Evaluating the Effects of Endocrine Disruptors on Endocrine Function duringDevelopmentRobert Bigsby,1 Robert E. Chapin,2 George P. Daston,3 Barbara J. Davis,2 Jack Gorski,4 L. Earl Gray,5 Kembra L.Howdeshell,6 R. Thomas Zoeller,7 and Frederick S. vom Saal6'Indiana University School of Medicine, Indianapolis, Indiana USA; 2National Institute of Environmental Health Sciences, Research TrianglePark, North Carolina USA; 3Procter & Gamble Company, Cincinnati, Ohio USA; 4University of Wisconsin, Madison, Wisconsin USA; 5U.S.Environmental Protection Agency, Research Triangle Park, North Carolina USA; 6University of Missouri, Columbia, Missouri USA; 7Universityof Massachusetts, Amherst, Massachusetts USA

The major concerns with endocrine disruptors in the environment are based mostly on effects thathave been observed on the developing embryo and fetus. The focus of the present manuscript ison disruption of three hormonal systems: estrogens, androgens, and thyroid hormones. Thesethree hormonal systems have been well characterized with regard to their roles in normaldevelopment, and their actions during development are known to be perturbed by endocrine-disrupting chemicals. During development, organs are especially sensitive to low concentrations ofthe sex steroids and thyroid hormones. Changes induced by exposure to these hormones duringdevelopment are often irreversible, in contrast with the reversible changes induced by transienthormone exposure in the adult. Although it is known that there are differences in embryonic/fetal/neonatal versus adult endocrine responses, minimal experimental information is available to aidin characterizing the risk of endocrine disruptors with regard to a number of issues. Issuesdiscussed here include the hypothesis of greater sensitivity of embryos/fetuses to endocrinedisruptors, irreversible consequences of exposure before maturation of homeostatic systems andduring periods of genetic imprinting, and quantitative information related to the shape of thedose-response curve for specific developmental phenomena. Key words: androgen, development,embryo, endocrine disruptors, estrogens, fetus, thyroid. - Environ Health Perspect 107(suppl 4):613-618(1999).http://ehpnetl.niehs.nih.gov/docs/1999/suppl-4/613-618bigsby/abstract.html

An endocrine disruptor has recently beendescribed as "an exogenous chemical sub-stance or mixture that alters the structure orfunction(s) of the endocrine system andcauses adverse effects at the level of the organ-ism, its progeny, populations, or subpopula-tions of organisms, based on scientificprinciples, data, weight-of-evidence, and theprecautionary principle" (1). To address con-cerns of potential effects of endocrine disrup-tors, the National Institute of EnvironmentalHealth Sciences and other co-sponsors held aworkshop to characterize the effects fromenvironmental exposures to endocrine dis-ruptors on human health. The workshopprovided a forum to discuss methods anddata needed to improve risk assessments ofendocrine disruptors. This article is theproduct of one of six subgroups from thatworkshop. It is based on the work group'sdiscussion of a set of questions provided bythe organizing committee of the workshop.The following is a list of questions posed tothe working group on endocrine functionduring development that served as the basisfor the information discussed in this report.* What should be included in a baseline

model to describe quantative relation-ships among the processes controllingnormal development?

* How do perturbations at critical stages ofdevelopment lead to adverse effects, e.g.,

impaired reproductive function, neurologiceffects, cancer?

* How can these changes be quantified?* By what mechanisms do endocrine dis-

ruptors perturb endocrine function dur-ing development and alter risks fromnormal levels of endogenous hormones?

* What are the principal mechanisms bywhich endocrine disruptors are thought toact on the developing reproductive tract?

* Are there effective repair mechanismsoperating during development to reducethe effects of endocrine disruptors?

* Are there adequate/relevant animal modelsfor evaluating potential human effects?We focused on the regulatory processes of

normal development and on how exposure tolow doses (that is, doses encountered in theenvironment) of endocrine disruptors at criti-cal stages of development can lead to adversehealth effects. We also discussed areas whereinformation is needed to permit better evalu-ation of the risks of endocrine disruptors.

The authors feel that additional research infive areas is essential: a) mechanisms of normaldevelopment; b) differences of endocrine dis-ruptor effects between embryo/fetus/neonateand adult; c) mechanisms of endocrine disrup-tion; d) dose-response assessment involvingexamination over a wide range of doses, fromlevels encountered in the environment throughdoses that produce acute toxicity; and e) the

design of screens to accurately predict uniquedevelopmental effects.

Mechanisms of NormalDevelopmentBasic information is needed on the normalmolecular, cellular, and physiologic develop-mental mechanisms perturbed by alteredendocrine function during organogenesis(2-4). Some of the resultant developmentalchanges may not be detectable until later inlife (5). Also, knowledge acquired through thestudy of developmental perturbation is likelyto lead to a better understanding of normalprocesses occurring during that time in life.

Information is required for both humansand other animals. Knowledge of mechanismsaffected by endocrine perturbation due eitherto congenital defects, induding experimentalgene knockout systems, or to application ofsynthetic or naturally occurring endocrine-mimicking compounds would be useful.We recognize that development is

epigenetic, which refers to changes in geneactivity during development that are mediatedby environmental (chemical) signals (6).Autocrine, paracrine (such as growth factors),and endocrine (such as steroid) signals coordi-nate the direction of differentiation of tissuesduring critical periods in development. Thedifferentiation of organs thus involves a com-plex cascade of signals whose action is depen-dent on being released at precise times andwithin a specific dose range. Coordination ofthese processes depends on the transcriptionof genes coding for these signaling moleculesand their receptors at appropriate times andappropriate rates (7-9).

In the field of endocrine disruption,particular regulatory emphasis has been placedon processes or tissues affected by estrogens,

This report was developed at the Workshop onCharacterizing the Effects of Endocrine Disruptors onHuman Health at Environmental Exposure Levels held11-13 May 1998 in Raleigh, North Carolina.

Address correspondence to F.S. vom Saal,1 14Lefevre Hall, Division of Biological Sciences, Universityof Missouri, Columbia, MO 65211. Telephone: (573)882-4367. Fax: (573) 884-5020. E-mail: vomsaal@mis-souri.edu

Received 25 September 1998; accepted 18 May1999.

Environmental Health Perspectives * Vol 107, Supplement 4 * August 1999 613

BIGSBY ET AL.

androgens, and thyroid hormones, as well astheir antagonistic analogs. Organ systemsresponsive to the sex steroids include the maleand female reproductive organs, the centralnervous system, and the immune system,whereas thyroid hormone affects most tissues.The work group focused only on these threehormone groups. This decision was based onthe extensive literature that is availableregarding these developmentally importantsignaling molecules. The current identifica-tion of particular endocrine-disrupting chem-icals as mimics or antagonists of the sexsteroid (estrogen and androgen) and thyroidhormones, and their respective functions,facilitates the work group's goal toward anunderstanding of the mechanism of action ofthese known endocrine disruptors.

Quantitative aspects of these threecomponents of the endocrine system must becarefully considered to determine if certaindevelopmental events and tissues are particu-larly sensitive to the test compounds. Withspecific regard to the dose issue, a criticalquestion that remains to be resolved iswhether higher doses may actually inhibitsome responses that are stimulated by muchlower doses, causing what has been describedas an inverted U-shaped dose-response curve(7). To understand this phenomenon, thenormal concentration range for hormonesbeing disrupted must be characterized withregard to a variety of responses (Figure 1).

CD)o j ADI NOEL

0 1 2 3Untreated Dose

H - Estrogenic activity

Figure 1. An inverted-U (nonmonotonic) dose-responsefunction associated with an increase in total estrogenicactivity in the blood. As shown here, there is already aresponse occurring at zero dose of exogenous estrogenicendocrine disruptor, due to the presence of endogenousestrogen that is circulating in blood at a concentrationabove the threshold for the response [based on data inSheehan et al. (38)]. On the basis of the assumption of amonotonic dose-response function, which may not be avalid assumption for endocrine disruptors, the conclu-sion would be that dose 1 represents a threshold dosebelow which no effect occurs (the response is at thecontrol level), and lower doses are then not tested. Thelabeling of dose 1 in this figure as the NOEL (noobserved effect level) on the basis of testing three highdoses is only valid if the dose-response function doesnot form an inverted U. Similarily, the use of the NOEL toestimate an acceptable daily intake (ADI) dose would beinvalid if there were an inverted -U dose-response curve.The figure is based on data for prostate weight in adultmale mice following exposure to different doses of estro-genic chemicals during fetal life (7).

Examples of additional information neededon normal development indude the effects ona) spatial (9) and chronologic patterns ofexpression of relevant nuclear receptors(including isoforms) and of genes known tointegrate cellular processes of development,such as the homeobox genes Hox, Wnt, Pit-i,Pou, etc., and b) hormone-synthesizing andhormone-catabolizing enzymes after treatmentwith hormone analogs or endocrine-disruptingchemicals (4,8). Quantitative analyses of suchresponses should be stressed in an attempt toallow formulation of predictive hypotheses.

Differences between theEmbryo/Fetus/Neonateand AduftDuring the differentiation of reproductiveorgans, hormones, growth factors, and otherendogenous chemical mediators regulategene expression and direct differentiation(10). One marked difference between expo-sure to endocrine disruptors during criticalperiods in development versus during adult-hood is the irreversibility of an effect duringdevelopment (9,11,12).

Evidence indicates that changes in concen-trations of androgen and estrogen (two hor-mones involved in differentiation of thereproductive organs) result in permanentchanges in cell function. For example, thehigher circulating levels of testosterone (by2-3 ng/ml) in male mouse fetuses relative tofemale fetuses result in the differentiation oftissue in the cranial region of the urogenitalsinus into prostatic tissue as opposed to vagi-nal tissue. Many other differences betweenmales and females are also mediated by thissmall sex difference in testosterone (12). Inaddition, a small increase in total circulatingestradiol (about 50 pg/ml) permanentlyaltered prostate size in mice (7). It is thusplausible that disruption of the action of estro-gen or androgen during critical periods canlead to permanent alterations in the develop-ment of reproductive organs and other tissueswith receptors for these hormones. Some ofthese effects can be unique to the time duringdevelopment in which the hormonal alter-ation occurred (11). This contrasts with cyclicchanges in hormones that occur normally inadult females during the menstrual cycle thatdo not produce permanent effects.

Although development is a period ofchange, there are regulatory processes involvedin developmental processes, such as changes inplasma-binding proteins during pregnancy,that alter bioavailability of circulating steroids(13,14). However, the principle of homeo-stasis, which implies a level of constancy, isdifficult to apply during development.

Diczfalusy (15) initiated the concept ofthe maternal-placental-fetal unit. It is nowaccepted that pregnancy in mammals

represents the interaction of three endocrinesystems, all ofwhich are changing throughoutpregnancy. The vast differences in gestationlength, hormone production, and the degreeof intimacy of fetal-maternal blood suppliesrepresent important barriers to understandingthe complex interactions between these sys-tems in one species on the basis of informa-tion obtained in another. Little is knownconcerning the regulation of protein andsteroid hormones by the placenta in mostspecies, and this lack of information limitspredictions concerning the effects ofendocrine-disrupting chemicals on the func-tioning of the maternal-fetal-placental unit.What is known, however, is that regardless ofthe species, outcomes of endocrine mani-pulations in adults are not predictive ofendocrine changes in fetuses (11).

Mechanisms of EndocrineDisruptionNumerous mechanisms of endocrine functionhave been disrupted by endocrine disruptors.Consideration of these end points allows theidentification of end point measures thatcan be used in specific screens and tests. Endpoints for the three hormonal systems that arethe focus here are also the focus ofnew regula-tions currently being developed by the U.S.Environmental Protection Agency (U.S. EPA)under congressional mandate and named theEndocrine Disruptor Screening and TestingProgram (1). Examples of end points includethe following.

Steroids (Estrogen/Androgen)Receptor binding and function. Thisincludes both activation and inhibition and isan important mechanism of endocrinedisruption (14,16).

Steroid synthesis inhibition. This is awell-known mechanism by which steroid(estrogen/androgen) hormone systems aredisrupted (17).

Plasma transport and rate ofmetabolismand ckarance. An example is the free concen-tration of steroid (not bound to plasma-bind-ing proteins) in blood, which changesdramatically between development and adult-hood in rodents (18). Differences betweenendogenous steroids and endocrine disruptorsin binding to plasma-binding proteins can dra-matically alter the potency of endocrine dis-ruptors compared to the hormone, such asestradiol, being mimicked by the endocrinedisruptor (19). Endocrine disruptors mayrequire metabolic activation in order to inter-act with one of these mechanisms (16).

ThyroidReceptor binding andfunction. Currentlythere are no reports of xenobiotics binding tothe thyroid hormone receptor.

Environmental Health Perspectives a Vol 107, Supplement 4 - August 1999614

ENDOCRINE FUNCTION DURING DEVELOPMENT

Synthesis inhibitors. Several classes ofendocrine disruptors fall into this category,including compounds that block thyro-peroxidase (TPO), iodide uptake, and thedeiodinases (20).

Plasma transport and rate ofmetabo-lism and clearance. Thyroid hormone mustbe carried through the blood on serum pro-teins. Some endocrine-disrupting chemicals(polychlorinated biphenyls and dioxin)inhibit thyroid hormone binding to plasmatransport proteins, resulting in more rapidclearance and reduced thyroid hormonelevels (21).

Several types of endogenous hormonesand endocrine disruptors have been foundthat interact with more than one componentof the endocrine system. An example involvescompounds, such as genistein in soy, that areweak estrogens but that also block TPO (20).Another example is that at a higher thanphysiologic concentration, estradiol binds toandrogen receptors (22). Similarly, someestrogenic endocrine disruptors, such as thebis-hydroxy metabolite of the insecticidemethoxychlor, also bind to the androgenreceptor (23,24). Endocrine disruptors thatbind to steroid receptors such as p,p'-DDE(the persistent in vivo metabolite of the insec-ticide DDT) thus show the highest affinityfor one steroid receptor (in this case, andro-gen receptors) but also show a lower bindingaffinity for other receptors (estrogen recep-tors) (25,26). As the dose of p,p'-DDE ormethoxychlor increases, they will thus bind tomultiple receptors. As a result, the change insome end point to increasing doses of anendocrine disruptor may reflect its action ondifferent components of the endocrine sys-tem, and each componeni may contribute toa composite dose response. For this reason,the response to a dose on the high end of thedose-response curve may be qualitatively dif-ferent from and may not be a reliable predic-tor of the response at much lower doses.

Endocrine disruptors that act to disruptthe estrogen, androgen, and thyroid systemshave been the focus of the design of screensand tests for detecting potential endocrine-disrupting chemicals (25,27,28). However,we know that these mechanisms do not repre-sent the full range of potential endocrine dis-ruption. Therefore, it is essential to recognizethat endocrine disruptors may interfere withhormone actions in ways that would not beidentified in the assays currently contained inthe new U.S. EPA testing program (1).Moreover, there are many potential mecha-nisms by which endocrine disruptors couldproduce nonlinear dose-response curves (29).

Dose-Response AssessmentThe dose issue refers to the application of theprevious concepts to characterize the full

spectrum of the dose-response curve forendocrine disruptors. The issues are asfollows: first, are current risk assessment pro-cedures adequately evaluating the adverseeffects of endocrine disruptors by examiningonly a few doses that may be millions of timeshigher than those typical of exposure byhuman or wildlife? Second, there has beenconsiderable interest in the shape of dose-response curves for endocrine disruptors thatbind to intracellular receptors for endogenoussteroid hormones. However, until now, theestablishment of the dose range in toxicologicstudies on these chemicals has not been basedon an estimation of whether the doses admin-istered would result in doses within target tis-sues that would be below or above levels thatwould saturate available receptors for theendogenous hormone(s) being mimicked orantagonized. In a multigenerational study inwhich adults are administered a chemicalbefore and during the production of off-spring, and then the offspring continue to bedosed after weaning (the procedure is thenrepeated for two generations), three doses areusually examined (30). The lowest dose inthese experiments is typically a maximum of50-fold below the highest dose. The highestdose used in toxicologic experiments is basedon some index of acute toxicity, such as adecrease in body weight without other signsof overt toxicity.

With regard to the shape of the dose-response curve at low levels for endocrine dis-ruptors that interact reversibly with hormonereceptors (and other regulatory macromole-cules such as enzymes), consideration shouldbe given to characterizing the dose-responsecurve within the predicted dose range forregulating receptor activity on the basis of therelative potency of the endocrine disruptorand the endogenous hormone it mimics.

Third, the issue of the type of health riskposed by endocrine disruptors has generatedmuch discussion. There is evidence thatendocrine disruptors pose risks to functionalend points, such as neuromuscular andbehavioral changes (21,31,32), and organfunction (5,7,33). On the basis of these find-ings, the U.S. EPA will now require tests forendocrine disruptors that focus on adverseeffects on organ function (1).

Traditional approaches to determinedeleterious effects on the developing fetusfocused on high doses of compounds that maycause fetal death, malformations, or completeloss of function (such as infertility) (34,35).Tests commonly employed include classicalteratology tests. Such tests are referred to inthe industry as Segment 2 studies in whichgross malformations or death are the endpoints.These studies involve administrationof a chemical for a short period in pregnancy.Multigenerational studies have been conducted

for relatively few of the chemicals thatwill be screened by the U.S. EPA forendocrine-disrupting activity (36). Whethermultigenerational studies conducted with afew high doses will detect effects similar tothose seen with much lower doses is currentlybeing investigated for a few endocrine-dis-rupting chemicals in studies being conductedby the National Toxicology Program withinthe National Institute of EnvironmentalHealth Sciences.

Data for the mechanism of action of theendocrine disruptor in question provide abasis for predicting the types of adverse effectsthat may occur. However, these types of datahave not been available for most multigenera-tional studies that have been conducted, or ifknown, were not applied in the determina-tion of doses to be examined [for contrastingapproaches in examining a chemical used inplastic, bisphenol A (14,34)]. At present, lim-iting factors in using multigenerationalstudies to determine adverse developmentaleffects include the time required to completethese studies, interpretation of the extensiveamount of data generated, and cost effective-ness of such studies with respect to theknowledge gained about the effects. Anincrease in the number of doses used in thesestudies would increase costs unless accompa-nied by the use of smaller numbers of animalsper group. A resolution of these complexissues will require more information than isnow available.

The limitations of traditional teratologicand multigenerational studies led the workinggroup to suggest the following research needs:first, relevant and sensitive quantitative endpoints must be identified and tested over amuch wider range of doses than have previ-ously been examined. Second, the design ofthese experiments should require knowledgeof the variability of the end points in the con-trol population to adequately assess the num-bers of animals that should be examined (i.e.,conduct statistical power analysis). Third, theshape of the dose-response curve for specificresponses should be determined with respectto endocrine disruptors within a particularclass (for example, endocrine disruptors thatbind to estrogen receptors and show full ago-nistic activity). Fourth, the mechanisms ofreceptor binding and activation (and othermechanisms) should be determined over thefull range of dose responses. And finally, newstrategies and models for dose-responseassessment should be developed as databecome available.

In toxicologic studies, the current modelfor endocrine disruptors is based on thehypothesis that a) as dose increases, responsewill increase or stay the same (a monotonicdose-response curve is assumed), and b) athreshold exists below which there is no

Environmental Health Perspectives * Vol 107, Supplement 4 a August 1999 615

BIGSBY ET AL.

increase in risk (relative to controls) due toexposure (37). These assumptions, which arebased on studies conducted with high dosesof chemicals, have been challenged by theresults of experiments involving low doses ofendogenous hormones and endocrinedisruptors (7,14,29,38).

There are currently only a few ongoingstudies, including multigenerational studies,that have been designed to address some ofthese modeling needs and questions. Byaddressing these issues, information will beprovided concerning the need to expand thedose range for some chemicals. It will beimportant to determine which properties ofchemicals might predict whether theirdose-response relationships will behave in acomplex fashion. Finally, regulatory agencieswill have to assess the impact that this infor-mation will have on regulatory policies thatdrive the design of toxicologic studies (1).

Ability of Screens to PredictEmbryonic EffectsCurrent hazard identification (for example,identification of whether a chemical is anendocrine disruptor) and, more generally, riskassessment paradigms need to be reevaluatedto determine their effectiveness at assessingeffects of low doses of potential endocrine dis-ruptors on the developing organism. Althoughscreening systems can be designed to identifyendocrine-disrupting chemicals that eliciteffects at low doses, an additional concern iswhether there are unique effects of exposure tothese endocrine disruptors during critical peri-ods of development (i.e., organogenesis). Theconcern is that effects caused by exposure toendocrine disruptors during critical periods indevelopment may not be predicted by studiesconducted at later times in life (after weaning)and also may not be detected by in vitroscreens. There are data that support this possi-bility (5,17,24,39-411). Additionally, the iden-tification of which end points in which tissuesshould be evaluated for unique effects due toexposure during development needs to bemore carefully examined.

Proposed Chemicals toAddress the Issue of Dose inTests for Endocrine DisruptorsConsiderably more empirical data are neededthat directly compare the high end of thedose-response curve with the low end. Toaddress this issue, the work group suggestedthe following compounds for initial evalua-tion: diethylstilbestrol (DES), methoxychlor,bisphenol A, octylphenol, phthalates, keto-conazole, flutamide, propylthiouracil (PTU),and genistein. These compounds are pro-posed because much is already known abouttheir effects and mechanisms of action andbecause they present different spectra of

effects and mechanisms. Specifically, DES is apotent ligand for the estrogen receptor (ER).Methoxychlor has both estrogenic andantiandrogenic effects and must be metabo-lized to be active (24). Bisphenol A is anestrogenic chemical that binds to the ER withmodest affinity (14) and has been reported toresult in prostate enlargement and otherchanges in the reproductive system in mice(42) and changes in pituitary function in rats(43). Octylphenol also binds to the ER and isestrogenic in in vitro and in vivo assay systems(44) but shows significantly different bindingto plasma steroid-binding proteins thanbisphenol A (14). Some phthalates, such asdibutyl phthalate, show evidence of non-receptor-related effects on the androgen sys-tem (17). Ketoconazole blocks androgensynthesis and thus is antiandrogenic by areceptor-independent mechanism (17).Flutamide is a relatively pure androgenreceptor blocker and provides a positive con-trol antiandrogen (25). PTU produces thy-roid effects by inhibiting thyroid hormonesynthesis (45). Genistein has many actions,among which are binding to and activation ofthe ER as well as tyrosine kinase inhibition(19, 46). Thus, while a common thread ofhormone-related activities runs through thisgroup of chemicals, they present a sufficientspectrum of effects to allow a more broadassessment of the possibility of low-doseeffects and qualitative differences in responseacross the dose-response curve associated withnonmonotonic functions.

The doses used for these in vivo studiesshould cover the dose ranges from just belowovertly toxic (using the current method ofhigh dose selection) to approximately 6orders of magnitude lower. This dose rangeshould be sufficient to provide some informa-tion on the likelihood of nonmonotonicdose-response functions.

Although the end points measured inthese studies should be relevant to the com-pound being tested, whenever possible, anattempt to link end points to currentlyaccepted indices of toxicity should be made.At least some of the end points measuredshould take advantage of what is knownabout the molecular effects and mechanismsof each compound (i.e., levels of hormonesbeing mimicked, receptor number and actionin specific target tissues), whereas othersshould be more organ-level and whole-ani-mal-level end points (i.e., development of thereproductive or thyroid systems, gametenumbers, or rate of growth). The purpose ofexamining more sensitive end points for eachcompound against the more traditional endpoints in toxicologic studies is to establishwhether an effect is adverse by traditional cri-teria. However, it is also recognized that partof the new paradigm that has been developed

by the U.S. EPA in its endocrine disruptorscreening and testing program is a focus on adifferent set of outcomes from those previ-ously used in most toxicologic studies (1).

The development of this database willprovide important information regarding theprevalence of nonmonotonic dose-responsecurves and unique low-dose effects. As thisinformation becomes available, it can bedecided if current dose-response assessment,hazard identification, and risk assessment par-adigms need to be further modified. This willbe possible, as the endocrine disruptor screen-ing and testing program is designed to be aprocess that can be modified as new informa-tion becomes available [1). Future decisionsmust be based on data, not on presumptionand extrapolation.

The question of whether mixtures ofcompounds have a profile of toxicity thatdiffers qualitatively from that of its compo-nents has also been a concern with regard toendocrine disruptors. This question is of spe-cial importance given new regulatory man-dates (i.e., Food Quality Protection Act) tocarry out risk assessments based on the accu-mulated exposures to agents that exert theirtoxicity by a common mechanism. In prac-tice, the default approach to cumulative riskassessments is to consider the effects of indi-vidual components to be additive if theyinduce similar effects and there is no contra-dictory evidence to suggest a nonadditiveinteraction. However, if the dose-responserelationships are complex and nonlinear forthe components, then this practice would notbe appropriate. This issue must be addressed,particularly if it is determined that endocrinedisruptors have complex, nonmonotonicdose-response relationships.

Prostate Development as anExample of an Endocrine-Mediated Process Subjectto Endocrine DisruptionProstate development in the male mouse servesas a good example of the potential seriousnessof endocrine disruptors for the developingfetus. The prostate gland develops from theurogenital sinus (UGS) under the influence ofandrogens. In the day-14 male mouse embryo,testicular testosterone secretion increases, buttestosterone must be converted to 5a-dihy-droxytestosterone (DHT) by 5a-reductase fornormal prostate development to occur. DHTstimulates androgen receptor-positive mes-enchyme cells to induce glandular epithelialbudding. Thus, the critical parameters formodeling are fetal circulating testosteronelevels, UGS mesenchymal 5a-reductase activ-ity, androgen receptor content of UGS mes-enchyme, and mass of UGS mesenchyme atthe time of initial prostate organogenesis (10).

Environmental Health Perspectives * Vol 107, Supplement 4 * August 1999616

ENDOCRINE FUNCTION DURING DEVELOPMENT

With regard to the measurement ofestradiol and testosterone in plasma, the totalconcentration in plasma can be misleading.In rodents, the concentration of estrogenbinding plasma proteins (alphafetoprotein)that modify uptake into tissues is dramaticallyhigher in fetuses than adults. As a result, thefree (unbound to plasma proteins) concentra-tion of estradiol is approximately 10-foldlower in fetuses than in adults (18). This isimportant because endocrine-disruptingchemcials show significant differences in bind-ing to these plasma proteins relative to estra-diol, and thus show substantially differentuptake into fetal tissues than predicted by invitro assays that do not take this into accountin assessing the potency of these chemicals(19). In sharp contrast, the levels of freetestosterone are quite high in fetuses, asrodents do not have a high-affinity plasma-binding protein for testosterone (47).

A small increase in circulating estradiolin male fetuses permanently increases adultprostate weight via an increase in gland gen-esis (7). Specifically, a 50% increase in cir-culating estradiol led to a 40% increase inthe number of prostatic epithelial buds atthe end of the first day of prostate differenti-ation. Fetal testosterone levels were notincreased by estradiol treatment, but prosta-tic androgen receptors were permanentlyincreased (7). However, whether UGS 5a-reductase activity was influenced by estradioltreatment and whether there was any changein the mass of underlying UGS mesenchymehas not been examined. Quantitative analy-ses of these additional parameters wouldassist in formulating predictive hypotheses.

The relevance of the prostate model tothe assessment of endocrine-disrupting com-ponds is that it provides an example of anorgan that has been extensively studied withregard to the impact of alteration in enzymeactivity (50c-reductase) and circulatingsteroid levels (namely, testosterone andestradiol). Levels of these steroids can beadded to or interfered with by endocrinedisruptors that bind to their receptors andact as mimics or antagonists, respectively.Alternatively, circulating levels of thesesteroids could be altered by disruption ofsynthesis of or by competition for binding toplasma steroid-binding proteins. The impor-tance of the interaction of mesenchyme andepithelium in the UGS has been studied ingreat detail, and there are data on theontogeny of steroid receptors (4,48,49). Adetailed understanding of the mechanisms ofdevelopment is required to fully understandthe mechanisms of endocrine disruption,particularly with regard to understanding inmolecular detail the possibility that lowdoses of a particular chemical might interferewith normal developmental processes (30).

ConclusionsMuch of the controversy surrounding theproblem of endocrine disruptors in the envi-ronment is related to potential effects on theembryo and fetus. This working group deter-mined that we have limited information onboth the normal role of the hormones indevelopment and on potential endocrine dis-ruptors. Multigenerational assays have beenthe only means of assessing the potential fordisrupting normal development by endocrine-disrupting chemicals. The principal conclu-sion is that there is a need for more basicinformation about hormonal involvement indevelopment and for new methods to assess avariety of compounds for endocrine disrup-tor activity, particularly during criticalperiods in organogenesis.

REFERENCES AND NOTES

1. U.S. EPA. Endocrine Disruptor Screening and Testing AdvisoryCommittee (EDSTAC) Final Report. Washington, DC:U.S.Environmental Protection Agency, 1998

2. Kelce WR, Gray LE. Endocrine disruptors: effects on sex steroidhormone receptors and sex development. In: Drug Toxicity inEmbryonic Development, Vol II (Kavlock RJ, Daston GP, eds).Heidelberg:Springer Verlag, 1997;435-474.

3. Taguchi 0, Cunha GR, Robboy SJ. Experimental study of theeffect of diethylstilbestrol on the development of the humanfemale reproductive tract. Biol Res Pregnancy Perinatol 4:56-70(1983).

4. Cunha GR, Cooke PS, Bigsby R, Brody JR. Ontogeny of sexsteroid receptors in mammals. In: Nuclear Hormone Receptors:Molecular Mechanisms, Cellular Functions, ClinicalAbnormalities (Parker MG, ed). London:Academic Press,1991;235-268.

5. Newbold R. Cellular and molecular effects of developmentalexposure to diethylstilbestrol: implications for other environ-mental estrogens. Environ Health Perspect 103(suppl 7):83-87(1995).

6. Gilbert SF, Opitz JM, Raff RA. Resynthesizing evolutionary anddevelopmental biology. Dev Biol 173:357-372 (1996).

7. vom Saal FS, Timms BF, Montano MM, Palanza P, Thayer KA,Nagel SC, Dhar MD, Ganjam VK, Parmigiani S, Welshons WV.Prostate enlargement in mice due to fetal exposure to lowdoses of estradiol or diethylstilbestrol and opposite effects athigh doses. Proc NatI Acad Sci USA 94:2056-2061 (1997).

8. Taylor HS, Vanden Heuvel GB, Igarashi P. A conserved Hox axisin the mouse and human female reproductive system: lateestablishment and persistent adult expression of the Hoxa clus-ter genes. Biol Reprod 57:1338-1345(1997).

9. Greco T, Duello T, Gorski J. Estrogen receptors, estradiol, anddiethylstilbestrol in early development: the mouse as a modelfor the study of estrogen receptors and estrogen sensitivity inembryonic development of male and female reproductive tracts.Endocr Rev 14:59-71 (1993).

10. Cunha GR, Boutin EL, Turner T, Donjacour AA. Role of mes-enchyme in the development of the urogenital tract. In:Chemically Induced Alterations in Sexual and FunctionalDevelopment: The Wildlife/Human Connection (Colborn T,Clement C, eds). Princeton, NJ:Princeton Scientific Publishing,1992;85-105.

11. McLachlan JA, Newbold RR. Estrogens and development.Environ Health Perspect 75:25-27 (1987).

12. vom Saal FS. Sexual differentiation in litter-bearing mammals:influence of sex of adjacent fetuses in utero. J Anim Sci67:1824-1840 (1989).

13. Siiteri PK, Murai JT, Hammond GL, Nisker JA, Raymoure WJ,Kuhn RW. The serum transport of steroid hormones. RecentProg Horm Res 38:457-510 (1982).

14. Nagel SC, vom Saal FS, Thayer KA, Dhar MG, Boechler M,Welshons WV. Relative binding affinity-serum modified access(RBA-SMA) assay predicts the relative in vivo bioactivity of thexenoestrogens bisphenol A and octylphenol. Environ HealthPerspect 105:70-76 (1997).

15. Diczfalusy E. Endocrine functions of the human fetoplacentalunit. Fed Proc 23:791-798 (1964).

16. Wong C, Kelce WR, Sar M, Wilson EM. Androgen receptorantagonist versus agonist activities of the fungicide vinclozolinrelative to hydroxyflutamide. J Biol Chem 270:19998-20003(1995).

17. Gray LE Jr, Wolf C, lambright C, Mann P, Price M, Cooper RL,Ostby J. Administration of potentially antiandrogenic pesticides(procymidone, linuron, iprodione, chlozolinate, p,p'-DDE, andketoconazole) and toxic substances (dibutyl- and diethylhexylphthalate, PCB 169, and ethane dimethane sulphonate) duringsexual differentiation produces diverse profiles of reproductivemalformations in the male rat. Toxicol Ind Health 15:94-118(1999).

18. Montano MM, Welshons WV, vom Saal FS. Free estradiol inserum and brain uptake of estradiol during fetal and neonatalsexual differentiation in female rats. Biol Reprod 53:1198-1207(1995).

19. Nagel SC, vom Saal FS, Welshons WV. The effective free frac-tion of estradiol and xenoestrogens in human serum measuredby whole cell uptake assays: physiology of delivery modifiesestrogenic activity. Proc Soc Exp Biol Med 217:300-309(1998).

20. Divi Rl, Chang HC, Doerge DR. Anti-thyroid isoflavones fromsoybean: isolation, characterization and mechanism of action.Biochem Pharmacol 54:1087-1096 (1997).

21. Brouwer A, Morse DC, Lans MC, Schuur AG, Murk AJ, Klasson-Wehler E, Bergman A, Visser TJ. Interactions of persistent envi-ronmental organohalogens with the thyroid hormone system:mechanisms and possible consequences for animal and humanhealth. Toxicol Ind Health 14:59-84(1998).

22. Fox TO. Androgen and estrogen binding macromolecules indeveloping mouse brain: biochemical and genetic evidence.Proc NatI Acad Sci USA 72:4303-4307 (1975).

23. Wailer CL, Juma BW, Gray LE Jr, Kelce WR. Three-dimensionalquantitative structure-activity relationships for androgen recep-tor ligands. Toxicol AppI Pharmacol 137:219-227 (1996).

24. Gray LE Jr, Ostby J, Cooper RL, Kelce WR. The estrogenic andantiandrogenic pesticide methoxychlor alters the reproductivetract and behavior without affecting pituitary size or LH andprolactin secretion in male rats. Toxicol Ind Health 15:37-47(1999).

25. Kelce WR, Stone CR, Laws SC, Gray LE, Kemppainen JA,Wilson EM. Persistent DDT metabolite p,p'-DDE is a potentandrogen receptor antagonist. Nature 375:581-585 (1995).

26. Soto AM, Fernandez MF, Luizzi MF, Oies Karasko AS,Sonnenschein C. Developing a marker of exposure to xenoestro-gen mixtures in human serum. Environ Health Perspect 105(suppI 3):647-654 (1997).

27. Colborn T, vom Saal, FS, Soto AM. Developmental effects ofendocrine disrupting chemicals in wildlife and humans. EnvironHealth Perspect 101:378-384 (1993).

28. Colborn T, Smolen MJ, Rolland R. Environmental neurotoxiceffects: the search for new protocols in functional teratology.Toxicol Ind Health 14:9-23 (1998).

29. vom Saal FS, Sheehan DM. Challenging risk assessment. ForumAppI Res Pub Policy 13:11-18 (1998).

30. Welshons WV, Nagel SC, Thayer KA, Judy BM, vom Saal FS.Low-dose bioactivity of xenoestrogens in animals: fetal exposureto low doses of methoxychlor and other xenoestrogens increasesadult prostate size in mice. Toxicol Ind Health 15:12-25 (1999).

31. Jacobson, JL, Jacobson SW. Intellectual impairment in childrenexposed to polychlorinated biphenyls in utero. N EngI J Med335:783-789 (1996).

32. Lonky E, Reihman J, Darvill T, Mather J Sr, Daly H. Neonatalbehavioral assessment scale performance in humans influencedby maternal consumption of environmentally contaminated LakeOntario fish. J Great Lakes Res 22:198-212 (1996).

33. Peterson RE, Moore RW, Mably TA, Bjerke DL, Goy RW. Malereproductive system ontogeny: effect of perinatal exposure to2,3,7,8-tetrachlorodibenzo-p-dioxin. In: Chemically-inducedAlterations in Sexual and Functional Development: TheWildlife/Human Connection (Colborn T, Clement C, eds).Princeton, NJ:Princeton Scientific Publishing, 1992;175-193.

34. Morrissey RE, Lamb JC IV, Morris RW, Chapin RE, Gulati DK,Heindel JJ. Results and evaluations of 48 continuous breedingreproduction studies conducted in mice. Fundam AppI Toxicol13:747-777 (1989).

35. Kelce WR, Monosson E, Gamcsik MP, Laws SC, Gray LE Jr.Environmental hormone disruptors: evidence that vinclozolindevelopmental toxicity is mediated by antiandrogenic metabo-lites. Toxicol AppI Pharmacol 126:276-285)1994).

36. EDF. Toxic Ignorance. Baltimore:Environmental Defense Fund,1997.

37. Beck BD, Rudel R, Calabrese EJ. The use of toxicology in theregulatory process. In: Principles and Methods of Toxicology(Hays AW, ed). New York:Raven Press, 1994;19-58.

Environmental Health Perspectives * Vol 107, Supplement 4 * August 1999 617

BIGSBY ET AL.

38. Sheehan DM, Willingham E, Gaylor D, Bergeron JM, Crews D.No threshold dose from estradiol-induced sex reversal of turtleembryos: how little is too much? Environ Health Perspect107:155-159 (1999).

39. Bern HA. The fragile fetus. In: Chemically-induced Alterations inSexual and Functional Development: The Wildlife/HumanConnection (Colborn T, Clement C, eds). Princeton, NJ:PrincetonScientific Publishing, 1992;9-15.

40. Hajek RA, Robertson AD, Johnston DA, Van NT, Tcholakian RK,Wagner LA, Conti CJ, Meistrich ML, Contreras N, Edwards CL,Jones LA. During development, 17a-estradiol is a potent estro-gen and carcinogen. Environ Health Perspect 105(suppl3):577-581 (1997).

41. Newbold RH, Hanson RB, Jefferson WN, Bullock BC, HasemanJ, McLachlan JA. Increased tumors but uncompromisedfertility in the female descendants of mice exposed

developmentally to diethylstilbestrol. Carcinogenesis19:1655-1663 (1998).

42. vom Saal FS, Cooke PS, Buchanan DL, Palanza P, Thayer KA,Nagel SC, Parmigiani S, Welshons WV. A physiologically basedapproach to the study of bisphenol A and other estrogenicchemicals on the size of reproductive organs, daily sperm pro-duction, and behavior. Toxicol Ind Health 14:239-260 (1998).

43. Steinmetz R, Brown NG, Allen DL, Bigsby RM, Ben-Jonathan N.The environmental estrogen bisphenol A stimulates prolactinrelease in vitroand in vivo. Endocrinology 138:1780-1786 (1997).

44. Sharpe RM, Fisher JS, Millar MM, Jobling S, Sumpter JP.Gestational and lactational exposure of rats to xenoestrogensresults in reduced testicular size and sperm production. EnvironHealth Perspect 103:1136-1143 (1995).

45. Cooke PS, Hess RA, Kirby JD, Bunnick DB, Hardy MP. Neonatalpropylthiouracil (PTU) treatment as a model system for studying

factors controlling testis growth and sperm development. In:Function of Somatic Cells of the Testis (Bartke A, ed). NewYork:Springer-Veriag, 1994;400-407.

46. Akiyama T, Ishida J, Nakagawa S, Ogawara H, Watanabe S,Itoh N, Shibuya M, Fukami Y. Genistein; a specific inhibitor oftyrosine specific protein kinases. J Biol Chem 262:5592-5595(1987).

47. Danzo BJ, Eller BC. The ontogeny of biologically active andro-gen-binding protein in rat plasma, testis, and epididymis.Endocrinology 117:1380-1388 (1985).

48. Cooke PS, Young P, Cunha GR. Androgen receptor expression indeveloping male reproductive organs. Endocrinology128:2867-2873 (1991).

49. Cooke PS, Young P, Hess RA, Cunha GR. Estrogen receptorexpression in developing epididymis, efferent ductules and othermale reproductive organs. Endocrinology 128:2874-2879 (1991).

618 Environmental Health Perspectives * Vol 107, Supplement 4 * August 1999