Molecular pathways in placental-fetal development and disruption

Abstract

The first trimester of pregnancy ranks high in priority when minimizing harmful exposures, given the wide-ranging types of organogenesis occurring between 4- and 12-weeks’ gestation. One way to quantify potential harm to the fetus in the first trimester is to measure a corollary effect on the placenta. Placental biomarkers are widely present in maternal circulation, cord blood, and placental tissue biopsied at birth or at the time of pregnancy termination. Here we evaluate ten diverse pathways involving molecules expressed in the first trimester human placenta based on their relevance to normal fetal development and to the hypothesis of placental-fetal endocrine disruption (perturbation in development that results in abnormal endocrine function in the offspring), namely: human chorionic gonadotropin (hCG), thyroid hormone regulation, peroxisome proliferator activated receptor protein gamma (PPARγ), leptin, transforming growth factor beta, epiregulin, growth differentiation factor 15, small nucleolar RNAs, serotonin, and vitamin D. Some of these are well-established as biomarkers of placental-fetal endocrine disruption, while others are not well studied and were selected based on discovery analyses of the placental transcriptome. A literature search on these biomarkers summarizes evidence of placenta-specific production and regulation of each biomarker, and their role in fetal reproductive tract, brain, and other specific domains of fetal development. In this review, we extend the theory of fetal programming to placental-fetal programming.

1. Introduction

Understanding the causal effects of maternal exposures to endocrine disrupting chemicals on placental-fetal endocrinology and long-term child health outcomes is a priority area in public health, obstetrics, and pediatrics (Buckley et al., 2020; Braun, 2017). For ethical reasons, experimental studies in pregnancy are not generally feasible as a source of causal knowledge in humans, animal studies fall short due to between species differences, and primary tissue in vitro models are not able to accurately recreate the signaling of maternal, placental and fetal tissues in human development (Park et al., 2023). Fortunately, however, molecular biomarkers measured in human tissues in the first trimester and during critical windows of organogenesis can offer valuable insight into developmental processes that might otherwise benefit from experimental manipulation. Such biomarkers can connect maternal exposures and child outcomes separated by large spans of time in development. The aim of this review is to summarize what is known about 10 specific molecular pathways considered to be of relevance in impacting placental-fetal development with consequences for maternal, perinatal, and child outcomes.

The first trimester (up to 14 weeks gestation, i.e., from 0 to 12 weeks post-conception) is a privileged developmental window to measure tissue-specific biomarkers that are informative of specific types of exposures and perturbations of developmental pathways. Most fetal organs are fully formed by 12 weeks’ gestation although functional maturation occurs throughout pregnancy (Gilbert, 2003). During this early period, there is a complex system of fluid exchange, molecular transport, and tissue differentiation that includes the embryo, the placental villi, the chorion and the amnion (Adibi et al., 2021). Biomarkers, such as hormones, growth factors, cytokines and antibodies, generated by and transported into the gestational sac (includes placenta, chorion, amnion, exocoelomic cavity, yolk sac, umbilical cord) are relevant to this scenario as they supply essential hormones for fetal development (Jauniaux and Gulbis, 2000; Jauniaux et al., 2006; Nagy et al., 1994). Developmental toxicology studies can assess the direct effects of chemical exposures on fetal tissue; however, toxicity within non-fetal tissues in the gestational sac may be equally or more important in the first trimester.

A circulating serum biomarker is a cumulative measure of molecule synthesis, secretion, metabolism, and excretion. Thus, a correlation between a serum concentration and levels in the target tissue cannot be assumed without validation studies (in this case, the placenta and/or a fetal organ). Often measuring multiple molecules within a pathway of interest is more informative than assessment of a single molecule. It increases confidence that what is measured is real as opposed to artifact. Understanding if effects are autocrine (within cell), paracrine (in the context of this review defined as placenta-fetal and within gestational sac), or endocrine (placental-fetal-maternal) is a crucial aspect in designing a biomarker panel.

Fetal genetic sex, defined as the presence or absence of the Y chromosome, can affect variation in Y-linked and X-linked genes, or in imprinted genes. Since the trophoblast cells originate from the embryo (Benirschke and Kaufmann, 1995), placental sex is the same as fetal sex. Fetal sex can operate as a source of genetic and/or epigenetic variation in how a gene is transcribed, and how a protein is translated, secreted, metabolized or regulated. Later in pregnancy, these differences could contribute to paracrine/endocrine mechanisms by which the nascent fetal hypothalamus, pituitary, thyroid, adrenal glands and gonads engage in feedback mechanisms with the placenta.

Another critical issue in designing first trimester biomarker studies is to consider what happens at 10-weeks gestation (8 weeks post-conception) when maternal blood flow to the placenta begins (Hustin and Schaaps, 1987; Genbacev et al., 1997). This timepoint can reflect change in the expression of molecules involved in the response to oxygen tension such as HIF-1α and EGF (Genbacev et al., 1997; Caniggia et al., 2000; Caniggia et al., 2000; Armant et al., 2006), and the change in the movement of molecules within the gestational sac and between maternal-placental-fetal circulation (Burton et al., 2002). The 10th week of gestation is a well-established inflection point whereby maternal serum human chorionic gonadotropin (hCG) levels reach their peak and then start decreasing (Nagy et al., 1994). Ten weeks also coincides with the key fetal developmental milestones, including initiation of steroid hormone synthesis in the fetal gonad, adrenal, and pituitary (Moore et al., 2008; Yen et al., 2014; Scott et al., 2009). Importantly, events which are understood as occurring separately in the maternal, placental and fetal compartments may be mechanistically related.

The aim of the review is to extend the theory of fetal programming to placental-fetal programming. We organize knowledge from diverse disciplines into a theoretical framework which can be applied widely to the study of placental-fetal endocrine disruption in early pregnancy. We chose 10 molecular pathways to review that were of particular interest for their known or suspected relationship to 1) hCG regulation or function, 2) endocrine disruption, and 3) aspects of early fetal development. While some pathways address other fetal organs, we narrowed our review and discussion of hCG and fetal development to the reproductive tract and neurodevelopment. Literature on the toxic effects of endocrine-disrupting chemicals on the placenta was excluded here as it has been extensively summarized elsewhere (Padmanabhan et al., 2021; Tang et al., 2020; Street and Bernasconi, 2020; Rolfo et al., 2020; Gingrich et al., 2020; Basak et al., 2020). In the process of conducting this review, we identified and highlighted critical gaps in current understanding of the basic biology of placental-fetal endocrinology. The established and well-tested model on placental hCG as a necessary signal to initiate male fetal steroidogenesis is one example of a type of relationship that may be applied more widely (Scott et al., 2009; Huhtaniemi et al., 1977). The framework presented here is designed for application in biomarker-based observational research, and in experimental studies employing cellular and animal models to further elucidate causal relationships that connect the placenta and the fetus. This knowledge can inform the design of epidemiologic studies employing molecular biomarkers, analyses of the placental and fetal transcriptomes and proteomes, and the selection of relevant endpoints to measure when modeling these relationships in vitro. Notably, more detailed background on hCG is given since hCG is of primary interest in our hypotheses, and is discussed in relation to the other 9 pathways.

2. Methods

hCG was used as a model molecule in the design of the search terms in a non-systematic review. To ‘systematize’ the approach, the same search terms were applied to all 10 categories. Format of search was: (hCG [tiab] OR human chorionic gonadotropin [tiab]) AND (newborn or neonatal) genitalia AND pregnancy). For each of the categories, 9 searches were conducted in PubMed that varied the name of the molecule and the outcome. The outcome variables were: genitalia, anogenital distance, hypospadias, cryptorchidism, reproductive tract, cerebral palsy, neurodevelopment, brain, fetal or prenatal origins. These outcomes were specifically chosen as they have been reported on previously as being associated with hCG (Eskild et al., 2018; Peycelon et al., 2020; Burton et al., 1987; Chedane et al., 2014; Kiely et al., 1995; Schneuer et al., 2016; Adibi et al., 2021; Adibi et al., 2015). For neurodevelopmental outcomes, ‘newborn or neonatal’ was replaced with ‘fetal.’ This was a starting point to which literature iteration was added from previous searches, and further discovered through review papers and bibliographies (Table 1). The average number of citations reviewed per topic was 141 which includes English-language literature (original research articles and review papers) from different fields of clinical medicine, endocrinology, developmental biology and epidemiology. Where relevant (Islam et al., 2022; Rulli et al., 2018; Rao et al., 2003), studies conducted in animals were included. The Human Protein Atlas (referred to as HumProtAtlas) is an online resource that is used throughout this analysis as a reference for mRNA and protein expression, primarily to 1) identify molecules that are expressed in the human placenta (yes/no) and 2) to compare expression between the placenta and other organs (i.e. placenta and relevant target organ). It is a comparative analysis of RNA, measured by sequencing, from 256 tissue types; and protein, measured by antibody-based assays and by mass spectrometry, from 44 tissue types. The Atlas includes the analysis of term placental tissue (N = 8, mean maternal age 33 years, standard deviation 4 years) (Atlas, 2005; Fagerberg et al., 2014; Uhlen et al., 2015; Sjostedt et al., 2018). Unfortunately, the HumProtAtlas does not include fetal organs. Details on the donors or the pregnancies are not provided. Since placental tissues may be contaminated by decidual cells or contain proteins expressed in other tissues, we interpret the genes to be expressed in placenta only when expression is reported both in mRNA and protein levels. The totals included in Table 1 were the numbers of papers reviewed before synthesis and citation. The searches were carried out between January and August 2021, and updated in 2023.

Table 1.

List of molecular pathways reviewed, and total number (N) of references reviewed in each category. These pathways were chosen due to role in 1) hCG regulation or function, 2) endocrine disruption, and 3) aspects of early fetal development.

N	Category	Abbreviations	N citations reviewed
1	Human chorionic gonadotropin	hCG	127
2	Thyroid hormones	TSH, FT4, T4, T3	88
3	Peroxisome proliferator activated receptor gamma	PPARγ	102
4	Leptin	LEP	217
5	Transforming growth factor beta	TGFβ	235
6	Epiregulin	EREG	0a
7	Growth differentiation factor	GDF15	19
8	Small nucleolar RNA	snoRNA, SNORA, SNORD	49
9	Serotonin	5-HT	291
10	Vitamin D	(25-OH) Vit D	286

FT4 – free T4, TSH – Thyroid Stimulating Hormone, T4 – thyroxine, T3 – triiodothyronine.

^aThere were no citations identified according to the search terms used for this analysis.

3. Review of 10 molecular pathways

3.1. Human chorionic gonadotropin (hCG)

3.1.1. Key properties of hCG

hCG is expressed within hours of conception by the embryo and by the trophoblast and is a diagnostic marker of pregnancy (Hussa, 1980; Woodward et al., 1993; Butler et al., 2013). hCG is generally considered specific to human pregnancy and the placenta; although it is sometimes expressed at lower levels elsewhere, e.g., by certain types of tumors, including molar pregnancy and germline tumors, and by the pituitary (Demir et al., 2019; Merhi and Pollack, 2013, Schmid et al., 2013). It is a heterodimeric protein that consists of an alpha-subunit (hCGα), shared with other glycoprotein hormones, and a unique beta-subunit (hCGβ). While the alpha subunits of hCG and thyroid stimulating hormone (TSH), luteinizing hormone (LH), and follicle stimulating hormone (FSH) are encoded by the same CGA gene and are identical in protein sequence, their glycosylation patterns and regulation of expression may be different, reflecting the tissues where expressed (Blithe, 1990a,b; Blithe, 1990a,b; Blithe and Iles, 1995; Matari et al., 2021; Winters and Troen, 1988; Albanese et al., 1996). Unlike the single CGA gene, there are several highly similar genes (see below) encoding hCGβ, and many of these contribute to placental hCGβ expression (Rull and Laan, 2005). During early pregnancy, placental extravillous trophoblasts produced the hyperglycosylated form of hCG, which is considered important for implantation and invasion of the placental cells into myometrium (Berndt et al., 2013; Kovalevskaya et al., 2002). Later, between 5 and 10 weeks of pregnancy, the expression of the non-hyperglycosylated form of hCG by syncytiotrophoblasts takes over (Fournier et al., 2015). Trophoblastic hCG stimulates the maternal corpus luteum to produce large quantities of progesterone in early pregnancy up until 8 weeks (Yen et al., 2014). This is necessary to prevent shedding of the uterine lining while implantation occurs, and before the placenta has developed sufficiently to begin its own production of progesterone. Concentrations of hCG and its free subunits in the first trimester are highest by orders of magnitude in the coelomic fluid as compared to the maternal serum and the amniotic cavity (Jauniaux and Gulbis, 2000); yet maternal serum levels are also increasing dramatically over this time period. Intact hCG (i.e., dimer of hCGα and hCGβ) and free hCGβ reach their maximum levels in maternal blood at 10 weeks gestation and then start to decrease; whereas hCGα continues to rise over the course of pregnancy (Nagy et al., 1994a,b; Stenman et al., 2006). As early as 3 weeks post conception, circulating levels of intact hCG were 19% higher in the serum of women carrying female vs. male embryos (Yaron et al., 2002), as measured by an assay detecting both hCG dimer and the free beta subunit.

3.1.2. Measurement of hCG

CGA and CGB messenger RNAs encode the alpha and beta subunits of hCG. They are transcribed separately, and their encoded proteins associate non-covalently to form intact hCG (Fournier et al., 2015). Six non-allelic CGB genes are clustered on chromosome 19q13.3 (CGB1, CGB2, CGB3, CGB5, CGB7, CGB8), and are expressed at varying degrees in placental tissue (Rull and Laan, 2005; Jameson and Lindell, 1988). Some of the genes are methylated at the promoter (Buckberry et al., 2014). The intact hCG dimer, hCGα, hCGβ, and hyperglycosylated hCG can be measured in maternal serum using different antibodies (Adibi et al., 2021; Stenman et al., 2011; Berger et al., 2013). Antibody-based clinical hCG assays can generate variable results, partly because different antibodies recognize different epitopes. Discussion of the quality control issues in antibody-based assays has been covered extensively elsewhere (Stenman et al., 2006; Sturgeon et al., 2009; Whittington et al., 2010). Mass spectrometry-based analysis is less commonly used for hCG but may bypass some of the issues with specificity in antibodies (Al Matari, Goumenou, Combes et al., 2021) and could facilitate the detection of proteins encoded by different CGB genes, undistinguishable by the present immunoassays. In some cases, there is evidence that hCG is synthesized in fetal tissues, and sequestered and adsorbed to receptors in non-placental tissues giving an antibody signal (Abdallah et al., 2004; Huhtaniemi et al., 1978; Rothman et al., 1992). In prenatal screening, the intact form of hCG is most commonly measured in maternal blood or urine, while the beta subunit is measured less frequently (Stenman et al., 2006; Sturgeon et al., 2009; Stenman and Alfthan, 2013; Nerenz et al., 2016).

3.1.3. hCG as an example of placental-fetal programming

hCG in the first trimester acts as a strong endocrine and paracrine signal, and is essential in the process of embryo implantation and later in stimulating trophoblast invasion into the myometrium (Filicori et al., 2005; Licht et al., 2007). hCG is secreted by the trophoblast into the exocoelomic fluid (a fluid cavity between the amnion and the chorion that only exists during the period of organogenesis) and the intervillous space. It travels to the fetal gonad where it stimulates the onset of steroidogenesis by binding to the luteinizing hormone/chorionic gonadotropin receptor (LHCGR) (Huhtaniemi et al., 1977; Molsberry et al., 1982; Tapanainen et al., 1981).

The period from 8 to 12 weeks gestation is considered to be a masculinization programming window (MPW) (Welsh et al., 2008; Scott et al., 2008). This is based on extrapolation from rodent studies. Human studies are supportive of the MPW and there is indirect evidence to determine the timing in humans (Fowler et al., 2011), and reviewed in (Scott et al., 2009). What occurs within the MPW is an example of fetal programming whereby specific signals to the fetal cells within a narrow time frame can have long-term effects on diverse aspects of sex-specific male development, including the male reproductive tract and external genitalia (Scott et al., 2009). The theory presented here of placental-fetal programming includes placental contributions to the development of the reproductive tract and genitalia, and sex differentiation of the brain. This theory has been extended recently with findings that placental progesterone in the second trimester is essential to fetal androgen synthesis and the masculinization of the male external genitalia (O’Shaughnessy et al., 2019).

The concept of the fetal placenta as an essential source of endocrine signaling to the fetus was demonstrated originally in second trimester (14–20 weeks gestation) experiments with five human fetal testes explants in which controlled exposures to hCG were correlated with secreted levels of testosterone in the conditioned media in a dose-dependent manner (Huhtaniemi et al., 1977). Decades later, transcriptomic studies of the human fetal gonad confirmed that the mRNA that encodes the hCG receptor (LHCGR) has an inflection point at 10 weeks gestation, consistent with the MPW (Del Valle, Buonocore, Duncan et al., 2017; Lecluze et al., 2020; Mamsen et al., 2017). Gonad gene expression was measured as a function of change over gestational time in the first and second trimesters. There was a unique molecular signature associated temporally with the onset of steroidogenesis. All three independent studies confirmed that LHCGR mRNA, along with CYP17A1 mRNA and other steroidogenic enzymes, are upregulated in the male testis at approximately 10 weeks. Causality in terms of testosterone production is inferred by charting change over gestational time in the increased expression of mRNAs that encode fetal receptors to the placental hormone hCG, simultaneous with the upregulation of mRNAs that encode steroidogenic enzymes. In a review paper, these relationships were depicted also by charting changing levels of secreted hormones over time in the first and second trimesters (Scott et al., 2009). The gold standard approach would be (yet to be done), is to establish these placental-fetal connections during the MPW by charting change over time in fetal and placental transcriptomes and proteomes, ideally with tissues sampled from the same pregnancies.

Parallel studies of the fetal ovary transcriptome confirm that the molecular mechanism that drives steroidogenesis and sex differentiation in the male is not present in the female fetus, operating under the assumption that LHCGR expression is a hallmark of steroidogenesis. For example, LHCGR mRNA expression was detectable in the fetal ovary but did not change over the period 7.7–11.3 weeks gestation (Mamsen et al., 2017). At the protein level, only one study was identified conducted in the rabbit fetus. The rabbit also depends on LH-LHCGR binding to initiate testosterone synthesis (Molsberry et al., 1982, Catt et al., 1975). In this study, they compared the ovary and the testis over a time course in pregnancy. The LHCGR receptor was non-detectable in the fetal rabbit ovary at the time in which LH-stimulated androgen synthesis began in the rabbit testis. At day 18 when the LH-LHCGR binding was robustly detected in the male testis, the female ovary had minimal to negligible levels of LH-LHCGR binding at 3–9% percent of that in the testis. In mature human ovaries hCG (as well as FSH and LH) exert via LHCGR steroidogenic effects in granulosa and theca cells (Casarini et al., 2018). The LHCGR is not operating in the female fetus as it is in the male or the adult female. It is curious how such an essential mechanism of placental-fetal development could be present in one half of the species and not the other.

The general assumption, dating to 1953 is that the female reproductive tract and genital development is a default pathway that occurs in the absence of androgens and which is not controlled by female ovarian hormones [reviewed in (Weiss et al., 2012; Rey et al., 2000)]. More recent theories state that there are active processes within the first trimester female ovary that are necessary for sex differentiation, genital and reproductive tract development; yet the molecular mechanism is not yet fully understood (Weiss et al., 2012; Nef et al., 2005; Cunha et al., 2018). Within the first trimester, the female fetal genitalia and reproductive tract undergo differentiation (Cunha et al., 2018). Evidence of this phenomena is the well-established causal effects of diethyl-stilbesterol exposure (DES), a xenoestrogen administered in the first trimester, on formation of the vaginal epithelium (Cunha et al., 2018). DES (a canonical endocrine disruptor) treatment caused the expression of the progesterone receptor and the estrogen receptor alpha (ESR1) genes in female reproductive tract tissues where they are not normally expressed. Morphologically, DES stimulated endometrial/cervical glands formation, increased folding of tubal epithelium, elicited stratified squamous maturation of vaginal epithelium and vaginal adenosis. All of these effects were observed in a xenograft model whereby human fetal reproductive tract tissue was transplanted into a non-pregnant mouse; and there was no placenta present, and ovaries were removed (Cunha et al., 2018). It is important in future work that these findings be translated to a model of pregnancy that includes the placenta and the maternal ovary. The implication is that endocrine-disrupting chemicals may activate a molecular pathway in the female fetus that is otherwise not expressed. Would the placenta have a role in this mechanism?

Similar to above, we reviewed the fetal ovary transcriptome findings with an eye towards consistent findings to support a novel placental-fetal mechanism. The first wave of mRNAs upregulated in the female gonad in the 9th week gestation fell into the categories of cell differentiation and cell receptor signaling generally (Lecluze et al., 2020). During gestational weeks 14–19, mRNAs were upregulated related to meiosis and oogenesis. This is consistent with commitment of the female germ cells to meiosis during this period. In 2 of the 3 gonad transcriptome studies, olfactory receptors and molecules involved in neurogenesis and neurotransmission were upregulated in the female ovary as compared to the male testes (Del Valle et al., 2017; Lecluze et al., 2020). These include OR10G9, GABRG1, ORD45, NPNT, CNTAP4, NPY, NRXN3, CNTN1. Neuropeptide Y (NPY) was consistently upregulated at 9–10 weeks in somatic cells (Del Valle et al., 2017; Lecluze et al., 2020). NPY is known to control female reproduction at the level of ovarian cell proliferation and apoptosis (Sirotkin et al., 2015). In the third study which relied on microarray technology (not complete enumeration of the transcriptome) as opposed to RNA sequencing (complete enumeration), they did not report upregulation of these same neuropeptide and neurotransmission mRNAs in the female gonad (Mamsen et al., 2017). However, they reported upregulation of 2 WNT signaling mRNAs (WNT2B, ETV5) in the female ovary as compared to the male testes over the 9–12 gestational week window (Mamsen et al., 2017).

3.1.4. Placental hCG and male reproductive health endpoints

The implication from the above findings is that factors that affect placental hCG synthesis in the period between 10 and 12 weeks can have lasting impacts on the masculinization of the male reproductive system, and on the external genitalia specifically. Extreme scenarios including mutations in the LHCGR, whereby the hCG signal is muted or absent, are associated with hypogonadism and 46XY disorder/difference of sex development (DSD) in males at birth, conditions treatable to some degree by postnatal hCG stimulation (Apn and Huhtaniemi, 2000; Kremer et al., 1995). In the late second trimester, the fetal pituitary begins to secrete its own gonadotropin, as hCG levels decrease [reviewed in (Scott et al., 2009)]. In studies of male anencephalic fetuses in which there is no fetal pituitary, the external genitalia developed normally [reviewed in (Rabinovici and Jaffe, 1990)]. This evidence argues that placental hCG is the primary source of gonadotropin for development of the external genitalia, whilst pituitary gonadotrophins are required for subsequent growth (Howard and Dunkel, 2018).

Based on this, one might predict that lower hCG levels in maternal and fetal circulation would be associated with a higher risk of male genital abnormalities; however empirical evidence suggests the opposite may be true. High first trimester serum hCG levels were associated with a higher risk of hypospadias (anti-androgenic action), and only in relation to proximal hypospadias (Peycelon et al., 2020; Schneuer et al., 2016). In another study, mean levels of hCG were 1.5-fold higher in women who gave birth to boys with hypospadias (Chen et al., 2019, and hCG levels were higher still in mothers who gave birth to boys with hypospadias and growth restriction (Yinon et al., 2010).

Consistent with the hypospadias findings, male neonatal anogenital distance (AGD), a genital marker of masculinization in utero, was inversely associated with first trimester serum hCG levels in 354 pregnancies (Adibi et al., 2015). Anogenital distance is the distance from the anus to the scrotum/penis in males and the distance from the anus to the fourchette/vagina in females (Salazar-Martinez et al., 2004). It can be measured non-invasively in the newborn infant as a sensitive marker of hormonal disruption pregnancy; and hence has become a hallmark measure of fetal endocrine disruption and predictive marker of adult reproductive health (Dean and Sharpe, 2013; Sathyanarayana et al., 2015). Higher hCG levels were associated with shorter (i.e., less masculinized) AGD in male infants; whereas in female infants higher hCG was associated with longer AGD (i.e., more masculinized) (Adibi et al., 2015). First, it is important to note that these are maternal serum levels of hCG which may or may not be correlated with fetal testicular concentrations. Second, these studies measured the intact and beta forms of hCG, and not the alpha subunit. Third, it is possible that there is a feedback mechanism between fetus and placenta such that hCG production decreases if testosterone production is adequate which could explain a negative correlation between maternal serum hCG and the risk of hypospadias. Lastly, it could also be a feedback mechanism in gonadotropin levels between the placenta and the fetal pituitary, suppressing hCG.

Cryptorchidism, a male birth defect in which one of the testes does not descend, has been studied as a relevant outcome in the context of hCG. There was no association of first trimester hCG and the risk of cryptorchidism in three out of four epidemiological studies identified (Burton et al., 1987; Kiely et al., 1995; Bernstein et al., 1988). In the fourth study, there was a 20% lower mean serum hCG, measured between 12 and 16 weeks gestation in mothers who delivered a cryptorchid son versus controls (Chedane et al., 2014). Cases in that study were tightly matched to controls on gestational age at time of blood draw, maternal age, birth size, and gestational age at birth. Control for gestational age at birth may have introduced a specific type of bias, referred to as collider bias (Rothman et al., 2021). Collider bias occurs when an exposure-outcome model is adjusted for factors that potentially occurred after the exposure and/or outcome. This can induce an arti-factual association. More work is needed to evaluate the unbiased association of hCG and the risk of cryptorchidism.

Based on these counter-intuitive associations of higher serum hCG and higher risk of certain anti-androgenic outcomes, we speculate that the mechanism is not as simple as originally hypothesized. hCG (as measured in these studies) was not positively associated with testosterone and/or masculinization of the male genitalia. From our in vitro studies of the first trimester placenta and the fetal gonad, we found that it was specifically the free alpha subunit of hCG (measured in placental conditioned media which was placed on the fetal gonad culture or as mRNA, CGA, in placental explants) that was positively associated with fetal testosterone secretion (Adibi et al. unpublished). This may offer a possible explanation why hCG has pro-androgenic effects in some studies, but not in all studies. This relationship may depend on which form of hCG or its subunits is measured or used in the dosing studies, and the effect may be detectable with some but not all forms of hCG. To note, production of hCGα has been stated to be the rate-limiting step in gonadotropin heterodimer formation (Casarini et al., 2018).

3.1.5. Placental hCG and female reproductive health endpoints

As described above, there is minimal empirical evidence that first trimester placental hCG or any other placental hormone plays a role in female fetal development, comparable to the role it plays in the male. Female fetuses can be virilized in utero as a result of increased maternal and/or fetal androgen levels. Outcomes at birth which are reflective of this are cliteromegaly (Castets et al., 2021), and possibly longer anogenital distance (Callegari et al., 1987). These types of developmental deviations are thought to originate later, after 13 weeks, when androgen production is under the control of the fetal hypothalamic-pituitary-gonadal (HPG) axis and not relying on placental signals (Castets et al., 2021). Congenital adrenal hyperplasia (CAH) is a genetic disorder that causes defects in fetal adrenal steroidogenesis, resulting in abnormally high androgen levels and virilization of female fetuses (Forest, 2004).

From the sparse literature on female fetal reproductive development, we seek clues on molecular mechanisms of female fetal virilization. These are all studies that did not measure placental tissue hCG. High levels of hCG produced by ovarian cysts can cause high levels of androstenedione and testosterone in the maternal compartment, but without virilization of the female fetus (Bradshaw et al., 1986; Csapó et al., 1999; Erkkola et al., 1985). Whereas, another type of ovarian cyst called a Krukenberg cyst was associated with high levels of hCG, high testosterone and a female infant born with ambiguous genitalia in a case study (Bustamante et al., 2017). The difference between these two scenarios may relate to production of hCG by the fetal ovary vs. the fetal placenta. They could be different forms of hCG or be distributed differently in fetal circulation. More work is needed to understand the relationship of hCG and androgens in pregnancies with female fetuses.

As a first step in filling this knowledge gap of a molecular mechanism to connect the placenta to the female fetus, we speculate there is a mechanism that involves neuroendocrine signals, and not androgens, and which is distinct from the steroidogenic signaling in the male. These signals might originate in the maternal neuroendocrine system, or the placenta-fetus may simply employ similar molecules for its own and different purposes. NPY mRNA was reported in two independent studies as being upregulated in the fetal ovary in early pregnancy (Del Valle et al., 2017; Lecluze et al., 2020). NPY encodes a neuropeptide characterized in the adult as important to stress response and food intake. The finding on fetal ovary NPY expression is curious as it is not expressed in the adult ovary. In adults, it is restricted to the adrenals, the cerebral cortex, and the male reproductive tract (HumProtAltas). NPY neurons in the adult are regulated by leptin levels which may be an essential clue as to its relationship to the placenta (Elias et al., 1999). Leptin is expressed highly in the first trimester placenta and is higher in female vs. male placentas (Helland et al., 1998).

The overall conclusion is that mechanisms of male placental-fetal sex differentiation are supported historically by reports in the literature and corroborated by recent RNA sequencing data; yet female fetal sex differentiation in early pregnancy is not well understood nor are there existing theories on placental regulation of female development. Data generated by RNA sequencing of fetal ovaries, combined with associations of secreted biomarkers reported in cohorts, give rise to a new theory on the existence of an axis that may connect neuroendocrine molecular signals (i.e., stress) with placental function (i.e., leptin) and female fetal reproductive tract development (i.e., NPY and AGD or other genital measures). The analysis of maternal psychosocial experience and/or endocrine disrupting chemicals that can affect neuroendocrine pathways, placental biomarkers in pregnant women and their placentas (e.g., leptin mRNA or protein), and quantitative measures in the female offspring (e.g., NPY or anogenital distance) is one approach to ‘connect these dots’ and to quantitatively evaluate this theory.

3.1.6. Placental hCG and fetal neurodevelopment

At least three hypotheses are distilled from the literature and offered here to connect placental hCG and fetal brain development. One theory is based on the idea that placental hCG binds to embryo/fetal receptors in the brain to stimulate and regulate brain development (i.e., direct effect). Another theory is that hCG regulates other endocrine organs that in turn regulate brain development, namely the maternal thyroid and the fetal gonad (i.e., indirect effect) (Adibi et al., 2021a,b,c). A third theory is that general placental function (nutrient transfer, gas exchange, detoxification, immune tolerance), measured as morphologic differences, can impact both hCG concentrations and brain development and child cognition (Misra et al., 2012). The first two of these theories involve circulating biomarkers in the first trimester and are discussed in more detail below.

Curiosity regarding a direct effect of hCG during pregnancy on fetal neurodevelopment is based in part on the finding that the hCGα subunit specifically, and not the intact or the beta forms, was positively associated with infant cognition at 1 and 3 years (Adibi et al., 2021b). This was a small pilot study (N = 50) in which second trimester serum hCG was analyzed jointly with serum thyroid hormone. hCGα, independent of thyroid hormone, was positively associated with infant cognition at 1 year and with expressive communication at 3 years, but not associated with infant cognition at 3 years. Interestingly, the association of maternal serum thyroid stimulating hormone (TSH) and infant cognition at 3 years of age was 3-fold stronger when measured jointly with hCGα versus without. FT4 (free thyroxine) and TSH are typically associated with development of the cortex and cognition (Moog et al., 2017; Korevaar et al., 2016). A mother’s gonadotropin levels may be correlated with her baby’s brain development by way of maternal overall health and well-being, as opposed to a ‘direct effect’ of hCG on brain development. This can only be teased apart with the measurement and comparative analyses of extensive placental molecular and morphologic measures, maternal levels of serum hCG, and their associations with brain development.

The placenta and the fetal pituitary both have high concentrations of the alpha subunit of hCG (Hagen and McNeilly, 1975; Kaplan et al., 1976; Kaplan and Grumbach, 1976; Dubois et al., 1978). This could be evidence that pituitary (fetal and maternal) and/or placental levels of the alpha subunit play an unrecognized and essential role in brain development. This idea of restricted alpha subunit expression by only two organs in the body, the placenta and the pituitary, was confirmed in the HumProt Atlas. CGA mRNA and protein were exclusively expressed in the anterior cells of the adult pituitary and in the trophoblast cells of the placenta (HumProtAtlas). hCGα levels in the gestational sac and in the fetal pituitary both reach peak levels in early pregnancy while levels of fetal LH, FSH, and TSH are low to non-detectable (Nagy et al., 1994a, b; Kaplan and Grumbach, 1976). Correlations between placental and pituitary forms of hCG (all forms) have not been studied. The maternal pituitary is also a source of hCGα which complicates the ability to isolate the source and the target in pregnancy (Lähteenmäki, 1978). The expression in maternal (or later in pregnancy fetal) pituitary is low compared to that in placenta (Jauniaux and Gulbis, 2000; Clements et al., 1976; Korhonen et al., 1997). It may be that these three production sites of hCGα are all important to fetal development, and possibly interact with each other.

Krieger outlined a group of pituitary peptides made by the placenta including hCG, prolactin (PRL), adrenocorticotrophic hormone (ACTH), pro-opiomelanocortin, and others with predictions on how they may function in autocrine signaling or in endocrine signaling (Krieger, 1982). The HumProt Atlas corroborates Krieger’s hypothesis with fully enumerated data on a small set of mRNAs/proteins (CGA, CGB8, CSHL1, GH2), that are highly and uniquely enriched in both the placenta and the pituitary. While there is no evidence, these proteins may further support the theory of placental inputs to the fetal pituitary and neurodevelopment. There may be a period, before fetal production of gonadotropins begins, when placental production of the hCGα subunit and other gonadotropins may indeed be essential to brain development (analogous to hCG and LHCGR-stimulated steroidogenesis in the testis). These essential molecules come from the placenta (vs. maternal tissues). We speculate that this is dictated by the fact that the molecules are too large to cross the placenta.

These and other molecules are elaborated on in a review that describes a class of ‘placental neurohormones’ and their role in placental function and placental disorders (Reis et al., 2001). This class of neuropeptides produced by the placenta includes hCG, thyroid releasing hormone (TRH), human placental lactogen (hPL), PRL, GHRH (growth hormone releasing hormone), ACTH, corticotropin-releasing factor (CRF), leptin and NPY. Maternal levels of NPY and leptin are positively associated with the risk of preeclampsia and negatively associated with the risk of fetal growth restriction [reviewed in (Reis et al., 2001)]. Preeclampsia is a disorder associated with diverse mechanisms including inflammation, immune tolerance, shallow invasion, poor vascularization (Borzychowski et al., 2006; Redman, 2014; Erlebacher, 2013), and will be referred to throughout the Review as it is well studied as a placental disorder. Preeclampsia is also relevant here as it is associated with high serum levels of hCG (Jelliffe-Pawlowski et al., 2015; Barjaktarovic et al., 2019), the hCG association with preeclampsia is stronger in male pregnancies (Steier et al., 2002) and preeclampsia is associated with higher risk of hypospadias and cryptorchidism in male babies (Arendt et al., 2018). Many of these same neuropeptide molecules are described in a review on placental ‘neuroactive factors’ and are described specifically in their role in mediating immune responses in pregnancy (Sun and Sun, 2022). ‘Neuroactive’ refers here to maternal neuroendocrine function, and not fetal brain development. However, the role of these placental neurofactors in fetal brain development may be important to consider; yet is not well studied.

There is additional evidence that the alpha subunit of hCG, made by the pituitary and the placenta, are indeed operating separately and distinctly from their dimeric proteins, beta subunits or other gonadotropins. The hCGα subunit measured in the fetal pituitary did not differ by fetal sex, but levels of LH and FSH were higher in female vs. male fetal pituitaries in early pregnancy, and then FSH reversed in late pregnancy such that male levels were higher (Rabinovici and Jaffe, 1990; Kaplan and Grumbach, 1976; Dubois et al., 1978). An analysis of the hCGα subunit in adult men with idiopathic hypogonadotrophic hypogonadism showed that the α subunit may be synthesized and secreted independently of the β subunits of glycoprotein hormones hCG and luteinizing hormone (LH) (Winters and Troen, 1988). The synthesis of the β subunit of hCG in the early conceptus/blastocyst rises steeply during the first 22 days after embryo implantation; whereas the alpha subunit rises later (Hay, 1985).

3.1.7. A global theory on placental gonadotropins

The ratio of the gonadotropin LH to sex steroids in adults is associated with neurogenesis, cognition, and dementia at the time of menopause and andropause [reviewed in (Atwood and Bowen, 2015)]. In the context of pregnancy, we propose that the analogous gonadotropin, hCG, is essential for embryogenesis in a global sense. There is plausibly a ratio measure of hCG to other hormones that is associated with aspects of fetal brain and reproductive development over a long period of time. The ratio of placental gonadotropins to sex steroids in fetal life may be an indicator of life-long function of the hypothalamic-pituitary axis (HPA). These associations in pregnancy of the ratio of hCG to sex steroids and developmental outcomes (brain, reproductive) have not been studied, to our knowledge. This is a theory which we put forth here.

This idea of an essential relationship between placental and fetal brain gonadotropin is supported by a study in mice in which corresponding genes in the placenta and the fetal and postnatal brain were measured by whole genome bisulfite sequencing (methylome) and RNAseq (transcriptome) (Islam et al., 2022). The aim of the study was to specifically identify the set of methylated genes and pathways representative of the ‘epiclock’ of brain aging, and which show coordinated expression between the placenta and the fetal brain. They studied males and females to identify sex bias in the mechanism. It was specifically the gonadotropin releasing hormone receptor (GNRHR) gene, and GNRHR pathway genes (Pbx1, Atf3, Rap1b, Fos, Grb2, Zeb1, Smad4, Cdc42, and Raf1) that had coordinated methylation and expression patterns between the placenta and the fetal brain, and the fetal brain and the postnatal brain. In this study, they compared the epiclock genes to the non-epiclock genes. An interesting finding was that only the epiclock genes had coordinated expression between the placenta and the fetal brain; whereas the non-epiclock genes did not. Methylation was notably higher in the female vs. the male tissues. GNRHR is expressed by the pituitary and is essential in the stimulation of LH and FSH production and in reproduction. GNRHR is expressed in the human placenta in early pregnancy and regulates the pulsatile secretion of hCG (Sasaki and Norwitz, 2011). These may be important relationships to study in human placental and fetal brain tissue to determine if this is a conserved and important mechanism of fetal origins of adult brain function and aging.

An evolutionary perspective, in line with the theory of a direct effect of placental proteins on fetal brain development, postulates that the placenta is an adaptation in primates that has allowed for the development of language and higher order executive functions in the neocortex (Montiel et al., 2013). It is possible that this is not a specific molecular mechanism but explained by the fact that the human placenta is enabling much greater transfer of nutrients, etc. needed to fuel development of a larger brain and higher order cognition. Following the molecular mechanism theory of Montiel et al., they report on a connectome analysis with gene expression restricted to placental and neocortical cells. They mined these data for candidate genes and pathways that share a molecular mechanism in fetal life, using knowledge of protein-protein interactions (Szklarczyk et al., 2011). They found that the placental chorionic gonadotropin beta isoform 5 (CGB5) and other placental peptides were predicted to interact with fetal cortical chromogranin A (CHGA). CHGA is a neuropeptide involved in the control of neurosecretion, and associated with inflammation, schizophrenia, and Alzheimer’s Disease.

Another reason to believe that hCG is targeting cells in the fetal brain is because the hCG receptor (LHCGR) is expressed in the central nervous system, including the brain and the spinal cord. LHCGR expression (mRNA, protein) was highest in the adult testis and in neurologic tissues (spinal cord, basal ganglia) (HumProtAtlat), consistent with the original findings in the rodent (Rao et al., 2003; Apaja et al., 2004) (the rodent expresses LH and not hCG but they both bind the same LHCG receptor).

A direct effect of hCG on brain development is hypothesized in the case of cerebral palsy (CP). Epidemiologic evidence demonstrates an association of low hCG with higher risk of CP (Eskild et al., 2018). Deprivation of oxygen to placental and/or fetal brain cells at different points in pregnancy are presumed to be the cause of CP. Experimental findings show that dosing fetal cells with hCG induced a neuroprotective effect, countering the effects of low oxygen (Movsas et al., 2017). The molecular mechanism that explains hCG effects on CP pathophysiology is not described.

The second theory by which hCG can affect fetal brain development is by way of its indirect actions on other endocrine tissues, namely maternal thyroid and fetal gonads. In turn, these hormones regulate aspects of fetal brain development (Rulli et al., 2018). In the following section, this indirect pathway via regulation of thyroid function is described in more detail.

3.2. Thyroid hormones

3.2.1. The maternal and fetal thyroid system during pregnancy

The thyroid system consists of thyroid-releasing hormone (TRH)-producing neurons located within the paraventricular nucleus of the hypothalamus, the pituitary which releases thyroid-stimulating hormone (TSH) and the thyroid gland which releases thyroid hormones (TH) in response to TSH stimulation; predominantly thyroxine (T4) and to a lesser extent triiodothyronine (T3). Circulating T3 and T4 are, for the most part, bound to binding proteins such as thyroxin binding globulin (TBG); less than 1% are present in free (unbound) form. Different tissues modulate the activity of circulating THs according to their current needs via three iodothyronine deiodinases (D1, D2, and D3).

The hormonal and metabolic adaptations produced by the state of pregnancy induce major changes in maternal thyroid physiology. During the first trimester, a surge in T4 and T3 concentrations occurs and a concurrent suppression of circulating TSH due to negative feedback on the pituitary. This pattern likely serves to ensure adequate levels of substrate to meet the increased distribution volume and is mediated by the rising concentrations of placental hCG (Korevaar et al., 2017), which is structurally similar to TSH (i.e., they share the same alpha subunit) and also has thyrotropic effects as described in more detail in section 3.2.2. Overall, there is an increased demand on the maternal thyroid with an increase in TH production by approximately 50% throughout gestation due to estrogen-dependent increases in TH binding by TBG and degradation of TH to inactive forms by D3 in the placenta (Braunstein, 2011). The onset of fetal thyroid function occurs around 16–20 weeks of gestation (Williams, 2008; de Escobar et al., 2004). However, even before onset of fetal thyroid function, significant levels of TH can be found in coelomic fluid, the fetal brain and other tissues, and levels continue to increase over the course of gestation (Chan et al., 2002; Kester et al., 2004). These observations suggest that active transport of maternal TH across the placenta has to occur during this early period of gestation, and TH transporters have been identified in human placental tissue (Loubiere et al., 2012; Chan et al., 2006). In terms of quantity, T4 is the primary TH that gets transferred across the placenta (Calvo et al., 2002). T4 has low biological activity, and its activation to the biologically highly active T3 is regulated by and depends on the availability of deiodinases D2 and D3 in specific tissues.

3.2.2. Regulation of maternal and fetal thyroid function by placental hCG

Thyroid hormone is a unique example of a 2-way relationship between the mother and the placental-fetal compartment. It was first proposed that placental hCG is ‘thyrotropic’ with the observation that rising serum levels of hCG in the first trimester were associated with increasing levels of free thyroxine (FT4) and decreasing levels of thyrotropin (TSH)(Nisula and Ketelslegers, 1974; Higgins and Hershman, 1978; Taliadouros et al., 1978; Harada et al., 1979; Pekonen and Weintraub, 1980; Guillaume et al., 1985; Pekonen et al., 1988). The relationship of hCG and TH was later convincingly demonstrated in Glinoer et al.(Glinoer et al., 1990), and more recently in the Generation R study where serum levels of hCG, FT4, and TSH in over 5000 pregnancies were measured. Intact hCG and FT4 followed a parallel trend over time, both reaching a peak at 10 weeks and then decreasing. hCG and TSH changed in opposite directions from 9 to 18 weeks with inflection points at 10 weeks and at 17 weeks (Korevaar et al., 2017). A higher amplitude and prolonged duration of the hCG peak, as seen for example in twin pregnancies, have furthermore been shown to be associated with a more profound suppression of TSH and increase in FT4 (Grun et al., 1997).

Further support for this relationship has been established by studying cases of extreme production of hCG. For instance, choriocarcinoma (trophoblastic tumor) can cause both the over secretion of hCG and hyperthyroidism in pregnant women (Higgins and Hershman, 1978). Preeclampsia is also associated with abnormally high levels of hCG (Barjaktarovic et al., 2019), and with hyperthyroidism (Millar et al., 1994). These associations strongly suggest a causal relationship. An in vitro study of human thyrocytes further proved that hCG has the ability to stimulate T3 secretion, and other functions of the thyroid cells (Kraiem et al., 1994). The basis for thyroid stimulation by hCG is likely related to the structural similarity between TSH and hCG, and their receptors, TSHR and LHCGR, such that hCG activates the same domain of the TSH receptor that TSH activates. This stimulation of the maternal thyroid system by a placental mechanism promotes the system’s adaptation to the increased demands of the gestational state and ensures the provision of developmentally necessary TH to the embryo/fetus at a time at which the fetal thyroid gland still lacks the capability of synthesizing TH.

hCG may further regulate fetal thyroid homeostasis by way of stimulating expression of the placental sodium iodide symporter (NIS) (Arturi et al., 2002; Burns et al., 2013; Li et al., 2007) leading to an increased iodide uptake. Iodide is essential for the synthesis of thyroid hormones and the fetal thyroid develops the ability to accumulate iodide by 10–12 weeks gestation (Burrow et al., 1994), corresponding with the hCG peak in normal pregnancies.

3.2.3. Placental transport and regulation of thyroid hormones

Transplacental transport of THs from maternal circulation into the fetal compartment is modulated by several factors including trans-membrane transporters in the placenta, deiodinase enzymes which metabolize TH, and TH binding proteins within trophoblast cells (Forhead and Fowden, 2014; James et al., 2007). Thyroid hormone transporters include members of the organic anion transporting peptides family (OATPs), L-amino acid transporters (LATs) and monocarboxylate transporters (MCTs) (Visser et al., 2008; Bernal, 2005), the majority of which are non-specific for TH. Several TH transporters have been demonstrated to be present in the placenta, including MCT-8, MCT-10, LAT-1, LAT-2, OATP1A2, and OATP4A1 (Chan et al., 2009). An in vitro investigation of the specific contributions of the individual transporters in placental cell lines suggests an important role of LATs and MCT-10 in T3 (but not T4) transport across the placenta, with the caveat that MCT-8 is not present in these cell lines (Chen, van der Sman, Groeneweg et al., 2022).

The placenta expresses two deiodinase enzymes which regulate fetal-placental TH levels. D2 is responsible for the local conversion of T4 to bioactive T3, whereas D3, the predominant deiodinase subtype in the placenta, inactivates T4 and T3 into reverse T3 or diiodothyronine, respectively (Forhead and Fowden, 2014). Placental D3 activity is highest in the beginning of pregnancy and declines over gestation, likely serving to prevent the fetus from excess TH and to provide the fetal thyroid with iodine recycled from the metabolism of iodothyronines (Moleti et al., 2014).

3.2.4. Thyroid hormones, placental hCG and fetal neurodevelopment

A wealth of animal studies with experimentally induced developmental hypothyroidism suggest that a lack of TH during prenatal and early postnatal development can result in significantly altered cortical cytoarchitecture, delayed cortical maturation, and abnormal intra- and extracortical structural and functional connectivity (Moog et al., 2017). Research is lacking on the association of maternal thyroid function in pregnancy on fetal, newborn or infant brain structure or function in humans. However, studies have observed an association between a maternal diagnosis of hypothyroidism and morphological alterations in the corpus callosum and hippocampus in school-aged children (Samadi et al., 2015; Willoughby, McAndrews and Rovet, 2014; Willoughby et al., 2014a,b; Willoughby, McAndrews and Rovet, 2014). As described above, hCG is an important regulator of maternal (and fetal) thyroid function and thus may moderate the association between TH and neurodevelopment during gestation. In an analysis of TH and child brain health at 8 years in Generation R, the effects of TH on child IQ and cortical plate thickness were independent of hCG concentrations (Korevaar et al., 2016). In a similar type of analysis, we studied the joint effects of hCG, TSH, and FT4 on infant cognition (Adibi et al., 2021b). We compared the model that treated hCG as a confounder vs. the model that treated hCGα and hCGβ as effect modifiers of TSH effects on cognition. There was a marked increase in the model fit in the effect modification model. From this, we conclude that hCG is more aptly characterized as an effect measure modifier vs. a confounder of the TH association with infant cognition. Of note, hCG and TH were measured in two different developmental windows in the two studies which may limit our ability to compare results.

3.2.5. TH and fetal reproductive tract development

The placental-fetal axes related to reproductive and neurodevelopment described here may not be operating independently of each other. For example, in two separate studies, both serum hCG and cord blood levels of thyroid hormone were associated with newborn AGD. In both studies, the associations differed in direction by the sex of the infant (Adibi et al., 2015; Liu et al., 2016). Preeclampsia was associated with higher risk of hypospadias and cryptorchidism in two population and hospital-based studies (Arendt et al., 2018; Wang et al., 2022); and in one study maternal hyperthyroidism was independently associated with the highest risk of hypospadias compared to hypertensive disorders and multiple births (Wang et al., 2022). Even though they did not measure hCG in these studies, high hCG (Barjaktarovic et al., 2019) and low TSH are associated with a higher risk of preeclampsia (Levine et al., 2009; Turunen et al., 2019). It is interesting to consider the ways in which placental hCG and maternal thyroid hormone regulation overlap during the fetal period, and jointly regulate fetal reproductive tract and brain development. hCG and thyroid hormone levels are correlated in the first trimester, both are vulnerable to maternal environmental exposures, and both may provide useful information on long-term risks to child health.

3.3. Peroxisome proliferator activated receptor protein (PPARγ)

PPARγ is a member of the nuclear receptor family that includes 48 human transcription factors (Lehrke and Lazar, 2005), activated by the binding of steroid and thyroid hormones, vitamins, lipid metabolites, and xenobiotics (Chawla et al., 2001). In pregnancy, PPARγ expression is coordinated between the placenta and the fetus such that placental expression is causally related to heart defects in rodents (Barak et al., 1999). In the human placenta, PPARγ is a transcriptional regulator of genes encoding and regulating placental hCG and progesterone and is highly expressed at the placental interface with maternal tissue beginning in the first trimester of pregnancy (Bogacka et al., 2015; Fournier et al., 2011; Handschuh et al., 2007; Hu et al., 2017; Rouault et al., 2016). Functions of PPARγ across different gestational tissues and cell types include lipid storage, macrophage maturation, embryo implantation, adipocyte differentiation, and inflammatory control (Michalik et al., 2002).

Within the placenta, a primary function of PPARγ is to promote differentiation of the mononucleated trophoblast cells to multi-nucleated syncytiotrophoblasts (hCG is a marker of this process), and to regulate the transfer of long-chain polyunsaturated fatty acids (LCPUFA) from maternal to fetal circulation (Schaiff et al., 2000; Uauy et al., 2000; Waite et al., 2000; Schaiff et al., 2005). Long-chain fatty acids are required for proper brain development (Uauy et al., 2000). Placental trophoblast cells do not make or store lipids or fatty acids. Once inside the trophoblast, LCPUFA can up-regulate expression of PPARγ which, in turn, can upregulate expression of the genes for fatty acid transporters, fatty acid acyl-CoA-synthetases and adipophilin or other enzymes involved in lipolysis, all of which is consequential for fetal levels of LCPUFAs and fetal brain development and possibly other developmental pathways (Gil-Sánchez et al., 2011; Muczynski et al., 2012).

A germline polymorphism in the PPARG gene called PPARG Pro12Ala, was associated with long-term differences in child neurodevelopment in 138 mother-infant pairs (important to remember here that the fetus and the placenta are the same genotype). At 18 months of life, the offspring with the wild-type PPARG allele had higher cognition, language and motor development than the offspring with the polymorphism. In this study, they did not detect differences by PPARG allele in fatty acid concentrations in blood or in placental tissue (Torres-Espinola et al., 2015). Given no differences were observed in these biomarkers, it remains an open question if there was a PPARG-related mechanism that explained the differences in brain development and how it occurred.

In another study, genetic variation in the PPARG gene was associated with adverse reproductive and developmental outcomes. Given the importance of PPARG in placental development and function, this might be another important subgroup to study PPARG-mediated mechanisms of placental-fetal endocrine disruption. In this study, 26 individuals from 15 independent families were identified based on a diagnosis of familial PPARG-related lipodystrophy (Gosseaume et al., 2023). All families carried a heterozygous PPARG variant which varied across the families (of the 15 families, only 2 carried the same variant). Of 16 females in the sample, 12 (75%) had polycystic ovarian syndrome (PCOS), a developmental disorder in females associated with overproduction of androgens. PCOS is generally believed to reflect an interaction of susceptibility genes and environmental factors in fetal and early postnatal life [reviewed in (Parker et al., 2021)]. One female in this study who had a homozygous variant in the PPARG gene presented at birth with signs of virilization, severe developmental delay secondary to perinatal brain damage, and with growth retardation. There was higher than average prevalence of adverse obstetric outcomes (preterm labor 53%, small for gestational age 43%, large for gestational age 36%), suggesting abnormal placental function. Consistent with the lipodystrophy diagnosis, all women in the study who conceived also had at least one metabolic disorder in pregnancy (i.e., diabetes, hypertension, hyper-triglyceridaemia, gestational diabetes mellitus, preelcampsia). It is not possible to draw conclusions on how these findings translate to ‘normal variation’ in the expression of PPARG and PPARG-regulated genes. Both of the above studies on DNA sequence did not include the study of mRNA and protein biomarkers in placental or fetal tissues. They also did not measure interactions of environmental factors and genotype. However, the phenotypes associated with PPARG gene variation support the theory that the PPARγ pathway is important in placental and fetal reproductive and neurodevelopment.

PPARγ expression is modulated by factors in the maternal environment. Diabetes in pregnancy can alter placental PPARγ expression and the transfer of omega-3 or n-3 fatty acid docosahexaenoic acid (DHA), changes associated with long-term child neurodevelopment (Judge et al., 2016) and also with hyperglycemia in the offspring (Zhao et al., 2019). Similarly, phthalate-induced changes in placental PPARγ expression, may be a cause of child health outcome by way of the reduced availability of essential fatty acids and lipids at key points during fetal development (Adibi et al., 2017a,b; Xu, Agrawal, Cook et al., 2006, 2008).

Based on correlation studies, there is a sex difference in the PPARG/PPARγ regulation of placental hCG that may provide a mechanistic basis for other types of sex differences that originate in the first trimester. In isolated trophoblast cells, sex-specific correlations of cellular and secreted PPARG/PPARγ and CGA/CGB/hCG were reported (Adibi et al., 2017b). In fetal liver cells, there were sex differences reported in PPARγ− mediated effects of gestational diabetes mellitus (GDM, induced in the pregnant dams) on triglycerides and cholesterol levels (Fornes et al., 2018). Male GDM fetuses had elevated triglycerides and cholesterol levels as compared to controls, and female fetuses had lower levels compared to controls. In male, but not in female, fetal livers, miR-130 mediated the effects of GDM on PPARG expression. Another study reported stronger PPARG-mediated effects in male offspring. This was a rodent model in which a high fat maternal diet in pregnancy had stronger post-weaning effects on male vs. female weight gain (Lecoutre et al., 2016). Fetal male-specific effects were measured as mild glucose intolerance, hyperinsulinemia, and hypercorticosteronemia, and correlated with the diminished PPARG mRNA levels. Future studies can consider sex differences in PPARG-related biomarkers in both placental and fetal tissues to gain a more complete understanding.

PPARγ regulation may involve an epigenetic mechanism such as small non-coding RNA (micro-RNA identified in above study) or DNA methylation, which is impacted differentially by environmental factors and by XX and XY karyotype in the placenta (Buckberry, Bianco-Miotto and Roberts, 2014,Inkster et al., 2021; Martin et al., 2017). It might not be genetically intrinsic to PPARγ regulation. Maternal diet can alter PPARγ expression in the fetal liver by way of methylation of its promoter (Simmons, 2007; Suter et al., 2014; Lendvai et al., 2016). In a birth cohort study, cord blood levels of the organophosphate pesticide chlorpyrifos were studied in relation to DNA methylation levels, measured in placental tissue by ChIP-qPCR and pyrosequencing. Based on adjusted associations, they concluded that prenatal chlorpyrifos exposure was positively associated with PPARG DNA methylation, and that trimethylation of lysine 4 of H3 (H3K4me3 level in R3) was positively associated with 1 (gross motor) of 7 neurodevelopmental domains evaluated at 2 years of age (Chiu et al., 2021). This study offers weak evidence as the specific placental epigenetic biomarkers identified as important in the exposure-mediator, and mediator-outcome associations do not match up. The authors did not conduct a mediation analysis. PPARγ mRNA and/or protein expression was not measured.

Obesogen theory states that activation of PPARγ by environmental chemicals in pregnancy is causing the reprogramming of fetal adipocytes and is a causal factor in the obesity epidemic (Grün and Blumberg, 2007; Desvergne et al., 2009; Janesick and Blumberg, 2011; Egusquiza and Blumberg, 2020). This type of programming likely occurs in the nascent neuroendocrine system. The Obesogen theory has not been worked out with respect to a placental mechanism. It is assumed that the obesogenic chemicals are acting directly on fetal tissue. Alternatively, the placenta may mediate the effects of the maternal environment on obesity risk in the offspring. Maternal exposures that alter expression of placental glucose transporters are proposed as one example of this (Lewis et al., 2013). Placental PPARγ may be a crucial and yet unidentified factor in the Obesogen paradigm.

Recent studies point to novel transport functions for PPARγ that may be relevant to fetal development. In non-pregnant rats, PPARγ was a mediator of the effects of GenX (a replacement chemical for perfluorooctanoic acid that is used in Teflon, a surface coating of non-stick cookware) on the activity of the P-glycoprotein (P-gp) transporter protein in rodent brain capillaries at the blood-brain barrier (Cannon et al., 2020). The P-gp is a transporter protein, encoded by the ABCB1 gene, that is expressed in the liver and kidney where it transports harmful substances into bile and/or urine. P-gp can transport diverse substances including chemotherapeutics, lipids, steroids, bilirubin, bile acids, platelet-activating factor, dietary flavonoids, and conjugated endogenous and xenobiotic metabolites (Cannon et al., 2020). In this study, GenX had a negative effect on the transporter activities in all cell types tested (rat brain capillaries, human ovarian and breast cancer cell lines). Since the P-gp generally transports substances out of cells and into excretory pathways, this would potentially result in more direct toxicity to the fetal brain. Simultaneous treatment with a PPARγ antagonist blocked the negative effects of GenX on P-gp transporter activity in the rat brain capillaries specifically, resulting in lower toxicity to brain cells.

P-gp is expressed in the first trimester placenta and decreases in expression over time in pregnancy [reviewed in (Walker et al., 2017)]. Specific studies on joint expression of placental PPARγ and placental P-gp were not identified, nor were studies that evaluated parallel expression of placental and fetal PPARγ and PPARγ-regulation. Based on these 2 studies, placental PPARγ can theoretically regulate the expression and activity of the placental and fetal P-glycoprotein transporter protein, and modulate fetal exposures to compounds from maternal circulation, both beneficial and harmful.

Placental PPARs may also play a mediating role in the case of the effects of viral infections on fetal brain development. On the fetal side, PPARs mediated the infection of neural progenitor cells by the Zika virus and gene expression changes in those cells (Thulasi Raman, Latreille, Gao et al., 2020). Fetal PPARγ was proposed as a therapeutic target to protect the fetus in the context of fetal lung development (Ren et al., 2021). In that case, it was assumed that increasing PPARγ expression pharmacologically using rosiglitazone or leptin would increase production of lung surfactant in neonates born with respiratory distress syndrome due to inadequate production of surfactant protein (Ren et al., 2021). This targeted intervention did not prove effective in sheep, and instead reduced surfactant phospholipid and protein production (Ren et al., 2021). These 2 papers did not consider the role of placental PPARs in their etiologic or intervention models. Placental PPARs are likely mediating viral and rosiglitazone effects on fetal PPARs. These theoretical and empirical models can be extended to include placental PPARs in the pathway from exposure to fetal effects.

The effects of PPARγ upregulation and/or downregulation on fetal brain development is an active area of study, but so far investigators have not considered joint placental-fetal PPARγ expression (Guo et al., 2022). PPARγ regulation in the fetal brain is implicated in angiogenesis impairment, cognitive impairments and anxiety behaviors, hippocampal neuroinflammation, and neurodevelopmental and neurocognitive impairment (Guo et al., 2022). All these defects are believed to have long-term consequences on brain development. Guo et al. summarized this research to propose that pharmacologic manipulation of PPARγ in pregnancy may be one avenue to prevent the harmful effects to the fetus of exposures, known to disrupt PPARγ. Given the abundant expression of PPARγ in the placenta and in multiple fetal tissues, this could have unintended and adverse consequences.

3.4. Leptin

Leptin is an adipokine hormone that is primarily synthesized by fat cells (adipocytes), and by the placental trophoblast (Chardonnens et al., 1999; Masuzaki et al., 1997). It provides a signal to the hypothalamus that regulates hyperphagia (desire to eat) and swallowing. This occurs by way of neurons which express neuropeptide NPY and pro-opiomelanocortin (POMC) (Elias et al., 1999; Benite-Ribeiro et al., 2016). The fetus begins producing leptin as early as 6 weeks gestation at the beginning of embryonic lipidogenesis and adipocyte differentiation (Atanassova and Popova, 2000). The trophoblast and the placenta express leptin mRNA as early as 6 weeks.

Synthesis of placental leptin is linked mechanistically to placental hCG and PPARγ (Islami et al., 2003; Lappas et al., 2005; Maymo et al., 2009). In primary cultures of early pregnancy placental tissues, leptin stimulated hCG secretion. In late pregnancy tissues, the relationship reversed and hCG stimulated leptin secretion. Whereas PPARγ positively regulates hCG synthesis; it may negatively regulate leptin synthesis. Leptin dosing of placental explants stimulated the release of cytokines and inflammatory markers. This effect was blocked by troglitazone, an agonist of PPARγ (Lappas et al., 2005).

Leptin in pregnancy comes from maternal, fetal and placental cells and is most commonly measured as a circulating plasma biomarker. Most of the leptin synthesized by the placenta is transported into the maternal circulation vs. 5% that is transported into the fetal circulation. This was determined by placental perfusion studies of term placentas (Lepercq et al., 2001). Fetal plasma leptin levels are correlated with fetal fat mass which also suggests that the fetus is producing its own supply, and not dependent on placental levels (Alexe et al., 2006). Levels of free leptin in maternal blood rise slightly in pregnancy over the 3 trimesters with a surge in late pregnancy (Lewandowski, Horn, O’Callaghan et al., 1999; Andersson-Hall et al., 2020). The late pregnancy surge is assumed to be caused by the onset of fetal stomach production of leptin in the rodent (Yau-Qiu et al., 2020). If soluble leptin concentrations are adjusted for levels of the circulating receptor, leptin levels actually decrease from the first to third trimester (Lewandowski et al., 1999; Andersson-Hall et al., 2020). Leptin production and distribution are regulated to control quantities available to fetal and maternal tissues. Leptin and adiponectin levels are inversely correlated in pregnancy and are often measured together. The ratio of leptin to adiponectin is a sensitive measure of their joint regulation (Andersson-Hall et al., 2020; Makker et al., 2023).

Placental leptin can serve as an indicator of poor placental development and function. Leptin levels as early as 8 weeks were elevated in cases of reduced vascular development (Leijnse et al., 2018). Leptin levels are generally elevated in high-risk situations, such as pregnancies complicated by preeclampsia and growth restriction (Herse et al., 2009; Tzschoppe et al., 2010). Plasma leptin and placental LEP mRNA were elevated in preeclampsia vs. controls, (Herse et al., 2009; Taylor et al., 2015; Laivuori et al., 2000). Preeclampsia is associated with inflammation, immune tolerance, shallow invasion, poor vascularization, which means placental leptin may play a role in any of the above processes. Both hyperleptinemia (too much Leptin) and hypoleptinemia (too little Leptin) in pregnancy are associated with poor outcome which does not help to narrow the scope of possible functions (Kratzsch et al., 2000). In these studies, it is not clear if abnormal leptin levels are a cause or a consequence of abnormal placental development and function. For all these reasons, studies that can link circulating leptin to LEP mRNA and/or protein in placental tissue, and which expand joint measurement of the molecules that are associated with leptin in the placenta, can increase insight into the unique functions of leptin in the placenta.

Other placental functions and malfunctions are associated with leptin. Placental leptin, also characterized as a cytokine, is studied as part of the inflammatory response (González et al., 2000). Elevated amniotic fluid levels of leptin were present with other inflammatory markers, and higher in the presenting twin (the twin who is lower in the birth canal at the end of pregnancy) (Lee et al., 2015). Placental regulation of glucose homeostasis is another possible function (Benzler et al., 2013; Hajagos-Tóth et al., 2017). In a meta-analysis of over 1000 studies, leptin levels in the first and early second trimester were 7.25 ng/ml higher on average in women with GDM vs. women without GDM (Bao et al., 2015). In a longitudinal study, the odds of GDM increased by 3–4 fold in women with high leptin vs. low leptin (Francis et al., 2020. In this same study the soluble receptor of leptin, was lower in GDM subjects early in pregnancy vs. non-GDM subjects (Francis et al., 2020). Placental leptin mechanisms in these associations are not understood (Schanton et al., 2017).

Leptin is widely pursued as an informative biomarker to evaluate the developmental origins of health and disease (DOHaD) hypothesis which states that environmental influences on changes in molecular pathways during critical periods of fetal development can have effects on long-term chronic disease risks (Barker, 1995; Adibi et al., 2021). From 2006 to 2019, there were at least 24 review papers outlining leptin as a target of maternal under or over nutrition in pregnancy, and leptin-related mechanisms of fetal programming of child and adult health (Alexe et al., 2006; Kratzsch et al., 2000; Henson and Castracane, 2006; McMillen et al., 2006; Hanson and Gluckman, 2008; Mühlhäusler, Adam and McMillen, 2008; Mostyn and Symonds, 2009; Svechnikov et al., 2010; Tamashiro and Moran, 2010; Muhlhausler and Smith, 2009; Vickers and Sloboda, 2012; Bourguignon et al., 2013; Breton, 2013; O’Keeffe and Kenny, 2014; Briffa et al., 2015; Dearden and Ozanne, 2015; Lecoutre and Breton, 2015; Pérez-Pérez et al., 2015; Sullivan et al., 2015; Edlow, 2017; Chatmethakul and Roghair, 2019; Wróblewski et al., 2019; Stolzenbach et al., 2020; Lukaszewski et al., 2013).

A key measure in this paradigm is umbilical cord leptin. Cord leptin is reflective of fetal, placental, and maternal production depending if the sample is taken from the umbilical artery (fetal, placental) vs. the umbilical vein (maternal) (Yang et al., 2002). Umbilical leptin levels are positively associated with maternal body mass index (Jaramillo et al., 2021), birthweight (Sanli et al., 2021), neonatal adiposity (Bagias et al., 2021). Evidence is inconsistent on associations of umbilical cord leptin and child fat mass. Summarized in a review of literature over 26 years, cord blood leptin was inversely associated with child fat mass up to 3 years of age and null at 5 years (Bagias et al., 2021). In a recent study, there was no association up to 5 years of age (Ashley-Martin et al., 2020). Umbilical cord blood leptin was not associated with measures of appetite and overeating at 7 years of age (Warkentin et al., 2020). There is not a clear picture on the degree to which placental/cord blood leptin causes and/or mediates fetal programming of child and adult health.

Leptin plays a role in fetal organogenesis, much of which occurs in the first trimester. Children (and their placentas) that suffer from a rare defect in the leptin gene develop morbid obesity in childhood, do not undergo puberty, and have abnormal expression of growth and thyroid hormone (Chan and Mantzoros, 2005). Hence, leptin may have global influence early in development by way of the neuroendocrine axis. Leptin receptors are expressed in the human placenta, brain, heart, and pancreas (Briffa et al., 2015), which may indicate where leptin is required for proper fetal development. Leptin in the first trimester is associated with formation of the adipose tissue (Mostyn and Symonds, 2009), yolk sac hematopoiesis (Bennett et al., 1996; Mikhail et al., 1997), brain development (Sullivan et al., 2015; Valleau and Sullivan, 2014), lung development (Ip et al., 2020, bone development (Javaid et al., 2005; Yarbrough et al., 2000), Leydig cell differentiation (Li et al., 2019). In cases of overgrowth of the baby, leptin levels are elevated (Dong et al., 2018; Ozdemir and Aksit, 2020). In cases of growth restriction, leptin levels are reduced (Yalinbas et al., 2019; Ben, Qin, Wu et al., 2001; Ren and Shen, 2010; Koistinen et al., 1997).

Fetal sex is a putative modifier of placental leptin effects in pregnancy. Circulating leptin levels are generally higher in women carrying female fetuses (Helland et al., 1998; Yang et al., 2002; Cagnacci et al., 2006). Most studies reviewed report that the effects of adipokines in pregnancy on child outcomes differed by sex. A few examples are cited here (Jaramillo et al., 2021; Ashley-Martin et al., 2020; Lima et al., 2020).

Epigenetic regulation of leptin may occur by DNA methylation or by microRNAs, or both. Placental DNA methylation of the leptin gene is associated with circulating levels of leptin (Hogg et al., 2013), and also associated with infant neurobehavior and brain size (Kennedy et al., 2020; Lesseur et al., 2013; Bekkering et al., 2019). Placental DNA methylation was associated with newborn secretion of cortisol over a time course while taking a neurobehavioral assessment within the first week of life (Reid et al., 2023). The direction of the association was consistent with the hypothesis that placental methylation of the LEP gene is associated with lower circulating leptin, and leptin is in a negative feedback loop with cortisol production by the infant adrenal. The placental microRNA mir-532 was identified in 2 birth cohorts that may regulate placental leptin and its effects on birthweight (Kennedy et al., 2020). Studying circulating leptin jointly with placental epigenetic biomarkers is one approach to increase accuracy and specificity in distinguishing placental from maternal leptin.

Placental leptin can be measured as an umbilical cord concentration or as the placental LEP mRNA. Leptin is often measured in tandem with adiponectin, an adipokine produced by adipose tissue. Leptin and adiponectin often show opposite associations with common factors in pregnancy, suggesting that they are jointly regulated (Makker et al., 2023; Makker et al., 2022). One study reported a negative association of methylation of the promoter of the adiponectin gene (ADIPOQ) and cord blood adiponectin suggesting that the placenta might be a source of synthesis or epigenetic regulation of both adipokines (Yang et al., 2022). Maternal undernutrition and overnutrition and maternal adiposity are the strongest exposures that impact leptin. These types of exposure effects can be mimicked by endocrine disrupting chemicals (Bourguignon et al., 2013; Walley and Roepke, 2018).

Leptin regulation in pregnancy has ancillary but potentially consequential effects on development of bone, reproductive tract, lung, and aspects of brain function. Placental leptin is widely measured in longitudinal birth cohort studies. With time, we will learn more on how strongly it can predict long-term offspring health either as a primary exposure or as a mediator (Gagne-Ouellet et al., 2020). Given the vast accumulation of evidence on the importance of leptin, it is being pursued as an intermediary outcome within intervention studies to change maternal patterns of exercise and diet in pregnancy (Daniels et al., 2020; Ferrari et al., 2020; Harreiter et al., 2019) and assess their impacts on fetal programming.

3.5. Transforming growth factor beta

TGFβ is a regulator of embryogenesis and hormonal regulation in the placenta (Krieglstein et al., 2002; Unsicker and Krieglstein, 2002; Wu and Hill, 2009). It is not specific to the placenta and is expressed widely across multiple tissue types in the adult (Fagerberg et al., 2014; Uhlen et al., 2015; Sjostedt et al., 2018). It is well studied as a cytokine that regulates immune function and tumorigenesis, and less well studied as a placental protein. TGFβ plays a critical role in placental hormone synthesis and secretion (Jones et al., 2006; Petraglia et al., 1996). TGFβ is localized to the decidua and to the syncytiotrophoblast, the extracelluar matrix, and the villous core (Graham et al., 1992). It is expressed in 3 forms: TGFβ1, TGFβ2, and TGFβ3, all with slightly different autocrine and paracrine functions. TGFβ1 inhibited cell proliferation in first trimester villous placenta explants (Graham et al., 1992). Within the villous tissue, TGFβ1 decreases hormone synthesis and promotes fusion of the cytotrophoblasts (mono-nucleated) to form syncytiotrophoblasts (multi-nucleated) (Jones et al., 2006). The different forms of TGFβ play different roles, depending on cell type and tissue type and time since conception (Jones et al., 2006; Chuva de Sousa Lopes et al., 2020). General functions within the placenta are to promote or inhibit cell proliferation and trophoblast invasion (Jones et al., 2006), and to regulate immune reactions at the placental-maternal interface (Chuva de Sousa Lopes et al., 2020; Dekker and Robillard, 2005).

TGFβ has value as a placental biomarker that may reflect placental function and placental contributions to fetal development. In one study, TGFβ was associated with the risk of preterm birth when measured earlier in pregnancy (Gargano et al., 2008), and not associated when measured later in another study (Matoba et al., 2009). TGFβ is essential to aspects of fetal development, namely development of the cerebral cortex (Colakoglu and Kukner, 2004; Kathuria et al., 2022), the fetal ovary (Hartanti et al., 2020; Hatzirodos et al., 2011; Liu et al., 2022; Azumah et al., 2022), and male reproductive function (hypothalamic-pituitary-gonad axis) in the mouse (Ingman and Robertson, 2007). TGFβ levels in cord blood were higher in babies born with congenital heart defects (Teratology of Fallot), and also associated with severity of the heart defect (Gomez et al., 2023). However, in all these studies the assumption is that fetal TGFβ is causal in the fetal defects. The studies did not account for the placental supply of TGFβ and its transport and use within fetal tissue. One study was identified that measured a placental microRNA (miR-876) in the TGFβ pathway which was associated with child growth trajectory from 0 to 5 years (Kennedy et al., 2022).

3.6. Epiregulin

Epiregulin (EREG). has been localized in the placenta (Toyoda et al., 1997); yet has primarily been studied in non-placental tissues as a member of the epidermal growth factor family and a ligand of the epidermal growth factor receptor (EGFR). In the adult ovary, epiregulin mediates the gonadotropin (i.e. luteinizing hormone) effect on cellular processes during ovulation such as inflammation and wound healing (Ashkenazi et al., 2005; Park et al., 2004). For this reason, we hypothesized it may also mediate the effects of hCG, an analogous gonadotropin, in the trophoblast. As outlined above, PPARγ is a placental molecule that is upstream of hCG. Conversely, epiregulin is potentially an informative molecule that is downstream of hCG.

In the placenta, there are autocrine functions within the trophoblast that allow for specialized signaling in a low oxygen (~2%) environment in the first trimester. These signals are essential for cell proliferation and for survival by way of blocking apoptosis (Genbacev et al., 1997; Caniggia et al., 2000a,b; Armant et al., 2006). In an experimental model that compared protein expression in the first trimester trophoblast at 2% O₂ vs. 20% O₂, heparin-binding EGF-like growth factor (HBEGF) and EREGwere 2 out of 6 EGF family members that were upregulated by low O₂. Furthermore, this response was shown to be essential in preventing apoptosis (Armant et al., 2006). An abnormal response to hypoxia, involving lower expression of HBEGF and EREG, is believed to be causally related to the risk of preeclampsia (Armant et al., 2006; Armant et al., 2015). In a follow-up study in which EREG was measured in placental tissue biopsies at term, EREG was highest in placentas associated with small for gestational age (SGA) infants, and predominantly expressed in the extravillous trophoblasts within the basal plate (Armant et al., 2015).

We postulate that EREG would be associated with placental hCG and with placental TGFβ that is a co-activator of the EGFR. To date, there is little written about epiregulin as a placental biomarker. There were no papers identified with the search terms used for this review, specifically related to placental-fetal reproductive tract and brain development. EREG may be a biomarker of risk of miscarriage in early pregnancy given its role in implantation and in the epithelial to mesenchymal transition of the trophoblast to its invasive subphenotype called the extravillous trophoblast (Cui et al., 2019; Barnea et al., 2012).

3.7. Growth differentiation factor (GDF15)

GDF15 is a member of the TGF-Beta family. It is also referred to as the macrophage inhibitory cytokine-1 (MIC-1), placental transforming growth factor-β (PTGF-β), placental bone morphogenic protein (PLAB), NSAID-activated gene-1 (NAG-1). GDF15 expression was highest in the placenta compared to all other tissues, as mRNA and protein (HumProtAtlas). In an in vitro model of the trophoblast, GDF15 serves as a secreted biomarker, along with hCG, to confirm trophoblast survival and function (Turco et al., 2018). Its concentration in the conditioned media rises over time in culture (Moore et al., 2000). GDF15 rises in maternal serum over the course of pregnancy, similar to hCGα, and is at its maximum level in amniotic fluid (Moore et al., 2000).

The function of placental GDF15 may be to provide a pregnancy-specific signal to the mother. GDF15 is expressed in large quantities and exclusively by the placenta during pregnancy (it is 10-fold higher than in the non-gravid state)(Welsh et al., 2022). The idea was put forth that GDF15 is not expressed constitutively, and when it is expressed, it acts primarily as a stress signal (Patel et al., 2019). This theory is supported by the evidence in different species, in different tissue types, and in different physiologic and pathologic conditions (Lockhart, Saudek and O’Rahilly, 2020). GDF15 levels in maternal cerebral spinal fluid (CSF) were correlated with GDF levels in serum in pregnant individuals, and not in non-pregnant individuals (Andersson-Hall et al., 2021). The serum and CSF GDF15 levels were correlated at the time of birth, but not 5 years later. Both serum and CSF GDF15 were higher in women carrying females vs. males, suggesting that the placenta was the primary source and karyotypic sex was a source of variation (Andersson-Hall et al., 2021). In pregnancy, GDF15 is associated with severe nausea or hyperemesis gravidarum (Petry et al., 2018). Lockhart and colleagues propose a theory that GDF15 signals to the maternal central nervous system to influence appetite, food preferences, and food intake, beginning early in the first trimester. The evolutionary rationale, on behalf of the placenta-fetus, is to direct her to eat less and eat less aggravating foods, and by doing so avoid teratogenic substances and protect the fetus during the most sensitive timepoints of organogenesis.

Relevant to the idea of GDF15 as a signal between the placenta and the maternal brain, we reported that GDF15 was positively associated with hCG and free thyroxine (FT4) in second trimester maternal serum (Adibi et al., 2021b), and most strongly with hCGα. hCGα is also produced by the maternal pituitary (Uhlen et al., 2015; Korhonen et al., 1997) and FT4 is primarily produced by the maternal thyroid with targets in the maternal and fetal brain (Glinoer, 1999). The GDF15-hCGα association differed in direction and magnitude between Black and White women. This may indicate that GDF15 plays a role in the response to racism, stress, adversity, and/or nutrition that may also differ between Black and White women. This is consistent with the theory behind the allostatic load (McEwen, 2004), a panel of physiologic biomarkers that are characterized as being informative of the specific types of chronic stress experienced by marginalized groups in the US (Hux and Roberts, 2014; Dyke et al., 2020).

GDF15 may have a variety of functions in the fetal, maternal, and non-pregnant brain. It has also been examined as a potential therapeutic target in treating dementia (McGrath et al., 2020) which is pointed out here as evidence of its specific and potent effects on brain function. In the first discovery of this molecule, it was proposed that it may mediate placental effects on fetal brain development (Hromas et al., 1997). In a preliminary study, GDF15 and infant cognition were not associated (Adibi et al., 2021b). In sum, GDF15 may be an important and novel biomarker to study in the first trimester to understand endocrine disruption and its effects on maternal brain health.

GDF15 is expressed in decidual tissue and relates to endometrial as well as placental function. Levels were lower in pregnancies that ended in miscarriage (Tong et al., 2004; Kaitu’u-Lino et al., 2013), and they were higher in preeclamptic and gestational diabetes mellitus (GDM) pregnancies versus normal pregnancies (Sugulle et al., 2009; Jacobsen et al., 2022; Jacobsen, Roysland, Strand et al., 2022, 2022). In another study, levels were lower in preeclamptics vs. controls (Chen et al., 2016). The first study measured GDF15 in plasma and in placental tissue, and the second study measured it in serum. GDF15 dysfunction in paternal-maternal-placental immune reactions may contribute to risk of implantation, miscarriage, and preeclampsia (Wischhusen et al., 2020). We do not currently have a theory on how placental GDF15 relates to fetal reproductive tract or brain development. It is a promising biomarker in the context of maternal neuroendocrine - placental function.

3.8. Small nucleolar RNA (snoRNA)

SnoRNAs are a class of highly expressed small, non-coding RNAs which are classified either as H/ACA box or C/D box snoRNAs (SNORDs) (Tyc and Steitz, 1989; Maxwell and Fournier, 1995). SNORDs are candidate biomarkers in the investigation of endocrine disruption in the first trimester placenta, as the SNORD genes have consistently been at the top of the list of differentially expressed genes in 3 transcriptome analyses done by our group of isolated cytotrophoblasts and bulk placental tissues sampled in the first trimester and at term (unpublished). The best understood overall function of SNORDs is to provide a scaffold for post-transcriptional modification of rRNA by an antisense mechanism (Watkins and Bohnsack, 2012). Interestingly, a subset of SNORDs lack any classical antisense elements, many of which are only found in placental mammals (Barlow and Bartolomei, 2014; Cavaille, 2017). Particularly, the altered level of these untypical SNORDS may be a primary cause of Prader-Willi syndrome (PWS)(Nicholls et al., 1989). PWS occurs in fetuses who harbor deletion of a cluster of paternally-imprinted genes on Chromosome 15. Two of the transcripts corresponding to the deleted chromosome fall into the category of SNORDs: SNORD115 and SNORD116. The deletion of paternal SNORD116 is strongly related to aspects of the PWS phenotype, namely: postnatal growth retardation and hyperphagia (Costa et al., 2019; Buiting et al., 1992,de los Santos et al., 2000; Cassidy et al., 2012). The ways in which this happens are not well understood. SNORD115 regulates the RNA editing and the alternative splicing of the serotonin receptor 2C (HTR2C) (Kishore and Stamm, 2006).

In our transcriptome analysis of isolated cytotrophoblasts and bulk placental tissues, conditions that were associated with differential expression of SNORDs were phthalate dose (experimental), phthalate exposure (observational), circulating hCG levels, fetal sex, and tissue type differences (chorion smooth vs. villous tissue) (unpublished). The PWS phenotype resembles an extreme version of the disorders associated with fetal endocrine disruption: hypogonadism, pubertal insufficiency, childhood obesity, mild intellectual disability, and behavioral abnormalities (Cassidy et al., 2012). For all these reasons, placental snoRNAs might be part of an epigenetic mechanism in the placenta that is relevant to placental-fetal development and endocrine disruption.

Placental snoRNAs may also not play a role in fetal development, and instead may regulate general placental function. Multiple forms of SNORD114 and SNORD113 were upregulated in syncytiotrophoblasts and downregulated in cytotrophoblasts sampled from preeclamptic vs. gestational age-matched placentas (Gormley et al., 2017). Cajal bodies are subnuclear organelles where snoRNA activity takes place (Gormley et al., 2017; Meier, 2017). As functional confirmation of this finding, authors visualized the Cajal bodies in the preeclamptic tissue by immunolocalization of the coilin protein. They observed 6-fold more Cajal bodies in the isolated cytotrophoblasts of the preeclamptic placentas versus non-preeclamptic placentas. It is not clear from this study if the increase in Cajal bodies and differential gene expression were cause or consequence of preeclampsia. snoRNAs have been further shown to play an important role in trophoblast cell migration and proliferation (Hernandez Mora, Sanchez-Delgado, Petazzi et al., 2018; Yang et al., 2019; Zhou et al., 2020).

Canonical snoRNAs can be measured in different forms based on structural properties. The SNORD genes guide the methylation of ribose rings. The SNORA genes are involved in a RNA editing (Bratkovič and Rogelj, 2014). The emerging roles for snoRNAs include regulation of mRNA abundance and alternative splicing (Bratkovic, Bozic and Rogelj, 2020). In the transcriptome analyses, both SNORD and SNORA transcripts were differentially expressed in placental tissue by endocrine disruptors and by fetal sex. Multiple new technologies have been developed to study snoRNA:mRNA and snoRNA:protein binding (Bratkovic et al., 2020; Ramanathan et al., 2019). There is not yet a clear understanding of ways to confirm snoRNA-specific effects by measuring corresponding mRNAs or proteins. Adverse prenatal maternal environment, generated experimentally in pregnant mice, altered the expression of the hippocampal serotonin receptor 2c (HTR2c) and the small nucleolar RNA MBII-52 in the brains of offspring at postnatal day 21(Chen et al., 2021). This effect was present in the male and not the female hippocampi. It strongly suggests that SNORD115 is an epigenetic target of maternal stress and regulates brain development, and is in the Serotonin pathway. This study did not consider placental expression of SNORD115.

3.9. Serotonin

Serotonin (5-HT) is monoamine neuropeptide that functions as a neurotransmitter. During the fetal period, serotonin contributes to cortical development by way of stimulating cell division, neuronal migration, cell differentiation, and synaptogenesis in fetal cells (Bonnin et al., 2007; Lonstein, 2019,Yang et al., 2014). Historically, the embryo/fetus was considered the primary source of serotonin required for proper brain development, as early as 7–8 weeks gestation (Sundstrom et al., 1993). Several enzymes control the synthesis and metabolism of serotonin (Karahoda et al., 2020). Tryptophan, an essential amino acid, is the precursor of serotonin. The enzymes tryptophan hydroxylase (TPH1 and 2) and amino acid decarboxylase (encoded by DDC gene) convert L-tryptophan to 5-HT. Monoamine oxidase A (MAO) catabolizes 5-HT to its inactivated form, 5-hydroxyindoleacetic acid (5-HIAA) (Rosenfeld, 2020).

In 2011, a paradigm shift was put forth whereby it was shown experimentally in the mouse that the placenta was a primary source of serotonin (Bonnin et al., 2011). They inferred that this was true in humans as well. Authors reported that they cultured a single human placenta sampled at 11 weeks gestation, and observed that it could be stimulated with a cofactor for TPH1/2 activity to secrete 5-HT and 5-HTP (Bonnin et al., 2011). In this study, they relied on a single sample (no biologic replication). The methods of dissection and culture were not provided (Bonnin et al., 2011). In a paper from another group with biologic replication and more detail on methods, it was demonstrated that isolated trophoblasts from the first trimester (10–12 weeks) and from term placentas (39–41 weeks) could be given tryptophan, along with a monoamine oxidase inhibitor, to stimulate secretion of intracellular and extracellular serotonin at 150–200% that of the control cells (Laurent et al., 2017). Additionally, they showed strong evidence that TPH1 and TPH2 were detectable at the mRNA and protein levels in primary placental tissue and in 4 different trophoblast cell lines.

Evidence also argues against the production of serotonin by the placenta. In the Human Protein Atlas, TPH1 was detected as protein by immunohistochemistry in two term placentas but was not detected as mRNA (Uhlen et al., 2015). TPH2 was not detected as mRNA or protein in term placentas (Uhlen et al., 2015). It is not clear from these studies if the protein or mRNA encoded by the DDC gene was detectable in the human placenta. DDC is described as a necessary enzyme (amino acid decarboxylase) in the placental conversion of 5-HTP or 5-HT (Bonnin and Levitt, 2011). Another placental cell biology research group presented an alternative model. Through extensive immunohistochemical staining of placentas of all gestational ages, they contest that the placenta is not synthesizing appreciable quantities of serotonin. Instead the placenta is transporting serotonin from maternal platelets into fetal circulation, through a more complex transport system than previously described (Kliman et al., 2018). Authors do not exclude the possibility of some placental production in their model but they suspect it is a minor source compared to the maternal platelets.

These conflicts were partially resolved when gestational age differences in the serotonin synthetic machinery were analyzed. In a study of 13 first trimester and 32 term placental tissue samples, investigators measured 16 enzymes and 5 transporters involved in tryptophan metabolism and transport and serotonin synthesis and transport (Karahoda et al., 2020). The main differences they reported are in the higher expression of TPH1 and SLC6A4 mRNA in the first trimester vs. term and the higher expression of indoleamine 2,3-dioxygenase-1 (IDO) as mRNA, protein and in its enzymatic activity at term vs. the first trimester. From this, they infer that serotonin production only occurs in a limited fashion in the first trimester placenta. Later in pregnancy, the machinery switches to either transport or metabolism of serotonin. The debate as to whether the human placenta is synthesizing serotonin denovo vs. transporting and/or metabolizing maternal serotonin is still not fully resolved (Peric et al., 2022). The 2022 review by Peric et al. outlines latest findings in the characterization of the placental serotonin system, namely differences between early and late pregnancy, cell types, between mRNA and protein findings, and between primary and single cell placental cultures.

Specific membrane receptors and transporters have been identified that are involved in the 5-HT signaling pathway in the placenta: 5-HT transporter (SERT/SLC6A4) in the apical membrane of the villous syncytiotrophoblast facing the intervillous space, and OCT3 (SLC22A3) in the fetus-facing basal membrane of syncytiotrophoblast (Prasad et al., 1996; Sata et al., 2005). By binding to trophoblast receptors, 5-HT(1A)R and 5-HT(2A)R (both encoded by the gene HTR2A), 5-HT extracellular levels are tightly controlled. Binding to these receptors also stimulates localized actions, some of which may be deleterious to the placenta (Rosenfeld, 2020). Upon binding to SERT, 5-HT is transferred into the trophoblast and taken out of the extracellular space. There can be competition for binding to the SERT receptor between endogenous serotonin and Selective Serotonin Reuptake Inhibitor (SSRIs), a class of pharmaceuticals given to pregnant women with depression (Hudon Thibeault et al., 2019; Borue et al., 2007; Davidson et al., 2009).

Placental serotonin provides a distal signal to stimulate embryonic/fetal brain development; and it is also considered to be operating locally as an essential autocrine and paracrine signal. The binding of 5-HT to SERT and the catabolism of 5-HT by the monoamine oxidase (MAO-A) are both mechanisms to prevent harmful build-up of 5-HT concentrations in the extracellular spaces that can cause the vasoconstriction of placental-maternal blood vessels (Rosenfeld, 2020). This has been most well studied in the pathophysiology of preeclampsia (Bertrand and St-Louis, 1999; Bottalico et al., 2004; Ugun-Klusek et al., 2011; Carrasco et al., 2000; Gujrati et al., 1996; Sivasubramaniam et al., 2002). MAO-A is less specific than the other molecules to serotonin synthesis as it is expressed widely in the body and plays a role in the deamination of multiple amines. The first likely evidence of placental serotonin without identifying it specifically as placental showed that very early increases in serotonin levels, measured as maternal urinary hydroxyindole acetic acid (5-HIAA), or changes in the metabolism of serotonin were associated with higher risk of spontaneous miscarriage. They confirmed this finding by treating pregnant rats with a serotonin antagonist and observing lower numbers of abortions, consistent with an association and treatment effect measured in human pregnancy (Sadovsky et al., 1972). Given recent findings, we can now assume that these effects on miscarriage risk were reflective of placental serotonin production. Serotonin measured in cord blood was positively associated in a linear shape with fetal growth (Zhai et al., 2023), and in a mouse model intrauterine growth restriction resulted in reduced levels of 5-HT in the fetal brain (Ye et al., 2021). It was proposed that urinary levels of 5-HIAA be monitored in pregnancy as indicators of risk of adverse obstetric outcome (Sadovsky et al., 1972; Lorenzo-Almoros et al., 2019).

Placental serotonin offers a unique opportunity to empirically evaluate the DoHAD theory because, for example, it offers a powerful way to simultaneously measure the effects of maternal exposures (diet, stress, endocrine disrupting chemicals) and future brain function in the offspring. To establish this relationship, it is ideal to measure and analyze variables along the full pathway of maternal exposure, molecules in placental serotonin production, transport and/or metabolism, and fetal brain-related outcome. This was accomplished in a paper that used an experimental mouse model and human fetal cross-sectional measures (Ceasrine et al., 2022). In the mouse model, they established that a high fat diet caused placental inflammation, measured by endotoxin levels. Placental inflammation caused increased microglial phagocytosis of 5-HT neurons and a decrease in 5-HT levels in fetal and adult brains. In the human model, they measured triglyceride accumulation in the placental tissue as a proxy for maternal diet. They measured levels of 5-HT levels in aborted fetal brain tissue. They established that there was an inverse association such that higher placental trigylceride levels were associated with lower fetal brain 5-HT levels. In both models, they found that the effects were specific to the male. Another study also evaluated the full pathway. Serotonin was assessed as the methylation of the SLC6A4 (serotonin transporter gene) in maternal and infant buccal cells collected at birth, as opposed to placental cells. Self-reported maternal stress related to conditions of the COVID pandemic was associated with higher SLC6A4 methylation (Provenzi et al., 2021). They measured methylation at 13 CpG sites on the SLC6A4 promoter. Seven of the sites, identified in a principal components analysis, were positively associated with maternal stress. These same biomarkers were negatively associated with infant temperament measured at 3 months, and specifically with a measure of surgency (distress, sadness, fear). It was notable that only infant, and not maternal, SLC6A4 methylation was associated with stress and with infant outcomes. Infant buccal cell methylation levels are likely more correlated with placental levels given common genotype and epigenome. In two review papers, Rosenfeld, outlines key components by which placental serotonin can be plugged into the DoHAD theory (Rosenfeld, 2020, 2021).

Both the under and over production of serotonin by the placenta can have negative effects on brain development. Both conditions are hypothesized as associated with autistic spectrum disorder (ASD) etiology (Yang et al., 2014; Sato, 2013). Maternal exposures that are well studied as affecting serotonin levels (not necessarily placental) are alcohol consumption in pregnancy (Weinberg et al., 2008; Lokhorst and Druse, 1993; Kelly and Dillingham, 1994; Riley et al., 2001), cocaine (Meyer et al., 1993), maternal stress (Kofman, 2002; van den Hove et al., 2011) and maternal diet (Koski et al., 1993; Lukas and Campbell, 2000). The pharmaceutical teratogens valproic acid and thalidomide caused increased serotonin levels in the hippocampus in the mouse (Narita et al., 2002). The child outcome most well studied according to this paradigm of placental serotonin is ASD, reviewed previously (Yang et al., 2014; Rosenfeld, 2020, 2021; Sato, 2013).

Diverse types of maternal exposures have been associated with biomarkers in the placental serotonin system. Methylation status of SLC6A4 DNA in cord blood has been pursued as a biomarker of maternal stress in utero (Dukal et al., 2015). It was not associated with maternal stress in 90 newborns overall, but mean levels were higher in female vs. male newborns. The number of studies that have identified maternal exposures and conditions with effects on placental serotonin increased considerably from 2021 to 2023. Maternal diet is a major source of differences in placental serotonin because tryptophan comes from the diet and is a precursor to serotonin synthesis. Placental tryptophan metabolism was associated with placental serotonin, with placental inflammation, and risk of preterm birth (Karahoda et al., 2021). An experimental study in mice did not detect effects of air pollution or a high fat diet on fetal brain serotonin levels (Ye et al., 2021). Maternal exposure to green space (measured by mapping residential address to a Land use map in Flanders, Belgium and quantifying the amount of vegetation and undeveloped land within close proximity) was associated with methylation of the placental serotonin receptor gene (HTR2A) in a dose dependent manner (Dockx et al., 2022). The effects of maternal exposures on the serotonin system are sex dependent. Tobacco use was negatively associated with the methylation of the HTR2A gene, but only in male placentas (Horvaticek et al., 2022). A 2021 review paper summarizes evidence of environmental toxins that have effects on the serotonin system (Sarrouilhe et al., 2021). Maternal factors such as BMI and age are also associated with different molecules within the placental serotonin system and according to fetal sex. Serotonin, measured in maternal blood in a nontargeted metabolomics analysis was positively associated with maternal BMI. Maternal age was negatively associated with the placental protein form of MAOA. In the same study, there was a positive association of maternal age and the placental protein form of the serotonin transporter (SERT) (Levitan et al., 2022). Methylation of the gene HTR2A was higher in male vs. female placentas, and associations of methylation level with maternal BMI, GDM status and tobacco use differed by sex of the fetus (Horvaticek et al., 2022).

We did not detect an association of circulating second trimester serotonin and infant cognition at 1 and 3 years, working against this theory (Adibi et al., 2021b). It may also support the idea that only extremes in prenatal serotonin levels (hyper- and hyposerotonemia) are associated with child brain development. The lack of association could also be because serotonin was measured in maternal and not in fetal circulation. The question of a causal relationship between maternal exposure, placental serotonin and child neurodevelopment is an active area of translational studies using biomarkers of placental serotonin synthesis and transport.

Strategies for measurement of serotonin biomarkers in human pregnancy can include 5-HT, the end product, as well as the synthetic precursors and/or the enzymes involved. Biomarkers reflective of the synthetic pathway and its regulation can be measured in maternal circulation (blood, urine) or in placental tissue. There are caveats in the measurement of circulating serotonin as it binds platelets (Mercado and Kilic, 2010). Samples need to be treated specially at the time of collection. Serotonin can be measured by antibody-based methods (Adibi et al., 2021a), by mass spectrometry (de Jong et al., 2010), and by HPLC (Lesniak et al., 2013). The HPLC method can measure serotonin, tryptophan and its metabolites (Lesniak et al., 2013). These variables can be modeled as relative ratios. Another strategy is to measure circulating plasma tryptophan in neonates as a proxy for brain and/or placental levels of serotonin (Manjarrez et al., 1998). Serotonin pathways intersect with other molecular pathways of interest that likely also reflect maternal-placental communication, namely TRH-TSH and estrogen synthesis and regulation (Hudon Thibeault et al., 2019; Borue et al., 2007; Davidson et al., 2009; Shallie and Naicker, 2019).

Serotonin and hCG have not been widely studied together to our knowledge, except for a study that did not detect an association of CGB and MAOA mRNA in a set of placentas collected at delivery (Blakeley et al., 2013). In the second trimester, we did not detect associations of circulating serotonin and hCG. Serotonin was elevated in the plasma of pregnant women with hyperemesis gravidarum (nausea) which is a condition that is associated with elevated levels of hCG and GDF15 (Lockhart et al., 2020; Petry et al., 2018; Cengiz et al., 2015). Another potential link between serotonin and hCG was made in a study in which pregnant Wistar rats were dosed with testosterone (Erdogan et al., 2023). They measured effects on the increased production of the serotonin metabolite 5-HIAA in the fetal brain, and also increased autistic traits; along with decreased oxytocin levels. If placental hCG regulates fetal testosterone production in the human (described above), and testosterone regulates serotonin synthesis in the placenta and/or the fetal brain, it is plausible that placental hCG levels can affect aspects of fetal brain development through the testosterone-serotonin pathway.

3.10. Vitamin D

Similar to serotonin, Vitamin D concentrations in pregnancy may be both maternal and placental in origin. Maternal exposures that determine her levels are diet, supplementation and UV radiation (Liu and Hewison, 2012). Endogenous Vitamin D is produced primarily by maternal cells. Synthesis is regulated in part by the placenta (Liu and Hewison, 2012). The synthetic pathway is complex and can be studied as enzymes and as precursor molecules. D3 is supplied via the diet (from fortified dairy products and fish oils) or is synthesized in the skin from 7-dehydrocholesterol by ultraviolet irradiation. D2 (ergocalciferol) is produced in the liver (Bikle, 2014). D2 and D3 are subject to enzymatic reactions to make the hormonally active form of Vitamin D, 1, 25-dihydroxyvitamin D (1,25(OH)2D).

Vitamin D was chosen as a candidate biomarker for placental-fetal endocrine disruption for a few reasons. Levels increase by 30–75% during pregnancy and stay high over gestation, which could suggest that there is a placental contribution (Reddy et al., 1983; Hollis et al., 2011). There are well-characterized differences in Vitamin D levels and prevalence of Vitamin D deficiency by skin color in the U.S (Bodnar et al., 2009, Ames et al., 2021; Chawla et al., 2019). This is a source of insight and controversy as to a possible physiologic basis for some portion of the persistent Black-White race disparities in birth outcomes (Ames et al., 2021; Bodnar and Simhan, 2010; Thayer, 2014; Bodnar and Mair, 2014). There is also evidence that Vitamin D synthesis and regulation overlaps with placental hormone production and regulation, including hCG (Barrera, Avila, Hernández et al., 2007, 2008). For all these reasons, Vitamin D synthesis and regulation in the first trimester placenta could be important in the fetal endocrine disruption hypothesis.

Vitamin D measurement that addresses placental contributions ideally would include multiple measures. As a marker of Vitamin D status, the serum concentration of 25(OH)D is the most stable form. 25 (OH)2D is the hormonally active form. Key enzymes include CYP27B1 which carries out the 25-hydroxylase and CYP24A1 which catabolizes both forms of Vitamin D. In the non-pregnant individual, the primary organs involved in Vitamin D production are the skin, the liver, and the kidney. The human placenta expresses all components for vitamin D signaling, including the vitamin D receptor (VDR), its cofactor the retinoid X receptor (RXR), 1a-hydroxylase (CYP27B1) and 24-hydroxylase (CYP24A1) (Shin et al., 2010; Christakos et al., 2010; O’Brien et al., 2014).

Expression of these enzymes is present in the trophoblast and in the decidua, and expression is higher in the first trimester versus later in pregnancy (Zehnder et al., 2002). Expression of the Vitamin D receptor (VDR) is also higher in early pregnancy. Expression of these enzymes in the decidual macrophages in the first trimester reflects their function in mediating maternal immune reactions to fetal antigens and preventing miscarriage. 1,25(OH)2D3 specifically has immunosuppressive effects that allow for trophoblast invasion of the myometrium. It was 10-fold higher in the decidua vs the placenta in the first trimester (Zehnder et al., 2002; Tamblyn et al., 2017). Taken together, this could indicate a specific and important role for Vitamin D in early pregnancy and at the trophoblast-maternal interface.

Similar to thyroid hormone, only the hormonally inactive form of Vitamin D is transferred from maternal circulation into the placenta, and into fetal circulation. It is further processed by the placenta. 1,25-dihydroxyvitamin D is not transferred by the placenta into fetal circulation (Ideraabdullah et al., 2019). In preeclamptic cases and controls, Vitamin D metabolite levels were measured and compared in placental tissue, in maternal sera, and in cord blood (Tamblyn et al., 2017). Based on these correlations, it was inferred that there were defects in the transfer of 25 (OH)D, as well as the activation and catabolism of Vitamin D in preeclampsia. Overall, 1,25(OH)2D can regulate cell cycle progression, cell differentiation and induce apoptosis in the placenta (Ganguly et al., 2018).

Placental tissue in the first trimester can be a gold standard to validate and optimize measurement strategy but is not feasible to collect in a cohort study; whereas maternal circulating concentrations are. Within the placental tissue, 25-hydroxyvitamin D3 [25(OH)D3] was strongly correlated (r = 0.83, P < 0.001) with 24,25-dihydroxyvitamin D3 (Zehnder et al., 2002). Moreover, these placental metabolites were strongly correlated (r ≤ 0.85, P ≤ 0.04) with their respective metabolites in maternal circulation. Positive associations (P ≤ 0.045) were also observed between placental mRNA abundance of vitamin D metabolic components and circulating vitamin D metabolites [i.e., 25-hydroxylase (CYP2R1) with 3-epi-25(OH)D3; 24-hydroxylase (CYP24A1) with 25 (OH)D3, 3-epi-25(OH)D3, and 1,25-dihydroxyvitamin D3 [1,25(OH) 2D3]; and 1α-hydroxylase [(CYP27B1) with 3-epi-25(OH)D3 and 1,25 (OH)2D3]. Maternal Vitamin D are reflective of her liver, kidney, and placental functions (Zehnder et al., 2002).

Vitamin D may relate to other biomarkers and molecular pathways expressed in the first trimester, reviewed above. In vitro, 1,25(OH)2D modulated hCG secretion in human trophoblasts, isolated from term placentas, in a time-dependent manner (Barrera et al., 2008). hCG-treatment in pregnant rats increased placental and kidney concentrations of 25-hydroxyvitamin D3 and 24,25-dihydroxyvitamin (Kidroni et al., 1984). To our knowledge, the association of hCG and Vitamin D has not been studied previously in a birth cohort study. A hypothesis has been put forward as to why Vitamin D and serotonin should be studied together (Patrick and Ames, 2014). Authors posit that adequate levels of Vitamin D are necessary for the activation of TPH2 which is an enzyme in the serotonin synthetic pathway.

Vitamin D is a moderately well studied candidate mediator in the DoHAD framework. It has obvious appeal because it can be safely and easily manipulated in pregnant women. Fetal growth restriction was associated with vitamin D deficiency in white women and non-obese women at high risk of preeclampsia (Gernand et al., 2014). Prenatal levels of vitamin D and/or vitamin D deficiency have been studied as causes of childhood asthma, eczema, and respiratory illness from 3 to 10 years of age (Allan et al., 2015; Camargo et al., 2007; Erkkola et al., 2009; Loddo et al., 2023; Liu et al., 2022). Bone development in the fetus starting around 19 weeks gestation is influenced by circulating Vitamin D levels (Hart et al., 2015; Ioannou et al., 2012; Javaid et al., 2006; Mahon et al., 2010; Moon, Green, D’Angelo et al., 2023). The evidence of the association of prenatal vitamin D and the risk of autism has been reviewed elsewhere (Amestoy et al., 2023,Uçar et al., 2020). It is considered equivocal due to the large potential for confounding bias by family history and the absence of experimental data. A systematic review of the literature concluded that there was some evidence for an association of vitamin D deficiency and language and motor development in young children (Janbek et al., 2019); and other neurodevelopmental outcomes (Sammallahti et al., 2023). Lower maternal 25 (OH)D in mid pregnancy is associated with higher abdominal subcutaneous adipose tissue volume (Tint et al., 2018). A deficiency in 25(OH) D3 in pregnancy was associated with increased risk of being overweight in offspring at age 1 year (Morales et al., 2015). In a Danish longitudinal cohort with Vitamin D measures in the first and second trimesters, in cord blood, and at 5 years of age. The association of Vitamin D with childhood blood pressure at 5 years was strongest when measured in the first trimester (Pedersen et al., 2022).

There are few epidemiologic studies that have measured placental molecules in the Vitamin D synthesis, transport, and metabolism pathway. Most report circulating levels of Vitamin or reports of Vitamin D supplementation. In one study, investigators measured cord blood calcidiol (precursor to calcitriol) and calcitriol (most active metabolite of Vitamin D), and placental expression of CYP27B1 mRNA (enzyme that converts Vitamin D to calcitriol), as mediators of risk. They reported an inverse association of Vitamin D levels and maternal blood pressure. Lower levels of Vitamin D were associated with higher risk of maternal urinary tract infection, and specifically in women carrying female fetuses(Olmos-Ortiz et al., 2021). Placental Vitamin D is operating to regulate immunity and inflammation in pregnancy as outlined in review paper (Zhang et al., 2022). Finally, placental and cord blood Vitamin D have been studied as mediators of the effects of air pollution on placental inflammation and risk of hyperinsulinemia (Wang et al., 2023), and other markers of fetal glucose and lipid homeostasis (Liu et al., 2022).

In summary, placental Vitamin D is a challenging pathway to study as it is a combination of synthesis, transfer, and metabolism. Placental and maternal Vitamin D levels were correlated and therefore it is not clear how to disentangle the two. More work is needed to understand maternal-placental feedback in Vitamin D regulation (Zhang et al., 2022). Vitamin D regulation overlaps with that of other steroid hormones and it is associated with multiple domains of development including neurodevelopment, immune system, respiratory health, and cardiovascular health. Unlike other pathways reviewed here, the Vitamin D pathway can be manipulated through nutritional supplementation.

4. Discussion

This review of 10 placental-fetal molecular pathways is intended as a tool for investigators who are employing placental biomarkers in the study of fetal development and endocrine disruption. This information can be useful in translational studies at the population level, in clinical studies, and in mechanistic studies. The objective is to fill a critical gap in our knowledge of how normal development and fetal endocrine disruption occurs in humans considering the growing number of exposures in pregnancy which operate as endocrine disruptors (chemical and non-chemical) and teratogens.

Placental hCG and male fetal steroidogenesis; and hCG-maternal thyroid hormone-fetal neurodevelopment are 2 placental-fetal molecular pathways that are well-established and described in detail here (Fig. 1A). They are currently under study as mechanisms of fetal endocrine disruption. In this review, we stimulate thought and summarize evidence to support other less studied pathways that may follow this paradigm (Fig. 1B). Placental PPARγ, serotonin, and leptin are the most well-studied but lack a clear fetal target. GDF15 is not well studied according to this paradigm but holds promise.

Fig. 1.

A. Established mechanisms of placental-fetal development; B. Molecules reviewed as part of a similar paradigm. Created with BioRender.com. Gestational sac illustration (Adibi et al., 2016). Genitalia illustration by Asklepios Medical Atlas.

We set out to offer a perspective of placental regulation of fetal reproductive tract and brain development. However, we learned in the process that many placental molecular pathways have not been studied with respect to these outcomes. In all cases, we identified literature connecting placental molecular pathways to the risk of preeclampsia. Preeclampsia is a canonical placental pathology that is extremely well studied. It is a helpful starting point when studying placental mechanisms. However, it is also a bias in the literature as the human placenta carries out many essential functions that are not part of the preeclampsia pathophysiology and which are therefore understudied.

In the case of all pathways studied here, we uncovered evidence that the placenta may be signaling as intensively to the mother as it is the fetus – complicating, at least, the directional mother-to-child causal direction that is presumed in the DOHaD and related literatures. The placenta is controlling nausea and food ingestion (GDF15, hCG), appetite and satiety (leptin), progesterone production (hCG), infection prevention (Vitamin D metabolites), mood (serotonin) and thyroid homeostasis and cognition (hCG). Ultimately, the baby’s health is dependent on the mother’s health and well-being during pregnancy. Studies of placental-fetal endocrine disruption should at a minimum consider maternal outcomes, and ideally measure at least one maternal outcome jointly with any given child health outcome.

Across multiple pathways reviewed here, it is becoming evident that in addition to measuring mRNAs and proteins studies should ideally include epigenetic markers. Placental epigenetic markers identified in this review include DNA methylation and non-coding RNAs. Fetal sex is a type of epigenetic/genetic regulation that cannot be ignored in the study of placental-fetal development and/or endocrine disruption. Fetal sex differences in mean levels of placental molecules were identified for hCG (higher in female), Leptin (higher in female), GDF15 (higher in female), DNA methylation serotonin receptor (higher in female). Sex differences in associations of placental-fetal biomarkers were only identified broadly in the case of serotonin (stronger in male), and hCG (stronger in male).

hCG regulation, synthesis and action are the molecular pathways that are outlined and summarized here in the greatest detail. The hCG pathway serves as a foundation upon which to develop mechanistic knowledge of human placental-fetal-maternal endocrine disruption. The review gave rise to a semi-formal theory on the hCG-fetal brain relationship, into which empirical data can be plugged as it becomes available. The three scenarios are 1) direct effects of placental molecules on the fetal brain; 2) indirect effects through other maternal or fetal endocrine organs; or 3) effects by way of general placental vascularization and transport functions. This theory could easily be extended to the development of other molecular pathways and/or fetal organs. It can be useful when designing analytical strategies using biomarker panels, as has been proposed more generally in the case of first trimester placental-fetal teratogenicity (Adibi et al., 2021a).

In reviewing the literature, we identified paradoxes that warrant further investigation. For example, according to the theory, hCG levels should be positively correlated with pro-androgenic outcomes given that hCG can stimulate male fetal testosterone production. However, serum hCG was negatively correlated with pro-androgenic outcomes in the male (i.e. anogenital distance, risk of hypospadias). This may be explained by the fact that hCG was measured in maternal serum and not in the fetal testes, and/or by the fact that clinical studies rely on measures of intact hCG or the beta subunit. It may be the alpha subunit or the hyperglycosylated form of hCG that is able to stimulate fetal androgen production.

There is currently no model of female placental-fetal development or endocrine disruption, unlike the well-developed and empirically proven male model of masculinization programming. Using recent publications of the female fetal ovary transcriptome, we propose here that there may be a molecular pathway that connects the placenta and the ovary (Table 2). The molecules in this pathway fall into the general category of neuroendocrine function and not into the category of steroidogenesis.

Table 2.

Recommendations for future research on placental-fetal development, its utility in prenatal screening for environmental exposures, and ideas for targeted intervention.

N	Molecular pathway	Recommendations for future studies	Candidates for prenatal screening for environmental risks to fetus	Safe method for modulation, post conception
1	hCG	Measure hCG as alpha and beta subunits, and as heterodimer (intact, hyperglycosylated). Investigate female corollary to male placental hCG-fetal LHCGR model: measure candidate mRNAs (NPY, LEP, WNT2B, ETV5 and relevant receptors and proteins) in matched female placental and fetal tissues. Measure and model hCGα in the placental-fetal and in the maternal neuroendocrine-placental axes, along with other ‘neuroactive’ molecules (CSH1, PRL, CRF, Leptin, GDF15, Serotonin). Include maternal and child health outcomes. Combine measures of hCG and the GNRH pathways (placental, fetal) to evaluate the presence of an epigenetically regulated human ‘epiclock’ between the placenta and the fetal brain (analogous to finding in mouse). Analyze ratio of serum hCG to sex steroids as a global indicator of HPA and HPG development that can carry over into postnatal health. Study more hCG biomarkers in the context of cerebral palsy to identify the mechanism by which placental hCG can be neuroprotective in setting of hypoxic injury.	Serum hCG is already in use in prenatal screening for fetal aneuploidies.	None identified.
2	TH	MCT8, MCT10, LAT1, LAT2, ABCB1, NIS may be specific transporters to study in the human placenta to quantify the hCG-TH-fetal brain relationship. Measure D2 and D3 expression to determine regulation of bioactive TH availability.	A combined measured of serum hCG and thyroid hormone, normalized for gestational age, could be informative.	Iodine and selenium supplementation
3	PPARγ	Evaluate sex differences in PPARγ regulation. Evaluate PPARγ (jointly with ABCB1 and P-glycoprotein) as transport mechanism of xenobiotics into placental and fetal tissues. Consider genotypic differences in PPARG, measured and analyzed jointly with placental biomarkers of PPARG expression. Include biomarkers of epigenetic regulation of the PPARγ pathway. Identify role of the placenta in obesogenic effects.	It is not measured in maternal circulation.	Quality and quantity of fats in maternal diet.
4	Leptin	Measure and model in tandem with adiponectin; analyze the two biomarkers as a ratio. Include placental epigenetic and transcriptional biomarkers to identify placenta-specific regulation and production. Adjust maternal serum levels for circulating leptin receptor levels.	It is not possible to distinguish maternal and fetal serum levels.	Diet and exercise
5	TGFβ	Develop theory and methods to distinguish placental TGFβ from maternal and fetal sources. Evaluate microRNA epigenetic regulation.	The evidence is not strong.	None identified
6	Epiregulin	Minimal evidence of placental expression. No evidence of role in fetal development.	There is no evidence to support.	None identified
7	GDF15	Highly promising given strong correlation with hCG. GDF15 and hCGα together may offer a specific panel of placental-maternal endocrine disruption. In non-pregnant individuals, it is being studied as a therapeutic target for dementia and hence may have specific effects on cells in fetal and/or maternal brain.	A potential screening analyte for health risks that are specific to the maternal neuroendocrine-placental axis.	None identified
8	snoRNA	Potentially a type of epigenetic regulation of placental-fetal endocrine disruption. Placental expression and its association with Prader Willi syndrome offer support for relevance to theory. Unclear how to measure and relate to other molecules (coding RNAs and proteins).	It may have potential as a screening analyte for risk of Prader-Willi Syndrome, and related phenotypes.	None identified
9	Serotonin	Literature on placental serotonin, like leptin, has exploded in recent years. It has exceptional appeal given its potential role in DoHAD and the difficulties in directly measuring fetal brain development. Still unclear what is the most salient molecule to measure that is representative of the placental serotonin system.	5-HIAA was proposed as a urinary biomarker of miscarriage and potentially other outcomes.	Tryptophan Maternal diet (triglycerides)
10	Vitamin D	Most important in the first trimester. Requires multiple measures. Challenging to tease apart maternal and fetal measures.	The evidence is not strong.	Supplementation Maternal diet

Leptin is the most well studied in recent years of the molecules reviewed here, based on the numbers of manuscripts identified by the search terms. There is no theory or method currently to disentangle maternal and placental leptin regulation of fetal development. However, they may bias each other when studied jointly. They are both important to fetal development but represent different forms of regulation. This is true of other circulating molecules reviewed here (i.e. Vitamin D, Serotonin, TGFB, Epiregulin). Also similar to other molecules reviewed here (hCG, Serotonin), leptin at the extremes (too low, too high) is associated with adverse outcomes which may also indicate different mechanisms at low vs. high vs. normal concentrations. This scenario requires additional work to determine the relevant upstream and downstream molecules, given differences in mechanisms.

As future directions to move this field of study forward, Table 2 includes suggestions for each of the pathways. Almost every paragraph in the Review ends with a specific recommendation on how to fill key gaps in our knowledge and move the field forward. A repeating observation is the lack of inclusion and/or consideration of the role of the placenta in studies of fetal development. As public health and clinical researchers, we are interested in these pathways as potential sources of screening analytes and/or intervention targets to protect the future health of the offspring and the mother. We did not specifically review the literature to make this determination or recommendation. This type of review could be pursued in future research with the goal of developing empirical strategies to answer these 2 critical questions: 1) can molecules in these pathways be used to identify pregnancies at risk of environmental toxicities to the fetus? 2) are there targeted strategies to safely intervene and modulate environmental risk at the level of the placenta? The second proposition is challenging given the fact that many of these molecules are pleiotropic and an effect on one function is likely to have unintended consequences on other functions. This is essentially what occurred in the DES scenario whereby placental-fetal pathway were pharmaceutically manipulated and the fetal effects through the germline are still being measured across 3 generations [reviewed in (Adibi et al., 2021a)]. To streamline further work in this area, we summarized recurring themes that we encountered in the review of the literature on these 10 pathways (Fig. 2).

Fig. 2.

Recurring themes in the review of placental-fetal pathways in development and endocrine disruption.

Where relevant, the review offers specific suggestions on measurement strategy for the 10 pathways. As the field progresses, this is also a topic that could benefit from high-level review and organization. hCG measurement is essentially regulated at the global level by the World Health Organization (WHO)(Sturgeon et al., 2009). In studies where the entire pathway is measured (exposure, placental mediator, fetal outcome), analytical strategies that combine all 3 into a causal mediation analysis can be useful (Xun et al., 2022; Hong et al., 2022; Qin and Yang, 2021; Discacciati et al., 2018; Vansteelandt and Daniel, 2017; VanderWeele, 2015).

In summary, this review paper can serve as a reference tool to investigators working in the field of placental-fetal development and endocrine disruption. We have organized empirical data and theories developed over the last 50 years. We offer new ideas to be further developed. The information and ideas here will continue to expand and undergo revision. These 10 pathways are useful in quantifying the reproductive and developmental toxicity of diverse types of exposures including endocrine disrupting chemicals, maternal stress, diet, air pollution, and infection. Taken together, this knowledge motivates continued work and innovation to screen for and intervene on exposures during pregnancy that work through the placenta to influence child health.

Data availability

No data was used for the research described in the article.