1
Scien i ic RepoR s | 6:39378 | DOI: 10.1038/s ep39378
www.na u e.com/scien i ic epo s
An IFIH1 gene polymo phism
associa ed wi h isk o
au oimmuni y egula es canonical
an i i al de ence pa hways in
Coxsackie i us in ec ed human
panc ea ic isle s
E na Domsgen1,*, Ka ha ina Lind1,*, Lingjia Kong2, Michael H. Hühn1,†, Omid Rasool2,
F ank an Kuppe eld3, Olle Ko sg en4, Rii a Lahesmaa2 & Malin Flods öm-Tullbe g1,5
The IFIH1 gene encodes he pa e n ecogni ion ecep o MDA5. A common polymo phism in IFIH1
( s1990760, A946T) con e s inc eased isk o au oimmune disease, including ype 1-diabe es (T1D).
Coxsackie i us in ec ions a e linked o T1D and cause be a-cell damage in i o. He e we demons a e
ha he s1990760 polymo phism egula es he in e e on (IFN) signa u e exp essed by human
panc ea ic isle s ollowing Coxsackie i us in ec ion. A s ong IFN signa u e was associa ed wi h high
exp ession o IFNλ1 and IFNλ2, linking s1990760 o he exp ession o ype III IFNs. In he high-
esponding geno ype, IRF-1 exp ession co ela ed wi h ha o ype III IFN, sugges ing a posi i e-
eedback on ype III IFN ansc ip ion. In summa y, ou s udy unco e s an in luence o s1990760 on he
canonical e ec o unc ion o MDA5 in esponse o an acu e in ec ion o p ima y human pa enchymal
cells wi h a clinically ele an i us linked o human T1D. I also highligh s a p e iously un ecognized
connec ion be ween he s1990760 polymo phism and he exp ession le el o ype III IFNs.
The in e e on-induced helicase-1 (IFIH1) gene encodes he melanoma di e en ia ion associa ed p o ein 5
(MDA5), an in acellula pa e n ecogni ion ecep o (PRR) o impo ance o he ecogni ion o ce ain i uses.
The MDA5 p o ein consis s o wo N- e minal caspase ec ui men domains (CARDs), a cen al helicase domain,
and a C- e minal domain (CTD). MDA5 monome s bind o long double s anded (ds)RNA molecules o med
du ing he i al eplica ion cycle. This ini ia es he o ma ion o a MDA5 ilamen along he dsRNA, whe e mul-
iple CARD domains assemble and ec ui he signalling adap o in e e on p omo e s imula o 1 (IPS1) (also
deno ed mi ochond ial an i i al-signalling p o ein, MAVS). Ac i a ed IPS1/MAVS ini ia es he downs eam sig-
nalling e en s leading o he ac i a ion o he ansc ip ion ac o s NF-κ B and IRF-3, ollowed by he p oduc ion
o ype I and III in e e ons (IFN)1–4.
The apid p oduc ion o IFNs is an essen ial componen o he immune esponse agains many i uses. Type
I and III IFNs ha e po en an i i al ac i i ies; hey ac in an au o- and pa ac ine manne o induce he exp es-
sion o in e e on-s imula ed genes (ISGs) leading o an an i i al s a e in unin ec ed cells. They also p omo e
inna e and adap i e immune esponses1–5. Ac i a ion o ype I and III IFN gene ansc ip ion is dependen on he
1The Cen e o In ec ious Medicine, Depa men o Medicine HS, Ka olinska Ins i u e , Ka olinska Uni e si y
Hospi al, S ockholm, 141 86, Sweden. 2Tu ku Cen e o Bio echnology, Uni e si y o Tu ku and Åbo Akademi
Uni e si y, Tu ku, 205 20, Finland. 3Vi ology Di ision, Depa men o In ec ious Diseases and Immunology, Facul y
o Ve e ina y Medicine, U ech Uni e si y, U ech , 3584, The Ne he lands. 4Depa men o Immunology, Gene ics,
and Pa hology, Rudbeck Labo a o y, Uppsala Uni e si y, Uppsala, 751 05, Sweden. 5Ins i u e o Biosciences and
Medical Technologies, Uni e si y o Tampe e, Tampe e, 33520, Finland. †P esen add ess: As a Zeneca AB R&D,
Peppa edsleden 1, Mölndal, 431 50, Sweden. *These au ho s con ibu ed equally o his wo k. Co espondence and
eques s o ma e ials should be add essed o M.F.-T. (email: [email p o ec ed])
Recei ed: 12 Augus 2016
Accep ed: 23 No embe 2016
Published: 21 Decembe 2016
OPEN
www.na u e.com/scien i ic epo s/
2
Scien i ic RepoR s | 6:39378 | DOI: 10.1038/s ep39378
Figu e 1. RNAseq analysis iden i ies di e ences in gene exp ession be ween CVB3 in ec ed panc ea ic
isle s ca ying he s1990760 TT and TC geno ypes. (a and b) Human isle s om 23 dono s we e in ec ed in
mock o wi h CVB3. Isle s and supe na an s we e collec ed a 48 h p.i. The i e s o eplica ing i us pa icles
in supe na an s om in ec ed isle s (a) we e measu ed by a s anda d plaque assay and p esen ed as Log10
(PFU/ml). The exp ession le els o i us RNA (b) in CVB3 in ec ed isle s we e measu ed using quan i a i e
eal- ime PCR. The exp ession le els we e no malized o ha o GAPDH and p esen ed as 2−(ΔC ) ( ela i e
www.na u e.com/scien i ic epo s/
3
Scien i ic RepoR s | 6:39378 | DOI: 10.1038/s ep39378
ac i a ion o speci ic ansc ip ion ac o s, NF-κ B and IRF-3. Besides hese wo, he ansc ip ion ac o IRF-7 is
equi ed o he ansc ip ional ac i a ion o IFNα and possibly also IFN-λ 2 and -3 genes. IRF-7 is cons i u i ely
exp essed by some immune cells, bu a ely by pa enchymal cells. IRF-7 is o en induced ollowing exposu e o
JAK/STAT-ac i a ing cy okines, such as he IFNs hemsel es, hus es ablishing a posi i e eedback loop o IFN
exp ession. IRF-1 is ano he IFN inducible ansc ip ion ac o . I has binding si es in he ype I and III IFN p o-
mo e s and can also con ibu e o a posi i e eedback on IFN gene ansc ip ion. Recen s udies ha e sugges ed
ha IRF-1 has a ole in ac i a ing ype III and no ype I IFNs du ing in ec ions wi h RNA i uses3,6,7.
The human IFIH1 gene con ains nume ous single nucleo ide polymo phisms (SNPs). The non-synonymous
SNP (nsSNP), s1990760 (A946T), is associa ed wi h an inc eased isk o de elopmen o se e al au oimmune
diseases including T1D diabe es, sys emic lupus e y hema osus and mul iple scle osis (e.g.8–10). Li le is known
on how his nsSNP egula es isk o de elopmen o hese in lamma o y diseases, and he e o e in o ma ion on
he biological e ec s o his nsSNP may p o ide use ul insigh s.
P e ious s udies using ec opic o e exp ession o MDA5 in cell lines ha e eached con as ing conclusions wi h
ega ds o he impac o he s1990760 nsSNP on he ac i i ies o MDA5. Some s udies epo ed no in luence on
unc ion11–13, while ano he s udy sugges ed ha he isk allele T (946 T) encodes a cons i u i ely ac i e MDA5
p o ein ha is unable o espond o a na u al ligand (encephalomyoca di is i us, EMCV14). These con as ing
esul s illus a e ha addi ional s udies a e equi ed o e eal i and how he s1990760 a ec s he unc ion o
MDA5.
Recen s udies ha e shown ha he unc ional impac o disease-associa ed SNPs may a y depending on he
cell ype s udied (e.g.15). This, oge he wi h he ecen no ion ha he an i i al e ec o mechanisms o MDA5
may a y in esponse o di e en ypes o i uses14,16,17, sugges ha s udies using ele an human p ima y cells
and pe inen na u al ligands (i.e. clinically ele an i uses a he han i al mimics such as poly I:C) a e needed
o gain a be e unde s anding o he unc ion o he disease-associa ed a ian s o IFIH1.
En e o i us in ec ions, in pa icula ly hose wi h Coxsackie i us (CVBs), ha e been associa ed wi h he
de elopmen o T1D in humans. The link be ween CVB in ec ions and T1D is suppo ed by case epo s, epi-
demiological da a and by obse a ions indica ing ha en e o i uses a e p esen in he emaining be a cells o
indi iduals wi h T1D diabe es18–25. We ha e shown ha human panc ea ic isle s exp ess MDA526, and oge he
wi h o he s demons a ed ha MDA5 is impo an in he hos immune esponse o CVBs27,28. Human panc ea ic
isle s a e easily and ep oducibly in ec ed by CVBs in i o (e.g.25,26,29). This makes he isle s a ele an and use ul
model o s udy a p ima y cell esponse o a CVB in ec ion and o assess whe he he s1990760 nsSNP a ec his
esponse. In he p esen s udy, we used human panc ea ic isle s (p ima y cells) and a i us o clinical ele ance
(Coxsackie i us) as na u al ligand, o p o ide a be e unde s anding o he unc ional consequences o his pol-
ymo phism. Ou s udy e eals p e iously no desc ibed ela ionships be ween he s1990760 polymo phism and
he magni ude o he inna e esponse o Coxsackie i us in ec ion.
Resul s
Whole ansc ip ome analysis iden i ies an IFN signa u e in CVB3 in ec ed isle s. In ec ions
wi h CVBs a e associa ed wi h human diseases, including asep ic meningi is, myoca di is and au oimmune
T1D. The se o ype B3 (CVB3) eplica es in human panc ea ic isle s, wi h a peak a ound 48–72 h pos in ec ion
(p.i.) (26,29, da a no shown). Mo eo e , obus changes in gene ansc ip ion can be measu ed a 48 h p. i. (29;
Supplemen a yFig.1). He e, panc ea ic isle s om a o al o 23 human dono s we e in ec ed wi h CVB3 o 48 h.
The isle s we e highly pe missi e o in ec ion as shown by he measu emen s o high i e s o in ec ious i us
pa icles eleased in o he issue cul u e media (Fig.1a) and concomi an exp ession o i al RNA in he in ec ed
isle s (Fig.1b).
Geno yping o s1990760 was pe o med on DNA isola ed om each indi idual isle ba ch. This analysis
demons a ed ha he isle coho consis ed o a mix u e o dono s ca ying ei he he homozygous T1D dia-
be es isk geno ype, TT, o TC, a geno ype associa ed wi h p o ec ion om disease10. None o he dono s ca -
ied he p o ec i e geno ype CC (Fig.1c). In pa allel, he isle coho was geno yped o he s3747517 SNP
(Supplemen a yTableS1). The esul s om his geno yping we e in line wi h he p e iously published s ong
linkage disequilib ium wi h s19907609,16,30,31.
exp ession). Da a is shown as mean ± SEM, ****p < 0.0001, Wilcoxons ma ched-pai s signed ank es . n.d.,
no de ec ed. (c) Human isle dono s included in he s udy we e geno yped o he polymo phisms s1990760
(A946T). Allele dis ibu ion is shown in numbe s and as pe cen age o he whole coho (n = 23). (d) Pie
cha showing he en ichmen o p ocess ne wo ks based on he 166 DEGs iden i ied by RNAseq. Da a was
gene a ed using he Me aCo eTM en ichmen analysis ool wi h a cu o p- alue o 0.05. The con en o he
p ocess ne wo ks is de ined and anno a ed by Thomson Reu e s scien is s. Each p ocess ep esen s a p e-se
ne wo k o in e ac ions cha ac e is ic o he p ocess. (e) Lis o 10 genes displaying he highes log2 old change
in exp ession in in ec ed human isle s e sus unin ec ed, as iden i ied by RNAseq. ( ) Pie cha showing he
numbe and pe cen age o he 166 DEGs iden i ied as ISGs o non-ISGs based on he In e e ome da abase.
(g) Venn diag ams showing he numbe o ISGs (uppe diag am, yellow) and non-ISGs (including IFNs, lowe
diag am, ed) commonly o di e en ially exp essed in he wo s1990760 geno ype popula ions (TT and TCs).
(h) Hea map o di e en ially exp essed ISGs (n = 134). Log2 (CPM) alues we e mean-cen ed and scaled o
each gene (whi e, low exp ession di e ence; blue high exp ession di e ence). The Pea son co ela ion was used
o compu e dis ances be ween genes and samples, and he clus e ing was pe o med using a e age linkage.
Each column co esponds o one isle dono wi h indica ed s1990760 geno ype and each ow co esponds o a
speci ic gene.
www.na u e.com/scien i ic epo s/
4
Scien i ic RepoR s | 6:39378 | DOI: 10.1038/s ep39378
To add ess whe he he s1990760 nsSNP has an in luence on he human isle esponse o CVB3 in ec ion,
isle s om eigh o he 23 dono s (TT, n = 4; TC, n = 4) we e subjec ed o whole- ansc ip ome analysis (RNAseq).
Using a s ingen s a is ical me hod (see Me hods) his analysis iden i ied a o al o 166 mRNA ansc ip s ha
we e signi ican ly al e ed in in ec ed compa ed o unin ec ed isle s (FDR < 0.05, Fig.1d, Supplemen a yTableS2).
Func ional anno a ion o he di e en ially exp essed coding genes (DEGs) using Me aCo eTM (Thomson Reu e s)
sugges ed signi ican en ichmen o biological pa hways ela ed o in lamma ion, inna e immuni y and esponse
o i al in ec ion (Fig.1d). Wi h he same so wa e, we iden i ied se e al ansc ip ion ac o s c i ical o obus
an i i al esponses (e.g. IRF-1, STAT-1 and -2), as well as bo h ype I (IFNβ ) and III IFN (IFNλ 1–3) genes (Fig.1e,
and Supplemen a yTableS2). We also obse ed an inc eased exp ession o genes encoding nega i e egula o s
Figu e 2. CVB3 in ec ion o human panc ea ic isle s esul s in he up egula ed exp ession o ype I and
III IFNs and ISGs. (a–c,e and ) mRNA exp ession le els o IFNβ (a), IFNλ 1 and IFNλ 2 (b), PRRs MDA5,
RIG-I and TLR3 (c), MxA (e) and CXCL10 ( ) we e measu ed a 48 h p.i. in he mock- o CVB3 in ec ed isle s
desc ibed in Fig.1 (n = 23) using quan i a i e eal- ime PCR. The exp ession le els we e no malized o ha
o GAPDH and p esen ed as 2−(ΔC ) ( ela i e exp ession) o all isle s. (d) P o ein was isola ed om a o al o 6
isle dono s. A 48 h p.i. he p o ein exp ession le els o MDA5 (3 dono s) and RIG-I (3 dono s) we e measu ed
in mock and CVB3 in ec ed isle s using wes e n blo . Ac in was used as loading con ol. Da a om one dono
is shown o each p o ein. The displayed igu es a e c opped images. Images o unc opped blo s can be iewed
in Supplemen a yFig.S2. (g) The amoun o sec e ed CXCL10 p o ein was measu ed in supe na an ha es ed
om mock-in ec ed and CVB3 in ec ed isle s (n = 23) using ELISA and exp essed as ng/ng RNA. Da a is shown
as mean ± SEM, ****p < 0.0001, Wilcoxons ma ched-pai s signed ank es . n.d., no de ec ed.
www.na u e.com/scien i ic epo s/
5
Scien i ic RepoR s | 6:39378 | DOI: 10.1038/s ep39378
o IFN exp ession and signalling (e.g. USBP18, TRIM21, GBP4, Supplemen a yTableS2)5. Because we no ed an
up egula ed exp ession o ype I and III IFNs, which a e known o induce a wide ange o genes (so-called ISGs),
we used he In e e ome 2.0 da abase32 o iden i y how many o he DEGs we e ISGs. This analysis e ealed ha
he majo i y (80.7%) o he 166 DEGs we e ISGs (Fig.1 ).
The magni ude o he human isle immune esponse o CVB3 in ec ion is linked o he
s1990760 geno ype. Mos o he DEGs iden i ied in he RNAseq analysis (97%) we e sha ed be ween
dono s ca ying he TT and TC geno ypes, wi h only 4 DEGs unique o dono s wi h he TC geno ype (Fig.1g).
Hie a chical clus e ing based on he 134 genes iden i ied as ISGs dis inguished wo majo clus e s, in which he
majo i y (3/4) o dono s ca ying he TT geno ype we e placed in a clus e wi h low ela i e gene exp ession
le els. The majo i y o isle s ca ying he TC geno ype (3/4) we e clus e ed oge he based on high ela i e gene
exp ession (Fig.1h). Taken oge he , he ansc ip ome analysis demons a ed ha CVB3 in ec ion induces a
obus immune esponse in human panc ea ic isle s. These s udies also indica ed ha he s eng h o his esponse
migh be egula ed by he s1990760 geno ype.
The s1990760 TC geno ype is associa ed wi h a high ype III IFN esponse and a s ong IFN
signa u e. We nex alida ed some o he esul s om he ansc ip ome analysis using con en ional qPCR.
By analysing cDNA om he whole isle coho (n = 23) we e i ied ha ype I (IFNβ ) and III IFNs (IFNλ 1 and
IFNλ) showed inc eased exp ession ollowing CVB3 in ec ion (Fig.2a and b). Simila ly, he exp ession le els o a
numbe o he iden i ied DEGs including PRRs (e.g. he IFIH1/MDA5 gene i sel and DDX58/RIG-I), and genes
encoding p o eins in ol ed in cellula an i i al de ence (e.g. Mx1/MXA) o ec ui men o lymphocy es and wi h
ele ance o T1D (CXCL1033) (Fig.2c,e, ) we e ele a ed. Wes e n blo o ELISA measu emen s e ealed ha he
induced gene exp ession ansla ed in o inc eased p o ein exp ession (Fig.2d) and sec e ion (Fig.2g).
We a emp ed o measu e sec e ed ype I and III IFNs, bu ou pilo s udies using an ELISA and a sensi i e
bioassay ailed o de ec IFNs in he cul u e media o bo h con ol and CVB3 in ec ed isle s (da a no shown).
This is likely due o ha en e o i uses, including CVB3, ha e de eloped mechanisms o impai ype I and III IFN
p oduc ion and elease34,35, and ha he sec e ed IFNs a e locally consumed due o binding o ecep o s exp essed
by he cul u ed cells.
When s a i ying ou dono s acco ding o he s1990760 geno ype no s a is ically signi ican di e ence was
de ec ed in he exp ession o IFNβ mRNA be ween he isle s ca ying he TT and TC geno ypes (Fig.3a). In
con as , isle s om dono s ca ying he TC geno ype exp essed signi ican ly highe le els o ype III IFNs when
compa ed o isle s ca ying he TT geno ype (p < 0.05; Fig.3b). Simila ly, TC isle s exp essed highe le els o he
s udied genes (ISGs), and he di e ence was obse ed also when compa ing he old induc ion (i.e. Δ Δ C - alues)
(p < 0.05 and p < 0.01; Fig.3c–g). In line wi h he gene exp ession da a, he sec e ion o he chemokine CXCL10
was also signi ican ly highe in isle s om dono s wi h he TC geno ype compa ed o hose wi h he TT geno ype
(Fig.3h). In con as , he exp ession le els o he s udied genes did no di e be ween he mock-in ec ed isle s
o he TT and TC geno ypes Fig.3c–g). The i e s o i us accumula ing in he media du ing he whole 48 h cul-
u e pe iod, as well as he exp ession o i al RNA a 48 h p.i., we e no s a is ically di e en be ween he dono s,
al hough he la e was close o eaching signi icance (p = 0.09, Fig.3i). Collec i ely, his shows ha he human
isle s ca ying he s1990760 TC geno ype ha e a highe ype III IFN and ISG esponse o CVB3 han isle s ca -
ying he s1990760 TT ( isk) geno ype.
A high ype III IFN esponse o CVB3 in ec ion in TC dono s is associa ed wi h a s ong exp es-
sion o IRF-1. IFN exp ession is egula ed by he ac i i ies o ansc ip ion ac o s as well as nega i e eg-
ula o s3,36,37. Ou ansc ip ome analyses iden i ied se e al ansc ip ion ac o s as being inducible by CVB3
in ec ion (Fig.4a). A ew o hese showed signi ican ly di e en exp ession le els be ween he TT and TC gen-
o ypes (BATF2 and STAT1) (Fig.4a). In addi ion o his, he exp ession o IRF-1 was signi ican ly inc eased
only in in ec ed isle s wi h he TC geno ype (p < 0.01; Fig.4a). Fou nega i e egula o s o ype 1 IFN signalling,
IFI35, IFIT1, OASL, USP185, showed a di e en le el o exp ession in he wo geno ypes (Fig.4b). All ou we e
exp essed a a highe le el in he TC geno ype (high esponde s) as compa ed o he TT geno ype (low espond-
e s). Thus, i is unlikely ha hese genes con ibu ed o a highe exp ession o ype III IFNs in isle s om dono s
wi h he TC geno ype pe se, and hese ou genes we e he e o e no u he s udied.
The ansc ip ion ac o IRF-1 has been shown o speci ically egula e ype III and no ype I IFN exp ession
du ing RNA i us in ec ions3,37. Thus, we nex analysed he exp ession o IRF-1 in he whole isle coho by
eal- ime PCR and con i med ha CVB3 in ec ion induces he exp ession o IRF-1 (Fig.5a). We ound no di e -
ence in he basal exp ession le els o IRF-1 when con ol isle s om TT and TC dono s we e compa ed (Fig.5a).
A e CVB3 in ec ion, isle s wi h he TC geno ype exp essed highe le els o IRF-1 mRNA compa ed o isle s wi h
he TT geno ype (Fig.5a). In ec ed isle s wi h he TC geno ype also had a s onge inc ease in hei IRF-1 mRNA
exp ession ela i e o unin ec ed con ol isle s (i.e. “ old induc ion”) as compa ed o isle s wi h he TT geno ype
(Fig.5a).
We nex in es iga ed whe he he e was a co ela ion be ween he le els o IRF-1 and ype I and III IFN
mRNA exp ession. When analysed o he en i e human isle coho he le els o exp essed IRF-1 mRNA co -
ela ed wi h he exp essed le els o IFNβ , IFNλ 1 and IFNλ 2 (Fig.5b). Howe e , when analysing he TT and TC
dono s sepa a ely i was ound ha he le els o IRF-1 mRNA co ela ed wi h he le els o he ype III IFNs (IFNλ
1 and IFNλ 2) only in TC and no TT dono s (Fig.5c,d). This sugges s ha highe IRF-1 exp ession in he TC
dono s con ibu es o he s onge exp ession o ype III IFNs ia a eedback mechanism (Fig.6).

www.na u e.com/scien i ic epo s/
6
Scien i ic RepoR s | 6:39378 | DOI: 10.1038/s ep39378
Figu e 3. Homozygosi y o he T (Th 946) allele o he s1990760 polymo phism is associa ed wi h
low human isle ype III IFN and ISG mRNA exp ession le els, as well as amoun s o sec e ed CXCL10,
ollowing CVB3 in ec ion. (a and b) IFNβ (a), IFNλ 1 and IFNλ 2 (b) mRNA exp ession le els measu ed a 48 h
p.i. in he CVB3 in ec ed isle s desc ibed in Fig.2 (n = 23) we e e-analysed and he dono s s a i ied acco ding
o he s1990760 geno ype (TT o TC). The exp ession le els we e no malized o ha o GAPDH and p esen ed
as 2−(ΔC ) ( ela i e exp ession) o all isle s. (c–g) mRNA exp ession le els o PRRs (MDA5, RIG-I, TLR3; (c–e),
genes in ol ed in an i i al de ence (MxA; ( ) o chemoa ac ion (CXCL10; (g), measu ed a 48 h p.i. in he
mock-in ec ed (le column, C) and CVB3 in ec ed ( wo igh columns, CVB3) isle s desc ibed in Fig.2 (n = 23)
we e e-analysed by di iding he dono s acco ding o he s1990760 geno ype (TT o TC). The exp ession le els
we e no malized o ha o GAPDH and p esen ed as 2−(ΔC ) ( ela i e exp ession; he wo le columns), o
no malized o GAPDH and compa ed o he exp ession le els in mock-in ec ed isle s om he same dono and
p esen ed as 2−(ΔΔC ) ( old induc ion: he igh column). Fold induc ion calcula ion was no applicable o IFNβ ,
IFNλ 1–2 and CXCL10 mRNA, which we e unde ec able in mock-in ec ed isle s). (h) The amoun o sec e ed
CXCL10 p o ein (see Fig.2g) was measu ed in supe na an ha es ed om mock-in ec ed (le column, C) and
CVB3 in ec ed ( igh column, CVB3) isle s using ELISA and exp essed as ng/ng isola ed RNA. The isle dono s
we e s a i ied acco ding o s1990760 geno ype TT o TC. (i) The da a desc ibed in Fig.2 was e-analysed
and he isle s we e di ided acco ding o he s1990760 geno ype (TT o TC). Ti e s o eplica ing i us in
supe na an s ha es ed a 48 h p.i. a e p esen ed as Log10(PFU/ml). The exp ession le els o i us RNA in CVB3
in ec ed isle s we e measu ed using quan i a i e eal- ime PCR. The exp ession le els we e no malized o ha
o GAPDH and p esen ed as 2−(ΔC ) ( ela i e exp ession). Da a is shown as mean ± SEM, *p < 0.05, **p < 0.01,
***p < 0.001, Mann-Whi ney es .
www.na u e.com/scien i ic epo s/
7
Scien i ic RepoR s | 6:39378 | DOI: 10.1038/s ep39378
The exp ession o T1D- ela ed genes does no di e be ween dono s wi h he s1990760 TT
and TC geno ypes. Risk o T1D de elopmen has been linked o polymo phisms in nume ous genes
besides IFIH1. Many o hese genes a e exp essed by panc ea ic be a cells and/o isle s38. Using ou gene an-
sc ip ion da a (Supplemen a yTableS1) and he T1Dbase.o g we iden i ied some o hese genes as ha ing an
al e ed exp ession le el ollowing CVB3 in ec ion, including he up egula ed exp ession o TAGAP, CD69 and
Figu e 4. The gene exp ession o se e al ansc ip ion ac o s and genes in ol ed in nega i e egula ion
o ype 1 IFN signalling is inc eased in human isle s upon CVB3 in ec ion. RNA sequencing was pe o med
on RNA isola ed om non-in ec ed and CVB3 in ec ed p ima y human isle s ha es ed a 48 h p.i. (8 dono s in
o al; 4 TT and 4 TC) Among he genes demons a ing a di e en ial exp ession be ween con ol and in ec ed
isle s, 10 we e iden i ied as ansc ip ion ac o s and 8 ha e been desc ibed as nega i e egula o s o ype I
IFN signalling5. The mRNA abundance is shown as log2(CPM) o iden i ied ansc ip ion ac o s (a) and
egula o s o IFN signalling (b) om con ol and CVB3 in ec ed isle s ca ying he s1990760 TT (TT con ol,
TT in ec ed) o TC geno ype (TC con ol, TC in ec ed). Da a is shown as mean ± SEM, *p < 0.05, **p < 0.01,
***p < 0.001, ****p < 0.0001, one-way ANOVA wi h Bon e oni co ec ion.
www.na u e.com/scien i ic epo s/
8
Scien i ic RepoR s | 6:39378 | DOI: 10.1038/s ep39378
Figu e 5. Highe exp ession le els o ype III IFNs in TC dono s co ela e wi h he exp ession le els o he
ansc ip ion ac o IRF-1. (a) mRNA exp ession le els o IRF-1 we e measu ed a 48 h p.i. in he mock- and
CVB3 in ec ed isle s desc ibed in Fig.1 (n = 23) using quan i a i e eal- ime PCR. The exp ession le els we e
no malized o ha o GAPDH and p esen ed as 2−(ΔC ) ( ela i e exp ession) o he whole isle coho ( i s
column). The esul s we e e-analysed in ha he dono s we e di ided acco ding o hei s1990760 geno ype
(TT o TC). The exp ession le els in he mock in ec ed (C, second column) and CVB3 in ec ed human isle s
(CVB3, hi d column) we e no malized o ha o GAPDH and p esen ed as 2−(ΔC ) ( ela i e exp ession), o
no malized o GAPDH and compa ed o he exp ession le els in mock-in ec ed isle s om he same dono and
p esen ed as 2−(ΔΔC ) ( old induc ion: ou h column). (b) Co ela ion analysis was pe o med by compa ing
he mRNA le els o selec ed genes exp essed by CVB3 in ec ed isle s a 48 h p.i. (n = 23). The 2−(ΔC ) alues
o IRF-1 we e co ela ed wi h ha o IFNλ 1, IFNλ 2 and IFNβ in CVB3 in ec ed isle s. (c,d) The co ela ions
desc ibed in (b) we e sepa a ely pe o med o dono s wi h he s1990760 TT (c) and TC (d) geno ypes. Linea
eg ession and he co ela ion coe icien , R2 and p alues we e calcula ed using Pea son co ela ion es (b–d).
p < 0.05 was conside ed s a is ically signi ican .
www.na u e.com/scien i ic epo s/
9
Scien i ic RepoR s | 6:39378 | DOI: 10.1038/s ep39378
NF-κB
IKKβ IKKα
MAVS
MDA5
IFNAR IFNλR
ISGF3
ISGs
IκB
Deg ada ion
Type I
in e e ons
CVB3
Type III
in e e ons
P
Pe oxisome
946T
Cell su ace
Nucleus
946T
TAK1 TBK
IKKε
MAVS
MDA5
TRAF6
P
IRF3
P
IRF1
IRF3
JAK
TYK2
JAK
TYK2
STAT1
STAT2
IRF9
TT ( s1990760) 946T
au oc ine, pa ac ine
TC ( s1990760) 946A/T
NF-κB
IKKβ IKKα
MAVS
MDA5
IFNAR IFNλR
ISGF3
IκB
Deg ada ion
CVB3
P
Pe oxisome
946A/T
Cell su ace
Nucleus
946A/T
TAK1 TBK
IKKε
MAVS
MDA5
TRAF6
P
IRF3
P
IRF1
IRF3
JAK
TYK2
JAK
TYK2
STAT1
STAT2
IRF9
Feedback
au oc ine, pa ac ine
Type I
in e e ons
Type III
in e e ons
ISGs e.g. IRF1
Figu e 6. P oposed model o ype I and ype III IFN induc ion and ac ion in isle s om dono s ca ying
he TT and TC s1990760 geno ypes. CVB3 in ec ion o isle s om dono s ca ying he s1990760 TT
geno ype (MDA5, 946 T) (uppe model) esul s in eplica ion o i al RNA and he gene a ion o a i al
in e media e (dsRNA). MDA5 binds dsRNA and ac i a es mi ochond ial and pe oxisomal associa ed MAVS.
This ini ia es a signalling cascade leading o he phospho yla ion and dime iza ion o IRF-3 h ough TBK-IKKε ,
and he deg ada ion o Iκ B om he NF-κ B complex h ough TRAF6 and TAK1. The ac i a ed ansc ip ion
ac o s IRF-3 and NF-κ B ansloca e o he nucleus whe e hey bind o he p omo e egions o ype I and
ype III IFNs and ini ia e hei exp ession. In addi ion, pe oxisome associa ed MAVS leads h ough unknown
mechanisms o he ac i a ion o he ansc ip ion ac o IRF-1, which binds solely o he p omo e egions
o ype III IFNs and selec i ely ini ia e hei exp ession. Exp essed and sec e ed ype I and ype III IFNs bind
in an au o- and pa ac ine manne o hei ecep o s IFNAR and IFNλ R, espec i ely. This in u n ac i a es
JAK-TYK2 kinases leading o he o ma ion o he ISGF3 complex and he exp ession o a la ge numbe o
ISGs. In ec ion o isle s om dono s ca ying he s1990760 TC geno ype (MDA5, 946 T/A) (lowe model)
esul s in s onge ype III IFN gene exp ession compa ed o ha induced in isle s om TT dono s, possibly
h ough he pe oxisome associa ed MAVS media ed ac i a ion o IRF-1. The s onge ype III IFN exp ession
is accompanied by a mo e obus ISG exp ession in TC dono s compa ed o TT dono s. This includes he
exp ession o IRF-1, which may u he ampli y he ype III IFN exp ession h ough a posi i e eedback
mechanism. Why MDA5 ac i a ion in dono s ca ying he s1990760 TC geno ype leads o a s onge ac i a ion
o pe oxisome associa ed MAVS compa ed o mi ochond ial associa ed MAVS is s ill o be disco e ed.