Bisphenol A (BPA) is used as the monomer to manufacture polycarbonate plastic, the resin that lines most food and beverage cans, dental sealants, and as an additive in other plastics (, ) [1
]. BPA is one of the highest volume chemicals produced worldwide; global BPA production capacity in 2003 was 2.2 million metric tons (over 6.4 billion pounds), with a 6-10% growth in demand expected per year [2
]. Heat and either acidic or basic conditions accelerate hydrolysis of the ester bond linking BPA monomers, leading to release of BPA and the potential for human and environmental exposure. Studies conducted in Japan [3
] and in the USA [4
] have shown that BPA accounts for the majority of estrogenic activity that leaches from landfills into the surrounding ecosystem.
Chemical structure of bisphenol A (BPA).
Table 1 Physical, chemical, and acute toxicological properties of bisphenol A (BPA) [1, 199].
BPA has been demonstrated in both in vivo
and in vitro
experiments to act as an endocrine disrupting chemical (EDC) (reviewed in [5
]). There is extensive evidence that BPA is an estrogen-mimicking chemical, although recent findings have revealed that BPA is a selective estrogen receptor modulator (SERM), since BPA and the potent endogenous estrogen 17β-estradiol (E2) do not always show identical effects, and in some studies BPA has been shown to antagonize the activity of E2 [7
]. There is evidence that, similar to other estrogens, BPA can bind to androgen receptors and inhibit the action of androgen [8
]. In addition, there is evidence for an anti-thyroid hormone effect of BPA [9
]. However, effects of BPA mediated by binding to androgen and thyroid hormone receptors appear to require higher doses than those required to elicit estrogenic or antiestrogenic responses [7
Chemicals classified as endocrine disruptors include not only hormone-mimics or antagonists that act via binding to receptors, but also chemicals that can interfere with hormone synthesis and clearance, as well as other aspects of tissue metabolism. Experiments have shown that BPA influences enzyme activity and thus metabolism in various tissues. Another mechanism of endocrine disruption is the alteration of hormone receptor expression, and experiments described below have shown that BPA alters hormone receptor numbers and hormone receptor gene activity in target tissues.
In this review, we summarize the recent literature on low-dose effects of BPA in laboratory animals. The majority of the studies used rats and mice; only a few used other mammalian species. We conclude with a series of statements expressing our level of confidence concerning various effects of BPA in laboratory animals at low doses.
Rationale for experimental design, analysis and reporting criteria
Endocrinology experiments with laboratory animals are particularly vulnerable to confounding effects. Among other difficulties, treatment effects can be masked by hormonally active components of feed, water, or caging; and species and strains differ greatly in their sensitivity to different hormonally active compounds. Therefore, proper reporting of experimental design is critical to evaluation of studies in the literature. Below, we discuss several of the most important aspects of experimental design, which we considered when evaluating the studies discussed in this review.
Published results must identify precisely the animal model and supplier being used. For example, Sprague-Dawley rats from different commercial breeders cannot be assumed to be the same, since it is an outbred stock. In particular, the outbred Sprague-Dawley CD rat from Charles River Laboratories [Crl:CD(SD)] has very low sensitivity to exogenous estrogens, and after more than 50 years of selective breeding by Charles River for large body size and litter size, it would be inappropriate to identify these rats as just Sprague-Dawley (). In contrast to the Crl:CD(SD) rat, male and female CD-1 (ICR) mice are highly sensitive to exposure to low doses of BPA during development as revealed by over 20 published studies (reviewed below) reporting significant effects of low doses of BPA in this outbred stock. This high sensitivity of the CD-1 mouse to BPA is predicted by the high responsiveness to positive control estrogens: E2, ethinylestradiol and diethylstilbestrol (DES), as revealed by both the in vivo
studies discussed here and other studies of CD-1 cells and organs in primary culture [10
Lack of effects reported in low-dose in vivo bisphenol A (BPA) research using the Charles River Laboratories Sprague-Dawley (Crl:CD(SD)) rat strain.
The marked difference in sensitivity of different animal models used in toxicological, pharmacological and endocrinological research is just one of many reasons why it is essential that experiments include appropriate positive controls, which is discussed in more detail in following sections. With regard to examination of the in vivo
estrogenic activity of BPA, which is the subject of this review, the sensitivity of the endpoint of interest in the chosen animal model should be characterized with a positive control such as E2 (appropriate if administration is by injection or subcutaneous capsule, due to very limited oral absorption of E2) or either DES or ethinylestradiol (appropriate if chemicals are administered orally, since they are orally active at very low doses). For example, in the CD-1 mouse, which is the animal model used by the U.S. National Toxicology Program, an appropriate positive control dose of ethinylestradiol or DES to detect a response to BPA within the low-dose range (below 50 mg/kg/day) would be an oral dose not greater than 5 μg/kg/day. This suggestion is based on numerous reports that for responses mediated by nuclear estrogen receptors, estimates of BPA potency in CD-1 mice range between 10 – 1,000-fold less than either ethinylestradiol or DES, depending on the specific response being measured [10
]. Also, doses of DES above 5 μg/kg/day can result in opposite effects relative to lower doses; this has been shown following developmental exposure for the prostate [16
] and uterus [18
]. High doses of these positive control estrogenic chemicals are thus not appropriate as a positive control for low-dose effects of BPA or other estrogenic endocrine disrupting chemicals.
Feed, water, and housing
The exact feed used must be identified. Ideally, the estrogenicity of the feed should be characterized, since estrogenic components have been demonstrated to occur in both soy-based and non-soy-based animal feeds. There is also the possibility of variation in estrogenic activity between different lots of feed commonly used in toxicological research [19
]. If possible, the same lot of feed (same mill date) should be used throughout an experiment.
The type of caging should be carefully selected to avoid estrogenic contamination of experimental animals. In particular, polycarbonate cages and water bottles should not be used, since they will leach uncontrolled concentrations of BPA to the experimental animals [20
]. Polypropylene cages have been used successfully in studies of estrogens in mice, and polysulfone cages are available to replace polycarbonate; polysulfone is a co-polymer containing BPA and sulfone, but it is reported to be more resistant to degradation at high temperatures relative to polycarbonate [20
]. Similarly, the source of drinking water must be free of BPA and other estrogens; reverse osmosis and carbon filtration is often necessary to achieve this requirement. In addition, if BPA is delivered in drinking water, the water must be free of chlorine or other reactive ions. Adding BPA to chlorinated water results in formation of tetrachlorobisphenol A.
Method of dosing
In all cases the precise method of dosing the animals and the time of dosing should be identified. Methods of administration of BPA include: 1. Oral (p.o.): by gavage, by adding BPA to feed or drinking water, or by feeding the chemical in oil; 2. Injection: subcutaneous (s.c.), intraperitoneal (i.p.), intracisternal, or intramuscular (i.m.) routes; and 3. Implantation of Silastic® capsules or Alzet® minipumps that lead to steady-state exposures. The rationale for the dosing method must also be stated. For example, minipump implants model continuous exposure and avoid the first-pass metabolism of BPA in the liver that results from oral exposure.
The positive control chosen must be compatible with the selected route of exposure. For example, E2 has very low activity when administered orally. For studies in which BPA is administered orally, DES or ethinylestradiol are appropriate positive controls. When BPA is administered in constant release capsules, E2 would be an appropriate positive control, while an estrogen agonist with a long half-life would be inappropriate. Route of administration influences the rate of metabolism of BPA, at least in adults [21
], and there is some evidence for a higher contribution from ingestion of BPA in humans relative to inhalation or absorption through the skin [22
]. However, the report that BPA levels in plasma collected from pregnant women in Germany show a range encompassing two orders of magnitude, between 0.1 and 10 parts per billion [23
], suggests the likely possibility of variable exposure to multiple sources of BPA. Thus, although oral delivery appears to be most relevant for extrapolation to humans, all delivery methods may reveal effects of BPA.
Design and Analysis Issues
When animals are assigned to groups, the litter must be taken into account. It is well established that for both outbred stocks and inbred strains, the litter is a significant source of variation and needs to be accounted for in assigning animals to groups. Ideally, one animal per litter should be used. In cases when this is not possible, there are statistical methods (such as including litter as a main effect variable and dividing the F value for treatment effects by the F value for litter effects) that can be used when more than one animal per litter is used, or litter can be used as a covariate in ANCOVA. The method used to avoid confounding litter effects must be reported.
For all experiments, the positive and negative controls must be clearly identified. Experiments using a replicate block design should include a positive and a negative control in each block, if possible. Experiments showing no effect and lacking positive control data cannot be interpreted.
When taking measurements from adult animals, care must be taken to normalize the reproductive state of the experimental animals. Males should be singly housed for 2-4 weeks before collection, in order to avoid physiological differences arising from differences in dominance status of the males. Normalization of adult female reproductive status can be achieved by assaying their estrous cycles and collecting females at the same stage, or by using ovariectomized females with or without hormone replacement [24
Definition of “Low Dose”
Low-dose effects of environmental endocrine disrupting chemicals generally refer to effects being reported at doses lower than those used in traditional toxicological studies for risk assessment purposes. “Low dose” is also commonly used to refer to environmentally relevant doses, i.e., doses resulting in serum levels close to those observed in human serum. For BPA, prior to 1997, the lowest dose studied for risk assessment purposes was 50 mg/kg/day, which in the USA remains the currently accepted lowest observed adverse effect level (LOAEL) that was used to calculate the current EPA reference dose (and FDA acceptable daily intake or ADI dose) of 50 μg/kg/day; this presumed “safe” dose is estimated by dividing the LOAEL by three 10-fold safety factors (i.e. by 1000) [25
]. Thus, we included in our analysis studies dosing with less than 50 mg/kg/day BPA ( and ).
Published studies since 1997 (reviewed in detail in ) that reported low-dose effects of bisphenol A (BPA) in mice and rats resulting from exposure during development or during adulthood.
Table 4 Published papers reporting biological effects in animal studies (excluding aquatic animals) for bisphenol A (BPA) in the low-dose exposure range (<50,000 μg/kg BW/day), sorted by tissue or endpoint examined. The information indicated in (more ...)
Timing of exposure
Exposures to endocrine disruptors have different effects depending on the life stage of the exposed animals. Effects resulting from adult exposure are generally reversible and are termed “activational”. Effects resulting from exposure during organ development (beginning during prenatal development and continuing in postnatal life through puberty) may result in persistent alterations of the affected systems, even in the absence of subsequent exposure; these effects are termed “organizational” [26
]. Some organizational effects are measurable immediately upon exposure and persist throughout the life of the animal [16
]. Other organizational effects are undetectable at the time of exposure, but they become apparent in subsequent adulthood [26
]. Windows of vulnerability, also known as critical periods, during which the developing system is most sensitive to exposure, are common features of organizational effects [28
]. Exposures occurring outside the critical periods will not elicit organizational effects. There is evidence that organizational effects of estrogenic endocrine disruptors such as BPA are mediated by epigenetic alterations in DNA [29
]. Organizational and activational effects on the same tissue often differ qualitatively as well as in duration and in the dose required to elicit effects. In this review, we will first discuss organizational effects of prenatal through pubertal BPA exposure, termed “developmental effects” and then activational effects of adult BPA exposure, termed “adult effects”.