MeHg Tolerance in Wild Strains
Throughout this study, tolerance to MeHg was determined by quantifying eclosion of larvae reared on various concentrations of MeHg in the food medium (, inset). To determine the relative MeHg tolerance among several wild-derived strains of flies, we first assessed MeHg tolerance in the standard Canton S strain. We observed a sharp decrease in eclosion rate with increases in MeHg concentration and an ~50% lethal concentration (LC50) of 7.5μM MeHg (). Complete lethality in the Canton S strain was observed with 20μM MeHg.
FIG. 1. Effect of MeHg exposure on eclosion of Canton S flies. Eclosion of Canton S flies was determined with first instar larvae placed on indicated concentration of MeHg in food (bars = SD of three trials, n = 150). Fifty percent eclosion occurred at ~7.5μM (more ...)
We next assayed MeHg tolerance in 47 individual strains from various geographic locations in North America and Australia (see “Materials and Methods” section). Each strain, descended from a single female, had been maintained as an inbred line for more than 20 generations. Results of the eclosion assay demonstrate that there are significantly different levels of MeHg tolerance between strains (). As exemplified by the BF and RIH lines, there is a wide range of tolerance among lines derived from a single geographic region. We further analyzed the variance in tolerance trait by doing a pairwise comparison of the most and least tolerant lines from each geographic region. Eclosion was assessed on various concentrations of MeHg food. A significant difference in MeHg tolerance is seen between the HF7/HF9, Sorell9/Sorell15, RIH12/RIH11, BF50/BF54, and HG21/HG25 strains (Supplementary figure 1
). These data further confirm that natural variation in MeHg tolerance has a significant genetic component and can be determined with confidence via this assay system.
FIG. 2. MeHg tolerance of wild strains derived from various geographic regions. Forty-seven isolines from the indicated geographical region, plus Canton S and Oregon R strains, were assayed. Eclosion assays were performed with first instar larvae on 7.5μM (more ...)
To determine transmission of the tolerance trait to the F1 generation, we performed matings between tolerant and susceptible wild-derived strains. These were done in two reciprocal crosses to reveal basic patterns of dominance and chromosomal inheritance. Reciprocal crosses were done with the most tolerant (RIH12) and the least tolerant (Sorrel15) strains identified previously by eclosion assays on various concentrations of MeHg food (see Supplementary figure 1
). The F1 progeny from the RIH12 male × Sorell15 female were seen to be as tolerant as the parent RIH12 line (). Likewise, the F1 progeny from the RIH12 female × Sorell15 male were as tolerant as the RIH12 strain. These results indicate that MeHg tolerance is a dominant trait in the RIH12 line. If the dominant allele(s) is (are) localized predominantly on the X chromosome, we would expect F1 males of the respective reciprocal crosses to differ in tolerance. This was not observed with the RIH12 and Sorrel15 wild strains (data not shown), indicating that the tolerance trait is autosomal dominant. Additional reciprocal crosses of laboratory-selected tolerant and nontolerant flies show the same profile of autosomal dominance of MeHg tolerance (see below).
FIG. 3. Transmission of the tolerance trait in the progeny of wild strains. Eclosion assays were performed as in using the indicated tolerant (RIH12) and non-tolerant (Sorell15 and Sor15) parental strains of wild flies. Larvae assayed were derived from (more ...) Experimental Selection for the MeHg Tolerance Trait
To obtain additional lines of MeHg-tolerant flies, we performed a selection experiment starting with a “synthetic” natural mass population of flies (see “Materials and Methods” section). Included in this founding population were the 47 wild lines tested above. Six novel strains were established from this population: three replicate control (nonselected and nontolerant) lines were derived by culturing on “zero” MeHg (E0, F0, and H0) and three replicate selection (tolerant) lines were derived by culturing on MeHg food (E20, F20, and H20; note that MeHg was incrementally increased to a final concentration of 20μM [see “Materials and Methods” section]). Each line was cultured more than 20 generations and maintained as separate populations in culture bottle(s).
Selection resulted in robust MeHg tolerance as determined by eclosion assays. The LC50
for the H0 (control) flies was ~12μM MeHg with near-complete lethality at 20μM MeHg (). In the H20 strain, LC50
was greater than 20μM MeHg (). A similar high degree of MeHg tolerance was seen in the accompanying E20 and F20 lines relative to the E0 and F0 controls (Supplementary figure 2
FIG. 4. Transmission of the tolerance trait in the progeny of laboratory-selected strains. Eclosion assays were performed as in using the indicated tolerant (H20) and nontolerant (H0) parental strains of laboratory-selected flies (see ). Larvae assayed (more ...)
We further tested transmission of the tolerance trait to the F1 generation by performing reciprocal crosses in the H0/H20 pair as described above for the wild strains. Analogous to the wild strains, a pattern of autosomal dominance for the MeHg tolerance trait was seen in the F1 generation of the H20 × H0 line crosses (). It is of note that the laboratory-selected strains exhibited a more robust MeHg tolerance than the wild-derived tolerant strains, which suggests some heterosis for tolerance in the synthetic strains compared with inbred strains (compare RIH12 and H20 in and , respectively).
Transcript Profiling of MeHg Tolerance in the Developing Nervous System
The large tolerance difference between the three selection strains and three control strains is ideal for exploring the genetic basis of the MeHg tolerance trait. With the goal of identifying candidate MeHg tolerance genes, we have taken the approach of whole-genome transcript profiling using Affymetrix GeneChips. Our overall hypothesis is that MeHg tolerance is conferred by differences in transcription levels of genes that act individually, or in concert, to avert the toxicity of MeHg. It is predicted that MeHg tolerance can arise by numerous possible combinations of transcript levels. Therefore, a simple pairwise comparison of steady state transcript levels of three replicates of tolerant and nontolerant fly strains is not anticipated to distinguish differentially expressed MeHg tolerance genes from differential expression because of natural variation between two strains. We therefore tested the transcriptional response to the stressor (MeHg) by measuring transcript levels in tolerant and nontolerant lines both “on” and “off” MeHg following an experimental design seen in . This factorial design (+/−selection and +/−MeHg exposure) would allow us to dissect the interaction between evolved and induced responses to MeHg tolerance. In addition, we restricted our analysis to transcripts in the third instar larval brain. Our rationale for investigating this tissue is the preferential susceptibility of developing neural tissue for MeHg toxicity (Clarkson and Magos, 2006
FIG. 5. Experimental design to study the transcript profiling of MeHg tolerance. Replicate selection and control strains were derived by rearing a starting population on either MeHg or control food (see “Materials and Methods” section) to generate (more ...)
The set of 12 GeneChips met Affymetrix quality criteria based on presence calls, scale factors, background, and 3′:5′ ratios. The probe set × sample expression matrix exhibited a treatment effect (p
= 0.002) based on a robust nonparametric method (unweighted multiresponse permutation procedure, Mielke and Berry, 2007
, based on the Euclidean distance function). Parametric methods sufficed to identify genes differentially expressed for four of five contrasts investigated at a false discovery threshold of 0.05 (data not shown).
To simplify discussion of the microarray results, the following nomenclature was adopted: S0 pertains to the grouping of the three nonselected strains (E0, F0, and H0) and S20: the grouping of the three selection line replicates (E20, F20, and H20) (see ). Four pairwise comparisons were made with the data set: (1) S0 in the presence versus absence of MeHg (S0 + MeHg), (2) S20 versus S0 in the absence of MeHg (S20 vs. S0), (3) S20 in the presence versus absence of MeHg (S20 + MeHg), and (4) S20 versus S0 in the presence of MeHg (S20 vs. S0, MeHg). We first assessed the global changes under each of these pairwise comparisons using the criteria of ≥ 1.5-fold change with a p
value of ≤ 0.05. The fold change values of these genes are presented in Supplementary table 3
. The overall transcript changes are represented in scatter plots in . We find that MeHg exposure to the S0 strains results in a change in 361 transcripts with over 90% of these being downregulated (). Comparison of basal expression in the S20 versus the S0 strains shows a change in 246 transcripts, with the majority (72%) downregulated because of selection in the S20 (). MeHg exposure in the S20 strains shows 249 transcripts change; however, 44% of these are upregulated (). Expression levels between the S20 and S0 strains in the presence of MeHg stress show 233 transcripts differ with a majority (74%) of these genes being upregulated in the S20 strain compared with the S0 strain (). Thus, relative to the nonselected population (S0), we observe an overall trend in the transcript profile we describe as “down-down-up,” i.e., “down” by MeHg exposure, “down” by selection, and “up” by selection + MeHg exposure.
FIG. 6. Global changes in transcripts with selection and MeHg exposure. Scatter plots illustrating the pairwise comparisons of the entire probe set intensity data using the criteria of ≥ 1.5-fold change (black dots, 1.5-fold boundary marked by solid lines) (more ...)
The highly polarized pattern of this expression profile is even more apparent when a more stringent threshold of twofold change is applied (values in parentheses in ). Relative to basal expression in S0, MeHg treatment results in 94% of transcripts downregulated () and selection gives 88% of transcripts downregulated (). In contrast, MeHg treatment of the S20 strains gives 61% of transcripts upregulated () and more than 75% of transcripts upregulated compared with the S0 strains under MeHg stress (). This global change in transcript profile indicates that MeHg tolerance is likely conferred by a preferential upregulation of expression of a distinct set of genes.
Unexpectedly, we find that of the 110 genes that are upregulated by MeHg exposure in the S20 group, nearly half of these (52 genes [47%]) are downregulated by MeHg exposure in the control S0 group (). Furthermore, of these same 110 genes upregulated by MeHg in the S20 group, more than half of these (63 genes [57%]) are downregulated by the selection process (i.e., S20 vs. S0 in the absence of MeHg, ). Forty-nine genes are shared between these sets of downregulated genes, indicating that alleles of these genes in the S0 group that respond to MeHg with downregulation are replaced by alleles that constitutively express at lower levels as a result of selection in the S20 group. Even more striking is the apparent “reversal” of transcriptional response to MeHg that distinguishes the nonselected versus selected alleles of these 49 genes. This polarized transcriptional response presents an effective screening tool for these alleles in wild populations.
FIG. 7. Overlap of transcript changes in S20 and S0 strains in the presence and absence of MeHg. (A) Comparison of S20 treated with MeHg and S0 treated with MeHg. Of the 110 genes upregulated in S20 upon MeHg exposure, 52 are otherwise downregulated in S0 treated (more ...)
In an attempt to identify candidate genes associated with the MeHg tolerance trait, we performed an annotation cluster analysis of the 233 genes that show differential expression between S20 versus S0 with MeHg present using DAVID (Dennis et al., 2003
). This analysis showed a highly significant enrichment for the monooxygenase/oxidoreductase functional category (Supplementary table 2
, enrichment 4.01, p
< 5.3 × 10−11
). Numerous CYP
genes contribute to this score. In addition, this enrichment score was maintained within the group of 172 upregulated transcripts alone (data not shown). CYPs
form a superfamily of genes that are ubiquitous enzymes central to phase I metabolism of a wide variety of xenobiotics. We therefore investigated the changes of expression in the CYP
family of genes. In Drosophila, there are 90 CYP
genes with 84 of these represented by probes on the Affymetrix Genome 2 chip. The overall changes in CYP transcript levels under the four pairwise comparisons can be seen in . Ten CYP
genes are upregulated in S20 versus S0 in the presence of MeHg (≥ 1.5-fold change, p
value of ≤ 0.05, ). No CYPs
are downregulated in this comparison. In contrast, only downregulation of CYPs
is seen with the S0 strain in the presence of MeHg (). As well, the S20 strain shows downregulated CYP
expression compared with S0 in the absence of MeHg (). With MeHg treatment of the S20 strain, four CYPs
are upregulated and three are downregulated (). Thus, the overall trend of transcript changes seen in the CYP family of genes adheres to the down-down-up profile elucidated above.
FIG. 8. Overall changes in CYP transcripts. Scatter plots illustrating the pairwise comparisons of the CYP family probe set intensity data using the criteria of ≥ 1.5-fold change (boundary marked by solid lines) and a p value of ≤ 0.05. (Axes (more ...)
Of the 10 upregulated CYP
genes, CYP6g1 shows the maximum fold change. In the S20 group, CYP6g1 is upregulated 3.8-fold by MeHg and 6.2-fold in S20 versus S0 (+ MeHg) ( and Supplementary table 3
). In the S0 group, CYP6g1 is knocked down 3.5-fold by MeHg ( and Supplementary table 3
). CYP6g1 is also expressed 2.1-fold lower in the S20 lines relative to the S0 strain without MeHg exposure ( and Supplementary table 3
); however, the latter does not reach the < 0.05 significance level (p
= 0.067). Thus, CYP6g1 transcript levels follow the overall down-down-up profile. The fact that CYP6g1 is the most highly expressed CYP undergoing change in expression makes it a good candidate for validating the array data using qPCR and also for probing for the polarized expression pattern in response to MeHg among wild-derived tolerant and nontolerant fly lines.
To validate the microarray data using qPCR, we selected genes to analyze by the criteria of (1) overall robust upregulation of expression in the S20 strain and (2) adherence to the overall trend of down-down-up expression profile. In addition, we considered whether functional annotation was available for the genes and whether they were functionally associated with stress response. The list of genes analyzed and the bulk of the qPCR results can be seen in the Supplementary figures 3A–H
. Shown here are results for two genes, CYP6g1
, the latter being the most highly upregulated gene with selection + MeHg. Each of the three strains within each group (S0 and S20) were analyzed for microarray probe intensity (expressed in log2
scale) and compared with qPCR signal (expressed as fold change relative to the control strain). For simplicity, pairwise comparison of E0/E20, F0/F20, and H0/H20 are presented despite each strain being raised independently of each other and comparisons of, e.g., E0 with F20 or E0 and H20 are equally relevant. (It should be noted that statistical characterization of the selection process [e.g., S0 vs. S20] accounted for all pairwise comparisons of individual selected and nonselected strains [see “Materials and Methods” section].) For CYP6g1, the overall trend of down-down-up can be seen in the probe intensity data (). A strong correlation between the microarray data and qPCR analyses is apparent with both CYP6g1 and TotA. TotA exhibits a slightly varied profile across the three replicates. Comparison of E0/E20 and H0/H20 demonstrates the exceptionally high upregulation of TotA in the selection + MeHg (). Although comparison of F0/F20 adheres to the down-down-up profile, it demonstrates a lower relative expression in the F20 with MeHg (). This latter comparison is largely influenced by the unusually high basal expression of TotA in the F0 line (see the microarray probe intensity; see ). Nonetheless, in all cases, the nonselected strains respond to MeHg with downregulation of CYP6g1 and TotA, whereas the selected strains respond to MeHg with an upregulation of these two genes. Strong agreement of the probe intensity and qPCR expression level was seen across eight additional genes, thus confirming the overall validity of the microarray results (Supplementary figures 3A–H
). As well, the profile of “down” in the nonselected strains and “up” in the selected strains in response MeHg is seen for all these genes (Supplementary figures 3A–H
FIG. 9. Validation of microarray data by qPCR. Microarray probe intensity (log2) of Cyp6g1 (Ai–iii) and TotA (Bi–iii) with (+) and without (−) MeHg exposure (15μM). Relative transcript levels (fold change relative to −selection/−MeHg) (more ...)
To confirm that laboratory selection yielded isolation of relevant tolerance genes, we asked whether CYP6g1 and TotA expression in wild-derived nontolerant and tolerant strains shows a similar “down” versus “up” response to MeHg. We analyzed transcript levels of TotA and CYP6g1 in the larval brains of five tolerant lines and five susceptible lines from the isolines lines analyzed in . In three of the five tolerant strains (HF9, Sorrel9, and HG21) CYP6g1 expression is seen to increase upon MeHg exposure (). TotA expression is also increased in two of these lines (Sorrel9 and HG21) and unchanged in a third (HF9) (). In contrast, CYP6g1 and TotA expression is repressed in three and four of the nontolerant strains, respectively, in response to MeHg exposure (). As a whole, the data support the general trend that MeHg tolerance is positively correlated with the direction of change in expression of CYP6g1 and TotA in response to MeHg exposure.
FIG. 10. Relative transcript levels of TotA and Cyp6g1 in wild-derived tolerant and nontolerant strains. The fold change of TotA and CYP6g1 expression determined by qPCR upon exposure to 15μM MeHg in larval brains is shown. Five tolerant (A) and nontolerant (more ...) Induction of MeHg Tolerance with TotA Overexpression
Transcriptional profiling results suggest that upregulated expression of a limited cohort of genes is capable of inducing MeHg tolerance. To test the possibility that individual genes in this cohort are capable of invoking MeHg tolerance, we turned our attention to TotA
, which shows the greatest relative upregulation (9.2-fold) with selection + MeHg (Supplementary table 3
is a previously described humoral response gene that is turned on in response to a number of stressors, including heat shock and bacterial infection (Ekengren and Hultmark, 2001
expression in metal toxicity has not been investigated nor has the functional activity of TotA been described. However, TotA overexpression has been shown to increase longevity of flies exposed to heat stress (Ekengren and Hultmark, 2001
). We therefore tested whether TotA can functionally support MeHg tolerance. Using the conventional Gal4-UAS system of gene expression (Brand et al., 1994
), we overexpressed TotA specifically in the nervous system using crosses of the elav-Gal4 neural-specific driver and UAS-TotA responder (EG4 > TotA). Control crosses employed the standard w1118 laboratory strain in place of the UAS responder (EG4 > 1118). Expression of TotA in the larval brain was enhanced more than 50-fold with the EG4 > TotA combination compared with control crosses (). Development and eclosion of TotA overexpressing larvae was then assayed. We observed no significant increase in the rate of eclosion of adults in the EG4 > TotA flies as compared with the EG4 > 1118 flies (, dotted/dashed lines), indicating that TotA alone does not induce robust MeHg tolerance. However, we observed that a significant fraction of the EG4 > TotA larvae progressed to a late stage of pupal development as determined by formation of dark pupae in the vials. Extracting these pupae from their cases revealed that normal metamorphosis had occurred as determined by the presence of complete adult eye, wing, and bristles structures (data not shown). In contrast, the EG4 > 1118 larvae showed a marked deficit in development of the pupae (i.e., higher proportion of “white” incompletely metamorphosed pupae). When scored as the number of larvae reaching or surpassing the dark pupae stage, the data demonstrate significantly enhanced MeHg tolerance in the TotA overexpressing flies (, solid lines). To confirm this effect, we examined TotA expression by an additional Gal4 driver line. Using the c754Gal4 driver that expresses predominantly in the fat body of the larvae (Hrdlicka et al., 2002
), we observe a similar overexpression of TotA (~45-fold, ). As well, tolerance to MeHg was observed through the greater number of dark pupae that resulted at the 10 to 20 μM Melts Treatments (). Together, these data demonstrate an activity for TotA in mechanisms of MeHg tolerance.
FIG. 11. Transgenic expression of TotA induces MeHg tolerance. (A) Relative levels of TotA expression (by qPCR) under Gal4-driven expression in the nervous system (EG4 > TotA) or whole larvae (c754G4 > TotA) compared with controls (EG4 > (more ...) Additional Candidate MeHg Tolerance Genes
Analysis of functional annotation clusters identified few additional clusters with significant enrichment scores among the 233 genes that change in the S20 versus S0 (+ MeHg) comparison. The next most significant cluster contained glycoproteins/secreted/signal peptide proteins with an enrichment score of 3.69 (p
< 2.4 × 10−7
) (Supplementary table 2
). A notable gene in this group is persephone (up 2.1-fold), a serine protease that is responsible for activation of the Toll pathway in an antifungal defense mechanism in flies (Ashok, 2009
). It is interesting to note that 14 genes are allocated to a cluster of defense response/humoral response/immune response, which carries a low enrichment score (enrichment 1.87, p
= 2.0 × 10−4
) (Supplementary table 2
). Nonetheless, in this group are genes of note, particularly persephone, thor, serpent, GSH S-transferase D3 (GSTD3), and UDP-glucoronysyltransferase (ugt86Di). Thor (up 1.5-fold) is a eukaryotic initiation factor 4E-binding protein (4E-BP) also involved in antibacterial humoral response and response to oxidative stress (Levitin et al., 2007
). Serpent (up 1.9-fold) is a transcriptional regulator that controls hematopoeisis (Waltzer et al., 2002
GSTD3 and ugt86Di are both proteins that function in phase II xenobiotic metabolism. GSTs and UGTs cooperate with the phase I activity of CYPs by conjugating xenobiotics with small molecules that aid in clearance from the cell (Smart and Hodgson, 2008
). There are 21 GST family members represented on Affymetrix Drosophila Genome 2.0 GeneChips. Two GST members GSTE3 and GSTD3 are upregulated 2.3- and 1.6-fold, respectively (p
< 0.005), in selection + MeHg (Supplementary table 3
). A third GST (GSTE7) is upregulated 1.6-fold with marginal significance (p
= 0.068, Supplementary table 3
). ugt86Di is one of 18 UDP-glycosyltransferases represented on the Affymetrix GeneChip and is upregulated fourfold with selection + MeHg (Supplementary table 3
). However, a second UGT (ugt86Dd) is downregulated 1.7-fold in the same comparison (Supplementary table 3
). It is interesting to note that the elevated expression of GSTD3, GSTE3, and GSTE7 is also maintained in the S20 strains compared with the S0 strains in the absence of MeHg (Supplementary table 3
), indicating that these genes are selected for a higher steady state expression.
An additional candidate identified by its adherence to the down-down-up profile is alcohol dehydrogenase (ADH). ADH shows 4.8-fold lower expression in S20 than S0 without MeHg and 3.0-fold lower expression in S0 when treated with MeHg (Supplementary table 3
). S20 treated with MeHg results in a 2.8-fold upregulation of ADH and accounts for an overall 1.8-fold higher expression of ADH in the selection + MeHg comparison (Supplementary table 3
Several additional potential candidates showing transcriptional change with selection and MeHg exposure were identified and are summarized in tables presented in Supplementary table 3