Modeling of an epilepsy-related neurodevelopmental disorder caused by de novo GNB1 missense mutations and identification of targeted treatments Poster Sophie Colombo1, Emmanuel K. Ozoruonye1,4, Sabrina Petri1, Mu Yang1,4, Michael J. Boland1,2, Wayne N. Frankel1,3 and David B. Goldstein1,3 #560.13 1Institute for Genomic Medicine, 2Department of Neurology, 3Department of Developmental Genetics, 4NeuroBehavior Core facility, Columbia University, New York, NY 10032 USA.

Rationale The Gnb1K78R mouse model phenocopies aspects of the human disease Methods

 Molecular diagnosis via whole exome sequencing has revealed that heterozygous de a) K78R/+ WT b) Body weight Surface righting Negative geotaxis (90 degrees) c) Ambulatory distance Vertical count Center time  CRISPR/Cas9 knock-in mouse model was generated for the K78R human variant. 8 30 30 2000 200 300 WT novo mutations in GNB1 are associated with a rare neurodevelopmental disorder WT The mutation was generated and is maintained and studied on the C57BL/6NJ strain ** K78R/+ *** ** K 78R/+ characterized by the following main features: 6 ** 1500 150 background for results in this presentation, except for Fig.1f (BL/6NJ x FVB F1). *** 20 20 200 * * • developmental delay 4 * 1000 100

• hypotonia 10 10 100 O pen field  2 500 50 C entertim e(s) Neurobehavioral tests: Verticalcount (#)

• seizures (g) Bodyweight Surface righting (s) righting Surface Negative geotaxis (s) geotaxis Negative Ambulatory distance (cm) distance Ambulatory • Pups were weighed and tested on even days for developmental milestones and odd • ophthalmological defects P6 0 0 0 0 0 0

4 6 8 4 6 8 4 6 8 1 0 20 3 0 40 5 0 60 1 0 20 30 40 5 0 60 1 0 20 30 40 5 0 60 10 10 10 days for ultrasonic vocalizations (USV). Developmental milestones tests included  48 affected children have been identified in the last 2 years, with 29 different PND PND PND Tim e (m in) Tim e (m in) Tim e (m in) surface righting reflex, negative geotaxis (90 and 180 degrees) and vertical screen mutations in GNB1 (2 frameshifts, 2 splice sites, and 25 missense) (Petrovski et al., Negative geotaxis (180 degrees) Vertical screen holding U ltrasonic Vocalization holding. Pups were allowed 2 trials per tests, and time to success was recorded, up to AJHG 2016, Steinbrucke et al., Neuroloy 2016, Brett et al., AJMG 2017, Lohmann et Paw print area size Max Intensity at % Stand tim e 40 25 250 30 seconds which corresponded to failure. The mean of the two trials was used for * 0.5 80 0.4 WT

al., HMG 2017, unpublished). )

20 200 2 K78R/+ analysis. For USVs pups were separated from the dam, placed in a box with an 30 0.4 60 0.3 ultrasound microphone and recorded for 3 minutes (Avisoft Bioacoustics software). Identified GNB1 Mutations * 15 150 0.3 20 40 0.2 10 100 * • CatWalk XT (Noldus Information Technology) is a modern automated test for gait K78R 0.2

G77S Catwalk D76G I80N 10 20 0.1

GNB1 Max intensity functions and locomotion. It consists of an illuminated walled glass walkway (130 cm x (CCDS34) G77R 5 50 Standtim e(s) Number of calls Number of 0.1 D76E I80T (cm area Print G77A (s) screen Vertical Negative geotaxis (s) geotaxis Negative

(duration/% of stand) of (duration/% 10 cm) and a high-speed camera underneath. Mice are habituated for three daily M101V 0 0 G53E A326T 0 0.0 0 0.0 C114Y 4 6 8 4 6 8 5 7 9 L95P 10 10 11 sessions before the experiment. Light is reflected and illuminates the stimulus PND PND PND (footprint) when downward pressure is applied. Mice are allowed to traverse the d) e) f) Forelimbs Hindlimbs Forelimbs Hindlimbs Forelim bs H indlim bs G64V K89R A106T Electroconvulsive minimal clonic forebrain seizures Electroconvulsive psychomotor seizures walkway as many times as needed to obtain at least 3 fluent crossings (without L30F R52G K337Q (high frenquency stimulus) (6Hz stimulus) D118G c.915_916del c.917-1G>T stopping or hesitations). Parameters automatically collected by the software include c.268-1G>T A92T 50 30 stride length, width, base of support, distance between ipsilateral prints, cadence, WT H91R A92D P96L **** Female Female support formulas, step sequence, phase lags, and walking speed. c.272_275del P94S M ale 25 M ale 40 • The open field is general test for locomotor activity. Each mouse is gently placed in the  Most mutations identified in the individuals were also found somatically mutated in a center of a clear Plexiglass arena (40 x 40 x 40 cm, Med Associates) lit with dim light 20 variety of cancers, mainly myeloid and B cell neoplasms (Yoda et al., Nat Medicine (~30 lux), and is allowed to ambulate freely for 60 min. Infrared (IR) beams embedded 2015). So far, only one child with a germline mutation was reported with both the 30 15 along the X, Y, Z axes of the arena automatically track distance moved, horizontal iRM S current neurodevelopmental syndrome and acute lymphoblastic leukemia (Brett et al., AJMG iRM S current movement, vertical movement, stereotypies, and time spent in center area. K78R/+ 2017). 20 10

WT WT  Seizure susceptibility was determined using electroconvulsive threshold (ECT) tests,  Our clinical study of the affected individuals’ seizure characteristics suggests that K78R/+ K78R/+ using the Ugo Basile Electroconvulsive Device and transcorneal electrodes, as GNB1 mutations may lead to Electrical Status Epilepticus during Sleep (ESES), Figure 1. Heterozygous Gnb1K78R/+ mice present with developmental delay, motor function deficits, susceptibility to generalized seizures and excessive SWDs. a) Gnb1K78R/+ pups are smaller than WT littermates. b) described in Frankel, 2001 Genomics (PMID: 11414758). High frequency ECT which is characterized by continuous (or near continuous) spike-wave discharges Developmental milestones and USV testing show developmental delay of Gnb1K78R/+ pups. Gnb1K78R/+ pups gain weight more slowly, do not show progress in the surface righting reflex test, have delayed progression in the negative settings: 299 Hz, 1.6 ms pulse width, 0.2 s duration, variable current. Low frequency (SWDs) during NREM sleep. It is often associated with an encephalopathy (ESES geotaxis test, and are very quiet throughout the USV test. c) Adult mice show hypoactivity in the open field test, while the catwalk test reveals gait anomalies. The mice seem to walk on their tiptoe for the forelimbs while they drag their ECT settings: 6 Hz, 0.2 ms pulse width, 3 s shock duration, variable current. Tests syndrome) characterized by epileptic seizures, continuous SWDs in NREM sleep, hindlimbs. d) Gnb1K78R/+ mice have a very low threshold to electroconvulsive minimal clonic forebrain seizures, compared to WT littermates but e) their electroconvulsive threshold in the 6 Hz psychomotor seizure model is unchanged. were performed approximately daily in individual mice until threshold was reached. global or selective regression of cognitive functions, and motor impairments. Many f) Preliminary video-EEG recordings of cortically implanted Gnb1K78R/+ and WT littermates (N=2 each genotype) suggest that Gnb1K78R/+ confers a significant number of spike-wave discharges (SWD), ranging in length from 0.5s to 15s Integrated root mean square (iRMS) was calculated from stimulus parameters to children will have absence seizures, some myoclonic seizures and others will have and appear in clusters. We have not yet analyzed the relationship between SWD incidence and awareness or diurnal activity. * p<0.05; ** p<0.01; *** p<0.001; **** p<0.0001 (Two-way ANOVA with multiple comparisons). describe threshold, and group means were calculated for genetic analysis. focal (partial) motor seizures, all of which were observed in our cohort. K78R  Spontaneous seizure events detected by video-EEG . Adult mice (6 wks or older) were Spontaneous activity in Gnb1 neurons shows aberrant firing pattern implanted with electrodes for continuous EEG monitoring. Subdural cortical electrodes were implanted under general anesthesia as indicated in the image below. Signal was DIV14 WT K78R+/- DIV14 WT K78R+/- GNB1 Function a) a) b) c) detected with a Grael 48 EEG amplifier and acquired on a computer using Profusion 5 software (Compumedics, USA), synchronized to a high resolution video camera.  GNB1 encodes an ubiquitously expressed β subunit (Gβ1) of membrane bound GTPases heterotrimeric G proteins (Gαβγ) which are activated by G protein-coupled receptors (GPCRs).  Activity analysis of cultured neurons from Gnb1K78R mice utilizing Maestro MEA system (Axion BioSystems). Briefly primary cortical neurons from P0 pups were dissociated MAP2/Gnb1/Dapi Gnb1/VGlut1/Dapi DIV14 WT K78R+/- DIV14 WT K78R+/- and plated on 48-well MEA plates, with each well containing 16 electrodes. Neurons were maintained in NBA/B27 media and recorded for 15 minutes every other day. Parameters including firing rate, bursting properties, and network properties were obtained using an in house program. Known epilepsy Neurodevelopmental gene, GNAO1 disorders: GNB5 LoF Nakamura K. et al., mutations Lodder E.M. et al., 2013 2016; GNB1 de novo mutations Petrovski S. et al., Gnb1/GFAP/Dapi Gnb1/Gad67/Dapi 2016 Adapted from Li, J. et al., 2002 d) e) f)

 Gβγ regulate a variety of effectors and signaling pathways, such as:

• Activation of MAPK and Pi3K pathways Gad67 Gnb1 • Activation of GIRK (K+) channels Non-induced D A I 2+ • Inhibition of Cav (Ca ) channels b) P0 cortex c) DIV15 DIV21 DIV15 • Activation of Adenylate cyclase • Activation of Phospholipase Cβ Gnb1  Most of the identified mutations localize in a region that has been shown to participate Gnb1 D: Dopamine to the interface of interaction of Gβ with Gα subunits and effectors like GIRK and Cav A: Acetylcholine I: Isoproterenol channels. We thus hypothesize that defective interaction with partner proteins - either β-actin β-actin increased or decreased - will affect neuronal activity.

Figure 2. Normal expression and localization of Gβ1 in Gnb1K78R/+ neurons. a) Figure 3. MEA reveals increased burst duration accompanied by decreased number of bursts in Gnb1K78R/+  Co-immunoprecipitation (Co-IP) assays were performed in HEK293 cells using flag- General Approach Immunocytochemistry on cortical primary neurons reveals somatic localization of Gβ1 in neurons neurons. Recordings from WT and Gnb1K78R/+ cortical neurons reveal a phenotype that emerges after neuronal tagged WT or mutant GNB1 and GFP-tagged partner proteins transfected following (MAP2), both excitatory (VGlut1) and inhibitory (Gad67), but not in astrocytes (GFAP). No staining maturation at DIV15. Data from 10 different litters averaged and normalized to WT. a) Gnb1K78R/+ neurons tend to fire the Lipofectamine 3000 protocol (ThermoFisher). Proteins were pulled down using a difference was observed between Gnb1K78R/+ and WT neurons. Gβ1 is more strongly expressed in more during the developmental period of the network but slow down during the mature period compared to WT b) rabbit polyclonal anti-GFP antibody that was bound to anti-rabbit magnetic beads Gad67- expressing inhibitory neurons than other neuronal types (bottom right panel). b) Western Blot of Gnb1K78R/+ neurons show decreased number of bursts at maturity c) while they show increased mean duration of bursts (ThermoFisher). Western Blot was performed with anti-Flag and anti-GFP antibodies. Transcriptional cortex shows normal Gβ1 protein expression. c) Western Blot of primary cortical neurons before or after d) and increased inter-burst intervals. e) Gnb1K78R/+ neurons also show increased mean spikes in bursts compared to WT Generate Assess & characterize profile of cortical Gnb1K78R/+ seizure phenotype neurons activation of GPCRs by neurotransmitters for 2 hours shows no difference in Gβ1 expression. f) and decreased mean frequency of spikes in bursts, which correlates with the very long burst duration observed. Summary In vitro: In vivo: MEA of neuronal Electroconvulsive Mutations in GNB1 affect interaction and activation of downstream effectors cultures, patch-  K78R/+ Threshold, video- Heterozygous Gnb1 mice show > 50% attrition at P0, while homozygous clamp on brain Children with EEG and behavior K78R/K78R K78R/+ slices and neuronal Gnb1 mice are embryonic lethal. Heterozygous Gnb1 mice present with GNB1-related analysis a) c) a) b) c) epilepsy cultures developmental delay, motor deficits and very low threshold to minimal Non-induced D A I electroconvulsive forebrain clonic seizures. Numerous clustered and long-lasting DIV15 DIV21 DIV15 Screen candidate SWDs are readily observed on video-EEG, which could be a sign of absence seizures. therapeutic compounds Overall, the K78R mouse model shares several aspects of the patient’s phenotype. Generate Non-induced D A I heterologous DIV15 DIV21 DIV15 pAkt cellular models Compare & contrast Interrogate  Gβ1 expression and localization is not affected by the K78R mutation. Gβ1 is In vitro & in vivo potential pathological changes Gnb1 Akt expressed in the somatic region of both glutamatergic and GABAergic neurons, with a misregulated pERK stronger expression in GABAergic neurons. Assess effectors β -actin molecular consequences  Gnb1K78R/+ cortical neurons burst less often than WT neurons, but for significantly of the mutations b) ERK P-p70S6K longer period of times. The concomitant increase in inter-burst intervals suggests a β -actin longer recovery period from the bursts.

Acknowledgements p70S6K  K78R and D76G variants increase Gβ1 interaction with Gαo subunits, but decrease it D: Dopamine P70S6K with other Gα types, suggesting potential deficits in Gαo-associated GPCR signaling. This study was funded, in part, by Grant # NIH R37 NS031348 (to WNF). A: Acetylcholine β-actin I: Isoproterenol K78R, D76G and I80N variants reduce interaction with Adcy2 and mTOR suggesting that these major pathways may contribute to the phenotype. The K78R variant shows The mouse model K78R was generated by Victor Lin at the Transgenic Mouse Core New D.C. and Wong Y.H. J. Mol. Signal. 2007 defective interaction with GIRK1, while D76G shows increased interaction. The facility at Columbia University Medical Center. differential effects of mutations on certain interactions may contribute to the Figure 4. Gβ1 interaction with Gα subunits and downstream effectors is affected by GNB1 mutations. a) Figure 5. Intracellular signaling pathway activation is affected by the K78R mutation. a) Schematics of phenotypic variability observed in affected individuals. Emmanuel Ozoruonye and Stephen Owens from the Neurobehavior Core facility at Representative Western Blot of Co-IP between Gβ1-WT or Gβ1-K78R and Gα subunits. b) Quantification of the the regulation of MAPK/Erk (top) and Pi3K/Akt/mTOR (bottom) pathways by GPCRs. b) Western Blot from Columbia University Medical Center helped with behavior studies. interaction between Gβ1-WT, Gβ1-K78R or Gβ1-D76G and Gα shows increased interaction with Gαo but decreased primary cortical neurons before or after activation of GPCRs for 2 hours with neurotransmitters shows that  Gnb1K78R/+ cortical neurons show defective activation of mTOR pathway following interaction with Gαs, Gαi and Gαq subunits. c) Quantification of the interaction between Gβ1-WT, Gβ1-K78R, Gβ1- activation of the MAPK pathway is not affected by the K78R mutation. c) Activation of Pi3K/Akt pathway is GPCR activation, while MAPK/Erk and Pi3K/Akt pathways are not affected. This D76G or Gβ1-I80N and β-arrestin1, Rack1, Adcy2, Kcnj3 (GIRK1) and mTOR shows variability of the interaction not affected while activation of mTOR pathway (p70S6K) is reduced in Gnb1K78R/+ neurons compared to WT suggests that Gβ1 can regulate mTOR pathway through its direct interaction with defects depending on the mutation. All mutations tested show decreased interaction with Adcy2 and mTOR. following GPCR activation, suggesting direct regulation of mTOR signaling by GNB1 (red dotted arrow in a)). mTOR (as seen in Co-IP assay).