IUGR and PE are two complex diseases which are frequently associated with maternal and foetal complications during pregnancy. A clear association between these disorders has been documented, as women suffering PE have an increased risk (up to 4-fold) of being affected by IUGR (Fox et al., 2014; Srinivas et al., 2009). Conversely, IUGR-affected individuals have an increased risk of being affected by PE (Mitani et al., 2009). PE and IUGR share pathophysiological mechanisms affecting the placenta and endometrial tissues, such as hypoxia, thrombosis, ischemia, impaired angiogenesis and inflammation (Armaly et al., 2018; Collinot et al., 2018; Garrido-Gomez et al., 2017; Gurugubelli and Vishnu, 2018; Shamshirsaz et al., 2012; Sharma et al., 2017). Several molecular pathways thus become simultaneously dysregulated, which may partly result from the dysfunction of key transcription factors acting in the endometrium and the placenta. Particular interest in FOXD1 has been highlighted as it has been shown to play a central role in mammalian embryo implantation and pregnancy maintenance (Laissue et al., 2016, 2009). FOXD1 mutations have led to embryo resorption in mice and RPL in humans by perturbing transcriptional networks in the endometrium and placenta. It was thus considered that FOXD1 was a coherent candidate gene in the present study as it is potentially related to other female reproductive phenotypes, such as RIF, IUGR and PE.
We focused our attention on FOXD1-p.His267Tyr and p.Arg57del from the 9 non-synonymous sequence variants identified in the present study since they are rare and had not been described previously in RPL women (Laissue et al., 2016) (Table 1). The c.799C > T (p.His267Tyr) variant was carried by a Colombian RIF patient. Since the Colombian population consists of a particular ethnic admixture, and its genetic composition/variability is not widely represented in public SNP databases, we screened this variant in a panel of 203 ethnically-matched controls. The variant was not found in this control population, thereby arguing in favour of an association with the disease’s aetiology. Furthermore, the His267 residue has been conserved during mammalian species’ evolution, strongly suggesting functional relevance (Additional file 1). Accordingly, SIFT and PolyPhen bioinformatics’ prediction tools gave scores compatible with a harmful effect (Table 1). Furthermore, replacing a histidine (His) with a tyrosine (Tyr) has been predicted to be potentially deleterious, since as His is an amino acid which is electrically charged with basic side chains, whilst Tyr is a large aromatic polar uncharged molecule. The p.His267Tyr mutation could thus have led to local or global changes regarding FOXD1’s physiochemical properties, thereby contributing to transcriptional disturbances.
We used a gene reporter system to explore this hypothesis as it facilitated assessing FOXD1’s transactivation capability regarding the C3 and PlGF promoters. C3 belongs to the complement system family of proteins which has at least 50 members and can be activated in several tissues by different mechanisms (Regal et al., 2015). Interestingly, complement factors (including C3) act at the crossroads of endometrium/placenta development and physiology, meaning that they can be considered key molecules potentially involved in various female reproductive disorders (Laissue et al., 2016; Regal et al., 2017, 2015). Recurrent studies in animal models hint to a central effect of C3 deregulation in placental pathophysiology (Girardi, 2018; Girardi et al., 2015; Qing et al., 2011; Wang et al., 2012).
In vitro, we showed that the protein’s WT version was able to transactivate the C3 promoter (1.9-fold, p = 0.024) (Fig. 1). The FOXD1-p.His267T and the pArg57del mutations led to statistically significant increases in C3 transcription activity compared to that induced by the WT version. This finding reinforced those described previously for the FOXD1-p.Ile364Met and p.429AlaAla mutations identified in RPL women, arguing in favour of this variant’s functional contribution to the phenotype (Laissue et al., 2016).
High C3 levels have been recorded in women having suffered three pregnancy losses, which might be linked to other inflammatory-related molecules’ local (endometrium and placental tissues) expressional disturbances. Interestingly, increased complement activation has been recorded in human placentas following spontaneous abortion, while CD46 and CD55 (complement regulators) became reduced (Banadakoppa et al., 2014; Regal et al., 2015). Here, we have identified the FOXD1-p.His267Tyr mutation in a RIF patient consistently with the hypothesis that FOXD1 plays an essential role in early pregnancy maintenance. This is consistent with our previous observations where the 66H-IRCS strain of mice (which carries the M. spretus–derived Foxd1-Thr152Ala mutation) presents with high rates of early embryonic death (Laissue et al., 2009).
Regarding the FOXD1-p.Arg57del mutation (which we identified in an IUGR patient), we also considered it as potentially having a functional impact because it had a low MAF in the gnomAD database. Similarly to other harmful FOXD1 missense mutations, FOXD1-p.Arg57del may lead to the protein’s three-dimensional conformational changes and functional disturbances. The FOXD1-p.Arg57del mutation increased 1.5-fold C3 promoter transcriptional activity. Although C3 levels have not been widely studied in women affected by IUGR, due to its relevant role during placental physiology we consider that the transcriptional increase observed in our experiments might also be found in vivo.
FOXD1 has already been shown to be a regulator of PlGF in mice and humans (Zhang et al., 2003, Laissue et al., 2016). We observed increased PlGF transcriptional activity (1.4 fold, p = 0.002) with the FOXD1-p.Arg57del mutation compared to that for the FOXD1-WT counterpart, thereby arguing in favour of a potential placental dysfunction leading to IUGR. Low plasmatic PlGF levels have been reported in PE women and it has been seen that FOXD1 mutations have led to reduced induction capacity on the PlGF promoter in recurrent pregnancy loss patients, whilst PlGF overexpression has been linked to enhanced angiogenesis in tumours (Laissue et al., 2016, Chau et al., 2017 and references therein). These findings and the results of the present work, suggest that fine-tuning PlGF expression is an essential condition contributing to placental/endometrial physiology; indeed its transcriptional dysregulation may contribute to different diseases pathogenesis.
Interestingly, two FOXD1 previously identified mutations (p.Ile364Met, p.Ala356Gly) were re-identified in the present study in IUGR patients. We have previously found them in RPL women and shown that they led to C3 promoter transactivation disturbances (Laissue et al., 2016). Indeed, similarly to that observed in our present FOXD1-p.His267Tyr and p.Arg57del experiments, the FOXD1-p.Ile364Met mutation also increased C3 promoter transcription activity ~ 5 fold (Laissue et al., 2016). These findings argue in favour of FOXD1 mutations possibly contributing to IUGR pathogenesis.
Surprisingly, contrary to that observed for the FOXD1-p.Arg57del mutation, FOXD1-p.Ala356Gly has been reported to decrease C3 promoter transcription activity (Laissue et al., 2016). Although complement cascade activation has been observed in PE patients, it has been postulated that the fine tuning of C3 expression may be an important factor regarding physiological gestation in mice and humans (Chow et al., 2009; Laissue et al., 2016; Lynch et al., 2012, 2011, 2008; Regal et al., 2017). C3 expression disturbances (up or down-regulation) due to FOXD1 mutations over/under specific thresholds might therefore contribute to RPL, PE, and/or IUGR. The functional differences amongst FOXD1 mutations might be related to specific physicochemical modifications triggered by particular amino acid changes and/or secondary to regulatory networks’ inherent downstream complexity. It should be also taken into account that other genetic (e.g. variants in other genes) and epigenetic (e.g. imprinting of paternal alleles, or consequences to variable environmental exposures) changes may modify FOXD1 mutations’ phenotypic effect.
Taken together, our results argue in favour of FOXD1 mutations’ central role in RPL, RIF, IUGR and PE pathogenesis via C3 regulation (Laissue et al., 2016). We consider that FOXD1 should be genotyped in larger panels of patients to establish an accurate genotype-phenotype correlation and to justify proposing it as a reliable, clinically useful biomarker.