JNEUROSCI.0136-24.2024.full-html.html

Calcium-dependent regulation of neuronal excitability is rescued in Fragile X

Syndrome by a tat-conjugated N-terminal fragment of FMRP

Xiaoqin Zhan

1,2

, Hadhimulya Asmara

1,2

, Paul Pfaffinger

and Ray W. Turner

1,2,3

Hotchkiss Brain Institute,

Alberta Children’s Hospital Research Institute,

Department

Clinical Neurosciences, University of Calgary, Calgary, Alberta, Canada T2N 4N1,

Department of Neuroscience, Baylor College of Medicine, Houston, TX 77030, USA

Corresponding author:

Ray W. Turner

Email: rwturner@ucalgary.ca

Abbreviated title:

Calcium-dependent regulation of I

depends on FMRP

Number of Figures: 8

Number of Extended Data Figures: 4

Number of words: 228 (Abstract), 465 (Introduction), and 1488 (Discussion)

Conflict of interest:

R.W. Turner is named on a pending patent application for the use of

FMRP-N-tat as a potential therapeutic treatment for Fragile X Syndrome filed through Innovate

Calgary, University of Calgary. None of the other authors declare competing financial interests.

Acknowledgments

This work was supported by the Canadian Institutes for Health Research

and Fragile X Canada / FRAXA, and a SFARI Explorer grant. Postdoctoral Fellowships (X.Z.)

JNeurosci Accepted Manuscript

Abstract (228 words)

Fragile X Syndrome arises from the loss of Fragile X Messenger Ribonucleoprotein (FMRP)

needed for normal neuronal excitability and circuit functions. Recent work revealed that FMRP

contributes to mossy fiber LTP by adjusting Kv4 A-type current availability through interactions

with a Cav3-Kv4 ion channel complex, yet the mechanism has not yet been defined. In this study

using wild-type and

Fmr1

knockout (KO) tsA-201 cells and cerebellar sections from male

Fmr1

KO mice, we show that FMRP associates with all subunits of the Cav3.1-Kv4.3-KChIP3

complex, and is critical to enabling calcium-dependent shifts in Kv4.3 inactivation to modulate A-

type current. Specifically, upon depolarization Cav3 calcium influx activates dual specific

phosphatase 1/6 (DUSP1/6) to deactivate ERK1/2 (ERK) and lower phosphorylation of Kv4.3, a

signalling pathway that does not function in

Fmr1

KO cells. In

Fmr1

KO mouse tissue slices

cerebellar granule cells exhibit a hyperexcitable response to membrane depolarizations. Either

incubating

Fmr1

KO cells or in vivo administration of a tat-conjugated FMRP N-terminus

fragment (FMRP-N-tat) rescued Cav3-Kv4 function and granule cell excitability, with a decrease

in the level of DUSP6. Together these data reveal a Cav3-activated DUSP signalling pathway

critical to the function of a FMRP-Cav3-Kv4 complex that is misregulated in

Fmr1

KO conditions.

Moreover, FMRP-N-tat restores function of this complex to rescue calcium-dependent control of

neuronal excitability as a potential therapeutic approach to alleviating the symptoms of Fragile X

Syndrome.

Significance Statement (116 words)

Changes in neuronal excitability and ion channel functions have been a focus in studies of

Fragile X Syndrome. Previous work identified ion channels that are regulated by FMRP through

either protein translation or direct protein-protein interactions. The current study reveals FMRP

as a constitutive member of a Cav3-Kv4 complex that is required for a Cav3-DUSP-ERK

signalling pathway to increase A-type current and reduce cerebellar granule cell excitability. In

Fmr1

KO cells, Cav3-Kv4 function and calcium-dependent modulation of A-type current is lost,

leading to a hyperexcitable state of cerebellar granule cells. Pretreating with FMRP-N-tat

restores all Cav3-Kv4 function and granule cell excitability, providing support for FMRP-tat

peptide treatment as a potential therapeutic strategy for Fragile X Syndrome.

JNeurosci Accepted Manuscript

Introduction (465 words)

Fragile X Syndrome (FXS) arises through a CGG repeat expansion in the

Fmr1

gene that blocks

expression of Fragile X Messenger Ribonucleoprotein (FMRP), representing the leading

monogenic cause of autism spectrum disorders (ASD). Cell and circuit function in FXS is

disrupted because FMRP exerts vast influence over protein translation and modulation of ion

channels that control neuronal excitability (Contractor et al., 2015; Deng and Klyachko, 2021).

The cerebellum acts as an important sensory-motor interface that impacts the activity of cortical

systems that contribute to ASD

(D’Mello and Stoodley, 2015; Gibson et al., 2023; G allagher and

Hallahan, 2012; Liu et al., 2021). A prominent layer of granule cells receive a massive array of

mossy fibre input carrying sensory information that is processed to enable time coding, sensori-

motor integration, and predictive expectation

(D’Angelo and De Zeeuw, 2009; Garrido et al.,

2013; Giovannucci et al., 2017; Sgritta et al., 2017; Wagner et al., 2017). Control over the

excitability of granule cells could then exert significant influence on higher levels of circuit

function relevant to ASD and FXS.

Membrane excitability in granule cells is known to be highly sensitive to the availability of Kv4

potassium channel mediated “A-type” current (I

). Bursts of mossy fibre input to granule cells

was shown to reduce I

through a mechanism that involves ERK1/2 (ERK), a kinase increasingly

recognized for its role in FXS and ASD (Curia et al., 2013; Osterweil et al., 2013; Asiminas et al.,

2019; Murari et al., 2023). A resulting increase in the rate of spike firing then contributes to long-

term potentiation of mossy fibre input (Zhan et al., 2020). Cerebellar granule cells also express a

Cav3-Kv4 complex that invokes calcium-dependent modulation of Kv4 availability through

association with a calcium-sensing KChIP3 protein (Anderson et al., 2010a; Heath et al., 2014;

Vierra and Trimmer, 2022). Recently FMRP was found to be closely associated with the Cav3-

Kv4 complex (Zhan et al., 2020), yet the manner in which FMRP integrates into the complex,

and how ERK is involved in controlling I

and membrane excitability is unknown.

The current study reveals a novel signalling pathway whereby Cav3.1 calcium influx activates

the phosphatase DUSP1/6 to reduce activated ERK and phosphorylation of subunits in the

complex that increases I

availability. FMRP proves to be critical to calcium-dependent

modulation of Kv4.3 as well as the expression levels of DUSP6, where loss of FMRP in

Fmr1

KO mice leads to hyperexcitability in cerebellar granule cells. Yet all aspects of Cav3-Kv4

complex function and granule cell excitability are rescued in

Fmr1

KO cells within 24 hr of

JNeurosci Accepted Manuscript

reintroducing a tat-conjugated N-terminal fragment of FMRP. The results thus identify a new

signalling pathway that depends on FMRP to control the influence of I

on intrinsic excitability of

a key cell type in receipt of sensory input.

100

101

Material and Methods:

102

103

Cell lines

104

tsA-201 cells originally derived from a female were purchased from Sigma-Aldrich (Cat#

105

96121229) and maintained in DMEM supplemented with 10% heat inactivated foetal bovine

106

serum and 1% pen

–streptomycin at 37 °C (5% CO2). The calcium phosphate method was used

107

to transiently transfect cDNAs of heterologous proteins. Cells were washed with fresh medium

108

16 -18 hr after transfection and then transferred to 32 °C (5% CO2) and cultured for up to 36 hr

109

prior to tests. Previous work has determined that tsA-201 cells we use do not endogenously

110

express Cav3.1 or Kv4 channels (X. Zhan, unpublished observations), and have no expression

111

of KChIP3 (

Figure 1-1

). For different experiments, cells were transfected with different

112

combinations of human cDNA Cav3.1 (2.2 µg), Kv4.3 (1.5 µg), KChIP3 cDNA (1.5 µg) as

113

indicated in re

sults and figures. In addition, cells were transfected with eGFP cDNA (1 μg) and

114

human Kir2.1 cDNA (1 μg) for electrophysiology and high potassium stimulation experiments,

115

respectively. CRISPR-Cas9 was used to create an

Fmr1

knockout tsA-201 cell line (

Figure 1-2

116

For this three short guide oligonucleotides were synthesised, annealed and cloned into the

117

pSpCas9(BB)-2A-GFP (pX458) vector (Addgene), and constructs were then transfected into

118

tsA201 cells with Lipofectamine 2000 (Invitrogen). After transfection, GFP containing cells were

119

sorted into 96 well plates by flow cytometry. Western blotting and DNA sequencing were used to

120

confirm functional knockout of the

Fmr1

gene. All

Fmr1

knockout tsA-201 cells used in the

121

experiments were maintained from clone 7. To verify that knockout of

Fmr1

did not alter subunit

122

expression, we conducted Western blot on cells coexpressing Cav3.1-Kv4.3-KChIP3 and found

123

no significant difference for any of the subunits between WT and KO cells (

Figure 1-3

124

125

Mouse lines

126

JNeurosci Accepted Manuscript

Wild-type (Jackson Lab stock #004828, FVB.129P2-Pde6b+ Tyrc-ch/ AntJ,

127

https://www.jax.org/strain/004828) and

Fmr1

knockout (Jackson Lab stock #004624,

128

FVB.129P2-Pde6b+ Tyrc-ch

Fmr1

tm1Cgr/J, https://www.jax.org/strain/004624) mice were

129

maintained in an Animal Resource Center of the University of Calgary in accordance with

130

guidelines of the Canadian Council of Animal Care. Animals had free access to water and food

131

with a temperature of 20 - 23 °C on a 12 hr light and dark cycle.

132

133

Cerebellar slice preparation

134

Sagittal cerebellar sections of 250 µm thickness were prepared from P21-P25 male mice as

135

previously described

. Briefly, after isoflurane inhalation anaesthesia animals were decapitated

136

and the cerebella were dissected out and placed in ice-cold artificial cerebrospinal fluid (aCSF)

137

composed of (in mM): 125 NaCl, 25 NaHCO

, 25 D-glucose, 3.25 KCl, 1.5 CaCl

, 1.5 MgCl

138

bubbled with carbogen (95% O

and 5% CO

) gas. Tissue slices were cut by vibratome (Leica

139

VT1200 S) and recovered for 20 - 30 min at 37 ºC before storing at room temperature (25ºC) in

140

carbogen-bubbled aCSF before transferral to a recording chamber on the stage of an Olympus

141

BX51W1 microscope maintained at 32 ºC. Whole-cell recordings were obtained in granule cells

142

of lobule 9.

143

144

Electrophysiology

145

Whole-cell patch recordings were obtained using a Multiclamp 700B amplifier, Digidata 1440A to

146

digitize at 40 kHz and analysed with pClamp 10.5 software. Glass pipettes of 1.5 mm O.D. (A-M

147

Systems) were pulled using a P-95 puller (Sutter Instruments; 4

–8 MΩ). For voltage clamp

148

recordings, series resistance was compensated to at least 70% and leak was subtracted offline

149

in pClamp software. Kv4 activation and inactivation plots were recorded from a holding potential

150

of −110 mV stepped in 10 mV (1000 ms) increments to 60 mV, followed by a return step to −30

151

mV (500 ms) as the test potential.

Inactivation curves were fitted according to the Boltzmann

152

equation:

Imax

) = 1/(1 + exp((

)), where

is the half inactivation potential and

153

the slope factor. Activation curves were fitted by the Boltzmann equation:

Gmax

) = 1/(1 +

154

exp((

)), where

is calculated with

Vrev

is the half activation potential and

155

JNeurosci Accepted Manuscript

is the slope factor. Inactivation and activation plots were constructed using Origin 8.0 (OriginLab,

156

Northampton, MA).

157

tsA-201 cells:

Recordings of Kv4.3 current expressed in tsA-201 cells that express FMRP (WT)

158

or in which

Fmr1

was knocked out by CRISPR-Cas9 (KO) (

Figure 1-2, 1-3

) were carried out at

159

room temperature with an external solution (pH adjusted to 7.3 with NaOH) composed of (mM):

160

125 NaCl, 3.25 KCl, 1.5 CaCl

, 1.5 MgCl

, 10 HEPES, 10 D-Glucose, and 2 TEA. Pipettes were

161

filled with a solution composed of (pH 7.3 with KOH, mM): 110 potassium gluconate, 30 KCl, 1

162

EGTA, 5 HEPES, and 0.5 MgCl

, with 5 di

–tris-creatine phosphate, 2 Tris-ATP, and 0.5 Na-GTP

163

added from fresh frozen stock each day. Where indicated, internal levels of calcium

164

concentration were adjusted according to MaxChelator calculations. Previous work (Turner et

165

al., 2016) and unpublished tests on the effects of exchanging internal electrolyte with drugs or a

166

buffered calcium concentration assure complete exchange of the effective internal levels of

167

calcium in 5-10 min of breaking into whole cell recording mode.

168

169

Cerebellar granule cells:

Current clamp recordings of granule cells from cerebellar sections

170

used an internal solution modified from Gall et al (Gall et al., 2003) composed of (mM): 126 K-

171

gluconate, 4 NaCl, 15 Glucose, 5 HEPES, 1 MgSO

, 0.15 BAPTA, 0.05 CaCl

, pH 7.3 via KOH,

172

with 5 di

–tris-creatine phosphate, 2 Tris-ATP, and 0.5 Na-GTP added from fresh frozen stock

173

each day. Bath solution was carbogen-bubbled aCSF maintained at 32ºC. A holding current of

174

less than 10 pA was applied to maintain a holding potential of -70 mV. Input resistance was

175

determined by hyperpolarizing cells with injections of 0 to -8 pA currents in five steps at an

176

increment of -2 pA. Cerebellar granule cell excitability was measured by depolarizing cells from -

177

70 mV with 1 sec steps of currents of 0 - 16 pA in 2 pA increments. Spike events were detected

178

and analysed with pClamp 10.5 software (Axon, Molecular Devices).

179

180

Chemicals and Proteins

181

Chemical compounds were obtained from Sigma-Aldrich unless otherwise indicated. Drugs were

182

either bath-applied or infused directly through the electrode with the 2PK+ Perfusion system

183

(ALA Scientific, Farmingdale, NY) (Zhan et al., 2020). Drugs were applied at the following

184

concentrations: 30 μM (E)-2-benzylidene-3-(cyclohexylamino)-2,3-dihydro- 1H-inden-1-one

185

(BCI), 0.1 μg/ml MEK1 activated extracellular signal-regulated kinase 1 and 2 (pERK1/2)

186

JNeurosci Accepted Manuscript

(Sign

alChem Biotech), 1 μM TTA-P2 (Alomone Labs), 1 μM Mibefradil (Sigma-Aldrich), and 3

187

nM Okadaic acid (Sigma-Aldrich). FMRP(1-297)-

tat

was prepared as previously described (Zhan

188

et al., 2020), and was added to the external medium of KO cells for 7 - 8 hr, or delivered in vivo

189

by tail vein injection 24 hr before tests (Zhan et al., 2020).

190

191

Immunoprecipitation (IP) and Co-Immunoprecipitation (co-IP)

192

Protein-protein associations and phosphorylation of the Cav3-Kv4 complex were tested at

193

different levels of [K]o with co-IP and IP as described previously (Zhan et al., 2020). Transfected

194

tsA-201 cells were washed with a low [K]o (LK) solution composed of (mM): 148 NaCl, 1 KCl, 10

195

HEPES, 3 CaCl

, 1 MgCl

, 10 Glucose, and equilibrated in LK solution for 10 min. Cells were

196

then exposed to a high [K]o (HK) solution composed of (mM): 148 NaCl, 50 KCl, 10 HEPES, 3

197

CaCl

1 MgCl

, 10 Glucose for 5 min before being lysed in Lysis Buffer containing (mM): 150

198

NaCl, 50 Tris, 2.5 EGTA, 1% NP-40, pH 7.5, phosphatase inhibitor (Sigma-Aldrich, Cat. P5726)

199

and proteinase inhibitor (Roche, Cat. 04693124001). Lysates containing 500 µg total protein

200

were incubated at 4°C for 2 - 3 hr with one of the following antibodies: Rabbit anti-Cav3.1 (1:50,

201

custom made and lab verified (Molineux et al., 2006)), rabbit anti-Kv4.3 (1:50, Abcam, Cat.

202

ab65794, RRID:AB_1140929), mouse anti-KChIP3 (1:50, Neuromab, Cat. K66/38,

203

RRID:AB_2877337), rabbit normal IgG (as a control, Abcam, Cat. ab172730,

204

RRID:AB_2687931), mouse normal IgG (as a control, Abcam, Cat. ab37355,

205

RRID:AB_2665484), rabbit anti-P-MAPK substrate (1:50, Cell Signaling Technology, Cat.

206

14378s, RRID:AB_2798468), rabbit anti-P-MAPK/CDK substrate (1:40, Cell Signaling

207

Technology, Cat. 2325s, RRID:AB_331820). Background control (BC) was used as a negative

208

control by incubating lysate with normal IgG from the same species as the IP antibody. The

209

immuno-

complexes were then incubated with 30 μl protein G sepharose beads (GE healthcare,

210

17-0618-01), which were pre-equilibrated with Lysis Buffer, at 4°C overnight. After four washes

211

with Lysis Buffer, proteins were eluted from sepharose beads by boiling at 95 - 100 °C for 5 min

212

with 40 µl sample buffer diluted from 4x sample buffer (Bio-Rad, 1610747) with Lysis Buffer. IP

213

samples were then resolved and analyzed with Western blot. All commercially obtained

214

antibodies were verified by suppliers. Full western blot data sets are available for viewing on the

215

repository Figshare: https://figshare.com/s/402f8335aa1816cc35c4.

216

217

JNeurosci Accepted Manuscript

Western Blot

218

Cell lysates and

IP samples were loaded to 6 - 12% SDS-polyacrylamide gel and resolved with

219

SDS-

PAGE. Proteins were transferred to a 0.2 μm PVDF membrane (Millipore) and probed with

220

primary antibodies overnight at 4°C, followed by a goat anti-mouse (1:3000, Invitrogen, Cat.62-

221

6520, RRID:AB_2533947) or donkey anti-rabbit IgG (1:5000, Cytiva, Cat.NA934-1ml,

222

RRID:RRID:AB_772206) HRP-conjugated secondary antibodies. Blot images were taken with a

223

ChemiDoc imager and band densities were analysed with Image Lab (Bio-Rad). Primary

224

antibodies used in this study include: Rabbit anti-FMRP (1:1000, Cell Signalling Technology,

225

Cat.7104s, RRID:AB_10950502), mouse anti-FMRP (1:1000, Abcam, Cat.ab230915,

226

RRID:AB_10950502), rabbit anti-pERK1/2 (1:1000, Cell Signalling Technology, Cat.4370s,

227

RRID:AB_2315112), rabbit anti-tERK1/2 (1:2000, Cell Signalling Technology, Cat.9102s,

228

RRID:AB_330744), mouse anti-GAPDH (1:2000, Invitrogen, Cat.39-8600, RRID:AB_2533438),

229

rabbit anti-DUSP6 (1:1000, Abcam, Cat.ab76310, RRID:AB_1523517), rabbit anti-

ɑTubulin

230

(1:2000, Abcam, Cat.ab176560, RRID:AB_2860019).

231

232

Experimental Design and Statistical Analysis

233

When there were multiple samples in one experiment, samples were randomly chosen to

234

receive different treatments. Experiments were performed with mouse genotypes and

235

experimental groups known to the persons who conducted them. No experimental data without

236

technical issues were excluded. Quantification of Western blot bands was conducted with

237

ImageLab software (Bio-

Rad) by normalising a target protein’s intensity to house-keeping

238

proteins. Patch-clamp recordings were analysed with pClamp 10.5 software. All figures were

239

prepared and statistical analyses performed with OriginPro 8 or GraphPad Prism 6, and Adobe

240

Illustrator CC 2021 software. Data normality was tested with the Shapiro-Wilk test. When

241

normally distributed, statistical significance was determined using two-

sample Student’s

-test for

242

unpaired samples or Welch’s test (

test with Welch correction), and a paired sample Student’s

243

tests for the same samples under different conditions. When normality was rejected, the Mann-

244

Whitney test was used for two sample comparisons.

245

246

Results

247

JNeurosci Accepted Manuscript

Previous work established that A-type current is subject to modulation by Cav3 T-type

248

calcium current, with a primary effect of modulating the voltage for half inactivation (Vh) of Kv4

249

channels and thus the availability of A-type current (I

) (Anderson et al., 2010a; Heath et al.,

250

2014). The calcium-dependence imparted on Kv4 channels is readily apparent when a reduction

251

of Cav3 channel conductance produces a negative shift in Kv4 Vh, a process that reduces the

252

amplitude of I

(Anderson et al., 2010a, 2010b, 2013; Heath et al., 2014). In the case of

253

cerebellar granule cells the Cav3-Kv4-KChIP3 complex has been shown to increase spike

254

output in relation to long-term potentiation (LTP) of mossy fibre inputs (Heath et al., 2014;

255

Rizwan et al., 2016; Zhan et al., 2020).

256

FMRP was recently found to exhibit coimmunoprecipitation (co-IP) and FRET with both

257

Cav3.1 and Kv4.3 (Zhan et al., 2020), suggesting a direct association between FMRP and at

258

least these two members of the Cav3-Kv4-KChIP3 complex. Importantly, these protein-protein

259

interactions can be reproduced in tsA-201 cells by co-expressing cDNA for the principal subunits

260

of Cav3.1-Kv4.3-KChIP3 (Anderson et al., 2010a, 2010b, 2013). To further test the molecular

261

interactions between FMRP and subunits of the complex we first transiently transfected cDNA

262

for human Cav3.1, Kv4.3 and KChIP3 in native tsA-201 cells that express FMRP (WT) or cells in

263

which FMRP was knocked out by CRISPR-Cas9 technology (KO) (

Figures 1-2, 1-3

). Kv4.3 Vh

264

was measured using a series of step commands (1 sec) from a holding potential of -90 mV to

265

+60 mV in 10 mV increments followed by a step back to -30 mV (500 msec).

266

267

Cav3.1 and FMRP regulate Kv4.3 availability

268

We verified the ability for calcium to modulate Kv4.3 Vh by first constructing voltage-

269

inactivation plots for Kv4.3 current in a medium containing 0 calcium (

Fig. 1A

). Addition of 1.5

270

mM calcium to the bathing medium resulted in a positive shift in Kv4.3 Vh (0 Ca

, -55.5 ± 1.9

271

mV; 1.5 Ca

, -49.9 ± 1.8 mV; n = 6;

= 0.001, paired

-test) and an average 142 ± 33.3% (n =

272

= 0.013, paired

-test) increase in current amplitude for a step of -40 mV to -30 mV (

Fig.

273

). Both the calcium-induced shift in Kv4.3 Vh and increase in I

amplitude were blocked by

274

omitting Cav3.1 cDNA from the transfection (

Fig. 1A

). Conversely, applying 1 µM TTA-P2 as a

275

selective Cav3 channel blocker in normal bathing medium containing 1.5 mM calcium negatively

276

shifted Kv4.3 Vh (control, -49.7 ± 1.0 mV; TTA-P2, -54.4 ± 1.7 mV; n = 9;

= 0.003, paired

277

JNeurosci Accepted Manuscript

test) without affecting Kv4.3 voltage for activation (Va) (n = 7,

= 0.124, paired

-test) (

Fig. 1B

278

The effects of TTA-P2 on Kv4.3 Vh was further reflected in a 42 ± 7.4% (n = 9;

= 4.5 x 10

-4

;

279

paired t-test) decrease in I

amplitude for a step from -40 mV to -30 mV (

Fig. 1B

). The negative

280

shift in Kv4.3 Vh induced by TTA-P2 was comparable to that obtained in 1.5 mM calcium

281

medium for application of either 1 µM mibefradil or 300 µM Ni

, with no significant effects on

282

Kv4.3 Va (

Fig. 1C

) (Anderson et al., 2010a).

283

We note that much of the work to date on the Cav3-Kv4-KChIP3 complex was conducted in

284

cerebellar granule cells or WT tsA-201 cells that express FMRP. To test the role of FMRP, we

285

used the

Fmr1

KO cell line. Using KO cells we found that in the absence of FMRP, the ability for

286

TTA-P2 application to induce a negative shift in Kv4.3 Vh in cells was lost (

Fig. 1D

). To further

287

test the effects of a loss of FMRP we recorded Kv4.3 current in KO cells coexpressing the

288

Cav3.1-Kv4.3-KChIP3 complex in standard internal electrode solution or in the presence of 3 nM

289

FMRP(1-297) in the recording electrode. In the presence of internal 3 nM FMRP(1-297), we

290

found no effect of applying 1 µM TTA-P2 on Kv4.3 Vh (

Fig. 1E

). We considered the possibility

291

that reinstating FMRP functions in KO cells might require more time than the 10-15 min typical of

292

whole-cell recordings. It was previously shown that attaching the cell permeating

tat

-peptide

293

moiety to FMRP(1-297) (FMRP-N-

tat

) facilitates its passage across cell membranes (Zhan et al.,

294

2020). We thus pretreated KO cells coexpressing Cav3.1-Kv4.3-KChIP3 with 70 nM FMRP-N-

tat

295

in the cell medium and found variable levels of recovery at 4 hr, but full recovery at 7 hr. When

296

recordings were subsequently conducted in these cells, the Kv4.3 Vh was not changed between

297

FMRP-N-

tat

pretreated cells to those without this treatment (

= 0.07) (

Fig. 1D, F

). However, in

298

cells pretreated for the longer time frame with FMRP-N-

tat

, applying 1 µM TTA-P2 now induced

299

a significant negative shift in Kv4.3 Vh (KO, -48.1 ± 1.4 mV; KO post TTA-P2, -54.8 ± 2.1 mV, n

300

= 9;

= 0.012, paired

-test) (

Fig. 1F

301

Together these tests are important in providing the first evidence that Cav3 calcium

302

conductance can induce a bidirectional shift in Kv4.3 Vh and I

amplitude in cells expressing the

303

Cav3-Kv4-KChIP3 complex. Moreover, Cav3.1 calcium-dependent modulation of Kv4.3 Vh

304

unexpectedly

depends

on the expression of FMRP, identifying a key role for this molecule in the

305

complex. The ability for FMRP-N-

tat

to rescue calcium-dependent modulation of Kv4.3 in KO

306

cells further indicates that the N-terminal portion of FMRP is sufficient to reinstate the normal

307

functions of the Cav3-Kv4 complex.

308

JNeurosci Accepted Manuscript

309

Cav3.1 and KChIP3 are necessary to induce a calcium-dependent modulation of Kv4 Vh

310

It is known that KChIP molecules can act as the calcium sensor for Kv4 channels with 4

311

KChIP subunits present in an assembled channel (An et al., 2000; Vierra and Trimmer, 2022)

312

and EF hands 3 and 4 available to bind calcium. Yet the exact role for KChIP3 in Cav3-

313

dependent modulation of Kv4 channels has not been determined.

314

To test the role of KChIP3 we first coexpressed Cav3.1 and Kv4.3 without KChIP3. In this

315

case applying TTA-P2 (1 µM) produced a very slight shift in Kv4.3 Vh (control, -48.6 ± 1.2 mV;

316

TTA-P2, -49.7 ± 1.1 mV; n = 10;

= 0.003, paired

-test) (

Fig. 2A, B

). However, coexpressing

317

KChIP3 with Cav3.1-Kv4.3 restored the ability for TTA-P2 application to induce a much larger

318

negative shift in Kv4.3 Vh (-4.8 ± 3.4 mV, n = 9;

= 0.005; two-sample

-test) (

Fig. 2B

). The

319

effect of TTA-P2 was lost if an interfering peptide

tat

pp1 (10 μg/ml) that impedes the Kv4.3-

320

KChIP3 interaction (Scannevin et al., 2004; Callsen et al., 2005) was pre-applied in the bath

321

solution (

Fig. 2B

). We further tested the role of specific EF hands by coexpressing KChIP3

322

constructs that were mutated to block function in EF hand 3 (KChIP3

(E186Q)

), EF hand 4

323

(KChIP3

(E234Q)

), or both. Neither of the two single KChIP3 mutants expressed alone with Cav3.1-

324

Kv4.3 fully prevented a TTA-P2-induced negative shift in Kv4.3 Vh (

Fig. 2C

). Coexpressing

325

Cav3.1-Kv4.3 and the double mutant KChIP3

(E186Q, E234Q)

had no effect on the resting value of

326

Kv4.3 Vh (WT Cav3.1-Kv4.3-KChIP3 Vh,

–49.7 ± 1.0 mV, n = 9; WT Cav3.1-Kv4.3-KChIP3

(E186Q,

327

E234Q)

Vh, -48.5 ± 0.6 mV, n = 8,

= 0.334, Welch’s test) but proved to fully block any effects of

328

applying TTA-P2 on Kv4.3 Vh (

Fig. 2C

). These data confirm that calcium sensitivity of the Cav3-

329

Kv4 complex also requires the actions of EF hands 3 and 4 of KChIP3.

330

Previous work established that FMRP forms a link to both Cav3.1 and Kv4.3 (Zhan et al.,

331

2020). To determine if FMRP forms a similar close relationship to KChIP3 as the calcium-

332

sensing component of the complex, we conducted co-IP tests in WT cells. These experiments

333

established that FMRP also co-IPs with KChIP3 (

Fig. 2E; Figure 2-1

). To determine if

334

membrane depolarization and Cav3.1 calcium influx induced any calcium-dependent change in

335

the relationship between KChIP3 and other members of the complex we compared co-IPs

336

obtained in the presence of 1 mM [K]o (Low K; LK) or 50 mM [K]o (High K; HK) solution to

337

promote Cav3.1 calcium influx. HK solution promoted a selective and striking loss of the co-IP

338

JNeurosci Accepted Manuscript

between KChIP3 and FMRP (

Fig. 2E

) but not between KChIP3 and either Kv4.3 or Cav3.1

339

(

Figure 2-1

340

To explore the functional outcome of KChIP3 and FMRP on Kv4 Vh we tested the effects of

341

different levels of internal calcium concentration and with or without FMRP. By coexpressing

342

Kv4.3-KChIP3 in the absence of Cav3.1 and buffering internal calcium concentration to 0 µM

343

(Maxchelator) we determined that adding KChIP3 had no effect on the baseline value of Kv4 Vh

344

in either WT or KO cells (

Fig. 2F

). We then repeated these tests by coexpressing Kv4.3-KChIP3

345

and recorded with internal solutions that had calcium concentrations buffered to either 0 or 13

346

µM to compare Kv4 Vh. In WT cells a higher internal calcium concentration induced a significant

347

positive shift in Kv4 Vh (WT 0 calcium, -40.4 ± 1.5 mV, n = 10; 13 µM calcium, -16.9 ± 1.0 mV, n

348

= 7;

= 5.3E-9, two-sample

-test) (

Fig. 2G

). However, repeating these tests in

Fmr1

KO cells

349

revealed no calcium-dependent shift in Kv4 Vh in the presence of elevated internal calcium (KO

350

0 calcium, -39.2 ± 2.0 mV, n = 8; 13 µM calcium, -41.2 ± 2.3 mV, n = 9;

= 0.64; two-sample

351

test) (

Fig. 2G

), indicating that FMRP is required to invoke a calcium-dependent positive shift in

352

Kv4 Vh.

353

Together these tests are important in indicating that Cav3 calcium-induced shifts in Kv4.3 Vh

354

and modulation of I

amplitude depends on all of Cav3.1, KChIP3 and FMRP, with a selective

355

and dynamic change in the FMRP-KChIP3 association with an elevation of internal [Ca].

356

357

Cav3.1 and FMRP regulate ERK phosphorylation

358

We previously found that a negative shift in Kv4 channel Vh in cerebellar granule cells

359

invoked by mossy fibre stimulation could be prevented by PD-98059, an inhibitor of mitogen-

360

activated protein kinase kinase (MEK) that phosphorylates and activates ERK (Rizwan et al.,

361

2016). Yet the factors that control ERK activation and the potential target(s) for ERK

362

phosphorylation of subunits in the Cav3-Kv4-KChIP3 complex are unknown. ERK has been

363

shown to phosphorylate 3 different sites on Kv4.2 that can cause shifts in Vh or Va to regulate I

364

availability (Adams et al., 2000; Schrader et al., 2002, 2006), but the potential actions of ERK on

365

Kv4.3 have not been reported. Interestingly, Cav3.1-mediated calcium influx has been reported

366

to decrease ERK activation in HEK293 cells (Choi et al., 2005).

367

JNeurosci Accepted Manuscript

To define any Cav3-mediated regulation of the complex by ERK we coexpressed Cav3.1-

368

Kv4.3-KChIP3 in WT cells and used Western blots to detect the effects of membrane

369

depolarization and calcium conductance on ERK1/2 (ERK) activation. For this we used an

370

antibody that targets phosphorylated ERK (pERK) to identify the level of pERK in relation to total

371

ERK (tERK) (

Fig. 3A

). These tests established a baseline ratio of pERK / tERK in low [K]o (1.5

372

mM), and that application of high [K]o decreased the pERK / tERK ratio by 53.7% within 10 min

373

(high [K]o 0.46 ± 0.08, n = 7,

= 4.75E-4, paired

-test ) (

Fig. 3A

). Moreover, this high [K]o-

374

induced reduction in pERK was blocked by pre-incubation in 1 µM TTA-P2 (high [K]o 0.97 ±

375

0.28, n = 4,

= 0.927, paired

-test )(

Fig. 3B

), indicating that a depolarization-induced reduction

376

of pERK depends on Cav3.1 calcium conductance. Repeating these tests in KO cells expressing

377

Cav3.1-Kv4.3-KChIP3 further revealed that the high [K]o-induced reduction in pERK levels was

378

lost in the absence of FMRP (KO high [K]o 0.91 ± 0.10, n = 4,

= 0.422, paired

-test ) (

Fig. 3C

379

These data establish that Cav3.1 calcium conductance lowers the level of activated ERK,

380

and that this process requires the expression of FMRP.

381

382

Cav3.1 activates DUSP to regulate ERK and set Kv4.3 availability

383

A Cav3.1-mediated reduction in pERK levels could contribute to a shift in Kv4.3 Vh and I

384

availability. To test this we recorded from WT cells expressing Cav3.1-Kv4.3-KChIP3 with or

385

without 1 μM TTA-P2 in the bath and measured Kv4.3 Vh. For this Kv4.3 current was first

386

measured using a control internal electrolyte that was subsequently replaced by direct electrode

387

infusion with an electrolyte containing 0.2 µg/ml pERK (

Fig. 4A

). When Cav3.1 conductance

388

was blocked in the presence of 1 μM TTA-P2 in the bath, infusion of pERK induced a significant

389

negative shift in Kv4.3 Vh (n = 6,

= 0.011, paired

-test) (

Fig. 4A

). However, when Cav3.1

390

conductance was intact (no TTA-P2 in medium), internal infusion of pERK had no significant

391

effect on Kv4.3 Vh (n = 3,

= 0.311, paired

-test) (

Fig. 4A

). These results are important in

392

suggesting that Cav3.1 calcium conductance can occlude the actions of pERK even when added

393

directly to the electrolyte. To assess the requirement for FMRP in these results we coexpressed

394

the complex in KO cells in the presence of 1 μM TTA-P2 in the bath to prevent calcium

395

conductance, and found that recording with 0.2 µg/ml pERK in the electrode again significantly

396

negative shifted the Kv4.3 Vh (control -45.7 ± 1.8 mV, n = 6; pERK1/2 -53.6 ± 1.5 mV, n = 8;

397

0.0069, two-sample

-test) (

Fig. 4B

). Together these results indicate the ability for pERK to

398

JNeurosci Accepted Manuscript

induce a negative shift in Kv4.3 Vh through a process that is actively prevented or reduced by

399

Cav3.1 calcium conductance.

400

Cav3.1 calcium influx might reduce pERK levels by activating a phosphatase to

401

dephosphorylate ERK to reduce its actions on downstream targets at serine/threonine (Ser/Thr)

402

sites. The process of dephosphorylation is most often achieved by Ser/Thr phosphatases that

403

include PP1/PP2A (Narayanan et al., 2007). Indeed, Cav3 calcium conductance has been

404

associated with activation of PP2A (Ferron et al., 2011), and PP2A has been shown to

405

dephosphorylate all of FMRP (Narayanan et al., 2007), ERK (Kim et al., 2008) or S6K1 (Hahn et

406

al., 2010). To test the role of PP2A we coexpressed the Cav3.1-Kv4.3-KChIP3 complex in WT

407

cells in the presence of 3 nM okadaic acid to block PP1 and PP2A. We then applied 1 µM TTA-

408

P2 to block Cav3.1 conductance and still found a negative shift in Kv4.3 Vh (control -46.8 ± 1.4

409

mV; TTA-P2 -52.2 ± 1.2 mV; n = 5;

= 4.98E-5, paired

-test) (

Fig. 4C

410

A key factor that specifically controls pERK levels is the family of dual specificity

411

phosphatases (DUSP) that dephosphorylate phospho-threonine and -tyrosine residues on

412

MAPK (Caunt et al., 2008). To test the potential role of DUSPs we coexpressed the Cav3.1-

413

Kv4.3-KChIP3 complex in WT cells and recorded in the presence of 30 µM BCI as an allosteric

414

inhibitor of DUSP1/6 (Molina et al., 2009). Under these conditions perfusion of 1 µM TTA-P2 had

415

no effect on Kv4.3 Vh (control -55.7 ± 1.7 mV; TTA-P2 -56.6 ± 1.6 mV; n = 7;

= 0.12, paired

416

test) (

Fig. 4D

), revealing that BCI occluded the effects of Cav3.1 conductance on Kv4.3. In

417

contrast, when Cav3.1-Kv4.3-KChIP3 was coexpressed in WT cells, 30 µM BCI application by

418

itself produced a significant negative shift in Kv4.3 Vh (control -49.8 ± 0.7 mV; BCI -54.6 ± 1.0

419

mV; n = 9;

= 9.8E-6, paired

-test) (

Fig. 4E

). In fact, the BCI-induced negative shift in Kv4.3 Vh

420

was not significantly different from that induced by TTA-P2 (BCI Vh shift, -4.8 ± 0.5 mV, n = 9;

421

TTA-P2 Vh shift, -4.8 ± 1.1 mV, n = 9,

= 0.99). These results were important in revealing that a

422

Cav3.1-mediated positive shift in Kv4.3 Vh (cf.

Fig. 1A

) involves Cav3.1 activation of DUSP1/6

423

and its influence on pERK. This conclusion was further verified using Western blots to repeat the

424

test of depolarizing WT cells using high [K]o medium (cf.

Fig. 3

), which revealed that inhibiting

425

DUSP1/6 with 30 µM BCI prevented a high [K]o-induced depolarization from reducing the

426

pERK/tERK ratio (

Fig. 4F

427

We further explored the extent to which these results could be reproduced by an increase in

428

[Ca]i (

Fig. 4G

). Using whole-cell recordings in WT cells coexpressing only Kv4.3-KChIP3 we

429

JNeurosci Accepted Manuscript

buffered [Ca] in the electrode (MaxChelator) to either 0 or 13 µM. Cells recorded in the presence

430

of 13 µM internal calcium exhibited a positive shift in Kv4.3 Vh (

Fig. 4G

). However, no effect on

431

Kv4.3 Vh was detected under these conditions in the presence of 30 µM BCI to block DUSP1/6

432

(

Fig. 4G

433

These results reveal that a Cav3.1-mediated activation of DUSP1/6 that downregulates the

434

level of pERK is accompanied by a positive shift in Kv4.3 Vh that will increase I

435

436

Depolarization induces differential effects of phosphorylation of Kv4.3 and FMRP

437

While the previous data revealed that a Cav3.1-DUSP-ERK signalling pathway regulates

438

calcium dependence of the Cav3-Kv4 complex, the actual targets of ERK-mediated

439

phosphorylation in this process are unknown. The tests already conducted enable predictions on

440

what elements may be important. While regulation of Cav3 channel isoforms by ERK has been

441

reported in fibroblasts and cardiomyocytes (Sharma et al., 2023), we have not detected any

442

change in Cav3.1 density or voltage dependence in relation to stimuli that modulate the Cav3-

443

Kv4 complex (Heath et al., 2014; Rizwan et al., 2016; Zhan et al., 2020). Several lines of

444

evidence implicate Kv4.3 as one important target for phosphorylation. Specifically, shifts in Kv4.3

445

Vh could be induced in WT cells by direct infusion of pERK in the presence of TTA-P2 to block

446

Cav3.1 conductance (

Fig. 4A

), or with a direct increase in [Ca]i in the absence of Cav3.1

447

expression (

Fig. 4G

). A negative shift in Kv4.3 Vh could also be induced by internal pERK

448

infusion in WT cells expressing only Cav3.1-Kv4.3 (

Fig. 4A

), suggesting that KChIP3 is not

449

required as an ERK target to enable the calcium-dependent effects studied here. In KO cells

450

coexpressing Cav3.1-Kv4.3-KChIP3 and that lack FMRP, both the calcium-dependent negative

451

shift in Kv4.3 Vh (

Fig. 2G

) and a high [K]o-induced reduction in pERK levels (

Fig. 3C

) were lost,

452

but the effects of pERK on Kv4.3 Vh remained (

Fig. 4C

). These results indicate that pERK

453

actions on Kv4.3 can be sufficient to modulate Kv4.3 Vh but that FMRP is also required for

454

regulating the actions of ERK that promote a calcium-dependent shift in Kv4.3 Vh.

455

ERK is a member of the MAPK family that can phosphorylate targets at either Serine

456

(PXS*P) or Threonine (PXpTP) motifs (Rubinfeld and Seger, 2005). To identify targets for

457

MAPK-mediated phosphorylation, WT cells coexpressing Cav3.1-Kv4.3-KChIP3 were

458

depolarized using high [K]o medium (HK). Cells were then processed for Western blot analysis

459

JNeurosci Accepted Manuscript

for IP between proteins reactive for antibodies specific to MAPK phosphorylated Serine (pSer) or

460

Threonine (pThr) motifs with each of Cav3.1, Kv4.3, KChIP3, and FMRP proteins (

Fig. 5

). These

461

tests revealed pThr sites on FMRP in LK conditions but no detectable difference in

462

phosphorylation levels between LK and HK conditions (

Fig. 5A, B

). No consistent pThr labelling

463

was detected for any of Cav3.1, Kv4.3 or KChIP3 (

Fig. 5A, B

). Tests of immunoprecipitation for

464

Serine motifs revealed pSer labelling for both FMRP and Kv4.3 under resting conditions of LK

465

medium (

Fig. 5C, D

). Exposure of cells to HK medium produced an increase in pSer labelling on

466

FMRP and a marked decrease in phosphorylation of Kv4.3 (

Fig. 5C, D

). Repeating tests for the

467

pSer motif of Kv4.3 further revealed that the reduction of Kv4.3 phosphorylation in HK medium

468

was prevented if cells were preincubated with either 1 µM TTA-P2 (

Fig. 5E

) or 30 µM BCI (

Fig.

469

). The increase in FMRP phosphorylation was also correlated to Cav3.1 calcium influx

470

although the exact mediator for phosphorylation is unknown since depolarization also lowers the

471

levels of pERK (

Fig. 3A

472

These data reveal that FMRP is phosphorylated at both pSer and pThr sites, and that

473

depolarization increases the pSer labelling. pSer labelling on Kv4.3 is instead reduced upon

474

depolarizing cells, a process that involves Cav3.1 activation of DUSP1/6 and its ability to

475

dephosphorylate ERK. The data thus suggest that Cav3.1 activation of DUSP1/6 and modulation

476

of FMRP and Kv4.3 phosphorylation enable KChIP3 to induce a depolarizing shift in Kv4.3 Vh

477

that increases I

in a calcium-dependent manner (cf.

Fig. 1A

478

479

Hyperexcitability of cerebellar granule cells in Fmr1 KO mice is rescued by FMRP-N-tat

480

Several aspects of excitability of cerebellar granule cells and their response to synaptic input

481

have been shown to depend on the Cav3.1-Kv4-KChIP3 complex (Heath et al., 2014; Rizwan et

482

al., 2016; Zhan et al., 2020). To further test the effects of FMRP-N-tat on excitability of neurons

483

we recorded from lobule 9 granule cells from P22 - 24 WT and

Fmr1

KO mice in cerebellar

484

slices prepared 1 day following tail vein injection of either vehicle or 1 mg/kg FMRP-N-

tat

485

Whole-cell recordings were used to deliver 1 sec current pulses to map the current-frequency

486

relationship of sodium spike discharge. In WT mice injected with vehicle alone, granule cells

487

fired with an initial frequency of 0.5 ± 0.54 Hz, showing a linear increase in firing with increments

488

of current injection up to 42.2 ± 8.7 Hz (n = 9) (

Fig. 6A, B

). The maximum firing frequency

489

reached during 16 pA current injection increased by 86% in

Fmr1

KO granule cells compared to

490

JNeurosci Accepted Manuscript

WT cells, further indicating a state of hyperexcitability (

Fig. 6C

). An increased sensitivity to

491

depolarizations was also apparent in a decrease in the rheobase to evoke spike firing in vehicle

492

injected

Fmr1

KO mice compared to WT mice (

Fig. 6D

). Finally, the increased spike frequency

493

was associated with a significant decrease in the decay tau (rate of repolarization) of evoked

494

spikes

(Fig. 6E

). By comparison there was no significant difference in the input resistance of

495

granule cells between WT and

Fmr1

KO mice (WT 1.05 ± 0.10 GΩ, n = 9;

Fmr1

KO 1.30 ± 0.17

496

GΩ, n = 7;

= 0.24, two-sample

-test). We then injected

Fmr1

KO mice with 1 mg/kg FMRP-N-

497

tat

and prepared tissue slices 24 hr later, a time previously shown to be sufficient for this

498

compound to distribute throughout the cerebellum for uptake by granule and Purkinje cells (Zhan

499

et al., 2020). Here we found complete rescue of the elevated firing rate of

Fmr1

KO compared to

500

WT granule cells over the full range of a current-frequency plot, with similar effects on rheobase

501

and rate of spike repolarization (

Fig. 6A-E

). These effects by FMRP-N-tat are important in that I

502

has previously been shown in granule cells to shape each of the parameters tested here (Heath

503

et al., 2014; Zhan et al., 2020).

504

505

An imbalance in expression levels of DUSP in Fmr1 KO cells is rescued by FMRP-N-tat

506

The function of the Cav3.1-DUSP-ERK signalling pathway here was defined primarily using WT

507

or KO tsA-201 cells. To explore the prevalence of DUSP6 expression

in situ

, we conducted

508

immunocytochemical labelling for DUSP6 in the cerebellum of P60 male mice. These tests

509

confirmed a widespread expression of this phosphatase in granule cells (

Fig. 7A

). The DUSP

510

family is interesting in establishing a reciprocal inhibitory system with pERK that will activate

511

nuclear transcription factors that modify DUSP levels (Rubinfeld and Seger, 2005; Cagnol and

512

Rivard, 2013; Caunt and Keyse, 2013). We used Western blot analysis to explore the potential

513

for a loss of FMRP to affect the levels of DUSP protein. This was first tested in WT and KO tsA-

514

201 cells exposed in culture medium to either vehicle alone or FMRP-N-tat (70 nM) for 7 hr (

Fig.

515

). These tests revealed a significant decrease in the level of DUSP6 in KO cells compared to

516

WT cells. However, the lower levels of DUSP6 in KO cells were rescued by pretreating cells with

517

70 nM FMRP-N-tat for 7 hr (

Fig. 7B

), a timeline that proved to overlap that required for FMRP-N-

518

tat to rescue Cav3.1 modulation of Kv4.3 in

Fmr1

KO cells (cf.

Fig. 1F

519

We then compared these results to the levels of DUSP detected in homogenates of whole

520

cerebellum from WT and

Fmr1

KO mice 24 hr following tail vein injection of either vehicle or

521

JNeurosci Accepted Manuscript

FMRP-N-tat (

Fig. 7C

). In this case we found that the level of DUSP6 in cerebellum was

522

significantly reduced in

Fmr1

KO mice compared to WT mice (

Fig. 7B

). Yet once again, tail vein

523

injection of

Fmr1

KO mice with 1 mg/kg FMRP-N-tat promoted a rescue of the level of DUSP6 in

524

the cerebellum when tested 24 hr later (

Fig. 7C

525

These data are important in identifying dual actions of FMRP-N-tat in reassociating with the

526

Cav3-Kv4 complex at the plasma membrane to restore calcium-dependent regulation (

Fig. 6

) as

527

well as restoring a disrupted level of DUSP (

Fig. 7

) that is critical to the function of the Cav3-Kv4

528

complex.

529

530

Discussion (1488 words)

531

Cav3 calcium channels form a nandomain association with a Kv4-KChIP complex to confer

532

calcium-dependent modulation of I

that can influence signal processing (Vierra and Trimmer,

533

2022). Despite our knowledge of the functional outputs of the Cav3-Kv4 complex, the molecular

534

mechanisms underlying calcium-dependent control of Kv4 channels had not been fully explored.

535

The current study identified a critical role for FMRP in enabling a new Cav3-DUSP signalling

536

pathway to regulate ERK-mediated phosphorylation in the complex. The physiological relevance

537

of FMRP to Cav3-Kv4 function was further shown by the rescue of a hyperexcitable state and

538

DUSP levels in cerebellar granule cells in the

Fmr1

KO mouse model for 24 hr by

in vivo

539

administration of FMRP-N-tat.

540

541

FMRP as a member of the Cav3-Kv4-KChIP complex

542

An association between Kv4 potassium channels and KChIP proteins as a potential calcium-

543

sensing element has long been recognized (An et al., 2000; Jerng et al., 2005; Pioletti et al.,

544

2006; Wang et al., 2007), followed by recognition of Cav3 calcium channels as a component of

545

the complex (Molineux et al., 2005). The role of FMRP as a critical element tightly associated

546

with this complex was identified through co-IP or FRET imaging. Thus, FMRP co-IPs with each

547

of Cav3.1, Kv4.3, and KChIP3 (

Figs. 2, 2-1

) and exhibits FRET with both Cav3.1 and Kv4.3

548

(Zhan et al., 2020). The FMRP-KChIP3 association proves to be unique in exhibiting a calcium-

549

dependent loss of co-IP (i.e. dissociation) (

Fig. 2

), a result not apparent for the co-IP between

550

FMRP and either Cav3.1 or Kv4.3 (

Figure 2-1

). The current work using a

Fmr1

KO tsA-201 cell

551

JNeurosci Accepted Manuscript

line provides evidence that modulation of Kv4.3 also involves Cav3.1 calcium-dependent

552

activation of DUSP to control the levels of pERK-mediated phosphorylation. It is interesting to

553

note that several studies have been carried out on how the Cav3-Kv4 complex shapes the

554

postsynaptic response of cerebellar stellate and granule cells (Molineux et al., 2005; Anderson

555

et al., 2010a, 2010b, 2013; Heath et al., 2014; Rizwan et al., 2016; Zhan et al., 2020). A

556

retrospective analysis indicates that the new findings of a role for FMRP in the Cav3-Kv4

557

complex do not conflict in any way with results of these studies. Rather, FMRP proves to have

558

been a hidden component that is central to the entire process by which Kv4 channels achieve

559

calcium-dependent regulation.

560

561

Calcium-dependent control of Kv4 and I

availability

562

Our available data supports a working model of actions that underlie calcium-dependent

563

regulation of Kv4.3 for the combination of FMRP, Cav3.1, Kv4.3 and KChIP3 (

Fig. 8

). Under

564

normal conditions FMRP is associated with each of the protein subunits of the Cav3.1-Kv4.3-

565

KChIP3 complex, with some resting level of phosphorylation of both Kv4.3 and FMRP at MAPK

566

consensus sites that maintains a relatively low amplitude I

(

Fig. 8A

). Membrane depolarization

567

induces Cav3.1-mediated calcium entry that activates DUSP1/6 to dephosphorylate ERK. The

568

decrease in pERK levels is then reflected in a reduction in phosphorylation of Kv4.3 along with

569

an increase in phosphorylation of FMRP. Associated with calcium influx is a dissociation of

570

FMRP-KChIP3, and a KChIP3-induced positive shift in Kv4.3 Vh that increases I

amplitude to

571

reduce membrane excitability (

Fig. 8B

). The converse effect of a hyperpolarizing shift in Kv4.3

572

Vh can be revealed by reducing Cav3.1 calcium conductance, decreasing [Ca]i, or by direct

573

intracellular infusion of pERK. In the

Fmr1

KO mouse cerebellum, the level of DUSP6 is

574

reduced, indicating a role for FMRP in regulating translation and expression of a phosphatase

575

important to Cav3-Kv4 function (

Fig. 8C

). The lack of FMRP further removes the ability to

576

phosphorylate FMRP and its dissociation from KChIP3, which together blocks all calcium-

577

dependent regulation of Kv4.3 Vh and I

amplitude (

Fig. 8C

). Thus it appears that FMRP is

578

required for KChIP3 to carry out its role of producing calcium-dependent shifts in Kv4.3 Vh. This

579

model helps account for the sensitivity of Kv4 channels to fluctuations of calcium conductance

580

that provides a bidirectional control over subunit phosphorylation by ERK and I

availability.

581

582

JNeurosci Accepted Manuscript

Distinct pathways to modulate I

and membrane excitability

583

It is well established that mGluR/NMDAR activation provides a source of calcium influx to

584

trigger the interplay between ERK, PP2A and S6K1 to modify the phosphorylation state of FMRP

585

(Bagni and Greenough, 2005; D’Incal et al., 2022). Bursts of mossy fiber input to stimulate

586

mGluR/NMDARs on granule cells was also shown to activate ERK that reduced I

to increase

587

spike output as part of mossy fiber LTP (Rizwan et al., 2016). Yet the current work reveals a new

588

Cav3.1 channel-activated signalling pathway that does not involve PP2A and instead

589

deactivates ERK to increase I

590

The data thus suggest the existence of two pathways that can separately employ ERK to

591

modulate I

availability. One is intrinsic to the cell in being triggered by Cav3.1 calcium influx to

592

activate DUSP and reduce pERK and phosphorylation of Kv4.3. This Cav3-dependent process

593

also depends on a dynamic change in an FMRP-KChIP3 association, with the net effect of

594

inducing a depolarizing shift of Kv4.3 Vh to increase I

and reduce spike frequency. The second

595

pathway requires an mGluR/NMDAR stimulus that activates ERK to induce a hyperpolarizing

596

shift in Kv4 Vh to decrease I

and promote a long term increase in excitability in a manner that is

597

also FMRP-dependent. At this time the exact means by which mGluR/NMDA activation results in

598

a pERK-mediated increase in excitability is unknown but is expected to involve both pre- and

599

postsynaptic mechanisms (Zhan et al., 2020). However, both the intrinsic and synaptically

600

evoked process depend on calcium-dependent modulation of Kv4 Vh that in turn relies on

601

FMRP.

602

603

Rescue of FMRP function in FXS

604

A number of strategies have been developed to reduce the symptoms of FXS through

605

pharmacology (Yamasue et al., 2019), reactivating FMRP expression (Xie et al., 2016; Graef et

606

al., 2019; Yrigollen and Davidson, 2019; Shah et al., 2023), or inducing FMRP expression

607

through delivery of viral constructs (Hampson et al., 2019). It was recently found that

608

reintroducing the N-terminus of FMRP as a tat conjugate peptide achieved rapid transport

609

across the BBB, rescuing circuit functions and alleviating symptoms of

Fmr1

KO mice (Zhan et

610

al., 2020). The current work now shows that a loss of FMRP and its actions in the Cav3-Kv4

611

complex increases the intrinsic excitability of cerebellar granule cells to promote a

612

hyperexcitable state of postsynaptic spike discharge. Tail vein injection of FMRP-N-tat fully

613

JNeurosci Accepted Manuscript

rescued granule cell excitability in terms of postsynaptic spike properties and the level of DUSP

614

even 24 hr after tail vein injection. It is striking that FMRP-N-tat is able to restore the actions of

615

FMRP both within an ion channel complex at the membrane while also modifying the expression

616

level of a phosphatase that is key to calcium-dependent modulation of I

. It is important to note

617

that the N-terminal region of FMRP has also been shown to restore GABAergic inhibition from

618

basket cells to Purkinje cells in

Fmr1

KO mice (Yang et al., 2018). Conditional expression of

619

FMRP in Purkinje cells of global

Fmr1

KO mice further restores Purkinje cell excitability

620

(Stoodley et al., 2017; Gibson et al., 2023). The potential thus exists for FMRP fragments to

621

exert therapeutic actions on several different parameters of cerebellar circuit dysfunction in FXS.

622

623

Cerebellar role in ASD in FXS

624

The behavioural and cognitive functions affected in FXS fall into the realm of ASD. The

625

cerebellum is highly implicated given its established role in higher levels of cognition

(D’Mello

626

and Stoodley, 2015; Stoodley et al., 2017; Mapelli et al., 2022; Gibson et al., 2023). Indeed, a

627

change in the output of cerebellar Purkinje cells in the right Crus I/II region has been shown to

628

account for a range of ASD symptoms that include hypersensitivity to sensory stimuli (Tsai et al.,

629

2012, 2018; D’Mello and Stoodley, 2015; Cupolillo et al., 2016; Peter et al., 2016; Stoodley et

630

al., 2017; Sundberg et al., 2018; Gibson et al., 2023). A loss of FMRP has also been shown to

631

alter the activity of granule (Zhan et al., 2020), Purkinje (Koekkoek et al., 2005; Gibson et al.,

632

2023; Martín et al., 2023), and basket cells (Yang et al., 2018). Our work was conducted on

633

granule cells in the vermis of lobule 9 in the posterior cerebellum, a region that has a role in

634

processing multiple sensory inputs and exhibits functional connectivity with limbic structures

635

(Sang et al., 2012; Markwalter et al., 2019). To our knowledge a direct link between granule cell

636

activity and ASD symptoms in FXS has not yet been defined as for Purkinje cells and ASD. Yet,

637

a change in the intrinsic excitability of granule cells will alter signal processing at the first stage

638

of mossy fiber input to the cerebellum that will affect cerebellar cortical activity, and finally the

639

output to neocortical regions that are predicted to contribute to ASD.

640

641

Author Contributions

. X.Z. and R.W.T. designed the experiments, X.Z. acquired and analysed

642

electrophysiology and protein biochemical data, P.P. designed and provided KChIP3 EF hand

643

mutant cDNA, and H.A. contributed protein biochemistry. R.W.T. and X.Z. supervised the study

644

JNeurosci Accepted Manuscript

Figure Legends

649

650

Figure 1.

Cav3.1-induced shifts in Kv4.3 Vh are bidirectional and depend on the

651

expression of FMRP.

Shown in this and subsequent figures are mean voltage-inactivation or

652

voltage-activation plots for Kv4.3 in

Fmr1

WT cells that express FMRP or in

Fmr1

KO cells in

653

which FMRP was knocked out by CRISPR-Cas9. The proteins coexpressed in each test are

654

indicated above the plots.

Insets

in (

A, B

) show superimposed records of Kv4.3 current evoked

655

by a step from -40 mV to -30 mV before and after the indicated tests.

Raising external

656

calcium from 0 to 1.5 mM induces a positive shift of Kv4.3 Vh (

arrow

) and an increase in Kv4.3

657

current amplitude that does not occur in the absence of Cav3.1 expression.

, Blocking Cav3.1

658

conductance with 1 µM TTA-P2 induces a negative shift of Kv4.3 Vh (

arrow

), but does not affect

659

the steady state activation of Kv4.3.

Mean bar plots comparing the effects of three Cav3

660

channel blockers (1 µM mibefradil, 300 µM Ni

, 1 µM TTA-P2) establish the ability to selectively

661

induce a negative shift of Kv4.3 Vh but not Va when applied in a bathing medium containing 1.5

662

mM calcium.

Fmr1

KO cells TTA-P2 (1 µM) fails to shift Kv4.3 Vh.

E, F,

A TTA-P2-induced

663

negative shift in Kv4.3 Vh is not enabled in

Fmr1

KO cells by including 3 nM FMRP(1-297) in the

664

pipette (

) but is produced by long term pre-incubation (4 - 8 hr) with 70 nM FMRP-N-

tat

(

665

Average values are mean ± S.E.M. with sample values (n) indicated on plots. n.s., not

666

significant, **

< 0.01, paired sample

-test. See also

Figure 1-1, 1-2, 1-3

667

668

Figure 2.

KChIP3 and FMRP are required to enable a calcium-dependent shift in Kv4 Vh.

669

Shown are the effects of TTA-P2 application (

A-D

) on

Fmr1

WT cells coexpressing the indicated

670

subunits.

A, B,

In the absence of KChIP3 coexpression TTA-P2 has only a small effect on Kv4.3

671

Vh (

A, B

). Coexpressing KChIP3 restores a larger TTA-P2-induced negative shift in Kv4.3 Vh

672

that is lost if preincubated cells with a

tat

-pp1 peptide that interferes with the KChIP3-Kv4.3

673

binding site (

C, D,

Mean bar plots showing a TTA-P2-induced shift in Kv4.3 Vh is partially

674

blocked in cells expressing either EF hand mutant, and fully blocked upon expression of a dual

675

KChIP3 EF hand mutant.

Western blot and mean bar plots showing co-IP between KChIP3

676

and FMRP in

Fmr1

WT cells expressing all subunits of the Cav3-Kv4 complex. The KChIP3-

677

FMRP co-IP detected in the presence of 1.5 mM [K]o (LK) is lost in the presence of 50 mM [K]o

678

(HK) medium to promote Cav3.1 calcium influx.

Coexpressing Kv4.3-KChIP3 in the absence

679

JNeurosci Accepted Manuscript

of Cav3.1 has no effect on Kv4.3 Vh in

Fmr1

WT or

Fmr1

KO cells (0 µM internal free calcium,

680

Maxchelator).

Coexpressing Kv4.3-KChIP3 in the absence of Cav3.1 reveals that recording in

681

the presence of 13 µM calcium in the electrode induces a positive shift in Kv4 Vh in

Fmr1

682

cells but not in

Fmr1

KO cells (

). BC, background control. Average values are mean ± S.E.M.

683

with sample sizes (n) indicated in brackets. n.s., not significant, *

< 0.05, **

< 0.01, ***

684

0.001; Paired-sample

-test (

C, D

) and Two-sample

-test (

B, F, G

). See also

Figure 2-1.

685

686

Figure 3. A Cav3.1-dependent reduction in phosphorylated ERK is impaired in Fmr1 KO

687

cells

. Shown are the levels of phosphorylated ERK1/2 (pERK) in relation to total ERK1/2 (tERK)

688

at rest and upon depolarization in tsA-201 cells expressing Cav3.1-Kv4.3-KChIP3. pERK levels

689

are quantified by Western blot and mean bar plots in the presence of 1.5 mM [K]o (LK) or 50 mM

690

[K]o (HK) medium.

WT cells a resting level of pERK is detected in LK medium. The pERK /

691

tERK ratio is reduced over 10 min upon depolarization in HK medium.

The decrease in pERK

692

/ tERK with HK-induced depolarization is prevented if Cav3.1 conductance is blocked by pre-

693

incubation in 1 µM TTA-P2.

The HK-induced reduction of pERK levels is absent in

Fmr1

694

cells. All cells in (

A, B, C

) were also transfected with 0.7µg Kir2.1 to maintain a negative resting

695

potential. Average values are mean ± S.E.M. with sample sizes (n) indicated in bar plots. n.s.,

696

not significant, ***

< 0.001; Student’s paired sample

-test.

697

698

Figure 4. Cav3.1 calcium influx activates DUSP to regulate pERK actions on Kv4.3 Vh.

699

Shown are results from whole-cell recordings and Western blots from

Fmr1

WT or KO cells

700

expressing the indicated subunits.

A, B,

The effects of direct pERK infusion through the

701

electrode in

Fmr1

WT (

) or

Fmr1

KO cells (

) with Cav3.1-Kv4.3-KChIP3 subunits

702

coexpressed. With 1 µM TTA-P2 in the bath to block T-type current, direct infusion of pERK

703

results in a negative shift in Kv4 Vh in both

Fmr1

WT and KO cells. However, pERK has no

704

effect in the absence of TTA-P2 to retain Cav3.1 conductance (

C-E,

The effects of

705

phosphatase blockers on Kv4.3 Vh. Bath application of 1 µM TTA-P2 induces a negative shift in

706

Kv4.3 Vh in the presence of 3 nM okadaic acid to block PP1/PP2A phosphatases (

), but not in

707

the presence of 30 µM BCI to block DUSP1/6 (

). Recordings in the presence of 30 µM BCI also

708

exhibit a negative shift in Kv4.3 Vh (

Western blots indicating that BCI (30 µM) in the bath

709

blocks a reduction of pERK normally stimulated by a high [K]o-mediated depolarization (HK) in

710

JNeurosci Accepted Manuscript

Fmr1

WT cells expressing the complex (cf.

Fig. 3A

The effects of blocking DUSP1/6 on

711

Kv4.3 Vh in the absence of Cav3.1 expression and direct manipulation of internal calcium

712

concentration in the electrode. Recording with a high level of buffered internal calcium (13 µM,

713

MaxChelator) in the electrode produces a positive shift in Kv4.3 Vh that is blocked in the

714

presence of 30 µM BCI in the bath to block DUSP1/6. Average values are mean ± S.E.M. with

715

sample sizes (n) indicated in bar plots. n.s., not significant, *

< 0.05; ***

< 0.001; Student’s

716

paired sample

-test (

A, C-F

) and Student’s two sample

-test (

B, G

717

718

719

Figure 5. Cav3.1 calcium influx reduces Kv4.3 phosphorylation through a DUSP-ERK

720

pathway.

721

Shown are the effects of testing

Fmr1

WT cells coexpressing Cav3.1-Kv4.3-KChIP3 in either low

722

[K]o (LK) or high [K]o (HK) on phosphorylation of the indicated subunits. Results are shown in

723

Western blots (

A, C, E

) and mean bar plots (

B, D, F

) of the indicated proteins after

724

immunoprecipitation with different ERK substrate antibodies.

An anti-PXpTP motif (pThr)

725

antibody detects phosphorylation of FMRP (

), but the level of phosphorylated FMRP is not

726

affected by HK treatment (

C, D,

An anti PXS*P or S*PXR/K motif (pSer) antibody detects

727

phosphorylation of both Kv4.3 and FMRP (

), with HK stimulation increasing the level of pSer on

728

FMRP and decreasing the level on Kv4.3 (

Representative western blots (

) and mean

729

bar plots of the extent of phosphorylation at pSer sites in LK vs HK conditions (

). Pretreating

730

cells with 1 µM TTA-P2 or 30 µM BCI blocks the HK-induced reduction of Kv4.3 phosphorylation.

731

Abbreviations: LK, low potassium; HK, high potassium; BC, background control; IP,

732

immunoprecipitation; IB, immunoblotting. Average values are mean ± S.E.M. with sample sizes

733

(n) indicated in bar plots. n.s., not significant, *

< 0.05; Student’s paired sample t-test.

734

735

Figure 6. FMRP regulates excitability in cerebellar granule cells.

The effects on spike

736

output of cerebellar granule cells in tissue slices of P22 -24 male WT and

Fmr1

KO mice 24 hr

737

following tail vein injection of vehicle or 1.0 mg/kg FMRP-N-tat.

, Representative whole-cell

738

recordings indicate that tonic firing frequency in response to current injection is markedly

739

elevated in vehicle injected

Fmr1

KO mice compared to WT mice. Firing frequency is rescued in

740

JNeurosci Accepted Manuscript

granule cells of

Fmr1

KO mice 24 hr after FMRP-N-tat injection.

Mean frequency-current (F/I)

741

plots of spike output for 1 sec long current injections.

, Bar plots of mean maximal firing

742

frequency measured from the first 4 spike outputs with 16 pA current injection.

, Bar plots of

743

rheobase current required to reach spike firing threshold.

, Bar plots of mean decay tau of

744

spike output as a measure of the rate of repolarization with the rheobase current injection

745

required to trigger spike firing. Average values are mean ± S.E.M. with sample sizes (n)

746

indicated in brackets. n.s.; not significant; *

< 0.05, **

< 0.01, and ***

< 0.001 (

C, D, E

); #

747

< 0.05 for

Fmr1

KO vehicle vs FMRP-N-tat injected firing rate (

); Student’s two sample

-test.

748

749

Figure 7. DUSP6 expression in cerebellum is disrupted in Fmr1 KO cells but rescued by

750

reintroducing FMRP-N-tat

Low power immunolabel images for MAP-2 as a structural label

751

and DUSP6 in tissue sections of P60 WT mouse lobule 9 cerebellum showing DUSP6

752

expression in granule cells.

B, C,

Two representative Western blots of the relative expression

753

levels of DUSP6 in homogenates of

Fmr1

WT and KO tsA-201 cells (

) and P24-26 mouse

754

cerebellar WT and

Fmr1

KO tissue (

). Protein levels are normalized to levels of GAPDH (

)

755

and

𝛂

-tubulin (

). DUSP6 levels are lower in tsA-201

Fmr1

KO cells (

) and

Fmr1

756

cerebellum compared to WT (

). Direct application of 70 nM FMRP-N-tat in tsA-201 culture cell

757

medium restores the level of DUSP6 (

), as found in

Fmr1

KO cerebellar tissue 24 hr after tail

758

vein injection of 1.0 mg/kg FMRP-N-tat (

). Average values are mean ± S.E.M. with sample

759

sizes (n) indicated in bar plots. n.s., not significant, *

< 0.05, **

< 0.01. Student’s two sample

760

-test. Mol, molecular layer; PCL, Purkinje cell layer; GCL, granule cell layer. Scale bars (

), 50

761

µm.

762

763

764

Figure 8.

Model of FMRP regulation of calcium-dependent modulation of Kv4 channels

765

via a DUSP-ERK pathway.

Interactions shown are relevant to mouse cerebellar lobule 9

766

granule cells where Cav3.1, Kv4.3, KChIP3 and FMRP form a tight association at the plasma

767

membrane.

A-C,

Shown are interactions found at rest and in response to depolarization in WT

768

cells (

A, B

), and the effects of depolarization in the case of a loss of FMRP in FXS (

Under

769

resting conditions, a low Cav3.1 conductance provides little activation of DUSP1/6 that leads to

770

an elevation of pERK. Both FMRP and Kv4.3 exhibit phosphorylation at MAPK consensus

771

JNeurosci Accepted Manuscript

sequence sites, promoting a hyperpolarizing shift in Kv4.3 Vh (

white arrow

) that reduces IA

772

amplitude.

Upon membrane depolarization Cav3.1 calcium conductance increases, activating

773

DUSP1/6 to reduce levels of pERK and MAPK phosphorylation of Kv4.3 at a pSer site, and an

774

increase in phosphorylation at a pSer site on FMRP. Accompanying this is a reduction of the

775

FMRP-KChIP3 association (

white arrow

) and a depolarizing shift in Kv4.3 Vh (

red filled arrow

)

776

that increases IA amplitude.

In Fragile X Syndrome a loss of FMRP from the complex and its

777

interactions with KChIP3 prevents all calcium-dependent regulation of Kv4.3 Vh and IA

778

amplitude (double arrows marked by X). In the

Fmr1

KO cerebellum the level of DUSP6 is

779

reduced but rescued along with function of the Cav3-Kv4 complex upon reintroducing the

780

FMRP(1-297) fragment as a tat-conjugated peptide.

781

782

783

784

785

786

787

788

789

JNeurosci Accepted Manuscript