In Vivo Expression of an SCA27A-Linked FGF14 Mutation Results in Haploinsufficiency and Impaired Firing of Cerebellar Purkinje Neurons
- 1Department of Medicine, Cardiovascular Division, Washington University School of Medicine, St. Louis, Missouri 63110
- 2Department of Biology, Miami University, Oxford, Ohio 45056
- 3Department of Developmental Biology, Washington University School of Medicine, St. Louis, Missouri 63110
-
Author contributions: J.L.R., S.P.B., M.X., D.M.O., and J.M.N. designed research; J.L.R., S.P.B., and M.X. performed research; D.M.O. and J.M.N. contributed unpublished reagents/analytic tools; J.L.R., S.P.B., M.X., D.M.O., and J.M.N. analyzed data; J.L.R., D.M.O., and J.M.N. wrote the paper.
Abstract
Autosomal dominant mutations in FGF14, which encodes intracellular fibroblast growth factor 14 (iFGF14), underlie spinocerebellar ataxia type 27A (SCA27A), a devastating multisystem disorder resulting in progressive deficits in motor coordination and cognitive function. Mice lacking iFGF14 exhibit similar phenotypes, which have been linked to iFGF14-mediated modulation of the voltage-gated sodium (Nav) channels that regulate high-frequency repetitive firing of cerebellar Purkinje neurons, the main output neurons of the cerebellar cortex. To investigate the in vivo mechanisms underlying SCA27A, we developed a targeted knock-in strategy to introduce the first point mutation identified in FGF14 into the mouse Fgf14 locus (Fgf14F145S). Current-clamp recordings from Purkinje neurons in acute cerebellar slices from adult male and female Fgf14F145S/+ mice revealed that high-frequency repetitive firing, which is characteristic of wild-type Purkinje neurons, was replaced by prolonged bursts of action potentials. A shift from tonic to burst firing was mimicked in wild-type Purkinje neurons by bath application of the Nav channel toxin, tetrodotoxin. Burst firing was also measured in heterozygous Fgf14 knock-out (Fgf14+/−) Purkinje neurons, suggesting that the impaired firing of Fgf14F145S/+ Purkinje neurons reflects reduced Nav channel availability, owing to the loss of the iFGF14 protein. Western blot analyses confirmed reduced iFGF14 protein expression in cerebellar lysates prepared from Fgf14F145S/+ (and Fgf14+/−) animals and voltage-clamp experiments revealed a hyperpolarizing shift in the voltage dependence of closed-state Nav channel inactivation in Fgf14F145S/+ (and Fgf14+/−) Purkinje neurons. Together, these results indicate that Fgf14 haploinsufficiency and reduced Nav channel availability underlie impaired firing in Fgf14F145S/+ Purkinje neurons.
- cerebellum
- FHF4
- fibroblast growth factor homologous factor
- iFGF14
- iFGF14F145S
- intrinsic excitability
- spinocerebellar ataxia type 27
- voltage-gated sodium (Nav) channels
Significance Statement
Autosomal dominant mutations in FGF14 underlie spinal cerebellar ataxia 27A (SCA27A), a neurological disorder associated with progressive motor and cognitive deficits. To explore the in vivo functional effects of SCA27A-linked mutations in FGF14, we developed a mouse model with targeted knock-in of the first point mutation identified in FGF14, which results in a single amino acid change (phenylalanine to serine) in the iFGF14 protein, iFGF14F145S. The experiments here revealed that spontaneous high-frequency repetitive firing, characteristic of wild-type Purkinje neurons, is impaired iFGF14F145S Purkinje neurons and that this impairment in firing properties reflects Fgf14 haploinsufficiency, reduced iFGF14 protein expression, and resulting alterations in Nav channel availability.
Introduction
Although sharing sequence homology with canonical fibroblast growth factors (FGFs), the intracellular fibroblast growth factors (iFGF11-14), also called fibroblast growth factor homologous factors (FHFs), are distinct (Smallwood et al., 1996; Goldfarb, 2005; Pablo and Pitt, 2016; Ornitz and Itoh, 2022). The primary physiological function identified for the iFGFs reflects binding to the intracellular C-terminal domain(s) of voltage-gated sodium (Nav) channel pore-forming (α) subunit(s) (Liu Cj et al., 2001; Liu et al., 2003; Wittmack et al., 2004; Lou et al., 2005; Goetz et al., 2009; Wang et al., 2011; Bosch et al., 2016) and modulation of Nav current properties (Wittmack et al., 2004; Lou et al., 2005; Goldfarb et al., 2007; Laezza et al., 2009; Yan et al., 2014; Bosch et al., 2015; Park et al., 2016; Abrams et al., 2020; Chakouri et al., 2022; Angsutararux et al., 2023). The functional effects, however, depend on the coexpressed iFGF and Nav α subunit and on the cellular environment (Di Re et al., 2017; Pablo and Pitt, 2017; Ransdell et al., 2024).
In vivo, iFGF14 localizes to the axon initial segments of hippocampal pyramidal neurons (Lou et al., 2005; Goldfarb et al., 2007; Laezza et al., 2007; Pablo and Pitt, 2017; Ransdell et al., 2024), as well as cerebellar Purkinje, granule, basket, and stellate neurons (Xiao et al., 2013; Yan et al., 2014; Bosch et al., 2015). Animals harboring targeted disruption of the Fgf14 locus (Fgf14−/−) display severe locomotor and learning deficits (Wang et al., 2002; Wozniak et al., 2007). Consistent with the locomotor phenotype (Wang et al., 2002), Fgf14−/− Purkinje neurons have severely attenuated repetitive firing (Shakkottai et al., 2009; Bosch et al., 2015), an effect also observed in adult Purkinje neurons after shRNA-mediated in vivo knockdown of Fgf14. Attenuated Fgf14−/− Purkinje neuron firing is linked to a hyperpolarizing shift in the voltage dependence of Nav current closed-state inactivation (Bosch et al., 2015). Acute shRNA-mediated knockdown of Fgf14 in isolated neonatal Purkinje neurons in vitro was also reported to increase the rate of transient Nav current inactivation and reduce resurgent Nav current (Yan et al., 2014).
Shortly after the description of the Fgf14−/− line (Wang et al., 2002), an autosomal dominant missense mutation in FGF14, resulting in a single (phenylalanine to serine) amino acid substitution in iFGF14, iFGF14F145S, was identified in a Dutch family presenting with ataxia and progressive cognitive decline (van Swieten et al., 2003). Several additional autosomal dominant mutations in FGF14, including frameshift (Dalski et al., 2005; Choquet et al., 2015), nonsense (Miura et al., 2019), deletion (Coebergh et al., 2014; Amado et al., 2017; Keller Sarmiento et al., 2025), and chromosomal translocations (Misceo et al., 2009; Shimojima et al., 2012), have been identified, and these are now collectively classified as spinocerebellar ataxia type 27A (SCA27A).
Modeling studies suggested that the FGF14F145S mutation yields an unstable protein (Olsen et al., 2003; van Swieten et al., 2003), resulting in haploinsufficiency. In vitro experiments, however, revealed that transfection of neonatal rat hippocampal neurons with a plasmid encoding a GFP-tagged human iFGF14F145S resulted in reduced Nav current amplitudes, compared with cells transfected with a plasmid encoding GFP-tagged wild-type human FGF14 (Laezza et al., 2007). In HEK-293 cells, FGF14F145S expression was shown to prevent wild-type iFGF14 from modulating Nav currents, suggesting that the mutant iFGF14F145S functions as a dominant negative (Laezza et al., 2007). As these experiments were performed on neonatal hippocampal neurons in dissociated cell culture or in heterologous cells, and involved overexpression, the relevance of this mechanism in vivo remained to be evaluated. To this end, we developed a mouse model with targeted knock-in of a single point mutation (T to C) in Fgf14, resulting in production of the iFGF14F145S mutant protein, and we examined the functional consequences of this in vivo manipulation directly. We report that high-frequency repetitive firing is impaired in Fgf14F145S/+ Purkinje neurons and that this impairment in firing properties reflects reduced iFGF14 expression and resulting alterations in Nav channel availability.
Materials and Methods
All reagents were obtained from Sigma-Aldrich unless otherwise indicated.
Experimental animals
All experiments involving animals were performed in accordance with the guidelines published in the National Institutes of Health Guide for the Care and Use of Laboratory Animals. All protocols were approved by the Washington University or Miami University Institutional Animal Care and Use Committees (IACUC). To construct an in vivo mouse model, we selected the first human mutation, FGF14F145S, identified (van Swieten et al., 2003) and mutated the codon for phenylalanine 145 (F145), as illustrated in Figure 1A. F145 is based on the FGF14A numbering system and corresponds to F150 in iFGF14B (Fig. S1). The F145 codon (TTT) is located in exon 4 and overlaps with a Dra1 restriction site (TTTAAA). The single T to C mutation results in replacement of phenylalanine (F) with serine (S) and destroys the Dra1 restriction site. The nucleotide sequence encoding F145 (F150) and the Dra1 restriction site is identical in the mouse and human sequences. The targeting vector depicted in Figure 1A along with homologous recombination methods using the SCC#10 ES cell line (ATCC SCRC-1020, RRID:CVCL_C259) were used to generate the F145 Fgf14 mutation. Clones that showed both 5′ and 3′ homologous recombination were screened for loss of the Dra1 restriction site. Wild-type DNA, amplified with primers flanking exon 4, were expected to cut with Dra1, while targeted DNA that contained the engineered mutation were expected not to cut with Dra1. Cell lines that contained the targeted mutation were then karyotyped. Cells with a normal, 40, XY karyotype that contained the F145S mutation were injected into C57BL/6 blastocysts by the Washington University Mouse Genetics Core. Resulting chimeric mice were bred and tested for germline transmission. Mice containing a targeted allele were bred to mice expressing Cre recombinase in the germline. This resulted in the excision of the LoxP-neo-LoxP cassette, leaving behind only a single loxP site (Fig. 1A).
Animals harboring a targeted disruption of the Fgf14 locus (Fgf14−/−), described in Wang et al. (2002), were generated by crossing heterozygous Fgf14+/− mice, congenic in the C57BL/6 background; genotypes were confirmed by PCR. Homozygous Fgf14F145S/F145S knock-in animals were generated by crossing heterozygous Fgf14F145S/+ (heterozygotes) in the C57BL/6 background. Genotypes were confirmed by PCR screening. Male and female mice [postnatal day (P) 12 to 8 weeks old] of the genotypes (Fgf14+/−, Fgf14−/−, Fgf14F145S/+, Fgf14F145S/F145S), along with wild-type controls, were used in the experiments completed here.
Preparation of acute brain slices
Acute (350 μM) parasagittal cerebellar slices were prepared from (5–8-week-old) animals using previously described methods (Bosch et al., 2015; Ransdell et al., 2017). Briefly, animals were anaesthetized with 1.25% Avertin and perfused transcardially with ice-cold cutting solution that contained the following (in mM): 240 sucrose, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 0.5 CaCl2, and 7 MgCl2 saturated with 95% O2/5% CO2. Following perfusion, animals were rapidly decapitated and the brain of each animal was removed and cut parasagittally in ice-cold cutting solution on a Leica VT1000S Vibratome (Leica Microsystems). Parasagittal cerebellar sections were incubated in warmed (34°C) artificial cerebrospinal fluid (ACSF) containing the following (in mM): 125 NaCl, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 2 CaCl2, 1 MgCl2, and 25 dextrose, ∼310 mOsmol/L, saturated with 95% O2/5% CO2, for 35 min and an additional 25 min–4 h at room temperature (25°C) prior to electrophysiological recordings.
Isolation of neonatal cerebellar Purkinje neurons
P12–P15 wild-type and Fgf14−/− animals were anesthetized with 1.25% Avertin, and brains were rapidly removed and placed in ice-cold isolation medium containing the following (in mM): 82 Na2SO4, 30 K2SO4, 5 MgCl2, 10 HEPES, 10 glucose, and 0.001% phenol red (at pH 7.4). Using a scalpel, the cerebellum was removed, minced into small chunks, and incubated in isolation medium containing 3 mg/ml protease XXIV at 33°C for 10–15 min. Following this incubation period, the tissue pieces were washed with enzyme-free isolation medium containing 1 mg/ml bovine serum albumin and 1 mg/ml trypsin inhibitor. After being transferred to oxygenated ACSF (22–23°C), the tissue pieces were triturated with a fire-polished glass pipette. An aliquot of the cell suspension was placed on a coverslip in the recording chamber and superfused with fresh ACSF (0.5 ml/min) saturated with 95% O2/5% CO2 for 25 min before beginning electrophysiological recordings.
Electrophysiological recordings
Whole-cell current-clamp and voltage-clamp recordings were obtained from visually identified cerebellar Purkinje neurons using differential interference contrast optics. During recordings from cells in parasagittal cerebellar slices, warmed ACSF (34 ± 1°C) saturated with 95% O2/5% CO2 was continuously perfused. Data were collected using a MultiClamp 700B patch-clamp amplifier interfaced with a Digidata 1332 and pCLAMP 10 software (Axon Instruments). In all experiments, tip potentials were zeroed prior to forming giga-ohm membrane-pipette seals. Pipette capacitances were compensated using the pCLAMP software. Signals were acquired at 50–100 kHz and filtered at 10 kHz prior to digitization and storage.
For current-clamp recordings, pipettes contained the following (in mM): 144 K-gluconate, 0.2 EGTA, 3 MgCl2,10 HEPES, 4 MgATP, 0.5 NaGTP, PH 7.3, ∼300 mOsmol/L. Spontaneous firing was recorded for 60 s and analyzed during the last
30 s. Evoked repetitive firing was measured during 0.5 s depolarizing current injections of varying amplitudes that were administered
immediately after a −0.5 nA prepulse. For voltage-clamp recordings, pipettes contained the following (in mM): 110 CsCl, 15
TEA-Cl, 5 4AP, 1 CaCl2, 2 MgCl2, 10 EGTA, 4 Na2-ATP and 10 HEPES, pH 7.25 ∼300 mOsmol/L. Recording pipettes had resistances of 2–4 MΩ. Series resistances were compensated
≥80%, and the voltage errors resulting from the uncompensated resistances were always ≤4 mV and were not corrected. To isolate
and measure Nav currents in Purkinje neurons in acute cerebellar slices, after formation of a gigaseal, the superfusing ACSF
was switched to a low Na+ ACSF containing the following (in mM): 25 NaCl, 100 TEA-Cl, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 2 CaCl2, 1 MgCl2, 0.1 CdCl2, and 25 dextrose, ∼310 mOsmol/L, saturated with 95% O2/5% CO2. Voltage-clamp protocols to isolate Nav currents were tested using a strategy outlined in Milescu et al. (2010) and optimized to minimize space-clamp errors and enable reliable measurements of Nav currents in Purkinje neurons. To determine
the voltage dependence of activation of the Nav currents, a three-step voltage-clamp protocol was used: a depolarizing prepulse
(0 mV, 5 ms), to inactivate the Nav currents in the distal neurites, was delivered first, followed by a brief (2.5 ms) hyperpolarizing
voltage step to −60 mV, to allow recovery of the Nav currents in the soma and proximal neurites, and a second depolarizing
pulse (for 8 ms) was used to measure (appropriately clamped) Nav currents at different test potentials. Peak Nav conductance
(at each potential) was calculated (using Ereversal = +43 mV) and normalized to the maximal Nav conductance. The mean ± SEM normalized Nav conductances were plotted as a function
of the test potential and fitted with a single (Eq. 1) Boltzmann function shown as follows:<mml:math display="block"><mml:mstyle displaystyle="true" scriptlevel="0"><mml:mrow><mml:mfrac><mml:mrow><mml:msub><mml:mi>G</mml:mi><mml:mrow><mml:mrow><mml:mi
mathvariant="normal">Na</mml:mi></mml:mrow></mml:mrow></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>G</mml:mi><mml:mrow><mml:mrow><mml:mi
mathvariant="normal">Na</mml:mi><mml:mo>,</mml:mo><mml:mi mathvariant="normal">max</mml:mi></mml:mrow></mml:mrow></mml:msub></mml:mrow></mml:mfrac></mml:mrow><mml:mo>=</mml:mo><mml:mstyle
displaystyle="true" scriptlevel="0"><mml:mrow><mml:mfrac><mml:mn>1</mml:mn><mml:mrow><mml:mn>1</mml:mn><mml:mo>+</mml:mo><mml:msup><mml:mi>e</mml:mi><mml:mrow><mml:mo
stretchy="false">[</mml:mo><mml:mo stretchy="false">(</mml:mo><mml:mi>V</mml:mi><mml:mi>h</mml:mi><mml:mo>−</mml:mo><mml:mi>V</mml:mi><mml:mi>m</mml:mi><mml:mo
stretchy="false">)</mml:mo><mml:mo>/</mml:mo><mml:mi>k</mml:mi><mml:mo stretchy="false">]</mml:mo></mml:mrow></mml:msup></mml:mrow></mml:mfrac></mml:mrow><mml:mo>,</mml:mo><mml:mspace
width="2em"></mml:mspace><mml:mo stretchy="false">(</mml:mo><mml:mn>1</mml:mn><mml:mo stretchy="false">)</mml:mo></mml:mstyle></mml:mstyle></mml:math>GNaGNa,max=11+e[(Vh−Vm)/k],(1)
where Vh is the membrane potential of half-maximal activation and k is the slope factor.
To determine the voltage dependence of steady-state inactivation of the Nav currents, a three step voltage-clamp protocol
was also used. Each cell was first depolarized to 0 mV for 5 ms to inactivate the Nav currents and subsequently hyperpolarized
briefly (2.5 ms) to various conditioning voltages to allow recovery of Nav channels from inactivation and, finally, depolarized
to −20 mV for 8 ms. The peak transient Nav current amplitudes evoked during this final step (to −20 mV) were measured and
normalized to the current amplitude evoked (at −20 mV) from a −120 mV conditioning voltage step (in the same cell). Mean ± SEM
normalized Nav currents were plotted as a function of the conditioning voltage. The voltage dependence of Nav current steady-state
inactivation were plotted and fit with single Boltzmann functions (Eq. 2):<mml:math display="block"><mml:mspace width="0.25em"></mml:mspace><mml:mstyle displaystyle="true" scriptlevel="0"><mml:mrow><mml:mfrac><mml:mrow><mml:msub><mml:mi>I</mml:mi><mml:mrow><mml:mrow><mml:mi
mathvariant="normal">Na</mml:mi></mml:mrow></mml:mrow></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>I</mml:mi><mml:mrow><mml:mrow><mml:mi
mathvariant="normal">Na</mml:mi><mml:mo>,</mml:mo><mml:mi mathvariant="normal">max</mml:mi></mml:mrow></mml:mrow></mml:msub></mml:mrow></mml:mfrac></mml:mrow><mml:mo>=</mml:mo><mml:mstyle
displaystyle="true" scriptlevel="0"><mml:mrow><mml:mfrac><mml:mn>1</mml:mn><mml:mrow><mml:mn>1</mml:mn><mml:mspace width=".1em"></mml:mspace><mml:mo>+</mml:mo><mml:mspace
width=".1em"></mml:mspace><mml:msup><mml:mi>e</mml:mi><mml:mrow><mml:mo stretchy="false">[</mml:mo><mml:mo stretchy="false">(</mml:mo><mml:mrow><mml:mi>V</mml:mi><mml:mi>h</mml:mi><mml:mo>−</mml:mo><mml:mi>V</mml:mi><mml:mi>m</mml:mi></mml:mrow><mml:mo
stretchy="false">)</mml:mo><mml:mo>/</mml:mo><mml:mi>k</mml:mi><mml:mo stretchy="false">]</mml:mo></mml:mrow></mml:msup></mml:mrow></mml:mfrac></mml:mrow><mml:mo>,</mml:mo><mml:mspace
width="2em"></mml:mspace><mml:mo stretchy="false">(</mml:mo><mml:mn>2</mml:mn><mml:mo stretchy="false">)</mml:mo></mml:mstyle></mml:mstyle></mml:math>INaINa,max=11+e[(Vh−Vm)/k],(2)
where Vh is the membrane potential of half-maximal inactivation and k is the slope factor.
Analysis of electrophysiological data
Data were analyzed using Clampfit (Molecular Devices), MATLAB (MathWorks), Microsoft Excel, and Prism (GraphPad Software). Input resistances (Rin) were determined from the change in membrane potential produced by a 20 pA hyperpolarizing current injection from the resting potential. Percentages of time firing were calculated using a custom MATLAB script in which 30 s of spontaneous (current-clamp) activity was analyzed and the elapsed times between action potentials (interspike interval) were categorized as “firing” if less than or equal to 50 ms. If the interspike intervals were >50 ms, the elapsed times between action potentials were characterized as “silent.” In this analysis, sodium-mediated spikes were recognized (and differentiated from calcium spikes) as having maximal rising phases of greater than or equal to 100 mV/ms. These values were based on previous experiments in our lab (data not shown) in which TTX (which blocks Nav channels) and cadmium (which blocks calcium channels) were used to determine that Purkinje neuron calcium-mediated spikes, measured after applying 1 µM TTX, have maximum rising dV/dt values that are <100 mV/ms and sodium-mediated spikes have maximum rising dV/dt values that are >100 mV/ms.
Western blot analysis
Lysates were prepared from adult (>8-week-old) animals in 20 mM HEPES + 150 mM NaCl buffer with 0.5% CHAPS and a protease inhibitor tablet (Roche) using established methods (Brunet et al., 2004; Norris and Nerbonne, 2010), fractionated on SDS-PAGE (12%) gels, transferred to polyvinylidene fluoride (PVDF) membranes (Biorad) and probed for iFGF14 expression using a rabbit polyclonal anti-iFGF14 antiserum (1:1,000; Covance Laboratories), validated as described previously (Bosch et al., 2016).
Elevated balance beam
The elevated balance beam assay was performed using previously described methods (Ransdell et al., 2017). Briefly, animals were trained for 3 consecutive days, followed by 5 consecutive days of testing. During each training day, animals were allowed to explore the Plexiglas platform and the enclosed box for 2 min on each surface. Subsequently, during training, animals were placed on the Plexiglas platform and allowed to transverse an 11-mm-wide beam into an enclosed box. After 3 d of training, animals began 5 consecutive days of testing. During testing days, animals (individually) crossed the 11-mm-wide beam (from the platform to the enclosed box) and both the time to cross and the number of hindlimb foot-slips during crossing were recorded by the experimenter. During all tests, the experimenter was blinded to animal genotypes.
Statistical analysis
Averaged and normalized data are presented as mean ± SEM. Statistical analyses were performed using Student's paired or unpaired t tests or two-way and one-way analysis of variance (ANOVA) as indicated in the text and/or the figure legends. Similarities in variances were assessed using F tests prior to paired and unpaired Student's t tests and Brown–Forsythe tests prior to one-way ANOVA tests. In all cases, p values from F tests and Brown–Forsythe tests were >0.05. Repeated-measures (RM) two-way ANOVA tests included Geisser–Greenhouse corrections.
Results
Development of a SCA27A mouse model expressing the Fgf14F145S mutation
A gene-targeting construct for knock-in expression of a single point mutation (T to C) in exon 4 of Fgf14 was developed to provide in vivo expression of the first SCA27A-linked mutant iFGF14 protein (Van Swieten et al., 2003; see Materials and Methods and Fig. 1A). Southern blot analysis (Xba1 digest) with 5′ and 3′ probes (locations labeled in Fig. 1A) allowed the identification of the mutant and wild-type alleles in two independent embryonic stem cell clones (Fig. 1B) and Sanger sequencing of the genomic PCR products confirmed the presence of the wild type (WT), the homozygous S145 mutation, and the heterozygous F145/S145 sequences (Fig. 1C). Animals (Fgf14F145S/+) heterozygous for the mutant allele were crossed to generate offspring that were heterozygous (Fgf14F145S/+) or homozygous (Fgf14F145S/F145S) for the Fgf14 mutation. Offspring (from this and subsequent crosses) were screened by PCR using the primers given in the legend to Figure 1; representative results are illustrated in Figure 1D. Heterozygous Fgf14F145S/+ animals were identified by the presence of bands corresponding to both the wild-type (280 base pairs) and the mutant (325 base pairs) alleles, and homozygous Fgf14F145S/F145S animals were identified by the presence of only the mutant (325 base pairs) allele (Fig. 1D).
Mouse model of SCA27: knock-in of a point mutation in Fgf14 to provide expression of the iFGF14F145S mutant protein. A, Schematic of the wild-type Fgf14 gene structure and the targeting strategy used to introduce the single T to C mutation to generate the (F145S) iFGF14 mutant protein. Top, Fgf14 genomic structure showing exons 1a/1b to 5. Below, the targeting construct with a neo cassette and flanking loxP sequences, the mutant allele produced by homologous recombination, and the mutant allele with neo deletion mediated by Cre. B, Southern blot analysis (Xba1 digest) with 5′ and 3′ probes (labeled in A) allowed the identification of the mutant and wild-type alleles (after neo deletion). C, Sanger sequencing chromatograms of genomic PCR products confirmed the wild-type F145 and homozygous S145 sequences. D, Representative gel illustrating the PCR genotyping results of littermates from a Fgf14F145S/+ × Fgf14F145S/+ cross with primers flanking the loxP site. Lane 1 contains only the wild-type allele (280 bp), consistent with a wild-type animal. Both the wild-type (280 bp) and mutant (325 bp) alleles are evident in lanes 2 and 4, consistent with heterozygous mutant (Fgf14F145S/+) animals, and lane 3 shows only the mutant (325 bp) allele, reflecting the homozygous mutant (Fgf14F145S/F145S; 325 bp). Human and mouse iFGF14A/iFGF114B protein sequences are shown/compared in Figure S1; the phenylalanine (F) residue in the core domain that was replaced by a serine (S) in the Fgf14F145S/+ mutant mouse line is highlighted in pink.
Sustained repetitive firing is impaired in Fgf14F145S/+ Purkinje neurons
To determine how expression of the iFGF14F145S mutant protein affects the intrinsic excitability of cerebellar Purkinje neurons, we obtained whole-cell current-clamp recordings from heterozygous (Fgf14F145S/+) mutant Purkinje neurons in acute parasagittal cerebellar slices prepared from adult (5–8-week-old) animals at 34 ± 1°C. These experiments revealed that the spontaneous, high-frequency repetitive firing, which is characteristic of wild-type Purkinje neurons in acute cerebellar slices (Häusser and Clark, 1997), was impaired in Fgf14F145S/+ Purkinje neurons. In contrast to the sustained repetitive spike activity typically measured in wild-type Purkinje neurons (Fig. 2A), Fgf14F145S/+ Purkinje neurons fired in lengthy bursts of action potentials punctuated by periods of quiescence (Fig. 2B).
Sustained high-frequency firing is impaired in Fgf14F145S/+ cerebellar Purkinje neurons and is mimicked by treatment of wild-type cells with TTX. A, B, Representative recordings of spontaneous firing obtained from adult Purkinje neurons in acute cerebellar slices prepared from (A) wild-type and (B) Fgf14F145S/+ animals are shown. The inset in panel B highlights the sodium-mediated (filled arrow) and calcium-mediated (unfilled arrow) spikes in the representative Fgf14F145S/+ Purkinje neuron. C, Representative recordings from an adult wild-type Purkinje neuron before and after the addition of a 1 nM TTX to the perfusing ACSF. The application of TTX converts the sustained repetitive firing of the wild-type neuron into biphasic periods of quiescence and firing, similar to the biphasic firing pattern observed (B) in Fgf14F145S/+ Purkinje neurons. D, Addition of 1 nM TTX markedly (p = 0.002, paired Student's t test, n = 6) reduced the percentage of time wild-type Purkinje neurons were firing action potentials (see Materials and Methods).
It seemed likely to us that the intermittent burst firing observed in Fgf14F145S/+ Purkinje neurons may reflect reduced Nav channel availability. To explore this possibility, we bath applied a concentration (1 nM) of TTX that was previously reported to attenuate, but not to eliminate, Nav currents in Purkinje neurons (Khaliq and Raman, 2006). As illustrated in Figure 2C, the characteristic spontaneous, high-frequency repetitive firing of a wild-type Purkinje neuron is altered dramatically on exposure to 1 nM TTX. Indeed, these experiments revealed that Purkinje cells in the presence of 1 nM TTX fire in prolonged bursts of action potentials, which also included fast, sodium-dependent and slow, calcium-dependent spikes (Fig. 2C), firing behavior that is remarkably similar to that observed in Fgf14F145S/+ Purkinje neurons (Fig. 2B). To quantify firing rates, we calculated the mean ± SEM percentage of time wild-type Purkinje neurons fired sodium-mediated action potentials during a 30 s whole-cell recording period, before and after exposure to 1 nM TTX. These analyses revealed markedly (p = 0.002) reduced sodium-mediated spike activity in the presence of 1 nM TTX. The loss of sustained high-frequency repetitive firing and the conversion to combined sodium-dependent and calcium-dependent spiking activity in wild-type Purkinje neurons in the presence of 1 nM TTX suggests that the intermittent burst firing phenotype reflects reduced Nav channel availability.
Effect of Fgf14F145S on Purkinje neuron repetitive firing phenocopies Fgf14 deletion
Additional experiments revealed that the spontaneous firing properties of Purkinje neurons lacking a single Fgf14 allele (Fgf14+/−) are very similar (Fig. 3B) to the spontaneous firing properties of Fgf14F145S/+ Purkinje neurons (Fig. 2B), with bursts of sodium- and calcium-mediated spikes that are interrupted by periods of quiescence. We also measured spontaneous firing in adult Purkinje neurons with homozygous knock-in of the Fgf14F145S mutant allele (Fgf14F145S/F145S) and found that most (11 of 13) of the Fgf14F145S/F145S Purkinje cells from which recordings were obtained were electrically silent (Fig. 3C), behavior similar to that described previously in Fgf14−/− Purkinje neurons (Shakkottai et al., 2009; Bosch et al., 2015), i.e., that the vast majority (>80%) of Fgf14−/− cells are electrically silent (see also Fig. 3D). Measurements of membrane capacitance and input resistance in Fgf14F145S/F145S (mean values of 261 ± 27 pF; 49.4 ± 3.0 MΩ, n = 13) and Fgf14−/− (mean values of 252 ± 16 pF; 42.5 ± 3.6 MΩ, n = 14) Purkinje neurons were also found to be similar to the values determined in wild-type Purkinje neurons (mean values of 296 ± 23 pF; 52.8 ± 2.2 MΩ, n = 18; p > 0.05, one-way ANOVA).
Repetitive firing properties of Fgf14F145S/+ and Fgf14F145S/F145S Purkinje neurons phenocopy the firing properties of Fgf14+/− and Fgf14−/− Purkinje neurons. A–C, Representative recordings of spontaneous firing from Purkinje neurons in acute cerebellar slices prepared from adult wild-type (A), Fgf14+/− (B), and Fgf14F145S/F145S (C) animals are shown. The mean ± SEM percentage of time each cell of each genotype was firing action potentials (during 30 s current-clamp recordings; see Materials and Methods), which were determined and are plotted in D. As is evident, repetitive firing is markedly attenuated in Fgf14+/− and Fgf14F145S/+ Purkinje neurons, each of which lack one copy of the wild-type Fgf14 allele, compared with wild-type Purkinje neurons. E, Representative Western blots of cerebellar lysates prepared from adult wild-type, Fgf14F145S/+, Fgf14F145S/F145S, Fgf14+/−, and Fgf14−/− animals probed with a rabbit polyclonal iFGF14 antiserum and with mouse monoclonal anti-tubulin antibody (loading control) reveal reduced iFGF14 expression in cerebellar tissues from animals with targeted knock-in of the Fgf14F145S mutant allele. Full Western blots are shown in Figure S2.
To compare the spontaneous firing properties of Fgf14F145S/+, Fgf14F145S/F145S, Fgf14+/−, and Fgf14−/− Purkinje neurons (illustrated in the representative records in Figs. 2A,B and 3A–C), we determined the proportion of time (during a 30 s current-clamp recording) that each Purkinje neuron fired repetitive, sodium-mediated action potentials. The duration between action potentials was categorized as repetitive if the interspike intervals were ≤50 ms, and spikes were determined to be sodium mediated if the rising phases had a peak dV/dt value of ≥100 mV/ms (see Materials and Methods). As the representative records suggest, the mean ± SEM percentages of time Fgf14F145S/+ (33 ± 7%) and Fgf14+/− (36 ± 11%) animals fired repetitively are similar (Fig. 3D) and intermediate to the mean ± SEM percentages of time that Fgf14F145S/F145S (4 ± 3%), Fgf14−/− (4 ± 4%), and wild-type (75 ± 5%) Purkinje neurons repetitively fired sodium-mediated action potentials (Fig. 3D).
Knock-in of the Fgf14F145S mutant allele reduces iFGF14 protein
The results of the current-clamp studies described above reveal that knock-in expression of the Fgf14F145S mutant allele has functional effects similar to the targeted deletion of Fgf14, suggesting loss of the iFGF14 protein. To explore this hypothesis directly, cerebellar tissue lysates from Fgf14F145S/F145S and Fgf14F145S/+ animals were fractionated, probed with an anti-iFGF14 antiserum, and compared directly to cerebellar lysates from wild-type animals and from animals with targeted deletion of one (Fgf14+/−) or both (Fgf14−/−) Fgf14 alleles; a representative Western blot is presented in Figure 3E. As is evident, these analyses revealed that iFGF14 expression is lower Fgf14F145S/+ and Fgf14+/− tissue lysates than in wild-type lysates (Fig. 3E). In addition, similar to Fgf14−/− tissue lysates, the iFGF14 protein is undetectable in cerebellar tissue lysates prepared from Fgf14F145S/F145S animals, indicating that the in vivo expression of the Fgf14F145S missense mutation results in loss of the iFGF14 protein.
Similar to Fgf14 deletion, knock-in of Fgf14F145S affects Nav current inactivation
To explore the molecular mechanisms underlying the repetitive firing phenotypes observed in Fgf14F145S/+ (Fig. 2) and Fgf14+/− (Fig. 3) cerebellar Purkinje neurons, we examined the voltage dependences of activation and inactivation of the fast-transient Nav currents (INaT) in these cells. To accurately measure Nav currents from mature Purkinje neurons in acute brain slices and avoid errors caused by inadequate space clamp that are typical in experiments involving cells with intact neurites, voltage-clamp experiments were performed (see Materials and Methods) with reduced extracellular Na+ and using a voltage-clamp strategy developed by Milescu et al. (2010) (see Materials and Methods). Briefly, cells were first depolarized to activate and inactivate the Nav channels in the soma and distal neurites, followed by brief hyperpolarization to allow recovery of Nav channels in/near the soma, and subsequent depolarization to various test potentials to measure Nav current amplitudes and the voltage-dependent properties of recovered Nav channels at different test potentials. Figure 4A depicts the voltage-clamp protocol (bottom) and representative Nav current records (top) used to examine the voltage dependence of Nav current steady-state inactivation. Peak current amplitudes, measured at the −20 mV test potential, were normalized to the current amplitude after a conditioning voltage step to −120 mV. The mean ± SEM normalized Nav current amplitudes are plotted as a function of the conditioning membrane potential (Fig. 4B), and Boltzmann fits of these plots (see Materials and Methods) reveal that 50% of the Nav channels are inactivated (V1/2) in Fgf14F145S/+ (V1/2 = −65 mV, n = 8) Purkinje neurons at voltages that are significantly (p ≤ 0.0001, two-way ANOVA) different from wild-type (V1/2 = −58 mV, n = 11) Purkinje neurons. Additional experiments on Fgf14F145S/F145S Purkinje neurons also revealed a marked hyperpolarizing shift in the V1/2 (equal to −71 mV) of Nav current steady-state inactivation, compared with wild-type and Fgf14F145S/+ cells. The slopes of the fits to the steady-state inactivation data (Fig. 4B) in Fgf14F145S/F145S, Fgf14F145S/+, and wild-type Purkinje neurons are 10.5 ± 0.6, 9.9 ± 0.7, and 11 ± 0.9, respectively. Conversely, plots of normalized Nav conductance activation versus membrane voltage are not significantly different in wild-type, Fgf14F145S/+, and Fgf14F145S/F145S Purkinje neurons (Fig. 4C) with V1/2 values from Boltzmann fits of −35 ± 0.5 mV, −32 ± 1 mV, and −31 ± 1.8 mV, respectively.
Expression of Fgf14F145S results in selective changes in Nav current inactivation in cerebellar Purkinje neurons that are similar to the changes identified in cells with targeted deletion of Fgf14. A, To analyze the voltage dependences of steady-state inactivation of INaT in Purkinje neurons across genotypes, peak INaT amplitudes evoked at −20 mV from a 2.5 ms voltage step to various conditioning potentials (−120 to −10 mV) were measured and normalized to the peak INaT measured from the −120 mV conditioning voltage step in the same cell; the voltage-clamp paradigm is illustrated below the representative current records. B, Mean ± SEM normalized peak INaT amplitudes in wild-type (black; n = 11), Fgf14F145S/+ (purple, n = 8), and Fgf14F145S/F145S (blue, n = 10) Purkinje neurons are plotted as a function of the conditioning voltage and fitted with first-order Boltzmann functions. C, To determine the voltage dependences of INaT activation, the mean ± SEM normalized peak INaT conductances calculated (as described in Materials and Methods) in wild-type (black; n = 8), Fgf14F145S/+ (purple, n = 7), and Fgf14F145S/F145S (blue, n = 7) Purkinje neurons are plotted as a function of the test potential and fitted with first-order Boltzmann functions. D, The voltage dependences of steady-state inactivation of INaT were also measured in adult Fgf14+/− (green, n = 10) and Fgf14−/− (red, n = 10) Purkinje neurons, plotted and fitted (as in panel B), and compared with the steady-state inactivation data for INaT in wild-type (black; n = 11) Purkinje neurons. E, Mean ± SEM normalized peak Nav conductances were also determined in Fgf14+/− (green, n = 7) and Fgf14−/− (red, n = 7) Purkinje neurons and are plotted and fitted (similar to panel C) in comparison with data from wild-type (black, n = 8) Purkinje neurons. F, To compare the kinetics of INaT decay, the waveforms of the currents evoked at 0 mV in wild-type, Fgf14−/−, and Fgf14F145S/F145S Purkinje neurons were fitted with single exponentials. As is evident, the mean ± SEM time constants determined from these fits indicate that the kinetics of INaT inactivation are similar (one-way ANOVA) in wild-type, Fgf14−/−, and Fgf14F145S/F145S Purkinje neurons.
Similar to Purkinje neurons with targeted knock-in of the Fgf14F145S mutation, targeted disruptions in one or both Fgf14 alleles caused significant (p ≤ 0.0001, two-way ANOVA) hyperpolarized shifts in the voltage dependence of Nav current steady-state inactivation, compared with wild-type controls (Fig. 4D). Boltzmann fits of normalized Nav current availability (plotted against the conditioning voltage) in wild-type (n = 11), Fgf14+/− (n = 10), and Fgf14−/− (n = 10) Purkinje neurons had V1/2 values of −58.0 ± 0.9 mV, −64.0 ± 0.9 mV, and −71.0 ± 1.0 mV, respectively, and slope factor values of 11.3 ± 0.9, 12.1 ± 0.9 and, 10.5 ± 1.0, respectively. Targeted deletion of Fgf14 did not affect the voltage dependence of Nav current activation (Fig. 4E). The kinetics of INaT inactivation was measured at a 0 mV in wild-type, Fgf14F145S/F145S, and Fgf14−/− Purkinje neurons by fitting the decay phases of the currents with single exponentials; the mean ± SEM time constant derived from these fits (Fig. 4F) reveal that INaT inactivation kinetics in Fgf14F145S/F145S (0.35 ± 0.02 ms) and Fgf14−/− (0.31 ± 0.02 ms) Purkinje neurons is not significantly different from wild-type (0.37 ± 0.03 ms) values (one-way ANOVA).
Loss of iFGF14 does not affect INaR or the kinetics of INaT inactivation in acutely isolated neonatal Purkinje neurons
Acute, shRNA-mediated knockdown of iFGF14 in isolated neonatal cerebellar Purkinje neurons was reported to accelerate the kinetics of INaT inactivation and to reduce the density of the resurgent Nav current (INaR; Yan et al., 2014). As illustrated in Figure 4C, we found that the mean ± SEM time constants of INaT inactivation in adult Fgf14F145S/F145S and Fgf14−/− Purkinje neurons in acute cerebellar slices are similar to wild-type controls. To further explore the functional impact of the loss of iFGF14 on Nav currents in Purkinje neurons directly, we obtained voltage-clamp recordings from acutely isolated wild-type and Fgf14−/− neonatal (P11–P16) Purkinje neurons (Fig. 5A) and measured INaT inactivation kinetics and INaR densities in these cells. Representative INaR waveforms recorded in wild-type and Fgf14−/− Purkinje neurons are shown in Figure 5B. As is evident, these experiments revealed INaR measured in wild-type and Fgf14−/− Purkinje neurons are similar in magnitude and voltage dependences of activation (Fig. 5C; two-way ANOVA, p = 0.20). These plots are also similar when INaR is normalized to peak INaT amplitudes measured at 0 mV (Fig. 5D; two-way ANOVA, p = 0.24) or whole-cell membrane capacitances (Fig. 5E; two-way ANOVA, p = 0.34). Similar to Figure 4F, INaT inactivation kinetics were measured by fitting the decay phase of INaT to first-order exponentials. Compared with isolated wild-type Purkinje neurons, the kinetics of INaT inactivation measured during 0 mV and −20 mV voltage steps in Fgf14−/− isolated Purkinje neurons were also indistinguishable from the rates measured in wild-type cells (Student’s unpaired t tests, p = 0.66 and 0.20, respectively). Mean ± SEM time constants of INaT inactivation at 0 mV and −20 mV measured in wild-type (0 mV, 0.37 ± 0.02 ms; −20 mV, 0.52 ± 0.04 ms) and Fgf14−/− (0 mV, 0.35 ± 0.03 ms; −20 mV, 0.45 ± 0.04 ms) Purkinje neurons are plotted in Figure 5F.
Targeted deletion of Fgf14 does not affect INaR or the kinetics of INaT inactivation in acutely isolated Purkinje neurons. A, Differential interference contrast (DIC) image of an acutely isolated neonatal cerebellar Purkinje neuron during whole-cell patch-clamp recording. B, Representative TTX-sensitive INaR recorded during a voltage step to −45 mV following a 5 ms depolarization to 0 mV in acutely isolated wild-type (black) and Fgf14−/− (red) Purkinje neurons with 151 mM Na+ in the bath; the voltage command is displayed below the current records. C–E, Mean ± SEM peak INaR amplitudes (C), peak INaR normalized to peak INaT at 0 mV (D), and INaR/INaT normalized to cell capacitance (E) measured in acutely isolated neonatal Fgf14−/− (red, n = 11) and wild-type (black, n = 15) Purkinje neurons and plotted as a function of voltage are indistinguishable in magnitude and voltage dependence. Note that in all cases, INaP, which measured 80 ms into the sweep, was digitally subtracted before determination of peak INaR. F, The mean ± SEM time constants of INaT inactivation, measured at −20 and 0 mV, are also similar in isolated neonatal wild-type (n = 7) and Fgf14−/− (n = 7) Purkinje neurons.
Fgf14F145S/F145S animals display motor deficits similar to Fgf14−/− animals
To assess balance and motor coordination, 6–8-week-old animals harboring the Fgf14F145S mutation were examined using the elevated balance beam test (Carter et al., 2001) and compared directly with age-matched animals with targeted deletion of Fgf14 and wild-type controls. A schematic of the elevated balance beam (described in Materials and Methods) is shown in Figure 6A. During 5 consecutive testing days, adult animals individually traversed a narrow (11-mm-wide) beam from a clear platform into an enclosed box. Fgf14F145S/F145S animals (N = 7) displayed impaired performance, compared with wild-type controls (p < 0.0001; two-way ANOVA), and these deficits, measured as time to cross and the number of hindlimb foot-slips, were similar to deficits measured in Fgf14−/− (N = 6) animals (Fig. 6B,C). Interestingly, Fgf14F145S/+ (N = 7) and Fgf14+/− (N = 8) animals performed similarly on the elevated balance beam to wild-type (N = 8) controls (p = 0.07; RM two-way ANOVA; Fig. 6B,C). Across the 5 d testing period, mean crossing times (±SEM) were 5.7 ± 0.8 s for wild-type animals, 4.8 ± 0.5 s for Fgf14F145S/+ mice, and 4.3 ± 0.4 s for Fgf14+/− mice. Older (16- and 32-week-old) Fgf14+/− animals showed performance comparable with age-matched wild-type controls (data not shown).
Expression of the Fgf14F145S mutant allele results in behavioral deficits similar to targeted Fgf14 deletion. A, Balance and motor coordination were evaluated and compared across genotypes by measuring performance on the elevated balance beam (see Materials and Methods). B, The mean ± SEM times to cross an 11 mm (wide) beam were significantly (p < 0.0001; two-way ANOVA) longer for Fgf14F145S/F145S (N = 7) and Fgf14−/− (N = 6) animals, than for wild-type (N = 8) and/or heterozygous (Fgf14+/−, N = 8, and Fgf14F145S/+, N = 8) animals. C, Analysis of the numbers of hindlimb foot-slips revealed significantly (p < 0.0001; two-way ANOVA) more errors for Fgf14F145S/F145S and Fgf14−/− animals than wild-type animals. The performance of heterozygous (Fgf14+/− and Fgf14F145S/+) animals was also similar to wild-type controls.
Evoked firing Purkinje neurons lacking iFGF14 is rescued by membrane hyperpolarization
The results of the voltage-clamp experiments presented in Figures 4 and 5 suggest the impaired excitability of Purkinje neurons with targeted Fgf14 deletion, or knock-in of the Fgf14F145S human mutation, is the result of the hyperpolarized shift in the voltage dependence of steady-state inactivation of INaT. Additional experiments were performed to determine directly if membrane hyperpolarization, which will drive recovery of inactivated Nav channels, is effective in rescuing high-frequency firing. Evoked firing, driven by depolarizing current injections, was measured after a −0.5 nA prepulse (Fig. 7A, right). The −0.5 nA prepulse silenced spontaneous firing across genotypes and resulted in similar resting membrane potentials (approximately −60 mV; Fig. 7B), prior to the depolarizing current injections. In Purkinje neurons from wild-type animals, the mean ± SEM evoked firing frequencies were significantly (p ≤ 0.0001, two-way ANOVA) higher than in Purkinje neurons from Fgf14+/− and Fgf14−/− animals (Fig. 7C) and also higher (p ≤ 0.001, two-way ANOVA) than in Purkinje neurons from Fgf14F145S/F145S and/or Fgf14F145S/+ animals (Fig. 7D). In addition, Purkinje neurons from animals with one wild-type Fgf14 allele (Fgf14F145S/+ and Fgf14+/−) displayed an intermediate evoked firing phenotype, firing at frequencies that were significantly higher (p ≤ 0.0001, two-way ANOVA) than those measured in Fgf14F145S/F145S and Fgf14−/− Purkinje neurons. The intermediate phenotypes in evoked firing rates measured in heterozygous animals are consistent with the intermediate shifts in the voltage dependences of Nav current steady-state inactivation measured in Fgf14F145S/+ and Fgf14+/− Purkinje neurons (Fig. 4). To test the hypothesis that the observed differences in the voltage dependence of Nav current steady-state inactivation (Fig. 4) underlie the reduced evoked firing rates in Fgf14+/− Purkinje neurons, we hyperpolarized these cells to more negative membrane potentials (i.e., negative to −60 mV), to enable further recovery of inactivated Nav channels. These membrane potentials are plotted as different colors in Figure 7E from two representative Fgf14+/− cells. As is evident, the results of these experiments indicate that evoked firing frequencies can be increased further if the cells are hyperpolarized to more negative potentials (than −60 mV), consistent with additional recovery of inactivated Nav channels.
Membrane hyperpolarization rescues firing in Fgf14F145S mutant and Fgf14 knock-out Purkinje neurons. A, Representative recordings of firing evoked by 0.5 and 1.5 nA current injections in an adult wild-type Purkinje neuron are shown. Prior to depolarizing current injections, −0.5 nA current injections were delivered to hyperpolarize the cells. Current injection protocols are shown below the voltage records. The −0.5 nA hyperpolarizing current injections resulted in membrane potential hyperpolarizations of similar magnitudes in Purkinje neurons of all genotypes, as illustrated in panel B. C, D, Mean ± SEM evoked repetitive firing frequencies measured in wild-type, Fgf14 knock-out, and Fgf14F145S knock-in Purkinje neurons are plotted against the amplitudes of depolarizing current injections. As is evident, evoked repetitive firing frequencies are similar in heterozygous mutant (Fgf14+/− and Fgf14F145S/+) and in homozygous mutant (Fgf14−/− and Fgf14F145S/F145S) Purkinje neurons. E, Evoked repetitive firing frequencies are plotted against the amplitudes of the depolarizing current injections presented after hyperpolarizing prepulses to varying membrane voltages (plotted in different colors) in two representative Fgf14+/− cells. In both Fgf14+/− Purkinje neurons, increasing the amplitude of the membrane hyperpolarization (prior to the depolarizing current injection) resulted in higher evoked firing frequencies. F, G, Durations of repetitive firing during the 0.5 s current injection in wild-type, Fgf14 knock-out, and Fgf14F145S knock-in adult Purkinje neurons are plotted against the depolarizing current injection. Similar levels of depolarization block is observed in Purkinje neurons from heterozygous (Fgf14+/− and Fgf14F145S/+) and homozygous (Fgf14−/− and Fgf14F145S/F145S) animals.
The rescue of evoked firing (after hyperpolarization) in Purkinje neurons lacking Fgf14 is expected to be temporary during depolarizing current injections because, during the sustained depolarizing current injections, Nav channels should again accumulate into the inactivated state. We measured the rate at which Purkinje neurons enter a state of depolarization block during the 0.5 s depolarizing current injections. The mean ± SEM firing duration is plotted against the depolarizing current injection in wild-type, FGF14F145S knock-in, and Fgf14 knock-out Purkinje neurons in Figure 7F,G. These experiments revealed that Fgf14−/− and Fgf14F145S/F145S Purkinje neurons undergo depolarization block more rapidly (p ≤ 0.0001, two-way ANOVA) than wild-type, Fgf14+/− (Fig. 7F), and Fgf14F145S/+ (Fig. 7G) Purkinje neurons.
Discussion
Fgf14145S is a loss-of-function mutation
We developed a knock-in mouse model to determine directly the effects of in vivo expression of the first mutation in FGF14, which results in a single (phenylalanine to serine) amino acid substitution, that was linked to familial spinocerebellar ataxia (van Swieten et al., 2003). Electrophysiological experiments revealed that heterozygous Fgf14F145S/+ mutant cerebellar Purkinje neurons did not display the spontaneous, high-frequency repetitive firing, which is a hallmark feature of wild-type Purkinje neurons. Rather, Fgf14F145S/+ cerebellar Purkinje neurons fire in lengthy bursts of action potentials punctuated by periods of quiescence. Similar firing patterns were observed in wild-type cerebellar Purkinje neurons following exposure to (1 nM) TTX and in heterozygous knock-out Fgf14+/− Purkinje neurons, suggesting that the impaired firing of Fgf14F145S/+ Purkinje neurons reflects reduced Nav currents. Voltage-clamp experiments revealed marked hyperpolarizing shifts in the voltage dependence of INaT inactivation in Fgf14F145S/+ and Fgf14+/− Purkinje neurons. Western blot analyses revealed reduced iFGF14 expression in cerebellar lysates prepared from Fgf14F145S/+ and Fgf14+/− animals and, in addition, that iFGF14 is undetectable in cerebellar lysates prepared from homozygous Fgf14F145S/F145S and Fgf14−/− animals. It is possible the anti-iFGF14 antiserum probe used for iFGF14 labeling is not effective in labeling iFGF14 carrying the F145S mutation; however, the Western blot data are consistent with effects of the iFGF14F145S mutation measured in voltage- and current-clamp experiments. Taken together, these results suggest that the mutant iFGF14F1455S protein is unstable in vivo and that the impaired firing properties observed in Fgf14F145S/+ Purkinje neurons reflect haploinsufficiency, consistent with the original suggestion that the amino acid change from a phenylalanine to a serine at position 145 (150), which lies in the central core of iFGF14A (B; Fig. S1), would be expected to reduce the stability and, therefore, result in iFGF14 protein haploinsufficiency (van Swieten et al., 2003).
Partial block of Nav channels using a low concentration (1 nM) of TTX impaired firing in a manner similar to that observed in Fgf14+/− and Fgf14F145S/+ Purkinje neurons (Fig. 2). Further support for the hypothesis that a loss of available Nav channels is responsible for the altered firing phenotypes of Fgf14+/− and Fgf14F145S/+ Purkinje neurons is provided by the results of experiments showing that membrane hyperpolarization, which drives recovery of inactivated Nav channels, resulted in increased evoked firing rates in Fgf14F145S/+ (and Fgf14+/−) Purkinje neurons (Fig. 7). It has been reported that shRNA-mediated knockdown of iFGF14 in dissociated Purkinje neurons in culture results in marked acceleration of the rate of inactivation INaT and reduced INaR, compared with cells expressing a scrambled, control shRNA (Yan et al., 2014). Accelerated INaT inactivation has also been reported in acutely isolated Fgf14−/− Purkinje neurons studied in the presence of TTX and with reduced (50 mM) extracellular sodium (White et al., 2019). Under the physiological recording conditions used here (151 mM extracellular sodium), however, we found that the kinetics of INaT inactivation, as well as the voltage dependence and the magnitude of INaR, is similar in acutely isolated Fgf14−/− and wild-type Purkinje neurons (Fig. 5B–D). Together, these results support the hypothesis that hyperpolarized shifts in the voltage dependence of INaT inactivation are responsible for the reduced firing in Purkinje neurons that lack two wild-type Fgf14 alleles.
It has previously been reported that transfection of isolated neonatal hippocampal neurons with cDNA constructs encoding the iFGF14F145S mutant reduced Nav currents and impaired evoked action potential firing, effects attributed to a dominant negative effect of expression of the iFGF14F145S mutant protein on the interaction between wild-type iFGF14 and Nav α subunits (Laezza et al., 2007). Transfection of isolated cerebellar granule neurons with a cDNA construct encoding the iFGF14F145S protein was also shown to exert dominant-negative effects on voltage-gated Ca2+ currents and vesicular recycling (Yan et al., 2013). The distinct results reported previously and the results presented here demonstrate there are clear and important differences in the effects of acutely expressed cDNA constructs encoding the mutant iFGF14F145S protein in cultured neurons and the in vivo effects of genomic Fgf14F145S expression. These differences may, at least in part, be driven by the sustained loss of surrounding glia (Schlag et al., 1998) and/or reflect in vitro transcriptomic or posttranscriptomic changes in isolated neurons (Jacobsen et al., 2009; Prilutsky et al., 2015; Wangzhou et al., 2020). Additionally, it is also possible the in vitro transfection experiments, which employed plasmids with very strong promoters (such as CMV) resulting in the marked “overexpression” of the mutant protein, cause dominant-negative or other functional effects that do not occur when the expression of iFGF14F145S occurs under native, physiological conditions.
The results here also revealed that adult (6–8-week-old) Fgf14+/− and Fgf14F145S/+ animals, which express one wild-type Fgf14 allele, performed similarly to age-matched wild-type controls on the elevated balance beam test, indicating that mice harboring a single Fgf14F145S mutation do not recapitulate the motor phenotypes expressed in humans. Conversely, adult (6–8-week-old) Fgf14−/− and Fgf14F145S/F145S animals had severe motor deficits. The absence of a motor phenotype in Fgf14F145S/+ and Fgf14+/− animals was initially surprising, given the marked changes in the spontaneous firing properties of adult Fgf14+/− and Fgf14F145S/+ Purkinje neurons. Additional elevated balance beam testing was performed on Fgf14+/− cohorts (compared with wild-type controls) at 16 and 32 weeks-old and, similar to results measured in 6–8 week-old Fgf14+/− and Fgf14F145S/+ animals, revealed no differences with age-matched wild-type control groups (data not shown). While the lack of a motor phenotype in heterozygous iFGF14 mutant mice likely reflect functional differences between mice and humans, it is also possible that the preparation of acute sagittal brain sections, and the disruption of physiological synaptic and neuromodulatory influences on Purkinje neuron firing, exposes differences in Fgf14+/− and Fgf14F145S/+ Purkinje neuron membrane excitability, when compared with wild-type Purkinje neurons; however, in the intact cerebellar circuits, these changes in membrane excitability are insufficient to disturb cerebellar function. In vivo electrophysiological experiments will be necessary to test this possibility directly.
Human FGF14 mutations have now been linked to early onset, autosomal dominant episodic ataxia type 9, EA9 (Choquet et al., 2015; Piarroux et al., 2020; Graves et al., 2024), and GAA repeat expansions in FGF14 were recently identified as one of the most common genetic causes of late-onset ataxias, now referred to collectively as SCA27B (Rafehi et al., 2023; Wilke et al., 2023; Kartanou et al., 2024; Pellerin et al., 2024; Iruzubieta et al., 2025). The effects of these types of FGF14 mutations on membrane excitability and the mechanisms driving these effects have not been explored. While this study focused on motor phenotypes associated with SCA27A, it is important to also consider that targeted Fgf14 deletion in rodents has been linked to aggression and sexual dysfunction (Hoxha et al., 2019), as well as to cognitive impairments, including deficits in learning, memory, and spatial navigation (Brusse et al., 2006; Wozniak et al., 2007).
Beyond its well-characterized role in modulating Nav channels, iFGF14 knockdown in cultured granule neurons has also been shown to result in reduced P/Q-type calcium (CaV2.1) channels/currents, resulting in diminished excitatory postsynaptic currents at the granule (parallel fiber)→Purkinje synapse (Yan et al., 2013). Moreover, iFGF14 expression has been shown to affect voltage-gated potassium channels, particularly KCNQ2/3 (Kv7.2/7.3), which underlie the M-current. This was initially demonstrated in heterologous systems (HEK293 cells) and supported by immunocytochemistry in primary hippocampal neurons cultured from neonatal rats where iFGF14 influenced KCNQ2 localization at the axon initial segment (Pablo and Pitt, 2017). Notably, recent work has also shown Fgf14 deletion leads to increased excitability in adult mouse CA1 pyramidal neurons without changes in Nav channel inactivation properties, in stark contrast to the reduced excitability and hyperpolarized shifts in the voltage dependence of INaT inactivation observed in Purkinje neurons (Ransdell et al., 2025). These opposing effects highlight the cell type-specific contributions of iFGF14 to neuronal excitability, which likely reflect distinct repertoires of expressed ion channel pore-forming and accessory subunits in different neuronal populations.
The loss of iFGF14, and iFGF14 haploinsufficiency, likely drives changes in the expression properties of other Nav channel accessory subunits and Nav pore-forming alpha subunits. In future work, it will be important to examine whether other intracellular FGFs (e.g., iFGF11, 12, 13) are differentially regulated in Fgf14-deficient models and contribute to compensatory mechanisms across brain regions; however, given the context-dependent nature of iFGF14 effects, changes in the expression of other Nav channel subunits, and upstream regulators of these subunits (Shavkunov et al., 2013; Wadsworth et al., 2020; Singh et al., 2024), should be accessed in a cell type-specific manner.
Footnotes
- Received October 21, 2024.
- Revision received January 2, 2026.
- Accepted January 8, 2026.
-
We thank Drs. A. Burkhalter and T. Hermanstyne for many helpful discussions. We also thank R. Wilson, R. Mellor, and C. Smith for expert technical assistance. This work was supported by the National Institutes of Health (R01NS065761 to J.M.N. and D.M.O., F32NS090765 to J.L.R., R15NS125560 to J.L.R.).
-
The authors declare no competing financial interests.
-
This paper contains supplemental material available at: https://doi.org/10.1523/JNEUROSCI.2016-24.2026
- Correspondence should be addressed to Joseph L. Ransdell at ransdejl{at}miamioh.edu.
- Copyright © 2026 the authors
