In cranial neural tube slices of White Leghorn chick embryos, Ahlgren et al. (85) performed retroviral gene transfer to overexpress the avian homolog of FoxG1, V-qin, in the telencephalon and to ectopically express it in the mesencephalon, rhombencephalon, and spinal cord (85). The ectopic expression of FoxG1 resulted in a selective overgrowth of the telencephalon and mesencephalon (midbrain) but not in more posterior brain regions. As well, there was a marked thickening of the neuroepithelium. Interestingly, a separate experiment demonstrated that retroviral expression of FoxG1^NHR−AAA (virus containing the FoxG1 construct with the DNA binding domain inactivated) resulted in no observable phenotype. This finding suggested that the brain overgrowth is mediated through the DNA-binding domain of FoxG1 (85). Ahlgren et al. (85) concluded that the observed overgrowth was not due to an increase in proliferation rates (85). Embryos examined 2–3 days after retroviral infection demonstrated no significant increase in BrdU incorporation in the neural tube. Similarly, there was no detectable effect of FoxG1^NHR−AAA on proliferation as measured by BrdU or mitotic index (85). Rather than uncontrolled proliferation, Ahlgren et al. suggested that the absence of normal programmed cell death was associated with the brain overgrowth observed in FoxG1 overexpressing chicks (85). The authors used DAPI as an indicator of apoptotic nuclei and observed that dying cells appeared small and bright. Cellular counts revealed a significant decrease in the number of apoptotic nuclei in the anterior neural tube including both the telencephalon and mesencephalon. Furthermore, control retroviruses and FoxG1^NHR−AAA did not yield a significant change in the apoptotic index compared to embryos with no virus infection.

In Xenopus laevis embryos in which FoxG1 (known as XBF-1) is overexpressed, studies revealed an expansion of the telencephalic progenitor population (90, 91). According to Hardcastle and Papalopulu, embryos injected with a high XBF-1 concentration show increased proliferation over an area of expanded or ectopic neuroectoderm, such that the normally bilayered neuroectoderm becomes multilayered (91). XBF-1 injected embryos also demonstrated proliferating neural precursor cells in lateral domains where post-mitotic cells would normally be found. Overall, the authors show that a high dose of XBF-1 causes tissue outgrowths in the ectoderm, increases proliferation, and inhibits the expression of the cyclin-dependent kinase (cdk) inhibitor p27XIC (91). Bourguignon et al. demonstrated that neuronal differentiation is specifically suppressed in cells in which XBF-1 is expressed at high levels (90). This was seen also via the thickening of the ectoderm in developing Xenopus models as well as through an increase in the number of proliferating progenitor cells in place of differentiated neurons in the anterior neural plate (90). These studies support a model whereby increased XBF-1 leads to more proliferation of precursor cells possibly through suppression of neuron differentiation.

“Natural” experiments exist in humans whereby mutations have arisen on chromosome 14q12 where FOXG1 is duplicated leading to three gene copies instead of two. Pontrelli and others recently reviewed 15 cases with duplications on chromosome 14q12 all of which included FOXG1 (92), where epilepsy and cognitive impairment with dysmorphic features are the common phenotypes of this cohort. There was no identifiable microcephaly, though there is also no macrocephaly, arguing against a simple balanced model of FOXG1 to drive proliferation, at least in human. Some amount of FOXG1 may be required to ensure enough progenitor cells are made. Too much FOXG1 however, may not affect this specific process which may involve an interaction with specific proteins that govern the generation of progenitor cells. The epilepsy and intellectual disability phenotypes in the duplication cases may arise from completely different mechanisms than from loss of FOXG1 dosage, i.e., the interaction of FOXG1 with different molecules or with different genomic regions.

Cancers are broadly defined as a group of diseases that involve abnormal cell growth with the potential to invade or spread to other parts of the body. Tumors are large masses that are often the result of this abnormal growth of cells. Resistance to cell death is an important feature of cancers, where apoptosis has been established as a mechanism of anti-cancer defense. Gliomas are a common form of brain cancer characterized by excessive cell proliferation and aggressive infiltration (93). Notably, FOXG1 has been shown to be upregulated in glioma as well as ovarian cancer and medulloblastoma (94–96), and to have important driver effects. To examine the role of FOXG1 in glioma, Chen et al. examined FOXG1 expression in two cultured glioma cell lines (U87MG and SHG44) and found elevated FOXG1 expression in U87MG cells (93). A lentivirus-mediated expression system was used to overexpress FOXG1 in SHG44 cells and a lentivirus-mediated shRNA was used to knock down FOXG1 in U87MG. The results of these expression studies demonstrated that cell proliferation was decreased as a function of downregulated FOXG1. Similarly, increased cell proliferation was associated with increased FOXG1 expression. The authors further questioned whether the change in the proliferation rate was attributed to altered apoptotic activity. In FOXG1-overexpressing SHG44 cells, apoptosis appeared to be reduced given by the decreased expression of caspase-9, 8 and 3 and the cleaved versions of these pro-apoptotic proteins. Furthermore, expression of these caspases was elevated in the FOXG1 knockdown U87MG cells, indicating increased apoptotic activity. Together, these results suggest that FOXG1 has a pro-survival function and that expression is negatively correlated with glioma cell apoptosis (93).

Given the purported pro-proliferation and anti-differentiation activity of FOXG1, a study by Wang et al. (97) hypothesized that FOXG1 expression supported the resistance of glioblastoma multiforme (GBM) cells against temozolomide (TMZ) treatment. TMZ is a DNA methylation agent and drug resistance-modifying agent that induces G2/M arrest and apoptosis. Upon TMZ treatment, viability of GBM cells was assessed using an MTT assay (apoptotic assay) which demonstrated significantly reduced cell viability—defined as the ratio of initial cell number minus dead cell number to the initial cell number (97). GBM cells transiently overexpressing FOXG1 in combination with TMZ treatment showed significantly improved cell viability, indicating that FOXG1 resisted the anti-proliferation ability of TMZ treatment (97).

In a smaller study, Adesina et al. (98) demonstrated that FOXG1 is significantly differentially overexpressed in aggressive medulloblastoma subtypes from four publicly available gene expression profiling data sets. As a result, the authors attempted to examine the genome-wide effect of down-regulating FOXG1 expression in DAOY (a medulloblastoma cell line) by running an mRNA expression profile of 44,000 genes using the shFOXG1, shLuciferase, and the UT DAOY cell lines. Whole-genome expression analyses revealed pathways affected by decreased FOXG1 including those involved in cell adhesion and migration (98). As expected, changes in expression were seen in genes previously implicated in cancer. There also appeared to be a variety of altered genes involved in cell survival or anti-apoptotic activity (98). In a separate experiment, the authors demonstrated that mice xenografts injected with DAOY cells demonstrated enhanced survival when transfected with shFOXG1 knockdown constructs as opposed to sh-Luciferase. Overall, these studies offer evidence for the overexpression of FOXG1 in mediating excessive cell survival in glioma and medulloblastoma, respectively (93, 98).

Idiopathic autism spectrum disorder refers to individuals where no underlying cause for the disorder has been identified. (99) suggest that FOXG1 may act as a convergence point for these ASDs associated with macrocephaly and modeled these patients in human stem cells. Gene expression profiling of neurons derived from different patient lines revealed that overexpression of FOXG1 was ubiquitous in their transcriptomic profiles. The authors also observed an excitatory/inhibitory neuron imbalance in brain organoids generated from proband iPSCs, such that FOXG1 may be partially involved in the overproduction GABAergic neurons.

Studies from model organisms, human duplication cases, neurodevelopmental disorders with macrocephaly and no mutation in FOXG1, and human tumors suggest that cell survival and tissue growth are sensitive to FOXG1 gene dosage. Tipping the balance of FOXG1 toward overexpression leads to a reduction in cell death and tissue overgrowth. Brain overgrowth and tumor formations are logical consequences of FoxG1 overexpression as its role in promoting proliferation and cell survival are amplified beyond normal levels. This idea needs to be tempered with the results from the human duplication cases where no brain overgrowth was observed (92), arguing against a simple FOXG1 dosage model. While the data do support a FOXG1 dosage sensitive model in brain cells, it may be that the mechanism important in reduced dosage of FOXG1 operates on different molecules than those that are important where there is too much FOXG1, something we call an asymmetric dosage sensitivity model.

Foxg1 heterozygous mice were first generated while making Foxg1 homozygous KO mice and were considered as controls (i.e., before the human heterozygous deletion syndrome was identified, highlighting the importance of heterozygotes). This was attributed to the fact that several initial studies reported that mice with a single allele of Foxg1 develop an apparently normal cerebral cortex (5, 33, 50, 86, 100). Foxg1 heterozygous mice did not exhibit the severe cortical defects in patterning observed in the null mice (100), at least on cursory observation. Closer investigation of Foxg1^+/− mice identified smaller cortical volumes and Foxg1 heterozygous mice showed a reduction in layer II/III thickness associated with microcephaly and impaired hippocampal neurogenesis (62, 101). The Foxg1^+/− model also showed hyperlocomotion, impaired habituation in the open field and a severe deficit in contextual fear conditioning (62, 63, 101). The cerebral cortex, hippocampus and striatum were observed to have reduced volumes in the Foxg1^+/− mice (62, 63), though this may be strain or genetic background dependent. For example, the forebrain of heterozygous Foxg1 mice maintained on the C57BL/6J background had severely impaired development. However, Foxg1^+/− mice of the Foxg1-tet line and Foxg1-lacZ and Foxg1-cre mice maintained on a mixed background, did not display reduced cortical thickness. This suggests that reduced but not absent Foxg1 in mice displays complex interactions with brain development.

Clinical data on several FOXG1 deletion syndrome patients have been reviewed and discussed in this review; however, understanding why a loss or mutation in one copy of FOXG1 leads to microcephaly and severe intellectual disability in humans is unknown. Human-derived iPSCs now make it feasible to generate isogenic, patient-derived neurons to investigate neurodevelopment and to perform functional genetic studies (102). Patriarchi et al. generated iPSC-derived neurons from FOXG1^+/− patients and suggested that there is an imbalance in excitatory/inhibitory (E/I) synaptic protein expression in patient neurons compared to controls (103). However, these data do not explore the dynamics of FOXG1 dose as neurons develop. It seems reasonable to suspect that the molecular mechanism of disease will arise early on as cells differentiate and any overt cellular phenotype at a mature cell stage is a passenger effect to an earlier problem in cell differentiation. It is these early molecular mechanisms that need to be assessed to understand how FOXG1 dose leads to a reproducible, robust cellular phenotype. To this end, a recent study was able to generate human stem cells where FOXG1 dose could be fine-tuned (104). Studies such as these will become important in titrating specific doses at specific times for in vitro neurodevelopment.

FoxG1 is known to interact directly with Groucho (Gro)/Transducin-like enhancer of split-1 (TLE1) by forming a transcription repression complex with co-repressors of the TLE family (105, 106). TLE family members are transcriptional repressors that lack a DNA binding motif and so are dependent on other factors for this function, like Foxg1. Among mammals, there are four full-length TLE family members (TLE1-4) and two shorter isoforms–Groucho-related gene product (Grg) 5 and 6. Only full-length TLE and Grg6 proteins contain a conserved C-terminal WD40 repeat domain mediating interaction with FoxG1. Grg6 acts as a dominant-negative regulator of FoxG1:TLE transcriptional repressor complexes (107). Grg6 interferes with the binding of TLE1 to FoxG1 and does not repress transcription when targeted to DNA. Moreover, co-expression of Grg6 and FoxG1 in cortical progenitor cells leads to a decrease in the number of proliferating cells and increased neuronal differentiation (107). Furthermore, Roth et al. show that Xenopus tropicalis TLE2 (a closely related family member to TLE1) physically interacts with FoxG1 in the ventral telencephalon (subpallium) (108) via a conserved N-terminal Engrailed Homology 1 (EH1) motif. Knocking down TLE2 leads to impaired development of the ventral telencephalon, similar to the knockdown of FoxG1. This suggests that TLE2 is a spatially restricted member of the Groucho/TLE family, which interacts with FoxG1 to specify and promote the development of the ventral telencephalon. The dynamic interplay of TLE and Grg proteins shows just how dynamic altering the total level of Foxg1 protein could be. The binding affinities of each would be critical to determining outcome, and suggest that a simple linear model (more expression of FOXG1 = more binding with TLE) is not necessarily correct.

FOXG1 cooperates with KDM5B (previously JARID1B or PLU-1), a histone demethylase, to potentially regulate cell proliferation and differentiation. The interaction between KDM5B and FOXG1 is mediated by a conserved interacting motif (Ala-X-Ala-Ala-X-Val-Pro-X₄-Val-Pro-X₈-Pro; termed the VP motif) in both proteins (109). The interaction between FOXG1 and the transcriptional repressor KDM5B is of functional importance for early brain development (110). In particular, during mouse embryogenesis, KDM5B expression overlaps with FOXG1 expression both spatially and temporally (111). While the two interact directly, KDM5B also acts as a repressor of FOXG1 expression (112). KDM5B then can both regulate the expression of FOXG1 and bind to FOXG1 protein, possibly forming important regulatory loops. Dosage change in FOXG1 would thus have important consequences on the activities of KDM5B.

KDM5B is an H3K4 demethylase (mono-, di-, and tri-), and therefore has a role in removing an important mark of actively transcribed regions. KDM5B is classified as a repressive chromatin writer and so loss of function would lead to a more permissive (i.e., more gene expression) chromatin state. KDM5B is predominantly expressed during embryonic development, including embryonic stem cells (ESCs) and also the adult brain, testis, eye, spleen and thymus (111–113). It has also been identified as an oncogene in many cancer types (114). This suggests that while it is essential for normal development, any perturbation to gene expression may induce abnormal phenotypes related to growth and survival.

FOXG1 and BMI-1 expression levels are tightly correlated with each other in a close expression loop to affect neural progenitor cell survival (115). BMI-1 is part of the polycomb repressive complex 1, a transcriptional repressor complex known to interact with multiple proteins (116). PRC1 is thought to repress gene expression by affecting the level of histone H2A variants in nucleosomes (117), levels of which determine the stem-like state of a cell. While FOXG1 and BMI-1 may not physically interact, their tight regulatory relationship might suggest that the proper dose of FOXG1 is important for PRC1-mediated gene repression in nerve cells, perhaps as a proper guide to the correct genomic coordinates.

FGF8 is thought to directly affect FOXG1 during neurodevelopment to specify and pattern the ventral telencephalon (40, 84, 118). FGF8 is a morphogen meaning it derives its function from amount or concentration along a specific gradient. Morphogens are an appealing model to explain why FOXG1 dose may be important since more or less FOXG1 might lead to more or less FGF8, or vice versa. At neural plate stages, Fgf8 induces and/or maintains Foxg1 expression in the anterior neural ridge (119). Foxg1 then restricts the expression of Bmp4 to the midline where BMP4 is believed to induce apoptosis. Analysis of serial sections of forebrains from normal, Fgf8 function eliminated, and Fgf8 function reduced animals confirmed that the Foxg1 and Bmp4 expression domains in the midline were complementary. These observations support the hypothesis that FGF8 regulates telencephalic cell survival in part via a Foxg1 pathway and that either eliminating or increasing Fgf8 expression decreases Foxg1 pathway activity; whereby reducing Fgf8 expression increases it (119).

The PI3K (phosphatidylinositide-3′-OH kinase)-Akt pathway (120) promotes neuronal survival by inactivating the cell death machinery and repressing pro-apoptotic gene expression (121, 122), likely through IGF-1 (123). This signaling cascade activates CK1 and AKT, both of which can target FOXG1 (122). Foxg1 may be imported into the nucleus of cells through its phosphorylation by CKI which promotes NPC differentiation into neurons, while Foxg1 phosphorylation by Akt at Thr271 leads to Foxg1 nuclear export. Loss of FOXG1 dose may mean there are less FOXG1 targets available to be phosphorylated and that a critical mass of FOXG1 may need to be phosphorylated for AKT to execute its cellular programming.

The transforming growth factor-β (TGF-β) pathway consists of multiple cytokines that control a wide variety of biological activities including apoptosis, cell proliferation, differentiation, cell adhesion, and embryonic development through TGF-β and other receptors and Smad transducer proteins (124, 125). Studies have shown that FoxG1 may act as a negative regulator of TGF-β signaling pathway by binding to the MH2 of Smads -1, -2, -3, and -4 (96, 126). This association blocks the binding of Smad proteins to DNA and results in the inhibition of TGF-β signaling (127). FoxG1 has been shown to inhibit expression of the cyclin-dependent kinase (CDK) inhibitor p21^WAF1/CIP1, which is normally transcriptionally activated by TGF-β signaling, in glioblastoma and the neuroepithelium (96, 128). While TGF-β signaling is complex and context dependent, some of the effects of decreased FOXG1 dose could be exerted through its interactions with the SMAD proteins.