Article Text
Abstract
Background Ichthyosis and neurological involvement occur in relatively few known Mendelian disorders caused by mutations in genes relevant both for epidermis and neural function.
Objectives To identify the cause of a similar phenotype of ichthyotic keratoderma, spasticity, mild hypomyelination (on MRI) and dysmorphic features (IKSHD) observed in two unrelated paediatric probands without family history of disease.
Methods Whole exome sequencing was performed in both patients. The functional effect of prioritised variant in ELOVL1 (very-long-chain fatty acids (VLCFAs) elongase) was analysed by VLCFA profiling by gas chromatography–mass spectrometry in stably transfected HEK2932 cells and in cultured patient’s fibroblasts.
Results Probands shared novel heterozygous ELOVL1 p.Ser165Phe mutation (de novo in one family, while in the other family, father could not be tested). In transfected cells p.Ser165Phe: (1) reduced levels of FAs C24:0-C28:0 and C26:1 with the most pronounced effect for C26:0 (P=7.8×10−6 vs HEK293 cells with wild type (wt) construct, no difference vs naïve HEK293) and (2) increased levels of C20:0 and C22:0 (P=6.3×10−7, P=1.2×10−5, for C20:0 and C22:0, respectively, comparison vs HEK293 cells with wt construct; P=2.2×10−7, P=1.9×10−4, respectively, comparison vs naïve HEK293). In skin fibroblasts, there was decrease of C26:1 (P=0.014), C28:0 (P=0.001) and increase of C20:0 (P=0.033) in the patient versus controls. There was a strong correlation (r=0.92, P=0.008) between the FAs profile of patient’s fibroblasts and that of p.Ser165Phe transfected HEK293 cells. Serum levels of C20:0–C26:0 FAs were normal, but the C24:0/C22:0 ratio was decreased.
Conclusion The ELOVL1 p.Ser165Phe mutation is a likely cause of IKSHD.
- VLCFA
- ELOVL1
- skin disease
- neurological disease
- de novo mutation
Statistics from Altmetric.com
Introduction
Inherited ichthyosis occurs in a group of Mendelian disorders caused by mutations of relatively large number of genes affecting epidermal function.1 Ichthyosis may be syndromic; that is, it coexists with extracutaneous involvement, for example, neurological symptoms (Sjogren-Larsson syndrome (SLS; MIM: 270200), myopathy (Chanarin-Dorfman syndrome (CDS; MIM: 275630) or sensorineural hearing impairment (diseases caused by dominant GJB2 mutations MIM: 148210).
Hypomyelinating disorders have been recently recognised as diseases with varying neurological manifestations that display reduced myelin content on brain MRI. At present, this group includes 12 disorders associated with mutations in 13 genes with diverse functions as well as a chromosomal aberration (18q-syndrome).2
Recently, Aldahmesh et al described a neurological disease with delayed myelination accompanied by ichthyosis (ichthyosis, spastic quadriplegia and mental retardation (ISQMR) MIM: 614457). ISQMR was shown to be caused by recessive mutations of elongase ELOVL4, thus emphasising the role of undisturbed synthesis of very-long-chain fatty acids (VLCFAs) for the proper development of both myelin and epidermis.3
The purpose of our study was to search for the cause of disease ascertained in two unrelated paediatric probands with overlapping dysmorphic features, pronounced ichthyotic keratoderma and an early onset progressive neurological disease with mild hypomyelination.
Patients
Patient 1 was a boy born after 40 weeks of non-complicated pregnancy with weight 3650 g (50th–75th percentile), length 57 cm (>95th percentile), occipital frontal circumference (OFC) 36 cm (50th–75th percentile) and 10 points in Apgar score. During the first 6 months, his development was normal. At 7 months of age, the skin changed colour into orange, became very dry and his psychomotor development slowed. Spasticity and dryness of the skin were progressing, and ichthyosis was diagnosed at 1.5 years. Histopathological examination at 2 years revealed reticular hyperkeratosis, focal parakeratosis and focal acanthosis of epidermis with ‘folded straw’ hairs. Subsequently, slowed speech, photophobia, bilateral high frequency sensorineural hearing loss (70% at 3500 Hz) and narrowing of the visual field (30% of normal value) appeared. Since the age of 8 months, there was intermittent rotary nystagmus that became fixed at approximately 10 years of age. Abnormal Babinski sign (ie, persisting beyond 18 months of age) has been present on all exams. A cradle cap on patient’s head occurred in first 3 years of life. There were mild dysmorphic features (figure 1).
The initially considered diagnoses of SLS, CDS or GJB2 associated diseases were excluded by sequencing of the respective genes. Furthermore, PLP1 sequencing (prompted by rotary nystagmus) excluded Paelizaeus-Merzbacher disease, while Refsum disease and biotinidase deficiency were excluded by normal concentration of plasma phytanic acid and biotinidase, respectively. Results of organic acids analysis by gas chromatography–mass spectrometry (GC–MS) and amninoacids/acylcarnitin profiling by tandem mass spectrometry (tandem MS) were normal.
At 12 years, the patient was a wheelchair user with spastic paraplegia of lower limbs, discrete signs of ataxia (dysmetria, slowed speech and nystagmus) and pale optic discs. He complained of dryness in mouth Content and pain during swallowing. Head circumference, echocardiography and ultrasonography of abdomen, EEG and renal function were normal. Mental development is normal; he follows regular school curriculum, excels in mathematics and enjoys playing chess.
Patient 2 attracted our attention due to dysmorphic features similar to patient 1 (figure 1). His family was unrelated to patient 1 and lived in a different geographic region of Poland. He was born after 39 weeks of non-complicated pregnancy (weight 3340 g, >50th percentile; length 53 cm, ~95th percentile; OFC 34 cm, >25th percentile) and Apgar score of 10. Audiological examination after birth was normal. From the beginning, the skin was very dry; cow milk’s allergy was suspected. At the age of 9 months, ichthyosis was diagnosed. He began to sit with help at 6 months, he did not stand and had spasticity of legs. At 1 year, neurological examination revealed rotating nystagmus, spasticity of legs, decreased tonus of head-corpus axe and exaggerated deep tendon reflexes. Eye fundus, organic acids profile by GC–MS, aminoacids and acylocarnitine profile by tandem MS were all normal. At 3 years 4 months, he walked with help or in orthoses, had spastic paraplegia of legs and discrete signs of ataxia (with discrete dysarthria – slowed speech, nystagmus and mild tremor). Intellectual development was good (at 4–5 years Columbia Mental Maturity Scale scores between 25th and 75th percentile). Head circumference and height were normal (25th–50th percentile, 10th–25th percentile, respectively). On dermatological examination, he presented dry, hyperkeratotic skin with ichthyotic scales without inflammation (figure 1). Skin biopsy from the knee area showed hyperkeratosis with focal parakeratosis, preserved granular layer, acanthosis and mixed infiltrations in the papillary dermis (figure 1). There was bilateral high frequency sensorineural hearing loss (70 dB at 6 kHz).
The first brain MRIs performed at the age of 15 (patient 1) and 13 (patient 2) months showed hypomyelination of white matter in both infratentorial and supratentorial regions (online supplementary figure S1). On T2- weighted images normally myelinated white matter is hypointense and a severe myelin deficit – hyperintense.2 Thus, the hypomelination observed in both p atients was relatively mild. MRIs performed at 4 years showed similar pattern of incomplete myelination; additionally signal intensity of the corpus callosum changed into less hypointense in both probands (figure 2). The latest MRI was performed at 13 years in patient 1. During 12 years of observation, there was slight atrophic progression involving cerebral hemispheres and corpus callosum with appearance of abnormal signal hyperintensity of the posterior limb of the internal capsule (PLIC; online supplementary figure S2). Comparing the patients’ MRI results at similar age (figure 2, and online supplementary figure S1) revealed similar findings indicating globally impaired myelination, both supratentorially and infratentorially. Overall, the MRI results suggest that both patients have the same disorder with moderate–mild myelin deficit. Findings of brain MRIs are summarised in online supplementary table S1.
Supplementary file 1
In both patients, nerve conduction studies of the lower and upper limb nerves were normal both proximally and distally.
All individuals provided informed consent after the possible consequences of the study were explained, in accordance with the Declaration of Helsinki.
Methods
Whole exome sequencing (WES) was performed on HiSeq 1500 (whole exome sequencing, online supplementary material). In silico analysis of ELOVL1 p.Ser165Phe mutation effect on protein structure was done using all atom model of canonical splice variant of human ELOVL1 obtained by homology modelling (computational modelling, online supplementary material). In order to study the effect of the ELOVL1 mutation on cellular VLCFAs, wild type (wt) and mutant cDNA were cloned into pKK-FLAG-TEV and pKK-TEV-EGFP under a Tet inducible promoter using SLIC method4 (see also obtaining stable Flip-In Hek293 cell lines, online supplementary material). Expression of ELOVL1 protein in transfected cells was verified by immunofluorescence (immunofluorescence and laser scanning confocal microscopy, online supplementary material) and by Western blot (WB) using HPA056557 antibody (Sigma-Aldrich; Western blot analysis, online supplementary material). The same WB protocol was used for studying ELOVL1 expression in cultured fibroblasts. Analysis of ELOVL1 mRNA in fibroblasts was performed by real-time PCR (in order to compare total amount of transcript between patient and control) and by Sanger sequencing of PCR product (generated from cDNA) encompassing mutation site (in order to compare transcription of the wt and mutated allele (analysis of mRNA, online supplementary material). VLCFA profiling of transfected cells and cultured fibroblasts was performed by GC–MC using an established protocol5 (see also analysis of fatty acids, online supplementary material). Search for of abnormal lipid storage in studied cells was performed by transmission electron microscopy (electron microscopy, online supplementary material).
Results
WES analysis
First, in each patient, we searched for known pathogenic variants (ie, those annotated in the Human Gene Mutation Database (HGMD)) as well as potentially pathogenic variants remaining after filtering by function (affecting aminoacid sequence or splice site) and frequency in Genome Aggregation Database (GnomAD) and an inhouse database of >500 exomes. In particular, we considered 19 variants in 13 genes in patient 1 and 21 variants in 16 genes in patient 2 (online supplementary table S4). Since they were all rejected as candidates, we next searched for rare (<1%) variants located in the same gene in both patients. In this analysis, ELOVL1 p.Ser165Phe (chr1:43830119 G>A, NM_001256399.1: c.494C>T) heterozygous mutation emerged as sole plausible candidate. The presence of this variant in both patients was confirmed by Sanger sequencing. The typing of available family members showed that the mutation’s origin was consistent with de novo events (the variant absent in mother and sister of patient 1 and both parents and the brother of patient 2, figure 3A,B respectively). The father of patient 1 or his relatives could not be tested. In the family of patient 2, paternity was confirmed analysing 17 hypervariable Short Tandem Repeats (STRs) (AmpFLSTR NGM SElect PCR Amplification Kit, ThermoFisher Scientific, odds in favour of paternity >106).
The c.494C>T ELOVL1 mutation has been deposited in GeneBank (BankIt2077138 ELOVL1 MG766453).
When specifically questioned, both families confirmed that they were not related. The relatedness between patients was also excluded by calculation of kinship coefficient from WES data using KING package6 (result: −0.4647, no relationship) and by finding that patients did not share any ultra rare variants (ie, variants with frequency 0 in available databases) other than ELOVL1 p.Ser165Phe.
The ELOVL1 p.Ser165Phe variant was absent from the Exome Aggregation Consortium (ExAC) database and gnomAD, which include 123 136 exomes and 15 496 genomes as well from our inhouse database of >500 Polish exomes. The analysis of ELOVL1 homologues from different organisms showed that p.Ser165 is highly conserved being present in all the analysed species (n=80, figure 3D). The p.Ser165Phe mutation is located in the region that may undergo myristoylation7 (figure 3D), although experimental data are necessary to verify this prediction.
ELOVL1 computational modelling
ELOVL1 has not been crystallised, but we generated a protein structure prediction using homology modelling (figure 3C). ELOVL1 was predicted to exhibit 7 transmembrane alpha helices with the N-terminus in the ER lumen and the C-terminus in the cytosol consistent with experimental data for yeast Sur4 (Elovl3) protein.8 The mutation site is at the beginning of H5 helix, close to L45 loop region. Molecular dynamics modelling suggested that the p.Ser165Phe mutation causes rearrangement of protein structure, in particular decrease of H4-H5 distance (online supplementary figure S3), and relocation of hydrophobic Phe residue deeper into the membrane region, which may affect catalytic activity.
Functional tests
In order to study the functional effect of ELOVL1 p.Ser165Phe, we stably expressed constructs of mutated and wt protein tagged with eGFP or FLAG in HEK293 cells under control of tetracycline-responsive promoter (see online supplementary material for the advantages of stable vstransient transfection). The following lines of transformed cells were generated: 140: WT-eGFP, 141: mutant-eGFP, 142: WT-FLAG and 143: mutant-FLAG (online supplementary materials). The expression of ELOVL1 in HEK293 cells was verified by WB (online supplementary figure S4A) and immunofluorescence (online supplementary figure S5). We have observed no major effect on cell viability in either cell line. Using GC–MS, we analysed the profile of C20:0-C28:0 as well as C24:1 and C26:1 fatty acids (FAs) in the four lines of transformed cells and in untransformed HEK293 cells (figure 4). We found that levels of C26:0 were distinctly higher (approximately sixfold) in cells transformed with wt ELOVL1 compared with both the cells transformed with the p.Ser165Phe mutant (P=7.8×10−6) and untransformed cells (P=1.3×10−9). Similar but less pronounced differences/trends were observed for C26:1 (P=0.001 and P=3.6×10−5, respectively), C24:0 (P=0.021 and P=0.02, respectively) and C28:0 (NS, P=2.0×10−4, respectively). For all these FAs (C24:0-C28:0 and C26:1), the levels in HEK293 cells transfected with the p.Ser165Phe mutant were similar as in untransformed cells. In HEK293 transfected with wt ELOVL1, we also found a moderate decrease of C20:0 and C22:0 versus naïve HEK293 (P=0.04 and P=0.01, for C20:0 and C22:0, respectively, figure 4), which we interpreted as secondary to enhanced synthesis of longer VLCFAs such as C24:0-C28:0.
These result show that the p.Ser165Phe mutation abolishes/diminishes the enzymatic activity of ELOVL1 necessary for synthesis of C24:0-C28:0 and C26:1. Interestingly, in HEK293 cells transfected with the p.Ser165Phe construct, there was a strong increase of C20:0 and C22:0 versus the cells transfected with wt ELOVL1 (P=6.3×10−7, P=1.2×10−5, for C20:0 and C22:0, respectively) and, interestingly, in comparison with naïve HEK293 cells (P=2.2×10−7, P=1.9×10−4, for C20:0 and C22:0, respectively, figure 4). These result suggest that the p.Ser165Phe mutation while abolishing ELOVL1 activity in synthesis of C26:0-28:0 and C26:1 FAs enhances or at least does not affect the enzyme’s activity in synthesis of C20:0-C22:0 (if the mutated construct encoded protein devoid of function it should not influence FA profile in any way and there should be no difference between transfected and naïve cells).
In parallel to the studies in HEK293 cells, we studied fibroblasts obtained from a skin biopsy in patient 2 and control human fibroblasts (n=4). The expression of ELOVL1 in patient’s and control fibroblasts was similar both at the level of mRNA (assessed by qPCR, mean ΔCt=4.2±0.67 and 4.0±0.22, respectively, P=0.31) and protein (assessed by WB, online supplementary figure S4B). We also found that in patient’s fibroblasts, the ELOVL1 mutation was expressed at the mRNA level (online supplementary figure S4C). VLCFA analysis showed that patient’s fibroblasts had significantly lower level of C26:1 (P=0.014) and C28:0 (P=0.001) as well as significantly higher level of C20:0 (P=0.033) in comparison with controls (figure 5). Whereas the levels of other tested FAs did not show statistically significant differences between patient and control fibroblast, there was a correlation (r=0.92, P=0.008) between the trends found in patient versus control fibroblasts for individual FAs with those observed in the HEK293 cells transfected with ELOVL1 p.Ser165Phe mutant versus the wt ELOVL1 construct (figure 6). As shown in figure 6, both in fibroblasts and HEK293 cells, the presence of ELOVL1 p.Ser165Phe decreased the concentration of longest FAs (C24:0-C28:0, C26:1) and increased the levels of those with shorter chains (C20:0-C22:0).
Analysis of fatty acids
We also examined C20:0–C26:0 FAs concentration in patient’s serum (online supplementary table S2). Two independent tests were performed in each patient. In three out of the four tests performed the C24:0/C22:0 ratio was below the reference value and in one it was borderline low. Thus, whereas individual VLCFA levels in serum may not be diagnostic of the ELOVL1 p.Ser165Phe syndrome, a suspicion of the disease may come from lowered C24:0/C22:0 ratio, one of the major biomarkers for peroxisomal diseases.
Discussion
The progressing dermatological symptoms leading to a relatively late (ie, at the age of 9–18 months) onset of ichthyosis, central nervous system involvement with spastic paraplegia of legs and a distinctive sign of rotary nystagmus in infancy in conjunction with dysmorphic features present in two unrelated patients allowed a clinical suspicion of a novel genetic syndrome. Further genetic and functional studies showed that this syndrome of ichthyotic keratoderma, spasticity, mild hypomyelination and dysmorphic features (IKSHD) is caused by the p.Ser165Phe mutation in the ELOVL1 gene.
ELOVL1 is constitutively expressed9 with particular abundance in skin, kidney, stomach lung and highly myelinated parts of central nervous system (CNS).10 11 ELOVL1 is an elongase, that is, an enzyme catalysing the rate limiting step in synthesis of VLCFAs. In humans, there are seven elongases (ELOVL1-7) that differ in substrate specificity and tissue expression.12 ELOVL1 functions in the synthesis of saturated and unsaturated FAs with 20 and more carbons (C20:0, C22:0 and so on) with particularly strong activity in synthesis of C24:0-C28:0 and C26:1.12–14
Elovl1 was cloned in mice by Tvrdik et al, who also showed that it played a role in biosynthesis of C26 FA and sphingolipids.11 In humans, ELOVL1 was suggested to be the single elongase synthesising C26:0.13 In fibroblasts from patients with X-linked adrenoleukodystrophy, ELOVL1 knockdown reduced elongation of C22:0 to C26:0 causing a pronounced lowering of C26:0 levels.13 The role of ELOVL1 in synthesis of C24:0 and C26:0 is also evident from results of its pharmacological inhibition.15 Elovl1 knockout mice die shortly after birth due to epidermal barrier defect.16
The importance of the elongases for neurological and/or epidermal barrier function is underscored by emerging descriptions of their genetic defects. Autosomal dominant mutations in ELOVL4 cause Stargardt disease 3 (MIM 600110)17 and spinocerebellar ataxia 34 (MIM: 133190),18 whereas recessive ELOVL4 mutations cause ISQMR (MIM: 614457).3 Autosomal dominant mutations in ELOVL5 cause spinocerebellar ataxia 38 (MIM: 615957).19
ELOVL1-associated IKSHD described here differs from other diseases affecting both skin and nervous system. The difference between IKSHD and both SLS and ISQMR is the lack of mental impairment in IKSHD; from CDS and Keratitis-ichthyosis-deafness (KID) syndromes, IKSHD is distinguished by absence of Jordan’s bodies and hepatosplenomegaly (both typical for CDS) and lack of corneal abnormalities (found in KID).
At present, it is not clear how the p.Ser165Phe ELOVL1 mutation causes disease. Lack of activity of mutated enzyme towards synthesis of C26:0-28:0 and C26:1 is compatible with pathogenesis due to shortage of these (and perhaps longer) VLCFAs. It may also be speculated that the mutation has a greater impact on VLCFA levels in brain and skin than fibroblasts or plasma. However, according to ExAC database, the evidence that ELOVL1 is a haploinsufficient gene is only moderate: probability of being loss-of-function intolerant(PLI)=0.22, three loss-of-function (LOF) variants observed, 11.2 expected. Furthermore, it is unclear why both probands should have the same missense mutation, especially at a non-CpG nucleotide, if its only effect was LOF.
Whereas ELOVL1 haploinsufficiency cannot be fully excluded as the disease mechanism, the data from FAs profiling of transfected HEK293 cell and patient’s fibroblast suggest that the p.Ser165Phe mutation results also in a gain-of-function of the enzyme manifested by increase of C20:0 and C22:0. Whereas C20:0 and C22:0 are not known to be toxic, we cannot exclude that mutated ELOVL1 causes accumulation of some other lipid compound(s) that may be detrimental to neuronal and epidermal cells.
In conclusion, our genetic and functional studies indicate that the disease that we ascertained in two unrelated probands and called IKSHD is caused by dominantly acting ELOVL1 p.Ser165Phe mutation.
Acknowledgments
We would like to thank Michał Korostyński (Intelliseq) for his help in calculation of kinship coefficient.
References
Footnotes
AK-K and MR contributed equally.
Contributors Conceptualisation: RP, MR and AK-K. Methodology: RP, WN, RJ, KK-K and JG. Software: WN and RJ. Validation: AP, PG and AW. Formal analysis: RP, TJS, WN, RJ, TS and AM. Investigation: RP, MR, AP, PG, AW, JK, TJS, WN, RJ, AC, KK, JG, TS, AM, DŚ, AK-K, EJ-S, EJ,CK, TK and EO. Resources: WN, RJ, DS and EJ. Writing: original draft preparation: RP, MR, TJS, AK-K and EJ. Writing: review and editing: RP, MR, TJS, WN, RJ, TS, AM, AK-K and EJ. Visualisation: RP, MR, WN, RJ and EJ. Supervision: RP and AD. Project administration: RP. Funding acquisition: RP and AD.
Funding The study was supported by the National Science Centre (NCN) Poland grant 2013/11/B/NZ7/04944 to RP. The work in the AD laboratory was supported by Foundation for Polish Science grant TEAM/2016-1/3. Some of the experiments were carried out with the use of CePT infrastructure (Innovative economy 2007–13, Agreement POIG.02.02.00-14-024/08-00). Calculations were possible due to ICNT UMK infrastructure. Support by Polish Ministry for Science and Higher Education Grant 0003/ID3/2016/64 (Ideas Plus II) is acknowledged (RJ).
Competing interests None declared.
Patient consent Guardian consent obtained.
Ethics approval The study protocol was approved by the Ethical Committee at Warsaw Medical University.
Provenance and peer review Not commissioned; externally peer reviewed.