A KCNC3 mutation causes a neurodevelopmental, non-progressive SCA13 subtype associated with dominant negative effects and aberrant EGFR trafficking

The autosomal dominant spinocerebellar ataxias (SCAs) are a diverse group of neurological disorders anchored by the phenotypes of motor incoordination and cerebellar atrophy. Disease heterogeneity is appreciated through varying comorbidities: dysarthria, dysphagia, oculomotor and/or retinal abnormalities, motor neuron pathology, epilepsy, cognitive impairment, autonomic dysfunction, and psychiatric manifestations. Our study focuses on SCA13, which is caused by several allelic variants in the voltage-gated potassium channel KCNC3 (Kv3.3). We detail the clinical phenotype of four SCA13 kindreds that confirm causation of the KCNC3R423H allele. The heralding features demonstrate congenital onset with non-progressive, neurodevelopmental cerebellar hypoplasia and lifetime improvement in motor and cognitive function that implicate compensatory neural mechanisms. Targeted expression of human KCNC3R423H in Drosophila triggers aberrant wing veins, maldeveloped eyes, and fused ommatidia consistent with the neurodevelopmental presentation of patients. Furthermore, human KCNC3R423H expression in mammalian cells results in altered glycosylation and aberrant retention of the channel in anterograde and/or endosomal vesicles. Confirmation of the absence of plasma membrane targeting was based on the loss of current conductance in cells expressing the mutant channel. Mechanistically, genetic studies in Drosophila, along with cellular and biophysical studies in mammalian systems, demonstrate the dominant negative effect exerted by the mutant on the wild-type (WT) protein, which explains dominant inheritance. We demonstrate that ocular co-expression of KCNC3R423H with Drosophila epidermal growth factor receptor (dEgfr) results in striking rescue of the eye phenotype, whereas KCNC3R423H expression in mammalian cells results in aberrant intracellular retention of human epidermal growth factor receptor (EGFR). Together, these results indicate that the neurodevelopmental consequences of KCNC3R423H may be mediated through indirect effects on EGFR signaling in the developing cerebellum. Our results therefore confirm the KCNC3R423H allele as causative for SCA13, through a dominant negative effect on KCNC3WT and links with EGFR that account for dominant inheritance, congenital onset, and disease pathology.


Introduction
Patients with dominant cerebellar ataxias display adult-onset, progressive motor incoordination, and cerebellar atrophy [1][2][3][4][5]. Previously, we reported causation of the autosomal dominant disorder SCA13 by mutations in the voltage-gated potassium channel gene, KCNC3 (MIM: 176264, Kv3.3) [6][7][8]. This tetrameric-delayed rectifier channel facilitates rapid firing of action potentials in the cerebellum, hippocampus, and brainstem [9][10][11][12]. Two allelic forms of SCA13 [7,8,13], p.Arg420His (KCNC3 R420H ) and p.Phe448Leu (KCNC3 F448L ) [7], have been described. Their phenotypes are distinct in that KCNC3 R420H results in a slowly progressive, adult-onset ataxia, whereas KCNC3 F448L presents in childhood with delayed motor milestones. Screening of ataxia DNA repositories identified a third mutation, g.10693G>A; p. Arg423His (KCNC3 R423H ), displaying early-onset SCA13 [14][15][16]. We report a detailed phenotypic description of this allelic form in a child who presented at age 7 months and in three additional multigeneration kindreds with multiple affected persons. Features include infantile onset with delayed gross and/or fine motor milestones, tremor, seizures, cognitive impairment, gait and/or appendicular ataxia, and dysarthria. Magnetic resonance imaging (MRI) confirms marked cerebellar hypoplasia as early as 10 months of age. Longitudinal follow-up demonstrates non-progressive cerebellar hypoplasia, with lifetime improvement in motor and cognitive function. Therefore, SCA13 R423H may be considered a congenital ataxia causing a fixed deficit, that is, it is partially overcome by normal development, with eventual accomplishment of motor and cognitive milestones. We provide supportive evidence for neurodevelopmental onset through a Drosophila model. Biophysical studies and experiments on cellular localization also address channel activity and protein trafficking. Our studies also speak to a cellular basis for dominant inheritance and congenital cerebellar hypoplasia.

Human genotyping and MRI
Patient DNA was isolated from blood (QIAamp Blood kit) or saliva (Oragene) following written informed consent and approval from the Institutional Review Board of the University of Florida, Gainesville, Florida (IRB project . Written informed consent for minors was obtained from the next of kin. This study was performed in accordance with the Declaration of Helsinki. All patient-derived sequencing was performed with specific primers (S1 Table) at the DNA Sequencing Core, University of Florida. Midline sagittal T1-weighted MR images of patients were collected at multiple institutions.

Drosophila genetics and imaging
Fly stocks were obtained from the Bloomington Drosophila Stock Center (flystocks.bio.indiana.edu, S2 Table). All crosses were at 25˚C unless indicated otherwise. Adult wings (n = 6) and frozen adult eyes (n = 7) were processed and imaged, as previously described [19].

Immunoblot analysis
Protein from individual 2-day-old flies or CHO or U87 cells was analyzed, as previously described, [19] with the modification of resolution with 3-8% Tris-acetate gels (Life Technologies). Co-immunoprecipitation experiments were performed with Dynabeads Protein G (Thermo Fisher Scientific) using α-EGFR (Abcam) for binding and α-EGFR (EMD Millipore) and α-KCNC3 (Alomone Labs) for detection. Full-length blots are presented in S1 Fig.

Statistical analysis
Statistical analysis with SEM values are presented using an unpaired, two-tailed t-test for significance ( ÃÃÃ , p<0.001). All analyses were performed using Microsoft Excel and GraphPad Prism.
The MR images for individuals 423-3, III-2 at 7 and 17 years (Fig 1I and 1J) and 423-4, III-4 at 16 and 26 years (Fig 1K and 1L) similarly illustrate severe cerebellar hypoplasia. Strikingly, these two individuals display minimal age-dependent atrophy in 10-year serial scans (compare Fig 1I to 1J; 1K to 1L). This contrasts to the KCNC3 R420H allele where patients demonstrate marked progression of cerebellar atrophy over time with concomitant symptom progression  [8]. Furthermore, individual III-2 (423-3, Fig 1C, 1I and 1J) was wheelchair-bound during early adolescence, transitioning to a walker, then to a cane, and ambulating independently by late teens, although not completely normally. Patient III-4 (423-4, Fig 1D, 1K and 1L) was initially diagnosed with cerebral palsy and moderate cognitive impairment, but progressed to unassisted ambulation and/or running and normal-range cognition by age 26 years. A motor evaluation in 2013 yielded a SARA (Scale for the Assessment and Rating of Ataxia) score [21] of 13, which remained unchanged at12-month follow-up. Both patients therefore display nonprogressive cerebellar hypoplasia, along with displaying many other activity-dependent improvements in motor and cognitive milestones. Phenotypic data and ataxic parameters (S3 and S4 Tables), demonstrate uniformity in clinical features within and across families. These kindreds establish several novel aspects not typical of other SCAs [4,[22][23][24]: a neurodevelopmental pattern with infantile onset, non-progression of cerebellar atrophy, and clinical symptoms accompanied by cognitive and motor improvement suggestive of compensatory neural mechanisms despite severe cerebellar hypoplasia.

Expression of human KCNC3 R423H in Drosophila melanogaster wing and eye
Developmental consequences of R423H expression were examined in strains of Drosophila melanogaster expressing human KCNC3 WT and KCNC3 R423H controlled by the yeast UAS/ Gal4 expression system [25]. All driver and responder fly strains are summarized in S2 Table. Expression of KCNC3 WT and KCNC3 R423H in transgenic flies was verified by immunoblot (Fig 2A) from individual 2-day-old flies controlled by the ubiquitous daughterless (da)-Gal4 driver [26]. We expressed LacZ, KCNC3 WT , or KCNC3 R423H in the developing wing ( Fig 2B-2G), along the anteroposterior boundaries extending to the anterior compartment using the decapentaplegic (dpp)-Gal4 driver. Gal4/KCNC3 R423H expression caused complete loss of the anterior cross vein (ACV) and partial loss of vein L3 ( Fig 2G). The flies were kept at 29˚C to achieve maximal Gal4-transgene expression with minimal effects on viability and fertility [27]. KCNC3 R423H expression in the developing wing pouch, using the A9-Gal4 driver [28], resulted in severely altered wing morphology and vein patterning (Fig 2H-2J).
To test the mutant allele in a developing neuronal lineage, we utilized the eye-specific driver gmr-Gal4 [29] for expression throughout eye development. KCNC3 R423H expression displayed marked eye dysmorphology and a profound reduction in size (Fig 2M), disrupted ommatidial organization, fused ommatidia, and malformed eye bristles (Fig 2P and 2S) compared to those of a normal eye phenotype in KCNC3 WT and a control LacZ (Fig 2K, 2L, 2N, 2O, 2Q and 2R). Sagittal sections of the KCNC3 WT eyes illustrate regular arrangement of each lens at the apex of the elongated ommatidium, in contrast with the disrupted pattern observed in the KCNC3 R423H eyes (S2A and S2B Fig). In addition to the adult eye, we stained the eye imaginal discs from wandering third instar larvae for the pan-neuronal marker elav [30]. Control KCNC3 WT larval imaginal discs revealed expected organized ommatidia (S2C Fig), whereas KCNC3 R423H larvae showed disorganized and fused ommatidia with maldeveloped smaller ommatidial clusters (S2D Fig). Detection of chaoptin, a photoreceptor cell-and axon-specific membrane protein required for cell morphogenesis [31], illustrates normal organization of the in 423-2. Midline T1-weighted sagittal magnetic resonance images (MRIs) of (E) a 35-year-old control; (F) patient 423-1, II-3 at age 42 years; (G) patient 423-1, III-1 at age 10 months (inset shows age-matched control); (H) de novo patient 423-2, II-2 at age 21 months. Midline T1-weighted sagittal MRIs of (I,J) patient 423-3, III-2 at age 7 and 17 years, respectively; and (K,L) patient 423-4, III-4 at age 16 and 26 years, respectively, demonstrating the lack of progressive cerebellar hypoplasia and/or atrophy.

KCNC3 R423H intracellular location and biophysics
We recently reported that the KCNC3 R420H allele [7,8,32] displayed altered post-translational modifications with aberrant retention in the Golgi [19]. CHO cells, with no detectable endogenous KCNC3, transiently transfected with human KCNC3 WT (Fig 3A), show normal plasma membrane localization by immunofluorescence. In contrast, cells expressing KCNC3 R423H (Fig 3C; compare to KCNC3 R420H in Fig 3B), demonstrate aberrant trafficking with strong perinuclear staining analogous to the KCNC3 R420H allele [19]. As illustrated with other representative cells (Fig 3B and 3C insets) aberrant trafficking is consistently observed with both mutants. To corroborate the aberrant trafficking, we then generated C-terminal Clover-tagged vectors [17] harboring. Confocal microscopy illustrates normal plasma membrane localization ( Fig 3D) for KCNC3 WT , in contrast to absent plasma membrane localization and clear perinuclear retention for each mutant (Fig 3E and 3F). This aberrant trafficking is found in all cells from multiple independent transfection experiments. For further comparison, we also expressed the Clover-tagged SCA13 mutation from a French pedigree, KCNC3 F448L , which displays normal plasma membrane trafficking (Fig 3G), although associated with altered biophysical properties [7].
To demonstrate that protein overexpression was not responsible for aberrant localization, we show that over about an 8-fold range of transfected plasmid concentrations, with vectors harboring human KCNC3 WT and KCNC3 R423H C-terminally tagged with mCerulean3 [18], KCNC3 R423H-mCerulean3 remains intracellularly localized, with no detectable plasma membrane trafficking, in contrast to normal localization with equivalent concentrations of the KCNC3 WT  This finding clearly demonstrates that channel mis-trafficking is not attributable to protein expression levels.
To determine the intracellular localization of the mutant channel, we used fluorescently tagged markers of endoplasmic reticulum (ER), Golgi, and intracellular vesicles. Sigma receptor 1 (SIGMAR1) is a highly conserved, transmembrane chaperone protein located in the ER membrane [33]. On co-expression of the cyan fluorescent protein (CFP) tagged SIGMAR1 (SIGMAR1 CFP ) with KCNC3 R423H-mRuby2 (Fig 3H-3J), we observed no co-localization, which rules out aberrant retention in the ER. JMY is linked to cytosolic actin assembly and acts as a DNA damage-induced transcriptional co-activator of p53 [34,35]. Recently, Schlüter et al. [36] also demonstrated JMY's role in vesicular trafficking at the trans-Golgi network through actin-dependent elongation and/or tubulation of anterograde vesicles. Co-expression of JMY GFP with KCNC3 R423H-mRuby2 resulted in co-localization (Fig 3K-3M) in vesicular structures, implying the retention of KCNC3 R423H in trafficking vesicles. PI4K2A, a membranebound phosphatidylinositol-4 kinase, localizes to the trans-Golgi network (TGN) and early endosomes [37,38] and is responsible for phosphorylation of phosphatidylinositol (PI) to phosphatidylinositol 4-phosphate (PI4P), a critical lipid in endocytosis, Golgi function, protein sorting, and membrane trafficking [39]. Co-expression of KCNC3 R423H-mRuby2 with PI4K2A GFP clearly demonstrates that the mis-trafficked mutant channel accumulates in the intravesicular space of PI4K2A GFP -positive vesicles (Fig 3N-3S). These results suggest that KCNC3 R423H is retained in the interior of anterograde or endosomal vesicles rather than being incorporated into the vesicular membrane.
As an integral membrane protein, N-glycans are added to the nascent KCNC3 protein in the ER with trimming of the oligosaccharide precursor, a critical quality-control measure for proper glycoprotein folding and vesicular trafficking through the Golgi en route to the plasma membrane. The expression of human KCNC3 R423H in CHO (Fig 3T) or U87 cells (Fig 3U) demonstrates altered glycosylation patterns consistent with our previous studies on KCNC3 R420H [19].
As a voltage-dependent potassium channel, KCNC3/Kv3.3 displays high activation thresholds with fast activation and deactivation kinetics, conveying the property of sustained trains of high-frequency action potentials in neurons expressing them [40]. To assess the functional consequences of the KCNC3 R423H mutation, we performed electrophysiology in CHO cells transiently expressing the human KCNC3 WT or KCNC3 R423H channels. For KCNC3 WT , Fig  3V illustrates the slow inactivating outward current evoked during application of depolarizing voltage steps (increments of 10 mV from −80 mV to +70 mV). In contrast, cells expressing the KCNC3 R423H channels demonstrate the complete absence of current conductance by the mutant channel over the same voltage range (Fig 3W), consistent with aberrant glycosylation and intracellular retention.

KCNC3 R423H has a dominant effect on KCNC3 WT
The consequences of an increased dosage of the KCNC3 R423H allele were determined by creating flies harboring 2× copies of KCNC3 R423H , resulting in a more severe eye phenotype entirely lacking ommatidia (Fig 4B compared to Fig 4A). Moreover, simultaneous expression of exogenous KCNC3 WT with KCNC3 R423H did not overcome the eye phenotype (Fig 4C compared to  Fig 4A), which implicates the dominant nature of the R423H allele. Similarly, KCNC3 R423H self-cross, driven by dpp-Gal4 in the wing, led to disruption of veins L2, L3, and ACV in the wing (Fig 4E compared to Fig 4D), whereas co-expression of KCNC3 WT with KCNC3 R423H resulted in a phenotype similar to KCNC3 R423H alone (Fig 4F compared to Fig 4D). Studies in the eye and wing thus support the strong dominant negative effects exerted by KCNC3 R423H .
KCNC3 functions as a tetrameric voltage-gated potassium channel [41,42]. This disease presents as an autosomal dominant phenotype, [7,32] presumably resulting from the formation of a heterotetramer composed of WT and mutant monomers. We have determined the probabilities of heterotetramer interactions based on WT:423 ratios of 1:1 to 6:1 (S4A and S4B  Fig). Biophysical studies of CHO cells co-expressing KCNC3 WT : KCNC3 R423H at a ratio of 1:1 demonstrate a significant reduction in current amplitude and mean current densities compared to those of cells expressing KCNC3 WT alone (Fig 5A, 5B and 5C), thereby showing a dominant electrophysiological effect. Studies using other model systems have implied that the dominant phenotypic effects manifested by KCNC3 R423H and KCNC3 R420H mutations are based on alterations in channel electrophysiology [43]. These results are not supported by our data [19] (Fig 3A-3S), which demonstrate that neither mutation is appreciably trafficked to the plasma membrane. To address the mechanism underlying the dominant phenotypic effects of KCNC3 R423H , we also used differentially C-terminally labeled mutant and WT channels to explore intracellular trafficking and tetrameric protein association. KCNC3 WT-Clover was coexpressed with KCNC3 R423H-mRuby2 in cells at ratios from 1:1 to 6:1 and visualized by confocal microscopy (Fig 5D-5S). As expected, KCNC3 WT-Clover alone traffics normally to the plasma membrane (Fig 5D), whereas KCNC3 R423H-mRuby2 alone displays complete retention in intracellular vesicles (Fig 5G). Co-expression with increasing ratios of KCNC3 WT : KCNC3 R423H demonstrates complete co-localization (Fig 5H-5S) suggestive of tetrameric co-assembly, along with complete retention of the KCNC3 WT protein in the same intracellular vesicles even at a ratio of 6:1 (Fig 5Q-5S).

dEgfr rescues the Drosophila KCNC3 R423H eye phenotype
To identify the Drosophila pathway(s) affected by KCNC3 R423H , we evaluated modifiers of the mutant phenotypes upon co-expression with a series of eye and wing specific determinants, including dEgfr, Ras, Rolled (mitogen-activated protein kinase [MAPK]) and Notch [44][45][46] (S2 Table). Effects of Egfr overexpression (UAS-Egfr.B) [47] were evaluated on a gmr-Gal4 (Fig 6A and 6B) or gmr-Gal4, UAS-KCNC3 R423H background (Fig 6C and 6D). Elevated expression of dEgfr with the UAS-Egfr.B allele resulted in a striking rescue of the KCNC3 R423H eye phenotype (Fig 6D), which supports a link between the mutant voltage-gated potassium channel and Egfr or its downstream signaling pathway. To further demonstrate interactions between KCNC3 R423H and Egfr in eye development, we expressed two dEgfr RNAi strains in the context of  [48,49]. The MF then travels anteriorly, where Notch signaling establishes specification of the initial R8 photoreceptor neurons as the founding cells of the developing ommatidia. Signaling from the Egfr pathway leads to the recruitment of the remaining photoreceptor cells in a pairwise manner (R2/5, R3/4, R1/ 6, and finally R7), along with non-neuronal cone and pigment cells [46]. Therefore, we also evaluated the effects of Notch overexpression and found no effect on the KCNC3 R423H eye phenotype (S7D and S7H Fig), further supporting specificity for the interaction of the mutant potassium channel allele with Egfr. The connection with EGFR was further corroborated with experiments in the wing demonstrating that co-expression of Egfr-RNAi and KCNC3 R423H Pathological mechanisms associated with a congenital non-progressive SCA13 causative mutation causes more significant abnormalities in wing vein development than either strain alone ( Fig  6E-6G). These results thus provide evidence that the developmental effects of KCNC3 R423H expression may be mediated through disruption of EGFR signaling.

KCNC3 R423H allele contributes to aberrant trafficking of EGFR
To test whether the potassium channel directly interacts with EGFR in mammalian cells, we transiently transfected U87 cells, known to express EGFR [50], with either human KCNC3 WT or KCNC3 R423H .Cellular lysates were subjected to immunoprecipitation with EGFR-specific antibodies. As expected, U87 cells expressed EGFR (Fig 7A, left panel), while immunoblot analyses of the identical immunoprecipitation lysates showed no detectable bands in either the KCNC3 WT or KCNC3 R423H transfected cells (Fig 7A, right panel), despite clear overexpression of KCNC3 (Fig 7B).
With no apparent direct protein-protein interaction, we examined the potential influence of aberrant KCNC3 R423H cellular trafficking on EGFR transit through the ER/Golgi to the plasma KCNC3 R423H-Clover , and in those transfected with both constructs KCNC3 WT-Clover :KCNC3 R423H-mRuby2 in a 1:1 ratio. (B) Mean current densities recorded in CHO cells expressing either wild-type KCNC3 WT-Clover (n = 7) or KCNC3 R423H-Clover (n = 5) and in those expressing both KCNC3 WT-Clover : KCNC3 R423H-mRuby2 (1:1) constructs (n = 6). Current density was calculated by dividing the peak current evoked by a step from −70 to +70 mV by cell capacitance. Values are shown as mean±SEM, and significance was tested using a one-way ANOVA.  Pathological mechanisms associated with a congenital non-progressive SCA13 causative mutation membrane. Cells were co-transfected with KCNC3 WT-mCerulean3 or KCNC3 R423H-mCerulean3 and with varying concentrations of a Citrine-tagged human EGFR [51] construct, using KCNC3: EGFR molar ratios from 5:1 to 1:1 (Fig 7C-7Q). Representative confocal Z-stacks of cells expressing both KCNC3 WT and EGFR (Fig 7C-7E) illustrate that, despite an excess of KCNC3 WT (5:1), both proteins trafficked normally to the plasma membrane. Conversely, coexpression of KCNC3 R423H and EGFR at ratios of 5:1 (Fig 7F-7H) and 4:1 (Fig 7I-7K) reproducibly led to aberrant trafficking of EGFR with sequestration in intracellular vesicles that co-register with KCNC3 R423H (Fig 7H and 7K). As the molar ratio of KCNC3 R423H :EGFR is decreased to 2:1 (Fig 7L-7N), EGFR is found both intracellularly sequestered and at the plasma membrane. At a ratio of 1:1, EGFR traffics normally to the plasma membrane, while in some cells the receptor continues to intracellularly co-register with KCNC3 R423H (Fig 7O-7Q). The titration was repeated with KCNC3 R423H tagged with a different fluorescent protein, mRuby2, co-expressed with EGFR Citrine at ratios of 3:1 (S8 Fig). These results further illustrate that complete normal trafficking of the receptor to the plasma membrane is achieved only with reduction of the molar ratio of KCNC3 R423H :EGFR to 0.7:1. To determine whether the mutant channel could cause aberrant trafficking of other membrane proteins, we co-expressed KCNC3 WT or KCNC3 R423H with N-cadherin (Cadherin-2 [CDH2]) fused with eGFP [52] at molar ratios of 3.5:1 and 2.5:1, with no effects on plasma membrane localization for cadherin (S9 Fig). Together with the rescue of the Drosophila eye phenotype (Fig 6E and 6G), these observations strongly implicate a specific link between this causative mutant allele and EGFR.

Discussion
Previous studies [43,53,54] identified KCNC3 R423H by screening index ataxia patients from US [15] and European [14,16] DNA repositories. Our analysis of four pedigrees (3 US, 1 Swedish) unequivocally demonstrates SCA13 causation for this allele. The unifying endophenotype includes infantile onset, non-progressive cerebellar hypoplasia, lifetime improvement of motor and cognitive function, bradyphrenia, dysarthria, tremor, and the classic SCA feature of limb, truncal, and gait ataxia (S3 and S4 Tables). A striking clinical observation is the timeand activity-dependent improvement in motor and cognitive function, despite severe congenital cerebellar hypoplasia (Fig 1).
To address this unique clinical pathology, we have developed fly and cellular models to investigate the underlying developmental, cellular, and biochemical events illuminating mechanisms of neurodevelopmental onset and dominant inheritance. Without SCA13 autopsy specimens, our current understanding of human pathology derives from clinical history and MRI-based cerebellar hypoplasia.
Drosophila eye differentiation has previously been used to model alleles causative in SCAs [23,[55][56][57][58]. Consistent with neurodegeneration, all these mutant alleles display large, Pathological mechanisms associated with a congenital non-progressive SCA13 causative mutation disorganized eyes or a "rough eye" phenotype. In contrast, overexpression of human KCNC3 R423H in adult flies results in small, maldeveloped eyes exhibiting fused and disorganized ommatidia and disordered eye bristles, along with aberrant wing veins and shrunken wing formation, which suggests a neurodevelopmental effect. Consistent with a congenital phenotype, expression of KCNC3 R423H in third instar larvae caused disturbed patterning and fusion of ommatidial clusters, axonal bundle thinning, and reduced photoreceptor cell clusters. Collectively, these data support the neurodevelopmental nature of KCNC3 R423H in SCA13 patients.
Normal trafficking of plasma membrane-targeted proteins involves glycosylation and folding in the ER compartment believed to be followed by vesicular transport to the Golgi for glycan trimming and protein sorting [59]. Vesicles originating at the trans-Golgi network are involved in anterograde transport to the plasma membrane through close association with microtubules and actin filaments [60][61][62]. Our data show that KCNC3 R423H expressed in cell culture is abnormally glycosylated and does not reach the plasma membrane. Co-localization results demonstrate that KCNC3 R423H is sequestered in the intravesicular space of either anterograde or endosomal vesicles based on striking co-expression with PI4K2A, a vesicular membrane marker. Consistent with this intracellular sequestration are biophysical studies unequivocally demonstrating absent current conductance in cells expressing this mutation. These data provide an explanation for the deleterious effects of the mutation but alone do not address the dominant inheritance.
To explore the dominant phenotype in SCA13 in light of its tetrameric nature, we calculated the probabilities of tetramer formation relative to expressed ratios of KCNC3 WT and KCNC3 R423H (S4A and S4B Fig). We provide three lines of experimental evidence that KCNC3 R423H exerts a strong dominant negative effect on KCNC3 WT monomers (Figs 4 and 5). Self-crosses in the fly eye and wing show that KCNC3 WT does not overcome the effects of the mutant allele, along with co-expression in cell culture, where at a ratio as high as 6:1 (WT: R423H), KCNC3 WT remains mis-trafficked and intracellularly localized. Functionally, these data are further supported with biophysical data showing that <20% current conductance is detectable at a 1:1 ratio of WT:R423H. Collectively, these data form a strong basis for dominant inheritance but do not address the neurodevelopmental phenotype.
The Egfr and Notch morphogenetic pathways are prominent pathways involved in Drosophila eye and wing development [46,63,64]. dEgfr is required for recruitment of neuronal and non-neuronal cells in the ommatidium [29,65]. Because the profound KCNC3 R423H eye phenotype can be almost completely rescued by elevated Egfr expression and R423H/Egfr-RNAi co-expression accentuates wing malformations, a direct or indirect association is strongly implicated between KCNC3 and Egfr, as well as a link between this potent growth factor receptor and cerebellar hypoplasia. Consistent with the fly data, co-expression of EGFR with KCNC3 R423H in mammalian cells leads to aberrant intracellular retention of EGFR that co-registers in vesicles with the mutant channel, the same anterograde and/or endosomal vesicles positive for PI4K2A. Normal EGFR plasma membrane localization is overcome only with a decrease in the concentration of the mutant channel below a ratio of 1:1. Negative co-immunoprecipitation data confirms that the intracellular retention of EGFR by KCNC3 R423H is not likely to be dependent on a direct protein-to-protein interactions. Therefore, our current hypothesis involves the intracellular sequestration of EGFR through yet unknown effects of KCNC3 R423H in anterograde or endosomal vesicles during cerebellar development. This hypothesis can be rationalized in part by previous studies in rats proving the presence of EGFR in Purkinje cells during late cerebellar development [66,67]. It is also relevant to note that PI4K2A has been shown to co-localize with protein markers of the late endosome and is required for endocytic trafficking and degradation and/or downregulation of EGFR [68].
Future studies to address the interplay of KCNC3 R423H , EGFR, and PI4K2A will provide important insights into the mechanisms governing normal cerebellar development and cerebellar hypoplasia in SCA13.
Supporting information S1