The POPs mixture, previously described (Lessard et al., 2019), was designed to resemble the contaminant composition and concentration present in northern Canadian populations (Muir, 1995; Anas et al., 2005). The experimental dose, therefore, represents an environmentally relevant exposure and is reported as 500 µg PCBs/kg body weight plus other POPs in relative proportions. The detailed composition of the mixture is presented in Table 1.

As described previously (Herst et al., 2019; Lessard et al., 2019; Navarro et al., 2019), the animals were fed diets based on the AIN-93G formula, designed to promote growth, pregnancy, and lactational phases (Reeves et al., 1993) and the diets differed only by their level of FA. The FA control diet contained 2 mg FA/kg diet, reflecting the AIN-93G diet, and the supplemented FA diet contained 6 mg FA/kg diet (# 110700 and #117819 respectively by Dyets, Inc., Bethlehem, PA). The FA control diet was assumed to approximate the recommended dietary intake for human adults (0.4 mg/day), whereas the supplemented FA diet approximated intake by a Canadian woman in the postfortification era from FA-enriched foods and prenatal supplements (≈1.2 mg/day) (Swayne et al., 2012a).

All animal protocols were conducted in accordance with the guidelines of the Canadian Council on Animal Care and approved by the Université Laval Animal Research Ethics Committee (certificate # 2015010-2) as described previously (Herst et al., 2019; Lessard et al., 2019; Navarro et al., 2019). Thirty-two Sprague-Dawley female rats (aged 45 days and weighing 140–180 g) (Charles River Laboratories; Saint-Constant, Quebec, Canada) constituted the F0 founder dams and were pair-housed in standard cages at 22°C (50% humidity) with 12-h light-dark cycle. Food and water were provided ad libitum. After 10 days of acclimatization, the F0 females were randomly assigned to one of four treatments groups (Figure 1A): 1) the control group (“CTRL”; n = 8) was gavaged with corn oil (no POPs) and fed the control FA diet; 2) the POPs-treated group (“POPs”; n = 8) was gavaged with the POPs mixture and fed the control FA diet; 3) the no POPs-FA supplemented group (“FAS”; n = 8) was gavaged with corn oil and fed the FA supplemented diet; 4) the POPs-treated and FA-supplemented group (“POPs + FAS”; n = 8) was gavaged with the POPs mixture and fed the FA supplemented diet.

The F0 females were weighed and gavaged thrice weekly with corn oil (CTRL and FAS groups) or the POPs mixture (POPs and POPs + FAS groups) for 5 weeks before mating with untreated males (no POPs, control FA diet), aged 90 days and weighing 200–240 g (Charles River Laboratories). Mating was confirmed by the presence of spermatozoa in vaginal smears, which were performed every morning, and the day a female was found sperm-positive was designated Gestational Day 0.5 (GD 0.5). Pregnant females were housed individually, and gavaged trice weekly until parturition. After the birth of the F1 litters, all F0 females and subsequent generations received the FA control diet. At Postnatal Day (PND) 21, the pups were weaned, and 12 randomly selected males (two for each of the six F0 dam that gave birth) were kept and pair-housed (a schematic of this experimental design is shown in Figure 1B).

At PND 90, F1 males were mated with untreated (no POPs, FA control diet) nulliparous and sexually mature females (aged 63–74 days, Charles River Laboratories) to generate F2 (Figure 1B). The F3 and F4 lineages were generated in the same manner. At GD 19.5, females were sacrificed to collect fetuses (F1: n = 2 pregnant females per group; F2-F3-F4: n = 6 pregnant females per group) and the remaining females gave birth (n = 6).

In the current study, F0 mothers received the interventions during pregnancy, therefore the F1 and F2 pups were directly exposed (intergenerational exposure), through the placenta and lactation for the F1 pups, and through the germ cells of the F1 fetuses for the F2 pups. The F3 and F4 generations, however, were not exposed to the treatment and will hereafter be referred to as transgenerational exposure.

Dams were anesthetized with 3% isoflurane, then euthanized by exsanguination via cardiac puncture and CO₂ asphyxiation. The fetuses were collected by caesarean-section at GD19.5. This time point was chosen to ensure that malformed fetuses were assessed that could have otherwise been cannibalized at birth by the dam. Eight fetuses per group (2♂ and 2♀ each from 2 different litters) were collected in F1, and 12 fetuses per group (1♂ and 1♀ each from 6 different litters) were collected in F2-F3-F4. Fetuses were sexed by measuring the anogenital distance with an electronic micrometer, then confirmed by the presence of testes or ovaries after abdominal dissection. The fetuses were placed in 95% ethanol at room temperature for maceration and storage until staining.

Double staining of the fetal bone and cartilage was performed with alcian blue and alizarin red S stains. Our protocol was adapted for rats by empirically modifying dye concentration, incubation time and KOH concentration from previously described procedures (Menegola et al., 2001; Lambrot et al., 2013; Rigueur and Lyons, 2014). Briefly, fetuses (GD19.5) were removed from 95% ethanol and put in a warm bath for up to a minute to facilitate removal of the skin, muscles, and viscera. The cleared skeletons were placed in 100% acetone overnight, and then stained with alcian blue solution 0.03% (80% ethanol, 20% acetic acid) for 12 h. The fetuses were washed in 70% ethanol to remove the blue dye and stored in 95% ethanol overnight. The skeletons were then transferred into a 2% KOH solution for 24 h, after which they were stained in the alizarin red S solution 0.005% (KOH 1%) for 4 h. The double-stained skeletons were stored in a glycerin and 70% ethanol solution (1:1).

The degree of ossification and, where possible, bone lengths were quantitatively assessed for skeletal sections (head, sternum, vertebrae, and limbs). Data were scored based on Kaufman (1994) and Menegola et al. (2001), and presented as a percentage (total score per section/maximum score). Analysis of the head ossification was divided into three measures: the skull bones (frontal, parietal, interparietal, supraoccipital, exoccipital, and basioccipital), the mandible and the premaxilla. Sternum development was evaluated by the degree of sternebrae ossification. The spinal evaluation comprised the cervical, thoracic, lumbar, and sacral vertebrae. Both front and hind paws were examined at distal, intermediate, and proximal phalanges. The metacarpals for the front paws and the metatarsals for the hind paws were evaluated. The front and hind paws, each composed of five digits, were numbered from 1 to 5, the thumb being 1 and the smallest digit being 5. The analysis was based on the average of left and right front and hind paws for each individual.

The bone lengths were measured with ImageJ (Image Processing and Analysis in Java, public domain, version 1.52 m) as shown in Figure 2. These included measures of head structures (craniofacial distance and mandible length) and the length of the main long bones (humerus, radius, femur, and tibia).

For each treatment group, data are expressed as mean ± standard error of the mean (SEM). One-way analysis of variance (ANOVA) for a 2 × 2 factorial design using the MIXED procedure in SAS^® University 9.4 Edition (SAS Institute Inc., Cary, United States) was performed. The dam effect (F0 founder females and subsequent F1-F2-F3 dams), the sex of the pups and the number of pups per litter were controlled in the model. When the data showed non-significant sex effects, male and female data were combined. Throughout the four generations, male lineages were derived from an equivalent number of the F0 founder females to increase diversity and minimize the effect of the F0 dams. The main effects of POPs (in POPs and POPs + FAS groups) and FAS (in FAS and POPs + FAS groups) as well as their interaction (POPs*FAS) were determined. If there was no significant interaction, the main effect of the POPs and/or FAS were interpreted individually. The assessment of the skeletal ossification and structural measurements were conducted using a blind analysis.

To determine if maternal FAS can protect against the expected negative impacts of prenatal and ancestral paternal POPs exposure on the bone ossification of four generations of rat fetuses, a quantitative assessment of the degree of skeletal ossification was performed. The number of fetuses obtain throughout the four generations were always between 8 and 18 per litter, with a mean of 13,2 fetuses per litter (Table 2). The results are presented in Table 3; Figure 3 shows examples of reduced ossification in the sternum, front paws, and hind paws in the POPs-exposed and/or FAS-treated groups. No difference was noted regarding the ossification profile between males and females in any generation. There was no difference in the degree of ossification of the vertebral column among the four treatment groups within any generation. The ossification of the sternum (p = 0.05) and the front (p = 0.05) and (p = 0.03) hind paws, specifically at the digits, is reduced in POPs-exposed F1 fetuses, with no beneficial effect of FAS. Indicating a specific effect of folic acid supplementation alone, the front paws only of FAS-treated F1 fetuses present a similar decreased ossification (p = 0.003) as the POPs-treated groups. Reduced ossification induced by POPs exposure in the sternum (p = 0.04) persisted in the F2 generation. No significant differences in ossification were noted in the F3-F4 fetuses.

Head structure and long bone lengths were measured to evaluate the consequence of exposure to POPs and/or FAS treatment on bone growth (Table 4; Figure 4). No difference was noted in bone structure measures between males and females in any generation.

In F1 fetuses, there was an interaction between POPs and FAS (POPs*FAS p = 0.004), where the POPs exposure reduced the length of the mandible, but the co-intervention with FAS resulted in lengths similar to the CTRL measurements. As for long bone length, a similar pattern of reduction was observed in the humerus (p = 0.001), radius (p = 0.003), femur (p = 0.03), and tibia (p = 0.005) following POPs exposure.

In F2 fetuses, the mandible (p = 0.03) length was smaller in the POPs-treated lineages. Similar to the F1 generation, POPs exposure also reduced the length of the F2 fetal humerus (p = 0.05), radius (p = 0.005), and femur (p = 0.03), without effect on the tibia. Furthermore, the co-intervention of FAS did not counteract the consequences of POPs exposure. Similar to the bone ossification results, there was no difference in the lengths of head structures or in the lengths of the long bones in the F3 generation.

In the F4 generation, the POPs ancestral exposure resulted in longer humerus (p = 0.01) and radius (p = 0.003), contrasting with the reductions seen in F1 and F2. Moreover, ancestral FAS treatments had the same increasing effect as the POPs for the humerus (p = 0.02) and radius (p = 0.02).

Our data show that direct exposure to POPs resulted in delayed ossification of the sternum (F1-F2) and front and hind paws (F1). At GD19.5, the sternum is expected to be only partially ossified, but ossification centers are usually present in most sternebrae (Menegola et al., 2001). Exposure to our POPs mixture resulted in a poorly ossified sternum over two generations (F1-F2). Another study reported a similar phenotype in the first generation of late term rat fetuses perinatally exposed to an environmentally relevant mixture of brominated flame retardants (BFRs), including polybrominated diphenyl ethers (PBDEs), that are also POPs (Berger et al., 2014).

Direct exposure to POPs also affected the length of the limb long bones over multiple generations (F1-F2; F4) as it resulted in shorter humerus, radius and femur bones in F1 and F2 fetuses and in F1 tibia. Surprisingly, F4 fetuses with ancestral POPs exposure had longer front limb long bones. Previous studies demonstrated bone alterations related to POPs exposure up until only the first generation. Indeed, perinatal exposure to a similar environmentally-relevant POPs mixture including methylmercury, resulted in delayed F1 femur growth and ossification, and reduced mechanical strength in a rat model (Elabbas et al., 2011a). The same team further explored the specific role of PCBs in their mixture, namely Aroclor 1254, a technical mixture of several PCBs. They observed the same consequences as with the full mixture, suggesting that bone-related consequences could be attributed to the PCBs in the POPs mixture (Elabbas et al., 2011b). Indeed, certain PCBs are dioxin-like and can mimic the action of dioxins (Giesy and Kannan, 1998), which have been previously shown to be detrimental to bone health (Jämsä et al., 2001). Although the mechanisms of bone toxicity caused by dioxin-like chemicals are not totally understood, studies suggest that they impair the normal function of the aryl hydrocarbon receptor (AhR), implicated in bone remodelling and general bone homeostasis (Herlin et al., 2010; Yu et al., 2018). The anti-estrogenic PCB 126 has also been linked to impaired mineralisation and reduced length of long bones (Lind et al., 1999; Lind et al., 2000).

Furthermore, dissecting the impacts of each component suggest that the mixture used here can perturb several hormone-signalling pathways because of its: 1) estrogenic activity, due to aldrin, dieldrin, DDT/DDE, B-HCH, and toxaphene; 2) strong antiandrogenic activity, due to PCBs and DDT/DDE compounds blocking the androgen receptor; 3) dioxin-like activity, due to certain PCBs (Donaldson et al., 2010). Androgen hormones have been shown to modulate longitudinal growth in long bones (Wiren, 2005). Androgen receptors have been found on osteoblasts, osteocytes, and chondrocytes which further implies a direct role of androgen on bone and cartilage (Chen and Guo, 2013). The strong inhibition of the androgen receptor by the Arctic POPs mixture could explain why the growth of the long bones was impaired in this model. Future investigation should assess the interaction among these toxicants, the endocrine system, and bone development.

Although these endocrine disrupting mechanisms could partly explain the bone variations observed in F1, it is not clear how the paternal exposure to POPs and FAS influences the subsequent, non-directly exposed generations. An increasing body of literature demonstrates that the paternal environment can affect the development of his offspring over several generations through epigenetic perturbations of the male germline (Lambrot et al., 2013; Braun et al., 2017; Ly et al., 2017). Our team previously demonstrated that prenatal and ancestral POPs exposure led to intergenerational alterations of specific sperm miRNA profiles, known to target genes implicated in embryonic organ development, among others (Herst et al., 2019). We also observed reduced sperm parameters throughout three unexposed generations following prenatal and ancestral POPs treatments (Lessard et al., 2019). The increased long bone lengths in POPs-exposed F4 fetuses reported in the present study is consistent with our previous surprising observation that the most differentially expressed genes in two-cell embryos and the poorest pregnancy outcomes also occurred in the F4 litters (Lessard et al., 2019). Indeed, environmentally sensitive epigenomic regions have been identified in human sperm that could be transmitted to subsequent generations (Chan et al., 2019). Using an inheritance simulation statistical package to assess the rat sperm DNA methylome demonstrated that early-life paternal exposure to the same POPs mixture used in this study leads to inherited methylation changes that are inter- and transgenerationally transmitted (Belleau et al., 2018).

In this study, following our various publications, we hypothesized that the phenotypes observed at the bone level could potentially be the result of alterations in the sperm epigenome, including the methylome and miRNAs. In this same vein, another potential mechanism that was not explored in this study is the role of the stress hormone corticosterone. This hormone known as a modulator of transgenerational epigenetic inheritance could be an adjacent or complementary mechanism to POPS-induced changes observed across generations (Short et al., 2016; Yeshurun et al., 2017).

Maternal supplementation of FA is recommended to ensure adequate fetal development (Antony, 2007). The placenta concentrates folate into the fetal circulation and the maternal-to-fetal folate transfer is mediated by placental folate receptors (Henderson et al., 1995; Antony, 2007). Sufficient folate in utero is essential for the neuroskeletal development of the fetus, where the brain, skull and spine are growing interdependently (Morse, 2012). Osteogenesis takes place during an intensive cell proliferation and DNA synthesis period, which involves one-carbon metabolism where folate acts to carry and donate one-carbon units to support biosynthetic reactions including for de novo nucleotide and methionine synthesis (Crider et al., 2012). Therefore, as skeletal development occurs mainly during gestation and early life, we hypothesized that prenatal folic acid supplementation (FAS) would reduce and/or counteract the harmful bone effects induced by POPs assuming that they were related to cell proliferation and differentiation.

Our team previously reported some protective effect of FAS on sperm parameters and sperm miRNA profiles across multiple generations (Herst et al., 2019; Lessard et al., 2019). In this study, however, we observed few noticeable protective effects of direct or ancestral FAS exposure on skeletal development. Prenatal dietary FAS did not counteract delayed ossification and reduced length of long bones in fetuses exposed to POPs in any generation. Moreover, FAS alone led to a similar phenotype to POPs exposure in the F1 generation, with decreased ossification of the front paws. FAS also led to negative effects on pregnancy outcomes, namely fertility rates and preimplantation losses in F3 fathers (Lessard et al., 2019). Our team also showed that FAS could not prevent the deleterious consequences of POPs exposure on lipid homeostasis in male descendants (Navarro et al., 2019). Taken together, therefore, FAS inconsistently prevents the negative effects associated with POPs exposure that occurred over several generations. The timing of exposure may have been insufficient since the FAS lasted only as long as direct exposure to POPs. It is also possible that FA sufficiency, as represented by the control diet, provided a maximal protective effect against POPs and therefore FAS provided no additional benefits. It is possible that a more severe phenotype would have been observed in rats exposed to POPs and a FA-insufficient diet.