Article Text

Download PDFPDF

Original article
De novo mutation in ELOVL1 causes ichthyosis, acanthosis nigricans, hypomyelination, spastic paraplegia, high frequency deafness and optic atrophy
Free
  1. Noomi Mueller1,2,
  2. Takayuki Sassa3,
  3. Susanne Morales-Gonzalez1,
  4. Joanna Schneider2,4,
  5. Daniel J Salchow5,
  6. Dominik Seelow1,4,
  7. Ellen Knierim1,2,
  8. Werner Stenzel6,
  9. Akio Kihara3,
  10. Markus Schuelke1,2
  1. 1 NeuroCure Clinical Research Center, Charité–Universitätsmedizin Berlin, Berlin, Germany
  2. 2 Department of Neuropediatrics, Charité–Universitätsmedizin Berlin, Berlin, Germany
  3. 3 Faculty of Pharmaceutical Sciences, Laboratory of Biochemistry, Hokkaido University, Sapporo, Japan
  4. 4 Berlin Institute of Health, Berlin, Germany
  5. 5 Department of Ophthalmology, Charité–Universitätsmedizin Berlin, Berlin, Germany
  6. 6 Institute of Neuropathology, Charité–Universitätsmedizin Berlin, Berlin, Germany
  1. Correspondence to Professor Akio Kihara, Faculty of Pharmaceutical Sciences, Laboratory of Biochemistry, Hokkaido University, Sapporo 060-0812, Japan; kihara{at}pharm.hokudai.ac.jp and Professor Markus Schuelke, Department of Neuropediatrics, Charité Universitätsmedizin Berlin, Berlin D-13353, Germany; markus.schuelke{at}charite.de

Abstract

Background Very long-chain fatty acids (VLCFAs) are essential for functioning of biological membranes. ELOVL fatty acid elongase 1 catalyses elongation of saturated and monounsaturated C22-C26-VLCFAs. We studied two patients with a dominant ELOVL1 mutation. Independently, Kutkowska-Kaźmierczak et al. had investigated the same patients and found the same mutation. We extended our study towards additional biochemical, functional, and therapeutic aspects.

Methods We did mutation screening by whole exome sequencing. RNA-sequencing was performed in patient and control fibroblasts. Ceramide and sphingomyelin levels were measured by LC-MS/MS. ELOVL1 activity was determined by a stable isotope-labelled [13C]malonyl-CoA elongation assay. ELOVL1 expression patterns were investigated by immunofluorescence, in situ hybridisation and RT-qPCR. As treatment option, we investigated VLCFA loading of fibroblasts.

Results Both patients carried an identical heterozygous de novo ELOVL1 mutation (c.494C>T, NM_001256399; p.S165F) not deriving from a founder allele. Patients suffered from epidermal hyperproliferation and increased keratinisation (ichthyosis). Hypomyelination of the central white matter explained spastic paraplegia and central nystagmus, while optic atrophy was causative for reduction of peripheral vision and visual acuity. The mutation abrogated ELOVL1 enzymatic activity and reduced ≥C24 ceramides and sphingomyelins in patient cells. Fibroblast loading with C22:0-VLCFAs increased C24:0-ceramides and sphingomyelins. We found competitive inhibition for ceramide and sphingomyelin synthesis between saturated and monounsaturated VLCFAs. Transcriptome analysis revealed upregulation of modules involved in epidermal development and keratinisation, and downregulation of genes for neurodevelopment, myelination, and synaptogenesis. Many regulated genes carried consensus proliferator-activated receptor (PPAR)α and PPARγ binding motifs in their 5’-regions.

Conclusion A dominant ELOVL1 mutation causes a neuro-ichthyotic disorder possibly amenable to treatment with PPAR-modulating drugs.

  • hypomyelination
  • very long-chain fatty acids
  • fatty acid synthesis
  • ichthyosis

Statistics from Altmetric.com

Introduction

Very long-chain fatty acids (VLCFAs) are fatty acids (FAs) with carbon chain lengths of ≥C21.1 2 Depending on the number of double bonds, VLCFAs are divided into saturated, monounsaturated, and polyunsaturated FAs. VLCFA are synthesised from diet-derived C11-C20 precursors or de novo as C16-FAs by cytosolic FA synthase.2 On activation to acyl-CoA, VLCFAs are elongated by four interacting enzymes, the elongase complex, in the endoplasmic reticulum (ER).3 The first and rate-limiting step, the condensation of malonyl-CoA+acyl CoA=>3 ketoacyl-CoA is catalysed by elongases. Seven members of the mammalian FA elongase family (ELongation Of VLCFAs proteins; ELOVL1–7) have distinct substrate preferences4 and perform repetitive rounds of the elongation cycle until a defined VLCFA length has been achieved.3 ELOVL1 is particularly active towards saturated and monounsaturated C20-C24-VLCFAs,5 thus responsible for the synthesis of C22-C26-VLCFAs.

VLCFAs are essential for biological processes that cannot be maintained by shorter FAs, such as formation of myelin, photoreceptors, and the epidermal and tear film permeability barrier.1 2 6 7 Different tissues require VLCFAs of different length and their synthetic machinery is regulated on the transcriptome level. ELOVL1, 5 and 6 are expressed ubiquitously, while ELOVL2–4 and 7 are tissue-specific.4 VLCFAs sparsely operate as free molecules, but rather as components of sphingolipids (ceramide and sphingomyelin) and glycerophospholipids.8 Ceramide is composed of sphingosine plus FA and sphingomyelin of ceramide plus phosphocholine. Ceramides with ≥C26-FAs are principal components of the epidermal permeability barrier.9 Elovl1 knockout mice died neonatally due to skin barrier defects, exhibiting decreased epidermal ceramide (≥C26-FA) and sphingomyelin (≥C24-FA) levels, likewise in brain and various other tissues.1 ELOVL1-generated VLCFAs are essential for myelin formation and maintenance. While neurons preferentially synthesise C18-sphingolipids,10 myelin-producing cells (oligodendrocytes and Schwann cells) mainly use C24-sphingolipids.6 11

Here, we report on clinical phenotype and the biochemical and molecular consequences of a heterozygous ELOVL1 de novo mutation in two children. During preparation of this manuscript, we learnt about a publication by Kutkowska-Kaźmierczak et al.,12 who had independently investigated the same two families. It is not uncommon for patients with rare diseases to move internationally in search for diagnosis and therapy. Regarding the dominant de novo mutation in ELOVL1, we fully confirm their findings. Our study is complementary for understanding of the disease, investigating additional clinical, pathophysiological and molecular aspects. Since ceramides and sphingomyelins are the relevant building blocks made from VLCFAs, we examined their composition in skin as one of the primarily affected tissues. We directly tested mutant and wild-type ELOVL1 enzymatic activity using highly sensitive assays. Furthermore, we aimed to improve understanding of involved regulatory networks by transcriptome and tissue expression analysis, and to explore therapeutic approaches by VLCFA supplementation.

Materials and methods

The patients' parents provided written informed consent for the publication of photographs and all aspects of the study according to the Declaration of Helsinki.

For details of clinical and molecular genetics investigations (WES, virtual gene panel analysis, founder haplotype analysis, global mRNA-expression profiling, in situ hybridisation), cloning of constructs, biochemical investigations (ceramide and sphingomyelin analysis by LC-MS/MS, FA elongation assay, VLCFA treatment of fibroblasts), microscopy (immunomicroscopy and electron microscopy), and bioinformatic analysis (prediction of ELOVL1 secondary structure, identification of PPAR response elements), please refer to online supplementary methods.

Results

Case histories

The comparison of the clinical phenotype of both patients is depicted in online supplementary table S1.

Patient 1 (A:II.2)

The adolescent aged 15 years originates from Poland and was born to healthy non-consanguineous parents. A younger sister is healthy and no further family members were affected. His growth parameters were normal. Parents observed orange skin discolouration at 10 months of age. At 18 months, the boy presented with pruritic, dry, scaly, and thickened skin over the entire body, particularly over extensor surfaces of large joints, hands, and feet. He developed nuchal, axillary, and umbilical hyperpigmentation (acanthosis nigricans), with few additional hypopigmented skin areas. His fingernails were thickened, splitting into horizontal layers. Palms, soles, mucous membranes, and scalp were unaffected (figure 1B,D,E). Parents noticed an occasional strange body odour and hypohidrosis. Skin biopsy showed ichthyosis with thickening of epidermal stratum corneum and spinosum (figure 2G). Electron microscopy revealed numerous lysosomes filled with lipopigment or melanosomes (figure 2L,M). Motor development was delayed. Due to progressive lower limb spasticity (see online supplementary video 1), he was never able to walk and became wheelchair-dependent (figure 1A). In contrast, his arms remained strong to make him an elite competing athlete. He is socially well integrated and attends normal school with good performance. Progressive high frequency hearing deficit started at 7 years (figure 2I). His language has slowly become dysarthric and he mentioned intermittent stool incontinence and incomplete bladder voiding. Urodynamic investigation at 10 years showed a neurogenic bladder with detrusor-sphincter dyssynergia. At 14 years, neurological examination revealed hyperreflexia, upgoing plantar reflexes, contractures of the large joints, and equinovarus foot deformity. His arms were less severely affected, although with some contractures and muscle atrophy (figure 1C,E). During early childhood, parents noticed horizontal, high-frequency, low-amplitude nystagmus (see online supplementary video 1), pronounced light sensitivity, progressive visual problems in dim illumination, constriction of the visual field (figure 2H), high astigmatism of −5 dpt, and decreased visual acuity (0.1 at 15 years). Circular spectral-domain optical coherence tomography (SD-OCT) scanning of the peripapillary retina showed thinning of the retinal nerve fibre layer (RNFL) with a global thickness of 71 µm as a sign of optic atrophy. Retinal morphology (see online supplementary figure S2) and function (ERG, online supplementary figure S3) were normal. Fine motor coordination of his hands and sensory function were normal, as was neurography of peripheral motor and sensory nerves. Cranial MRIs (at 1, 4, 11, and 14 years of age, online supplementary figure S4) showed non-progressive T2-signal hyperintensities (hypomyelination) in the posterior limb of the internal capsule (PLIC) and in the optic radiation (OR) without contrast enhancement (figure 2C). Corpus callosum and temporal lobes were hypoplastic (figure 2C,E). Spinal MRI was normal (see online supplementary figure S5). Consistent with a central myelination defect, somatosensory and acoustic evoked potentials revealed brainstem and central conduction deficit. Concentrations of plasma FAs and VLCFAs were normal (see online supplementary table S2), as was saliva cortisol (morning sampling: 0.27–0.75 µg/dL; N 0.09–1.04 µg/dL). Normal serum lactate and alanine levels made a mitochondrial disease unlikely. Funduscopy, perimetry and hearing tests in his parents were normal.

Figure 1

Clinical phenotype. (A–E) Images of patient 1 (15 years), wheelchair-dependent but being able to move autonomously (A). Acanthosis nigricans in the armpit and at the upper arm (B). Spastic paraplegia with bilateral knee contractures of 45 degrees (C). Hyperkeratotic ichthyosis on the dorsum of the hand, especially over the extensor surfaces of the small finger joints (D), while sparing the palms; mild thenar atrophy (E). (F–K) Images of patient 2 (4 years). Passive sitting with spinal kyphosis and knee contractures of 30 degrees (F). Perioral hyperkeratotic-ichthyotic skin lesions (G). Acanthosis nigricans of the lower abdominal skin and patchy skin depigmentation (white spots) on the forearm (H). ‘Scissoring’ of the legs due to spasticity of the adductor muscles (I). Grossly thickened and coarsened skin relief over both knees (K).

Figure 2

Clinical investigations. The (A) T2-weighted and (B) FLAIR-dark-fluid weighted cMRT images of patient 2 (4 years) with areas of hypomyelination along the optic radiation (OR) and the posterior limb of the internal capsule (PLIC) without contrast enhancement. The (C) T2-weighted image of patient 1 (14 years) also with PLIC and OR hypomyelination. Both genu (GCC) and splenium (SCC) of the corpus callosum and the temporal lobes are hypoplastic. For comparison, (D) the image of an age-matched and sex-matched control. (E) Hypoplasia of the corpus callosum in patient 1 (open arrowheads). (F, G) H&E stained skin sections of patient 1 and control for comparison depict the four layers of the epidermis, of which the stratum corneum and spinosum are grossly broadened in the patient. sc, stratum corneum; ss, stratum spinosum; sg, stratum granulare; sb, stratum basale. In patient 1, the epidermal thickness is increased to 100–150 µm (N 30–50). (H) Perimetry in patient 1 (10 years) reveals pronounced visual field constriction to <30 degrees. (I) Pure-tone threshold audiometry of patient 1 with moderately severe hearing loss above >2000 Hz. (K) Optical coherence tomography (OCT) of patient 2. The circular peripapillary scan centred on the optic disc reveals thinning of the retinal nerve fibre layer nasally and inferiorly. The colour scheme represents age-matched percentiles for the normal range. T, temporal; S, superior; N, nasal; I, inferior. (L–N) Transmission electron microscopic images, (L) skin of patient 1 showing numerous accumulations of lipopigment or melanosomes (white arrowheads in the inset), partially being degraded in lysosomes, (M) cultured fibroblasts of patient 1, depicting lysosomes filled with lipopigment or melanosomes (white arrowheads) and (N) isolated lymphocytes of patient 2 that are loaded with numerous lysosomes.

Patient 2 (B:II.2)

The boy aged 5 years, also from Poland, was born to healthy, non-consanguineous parents. No other family member was affected. His growth parameters were normal. Interestingly, both families had contacted each other via an internet skin disease forum. In the neonatal period, parents noticed dry and thickened skin, mainly over knee joints. Palms, soles, mucous membranes, hair, and nails remained unaffected. He developed hyperpigmented and hypopigmented skin areas (figure 1G,H,K). Parents reported occasional strange body odour. Lower limb spasticity was noticed from 8 months onwards (see online supplementary video 2). He is able to walk with calipers and support, but generally is wheelchair-dependent. Hyperreflexia, muscle atrophy, pyramidal tract signs (see online supplementary video 2) and large joint contractures were more pronounced in the legs (figure 1F,I). The patient had dysarthria, but cognitive and language development was normal. A conjugate horizontal nystagmus was noticed at 12 months. Circular SD-OCT revealed sectoral thinning of the peripapillary RNFL nasally (41 µm) and inferiorly (36 µm), but a normal macula ((figure 2). He had a visual acuity of 0.3–0.4 and a high astigmatism of −4.5 dpt. High frequency hearing deficit >2000 Hz was diagnosed at 4 years. cMRI showed bilateral T2-hyperintensities in the white matter of the PLIC and OR (Figure 2B). Visually and acoustic evoked potentials confirmed a central conduction defect. Motor and sensory nerve conduction velocities were normal. Plasma VLCFAs (see online supplementary table S2), urine organic acids, and saliva cortisol levels (morning sampling: 0.24–0.53 µg/dL; N 0.09–1.04 µg/dL) were normal, as were serum lactate and alanine levels. Funduscopy, perimetry and hearing tests in his parents were normal.

Molecular genetics

In both patients, we found the identical heterozygous de novo ELOVL1-mutation, chr1:43830119G>A (GRCh37); c.494C>T, NM_001256399; p.(S165F), present on DNA and mRNA level and absent in parents and siblings (Figure 3A–C). Haplotype analysis excluded a common ancestral origin of the mutation (Figure 3D). Mutations in other genes known to be associated with characteristic symptoms were excluded by analysis of virtual gene panels (see online supplementary material).

Figure 3

Molecular genetic findings. (A) Pedigrees of both families. The genotype of each individual is plotted below the symbols. The healthy infant B.II:3 did not undergo any genetic testing. (B) Sequence electropherograms of patient 1 and both his parents verify a heterozygous de novo ELOVL1 mutation. Same was found for patient 2 (data not shown). (C) Segregation of the ELOVL1 c.494C>T mutation with the clinical phenotype by restriction enzyme analysis. The mutation removes a BanII restriction site. Only half of the PCR products are cleaved in the heterozygous patients. gDNA: 210 bp=>125+85 bp; cDNA 554 bp=>402+152 bp. Presence of the ELOVL1 variant on mRNA level excludes selective degradation of the mutant mRNA. (D) Haplotype analysis of SNPs around the mutant position at the ELOVL1 locus rules out a co-segregating element and thereby a potential founder mutation. Each dot represents the location of a SNP deviating from RefSeq GRCh37.p11 (hg19). (E) Multispecies alignment of the ELOVL1 orthologues showing high evolutionary conservation of Ser165 down to amphibians.

Global mRNA expression analysis

Analysis of two patients and four controls revealed n=44 genes to be upregulated and n=31 to be downregulated (see online supplementary table S3). Functional annotation clustering with DAVID V.6.7 (https://david.ncifcrf.gov/)13 revealed a highly significant upregulation of genes involved in epithelial differentiation (enrichment=4.55, Benjamini p=3.3E-4), epidermis development (Benjamini p=6.2E-4), and keratinisation (Benjamini p=3.4E-3). A second group (enrichment=2.75) comprised genes of inflammation (Benjamini p=1.6E-3) and wound healing (Benjamini p=1.5E-2). A third cluster (enrichment=2.39) comprised genes involved in calcium handling (Benjamini p=2.5E-2). Interestingly, only CERS3 and ELOVL2 were upregulated in patients, while all the other ELOVL-isoenzymes remained unaltered. Among downregulated genes we only found a single gene cluster (PODXL, CDH18, CDH8, C11ORF87) with Benjamini p=5.2E-2 (marginally significant). Beyond that a group of cGMP-pathway genes (CNGA3, PDE1A, PDE1C) being weakly enriched (enrichment=1.5, Benjamini p=1.2E-2). Online supplementary table S3 reveals a high percentage (>40%) of downregulated genes that are involved in neurodevelopment (CHD18, FOLR1, ADRA2A, CNGA3, RIMS1, CDH8, PODXL, BRINP1, S1PR1, IGFBP3, NOG, FGF13, APBB1IP).

Determination of PPAR binding incidences in regulated genes

Bioinformatic analysis of the promoter regions of upregulated and downregulated genes from online supplementary table S3 for high efficiency (>80%) PPAR binding incidences, revealed numerous PPAR response elements (PPREs) in 32% of the genes for PPARα (n=8) and PPARγ (n=20) (see online supplementary table S4). Highest PPAR binding efficiency of 100% was predicted for CHD8, CERS3, KRTDAP and LIPG.

Biochemical investigations

Lipid analyses of patient fibroblasts and skin samples

Total fibroblast ceramide and sphingomyelin levels were similar in patients and controls. We found reduced C26-ceramides and sphingomyelins behind the ELOVL1 defect, while C20- and C22-sphingomyelins were increased. This was also reflected in the decrease of C24:0/C22:0 and C26:0/C22:0 ratios of ceramide and sphingomyelin in skin/keratinocytes and fibroblasts (Figure 4A).

Figure 4

Biochemical analysis of fatty acid (FA) composition of ceramides and sphingomyelins. (A) Ceramides are composed of a sphingosine molecule (in black) coupled through an amide bond to a FA of various length (‘R’ in red). Lipids were extracted from patient and control fibroblasts and from differentiated keratinocytes, which served as controls for the patient skin biopsy samples. C24:0-, C26:0-, and C26:1-ceramides were significantly diminished in patients. Since ELOVL1 specifically catalyses elongation of C22:0-CoA to C24:0-C26:0-CoAs, we measured C24:0/C22:0 and C26:0/C22:0 ceramide ratios in fibroblasts and skin biopsy samples, which were significantly reduced in patients. *p<0.05; ***p<0.001 (one-tailed independent t-test). (B) Sphingomyelins are composed of a ceramide molecule coupled to phosphocholine (in green). C26:0- and C26:1-sphingomyelins were decreased, while C22:0sphingomyelins in front of the biochemical defect were increased. C24:0/C22:0 and C26:0/C22:0 sphingomyelin ratios were reduced in patient fibroblasts and skin samples. *p<0.05; **p<0.01; ***p<0.001 (one-tailed independent t-test). (C) To investigate ELOVL1 enzymatic activity for in vitro elongation of C20:0-CoA and C22:0-CoA we performed an FA elongase assay using total membrane fractions with [13C]-labelled malonyl-CoA and unlabelled C20:0-CoA or C22:0-CoA. Cells were transfected with empty control vector versus constructs expressing ELOVL1(wild-type), ELOVL1(p.S165F), and ELOVL1(p.H144A|p.H145A) carrying a loss-of-function mutation in its catalytic centre. ELOVL1-EGFP, construct expressing wild-type or mutant ELOVL1 protein fused to a C-terminal EGFP; 3xFLAG-ELOVL1, construct expressing wild-type or mutant ELOVL1 protein fused to a triple N-terminal FLAG tag. Values presented are the means, error bars depict SD. *p<0.05; **p<0.01; ***p<0.001 (Tukey’s honestly significant difference test).

In vitro FA elongation assay of wildtype and mutant ELOVL1

Mutant ELOVL1(p.S165F) does not exhibit any enzymatic activity. Cells transfected with the ELOVL1(p.S165F) constructs and a construct with mutated active site, ELOVL1(p.H144A/p.H145A), failed to incorporate [13C]malonyl-CoA as C2 donor into C20:0-CoA or C22:0-CoA (Figure 4C). Correct expression of ELOVL proteins in transfected cells was verified by Western blot analysis (see online supplementary figure S6).

FA treatment of patient fibroblasts

In control and patient fibroblasts, loading with C22:0-FAs (figure 5A), increased the C22:0- and C24:0-ceramide and sphingomyelin levels. Interestingly, C22:1-, C24:1-, C26:1-ceramides and C24:1-, C26:1-sphingomyelins were decreased. Loading with C22:1-FAs killed both patient and control cells. Loading with C24:0-FAs (figure 5B) did not significantly alter ceramide or sphingomyelin content, which is likely due to the combined effect of scarce incorporation into cells and low solubility in the media. Loading with C24:1-FAs (figure 5C) significantly increased C24:1-sphingomyelins, but decreased C24:0 -ceramides and sphingomyelins.

Figure 5

Loading of control and patient fibroblasts with C22:0-, C24:0-, and C24:1-FAs. (A) Loading with behenicacid (C22:0) significantly increased saturated C22:0- and C24:0-ceramides and sphingomyelins, while it reduced monounsaturated C24:1 -ceramides and sphingomyelins. (B) Loading with lignoceric acid (C24:0) did not change the ceramide and sphingomyelin levelsof patient and control cells. (C) Loading with nervonic acid (C24:1) did only increase C24:1-sphingomyelins, but not C24:1-ceramides. In control cells, C24:1-FA treatment led to a significant reduction of C24:0 -eramide and sphingomyelin. *p<0.05; **p<0.01; ***p<0.001 (two-tailed paired t-test).

Immunofluorescence

Subcellular localisation of the wild-type and mutant ELOVL1 protein was studied by fluorescence microscopy in transiently transfected HeLa and COS-1 cells (figure 6A,B). As expected, ELOVL1 protein was found in the ER compartment, which has a continuous transition to the nuclear envelope.5 Mutated and wild-type ELOVL1 protein colocalised, thereby excluding a mislocalisation of the mutant protein.

Figure 6

Elovl1 gene expression studies. (A) Confocal images of COS-1 cells after simultaneous transfection with ELOVL1(wild-type)-EGFP and ELOVL1(p.S165F)-RFP constructs do not reveal any difference in subcellular localisation between wild-type and mutant protein. (B) Mutant ELOVL1 protein is localised in the ER as confirmed by colocalisation of calnexin as an ER marker with the FLAG immune signal deriving from the 3xFLAG-tagged ELOVL1 protein. The same applies to wild-type ELOVL1 protein (data not shown). (C) In situ hybridisation of cryosections from a E16.5 mouse embryo demonstrates nearly ubiquitous Elovl1 expression, although at different levels. Highest Elovl1 expression levels are found in epidermis, brain cortex, pituitary, and spinal cord. (D) In the adult mouse brain, highest Elovl1 expression levels are found in the hippocampus (the CA1-3 regions and the gyrus dentatus), the granule and Purkinje cells of the cerebellum and the neocortical neurons, especially in the layer IV pyramidal neurons. Cb, cerebellum; Cor, brain cortex; E, oesophagus; Epi), epidermis; H, heart; Liv, liver; Lu, lung; MO, medulla oblongata; Pc, pancreas; Pit, pituitary; SC, spinal cord; SMG, submandibular salivary gland; T, trachea; Th, thalamus; Tm; thymus; VC, vertebral column. (E) Reverse transcription (RT)-qPCR for Elovl1 gene expression in different mouse tissues. Error bars represent the SD from the measurement of three replicates.

Elovl1 gene expression studies

In situ hybridisation

Elovl1 was ubiquitously expressed with highest expression in the epidermal layer. High expression was also seen in the cerebral cortex, spinal cord, hippocampus, granule and Purkinje cells of the cerebellar cortex, spinal cord, thyroid and pituitary glands (figure 6C,D)

RT-qPCR analysis

Elovl1 mRNA expression was ubiquitous in different mouse tissues, although at different expression levels varying by more than two orders of magnitude. Lowest expression was seen in heart and skeletal muscle and highest in the eye, cerebral white matter, brainstem, spinal cord, and peripheral nerve (figure 6E).

Discussion

We report on two unrelated patients with a severe congenital condition affecting skin, central nervous system (CNS), and sensory organs, who carried the identical de novo mutation in ELOVL1 (c.494C>T|p.S165F) (Figure 3A–C). ELOVL1 encodes 279AA-protein ELOVL fatty acid elongase 1, catalysing elongation of VLCFAs in the ER, preferably of saturated and monounsaturated C22- and C24-acyl-CoAs.5 De novo occurrence was verified by DNA analysis of both patients’ parents and siblings. Given the close geographic origin of both patients, we considered the remote possibility of a founder mutation transmitted via germline mosaicism. In this case, we would expect a common founder allele, which was, however, ruled out (Figure 3D). It is intriguing for this de novo mutation to occur at exactly the same location in both patients. There seem to be preferred genomic sites for de novo mutations, as reported for SLC25A4 and ALDH18A1.14 15 Mutant ELOVL1 Ser165 along with its flanking amino acids is highly conserved throughout vertebrates (Figure 3E) and was predicted to be located at the border between ER lumen and transmembrane helix 5 (see online supplementary figure S7). This is in line with known dominant missense mutations of ELOVL4 (p.W246G) and ELOVL5 (p.G230V),16 17 located at the border between the small loop in ER lumen and helix 7, both causing spinocerebellar ataxia type 34 and 38 (OMIM#133190, OMIM#615957).

The finding of p.S165F ELOVL1 as causative dominant mutation is supported (1) by a decrease of ELOVL1 enzymatic activity, reflected by (i) loss of enzymatic function in the FA-elongation assay, (ii) reduction of the enzymatic products (C26:1- and C26:0-ceramides and sphingomyelins, Figure 4A,B) behind and accumulation in front of the enzymatic block (C22:0-sphingomyelins), (iii) decrease of C24:0/C22:0 and C26:0/C22:0 sphingomyelin and ceramide ratios, especially in patient skin samples. Further support comes from the (2) genotype-phenotype correspondence of two unrelated patients carrying the same de novo mutation on different genetic backgrounds, which was absent in the gnomAD database and (3) an Elovl1-knockout mouse model with similar skin alterations.1 (4) Finally, Elovl1-mRNA is highly expressed in the predominantly affected tissues, such as white matter of the brain, eye, and skin (figure 6C–E).

The dominant inheritance pattern suggests either haploinsufficiency or gain of pathologic function as underlying pathomechanism. In dominant ELOVL4 nonsense mutations, gain of pathologic function has been verified for Stargardt macular degeneration (OMIM#600110), where a C-terminally truncated ELOVL4 protein mislocalised due to loss of its ER retention motif,18 aggregated with wild-type ELOVL4 and accumulated in aggresomes.19

In contrast, recessive Elovl1, Elovl4 and ELOVL4 mutations completely abrogate expression of both mutant alleles, either in mice by knockout of an entire exon,1 20 or in humans by nonsense or frameshift mutations located 5’ to the final exon. In humans, recessive ELOVL4 mutations cause ichthyosis, spastic quadriplegia, and mental retardation (OMIM#614457).21 In humans and mice, homozygous individuals were always severely affected, while heterozygous carriers were healthy. This suggests that one functional allele might suffice to maintain enzymatic function in the ELOVL-enzyme group. The mutation of our patients entirely abrogated ELOVL1 enzymatic activity (Figure 4C). However, as haploinsufficiency may not explain the disease, we suspected an additional gain of pathologic function, searched for mislocalisation of ELOVL1(p.S165F) protein, but found none (figure 6A,B). Further studies are needed to investigate whether complex formation between mutant and wild-type ELOVL1 protein might impede overall enzymatic function.

Neuronal development and myelination crucially depend on VLCFA availability and on strict control of their synthesis and degradation. Failure to degrade VLCFAs, as seen in X linked adrenoleukodystrophy (OMIM#300100), leads to widespread white matter lesions possibly via aberrant microglia activation.22 Failure to synthesise VLCFAs might lead to hypomyelination as seen in ELOVL4 mutations.21 VLCFAs are de novo synthesised cell autonomously by oligodendrocytes or Schwann cells.23 24 In the CNS, precursor lipids required for large-scale VLCFA synthesis are provided by neighbouring astroglia.24 Myelin membranes contain higher proportions of ELOVL1-generated saturated and monounsaturated C24-VLCFAs than other membranes.25 For C24-ceramide synthesis, CERS2 uses C24-VLCFAs produced by ELOVL1.8 Cers2 knockout mice showed significant decrease in myelin-specific C24-sphingolipids impairing their myelin sheath stability.6 We found central hypomyelination in our patients, although only mild and static, clinically manifesting as spastic paraplegia. However, their peripheral nervous system (PNS) was entirely unaffected. Normal ultrastructure of peripheral nerves (online supplementary figure S8) and normal peripheral motor and sensory nerve conduction velocities were testament to this, whereas central conduction velocities were all pathological. Yet, ELOVL1-mRNA expression in PNS and CNS is similar (figure 6E), as are relative amounts of C24-sphingomyelin.26 One hypothetical explanation may be the Schwann cells' localisation outside the blood-brain barrier, in contrast to oligodendrocytes,27 which would guarantee access to the bloodstream for FA uptake. Alternative FA sources for Schwann cells may derive from lipolysis of epineurial adipocytes23 or from compensatory synthesis via an alternative pathway. The latter phenomenon was observed in patients with fatty acid 2-hydroxylase deficiency (OMIM#612319), whose PNS was also spared. The authors discovered compensatory FA-hydroxylation activity in peripheral cells that was absent in CNS.28 Interestingly, VLCFA levels in plasma were normal in both our patients (see online supplementary table S2), possibly reflecting rather dietary resorption than cell-autonomous synthesis.29

The patients' audiograms (figure 2I) with high frequency hearing loss resemble those of sensory presbycusis, caused by degeneration of the distal-most segments of the auditory neurons or their synaptic contacts.30 Not much is known about VLCFA requirements for inner ear development and function. Of note, one epidemiological study has observed an inverse correlation between presbycusis and high dietary intake and plasma levels of long-chain ω-3 polyunsaturated fatty acids (PUFAs).31

Ichthyosis of our patients corresponds to skin alterations of Elovl1-/- mice that die shortly after birth due to ichthyosis-like symptoms, with impaired skin barrier formation causing excessive transcutaneous water loss.1 VLCFAs provided by ELOVL1 (particularly ≥C26-FAs) are essential components of epidermal lipid lamellae, a hydrophobic structure between corneocytes of the epidermal stratum corneum, constituting the skin permeability barrier.9 Interestingly, we found compensatory upregulation of ceramide synthase 3 (CERS3) in our ELOVL1-deficient patient fibroblasts (see online supplementary table S3). In cooperation with ELOVL1, CERS3 is required for synthesis of ≥C26-sphingolipids. Cers3 -/- knockout mice die from a similar skin barrier defect as Elovl1-/- mice,32 human recessive CERS3 mutations cause ichthyosis.33 34 Moreover, thickened epidermal layer with increased cornification of patient skin (figure 2G) can be explained by the massive upregulation of an entire group of genes enhancing keratinocyte growth and differentiation (CERS3, GJB6, KRT75, CALML5, SPRR3, KRTDAP, SPRR2A, CRCT1, LCE3E, SPRR4, KRT5, KRT14, CDH3, SOX11, PDPN, online supplementary table S3).

Furthermore, we found numerous lysosomes filled with electron dense bodies (either lipopigment or melanosomes) in patient skin (figure 2L), supposedly the morphological correlate of acanthosis nigricans, a poorly understood phenomenon observed in diseases associated with insulin resistance. Pathogenesis includes insulin-mediated activation of insulin-like growth 1 factor receptors (IGF1R) on keratinocytes and fibroblasts, and increased levels of circulating insulin-like growth factor 1 (IGF1).35 Indeed, in patient fibroblasts we found upregulation of IGF-like family member 1 mRNA (IGFL1), and a downregulation of IGF-binding protein 3 (IGFBP3), normally inhibiting IGF1-stimulated IGF1R activity36 (see online supplementary table S3). Interestingly, acanthosis nigricans has been successfully treated by oral supplementation with ω-3 PUFA-rich fish oil.37

We wondered what mechanism might link ELOVL1 dysfunction to loss of peripheral vision and visual acuity observed in our patients. Recently, ELOVL1 was found to protect photoreceptors from light-induced degeneration by reducing activity of retinoid isomerohydrolase 65 kDa protein (RPE65) in the retinal pigment epithelium.38 We did not find evidence for retinal degeneration because ophthalmological inspection, macular SD-OCT, and retinal function on ERG were normal. Still, a slow retinal degeneration may not be detectable at the patients’ young age and may require long-term follow-up. However, circular SD-OCT revealed thinning of the retinal nerve fibre layers in both patients and confirmed the clinical impression of optic atrophy. The requirement of VLCFAs for retinal ganglion cell and optic nerve function and survival is not well understood. A quantitative study of the relative contribution of phospholipids in the neural retina, optic nerve head, and optic nerve proper revealed concentrations of sphingomyelins that were threefold higher in the optic nerve and nerve head in comparison to the neural retina.39 Of note, axons of retinal ganglion cells do not possess intraocular myelin sheaths; they are only present after the fibres have left the optic bulb through the lamina cribrosa. Given the importance of ceramides and sphingomyelins for myelination in general,40 the importance of C24:1-sphingomyelin for juvenile forebrain development,41 and the high expression of ELOVL1-mRNA in the eye (figure 6E), it is conceivable that presence of VLCFAs would be crucial for optic nerve fibre myelination and maintenance. To our knowledge, astigmatism has not been reported in patients with ELOVL mutations and is a novel ocular finding, while nystagmus appears to be a consistent feature in patients with ELOVL mutations.16 42 43 In contrast to earlier studies that reported supranuclear gaze palsy,16 esotropia, and slow abduction16 42 43 in patients with ELOVL4 and ELOVL5 mutations, these ocular motor disturbances were not observed in our patients with an ELOVL1 mutation.

Comparison of gene expression patterns showed the heterozygous ELOVL1 mutation to exert an influence on entire gene expression modules, which can be subsumed under ‘keratinocyte differentiation and skin formation’, ‘inflammation’, ‘cell growth and apoptosis’ and ‘neurodevelopment’ (see online supplementary table S3). Our analysis was done in fibroblasts and not in highly specialised cells, such as keratinocytes or neurons. However, genome-wide expression data from patient-derived non-neuronal tissue corresponded well to those of the human brain transcriptome.44 The most striking upregulation was found in genes implicated in keratinocyte differentiation as discussed above. Among the downregulated group, we found an enrichment of genes involved in brain development (APBB1IP, BRINP1, CDH8, CDH18, CLIC3, NOG, PODXL), myelination (FOLR1, S1PR1, IGFBP3), and synaptic function and plasticity (ADRA2A, RIMS1, CNGA3, FGF13, online supplementary table S3). This makes teleological sense as neuronal maturation and myelination cannot proceed unimpeded in the absence of the required building blocks.

Changed expression of entire gene modules in states of VLCFA deficiency suggests FA involvement in the transcription machinery. Activation and suppression of gene transcription by long-chain FA (LCFA) is well studied in liver.45 LCFA bind to PPARs, members of the nuclear receptor superfamily. On LCFA binding, PPARs change their conformation, recruit cofactors and either enhance or suppress transcription. PPARα has a high binding affinity for VLCFA-CoA thioesters.46 Inhibition versus stimulation strongly depends on the type of VLCFA-CoA bound. Notably, nervonoyl-(C24:1)-CoA reduced coactivator recruitment, thereby suppressing PPARα-mediated transcription.46 Indeed, we discovered multiple PPRE 5 kbp upstream of the transcription start site of many genes with altered expression levels in our patients. Interestingly, genes encoding ceramide synthase 3 (CERS3) and keratin differentiation associated protein (KRTDAP) had the highest probabilities for PPARα binding. Deficiency of C24-FAs may thus interfere with the PPARα transcriptional network, possibly contributing to the strong upregulation of genes enhancing epidermal proliferation and cornification. This regulatory mechanism has not been studied in brain and keratinocytes, but PPARα is highly expressed and active in these tissues.47 48 FAs also modulate PPARγ activity, especially in the PNS. In mice, decreased C16-FA availability downregulated the PPARγ transcriptional network in Schwann cells.23 Restoring PPARγ activity in those mice rescued PNS myelination.

Finally, we aimed to overcome the biochemical defect of our patients by exploiting our knowledge about ELOVL1 pathophysiology. As transmembrane transport of VLCFAs is only minimal, we treated patient fibroblast cultures with C22 and C24 precursor FAs. C22:1-FA could not be used, because it was toxic to the cells. Indeed, C22:0-supplementation significantly increased C22:0- and C24:0-, but unexpectedly decreased concentrations of C24:1- and C26:1-ceramides and sphingomyelins, possibly through competitive inhibition between saturated and monounsaturated elongase substrates. C24:0 loading showed no effect. C24:1 loading increased C24:1-sphingomyelins, likewise with decrease of C24:0-ceramides and sphingomyelins. Our results show that smaller size VLCFAs (C22:0) can be transported into cells and lead to an increase of products of the FA elongation cycle, although only one step further towards C24:0-FAs. We did not see any increase of C26-ceramides or sphingomyelins. Importantly, when establishing VLCFA treatment, we have to consider competitive inhibition between saturated and monounsaturated elongase substrates that might diminish ceramide and sphingomyelin synthesis. Of note, Lorenzo’s oil, a mixture of C18:1- and C22:1-VLCFA-containing triglycerides, inhibits ELOVL1 activity,49 thereby normalising increased C24:0-26:0 levels in X-ALD patients. This mixture would be contraindicated in our patients. Our findings call for more systematic investigations to determine the optimum mixture of saturated and monounsaturated FAs before any therapeutic intervention. Such therapy could then be tested as topical application on the skin, for example, by incorporating the highly hydrophobic VLCFA-mixture into liposomes for improved skin resorption.

We here describe the clinical phenotype and molecular mechanism of a heterozygous ELOVL1 mutation, which completely abrogated ELOVL1 enzymatic activity and reduced ELOVL1 products. Transcriptome analysis demonstrated modulation of genes involved in epidermal development and neurodevelopment, many of which being dependent on PPARs. The implication of the PPAR-pathway opens directions for further research and treatment. Many FDA-approved drugs with effect on PPAR activity are available and should be screened for positive effects. In addition, we show promises and caveats of VLCFA supplementation, which still needs to be explored systematically.

Acknowledgments

The authors would like to thank the patients and their parents for participation in the study and Angelika Zwirner and Esther Gill for excellent technical assistance.

References

Footnotes

  • Contributors NM: analysed and clinically verified the NGS data, did the segregation analyses in the family, cloned the ELOVL1-expression vectors, did the subcellular colocalisation analysis of mutant and wild-type ELOVL1, did the founder haplotype analysis, did the RT-qPCR analysis of mouse tissue, performed the transcriptome analysis, together with MS and TS wrote the first draft of the manuscript. TS: performed the measurement of ceramides and sphingomyelins in the patient fibroblast and skin, the fatty acid elongation assays, and investigated the subcellular localisation of wild-type and mutant ELOVL1 protein. Together with MS and NM wrote the first draft of the manuscript. SM-G: performed cell culture, immunostainings and in situ hybridisation. JS: investigated the patients, provided patient relevant health data. DJS: performed the ophthalmological investigations. DS: provided and programmed software for bioinformatic analyses. EK: supervised patient care and investigation, did molecular genetic analyses. WS: did the light and electron microscopic investigations. AK: directed and supervised the biochemical investigations. MS: conception and supervision of the study, investigated the patients neurologically, performed the bioinformatic analyses of WES and transcriptome analysis, did the genetic counseling of the family, together with NM and TS wrote the first draft of the manuscript. All authors read the final version of the manuscript for intellectual content and consented to its publication.

  • Funding The project was funded by the Deutsche Forschungsgemeinschaft (grant number SFB 665 TP C4 to MS), and the NeuroCure Center of Excellence (grant number Exc 257) as well as by the Advanced Research and Development Programs for Medical Innovation from the Japan Agency for Medical Research and Development (AMED) (AMED-CREST, grant number JP18gm0910002 to AK) and from the Japan Society for the Promotion of Science (JSPS) (KAKENHI grant number JP18H04664 to AK and grant number JP16K08220 to TS), NM was supported by the ’Studienstiftung des deutschen Volkes' and the ’Villigst Studienwerk'.

  • Competing interests None declared.

  • Patient consent Parental/guardian consent obtained.

  • Ethics approval The study was approved by the IRBs of the Charité (EA2/107/14) and the Faculty of Pharmaceutical Sciences of Hokkaido University (2017–002).

  • 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.