Movement disorder in GNAO1 encephalopathy associated with gain-of-function mutations

Abstract

Objective:

To define molecular mechanisms underlying the clinical spectrum of epilepsy and movement disorder in individuals with de novo mutations in the GNAO1 gene.

Methods:

We identified all GNAO1 mutations reported in individuals with epilepsy (early infantile epileptiform encephalopathy 17) or movement disorders through April 2016; 15 de novo mutant alleles from 25 individuals were introduced into the Gα_o subunit by site-directed mutagenesis in a mammalian expression plasmid. We assessed protein expression and function in vitro in HEK-293T cells by Western blot and determined functional Gα_o-dependent cyclic adenosine monophosphate (cAMP) inhibition with a coexpressed α_2A adrenergic receptor.

Results:

Of the 15 clinical GNAO1 mutations studied, 9 show reduced expression and loss of function (LOF; <90% maximal inhibition). Six other mutations show variable levels of expression but exhibit normal or even gain-of-function (GOF) behavior, as demonstrated by significantly lower EC₅₀ values for α_2A adrenergic receptor–mediated inhibition of cAMP. The GNAO1 LOF mutations are associated with epileptic encephalopathy while GOF mutants (such as G42R, G203R, and E246K) or normally functioning mutants (R209) were found in patients with movement disorders with or without seizures.

Conclusions:

Both LOF and GOF mutations in Gα_o (encoded by GNAO1) are associated with neurologic pathophysiology. There appears to be a strong predictive correlation between the in vitro biochemical phenotype and the clinical pattern of epilepsy vs movement disorder.

Epilepsy is one of the most common neurologic disorders in the United States.¹ Severe early-onset seizures can result in epileptic encephalopathy.² There are at least 52 different gene mutations that cause early infantile epileptiform encephalopathy (EIEE).³ Mutations in the same genes also cause other neurodevelopmental abnormalities.^4,5 A key challenge in genetic epilepsies has been understanding the genotype–phenotype relationships of causal genes; this may require biochemical analysis.⁶

Mutations in the heterotrimeric G protein Gα_o (GNAO1 gene) cause an autosomal dominant epileptiform encephalopathy (EIEE17, OMIM: 615473).⁷ In this original article, all 4 mutations were characterized as having loss of function (LOF).⁷ More recently, an extended spectrum of GNAO1 encephalopathies was identified in which individuals had movement disorders but minimal to no seizures.^8,9 Here, we use GNAO1 encephalopathy to describe the entire clinical spectrum of individuals with pathologic GNAO1 mutations. A genotype–phenotype correlation was also recently noted¹⁰; certain mutations (e.g., E246K and several R209 alleles) were found specifically in children with hypotonia, developmental delay, and chorea but no epilepsy. However, the mechanistic basis for this genotype–phenotype correlation remains unknown.

GNAO1 encodes the α subunit of G_o, a heterotrimeric G protein (consisting of α and βγ subunits) that is highly abundant in the CNS, comprising about 1% of brain membrane protein.¹¹ G_o mediates signals from a wide range of inhibitory receptors including GABA_B, α_2A adrenergic, adenosine A₁, and dopamine D₂ receptors. A canonical function of Gα_i/o family proteins is inhibition of cyclic adenosine monophosphate (cAMP).¹² The identification of GOF mutations in ADCY5 (which encodes adenylate cyclase 5, the enzyme that produces cAMP) in dyskinesia and chorea patients directly links reduced cAMP to involuntary movement disorders.^13–17 This is consistent with GOF behavior of Gα_o.

Here we assessed expression and function of human mutations in GNAO1.^{7–9,18–27} Our biochemical analysis identified both LOF and gain-of-function (GOF) behaviors; the latter are associated with movement disorders while the former are primarily found in individuals with epileptiform encephalopathies. This mechanistic insight has important implications for therapies of GNAO1 encephalopathies.

METHODS

The detailed sources of reagents and antibodies, and the methods for mutagenesis, cell culture, transfections, Western blot, and cAMP assays, are described in the e-Methods at Neurology.org. The primer sets used in this study are listed in table e-1.

RESULTS

Most pathogenic GNAO1 mutations cause reduced Gα_o protein expression.

To evaluate 15 mutations in GNAO1 (figures 1A and e-1) that were previously identified in patients with epilepsy or other neurodevelopmental disorders,^{7–9,18–27} we performed Western blots in HEK-293T cells transiently transfected with each mutant. The majority of mutants (12) showed significantly lower protein levels than wild-type (WT) Gα_o, whereas 3 separate Arg²⁰⁹ mutant alleles showed essentially normal expression, as did the previously described GOF mutant G184S²⁸ (figure 1, B–E).

Figure 1. Location and protein expression levels of human GNAO1 mutations related to epileptic encephalopathy.

(A) Location of 15 mutations (G40R, G42R, D174G, T191_F197del, L199P, G203R, R209C, R209G, R209H, A227V, Y231C, E246K, N270H, F275S, and I279N) mapped on the Gα_o amino acid sequence. (B, C) Representative Western blots of Gα_o protein expression from HEK293T cells transiently transfected with each Gα_o mutant. (D, E) Quantification of relative protein levels of each Gα_o mutant compared to wild-type Gα_o. Graphs are the result of 3 independent experiments and data are presented as mean ± SEM. **p < 0.01, ****p < 0.0001 using 1-way analysis of variance with Bonferroni post hoc test for pairwise comparison.

Validation of an in vitro assay to assess function of GNAO1 mutations.

We used inhibition of forskolin-stimulated cAMP levels as a functional readout to allow efficient quantification of Gα_o effects with full concentration curves for agonist-mediated signaling. We cotransfected Gα_o plasmids with α_2A adrenergic receptor (α_2A AR) cDNA.²⁹ Robust inhibition of cAMP by the α₂ adrenergic agonist UK14,304 depends on both the transfected receptor and the Gα_o protein (figure e-2, A and B). A pertussis toxin (PTX)–insensitive Gα_o (C351G)³⁰ was used to create all mutant constructs and WT, enabling inactivation of endogenous G_i/o proteins using PTX.

When PTX eliminates the inhibitory signaling through G_i/o, α_2A AR couples weakly to G_s and stimulates adenylate cyclase (AC),³¹ resulting in increased cAMP levels after PTX treatment in the absence of a transfected Gα_o (figure e-2B). Consequently, the fractional inhibition of AC by Gα_o mutants was assessed as the decrease from the high control level of cAMP (PTX but no Gα_o—0%) to the low level with WT Gα_o (PTX and PTXi Gα_o—100%). This is termed normalized % inhibition. PTX treatment did not alter the ability of the PTX-insensitive Gα_o to mediate α_2A AR-stimulated cAMP inhibition (figure e-2C). Hence, this is a good system to study functional consequences of Gα_o mutations.

Nine GNAO1 mutations result in LOF or partial LOF.

Six mutants showed essentially complete LOF with normalized inhibition below 40% (figure 2A and table 1). All of these mutants also showed low expression levels (11%–35% of control; figure 1). Three mutants (A227V, Y231C, and I279N; figure 2, A and E, and table 1) had intermediate effects and were classified as partial LOF (PLOF) mutants. The I279N mutation showed a modestly reduced maximal inhibition of cAMP levels (figure 2E and table 1). This mutant also produced a very low EC₅₀ value (0.7 nM vs 25 nM for WT Gα_o), which might explain the discrepancy between the low expression levels (15% of WT) while maintaining good maximal inhibition in the cAMP inhibition assay. Based on the maximum inhibition below 90% of control, however, we classified this mutation as PLOF (table 1).

Figure 2. Effect on α_2A adrenergic receptor (α_2A AR)–mediated cyclic adenosine monophosphate (cAMP) inhibition by GNAO1 mutants.

(A–C) Dose–response curves of representative GNAO1 mutants. (A) Dose–response curves of loss-of-function (LOF) and partial loss-of-function (PLOF) mutants (G40R, L199P, N270H, F275S, A227V, Y231C) show changes in cAMP production in response to the adenylate cyclase activator forskolin and α₂ AR agonist UK14,304, compared to the positive control (wild-type [WT]) and negative control (pcDNA). (B) Dose–response curves of functioning Gα_o mutants show changes in cAMP production in response to the α₂ AR agonist UK14,304. All dose–response curves are shown in comparison with WT and G184S. (C) G42R displays a biphasic dose–response curve with cAMP inhibition at low concentrations (gain of function [GOF]), followed by enhancement of cAMP levels at higher concentrations of UK14,304. (D) Quantification of EC₅₀ of functioning Gα_o mutants. G42R, G203R, and E246K exhibit significantly increased potency for α_2A AR–mediated cAMP inhibition similar to the known GOF mutation G184S. *p < 0.05, **p < 0.01, ***p < 0.001 using paired t test between WT and each mutant separately (figure e-4). (E) Percentage of maximum inhibition (n = 5) was normalized to pcDNA (0%; resulting in activation of cAMP) and WT (100%). ****p < 0.0001 using 1-way analysis of variance with Bonferroni post hoc test for pairwise comparison. Note that the maximum inhibition of G42R was calculated at UK14,304 of 15.8 nM. NF = normal function; PTX = pertussis toxin.

Table 1.

Functional data for loss-of-function and partial loss-of-function mutants

The dominant nature of the clinical picture in the GNAO1 encephalopathies raised the question of whether the LOF mutations are actually dominant negative mutations that interfere with the function of the remaining normal Gα_o protein expressed in heterozygous individuals. However, in coexpression studies of WT and mutant Gα_o at plasmid ratios of 1:1 or 1:2, there was no evidence of a dominant negative action (figure e-3). This suggests that the effect of LOF mutations is through a haploinsufficiency mechanism rather than a dominant negative one.

Six GNAO1 mutations result in GOF or normal function.

Unexpectedly, a significant number of pathologic GNAO1 mutants showed essentially normal or even GOF behavior (figure 2, B–E, and table 2). As a benchmark for GOF behavior, we used our previously described RGS-insensitive G184S mutant,^28,29,32,33 which shows a mild seizure phenotype in mouse models.³² It produced a small increase in the maximum inhibition of forskolin-stimulated cAMP levels in response to UK14,304 (table 2). More importantly, it had a significantly more potent response to the α₂AR agonist UK14,304 (figures 2D and e-4A). This represents a 2- to 3-fold increase in signal strength at low agonist concentrations. Three of the human pathologic GNAO1 mutants also showed GOF behavior by this criterion. The G203R, E246K, and G42R mutants produced robust inhibition of cAMP with significantly lower EC₅₀ values for the α₂AR agonist UK14,304 (figures 2D and e-4, and table 2). G203R and E246K showed normal inhibition with modest decreases in EC₅₀ (table 2 and figures 2, D and E, and e-4). This is similar to the effect on EC₅₀ seen for the bona fide GOF mutant G184S. The G42R mutant showed the lowest EC₅₀ of any of the mutants (figure e-4), at least 50-fold lower than the WT protein (figure 2, C and D). However, the inhibition mediated by G42R is followed by activation of cAMP with increasing concentrations of UK14,304 (figure 2C). The calculated normalized % inhibition for the G42R mutant is essentially identical to that of WT Gα_o, which combined with its very high potency for agonist-mediated inhibition suggests GOF behavior.

Table 2.

Functional data for normal and gain-of-function mutants

Three other patient-derived mutations (R209G, R209H, and R209C) showed almost completely normal function (NF) and nearly normal expression levels and are designated as NF mutants (figures 2, B, D, and E, and e-4, and table 2). The EC₅₀ values for R209G and R209H mutant were not significantly different from WT, while the value for R209C was modestly but significantly higher (table 2 and figures 2D and e-4).

Clinical correlation with biochemical behavior of mutant GNAO1 alleles.

To address genotype–phenotype correlations for GNAO1 encephalopathy, we reviewed the case reports of all 25 individuals who had GNAO1 mutations that had been reported by April 2016. They have a range of clinical patterns, which extend from early severe epileptic encephalopathy with prominent tonic seizure activity to individuals with a dominant choreoathetotic movement disorder with virtually no evidence of seizures. There are also individuals,^18,21 including one of the original 4 cases⁷ (patient 13—G203R; table 3), who had multiple seizures but also showed prominent choreoathetosis.

Table 3.

Correlation between cyclic adenosine monophosphate (cAMP) inhibition and clinical diagnosis

In 2016, a clinical report⁹ described a series of 6 patients with GNAO1 mutations and a pronounced movement disorder, virtually without seizures. They had global developmental delay and hypotonia from infancy and all developed chorea by ages 4–11 years. In the majority of cases it was intractable, leading to death in 2 cases. Four patients carried the E246K allele, which we have found to be a GOF mutation. The other 2 mutations found in this group (R209G and R209C) exhibited essentially normal function in our cAMP inhibition measurements. There are several other reports^8,9,18–27 of GNAO1 mutations in individuals with a predominant movement disorder with or without seizures. This distinction of clinical patterns based on certain mutant alleles in GNAO1 encephalopathy patients was noted very recently¹⁰ but without information about biochemical mechanisms. Table 3 summarizes the Gα_o biochemical function from the present report and its relation to seizure disorder or movement disorder in literature reports for these mutations. GOF and NF mutations are nearly always found when movement disorder is the predominant feature of the clinical pattern. Mutations that have pure LOF or PLOF biochemical phenotypes are seen in individuals with epileptic encephalopathy without pronounced choreoathetosis. A number of patients exhibit both seizures and movement disorder. We have indicated in table 3 with major symptoms or minor symptoms which of these features is predominant or less so. Further studies will be needed based on new cases or additional mutations, but there does appear to be a clear pattern emerging about a genotype–phenotype correlation that is driven by a GOF/LOF difference in mutant GNAO1 alleles.

Location of mutations linked to GNAO1 encephalopathies in the Gα_o protein.

To investigate the structural basis for the effects of mutations in Gα_o, the mutations were modeled onto the published crystal structure of Gα_o in complex with RGS16 (PDB: 3C7K; figure e-5). The locations of mutations within the Gα_o structure segregated according to their function. The GOF mutations are all near G184S and close to the ribose and phosphate moieties of the bound GDP. The LOF mutants are more broadly scattered throughout the GTPase domain and may destabilize protein folding or stability consistent with their markedly reduced expression levels. The PLOF mutations are clustered in the GTPase domain but away from the bound GDP. This striking structure–function correlation may facilitate prediction of the function of new mutations but ultimately a rigorous biochemical analysis will provide definitive understanding of function.

DISCUSSION

The concept of a GNAO1 encephalopathy has developed based on the identification of at least 15 different mutations in the GNAO1 gene^{7–9,18–27} associated with various combinations of epilepsy, developmental delay, hypotonia, and choreoathetotic movement disorders. Our study demonstrates GOF as well as LOF mutations in GNAO1 and describes a clear correlation between biochemical and clinical characteristics. The existence of these unexpected GOF mutations has important therapeutic implications. Specifically, one might expect that different approaches to therapy would be needed for different mutations (i.e., agonists for LOF and antagonists for GOF mutants).

We chose inhibition of cAMP production as the functional readout to assess the Gα_o mutants because of the robust measurements permitting complete agonist concentration response studies. This was critical to our findings since the GOF mutants were detected primarily through their ability to increase signals at low agonist concentrations (figure 2 and table 2). Our previously studied GOF mutant (G184S), which is insensitive to the inhibitory influence of RGS proteins, shows such a left shift of agonist concentration response curves in vitro^28,34 and in vivo^29,35 and also has a mild seizure phenotype in a mouse model.³² One might argue that cAMP is not the best choice of functional measures for epilepsy since N-type Ca⁺⁺ channels or the synaptic release mechanism proteins are critical for regulation of neurotransmitter release. However, the apparent correlation of clinical patterns with the biochemical behavior in our cAMP assay suggests that function assessed in this way is relevant to functionality in humans. The clear pathologic effect of the R209 mutations (with at least 3 individuals carrying distinct alleles), however, does raise the question of why a protein with normal expression and function would cause pathology. It is possible that the R209 mutations have a selective loss of one of the other functional outputs while retaining a normal ability to inhibit AC. Alternatively, there may be selective alterations in expression or localization in neurons that are not accurately reflected in our HEK-293T cell studies of cAMP regulation. A full understanding of the causal mechanisms in GNAO1 encephalopathies requires additional studies of these mutant Gα_o proteins in neurons and with different functional readouts.

The locations of mutations in the protein structure may partially explain their functional influences. All functioning mutants (NF and GOF) are located around the RGS binding domain, while most of the LOF or PLOF mutants are near the GDP binding region. Two exceptions are D174G and T191_F197del. D174 forms a salt bridge with R162 and mutations in this position may disrupt this interaction. T191_F197del truncates 2 β sheets as well as their linking region, which would be expected to decrease protein stability of Gα_o.

A dominant genetic effect from GOF mutants is not unusual but the fact that the LOF mutations result in a severe autosomal dominant disorder is a bit surprising. We have ruled out a biochemical dominant negative mechanism of these mutations, at least for cAMP regulation, suggesting a haploinsufficiency mechanism. In mice, homozygous Gα_o^−/− knockouts exhibit seizures as well as hyperactive turning behavior.³⁶ We did not, however, observe spontaneous seizures or an increased sensitivity to pentylenetetrazol kindling in heterozygous Gα_o^+/− knockouts.³² This suggests that humans are more susceptible to haploinsufficiency of Gα_o than are mice. In contrast, we observed enhanced kindling sensitivity and reduced survival in our Gnao1^+/G184S knock-in mouse model possibly due to seizures.³² Furthermore, these mice display early neonatal lethality³² of unclear mechanism, which may be similar to the hypotonia seen in human patients carrying GOF mutations. We do not know whether the abnormalities in these mice are due to brain developmental abnormalities or acute signaling effects. Further studies are needed to better understand this and to determine whether our Gα_o^+/G184S mutant mouse might represent a useful preclinical model for individuals with GNAO1 GOF mutations.

To date, all characterized GNAO1 mutants have been reported as LOF mutations. The G203R mutant in the original article⁷ was reported as a LOF mutant for regulation of N-type Ca⁺⁺ channels. Similarly, a G42R mutation in Gα_i1 was reported as a LOF mutant based on biochemical studies.³⁷ The unique approach that we have taken with detailed cAMP dose–response studies in a mammalian cell model permitted our recognition of the GOF mechanisms (e.g., the Gα_o G42R mutation). The patient with the G203R mutation, which we found to have GOF for cAMP inhibition, had a very different clinical pattern than the other 3 patients in the original study.⁷ She had a much later onset of disease (7 months) as well as developmental delay and severe chorea with only localized seizures.⁷ A similar clinical pattern was observed in 2 more recently described patients with this same mutation.^18,21 All patients carrying the GOF mutations identified here appear distinct from the strict EIEE pattern (table 3). In comparison, patients with LOF or PLOF mutations were diagnosed with either Ohtahara syndrome (Y231C, I279N, D174G, T191_F197del) or EIEE (A227V, L199P, N270H, F275S). Thus GOF and LOF mutations in Gα_o appear to result in different disease mechanisms likely requiring different therapeutic approaches.

It has remained challenging to convert knowledge about genetic epilepsy mutations into therapies. The GNAO1 encephalopathies may be different because of the eminently druggable nature of the receptors that drive Gα_o signaling pathways. A critical question then becomes which receptors might be involved. Interestingly, activation of many G_i/o-coupled receptors is associated with suppression of seizures. Adenosine A1 receptors may play a role in the efficacy of the ketogenic diet³⁸ and agonists at group II metabotropic glutamate receptors are anticonvulsant in various models.³⁹ The opposite situation is also seen; GABA_BR agonists exacerbate absence seizures while GABA_BR antagonists suppress them.⁴⁰ Identifying which receptors or downstream signaling effectors of Gα_o contribute to mechanisms of encephalopathy from LOF or GOF GNAO1 mutations could therefore reveal potential targets for novel anticonvulsant drug development.

We have identified distinct biochemical mechanisms of pathogenic human GNAO1 mutations that may improve the understanding of the heterogeneous clinical spectrum of GNAO1-associated epilepsy and movement disorders. Furthermore, these results also carry significant implications for personalized therapeutics in GNAO1 encephalopathies.

GLOSSARY

α_2A AR

α_2A adrenergic receptor

AC

adenylate cyclase

cAMP

cyclic adenosine monophosphate

EIEE

early infantile epileptiform encephalopathy

GOF

gain of function

LOF

loss of function

NF

normal function

PLOF

partial loss of function

PTX

pertussis toxin

WT

wild-type

AUTHOR CONTRIBUTIONS

Huijie Feng: research design, acquisition, analysis, and interpretation of data, manuscript writing. Benita Sjögren: research design, acquisition, analysis, and interpretation of data, critical revision of manuscript for intellectual content. Behirda Karaj: acquisition of data. Vincent Shaw: interpretation of data. Aysegul Gezer: acquisition of data. Richard R. Neubig: research design, interpretation of data, critical revision of manuscript for intellectual content.

STUDY FUNDING

Study funded by the NIH (R01 GM039561 and T32 GM092715).

DISCLOSURE

The authors report no disclosures relevant to the manuscript. Go to Neurology.org for full disclosures.