Autosomal dominant transmission of GLUT1 deficiency

Abstract

GLUT1 deficiency is caused by a defect in the facilitative glucose transporter GLUT1. Impaired glucose transport across brain tissue barriers is reflected by hypoglycorrhachia and results in an epileptic encephalopathy with developmental delay and motor disorders. Recently heterozygous mutations in the GLUT1 gene (1p35–31.3) have been reported in sporadic patients. Parents and siblings carried the GLUT1 wild-type, suggesting a de novo, autosomal dominant condition resulting from GLUT1 haploinsufficiency. We report a father and two children from separate marriages affected by GLUT1 deficiency and carrying a novel heterozygous missense mutation (G272A) in the GLUT1 gene. Mutations were identified by polymerase chain reaction and DNA sequencing and confirmed by restriction fragment digest. The predicted amino acid change (Gly91Asp) affects an Arg-X-Gly-Arg-Arg motif between helices 2 and 3 that represents a cytoplasmic anchor point and is highly conserved among transporters of the major facilitator superfamily down to yeast and bacteria. GLUT1 immunoreactivity was normal, but 3-O-methyl-d-glucose uptake into erythrocytes was significantly reduced, suggesting a quantitatively normal, but functionally impaired, GLUT1 protein at the cell membrane. This is the first report of autosomal dominant transmission of GLUT1 deficiency, confirming that this condition is the result of haploinsufficiency. The Gly→Asp mutation within a highly conserved sequence highlights its importance for GLUT1 function. GLUT1 deficiency should be considered in patients with epilepsy, mental retardation and motor disorders. Our observations have bearing on the identification of this treatable disorder in pediatric and adult patients, will modify current biochemical protocols which use parental controls and will enable genetic counseling of affected families.

Received 31 August 2000; Revised and Accepted 26 October 2000.

INTRODUCTION

GLUT1, a member of the facilitative glucose transporter (GLUT) family, exclusively mediates glucose transport across the blood–brain barrier (BBB) (1,2). In the brain, GLUT1 and other GLUT isoforms are involved in intracerebral glucose transfer (for reviews see refs 3–8). GLUT1, a plasma membrane protein of 492 amino acids, has 12 transmembrane domains with both N- and C-termini located in the cytosol (9). The GLUT1 gene (1p35–31.3) is composed of ten exons and nine introns with an approximate length of 35 kb (9,10).

In 1991, De Vivo et al. (11) described two children with infantile seizures, developmental delay and acquired microcephaly. Analyses of the cerebrospinal fluid (CSF) showed unexplained hypoglycorrhachia (low CSF glucose concentration). CSF lactate concentrations were low to normal, suggesting intact intracellular pathways for glucose utilization. Based on these findings, a defect in GLUT1-mediated glucose transport into brain was assumed. Cytochalasin B binding studies, immunoblot analyses, and 3-O-methyl-d-glucose (3OMG) uptake into erythrocytes confirmed the GLUT1 defect. The novel disease was termed GLUT1 deficiency, or De Vivo disease (McKusick 138140). An effective treatment was developed by means of a ketogenic diet, as ketone bodies readily penetrate the BBB and serve as an alternative fuel for the brain (12,13). Molecular analyses identified a wide spectrum of heterozygous mutations, including nonsense, missense and splice site mutations, as well as hemizygosity of the GLUT1 gene in 15 patients (14–16). Parents and siblings were unaffected, suggesting a de novo, autosomal dominant condition.

We present an index patient with a complex neurological picture including intractable epilepsy and hypoglycorrhachia. Her father and half-sister (Fig. 1) also suffered from epilepsy and molecular analyses of the GLUT1 gene showed an identical heterozygous missense mutation in all three, suggesting an autosomal dominant trait caused by haploinsufficiency.

RESULTS

The lumbar punctures in II:3 and II:5, performed in the fasting state, showed a CSF:blood glucose ratio of 0.29 and 0.35, respectively. In II:3, initial single-stranded DNA conformation polymorphism (SSCP)analyses of exons 1–10 did not yield abnormal results. Consequently, polymerase chain reaction (PCR) fragments were forwarded to automated sequencing analyses. In II:3, I:2 and II:5 a heterozygous single-base G→A transversion in exon 3 at position 272, substituting aspartate for glycine (Gly→Asp) was detected (GenBank accession no. M20681) starting at ATG at positions 180–182 (Fig. 2A). PCR amplification and DNA sequencing of all remaining exons in II:3 did not show any additional mutations. In particular, DNA sequencing of exon 8 containing the coding sequence for the second R-X-G-R-R motif between helices 8 and 9 identified the GLUT1 wild-type in all affected individuals (data not shown).

In exon 3, the G→A transversion eliminated an HaeIII restriction site. Consequently, restriction analysis identified a novel 57 bp fragment in I:2, II:3 and II:5 in addition to the expected 43 and 36 bp wild-type fragments (Fig. 2B). This 57 bp fragment could not be observed in I:1 and in an unrelated control (Fig. 2B).

Quantitative western blotting of membranes from erythrocytes of patients I:2, II:3 and II:5 visualized against band 3 showed no difference in GLUT1 expression compared with I:1 and a control (Fig. 3).

Glucose uptake studies confirmed a functional defect of GLUT1. 3OMG uptake versus time in seconds, expressed as the slope of the curve as described previously (17), was 42% of controls in the index patient (II:3) and 57% of controls in I:2 and II:5 (Table 1).

DISCUSSION

Cerebral metabolism is unique because no other tissue except for the testis has been found to rely predominantly on carbohydrates for energy (18,19). Consequently, a defect of glucose transport across the BBB and into brain cells significantly impairs brain function and development as reflected in GLUT1 deficiency (11,12,14). A total of 15 cases of heterozygous private mutations in the GLUT1 gene have been described recently (14–16). The clinical presentation was extremely variable and preliminary analyses gave no evidence for phenotype–genotype correlations (20). Parents and siblings were clinically unaffected and carried the GLUT1 wild-type, suggesting that this disease results from a de novo genetic defect and is caused by haploinsufficiency. This is the first report of GLUT1 deficiency within one family with the father and two daughters from two separate marriages affected. All patients exhibited characteristic clinical features of the disease and hypoglycorrhachia was shown in two of them. Interestingly, patient I:2 (father) and II:5 (sister of index patient) represent the first cases reported with GLUT1 deficiency who have been diagnosed as adults and reflect the clinical presentation of this condition in adulthood. The three patients within that family shared a missense mutation within a conserved motif of the GLUT1 gene, representing a novel mechanism of transmission for this clinical entity.

Conserved motifs in the GLUT family include two R-X-G-R-R motifs between helices 2 and 3 in the N-terminal half, and correspondingly between helices 8 and 9 in the C-terminal half (Fig. 4), and two E-X-X-X-X-X-X-R motifs between helices 4 and 5, and correspondingly between helices 10 and 11. The repetition of these motifs between the two halves of the protein also suggests that duplication of a gene encoding an ancestral six-membrane-spanning helical protein may have produced the two-domain 12-membane-spanning helical structure that is so highly conserved in the GLUT family (5). Most of the intracellular R-X-G-R-R motif between helices 2 and 3 is present in GLUT1–8 and throughout multiple members of the facilitative superfamily of membrane transporters (Table 2) (3). These motifs may be conserved to maintain conformational stability of the protein and may be involved in salt-bridging between helices (5,21). Recently the importance of the amino acid charges for membrane topology within the R‐X-G-R-R-motif has been analyzed (22). Expression of site-directed mutant GLUT1 transporters in Xenopus oocytes showed that the substitution of the positively charged arginine with glycine completely abolished transport activity as the result of a local perturbation in the membrane topology, in which the cytoplasmic loop was aberrantly translocated into the exoplasm along with the two flanking transmembrane segments. The authors concluded that the positive charges in the R-X-G-R-R motif are likely to play no direct role in the transport mechanism, but may function as critical local cytoplasmic anchor points involved in determining the membrane topology of GLUT1. Based on these data, the identified mutation (G91D) substituting glycine, a particularly conserved small, hydrophobic and neutral amino acid, with aspartate, a long, hydrophilic and acidic, negatively charged amino acid, is likely to perturb the membrane topology of the GLUT1 transporter protein. Our data support the hypothesis of a heterozygous, non-functional GLUT1 transporter: 3OMG uptake was significantly impaired in all affected family members, indicating impaired function of transporters at the cell membrane. Membrane topology was not investigated in this study, but normal immunoreactivity of GLUT1 isolated from erythrocyte membranes of patients, mother and control suggests that the mutant transporters are fully expressed at the cell membrane (Fig. 3). Normal immunoreactivity in the presence of missense mutations has been reported previously in a GLUT1-deficient patient carrying a missense mutation in exon 7 of the GLUT1 gene (14), and for mutations within the active, sodium-dependent glucose transporter (SGLT1) gene, resulting in glucose–galactose malabsorption (23). In contrast, hemizygosity and nonsense mutations have been shown to reduce immunoreactivity (15).

It is currently unclear whether the GLUT1 defect is limited to the brain. GLUT1 is also highly expressed in fetal tissues, retina, placenta and testis (3,5), but to date no evidence of impaired GLUT1 function was observed in these tissues in GLUT1-deficient patients. In the family reported, the mutation apparently did not affect the reproductive function of the father.

In a recent report, Wang et al. (16) describe a nonsense mutation in the second R-X-G-R-R-motif in one patient. Parents carried the GLUT wild-type, suggesting that in this case a de novo mutation occurred. Their data also indicate that de novo mutations may be much more common than mutations transmitted in an autosomal dominant mode. However, our observation of GLUT1 deficiency within one family suggests a change in the diagnostic approach to this disease. For immunoblotting and glucose uptake studies, parents and siblings have previously been used as controls (14–16). Our data suggest that this might not be optimal as patients might be misdiagnosed as normal, especially as clinical information about family members is not always available in the laboratory, and the clinical spectrum of GLUT1 deficiency in adults not yet established. Consequently, for diagnostic analyses of GLUT1 deficiency we now use samples from independent individuals as intra-assay controls only.

In conclusion, this first report of autosomal dominant transmission of GLUT1 deficiency within three family members represents a novel mechanism of transmission. All patients shared a missense mutation within a conserved R-X-G-R-R-motif of the facilitator superfamily of membrane transporters. Immunoreactivity and glucose uptake studies suggested that the GLUT1 protein is present, but non-functional. Since GLUT1 deficiency is autosomally and dominantly transmitted, caused by haploinsufficiency, and symptoms possibly ameliorate with age, this treatable disorder should be considered in the diagnostic workup in pediatric and adult patients with epilepsy, mental retardation and motor disorders. In particular, special attention should be paid to unexplained hypoglycorrhachia, the biochemical hallmark of the disease.

MATERIALS AND METHODS

Clinical summaries

The index patient (II:3), now a 10-year-old girl, was born to non-consanguineous parents at term after an uneventful pregnancy and delivery. Mild spastic diplegia was observed at the age of 9 months. Over the next 2 years, developmental delay gradually became obvious. At 3 years, she developed complex partial seizures that were initially resistant to anticonvulsive treatment and eventually responded to valproate and acetazolamide. Phenobarbital was never used. Her mood was generally well, but there was often unpredictable and agressive behavior. Clinical features were consistent throughout the day and not related to meals or fasting. Physical examination showed moderate mental retardation, nystagmus, mild pyramidal signs of the legs and cerebellar ataxia. Head circumference was 51.0 cm (P20). Recent neuropsychological evaluation showed a total IQ score of 56 (WISC-R). Except for hypoglycorrhachia, extensive laboratory investigations were normal. Magnetic resonance imaging of the brain showed no abnormalities.

Patient II:5, the half-sister of patient 1, showed no abnormalities until the second year of life, when developmental delay and spastic diplegia became prominent and generalized tonic–clonic seizures occurred. Epilepsy continued since with at least one seizure per week despite anticonvulsive drug polytherapy. Her physical examination at 22 years revealed moderate mental retardation, cerebellar ataxia and spastic tetraplegia that predominantly involved the legs. She was able to walk unsupported. Head circumference was 54.2 cm (P30). As in the index patient, extensive laboratory investigations never revealed abnormalities except hypoglycorrhachia. Computed tomography scanning of the brain was normal at the ages of 7 and 19 years.

Patient I:2, age 46 years, is the father of patients 1 and 2. At the age of 3 years he developed generalized tonic–clonic seizures and myoclonic seizures. During childhood, seizures were more prominent and required anticonvulsant polytherapy. Currently seizures of one short myoclonic episode per day were strongly related to psycho-emotional circumstances and responded to valproate. Additional clinical features included nervousness and mood disturbances (depression), migraine and hypercholesterolemia. Fasting did not affect his clinical state. As a child he attended a school for mentally handicapped children and he has a job in a sheltered environment. His parents, who died some years ago, and his healthy daughters never experienced neurological symptoms. Except formild mental retardation his physical examination was normal. The patient did not agree to a lumbar puncture and no neuroimaging was performed. Head circumference was 58.5 cm (P60).

Lumbar puncture

Lumbar punctures were performed following a 4 h fast. Blood glucose was determined shortly before the procedure to avoid stress-related hyperglycemia. CSF was analyzed for cell number and type, and the concentrations of glucose, protein and lactate were determined.

Zero-trans-3OMG influx into erythrocytes

The assay has been described in detail elsewhere (17). Briefly, blood specimens were collected in citrate-dextrose-phosphate solution, immediately put on wet ice, and processed within 10 days. All procedures were performed at 4°C. Blood samples were washed three times in phosphate-buffered saline and aliquots were incubated with ¹⁴C-3-OMG. Uptake was terminated at 5 s intervals, the aliquots were washed twice, lysed, bleached and counted in a scintillation counter (Tricarb 2300; Canberra Packard).

Immunoblot

Following the isolation of red cell membranes from blood samples collected in citrate-dextrose-phosphate (5:1), 2 µg of protein was separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis and electroblotted onto a nitrocellulose membrane by standard protocols. Membranes were stained with Ponceau S to confirm the uniform efficiency of transfer. The GLUT1 protein was identified using a polyclonal rabbit GLUT1 antibody (Cymbus Biotechnology), raised against the C-terminus of GLUT1. Blots were also exposed to a monoclonal anti-human band 3 antibody (mouse UgG2a isotype) recognizing an epitope located in the cytoplasmic pole of the band 3 molecule (Sigma). Primary antibodies were detected by enhanced chemiluminescence (ECL; Amersham Life Science) and GLUT1 immunoreactivity was visualized against the immunoreactivity of band 3.

GLUT1 genomic analyses

DNA was isolated from white blood cells by standard protocols. Genomic DNA was used as a template for PCR with specific primers of the human GLUT1 gene as described (10,14). Amplification products were visualized by ethidium bromide staining on agarose gels and analyzed for SSCPs as described (24). No shifts were detected. Subsequently, PCR products were purified and forwarded to automated DNA sequencing. Following the analyses of exons 3, 4, 7 and 8, a mutation was found in exon 3 of II:3 (see Results). For confirmation, a separate set of PCR fragments of exon 3 was digested with HaeIII and the reaction products visualized by ethidium bromide staining in an 8% acrylamide gel. In addition, SSCP analyses and DNA sequencing were performed on all exons in the index patient II:3.

ACKNOWLEDGEMENTS

We acknowledge the dedicated work of Christine Fischer-Lahdo and Danute Bergmann as laboratory technicians. This work was supported by a grant provided by the Deutsche Forschungsgemeinschaft (DFG KL 1102/2-1).

+

These authors contributed equally to this work

§

To whom correspondence should be addressed. Tel: [phone]; Fax: [phone]; Email: [e-mail]

Figure 1. Pedigree of the reported family.

Figure 2. Automated DNA sequence showing the missense mutation (G272A) in II:3. (A) A fragment of ∼230 bp (exon 3 and intron–exon boundaries) was amplified from genomic DNA. The sense G→A transversion and the wild-type are indicated by arrows (nucleotide 272, numbering according to GenBank accession no. M20681, starting at ATG at positions 180–182). (B) HaeIII restriction map and restriction digest of the exon 3 PCR product, resolved on an 8% acrylamide gel, showing an additional fragment of 57 bp in I:2, II:3 and II:5. This fragment results from the loss of the HaeIII site between fragments 21 and 36 bp. The additional 57 bp fragment could not be observed in I:1 and an unrelated control (C).

Figure 3. Western blot analyses. Membranes were generated from erythrocytes, the protein concentration determined and 2 µg of protein submitted to SDS–acrylamide gel electrophoresis, transferred to a polyvinylidene fluoride (PVDF) microporous membrane and probed with antibodies specific for the GLUT1 C-terminus and band 3. No significant differences in immunoreactivity are seen in patients I:2, II:3 and II:5 compared with I:1 and an unrelated control (C).

Figure 4. Conformational model for GLUT1 in the membrane, based on Mueckler et al. (9). The protein is predicted to contain 12 transmembrane helices, with both the N- and C-termini intracellularly disposed. The two R-X-G-R-R-motifs within the intracellular loops joining transmembrane segments 2 and 3 and 8 and 9 are indicated as black boxes. The R-X-G-R-R motif carrying the mutation is shown enlarged with the nucleotide (G272A) and the amino acid exchange (G91D) indicated.

Table 1.

Clinical and biochemical findings

	Patient II:3	Patient II:5	Patient I:2
Age (years)	10	22	46
Mental retardation	Moderate	Moderate	Mild
Microcephaly	–	–	–
Seizures	Complex partial	Generalized tonic–clonic	Generalized tonic–clonic, myoclonic
Cerebellar ataxia	+	+	–
Nystagmus	+	–	–
Pyramidal signs	+	+	–
CSF glucose (mmol/l)	1.9	1.9	No lumbar puncture
Blood glucose (mmol/l)	6.6	5.4
Ratio CSF/blood glucose	0.29	0.35
CSF lactate (mmol/l)	1.0	1.5
3OMG uptake (% of controls)	42	57	57

Table 2.

Properties of the major facilitator superfamily of transporters related to the mammalian GLUT family and conserved R-X-G-R-R sequence within the intracellular loop between helices 2 and 3

Designation	Organism	Substrate	Aligned sequence
			— Helix 2 — │...loop....│ — Helix 3 —
GLUT1	Mammalian	Glucose	NRFGRRNS
GLUT2	Mammalian	Glucose	DTLGRIKA
GLUT3	Mammalian	Glucose	NRFGRRNS
GLUT4	Mammalian	Glucose	QWLGRKRA
GLUT5	Mammalian	Fructose	NKFGRKGA
GLUT8	Mammalian	Glucose	VDRAGRKLS
SNF3	Saccharomyces cerevisiae (yeast)	Glucose	DSYGRKPT
GAL2	S.cerevisiae	Galactose	DMYGRKKG
HXT1	S.cerevisiae	Glucose	DMYGRRIG
HXT2	S.cerevisiae	Glucose	DMYGRRIG
RAG1	Kluyceromyces lactis (yeast)	Glucose	DRWGRRIG
MAL61	S.carlsbergensis (yeast)	Maltose	DYMGNRYT
LAC12	K.lactis	Lactose	DWKGRKPA
ITR1	S.cerevisiae	Myo-inositol	DIFGRKRC
ITR2	S.cerevisiae	Myo-inositol	DVFGRRPC
HUP1	Chlorella kessleri (green alga)	Glucose	RNWGRKVT
STP1	Arabidopsis thaliana (green plant)	Glucose	RKFGRRLS
glcP	Synechoczystis PCC6803 (cyanobacterium)	Glucose	DRHGRIKT
Pro-1	Leishmania enriettii (protozoan)	Unknown	SKIGARLS
qa-y	Neurospora grassa (fungus)	Quinate	YFLGRRKS
qutD	Aspergillus nidulans (fungus)	Quinate	HFWGRRWG
AraE	Escherichia coli (bacterium)	Arabinose	FRLGRKYS
GalP	E.coli	Galactose, glucose	FKLGRKKS
XylE	E. coli	Xylose	NRFGRRDS
glf	Zymomonas mobilis (bacterium)	Glucose	IRFGRRGG
citA	E.coli pWR61 (bacterial plasmid)	Citrate	DRIGRRKG
TetC	E.coli pBR322 (bacterial plasmid)	Tetracyclines	DRFGRRPF

-X- can be any amino acid and R may be replaced by K [modified from Baldwin et al. (3)].