• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Mol Ecol. Author manuscript; available in PMC Jul 1, 2010.
Published in final edited form as:
PMCID: PMC2705469

The Effect of Genetic and Environmental Variation on Metabolic Gene Expression


What is the relationship between genetic or environmental variation and the variation in mRNA expression? To address this, microarrays were used to examine the effect of genetic and environmental variation on cardiac mRNA expression for metabolic genes in three groups of Fundulus heteroclitus: (1) individuals sampled in the field (field), (2) field individuals acclimated for six months to laboratory conditions (acclimated) or (3) individuals bred for ten successive generations in a laboratory environment (G10). The G10 individuals have significantly less genetic variation than individuals obtained in the field and had a significantly lower variation in mRNA expression across all genes in comparison to the other two groups (p ≤ 0.001). When examining the gene specific variation, twenty-two genes had variation in expression that was significantly different among groups with lower variation in G10 individuals than in acclimated individuals. Additionally, there were fewer genes with significant differences in expression among G10 individuals versus either acclimated or field individuals: 66 genes have statistically different levels of expression versus 107 or 97 for Acclimated or Field groups. Based on the permutation of the data, these differences in the number of genes with significant differences among individuals within a group are unlikely to occur by chance (p < 0.01). Surprisingly, variation in mRNA expression in field individuals is lower than in acclimated individuals. Relative to the variation among individual within a group, few genes have significant differences in expression among groups (seven, 2.3%) and none of these are different between acclimated and field individuals. The results support the concept that genetic variation affects variation in mRNA expression and also suggests that temporal environmental variation associated with estuarine environments does not increase the variation among individuals or add to the differences among groups.

Keywords: Fundulus heteroclitus, microarray, gene expression evolutionary genomics, genetic variation


Variation in mRNA expression should be a function of genetic and environmental variation. The quantification of variance due to the additive effects of genes is important as natural selection acts on this genetic component of variance (Falconer, Mackay, 1996). In outbred populations of the teleost fish Fundulus heteroclitus there is substantial variation in mRNA expression within and between groups of individuals (Crawford, Oleksiak, 2007; Oleksiak et al., 2002; Oleksiak et al., 2005; Whitehead, Crawford, 2006b). Although much of this variation appears to be due to random neutral evolutionary processes, a significant fraction of variation in expression is best explained by evolution by natural selection (Whitehead, Crawford, 2006a). These evolutionarily adaptive patterns suggest that the variation in mRNA expression is biologically important because any trait evolving by natural selection must affect fitness (Whitehead, Crawford, 2006a). Investigation into the functional importance of natural variation in mRNA expression revealed that variation in mRNA expression explains differential use of metabolic substrates among groups of individuals providing additional evidence for the biological significance for otherwise seemingly chaotic patterns of expression (Oleksiak et al., 2005). While these studies assume that variation in mRNA expression among individuals is genetically based due to the rearing of fish in a common environment, no studies have been performed to test this hypothesis.

To provide evidence supporting this hypothesis, we examine the variation in metabolic mRNA expression among groups that have different levels of genetic and environmental variation. If mRNA expression is heritable and genotypic effects dominate the variation in expression then more genetically similar individuals should have less variation and fewer significant differences in mRNA expression than among unrelated individuals. Notice, we are not asking whether mRNA expression differs in different environments, or between genotypes. Instead, we are asking whether the variation in mRNA expression is a function of environmental and genetic variation. We demonstrate that variation in mRNA expression is significantly lower among closely related individuals compared to outbred fish raised in similar environments. Surprisingly, increased environmental heterogeneity in unacclimated individuals sampled in the field did not increase the variation in mRNA expression among the outbred samples. These observations suggest that the normal environmental variation associated with tidal fluxes in estuarine environments does not substantially add to the variation in mRNA expression in F. heteroclitus.

Methods and Materials


Fundulus heteroclitus were caught from wild populations in Beaufort, North Carolina, USA (34°43’34”N, 76°40’62”S) by allowing fish to swim into seine nets on an outgoing tide. Seine nets were used to catch a few fish (10-20) to minimize handling time and stress. Upon capture in the field five males and five females were placed in a bucket, weighed, sexed, killed by cervical dislocation, their hearts removed and stored in RNA later (Ambion, Inc.). From the same location and tide, more fish were capture and transported to the University of Miami where they were acclimated to 20°C and 15ppt in laboratory aquaria for approximately 6 months. These fish were compared to fish trapped at the same location, and raised at 20°C and 15ppt and allowed to interbreed for ten successive generations (G10) at the Aquatic Biotechnology and Environmental Laboratory at the University of Georgia. For the purposes of this experiment, 5 males and 5 females from each of the following groups were used: field caught (field), field caught then acclimated for 6 months at the University of Miami (acclimated) and fish raised for ten generations (G10).

Genetic Diversity

The G10 fish were started from a group of approximately 16 adults and were allowed to freely interbreed for 10 generations. In order to characterize levels of genetic diversity and pair wise relatedness within and between the G10 and field caught (field) individuals, we genotyped 49 G10 and 109 field individuals (including individuals used in the microarray experiments) at 10 microsatellite loci for F. heteroclitus (Adams et al., 2005).

DNA was extracted from dried fin clips. The tissue was placed in 300 μL lysis buffer (75 mM NaCl, 25mM EDTA, 1%SDS) and incubated with 0.1 mg Proteinase K at 55°C for 2 hours. Proteins were precipitated by adding a half volume of 7.5 M ammonium acetate and centrifugation for 10 minutes. DNA was precipitated from the supernatant by adding 0.7 volumes of isopropanol and centrifugation for 15 minutes. The DNA pellet was washed with 70% ethanol then allowed to air dry for 30 minutes followed by re-suspension in 50 μL 10 mM Tris-HCl pH 8.5.

Microsatellite loci were amplified in four fluorescently labeled multiplex primer groups containing the following final concentrations: A – (0.15 μM CA-1, 0.07 μM CA-A3, 0.20 μM C-1), B – (0.10 μM ATG-18, 0.10 μM ATG-B4), C – (0.07 μM ATG-25, 0.07 μM ATG-6), D – (0.07 μM ATG-B128, 0.15 μM CA-21). Locus ATG-20 was amplified alone at a final concentration of 0.5 μM. The 10 μL reaction contained 2.5 mM MgCl2, 1X PCR buffer, 0.2 mM dNTPs, 0.4 units Taq DNA polymerase, 70 ng DNA, and one of the five primer combinations (see above for concentrations). The PCR thermal cycling profile consisted of 94°C for 2 minutes, followed by 31 cycles of 94°C for 15 seconds, 55°C (A, C, ATG-20) or 55°C (B and D) for 15 seconds, and 72°C for 30 seconds, ending with a 5 minute extension step at 72°C. Following PCR amplification, the products from A, C, and D were co-loaded, as were ATG-20 and B, before being electrophoreses on an ABI 3730XL Genetic Analyzer (Applied Biosystems). GENEMAPPER v4.0 (Applied Biosystems) was used to score the genotypes. All genotypes were checked by two individuals.

RNA isolation, labeling and hybridization

Total RNA was isolated from using 4.5M guanidinium thiocyanate, 2% N-lauroylsarcosine, 50mM EDTA, 25mM Tris-HCl, 0.1M β-Mercaptoethanol and .2% Antifoam A. The extracted RNA was further purified using a Qiagen RNeasy Mini kit in accordance with the manufacturer’s protocols. The quantity and quality of the RNA was determined using a spectrophotometer (Nanodrop, ND-1000 V3.2.1) and by electrophoresis with the use of a bioanalyzer (Agilent 2100). RNA was then converted into amino allyl labeled RNA (aRNA) using the Ambion Amino Allyl MessageAmp II aRNA Amplification kit. This method converts poly-A RNA into cDNA with a T7 RNA polymerase binding site and T7 is used to synthesize many new strands of RNA (in vitro transcription) (Eberwine, 1996). During this in vitro transcription of aRNA, an amino allyl UTP (aaUTP) is incorporated into the elongating strand. The incorporations of aaUTPs allows for the coupling of Cy3 or Cy5 dyes (GE biosciences) onto aRNA for microarray hybridization.

Dye labeled aRNA aliquots for each hybridization (30 pmol each of Cy3 and Cy5) were vacuum dried together and resuspended in 15μl hybridization buffer (final concentration of each labeled sample = 2 pmol/μl). Hybridization buffer consisted of 5X SSPE, 1% SDS, 50% formamide, 1mg/ml polyA, 1mg/ml sheared herring sperm carrier DNA, and 1mg/ml BSA. Slides were washed in sodium borohydride solution in order to reduce autofluorescence. Following rinsing, slides were boiled for 2 minutes and spin-dried in a centrifuge at 800 rpm for 3 minutes. Samples (15μl) were heated to 90°C for 2 minutes, quick cooled to 42°C, applied to the slide (hybridization zone area was 350mm2), and covered with a cover slip. Slides were placed in an airtight chamber humidified with paper soaked in 5X SSPE and incubated 24-48 hours at 42°C.


The amount of gene specific mRNA expression was measured using microarrays with four spatially separated replicates per gene on each array. Microarrays were printed using 384 Fundulus heteroclitus cDNAs that included 329 cDNAs that encode essential proteins for cellular metabolism (Table 1, (Paschall et al., 2004)). Average lengths of cDNAs were 1.5Kb with a majority including the N-terminal methionine. Table one provides a summary of ESTs used for printing where the most meaningful GO term is used to categorized the annotation (Paschall et al., 2004). These cDNAs were amplified with amine-linked primers and printed on 3-D Link Activated slides (Surmodics Inc., Eden Prairie, MN) at the University of Miami core microarray facility.

Table 1
384 Microarray Pathways

Dye coupled aRNA from thirty individuals from three groups were hybridized to slides using two loops (Kerr, Churchill, 2001; Oleksiak et al., 2002); one for males and one for females (Supplemental Figure 1). The female loop is as follows; G1→ A1→ F1→ G2→ A2→ F3→ G7→ A5→ F6→ G4→ A10→ F15→ G5→ A12→ F18→ G1. The male loop is as follows; G12→ A6→ F2→ G14→ A7→ F4→ G15→ A8→ F5→ G6→ A9→ F7→ G10→ A11→ F8→ G12. Each arrow represents a single hybridization between an individual labeled with Cy3 at the base of the arrow and an individual labeled with Cy5 at the head of the arrow. The first letter represents either a G10 (G), acclimated (A) or field caught (F) individual. The number represents each individual used in the study. Three males (A6, F12 and G10) were removed from the analysis due to poor hybridization or too strong of signals (i.e, where too many different genes had signals that saturated the PMT = 65,535). Therefore, a total of 12 males (4 per group) and 15 females (5 per group) were used for the analysis.

The microarray slides were scanned using ScanArray Express. The raw TIFF-image data was quantified using Imagene (v5). If a gene had a fluorescent signal that was too low or too high it was eliminated from the analysis for all individuals. Fluorescent signals were consider too low if the average across all samples were within 2 standard deviations of the average signal from the Ctenophore negative controls. Fluorescent signals were considered too high if the average signal plus two standard deviations exceeded 55,000. This procedure is based on empirical analyses of data and removes fluorescent signals that saturate the photomultiplier tube (maximum signal is 65,565). Using these criteria 100 genes were eliminated from all individuals leaving 284 genes.


Microsatellite loci were tested for deviation from Hardy-Weinberg equilibrium and for linkage disequilibrium using GENEPOP version 3.3 (Raymond, Rousset, 1995). The number of alleles (NA), observed heterozygosity (HO), and expected heterozygosity (HE) were calculated using GENALEX 6 (Peakall, Smouse, 2006). Allelic richness (AR) for each group (G10 and field) was calculated with FSTAT version 2.9.3 (Goudet, 1995) with a sample size adjustment of n = 49 individuals (the smallest sample size). We compared average measures of genetic diversity calculated across loci between G10 and field individuals by randomizing locus specific values between groups and recalculating the difference in mean values 5,000 times to generate a random distribution of mean values. The location of the observed mean difference within this random distribution was used to determine the probability that it was significantly different from the random distribution.

Genetic similarity between individuals within groups was estimated by the relatedness coefficient R of Queller and Goodnight using RELATEDNESS 5.0 (Queller, Goodnight, 1989). The allele frequencies used to calculate relatedness coefficients came from the entire sample of G10 and field individuals. Standard errors of the estimates were obtained by jackknifing over loci (Sokal, Rohlf, 1995). We compared the average relatedness of G10 and field individuals by jackknifing over the unpaired R difference using RELATEDNESS. We also estimated genetic similarity by the proportion of shared alleles (Bowcock et al., 1994). Significance was determined by permuting individuals between the groups and recalculating the mean proportion of alleles shared between individuals 1,000 times to construct the 95% CI around the random expectation. Ninety-five percent confidence intervals were also calculated around each mean by bootstrapping values within each group 1,000 times.

Statistical analyses of the mRNA expression data were carried out using JMP genomics (SAS JMP Genomics v.7.0.2). All analyses used fluorescent measures that were log2 transformed and loess normalized. These normalized fluorescent measures showed nearly identical distributions among all individuals (Supplemental Figure 2). Standardization of data (mean signal with an average intensity equal to zero) or further normalization using ANOVA or a mixed model did not substantially affect the distribution of fluorescences nor did it affect the relative frequency of genes with significant differences in expression among individuals. Thus, the simpler of the normalizations (log2-loess) were used for parsimony and clarity of results.

For each gene, an ANOVA for the significant differences in mRNA expression among individuals within each group used the linear mixed model (Kerr, Churchill, 2001; Patterson et al., 2006; Wolfinger et al., 2001; Yu et al., 2004): yijk=μ + Ai + Dj + Ikijk; where yijk represents the fluorescence intensities on a log scale and μ is a constant. The fixed effect is Ik for kth individual (for one of the nine individuals per group) and the random effects are Dj for the jth dye of the two Cy dyes, Ai is the ith array (for one of the 27 different arrays) and εijk are random residual term. This analysis, to define difference among individuals, was applied to the nine individuals in each group. With nine individuals per group and eight replicates (four replicates per array and two dyes) there are 8 and 52 degrees of freedom. To determine if the number of genes with significantly different expression among individuals in each group was statistically meaningful, this mixed model analysis was run on all 126 possible combinations of 5 out of nine individuals for each group (4 and 30 degrees of freedom). From these 126 ANOVAs, the average number of genes and the confidence intervals that were significantly different among individuals was calculated. In these and other statistical analyses of mRNA expression a p-value of 0.01 is used (with and without Bonferonni’s correction). This arbitrary value is chosen because it provides more a conservative value than a p-value of 5% and it allows the reader to quickly judge the 1% frequency of type 1 errors.

Statistical analysis of group or sex effects used the least square means from the linear mixed model. This model, which provides a single measure of expression for each gene for each individual, is identical to the model described above except the linear model used all 27 individuals across all 3 groups (instead of the 9 in each group; resulting in 26 and 161 degrees of freedom).

Using the least square means for each individual, an ANOVA was used to determine the significant differences in mRNA expression between groups or sex. Notice that with the least square means there are no dye nor array effects (only one measure for each individual and each gene from the mixed model with dyes and arrays as random factors), thus, these factors are not included. For the ANOVA with sex as a fixed effect there were 1 and 25 degrees of freedom. For the ANOVA with group as fixed effect there were 2 and 24 degrees of freedom.

Hierarchical clustering of gene expression uses Macintosh’s version [104] of Eisen’s Cluster and Treeview [105]. Volcano plots are from SAS-JMP and use a t-test for comparisons between any two individuals or groups.


The genetic variation and expression of mRNA was measured in three groups: Fundulus caught in the field and immediately sacrificed (field), Fundulus caught in the field and acclimated for six months to common laboratory conditions (acclimated) and Fundulus bred for ten successive generations (G10) in a common laboratory environment. All fish originated from the same location in Beaufort, North Carolina. Most of the mRNAs quantified encode metabolic genes (Table 1).

Genetic Diversity and Relatedness

The microsatellite loci in the field and acclimated samples (outbred groups) were highly polymorphic and in Hardy-Weinberg (p = 0.70) and genotypic linkage equilibrium (p > 0.20 in all cases). The G10 individuals had significantly lower mean genetic diversity values than the field caught individuals for three of our four measures (AR p = 0.001; HO p = 0.001; HE p = 0.001; FIS p = 0.11) (Table 2). The G10 fish were also in Hardy Weinberg equilibrium (p = 0.14), but 13 of the 45 pair-wise comparisons between loci had significant linkage disequilibrium after a Bonferroni’s correction for multiple comparisons. Relatedness among individuals was significantly higher than zero and close to the level of full siblings for the G10 individuals (R = 0.43; 95% CI 0.33 – 0.53). This contrasts with the relatedness of field individual that were not significantly different from zero (R = 0.03; 95% CI -0.004 – 0.066). The proportion of shared alleles between individuals revealed a similar pattern; individuals within the G10 group shared more alleles (mean = 0.545; 95% CI 0.536 – 0.553) than individuals within the field group (mean = 0.315; 95% CI 0.312 - 0.317).

Table 2
Genetic diversity values for laboratory bred (G10) individuals (n = 49) and field caught individuals (F) (n = 109)

We also compared the proportion of shared alleles between the individuals that were used only in the microarray experiment (Fig. 1). The G10 individuals used in the microarray experiment shared a significantly higher proportion of their alleles (0.46, 95% CI 0.42 – 0.51) than either the acclimated (0.36, 95% CI 0.34 – 0.38) or the field (0.30, 95% CI 0.28 – 0.33). Although the acclimated group had a slightly higher level of shared alleles than the field group, both fell within the 95% CI expected for random pairings of individuals (Fig 1).

Figure 1
Average Proportion of Shared Alleles within Groups

Significant Difference between sexes in mRNA expression

Gene expression was measured in a total of 12 males and 15 females using two separate hybrization loops. To test for difference between the sexes or the confounding affect of loops, we applied an ANOVA using the least square means from the linear model with individual as fixed effect and array and dyes as random affect. Only three genes have significant differences (p < 0.01) in mRNA expression between the two sexes. None are significant with Bonferroni’s correction for multiple comparisons (p < 3.5 * 10-5). With so few differences in gene expression between the sexes, we analyzed males and females together.

Significant Differences within Groups in mRNA expression

Among all individuals (ignoring group effects) 281 of the 284 genes have significant differences in expression (Fig. 2A, p < 0.01). Within each group the number genes with significant differences in mRNA expression is a measure of variation and thus there is an expectation that the number of significant genes will be related to genetic and environmental variation (Fig.2, Supplemental Table 1). A mixed-model ANOVA was used to test if there are significant differences in mRNA expression among individuals within each group (p < 0.01, Table 3). G10 individuals had approximately two-thirds the number of significantly different genes (66 or 23%, 23 significant with Bonferroni’s corrected p < 3.5 * 10-5) as did acclimated individuals (107 or 38%, 46 with Bonferroni’s corrected p < 3.5 * 10-5) even though both groups were raised in similar laboratory conditions. Surprisingly, field individuals had an intermediate number of significant genes (97 or 34%, 29 with Bonferroni’s corrected p < 3.5 * 10-5).

Figure 2
Hierarchical Clustering of Individual Gene Expression
Table 3
Number of Genes with Significantly Different mRNA Expression

The patterns of variation among individuals in mRNA expression are similar in all three groups (Fig. 2B). This supposition is supported by hierarchical cluster of all individual ignoring groups (Fig. 2A): although groups differ in the number of significant genes, the patterns of variation are shared among individuals from different groups. Additionally, these patterns (Fig. 2) suggest that the significant differences in mRNA expression among individuals are not due to just one or a few individuals (Fig 2B). This later supposition is supported by the permutation analysis (see below) where all possible 5 out of nine individuals have similar patterns of differential mRNA expression (see below). Examples of the magnitude and associated p-values with these differences are shown in the volcano plots (Fig. 3). Although only three of the possible 36 paired comparison in each group are shown in figure three, these differences are a representative samples and suggest that the difference among acclimated individuals tend to be larger (x-axis, log2 differences in the least squared mean) and are more significant (y-axis, negative log10(p-values).

Figure 3
Volcano Plots among Three Individuals within each Group

To test if the number of genes with significant differences in expression were meaningful, all 126 possible combinations of five out nine individuals per group were examined (Table 3). This permutation of the data allows us to calculate means, variance and test for differences between groups for the number of genes with significantly different mRNA expression. Among these combinations, the average number of genes with a significant difference in expression share the same pattern as the analysis of all nine individuals: for the number of genes with significant differences in mRNA expression acclimated > field > G10. Notice that the overall number of significant genes is less because only five individuals were examine at a time and thus there are fewer degrees of freedom. For each group, mean numbers of genes with significant differences in expression among these 126 combinations have confidence intervals that do not overlap (Table 3) and these means are statistically different (Kruskal-Wallis non-parametric test p < 0.001). Thus, there is statistical support that there are a greater number of genes with significant difference in expression in acclimated versus field or G10, and field versus G10.

Variance in mRNA expression Across Genes

Another test of how the variation in mRNA differs among groups is to examine compare the mean of the variance estimates across all genes among individuals. That is, rather than testing for the quantitative differences in mRNA expression, we tested whether the mean variation in mRNA expression across all 284 genes among individuals was different among groups. Because the variance is a function of the magnitude of the mean, the measures of expression of all 284 genes was normalized so that the average expression for each gene was equal to one: lsmeanpg/(Avgg) where the lsmeanp is the least square mean for the pth individual and the gth gene, and these measures are divided by Avgg, the average least square mean for the gth gene. For each individual the mean variance for all 284 measures of expression using these normalized values was calculated. This mean variation in expression among the nine individuals is significantly different between groups (Kruskal-Wallis test, p < 0.001) with mean variance of 0.575 (stdev = 0.346), 0.422 (stdev. = 0.256) and 0.386 (stdev = 0.132) for acclimated, field and G10 respectively. It is interesting that the standard deviation of the variance is greatest in the acclimated group, suggesting greater differences in the variation among individuals in this group.

Homogeneity of Variance

A third test of how the variation among individuals for mRNA expression differs among the three groups is to examine the similarity of the variance for each gene. We tested the similarity for each gene by applying the Barlett’s test for homogeneity of variance among groups using the least squared means for each individual (Table 4). Of the 284 measures of mRNA expression used in this experiment, 22 (7%), had an unequal variance among groups (p < 0.01). Among these 22 genes with significant differences in the individual variation in mRNA expression, 17 (77%) of genes had greater expression variation in acclimated individuals than G10 individuals. This bias of larger variance in mRNA expression in the acclimated group versus G10 is unlikely to occur by chance (χ2 p < 0.01). For the acclimated versus field, or field versus G10 there are the same (50%) or nearly the same (59%) number of genes with greater variance in mRNA expression.

Table 4
Homogeneity of Variance

Significant differences between groups in mRNA expression

We expected a difference in the variance in mRNA expression between groups, but did not necessarily expect a difference in the mean of mRNA expression. Using a p-value of 0.01, 7 genes (2.5%) have a significant difference in mRNA expression among groups (Fig. 4). None of these are significant with a Bonferroni’s corrected p-value of 3.5 * 10-5. To determine which groups differ in their gene specific mRNA expression, t-tests were applied between the 3 pairs of comparisons (Acc vs. Fld, Acc vs. G10 and Fld vs. G10). There are no significant differences between the acclimated and the field groups (Fig. 4A) and there are ten significant differences in mRNA expression between either the acclimated or field versus G10 group (Fig. 4B).

Figure 4
Differences Among Groups


The genetic basis for the variation in mRNA expression among natural populations, including F. heteroclitus, is not well understood. In other species, our understanding of the genetics of mRNA expression has relied on the study of inbred strains (Gibson, Weir, 2005; Schadt et al., 2003; Wayne et al., 2004) or cell culture (Monks et al., 2004). Using these systems, the variation in mRNA expression measured by microarrays appears to be genetically based: it differs between inbred lines, is associated with QTLs and has narrow sense heritability (h2) greater than 30% (Cheung et al., 2003; Gibson, Weir, 2005; Rockman, Kruglyak, 2006; Sharma et al., 2005; Tan et al., 2005). Heritability of mRNA expression has been measured in a variety of organisms. For example, in ten lines of Drosophila, 663 of 7886 measured genes (8%) had significant genetic variation with a medium h2 = 0.47 (quartile range 0.39-0.60) (Wayne et al., 2004). Among 112 Sacchoromyces cerevisiae segregants, 3,546 out of 5,727 measured genes (62%) had a h2 > 0.69 (Brem, Kruglyak, 2005). Using lymphoblast human cell lines, among 15 families, 762 out of 2,430 (31%) of differentially expressed genes had a significant h2 with median of 0.34 (Monks et al., 2004; Williams et al., 2007).

These studies on inbred lines or cell culture are informative. However, they do not provide data on the genetics of the variation in mRNA expression in outbred species. For humans, twin-studies (Sharma et al., 2005; Tan et al., 2005) and replicate measures of the same individuals over time (Cobb et al., 2005; Eady et al., 2005; Radich et al., 2004; Whitney et al., 2003) suggest a strong genetic component to the natural variation in mRNA expression. For natural populations of Fundulus heteroclitus, it is unclear if differences within and among populations (Crawford, Oleksiak, 2007; Oleksiak et al., 2002; Oleksiak et al., 2005; Whitehead, Crawford, 2005; Whitehead, Crawford, 2006a; Whitehead, Crawford, 2006b) are a function of genetic variation or other less evolutionarily important parameters. The data presented here supports the hypothesis that much of the variation in mRNA expression is a function of genetic variation.

The genetic variation based on microsatellite markers in F. heteroclitus from a single North Carolina population is greater in the outbred groups (acclimated and field) than in the inbred G10 individuals (Fig. 1). Among G10 individuals they have half the allelic richness and 75% of the heterozygosity of the outbred group. Additionally, the relatedness among G10 individuals is nearly equal to full sibs (R = 0.43) but among outbred individuals the relatedness is not different from zero. The reduced genetic variation is expected in the G10 individuals because they originated from fewer than 16 individuals and were inbred for ten generations; whereas the field caught individuals have effective population sizes that exceed 105 (Adams et al., 2006). The only measure of genetic variation that is not different for G10 is FIS where FIS is the fixation index relative to individuals within a subpopulation or group. The lack of a difference in FIS is reasonable because each generation of siblings of the G10 group was allowed to breed randomly which allowed for the re-establishment of Hardy-Weinburg equilibrium (Table 1).

Among G10 individuals there is also a lower variation in mRNA expression relative to acclimated individuals: fewer genes have significant differences in mRNA expression among individuals (Table 3, Fig. 2 and and3),3), and the mean variance across all genes is significantly less. Additionally when examining the gene specific variation in mRNA expression among individual within groups, 22 genes have significant differences in the variation among groups and for 77% of these genes the variations in mRNA expression are lower in G10 than in acclimated individuals. These differences are found among individuals raised or acclimated to a laboratory environment with constant food, salinity, temperature, oxygen and lack of predators. These data support the supposition that outbred acclimated individuals have greater variation in mRNA expression than the inbred G10 individuals even though both groups share common stable environment.

Among outbred individuals (acclimated and field) there are also differences in the variation in mRNA expression: acclimated individuals had more genes with significant differences in mRNA expression (107 vs. 97) and greater variation in mRNA expression across all 284 genes. However, when comparing the variance among individual per gene, there is not a difference in the magnitude of the gene specific variation in the acclimated and field groups (Table 3). These data indicate that acclimated individuals appear to have greater or nearly equal variation as field individuals. Thus, these data support a surprising conclusion: the environmental variation in the field (tidal changes, spatial and temporal changes in salinity, food availability, oxygen, etc. (Marshall, 2003; Marshall et al., 2005)) does not have a major affect on the variation in gene expression. Notice, we are not addressing whether changes in the environment alters mRNA expression. Instead, our data suggest that variation in environment does not have a substantial effect on the variation in mRNA expression.

Among the three groups (acclimated, field and G10) there are few differences in expression: seven genes have significant difference in mRNA expression with a critical p-value of 1%, none with Bonferroni’s correct p-value. This supposition of few differences among groups is supported by patterns of mRNA expression (Fig 2A): individuals in the three groups do not cluster together. Instead individuals in each group share common patterns of expression with individuals in other groups (Fig. 2A). With t-test between each pair of groups (which suffer from high false positive rate) the differences in expression are only significant between G10 and the two outbred groups (Fig. 4). These data suggest that the laboratory versus the field environments does not have a large affect on mRNA expression. If we propose that few significant differences between G10 and the outbred groups are important, it is difficult to imagine a common environmental factor for the field and acclimated groups that could explain the difference between these groups versus the G10 group. Alternatively, if much of the variation in mRNA expression is genetically based, as suggested by the correlation between genetic variation and the variation in expression, then one would have to speculate that the G10 individuals have different or less frequent genotypes that affect the expression of these seven mRNAs. This difference is most parsimoniously explained by due random drift due to recent bottleneck.

These patterns of variation in mRNA expression in acclimated, field and G10 individuals are consistent with the hypothesis that much of the variations among individuals are due to genetic variation. Specifically, there are fewer differences among G10 individuals that share 43% of their alleles versus completely outbred individuals (acclimated) even though both were subjected to similar laboratory condition for at least six months. Additionally, there is little support for environmental variation effect on mRNA expression: acclimated individuals have equal or more variation in mRNA expression versus field individuals that suffers the slings and arrows of environmental variation associated with estuarine environments. Added to these observations is that there is little significant variation in mRNA expression when the same individual is repetitively measure over a six-week period (i.e. mRNA expression from blood sample every 2 weeks over six-week period; (Scott et al., 2009)). That is, adult mRNA expression has little temporal variation. Together these data strongly support the hypothesis that the large inter-individual variations in gene expression measure here and elsewhere (Crawford, Oleksiak, 2007; Oleksiak et al., 2002; Oleksiak et al., 2005; Whitehead, Crawford, 2005; Whitehead, Crawford, 2006a; Whitehead, Crawford, 2006b) are unlikely to reflect environmental variation and is more reasonable assigned to genetic variation.

One of the assumptions in this work is that acclimation removes most, if not all, of the physiological differences among individuals. Clearly, acclimation to a common environment can remove many physiological differences especially differences in enzyme expression (Crawford, Powers, 1989; Hochachka, Somero, 1984; Pierce, Crawford, 1997; Prosser, 1986; Schmidt-Neilsen, 1990; Segal, Crawford, 1994). However, these observations do not refute that other non-heritable mechanisms can effect mRNA expression. For example in clones of sea anemones, metabolic rates are not affected by acclimation to a common temperature. Instead metabolisms among genetically identical individuals reflect an individual’s developmental temperature. Similarly, the maximum expression of heat shock proteins in sea urchins (Strongylocentrotus purpuratus) was unaffected by acclimation temperature but appeared to be influenced by irreversible acclimation at early life stages (Osovitz, Hofmann, 2005). Additionally, an individual’s phenotype can be influenced by maternal and other epigenetic effects. Thus, one could suggest that the inbred individuals (G10) whose parents all experience the same environment affected mRNA expression differently than acclimated individual whose parents experience a wide range of environments. However, there are few differences in expression among G10, acclimated or field individuals. Thus, epigenetic affects causing a difference in gene expression are not supported. Alternatively, one could suggest that the variation in mRNA expression (but not a different in the mean expression) is related to the environmental variation experience by parents or developing embryo. That is, the more variable the paternal or developmental environment the greater the variation in mRNA expression. To explain most of the data, this hypothesis that the variation in expression is a function of parental or developmental environmental variation, would require that greater environmental variation produces many different adult phenotypes, and thus there is a greater variation in mRNA expression. This hypothesis is unlike any currently available data, and cannot be readily rejected. However, it seems more parsimonious to suggest that the larger inter-individual variation in mRNA expression is related to genetic variation, rather then a novel epigenetic mechanism that does not affect the mean mRNA expression but instead creates greater individual variation.

To summarize: our data support the hypothesis that variation in metabolic mRNA expression is primarily related to the genetic variation among individuals. For G10 individuals, high amounts of relatedness and low levels of allelic richness are associated with less variation in mRNA expression. Surprisingly, the variation in metabolic mRNA expression is either lower or, at the very least, similar among field versus acclimated individuals. These data indicate that much of the variation in mRNA expression is related to genetic variation and less of the variation is in response to environmental variation.

Supplementary Material

Supp Fig S1

Supp Fig S2

Supp Table S1


The authors thank Richard Winn and Michelle Norris at the Aquatic Biotechnology and Environmental Laboratory at the University of Georgia for providing the G10 fish. We thank Dr. Marjorie Oleksiak for the creation of Fundulus microarrays, and Jeffrey VanWye for EST sequencing and printing of the microarray. This project was supported by NSF (OCE 0308777 BE/GEN-EN Functional Genomics) and by a National Institutes of Health grant (NHLBI R01 HL065470) to Douglas L. Crawford.


  • Adams SM, Lindmeier JB, Duvernell DD. Microsatellite analysis of the phylogeography, Pleistocene history and secondary contact hypotheses for the killifish, Fundulus heteroclitus. Molecular Ecology. 2006;15:1109–1123. [PubMed]
  • Adams SM, Oleksiak MF, Duvernell DD. Microsatellite primers for the Atlantic coastal killifish, Fundulus heteroclitus, with applicability to related Fundulus species. Molecular Ecology Notes. 2005;5:275–277.
  • Bowcock AM, Ruizlinares A, Tomfohrde J, et al. High-Resolution of Human Evolutionary Trees with Polymorphic Microsatellites. Nature. 1994;368:455–457. [PubMed]
  • Brem RB, Kruglyak L. The landscape of genetic complexity across 5,700 gene expression traits in yeast. Proceedings of the National Academy of Sciences of the United States of America. 2005;102:1572–1577. [PMC free article] [PubMed]
  • Cheung VG, Jen KY, Weber T, et al. Genetics of quantitative variation in human gene expression. Cold Spring Harb Symp Quant Biol. 2003;68:403–407. [PubMed]
  • Cobb JP, Mindrinos MN, Miller-Graziano C, et al. Application of genome-wide expression analysis to human health and disease. Proc Natl Acad Sci U S A. 2005;102:4801–4806. [PMC free article] [PubMed]
  • Crawford DL, Oleksiak MF. The biological importance of measuring individual variation. Journal of Experimental Biology. 2007;210:1613–1621. [PubMed]
  • Crawford DL, Powers DA. Molecular basis of evolutionary adaptation at the lactate dehydrogenase-B locus in the fish Fundulus heteroclitus. Proceedings of the National Academy of Sciences of the United States of America. 1989;86:9365–9369. [PMC free article] [PubMed]
  • Eady JJ, Wortley GM, Wormstone YM, et al. Variation in gene expression profiles of peripheral blood mononuclear cells from healthy volunteers. Physiol Genomics. 2005;22:402–411. [PubMed]
  • Eberwine J. Amplification of mRNA populations using aRNA generated from immobilized oligo(dT)-T7 primed cDNA. Biotechniques. 1996;20:584–&. [PubMed]
  • Falconer DS, Mackay TFC. Introduction to Quantitative Genetics. Fourth. Pearson Education Limited; Harlow: 1996.
  • Gibson G, Weir B. The quantitative genetics of transcription. Trends in Genetics Electronic vers. 2005 InPress.
  • Goudet J. FSTAT (Version 1.2): A computer program to calculate F-statistics. Journal of Heredity. 1995;86:485–486.
  • Hochachka PW, Somero GN. Biochemical Adaptation. Princeton University Press; Princeton, NJ: 1984.
  • Kerr MK, Churchill GA. Experimental design for gene expression microarrays. Biostatistics. 2001;2:183–201. [PubMed]
  • Marshall WS. Rapid regulation of NaCl secretion by estuarine teleost fish: coping strategies for short-duration freshwater exposures. Biochimica et Biophysica Acta (BBA) - Biomembranes. 2003;1618:95. [PubMed]
  • Marshall WS, Ossum CG, Hoffmann EK. Hypotonic shock mediation by p38 MAPK, JNK, PKC, FAK, OSR1 and SPAK in osmosensing chloride secreting cells of killifish opercular epithelium. J Exp Biol. 2005;208:1063–1077. [PubMed]
  • Monks SA, Leonardson A, Zhu H, et al. Genetic inheritance of gene expression in human cell lines. American Journal of Human Genetics. 2004;75:1094–1105. [PMC free article] [PubMed]
  • Oleksiak MF, Churchill GA, Crawford DL. Variation in gene expression within and among natural populations. Nature Genetics. 2002;32:261–266. [PubMed]
  • Oleksiak MF, Roach JL, Crawford DL. Natural variation in cardiac metabolism and gene expression in Fundulus heteroclitus. Nature Genetics. 2005;37:67–72. [PMC free article] [PubMed]
  • Osovitz CJ, Hofmann GE. Thermal history-dependent expression of the hsp70 gene in purple sea urchins: Biogeographic patterns and the effect of temperature acclimation. Journal of Experimental Marine Biology and Ecology. 2005;327:134.
  • Paschall JE, Oleksiak MF, VanWye JD, et al. FunnyBase: a Systems Level Functional Annotation of Fundulus ESTs for the Analysis of Gene Expression. BMC Genomics. 2004;5:96. [PMC free article] [PubMed]
  • Patterson TA, Lobenhofer EK, Fulmer-Smentek SB, et al. Performance comparison of one-color and two-color platforms within the MicroArray Quality Control (MAQC) project. Nat Biotechnol. 2006;24:1140–1150. [PubMed]
  • Peakall R, Smouse PE. GENALEX 6: genetic analysis in Excel. Population genetic software for teaching and research. Molecular Ecology Notes. 2006;6:288–295. [PMC free article] [PubMed]
  • Pierce VA, Crawford DL. Phylogenetic analysis of thermal acclimation of the glycolytic enzymes in the genus Fundulus. Physiological Zoology. 1997;70:597–609. [PubMed]
  • Prosser CL. Adaptational Biology: From Molecules to Organisms. Wiley; New York: 1986.
  • Queller DC, Goodnight KF. Estimating Relatedness Using Genetic-Markers. Evolution. 1989;43:258–275.
  • Radich JP, Mao M, Stepaniants S, et al. Individual-specific variation of gene expression in peripheral blood leukocytes. Genomics. 2004;83:980–988. [PubMed]
  • Raymond M, Rousset F. Genepop (Version-1.2) - Population-Genetics Software for Exact Tests and Ecumenicism. Journal of Heredity. 1995;86:248–249.
  • Rockman MV, Kruglyak L. Genetics of global gene expression. Nature Reviews Genetics. 2006;7:862–872. [PubMed]
  • Schadt EE, Monks SA, Drake TA, et al. Genetics of gene expression surveyed in maize, mouse and man. Nature. 2003;422:297–302. [PubMed]
  • Schmidt-Neilsen K. Animal Physiology: Adaptation and Environment. Fourth. Cambridge University Press; New York, NY: 1990.
  • Scott CP, VanWye JD, McDonald D, Crawford Douglas L. Technical Analysis of Fundulus heteroclitus cDNA Microarrays. PloS One. 2009 Accepted.
  • Segal JA, Crawford DL. LDH-B enzyme expression: the mechanisms of altered gene expression in acclimation and evolutionary adaptation. American Journal of Physiology. 1994;267:R1150–1153. [PubMed]
  • Sharma A, Sharma VK, Horn-Saban S, et al. Assessing natural variations in gene expression in humans by comparing with monozygotic twins using microarrays. Physiological Genomics. 2005;21:117–123. [PubMed]
  • Sokal RR, Rohlf FJ. Biometry: The Principles and Practice of Statistics in Biological Research. Third. W.H. Freeman and Company; New York: 1995.
  • Tan Q, Christensen K, Christiansen L, et al. Genetic dissection of gene expression observed in whole blood samples of elderly Danish twins. Human Genetics. 2005;117:267–274. [PubMed]
  • Wayne ML, Pan YJ, Nuzhdin SV, McIntyre LM. Additivity and trans-acting effects on gene expression in male Drosophila simulans. Genetics. 2004;168:1413–1420. [PMC free article] [PubMed]
  • Whitehead A, Crawford D. Variation in tissue-specific gene expression among natural populations. Genome Biology. 2005;6:R13.11–13.14. [PMC free article] [PubMed]
  • Whitehead A, Crawford DL. Neutral and adaptive variation in gene expression. Proc Natl Acad Sci U S A. 2006a;103:5425–5430. [PMC free article] [PubMed]
  • Whitehead A, Crawford DL. Variation within and among species in gene expression: raw material for evolution. Molecular Ecology. 2006b;15:1197–1211. [PubMed]
  • Whitney AR, Diehn M, Popper SJ, et al. Individuality and variation in gene expression patterns in human blood. Proceedings of the National Academy of Sciences of the United States of America. 2003;100:1896–1901. [PMC free article] [PubMed]
  • Williams RBH, Chan EKF, Cowley MJ, Little PFR. The influence of genetic variation on gene expression. Genome Research. 2007;17:1707–1716. [PubMed]
  • Wolfinger RD, Gibson G, Wolfinger ED, et al. Assessing gene significance from cDNA microarray expression data via mixed models. J Comput Biol. 2001;8:625–637. [PubMed]
  • Yu X, Chu TM, Gibson G, Wolfinger RD. A mixed model approach to identify yeast transcriptional regulatory motifs via microarray experiments. Stat Appl Genet Mol Biol. 2004;3 Article22. [PubMed]
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...