This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright
Transcript
This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies areencouraged to visit:
Immune parameters correlating with reduced susceptibility topancreas disease in experimentally challenged Atlantic salmon(Salmo salar)
Søren Grove a,1, Lars Austbø b,1, Kjartan Hodneland c, Petter Frost c, Marie Løvoll a,Marian McLoughlin c, Hanna L. Thim d, Stine Braaen e, Melanie König b, Mohasina Syed b,Jorunn B. Jørgensen d, Espen Rimstad e,*
a Section for Immunology, Norwegian Veterinary Institute, Oslo, NorwaybDepartment of Basic Science and Aquatic Medicine, Norwegian School of Veterinary Science, Oslo, NorwaycMSD Animal Health, Thormøhlensgate 55, 5008 Bergen, NorwaydNorwegian College of Fishery Science, University of Tromsø, 9037 Tromsø, NorwayeDepartment of Food Safety and Infection Biology, Norwegian School of Veterinary Science, P.O. Box 8146 Dep, N-0033 Oslo, Norway
a r t i c l e i n f o
Article history:Received 21 December 2011Received in revised form23 November 2012Accepted 10 December 2012Available online 7 January 2013
Two strains of Atlantic salmon (Salmon salar) with different susceptibility to infectious salmon anaemia(ISA) were challenged with salmon pancreas disease virus (SPDV), the etiological agent of salmon pan-creas disease (PD), by cohabitation. Serum and tissues were sampled at 0, 1, 3, 6 and 8 weeks post-challenge. Experimental challenge with SAV did not cause mortality, but virus loads and assessment ofhistopathology indicated that the fish more resistant to ISAV (ISAHi) also was more resistant to PD. Eightweeks post-challenge, the ISAHi strain had higher titres of SAV-neutralising antibodies than the lessresistant strain (ISALo). Transcript levels of four adaptive and six innate immune parameters wereanalysed by real-time RT-PCR in heart, head kidney (HK) and gills of both strains. Secretory IgM (sIgM)and CD8 levels differed most between the two salmon strains. The ISAHi strain had significantly higherlevels of sIgM in HK at all samplings, and significantly higher CD8 levels in gills at most samplings. Inheart, both sIgM and CD8 levels increased significantly during the challenge, but the increase appearedearlier for the ISALo strain. By hierarchical clustering analysis of mRNA levels, a clear segregation wasobserved between the two strains prior to the virus challenge. As the viral infection developed, theclustering divide between fish strains disappeared, first for innate and later for adaptive parameters. Ateight weeks post-challenge, the divide had however reformed for adaptive parameters. Possible pair-wise correlation between transcript levels of immune parameters was evaluated by a non-parametricstatistical test. For innate parameters, the extent of correlation peaked at 3 wpc in all tissues; thiscame rapidly for ISALo and more gradual for ISAHi. The ISAHi strain tended to show higher correlation forinnate parameters in heart and gill than ISALo at early sampling times. For adaptive immune parameters,little correlation was observed in general, except for ISAHi in heart at 6 wpc.
Overall, the observed differences in immune parameters may provide important clues to the causesunderlying the observed difference in susceptibility to PD.
� 2013 Elsevier Ltd. All rights reserved.
1. Introduction
Salmon pancreas disease (PD) is caused by an alphavirus calledSalmon pancreas disease virus [1]. Due to its taxonomic placement,the name Salmonid alphavirus (SAV) has been proposed and is nowcommonly used [2]. Different isolates of SAV can only be dis-criminated by genetic analysis. Based on genotypic variations, it iscurrently proposed that SAV should be divided into six genogroups(subtypes) [3], each correlating with geographical separation.
Abbreviations: ISAHi, fish strain with relative high resistance to infection withISAV; ISALo, fish strain with relative high resistance to infection with ISAV.* Corresponding author. Tel.: þ47 22 96 47 66.
E-mail address: [email protected] (E. Rimstad).1 These authors contributed equally to this work.
Contents lists available at SciVerse ScienceDirect
Fish & Shellfish Immunology
journal homepage: www.elsevier .com/locate / fs i
1050-4648/$ e see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.fsi.2012.12.014
Whereas PD refers to the disease in seawater farmed Atlanticsalmon (Salmo salar) and rainbow trout (Oncorhynchus mykiss) inthe British Isles/Norway, a related disease called sleeping disease(SD) is mainly found in rainbow trout reared in freshwater (CentralEurope). SD in trout caused by subtype SAV2 and SPD in the BritishIsles is most commonly caused by SAV1, while the Norwegianoutbreaks of SAV have been restricted to the subtype 3 [4].
PD is an emerging disease in the Atlantic salmon farming in-dustry in Europe. In Norway a significant increase of disease out-breaks in the southern part of the country has occurred since 2003[5]. In field outbreaks of PD, clinical signs and severity of patho-logical changes in the key target organs pancreas, heart and skeletalmuscle, will vary between individuals [6]. Clinical signs includelethargy and impaired swimming performance; the latter inter-preted as the “sleeping” behaviour typical for SD. Depending ontemperature and route of infection, infected fish within daysdevelop viraemia that normally lasts for 2 weeks [7,8]. In SAVcohabitant challenge experiments, SAV infection and PD pathologyis found from 3 wpc after shedding [9], but generally experimentalchallenges do not induce mortality. Waning of viraemia has beenshown to coincide in time, but not overlap, with the emergence ofneutralizing antibodies (NAb) in the bloodstream [10], which mayhappen as early as 10 days after challenge by intraperitoneal in-jection [11]. Tissues [7,8] and blood cells [7] become virus positivewithin few days, and virus may remain in these cellular compart-ments weeks after virus have been cleared from the serum. DuringSAV infection, lesions develop in a definite temporal manner,affecting first pancreas and heart followed by skeletal musculatureand possibly CNS [6,10]. Atlantic salmon surviving experimental PDdevelops protective immunity that can last at least nine months[12]. As reoccurrence of PD has not been reported from previouslyinfected field populations, the protection obtained from natural PDis likely long-lasting, at least on population level [6]. This protectiveimmunity is likely conferred by NAb, as indicated by the protectiongained in fish receiving injection of serum from PD convalescentfish [13]. Susceptibility to PD has been shown to vary betweencommercial strains of Atlantic salmon [10], but the underlyingreason is not well known. In response to experimental infection,leukocyte phagocytic activity was shown to increase, as were levelsof lysozyme and complement [14]. Reduced PD specific patho-logical lesions and SAV3 levels in experimentally infected Atlanticsalmon injected with the toll-like receptor (TLR)-ligands polyI:Cand CpG have recently been reported, and it was demonstrated thatthis correlated with transcriptional up-regulation of IFNa1, IFNg,Mx and the chemokine CXCL10 [15]. The study found that the IFNsystem participates in the host defence against SAV, as also indi-cated by the increased levels of IFNa1 and Mx in SAV challengedsalmon [16]. Mammalian alphaviruses are also strong inducers ofIFN-I resulting in transcriptional up-regulation of genes with anti-viral activities [17]. Studies of alphavirus infection in mice deficientin IFN-I signalling have indicated that this pathway is a primaryprotective response [17,18]. The viral non-structural protein nsP2 ofmammalian alphaviruses is an important regulator of virusehostcell interactions and plays a significant role in suppressing theantiviral response [19]. Experimental vaccines using formalininactivated [6,20] or attenuated SAV [21] induce protection, butcommercial vaccines have not been able to fully control the diseasein the field. Description of SAV-specific induction of innate andacquired immune responses, and in particular the nature of pro-tective response may aid the development of efficient prophylaxis.
The possibility that increased resistance to a specific pathogenin a bred fish strain may affect susceptibility to other pathogens hasbeen given little attention. In the present study, two strains ofAtlantic salmonwith differing susceptibility to ISAwere challengedwith SAV by cohabitation. Based on morbidity parameters and
quantification of virus RNA, the two fish strains were shown todiffer in susceptibility to PD. In an attempt to resolve the nature ofimmune factors underlying this differential disease susceptibility,an array of adaptive and innate immune parameters were analysedby real-time reverse transcription quantitative PCR (RT-qPCR) andby serological methods.
2. Materials and methods
2.1. Fish
Atlantic salmon (S. salar) of two different strains were obtainedfrom SalmoBreed AS (Bergen, Norway) and kept at the ILAB facility(Bergen, Norway). One strain (ISAHi) was characterised by relativehigh resistance to infection with ISAV (estimated breeding value of26.3%) whereas the other (ISALo) had relative low resistance(estimated breeding value of �18%). A third strain of A. salmon(ILAB/08/004) was obtained from ILAB and used as virus shedderfish in cohabitation challenge. Prior to use, the fish were tested andfound negative for presence of infectious pancreas necrosis virus(IPNV) and SAV by real-time RT-PCR. The fish were kept at 10e13 �Cand fed ad libitum during the challenge.
2.2. Cohabitation challenge and sampling
For the cohabitation challenge, a Norwegian isolate of SPDV(PD03 13p2) passed twice in CHSE-214 cells, was used to inject theshedder fish. The challenge was performed in two tanks, tank Acontaining 100 salmon of strain ISAHi (meanweight 35 g) and tankB containing 100 salmon of strain ISALo (meanweight 35 g). At day0 (0 wpc), each of tanks A and B received 25 shedders of strain ILAB/08/004 that had got an intraperitoneal (ip) injection of 0.2 ml virussuspension the same day. Although the exact timing of infection isnot possible to pin-point in a cohabitant challenge model this waspartly compensated for by using a relatively large number ofshedder fish.
At day 0, prior to infection, blood and tissue (head kidney (HK),heart and gill) were sampled from 10 fish from each of strain ISAHiand ISALo. Blood samples were taken from anaesthetized fish usingheparin-containing Vacutainers and serum was prepared by stan-dard centrifugation protocol and stored at �20 �C. Tissue sampleswere immediately transferred to ice-cold RNAlater (Ambion), thenincubated at 4 �C for 24 h and finally stored at �20 �C. During thesubsequent challenge, identical samplings were performed at 1, 3, 6and 8 wpc.
At separate samplings, at 4 and 6 wpc, additional heart tissuesamples were taken from 15 fish from each of strain ISAHi andISALo and fixed in 3.5% formaldehyde in buffered saline (pH 7.0).
2.3. Histopathological examination
Formalin fixed heart tissue was processed for haematoxylineeosin staining using standard protocol. A score system was usedto evaluate the severity of SAV induced heart lesions, i.e. no lesion:minimal: 1, mild: 2, moderate: 3 and severe: 4 [22]. Scores �2 areconsidered to be specifically to clinical PD infection. Moribund fishand fish that have died from PD are typically scored as severe (4).The scoring was done as a blinded experiment.
2.4. RNA extraction and cDNA synthesis
Tissue samples in RNAlater were distributed between fourcontributing laboratories (AeD), where total RNA was isolated byuse of RNeasy Mini Kit (Qiagen) and eluted in 50 ml of RNase freeH2O. The RNA output was checked by gel electrophoresis for the
S. Grove et al. / Fish & Shellfish Immunology 34 (2013) 789e798790
absence of degradation and quantified spectrophotometrically us-ing Nanodrop ND-1000 (Thermo Scientific) before storageat�80 �C. cDNAwas synthesised from 1 mg of RNA using QuantiTectReverse Transcription Kit (Qiagen) following the manufacturer’srecommendations and stored at �20 �C until use. Prepared cDNAwere exchanged so that each laboratory was provided with cDNAfrom all samples and time-points.
2.5. Quantitative two-step RT-PCR
The RT-qPCR was performed using TaqMan Gene ExpressionMaster Mix (Applied Biosystems). In laboratory A, samples wereanalysed for transcripts of CD8 and secreted immunoglobulin M(sIgM). In laboratory B, samples were analysed for Viperin (Vip),MHCI and MHCII. In laboratory C, samples were analysed for IFNa1,IFNg, TLR8, Mx protein (Mx) and CXC ligand 10 (CXCL10). Labora-tory D performed RT-PCR based detection and semi-quantificationof SAV. The primers and probe are as listed in Table 1. Elongationfactor 1ab (EF1ab) [23] was used as the reference gene (RG). Aprevious study has established that the EF1ab gene was satisfac-torily stable expressed during PD infection and that inclusion ofmore reference genes did not improve overall stability [24]. The RT-qPCR was carried out with cDNA corresponding to 15 ng of RNA.Each quantification target was amplified in triplicate with negativecontrol lacking the template, for each master mix. The transcriptlevels were normalized to EF1ab expression using the method ofPfaffl [25]. Further, the EF1ab-normalized transcript levels were
normalized against a calibrator which was the mean of ISAHi andISALo transcript levels at 0 wpc.
Detection of the SAV3 nsP1 gene was used for quantitative real-time RT-PCR analysis, and the PCR parameters used have beendescribed earlier [26]. Each sample was tested in triplicate andconsidered as positivewhen the Cq [27] value was�37.5. A fish wasdetermined SAV positive if a sample from that fish was foundpositive. To assess PCR efficiency, template dilutions were used togenerate standard curve. The load of SAV RNA was estimated intissue samples at 0, 1, 3, 6 and 8 wpc and in serum at 3 wpc.
2.6. Neutralization assay
Five serum samples from each sampling (0, 1, 3, 6 and 8 wpc)were analysed for their SAV-neutralizing activity. Serum was heatinactivated at 43 �C for 45 min and then two-fold dilutions of theserum in L-15 cell culture medium were mixed with an equal vol-ume of Irish SAV1 (F93-125) [1] containing 5 � 102 TCID50 ml�1,passed 9 times in CHSE-214 cells. Following incubation at 20 �C for2 h, 50 ml of each serumevirus mixture was added in three parallelsto Chinook salmon embryo cells (CHSE) grown in 96-well plates.After 10 days of incubation at 20 �C, the cells were fixed in 80%acetone and stained for SAV using an a-SAV1 mAb (5A5) as earlierdescribed [28]. Neutralizing activity was expressed as the recip-rocal value of the serum dilution that inhibited virus detection in50% of the inoculated cultures.
2.7. Statistical analysis
Statistical analyses and hierarchical clustering were performedusing the JMP� software (SAS Institute Inc., North Carolina, USA).Both RT-PCR based virus detection data and histopathology scoredata were analysed using the non-parametric ManneWhitney Utest. Data obtained by quantitative RT-PCR showed a significantlogarithmic distribution. These data were log2 transformed andthis resulted in a satisfactory normal distribution as judged by useof Normal Quantile plotting. The log2 transformed data were ana-lysed by ANOVA and a possible pair-wise correlation betweenmRNA levels of immune parameters was evaluated by the non-parametric Kendall’s Tau test. Hierarchical clustering analysis wasperformed using JMP 10 (SAS institute) and the Ward’s minimumvariance method on data standardised by mean and standarddeviation.
3. Results
3.1. Course of infection
Histopathologic evaluation of heart tissue sampled 4 and 6 wpcshowed little variation in the mean histology score for the ISAHistrain. In contrast, the ISALo group had a significant higher score at4 wpc (ManneWhitney U test, p < 0.01) (Fig. 1). Comparing thestrains, ISALo had a significant higher mean histology score at4 wpc but not at 6 wpc (ManneWhitney U test, p < 0.01). Theprevalence of fish having a histology score �2 (i.e. with clinical PD)was also significantly higher in ISALo (ManneWhitney U test,p < 0.05) as compared to the ISAHi strain (data not shown). Nomortality was observed during the experiment.
3.2. SAV in serum, heart, gills and head kidney
The load of SAV RNA was estimated by real-time PCR in tissuesamples from both ISALo and ISAHi fish at 0, 1, 3, 6 and 8 wpc and inserum at 3 wpc (Table 2). All fish sampled at 0 wpc and 1 wpc werenegative for SAV. Tissue samples from ISALo fish were more
Table 1Primers and probes used in Atlantic salmon quantitative RT-PCR. Primers and probewere design to span intron section and exon-junction sites. FAM: 6-carboxyfluorescein, MGB: minor groove binder, BHQ: black hole quencher, NFQ:non-fluorescent quencher (Applied Biosystems).
frequently positive for SAV RNA (SAVþ) at samplings 3, 6 and 8wpc,as compared to ISAHi fish. A single exception to this was seen in HKsamples at 6 wpc. This difference was however, only statisticallysignificant for heart tissue sampled at 8 wpc (ManneWhitney Utest, p < 0.01). There was no statistical significant difference in theprevalence or load of SAV in serum samples at 3 wpc for the twogroups. Compared to ISAHi group, the Cq values in SAV positiveISALo fish were lower, suggesting a higher virus load in the latterstrain. Within the ISAHi and ISALo strains, the number of SAVþsamples was relatively equal between the tissue types at 3 wpc. Theload of virus in the two groups, estimated by combining thenumber of RT-qPCR positive and the individual virus loads, peakedat 3 wpc. At 3 wpc, 65 and 75% of the ISAHi and ISALo fishrespectively, were viraemic which clearly indicated that the infec-tion was well-established in the groups. At 6 wpc, the number ofpositive head kidney samples had decreased in both strains butincreased for heart samples. At 8 wpc the virus was no longer foundin head kidney and only in a minor extent in the gills, but was stillpresent in 9 out of 10 heart samples from the ISALo group.
3.3. Neutralizing antibodies
No SAV-neutralising antibodies were detected at the time ofchallenge (0wpc) or at 1wpc (Table 3). At 3wpc,1 out of 5 fish from
each strain had a moderate neutralising antibody titre(50 < X < 80). At 6 wpc, 4 out of 5 fish from each strain haddeveloped high neutralising activity (>80). The two strains differedsubstantially only at 8 wpc, where 5 out of 5 ISAHi fish showed highneutralising activity (>80) compared to only 1 out of 5 fish for theISALo strain.
3.4. Innate immune parameters
3.4.1. mRNA levelsWith few and mostly modest exceptions, the mRNA levels of the
innate immune parameters IFNab, CXCL10, Mx, Vip, TLR8 and IFNg,differed little between ISAHi and ISALo strains irrespective of tissuetype and sampling time (Fig. 2). At 0 wpc however, the level ofCXCL10 was approximately 16 times higher in the heart of ISAHithan ISALo, but the difference disappeared in the course of theexperiment. A similar but much less pronounced trend wasobserved in heart and HK for Vip, Mx, TLR8, IFN-I and IFNg. ForIFNg, the levels in gills of both strains tended to increase slightlytowards 3 wpc. At 3 wpc Vip, Mx, CXCL10 and IFNg mRNA levelsincreased more abruptly, but notably the increase arose exclusivelyfrom fish inwhich SAV had been demonstrated in all three analysedtissues (SAV triple-positive).
3.4.2. ClusteringThe hierarchical clustering analysis of innate genes showed that
both the ISALo and ISAHi strains formed distinct clusters at 0 wpc(Fig. 3A). The ISAHi cluster, which was particular distinct, wascharacterised by relative high levels of innate transcripts in heartand head kidney. At 1 wpc, the distinction between the fish strainswas less pronounced and the strains now formed a looser clusterthat primarily was distinguished by relative high transcript levels ofall innate genes in the gills. Analysis of the samplings at 3 and 6wpc(Fig. 3B) showed a very distinct cluster formed by SAV triple-
Table 2RT-PCR based detection of SAV RNA in gill, head kidney (HK) and heart tissue of ISAHi (Hi) and ISALo (Lo) strains at 3, 6 and 8 wpc. Ratio in bold indicate the number of SAV RNApositive samples over the total number analysed. The Cq range indicates minimum and maximum Cq values. The numbers in the lower row of the sub-table for each samplingtime indicate fromwhich individual fish the SAV positive sample was taken. All fish sampled at 0 wpc and 1 wpc were negative for SAV and therefore not included in the table.
Gill-Lo Gill-Hi HK-Lo HK-Hi Heart-Lo Heart-Hi
No Pos Cq range No Pos Cq range No Pos Cq range No Pos Cq range No Pos Cq range No Pos Ct range
Ratio shown in bold numbers indicates the number of SAV RNA positive samples over the total number of samples analysed.
Fig. 1. Histology score for heart pathology at 4 and 6 weeks post-challenge for ISAHi(Hi) and ISALo (Lo) strains. Y-axis: histology score. Error bars represent SEM.
Table 3SAV-neutralising capacity of serum samples from groups ISAHi (Hi) and ISALo (Lo).Data is given as number of sera that had no/little neutralising activity (<20), in-termediate neutralising activity (<50) or high neutralising activity (�80). (N ¼ 5 foreach group at each sampling).
a One of the tested sera was virus positive, i.e. neutralizing effect could not beestimated in this serum.
b Two of the tested sera were virus positive, i.e. neutralizing effect could not beestimated in these sera.
S. Grove et al. / Fish & Shellfish Immunology 34 (2013) 789e798792
Author's personal copy
Fig. 2. RT-PCR quantification of Mx, TLR, Vip, CXCL10, IFNab and IFNg mRNA in tissue samples from Atlantic salmon challenged with SAV. X-axis: sampling time wpc (number) andfish strain (Hi/Lo). Y-axis: log2 transformed relative ratios (relative to mean of ISAHi and ISALo at 0 wpc). Due to the log2 transformation, the Y-value zero does not represent themean of ISAHi and ISALo at 0 wpc. Results for individual fish are indicated by diamonds; open diamonds indicate fish that are positive for SAV in all analysed tissues (triple-positive).For each organ, the horizontal line indicates the grand mean (of all samples). Mean diamonds show the means at each sampling time (midmost horizontal line) and the 95%confidence intervals (upper and lower corner of diamond). Sampling means differ significantly if the Overlap marks (upper and lower horizontal line in mean diamonds) do notoverlap in the vertical dimension.
Author's personal copy
Fig. 3. Hierarchical clustering analysis of innate and adaptive immune parameters in tissue samples from Atlantic salmon challenged with SAV. A. Innate genes at 0 and 1 wpc. B.Innate genes at 3 and 6 wpc. C. Adaptive genes at 0 and 1 wpc. D. Adaptive genes at 3 and 6 wpc. X-axis shows the clustering of the immune parameter mRNA levels in heart (He),head kidney (HK) and gills (Gi), respectively. Y-axis shows the clustering of individual fish with their sampling time (number) and strain (Hi/Lo). Each combination of sampling timeand fish strain is indicated by a specific colour. Symbols þ and � indicate SAV detection by RT-PCR in He, HK and Gi tissues, respectively. The þ indicates that SAV was detected byRT-PCR while the �, that SAV was not detected.
Author's personal copy
positive fish from both strains sampled 3 wpc. Fish in this clustergenerally had a high level of all innate transcripts in all tissues. Asecond cluster was formed by fish sampled at 6 wpc that were SAVpositive in at least one tissue. While a certain distinction betweenISAHi and ISALo strains could still be made at 3 and 6 wpc, thesedifferences were now reduced to the sub-cluster level(Supplementary Fig. 1).
3.4.3. CorrelationFor the innate parameters, the tendency to correlation varied
greatly with the sampling time, revealing a peak at 3 wpc for alltissues and both fish strains (Supplementary Fig. 2). In heart, thecorrelation peak appeared abruptly at 3 wpc for ISALo, whereas ithad a more gradual development in the ISAHi strain. In gill, theISAHi strain tended to have more pronounced correlation at earlysamplings than ISALo. Comparison of SAV negative with SAV posi-tive fish, irrespective of fish strain and sampling time, clearlyrevealed that correlation of innate immune parameters were muchstronger in the SAV positive fish in all tissues (data not shown).
3.5. Adaptive immune parameters
3.5.1. mRNA levelsFor MHCI andMHCII, transcript levels remained relatively stable
throughout the experiment, with little differences between fishstrains and sampling times (Supplementary Fig. 3). At 0 wpchowever, mRNA levels for both MHCI and MHCII were significantlyhigher for the ISAHi strain in all analysed tissues. By 1 wpc thesedifferences were less pronounced and only significant for MHCI inheart and for MHCII in gills. For sIgM and CD8, changes in transcriptlevels were more prominent (Fig. 4). Both ISAHi and ISALo strains
showed an increase in sIgM transcripts levels in HK from 0 wpc to1 wpc, but the ISAHi strain had consistently a significant highertranscript level at all samplings. While both strains had a similarearly increase in sIgMmRNA in gills from 0wpc to 1 wpc, there wasno general difference in the transcript levels between the twostrains. In heart tissue, sIgM levels in ISAHi and ISALo were similarat early time points but increased significantly at 3 wpc for ISALoand 6 wpc for ISAHi strains. For CD8, the transcript levels in HKshowed a slight and gradual decrease towards 6wpc/8 wpc for bothstrains. In gills, levels of CD8 mRNA were generally higher in theISAHi strain except for 6 wpc, where this strain experienced a sig-nificant drop. In heart tissue, CD8 transcript levels were similar inthe strains at early time point, but experienced an increase thatcoincided with the increase in sIgM mRNA, i.e. at 3 wpc for ISALoand at 6 wpc for ISAHi. In gill tissue, however, the minimumtranscript level for CD8 for ISAHi was at 6 wpc, i.e. the same timepoint where maximum was found in heart tissue.
3.5.2. ClusteringHierarchical clustering analysis of the adaptive genes at 0 and
1 wpc showed a clustering that primarily followed the fish strainand secondarily the sampling time, producing four distinct clusters(Fig. 3C). Fish sampled 3 and 6 wpc showed less distinct clusteringwhere the strains formed clusters at 3 wpc but not at 6 wpc(Fig. 3D). Clustering analysis of the individual sampling times, i.e. 3and 6 wpc respectively, confirmed the distinct clustering alongstrain at 3 wpc but not at 6 wpc (Supplementary Fig. 4). At 8 wpc,analysis based on sIgM and CD8 only, showed that the clusteringalong strains was again formed. In contrast to the innate genes,clusters based on adaptive gene data did at no time point follow anyobvious pattern in viral status.
Fig. 4. RT-PCR quantification of sIgM and CD8 mRNA in tissue samples from Atlantic salmon challenged with SAV. (See Fig. 2 for explanatory text).
S. Grove et al. / Fish & Shellfish Immunology 34 (2013) 789e798 795
Author's personal copy
3.5.3. CorrelationFor both ISAHi and ISALo strains, correlation between adaptive
parameters tended to be low irrespective of tissue and samplingtime (data not shown). One noteworthy exception was observed inheart at 6 wpc, where adaptive immune parameters were bothmore and statistically stronger correlated in the ISAHi strain than inISALo (Supplementary Fig. 5).
4. Discussion
In the present SAV cohabitant challenge experiment it wasfound that two strains of Atlantic salmon bred to high and lowresistance to ISA, showed a similar difference in susceptibility to PD.The disease was more severe in the ISALo group compared to theISAHi group, manifested by both higher prevalence and load of SAVin the ISALo group. These differences, however, levelled out from6 wpc and onwards.
It was shown that the two fish strains differed in immuneresponse to the infection, regarding both adaptive immune pa-rameters and the overall level of coordination of both innate andadaptive immune parameters. These immune parameter differ-ences likely contribute to the unequal PD susceptibility observed inthe two fish strains.
The mean histology score, as well as the prevalence of fish withclinical PD, was significantly higher in the ISALo group at 4 wpc.Regarding presence of SAV RNA, more fish of the ISALo strain wereSAV positive at 3 wpc and 8 wpc, but the difference was only sta-tistically significant in heart tissue at 8 wpc. Moreover, the Cq rangefor the SAV positive fish was lower in ISALo, suggesting higher virusRNA levels in this strain. A similar difference in prevalence, severityof lesions and in serological responses, due to an SAV infection, haspreviously been shown between commercial strains of Atlanticsalmon in Ireland [10].
Neutralising antibodies were detectable from 3 wpc in both fishstrains, but a difference in NAb levels was only observed at 8 wpcwhere ISAHi had the highest prevalence and titres. In contrast, theobserved difference in disease susceptibility was already man-ifested at 3e4wpc, as observed by histopathological and virologicalparameters, suggesting that the causal factors leading to this werealready at work prior to 3 wpc. It is hence less likely that differencesin NAb levels played a primary role for ISAHi being the less sus-ceptible strain. In accordance with this observation, a previousstudy could not explain demonstrated differences in SAV1 sus-ceptibility in three strains of Atlantic salmon by differences in levelsof NAb [10]. Neutralizing antibodies became detectable only afterSAV had been cleared fromplasma of infected fish [10,14]. The ISAHifish had high titres of NAb and few SAVþ hearts at 8 wpc, while theopposite was true for ISALo, which may suggest that NAb areimportant for SAV clearance in tissues. As demonstrated by passiveimmunisation experiments, fish receiving injection of NAb (serum)from PD convalescent fish are completely protected against sub-sequent SAV challenge [13,29].
Transcript levels of sIgM showed significant differences in tworespects during the challenge experiment. Firstly, the ISAHi strainconsistently had significantly higher sIgM transcript levels in headkidney. Secondly, sIgM transcript levels increased relative to 0 wpcin all tissues of both strains, although this increase came at differenttime points in the heart for the two strains. The difference in headkidney sIgM transcript levels between the strains is interestingbecause it appeared in a tissue that is central to systemic humoralimmunity, but also because it was present prior to the firstencounter with SAV and was maintained after the first increasefrom 0 wpc to 1 wpc. Taken together, a non-transient difference insIgM transcript levels between the two fish strains was present andcould thus be related to the difference in PD susceptibility. At 0wpc,
fish were naïve to SAV and hence the higher PD resistance observedin ISAHi is not likely based on pre-existing specificity to SAV anti-gens. A possibility could be that the observed higher level of sIgMalso reflects a larger population of B cell lineage cells in head kidneyof ISAHi fish. From a larger pool of B cells with more antigenspecificities to select from, a resulting specific antibody responsemay be more efficient, likely improving disease resistance. Ina recent study, increased levels of C4 in serum, an important factorfor classical complement activation, were found in SAV infectedsalmon [15]. In the same study, C4 levels were depleted in vacci-nated groups which showed high protection against PD. C4 con-sumption is a sensitive measure of classical activation, suggestinga role of antibody-mediated complement activation in protectionagainst piscine alphaviruses. The initial higher levels of sIgM foundin the ISAHi strain may lead to more efficient activation of com-plement mediated protection against SAV. In heart, but not in headkidney and gills, the increase of sIgM observed coincided closelywith a significant increase in CD8 mRNA levels. In ISAHi fish, sIgMand CD8 transcripts were further significantly correlated at both 6and 8 wpc. The observed increases in heart and gills suggest aug-mented activity and/or influx of plasmacells or precursors andcytotoxic T cells, respectively, and for sIgM it specifically indicateslocal antibody production. Similar peripheral antibody productionhas been demonstrated in Atlantic halibut (Hippoglossus hippo-glossus) where it was indicated that target tissues for nodavirusbecame major sites of antibody production [30]. The late occur-rence of the sIgM and CD8 mRNA increase in heart favours thepossibility that plasmacells/blasts and cytotoxic T cells migrating tothe infected target tissue are indeed SAV antigen specific. Con-versely, the early sIgM increase in gills suggests influx or activationof cells from the B cell lineage that are not yet antigen specific. AsIgM is the most abundant immunoglobulin type in salmonids [31],the levels of sIgM transcript are likely associated with the observedtitres of NAb. The level of CD8mRNAwas significantly higher in gillsof ISAHi at most samplings including at 0 wpc. Atlantic salmon gillshave recently been shown to harbour a lymphoid-like tissuehosting many T cells [32] and may thus be an important compart-ment for development of T cell responses. Similarly to sIgM in headkidney, the high levels of CD8 in ISAHi gills may be interpreted asa distinct presence of the cell type indicated by the CD8 cell marker,i.e. cytotoxic T cells. Again, a larger pool of T cells and antigenspecificities could endow ISAHi with a better capacity to launch anearlier and/or more efficient T cell response.
Compared to previous studies with viral salmon diseases, theinnate response observed in the current experiment was modest[33,34]. Transcript levels of several innate genes were at 0 wpcsignificantly higher in heart of ISAHi compared to ISALo, but themagnitude of these differences was relatively small. The mostpronounced response was seen at 3 wpc and both RT-qPCR dataand clustering analysis clearly indicate that the most distinct re-sponders were SAV triple-positive fish from both fish strains. Whilethis finding suggests a late innate response associated primarilywith systemic infection, it is possible that the sampling frequency(weeks) used in the experiment failed to identify an early (short-lived) innate response. For viperin and Mx, which are induced byIFN type I, a transcript level up-regulation was observed at 3 wpc.This however, was not accompanied by an increase in of IFNa1expression at this time point, probably reflecting that the IFNa1response had already waned. This latter would be in accordancewith previous results from Skjaeveland et al. [16], where thetranscript levels of this gene were higher at day 7 than at 4 wpc. Inmammals, alphaviruses are strong inducers of IFN type I, resultingin transcriptional up-regulation of genes with antiviral activities[17], but the virus may also strongly inhibit host synthesis of mRNAand protein. Such a cytotoxic effect and shut-down of protein
S. Grove et al. / Fish & Shellfish Immunology 34 (2013) 789e798796
Author's personal copy
synthesis, tentatively exerted by the SAV capsid protein, have beendescribed [1] and might have influenced the induction of immunegenes. As opposed to the IFN type I induced genes, salmon CXCL10is an IFNg-inducible gene, and its expression along with the otherinnate response genes was correlated with the expression of IFNgof both strains. A recent report shows that salmonid IFNg hasa potent antiviral activity against SAV and that this may in part alsobe linked to up-regulation of other innate parameters such as Mx,ISG15 and viperin [35].
Whereas the individual innate parameters may not convincinglybe correlated to the observed difference in disease susceptibility,the demonstrated changes and differences in correlation betweenthese parameters could be of interest. While the individual pair-wise correlations have to be interpreted with great caution, therevealed differences in range of correlationmay distinguish the twosalmon strains regarding their immunological response. Forexample, the observation of more correlated innate immune pa-rameters in gills of the ISAHi strain at 0 wpc is noteworthy. At 0 wpcthe fish had not been exposed to SAV and the correlation disparitymay thus reflect a difference in coordination of innate factors ingills. Better coordination of innate immune factors in this tissue,which is a possible port of virus entry, could make the ISAHi strainless susceptible to SAV infection.
In heart, correlations tended to appear very abrupt in ISALo andmore gradual, and with a less distinct peak in ISAHi. The extensivecorrelation pattern, generally observed at 3 wpc, most likely re-flects a major activation of the innate antiviral genes due to theintruding virus. As the ISAHi group was more resistant, it can bespeculated that a more gradual, and perhaps more host-controlled,development of the innate response in the SAV target tissue couldbe advantageous.
While innate immune parameters were highly correlated inboth ISAHi and ISALo strains in all tissues at 3 wpc, only the ISAHistrain showed a comparable correlation trend for adaptive pa-rameters and then only in heart at 6 wpc. At 8 wpc, only sIgM andCD8 were analysed, and their levels were highly correlated in ISAHibut not in ISALo. The correlations suggest a coordinated activity ofsIgMþ plasmacells/-blasts, CD8þ cytotoxic T cells and MHCIIþ an-tigen presenting cells on-site in the infected target organ.
The clustering analysis showed that ISAHi and ISALo formedseparate clusters at 0 wpc for both innate and adaptive immuneparameters.Whereas this ISAHi/ISALo clustering wasmaintained at1 and 3 wpc for the adaptive parameters, it was only partial at1 wpc for the innate parameters, and by 3 wpc it was completelyabsent. The clustering along strains at 0 wpc, prior to the encounterof SAV, suggests a pre-existing difference in immune parameters.
The fact that the distinct strain-specific clustering was resumedat 8 wpc (adaptive), implies that the differences seen at 0 wpc werenot accidental, but rather was a basic non-transient difference thatcan be correlated to the observed difference in disease resistance inthe two strains. The higher mRNA level of innate parameters seenin the ISAHi cluster at 0 wpc may suggest a generally increasedtranscription activity in this strain. In a recent QTL study on Atlanticsalmon, the transcription factor HIV-EP2/MBP-2 was suggested asone of the strong correlates for ISA resistance [36]. The collapse ofthe strain-specific clustering at 3 and 6 wpc for innate and at 6 wpcfor adaptive parameters most likely reflects the general impact ofthe viral infection on the immune system. The very distinct clusterformed by SAV triple-positive fish at 3 wpc (innate), is noteworthyas it seem to show that widespread (systemic) infection induceinnate parameters (at a late time point) while more confinedinfection do not.
The present study demonstrates that two strains of Atlanticsalmon genetically selected for resistance to ISA, also differ inresistance to pancreas disease in an experimental cohabitation
challenge. The differences in immune parameters observed be-tween the two salmon strains may provide important clues to thecauses underlying this resistance difference. As the ISA and SAVviruses differ in structure, virulence strategy and pathogenehostinteraction, the observed resistance increase likely includes fac-tors of a more general antiviral nature. Future comparison of im-mune responses in the ISAHi and ISALo strains to ISAV and SAVinfections will facilitate identification of immune parameters andresponses that are important to antiviral protection, both regardingthese specific viruses and viruses in general.
Acknowledgements
We would like to thank Ingebjørg Modahl, Guro Seternes, RandiFaller, Ingebjørg Sævareid and Irene Gabestad for excellent tech-nical assistance. Fish strains were kindly provided by SalmoBreedAS (Bergen, Norway). The work was supported by the ResearchCouncil of Norway, grant # 183196/S40, and byMSDAnimal Health.
Appendix A. Supplementary material
Supplementary data related to this article can be found online athttp://dx.doi.org/10.1016/j.fsi.2012.12.014.
References
[1] Nelson RT, McLoughlin MF, Rowley HM, Platten MA, McCormick JI. Isolation ofa Toga-like virus from farmed Atlantic salmon Salmo salar with pancreasdisease. Dis Aquat Org 1995;22:25e32.
[2] Weston J, Villoing S, Bremont M, Castric J, Pfeffer M, Jewhurst V, et al. Com-parison of two aquatic alphaviruses, salmon pancreas disease virus andsleeping disease virus, by using genome sequence analysis, monoclonalreactivity, and cross-infection. J Virol 2002;76:6155e63.
[3] Fringuelli E, Rowley HM, Wilson JC, Hunter R, Rodger H, Graham DA. Phylo-genetic analyses and molecular epidemiology of European salmonid alpha-viruses (SAV) based on partial E2 and nsP3 gene nucleotide sequences. J FishDis 2008;31:811e23.
[4] Hodneland K, Bratland A, Christie KE, Endresen C, Nylund A. New subtype ofsalmonid alphavirus (SAV), Togaviridae, from Atlantic salmon Salmo salar andrainbow troutOncorhynchusmykiss in Norway. Dis Aquat Org 2005;66:113e20.
[5] Kristoffersen AB, Viljugrein H, Kongtorp RT, Brun E, Jansen PA. Risk factors forpancreas disease (PD) outbreaks in farmed Atlantic salmon and rainbow troutin Norway during 2003e2007. Prev Vet Med 2009;90:127e36.
[6] McLoughlin MF, Graham DA. Alphavirus infections in salmonids e a review.J Fish Dis 2007;30:511e31.
[7] Houghton G. Kinetics of infection of plasma, blood leukocytes and lymphoid-tissue from Atlantic salmon Salmo salar experimentally infected with pancreasdisease. Dis Aquat Org 1995;22:193e8.
[8] Villoing S, Castric J, Jeffroy J, Le Ven A, Thiery R, Bremont M. An RT-PCR-basedmethod for the diagnosis of the sleeping disease virus in experimentally andnaturally infected salmonids. Dis Aquat Org 2000;40:19e27.
[9] Graham DA, Frost P, McLaughlin K, Rowley HM, Gabestad I, Gordon A, et al.A comparative study of marine salmonid alphavirus subtypes 1e6 using anexperimental cohabitation challenge model. J Fish Dis 2011;34:273e86.
[10] McLoughlin MF, Graham DA, Norris A, Matthews D, Foyle L, Rowley HM, et al.Virological, serological and histopathological evaluation of fish strain sus-ceptibility to experimental infection with salmonid alphavirus. Dis Aquat Org2006;72:125e33.
[11] McLoughlin MF, Nelson RT, Rowley HM, Cox DI, Grant AN. Experimentalpancreas disease in Atlantic salmon Salmo salar post-smolts induced bysalmon pancreas disease virus (SPDV). Dis Aquat Org 1996;26:117e24.
[12] Houghton G. Acquired protection in Atlantic salmon Salmo salar parr andpost-smolts against pancreas disease. Dis Aquat Org 1994;18:109e18.
[13] Houghton G, Ellis AE. Pancreas disease in Atlantic salmon: serum neutralisa-tion and passive immunisation. Fish Shellfish Immunol 1996;6:465e72.
[14] Desvignes L, Quentel C, Lamour F, Le Ven A. Pathogenesis and immune responsein Atlantic salmon (Salmo salar L.) parr experimentally infected with salmonpancreas disease virus (SPDV). Fish Shellfish Immunol 2002;12:77e95.
[15] Strandskog G, Villoing S, Iliev DB, Thim HL, Christie KE, Jorgensen JB. For-mulations combining CpG containing oliogonucleotides and poly I:C enhancethe magnitude of immune responses and protection against pancreas diseasein Atlantic salmon. Dev Comp Immunol 2011;35:1116e27.
[16] Skjaeveland I, Iliev DB, Strandskog G, Jorgensen JB. Identification and char-acterization of TLR8 and MyD88 homologs in Atlantic salmon (Salmo salar).Dev Comp Immunol 2009;33:1011e7.
S. Grove et al. / Fish & Shellfish Immunology 34 (2013) 789e798 797
Author's personal copy
[18] Ryman KD, Meier KC, Gardner CL, Adegboyega PA, Klimstra WB. Non-patho-genic Sindbis virus causes hemorrhagic fever in the absence of alpha/beta andgamma interferons. Virology 2007;368:273e85.
[19] Frolova EI, Fayzulin RZ, Cook SH, Griffin DE, Rice CM, Frolov I. Roles of non-structural protein nsP2 and alpha/beta interferons in determining the out-come of Sindbis virus infection. J Virol 2002;76:11254e64.
[20] Lopez-Doriga MV, Smail DA, Smith RJ, Domenech A, Castric J, Smith PD, et al.Isolation of salmon pancreas disease virus (SPDV) in cell culture and its abilityto protect against infection by the ‘wild-type’ agent. Fish Shellfish Immunol2001;11:505e22.
[21] Moriette C, LeBerre M, Lamoureux A, Lai TL, Bremont M. Recovery of a re-combinant salmonid alphavirus fully attenuated and protective for rainbowtrout. J Virol 2006;80:4088e98.
[22] Christie KE, Graham DA, McLoughlin MF, Villoing S, Todd D, Knappskog D.Experimental infection of Atlantic salmon Salmo salar pre-smolts by i.p. in-jection with new Irish and Norwegian salmonid alphavirus (SAV) isolates:a comparative study. Dis Aquat Org 2007;75:13e22.
[23] Olsvik PA, Lie KK, Jordal AEO, Nilsen TO, Hordvik I. Evaluation of potentialreference genes in real-time RT-PCR studies of Atlantic salmon. BMC Mol Biol2005;6.
[24] Lovoll M, Austbo L, Jorgensen JB, Rimstad E, Frost P. Transcription of referencegenes used for quantitative RT-PCR in Atlantic salmon is affected by viralinfection. Vet Res 2011;42.
[25] Pfaffl MW. A new mathematical model for relative quantification in real-timeRT-PCR. Nucleic Acids Res 2001;29.
[26] Hodneland K, Endresen C. Sensitive and specific detection of salmonidalphavirus using real-time PCR (TaqMan (R)). J Virol Methods 2006;131:184e92.
[27] Bustin SA, Benes V, Garson JA, Hellemans J, Huggett J, Kubista M, et al. TheMIQE guidelines: minimum information for publication of quantitative real-time PCR experiments. Clin Chem 2009;55:611e22.
[28] Graham DA, Jewhurst VA, Rowley HM, McLoughlin MF, Todd D. A rapidimmunoperoxidase-based virus neutralization assay for salmonid alpha-virus used for a serological survey in Northern Ireland. J Fish Dis 2003;26:407e13.
[29] Murphy TM, Drinan EM, Gannon F. Studies with an experimental model forpancreas disease of Atlantic salmon Salmo Salar L. Aquacult Res 1995;26:861e74.
[30] Grove S, Johansen R, Reitan LJ, Press CM, Dannevig BH. Quantitative inves-tigation of antigen and immune response in nervous and lymphoid tissues ofAtlantic halibut (Hippoglossus hippoglossus) challenged with nodavirus. FishShellfish Immunol 2006;21:525e39.
[31] Salinas I, Zhang Y-A, Sunyer JO. Mucosal immunoglobulins and B cells ofteleost fish. Dev Comp Immunol 2011;35:1346e65.
[32] Koppang EO, Fischer U, Moore L, Tranulis MA, Dijkstra JM, Kollner B, et al.T cells assemble in the thymus, spleen and in novel interbranchial lymphoidtissue. J Anat 2010;217:728e39.
[33] Ingerslev HC, Ronneseth A, Pettersen EF, Wergeland HI. Differentialexpression of immune genes in Atlantic salmon (Salmo salar L.) challengedintraperitoneally or by cohabitation with IPNV. Scand J Immunol 2009;69:90e8.
[34] Jorgensen SM, Hetland DL, Press CM, Grimholt U, Gjoen T. Effect of early in-fectious salmon anaemia virus (ISAV) infection on expression of MHC path-way genes and type I and II interferon in Atlantic salmon (Salmo salar L.)tissues. Fish Shellfish Immunol 2007;23:576e88.
[35] Sun B, Skjaeveland I, Svingerud T, Zou J, Jorgensen J, Robertsen B. Antiviralactivity of salmonid gamma interferon against infectious pancreatic necrosisvirus and salmonid alphavirus and its dependency on type I interferon. J Virol2011;85:9188e98.
[36] Li J, Boroevich KA, Koop BF, Davidson WS. Comparative genomics identifiescandidate genes for infectious salmon anemia (ISA) resistance in Atlanticsalmon (Salmo salar). Mar Biotechnol (NY) 2011;13:232e41.
S. Grove et al. / Fish & Shellfish Immunology 34 (2013) 789e798798
