Steroid biosynthetic enzyme activities in leachate-exposed female perch (Perca fluviatilis) as biomarkers for endocrine disruption

ent 366 (2006) 638–648www.elsevier.com/locate/scitotenv

Science of the Total Environm

Steroid biosynthetic enzyme activities in leachate-exposed femaleperch (Perca fluviatilis) as biomarkers for endocrine disruption

Maria Linderoth a,⁎, Anna Norman b, Erik Noaksson c, Yngve Zebühr a,Leif Norrgren b, Lennart Balk a

a Department of Applied Environmental Science (ITM), Stockholm University, SE-106 91 Stockholm, Swedenb Department of Biomedical Sciences and Veterinary Public Health, Swedish University of Agricultural Sciences, SE-750 07 Uppsala, Sweden

c Jegrelius Research Centre, Department of Research and Development, Jämtland County Council, Studentplan 4:D, Campus,SE-831 40 Östersund, Sweden

Received 30 September 2005; received in revised form 12 January 2006; accepted 24 January 2006Available online 15 March 2006

Abstract

Studies have shown that adult female perch in a freshwater lake, Molnbyggen, Sweden, have a reproductive disorder caused byunidentified endocrine disrupting compounds (EDCs) leaching from a local refuse dump. The adverse effects include shallow opensores, low ratio of sexually mature individuals, low gonadosomatic index and low circulating levels of androgens. We hypothesisedthat the low androgen levels could be a result of impaired production and/or stimulated excretion of androgens by EDCs.

From October 2000 to November 2001, at time-points important in the perch reproductive cycle, adult female perch werecollected in Molnbyggen and in the reference lake, Djursjön. The activities of three key enzymes in androgen biosynthesis: 17α-hydroxylase (17OHlase), 17,20-lyase (lyase) and 17β-hydroxysteroid dehydrogenase (17βHSD) were determined in head kidneyor ovary. The relationship between enzyme activities and plasma steroid concentrations was examined. Ovarian histopathology andthe determination of brain aromatase activity were also included in the study.

Similar 17OHlase, 17βHSD and aromatase activities were found in Molnbyggen females and reference fish throughout the year.Head kidney 17OHlase showed a positive correlation to cortisol levels (r=0.754; pb0.001) but not to androgen levels.Molnbyggen females exhibited lower ovarian lyase activity during vitellogenesis than reference fish. Atretic oocytes were on mostoccasions more frequent in sexually immature than in sexually mature females. The results suggest that neither 17OHlase, 17βHSDnor aromatase is the target for EDCs disrupting the androgen homeostasis of exposed female perch. Further investigation is neededto establish the role of decreased ovarian lyase activity in endocrine homeostasis, but the possibility of increased excretion ofandrogens should also be examined.© 2006 Elsevier B.V. All rights reserved.

Keywords: Reproductive failure; Aromatase; 17,20-lyase; 17α-hydroxylase; 17β-hydroxysteroid dehydrogenase; Refuse dump leachate

⁎ Corresponding author. Tel.: +46 8 6747720; fax: +46 8 6747638.E-mail address: [email protected] (M. Linderoth).

0048-9697/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.scitotenv.2006.01.023

1. Introduction

In previous field studies we have reported effects ofexposure to landfill leachate on fish reproduction in alake, Molnbyggen, in Sweden. In adult female perch(Perca fluviatilis) reproductive responses included a

639M. Linderoth et al. / Science of the Total Environment 366 (2006) 638–648

high level of sexually immature (SIM) fish, as well asreduced gonad size (GSI), decreased brain aromataseactivity and low plasma 17β-estradiol (E2) and evenlower testosterone (T) levels among the fish that weresexually mature (SM) (Noaksson et al., 2001). Othereffects included high frequencies of fin erosion andshallow open sores. Similar adverse effects on thereproductive system of brook trout (Salvelinus fontina-lis) inhabiting the leachate-contaminated stream Vad-bäcken, which empties into Molnbyggen, were taken asevidence that the observed reproductive impairmentswere an effect of exposure to unidentified pollutantsleaching from the landfill in Lindbodarna (Noaksson etal., 2003a). Although the role of androgens in femalereproduction is still unclear, T has been found to sti-mulate both the production and release of luteinisinghormone (LH) in pituitary cells cultured from immatureEuropean silver eel (Anguilla anguilla) (Huang et al.,1997). Furthermore, androgens in combination withestradiol stimulated LH and gonadotropin-releasing hor-mone (GnRH) levels in the same species in vivo(Montero et al., 1995). Also, low doses of T implantedinto female orange-spotted grouper (Epinepheluscoioides) after spawning, induced ovarian developmentand GSI was higher than in controls after 90 days (Yehet al., 2003). It is thus possible that the T levels areinsufficient for activating the brain–pituitary–gonadal(BPG) axis in the major part of adult female perch inMolnbyggen, leaving them reproductively inactive.

Sweden still has about 500 active landfills andalthough each refuse dump's leachate is unique in itscomposition, we have found low frequency of SM incombination with sores and endocrine dysfunction in atleast one other lake when investigating female perch infour other leachate-contaminated lakes in Sweden(Noaksson et al., 2005). In both these cases the identityof the responsible pollutant(s) is unknown. Chemicalanalysis showed that levels of PAHs and PCBs are nothigher in Molnbyggen than in the reference Djursjön(Noaksson et al., 2003a). A study by Baun et al. (2004)illustrates the difficulty in predicting toxicity from dataon analysed pollutants. Existing toxicity data regardingthe 65 compounds identified in leachates from tenDanish landfills could explain only 2–37% of the algaltoxicity observed in the study. The opposite is also true—it is unlikely that the identity of responsible com-pounds will be revealed by non-specific biotests. Thus,in order to explore what compounds cause the disorderin fish in Molnbyggen, a further understanding of themechanism behind the low androgen levels is needed.An in vitro system could then be developed in whichtesting of fractionated matrices could facilitate the

identification of responsible pollutants. Our earlier in-vestigations of the female perch in Molnbyggen andbrook trout in Vadbäcken indicated that the P450sccactivity was unimpaired in exposed fish since proges-terone (P) levels were at least as high in exposed femalesof both these species compared to reference fish(Noaksson et al., 2003b). Steroid homeostasis couldtherefore be disrupted somewhere below the synthesisof P, either through decreased synthesis or increasedexcretion. Data from our year-long study of the plasmaprofiles of several sex steroids in female perch fromMolnbyggen and Djursjön also supported this conclu-sion (Noaksson et al., 2004). Inhibited brain aromatase(P450arom) was at first considered a promisingbiomarker, but later we concluded that the decrease inP450arom was probably a result of a down-regulation ofthe enzyme in response to the low androgen levels(Noaksson et al., 2001, 2003a).

To investigate the function of androgen biosynthesisin female perch, methods for the measurement of threekey enzyme activities–17α-hydroxylase (17OHlase),17,20-lyase (lyase) and 17β-hydroxysteroid dehydro-genase (17βHSD)–were established. The three activitieswere analysed in either head kidney or ovary in leachate-exposed female perch from Molnbyggen and referencefish from Djursjön. Analysis of brain P450arom activitywas also included. The activities were measured atdifferent time-points important in the perch reproductivecycle: at the start of gonadal recrudescence, at early andlate vitellogenesis and at spawning. We also investigatedthe relationship between enzyme activities and plasmasteroid concentrations. Histological analysis of ovariantissue was included to see whether the low androgensaffected oocyte maturation in SM fish and to confirm thesexual maturity status of the fish. The level of atresia(resorption of oocytes), a possible indicator of pollutantexposure was also recorded.

2. Materials and methods

2.1. Study locations and fish sampling

The study sites as well as the sampling procedureof the fish are described fully by Noaksson et al.(2004). In this study female perch were collected fromthe leachate-exposed lake Molnbyggen and thereference lake Djursjön on six time-points betweenOctober 2000 and November 2001 to cover the dif-ferent stages in the reproductive cycle: vitellogenesis,spawning and gonadal recrudescence. A public muni-cipal refuse dump, in use since 1980, is located on amountainside at Lindbodarna (60°41′N, 14°52′E),

640 M. Linderoth et al. / Science of the Total Environment 366 (2006) 638–648

Leksand, and approximately 3 km from Molnbyggen.The dump was intended for household refuse, butindustrial refuse of unknown origin and amount hasbeen deposited, especially prior to 1991 when strictermanagement was introduced. It has been estimated thatthe refuse dump generates 18,000 m3 of leachate yearly,one-third of which at least until 1998 was uncontrol-lably discharged to the surrounding environment andknown to contaminate Molnbyggen through Vadbäckenand Våtån (T. Lundgren, unpublished data). Femaleperch were divided into sexually mature (SM) andsexually immature (SIM) fish based on the occurrenceof maturing oocytes in the gonad. In July, in the verybeginning of the gonadal recrudescence this cannot bedetermined and all females were classified as SIM. SIMfemales were easily distinguished from SIM males bythe distinct morphological difference between thefemale ovary (single structure) and male testis (pairedbilaterally) in this species.

2.2. Tissue preparation

The head kidney, ovary and brain were homogenisedwith four times their weight of 0.1 M sodium-phosphatebuffer (0.25 M sucrose, pH 7.4) in a Potter-Elvehjelmhomogeniser. The head kidney was homogenised usingsix up-and-down strokes at 420 rpm, while the ovaryand brain were homogenised using four up-and-downstrokes at 400 rpm. The homogenates were centrifugedat 9000 gav for 10 min. The supernatants were mixedwith glycerol to a final concentration of 20%, to mini-mise adverse effects of freezing and thawing. Aliquotswere placed in cryo vials and immediately frozen inliquid nitrogen. All tissue preparation steps were carriedout at 0–4 °C in an ice bath. Supernatants were stored at−140 °C until use.

Slices (b3 mm) of the central part of the ovary werefixed in phosphate-buffered formalin, dehydrated in agraded series of ethanol and embedded in paraffin. Fromthree levels per ovary, 3 μm thick sections were cut,stained with eosin-haematoxylin and morphologicallyexamined by light microscopy. The morphologicalexamination included maturational staging and record-ing the presence of atretic oocytes. Oocytes were clas-sified as belonging to one of five maturational stages(I–V) according to Yamamoto (1956) and Treasurer andHolliday (1981), i.e. I: perinucleolar stage, II: corticalalveolar stage, III: early-mid vitellogenin stage, IV: latevitellogenin stage, V: ripe stage. Individuals withoocytes only in the pre-vitellogenic stages I–II wereconsidered SIM. The volume density (Vv) of oocytes inthe different maturational stages was estimated utilising

a point-count method based on the Delesse principle ofstereology (Weibel, 1979). For an individual fish, thevolume density, or proportional volume, was computedbased on the combined total counts from all micro-graphs from all three levels for that fish.

2.3. Enzyme assays

The measurements of the 17OHlase, lyase and17βHSD activities were based on the method by Sikkaet al. (1985) with modifications. The 17OHlase activitywas determined for 5 time-points during the reproduc-tive cycle by the in vitro conversion of progesterone(0.45 μM, 5×104 dpm) to 17α-hydroxyprogesterone.Each 1-ml incubation contained 8–30 μg head kidneyS9 protein in a 0.1 M sodium-phosphate buffer (0.25 Msucrose, 20% glycerol; pH 7.4). The detection limit was1.0×10−12 mol. Lyase activity was determined for3 time-points during the reproductive cycle by the invitro conversion of 17α-hydroxyprogesterone (0.93 μM,1×106 dpm) to androstenedione. Each 1-ml incubationcontained 4–29 mg ovary S9 protein in a 0.- M sodium–phosphate buffer (0.25 M sucrose, 20% glycerol;pH 7.4). The detection limit was 1.4×10−11 mol. The17βHSD activity was determined for 5 time-pointsduring the reproductive cycle by the in vitro conversionof androstenedione (4.5 μM, 5×104 dpm) to testoster-one. Each 1-ml incubation contained 0.14–1.0 mg headkidney S9 protein in a 0.1 M sodium-phosphate buffer(0.25 M sucrose, 20% glycerol; pH 7.4). The detectionlimit was 4.7×10−12 mol. All reactions were started bythe addition of NADPH to a final concentration of0.5 mM and terminated (after 20 min for 17OHlase and17βHSD, 20–50 min for lyase) by the addition of 0.1 mlof 1 M NaOH. Incubations took place at 30 °C in ashaking water bath. Products and remaining substratewere extracted with 5 ml ethyl acetate by vortexing for5 min in a multitube vortexer. After centrifugation for15 min at 3000 rpm an aliquot of the solvent wasevaporated and the residues resolved in a small amountof ethanol containing unlabelled reference steroids. Theproducts were separated by thin layer chromatography(TLC) in either of the following combinations of solventsystems (a) chloroform/diethylether (7:1) and toluene/methanol (9:2) or (b) toluene/methanol/acetone (9:2:1)and toluene/methanol (18:7). The thin layer plates wereactivated for 20 min in 120 °C prior to sample appli-cation. Products were tentatively identified based ontheir co-migration with UV–visible reference standardsand then detected on a Phosphor Imager (MolecularDynamics, CA, USA). The radioactive spots on theautoradiograms were quantified with ImageQuant 5.0

641M. Linderoth et al. / Science of the Total Environment 366 (2006) 638–648

computer software (Molecular Dynamics, CA, USA).Values of each radioactive steroid were expressed inpercentages, with 100% being the total radioactivityrecovered from a single TLC lane. From this, the pico-mole conversion of the substrate to the relevant productswas calculated. The identities of the quantified metabo-lites were confirmed by HRGC/HRMS analysis of themetabolites formed in incubations using only unlabelledsubstrate. Aromatase activity, determined for 5 time-points during the reproductive cycle, was measured asthe specific release of tritiated water accompanyingconversion of 1β-H3-androstenedione to estrone, pri-marily as described by Jeyasuria et al. (1994) with somemodifications (Noaksson et al., 2001). The 17OHlase,lyase and 17βHSD activities were determined for SMfemale perch but since all females were classified as SIMin July, the activities were analysed in SIM fish at thistime-point. Aromatase was determined for both SM andSIM female perch. All enzyme activities were demon-strated to be linear with time and amount of proteinunder the conditions employed and appropriate back-ground and control incubations were routinely per-formed. All enzymatic measurements were carried out induplicate. Enzyme activities are normalised to theprotein content which was measured using the techniquedescribed by Lowry et al. (1951) with bovine serumalbumin as standard.

2.4. Steroid analysis with HRGC/HRMS

Steroid metabolites in incubations using only un-labelled substrate were qualitatively identified withHRGC/HRMS according to Hartmann and Steinhart(1997), with some modifications. Briefly, this techniqueinvolves the precipitation of proteins with methanol,followed by the reduction of lipids by sequential ex-traction with n-hexane. Solid phase extraction (SPE)-cartridges were employed for further clean-up. C18-phases were used for separating steroids from fatty acidsand mono- and diglycerids. Both polar and non-polarsteroids were retained on the C18-phase and eluted withmethanol. The sample was then evaporated to drynessunder vacuum, re-dissolved in ethylacetate/methanol(8:2) and applied to a NH2-phase for further clean-up.Steroids were derivatised with N-methyl-N-(trimethyl-silyl)trifluoroacetamide (MSTFA)/iodotrimethylsilane(TMSI)/dithioerythritol (DTE) (1000:2:2) in 80 °C for2 h. The derivatisation mixture was removed by aliquid–liquid extraction prior to HRGC/HRMS analysis.The samples were on column injected with a siltekdeactivated silica column (2 m×0.53 mm; Restek,Bellefonte, PA, USA) as the retention gap. The detection

was carried out in the selected ion-monitoring mode,monitoring M+ and (M-15)+. The instrument was set toa resolution of 10000 and an acceleration voltage of8000 V.

2.5. Chemicals and materials

NADPH (N-7505), bovine serum albumin (A-7030),activated charcoal (C-5620) and chemicals for thederivatisation, as well as all reference steroids and un-labeled substrate, progesterone (P-0130), 17α-hydro-xyprogesterone (H-5752), androstenedione (A-9630)and testosterone (T-1500) were bought from SigmaChemical Company, St. Louis, USA. The radiolabelledsubstrates, 4-14C-progesterone (NEC-081, 0.02 mCi/mmol), 17α-[1, 2-3H(N)]-hydroxyprogesterone (NET-332, 1.0 mCi/mmol), 4-14C-androstenedione (NEC-136,0.02 mCi/mmol) and 1β-3H-androstenedione (NET-92623.1 mCi/mmol) were bought from NEN, Life ScienceProducts, Boston, USA. Thin layer chromatographyplates coated with silica gel G (0.25 mm) were obtainedfrom Tamro MedLab, Mölndal, Sweden. Solvents usedfor the TLC separation of steroids were of analyticalgrade and purchased from KEBO, Stockholm, Sweden.Methanol and ethyl acetate used for the clean-up prior toanalysis on HRGC/HRMS were obtained from ScharlauChemie S.A., Barcelona, Spain, and hexane and aceto-nitrile from Riedel de Häen, Seelze GmbH, Germany.Deuterated steroid standards, [16, 16, 17]-D3-T, [2, 4,16, 16]-D4-E2 and [9, 11, 12, 12]-D4-C were purchasedfrom Cambridge Isotope Laboratories, Inc., Andover,MA, USA. SPE-cartridges (C18 and NH2) were ob-tained from Supelco, Bellefonte, PA, USA. All otherchemicals were purchased from common commercialsources and were of analytical purity.

2.6. Equipment and apparatus

A refrigerated Eppendorf 5804R centrifuge was usedfor the preparation of head kidney, brain and ovarysupernatants. Liquid scintillation spectroscopy was per-formed employing a Packard Tri-carb 2100 TR liquidscintillation counter from Packard Instrument Company,while the protein determinations were done using aHitachi U-3200 spectrophotometer. A HRGC (HP6890,Hewlett Packard, Avondale, PA, USA) coupled to anAutospec Ultima HRMS (Micromass, Ultricham, UK)was employed for the steroid analysis. A PTE-5 fusedsilica capillary column (30 m×0.25 mm i.d., 0.25 μmfilm thickness from Supelco) was used for the separationof steroids. The mass spectrometer was operated inelectron impact mode at 32 eV.

642 M. Linderoth et al. / Science of the Total Environment 366 (2006) 638–648

2.7. Sex steroids

Plasma concentrations of A, T, E2 and C for the fishused in this study were previously reported (Noaksson etal., 2004). They are used here to determine whether arelationship exists between plasma steroid concentra-tions and the enzyme activities determined in this study.Briefly, both A and T levels increased in SM femaleperch from their lowest values right after spawning tothe highest values in April/May just prior to spawning.Average A levels in female perch were lower inMolnbyggen than in Djursjön, significantly so on 4out of 13 occasions. Similarly, average T levels werelower in SM female perch from Molnbyggen, signifi-cantly so on 5 out of 13 occasions.

2.8. Statistical analysis

The data on enzyme activities is presented as means±standard errors (SE) of the values obtained with anumber (n) of fish. Female perch from Molnbyggenwere compared with reference perch from Djursjön foreach sampling occasion. Significant differences inmeans of the enzyme activities were tested with theStudent's t-test or ANOVA or, when unequal variancespersisted after log-transformation, with the Mann-Whitney U-test or the Kruskal-Wallis test. For the sta-tistical analysis, enzyme activities that were below thedetection limit were considered equal to 50% of is value.

0

10

20

30

40

50

60

70

80

90

100

Oct Nov Dec Jan Feb Mar Apr

17O

Hla

se a

ctiv

ity

(pm

ol/m

in*m

g pr

otei

n)

0200

400600800

100012001400

16001800

0 50 100 15017OHlase (pmol/min/mg protein)

Cor

tiso

l (ng

/g p

lasm

a) Dj

Mb

Fig. 1. Head kidney 17α-hydroxylase activity (17OHlase) in female perch (Pexposed lake Molnbyggen (Mb). Activities were determined for five timeNovember 2001. The data is presented as mean±standard error for the valueswere classified as sexually immature. The activity was determined in 2–4relationship of plasma cortisol to 17OHlase activity for Dj (r=0.393, p=0.0

One-tailed Pearson correlation was used to examine therelationship between enzyme activities and plasmasteroid concentrations. The frequency distribution ofoocytes in different maturational stages is presented asmean±SE and was examined with the Student's t-testafter arcsin transformation. Differences in the propor-tions of fish with atretic oocytes were examined withFisher's exact test. The level of significance in all testswas set at pb0.05. The statistical software employedwas SPSS Inc. version 11.5 for Windows, Chicago, IL,USA.

3. Results

3.1. 17OHlase

In the present study the head kidney 17OHlaseactivity was detected throughout the year in femaleperch from both lakes investigated (Fig. 1) but nosignificant differences were detected at any time. Theactivity was stable in reference fish throughout vitel-logenesis and spawning (October to May) with activitypeaking at 73 pmol/min/mg protein at the end of Julyat the onset of gonadal recrudescence. In May, bothpre-spawning and post-spawning fish showed similarlevels of 17α-hydroxylase activity. 17OHlase activitywas positively correlated to cortisol levels in femalesfrom both Djursjön (r=0.393, p=0.039) and Molnbyg-gen (r=0.754, pb0.001) (Fig. 1, superimposed graph)

May Jun Jul Aug Sep Oct Nov

Dj

Mb

Dj (spawned)

Mb (spawned)

200

erca fluviatilis) from the reference lake Djursjön (Dj) and the leachate--points in the perch reproductive cycle between October 2000 andderived from 6 to 10 sexually mature fish, except in July when all fishpost-spawning fish in May. Superimposed is a graph showing the

39) and Mb (r=0.754, pb0.001).

0

20

40

60

80

100

120

140

Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov

Lya

se a

ctiv

ity

(fm

ol/m

in*m

g pr

otei

n) Dj

MbDj (spawned)Mb (spawned)

*

N.A. N.D.N.A. N.D.

Fig. 2. Ovarian 17,20-lyase activity (lyase) in female perch (Perca fluviatilis) from the reference lake Djursjön (Dj) and the leachate-exposed lakeMolnbyggen (Mb). Activities were determined for three time-points in the perch reproductive cycle betweenMay 2001 and November 2001. The datais presented as mean±standard error for the values derived from 6 to 10 sexually mature fish, except in July when all fish were classified as sexuallyimmature. The activity was determined in 3–4 post-spawning fish in May. *Significantly different from female perch in Djursjön when comparedwith the Student's t-test. The level of significance was set to pb0.05. NA=not analysed, ND=not detected.

Fig. 3. Representative autoradiogram of a thin layer chromatogram ofproducts of the incubation of ovarian S9 protein from pre-spawningfish caught in Djursjön at spawning time in May, with tritiated 17α-hydroxyprogesterone (17OHP). A major unidentified metabolite (X)was detected and separated from 17α,20β-dihydroxy-4-pregnane-3-one (17,20β-P) and 11-deoxycortisol (11-DC).

643M. Linderoth et al. / Science of the Total Environment 366 (2006) 638–648

whereas no correlation with androstenedione (Djursjön,r=−0.333, p=0.065; Molnbyggen, r=−0.048, p=0.422) was evident (not shown).

3.2. Lyase

The level of the ovarian lyase activity was sig-nificantly decreased by 30% (p=0.012) in the exposedfemale perch in November 2001 (Fig. 2). In May theactivity could be detected in only 14% (1/7) of referencefish and 50% (3/6) of exposed pre-spawning females,while none (0/3 and 0/4) of the post-spawning femaleshad detectable activities. Lyase activity was below thedetection limit in all fish sampled in July, resulting in toofew individuals with lyase activity to examine itsrelationship to plasma androstenedione and testosteronelevels. In May, however, instead of androstenedioneproduction both pre- and post-spawning fish fromDjursjön and Molnbyggen showed a marked increasein the in vitro production of a major metabolite (Fig. 3).This metabolite is, as yet unidentified, but distinguish-able from the known maturation inducing hormones(MIHs), 17α,20β-dihydroxy-4-pregnane-3-one(17,20β-P) and 11-deoxycortisol (11-DC).

3.3. 17βHSD

There were no significant differences in the headkidney 17βHSD activity between the exposed and ref-erence fish at any time in the reproductive cycle even

though the average activity was higher in MolnbyggenSM females at four out of five time-points (Fig. 4).There were no marked peaks in the activity of 17βHSD.

0

2

4

6

8

10

12

14

Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov

17ββH

SD a

ctiv

ity

(pm

ol/m

in*m

g pr

otei

n) Dj

Mb

Dj (spawned)

Mb (spawned)

Fig. 4. Head kidney 17β-hydroxysteroid dehydrogenase activity (17βHSD) in female perch (Perca fluviatilis) from the reference lake Djursjön (Dj)and the leachate-exposed lake Molnbyggen (Mb). Activities were analysed for five time-points in the perch reproductive cycle between October 2000and November 2001. The data is presented as mean±standard error for the values derived from 6 to 10 sexually mature fish, except in July when allfish were classified as sexually immature. The activity was determined in 2–4 post-spawning fish in May.

644 M. Linderoth et al. / Science of the Total Environment 366 (2006) 638–648

In May both pre- and post-spawning females had thesame level of activity. 17βHSD did not correlate toplasma concentrations of androstenedione (Djursjön,r =− 0.187, p = 0.208; Molnbyggen, r = − 0.150,p=0.270) or testosterone (Djursjön, r=−0.134,p=0.281; Molnbyggen, r=−0.147, p=0.274) (notshown).

0

10

20

30

40

50

60

70

Oct Nov Dec Jan Feb Mar Apr

Aro

mat

ase

acti

vity

(fm

ol/m

in*m

g pr

otei

n)

0

2

4

6

8

10

12

14

16

0 20 40 60Aromatase (fmol/min/mg protein)

Est

radi

ol (

ng/g

pla

sma)

Dj

Mb

Fig. 5. Brain aromatase activity (P450arom) in sexually mature (SM) and sexulake Djursjön (Dj) and the leachate-exposed lake Molnbyggen (Mb). Activibetween October 2000 and November 2001. The data is presented as mean±sand 3–5 post-spawning fish. Superimposed is a graph showing the relationshp=0.003) and Mb (r=0.152, p=0.249).

3.4. Aromatase

Brain aromatase activity was not significantlydifferent between SM female perch in the two lakesat any time-point analysed throughout the year (Fig. 5).The activity seems to be elevated at early vitellogen-esis reaching as high as 55 fmol/min/mg protein in

May Jun Jul Aug Sep Oct Nov

Dj

Mb

Dj (spawned)

Mb (spawned)

Dj (SIM)

Mb (SIM)80

ally immature (SIM) female perch (Perca fluviatilis) from the referenceties were analysed for five time-points in the perch reproductive cycletandard error for the values derived from 9–21 SM fish, 1–14 SIM fiship of plasma 17β-estradiol (E2) to aromatase activity for Dj (r=0.52,

Table 1Frequency distributions of oocytes in the different maturational stages in (A) sexually immature and (B) sexually mature female perch from lakesDjursjön (D) and Molnbyggen (Mb) from October 2000 to November 2001

Month Lake Oocyte maturational stage a, b

I II III IV V n

(A) Sexually immatureOctober 2000 D 91 9 1

Mb 84±4 15±3.5 10January 2001 D 86±1.6 14±1.7 2

Mb 77±4.7 19±3.1 10May 2001 D 0

Mb 88±2.9 12±2.9 12July 2001c D 64±6.8 36±6.8 7

Mb 64±2.3 36±2.7 9September 2001 D 0

Mb 88±7.5 3±2 5November 2001 D 84 9 1

Mb 83±3.0 15±2.9 14

(B) Sexually matureOctober 2000 D 6±1 94±0.9 9

Mb 8±1 0.1±0.04 92±1.2 8January 2001 D 2±0.2 98±0.3 10

Mb 2±0.4 98±0.4 8May 2001 D 1±1 49±17 50±18 8

Mb 1±0.4 59±24 40±24 5September 2001 D 18±1.4 0.1±0.05 82±1.4 19

Mb 15±1.2 0.1±0.04 85±1.3 13November 2001 D 4±1 96±1.0 11

Mb 5±1 95±1.4 10a The stage with the highest frequency for the time-point in bold.b Data is presented as mean±SE.c SIM and SM fish could not be distinguished at this time, all fish were classified as SIM.

Table 2Proportion (%) of (n) female perch from lakes Djursjön andMolnbyggen with atretic oocytes

Month SM SIM

Djursjön Molnbyggen Djursjön Molnbyggen

October 2000 0 (9) 0 (8) 0 (1) 30 (10)February 2001 0 (10) 0 (9) 0 (2) 11 (9)May 2001 0 (8) 0 (5) 17 (12)July 2001 a 29 (7) 33 (9)September 2001 26 (19) 18 (11) 29 (7)November 2001 18 (11) 10 (10) 100 (1) 29 (14)a SM and SIM could not be distinguished at this time, all fish were

classified as SIM.

645M. Linderoth et al. / Science of the Total Environment 366 (2006) 638–648

Djursjön females in October 2000. At spawning andgonadal recrudescence the levels were lower, 18 and16 fmol/min/mg protein, respectively. SIM femalesfrom Djursjön had lower average aromatase activitiesthan SM fish, but a similar annual pattern. Post-spawn-ing fish in May had brain aromatase activities com-parable to those who had yet to spawn. For SM fishthere was a positive correlation between brain aroma-tase activity and E2 in reference fish (r=0.52, p=0.003)but not in exposed (r=0.152, p=0.249) (Fig. 5, super-imposed graph).

3.5. Histological analysis

In July, no distinction between SM and SIM femaleperch can be made, neither by ocular inspection nor byhistological analysis. At this time all specimens fromboth lakes had oocytes in stages I (approx. 2/3) and II(approx. 1/3) (Table 1A). In September (Table 1A andB), the two groups were easily distinguished, SM fishwith oocytes mostly in stage III and SIM fish withmostly stage I-oocytes.

SM fish from both lakes had similar distributions ofoocytes in the five maturational stages at most of thetime-points sampled (Table 1B). The oocytes in SIMfish from both lakes were predominately arrested instage I at all time-points (Table 1A).

The number of investigated individuals was too lowto perform χ2-test on the proportions of fish with atresia.Between October 2000 and May 2001 no atretic oocytes

646 M. Linderoth et al. / Science of the Total Environment 366 (2006) 638–648

were present in SM fish from either lake or in the fewSIM fish from Djursjön, while atresia was present in aproportion of Molnbyggen SIM fish (Table 2). From thestart of the gonadal recrudescence in late July 2001 toNovember 2001 atretic oocytes were present in both SMand SIM fish at similar levels, in November a bit morepronounced in SIM fish (Table 2).

4. Discussion

Ovarian lyase showed a significantly altered activityin leachate-exposed SM female perch. This activity wasdecreased by 30% in November 2001. As reported inNoaksson et al. (2004), the average plasma androstene-dione level was, although not statistically significant,about 50% lower in females from Molnbyggen at thattime, implying that inhibition of lyase could be amechanism causing the low androgen levels. Lyaseactivity could in May only be detected in some of thepre-spawning females from both the reference (1/7) andfrom the contaminated site (3/6), while none of the post-spawning females had lyase activities over the detectionlimit, which is probably a result of the steroidogenicshift that occurs in the ovarian tissues right beforeovulation. This is supported by the fact that noindividual fish with oocytes mainly in the finaldevelopmental stage, V, had detectable lyase activity,while some of the less mature fish did, demonstratingthat the shift in synthesis is occurring before finalmaturation. The final oocyte maturation in fish requiresthe stimulation by a maturation-inducing hormone(MIH) (Nagahama, 1997). The proposed MIHs are allproduced from 17OHP, as is androstenedione, and assynthesis of MIH is induced, lyase activity is reduced(Senthilkumaran et al., 2004). Lyase activity wasanalysed in a few SIM fish from May but the activitieswere never over the detection limit indicating that thisenzyme is indeed inactive in SIM fish (data not shown).

The fact that in July no lyase activity could bedetected in fish from both lakes corresponds with theoverall low production of androgens and estrogensobserved at that time (Noaksson et al., 2004) and is notnecessarily an indication of an enzyme inhibited byEDCs. It could also be that important differences inlyase activity exist at this time, although non-detectable.

No differences in either the head kidney 17OHlase or17βHSD activity were found between female perchfrom the reference Djursjön and those from the leachate-exposed Molnbyggen. Similar results were obtained forSM female perch in November 1999 (Haarstad et al.,2002), when both activities were at approximately thesame level in both SIM fish and SM fish. Together with

the present work this suggests that a dysfunction of17OHlase or 17βHSD is not the main mechanismcausing the low androgen levels.

Since the 17OHlase and the lyase activities arecatalysed by the same active site on the same protein,P450c17 (Ahmed, 2004), different responses in headkidney 17OHlase and ovarian lyase indicate that EDC-action can not be through a general inhibition or down-regulation of the protein. Instead an EDC could either(a) selectively affect the lyase activity of the protein or(b) selectively affect the BPG axis but not the brain–pituitary–interrenal (BPI) axis and thus down-regulatethe ovarian form of the protein. 17OHP, synthesised by17OHlase, can either be converted further by lyase toandrostenedione and consequently to other sex steroidsor be released and used for the synthesis of glucocorti-coids or maturation-inducing hormones. Hence, there isa need for separate regulation of the both activities indifferent tissues and at different times in the reproduc-tive cycle. The basal lyase activity is usually much lowerthan the 17OHlase activity and is selectively increasedby stimulation of the electron transfer reactions, mostextensively studied with the rat or human form of theprotein. Such stimulation includes an increase in theabundance of the electron donor flavoprotein P450-oxidoreductase (Ogishima et al., 2003), cytochrome b5(Katagiri et al., 1995; Brock and Waterman, 1999), andpost-translational modification of the protein by cAMP-dependent phosphorylation of certain serine/threosineresidues (Zhang et al., 1995). Head kidney lyaseactivities are to low to be detected in reference femaleperch (data not shown), which neither supports norcontradicts an EDC acting selectively on the lyaseactivity. The determination of ovarian 17OHlase inreference and exposed female perch would be valuablein evaluating the hypotheses.

Head kidney 17OHlase activity is necessary for thesynthesis of glucocorticoids such as cortisol evidencedby the positive correlation between 17OHlase andcortisol over the whole period and the distinct peaksin both enzyme activity (this study) and cortisol levels(Noaksson et al., 2004) in late July.

Aromatase activity in SM female perch fromMolnbyggen was not significantly different fromreference fish at any time measured throughout thereproductive cycle. On the other hand aromatase activityin SIM fish from Molnbyggen was lower than in SMfish, which is consistent both with earlier investigations(Noaksson et al., 2001, 2003a,b) and with our con-clusion that the previously reported decrease in aro-matase activity in SM fish was probably due to down-regulation rather than inhibition of the enzyme. Our

647M. Linderoth et al. / Science of the Total Environment 366 (2006) 638–648

results indicate higher brain aromatase activity duringearly vitellogenesis which is similar to what Gelinas etal. (1998) found when investigating seasonal expressionof brain P450arom cDNA in goldfish (Carassiusauratus). Kumar et al. (2000) also found the highestpeak in P450arom transcripts in the ovary of femalechannel catfish (Ictalurus punctatus), at the onset ofovarian recrudescence, probably a response to theincrease in FSH recorded in the same fish (Kumar andTrant, 2004), while a lesser peak was observed duringmid-vitellogenesis. E2 levels in fish from both lakesrose from low levels after spawning, peaked in Octoberand were relatively constant until spawning (Noakssonet al., 2004). The observed positive linear relationshipbetween E2 plasma concentrations and aromataseactivity in Djursjön SM females was not found forMolnbyggen SM females, which implies that even whenP450arom is present in high amounts less E2 is formedin exposed fish, maybe as a result of low availability ofthe substrate T.

It is noteworthy that in July all individual fish, fromboth lakes, have about 30% oocytes in stage II and thatthey at this point cannot be divided into two groups. Atlater samplings, the frequency of oocytes in stage II islower both in SM and SIM. In SM fish the oocytespresumingly proceed into later maturational stages. InSIM fish there is a return to more stage I oocytes, withstage II oocytes reaching a maximum of 19% at laterparts of the reproductive cycle, indicating theirdegeneration. This suggests a similar development forall fish at least until the end of July, after which thetwo groups diverge and are easily distinguishable byearly September.

The presence of atretic oocytes in Molnbyggen SIMfemales but not inMolnbyggen and Djursjön SM femalesat all time-points examined could be taken as an in-dication of greater contaminant exposure. But this couldalso be a function of being SIM. In SM fish atresia seemsto be more common during the first part of vitello-genesis, while it is found all year round for SIM fish.

In conclusion, this study shows that an alteration oflyase activity could be affecting androgen homeostasisin Molnbyggen female perch, but the results are notunambiguous and should be investigated further. Otherpossible explanations, such as increased metabolism andexcretion of androgens, should be considered.

Acknowledgements

Funding for this study was provided in part by theSwedish Environmental Protection Agency, the munic-ipality of Leksand, Sweden and by the Sven and Lilly

Lawski foundation. We thank Olle Bergfors, RolfGamberg and co-workers at the Environmental officeof the municipality of Leksand for their technical andpractical support in collecting the fish. We also wish tothank Birgitta Liewenborg, Ulla Tjärnlund and GunÅkerman for their assistance in collecting and preparingthe tissue samples and Marsha Hanson for improvingthe English of this manuscript.

References

Ahmed S. The use of the novel substrate-heme complex approach inthe derivation of a representation of the active site of the enzymecomplex 17alpha-hydroxylase and 17,20-lyase. Biochem BiophysRes Commun 2004;316(3):595–8.

Baun A, Ledin A, Reitzel LA, Bjerg PL, Christensen TH. Xenobioticorganic compounds in leachates from ten Danish MSW landfills-chemical analysis and toxicity tests. Water Res 2004;38:3845–58.

Brock BJ, Waterman MR. Biochemical differences between rat andhuman cytochrome P450c17 support the different steroidogenicneeds of these two species. Biochemistry 1999;38:1598–606.

Gelinas D, Pitoc GA, Callard GV. Isolation of a goldfish brain cytochromeP450 aromatase cDNA: mRNA expression during the seasonal cycleand after steroid treatment. Mol Cell Endocrinol 1998;138:81–93.

Haarstad K, Borch H, Linderoth M. Biotests and guidelines forinvestigations of biochemical disorders based upon the character-ization of pollutants in landfill leaches. NT Tec Report 518, Espoo,Finland: Nordtest; 2002, 36 pp.

Hartmann S, Steinhart H. Simultaneous determination of anabolic andcatabolic steroid hormones in meat by gas chromatography massspectrometry. J Chromatogr B 1997;704:105–17.

Huang YS, Schmitz M, LeBelle N, Chang CF, Querat B, Dufour S.Androgens stimulate gonadotropin-II beta-subunit in eel pituitarycells in vitro. Mol Cell Endocrinol 1997;131:157–66.

Jeyasuria P, Roosenburg WM, Place AR. Role of P-450 aromatase insex determination of the diamondback terrapin, malaclemysterrapin. J Exp Zool 1994;270:95-111.

Katagiri M, Kagawa N, Waterman MR. The role of cytochrome b5 inthe biosynthesis of androgens by human P450c17. Arch BiochemBiophys 1995;317:343–7.

Kumar RS, Trant JM. Hypophyseal gene expression profiles of FSH-beta, LH-beta, and glycoprotein hormone-alpha subunits inIctalurus punctatus throughout a reproductive cycle. Gen CompEndocrinol 2004;136:82–9.

Kumar RS, Ijiri S, Trant JM. Changes in the expression of genesencoding steroidogenic enzymes in the channel catfish (Ictaluruspunctatus) ovary throughout a reproductive cycle. Biol Reprod2000;63:1676–82.

Lowry OH, Roseborough NJ, Farr AL, Randall RJ. Protein measure-ment with the folin phenol reagent. J Biol Chem 1951;193:265–75.

Montero M, LeBelle N, King JA, Millar RP, Dufour S. Differentialregulation of the 2 forms of gonadotropin-releasing-hormone(mGnRH and cGnRH-II) by sex steroids in the European femalesilver eel (Anguilla anguilla). Neuroendocrinology 1995;61:525–35.

Nagahama Y. 17 alpha,20 beta-dihydroxy-4-pregnen-3-one, a matu-ration-inducing hormone in fish oocytes: mechanisms of synthesisand action. Steroids 1997;62:190–6.

Noaksson E, Tjärnlund U, Bosveld ATC, Balk L. Evidence for endo-crine disruption in perch (Perca fluviatilis) and roach (Rutilus

648 M. Linderoth et al. / Science of the Total Environment 366 (2006) 638–648

rutilus) in a remote Swedish lake in the vicinity of a public refusedump. Toxicol Appl Pharmacol 2001;174:160–76.

Noaksson E, Linderoth M, Bosveld ATC, Norrgren L, Zebühr Y, BalkL. Endocrine disruption in brook trout (Salvelinus fontinalis)exposed to leachate from a public refuse dump. Sci Total Environ2003a;305:87-103.

Noaksson E, Linderoth M, Bosveld ATC, Balk L. Altered steroidmetabolism in several teleost species exposed to endocrinedisrupting substances in refuse dump leachate. Gen CompEndocrinol 2003b;134:273–84.

Noaksson E, Gustavsson B, Linderoth M, Zebühr Y, Broman D, BalkL. Gonad development and plasma steroid profiles by HRGC/HRMS during one reproductive cycle in reference and leachate-exposed female perch (Perca fluviatilis). Toxicol Appl Pharmacol2004;195:247–61.

Noaksson E, Linderoth M, Tjärnlund U, Balk L. Toxicological effectsand reproductive impairments in female perch (Perca fluviatilis)exposed to leachate from Swedish refuse dumps. Aquat Toxicol2005;75(2):162–77.

Ogishima T, Kinoshita JY, Mitani F, Suematsu M, Ito A. Identification ofouter mitochondrial membrane cytochrome b5 as a modulator forandrogen synthesis in Leydig cells. J Biol Chem 2003;278:21204–11.

Senthilkumaran B, Yoshikuni M, Nagahama Y. A shift in steroido-genesis occurring in ovarian follicles prior to oocyte maturation.Mol Cell Endocrinol 2004;215:11–8.

Sikka S, Swerdloff RS, Rajfer J. Assay system for simultaneousmeasurement of three steroidogenic enzyme activities in rat andhuman testis—effect of human chorionic gonadotropin. AnalBiochem 1985;149:72–86.

Treasurer JW, Holliday FGT. Some aspects of the reproductive biologyof perch Perca fluviatilis L.: a histological description of thereproductive cycle. J Fish Biol 1981;18:359–76.

Weibel ER. Stereological methods. Practical Methods for BiologicalMorphometry, vol. 1. London: Academic Press; 1979. 256 pp.

Yamamoto T. Studies on the formation of fish eggs: VII. The fate of theyolk vesicle in the oocyte of the smelt, Hypomesus japonicus,during vitellogenesis. Embryologia 1956;3:131–8.

Yeh SL, Kuo CM, Ting YY, Chang CF. The effects of exogenousandrogens on ovarian development and sex change in femaleorange-spotted protogynous grouper, Epinephelus coioides. Aqua-culture 2003;218:729–39.

Zhang LH, Rodriguez H, Ohno S, Miller WL. Serine phosphorylationof human P450c17 increases 17,20-lyase activity: implications foradrenarche and the polycystic ovary syndrome. Proc Natl Acad SciU S A 1995;92:10619–23.