Proper localization of the SCA13-linked K⁺ channel in Purkinje neurons requires a microtubule-stabilizing protein

Abstract

Kv3.3 voltage-gated K⁺ (Kv) channels are highly expressed in cerebellar Purkinje neurons and some hippocampal neurons, aligning with the motor and cognitive impairments observed in spinocerebellar ataxia 13 (SCA13) caused by Kv3.3 mutations. Despite their functional significance, the mechanisms governing Kv3.3 subcellular localization remain poorly understood. Here we report microtubule-associated protein 6 (MAP6) regulates Kv3.3 axon-dendrite targeting. MAP6 deletion reduces Kv3.3 levels in the processes of Purkinje neurons. Mechanistically, MAP6’s 1st and 2nd Mn modules directly bind the external surface of the Kv3.3 N-terminal T1 tetramer, while its 3rd Mn module indirectly associates with Cav2 Ca²⁺ channels. In Purkinje neurons, shRNA-mediated MAP6 knockdown decreases somatodendritic levels of both Kv3.3 and Cav2.1 (associated with SCA6). Notably, expression of Mn1/2-GFP selectively reduces Kv3.3, but not Cav2.1, levels. Purkinje neuron burst firing is reduced in both conditions. These findings uncover a MAP6-dependent mechanism for targeting two key ion channels linked to SCAs.

1. Introduction

Kv3.3 (KCNC3) voltage-gated K⁺ (Kv) channel proteins are widely expressed across multiple brain regions, with particularly high concentrations in the dendrites of cerebellar Purkinje neurons and the axons of the hippocampal dentate granule cells (or known as mossy fibers) (Chang et al., 2007; Goldman-Wohl et al., 1994; Gu and Barry, 2011; Kaczmarek and Zhang, 2017; Martina et al., 2003). As a member of the Kv3 subfamily, Kv3.3 channels possess distinct biophysical properties, including a more depolarized activation potential (~ −10 mV) and fast activation and deactivation kinetics compared to other Kv channels (Gu et al., 2018; Rudy and McBain, 2001; Zhang et al., 2016). Unlike some other Kv3 channels, Kv3.3 also displays inactivation, contributing to complex spikes or high-frequency burst firing of action potentials in Purkinje neurons (Hurlock et al., 2008; Zagha et al., 2008). Multiple missense mutations in the Kv3.3 gene are linked to spinocerebellar ataxia type 13 (SCA13), a neurodegenerative disorder characterized by motor and cognitive impairments (Figueroa et al., 2011; Waters et al., 2006; Zhang and Kaczmarek, 2016), consistent with the enriched expression of Kv3.3 in the cerebellum and hippocampus. Early studies identified a PDZ-binding motif in the Kv3.3C-terminus that mediates its dendritic localization in both cerebellum and electrosensory lateral line lobe of the weakly electric fish Apteronotus leptorhynchus (Deng et al., 2005). While our recent research has uncovered several molecular mechanisms underlying Kv3.1 trafficking and axonal targeting (Gu et al., 2011; Ma et al., 2023; Xu et al., 2007; Xu et al., 2010), Kv3.3 channels differ in that they localize to both axons and dendrites, without a strong axonal preference. Despite their important functions, the molecular mechanisms governing Kv3.3 subcellular localization in neurons remain largely unknown.

Microtubule-associated protein 6 (MAP6) was originally identified for its unique ability to confer remarkable cold stability to microtubules and subsequently recognized as a vertebrate-specific, multi-functional protein (Cuveillier et al., 2021). Reduced MAP6 expression has been reported in humans with mental disorders (Chen et al., 2021; Martins-de-Souza et al., 2009; Shimizu et al., 2006; Wei et al., 2016; Xiao et al., 2015). Our recent studies demonstrated that MAP6 physically binds Kv3.1 channels, stabilizing them in the soma and axons of parvalbumin-positive (PV+) GABAergic interneurons, thereby influencing behavior (Ma et al., 2023). This interaction aligns with behavioral phenotypes observed in both MAP6^−/− and Kv3.1^−/− mice, which display hyperactivity and avoidance reduction (Ma et al., 2023). Interestingly, only MAP6^−/− mice exhibit pronounced deficits in motor coordination, balance, and spatial memory, suggesting that MAP6 may regulate motor and cognitive functions through additional ion channels beyond Kv3.1 (Ma et al., 2023). We found that MAP6 binds Kv3.1 through its N-terminal T1 domain—a region highly conserved across the Kv3 channel subfamily (Ma et al., 2023). This raises the intriguing possibility that MAP6 may also bind Kv3.3 channels, potentially influencing their localization and function in CNS neurons.

MAP6 also interacts with Cav2 (Cav2.1-Cav2.3) voltage-gated Ca²⁺ (Cav) channels, a connection first identified through proteomic analysis of the nano-environment surrounding these channels (Muller et al., 2010). Subsequent studies showed that the Mn3 module of MAP6 directly binds to Tctex1, a dynein light chain known to associate with Cav2 channels, suggesting a role for MAP6 in the trafficking and sorting of these channels (Brocard et al., 2017). Supporting this, calcium signaling is disrupted in MAP6 knockout (KO) neurons (Brocard et al., 2017). Notably, MAP6 interacts with different channels through distinct modular domains: Kv3.1 channels bind to the first two Mn modules (Mn1 and Mn2) (Ma et al., 2023), while Cav2 channels, via Tctex1, associate with the third Mn module (Mn3) (Brocard et al., 2017). Among Cav2 family members, Cav2.1 (P/Q-type) channels are crucial for neurotransmitter release in the CNS and are prominently expressed in the somatodendritic compartments of cerebellar Purkinje neurons (Westenbroek et al., 1995). Importantly, expansion of CAG repeats (encoding a polyglutamine tract) within the Cav2.1 gene leads to spinocerebellar ataxia type 6 (SCA6) (Watase et al., 2008; Zhuchenko et al., 1997). However, whether MAP6 regulates the dendritic targeting of Cav2.1 channels in cerebellar Purkinje neurons remains an open question.

In this study, we demonstrate that Kv3.3 and Cav2.1 channels interact with distinct domains of microtubule-stabilizing protein MAP6, which is essential for their proper targeting to dendrites of cerebellar Purkinje neurons. In addition to its known interaction with Kv3.1, MAP6 also binds Kv3.3 via the N-terminal T1 domain conserved among Kv3 subfamily members, thereby regulating both axonal and dendritic targeting of Kv3.3 in Purkinje neurons. Disruption of the Kv3.3-MAP6 interaction leads to a marked reduction in burst firing in these neurons, highlighting the functional importance of this molecular interaction.

2. Methods

2.1. Antibodies

Mouse monoclonal antibodies: the anti-MAP6 antibody (STOP (175), mAb #4265, Cell Signaling Technology, Inc), anti-6xHis antibody (N144/14, NeuroMab), anti-Kv3.1b antibody (N16b/8, NeuroMab), anti-Kv3.3 antibody (N375/67, NeuroMab), anti-β-tubulin Antibody (Cat. No. 05–661-I, Millipore), and anti-Calbindin-D28K antibody (Cat. 66,394–1-Ig, Proteintech). Rabbit polyclonal antibodies: the anti-MAP6 antibody (NBP214220, Novus Biologicals), anti-Kv3.1b antibody (APC-014, Alomone Labs), anti-Kv3.3 antibody (APC-102, Alomone Labs), and anti-CACNA1A (anti-Cav2.1) antibody (ACC-001, Alomone Labs). Secondary antibodies: AffiniPure Donkey Anti-Mouse IgG (H + L) Horseradish Peroxidase (715–035–151, Jackson ImmunoResearch Inc), AffiniPure Donkey Anti-Mouse IgG (H + L) Alexa Fluor^® 594 (715–585-151 Jackson ImmunoResearch Inc), AffiniPure Donkey Anti-Mouse IgG (H + L) Alexa Fluor^® 647 (715–605-151, Jackson ImmunoResearch Inc), AffiniPure Donkey Anti-Rabbit IgG (H + L) Horseradish Peroxidase (711–035-152, Jackson ImmunoResearch Inc).

2.2. Maintenance of knockout (KO) mouse lines and genotyping procedures

All animal experiments have been conducted in accordance with the NIH Animal Use Guidelines and were approved by the Ohio State University Institutional Animal Care and Use Committee (IACUC). The MAP6^−/− mouse line was kindly provided by Dr. Annie Andrieux (Andrieux et al., 2002) and used in our recently published paper (Ma et al., 2023). MAP6^+/−;Thy1-YFP and MAP6^−/−;Thy1-YFP were obtained by crossing the MAP6^+/− and Thy1-YFP-H transgenic mice (The Jackson Laboratory Stock #003782). These mouse lines have been maintained using a PCR-based genotyping procedure. The following primers were used for MAP6 KO: reverse primer (for both WT and MAP6 KO, 5′-CTGGGAAAACCAGTGTGGAAACTGTTA-3′), forward primer 1 (for WT, 5′-GCAGATGCCCTCAACAGGCAAATCCGG-3′), and forward primer 2 (for MAP6 KO, 5′-GATTCCCACTTTGTGGTTCTAAGTACTG-3′). WT band: 270 bp. Mutant band: 400 bp. On the B6 background, the proportion of MAP6^−/− mice among the offspring is substantially lower than expected by Mendelian inheritance, likely due to increased mortality in homozygous animals. Therefore, our study primarily utilizes MAP6^+/− mice. For Thy1-YFP-H transgenic mice, forward primer (oIMR-1258): 5′-TCTGAGTGGCAAAGGACCTTAGG-3′; reverse primer (oIMR-1260): 5′-CGCTGAACTTGTGGCCGTTTACG-3′. Band size: 300 bp.

Kv3.3 KO (Kv3.3^−/−) mouse line was purchased from the Jackson Laboratory (Stock #: 048285-UCD; MMRRC). The following primers were used for genotyping Kv3.3 KO mouse line: reverse primer 1 (for Kv3.3 KO, 5′-CACCCTACTCTGCCTGAACC-3′), reverse primer 2 (for WT, 5′-CACCGTCTTGTTGCTGATGT-3′), and forward primer (for both WT and Kv3.3 KO, 5′-CCTGAGCCTCTGTTCCTGTC-3′). WT band: 215 bp. Mutant band: 283 bp.

2.3. Cardiac perfusion, fixation, and immunostaining

We have described the procedures of cardiac perfusion, fixation, sectioning, staining, and imaging in detail in our published papers (Barry et al., 2014; Gu et al., 2017; Jukkola et al., 2012; Jukkola et al., 2013; Ma et al., 2023; Sun et al., 2022). In brief, mice (3–5 months old) were anesthetized with Ethasol (Virbac; 100 mg/kg) and were transcardially perfused with 20 mL of PBS followed by 20 mL of a 4% formaldehyde/PBS solution. Then, mouse brains were removed, post-fixed for 1 h in 4% formaldehyde/PBS solution, cut into 3 mm blocks using an acrylic brain matrix (Braintree Scientific, Braintree, MA, USA), and cryoprotected in a 30% sucrose/PBS solution for 1–3 days. Brain blocks were embedded in optimal cutting temperature (OCT) media (Sakura Finetek USA, Inc., Torrance, CA, USA) and were stored at −80 °C until sectioning. We used a Microm HM550 cryostat (Thermo Scientific, Waltham, MA, USA) to cut brain blocks into 40-μm-thick slides that were then collected on Superfrost Plus microscope slides (FisherScientific, Pittsburgh, PA, USA) for storage at −20 °C. After immunostaining, we mounted the sections with glass coverslips using tris-buffered Fluoro-Gel mounting media (Electron Microscopy Sciences, Hatfield, PA, USA).

2.4. Conventional fluorescence microscopy and confocal microscopy

Low magnification fluorescence images were captured with a Spot CCD camera RT slider (Diagnostic Instruments) in a Nikon inverted microscope, Axiophot, using Plan Apo objectives 20×/0.75 and 100×/1.4 oil, saved as 16-bit TIFF files, and analyzed with NIH ImageJ and SigmaPlot 14.0 for fluorescence intensity quantification and colocalization, as previously described (Gu et al., 2017; Ma et al., 2023; Rice et al., 2019; Sun et al., 2022; Xu et al., 2010). Exposure times were controlled so that the pixel intensities within the image field were below saturation, but the same exposure time was used for each fluorophore in an experiment. Fluorescence confocal microscopy and image stack capturing were carried out using Zeiss LSM 900 confocal system (Carl Zeiss AG, Oberkochen, Germany) with Airyscan SR. Images(.czi file) were taken using Zeiss 63× oil objectives (numerical aperture at 1.40) with ZEN 3.0 (blue edition) software. We captured Z-stack images at ~0.20 μm steps for the selected areas of interest. Image stacks were analyzed with NIH ImageJ. Staining and imaging have been replicated at least for 3 rounds.

2.5. cDNA plasmids

Kv3.3, GST-Kv3.1 T1, GST-Kv3.3 T1, His-3.1 T1, GST, MAP6N-GFP (aa 1–952), His-MAP61 (aa 1–221), His-MAP62 (aa 36–221), His-MAP63 (aa 222–451), and His-MAP64 (aa 452–615) were published in our papers (Gu et al., 2018; Ma et al., 2023; Xu et al., 2007; Xu et al., 2010). GFP-Mn, YFP-Mn, and CFP-Mn were constructed by inserting the corresponding PCR fragments into pGFP-C1/pYFP-C1/p-CFP-C1 between BglII and HindIII. YFP-Kv3.3 was constructed by inserting the corresponding PCR fragments into pYFP-C1 between BglII and SalI.

2.6. Tissue co-immunoprecipitation (co-IP)

Tissue co-IP and Western blotting were prepared as previously described(Ma et al., 2023; Xu et al., 2010). In brief, mouse brains were homogenized in homogenization buffer (50 mm Tris buffer, pH 8.0, 1 mm EDTA, and a Complete protease inhibitor tablet (Cat. #5892970001, Roche, Basel, Switzerland). Crude membranes were pelleted from the suspension by high-speed centrifugation at 10350 rpm for 30 min at 4 °C, solubilized in IP buffer (50 mm Tris-Cl, pH 7.4, 150 mm NaCl, 1% Triton X-100, and a Complete protease inhibitor tablet) for 2 h at 4 °C, and centrifuged at 10350 rpm for 30 min at 4 °C. The supernatant (1 mL in each condition) was incubated (4 h, 4 °C) with 40 μl of protein G-agarose beads (Cat. #11719416001, Roche, Basel, Switzerland) and 3 μl of mouse monoclonal anti-Kv3.1b antibody (Cat. #AB2877376, Neuromab, CA, USA), mouse monoclonal anti-Kv3.3 antibody (N375/67, Neuromab, CA, USA), mouse monoclonal anti-MAP6 antibody (Cat. #4265, Cell Signaling Technology, Massachusetts, USA), or control mouse IgG (Sigma). The beads were washed six times with the IP buffer and eluted with 2× sample buffer. The immunoprecipitants were resolved by SDS-PAGE, transferred to a polyvinylidene difluoride membrane, and subjected to Western blotting with the mouse monoclonal anti-MAP6 antibody (1:1000 dilution). Each experiment was repeated 3 times with consistent results.

2.7. Protein pulldown assays and Western blotting

Procedures of protein pulldown assays and Western blotting were used as previously described (Ma et al., 2023; Xu et al., 2007; Xu et al., 2010). The expression of GST- or 6xHis-tagged proteins was induced with 1 mM Isopropyl β-D-1thiogalactopyranoside (IPTG) in E. coli BL21 cells for 4 h at 37 °C. The bacterial pellets were solubilized with sonication in a pull-down buffer (50 mm Tris-HCl, pH 7.4, 150 mm NaCl, 1% Triton X100, and a Complete protease inhibitor tablet) at 4 °C, and centrifuged at 10,350 ×g for 30 min at 4 °C. The supernatants were incubated with 50 μl of glutathione beads at 4 °C for 2 h. After extensive washing, the beads coated with purified GST fusion proteins were further incubated with bacterial lysate supernatant containing 6xHis-tagged fusion proteins. The proteins precipitated were eluted with a 2× sample buffer, resolved in SDS-PAGE, transferred to PVDF membrane, and subjected to immunoblotting with an anti-His antibody (Cat. #AB2877254, Neuromab, CA, USA). The SDS-PAGE gels with GST fusion inputs were stained with the Colloidal Blue staining kit (Cat. #LC6025, Invitrogen, Massachusetts, USA).

2.8. Prediction of protein complex structures using AlphaFold 3

We used PyMOL 3.0.3 (Schrödinger LLC) to visualize the published structure of the entire Kv3.1a tetrameric channel complex (PDB code: 7PHI) (Chi et al., 2022). Structural complexes of the Mn1/2 domain of MAP6 (a.a. 115–221) with either the T1 domain of Kv3.1 (a.a. 2–124) or the one of Kv3.3 (a.a. 84–208) were predicted using AlphaFold 3 (https://alphafoldserver.com/, developed by DeepMind and Isomorphic Lab) (Abramson et al., 2024). The inputs were provided in FASTA format, and the predicted structures were downloaded in CIF format. The CIF files from AlphaFold 3 were opened with PyMOL and exported as PDB (.pdb) for structural analysis. Labeling of individual subunit or amino acid residues were performed using PyMOL.

2.9. Hippocampal neuron culture, transfection, and imaging

Hippocampal neuron culture was prepared as previously described (Barry et al., 2014; Gu et al., 2006; Ma et al., 2023). In brief, 2 d after neuron plating, 1 μm cytosine arabinose (Sigma) was added to the neuronal culture medium to inhibit glial growth for the subsequent 2 d, then replaced with the normal neuronal culture medium. The culture medium was replenished twice a week by replacing half the volume. For transient transfection, neurons in culture at 5–7 DIV were incubated in Opti-MEM containing 0.8 μg of cDNA plasmid and 1.5 μl of Lipofectamine 2000 (Invitrogen, Massachusetts, USA) for 20–30 min at 37 °C. The procedures of immunostaining and fluorescence microscopy were described previously (Barry et al., 2014; Gu et al., 2006). In brief, the transfected neurons expressing Kv3.1bHA (or its mutants) and MAP6N-GFP (or its truncation) were fixed, permeabilized, and stained with a rat monoclonal anti-HA antibody (Roche, Basel, Switzerland)

2.10. Fluorescence Resonance Energy Transfer (FRET) imaging

To further evaluate the binding between Kv3.3 channel and MAP6 Mn modules, we performed fluorescence resonance energy transfer (FRET) imaging as we described in our prior papers (Barry et al., 2013; Gu et al., 2001). YFP-Kv3.3 and Mn-CFP cDNAs were co-transfected into cultured hippocampal neurons at 5 DIV. FRET imaging was performed 2–4 days later.

2.11. AAV-shRNA knockdown of MAP6 and AAV-mediated expression of Mn-GFP

Recombinant adeno-associated viral vectors (AAV) carrying GFP and shRNA, or Mn-GFP, were used in this study. AAV-Control-GFP, AAV-MAP6-shRNA, and AAV-Mn-GFP (aa 115–221 from MAP6N) were published in our previous paper (Ma et al., 2023). The viruses were packaged and purified by the Boston Children’s Hospital virus core.

2.12. Stereotaxic microinjection of AAV into different brain regions in mice

The surgery and microinjection procedures are similar to those described in earlier publications (Barry et al., 2014; Ma et al., 2023). Briefly, mice were anesthetized with a mixture of 100 mg/kg ketamine and 20 mg/kg xylazine (Sigma-Aldrich, St. Louis, MO). Mice were mounted on a stereotaxic frame, and a small straight incision was made along the midline of the head to expose the underlying skull. The bregma coordinates were recorded and used as the reference point. The stereotaxic injection setup consisted of a Hamilton syringe connected to a 33-gauge injector cannula (Plastics One, Roanoke VA). A volume of 2–10 μl of the virus was injected per mouse, at a rate of 0.1–0.25 μl/min controlled by a syringe pump. Viral vectors were injected into the cerebellum (AP: −6.6 mm; LR: −2 mm; DV: −2.5 mm). After the infusion was finished, the injector was left in place for 2 min before being raised. The mice were sutured after the surgery and administered post-operative care for one week. After 3–4 weeks of recovery, the GFP fluorescence was used to verify the injected brain region, AAV infection, and MAP6 knockdown.

2.13. Whole-cell recording of Purkinje neurons from mouse cerebellar slices

Mice aged 8–11 weeks were first anesthetized with intraperitoneal ketamine/xylazine (100 mg/kg / 10 mg/kg) and then perfused for 2 min with ice-cold artificial cerebrospinal fluid (ACSF) containing (in mM): 100 sucrose, 80 NaCl, 3.5 KCl, 24 NaHCO₃, 1.25 NaH₂PO₄, 4.5 MgCl₂, 0.5 CaCl₂, and 10 glucose, saturated with 95% O₂ and 5% CO₂. After decapitation, the brain was dissected in ice-cold ACSF. Sagittal slices of cerebellar cortex (300 μm) were cut using a Leica VT 1200S vibratome (Leica Microsystems) and incubated in the above solution at 35 °C for 30 min post-dissection. Slices were then maintained at room temperature until use in recording ACSF containing (in mM): 130 NaCl, 3.5 KCl, 24 NaHCO₃, 1.25 NaH₂PO₄, 1.5 MgCl₂, 2.5 CaCl₂, and 10 glucose, saturated with 95% O₂ and 5% CO₂.

For recording, slices were constantly perfused with ACSF at 2 mL/min at a temperature of 31–33 °C. Cells were visualized using an upright microscope (Scientifica SliceScope) with a 40× water-immersion objective (Olympus), camera (Scientifica SciCam Pro), and Oculus software. GFP-expression was identified by fluorescence (CoolLED pE-300ultra). Recording pipettes were pulled from borosilicate glass (World Precision Instruments) to a resistance of 3–5 MΩ using a vertical pipette puller (Narishige PC-100). Whole-cell patch clamp recordings were acquired using a Multiclamp 700B amplifier (Molecular Devices), filtered at 3 kHz (Bessel filter), and digitized at 20 kHz (Digidata 1550B and pClamp v11.1, Molecular Devices). Recordings were not corrected for liquid junction potential. Series resistance (5–25 MΩ) was closely monitored, and recordings were excluded from analysis if series resistance exceeded 25 MΩ. In current-clamp mode, cells were biased to a membrane potential (Vm) of −70 mV. The internal solution used to collect intrinsic membrane properties contained (in mM): 130 K-gluconate, 5 KCl, 2 NaCl, 4 MgATP, 0.3 NaGTP, 10 phosphocreatine, 10 HEPES, 0.5 EGTA, and 0.2% biocytin (Sigma-Aldrich catalog #B4261).

To calculate burst hyperpolarization, the membrane potential was biased to −70 mV (V_initial) followed by injection of a 1-s-duration positive current step at twice the current amplitude required to reach action potential threshold. The local minima of the complex spike located nearest 667 ms of the pulse duration (V_min) were then used to calculate the burst hyperpolarization as (V_min – V_initial). The bursts/current step was calculated by counting the number of complex spikes that occurred during the 1-s current step at twice the action potential threshold. The spikes/burst value was calculated by counting the number of spikes fired in the complex spike located nearest 667 ms of the pulse duration. Data from individual cells were treated as independent measurements, and each condition was replicated in two mice.

2.14. Statistical analysis

Sample sizes were chosen based on our previous experiments and publications, as well as related literature in the field. No formal randomization was used to allocate samples to the experimental condition. Whenever possible, the investigator was blind to the sample conditions. In AAV injection experiments, post hoc analyses of viral infection and protein expression were used to exclude mice with no or incorrect AAV infection from behavioral results. Statistical analysis was conducted using SigmaPlot 15.0 software. To assess normality, the Shapiro-Wilk test was employed, and the homogeneity of variances was evaluated using the Brown-Forsythe test. Results were presented as the mean ± SEM. Two-tailed Student’s t-test was used for comparisons between two groups. One-way ANOVA followed by Dunnet’s test was used for comparing two or more groups to one control group. (*) p < 0.05, (**) p < 0.01, and (***) p < 0.001 were considered statistically significant. Two-Way ANOVA followed by Holm-Sidak multiple comparisons testing, and (*) p < 0.05, (**) p < 0.01, and (***) p < 0.001 were considered statistically significant.

3. Results

3.1. Colocalization of Kv3.3 and MAP6 in dendrites and axons of cerebellar neurons

To investigate the potential colocalization of MAP6 and Kv3.3 in the cerebellum, we performed co-immunostaining on mouse cerebellar tissue. In wild-type (WT) mice, MAP6 was predominantly localized to the primary dendrites of Purkinje neurons, with minimal presence in the soma. In contrast, Kv3.3 channel proteins were enriched in both soma and dendrites of Purkinje neurons (Fig. 1A). Colocalization of MAP6 and Kv3.3 was also observed in a subset of neurons and their processes within the deep cerebellar nuclei (DCN). Interestingly, some neuronal cell bodies in this region contained both MAP6 and Kv3.3, whereas others expressed only Kv3.3 (Fig. 1B). High-magnification confocal imaging further revealed strong colocalization of Kv3.3 and MAP6 in the primary dendrites of Purkinje neurons in WT mice. In contrast, Kv3.3—but not MAP6—was enriched in Purkinje neuron soma. Additionally, MAP6 and Kv3.3 were found to colocalize in certain axonal structures from both Purkinje neurons and other cerebellar neurons (Fig. 1C). Notably, unlike neurons in the DCN, Purkinje neurons consistently lacked MAP6 enrichment in their soma. In Kv3.3^−/− mice, Kv3.3 staining was absent, while MAP6 remained predominantly localized to dendrites rather than the soma of Purkinje neurons (Fig. 1D). In MAP6 heterozygous and homozygous knockout (KO) mice (MAP6^+/− and MAP6^−/−), Kv3.3 protein levels were significantly reduced in both somatic and dendritic compartments of Purkinje neurons (Fig. 1E). These findings indicate that MAP6 and Kv3.3 proteins colocalize within dendrites and axons of cerebellar Purkinje neurons and suggest that MAP6 is required for the proper targeting or stabilization of Kv3.3 channels in these cells.

Fig. 1.

Dendritic targeting of Kv3.3 channels in cerebellar Purkinje neurons is disrupted in MAP6 KO mice.

A, Colocalization of Kv3.3 and MAP6 in the low magnification image of cerebellar cortex of WT mice.

B, Colocalization of Kv3.3 and MAP6 in some neurons in the deep cerebellar nuclei (DCN).

C, Confocal image of Kv3.3 and MAP6 staining signals in WT Purkinje neurons.

D, Confocal image of Kv3.3 and MAP6 staining signals in Kv3.3^−/− Purkinje neurons. While Kv3.3 (black), MAP6 (magenta), and Hoechst (yellow) signals are colored in merged images in A-D (left), signals are inverted in gray-scale images.

E, Kv3.3 somatodendritic levels in Purkinje neurons in WT (left), heterozygous MAP6 KO (MAP6^+/−; middle), and homozygous MAP6 KO (MAP6^−/−; right) mice. The background in white, Kv3.3 staining signals in black, and Hoechst staining signals in cyan.

Scale bars: 250 μm in A, B, and E; 40 μm in C and D. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Our recent studies showed that MAP6^−/− mice, but not Kv3.1^−/− mice, exhibited significant deficits in recognition memory, suggesting that MAP6 may influence the function of proteins beyond Kv3.1 (Ma et al., 2023). Although Kv3.1 and Kv3.3 belong to the same Kv channel subfamily, they display distinct expression patterns in the CNS. In the hippocampus, for instance, Kv3.3 is expressed in the mossy fibers of the dentate gyrus, whereas Kv3.1 is not, although both are present in PV+ GABAergic interneurons (Gu and Barry, 2011; Kaczmarek and Zhang, 2017; Rudy and McBain, 2001). To further evaluate the impact of MAP6 reduction on overall Kv3.3 expression in the brain, we performed Western blot analysis. This revealed a significant decrease of Kv3.3 protein levels in whole brain lysates from MAP6^+/− mice, and a complete absence in Kv3.3^−/− samples (Figs. 2 and S1). Taken together, these results suggest that MAP6 may play an important role in regulating Kv3.3 protein subcellular localization and overall abundance in the brain.

Fig. 2.

Reduced expression of Kv3.3 in hippocampal mossy fibers in MAP6 KO mice.

A, Western blotting of Kv3.3 expression (top) in the brain lysates from WT, MAP6^+/− , and Kv3.3^−/− mice. Western blotting of β-tubulin (bottom) is the loading control. Molecular weights are on the right in kDa.

B, Summary of the Western blotting results. One-Way ANOVA followed by Dunnetťs test: *, p < 0.05; ***, p < 0.001. n = 4.

3.2. Direct binding between MAP6 and Kv3.3

Our recent studies demonstrated that MAP6 directly binds Kv3.1 via its first two Mn modules, which interact with the N-terminal T1 domain of Kv3.1 (Ma et al., 2023). The T1 domains are highly conserved across the Kv3 subfamily (Kv3.1-Kv3.4) and are responsible for channel tetramerization (Gu and Barry, 2011; Kaczmarek and Zhang, 2017). This raised the possibility that MAP6 might also bind directly to Kv3.3. Supporting this hypothesis, tissue immunoprecipitation using an anti-MAP6 antibody pulled down both Kv3.1 and Kv3.3 (Figs. 3A and S2A). To map the binding site, we used four 6xHis-tagged MAP6 truncation constructs (His-MAP61 to His-MAP64) containing different domains of the protein (Fig. 3B). Three GST-fusion proteins, including GST-Kv3.1 T1 (positive control), GST-Kv3.3 T1, and GST alone (negative control), were expressed in bacteria, purified using glutathione beads, and used in pull-down assays with bacterial lysates expressing the His-tagged MAP6 fragments. Both His-MAP61 and His-MAP62, which include the 1st and 2nd Mn modules, were pulled down by the T1 domains of Kv3.1 and Kv3.3. In contrast, His-MAP63 (containing the Mc modules) and His-MAP64 (containing the 3rd Mn module) showed no interaction (Figs. 3C,D and S2B–D). These results indicate that Kv3.3, like Kv3.1, directly binds the first two Mn modules of MAP6. By contrast, our previous studies showed that Cav2 channels interact with Tctex1, which binds the 3rd Mn module of MAP6 (Brocard et al., 2017). Taken together, these findings suggest that MAP6 engages distinct ion channels through different Mn modules—Mn1/Mn2 for Kv3 channels and Mn3 for Cav2 channels.

Fig. 3.

Kv3.3 channel N-terminal T1 domain directly binds MAP6.

A, Tissue IP experiment. The soluble fraction (Input SP) and crude membrane (CM) of WT mouse brains were initially separated. The supernatants (Input CM) of solubilized CMs were pulled down by an anti-Kv3.1b antibody, an anti-Kv3.3 antibody, mouse IgG (mIgG as the negative control), or an anti-MAP6 antibody (10% loaded as the positive control), resolved with SDS-PAGE, and blotted with the anti-MAP6 antibody.

B, Structural diagrams of the full-length MAP6N and its truncations. Mn modules: red bars. Mc modules: the gray bar. The C-terminal domain: the open bar. The residue positions of the N- and C-termini of each construct are provided in numbers.

C, Purified GST-fused Kv3.1 T1 domain (GST-Kv3.1 T1) or Kv3.3 T1 domain (GST-Kv3.3 T1), but not GST, pulled down bacterially-expressed His-MAP62 and His-Kv3.1 T1 but not His-MAP64. Inputs of His-MAP62, His-MAP64, and His-Kv3.1 T1 are shown on the top panel. The pulldowns were resolved in SDS-PAGE and Western blotting with an anti-6XHis antibody (middle). Inputs of GST fusion proteins are shown with Colloidal blue staining (bottom).

D, Purified GST-fused Kv3.1 T1 domain (GST-Kv3.1 T1) and Kv3.3 T1 domain (GST-Kv3.3 T1) pulled down His-MAP61 but not His-MAP63 or His-MAP64. Numbers on the right: molecular weights in kDa.

E, Structural diagram of MAP6 with different binding sites for Kv3 channel T1 domains (Kv3 T1) and Tctex1/Cav2 channels. The binding results between MAP6 and Kv3.1 are used as a positive control here and were previously published (Ma et al., 2023). The binding site between MAP6 and Tctex1/Cav2 was previously identified (Brocard et al., 2017). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.3. Molecular insights into the interaction between Kv3 T1 domains and MAP6 Mn modules

The full structure of the tetrameric Kv3.1a channel complex was recently resolved using Cryo-EM (Chi et al., 2022) (Fig. 4A left). Each Kv3.1 subunit features an N-terminal T1 domain that forms a tetrameric assembly with a central Zn²⁺-binding site located at each interface (Fig. 4A, middle and right). Opposite to the Zn²⁺-binding site, five residues (W36, P85, E89, F93, and E98 in Kv3.1) located on the outer surface of T1 tetramer complex are highlighted in red (Fig. 4A, middle and right). Sequence alignment reveals a high degree of homology between the T1 domains of Kv3.1 and Kv3.3 (Fig. 4B, top). Among the 5 residues mentioned above, only W36 differs in Kv3.3, where a glycine (G) replaces the tryptophan (W). Additionally, the D129N mutation in human Kv3.3, associated with SCA13 (Duarri et al., 2015), corresponds to D128 in mouse Kv3.3 (Fig. 4B). Notably, the amino acid sequences of the 1st and 2nd Mn modules in MAP6 are identical in mice, rats, and humans (Fig. 4B, bottom). To model the interaction between Kv3 T1 domains (aa 2–124 for Kv3.1; aa 84–208 for Kv3.3) and MAP6 (aa 115–221; the 1st and 2nd Mn modules contain residues 115–174), we performed AlphaFold 3 simulations. The predicted complexes for Kv3.1 T1-MAP6(115–221) and Kv3.3 T1-MAP6(115–221) revealed subtle differences, but in both cases, only the Mn1 and Mn2 modules (blue) directly contacted the T1 domain surface (Fig. 4C,D). The adjacent MAP6 region (aa 175–221; shown in black) did not participate in binding, consistent with our biochemical pull-down data (Fig. 3)(Ma et al., 2023). Interestingly, the 5 residues located on the external surface of the T1 tetramers of Kv3.1 or Kv3.3 are positioned to face the Mn modules of MAP6 in the predicted complexes (Fig. 4). Notably, the disease-linked residue D128(Kv3.3) lies at the Kv3.3 T1-MAP6 Mns interface (Fig. 4C, D, F), raising the possibility that the SCA13-associated D128N mutation may impair MAP6 binding, potentially contributing to disease mechanisms. Both structural models consistently show that the MAP6 Mn modules bind to the external surface of T1 tetramer, rather than the Zn²⁺-binding interfaces (Fig. 4C–F). While previous results suggest MAP6 might interact with the T1 domain interface, given its ability to bind both monomeric and tetrameric Kv3.1 T1 domains (Ma et al., 2023), our new simulation results do not support this model. In conclusion, our results suggest that the first two Mn modules of MAP6 directly bind the outer surface of the T1 tetramer of the Kv3.1 or Kv3.3 channels, providing new structural insights into their interaction.

Fig. 4.

Alpha-Fold simulations predict contact surfaces between Kv3 T1 tetramers and the first two Mn modules in MAP6.

A, The side view of Kv3.1 tetramer channel complex with one subunit in green (left), the side view of a single Kv3.1 T1 domain in green with a Zn²⁺ in purple (middle), and the top view of a single Kv3.1 T1 domain (right). Five residues on the external surface are indicated. The tetramer interface on the T1 domain is indicated in gray dashed curves with a Zn²⁺ in purple.

B, Amino acid sequence alignment of Kv3.1 and Kv3.3 T1 domains (top) and of the first two Mn modules of MAP6 from mouse, human, and rat (bottom). Five residues on the external surface, red arrowheads. The red arrow indicates the residue D128 in Kv3.3, which was mutated to N in some SCA13 patients. The positions of residues at both N- and C-termini are provided by the adjacent numbers. The net charges of gray-line-indicated sequences are provided below.

C, Side views of the binding complexes of Kv3.1 T1/MAP6 115–221 (top) and of Kv3.3/MAP6 115–221 (bottom) predicted by alpha-fold simulation.

D, Top views of the binding complexes of Kv3.1 T1/MAP6 115–221 (left) and of Kv3.3/MAP6 115–221 (right).

E, The binding interface between Kv3.1 T1 (green; the 5 residues in red) and MAP6 two Mns (blue).

F, The binding interface between Kv3.3 T1 (green; the 5 residues and D128 in red) and MAP6 two Mns (blue). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.4. Overexpression of Mn-GFP reduces the Kv3.3 axonal level in cultured hippocampal neurons

To investigate whether disrupting the interaction between Kv3.3 and MAP6 affects Kv3.3 localization in neurons, we used primary cultures of rodent hippocampal neurons, as we previously described (Gu et al., 2017; Ma et al., 2022; Ma et al., 2023). Cultured neurons at 5–7 days in vitro (DIV) were co-transfected with Kv3.3 and MAP6N-GFP cDNAs, and fixed 2 days later for immunostaining with an anti-Kv3.3 antibody. Kv3.3 channels colocalized with MAP6N-GFP in both somatodendritic compartments and axons (Fig. 5A). When Kv3.3 was co-expressed with GFP alone, it localized to the soma, dendrites, and axons (Fig. 5B). In contrast, when Kv3.3 was co-expressed with Mn-GFP (MAP6 aa115–221, encompassing the first two Mn modules), it was restricted to the soma and dendrites with a marked reduction in axons (Fig. 5C). Quantification confirmed that Mn-GFP acted as a dominant-negative mutant and significantly decreased the axonal levels of Kv3.3 compared to GFP controls (Fig. 5B–E), consistent with its competitive interference in the interaction between Kv3.3 channels and endogenous MAP6 proteins.

Fig. 5.

The axonal level of Kv3.3 in cultured hippocampal neurons is reduced by the over-expression of MAP6 Mn1/2 modules (Mn-GFP).

The hippocampal neurons were cultured from the P0 pups of WT mice, transfected with cDNA plasmids at 5–7 DIV, and fixed for immunostaining at 7–10 DIV.

A, Expressed Kv3.3 (black) and MAP6N-GFP (cyan) colocalized in the soma and axons of transfected neurons.

B, Expressed Kv3.3 (black) and GFP (cyan) colocalized in the soma and axons of transfected neurons.

C, In the presence of Mn-GFP, the axonal level of Kv3.3 was markedly reduced. Red arrows: axons of transfected neurons.

D-E, Summary of axonal (F_axon/F_soma) and dendritic (F_dendrite/F_soma) levels of Kv3.3 in the presence of either GFP or Mn-GFP. Unpaired t-test: **, p < 0.01.

F, FRET signals between YFP-Kv3.3 and CFP-MAP6Mn indicate the direct interaction of the two fusion proteins (n = 16, all neurons had significant FRET signals over the background).

G, Fluorescence intensity profiles along the line indicated in the lower-right panel in (F).

Scale bars: 100 μm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

To further confirm the binding between Kv3.3 channel and MAP6 Mn modules, we performed fluorescence resonance energy transfer (FRET) imaging, as we previously described (Barry et al., 2013; Gu et al., 2001). We co-transfected YFP-Kv3.3 and CFP-MAP6Mn into cultured hippocampal neurons. Two days after transfection, the expression of both fusion proteins was detected. YFP-Kv3.3 and CFP-MAP6Mn colocalized, and FRET signals were detected in a subset of transfected neurons (Fig. 5F,G). These results support a direct interaction between Kv3.3 channels and the Mn modules of MAP6 in a cellular context.

3.5. MAP6 deletion, but not Mn-GFP overexpression, disrupts Cav2.1 dendritic targeting in cerebellar Purkinje neurons

To investigate whether disruption of MAP6 alters Cav2.1 targeting in Purkinje neurons, we employed AAV-mediated gene delivery targeting the mouse cerebellum (Fig. 6A), a strategy we previously used successfully in the hippocampus and amygdala (Ma et al., 2023). Mice were injected with either AAV-Control-GFP, AAV-MAP6-shRNA, or AAV-Mn-GFP into the cerebellum. One month post-injection, mouse brains were perfused, and cerebellums were fixed for immunostaining. In the cerebellum partially infected with AAV-MAP6-shRNA, MAP6 expression was significantly reduced in GFP-positive (infected) Purkinje neurons, while neighboring uninfected (GFP-negative) Purkinje neurons retained strong MAP6 signals in their dendrites (Fig. 6B). These results confirm efficient MAP6 knockdown by the AAV-shRNA construct, consistent with our previous findings (Ma et al., 2023). We next assessed the expression of the Cav2.1 channel in Purkinje neurons. Cav2.1 channels were mainly concentrated in the soma and dendrites of uninfected Purkinje neurons (Figs. 6B and S3). In contrast, GFP-positive neurons infected with AAV-MAP6-shRNA displayed a significant reduction in Cav2.1 levels in both compartments (Figs. 6B–D and S3), indicating impaired Cav2.1 targeting upon MAP6 knockdown. To determine whether overexpression of Mn-GFP disrupts Cav2.1 localization in Purkinje neurons, we performed immunostaining and imaging on the cerebellum partially infected with AAV-Mn-GFP. Despite strong Mn-GFP expression in a subset of Purkinje neurons, Cav2.1 levels in soma and dendrites of GFP-positive (infected) neurons remained unchanged (Fig. 6E–G). Cav2.1 levels in GFP-positive (infected) and GFP-negative (uninfected) Purkinje neurons were similar. These results demonstrate that deletion of MAP6, but not overexpression of its first two Mn modules, disrupts Cav2.1 somatodendritic targeting in Purkinje neurons (Fig. 6G). This finding aligns with previous studies showing that MAP6 regulates the trafficking and targeting of Cav2 channel proteins in neurons through their indirect association via MAP6’s 3rd Mn module and Tctex1 (Brocard et al., 2017).

Fig. 6.

MAP6 deletion but not Mn-GFP overexpression markedly reduces the dendritic level of Cav2.1 channels in Purkinje neurons.

A, Sagittal diagram of AAV (AAV-control-GFP, AAV-MAP6-shRNA, or AAV-Mn-GFP) microinjection into the cerebellum of WT B6 mice. Approximately 3 weeks after the injection, the mice were perfused, and their brains were fixed and sectioned for immunostaining studies.

B, Low magnification image of mouse cerebellar cortex with partial AAV infection. Expression levels of Cav2.1 (black) and MAP6 (magenta) in dendrites and soma were markedly reduced in the Purkinje neurons infected with AAV-MAP6-shRNA (GFP in cyan). Hoechst staining signals: yellow.

C, Confocal image of Cav2.1 staining signals (black) in the dendrites and soma of Purkinje neurons infected with AAV-control-GFP (GFP in cyan).

D, Confocal image of reduced Cav2.1 staining signals (black) in the Purkinje neurons infected with AAV-MAP6-shRNA.

E, Expression levels of Cav2.1 (black) and MAP6 (magenta) in dendrites and soma remain unchanged in the Purkinje neurons infected with AAV-MAP6-shRNA (GFP in cyan).

F, Confocal image of Cav2.1 (black) and MAP6 (magenta) staining signals in the dendrites and soma of Purkinje neurons infected with AAV-Mn-GFP (GFP in cyan).

G, Summary of Cav2.1 level changes in cerebellar Purkinje neurons in the presence and absence of three different AAVs. Cav2.1 staining intensities were normalized with that of uninfected Purkinje neurons for each AAV. Two-way ANOVA [Effect of AAV treatment, F (2, 262) = 6.682, p = 0.001; Effect of infection status, F (1, 262) = 31.246, p < 0.001; Effect of interaction, F (2, 262) = 6.682; p = 0.001], followed by Holm-Sidak multiple comparisons testing: *, p < 0.05; ***, p < 0.001.

Scale bars: 500 μm in B, E; 50 μm in C, D, F. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.6. Both MAP6 deletion and Mn-GFP overexpression disrupt Kv3.3 axon-dendrite targeting in Purkinje neurons

To determine how AAV-mediated disruption of MAP6 affects Kv3.3 channel targeting in Purkinje neurons, we performed immunostaining for endogenous Kv3.3 in partially infected mouse cerebellar tissues, as described in the prior section. In the cerebellum partially infected with AAV-Control-GFP, Kv3.3 staining intensities in Purkinje cell soma and dendrites were comparable to those in neighboring uninfected neurons. Confocal imaging confirmed that these GFP+ cells are Calbindin+ Purkinje neurons with strong Kv3.3 expression (Figs. S4A, S5A, and 7A). Calbindin is a critical calcium-binding protein heavily expressed in cerebellar Purkinje neurons, essential for buffering rapid calcium influxes and regulating synaptic plasticity. In the cerebellum partially infected with AAV-MAP6-shRNA, Kv3.3 levels were significantly reduced in the soma and dendrites of infected (GFP+/Calbindin+) Purkinje neurons, while uninfected (GFP−) neurons retained normal Kv3.3 expression (Figs. S4B, S5B, 7B, 7D and 7E). This result is consistent with the reduction of Kv3.3 observed in Purkinje neurons of MAP6^+/− and MAP6^−/− mice (Fig. 1). Similarly, in the cerebellum partially infected with AAV-Mn-GFP, Kv3.3 expression in the soma and dendrites of infected (GFP+/Calbindin+) Purkinje neurons was also significantly decreased, while neighboring uninfected (GFP−) neurons remained unaffected (Figs. S4A, S5A, 7C, 7D and 7E). These findings indicate that overexpression of Mn-GFP is sufficient to disrupt Kv3.3 dendritic targeting, likely through the dominant-negative inhibition by a competitive interaction with endogenous MAP6.

Fig. 7.

Both dendritic and axonal levels of Kv3.3 channel proteins in Purkinje neurons are reduced by either MAP6 deletion or Mn-GFP overexpression.

A, Confocal image of WT mouse cerebellar cortex infected with AAV-Control-GFP and stained for Kv3.3, calbindin, and Hoechst.

B, Confocal image of WT mouse cerebellar cortex infected with AAV-MAP6-shRNA and stained for Kv3.3, calbindin, and Hoechst.

C, Confocal image of WT mouse cerebellar cortex infected with AAV-Mn-GFP and stained for Kv3.3, calbindin, and Hoechst.

While Kv3.3 (black), GFP (cyan), Calbindin (magenta), and Hoechst (yellow) signals are colored in merged images in A-C (left), signals are inverted in gray-scale images.

D, Summary of the changes of Kv3.3 staining signals in the dendrites of Purkinje neurons. Uninfected neurons (closed bars) and infected neurons (open bars). Two-way ANOVA [Effect of AAV treatment, F(2,66) = 14.72, p < 0.001; Effect of infection status, F (1,66) = 73.89, p < 0.001; Effect of interaction, F(2,66) = 14.72, p < 0.001], followed by Holm-Sidak multiple comparisons testing; ** p < 0.01, *** p < 0.001.

E, Summary of the changes of Kv3.3 staining signals in the soma of Purkinje neurons. Uninfected neurons (closed bars) and infected neurons (open bars). Two-way ANOVA [Effect of AAV treatment, F (2,70) = 2.28, p = 0.11; Effect of infection status, F (1,70) = 25.07, p> < 0.001; Effect of interaction, F (2,70) = 2.12, p = 0.13], followed by Holm-Sidak multiple comparisons testing; * p < 0.05, *** p < 0.001.

F, The Kv3.3 expression levels in Purkinje neuron axons of cerebellar cortex infected with AAV-Control-GFP (left), AAV-MAP6-shRNA (middle), and AAV-Mn-GFP (right). In merged images (left), Kv3.3 (black), GFP (cyan), and Hoechst (yellow) signals are colored.

G, Summary of the changes of Kv3.3 staining signals in the axons of Purkinje neurons. Uninfected neurons (closed bars) and infected neurons (open bars). Two-way ANOVA [Effect of AAV treatment, F (2,78) = 4.45, p = 0.015; Effect of infection status, F (1,78) = 6.59, p = 0.012; Effect of interaction, F (2,78) = 3.41, p = 0.038], followed by Holm-Sidak multiple comparisons testing; *, p < 0.05, ** p < 0.01, *** p < 0.001.

Scale bars: 50 μm in A-C; 15 μm in F. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Kv3.3 channels were also observed in the axons of Purkinje neurons. In the Purkinje neurons infected with AAV-Control-GFP, Kv3.3 staining signals were detected in GFP+ axons. In contrast, Kv3.3 staining signals were significantly reduced in GFP+ axons from Purkinje neurons infected with either AAV-MAP6-shRNA or AAV-Mn-GFP (Fig. S5D, 7F and 7G). Taken together, these results show that either MAP6 deletion or Mn-GFP overexpression impairs Kv3.3 localization to both dendrites and axons of Purkinje neurons. This is consistent with our biochemical findings that Kv3.3 directly binds the 1st and 2nd Mn modules of MAP6 (Figs. 3 and 4). In contrast, Cav2.1 trafficking depends on the 3rd Mn module of MAP6 through an indirect interaction mediated by Tctex1 (Brocard et al., 2017), as evidenced by the selective effects of Mn-GFP overexpression on Kv3.3 and Cav2.1 localizations (Figs. 6 and 7).

3.7. MAP6 disruption impairs burst firing of cerebellar Purkinje neurons

To determine how MAP6 knockdown and Mn-GFP expression affect the intrinsic excitability of cerebellar Purkinje neurons, we performed whole-cell recordings in cerebellar slices. We injected AAV-Control-GFP, AAV-MAP6-shRNA, or AAV-Mn-GFP vectors into the cerebellum, and then three weeks later, cut sections for whole-cell recordings of GFP+ Purkinje neurons in current-clamp mode. We biased the resting membrane potential for each neuron to −70 mV, and then injected constant step depolarizing current pulses to determine the rheobase threshold for action potential initiation. At twice rheobase, we observed repetitive bursts of complex spikes in most Purkinje neurons infected with AAV-Control-GFP (Fig. 8A). However, in mice injected with AAV-MAP6-shRNA or AAV-Mn-GFP, fewer infected Purkinje neurons generated bursts, which were altered relative to those observed in control neurons (Fig. 8B). In particular, we found that the membrane potential following bursts was depolarized relative to controls (Fig. 8C). To quantify this across the population of neurons, we measured the membrane potential relative to baseline (−70 mV) following bursts that occurred during the final third of the current step, during which bursting was relatively stable (Fig. 8C). This revealed that in neurons infected with AAV-MAP6-shRNA or AAV-Mn-GFP, the membrane potential following bursts was significantly more hyperpolarized than in controls (Fig. 8D). These data are consistent with previous work showing that Kv3 channels play an important role in repolarizing both Na⁺ and Ca²⁺ channels necessary to generate repetitive bursting (Llinas et al., 1989; McKay and Turner, 2004; Schmolesky et al., 2002). Concomitant with impaired repolarization, we found that the number of bursts generated per current pulse was also reduced in these neurons (Fig. 8E). However, the number of individual spikes within each burst was only significantly increased in neurons infected with the AAV-MAP6-shRNA vector (Fig. 8F). In summary, these data show that disrupting Kv3.3 channel expression via MAP6 knockdown or Mn overexpression alters the physiology of Purkinje neurons and may impair how they transmit information via spike bursts.

Fig. 8.

MAP6 deletion or Mn-GFP overexpression impairs membrane potential repolarization to generate burst firing.

A, Representative recordings from two Purkinje neurons infected with the AAV-GFP control construct. The depolarizing current pulses are twice the threshold amplitude necessary to elicit action potentials. (Bottom), Proportion of neurons that generated bursts during the current pulse.

B, Representative recordings from Purkinje neurons infected with AAV-Mn overexpression or AAV-MAP6-shRNA knockdown vectors. The depolarizing current pulses are twice the threshold amplitude necessary to elicit action potentials. (Bottom), Proportion of neurons that generated repetitive bursts during the current pulse.

C, In control and MAP6 knockdown recordings, the membrane potential following bursts is less hyperpolarization relative to baseline (−70 mV) compared to controls (red arrows).

D, Population quantification of membrane potential hyperpolarization following spike bursts. One-way ANOVA: F(2,24) = 8.068, p = 0.0021; Tukey’s post hoc test.

E, Population quantification of the number of bursts per current pulse at twice the action potential threshold. One-way ANOVA: F(2,24) = 7.528, p = 0.0029; Tukey’s post hoc test.

F, Population quantification of the number of spikes per burst. One-way ANOVA: F(2,24) = 4.092, p = 0.0296 Tukey’s post hoc. For (D–F), Control n = 14, Mn-GFP+ n = 9, and shRNA MAP6 n = 4. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

4. Discussion

In the present study, we show that MAP6 directly binds Kv3.3 to regulate the channel targeting in dendrites and axons of CNS neurons, particularly the cerebellar Purkinje neurons. In MAP6 KO mice, Kv3.3 levels were significantly reduced in Purkinje neuron dendrites and soma. The results of protein biochemistry assays and AlphaFold simulation revealed that the Kv3.3-MAP6 binding is mediated by Kv3.3 N-terminal T1 tetramer and MAP6’s 1st and 2nd Mn modules. In contrast, the Cav2-MAP6 binding is mediated by different sites. Furthermore, we show that AAV-mediated MAP6 knockdown significantly reduced the somatodendritic level of Kv3.3 and Cav2.1 in Purkinje neurons. Overexpression of Mn-GFP (containing the 1st and 2nd Mn modules) selectively reduced the level of Kv3.3, but not Cav2.1, in Purkinje neurons. Consistent with our immunostaining results, Purkinje neuron burst firing was reduced in both conditions—MAP6 knockdown and Mn-GFP overexpression. Thus, our findings provide new insights into Purkinje neuron targeting of two important ion channels whose genes are linked to SCAs.

Our recent studies showed that despite shared phenotypes including hyperactivity and avoidance reduction (Ma et al., 2023), MAP6^−/− but not Kv3.1^−/− mice display significant differences in motor coordination and balance, and spatial learning. This discrepancy suggests that MAP6 may influence motor functions and spatial memory through other types of ion channels, besides Kv3.1 (Ma et al., 2023). In this study, we have provided multiple lines of evidence showing MAP6 knockdown reduces the overall level of Kv3.3 in the brain, including its dendritic and axonal levels in cerebellar Purkinje neurons (Figs. 1, 2, and 7). Thus, MAP6 deletion reduces the axon-dendrite targeting of both Kv3.1 and Kv3.3 channels, likely leading to additional deficits compared to Kv3.1 deletion. This notion is supported by the finding that Kv3.1 and Kv3.3 double KO mice display much more pronounced deficits than any of the single KO mice, while displaying no changes in gross brain anatomy (Espinosa et al., 2001). The deficits observed in Kv3.1^−/−;Kv3.3^−/− double KO mice but not in single KO mice include severe ataxia, myoclonus, and severe motor-skill deficits (Espinosa et al., 2001). Furthermore, it has been reported that MAP6 can bind other proteins (Cuveillier et al., 2021). For instance, the Mn3 module of MAP6 indirectly binds Cav2 channels via Tctex1 (Brocard et al., 2017). MAP6 may regulate calcium signaling via influencing the sorting and trafficking of Cav2 channel proteins (Brocard et al., 2017). Indeed, our findings show that Cav2.1 expression level in the dendrites of cerebellar Purkinje neurons is significantly reduced (Fig. 6). Together, these alterations in ion channels may lead to multiple behavioral impairments observed in MAP6^−/− mice. These results are consistent with the observation that MAP6 knockout mice display significant motor deficits.

Our results show that the Kv3.3 axonal but not dendritic level was significantly reduced in cultured hippocampal neurons in the presence of Mn-GFP compared to GFP (Fig. 5B–E). This in vitro result of Kv3.3 reduction in axons is consistent with the result that MAP6 deletion significantly reduced the Kv3.3 level in Purkinje neuron axons (Figs. 7 and S5). However, the in vitro result of Kv3.3 dendritic level appears different from the in vivo results on cerebellar Purkinje neurons, where Mn-GFP expression reduced the Kv3.3 expression in the dendrites of Purkinje neurons (Figs. 7 and S5). The difference may result from the following potential reasons: (1) different types of neurons (e.g. hippocampal vs cerebellar neurons) may differ in additional factors involved in regulation of Kv3.3 dendritic targeting, (2) neurons cultured in vitro may differ from the ones in vivo, and (3) the cultured hippocampal neurons were not completely mature yet at the time of fixation and staining (around 7–10 DIV), compared to the brain sections from young adult mice (around 2–4 months old).

Moreover, in cerebellar Purkinje neurons, MAP6 knockdown or Mn-GFP expression not only reduced Kv3.3 dendritic and axonal levels, but also its level in the cell bodies (Fig. 6). While MAP6 knockdown may affect the protein synthesis of Kv3.3, this is likely the consequence of the disruption of MAP6-mediated targeting and/or retention of Kv3.3 protein. This is similar to the effects of MAP6 knockdown on Kv3.1 expression in PV+ GABAergic interneurons (Ma et al., 2023). Our results have shown that disrupting the Kv3.3-MAP6 interaction reduces the axonal levels of Kv3.3, as well as its dendritic levels in Purkinje neurons (Figs. 7 and S5). This causes reduced membrane potential hyperpolarization following complex spike bursts, with concomitant impairment of repetitive burst firing (Fig. 8). It remains to be determined whether MAP6 regulates the axon-dendrite targeting of Kv3.3 in other neurons in future studies. Furthermore, our finding that MAP6 is required for proper targeting of Kv3.3 to both axons and dendrites does not imply an identical mechanism underlying both axonal and dendritic targeting, since additional players involved in the two mechanisms may differ. It is also important to note that Purkinje neurons infected by AAVs at moderate and low levels express a normal level of Calbindin. However, when being highly infected by AAV, especially by AAV-MAP6-shRNA, reflected by high levels of GFP fluorescence intensity, Purkinje neurons can exhibit a decreased level of Calbindin expression. Finally, the present study mainly focuses on Kv3.3 channel and examines Cav2.1 as a comparison. The MAP6-Cav2 interaction was previously characterized extensively (Brocard et al., 2017). Whether and how MAP6 downregulation affects the axonal levels of Cav2.1 in CNS neurons should be an interesting topic for future investigation.

To gain a better insight into the interaction between MAP6’s first two Mn modules and Kv3 T1 domains, we used the AlphaFold 3 program to simulate the potential structures of Kv3.1 T1-MAP6 Mn1/2 and Kv3.3 T1-MAP6 Mn1/2 complexes. Our previous results suggest that MAP6 may bind the T1 tetramer interface, since MAP6 appeared to bind both monomers and tetramers of Kv3.1 T1 domains efficiently, in the presence and absence of EDTA to chelate Zn²⁺ ions that are required for T1 tetramerization (Ma et al., 2023). Our new simulation results do not support this possibility. Our results show that the first two Mn modules of MAP6 directly bind the external surface of the T1 tetramer of the Kv3.1 or Kv3.3 channel (Fig. 4). The Kv3 T1 and MAP6 Mn binding is likely mediated by the electrostatic interaction between negatively-charged residues on the external surface of the T1 tetramer and positively-charged residues in the two Mn modules of MAP6 (Fig. 4). Interestingly, the point mutation D128N in Kv3.3 T1 domain was linked to SCA13 with both motor deficits and intellectual disability (Duarri et al., 2015). Based on our simulation results, this residue D128 is on the external surface of Kv3.3 T1 domain and potentially resides within the interface between Kv3.3 T1 and MAP6 Mn1/2 (Fig. 4). The previous study suggests that Kv3.3 D128N mutation may not only alter channel biophysical properties but also increase the intracellular pool of the channel protein (Duarri et al., 2015), which is perhaps related to its disrupted interaction with MAP6. Although the predicted structures of these protein complexes may differ from the structures in vivo, these simulations can provide important guidelines for future investigations. For instance, since our previous studies showed that Kv3.1 T1 domain also binds the tail domain of KIF5B (kinesin 1) (Xu et al., 2010), it remains to be determined how KIF5B may bind Kv3 T1 domains in future studies.

Finally, missense mutations in Kv3.3 gene and CAG expansion in Cav2.1 gene have been linked to SCA13 and SCA6, respectively. Kv3.3 (KCNC3) and Cav2.1 (CACNA1A) genes are located in Chromosome 19 in humans, but in different chromosomes in mice, Chromosome 7 and 8, respectively. The MAP6 gene is located on Chromosome 11 in humans and Chromosome 7 in mice. Although mutations in MAP6 have not yet been linked to SCA or any other diseases in humans, dysregulation of microtubules has been implicated in SCA. For instance, Tau tubulin kinase 2 is encoded by the TTBK2 gene and functions as a serine-threonine kinase that putatively phosphorylates tau and tubulin proteins. Mutations in TTBK2 gene have been linked to SCA11 (Edener et al., 2009; Houlden et al., 2007). Nonetheless, our findings provide new insights into Purkinje neuron targeting of two important ion channels whose genes are linked to SCAs.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.nbd.2026.107429.

Data availability

Data will be made available on request.