ATP1A2: more than a fraternal twin 

ATP1A2. We have previously written about ATP1A3, one of the neuronal ATPases, which, together with ATP1A2, actively works to maintain the electrochemical gradient across cell membranes. In our previous post, we likened ATP1A3 to Sisyphus, who was punished by the gods to spend eternity exerting his strength to push a boulder up a hill, only for the boulder to roll back down, over and over again. Similarly, the ATPases utilize energy (ATP) to transport Na+ out of the cell and K+ into the cell, even while other voltage-gated channels effortlessly open to allow Na+ and K+ to move in the opposite direction, forcing the boulder back down the hill. However, from another perspective we can see that far from being Sisyphus, the ATPases are the noblest of facilitators, proudly maintaining the membrane electrochemical gradient in order to enable all other cell processes to function. If you will allow me further anthropomorphization of submicroscopic proteins – far from being inconsiderate to the task of the ATPases, our voltage-gated Na+, K+, and Ca++ channels are reliant on these stoic transporters to get their jobs done. It is because of the electrochemical gradient that cells are able to propagate action potentials down an axon, signal axon terminals to release neurotransmitters, and continue to send electrochemical signals on to subsequent cells. Both ATP1A2 and ATP1A3 code for subunits of the Na+-K+ ATPase, where ATP1A2 codes for the α2 subunit and ATP1A3 for the α3 subunit. Both ATP1A3 and ATP1A2 are expressed in neurons during embryonal brain development, and while ATP1A3 remains neuronal postnatally, ATP1A2 is relegated to the glial cells postnatally and into adulthood. In addition to their role in active ion transport, Na+-K+ ATPases are thought to play a part in regulating signaling pathways and gene transcription. Given its localization to astrocytes, which act as a so-called “glutamate sink” to prevent glutamate excitotoxicity within the synapse, ATP1A2 is thought to be critical for the process of glutamate clearance and prevention of excitotoxicity (Du et al 2020). 

ATP1A2. Like ATP1A3, this protein also maintains electro-chemical gradients across cell membranes with the use of ATP. ATP1A2 is primarily expressed in glial cells, while ATP1A3 is primarily expressed in neurons.

Spectrum of disease. There has been a wide spectrum of disease conditions associated with variants in ATP1A2, the most prominent of which is familial hemiplegic migraine (HM). Rarely for a neurodevelopmental gene, it has been implicated in both autosomal dominant and autosomal recessive conditions, where autosomal recessive conditions are characterized by a significantly more severe phenotype. There is no evidence that ATP1A2 is related to progressive or neurodegenerative conditions. 

Hemiplegic migraine. ATP1A2 was first discovered in relation to HM, which may range from mild, characterized by transient hemiparesis associated with migraine headache, to severe, with potentially fatal unihemispheric cerebral edema. On the more severe end of the spectrum, ATP1A2 variants may also be associated with epilepsy and intellectual disability. On review of the literature, there seems to be considerable genotype-phenotype correlation with phenotypic homogeneity within recurrent variants, though no clear pattern relating structure of the protein to clinical symptoms has emerged. Moya-Mendez and colleagues (2021) have proposed that variants localized to sodium binding domains may have more severe phenotypes. The R65W, R202Q, R593W, and G762S variants have been associated with pure HM (Li et al., 2021). Certain variants, including M813K, have been associated with severe HM with cerebral edema, in addition to a severe neurodevelopmental phenotype. Other variants have been reported in association with cerebral edema, in otherwise neurotypical individuals (E825K in Du et al 2020).  

Of note, CACNA1A is the other principal gene implicated in HM, with a similar phenotypic range of mild to severe, as well as a variable association with epilepsy and neurodevelopmental differences. ATP1A3, on the other hand, is implicated in alternating hemiplegia of childhood (AHC), a phenotypically distinct condition which typically emerges at a younger age, and in which it is quite rare to have associated cerebral edema. 

Epilepsy. Epilepsy is present in a subset of patients with ATP1A2-related conditions, where all individuals also appear to have HM episodes in addition to epilepsy. Both Calame and colleagues (2021) and Moya-Mendez and colleagues (2021) have described a recurrent variant (M813K) in individuals who had infantile onset hemiplegic episodes associated with unilateral cerebral edema, epilepsy, movement disorder, and ataxia. Moya-Mendez and colleagues have also described other, non-recurrent variants in ATP1A2 associated with early onset epilepsy. Per Li and colleagues (2021), certain variants, including R548C, E825K, and R938P have been associated with HM with epilepsy, while others, namely T378N, G615R, and D718N have been associated with HM, epilepsy, and intellectual disability.  

Structural brain abnormalities. Severe structural brain abnormalities associated with severe phenotypes (early onset epilepsy, intellectual disability, and early mortality) have been described in both heterozygous missense variants (Vetro et al., 2021) and in biallelic variants (Monteiro et al 2019). Given the phenotypic similarities in these presentations despite drastically different genotypes, the likely functional mechanism of the heterozygous variants is dominant negative.  

Treatment recommendations. Acetazolamide (Diamox) and flunarizine have been used as preventive medications for ATP1A2-related HM, though there is no evidence as to their efficacy for other ATP1A2-related conditions. Memantine treatment has been proposed as a daily medication for individuals with more severe phenotypes associated with ATP1A2 variants, with the suggestion of improved cognition with memantine therapy (Moya-Mendez et al 2021) – though of note, this is all self-reported data with therefore limited objective evidence. There is no clear evidence for the efficacy of memantine in treating seizures due to ATP1A2 variants. For those with severe epilepsy associated with ATP1A2, proposed anti-seizure therapies include topiramate and the ketogenic diet. Treatment for acute HM may include acetazolamide as well as common migraine medications, including ibuprofen and magnesium. With regard to treatment for acute cerebral edema secondary to ATP1A2-related HM, Du and colleagues (2020) propose the use of NMDA-receptor antagonists (such as memantine or ketamine) to reduce glutamate excitotoxicity, as well as targeted treatments aimed to protect mitochondrial function – though again, there is limited evidence showing efficacy in these therapies given a very limited sample size.  

This is what you need to know. While structurally and functionally similar to its neuronally-based sister, ATP1A3, ATP1A2 is associated with distinct phenotypic features, perhaps related to its localization in glial cells where it serves as a buffer against glutamate-induced excitotoxicity. The majority of reported variants in ATP1A2 are missense; though protein-truncating variants have been reported as well (and it should be noted that gnomAD scores the pLI as 1), many of the loss-of-function variants reported in ClinVar are present in population databases and may only cause disease in homozygous or biallelic state. This bewildering pattern of pathogenicity – as well as the broad phenotypic spectrum – lies at the root of the challenge for clinicians in interpreting results pertaining to ATP1A2 and we therefore urge continued caution during interpretation. As we concluded in the case of ATP1A2’s sister, our task is to continue to gather clinical data and variant information in order to gradually sort out genotype-phenotype relationships with the aim for more precise therapies in the future. 

Alexis Karlin

Alexis Karlin is a pediatric neurologist at the Children’s Hospital of Philadelphia.