AOP key event relationship report: Linking androgen receptor antagonism with nipple retention

Graphical abstract


Introduction
Areola/nipple retention (NR) in male rat, or mouse, offspring is considered a biomarker for incomplete masculinization during fetal development. This is because male rats and mice normally do not display nipples, in contrast to female rats and mice that have 12 and 10 nipples, respectively Mayer et al., 2008). This sexual dimorphism is believed to be largely due to differences in androgen signaling during development and thus NR in males can be considered a readout for compromised androgen action during critical developmental stages, as recently reviewed (Schwartz et al., 2021). Consequently, NR is included as a mandatory endpoint in several OECD test guidelines (OECD, 2008, OECD, 2018 on assessment of developmental and reproductive toxicity, not least to detect antiandrogenicity. As NR is measured in vivo, it is desirable to describe in  transcription factor belonging to the steroid hormone nuclear receptor family (Davey & Grossmann, 2016). The AR has three domains: the Nterminal domain, the DNA-binding domain, and the ligand-binding domain, with the latter being most evolutionary conserved. Apart from the essential role AR plays for male reproductive development and function (Walters et al., 2010), the AR is also expressed in many other tissues and organs such as bone, muscles, ovaries and the immune system (Rana et al., 2014).

AR antagonism as key event
The main function of the AR is to activate gene transcription in cells. Canonical signaling occurs by ligands (androgens) binding to AR in the cytoplasm which results in translocation to the cell nucleus, receptor dimerization and binding to specific regulatory DNA sequences (Heemers and Tindall, 2007). The gene targets regulated by AR activation depends on cell/tissue type and what stage of development activation occur, and is, for instance, dependent on available co-factors. Apart from the canonical signaling pathway, AR can also function through non-genomic modalities, for instance rapid change in cell function by ion transport changes (Heinlein & Chang, 2002). However, with regard to this specific KE the canonical signaling pathway is what is referred to.

How it is measured or detected
AR antagonism can be measured in vitro by transient or stable transactivation assays to evaluate nuclear receptor activation. There is already a validated assay for AR (ant)agonism adopted by the OECD, Test No. 458: Stably Transfected Human Androgen Receptor Transcriptional Activation Assay for Detection of Androgenic Agonist and Antagonist Activity of Chemicals (OECD, 2020). The stably transfected AR-EcoScreen TM cells (Satoh et al., 2004) should be used for the assay and is freely available for the Japanese Collection of Research Bioresources (JCRB) Cell Bank under reference number JCRB1328.
Other assays include the AR-CALUX reporter gene assay that is derived from human U2-OS cells stably transfected with the human AR and an AR responsive reporter gene (van der Burg et al., 2010), the MDA-kb2 cell line (Wilson et al 2004) and various other transiently transfected reporter cell lines (Körner et al., 2004), and more. Recently developed AR dimerization assay may soon be included in TGs for its improved ability to measure potential stressor-mediated dimerization/ activation (Lee et al., 2021).

Domain of application Overview
Both the DNA-binding and ligand-binding domains of the AR are highly evolutionary conserved, whereas the transactivation domain show more divergence which may affect AR-mediated gene regulation across species (Davey & Grossmann, 2016). Despite certain inter-species differences, AR function mediated through gene expression is highly conserved, with mutations studies from both humans and rodents showing strong correlation for AR-dependent development and function (Walters et al., 2010).

Evidence for perturbation of this MIE by stressor
A large number of drugs and chemicals have been shown to antagonize the AR using various AR reporter gene assays. The AR is specifically targeted in AR-sensitive cancers, for example the use of the antiandrogenic drug flutamide in treating prostate cancer (Alapi & Fischer, 2006). Flutamide has also been used in several rodent in vivo studies showing anti-androgenic effects (feminization of male offspring) evident by e.g., short anogenital distance (AGD) in males (Foster & Harris, 2005;Hass et al., 2007;Kita et al., 2016). Quantitative Structure-Activity Relation (QSAR) models can predict AR antagonism for a wide range of chemicals, many of which have shown in vitro antagonistic potential (Vinggaard et al., 2008).

Key event description
In common laboratory strains of rats and mice, females typically have 6 (rats) or 5 (mice) pairs of nipples along the bilateral milk lines. In contrast, male rats and mice do not have nipples. This is unlike e.g., humans where both sexes have 2 nipples (Schwartz et al., 2021).
In laboratory rats, high levels of dihydrotestosterone (DHT) induce regression of the nipples in males (Imperato-McGinley & Gautier, 1986;Kratochwil, 1977;Kratochwil & Schwartz, 1976). Females, in the absence of this DHT surge, retain their nipples. This relationship has also been shown in numerous rat studies with perinatal exposure to antiandrogenic chemicals (Schwartz et al., 2021). Hence, if juvenile male rats and mice possess nipples, it is considered a sign of perturbed androgen action early in life.

How it is measured or detected
Nipple retention (NR) is visually assessed, ideally on postnatal day (PND) 12/13 (OECD, 2018; Schwartz et al., 2021). However, PND 14 is also an accepted stage of examination (OECD, 2013). Depending on animal strain, the time when nipples become visible can vary, but the assessment of NR in males should be conducted when nipples are visible in their female littermates (OECD, 2013).
Nipples are detected as dark spots (or shadows) called areolae, which resemble precursors to a nipple rather than a fully developed nipple. The dark area may or may not display a nipple bud . Areolae typically emerge along the milk lines of the male pups corresponding to where female pups display nipples. Fur growth may challenge detection of areolae after PND 14/15. Therefore, the NR assessment should be conducted prior to excessive fur growth. Ideally, all pups in a study are assessed on the same postnatal day to minimize variation due to maturation level (OECD, 2013).
NR is occasionally observed in controls. Hence, accurate assessment of NR in controls is needed to detect substance-induced effects on masculine development (Schwartz et al., 2021). It is recommended by the OECD guidance documents 43 and 151 to record NR as a quantitative number rather than a qualitative measure (present/absent or yes/no response). This allows for more nuanced analysis of results, e.g., high control values may be recognized (OECD, 2013, OECD, 2018. Studies reporting quantitative measures of NR are therefore considered stronger in terms of weight of evidence. Reproducibility of NR results is challenged by the measure being a visual assessment prone to a degree of subjectivity. Thus, NR should be Table 2 List of chemicals causing NR in male rat offspring (in vivo) due to exposure to an AR antagonist during development. Several of the chemical stressors have also been shown to antagonize AR in vitro; these are noted in the far-right column. Additional information, including species, strain, exposure period, time of NR measurement as well as the magnitude of NR at the effect dose is presented. * (p < 0.05).  assessed and scored blinded to exposure groups and ideally be performed by the same person(s) within the same study.

Biological domain of applicability
The applicability domain of NR is limited to male laboratory strains of rats and mice from birth to juvenile age.

Regulatory significance of the adverse outcome
NR is recognized by the OECD as a relevant measure for antiandrogenic effects and is mandatory in the test guidelines Extended One Generation Reproductive Toxicity Study, TG 443 (OECD, 2018) and the two screening studies for reproductive toxicity, TGs 421/422 (OECD, 2016a, OECD, 2016b). The endpoint is also described in the guidance documents 43 (OECD, 2008) and 151 (OECD, 2013). Furthermore, NR data can be used in chemical risk assessment for setting the No Observed Adverse Effect Level (NOAEL) as stated in the OECD guidance document 151 (OECD, 2013): "A statistically significant change in nipple retention should be evaluated similarly to an effect on AGD as both endpoints indicate an adverse effect of exposure and should be considered in setting a NOAEL".

Stressors
An overview of chemical stressors causing nipple retention in male rats is provided in the main text and supplementary data (S1, Table 2) of Schwartz et al. (2021). The stressors and intrauterine exposure levels resulting in NR are listed below (Table 1).

Biological domain of applicability
Taxonomic applicability: Rats and mice. Life stage applicability: Developmental. Sex applicability: Male.

KER description
Several chemicals can antagonize the androgen receptor (AR) in vitro, resulting in decreased AR activation. Decreased AR activation can lead to incomplete reproductive development in males, which can be expressed in several ways. One endpoint affected is areola/nipple retention (NR), which in vivo studies have shown to be linked to suppressed AR activation. NR in rat and mouse toxicity studies is considered an adverse effect (i.e., an AO).

Biological plausibility
The biological plausibility of a link between decreased AR activation and increased NR in male rats is high. The relationship is supported by numerous studies showing that several potent AR antagonists in vitro induce NR in vivo. However, in the literature review conducted for this KER, no studies in mice were found to fulfill the inclusion criteria. The present KER is hence exclusively a description of the situation in rats, although it is believed that the link also exists in mice.
The AR is activated through binding of either testosterone or dihydrotestosterone (DHT), the latter having the highest affinity for the AR. Upon binding, the AR translocates to the target cell nucleus where it acts as a transcription factor (Albert, 2018).
NR has been shown to be more dependent on DHT-signaling, which suggests that chemicals inducing increased NR also have a higher affinity for the AR than DHT in order to outcompete DHT for AR binding, although supra-high doses of chemicals with lower AR affinity could be speculated to also outcompete T or DHT. The general principle of higher affinity, however, has been confirmed by in vitro studies (Gray et al., 2019;Hass et al., 2012;McIntyre et al., 2000). Table 2 lists chemical stressors shown to antagonize the AR in vitro as well as causing NR in male rat offspring in vivo. Additional information from the in vivo studies, including the animal species and strain, as well as the doses tested, the dosing period and the time of measurement of NR are specified in this table. The lowest dose yielding a significant increase of retained nipples in male rat pups is defined as the LOAEL. Conversely, the NOAEL represents the highest tested dose yielding no significant increase in NR. Note that the given NOAEL and LOAEL values are highly dependent on study design. Significant values are marked with an asterisk. Table 3 shows a list of stressors shown to have AR antagonistic properties in vitro or in other in vivo studies, but for which the doses tested in vivo did not produce a significant effect on NR. In this list, the lowest tested dose is reported, and the NOAEL presents the highest dose tested which produced no statistically significant effect on NR. Apart from the NOAEL, the information given in Table 3 is identical to Table 2.
the same study. To minimize these sources of uncertainty, assessors must be trained to recognize areolae and not look for fully developed nipples. Moreover, the number of assessors should be limited to one or two, and they should always be blinded to exposure groups. Another factor that may affect NR results is the age of the rat pups at the time of assessment. OECD guidelines have standardized the time for measuring occurrence of NR to be optimal at PD 12 or 13, when they are visible in female littermates (OECD, 2013). However, assessment of permanent NR is not included in any international guidelines. Hence, if NR is measured in older offspring, the time of measurement is not consistent between studies and varies between PD 20 and PD 100. Thus, conclusions on whether NR is permanent or not may differ based on study design. This distinction between a transient and a permanent effect is important from a regulatory perspective, since only a permanent effect will be categorized as a malformation according to OECD guidance document 43 (OECD, 2008).

Quantitative understanding
The quantitative understanding of the relationship between decreased AR activity and NR is challenged by the fact that the potency of AR antagonism in vitro is not proportional to the magnitude of NR observed in vivo (Gray et al., 2019). Hence, predicting in vivo effects based on in vitro data is not yet possible. However, in vitro studies can give indications of which chemicals might exhibit anti-androgenic effects in vivo and should be subject to further testing (Hass et al., 2012). Development of more representative in vitro models is necessary if in vivo tests are to be phased out entirely.
Timescale NR manifest in juvenile male rat pups in response to reduced androgen signaling, e.g. resulting from exposure to an anti-androgenic chemical stressor during fetal life. Developmental sensitivity during fetal development is highest during the so-called male masculinization programming window (MPW) which in rats is between gestational day (GD) 15 and 19 (Welsh et al., 2008).
A study in which pregnant rat dams were exposed to the AR antagonist vinclozolin for two-day periods during gestation showed that GD 16-17 was the most sensitive period for increased NR in male offspring (Wolf et al., 2000). A similar study using di-n-butyl phthalate (reduces testosterone levels) also showed that GD 16-17 was the most sensitive period for increased NR in male rats (Carruthers & Foster, 2005). However, to determine if other chemical stressors also have the highest antagonistic potential towards the AR during GD 16-17, further studies with a similar design would be informative.
NR can only be recorded when pups are old enough to display them, yet before excessive fur has developed. Hence, the most accurate results can be obtained from assessing the number of nipples on PD 12-14 depending on rat strain and the time of female littermates displaying nipples (OECD, 2013).

Known modulating factors
One factor that may influence NR counts in toxicity studies is the rat strain. In the studies included for development of the present KER, Wistar and Sprague-Dawley rats are the most widely used. Additionally, some studies have reported effects in Long Evans hooded rats and Holtzman rats.
An extensive review on NR effects reports no major differences on the magnitude of effect between Sprague-Dawley and Wistar rats (Schwartz et al., 2021). Thus, results from the two rat strains appear comparable. However, attention should be paid when comparing of NR between Sprague-Dawley and Long Evans hooded rats. For example, when exposed to flutamide or p,p'-DDE during GD 14-18, the Sprague-Dawley and Long Evans strains are equally sensitive to flutamide exposure, but Sprague-Dawley rats are more sensitive towards exposure to p,p'-DDE (You et al., 1998). The LOAEL for p,p'-DDE exposure in Sprague-Dawley rats was estimated to be 10-fold lower than in Long Evans rats. This finding is supported by another study showing that Sprague-Dawley rats present 4 times as many nipples as Long Evans hooded rats when exposed to 100 mg/kg bw/day p,p'-DDE during GD 14-18 (Wolf et al., 1999).

Response-response relationship
No response-response relationship has been identified.

Known feedback loops influencing this KER
No feedback loops that could influence the KER have been identified.

Classification of quantitative understanding
The quantitative understanding of the present KER remains low.

Conclusion
We conclude that a causal relationship exists between the KEs "androgen receptor (AR) antagonism" and "areola/nipple retention" in male rat offspring. These are non-adjacent KEs essentially linking a MIE with an AO. Based on a semi-systematic review of available literature, evidence for this qualitative relationship is strong. Methods for reliably quantifying how much the AR must be inhibited before significant NR manifest remains insufficient. Available in vivo studies report NR with a high degree of variability, which can be attributed to different sources of uncertainty. It is hence not possible to isolate and correct for confounding factors at present. Further studies following OECD guidelines Table 3 List of chemicals that caused no significant effect on NR in vivo despite being known to have AR antagonistic properties in in vitro studies or previous in vivo experiments. The highest dose tested that led to no significant effect is presented as the NOAEL. Additional information, including species, strain, exposure period, time of NR measurement as well as the magnitude of NR at the NOAEL is presented. Based on semi-systematic literature review. Abbreviations: SD = Sprague-Dawley; GD = Gestational Day; PD = Pup Day; NOAEL = No Observed Adverse Effect Level. p,p'-DDE, dichlorodiphenyldichloro ethylene. a This study had a control group with NR = 2.08, which can explain the non-significance compared to the exposure group despite the high NR value. and, ideally, conducting both in vitro and in vivo assays are necessary before a general model can be developed and applied for predictive toxicology with quantitative power. OECD test guidelines relevant to AOP 344 include in vitro and in vivo detection of AR antagonism (TG 458 and 441, respectively) and rodent reproductivity toxicity studies assessing NR (TG 443; TG 421/422) (OECD, 2009, OECD, 2016a, OECD, 2016b, OECD, 2018, OECD, 2020.

Funding
This work was funded by grants from The Danish Veterinary and Food Administration (DVFA), Project "Feminix", and The Nordic Working Group on Chemicals, Environment and Health (NKE) subgroup Nord UTTE (project 2020-027 and 2021-020).

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.