SEARCH

SEARCH BY CITATION

Keywords:

  • Coregonus ;
  • gene transcription;
  • metabolism;
  • oxidative phosphorylation;
  • phenotypic divergence;
  • physiology;
  • speciation

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. Data accessibility
  10. References
  11. Supporting Information

Across multiple lakes in North America, lake whitefish ( Coregonus clupeaformis) have independently evolved ‘dwarf’ and ‘normal’ sympatric species pairs that exhibit pronounced phenotypic and genetic divergence. In particular, traits associated with metabolism have been shown to be highly differentiated between whitefish species. Here, we examine the transcription of genes associated with the five mitochondrial and nuclear genome-encoded oxidative phosphorylation (OXPHOS) complexes, the primary physiological mechanism responsible for the production of ATP, in whitefish species pairs from Cliff Lake and Webster Lake in Maine, USA. We observed OXPHOS gene transcription divergence between dwarf and normal whitefish in each of the two lakes, with the former exhibiting transcription upregulation for genes associated with each of the OXPHOS complexes. We also observed a significant influence of lake on transcription levels for some of the genes, indicating that inter-lake ecological or genetic differences are contributing to variation in OXPHOS gene transcription levels. Together, our results support the hypothesis that metabolic divergence is a critical adaptation involved in whitefish speciation and implicate OXPHOS gene upregulation as a factor involved in meeting the enhanced energetic demands of dwarf whitefish. Further studies are now needed to evaluate the contribution of genetically vs. plasticity driven variation in transcription associated with this critical physiological pathway.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. Data accessibility
  10. References
  11. Supporting Information

Physiological traits represent some of the most promising for studies of adaptation, given that their function is tightly linked to environmental conditions (Pörtner et al., 2010; Seebacher et al., 2010; Storz et al., 2010). Indeed, the oxidative phosphorylation (OXPHOS) pathway is now considered a prime candidate for studies of eco-physiological adaptation (Gershoni et al., 2009; Ballard & Melvin, 2010). The OXPHOS pathway is composed of a series of five protein complexes that are embedded within the inner mitochondrial membrane layer and which ultimately generate the majority of energy (ATP) available for use in cellular processes (Saraste, 1999; Kadenbach, 2003). During OXPHOS, electrons derived from the oxidation of nutrients such as glucose are passed along the first four complexes, which is also known as the electron transport chain: NADH or ubiquinone oxidoreductase (complex I), succinate/ubiquinone reductase (complex II), cytochrome bc1 (complex III) and cytochrome oxidase (complex IV; Saraste, 1999). The passage of electrons between the complexes collectively generates a proton gradient across the inner mitochondrial membrane, and the kinetic energy associated with this proton gradient is harnessed by OXPHOS complex V, ATP synthase, to generate ATP from ADP and inorganic phosphate (Saraste, 1999). Importantly, recent growth in the number of genomic tools available for nonmodel organisms has now made it possible to study molecular and transcription-level variation in the five OXPHOS complexes and the potential role of OXPHOS in eco-physiological interactions.

A number of studies have indicated that both coding and transcription-level variation in OXPHOS genes represent physiological responses to environmental variation. For instance, nonsynonymous nucleotide substitutions have been observed at OXPHOS genes in a variety of taxa, and it has been suggested that these genetic changes translate into adaptive differences in aerobic capacity or thermal tolerance (e.g. Mishmar et al., 2003; Ruiz-Pesini et al., 2004; Dalziel et al., 2006; Scott et al., 2010; Garvin et al., 2011). At the transcription level, rufous-collared sparrows (Zonotrichia capensis) living at high altitudes show OXPHOS gene upregulation compared to populations found at low altitudes, a difference that may facilitate adequate energy production under extreme environmental conditions including hypoxia and cold (Cheviron et al., 2008). Similarly, cytochrome c oxidase gene upregulation is an important mechanism used to cope with the increased energetic demands found in some populations of marine snails (Littorina saxatilis; Martinez-Fernandez et al., 2010). Thus, inter-population variation at genes associated with the OXPHOS complexes may accurately reflect functional genomic responses to divergent environments.

Adaptation to diverging environments may also result in the formation of locally adapted gene complexes and reproductive barriers. It has been suggested that OXPHOS genes may play a particularly important role in the evolution of reproductive barriers among populations, and as a mechanism leading to speciation (Gershoni et al., 2009). Although the OXPHOS complexes are embedded within the inner mitochondrial membrane layer, they are encoded by both the nuclear and mitochondrial genomes, which are expected to be tightly co-adapted (Blier et al., 2001; Gershoni et al., 2009). In the marine copepod Tigriopus californicus, inter-population hybridization leads to a reduction in cytochrome c oxidase enzyme activity in hybrids, suggesting that the disruption of co-adapted nuclear and mitochondrial OXPHOS genes may adversely affect fitness and promote reproductive barriers in this species (e.g. Rawson & Burton, 2002; Ellison & Burton, 2008, 2010). Studies in anthropoids have shown that nuclear and mitochondrial genes associated with OXPHOS complexes III and IV exhibit elevated rates of positive selection compared to other complexes, which may indicate that these regions of the genome are co-evolving (e.g. Rawson & Burton, 2002; Grossman et al., 2004). Furthermore, genes associated with OXPHOS complex V have been implicated in Atlantic eel species (Anguilla spp.) reproductive isolation via cytonuclear incompatibility (Gagnaire et al., 2012). Indeed, it has been hypothesized that mismatched mitochondrial and nuclear components may lead to the enhanced production of reactive oxygen species (ROS), which may affect DNA or functional molecules within cells, or even result in complete mitochondrial dysfunction (Grossman et al., 2004; Bayona-Bafaluy et al., 2005; Reinecke et al., 2009). Taken together, these studies indicate that the co-adaptation of nuclear and mitochondrial OXPHOS components play critical roles in delineating boundaries to gene flow among populations.

Sympatric dwarf and normal species pairs of the lake whitefish (Coregonus clupeaformis) provide an ideal system for investigations of physiological traits, such as OXPHOS, and their roles in ecologically driven divergence. In at least six lakes in the St. John River basin in eastern North America, dwarf and normal whitefish have undergone repeated phenotypic divergence accompanied by variable degrees of reproductive isolation following the recession of the last glaciation event ~15 000 years BP (Pigeon et al., 1997; Bernatchez et al., 2010). Dwarf whitefish have evolved from the ancestral benthic form in response to competition deriving from the secondary contact of two previously allopatric benthic populations (Bernatchez et al., 2010). Dwarfs allocate a greater proportion of their energy budget to metabolism, exhibiting a lower bioenergetic conversion efficiency compared to normal whitefish (growth rate/consumption rate ratio; Trudel et al., 2001). Furthermore, microarray- and qPCR-based studies of gene transcription in whitefish have revealed marked differences between the dwarf and normal whitefish, with dwarfs typically exhibiting upregulation at genes associated with metabolism (Derome et al., 2006; St-Cyr et al., 2008; Jeukens et al., 2009) and oxygen transport (Evans et al., 2012). These differences may be genetically determined as other studies have shown divergence between dwarf and normal whitefish in genes associated with energy production (Renaut et al., 2010, 2011). The allocation of energy towards metabolism by the dwarf whitefish is likely to have facilitated the alternative foraging and predator avoidance tactics needed to colonize the limnetic zone (Rogers et al., 2002; Landry et al., 2007; Landry & Bernatchez, 2010; also see Kahilainen & Lehtonen, 2003). Thus, studies of candidate genes associated with the OXPHOS pathway should help to further elucidate the key traits associated with whitefish expansion into divergent ecological niches and the factors contributing to the maintenance of reproductively isolated populations in sympatry.

Here, we test for OXPHOS complex I–V gene transcription divergence between dwarf and normal lake whitefish from Cliff and Webster lakes in Maine, USA. We predicted that dwarf whitefish would show higher levels of OXPHOS gene transcription compared to the normal whitefish, given the previously observed differences in metabolism between the two species. Whitefish from Webster Lake have not previously been examined for transcription levels at any genes. However, morphological and population genetic differentiation is lower, and levels of gene flow are higher, between dwarf and normal whitefish from Webster Lake compared to Cliff Lake (Campbell & Bernatchez, 2004; Renaut et al., 2011). Thus, we predicted that dwarf and normal whitefish from Cliff Lake would exhibit greater divergence in OXPHOS gene transcription levels compared to Webster Lake. We explored transcription correlations across all of the OXPHOS complexes using multivariate and correlation analyses, as these genes are expected to be tightly co-adapted. In particular, we highlight our results for the correlations between the nuclear and mitochondrial genes within complexes I, and III–V, given the potential role of genomic uncoupling in promoting species boundaries. Gene transcription coupling (correlation) was examined separately in each of the lakes, as the relatively high levels of gene flow in Webster compared to Cliff Lake could contribute to the breakdown gene transcription coordination within OXPHOS complexes (Ellison & Burton, 2008).

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. Data accessibility
  10. References
  11. Supporting Information

Sample collection

Dwarf and normal lake whitefish were sampled from Cliff (46°23′51″ N, 69°15′05″ W) and Webster lakes (46°09′21″ N, 69°04′45″ W) in Maine, USA, in June 2010. Fish were caught live with gillnets, euthanized and immediately dissected to obtain fresh tissue samples and to determine sex. Approximately 30 mg of fresh muscle tissue was collected from each individual. Muscle was collected from the left side of each fish immediately above the lateral line and just anterior to the dorsal fin. Muscle samples were flash frozen in liquid nitrogen and later stored at −80 °C. We analysed gene transcription values in tissue samples from a total of 10 dwarf and 10 normal whitefish from Cliff Lake and 12 dwarf and 11 normal whitefish from Webster Lake.

Whitefish from Cliff and Webster lakes bear mitochondrial DNA descended from Atlantic and Acadian lineages (Pigeon et al., 1997). Previous studies have shown that Cliff Lake dwarfs and normals are, respectively, fixed for the ancestral Acadian and Atlantic mitochondrial lineages (Bernatchez & Dodson, 1990; Pigeon et al., 1997). However, both mitochondrial lineages are present in each of the dwarf and normal whitefish from Webster Lake (Pigeon et al., 1997) due to introgressive hybridization in that lake (Lu et al., 2001). To account for the potential influence of ancestral mitochondrial genomic background on gene transcription, we examined the mitochondrial lineage of six of the dwarf and eight of the normal whitefish from Webster Lake. Mitochondrial lineage was determined by sequencing a portion of the mitochondrial control (D-loop) region from each fish (Lu et al., 2001). Briefly, DNA was isolated from fin clips using a salt extraction technique (Aljanabi & Martinez, 1997), and the primers TPro2 (5′-ACCCTTAACTCCCAAAGC-3′) and HN20 (5′-GTGTTATGCTTTAGTTAAGC-3′) were used to amplify the target gene (Gilbert et al., 1988; Bernatchez & Danzmann, 1993).

RNA extraction and cDNA preparation

Total RNA was isolated from the muscle using the Ambion PureLink RNA Mini Kit according to the manufacturer's instructions (Ambion, Austin, TX, USA). All RNA extractions were treated with Superase-In RNase inhibitor and DNase I following the manufacturer's instructions (Ambion). RNA was quantified with the Nanodrop 2000 Spectrophotometer (Nanodrop Technologies, Wilmington, DE, USA), and RNA integrity was confirmed using the Experion Automated Electrophoresis system (Bio-Rad, Mississauga, ON, Canada). Complementary DNA (cDNA) synthesis was performed using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA). Reverse transcription reactions were performed in 20-μL volumes containing 500 ng RNA and using random primers.

Oxidative phosphorylation gene transcription assays

Annotated whitefish transcriptome sequences (see Renaut et al., 2010) were used as templates in the design of primers and TaqMan minor groove binder probes for use in quantitative PCR (qPCR) evaluation of OXPHOS gene transcription (Table 1, Ambion). Primers and probes were designed using Primer Express v.3.0 software (Applied Biosystems), and primer specificity was tested using conventional PCR and visualization of a single amplification product on agarose gel. In total, a set of 10 OXPHOS gene assays were developed. Specifically, assays were developed to evaluate transcription levels for one nuclear and one mitochondrial-encoded gene associated with complexes I, III and IV, one nuclear and two mitochondrial-encoded genes associated with complex V and one nuclear gene associated with complex II (Gershoni et al., 2009; Table 1). However, gene transcription levels for the complex II assay were too low to be reliably evaluated and thus were excluded from our analyses.

Table 1. TaqMan qPCR assays used to evaluate oxidative phosphorylation (OXPHOS) transcription levels in the lake whitefish (Coregonus clupeaformis). TaqMan assays were designed in Primer Express v. 3.0, and the efficiency of each assay was tested in a validation experiment. Probe, forward (F) and reverse (R) primer sequences are indicated
Target Genea Encoding Genomeb OXPHOS ComplexPrimers (5′–3′)Probe (5′–3′)Efficiency (%)c
  1. a

    The name of the putative target gene is based on sequence alignment with the annotated lake whitefish transcriptome. The gene name abbreviation is shown in brackets.

  2. b

    The coding genome was determined based on the alignment of primer and probe sequences with the Coregonus lavaretus mitochondrial DNA sequence.

  3. c

    TaqMan efficiencies can occasionally exceed 100% when RT reactions are saturated at higher template concentration points in the calibration test.

NADH-ubiquinone oxidoreductase 75-kDa subunit (NADH75-N-I)NuclearIF-CAGGCCAACCAACTCTCACAR-GAGGAGAAACGAGAGGATTTGCTGTGAATCAGACTCTTC100.7
Succinate dehydrogenase (SUCC-N-II)NuclearII

F-GCTACTGGTGGAATGGAGACAAG

R-CCAACGGTAGGCCTGCATAA

ACCTGGGACCCGCTGT98.2
Cytochrome b–c1 complex subunit 8 (CYTB-N-III)NuclearIII

F-GCCAAGATCAGGCATGTGATT

R-GGAAAGCCCTCTGCTCGAA

CCTACAGTCTGTCCCC100.9
Cytochrome c oxidase subunit 6A (CYTC-N-IV)NuclearIV

F-CTTACCTGGTGTGGCCGTATG

R-GGAGTGCTGCTGCATCTTCA

ATCGCCAACGCCTAC100.1
ATP synthase subunit d (ATP-N-V)NuclearV

F-AGCCGCGATCAACACACA

R-GACGTAGGCAACAGCAGCTTT

AGGCCGAGGCGAA119.4
NADH-ubiquinone oxidoreductase chain 5 (NADH5-M-I)MitochondrialI

F-GGTCACCATCGGACTTAACCA

R-GGCGTGGGTACAGATGTGAA

CCACAGCTGGCCTT104.3
Cytochrome b (CYTB-M-III)MitochondrialIII

F-GCCATATTTGCCGAGATGTCA

R-TGCTCCATTGGCGTGAATATT

CGGCTGACTTATCCG100.8
Cytochrome c oxidase subunit 3 (CYTC-M-IV)MitochondrialIV

F-GAGAGGGCACATTCCAAGGA

R-TGCCGTAGCGTAGGCCTTT

ACCACACGCCCCC98.4
ATPase subunit 6 (ATP6-M-V)MitochondrialV

F-TCCGCCCCCTTGCTCTAG

R-TAGAAGGTGGCCTGCCGTAA

CGTACGGCTTACAGCCA86.1
ATPase subunit 8 (ATP8-M-V)MitochondrialV

F-CTAACTGTTATTCCCCCCAAAGTC

R-TGCTTTGTGAGGTAGGCTCATTT

TTGGCCACACCTTC95.4

qPCR analysis

We assessed the efficiency of each of the qPCR assays through a validation experiment (Applied Biosystems). We then utilized the comparative C T method (ΔΔ C T method, Applied Biosystems) to quantify inter-individual variation in the relative amounts of target OXPHOS cDNAs in the muscle tissue. This method calculates the relative quantity (RQ) of cDNA compared to a reference sample (calibrator), and RQ values represent fold changes in transcription relative to the reference sample, which has a value of one (Applied Biosystems). We amplified target OXPHOS genes from the cDNA in triplicate using the ABI 7500 Fast Real-Time PCR System (Applied Biosystems) and TaqMan Universal PCR Master Mix (Applied Biosystems). All quantitative PCRs (validation and comparative C T) were conducted in 10-μL reactions, and we used the default thermocycler setting for Fast qPCR TaqMan assays, consisting of an initial denaturation step at 95 °C for 20 s, and then 40 cycles of 95 °C for 3 s and 60 °C for 30 s of annealing/extension. The quantity of amplification product was assessed following each PCR cycle. We used ABI's Human Euk 18S rRNA gene assay (Applied Biosystems) as an endogenous control to normalize the total quantity of cDNA loaded into the PCR wells. The Human Euk 18S gene has previously been shown to be appropriate for the normalization of transcription assays in whitefish (Jeukens et al., 2009) and in other salmonids (Olsvik et al., 2005; Jørgensen et al., 2006). We ran an RT reaction without reverse transcriptase and tested for amplification in the qPCRs to ensure the efficacy of our DNase treatment (Bustin et al., 2009). We also included negative controls (no template) in each qPCR.

Statistical analyses

We examined the influence of mitochondrial lineage (Atlantic, Acadian) on gene transcription levels (RQ) in Webster Lake using a multivariate analysis of variance (manova) that included all nine OXPHOS genes as dependent variables, and species (dwarf or normal) and mitochondrial lineage as fixed factors. The identity function was used as the response specification. manova was also used to investigate differences in transcription of the nine OXPHOS genes between the two lakes and species. In this model, lake and species (main effects), and the interaction between lake and species, were included as fixed factors. Sex may affect the transcription of genes in the lake whitefish (Jeukens et al., 2009), and thus, we also included sex as a fixed factor in the model. As the multivariate gene transcription response variable was not normally distributed, we examined the significance of the manova model using Pillai's Trace statistic, as this statistic is robust to deviations from normality (Zar, 1999). To visually examine standardized multivariate gene transcription differences among individual whitefish, we calculated canonical scores using the eigenvectors associated with the first two manova roots, which were significant for both Cliff and Webster lake whitefish. To ensure that we had sufficient power to detect any differences in transcription levels among the groups, we ran a post hoc power analysis on our data set in G*Power v.3.0 (Faul et al., 2007). Using a manova test with repeated measures and within–between interaction specifications at α = 0.05, our power (1 − β) to detect differences among the groups was 0.96.

To confirm any observed relationships between the lake, species and lake × species factors and OXPHOS gene transcription, we conducted a permutation test in SAS (SAS Institute Inc., Cary, NC, USA). Permutation tests are distribution-free tests, and thus, the results of this analysis are robust to violations of normality. In the permutation test, the F-statistic associated with each factor, derived from the original manova test, was compared to a distribution of F-statistics calculated from a randomization of group assignments (lake, species and lake × species) across the observed gene transcription values. The randomization was repeated for 1000 iterations to generate the distribution of F-statistics. When the observed (original) F-statistic fell significantly (< 0.05) outside of the permutation distribution, this was taken as evidence of a relationship between that factor and gene transcription levels (see Good, 2000).

Univariate two-way anovas were run following the manova to examine the influence of lake, species, the lake × species interaction and sex on the transcription of individual genes. We used Spearman's rank correlations to examine gene transcription-level similarities among the complexes and between the mitochondrial and nuclear genes within each of the OXPHOS complexes in Cliff and Webster lakes. All statistical tests, excluding the permutation test, were run in JMP v.9.0 using an alpha = 0.05 as the threshold for statistical significance. Means are reported ±1 SD.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. Data accessibility
  10. References
  11. Supporting Information

Transcription levels between species and lakes

We did not observe an effect of mitochondrial lineage on OXPHOS gene RQ in whitefish from Webster Lake (manova model: Pillai's Trace = 1.52, F 8,18 = 1.41, = 0.321; species: F 3,9 = 1.68, = 0.366; mtDNA lineage: F 3,9 = 0.75, = 0.678). However, across the two lakes, we observed significant lake, species and lake × species effects on RQ for the complement of OXPHOS genes (Table 2). The significant interaction between lake and species indicates that while the levels of OXPHOS gene transcription found in dwarf and normal whitefish differ, the effect of species on transcription levels of each of the genes also varies by lake.

Table 2. Results of manova examining the influence of lake and species on oxidative phosphorylation (OXPHOS) gene transcription in the lake whitefish (Coregonus clupeaformis). Sex was also included as a fixed factor in the model. The overall model was tested using Pillai's Trace statistic and each factor using approximate F-statistics. Numerator and denominator degrees of freedom (Num d.f., Denom d.f.) and the significance of the relationship between gene transcription and each factor (P-value) are indicated. Significant P-values (<0.05) are indicated in bold
Pillai's Trace F Num d.f., Denom d.f. P
Model1.422.0136,132 0.002
Intercept33.879,300.001
Lake3.269,300.001
Species2.419,30 0.034
Lake × Species2.259,30 0.047
Sex0.579,300.812

In the permutation analysis, the observed F-statistics associated with the lake, species and lake × species factors were significantly higher than expected when compared to the distribution of F-statistics generated when lake, species and lake × species identities were randomized across the observed gene RQ values (lake: < 0.001; species: P = 0.020; lake × species: = 0.031). These results confirmed the manova results, indicating that significant differences in OXPHOS gene transcription occur both between dwarf and normal whitefish and between Webster and Cliff lakes, independent of species.

Given the significant interaction effect in the manova model, we present the results of this analysis separately for Webster and Cliff lakes. For Cliff Lake, canonical axes one and two explained 31% and 28% of the variation in multivariate OXPHOS gene RQ, respectively. For Webster Lake, axes one and two explained 38% and 33% of the variation in OXPHOS gene RQ, respectively. Projections of the canonical scores for the dwarf and normal whitefish show clear gene transcription differentiation between species in each of the lakes (Fig. 1a,b). For Cliff Lake, dwarf and normal whitefish exhibited group differentiation primarily along axis one, with dwarfs exhibiting higher canonical scores than normals (Fig. 1a). Higher canonical scores along axis one in Cliff Lake are associated with higher RQ for all nine of the OXPHOS genes included in the model, all of which showed positive loadings along this axis (Fig. 1c). In Webster Lake, the canonical scores for dwarf and normal whitefish gene RQ exhibited more overlap than in Cliff Lake (Fig. 1b). Here, gene transcription differentiation is driven along both axes, with Webster dwarfs tending to exhibit lower and higher canonical scores, respectively, along axes one and two compared to the normal whitefish (Fig. 1b). In particular, lower scores along axis one are associated with higher RQ at the NADH5-M-I, CYTB-N-III, ATP8-M-V and NADH75-N-I genes, whereas higher scores along axis two are associated with higher RQ at the CYTC-M-IV and ATP8-M-V genes (Fig. 1d).

image

Figure 1. manova results examining the influence of species (dwarf, normal) on oxidative phosphorylation (OXPHOS) gene transcription levels in lake whitefish ( Coregonus clupeaformis) from Cliff and Webster lakes in Maine, USA. (a, b) show OXPHOS gene transcription canonical scores projected onto canonical axes one and two, for dwarf (black circles) and normal (open circles) whitefish examined in Cliff and Webster lakes, respectively. The canonical scores for each individual were calculated from the eigenvectors of the first two manova roots. The stars within each figure represent the mean canonical scores, and the ellipses encompass 75% of the data points for each species. (c, d) show the mean centroid values for each of the nine OXPHOS genes plotted across the first two canonical axes for Cliff and Webster lakes, respectively.

Download figure to PowerPoint

The univariate anovas revealed a lake (main) effect on RQ of both of the complex I genes (NADH5 and NADH75) and the ATP6-M-V gene, with both dwarfs and normals in Cliff Lake exhibiting significantly higher transcription levels than observed in Webster Lake whitefish (Table 3; post hoc Student's t-test: < 0.05). We observed a significant species (main) effect on RQ levels of OXPHOS genes NADH5-M-I, CYTB-M-III and ATP8-M-V (Table 3) with dwarfs showing gene upregulation in both Cliff and Webster lakes (post hoc Student's t-test: < 0.05, Fig. 2). We also observed a near-significant trend for dwarf whitefish to exhibit higher CYTC-M-IV RQ than normals (Table 3, Fig. 2). Higher-order lake × species effects were observed at the CYTB-N-III and CYTC-N-IV genes. Specifically, Cliff dwarfs exhibited CYTB-N-III gene upregulation compared to Webster dwarfs and normal whitefish from both lakes (Tukey's post hoc test: P < 0.05). Cliff dwarfs also exhibited significant gene upregulation compared to Cliff normals at the CYTC-N-IV gene (Tukey's post hoc test: P < 0.05; Table 3, Fig. 2), whereas no difference was observed between dwarfs and normals from Webster Lake. Sex was not a significant predictor of RQ for any of the genes (Table 3).

image

Figure 2. Variation in oxidative phosphorylation (OXPHOS) gene transcription levels in dwarf and normal lake whitefish ( Coregonus clupeaformis) from Cliff and Webster lakes in Maine, USA. Relative gene transcription quantities (relative quantity, RQ) ± 1 SD are shown for each of the nine OXPHOS genes examined in Cliff Lake dwarfs (black bars), Cliff Lake normals (light grey bars), Webster Lake dwarfs (dark grey bars) and Webster Lake normals (white bars). The gene name abbreviations shown on the x-axis correspond to those outlined in Table 1. See Table 3 for the results of the univariate anova models examining the contribution of species (dwarf, normal) and lake to variation in OXPHOS gene transcription. Note that RQ cannot be compared among genes as different calibration samples were used across some of the gene transcription assays.

Download figure to PowerPoint

Table 3. Results of univariate anova tests examining the contribution of lake and species to the transcription of oxidative phosphorylation (OXPHOS) genes in lake whitefish (Coregonus clupeaformis). Definitions for the abbreviated gene names can be found in Table 1. For each factor, the F-statistic and P-value (in brackets) are indicated. Model and error degrees of freedom (d.f.) are indicated for each model, and d.f. for each factor are indicated. Significant P-values (<0.05) are indicated in bold text
DFNADH75-N-INADH5-M-ICYTB-N-IIICYTB-M-IIICYTC-N-IVCYTC-M-IVATP-N-VATP6-M-VATP8-M-V
Model4,381.89 (0.131)3.65 (0.013)11.74 (<0.001)2.14 (0.094)4.17 (0.006)1.31 (0.285)1.63 (0.186)10.44 (<0.001)2.91 (0.034)
Lake15.64 (0.023)4.41 (0.043)23.97 (<0.001)0.60 (0.443)0.68 (0.414)0.01 (0.987)1.99 (0.166)28.76 (<0.001)2.72 (0.107)
Species11.07 (0.307)4.30 (0.045)17.30 (<0.001)7.43 (0.001)7.16 (0.011)3.89 (0.056)0.01 (0.982)3.42 (0.072)7.08 (0.011)
Lake × Species11.07 (0.307)2.63 (0.113)4.22 (0.047)0.01 (0.998)7.59 (0.009)0.18 (0.672)0.09 (0.756)2.69 (0.109)0.04 (0.838)
Sex10.04 (0.828)1.46 (0.235)0.89 (0.895)1.21 (0.279)1.26 (0.269)1.26 (0.267)2.37 (0.131)0.84 (0.364)0.86 (0.361)

Nuclear and mitochondrial gene transcription correlation

Relative transcription quantities (RQ) for the nuclear and mitochondrial genes associated with complex I were not significantly correlated in either Cliff or Webster Lake (Table 4). However, the RQ of nuclear and mitochondrial genes within complexes III and IV were significantly correlated (Table 4). For complex V, the nuclear and mitochondrial genes exhibited uncorrelated RQ, whereas the two mitochondrial complex V genes, ATP6 and ATP8, showed significantly correlated RQ in both lakes (Table 4). Examining all genes, we observed 18 significant and three nearly significant correlations in transcription levels between genes in Cliff Lake and 17 significant and two nearly significant correlations in transcription levels between genes in Webster Lake (Table S1).

Table 4. Correlation of gene transcription quantities (relative quantity) within each of the nuclear and mitochondrial-encoded oxidative phosphorylation (OXPHOS) complexes (I, III–V) in the lake whitefish (Coregonus clupeaformis) from Cliff and Webster lakes in Maine, USA. Spearman's rank correlation coefficients (ρ) and P-values for each test are shown. Significant P-values (<0.05) are indicated in bold text
ComplexGenesCliff LakeWebster Lake
ρ P ρ P
INADH5-M and NADH75-N−0.0140.9550.3720.081
IIICYTB-M and CYTB-N0.596 0.006 0.723<0.001
IVCYTC-M and CYTC-N0.553 0.011 0.668<0.001
VATP6-M and ATP-N−0.0180.9390.0780.723
VATP8-M and ATP-N−0.0570.8110.2280.295
VATP6-M and ATP8M0.532 0.016 0.610 0.002

Eight and 10 of these significant correlations were observed between the five mitochondrially encoded genes in Cliff Lake and Webster Lake, respectively.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. Data accessibility
  10. References
  11. Supporting Information

Parallelism in OXPHOS gene transcription

The pronounced parallel divergence of metabolic traits between multiple pairs of dwarf and normal whitefish has led to their recognition as key attributes involved in ecological speciation (Bernatchez et al., 2010). Here, we found strong evidence of differential gene transcription in the muscle tissues of dwarf and normal whitefish from Cliff and Webster lakes through both multivariate and univariate analyses of gene transcription patterns. Indeed, at least one gene from each of OXPHOS complexes I, III, IV and V exhibited significant or near-significant upregulation in dwarf compared to normal whitefish in both lakes. These results support previous research showing higher levels of transcription in dwarf compared to normal whitefish for genes associated with metabolism (Derome et al., 2006; St-Cyr et al., 2008). For example, Derome et al. (2006) used the salmonid 3557 cDNA microarray to examine differential gene transcription in the muscle tissues of lake whitefish from Cliff Lake and Indian Pond in Maine, USA, and found evidence of upregulation in Cliff dwarfs for two ESTs associated with cytochrome c oxidase (complex IV), albeit the same pattern was not observed in Indian Pond dwarfs. However, the authors did observe gene transcription upregulation in dwarfs from both populations at an EST associated with ATP synthase (Derome et al., 2006). Therefore, our results confirm previously observed patterns of OXPHOS gene transcription upregulation in dwarf whitefish from microarray studies. Moreover, the observed parallelism in OXPHOS gene transcription divergence between dwarf and normal whitefish from Cliff and Webster lakes lends further support to the hypothesis that metabolic traits play a key role in the diversification of the lake whitefish.

Diverging ecological factors, both biotic and abiotic, could contribute to the general pattern of OXPHOS gene upregulation observed in dwarf compared to normal whitefish (see Guderley, 2004). For instance, the upregulation of genes associated with OXPHOS has been observed in other systems, in response to both cold temperatures and hypoxia-induced stress (e.g. Gracey et al., 2001, 2004; Storey, 2007; Cheviron et al., 2008). While hypoxia and temperature differences do occur between lake zones in Webster and Cliff lakes, these factors are unlikely to explain the observed upregulation of gene transcription in the dwarf whitefish, as, at least in the summer months, dwarf and normal whitefish experience similar temperatures and oxygen saturation levels (Landry et al., 2007). However, there could be divergent energetic demands placed on dwarf and normal whitefish due to differences in prey utilization or predation pressures. Studies have shown that dwarf whitefish typically forage in the limnetic environment on zooplankton, whereas normal whitefish forage on benthic macroinvertebrates (Trudel et al., 2001; Landry et al., 2007; Landry & Bernatchez, 2010; also see Robinson & Wilson, 1994; Kahilainen et al., 2004). While we are not currently aware of any studies examining the influence of diet of OXPHOS gene transcription in wild populations, a recent study in cattle showed an influence of feed type on the transcription of genes associated with OXPHOS (Kelly et al., 2011). Little is currently known about predation pressures faced by the dwarf and normal whitefish in Cliff and Webster lakes or the potential influence of predation on lake whitefish metabolism. However, studies of European whitefish (C. lavaretus) have shown that dwarfs face greater predation pressures (Kahilainen & Lehtonen, 2003), which may contribute to the evolution of their smaller, more streamlined size and associated differences in behaviour and metabolism compared to normal whitefish (Trudel et al., 2001). Furthermore, increased predation risk is associated with the upregulation of genes associated with metabolism in the stickleback (Gasterosteus aculeatus), albeit this has only been examined in brain tissues (Sanogo et al., 2011). Taken together, the results from this and other studies suggest that ecological factors, including differences in diet and predation risk, are important drivers of OXPHOS gene transcription divergence between dwarf and normal whitefish.

Nonparallel patterns of OXPHOS gene transcription between lakes

Natural replicates of ecological speciation events provide an exceptional opportunity to examine not only the factors driving trait parallelism but also those contributing to nonparallel variation (Landry & Bernatchez, 2010; Rosenblum & Harmon, 2011; Fan et al., 2012; Kaeuffer et al., 2012). While we observed parallel patterns of transcription divergence between dwarf and normal whitefish in Cliff and Webster lakes for some of the OXPHOS genes, we also observed significant differences between the lakes, independent of species (Landry & Bernatchez, 2010; also see Kaeuffer et al., 2012). First, whitefish from Cliff Lake generally exhibited higher levels of gene transcription than in Webster Lake for the nuclear and mitochondrial complex I genes and the ATP subunit 6 gene. We also observed significant upregulation in dwarfs at the nuclear cytochrome b and cytochrome c genes in Cliff Lake but not in Webster Lake. Ecological variation between the two lakes examined in this study could have profound effects on the patterns of observed OXPHOS gene transcription. For example, Webster and Cliff lakes are significantly diverged in prey availability and size, with Webster Lake exhibiting larger prey at lower densities than is found in Cliff Lake (Landry & Bernatchez, 2010). Furthermore, gill raker morphology shows more overlap between dwarf and normal whitefish in Webster than in Cliff Lake, suggesting that that two species exploit relatively similar foraging opportunities in Webster Lake (Landry et al., 2007). Our results suggest that natural replicates of whitefish species pairs exhibit as many differences as they do similarities in patterns of OXPHOS gene transcription, which could be related to the nonparallel ecological conditions experienced by whitefish among lakes.

In addition to ecological factors, nonparallel patterns of OXPHOS gene transcription could be related to differences in the genetic background of the whitefish populations. Dwarf and normal whitefish exhibit lower levels of genetic differentiation and more gene flow in Webster than in Cliff Lake (Bernatchez & Dodson, 1994; Pigeon et al., 1997; Lu & Bernatchez, 1999; Campbell & Bernatchez, 2004; Renaut et al., 2011), which could contribute to the less-pronounced divergence we observed in OXPHOS gene transcription between the species in Webster Lake. We found no evidence that ancestral mitochondrial lineage influences gene transcription levels, albeit our ability to relate mitochondrial haplotype to transcription was somewhat constrained by sample size and thus requires further study. However, it is also possible that more recently derived genetic differences between the populations could play a role in mediating OXPHOS gene transcription. For instance, a recent study of functional genetic differentiation between dwarf and normal whitefish showed that single nucleotide polymorphism (SNP) markers associated with NADH-ubiquinone oxidoreductase chain 5 (complex I) and cytochrome c oxidase subunit 3 (complex IV) are highly diverged between dwarf and normal whitefish in Cliff Lake, but exhibit no differentiation in Webster Lake (Renaut et al., 2011). Thus, there is at least some evidence that coding regions associated with OXPHOS genes exhibit differing genetic backgrounds, which could lead to variation in gene transcription levels between the populations.

Gene translation, whereby the products of gene transcription (messenger RNA) are decoded into amino acids, has also been implicated as a potentially important mechanism contributing to metabolic capacities in fish species. Indeed, differences in mitochondrial content among individuals may relate as much to translational differences as gene transcription (Hardewig et al., 1999; Moyes, 2003). In a study of billfish (Xiphiidae and Istiophoridae) and tuna (Thunnus spp.), Dalziel et al. (2005) observed no difference between the two groups in transcription levels of citrate synthase, an indicator of mitochondrial content in cells. However, the groups exhibited significant differences in citrate synthase enzyme content in their muscle, suggesting that gene translation is the major factor regulating levels of this enzyme (Dalziel et al., 2005). Here, we observed differences among populations in the transcription profiles of OXPHOS genes, but it is possible that dwarf whitefish from Cliff and Webster lakes take different functional pathways to colonize the limnetic zones in each lake (see Derome et al., 2006). Thus, the functional significance of our results must be interpreted with some caution until studies of enzymatic function in dwarf and normal whitefish from each of the lakes can be conducted.

Nuclear and mitochondrial OXPHOS gene co-transcription

It has been suggested that the co-adaptation of nuclear and mitochondrial genes of the OXPHOS pathway is an important mechanism contributing to the formation of reproductive barriers between locally adapted populations (Blier et al., 2001; Gershoni et al., 2009). In whitefish, we found evidence of correlated nuclear and mitochondrial gene transcription within OXPHOS complexes III and IV, suggesting that the nuclear and mitochondrial genomic regions encoding this portion of the OXPHOS pathway are co-adapted. In contrast, the transcription of the nuclear and mitochondrial genes examined from complexes I and V appears to be uncorrelated. Complex IV is thought the be the rate-limiting complex in OXPHOS, and complex III is one of the primary producers of ROS; therefore, the consequences of nuclear and mitochondrial transcription decoupling for these complexes may be of particular importance to individual fitness (see Lenaz, 2001; Grossman et al., 2004). Similar to our results, studies in humans have shown that OXPHOS complex I genes exhibit uncorrelated transcription, and it has been suggested that the widespread genomic distribution of the genes encoding for the ~43 complex I subunits may complicate coordinated transcription (Garbian et al., 2010; also see Van Waveren & Moraes, 2008). Thus, it is possible that our results for complex I and complex V, which is composed of the largest number of subunits save for complex I (Ballard & Melvin, 2010), reflect similarly complicated regulation of nuclear and mitochondrial gene transcription. Overall, our results show that correlated transcription of nuclear and mitochondrial-encoded genes occurs in the lake whitefish and suggest that transcription coordination is of particular importance in the central complexes of the OXPHOS pathway.

In contrast to our prediction of lower nuclear and mitochondrial gene transcription correlation in Webster compared to Cliff Lake, we observed similar patterns in the two lakes. This result suggests that gene flow between dwarf and normal whitefish may not be an important factor influencing the coordination of OXPHOS gene transcription. However, it is possible that the natural rates of gene flow between dwarf and normal whitefish are too low in either of the lakes examined to cause a breakdown in OXPHOS gene transcription. Indeed, estimates of migration rate (per generation) in Cliff and Webster lakes are only 0.0004 and 0.0006, respectively (Campbell & Bernatchez, 2004). Experimental studies examining OXPHOS gene transcription in dwarf–normal hybrids are needed to investigate how mitochondrial and nuclear genomic mismatch may contribute to the breakdown of co-adapted OXPHOS genes, particularly in complexes III and IV, and to test whether the decoupling of genes contributes to reduced fitness and reproductive isolation.

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. Data accessibility
  10. References
  11. Supporting Information

This study points to the importance of eco-physiological interactions in the repeated ecological speciation of the lake whitefish. Dwarf whitefish showed OXPHOS gene upregulation in both Webster and Cliff lakes, suggesting that parallel ecological conditions faced in the two lakes lead to the requirement for enhanced ATP production via OXPHOS in dwarfs. Further studies are needed to examine the extent to which these differences are regulated via plastic responses to environmental variation and evolutionary genetic changes in each population. Nevertheless, our results, in combination with previous studies that have shown a heritable basis for transcription variation in other populations, point to OXPHOS gene transcription divergence as a factor potentially involved in delineating reproductive barriers between dwarf and normal whitefish (St-Cyr et al., 2008; Whiteley et al., 2008; see Gershoni et al., 2009). The patterns of OXPHOS gene transcription were also highly variable between the two lakes potentially reflecting divergence in response to nonparallel ecological conditions, differing evolutionary (genetic) backgrounds, or variation in the functional pathways used to achieve similar phenotypes in each lake. Contemporary ecological and historical genetic differences across lakes could lead to an evolutionary trade-off that influences the trajectory of gene transcription divergence in each species pair. Our results thus highlight the importance of investigating parallel and nonparallel aspects of trait variation to fully understand how ecologically driven selection produces optimal phenotypes in nature.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. Data accessibility
  10. References
  11. Supporting Information

We thank W. Adam, D. Basley and G. Côté for assistance in the laboratory and field. Thanks to A.-S. Julien for statistical advice and for assistance in running the manova permutation tests. S. Pavey, M. Filteau, T. Flatt and three anonymous reviewers provided helpful comments on the manuscript. This work was supported by a Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant and Canada Research Chair Grant to L. Bernatchez and an NSERC PDF award to M. Evans. The authors report no conflict of interest.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. Data accessibility
  10. References
  11. Supporting Information

Data deposited at Dryad: doi: 10.5061/dryad.s4kc8

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. Data accessibility
  10. References
  11. Supporting Information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

FilenameFormatSizeDescription
jeb2570-sup-0001-TableS1.docxWord document38K Table S1 Pairwise correlations of transcription levels for the nine oxidative phosphorylation genes examined in lake whitefish (Coregonus clupeaformis) from Cliff and Webster lakes in Maine, USA.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.