RESEARCH ARTICLE Open Access
A lan ic salmon popula ions e eal adap i e
di e gence o immune ela ed genes - a
duplica ed genome unde selec ion
E ik Kjæ ne -Semb
1,2*
, Fe nando Ayllon
1
, Tomasz Fu manek
1
, Vida Wenne ik
1
, Gei Dahle
1
, Ee o Niemelä
3
,
Mikhail Oze o
4
, Juha-Pekka Vähä
4,6
, Ke in A. Glo e
1,2
, Ca l J. Rubin
5
, Anna Wa gelius
1
and Rol B. Ed a dsen
1*
Abs ac
Backg ound: Popula ions o A lan ic salmon display highly signi ican gene ic di e ences wi h un esol ed molecula
basis. These di e ences may esul om sepa a e pos glacial coloniza ion pa e ns, di e si ying na u al selec ion and
adap a ion, o a combina ion. Adap a ion could be in luenced o e en acili a ed by he ecen whole genome
duplica ion in he salmonid lineage which esul ed in a pa ly e aploid species wi h duplica ed genes and egions.
Resul s: In o de o elucida e he genes and genomic egions unde lying he gene ic di e ences, we conduc ed a
genome wide associa ion s udy using whole genome esequencing da a om eigh popula ions om No he n
and Sou he n No way. F om a o al o ~4.5 million sequencing-de i ed SNPs, mo e han 10 % showed signi ican
di e en ia ion be ween popula ions om hese wo egions and en selec i e sweeps on ch omosomes 5, 10, 11,
13–15, 21, 24 and 25 we e iden i ied. These comp ised 59 genes, o which 15 had one o mo e di e en ia ed
missense mu a ion. Ou analysis showed ha mos sweeps ha e pa alogous egions in he pa ially e aploid
genome, each lacking he high numbe o signi ican SNPs ound in he sweeps. The mos signi ican sweep was
ound on Ch 25 and ca ied se e al missense mu a ions in he an i i al mx genes, sugges ing ha hese popula ions
ha e expe ienced di e ing i al p essu es. In e es ingly he second mos signi ican sweep, ound on Ch 5, con ains
wo genes in ol ed in he NF-KB pa hway (nkap and nk ), which is also a known pa hogen a ge ha con ols a la ge
numbe o p ocesses in animals.
Conclusion: Ou esul s show ha na u al selec ion ac ing on immune ela ed genes has con ibu ed o gene ic
di e gence be ween salmon popula ions in No way. The di e ences be ween popula ions may ha e been acili a ed
by he plas ici y o he salmon genome. The obse ed signa u es o selec ion in duplica ed genomic egions sugges
ha he ecen ly duplica ed genome has p o ided aw ma e ial o e olu iona y adap a ion.
Keywo ds: Whole genome duplica ion, Adap a ion, Aquacul u e, Immune sys em, GWAS, Resequencing, Selec i e
sweep, SNPs, Salmo sala
Backg ound
In addi ion o being one o he mos highly p ized esh-
wa e ish o ec ea ional ishing, he A lan ic salmon
(Salmo sala L.) is one o he mos economically impo -
an aquacul u e species wo ldwide. I s na u al dis ibu ion
is h oughou he No h A lan ic, anging om Long Is-
land Sound o Unga a Bay in he wes and om No he n
Po ugal o he Ba en s Sea in he eas [1]. This
dis ibu ion is he esul o pos glacial coloniza ion o
ecosys ems ha became a ailable when he glacial ice
e ea ed abou 10,000 yea s ago [2].
A lan ic salmon is cha ac e ised by highly signi ican ,
hie a chically s uc u ed popula ion gene ic di e gence,
wi h he la ges di e ences obse ed be ween he Eu opean
and No h Ame ican lineages [3–5]. This di e gence is also
obse ed on a egional scale, p esumably as a consequence
o he coloniza ion p ocess associa ed wi h he e ea o
he glacie [6, 7]. Mo eo e , local scale di e en ia ion ex-
is s, o example be ween neighbou ing i e s [8–10] and
* Co espondence: [email p o ec ed];[email p o ec ed]
1
Ins i u e o Ma ine Resea ch, Be gen, No way
Full lis o au ho in o ma ion is a ailable a he end o he a icle
© 2016 The Au ho (s). Open Access This a icle is dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0
In e na ional License (h p://c ea i ecommons.o g/licenses/by/4.0/), which pe mi s un es ic ed use, dis ibu ion, and
ep oduc ion in any medium, p o ided you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o
he C ea i e Commons license, and indica e i changes we e made. The C ea i e Commons Public Domain Dedica ion wai e
(h p://c ea i ecommons.o g/publicdomain/ze o/1.0/) applies o he da a made a ailable in his a icle, unless o he wise s a ed.
Kjæ ne -Semb e al. BMC Genomics (2016) 17:610
DOI 10.1186/s12864-016-2867-z
among ibu a ies wi hin he same i e which migh be
explained by es ic ed gene low, gene ic d i and
adap a ion [11–13].
A lan ic salmon exhibi a ela i ely complex li e his o y
ha includes spawning and ju enile ea ing in eshwa e
ollowed by ex ended ocean mig a ions o he eeding
g ounds [14]. As a consequence, salmon go h ough se -
e al dis inc ansi ions ha a e cha ac e ized by changes
in beha iou and physiology [15]. They a e also able o
adap o a ying local condi ions h oughou hei ange
o en i onmen s [16], exempli ied by hei abili y o in-
habi i e s wi h a wide ange o empe a u es, om Spain
o he colde A c ic la i udes [17]. P e ious s udies ha e
shown di e ences in empe a u e and clima e o be as-
socia ed wi h gene ic di e ences be ween salmon popu-
la ions [7, 18], and la i ude also seems o be co ela ed
wi h allele equencies o ma ke s ele an o immune
esponse in Ame ican and Eu opean A lan ic salmon
popula ions, possibly due o empe a u e induced di e -
ences in pa hogen-d i en selec ion o o he en i onmen al
ac o s [19–21].
In he wild, A lan ic salmon a e cons an ly con on ed
wi h a ange o pa hogens, and ha e consequen ly de el-
oped nume ous inna e and adap i e immune mechanisms
o o e come in ec ious challenges [22]. Recen s udies
sugges ha he p e alence o pa asi es and in ec ious dis-
eases is inc easing in wild popula ions pa ly due o global
wa ming [23, 24]. Gi en he comme cial ele ance o
A lan ic salmon and he ecen elease o a e e ence
genome [25], pa icula e o should be made o iden-
i y genes a ge ed by na u al selec ion in wild A lan ic
salmon popula ions ha ul ima ely can lead o op i-
mized aquacul u e p ac ices. The po en ial ele ance o
hese indings o he A lan ic salmon a ming indus y
is exempli ied by he iden i ica ion o In ec ious Pan-
c ea ic Nec osis (IPN) Vi us esis ance [26] and age a
ma u i y associa ed genes [27, 28]. A ela i ely ecen
whole genome duplica ion occu ed in he salmonid
lineage some 80 million yea s ago [29], esul ing in a
pa ly e aploid genome unde going ediploidiza ion.
Consequen ly he genome con ains many pa alogous
egions ha could p o ide aw ma e ial o e olu ion as
pa alogous genes and egions can di e si y and acqui e
new unc ions [30].
Based upon he analysis o mic osa elli e and SNP
ma ke s, se e al s udies ha e demons a ed ha he e a e
highly signi ican gene ic di e ences be ween A lan ic
salmon popula ions loca ed in he no h and sou h o
No way [31–33]. Howe e , he genomic egions and
genes behind he di e ences ha e no been in es iga ed
in de ail, and consequen ly, he po en ial adap i e sig-
ni icance o his gene ic di e gence emains elusi e.
Recen ly, a genome wide associa ion s udy (GWAS)
based upon whole genome esequencing da a e ealed a
selec i e sweep in A lan ic salmon s ongly associa ed
wi h age o ma u a ion [27]. Using a simila me hodo-
logical app oach, he p esen s udy aimed o iden i y genes
and genomic egions di e ging be ween A lan ic salmon
popula ions in he no h and sou h o No way. In o de o
achie e his objec i e, salmon popula ions inhabi ing he
ou i e s Tanael a, Laksel , Al ael a and Reisael a om
No he n No way and he ou i e s Gloppenel a,
Eidsel a, Suldalslågen and Å dalsel a om Sou he n
No way we e chosen o esequencing using DNA
pools (n= 30 ish pe i e , Fig. 1). The majo inding in
his s udy was he obse a ion ha di e si ying na u al se-
lec ion has ac ed on immune ela ed genes causing adap-
i e di e gence be ween popula ions in he no h and
sou h o No way.
Resul s and discussion
Wholegenomesequenceda a om eigh selec ed i -
e s along he No wegian coas (Fig. 1) was mapped o
he mos ecen A lan ic salmon e e ence genome
(AKGD00000000.4). This yielded a 26.7× a e age dep h o
co e age o uniquely mapped eads pe i e . SNP calling
e ealed 4,450,990 high quali y SNPs. To quan i y he gen-
e ic di e ence be ween popula ions o he chosen i e s,
Hudson’s es ima o o W ig h’s ixa ion index (F
ST
) [34]
was calcula ed (Addi ional ile 1: Table S1). A phylogene ic
ee was made using his dis ance ma ix o illus a e and
con i m he epo ed la ge gene ic di e ence be ween he
no he n and sou he n popula ions o A lan ic salmon in
No way (Fig. 1). S a is ical analysis using he Coch an-
Man el-Haenszel es o di e en allele equencies be-
ween no he n and sou he n A lan ic salmon popula ions
e ealed 474,410 SNPs wi h signi ican ly di e en allele
equencies (0.1 % FDR, Fig. 2a). Genomic egions sub-
jec ed o ecen posi i e selec ion a e expec ed o ha e
lowe he e ozygosi y han o he egions, and i he selec i e
p essu e di e s be ween popula ions, highe F
ST
is ob-
se ed [35]. An app oach calcula ing F
ST
and he e ozygosi y
in 50 kb sliding windows has p e iously been used o iden-
i y genomic egions unde selec ion (selec i e sweeps) [36].
Thisme hodwasused o indselec i esweepswhichdi e
be ween no he n and sou he n salmon popula ions in
No way (Addi ional ile 1: Figu e S1). The combined F
ST
/
he e ozygosi y app oach sugges ed 10 selec i e sweeps ha
di e ed be ween he wo geog aphical egions. The sweeps
anged om 75,000 o 575,000 bp in size, and we e ound
in ch omosomes 5, 10, 11, 13–15, 21, 24 and 25 (Table 1).
These sweeps con ained in o al 59 genes in ol ed in a
numbe o di e en biological p ocesses including cell di -
ision, cy okinesis, angiogenesis, de elopmen , ansc ip-
ional egula ion and immune esponse. Fo a de ailed lis
o gene ID and sho desc ip ion o unc ion see Addi ional
ile 1: Table S2.
Kjæ ne -Semb e al. BMC Genomics (2016) 17:610 Page 2 o 12
The high numbe o SNPs and genes in he selec i e
sweeps complica es he ask o pin-poin ing he mos
impo an gene ic di e ences. The e o e, we ocused on
missense mu a ions ha induce amino acid changes in
p o eins, since hese a e mo e likely o con e a di e -
ence in biological unc ion. Wi hin he iden i ied sweeps
20 signi ican ly di e en ia ed missense SNPs we e ound,
comp ising 15 di e en genes dispe sed in 6 selec i e
sweeps (Table 2). Th ee missense mu a ions we e ob-
se ed in he sweep on Ch 10, all in a single gene, anln,
encoding an ac in-binding p o ein equi ed o cy okin-
esis. Th ee genes on Ch 13 ha bo missense mu a ions:
pc2, in ol ed in chemosenso y ansduc ion, and in e -
es ingly knockou mice display changes in hei sexual,
agg essi e, and pa en ing beha io s [37]; m1,anenzyme
essen ial o he p oduc ion o deoxy ibonucleo ides; b1,
which p omo es G0-G1 ansi ion when phospho yla ed
by CDK3/cyclin-C ac s as a ansc ip ional ep esso o
E2F1 a ge genes. Also in he Ch 14 sweep he e a e
h ee genes wi h missense mu a ions: adnp,ahomeo-
domain con aining DNA binding ansc ip ion ac o ;
cps 1, encoding a componen o he clea age and polya-
denyla ion speci ici y ac o complex; pa p10, encoding
a ADP- ibosyl ans e ase in ol ed in apop osis, NF-kB
signaling, and DNA damage epai [38]. The sweep on
Ch 21 con ains one gene, naseh2b, which is linked o
a ch onic in lamma o y diso de in humans [39].
The second mos signi ican selec i e sweep was ound
on Ch 5 (Fig. 2b) and included he s ess and immune
esponse ansc ip ion ac o genes nk and nkap;zb b33
encoding a ansc ip ional egula o binding o me hyla ed
CpG dinucleo ides, and a gene wi h unknown unc ion,
sowahc. Bo h Nk and Nkap a e ansc ip ion ac o s
which egula e he NF-kB pa hway in which Nkap ac i-
a es many cell p ocesses including in lamma ion, im-
muni y, di e en ia ion, cell g ow h and apop osis, while
Fig. 1 Geog aphical o e iew o sampled salmon popula ions. Resequenced genomes o A lan ic salmon om ou popula ions in No he n No way
and ou popula ions in Sou he n No way we e analyzed in his s udy and a e shown as black do s on a map. A phylogene ic ee based on pai wise
calcula ions o ixa ion index (F
ST
) illus a es he gene ic dis ances be ween he sequenced popula ions. 19 addi ional popula ions om i e s
along he No wegian coas we e analyzed using geno yping assays and he geog aphical sampling loca ions o hese a e indica ed acco ding
o numbe s in he map
Kjæ ne -Semb e al. BMC Genomics (2016) 17:610 Page 3 o 12
Nk media es ansc ip ional ep ession o ce ain
Nkap esponsi e genes. Since NF-kB signaling pa hways
ac i a e heimmunesys emin hehos , hesep o eins
a e key a ge s o p o eases exp essed by in ading
pa hogens [40]. Func ional s udies o he Nkap p o ein
ha e e ealed oles o his p o ein in T-cell ma u a ion
[41] and mRNA splicing [42]. To ou knowledge, no
p e ious s udies ha e iden i ied unc ionally signi ican
SNPs associa ed wi h any o he ou genes loca ed
wi hin his sweep, howe e one o he SNPs ound in
nkap is loca ed in a highly conse ed egion necessa y
o ansc ip ional ep ession. He e he aline is conse ed
in o he species ep esen ing he ances al a ian while
in No he n No way me hionine is mos common
(Addi ional ile 1: Figu e S2). This inding may be e-
la ed o di e ences in immune de ense be ween salmon
om hese wo egions, a sugges ion suppo ed by he
ac ha he NF-kB pa hway is di e en ly egula ed in
IPN esis an salmon [43]. Fu he s udies will e eal
how hese SNPs modula e he unc ion o NF-kB and
i us esponse o i o he unc ional p ope ies a e as-
socia ed wi h he selec i e sweep on Ch 5.
The mos signi ican sweep was ound on Ch 25 and
con ained a clus e o i e mx (myxo i us esis ance)
genes known o be in ol ed in de ense agains i uses.
Th ee o hese mx genes con ained missense mu a ions;
mx1-1,mx1-2 and mx2-1 (Fig. 2c). These p o eins a e
dynamin-like GTPases induced upon i us in ec ion
h ough he inna e in e e on sys em. I has been shown
ha hey can ac b oadly agains bo h DNA and RNA i-
uses and speci ically agains ce ain i uses [44] and
s udies in mouse, human and chicken ha e shown ha
single missense mu a ions in Mx1 and Mx2 can con e
such speci ic esponses [45–48]. I is possible ha he
iden i ied missense SNPs in he mx genes e lec speci ic
adjus men s o di e en i al disease p essu es be ween
Fig. 2 Iden i ica ion o genomic egions unde selec ion in No he n and Sou he n No way. aManha an plo showing di e en ia ed SNPs be ween
no he n and sou he n popula ions o A lan ic salmon in No way. The x-axis indica es ch omosomal posi ions; he y-axis p esen s he nega i e
loga i hm o he P- alue o allele equencies being di e en be ween he wo geog aphical egions. SNPs in selec i e sweep egions iden i ied using
FST/he e ozygosi y a e indica ed by ed do s. SNPs abo e he dashed ho izon al line (p< 1.0658e-4, 0.1 % FDR) ha e signi ican ly di e en
allele equencies be ween he wo geog aphical egions. band cMagni ica ion showing 500 kb o selec i e sweeps on Ch 5 and Ch 25.
SNPs a e indica ed as black do s and missense mu a ions a e ma ked wi h ed squa es. The ack labeled “HET”shows he he e ozygosi y o
salmon om No he n (blue) and Sou he n No way (g een) in 3 kb windows. The ack labeled “FST”shows he F
ST
be ween popula ions in
No he n and Sou he n No way in 3 kb windows. In he bo om, iden i ied genes a e shown, wi h genes con aining di e en ia ed missense
mu a ions colo ed black. The x-axis shows he ch omosomal posi ions gi en in kb
Kjæ ne -Semb e al. BMC Genomics (2016) 17:610 Page 4 o 12
no he n and sou he n popula ions o salmon. We iden-
i ied missense SNPs in all egions o he p o ein includ-
ing a SNP in he an i i al speci ici y domain in exon 13
(Addi ional ile 1: Figu e S3). This SNP ep esen s a
s uc u ally ele an amino acid subs i u ion, whe e
a ginine seems o be he ances al a ian and cys eine
he de i ed a ian domina ing in he no he n popula-
ion (Ch 25 posi ion: 47,120,121). Likewise, SNPs in his
domain ha e been associa ed wi h speci ic i us esis -
ance in chicken [49, 50] and pig [51]. SNPs in mx genes
Table 1 Selec i e sweeps. F
ST
and he e ozygosi y es ima ions in sliding windows we e used o iden i y di e en ia ed loci unde going
selec ion in ei he no he n o sou he n popula ions o A lan ic salmon in No way. Genes ca ying di e en ia ed missense mu a ions a e
shown in bold
Ch omosome Ch omosomal egion Sweep leng h (bp) Genes in selec i e sweeps
5 22,475,000 –22,800,000 325,000 nmd, el 1, nk , sep 6, sowahc, pl39, up 3b, nkap,zb b33,
a p1b4, lamp2, cul4b, mc s1, c1gal 1c1, clic2
10 76,250,000 –76,450,000 200,000 sep 7, anln, p m 7, coq9, pol 2c, dga 1, nedd4, ada 1, spi e2
11 19,225,000 –19,300,000 75,000 numa1
13 78,325,000 –78,425,000 100,000 pc2, m1, slc6a7, b1, lpa 6
13 81,025,000 –81,150,000 125,000 eda2 , a , ophn1
14 64,700,000 –65,275,000 575,000 pa d6g, bloc1s4, nu 2, adnp2, xnl4a, pqlc1, kcng2, c dp1, adck5,
cps 1,pa p10, s 3gal1, khd bs3
15 46,925,000 –47,200,000 275,000 mdga1
21 24,850,000 –25,075,000 225,000 im13, naseh2b, n 0b1, il1 apl1
24 34,225,000 –34,525,000 300,000 edil3
25 47,075,000 –47,225,000 150,000 s xbp5l, g 2e1, mx1-1,mx1-2,mx2-1, smcp-1a, mx1-3
Table 2 Missense mu a ions in selec i e sweeps. Se e al missense mu a ions we e disco e ed in he selec i e sweeps. The able lis s
esequencing de i ed e e ence alleles equencies in he no he n (N) and sou he n (S) popula ions o salmon in No way. SNPs om
Ch 5 and Ch 25 selec ed o geno yping a e shown in bold
Ch Posi ion -log
10
P Re e ence allele
equency (S/N)
Re /al nucleo ide Re /al amino acid Gene Desc ip ion o gene
5
a
22,641,277 29.6 1.00/0.34 A (G) Phe (Se ) nk NF-kB ep essing ac o .
5 22,691,092 27.5 0.95/0.29 G (T) Ala (Se ) sowahc Unknown unc ion.
5
a
22,708,279 28.8 1.00/0.33 C (T) Val (Me ) nkap NF-kB ac i a ing p o ein.
5
a
22,719,453 19.4 0.79/0.27 T (A) Val (Glu) zb b33 T ansc ip ional egula o .
10
b
76,272,006 18.7 0.94/0.46 T (G) Asp (Glu) anln Ac in binding p o ein.
10 76,272,245 18.4 0.96/0.49 C (A) P o (Gln) anln Same as abo e.
10 76,278,414 14.5 0.32/0.76 G (A) Ala (Th ) anln Same as abo e.
13 78,341,469 24.1 0.97/0.43 G (A) Gly (A g) pc2 T ansien Recep o Ca ion Channel.
13 78,349,376 20.3 0.98/0.40 T (G) Glu (Ala) m1 Ribonucleoside-diphospha e educ ase.
13
b
78,413,379 20.7 0.98/0.48 A (T) Gln (Leu) b1 Regula o o en y in o cell di ision.
14 64,739,123 20.0 0.55/0.01 T (C) Asn (Ile) adnp2 T ansc ip ion ac o .
14
b
64,988,859 11.9 0.36/0.79 A (T) Asn (Lys) cps 1 P e-mRNAs p ocessing.
14
b
65,006,543 20.6 0.38/0.94 A (C) Asp (Glu) pa p10 ADP-Ribosyl ans e ase.
21
b
24,974,056 19.7 0.98/0.46 T (A) Me (Lys) naseh2b Non ca aly ic subuni o RNase H2.
25
a
47,105,598 26.0 0.84/0.03 A (G) Th (Ala) mx1-1 In e e on-induced an i i al.
25 47,108,912 27.2 0.91/0.00 G (A) Val (Ile) mx1-1 Same as abo e.
25 47,111,137 18.0 0.79/0.06 G (A) Val (Me ) mx1-1 Same as abo e.
25
a
47,120,121 33.1 0.83/0.06 C (T) A g (Cys) mx1-1 Same as abo e.
25 47,147,348 27.2 0.22/0.98 G (T) P o (His) mx1-2 Same as abo e.
25 47,181,001 25.4 0.85/0.00 T (C) His (A g) mx2-1 Same as abo e.
a
Allele equencies om geno yping a e illus a ed in Fig. 4
b
Allele equencies om geno yping a e illus a ed in Addi ional ile 1: Figu e S5
Kjæ ne -Semb e al. BMC Genomics (2016) 17:610 Page 5 o 12

ha e also been in es iga ed in ano he ish species, he
u bo [52], howe e , p ope ies ela ed o p o ec ion
agains i uses we e no in es iga ed in his s udy. In
ainbow ou (Onco hynchus mykiss) gene ic a ia ion
in mx be ween s ains in exon 3–6, was co ela ed wi h
suscep ibili y o in ec ious hema opoie ic nec osis i us
(IHNV) [53]. This i us also in ec s A lan ic salmon
and ou disco e y o a missense mu a ion in exon 6 sug-
ges s ha salmon could ha e adap ed o he IHNV
(Addi ional ile 1: Figu e S3). In addi ion, di e en
s ains o ainbow ou display a iable suscep ibili y
o his i us [54]. In his s udy we canno elucida e he
unc ional signi icance o he acqui ed SNPs in mx in
No he n No way, howe e , u he s udies will e eal
whe he any o hese changes ha e been in ol ed in
hos - i us adap a ion [55].
We also in es iga ed whe he he selec i e sweeps on
Ch 5 and Ch 25 had pa alogous egions in he pa ially
e aploid salmon genome [56]. In silico analysis showed
ha bo h sweeps ha e pa alogous egions loca ed on
o he ch omosomes. The Ch 5 sweep has a pa alogous
egion on Ch 9 (Addi ional ile 1: Figu e S4, posi ion
51,349,279 o 51,849,279), which did no con ain any di -
e en ia ed SNPs. The syn eny is conse ed in o he
species, and he exis ence o only one copy in zeb a ish
(Danio e io), combined wi h he obse a ion ha mis-
sense mu a ions on Ch 5 a e no p esen in he pa alo-
gous genes on Ch 9, indica e ha he mu a ions a ose
a e he salmonid speci ic whole genome duplica ion
(WGD). Based upon his obse a ion, i is possible o
specula e ha he WGD p o ided pa alogous egions
whe e one copy was ee o sub- o neo- unc ionalize,
much like he heo y o duplica ed genes [57] which has
been sugges ed o be impo an o e olu iona y adap-
a ion and inno a ion in salmon [58], in eleos s [59]
and in gene al [30]. A simila pic u e is seen o he
sweep on Ch 25 whe e he pa alogous egion ha bo s
aclus e o h eemx genes on Ch 12 (posi ion
66,552,602 o 67,052,602), bu ca ies no di e en ia ed
SNPs o missense mu a ions. While he sweeps on Ch s
11, 15, 21 and 24 ha e no clea pa alogous egions, he
sweeps on Ch 10, Ch 14 and he wo sweeps on Ch
13 also ha e pa alogous egions wi h e y ew signi ican ly
di e en ia ed SNPs, on Ch 16, 27 and 4, espec i ely
(Fig. 3). Simila ly, in ou ecen disco e y o he loci in
Ch 25 con olling age a ma u i y [27] we in es iga ed
he wo pa alogous egions in Ch 21, bo h o which we e
wi hou SNPs associa ed wi h he ai . Toge he , hese
indings indica e ha he pa ially e aploid s age may be
bene icial o adap a ion, since one gene copy o gene
clus e can keep he o iginal unc ion while he o he can
adap o a new si ua ion such as no el disease p essu es.
In his s udy, he ini ial esequencing was based only
upon males. This is because i allowed eusing sequence
da a om ou p e ious wo k [27]. The a ge ed SNP
analysis, used o alida e he esul s om esequencing
in a la ge independen se o i e s, was conduc ed using
bo h males and emales (Figs. 1 and 4). Geno yping o
mixed sex salmon om 19 i e s (n= 20 salmon/ i e )
along he No wegian coas (Fig. 1) o i e missense SNPs
on Ch 5 and 25 con i med s ong gene ic di e en ia ion
be ween salmon popula ions om he no h and sou h o
No way (Fig. 4). Popula ions om no he n i e s (1–9)
displayed allele equencies in he ange 0–0.7, while hose
om sou he n i e s (11–19) we e close o ixa ion o
one allele a hese wo loci. Salmon om i e 10, Målsel ,
shows in e media e equencies, which co esponds well
wi h wha has been epo ed in he li e a u e [31, 32].
These esul s also con i m allele equency es ima ions
om he pooled esequencing (Table 2). In addi ion, we
designed Sequenom assays o i e o he missense SNPs
in o he egions; one SNP each o sweeps on Ch s 10, 13
and 21, and wo SNPs in Ch 14. Geno yping was pe -
o med o all 19 i e s (Addi ional ile 1: Figu e S5). The
allele equencies showed he same clea di e ence be-
ween he no he n and sou he n popula ions. Fo he
SNPs on Ch 14 he e appea s o be an addi ional gene ic
shi be ween he i e s 14, S jø dalsel a and i e s sou h
o his. In addi ion o he da a p oduced wi hin he
p esen s udy, esequencing da a om a ecen publica-
ion was downloaded and compa ed o ou esul s [28].
The downloaded da a include h ee indi idually se-
quenced salmon om 4 sou he n and 3 no he n salmon
i e s in No way. These da a co obo a e ou esequen-
cing and geno yping esul s (Addi ional ile 1: Table S3).
Ou su eyed SNPs he e o e also ep esen obus and
good gene ic ma ke s o dis inguishing no he n and
sou he n popula ions o A lan ic salmon in No way. Fu-
u e s udies on an ex ended se o popula ions may e eal
i hese a e also obus ma ke s o de ec ing gene ic
s uc u ing in o he pa s o he dis ibu ion ange o he
species.
A lan ic salmon aquacul u e in ol es ea ing domes i-
ca ed ish ha o igina e om comme cial b eeding p o-
g ams. Fo y wild popula ions om bo h he no h and
sou h o No way we e sampled when es ablishing he
na ional b eeding p og ams o salmon [60]. Howe e ,
analyses o gene ic ma ke s demons a e ha he e is a
dominance o salmon om Sou he n No way in he do-
mes ica ed lines cu en ly in p oduc ion [61]. Gene ic
analyses o a med salmon escapees in No way ha e un-
co e ed gene ic in og ession in o na i e salmon popula-
ions in bo h No he n and Sou he n No way, bu he
biological consequence emains unknown [32, 61, 62].
Consequen ly he esul s om he p esen s udy, whe e
adap i e gene ic di e gence be ween wild salmon om
popula ions loca ed in he no h and sou h o No way
was e ealed, i is likely ha he po en ial nega i e
Kjæ ne -Semb e al. BMC Genomics (2016) 17:610 Page 6 o 12
gene ic impac o domes ica ed salmon in og ession is
g ea e in popula ions loca ed in no he n egions, since
he a med ish o igina e mos ly om wild Sou he n
No way popula ions.
Conclusion
In his s udy we pe o med a GWAS by genome ese-
quencing wi h he aim o sc een he A lan ic salmon
genome o gene ic di e en ia ion be ween he no he n
and sou he n popula ions in No way. By in es iga ing
eigh i e s we unco e ed en pa icula ly s iking
sweeps including wo clus e s o immune ela ed genes
ha bo ing missense mu a ions. A easible in e p e a ion
is ha di e en popula ions o A lan ic salmon ha e his-
o ically been exposed o di e en selec ion p essu es in
he o m o pa hogens. Some o hese adap ed alleles
could be ad an ageous o aquacul u e p oduc ion which
is cu en ly hampe ed by a numbe o diseases, including
i us in ec ions [63]. Fu u e s udies should include gene
edi ing o immune genes ound in hese selec i e sweeps
[64, 65] in combina ion wi h i al exposu e expe imen s.
Wi hin hese expe imen s, i uses ele an o salmon
aquacul u e should be he p ima y ocus since inding
speci ic esis ance alleles can be o signi ican alue o
he indus y and could also be used o p o ec ing wild
ish agains high disease p essu es posed by open cage
aquacul u e [66]. Upon inding he p o ec i e alleles, se-
lec i e b eeding on indi iduals wi h bene icial haplo ypes
could lead o inc eased wel a e o aquacul u e salmon,
dec eased disease p essu e on wild popula ions and
could also be economically a o able o he indus y.
On he o he hand, u he s udies should in es iga e he
impac o gene ic in og ession om e ile aquacul u e
escapees on he adap i e gene ic p ope ies in wild pop-
ula ions. To educe he isk o his unwan ed loss o
local adap a ion and al e a ion o i ness- ela ed ai s, a
Fig. 3 Compa ison o sweeps and pa alogous egions. Rep esen a ion o he selec i e sweeps ha ing pa alogous egions in he salmon genome,
displaying a 500 kb o e iew o he SNPs in he selec i e sweeps (le side) and in he co esponding pa alogous egions ( igh side). The y-axis
shows –log
10
P o he SNPs being di e en in popula ions be ween he no h and sou h o No way, and he x-axis ep esen s he posi ion in he
gi en ch omosome. The dashed lines indica e he genome wide signi icance h eshold (0.1 % FDR)
Kjæ ne -Semb e al. BMC Genomics (2016) 17:610 Page 7 o 12
sus ainable solu ion would be he use o s e ile ish in
aquacul u e, especially in No he n No way. Fu u e s ud-
ies should also in es iga e whe he pa alogous egions o
selec i e sweeps ha e unde gone posi i e selec ion o no ,
as he la e scena io would sugges an e olu iona y mech-
anism which p o ides highe adap i e possibili ies when a
genome is pa ially e aploid.
Me hods
Samples and sampling
Scales om 30 A lan ic salmon males pe i e we e selec ed
om a sample se o 26,000 samples collec ed in coas al
ishe iesinNo he nNo way.In he Kola c ic Salmon p o-
jec (h p://p osjek . ylkesmannen.no/Kola c icsalmon), he
mul ilocus geno ypes o all indi iduals we e compa ed o a
gene ic baseline consis ing o o e 180 i e s om No he n
Russia and No way and we e assigned o i e o o igin.
Samples ha we e assigned wi h high p obabili y o ou i -
e s in No he n No way; Al ael a, Reisael a, Laksel and
Tanael a we e gene ously made a ailable o his s udy. 30
salmon males om each o ou di e en i e s in Sou he n
No way, including Å dalsel a, Eidsel a, Gloppenel a
and Suldalslågen we e sampled and esequenced in a
ecen s udy [27]. In addi ion o hese, we also used
male and emale salmon DNA om 19 i e s along he
No wegian coas . These included 20 pa indi iduals om
each o he i e s G ense Jakobsel , Neiden, Be gebyel a,
Komagel a, Kongs jo del a, Lang jo del a, Bø sel a, S ab-
bu sel a, Reppa jo dsel a, Målsel , Laukhelle, Al s åg ass-
d age , Å gå ds assd age , S jø dalsel a, Jøls a, Lyseel a,
Bje k eimsel a, S o el a and Enningdalsel a ( ep esen ed
by numbe s in Fig. 1). Wi h he excep ion o Enningdalsel a
whe e he sample was ob ained om scales collec ed by
ec ea ional ishe ies, hese samples we e ob ained om
ins collec ed by elec o ishing o ju enile salmon om
muli ple loca ions in he i e s.
DNA ex ac ion and sequencing
DNA om he 19 i e s o geno yping was ex ac ed
om scales o in samples using Qiagen DNeasy Blood
and Tissue Ki (Qiagen, Hilden, Ge many) acco ding o
manu ac u e ’s ecommenda ions. F om salmon belong-
ing o he ou popula ions in No he n No way o al
DNA was ex ac ed om scales using Qiagen DNeasy
Blood and Tissue Ki . Equal amoun s o DNA om en
indi iduals we e pooled o make h ee pools pe i e ,
o aling 30 indi iduals om each i e . Pai ed-end lib a ies
we e cons uc ed using he Genomic DNA Sample P epa -
a ion Ki (Illumina, CA, USA) acco ding o manu ac u e ’s
ins uc ions and sequenced on he Illumina HiSeq2000
pla o m (Illumina, CA, USA) a he No wegian Sequen-
cing cen e (h ps://www.sequencing.uio.no, Oslo, No way)
wi h each pool sequenced in sepa a e lanes.
Sequence mapping and SNP calling
To ensu e high quali y sequences, sequenced eads
we e inspec ed wi h Fas QC (h p://www.bioin o ma ics.
bab aham.ac.uk/p ojec s/ as qc/). Adap e sequence e-
mo al and quali y imming was done wi h Cu adap
[67], esul ing in 1,077,839,448 (SD 40,407,492) pai ed
eads on a e age pe i e . Sequenced eads we e
mapped o he mos ecen elease o he salmon genome
Fig. 4 Allele equencies o i e missense mu a ions in popula ions along he No wegian coas . Th ee missense mu a ions iden i ied in selec i e
sweeps on Ch 5 and wo on Ch 25 we e used in a geno yping assay o A lan ic salmon popula ions along he No wegian coas . The g aphs
show equencies o he e e ence alleles. Ri e numbe s a e explained in Fig. 1. SNPs on Ch 5 included missense mu a ions in he genes nk ,
nkap and zb b33 (g een lines) and missense mu a ions on Ch 25 we e loca ed in he mx1-1 gene (blue lines)
Kjæ ne -Semb e al. BMC Genomics (2016) 17:610 Page 8 o 12
(AGKD0000000.4) using Bow ie2 ( .2.1.0) [68] wi hou
so clipping (end- o-end mode). To inc ease he sensi i -
i y o he mapping, seed leng h (−L pa ame e ) was se o
18 and he in e al be ween ex ac ed seeds (−ipa ame e )
was se o S,1,1.5 co esponding o he unc ion (L) = 1 +
1.5*sq (L), whe e L is he ead leng h. Addi ionally, he
maximum numbe o misma ches pe seed (−N pa ame e )
was se o L,0,0.1, co esponding o he unc ion (L) = 0 +
0.1*L, whe e L is he ead leng h, and minimum alignmen
sco e (−− sco e-min pa ame e ) was se o L,-0.6,-0.4, co e-
sponding o he unc ion (L) = −0.6 + −0.4*L, whe e L is
he ead leng h. To emo e ambiguously mapped eads he
mapping quali y h eshold was se o 20. To ob ain highe
sequence co e age, he h ee sequenced pools pe i e
we e me ged o a single BAM ile using SAM ools me ge.
SNPs we e called using SAM ools mpileup [69] and he
ou pu was pa sed using he PoPoola ion2 package
(mpileup2sync.ja ) [70] wi h a minimum base quali y
h eshold o 20. Fo a SNP o be included in he inal
se o high quali y SNPs, minimum co e age o 10 and
maximum co e age o 50 (99 % pe cen ile) was e-
qui ed o each i e . In addi ion, he o al numbe o
obse ed mino alleles was equi ed o be a leas 8. Re-
cen ly published whole genome esequencing da a om
indi iduals [28] was downloaded and mapped o he
e e ence genome. The da a included h ee salmon om
each o he i e s Tanael a, Reppa jo del a, Al ael a,
Namsenel a, Å gå ds assd age , Naus a and Jøls a, whe e
he i s h ee ep esen popula ions in No h e n No way
and he las ou ep esen Sou he n No way. Accession
numbe s o he samples a e shown in he cap ion o
Addi ional ile 1: Table S3.
S a is ical analysis
Pai wise ixa ion index (F
ST
) be ween all eigh sequenced
popula ions was calcula ed o all high quali y SNPs
using Hudson’s es ima o o F
ST
[34]. F
ST
alues we e
a e aged o e all SNPs in each popula ion o gene a e a
dis ance ma ix using F
ST
as gene ic dis ance. This ma ix
was con e ed o a newick ee using NEIGHBOR om
hePhylippackage[71]andaphylogene ic eewas
c ea ed wi h NJplo [72]. To ind SNPs wi h signi ican ly
di e en allele equencies (0.1 % FDR) be ween popula-
ions om No he n and Sou he n No way he Coch an-
Man el-Haenszel es o epea ed es s o independence
om he PoPoola ion2 package (cmh- es .pl) [70] was
used. The FDR h eshold was de e mined using he
me hod desc ibed in [73]. Allele coun s o each i e
we e me ged o ge he o al allele coun pe SNP in
No he n and Sou he n No way, co esponding o 120
indi iduals pe geog aphical egion. F om his, F
ST
alues
be ween he no he n and sou he n popula ions we e es i-
ma ed using he F
ST
calcula ion om he PoPoola ion2
package ( s -sliding.pl) o each SNP, wi h –pool-size
pa ame e se o 120. Genomic egions wi h low alues o
he e ozygosi y may indica e SNPs unde selec ion. The e-
o e he e ozygosi y alues we e es ima ed o no h and
sou h o No way, sepa a ely, o each SNPs as 2 * (majo
allele equency * mino allele equency). Sliding windows
o 50 kb wi h s eps o 25 kb was used o ind genomic e-
gions wi h high F
ST
alues and wi h low he e ozygosi y
alues in ei he No he n o Sou he n No way. This ap-
p oach is simila o one used o disco e genomic egions
unde selec ion in o he animals [36]. To iden i y pu a i e
selec i e sweeps i was equi ed ha he a e age F
ST
alue
o he window was a leas 0.17 (abo e 99.9 % pe cen ile)
and ha a e age he e ozygosi y o he window in ei he
No he n o Sou he n No way was a mos 0.15 (below
5 % pe cen ile) (Addi ional ile 1: Figu e S1). The h esh-
olds we e chosen wi h ocus on cap u ing he ou lie s in
he FST and he e ozygosi y dis ibu ions. Pu a i e sweeps
we e ex ended o he sides o as long as he neighbo ing
windows had ei he a e age F
ST
o a leas 0.17 o he e o-
zygosi y o a mos 0.15 in ei he No he n o Sou he n
No way. I iden i ied sweeps we e less han 50 kb apa
hese we e joined o a oid agmen a ion o he pu a i e
selec i e sweeps. Genomic windows con aining mo e han
10 % ambiguous bases (Ns) in he e e ence assembly
we e disca ded o exclude egions wi h high le els o
unce ain y.
SNP anno a ion
Genes in he sweep egions we e ob ained om he o i-
cial genome anno a ion (NCBI Salmo sala Anno a ion
Release 100). Missense mu a ions in selec i e sweep e-
gions we e iden i ied by manual inspec ion o he coding
sequences. Amino acid sequences o i e mx genes ound
in a selec i e sweep on Ch 25 we e aligned o he ho-
mologs Mx1 and Mx2 om human and MxD and MxG
om Zeb a ish using BLASTP (de aul pa ame e s).
Func ional domains in he Mx p o eins we e assigned
using domain in o ma ion o human Mx1 om Uni-
P o . Amino acid sequences om ou genes con aining
missense mu a ions in a selec i e sweep on Ch 5 (nk ,
sowahc,nkap and zb b33) we e aligned o homologous
zeb a ish and No he n Pike genes using BLASTP wi h
de aul pa ame e s. Syn eny be ween genes in he sweep
on Ch 5 and o he animals was ound using he UCSC
genome b owse (h ps://genome.ucsc.edu) o inspec
he syn enic egions o zeb a ish, human and mouse.
Pa alogous egions o he sweeps we e iden i ied using
TBLASTN (de aul pa ame e s) wi h he genes in he
sweeps agains he salmon genome.
Geno yping
Twen y salmon om 19 i e s along he No wegian
coas line (n= 380) we e geno yped using en o he
mos signi ican missensemu a ionsonaSequenom
Kjæ ne -Semb e al. BMC Genomics (2016) 17:610 Page 9 o 12