Endocrine disrupting chemicals (EDCs) are compounds, or mixtures of compounds, that interfere with hormone actions by altering hormone synthesis, secretion, binding, metabolism, or elimination [1, 2]. Scientists that study EDCs in controlled laboratory experiments must take concerted efforts to control contaminations from environmental chemicals that could interfere with the assays being conducted and/or the endpoints being evaluated [3, 4]. These contaminations can be introduced by standard laboratory consumables (disposable plastics, storage containers), disinfectants, laboratory equipment, and other sources. In vitro assays designed to examine the role of estrogens in breast cell proliferation were some of the first studies to show that xenoestrogens can leach from commercially available plastics [5, 6]. Further, studies in mice that were originally intended to evaluate age-dependent changes in ovarian function and the timing of pubertal onset were disrupted by the unanticipated leaching of estrogenic chemicals from animals’ caging materials [7–10].

Experiments using laboratory animals must consider a number of different potential sources of EDCs including caging materials, water bottles (or tubing to deliver water directly to the cage), bedding, enrichment materials, food, water, and even the air in the animal facility [3, 11, 12]. A failure to rule out contamination can make it difficult, or impossible, to draw accurate conclusions about the endocrine disrupting properties of test chemicals. For example, if an experimenter wants to evaluate whether a test chemical displays estrogenic properties but all animals (including the unexposed negative controls) are inadvertently exposed to a xenoestrogen from the environment, the treated animals may appear to be non-responsive due to maximal stimulation of all animals [13, 14].

Significant efforts have been invested to identify the most sensitive endpoints to reveal the effects of EDCs [15, 16]. Whereas traditional endpoints of toxicity (e.g. significant loss of body weight, altered organ weight) are typically used to evaluate chemicals for regulatory purposes, numerous studies have demonstrated that the endpoints evaluated in test guidelines are not sufficiently sensitive to predict the effects of chemicals on humans, particularly when exposures occur during vulnerable periods of development [17–19]. Studies that have directly compared the effects of traditional guideline endpoints with non-guideline endpoints are relatively rare [20], but a large volume of data suggests that non-guideline endpoints can identify adverse effects of EDCs at doses far below those that alter organ weight [21–23].

Our own work, and work from multiple other laboratories, has shown that the rodent mammary gland is sensitive to EDCs, especially when exposures occur during critical windows of development [24–28]. We recently showed that male CD-1 mice exposed to bisphenol A (BPA), a well-known xenoestrogen, during the perinatal period develop larger and more elaborated mammary epithelial trees in adulthood [29]. These findings were particularly interesting because male mice do not have nipples, leading some researchers to conclude that they do not retain any epithelial tissues [24]. The male CD-1 mouse mammary gland may in fact be a sensitive bioassay that could be used to evaluate novel compounds as well as evaluate the potential for inadvertent contamination of animal populations with xenoestrogens, and perhaps compounds with other endocrine disrupting properties [30, 31].

While conducting controlled laboratory experiments, we have collected mammary glands from adult male CD-1 mice that were born and raised in a commercial laboratory. We noted that the epithelial trees in these males were much larger and more developed than what we previously reported for ‘control’ adult male mice; rather, these males had mammary tissues that more closely resembled animals developmentally exposed to BPA [29]. Based on those observations, we asked if the morphology of the mouse mammary gland can be used to distinguish colonies of mice raised in different environments including groups that may have experienced EDC exposures. Here, we report the results of our evaluations of two populations of mice: CD-1 mice ordered from commercial laboratories, and CD-1 mice raised in our animal facility for two generations under controlled conditions with limited environmental chemical exposures (e.g. exposures that were minimized wherever possible). We find that these populations are distinguishable, with the most striking differences in the adult male mammary glands, suggesting this may be an appropriate sentinel organ to evaluate an animal’s prior environmental exposure to xenoestrogens and other EDCs.

Our initial observations of male CRL mammary glands were that these glands were larger than expected, consistent with developmental exposures to EDCs like BPA [29]. In fact, the mammary glands collected from CRL animals were most similar to CD-1 male mice from prior studies that were developmentally exposed to BPA in the range of 2.5 – 25 μg/kg/day [29]. Further quantifications revealed that our F2 animals, raised in controlled conditions with concerted efforts to limit EDC exposures, display small ductal trees similar to what have been reported previously in other controlled environments [29–31, 36].

At first glance, the female data show that mammary glands from CRL females are smaller at puberty, which might be interpreted as suggesting that their chemical exposures are lower than those experienced in the F2 females. However, studies examining the effects of BPA on the pubertal mammary gland have shown that lower doses can increase ductal area whereas higher doses can decrease ductal area [35, 37]. These studies also demonstrate that BPA doses in the ng/kg and μg/kg range can significantly shift the density of TEBs in the mammary gland, to produce densities consistent with what we have observed in the CRL females. Thus, it is highly likely that all of the CRL animals experienced developmental exposures to xenoestrogens.

Is it possible that the CRL and F2 animals have some other feature that distinguishes these two populations, e.g. could our observations be due to genetic drift? Non-environmental differences between CRL and F2 animals seem highly unlikely, especially considering that all of our F2 animals were derived from animals that were ordered from the commercial supplier (e.g. F2 animals are only two generations removed from CRL animals), and it is unexpected that such drastic genetic changes could occur in just two generations of controlled breeding. However, our study is limited because we do not know what chemicals CRL animals might have been exposed to during development or the specific source of chemical exposures. Although our results are both qualitatively and quantitatively consistent with exposures to BPA in the low dose range [38], similar effects on the mammary gland are expected from exposures to other xenoestrogens including common chemical contaminants in animal feed, phytoestrogens, or other compounds that might leach from plastics used in cages, water bottles, or other housing devices [39–41]. Furthermore, our studies suggest that commercial laboratories are raising animals in environments that are not well defined when it comes to EDC exposures. We expect that these exposures are low enough that they are not affecting fertility or fecundity, as commercial animal suppliers are acutely aware of these health outcomes; our results here suggest that xenoestrogen exposures in vendor laboratories that have modest (or perhaps no) effect on reproductive outcomes could still influence other hormone-sensitive endpoints including the mammary gland.

It is theoretically possible to determine what chemicals the CRL animals were exposed to during early development. First, some chemicals with longer half-lives may still be present in the blood of these animals at the time of necropsy, or additional animals could be sacrificed immediately upon arrival for the collection of blood and tissues. It is important to note that collecting serum for these purposes can be difficult because of the potential for contamination, the requirement for large volumes of serum to do broad screening rather than measures for single compounds, and concerns that the most relevant exposures likely occurred during embryonic and fetal development and thus may no longer be present in the blood of animals that are weeks or months old [42–44]. Recent studies have shown that novel metabolomics methods can reveal persistent changes in metabolic “fingerprints” including variations in glucose, amino acids, and neurotransmitters in the serum of animals exposed to BPA during early development [45]. Altered metabolomic profiles were maintained for at least 180 days after exposures ceased [46]. These studies suggest the possibility that features of an animals’ serum could be used to not only determine what compound(s) they were exposed to, but also what dose. However, until such metabolomics profiles are available for the most common environmental contaminants, their use in the retrospective determination of chemical exposures will remain very limited.

Certainly, a priori screening of items in the animal facility is the best approach to avoid environmental chemical contamination of animals to be used in laboratory experiments. However, the ability to perform a posteriori evaluations to determine whether inadvertent contamination has occurred remains important, especially when new materials are introduced in animal care settings without the knowledge of the experimenter. In these cases, post hoc evaluative approaches using metabolomics are very resource intensive and require extensive training in sensitive molecular biology techniques. In contrast, analysis of mammary gland morphology is straightforward and inexpensive. For this reason, the tools we have demonstrated in this manuscript may prove to be sufficient to distinguish populations with distinct environmental histories that include chemical exposures during vulnerable periods of development.

There are two important consequences of our findings which are consistent with low-level xenoestrogen contamination in commercially available animals. The first arises from the use of adult animals for toxicity screening of environmental chemicals. If an experimental protocol requires that adult animals be exposed to test compounds, developmental exposures to EDC mixtures could interfere with the accuracy of this testing. There are a number of studies that have demonstrated that early life exposure to BPA alters responsiveness of animals to hormones or carcinogen challenges at puberty or in adulthood [35, 47–52]. This “two hit” model suggests that, for some endpoints, developmental exposures may not be sufficient to induce adverse effects, but the consequences of EDC exposures are best seen when a secondary challenge is experienced [22]. In this case, the uncontrolled EDC exposures that occur in commercial animal facilities could represent “hit one” and the controlled exposure experienced during toxicity testing would actually be “hit two”.

The second concern related to uncontrolled EDC exposures is the use of historical controls in toxicity testing in commercial laboratories. In some testing protocols, when the negative controls have unexpected responses, experimenters will instead use “historical controls”, e.g. unexposed controls collected in previous experiments, including experiments that may have been conducted years prior [53]. A report from the Historical Control Working Group from the Society of Toxicologic Pathology noted that “[s]tudy design-related parameters such as laboratory, species/strain, route of administration, vehicle, feed, feeding practices, study duration, and housing have a potential to impact study outcomes and control findings. These parameters should be considered when selecting the appropriate studies” to generate historical control data [54]. Yet, differences in background EDC exposures may occur even when experimenters think they are following the guidance of this working group. Other groups have highlighted differences in endocrine endpoints in historical controls from different testing facilities used to evaluate chemicals for endocrine disrupting activities [55, 56], suggesting that this issue may be more widespread than the single commercial laboratory we examined. Our results suggest that ‘control’ animals may not be unexposed; thus, using historical controls which might have one set of background EDC exposures to compare to experimental animals with an entirely different environmental history should not be acceptable practice [11].

Here, we compared mammary gland morphologies for two “control” populations with distinct environmental histories and found significant differences in adult males, prepubertal females, and pubertal females. These results suggest that the mammary gland may be a sensitive organ to probe populations for unknown environmental EDC exposures during vulnerable periods of development. Our results also dispute the use of historical controls because environmental EDC exposures may differ between experiments due to changes in cages, water bottles, animal feed, water, or other environmental factors that might otherwise appear benign.