A de novo EGR2 variant, c.1232A > G p.Asp411Gly, causes severe early-onset Charcot-Marie-Tooth Neuropathy Type 3 (Dejerine-Sottas Neuropathy)

EGR2 (early growth response 2) is a crucial transcription factor for the myelination of the peripheral nervous system. Mutations in EGR2 are reported to cause a heterogenous spectrum of peripheral neuropathy with wide variation in both severity and age of onset, including demyelinating and axonal forms of Charcot-Marie Tooth (CMT) neuropathy, Dejerine-Sottas neuropathy (DSN/CMT3), and congenital hypomyelinating neuropathy (CHN/CMT4E). Here we report a sporadic de novo EGR2 variant, c.1232A > G (NM_000399.5), causing a missense p.Asp411Gly substitution and discovered through whole-exome sequencing (WES) of the proband. The resultant phenotype is severe demyelinating DSN with onset at two years of age, confirmed through nerve biopsy and electrophysiological examination. In silico analyses showed that the Asp411 residue is evolutionarily conserved, and the p.Asp411Gly variant was predicted to be deleterious by multiple in silico analyses. A luciferase-based reporter assay confirmed the reduced ability of p.Asp411Gly EGR2 to activate a PMP22 (peripheral myelin protein 22) enhancer element compared to wild-type EGR2. This study adds further support to the heterogeneity of EGR2-related peripheral neuropathies and provides strong functional evidence for the pathogenicity of the p.Asp411Gly EGR2 variant.


Results
Clinical history and neurological examination. The proband is a 56-year-old male who was diagnosed with DSN at six years of age following a nerve biopsy (results not available). He had no family history of peripheral neuropathy, and no family consanguinity. His father and mother had normal nerve conduction studies (NCS) in their fifties. His eldest sister also had a normal neurological examination and NCS at age 25, and the proband reported that this sister has not developed symptoms of peripheral neuropathy in her fifties.
At two years of age, his mother noted that he was consistently fatigued and was repeatedly falling when attempting to walk. He also reported loss of sensation in his feet in early childhood and was never able to run. He was not a candidate for ankle-foot orthoses, and subsequently had a triple arthrodesis at age 15 to stabilise his left ankle. He was also diagnosed with scoliosis and had a Harrington rod inserted at age 16 to correct this. During his teenage years, he had multiple unexplained episodes of hemiparesis and double vision after severe migraine pain behind his eyes, and these episodes would last for approximately two weeks. He reports his mother and sister also suffered from frequent severe migraines, particularly between the ages of 30-45. On examination at the age of 31, he presented with muscle wasting of the hands and feet, as well as pes cavus, hammer toes, corns, and callouses. He walked with a high steppage gait. He had mildly weak hip flexion, and mildly weak ankle dorsiflexion (3/5). He had absent reflexes in the knees and ankles, with distal loss of vibration sense to the knees. There was loss of pain sensation to above the ankles and decreased joint position sense in the toes. He had a tremor of the hands, with weakness of wrist and finger extension. NCS conducted when the patient was 36 years old (Table 1) showed a severe demyelinating motor and sensory neuropathy.
Autonomic testing conducted at this time showed no postural hypotension, normal cardiovascular reflexes, and normal sudomotor reflexes in both upper and lower limbs indicating no significant autonomic neuropathy. An audiogram showed mild bilateral hearing loss above 1500 Hz, and he also reported occasional tinnitus. Right and left ear brainstem auditory evoked potential (BAEP) waveforms were poorly formed with a bilateral delay in wave V. Right and left full and half field visual evoked potentials (VEP) showed delayed P100 waveforms. An MRI of the brain showed multiple subcortical and periventricular white matter hyperintensities on T2 and FLAIR sequences.
On re-examination at age 52, NCS of the left median sensory, left median motor, and left peroneal motor nerves showed no response. He also reported difficulty opposing his thumb and fingers, and a progressive loss of sensation in his right hand. On examination, there was slight curling of the fingers and wasting of the intrinsic hand muscle. Dorsiflexion in his fingers was mildly weak (grade 3/5), as was his finger abduction (grade 3/5). An additional NCS conducted 12 months later additionally showed no response of the left ulnar sensory nerve, and a severely reduced latency of 19.5 m/s in the left ulnar motor nerve. Four years later at the age of 56, his ability to oppose his thumb and fingers had worsened, and he reported that he was unable to write, or use a knife and fork. He additionally had allodynia below his knees. Another brain MRI was conducted at this time ( Fig. 1),  which again showed multiple bilateral inactive demyelination foci of T2/FLAIR white matter intensity which were mostly periventricular, however they were also present subcortically (mostly in the frontal lobes), at the callososeptal interface and in the right thalamus. There was no restricted diffusion or suspicious enhancement, and the findings were similar to those from an MRI performed age 37 years. A lumbar puncture for oligoclonal bands was not performed.
Due to the additional phenotype of hemiplegic migraine in our proband, as well as migraine in his mother and eldest sister, variant filtering of WES was also conducted for genes related to familial hemiplegic migraine (FMH). As variants in CACNA1A can also cause an FMH phenotype in addition to an inherited neuropathy, segregation of the CACNA1A c.6692G > A variant was conducted. However, it was present in the mother and absent in the FMH-affected eldest sister. Additionally, the proband had no variants in the ATP1A2 or SCN1A genes, which are both known to cause FMH.

In silico analyses show that p.Asp411Gly EGR2 affects a highly conserved residue and that it
is likely pathogenic. Amino acid sequence alignment demonstrates that the p.Asp411Gly variant occurs at a highly conserved residue in EGR2 with surrounding amino-acid residues being conserved between orthologues in different vertebrate species (Fig. 3). Multiple in silico analysis techniques were utilised to predict the conservation of the Asp411 amino-acid residue and pathogenicity of the p.Asp411Gly variant. GERP 24 , phast-Cons 25 , and PhyloP 26 scores all provided further support that this amino-acid residue is highly evolutionarily constrained ( Table 2). The functional effect of the p.Asp411Gly variant was predicted to be damaging by Polyphen2 27 , PROVEAN 28 , SIFT 29 , and MutationTaster2 30 ( Table 2). p.Asp411Gly EGR2 has reduced transcriptional regulatory activity compared to wild-type EGR2 in cultured schwann cells. To determine the relevant functional consequences of the p.Asp411Gly EGR2 variant, the ability of wild-type and p.Asp411Gly EGR2 to induce activity of a previously reported EGR2 response element at the PMP22 locus was tested in a Schwann cell line 31 . A construct expressing wild-type or p.Asp411Gly EGR2 was transfected into cultured rat Schwann (S16) cells along with a construct harbouring an EGR2 response element upstream of a minimal promoter 32 and a firefly luciferase (Fluc) reporter gene. To control www.nature.com/scientificreports www.nature.com/scientificreports/ for cell viability and transfection efficiency, a pCMV-Renilla luciferase (Rluc) construct was also transfected. The fold induction of Fluc, measured as the ratio of Fluc activity to Rluc activity, was employed to test the ability of EGR2 (wild-type or p.Asp411Gly) to activate the PMP22 enhancer.
These assays demonstrated that wild-type EGR2 was able to increase expression of Fluc (~110 fold increase over the empty vector; Fig. 4) consistent with previous reports that the PMP22 enhancer includes an EGR2 response element. In contrast, p.Asp411Gly EGR2 induced Fluc activity at a significantly lower level (~40 fold increase over the empty vector p = 2.1 × 10 −13 compared to wild-type; Fig. 4) indicating that the p.Asp411Gly variant reduces the capacity for EGR2 to appropriately activate the PMP22 enhancer. These results reveal a functional effect of p.Asp411Gly EGR2 and support the pathogenicity of this variant.

Discussion
The spectrum of peripheral neuropathy associated with disease-causing EGR2 variants ranges from severe early onset DSN and CHN, to later onset demyelinating CMT1 and axonal CMT2. Using WES and functional testing we have determined that a de novo missense EGR2 variant, c.1232A > G p.Asp411Gly, causes a severe and early onset DSN phenotype by reducing the capacity for EGR2 to function as a transcription factor. This variant can be classified as 'pathogenic' according to guidelines determined by the American College of Medical Genetics and Genomics (PS2, PS3, PM1, PM2, PP2, PP3) 33 . This pathogenic variant affects the third zinc-finger domain of EGR2, which is highly conserved among species and is a DNA-binding motif crucial for its role as a transcription factor in the regulation of myelin genes 34 . All reported pathogenic variants are within these three zinc-finger regions except for p.Ile268Asn, which causes recessive CHN and is within an inhibitory domain (Table 3) 4 .
Functional testing conducted here with luciferase-based reporter assays demonstrated that p.Asp411Gly EGR2 activates transcription less effectively than wild-type EGR2, similar to what has been previously reported for pathogenic EGR2 variants located in the zinc-finger domains 13,15,34,35 . EGR2 function is crucial for myelination of the peripheral nervous system, and Krox20 −/− (murine EGR2 orthologue) mice demonstrated arrested   www.nature.com/scientificreports www.nature.com/scientificreports/ differentiation of Schwann cells at an early stage 17 . Interestingly, heterozygous Krox20 +/− mice are phenotypically normal 17 , yet all reported pathogenic EGR2 variants in the zinc-finger domains cause peripheral neuropathy in a heterozygous state (Table 3) 13 . This is likely explained by the finding that in addition to pathogenic EGR2 variants resulting in a decrease or loss of EGR2 function, it has been shown that pathogenic EGR2 variants in all three zinc-finger domains (p.Arg359Trp, and p.Ser382Arg + p.Asp383Tyr, and p.Arg409Trp) result in the dominant-negative inhibition of wild-type EGR2 15,18 and SOX10 18 . These results are consistent with the reduction in transcriptional activation we observed in our luciferase assays with p.Asp411Gly EGR2.
The severity and onset of peripheral neuropathy caused by EGR2 variants is highly heterogenous, and it is yet to be determined why this broad phenotypic variability exists. This variation is exemplified by the p.Arg359Trp variant, which causes demyelinating CMT1D 3 , DSN 5,36 , and CHN 5 . This is also true for different variants of the amino acid adjacent to the p.Asp411Gly variant we have reported here, where p.Glu412Gly causes both demyelinating CMT1D 12 and axonal CMT 14 , and p.Glu412Lys causes DSN 35 . Given these observations, it is possible that additional genetic factors exist which modify the phenotype of EGR2 variant, such as those recently reported for CMT1A 37 .
Our patient had findings on brain MRI and evoked potentials that may be consistent with multiple sclerosis, however it was determined that he did not fulfil diagnostic criteria. White matter lesions have been reported in some patients with CMT [38][39][40][41] , and the proband had no contrast enhancing lesions suggesting active demyelinating disease in both brain MRIs conducted. In CMTX1, which is caused by pathogenic variants in GJB1, white matter lesions were reported in 7.5% of a large cohort 42 . Given that EGR2 mediates transcription of GJB1, it is possible that the reduced efficiency of p.Asp411Gly EGR2 in activating transcription resulting in a phenotype similar to that seen in some CMTX1 patients. Additionally, the FMH phenotype may be a chance association, or due to another genetic cause that has not been identified, as there were no variants in hemiplegic migraine genes that segregated with the phenotype. White matter lesions have also been reported in patients with migraines and patients with sporadic hemiplegic migraine [43][44][45][46] , further complicating the phenotype seen in our proband.
In this study we report a novel de novo EGR2 variant, c.1232A > G p.Asp411Gly, as the cause of severe, early-onset CMT3 (DSN). We provide robust functional data to support the deleterious functional effect of this mutation. The phenotype is complicated by clinical features of white matter lesions and FMH which may be a chance association, although it would be important to consider the possibility of central nervous system involvement in future cases of EGR2-related neuropathies.

Methods
Subjects. The proband and three family members were recruited and informed consent was obtained for this study using protocols approved by the Sydney Local Health District Human Ethics Research Committee (SLHD HERC). All experiments were performed in accordance with relevant guidelines and regulations from SLHD HREC. Genomic DNA was extracted from peripheral blood using the PureGene Kit (Qiagen) following manufacturer's instructions.
Whole-exome sequencing (WES). WES was performed on genomic DNA (2.5 µg) in the proband and was outsourced to Macrogen (South Korea) as previously described 19 . Wild-type and p.Asp411Gly EGR2 were evaluated for regulatory activity on a previously reported EGR2 response element at the PMP22 locus. Constructs (conditions indicated along the bottom) were transfected into cultured Schwann (S16) cells and tested for activity in luciferase assays normalized to an empty vector containing no genomic insert (far left column). Wild-type EGR2 and p.Asp411Gly EGR2 were each tested using two independently generated expression constructs (A and B) and each condition was tested in at least 24 biological replicates. The fold induction of Fluc activity is indicated along the y-axis and error bars indicate standard deviations. Statistical significance (p-value) was assessed using a two-tailed Student's T-test. Cell culture, transfection, and luciferase assays. Cultured rat Schwann cells (S16) 47 were grown in Dulbecco's Modified Eagle's Medium (DMEM) with 10% (v/v) fetal bovine serum (Gibco), 2 mM L-glutamine (Gibco), 50 U/mL penicillin, and 50 g/mL streptomycin. For luciferase assays, ~1 × 10 4 S16 cells were plated in each well of a 96-well untreated cell culture plate and grown overnight at 37 °C with 5% CO 2 . After 24 h, the cells were transfected with the experimental constructs using Lipofectamine 2000 (Invitrogen) diluted 1:100 in OptiMem (Life Technologies). Cells in each well received 200 ng of an expression construct: either empty pEIB (expression vector containing a Fluc reporter gene with no genomic insert) 32 or the pEIB-Forward vector with a PMP22 enhancer 31 upstream of a Fluc reporter gene. S16 cells were co-transfected with 100 ng per well of wild-type or mutant EGR2 constructs in addition to the PMP22 enhancer construct. For an internal control for transfection efficiency, 2 ng of a pCMV-Rluc construct were transfected into cells in all wells. Each experimental construct was diluted in OptiMem and incubated with an equal volume of Lipofectamine 2000 in OptiMem for  www.nature.com/scientificreports www.nature.com/scientificreports/ 20 min at room temperature before being applied to cells. Cells were incubated with transfection reagents for 4 h at 37 °C with 5% CO 2 , at which point the transfection reagents were removed and replaced with standard growth medium (above) and allowed to grow for 48 h at 37 °C with 5% CO 2 .
After 48 h, growth medium was removed, and cells were washed with 1X PBS. The cells were then lysed for 1 h at room temperature in 1X Passive Lysis Buffer (Promega), and 10 uL of lysate from each well was transferred into a white polystyrene 96-well plate (Corning). A Dual Luciferase Assay was performed using a Glomax Multi-Detection System and a Dual Luciferase Reporter 1000 Assay System kit (Promega) to determine Fluc and Rluc activities.
The ratio of Fluc to RLuc activity in each well was calculated, and ratios from individual wells were normalized relative to the average Fluc to Rluc ratio from the empty pEIB vector. The mean normalized ratio is a readout for the fold induction of Fluc by each experimental construct, and this is shown with standard deviation in the figure. Each variant was tested with two independently generated constructs (A and B in the figure) to rule out expression changes due to variation in the expression construct backbone, and each construct was tested in at least 24 individual experiments. Statistical significance was assessed using a two-tailed Student's T-test.

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.