Maternal exposure to perfluorobutane sulfonate (PFBS) during pregnancy: evidence of adverse maternal and fetoplacental effects in New Zealand White (NZW) rabbits

Abstract

Perfluorobutanesulfonic acid (PFBS) is a replacement for perfluorooctanesulfonic acid (PFOS) that is increasingly detected in drinking water and human serum. Higher PFBS exposure is associated with risk for preeclampsia, the leading cause of maternal and infant morbidity and mortality in the United States. This study investigated relevant maternal and fetal health outcomes after gestational exposure to PFBS in a New Zealand White rabbit model. Nulliparous female rabbits were supplied drinking water containing 0 mg/l (control), 10 mg/l (low), or 100 mg/l (high) PFBS. Maternal blood pressure, body weights, liver and kidney weights histopathology, clinical chemistry panels, and thyroid hormone levels were evaluated. Fetal endpoints evaluated at necropsy included viability, body weights, crown-rump length, and liver and kidney histopathology, whereas placenta endpoints included weight, morphology, histopathology, and full transcriptome RNA sequencing. PFBS-high dose dams exhibited significant changes in blood pressure markers, seen through increased pulse pressure and renal resistive index measures, as well as kidney histopathological changes. Fetuses from these dams showed decreased crown-rump length. Statistical analysis of placental weight via a mixed model statistical approach identified a significant interaction term between PFBS high dose and fetal sex, suggesting a sex-specific effect on placental weight. RNA sequencing identified the dysregulation of angiotensin (AGT) in PFBS high-dose placentas. These results suggest that PFBS exposure during gestation leads to adverse maternal outcomes, such as renal injury and hypertension, and fetal outcomes, including decreased growth parameters and adverse placenta function. These outcomes raise concerns about pregnant women’s exposure to PFBS and pregnancy outcomes.

Per- and polyfluoroalkyl substances (PFAS) are a class of manmade chemicals widely used in manufacturing and commercial products due to their water- and stain-resistant properties (ATSDR, 2019; ITRC, 2020). Following 5 decades of unregulated production and disposal processes along with long chemical half-lives, PFAS are widely detected in the environment and have been measured in air, water, soil, and household dust (Brusseau et al., 2020; Domingo and Nadal, 2019; Hall et al., 2020; Stoiber et al., 2020). Amid hundreds of PFAS in production, most research has focused on perfluorooctanoic acid (PFOA) and perfluorooctanesulfonic acid (PFOS) due to lengthy environmental persistence and their association with detrimental health effects. Epidemiology and studies in rats and mice have demonstrated links between PFOS and PFOA exposure and adverse maternal and fetal health outcomes, including metabolic diseases, fetal growth restriction, and increased preeclampsia risk (Bach et al., 2015; Bommarito et al., 2021; Stein et al., 2009; Szilagyi et al., 2020; Wikstrom et al., 2019). In the early 2000s, PFOS and PFOA were phased out of U.S. production (ATSDR, 2021). Since the phase out, replacement PFAS have flooded the market—and consequently, the environment. There is a limited understanding of replacement PFAS toxicity, but recent in vivo studies have demonstrated that exposure to replacement PFAS can result in health effects resembling that of legacy PFAS (Blake et al., 2020; Yue et al., 2020), suggesting these replacements may be of concern for human health.

Perfluorobutanesulfonic acid (PFBS) is a replacement PFAS made up of a 4-carbon chain and a sulfonic acid functional group, resembling legacy PFOS (Figure 1). PFBS has a shorter half-life than PFOS (∼28 days vs 3–5 years, respectively) (Olsen et al., 2007b, 2009; Xu et al., 2020). However, PFBS is increasingly detected in drinking water worldwide (Li et al., 2019, 2021; Ünlü Endirlik et al., 2019; Zafeiraki et al., 2015). In the United States, PFBS is the fourth most frequently detected PFAS in water systems, and the detection of PFBS in the environment has increased over time (Podder et al., 2021; Sutherland-Ashley et al., 2021). Furthermore, PFBS is detectable in human serum, including children (Bao et al., 2014; Gyllenhammar et al., 2019; Liu et al., 2020). High industrial demand coupled with a lack of regulation predicts human PFBS exposure levels will continue to rise (Podder et al., 2021; Sutherland-Ashley et al., 2021). Due to widespread exposure and emerging evidence of adverse human health effects, the U.S. Environmental Protection Agency (EPA) recently released the first drinking water health advisory levels for PFBS (2000 ppt), the level at which adverse health effects are anticipated to occur (EPA, 2022).

Figure 1.

Chemical structure of legacy perfluorooctanesulfonic acid (PFOS) and perfluorobutane sulfonate (PFBS).

Although there are limited human toxicity studies of PFBS, several key epidemiology studies have associated exposure to PFBS with adverse pregnancy outcomes, such as risk of endometriosis-related infertility (Liu et al., 2020) and hormone disruption (Huang et al., 2019b; Liu et al., 2020; Wang et al., 2017). Importantly, 3 studies have recently associated PFBS exposure with increased risk of hypertensive disorders of pregnancy (HDP) (Huang et al., 2019b; Huo et al., 2020; Liu et al., 2022), and 2 found significant associations between PFBS and increased risk for women to develop HDP such as preeclampsia (Huang et al., 2019b; Liu et al., 2022).

Thus far, in vivo reproductive studies have linked PFBS exposure with adverse effects on liver and kidneys (300 and 1000 mg/kg/day [Lieder et al., 2009b]; 1000 mg/kg/day [NTP, 2019a]), changes in thyroid and sex hormones (200 and 500 mg/kg/day [Feng et al., 2017]; 1000 mg/kg/day [NTP, 2019a]), decreased fetal weight (200 mg/kg/day [Feng et al., 2017]; 1000 mg/kg/day [York, 2002]; and 2000 mg/kg/day [York, 2003]), and delayed development of rat pups (1000 mg/kg/day [York, 2002]). However, the effects of PFBS on the development of HDP and preeclampsia and related reproductive effects have not been closely investigated in vivo.

A primary reason for adverse pregnancy outcomes including preeclampsia and fetal growth restriction is dysregulated placentation (Brosens et al., 2011; Roberts and Escudero, 2012). A fully functional placenta supplies necessary nutrients, removes waste, and provides a communicative interface and toxicant barrier between mother and child. For example, the placenta plays a major role in transporting and metabolizing maternal thyroid hormones for delivery to the fetus, which is critical for fetal growth and development (Eerdekens et al., 2020). Disruption of these processes can manifest as adverse health effects in both the mother and child during pregnancy. Subsequently, in vitro studies in our lab have demonstrated that PFBS exposure can alter processes required for successful placentation, including trophoblast cell invasion and migration as well as angiogenic processes with underlying changes in cellular transcriptomics related to disease (Marinello et al., 2020; Pham et al., 2020).

Taken together, the evidence suggests that maternal PFBS exposure may lead to adverse maternal and fetal health outcomes during pregnancy, mediated by suboptimal placental function. The EPA’s new drinking water health advisory raises concern that PFBS can have more significant health effects than previously thought, although there are limited reproductive studies on this replacement PFAS. Thus, this study evaluates the effects of PFBS exposure through drinking water during pregnancy on maternal and fetal health effects, with a focus on altered placental function as a potential mediator of adverse pregnancy and birth outcomes.

Materials and methods

Methodology in this study closely follows our previous rabbit study (Crute et al., 2022). Variations and additional details for clarity are described below.

 

 

Animals

Nulliparous 6-month-old female New Zealand White (NZW) rabbits were acquired from Covance Research Products/Envigo (Envigo, Denver, Pennsylvania) and housed within an AAALAC-accredited facility within the Division of Laboratory Animal Resources (DLAR) at the Duke University Medical Center (DUMC). Rabbits were singly housed in 2 adjoining, non-ventilated racks (Euro Cage, Allentown, LLC, Allentown, New Jersey). All rabbits received reverse osmosis filtered tap water from a glass water bottle (±components described below) and were maintained on a Laboratory Rabbit Diet (LabDiet 5321; LabDiet, St. Louis, Missouri). The room was kept on a 12:12 h light-dark cycle at 70°F (±1°F), and relative humidity was maintained at approximately 40% (±1%). All experimental conditions and procedures were approved by the DUMC Institutional Animal Care and Use Committee (IACUC; Protocol Registry Number A214-20-11).

PFBS dosing through drinking water

In this study, rabbits were exposed to 1 of 3 preparations of drinking water containing 0, 10, or 100 mg/l PFBS. A previous pilot rabbit study (Feng Laboratory, unpublished) demonstrated that PFBS exposure at these levels led to changes in placental size and thyroid hormone disruption. Thus, these PFBS concentrations were selected for use in the present study.

Although these PFBS doses exceed the concentrations of PFBS typically found in U.S. drinking water (reported at levels up to 2.2 µg/l), they are lower than previous in vivo studies ((EWG), 2021; Feng et al., 2017; Lieder et al., 2009a,b; NTP, 2019b). In addition, due to species differences in toxicokinetics and toxicodynamics, higher concentrations of PFBS are typically necessary in animal studies to observe effects. PFAS in particular has much shorter elimination half-lives in animals than humans (Huang et al., 2019a; Olsen et al., 2007a, 2009).

PFBS was procured as a potassium salt with a 95%–98% purity from SynQuest Laboratories (CAS Number: 29420-49-3; Alachua, Florida). Neat PFBS was added to 50 l of ultra-pure water into polypropylene carboys to a final concentration of either 10 or 100 mg/l PFBS, and shaken overnight until dissolved. Polypropylene 50-L carboys with stopcocks (VWR, Wayne, Pennsylvania) were used to store prepared water and were kept at room temperature in the animal facility throughout the study period. Before use, carboys were rinsed with methanol (Optima grade, Fisher Chemical, Chicago, Illinois) and ammonium hydroxide (Optima grade, Fisher Chemical), followed by several successive rinses of methanol alone, hot tap water and Liquinox detergent (Alconox, Inc, New York, New York), tap water, deionized water, and ultra-pure water. Carboys were then sterilized by autoclave. The concentration of the PFBS-dosed water was confirmed using LC-MS/MS analysis as previously described (Crute et al., 2022) and as briefly summarized in Supplementary Material.

Study design

This study was performed over 2 blocks with a total of 10 control, 7 PFBS-low dose, and 7 PFBS-high dose rabbits. All dams were randomly assigned to treatment groups. Experimental groups were staggered by 3-4 weeks each due to housing and experimental constraints. Investigators were blinded to live animal procedures, but they were not blinded to experimental group during necropsy due to the staggered experimental blocks design. Each group completed a total study duration of 32 days, as depicted in Figure 2. Briefly, baseline body weight (BW), blood pressure measurements, and blood were collected before day 1 of exposure to PFBS-dosed water. For these procedures, rabbits were sedated with acepromazine (1 mg/kg intramuscularly; IM), and EMLA cream (mixture of lidocaine 2.5% and prilocaine 2.5%) was applied topically to the ears to serve as a local anesthetic. After 15 min to allow medications to take effect, rabbits were placed in a rabbit restrainer (TBJ, Inc, Chambersburg, Pennsylvania or Otto Environmental, Greenfield, Wisconsin), and approximately 12 ml of blood was collected from the marginal ear vein. Direct blood pressure measurements were collected via a 22-gauge catheter in the central auricular artery, which was connected to saline-filled tubing and a calibrated pressure transducer. Measurements were recorded starting approximately 2 min after the transducer was zeroed and recorded every 30 s for a total of 3 min.

Figure 2.

New Zealand White rabbits were exposed to control or PFBS-contaminated drinking water for 32 days over the course of the study. Blood and baseline measurements, including body weight and blood pressure, were taken at timepoint 0. Only body weight measurements and blood were taken at timepoints 1 and 2, whereas blood was collected and measurements of body weight and blood pressure were assessed again at timepoint 3. Rabbits were bred 7 days after exposure began, and an ultrasound to confirm pregnancy occurred at approximately gestational day (GD) 15. Rabbits were necropsied before parturition on GD 25. Abbreviation: PFBS, perfluorobutanesulfonic acid.

When not performed alongside blood pressure measurements, blood was collected from the central auricular artery following sedation and application of the topical anesthetic cream, as described above, and placed into serum separator tubes and EDTA tubes (Becton-Dickinson, Franklin Lakes, New Jersey). Blood components were separated via centrifugation, and serum and plasma were stored at −80°C. Direct blood pressure was measured at baseline and study day 32 (gestational day [GD] 25), and blood was also collected at baseline, GD 6, and GD 15 as described above.

Rabbits were provided control (ultra-pure drinking water) or the PFBS-dosed drinking water via bottles on their cages according to their exposure group on study day zero. Water intake by mass was measured twice daily. Seven days after exposures were initiated, rabbits were bred by live cover. Breeding was performed for all rabbits in the morning and once again in the afternoon on study day 7 (GD 0) to ensure successful breeding within the ovulation window (Landon et al., 2021). On approximately study day 25 (GD 15), all rabbits were sedated as described above and anesthetized with isoflurane by mask for pregnancy confirmation by ultrasound (SonoSite M-Turbo, FUJIFILM SonoSite Inc, Bothell, Washington).

Necropsy was performed on study day 32 (GD 25), shortly before live birth to ensure fetuses, placentas, and associated tissues could be collected, as previously described (Crute et al., 2022). Rabbits were anesthetized with a combination of ketamine (30–40 mg/kg) and xylazine (3–5 mg/kg) administered IM or subcutaneously. Once an anesthesia plane was confirmed, terminal blood collection was performed via the intracardiac (IC) route. Dams were then euthanized by an overdose of barbiturates (1 ml/10 lbs Euthasol given IC; Virbac AH Inc, Fort Worth, Texas). Following removal from the uterus, all fetuses were monitored for breathing and responsiveness to toe pinch. If necessary, supplemental isoflurane was administered via face mask prior to blood collection via cardiac puncture, and Euthasol (IC) was administered to confirm euthanasia.

Necropsy and sample collection

Following terminal blood collection and euthanasia, a gross necropsy was performed, as previously described (Crute et al., 2022). Briefly, the intact gravid uterus was weighed and examined for macroscopic lesions. Fetuses and corresponding placentas were removed from the gravid uterus and labeled according to their position within the left and right uterine horns. Placentas were assessed for adverse morphology (ie, disconnected lobes, uneven lobes, lesions). Next, the degenerative maternal tissue was removed from the fetal side of the placenta, and fetal placental weight (PW) was recorded. Fetal BW and brain weight (BrW) were recorded, and crown-rump length (CRL) was determined by measuring the length from the top of head (crown) to the tail (rump). Fetal whole blood was collected via cardiac puncture and stored in EDTA tubes. All fetuses were sexed via internal gonad assessment. Finally, dam liver and left kidney were examined and weighed. All collected organs were sectioned and stored in RNALater for gene expression analysis; on dry ice for chemical, hormone, and protein analysis; or in 10% formalin for histologic evaluation.

Histopathology

Tissues were collected for histologic evaluation, including sections of dam livers (left lateral lobe) and kidneys (right kidney), fetal liver, fetal kidney, and placenta, and any observed macroscopic lesions, and were immersion fixed in 10% neutral buffered formalin. Tissues were then processed routinely for paraffin embedment, cut at 5 µM, and sliced into microscope slides, which were further stained with hematoxylin and eosin (IDEXX, Columbia, Missouri). Slides were evaluated by a board-certified veterinary pathologist with toxicologic pathology experience to identify histopathologic changes. Pathologic evaluation assessed many structural features but primarily focused on degenerative changes in labyrinth region. A first pass examination of tissues was conducted with knowledge of allocation of control and treatment groups, and then a secondary read was performed in a blinded fashion. Histological placental slides were whole-slide imaged at 40× (Aperio AT2, Leica) and labyrinth zone measurements made using measuring tool in Imagescope software (Leica Biosystems Imaging, Inc). Labyrinth zone thickness was assessed by pathologist on sections through the center of the disc.

Internal dosimetry and clinical serum measures

Maternal serum samples from the control animals were analyzed for PFBS as previously described (Crute et al., 2022). Maternal serum samples from the PFBS-exposed dams were diluted 1:10 in a 30:70 (volume:volume [v:v]) mixture of methanol and 2 mM ammonium acetate, and then analyzed for PFBS concentration as described in Crute et al. (2022). In brief, samples were analyzed by solid-phase extraction and LC-MS/MS using modified methods from Liu et al. (2015). Further details on the serum PFBS LC-MS/MS analysis are included in the Supplementary Material.

Additional maternal serum samples were submitted to Duke Veterinary Diagnostic Laboratory (VDL) for analysis of clinical chemistry measures. Briefly, serum was aliquoted into an Element DC5X Veterinary Chemistry Analyzer (Heska Cooperation, Loveland, Colorado) and analyzed via colorimetric endpoint and ion selective electrodes tests, using the associated rabbit DRI-CHEM slides (Heska Cooperation).

Thyroid hormone measurements

Thyroid hormones rT3, T3, and T4 were measured in maternal serum and kit whole blood following established methods (Crute et al., 2022; Wang and Stapleton, 2010). Maternal samples were pooled into 7 samples and kit whole blood was pooled into 3 female and 3 male samples per dose group. In brief, serum or whole blood (1.0 ml) was spiked with isotopically labeled internal standards, extracted with acetone: water, purified using solid-phase extraction, and analyzed for thyroid hormones using LC-MS/MS. Further methodological details are included in Crute (2022) and also in Supplementary Material.

Gene expression analysis via quantitative PCR

To determine gene expression levels in rabbit placental samples, previously designed and optimized assays were used (Crute et al., 2022). Briefly, DNA/RNA was simultaneously extracted from the same tissue sample using placenta samples from 2 female and 2 male fetuses per dam with the Qiagen AllPrep kit (Germantown, Maryland) and included a DNase treatment step. The 2 RNA samples of the same sex were combined into one tube to provide for one pooled male and one pooled female RNA placenta sample per dam. To evaluate gene expression, a 2-step qPCR was performed. First, cDNA was created from 500 ng RNA in a reaction containing 4 µl iScript Reverse Transcription supermix (Bio-Rad, Hercules, California), and brought to a total volume of 20 µl with RNase-free water. Reactions were run on a CFX Connect Real-Time system (Bio-Rad). Each reaction contained: 1 µl cDNA, 1 µl each of the 10 mM forward and 10 µM reverse primer, 12.5 µl iQ SYBR Green Supermix (Bio-Rad), and 9.5 µl water. The reactions were run in a real-time thermocycler at the following conditions: 95°C for 3 min followed by 39 cycles of 95°C for 30 s and 60°C for 40 s. Each reaction was run in duplicate. Expression levels for each individual sample were normalized to GAPDH, and the relative expression was analyzed using the 2^−ΔΔCt method (Livak and Schmittgen, 2001).

Full transcriptome sequencing of placental mRNA

Whole transcriptome sequencing and analysis followed previously published methods (Crute et al., 2022) and are described in detail in the Supplementary Material. Briefly, placental RNA purified from the placentas of 1 male and 1 female fetus per dam were combined into 3 PFBS-high, 3 PFBS-low, and 3 control pooled samples. Library preparation was performed using the Roche KAPA Stranded mRNA-Seq Kit following the protocol of the manufacturer (Indianapolis, Indiana). Libraries were indexed using a dual indexing approach allowing for multiple libraries to be pooled and sequenced on the same Ilumina NovaSeq 6000 sequencing platform (San Diego, California). Once generated, sequence data were demultiplexed and Fastq files generated using Bcl2Fastq conversion software provided by Illumina (San Diego, California). Raw data files are accessible through the Duke Data Repository (https://doi.org/10.7924/r4w66k88z; last accessed December 7, 2022).

Raw data Fastp files were run through fastp version 0.20.1 to control and improve the sequence read quality before starting downstream analysis (Chen et al., 2018). Alignment of the trimmed reads was performed using STAR version 2.7.9a (Dobin et al., 2013). The reference genome assembly OryCun2.0 for Oryctolagus cuniculus (rabbit) was used with the corresponding RefSeq annotation (collected from NCBI datasets) for generating the STAR index. Read counts produced by the STAR aligner were gathered to a count matrix that was then used as input for differential gene expression analysis using DESeq2 version 1.28.1 (Love et al., 2014). Adaptive shrinkage was used for adjusting the expression differences observed for genes with low expression. Significant differentially expressed genes were identified by filtering the adjusted p-value <.1 (Stephens, 2017).

Statistical analysis

All data endpoints and sample sizes are listed in the table or figure legends. A p < .05 was considered statistically significant unless otherwise noted. Data were analyzed using GraphPad Prism 9.1 (La Jolla, California). All data were examined for parametric distribution via the Anderson-Darling test. All maternal measures, including necropsy endpoints, dosimetry, clinical chemistry, and thyroid hormones were analyzed using a 1-way ANOVA with Tukey’s multiple comparison test or Kruskal-Wallis test with Dunn’s multiple comparison test for parametric or nonparametric data, respectively. Incidence of histopathology was evaluated using Prism’s contingency table function with Fisher’s exact test. When analyzing dam PFBS internal dosimetry, values below the MDL were imputed as half the MDL. Incidence of histopathology and gross morphology was evaluated using Prism’s contingency table function with Fisher’s exact test.

All fetal measures (eg, BW, CRL, BrW, fetal PW), thyroid hormones, and placental qPCR and thickness of labyrinth zone data were analyzed using a 1-way ANOVA with Tukey’s multiple comparison test or Kruskal-Wallis test with Dunn’s multiple comparison test for parametric or nonparametric data, respectively. To investigate sex differences, data were further stratified by fetal sex and analyzed via 2-way ANOVA with Sidak’s multiple comparisons test. Incidence of histopathology and gross morphology was evaluated using Prism’s contingency table function with Fisher’s exact test. From the same dam, the mean of all fetuses, or all male fetuses, or all female fetuses, were considered as N = 1 in all analyses.

To statistically account for effects of biologically relevant predictors, including both fixed effects (eg, litter size, kit sex) and random effects (eg, experimental block, dam), data corresponding to kit and placental growth were evaluated using mixed effects models, as previously published (Blake et al., 2020). Data were analyzed in R (version 4.0.3; R Development Core Team). Models were fit in a stepwise procedure using Satterthwaite’s method for fixed and random effects. Models included an interaction term for treatment group and kit sex using the lmerTest package (Kuznetsova et al., 2017). All final models included treatment group and litter size as fixed effects and dam as a random effect, but models were allowed to vary in the inclusion of the treatment group, kit sex interaction term, and kit sex as an independent fixed effect. No significant random effects of experimental block were identified across all kit or placenta growth endpoints, so this random effect term was not included in any final models. Using the Wald method, point estimates and 95% confidence intervals were determined from the final model.

Results

 

Water consumption and internal dosimetry

The amount of water consumed did not significantly differ between dams in control (230.3 ± 39.2 g), PFBS-low (235.5 ± 49.6 g), and PFBS-high (245.7 ± 51.0 g) groups (Table 1). Given the PFBS concentration and amount of water consumed, the daily PFBS dose was estimated at 0.65 (±0.11) mg/kg BW/day for the PFBS-low dose group and 6.8 (±1.4) mg/kg BW/day for the PFBS-high dose group. To evaluate circulating exposure, PFBS was measured in dam serum from the last study day for each dam. PFBS was detected in 100% of dams, at a mean and standard deviation in control dams at 0.876 ± 0.297 ng/g serum, PFBS-low dams at 306.1 ± 354.7 ng/g, and PFBS-high dams at 2036.0 ± 1478.1 ng/g (Table 1). Using a 1-way ANOVA to determine significance between means of serum exposure, the PFBS-low dam mean did not significantly vary from controls, whereas the high-dose mean significantly varied from controls (p = .0001) and PFBS-low dams (p = .002). Water consumption and internal dosimetry by dam are included in Supplementary Table 1.

Table 1.

Observed daily water consumption (in grams) and estimated PFBS intake (in mg/kg BW/day) are reported for each dose group (means and standard deviation)

		Control	PFBS-low	PFBS-high
		(0 mg/l)	(10 mg/l)	(100 mg/l)
		(n = 10)	(n = 7)	(n = 7)
Water	Daily water consumption (g)	230.3 ± 39.2	235.5 ± 49.6	245.7 ± 51.0
	Estimated PFBS dose (mg/kg BW/day)	0.0 ± 0.0	0.65 ± 0.11	6.8 ± 1.4
Serum	Detection frequency in serum	100%	100%	100%
	Serum concentration
	Min–max (ng/g)	0.512–1.570	57.0–1075.5	356.9–4538.0
	Median (ng/g)	0.819	153.2	2039
	Mean (ng/g)	0.878	306.1	2036

Dam serum PFBS concentration at necropsy is also reported (in ng/g) as a measure of internal dose. The PFBS method detection limit (MDL) for serum was 0.242 ng/g.

Maternal blood pressure

Maternal blood pressure was evaluated by direct auricular arterial measurements, which allowed for collection of systolic, diastolic, and average arterial pressure measures. All measures were adjusted to percent change, calculated as [(study day 32 − baseline)/baseline]. Neither systolic, diastolic, or average arterial pulse pressure varied significantly between groups (Supplementary Figure 1). However, percent change of pulse pressure (calculated as [systolic − diastolic blood pressure]) was significantly increased in PFBS-high dams (p = .028) (Figure 3). Additionally, a renal resistive index (RI) measure (calculated as [diastolic/systolic blood pressure]) was significantly decreased in PFBS-high dams (p = .018) (Figure 3).

Figure 3.

Maternal blood pressure measures, illustrated as pulse pressure (A) and renal resistive index measure (B). All data illustrated as percent change from baseline to the last study day, and analyzed via 1-way ANOVA with Dunnett’s test for multiple comparisons.  p < .05 is considered significant.

Maternal body weight

Single-observation maternal outcomes were measured at necropsy and are presented in Supplementary Table 2. Absolute dam BW did not significantly differ between controls (3.75 ± 0.24 kg), PFBS-low (3.61 ± 0.24 kg), and PFBS-high (3.64 ± 0.27 kg) group dams. Additionally, absolute maternal BW correcting for the weight of the corresponding gravid uterus was not significantly different between groups (p > .05). Maternal liver and kidney weights showed no significant difference in PFBS-exposed dams, in either absolute or BW-adjusted values.

Maternal histopathology

Liver histopathology results showed no changes in dams in control or PFBS-low group, but 3 dams in the PFBS-high group exhibited minimal cytoplasmic vacuolation (Figure 4). Using a Fisher’s exact test, incidence of histopathological change in PFBS-high dams was not statistically significant compared with controls (p = .051). No kidney lesions were observed in the control dams, but PFBS-low and PFBS-high exposures produced minimal-to-mild renal tubular epithelial changes (nephropathy) in 5 dams (p = .003) and 4 dams (p = .015), out of 7 total dams per group, respectively (Figure 4).

Figure 4.

Maternal kidney and liver histopathology. A, Representative images of maternal kidney histopathology results under light microscopy. Kidney from a PFBS-high dose exposed dam shows a mild case of nephropathy relative to the control. Arrows designate basophilic tubules. B, Livers of exposed dams showed minimal changes in cytoplasmic vacuolation (arrow) at low incidence. Abbreviation: ND, not detected.

Maternal chemistry

No clinical chemistry measures were significantly different between control, PFBS-low, and PFBS-high rabbits from maternal serum collected on GD 25 (Supplementary Table 3). However, several relevant trends were apparent in PFBS-high rabbits, including lower blood urea nitrogen to creatinine ratios (p = .12) and lower glucose (p = .11). Differences in thyroid hormones rT3, T3, T4, and the ratio of T3:T4 measured in maternal serum did not reach statistical significance (Supplementary Figure 2). However, trends of lower T4 levels (p = .25) and higher T3:T4 ratios (p = .22) were observed in PFBS-low exposed dams, and lower T3 levels (p = .31) and T4 levels (p = .25) were observed in PFBS-high exposed dams.

Fetoplacental growth

To evaluate fetoplacental outcomes of PFBS exposure, the number of viable fetuses, fetal sex, number of resorptions, fetal BW, CRL, BrW, and PW were evaluated at necropsy. PFBS treatment exhibited a decreased dose-dependent trend from control to PFBS-low to PFBS-high fetuses in BW, CRL, BrW, and fetal PW (Table 2, Supplementary Table 4 and Figure 3). However, only CRL demonstrated a statistically significant decrease in PFBS-high dose fetuses (p = .041). Furthermore, no statistically significant changes were detected when analyzing BW-adjusted CRL, BrW, or PW.

Table 2.

Fetoplacental measures at necropsy (GD 25)

Measurement	Control	PFBS-low	PFBS-high
	(n = 10)	(n = 7)	(n = 7)
Viable fetuses	7.80 (2.61)	7.00 (2.77)	8.57 (1.90)
Male fetuses	4.40 (1.83)	4.29 (1.38)	4.57 (1.72)
Female fetuses	3.40 (1.08)	2.71 (1.80)	4.00 (1.53)
Dams with RES present	40%	43%	43%
BW (g)	19.60 (1.94)	19.53 (2.30)	18.57 (1.63)
CRL (mm)	72.62 (3.05)	71.83 (4.26)	68.67 (2.23)*
BrW (g)	0.557 (0.047)	0.560 (0.057)	0.548 (0.030)
Fetal PW (g)	3.14 (0.55)	3.08 (0.52)	3.01 (0.42)

All measurements are displayed as mean of means (standard deviation).

Abbreviations: RES, resorptions; BW, body weight; CRL, crown-rump length; BrW, brain weight; PW, placenta weight.

* p < .05.

Analysis via a mixed-effects model further evaluated fetal BW, CRL, BrW, and PW using litter size and fetal sex as variables. These results recapitulated the findings that there were no significant impacts of PFBS-low or PFBS-high dose on overall BW, BrW, or PW compared with controls (p > .05), and that CRL is significantly decreased in the PFBS-high group when compared with controls (p = .020) (Figure 5).

Figure 5.

Fetal body weight and crown-rump length at necropsy (GD 25). Best fit mixed model showing beta estimates and 95% confidence intervals for fixed effects of PFBS exposure and litter size on (A) fetal body weight (BW) and (B) crown-rump length (CRL) (*p < .05). PFBS-H = 100 mg/l PFBS group, PFBS-L = 10 mg/l PFBS group; effect estimates are centered around the control group (white vertical line at y = 0). Abbreviation: GD, gestational day.

For whole PW (combined maternal and fetal components), fetal sex was significant as a predictor of whole PW (p = .017) (Figure 6A, left), demonstrating male fetuses generally tended to have heavier placentas (Figure 6A, right). A significant interaction was identified between PFBS-high exposure and sex for full placenta weight (p = .029) (Figure 6A, left). This significant interaction term demonstrates that the PW is higher in PFBS-exposed male fetuses, whereas PFBS-exposed female fetuses are very similar to controls (Figure 6A, right). Furthermore, PFBS-high dose exposure had a significant main effect on the fetal BW: PW ratio (p = .037) (Figure 6B, left). Fetal sex was significant as a predictor of whole PW, as males had lower fetal BW: PW ratios (p = .0004) (Figure 6B, right). Again, a significant interaction was identified between PFBS-high exposure and sex for fetal BW: PW ratios (p = .010) (Figure 6B, left). Finally, all fetal outcomes (BW, CRL, BrW, and PW) were significantly impacted by litter size (p > .05), in which each measure was decreased with larger litters (Figures 5 and 6).

Figure 6.

Maternal PFBS exposure effects on whole placenta and body weight to placenta ratio at necropsy (GD 25). Mixed models showing beta estimates and 95% confidence intervals for fixed effects of PFBS exposure, fetal sex, litter size, and the interaction effect of PFBS treatment and fetal sex on (A) whole PW and (B) fetal BW:PW ratio, with plots of interaction term between PFBS-high and fetal sex. (*p < .05, **p < .01). PFBS-H = 100 mg/l PFBS group, PFBS-L = 10 mg/l PFBS group; effect estimates are centered around the control group (white vertical line at y = 0). Abbreviation: GD, gestational day.

Fetoplacental histopathology

Histopathological lesions were not detected in PFBS-exposed or control kit liver, kidney, or placentas. Placenta morphology did not differ between controls and PFBS-low or PFBS-high dose, and no difference was detected when incidence of abnormal morphology was stratified by sex (p > .05). There were no significant differences in thickness of placental labyrinth zone among groups (Supplementary Figure 4, p = .98).

Fetal thyroid hormones

Although we attempted to measure thyroid hormones rT3, T3, and T4 in kit whole blood, only T4 levels were above the MDL. There were no significant differences between PFBS-low or PFBS-high fetus and controls, nor upon further stratification by sex (Supplementary Figure 5).

Placental gene expression

Investigation of mRNA levels for 6 genes involved with thyroid hormone metabolism in fetal placentas were analyzed both stratified by sex and with sexes combined. PFBS-low treatment disrupted expression of DIO2 (females, increased, p = .021; males, increased, p = .004) and SCARB1 (males, decreased, p = .012) when compared with controls (Figure 7). PFBS-high treatment disrupted expression of SCARB1 (males, decreased, p = .005) and SULT1C2 (females, increased, p = .035) when compared with controls (Figure 7). When data were analyzed by combining sexes, DIO2 increased in PFBS-low (p = .009) and SCARB1 decreased in PFBS-low (p = .043) and PFBS-high (p = .025) remained significant. Additionally, trends in differential expression of MCT8 in both PFBS-low (increased) and PFBS-high (decreased) versus controls were detected (p < .1) (Supplementary Figure 6). RNA sequencing of PFBS-low versus control placentas did not identify any significant changes in gene expression with a Bonferroni adjusted p-value <.05. However, for PFBS-high versus control placentas, the expression of AGT was significantly increased by 6.6-fold (Bonferroni-adjusted p-value = .0006).

Figure 7.

Gene expression differences in the placenta following PFBS exposure. Genes studied were involved in thyroid hormone transport and metabolism in the placenta. Data were analyzed using a 1-way ANOVA with Tukey’s multiple comparison test; p < .05 is considered significant. Abbreviation: PFBS, perfluorobutanesulfonic acid.

Discussion

This study investigated the effects of PFBS exposure during pregnancy on maternal health, fetal development, and placentation using a rabbit model. Rabbits are the requisite non-rodent model in regulatory toxicity testing and have been used extensively in studies of pregnancy (Foote and Carney, 2000). Blood pressure increases during rabbit pregnancy, mimicking human gestational hemodynamics, and blood pressure measurements are readily obtained in these animals (Foote and Carney, 2000). We report that gestational PFBS exposure resulted in higher pulse pressure and decreased renal resistance index in dams, disrupted fetal-placental growth, and altered the expression of placental genes. This work builds on a growing body of evidence in animal studies demonstrating adverse maternal, placental, and fetal outcomes after gestational exposure to PFAS.

PFBS exposure elicited higher pulse pressure in dams exposed to the high dose, which may be reflective of atherosclerosis, arterial wall fibrosis, and/or arterial calcification (de Simone et al., 2005). Increased pulse pressure has been associated with hypertensive complications of pregnancy (Antza et al., 2018) and later life cardiovascular disease (Asmar et al., 2001; Baguet et al., 2000; de Simone et al., 2005; Tozawa et al., 2002). Thus, PFBS-induced increased pulse pressure in the dams could alter arterial structure, increasing risk for cardiovascular diseases post-partum.

PFBS-high dose exposure significantly dysregulated the renal resistive index measure, which is a reproducible, noninvasive measure of arterial compliance or resistance (Andrikou et al., 2018). Renal RI is used as a nonspecific prognostic marker in vascular diseases that affect the kidney, such as hypertension, and can indicate renal disease progression not detected by traditional measures (Andrikou et al., 2018). Additionally, renal RI has been used as a prognostic marker for cardiovascular disease-induced morbidity, and mortality (Andrikou et al., 2018; Radermacher et al., 2002). PFBS-high dose dams demonstrated a decreased renal RI, which is associated with increased vascular resistance (Akaishi et al., 2020). Taken together, PFBS-high dose dams exhibited signs of increased vascular resistance, which could foreshadow hypertension or cardiovascular disease (Akaishi et al., 2020; Andrikou et al., 2018).

Dams in both PFBS exposure groups exhibited significant nephropathy relative to the control dams. These results add to a growing body of evidence that PFBS can disrupt kidney structure and function. For example, higher kidney weights were observed in rats orally exposed to 500 mg/kg/day PFBS ((NTP), 2018) and histopathological changes related to papillary edema and hyperplasia were observed in rats exposed to 300 mg/kg/day and greater doses of PFBS (Lieder et al., 2009a,b).

Gestational PFBS exposure disrupted fetal and placental growth. CRL is a measure of fetal growth corelated with BW, but measures a different fetal growth outcome, and is a common estimate of gestational age (Hoberman and Lewis, 2017). First, PFBS-high dose fetuses had lower CRL compared with controls. This is the first study to report on the effects of in utero PFBS exposure on CRL and provides evidence that PFBS exposure during pregnancy can lead to small for gestational age offspring.

In humans and animals, both PW and fetal-PW ratios are clinically relevant endpoints that have been associated with adverse health outcomes such as fetal growth restriction, preterm birth, perinatal death, and intrapartum distress (Hutcheon et al., 2012; Thornburg et al., 2010). In fact, PW is considered a better predictor of adverse perinatal outcomes than birth weight (Hutcheon et al., 2012). Previous in vivo studies have detected changes in PWs with PFOA and GenX exposure, but direction of the effect varies (Blake et al., 2020; Li et al., 2016). Here we implemented a more sophisticated statistical model to evaluate effects of exposure, fetal sex, and litter on fetal endpoints. A significant interaction between PFBS-high dose and fetal sex was identified for whole placenta weight, indicating that PFBS exposure at the high dose resulted in reduced placenta weights for males only. Increased and decreased PWs are associated with adverse outcomes; increased PW may indicate impaired nutrient supply to the fetus, and a larger placenta has been associated with high childhood blood pressure (Blake and Fenton, 2020; Hemachandra et al., 2006), whereas decreased PW is associated with a lesser nutrient supply to the fetus and subsequent fetal growth restriction (Liu et al., 2021; Salafia et al., 2007).

The feto-PW ratio is used as a proxy measurement of placental efficiency (Hayward et al., 2016; Hutcheon et al., 2012; Risnes et al., 2009; Thornburg et al., 2010). The effect of PFAS exposure on feto-PW ratio has not been extensively investigated in human or in vivo studies, and only one prior study has demonstrated the ability of PFAS to interfere with this marker of placental efficiency (Blake et al., 2020). Here we showed PFBS-high dose exposure led to a significant decrease in BW:PW ratio. There was also a significant interaction between PFBS exposure and sex where female BW:PW ratios were reduced at the high dose. Overall, the present study is the first to provide evidence that PFBS exposure can disrupt PW and feto-PW ratio in an in vivo model, 2 clinically relevant endpoints for adverse pregnancy outcomes.

This study is the first to investigate effects of PFBS on placental thyroid hormone gene expression in the rabbit placenta. This placenta is crucial for transport and metabolism of maternal thyroid hormones that influence fetal growth and development (Chan et al., 2009). Expression of DIO2, a membrane-bound protein that converts pro-hormone thyroxine (T4) to the biologically active T3 molecule (Arrojo et al., 2013; Arrojo and Bianco, 2011; Bianco et al., 2019), was significantly increased in PFBS-low dose placentas. Expression of SCARB1, a plasma membrane receptor recently implicated in thyroid hormone transport (Landers et al., 2018, 2019; Ma et al., 2020), was significantly decreased in both exposure groups. As SCARB1 is involved with T4 uptake and transport by placental trophoblast cells, (Landers et al., 2018), PFBS exposure may decrease transplacental transport of T4. Finally, expression of MCT8 trended toward significance in both exposure groups, which may indicate biological significance. Because T3 and T4 cannot cross the cellular barrier, the MCT8 transporter is crucial for influx and efflux of thyroid hormones T3 and T4, and increased placental MCT8 expression has been associated with intrauterine growth restriction (Bianco et al., 2019; Chan et al., 2006; Visser et al., 2011). Overall, this study provides initial evidence that PFBS exposure can disrupt processes involved with fetal thyroid hormone bioavailability.

Exploratory RNA sequencing produced one differentially expressed gene, AGT, in PFBS-high placentas when compared with controls. AGT encodes the angiotensinogen precursor, which is involved in the renin-angiotensin system (RAS) for maintaining blood pressure (Delforce et al., 2019) and is well-characterized for its role in placentation and preeclampsia. In the placenta, RAS system components regulate placental angiogenesis and extravillous trophoblast migration and invasion (Delforce et al., 2019; Pringle et al., 2011). Dysregulated AGT expression has been associated with fetal growth restriction, preeclampsia, and hypertension (Delforce et al., 2019; Mistry et al., 2019; Vefring et al., 2010). Therefore, the detected change in AGT expression in PFBS-high dose placentas likely signifies a disruption in the RAS which may lead to clinically relevant outcomes for maternal and fetal health. It is likely that only one gene was detected as significantly differentially expressed due to the pooling of placentas and the small number (n = 3) of samples for RNAseq. However, these constraints pointed to a large biological effect of PFBS exposure on AGT expression, and future experiments should expand on this finding. Additionally, future RNAseq experiments should investigate individual samples by group and stratified by sex. Although the expression levels of AGT in the regions of rabbit placentas is unknown, the sizes of placental layers or regions are less likely responsible for the changes as we did not observe differences in sizes of placental layers among groups.

One major strength of this study is that the PFBS doses used are lower than previous in vivo studies (Feng et al., 2017; Lieder et al., 2009a,b; NTP, 2019a), filling a critical knowledge gap on the range of doses known to induce adverse outcomes in animals. PFBS levels in drinking water in the United States have been reported as high as 185.9 ppt (Brunswick County, North Carolina) ((EWG), 2021), and the current EPA drinking water lifetime health advisory for PFBS is 0.002 mg/l, or 2000 ppt. Although the doses used in this study exceed human exposure levels, the data are valuable as the sensitivity of NZW rabbits relative to humans is still not known for PFBS. It is possible that NZW rabbits are less sensitive than other model species, which would require using high exposure levels to investigate various health effects. Animals are known to have much shorter elimination half-lives for PFAS than humans (Huang et al., 2019a; Olsen et al., 2007a; Olsen et al., 2009) so differences in toxicokinetics and toxicodynamics also necessitate the use of higher doses of PFAS through drinking water in animal studies.

Additionally, humans are likely co-exposed to PFBS through drinking water and via other routes, including diet, household dust, consumer products, and/or occupational exposures. Here, rabbits were exposed acutely over 32 days. To extrapolate results, higher doses are required to mimic chronic human exposure. Finally, this study evaluated maternal effects during pregnancy. Pregnant women are more vulnerable to environmental stressors. Higher doses are critical in risk assessment studies to ensure even the most vulnerable individuals are protected.

Limitations of this study include that there are interspecies differences between rabbits and humans, including gestational length, and thus more limited timeframe for fetal exposure; number of offspring; and uncertainty regarding the half-life of PFBS in rabbits, which likely differs from humans. Another limitation is that fetal PFBS blood levels were not measured due to the small volume of fetal blood collected. However, it is well established that PFBS cross the placental barrier, as evidenced by PFBS measurements in human cord blood (Huang et al., 2019b, 2020; Yao et al., 2019; Zheng et al., 2022). Toxicokinetic studies are required to investigate how rabbits differ from humans.

In conclusion, this research utilized a maternal rabbit model to provide fundamental evidence that in utero PFBS exposure leads to adverse effects on maternal, fetal, and placental outcomes. PFBS-exposed dams demonstrated dysregulated pulse pressure and renal RI measures, as well as histopathological changes to the kidneys, whereas PFBS-high dose reduced fetal growth. Placentas from PFBS-high dose male fetuses exhibited lower weights. The BW:PW ratio was overall decreased in PFBS-high dose fetuses, and this ratio showed sex-specific effects with exposure. Finally, we demonstrated altered expression of genes involved in thyroid hormone placental transfer and significant disruption of a gene critical to blood pressure regulation. As PFBS exposure led to disrupted fetal and placental endpoints, future animal and population studies should incorporate placental endpoints and related outcomes for offspring exposed to PFBS during pregnancy.

Supplementary data

Supplementary data are available at Toxicological Sciences online.

Data availability

Full transcriptome sequencing of placental mRNA raw data files are accessible through the Duke Data Repository: https://doi.org/10.7924/r4w66k88z.