Onco a ge 17430
www.impac jou nals.com/onco a ge
www.impac jou nals.com/onco a ge / Onco a ge , Vol. 6, No. 19
C/D-box snoRNA-de i ed RNA p oduc ion is associa ed wi h
malignan ans o ma ion and me as a ic p og ession in
p os a e cance
Elena S. Ma ens-Uzuno a1, You i Hoogs a e1, An on Kalsbeek1, Bas Pigmans1,
Mi ella V edenb eg - an den Be g1, Na asja Di s1, Sø en Jensby Nielsen2,3, Adam
Bake 2,4, Tapio Visako pi5, Ch is Bangma1 and Guido Jens e 1
1 Depa men o U ology, E asmus MC, Ro e dam, The Ne he lands
2 Exiqon A/S, Vedbaek, Denma k
3 Nue olu ion A/S, Copenhagen, Denma k
4 Ch . Hansen A/S, Hø sholm, Denma k
5 Ins i u e o Biosciences and Medical Technology - BioMediTech, Uni e si y o Tampe e and Tampe e Uni e si y Hospi al,
Tampe e, Finland
Co espondence o: Elena Ma ens-Uzuno a, email: [email p o ec ed]
Keywo ds: snoRNA, sdRNA, SNORD78, GAS5, p os a e cance
Recei ed: Feb ua y 05, 2015 Accep ed: May 02, 2015 Published: May 19, 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
Small nucleola RNAs (snoRNAs) a e dynamically egula ed in di e en issues
and a ec ed in disease. SnoRNAs a e p ocessed u he o s able smalle RNAs.
We sequenced he small RNA ansc ip ome o p os a e cance (PCa) a di e en
PCa s ages and gene a ed a quan i ied ca alogue o 3927 small non-coding RNAs
(sncRNAs) de ec ed in no mal and malignan p os a e issue. F om hese, only 1524
a e mic oRNAs. The emaining 2401 sncRNAs ep esen s able sncRNAs species ha
o igina e om snoRNA, RNA and o he sncRNAs. We show ha snoRNA-de i ed RNAs
(sdRNAs) display s onge di e en ial exp ession han mic oRNAs and a e massi ely
up egula ed in PCa. SdRNAs accoun o a leas one hi d o all small RNAs wi h
exp ession changes in umo compa ed o no mal adjacen issue. Mul iple sdRNAs can
be p oduced om one snoRNA in a manne ela ed o he conse a ion o s uc u al
snoRNA mo i s. Q-PCR analysis in an independen pa ien coho (n=106) con i med
he p ocessing pa e ns o selec ed snoRNAs (SNORD44, SNORD78, SNORD74 and
SNORD81) and he cance -associa ed up- egula ion o hei sdRNAs obse ed in
sequencing da a. Impo an ly, exp ession o SNORD78 and i s sdRNA is signi ican ly
highe in a subse o pa ien s ha de eloped me as a ic disease demons a ing ha
snoRNA and sdRNAs may p esen as no el diagnos ic and/o p ognos ic bioma ke s
o PCa.
INTRODUCTION
Malignan ans o ma ion and cance p og ession
cause changes in he exp ession and unc ion o
mic oRNAs (miRNAs) [1, 2]. Howe e , he e ec s
o hese p ocesses on o he small non-coding RNAs
(sncRNAs) a e less unde s ood. Recen ly, we
demons a ed he abundance and di e en ial exp ession
o small nucleola RNA-de i ed RNAs (sdRNAs) in he
small ansc ip ome o p os a e cance (PCa) [3]. I is
gene ally accep ed ha small nucleola RNAs (snoRNAs)
a e housekeeping, p o ein-noncoding molecules ha
associa e wi h speci ic se s o p o eins o main ain p ope
ibosomal ma u a ion in he nucleolus.
S ill, se e al epo s show ha snoRNAs ha e
issue-speci ic exp ession [4, 5], and may p esen as
no el cance bioma ke s. Fo example, he H/ACA-
box snoRNA SNORA42 is commonly o e exp essed in
non-small cell lung cance (NSCLC) and i s exp ession
is signi ican ly in e sely co ela ed wi h su i al [6, 7].
Onco a ge 17431
www.impac jou nals.com/onco a ge
Simila ly, he le els o C/D-box snoRNAs SNORD33,
SNORD66 and SNORD76 a e signi ican ly ele a ed in
plasma om NSCLC pa ien s compa ed wi h cance - ee
con ols and can p o ide po en ial bioma ke s o ea ly
de ec ion [8]. In ch onic lymphocy ic leukemia (CLL),
he e ogeneous snoRNA exp ession pa e ns disc imina e
majo CLL subg oups and can s a i y pa ien s in di e en
p ognos ic g oups [9], while in mul iple myeloma snoRNA
exp ession pa e ns a e associa ed wi h dis inc molecula
sub ypes o he disease [10].
Fu he mo e, esen esea ch demons a es ha he
molecula al e a ions o snoRNA a e unc ionally linked o
basic cellula p ocesses associa ed wi h cance p oposing
ei he umo supp esso o oncogene ole o di e en
snoRNAs. In NSCLC, SNORA42 ac s as a pu a i e
oncogene. I s o e exp ession enhances cell p oli e a ion
and g ow h in b onchial epi helium and cance cells, while
i s knockdown in NSCLC cells inhibi s colony o ming
[7]. In acu e p omyelocy ic leukemia, he SNORD112–114
is speci ically ac i a ed in a subse o pa ien s and may
in luence cell g ow h h ough a nega i e egula ion o
he cell cycle and he Rb pa hway [11]. On he con a y,
in pe iphe al T-cell lymphoma, o e -exp ession o he
candida e p ognos ic ma ke SNORD71 (HBII-239) is
associa ed wi h a o able ou come [12]. The C/D box
snoRNA SNORD50, a ansloca ion pa ne o BCL6 in
B-cell lymphoma [13], is a candida e umo supp esso
signi ican ly associa ed wi h clinically ele an p os a e
[14] and b eas [15] cance . In hepa ocellula ca cinoma
(HCC), SNORD113-1 has been iden i ied as a umo
supp esso [16]. Down- egula ion o his snoRNA is
associa ed wi h dec eased su i al o HCC pa ien s, while
econs i u ion o i s exp ession supp esses umo igenesis
in i o and in i o. In glioblas oma, dec eased exp ession
o he GAS5 encoded SNORD76 is associa ed wi h an
agg essi e pheno ype [17]. Ec opic exp ession o his
umo supp esso snoRNA inhibi s umo igenici y by
a es ing cance cells in S phase in i o and inhibi s
o ho opic umo g ow h in i o. In b eas cance and head
and neck squamous cell ca cinoma he low exp ession o
ano he GAS5 encoded snoRNA, SNORD44, co ela es
wi h ma ke s o agg essi e pa hology and poo p ognosis
[18, 19].
A p esen , li le is known abou he pa hways o
snoRNA u no e . Appa en ly, snoRNAs a e u he
p ocessed o sdRNAs in a as a ie y o o ganisms [20].
I is ye unclea whe he sdRNAs a e no el unc ional
en i ies o oo p in -p oduc s o snoRNA downs eam
p ocessing shielded om deg ada ion by snoRNA-
in e ac ing p o eins. A miRNA-like ac i i y has been
p oposed o sdRNAs de i ed om H/ACA-box snoRNAs
(H/ACA-sdRNAs) based on hei appa en size o 20-24 n
equi alen o miRNAs, he abili y o p omo e ep ession
o complemen a y a ge s in i o, and he associa ion wi h
Dice and AGO complexes [21-27]. In con as , a bimodal
size dis ibu ion o 17-20 n and 27-30 n has been epo ed
o sdRNAs de i ed om C/D-box snoRNAs (C/D-
sdRNAs) [3, 22, 23]. C/D-sdRNAs a e no e icien ly
inco po a ed in AGO2 sugges ing a di e en unc ion o
his ype o sdRNAs [28]. In addi ion, i has been epo ed
ha he highly abundan in b ain ‘o phan’ snoRNAs,
SNORD115 and SNORD116, a e p ocessed in o la ge
sdRNAs (34-73 n ) ha complex wi h spliceosomal
p o eins and may egula e he al e na i e splicing o a ge
mRNAs [29, 30]. Associa ion o C/D-box snoRNAs wi h
no el RNPs and in ol emen in al e na i e splicing has
been p e iously obse ed in mice o he b ain speci ic
MbII-52 [31]. In e es ingly, bo h MbII-52 and i s human
o holog SNORD115 p oduce la ge sdRNAs (34-73
n ). Simila obse a ion has also been made o sdRNA
egions o SNORD88C, which can in luence he al e na i e
splicing o FGFR3 p e-mRNA [32]. A he same ime,
s udies in D osophila sp. and in human cells show ha
snoRNAs a e s ongly en iched in he nuclea ac ions o
ch oma in-associa ed RNA and possibly in ol ed in he
main enance o open ch oma in s uc u e [33].
He e, we epo he deep sequencing o pa ien -
de i ed samples om no mal p os a e, and PCa in
di e en disease s ages, which e eals sdRNA p oduc ion
om he as majo i y o known human snoRNAs. A leas
78 o he de ec ed sdRNAs demons a e s ong di e en ial
exp ession in cance . Fu he mo e, he exp ession o some
sdRNAs and hei p ecu so s is associa ed wi h clinical
p og ession and me as a ic occu ence.
RESULTS
Lib a y p epa a ion and sequencing
We gene a ed 10 sncRNA lib a ies om no mal
adjacen p os a e (NAP), benign p os a e hype plasia
(BPH), di e en s ages o PCa, and me as a ic lymph
node (LN) p epa ed om esh- ozen pa ien ma e ial
(FF) (Supplemen a y Table 1). To es ima e he in luence
o sample s o age on sncRNA abundance and s abili y, we
p epa ed a eplica e lib a y om o malin- ixed, pa a in-
embedded issue (FFPE) om umo samples used o
one o he esh- ozen lib a ies (g oup 3). All sequencing
eac ions yielded app oxima ely 14 million aw eads
each (13,468,284 o 15,393,670) wi h he FFPE lib a y
p oducing he highes aw ead numbe (Figu e 1a).
Anno a ion o he sncRNA ansc ip ome
The co ec mapping o sncRNA eads is challenged
by he ac ha p edominan iso o ms o miRNAs
and o he sncRNAs such as snoRNAs may a y om
he ma u e sequences anno a ed in public da abases.
Di e ences can be caused by al e na i e 3’-end
modi ica ions [34] o al e na i e 5’-/3’-end posi ions o
Onco a ge 17432
www.impac jou nals.com/onco a ge
he de ec ed sncRNA. Addi ionally, he leng h o ma u e
sncRNA ansc ip s can be ambiguously anno a ed in
di e en public da abases. To map as many sequence
eads as possible, we cons uc ed a cus om small non-
coding RNA da abase (sncRNAdb) ha consis s o 2271
unique small non-coding RNA species co esponding o
2356 unique genomic loci (Supplemen a y Figu e 1 and
Supplemen a y File 1).
Mapping o sncRNAdb esul ed in he de ec ion o
a o al o 1637 unique sncRNAs exp essed ac oss any o
he 11 lib a ies wi h an a e age o 1229 pe lib a y. 70%
o 84% o he eads gene a ed om esh- ozen samples
and only 52% o he eads gene a ed om FFPE could be
anno a ed by sncRNAdb (Figu e 1a and Supplemen a y
Table 2). The majo i y o anno a ed eads mapped o 873
p e-miRNAs (85.5 - 95.6%), 385 RNAs (1.89 - 7.4%),
228 C/D-box snoRNAs (0.3 - 1.9%), and 91 H/ACA-box
snoRNAs (0.0 - 0.1%) (Figu e 1b, 1c and Supplemen a y
Tables 2 and 3).
In e es ingly, in PCa samples we de ec ed up o
27% mo e C/D-box and up o 52% mo e H/ACA-box
snoRNAs compa ed o NAP o BPH. Fu he mo e, o al
snoRNA ead-coun s we e inc eased a leas wo- old,
indica ing possible ac i a ion o snoRNA-gene exp ession
in esponse o malignan ans o ma ion. In con as , he
numbe o de ec ed miRNAs emained ela i ely s able
and he o al miRNA ead-coun s changed by no mo e
han 19% (min. 9,202,300, max. 11,367,682) (Figu e 1c,
Supplemen a y Figu e 2 and Supplemen a y Tables 3 and
4).
We also examined he ead-leng h associa ed wi h
di e en ypes o sncRNAs. As expec ed, miRNA eads
had a na ow size dis ibu ion be ween 21 and 23 n in
all lib a ies. Simila size ange was obse ed o snRNA-
and scaRNA-de i ed RNAs in esh- ozen lib a ies. In
conco dance wi h ou p e ious esul s [3], we de ec ed
a size peak a 23 n and a pla eau be ween 26-28 n o
eads mapping o C/D-box snoRNAs. In e es ingly, eads
mapping o H/ACA snoRNAs and RNAs demons a ed
a shi in size dis ibu ion be ween no mal and malignan
Figu e 1: Summa y o sncRNA sequencing da a om PCa pa ien samples. (a) Numbe o e ie ed aw, ex ac ed, anno a ed,
and unique eads gene a ed o each one o he sequencing lib a ies. (b) Numbe o de ec ed sncRNA-species pe lib a y. (c) Rela i e
abundance o di e en sncRNA- ypes pe lib a y. Read-leng h dis ibu ion in no mal (d) and cance lib a ies (e) de i ed om esh- ozen,
( ) and FFPE ma e ial. Each sncRNA ype is ep esen ed by di e en colo : miRNA ( ed), SNORD (da k blue), SNORA (o ange), RNA
(g een), scaRNA (black), RNA (g ay), snRNA (yellow), scRNA (magen a), o he miscellaneous RNAs (ligh blue).
Onco a ge 17433
www.impac jou nals.com/onco a ge
samples (Figu e 1d, 1e and Supplemen a y Figu e 3)
sugges ing cance -associa ed al e a ions in sncRNA
p ocessing.
Compa ison o he sncRNA composi ion o
he FFPE lib a y wi h i s esh- ozen coun e pa
demons a ed ha he ela i e miRNA ead-con en
in FFPE dec eased 3.9- old om 92% o 24% o he
o al anno a ed eads. On he con a y, he numbe o
eads mapping o o he sncRNA species was s ongly
ele a ed i.e. sequence ead-coun s we e inc eased 152-
old o snRNAs, 12.7- old o H/ACA-box snoRNAs,
5.6- old o RNAs, and 2.7- old o C/D-box snoRNAs
(χ2 es , p < 0.0001 o all es ed g oups) (Figu e 1c,
Supplemen a y Figu e 2 and Supplemen a y Table 4). The
size dis ibu ion o ead-leng h in FFPE ma e ial was also
s ongly a ec ed o all examined ncRNA g oups excep
o miRNAs. (Figu e 1e, and Supplemen a y Figu e 3).
These obse a ions can be explained wi h he highe le el
o RNA deg ada ion in FFPE o ansc ip s longe han
miRNA [35, 36].
Mapping, and anno a ion o sncRNA-de i ed
RNAs (sncdRNAs)
The majo i y o miRNA eads in small RNA
sequencing da a map o he speci ic loca ion on hei p e-
miRNA co esponding o he ma u e miRNA. Simila ly,
eads mapping o o he sncRNAs, o igina e om
speci ic posi ions on hei p ecu so a he han being
andomly de i ed and can ep esen speci ic, biologically
unc ional, smalle RNA species, e.g. sdRNAs o RNA
agmen s ( RFs) [37]. Ne e heless, he assignmen o
RNA-seq sequence- eads o speci ic sdRNAs o RFs o
quan i a ion pu poses is hampe ed by he lack o p ope
anno a ion. Fu he mo e, many sncRNAs p oduce mul iple
agmen s [29, 30] ha may o e lap each o he , which
u he complica es he exac de e mina ion o hei o igin
loci and a subsequen quan i a i e analysis.
To co ec ly de e mine he bounda ies o sdRNAs,
RFs and o he sncRNA-de i ed RNAs (sncdRNAs) in ou
da ase and anno a e hei speci ic loca ion on he p ecu so
sequence, we applied he compu a ional algo i hm
Onco a ge 17434
www.impac jou nals.com/onco a ge
Figu e 2: FlaiMappe esul s. (a)To al numbe o de ec ed sncRNA p ecu so s pe RNA ype and sequencing lib a y. (b) To al numbe
o sncdRNAs pe p ecu so ype and sequencing lib a y. (c, d, e, ) Di e en ypes o sncRNAs p oduce di e en numbe o agmen s. (g)
Rela ion be ween he numbe o agmen s p oduced pe p ecu so RNA and he exp ession le els o indi idual agmen s. (h) Rela ion
be ween he numbe o agmen s p oduced pe p ecu so RNA and he exp ession le els o he mos abundan agmen pe p ecu so .

Onco a ge 17435
www.impac jou nals.com/onco a ge
F agmen Loca ion Anno a ion and Iden i ica ion Mappe
(FlaiMappe ) and e alua ed i s pe o mance in his da a
se as desc ibed [38]. Sho ly, FlaiMappe p edic ed 5’-
and 3’-miRNA ends we e compa ed wi h he 5’- and
3’-end bounda ies o co esponding ma u e miRNAs in
MiRBase, 17 [39]. 82% o he de ec ed miRNAs had a
co ec ly de e mined 5’-end exac ly ma ching miRBase
anno a ion. An addi ional 11 % had an o se o 1 n . In
ag eemen wi h p e ious obse a ions [39], 3’-ends o
ma u e miRNAs had highe a iabili y and ma ched
miRBase anno a ions o 45%. F om he in es iga ed
miRNAs addi ional 33% had 1 n o se , and 14%, 2 n
o se (Supplemen a y Figu e 4).
Gi en he high con idence wi h which FlaiMappe
iden i ied 5’- and 3’-end bounda ies o bona ide miRNAs,
we pe o med anno a ion o all sncdRNAs in ou esh-
ozen lib a ies. We de ec ed 3927 unique sncdRNAs
de i ed om di e en p ecu so classes. F om hese, 1524
Figu e 3: Global exp ession changes o sncdRNAs in no mal and malignan p os a e issue. Uppe and middle panels
p esen sca e plo s compa ing he no malized exp ession alues o indi idual sncdRNA (do s) in each p os a e cance lib a y (PCa)
du ing p og essing disease o hese in he lib a y p epa ed om no mal adjacen p os a e issue (NAP). The exp ession o sncdRNAs in
he ho mone- e ac o y, ansu e h al esec ion o he p os a e (TURP HR) lib a y is also compa ed o he benign p os a e hype plasia
(BPH) lib a y since he la es ep esen s he no mal coun e pa o malignan ansu e h al esec ion o he p os a e ma e ial. Di e ences
in sncdRNA exp ession be ween biological eplica es o Gleason 6 cance s (PCa 6) as well as compa ison o a esh- ozen lib a y (FF)
wi h i s o malin- ixed, pa a in-embedded (FFPE) coun e pa de i ed om he same pa ien s a e p esen ed in he lowe panels. Each
sncdRNA ype is p esen ed by a di e en colo . Diagonal lines ac oss each sca e plo ep esen old change di e ence in exp ession.
Middle line, c ossing he ho izon al and e ical axes a 0, no exp ession change; lines c ossing he e ical and ho izon al axes a 2, wo-
old exp ession change; lines c ossing he e ical and ho izon al axes a 4, ou - old exp ession change. Cu ed, no disease elapse a e
adical p os a ec omy; ec., ecu en disease, biochemical o me as a ic elapse a e su ge y; LN, me as a ic lymph node sample; TERG+,
TMPRSS2-ERG usion gene e en ; TERG-, no TMPRSS2-ERG usion e en ; Numbe s (6, 7 o 8) a e PCa indica e he pa hological
Gleason sco e o he umo s in he espec i e g oup
Onco a ge 17436
www.impac jou nals.com/onco a ge
o igina ed om miRNAs, 1175 - om RNAs, 657 – om
C/D-box snoRNAs, 244 – om H/ACA-box snoRNAs,
and 327 – om o he sncRNA species (Supplemen a y
Table 5 and Supplemen a y File 2). The o al numbe o
de ec ed unique sncdRNAs was highe han he numbe o
de ec ed unique p ecu so species showing ha indi idual
sncRNA p ecu so s p oduce mo e han one sncdRNA
(Figu e 2a, 2b and Supplemen a y Figu e 5). Fo example,
he majo i y o p e-miRNAs p oduced one o wo miRNAs
co esponding o he guide and passenge s and. Fo C/D
box snoRNAs we de ec ed be ween 1 and 6 sdRNAs
o igina ing om he same p ecu so , wi h he excep ion o
he unusually long SNORD3A, SNORD3B and SNORD3C,
which gi e ise o 10 o 13 C/D-sdRNAs. Mos H/ACA-
box snoRNAs p oduced be ween 1 and 3 sdRNAs, while
o RNAs we de ec ed be ween 1 and 6 RFs pe p ecu so
(Figu e 2d-2 ). O he examined sncRNAs in ou lib a ies
p oduced a a ying numbe o agmen s anging om 3
o he elome ase RNA componen o 28 o he small
nuclea 7SK RNA (Figu e 2c; Supplemen a y Figu e 6 and
7).
We nex a gued ha he exp ession le el o he
sncRNA-p ecu so migh posi i ely in luence he numbe
o sncdRNAs de ec ed pe sncRNA. We examined he
dis ibu ion o exp ession alues o indi idual sncdRNAs
in ela ion o he numbe o sncdRNAs de i ed pe
sncRNA and could no obse e a s ong dependency
be ween he median exp ession le els o sncdRNAs and
he o al numbe o sncdRNAs p oduced pe sncRNA.
We ob ained simila esul s when he exp ession le el o
he mos abundan sncdRNA pe p ecu so was used as
a su oga e measu e o he exp ession o he p ecu so
RNA (Figu e 2g, 2h). Based on hese esul s, we can
conclude ha mul iple sncdRNAs o igina ing om he
same sncRNA can be de ec ed independen ly o hei
(low) exp ession le el o he exp ession le el o hei
p ecu so . Vice e sa, di e en p ecu so RNAs can
p oduce only one sncdRNAs wi h e y high abundance.
Hence, i can be assumed ha he numbe and quan i y o
di e en sncdRNAs do no di ec ly e lec he abundance
o hei p ecu so bu , like miRNAs, a e p obably also
in luenced by addi ional aspec s o cellula me abolism,
e.g. associa ion wi h p o ein complexes and/o u no e
a es.
The size o unique sdRNAs anged be ween 15
and 29 n (Supplemen a y Figu e 6). Howe e , when
he exp ession o indi idual sdRNAs o he same leng h
we e accoun ed, we obse ed a p edominan size o 23
n o he majo i y H/ACA-sdRNAs and a binominal size
dis ibu ion o C/D-sdRNAs wi h wo p edominan sizes
o 22-23 n and 28 n (Supplemen a y Figu e 7), which is
in ag eemen wi h ou p e ious indings and o he epo s
[3, 22, 23, 40]. C/D-sdRNAs demons a ed a b oade size
dis ibu ion, which howe e could be a e lec ion o he
b oade size ange o hei p ecu so s.
sdRNAs a e di e en ially exp essed in p os a e
cance
P e iously, we obse ed di e en ial exp ession
o sdRNAs be ween PCa specimens [3]. To examine i
such changes a e a cance -speci ic e en we compa ed
he exp ession o FlaiMappe de ined sncdRNAs
be ween no mal (NAP and BPH) and malignan issues
o p og essing disease (PCa, LN, TURP). We de ec ed
be ween 34 and 202 sncdRNAs wi h signi ican di e en ial
exp ession (Table 1, Figu e 3, and Supplemen a y File 3).
App oxima ely one hi d o he di e en ially exp essed
RNAs in each compa ison comp ised C/D-sdRNAs
up egula ed in cance (Figu e 3). In con as , only one
sdRNA was di e en ially exp essed be ween non-
malignan samples (NAP and BPH) and only i e, be ween
biological eplica e samples (PCa, Gleason 6, g oups 3,
4, and 10). This sugges s ha he accumula ion o C/D-
sdRNAs is p ima ily d i en by malignan ans o ma ion.
To examine he e ec o sample s o age on
agmen abundance we compa ed he exp ession o
sncdRNAs be ween he FFPE sample and i s esh- ozen
(FF) coun e pa . We limi ed compa ison analyses o
sncRNAs de ec ed in any o he FF lib a ies. MiRNAs
had dec eased exp ession in FFPE compa ed wi h
sdRNAs, RFs and o he sncdRNAs (Table 1, Figu e 3 and
Supplemen a y Figu e 8). Ne e heless, he educ ion o
miRNA exp ession in FFPE appea s o be he esul o a
global dec ease in miRNA ead-coun s compa ed o ead-
coun s o o he sncdRNAs (Figu e 1c) since he ela i e
exp ession o miRNAs co ela ed s ongly be ween bo h
condi ions (Pea son ρ = 0.9289) (Supplemen a y Figu e
8). This was no obse ed o sdRNAs (Pea son ρ = 0.6557
o C/D-sdRNAs and 0.3895 o H/ACA-sdRNAs) o
o he sncdRNAs, which ha e longe p ecu so s and may
be mo e suscep ible o deg ada ion in FFPE ma e ial.
SdRNAs demons a e speci ic global p ocessing
pa e ns in p os a e issue
Gi en he disc e e size and speci ic exp ession
o sdRNAs, we examined de ec ed snoRNAs o he
p esence o a common p ocessing pa e n. To be able o
compa e wi h miRNAs, we aligned all snoRNA and p e-
miRNA sequences and isualized he posi ion and ela i e
abundance o he co esponding sdRNAs and miRNAs
(Figu e 4, Supplemen a y ile 4 and 5). The majo i y o
sdRNAs o igina ed om equi alen loca ions o hei
p ecu so s. O en, one p edominan sdRNA was obse ed
pe p ecu so . The posi ion o hese p edominan sdRNAs
was no dependen on he o al numbe o smalle species
de ec ed pe p ecu so sequence, showing a a he uni o m
agmen a ion pa e n consis en wi h he p ecu so - ype.
This is in ag eemen wi h p e iously sugges ed speci ic
snoRNAs p ocessing and accumula ion o smalle RNAs
Onco a ge 17437
www.impac jou nals.com/onco a ge
obse ed in cell lines [28, 32].
In ou pa ien samples, p edominan H/ACA-
sdRNAs o igina e om ei he he 5’-a m o he i s H/
ACA-snoRNA hai pin (38.5%) o he 3’-a m o he
second hai pin including he egion o he ACA-box (31%)
(Figu e 4b). C/D-box snoRNA p oduce wice as many
p edominan sdRNAs o igina ing om he 5’- e minus
ha con ain a C-box (60.1%) compa ed o 3’- e minal
sdRNAs ha con ain a D-box (30.1%) (Figu e 4a).
In e es ingly, indi idual C/D-sdRNAs wi h
highly simila sequences demons a e almos iden ical
agmen a ion pa e n, which is also dependen on he
conse a ion o snoRNA s uc u al ea u es. Fo example,
snoRNAs om he highly conse ed, mul iple gene-copy
SNORD116 amily (HBII-85), which ha e a degene a ed
C’-box (UGAGUGA) p oduce ou sdRNAs whe e he
mos abundan one maps o he 5’- egion co e ing he
C-box. SnoRNAs om he SNORD115 (HBII-52) amily
wi h conse ed C’/D’-boxes p oduce h ee sdRNAs, wi h
he mos abundan ones mapping o he middle- egion and
co e ing he en i e K-loop including he C’/D’-box. In
con as , he la ge snoRNAs om he SNORD3 amily,
which lack a conse ed C-box, p oduce be ween 10 and 13
o e lapping sdRNAs wi h he mos p edominan mapping
o he 3’-end.
SNORD115 and SNORD116 sdRNAs di e in size
and posi ion om he p e iously epo ed highly abundan
psnoRNAs p ocessed om he o hologous MBII-52 and
MBII-85 de ec ed by RNase p o ec ion assays [29, 30,
41]. This disc epancy could be explained by he implici
me hodology di e ences be ween sncRNA sequencing and
RNase p o ec ion assays. Howe e , hese di e ences could
be also caused by issue-speci ic sdRNA accumula ion
as p e iously desc ibed o sdRNAs o igina ing om
SNORD88C (HBII-180C) [32] o by he dependence o
p ocessing mechanisms on he s uc u al conse a ion o
C/D-box snoRNAs. O no e, SNORD115, SNORD116, o
SNORD88C-o igina ing sdRNAs we e de ec ed a low
abundance in ou samples.
P ocessing and exp ession o sdRNAs o igina ing
om GAS5 encoded C/D-box snoRNAs is ela ed
o he conse a ion o s uc u al C’/D’-boxes
We in es iga ed whe he he agmen a ion pa e n
o o he C/D-box snoRNA is also dependen on s uc u al
ea u e conse a ion. Fo his we analyzed he posi ional
o igin o a highly abundan sdRNA p oduced om he 3’-
end o SNORD78 [3] and o he sdRNAs om he same
locus. SNORD78 is in onically encoded by he G ow h
A es Speci ic 5 gene (GAS5) oge he wi h 9 o he C/D-
box snoRNAs [42]. All 10 SNORDs a e p esumably
simul aneously ansc ibed as a GAS5 p ecu so - ansc ip ,
which unde goes in on emo al and pos ansc ip ional
p ocessing. We could de ec sdRNAs om all 10 GAS5-
encoded snoRNAs. Howe e , only ou (SNORD44,
SNORD78, SNORD74 and SNORD81) snoRNAs p oduced
abundan sdRNAs (Figu e 5 and Supplemen a y Figu e 9).
In e es ingly, SNORD74 and SNORD81 p oduced
h ee abundan sdRNAs wi h simila , ela i ely low
exp ession le els ha mapped o he 5’-, 3’-, and middle
egion o he snoRNAs. The 3’- and middle sdRNAs
o e lapped each o he and co e ed he K-loop and he
conse ed canonical C’/D’-box (Figu e 5). In con as ,
SNORD78 and SNORD44, which lack he canonical C’/
D’-box, p oduced p edominan ly one 28 n long sdRNA
each, mapping o he 3’-a m o SNORD78 (sd78-3’) o he
5’-a m o SNORD44 (sd44-5’). sd78-3’ and sd44-5’ we e
s ongly up egula ed in samples p epa ed om malignan
issue compa ed o no mal o benign, while middle- and
opposi e a m-de i ed sdRNAs we e p esen only a e y
low ead-coun s in all lib a ies (Supplemen a y Figu e 9a
and 10).
SNORD78 and sd78-3’ exp ession is associa ed
wi h me as a ic PCa
To alida e ou sequencing da a we es ed he
exp ession o SNORD44, SNORD78, SNORD74,
SNORD81, and hei de i a e sdRNAs, in an independen
pa ien coho o 106 esh- ozen clinical samples
by quan i a i e eal- ime PCR (Q-PCR). To e alua e
whe he inc eased sdRNA exp ession is a esul o a
gene al ac i a ion o he GAS5 locus, we also measu ed
he exp ession o he spliced GAS5 ansc ip (Figu e
6 and Supplemen a y Figu e 9b). All es ed snoRNAs
and sdRNAs we e up egula ed in o gan-con ined PCa
compa ed o no mal adjacen con ols. This was no
ela ed o an ele a ion o he spliced GAS5 ansc ip ,
which did no demons a e p onounced exp ession changes
be ween NAP and PCa. In e es ingly, o e lapping sdRNAs
o igina ing om he same snoRNA as well as ull-leng h
snoRNAs we e simul aneously de ec able by Q-PCR
sugges ing he exis ence o mul iple con o ma ional s a es
o hese snoRNAs.
Sd78-3’, SNORD78 and GAS5 exp ession was also
de ec able in di e en no mal basal p os a e epi helium
cell lines (PNT2C2, RWPE) p os a e cance cell lines
(PC346C, LAPC4, VCAP, LNCAP, 22RV1, PC3,
and DU145N) as well as in hepa ocellula ca cinoma
(HEP3B) and colon adenoca cinoma (COLO205) cells
demons a ing ha SNORD78 p ocessing o sd78-3’ is no
es ic ed o p os a e issue o cells. Simila ly o pa ien
da a, he exp ession le els o sd78-3’ and SNORD78 we e
no co ela ed o he exp ession o he GAS5 hos gene
(Supplemen a y Figu e 11).
Consis en wi h ou p e ious esul s [3] sd78-3’
was up egula ed in he LN lib a y gene a ed in his s udy,
sugges ing associa ion o his sdRNA wi h agg essi e
disease. The e o e, in he alida ion coho we s a i ied
pa ien s wi h o gan-con ined disease a he ime o adical
Onco a ge 17438
www.impac jou nals.com/onco a ge
Figu e 4: Global p ocessing pa e ns and ela i e abundance o sdRNAs and miRNAs exp essed in p os a e (cance )
issues. (a) Full-leng h C/D-box snoRNAs a e aligned ela i e o he middle nucleo ide o each sequence. (b) H/ACA-box snoRNAs a e
aligned based on he posi ion o he H-box. (c) P e-miRNAs a e aligned ela i e o he middle nucleo ide o each sequence. A g een line
ep esen s each ull-leng h sncRNA. Sequences a e ex ended 10 n a each end o a oid mapping ambigui y caused by inco ec anno a ion.
Posi ions o de ec ed conse ed H/ACA-boxes o C/D-boxes a e shown in blue and ed. Ligh and da k g ey lines indica e he posi ional
o igin o sdRNAs, miRNAs and miRNAs*. The colo in ensi y co esponds o he ela i e abundance o sncdRNAs o igina ing om he
same p ecu so ( ead-coun as a pe cen age o he o al ead-coun pe p ecu so ), e.g. i only one sdRNA pe snoRNA-p ecu so is de ec ed
i is assigned 100% abundance, i wo o mo e sdRNAs o igina e om he same snoRNA he sdRNA wi h he highes ead-coun is gi en
he da kes colo and he sdRNA wi h he lowes ead-coun - he ligh es . Thin dashed lines sepa a e each panel in o h ee subg oups whe e
sncRNAs p oducing only one sncdRNA a e on op, sncRNAs p oducing wo sncdRNAs a e in he middle and hose p oducing he e o mo e
sncdRNAs a e on he bo om. The highly sequen ially conse ed, mul iple gene-copy C/D-box snoRNAs om he SNORD116 (HBII-85)
and SNORD115 (HBII-52) amilies a e g ouped oge he below o he C/D-box snoRNAs. The X-axis indica es he posi ion o sncdRNAs
ela i e o he cen e o hei p ecu so sequence. The Y-axis depic s he numbe o ull-leng h sncRNA p ecu so s.