Identification of Quantitative Trait Loci Associated with ...

Identification of Quantitative Trait Loci Associated with Resistance to Viral Haemorrhagic Septicaemia (VHS) in Turbot (Scophthalmus maximus): A Comparison Between Bacterium, Parasite and Virus Diseases Silvia T. Rodríguez-Ramilo, Roberto De La Herrán, Carmelo Ruiz-Rejón, Miguel Hermida, et al. Marine Biotechnology An International Journal Focusing on Marine Genomics, Molecular Biology and Biotechnology ISSN 1436-2228 Mar Biotechnol DOI 10.1007/s10126-013-9544-x

1 23

Your article is protected by copyright and all rights are held exclusively by Springer Science +Business Media New York. This e-offprint is for personal use only and shall not be selfarchived in electronic repositories. If you wish to self-archive your article, please use the accepted manuscript version for posting on your own website. You may further deposit the accepted manuscript version in any repository, provided it is only made publicly available 12 months after official publication or later and provided acknowledgement is given to the original source of publication and a link is inserted to the published article on Springer's website. The link must be accompanied by the following text: "The final publication is available at link.springer.com”.

1 23

Author's personal copy Mar Biotechnol DOI 10.1007/s10126-013-9544-x

ORIGINAL ARTICLE

Identification of Quantitative Trait Loci Associated with Resistance to Viral Haemorrhagic Septicaemia (VHS) in Turbot (Scophthalmus maximus ): A Comparison Between Bacterium, Parasite and Virus Diseases Silvia T. Rodríguez-Ramilo & Roberto De La Herrán & Carmelo Ruiz-Rejón & Miguel Hermida & Carlos Fernández & Patricia Pereiro & Antonio Figueras & Carmen Bouza & Miguel A. Toro & Paulino Martínez & Jesús Fernández

Received: 11 March 2013 / Accepted: 12 September 2013 # Springer Science+Business Media New York 2013

Abstract One of the main objectives of genetic breeding programs in turbot industry is to reduce disease-related mortality. In the present study, a genome scan to detect quantitative trait loci (QTL) affecting resistance and survival to viral haemorrhagic septicaemia (VHS) was carried out. Three full-sib families with Electronic supplementary material The online version of this article (doi:10.1007/s10126-013-9544-x) contains supplementary material, which is available to authorized users. S. T. Rodríguez-Ramilo Departamento de Bioquímica, Genética e Inmunología, Facultad de Biología, Universidad de Vigo, 36310 Vigo, Spain S. T. Rodríguez-Ramilo : J. Fernández (*) Departamento de Mejora Genética Animal, Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria, Ctra. A Coruña Km. 7.5, 28040 Madrid, Spain e-mail: [email protected] R. De La Herrán : C. Ruiz-Rejón Departamento de Genética, Facultad de Ciencias, Universidad de Granada, Avenida Fuentenueva s/n, 18071 Granada, Spain M. Hermida : C. Fernández : C. Bouza : P. Martínez Departamento de Genética, Facultad de Veterinaria, Universidad de Santiago de Compostela, Campus de Lugo, 27002 Lugo, Spain P. Pereiro : A. Figueras Instituto de Investigaciones Marinas, CSIC, Eduardo Cabello 6, 36208 Vigo, Spain M. A. Toro Departamento de Producción Animal, ETS Ingenieros Agrónomos, Universidad Politécnica de Madrid, Ciudad Universitaria, 28040 Madrid, Spain

approximately 90 individuals each were genotyped and evaluated by linear regression and maximum likelihood approaches. In addition, a comparison between QTL detected for resistance and survival time to other important bacterial and parasite diseases affecting turbot (furunculosis and scuticociliatosis) was also carried out. Finally, the relationship between QTL affecting resistance/survival time to the virus and growth-related QTL was also evaluated. Several genomic regions controlling resistance and survival time to VHS were detected. Also significant associations between the evaluated traits and genotypes at particular markers were identified, explaining up to 14 % of the phenotypic variance. Several genomic regions controlling general and specific resistance to different diseases in turbot were detected. A preliminary gene mining approach identified candidate genes related to general or specific immunity. This information will be valuable to develop marker-assisted selection programs and to discover candidate genes related to disease resistance to improve turbot production. Keywords QTL . Turbot . Viral haemorrhagic septicaemia . Linear regression . Maximum likelihood

Introduction The great commercial value of turbot (Scophthalmus maximus) has promoted its intensive culture during the last decade. In Europe, production reached 9,142 tons in 2009 (FEAP 2010), and it is expected to double this production in 2014. This species, native to Europe, was also introduced in

Author's personal copy Mar Biotechnol

Chile and, more recently, in China, which has produced an annual level of 50,000–60,000 tons in the past few years — becoming the largest producer of turbot in the world (FAO 2010). Currently, the main objectives of genetic breeding programs in turbot are: (1) increasing growth rate, (2) controlling sex ratio and (3) improving disease resistance. Nowadays, the culture of this fish is a well-established process. However, several infectious diseases including those caused by parasites (Álvarez-Pellitero 2008), bacteria (Toranzo et al. 2005) and viruses (Barja 2004), comprise one of the most relevant limiting factors, causing severe economic losses in many cases. Viral haemorrhagic septicaemia virus (VHSV, Novirhabdovirus) is responsible of an important disease affecting different fish species like salmonids (Castric and de Kinkelin 1980; Hørlyck et al. 1984; Wolf 1988) and marine species (Meyers and Winton 1995; Mortensen et al. 1999). The VHSV has been also detected in flatfish such as Japanese flounder (Paralichthys olivaceus; Isshiki et al. 2001) and turbot (Schlotfeldt et al. 1991; Ross et al. 1994). This etiological agent is a serious threat in farmed fish with a potential economic impact in turbot culture as was previously reported (Ross et al. 1994). Although the available information about natural outbreaks of VHS in turbot is scarce, numerous experimental infections have revealed the devastating capacity of this virus and, as consequence, the powerful risk for the turbot farming industry (Snow and Smail 1999; Kin et al. 2001; López-Vazquez et al. 2007). Furthermore, different well established S. maximus families showed differential susceptibility to VHSV challenge, reflecting the importance of the genetic selection (Díaz-Rosales et al. 2012). Currently, there are no vaccines available to prevent VHS disease, which has promoted the idea of achieving a more robust broodstock resistant to VHSV through genetic breeding programs. For this reason, several studies have been conducted to investigate the genomic architecture underlying the resistance to this etiological agent in several aquaculture species, for example rainbow trout (Slierendrecht et al. 2001; Verrier et al. 2013), Japanese flounder (Byon et al. 2006) and Atlantic salmon (Acosta et al. 2005). Ongoing breeding programs in aquaculture usually involve traits associated with growth and cold tolerance, but examples also exist on disease resistance-related traits (Gjedrem 2010; Rye et al. 2010). In aquaculture species, the heritability estimates for resistance to different viral species range from 0.04 to 0.79 (reviewed by Ødegård et al. 2011). Henryon et al. (2005) reported a heritability estimate for VHSV resistance of 0.57 in rainbow trout. These estimates indicate an important additive genetic component susceptible to be employed for obtaining more robust broodstock against VHSV. The variability of a quantitative trait could be attributable both to many loci with a small effect and also to a reduced number of genes with a large effect. Quantitative trait loci (QTL) designate those genes or genomic regions with a significant

effect which can be detected through the association between trait phenotypes and genetic markers to which they are tightly linked (Mackay 2001). The allele segregation at genetic markers can be employed in a genome scan to uncover the number, position and effect of QTL (Lynch and Walsh 1998). Consequently, genetic markers associated with QTL can be employed to perform marker-assisted selection (MAS) or gene-assisted selection (GAS) if the actual gene affecting the trait could be determined. In recent years, several QTL for disease resistance have been reported in other aquaculture species such as the Eastern oyster, European flat oyster and Japanese flounder (reviewed by Ødegård et al. 2011), rainbow trout (Ozaki et al. 2001), sea bass and sea bream (Massault et al. 2010, 2011). Moreover, two examples of MAS programs for resistance to infectious pancreatic necrosis (IPN) and lymphocystis disease is being implemented in the Atlantic salmon (Houston et al. 2008, 2009; Moen et al. 2009) and Japanese flounder (Fuji et al. 2007; Ozaki et al. 2012) industries, respectively. Having an accurate linkage map is crucial for the task of searching for QTL, where the density of genetic markers is an important factor in its construction and power to detect QTL. Bouza et al. (2008) reported a consensus map in turbot, including a total of 273 microsatellites clustered in 26 linkage groups, comprising 1,343.2 cM length, with an average distance between markers of 6.5 cM. This reported consensus map has proved to be dense enough to detect QTL in turbot (e.g., Sánchez-Molano et al. 2011; RodríguezRamilo et al. 2011, 2013), and it is within the range suggested by Dekkers and Hospital (2002) to obtain reliable results. Obviously, increasing marker density could help to reduce the average distance between markers, refine the maps and, consequently, the accuracy in the detection of disease resistance genomic regions. The detection of QTL in turbot has been initiated recently to increase growth rate (Ruan et al. 2010; Sánchez-Molano et al. 2011), to control sex ratio (Martínez et al. 2009) and to improve disease resistance to furunculosis caused by the bacteria Aeromonas salmonicida (Rodríguez-Ramilo et al. 2011) and to scuticociliatosis originated by the parasite Philasterides dicentrarchi (RodríguezRamilo et al. 2013). The main objective of this study is to carry out a genome scan to detect QTL affecting resistance and survival time to VHSV in three turbot families applying linear regression (LR) and maximum likelihood (ML) approaches. In addition, taking advantage of previous QTL studies on turbot disease resistance (Rodríguez-Ramilo et al. 2011, 2013) and growth (Sánchez-Molano et al. 2011), we try to identify specific and general disease-resistance genomic regions, and to evaluate the possible relationship between growth and VHSV resistance. Finally, we addressed a preliminary comparative mapping approach with model fish species to identify candidate genes related to VHSV resistance on suitable selected QTL.


Materials and Methods

Genetic Map for QTL Screening

Families Three full-sib turbot families with approximately 90 individuals each were used to identify QTL. These families were obtained from the breeding program of Stolt Sea Farm S.A., a company located in NW Spain. Since no selected strains for disease resistance exist in turbot, families were chosen trying to pick up the highest genetic and phenotypic differences among offspring by selecting parents of divergent origins. Thus, families were founded with unrelated grandparents coming from different Atlantic origins when feasible. Therefore, a three-generation pedigree was available for all of them. This enabled us to know the linkage-phase between markers for a more reliable statistical analysis. This process finally led to three independent full-sib family analyses and to a joint analysis with all the data set.

The panel of markers used for genotyping the selected animals and for QTL identification was reported by Martínez et al. (2009) and is based on the consensus map by Bouza et al. (2007), the new EST-linked microsatellites by Bouza et al. (2008) and the centromere mapping by Martínez et al. (2008). All families were analyzed taking as reference the same consensus genetic map, thus maintaining the same genetic distances between markers. Table 1 shows the number, location and coverage of the analyzed microsatellites for each family and for the total data set. The average distance between markers for each family ranged between 16.98 and 19.41 cM, being below the minimum distance proposed for QTL detection (Dekkers and Hospital 2002). For the total data set, the average distance was lower (16.01 cM), because the number of genotyped markers involving the three families was higher.

Trait Measurement

QTL Analyses

All offspring of each family (approximately 150 individuals per family) were intracelomically injected with a highly virulent strain of VHSV (UK-860/94) (Ross et al. 1994) when individuals showed a mean weight around 5 g. The dose was adjusted to 5×106 TCID50/g of body weight for each family, resulting in 5×107 TCID50/fish in a volume of 100 μl for the family Fam -1 and 1.5×107 TCID50/fish in 50 μl for the families Fam -2 and Fam -3 . After injection, two disease resistance related traits were evaluated: resistance (Re) and survival time (Su). Around 90 individuals for each family were selected to evaluate both traits (approximately the 45 most resistant and the 45 most sensitive individuals). The dichotomous trait Re was defined as the survival vs. nonsurvival status of individuals at the end of the experiment. The trait Su was defined as the number of elapsed days until the individual died or the experiment finished. Consequently, this is a censored trait since all individuals still alive at the end of the experiment were scored with the same value for survival time. The experiment was prolonged up to a maximum of 30 days in total (when no deaths were observed during several days), trying to increase phenotypic variance for survival time. Also, total length and weight were measured in each individual at the time of death for non-resistant and at the end of the experiment for resistant ones. From these data, weight, length and Fulton's condition factor (Fulton 1902) were obtained for each individual. Fulton's condition factor is a measure of a fish fatness computed as 100 × W e/L e, where W e is the body weight of the fish (in grams) and Le is the length of the fish (in centimetres). Non-parametric correlations were obtained between diseaseand growth-related traits to evaluate the relationship between both data sets.

The main goal of this study was to identify VHSV resistancerelated QTL, but growth-related QTL could also be detected using information on the length and weight of each fish. This enabled to evaluate a possible relationship between growth and disease resistance. The programs GridQTL (Seaton et al. 2006) and QTLMap (Gilbert et al. 2008) were used to detect QTL. The former software (http://www.gridqtl.org.uk/) implements an LR methodology, considering the linkage phase between markers according to pedigree information. The default regression method (Haseman and Elston 1972) for QTL linkage analysis was applied, and the genome and chromosomewide significance thresholds at p =0.05 and p =0.01 were estimated with a permutation test of 10,000 iterations. QTLmap software (http://www.inra.fr/qtlmap) detects QTL through interval mapping using an ML test. To determine the significance level, a simulation with 10,000 iterations was performed for each trait and linkage group, with heritability set at 0.10 for disease traits (Wang et al. 2010) and at 0.45, 0.30 and 0.20 for weight, length and Fulton's factor, respectively

Table 1 Screening figures in the three families analyzed for QTL identification Family

NM

LG

MLe

D

NMG

Fam-1 Fam-2 Fam-3 Total

83 78 91 93

23 23 23 23

1,186.85 1,218.74 1,301.17 1,323.02

17.69 19.41 16.98 16.01

3.61 3.39 3.96 4.04

NM number of markers, LG total number of linkage groups, MLe map length (in cM), D average distance between markers (in cM), NMG mean number of microsatellites per linkage group


(Gjerde et al. 1997). The consensus turbot map marker positions were used for both methods to identify QTL and establish their confidence intervals. An outbred full-sib model was used, and a QTL was considered suggestive when significance was between 5 % and 1 % at chromosome-wide level, and significant when significance was below 1 % at chromosome-wide level or when significance was below 5 % at genome-wide level. These thresholds also allowed establishing a confidence interval around the estimated position of the QTL. Weight and length were included as covariates within the model to reduce stochasticity when detecting QTL for resistance and survival time to VHSV. The model applied to detect QTL was y =μ +X β +Zq +e, where y is the vector of phenotypic values, μ is the overall mean, β is the vector of covariates, q is the vector of QTL associated effects, e is the vector of random errors, and X and Z are incidence matrices. Body length and weight were included as covariates. Association Analyses Within the linkage groups where a significant or suggestive QTL was found, a one way ANOVA was performed on the phenotypic values of the progeny for each family using individual genotypes for single markers. The objective was to detect associations between markers and traits by estimating the between-genotype component of the observed phenotypic variance (i.e., differences attributable to the different marker genotypes). To avoid false positives due to multiple testing, a simple Bonferroni correction was performed for all tests involving markers within the same linkage group. Each ANOVA also provided a corrected R 2 value useful in estimating the reduction of the overall phenotypic variance of the trait due to the model fitting, thus providing the proportion of the trait variance explained by the given marker genotypes. Comparative Mapping Comparative mapping between the turbot genetic map against model Acanthopterygii teleost chromosomes (Bouza et al. 2012; Tni, Tetraodon nigroviridis v.8.61; Tru, Takifugu rubripes v.5; Ola, Oryzias latipes v.1.61; Gac, Gasterosteus aculeatus v.1.61) was used for identifying putative candidate genes around selected QTL for VHS resistance traits: (1) VHS specific QTL or (2) colocalized VHS-QTL in more than one family background with previously reported QTL for resistance to bacterial and parasite diseases in turbot (RodríguezRamilo et al. 2011, 2013). Gene mining and functional analysis of the extracted gene lists was carried out by using the BioMart bioinformatic tool (www.ensembl.org) starting from significantly associated or closely linked markers to the selected QTL positions. This approach provided significant

sequence matches against model genomes, focusing on particularly informative syntenic patterns with Ola and Gac chromosomes (Bouza et al. 2012).

Results and Discussion Trait Values Table 2 shows the statistics of the resistance-related traits (survival time and resistance) and growth-related traits (weight, length and Fulton's condition factor) for each family and for all data set. The standard error ranged between 0.53 and 1.04 days for survival time and 0.03 – 0.05 for resistance. A greater power for detection of QTL for resistance relatedtraits is expected when 50 % survival is achieved after challenges, as in family Fam-1. However, although a lower mean resistance was observed for families Fam -2 and Fam-3 and for the total data set (9 %, 25 % and 26 %, respectively), this did not preclude the detection of QTL for resistance relatedtraits in those data sets (Table 3). Mean weight, length and Fulton's condition factor ranged between 2.58 and 9.46 g, 5.47 and 8.36 cm and 1.56 and 1.65, respectively, with family 1 showing the highest size figures (Table 2). QTL Analyses Table 3 shows the location of the detected QTL for resistancerelated traits with both methodologies (LR and ML) within each linkage group (LG). No QTL were detected at genomewide level. In contrast with our previous data with other pathogens (Rodríguez-Ramilo et al. 2011, 2013) low correspondence between both statistical methods was observed. Several QTL detected in the single family analysis could not be detected when analyzing the total data set (LG5, LG8, LG15, LG17, LG20, LG21), and only one QTL not found in the single family analysis was detected in this analysis (LG2). The reduced power of detection with the whole data set was in part due to the fact that founders of each family come from different strains and, probably different informative QTL and genetic backgrounds exist between families, which may influence QTL detection. Moreover, differences observed for the means of the recorded traits (see Table 2) point towards the existence of other uncontrolled factors in each family (temperature, salinity, oxygenation, etc.), which could mask the effects of QTL and, thus, reduce the ability for the detection of QTL related with resistance. Consequently, these factors could explain the loss of power when using all the information together and the scarce correspondence between results in different families. Since we are dealing with outbred populations, different markers may be segregating in different families and, therefore, their informativeness is not the same. The detection of

Author's personal copy Mar Biotechnol Table 2 Statistics of the measured traits in the three families analyzed (±standard error)

Re resistance (survival vs. nonsurvival status of individuals after 30 days post-challenge), Su survival time (number of elapsed days until the individual died or after 30 days post-challenge), N number of individuals

Family

Weight (g)

Length (cm)

Fulton's condition factor

Trait

N

Mean

Fam-1

9.46±0.30

8.36±0.08

1.57±0.01

Fam-2

2.58±0.07

5.47±0.04

1.56±0.02

Fam-3

2.99±0.06

5.65±0.04

1.65±0.02

Re Su Re Su Re

91 91 93 93 96

0.45±0.05 19.81±1.04 0.09±0.03 13.20±0.60 0.25±0.04

Su Re Su

96 280 280

14.19±0.59 0.26±0.03 16.49±0.53

Total

4.96±0.21

6.47±0.09

the same QTL in different families depends on the frequencies of the involved alleles. It is even possible that the causal mutation is fixed in one family and, thus, the QTL cannot be detected. Consequently, when QTL are detected in similar positions in different families (LG6 in Fam -1 and Fam-3; see Table 3) the confidence on the QTL increases. Also, since we are using outbred families instead of, for example, an F2 design, the power of QTL analysis in our study is limited. However, considering that families come from unrelated and genetically divergent grandparents from natural populations of the Atlantic area, identified QTL in our study could shed reliable information on the genetic architecture of disease resistance related traits in turbot. Nevertheless, the markers associated with the detected QTL in the present study should be genotyped (and their effect verified) in other turbot families before their use in breeding programs. Additionally, new markers are to be developed in the detected regions and used to refine the position of the QTL.

1.59±0.01

As shown in Table 3, several QTL were detected for resistance and survival time at close positions within the same linkage group. The main reason for this concordance, as previously reported by Rodríguez-Ramilo et al. (2011, 2013), is the high correlation expected between both traits, the same genes being likely involved in their underlying mechanisms. In fact, Spearman correlation between resistance and survival time ranged between 0.75 and 0.91 (p