Introduction

The ARX gene is located on the short arm of the X chromosome, and since its discovery in 2002,1, 2, 3 it has been implicated in a number of different syndromes and non-syndromic mental retardation.4 Limited human and extensive mouse gene-expression studies show high levels of ARX expression in the foetal brain as well as in some parts of the adult brain. In particular, ARX expression is high in the neuronal precursors of the germinal matrix and ventricular zones at all stages of development. High levels of expression are also observed in the subventricular zone, the caudate nucleus, putamen, substantia nigra, corpus callosum, amygdala and hippocampus.1, 2, 3, 5, 6 This expression during early development and predilection for neuronal tissue suggests that ARX has a pivotal function in neurodevelopment.

We report a family with two male cousins (IV-1 and IV-2, Figure 1a), who were diagnosed with West syndrome (WS) and Ohtahara syndrome (OS), respectively. There is an additional family history of stillbirth, perinatal death and epilepsy in males, compatible with X-linked inheritance. An ARX mutation, c.81C>G in exon 1, was identified in both cousins and their mothers. This mutation represents the first ARX protein truncation mutation that causes a non-malformation phenotype. This apparent lack of a severe malformation is probably because of the re-initiation of the translation from p.M41, allowing an N-terminally truncated, partially functional ARX protein to be produced.

Figure 1
figure 1

(a) The pedigree is compatible with an X-linked mode of inheritance. Affected males (IV-1, IV-2) and obligate carrier females (I-2, II-8, III-6 and III-9) are shown, with the proband (IV-1) designated by an arrow. Individuals II-4 and II-5 are presumed to have been affected. (b) Representative sequence traces are shown for all of the individuals tested for ARX mutations, homozygous ARX c.81C/C wt females (I-3, II-6, II-9, III-1), heterozygous ARX c.81G/C carrier females (II-8, III-6, III-7, III-9) and hemizygous ARX c.81G males (IV-1, IV-2).

Materials and methods

Patient data

The proband (IV-1) is the maternal first cousin of IV-2, as their mothers are sisters. Both mothers are of normal intelligence, in good health and have no history of seizures. There are no surviving males in their mother's or grandmother's generations. However, two maternal uncles (III-2 and III-3, Figure 1a) died perinatally and the maternal grandmother had four male siblings, two of whom (II-2 and II-3, Figure 1a) died shortly after birth. The other two (II-4 and II-5, Figure 1a), who were epileptic, died in childhood. A maternal aunt (III-4, Figure 1a) was also stillborn for unknown reasons.

The proband (IV-1) was born at term by normal vaginal delivery weighing 3.1 kg and was noted to have a left congenital cataract. He presented with focal seizures at the age of 4 weeks. On examination, no other abnormality was noted apart from the cataract. His genitalia were normal. At 4 months, he developed infantile spasms and an electroencephalogram (EEG) confirmed hypsarrhythmia (Figure 2a). He was also found to have gallstones. At the age of five, he is profoundly delayed in all areas of development, is unable to feed orally and has remained refractory to anticonvulsant therapy. He continues to have daily seizures in the form of myoclonic jerks and occasional tonic-clonic fits. The most recent EEG shows a severe epileptic encephalopathy with subclinical status epilepticus in sleep and frequent myoclonic jerks associated with general discharges when awake (data not shown).

Figure 2
figure 2

(a) EEG of individual IV-1 showing a hypsarrhythmia pattern with short periods of suppression. (b) EEG of individual IV-2 is grossly abnormal showing a burst-suppression pattern, consistent with a diagnosis of OS.

Individual IV-2 was a male infant born at 42 weeks by emergency Caesarian section for failure to progress, but did not require resuscitation. Birth weight was 3.54 kg. He presented with tonic-clonic seizures and myoclonic jerks on day five. An EEG at 3 weeks was grossly abnormal showing a burst-suppression pattern consistent with a diagnosis of OS (Figure 2b), a severe early infantile epileptic encephalopathy (EIEE). By 5 months of age, he had axial hypotonia and hypertonia of all four limbs. His genitalia were normal. He made no developmental progress, was visually unaware and died aged 8 months from intractable seizures.

Cytogenetic analysis and extensive metabolic investigations were normal in both boys. Magnetic resonance imaging (MRI) of the brain was performed on both patients (Figure 3a and b). The MRI brain scan of patient IV-1 at the age of 5 months showed mild cerebral atrophy, with no evidence of pachygyria or lissencephaly and normal myelination. A repeat MRI scan at the age of 21 months (Figure 3a) showed slight progression of the atrophy, an atrophic posterior corpus callosum (arrows on right image of Figure 3a), abnormal white matter signal for age and positional plagiocephaly. The MRI brain scan of patient IV-2 at the age of 4 weeks (Figure 3b) showed a normal gyral pattern with no evidence of pachygyria or lissencephaly and a hypoplastic corpus callosum (arrows on right image of Figure 3b) with absent rostrum, splenium and posterior body.

Figure 3
figure 3

(a) MRI brain scan of individual IV-1 at 21 months: axial T2 and sagittal T1 weighted images showing no lissencephaly, but cerebral atrophy, abnormal white matter signal, an atrophic posterior corpus callosum (arrows) and plagiocephaly. (b) MRI scan of individual IV-2 at 4 weeks: Axial and sagittal T2 weighted images showing normal gyral pattern and a hypoplastic corpus callosum (arrows).

Molecular analysis

ARX mutation analysis

After the birth of IV-1 and having reviewed the family history (Figure 1a), X-linked recessive inheritance seemed likely and the possibility of an ARX-related disorder was explored. ARX gene mutation analysis was performed, as described earlier.1 Each exon of the ARX gene (including 30 bp of flanking intronic sequence) was amplified by PCR and sequenced on an automated capillary sequencer (ABI 3100, Applied Biosystems, Foster City, CA, USA) and the sequence compared with the ARX reference sequence (NM_139058) using SeqMan module of the Lasergene DNA and protein analysis software package (DNAStar Inc., Madison, WI, USA).

Generation of an ARX c.81C>G expression construct

pCMV-Myc-ARX wild-type (wt) vector was obtained7 and the c.81C>G mutation was introduced by site-directed mutagenesis using the QuikChange site-directed mutagenesis kit (Stratagene, Cedar Creek, TX, USA) following manufacturer's instructions and using the following mutagenic primer: 5′-CAA CTT TGC TCT CCT CCT AgT GCA TCG ACA GCA TCC TGG G-3′. The entire open reading frame of the construct was then sequenced to ensure no other mutations had been introduced.

Transfection studies and cell culture

HEK-293T cells were transfected with 1 μg of either pCMV-Myc-ARX wt or c.81C>G vector and cultured using earlier described techniques.7, 8

ARX polyclonal antibody production

We produced a polyclonal antibody against a small peptide based on p.161_179 of mature human ARX (KISQAPQVSISRSKSYREN). The peptide was synthetically manufactured (Thermo Fisher Scientific, Waltham, MA, USA) and conjugated to diphtheria toxoid. Polyclonal antibodies were produced by immunizing cross-bred Merino sheep with four injections of 2 mg conjugated peptide in adjuvant at three-weekly intervals (IMVS, Adelaide, SA, Australia). Serum was collected and antibodies were then purified by protein G affinity chromatography.

ARX protein analysis

293T cells transfected with 1 μg pCMV-Myc-ARX constructs were harvested and protein extracted for western immunoblot. Lysates were then subjected to SDS–PAGE, transferred to nitrocellulose and analyzed by immunoblotting. The Sheep α-ARX polyclonal antibody was diluted 1/2000 and detected by a Donkey α-Sheep IgG-HRP-conjugated secondary antibody (Millipore, Billerica, MA, USA) diluted 1/2000. The membrane was stripped in 0.2 M NaOH for 5 min after ARX detection to enable immunoblotting of the same membrane with the cMYC antibody. The mouse α-cMYC monoclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) was diluted 1/1000 and was detected by an α-mouse IgG-HRP-conjugated secondary antibody (Dako, Glostrup, Denmark) diluted 1/2000. Signal was detected through chemiluminescence.

Results

We identified a novel ARX nucleotide change, c. 81C>G in exon 1 of the gene in patient IV-1. Subsequently, mutation analysis on DNA extracted from peripheral blood from family members I-3, II-6, II-8, II-9, III-1, III-6, III-7, III-9 and IV-2 (using Guthrie card sample) was performed. Both affected males (IV-1 and IV-2) were found to be hemizygous for the c.81C>G mutation (Figure 1b), whereas their mothers (III-6, III-9), aunt (III-7) and maternal grandmother (II-8) were shown to be heterozygous carriers (Figure 1b). The two maternal great aunts (II-6 and II-9) were found to be homozygous for the wt allele (Figure 1b). The great grandmother (I-2) was inferred to be a carrier, whereas her sister (I-3) and granddaughter (III-1) are homozygous for the wt allele.

The c.81C>G nucleotide change is predicted to cause a replacement of the tyrosine (p.Y27) codon (UAC) with a premature termination codon (PTC; UAG) leaving the ARX protein only 26 amino-acids long (Figures 1b and 4a). As such, this p.Y27X change is highly likely to be deleterious and, thus, considered to be the mutation responsible for the observed phenotype in this family.

Figure 4
figure 4

(a) Schematic of ARX wt and p.Y27X mutant myc-tagged ARX proteins (adapted from Ref. 7). Shaded dark grey is the 13 amino acids (aa) of the MYC tag with a further 26aa of leader peptide before the p.M1 of ARX. The total number of aa added by the MYC tag and leader sequence is 49. The ARX protein (grey) has aa residue numbers in black below each schematic and contains the following domains: OP, octapeptide domain (horizontal hatch); NLS1-3, nuclear localization sequence (black); pA1-4, polyalanine tracts (white); acidic domain (vertical hatch); homeodomain (diagonal hatch) and OAR Aristaless domain (cross-hatched). Initiation of translation is indicated by an arrow above the protein schematics. A premature N-terminal termination codon is introduced by the c.81C>G mutation, as represented in the lower protein schematic. The star indicates the predicted residue (p.M41) at which re-initiation of translation most likely occurs. The predicted protein sizes that result from over-expression from a CMV promoter in transiently transfected HEK-293T are MYC-ARXwt—611 aa (62 kDa), MYC-ARX(p.M1_S26)—75 aa (4 kDa), ARX(p.M41_C562)—522 aa (53.7 kDa). (b). SDS–PAGE western immunoblot of pCMV-Myc-ARX transfected HEK-293T cell lysates. Lanes were loaded as follows: M—marker 10 μl of precision plus protein standards, un—20 μg of mock transfected cells, wt—2 μg of pCMV-Myc-ARXwt transfected cells, mut—20 μg of pCMV-Myc-ARX c.81C>G transfected cells. The same membrane was first probed with the ARX polyclonal antibody, stripped, then re-probed with the cMYC antibody.

An N-terminally truncated ARX was detected in lysates of HEK-293T cells transiently transfected with a vector (pCMV-Myc) that over-expresses the c.81C>G mutant form of ARX (Figure 4). This truncated ARX, most likely ARX p.M41_C562, is detected by an ARX polyclonal antibody, but not by a cMYC antibody that detects the N-terminal MYC tag. As there is 10 times more protein extract in the c.81C>G lane and the truncated ARX protein is detected at approximately half the level of the wt, there is approximately less than 1/20th the amount of the re-initiation product compared with wt. We predict that the MYC-tagged ARX p.S26 peptide is degraded, and as such could not be detected by the cMYC antibody.

Discussion

The ARX mutation identified in this family is predicted to cause premature protein truncation. Other ARX mutations predicted to cause an early truncation of the ARX protein (EX2-5del, c.196+2T>C, c.232G>T; EX1-2.1del; reviewed in Ref. 4) have been described. Interestingly, virtually all of these cause the X-linked lissencephaly with ambiguous genitalia (XLAG) phenotype, which represents the severe malformation end of the phenotypic spectrum of ARX-associated disorders.9, 10 All of these mutations can also be considered ARX null mutations, which is an assumption supported by the original discovery of the human XLAG mutations that mimic the Arx mouse knockout model.3 We predict that the c.81C>G/p.Y27X mutation is also akin to a null mutation. The resultant p.Y27X ARX protein would lack all of its functional domains, including the octapeptide domain (aa 27–34), and would either not be made, be non-functional or rapidly degraded. In our family, although minor brain abnormalities were visible on the MRI brain scans in the affected individuals, there is a striking lack of the hallmark lissencephaly and ambiguous genitalia associated with the XLAG phenotype. After detecting an N-terminally truncated ARX protein generated from the over-expression of the p.Y27X mutant, we speculated that ARX mRNA translation re-started at a more downstream AUG/Methionine codon. Only one such in-frame codon, at position c.121_123 of ARX, fulfilled this role, suggesting that the N-terminally truncated ARX protein consists of p.M41_C562. Re-initiation of mRNA translation is a known, although still poorly understood, phenomenon (for review see Ref. 11). Translation re-initiation may explain the relatively ‘mild’ non-malformation phenotype observed in this family, consistent with the observations made in other diseases.12 However, this explanation is subject to the proposition that the in vivo c.81C>G containing ARX mRNA escapes degradation by the nonsense-mediated decay (NMD) pathway. We predict that NMD partially degrades the c.81C>G ARX mRNA, as is seen with other PTC containing mRNAs,13 and is most likely protected from complete degradation because of the proximity of the PTC to the translation initiation codon.14 Unfortunately, we were unable to test ARX mRNA or protein expression in vivo because of the lack of a relevant ARX-expressing tissue source from the patients. Provided that ARX mRNA translation does re-initiate at codon c.121_123 (aa 41 of wt ARX), such an N-terminally truncated protein would completely lack the octapeptide domain (aa 27–34). This domain is an important transcription co-repression domain of ARX and is involved in binding to the Groucho/transducin-like enhancer of split co-factor proteins.8 The p.L33P mutation of the octapeptide domain is known to give a non-syndromic phenotype in the MRX54 family,2 but perhaps a more severe phenotype would have been observed if the entire octapeptide was absent.

In summary, we speculate that the lack of a severe malformation phenotype observed for these patients could be explained by a partially functioning, N-terminally truncated ARX protein (p.M41_C562). This truncated ARX protein might be less abundant than the wt (Figure 4), probably because of partial NMD degradation of the PTC containing ARX mRNA and the inefficiency of translation re-initiation from the c.121_123 AUG codon.

Another possible mechanism behind the partial phenotypic rescue of these patients is RNA editing of the c.81C>G containing ARX mRNA, which would allow translation of a full-length ARX mRNA (reviewed in Ref. 15). However, this could not be tested because of the unavailability of a suitable patient material.

Investigations did not reveal an underlying cause for the cataract or gallstones in patient IV-1, neither of which has been described earlier in patients with an ARX gene mutation and the significance of these features remains unclear.

OS is an EIEE with the earliest known age of onset for any age-dependant epileptic syndrome, often presenting within days of birth or even prenatally.16 OS is characterized by frequent minor generalized seizures and ‘burst suppression’ on the EEG. ‘Bursts’ are high amplitude slow activities, often with sharp waves or spikes, interspersed with periods of ‘suppression’ with little or no activity. The seizures are intractable and are co-morbid with severe psychomotor retardation. Prognosis is poor with about one in three patients dying before the second year of life. OS can be distinguished from other neonatal epileptic encephalopathies by the disease's progression during development.16 OS progresses into WS in 75% of cases at approximately 3–4 months of age. After this, 59% of WS cases develop into Lennox–Gastaut syndrome (LGS),17 distinguishing it from early myoclonic encephalopathy, another EIEE, which neither progress into WS nor LGS.16 The aetiology of OS is heterogeneous,16 and can be caused by structural abnormalities such as hemimegencephaly, porencephaly, hydrocephaly and lissencephaly and includes syndromes such as Aicardi syndrome. In addition, a cryptogenic non-malformation form occurs. Similar to mental retardation, there is a 30% excess of males with OS, indicative of the involvement of genes located on the X chromosome,18 and an ARX polyalanine expansion mutation has recently been described as a cause of OS in two out of three sporadic male cases.19 ARX is expressed in GABAergic neurons, and Kato et al19 hypothesized that dysfunction of the GABAergic system is critical to the neuropathology of EIEE and WS. In addition, heterozygous mutations in the STXBP1 gene seem to cause OS in both males and females.20

OS and XLAG have many features in common such as early onset during infancy, frequent intractable seizures, abnormal burst-suppression EEG patterns, hypotonia, impaired motor function and profound developmental delay. Mutations in ARX have been reported to account for up to 94% of XLAG9 and be responsible for X-linked WS.1 On the basis of our findings, we suggest that impaired function of the ARX protein can also result in OS. Perhaps this finding is not surprising given the clinical overlap between XLAG and OS, as well as the progression of most OS cases into WS. In addition, this family shows intrafamilial clinical heterogeneity, indicating the influence of different genetic and environmental background on the clinical expression of a given ARX gene mutation. Specifically in this case, we suggest that subtle differences in genetic background and environment affected the extent to which NMD or translation re-initiation occurred in the two affected individuals, leading to the different clinical manifestations. To our knowledge, this is only the second report of an ARX gene mutation causing OS and the first with an inherited mutation. Interestingly, in the cases reported by Kato et al,19 both had a 33-bp duplication in exon 2 of ARX rather than a protein truncating mutation. This supports the conclusion that modified function rather than complete loss of function of the ARX protein may be the cause of OS in our family. This mutation represents the first ARX protein truncation mutation causing non-malformation phenotype and points towards the complexities of the interplay between the mutation, molecular mechanism and the genetic background. It also reminds us to exercise caution when drawing conclusions about genotype–phenotype correlation. We suggest that ARX screening should be considered in affected males with OS.