Journal of Heredity, 2016, 82–89doi:10.1093/jhered/esv069
Symposium ArticleAdvance Access publication August 21, 2015
Symposium Article
Plastic and Evolutionary Gene Expression Responses Are Correlated in European Grayling (Thymallus thymallus) Subpopulations Adapted to Different Thermal EnvironmentsHannu Mäkinen, Spiros Papakostas, Leif Asbjørn Vøllestad, Erica H. Leder, and Craig R. Primmer
From the Department of Biology, University of Turku, Pharmacity, Itäinen Pitkäkatu 4, 20014 University of Turku, Finland (Mäkinen, Papakostas, Leder, and Primmer); and Centre for Ecological and Evolutionary Synthesis, Department of Biosciences, University of Oslo, OSLO, Norway (Vøllestad).
Address correspondence to Hannu Mäkinen at the address above, or e-mail: [email protected].
Received March 31, 2015; First decision April 20, 2015; Accepted July 17, 2015.
Corresponding Editor: Robin Waples
Abstract
Understanding how populations adapt to changing environmental conditions is a long-standing theme in evolutionary biology. Gene expression changes have been recognized as an important driver of local adaptation, but relatively little is known regarding the direction of change and in particular, about the interplay between plastic and evolutionary gene expression. We have previously shown that the gene expression profiles of European grayling (Thymallus thymallus) populations inhabiting different thermal environments include both plastic and evolutionary components. However, whether the plastic and evolutionary responses were in the same direction was not investigated in detail, nor was the identity of the specific genes involved. In this study, we show that the plastic changes in protein expression in response to different temperatures are highly correlated with the evolutionary response in grayling subpopulations adapted to different thermal environments. This finding provides preliminary evidence that the plastic response most likely facilitates adaptation during the early phases of colonization of thermal environments. The proteins that showed significant changes in expression level between warm and cold temperature treatments were mostly related to muscle development, which is consistent with earlier findings demonstrating muscle mass differentiation between cold and warm grayling populations.
Subject areas: Gene action; Regulation and transmission; Molecular adaptation and selectionKey words: evolutionary adaptation, gene expression, plasticity, proteomics.
It is widely accepted that phenotypic plasticity is a characteristic of most living organisms and it can aid population persistence dur-ing periods of rapid environmental change (Wund 2012; Forsman
2014; Murren et al. 2015). Recently, there has been growing inter-est in understanding various aspects of plasticity and its interplay with evolutionary adaptation in the ability to cope with changing
Downloaded from https://academic.oup.com/jhered/article-abstract/107/1/82/2622829by gueston 13 April 2018
environmental conditions, but the picture is not yet clear (Forsman 2014; Murren et al. 2015). There are relatively few examples where the relative contributions of plasticity and evolutionary adapta-tion in phenotypic change have been dissected, especially in a cli-mate change context (Gienapp et al. 2008; Merilä 2012; Merilä and Hendry 2014). Although several ecological responses such as phenological changes to earlier breeding season and range shifts to higher latitudes and altitudes have been commonly detected, the relative roles of plasticity and evolutionary adaptation behind these responses are poorly understood (Skelly and Freidenburg 2010; Merilä and Hendry 2014). Studies based on model systems mimick-ing very early stages of adaptation can give insight on how popula-tions adapt to rapid anthropogenic change.
Recent conceptual and theoretical work has also revitalized the importance of plasticity at the early stages of speciation and diversi-fication (Chevin et al. 2010; Fusco and Minelli 2010; Pfennig et al. 2010; Draghi and Whitlock 2012; Wund 2012). The immediate plastic response induced by the environment may produce adaptive pheno-types without the need for novel mutations for which the waiting time can be long (Pfennig et al. 2010). Theoretical work suggests that plastic-ity can indeed induce a faster rate of adaptation by maintaining genetic variation and buffering against population bottlenecks (Espinosa-Soto et al. 2011; Fierst 2011; Draghi and Whitlock 2012; Gomez-Mestre and Jovani 2013). Inferring the role of plasticity in adaptation is not however straightforward due to the complexity of possible responses to the new environment (Ghalambor et al. 2007; Fitzpatrick 2012; Morris and Rogers 2013). Plasticity may also drive the population further away from the new optimum resulting in a nonadaptive (or maladaptive) response. In this case, the nonadaptive response results in lower fitness in the new environment relative to the ancestral environ-ment. Adaptive plasticity on the other hand, can drive the population to the new optimum and may thus facilitate adaptation (Ghalambor et al. 2007; Fitzpatrick 2012; Morris and Rogers 2013). In other words, the reaction norm of adaptive plasticity is in the same direc-tion as would be expected if the trait would evolve under directional selection, while for nonadaptive plasticity the opposite pattern could be expected (Ghalambor et al. 2007; Fitzpatrick 2012). The adaptive plas-tic response may not always result in a perfect match but rather a step closer to the new optimum (Ghalambor et al. 2007). Plastic responses can also be variable with respect to different traits in a given organism and maladaptive response can be compensated by various mechanisms (Morris and Rogers 2013). For example, a maladaptive phenotype may not be visible due to expression of other adaptive phenotypes induced by plastic compensation (Morris and Rogers 2013).
Gene expression regulation is a flexible mechanism for adjust-ing rapidly to the local environment, but it can also be associated with long-term evolutionary response (Romero et al. 2012; Alvarez et al. 2015). Thus studying gene expression may provide information about the importance of plasticity in the early stages of adaptation (Morris et al. 2014). Gene expression plasticity can be estimated by treatment with ecologically important variables such as temperature in standardized conditions. Evolutionary response in gene expression can be inferred by comparing gene expression in divergent popula-tions adapted to different ecological conditions (Hodgins-Davis and Townsend 2009). Understanding gene expression in an evolutionary context may be challenging due to the complexity of gene expression regulation mechanisms. For example, mRNA and protein expression are not always well correlated probably indicating that they are evolv-ing under different evolutionary constraints (Diz et al. 2012; Khan et al. 2013). Furthermore, applying a null model for gene expres-sion evolution to infer selection on gene expression change is not
always straightforward (Fraser 2011). One way to overcome these difficulties, at least partially, is to measure expression at the protein level. It has been suggested that gene expression evolution toward optimized gene function and cell growth may be more relevant at the proteome than in the transcriptome level (Diz et al. 2012; Li et al. 2014). Studying gene expression responses directly from the pro-teome can be a good starting point to elucidate hypotheses regarding gene expression plasticity and evolution of rapid local adaptation.
We have used European grayling (Thymallus thymallus) sub-populations in a recently colonized mountain lake in Norway as a model system to investigate the interplay between plastic and evolutionary responses in protein expression. Earlier research has shown that there have been both evolutionary and plastic responses in gene expression during adaptation to warm and cold tributar-ies (Papakostas et al. 2014). This observation provides a starting point for elucidating in more detail whether plasticity is adaptive or maladaptive in these recently diverged populations. Our hypothesis is based on conceptual work suggesting that if the reaction norm of the plastic response is in the same direction as the evolutionary response it is indicative of adaptive plasticity (Ghalambor et al. 2007; Fitzpatrick 2012). The opposite pattern might reflect plasticity most likely being maladaptive (Ghalambor et al. 2007; Fitzpatrick 2012). In order to test this hypothesis, we used proteome expres-sion profiles of subpopulations adapted to different thermal regimes as a proxy for evolutionary response. We estimated plastic response in subpopulations originating from different thermal environments and compared expression profiles to the evolutionary response. We also aimed to identify proteins that were differentially expressed between thermal environments and considered their adaptive impor-tance in relation to previous studies on phenotypic divergence in the same subpopulations (Gregersen et al. 2008; Kavanagh et al. 2010; Thomassen et al. 2011).
Materials and Methods
Study SystemLake Lesjaskogsvatnet is a small mountain lake (611 m above sea level, 4.52 km2 and 24 m maximum depth) situated in Norway (Figure 1). It was colonized by grayling from the river Gudbrandsdalslågen in the late 1800s (ca. 25 generations ago) fol-lowing a temporary watercourse manipulation. The grayling sub-populations spawning in the various tributaries have been evolving in isolation from the ancestral population since the initial coloniza-tion (Figure 1) (Haugen and Vøllestad 2001). The subpopulations can be roughly classified as “large-cold” or “small-warm” according to the river temperature profiles during the grayling-spawning sea-son and embryo development period in the spring and early sum-mer (Gregersen et al. 2008). Here, we study 2 cold subpopulations (Valåe and Hyrjon) and 2 warm subpopulations (Sandbekken and Steinbekken) (Thomassen et al. 2011). Timing of spawning in the cold and warm subpopulations may differ by 2–5 weeks, due to the differences in water temperature and flow (Junge 2011). The long-term average temperature difference is 1–1.5 °C between cold and warm tributaries, but the differences may be substantially larger in a given year. In total these differing temperatures translate into substantial differences in the cumulative temperature sum (Figure 1, inset fig) (Gregersen et al. 2008; Thomassen et al. 2011). Although divergence in neutral genetic markers is low, the warm and cold sub-populations show divergence in early life-history traits associated with the river temperature during breeding the season (Kavanagh et al. 2010; Junge et al. 2011; Thomassen et al. 2011).
Journal of Heredity, 2016, Vol. 107, No. 1 83
Downloaded from https://academic.oup.com/jhered/article-abstract/107/1/82/2622829by gueston 13 April 2018
Common Garden ExperimentA reciprocal common garden experiment was conducted to esti-mate plastic and evolutionary responses in standardized conditions. A detailed description of the experimental set-up can be found in Thomassen et al. (2011). Briefly, gametes of mature individuals origi-nating from 2 cold subpopulations and 2 warm subpopulations were used to produce fertilized eggs for the experiment. The developing embryos were reared in temperatures reflecting the lower (6 °C) and upper (10 °C) range of temperatures juvenile grayling experiencing in the wild (Thomassen et al. 2011). Earlier studies have shown that mini-mum and maximum temperatures for successful embryo development in these populations are 5 °C and 12 °C (Kavanagh et al. 2010; Junge 2011). Temperature treatments were chosen to reflect native conditions (warm origin population in 10 °C and cold origin population in 6 °C) and non-native temperatures (warm origin population in 6 °C and cold origin population in 10 °C). Embryos were sampled based on the number of degree-days in order to assure they were at similar develop-mental stages in relation to the average number of degree-days to 50% hatching in the subpopulation—temperature treatment (Papakostas et al. 2014). This approach can provide a reasonable approximation of the developmental stage when different temperatures are used in an experiment (Chezik et al. 2014). Plastic response in protein expression was estimated by comparing the expression profiles of subpopulations reared in their native temperature to the expression in non-native tem-perature. Evolutionary response was estimated by comparing cold- and warm-origin subpopulations reared in similar temperatures.
Protein Extraction and Mass SpectrometryFull details about the proteomics methods are available in Papakostas et al. (2014). Briefly, 3 embryos per subpopulation per common garden temperature treatment, altogether 24 samples (3 embryos × 4 subpopulations × 2 temperature treatments) were used for this study. Samples were labeled by iTRAQ, fractionated by strong
cation exchange (SCX) chromatography, and combined accordingly to minimize batch effects. To extract whole-organism proteins from the fish embryos, we employed a standard SDS-based protocol, and concentration was determined with a NanoDrop ND-1000 spec-trophotometer. Proteins were then trypsin-digested and peptides were labeled with isobaric tags by iTRAQ reagents (4-plex, Applied Biosystems) following the manufacturer’s protocol. To increase proteome coverage, each sample was peptide-fractionated into 4 fractions using a SCX column and buffers supplied as part of the ICAT Cation Exchange Buffer Pack according to the manufacturer’s protocol (Applied Biosystems). Each fraction was then separated by reverse-phase chromatography using an Easy-nLC II nanoflow system connected to an Orbitrap Velos mass spectrometer (Thermo Fisher Scientific). The mobile phases were 0.2% formic acid/2% ace-tonitrile (A) and 0.2% formic acid/95% acetonitrile (B). Peptides were separated at a flow rate of 300 nL/min with 102 min gradients as follows: initially 2% B to 25% B (60 min), 40% B (90 min), and 100% B (92–102 min). Data were acquired in the data-dependent mode with up to 10 MS/MS scans being recorded for each precursor ion scan. Precursor ion spectra were recorded in profile mode in the Orbitrap (m/z 300–1800, R = 30 000 at m/z 400, max injection time 100 ms, and max 1 000 000 ions), and MS/MS spectra were acquired in centroid mode in the Orbitrap (R = 7500 at m/z 400, max injec-tion time 200 ms, max 50 000 ions, HCD stepped normalized CE 40% and 50%, isolation width 2, activation time 0.1 ms, and the first mass fixed at m/z 100). The instrument control software LTQ Tune Plus version 2.6.0.1050 was used in this study, and the HCD energy of 40 on that version would be equivalent to 35 on the updated version 2.7.0.1093. Mono-isotopic precursor selection was enabled, singly charged ions and ions with an unassigned charge state were rejected, and each fragmented ion was dynamically excluded for 90 s. The lock-mass option was enabled (m/z 445.120025). In every step of the molecular work, samples were properly randomized to minimize batch effects.
Figure 1. Map of the sampling sites and the temperature profiles of cold and warm subpopulations (inset figure) during the grayling fertilization, embryo and juvenile development periods. Horizontal axis in the inset map refers to the time points (date) of the temperature records smoothed across years (square = cold origin subpopulation, circle = warm origin subpopulation).
84 Journal of Heredity, 2016, Vol. 107, No. 1
Downloaded from https://academic.oup.com/jhered/article-abstract/107/1/82/2622829by gueston 13 April 2018
Measuring Relative Protein Expression LevelsProteins were identified and quantified using the ProteinPilot v.4 program (Applied Biosystems). As a search database, we used all Atlantic salmon (Salmo salar) amino acid sequences submitted to the UniProt database (www.uniprot.org) as of March 2012 (13 035 amino acid sequences). The mass spectrometry files (*.RAW), the processed peaklists (*.mgf), the search databases (*.fasta), and the results of the ProteinPilot searches (*.xml) for both the validation and the actual experiment have been deposited in the ProteomeXchange consortium (http://proteomecentral.proteomex-change.org) via the PRIDE partner repository (Vizcaíno et al. 2013) with the data set identifier PXD000368. Maximum false discov-ery rate correction for protein identification was performed using a decoy database as implemented in ProteinPilot and was set to 5%, whereas minimum confidence for peptide identification was 95%. A collection of 248 sequences of common- contaminant proteins, provided by Applied Biosystems, was also included in the search database. Contaminant and decoy hits were filtered and samples were divided into 4 groups of 6 samples each, namely the grayling embryos from each of the cold/warm thermal origin subpopula-tions with each of the 6 °C/10 °C temperature treatments. For the statistical analyses, proteins with less than 3 valid ratios per ther-mal origin—temperature treatment combination (6 samples) were excluded from the dataset. Ratios in this filtered dataset were then log2-transformed and then loess-normalized across biological repli-cates using the median values as a reference set. The few remaining missing expression values in the filtered dataset were replaced with normalized expression values imputed from the normal distribu-tion. For these transformations, we used DanteR, an R package for the analysis of proteomic data (Taverner et al. 2012). The variation in protein expression across individuals and treatments are shown in Supplementary Figure 2 online. In fulfillment of data archiving guidelines (Baker 2013), we have deposited the primary data under-lying these analyses into the Dryad database as a Supplementary Table online (doi:10.5061/dryad.7bp03).
Statistical MethodsAltogether 3 biological replicates were used for protein quantifi-cation of expression from each subpopulation and temperature treatment. Data from the same temperature origin subpopulations were combined for the data analyses, that is, Hyrjon-Valåe and Steinbekken-Sandbekken to increase the sample size in differen-tial gene expression analyses. Fold changes in protein expression between temperature treatments were used to estimate correlation between plastic and evolutionary responses. Pearson, Kendall and Spearman coefficients were used to estimate correlation and permu-tation tests were used to estimate the significance of each correla-tion coefficient. A permutation test was used to test if correlations were solely caused by common terms in the fold change equations. Thousand permutations were used by randomly assigning treatment labels to the expression values of the samples. Permutation tests were carried out in the R-environment using a custom made script. Fold change was calculated as expression in non-native temperature divided by the expression in native temperature. In the first com-parison, the plastic change was estimated by comparing cold-origin subpopulations reared in 6 °C and 10 °C (cold plastic response: Corigin10/Corigin6), and the evolutionary response by comparing pro-tein expression of cold and warm origin subpopulations reared in 6 °C (cold evolutionary response: Worigin6/Corigin6). In the second comparison, the plastic response was estimated by comparing the
protein expression of warm-origin subpopulations reared in 10 °C and 6 °C (warm plastic response: Worigin6/Worigin10), whereas the evo-lutionary response by comparing the expression of warm and cold origin subpopulations reared in 10 °C (warm evolutionary response: Corigin10/Worigin10). Finally, a comparison averaging the plastic and evolutionary responses across subpopulations [(Corigin6 + Worigin6)/(Corigin10 + Worigin10)]/[(Corigin6 + Corigin10)/(Worigin6 + Worigin10)] was conducted to investigate if there was any contribution of environ-mental effects, that is, plasticity itself, in the evolutionary compo-nent. The design for the plastic response should reflect a case when cold or warm adapted individuals might occasionally migrate to the opposite temperature regimes for spawning. Evolutionary responses reflect the differentiation in gene expression after 20–25 generations of divergence from a common ancestral population.
The differential gene expression of specific proteins was esti-mated by fitting a linear model using the lmfit function as imple-mented in the limma R package (Smyth 2004). Separate analyses were conducted for plastic and evolutionary responses. Both mod-els included the plastic and evolutionary contrasts described above and their interaction. A moderated t-statistic was estimated with the eBayes function to obtain P-values for each protein as implemented in the limma R-package (Smyth 2004). All P-values were corrected for multiple testing according to Benjamini–Hochberg procedure at the 5% significance level. A gene ontology (GO) category enrichment test was carried out for those proteins that were identified as differ-entially expressed in linear model analyses. The search for enriched GO terms was done by using statistically significant proteins as tar-gets and the whole data set (408 proteins) as a background reference and contrasted against the human protein ortholog database. The GO enrichment analyses were conducted using the GOrilla web tool (Eden et al. 2009). The enrichment results were corrected for multi-ple testing by applying false discovery rate (Benjamini–Hochberg) at the 5% significance level.
Results
There were strong and significant (P value < 0.001 or < 0.01) posi-tive correlations between both plastic and evolutionary responses in fold changes of the 408 protein expression profiles (Figure 2a–c, Table 1). All correlations were robust across Kendall, Spearman, and Pearson correlation coefficients and the observed estimates were located in the upper tail of the permuted distribution (Supplementary Figure 1a–c online). The correlation estimates for cold plastic versus cold evolutionary responses were 0.84, 0.54, and 0.71 for Pearson, Kendall, and Spearman coefficients, respectively. The correlation estimates for warm plastic and warm evolutionary responses were in the same range (Table 1). Also the comparison of the average plastic and evolutionary responses across treatments/subpopulations revealed high and positive correlations (Figure 2c). The majority of the significantly differentially expressed proteins had subpopula-tion or treatment interactions (Figure 3 and Supplementary Table online). Altogether 5 and 25 proteins were significantly differentially expressed in evolutionary and plastic response comparisons respec-tively, after correcting for multiple testing (FDR q-value < 0.05) (Figure 3). In the evolutionary response comparison all 5 proteins had significant interaction (Figure 3). In the plastic comparison 17 proteins had a significant interaction whereas one protein had a sig-nificant response to both cold and warm thermal treatments as well their interaction. Three proteins had both significant effect in cold plastic response and an interaction. Four proteins had significant warm plastic response and interaction effects (Figure 3). Significant
Journal of Heredity, 2016, Vol. 107, No. 1 85
Downloaded from https://academic.oup.com/jhered/article-abstract/107/1/82/2622829by gueston 13 April 2018
enrichment for 9 and 13 GO categories was detected after correc-tion for multiple testing (FDR q-value < 0.05) for evolutionary and plastic responses, respectively (Table 2). All categories were related to muscle function processes but each GO category was associated to a combination of a small number of genes (2–4). These genes were TPMI1 (tropomyosin), TNNI2 (troponin I type 2), TNNT3 (tro-ponin t type 3), and MYLPF (myosin light chain).
Discussion
Our study demonstrates that the reaction norms of plastic and evo-lutionary responses in protein expression were positively correlated in grayling subpopulations adapted to different thermal regimes. This finding suggests that the plastic protein expression response to a temperature treatment is in the same direction as the between-population protein expression profile, taken as a proxy of evolution-ary response to different thermal environments. This experimental observation is in line with previous conceptual work suggesting that the adaptive plastic reaction norm of a trait should be in the same direction as would be expected if the trait would be evolv-ing under directional selection (Ghalambor et al. 2007; Fitzpatrick 2012). Therefore, our results suggest that gene expression plastic-ity among the grayling subpopulations is adaptive and driving the population closer to or to the thermal gene expression optimum. We further show that the differentially expressed genes are mostly related to muscle development and activity. This finding is in line with phenotypic differentiation in muscle-related traits reported earlier between the cold and warm subpopulations (Kavanagh et al. 2010; Thomassen et al. 2011).
Earlier studies have shown that the expression of plastic or environmentally induced traits can later be fixed or reduced due to genetic changes commonly known as genetic accommodation or assimilation (Gomez-Mestre and Buchholz 2006; Scoville and Pfrender 2010). For example, Scoville and Pfrender (2010) showed that the expression of pigmentation genes (Ddc and Ebony) in Daphnia were plastic in ancestral populations lacking predatory fish whereas in derived populations in the presence of predator fish the plasticity in expression was lost or genetically fixed. It appears that the contemporary grayling subpopulations in L. Lesjaskogsvatnet have a plastic response to temperature treatment, but whether this has evolved after the colonization event or was present in the ancestral population needs further studies. In particular, without knowledge about the level of plasticity in the ancestral population we cannot test whether plasticity has evolved as predicted by the genetic accommodation hypothesis. Conceptual work has suggested that there are more complex alternatives for plasticity to promote population persistence when facing novel environmental conditions
−0.8 −0.4 0.0 0.4 0.8−0.8
−0.4
0.0
0.4
0.8
−0.8
−0.4
0.0
0.4
0.8
−0.8 −0.4 0.0 0.4 0.8
−0.8 −0.4 0.0 0.4 0.8−0.8
−0.4
0.0
0.4
0.8
a
b
c
Cold evolutionary response
Col
d pl
astic
res
pons
e
Warm evolutionary response
War
m p
last
ic r
espo
nse
Average evolutionary response
Ave
rage
pla
stic
res
pons
e
Figure 2. The correlation between the plastic and evolutionary responses of 408 proteins as measured in fold change. The correlation between cold plastic response and cold evolutionary response is shown in (a), whereas (b) shows the correlation between warm plastic response and warm evolutionary response. c shows the correlation when plastic and evolutionary responses are averaged across the subpopulations. A detailed explanation for these comparisons is given in the material and methods.
Table 1. Observed correlation coefficients of the plastic versus evolutionary protein expression changes for cold and warm comparisons, as well as the average of both responses
Co-efficient Cold responsea Warm responseb Average responsec
Spearman 0.715*** (−0.500 to 0.500) 0.750*** (−0.480 to 0.480) 0.577** (−0.452 to 0.487)Pearson 0.844*** (−0.556 to 0.556) 0.835*** (−0.524 to 0.525) 0.806*** (−0.501 to 0.536)Kendall 0.537*** (−0.362 to 0.362) 0.573*** (−0.345 to 0.346) 0.423** (−0.324 to 0.352)
95% confidence intervals derived from the permutations are shown in the parenthesis. aCold plastic response versus cold evolutionary response.bWarm plastic response versus warm evolutionary response.cAverage plastic and evolutionary responses across treatments and subpopulations.***P < 0.001, **P < 0.01.
86 Journal of Heredity, 2016, Vol. 107, No. 1
Downloaded from https://academic.oup.com/jhered/article-abstract/107/1/82/2622829by gueston 13 April 2018
(Morris and Rogers 2013). Although a given trait can lack plastic capacity or its response is maladaptive, the response can be com-pensated in another complementary trait via plastic or genetic com-pensation (Morris and Rogers 2013). A previous study on the same grayling subpopulations showed accelerated muscle development but at the cost of delayed skeletal differentiation which may indicate that plastic or genetic compensation might play a role in the grayling study system (Kavanagh et al. 2010).
One critical aspect in our correlation comparisons is the esti-mation of evolutionary response in gene expression (Fraser 2011). There might be an unknown environmental component contribut-ing to the evolutionary response and thus this might create spurious correlations when contrasted against the plastic response. Several lines of evidence suggest that our between-population comparisons reflect true divergence in gene expression between cold and warm grayling subpopulations. Previous analyses using a general linear mixed model (GLMM) approach on the same data have shown that there is a significant effect of thermal treatment and the temperature origin of the subpopulation, but no interaction between them. This suggests that there is a plastic component associated with the tem-perature treatment and also an evolutionary component associated with the subpopulations adapted to different thermal conditions (Papakostas et al. 2014). Averaging the plastic and evolutionary responses across subpopulations also revealed a high correlation suggesting that environmental effects have little effect on the evo-lutionary component. Second, comparisons in a QST–FST framework
on the same subpopulations revealed higher differentiation in gene expression (QST ~ 0.08) than for neutral microsatellite markers (FST ~ 0.02) (Papakostas et al. 2014). QST higher than expected under neu-trality (FST) is in line with the interpretation that the gene expression differentiation between cold and warm subpopulations is adaptive. Finally, maternal effects might confound some of the expression pat-terns due to the fact the eggs were sampled directly from the wild. However, maternal effects have been found to play a minor role in phenotypic divergence between grayling populations (Haugen and Vøllestad 2001).
Numerous studies have investigated the importance of gene expression in relation to ecologically important variables such as temperature and salinity but have not explicitly addressed the cor-relation between plastic and evolutionary responses (Alvarez et al. 2015). A similar correlation between plastic and evolutionary responses as reported here has been observed to best of our knowl-edge only in sailfin mollies (Poecilia latipinna) (Fraser et al. 2014). In this species, some genotypes exhibit plastic behavior by switching between courtship and sneaking mating tactics while other geno-types are fixed to one of the behavioral types. Gene expression levels were highly correlated between the plastic and fixed behavioral types indicating that plasticity can facilitate adaptation in the sailfin mol-lies behavioral types. The authors further concluded that if plastic and evolutionary responses share a common molecular mechanism, it is likely to result in adaptive plasticity. This conclusion has also been supported by simulation studies on gene regulatory networks
warm_evol cold_evol
interaction
4035
0
0
0
0
0
0
cold_plastic warm_plastic
interaction
38317
0
4
0
3
0
1
Figure 3. Venn diagrams showing the number of significantly differentially expressed proteins in plastic and evolutionary comparisons. Each number refers to the number of significant tests and the number outside the figure to the nonsignificant tests.
Table 2. Significant (FDR q-value < 0.05) GO categories based on the differentially expressed proteins for plastic and evolutionary re-sponses
GO term Description Evolutionary responsea Plastic responseb
GO:0006937 Regulation of muscle contraction 0.0482 (3) 0.0147 (3)GO:0006941 Striated muscle contraction NS 0.0118 (3)GO:0030049 Muscle filament sliding 0.0461 (3) 0.0078 (3)GO:0033275 Actin-myosin filament sliding 0.0384 (3) 0.0059 (3)GO:0070252 Actin-mediated cell contraction 0.0329 (3) 0.0047 (3)GO:0090257 Regulation of muscle system process 0.0288 (3) 0.0039 (3)GO:0030048 Actin filament-based movement 0.0382 (3) 0.0054 (3)GO:0003009 Skeletal muscle contraction 0.0471 (2) 0.0173 (2)GO:0044057 Regulation of system process NS 0.0115 (3)GO:0043462 Regulation of ATPase activity NS 0.0388 (2)GO:0006936 Muscle contraction 0.0403 (4) 0.0235 (3)GO:0003012 Muscle system process 0.0263 (4) 0.0260 (3)GO:0030029 Actin filament-based process NA 0.0386 (3)
The number in the parenthesis indicates the number of genes associated with a given category.a,bFDR corrected q-value (<0.05).
Journal of Heredity, 2016, Vol. 107, No. 1 87
Downloaded from https://academic.oup.com/jhered/article-abstract/107/1/82/2622829by gueston 13 April 2018
(Espinosa-Soto et al. 2011; Fierst 2011). The fact that we observed a similar pattern in plastic and evolutionary responses, as did Fraser et al. (2014), might suggest that plastic and evolutionary responses share a common molecular mechanism in the grayling subpopula-tions but this needs to be verified by further studies on the regulatory mechanisms of gene expression.
The functions of the differentially expressed proteins are in con-cordance with previous studies on phenotypic differentiation in the same subpopulations. There is considerable divergence in ontoge-netic traits between cold and warm subpopulations. Cold adapted subpopulations had higher growth rates, muscle mass development, and yolk conversion efficiency than warm adapted subpopula-tions (Kavanagh et al. 2010; Thomassen et al. 2011). The similarity between the observed differentiation in phenotypic traits and pro-teome is in line with the argument that proteomic studies may miti-gate some of the challenges of identifying adaptive genetic variation (Diz et al. 2012). The proteome is closer to the phenotype where selection is operating and thus possibly able to capture genetic vari-ation underlying adaptive divergence. Verifying this result would however require information on different levels of biological organi-zation. Several studies have focused on transcription level changes using temperature treatments and have documented a large number genes associated with thermal change and stress (Quinn et al. 2011; Morris et al. 2014; Narum and Campbell 2015). These differentially expressed genes represent a diverse set of molecular pathways, but quite often highlight the importance of heat shock genes (Quinn et al. 2011; Schoville et al. 2012; Silvestre et al. 2012; Narum et al. 2013; Morris et al. 2014; Narum and Campbell 2015). Heat shock proteins are crucial for response to several types of environmental stressors such as UV-light or elevated temperatures (Kregel 2002). Two heat shock proteins (transcription factors 1 and 2, hsf1 and hsf2) were detected in our protein assay but they were not identi-fied as differentially expressed. However, a previous study identi-fied heat shock transcription factors 1 and 2 as upstream regulators of overall gene expression changes in this system (Papakostas et al. 2014). Other studies have also raised the possibility that only a small difference in gene expression might be biologically significant. For example, 20% difference in KITLG locus expression measured in mRNA level explains human hair color variation (Guenther et al. 2014; Hoekstra 2014). Further, other factors such as pleiotropic interactions, rather than fold-change per se may also be important (Papakostas et al. 2014) in gene expression adaptation. Overall, this may indicate that the link between statistical and biological sig-nificance in gene expression is not always straightforward and also highlights some of the difficulties in identifying the adaptive compo-nent of gene expression variation (Fraser 2011).
One of the key questions in relation to anthropogenic environ-mental change is how species and populations can cope with higher temperature and the associated adverse effects on other environ-mental variables (Franks and Hoffmann 2012). Populations with small effective population size living in isolated habitats may have an increased extinction risk without the necessary genetic varia-tion to cope with environmental perturbations (Willi et al. 2006). L. Lesjaskogsvatnet grayling subpopulations harbor relatively little neutral genetic variation but nevertheless show adaptive potential to different thermal conditions in a very short timescale (Kavanagh et al. 2010; Junge et al. 2011). Our study indicates that plasticity in gene expression is adaptive and thus may facilitate adaptation to dif-ferent thermal conditions. Plasticity in gene expression might there-fore be an important component for small and genetically depleted populations to cope with rapid anthropogenic change.
Supplementary Material
Supplementary material can be found at http://www.jhered.oxford-journals.org/
Funding
Academy of Finland (272836, 258048, 137710, and 136464) and the Research Council of Norway (177728).
AcknowledgmentsWe thank T. Haugen, G. Thomassen and N. Barson for fieldwork assistance and sampling the grayling embryos, Turku Biocenter Proteomics facility, L. Peil, R. Moulder and M. Ning for help with the proteomics experiment and the Finnish Centre for Scientific Computing for providing computational resources. We would also like to thank 3 anonymous reviewers for their very constructive comments on earlier versions of this manuscript.
Data Availability
Data deposited at Dryad: http://dx.doi.org/doi:10.5061/dryad.7bp03
ReferencesAlvarez M, Schrey AW, Richards CL. 2015. Ten years of transcriptomics in
wild populations: what have we learned about their ecology and evolu-tion? Mol Ecol. 24:710–725.
Baker CS. 2013. Journal of heredity adopts joint data archiving policy. J Hered. 104:1.
Chevin LM, Lande R, Mace GM. 2010. Adaptation, plasticity, and extinc-tion in a changing environment: towards a predictive theory. PLoS Biol. 8:e1000357.
Chezik KA, Lester NP, Venturelli PA. 2014. Fish growth and degree-days I: selecting a base temperature for a within-population study. Can J Fish Aquat Sci. 55:47–55.
Diz AP, Martínez-Fernández M, Rolán-Alvarez E. 2012. Proteomics in evo-lutionary ecology: linking the genotype with the phenotype. Mol Ecol. 21:1060–1080.
Draghi JA, Whitlock MC. 2012. Phenotypic plasticity facilitates mutational variance, genetic variance, and evolvability along the major axis of envi-ronmental variation. Evolution. 66:2891–2902.
Eden E, Navon R, Steinfeld I, Lipson D, Yakhini Z. 2009. GOrilla: a tool for discovery and visualization of enriched GO terms in ranked gene lists. BMC Bioinform. 10:48.
Espinosa-Soto C, Martin OC, Wagner A. 2011. Phenotypic plasticity can facil-itate adaptive evolution in gene regulatory circuits. BMC Evol Biol. 11:5.
Fierst JL. 2011. A history of phenotypic plasticity accelerates adaptation to a new environment. J Evol Biol. 24:1992–2001.
Fitzpatrick BM. 2012. Underappreciated consequences of phenotypic plastic-ity for ecological speciation. Int J Ecol. 2012:Article ID 256017.
Forsman A. 2014. Rethinking phenotypic plasticity and its consequences for individuals, populations and species. Heredity. Advance Access published October 8, 2014, doi:10.1038/hdy.2014.92
Franks SJ, Hoffmann AA. 2012. Genetics of climate change adaptation. Annu Rev Genet. 46:185–208.
Fraser BA, Janowitz I, Thairu M, Travis J, Hughes KA. 2014. Phenotypic and genomic plasticity of alternative male reproductive tactics in sailfin mol-lies. Proc R Soc B Biol Sci. 281:20132310.
Fraser HB. 2011. Prospects & overviews genome-wide approaches to the study of adaptive gene expression evolution. BioEssays. 33:469–477.
Fusco G, Minelli A. 2010. Phenotypic plasticity in development and evolu-tion: facts and concepts. Introduction. Philos Trans R Soc Lond B Biol Sci. 365:547–556.
Ghalambor CK, McKay JK, Carroll SP, Reznick DN. 2007. Adaptive versus non-adaptive phenotypic plasticity and the potential for contemporary adaptation in new environments. Funct. Ecol. 21:394–407.
88 Journal of Heredity, 2016, Vol. 107, No. 1
Downloaded from https://academic.oup.com/jhered/article-abstract/107/1/82/2622829by gueston 13 April 2018
Gienapp P, Teplitsky C, Alho JS, Mills JA, Merilä J. 2008. Climate change and evolution: disentangling environmental and genetic responses. Mol Ecol. 17:167–178.
Gomez-Mestre I, Buchholz DR. 2006. Developmental plasticity mirrors differ-ences among taxa in spadefoot toads linking plasticity and diversity. Proc Natl Acad Sci USA. 103:19021–19026.
Gomez-Mestre I, Jovani R. 2013. A heuristic model on the role of plasticity in adaptive evolution: plasticity increases adaptation, population viability and genetic variation. Proc Biol Sci. 280:20131869.
Gregersen F, Haugen TO, Vøllestad LA 2008. Contemporary egg size diver-gence among sympatric grayling demes with common ancestors. Ecol. Freshw. Fish 17:110–118.
Guenther CA, Tasic B, Luo L, Bedell MA, Kingsley DM. 2014. A molecular basis for classic blond hair color in Europeans. Nat Genet. 46:748–752.
Haugen TO, Vøllestad LA. 2001. A century of life-history evolution in gray-ling. Genetica. 112–113:475–491.
Hodgins-Davis A, Townsend JP. 2009. Evolving gene expression: from G to E to GxE. Trends Ecol Evol. 24:649–658.
Hoekstra H. 2014. The secret of a natural blond. Nat Genet. 46:660–661.Junge, C. 2011. Spatio-temporal population structuring in complex environ-
ments: insights from the European grayling (Thymallus thymallus). PhD-thesis, Faculty of Mathematics and Natural Sciences. University of Oslo, Oslo.
Junge C, Vøllestad LA, Barson NJ, Haugen TO, Otero J, Sætre G-P, Primmer CR. 2011. Strong gene flow and lack of stable population structure in the face of rapid adaptation to local temperature in a spring-spawning sal-monid, the European grayling (Thymallus thymallus). Heredity. 106:460–471.
Kavanagh KD, Haugen TO, Gregersen F, Jernvall J, Vøllestad LA. 2010. Con-temporary temperature-driven divergence in a Nordic freshwater fish under conditions commonly thought to hinder adaptation. BMC Evol Biol. 10:350.
Khan Z, Ford MJ, Cusanovich DA, Mitrano A, Pritchard JK, Gilad Y. 2013. Primate transcript and protein expression levels evolve under compensa-tory selection pressures. Science. 342:1100–1104.
Kregel KC. 2002. Heat shock proteins: modifying factors in physiological stress responses and acquired thermotolerance. J Appl Physiol (1985). 92:2177–2186.
Li GW, Burkhardt D, Gross C, Weissman JS. 2014. Quantifying absolute protein synthesis rates reveals principles underlying allocation of cellular resources. Cell. 157:624–635.
Merilä J. 2012. Evolution in response to climate change: in pursuit of the miss-ing evidence. Bioessays. 34:811–818.
Merilä J, Hendry AP. 2014. Climate change, adaptation, and phenotypic plas-ticity: the problem and the evidence. Evol Appl. 7:1–14.
Morris MR, Richard R, Leder EH, Barrett RD, Aubin-Horth N, Rogers SM. 2014. Gene expression plasticity evolves in response to colonization of freshwater lakes in threespine stickleback. Mol Ecol. 23:3226–3240.
Morris MRJ, Rogers SM 2013. Overcoming maladaptive plasticity through plastic compensation. Curr. Zool. 59:526–536.
Murren CJ, Auld JR, Callahan H, Ghalambor CK, Handelsman CA, Heskel MA, et al. 2015. Constraints on the evolution of phenotypic plasticity: limits and costs of phenotype and plasticity. Heredity. Advance Access published February 18, 2015, doi:10.1038/hdy.2015.8
Narum SR, Campbell NR. 2015. Transcriptomic response to heat stress among ecologically divergent populations of redband trout. BMC Genom-ics. 16:103.
Narum SR, Campbell NR, Meyer KA, Miller MR, Hardy RW. 2013. Thermal adaptation and acclimation of ectotherms from differing aquatic climates. Mol Ecol. 22:3090–3097.
Papakostas S, Vøllestad LA, Bruneaux M, Aykanat T, Vanoverbeke J, Ning M, Primmer CR, Erica LH. 2014. Gene pleiotropy constrains gene expression changes in fish adapted to different thermal conditions. Nat Commun. 5:4071.
Quinn NL, McGowan CR, Cooper GA, Koop BF, Davidson WS. 2011. Identi-fication of genes associated with heat tolerance in Arctic charr exposed to acute thermal stress. Physiol Genomics. 43:685–696.
Romero IG, Ruvinsky I, Gilad Y. 2012. Comparative studies of gene expres-sion and the evolution of gene regulation. Nat Rev Genet. 13:505–516.
Schoville SD, Barreto FS, Moy GW, Wolff A, Burton RS. 2012. Investigat-ing the molecular basis of local adaptation to thermal stress: population differences in gene expression across the transcriptome of the copepod Tigriopus californicus. BMC Evol Biol. 12:170.
Scoville AG, Pfrender ME. 2010. Phenotypic plasticity facilitates recurrent rapid adaptation to introduced predators. Proc Natl Acad Sci USA. 107:4260–4263.
Silvestre F, Gillardin V, Dorts J. 2012. Proteomics to assess the role of phe-notypic plasticity in aquatic organisms exposed to pollution and global warming. Integr Comp Biol. 52:681–694.
Skelly DK, Freidenburg KL. 2010. Evolutionary responses to climate change. In Evolutionary responses to climate change. In: Encyclopedia of life sci-ences (ELS). Chichester (UK): John Wiley & Sons. pp. 1–7.
Smyth GK. 2004. Linear models and empirical bayes methods for assessing dif-ferential expression in microarray experiments. Stat Appl Genet Mol Biol. 3.
Taverner T, Karpievitch YV, Polpitiya AD, Brown JN, Dabney AR, Anderson GA, Smith RD. 2012. DanteR: an extensible R-based tool for quantitative analysis of -omics data. Bioinformatics. 28:2404–2406.
Thomassen G, Barson NJ, Haugen TO, Vøllestad LA. 2011. Contemporary divergence in early life history in grayling (Thymallus thymallus). BMC Evol Biol. 11:360.
Vizcaíno JA, Côté RG, Csordas A, Dianes JA, Fabregat A, Foster JM, Griss J, Alpi E, Birim M, Contell J, et al. 2013. The Proteomics Identifications (PRIDE) database and associated tools: status in 2013. Nucleic Acids Res. 41:1063–1069.
Willi Y, Van Buskirk J, Hoffmann AA. 2006. Limits to the adaptive potential of small populations. Annu Rev Ecol Evol Syst. 37:433–458.
Wund MA. 2012. Assessing the impacts of phenotypic plasticity on evolution. Integr Comp Biol. 52:5–15.
Journal of Heredity, 2016, Vol. 107, No. 1 89
Downloaded from https://academic.oup.com/jhered/article-abstract/107/1/82/2622829by gueston 13 April 2018
