Research Grants/Fellowships/SBIR

Final Report: Frog Deformities: Role of Endocrine Disruptors During Development

EPA Grant Number: R827398
Title: Frog Deformities: Role of Endocrine Disruptors During Development
Investigators: Gardiner, David M. , Blumberg, Bruce
Institution: University of California - Irvine
EPA Project Officer: Turner, Vivian
Project Period: October 1, 1999 through September 30, 2002
Project Amount: $1,194,536
RFA: Endocrine Disruptors (1999) RFA Text |  Recipients Lists
Research Category: Economics and Decision Sciences , Endocrine Disruptors , Health , Safer Chemicals



The objectives of this research project were to: (1) assess the significance of endocrine disruptors that activate retinoid signaling pathways for their role in causing limb developmental deformities in frogs; and (2) understand their mechanism of action to assess their implications for human health.

Summary/Accomplishments (Outputs/Outcomes):

Part I – Developmental Toxicology of Retinoid-Signaling Disruptors

There are Multiple Developmental Windows During Xenopus and Rana Early Limb Bud Development When Disruption of Retinoid Signaling Induces Multiple Phenotypes

During the grant period, we were able to optimize the animal holding/spawning conditions (temperature, water quality, and diet) for Xenopus laevis. We now spawn and raise Xenopus with a deformity rate of zero, and a mortality rate of less than 5 percent from hatching to the end of metamorphosis for control samples. This low rate of malformation and mortality has allowed us to develop a toxicology assay for the presence of chemicals that disrupt normal retinoid signaling.

We successfully tested the effects of several retinoids on a range of developmental stages of X. laevis. A variety of effects have been observed, consistent with pervious studies published in the literature on retinoid effects on amphibian development. The observed phenotypes include normal development, lethality, duplicated limb buds, bony triangles (BT), and truncated limbs. For the detailed characterization of the effects of retinoid disruption, we used p-[(E)-2-(5,6,7,8,-tetrahydro-5,5,8,8-tetramethyl-2-naphthalenyl)-1-propenyl]benzoic acid (TTNPB), a retinoic acid receptor (RAR)-specific ligand, as a model retinoid. TTNPB induces all the limb malformations observed in the field, including duplicated limb buds. TTNPB is very stable, and it induces malformations after both short (3 hours) and long (2 weeks) exposures.

Limb dysplasias have only been observed when larvae are treated during stages of limb bud development (Xenopus stages 51-55). Exposures at early developmental stages induced malformations in other organ systems (e.g., craniofacial and axial), but not in limbs. Treatment at all stages of embryonic and larval development is developmentally toxic at high doses and long exposures, as evidenced by the death of larvae a few weeks after exposure. At lower dosages, survival is high; however, for most stages, exposures that are not lethal do not induce limb malformations. In contrast, exposures between stages 51-55 induce limb malformations at high frequency.

In addition to the overall developmental window for induction of limb deformities (stages 51-55), there are smaller, specific windows associated with specific types of deformities. Exposure at stage 51 induces duplicated limb buds. Exposure at stage 51/52 induces BT at high frequencies. Exposure during stages 52/53 induces the loss of more distal limb structures (lower leg and foot), and beyond stages 53, TTNPB exposure affects only digit development. Although both forelimbs and hindlimbs can be affected, hindlimb development is more frequently affected, and it appears to be more sensitive to retinoid-induced malformations.

The Induction of Limb Malformations at Sensitive Stages of Limb Development is Mediated by RAR Signaling

Although all of the tested retinoids (all-trans retinoic acid [RA], retinol, palmitate, retinal, and TTNPB) can induce amphibian limb malformations, it is difficult to identify the direct mode of action because of the complex metabolism of retinoids in vivo. Our finding that all the malformations can be induced by TTNPB is significant because TTNPB functions through specific binding and activation of the RAR.

BTs Are Diagnostic of Exposure to a Retinoid

The induction of BTs appears to be specifically correlated with disruption of retinoid signaling in all vertebrate embryos, including humans. Retinoids appear to induce BTs by altering the positional identity of cells around the circumference of the limb bud, because surgical rotation of the limb bud to juxtapose cells from the opposite sides of the limb can induce BTs. Similar grafting of tissues to juxtapose anterior and posterior cells induces duplicated limbs in many vertebrate embryos, and it is mechanistically equivalent to the induction of extra limbs by exposure to retinoic acid. Thus, the high frequency of BT induction in response to exposure to TTNPB is consistent with the conclusion that BTs are diagnostic of exposure to a retinoid.

TTNPB Induces BTs in Rana

Although Xenopus is a valuable laboratory model organism for experimental work, it is not a species that is native to North America, and thus, is not among the affected species found in the natural populations that are exhibiting malformations. For this reason, we conducted parallel experiments in Rana pipiens and Rana sylvatica, which are native species, but are not as amenable to laboratory experimentation. Previous experiments by others and us have demonstrated that retinoids, including TTNPB, retinol palmitate, and all-trans RA can induce limb malformations, including BTs, in Rana. In our experimental work during this project, we demonstrated that TTNPB will induce BTs at high frequencies in R. sylvatica, indicating that, as in Xenopus, BT induction is mediated by RAR signaling.

BTs Arise Early in Limb Bud Development. Although BTs can be observed and described in postmetamorphic frogs, the sensitive stages for induction of this dysplasia are about 6 weeks prior to metamorphic climax. We observed tadpoles at multiple time points after exposure, but prior to metamorphic climax. We observed that BTs form while the limb bud is initially growing out and developing. BT formation does not occur as a secondary event after limb development is nearly completed at the time of metamorphic climax. Similarly, duplicated limbs arise from duplicated limb buds that appear about 3 weeks after TTNPB exposure.

Exposure to TTNPB Induces Both Bilateral and Unilateral BTs. Because experimental animals are exposed to TTNPB that is added to the aquarium, they are treated with a uniform, external exposure that often induces malformation in both hindlimbs. However, unilateral malformations often are observed in which one limb is normal. In addition, malformations can be observed in both forelimbs and hindlimbs; however, hindlimbs appear to be more sensitive, and many animals with malformed hindlimbs develop normal forelimbs. Some investigators have assumed that exposure to a chemical agent would necessarily induce bilaterally symmetrical deformities in both forelimbs and hindlimbs. Based on this assumption, it has been argued that the presence of asymmetrical, unilateral deformities is evidence against chemical exposure as a cause of frog deformities. Our results demonstrate that neither the assumption nor the conclusion is correct.

Organic Extracts of Water Samples and Known Pesticides Can Induce Hindlimb Malformations in R. pipiens

In a series of experiments that we conducted in collaboration with the U.S. Geological Survey Columbia Environmental Research Center, we have observed experimental animals with a variety of limb malformations, including BTs. These animals had been exposed to organic compounds extracted from environmental water samples, as well as to a mixture of known pesticides. As in deformed frogs in nature, limb malformations were only induced in hindlimbs. In some experimental animals exposed to environmental organic compounds or pesticides, BTs were induced in both the left and right hindlimbs (bilateral); however, as observed in deformed frogs in nature, some animals exhibited unilateral BTs. The results are comparable to those observed in response to TTNPB exposure.

Screens for Disruption of Retinoid Signaling by Known Pesticides

Using the parameters that we established for the effects of TTNPB on Xenopus limb development, we established a developmental toxicology assay to screen for the ability of a large number of known pesticides to disrupt retinoid signaling. Chemicals were selected by a variety of criteria including known effects on retinoid-sensitive signaling pathways and known presence in water samples from sites with deformed frogs. We tested a large number of compounds (see Table 1), and we did not detect the induction of any significant, reproducible limb malformations in response to exposure to any of these compounds.

Among the chemicals that we have identified as being present in the water at our study site in Minnesota (Crow Wing B [CWB]), we determined that N,N-diethyl-m-toluamide (DEET) is present at very high levels (discussed below). For this reason, we tested the effect of DEET on limb development in both Xenopus and Hyla regilla, another species of amphibian native to North America. We did not detect the induction of any significant limb malformations in response to exposure to DEET.

Table 1. Compounds Tested for the Ability To Disrupt Retinoid-Sensitive Signaling Pathways During Amphibian Limb Development

Atrazine Aatrax Nonylphenol
Bromoxynil Tetrachlorobisphenol A Tetrabromobisphenol A
Bisphenol A Nonylphenol plus TTNPB Nonylphenol plus coumaphos
Nonylphenol plus DEET Sevin (carbaryl) Bladex
MCPA Triphenyltin acetate 4-Heptylbenzoic acid
Phenothiazine p-Terphenyl Aximphos-ethyl
Permethrin Nitrite Maneb
Tributyltin Triethyltin bromide POE (8)
Polyoxyethylene (POE) (9-10) POE (10-11) POE (12)
Dicamba Fenitrothion DEET

Finally, we tested several environmentally derived samples for the ability to induce limb malformations. In addition to water from the CWB site and partially purified fractions from that water (see below), we also tested water from a site in Southern California that we have monitored for the past 3 years. Pacific tree frogs (H. regilla) at that site exhibited typical limb malformations at seasonally high frequencies. We also tested aqueous extracts of cow manure obtained from the dairy located adjacent to the CWB site in Minnesota as a possible source of a chemical agent that could induce limb malformations. We did not detect the induction of any significant limb malformations in response to exposure to any of these samples.

Malformations Persist in Nature and Populations of Amphibians at Affected Sites Continue to Decline

Although malformed frogs have been widely studied since 1995, continued annual studies of individual sites to reveal malformation recurrence and impacts on frog populations are not common. A central Minnesota site was studied beginning in 1996 and continuing to the present as part of our investigation. CWB is among the worst sites in the state regarding frog malformation severity and frequency. Malformation frequencies were highest (5-85 percent) in highly aquatic mink frogs and green frogs, moderate (11-20 percent) in semiaquatic northern leopard frogs, and lowest (5 percent collectively) in highly terrestrial wood frogs, gray tree frogs, spring peepers, and American toads. High mortality of larval and perimetamorphic frogs (mostly mink frogs), fathead minnows, and brook sticklebacks was observed in June and July of 1997-2002. Frog mortality appeared to be independent of malformations. Tape recordings, surveys, and capture data revealed a progressive population decline in northern leopard frogs and mink frogs (near absence of both species by 2000 and 2002, respectively) and disappearance of wood frogs and American toads.

Our field studies in Year 3 of the project indicated that populations of amphibians continued to decline, and new populations of deformed frogs continue to be discovered. Intensive field surveys at the CWB site (tape recorder to monitor calling by breeding frogs and wading census) throughout the spring and summer months revealed that R. pipiens is still nearly absent. The previously low breeding populations of adult Rana clamitans, Rana septentrionalis, and Bufo americanus are now absent. Continued weekly surveys at the CWB site through prospective mink, green, and leopard frog metamorphosis period to collect data on newly metamorphosed animals indicated that the frequency of malformations also has decreased relative to levels at the time our research project was initiated. The array of malformation types also was somewhat different, with the predominant type being abnormal jaws, followed by animals with unresorbed tails with adult pigment (whereas in previous years, the predominant types of malformations were skin webbings and hindlimb malformations). We also surveyed eight other sites where malformed frogs were observed in previous years, and observed that although the frequency was significant, it was lower than in the past (< 5 percent as compared to 5-10 percent).

Part II – Isolation, Purification, and Identification of an Environmental Retinoid

CWB Water Contains an Organic Chemical That Activates RAR

Because chemicals that activate RAR (retinoids by definition) can induce all the malformations observed in natural populations (discussed above), we hypothesized that a chemical with RAR-activating activity would be present in water samples from sites where deformed frogs are found. We have found such an activity, and have made significant progress in purifying the chemical that is responsible for this activity. This activity is associated with a chemical in solution in the water, and is not associated with particulate material (greater than 0.2 µm) in the water samples (e.g., plant debris or bacteria). We have used a RAR-reporter assay to guide the fractionation and purification of the chemical associated with this activity.

Fractionation and Purification Techniques Have Been Developed

To identify the RAR-activating chemical, we need to purify larger quantities of this chemical. We determined that we can use XAD-2 resin in a solid-phase extraction procedure to obtain sufficient quantities of the purified chemical. CWB lake water was pumped over the resin to capture organic substances. Substances were eluted from the XAD-2 resin by sequential elution with methanol, acetone, hexane, and methylene chloride, concentrated by rotor evaporation. This extract then was fractionated over four sequential reverse phase C18 and C8 high performance liquid chromatography (HPLC) columns. Fractions were assayed for RAR activity using GAL-RARa and UAS-luciferase reporter constructs transiently transfected into Cos-7 cells. The activity profile was used to guide the fractionation. A prominent activity has been consistently present in lake water samples. On analytical C18 reverse-phase columns developed with shallow mixed methanol-acetonitrile gradients, the activity was resolved into three distinct activity peaks.

Electrospray mass spectroscopy (ES-MS) data on the mixture in these fractions showed a number of mass peaks (see Table 2) present in the fractions. Exact masses have been obtained for all ions, and putative molecular formulae have been deduced.

Frequently, the most intense ions detected in these fractions by ES-MS are a series of ethoxyl polymer derivatives (series A and B) separated by 44 a.m.u. We have positively identified the most abundant of these to be nonylphenol polyethylene glycol (n=2-13) using various mass spectroscopic methods, including gas chromatography (GC) electron impact (EI)-MS and by retention time and single-ion monitoring on HPLC ES-MSMS in comparison to authentic standards. Nonylphenol polyethylene glycol is a manmade polymer that is extensively utilized in various industries as a surfactant. Various polyethylene glycol polymers are major components in many agrochemical pesticide formulations where they are used as spreading/wetting agents. Although the high molecular weight polymers are considered inert, degradation products can exhibit endocrine disrupting activities. Nonylphenol, for instance, is a known weak environmental estrogen suspected to be linked to feminization of salmonids.

Table 2. Prevalent Ions Present in the CWB XAD-2 Active Fractions

Mass Peaks (m/z+) Detected:  
399.12/473.14 Related to 397.12/473.14 by saturation of a double bond?
331.15/375.17/419.18 Ethoxyl polymer series A
551.23/629.26/683.28/727.31/771.30 Ethoxyl polymer series B
(Nonylphenol polyethylene glycol)

We have tested reagent grade nonylphenol polyethylene glycol as well as numerous closely related polymer series (polymers with altered side chains, different polymeric size distributions and functional groups) in our RAR transactivation assay, but have not observed any significant activation to date. One caveat, however, is that in commercial preparations of these polymers (e.g., Tergitol NP-9), individual geometric isomers of a specific polyethoxyl chain length comprise only a small fraction by mass of the mixture. Activation by a minor component may therefore be masked by cytotoxicity of the mixture in the in vitro assay.

Additional Compounds Have Been Identified. In addition to the major RAR activities described above, we also have characterized a minor RAR activity that elutes towards the end of the HPLC chromatograms. Components in this region are extremely hydrophobic requiring dichloromethane modifier to elute from the column. Analysis by ES-MS and the distinctive photodiode array ultraviolet spectra, identified this component as a dihydroxy-carotenoid, most likely alloxanthin (M.Na+ = 587.64; UV max 425, 452, 481 nm). The source of this active carotenoid is likely of natural origins, either from suspended decaying plant matter or of algal origin. A spun polyester fiber prefilter was used to prevent fouling of the XAD-2 resin columns, but was not completely effective at removing particulates. Some entrapment of carotenoid-containing material was expected. In contrast, the main RAR activity was too polar to be a carotenoid, and it was not associated with the particulate matter, because a 0.2-µm filter was ineffective at removing it from the sample.

Analysis by HPLC ES-MS and GC EI-MS of total hydrophobic components in water samples from CWB also positively identified DEET as a prevalent contaminant of the lake water, but not the adjacent well water. DEET is the leading consumer insect repellent and a common emerging contaminant in U.S. waterways. Personnel did not use DEET during water sample collection. We tested DEET on a battery of nuclear hormone receptors in transient transfection assays for its ability to activate in a ligand-dependent manner. Only the Xenopus benzoate X receptor (BXR) was activated in a dose-dependent manner. The BXRs are activated by small benzoate-like molecules. Their exact role in amphibian development and adult life is unknown. We extensively tested DEET for developmental effects in our Xenopus limb assay, but did not observe any significant malformations (discussed above).

Part III – Conclusions

• Chemicals that activate the RAR have the potential to induce all of the malformations observed in wild populations of deformed frogs. By definition, these chemicals are retinoids, and they act to disrupt normal retinoid signaling in developing frog tadpoles.

• Retinoids only induce malformations during specific developmental windows of sensitivity. Different types of malformations are induced depending on the stage of exposure. Different doses of retinoids induce malformations depending on the duration of exposure.

• The types of malformations depend on the developmental stage at which the animal was exposed. The developmental response ranges from "normal" to "malformed" to "dead." Thus, variations in dose and time of exposure relative to developmental stage results in the induction of all the malformation phenotypes observed in natural populations of frogs.

• We have developed a limb teratogenesis assay to screen for the ability of compounds to induce the various types of malformations observed in natural populations of frogs. To date, none of the tested compounds have yielded positive results in the limb teratogenesis assay.

• We have developed a protocol for the fractionation and purification of compounds from water samples collected in the field. This protocol is designed to identity compounds that activate RAR, and thus would be prime candidates for the compound that induces malformations in natural populations of frogs.

• We have succeeded in identifying some of the compounds that are present in the most highly purified fractions; however, these compounds do not activate RAR, and they are unlikely to be candidates for the teratogenic chemical of interest.

• Efforts to increase the yield of the remaining, unidentified compounds are in progress.

• There has been a steady decline in the number of frogs during the period of time that we have studied the site. We presume this is the consequence of the high rates of mortality observed in the laboratory in response to a range of doses and durations of exposure to retinoids.

Journal Articles on this Report : 1 Displayed | Download in RIS Format

Other project views: All 33 publications 1 publications in selected types All 1 journal articles
Type Citation Project Document Sources
Journal Article Gardiner D, Ndayibagira A, Grun F, Blumberg B. Deformed frogs and environmental retinoids. Pure and Applied Chemistry 2003;75(11-12):2263-2273. R827398 (Final)
not available
Supplemental Keywords:

human health, dose response, teratogen, amphibian, chemicals, pesticides, retinoids, organics, environmental chemistry, developmental biology, biology, zoology, Northeast, Midwest, Minnesota, MN, pollution, herbicides, toxics, cancer, birth defects, carcinogen, methoprene, skeletal dysplasia, endocrine disruptors, pesticides, agrochemicals, altered gene expression, animal models, bioindicator, biomarkers, boney triangles, BTs, developmental biology, endocrine disrupting chemicals, EDCs, limb deformities., RFA, Health, Scientific Discipline, Toxics, Environmental Chemistry, pesticides, Endocrine Disruptors - Environmental Exposure & Risk, endocrine disruptors, Risk Assessments, Biochemistry, Children's Health, Ecological Risk Assessment, Biology, Endocrine Disruptors - Human Health, bioindicator, altered gene expression, biomarkers, limb deformities, EDCs, endocrine disrupting chemicals, boney triangles, retinoid signaling pathways, developmental biology, animal models, agrochemicals, developmental disorders

Relevant Websites: Exit Exit

Progress and Final Reports:
Original Abstract
2001 Progress Report