Characterization of Cognitive Deficits in Mice With an Alternating Hemiplegia-Linked Mutation

Cognitive impairment is a prominent feature in a range of different movement disorders. Children with Alternating Hemiplegia of Childhood are prone to developmental delay, with deficits in cognitive functioning becoming progressively more evident as they grow older. Heterozygous mutations of the ATP1A3 gene, encoding the Na+,K+-ATPase α3 subunit, have been identified as the primary cause of Alternating Hemiplegia. Heterozygous Myshkin mice have an amino acid change (I810N) in Na+,K+-ATPase α3 that is also found in Alternating Hemiplegia. To investigate whether Myshkin mice exhibit learning and memory deficits resembling the cognitive impairments of patients with Alternating Hemiplegia, we subjected them to a range of behavioral tests that interrogate various cognitive domains. Myshkin mice showed impairments in spatial memory, spatial habituation, locomotor habituation, object recognition, social recognition, and trace fear conditioning, as well as in the visible platform version of the Morris water maze. Increasing the duration of training ameliorated the deficit in social recognition but not in spatial habituation. The deficits of Myshkin mice in all of the learning and memory tests used are consistent with the cognitive impairment of the vast majority of AHC patients. These mice could thus help advance our understanding of the underlying neural mechanisms influencing cognitive impairment in patients with ATP1A3-related disorders.

an average full-scale IQ of 62.5 Ϯ 14.0 (mild intellectual disability), with less impairment of cognitive and adaptive skills in younger patients. Additional deficits in speech, attention, psychomotor abilities, and psychosocial functioning have also been noted (Neville & Ninan, 2007;Shafer, Mayfield, & McDonald, 2005). Flunarizine, a calcium channel blocker, is reported to alleviate the frequency and severity of hemiplegia in some patients, but there is no evidence that it has any effect on the persistent cognitive deficits that add to the burden of disease (Silver & Andermann, 1993).
Heterozygous mutations of the ATP1A3 gene, encoding the Na ϩ ,K ϩ -ATPase ␣3 subunit, have been identified as the primary cause of AHC (Heinzen et al., 2014). Na ϩ ,K ϩ -ATPases are membrane-bound transporters that harness the energy of ATP hydrolysis to pump three Na ϩ ions out of the cell in exchange for two K ϩ ions moving inward. The main role of the ␣-subunit is to bind and transport Na ϩ and K ϩ (Benarroch, 2011). The Na ϩ ,K ϩ -ATPase ␣3 subunit is exclusively present in neurons, where it is thought to be responsible for restoration of basal intracellular Na ϩ concentrations after sustained activity (Azarias et al., 2013). This Na ϩ ,K ϩ -ATPase consumes about 50% of the energy of neurons and is vulnerable to metabolic stress (Bøttger et al., 2011). Na ϩ ,K ϩ -ATPase ␣3 immunofluorescence is found preferentially localized to ␥-aminobutyric acid (GABA)ergic neurons in human and rodent brains (Bøttger et al., 2011;Paciorkowski et al., 2015). To date, 59 different de novo missense mutations in ATP1A3 have been identified in AHC patients, including three affecting the isoleucine at position 810: I810F, I810N, and I810S (Heinzen et al., 2014;Panagiotakaki et al., 2015;Sasaki et al., 2014;Ulate-Campos et al., 2014;Viollet et al., 2015;Yang et al., 2014).
During seizures, there is often an energy crisis in neurons, which can lead to Na ϩ ,K ϩ -ATPase pump failure (Araujo et al., 2014). Such failure could influence the pathophysiology of seizure activity of any etiology, as well as contribute to its long-term consequences and possibly even to sudden unexpected death in epilepsy (SUDEP; Richter et al., 2012). Indeed, two further missense mu-tations in ATP1A3 have been found in a child with catastrophic early life epilepsy (G358V), and in another child with epilepsy, episodic prolonged apnea, postnatal microcephaly, and severe developmental disability (I363N; Paciorkowski et al., 2015).
Heterozygous Myshkin (Atp1a3 Myk/ϩ ; Myk/ϩ) mutant mice have an I810N amino acid substitution in transmembrane ␣-helix TM6 that is identical to that present in AHC (Clapcote et al., 2009;Panagiotakaki et al., 2015;Yang et al., 2014). I810N results in a normally expressed, but inactive, ␣3 protein and a subsequent 42% reduction of total Na ϩ ,K ϩ -ATPase activity in the mouse brain (Clapcote et al., 2009). Molecular modeling of I810N has shown that it brings about severe structural effects on Na ϩ ,K ϩ -ATPase ␣3, including the capacity for efficient K ϩ movement along the K ϩ access pathway (Kirshenbaum et al., 2013). We previously reported that Myk/ϩ mice show deficits in contextual fear conditioning, delay cued fear conditioning, and conditioned taste aversion tests of learning and memory (Kirshenbaum et al., 2013), but their normal head tracking in an optokinetic drum suggests that their vision is not impaired (Kirshenbaum, Clapcote et al., 2011). These mice also show deficits in both frontal cortex functioning (hypofrontality) and thalamocortical functional connectivity, as revealed by glucose metabolism imaging (Kirshenbaum et al., 2013). However, hippocampal slices from Myk/ϩ mice showed wild-type-like levels of burst stimulation-evoked long-term potentiation (LTP; Clapcote et al., 2009), a persistent increase in synaptic strength that is widely considered one of the major cellular mechanisms that underlies learning and memory (Cooke & Bliss, 2006). The purpose of the present study was to extend our phenotypic characterization of cognitive functioning in Myk/ϩ mice, by subjecting them to a range of behavioral tests that interrogate various cognitive domains.

Behavioral Procedures
Myk/ϩ and ϩ/ϩ littermates were tested at 8 -14 weeks of age in four cohorts. Males and females were included in balanced numbers, apart from in the social recognition and open field tests. Subjects were handled daily for 5 min/day for 7 days before behavioral testing. Experimenters were blinded to genotype during behavioral testing. Behavioral testing apparatus were cleaned with 70% ethanol between subjects. The order of tests, with a rest period of 3-5 days between each test, was as follows: Cohort 1 (n ϭ 10/genotype): Y maze ¡ object recognition ¡ T maze (10-min); Cohort 2 (n ϭ 10/genotype): T maze (1 hr) ¡ Morris water maze ¡ trace fear conditioning; Cohort 3 (n ϭ 9/genotype; males only): social recognition (2-min); Cohort 4 (n ϭ 6/genotype; males only): social recognition (10-min) ¡ open field. All procedures were approved by the University of Leeds Ethical Review Committee and were conducted in accordance with the requirements of the Animals (Scientific Procedures) Act, 1986. Y-maze spontaneous alternation. The Y-maze spontaneous alternation test is based on the natural tendency of rodents to explore a novel environment. The Y-maze apparatus made of matt white acrylic sheet had three identical arms (35 ϫ 5 ϫ 15 cm) placed at 120°from each other. Each mouse received one trial, in the course of which the subject was placed into one of the three arms and allowed free exploration of the maze for 5 min. The number of arm entries and the number of alternations were recorded to calculate the percentage of alternation. Over the course of multiple arm entries, mice typically exhibit a tendency to explore the least recently visited arm, and thus, tend to alternate visits between the three arms. For efficient alternation, mice need to use working memory, and thus, they should maintain an ongoing record of most recently visited arms, and continuously update such a record. A mouse with an impaired working memory cannot remember which arm it has just visited, and thus shows decreased spontaneous alternation (King & Arendash, 2002). Alternation was defined as the number of arm choices that differed from the previous two choices. The alternation percentage was calculated as: "number of alternations" divided by "total number of arm entries" minus 2. Thus, if a subject made the following sequence of arm choices (3, 2, 1, 2, 3, 2, 1, 3), the total number of alternation opportunities would be six (total entries minus 2) and the percentage alternation would be 67% (4 of 6).
Object recognition. The object recognition test was conducted as previously described (Ricceri, Colozza, & Calamandrei, 2000). The test exploits the natural exploratory activity of rodents toward spatial novelty to assess the detection of spatial relocation of a known object and is critically dependent on the hippocampus (Stupien, Florian, & Roullet, 2003). Mice were placed into a testing arena consisting of an open field made of clear acrylic sheet (41.25 ϫ 41.25 ϫ 31.25 cm) in dim light. After 10-min of habituation, mice were removed from the arena and returned to the home cage. Four identical objects (1-4; glass Petri dishes) were placed into the arena, each in a separate corner 4-cm away from the wall. After 2 min, mice were returned to the arena, and left to explore for 15 min. Mice were then removed from the arena and returned to the home cage. Two of the objects (3 and 4) in the arena were then displaced from their original locations. After 2 min, mice were placed back into the arena, and left to explore for 5 min. Mice were then removed from the arena and returned to the home cage. Three of the original objects (2-4) were then removed from the arena and replaced with a single novel object (white porcelain cup). After 2 min, mice were placed back into the arena, and left to explore for 5 min.
T-maze spatial habituation. Exposure to a spatial location leads to habituation of exploration such that, in a novelty preference test, rodents subsequently prefer exploring a novel location to the familiar location. In the T-maze test of spatial habituation, each mouse was allowed to explore two arms of a T-maze made of acrylic sheet (0.6 mm thick) painted a light beige color. The main runway (65 ϫ 14 ϫ 30 cm) was connected to two side (goal) arms (30 ϫ 14 ϫ 30 cm). After 10 min (Cohort 1) or 1 hr (Cohort 2) of exposure training, a guillotine door was manually removed to reveal the third (novel) arm. The ambulation of the subject in each arm was then observed for 5 min. The exploration ratio was calculated as: "time in novel arm" divided by "time in familiar arms." Visible platform water maze. The visible platform version of the Morris water maze is a simple nonspatial learning task that is believed to be independent of hippocampal function (D'Hooge & De Deyn, 2001). The Morris water maze consisted of a cylindrical tub of ivory-colored acrylic sheet (117 cm diameter; 30 cm depth) that was filled with water (26 Ϯ 1°C temperature) to 11 cm below the rim, as described previously (Clapcote et al., 2005). The water was rendered opaque by the addition of white, nontoxic paint. A circular platform (10 cm diameter) made of transparent acrylic sheet was submerged 1 cm below the water surface at the center of the pool. The platform location was indicated by a high-contrast striped marker (1 cm diameter rod with a ping-pong ball affixed to the top) rising 13 cm above the water surface. Each subject was given four training trials. At the start of each trial, the mouse was placed by the tail into the water, immediately facing the perimeter, at one of the cardinal compass points (north, south, east, or west), and then was allowed a maximal time of 90 s to locate the platform. Finding the platform was defined as staying on it for at least 2 s.
Trace fear conditioning. Fear conditioning takes advantage of the natural tendency of rodents to freeze in response to fearful stimuli. Decreased freezing behavior is used as an indicator of a rodent's recognition of a potentially aversive stimulus, and thus, an indicator of memory function. In trace conditioning, a conditioned stimulus (CS), such as an auditory tone, is presented. After a stimulus-free period called the trace interval, an unconditioned stimulus (US), such as a footshock, is presented. During the interval, the subject must sustain attention to learn a CS-US association. The fear conditioning apparatus used has been described previously (Kirshenbaum et al., 2013). For the training phase, mice were placed in the chamber for 2 min, after which the CS (auditory tone, 3,600 Hz, 95 dB) was presented for 30 s. There was then a 3-s interval before the US (continuous scrambled footshock, 1.0 mA) was presented for 2 s. Mice then remained in the chamber for an additional 30 s before being returned to their home cage. Twenty-four hours after training, the chamber was altered to evaluate freezing to the auditory tone. To disguise the chamber, the grid was covered with a smooth white acrylic sheet, the walls were altered by inserting a white acrylic sheet triangle, a 1% acetic acid odor was applied and the lights in the room were switched off. Subjects were placed in a chamber for 3 min to explore the new environment, followed by a 3 min presentation of the CS while freezing was recorded.
Social recognition. Social recognition is the ability to distinguish familiar from novel conspecifics. Social recognition by a subject mouse is defined by decreased investigation of a previously encountered mouse. The social recognition procedure was modified from a previously described protocol (Kogan, Franklandand, & Silva, 2000). To avoid potentially confounding effects of sexual behavior, only male mice were used for this experiment. On Day 1, training sessions consisted of a 15-min habituation to a new cage, followed by a 2-min (Cohort 3) or 10-min (Cohort 4) presentation of a novel juvenile male C57BL/6NCrl mouse (Charles River). Social investigation of the juvenile (mainly sniff-ing and licking of the anogenital region of the juvenile) by the adult mouse was observed continuously by an observer who scored the duration of investigation behavior with Observer software (Noldus). The social interaction time was measured, after which the mice were returned to their home cages. On Day 2, mice were tested 24 hr after the initial interaction using the same juvenile male. Any aggressive encounter between mice was immediate cause for terminating the experiment and excluding data from an adult from the analysis. The recognition ratio was calculated as: "social interaction time on Day 2" divided by "social interaction time on Day 1." Open field test of locomotor habituation. When placed in a novel environment, mice tend to explore for a period of time, and then reduce the level of exploration. This reduction in locomotor or exploratory behavior is known as habituation. Habituation to an open field arena made of clear acrylic sheet (41.25 ϫ 41.25 ϫ 31.25 cm) was measured as a decrement in exploratory locomotor activity over time. Mice were placed in the center of the floor, and allowed to freely explore the open field for 1 hr on each of four consecutive days. Locomotor activity was indexed by the total distance traveled (in meters) in each 1-hr session using the VersaMax Animal Activity Monitoring System (AccuScan Instruments).

Statistical Analysis
Data were subjected to analysis of variance (ANOVA) to examine performance by genotype and sex as between-subjects factors. Tukey-Kramer post hoc multiple comparison tests were performed on any significant main effect or interaction. For each genotype in the object recognition test, one-way ANOVA was used to examine the effect of object location on the percentage of exploration time spent with each identical object. Student's t tests were used in the Y-maze spontaneous alternation and object recognition tests to compare the performance of each genotype versus chance levels, in the trace fear conditioning test to compare freezing levels between various phases, in the social recognition test to compare the recognition ratio for each genotype versus a "null recognition" ratio of 1.0, and in the open field locomotor habituation test to compare between days. Significance level was set to 0.05. All values reported in the figures are expressed as mean Ϯ SEM.

T-Maze Spatial Habituation
In the T-maze test of spatial habituation, Cohort 1 was allowed to explore two arms for 10 min, after which the third (novel) arm was revealed and the ambulation of mice in the novel versus familiar arms was observed for 5 min. A main effect of genotype was observed for 10-min exploration ratio, F(1, 16) ϭ 6.10, p Ͻ .05, with no sex effect, F(1, 16) ϭ 1.13, ns or sex by genotype interaction, F(1, 16) ϭ 0.11, ns observed. With 10 min of exposure, Myk/ϩ mice spent proportionately less time in the novel arm than ϩ/ϩ mice (see Figure 3). Cohort 2 was allowed to explore two arms for 1 hr before the third (novel) arm was revealed. A main effect of genotype was observed for 1-hr exploration ratio, F(1, 16) ϭ 18.19, p Ͻ .001, with no sex effect, F(1, 16) ϭ 0.11, ns or sex by genotype interaction, F(1, 16) ϭ 0.17, ns observed. Even with 1 hr of exposure, Myk/ϩ mice still spent proportionately less time in the novel arm than ϩ/ϩ mice (see Figure 3), indicating a deficit in spatial habituation. Only ϩ/ϩ mice spent proportionately more time in the novel arm after 1 hr versus 10 min of exposure to the familiar arms (ϩ/ϩ: t(9) ϭ Ϫ3.75, p Ͻ .01; Myk/ϩ: t(9) ϭ Ϫ1.00, ns; Figure 4).

Trace Fear Conditioning
We previously observed that Myk/ϩ mice are impaired in delay fear conditioning (Kirshenbaum et al., 2013), in which the CS is immediately followed by the US. In trace conditioning, the end of the CS and the start of the US are separated by a trace interval. In this procedure, a main effect of genotype was observed for baseline (test-naïve) freezing, F(1, 16) ϭ 17.39, p Ͻ .001 and freezing during the auditory tone (CS) in the training phase, F(1, 16) ϭ 14.06, p Ͻ .01, with no sex effect (F(1, 16) Ͻ 1.52, ns), or sex by genotype interaction (F(1, 16) Ͻ 0.86, ns) observed. Both genotypes displayed baseline freezing of less than 10%, but that displayed by Myk/ϩ mice was significantly lower compared with ϩ/ϩ mice ( Figure 4C). Twenty-four hours after training, subjects were placed into an altered chamber and the CS was presented again. No effect of genotype, F(1, 16) ϭ 2.13, ns, sex, F(1, 16) ϭ 4.22, ns or sex by genotype interaction, F(1, 16) ϭ 0.02, ns was observed for freezing in the altered chamber before the auditory tone (Pre-CS). Both genotypes increased their Pre-CS freezing relative to baseline freezing (ϩ/ϩ: t(9) ϭ 3.02, p Ͻ .01; Myk/ϩ: t(9) ϭ 3.41, p Ͻ .01), suggesting that both genotypes showed some fear response to the altered chamber. During presentation of the auditory tone (CS), a main effect of genotype was observed for CS freezing, F(1, 16) ϭ 10.17, p Ͻ .01, with no sex effect (F(1, 16) Ͻ 1.52, ns) or sex by genotype interaction (F(1, 16) Ͻ 0.86, ns) observed. Both genotypes increased their CS freezing level relative to Pre-CS freezing (ϩ/ϩ: t(9) ϭ 10.29, p Ͻ .0001; Myk/ϩ: t(9) ϭ 10.84, p Ͻ .0001), although Myk/ϩ mice showed a significantly lower level of CS freezing compared with ϩ/ϩ mice ( Figure 4C).

Locomotor Habituation in Open Field
Male Myk/ϩ and ϩ/ϩ mice that had previously been tested in the social recognition test (Cohort 4) were subjected to the open field test of locomotor habituation. Main effects of genotype, F(1, 64) ϭ 156.51, p Ͻ .0001, day, F(7, 64) ϭ 2.60, p Ͻ .05 and genotype by day interaction, F(7, 64) ϭ 3.71, p Ͻ .01 were observed for locomotor activity in each 1-hr session. The locomotor activity of Myk/ϩ mice was significantly greater than that of ϩ/ϩ mice on all four test days ( Figure 5B). The locomotor habituation deficit of Myk/ϩ mice, as suggested by the presence of a significant genotype by day interaction, was confirmed by the finding that only ϩ/ϩ mice showed a significant decrement in locomotor activity between Day 1 and Day 4 (ϩ/ϩ: t(5) ϭ Ϫ17.54, p Ͻ .0001; Myk/ϩ: t(5) ϭ 2.46, ns).

Discussion
Myk/ϩ mice have an I810N amino acid substitution in Na ϩ ,K ϩ -ATPase ␣3 that is identical to that present in AHC (Clapcote et al., 2009;Panagiotakaki et al., 2015;Yang et al., 2014). To investigate whether Myk/ϩ mice exhibit learning and memory deficits resembling the cognitive impairments of patients with ATP1A3-related disorders (Barbano et al., 2012;Paciorkowski et al., 2015Paciorkowski et al., , 2010Sweney et al., 2009), we subjected them to a range of behavioral tests that interrogate various cognitive domains.
In the Y-maze spontaneous alternation test, Myk/ϩ mice showed increased exploratory locomotion, consistent with our previous observations in the hole-board test (Kirshenbaum, Clapcote et al., 2011). Although Myk/ϩ and ϩ/ϩ mice showed roughly comparable levels of alternation, only ϩ/ϩ mice exceeded the chance level, suggesting that Myk/ϩ mice were deficient in this measure of short term spatial memory. In the object recognition test, Myk/ϩ mice were significantly worse than ϩ/ϩ mice at recognizing displaced objects, and did not perform above chance level. Myk/ϩ mice were better at recognizing novel objects (exceeding chance level in this task), but were still significantly worse than ϩ/ϩ mice.
In the T-maze test of spatial habituation, Myk/ϩ mice showed a spatial habituation deficit after 10 min of training. Increasing the duration of training to 1 hr significantly increased the spatial habituation of ϩ/ϩ mice, but had no effect on Myk/ϩ mice. Similarly, Myk/ϩ mice were deficient in the open field test of locomotor habituation. Unlike ϩ/ϩ mice, they did not demonstrate locomotor habituation by reducing their exploratory activity over the 4 days of testing. Increasing the duration of training from 2 min to 10 min did, however, improve the performance of Myk/ϩ mice in the social recognition test. With 2 min of training, only ϩ/ϩ mice demonstrated social recognition by showing less social interaction on Day 2 than on Day 1. When training was increased to 10 min, both genotypes demonstrated social recognition, indicating that Myk/ϩ mice required more training than ϩ/ϩ mice.
In the Morris water maze, Myk/ϩ mice were impaired in navigating to a visible escape platform. Swim speed and duration of floating were normal, but thigmotaxis was significantly increased in Myk/ϩ mice. Increased thigmotaxis has previously been observed in mice with retinal degeneration (Clapcote et al., 2005a), but the normal head tracking of Myk/ϩ mice in an optokinetic drum suggests that their vision is not impaired (Kirshenbaum, Clapcote et al., 2011).
In the trace fear conditioning test, there was no genotypic difference in freezing in the altered chamber (Pre-CS freezing), but Myk/ϩ mice showed less freezing than ϩ/ϩ mice to the auditory tone (CS freezing), suggesting an impairment in trace fear conditioning. The mean CS freezing level of 47% exhibited by Myk/ϩ mice in the trace fear conditioning test suggests that their ability to exhibit freezing was not grossly affected by their whole body tremor (Kirshenbaum et al., 2013). We previously reported that Myk/ϩ mice show a deficit in delay fear conditioning (Kirshenbaum et al., 2013), in which the CS is immediately followed by the US during training. This deficit was maintained regardless of whether the footshock (US) was 0.75 or 1.0 mA (Kirshenbaum et al., 2013).
We previously reported that Myk/ϩ mice show hyperambulation and a chaotic walking path in response to a novel open field (Kirshenbaum, Clapcote et al., 2011). This response to a novel environment is likely to have some influence on the performance of Myk/ϩ mice in the Y-maze, T-maze, open field, objection recognition and social recognition, as these tests are dependent on locomotor activity. It likely also explains the decreased baseline freezing of Myk/ϩ mice when put into the novel environment of the fear conditioning chamber. The apparent cognitive impairment of Myk/ϩ mice is likely not a mere consequence of their locomotor hyperactivity, though. We previously reported that Myk/ϩ mice show a deficit in conditioned taste avoidance (Kirshenbaum et al., 2013), a test of implicit memory that takes place in the home cage rather than a novel environment, and is not dependent on the locomotor activity or freezing of the subject.
Factor analysis permits recognition of clusters of strongly intercorrelated variables termed factors, and dissociation of them from other (minimally intercorrelated) factors. In an attempt to dissociate cognitive function from noncognitive performance characteristics, future work will apply factor analysis to a selection of variables known to best characterize the behavioral variability of Myk/ϩ and ϩ/ϩ mice in several tests, as undertaken previously with prion protein-deficient mice (Valenti et al., 2001).
Although Myk/ϩ mice showed deficits in all of the cognitive tests in the present study, one form of learning is known to be intact in Myk/ϩ mice. We previously reported that a "genotype by day" interaction was not observed when we tested Myk/ϩ and ϩ/ϩ mice on an accelerating rotarod on five consecutive days (Kirshenbaum et al., 2013). Myk/ϩ mice demonstrated intact motor learning, the ability to acquire new motor skills through practice. Rodents with lesions that selectively destroy the hippocampus are able to acquire a motor learning task, but show profound impairment in a novel object recognition memory task (Gould et al., 2002). Because young adult Myk/ϩ mice are known to exhibit medial temporal sclerosis and aberrant membranous whorls in hippocampal pyramidal neurons in CA3 and CA1 (Clapcote et al., 2009), the difference in the outcome of the rotarod versus the other cognitive tests used is likely because of the relative importance of various brain structures in each test.
Several rodent studies have used a pharmacological approach to perturb Na ϩ ,K ϩ -ATPase function in the brain. Intracranial injection of ouabain, a specific Na ϩ ,K ϩ -ATPase inhibitor that does not cross the blood-brain barrier (Sugimoto et al., 2011), has been shown to impair learning and memory in laboratory rats and mice (Mizumori et al., 1987;Sato et al., 2004;Tucker & Gibbs, 1979;Wang et al., 2013Wang et al., , 2014Zhan et al., 2004). These studies all point to a relationship between inhibition of Na ϩ ,K ϩ -ATPase in the brain and impairment of learning and memory, but they are limited by the similar affinities of the ␣2 and ␣3 subunits for ouabain (Sweadner, 1989); thus, precluding Na ϩ ,K ϩ -ATPase ␣3 specificity.
Other mouse genetic studies provide support for the involvement of Na ϩ ,K ϩ -ATPase ␣3 in memory formation. Heterozygous Atp1a3 D801N/ϩ mice, which carry the most common mutation causing AHC (D801N), were deficient in a novel object recognition task and in the visible platform version of the Morris water maze (Hunanyan et al., 2015). Though they showed unimpaired contextual fear conditioning and motor learning on the accelerating rotarod (Hunanyan et al., 2015). Heterozygous Atp1a3 tm1Ling/ϩ mice, which have a reduction of hippocampal ␣3 protein expression of around 60% and a reduction of total brain Na ϩ ,K ϩ -ATPase activity (of ␣1, ␣2, and ␣3 combined) of around 16%, showed deficient spatial learning in the Morris water maze but normal memory in a novel object recognition task (Kirshenbaum, Saltzman et al., 2011;Moseley et al., 2007). After exposure to chronic variable stress, consisting of one or two unpredictable mild stressors per day for 6 weeks, Atp1a3 tm1Ling/ϩ mice showed a greater reduction of total brain Na ϩ ,K ϩ -ATPase activity (of around 33%) and a pronounced deficit in novel object recognition memory (Kirshenbaum, Saltzman et al., 2011).
An obvious difference between the published pharmacological and genetic approaches that perturb Na ϩ ,K ϩ -ATPase function in vivo is that the genetic mutations are present from conception. While it is therefore possible that Na ϩ ,K ϩ -ATPase ␣3 mutations could affect the nervous system throughout development, we previously reported that the ␣1 and ␣2 isoforms contribute more than ␣3 to total brain Na ϩ ,K ϩ -ATPase activity in E18 fetuses, whereas ␣3 is the functionally dominant Na ϩ ,K ϩ -ATPase isoform in the young adult brain (Clapcote et al., 2009). A difference in the functional importance of ␣3 versus the other Na ϩ ,K ϩ -ATPase isoforms in utero and postnatally may explain why homozygous Myk/Myk pups all die perinatally, despite being born alive and appearing grossly normal at birth (Clapcote et al., 2009).
One factor that may contribute to the impaired cognition of Myk/ϩ mice is altered neuronal excitability. Within the CA1 stratum radiatum of the hippocampus, following burst stimulation (a plasticity induction protocol), the magnitude of the relationship between the population action potential to the field excitatory postsynaptic potential was increased compared with ϩ/ϩ mice (Clapcote et al., 2009). Hence, after a high frequency train of stimuli, the CA3-CA1 pathway exhibits signs of hyperexcitability. It is possible that this aberrant synaptic activity is, in part, responsible for altered neural coding during behavior. Similar hyperexcitability within the dentate gyrus and CA3 caused by brain-derived neurotrophic factor overexpression in mice was associated with impaired performance in the passive avoidance test (Croll et al., 1999). However, the overall magnitude of long-term potentiation in Myk/ϩ mice is similar, albeit more variable, to that of ϩ/ϩ mice (Clapcote et al., 2009), arguing against cognitive impairment being caused merely by a lack of synaptic plasticity. Further work is thus required to elucidate the link between hippocampal hyperexcitability and the cognitive deficits exhibited by Myk/ϩ mice.
Because Myk/ϩ mice showed deficits in all of the tests used in the present study, we conclude that they do exhibit learning and memory deficits resembling the broadly defined cognitive impairments of patients with ATP1A3-related disorders. A Chinese AHC patient with the same I810N mutation as Myk/ϩ mice is reported to have developmental delay (Yang et al., 2014), while a Belgian AHC patient with I810N has moderate intellectual disability and autism (Panagiotakaki et al., 2015). Specific deficits in intellectual, academic, memory, attention, and executive functioning exhibited by a child with AHC of unknown ATP1A3 status have been reported (Shafer, Mayfield, & McDonald, 2005). Until similar comprehensive neuropsychological data on the I810N patients and other AHC patients with identified ATP1A3 mutations become available, it would be unreliable to draw specific parallels with aspects of the Myk/ϩ cognitive phenotype at this stage. Nevertheless, by better defining the cognitive profile of Myk/ϩ mice, the present study has widened the range of AHC symptoms for which the Myk/ϩ model could be used as a tool in the identification of novel therapeutic strategies.