Article Text


Phenotypic variability in Rett syndrome associated with FOXG1 mutations in females
  1. C Philippe1,
  2. D Amsallem2,
  3. C Francannet3,
  4. L Lambert4,
  5. A Saunier1,
  6. F Verneau1,
  7. P Jonveaux1
  1. 1Laboratoire de Génétique Médicale, Nancy Université, Centre Hospitalier Régional et Universitaire, Vandoeuvre-les-Nancy, France
  2. 2Service de Neuropédiatrie, Hôpital St Jacques, Besançon, France
  3. 3Service de Génétique Médicale, CHU Hôtel-Dieu, Clermont-Ferrand I, France
  4. 4Service de Médecine Infantile 1, Hôpital d'Enfants, Centre Hospitalier Régional et Universitaire, Vandoeuvre-les-Nancy, France
  1. Correspondence to Dr C Philippe, Laboratoire de Génétique Médicale, Nancy Université, EA 4002, Centre Hospitalier Régional et Universitaire, Rue du Morvan, 54511 cedex 1, Vandoeuvre-les-Nancy, France; c.philippe{at}


Background The FOXG1 gene has been recently implicated in the congenital form of Rett syndrome (RTT). It encodes the fork-head box protein G1, a winged-helix transcriptional repressor with expression restricted to testis and brain, where it is critical for forebrain development. So far, only two point mutations in FOXG1 have been reported in females affected by the congenital form of RTT.

Aim To assess the involvement of FOXG1 in the molecular aetiology of classical RTT and related disorders.

Methods The entire multi-exon coding sequence of FOXG1 was screened for point mutations and large rearrangements in a cohort of 35 MECP2/CDKL5 mutation-negative female patients including 31 classical and four congenital forms of RTT.

Results Two different de novo heterozygous FOXG1-truncating mutations were identified. The subject with the p.Trp308X mutation presented with a severe RTT-like neurodevelopmental disorder, whereas the p.Tyr400X allele was associated with a classical clinical RTT presentation.

Conclusions These new cases give additional support to the genetic heterogeneity in RTT and help to delineate the clinical spectrum of the FOXG1-related phenotypes. FOXG1 screening should be considered in the molecular diagnosis of RTT.

Statistics from

Rett syndrome (RTT, OMIM 312750) is a disorder of postnatal neurodevelopment mainly caused by de novo mutations in the MECP2 gene encoding methyl-CpG-binding protein 2.1 2 MECP2 is involved in up to 90% of classical cases of RTT, but its mutation rate is low (∼20%) in variant RTT.3 RTT is a clinically defined condition with a large spectrum of phenotypes. In the past few years, this disease has been found to be genetically heterogeneous. Microdeletions found by comparative genomic hybridisation (CGH) array and balanced translocations have allowed the identification of genes responsible for phenotypes that overlap strongly with RTT. A translocation that disrupted the Netrin G1 gene (NTNG1) was reported in a patient with the early-seizure variant of RTT.4 However, mutations in the NTNG1 gene appear to be a very rare cause of RTT, as no further deleterious mutations have been described since then. Mutations in the X-linked cyclin-dependent kinase-like 5 (CDKL5/STK9, OMIM 300203) gene have been identified in both females and males with the Hanelfed variant of RTT and/or X-linked infantile spasms (OMIM 308350).5–10 More recently, the FoxG1 transcription factor (formerly brain factor 1 (BF1)) was found to be altered in two girls with clinical features consistent with congenital RTT.11 FoxG1 is a DNA-binding transcription factor with a fork-head-binding domain, which represses target genes during development of the brain, with a key role in early telencephalon patterning.12 It recruits transcriptional corepressor proteins via two protein-binding domains: JARID1B-binding domain (JBD) and Groucho-binding domain (GBD).

In a diagnostic setting, we have analysed MECP22 13 14 and CDKL5 genes9 in a very large cohort of patients with typical and variant forms of RTT. Here, we report molecular screening of the FOXG1 gene in a cohort of 35 female patients with a typical form or a congenital variant of RTT. We describe two females heterozygous for a FOXG1 mutated allele.

Patients and methods


In this study, we included 35 female patients who were recruited at different clinical genetic centres throughout France with a clinical diagnosis of RTT based on revised criteria for this disorder.15 All patients with classical RTT satisfied at least seven required criteria together with supportive criteria. Thirty-one cases were considered to be typical RTT, and four to be congenital or early-onset RTT when the normal perinatal period was absent or shorter than 6 months without any regression period, respectively. Mutations in the MECP2 and CDKL5 genes were not detected in these patients by direct sequencing of the entire coding sequences and exon/intron boundaries and by multiplex ligation-dependent PCR amplification. Blood samples were taken from the patients after informed consent had been obtained. Total genomic DNA was extracted from peripheral blood using the Nucleon BACC genomic DNA extraction kit (GE Healthcare, Orsay, France).

Mutation screening of FOXG1 by direct sequencing of PCR products

We analysed the entire coding sequence by direct sequencing of PCR products. Exons 1–5 of the FOXG1 gene were PCR amplified (table 1). Primers were modified by the addition of either M13F (5′-TGTAAAACGACGGCCAGT-3′) or M13R (5′-CAGGAAACAGTCATGACC) sequences at their 5′ end. The coding sequence was screened by direct DNA sequencing with M13F and M13R primers as described previously13 except for exon 4, for which we used internal primers because of a polyT tract at the end of intron 3 (table 1). Sequences were automatically analysed with the Seqscape 2.5 software (Applied Biosystems, Courtaboeuf, France). Sequence variants are numbered starting from the first base of the ATG codon, with numbering based on reference sequence NM_005249.3. Naming of variants with the Alamut 1.4 software (Interactive Biosoftware, Rouen, France) follows the Human Genome Variation Society nomenclature. Mutations reported in this study have been deposited in the Italian Rett database (

Table 1

Primers in FOXG1 used in this study

Screening for large rearrangements of FOXG1 by quantitative multiplex PCR of short fluorescent fragments (QMPSF)

Detection of large rearrangements of the FOXG1 gene was performed by QMPSF. The QMPSF analysis was performed as described previously13 with a multiplex PCR amplifying exons 1, 3, 4 and 5 of the FOXG1 gene.16 Table 1 shows the primer sequences. One multiplex PCR with five amplicons was performed in a 25 µl reaction volume with 2 mM MgCl2, 200 µM dNTPs, 0.15–0.6 µM primers, 2 M betaine, 2.5 U Taq polymerase (Thermoprime; Abgene, Epsom, Surrey, UK) and 200 ng genomic DNA. PCR consisted of an initial denaturation step for 4 min at 95°C followed by 24 cycles: 95°C for 30 s, 58°C for 30 s, 72°C for 30 s. The PCR was ended by an elongation step for 7 min at 72°C.

RT-PCR of the FOXG1 transcripts

Total RNAs from whole blood were extracted with the PAXgene blood RNA system (PreAnalytiX, Hombrechtikon, Switzerland) or from lymphoblastoid cells with the TRIzol reagent (Invitrogen, San Diego, California, USA). Before reverse transcriptase (RT)-PCR, RNAs were treated with DNase I (Sigma, St Louis, Missouri, USA) at room temperature for 15 min; DNase I was inactivated at 70°C for 10 min. RT-PCR with primers located in exons of the different alternative transcripts was performed with the QIAGEN (Courtaboeuf, France)OneStep kit.


Identification of mutations in the FOXG1 gene

This molecular study resulted in the finding of premature stop codons (PTCs) at amino acids 308 (c.924A>G; figure 1A) and 400 (c.1200C>G; figure 1C) in patients 1 and 2, respectively. Testing of parents revealed that both mutated alleles were de novo. It was not possible to evaluate the effect of the PTCs on the stability of the FOXG1 mRNAs, as we failed to RT-PCR amplify any transcripts for this gene in leucocytes and lymphoblastoid cells.

Figure 1

FOXG1 mutations identified in this study. (A) FOXG1 mutation c.924G>A in patient 1. Sequence trace obtained with primer 5′-ACGCTCAACGGCATCTACGA-3′ (amplicon exon 1–3) in patient 1 showing a heterozygous base (G/A) at position c.924 resulting in a stop codon at amino acid 308. (B) Brain MRI data in patient 1 at 15 years of age. Axial (1) and sagittal (2) T1-weighted images: frontal lobes are hypoplastic, and the corpus callosum is thin. Decreased white matter volume results in an enlarged occipital horns aspect. The cortex is thick with a paucity of gyral development. (C) FOXG1 mutation c.1200C>G in patient 2. Sequence trace obtained with primer 5′-TCAGAACTCGCTGGGCAACA-3′ (amplicon exon 1–4) in patient 2 showing a heterozygous base (C/G) at position c.924 resulting in a stop codon at amino acid 400.

Clinical reports

Patient 1 is a woman aged 22, the only child of healthy and unrelated parents. She was born by spontaneous delivery after an uneventful pregnancy, with Apgar scores of 8 at 1 min and 10 at 5 min. Birth weight (3400 g), length (54 cm) and head circumference (34 cm) were normal. Head growth deceleration was noted at 2 months. She was considered to be a normal child until the age of 4 months when she presented with strabismus, feeding difficulties and interaction dysfunction. She also presented with global hypotonia, foot deformity (equinovarus), choreic movements, sialorrhoea and night screaming fits. At 2 years of age, she was diagnosed as having severe epilepsy with hyperthermia-induced seizures. She developed spontaneous myoclonic seizures with age. Interictal awaken recordings showed background activity in the θ frequencies (4–6 Hz), symmetrical, continuous, with a voltage of 20–30 µV, attenuated by eye opening. Between 4 and 21 years of age, background activity stayed the same. At age 4–10, EEGs showed widespread biphasic waves in the δ or θ frequencies occurring in brief bursts, without clinical signs or sometimes accompanied by clonic jerks of the head. At age 11, spike-and-wave oscillations appeared in these bursts, which became polyspikes at age 15. Common sleep patterns were absent until the age of 17, when spindles were first seen. In light and deep sleep, background activities were in the θ then δ frequencies, with slower frequencies on the right hemisphere. EEGs remained organised during slow sleep despite consistent activation of paroxysmal patterns as well as widespread bisynchronous waves, with high amplitude on anterior regions. These bifrontal waves were asymptomatic when in brief bursts, or followed by conjugate lateral ocular deviations with unilateral clonic jerks of the superior limbs when in longer and rhythmic trains. These patterns were different from those seen in Lennox–Gastaut syndrome. Two bone fractures (of the femur and humerus at the ages of 5 and 10 years, respectively) may suggest osteoporosis. She never acquired speech or purposeful hand skills. Cranial MRI at age 15 showed ventricular enlargement and hypoplasia of the corpus callosum with decreased frontal and occipital white matter volume and a paucity of gyral development (figure 1B). At re-evaluation at 22 years of age, she presented with inexplicable episodes of laughing, hand stereotypies and bruxism. She had feeding problems resulting from swallowing difficulties, and had a body weight of 44 kg (body mass index 17). The seizures are controlled by clonazepam and oxcarbamazepine antiepileptic drugs.

Patient 2 is a 10-year-old girl who was born by spontaneous delivery after an uneventful pregnancy. She is the first child of healthy and non-consanguineous parents. Her birth weight was 3880 g, and her length was 52 cm with a normal head circumference (34 cm). She was considered to be a normal child until age 6 months, when developmental delay was noticed. She was not able to walk until the age of 30 months. Postnatal deceleration of head growth started at the age of 9 months; her head circumference was 48.5 cm (−2 SD) at 10 years of age. The clinician noticed repetitive stereotyped hand movements, absence of speech, frequent and inappropriate episodes of laughing, and hemihypotrophy of the left half of the body at the age of 5. A normal methylation pattern at the 15q11–q12 region did not support a diagnosis of Angelman syndrome and a diagnosis of RTT was proposed. At the age of 9 years, she presented with autistic behaviour with poor eye contact, ataxia and abundant drooling in addition to self-abusive behaviour. She was capable of saying a few words (mummy and daddy), and bit her hands. Hand stereotypies were no longer noted, and hand skills were improved. Breathing, sleeping pattern and peripheral vasomotor function were normal. No seizures have been noted by her parents. In addition, she has foot deformity (equinovarus) and minor anomalies, with protruding ears, anteverted nostrils and an open mouth appearance. Brain MRI did not reveal any malformation.


FOXG1 located in 14q12 was cloned as a gene containing the fork-head domain (HFK1 (human fork head 1)) in 1994.17 It encodes a developmental transcription factor with repressor activities. Its expression profile is restricted to brain and testis. In the mouse, Foxg1 transcription factor is critical for forebrain development. In concert with Fgf (fibroblast growth factor) signalling, it participates in early telencephalon patterning at approximately E8.5 in mouse, and is essential for the establishment of ventral identity in the telencephalon.12 The Foxg1 transcription factor was also implicated in the development of the olfactory system18 and in sex hormone signalling by interacting with the androgen receptor protein.19 FOXG1 is subjected to alternative splicing, with the FOXG1B major transcript expressed in fetal and adult brain.16 17 Four additional FOXG1B splicing variants have recently been characterised exclusively in fetal brain tissue.16

We report on two nonsense mutations in the FOXG1 gene in two female subjects. Patient 1, who has a p.Trp308X mutation, has a severe RTT-like neurodevelopmental disorder with brain malformations, whereas the more C-terminal p.Tyr400X stop codon is associated with a classical RTT in patient 2. The overall mutation rate of FOXG1 in our series is 3.2% (1/31) in females with a classic RTT, whereas one in four females with congenital RTT is positive for FOXG1. Patient 1 has a severe RTT-like phenotype suggestive of congenital RTT with MRI anomalies. The clinical presentation in patient 1 is very similar to that of patients reported on by Ariani et al.11 This severe RTT-like phenotype prompted a CDKL5 screen, although patient 1 did not show early-onset epileptic encephalopathy. Patient 2 can be considered to have classical RTT (table 2) according to the revised diagnostic criteria for classical and variant RTT.15

Table 2

Clinical features of female patients with alterations in the FOXG1 gene

The FOXG1B transcript described by the NM_005249 mRNA sequence is encoded by a single exon. We can speculate that the abnormal FOXG1B mRNAs in patients 1 and 2 are not affected by the nonsense-mediated mRNA decay (NMD) mRNA surveillance pathway and give rise to truncated proteins missing different parts of the C-terminal region of FoxG1. The FoxG1 transcription factor represses the transcription of target genes through recruitment of Groucho and JARID1C corepressors. The GBD interacts with the Groucho protein, which is widely used by many developmentally important repressors for silencing their various targets. The JBD recruits the JARID1C demethylase involved in demethylation of trimethylated and dimethylated Lys 4 histone 3 (H3K4). Demethylation of H3K4 is associated with silenced chromatin. In patient 1, the mutated FoxG1 transcription factor misses both the GBD and JBD essential for recruiting transcriptional corepressors. The clinical presentation in patient 1 is very similar to that of both cases reported by Ariani et al, with absent or no functional Groucho and JARID1C-binding domains.11 The PTC is more C-terminal in patient 2. The truncated FoxG1 protein contains an intact GBD and therefore may retain residual repression activity in patient 2. This residual activity may explain the milder phenotype in this patient as compared with the more severe RTT in patients heterozygous for a non-functional FoxG1 protein incapable of recruiting any corepressors. FOXG1 alternative transcripts consist of multi-exon mRNAs expressed in fetal brain. In both patients, the PTCs may trigger the NMD for these transcripts in fetal brain. It was not possible to estimate the effect of PTCs on the stability of the FOXG1 splice variants B2–B5,16 as we failed to amplify these transcripts on total RNAs extracted from leucocytes or lymphoblastoid cells.

Before our study, to the best of our knowledge a total of 15 FOXG1 alterations were described including 12 large deletions, one t(2;14)(p22;q12) with a breakpoint in intron 3 of FOXG1, and two nonsense mutations (figure 2). Initially, chromosome abnormalities revealed by conventional karyotyping or CGH array were reported to affect the FOXG1 locus. Seven large deletions of the 14q proximal region encompassing the FOXG1 locus were not precisely mapped.20–24 In these cases, the absence of precise breakpoint mapping and the high number of genes in the deleted region hampered any genotype–phenotype correlations. A 3 Mb deletion was identified by CGH array analysis in a girl with severe mental retardation and many RETT-like features.25 This was the first report to show a link between the FOXG1 gene and RTT in a patient who initially tested negative for MECP2 and CDKL5. Brain MRI scans were normal. In a compilation of nine patients with a 14q rearrangement, breakpoint mapping showed that two male patients (No 3 and No 5 in the report by Kamnasaran et al28) were heterozygous for deletions encompassing the FOXG1 locus.26–28 Case 3 showed autistic-like features, microcephaly and generalised hypotonia together with additional symptoms probably related to other contiguous genes in the 4 Mb deleted area (from D14S275 to D14S975). The 10 Mb deletion in case 5 (from D14S80 to AFM205XG5) was associated with severe psychomotor retardation, hypotonia and brain malformations (agenesis of the corpus callosum and asymmetric ventricles). CGH array analysis in a girl with a maternally inherited balanced (X;3) translocation revealed an interstitial 14q12 deletion removing FOXG1.29 The deletion had a 2.65 Mb minimal size and was associated with microcephaly, hypermetropia, epilepsy and facial dysmorphic features. At the age of 14 years, she showed a normally developed brain. Haploinsufficiency of the fetal brain-specific FOXG1 splicing variants, as a result of a t(2;14) with a chromosome 14 breakpoint in intron 3 of the FOXG1 gene, has been associated with severe mental retardation, brain malformations (asymmetric enlargement of the lateral ventricles and complete agenesis of the corpus callosum) and microcephaly (noted at the age of 6 months).16 Recently, a FOXG1-truncating mutation was found in two female patients affected by the congenital variant of RTT.11 For patients with a large deletion encompassing additional genes, it is not clear to what extent clinical features are the result of FOXG1 haploinsufficiency. A compilation of all phenotypes associated with FOXG1 mutated alleles shows a high prevalence of specific brain anomalies, including microcephaly, dysgenesis of the corpus callosum, abnormal ventricles and abnormal white matter.11 16 20 21 23 27 30 31 Foxg1 is essential for correct brain development in the mouse. A heterozygous Foxg1+/−-cre line mouse has microcephaly with a reduction in the volume of the neocortex, hippocampus and striatum.32 Using another Foxg1+/− mouse model, Shen and collaborators33 showed that Foxg1 is also expressed in areas of postnatal neurogenesis. The postnatal expression of FoxG1 may explain the acquired microcephaly in patients heterozygous for a mutated allele. The brain phenotype in these mouse models is evocative of, although not identical with, that of patients with a FOXG1-related phenotype. It is noteworthy that a complete heterozygous deletion of FOXG1 does not necessarily result in brain malformations in male and female patients (table 2).25 29

Figure 2

Deleterious mutations identified to date in FOXG1. The 489-amino acid long FoxG1 transcription factor is encoded by the intronless FOXG1 coding region (part of exon 1). Four alternative transcripts involving exons 2–5 have been identified in fetal brain. Fourteen pathogenic alleles have been described: (1) four nonsense mutations in exon 1 (Ariani et al11 and this study); (2) nine large deletions encompassing FOXG120–23 25–31; (3) one translocation with a breakpoint interrupting the FOXG1 gene.16 FHD, fork-head domain; GBD, Groucho- binding domain; JBD, JARID1B-binding domain.

To date, four point mutations in FOXG1 have been found in females initially diagnosed as having typical RTT, congenital RTT or a severe RTT-like neurodevelopmental disorder. They are associated with a severe phenotype with brain MRI anomalies when the truncated FoxG1 transcription factor is devoid of corepressor-recruiting domains (table 2). Clinical heterogeneity resulting from allelic heterogeneity is common in genetic conditions. We report on a late-truncating nonsense mutation associated with classical RTT. This finding highlights the need to analyse the FOXG1 gene not only in congenital RTT with brain anomalies. Missense mutations in FoxG1 functional domains may result in milder phenotypes, including typical and atypical RTT, syndromic mental retardations with telencephalon abnormalities, and non-syndromic mental retardations. FOXG1 mutation screening should be considered in male patients with a neurological disorder. The FoxG1 protein interacts with the androgen receptor and may regulate sex-hormone signalling.19 Abnormal testis development was noticed in two male patients with an interstitial 14q12 deletion encompassing FOXG1.20 26 Therefore, males with brain anomalies, acquired microcephaly, clinical features of the RTT series and testis anomalies would be good candidates for FOXG1 mutation screening.

We applied a two-step strategy for the search for point mutations and large rearrangements affecting exons 1–5 of FOXG1. A balanced de novo t(2;14) with a 720 kb inversion in FOGX1 was detected in a female patient with syndromic mental retardation, a phenotype reminiscent of clinical pictures of female subjects with PTCs in FOXG1.16 A priori, the chromosome 14 breakpoint located within the FOXG1 gene affects only alternative multi-exon transcripts, although we cannot exclude a position effect with a new chromatin environment incompatible with transcription of the intronless FOXG1 transcript. Point mutations or large rearrangements in exons 2–5 of FOXG1 may also be responsible for encephalopathies with brain anomalies and clinical features of RTT. We did not find any deleterious alleles (point mutations or large rearrangements) in exons 2–5 of the FOXG1 gene in a cohort of 35 female patients with classical or congenital RTT. Additional patients need to be screened to estimate the potential implication of fetal brain-specific FOXG1 transcripts in the molecular aetiology of RTT or RTT-like phenotypes.

In conclusion, our data give additional support to the involvement of FOXG1 in the molecular aetiology of RTT and help to delineate the clinical spectrum of FOXG1-related phenotypes.


We gratefully acknowledge clinicians who referred patients for MECP2/CDKL5/FOXG1 analysis.


View Abstract


  • Funding This study was supported by a grant from GIS-Institut des Maladies Rares.

  • Competing interests None.

  • Provenance and Peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.