Differential display of hepatic mRNA from killifish(Fundulus heteroclitus) inhabiting a Superfund estuary
Joel N. Meyer1, David C. Volz, Jonathan H. Freedman, Richard T. Di Giulio∗
Nicholas School of the Environment and Earth Sciences, and Integrated Toxicology Program, Duke University,Durham, NC 27708-0328, USA
Received 15 December 2004; received in revised form 22 March 2005; accepted 28 March 2005
Abstract
Fundulus heteroclitus(Atlantic killifish, mummichog) from a highly contaminated site on the Elizabeth River (VA, USA) areresistant to the toxicity of sediment from the site. However, the mechanistic changes that confer resistance to the toxicity arenot yet well understood. We utilized differential display to identify mRNAs that are differentially expressed in hepatic tissue ofmale and female killifish from the Elizabeth River environment, compared to killifish from a non-contaminated reference site,King’s Creek (VA, USA). Seventy-four differentially expressed mRNAs were initially identified (including sex and population-specific differences), and 65 of these were isolated and sequenced. A reverse northern blot array constructed of these cDNAs(plus an additional 15 previously sequenced mRNAs of interest) was used to confirm and quantify expression differences. Highinterindividual variability was observed in mRNA expression, but multiple differentially expressed mRNAs were identified,including 11 population-specific differences occurring in both sexes, 24 population-specific differences occurring in only one
E-mail address:[email protected] (R.T.D. Giulio).1 Present address: Laboratory of Molecular Genetics, National In-
stitute of Environmental Health Sciences, Research Triangle Park,NC 27709, USA.
1. Introduction
Fundulus heteroclitus(Atlantic killifish, mummi-chog) from a highly contaminated site on the ElizabRiver (VA, USA; adjacent to the former siteAtlantic Wood Industries, now a US EPA-designaSuperfund site) are resistant to the toxicity of sedimfrom the site. This sediment is teratogenic to embrand lethal to larval killifish bred from adults collect
from relatively pristine reference sites (Meyer and DiGiulio, 2002, 2003; Ownby et al., 2002). It is likely thatthe resistant phenotype is based upon both heritableand nonheritable mechanisms (Meyer and Di Giulio,2002, 2003). The Elizabeth River killifish show alter-ations in several biochemical parameters that may berelated to their ability to survive in their contaminatedenvironment. These include an increased expression ofintestinal glutathioneS-transferase andp-glycoprotein(Van Veld et al., 1991; Cooper et al., 1996, 1999;Armknecht et al., 1998), upregulated hepatic UDP-glucuronyltransferase (Gaworecki et al., 2004),upregulated antioxidant defenses (Meyer et al., 2003a;Bacanskas et al., 2004), and decreased inducibilityof cytochrome P4501A mRNA, protein and catalyticactivity (Van Veld and Westbrook, 1995; Meyer andDi Giulio, 2002; Meyer et al., 2002, 2003b). However,existing evidence does not support the hypothesis thatlack of cytochrome P4501A inducibility is adaptive(Meyer et al., 2002; Hawkins et al., 2002; Wassenbergand Di Giulio, 2004a,b). In addition, some of the otherdifferences are not heritable for multiple generationsunder non-contaminated conditions. Of the biochem-ical differences tested, the only large difference thatwas heritable for more than one generation in Eliza-beth River killifish raised in clean conditions was anincreased expression of manganese superoxide dismu-tase protein in whole-body larval homogenates. Thus, itis likely that additional, as-yet undiscovered mechanis-tic changes are involved in conferring heritable resis-t ts.
are areas of increasing interest (Anderson et al., 1994;Forbes, 1999; Elskus, 2001; Weis, 2002).
The goal of the experiments described here wasto identify mRNAs that are differentially expressedin hepatic tissue of killifish adapted to the ElizabethRiver environment, compared to killifish from a ref-erence site, King’s Creek (VA, USA). Differentiallyexpressed mRNAs were identified utilizing differen-tial display. This technique has been successfully usedto identify differences in gene expression after expo-sure to xenobiotics in aquatic (Denslow et al., 2001;Larkin et al., 2003; Maples and Bain, 2004; Roling etal., 2004) and other models (e.g.,Liao and Freedman,2002; Stein and Liang, 2002). Liver samples were usedbecause of the important role played by hepatic en-zymes in metabolism of PAHs and the importance ofthis organ as a target of PAH exposure and toxicity.
2. Materials and methods
2.1. Animal capture and tissue extraction
All killifish used in these experiments were feralkillifish trapped from the Elizabeth River (N36◦48′28.2′′, W076◦17′38.1′′) and King’s Creek (N37◦18′05.9′′, W076◦25′ 21.4′′) in baited minnow traps in May2002. King’s Creek is a tributary of the York River(VA, USA) that has been used previously as a referencesite for the Elizabeth River (Van Veld and Westbrook,1 eta tiona pm( ts eka fishf ticd to ber phicd
me ateri d.A erv-a s ori reg e ofm ere
ance to the toxicity of the Elizabeth River sedimenPolycyclic aromatic hydrocarbon (PAH) conce
rations in the sediments at the Elizabeth Riverre among the highest observed, in some cases000 ppm (Padma et al., 1998). PAHs represent a ubiitous, toxicologically important class of polluta
hat is growing in importance in the environment, pially due to the increased rate of fossil fuel consuion (Van Metre et al., 2000). Thus, it is of considerble interest to understand the mechanisms by w
he Elizabeth River population of killifish has adapo high levels of PAH. In addition, the Elizabeth Rivopulation of killifish is one of several populationsillifish identified as having adapted to highly contaated environments (reviewed inWirgin and Waldman004). The ability of vertebrates to adapt to contaated environments, and the implications of the phy
ogical and genetic changes associated with adapta
995; Meyer and Di Giulio, 2003; Gaworeckil., 2004). The average sediment PAH concentrat the King’s Creek sites was reported as 3 ptotal PAH; Van Veld and Westbrook, 1995). Recentudies have indicated that killifish from King’s Crere genetically suited for use as reference killi
or Elizabeth River killifish; i.e., detectable geneistance between the two populations appearedelated to PAH contamination, rather than geograistance (Mulvey et al., 2002, 2003).
Twelve male and twelve female adult killifish froach site were anesthetized by immersion in ice w
mmediately following capture, weighed, and killell fish used were in good health and free of obsble external or internal (upon dissection) lesion
nfections. All females from both populations weravid, or close to gravid, based on the presencature eggs in ovaries upon dissection. All fish w
adults, weighing between 6 and 10 g. Livers wereremoved, rinsed in ice-cold KCl (0.15 mM, pH 7.4),blotted dry, weighed, placed in RNAlater (Qiagen;Valencia, CA, USA), and stored at 4◦C.
2.2. RNA extraction
Total RNA was extracted from liver stored less thanone week using an RNeasy kit (Qiagen), accordingto the manufacturer’s instructions. The purified totalRNA was dissolved in water, aliquoted, and stored at−80◦C. RNA quality was verified by formaldehydeagarose gel electrophoresis, and RNA concentrationswere quantified using an Eppendorf Biophotometer(Hamburg, Germany). For reverse northern blot exper-iments, poly(A+) mRNA was subsequently isolatedusing the Oligotex mRNA Midi Kit (Qiagen). PossibleDNA contamination of RNA samples was removedusing RNAse-free DNAse prior to differential displayanalysis.
2.3. Differential display of mRNA
Differential display was carried out as described byLiao and Freedman (1998). Equal amounts of RNAfrom two adults of the same sex were pooled, andthree pools per sex per population were compared foreach primer set. First-strand cDNAs were generated inreverse transcriptase reactions with the RNAimage kit(GenHunter, Nashville, TN, USA), using 0.2�g totalRp le.P col,u erset ers,a s,P forr erepf anda iedo per( r sexp ereu itsG
lts ofa m-
parisons were made on the basis of pools of six adultlivers; confirmation with reverse northern showed thatless than 30% of the bands initially identified by differ-ential display were truly differentially expressed. Wesuspected that sex differences would be important, andincorporated them into the experimental design, basedon the significant effect of sex on various parametersstudied in previous experiments with these populationsof fish (Meyer et al., 2002, 2003a,b).
PCR products were separated in adjacent lanes on6%, 8 M urea/acrylamide gels, and gels were then driedon 3 M chromatography paper (Whatman Inc., Clifton,NJ, USA). The location of the differential expressedcDNAs was visualized by exposing Biomax MR film(Kodak) to the gels for approximately 24 and 72 h. Dif-ferentially expressed cDNAs (∼150 or more nt) weresubsequently isolated from the gels, re-amplified andpurified using the QIAEX II kit (Qiagen) or GenElutespin-columns (Sigma-Aldrich Corporation, St. Louis,MO, USA). Expression was defined as different onlyif differential band intensity was observed in bothPCR replicates of at least two of the three individu-als examined per sex per population (Stein and Liang,2002).
2.4. Subcloning and sequencing
Purified cDNAs were subcloned into pGEM-Tcloning vector (Promega, Madison, WI, USA) andsequenced. Sequencing reactions were carried outu dingp nesa ybase( sB theg forB
2b
2rm
a li-c alea mR-N en-t ave
NA and three different one-base-anchored H-T11Mrimers (where M may be G, A, or C) per sampCRs were carried out per the RNAimage protosing the same anchored primer as in the rev
ranscriptase reaction, one of 16 arbitrary 13mnd 10�Ci �-[35S]dATP (Amersham Bioscienceiscataway, NJ, USA) per reaction. To control
andom differences in cDNA synthesis, PCRs werformed in duplicate (Stein and Liang, 2002). Thus,
or each primer pair (combination of anchoredrbitrary primer), 24 PCR amplifications were carrut, corresponding to two replicate PCR reactionspooled) sample, and three pooled samples peer population. Forty-eight primer combinations wsed (arbitrary primers H-AP1 through H-AP16, k501 and G502, from GenHunter).This experimental design was based on the resu
n earlier pilot study in which mRNA expression co
sing T7 and SP6 primers. Sequences (exclurimers) were analyzed for similarity to known gend expressed sequence tags (ESTs) using funnhttp://funnybase.umkc.edu/supercraw3/), as well aLAST analysis using non-redundant, EST, andenomicFugu rubripesdatabases (National Centeriotechnology Information).
.5. cDNA macroarrays (reverse northern slotlots)
.5.1. Membrane (array)cDNA macroarrays were constructed to confi
nd quantify differential mRNA expression in repate King’s Creek and Elizabeth River-collected mnd female livers. The macroarray contained 60As identified as differentially expressed by differ
previously been shown to be toxicant-responsive, and2 loading control cDNAs.
These genes and the primers used to amplify thecorresponding cDNAs for incorporation onto the array,and the expected length of the cDNAs were as follows:
The preceding primers were designed de novo,and PCR products were sequenced to confirm theidentity of the amplified target. Previously publishedprimers were used for the following genes: CytochromeP4501A (CYP1A), Aryl Hydrocarbon Receptor 1( )f )faf
NaOH/100 mM EDTA solution, heating at 100◦C for10 min, and chilled on ice for 2 min. Denatured cDNA(25 ng) was spotted under vacuum on four replicatenylon membranes (Roche, Indianapolis, IN, USA) us-ing a 96-well dot blot microfiltration apparatus. Eachmembrane was then neutralized in 0.5 M Tris/1.5 MNaCl for 4 min and washed in sterile water for 2 min.Membrane-bound cDNAs were UV-cross-linked witha Spectrolinker XL 1000 (Spectronics Corporation,Westbury, NY, USA) and dried overnight before hy-bridization.
primer and 1�g poly(A+) mRNA from two replicateKing’s Creek or Elizabeth River male or female livers(derived from different fish than those used for differen-tial display procedure above, and thus independent bio-logical replicates) were pooled (2�g poly(A+) mRNAtotal), diluted with 4.25�L DEPC-treated water, andheat-denatured at 65◦C for 5 min. After chilling on icefor 2 min, 16.75�L stock solution (1X reverse tran-scriptase buffer, 20 U RNase inhibitor, 500�M dNTP(-dCTP), 100�Ci [�-32P] dCTP) was added and heatedat 37◦C for 10 min, followed by addition of 200 UMoloney murine leukemia virus (MMLV) reverse tran-scriptase and heating at 37 and 95◦C for 60 and 5 min,respectively. Radiolabeled cDNAs were purified usingMicroSpinTM G-25 columns (Amersham Biosciences)following the manufacturer’s instructions, and radioac-t
AHR1), AHR2, AHR Nuclear Translocator (ARNTrom Powell et al. (2000); AHR Repressor (AHRRrom Karchner et al. (2002); HIF-2� (primers HifQf2nd HifQb2) fromPowell and Hahn (2002)and�-actin
rom Meyer et al. (2003b).cDNA derived from hepatic mRNA was used
CRs with all of these primers, except for AHR1,hich heart-derived cDNA was used. SubclonedAs were amplified directly from miniprep produsing T7 and Sp6 primers, and additional cDNAs wroduced from mRNA by reverse transcriptase Psing custom or previously published primers (aboll PCR products were resolved on 1.5% agarose
o confirm size and purity. Subsequently, PCR pcts were purified with QIAquick spin columns (Qgen) and quantified spectrophotometrically. For eurified cDNA, 120 ng was diluted in 100�L sterileater, denatured following the addition of 10�L 4 N
ivity quantified using a scintillation counter.Three replicate site- and sex-specific hybridizat
ere performed on each of four membranes (oneor King’s Creek male, King’s Creek female, Elizabiver male, and Elizabeth River female) using ra
abeled cDNA probe (the three replicates were derrom pooled livers from different fish). Each membraas pre-hybridized at 72◦C for 2 h in ExpressHyb Hyridization Solution (BD Biosciences, San Jose,SA), 0.2X SSC, and 0.5 mg denatured salmon spNA. Equal concentrations of radiolabel cDNA pro
5× 106 cpm) were added to 0.2 mg salmon spNA and 20X SSC, denatured, and added to eembrane. Membranes were incubated with radeled cDNA overnight at 72◦C. Following hybridiza
ion, membranes were washed 30 min each forimes with primary wash solution (2X SSC, 0.1% SDnd twice with secondary wash (0.5X SSC, 0.1% S
at 65◦C while mixing. The level of bound probe wasdetermined following PhosphorImage analysis usingImageQuant imaging software (Molecular Dynamics,Sunnyvale, CA, USA). For re-use of membranes foradditional replicates, membranes were stripped of ra-diolabeled cDNA in 0.4 M NaOH for 30 min at 45◦C,washed in 0.5% SDS for 2 h at 65◦C, and neutralizedin 100 mM Tris (pH 8) at room temperature for 5 min.All stripped membranes were checked for probe elim-ination before proceeding to the next hybridization.
2.6. Statistical analysis
The signal (level of expression) of each mRNA afterbackground correction was normalized by dividingby the average signal for 16S rRNA. Three signals foreach mRNA were obtained since each hybridizationrepresented a biologically independent sample. Theexpression level for each mRNA was compared for
F ons of m one primers r (ER) R reactionsf red 1– shows a sexd pressio n.
the two populations and the two sexes both on thebasis of a two-fold difference in expression and by2-factor ANOVA. The criterion of either a greaterthan two-fold difference in expression or ap-value ofless than 0.05 (main effects) or 0.1 (interactive effect;Snecedor and Cohcran, 1967) was used to categorize“significant” differences in expression inTable 1and inthe text.
3. Results
3.1. Identification of differentially expressedmRNAs
Forty-eight primer pair combinations were utilized;an example of the gel image generated from the ampli-fication products of one set of primers is presented inFig. 1. Differentially expressed mRNAs were identi-
ig. 1. Representative comparison of replicate PCR amplificatiet. King’s Creek (KC) samples in left 12 lanes, Elizabeth Riveor each pooled sample run in adjacent lanes (samples numbeifference, third shows no differences and high variability in ex
ale and female (pooled) samples from both populations, usingin right 12 lanes, males and females as labeled, duplicate PC12). First box from top shows a population difference, secondn, fourth shows no differences and low variability in expressio
fied on the basis of (1) population only (i.e. populationdifference consistent in both sexes), (2) sex only (i.e.sex difference consistent in both populations), and(3) both population and sex (i.e. expression alteredin only one sex of a given population). Examples ofmRNAs expressed differently in a population-specificas well as a sex-specific fashion are highlighted in therepresentative gel shown inFig. 1, as are examplesof mRNAs not differentially expressed, and a mRNAthat showed very high variability in expression.
All cDNAs representing mRNAs that were ex-pressed differentially based on population, or in one sexof a population, were cut out. In addition, some but notall of the cDNAs apparently expressed in a sex-specificfashion (irrespective of population) were cut out. In to-tal, 74 cDNAs were excised; the primer combinationscorresponding to the cDNAs that were eventually in-cluded on the macroarray are listed inTable 1.
3.2. Subcloning and sequencing of cDNAs
Seventy of 74 differentially expressed cDNAswere reamplified, and 65 of those PCR products
were successfully cloned and sequenced. The ac-cession numbers for all arrayed cDNAs are indi-cated inTable 1, along with BLAST search results.BLAST analysis showed that 30% of the arrayedcDNAs did not correspond to any previously iden-tified sequence, 60% had high sequence identity tosequences in the database, and 10% were matchedESTs.
3.3. Reverse northern analysis
Three hybridizations, utilizing three biologi-cal replicates (with each replicate representing apool of two same sex individuals), were analyzed. Arepresentative hybridization is shown inFig. 2, with se-lected cDNAs highlighted. Twenty-five of the mRNAsshowed a statistically significant differential expression(p< 0.05); 38 of the mRNAs showed a greater than two-fold change in expression (Table 1). Most of the differ-ences observed were either population differences ob-served in only one sex (24 mRNAs) or sex differences(22 mRNAs); only 11 population differences present inboth sexes were observed (Fig. 3; using either two-fold
F entially n femalesf g’s Cre s Creekm xes.
ig. 2. Representative macroarray highlighting selected differrom both populations, UDPGP and Factor XI are higher in Kinales), and G6P is higher in Elizabeth River killifish of both se
regulated cDNAs. Vitellogenin and choriogenin are higher iek killifish of both sexes (with Factor XI especially high in King’
Fig. 3. Frequency of different kinds of altered expression (as definedby a two-fold increase/decrease in expression, orp< 0.05). “ER/KCboth sexes” indicates population differences present in both sexes;“ER/KC female” indicates population differences present only in fe-males; “ER/KC male” indicates population differences present onlyin males; “Male/female both” indicates sex differences present inboth populations.
or p-value as the criterion for a “real” difference).Eighteen of the 24 sex-specific population differenceswere observed in males. Twenty-two of the 65 total dif-ferences initially identified by differential display andsubsequently cloned and sequenced were confirmed byreverse northern analysis (again using either two-foldor p-value as the criterion for a significant difference).Five of the cDNAs that were spotted showed very low orundetectable signals following hybridization and wereexcluded from the analysis: HIF2�, XIR, AHR1, VIT2,and SPGP.
Significant variability was observed in the ex-pression levels of many of the genes examined, de-spite the fact that each mRNA sample was a poolof two individuals. This result is consistent withour differential display results (observable inFig. 1),as well as with previous studies that demonstrateda high degree of genetic variability present in thisspecies (Smith and Fujio, 1982), and a high degreeof interindividual variability in gene expression, evenwithin populations, of this species (Oleksiak et al.,2002).
We identified a number of genes that are differen-tially regulated in male, female, or killifish of bothsexes from a contaminated site on the Elizabeth River.Two principal observations can be made about the pat-terns of gene expression identified in this study. First,few of the differences identified corresponded to genesfrequently studied by toxicologists, suggesting novelhypotheses regarding the physiological response ofthe killifish from this contaminated site. Furthermore,many of the sequences obtained are novel. This resulthighlights the utility of using a multigene, hypothesis-free approach such as differential display to under-standing responses to contaminant exposure. Second,many of the population differences observed were sex-specific, suggesting that sex is important in mediatingcontaminant response for a large number of genes.
4.2. Identity of differentially expressed genes
One of the most striking population differences wasthe decreased expression of Factor XI (plasma throm-boplastin antecedent) in Elizabeth River killifish. Fac-tor XI is an endopeptidase involved in the intrinsiccoagulation pathway, and Factor XI deficiency leadsin humans to a facile bleeding disorder (O’Connell,2 byF m-b brinp na woo tedp d at kil-l nd2 not6 eren y,b ver-a ness iverk ea havec stic)
004). Factor XI is activated immediately upstreamactor XIIa, and contributes to the formation of throin, an enzyme that directly cleaves fibrinogen to fieptides (Butenas and Mann, 2002). Reverse northernalysis indicated that in addition to Factor XI, tut of the three other mRNAs for coagulation-relaroteins identified in this experiment also exhibite
rend towards downregulation in Elizabeth Riverifish: cDNA 6 (fibrinogen beta chain precursor) a0 (fibrinogen alpha and alpha-E subunits), but1/70/71 (plasminogen activator). These results wot statistically significant individually in this studut may merit further investigation based on the oll pattern of three out of four coagulation-related gehowing a pattern of downregulation in Elizabeth Rillifish liver. Although in this study we did not usny fish bearing obvious lesions, previous studiesharacterized external (inflammatory and neopla
and internal (preneoplastic and neoplastic) lesions inwild-caught Elizabeth River killifish (Vogelbein et al.,1990; Huggett et al., 1992), and it appears possible thatthe frequency and severity of lesions in Elizabeth Riverkillifish is exacerbated by poor blood-clotting capacity.To our knowledge, no studies of coagulative capacityhave been performed in this population of killifish.
Glucose 6-phosphate dehydrogenase (G6PDH,cDNA 23) is postulated to play an important role inxenobiotic detoxification and antioxidant defenses dueto its role in production of NADPH, which in addi-tion to other biological functions (e.g., macromolecu-lar biosynthesis), is needed for reduction of glutathioneand other cellular antioxidants, as well as for vari-ous xenobiotic metabolism pathways (Winzer et al.,2002). The sex difference we observed (i.e., higherin males) has been previously observed in Europeanflounder, and correlated to vulnerability to oxidativestress (Platichthys flesus; Winzer et al., 2002a). It isalso interesting that expression of additional carbohy-drate metabolism genes was altered, with UDP-glucosepyrophosphorylase (UDPGP, cDNA 1) lower, and glu-cose 6-phosphatase (G6P, cDNA 33) higher, in Eliz-abeth River killifish. UDPGP catalyzes a key step inglycogen formation, while G6P is involved in the lib-eration of glucose into the circulation from glycogen.Together, these results suggest that the Elizabeth Riverfish are more reliant on glycolytic energy metabolismthan reference killifish. Recent research demonstratesthat at least some PAHs may be capable of altering mi-t l.,2 im-p inE ther eha
wn-r iverk er-v erka edM Hsc ta lli fishi ion.
Kelly-Reay and Weeks-Perkins (1994)reported anincreased level of macrophage activity in wildcaughtElizabeth River killifish, andFrederick et al. (2003)reported both a decrease in circulating antibody titerfor five specific marine bacteria, and an increasein expression of lysozyme and cyclooxygenase twoproteins (markers of innate immune response), inwildcaught Elizabeth River killifish.
A significant number of the differentially expressedgenes did not correspond to sequences in the availabledatabases. These sequences may represent novel mR-NAs, or mRNAs that in the killifish are sufficientlydivergent from the cognate gene in other organisms topreclude identification. The latter possibility is strong,since it is common with differential display to sequenceonly the 3′ untranslated region of the gene, and thesequences of these regions are often less conservedthan the sequences of protein-coding regions of the se-quence (Li, 1997).
4.3. Role of sex in mediating differential geneexpression
Many toxicological studies do not include sex aspart of the experimental design, either using only onesex, pooling sexes, or ignoring sex as a variable. In-creasing numbers of studies with aquatic organismsare identifying sex as a significant factor in expres-sion of toxicologically important genes such as CYP1A(Stegeman and Chevion, 1980; Forlin and Hansson,1S dC ra 000a ;M yere ser-v DHa nonm rec-o asm cesb killi-fi llye com-p to-c tionf han
ochondrial function (Salazar et al., 2004; Xia et a004). These differences suggest the possibility ofaired aerobic (mitochondrial) energy metabolismlizabeth River killifish and are consistent with
educed ability of Elizabeth River killifish to survivypoxia, as compared to King’s Creek killifish (Meyernd Di Giulio, 2003).
Complement components C3 and C9 were doegulated in one or both sexes of Elizabeth Rillifish. This may contribute to the frequent obsation of poor health in wild-caught Elizabeth Rivillifish after being held in the laboratory (Meyernd Di Giulio, 2003; Van Veld and Nacci, accept;ichael Newman, personal communication). PA
an have immunosuppressant properties (Harper el., 1996; Carlson et al., 2004). However, the overa
mmune system response in Elizabeth River killis likely more complicated than simple suppress
982; many later studies e.g. as referenced inNavas andegner, 2001; Elskus, 2004), CYP3A30 (Hegelund anelander, 2003), glutathioneS-transferase (Gallaghend Sheehy, 2000; Gunawickrama and Goksøyr, 2),nd antioxidant enzymes (Livingstone et al., 1995cFarland et al., 1999; Winzer et al., 2002a,b; Met al., 2003). Our results support some of these obations (for example, lower CYP1A and lower G6Pctivity in females) and suggest that this phenomeay be even more widespread than previouslygnized; we observed approximately 2–3 timesany sex-specific as both-sex population differenetween the Elizabeth River and reference sitesh. Additional identified genes that were differentiaxpressed according to sex included complementonents C3 and C9, tributyltin-binding protein, hepayte growth factor activator, MDR, plasma coagulaactor XI (all expressed at higher levels in males t
in females), CYP1B1 and CYP3A30 (expressed higherin females but not males from the Elizabeth River, ascompared to their King’s Creek counterparts), as wellas multiple novel ESTs (e.g., cDNAs 13, 41, 57). Thus,our results along with those of many other researchers(e.g., as cited above) collectively suggest that sex of-ten plays a large role in modulating fish’s response toxenobiotics, at least in reproductively active individ-uals. Studies that ignore sex as a factor are likely toeither miss important information and/or suffer fromhigh “noise” that is in fact sex-related.
4.4. Creation of a pollutant-specific cDNA array
We have identified a number of genes that are ex-pressed in an altered fashion in male, female, or maleand female killifish from a Superfund site character-ized by contamination with creosote and other xeno-biotics. The differentially expressed mRNAs identifiedand incorporated into the array should provide a useful“signature” in gene expression, which can be comparedwith the signature created by other exposures.
In general, gene expression analysis will likely be ofparticular value in investigating mechanisms of toxicityand adaptation when those mechanisms and responsesare pleiotropic in nature. While it is not yet knownwhether the resistant phenotype of the Elizabeth Riverkillifish is based on one, a few, or many differentiallyexpressed genes, it is likely that the development of anarray of genes differentially expressed in the wild willa alsoa efuli osedt ronicc so-l ifiedb ,2a ech-n lysisi oesn
4
pli-c ationo nd
has highlighted the important role of sex in modulatinggene expression results. This study has also identifiednew genes and pathways as candidates for contributingto the adaptive phenotype exhibited by the ElizabethRiver killifish, including both novel mRNAs as wellas known mRNAs not previously postulated to be dif-ferentially regulated in these populations. We expect tofurther investigate the role of these genes in modulatingthe toxicity of the Elizabeth River sediments. Finally,we plan to continue adding genes to the array, and hopeto combine the use of this array with other emerginggene expression tools available for this species.
Acknowledgements
We gratefully acknowledge Vivian Liao, Seth Kull-man, David Bencic, Deena Wassenberg, Lee Barber,and Michael Mattie for their technical assistance. Thisresearch was supported in part by National Institutes ofHealth grants P42 ES10356 and T32-ES-07031, andthe Office of Naval Research grant no. 00014-00-1-0315.
References
Anderson, S., Sadinski, W., Shugart, L., Brussard, P., Depledge, M.,Ford, T., Hose, J., Stegeman, J., Suk, W., Wirgin, I., Wogan, G.,1994. Genetic and molecular ecotoxicology: a research frame-
A ted
B tress
. 58,
B y 67,
C ofon-
ron.
C isto-maluesFL,
C teing a
ssist in elucidating the mechanisms involved. Wenticipate that this collection of genes may prove us
n studying other populations that have been expo, and in some cases adapted to, high levels of chontamination. Finally, the cDNAs that we have iated can be added to the growing list of genes identy various researchers in killifish (e.g.,Oleksiak et al.001; Maples and Bain, 2004; Roling et al., 2004) thatre available for inclusion in gene arrays. These tologies permit comparative gene expression ana
n laboratory and field populations, an ability that dot exist with many model species.
.5. Conclusion
This study has permitted insight into the comated nature of altered gene expression in a populf killifish inhabiting a highly contaminated site, a
work. Environ. Health. Perspect. 102 (Suppl. 12), 3–8.rmknecht, S.L., Kaatari, S.L., Van Veld, P.A., 1998. An eleva
glutathioneS-transferase in creosote-resistant mummichog (Fun-dulus heteroclitus). Aquat. Toxicol. 41, 1–16.
acanskas, L.R., Whitaker, J., Di Giulio, R.T., 2004. Oxidative sin two populations of killifish (Fundulus heteroclitus) with dif-fering contaminant exposure histories. Mar. Environ. Res597–601.
Denslow, N.D., Bowman, C.J., Ferguson, R.J., Lee, H.S., Hem-mer, M.J., Folmar, L.C., 2001. Induction of gene expression insheepshead minnows (Cyprinodon variegatus) treated with 17�-estradiol, diethylstilbestrol, or ethinylestradiol: the use of mRNAfingerprints as an indicator of gene regulation. Gen. Comp. En-docrinol. 121, 250–260.
Elskus, A.A., 2001. Toxicant resistance in wildlife: fish populations.In: Robertson, L.W., Hansen, L.G. (Eds.), PCBs: Recent ad-vances in Environmental Toxicology and Health Effects. Uni-versity Press of Kentucky, Lexington, KY, USA, pp. 273–276.
Elskus, A.A., 2004. Estradiol and estriol suppress CYP1A expressionin rainbow trout primary hepatocytes. Mar. Environ. Res. 58,463–467.
Forbes, V.E., 1999. Genetics and ecotoxicology—insights from theinterface. In: Forbes, V.E. (Ed.), Genetics and Ecotoxicology:Current Topics in Ecotoxicology and Environmental Chemistry.Taylor and Francis, Philadelphia, PA, USA, pp. 1–8.
Forlin, L., Hansson, T., 1982. Effects of oestradiol-17 beta and hy-pophysectomy on hepatic mixed function oxidases in rainbowtrout. J. Endocrinol. 95, 245–252.
Frederick, L.A., Van Veld, P.A., Rice, C.D., 2003. Low antibodyresponses and elevated bioindicators of innate immunity increosote-adapted muumichogs,Fundulus heteroclitus. Abstract86, 12th International Symposium on Pollutant Responses in Ma-rine Organisms, Tampa, FL, USA.
Gallagher, E.P., Sheehy, K.M., 2000. Altered glutathione-S-transferase catalytic activities in female brown bullheads froma contaminated central Florida lake. Mar. Environ. Res. 50,399–403.
Gaworecki, K.M., Rice, C.D., van den Hurk, P., 2004. Induc-tion of phenol-type sulfotransferase and glucuronosyltransferasein channel catfish and mummichog. Mar. Environ. Res. 58,525–528.
Gunawickrama, S.H.N.P., Goksøyr, A., 2000. CYP1A and phase IIresponses ofScophthalmus maximusupon long-term repeated
iron.
H con-3F1
H son,ly-in-,
H tic ex-
H gel-sed-entSA,
K eg-e arylR2.
Kelly-Reay, K., Weeks-Perkins, B.A., 1994. Determination of themacrophage chemiluminescent reposne inFundulus heteroclitusas a function of ollution stress. Fish Shellfish Immun. 4, 95–105.
Larkin, P., Folmar, L.C., Hemmer, M.J., Poston, A.J., Denslow, N.D.,2003. Expression profiling of estrogenic compounds using asheepshead minnow cDNA array. Environ. Health Perspect. 111,839–846.
Li, W.-H., 1997. Molecular Evolution. Sinauer Associates, Sunder-land, MA, USA, pp. 184.
Liao, V.H.-S., Freedman, J.H., 1998. Cadmium-regulated genesfrom the nematode Caenorhabditis elegans. J. Biol. Chem. 273,31962–31970.
Liao, V.H.-S., Freedman, J.H., 2002. Differential display analysisof gene expression in invertebrates. Cell. Mol. Life Sci. 59,1256–1263.
Livingstone, D.R., Lemaire, P., Matthews, A., Peters, L.D., Porte, C.,Fitzpatrick, P.J., Forlin, L., Nasci, C., Fossatto, V., Wootton, N.,Goldfarb, P., 1995. Assessment of the impact of organic pollu-tants on Goby (Zosterisessor ophicephalus) and mussel (Mytilusgalloprovincialis) from the Venice Lagoon Italy: biochemicalstudies. Mar. Environ. Res. 39, 235–240.
Maples, N.L., Bain, L.J., 2004. Trivalent chromium alters gene ex-pression in the mummichog (Fundulus heteroclitus). Environ.Toxicol. Chem. 23, 626–631.
McFarland, V.A., Inouye, L.S., Lutz, C.H., Jarvis, A.S., Clarkem,J.U., McCant, D.D., 1999. Biomarkers of oxidative stress andgenotoxicity in livers of field-collected brown bullhead,Ameri-urus nebulosus. Arch. Environ. Contam. Toxicol. 37, 236–241.
Meyer, J.N., Smith, J.D., Winston, G.W., Di Giulio, R.T., 2003a.Antioxidant defenses in killifish (Fundulus heteroclitus) exposedto Superfund sediments and model prooxidants: short-term andheritable responses. Aquat. Toxicol. 65, 377–395.
Meyer, J.N., Wassenberg, D.M., Karchner, S.I., Hahn, M.E., DiGiulio, R.T., 2003b. Expression and inducibility of aryl hydro-carbon receptor pathway genes in wildcaught killifish (Fundulus
En-
M soci-
M 01A
sed-
M sednesis
, VA
M e-duat.
M by,rsityn-
exposure to B(a)P and B(a)P-PCB77 mixtures. Mar. EnvRes. 50, 68–69.
arper, N., Steinberg, M., Safe, S., 1996. Immunotoxicity of a restituted polynuclear aromatic hydrocarbon mixture in B6Cmice. Toxicology 109, 31–38.
egelund, T., Celander, M.C., 2003. Hepatic versus extrahepapression of CYP3A30 and CYP3A56 in adult killifish (Fundulusheteroclitus). Aquat. Toxicol. 64, 277–291.
uggett, R.J., Van Veld, P.A., Smith, C.L., Hargis Jr., W.J., Vobein, W.K., Weeks, B.A., 1992. The effects of contaminatediments in the Elizabeth River. In: Burton, G.A. (Ed.), SedimToxicity Assessment. Lewis Publishers, Boca Raton FL, Upp. 403–431.
archner, S.I., Franks, D.G., Powell, W.H., Hahn, M.E., 2002. Rulatory interactions among three members of the vertebrathydrocarbon receptor family: AHR repressor, AHR1, and AHJ. Biol. Chem. 277, 6949–6959.
heteroclitus) with different contaminant-exposure histories.viron. Toxicol. Chem. 22, 2337–2343.
eyer, J.N., Di Giulio, R.T., 2003. Heritable adaptation and asated fitness costs in killifish (Fundulus heteroclitus) inhabiting acontaminated estuary. Ecol. Appl. 13, 490–503.
eyer, J.N., Nacci, D.E., Di Giulio, R.T., 2002. Cytochrome P45(CYP1A) in killifish (Fundulus heteroclitus): heritability of al-tered expression and relationship to survival in contaminatediments. Toxicol. Sci. 68, 69–81.
eyer, J., Di Giulio, R., 2002. Patterns of heritability of decreaEROD activity and resistance to PCB 126-induced teratogein laboratory-raised offspring of killifish (Fundulus heteroclitus)from a creosote-contaminated site in the Elizabeth RiverUSA. Mar. Environ. Res. 54, 621–628.
ulvey, M., Newman, M.C., Vogelbein, W., Unger, M.A., 2002. Gnetic structure ofFundulus heteroclitusfrom PAH-contaminateand neighboring sites in the Elizabeth and York Rivers. AqToxicol. 61, 195–209.
ulvey, M., Newman, M.C., Vogelbein, W., Unger, M.A., OwnD.R., 2003. Genetic structure and mtDNA sequence diveof Fundulus heteroclitusfrom polycyclic aromatic hydrocarbocontaminated sites. Environ. Toxicol. Chem. 22, 671–677.
Powell, W.H., Bright, R., Bello, S.M., Hahn, M.E., 2000. Devel-opmental and tissue-specific expression of AhR1, AhR2, andARNT2 in dioxin-sensitive and -resistant populations of the ma-rine fishFundulus heteroclitus. Toxicol. Sci. 57, 229–239.
Powell, W.H., Hahn, M.E., 2002. Identification and functional char-acterization of hypoxia-inducible factor 2� from the estuarineteleostsFundulus heteroclitus: Interaction of HIF-2� with twoARNT2 splice variants. J. Exp. Zool. 294, 17–29.
Roling, J.A., Bain, L.J., Baldwin, W.S., 2004. Differential gene ex-pression in mummichogs (Fundulusheteroclitus) following treat-ment with pyrene: comparison to creosote contaminated site.Mar. Environ. Res. 57, 377–395.
Salazar, I., Pavani, M., Aranda, W., Maya, J.D., Morello, A., Ferreira,J., 2004. Alterations of rat liver mitochondrial oxidative phospho-
.
S sts:itat
S Iowa
S me P-ture
Stein, J., Liang, P., 2002. Differential display technology: a generalguide. Cell. Mol. Life Sci. 8, 1235–1240.
Van Metre, P.C., Mahler, B.J., Furlong, E.T., 2000. Urban sprawlleaves its PAH signature. Environ. Sci. Technol. 34, 4064–4070.
Van Veld, P.A., Ko, U., Vogelbein, W.K., Westbrook, D.J.,1991. Glutathiones-transferase in intestine, liver, and hep-atic lesions of mummichog (Fundulus heteroclitus) from acreosote-contaminated environment. Fish Physiol. Biochem. 9,369–376.
Van Veld, P.A., Westbrook, D.J., 1995. Evidence for depressionof cytochrome P4501A in a population of chemically resis-tant mummichog (Fundulus heteroclitus). Environ. Sci. 3, 221–334.
Van Veld, P., Nacci, D., 2005. Toxicity Resistance. In: Di Giulio,R.T., Hinton, D.E. (Eds.), The Toxicology of Fishes. Taylor andFrancis, Washington, DC.
Vogelbein, W.K., Fournie, J.W., Van Veld, P.A., Huggett, R.J., 1990.Hepatic neoplasms in the mummichogFundulus heteroclitusfrom a creosote-contaminated site. Cancer Res. 50, 5968–5978.
Wassenberg, D.M., Di Giulio, R.T., 2004a. Synergistic embryotox-icity of polycyclic aromatic hydrocarbon aryl hydrocarbon re-ceptor agonists with cytochrome P4501A inhibitors inFundulusheteroclitus. Environ. Health Perspect. 112, 1658–1664.
Wassenberg, D.M., Di Giulio, R.T., 2004b. Teratogenesis inFun-dulus heteroclitusembryos exposed to a creosote-contaminatedsediment extract and CYP1A inhibitors. Mar. Environ. Res. 58,163–168.
Weis, J.S., 2002. Tolerance to environmental contaminants in themummichog Fundulus heteroclitus. Hum. Ecol. Risk Assess. 8,933–953.
Winzer, K., Van Noorden, C.J.F., Kohler, A., 2002a. Glucose-6-phosphate dehydrogenase: the key to sex-related xenobiotic tox-icity in hepatocytes of European flounder (Platichthys flesusL.)?Aquat. Toxicol. 59, 275–288.
Winzer, K., Van Noorden, C.J.F., Kohler, A., 2002b. Sex-specificre of
W orth
X tas,undspli-ect.
rylation and calcium uptake by benzo[a]pyrene. Toxicol. ApplPharmaol. 198, 1–10.
mith, P.J., Fujio, Y., 1982. Genetic variation in marine teleohigh variability in habitat specialists and low variability in habgeneralists. Mar. Biol. 69, 7–20.
nedecor, G.W., Cochran, W.G., 1967. Statistical Methods.State University Press, Ames, IA, pp. 346–349.
tegeman, J.J., Chevion, M., 1980. Sex differences in cytochro450 and mixed-function oxygenase activity in gonadally matrout. Biochem. Pharmacol. 29, 553–558.
biotransformation and detoxification after xenobiotic exposuprimary cultured hepatocytes of European flounder (PlatichthysflesusL.). Aquat. Toxicol. 59, 17–33.
irgin, I., Waldman, J.R., 2004. Resistance to contaminants in NAmerican fish populations. Mutat. Res. 18, 73–100.
ia, T., Korge, P., Weiss, J.N., Li, N., Venkatesen, M.I., SiouC., Nel, A., 2004. Quinones and aromatic chemical compoin particulate matter induce mitochondrial dysfunction: imcations for ultrafine particle toxicity. Environ. Health Persp112, 1347–1358.
