Kv1.1 deficiency alters repetitive and social behaviors in mice and rescues autistic‐like behaviors due to Scn2a haploinsufficiency

Abstract Background Autism spectrum disorder (ASD) and epilepsy are highly comorbid, suggesting potential overlap in genetic etiology, pathophysiology, and neurodevelopmental abnormalities; however, the nature of this relationship remains unclear. This work investigated how two ion channel mutations, one associated with autism (Scn2a‐null) and one with epilepsy (Kcna1‐null), interact to modify genotype–phenotype relationships in the context of autism. Previous studies have shown that Scn2a +/– ameliorates epilepsy in Kcna1 –/– mice, improving survival, seizure characteristics, and brain–heart dynamics. Here, we tested the converse, whether Kcna1 deletion modifies ASD‐like repetitive and social behaviors in Scn2a+/– mice. Methods Mice were bred with various combinations of Kcna1 and Scn2a knockout alleles. Animals were assessed for repetitive behaviors using marble burying, grooming, and nestlet shredding tests and for social behaviors using sociability and social novelty preference tests. Results Behavioral testing revealed drastic reductions in all repetitive behaviors in epileptic Kcna1 –/– mice, but relatively normal social interactions. In contrast, mice with partial Kcna1 deletion (Kcna1 +/–) exhibited increased self‐grooming and decreased sociability suggestive of ASD‐like features similar to those observed in Scn2a +/– mice. In double‐mutant Scn2a +/–; Kcna1 +/– mice, the two mutations interacted to partially normalize ASD‐like behaviors associated with each mutation independently. Conclusions Taken together, these findings suggest that Kv1.1 subunits are important in pathways and neural networks underlying ASD and that Kcna1 may be a therapeutic target for treatment of Scn2a‐associated ASD.


| INTRODUC TI ON
Autism spectrum disorder (ASD) and epilepsy are common neurological diseases that often co-occur in patients for reasons that are not well understood. The two main hallmarks of ASD are impaired social interaction and communication and the presence of restricted repetitive behavior, whereas epilepsy is characterized by the presence of spontaneous recurrent seizures (Stafstrom & Carmant, 2015;de la Torre-Ubieta et al., 2016). About 6%-27% of people diagnosed with ASD also have epilepsy, and there is a sevenfold increased risk of epilepsy in ASD patients (Jeste & Tuchman, 2015;Thomas et al., 2017).
Similarly, about 37% of children <5 years of age with epilepsy test positive for autism, and there is a 6-to 8-fold increased risk of ASD in epilepsy patients overall (Fisher et al., 2012;Strasser et al., 2018).
One explanation for the high comorbidity rate of ASD and epilepsy is that the two disorders have a common underlying neuropathophysiology, which could be caused by a shared genetic basis since many genes associated with epilepsy are also linked to autism (Jeste & Tuchman, 2015;Srivastava & Sahin, 2017).
In addition to rare monogenic causes, both ASD and epilepsy have significant genetic etiologies involving complex interactions between multiple genetic variants which can be inherited or arise de novo (Klassen et al., 2011;de la Torre-Ubieta et al., 2016). Although these mutations are typically thought to add together in a deleterious fashion causing disease when some risk threshold is crossed, the opposite can also be true, that mutations can exert protective effects that mask disease (Devlin & Scherer, 2012;Noebels, 2015;de la Torre-Ubieta et al., 2016). Indeed, in both ASD and epilepsy, beneficial suppressor mutations have been identified that are capable of ameliorating the harmful effects of known disease-causing variants (Glasscock et al., 2007;Leblond et al., 2012). Protective genetic modifiers represent valuable opportunities for improving diagnostic risk assessment, identifying new therapeutic targets, and revealing underlying disease mechanisms. Complex interplay between mutations associated with epilepsy and autism (i.e., epistatic effects) could partially explain the high degree of comorbidity for the two conditions, as well as the absence of disease when high-risk variants are present.
One gene that has come to light as an important molecular cause of both epilepsy and autism is Scn2a, which encodes neuronal Nav1.2 voltage-gated sodium channel α-subunits that regulate action potential initiation and propagation in axons and action potential backpropagation in somatodendritic compartments (Hu et al., 2009;Spratt et al., 2019). Nav1.2 is widely expressed throughout the central nervous system with subcellular localization in axons and somatodendritic regions, including unmyelinated axons and axon initial segments (AIS) of excitatory neurons (and some interneurons) (Li et al., 2014;Sanders et al., 2018). Loss-of-function mutations in Scn2a are strongly associated with ASD and intellectual disability in patients, whereas gain-of-function mutations cause encephalopathy with early-onset infantile epilepsy (Sanders et al., 2012(Sanders et al., , 2018. In some individuals, Scn2a mutations can lead to both epilepsy and autistic features (Kamiya et al., 2004;Wolff et al., 2017). Several recent studies have shown that heterozygous Scn2a knockout mice (Scn2a +/-) faithfully model features of autism without epilepsy, including impaired social behavior and synaptic dysfunction (Léna & Mantegazza, 2019;Spratt et al., 2019;Tatsukawa et al., 2019).
In addition to causing autistic-like phenotypes, heterozygosity for the Scn2a knockout allele (i.e., Scn2a +/-) also acts as a protective modifier of epilepsy and sudden death in mice carrying a deletion of the Kcna1 gene (i.e., Kcna1 -/-), which encodes Kv1.1 voltage-gated potassium channel α-subunits that regulate axonal excitability (Mishra et al., 2017). In Kcna1 -/mice, the Scn2a +/mutation increases survival, decreases seizure durations, and improves brain-heart dynamics, likely as a result of the channels' complementary functions and expression patterns (Mishra et al., 2017). Given the beneficial epistatic interactions between these two ion channel mutations in the context of epilepsy, in this work a battery of behavioral tests was employed to investigate whether the Scn2a and Kcna1 mutations interact to modify genotype-phenotype relationships in the context of autism. First, we examined whether Kcna1 -/mice exhibit autistic-like tendencies in measures of repetitive and social behaviors. Second, we generated double-mutant Scn2a +/mice with various combinations of Kcna1-null alleles to test the hypothesis that Kcna1 deletion can act as a protective modifier of Scn2a-related autistic-like phenotypes. The results of this study provide a window into the types of complex genetic interactions underlying autismepilepsy comorbidity, while also identifying Kcna1 as a modifier of ASD susceptibility and Kv1.1 subunits or related networks and pathways as potential therapeutic targets in autism due to Scn2a mutations.

| Animals and genotyping
The Kcna1 knockout (KO) allele was generated by targeted deletion of the entire open reading frame of the Kcna1 gene (chromosome 6), as described previously (Smart et al., 1998). Kcna1 mice are maintained on a Black Swiss (Tac:N:NIHS-BC) genetic background. The initial behavioral analyses of Kcna1 −/− knockout mice and controls described in the results were performed on animals from the Tac:N:NIHS-BC genetic background. The Scn2a KO allele was created by targeted deletion of the first half of exon 1 of the Scn2a gene (chromosome 2), as described previously (Planells-Cases et al., 2000). backgrounds. Although these percentages should be similar between animals, they are not identical and could influence the outcome of the behavioral tests. Mice were group-housed at ~22°C, fed ad libitum, and submitted to a 12-hr light/dark cycle. All procedures were performed in accordance with the guidelines of the National Institutes of Health (NIH), as approved by the Institutional Animal Care and Use Committee of the Louisiana State University Health Sciences Center-Shreveport.
For genotyping, genomic DNA was isolated by enzymatic digestion of tail clips using Direct PCR Lysis Reagent (Viagen Biotech).
In behavioral experiments, age-matched juvenile mice of both sexes were tested at 28-34 days old. The total number of animals tested was between 6 and 12 per genotype per experiment, as specified. To obtain greater statistical power, mice of both sexes were grouped for analyses. For tests of repetitive behavior, 2-10 males and 2-7 females per genotype were used. For tests of social interactions, 3-9 males and 1-5 females per genotype were used. The exact numbers of mice of each sex per genotype are indicated in the figure legends. All experiments were conducted during the daytime and analyzed by individuals blinded to genotype. Animals displaying overt behavioral seizure activity immediately before or during behavioral assessment were excluded from analysis; however, only Kcna1 −/− and Scn2a +/-; Kcna1 -/mice exhibit seizures. Animals were exposed to each test one time over 3 days in the following order: marble burying (day 1); self-grooming (day 1, 25 min after marble burying test); sociability and social novelty (day 2); and nestlet shredding (day 3).

| Marble burying
Mice were individually placed in a standard shoebox rat cage (~143-in 2 floor space; Allentown, Inc.) containing a 4-cm layer of fresh bedding (Sani-Chips; Lab Supply) with 20 marbles arranged in 5 equally spaced rows of 4 marbles each (i.e., a 4 × 5 grid pattern), as previously described (Angoa-Pérez et al., 2013). Marbles were scored as buried if they were at least two-thirds covered at the end of a 30-min test session. The number of buried marbles was scored by two individuals independently, and their results averaged to give a value for each animal.

| Nestlet shredding
Mice were individually placed in a standard shoebox mouse cage (~75-in 2 floor space; Allentown, Inc.) with an intact preweighed cotton nestlet (2-in square; Lab Supply) and left overnight for 16 hr.
At the end of the session, the remaining intact (nonshredded) portion of the cotton nestlet was removed from the cage, allowed to dry overnight, and weighed. The percentage of the nestlet shredded was then calculated using the measurements of the intact nestlet mass before and after introduction of the mouse, as previously described (Angoa-Pérez et al., 2013).

| Self-grooming
Mice were individually placed in a standard shoebox mouse cage with ~1 cm of fresh bedding (Sani-Chips; Lab Supply) and allowed to habituate for 5 min. Then, the mice were observed for 15 min to quantify grooming behaviors, as previously described (Silverman et al., 2010). Grooming behavior was defined as stroking or scratching of face, head, or body with two forelimbs, or licking body parts.
Total grooming time was scored by two trained observers independently using a stopwatch, and the average for each animal was calculated.

| Sociability and social novelty
Mice were tested for social interactions using a rectangular plexiglass three-chamber box (66-cm length × 45-cm width × 22-cm height), composed of a middle chamber separated by dividing walls with retractable doorways from two flanking and equalsized side chambers (20-cm length × 45-cm width), as previously described (Crawley, 2007;Moy et al., 2004). First, the test mouse was placed in the middle chamber for 5 min to habituate. Then, a stranger mouse (S1) was placed in one of the side chambers under an inverted wire pencil cup (10-cm diameter × 10-cm height). In the other, side chamber was placed an empty inverted wire pencil cup (E) as a nonsocial object. The doorways were then raised, and the test mouse allowed to explore the side chambers for 10 min.
The inverted wire pencil cups were composed of vertical wires that were spaced widely enough (~7 mm) to allow the test mouse to see and sniff the stranger mouse while preventing aggressive or sexual interactions. Plastic cups with weights in them were placed on top of the inverted pencil cups to prevent them from moving during the test. Sociability was quantified as the preference of the test mouse for the stranger mouse (S1) versus the inanimate object alone (E). For testing social novelty, immediately at the conclusion of the sociability test, a second stranger mouse (S2) was placed under the empty inverted wire pencil cup. The test mouse was then allowed access to the side chambers for another 10 min. Social novelty was measured as the preference of the test mouse for the new stranger mouse (S2) compared to the previous stranger mouse (S1).
For both tests, the following variables were measured during the 10-min trial periods to quantify chamber preference and social interaction: the total time spent in each chamber; the total time the test mouse spent sniffing the stranger mouse or the empty wire cup; and the number of entries into each chamber.
The time spent in each chamber provides a general gauge of interest, while sniffing time gives a more specific measure of social investigation. The number of chamber entries between compartments provides a general measure of locomotion and activity. Times and counts were measured manually by offline video observation performed by two individuals blinded to genotype whose results were then averaged to give a value for each animal.
Two mice (one Scn2a +/-; Kcna1 -/mouse and one WT mouse from the mixed background) that failed to explore each of the two side chambers for a minimum of 90 s (15% of the total test time) were excluded from analysis.
Stranger mice were age-and sex-matched to the test mice. In addition, the stranger mice (which were selected to always have a Kcna1 +/genotype) were from different litters than the test mice, and they were never housed together with the test mice or with the other stranger mice used in the same experiment. The location of stranger mice was alternated between the left and right chambers for each experiment for a given genotype to eliminate potential side bias. New clean wire pencil cups were used for each experiment, and the surfaces of the three-chamber box were wiped clean with Clidox in between animals.

| Statistical analysis
Statistical analysis was performed with Prism 8 for Windows (GraphPad Software Inc, La Jolla, CA), and the data are expressed as means ± standard error of the mean (SEM). The normality of the data was checked using the D'Agostino-Pearson test. For comparison of repetitive behavior and social interaction, unpaired two-tailed Student's t tests (involving only two groups) and one-way ANOVA with Tukey's multiple comparison post hoc tests (involving three or more groups) were performed, as appropriate. p Values < .05 were considered statistically significant.

| Kcna1 -/mice exhibit normal social interactions
To determine whether Kcna1 -/mice (Tac:N:NIHS-BC genetic background) exhibit abnormal social interactions, sociability and an F I G U R E 1 Kcna1 -/mice exhibit drastic reductions in repetitive behaviors. Repetitive behaviors of Kcna1 -/mice (n = 12:5 males, 7 females) and wild-type (WT) control animals (n = 13:6 males, 7 females) were evaluated in tests of marble burying (a), nestlet shredding (b), and self-grooming (c). **, p < .01; ***, p ≤ .001 (unpaired two-tailed Student's t test) interest in social novelty were assessed using a three-chamber social approach test. Three-chamber tests are routinely used for the assessment of ASD-like social deficits in mice (Kaidanovich-Beilin et al., 2011;Nadler et al., 2004;Silverman et al., 2012). In the sociability phase of the examination, the preference of the mouse was compared for a novel social object (a stranger mouse in an inverted wire cup; S1) versus a novel nonsocial object (an empty inverted wire cup; E). In tests of sociability, both Kcna1 -/-(n = 7) and WT mice (n = 7) spent significantly more time exploring the chamber containing the stranger mouse than the chamber with the empty cup, indicative of normal social behavior in the animals (two-tailed Student's t test, WT: t(12) = 2.83, p = .015; Kcna1 -/-: t(12) = 4.65, p = .0006; Figure 2a). In addition, both genotypes spent significantly more time sniffing the cup with the mouse in it than the empty cup, further suggesting a preference for social interaction (two-tailed Student's t test, WT: t(12) = 4.67, p = .0005; Kcna1 -/-: t(12) = 5.40, p = .0002; Figure 2a). Finally, Kcna1 -/and WT mice both exhibited similar numbers of entries into each of the side chambers demonstrating similar levels of exploratory activity and a lack of innate side preference. Thus, Kcna1 -/mice exhibited normal sociability.
In the social novelty phase of the experiment, the preference of the test mouse was compared for the original stranger mouse (S1) versus a novel stranger mouse (S2). Kcna1 -/mice (n = 7) spent similar amounts of time in each of the two side chambers suggesting a lack of preference for the unfamiliar mouse (two-tailed Student's t test, t(12) = 0.37, p = .72). However, WT mice (n = 7) also unexpectedly exhibited a similar lack of preference (two-tailed Student's t test, t(12) = 0.20, p = .85; Figure 2b), which limits the interpretation of social novelty features in Kcna1 -/mice. Both genotypes exhibited increased time spent sniffing the new stranger mouse (S2), but these differences did not reach statistical significance (two-tailed Student's t test, WT: t(12) = 1.31, p = .21; Kcna1 -/-: t(12) = 0.95, p = .36; Figure 2b). The number of transitions into the two side chambers was indistinguishable between genotypes suggesting normal exploration. Thus, Kcna1 -/exhibit an interest in social novelty similar to their WT littermates.

females) and
Kcna1 -/mice (n = 7:3 males, 4 females) using three-chamber tests of sociability (a) and social novelty preference (b). In the sociability test, both genotypes exhibited a preference for the stranger mouse (S1) over the empty cup (E) as evidenced by the significantly greater amount of time spent in the S1 chamber (versus the E chamber) and in sniffing the S1 mouse (versus the empty cup). In the subsequent social novelty test, the preference for the original S1 mouse disappeared with the introduction of a second novel stranger mouse (S2) as both genotypes spent similar amounts of time exploring the side chambers and sniffing both stranger mice. Both genotypes exhibited a similar number of transitions from the middle chamber into the two side chambers suggesting no innate side preference. *, p < .05; ***, p < .01 (unpaired twotailed Student's t test) allelic combinations of Kcna1 and Scn2a were then subjected to the same battery of tests for repetitive and social behaviors as performed above.
In the tests of social novelty, Scn2a +/mice failed to show a preference for the novel stranger mouse over the original familiar stranger mouse, spending significantly more time in the chamber with the original mouse (two-tailed Student's t test, t(18) = 3.97, p = .0009; Figure 4b). In addition, they sniffed both the new and old stranger mice for similar amounts of time further demonstrating an abnormal lack of interest in the socially novel animal. The number of entries into the side chambers was the same for each social interaction test in Scn2a +/mice, as well as for all other genotypes detailed below ( Figure 4). Thus, the increased repetitive self-grooming behavior and abnormal interest in social novelty in Scn2a +/mice are consistent with autistic-like features, providing further support for their utility as a mouse model of autism. In addition, their significant reductions in marble burying could be a correlate of the restricted interests observed in ASD.
Interestingly, nonepileptic heterozygous Kcna1 +/mice exhibited behavioral abnormalities that were similar to Scn2a +/mice, suggesting that partial Kcna1 gene deletion may cause haploinsufficiency for autistic-like phenotypes. With regard to repetitive behaviors, Kcna1 +/mice showed about 50% increases in grooming behavior (1-way ANOVA, F(5,58) = 11.70, p < .0001; Tukey's post hoc, q(58) = 4.38, p = .0341; Figure 3c) relative to WT, which was comparable to the increases in Scn2a +/mice. In addition, they also exhibited 73% reductions in nestlet shredding (1-way ANOVA, F(5,58) = 9.17, p < .0001; Tukey's post hoc, q(58) = 4.91, p = .0121; Figure 3b) versus WT, a deficit not present in Scn2a +/mice. However, marble burying in Kcna1 +/mice was normal, but in Scn2a +/mice it was decreased. Whereas Scn2a +/mice exhibited abnormal social interactions manifested as a lack of interest in social novelty, Kcna1 +/mice displayed a lack of sociability showing no significant preference for the stranger mouse when measured as either chamber time or sniffing. Thus, double-mutant Scn2a +/-; Kcna1 +/mice represent a unique situation in which autistic-like features are mutually ameliorated by the combination of two gene mutations that by themselves produce autistic-like behavioral deficits.

| D ISCUSS I ON
This study uncovers a potential new role for the Kcna1 gene in ASD pathophysiology. Kcna1 deletion modified ASD-relevant repetitive and social behaviors in mice in a gene dosage-dependent manner.
Complete gene deletion in Kcna1 -/mice led to drastic reductions in all repetitive behaviors tested, including decreased grooming, marble burying, and nestlet shredding. In contrast, partial gene deletion in heterozygous Kcna1 +/mice produced an opposite behavioral phenotype consistent with ASD, including increased grooming and decreased sociability. Taken together, these findings suggest that Kv1.1 subunits are important in pathways and neural networks underlying ASD and that Kcna1 may be a modifier gene for ASD susceptibility. Interestingly, double-mutant mice that are heterozygous for the Kcna1 deletion mutation and carry the ASD-associated Scn2a +/mutation (i.e., Scn2a +/-; Kcna1 +/-) exhibited an ameliorated ASD phenotype due to potential reciprocal masking interactions between the two genes. Thus, Kv1.1 subunits may also represent a potential therapeutic target for ASD due to Scn2a mutations. These findings illustrate the complexity of genotype-phenotype relationships in ASD, demonstrating that genetic variants that increase disease risk can also exert beneficial effects depending on the genetic milieu.
The ability of Kcna1 and Scn2a mutations, which both cause ASD-like behaviors, to mutually suppress one another resulting in a double-mutant mouse with relatively normal behavior underscores the importance of elucidating protective modifier interactions for improving genomic risk assessment and identifying new therapeutic targets. ASD has a significant and complex genetic etiology often involving interactions between multiple loci, such as the up to 1,000 genes estimated to influence ASD susceptibility (Krumm et al., 2015;Pickles et al., 1995;Risch et al., 1999;Spratt et al., 2019). Although genetic risk factors are usually thought of in the negative sense, as deleterious variants that add together to cause disease when a threshold is crossed, some genetic variants may be beneficial modifiers that reduce disease risk. Clinical studies of ASD leave room for the possible presence of these protective variants. For example, the most common genetic anomaly associated with ASD is deletion of the 16p11.2 chromosomal region; however, ~75% of carriers are asymptomatic for the disease, including carrier parents of children with ASD (Hanson et al., 2015). Mutations in the SHANK2 gene are another prevalent molecular cause of ASD in patients; however, many individuals have been identified that carry potentially deleterious SHANK2 mutations without any apparent psychiatric disorder, leading to the hypothesis that these healthy individuals may possess counteracting suppressor mutations (Leblond et al., 2012).
Interestingly, in double-mutant mice carrying the epilepsy-causing TA B L E 1 Summary of behavioral phenotypes by genotype Notes.: Downward red arrowheads ( ) and upward green arrowheads ( ) indicate statistically significant decreases or increases, respectively, relative to WT controls. Circles with lines through them (Ø) indicate no significant change relative to WT controls.
Kcna1 -/mutation in addition to another epileptogenic mutation in the P/Q-type calcium channel gene Cacna1a (Cacna1a tottering ), epilepsy was mutually suppressed (Glasscock et al., 2007). Similarly, this work finds masking of ASD-related features in mice by combining two ion channel mutations associated with ASD-like behaviors.
However, additional work will be needed to identify where exactly in the brain Scn2a and Kcna1 mutations interact to affect neuronal function. Two possibilities are the neocortex and hippocampus where both Nav1.2 and Kv1.1 are prominently expressed and exhibit shared subcellular distributions (Duménieu et al., 2017). For example, in neocortex Nav1.2 and Kv1.1 are both expressed on axon initial segments and dendrites of pyramidal cells, whereas in hippocampus both are present on unmyelinated Schaffer collateral axons (Guan et al., 2006;Lorincz & Nusser, 2008;Monaghan et al., 2001;Spratt et al., 2019;Tian et al., 2014;Westenbroek et al., 1989). Given this potential expression overlap, coexisting mutations in the Scn2a and Kcna1 genes would be expected to have dramatic but opposing complementary effects on action potential waveform characteristics (Duménieu et al., 2017).
Protective modifiers are important from a treatment standpoint since each one represents a potential therapeutic target. Variants in genes that support synaptic function are highly implicated in ASD. Therefore, a key focus of drug development in ASD has been identifying compounds that improve synaptic processes (Ghosh et al., 2013;Spooren et al., 2012). Even though ASD is a neurodevelopmental disorder, studies in mice show that symptoms can be reversed by treatment even in adulthood. Several mouse models of ASD and genetic syndromes that are highly associated with ASD (including Rett syndrome, FXS, and tuberous sclerosis) show improvement of behavioral deficits, synaptic plasticity, and cognitive measures when treated in adulthood (Ehninger et al., 2008;Guy et al., 2007;Mei et al., 2016;Michalon et al., 2012;Silverman et al., 2012). The genetic amelioration of Scn2a +/-ASD phenotypes by partial Kcna1 deletion suggests that modulating synaptic function by targeting Kv1.1 channels may represent a new strategy for treatment of Scn2a-associated ASD.
The heterozygous Scn2a +/mice in our study exhibited autis- adult Scn2a +/mice (12 weeks old) in their study did not exhibit differences in three-chamber tests of sociability and social novelty.
Finally, Spratt et al. tested for ASD-like behavior in adult Scn2a +/mice (3-4 months old) and found no major deficits in nesting, grooming, and two-chamber social interaction tests between the test mouse and a stranger mouse (Spratt et al., 2019). Phenotypic differences between the Scn2a +/mice in our study and the others could be largely attributed to differences in age and genetic background.
Our animals were tested at a young age (28-34 day) when ASD-like behaviors in Scn2a +/mice are the most robust, whereas most of the other studies were performed in adults. Furthermore, the animals in our experiments were tested on a mixed Tac:N:NIHS-BC x C57BL/6J genetic background compared to a straight C57BL/6J background in the other studies. Despite some differences in results between studies, the pattern that emerges is that Scn2a +/mice display age-dependent deficits in repetitive and social behavior consistent with a role for the gene in ASD pathophysiology.
In addition to the repetitive and social behavioral abnormali-  Roundtree et al., 2016). About 9% of autistic children exhibit brain overgrowth with enlargements in the cerebral hemispheres, cerebellum, caudate nucleus, and in some studies the amygdala and hippocampus (Sacco et al., 2015;Schumann et al., 2004). In mice, deletion or truncation of Kcna1 causes megalencephaly due to enlargement of the hippocampus and ventral cortex beginning during the first postnatal month (Diez et al., 2003;Persson et al., 2007). In autism, one of the most consistently abnormal brain regions is the cerebellum, and one of the most common microscopic findings is a decrease in the number of Purkinje cells, the sole source of output from the cerebellar cortex (Fatemi et al., 2012). Kv1.1 channels are prominently expressed in the cerebellum, including high levels in basket cell axons which provide GABAergic inhibition to Purkinje cells (Wang et al., 1994). In mice, Kcna1 deletion and missense mutations cause increased basket cell excitability leading to excessive inhibition of Purkinje cells, which can lead to stress-induced motor dysfunction resembling the episodic ataxia in humans due to KCNA1 mutations (Browne et al., 1994;Herson et al., 2003;Zhang et al., 1999). FXS, caused by mutations in the FMR1 gene, is one of the most prevalent causes of monogenic autism representing about 2%-5% of all ASD diagnoses (Mila et al., 2018 Recent clinical findings in epileptic encephalopathy patients point to KCNA1 mutations being a potential molecular risk factor for ASD susceptibility. Epileptic encephalopathy (EE) refers to the process whereby early-onset seizures and epileptiform activity impair brain function leading to cognitive, behavioral, and language deficits (Howell et al., 2016). ASD has an especially high prevalence in patients with epileptic encephalopathy. At least 34 of the 62 genes implicated in EE have also been associated with ASD, including sodium (e.g., SCN1A and SCN2A) and potassium channel genes (e.g., KCNQ2 and KCNQ3), and several EE genes have ASD prevalence rates as high as 70%-100% (Srivastava & Sahin, 2017). Whole exome sequencing has now identified KCNA1 mutations as the cause of EE in 5 child patients (Rogers et al., 2018;Verdura et al., 2019). In one patient with infantile EE, a de novo single-nucleotide variant in KCNA1 was found in the highly conserved Pro-Val-Pro motif of Kv1.1 that is essential for channel gating (Paulhus et al., 2020;Rogers et al., 2018).
Interestingly, at age 4, this individual was also diagnosed with pervasive developmental disorder (PDD), which is an umbrella term for the group of neurodevelopmental disorders that includes ASD (i.e., autistic disorder, Asperger's disorder, and PDD-not otherwise specified), childhood disintegrative disorder, and Rett's syndrome (McPartland & Volkmar, 2012;Rogers et al., 2018). Thus, the ability of Kcna1 mutations to cause ASD-like features in animal models and now a patient suggests that KCNA1 should be considered as a new genetic risk factor for ASD phenotypes.
Sex-specific differences in behavior and genetic interactions in our work cannot be ruled out because mice of both sexes were grouped together for analyses of each genotype to achieve greater statistical power. The drawback to this approach is that if behavioral sex differences are present, they could influence the results. Notably, however, other studies have demonstrated that Scn2a +/mice do not exhibit significant sex differences in social and repetitive behaviors (including nesting and grooming) (Spratt et al., 2019). In addition, for the genotypes with larger sample sizes in our study (e.g., 5-7 per sex), major sex differences were not apparent (Table S1). Nonetheless, in future studies, it may be important to examine potential effects of sex on the observed behavioral genetic interactions between Kcna1 and Scn2a.

| CON CLUS ION
These findings demonstrate the importance of elucidating interactions between genetic variants to improve neuropsychiatric disease risk assessment. As shown here, even mutations that independently produce ASD-like behavioral deficits can act as protective variants in the right genetic milieu leading to mutual normalization of ASD phenotypes. This work also sheds light on the complexity of ASDepilepsy comorbidity, showing that epilepsy-associated mutations (such as Kcna1 deletion) can produce epileptic or autistic phenotypes depending on gene dosage. The presence of ASD-like behaviors in both mice and an epileptic encephalopathy patient due to KCNA1 mutations suggests that KCNA1 may represent a new susceptibility gene for ASD phenotypes. In addition, the ability of partial deletion of Kcna1 to improve ASD phenotypes in Scn2a +/suggests Kv1.1 subunits as a novel target for therapy in autism due to Scn2a channelopathy.

ACK N OWLED G M ENTS
This work was supported by the National Institutes of Health (grant numbers R01NS100954 and R01NS099188 to E.G.).

CO N FLI C T O F I NTE R E S T
The authors declare they have no conflict of interest.

AUTH O R CO NTR I B UTI O N S
E.G. and J.I. conceptualized the study and wrote the paper. J.I., A.P., N.M.G., and K.C. performed the experiments. E.G., J.I., and A.P. analyzed the data.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.