GLUT1-DSNews:Treatment,Diagnosis,Mechanisms
FannyMochel,MD,PhDCentredeRéférenceNeuroMétaboliqueAdulte
UniversityHospitalLaPitié-Salpêtrière,Paris–France
The multiple faces of GLUT1-DS
Grasetal,RevNeurol2013
Classical
Permanent
Paroxysmal
ASTROCYTE
NEURON
vessel
NEURON
KC
KC
glucose glucoseGLUT3
glucose-6-P
pyruvatelac
glucose-6-P
lactatepyr
acetyl-CoAacetyl-CoA
OAA
Ca2+
glutamate
Na+
glutamineglutamine glutamate
Glutamine synthetase
Glutaminase
MCT1
MCT4
MCT2
vessel
glucoseglucose
GLYCOLYSIS
OXIDATIVE PHOSPHORYLATION
C5KBHeptanoate
Glucose (astrocytic) Transporter Deficiency
Mochel,JNeurosciRes2017
TRIHEPTANOIN CYTOSOL
MITOCHONDRIA
Oxaloacetate
α-ketoglutarate
Citrate
ELECTRON TRANSPORT
CHAIN
ATP Succinyl-CoA KREBS
CYCLE
AcetylCoA Propionyl-CoA
3-ketopentanoate* 3-OHpentanoate*
*C5 ketone bodies
Glycerol
Pyruvate
Pyruvate
Heptanoate
Heptanoyl-CoA
Pentanoyl-CoA
Heptanoyl-CoA
Triheptanoin: Anaplerotic Approach
Mochel,JNeurosciRes2017
TRIHEPTANOIN CYTOSOL
MITOCHONDRIA
Oxaloacetate
α-ketoglutarate
Citrate
ELECTRON TRANSPORT
CHAIN
ATP Succinyl-CoA KREBS
CYCLE
Glycerol
Pyruvate
Pyruvate
Heptanoate
Heptanoyl-CoA
Pentanoyl-CoA
Heptanoyl-CoA
Triheptanoin: Anaplerotic Approach
AcetylCoA Propionyl-CoA
Mochel,JNeurosciRes2017
Phase II Open Label Study in GLUT1-DS
PHASE2• Triheptanoin• Patientdiary
2 months
Visit2 Visit3
PHASE1• Usualdiet
• Patientdiary
2 months
PHASE 3 • Backtousualdiet
• Patientdiary
2 months
Visit1
• 6minwalktest• 9holepegboard
• 6minwalktest• 9holepegboard• CGIphysicianandpatient• PlasmaC5-ketonebodies• 31P-MRS
Visit4
Primaryoutcomemeasure
§ Numberofepisodes
Secondaryoutcomemeasures
§ CGIscores§ 6minwalktest§ 9holepegboard§ PlasmaC5ketonebodies§ 31PMRS
Effect of Triheptanoin on Paroxysmal Manifestations
2.3 « much better »
5.7 « much worse » CGI
90%
0.0
0.1
0.2
0.3
0.4
0.5
0.6
Baseline Treatment Withdrawal
Rest Activation Recovery
*
0
10
20
30
40
50
Baseline Treatment Withdrawal
*
*
A
B
Num
bero
fparoxysmaleventsp
erm
onth
Mocheletal,JNeurolNeurosurgPsychiatry2016
Mirror
Adjustable handle
31P-MR spectroscopy: a window on brain energy metabolism
0.00
0.10
0.20
0.30
0.40
0.50
0.60
Mov Dis 2012 Neurology 2015
Rest Activation Recovery
* *
* *
ADP+PCr ATP+Pi
0.0
0.1
0.2
0.3
0.4
0.5
0.6
Baseline Treatment Withdrawal
Rest Activation Recovery
*
0
10
20
30
40
50
Baseline Treatment Withdrawal
*
*
A
B
Brain Metabolic Response to Triheptanoin
Mocheletal,JNeurolNeurosurgPsychiatry2016
Sustained Therapeutic Benefit with Triheptanoin in GLUT1-DS
Improvedcognitiveperformance(after2years)- 26yoldwoman:TIQ72(63),PIQ75(67),VIQ72(62)- 9yoldboy:improvedschoolingperformance
0
10
20
30
40
50
Baseline Treatment Withdrawal Resumption short term
Resumption long term
Num
ber
of p
arox
ysm
al e
vent
s
* *
*
* *
90%
90%
PhaseIIIStudy
Mocheletal,unpublished
Grasetal,AnnNeurol2017
Red blood cells test for GLUT1-DS
Flowcytometry
a développé glut1
Un diagnostic en seulement 24h
METAglut1 est un test de diagnostic in vitro permettant de quantifier le niveau d’expression de GLUT1 directement à la surface de globules rouges, par cytométrie en flux.
Ce test devrait permettre d’améliorer drastiquement le diagnostic de GLUT1DS, en détectant les patients le plus tôt possible, et en leur évitant une longue errancemédicale.
La sensibilité et la spécificité sont évaluées à 80% et 100% respectivement –publication en cours.
Phénotypeclassique
Phenotypesintermédiaires
?? Autres phénotypes ??
Diagnostic actuel
Capacité de criblage de METAglut1
Detection rate: 80% - Specificity 100%
ForfaitInnovation
GLUT1-DS: not just energy deficiency
Tangetal,NatCommun2017
Recognizing that our experiments may not have detected subtlealterations in the Glut1 DS BBB, we carried out the followingadditional experiments. First, we substituted the albumin andIgG molecules with the labelled tracer, TMR-biocytin, which,owing to its much smaller size (B900 daltons) is expectedto traverse a leaky BBB with much greater ease, thereforeenhancing the sensitivity of our assay21. Next, we used our modelmice to examine expression levels of a subset of signatureBBB endothelial genes, perturbations of which are known todisrupt the BBB22–28. Finally, we investigated CSF levelsof proteins normally restricted to the serum in a cohort ofGlut1 DS patients. Consistent with our earlier findings, we foundno difference in levels of fluorescently labelled biocytin in thebrain parenchyma of mutant and control animals, whereasabundant and intense signal was detected in the CNS of micetreated with the BBB-disrupting agent kainic acid (SupplementaryFig. 3a). In support of an intact BBB, as assessed by our labelledtracer assays, we detected no major alterations in the cerebralexpression of Cldn3, Cldn5 and Cldn12 (tight junction proteins),Abcg2 and Abcb1b (active efflux transporters), Cdh5 (adherensjunction protein) and Pvlap (plasmalemma vesicle-associatedprotein) (Supplementary Fig. 3b–h). Finally, we foundno evidence of elevated CSF serum proteins and thusa compromised BBB in Glut1 DS patients evaluated by usbetween the ages of 6 months and 10 years (Mean CSF glucosein mg dl! 1: 32.58±0.57, N¼ 44; Mean CSF lactate in mmol l! 1:0.93±0.03, N¼ 44; Mean CSF protein concentration in mg dl! 1:
20.52±1.10 versus 22±5 (ref. 29) P¼ 0.19, one sample t test,N¼ 44 and N¼ 599 respectively). We conclude that althoughthere is a distinct loss of the brain microvasculature in adultGlut1 DS mice, the BBB in these animals and likely in humanpatients remains largely intact.
Although our results clearly demonstrate that Glut1 deficiencyresults in a poorly developed brain microvasculature, how thisevolves is unclear. Accordingly, in a concluding set of analyses, wesought to explore potential mechanisms linking Glut1 to theprocess of angiogenesis. Two results appear to establish sucha link. First, we discovered a sharp decline in levels of Vegfr2in 2-week old Glut1 DS mutants. Transcripts as well as proteinwere diminished in expression. Importantly, the relative paucityof the molecule was restricted to blood vessels, remainingunchanged in the capillary-depleted fraction (parenchyma) ofthe mutant brain (Fig. 3a–c). Vegfr2 is a major positive-signaltransducer expressed in ECs as capillaries form and expand,establishing it as a potential mediator of the microvasculaturedefects in Glut1 DS30,31. We also investigated the effects of Glut1deficiency on glycolytic flux, a modest inhibition of which isknown to perturb blood vessel formation32,33. To do so, we usedastrocytic cells from Glut1 DS mutant mice and WT controls toassess rates of extracellular acidification (ECAR) and oxygenconsumption (OCR), measures of glycolytic flux andmitochondrial respiration respectively34. Both basal glycolyticflux and maximum respiratory capacity declined significantly inmutant cells (Fig. 3d; Supplementary Fig. 4a,b). This suggests that
08 weeks 20 weeks
WTMut
*** ***WT
*** ***
Glu
t1 D
S m
utan
t
2 weeks 8 weeks 20 weeks
Wild
type
25 µm
0
5
10
15
WTMut
8 weeks 20 weeks
*** ***
Bra
in c
apill
ary
dens
ity(a
ggre
gate
leng
th in
µm p
er u
nit v
olum
e)
1,800
800
1,600
1,400
1,200
1,000
20 wk2 wk 8 wkAge (weeks)
Mut
Ave
rage
ves
sel s
ize
(µm
)
100
80
60
40
20
Num
ber
of v
esse
l bra
nch
poin
ts
a
b c d
Figure 1 | Cerebral microvasculature defects in Glut1 DS. (a) Thalamic sections of Glut1 DS model mice and littermate controls stained with labelled lectinreveal cerebral angiogenesis defects and diminution of the microvasculature of mutants at 8 and 20 weeks of life. Graphs quantifying (b) aggregate cerebralcapillary length, (c) mean vessel size and (d) average capillary branches in the mutants and controls illustrate the diminished size of the microvasculaturein adult Glut1 DS mice. ***, Po0.001, t-test, nZ9 regions from each of NZ3 mice of each genotype examined.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14152 ARTICLE
NATURE COMMUNICATIONS | 8:14152 | DOI: 10.1038/ncomms14152 | www.nature.com/naturecommunications 3
BrainmicrovasculaturedefectinGLUT1-DSmouse
a truly effective treatment for the disorder, and much remainsto be learned about the cellular pathology linking Glut1mutations to the characteristic brain dysfunction seen in patients.Here, we attempted to address these deficiencies in model mice.Three principal findings emerge from our study. The firsthighlights novel defects of the brain microvasculature in theGlut1 deficient organism. These involve both a delay in the initialexpansion of the cerebral capillary network as well as a laterimpairment in its maintenance. Thus in mature mutants,a striking diminution of the capillary network became evident.Remarkably, though, no discernible alteration in BBB integritywas noted. Our second salient finding demonstrates that genereplacement using an AAV9 vector constitutes a relativelystraightforward, safe and highly effective means of treatingGlut1 DS. Glut1 repletion in neonates had a major therapeuticeffect on mutant mice, arresting the onset of a motor phenotype,enabling the development of the brain microvasculature andrestoring parameters typically perturbed in the mutant to thewild-type state. Our final notable result addresses the temporalrequirement for the Glut1 protein and allows one to definea window of opportunity to treat Glut1 DS by means of restoringthe protein. Initiating a treatment in juvenile animals that were atleast partially symptomatic, reversed hypoglycorrhachia,improved motor performance, accelerated the growth of thebrain and facilitated the expansion of the cerebral microvascu-lature. Akin to effecting repletion during neonatal and earlypostnatal life, restoring the protein to the fully symptomaticadult mutant raised characteristically low brain glucose levels.However, in contrast to the outcome of early repletion,augmenting the Glut1 protein late failed to mitigate any othermajor feature of the Glut1 DS phenotype. We suggest that theaccumulating damage sustained by the Glut1 deficient brain as itattempts to establish and refine important neural circuitseventually precludes the possibility of therapeutic rescue even ifoverall cerebral Glut1 expression is eventually restored. Ourresults predict that Glut1-deficient patients will be mostresponsive to gene replacement-type treatments relatively early
in the course of the disease. Nevertheless, there exists aperiod during the symptomatic phase of the disease when Glut1DS can be effectively treated. These findings argue for the useof new-born screens to identify the pre-symptomatic Glut1DS patient.
Although widely expressed, Glut1 is particularly abundant inthe ECs of the brain microvasculature47. Moreover, braindysfunction is a signature feature of Glut1 DS. Examining thecapillary network of our mutant mice therefore appeared tobe logical way to initiate our investigation. Still, we were startledby how profoundly loss of one copy of the Slc2a1 genehad affected the elaboration of the brain microvasculature.Myriad genes govern CNS angiogenesis (ref. 48 and referencestherein), but few trigger such marked defects without a totalablation of their activities. Interestingly, the diminution of thebrain capillary network that we observed became evident only inmature mice and did not compromise BBB integrity. The firstfinding—that of angiogenesis defects—is consistent with those oftwo prior reports, in which Glut1 deficient model fish and micerespectively were studied49,50. The second—defects ofbarriergenesis—is not. While the distinct findings in Glut1deficient fish has a logical explanation and likely stems from amuch greater (B90% versus B50% in our study) level of Glut1knockdown49, the report of Winkler et al.50 is more curious anddifficult to reconcile with ours. In their study, massive (410-fold)extravasation of serum proteins was observed as early as 2 weeks.Cerebral vasogenic edema ought to have followed but,surprisingly, was not reported. More importantly, such edemahas neither been cited in the literature as characteristic ofGlut1 DS patients nor detected by us over a three decade periodin the clinic. These clinical observations correlate to a greaterextent with our findings of an intact BBB in model mice thanthey do with those of Winkler and colleagues. Still, oneoft-overlooked but important factor that might explain thedistinct findings in the study of Winkler et al. is a difference inmouse strain background. Whereas our mice were maintained ona pure 129/SvJ background, the mutants in the study by Winkler
Mutant (PND3) + AAV9-Glut1
Lect
in
Mutant (PND3) + vehicle
25 µm
Mut + vehicle
Mut (2wk) + AAV9-Glut1
Wild typeMut (8wk) + AAV9-Glut1
Mut (PND3) + AAV9-Glut1
0
NS
NS***
c
0
NS
NS
***
***
b d
NS
NS***
0
5
Wild type Mutant (2 wk) + AAV9-Glut1 Mutant (8 wk) + AAV9-Glut1
Bra
in c
apill
ary
dens
ity(a
ggre
gate
leng
th in
µm p
er u
nit v
olum
e)
2,000
1,500
1,000
500
Ave
rage
ves
sel s
ize
(µm
)
100
80
60
40
20 Num
ber
of v
esse
lbr
anch
poi
nts
15
10
a
Figure 7 | Brain capillary network restored following early but not late repletion of the Glut1 protein. (a) Representative immuno-histochemicalphotomicrographs of the thalamic microvasculature in relevant controls and mutants treated with AAV9-Glut1 at various stages of the disease.Note persistent fragmentation and reduced complexity of the capillary network in vehicle-treated mutants and mutants treated at 8 weeks of age,an intermediate capillary density in cohorts treated at 2 weeks of age and a relatively normal microvasculature in mice administered virus at PND3.Quantitative estimation of (b) aggregate cerebral capillary length, (c) mean vessel length and (d) number of branches per capillary in the different cohortsof mice. ***, Po0.001, one-way ANOVA, nZ9 regions from NZ3 mice in each cohort examined.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14152
10 NATURE COMMUNICATIONS | 8:14152 | DOI: 10.1038/ncomms14152 | www.nature.com/naturecommunications
RescuebyearlyrepletionofGLUT1-DS