Type of the Paper (Article

0 downloads 0 Views 1MB Size Report
Sep 6, 2017 - 1,2-dioleoyl-sn-glycero-3-phospho-L-serine DOPS 40%, ...... partitioning between water and lipid bilayer, Langmuir : the ACS journal of ...

1

Article BBA Biomembrane: available online: 6-SEP-2017

3

Functional reconstitution of cell-free synthetized purified Kv channels

4 5

Stéphane Renauld 1-3, Sandra Cortes 4, Beate Bersch 1-3, Xavier Henry 1-3, Michel De Waard 5, 6, Béatrice Schaack 1-3*

2

6 7 8 9 10 11 12 13 14

1

15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

Academic Editor: name Received: date; Accepted: date; Published: date

33 34

Univ. Grenoble Alpes, IBS, F-38044 Grenoble, France. CNRS, IBS, F-38044 Grenoble, France. 3 CEA, IBS, F-38044 Grenoble, France; [email protected]; [email protected]; [email protected]; [email protected] 4 Synthelis; 5, avenue du Grand Sablon 38700 La Tronche, France; [email protected] 5 Inserm UMR 1087 / CNRS UMR 6291, Institut du Thorax, 44007 Nantes, France; [email protected]niv-nantes.fr 2

6

Smartox Biotechnology, 570 Rue de la Chimie, 38400 Saint-Martin d’Hères, France.

* Correspondence to: [email protected]; Tel.: +33-457-42-8670

Abstract: The study of ion channel activity and the screening of possible inhibitor molecules require reliable methods for production of active channel proteins, their insertion into artificial membranes and for the measurement of their activity. Here we report on cell-free expression of soluble and active Kv1.1 and Kv1.3 channels and their efficient insertion into liposomes. Two complementary methods for the determination of the electrical activity of the proteoliposome-embedded channels were compared using Kv1.1 as a model system: (1) single channel recordings in droplet interface bilayers (DIB) and (2) measurement of the membrane voltage potential generated by a potassium ion diffusion potential using the voltage-sensitive fluorescent dye oxonol VI. Single channel recordings in DIBs proved unreliable because of the non-reproducible fusion of proteoliposomes with an artificial membrane. Therefore, the use of the optical indicator oxonol VI was adapted for 96 well microtiter plates using the ionophore valinomycin as a positive control. The activity of Kv1.1 and Kv1.3 channels was then monitored in the absence and presence of different venom toxins, demonstrating that fluorescent dyes can be used very efficiently when screening small molecules for their channel blocking activity. Keywords: Kv channels; DIB; oxonol VI; proteoliposomes 1. Introduction

35

In humans, potassium channels are involved in many important physiological processes and

36

therefore offer important opportunities for the development of new drugs [1]. This involves the

37

development of sensitive and reproducible screening assays for potassium channel activity. In

38

particular, the Kv1.3 channel has been related to auto-immune diseases, especially because of its role

39

in immune response at the plasma membrane of effector memory T lymphocytes (TEM). The

40

voltage-gated potassium Kv1.3 channel regulates the membrane potential of TEM cells and provides

41

the counterbalancing K+ efflux that regulates Ca2+ signalling during their activation response. The

2 of 19

42

driving force for Ca2+ entry is restored by membrane hyperpolarization brought about by the opening

43

of Kv1.3 channels in response to membrane depolarization. In patients with multiple sclerosis, type-1

44

diabetes, rheumatoid arthritis, psoriasis, or chronic asthma, disease-associated TEM cells are

45

Kv1.3-dependent, and selective Kv1.3 inhibitors suppress the proliferation of these cells and their

46

cytokine production [2]. Hence, its important role within TEM cells has made Kv1.3 an attractive

47

therapeutic target for immunomodulation. Kv1.1 channel play a role in the repolarisation of neuronal

48

membranes and are known to be blocked by several toxins like the α-dendrotoxin from Dendroaspis

49

angusticeps snake. Both Kv1.1 and Kv1.3 proteins are expected to have very similar structures as

50

compared to the known structure of Kv1.2 [3].

51

Several studies using animal preclinical models, as well as human patients in clinical trials,

52

have already been conducted in order to test the immunomodulatory efficiency of selective Kv1.3

53

inhibitors, in particular those peptides issued from venom sources [2, 4, 5]. Several peptides, like

54

margatoxin from the venom of the Centrutoides margaritatus scorpion and ShK from the venom

55

of Stichodactyla helianthus sea anemone, displayed strong affinities for the channel pores, and have

56

been identified by means of patch clamp techniques using recombinant cells.

57

Functional membrane insertion of Kv1 channels has already been evaluated through

58

electrophysiological experiments. These included the electrophysiological response of purified Kv1

59

channels incorporated into artificial membranes, patch clamping and whole-cell current experiments

60

performed on endogenous channels [5, 6]. Kv1.1 and Kv1.3 channels have so far been mainly analysed

61

using recombinant eukaryotic cells and bacteria [7]. In this paper we report a more time-efficient

62

method using cell free protein expression in presence of a detergent followed by purification and

63

reconstitution into liposomes with a well-defined lipid composition. For channel activity

64

measurements, we first attempted an electrophysiological technique described by Baley et al. [4]

65

where the protein is embedded in a bilayer at the interface of two droplets dipped in oil (droplet

66

interface bilayer DIB). Use of this technique combined with the injection of small volumes of venom

67

for screening of Kv1.3 channel activity was already described by [8]. This technique was found not

68

suitable for high throughput studies, because of difficulties encountered with protein reconstitution

69

and the addition of very tiny amounts of venom into the droplets. Therefore, we established an

70

alternative approach using a voltage-sensitive fluorescent dye, oxonol VI, in microtiter plates [9-11].

71

We exploited the inside-positive membrane potential generated by K+ fluxes through active Kv

72

channels in the presence of a K+ concentration gradient across the membrane. This allowed

73

monitoring of the activity of the Kv1.3 channel and its inhibition by ShK and margatoxin [12]. Herein

74

we report reproducible and encouraging results, which pave the way for a more extended study

75

aiming towards the screening of different venom fractions. We have also used the human K v1.1

76

channel for proof of concept for DIB studies as well as for fluorescent membrane potential

77

measurements.

3 of 19

78

2. Experimental Section

79

2.1. Molecular biology

80

Human Kv1.1 cDNA sequence (gene ID: NM_000217) was codon-optimized for expression

81

in E. coli (GeneArt, Germany). The resulting 1.812 kb cDNA was subcloned into pIVEX 2.4a

82

plasmid (Roche Applied Science, Germany) between Sac II and Xho I restriction sites, respectively.

83

This construction encoded Kv1.1 channel with a His6-tag upstream of the first methionine. Plasmid

84

construction was verified by DNA sequencing (Beckman Coulter Genomics, USA) and sequence

85

analysis was performed with Serial Cloner freeware (SerialBasics, USA). Plasmids were prepared

86

using Nucleobond Xtra Midi prep (Macherey Nagel, Germany). Mouse Kv1.3 protein production

87

(gene ID: NM_019270.3) was provided free of charge by Synthelis [13].

88

2.2. Cell free expression of Kv1.1 and Kv1.3 channels

89

Kv1.1 RNA transcription and protein translation were performed in a reaction volume of 2

90

mL in batch mode, using an E. coli S30 extract, as described by Schwartz et al. [14]. Each reagent was

91

carefully stored and handled in an RNase-free environment. The T7-based transcription was obtained

92

using a 16 µg/mL plasmid solution. The Kv1.1 channel was expressed in the presence of 0.9 mM Brij

93

35 detergent (CMC = 0.09 mM) (Sigma Aldrich, USA). The reaction mix was incubated in a bench

94

shaker (Eppendorf, USA) for 5 hrs at 22°C. Kv1.3 RNA transcription and protein translation were

95

performed in a reaction volume of 2 mL in batch mode. The Kv1.3 protein was expressed in presence

96

of 1.5 mM n-Dodecyl β-D-maltoside (DDM) detergent (CMC = 0.15 mM) (Anatrace, USA).

97

2.3. Purification of Kv1.1 and Kv1.3 channels

98

The cell free synthesis reaction mixture was centrifuged at 13,000 rpm for 5 min at 4°C. The

99

supernatant was then recovered and diluted 5 times with 8 mL of 150 mM NaCl, 50 mM Tris, pH 8,

100

25 mM imidazole, 0.9 mM Brij 35 or 1.5 mM DDM for Kv1.1 or Kv1.3, respectively and loaded on 1

101

mL Ni NTA gel overnight (batch mode). Washes were carried out in the same buffer but using less

102

detergent (0.2 mM Brij 35 or 0.3 mM DDM). Elution was done in four 2.5 mL fractions containing

103

150 mM KCl (for the electrophysiology experiments) or 150 mM NaCl (for the fluorescent

104

experiments), 10 mM Pipes, pH 7.4, 0.2 mM Brij 35 or 0.3 mM DDM and 300 mM imidazole. Protein

105

quantification and quality were assessed by UV absorbance at 280 nm and Western blot analysis

106

using a rabbit HRP-coupled anti His tag antibody.

107

2.4. Empty liposomes preparation

108

Pure

lipids

(1,2-diphytanoyl-sn-glycero-3-phosphocholine

DPhPC,

109

1,2-dioleoyl-sn-glycero-3-phosphocholine

110

(1,2-dioleoyl-sn-glycero-3-phosphoethanolamine

111

1,2-dioleoyl-sn-glycero-3-phospho-L-serine DOPS 40%, 1,2-dimyristoyl-sn-glycero-3-phosphate

112

DMPA 20%, cholesterol 20 %) were purchased as chloroform solutions (Avanti Polar Lipids, USA).

DOPC),

and DOPE

L4

mix 20%,

4 of 19

113

For liposome preparation, the desired amount of lipids was transferred to a glass vessel and the

114

chloroform was evaporated under nitrogen flux in a fume hood and then residual chloroform was

115

removed under vacuum in a freeze dryer (Christ, Germany) under 0.055 mbar atmosphere at -55°C

116

for 12 hrs. Dried lipids were suspended as multi-lamellar vesicles to a final concentration of 10

117

mg/mL in 10 mM Pipes pH 7.4 and 150 mM KCl (buffer B) for the DPhPC liposomes used in

118

electrophysiology or 10 mM Pipes pH 7.4 and 150 mM NaCl (buffer N) for the L4 liposomes used in

119

fluorescent assays. Suspensions of large unilamellar vesicles (LUVs) were obtained by extrusion,

120

using successively 200 nm and 100 nm nucleopore polycarbonate membranes (Whatman, USA),

121

mounted between two semi-chambers (Avanti Polar Lipids, USA) in accordance with the

122

manufacturer’s instructions. The suspension of LUVs was monomodal with a mean hydrodynamic

123

diameter of 160 nm measured by dynamic light scattering analysis (Wyatt, USA). The polydispersity

124

was typically 20%, which means that the diameter of the liposomes varied between 90 and 200 nm.

125

2.5. Lipid titration curve by detergents

126

Before attempting protein reconstitution, we controlled the liposome solubilisation process

127

using the detergent octyl glucoside (OG) (Anatrace, USA) as described by Paternostre et al. [15] for

128

Kv1.1 and the detergent Triton X100 (Sigma Aldrich, USA) in the case of Kv1.3. The turbidity of the

129

liposome preparations was measured by optical density at 550 nm at fixed phospholipid (2 mg/mL)

130

and detergent (0.2 mM Brij 35, 0.3 mM DDM) concentrations while varying the amount of added OG

131

or Triton X100. From these measurements, we determined the amount of detergent necessary for

132

complete saturation (Rsat) and solubilisation (Rsol) of the lipid phase. Optical density was measured

133

30 s after detergent addition to the liposome suspension.

134

2.6. Protein insertion in liposomes

135

Liposomes were first destabilized adding 23 mM OG or 1 mM Triton X100 to the liposome

136

DPhPC or L4, respectively. These detergent concentrations, halfway between the Rsat and Rsol,

137

yielded an isotropic solution of mixed phospholipid-detergent micelles. Protein insertion in the

138

liposomes DPhPC-OG or L4-Triton X100 was then obtained by adding the solubilized Kv1.1 or

139

Kv1.3, respectively, for 15 min at 21 °C. As described by Geertsma et al. [16], a protein/lipid ratio of

140

1/10 w/w (1/ 3,500 mol/mol) was chosen: 0.05 mg freshly purified and solubilized Kv1.1 or Kv1.3

141

were mixed with 0.5 mg liposome in 0.5 mL of buffer B or buffer N in a small Eppendorf tube,

142

respectively. In the case of Kv1.1-proteoliposomes, detergents (OG and Brij35) were removed by

143

successive 1 h and overnight dialyses using MIDI GeBa flex-tube (MWCO 3.5 kDa) in 1 L buffer B.

144

For the formation of Kv1.3 proteoliposomes, 100 mg of biobeads (Biorad, USA) were added 4 times

145

to the 0.5 mL protein/lipid/detergent mixture in buffer N at 21 °C (first time) and at 4 °C (three

146

subsequent times), as described in [16]. During detergent removal, the proteins spontaneously

147

associated with lipids to form proteoliposomes.

148

2.7. Quantification of the protein insertion in proteoliposome using sucrose gradient

5 of 19

149

Proteoliposomes formed by Kv1.3 in L4 liposomes were purified on a sucrose gradient in

150

order to quantify protein insertion. We used a 2.2 mL discontinuous sucrose gradient (10 to 40%

151

sucrose in buffer N). The proteoliposome solution obtained after detergent removal was loaded onto

152

the top of the gradient. This gradient was centrifuged for 15 h at 366 000 RCF (rotor SW 55 Ti,

153

Beckman, USA) at 4 °C. After centrifugation, the 2.2 mL gradient was split into 5 fractions (400 µL

154

each) and the amount of His-tagged recombinant protein of each fraction was assessed using 3 µL dot

155

blots on nitrocellulose. Blots were performed in triplicate using 3 µL of the proteoliposome solution,

156

3 µL of each 400 µL gradient fraction (fractions 10/20/30/40 in Fig. 1C) and 3 µL of the pellet which

157

was suspended in 400 µL Laemmli buffer (0.1% SDS). The membrane was then incubated for 1 h in

158

2% milk-tris-buffered saline-Tween 20 solution (TBST) containing a 1/5000 rabbit horseradish

159

peroxidase (HRP)-coupled anti-His tag antibody. HRP signal was revealed using enhanced

160

chemiluminescence and quantified using the Chemidoc imager (Biorad, USA) and the Image Lab

161

software.

162

2.8. Channel orientation in proteoliposomes

163

The orientation of the protein in the membrane of the liposome determines the possibility to

164

block the channel using a toxin on the outside of the proteo-liposome. The His-tag is situated on the

165

N-terminal part of the Kv1.3 channel. Therefore the extent of the binding of an HRP-coupled anti-His

166

tag antibody to the proteoliposomes identifies the direction of the protein, when an absence of signal

167

indicates that the tag is inside the liposome, and when a limited signal indicates that part of the tag lies

168

outside of the liposome. The proteoliposomes were incubated for 30 min (100 µL of the 10% sucrose

169

fraction) with a 1/5000 rabbit HRP-coupled anti-His tag antibody followed by 30 min

170

ultracentrifugation (TLA 100 rotor, Beckman, USA, 8700 RCF). The pellet was washed twice with

171

buffer N, then re-suspended in 100 µL Laemmli buffer. The HRP signal was then assessed by spotting

172

3 µL of the re-suspended pellet on nitrocellulose. HRP signal was revealed using enhanced

173

chemiluminescence and quantified using the Chemidoc imager (Biorad, USA) and the Image Lab

174

software.

175

2.9. Single channel recordings

176

The recordings were performed at room temperature using an inverted microscope within a

177

Faraday cage. DIB experiments were carried out according to Bayley et al. [17] with slight

178

modifications. 250 μm-diameter chlorurated Ag/AgCl electrodes covered with 5% agarose were

179

handled using a micromanipulator (Burleigh, Thorlabs, USA). These electrodes were plunged in 3-4

180

mL dodecane or hexadecane (Sigma Aldrich, USA) contained in a 5 cm diameter Petri dish. This oil

181

did not contain lipids, an issue differing from [17]. 350 nL droplets containing Kv1.1 proteoliposomes

182

in buffer B were placed on each electrode. The fusion of some liposomes (5 mg/mL) in both droplets

183

was producing a lipid monolayer at the interface buffer/oil. During this stage, however, the proteins

6 of 19

184

originally contained in the proteoliposomes were lost and could not be embedded in a single lipid

185

bilayer.

186

After 10 min, a physical contact between both monolayers was established and a lipid bilayer

187

was formed at the interface. Resistance and capacitance were assessed by applying a 10 mV square

188

pulse at 100 Hz with the “membrane test” tool of pClamp 10 (Molecular Devices, USA). A bilayer is

189

considered stable when its resistance is within the 1 GΩ range. Lower resistance indicates that the

190

layers are not tightly assembled and/or that the membrane is fragile, ready to disrupt; tenth of GΩ

191

resistance indicates multilayers instead of bilayers. These last two cases correspond to membranes

192

that are not suitable for protein insertion. Membrane capacitance, directly proportional to bilayer

193

surface is typically between 200 and 500 pF. Single channel currents were recorded using an

194

Axopatch 200B amplifier (Molecular Devices, USA), and acquired at 2-3 kHz with a low-pass Bessel

195

filter set at 1 kHz. Recordings were performed using Clampex software (pClamp 10 suite, Molecular

196

Devices, USA) and digidata 1322A interface analogue-to-digital converter (Molecular Devices,

197

USA). Single channel current conductance and open probability were analyzed using ClampFit

198

software (Molecular Devices, USA). By convention the Trans electrode is connected to the ground

199

and voltage is clamped with the Cis electrode. Variable volumes of tested drugs (from 2.3 to 9.2 nL)

200

were injected into the droplets using Nanoject II (Harvard Apparatus, USA).

201

2.10. Fluorescent measurements

202

The oxonol VI (Sigma-Aldrich, USA) stock solution was kept at 0.32 M in ethanol at -20°C.

203

Oxonol VI was diluted in buffer N at 0.15 µM just before the experiments. 50 nM valinomycin

204

(Sigma-Aldrich, USA) was used as positive control. Liposomes, as well as Kv1.1- or

205

Kv1.3-proteoliposomes were diluted to a lipidic concentration of 50 µg/mL in KCl-free buffer N.

206

Experiments were performed in duplicate. First, 2.5 µL of KCl at different concentrations was added

207

to the wells of a 96-well half area Greiner black microtiter plate (Greiner Bio-One, Germany) in order

208

to vary the KCl final concentration between 0 and 150 mM. Then 47.5 µL of a solution containing the

209

Kv1.1- or Kv1.3- (or valinomycin-) liposomes + oxonol VI in buffer N was added and mixed.

210

Immediately afterwards the fluorescence intensity of each well was measured using a Clariostar

211

(BMG, USA) spectrophotometer (400 flashes, excitation wavelength 544 +/- 8 nm, emission

212

spectrum between 620 +/- 4 and 650 +/- 4 nm). The fluorescence intensity F was measured at 640 nm

213

as a function of the inside-positive membrane potential created by the entrance of K+ ion into the

214

proteoliposomes. The reference fluorescence intensity F0 in the absence of KCl was measured in

215

separate wells with the same proteoliposome concentration. The fluorescence increase as a function

216

of an inside-positive membrane potential was expressed and normalized by (F-F0)/F0.

217

2.11. Toxin inhibition experiments

218

The toxin ShK and margatoxin were provided by Smartox Biotechnology [12]. These

219

lyophilized peptides were diluted in water and aliquots of the stocks were kept frozen at -80°C at 1.22

7 of 19

220

mM and 0.24 mM, respectively. For the toxin inhibition experiments, we first added the toxin to the

221

Kv1.3-proteoliposome during 5 min, before addition of oxonol VI, and finally KCl. The fluorescence

222

response was measured at 640 nm as described above.

223

3. Results and Discussion

224

3.1. Proteoliposome formation

225

Deal et al [7] have reported that the recombinant rat Kv1.1 channel is represented by two

226

molecular species with molecular weights of 57 and 59 kDa. They concluded that pulse-chase

227

labelling studies in cells were consistent with a precursor-product relationship with the 57 kDa

228

species giving rise to the 59 kDa protein within several minutes of synthesis. There was no

229

glycosylation difference between both species indicating that the difference in molecular weight was

230

not related to glycosylation. After in vitro expression, we also detected two forms of human Kv1.1

231

proteins. Both bands proved to be positive in Western blot analysis using either a rabbit HRP-coupled

232

anti-His tag antibody (Fig. 1A) recognizing the N-terminal part of the Kv1.1 channel or a specific

233

N-terminal Kv1.1 antibody (SI). This latter analysis confirmed that the second band did not

234

correspond to a partial synthesis. In contrast, mouse Kv1.3 in vitro expression resulted in a single band

235

at 60 kDa (Fig. 1A). Bacterial expression produced recombinant proteins without glycosylation.

236

However, it has been shown previously by using a Kv1.1 mutant protein that absence of glycosylation

237

had little effect on channel synthesis (two bands still detected), turnover, or function [7].

238

Each Kv channel necessitated a specific detergent for its successful solubilisation during cell

239

free synthesis (Brij 35 for Kv1.1, DDM for Kv1.3). For the destabilization of the liposomes we used

240

OG (for Kv1.1) or Triton X-100 (for Kv1.3). Different lipids were also used: DPhPC for Kv1.1 in order

241

to perform electrophysiological studies where the diphytanoyl fatty acid chains produced stable

242

planar lipid membranes and L4 mix for Kv1.3, containing cholesterol in an anionic environment. The

243

high critical micelle concentration (CMC) detergent OG was removed using dialysis. The low CMC

244

detergent Triton X-100 was removed using BioBeads. Triton X-100 has been reported to yield better

245

results than OG in many cases [15, 16] and might be considered as a better choice for liposome

246

destabilization. Indeed, in our case, a lower quantity of Triton X100 than OG was necessary (1 mM

247

Triton X100 versus 23 mM OG) for the insertion of the Kv1.1 and Kv1.3 channels into liposomes. As

248

these detergents had to be removed from the liposome and from the medium, a lower concentration

249

was more suitable. For this removal and in order to proceed faster and avoid protein degradation, we

250

therefore recommend to use BioBeads instead of dialysis. Kv1.3 proteoliposomes were then separated

251

from precipitated protein using a sucrose gradient sedimentation. Nevertheless, ultracentrifugation at

252

29,000 RCF induced fusion of the proteoliposomes. Indeed, a doubling of the liposome size could be

253

demonstrated by dynamic light scattering (DLS). Approximately 75% of the available Kv1.3 was

254

found inserted into the liposomes. The fractions containing Kv1.3-containing proteoliposomes were

255

identified using a dot blot analysis during which the nitrocellulose membrane was treated with Tween

8 of 19

256

20 detergent. This results in the solubilisation of the liposomes and therefore exposition of the His-tag

257

which is subsequently detected by ECL. This analysis demonstrated that the white band visible in the

258

10% sucrose fraction corresponds to the Kv1.3 proteoliposomes. The yield of inserted protein was

259

determined using detergent-solubilized protein as a control, as shown in Fig. 1C.

260 261

Figure 1. A: Representative anti-His Western blot of purified human Kv1.1 and mouse Kv1.3. One µg of protein

262

separated in 10% denaturing SDS/Page gel and transferred onto a PVDF membrane. The monomeric form of

263

Kv1.1 shows two bands, with maximum intensity at 56 kDa. The Kv1.3 monomer appears at 60 kDa (red arrows).

264

B: Typical Triton X100 saturation curve of L4 lipids. Increasing concentrations of detergent were added to the

265

liposome suspension (2 mg/mL). The turbidity curve allows identification of the detergent-to-lipid ratio for

266

which the turbidity started to decrease (R experimental = Rexp = 1 mM). Channels were incorporated at Rexp.

267

Rsol corresponds to the detergent-to-lipid ratio for which total solubilisation was reached (Rsol = 5 mM). C: Dot

268

blot analysis of Kv1.3 insertion into L4 liposomes. After protein insertion and detergent removal using

269

BioBeads, proteoliposomes were deposited on a sucrose gradient. Five fractions of 400 µL corresponding to

270

each sucrose concentration were collected and 3 µL of each fraction was spotted in triplicates on a nitrocellulose

271

membrane, along with a 3 µL sample of the original solution (starting material). Kv1.3 was detected with

272

anti-His tag antibody. The nitrocellulose membrane was incubated in presence of Tween 20 detergent which

273

leads to the destruction of the proteoliposomes, making the His-tag accessible to the antibody. The protein was

274

detected in the 10% sucrose fraction that contained lipids (white band appearing in the tube). D:

275

Proteoliposomes (PL) immune-sedimentation using anti-His antibody in different experimental conditions and

276

after purification by sucrose sedimentation. Top panel (liposome only): showing the background signal in the

277

absence of Kv1.3. Middle panel: PL obtained using 1 mM Triton X100, the signal was equal to background noise,

278

demonstrating that the His-tag is located at the inside of the proteoliposome. Bottom panel: PL obtained using

279

10 mM Triton X100, the significant signal increase illustrates an important proportion of the His-tag localized

280

outside the liposome. Conditions described for middle panel were solely used throughout the manuscript.

281

In order to determine the orientation of the Kv1.3 channel in the liposome membrane we

282

performed an immuno-sedimentation experiment. The proteoliposomes, issued from the sucrose

283

gradient shown in Fig. 1C, were first incubated with a rabbit HRP-coupled anti-His tag antibody. The

284

resulting protein-liposome-antibody complexes were sedimented by centrifugation while unbound

9 of 19

285

antibodies remained in the supernatant. The pellet was washed in order to remove residual free

286

antibodies. Centrifugation was done at 8,700 RCF in order to avoid bursting of the proteoliposomes.

287

Proteoliposomes obtained with different concentrations of destabilizing Triton X100 were compared.

288

This analysis revealed (Fig. 1D) that the channel orientation was dependent on Triton X100

289

concentration used for destabilization of the liposomes during channel reconstitution.

290

Proteoliposomes prepared using 1 mM Triton X100 did not yield an HRP signal, indicating that the

291

N-terminal histidine tag of Kv1.3 was not accessible to the HRP-coupled anti-His tag antibody. This

292

means that the channels are inserted in an outside-out orientation. The use of 10 mM Triton X100, a

293

concentration at which the lipids are completely solubilized, an HRP-positive signal was recorded

294

showing that the N-terminal histidine tag was partially located outside the liposome. As reported in

295

previous articles [16, 18], a high Rexp value results in an undefined orientation of the protein within

296

the proteoliposomes.

297

The production of Kv1.1 proteoliposomes was performed in a similar way but with different

298

detergents (Brij35 was used for the solubilisation of the protein during cell free expression and OG

299

was used for the protein insertion in liposomes composed of DPhPC).

300

3.2. Electrophysiological recording of Kv1.1 channel activity

301

Electrophysiological characterization of Kv1.1 channel was performed in vitro at the single

302

channel level using bilayers obtained by DIB. The principle of DIB is illustrated in Fig. 2A. The

303

membranes formed between the proteoliposome-containing droplets were first characterized by

304

measuring their resistance and capacitance. Indeed, the microscopic observation of this interface was

305

insufficient to conclude on bilayer formation. Large variations of resistance and capacitance of these

306

bilayers were observed depending on individual experimental setup (Table 1 SI). Frequently, a

307

resistance superior to 10 G was measured (95% of the experiments). The corresponding membranes

308

did not explode when high-voltage pulses of 1.2 V were applied. We therefore concluded that these

309

membranes corresponded to unsuitable multilayer membranes where channels could not properly be

310

inserted. For activity measurements, membranes displaying a resistance of 1-4 GΩ were chosen (5%

311

of events). We also observed that the specific capacitance (Cs) of suitable bilayers formed in the

312

presence of dodecane versus hexadecane were different (0.48 and 0.56 µF/cm²). Gross et al. had

313

already determined membrane capacitance by dynamic control of droplet interface bilayer area [19]

314

and reported that the Cs measured in hexadecane was larger than the one in dodecane. In our studies,

315

the Cs in dodecane was similar to the one reported [19], whereas the one measured in hexadecane was

316

slightly lower (0.56 versus 0.65 µF/cm²). In addition, detected electric noise was higher for

317

membranes prepared in hexadecane than for those prepared in dodecane (2 pA versus 3 pA).

318

Therefore, we concluded that DIB experiments should be performed using dodecane and with

319

bilayers characterized by a resistance of 1 to 4 GΩ.

10 of 19

320

We did not observe the spontaneous fusion of the Kv1.1 proteoliposomes (located in the

321

droplet) with the DPhPC bilayer at the droplet interface. Trials using fusogenic agents like ergosterol

322

or osmotic shocks (as described by Woodbury et al. [20] [21]) did not result in protein insertion either.

323

Fusion could only be obtained by pulse stimulations of 250 ms and 200 mV causing electrical rupture.

324

A bilayer could then be reformed and occasionally ion-channel activity was observed as described by

325

Zagnoni et al. [22]. Using another method, the Droplet-Hydrogel-Bilayers described by Leptihn et al.

326

[23], we could not increase our yield of channel insertion in the artificial membrane. In Fig. 2B an

327

example of Kv1.1 ohmic current is shown. We recorded the intensity of the current at different

328

voltages between -50 and +70 mV in symmetric 150 mM KCl conditions and plotted the I-V curve in

329

Fig. 2C. A conductance of 39 pS was deduced for Kv1.1 channel by calculating the

330

slope-conductance value of the linear regression of this I–V relationship. This value was higher than

331

the 12 pS previously observed [22]. There are several possible explanations for this discrepancy.

332

Firstly, we assumed that the lipid composition may have influenced channel conductance because of

333

the loss of specific protein/lipid interaction, and secondly it is possible that the detection of a 12 pS

334

conductance was hindered by the low resistance of the lipid bilayer and the associated noise level (1-2

335

pA, observed in Fig. 2B) that may have masked the opening events at 12 pS. Records displayed at

336

high resolution clearly demonstrated the occurrence of unitary events. We also observed that the open

337

probability (0.8) of the channel did not depend on the membrane potential. Therefore, we did not

338

observe voltage-dependence or inactivation of the channel. This is not unusual for in vitro channel

339

recording and has for example been described recently by Rosholm et al. for KcsA channel and

340

earlier for N-type calcium channels in bilayers [25] [26]. We suggest that the unusually high

341

conductance (versus the one of 12 pS reported by Bretschneider et al. [24]), the low success rate of

342

protein insertion (only 4% successful experiments with channel insertion), the voltage-independence

343

of the channel and the absence of inactivation over time could be explained by the identity of the

344

lipids used, namely the rigidity of the acyl chain of the DPhPC lipid. Indeed, the hydrophobic

345

interactions between a membrane protein and the host lipid bilayer usually provide an energetic

346

coupling that could explain the small amount of active channels in the DIB membrane and a

347

non-conventional structural position of the gate allowing voltage-independence of the rare active

348

channels [27]. Particularly for voltage-gated potassium channels, a functional role of the surrounding

349

lipids has previously been suggested: in late repolarization, the inactivation has been shown to be the

350

result of contacts between pore domain and the cellular membrane [28]. We hypothesized that

351

voltage-sensitivity could be preserved in vitro in different lipid environments, as lipid-protein

352

interactions influence channel conformation and function through close association with the voltage

353

sensor domains. Indeed, the structure of the Kv1.2 revealed that the pore and voltage sensors were

354

closely embedded in a membrane-like arrangement of lipid molecules [29]. A similar mechanism was

355

expected for the Kv1.1 channel. DPhPC is indeed a zwitterionic lipid, therefore no cationic gating

356

controlled the voltage-dependent opening of the channel through the reorientations of positive

11 of 19

357

charges in the S4 helix [30] [31] and the channel was embedded either in its closed state or (rarely) in

358

its open state.

359 360

Figure 2. A: Schematic of a Droplet Interface Bilayer system described originally by Baley et al. [17]. The oil is

361

represented in yellow and the DPhPC lipid bilayer in blue. 350 nL of proteoliposome (blue dots in the droplets)

362

solution were pipetted onto the electrodes, the droplets were manipulated using a micromanipulator and

363

moved down once a lipid monolayer had been formed (after 10 min), the droplets were then brought into

364

contact in order to form a bilayer (bottom zoom). The insertion of a Kv channel allows the transport of

365

potassium ions through the bilayer. This occurs only when proteoliposomes fuse with the bilayer. B: Typical

366

traces of a single Kv1.1 channel activity recorded with DIB system at -50, 0, +50 and +70 mV. 0 and 1 indicate the

367

closed and open states of the channel, respectively. The open probability (Po) was not affected by the membrane

368

potential applied as indicated above each trace. C: I/V curve of Kv1.1 channel. The slope of the curve gave the

369

unitary conductance of the channel: 39 pS. D: The pictures show the diffusion of 9.2 nL bromophenol blue

370

inside the cis droplet (= 350 nL) in order to illustrate the diffusion after an injection in the DIB system. Blue scale

371

bar, 1 mm. These five pictures were taken at the same time points as the electric traces shown in E. E: Complete

372

recording at +60 mV of Kv1.1 activity. Top: whole 2 min recording after αDTX injection (same time point as the

373

large electric artefact) of 9.2 nL (αDTX final concentration 50 nM). Below: enlargement of four multiple channel

374

activities: a. before injection; and after toxin injections; b. 80 s, c. 120 s and d. 140 s, respectively.

375

In order to demonstrate that DIB can be used to screen ultrasmall amounts of venoms, we first

376

measured the kinetic of the diffusion of a small injection of bromophenol blue in the cis drop of a DIB

377

using a nanoject II pipette (Fig. 2D). We concluded that bromophenol blue diffused completely in the

378

350 nL drop within two minutes. We then used α-dendrotoxin, (αDTX) (courtesy of N. Gilles), a 60

379

amino acid long peptide isolated from the green mamba snake venom [32], to block the DIB the Kv1.1

380

channels, as reported by Wulff et al. [1]. Using the nanoject II we injected a volume of 9.2 nL αDTX

381

in the cis 350 nL drop of a DIB to obtain a 50 nM final solution. This injection blocked two active

12 of 19

382

channels within 1 min as shown at the single channel level in Fig. 2E (top record and four zoomed

383

areas) which is in accordance with our diffusion time determined using bromophenol blue. We could

384

conclude that the NPo (the number of active channels times the open probability) decreased from

385

0.83 to 0.25 until all channels reached closed state during this blockage. For each experiment, we

386

verified that the absence of current was the result of a pharmacological action and not caused by

387

electric circuit failure. To do so, we simply applied a short 1.2 V pulse (zap) to collapse the droplet.

388 389

3.3. Recording of Kv1.1 and Kv1.3 channel activities in liposomes using an optical indicator of membrane potentials

390

Discovery of small molecule pharmacological agents directed against K+ channels depends

391

on the availability of robust, affordable and high-throughput assays. To pursue this aim we have

392

established a simple and cheap assay using oxonol VI, a potential-sensitive fluorescent dye, in order

393

to monitor concentration gradient-driven potassium entry into liposomes through potassium-specific

394

ionophores (valinomycin) or Kv channels. In these experiments, the dye was added from the outside of

395

the liposomes. Precise concentrations of dye (0.15 µM) and lipid (50 µg/mL) were used in order to

396

maximize the partition-dependent fluorescence response in the presence of an inside-positive

397

membrane potential. Previously, an apparent partition coefficient γ = CLipid/Cwater of about 19,000 was

398

calculated by Apell et al. using similar conditions, in the absence of a membrane potential [9]. For the

399

measurements in microtiter plates, the protocol was adapted using valinomycin as a positive control.

400

Experiments were carried out in 3 steps: first the oxonol VI was added to the liposomes where it

401

partially partitions in the lipidic phase, then valinomycin was added, and finally a KCl concentration

402

gradient was created across the membrane of the valinomycin-liposome by the addition of small

403

volumes of KCl solution. The resulting entry of potassium ions in the liposomes triggered an

404

inside-positive membrane potential, which resulted in an increased partition of oxonol VI to the

405

membrane. Consequently, a red shift of the oxonol VI spectrum was observed, as shown in Fig. 3A.

406

Normalized fluorescence intensity changes (F-F0/F0) were measured at an emission wavelength of

407

640 nm and as a function of the extracellular K+ concentration (up to 150 mM KCl) in Fig. 3B.

408 409

Figure 3. Oxonol VI fluorescence in response to inside-positive membrane potentials in liposomes in the

410

presence of potassium selective valinomycin (Val) ionophore (50 nM). A: Fluorescence emission spectra at

13 of 19

411

different KCl concentrations outside the valinomycin-liposomes. The excitation wavelength was 544 ± 8 nm. B:

412

(F- F0)/F0 normalized fluorescence intensity ratio at different KCl concentrations outside liposomes containing

413

valinomycin (blue, grey and yellow) and no valinomycin (orange). The emission wavelength was 640 nm.

414

Different lipids were used: EPC = egg phosphatidylcholine in blue and red; PEPG = mix POPE + POPG (3:1) in

415

grey; L4 mix (DOPE 20%, DOPS 40%, DMPA 20%, cholesterol 20%) in yellow.

416

The entry of potassium ions through valinomycin embedded in membranes of different lipid

417

compositions was monitored by measuring the (F- F0)/F0 normalized fluorescence intensity ratio

418

using identical concentrations of valinomycin and oxonol (Fig. 3C). Similar plateaus at (F-F0)/F0 =

419

0.6 were reached when adding 150 mM KCl outside the liposomes in the presence of valinomycin,

420

indicating that a similar amount of potassium ions had entered the different valinomycin-liposomes.

421

The presence of the anionic lipid POPG did not induce changes of potassium entry versus liposomes

422

containing only PC. Nevertheless, the presence of cholesterol in the L4 mix seemed to delay the entry

423

of potassium. This result could be due to the insertion of less valinomycin in the membrane

424

containing cholesterol. We also verified that the potential across the membrane liposomes containing

425

no

valinomycin

did

not

change

when

adding

KCl.

426 427

Figure 4. Activity measurements of Kv1.1 and Kv1.3 reconstituted in liposomes by the voltage-sensitive dye

428

oxonol VI. A. Fluorescence emission spectra at different KCl concentrations outside the Kv1.3 proteoliposomes

429

(PL). B: (F- F0)/F0 normalized fluorescence intensity ratio using fixed 150 mM KCl concentration outside the PL

430

and as a function of Kv1.3-liposome concentration (expressed as lipid concentration evaluated by thin layer

431

chromatography (data not shown)). C. (F- F0)/F0 normalized fluorescence intensity ratio at different KCl

432

concentrations outside the PL containing Kv1.3 (blue), Kv1.1 (grey) and control experiment in the absence of

433

protein (red).

434

The fluorescence assay was then used to assess the activity of potassium channels. Potassium

435

channels were inserted in liposomes as described in Fig. 1. Active channels would lead to an

436

inside-positive membrane potential in the presence of a K+ concentration gradient across the

437

membrane, in analogy to the K+-specific ionophore valinomycin. This is independent of channel

438

orientation within the membrane. Fluorescence changes upon addition of up to KCl 150 mM were

439

monitored. However, in the case of Kv1.3-liposomes, we did not control the number of potassium

14 of 19

440

channels per liposome. The optimal amount of proteoliposomes was therefore determined as shown

441

in Fig. 4B. Accordingly, a lipid concentration of 20 µg/mL was used in the activity assays described

442

in Fig. 4A and B. We performed the same test for human Kv1.1 channel (data not shown).

443

Fluorescence intensity changes resulting from channel activity are shown in Fig. 4C as a function of

444

added KCl concentration. We did not control the number of channels per liposome, which could

445

explain the difference voltage-dependent response found for Kv1.1 and Kv1.3 proteoliposomes.

446 447

Figure 5. Oxonol VI fluorescence response to membrane potential changes in Kv1.3 proteo-liposomes blocked

448

by toxins. A. (F- F0)/F0 normalized fluorescence intensity ratio using 150 mM KCl concentrations outside the

449

Kv1.3-liposomes and 0, 3 and 67 nM ShK, respectively. B and C. (F- F0)/F0 normalized fluorescence intensity

450

ratio using 150 mM KCl concentrations outside the Kv1.3-liposomes and various concentrations of margatoxin

451

(MTX).

452

Testing toxins using channels embedded in liposomes requires that their binding site is

453

accessible outside the liposomes. Using the channel orientation protocol, we have shown that the

454

Kv1.3 channel was oriented with the N-terminal tag inside the liposome, making the toxin binding site

455

accessible on the outside [1]. Therefore, channel blockage using ShK toxin and margatoxin could be

456

assayed. These peptides were already known to block Kv1.3 [33, 34]. The ShK toxin was found in the

457

sea anemone Stichodactyla helianthus and is a known, potent blocker of Kv1.3. It is currently being

458

developed as a therapeutic peptide for autoimmune diseases. Margatoxin (MTx) is a toxin from the

459

scorpion Centruroides margaritatus and is a high affinity inhibitor of Kv1.3 (Kd = 11.7 pM using

460

patch clamp electrophysiological recording). This toxin, however, is not selective for Kv1.3. The

461

fluorescent dye oxonol VI was used in order to monitor channel activity in the presence of these two

462

toxins. We found in Fig. 3A that 3 nM and 67 nM Shk were both able to block mouse Kv1.3. No K+

463

flux across the membrane was observed in presence of a K+ gradient as deduced from the

464

fluorescence intensity of oxonol VI at 640 nm which was identical to that of Kv1.3-liposome in the

465

absence of KCl (inactive channel). For margatoxin we showed in Fig. 3B that the channel was

466

blocked at 5 nM and calculated that the IC50 in our assay was 1 nM (Fig. 5C). This value is higher than

467

the one found in electrophysiological experiments but both assays cannot be compared. The

468

fluorescent assay concerned channels population in wells, the patch clamp assay focused on a single

469

channel.

470

4. Conclusion

15 of 19

471

Previous work has shown that cell free expression can be used to produce Kv channels

472

embedded in liposomes [6]. The orientation of the protein is crucial to perform pharmacological

473

studies. The present study demonstrates that channels expressed through in vitro techniques can be

474

reconstituted in liposomes and that they are electrophysiologically active. Here we compared two

475

complementary approaches: electrophysiological measurements in droplet interface bilayers and

476

optical determination of the build-up of an inside positive membrane potential due to K+ influx into

477

liposomes using a voltage-sensitive fluorescent dye. In the case of electrophysiological

478

measurements using the DIB technique, formation of suitable bilayers and consequent insertion of K v

479

channels seldom occurred, thus excluding this approach for systematic screening of a large number of

480

different compounds. Therefore, we developed another assay that could be performed in microtiter

481

plates using the fluorescent dye oxonol VI. This approach was based on the detection of changes in

482

membrane potential due to K+ influx through active Kv1.1 and Kv1.3 channels reconstituted in

483

liposomes. This assay was developed using valinomycin as positive control for K+ entry in liposomes

484

and was reported to let 104 ions/s through one molecule which is less than the 108 ions/s value

485

reported for one Kv1 channel [11]. Oxonol fluorescence probes the partition between water and

486

membranes, which is voltage-dependent. Smith [35] reported that oxonol VI gave the largest spectral

487

shift, with an isosbestic point at 603 nm. In addition, oxonol VI responds more rapidly to changes in

488

potential than oxonol V and is therefore considered to be a better probe for measuring fast potential

489

changes.

490

While the build-up of the inside positive membrane potential is independent of the channel

491

orientation in the liposome membrane, only channels with an accessible toxin binding site (i.e. at the

492

outside of the liposomes) can be blocked. Care has therefore to be taken that channels are inserted in

493

the correct orientation into the liposomes, which in our case could be demonstrated. Under these

494

conditions, the use of Oxonol VI fluorescent dyes can provide valuable information in order to

495

screen small molecules for their channel blocking activity.

496

Altrichter et al. [36] have recently demonstrated that electrostatic attraction between

497

the protein charges and the lipid polar heads are crucial during protein insertion into

498

liposomes. This step precedes the hydrophobic residues insertion into the apolar hydrocarbon

499

chains of the membrane. Therefore the role of anionic lipids in the efficacy of the protein

500

insertion should be considered in the future.

501

This work used the platforms of the Grenoble Instruct Center (ISBG; UMS 3518

502

CNRS-CEA-UJF-EMBL) with support from FRISBI (ANR-10-INSB-05-02) and GRAL

503

(ANR-10-LABX-49-01) within the Grenoble Partnership for Structural Biology (PSB). Research

504

reported in this publication was supported by the ANR VenomPicoScreen (ANR-11-RPIB-022-04).

505

We thank Nicolas Gilles who was the soul of this project. M.D.W. is also supported by the Labex Ion

506

Channels, Science and Therapeutics (ANR-11-LABX-0015). We thank Lionel Imbert for

16 of 19

507

assistance and access to the cell-free expression platform, Cécile Breyton and Christine Ebel for

508

scientific discussion.

509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526

Supplementary figure and table: Anti His tag 130 kDa 70 kDa

55 kDa

Fig. 1S: Kv1.1 protein quality was assessed by Western blot using a rabbit HRP-coupled anti His tag antibody on the left and with C-terminal antibody on the right (Abcam, USA, ab177481).

Resistance

527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543

Anti Kv1.1 Cter

Cs12 Cs16 1-4 GΩ 1-4 GΩ 0,48±0,08 0,56±0,02a µF/cm2 µF/cm2 2 pAd 3 pAd

Capacitance (0 mV) Noise intensity (0 mV) Table 1: resistance and specific capacitance (Cs) of DIB membranes. n=4 for C12 (dodecane) and C16 (hexadecane).

[1] H. Wulff, N.A. Castle, L.A. Pardo, Voltage-gated potassium channels as therapeutic targets, Nat Rev Drug Discov, 8 (2009) 982-1001. [2] S. Chhabra, S.C. Chang, H.M. Nguyen, R. Huq, M.R. Tanner, L.M. Londono, R. Estrada, V. Dhawan, S. Chauhan, S.K. Upadhyay, M. Gindin, P.J. Hotez, J.G. Valenzuela, B. Mohanty, J.D. Swarbrick, H. Wulff, S.P. Iadonato, G.A. Gutman, C. Beeton, M.W. Pennington, R.S. Norton, K.G. Chandy, Kv1.3 channel-blocking immunomodulatory peptides from parasitic worms: implications for autoimmune diseases, FASEB journal : official publication of the Federation of American Societies for Experimental Biology, 28 (2014) 3952-3964. [3] R. Chen, A. Robinson, D. Gordon, S.H. Chung, Modeling the binding of three toxins to the voltage-gated potassium channel (Kv1.3), Biophysical journal, 101 (2011) 2652-2660. [4] C. Beeton, K.G. Chandy, Potassium channels, memory T cells, and multiple sclerosis, The Neuroscientist : a review journal bringing neurobiology, neurology and psychiatry, 11 (2005) 550-562.

17 of 19

544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586

[5] K.G. Chandy, H. Wulff, C. Beeton, M. Pennington, G.A. Gutman, M.D. Cahalan, K+ channels as targets for specific immunomodulation, Trends in pharmacological sciences, 25 (2004) 280-289. [6] Z. Su, E.C. Brown, W. Wang, R. MacKinnon, Novel cell-free high-throughput screening method for pharmacological tools targeting K+ channels, Proceedings of the National Academy of Sciences of the United States of America, 113 (2016) 5748-5753. [7] K.K. Deal, D.M. Lovinger, M.M. Tamkun, The brain Kv1.1 potassium channel: in vitro and in vivo studies on subunit assembly and posttranslational processing, J Neurosci, 14 (1994) 1666-1676. [8] S. Leptihn, J.R. Thompson, J.C. Ellory, S.J. Tucker, M.I. Wallace, In vitro reconstitution of eukaryotic ion channels using droplet interface bilayers, J Am Chem Soc, 133 (2011) 9370-9375. [9] H.J. Apell, B. Bersch, Oxonol VI as an optical indicator for membrane potentials in lipid vesicles, Biochimica et biophysica acta, 903 (1987) 480-494. [10] B. Chanda, M.K. Mathew, Functional reconstitution of bacterially expressed human potassium channels in proteoliposomes: membrane potential measurements with JC-1 to assay ion channel activity, Biochimica et biophysica acta, 1416 (1999) 92-100. [11] P. Pouliquin, J. Grouzis, R. Gibrat, Electrophysiological study with oxonol VI of passive NO3- transport by isolated plant root plasma membrane, Biophysical journal, 76 (1999) 360-373. [12] in. https://www.smartox-biotech.com/product/potassium-channel-blocker/margatoxin [13] in. http://www.synthelis.com/products/mouse-kv1-3-proteoliposome-pl_kv1-3_004/ [14] D. Schwarz, C. Klammt, A. Koglin, F. Lohr, B. Schneider, V. Dotsch, F. Bernhard, Preparative scale cell-free expression systems: new tools for the large scale preparation of integral membrane proteins for functional and structural studies, Methods, 41 (2007) 355-369. [15] M.T. Paternostre, M. Roux, J.L. Rigaud, Mechanisms of membrane protein insertion into liposomes during reconstitution procedures involving the use of detergents. 1. Solubilization of large unilamellar liposomes (prepared by reverse-phase evaporation) by triton X-100, octyl glucoside, and sodium cholate, Biochemistry, 27 (1988) 2668-2677. [16] E.R. Geertsma, N.A. Nik Mahmood, G.K. Schuurman-Wolters, B. Poolman, Membrane reconstitution of ABC transporters and assays of translocator function, Nature protocols, 3 (2008) 256-266. [17] H. Bayley, B. Cronin, A. Heron, M.A. Holden, W.L. Hwang, R. Syeda, J. Thompson, M. Wallace, Droplet interface bilayers, Mol Biosyst, 4 (2008) 1191-1208. [18] C. Menager, D. Guemghar, V. Cabuil, S. Lesieur, Interaction of n-octyl beta,D-glucopyranoside with giant magnetic-fluid-loaded phosphatidylcholine vesicles: direct visualization of membrane curvature fluctuations as a function of surfactant partitioning between water and lipid bilayer, Langmuir : the ACS journal of surfaces and colloids, 26 (2010) 15453-15463. [19] L.C. Gross, A.J. Heron, S.C. Baca, M.I. Wallace, Determining membrane capacitance by dynamic control of droplet interface bilayer area, Langmuir : the ACS journal of surfaces and colloids, 27 (2011) 14335-14342.

18 of 19

587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629

[20] D.J. Woodbury, Nystatin/ergosterol method for reconstituting ion channels into planar lipid bilayers, Methods in enzymology, 294 (1999) 319-339. [21] M. Zagnoni, Miniaturised technologies for the development of artificial lipid bilayer systems, Lab on a chip, 12 (2012) 1026-1039. [22] M. Zagnoni, M.E. Sandison, P. Marius, A.G. Lee, H. Morgan, Controlled delivery of proteins into bilayer lipid membranes on chip, Lab on a chip, 7 (2007) 1176-1183. [23] S. Leptihn, O.K. Castell, B. Cronin, E.H. Lee, L.C. Gross, D.P. Marshall, J.R. Thompson, M. Holden, M.I. Wallace, Constructing droplet interface bilayers from the contact of aqueous droplets in oil, Nature protocols, 8 (2013) 1048-1057. [24] F. Bretschneider, A. Wrisch, F. Lehmann-Horn, S. Grissmer, Expression in mammalian cells and electrophysiological characterization of two mutant Kv1.1 channels causing episodic ataxia type 1 (EA-1), The European journal of neuroscience, 11 (1999) 2403-2412. [25] M. De Waard, D.R. Witcher, K.P. Campbell, Functional properties of the purified N-type Ca2+ channel from rabbit brain, The Journal of biological chemistry, 269 (1994) 6716-6724. [26] K.R. Rosholm, M.A. Baker, P. Ridone, Y. Nakayama, P.R. Rohde, L.G. Cuello, L.K. Lee, B. Martinac, Activation of the mechanosensitive ion channel MscL by mechanical stimulation of supported Droplet-Hydrogel bilayers, Scientific reports, 7 (2017) 45180. [27] J.A. Lundbaek, Regulation of membrane protein function by lipid bilayer elasticity-a single molecule technology to measure the bilayer properties experienced by an embedded protein, Journal of physics. Condensed matter : an Institute of Physics journal, 18 (2006) S1305-1344. [28] E.A. van der Cruijsen, D. Nand, M. Weingarth, A. Prokofyev, S. Hornig, A.A. Cukkemane, A.M. Bonvin, S. Becker, R.E. Hulse, E. Perozo, O. Pongs, M. Baldus, Importance of lipid-pore loop interface for potassium channel structure and function, Proceedings of the National Academy of Sciences of the United States of America, 110 (2013) 13008-13013. [29] S.B. Long, X. Tao, E.B. Campbell, R. MacKinnon, Atomic structure of a voltage-dependent K+ channel in a lipid membrane-like environment, Nature, 450 (2007) 376-382. [30] Y. Jiang, V. Ruta, J. Chen, A. Lee, R. MacKinnon, The principle of gating charge movement in a voltage-dependent K+ channel, Nature, 423 (2003) 42-48. [31] D. Schmidt, Q.X. Jiang, R. MacKinnon, Phospholipids and the origin of cationic gating charges in voltage sensors, Nature, 444 (2006) 775-779. [32] A.L. Harvey, Twenty years of dendrotoxins, Toxicon : official journal of the International Society on Toxinology, 39 (2001) 15-26. [33] A. Bartok, A. Toth, S. Somodi, T.G. Szanto, P. Hajdu, G. Panyi, Z. Varga, Margatoxin is a non-selective inhibitor of human Kv1.3 K+ channels, Toxicon : official journal of the International Society on Toxinology, 87 (2014) 6-16. [34] V. Chi, M.W. Pennington, R.S. Norton, E.J. Tarcha, L.M. Londono, B. Sims-Fahey, S.K. Upadhyay, J.T. Lakey, S. Iadonato, H. Wulff, C. Beeton, K.G. Chandy, Development of a sea anemone toxin as an immunomodulator for therapy of autoimmune diseases, Toxicon : official journal of the International Society on Toxinology, 59 (2012) 529-546.

19 of 19

630 631 632 633 634 635 636

[35] J.C. Smith, Potential-sensitive molecular probes in membranes of bioenergetic relevance, Biochimica et biophysica acta, 1016 (1990) 1-28. [36] S. Altrichter, M. Haase, B. Loh, A. Kuhn, S. Leptihn, Mechanism of the Spontaneous and Directional Membrane Insertion of a 2-Transmembrane Ion Channel, ACS chemical biology, 12 (2017) 380-388.