1115

Environmental Toxicology and Chemistry, Vol. 22, No. 5, pp. 1115–1121, 2003Printed in the USA

0730-7268/03 $12.00 1 .00

EFFECTS OF AMMONIUM PERCHLORATE ON THE REPRODUCTIVE PERFORMANCEAND THYROID FOLLICLE HISTOLOGY OF ZEBRAFISH

REYNALDO PATINO,*†‡§ MELODY R. WAINSCOTT,‡ EMILIA I. CRUZ-LI,zz SARAVANAN BALAKRISHNAN,‡CYNTHIA MCMURRY,‡ VICKI S. BLAZER,†# and TODD A. ANDERSONzz

†U.S. Geological Survey, ‡Texas Cooperative Fish and Wildlife Research Unit, §Departments of Range Wildlife and Fisheries Managementand of Biological Sciences, zzInstitute of Human and Environmental Health, Texas Tech University, Lubbock, Texas 79409, USA

#National Fish Health Laboratory, Leetown Science Center, 1700 Leetown Road, Kearneysville, West Virginia 25430, USA

(Received 30 April 2002; Accepted 10 October 2002)

Abstract—Adult zebrafish were reared up to eight weeks in control water or in water containing ammonium perchlorate (AP) atmeasured perchlorate concentrations of 18 (environmentally relevant, high) and 677 ppm. Groups of eight females were pairedwith four males on a weekly basis to assess AP effects on spawned egg volume, an index of reproductive performance. All treatmentswere applied to four to five spawning replicates. At 677 ppm, spawn volume was reduced within one week and became negligibleafter four weeks. At 18 ppm, spawn volume was unaffected even after eight weeks. Also, perchlorate at 18 ppm did not affectpercentage egg fertilization. Fish were collected at the end of the exposures (677 ppm, four weeks; control and 18 ppm, eightweeks) for whole-body perchlorate content and thyroid histopathological analysis. Fish perchlorate levels were about one-hundredthof those of treatment water levels, indicating that waterborne perchlorate does not accumulate in whole fish. At 677 ppm for fourweeks, perchlorate caused thyroid follicle cell (nuclear) hypertrophy and angiogenesis, whereas at 18 ppm for eight weeks, itseffects were more pronounced and included hypertrophy, angiogenesis, hyperplasia, and colloid depletion. In conclusion, an eight-week exposure of adult zebrafish to 18 ppm perchlorate (high environmentally relevant concentrations) affected the histologicalcondition of their thyroid follicles but not their reproductive performance. The effect of 677 ppm perchlorate on reproduction maybe due to extrathyroidal toxicity. Further research is needed to determine if AP at lower environmentally relevant concentrationsalso affects the thyroid follicles of zebrafish.

Keywords—Perchlorate Thyroid Fish Reproduction Fertilization

INTRODUCTION

Ammonium perchlorate (AP) is a strong oxidizer that isused in the manufacture of rocket propellants, fireworks, andother industrial items [1]. Perchlorate ions derived from APand other perchlorate salts are stable and mobile in water andcan persist in the environment for many years [2]. Increasingreports of the presence of perchlorate in groundwater and sur-face water sites nationwide have led to concerns about itspotential effects on biotic resources and human health [2,3].A well-known biological effect of perchlorate is alteration ofthyroid gland function [4,5]. Elemental iodide is required forthe synthesis of thyroid hormones. Perchlorate competitivelyinhibits the uptake of iodide by thyroid follicles, thus inhibitingtheir ability to produce thyroid hormones [5]. The reductionin circulating levels of thyroid hormones causes an increasedproduction of thyroid-stimulating hormone by the pituitary,which in turn hyperstimulates the thyroid gland. Goiter andother abnormalities can develop if this condition persists fora prolonged period of time [5].

In fishes, growth [6] and reproduction [7] are believed tobe at least partly under the control of thyroid hormones. There-fore, these thyroid-dependent or thyroid-associated functionscould be affected by exposure to perchlorate. In fact, per-chlorate is commonly used to examine the role of thyroidhormones in animal development [8–12] and has also beenapplied for many years in the clinical treatment of certainhuman thyroid dysfunctions [4]. Recently, it was shown thatexposure to perchlorate at environmentally relevant concen-

* To whom correspondence may be addressed(reynaldopatino@ttu.edu).

trations can affect thyroid-dependent developmental events inamphibians [13,14]. However, the toxicological effects of per-chlorate on biological activities other than development, suchas reproduction, are not well understood. In a recent studywith rats, Rattus norvegicus, exposure to AP over two gen-erations seemed to have little effect on their reproductive suc-cess [15]. To our knowledge, similar information is unavailablefor nonmammalian vertebrates.

The principal objective of the present study is to determinethe effects of AP on reproductive performance of zebrafish,Danio rerio. Ammonium perchlorate was tested at a nominalwater concentration of 14 ppm, which is within the range ofenvironmentally relevant perchlorate concentrations reportedin some AP-contaminated sites [3], and at nominal concen-tration of 529 ppm, which corresponds to the estimated 5-dmedian lethal concentration (LC50) of zebrafish embryos/lar-vae (C. McMurry, R. Patino, Texas Tech University, Lubbock,TX, USA, unpublished data). Indices of reproductive perfor-mance measured in this study included packed egg volume(estimate of egg production) and egg fertilization rates (esti-mate of egg quality). Thyroid follicles were examined histo-logically at the conclusion of the exposure period as markerof perchlorate exposure and thyroidal effects.

METHODS

Fish and standard rearing conditions

All animal procedures used in this study were reviewed andapproved by the Texas Tech University Animal Care And UseCommittee (Lubbock, TX, USA). Three-month-old, adult ze-brafish (D. rerio) were obtained from Aquatic Research Or-

1116 Environ. Toxicol. Chem. 22, 2003 R. Patino et al.

ganisms (Hampton, NH, USA) and allowed a four-week ac-climation period before the spawning trials began. Groups ofeight females or 12 males were separately reared in 28-L aquar-ia containing 60 mg of Instant Oceant (Aquarium Systems;Mentor, OH, USA) per liter of deionized water (pH 6–8.5).Each aquarium was fitted with two internal biofilters consistingof 250-ml glass beakers filled with fiberglass and glass beadsand driven by air delivered through a glass serological pipette.Temperature was maintained at 27 to 288C (with submersible200-W glass heaters with plastic parts coated with silicon) anda photoperiod at 14:10-h light:dark (lights on at 0700). Fishwere fed adult brine shrimp twice daily to satiation, at whichtime they were also observed for general feeding behavior andhealth. Dissolved oxygen, water temperature, and salinity wererecorded once daily (YSIt model 85 meter; Yellow Springs,OH, USA). Uneaten food and other debris were siphoned witha glass pipette at the end of each day. One-half the watervolume (14 L) was removed and replaced with clean water(with salts, at 27–288C) twice weekly. Un-ionized ammoniain tank water was measured before each water change (Hachtspectrophotometer model DR/2000; Loveland, CO, USA).

Spawning procedure and basic experimental design

Fish were spawned in floating containers (;30 cm long by15 cm wide) made of silicon-coated plastic. The spawningcontainers consisted of an upper chamber for holding the fish(;5 cm deep) and a lower chamber for collecting the eggs.The two chambers were separated with a silicon-coated plasticmesh. Containers were placed in the appropriate female fishaquaria the evening before the scheduled spawning event, andin it were placed the eight females from the aquarium and fourmales from the same AP treatment regime. Fish were removedfrom the spawning container and returned to their originaltanks 3 to 4 h after lights on the following morning. Debriswas removed from the egg slurry, and packed egg volume wasdetermined by volume displacement in a graduated 5-ml glasscylinder.

Fish were spawned once weekly. They were allowed tospawn for two weeks prior to AP exposure. Six female tanks(eight fish each) and two male tanks (12 fish each) were pre-pared for each AP treatment regime. Spawn volume data werecollected from five female tank replicates for each treatment,with the sixth replicate (also spawned weekly) providing re-placement fish in the event of mortality. Spawn data werecollected for a period of eight weeks after onset of AP exposureor until spawning ceased.

Ammonium perchlorate exposures

Ammonium perchlorate (CAS 7790-98-9; Aldrich Chem-ical, Milwaukee, WI, USA) was added to the appropriate maleand female fish aquaria at the beginning of the exposure periodat nominal concentrations of 14 and 529 ppm. When waterwas replaced with freshwater during routine aquarium main-tenance, the appropriate volume of AP stock solution was add-ed directly to tank water to maintain the desired concentrations.Initially, water samples were taken for confirmation of per-chlorate concentration (by ion chromatography) before andafter each water change (see preceding paragraph). Once pre-liminary measurements indicated that the measured perchlo-rate levels were stable and close to nominal values, watersamples were subsequently taken only after the water changes.Water pH was measured daily during the AP exposures (Oak-tont pH meter; Gresham, OR, USA). It was somewhat unstable

and tended to decrease over time in the AP-treated aquaria. Itwas maintained close to a value of 7 by adding the appropriatevolume of 1-M NaOH solution to aquarium water as needed.

Percentage egg fertilization

In the period between four and eight weeks of treatment,eggs collected from control groups and from groups exposedto 14 ppm AP (nominal) were used to determine and comparenatural fertilization rates. Unfertilized eggs were regarded asthe fraction of dead (opaque) eggs in 1 ml of packed eggvolume. Observations were made with a dissecting scope 3 to4 h after spawn collection (7–8 h after lights on). Percentagefertilization per spawning replicate was estimated using theformula, 100 3 (total eggs 2 dead eggs) 4 total eggs. Eggswere maintained in petri dishes containing regular aquariumwater during processing.

Fish sampling

At the conclusion of the exposure period, fish were takenfrom their respective aquaria and allowed to swim (rinse) inthree consecutive baths of fresh (AP free) tank water beforebeing euthanized in a fourth bath of anesthesia (1 g/L MS-222 in fresh tank water). The total length of the rinse was 6to 10 min, and its purpose was to remove excess AP notspecifically associated with external surfaces (skin, oral cavity,gills). Fork length and weight were measured on all fish. Abouthalf the fish in each aquarium were frozen in liquid nitrogenfor perchlorate body burden analysis [16], and the other halfwere placed in Bouin’s fixative solution for histological anal-yses (see next paragraph) following an abdominal incision tofacilitate fixative penetration.

Thyroid histopathology

Whole fish remained in Bouin’s solution for 48 h at 48C,were rinsed in tap water for several hours, and were then placedin 70% ethanol until further processing. The head region wasseparated from the trunk and processed whole into paraffinblocks using standard procedures [17]. Like in other teleostfishes, thyroid follicles in zebrafish are not contained withina discrete gland but are found dispersed among the afferentbranchial arterioles of the throat region. Preliminary analysesindicated that the greatest abundance of thyroid follicles isfound in cross sections of the head where the posterior portionof the eyes overlaps with the anterior portion of the gill arches(R. Patino, Texas Tech University, Lubbock, TX, USA, un-published observations). Serial sections (6 mm) were thus col-lected from this region of the head and stained with hematox-ylin and eosin.

All histological observations were collected using the sametissue section for each fish. The section was randomly chosenfrom those available on slide preparations. because of the lossof follicular structure in some samples (see Results), the stan-dard practice of measuring epithelial height as index of hy-pertrophy could not be applied in this study. Thus, nuclearsize (as monitored by nuclear section area) was used instead.Nuclear size (hypertrophy) was assessed quantitatively bymeasuring the long and short diameters of at least 30 folliclecell nuclei from at least three different follicles at a total mag-nification of 31,000 or from three separate areas of the sectionin cases where follicular structure was unclear or absent. Thetwo diameters were used to estimate nuclear cross-section areabased on the formula of an ellipse (width 3 length 3 p 4 4).Hyperplasia was ranked positive or negative according to the

Effects of ammonium perchlorate on zebrafish Environ. Toxicol. Chem. 22, 2003 1117

presence or absence of follicle cell masses or irregular clusters(not forming follicles). Colloid depletion was ranked positiveor negative according to the presence or absence of reducedcolloid volume or collapsed follicles. Observations for hy-perplasia and colloid depletion were made at a total magni-fication of 3200.

Preliminary observations suggested the presence of greaternumbers of small blood vessels in association with the thyroidtissue of AP-treated fish. To examine if increased formationof blood vessels (angiogenesis) had occurred, the tissue sec-tions were placed under a 1-mm2 ocular grid at a total mag-nification of 3200 so as to include within the grid the greatestconcentration of thyroid tissue in the section. Angiogenesiswas ranked positive or negative according to the presence orabsence within the grid of more than five blood vessels whoserespective diameter fit within a 50 3 50-mm square (,2,500mm2). This cutoff for number and size of small vessels waschosen according to a preliminary screening of control fishindicating that it covered the values for these fish. Larger bloodvessels are normally associated with thyroid tissue and weredisregarded for this assessment.

Statistical analyses

For analysis of follicle cell nuclear size, the average of the30 nuclear area measurements per fish was designated as thefish value, and the average of all fish values obtained withina tank was designated as the tank value (unit of replication).Tank values were also used for the analysis of fork length andweight. This approach was applied to account for the possi-bility of tank effects. Packed egg volume and percentage fer-tilization and rank data (for each tank replicate) for hyperpla-sia, colloid depletion, and angiogenesis already represent tankvalues. Thus, the sample size for packed egg volume and per-centage fertilization corresponded to the number of female fishtanks and for thyroid histopathological analyses to the com-bined number of male and female tanks. Fish weights andlengths were analyzed separately for male and female tanks.

All analyses were conducted using the Statisticat for Win-dows 1998 software package (StatSoft, Tulsa, OK, USA).Packed egg volume, percentage fertilization, fork length,weight, and nuclear size data were analyzed with one- or two-way analysis of variance (ANOVA) or Student’s t tests asappropriate. For these parametric tests, homogeneity of vari-ances was assessed using the Cochran C statistic and Bartlettchi-square test, and appropriate data transformations were usedif needed to correct nonhomogeneities. Statistical differenceswere considered significant at overall a of 0.05. Nonpara-metric, planned comparisons were used for rank measurementsusing Kruskal–Wallis and median tests. The critical value ofa was adjusted according to the number of comparisons (a 50.025 for two planned comparisons).

RESULTS

Measured water and whole-fish perchlorate concentrations

The measured levels of perchlorate in tank water were closeto the nominal concentrations. The average (6 standard error[SE]) values for all water samples taken from all tank replicatesduring the exposure period were 6 6 1 ppb, 18 6 0.2 ppm,and 677 6 22 ppm for the nominal concentrations of 0(control), 14, and 529 ppm, respectively (detection limit, 1ppb). For whole-body perchlorate content analysis, tank valueswere obtained by pooling fish from the same tank before pro-

cessing. Including the two male tanks for each treatment re-gime, sample size thus was 6, 7, and 6 for the control, 18-ppm, and 677-ppm groups, respectively (see also next para-graph). Whole-body perchlorate contents (ppm) at the end ofthe respective exposure periods (see Results) were 0.0 6 0.0,0.389 6 0.154, and 7.771 6 1.863 (mean 6 SE) for the control(6 ppb), 18-ppm, and 677-ppm fish, respectively. In the lattertwo groups, the accumulation of perchlorate in whole body isequivalent to 2% and 1% of the treatment (water) concentra-tion.

General effects of AP exposure

Exposure to 677 ppm perchlorate caused acute mortalitiesin two of the six female tanks for this treatment (three to fourfish of the original eight fish per tank). This loss of fish requireda reduction in the number of spawning replicates from five tofour for analysis of packed egg volume since the fish in thetwo affected tanks were subsequently designated as replace-ment fish. (In all treatments, one tank replicate was reservedto replace normal mortalities in the other replicates during thecourse of the experiment in order to maintain the number ofexperimental females at n 5 8 per spawning tank replicate.)These fish resumed feeding within 1 to 2 d, and no differentialmortality was subsequently observed relative to the other treat-ments. However, their appetite was chronically reduced, andthey appeared to be less active throughout the exposure period.Mortality in the control and 18-ppm groups was random andrare throughout the experimental period, and no differenceswere observed in their appetite or behavior. One control rep-licate was lost after the four-week spawning measurement be-cause of a heater malfunction (the number of replicates foranalysis of reproductive performance in the control group wastherefore also reduced to four during the remaining five- toeight-week exposure period).

Because of the differential effects of AP at different con-centrations on spawning performance, the 677-ppm group wassampled four weeks earlier than the control and 18-ppm groups(see next paragraph). Under the guidance of this caveat, allthree AP exposure groups were included in the comparisonsof fish length and weight at the end of the exposure periods.Two-way ANOVA indicated an effect of sex (p 5 0.000003)but not of treatment (p 5 0.919570) or their interaction (p 50.645548) on fork length. Likewise, an effect of sex (p 50.000001) but not of treatment (p 5 0.825961) or their inter-action (p 5 0.901343) was observed on weight. Overall forklengths and weights (mean 6 SE) were 3.83 6 0.04 cm and0.63 6 0.02 g for males and 4.38 6 0.04 cm and 1.20 6 0.04g for females.

Reproductive effects of AP exposure

Spawned egg production became negligible four weeks af-ter the onset of exposure to 677 ppm perchlorate (Fig. 1A).This exposure regime was therefore discontinued at fourweeks. Exposures for the control and 18-ppm groups continueduninterrupted for four additional weeks (total, eight weeks).Because of the resulting unequal numbers of weekly obser-vations, packed egg volumes had to be separately comparedbetween the control and 677-ppm groups and the control and18-ppm groups. Two-way ANOVA (treatment 3 length ofexposure) indicated that 677 ppm perchlorate (p 5 0.000001)but not length of exposure or their interaction (p 5 0.275520and 0.201727, respectively) significantly affected weeklypacked egg volume (Fig. 1A). No effects of 18 ppm perchlo-

1118 Environ. Toxicol. Chem. 22, 2003 R. Patino et al.

Fig. 1. Effect of perchlorate at 6 ppb (control), 18 ppm, or 677 ppmon weekly (A) or cumulative (B) packed egg volume. Each pointrepresents the mean spawn volume (6 standard error) produced byfour to five spawning group replicates per treatment, and each replicateconsisted of eight adult females and four adult males. Weekly spawnvolumes differed between the control and 677-ppm group (p 50.000001) but not between the control and 18-ppm group (p . 0.05).Likewise, cumulative spawn volume differed (*) between the controland 677-ppm group (p 5 0.000218; four weeks) but not between thecontrol and 18-ppm group (p . 0.05; eight weeks). See text for detailsof statistical analysis.

Fig. 2. Effects of perchlorate on the histological structure of thyroidfollicles. Control fish (A) had ovoid follicles filled with colloid (Coll)and lined with squamous (lower-right follicle) or cuboidal (upper-leftfollicle) follicular epithelia. Fish treated with 18 ppm perchlorate (B–D) had follicles with thickened (columnar) epithelia (B), nuclear hy-pertrophy (compare A with B and C), follicle cell hyperplasia, in-creased vascularization (B–D), and colloid depletion or follicular col-lapse (B and D). Angiogenesis was also recognized within the follicles(asterisk in B). Signs of hyperplasia included the proliferation offollicle cell clusters in extravascular space (thin single arrows in Band C) and the growth of follicle cell masses projecting outward fromthyroid follicles (thin double arrows in B) or inward into the lumen(arrowhead in D). Follicular collapse (colloid depletion) ranged frompartial (B) to complete (block arrow in D). Thyroidal tissue oftenlacked clear follicular structure and organization (D). Double asterisksindicate some of the secondary circulation vessels in the sections. Bv5 blood vessel.

rate, length of exposure, or their interaction were observed (p5 0.191801, 0.490303, and 0.545029, respectively; Fig. 1A).An analysis of the cumulative egg volume produced by eachtank replicate was also conducted using Student’s t tests (Fig.1B). Again, significant differences were observed in the four-week cumulative egg volume between control and 677-ppmfish (p 5 0.000218) but not in the eight-week cumulativevolume between control and 18-ppm fish (p 5 0.123090; thecontrol replicate that was lost after four weeks was not includedin this cumulative volume analysis).

Data for percentage fertilization were subjected to an arc-sine transformation to achieve homogeneity of variances. Two-way ANOVA (treatment 3 length of exposure) showed noeffect of 18 ppm perchlorate, length of exposure (four to eightweeks), or their interaction on percentage fertilization relativeto control groups (p 5 0.971720, 0.317797, and 0.938499,respectively). The mean (SE) percentage fertilization rates atfour, five, six, seven, and eight weeks of treatment were 92(3), 94 (5), 87 (11), 94 (4), and 97 (3) for the control groupand 87 (5), 98 (1), 93 (4), 95 (1), and 97 (2) for the 18-ppmgroup. During the processing of eggs, it was noted that thenumber of eggs per milliliter of packed spawn volume wasslightly higher in the 18-ppm than in the control group (531.66 12.3 and 484.7 6 7.3 eggs/ml, respectively; mean 6 SE).Two-way ANOVA indicated effects of treatment (p 50.002140) but not of length of exposure (p 5 0.214555) ortheir interaction (p 5 0.413848) on this parameter.

Thyroidal effects of AP exposure

One-way ANOVA (factor, sex) of nuclear-size data usingtreatment regime as covariate showed no differences betweenmale and female tanks. Thus, male and female tanks weresubsequently combined for the statistical analysis of thyroidhistology. Because the effects of AP on thyroid condition de-

pend on the length of exposure [18], the different samplingschedules do not allow for direct comparisons of thyroid his-topathology between the 677-ppm and 18-ppm groups. Thus,treatment effects on thyroid histopathology were determinedusing the following planned comparisons: control fish and 18-ppm fish and control fish and 677-ppm fish (with critical aadjusted to 0.025). The latter comparison is based on the as-sumption that thyroid histology of control fish is unlikely tohave changed significantly between week 4 (677-ppm sam-pling) and week 8 (control sampling). As was the case forwhole-body perchlorate content analysis, a total of six, seven,and six tank replicates were used for the control, 18-ppm, and677-ppm treatment regimes, respectively.

Control fish had oval thyroid follicles of variable size filledwith colloid. The follicles were lined with squamous or cu-boidal follicle cells (Fig. 2A). Treatment with AP induced athickening of the follicle cell layer in those follicles with nor-mal structures. This appeared to be due to an increase in cellsize (Fig. 2B). Analysis of nuclear size showed that AP inducednuclear hypertrophy at perchlorate concentrations of 18 ppm(Student’s t test, p 5 0.000462) and 677 ppm (Student’s t test,p 5 0.000228) (Fig. 3). The number of small blood vesselswas significantly increased in the 18-ppm (Kruskal–Wallis, p5 0.0019; median, p 5 0.0020) and 677-ppm groups (Kruskal–Wallis, p 5 0.0125; median, p 5 0.0209) relative to the control

Effects of ammonium perchlorate on zebrafish Environ. Toxicol. Chem. 22, 2003 1119

Fig. 3. Effect of perchlorate at 6 ppb (control), 18 ppm, or 677 ppmon nuclear size of thyroid follicle cells. Long and short nuclear di-ameters were used to estimate the cross-section area of at least 30follicle cell nuclei per fish. These measurements were averaged toobtain a fish nuclear size value, and the values obtained for three tosix fish per tank were used to estimate tank values. Tank values wereused as unit of replication in the statistical analyses. Bars for eachtreatment regime represent the mean (6 standard error) of six to seventank replicates. Planned comparisons were made between the controland 18-ppm groups and the control and 677-ppm groups using Stu-dent’s t tests (critical a adjusted to 0.025). Statistically significantdifferences from the control are shown with an asterisk. See text fordetails of statistical analyses and results.

Fig. 4. Effect of perchlorate at 6 ppb (control), 18 ppm, or 677 ppmon the percentage of fish showing signs of thyroidal angiogenesis,hyperplasia, and colloid depletion. Bars for each treatment regimerepresent the mean percentages (6 standard error) of six to seventank replicates (see Fig. 3). Planned comparisons were made betweenthe control and the 18-ppm group and the control and 677-ppm groupusing Kruskal–Wallis and median tests (critical a adjusted to 0.025).Statistically significant differences from the control group are shownwith an asterisk. See text for details of statistical analyses and results.ND 5 not detected.

group (Fig. 4). In some of the samples from the 18-ppm group,the small vessels were difficult to define and quantify but wereobviously numerous (Fig. 2B and C). Small blood vessels werealso present within the follicular epithelium, most often at thebase of intact follicles, in fish exposed to either 18 ppm (Fig.2B) or 677 ppm perchlorate. Two out of four fish in one ofthe control female tank replicates had six and seven smallblood vessel counts, respectively, and were therefore rankedas positive for angiogenesis. The cutoff for a positive classi-fication had been set at five small vessels (see Methods) priorto the final evaluation. All other control fish in the remainingreplicates ranked negative for angiogenesis. Colloid depletionand accompanying follicular collapse were observed mostclearly in the 18-ppm group (Fig. 2B and D). The differencebetween the control and 18-ppm group was statistically sig-nificant (Kruskal–Wallis, p 5 0.0016; median, p 5 0.0020)but not that between the control and 677-ppm group (Kruskal–Wallis, p 5 0.0585; median, p 5 0.0455; the critical a forthese tests is 0.025) (Fig. 4). Follicular collapse in the 18-ppmgroup ranged from partial (Fig. 2B) to complete (Fig. 2D), andfollicular structure was often irregular and difficult to define(Fig. 2D). Hyperplasia was also most clearly observed in the18-ppm group (Fig. 2B to D). Hyperplasia was recognized asthe presence of follicle cell masses associated with the follicles(Fig. 2B and D) or as small clusters in the extravascular spacearound blood and secondary circulation vessels (Fig. 2B andC). The difference in hyperplasia between the control and 18-ppm group was statistically significant (Kruskal–Wallis, p 50.0015; median, p 5 0.0083) but not that between the controland 677-ppm group (Kruskal–Wallis, p 5 0.1397; median, p5 0.1213) (Fig. 4).

DISCUSSION

The results of the present study showed that AP-derivedperchlorate at high environmentally relevant concentrations

affected the histological condition of thyroid follicles but notthe reproductive performance of adult zebrafish. Although thehistological condition of thyroid follicles was greatly disruptedby an eight-week exposure to 18 ppm perchlorate (see laterdiscussion), packed egg volume (index of spawning success)and rate of natural egg fertilization (index of egg quality) werenot affected. Body size and weight also did not differ betweenthe control and 18-ppm group (sorted by sex), which is con-sistent with the conclusion that 18 ppm perchlorate did notaffect the general health of zebrafish during the eight-weekexposure period. This conclusion is consistent with the resultsof a recent study with rats, where AP exposure over two gen-erations was reported to have little effect on reproductive per-formance at dosages that affected the thyroid gland [15].

More eggs per unit volume were apparent in spawns fromthe 18-ppm group than the control group. Although no directmeasurements of egg size were taken, this observation suggeststhat eggs from the 18-ppm group were smaller. Since eggquality in some teleosts may correlate with egg size [19], itis conceivable that embryo or larval survival could be com-promised by maternal zebrafish exposures to 18 ppm perchlo-rate. Direct measurements of egg size and of embryo and larvalsurvival after maternal exposure to AP are needed to clarifythis question.

The role of thyroid hormones in vertebrate reproduction isunclear. Chemically induced hypothyroidism in prepubertalrats appeared to inhibit ovarian follicle development [20], andin adult rats it disrupted reproductive cycles [21]. Conversely,in a two-generation study with rats, AP-induced hypothyroid-ism did not affect reproductive performance [15]. Also, ad-ministration of exogenous thyroid hormone inhibited gonad-otropin-dependent ovarian follicle development in thyroidec-tomized immature rats [22] and suppressed gonadal devel-opment in male and female immature chicken, Gallus gallus[23] and ovarian follicle growth in the frog Rana cyanophlyctis[24]. In cows, Bos taurus, ovarian function was not affected

1120 Environ. Toxicol. Chem. 22, 2003 R. Patino et al.

by either hyper- or hypothyroid conditions [25]. In some teleostfishes, a positive effect of thyroid hormones on the gonadalresponsiveness to reproductive hormone stimuli has been re-ported (reviewed in [7]). Also, hyperthyroidism induced byadministration of exogenous thyroid hormone enhanced ovar-ian follicle growth in guppy, Poecilia reticulata [26], andaccelerated spermatogenesis in prepubertal carp, Cyprinuscarpio [27]. However, other studies with teleosts have failedto show an effect of thyroid hormones on the gonadal sensi-tivity to reproductive stimuli [28,29]. Further, seasonal or de-velopmental hormone profiles that are inconsistent with a rolefor thyroid hormones in the regulation of gonadal developmenthave been reported [30,31]. If the remarkable structural dis-ruption of thyroid follicles induced by 18 ppm perchlorate (seelater discussion) were related to an impairment in thyroid hor-mone production, this finding would suggest that thyroid hor-mones do not play a major role in the regulation of repro-ductive function in adult zebrafish. However, we did not mea-sure thyroid hormones in the present study, and thus this sce-nario remains speculative until the effects of perchlorate onthyroid hormone production are directly determined in ze-brafish.

Perchlorate at the very high concentration of 677 ppm sup-pressed spawning activity, but this response may have beendue primarily to toxic effects of AP at sites other than thethyroid follicles. Some acute mortality was observed in the677-ppm perchlorate group on first addition of AP to the tankwater, and the appetite and activity level of the treated fishwere chronically depressed during the period of exposure. En-vironmental stress and nutrient deprivation, as experienced bythe 677-ppm fish of this study, are known to suppress repro-ductive performance in fishes [19]. The 677-ppm fish did notappear to have lost weight at the conclusion of their four-weekexposure (when cautiously compared to control fish sampledfour weeks later), but this observation may be explained bythe energy saved through the almost complete shutdown ofreproductive activity and gamete production. Either perchlo-rate or ammonium ions could be responsible for the repro-ductive toxicity of AP at very high concentrations. Perchloratehas extrathyroidal, direct effects on animal tissues [32–34],and un-ionized ammonia at high concentrations is deleteriousto fish health [35]. The observations that exposure to 677 ppmperchlorate for four weeks is less disruptive of thyroid func-tional structure than exposure to 18 ppm for eight weeks andthat the reproductive toxicity of 18 ppm was negligible areconsistent with the conclusion that the severe reproductivetoxicity of 677 ppm perchlorate is mediated by nonthyroidalmechanisms.

Zebrafish exposed to AP underwent changes in thyroid his-tology consistent with those reported in thyroid follicles ofother animals exposed to perchlorate [4,5,18]. Major changesobserved included follicle cell (nuclear) hypertrophy, hyper-plasia, angiogenesis, and colloid depletion. One notable aspectof this study with zebrafish is the high degree of thyroidalvascularization induced by AP. Especially in fish from the 18-ppm treatment, thyroidal tissue sections sometimes appearedto be mostly composed of small blood vessels and secondarycirculation vessels (the latter are similar to lymphatic vesselsof other vertebrates) and unstructured clusters of follicle cellsfilling the extravascular space. Increased production of pitui-tary thyroid-stimulating hormone, such as when the levels ofthyroid hormones are reduced, causes angiogenesis in themammalian thyroid gland [36]. Also, angiogenesis in the mam-

malian thyroid has been correlated with hyperplastic and neo-plastic follicle cell transformations [36]. In the present studywith zebrafish, angiogenesis in the 18-ppm/eight-week groupwas accompanied by the appearance of follicle cell massesassociated with thyroid follicles, irregular follicular patterns,and unstructured clusters of follicle cells. It is uncertain wheth-er these changes in zebrafish thyroid follicles indicate the onsetof neoplasia, but they share some similarities with those de-scribed during the progression from follicular hyperplasia toadenoma and carcinoma in the mammalian thyroid [37].

The effects of AP on thyroid follicles depend on the lengthof exposure [18]. This observation may explain why exposureto 18 ppm perchlorate for eight weeks had a much greatereffect on follicular structure than exposure to 677 ppm (40-fold higher concentration) for four weeks. It is thus conceivablethat perchlorate at concentrations much lower than 18 ppmcould also affect thyroid function if the exposure periods wereextended for a few more weeks. Perchlorate at a nominal waterconcentration of 500 ppm inhibited iodide accumulation bythe thyroid region of young zebrafish exposed from day 12 today 33 postfertilization, but the condition of the thyroid fol-licles was not examined [10]. Information about the effects ofAP at lower concentrations than used in the present study,such as those more commonly reported in the environment, isneeded to assess the ecological risks associated with perchlo-rate contamination of the aquatic habitat.

Perchlorate levels in whole body were about two orders ofmagnitude lower than in treatment water. Given the length ofthe exposure regimes (four to eight weeks), it seems likelythat these levels represent steady-state equilibria between wa-ter and fish. Thus, perchlorate does not seem to accumulatein whole-body zebrafish. However, the present results do notrule out the possibility of tissue-specific accumulation of per-chlorate (e.g., in thyroid follicles).

CONCLUSIONS

An eight-week exposure to perchlorate at high environ-mentally relevant levels (18 ppm) does not appear to affectzebrafish reproduction, but it greatly disrupts thyroid histo-logical condition. Further research is needed to determine theinteraction between perchlorate concentration and length ofexposure as they affect the normal structure and function ofzebrafish thyroid follicles.

Acknowledgement—We thank Cathy Bens and Ronald Kendall fortheir logistical assistance. James Carr, Chris Theodorakis, and SandeepMukhi provided critical reviews of an earlier version of this manu-script. This research was supported by funding from the U.S. De-partment of Defense through the Strategic Environmental Researchand Development Program under a cooperative agreement with theU.S. Air Force, Institute for Environmental Safety and OccupationalHealth, Brooks AFB, Texas, and from the U.S. Geological Surveythrough the Texas Cooperative Fish and Wildlife Research Unit, Lub-bock, Texas. The views and conclusions contained herein are thoseof the authors and should not be interpreted as necessarily representingthe official policies or endorsements, either expressed or implied, ofany office of the U.S. government.

REFERENCES

1. Von Burg R. 1995. Perchlorates. J Appl Toxicol 15:237–241.2. Urbansky ET. 1998. Perchlorate chemistry: Implications for anal-

ysis and remediation. Bioremediation J 2:81–95.3. Smith PN, Theodorakis CW, Anderson TA, Kendall RJ. 2001.

Preliminary assessment of perchlorate in ecological receptors atthe Longhorn Army Ammunition Plant (LHAAP), Karnack, Tex-as. Ecotoxicology 10:305–313.

Effects of ammonium perchlorate on zebrafish Environ. Toxicol. Chem. 22, 2003 1121

4. Wolff J. 1998. Perchlorate and the thyroid gland. Pharmacol Rev50:89–105.

5. Capen CC. 2001. Toxic responses of the endocrine system. InKlaassen CD, ed, Casarett and Doull’s Toxicology: The BasicScience of Poisons. McGraw-Hill, New York, NY, USA, pp 711–759.

6. Power DM, Llewellyn L, Faustino M, Nowell MA, BjornssonBT, Einarsdottir IE, Canario AVM, Sweeney GE. 2001. Thyroidhormones in growth and development of fish. Comp BiochemPhysiol C 130:447–459.

7. Cyr DG, Eales JG. 1996. Interrelationships between thyroidal andreproductive endocrine systems in fish. Rev Fish Biol Fisher 6:165–200.

8. Michael MI, Aziz FK. 1976. Effect of sodium perchlorate on therestoration of the limb regenerative ability in a metamorphic stageof Bufo regularis Reuss. Folia Biol Krakow 24:309–315.

9. Miranda LA, Paz DA, Dezi RE, Pisano A. 1995. Immunocyto-chemical and morphometric study of TSH, PRL, GH, and ACTHcells in Bufo arenarum larvae with inhibited thyroid function.Gen Comp Endocrinol 98:166–176.

10. Brown DD. 1997. The role of thyroid hormone in zebrafish andaxolotl development. Proc Natl Acad Sci USA 94:13011–13016.

11. Manzon RG, Eales JG, Youson JH. 1998. Blocking of KClO4-induced metamorphosis in premetamorphic sea lampreys by ex-ogenous thyroid hormones (TH): Effects of KClO4 and TH onserum TH concentrations and intestinal thyroxine outer-ring de-iodination. Gen Comp Endocrinol 112:54–62.

12. Saito M, Seki M, Amemiya S, Yamasu K, Suyemitsu T, IshiharaK. 1998. Induction of metamorphosis in the sand dollar Peronellajaponica by thyroid hormones. Dev Growth Differ 40:307–312.

13. Goleman WL, Carr JA, Anderson TA. 2002. Environmentallyrelevant concentrations of ammonium perchlorate inhibit thyroidfunction and alter sex ratios in developing Xenopus laevis. En-viron Toxicol Chem 21:590–597.

14. Goleman WL, Urquidi LJ, Anderson TA, Smith EE, Kendall RJ,Carr JA. 2002. Environmentally relevant concentrations of am-monium perchlorate inhibit development and metamorphosis inXenopus laevis. Environ Toxicol Chem 21:424–430.

15. York RG, Brown WR, Girard MF, Dollarhide JS. 2001. Two-generation reproduction study of ammonium perchlorate in drink-ing water in rats evaluates thyroid toxicity. Int J Toxicol 20:183–197.

16. Anderson TA, Wu TH. 2002. Extraction, cleanup, and analysisof the perchlorate ion in tissue samples. Bull Environ ContamToxicol 68:684–691.

17. Luna LG. 1992. Histopathological Methods and Color Atlas ofSpecial Stains and Artifacts. American Histolabs, Gaithersburg,MD, USA.

18. Fernandez Rodriguez A, Galera Davidson H, Salguero VilladiegoM, Moreno Fernandez A, Martin Lacave I, Fernandez Sanz J.1991. Induction of thyroid proliferative changes in rats treatedwith antithyroid compound. Anal Histol Embryol 20:289–298.

19. Bromage NR. 1995. Broodstock management and seed quality—General considerations. In Bromage NR, Roberts RJ, eds, Brood-stock Management and Egg and Larval Quality. Blackwell Sci-ence, Osney Mead, Oxford, UK, pp 1–24.

20. Dijkstra G, de Rooij DG, de Jong FH, van den Hurk R. 1996.Effect of hypothyroidism on ovarian follicular development, gran-

ulosa cell proliferation and peripheral hormone levels in the pre-pubertal rat. Eur J Endocrinol 134:649–654.

21. Armada-Diaz L, Carvalho JJ, Breitenbach MM, Franci CR, MouraEG. 2001. Is the infertility in hypothyroidism mainly due to ovar-ian or pituitary functional changes? Braz J Med Biol Res 34:1209–1215.

22. Tamura K, Hatsuta M, Watanabe G, Taya K, Kogo H. 1998.Inhibitory regulation of inhibin gene expression by thyroid hor-mone during ovarian development in immature rats. BiochemBiophys Res Commun 242:102–108.

23. Majeed SK, Injidi MH, Forbesa JM. 1984. Effects of thyroxineon the reproductive organs and growth of young chickens. Ex-perientia 40:281–283.

24. Saidapur SK, Kupwade VA. 1993. Effect of thyroxine, prolactinand adrenalin on the gravid ovaries of the frog Rana cyanophlyctis(Schn.). Indian J Exp Biol 31:401–405.

25. De Moraes GV, Vera-Avila HR, Lewis AW, Koch JW, NeuendorffDA, Hallford DM, Reeves JJ, Randel RD. 1998. Influence ofhypo- or hyperthyroidism on ovarian function in Brahman cows.J Anim Sci 76:871–879.

26. Lam TJ, Loy GL. 1985. Effect of L-thyroxine on ovarian devel-opment and gestation in the viviparous guppy, Poecilia reticulata.Gen Comp Endocrinol 60:324–330.

27. Timmermans LP, Chmilevsky DA, Komen H, Schipper H. 1997.Precocious onset of spermatogenesis in juvenile carp (Cyprinuscarpio L., Teleostei) following treatment with low doses of L-thyroxine. Eur J Morphol 35:344–353.

28. Mylonas CC, Sullivan CV, Hinshaw JM. 1994. Thyroid-hormonesin brown trout (Salmo trutta) reproduction and early development.Fish Physiol Biochem 13:485–493.

29. Reddy PK, Renaud R, Leatherland JF. 1999. Effects of cortisoland triiodo-L-thyronine on the steroidogenic capacity of rainbowtrout ovarian follicles at two stages of oocyte maturation. FishPhysiol Biochem 21:129–140.

30. Pavlidis M, Dessypris A, Christofidis I. 1991. Seasonal fluctua-tions in plasma thyroid hormones, in two strains of rainbow trout(Oncorhynchus mykiss), during the first reproductive and secondreproductive cycle—Relation with their photoperiodically alteredspawning time. Aquaculture 99:365–385.

31. Sohn YC, Yoshiura Y, Kobayashi, M, Aida K. 1999. Seasonalchanges in mRNA levels of gonadotropin and thyrotropin subunitsin the goldfish, Carassius auratus. Gen Comp Endocrinol 113:436–444.

32. Yen MH, Chow SY, Chang LR. 1973. Perchlorate and renal ex-cretory functions. Chin J Physiol 21:205–214.

33. Moonga BS, Datta HK, Bevis PJ, Huang CL, MacIntyre I, ZaidiM. 1991. Correlates of osteoclast function in the presence ofperchlorate ions in the rat. Exp Physiol 76:923–933.

34. Huang CL. 1998. The influence of perchlorate ions on complexcharging transients in amphibian striated muscle. J Physiol 506:699–714.

35. Wedemeyer GA. 1996. Physiology of Fish in Intensive CultureSystems. Chapman & Hall, New York, NY, USA, p 232.

36. Ramsden JD. 2000. Angiogenesis in the thyroid gland. J Endo-crinol 166:475–480.

37. Hardisty JF, Boorman GA. 1990. Thyroid gland. In Pathology ofthe Fischer Rat: Reference and Atlas. Academic, New York, NY,USA, pp 519–532.