Action potential broadening in a presynaptic channelopathy

Abstract

Brain development and interictal function are unaffected in many paroxysmal neurological channelopathies, possibly explained by homoeostatic plasticity of synaptic transmission. Episodic ataxia type 1 is caused by missense mutations of the potassium channel Kv1.1, which is abundantly expressed in the terminals of cerebellar basket cells. Presynaptic action potentials of small inhibitory terminals have not been characterized, and it is not known whether developmental plasticity compensates for the effects of Kv1.1 dysfunction. Here we use visually targeted patch-clamp recordings from basket cell terminals of mice harbouring an ataxia-associated mutation and their wild-type littermates. Presynaptic spikes are followed by a pronounced afterdepolarization, and are broadened by pharmacological blockade of Kv1.1 or by a dominant ataxia-associated mutation. Somatic recordings fail to detect such changes. Spike broadening leads to increased Ca²⁺ influx and GABA release, and decreased spontaneous Purkinje cell firing. We find no evidence for developmental compensation for inherited Kv1.1 dysfunction.

ARTICLE

Received 16 Oct 2015 | Accepted 27 May 2016 | Published 6 Jul 2016

Brain development and interictal function are unaffected in many paroxysmal neurological channelopathies, possibly explained by homoeostatic plasticity of synaptic transmission. Episodic ataxia type 1 is caused by missense mutations of the potassium channel Kv1.1, which is abundantly expressed in the terminals of cerebellar basket cells. Presynaptic action potentials of small inhibitory terminals have not been characterized, and it is not known whether developmental plasticity compensates for the effects of Kv1.1 dysfunction. Here we use visually targeted patch-clamp recordings from basket cell terminals of mice harbouring an ataxia-associated mutation and their wild-type littermates. Presynaptic spikes are followed by a pronounced afterdepolarization, and are broadened by pharmacological blockade of Kv1.1 or by a dominant ataxia-associated mutation. Somatic recordings fail to detect such changes. Spike broadening leads to increased Ca2 inux and GABA release, and decreased spontaneous Purkinje cell ring. We nd no evidence for developmental compensation for inherited Kv1.1 dysfunction.

DOI: 10.1038/ncomms12102 OPEN

Action potential broadening in a presynaptic channelopathy

Rahima Begum1, Yamina Bakiri1, Kirill E. Volynski1 & Dimitri M. Kullmann1

1 UCL Institute of Neurology, University College London, Queen Square, London WC1N 3BG, UK. Correspondence and requests for materials should be addressed to D.M.K. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 7:12102 | DOI: 10.1038/ncomms12102 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12102

Kv1 channels are mainly expressed in axons throughout the central nervous system and play important roles in action potential initiation and repolarization13. Several

heterozygous mutations in the KCNA1 gene, which encodes the Kv1.1 subunit, have been identied in episodic ataxia type 1 (EA1), a familial disorder characterized by paroxysmal cerebellar incoordination and interictal myokymia, and occasionally with other features such as epilepsy48. Consistent with the prominent features of the syndrome, Kv1.1 is most abundantly expressed in the terminals of cerebellar basket cells912.

Heterologous expression studies have shown that mutations disrupt channel function through a variety of changes in assembly, trafcking and kinetics, often with dominant negative effects1315. The consequences depend on the channel composition, as Kv1.1 co-assembles with other members of the Kv1 family and with beta subunits, and it is therefore difcult to predict the consequence of a given mutation in situ from heterologous expression studies. Indeed, some mutations may interfere with fast inactivation, offsetting other loss-of-function effects16,17. A mouse knock-in model of EA1 (Kcna1V408A/ )

bypasses these limitations of heterologous expression, and exhibits impaired cerebellar coordination when stressed18. Importantly, and unlike most mouse models of human genetic disease, the disease phenotype is recapitulated in the heterozygous state18.

How does Kv1.1 channel dysfunction affect synaptic and circuit function, leading to cerebellar incoordination? In the neocortex, Kv1 channels play a key role in action potential repolarization1, and also determine the latency to the rst action potential in fast-spiking interneurons2,19. Little is known, however, of the consequences of Kv1.1 mutations in the cerebellar cortex. A study using targeted patch-clamp recordings showed that a-dendrotoxin (a-DTx), a blocker of

Kv1.1 and Kv1.2 channels20, attenuates K currents in voltage-clamp recordings from basket cell terminals, where both channel subunits are highly concentrated21. However, the consequences for action potential shape are unknown, and current-clamp recordings have not been reported. Although a-DTx increased spontaneous inhibitory postsynaptic currents (IPSCs) in Purkinje cells22, in another study it had no effect on action potential-dependent Ca2 uorescence transients measured from basket cell boutons23. Moreover, it is difcult to extrapolate from acute pharmacological blockade of Kv1.1 and Kv1.2 channels to constitutive dysfunction of Kv1.1 with dominant negative effects that depend on channel stoichiometry1317. The Kcna1V408A/ mouse shows an increase in spontaneous IPSC frequency in Purkinje cells relative to wild-type littermates18, but how this relates to action potential-dependent GABA release from basket cell terminals is unclear. Furthermore, abundant evidence exists for developmental homoeostatic compensation for pharmacological manipulation of ion channels and receptors24. Do similar compensatory mechanisms occur in response to inherited mutations of presynaptic ion channels? To address these questions, we rst characterized presynaptic action potentials at cerebellar basket cell terminals of wild-type mice, and then compared the effects of pharmacological and genetic manipulations of Kv1.1 on presynaptic spike shape, action potential Ca2 inux, and GABAergic inhibition of Purkinje cells.

ResultsPresynaptic spikes are broadened by Kv1.1 blockade. To understand the normal role of Kv1.1 in basket cells we examined the effect of the highly specic Kv1.1 blocker DTx-K25 on action potential shape. Targeted patch-clamp recordings were

obtained from somata or terminals of basket cells in acute cerebellar slices from wild-type mice, bred on the same background as Kcna1V408A/ mice described below18.

Somatic recordings from basket cells showed no effect of 200 nM DTx-K on action potential width (Fig. 1a,b). A likely explanation is that Kv1.1 is predominantly expressed in axons and terminals of basket cells9,11, and plays only a small role in somata. Patch-clamp recordings from basket cells have indeed detected potassium currents with biophysical and pharmacological properties consistent with Kv1.1 or Kv1.2 (refs 21,26). We therefore targeted basket cell terminals for patch-clamp recordings.

Presynaptic terminals were initially identied under infrared differential interference contrast (DIC), and had a high input resistance, little or no sag potential, and were able to spike upon 1-ms current injection (passive membrane properties are given in Supplementary Table 1). Prolonged depolarizing current pulses failed to evoke stable trains of action potentials (see also Supplementary Fig. 1a). Instead, repeated brief pulses reliably triggered action potentials at either 55 or 100 Hz (see also Supplementary Fig. 1b,c), with modest progressive spike broadening. Evoked presynaptic action potentials were followed by a prominent afterdepolarization (ADP), which reversed between 50 and 40 mV (see also Supplementary Fig. 2a,b). Because GABAA autoreceptors are prominent in another type of cerebellar interneurons, stellate cells27, we asked whether the ADP was abolished by blocking these receptors. Bath application of picrotoxin (PTx, 100 mM), however, only led to a small albeit signicant decrease in the area of the ADP (Supplementary Fig. 2c,d), suggesting that GABA release from basket cell terminals acting on autoreceptors plays only a small role.

In contrast to somatic action potentials, presynaptic spike width (measured at 30 mV) was robustly prolonged by DTx-K (Fig. 1a,b; baseline: 0.960.03 ms; DTx-K: 1.150.02 ms; Po0.001, paired t-test, n 11). The ADP was however

unaffected. We compared the effects of DTx-K to those of charybdotoxin (ChTx), a blocker of the large conductance calcium- and voltage-activated potassium channel BKCa, which

is usually abundantly expressed at presynaptic terminals in close association with the active zone28 (but see ref. 29). Unexpectedly, ChTx (100 nM) led to spike broadening at the soma (Fig. 1c: baseline: 1.280.10 ms; ChTx: 1.550.13 ms; Po0.05, paired t-test, n 7) but not in presynaptic boutons. Thus, two K

channels show complementary roles in determining spike shape in basket cells.

Kv1.1 modulates presynaptic Ca2 inux and GABA release. A previous study reported no effect of a-DTx on action potential-evoked Ca2 uorescence transients measured in basket cell terminals23. We re-examined the role of Kv1.1 in presynaptic spike-evoked Ca2 inux by patch clamping basket cell bodies in wild-type mice. Two-photon uorescence excitation microscopy was used to image individual boutons apposed to Purkinje cell somata (Fig. 2a). Line scans were taken before and after bath perfusion of 200 nM DTx-K (Fig. 2b,c, see Methods for detailed protocols and calibration of Ca2 responses). To improve the signal-to-noise ratio, we took the integral of the Fluo-4 uorescence signal for 200 ms from the rst action potential, as a measure of total action potential-evoked Ca2 inux D[Ca2 ].

A non-stationary single-compartment model30 incorporating the Ca2 buffer parvalbumin provided a good t to the uorescence transients (Fig. 2c), conrmed that the 200 ms integral varied linearly with D[Ca2 ] (see also Supplementary Fig. 3), and further yielded an estimate of the absolute Ca2 concentration change.

2 NATURE COMMUNICATIONS | 7:12102 | DOI: 10.1038/ncomms12102 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12102 ARTICLE

a

G

P

Soma Bouton

Control

DTx-K

30 mV

30 mV

70 mV 70 mV

b

Soma

Bouton

1.5 n = 7

n = 11

***

AP width at 30 mV (ms)

1.2

0.9

0.6

0.3

0

+

DTx-K +

c

Soma Bouton

+ +

2.5

ChTx

n = 7 n = 4

*

AP width at 30 mV (ms)

2.0

1.5

1.0

0.5

0.0

Figure 1 | Compartment-specic roles of K channels in action potential shape in cerebellar basket cells. (a) Representative action potentials recorded from the soma (left) or bouton (right) of cerebellar basket cells, before (black trace) and after (green trace) Kv1.1 blockade with DTx-K. Top: schematics illustrating recording sites. Inset: action potentials at slow time base, showing a prominent ADP at the bouton. Scale bar, 20 mV; 2 ms (main traces); 40 mV, 200 ms (insets). The uorescence image at right shows a cerebellar basket cell bouton labelled with Alexa 568. G and P indicate granule cell and Purkinje cell layer, respectively. Scale bar, 10 mm. (b) Summary data showing selective effect of DTx-K on presynaptic spike width measured at 30 mV. Circles show individual experiments. (c) BKCa blockade with ChTx led to somatic action potential broadening. *Po0.05, ***Po0.001; paired t-tests.

DTx-K perfusion led to a 219% increase in the normalized Ca2 uorescence integral (Po0.05, Wilcoxon signed-rank test for paired data), which was not seen in control experiments followed for the same time (Fig. 2d). Dividing the estimated total Ca2 concentration change (D[Ca2] B10 mM per action potential under baseline conditions) into the approximate bouton volume (estimated from DIC or Alexa images as roughly 1 fL), we further estimated that B6.0 106 Ca2 ions (equivalent to a

charge of 1.93 fC) enter the bouton for each action potential under baseline conditions. A six-state kinetic model of P/Q-type Ca2 channels31, which predominate in basket cells32, yields an estimate of their gating kinetics when driven by a presynaptic action potential waveform that incorporates the ADP recorded in boutons. Taking into account the single channel conductance and driving force30, we estimate that B 0.04 fC enters via each Ca2 channel. This yields an estimate of B50 P/Q-type Ca2 channels present in a typical bouton. Although the driving force for Ca2 entry increases following repolarization, we found that subtracting the ADP from the action potential waveform

actually led to a B3% decrease in the total Ca2 inux (Supplementary Fig. 4), consistent with accelerated deactivation.

In contrast, prolonging the decay phase of the action potential led to a linear increase in Ca2 inux over a wide range (Fig. 2e).

This gives an independent estimate of the effect of spike broadening due to DTx-K, corresponding to an 18% increase in Ca2 inux.

Two independent approaches (Ca2 uorescence imaging and kinetic modelling of P/Q-type channels), thus converge on the conclusion that Kv1.1 channel blockade causes B20% more

Ca2 inux per action potential.

Previous studies using a-DTx revealed a large increase in spontaneous IPSC amplitude and frequency in Purkinje cells22,23. We asked how DTx-K affects evoked IPSCs by recording from Purkinje cells, while activating axons in the Purkinje cell layer. Kv1.1 blockade with DTx-K led to a 4519% increase in pharmacologically isolated monosynaptic IPSCs (n 15, Po0.05, Wilcoxon matched pairs signed-rank test;

Supplementary Fig. 5). This is consistent with a B20%

NATURE COMMUNICATIONS | 7:12102 | DOI: 10.1038/ncomms12102 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12102

a

M G

P

b Before DTx-K After DTx-K

c

[Ca2+] = 9.9 M

[Ca2+] = 10.8 M

Int-F = 22.0103 RFU ms

Int-F = 24.5103 RFU ms

Fmax = 465

Fmax = 538

F = 0

F = 0

d

e

C0 C1

[afii9825]1(V)

[afii9826]1(V)

[afii9825][notdef]C2

2(V) [afii9826]2(V)

[afii9825][notdef]C3

3(V) [afii9826]3(V)

[afii9825][notdef]C4

4(V) [afii9826]4(V)

*

[afii9825]

[afii9826]

O

1.5

2.5

0.8 1 1.2

Normalized int-F

1

30 mV

I Ca(pA ms)

2.0

n = 6 n = 6

0.5

1.5

ICa

0

1.0

DTx-K

+

AP width at 30 mV (ms)

Figure 2 | Kv1.1 modulates presynaptic action potential-evoked Ca2 inux and GABA release. (a) Cerebellar basket cell lled with Alexa 594, under multi-photon uorescence microscopy. M, P and G indicate molecular, Purkinje cell and granule cell layers, respectively. Middle: axon superimposed on transmitted light image, showing boutons apposed to the soma of a Purkinje cell. Inset (expanded at right) shows imaged bouton and line scan position (white dashed line). Scale bar, (left) 20 mm; (middle) 5 mm; (right) 2 mm. (b) Representative line scans (averages of four trials), showing Fluo-4 uorescence response to a train of four action potentials at 40 Hz (arrows) elicited at the soma before (left) and 10 min after DTx-K perfusion (right). Scale bar, 75 ms. (c) Fluo-4 uorescence time courses (same bouton as in b) elicited by four action potentials at 40 Hz followed by 50 action potentials at 100 Hz to saturate Fluo-4 (Scale bar, 200 relative uorescence units (RFU); 200 ms). Insets: zoomed responses to four action potentials (Scale bar, 50 RFU; 100 ms). Black and green lines represent non-stationary single-compartment model ts before and after application of DTx-K, respectively. Model-predicted values for the total action potential-evoked Ca2 inux (D[Ca2 ]) are shown next to each trace. Shaded areas under the traces indicate the Fluo-4 uorescence integrated over 200 ms from the beginning of stimulation (Int-DF), which is proportional to D[Ca2 ] (Supplementary Fig. 3). (d) Summary of the normalized effects of DTx-K on Int-DF, compared with control experiments where DTx-K was not applied. *Po0.05, Wilcoxon signed-rank test for paired data. (e) Top, gating model for P/Q-type presynaptic Ca2 channels31. Left, average time course of Ca2 current through a single P/Q-type channel (10,000 simulations including failures) elicited by action potentials of different widths. Modied action potential waveforms (grey) were generated by scaling the repolarization phase of the experimental control trace (black) between 0.9 and 1.4. Right, calculated dependence of evoked Ca2 current integral (ICa2) on action potential width measured at 30 mV.

increase in Ca2 inux, assuming a Ca2 current cooperativity (m) of B2 (refs 33,34).

Thus, delayed repolarization secondary to Kv1.1 blockade leads to increased Ca2 inux and enhanced neurotransmitter release.

How does the acute effect of manipulating Kv1.1 channels compare to genetic disruption of Kv1.1 in EA1?

Increased spike width and GABA release in Kcna1V408A/ mice. We repeated action potential recordings from basket cells in Kcna1V408A/ mice, which harbour a missense mutation that underlies EA1 (ref. 18), and their wild-type littermates (Kcna1 / ). Data were acquired and analysed blind to genotype.

Passive membrane properties and current threshold for eliciting

4 NATURE COMMUNICATIONS | 7:12102 | DOI: 10.1038/ncomms12102 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12102 ARTICLE

a

Soma

WT Kcna1V408A/+

Bouton

30 mV

30 mV

70 mV

70 mV

b

1.5

WT Kcna1V408A/+

**

1.2

AP width at 30 mV (ms)

0.9

0.6

Soma Soma

n = 20 n = 13

Bouton Bouton

0.3

n = 10

n = 6

0

Figure 3 | Action potentials recorded in cerebellar basket cells of Kcna1V408A/ mice. (a) Superimposed representative action potentials recorded from somata (left) or boutons (right) of cerebellar basket cells from wild-type or Kcna1V408A/ mice. Scale bar, 10 mV; 1 ms.

(b) Presynaptic spike width at 30 mV was longer in Kcna1V408A/ boutons. **Po0.01, unpaired t-test.

action potentials were unaffected by the mutation (Supplementary Table 1). Recordings of somatic action potentials also failed to reveal a signicant difference in action potential width between genotypes (Fig. 3a). A robust difference in duration was however observed in presynaptic boutons. Action potential width was B35% greater in Kcna1V408A/ mice(1.260.08 ms, n 13) than in Kcna1 / mice (0.930.03 ms,

n 20; unpaired t-test: Po0.002; Fig. 3a,b). Trains of action

potentials elicited at 55 and 100 Hz showed a similar spike broadening as in wild-type mice (Supplementary Fig. 1c), showing no evidence of occlusion between the effects of repetitive spiking and genotype. DTx-K failed to broaden action potential duration recorded from the presynaptic terminals of basket cells in Kcna1V408A/ mice, consistent with the loss of function of

Kv1.1-containing channels (Supplementary Fig. 6). We also observed no difference in the ADP between the genotypes.

The Kcna1V408A/ mutation thus broadens the spike at least as much as acute application of DTx-K in wild-type boutons. We observed no evidence of homoeostatic compensation by other channels correcting for loss of Kv1 channel function in the knock-in model of EA1.

Increased inhibition of Purkinje cells in Kcna1V408A/ mice. An increase in spontaneous GABAergic IPSCs has previously been reported in Purkinje cells of the Kcna1V408A/ mouse18. It is technically difcult to compare presynaptic Ca2 inux between genotypes, and the amplitudes of evoked IPSCs are uninformative when evoked by extracellular stimulation of multiple axons. Instead, we compared the paired-pulse ratio of

IPSCs recorded in Purkinje cells as a surrogate measure of action potential-evoked neurotransmitter release probability. Paired-pulse ratio was signicantly lower in Kcna1V408A/ mice(0.490.06, n 6) than in Kcna1 / mice (0.770.07, n 7;

Po0.01, unpaired t-test; Fig. 4a,b), consistent with an increased release probability. DTx-K had no signicant effect on IPSC amplitude in Kcna1V408A/ mice (45%, n 5). To examine the

downstream consequences for Purkinje cells, we compared their spontaneous activity between Kcna1V408A/ and Kcna1/ littermates, using cell-attached recordings to minimize disruption of intrinsic excitability. All Purkinje cells irrespective of genotype exhibited periods of tonic ring alternating with periods of quiescence or burst ring (Fig. 4c), as previously reported in wild-type mice and rats35. Overall, spike frequency recorded in Kcna1V408A/ mice was signicantly lower (55.410.7 Hz, n 13) compared with Kcna1 / littermates (96.312.1 Hz,

n 20, Po0.02, MannWhitney U test). The average inter-spike

interval during tonic ring was signicantly longer in Kcna1V408A/ mice (20.62.8 ms, n 19) than in Kcna1 /

mice (10.91.3 ms, n 20; Po0.01, MannWhitney U test;

Fig. 4d,e). The number of spikes per burst was also lower in mutant mice (10.93.3, n 5) than in wild-type mice (27.93.0,

n 11; Po0.01, unpaired t-test; Fig. 4f). No signicant difference

was observed in burst duration or the inter-burst interval between the genotypes. There was also no difference in action potential width in Purkinje cells between the genotypes, as estimated by integrating the cell-attached recordings36 (Supplementary Fig. 7).

Finally, we asked if enhanced spontaneous GABA release contributed to the lower activity of Purkinje cells. Bath application of blockers of GABAA and GABAB receptors (100 mM PTx and 1 mM CGP 52432, respectively) led to a greater decrease in inter-spike interval in Kcna1V408A/ mice(33.74.6%, n 10) than in Kcna1 / mice (17.14.1%,

n 8; Po0.05, unpaired t-test; Fig. 4g,h). Indeed, the inter-spike

interval in the presence of GABA receptor blockers was not signicantly different between wild-type and Kcna1V408A/ mice(11.31.9 ms and 15.62.1 ms; P 0.36, unpaired t-test). We

thus conclude that enhanced GABA release indeed contributes to decreased spontaneous activity of Purkinje cells, with no evidence for a homoeostatic compensation in the EA1 mouse model.

DiscussionThe present study reveals a major role of Kv1.1 channels in action potential repolarization at basket cell terminals, which was not apparent when recording from somata. We provide a quantitative account of presynaptic spike-evoked Ca2 transients. EA1 modelled in heterozygous knock-in mice led to a similar effect on spike shape and neurotransmitter release, as acute pharmacological blockade of Kv1.1 channels in wild-type mice. GABAergic inhibition of Purkinje cell ring was enhanced in Kcna1V408A/ mice, with no evidence of homoeostatic compensation for this presynaptic channelopathy.

Central to this study is the ability to target small GABAergic presynaptic terminals for whole-cell patch-clamp recordings. This has only previously been achieved in voltage-clamp mode21. Presynaptic recordings have yielded invaluable insights into mechanisms linking action potentials and K channels to neurotransmitter release, but hitherto such studies have been restricted to giant calyceal synapses in the auditory brainstem37 or large hippocampal or cerebellar mossy bre boutons38,39. Small boutons can also be targeted for whole-cell patch-clamp using scanning ion conductance microscopy, but this is restricted to cultured neurons40. Less invasive methods include bouton-attached recordings41 and voltage-sensitive dye imaging. A study using both of these methods in cerebellar stellate cells revealed an

NATURE COMMUNICATIONS | 7:12102 | DOI: 10.1038/ncomms12102 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12102

a

WT

b

1.2 **

0.8

PPR

Kcna1V408A/+

0.4

n = 7

n = 6

0 WT

Kcna1V408A/+

c d

WT

Kcna1V408A/+

1,500

1,000

Count

500

0 0 10 20 30 40

Interspike interval (ms)

e f

Interspike intervals (ms)

No. of spikes per burst

60

**

60

**

40

40

20

20

0 n = 20

0 n = 11

Kcna1V408A/+

WT

n =19

Kcna1V408A/+

WT

n = 5

g

WT

Kcna1V408A/+

h

80

*

Control

PTx CGP

Decrease in ISI (%)

60

40

20

0 n = 8

n = 10

WT

Kcna1V408A/+

Figure 4 | Increased GABA release at basket cell terminals in Kcna1V408A/ mice attenuates spontaneous Purkinje cell ring. (a) IPSCs recorded in cerebellar Purkinje cells of wild-type and Kcna1V408A/ mice. Grey and orange traces: individual sweeps. Black and turquoise traces: averages. Scale bar, 200 pA; 10 ms. (b) PPR was signicantly lower in Kcna1V408A/ than wild-type Purkinje cells. (c) Representative cell-attached recordings from wild-type and Kcna1V408A/ Purkinje cells, illustrating tonic and burst ring, and periods of quiescence. Scale bar, 500 ms; 50 ms (expanded traces). (d) Inter-spike interval distribution during tonic ring (same cells as in c). (e) Average inter-spike interval was signicantly longer in Kcna1V408A/ than wild-type Purkinje cells. (f) Average number of action potentials per burst was signicantly higher in wild-type neurons. (g) Representative traces showing a greater decrease in inter-spike interval following PTx (100 mM) and CGP (1 mM) perfusion in Kcna1V408A/ than wild-type neurons (Scale bar, 20 ms). (h) Summary of effect of GABA receptor blockade. *Po0.05; **Po0.01, unpaired t-tests (b,f,h) or MannWhitney U test (e).

important role of Kv1 channels in action potential shape and activity-dependent broadening at the axon initial segment42. However, in striking contrast to the present results, Kv1 family channels were found to play only a minor role in boutons. The difference is consistent with the intense staining for Kv1.1 and Kv1.2, specically in the terminals of basket cells912. Several other roles have been inferred for axonal Kv1.1 channels on the basis of somatic recordings. These include delayed spiking in some forebrain interneurons2,19 and subthreshold modulation of neurotransmitter release43.

The direct recordings from basket cell boutons reported here revealed an increase in spike width by acute application of the Kv1.1-specic blocker DTx-K, which was qualitatively similar to the effect of the Kcna1V408A/ mutation. The Kcna1V408A/

mutation, however, had a larger effect than pharmacological blockade (spike width at 30 mV: 1.260.08 ms and1.150.02 ms, respectively), consistent with a dominant negative effect and heteromultimeric assembly of Kv1 channels. Mutant mice also exhibit an increase in spontaneous IPSCs, but no difference in basket cell ring rate or in miniature IPSC amplitude18. We failed to observe a compensation for the genetic lesion at the level of spike shape, GABA release or downstream effects on Purkinje cell ring. This is arguably unexpected, given the abundant evidence that blocking neurotransmitter receptors or Na channels for a few days results in extensive synaptic plasticity, contributing to homoeostatic regulation of neuronal activity24. Most attention has hitherto been given to homoeostatic plasticity at

6 NATURE COMMUNICATIONS | 7:12102 | DOI: 10.1038/ncomms12102 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12102 ARTICLE

glutamatergic synapses. However, GABAergic synaptic strength has also been shown to undergo activity-dependent plasticity that compensates for altered activity4446. The results of the present study question the importance of these phenomena for an inherited presynaptic K channelopathy. A possible explanation is that basket cell synapses are intrinsically less plastic than synapses in the forebrain, although the nding that spontaneous Purkinje cell ring was depressed in Kcna1V408A/ mice further argues against a major role of homoeostatic plasticity in the cerebellar cortex.

The present study stresses the role of Kv1.1 in ensuring rapid repolarization. It is tempting to speculate that the very high density of Kv1.1 channels in basket cell terminals is an evolutionary adaptation related to the electrotonic properties of a relatively short but profusely arborizing axon, compromising the ability of passive charge dissipation to terminate action potentials. Basket cell terminals also have a prominent ADP that is not affected by either pharmacological or genetic manipulation of Kv1.1, or GABAA receptor blockade, and whose biophysical basis remains to be elucidated. Another anatomical peculiarity of basket cell terminals, the pinceau formation, where ephaptic transmission also contributes to inhibition of Purkinje cells47,48, only forms at a later developmental stage49, and so is unlikely to contribute to spike shape in the present study.

We observed a complementary compartmentalization of BKCa

channels, blockade of which affected somatic, but not presynaptic spike width. A pronounced effect of ChTx on somatic K currents has not been reported previously. Although BKCa

channels are usually abundantly expressed presynaptically, a detailed immunhistochemical study identied cerebellar basket cell synapses as the sole exception to this rule, with greater post-than presynaptic expression29. The effect of BKCa channel blockade on somatic action potential width is also unexpected because they generally require a large depolarization and a high local Ca2 concentration to open50. Although BKCa channels are present at cerebellar mossy bre terminals, their role in action potential repolarization is masked by recruitment of fast Kv3 channels51. We cannot exclude a possible role for presynaptic BKCa channels in basket cells under conditions when other K channels are non-functional.

Acute blockade of Kv1.1 channels with DTx-K led to an increase in action potential-evoked Ca2 inux, which was not observed in a previous study23. This is consistent with spike width as a major determinant of Ca2 channel activation and total

Ca2 inux52. Indeed, the simulations suggest that the ADP, by comparison, has little effect on Ca2 inux. We did not attempt to compare presynaptic Ca2 inux in wild-type and

Kcna1V408A/ mice because of the large variability among boutons. Nevertheless, both pharmacological and genetic manipulations of Kv1.1 led to a marked increase in evoked and spontaneous GABA release, consistent with a supralinear relationship between Ca2 inux and exocytosis33,34. In common with many other neurological channelopathies, the paroxysmal nature of EA1 remains unexplained. Although we have only examined in detail one synapse in the cerebellar circuit, where Kv1.1 is most abundantly expressed, we cannot exclude a role for other elements of the cerebellar circuitry. Strikingly, cerebellar incoordination both in patients with EA1 and in mice harbouring the V408A mutation is triggered by stress, suggesting a role for neuromodulators.

Methods

Cerebellar slices. Kcna1V408A/ mice were a generous gift from J. Maylie. Breeding pairs were set-up between a heterozygous Kcna1V408A/ female and a wild-type C57/BL6 mouse (Harlan), and refreshed after 6 months. This study was performed in accordance with the Animals (Scientic Procedures) Act 1986.

All experiments were performed and analysed blind to genotype. Acute parasagittal cerebellar slices (300-mm thick) were prepared from postnatal day 1420 mutant and wild-type littermates using a vibrating microtome (Leica VT 1200) in an ice-cold solution containing (in mM): 75 sucrose, 87 NaCl, 2.5 KCl, 25 NaHCO3, 1.25 NaH2PO4, 7 MgCl2, 0.5 CaCl2 and 20 glucose, (pH 7.4 when

gassed with 95% O2:5% CO2). Slices were warmed to 32 C for 10 min and then stored in a carbogen-gassed solution at room temperature containing (in mM): 126 NaCl, 2.5 KCl, 25 NaHCO3, 1.25 NaH2PO4, 1 MgCl2, 2.0 CaCl2 and 10 glucose.

The same solution was used to perfuse slices during electrophysiological recordings. Slices were used within 3 h of preparation.

Electrophysiology. Slices were anchored in a recording chamber mounted on the stage of an upright microscope (BX51WI, Olympus) and visualized with infrared differential interference contrast optics with a 20 water immersion objective.

Slices were continuously perfused at a rate of 3 ml min 1 and recordings were obtained at 3335 C. AMPA and NMDA receptors were blocked throughout. Purkinje cells and basket cell somata were patch clamped with 46-MO glass pipettes, to achieve an access resistance o20 MO. Bouton recordings were obtained with 911-MO glass pipettes, to achieve an access resistance of o50 MO. Basket cell boutons were identied by their size, shape, apposition to Purkinje cell somata and input resistance B1 GO.

For current-clamp experiments the pipette solution contained (in mM): 134 K-gluconate; 5 KCl, 10 phosphocreatine, 10 HEPES, 25 EGTA, 0.3 Na3GTP and 4

MgATP, as well as 2 mg ml 1 biocytin or 200 nM Alexa 568 (adjusted to pH7.27.3 with KOH). If necessary, current was injected to maintain the membrane potential between 70 and 75 mV. Boutons could be made to re one-to-one in response to 0.5- or 1-ms current injections delivered at 55 or 100 Hz.

For voltage-clamp experiments, the pipette solution contained (in mM): 140 CsCl; 1 EGTA, 10 HEPES, 1 MgCl2, 0.3 Na3GTP, 5 MgATP, 12 QX314-Br and pH 7.27.3 with CsOH. Purkinje cell somata were voltage clamped at 70 mV. Whole-cell capacitance and series resistance were compensated. IPSCs were elicited in Purkinje cells using an extracellular stimulating electrode placed in the molecular layer.

To measure the spontaneous ring rates of Purkinje cells, pipettes were lled with the perfusion solution, and voltage-clamp recordings were obtained in the cell-attached mode.

Data were acquired using a Multiclamp 700B amplier (Molecular Devices) and custom software (National Instruments LabView). Data were low-pass ltered (20 kHz) and digitized at 50 kHz.

All data from Kcna1V408A/ mice were compared with data from their wild-type littermates. Data were analysed using LabView and are presented as means.e.m. The action potential width was measured at 30 mV. Recordings where action potentials did not overshoot 0 mV were discarded. To calculate the area under the ADP curves the initial 20 ms were used. In graphs, open circles represent individual experiments, bars illustrate the averages obtained from all experiments. The paired-pulse ratio was calculated as the ratio of the second IPSC to the second IPSC evoked by two extracellular stimuli at 40 Hz. The voltage threshold was estimated from the maximum of the second time derivative of the voltage.

Two-photon excitation uorescence imaging. Basket cells held in current-clamp whole-cell mode were loaded with both a morphological dye (Alexa 594, 50 or 100 mM) and the Ca2 indicator Fluo-4 (200 mM), and imaged in two-photon excitation mode at 800 nm using a femtosecond Ti:sapphire-pulsed laser (MaiTai,

Spectra-Physics). Recordings started at least 20 min after obtaining the whole-cell conguration. We followed the axon and focused on presynaptic boutons in the Purkinje cell layer. Four 1-ms long depolarizing current pulses were injected at40 Hz to evoke action potentials, and the Fluo-4 uorescence was measured using line scans across the bouton width. Three or four trials were recorded with an inter-trial interval between 30 s and 1 min, and the saturated uorescence (Fmax)

was measured with a train of 50 action potentials at 100 Hz. The sequence was repeated at least 10 min after washing in DTx (200 nM). To improve the signal-to-noise ratio, we took the integral of the Fluo-4 uorescence signal for 200 ms from the rst action potential (Int-DF), as a measure of total action potential-evoked

Ca2 inux D[Ca2 ].

Presynaptic Ca2 kinetics. The total magnitude of action potential-evoked presynaptic Ca2 inux D[Ca2] was estimated using a non-stationary single-compartment model30,53,54. We assumed that, in addition to Fluo-4, the presynaptic bouton contained 150 mM of total parvalbumin (equivalent to 8 mM of free buffer at rest, where [Ca2 ]rest 10 nM and [Mg2 ]rest 0.5 mM). The

model operated with only two adjustable (free) parameters: D[Ca2 ] and the Ca2 removal rate, which have virtually independent effects on the Ca2 uorescence signal54, and thus were constrained by a direct tting procedure to match the simulated and experimental uorescence proles. Other model parameters were constrained by measuring the maximal Fluo-4 uorescence, which was used to estimate [Ca2 ]rest and also the total Fluo-4 concentration at the time of each measurement. This was necessary because the Ca2 indicator and morphological tracer did not fully reach a steady state during the time course of the

NATURE COMMUNICATIONS | 7:12102 | DOI: 10.1038/ncomms12102 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12102

experiment (that is, between 20 and 50 min after obtaining the whole-cell conguration; ref. 53). Ca2 and Mg2 binding/unbinding rates of parvalbumin were as in ref. 55, and Ca2 binding/unbinding rates of Fluo-4 were as inrefs 30,53,54.

Ca2 current modelling. Action potential-evoked Ca2 currents through P/Q-type Ca2 channels were modelled in the NEURON simulation environment56, using a six-state Ca2 channel model developed for hippocampal mossy bre boutons31 as described in detail previously (ref. 30).

Drugs. To block AMPA, NMDA, GABAA and GABAB receptors, 5 mM 2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo[f]quinoxaline-2,3-dione, 50 mM (2 R)-amino-5-phosphonovaleric acid (APV), 100 mM PTx and 1 mM CGP 52432, were added to the perfusate. DTx-K (200 nM) was used to block Kv1.1 channels. ChTx (100 mM) was used to block BKCa channels. APV was from Ascent (UK). DTx-K was from Alomone lab (UK). All other compounds were purchased from Tocris (Bristol).

Statistics. Data sets that passed the ShapiroWilk test for normality were analysed with Students paired or unpaired t-test. Non-parametric tests (Wilcoxon matched pairs signed-rank or MannWhitney U) were applied in all other cases.

Data availability. The data that support the ndings of this study are available from the corresponding author upon request.

References

1. Kole, M. H. P., Letzkus, J. J. & Stuart, G. J. Axon initial segment Kv1 channels control axonal action potential waveform and synaptic efcacy. Neuron 55, 633647 (2007).

2. Goldberg, E. M. et al. K channels at the axon initial segment dampen near-threshold excitability of neocortical fast-spiking GABAergic interneurons.

Neuron 58, 387400 (2008).3. Jan, L. Y. & Jan, Y. N. Voltage-gated potassium channels and the diversity of electrical signalling. J. Physiol. 590, 25912599 (2012).

4. Browne, D. L. et al. Episodic ataxia/myokymia syndrome is associated with point mutations in the human potassium channel gene, KCNA1. Nat. Genet. 8, 136140 (1994).

5. Browne, D. L. et al. Identication of two new KCNA1 mutations in episodic ataxia/myokymia families. Hum. Mol. Genet 4, 16711672 (1995).

6. Zuberi, S. M. et al. A novel mutation in the human voltage-gated potassium channel gene (Kv1.1) associates with episodic ataxia type 1 and sometimes with partial epilepsy. Brain J. Neurol. 122, 817825 (1999).

7. Eunson, L. H. et al. Clinical, genetic, and expression studies of mutations in the potassium channel gene KCNA1 reveal new phenotypic variability. Ann. Neurol. 48, 647656 (2000).

8. Rajakulendran, S., Schorge, S., Kullmann, D. M. & Hanna, M. G. Episodic ataxia type 1: a neuronal potassium channelopathy. Neurotherapeutics 4, 258266 (2007).

9. Laube, G. et al. Ultrastructural localization of Shaker-related potassium channel subunits and synapse-associated protein 90 to septate-like junctions in rat cerebellar Pinceaux. Brain Res. Mol. Brain Res. 42, 5161 (1996).

10. Trimmer, J. S. Subcellular localization of K channels in mammalian brain neurons: remarkable precision in the midst of extraordinary complexity.

Neuron 85, 238256 (2015).11. Wang, H., Kunkel, D. D., Martin, T. M., Schwartzkroin, P. A. & Tempel, B. L. Heteromultimeric K channels in terminal and juxtaparanodal regions of neurons. Nature 365, 7579 (1993).

12. Wang, H., Kunkel, D. D., Schwartzkroin, P. A. & Tempel, B. L. Localization of Kv1.1 and Kv1.2, two K channel proteins, to synaptic terminals, somata, and dendrites in the mouse brain. J. Neurosci. 14, 45884599 (1994).

13. Adelman, J. P., Bond, C. T., Pessia, M. & Maylie, J. Episodic ataxia results from voltage-dependent potassium channels with altered functions. Neuron 15, 14491454 (1995).

14. DAdamo, M. C., Liu, Z., Adelman, J. P., Maylie, J. & Pessia, M. Episodic ataxia type-1 mutations in the hKv1.1 cytoplasmic pore region alter the gating properties of the channel. EMBO J. 17, 12001207 (1998).

15. Rea, R., Spauschus, A., Eunson, L. H., Hanna, M. G. & Kullmann, D. M. Variable K( ) channel subunit dysfunction in inherited mutations of KCNA1.

J. Physiol. 538, 523 (2002).16. Maylie, B., Bissonnette, E., Virk, M., Adelman, J. P. & Maylie, J. G. Episodic ataxia type 1 mutations in the human Kv1.1 potassium channel alter hKvbeta 1-induced N-type inactivation. J. Neurosci. 22, 47864793 (2002).

17. Imbrici, P., DAdamo, M. C., Kullmann, D. M. & Pessia, M. Episodic ataxia type 1 mutations in the KCNA1 gene impair the fast inactivation properties of the human potassium channels Kv1.4-1.1/Kvbeta1.1 and Kv1.4-1.1/Kvbeta1.2. Eur. J. Neurosci. 24, 30733083 (2006).

18. Herson, P. S. et al. A mouse model of episodic ataxia type-1. Nat. Neurosci. 6, 378383 (2003).

19. Dehorter, N. et al. Tuning of fast-spiking interneuron properties by an activity-dependent transcriptional switch. Science 349, 12161220 (2015).

20. Harvey, A. L. & Robertson, B. Dendrotoxins: Structure-Activity Relationships and Effects on Potassium Ion Channels. Curr. Med. Chem. 11, 30653072 (2004).

21. Southan, A. P. & Robertson, B. Electrophysiological characterization of voltage-gated K( ) currents in cerebellar basket and purkinje cells: Kv1 and Kv3 channel subfamilies are present in basket cell nerve terminals. J. Neurosci. 20, 114122 (2000).

22. Southan, A. P. & Robertson, B. Modulation of inhibitory post-synaptic currents (IPSCs) in mouse cerebellar Purkinje and basket cells by snake and scorpion toxin K channel blockers. Br. J. Pharmacol. 125, 13751381 (1998).

23. Tan, Y. P. & Llano, I. Modulation by K channels of action potential-evoked intracellular Ca2 concentration rises in rat cerebellar basket cell axons.

J. Physiol. 520(Pt 1): 6578 (1999).24. Turrigiano, G. Too many cooks? Intrinsic and synaptic homeostatic mechanisms in cortical circuit renement. Annu. Rev. Neurosci. 34, 89103 (2011).

25. Robertson, B., Owen, D., Stow, J., Butler, C. & Newland, C. Novel effects of dendrotoxin homologues on subtypes of mammalian Kv1 potassium channels expressed in Xenopus oocytes. FEBS Lett. 383, 2630 (1996).

26. Southan, A. P. & Robertson, B. Patch-clamp recordings from cerebellar basket cell bodies and their presynaptic terminals reveal an asymmetric distribution of voltage-gated potassium channels. J. Neurosci. 18, 948955 (1998).

27. Trigo, F. F., Chat, M. & Marty, A. Enhancement of GABA release through endogenous activation of axonal GABA(A) receptors in juvenile cerebellum.J. Neurosci. 27, 1245212463 (2007).28. Hu, H. et al. Presynaptic Ca2 -activated K channels in glutamatergic hippocampal terminals and their role in spike repolarization and regulation of transmitter release. J. Neurosci. 21, 95859597 (2001).

29. Knaus, H. G. et al. Distribution of high-conductance Ca(2 )-activatedK channels in rat brain: targeting to axons and nerve terminals. J. Neurosci.

16, 955963 (1996).30. Ermolyuk, Y. S. et al. Differential triggering of spontaneous glutamate release by P/Q-, N- and R-type Ca2 channels. Nat. Neurosci. 16, 17541763 (2013).

31. Li, L., Bischofberger, J. & Jonas, P. Differential gating and recruitment of P/Q-, N-, and R-type Ca2 channels in hippocampal mossy ber boutons. J.

Neurosci. 27, 1342013429 (2007).32. Stephens, G. J., Morris, N. P., Fyffe, R. E. W. & Robertson, B. The Cav2.1/a1A (P/Q-type) voltage-dependent calcium channel mediates inhibitory neurotransmission onto mouse cerebellar Purkinje cells. Eur. J. Neurosci. 13, 19021912 (2001).

33. Matveev, V., Bertram, R. & Sherman, A. Ca2 current versus Ca2 channel cooperativity of exocytosis. J. Neurosci. 29, 1219612209 (2009).

34. Bucurenciu, I., Kulik, A., Schwaller, B., Frotscher, M. & Jonas, P. Nanodomain coupling between Ca2 channels and Ca2 sensors promotes fast and efcient transmitter release at a cortical GABAergic synapse. Neuron 57,536545 (2008).

35. Womack, M. D. & Khodakhah, K. Dendritic control of spontaneous bursting in cerebellar Purkinje cells. J. Neurosci. 24, 35113521 (2004).

36. Sasaki, T., Matsuki, N. & Ikegaya, Y. Targeted axon-attached recording with uorescent patch-clamp pipettes in brain slices. Nat. Protoc. 7, 12281234 (2012).

37. Nakamura, Y. & Takahashi, T. Developmental changes in potassium currents at the rat calyx of Held presynaptic terminal. J. Physiol. 581, 11011112 (2007).

38. Bischofberger, J., Engel, D., Li, L., Geiger, J. R. P. & Jonas, P. Patch-clamp recording from mossy ber terminals in hippocampal slices. Nat. Protoc. 1, 20752081 (2006).

39. Rancz, E. A. et al. High-delity transmission of sensory information by single cerebellar mossy bre boutons. Nature 450, 12451248 (2007).

40. Novak, P. et al. Nanoscale-targeted patch-clamp recordings of functional presynaptic ion channels. Neuron 79, 10671077 (2013).

41. Kawaguchi, S. & Sakaba, T. Control of inhibitory synaptic outputs by low excitability of axon terminals revealed by direct recording. Neuron 85, 12731288 (2015).

42. Rowan, M. J. M., Tranquil, E. & Christie, J. M. Distinct Kv channel subtypes contribute to differences in spike signaling properties in the axon initial segment and presynaptic boutons of cerebellar interneurons. J. Neurosci. 34, 66116623 (2014).

43. Bialowas, A. et al. Analog modulation of spike-evoked transmission in CA3 circuits is determined by axonal Kv1.1 channels in a time-dependent manner. Eur. J. Neurosci. 41, 293304 (2015).

44. Xu, C., Zhao, M., Poo, M. & Zhang, X. GABA(B) receptor activation mediates frequency-dependent plasticity of developing GABAergic synapses. Nat. Neurosci. 11, 14101418 (2008).

8 NATURE COMMUNICATIONS | 7:12102 | DOI: 10.1038/ncomms12102 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12102 ARTICLE

45. Maffei, A., Nataraj, K., Nelson, S. B. & Turrigiano, G. G. Potentiation of cortical inhibition by visual deprivation. Nature 443, 8184 (2006).

46. Nahmani, M. & Turrigiano, G. G. Deprivation-induced strengthening of presynaptic and postsynaptic inhibitory transmission in layer 4 of visual cortex during the critical period. J. Neurosci. 34, 25712582 (2014).

47. Blot, A. & Barbour, B. Ultra-rapid axon-axon ephaptic inhibition of cerebellar Purkinje cells by the pinceau. Nat. Neurosci. 17, 289295 (2014).

48. Kole, M. J. et al. Selective loss of presynaptic potassium channel clusters at the cerebellar basket cell terminal pinceau in adam11 mutants reveals their role in ephaptic control of purkinje cell ring. J. Neurosci. 35, 1143311444 (2015).

49. Sotelo, C. Development of Pinceaux formations and dendritic translocation of climbing bers during the acquisition of the balance between glutamatergic and gamma-aminobutyric acidergic inputs in developing Purkinje cells. J. Comp. Neurol. 506, 240262 (2008).

50. Fakler, B. & Adelman, J. P. Control of K(Ca) channels by calcium nano/ microdomains. Neuron 59, 873881 (2008).

51. Alle, H., Kubota, H. & Geiger, J. R. P. Sparse but highly efcient Kv3 outpace BKCa channels in action potential repolarization at hippocampal mossy ber boutons. J. Neurosci. 31, 80018012 (2011).

52. Bischofberger, J., Geiger, J. R. P. & Jonas, P. Timing and efcacy of Ca2 channel activation in hippocampal mossy ber boutons. J. Neurosci. 22, 1059310602 (2002).

53. Ermolyuk, Y. S. et al. Independent regulation of basal neurotransmitter release efcacy by variable Ca2 inux and bouton size at small central synapses. PLoS

Biol. 10, e1001396 (2012).54. Scott, R. & Rusakov, D. A. Main determinants of presynaptic Ca2 dynamics at individual mossy ber-CA3 pyramidal cell synapses. J. Neurosci. 26, 70717081 (2006).

55. Eggermann, E. & Jonas, P. How the slow Ca(2) buffer parvalbumin affects transmitter release in nanodomain-coupling regimes. Nat. Neurosci. 15, 2022 (2012).

56. Hines, M. L. & Carnevale, N. T. The NEURON simulation environment. Neural Comput. 9, 11791209 (1997).

Acknowledgements

We are indebted to J. Maylie for the gift of the Kcna1V408A/ mice, to M. Cano andS. Martin for the help with breeding and genotyping, to Y. Timofeeva for the help with simulations, to D.A. Rusakov for the help with Ca2 imaging and to S. Schorge,

M.C. Walker, D. Attwell and R.A. Silver for helpful comments. This work was supported by the Medical Research Council, European Research Council and Wellcome Trust.

Author contributions

R.B. performed the electrophysiological experiments. Y.B. performed the Ca2 imaging experiments. K.E.V. analysed the Ca2 uorescence data and modelled presynaptic

Ca2 kinetics. D.M.K. designed the project. D.M.K. and R.B. wrote the manuscript with revisions from co-authors. The authors declare no competing nancial interests.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Begum, R. et al. Action potential broadening in a presynaptic channelopathy. Nat. Commun. 7:12102 doi: 10.1038/ncomms12102 (2016).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

r The Author(s) 2016

NATURE COMMUNICATIONS | 7:12102 | DOI: 10.1038/ncomms12102 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9