Introduction

Transcription regulation depends on correct transcription factors (TFs) functioning and chromatin modulation. Two main categories (initiation and elongation) of TFs have been defined depending on their role during transcriptional regulation [1,2,3]. These proteins do not usually act alone as they bind cofactors, thereby forming protein complexes to activate or repress target gene transcription [4, 5]. Epigenetic modifications orchestrated by histone-modifying enzymes are also necessary to transform nucleosome dynamics and structure for producing interaction spaces for transcription complexes. Specific gene expression programmes are greatly related to particular sets of TFs’ temporal and spatial activity leading to cells’ inherent functional properties. Regulation in specific cells and during developmental stages appears to be particularly intricate in some cases, as different TF combinations can bind various genome sites to form regulatory networks [6,7,8,9,10,11].

TFs participate in a wide range of cellular processes due to their function as essential regulatory proteins. Numerous human diseases have been linked to their dysfunction (secondary to encoding or regulatory mutations). Greater than 15% of reported TFs have been related to Online Mendelian Inheritance in Man (OMIM-FOXD1: 601091) links. However, computational exploration of databases has predicted thousands of disease-associated TFs [12]. The amount of this kind of proteins has not yet been stated accurately, but it has been estimated that at least ~ 1400 have high confidence sequence-specific DNA binding domains (DBD) [13,14,15]. TFs are classified into families and subfamilies depending on their DBD sequence/structure and function [12, 16, 17].

The forkhead box (FOX) family of TFs belongs to the “winged helix” superfamily, consisting of proteins sharing a related winged helix-turn-helix DNA-binding motif. FOX genes have been extensively present during vertebrates and invertebrates’ evolution, participating in numerous physiological processes and biological functions, such as embryonic development and organogenesis, cell cycle regulation, metabolism control, stem cell niche maintenance, signal transduction, and many others [18, 19]. A wide range of phenotypes are associated with FOX genes’ dysregulation and mutations in human species [19,20,21,22,23,24,25,26,27,28,29].

Different key biological processes are related to FOXD1 function (also known as BF-2 such as kidney, retina, facial growth centres, hypothalamus, cardiomyocytes, and lung pericyte development, as well as embryo implantation. FOXD1 dysfunction, secondary to FOXD1 coding mutations and expression disturbances, has been linked to different diseases, suggesting it as a possible diagnostic biomarker and a promising target for future therapies.

This paper aims to present, for the first time, a comprehensive review of FOXD1’s role in mouse development (e.g., kidney and retina) and human disease. Molecular, structural, and functional aspects of FOXD1 are presented in the light of physiological and pathogenic conditions, including its role in human disease aetiology, such as cancer and recurrent pregnancy loss (RPL). Taken together, the information given here should enable a better understanding of FOXD1 function for basic science researchers and clinicians.

FOXD1 and the forkhead family of transcription factors

The first Forkhead (FKH) factor was described almost 30 years ago when a new mutant protein was shown to be involved in the formation of the spiked head phenotype in D. melanogaster [30]. This morphological modification resulted from a dysfunction resembling a “forked head” of animals’ foregut during their physiological development. Members of this family have been named forkhead (or fox) proteins since then. Their characteristic DBD (forkhead) is a ~ 100 amino acid long conserved region, defined because of its similarity with a hepatocyte nuclear factor 3 (HNF3) region [31]. Thereafter, Clark et al., used X-ray crystallography at 2.5 Å resolution to describe the HNF-3/forkhead DNA-recognition motif’s DBD three-dimensional structure [32]. FOX gene amount varies considerably in each species, although they are widely distributed amongst vertebrate and non-vertebrate organisms (> 2000 members throughout > 100 species) [19, 22]. At least 19 FOX protein subclasses (A through S) have been defined, depending on their conservation and DBD similarity [18, 33, 34]. FOX proteins’ DBD recognises and binds to the 5′-(G/A)(T/C)(A/C)AA(C/T)A-3′ core sequence to transactivate selected target gene promoters [22, 35]. However, the transactivation domain and accurate mechanisms for regulating target genes have not yet been completely elucidated for most of them.

Regarding human disease, numerous disorders are related to FOX gene dysfunction, such as cancer (e.g. FOXA1, FOXA2, FOXC2, FOXE1, FOXM1, FOXO1, FOXO3, FOXD1), vitiligo, lymphedema-distichiasis syndrome (FOXC2), thyroid dysgenesis, spiky hair and cleft palate (FOXE1, FOXF2, FOXP2, FOXC2 and FOXG1), eye abnormalities (FOXE3), Rett syndrome (FOXG1), heart disease (FOXH1), blepharophimosis, ptosis and epicanthus inversus syndrome (BPES) and non-syndromic primary ovarian insufficiency (FOXL2), intellectual disability (FOXP1), speech-language disorders (FOXP2) and RPL (FOXD1) [19, 36,37,38,39,40,41,42,43,44] (and references therein).

FOXD1 was identified and described for the first time within the forebrain neuroepithelium, and it has been suggested to be an important factor during retina development [45]; this study showed that the BF-2 DBD was closely related to those from other winged helix proteins, which were described thereafter as forkhead transcription factors (e.g. HFH-2 or FOXD3, FKH2 or FOXG1). At chromosome level, FOXD1 is an autosomal monoexonic gene located on chromosome 13 in mice and 5q12-q13 in humans. FOXD1 protein consists of 465 and 456 amino acids in mouse and human, respectively. They have similarity regarding their DBD (Supplementary Fig. S1). FOXD1 has various GC guanine-cytosine (GC)-rich regions encoding different lengths of homopolymeric polyamino acid stretches (e.g. Ala, Pro) having unknown function (Supplementary Fig. S1). These DNA regions are prone to forming hairpin DNA structures and suffering polymerase slippage, thereby hampering PCR, cloning and sequencing of coding regions. Apart from the forkhead domain, FOXD1 has a 122 amino acid-long hyperacidic NH2-terminal domain which might have transactivation properties [46]. Two polyproline runs (positions 241–312) are located within the COOH-terminal domain and described as potential effector sites. FOXD1 is expressed in different tissues and cells such as testes, kidneys, the central nervous system (CNS) (e.g. optic chiasm, retina, pituitary gland), the mesenchyme of facial growth centres, neuroepithelial cells of the prethalamus and hypothalamus, cadiomyocytes, lung pericytes and placenta, while overexpression has been observed in certain types of cancer (see below) [32, 41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71]. Due to its inherent nature as a transcription factor, FOXD1’s subtle expression is mandatory in regulating gene targets belonging to intricate molecular cascades. Several genes have been described as being direct or indirect targets for FOXD1, such as DCN, PGF, C3, DAX1, HMX1 (GH6), HMX4 (SOHo), EPHA3, EPHA6, LHB, PAI-1, PRKAR1A (Riα), p27, ALDH1A3, NKX2–2, AGT and REN [36, 49, 65, 66, 70, 72,73,74,75,76,77,78,79,80] (Fig. 1). In mice, Gli2 is an upstream direct regulator of Foxd1 expression during medulloblastoma pathogenesis [66]. Moreover, Shh transcriptionally regulates Foxd1 during mitogenic activity using Gli2−/−3−/− irradiated mouse embryonic fibroblast (iMEF) cells. In addition, Fgf, Wnt and TGFβ regulate Foxd1 expression in developing chick retina [60, 80, 81]. FOXD1 protein-protein interactions have not yet been validated experimentally.

Fig. 1
figure 1

FOXD1 direct and indirect regulatory proteins and target genes. a FOXD1 direct and indirect regulatory proteins. A1 Fgf8 and Wnt signalling repress FoxD1 expression in optic vesicles. A2 transcription factors such as Ets-1, Wt1 and Wt11, along with the p53 protein, are upstream regulators of Foxd1 gene expression during kidney development. A3 TGFβ and WNT5B factors involved in Hodgkin lymphoma have a transcriptional effect on FOXD1 expression. A4 Gli2 is an upstream direct regulator of Foxd1 expression during medulloblastoma pathogenesis. b FOXD1 direct and indirect target genes. B1 Retina development. B2 Kidney development. B3 Placenta and endometrium. B4 Testes and adipocytes. B5 Induced-pluripotent stem cells (iPSCs) b6 Different types of human cancer. *DEG genes, red: direct FOXD1 regulators and target genes

Full size image

FOXD1 in kidney development

The kidneys develop during three stages in mammalian embryogenesis: pronephros, mesonephros and metanephros. Kidney development in mice begins in E8.5 and lasts to E11.5 whilst it becomes initiated during the 4th week to 35th week of gestation in humans. The metanephros develops during the last stage of renal embryogenesis, giving rise to the mature kidneys, which consist of the ureteric bud (UB) and the metanephric mesenchyme (MM) (Fig. 2). MM stroma mediates UB development, thereby enhancing its differentiation into ureters, pelvises, calyces and the collecting system. The branching UB induces the MM to form the nephrons’ epithelia and renal stroma [50, 82]. Foxd1 expression in mice is constrained to the MM cortical stroma in the developing kidneys (where nephrons differentiate), and it has lower expression in the medulla stroma around the collecting system (Fig. 2). Osr1 is an early marker of kidney development which acts as an upstream regulator of the expression of Six-2 and Pax-2 transcription factors. The Osr1+ cell population is divided into various groups of cells in developing kidneys (E11.5): Foxd1+ cells (which are stromal cell lineage progenitors), Pax-2+ epithelial cell progenitors and Six-2+ cell cap mesenchyme cell progenitors [57, 83,84,85,86]. Foxd1+ and Six-2+ coordinate cell differentiation into distinct kidney compartments [86].

Fig. 2
figure 2

WT and mutant forms of mouse Foxd1 affecting kidney development. a Foxd1 WT. Foxd1 during normal kidney development. Foxd1 expression in mice is constrained to the MM cortical stroma. Hox10, Tcf21, Ecm1, Fat4, Alcam and Rarb are implicated in UB branching. Ets-1, Wt1 and Wt11, along with the p53 protein regulates Foxd1 gene expression. Foxd1+ together with Six-2+ coordinate cell differentiation into distinct kidney compartments. Dcn becomes repressed by Foxd1 in cortical cells required for regulation of Bmp/Smad signalling and Cited1+. Foxd1-expressing population cells (Foxd1+ progenitors) can differentiate into mesangial cells, vascular smooth muscle cells, perivascular fibroblast, renin cell precursors and pericytes. b Foxd1 KO models. The lack of Foxd1 causes small kidneys and alterations in UB epithelium and the MM. Sfrp1 and Raldh2 gene expression is absent, and there is an increase in non-capsular (Bmp4+) and endothelial cells, accompanied by overexpression of Ret in the UB. Dcn is ectopically overexpressed altering Bmp/Smad signalling. Foxd1-null kidneys have shown decreased expression of Agt, Ren, Ace and At1r. MM: metanephric mesenchyme, UB: ureteric bud, red: red: direct FOXD1 regulators and target genes

Full size image

Stroma cells play a pivotal role in signalling pathway activation and the expression of molecules enabling ureters and collecting system development, as well as epithelial and capsule differentiation regulation. Studies using reporter gene activity and a knockout murine model (C57Bl6 Wnt11−/−) have shown that transcription factors such as Ets-1, Wt1 and Wt11, along with the p53 protein, are upstream regulators of Foxd1 gene expression during kidney development [46, 71, 87] (Figs. 1 and 2). The Wnt11−/− murine model has revealed alterations in renal tubule morphology due to downregulation of Foxd1 [87].

Proteins such as Hox10, Tcf21, Ecm1, Fat4, Alcam and Rarb have also been implicated in stroma differentiation and UB branching [88,89,90]. Moreover, Hox10 (a homebox transcription factor) enables proper Foxd1+ cell location within the kidneys. Studies using Hox10-deficient mice models have shown that Foxd1+ cell expression was restricted to the kidney periphery, thereby impairing cortical stromal cell integration during renal embryogenesis (Fig. 2) [85, 89].

Studies of renal development using Foxd1 murine genetic modified models, during E14.5, have shown alterations in UB epithelium due to a small collecting system and inhibited nephron development in MM cortical stroma [50, 58, 85] (Table 1). Hatini et al. used KO mice to determine Foxd1 role during kidney development; they showed that Foxd1 was the earliest genetic marker of kidney stromal cell lineage and that its absence produced defects in UB epithelium and the MM [50]. These authors demonstrated inhibited UB growth and branching, as well as increased condensation in the mesenchyme with a decreased amount of nephrons. The KO mice had misplaced small kidneys which were separated from the adrenal glands.

Table 1 FOXD1 mutant models
Full size table

Levinson et al. used in situ hybridisation and immunohistochemistry to show that Foxd1−/− animals’ kidneys were fused by defects in the renal capsule’s stromal cell population, thus inhibiting their detachment from the midline; they also showed that capsule development was impaired. The renal capsule normally expresses Sfrp1 and Raldh2 genes. However, these proteins were absent in Foxd1-deficient mice. Instead, there was expression of non-capsular (Bmp4+) and endothelial cells, accompanied by overexpression of Ret in the UB, thereby altering its branching. Bmp4 overexpression contributed to ectopic location of the glomeruli in the kidneys [58]. All these defects in kidney development might have been due to alterations in the signalling pathways required for renal capsule maturation [58, 85].

Other studies, using Foxd1 murine models (Foxd1lacZ/+ and Foxd1GFPCre), ChIP and reporter gene assays, have found that decorin (Dcn) became a repressed direct transcriptional target of Foxd1 in cortical cells [74] (Table 1; Fig. 1). Dcn in mice regulates the activation of Bmp/Smad signalling required for Cited1+ nephron progenitor cell differentiation to epithelial cells. Such mice’s differentiated kidneys retained cap mesenchyme cells within the Cited1+ compartment, thus inhibiting nephron progenitor differentiation. Furthermore, Dcn was ectopically overexpressed, thereby altering nephron differentiation by abnormal Bmp/Smad signalling (Bmp-4 and Bmp-7) and epithelial transition induction (Fig. 2) [74, 85, 95].

Hum et al. used a Foxd1-eGFP/Cre transgenic mouse line, immunohistochemistry and in situ hybridisation to determine the role of the renal stroma in patterning developing kidneys (Table 1). They found that Foxd1 played a key role in maintenance of renal stroma integrity [90]. Furthermore, they replicated previous studies’ findings regarding disorganised stromal pattern, loss of limits between nephrons, reduced ureteric branching, differentiation of nephrons from the medulla and abnormal vasculogenesis patterns [50, 58, 85, 90]. This suggests that Foxd1 might be used as an early biomarker of stromal cell lineage during kidney embryogenesis, enabling proper cell differentiation within the renal cortex and capsule.

Foxd1-expressing population cells (Foxd1+ progenitors) can differentiate into mesangial cells, vascular smooth muscle cells, perivascular fibroblast, renin cell precursors (expressing renin-Ren+) and pericytes once a stromal cell lineage has been developed (Fig. 2) [57, 96,97,98]. Foxd1+ cells in mice are endothelial (EPC) and mural renal arteriole progenitor cells, located in peritubular capillaries. Mukherjee et al. isolated Foxd1+ stroma from the Foxd1-eGFP/Cre transgenic mouse strain to grow them as organoids, thereby confirming that the endothelium was differentiated from Foxd1+ cell renal stroma [91, 99]. Moreover, these cells are angiogenic regulators which can synthesise erythropoietin and also play a key role during the arterial tree’s suitable expansion and orientation in the kidneys [90, 92, 94, 97, 100,101,102].

Foxd1+ cell differentiation into mesangial and vascular cells depends on the recombining binding protein suppressor (Rbpj) which is the downstream effector in the Notch signalling pathway [103, 104]. Sequeira-Lopez et al. have used Foxd1-GFP/Cre mice to study whether Foxd1+ stromal cells were renal artery-derived vascular progenitor precursors regulating kidney vascular morphogenesis and orientation. The authors showed that these cells differentiated into Ren+ cells, making cellular networks with capillary epithelium and thus regulating vascular resistance, blood pressure and fluid-electrolyte balance [92].

Moreover, studies using Foxd1-null kidneys have shown that a reduced amount of Ren+ cells can alter renal perfusion due to these cells’ detachment from the capillary walls, as well as decreased expression of Agt (angiotensinogen), Ren (renin), Ace (angiotensin I converting enzyme) and At1r (angiotensin II, type I receptor-associated protein). It is worth noting that Agt and Ren promoters have putative binding sites for Foxd1. Foxd1 therefore might act as an upstream regulator of the renin-angiotensin system [76] (Figs. 1 and 2). Renin-angiotensin system proteins are key molecules during UB branching morphogenesis. Their dysregulation due to the absence of Foxd1 might thus contribute toward abnormal UB development [85, 91, 92, 96].

Some studies have found that Foxd1+ cells, such as pericytes and mesangial cells, are involved in myofibroblast and endothelial cell development, causing kidney fibrotic diseases and a renal pro-inflammatory environment [93, 94, 96, 100, 105,106,107,108]. Humphreys et al. (2010) used a FoxD1/Cre knock-in mouse model for labelling the renal stroma and determining myofibroblast development during fibrosis (Table 1). It was found that Foxd1+-derived pericytes differentiated into smooth muscle actin-positive myofibroblast during fibrogenesis [93].

A recent study by Chang et al. used a Foxd1Cre/+ model and methylation techniques to demonstrate the molecular mechanism regulating Epo expression in fibrotic kidneys. They used Foxd1+ progenitor-derived Col1a1-GFP+Pdgfrβ+ pericytes to differentiate them into myofibroblasts and reported that pericytes expressed Epo, though its expression was inhibited by its promoter’s methylation when differentiating into myofibroblasts during kidney fibrogenesis (Table 1) [94]. Foxd1+ cell homeostatic functions have also been seen to become lost during kidney fibrinogenesis which involved upregulation of Adamts-2 and downregulation of Timp-3 [91, 108].

Foxd1+ cells have thus been defined as key players during proper kidney organogenesis. They contribute toward activating the signalling pathways necessary for the development, differentiation and compartmentalisation of renal cells belonging to the stroma, renal capsule, endothelium and vascular system. FOXD1 in humans may be a useful marker for cell fate maps from kidney embryogenesis to possible kidney diseases. However, how FOXD1 variants affect kidney development in humans or its pathogenesis still needs elucidating.

FOXD1 in retina development

The retina is the part of the eye in close contact with the CNS; its main function is to convert light into electrical signals. This organ consists of photoreceptors, rods and cones. These cells send a light signal to the bipolar cells which, in turn, are in contact with the retinal ganglion cells (RGCs) located in the inner retina. RGC somata occupy ganglion cell layers; their dendrites project into the inner plexiform layer and their axons extend through the optic disk to exit the eye to reach the optic nerve, which transfers light signals to the brain’s visual centres (the lateral geniculate nucleus and the superior colliculus) [109,110,111,112].

RGCs are the first cells to develop from the multipotent progenitor cells in the dorsocentral retina and migrate to the diencephalon. In mice, these cells develop between E11 and P0, peaking in E14.5, while in humans, they develop between the 5th and 18th weeks of gestation [109, 113]. These cells’ development in mice depend on fibroblast growth factors Fgf3 and Ffg8 expression in the neural retina as well as Notch signalling inhibition by secreted frizzled-related proteins and the Delta signalling pathway [109, 114,115,116]. Transcription factors such as Atoh7, Neurog2, Ascl1 and downstream genes such as Brn3b and Isl-1 are expressed to enhance RGCs survival and differentiation [109, 117,118,119]. RGC growth through the optic nerve is achieved by a cAMP-dependent mechanism and the Eph/Ephrin signalling pathway.

Five per cent of the RGCs’ axons remain ipsilateral while 95% decussate to the contralateral optic tract once they approach the optic chiasm in mice, around E10 and E11 [109, 113]. Conversely, 50% of nasal retina RGC axons in humans decussate to the contralateral optic tract, while the remaining 50% of temporal retina RGC axons continue in the ipsilateral tract. It has been stated that activating Shh signalling and transcription factor Isl2 expression are needed for proper decussation. Transcription factor Zic2 promotes Ephb1 expression for RGC ipsilateral projection.

Foxd1 and Foxg1 play a key role during binocular vision development as they participate in regulating RGC ipsilateral and contralateral projection. Foxd1 is expressed in the temporal retina and Foxg1 in nasal retina, as well as in the ventral diencephalon (Fig. 3) [45, 51, 70, 113, 120]. Zic2, a zinc-finger transcription factor, regulates RGC ipsilateral projection from the ventrotemporal (VT) retina to the optic chiasm, while EphB1/ephrin-B2 guides RGC axons ipsilateral projection through the optic tract [121, 122]. Foxd1 is expressed predominantly in the VT retina, near to the midline of the optic chiasm, where it regulates Zic2 and EphB1 expression (Figs. 1 and 3) [51].

Fig. 3
figure 3

WT and mutant forms of mouse Foxd1 affecting retina development. a Foxd1 WT. Foxd1 is expressed in the temporal retina and Foxg1 in nasal retina. Isl2 expression is needed for proper decussation. Transcription factor Zic2 promotes Ephb1 expression for RGC ipsilateral projection. Zic2 regulates RGC ipsilateral projection from the ventrotemporal (VT) retina to the optic chiasm, while EphB1/ephrin-B2 guides RGC axons through the optic tract. b Foxd1 KO model. Foxd1-deficient mice showed RGC axons growing arrested toward the midline of the optic chiasma. These animals had alterations regarding the crossed/uncrossed ratio of RGC axons projecting ipsi- or contralaterally and abnormalities in chiasma shape. Mice also showed ectopic expression of Foxg1 and downregulation of Zic2 and Ephb1 expression. Other genes such as Islet1, Slit2. Boc and EphA/ephrinA also had altered expression. Dashed lines: genes with altered expression due to Foxd1 KO

Full size image

Herrera et al. used a Foxd1lacZ/lacZ murine model to study the retinofugal pathway, and this transcription factor’s role during chiasma development in embryos from E12.5 to P0 (Table 1; Fig. 3). This model showed that RGC axons grew in the same way as those from wild-type mice when they exited the retina to enter the optic stalk during E12.5 and E13.5. However, following this embryonic age, Foxd1-deficient mice showed RGC axons growing toward the midline of the optic chiasma where their development became subsequently arrested. These animals had alterations regarding the crossed/uncrossed ratio of RGC axons projecting ipsi- or contralaterally and abnormalities in chiasma shape. These results demonstrated that Foxd1 is a key regulator of RGC axon migration through the optic chiasm [51]. Moreover, there was an increase of ipsilateral projections as well as the ectopic expression of Foxg1 in areas previously occupied by Foxd1, thereby affecting cell type organisation and the expression of guide factors such as Slit2 (Fig. 3). It was shown that Zic2 and Ephb1 expression was downregulated in the VT retina, indicating that these molecules might be Foxd1 targets in the retina (Fig. 1) [51, 123].

Other genes such as Islet1, Slit2. Boc and EphA/ephrinA also had altered expression, suggesting that Foxd1 is a key regulator of signalling pathways involved in VT RGC axons’ ipsilateral retinal projection [51, 70, 109, 124, 125].

Takahashi et al. studied genetic FoxD1 mechanisms for establishing temporal specificity in chick retina using in ovo electroporation and in situ hybridisation (Table 1). They found that FoxD1 played a crucial role in determining temporal projection from the retina. Moreover, these authors showed that Fgf8 and Wnt signalling repressed FoxD1 expression in optic vesicles (Fig. 1). Altered FoxD1 expression resulted in abnormal projection of RGC axons throughout the nasotemporal retina by dysregulation of downstream target proteins such as GH6, SOHo1 and EphA3 located in the temporal retina [70, 77, 80]. Foxd1 promotes temporal characteristics in the retina during optic vesicle protrusion in response to Fgf and Shh activity in the zebrafish model (Table 1; Fig. 1) [126]. Inhibiting Shh activity produced Foxd1 downregulation and subsequent abnormal routing of retinal axon projections in the mesencephalon.

Interestingly, FOXD1 (xbf-2) can also act as a transcriptional repressor participating in the development of neural plate via Bmp-4 downregulation in the Xenopus model [127]. More recently, Yaklichkin et al., (2007) have predicted the transcriptional repressor nature of Fox TFs in metazoan organisms. The authors showed that the eh1 repressor motif interacts with the Groucho transcriptional co-repressors at the C-terminal region of different Fox family members, thus potentially repressing their transcriptional activity [128]. It is highly probable that FOXD1 also functions in mammalian species as a transcriptional repressor of genes related to various processes involved in embryo development and organogenesis.

All the aforementioned results reported in different species have supported the fact that Foxd1 might be a key player during retinal development and is crucial for the suitable routing of ipsilateral RGC axon projections in mammals. Mutations within FOXD1 in humans might thus alter its functionality, causing alterations in binocular vision due to abnormal ipsilateral projections through the brain’s visual centres.

FOXD1 and the biology of cancer

The biology of cancer is genetically based on the accumulation of mutations and epigenetic modifications leading to cell metabolism dysfunction and uncontrolled proliferation. Genetic abnormalities (e.g. point mutations, gene amplifications, chromosomal translocations) and disturbances in FOX gene expression have been linked to a variety of cancer types [18, 33, 129,130,131]. FOXD1 expression disturbances, rather than mutations in its open reading frame, have been related to the biology of different types of cancer, such as prostate, lung, gastric, hepatocellular carcinoma, glioma, medulloblastoma, ovary, breast and Hodgkin lymphoma [47, 48, 52, 55, 60, 61, 63, 66, 69, 72, 75, 132].

FOXD1 expression (and that of 11 further FOX genes) in prostate cancer has been assessed by qPCR in neoplastic (malignant) prostate tissue, lymph node metastases, benign prostatic hyperplasia, xenografts and prostate cell lines [52]. That study showed that FOXD1 was very poorly expressed in normal prostate tissue but overexpressed in cancer cells and lymph node metastases, thereby seeming to contribute tumour biology (Fig. 1; Table 2). However, differential FOXD1 levels could not be related to the disease’s progression and their quantification was not defined as a prognostic marker of clinical usefulness. Comparative analysis (i.e. microarray analysis) and knockdown assays concerning lung cancer transcriptome have shown that FOXD1 expression was related to abnormal cell proliferation [61].

Table 2 FOXD1 in cancer
Full size table

Microarray analysis has revealed that FOXD1 expression in non-small-cell lung cancer was related to the PTEN signature. Patients having high FOXD1 expression levels had lower survival rates than those having normal FOXD1 transcription. Indeed, high FOXD1 expression levels were linked to different clinical features, such as being male, having a history of smoking, the presence of squamous cell carcinoma and a lack of EGFR mutations (Fig. 1; Table 2) [61].

FOXD1 has been found to be differentially expressed in gastric cancer between metastatic and non-metastatic tumours [47]. These findings led to proposing it as a factor having an important role in complex regulatory networks, especially those involved in cell differentiation. A more recent study using previous gene expression patterns in gastric cancer and robust computational analysis has shown that FOXD1 is a key transcription factor, regulating numerous downstream target genes related to the disease’s progression [63]. A similar study using 8 hepatocellular carcinoma expression profiles proposed that at least ten TFs (including FOXD1) were responsible for most downstream differentially expressed genes [132] (Fig. 1; Table 2).

FOXD1 downregulation in the ovary was related to chemoresistant tumours [55]. Three studies have reported an association between FOXD1 expression and cancer biology in CNS tumours (glioma and medulloblastoma) [48, 66, 72]. Cheng et al. reported a key role for FOXD1-aldehyde dehydrogenase 1A3 (ALDH1A3) signalling in the development of the mesenchymal glioma stem-like cell (MES) population (a malignant glioblastoma subtype) (Fig. 1; Table 2). In vitro and in vivo assays have shown that FOXD1 knockdown led to MES clonogenic properties. Clinically, high FOXD1 expression levels in high-grade glioma were related to patients’ poorest prognosis [72]. FOXD1 was found to be overexpressed in glioma tissue compared to normal brain samples using primary tumours and gene knockdown experiments involving cell lines [48]. Furthermore, siRNA transfection against FOXD1 has led to a decrease in cell proliferation/migration rates and apoptotic features. A very recent study involving mice described Foxd1 as a relevant regulator of the Shh/Gli2-Nkx2-2 molecular pathway regarding medulloblastoma biology (Fig. 1; Table 2) [66]. FOXD1 has also been studied in haematopoietic malignancies, being overexpressed in six Hodgkin lymphoma cell lines and in primary cancer tissues [60]. Bioinformatics functional clustering of transcriptomic data has indicated a potential activator output for the JAK-STAT pathway in FOXD1 transcription. Specific molecules (i.e. TGFβ, WNT5B) belonging to signal transduction pathways involved in Hodgkin lymphoma biology have had a transcriptional effect on FOXD1 expression [60]. FOXD1 has been associated with tumourigenesis and the disease’s progression in malignant breast neoplasms [75]. It has been found to be upregulated in human breast cancer samples [75]. Functional in vitro studies shown that reduced FOXD1 expression has led to decreased proliferation and chemoresistance in breast cancer cell lines. Interestingly, this study also described an important role for FOXD1 in cell cycle regulation as it induces G1 to S transition via p27 direct regulation (Fig. 1; Table 2) [75].

All the above findings might suggest that FOXD1’s could have a major role in cancer biology and provide the basis for proposing it as a coherent target for future therapeutic approaches.

FOXD1 in mammalian implantation and recurrent pregnancy loss

Mammalian reproductive success can be considered the result of several finely regulated physiological and cellular processes in terms of gene expression. Numerous molecular pathways participate in the steps involved, from gamete generation/fecundation to birth. Embryo implantation (a critical stage for fertility and reproduction) involves hundreds of genes participating in complex regulatory networks [133,134,135]. Different mouse models have been used to describe its molecular basis due to inherent ethical limitations in studying implantation in human species. The interspecific recombinant congenic strains (IRCS) model, enabling quantitative trait loci (QTL) mapping, has been particularly successful, as it has led to identifying short genomic regions and genes (e.g. Foxd1 and Alpp) as being relevant molecular actors involved in implantation [36, 136,137,138,139,140,141].

Briefly, the IRCS model was created from an initial cross between two evolutionarily distant parental species of mice (Mus musculus and Mus spretus), followed by two backcrosses involving Mus musculus which reduced the percentage donor (M. spretus) genome introgressed into the receptor’s genetic background (Mus musculus) [137, 139, 140]. Consanguineous (brother-sister) crosses were then performed during > 30 generations to establish 53 IRCS strains, each having an average of 98% M. musculus and 2% M. spretus genomes (fixed at homozygous state). Complex traits can be studied in this model and QTL mapped (using genome markers) by comparing specific phenotypes between strains and parental Mus. musculus animals.

Laissue et al. studied embryonic death/resorption in 39 IRCS strains and 4 sub-strains via in vivo ultrasonography of pregnant females and identified 3 QTL potentially related to the phenotype [138]. One of these (~ 5 Mb in length), located on murine chromosome 13 (MMU13), encompassed 31 genes, three of which were transcription factors (Btf3, PolK and Foxd1). A more recent study determined that the drastic Foxd1-p.Thr152Ala mutation, located in the Foxd1-DBD, became fixed during M. spretus genome evolution (Table 1) [36]. This finding was related to transcriptomic disturbances of endometrial/placental tissues in IRCS-MMU13 females and evoked a potentially major role for Foxd1 in embryonic resorption and litter size modulation.

Interestingly, the Foxd1-p.Thr152Ala mutation has led to transactivation disturbances of direct target genes such as Pgf and C3, two key molecules involved in embryo implantation physiology (Fig. 1). These findings led to proposing FOXD1 as a coherent RPL-related candidate gene. FOXD1 open reading frame direct sequencing involving a panel of 556 RPL-affected women revealed 10 non-synonymous heterozygous mutations potentially related to the phenotype [36]. RPL-related FOXD1-p.Ala356Gly, FOXD1-p.Ile364Met and FOXD1-p.Ala428_Ala429InsAlaAla mutations have shown transactivation disturbances in PGF and/or C3 promoters, similarly to the Foxd1-p.Thr152Ala change. Interestingly, it has been stated that RPL individuals carrying FOXD1 mutations have 10.3 times higher relative risk than women having wild-type alleles, while the FOXD1-p.Ala88Gly variant seems to be related to a protective effect. Taken together, these results have demonstrated that FOXD1 might be a key molecule participating in human implantation and pregnancy maintenance and could be used as a molecular marker in clinical environments.

Concluding remarks and future directions

FOXD1 has demonstrated multi-level roles during normal development, adult physiology and several diseases’ pathogenesis. It is crucial for kidney and retina organogenesis, facial growth centers development in normal conditions, as well as for female fertility. Interestingly, coding mutations in human species have been uniquely associated with RPL, while different kinds of murine models have shown retina, kidney and fertility dysfunction. These differences might have been due to inherent interspecific genomic/transcriptomic/proteomic differences and to the nature of FOXD1 mutant forms (e.g. KO, point mutations, in or out the DBD).

It should be noted that the embryonic resorption phenotype was not studied in Foxd1-KO animals, and kidney and optic dysfunctions were not assessed in animals carrying the Foxd1-p.Thr152Ala mutation [36, 50, 51, 58, 74, 76, 90]. Thus, it cannot be ruled out that these animals could have dysfunctions affecting multiple organs which have not yet been described. FOXD1 variants found in human RPL patients were located exclusively outside the DBD, arguing in favour of drastic mutations possibly being negatively selected or related to another phenotype (e.g. renal or visual). It might also be possible that severe mutations could cause a syndromic disorder affecting various tissues. Consequently, FOXD1 encoding regions should be directly sequenced in patients having different combinations of renal, visual and/or implantation dysfunction.

FOXD1 expression disturbances have been associated with the disease’s pathogenesis in cancer. Specific transcriptome studies and promoter sequence analysis concerned with enlarged types of tumours would thus be interesting. Genome-editing approaches using somatic cells might be considered as a future therapeutic alternative. For instance, CRISPR technology might be applied in vitro and in vivo to modifying FOXD1-relevant genomic regions’ underlying functional characteristics. FOXD1 upstream and downstream regulation might also be explored using additional experimental techniques, such as yeast one-hybrid (Y1H) screening for DNA-protein interactions and chromatin immunoprecipitation sequencing (ChIP-Seq). Finally, it is worth stressing that FOXD1 structure elucidation and novel protein partner discoveries are needed at proteomic level for a better understanding of its function.