Skip to main content
NCBI home page
Genetics logoLink to Genetics
. 2010 Sep;186(1):97–107. doi: 10.1534/genetics.110.118158

The Transcriptional Landscape of Cross-Specific Hybrids and Its Possible Link With Growth in Brook Charr (Salvelinus fontinalis Mitchill)

Bérénice Bougas *,1, Sarah Granier †, Céline Audet †, Louis Bernatchez *
PMCID: PMC2940315  PMID: 20551437

Abstract

The genetic mechanisms underlying hybridization are poorly understood despite their potentially important roles in speciation processes, adaptative evolution, and agronomical innovation. In this study, transcription profiles were compared among three populations of brook charr and their hybrids using microarrays to assess the influence of hybrid origin on modes of transcription regulation inheritance and on the mechanisms underlying growth. We found that twice as many transcripts were differently expressed between the domestic population and the two wild populations (Rupert and Laval) than between wild ones, despite their deeper genetic distance. This could reflect the consequence of artificial selection during domestication. We detected that hybrids exhibited strikingly different patterns of mode of transcription regulation, being mostly additive (94%) for domestic × Rupert, and nonadditive for Laval × domestic (45.7%) and Rupert × Laval hybrids (37.5%). Both heterosis and outbreeding depression for growth were observed among the crosses. Our results indicated that prevalence of dominance in transcription regulation seems related to growth heterosis, while prevalence of transgressive transcription regulation may be more related to outbreeding depression. Our study clearly shows, for the first time in vertebrates, that the consequences of hybridization on both the transcriptome level and the phenotype are highly dependent on the specific genetic architectures of crossed populations and therefore hardly predictable.

LONG viewed as a dead end in animals, hybridization is increasingly considered as potentially positive from an evolutionary perspective (Barton 2001; Burke and Arnold 2001; Rieseberg et al. 2003; Tallmon et al. 2004; Nolte et al. 2006; Baak and Rieseberg 2007). Hybridization is also of interest for plant and animal production, because it may generate desirable phenotypic novelty through heterosis, in which the offspring have a more advantageous phenotype than their parents. However, hybridization can also lead to outbreeding depression, by which the offspring express a disadvantageous phenotype relative to their parents. Such hybrid phenotypic traits, expressed outside the range normally observed in the parental lines, are also referred to as transgressive phenotypes.

Numerous genetically based mechanisms have been invoked to explain transgressive phenotypes, including epistasis, overdominance, and additive allelic interactions (reviewed by Rieseberg et al. 1999). It has also been hypothesized that the crossing of inbred lines characterized by distinct gene expression levels could lead to advantageous transgressive phenotypic (Birchler et al. 2003; Cui et al. 2006; Stupar and Springer 2006; Swanson-Wagner et al. 2006; Stupar et al. 2008). One approach used to study the genetics of phenotypic traits is to measure transcript expression in the hybrids relative to their parents, but such studies have been in the past limited to mouse and Drosophila sp. (Reiland and Norr 2002; Michalak and Noor 2003; Gibson et al. 2004; Ranz et al. 2004; Cui et al. 2006; Hughes et al. 2006; Rottscheidt and Harr 2007). Other studies have also focused on elucidating the genomic basis of heterosis of mollusc production (Hedgecock et al. 2007) and plant production (Auger et al. 2005; Swanson-Wagner et al. 2006; Lippman and Zamir 2007; Ge et al. 2008; Stupar et al. 2008). Comparative studies of genome-wide transcriptomes of parental and hybrid crosses also revealed that hybrids tended to express numerous genes at a level outside the range observed in their parental lines (Gibson et al. 2004; Ranz et al. 2004; Auger et al. 2005; Cui et al. 2006; Hughes et al. 2006; Stupar and Springer 2006; Swanson-Wagner et al. 2006; Hedgecock et al. 2007; Rottscheidt and Harr 2007; Ge et al. 2008; Roberge et al. 2008; Stupar et al. 2008; Normandeau et al. 2009; Renaut et al. 2009). Thus, divergent lineages can accumulate divergences in gene regulation networks through time and these divergences often result in misexpressed genes (reviewed by Landry et al. 2007), which could affect the viability or fitness of individuals. In hybrids from subspecies of Mus, an overabundance of misexpressed transcripts was found in the testis relative to other tissues, including brain and liver (Rottscheidt and Harr 2007). In addition to this variation across tissues, the proportions of additive vs. nonadditive expression inheritance may vary between hybrids from different populations within species (Rottscheidt and Harr 2007; Stupar et al. 2008), sometimes in a contradictory manner. For example, in Drosophila, Gibson et al. (2004) found a prevalence of nonadditivity, whereas Hughes et al. (2006) found the opposite. These varying results seemed to depend on both methodological and genetic factors. Thus, the prevalence of nonadditivity could be linked with the X chromosome, as a bias was observed between males and females and very few additive genes were X linked (Gibson et al. 2004). Furthermore, these differences could also be linked to the highly homozygous lines used by Hughes et al. (2006), a situation which tends to increase the additive portion of expression inheritance observed. These two studies also differed in the methods used for the analysis; they used different criteria to calculate the dominance effects and different α level in their statistical analysis. However, Hughes et al. (2006) suggested that the differences between these two studies likely lie in genetic architecture divergence among the lines being crossed.

Genetic architecture is defined as the sum of the interacting genetic dimensions that lead to a given phenotype (Hansen 2006; Lynch 2007) and has been shown to vary in a population-specific manner (Lavagnino et al. 2008). As a result, it is especially difficult to predict the phenotype of hybrids. To date, very few studies have investigated the influence of population-specific genetic architecture on gene transcription regulation among populations of a same species. To our knowledge, only one study in maize (Stupar et al. 2008) examined hybrids between more than two crosses. These authors aimed to evaluate whether there was a link between heterosis at phenotypic traits and the genetic distance of the parental strains crossed, as well as the observed patterns of transcription regulation inheritance. However, they did not find any solid link between these parameters.

Brook charr (Salvelinus fontinalis) is an economically important salmonid native to eastern North America; in Québec, it represents 50% of all freshwater aquaculture production. Using microarrays, the general objective of this study was to document the differences between pure brook charr populations in terms of gene expression and their modes of transcription regulation inheritance in the F1 hybrids. Thus, F1 hybrid crosses descending from three genetically distinct populations were performed in order to investigate the phenotypic and genomic responses of these F1 hybrids relative to their parental populations. Our results show that the three parental populations and their hybrids differ strikingly in their overall patterns of gene expression. Namely, the number of significant transcripts and impacted biological functions vary in the three possible comparisons among pure and hybrid crosses. Moreover, the mode of transcription regulation (additivity or nonadditivity) in hybrids was dependent on which parental population it was compared to. We interpreted these patterns of genome-wide transcription in terms of their possible link to the contrasting size phenotypes observed in the same hybrid crosses.

MATERIALS AND METHODS

Fish crosses:

Juvenile brook charr (yolk sac resorption stage) were produced using breeders from three genetically distinct populations: a domestic (D) population, in use for aquaculture in Québec, Canada, for more than a hundred years, as well as the Laval (L) and Rupert (R) populations. The L population is derived from an anadromous population originating from the Laval River near Forestville (north of the St. Lawrence River, Québec) (Castric and Bernatchez 2003), whereas the R population originates from a freshwater wild population of the Rupert River, which drains into James Bay in northwestern Quebec (Fraser et al. 2005). On the basis of estimates of Shriver et al. (1995) using microsatellite data, these three populations were highly differentiated; L and R populations were separated by 13.3 Dsw (genetic distance), with the D population being about equally genetically distant (about 6.7 Dsw) from the two others (Martin et al. 1997).

Breeders from the L population were kept in captivity for three generations at the Institut des Sciences de la Mer à Rimouski (Québec) and at the Laboratoire de Recherche des Sciences Aquatiques (Laboratoire de Recherche des Sciences Aquatiques, LARSA, Laval University, Québec) for the R population. Breeders from the D population were obtained from the Pisciculture de la Jacques Cartier (Cap-Santé, Québec). In 2005, 10 sires of each population (L, R, or D) were crossed with 10 dams (L, R, or D) to generate 10 full-sib outbreed families per pure and hybrid crosses. Three pure and three hybrid crosses were generated (RL, LD, DR, L, D, and R). For the hybrid crosses, the first letter corresponds to the mother's origin. All families were kept separately at the LARSA under identical controlled conditions. Fertilized eggs were incubated at 6°. After hatching, the progeny were kept at 8°, with a photoperiod of 12 hours of light and 12 hours of darkness.

Microarray experiment:

Sampling:

One sexually undifferentiated juvenile was randomly sampled for eight families in each of the crosses and immediately frozen in a mix of dry ice and alcohol. Experiments were conducted on whole, unpooled individuals.

RNA extraction, labeling, and cDNA hybridization:

Total RNA was extracted with the PureLink Micro-to-Midi total RNA purification system kit and then treated using DNase I, amplification grade (1 unit/μl) (Invitrogen, Carlsbad, CA) following the manufacturer's protocol. Total RNA was further purified by ultracentrifugation using microcon (Millipore) spin columns. RNA quality and integrity was controlled with an Experion automated electrophoresis station and RNA HighSens chips (Bio-Rad, Hercules, CA). For each sample, 12 μg total RNA was retrotranscribed and the cDNA samples labeled using Genisphere 3DNA Array 50 kit, Invitrogen's Superscript II retrotranscriptase, and cyanine 3 and Alexa 647 fluorescent dyes (Genisphere), following the procedures at http://web.uvic.ca/cbr/grasp/ (Genisphere Array 50 Protocol).

The transcription profiles were measured using cDNA microarrays produced by the consortium for Genomic Research on All Salmon Project (cGRASP, http://web.uvic.ca/grasp/, Rise et al. 2004). The chip contains 16,006 salmonid cDNAs of Atlantic salmon and rainbow trout (von Schalburg et al. 2005) and was successfully tested and applied to other salmonids species, including brook charr (Rise et al. 2004; von Schalburg et al. 2005; Koop et al. 2008; Mavarez et al. 2009; Sauvage et al. 2010). Sequence differences between samples are not likely to result in spurious differential expression since the divergence between the studied populations are similar among themselves relative to Atlantic salmon and much less important than is found among salmonid species, which showed similar hybridization performances for this chip (von Schalburg et al. 2005; Koop et al. 2008).

A total of 48 microarrays were analyzed. Each individual was technically replicated on two bicolored distinct microarrays and dye swapped. A loop design (Figure 1) was used to include pairwise direct comparisons between a hybrid and its parental population and indirect comparisons between pure crosses. This design allowed the comparison of all the groups with a similar statistical power while minimizing the use of individuals and microarrays.

Figure 1.—

Figure 1.—

Open in a new tab

Loop experimental design including direct comparisons between hybrids and their parental populations and indirect comparisons between parental populations. Each sample was hybridized twice using dye swap. L, Laval; R, Rupert; D, domestic; ind., individuals.

Data acquisition, preparation, statistical analysis, and functional classification of genes:

Microarrays were scanned using a ScanArray scanner (Packard BioScience). Spots were localized and quantified with the QuantArray 3.0 software, using the histogram quantification method. Local background and the data from bad spots were removed. Spots with signal intensities smaller than the mean intensity of empty spots plus twice their standard deviations were removed from the analysis, leaving a total of 3263 analyzed spots. After a log2 transformation, the data were normalized with the lowess method (regional lowess procedure) implemented in the R/MAANOVA package (Kerr et al. 2000) to remove signal intensity-dependant dye effects on each slide (regional lowess procedure). Normalized data for each slide/dye combination were median centered to allow comparable expression values within and between slides without global patterns of expression showing any individual or population biases (Swanson-Wagner et al. 2006).

To detect differences between crosses, data were analyzed using a mixed-model ANOVA (Wolfinger et al. 2001) and the R/MAANOVA package (Kerr et al. 2000; Kerr et al. 2002). We tested the presence of cross-type effects with the following ANOVA model,

graphic file with name M1.gif

with A, array; D, dye; C, cross type; G, gene; S, sample (two replicates per samples); and terms in parentheses are interaction terms. This model included “array” and “sample” as random terms and “dye” and “cross type” as fixed terms.

We tested the null hypothesis that the residuals for each gene were normally distributed using the Kolmogorov test from the R Nortest package. We also evaluated departure from homoscedasticity of gene-specific variance using the Fligner–Killeen test, as implemented in R. A permutation-based F-test (Fs, with 1000 sample ID permutations) was then performed, and restricted maximum likelihood was used to solve the mixed-model equations (Cui et al. 2005). A false discovery rate correction (FDR = 0.05) was applied using the R/MAANOVA package, to generate a first list of transcripts with significant differential expression between crosses. Further contrast tests were conducted on this list with 1000 permutations for the 12 possible comparisons among the six crosses. A FDR of 0.05 was also applied for each contrast test.

All differentially expressed transcripts between parental crosses were examined to assess their mode of transcription regulation in the hybrid crosses. Each transcript was classified as additive if the contrast P value did not allow rejecting the null hypothesis of no statistical difference in expression between hybrid and the average value of its parental populations. Transcripts were classified as nonadditive if the contrast P value indicated that the hybrid was expressing the transcript at a significantly different level than the average of the parental populations (FDR = 0.05). Among nonadditive transcripts, dominant transcripts were defined as transcripts for which the contrast P value with one of the parental populations was nonsignificant. Transcripts were defined as maternal dominant if the mother/hybrid crosses contrast was nonsignificant and paternal dominant if the father/hybrid crosses contrast was nonsignificant (FDR < 0.05). Nonadditive transcripts were defined as underdominant (UD) if their expression was significantly lower than both parental populations and as overdominant (OD) if their expression was significantly higher than both parental populations (contrast P value < 0.05). For transcripts that were not differentially expressed between parental populations, all nonadditive transcripts were classified as UD or OD in the same way. The OD and UD transcripts were also defined as transgressive. A chi-square test (exact binomial with 1 d.f.) was applied to determine whether the modes of transcription regulation differed among the crosses, using the R software package.

Functional classification and assessment of significant differential representation of functional classes were performed in the DAVID and PANTHER environments. DAVID 2.1 gene accession conversion tool was first used to convert gene ontology-linked identifications from the salmonid microarrays to UNIGEN clusters. Assessment of significant differential representation of functional categories was performed using PANTHER (significance threshold: P = 0.05). The overrepresentation of a given function corresponds to a significant differentially overrepresented function among the ones represented by all significantly expressed transcripts (3263). Gene ontology annotation was possible for 45% of the significant cDNA clones and 49% of all clones on the cGRASP microarray.

Body size measurement:

Twenty juveniles per family (10 families for RR, RL, LL, LD, and DD crosses, and 9 families for DR cross) were sampled and their fork lengths measured. Normality and homoscedasticity of data were tested respectively with the Kolmogorov–Smirnov and the Brown and Forsythe tests (Quinn and Keough 2005). To detect cross effects, data were analyzed with the following ANOVA mixed model implemented in ASReml version 2 (VSN International, UK),

graphic file with name M2.gif

with C (cross type) as the fixed effect and F (family) nested within C as the random effect followed by a posteriori analysis when significant. If a cross effect was confirmed, the differences between parental populations and their hybrid crosses were tested to detect the presence of heterosis or outbreeding depression with the one-way ANOVA model (Yijk = μ + Ci + ɛijk with C as the fixed effect) using Statistica version 6.0 (StatSoft, Tulsa, OK, USA). The mean differences were tested with Tukey tests or Games and Howell tests for nonhomoscedastic data, using SPSS version 13.0. The intensity of heterosis or outbreeding depression is expressed in percentage as [(f1m-1) − 1] × 100 for heterosis and as [1 − (f1m-1)] × 100 for outbreeding depression (Shikano and Taniguchi 2002), where f1 and m represent the mean hybrid value and the mean pure-population value, respectively. Size data per family are presented in supporting information, File S1.

RESULTS

Overall sample differences:

Results from the Kolmogorov test revealed that 4.68% (153 of 3263) transcripts showed departure of normality at α = 0.05. However, the permutation-based ANOVA makes no assumption about normality, and although normality is assumed in the procedure used for assessing genetic parameters, restricted maximum-likelihood estimators, such as the one we used for the Fs statistic, are robust to deviations from the assumption of normality (Kruuk 2004; Cui et al. 2005). Results from the Fligner–Killeen test revealed that only 4.1% (135 out of 3263) transcripts showed departure from homogeneity of gene-specific variance at α=0.05, which should not significantly affect our main conclusions.

A total of 863 of 3263 (26.4%) analyzed transcripts showed significant differential expression in at least one of the six crosses (P < 0.015, FDR < 0.05). Among the 863 significant transcripts, 129 (15%) were labeled “unknown” because they did not generate any significant BLAST hits (von Schalburg et al. 2005, 2008). A total of 151 functional categories were represented by the 3263 significantly expressed transcripts. However, only those that were identified as significantly overrepresented in parental populations and their hybrid comparisons by the Panther analysis are presented (Figure 2) and interpreted, whereas details about the genes representing other functional groups are presented in supporting tables.

Figure 2.—

Figure 2.—

Open in a new tab

Pie chart representation of overrepresented biological functions as revealed by Panther analysis in the comparisons between the parental populations and hybrid crosses and parental populations. The numbers inside the triangles correspond to the number of significant genes with differential expression in these comparisons, and those near the pie splices correspond to the number of genes in each functional category labeled in the legend. (A) Comparison between the three parental populations. (B) Comparison between DR hybrids and parental populations. (C) Comparison between LD hybrids and parental populations. (D) Comparison between RL hybrids and parental populations.

Comparing the parental populations:

The three parental populations differed substantially in terms of the number of differentially expressed genes as well as transcribed functional categories. Between the R and L populations, 104 transcripts representing 72 different genes (23 unknown) were differentially expressed (Table S1). In contrast, 265 transcripts representing 178 different genes (Table S2) including 43 unknown were differentially expressed between R and D populations, and 276 transcripts representing 191 different genes (Table S3) with 45 unknown between L and D populations. These differences were all significant (chi-square test, exact binomial with 1 d.f: R/L vs. L/D P < 2.2 E-16, R/L vs. R/D P < 2.2 E-16).

Gene ontology analysis identified nine functional categories that were overrepresented among parental populations (Figure 2A). Between the R and D populations, three functional categories (cell adhesion, pyrimidin metabolism, and rRNA metabolism), were overrepresented. Pyrimidin metabolism and rRNA metabolism genes were overtranscribed in the D relative to the R population. Genes falling in the cell adhesion category were either undertranscribed or overtranscribed in the D relative to the R population (Table S2). Between L and D populations, five functional categories (calcium metabolism, cell motility, chromosome segregation, lipid and fatty acid metabolism, and signal transduction) were overrepresented. The genes playing a role in the calcium metabolism and the chromosome segregation categories were undertranscribed in the L relative to the D population. Genes related to the cell motility, lipid and fatty acid metabolism, and signal transduction categories were either undertranscribed or overtranscribed in the L relative to the D population (Table S3). Finally, two functional categories (carbohydrate metabolism and pyrimidin metabolism) were overrepresented among the genes differently expressed between the L and R populations. Pyrimidin metabolism genes were undertranscribed in the R relative to the L populations. Inversely, genes related to the carbohydrate metabolism category were overtranscribed in the R relative to the L population (Table S1). Only one category, pyrimidin metabolism, was overrepresented in two parental population comparisons. Gene ontology annotation was possible for only 28 of 72 significant genes between R and L, 62 of the 178 significant genes between D and R, and 83 out of 191 genes between L and D populations, respectively.

Comparing hybrid and parental crosses:

The differences observed between the three hybrid crosses and their parental populations varied considerably depending on which of the three hybrids was compared to its parental populations. For the DR hybrids, the number of transcripts differentially expressed between the hybrids and their parental populations was smaller than between the latter ones; 130 (89 genes) and 155 (113 genes) transcripts were differentially expressed between DR vs. D and DR vs. R, respectively, compared to 265 differentially expressed transcripts (178 genes) between the parental populations. In contrast, the LD hybrids showed twice as many differently expressed transcripts when compared to their D parental population than when compared to L; 177 transcripts (125 genes) were differentially expressed between LD vs. L, compared to 455 (302 genes) between LD and D. The RL hybrids presented a third distinct pattern of transcription profile, whereby the number of significant differences in transcript expression between hybrids and either parents was higher than in the parental populations comparison. Thus, 350 and 294 transcripts were differentially expressed between RL vs. L and RL vs. R, representing 231 and 195 different genes, respectively, compared to 104 differentially expressed transcripts (72 genes) between parental populations.

Gene ontology analysis identified 16 different functional categories that were overrepresented in comparisons between hybrid and parental populations (Figure 2, B, C, and D). Between DR and D, one functional category, calcium-mediated signaling, was overrepresented whereas between DR and R, only the ion transport category was overrepresented (Figure 2B). Moreover, no category was overrepresented in both of the hybrids vs. parental and the parental population comparisons. Gene ontology annotation was possible for 44 out of 89 genes for DR vs. D and 46 out of 113 genes for DR vs. R. Two functional categories, nuclear transport and sulfur metabolism, were overrepresented when LD was compared to L and three (calcium-mediated metabolism, lipid and fatty acid metabolism, and pre-mRNA processing) when LD was compared to D (Figure 2C). Two functional categories were overrepresented in both hybrids vs. parental and parental population comparisons: calcium mediated metabolism and lipid and fatty acid metabolism. Gene ontology annotation was possible for 64 out of 125 genes for LD vs. L and 141 out of 302 genes for LD vs. D. Ten functional categories (calcium-mediated signaling, cell proliferation and differentiation, chromosome segregation, intracellular signaling cascade, ligand-mediated signaling, mRNA end-processing and stability, neurotransmitter release, pyrimidin metabolism, signal transduction, and intracellular transport) were overrepresented between RL and L crosses (Figure 2D). Seven functional categories (calcium-mediated signaling, cell motility, cell proliferation and differentiation, chromosome segregation, neurotransmitter release, signal transduction, and intracellular transport) were overrepresented when RL hybrid was compared to the R population (Figure 2D). Only one category, pyrimidin metabolism, was overrepresented in both hybrid vs. parental and parental population comparisons. Gene ontology annotation was possible for 108 of 231 genes for RL vs. L and 91 of 195 genes for RL vs. R.

Additivity vs. nonadditivity:

Differentially expressed transcripts between parental populations:

The majority of transcripts differently expressed between the parental populations exhibited an additive mode of transcription regulation but the proportion of additivity differed among crosses: 94% of transcripts in DR, compared to 54.3% in LD and 62.5% in RL hybrids (Table 1). The proportion of additivity in the DR cross was statistically different from RL and LD crosses (chi-square test, exact binomial with 1 d.f.: DR vs. LD P < 2.2 E-16, DR vs. RL P < 2.2 E-16). Maternal influence was also important in LD hybrids since 88% of the transcripts with a dominant mode of transcription regulation presented maternal dominance, compared with 55 and 25% for RL and DR hybrids, respectively. Very few transcripts were over- or underdominant (Table 1).

TABLE 1.

Additivity, dominant (maternal and paternal), and over-/under dominantly (OD/UD) transcript expression among differentially expressed transcripts between parental populations and OD/UD transcript expression among nondifferentially expressed transcripts between parental populations in hybrid crosses (FDR = 0.05)

Differentially expressed transcripts between parental populations
Dominant
 Nondifferentially expressed transcripts between parental populations
Hybrid strains Additive Maternal Paternal OD UD OD UD
DR 249 4 12 0 0 10 7
LD 150 110 14 2 0 59 55
RL 65 21 17 0 1 90 56
Open in a new tab

DR, domestic × Rupert hybrids; LD, Laval × domestic hybrids; RL, Rupert × Laval hybrids.

Nondifferentially expressed transcripts between parental populations:

All transcripts that were differentially expressed in hybrids, but did not show significant differential expression between parental populations, were included in this analysis. The number of transcripts showing either OD or UD modes of transcription was variable among crosses with the highest number for RL hybrids (Table 1). The numbers of OD vs. UD transcripts were only significantly different in the RL hybrid (P < 0.05, chi-square test, exact binomial with 1 d.f.). The fold change of expression for the transcripts showing OD in LD hybrids was significantly different from DR and RL hybrids crosses (LD: 1.68, DR: 1.19, RL: 1.31, t-test, P < 0.05), whereas the fold change of the transcripts showing UD was not significantly different between the three hybrids crosses (LD: 0.81, DR: 0.81, RL: 0.79). The list of misexpressed transcripts and their functional categories are available in supporting Table S4.

Cross-specific modes of transcription regulation:

A total of 116 transcripts (62 genes) involved in various biological functions showed varying modes of transcription regulation (additivity vs. nonadditivity) among the different hybrid crosses. For example, the Heat shock protein HSP 90-beta gene had an additive mode of transcription regulation in DR whereas it presented a nonadditive mode of transcription regulation in the LD hybrid. The Apolipoprotein B-100 precursor gene had a nonadditive mode of transcription regulation in DR and RL hybrids but an additive one in the LD hybrid. The CD63 antigen gene also showed different modes of transcription regulation: additive in DR and nonadditive in RL hybrids. Only the genes for which gene ontology annotation could be found are presented in Table 2, while the rest of the list is presented in Table S5. When genes were represented by several clones, all the repeated clones within a given hybrid cross displayed the same inheritance pattern, except for the three following genes: H-2 class II histocompatibility antigen gamma chain, ADP/ATP translocase 2, and Tubulin alpha chain (Table S5).

TABLE 2.

Genes exhibiting an additive mode of transcription regulation in one hybrid cross and a nonadditive mode in one or two of the other hybrid crosses

Gene names No. clones DR LD RL Functional categories
Peroxiredoxin-5, mitochondrial precursor 1 A N-A * Antioxydation
Tubulin α-1A chain 1 * A N-A* C.M./I.P.T./chromosome segregation
Tubulin α-1C chain 1 A A N-A* C.M./I.P.T./chromosome segregation
UV excision repair protein RAD23 homolog B 1 * N-A A DNA repair
Heat shock protein HSP 90-β 1 A N-A * Immunity and defense
Peroxisomal multifunctional enzyme type 2 1 A N-A * L.F.A metabolism
H-2 class II histocompatibility antigen γ chain 2 N-A A A MHCII-mediated immunity
H-2 class II histocompatibility antigen γ chain 1 A N-A * MHCII-mediated immunity
NADH dehydrogenase [ubiquinone] 1 α subcomplex subunit 8 1 A N-A * Oxydative phosphorilation
Heterogeneous nuclear ribonucleoprotein A/B 1 * A N-A Pre-mRNA processing
Heterogeneous nuclear ribonucleoprotein A0 1 A N-A * Pre-mRNA processing
26S proteasome non-ATPase regulatory subunit 14 1 * A N-A Protein metabolism and modification
40S ribosomal protein S2 1 A N-A * Protein metabolism and modification
Elongation factor 1-δ 1 A N-A N-A* Protein metabolism and modification
Nucleoside diphosphate kinase A 5 A * N-A Pyrimidin metabolism
Nucleoside diphosphate kinase B 4 A * N-A Pyrimidin metabolism
Cold-inducible RNA-binding protein 2 A N-A * RNA binding
CD63 antigen 1 A * N-A Signal transduction
Acyl-CoA-binding protein 1 A * N-A Intracellular transport
ADP/ATP translocase 2 2 A N-A * Intracellular transport
ADP/ATP translocase 2 1 A N-A N-A* Intracellular transport
Myelin expression factor 2 1 A * N-A Intracellular transport
Oncorhynchus mykiss invariant chain S25-7 mRNA 1 N-A A * Intracellular transport
Vacuolar ATP synthase 16-kDa proteolipid subunit 2 A N-A * Intracellular transport
Apolipoprotein B-100 precursor 2 N-A A N-A* Intracellular transport/L.F.A. metabolism
Open in a new tab

Only genes for which gene ontology annotation could be found are presented here. Genes without gene ontology annotation can be found in Table S5.

No. clones, number of clones; DR, Domestic × Rupert hybrids; LD, Laval × Domestic hybrids; RL, Rupert × Laval hybrids; N-A, nonadditive mode; A, additive mode; *, transcripts did not show significant difference between their parental strains. Abbreviation for the functional categories: C.M., cell motility, I.P.T., intracellular protein traffic, L.F.A., Lipid and fatty acid. The functional category is defined for unique gene in Panther online classification system (http://www.pantherdb.org).

Body size differences among crosses:

Fork lengths at the yolk-sac resorption stage were statistically different between the three parental populations (Tukey tests, all P < 0.05). The D population was the smallest with 22.3 ± 0.9 mm of length. The L population was the longest (25.7 ± 1.0 mm) whereas the length of the R population was of 23.9 ± 1.4 mm. The DR hybrids, with a fork length of 23.1 ± 1.2 mm, displayed intermediate size (additivity) compared to the average of the juvenile from their parental populations (Figure 3A). Heterosis was observed in the LD hybrids, where hybrids were 19.6% longer (28.7 ± 1.7 mm) than the average of the juveniles from their parental populations (Tukey tests, P < 0.05) (Figure 3B). In contrast, RL hybrids showed outbreeding depression, whereby the hybrids were 9.2% shorter (22.5 ± 1.4 mm) than the average of the juveniles from their parental populations (Tukey tests, P < 0.05) (Figure 3C).

Figure 3.—

Figure 3.—

Open in a new tab

Comparison of fork length at yolk-sac resorption stage between hybrid crosses and parental populations. Means ±SD. Different letters indicate significantly different means (P < 0.05). (A) DR hybrid compared to D and R parental populations. (B) LD hybrid compared to L and D parental populations. (C) RL hybrid compared to R and L parental populations.

DISCUSSION

Differentiation among the parental populations:

On the basis of microsatellite markers, the three parental populations were previously found to be highly genetically distinct from one another with the domestic population being genetically intermediate to the L and R populations (Martin et al. 1997). However, genetic information obtained from microsatellites is more likely to reflect the outcome of differentiation resulting from neutral processes. Contrary to Martin et al. (1997), our results reveal that, from a transcriptome standpoint, the D population is the most differentiated, with the R and L populations being more similar to each other. Previous studies have suggested that much of the variation in transcription profiles among populations or species resulted from neutral divergence, while others considered that this is more likely the result of selection (reviewed by Rodríguez-Trelles et al. 2005). Whitehead and Crawford (2006) also showed that only variation exceeding the phylogenetic variance may be considered as resulting from selection. Our results, combined with those of Martin et al. (1997), show that transcription profiles and neutral patterns of divergence among the three parental populations of brook charr are distinct. This has also been reported in other studies on fish, including salmonids (Oleksiak et al. 2002; Giger et al. 2006). Although further investigation will be necessary to test this hypothesis, the more pronounced transcriptional distinctness observed in the D population could be a direct consequence of artificial selection or of inbreeding depression caused by the domestication process. Although we cannot strictly rule it out, the hypothesis of inbreeding depression may be less compelling than the artificial selection hypothesis, since the inbreeding coefficient has been found to be smaller for the D (F = 0.18) relative to the Rupert and Laval populations (mean F = 0.35) (Martin et al. 1997).

Previous studies in Atlantic salmon and brook charr demonstrated that only four to seven generations of domestication could lead to significant changes in transcription profiles (Roberge et al. 2006; Sauvage et al. 2010). Since the domestic (D) population used in this study has undergone at least 15 generations of domestication, it is plausible that some of the differentiation has been generated by directional selection for traits of commercial interest (i.e., growth, disease resistance, or swimming resistance). For instance, the growth factor gene (CTGF) and genes involved in lipid metabolism were significantly overtranscribed in the D population. Namely, the D population showed 34 and 30% overtranscription of the CTGF gene, 27 and 19% overtranscription of the apolipoprotein B-100 precursor gene, and 55 and 67% overtranscription of the apolipoprotein A-I precursor gene against L and R populations, respectively (Table S2 and Table S3). Similar results were obtained by Roberge et al. (2006), who observed 23% of overtranscription for GH gene and 89% of overtranscription for apolipoprotein A-I-1 precursor gene in farmed salmon. Interestingly, when comparing their results on domestication effects in brook charr to those on Atlantic salmon, Sauvage et al. (2010) also observed that genes with similar biological functions were found to be under selection in both studies. Finally, the D population is known for its disease susceptibility, for example to furuncunlosis (Cipriano et al. 2002). This susceptibility could be partially linked to the fact that the MHC antigen coding genes were undertranscribed in the domestic relative to the other two populations (Table S2 and Table S3).

The transcriptional differences observed between the R and L populations may partially reflect adaptive responses to their distinct environment (local adaptation). For example, R and L populations exhibited significant bioenergetic-related physiological differences at young ages, which could be the result of environmental constraints. Thus, some carbohydrate metabolism genes are undertranscribed, whereas genes related to lipid metabolism (apolipoprotein A, fatty acid-binding protein) are overtranscribed in the L population. This could reflect a reduction in basal metabolism and higher growth-related resource allocation, which, in turn, would result in higher growth performance, as observed in the L population, which is bigger at this life stage than the R population. Modifications of lipid metabolism have also been suggested to be linked to migration-related osmoregulatory changes associated with transition from fresh water to salt water, which anadromous salmonids undergo (Sheridan et al. 1985; Li and Yamada 1992). Thus, the differential expression of lipid metabolism-related genes between R and L populations could also be a consequence of differential adaptation to a purely freshwater lifecycle in the former vs. anadromous life cycle in the latter. This, however, remains to be rigorously investigated. In summary, the Martin et al. (1997) study on neutral markers and our gene expression data in controlled conditions showed that the three pure populations used in this study are genetically very distinct. They also suggest that both artificial selection and adaptation by natural selection to their respective environments may be responsible for this differentiation.

Hybrid response:

Despite the important role potentially played by hybridization in speciation processes, adaptative evolution, and agronomical innovation, little is known about transcription regulation inheritance in hybrids, particularly in fishes. Our results showed that the differences in numbers of significant genes, as well as affected biological functions, varied considerably depending on which of the three hybrids was compared to its parental populations. In addition, although individuals from the three hybrid crosses expressed a majority of transcripts in an additive manner, the proportions of additivity in transcription regulation inheritance varied among the hybrid crosses. Previous studies using single species reported contrasting results about transcription regulation inheritance. These reported a predominance of either additivity (fruit flies, Hughes et al. 2006; mice, Cui et al. 2006 and Rottscheidt and Harr 2007; maize, Stupar and Springer 2006; Swanson-Wagner et al. 2006; Stupar et al. 2008) or nonadditivity (fruit flies, Gibson et al. 2004; oyster, Hedgecock et al. 2007; maize, Auger et al. 2005; rice, Ge et al. 2008; Salmonid species, Roberge et al. 2008; Normandeau et al. 2009; Renaut et al. 2009) as the major mode of transcription regulation inheritance. As also noted by Rottscheidt and Harr (2007), the variability in the observed major mode of transcription regulation inheritance seemed dependent on different factors such as the methods, tissues, and types of hybrids (backcross or F1) used. Hybrids from divergent species exhibited numerous misexpressed transcripts resulting from the accumulation of regulatory incompatibilities (Landry et al. 2007). However, Rottscheidt and Harr (2007) showed that neither the level of inbreeding nor time since divergence could be associated in previous studies with the frequency of nonadditivity. The only exception to this was found in patterns of gene expression in testis, which were interpreted as a consequence of the role of sexual organ differentiation in reproductive isolation. In contrast, within inbred lines of maize, the hybrids from less distant parental lines exhibited the greatest proportion of nonadditivity of transcripts (Stupar et al. 2008). Moreover, frequency of additivity varied highly depending on the tissue studied. Thus, genetic distance has not yet firmly been linked with the frequency of additivity at the intraspecific level of comparison. Here, our results regarding the extent of nonadditive transcription regulation inheritance observed in hybrids are suggestive of a positive correlation with genetic distance between the parental populations. Thus, hybrids produced by crosses between the two most distant populations (R and L) showed more nonadditive transcripts than any other hybrid crosses. Predominance of nonadditivity was also observed in hybrids resulting from crosses between two divergent populations of lake whitefish (Coregonus clupeaformis), another salmonid species (Renaut et al. 2009). In addition, we found that transcription regulation inheritance (additive or nonadditive modes of transmission regulation) was often not correlated for a given gene among the three hybrid crosses (Table 2). Transcription regulation inheritance is affected by many factors, such as DNA binding sites, transcription factor abundance, and the affinity of the latter for the binding sites. However, epigenetic changes could be responsible for transcriptional changes among hybrids (reviewed by Michalak 2009). These parameters and their complex interactions result in the diversity of transcript expression patterns that in turn contribute to the biological diversity observed at the population level (Brown 2006; Maston et al. 2006; Gerke et al. 2009). Moreover, because of interactions in the regulatory network, hybrids with different phenotypes than those of their parents can be obtained from crossing parental populations with equivalent phenotypes (Brem and Kruglyak 2005; Landry et al. 2007). Here, the hybridization process resulted in complex patterns of transcript regulation and affected biological functions among crosses. Given the differences detected between parental populations, such variability of the transcriptome among the three hybrid crosses could depend on the unique genetic architectures of each of the parental populations.

Hypothetical links between patterns of gene expression and hybrid phenotypes:

In this section, we discuss how size phenotypes in the hybrid crosses could be indirectly linked to the differences in inheritance patterns of transcripts that we observed. Thus, as observed at the transcriptome level, we observed three different inheritance patterns of size at age in hybrids relative to their parental populations (Figure 3). Namely, the DR hybrids, which showed an additive size phenotype response, also exhibited over 94% of additively expressed transcripts. This was in sharp contrast with LD and RL hybrids, which respectively showed heterosis and outbreeding depression for size, as well as a higher proportion of nonadditivity (dominant and over-/underdominance) in transcription regulation. What could the molecular causes underlying the expression of heterosis vs. outbreeding depression at the phenotypic level be? Among the most common hypotheses regarding the mechanisms underlying heterosis are: (i) dominance, which explains heterosis by masking the effect of deleterious alleles by superior alleles from the parents; (ii) overdominance, for which allelic interactions at heterozygous locus result in positive effect superior to homozygote; (iii) pseudo-overdominance, which is a positive dominance complementation of linked alleles (Lippman and Zamir 2007). Moreover, dominant patterns of expression regulation have been proposed to result from allelic dosage effects (Auger et al. 2005), monoallelic expression (Birchler et al. 2003), or epistatic gene interactions (Hedgecock et al. 2007). The main differences observed between LD and RL hybrids lie in the proportions of maternal/paternal dominant transcripts and over-/underdominance (Table 1). Given that LD hybrids present a higher number of dominant transcripts relative to OD and UD compared to RL hybrids, the dominance hypothesis could explain the predominance of size heterosis observed in LD hybrids. In particular, there was a clear prevalence of maternal population dominance in the LD hybrids. Thus, gene dominance from the biggest parental population may explain the performance of these hybrids. At early life stages, the maternal genome may influence size through gene products contained in the egg (Hebert et al. 1998; Nakajima and Taniguchi 2002). Indeed, previous studies on the L population have revealed that maternal effects are acting until yolk sac resorption and are correlated with the size of the mother (Perry et al. 2005). Heterosis may be further enhanced by the presence of numerous overdominant transcripts in hybrids (Table S4). Therefore, at this stage of development, heterosis could result from advantageous dominance and overdominance of transcripts inheritance, and maternal effects could further contribute to the heterosis observed in the LD individuals.

On the other hand, outbreeding depression can also result in the loss of favorable genetic interactions or/and the disruption of epistatic interaction between coevolved genes (Mcclelland and Naish 2007). Transgressive phenotypes in hybrids may thus be the consequence of a deeper genetic distance between parental populations (Stelkens and Seehausen 2009). For instance, in rainbow trout (Oncorhynchus mykiss), outbreeding depression did not appear in introgressed genomes, possibly because coadapted genotypes were too similar for recombination to have a negative impact (Tymchuk et al. 2007). Moreover, divergent genomes may accumulate regulatory incompatibilities resulting in occurrence of misexpressed transcripts (OD or UD) in hybrids (Landry et al. 2007). Here, the deeper genetic distance between the R and L populations could explain the higher prevalence of transgressive transcripts involved in many biological functions in their hybrids, as compared to those involving the D population. It is also noteworthy that outbreeding depression for length was observed in the RL hybrids, which also showed the highest number of transgressive transcripts (either UD or OD) (Table 1).

Our results thus support the hypothesis that transgressivity of phenotypes may depend both on genetic distance between the parental populations and on interactions between dominant vs. transgressive transcription regulation mechanisms arising from divergent genetic architectures. Therefore, prevalence of dominance combined with overdominance in transcription regulation seems related with heterosis, while prevalence of transgressive transcription regulation seems to be more related with outbreeding depression. To conclude, while our results suggest that size and transcriptional regulation phenotypes are at least partially associated, they also illustrate the complexity of predicting hybrid phenotypes on the sole basis of previous knowledge of parental phenotypes or their genetic divergence.

Acknowledgments

We thank the staff of the LARSA (Laboratoire de Recherche des Sciences Aquatiques), and of the ISMER (Institut des sciences de la mer de Rimouski) station, as well as the personnel from Bernatchez' and Audet's Laboratories for their assistance in rearing and sampling the fish used in this study. We also thank N. N. Brodeur, A. Nolte, E. Normandeau, N. Maillet, J. Mavarez, C. Roberge, and A. R. Whiteley for rich discussions and assistance with the analyses. We are grateful to Lauren McIntyre and two anonymous referees for their constructive comments on an earlier version of the manuscript. This study was supported by a research grant form Science and Engineering Research Canada (NSERC strategic program) to L. Bernatchez, C. Audet, and D. Cyr and by financial contribution from the Société de recherche et de développement en aquaculture continentale (SORDAC), Inc. This work is a contribution to the research programme of the RAQ (Réseau Aquaculture Québec).

Supporting information is available online at http://www.genetics.org/cgi/content/full/genetics.110.118158/DC1.

Expression data from this article have been deposited with the NCBI/Gene Expression Omnibus under accession no. GSE22008.

References

  1. Auger, D. L., A. D. Gray, T. S. Ream, A. Kato, E. H. Coe et al., 2005. Nonadditive gene expression in diploid and triploid hybrids of maize. Genetics 169 389–397. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Baack, E. J., and L. H. Rieseberg, 2007. A genomic view of introgression and hybrid speciation. Curr. Opin. Genet. Dev. 17(6): 513–518. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Barton, N. H., 2001. The role of hybridization in evolution. Mol. Ecol. 10 551–568. [DOI] [PubMed] [Google Scholar]
  4. Birchler, J. A., D. L. Auger and N. C. Riddle, 2003. In search of the molecular basis of heterosis. Plant Cell 15 2236–2239. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Brem, R. B., and L. Kruglyak, 2005. The landscape of genetic complexity across 5,700 gene expression traits in yeast. Proc. Natl. Acad. Sci. USA 102 1572–1577. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Brown, T. A., 2006. Genomes 3. Garland Science, New York.
  7. Burke, J. M., and M. L. Arnold, 2001. Genetics and the fitness of hybrids. Annu. Rev.Genet. 35 31–52. [DOI] [PubMed] [Google Scholar]
  8. Castric, V. and L. Bernatchez, 2003. The rise and fall of isolation by distance in the anadromous brook charr (Salvelinus fontinalis Mitchill). Genetics 163 983–996. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Cipriano, R. C., D. Marchant, T. E. Jones and J. H. Schachte, 2002. Practical application of disease resistance: a brook trout fishery selected for resistance to furunculosis. Aquaculture 206 1–17. [Google Scholar]
  10. Cui, X. G., J. T. G. Hwang, J. Qiu, N. J. Blades and G. A. Churchill, 2005. Improved statistical tests for differential gene expression by shrinking variance components estimates. Biostatistics 6 59–75. [DOI] [PubMed] [Google Scholar]
  11. Cui, X. Q., J. Affourtit, K. R. Shockley, Y. Woo and G. A. Churchill, 2006. Inheritance patterns of transcript levels in F-1 hybrid mice. Genetics 174 627–637. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Fraser, D. J., P. Duchesne and L. Bernatchez, 2005. Migratory charr schools exhibit population and kin associations beyond juvenile stages. Mol. Ecol. 14 3133–3146. [DOI] [PubMed] [Google Scholar]
  13. Ge, X., W. Chen, S. Song, W. Wang, S. Hu et al., 2008. Transcriptomic profiling of mature embryo from an elite super-hybrid rice LYP9 and its parental lines. BMC Plant Biol. 8 114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Gerke, J., K. Lorenz and B. Cohen, 2009. Genetic interactions between transcription factors cause natural variation in yeast. Science 323 498–501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gibson, G., R. Riley-Berger, L. Harshman, A. Kopp, S. Vacha et al., 2004. Extensive sex-specific nonadditivity of gene expression in Drosophila melanogaster. Genetics 167 1791–1799. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Giger, T., L. Excoffier, P. J. R. Day, A. Champigneulle, M. M. Hansen et al., 2006. Life history shapes gene expression in salmonids. Curr. Biol. 16 R281–R282. [DOI] [PubMed] [Google Scholar]
  17. Hansen, T. F., 2006. The evolution of genetic architecture. Annu. Rev. Ecol. Evol. System. 37 123–157. [Google Scholar]
  18. Hebert, K. P., P. L. Goddard, W. W. Smoker and A. J. Gharrett, 1998. Quantitative genetic variation and genotype by environment interaction of embryo development rate in pink salmon (Oncorhynchus gorbuscha). Can. J. Fisheries Aquatic Sci. 55 2048–2057. [Google Scholar]
  19. Hedgecock, D., J.-Z. Lin, S. DeCola, C. D. Haudenschild, E. Meyer et al., 2007. Transcriptomic analysis of growth heterosis in larval Pacific oysters (Crassostrea gigas). Proc. Natl. Acad. Sci. USA 104 2313–2318. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hughes, K. A., J. F. Ayroles, M. M. Reedy, J. M. Drnevich, K. C. Rowe et al., 2006. Segregating variation in the transcriptome: cis regulation and additivity of effects. Genetics 173 1347–1355. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Kerr, M. K., M. Martin and G. A. Churchill, 2000. Analysis of variance for gene expression microarray data. J. Comput. Biol. 7 819–837. [DOI] [PubMed] [Google Scholar]
  22. Kerr, M. K., C. A. Afshari, L. Bennett, P. Bushel, J. Martinez et al., 2002. Statistical analysis of a gene expression microarray experiment with replication. Statist. Sinica 12 203–217. [Google Scholar]
  23. Kruuk, L. E. B., 2004. Estimating genetic parameters in natural populations using the ‘animal model’. Philos. Trans. R. Soc. Lond. B Biol. Sci. 359 873–890. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Koop, B. F, K. R. von Schalburg, J. Leong, N. Walker, R. Lieph et al., 2008. A salmonid EST genomic study: genes, duplications, phylogeny and microarrays. BMC Genomics 9 545. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Landry, C. R., D. L. Hartl and J. M. Ranz, 2007. Genome clashes in hybrids: insights from gene expression. Heredity 99 483–493. [DOI] [PubMed] [Google Scholar]
  26. Lavagnino, N. J., R. R. H. Anholt and J. J. Fanara, 2008. Variation in genetic architecture of olfactory behaviour among wild-derived populations of Drosophila melanogaster. J. Evol. Biol. 21 988–996. [DOI] [PubMed] [Google Scholar]
  27. Li, H. O., and J. Yamada, 1992. Changes of the fatty acid composition in smolts of masu salmon (Oncorhynchus masou) associated with desmoltification and sea-water transfer. Comp. Biochem. Physiol. 103 221–226. [Google Scholar]
  28. Lippman, Z. B., and D. Zamir, 2007. Heterosis: revisiting the magic. Trends Genet. 23 60–66. [DOI] [PubMed] [Google Scholar]
  29. Lynch, M., 2007. The Origins of Genome Architecture. Sinauer Associates, Sunderland, MA.
  30. Martin, S., J.-Y. Savaria, C. Audet and L. Bernatchez, 1997. Microsatellites reveal no evidence for inbreeding effects but low inter-stock genetic diversity among brook charr stocks used for production in Québec. Bull. Aquaculture Assoc. Can., 21–23.
  31. Maston, G. A., S. K. Evans and M. R. Green, 2006. Transcriptional regulatory elements in the human genome. Annu. Rev. Genomics Hum. Genet. 7 29–59. [DOI] [PubMed] [Google Scholar]
  32. Mavarez, J., C. Audet and L. Bernatchez, 2009. Major disruption of gene expression in hybrids between young sympatric anadromous and resident populations of brook charr (Salvelinus fontinalis Mitchill). J. Evol. Biol. 8 1708–1720. [DOI] [PubMed] [Google Scholar]
  33. McClelland, E. K. and K. A. Naish, 2007. What is the fitness outcome of crossing unrelated fish populations? A meta-analysis and an evaluation of future research directions. Conserv. Genet. 8 397–416. [Google Scholar]
  34. Michalak, P., 2009. Epigenetic, transposon and small RNA determinants of hybrid dysfunctions. Heredity 102 45–50. [DOI] [PubMed] [Google Scholar]
  35. Michalak, P., and M. A. F. Noor, 2003. Genome-wide patterns of expression in drosophila pure species and hybrid males. Mol. Biol. Evol. 20 1070–1076. [DOI] [PubMed] [Google Scholar]
  36. Nakajima, M., and N. Taniguchi, 2002. Genetic control of growth in the guppy (Poecilia reticulata). Aquaculture 202 393–405. [Google Scholar]
  37. Nolte, A., J. Freyhof and D. Tautz, 2006. When invaders meet locally adapted types: rapid moulding of hybrid zones between sculpins (Cottus, Pisces) in the Rhine system. Mol. Ecol. 15 1983–1993. [DOI] [PubMed] [Google Scholar]
  38. Normandeau, E., J. A. Hutchings, D. J. Fraser and L. Bernatchez, 2009. Population-specific gene expression responses to hybridization between farm and wild Atlantic salmon. Evol. Appl. 2 489–503. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Oleksiak, M. F., G. A. Churchill and D. L. Crawford, 2002. Variation in gene expression within and among natural populations. Nat. Genet. 32 261–266. [DOI] [PubMed] [Google Scholar]
  40. Perry, G. M. L., C. Audet and L. Bernatchez, 2005. Maternal genetic effects on adaptive divergence between anadromous and resident brook charr during early life history. J. Evol. Biol. 18 1348–1361. [DOI] [PubMed] [Google Scholar]
  41. Quinn, G. P., and M. J. Keough 2005. Experimental design and data analysis for biologists. Fourth edition. Cambridge University Press, Cambridge, UK.
  42. Ranz, J. M., K. Namgyal, G. Gibson and D. L. Hartl, 2004. Anomalies in the expression profile of interspecific hybrids of Drosophila melanogaster and Drosophila simulans. Genome Res. 14 373–379. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Reiland, J., and M. A. Noor, 2002. Little qualitative RNA misexpression in sterile male F1 hybrids of Drosophila pseudoobscura and D. persimilis. BMC Evol. Biol. 2 16. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Renaut, S., A. W. Nolte and L. Bernatchez, 2009. Transcriptomic investigation of post-zygotic isolation in lake whitefish (Coregonus clupeaformis). Mol. Biol. Evol. 26 925–936. [DOI] [PubMed] [Google Scholar]
  45. Rieseberg, L. H., M. A. Archer and R. K. Wayne, 1999. Transgressive segregation, adaptation and speciation. Heredity 83 363–372. [DOI] [PubMed] [Google Scholar]
  46. Rieseberg, L. H., A. Widmer, A. Michele Arntz and B. Burke, 2003. The genetic architecture necessary for transgressive segregation is common in both natural and domesticated populations. Philos. Trans. R. Soc. Lond. B Biol. Sci. 358 1141–1147. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Rise, M. L., K. R. von Schalburg, G. D. Brown, M. A. Mawer, R. H. Devlin et al., 2004. Development and application of a salmonid EST database and cDNA microarray: data mining and interspecific hybridization characteristics. Genome Res. 14 478–490. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Roberge, C., S. Einum, H. Guderley and L. Bernatchez, 2006. Rapid parallel evolutionary changes of gene transcription profiles in farmed Atlantic salmon. Mol. Ecol. 15 9–20. [DOI] [PubMed] [Google Scholar]
  49. Roberge, C., E. Normandeau, S. Einum, H. Guderley and L. Bernatchez, 2008. Genetic consequences of interbreeding between farmed and wild Atlantic salmon: insights from the transcriptome. Mol. Ecol. 17 314–324. [DOI] [PubMed] [Google Scholar]
  50. Rodríguez-Trelles, F., R. Tarrío and F. J. Ayala, 2005. Is ectopic expression caused by deregulatory mutations or due to gene-regulation leaks with evolutionary potential? BioEssays 27 592–601. [DOI] [PubMed] [Google Scholar]
  51. Rottscheidt, R. and B. Harr, 2007. Extensive additivity of gene expression differentiates subspecies of the house mouse. Genetics 177 1553–1567. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Sauvage, C., N. Derôme, E. Normandeau, J. St-Cyr, C. Audet et al., 2010. Fast transcriptional responses to domestication in the brook charr Salvelinus fontinalis. Genetics 185 105–112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Sheridan, M. A., N. Y. S. Woo and H. A. Bern, 1985. Changes in the rates of glycogenesis, glycogenolysis, lipogenesis, and lipolysis in selected tissues of the coho salmon (Oncorhynchus kisutch) associated with parr-smolt transformation. J. Exp. Zool. 236 35–44. [DOI] [PubMed] [Google Scholar]
  54. Shikano, T. and N. Taniguchi, 2002. Heterosis for neonatal survival in the guppy. J. Fish Biol. 60 715–725. [Google Scholar]
  55. Shriver, M. D., L. Jin, E. Boerwinkle, R. Deka, R. E. Ferrell et al., 1995. A novel measure of genetic distance for highly polymorphic tandem repeat loci. Mol. Biol. Evol. 12 914–920. [DOI] [PubMed] [Google Scholar]
  56. Stelkens, R., and O. Seehausen, 2009. Genetic distance between species predicts novel trait expression in their hybrids. Evolution 63 884–897. [DOI] [PubMed] [Google Scholar]
  57. Stupar, R. M., and N. M. Springer, 2006. Cis-transcriptional variation in maize inbred lines B73 and Mo17 leads to additive expression patterns in the F1 hybrid. Genetics 173 2199–2210. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Stupar, R. M., J. M. Gardiner, A. G. Oldre, W. J. Haun, V. L. Chandler et al., 2008. Gene expression analyses in maize inbreds and hybrids with varying levels of heterosis. BMC Plant Biol. 8 33. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Swanson-Wagner, R. A., Y. Jia, R. DeCook, L. A. Borsuk, D. Nettleton et al., 2006. All possible modes of gene action are observed in a global comparison of gene expression in a maize F-1 hybrid and its inbred parents. Proc. Natl. Acad. Sci. USA 103 6805–6810. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Tallmon, D. A., G. Luikart and R. S. Waples, 2004. The alluring simplicity and complex reality of genetic rescue. Trends Ecol. Evol. 19 489–496. [DOI] [PubMed] [Google Scholar]
  61. Tymchuk, W. E., L. F. Sundström and R. H. Devlin, 2007. Growth and survival trade-offs and outbreeding depression in rainbow trout (Oncorhynchus mykiss). Evolution 61–5 1225–1237. [DOI] [PubMed] [Google Scholar]
  62. von Schalburg, K. R., M. L. Rise, G. A. Cooper, G. D. Brown, A. R. Gibbs et al., 2005. Fish and chips: various methodologies demonstrate utility of a 16,006-gene salmonid microarray. BMC Genomics 6 126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. von Schalburg, K. R., G. A. Cooper, J. Leong, A. Robb, R. Lieph et al., 2008. Expansion of the genomics research on Atlantic salmon (Salmo salar) L. project (GRASP) microarray tools. J. Fish Biol. 72 2051–2070. [Google Scholar]
  64. Whitehead, A., and D. L. Crawford, 2006. Neutral and adaptive variation in gene expression. Proc. Natl. Acad. Sci. USA 103 5425–5430. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Wolfinger, R. D., G. Gibson, E. D. Wolfinger, L. Bennett, H. Hamadeh et al., 2001. Assessing gene significance from cDNA microarray expression data via mixed models. J. Comput. Biol. 8 625–637. [DOI] [PubMed] [Google Scholar]

Articles from Genetics are provided here courtesy of Oxford University Press

RESOURCES