Biallelic variation in the choline and ethanolamine transporter FLVCR1 underlies a pleiotropic disease spectrum from adult neurodegeneration to severe developmental disorders

FLVCR1 encodes Feline leukemia virus subgroup C receptor 1 (FLVCR1), a solute carrier (SLC) transporter within the Major Facilitator Superfamily. FLVCR1 is a widely expressed transmembrane protein with plasma membrane and mitochondrial isoforms implicated in heme, choline, and ethanolamine transport. While Flvcr1 knockout mice die in utero with skeletal malformations and defective erythropoiesis reminiscent of Diamond-Blackfan anemia, rare biallelic pathogenic FLVCR1 variants are linked to childhood or adult-onset neurodegeneration of the retina, spinal cord, and peripheral nervous system. We ascertained from research and clinical exome sequencing 27 individuals from 20 unrelated families with biallelic ultra-rare missense and predicted loss-of-function (pLoF) FLVCR1 variant alleles. We characterize an expansive FLVCR1 phenotypic spectrum ranging from adult-onset retinitis pigmentosa to severe developmental disorders with microcephaly, reduced brain volume, epilepsy, spasticity, and premature death. The most severely affected individuals, including three individuals with homozygous pLoF variants, share traits with Flvcr1 knockout mice and Diamond-Blackfan anemia including macrocytic anemia and congenital skeletal malformations. Pathogenic FLVCR1 missense variants primarily lie within transmembrane domains and reduce choline and ethanolamine transport activity compared with wild-type FLVCR1 with minimal impact on FLVCR1 stability or subcellular localization. Several variants disrupt splicing in a mini-gene assay which may contribute to genotype-phenotype correlations. Taken together, these data support an allele-specific gene dosage model in which phenotypic severity reflects residual FLVCR1 activity. This study expands our understanding of Mendelian disorders of choline and ethanolamine transport and demonstrates the importance of choline and ethanolamine in neurodevelopment and neuronal homeostasis.


Abstract:
FLVCR1 encodes Feline leukemia virus subgroup C receptor 1 (FLVCR1), a solute carrier (SLC) transporter within the Major Facilitator Superfamily.FLVCR1 is a widely expressed transmembrane protein with plasma membrane and mitochondrial isoforms implicated in heme, choline, and ethanolamine transport.While Flvcr1 knockout mice die in utero with skeletal malformations and defective erythropoiesis reminiscent of Diamond-Blackfan anemia, rare biallelic pathogenic FLVCR1 variants are linked to childhood or adult-onset neurodegeneration of the retina, spinal cord, and peripheral nervous system.
We ascertained from research and clinical exome sequencing 27 individuals from 20 unrelated families with biallelic ultra-rare missense and predicted loss-of-function (pLoF) FLVCR1 variant alleles.We characterize an expansive FLVCR1 phenotypic spectrum ranging from adult-onset retinitis pigmentosa to severe developmental disorders with microcephaly, reduced brain volume, epilepsy, spasticity, and premature death.The most severely affected individuals, including three individuals with homozygous pLoF variants, share traits with Flvcr1 knockout mice and Diamond-Blackfan anemia including macrocytic anemia and congenital skeletal malformations.Pathogenic FLVCR1 missense variants primarily lie within transmembrane domains and reduce choline and ethanolamine transport activity compared with wild-type FLVCR1 with minimal impact on FLVCR1 stability or subcellular localization.Several variants disrupt splicing in a mini-gene assay which may contribute to genotype-phenotype correlations.Taken together, these data support an allele-specific gene dosage model in which phenotypic severity reflects residual FLVCR1 activity.This study expands our understanding of Mendelian disorders of choline and ethanolamine transport and demonstrates the importance of choline and ethanolamine in neurodevelopment and neuronal homeostasis. .

Text:
Solute transport across lipid bilayers is critical for biological homeostasis and requires specialized transmembrane proteins 1 .The largest group of membrane transport proteins is the solute carrier (SLC) superfamily.The superfamily consists of 458 genes divided into 65 families 1 .SLC proteins transport a vast assortment of solutes including amino acids, sugars, ions, nucleotides, vitamins, and neurotransmitters.Genetic variation in SLC genes influences common diseases, metabolic traits, and an expanding spectrum of Mendelian disorders [2][3][4] .As solute transport disorders are responsive to solute supplementation, dietary interventions, and gene therapies in preclinical models and humans, they are attractive therapeutic targets [5][6][7] .Yet, the function of most SLC genes in human biology remains poorly characterized.
One SLC family member associated with Mendelian neurodegenerative disorders is FLVCR1 (also known as SLC49A1 and MFSD7B).First recognized as the receptor for Feline Leukemia Virus Subgroup C (FeLV-C), a retrovirus which causes aplastic anemia in cats 8 , FLVCR1 is highly conserved and encodes two isoforms in humans: a full-length 555 amino acid (aa) plasma membrane isoform FLVCR1a and a small mitochondrial membrane isoform FLVCR1b (aa 277-555) 9 .FLVCR1 was initially described as a heme exporter in erythroid cells and other cell lineages 10 .An essential requirement for FLVCR1 was subsequently demonstrated through knockout of the mouse ortholog Flvcr1 11 .Germline knockout of Flvcr1 results in embryonic lethality, absent erythropoiesis, craniofacial and limb malformations, and hepatic iron accumulation, whereas neonatal deletion causes severe macrocytic anemia 11 .Both FLVCR1 isoforms are required for murine viability; mice retaining FLVCR1b but lacking FLVCR1a have normal erythropoiesis but develop hemorrhages, edema, and skeletal malformations 9 .The physiologic significance of FLVCR1 heme transport has been called in question especially considering recent evidence that FLVCR1 is a choline and ethanolamine transporter 4,[12][13][14][15][16] .
. The severe anemia and craniofacial, limb and digital malformations in Flvcr1 knockout mice resemble humans with Diamond-Blackfan anemia (DBA) [MIM: 105650] 11,17 .A further link between FLVCR1 and DBA was suggested by the observation that FLVCR1 splicing is dysregulated in erythroid cells from patients with DBA 18 .However, biallelic variation in FLVCR1 has not been identified in individuals with DBA, and DBA is now recognized as primarily a disorder of ribosome biogenesis 17 .Instead, biallelic pathogenic variation in FLVCR1 causes rare recessive neurodegenerative disorders of the retina, spinal cord, and peripheral nerves: posterior column ataxia with retinitis pigmentosa (PCARP) [MIM: 609033], isolated retinitis pigmentosa (RP), and hereditary sensory and autonomic neuropathy (HSAN) [19][20][21] .FLVCR1related neurodegenerative disorders can manifest from childhood into adulthood.To date, at least 39 individuals from 22 families have been identified with FLVCR1-related neurodegeneration; most have biallelic FLVCR1 missense variants, and predicted loss-offunction variants have only been identified in trans with missense variants (Supplemental Table S1).Despite the severe multi-organ consequences of Flvcr1 knockout in mice, extraneurological findings or severe developmental disorders have only been described in one individual with biallelic FLVCR1 variants, a compound heterozygote (c.574T>C p.Cys192Arg; c.610del p.Met204Cysfs*56) with severe developmental disabilities, chronic macrocytic anemia, liver disease, self-mutilation, and sensory neuropathy 21 .The precise reason(s) for the apparent discrepancies between human and mouse FLVCR1-related phenotypes remains unclear but could suggest that FLVCR1 knockout is incompatible with human life.
The index case who initiated this study is a male child with severe developmental delay, epileptic encephalopathy, and microcephaly (Individual 1, Fig. 1, Fig. 2A-E, Table 1).He was born to non-consanguineous parents from a country in South Asia.He had a history of infantile spasms, self-mutilation, osteomyelitis, and absent sensory nerve responses on nerve conduction studies.Brain magnetic resonance imaging (MRI) showed corpus callosum thinning, brainstem and pontine thinning, prominent extra-axial fluid spaces, and reduced white matter volume (Fig. 2A-E).Uric acid and purine levels, measured to assess for Lesch-Nyhan syndrome [MIM: 300322] due to the history of severe self-injury, were normal.Clinical trio exome sequencing (cES) did not identify any variants in HPRT1 nor pathogenic variants in known disease genes.Hence, these cES data underwent research reanalysis through the Baylor College of Medicine Genomics Research Elucidates the Genetics of Rare diseases (BCM-GREGoR) which prioritized the homozygous FLVCR1 missense variant c.1390G>A p.G464S given the overlap between the patient's known sensory neuropathy and known FLVCR1-related disorders.
To comprehensively characterize the phenotypic spectrum of FLVCR1-related disorders in humans, we reanalyzed ES and genome sequencing (GS) data from the 29,766 individuals within the BCM-GREGoR and Baylor Genetics clinical diagnostic databases 22 , utilized the online matchmaking program GeneMatcher 23,24 , and searched other research and diagnostic lab datasets.We identified 27 individuals from 20 unrelated families with neurological disorders and biallelic FLVCR1 variants (Fig. 1, Table 1).A pedigree was not available for Individual 27, an elderly woman with isolated retinitis pigmentosa and dystonia.Unlike prior reports describing biallelic FLVCR1 variants in children or adults with neurodegenerative disorders, Individuals 1-17 had severe developmental disorders.Notably, three individuals with severe developmental disorders had homozygous FLVCR1 pLoF variants (Individuals 4, 15, and 17).Individuals 18-27 had childhood or adult-onset neurodegenerative disorders including PCARP, hereditary spastic paraplegia, and RP (Table 1).Most families (14/20) are consanguineous by clinical history.In Families 1 and 7, the same homozygous missense variant c.1390G>A p.G464S was identified.
Both families originate in the same country in South Asia and are unrelated.Neither family are consanguineous by clinical history.Absence-of-heterozygosity (AOH), a surrogate measure from ES data for runs-of-homozygosity (ROH), was calculated for Family 1; total autosomal .AOH was 150.1 Mb, and the variant was surrounded by a 4.9 Mb AOH block (Supplemental Fig. 1) 25 .These data suggest FLVCR1: c.1390G>A p.G464S may represent a South Asian founder allele.Compound heterozygous variants were identified in the three nonconsanguineous families and segregated according to Mendelian expectations within Family 8.
All clinical and research data were acquired in accordance with ethical standards upon informed consent and with the approval of the collaborative institutional review boards (Supplemental

Methods).
The phenotypic features of the 27 individuals are summarized in Table 1, Fig. 2, and Supplemental Table 1.Individuals 1-17 exhibited severe developmental delay: all were nonverbal and achieved no developmental motor milestones.All were microcephalic (median Zscore -4.45, range -2.5 to -10.5) and had reduced brain volume on brain magnetic resonance imaging (MRI).Individual 16 was microcephalic at birth (29 cm, Z-score -3.9) but was normocephalic at 3 years old (50.2 cm, Z-score +1.05).Brain MRI findings varied considerably.Some individuals had only a mild reduction in white matter volume, corpus callosum thinning, and/or pontine and brainstem thinning (Fig. 2E).Others had a severe reduction in brain volume with simplified gyral pattern (Fig. 2H, J, K).In Family 8, cystic encephalomalacia was seen in all three affected siblings (Fig. 2O-R); cystic encephalomalacia was detected on fetal MRI suggesting a very early developmental defect.Individuals 16 and 17 had only a thin rim of cerebral cortex reminiscent of hydrancephaly.Another recurring MRI finding observed in Individuals 1 and 3 was T2 hyperintensity of the posterior columns on spine MRI as previously reported in PCARP (Fig. 2L).Premature death before adulthood was common (14/17).
Hypotonia and epilepsy were nearly universal in individuals who survived the neonatal period.
Other common traits include cortical visual impairment, optic disk atrophy, and spasticity.
Features previously associated with FLVCR1 including RP and sensory neuropathy were observed but uncommon; this could represent age-dependent penetrance due to the young age of the cohort or under-ascertainment.Individuals 1 and 8 with the homozygous p.G464S variant had a history of self-mutilation, osteomyelitis, and sensory neuropathy; congenital insensitivity to pain was reported in Individual 3, but nerve conduction studies were not performed.Individual 16 also engaged in self-injurious behavior including tongue and lip biting.
The most profoundly affected individuals exhibited considerable phenotypic overlap with Flvcr1 knockout mice.Individuals 9-11 in Family 8 were stillborn and exhibited craniofacial, limb, and digital malformations; they were recently identified in a large cohort that investigated the utility of long-read whole genome sequencing 26 (Fig. 2X-g, Supplemental Data).Individual 17 from Family 12 died on in the neonatal period.Fetal MRI and autopsy demonstrated craniofacial, limb, and digital malformations in addition to congenital heart disease, renal agenesis, hepatosplenomegaly, and paper-thin cerebral cortex with hydrocephalus and multiple hemorrhages (Fig. 2X-d, Supplemental Data).Bone marrow hematopoiesis was normal.
Hepatomegaly and milder digital malformations including polydactyly and arthrogryposis were also observed in other individuals with FLVCR1-associated severe developmental disorders.
Unexplained macrocytic anemia was present in 5 of 12 individuals who survived the neonatal period and for whom red blood cell studies were performed.Reduced fetal movements were reported in Families 8 and 11.
Families 13-20 had mild FLVCR1-related diseases including typical phenotypes (PCARP, HSAN, RP) as well as developmental delay and the novel phenotype hereditary spastic paraplegia.While spasticity is a common feature of severe FLVCR1-related developmental disorders, it is uncommon in mild FLVCR1-related disease 27 .Two recurrent variants, p.Y128N and p.I343T, were observed in unrelated Persian families.Brain MRI was not performed in most instances of mild FLVCR1-related disease but was normal in individual 20 with FLVCR1-related sensory neuropathy, RP, and mild intellectual disability.The two siblings in Family 17 were evaluated for developmental delay, microcephaly, hypotonia, and hyporreflexia.
. There was intrafamilial variability with the younger brother exhibiting more developmental impairment than the older brother (Supplemental Fig. 2).In Family 18, the young adult asymptomatic sister of Individual 25 was also homozygous for FLVCR1: c.502C>G p.L168V; as age of onset as late as the third or fourth decade of life has been described, this likely represents age-related penetrance 28 .
To assess the impact of c.884-3C>G on splicing, whole blood RNA was isolated from the carrier mother of Individual 16, and reverse transcriptase polymerase chain reaction (RT-PCR) was performed.RT-PCR produced two bands: a larger band seen in controls and a smaller band absent in controls (Supplemental Fig. 4).Sanger sequencing confirmed that the larger band represented wild-type FLVCR1, whereas the smaller band represented a heterozygous deletion of exon 3 (r.884_1024del,p.A295_Y341del).The deletion spans portions of transmembrane domains 6 and 7 and completely removes the fourth cytoplasmic loop (Fig. 3B).The splicing variant c.1593+5_1593+8del is strongly predicted to alter splicing by SpliceAI (donor loss Δ score 0.97) and Pangolin (splice loss Δ score 0.88).A mini-gene splicing assay confirmed this .variant causes exon 9 skipping which is predicted to result in protein truncation (r.1526_1593del; p.A509Dfs*4) (Supplemental Fig. 5 & 9).Pathogenic missense variants are anticipated to disrupt gene function and involve more highly conserved functional regions than benign missense variants within the human population and non-human primates.We therefore compared pathogenicity predictions and conservation metrics between FLVCR1 missense variants identified in individuals with FLVCR1-related disorders and FLVCR1 missense variants present in human and primate populations in gnomAD v2.1.1 and primAD v1.0 (Fig. 3C) 30,33 .CADD, REVEL, MetaRNN, phyloP100way, and GERP scores were all significantly greater in cases than controls.In contrast, there was no difference in CADD, REVEL, or GERP scores between missense variants identified in mild versus severe FLVCR1-related disorders (Fig. 3D).Comparison of the distribution of severe versus mild disease-associated missense variants demonstrated that severe diseaseassociated missense variants form a cluster within transmembrane domains 9-11 (5 of 8 severe disease-associated variants vs. 0 of 15 mild disease-associated variants).Similarly, 87.5% of severe disease-associated missense variants (7/8) lie within both the plasma membrane isoform FLVCR1a and the mitochondrial isoform FLVCR1b, whereas only 53.3% of mild disease-associated missense variants (8/15) fell within both isoforms.
To investigate the molecular consequence of FLVCR1 missense variants on gene function, we generated missense variants using site-directed mutagenesis of human FLVCR1 cDNA and examined each variant's choline and ethanolamine transport activity (Fig. 4) 14 .
Briefly, human FLVCR1 cDNA was co-transfected into HEK293 cells along with human choline kinase A (CHKA) or ethanolamine kinase 1 (ETNK1) cDNA because the sole expression of FLVCR1 cDNA only slightly increased choline or ethanolamine transport activity.HEK293 cells were then incubated with [ 3 H] choline or [ 14 C] ethanolamine, washed, and cellular radioactivity levels were measured.CHKA catalyzes choline phosphorylation into phosphocholine while .ETNK1 catalyzes ethanolamine into phosphatidylethanolamine; their co-expression along with FLVCR1 greatly increases choline or ethanolamine import 14 .In addition to the FLVCR1 missense variants identified in this publication, we also examined p.Q124L and p.G219C, two FLVCR1 variants recently identified in an individual with HSAN 19,34 .These assays demonstrated nearly all FLVCR1 missense variants significantly reduced choline and ethanolamine transport activity relative to wild-type FLVCR1 with transport activity ranging from 0% to 55.38% (choline) and 0% to 48.80%.The transport activity of two variants, p.M151V and p.D421N, was comparable to wild-type FLVCR1.The p.M151V variant occurred in the homozygous state in an individual within the BCM-GREGoR database.This individual had a blended developmental disorder phenotype resulting from multi-locus pathogenic variation in three genes: AP4B1, AMPD2, and NOTCH2 35 .As the individual's phenotype is explained by other gene variants and the p.M151V variant's transport activity is normal, p.M151V may represent a rare benign polymorphism.The p.D421N variant was identified in compound heterozygosity with a nonsense variant in Individual 3.Although p.D421N did not alter choline or ethanolamine transport activity, the variant is weakly predicted to cause donor loss by SpliceAI (score 0.22).A mini-gene splicing assay of this variant (c.1261G>A p.D421N) demonstrated a strong effect on exon 6 skipping (Supplemental Fig. 5 & 6).The phenotypic overlap of Individual 3 with other severe FLVCR1 cases and the presence of typical FLVCR1 features including RP and posterior column T2 hyperintensity supports the variant's pathogenicity.Another missense variant in the same exon (c.1235G>A p.G412A) also showed a strong effect on exon 6 skipping in addition to its effect on choline and ethanolamine transport (Supplemental Fig. 5 & 6).Western blot analysis of HEK293 cells overexpressing wild-type and variant FLVCR1 showed similar levels of FLVCR1 protein indicating that FLVCR1 missense variants do not impact protein stability (Fig. 4).Immunostaining demonstrates FLVCR1 missense variants localize to the plasma membrane like wild-type FLVCR1; only p.G412A exhibited abnormal intracellular accumulation (Fig. 4).As we previously characterized the choline transport activity of all FLVCR1 missense variants .reported in the literature 36 , we compared transport activity between mild and severe phenotypeassociated variants but found no significant difference (mean choline transport activity as a percentage of wild-type FLVCR1: mild phenotypes, 34.3±21.9;severe phenotypes, 35.5±21.9;p=0.9222).
Here, we demonstrate rare genetic variation in the choline and ethanolamine transporter gene FLVCR1 causes a broad and pleiotropic recessive disease spectrum ranging from adult neurodegeneration to severe developmental disorders.Choline is an essential nutrient which plays an integral role in methyl group metabolism, phosphatidylcholine (PC) synthesis via the Kennedy pathway, and acetylcholine synthesis 37 .While de novo choline synthesis occurs in the liver through the phosphatidylethanolamine N-methyltransferase pathway, this pathway is inadequate to meet the needs of the human organism 37 .Choline is found in animal products, cruciferous vegetables, and beans 37 .The dietary intake of choline in most individuals in the United States and Europe falls below the United States Institute of Medicine's adequate intake level 37,38 .Choline insufficiency is particularly common in low-income countries, and choline is neglected in nutrient-fortified food aid 39 .Furthermore, individual dietary requirements vary considerably due to multiple factors including genotype, sex, developmental stage, and dietary intake of vitamins involved in methyl group metabolism including folate and B12 40 .Single nucleotide polymorphisms in genes involved in the phosphatidylethanolamine Nmethyltransferase pathway, folate and methyl group metabolism, and choline transport including PEMT, MTHFR, MTR, MTRR, and SLC44A1 have been shown to modulate choline requirements and risk of choline deficiency 40 .Similarly, ethanolamine cannot be synthesized in humans and is a precursor for phosphatidylethanolamine (PE) synthesis via the Kennedy pathway 41 .PE and PC are abundant membrane phospholipids required for membrane integrity, cell division, and mitochondrial respiratory function.
. Choline is required for normal neurodevelopment 42 .Maternal choline deficiency impairs hippocampal development as well as neuronal progenitor cell and retinal progenitor cell expansion and differentiation in mouse embryos 43,44 .Choline deficiency also causes anemia, liver disease, growth retardation, and immune deficiency [45][46][47] .Neurodevelopment is also disrupted by defective choline uptake 48  Finally, the protein encoded by closely related gene FLVCR2 is expressed at the blood-brain barrier and transports choline and ethanolamine 16,52 .Pathogenic variation in FLVCR2 causes Fowler syndrome, a severe AR disorder characterized by proliferative vasculopathy, hydranencephaly, fetal akinesia deformation sequence, and prenatal lethality [MIM: 225790] 53 .
The identification of FLVCR1 as a neurodevelopmental choline and ethanolamine transport disorder further reinforces the critical role of choline and ethanolamine in brain development.
Additionally, the identification of multiple neurodevelopmental choline and ethanolamine transport disorders demonstrates considerable complexity to choline and ethanolamine regulation within the developing brain and warrants further study.Therapeutic choline and ethanolamine supplementation has not been attempted in patients with choline and ethanolamine transport disorders but could in theory be efficacious as in riboflavin transporter deficiency 6 .The existence of multiple choline transporters allows for alternative routes for choline uptake 1 , and supplementation could potentially boost choline intake through hypomorphic transporters.Randomized controlled trials of choline supplementation in fetal alcohol syndrome have shown choline supplementation is well-tolerated and can have beneficial .neurocognitive effects 54,55 .Measurement of choline and ethanolamine levels in biospecimens from individuals with FLVCR1-related disorders is needed to inform choline and/or ethanolamine supplementation as a potential therapeutic modality.
Based on these findings, we propose an allele-specific gene dosage model in which disease severity is a function of residual FLVCR1 activity.This proposal stems from three key observations.First, we identified three individuals with severe developmental disorders and homozygous FLVCR1 pLoF variants.These individuals are thus genetically equivalent to Flvcr1 knockout mice and zebrafish, which also exhibit severe developmental phenotypes 11,56 .Biallelic pLoF FLVCR1 variants have not been identified in individuals with milder FLVCR1-related phenotypes like PCARP; nearly all reported individuals instead have biallelic missense variants.
Second, we show most pathogenic FLVCR1 missense variants reduce but do not completely eliminate choline and ethanolamine transport and thus likely represent hypomorphic alleles 36 .
Flvcr1 was previously shown to exhibit highest central nervous system expression in the retina followed by the posterior column of the spinal cord and cerebellum; this may parsimoniously explain the susceptibility of these tissues to mild hypomorphic alleles 19 .Third, we observe evidence of complex compound inheritance in which FLVCR1 variants may contribute to mild phenotypes in homozygosity but severe phenotypes in compound heterozygosity with a more severe variant (e.g., FLVCR1 p.C192R) or severe phenotypes in homozygosity but mild phenotypes when paired with a less deleterious variant (e.g., FLVCR1 c.1593+5_1593+8del) 19,21,29 .The allele-specific gene dosage model is well-described in recessive disorders but can confound diagnostic personalized genome analysis when the clinically reported phenotype diverges substantially from the disease trait clinical synopsis provided in the literature or online databases like Online Mendelian Inheritance in Man (OMIM).Indeed, the severely affected individuals reported here had all undergone clinical or research exome or genome sequencing which identified the reported FLVCR1 variants, yet in each case .the variants were previously felt either non-contributory or of uncertain significance given the apparent phenotypic mismatch.Such false assumptions illustrate the importance of incorporating model organism data into personalized genome analysis for rare diseases and the need to anticipate more severe and milder phenotypes associated with each disease gene locus to maximize the yield of diagnostic genetic testing.
We did not observe a correlation between FLVCR1 missense variant choline transport activity and phenotypic severity in our functional assay.There are several possible explanations.First, our choline transport assay may lack the sensitivity to discriminate between mild and severe phenotype-associated variants.As the transport assay is reliant on overexpression of FLVCR1 cDNA, it is insensitive to splicing defects.Indeed, we demonstrate through mini-gene splicing assays that several FLVCR1 single nucleotide variants and indels disrupt splicing (Supplemental Figures 5-9).Second, the cellular phenotype assessed by the choline transport assay must be distinguished from the organismal phenotypes characterized in humans with pathogenic FLVCR1 variation.Third, FLVCR1 was found to transport additional ligands like ethanolamine in both this study and previous studies 14 .While a reduction in both choline and ethanolamine transport was observed with most variants, some like p.S305R had a more severe impact on ethanolamine than choline transport.Fourth, as both plasma membrane and mitochondrial isoforms FLVCR1a and FLVCR1b are required for murine viability 9 , a missense variant involving both isoforms may have a greater phenotypic impact than one only impacting the plasma membrane isoform FLVCR1a.In line with this hypothesis, we observed nearly all severe disease associated FLVCR1 missense variants are predicted to impact both isoforms.Another possibility is maternal and/or fetal deficiency of choline, ethanolamine, folate, and/or B12 due to poor dietary intake or maternal-fetal genotypes could act as phenotypic modifiers 40 .Finally, it is important to note the phenotypes associated with homozygous first exon frameshifts in Individuals 4 and 15 were not as severe as those seen in Individual 17 with .the homozygous exon 4 nonsense variant nor in individuals 12-14 with the homozygous splice site variant c.1593+5_1593+8del.This defines a gene polarity effect 57 .One possible explanation is first exon pLoF variants may not represent true null alleles.First exon premature termination codons (PTCs) can escape NMD and generate a functional or partially functional truncated protein if a downstream methionine allows for in-frame translation re-initiation 58 .
In summary, our report reveals a broad and pleiotropic phenotypic spectrum resulting from biallelic FLVCR1 variants; these range from adult neurodegeneration to severe developmental disorders with variable anemia and skeletal malformations.Combined with recent studies demonstrating FLVCR1 encodes a choline and ethanolamine transporter 4,13,16,36 , these data suggest choline and ethanolamine transport into the central and peripheral nervous systems is essential to prevent neurodegeneration and required for neurodevelopment.The observation that the most severe FLVCR1-related phenotypes cause stillbirth and resemble Diamond-Blackfan anemia establishes FLVCR1 should be considered in the differential diagnosis of recurrent miscarriage, multiple congenital anomalies, and severe Diamond-Blackfan anemia-like phenotypes.Further studies are required to understand the impact of gene x environmental (GxE) interactions on FLVCR1-related phenotypes and the therapeutic implications of choline or ethanolamine supplementation in FLVCR1-related diseases.

Figure 1 :
Figure 1: Pedigrees of families with FLVCR1-related developmental and

Figure 2 :F)
Figure 2: Phenotypic features of FLVCR1-related developmental and neurodegenerative

Figure 3 :
Figure 3: Summary of the FLVCR1 allelic series