The developmental origins of health and disease (DOHaD) hypothesis broadly considers how environmental influences incurred during important early developmental stages impart lasting deficits throughout life (Haugen et al., 2015). These deficits may include metabolic, neurocognitive, or structural effects including skeletal effects. The impact of the environment is not inconsequential as, according to the World Health Organization, almost one quarter of global deaths are linked to environmental factors. One subset of consequential environmental factors includes endocrine disrupting chemicals (EDCs), defined by the Endocrine Society as an exogenous chemical, or a mixture of chemicals, that interferes with any aspect of hormone action (Gore et al., 2015). Although, historically, the vast majority of EDC research has focused on reproductive and neural targets, emerging work suggests that metabolic reprogramming is also a primary consequence of developmental EDC exposure (Janesick and Blumberg, 2011; Schneider et al., 2014; Heindel et al., 2017; Sargis and Simmons, 2019). It has been hypothesized that, based on in vitro work, some EDCs can promote adiposity at the expense of osteoblast production from mesenchymal stem cells (Heindel et al., 2022). We have shown that developmental exposure to the commercial flame-retardant (FR) mixture Firemaster 550 (FM 550) sex-specifically compromises bone composition in Wistar rats (Macari et al., 2020). Here, we followed up on that aforementioned work to determine which chemical class in the commercial mixture may be driving the phenotype.

FRs are used in a myriad of household products such as carpets, baby and children’s products, electronics, building materials, and polyurethane foam used in gyms, furniture, and mattresses. As such, exposure is widespread and higher in children than in adults (Stapleton et al., 2008; Stapleton et al., 2012; Carignan et al., 2013). The use of FM 550 has changed dramatically in less than two decades. It was the second most detected FR in furniture products in the United States in the early 2000s, but its use appears to have decreased in favor of contemporary mixtures containing either a subset of FM 550 components with other FRs or the same components at different concentrations (Stapleton et al., 2008; Stapleton et al., 2009; Stapleton et al., 2012; Phillips et al., 2017). Thus, understanding the potential toxicity of the component classes is critically important. FM 550 is a roughly 50:50 mixture of brominated (BFR) and organophosphate ester (OPE) flame-retardant (OPFR) chemicals, some of which are also used in other applications, including personal care products (Mendelsohn et al., 2016; Phillips et al., 2017). In cellular and animal models, FM 550 has been characterized as obesogenic and endocrine-disrupting (Krumm et al., 2018; Heindel et al., 2022), with our group and others showing that developmental FM 550 exposure sex-specifically impairs neurodevelopment, socioemotional behaviors, and, possibly, skeletal development (Patisaul et al., 2013a; Belcher et al., 2014; Gillera et al., 2019; Macari et al., 2020; Witchey et al., 2020).

In vitro, FM 550 has been shown to interact with PPARγ to promote adipogenesis in mesenchymal stem cells at the expense of osteogenesis, even when cultured under pro-osteogenic conditions. Upon differentiating 3T3-L1 cells in the presence of triphenyl phosphate (TPHP), an OFPR found in the FM 550 mixture, lipid size and volume were increased and the expression of multiple pro-adipogenic genes was upregulated (Tung et al., 2017a; Tung et al., 2017b; Cano-Sancho et al., 2017; Stapleton et al., 2017). This suggests that FM 550 could shift metabolic programming toward adipose development at the expense of bones. Given the widespread use of FM 550 components and their known adverse effects during development on multiple systems, it is important to further understand their potential effects on skeletal health.

In fish models, abnormal skeletal development, gene network dysregulation, and gross morphological impairment of skeletal organization has been observed in response to developmental BFR or OPFR exposure (Macaulay et al., 2017; Alzualde et al., 2018). Aquatic species, such as medaka and zebrafish, are often used for high-throughput screening and are accepted models of skeletal development due to conserved pathways with human development (Padilla et al., 2009; Shanthanagouda et al., 2014; Witten et al., 2017; Bergen et al., 2019; Lleras-Forero et al., 2020). In medaka, developmental exposure to two OPFRs present in FM 550 induced abnormal pectoral fin development, spinal curvature, and dysregulation of BMP signaling networks critical for early cartilage and bone development (Hong et al., 2021). In zebrafish, major signaling pathways such as MAPK were also disrupted following OPFR exposure (Kling and Förlin, 2009). Both species also displayed sex-specific disruption of the hypothalamic–pituitary–gonadal (HPG) axis (Liu et al., 2013; Saunders et al., 2015; Zhu et al., 2018), as well as behavioral disruptions akin to those observed in mammalian models, demonstrating species concordance (Glazer et al., 2018; Gillera et al., 2019). Evidence of sex specificity has also been reported. For example, in response to exposure to an OPFR mixture, adult male zebrafish had increased corticotropin-releasing hormone (crh) and thyroid-stimulating hormone (tsh) transcription in the brain and downregulation of thyroglobulin and deiodinase 2 in the thyroid and liver, while females displayed downregulated crh and tsh (Wang et al., 2013; Kim et al., 2015; Liu et al., 2019).

In rodent models, gestational and lactational exposure to FR mixtures negatively affects neurodevelopment, liver and thyroid function, global metabolism, and skeletal development (Patisaul et al., 2013a; Phillips et al., 2016; Adams et al., 2020; Macari et al., 2020). Sprague–Dawley rat pups exposed to an environmentally relevant BFR-based mixture throughout gestation and lactation failed to completely ossify vertebrae and phalanges by PND 4 (Berger et al., 2014). In an ex vivo murine limb bud model, exposure to OPFRs or BFRs severely stunted the organization of COL10A1, an early marker of chondrogenesis in mammals that is critical to proper skeletal growth, and antagonized Hedgehog signaling (Yan and Hales, 2019; Yan and Hales, 2020). It is unclear if or how these deficits persist into adulthood.

Previously, our group showed in a pilot study that the gross bone mineral density and microarchitecture in cortical and trabecular structures was disrupted in male (but not female) Wistar rats that were gestationally and lactationally exposed to a single dose of FM 550 (1,000 μg/kg bw; ∼3.3 mg/kg bw to the dam) (Macari et al., 2020). FM 550-exposed males had altered histomorphology including a decreased number of osteoblasts and increased proportion of yellow marrow (Macari et al., 2020). No changes in osteoclasts were observed in either sex. While this aforementioned study established FM 550 as an environmental contaminant that can sex-specifically impair skeletal outcomes, it is unclear which chemical class in the mixture is primarily responsible for the effects or if it is a unique effect of the full mixture. To begin to address this, here, we assessed the effects of each FR class and the full mixture on cortical and trabecular bone and morphology of the femur in Wistar rats. The approach utilized advanced X-ray computed tomography instrumentation which was unavailable for our prior related study; thus, we included a subset of those animals in the present study to validate our updated approach (Macari et al., 2020).

The identification of serum biomarkers for compromised skeletal integrity would be useful for future studies probing the mechanism of action and potential human risk. Consequently, we examined circulating serotonin (5-HT) and osteocalcin (OCN) levels. Our group has previously shown that developmental FM 550 exposure alters serotonergic (5-HTergic) innervation of the rat fetal forebrain, disrupts placental tryptophan metabolism, and promotes anxiety in female rodents (Baldwin et al., 2017b; Rock et al., 2018). Although this may appear tangential to the present study objectives, 5-HT also impacts aspects of bone physiology. Bone is sensitive to most neurotransmitters, including 5-HT, which promotes bone accrual when produced by the brain and inhibits bone formation when produced peripherally by the gut (Westbroek et al., 2001; Yadav et al., 2008; Ducy and Karsenty, 2010; Pawlak et al., 2019). LRP5, a co-receptor for Wnt signaling, also controls 5-HT synthesis in the gut, and the inhibition of lrp5 decreases bone mass in mice (Yadav et al., 2008). 5-HT also controls the transcription factor FOXO1 which forms separate complexes with either ATF4, to promote osteoblast proliferation, or CREB, to inhibit proliferation (Rached et al., 2010; Pawlak et al., 2019). The ATF4 complex is favored in environments of increased 5-HT (Rached et al., 2010; Pawlak et al., 2019). Similarly, the non-collagenous protein hormone OCN is produced and secreted exclusively by osteoblasts and primarily acts locally to influence matrix mineralization in skeletal tissue and can influence whole-system metabolism (Guntur and Rosen, 2012; Agas et al., 2013; Oldknow et al., 2015; Wei and Karsenty, 2015; Shan et al., 2019). The main osteocalcin target receptor, Gpr158, is expressed in the hypothalamus and pituitary gland, where it acts to promote spatial learning and memory and inhibits anxiety-like behavior.

Because, in its own internal studies, the chemical manufacturer found adverse skeletal effects in rats exposed to only the BFR components, albeit at doses at much higher exposures, we hypothesized that the BFRs could be most consequential (MPI Research, 2008). At doses of 100 mg/kg/day, inappropriate fusion of cervical vertebral neural arches was observed, and at doses of 300 mg/kg/day, further ossification variations in cervical vertebral neural arches, incomplete ossification of the skull, and unossified sternum were observed (MPI Research, 2008). We also hypothesized that the BFRs and OPFRs might induce unique sex- and FR class-specific effects on bone morphology and circulating serum markers, given their propensity to be metabolically disruptive in vitro and in zebrafish assays.

Gestational and lactational exposure to FM 550 or its OPFR or BFR components resulted in sex- and chemical-specific evidence of altered skeletal tissue organization in both the full and distal femur analyses. Unexpectedly, neither the OPFRs nor the BFRs produced a profound phenotype, suggesting that each contributes aspects to the effects of the full mixture. This finding is significant because the OPFR and BFR components are used in a wide variety of applications in different combinations. The mechanisms by which each class contributes to the structural phenotype remain to be established. The FM 550 phenotype in males was similar but not as severe as in our prior study using a lower FM 550 dose and assessing animals as younger adults (Macari et al., 2020). Critically, we show that we were able to recapitulate the phenotype using the revised approach on a subset of the males from the prior study and demonstrate that the discordance is not due to technical issues related to the updated CT approach but, instead, likely results from other experimental differences, including the older age at analysis and the higher FM 550 dose used. A comprehensive study incorporating additional doses and ages at examination to fully characterize the trajectory of the phenotype over the dose and life stage is merited. The present study expands on our prior study by elucidating how OPFR and BFR components of the FM 550 mixture sex-specifically disrupt skeletal development and by contributing hazard information at a higher dose. Our data also add to the limited but growing body of literature focusing on bone as a potential target for endocrine disruption.

Fetal skeletogenesis in the rat begins on GD 15, and the skeleton is fully formed by PND 2 (DeSesso and Scialli, 2018). Rapid bone growth and development occur postnatally with the appearance of the primary ossification center in the femur on PND 8. Thus, our exposure window from GD 0 to PND 21 fully encompassed all sensitive time points for early skeletal development (DeSesso and Scialli, 2018). Importantly, the parent study from which the animals were obtained was designed to focus primarily on behavioral outcomes, many of which rely on healthy motor function. Those data showed no indication of any gross mechanical impairment, and no animals showed any signs of impaired movement or mobility during behavioral tasks conducted on or around PND 90 (Witchey et al., 2020). Additionally, no evidence of impaired gross motor function was observed in any of our other prior behavioral studies on FM 550 (Patisaul et al., 2013b; Baldwin et al., 2017a; Gillera et al., 2022). Regardless, disrupted femoral skeletal organization was apparent in PND 250 FM 550 males in our prior study and less so in the older males tested herein. Collectively, this suggests that structural effects may be more pronounced in younger animals, particularly males, but are not severe enough to influence gross motor activity. Ex vivo biomechanical testing would be useful to test this posit across development in future studies, as would the assessment of fine motor activity. Vulnerability to skeletal degeneration at an advanced age is also a possibility that should be examined in future studies.

While sex-specific effects were anticipated, why the effects in females were significant here and not in the prior study is unclear, as is why females appeared to be more greatly affected by the OPFRs, while the males appeared to be more affected by the BFRs. Two main differences between the present and our prior skeletal study are dose and age at testing. One hypothesis regarding the males is that the phenotype previously observed is not produced at the higher dose used here due to yet-to-be described compensatory mechanisms not triggered at lower exposures. Non-monotonic dose responses (NMDRs) commonly occur with EDCs via multiple mechanisms (Vandenberg et al., 2012). In some instances, repair mechanisms are activated at higher, but not lower, more human-relevant doses, and this may be the case here. In nuclear hormone receptor binding assays, FM 550 shows a strong NMDR at PPARy, which is suggested to be driven by the OPFR components in the mixture (Belcher et al., 2014). Alternatively, the animals in this study were approximately 3 months older than the animals assessed in the prior study. Thus, it is possible to resolve the male skeletal phenotype through growth and remodeling.

Why a phenotype emerged in the older females when we did not detect one previously in younger adults is unknown. Estrogen is largely protective of bone health; thus, one hypothesis is that gonadal estrogens may mask exposure-related effects in younger females that recede with age as estradiol (E2) levels naturally decline (Imai et al., 2013). In males, estrogens are also protective. ERα-mediated signaling is critical for skeletal maintenance, with aromatase producing estradiol locally from circulating testosterone (Bouillon et al., 2004; Sinnesael et al., 2011; Imai et al., 2013). Aromatase knockout (ArKO) mice display a strong sex-specific skeletal phenotype, with ArKO females showing higher rates of bone turnover than males (Oz et al., 2000; Oz et al., 2001). The assessment of ER levels in bone was beyond the scope of this study but should be examined in future works.

Other EDC studies report evidence of improvement with age. In a similar study with gestational and lactational bisphenol-A (BPA) exposure, skeletal insults recorded in males at 5 weeks of age were resolved by 52 weeks, suggesting that the phenotype may be reversible or resolved through remodeling as the animal ages (Lind et al., 2017; Lind et al., 2019). Similar studies with bone-damaging pharmaceuticals also show that skeletal recovery following insult is possible. For example, intraperitoneal injection of a chemotherapy agent induced apoptosis and inhibited cartilage proliferation at the growth plate in 6-week-old male rats (Xian et al., 2004; Xian et al., 2006). However, by 10–14 days post-exposure, the proliferative and hypertrophic zones of the growth plate returned to normal thickness, and by 4–10 days post-exposure, expression of bone matrix proteins and growth factors increased (Xian et al., 2006). Future studies will assess animals at earlier stages of development (<PND 100) to establish a timeline for the skeletal phenotype in each sex. The present studies were conducted during the peak of COVID-related restrictions, and associated logistical limitations prevented planned histological assessment of skeletal tissue in this study; hence, the quantification of osteoblast, yellow marrow, and osteoclast populations will be desirable in future studies.

The effects of FR exposure on serum biomarkers were minimal, but the identification of expected sex differences confirms that the study was sufficiently powered to detect biologically meaningful effects. Expected sex differences in serum calcium and ALP levels were found, with males having higher ALP and lower serum calcium levels than females. FM 550-exposed females had higher concentrations of osteocalcin than controls, and BFR-exposed males had increased serum calcium compared to controls. Osteocalcin is produced exclusively from osteoblasts and primarily deposited in the bone matrix. Increased osteocalcin levels in the FM 550 females could suggest increased bone turnover activity but would need to be confirmed (Agas et al., 2013; Hlaing and Compston, 2014). Because increased serum calcium can result from many biological factors, why serum calcium was higher in BFR-exposed males is unclear. Involvement of the thyroid hormone and parathyroid hormone is a possibility, as thyroid hormone disruption is a well-known consequence of BFR exposure (Darnerud, 2008; Wang et al., 2019). Increased OCN in FM 550-exposed females and increased calcium in BFR-exposed males do not correlate with any single skeletal phenotype. However, the relationship between the sex and exposure group is consistent with the previous results for skeletal effects, where BFR exposure more commonly affected males and FM 550 exposure more commonly affected females.

The sex-specific effects of developmental FR exposure found are consistent with known sex differences in skeletal physiology and many other EDCs. In mice exposed to either 200 μg/kg body weight BPA or bisphenol-S (BPS) orally via the dam throughout gestation and lactation, only the BPA-exposed males had an impaired skeletal structure (Dirkes et al., 2021a; Dirkes et al., 2021b). The effects on females were minimal. Similarly, male and female mice exposed through diet to 500 μg/kg diethylstilbestrol (DES) or greater had increased trabecular bone formation in the proximal femur; however, only males showed increased bone formation in the sternum (McAnulty and Skydsgaard, 2005). In a rat model of developmental exposure, dams were exposed to approximately 0.06, 20, and 60 mg/kg body weight/day of a PBDE and hexabromocyclododecane (HBCD) mixture incorporated in chow, throughout gestation and lactation (Berger et al., 2014). Offspring exposed to any of the BFR doses had increased incidents of skeletal variations including incomplete ossification of the sternum or fusion of the vertebral column at GD 20 (Berger et al., 2014). On PND 4, offspring had delayed ossification in phalanges and cervical vertebrae, although neither were sex-specific effects reported nor were later stages of development assessed for persisting skeletal malformations (Tung et al., 2016). All serum biomarkers examined showed no difference in control levels by PND 210, an outcome concordant with our serum data at PND 250 and supporting the hypothesis that the adverse early-life skeletal effects can improve with age (Tung et al., 2016).

To the best of our knowledge, skeletal outcomes of FM 550 exposure have not been assessed in human studies, although a limited but growing number of studies include bone structure assessment in populations exposed to broader categories of EDCs. Globally, BFR and OPFR chemicals are routinely detected in biological tissues such as nail clippings, urine and feces in adults and toddlers, and breastmilk in women (Norén and Meironyté, 2000; Hoffman et al., 2014; Kim et al., 2014; Sahlström et al., 2015). The average human exposure to FM 550 or its OPFR and BFR components varies widely, partially due to consumer and lifestyle choices and occupational exposure, although the two brominated chemicals in the BFR portion of FM 550, EH-TBB and BEH-TBPH, were quantified at geometric mean concentrations of 315.1 and 364.7 ng/g, respectively, in house dust samples of homes in the United States (Stapleton et al., 2008; Hoffman et al., 2014). In a similar study, the geometric mean concentration of TPHP in US house dust extract was quantified at 7,360 ng/g (Stapleton et al., 2009). Although controversial, the ratio between the length of the second and fourth digits is sometimes used as a population-level indicator of prenatal androgen exposure, with a higher digit ratio suggesting feminization (Lutchmaya et al., 2004; Zheng and Cohn, 2011; Auger et al., 2013). A single study found that toddlers prenatally exposed to PBDEs had a higher second digit-to-fourth digit ratio as measured at 4 years of age (Chen et al., 2021b). While digit ratios are not a conclusive marker of skeletal disruption, it is interesting to note that incomplete ossification in the phalanges was observed in young rats following gestational and lactational PBDE exposure (Berger et al., 2014).

The mechanisms by which FM 550 and its component impair skeletal physiology remain to be elucidated. As a mixture, Firemaster 550 has significant agonistic effects on PPARy, which favors adipogenesis at the expense of osteogenesis, an effect largely driven by the OPEs (Belcher et al., 2014). The BFRs, by contrast, may exert their effects via thyroid hormone disruption (Darnerud, 2008; Legler, 2008). RNA sequencing of cultured limb buds from mice exposed to 10 µM tert-butylphenyl diphenyl phosphate for 3 h revealed downregulated transcripts including gli1 and runx3 (Yan and Hales, 2020). Following 24 h of exposure, downregulation of transcripts involved in Hedgehog signaling was further amplified, suggesting a dose-responsive effect (Yan and Hales, 2020). Gli1 is a transcription factor belonging to the family of critical Gli proteins in the Hedgehog signaling pathway including endochondral ossification, and runx3 is critical for chondrocyte maturation also involved in this pathway (Yoshida et al., 2004; Day and Yang, 2008; Ohba, 2020). Tracing the lineage of Gli1-positive cells in mice revealed that Gli1 marks osteoprogenitor cells involved in bone development and repair (Shi et al., 2017). As most bone development and repair is mediated by chondrocytes and deposition of a cartilage matrix, the downregulation of runx3 may suggest that chondrocyte maturation may be susceptible to OPFR disruption (Yoshida et al., 2004).

Finally, it is possible that vascular attributes could be contributing to the observed phenotypes. Vasculature plays a critical role in maintaining skeletal homeostasis and integrity throughout life, and sex differences in skeletal morphology are considered to be at least partially influenced through the vasculature (Ahluwalia et al., 2014; Jani and Rajkumar, 2006; Kaigler et al., 2003; Ramasamy et al., 2016; Owen-Woods and Kusumbe, 2022; Eriksen, 2010; Goring et al., 2019). Endothelial cells (ECs) release angiocrine factors, which include vascular endothelial growth factors (VEGFs) and fibroblast growth factors (FGFs), to promote and regulate bone growth (Hu and Olsen, 2016; Kaigler et al., 2003; Ramasamy et al., 2016; Owen-Woods and Kusumbe, 2022). Limited studies exist on the ability of FRs to impact EC function, although BFRs and OPFRs have been shown to induce apoptosis and genotoxicity in human umbilical vein ECs (Chen et al., 2020; Saquib et al., 2021). Deletion of bone-derived VEGF in osteoblasts obtained from male mice produced immature mineral precursors in culture, whereas it produced mature mineral precursors from female-derived osteoblasts, suggesting a role of VEGF in influencing the skeletal matrix independent of sex hormones in a sex-specific manner (Goring et al., 2019).

Incorporation of bone structure assessments into a toxicology testing framework remains relatively rare. Thus, a significant risk to human health from developmental FR exposure, and EDC exposure generally, may possibly be undetected. Hence, our study fills a need for data on how early-life chemical exposures influence adult skeletal outcomes. We found that developmental FM 550 exposure via the dam causes lasting but limited skeletal effects in rats, which are both sex- and chemical class-specific. Adult rats (PND 250) showed deficits in adult femur organization including significantly decreased marrow area, increased bone volume fraction in FR-exposed males, and decreased bone mineral density in FR-exposed females. The presence of skeletal effects in late adulthood suggests the disruption of pathways involved in early skeletal organization, but this needs to be examined in younger animals, as well as the potential mechanisms by which these adverse effects occur. The FRs examined have been shown to disrupt estrogen and thyroid hormone signaling, and this is a relevant point to consider since both estrogen and thyroid hormones are critically involved in skeletal development and maintenance. Further examination of other long bone components, especially the vasculature, is also needed to better elucidate the complete phenotype in each sex across the lifespan.