-
51.5). The use o clinical ma e ial was app o ed by he
e hical commi ee o he Tampe e Uni e si y Hospi al and
he Na ional Au ho i y o Medicolegal A ai s. W i en
in o med consen was ob ained om he subjec s.
Cell lines and xenog a s
P os a e cance cell lines PC-3, LNCaP, DU145,
22R 1 and immo alized p os a e epi helial cell line
RWPE-1 we e ob ained om Ame ican Type Cell
Collec ion (Manassas, VA, USA). LAPC-4 cell line was
kindly p o ided by D . Cha les Sawye s (Uni e si y o
Cali o nia a Los Angeles, Los Angeles, CA, USA), VCaP
and DuCaP by D . Jack Schalken (Radboud Uni e si y
Nijmegen Medical Cen e , Nijmegen, he Ne he lands),
and EP156T by D . Va da Ro e (Weizmann Ins i u e
o Science, Reho o , Is ael). All cell lines we e cul u ed
unde ecommended condi ions. 22 p e iously es ablished
LuCaP-se ies xenog a s we e p o ided by R.L.V. These
xenog a s ha e been de i ed om p ima y and me as a ic
Onco a ge 6250
www.impac jou nals.com/onco a ge
38. Kim D, Pe ea G, T apnell C, Pimen el H, Kelley R,
Salzbe g S. TopHa 2: accu a e alignmen o ansc ip omes
in he p esence o inse ions, dele ions and gene usions.
Genome Biol. 2013;14(4):R36.
39. Venka achalan SP, Bushman JD, Me cado JL, Sanca F,
Ch is ophe son KR, Boileau AJ. Op imized exp ession
ec o o ion channel s udies in Xenopus oocy es and
mammalian cells using al al a mosaic i us. P luge s A ch.
2007;454(1):155–63.
40. Langmead B, Salzbe g SL. Fas gapped- ead alignmen wi h
Bow ie 2. Na . Me hods 2012;9(4):357–9.
41. The 1000 Genomes P ojec Conso ium. A map o human
genome a ia ion om popula ion-scale sequencing. Na u e
2010;467(7319):1061–73.
42. Wang K, Li M, Hakona son H. ANNOVAR: unc ional
anno a ion o gene ic a ian s om high- h oughpu
sequencing da a. Nucleic Acids Res. 2010;38(16):e164.
43. Bam o d S, Dawson E, Fo bes S, Clemen s J, Pe e R,
Dogan A, Flanagan A, Teague J, Fu eal PA, S a on
MR, Woos e R. The COSMIC (Ca alogue o Soma ic
Mu a ions in Cance ) da abase and websi e. B . J. Cance
2004;91(2):355–8.