The α3 subunit of NKA is encoded by the ATP1A3 gene. To date, four mouse models used to study the in vivo consequences of mutations in the ATP1A3 gene have been described. The creation of model animals—mice in which the α3 subunit gene promoter (Atp1a3) is used to control the expression of the fluorescent protein ZsGreen1 (a3NKA-ZsGreen1 mouse model) [23]—made it possible to determine the localization of the α3 subunit in brain tissue. It was shown that the signal intensity was highest in the neuronal bodies located in the stem structures, including the substantia nigra, some nuclei of the thalamus and cerebellum. No fluorescence was detected in astrocytes and brain white matter.

Mutations in the ATP1A3 gene have an autosomal dominant inheritance pattern. Homozygous mutants die shortly after birth. Therefore, viable and fertile heterozygotes are used to study all four in vivo models. These models display symptoms and endophenotypes similar to those seen in the manic and depressive phases of bipolar disorder, rapid-onset dystonia parkinsonism, epilepsy, alternating hemiplegia of childhood, and CAPOS syndrome to varying degrees (Table 1) [3].

Heterozygous Myshkin mutants (NKA13AMyk/+; Myk/+) (1.1 in Table 1) carry a missense mutation with an amino acid substitution at position 810 (I810 N). Such NKA α3 subunits are expressed normally but are not functionally active. Myshkin mutants were originally developed as a preclinical model of epilepsy because heterozygotes exhibited spontaneous seizures [24]. By crossing with seizure-resistant C57BL/6NCr mice, mutants that did not exhibit seizures were obtained [32]. So far, Myk/+ mutants have been shown to be valid models of mania [33]. In behavioral tests, Myk/+ mutants demonstrated hyperactivity, circadian rhythm and sleep disturbances [34], risk-taking tendencies, and increased sensitivity to D-amphetamine [25,35]—these symptoms are seen in patients in the manic stage of bipolar disorder. Additionally, administration of lithium and valproic acid, effective in mania therapy, has been shown to normalize behavior in heterozygous mice. However, it is not known at this time whether an endophenotype of depression is possible in this model in response to stressors. Myk/+ mice were also shown to exhibit a number of disturbances in circadian behavioral rhythms related to the processing of sensory visual information but without disturbances in the function of clock genes [36]. The authors suggested a link between the identified circadian rhythm abnormalities in this mouse model and the sleep disorders observed in parkinsonism. Some reviews on rush-induced dystonia-parkinsonism suggested the use of Myshkin heterozygotes as models of this disease [37]. The 4-week-old Myk/+ displays a different gait than the wild type, unstable with a shorter stride and accompanied by tremor. Tremor and gait problems are symptoms characteristic of parkinsonism. Changes in glucose metabolism and functional brain connectivity have also been shown in mice of this line. However, Myk/+ heterozygotes are not adequate models of RDP and parkinsonism; their endophenotype is more similar to that of alternating hemiplegia of childhood [26].

Heterozygous mutants of Mashlool (α+/D801N; Mashl+/−) (1.2 In Table 1) also carry a missense mutation with an amino acid substitution at position 810. A similar amino acid substitution at the same position is found in AHC patients [38]. Hyperactivity, reduced learning ability, memory problems, tremor, and shorter stride length have been shown for this line of mice compared to wild-type mice. Dystonia, hemiplegia, and hyperexcitability were found in Mashl+/−. In vivo electrophysiology data show that heterozygotes require fewer electrical stimulations for full excitation than wild-type animals; in addition, registration of electrical activity of the amygdala and hippocampus shows that the duration of full excitation of these structures after stimulation is significantly longer in heterozygotes than in wild-type mice. Mashl+/− mutants show spontaneous seizures and have an increased mortality [27]. Mashlool mutant data show that this lineage can serve as an AHC model with some reservations, but it is difficult to judge whether it can be an adequate model for studying bipolar disorder.

Heterozygous mutants with a point mutation in the fourth intron (NKA1A3tm1Ling, NKA1A3+/−, α+/KOI4) (1.3 in Table 1) show an approximately 60% reduction in α3-subunit expression in the hippocampus [28] because of aberrant splicing. At the same time, total NKA activity is reduced by 15% compared to the wild type. Behavioral features of intact (unstressed) heterozygotes are hyperactivity, decreased anxiety, and sensitivity to methamphetamine. No behavioral manifestations of neurological disorders were found in intact heterozygotes [29]. High-performance liquid chromatography showed no change in the levels of serotonin, dopamine, and their metabolites in the striatum in heterozygotes compared to wild-type animals. However, heterozygotes showed increased locomotor activity when presented with methamphetamine, which may be related to disturbances in the dopaminergic system [28]. α+/KOI4 mice exposed to chronic variable stress (CVS) exhibit behaviors similar to those observed in the depressive phase of bipolar disorder: anhedonia, despair-like behavior, weight changes, increased anxiety, and impaired memory and socialization. At the same time, NKA1A3 activity was reduced by 33% compared to the stressed wild type, consistent with the endophenotype of depression [26]. Thus, CVS-treated α+/KOI4 mutants can serve as a model for the depressive phase of bipolar disorder. In males with this mutation, however, no overt symptoms of parkinsonism or dystonia were found before or after stressors. However, for females, chronic stress was shown to induce coordination problems. In addition, rearing in stressed heterozygotes of both sexes was shown to have a negative correlation with levels of dopamine and its metabolites, which was not observed in wild-type mice [29].

Heterozygous Atp1a3tm2Kwk/+ mutants (1.4 in Table 1) have directional deletion of exons 2–6. Hyperactivity in both cell and open field tests was shown for them, but their anxiety level is not significantly different from that of wild-type animals. Heterozygotes have a higher level of coordination and motor balance compared to the wild type. Stressors do not cause dystonia-like symptoms, but microinjections of kainate into the cerebellar vermis induced a similar state. Electrophysiological studies on slides showed a connection of the mutation to the GABAergic system but not to the dopaminergic system [30]. Heterozygotes at 4 weeks of age show a shorter stride length compared to the wild type. Older heterozygotes (6–12 weeks old) do not show gait abnormality in the absence of stressors. However, when exposed to stressors, they begin to take shorter steps when moving, compared to controls. This is very similar to the manifestation of RDP, the symptoms of which in humans can be triggered by stress. It has been suggested that Atp1a3tm2Kwk/+ mutants may be a good model for RDP, although researchers have not reported dystonia or other symptoms of parkinsonism (postural instability, bradykinesia) [31].

For all four genetic models, increased impulsivity, a propensity for risk-taking behavior, and decreased habituation have been shown to varying degrees. All of these behavioral traits are symptoms of mania. The most striking symptoms of a mania-like state are noted in Myshkin mutants. However, there is currently insufficient information about dopamine levels in this line of mice. The depressive phase of bipolar disorder is best reproduced in CVS-exposed NKA1A3tm1Ling mutants. A correlation was found between the activity of stressed mice of this lineage and dopamine levels, but the relationship between dopamine levels and the mania-like state of unstressed heterozygotes carrying this mutation is not well understood.

Gait impairment is one of the symptoms of parkinsonism, including RDP. Gait abnormalities in mice were shown for three of the four models. The Atp1a3tm2Kwk/+ model is the closest to RDP, but it does not demonstrate the full range of classic parkinsonism symptoms. Thus, no genetic model associated with a mutation in the ATP1A3 gene can be called sufficiently reliable to study parkinsonism, at least for the time being. Nevertheless, the manifestation of both manic behavior and motor disorders simultaneously in the models may indicate that mutations in the α3-subunit of NKA can phenotypically manifest these two pathologies. Further research is needed to understand the mechanisms of the relationship between these pathologies.

Mutations that disrupt the α2-subunit of NKA, which is expressed in the brain in glial cells, can also lead to the development of various neurological and neuropsychiatric disorders. Variants in the ATP1A2 gene, which encodes the α2-subunit of NKA, are associated with familial hemiplegic migraine. For example, patients with the G301R mutation are affected by a complex syndrome characterized by migraine comorbidity with epilepsy, motor symptoms, and depression or obsessive–compulsive disorder [39,40]. This mutation was successfully replicated in mice, which displayed impaired glutamate uptake and altered inflammatory cytokine signaling [39,40].