Unmasking a new p ognos ic ma ke and he apeu ic a ge
om he GDNF-RET/PIT1/p14ARF/p53 pa hway in ac omegaly
Miguel Chenlo
a,j,1
, I ia A. Rod iguez-Gomez
b,c,j,1
,RamonSe ami o
d,j
, Angela R. Ga cia-Rendueles
a,j
,
Rocío Villa -Taibo
b,j
, E a Fe nandez-Rod iguez
, Siha a Pe ez-Rome o
a,j
, Ma ia Sua ez-Fa iña
a,j
,
Al edo Ga cia-Allu
d,j
, Jose M. Cabezas-Ag icola
b,j
, Ja ie Rod iguez-Ga cia
e,j
,PamelaV.Lea
g
,
Rosa M. Al a ez-San Ma in
h
, C is ina Al a ez-Escola
i
, Ignacio Be nabeu
b,j,
⁎, Cla a V. Al a ez
a,j,
⁎⁎
a
Neoplasia & Endoc ine Di e en ia ion P0L5, Cen o de In es igación en Medicina Molecula y En e medades C ónicas (CIMUS), Uni e si y o San iago de Compos ela (USC), Spain
b
Se icio de Endoc inología y Nu ición, Complejo Hospi ala io Uni e si a io de San iago de Compos ela (CHUS)-SERGAS, San iago de Compos ela, Spain
c
Se icio de Endoc inología y Nu ición, Hospi al HM Modelo, A Co uña, Spain
d
Se icio de Neu oci ugía, Complejo Hospi ala io Uni e si a io de San iago de Compos ela (CHUS)-SERGAS, San iago de Compos ela, Spain
e
Se icio de Análisis Clínicos, Complejo Hospi ala io Uni e si a io de San iago de Compos ela (CHUS)-SERGAS, San iago de Compos ela, Spain
Se icio de Endoc inología y Nu ición, Complejo Hospi ala io Uni e si a io de Ou ense, Spain
g
Depa men o Physiology, Ana omy and Gene ics, Ox o d Uni e si y, Ox o d, Uni ed Kingdom
h
Se icio de Endoc inología y Nu ición, Complejo Asis encial Uni e si a io de Leon, Spain
i
Se icio de Endoc inología y Nu ición, Hospi al Uni e si a io La Paz, Mad id, Spain
j
Ins i u o de In es igación Sani a ia (IDIS), San iago de Compos ela, Spain
abs ac a icle in o
A icle his o y:
Recei ed 28 Janua y 2019
Recei ed in e ised o m 2 Ap il 2019
Accep ed 2 Ap il 2019
A ailable online 8 Ap il 2019
Backg ound: Ac omegaly is p oduced by excess g ow h ho mone sec e ed by a pi ui a y adenoma o soma o oph
cells (ACRO). Fi s -line he apy, su ge y and adju an he apy wi h soma os a in analogs, ails in 25% o pa ien s.
The e is no p edic i e ac o o esis ance o he apy. New he apies a e in es iga ed using ew dispe sed umo
cells in acu e p ima y cul u es in s anda d condi ions whe e he cells do no g ow, o using a pi ui a y cell lines
ha do no main ain he ull soma o oph pheno ype. The RET/PIT1/p14ARF/p53 pa hway egula es apop osis in
no mal pi ui a y soma o ophs whe eas he RET/GDNF pa hway egula es su i al, con olling PIT1 le els and
blocking p14ARF (ARF) and p53 exp ession.
Me hods: We in es iga ed hese wo RET pa hways in a p ospec i e se ies o 32 ACRO and 63 non- unc ioning
pi ui a y adenomas (NFPA), s udying quan i a i e RNA and p o ein gene exp ession o molecula -clinical co e-
la ions and how he RET pa hway migh be implica ed in he apeu ic success. Clinical da a was collec ed du ing
pos -su gical ollow-up. We also es ablished new'humanized’pi ui a y cul u es, allowing 20 epea ed passages
and main aining he pi ui a y sec e o y pheno ype, and es ed fi e mul ikinase inhibi o s (TKI: Vande anib,
Len a inib, Suni inib, Cabozan inib and So a enib) po en ially able o ac on he GDNF-induced RET dime iza-
ion/su i al pa hway. An ibody a ays in es iga ed in acellula molecula pa hways.
Findings: In ACRO, he e was specific en ichmen o all genes in bo h RET pa hways, especially GDNF. ARF and
GFRA4 gene exp ession we e ound o be opposing p edic o s o esponse o fi s -line he apy. ARF cu -o le els,
calcula ed ca ego izing by GNAS mu a ion, we e p edic i e o good esponse (abo e) o esis ance (below) o
he apy mon hs la e . So a enib, h ough AMPK, blocked he GDNF/AKT su i al ac ion wi hou al e ing he
RET apop o ic pa hway.
In e p e a ion: Tumo ARF mRNA exp ession measu ed a he ime o he su ge y is a p ognosis ac o in ac omeg-
aly. The RET inhibi o , So a enib, is p oposed as a po en ial ea men o esis an ACRO.
Fund: This p ojec was suppo ed by na ional g an s om Agencia Es a al de In es igación (AEI) and Ins i u o
In es igación Ca los III, wi h pa icipa ion o Eu opean FEDER unds, o IB (PI150056) and CVA (BFU2016-
76973-R). I was also suppo ed ini ially by a g an om he In es iga o Ini ia ed Resea ch (IIR) P og am
(WI177773) and by a non- es ic ed Resea ch G an om Pfize Founda ion o IB. Some o he pi ui a y
Keywo ds:
Pi ui a y umo s
Soma o opinomas
Endoscopy su ge y
Ac omegaly
ARF
GFRA4
SSA- esis ance
Human pi ui a y cul u es
So a enib
EBioMedicine 43 (2019) 537–552
⁎Co espondence o: Ignacio Be nabeu, Se icio de Endoc inologia y Nu ición, Complejo Hospi ala io Uni e si a io de San iago de Compos ela (CHUS)-SERGAS, 15706 San iago de
Compos ela, Spain.
⁎⁎ Co espondence o: Cla a V Al a ez, Neoplasia & Endoc ine Di e en ia ion P0L5, Cen o de In es igación en Medicina Molecula y En e medades C ónicas (CIMUS), A Ba celona s/n,
15782 San iago de Compos ela, Spain.
E-mail add esses: ignacio.be nabeu.mo on@se gas.es (I. Be nabeu), cla a.al a [email protected] (C.V. Al a ez).
1
Equal co-fi s au ho s.
h ps://doi.o g/10.1016/j.ebiom.2019.04.007
2352-3964/© 2019 Published by Else ie B.V. This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/).
Con en s lis s a ailable a ScienceDi ec
EBioMedicine
jou nal homepage: www.ebiomedicine.com
ac omegaly samples we e collec ed in he amewo k o he Spanish Na ional Regis y o Ac omegaly (REMAH),
pa ially suppo ed by an un es ic ed g an om No a is o he Spanish Endoc ine Associa ion (SEEN).
CVA is also suppo ed om a g an o Medical Resea ch Council UK MR/M018539/1.
© 2019 Published by Else ie B.V. This is an open access a icle unde he CC BY-NC-ND license (h p://
c ea i ecommons.o g/licenses/by-nc-nd/4.0/).
1. In oduc ion
Pi ui a y umo s a e he 16% (11–22%) o in ac anial neoplasms [1].
Ac omegaly is a a e disease caused by a benign pi ui a y adenoma o
soma o oph cells (ACRO) sec e ing g ow h ho mone (GH) (O phane :
963; GARD-5725). The majo i y o cases a e spo adic, al hough amilial
cases ha e been desc ibed [2].Ac omegaly has symp oms de i ed om
expansion o he umo comp essing he pi ui a y o he su ounding
s uc u es such as he ca e nous sinus, op ic chiasma o hypo halamus.
Addi ionally, excess GH and consequen ly ele a ed IGF1 le els, p oduce
pe iphe al signs wi h g ow h o so issues and mul iple como bidi ies
(me abolic, ca dio ascula , oncological, e c.; ( e iewed elsewhe e
[3–5]). Fi s -line he apy ini ially comp ises endoscopic su ge y aiming
o cu e he disease wi hou damaging he emaining pi ui a y gland.
Howe e , he su gical cu e a e is only a ound 50% (40–69%), being in-
e sely ela ed o umo size and in asi eness ou side he sella u cica
[6].Pa ien s no cu ed by su ge y unde go adju an he apy wi h fi s -
gene a ion analogs o soma os a in ( gSSA), he main hypo halamic
neu opep ide ha inhibi s GH sec e ion [7]. Howe e , only hal o pa-
ien s no cu ed by su ge y espond o gSSA [8].In gSSA- esis an
cases second-line he apies include combined ea men wi h dopa-
mine analogs, peg isoman o pasi eo ide, and e en e-in e en ion o
adio he apy. These p ocedu es ha e a ange o side e ec s wi hou
necessa ily con olling he appa en ly benign umo .
RET is a y osine kinase ecep o ac i a ed by a ligand in he p es-
ence o a memb ane co- ecep o [9]. In mammals he e a e ou di e -
en ligands o RET (Glial cell line-De i ed Neu o ophic Fac o
(GDNF), Neu u in (NRTN), A emin (ARTM), Pe sephin (PSPN)) and
ou espec i e co- ecep o s (GFRA1–4) [10,11].In addi ion o he
espec i e ligand-co- ecep o in e ac ion, some c oss-in e ac ions
be ween ligands and co- ecep o s able o ac i a e RET ha e been de-
sc ibed. No mal human and oden soma o ophs exp ess RET, GDNF
and GFRA1 [12,13], wi h RET wo king as a dependen ecep o , a mech-
anism sha ed by a ew o he ecep o s [14].Thus, in he absence o
GDNF i s ligand RET is p ocessed a he memb ane by Caspase-3 and in-
duces o e exp ession o PIT1, leading o induc ion o he CDKN2A/ARF
p omo e and ARF mRNA exp ession; p14ARF p o ein hen binds o
and inhibi s MDM2, leading o p53 accumula ion and apop o ic cell
dea h [15,16].When GDNF is p esen , RET dime izes, ac i a ing i s y o-
sine kinase which induces AKT and cell su i al. Thus, soma o ophs
ely on he p esence o he RET ligand GDNF o su i al.
Immunohis ochemical s udies ha e shown ha ACRO a e he only
pi ui a y adenomas exp essing all h ee p o eins, RET, GDNF and
GFRA1 [13].Mo e ecen ly, in i o expe imen s in nine acu ely dis-
pe sed ACRO ha e sugges ed ha ACRO main ain he RET/PIT1/ARF/
p53 apop o ic pa hway and need GDNF exp ession o su i e [17].In a
Resea ch in con ex
E idence be o e his s udy
25% o pa ien s wi h a head pi ui a y umo sec e ing g ow h ho -
mone (GH), ac omegaly, will be esis an o i s -line he apy, su -
ge y+ gSSA. Mon hs la e hose pa ien s will su e adio he apy
o second/ hi d su ge y in he head while p esen ing mul iple me -
abolic, ca dio ascula , join and oncological body como bidi ies.
The e is no p edic i e ac o o esis ance o he apy. In he las
yea s we ha e had a sca ci y o new he apies o pi ui a y ac o-
megaly add essing he umo s. One o he easons is he di icul y
o es ablish ac omegaly p ima y cul u es main aining he
soma o oph sec e o y pheno ype and g owing in i o. To ex-
p ess and sec e e GH a cell needs o exp ess he ansc ip ion ac-
o PIT1. The ew a pi ui a y cell lines sec e ing GH do no
main ain he ull panel o ecep o s cha ac e is ic o
soma o ophs.
In no mal pi ui a y soma o ophs he RET ecep o loca ed a he
memb ane is implica ed in o wo opposi e pa hways. In one
hand, when he e is no GDNF ligand eaching he cell su ace,
RET beha es as a “dependence ecep o ”being in acellula ly p oc-
essed by caspase and inducing o e exp ession o PIT1, which ac-
i a es he CDKN2A/p14ARF p omo e leading o p53
accumula ion and cell dea h. Tha is he eason why he
soma o ophs a e “dependen ”on he GDNF ligand exp ession in-
ducing RET ecep o dime iza ion a he memb ane, ac i a ing i s
y osine-kinase ac i i y and leading o AKT phospho yla ion, con-
olling PIT1 le els, enough o GH sec e ion bu main aining
su i al.
Added alue o his s udy
We ca ied ou a p ospec i e s udy o mo e han se en yea s o
e alua e he impo ance o he RET ecep o pa hways in
soma o oph umo s analyzing and cul u ing esh issue ob ained
a ini ial su ge y in co ela ion wi h clinical da a du ing ollow-up.
GDNF, he RET ligand, is hea ily o e exp essed. Mo eo e ,
p14ARF- he p53 ac i a o - exp ession measu ed a he ime o
he su ge y is a p ognosis ac o o pos e io esis ance o he apy.
We es ablished a new “humanized”p o ocol o cul u ing human
soma o oph adenoma cells, whe e all componen s a e adjus ed
o concen a ions ound in human se um. Pa ien s' cells a e able
o g ow main aining he soma o oph pheno ype. Following es -
ing o i e mul ikinase inhibi o s, we ound ou ha a new pha ma-
co he apy wi h So a enib could be e ec i e o esis an
ac omegaly.
Implica ions o all he a ailable e idence
Mechanis ic s udies in human umo samples in pa allel wi h p i-
ma y cul u es om he same pa ien s ha e been e y ele an
o medicine. Un o una ely, when umo cells a e cul u ed unde
non ideal condi ions, he ele ance o he esul s is los .
We ha e been able o ob ain a ma ke in he umo o la e esis-
ance o he apy, p14ARF. Ul ima ely, he abili y o iden i y pa-
ien s who a e esis an o i s -line he apy and he ial o new
pha maceu icals should imp o e ou comes in ac omegaly. Ou
in i o e idence disca ds some kinase inhibi o s bu p oposes So-
a enib as a pu a i e ea men o hose esis an pa ien s. Using
his new cul u e sys em o he he apies, e en mo e selec i e o
RET, could be iden i ied o la e be es ed in clinical ials.
538 M. Chenlo e al. / EBioMedicine 43 (2019) 537–552
small g oup o ACRO issues, GDNF exp ession was in e sely co ela ed
o PIT1 exp ession [17].To da e, no s udies ha e shown whe he o he
ligands o GFRA ecep o s a e exp essed in ACRO, no a e he e any
da a ela ing o he clinical ele ance o he RET/PIT1/ARF/p53 apop o ic
o he RET/GDNF su i al pa hways in ac omegaly.
In his p ospec i e s udy ou goal was o in es iga e he ole o each
o hese RET pa hways in ac omegaly, specifically in ela ion o clinical
umo cha ac e is ics, esponse o ea men and p ognosis. Addi ion-
ally, we ca ied ou in i o es s on fi e mul ikinase inhibi o s able o
ac on he GDNF-induced RET dime iza ion/su i al pa hway. Ou
main finding was ha So a enib, a RET inhibi o , is a po en ial he apeu-
ic agen in esis an ACRO.
2. Ma e ials and me hods
2.1. S udy design and clinical da a
This s udy ollowed he Eu opean Con en ion on Human Righ s and
Biomedicine (ETS n°164) eflec ed in he Spanish Law o Resea ch in
Biomedicine (14/2007). This was a p ospec i e s udy o ac omegaly
(ACRO) pa ien s ec ui ed om Oc obe 2010 ill Feb ua y 2017 om
h ee hospi als in Spain: Hospi al Clinico Uni e si a io de San iago de
Compos ela (CHUS, N= 20), Hospi al Uni e si a io La Paz de Mad id
(N= 6) and Hospi al Uni e si a io de León (N= 4). Two pa ien s
unde wen eope a ion. The s udy was app o ed by bo h a na ional
e hics commi ee based in Hospi al Vi gen del Rocio (Se illa) (Ac 1/
2010), as pa o a na ional egis y o Ac omegaly (REMAH node
6) [18], and an ins i u ional commi ee a CHUS, whe e he molecula
analyses we e pe o med. App o ed w i en consen was ob ained
om each pa ien pa icipa ing in he s udy. Pa ien s unde wen endo-
scopic endonasal ansphenoidal su ge y. Non- unc ioning pi ui a y ad-
enomas (NFPA) we e also p ospec i ely collec ed in CHUS (N= 50) and
Leon (N= 13) o compa ison wi h ACRO a he molecula le el.
The s udy was designed ollowing REMARK guidelines [19](h p://
www.equa o -ne wo k.o g/ epo ing-guidelines/ epo ing-
ecommenda ions- o - umou -ma ke -p ognos ic-s udies- ema k/)
(adap ed o endoc ine umo s). N≥5 was calcula ed as he sample size
in a g oup o ob ain significan di e ences. Sample size calcula ions
we e based on he known ailu e o fi s -line he apy [6–8]: su ge y ail-
u e has been epo ed o a y om 31% o 60% in p e ious se ies; o e-
maining non-su gically cu ed pa ien s, hal will no espond o he apy
wi h gSSA, meaning ha 15.5% o 30% o pa ien s can be expec ed o be
Resis an . So o ha e N≥5 in he Resis an g oup he cu en s udy
needed o ha e a o al size o 17–32 pa ien s, and we he e o e included
32.
Clinical da a om ACRO we e collec ed a he end o ollow-up by
one Clinical Endoc inologis in a p e-designed Excel file. Va iables in-
cluded we e based on cu en guidelines [6,7,20] and a e shown in Sup-
plemen a y Table 1: gende ; age; GH a diagnosis and a e su ge y;
IGF1 index (defined as se um le els espec i e o he uppe le el o no -
mali y o he co esponding li e decade) a diagnosis and a e su ge y
( he mos ele an wi hin 4–24 weeks); maximal diame e o he ade-
noma in he magne ic esonance imaging (MRI), olume calcula ed
om h ee diame e s o he umo , Knosp (g ades o in asion o
ca e nous sinus), in asi eness (a MRI, ex ension o he umo in any
s uc u e ou side he sella u cica); cu ed a su ge y (when IGF1 index
4–24 weeks a e su ge y was no mal ≤1; i IGF1 index was epea edly
ound o be 1–1·5, an O al Glucose Tole ance Tes was pe o med o as-
sess cu ing); adju an he apy wi h SSA analogs ( gSSA, ea men wi h
fi s -gene a ion analogs, Oc eo ide-LAR o Lan eo ide au ogel); esid-
ual umo (p esence o umo es a con ol MRI 4–24 weeks a e su -
ge y); esponse o SSA analogs (Comple e: no maliza ion o IGF1 index
a e 6 mon hs o ea men ; Pa ial: when no no malized bu he e
was a educ ion o 50%); esponse o fi s -line he apy (The pa ien
was ei he cu ed by su ge y o esponded o adju an SSA analogs);
o he ea men s o hose who did no espond o fi s -line ea men
(peg isoman , adio he apy). Molecula da a we e analysed in pi ui a y
umo su gical su pluses by a single labo a o y in San iago de
Compos ela (CIMUS).
2.2. Tissue sample managemen
Adenoma issue su plus was imme sed di ec ly in RNA La e
(AM7021, Ambion, Ne he lands), kep a 4 °C o 24 h, du ing which i
was cou ie ed o San iago de Compos ela, whe e i was kep a −20
°C o ano he 24 h and s o ed a −80 °C un il u he use. F om 2014,
o su plus issue samples N10 mg a CIMUS, a small esh piece was
ans e ed o he cul u e oom o h7H p ima y cul u e (see below).
2.3. RNA and DNA ex ac ion
Nucleic acids we e ex ac ed using an AllP ep DNA/RNA Mini Ki
(80284, Qiagen, Ge many), in which DNA is fi s pu ified in an a fini y
column, and RNA is hen pu ified om he elu ed ac ion in a second a -
fini y column which includes ea men wi h an RNasa-F ee DNase Se
(79254, Qiagen, Ge many). DNA was QC and quan ified using a
Quan ifile Human DNA ki (4343895, Applied Biosys ems, UK). RNA
was QC and quan ified using a Nanod op 2000 spec opho ome e
(The mo).
2.4. qRT-PCR assay
500 ng o o al RNA we e incuba ed wi h 0.5 IU RNAse ee DNAse I
(EN0521, The mo), 1 mic oL 10× bu e and wa e as a o al olume o
10 mic oL, a 37 °C o 30 min. The eac ion was e mna ed by adding
1 mic oL 50 mM EDTA and hea ing a 65 °C o 10 min. cDNA was ob-
ained a e adding 1.5 mic oL wi h 300 IU MMLV (28025–013,
In i ogen, USA) and 6 mic oL 5× Fi s -S and Bu e and, ollowing
he supplie 's p o ocol, 1·5 mic oL 10 mM dNTPs, 0·1 mic oL Random
P ime s, 3 mic oL 0·1 M DTT, 1mic oL RNaseOUT™Recombinan Ribo-
nuclease Inhibi o (40 uni s/mic oL) and H2O in a 30 mic oL eac ion.
50 ng, 25 ng and 12·5 ng o a comme cial human Pi ui a y Gland Poly
A+ mRNA pool (1305204A, Clon ech, USA) was simila ly e e se-
ansc ibed in h ee independen eac ions o use as a echnical con ol.
Exp ession was de ec ed by qPCR using 1 mic oL o he cDNA eac ion
plus 6 mic oL 2× TaqMan Gene Exp ession Mas e Mix (4369016 Ap-
plied Biosys ems) and 6 mic oL dilu ed p ime s in 96 well pla es in a
7500 Real-Time PCR Sys em (4351105, Applied Biosys ems, USA). Fo
p ime s, when possible we used comme cial Taqman Gene exp ession
p ime se s (Applied Biosys ems) (see Supplemen a y Table 5). When
iso o ms o highly homologous genes we e in ol ed, (FSHB, LHB,
SSTR2 and SSTR5), we used p e iously alida ed p ime s (See Supple-
men a y Table 5) [1] wi h 6 mic oL B illian III Ul a-Fas SYBR
Mas e Mix (600,882, Agilen Technologies, USA). GH iso o m mRNA ex-
p ession was de ec ed by bo h assays. Each 96 well pla e was used o a
single gene eac ion including duplica e samples o a g oup o 10 ACRO,
20 NFPA oge he wi h echnical posi i e con ols, nega i e con ol
(ACRO9 e e se- ansc ibed in he absence o MMLV) and blank (all e-
agen s wi hou cDNA).
As con ol o gene al gene exp ession we used TBP based in pub-
lished wo ks [21,22](h p://www3.appliedbiosys ems.com/cms/
g oups/mcb_ma ke ing/documen s/gene aldocumen s/cms_042279.
pd ) and ou p e ious expe ience [7].ACTB is no conside ed a good
con ol because i shows highe a iabili y be ween samples om he
same and om di e en issues. TBP is conside ed as a gold s anda d
con ol o quan i a i e mRNA assays as i shows a mo e s able exp es-
sion pa e n in di e en human issues and cell ypes. The ΔC me hod
was used o exp ess he esul s o each gene (ΔC
gene
=C
gene
−C
TBP
). We calcula ed some o he da a using bo h TBP and ACTB
and ob ained simila esul s and s a is ical significance.
Genes ha we e unde ec able in a sample (C
gene
N40) we e classi-
fied acco ding o he C
TBP
. I o he samples wi h a simila C
TBP
had a
539M. Chenlo e al. / EBioMedicine 43 (2019) 537–552
de ec able C
gene
(C gene b40), he pa icula sample was conside ed o
ha e a e y low exp ession o ha gene and i s C
gene
ans o med o 42
( wo cycles o e he limi ). I a sample had a C
TBP
ha was highe han
any o he sample, i was conside ed o be echnically unde ec able o
ha gene (a los sample o s a is ical compa isons). The mean o he
h ee concen a ions o he echnical con ol (comme cial human pi u-
i a y gland pool) was conside ed as 1. The ΔΔC me hod, ΔC
gene-sample
−ΔC
gene- echnical con ol
, was used o calcula e he end alue o he
qPCR in each sample. Po ency o 2−
ΔΔC
was he end esul used o s a-
is ical compa isons.
2.5. GNAS mu a ional p ofile and STR analysis
The GNAS gene was amplified om 50 ng DNA by PCR (p ime s in
Supplemen a y Table 5) using 0·5 IU D eamTaq DNA Polyme ase
(EP0701 The mo Scien ific), 2·5 mic oL D eamTaq Bu e , 0·5 mic oL
10 mM dNTPs, 0·5 mic oL 5mic oM Fo wa d P ime and 0·5 mic oL 5
mic oM Re e se P ime and dH20 o a final olume o 25 mic oL. The
PCR P oduc was Sange -sequenced o check o mu a ions in nucleo-
ide C601/codon R201 o nucleo ide A680/codon Q227 (Supplemen a y
Fig. 1a, Supplemen a y Table 3).
Fo assessmen o simila i y be ween pi ui a y issue and p ima y cul-
u e DNA we used a combined panel o 16 STR AmpFiSTR NGM Selec ki
(4457889, Applied Biosys ems, Ge many) including Iden ifile Plus Panel
V1 (Applied Biosys ems) as he ead-ou pa e n. The 16 STR p ofiles ob-
ained we e compa ed o he 8 STR Ame ican Type Cul u e Collec ion
(ATCC) da abase, using he ATCC Ma ch algo i hm acco ding o ICLAC
s anda ds (h ps://www.lgcs anda ds-a cc.o g/en/STR_Da abase.aspx).
2.6. Long- e m p ima y cul u e o human pi ui a y adenomas unde hu-
manized (h7H) condi ions
P e iously [23,24], we designed a cell cul u e medium o human
ollicula hy oid cells, h7H, in which all componen s and addi i es
we e a concen a ions wi hin he human physiological anges (based
on e e ence alues o human se um om he Mayo Clinic whe e e
possible (h p://www.mayomedicallabo a o ies.com/ es -ca alog/)). In
cu en s udy, based in ou expe ience in sho - e m oden and
human p ima y pi ui a y cul u es [16,17,25,26], he h7H medium was
sligh ly modified om he ollicula hy oid medium: while insulin, so-
ma os a in and co isol we e main ained; bTSH and hGH we e e-
mo ed, and GHRH, Gh elin, EGF, bFGF, T3 and Glucagon we e added.
(Supplemen a y Table 6: h7H componen s a e lis ed in se en g oups
as osmolali y and ions, ho mones, g ow h ac o s, oligo elemen s, an i-
oxidan s and i amins, me aboli es and e hanol. The medium also in-
cluded 10% se um (5% FBS and 5% NBCS)).
The excised agmen o he adenoma su plus was diges ed in
1 mg/mL collagenase IV (C9891-AG, Sigma, Is ael) and 1× ypsin
(T4674, Sigma, USA) in PBS o 45 min, and hen fil e ed in o comple e
h7H medium (including se um, Supplemen a y Table 6) h ough a 41-
mic on fil e (NY4100010, Me ck Millipo e, I eland) o emo e undi-
ges ed issue agmen s. Cells we e washed se e al imes wi h comple e
h7H medium and hen seeded in one well o a 12-well TC- ea ed poly-
s y ene pla e (Cos a , The mo Scien ific, USA). E e y wo weeks, cells
we e ypsinized and dilu ed 1:2. We pe o med h7H cul u e in 11
NFPA y 4 ACRO, o which 8 NFPA and 3 ACRO g ew and we e main ained
o a leas 21–33 passages. A he las passage, cells s op g owing and
we e able o emain passi ely a ached o he pla e o weeks (looking
senescen ). Passages a ained by cul u es p esen ed in cu en wo k
we e: ACRO28 eached p21; ACRO30 eached p23; ACRO32 eached
p24; NFPA41 eached p33.
2.7. Analysis o sec e ed hGH and GDNF
Fo hGH sec e ion cul u e medium om pla es was collec ed be o e
passaging, a e h ee- ou days wi h he cells, and s o ed a −80 °C.
Immedia ely a e , cells we e ypsinized and coun ed o he nex pas-
sage, so he p ecise numbe o cells in he dish, and he p ecise ime o
incuba ion was known.
Clinical ki s o de e mine hGH in human se um can ha e nega i e in-
e e ences in samples o cul u e medium con aining bo inese um wi h
he e ophylic an ibodies and highe bio in concen a ions. This hap-
pened o us when we used he Immuli e 2000 (Siemens Heal hinee s,
E lange , Ge many) an IRMA ype assay using a abbi an i-hGH an i-
body conjuga ed o bo ine cal alkaline phospha ase and beads coa ed
wi h mu ine monoclonal an i-hGH an ibody; signal is de ec ed h ough
chemiluminiscence. This ki ha e been p e iously used o measu e hGH
egula ion in acu e cul u es o p ima y ACRO adenomas [27].In ou
hands, measu emen s using 25 mic oL o whole medium (incuding
10% bo ine se a) and his ki ga e unde ec able sec e ion in he
Immuli e Siemens analyze .
We decided o de ec sec e ed hGH by loading 15 mic oL o collec ed
medium on o a wes e n blo in compa ison wi h medium ha was no
pu in o he cells. A e a posi i e esul in quan i y, specifici y and
molecula weigh (Fig. 3), a second clinically alida ed assay was pe -
o med in he cul u e samples a e e ision o he li e a u e. The
COBAS hGH assay (Roche Diagnos ics, Mannheim, Ge many) had also
been p e iously used o measu e hGH in cul u e medium om pi ui a y
adenomas [28].COBAS is also an IRMA ype assay bu wi hou compo-
nen s o bo ine o igin. 40 mic oL o sample medium is incuba ed wi h
a bio in labelled monoclonal mu ine an i-hGH an ibody and a abbi
polyclonal an i-hGH an ibody conjuga ed o u henium chela e;
s ep a idin-magne ic beads on a magne ic elec ode pu i y he sand-
wiches, ollowed by an elec omagne ic impulse o ini ia e he ligh
emission by he u henium ha is de ec ed in he Elecsys E170 au o-
ma ic analyze .
Fo GDNF sec e ion, 10,000/well h7H-p ima y pi ui a y ac omegaly
(o NFPA) cells we e seeded in a 96-well dish, collec ed o di e en
ime-leng hs and kep a −80 °C. Sec e ed GDNF le els we e measu ed
wi h a Human GDNF DuoSe ELISA (DY212, R&D Sys ems, USA) acco d-
ing o he manu ac u e 's p o ocol. B iefly, a 98-well mic opla e was
coa ed wi h he cap u e an ibody and incuba ed o e nigh a oom em-
pe a u e. Wells we e blocked o 1 h a oom empe a u e and s an-
da ds and cul u e medium samples added and incuba ed o 2 h a
oom empe a u e. De ec ion an ibody was added du ing o 2 h a
oom empe a u e. A oiding exposu e o ligh , s ep a idin-HRP was
added o 2 min, ollowed by he subs a e solu ion o a u he
20 min. A e e mina ion o he enzyme eac ion wi h he s op solu ion,
he pla e was ead a A450/A550nm in a mic opla e eade (Mi h as LB
940, Be hold, Ge many).
2.8. Immunocy ochemis y
P ocedu es we e pe o med as desc ibed [24].B iefly, h7H-p ima y
pi ui a y ac omegaly cells we e seeded and g own on glass co e slips
in a 48-well dish. Wells we e washed and fixed wi h 10% neu al o ma-
lin o 30 min, washed once and hen incuba ed a oom empe a u e o
20 min in me hanol added a −20 °C. A e u he washing, pe me-
abiliza ion was ca ied ou using T i on 0.2% in H
2
Oonice o one
hou . P ima y and Seconda y An ibodies a e lis ed in Supplemen a y
Table 8.
2.9. Ra pi ui a y cell line GH4C1
The a GH4C1 soma o h oph cell line was cul u ed in DMEM
(D6046, Sigma, UK) con aining 10% FBS [17].Since no STR p ofile has
been s anda dized o such a cell lines, exp ession p ofiling was used
o assess pheno ype (see Supplemen a y Fig. 3 and p e ious e . [29]
o see cons ancy o pheno ype). T ans ec ion o RETL o RETS (Supple-
men a y Table 7) was pe o med using Nucleo ec o II (AAD-1001S,
Amaxa, Ge many) wi h he A-20 p og am and L-ki (VCA-1005, Lonza,
Ge many) as desc ibed [16].
540 M. Chenlo e al. / EBioMedicine 43 (2019) 537–552
2.10. E alua ion o an i-su i al ac ion o y osine kinase inhibi o s (TKIs)
ac i i y
TKIs (TINIB ools, Czeck Rep) s ocks we e p epa ed in DMSO and
kep a −20 °C. An essen ial s ep in ou assay was deciding which TKI
concen a ions es ed in i o we e ele an o hei in i o ac ion.
TKI concen a ions we e based on he Uni ed S a es Food and D ug Ad-
minis a ion (FDA) app o ed pha macological e iew o each d ug om
da a collec ed in Phase II-III clinical ials. We selec ed he mean concen-
a ion a s eady-s a e (Cm
SS
) de ec ed in ea ed pa ien s' se um
(Supplemen a y Table 4).
h7H-p ima y pi ui a y ac omegaly cells we e seeded (10,000/well
in mw48 o apop osis; 50,000/mw12 o RNA o p o ein ex ac s) in
ull h7H medium wi h 10% se um. Two days la e , cells we e washed
and he medium changed o a dep i ed medium (0·5% FBS wi hou
ho mones o addi i es). Human GDNF (100 o 500 ng/mL, s ock: 10
mic og/mL in PBS con aining 0·1%BSA; 345,872, Calbiochem, Is ael),
NRTN (100 ng/mL; s ock: 10 mic og/mL in PBS con aining 0.1%BSA;
1297-NE-025, R&D Sys ems, USA) o ehicle (PBS con aining 0·1%
BSA) was added o 24 h. A ull h7H medium (wi h se um and addi i es)
condi ion was included as a con ol o basal apop osis. App op ia e TKI
dilu ions o DMSO we e added o he co esponding ime wi h/wi hou
GDNF/NRTN.
Fo GH4C1 pi ui a y cells, 25,000 ans ec ed cells/well we e seeded
in a 48-well dish in ull medium. The ollowing day, cells we e washed
and cul u ed in DMEM+0·1% BSA o 48 h wi h/wi hou a GDNF
(50 ng/mL PF039-10ugmic og, Calbiochem, USA) and TKIs.
2.11. Apop osis assay
Medium con aining GDNF, TKIs o ehicles was added o 24 h
ollowed by addi ion o 5 mic oM Hoesch -33,258 (B1155, Sigma,
USA) o 1 h. Apop osis was measu ed by li e mic oscopy in an IX51
equipped wi h CellSens so wa e (Olympus). A leas six fields including
75–100 cells we e egis e ed pe well de ec ing condensed b illian blue
(apop o ic cells) and homogeneous so blue (li e cells).
2.12. 16-pad slide a ay
h7H-p ima y pi ui a y ac omegaly cells we e seeded and ea ed as
desc ibed abo e o one hou . Cells we e washed wi h ice-cold PBS and
lysed wi h 30 mic oL o ice-cold Cell Lysis bu e (#7018, Cell Signaling
Technology, USA). Mic oBCA P o ein Assay (23,235 The mo IL USA) was
used o measu e p o ein concen a ion. The 16-pad Pa hScan In acellu-
la Signaling a ay ki (#7323, Cell Signaling Technology, USA) was used
acco ding o he manu ac u e 's ins uc ions. In b ie , he a ay was
blocked using he a ay blocking bu e o 15 min a oom empe a u e.
60 mic og cell lysa es/100 mic oL A ay Diluen Bu e was hen added
and incuba ed o 2 h a oom empe a u e on an o bi al shake . A e
washing, De ec ion An ibody cock ail was added and incuba ed o 1 h
a oom empe a u e. A e washing, HRP-linked s ep a idin was
added and incuba ed o 30 min a oom empe a u e. A e washing,
he A ay glass slide was finally co e ed wi h 1× LumiGLO©/Pe oxide
solu ion and exposed o films in a linea ime-cou se. The a e age
pixel densi y o each ma ke was scanned and quan ified by Quan i y
One (Bio-Rad) and no malized o pixel densi ies o he posi i e con ol
spo s.
2.13. Cell ex ac s and Wes e n blo
A small agmen o ozen ac omegaly issue was mixed wi h 100
mic oL lysis bu e (0.05 M T is-HCl, 0.01 M EGTA, 1 mM EDTA,
16 mM T i on X-100, 1 mM sodium o ho anada e, 0.05 M sodium fluo-
ide, 0.01 M sodium py ophospha e and 0.25 M suc ose, adjus ed o 7·5
pH (all eagen s om Sigma, USA), wi h eshly added Comple e EDTA-
ee p o ease inhibi o cock ail able s (11,836,170,001 Roche
Diagnos ics, USA). A Pi -Mill issue lyse (85,300 Qiagen, Ge many)
was used o h ee minu es a 20 Hz. Lysa es we e cen i uged a
12000 pm o 30 min and supe na an s s o ed a −80 °C.
Cell lysa es ob ained o he abo e 16-pa h slide a ay we e consid-
e ed cy oplasmic. Fo ex ac ions ha included memb ane p o eins,
wells we e incuba ed o 20 min on ice wi h 30 mic oL o ice-cold T i on
lysis bu e (50 mM Hepes pH 7·5, 1% T i on X-100, 150 mM NaCl, 5 mM
EGTA, 1·5 mM MgCL, 10 mM Na py o os a e, 92 mic og/mL Na
3
VO
4
,2
mic og/mL ap o inin and 4 mM PMSF, all eagen s om Sigma, USA).
Fo o al cell ex ac s 15 mic oL o boiling 1% SDS/well we e added o
each well, which was hen sc aped and boiled a 95 °C o 5 min in a
he moblock, and finally dilu ed 1:3 in T i on lysis bu e [17].All lysa es
we e cen i uged a 12000 pm o 10 min and supe na an s s o ed a
−80 °C.
Mic oBCA P o ein Assay (23,235 The mo IL USA) was used o mea-
su e p o ein concen a ion. 15–20 mic og o p o ein ex ac s un in
10–15% SDS-PAGE we e elec o ans e ed o a PVDF memb ane
(Immun-Blo PVDF 162–0177; Bio-Rad, USA) wi h a semid y blo e .
Memb anes we e blocked wi h 0·2% T opix i-block (T2015, Applied
Biosys ems, USA) in 1× TBS o 2 h. An ibodies and dilu ions a e
shown in Supplemen a y Table 8. Enhanced chemiluminescence assay
(Pie ce ECL Wes e n Blo ing Subs a e, 32,106, The mo Scien ific,
USA) was used o signal de ec ion. Memb anes we e exposed o X-
ay film (4741 19,289, Fujifilm, Japan) and de eloped (De elope
G150, AGFA Heal hCa e, Belgium) and fixed (Manual Fixing G354,
AGFA Heal hCa e, Belgium). S a is ical analysis S a is ical analysis and
plo ing o g aphs we e pe o med using G aphPad P ism 6 (G aphPad
So wa e, USA), SPSS S a is ics 20 (IBM, USA) and Co el D aw G aphic
Sui e 2017. Analysis we e pe o med in an independen way by CVA’s
g oup and he Ma hema ics Medical Depa men a CHUS. G aphs
show mean±SEM. No mali y o each g oup o quan i a i e da a was
assessed using he Kolmogo o -Smino es wi h Dallal-Wilkinson-
Lillie o ‘p’ alue. The majo i y o he g oups we e non-pa ame ic.
Fo s a is ical compa isons we pe o med de ailed bi- a ian analysis.
As he ini ial s ep in analysis o he da a, co ela ions we e assessed
wi h Spea man’s Rho ( s) and significance. Quan i a i e a iables
we e compa ed by Mann-Whi ney es s when non-pa ame ic o un-
pai ed - es s when pa ame ic. Quali a i e a iables we e compa ed
wi h a Chi-squa e independence es , using he Pea son ‘p’ alue when
nN5 o he Fishe exac ‘p’ alue when nb5. ANOVA was used o com-
pa isons o mo e han wo g oups; o cul u e expe imen s, whe e
he e was one con ol wi h no ea men , Dunne ’s co ec ion was ap-
plied. Fo he mul i a ia e analysis ( esponse o fi s -line ea men
and ARF exp ession cu -o depending on GNAS mu a ion) we used
Chi-squa ed wi h mul i a ia e con ingency able analysis. To calcula e
he disc imina ing h eshold o ARF (and GFRA4) exp ession, a ROC
cu e was used.
3. Resul s
We analysed quan i a i e RET pa hway gene exp ession in a p o-
spec i e se ies o 32 spo adic ACRO su gical samples collec ed in
2010–2017, in ela ion o clinical cha ac e is ics, and exp ession o
known ACRO genes and he G-p o ein subuni α(GNAS) DNA mu a ion,
a ecu en mu a ion ound in spo adic ACRO (Supplemen a y Table 1
and Supplemen a y Fig. 1a). In summa y, 30 new pa ien s and wo un-
de going eope a ions we e ec ui ed om h ee hospi als (San iago de
Compos ela, Mad id and León), comp ising 66·6% emales, median age
45·6 yea s old and median ollow-up 4·2 yea s (Supplemen a y
Table 2). 25% o samples con ained mu a ed GNAS (Supplemen a y
Table 2 and 3). The clinical cha ac e is ics o ou se ies do no di e
om o he published se ies [30].The majo i y o pa ien s had a
mac oadenoma, wi h umo olume and diame e being in e sely e-
la ed o age bu posi i ely ela ed o GH and IGF1 le els, Knosp g ade
and in asion (Supplemen a y Table 2). Hal o he pa ien s (N= 16)
ha e ecei ed p e-su ge y he apy wi h gSSA, al hough no co ela ion
541M. Chenlo e al. / EBioMedicine 43 (2019) 537–552
o his wi h o he clinical a iables was ound (Supplemen a y Table 2).
P esence o GNAS mu a ion was no ela ed o any o he a iable. Suc-
cess o su ge y was co ela ed wi h smalle , less in asi e umo s ha
sec e ed less GH/IGF1. Resis ance o adju an he apy wi h gSSA was
jus significan ly g ea e in males han emales (p= 0·049) bu had
no o he significan co ela ion. An imp o ed esponse o combined
fi s -line ea men was obse ed in olde pa ien s (p = 0·049) and
smalle non-in asi e umo s bu had no o he significan co ela ion.
3.1. Molecula axonomy o he RET pa hway in ACRO
Quan i a i e RNA exp ession analysis, using TaqMan qRT-PCR, was
pe o med o he ollowing gene g oups: pi ui a y ho mones (GH iso-
o ms; PRL; POMC/ACTH; CG-alpha, he common alpha subuni o
FSH, LH and TSH; and he specific be a subuni s o he wo gonado o-
pins, FSHB and LHB), ou pa hways o in e es (RET ecep o as o al ex-
p ession RETN, long RETL and sho RETS iso o ms; i s ou co- ecep o s
GFRA1-2-3-4 and i s ou ligands GDNF, NRTN, ARTM and PSPN; PIT1
ansc ip ion ac o ; p14ARF iso o m om he CDKN2A gene (he e
called ARF); and p53), hypo halamic neu opep ide ecep o s (GHRHR
and soma os a in ecep o s SSTR2, SSTR5), o he pi ui a y ansc ip ion
ac o s (PROP1, SF1, TPIT), AIP as a gene implica ed in amilial pi ui a y
adenomas (FIPA) [2], and wo cell cycle genes p e iously s udied in
models o ac omegaly (CDKN1A/p21 and CDKN1B/p27) (see Supple-
men a y Fig. 1 o a comp ehensi e scheme o he genes s udied). We
compa ed all genes wi h a comme cial pool o pi ui a y poly-A mRNA
as a echnical con ol o each gene, o equalize measu emen e ficiency
om di e en pla es du ing he s udy.
Fi s , we iden ified a common p ofile o genes en iched in ACRO, o
allow us subsequen ly o e alua e indi idual samples. To assess en ich-
men , we compa ed he exp ession o each ACRO gene wi h a simila
p ospec i e se ies o Non-Func ioning Pi ui a y Adenomas (NFPA, n
= 63). As expec ed, GH (22KDa (GHTaq) and 20 KDa (GHSyb) iso-
o ms), GHRHR, SSTR2 and SSTR5 we e e y abundan in ACRO in com-
pa ison wi h NFPA (Fig. 1a). The e was no di e ence be ween ACRO and
NFPA in exp ession le els o o he non-soma o oph ho mones, such as
POMC (co ico oph ho mone), CG-alpha o LHB ( he be a subuni o
he LH, one o he wo gonado oph ho mones) (Supplemen a y
Fig. 1b).Howe e , FSHB ( he be a subuni o FSH, he o he gonado oph
ho mone) was significan ly en iched in NFPA (Supplemen a y Fig. 1b).
F om he RET ecep o downs eam, all genes in he RET pa hway
we e significan ly mo e highly exp essed in ACRO han in NFPA
(Fig. 1a). Especially abundan we e he ligand GDNF, he soma o oph
ansc ip ion ac o PIT1 and he umo supp esso ARF, while a smalle
di e ence was ound in p53, a ibu able o i s egula ion being a he
p o ein le el. O he h ee non-soma o oph ansc ip ion ac o s s ud-
ied, PROP1 p esen ed low exp ession in all umo s, al hough i was sig-
nifican ly highe in ACRO han in NFPA; as expec ed, SF1 was
significan ly highe in NFPA, while TPIT was poo ly exp essed in bo h
ACRO and NFPA, a ibu able o he ac ha i is cha ac e is ic o
co ico oph umo s (ACTH se ies) (Fig. 1a). AIP, a p o ein ela ed o a-
milial bu no spo adic ac omegaly [1], did no show di e ences in ex-
p ession be ween ACRO and NFPA (Fig. 1A). p21, a gene ac i a ed by
p53, was mo e abundan in ACRO (Fig. 1a), con a y o p27 ha did
no p esen di e ences (Supplemen a y Fig. 1b).
A e his ini ial cha ac e iza ion o he ACRO se ies, each sample
was indi idually e alua ed o i s le el o en ichmen o he mos
abundan soma o oph genes (GH, GHRH, PIT1) and deple ion o
non-soma o oph genes (TPIT, POMC, SF1, CG-alpha, FSHB, LHB) (Sup-
plemen a y Fig. 1c). PRL was conside ed a ‘neu al’gene as i can be
co-sec e ed wi h GH. All samples passed ou echnical con ol, al hough
wo samples, ACRO23 - a e-ope a ion ollowing adio he apy - and
ACRO39, exp essed some non-ACRO genes (Supplemen a y Fig. 1c).
Howe e , he le el o GH exp ession in hese wo samples was s ill
much highe han ha in NFPA and hus hey we e included. PRL was
co-exp essed wi h GH in h ee samples: ACRO19, 25 and 32. We
p oceeded wi h ou s udy in he confidence ha he issue om
which we had ex ac ed RNA was mainly ACRO.
Fig. 1b shows he RET pa hway in no mal soma o oph cells [15,16].
In he absence o he ligand GDNF, RET is a dependence ecep o : he in-
acellula egion is p ocessed by Caspase-3, gene a ing an in acellula
agmen IC-RET and ac i a ion o PKC del a (PKCd); his induces PIT1
o e exp ession, exceeding pe missi e le els o an al eady abundan
gene and inducing ARF exp ession. ARF binds o and blocks MDM2,
he p o ein ha ubiqui ina es p53 p io o i s des uc ion, leading o
p53 accumula ion and apop osis. I he ligand is p esen , GDNF induces
RET dime iza ion which inhibi s he apop o ic pa hway, ac i a ing AKT
and main aining egula ed PIT1 exp ession, which is su ficien o p o-
duce GH bu no o induce apop osis [15,16].This pa hway explains
he abundance o GDNF bo h in he no mal pi ui a y (con aining
50–60% o soma o oph cells) and o a g ea e ex en in ACRO (Fig. 1a).
RET has a long (RETL) and a sho (RETS) iso o m a ying in he
leng h o i s cy oplasma ic C- e minal ail, leading o di e en numbe
o y osines candida es o phospho yla ion and hence in e ac ion
wi h di e en subs a es (Fig. 1b, igh ). As s a ed abo e, each RET co-
ecep o binds a cha ac e is ic ligand (GDNF-GFRA1, NRTN-GFRA2,
ARTM-GFRA3, PSPN-GFRA4). (see STRING, h ps://s ing-db.o g/cgi/
ne wo k.pl? askId=Unm9pNIM Jex). I is well es ablished ha GDNF
ac i a es RET h ough GFRA1 [31,32].I has also been demons a ed
ha NRTN is able o ac i a e RET h ough GFRA2 mo e po en ly han
GDNF [33].Howe e , he e is also c oss-ac i a ion be ween ligands
and co- ecep o s, as shown by ans ec ion expe imen s and in p ima y
cul u e. Thus NRTN is able o ully ac i a e RET wi h equal po ency o
GDNF h ough GFRA1 in ans ec ion s udies al hough i does so sligh ly
less po en ly han GDNF in p ima y cells exp essing GFRA1 ecep o
(Fig. 1b) [33–36].ARTM and PSP a e less po en han GDNF o NRTN
bu s ill able o ac i a e RET h ough GFRA1 in cells exp essing he e-
cep o [35–37].
In he ACRO se ies s udied, exp ession o RETL and RETS iso o ms
we e simila as app oxima e hal es o o al RET exp ession (RETN)
(Fig. 1c). The co- ecep o mos abundan ly exp essed was GFRA1
ollowed by GFRA4 (Fig. 1c). This fi ed wi h he majo ligand exp essed
being GDNF, ollowed by NRTN and PSPN (Fig. 1d).
We calcula ed co ela ions be ween molecula a iables. Spea man
co ela ions we e used o in es iga e gene exp ession and GNAS mu a-
ion. Significan bu ele an co ela ions (Rho, s N0·36) a e shown in
Fig. 1e. Exp ession o GDNF was posi i ely co ela ed wi h NRTN and
ARTM ligands, and bo h RET iso o ms. Bo h GDNF and NRTN we e neg-
a i ely co ela ed wi h PIT1 bu posi i ely co ela ed wi h PROP1. PIT1
and PROP1 we e nega i ely co ela ed wi h each o he . PROP1 di ec ly
induces PIT1 exp ession du ing emb yonic de elopmen , and once
PIT1 is induced PROP1 is ep essed h ough a mechanism ha is
known no o be di ec ly ep essed by PIT1 [38,39].GFRA1 and SF1
we e also posi i ely co ela ed wi h PROP1. ARF exp ession was posi-
i ely co ela ed wi h GH, GHRHR, AIP, ARTN and TPIT. SSTR2 was
only co ela ed wi h GH exp ession. No co ela ions we e ound o
p53, which was no unexpec ed as p53 is egula ed mainly a he p o-
ein le el. All o he genes s udied (including p21 and p27) and GNAS
mu a ions we e ound no o be ele an ly co ela ed.
Molecula co ela ions we e somewha educed in significance
when he g oup o naï e samples (N= 16) was analysed sepa a ely
om he g oup wi h p e-su ge y ea men (N = 16) (Supplemen a y
Fig. 1d). This was expec ed as a consequence o hal ing he samples.
Howe e , he weakes co ela ions we e los including PIT1-GDNF,
PIT1-PROP1, PROP1-NRTN, ARF-AIP and GH-SSTR2 (Fig. 1e, yellow o
blue do s). When genes implica ed in hese los co ela ions we e
analysed o di e ences be ween bo h g oups we ound significan di -
e ences exclusi ely in PIT1 and GH exp ession ha is educed, pa icu-
la ly GH, in p e-su ge y ea ed samples (Fig. 1 ).
The mos ele an finding om hese da a was ha he GDNF-RET
su i al pa hway is s ongly exp essed in ac omegaly and ein o ced
by edundancy in exp ession o RET ligands and RET ecep o iso o ms,
542 M. Chenlo e al. / EBioMedicine 43 (2019) 537–552
543M. Chenlo e al. / EBioMedicine 43 (2019) 537–552
con olling PIT1 and inhibi ing he RET apop o ic pa hway. This
p omp ed us o u he explo e he RET pa hway in ela ion o clinical
ollow-up cha ac e is ics o he pa ien s.
3.2. ARF is a p ognos ic ac o o esponse o fi s -line combined ea men
(su ge y+ gSSA) in ac omegaly
We in es iga ed co ela ions be ween molecula and clinical a i-
ables, which we e di ided in o h ee g oups: diagnos ic, pa hologic
and p ognos ic/ ollow-up. In Fig. 2Asignifican and ele an ( s ≥0·5)
co ela ions a e shown.
O he analy ical diagnos ic a iables, only basal se um GH a diagno-
sis was significan ly co ela ed wi h GH, SSTR2, AIP and p53 RNA umo
exp ession, all o which we e abundan ly exp essed in ACRO (Fig. 2a
le ). P e-su ge y he apy was nega i ely co ela ed wi h PIT1 and GH
(Fig. 2a). This was expec ed since exp ession o hose wo genes we e
educed wi h he ea men (Fig. 1e).
Rega ding pa hologic a iables, wo genes showed impo an co e-
la ions, ARF and TPIT. ARF exp ession was nega i ely co ela ed wi h in-
asi eness, Knosp g ade and, impo an ly, wi h pos -su gical esidual
umo (Fig. 2a cen e ). TPIT exp ession was nega i ely co ela ed wi h
umo diame e / olume, in asi eness and Knosp g ade, bu posi i ely
co ela edwi hsu gicalcu e(Fig. 2a igh ).
In he g oup o p ognos ic a iables, esponse o gSSA du ing
ollow-up was s ongly and posi i ely co ela ed wi h ARF exp ession
( s N0·8), while GFRA4 was nega i ely co ela ed ( s N0·7). O he
genes less s ongly co ela ed i h esponse o gSSA we e GH, TPIT and
SSTR2 ( s 0·5–0·6). Since exp ession o he gene wi h he wo mos im-
po an co ela ions, ARF, was also ela ed o a a o able cou se o dis-
ease (nega i ely co ela ed wi h esidual umo and posi i ely
co ela ed wi h esponse o gSSA), we combined he wo p ognos ic
a iables in o a single ‘Good P ognosis’G oup 1, defined as esponsi e
o combined fi s -line he apy (su ge y + gSSA). When his g oup
was analysed o co ela ions, ARF exp ession showed he highes pos-
i i i y ( s N0·7) and significance, ollowed by TPIT ( s = 0·6), while
GFRA4 was nega i ely co ela ed ( s = 0·5).
Molecula -clinical co ela ions we e main ained, al hough wi h e-
duced p alues, when analyzing sepa a ely naï e and p e-su ge y he -
apy g oups (N= 16 each) (Supplemen a y Fig. 2a), wi h some
excep ions (Fig. 2a: yellow do s, los in p e-su ge y he apy; blue do s
los in naï e). Co ela ion be ween SSTR2 and basal GH a diagnosis,
and TPIT and olume did no exis in naï e samples. Co ela ions be-
ween ARF and in asi eness, and SSTR2 and Response o gSSA we e
los in he g oup wi h p e-su ge y he apy. All o he co ela ions, espe-
cially hose implica ing ARF and GFRA4 wi h esponse o 1s line he -
apy we e s ongly main ained in bo h g oups (Compa e Fig. 2aand
Supplemen a y Fig. 2a).
We hen compa ed G oup 1 wi h ‘Resis an ’G oup 0, defined as no
cu ed by su ge y and esis an o gSSA du ing ollow-up. ARF mRNA ex-
p ession alues we e much lowe in G oup 0 han G oup 1 (Mann
Whi ney es , pb0·0001, Fig. 2c le ). Simila esul s we e ob ained
when umo s no cu ed by su ge y we e analysed sepa a ely. ARF was
again mo e highly exp essed in samples esponding o gSSA han esis-
an samples (Fig. 2c igh ). Di e ences be ween ollow-up g oups o
he o he genes es ed (SSTR2, SSTR5, TPIT, SF1) ei he did no each
he ARF le els o significance o we e no significan (Supplemen a y
Fig. 2c-d). Mo eo e , esul s did no change o ARF when ACTB was
used as he con ol gene, ins ead o TBP o no malize RNA exp ession
(Supplemen a y Fig. 2e). We hen aimed o find a cu -o alue o
ARF exp ession which would classi y an ac omegaly as ei he Respon-
si e G oup 1 o Resis an G oup 0 a he ime o su ge y, by cons uc ing
a ROC cu e o ARF le els (Supplemen a y Fig. 2 ). We ound ha plac-
ing he cu o a ≥0·06 sepa a ed he g oups wi h 100% sensi i i y bu
85·7% specifici y. Placing he cu o a ≥0·1, howe e , sepa a ed g oups
wi h 96% o sensi i i y and 100% specifici y. Al hough hese alues we e
e y good as a clinical p edic o we wan ed o unde s and i he e we e
o he ac o s ha migh explain he blu ed sepa a ion o h ee samples
loca ed a cu -o le els. In G oup 1 he only alue in he ange 0·06–0·1
was in ac GNAS-mu a ed, while in G oup 0 alues in he same ange
we e non-GNAS-mu a ed. We did no find any co ela ions be ween
GNAS mu a ion and any molecula o clinical a iables s udied (Supple-
men a y Table 2). Mo eo e , while he median alues we e no di e en
be ween he GNAS-mu a ed and non-mu a ed g oups, he ange o ARF
exp ession in he mu a ed g oup was much smalle ( ange 0·042–0·64,
n= 8) han ha in he non-mu a ed g oup ( ange 0·003–1·08, n=24)
(Supplemen a y Fig. 2 g). The e o e, a lowe cu -o alue would be ex-
pec ed o he mu a ed GNAS samples. In acco dance wi h his, when
we sepa a ed G oup 0 and G oup 1 using a cu o o 0·1 o ARF exp es-
sion in non-mu a ed GNAS ACRO and 0·06 in mu a ed GNAS ACRO, and
pe o med a mul i a ian analysis using Chi-squa e es , we ound max-
imal significance (pb0·0000) (Fig. 2d op), wi h all samples showing
100% sensi i i y and specifici y (Fig. 2dlowe ).
gSSA include wo p esc ip ions, Oc eo ide LAR (Oc eo ide, n=5
pa ien s) and Lan eo ide Au ogel (Lan eo ide, n= 13 pa ien s). We sep-
a a ed pa ien s ecei ing each ea men and ound ha , again, ARF ex-
p ession was able o sepa a e G oup1 om G oup 0 o bo h ea men s
(Supplemen a y Fig. 2 h). We pe o med s a is ics in he Lan eo ide
g oup o pa ien s, and again di e ences in ARF le els we e clea ly signi -
ican be ween G oup 0 (Resis an ) and G oup 1 (Responsi e), in spi e o
hei educed numbe .
GFRA4 was he only nega i ely co ela ed gene du ing ollow-up
(Fig. 2b). When all samples we e aken in o accoun he e we e signifi-
can highe exp ession le els in G oup 0 han G oup 1, al hough he
gene was no a pe ec classifie (Fig. 2 le ). Selec ing samples ha
we e no cu ed wi h su ge y and ea men wi h gSSA (Resis an
G oup 0, Responsi e G oup 1), GFRA4 exp ession was again signifi-
can ly highe in G oup 0, wi h a cu o ≥0·1 (Fig. 2E igh ). The e was
a single sample in G oup 0 ha did no ollow his ule (yellow do =
ACRO36), which was om a eope a ion ollowing adio he apy in he
same pa ien as ACRO27. This could be because adio he apy o gSSA
Fig. 1. mRNA exp ession cha ac e izing ACROand he RET pa hway. a) Quan i a i e RNA exp ession compa ing mean ± SEM ACRO (n= 32, blue ba s) and NFPA (n=56–63, pink ba s). In
e e y sample, gene exp ession was no malized o a comme cial pool o pi ui a y poly-A mRNA ( echnical con ol; whi e ba s). Genes mos highly exp essed in ACRO we e cha ac e is ic o
unc ional pi ui a y soma o ophs, ela ed o GH sec e ion (GH
Taq
(22 KDa), GH
Syb
(22, 20 and 17 KDa)), o PIT1 and hypo halamic egula ion (GHRHR, SSTR2, SSTR5). Howe e , he RET
ecep o (RETN), i s ligand GDNF and genes in he RET pa hway egula ed a he RNA le el (ARF and PIT1) we e significan ly mo e abundan in ACRO han NFPA. p53 was sligh ly
up egula ed in ACRO. SF-1 was cha ac e is ic o NFPA, and T-PIT o ACTH-sec e ing adenomas (o ange ba ). (Mann-Whi ney es ). b) Ca oon ep esen ing iso o ms o he RET
ecep o , di e ing a he C- e minal ail (long RETL and sho RETS), i s ou ligands and i s GFR-alpha co- ecep o s (GFRA1–4). Al hough he e is p e e ence o a ligand o a co-
ecep o , c oss-in e ac ion exis s. In soma o ophs in he absence o ligand, RET is p ocessed by Caspase-3 gene a ing an in acellula agmen (IC-RET) ha igge s a cell-dea h
pa hway h ough o e exp ession o PIT1 gene inducing p14ARF exp ession, p53 accumula ion and apop osis. When he ligand is p esen , RET dime izes and ac i a es i s cy oplasmic
y osine kinase ac i i y leading o AKT phospho yla ion and su i al. c) In ACRO, bo h RETL and RETS a e exp essed in app oxima ely equal amoun s. Al hough all ou RET co-
ecep o s we e exp essed, he mos highly exp essed was GFRA1 (high a fini y o GDNF), ollowed by GFRA4. (ANOVA). d) GDNF was by a he mos highly exp essed ligand, wi h
NTRN and PSPN also abundan ly exp essed. (ANOVA). e) Significan co ela ions among all he genes s udied in ACRO (
s
N0·36) e ealed ha GDNF exp ession was significan ly and
posi i ely co ela ed wi h exp ession o RET iso o ms and he o he ligands. GDNF (and NRTN) we e nega i ely co ela ed wi h PIT1 exp ession bu posi i ely co ela ed wi h PROP1.
As expec ed, exp ession o bo h ansc ip ion ac o s PIT1 and PROP1 we e nega i ely co ela ed. PROP1 was also co ela ed wi h GFRA1 and SF1. ARF exp ession was significan ly
co ela ed wi h he soma o oph pheno ype (GH, GHRHR, AIP), T-PIT and he ligand ARTN. (Spea man es ). Weakes co ela ions we e los (pN.2) when he naï e g oup (N= 16)
was analysed sepa a ely om he g oup ecei ing p e-su ge y he apy (N = 16). Yellow do s: co ela ions los in p e-su ge y he apy g oup; Blue do s: co ela ions los in naï e
g oup. ) Genes implica ed in los co ela ions we e no di e en ially exp essed be ween bo h g oups excep o PIT1 and GH ha we e significan ly educed in he p e-su ge y
he apy g oup (all Mann-Whi ney es excep ARF -Tes ). (pb0·05; **, p b0·01; ***, p b0·001; ****, p b0·0001; 0·0000 means lowe p).
544 M. Chenlo e al. / EBioMedicine 43 (2019) 537–552
a ec ed exp ession o GFRA4, making i s alue as a p ognos ic ma ke
o esponse o gSSA limi ed o adenomas no cu ed by a fi s ope a ion.
To es whe he he esul s ob ained o gene exp ession we e e i-
fiable wi h p o ein exp ession we s udied ele an p o ein exp ession in
a sub-se o samples (Fig. 2 ). We compa ed one esis an sample
(ACRO27) wi h fi e esponsi e, ei he cu ed by su ge y (ACRO7, 8) o
esponsi e o gSSA he apy (ACRO9, 16 and 15). Only ACRO8 and
ACRO27 we e naï e samples ha had no been p e-su gically ea ed
wi h gSSA, a leas o a sho ime while wai ing o he su ge y (Sup-
plemen a y Table 1). As shown in Fig. 2g, ACRO27 con ained high le els
o RET p o ein bu negligible o low le els o PIT1, ARF and p53. Only
ACRO7, 8 and 27 had de ec able GFRA4 exp ession, bu he p o eins de-
ec ed we e o h ee di e en molecula weigh s. I has p e iously been
shown ha GFRA4 has a canonical p o ein GPI iso o m ha makes a GPI
link on he ex acellula aspec o he plasma memb ane and unc ions
as a RET co- ecep o , as well as a longe TM iso o m wi h a ansmem-
b ane domain, and a sho e sec e ed iso o m [40].Thus we could asce -
ain ha ACRO7 exp essed he sec e ed GFRA4 iso o m bu ACRO8 he
TM iso o m. Only ACRO27 p esen ed he canonical GPI iso o m able o
unc ion as a RET co- ecep o . Quan i a i e ARF p o ein exp ession, el-
a i e ei he o ACTB o GAPDH as loading con ols, was well co ela ed
o ARF mRNA le els (Fig. 2g bo om).
545M. Chenlo e al. / EBioMedicine 43 (2019) 537–552
[57] Casa -Bo o a O, Heck A, Schulz S, Nesland JM, Ramm-Pe e sen J, Lek a T, e al. Ex-
p ession o SSTR2a, bu no o SSTRs 1, 3, o 5 in soma o oph adenomas assessed
by monoclonal an ibodies was educed by oc eo ide and co ela ed wi h he
acu e and long- e m e ec s o oc eo ide. J Clin Endoc inol Me ab 2013;98:E1730–9.
[58] Ga o F, Feelde s RA, an de Pas R, K os JM, Waaije s M, Sp ij-Mooij D, e al. Immu-
no eac i i y sco e using an an i-ss 2A ecep o monoclonal an ibody s ongly p e-
dic s he biochemical esponse o adju an ea men wi h soma os a in analogs in
ac omegaly. J Clin Endoc inol Me ab 2013;98:E66–71.
[59] Du an-P ado M, Sa eanu A, Luque RM, Gahe e MD, G acia-Na a o F, Jaque P, e al.
A po en ial inhibi o y ole o he new unca ed a ian o soma os a in ecep o 5,
ss 5TMD4, in pi ui a y adenomas poo ly esponsi e o soma os a in analogs. J Clin
Endoc inol Me ab 2010;95:2497–502.
[60] Venegas-Mo eno E, Vazquez-Bo ego MC, Dios E, G os-He guido N, Flo es-Ma inez
A, Ri e o-Co és E, e al. Associa ion be ween dopamine and soma os a in ecep o
exp ession and pha macological esponse o soma os a in analogues in ac omegaly.
J Cell Mol Med 2018;22:1640–9.
[61] Theodo opoulou M, Tichomi owa MA, Sie e s C, Yassou idis A, A zbe ge T,
Houg and O, e al. Tumo ZAC1 exp ession is associa ed wi h he esponse o so-
ma os a in analog he apy in pa ien s wi h ac omegaly. In J Cance 2009;125:
2122–6.
[62] Ga o F, Bie masz NR, Feelde s RA, K os JM, Dogan F, an de Lely AJ, e al. L ow be a-
a es in exp ession co ela es wi h he esponsi eness o long- e m soma os a in
analog ea men in ac omegaly. Eu J Endoc inol 2016;174:651–62.
[63] Fougne SL, Lek a T, Bo o a OC, Hald JK, Bolle sle J, Be g JP. The exp ession o E-
cadhe in in soma o oph pi ui a y adenomas is ela ed o umo size, in asi eness,
and soma os a in analog esponse. J Clin Endoc inol Me ab 2010;95:2334–42.
[64] Denes J, Kasuki L, T i ellin G, Colli LM, Takiya CM, S iles CE, e al. Regula ion o a yl
hyd oca bon ecep o in e ac ing p o ein (AIP) p o ein exp ession by MiR-34a in
spo adic soma o opinomas. PLoS One 2015;10:e0117107.
[65] Kasuki L, Wildembe g LE, Ne o LV, Ma condes J, Takiya CM, Gadelha MR. Ki-67 is a
p edic o o ac omegaly con ol wi h oc eo ide LAR independen o SSTR2 s a us
and ela es o cy oke a in pa e n. Eu J Endoc inol 2013;169:217–23.
[66] Ross FA, Hawley SA, Auciello FR, Gowans GJ, A ih A, Lamon DJ, e al. Mechanisms o
pa adoxical ac i a ion o AMPK by he kinase inhibi o s SU6656 and So a enib. Cell
Chem Biol 2017;24:813–824.e4.
[67] Fuma ola C, Ca a a C, La Monica S, Gale i M, Alfie i RR, Ca azzoni A, e al. E ec s o
so a enib on ene gy me abolism in b eas cance cells: ole o AMPK-mTORC1 sig-
naling. B eas Cance Res T ea 2013;141:67–78.
[68] P ie o-Dominguez N, O donez R, Fe nandez A, Ga cía-Palomo A, Mun ané J,
González-Gallego J, e al. Modula ion o au ophagy by So a enib: e ec s on ea -
men esponse. F on Pha macol 2016;7:151.
[69] Ga cia-La andei a M, Que eda V, Flo es I, Saez C, Diaz-Rod iguez E, Japon MA, e al. A
GRFa2/P op1/s em (GPS) cell niche in he pi ui a y. PLoS One 2009;4:e4815.
[70] So nson MW, Wu W, Dasen JS, Flynn SE, No man DJ, O'Connell SM, e al. Pi ui a y
lineage de e mina ion by he p ophe o pi -1 homeodomain ac o de ec i e in
Ames dwa fism. Na u e 1996;384:327–33.
[71] Lindahl M, Po e yae D, Yu L, A umae U, Timmusk T, Bonga zone I, e al. Human glial
cell line-de i ed neu o ophic ac o ecep o alpha 4 is he ecep o o pe sephin
and is p edominan ly exp essed in no mal and malignan hy oid medulla y cells.
J Biol Chem 2001;276:9344–51.
552 M. Chenlo e al. / EBioMedicine 43 (2019) 537–552
