Articles in PresS. J Neurophysiol (September 5, 2012). doi:10.1152/jn.01033.

2011

ROLE

OF

Ih

THERMORECEPTOR ENDINGS

IN

THE

FIRING

PATTERN

OF

MAMMALIAN

COLD

3
4

Patricio Orio1, Andrs Parra2, Rodolfo Madrid3, Omar Gonzlez2,4, Carlos Belmonte2 and

Flix Viana2.

Universidad de Valparaso, 2360102 Valparaso, Chile.

Alicante, Spain.

Centro Interdisciplinario de Neurociencia de Valparaso CINV, Facultad de Ciencias,

Instituto de Neurociencias de Alicante, Universidad Miguel Hernndez-CSIC, 03550

10

11

Chile, 9160000 Santiago, Chile.

12

13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33

Departamento de Biologa, Facultad de Qumica y Biologa, Universidad de Santiago de

Fundacin de Investigacin Oftalmolgica, Instituto Fernandez-Vega, Oviedo (Spain).

Running Head: Role of h-currents in cold thermoreceptor endings

Author contributions: P.O., R.M., F.V., and C.B. designed the experiments. A.P., O.G.
and R.M. performed the experiments. P.O. and A.P. analyzed the experimental data. P.O.
performed and analyzed mathematical simulations. P.O. and F.V. wrote the paper.

Contact Information:
Dr. Patricio Orio
Centro Interdisciplinario de Neurociencia de Valparaso
Universidad de Valparaso
Gran Bretaa 1111
2360102 Valparaiso, Chile
Tel: 56-32-2508040
e-mail: patricio.orio@uv.cl

1
Copyright 2012 by the American Physiological Society.

34

ABSTRACT

35

Mammalian peripheral cold thermoreceptors respond to cooling of their sensory endings

36

with an increase in firing rate and a modification of their discharge pattern. We recently

37

showed that cultured trigeminal cold-sensitive (CS) neurons express a prominent

38

hyperpolarization activated current (Ih), mainly carried by HCN1 channels, supporting

39

subthreshold resonance in the soma without participating in the response to acute cooling.

40

However, peripheral pharmacological blockade of Ih, or characterization of HCN1-/- mice,

41

still reveals a deficit in cold perception. Here we investigated the role of Ih in CS nerve

42

endings, where cold sensory transduction actually takes place. Corneal CS nerve endings in

43

mice show a rhythmic spiking activity at neutral skin temperature that switches to bursting

44

mode when the temperature is lowered. The Ih blockers ZD7288 and Ivabradine alter the

45

firing patterns of CS nerve endings, lengthening the inter-spike intervals and inducing

46

bursts at neutral skin temperature. We characterized the CS nerve endings from HCN1-/-

47

mouse corneas and found that they behave similar to the wild type, although with a lower

48

slope in the firing frequency versus temperature relationship thus explaining the deficit in

49

cold perception of HCN1-/- mice. The firing pattern of nerve endings from HCN1-/- mice

50

was also affected by ZD7288, a result that we attribute to the presence of HCN2 channels

51

in the place of HCN1. Mathematical modeling shows that the firing phenotype of CS nerve

52

endings from HCN-/- mice can be reproduced by replacing HCN1 channels with the slower

53

HCN2 channels rather than by abolishing Ih. We propose that Ih carried by HCN1 channels

54

helps tuning the frequency of the oscillation and the length of bursts that underlies regular

55

spiking in cold thermoreceptors, having important implications for the neural coding of

56

cold sensation.
2

57
58

Keywords: Hyperpolarization-activated current, HCN channel, sub-threshold oscillation,

bursting, cold thermoreceptors.

59

INTRODUCTION

60

Mammalian cold thermoreceptors are sensory terminals specialized in the detection of

61

innocuous cold/cool temperatures impinging on the skin (Hensel 1981). The transduction of

62

cold stimuli takes place in the free nerve endings of the peripheral axons of cold

63

thermoreceptor neurons from trigeminal and dorsal root ganglia. They display spontaneous,

64

tonic or bursting, spike activity at neutral skin temperatures that is accelerated by cooling

65

the receptive field and suppressed by warming (Braun et al. 1980; Brock et al. 2001; Dykes

66

1975; Iggo 1969). The average frequency of their static discharge has a bell-shaped

67

relationship with respect to the skin temperature, with the maximum frequency of single

68

cold thermoreceptors ranging from 18C to 34C. Therefore, the action potential rate cannot

69

encode unambiguously the peripheral temperature under static conditions (Bade et al. 1979;

70

Dykes 1975; Hensel and Wurster 1970) suggesting that discrimination depends on the

71

temporal structure of the impulse sequence (Braun et al. 1980; Hensel 1981). At static

72

temperatures above ~30C the activity consists mainly in regularly fired single spikes,

73

while at lower temperatures the bursting activity prevails (Iggo 1969). The transitions

74

between these different spiking patterns are continuous, and their temporal structure

75

suggests that all the different patterns can be attributed to oscillations of the membrane

76

potential which are systematically altered by temperature (Braun et al., 1980; Braun et al.,

77

1990; Schafer et al., 1991). Acute cooling leads to a transient rise in mean frequency

78

caused by an immediate increase in the frequency of bursts followed by an increase in burst

79

duration and number of impulses per burst while the frequency of bursts decreases (Braun

80

et al. 1980). There is growing experimental evidence that bursting is involved in the

81

transmission of behaviorally relevant information in other sensory systems (Krahe and

4

82

Gabbiani 2004) and it has been proposed that bursting enhances the sensitivity and working

83

range of cold receptors (Iggo 1969). However, the contribution of this temporal firing code

84

to the psychophysical capacities for cold discrimination in vivo has not been established.

85

The cellular and molecular mechanisms that confer specificity for thermal detection at cold

86

nerve endings have been unveiled in recent times. Cooling and cooling compounds, like

87

menthol, activate TRPM8, a non-selective, calcium-permeable, cation channel of the

88

transient receptor potential family (reviewed in Babes et al. 2011; Latorre et al. 2011;

89

McKemy et al. 2002; Peier et al. 2002). This channel is selectively expressed in cold

90

sensitive (CS) neurons and is a major determinant of their cold sensitivity (Bautista et al.

91

2007; Colburn et al. 2007; Dhaka et al. 2007). CS neurons also possess background K+

92

channels that are closed by cooling contributing to depolarization and firing (Reid and

93

Flonta 2001; Viana et al. 2002), and it has been shown that these channels participate in

94

cold perception (Noel et al. 2009). In contrast, the ionic mechanisms underlying the

95

rhythmic firing activity exhibited by peripheral cold receptors are still unknown.

96

Experimental evidence suggests the involvement of slow, TTX-resistant persistent sodium

97

currents (Brock et al., 1998; Herzog et al., 2001) and low-threshold calcium channels

98

(Schafer et al., 1982; Schafer et al., 1991) in the underlying oscillations that generate

99

rhythmic firing. Mathematical simulations (Braun et al. 1998; Longtin and Hinzer 1996)

100

have shown that the usual effect of temperature on ion channel gating kinetics (Q10~3,

101

(Hille 2001)) is enough to reproduce the temperature-induced change of spiking pattern in a

102

slow wave bursting model (Plant 1981).

103

Recently, we described the biophysical characteristics of the hyperpolarization-activated

104

cation current (Ih) in peripheral cold thermoreceptors and its specific role in cold sensitivity
5

105

(Orio et al. 2009). The Ih is a key regulator of cellular excitability and contributes to

106

rhythmic firing in neurons at various levels of the nervous system (Pape 1996) and the heart

107

(DiFrancesco 2006; Schreiber 2003). For example, Ih participates in the generation of

108

spindle waves and single cell oscillations in the thalamus (Luthi et al. 1998; McCormick

109

and Pape 1990), and subthreshold resonance behavior in cortical cells (Hutcheon et al.

110

1996; Richardson et al. 2003) and hypoglossal motoneurons (van Brederode and Berger

111

2011). Ih is carried through channels of the hyperpolarization-activated and cyclic

112

nucleotide-gated (HCN) ion channel subfamily (reviewed by Hofmann et al. 2005; Pape

113

1996; Robinson and Siegelbaum 2003), which include four members (HCN1-4)

114

characterized by six membrane-spanning domains and one cyclic nucleotide-binding

115

domain (Ludwig et al. 1998; Santoro et al. 1998). Individual HCN subunits assemble as

116

homotetrameric channels with distinct voltage dependency and kinetic properties (reviewed

117

by Kaupp and Seifert 2001). Further diversity is obtained by co-assembly of the different

118

HCN subunits (Chen et al. 2001; Ulens and Tytgat 2001). In CS neurons, the kinetic and

119

activation properties of Ih suggest a major involvement of HCN1 channels with some

120

participation of the HCN2 isoform (Orio et al. 2009). Indeed, CS neurons from HCN1 gene

121

knock-out mice show no Ih or a slower Ih of reduced amplitude. Behavioral responses to

122

cooling are impaired in HCN1 null mice and after peripheral pharmacological blockade of

123

Ih (Orio et al. 2009), suggesting an important role of Ih in cold thermoreception. However,

124

the actual role of Ih in the process of thermotransduction and coding of cold temperatures

125

by sensory terminals remained obscure. It is imperative therefore to record the activity of

126

CS nerve endings and observe the effect of Ih manipulation to be able to assign a specific

127

role of this current in cold perception. Due to their small size and subsurface locations,

128

electrophysiological recordings from mammalian sensory endings are extremely difficult.

6

129

Fortunately, a preparation that allows long-term extracellular recordings from corneal

130

sensory nerve endings in-vitro allowed us to pursue this goal (Brock et al. 1998; Parra et al.

131

2010). Here, we characterize the effects of pharmacological blockade of Ih in the

132

spontaneous and temperature-dependent firing patterns of cold sensitive nerve endings in

133

the cornea, and look at the firing patterns in cold themoreceptors from HCN1-/- mice. A

134

change of the firing pattern upon blockade of Ih with ZD7288 or Ivabradine suggests an

135

important role of HCN channels in tuning and maintaining the slow oscillation underlying

136

regular firing patterns. In CS nerve endings from HCN1-/- mice corneas, firing patterns

137

show less marked changes than those we observed under pharmacological suppression of

138

HCN channels, when compared to wild-type data. Based on previous results (Orio et al.

139

2009), we propose that this is due to functional expression of HCN2 channels in CS nerve

140

endings, a hypothesis that is further supported by mathematical modeling. We conclude that

141

Ih plays a critical role in shaping the temperature-dependent firing patterns in cold

142

thermoreceptors, determining the oscillation period and the length of bursts that underlie

143

their rhythmic activity.

144

MATERIALS AND METHODS

145

Animals

146

All animal experiments were conducted at the Institute of Neurosciences in Alicante

147

(Spain). Experimental protocols were supervised and approved by the Director of Animal

148

Research Services, and were performed according to EC animal use guidelines

149

(2007/526/EC). 20 HCN1+/+ mice and 7 HCN1-/- mice were used. Animals were sacrificed

150

by exposure to CO2 and decapitated. Eyes were carefully removed and placed in a

151

recording chamber. The mouse line HCN1-/- was generated by the laboratory of E. Kandel

152

(Columbia University) and obtained from The Jackson Laboratory (stock number 101045).

153

Mice were genotyped by PCR.

154

Extracellular recording of single nerve terminal impulses

155

To characterize the firing of cold-sensitive nerve endings, we used an in vitro preparation

156

of mouse cornea developed in the laboratory, as described recently in Parra et al. (2010).

157

Briefly, the optic nerve and associated tissues of the eye were drawn into a suction tube at

158

the bottom of a small recording chamber and eyes were continuously perfused (1 ml/min)

159

with a solution of the following composition (mM): NaCl (128), KCl (5), NaH2PO4 (1),

160

NaHCO3 (26), and glucose (10). The solution was gassed with carbogen (95%O2, 5%CO2)

161

and maintained at the desired temperature with a manually-controlled Peltier device. Glass

162

micropipette electrodes filled with the same solution were applied to the cornea with light

163

suction to record nerve terminal activity. Cold sensory receptors were identified by their

164

relatively high level of spontaneous discharge, which is increased by cooling of the

165

superfusing solution (Brock et al. 2001). Signals were amplified with an AC amplifier

166

(Neurolog NL104, Digitimer Ltd, Welwyn Garden City, UK; gain x2000, high pass filter
8

167

set at 0.1 Hz). Data were captured and analyzed using a CED 1401 interface coupled to a

168

computer running Spike 2 software (Cambridge Electronic Design, Cambridge, UK).

169

Temperature was simultaneously acquired at 10 samples/s.

170

Analysis of burst patterns

171

Nerve terminal impulses (NTIs) were detected during acquisition with a threshold-based

172

criterion and further filtered manually to remove artifacts. Only single-unit recordings

173

(evidenced by a spikes having the same shape and similar size), or multi-unit recordings

174

where spike sorting was easily achieved by different spike shapes, were used. Following

175

sorting, inter-spike intervals (ISIs) for individual units were calculated. A burst is defined

176

to begin when the ratio of two consecutive ISIs (ISIn-1:ISIn) is higher than 2.5. Conversely,

177

a burst ends when the ratio ISIn-1:ISIn is lower than 0.4. Also, a maximum intra-burst

178

interval (MaxIntra) and a minimum inter-burst interval (MinInter) were defined such that if

179

ISI > MaxIntra it was automatically considered as an inter-burst interval, and if ISI <

180

MinInter it was automatically treated as an intra-burst interval. Typically, MinInter = 40 ms

181

and MaxIntra = 100 ms, but these values were adjusted case by case based on the ISI

182

histograms and the temperature at which the analyzed recording was obtained. The

183

following firing pattern parameters were defined: Mean firing frequency: average number

184

of NTIs recorded per second. Spikes per burst: number of NTIs observed within a burst.

185

Cycle time (CT): Time interval between the first NTI of a burst and the first NTI of the next

186

burst. When there is only one NTI per burst (no bursting), CT = ISI. To quantify better the

187

period of the slow oscillation, when a high skipping rate (see below) was evident a

188

histogram of CT values was obtained and the first peak (corresponding to the shortest CT)

189

was defined as the main cycle of the oscillation (1stCT). Skip events. A skip event (Braun et

190

al. 1980; Longtin and Hinzer 1996) was defined when ISI > 1stCT * 0.6.

191

Mathematical modeling of cold-sensitive nerve endings

192

The electrical activity of cold-sensitive nerve endings was modeled using a modified

193

version of the Huber-Braun model (Braun et al. 1998) that includes an hyperpolarization-

194

activated current (Ih) and a TREK1-like potassium current. The change in membrane

195

voltage (V) is given by

196

Cm

dV
= I l I d I r I sd I sr I h I trek g w ,
dt

(1)

197

where Cm is membrane capacitance, Ii are ionic membrane currents and gw is a noise term. Il

198

stands for leak current, such that Il=gl(V-El), being g the conductance and E the reversal

199

potential. Id and Isd are depolarizing and slow depolarizing currents, respectively. Id

200

represents a fast sodium channel that generates action potentials while Isd is a mixed sodium

201

and calcium slow conductance (Longtin and Hinzer 1996; Plant 1981); a simplification that

202

assumes separate calcium and sodium channels with similar activation properties. The

203

activation properties of Isd resembles NaV1.9 channels (Herzog et al. 2001) and the fraction

204

of calcium conductance is controlled by the parameter (see below). Ir and Isr are

205

repolarizing and slow repolarizing currents (i.e. potassium currents), respectively. Itrek is a

206

temperature-sensitive potassium current. Currents are given by

I i = g i a i (V E i ) ,

207

(2)

208

where a is an activation parameter and is a temperature scaling factor for the current. ar,

209

ah and asd are governed by

210

da i
a ai
1
= i
and a i =
.
dt
i
1 + exp ( s i (V Vi 0 ))
10

(3),(4)

211

where is a temperature factor for channel kinetics. ad is considered to be instantaneous

212

such that ad = a d , defined as in eq. (4)

213

asr resembles a calcium activated potassium channel with the expressions

a sr =

214
215

Ca n
Ca n + K d

I sd Ca
dCa
=
.
dt
Ca

(5),(6)

Finally,

0.8

atrek = 0.2 +

216

217

and

T T
1 + exp q10 m

10

(7)

Temperature scaling factors are given by

= 1.3

218

T 25
10

, = 3.0

T 25
10

219

The noise component (gw) was modeled as an exponentially correlated noise in the form of

220

an Ornstein-Uhlenbeck process:

dg w
g +
= w
dt
w

221
222

where is drawn at each integration step from a normal distribution with mean 0 and

223

variance D.

224

Parameters used for the wt model are:

225

Cm=1, gd=2.5, gr=2.8, gsd=0.21, gsr=0.28, gh=0.6, gtrek=0.06, gl=0.022, Ed=Esd=50,

226

Er=Esr=Etrek=-90, Eh=-30, El=-60, Vd0=Vr0=-25, Vsd0=-40, Vh0=-85, sd=sr=0.25, ssd=0.11,

227

sh=-0.14, r=2, sd=10, Ca=35, h=125, Kd=0.4, n=2, =14, =0.18, q10=5, Tm=29, w=1,

228

D=0.2.

229

For the slow Ih model the only differences are Vh0=-95, h=900.

230

Units are F/cm2, mS/cm2, mV, ms and M.

11

(8)

(9)

231

The model was implemented and solved in the NEURON simulation environment

232

(http://www.neuron.yale.edu) (Hines and Carnevale 1997). All the files necessary to run the

233

simulations

234

(http://senselab.med.yale.edu/ModelDB/, accession number 141063).

235

Statistical Analysis

236

Unless stated otherwise, data is expressed as mean S.E.M. Data was compared using

237

Students t-test. Paired tests were applied when appropriate.

238

Drugs

239

ZD7288 (4-Ethylphenylamino-1,2-dimethyl-6-methylaminopyrimidinium chloride) was

240

purchased

241

dimethoxybicyclo[4,2,0]octa-1,3,5-trien-7-yl)methyl]

242

tetrahydro-7,8-dimethoxy-2H-3-benzazepin-2-one, hydrochloride) was a generous gift of

243

the Institut de Recherches Internationales Servier (France).

presented

from

Tocris

here

are

Bioscience

244
245

12

available

(UK).

Ivabradine

in

ModelDB

(3-(3-{[((7S)-3,4-

methylamino}propyl)-1,3,4,5-

246

RESULTS

247

Blockade of Ih alters spontaneous firing activity but not acute response to cold in nerve

248

terminals.

249

To investigate the function of Ih at sensory nerve terminals, we took advantage of an in

250

vitro preparation of the adult mouse cornea that allows long term recordings of cold

251

thermoreceptor endings (Parra et al. 2010). Figure 1A shows extracellular recordings of

252

single nerve terminal impulses in a CS nerve terminal. The time course of the recorded

253

electrical activity in the same nerve ending can be seen in Figure 1B, which plots the firing

254

frequency, the inter-spike intervals, the number of spikes per burst and the temperature of

255

the bathing solution during the complete experiment. In the control condition (Figure 1A,

256

top left) the terminal shows spontaneous nerve impulses with a frequency around 7

257

impulses/s. When cooled transiently to 20C, the terminal responded with an increase in

258

mean firing frequency up to 30 impulses/s (Figure 1B). Following application of 20 M

259

ZD7288, a selective blocker of HCN channels (Harris and Constanti 1995), to the bath

260

solution perfusing the cornea, the most obvious effect was a slowdown of the mean spiking

261

frequency (an increase of the mean ISI) shortly followed by the appearance of bursts

262

(evidenced by the appearance of shorter ISIs and an increase in the number of spikes per

263

bursts, see Figure 1B and Figure 1A, right). On the other hand, application of ZD7288 did

264

not induce significant changes in the shape of NTI spikes (Figure 1D and Table 1),

265

indicating that ZD7288 is not affecting the active mechanisms responsible for action

266

potential generation and propagation.

267

Remarkably, the terminals still retained their dynamic responsiveness to cold, showing an

268

increased firing frequency during rapid cooling steps (Figure 1B, compare the two cold
13

269

pulses). The maximum firing frequency during the cooling stimulus showed no difference

270

between the control condition and when ZD7288 was present (Table 2). Moreover, Table 2

271

shows that application of the drug did not change significantly the threshold for the increase

272

in firing frequency in response to cold, in agreement with our previous findings in cultured

273

CS trigeminal neurons where ZD7288 did not change the temperature threshold for action

274

potential firing (Orio et al. 2009). Finally, ZD7288 did not change the bursting

275

characteristics of the response to cooling ramps (Table2).

276

From these observations we conclude that blockade of HCN channels in CS nerve terminals

277

does not affect the response to an acute cold stimulus, but alters profoundly the temporal

278

pattern of the spontaneous firing of action potentials in a steady-state temperature

279

condition. We also tested the effect of ZD7288 in CS nerve endings from guinea pig

280

corneas (Brock et al. 1998) and found very similar results (not shown).

281

Blockade of Ih alters stationary firing pattern at low temperatures

282

We wanted to examine whether the changes of firing pattern observed upon Ih blockade

283

could also be observed at temperatures lower than the resting temperature of 34C. Figure 2

284

shows the behavior of a mouse CS nerve ending that was exposed to 20 M ZD7288 while

285

maintained at a constant temperature of 27C. In this case, the terminal showed a bursting

286

firing pattern right from the beginning of the recording (Figure 2A, left, and 2B), typical for

287

temperatures lower than 30C. After some minutes in the presence of ZD7288, it becomes

288

evident that the firing pattern suffers the same alteration as seen at 34C; namely the

289

lengthening of the inter-burst period (the longest inter-spike intervals) (Figure 2A, right,

290

and also see the intervals in 2B) and augmented bursting. When summarizing the results of

291

6 similar experiments at 27C, we see the same overall result as obtained with the protocol

14

292

used in Figure 1: there is a significant increase in cycle time and in the number of spikes

293

per burst, while the firing frequency is not significantly affected (Figure 2C).

294

Overall, these results indicate that the participation of Ih in the oscillatory properties of CS

295

nerve endings is also required for their characteristic bursting behaviors at lower

296

temperatures, and therefore may be critical for the neural coding of stationary temperatures.

297

Firing patterns of corneal cold sensitive nerve endings in wild type and HCN1 null mice

298

To investigate further the potential role of HCN channels in the coding of cold-evoked

299

responses, we used an alternative approach to the pharmacological blockade of Ih, taking

300

advantage of an HCN1 null mouse line (Nolan et al. 2003). Figure 3A shows representative

301

recording traces and inter-spike intervals (ISI) histograms for the steady state firing of NTIs

302

at 34C, 30C and 25C, for terminals recorded in corneas from HCN1+/+ (left) and HCN1-/-

303

(right) animals. In both cases, the decrease in temperature is associated to longer cycle

304

times (the larger ISIs) and bursting becoming more prominent (higher number of events

305

with short ISI). Notably, bursting is absent at 34C in the wild type nerve ending while it is

306

evident at the same temperature in the HCN1 null animal. Figure 3B-E summarizes the

307

firing pattern parameters of several nerve endings over the entire temperature range (24-

308

34C). Although there was notable variability between individual nerve endings in terms of

309

firing patterns and mean firing rates, the overall trends are maintained within the

310

population. Nerve endings from HCN1 null mice have a higher mean firing frequency at

311

temperatures between 33C and 35C than those from wild type animals, but below 32C

312

the differences disappear (Figure 3B). This is explained by a higher number of spikes per

313

burst at 30-32C for HCN1 null terminals (Figure 3C), but more importantly, by a higher

314

rate of event skipping in wild type terminals (Figure 3D and see arrows in 3A). The latter

15

315

means that at 30C and above, there is a higher probability in wild type nerve endings of

316

skipping a cycle of the oscillation (Braun et al. 1980), a phenomenon that is evident in the

317

ISI histograms as peaks at approximately the double of the main ISI values observed (see

318

Figure 3A at 34C, arrow).

319

The data presented above suggests that corneal CS nerve endings from HCN1 null mice

320

have subtle but significant differences in firing pattern compared to wild type siblings.

321

Although the firing frequency versus temperature curves (Figure 3B) are not significantly

322

different (P=0.16 in a two-way ANOVA test), in the HCN1 null nerve endings there is a

323

significantly higher probability of firing action potentials at normal body surface

324

temperature (p=0.02 non paired t-test for the data at 34C). Moreover, the slope of the firing

325

frequency versus temperature relationship is lower in HCN1 null animals compared to wild

326

type (Figure 3B, fit lines). In other words, during moderate cooling from 34C to 28C a CS

327

nerve ending in a wild type animal increases its firing frequency by a factor of 3.1, while in

328

the case of a HCN1-/- animal this factor is only 1.7. This difference grows at lower

329

temperatures (Figure 3E).

330

Effect of Ih blockade in CS nerve endings from HCN1-/- mice

331

The results so far indicate that pharmacological blockade of Ih in CS nerve endings has a

332

more profound effect on the firing pattern than genetic deletion of HCN1. One possibility

333

to explain this observation is that HCN1-/- mice still have Ih in their CS nerve ending due to

334

the expression of the HCN2 channel (Orio et al. 2009). Thus, we tested the effect of

335

ZD7288 in HCN1-/- mouse corneas.

336

Figure 4A shows the behavior of a CS nerve ending from a HCN1 knockout mouse upon

337

exposure to 20 M ZD7288, in a similar experiment to that shown in Figure 1. In this case,

16

338

ZD7288 also induces a longer cycle time (Figure 4A grey rectangle and see also Figure 4B,

339

center) with an increased number of spikes per burst (Figure 4B, right). In the average,

340

however, there was a significant decrease of the mean firing frequency (Figure 4B left);

341

probably because the longer cycle time was accompanied by a modest increase in the

342

number of spikes per burst (Figure 4B, right) that failed to be significant. The latter may be

343

related to the fact that, as already mentioned, cold receptors from KO animals showed

344

bursting in the control condition. As in wild type nerve endings, ZD7288 produced no

345

significant change in the shape of the nerve terminal impulses recorded (see Table 1).

346

The effect of ZD7288 on CS nerve endings from HCN1-/- animals also raises the possibility

347

that this drug is acting on a molecular target other than HCN channels, as it has been

348

suggested previously at mossy fiber synapses (Chevaleyre and Castillo 2002).

349

Alternatively, in cold-sensitive terminals from HCN1 null animals there may still be a

350

significant Ih current, supported by expression of HCN2 channels, as it was observed in a

351

fraction of cultured CS trigeminal neurons obtained from these animals (Orio et al. 2009).

352

This in turn may explain why the firing pattern of HCN1-/- CS terminals is not much

353

different from that of HCN+/+ CS terminals, as there is still a hyperpolarization-activated

354

current albeit with a more negative half-activation voltage and slower kinetics (Santoro et

355

al. 2000). In an effort to discriminate between these two alternatives, we tested the effect of

356

Ivabradine, another HCN channel blocker (Bois et al. 1996; Bucchi et al. 2006), on the

357

electrical activity of wild type CS nerve endings. In this case we also found an increase of

358

the cycle time at 34C (Figure 5) as well as at 27C (from 293 36 ms to 356 26 ms;

359

p=0.05, n=4). The cycle time increase is less than that observed with ZD7288 and the

360

increase in bursting failed to be significant, suggesting a partial block of Ih by Ivabradine

361

(see below). It is important to emphasize that differences between effects of ZD7288 and
17

362

Ivabradine are only quantitative, with the same qualitative result (lengthening of cycle

363

time) produced by both blockers. The partial effect of Ivabradine can be explained by

364

differences in their blocking mechanisms, specifically, because Ivabradine needs repeated

365

open/close cycles to effectively block HCN1 channels while ZD7288 can readily block

366

open channels (Bucchi et al. 2006; Shin et al. 2001). Moreover we confirm that the effect of

367

Ivabradine can be attributed to a partial block of Ih using mathematical simulations (see

368

below). On the other hand, Cs+ ion can also block HCN channels (DiFrancesco 1982), with

369

additional effects on inward rectifier K+ channels (Hille 2001). We tested the effect of 3

370

mM CsCl on the firing pattern of the spontaneous activity of CS terminals and found that it

371

rapidly increased the cycle time and promoted bursts, resembling the effects of ZD7288 and

372

Ivabradine. However, this was followed by a dramatic increase of activity where bursts

373

could hardly be distinguished (Figure 6). This effect was poorly reversible, suggesting that

374

Cs+ may be acting intracellularly on inward rectifier potassium channels. Still, the overall

375

results obtained are consistent with the idea that the principal mechanism of action of

376

ZD7288 and Ivabradine is a blockade of Ih.

377

Definitive evidence for the presence of HCN2 channels in HCN-/- terminals would come

378

from the direct assessment of the activity of specific ionic conductances at mammalian

379

nerve terminals; however, this is currently not feasible due to their small size. To further

380

explore whether the compensation of HCN1 function by expression of HCN2 channels is a

381

plausible explanation for the results described above, we developed and analyzed

382

mathematical models of corneal CS nerve endings, simulating the expression of HCN-like

383

currents with biophysical properties resembling HCN1 and HCN2 channels.

18

384

Mathematical models of cold receptors bearing HCN1 or HCN2 channels

385

The rhythmic spiking activity of vertebrate cold thermoreceptors is thought to arise from an

386

underlying oscillation of the membrane potential, upon which an action potential is

387

generated each time firing threshold is reached (Braun et al. 1980; Schafer et al. 1990).

388

Cold temperatures would slow down the intrinsic oscillation, allowing more than one action

389

potential per cycle and thus giving rise to the bursting nature of the response. Experimental

390

evidence points to the involvement of slow, TTX-resistant persistent sodium currents

391

(Brock et al. 1998; Herzog et al. 2001) and low-threshold calcium channels (Schafer et al.

392

1982; Schafer et al. 1991) in the generation of rhythmic firing.

393

Mathematical models that account for the temperature-dependent change in spiking pattern

394

of cold-sensitive nerve fibers have been published and analyzed (Braun et al. 1998; Huber

395

et al. 2000; Longtin and Hinzer 1996), all based on Plants ionic model of slow wave

396

bursting (Plant 1981). In this model, a slow oscillation is driven by a slow voltage-

397

dependent sodium current that acts as positive feedback. Calcium entry through voltage-

398

activated calcium channels, and the consequent activation of calcium-activated potassium

399

channels, provides a negative feedback current. On top of this oscillation, fast voltage-

400

activated sodium and potassium currents generate action potentials. By adding the typical

401

effect of temperature on ion channel gating kinetics (a 3-fold speed up for every 10C

402

temperature rise), a model is obtained in which the spiking pattern changes with

403

temperature in a similar way as the cold-sensitive nerve endings and fibers.

404

We took the simple, classical, model developed by Braun (Braun et al. 1998) and added a

405

hyperpolarization-activated current with parameters similar to those determined

406

experimentally in cultured CS trigeminal neurons (Orio et al. 2009), specifically a half-

407

activation voltage of -85 mV and a time constant for activation of 125 ms at 25C (adjusted
19

408

by a Q10 factor of 3). Thereafter, the parameters were hand tuned to reproduce two key

409

observations of the firing pattern at static temperature in wt mice: the effect we observe in

410

the firing pattern after Ih blockade (Figures 1,2) and the firing pattern versus temperature

411

curves (Figures 3B-D). In the first case, the most important parameter change with respect

412

to the original Brauns model was an increase in the voltage dependency of the slow

413

depolarizing current activation (zsd), from an original value of 0.09 mV-1 to 0.11 mV-1 (at

414

34C, this is equivalent to 2.4 and 2.9 electron equivalents, respectively). Persistent, TTX-

415

resistant sodium currents have been described in trigeminal neurons with a voltage

416

dependency a little bit higher than 0.1 mV-1 (Herzog et al. 2001). In the second case, to fine

417

tune the model to fit the experimentally observed behavior of cold receptors we found it

418

necessary to add a background K+ current specifically modulated by temperature in the

419

form of a TREK1-like potassium channel (Maingret et al. 2000). Otherwise, a lower slope

420

of the mean firing frequency versus temperature relationship was obtained (not shown).

421

Figure 7A shows the firing patterns simulated by the model at three different temperatures,

422

34C, 30C and 24C, and the corresponding ISI histograms. As shown in Figure 7B, the

423

firing pattern parameters in the 24C 34C range of the model match very well the mean

424

curves obtained in CS terminals from wild type mice. In the case of the percentage of skip

425

events, the experimental curve can be regarded as a smoothed version of the simulated

426

curve. This result is typically obtained when stochastic contributions are not modeled

427

properly (Faisal et al. 2008), but tuning that aspect of the model is beyond the purpose of

428

this work. Figure 7C shows that the effect of ZD7288 on CS nerve endings can be

429

effectively mimicked by reducing the Ih conductance of the model. More specifically, a

430

progressive reduction of Ih results in a slow decrease of the mean firing frequency, an

431

increase of the cycle time (evidenced by the shift of the ISI main component) and the
20

432

induction of bursts, as observed in mice terminals (see Figure 1B grey rectangle). Finally, a

433

partial block of Ih (to a 16% of the original value) produces a behavior that resembles the

434

effect of Ivabradine: a modest increase of cycle time (from 123 ms to 180 ms) and very

435

little bursting (from 1 to 1.14 spikes/burst).

436

To simulate a HCN1-/- mouse nerve ending, rather than eliminating Ih we modified its

437

parameters to make it resemble HCN2 channels. The changes consisted in a shift of the

438

open probability curve towards negative potentials (Vh0 from -85 mV to -95 mV) and

439

slower channel kinetics (h at 25C changed from 125 ms to 900 ms). This is in agreement

440

with the properties of HCN2 channels in heterologous expression systems (Chen et al.

441

2001) and the characteristics of Ih recorded in cultured CS neurons from mice lacking

442

HCN1 (Orio et al. 2009). No other parameter of the model was changed.

443

Remarkably, as shown in Figure 8A and 8B, the sole change of HCN1- to HCN2-like Ih is

444

sufficient to make the model match the firing patterns of CS nerve endings from HCN1 null

445

mice between 24C and 34C. The progressive reduction of Ih conductance in the model

446

also reproduces the effect of ZD7288 on terminals lacking HCN1 (Figure 8C), namely an

447

increase of the cycle time, which combined with a small increase in the number of spikes

448

per burst results in a reduction of the mean firing rate. The effect of changing Ih parameters

449

is more strikingly evidenced in Figure 9, where the experimental data of Figure 3B-D

450

(circles) are plotted together with the data from both simulations (fast and slow Ih;

451

triangles).

452

In our preceding work (Orio et al. 2009) we found that Ih recorded in CS neurons from

453

HCN1(-/-) mice was greatly diminished in amplitude and was even absent sometimes. Thus,

454

we tested whether a model combining an Ih with slow kinetics and smaller maximum

21

455

conductance (gh changed from 0.6 to 0.2) could also reproduce the experimental results

456

obtained in HCN1 ko terminals. As shown in Figure 10, expression of one third of Ih

457

current with slow properties (HCN2-like) is enough to obtain a behavior similar to the

458

observed in HCN1-/-, and thus overexpression of HCN2 is not necessary for functional

459

compensation.

460

Analysis of the time course of model variables (see Appendix) shows that Ih provides a

461

critical depolarizing drive at the beginning of the oscillations. The apparently subtle change

462

in kinetic properties from HCN1 to HCN2 results in a diminished Ih current, thus delaying

463

the start of a new oscillation. At the same time, the reduced membrane conductance results

464

in a lengthening of the burst period. From these results, we can conclude that the cold

465

receptor phenotype observed in HCN1 null mice can be reasonably explained by a

466

compensation of HCN1 channels by the remaining HCN2 related subunit.

467

22

468

DISCUSSION

469

In a previous study (Orio et al. 2009) we characterized extensively a prominent h current

470

present in cultured cold-sensitive trigeminal neurons, hypothesizing about its potential role

471

in the rhythmic generation of action potentials in cold-sensitive nerve endings.

472

Pharmacological or genetic suppression of Ih in mice produced an altered cold perception in

473

behavioral studies, further supporting a possible role for Ih in the stimulus coding of cold

474

thermoreceptors (Orio et al. 2009). In this work, by taking advantage of the recently

475

described development of extracellular recordings in mice corneal sensory endings (Parra et

476

al. 2010), we have confirmed this hypothesis by directly determining the effect of

477

pharmacological Ih suppression on the firing patterns of cold-sensitive nerve endings from

478

mouse cornea. Moreover, we have characterized the firing pattern under static temperature

479

conditions of cold-sensitive nerve endings from knock-out mice for HCN1 channel

480

subunits, the principal molecular determinant of Ih in cold thermoreceptor neurons.

481

HCN1 channels are not only present in CS peripheral nerve endings and neurons, but also

482

in many other neurons within the central nervous system. Thus, it is possible that other

483

neurons involved in the behavioral response to cold are also affected by the lack of this

484

channel. Therefore, the behavioral deficits observed can be the result of changes in the

485

function of several neuronal types, rather than the sole changes in nerve ending firing

486

patterns. While we cannot rule out this possibility with the present evidence, still our results

487

show an important role of Ih in the initial sensory encoding of innocuous cold temperatures.

488

Moreover, Figure 3E shows that the apparently small differences in the firing of wildtype

489

and HCN1-/- nerve endings are associated to a large change in the increase of static firing

490

rate upon cooling.

23

491

Effects of Ih blockers on firing pattern

492

The most notorious effect of Ih block on CS terminal NTIs is a rapid alteration in their

493

firing pattern, occurring within a few minutes of exposure of the preparation to ZD7288 or

494

Ivabradine, widely used Ih blockers. This change in firing pattern consisted in a lengthening

495

of the mean cycle time and the induction of bursts. There were quantitative differences in

496

the effect of both blockers, characterized by a lesser effect of Ivabradine compared to

497

ZD7288. Differences between the effect of both compounds are to be expected, since they

498

have a different blocking mechanism (Bucchi et al. 2006; Shin et al. 2001). Specifically, a

499

lesser potency of Ivabradine is plausible, since it needs repeated open/close cycles to

500

effectively block the channel, a situation that is not likely to occur if, according to the

501

mathematical simulations, the nerve endings spend most of the time at less than -60mV.

502

The observed change in firing pattern reveals an important role for Ih in the rhythmic

503

generation of action potentials, which is a distinctive feature of CS nerve endings and fibers

504

(Braun et al. 1980; Carr et al. 2003). A longer cycle time and the appearance of bursting

505

suggest that the slow membrane potential oscillation thought to underlie this rhythmic

506

activity (Braun et al. 1980; Braun et al. 1998) becomes slower when Ih is blocked, as has

507

been shown in the inferior olive (Bal and McCormick 1997). Until direct recording of

508

membrane potential at the tiny nerve endings becomes experimentally feasible this

509

suggestion will remain unconfirmed, but the computer simulations that reproduce the effect

510

of ZD7288 (Figures 7C and also Figure 11, Appendix) also show this effect, supporting this

511

hypothesis. Since the reversal potential of Ih is about -30 mV, a hyperpolarization of the

512

nerve ending is expected to occur upon blockade, leading to a reduced rate of action

513

potential generation. However, the blockade of Ih also leads to a higher membrane

514

resistance that increases the amplitude of the oscillation and the length of bursts.
24

515

Role of HCN1 versus HCN2

516

Our study shows that application of ZD7288 to CS nerve endings affects their firing pattern

517

regardless of the expression of the HCN1 protein, which we showed to be responsible for a

518

major fraction of the Ih observed in the soma of cultured trigeminal CS neurons (Orio et al.

519

2009). This result could be interpreted as ZD7288 acting on another (unspecific) target, an

520

interpretation we cannot fully rule out. On the other hand, this finding could be explained

521

by the presence of another member of the HCN channel family in CS nerve endings,

522

partially fulfilling the role of HCN1. We think that the last possibility is a very plausible

523

explanation, as the Ih in CS neurons resembles a mixture of HCN1 and HCN2 channels and

524

a slow HCN2-like current was recorded in a fraction of CS neurons from HCN1-/- mice

525

(Orio et al. 2009). Moreover, our computational simulations show (Figure 8) that the

526

behavior of HCN1-/- CS nerve endings can be fully reproduced just by changing Ih

527

activation parameters to make it resemble HCN2 channels.

528

This result shows that the normal rhythmic firing of CS nerve endings requires the

529

expression of a hyperpolarization-activated current with specific activating kinetics.

530

Though the difference in spiking pattern between CS terminals from wt and HCN1-/- mice

531

may seem small, it can produce a significant difference in neural encoding of temperature.

532

Just from the point of view of mean static firing frequency, CS terminals from HCN1 null

533

animals have a lower slope of the firing frequency versus temperature relationship (see

534

Figure 4B). As the neural coding underlying cold perception is thought to involve changes

535

both in firing frequency and firing pattern (Spray 1986), a lower increase in firing frequency

536

upon cooling is likely to have an impact on cold perception.

25

537

New kinetic model of CS nerve terminals

538

The temperature-dependent firing patterns of CS nerve endings and fibers have been

539

previously modeled using conductance-based models, notably in the studies of Longtin &

540

Hinzer (1996) and Huber & Braun (Braun et al. 1998; Huber et al. 2000). These models are

541

based on Plants model of slow wave bursting (Plant 1981) in which a slow oscillation of

542

the membrane is driven by a slow persistent sodium/calcium current and a calcium-

543

activated potassium channel. By adding the usual temperature dependencies of channel

544

kinetics and conductance, the firing patterns of CS nerve endings at different static

545

temperatures can be qualitatively reproduced. Huber & Brauns model stands out for

546

simplified equations and low number of parameters, leaving only the essential while still

547

being biologically plausible.

548

Our results prompted the inclusion of a hyperpolarization-activated current in the model in

549

order to explain the results with HCN channel blockers as well as the differences observed

550

in the firing patterns between wt and HCN1-/- mice. Previous models omitted this current in

551

simulating the overall behavior of CS nerve endings while still being adequate for

552

explaining the qualitative aspect of firing pattern change with temperature. Our expanded

553

model reproduces quantitative aspects of the firing patterns during cooling (see Figures 7B

554

and 8B) as well as the effects of Ih blockade on the firing patterns. Furthermore, in the

555

process of tuning the model to make it dependent on the expression of Ih, we found that this

556

behavior is quite sensitive to the voltage dependency of the persistent sodium current (INa,p).

557

The value of the ssd parameter in the original Huber and Braun model (0.09) is lower than

558

the value reported for the INa,p measured in sensory neurons from dorsal root ganglion

559

(Herzog et al. 2001). A shallower P(O)/V relationship implies more channels open at

560

negative voltages, thus making it easier to start a new depolarizing cycle from the negative
26

561

part of the oscillation. With the higher voltage dependency we use (0.11, equivalent to 2.8

562

charges per channel at 25C), that agrees with the values reported by Herzog et al. (2001),

563

there are less sodium channels open at negative voltages and a new oscillation will take

564

longer to start unless some other conductance (e.g. Ih in our model) helps in depolarizing

565

the membrane again.

566

As a result, we present a model that comprehensively reproduces all the experimental

567

evidence here discussed. In doing so, it provides a plausible explanation for the effect of the

568

Ih blocker on the CS nerve ending from HCN1(-/-) animals, a result that at first appeared

569

contradictory. The model also shows that the Ih is important for tuning the correct period of

570

the slow oscillation: the opening of HCN channels adds a depolarizing drive and helps in

571

the onset of the oscillation while their contribution to the total membrane conductance help

572

to determine the burst duration. Replacing HCN1 channels with an HCN2-like

573

conductance, with a different time constant and half-activation voltage, results in longer

574

cycles with augmented bursting. Finally, the absence of Ih results in the longest cycles

575

because of a more profound hyperpolarization of the membrane during the resting phase of

576

the cycle.

577

The role of cold-modulated conductances

578

Previous models did not take into account the presence of a cold-inhibited hyperpolarizing

579

(i.e. potassium) current, such as TREK1 and/or TRAAK channels (Kang et al. 2005; Reid

580

and Flonta 2001; Viana et al. 2002), in CS neurons. We found that incorporating this type

581

of current to the model allowed for a better fine tuning of the mean firing frequency

582

versus temperature relationship. In fact, it has been reported that mice lacking TREK1 and

583

TRAAK channels have an impaired cold perception (Noel et al. 2009).

27

584

However, it is noteworthy that current models for temperature-induced changes in firing

585

pattern (including ours) do not consider the presence of TRMP8, a channel member of the

586

TRP channel family activated by decreases in temperature and menthol (Brauchi et al.

587

2004; McKemy et al. 2002; Peier et al. 2002; Voets et al. 2004). One may wonder how this

588

omission affects the interpretation of the results we obtained, considering the critical

589

importance of this current in cold thermoreceptor function (Bautista et al. 2007; Colburn et

590

al. 2007; Dhaka et al. 2007; Parra et al. 2010). TRPM8 is unspecific for monovalent

591

cations: its activation causes depolarization of the membrane and a consequent increase in

592

action potential firing. Under physiological recording conditions (i.e. normal external Ca2+),

593

the cold-induced inward current that can be recorded in cultured CS neurons and that is

594

blocked by the TRPM8 blocker BCTC (Madrid et al. 2006), shows a fast desensitization

595

upon a sustained cold stimulus (see Reid et al. 2002; Viana et al. 2002). This

596

desensitization depends on phospholipase C activation, caused in turn by an increase in

597

intracellular calcium (Daniels et al. 2009) and shows a time course similar to the adaptation

598

of the dynamic response of CS nerve endings. Thus, it is possible that TRPM8 activity is

599

mostly responsible for the transient responses to sudden temperature changes, playing a

600

lesser role in establishing the firing patterns observed at static temperatures. Considering

601

that the experimental work and the model here presented focus primarily on static

602

responses of CS nerve terminals, the absence of TRPM8 should be of minor impact.

603

Obviously, future models, trying to replicate dynamic aspects of thermoreceptor activity,

604

should also include TRPM8 currents.

28

605

Concluding remarks

606

The hyperpolarization-activated current (Ih) has been implicated in pacemaking and

607

oscillatory activity throughout the nervous system (Chan et al. 2004; Hutcheon et al. 1996;

608

Luthi et al. 1998; McCormick and Pape 1990; Thoby-Brisson et al. 2000). In addition, Ih

609

has been recorded in sensory neurons (Doan et al. 2004; Momin et al. 2008; Orio et al.

610

2009; Scroggs et al. 1994) and the expression of HCN channels has been demonstrated in

611

sensory neurons and axon terminals (Antal et al. 2004; Kouranova et al. 2008; Momin et al.

612

2008; Tu et al. 2004) with its abnormal expression associated with pathologic pain (Emery

613

et al. 2011; Luo et al. 2007). The high-pass filter properties of Ih have been implicated in

614

sensory coding by retinal networks (Cangiano et al. 2007; Publio et al. 2009) and in the

615

phase-locked response of hypoglossal motoneurons to sinusoidal stimulus (van Brederode

616

and Berger 2011). Here, we demonstrate a role of Ih in rhythmic and oscillatory activity of

617

cold thermoreceptor terminals. When put together with our previous observation of a

618

behavioral deficit in cold sensing in mice lacking HCN1 or in the presence of Ih blockers

619

(Orio et al. 2009), our results support the idea that the rhythmic discharge of action

620

potentials in CS nerve endings plays a significant role in the sensory coding process, as was

621

originally hypothesized more than 30 years ago by Braun and colleagues (Braun et al.

622

1980). Finally, our results emphasize the notion that cold temperature perception is a

623

complex sensory process that involves additional sensory events beyond the initial

624

transduction step.

625
626

29

627

APPENDIX

628

The role of Ih in the membrane potential oscillation

629

We examined closely the time course of the voltage and activation variables in the

630

simulation, trying to understand what the role of Ih can be and why a change in Ih activation

631

parameters can affect the bursting pattern and firing frequency. Figure 11A-C presents this

632

examination graphically. In the wt model (Figure 11A), at the beginning of a voltage

633

oscillation (dV/dt=0, vertical segmented line) the open probability of the h current (ah) is

634

increasing because of membrane hyperpolarization. Since the reversal potential for the h

635

current is -30mV, at negative membrane potentials it provides a depolarizing force that

636

precedes the positive feedback of the slow depolarizing current Isd, thus accelerating the

637

beginning of a new cycle (note that asd the open probility of Isd starts to grow when ah is

638

almost at its maximum). In the slow Ih model (Figure 11B), ah displays a lower value

639

during the cycle, because of a more negative half-activation voltage and the slower

640

activation kinetics, limiting the opening of h channels during the hyperpolarization. This

641

results in a weaker depolarizing drive for the onset of a new firing cycle. Finally, in the

642

absence of Ih (Figure 11C), the new cycle in the model is initiated very slowly by the

643

depolarizing positive feedback provided by Isd alone. This causes a very long inter-burst

644

period and cycle times as seen in the experimental results with Ih blockade. It is also worth

645

noting that the absence or reduction of Ih changes dramatically the voltage reached during

646

the hyperpolarization phase of the oscillation: while the wt model reaches a maximum

647

hyperpolarization of -70 mV, the slow Ih model goes to -75 mV and in the absence of Ih the

648

most negative value is -82 mV, closer to the reversal potential for the K+ conductance.

30

649

A third important observation in the simulations is the increase in burst duration produced

650

by the reduction or absence of Ih. This may result counter-intuitive, considering that Ih is

651

depolarizing for most of the time during the cycle and thus should contribute in maintaining

652

the firing of action potentials. However, we like to note that when the voltage is near -50

653

mV, Ih is only weakly depolarizing and as an extra membrane conductance it actually

654

dampens the depolarizing force of Isd. In the absence of Ih, more potassium conductance

655

has to build up to dampen Isd and this delays the end of the firing phase. The bottom row of

656

Figure 11 shows that the bursting phase always ends when the total membrane conductance

657

reaches a similar value of 0.3 mS/cm2. When the Ih is decreased or absent, it takes more

658

action potentials to reach this value because the initial conductance is lower.

659

This analysis not only explains the dramatic change in firing pattern when Ih is fully

660

blocked; it also shows how subtle changes in channel kinetics and activation curve given by

661

a shift from HCN1 to HCN2 channels introduce significant alterations in the firing

662

properties of cold thermoreceptor endings. In turn, this may explain the behavioral deficits

663

in cold perception observed in HCN1-/- animals (Orio et al. 2009).

31

664

ACKNOWLEDGMENTS

665

The authors thank E. Quintero, A. Perez and A. Miralles for excellent technical assistance

666

and Tansy Donovan-Rodriguez for participating in preliminary pharmacological

667

experiments.

32

668

GRANTS

669

This work was supported by funds from the Chilean Comisin Nacional de Investigacin

670

Cientfica y Tecnolgica CONICYT (PBCT PSD20 and FONDECYT 11090308 to P.O.,

671

FONDECYT 1100983 to R.M. and ACT-1113 to P.O. and R.M.), the Universidad de

672

Valparaso (DIPUV 51/2007 to P.O.), the Universidad de Santiago de Chile (VRID-

673

USACH to R.M.) and the Spanish MICIIN (BFU2008-04425 and CONSOLIDER-

674

INGENIO 2010 CSD2007-00023 to C.B., SAF2010-14990 to F.V.). The CINV is funded

675

by Grant P09-022-F of the Millennium Science Initiative (Ministerio de Economa, Chile).

676

33

677

References

678
679

Antal M, Papp I, Bahaerguli N, Veress G, and Vereb G. Expression of

680

hyperpolarization-activated and cyclic nucleotide-gated cation channel subunit 2 in axon

681

terminals of peptidergic nociceptive primary sensory neurons in the superficial spinal

682

dorsal horn of rats. Eur J Neurosci 19: 1336-1342, 2004.

683

Babes A, Ciobanu AC, Neacsu C, and Babes RM. TRPM8, a sensor for mild cooling in

684

mammalian sensory nerve endings. Curr Pharm Biotechnol 12: 78-88, 2011.

685

Bade H, Braun HA, and Hensel H. Parameters of the static burst discharge of lingual cold

686

receptors in the cat. Pflugers Arch 382: 1-5, 1979.

687

Bal T, and McCormick DA. Synchronized oscillations in the inferior olive are controlled

688

by the hyperpolarization-activated cation current I(h). Journal of neurophysiology 77:

689

3145-3156, 1997.

690

Bautista DM, Siemens J, Glazer JM, Tsuruda PR, Basbaum AI, Stucky CL, Jordt SE,

691

and Julius D. The menthol receptor TRPM8 is the principal detector of environmental

692

cold. Nature 448: 204-208, 2007.

693

Bois P, Bescond J, Renaudon B, and Lenfant J. Mode of action of bradycardic agent, S

694

16257, on ionic currents of rabbit sinoatrial node cells. British journal of pharmacology

695

118: 1051-1057, 1996.

696

Brauchi S, Orio P, and Latorre R. Clues to understanding cold sensation:

697

thermodynamics and electrophysiological analysis of the cold receptor TRPM8. Proc Natl

698

Acad Sci U S A 101: 15494-15499, 2004.

699

Braun HA, Bade H, and Hensel H. Static and dynamic discharge patterns of bursting cold

700

fibers related to hypothetical receptor mechanisms. Pflugers Arch 386: 1-9, 1980.
34

701

Braun HA, Huber MT, Dewald M, Schafer K, and Voigt K. Computer simulations of

702

neuronal signal transduction: The role of nonlinear dynamics and noise. Int J Bifurcat

703

Chaos 8: 881-889, 1998.

704

Brock JA, McLachlan EM, and Belmonte C. Tetrodotoxin-resistant impulses in single

705

nociceptor nerve terminals in guinea-pig cornea. J Physiol 512 ( Pt 1): 211-217, 1998.

706

Brock JA, Pianova S, and Belmonte C. Differences between nerve terminal impulses of

707

polymodal nociceptors and cold sensory receptors of the guinea-pig cornea. J Physiol 533:

708

493-501, 2001.

709

Bucchi A, Tognati A, Milanesi R, Baruscotti M, and DiFrancesco D. Properties of

710

ivabradine-induced block of HCN1 and HCN4 pacemaker channels. The Journal of

711

physiology 572: 335-346, 2006.

712

Cangiano L, Gargini C, Della Santina L, Demontis GC, and Cervetto L. High-Pass

713

Filtering of Input Signals by the Ih Current in a Non-Spiking Neuron, the Retinal Rod

714

Bipolar Cell. PLoS ONE 2: e1327, 2007.

715

Carr RW, Pianova S, Fernandez J, Fallon JB, Belmonte C, and Brock JA. Effects of

716

heating and cooling on nerve terminal impulses recorded from cold-sensitive receptors in

717

the guinea-pig cornea. J Gen Physiol 121: 427-439, 2003.

718

Colburn RW, Lubin ML, Stone DJ, Jr., Wang Y, Lawrence D, D'Andrea MR, Brandt

719

MR, Liu Y, Flores CM, and Qin N. Attenuated cold sensitivity in TRPM8 null mice.

720

Neuron 54: 379-386, 2007.

721

Chan CS, Shigemoto R, Mercer JN, and Surmeier DJ. HCN2 and HCN1 channels

722

govern the regularity of autonomous pacemaking and synaptic resetting in globus pallidus

723

neurons. J Neurosci 24: 9921-9932, 2004.

35

724

Chen S, Wang J, and Siegelbaum SA. Properties of hyperpolarization-activated

725

pacemaker current defined by coassembly of HCN1 and HCN2 subunits and basal

726

modulation by cyclic nucleotide. J Gen Physiol 117: 491-504, 2001.

727

Chevaleyre V, and Castillo PE. Assessing the role of Ih channels in synaptic transmission

728

and mossy fiber LTP. Proc Natl Acad Sci U S A 99: 9538-9543, 2002.

729

Daniels RL, Takashima Y, and McKemy DD. Activity of the Neuronal Cold Sensor

730

TRPM8 Is Regulated by Phospholipase C via the Phospholipid Phosphoinositol 4,5-

731

Bisphosphate. Journal of Biological Chemistry 284: 1570-1582, 2009.

732

Dhaka A, Murray AN, Mathur J, Earley TJ, Petrus MJ, and Patapoutian A. TRPM8

733

is required for cold sensation in mice. Neuron 54: 371-378, 2007.

734

DiFrancesco D. Block and activation of the pace-maker channel in calf purkinje fibres:

735

effects of potassium, caesium and rubidium. The Journal of physiology 329: 485-507, 1982.

736

DiFrancesco D. Serious workings of the funny current. Prog Biophys Mol Biol 90: 13-25,

737

2006.

738

Doan TN, Stephans K, Ramirez AN, Glazebrook PA, Andresen MC, and Kunze DL.

739

Differential distribution and function of hyperpolarization-activated channels in sensory

740

neurons and mechanosensitive fibers. J Neurosci 24: 3335-3343, 2004.

741

Dykes RW. Coding of steady and transient temperatures by cutaneous "cold" fibers serving

742

the hand of monkeys. Brain Res 98: 485-500, 1975.

743

Emery EC, Young GT, Berrocoso EM, Chen L, and McNaughton PA. HCN2 Ion

744

Channels Play a Central Role in Inflammatory and Neuropathic Pain. Science 333: 1462-

745

1466, 2011.

746

Faisal AA, Selen LP, and Wolpert DM. Noise in the nervous system. Nat Rev Neurosci 9:

747

292-303, 2008.
36

748

Harris NC, and Constanti A. Mechanism of block by ZD7288 of the hyperpolarization-

749

activated inward rectifying current in guinea pig substantia nigra neurons in vitro. J

750

Neurophysiol 74: 2366-2378, 1995.

751

Hensel H. Cutaneous Thermoreceptors. In: Thermoreception and Temperature Regulation,

752

edited by Hensel H. London: Academic Press, 1981, p. 33-61.

753

Hensel H, and Wurster RD. Static properties of cold receptors in nasal area of cats. J

754

Neurophysiol 33: 271-275, 1970.

755

Herzog RI, Cummins TR, and Waxman SG. Persistent TTX-resistant Na+ current affects

756

resting potential and response to depolarization in simulated spinal sensory neurons. J

757

Neurophysiol 86: 1351-1364, 2001.

758

Hille B. Ion Channels of Excitable Membranes. Sunderland, MA, USA: Sinauer Associates

759

Inc., 2001.

760

Hines ML, and Carnevale NT. The NEURON simulation environment. Neural Comput 9:

761

1179-1209, 1997.

762

Hofmann F, Biel M, and Kaupp UB. International Union of Pharmacology. LI.

763

Nomenclature and structure-function relationships of cyclic nucleotide-regulated channels.

764

Pharmacol Rev 57: 455-462, 2005.

765

Huber MT, Krieg JC, Dewald M, Voigt K, and Braun HA. Stochastic encoding in

766

sensory neurons: impulse patterns of mammalian cold receptors. Chaos Solitons & Fractals

767

11: 1895-1903, 2000.

768

Hutcheon B, Miura RM, and Puil E. Subthreshold membrane resonance in neocortical

769

neurons. JNeurophysiol 76: 683-697, 1996.

770

Iggo A. Cutaneous thermoreceptors in primates and sub-primates. J Physiol 200: 403-430,

771

1969.
37

772

Kang D, Choe C, and Kim D. Thermosensitivity of the two-pore domain K+ channels

773

TREK-2 and TRAAK. J Physiol 564: 103-116, 2005.

774

Kaupp UB, and Seifert R. Molecular diversity of pacemaker ion channels. Annu Rev

775

Physiol 63: 235-257, 2001.

776

Kouranova EV, Strassle BW, Ring RH, Bowlby MR, and Vasilyev DV.

777

Hyperpolarization-activated cyclic nucleotide-gated channel mRNA and protein expression

778

in large versus small diameter dorsal root ganglion neurons: correlation with

779

hyperpolarization-activated current gating. Neuroscience 153: 1008-1019, 2008.

780

Krahe R, and Gabbiani F. Burst firing in sensory systems. Nat Rev Neurosci 5: 13-23,

781

2004.

782

Latorre R, Brauchi S, Madrid R, and Orio P. A cool channel in cold transduction.

783

Physiology (Bethesda) 26: 273-285, 2011.

784

Longtin A, and Hinzer K. Encoding with bursting, subthreshold oscillations, and noise in

785

mammalian cold receptors. Neural Comput 8: 215-255, 1996.

786

Ludwig A, Zong X, Jeglitsch M, Hofmann F, and Biel M. A family of hyperpolarization-

787

activated mammalian cation channels. Nature 393: 587-591, 1998.

788

Luo L, Chang L, Brown SM, Ao H, Lee DH, Higuera ES, Dubin AE, and Chaplan SR.

789

Role of peripheral hyperpolarization-activated cyclic nucleotide-modulated channel

790

pacemaker channels in acute and chronic pain models in the rat. Neuroscience 144: 1477-

791

1485, 2007.

792

Luthi A, Bal T, and McCormick DA. Periodicity of thalamic spindle waves is abolished

793

by ZD7288,a blocker of Ih. JNeurophysiol 79: 3284-3289, 1998.

38

794

Madrid R, Donovan-Rodriguez T, Meseguer V, Acosta MC, Belmonte C, and Viana F.

795

Contribution of TRPM8 channels to cold transduction in primary sensory neurons and

796

peripheral nerve terminals. J Neurosci 26: 12512-12525, 2006.

797

Maingret F, Lauritzen I, Patel AJ, Heurteaux C, Reyes R, Lesage F, Lazdunski M,

798

and Honore E. TREK-1 is a heat-activated background K+ channel. Embo J 19: 2483-

799

2491, 2000.

800

McCormick DA, and Pape HC. Properties of a hyperpolarization-activated cation current

801

and its role in rhythmic oscillation in thalamic relay neurones. J Physiol 431: 291-318,

802

1990.

803

McKemy DD, Neuhausser WM, and Julius D. Identification of a cold receptor reveals a

804

general role for TRP channels in thermosensation. Nature 416: 52-58, 2002.

805

Momin A, Cadiou H, Mason A, and McNaughton PA. Role of the hyperpolarization-

806

activated current Ih in somatosensory neurons. J Physiol 2008.

807

Noel J, Zimmermann K, Busserolles J, Deval E, Alloui A, Diochot S, Guy N, Borsotto

808

M, Reeh P, Eschalier A, and Lazdunski M. The mechano-activated K+ channels TRAAK

809

and TREK-1 control both warm and cold perception. EMBO J 2009.

810

Nolan MF, Malleret G, Lee KH, Gibbs E, Dudman JT, Santoro B, Yin D, Thompson

811

RF, Siegelbaum SA, Kandel ER, and Morozov A. The hyperpolarization-activated

812

HCN1 channel is important for motor learning and neuronal integration by cerebellar

813

Purkinje cells. Cell 115: 551-564, 2003.

814

Orio P, Madrid R, de la Pea E, Parra A, Meseguer V, Bayliss DA, Belmonte C, and

815

Viana F. Characteristics and physiological role of hyperpolarization activated currents in

816

mouse cold thermoreceptors. J Physiol 587: 1961-1976, 2009.

39

817

Pape HC. Queer current and pacemaker: the hyperpolarization-activated cation current in

818

neurons. AnnuRevPhysiol 58: 299-327, 1996.

819

Parra A, Madrid R, Echevarria D, del Olmo S, Morenilla-Palao C, Acosta MC, Gallar

820

J, Dhaka A, Viana F, and Belmonte C. Ocular surface wetness is regulated by TRPM8-

821

dependent cold thermoreceptors of the cornea. Nat Med 16: 1396-1399, 2010.

822

Peier AM, Moqrich A, Hergarden AC, Reeve AJ, Andersson DA, Story GM, Earley

823

TJ, Dragoni I, McIntyre P, Bevan S, and Patapoutian A. A TRP channel that senses

824

cold stimuli and menthol. Cell 108: 705-715, 2002.

825

Plant RE. Bifurcation and resonance in a model for bursting nerve cells. J Math Biol 11:

826

15-32, 1981.

827

Publio R, Oliveira RF, and Roque AC. A Computational Study on the Role of Gap

828

Junctions and Rod Ih Conductance in the Enhancement of the Dynamic Range of the

829

Retina. PLoS ONE 4: e6970, 2009.

830

Reid G, Babes A, and Pluteanu F. A cold- and menthol-activated current in rat dorsal root

831

ganglion neurones: properties and role in cold transduction. J Physiol 545: 595-614, 2002.

832

Reid G, and Flonta M. Cold transduction by inhibition of a background potassium

833

conductance in rat primary sensory neurones. Neurosci Lett 297: 171-174, 2001.

834

Richardson MJE, Brunel N, and Hakim V. From Subthreshold to Firing-Rate

835

Resonance. JNeurophysiol 89: 2538-2554, 2003.

836

Robinson RB, and Siegelbaum SA. Hyperpolarization-activated cation currents: from

837

molecules to physiological function. Annu Rev Physiol 65: 453-480, 2003.

838

Santoro B, Chen S, Luthi A, Pavlidis P, Shumyatsky GP, Tibbs GR, and Siegelbaum

839

SA. Molecular and functional heterogeneity of hyperpolarization-activated pacemaker

840

channels in the mouse CNS. J Neurosci 20: 5264-5275, 2000.

40

841

Santoro B, Liu DT, Yao H, Bartsch D, Kandel ER, Siegelbaum SA, and Tibbs GR.

842

Identification of a gene encoding a hyperpolarization-activated pacemaker channel of brain.

843

Cell 93: 717-729, 1998.

844

Scroggs RS, Todorovic SM, Anderson EG, and Fox AP. Variation in IH, IIR, and

845

ILEAK between acutely isolated adult rat dorsal root ganglion neurons of different size. J

846

Neurophysiol 71: 271-279, 1994.

847

Schafer K, Braun HA, and Hensel H. Static and dynamic activity of cold receptors at

848

various calcium levels. J Neurophysiol 47: 1017-1028, 1982.

849

Schafer K, Braun HA, and Rempe L. Discharge pattern analysis suggests existence of a

850

low-threshold calcium channel in cold receptors. Experientia 47: 47-50, 1991.

851

Schafer K, Braun HA, and Rempe L. Mechanism of Sensory Transduction in Cold

852

Receptors. In: Thermoreception and Thermoregulation, edited by Bligh J, and Voigt K.

853

Berlin: Springer Verlag, 1990, p. 30-36.

854

Schreiber S. Influence of Ionic Conductances on Spike Timing Reliability of Cortical

855

Neurons for Suprathreshold Rhythmic Inputs. J Neurophysiol 91: 194-205, 2003.

856

Shin KS, Rothberg BS, and Yellen G. Blocker state dependence and trapping in

857

hyperpolarization-activated cation channels: evidence for an intracellular activation gate.

858

The Journal of general physiology 117: 91-101, 2001.

859

Spray DC. Cutaneous temperature receptors. Annu Rev Physiol 48: 625-638, 1986.

860

Thoby-Brisson M, Telgkamp P, and Ramirez JM. The role of the hyperpolarization-

861

activated current in modulating rhythmic activity in the isolated respiratory network of

862

mice. J Neurosci 20: 2994-3005, 2000.

41

863

Tu H, Deng L, Sun Q, Yao L, Han JS, and Wan Y. Hyperpolarization-activated, cyclic

864

nucleotide-gated cation channels: roles in the differential electrophysiological properties of

865

rat primary afferent neurons. J Neurosci Res 76: 713-722, 2004.

866

Ulens C, and Tytgat J. Functional heteromerization of HCN1 and HCN2 pacemaker

867

channels. JBiolChem 276: 6069-6072, 2001.

868

van Brederode JF, and Berger AJ. GAD67-GFP+ neurons in the Nucleus of Roller. II.

869

Subthreshold and firing resonance properties. Journal of neurophysiology 105: 249-278,

870

2011.

871

Viana F, de la Pea E, and Belmonte C. Specificity of cold thermotransduction is

872

determined by differential ionic channel expression. NatNeurosci 5: 254-260, 2002.

873

Voets T, Droogmans G, Wissenbach U, Janssens A, Flockerzi V, and Nilius B. The

874

principle of temperature-dependent gating in cold- and heat-sensitive TRP channels. Nature

875

430: 748-754, 2004.

876
877
878

42

879

FIGURE CAPTIONS

880

Figure 1. ZD7288 alters the firing pattern at 34C in CS nerve terminals from the mouse

881

cornea without impairing the acute response to cooling.

882

A. Extracellular recordings of single nerve terminal impulses from a CS nerve terminal

883

from the adult mouse cornea at 34C. Sample recordings of 1.5 seconds duration are shown

884

for the control condition (left) and after 10 min of exposure to 20 M ZD7288 (right). B.

885

Firing frequency, ISI values, number of spikes per burst and temperature trace for the same

886

CS terminal shown in A. Black and grey line boxes indicate the periods analyzed for graphs

887

in Figure 1C. In the spikes per burst plot, the thick gray line indicates the moving average.

888

C. Mean firing frequency, cycle time and number of spikes per burst at baseline

889

temperature (34C) in the control condition and after addition of 20 M ZD7288. Bars

890

represent mean SEM. The experimental conditions were paired with their respective

891

controls and were compared with a paired t-test. * P<0.05, **P<0.01. n=8. D.

892

Representative NTIs recorded in control situation (black) and after exposure to 20 M

893

ZD7288 (gray). 100 to 200 spikes from a steady-state firing period (rectangles in B) were

894

averaged in each case. Samples from two different experiments are shown. In some

895

experiments the NTIs in ZD7288 were of lower amplitude (left), but when rescaling (inset)

896

it is shown that the shape of NTIs is unaltered. Table 1 shows that this decrease in

897

amplitude was not significant.

898

Figure 2. ZD7288 alters the firing pattern at 27C.

899

A. Extracellular recordings from a CS nerve ending from mouse cornea at 27C, in the

900

control condition and after exposure to 20 M ZD7288. B. Firing rate, ISI values, spikes

901

per burst and temperature trace from the same experiment. In the spikes per burst plot, the
43

902

thick gray line indicates moving average. C. Mean firing frequency, cycle time and

903

number of spikes per burst at 27C in the control condition and after addition of 20 M

904

ZD7288. Bars represent mean SEM. The experimental conditions were paired with their

905

respective controls and were compared with a paired t-test. *P<0.05, **P<0l.01. n=6.

906

Figure 3. Comparison of static firing patterns between cold receptors from HCN1+/+ and

907

HCN1-/- mice.

908

A, Two-seconds sample traces and ISI histograms obtained at three static temperature

909

conditions in wild type and HCN1 null mice corneas. For constructing the ISI histograms,

910

intervals during the last 40 seconds before changing temperature were considered. The

911

arrows point to the evidence for skipping events, both in the recording and in the ISI

912

histogram at 34C. B-D, Summary of steady-state firing pattern for CS nerve endings from

913

wild type and HCN1 null mice corneas. Mean firing frequency (B), mean number of spikes

914

per cycle (C), and fraction of skipped events (D) are plotted against steady-state

915

temperature. Symbols represent mean SEM. * p<0.05, ** p<0.01. n=11-13 for HCN1;

916

n=8-10 for HCN1-/-. In B, lines indicate the best fit of a linear function to the points

917

between 25C and 34C. The slopes are -1.11 s-1C-1 for HCN1+/+ and -0.76 s-1C-1 for

918

HCN1-/-. E. Increase in mean firing frequency when cooling. The values plotted in B (mean

919

only) are expressed as the ratio to the value at 34C.

920

Figure 4. Effect of ZD7288 on NTI discharge in cold receptors from HCN null mouse

921

cornea.

922

A, Firing rate, ISIs, number of spikes per burst and temperature trace during the application

923

of ZD7288 to a CS nerve ending from a HCN1-/- mouse. Boxes indicate the period

924

considered for the graphs in B. B, Mean firing frequency, cycle time and mean number of
44

925

spikes per burst in experiments performed with 6 HCN1-/- animals. In each experiment, a

926

150-200 s time interval of regular spiking before (black) or after (grey) application of

927

ZD7288 was analyzed. Bars represent mean SEM. * p<0.05.

928

Figure 5. Effect of Ivabradine on the firing pattern of cold receptors.

929

A, Firing rate, ISIs, number of spikes per burst and temperature trace during the application

930

of 30 M Ivabradine to a CS nerve ending from a wildtype mouse. Boxes indicate the

931

period considered for the plots in B. B, Mean firing frequency, cycle time and mean

932

number of spikes per burst. Bars represent mean SEM. n=7 ** p<0.01.

933

Figure 6. Effect of CsCl on the firing pattern of mouse cold receptors.

934

Firing rate, ISIs, number of spikes per burst and temperature trace during the application of

935

3 mM CsCl to a CS nerve ending from a wildtype mouse. The box indicates a period of

936

increasing cycle time and augmented bursting shortly after Cs+ application. Note the long

937

lasting effects of Cs+ on firing. Representative experiment of 4 performed, all showing

938

similar results.

939

Figure 7. Simulations of NTI activity with a temperature-modulated slow wave bursting

940

model.

941

A, sample 2-seconds voltage traces (left) and ISI histograms (right) at three temperatures

942

simulated with a mathematical model (described in Methods). B, Plots of different firing

943

pattern parameters (defined in Methods section) versus temperature, comparing modeled

944

data with the mean of experimental values obtained in corneal cold receptors from wild

945

type mice. C, ISI plot, number of spikes per burst and firing rate of a 200 seconds

946

simulation at 33C where the Ih maximum conductance, gh, was slowly reduced until

947

elimination. At t = 60 s gh begun to decrease from its initial value and reaching 0 at t=120 s.
45

948

After t=120 and until the end of the simulation, gh was 0 (see top trace). Firing parameters

949

before and after gh reduction are, respectively: mean firing rate: 7.9 and 3.9 s-1; cycle time:

950

123 and 332 ms; spikes per burst: 1 and 1.7.

951

Figure 8. Simulations of NTI activity with a model bearing a HCN2-like Ih.

952

A, sample 2-seconds voltage trace (left) and ISI histograms (right) obtained from

953

simulations at three temperatures. The only differences with the model shown in Fig. 7 are

954

the parameters of the Ih (see text). B, Plots of different firing pattern parameters (defined in

955

Methods section) versus temperature, comparing modeled data with the mean of

956

experimental values obtained in corneal cold receptors from HCN1 null mice. C, ISI plot,

957

number of spikes per burst and firing frequency of a 200 seconds simulation at 33C where

958

the Ih maximum conductance, gh, was progressively reduced until elimination (top trace).

959

Firing parameters before and after gh reduction are, respectively: mean firing rate: 7.35 and

960

3.9 s-1; cycle time: 180 and 332 ms; spikes per burst: 1.34 and 1.7.

961

Figure 9. Comparison of experimental and simulated firing patterns.

962

Mean firing frequency, mean number of spikes per burst and cycle time at different

963

temperatures from 24C to 34C are plotted for experimental recordings (circles) and

964

simulated terminals (triangles). Note the similarity between HCN1+/+ mice and the wt

965

model (filled symbols) and between HCN1-/- mice and the slow Ih model (empty

966

symbols).

967

Figure 10. Behavior of the model with a HCN2-like Ih with reduced maximum

968

conductance.

969

In the slow Ih+reduced gh model, parameters are identical to the wt model (Figure 7),

970

except for V0h=-95 mV, h=900 ms, gh=0.2. The only difference with the slow Ih model
46

971

(Figure 8) is the gh value. A, comparison of the firing rate, cycle time, spikes per burst and

972

fraction of skip events at different temperatures. Note that the model with reduced gh

973

behaves almost identically to the slow Ih model. B, effect of blocking Ih in the model at

974

33C. Note that, as occurs with the slow Ih model, the reduction of gh causes an increase of

975

the cycle time without a noticeable induction of bursting, resulting in a reduction of the

976

firing rate. Firing parameters before and after gh reduction are, respectively: mean firing

977

rate: 7.9 and 3.9 s-1; cycle time: 220 and 332 ms; spikes per burst: 1.55 and 1.7.

978

Figure 11. Role of Ih in the membrane voltage oscillation.

979

Time course of a simulated voltage trace (top row), the activation variable of the slow

980

depolarizing current asd and the hyperpolarization activated current ah (continuous and

981

dashed lines, respectively, second row), the first derivative of the voltage (third row) and

982

the total membrane conductance (bottom row). A, wt model. B, slow Ih model. C, model

983

without Ih (gh=0). Traces were simulated with the temperature set at 28C to emphasize the

984

role of Ih in bursting behavior. The dashed vertical line marks the time when dV/dt=0,

985

representing the start of a new oscillation. The plot of total membrane conductance

986

excludes the fast, spike generating conductances gd and gr. Note the different time scale

987

used in C.

988

47

989
990

TABLES
Table 1. Effect of ZD7288 on NTI shape parameters.
Maximum

Minimum

+Half-

Half-

dV/dt

dV/dt

width

width

1.02 0.03

0.99 0.06

1.04 0.04

1.00 0.02

0.97 0.03

0.94 0.03

1.07 0.02

0.99 0.03

0.99 0.01

0.98 0.01

+Amplitude

Amplitude

HCN1(+/+)

0.97 0.03

HCN1(-/-)

1.01 0.02

991
992

NTI parameters are expressed as the ratio between ZD7288 (20 M) and the control

993

condition. In all cases, the paired differences were not statistically significant (P > 0.15).

994

Data is presented as mean S.E.M.

995

48

996

Table 2. Effect of ZD7288 on the dynamic response parameters of CS nerve endings.

Parameter
Max. Firing Rate
during response to
cold (impulses/s)
Temperature
Threshold (C)
Spikes/Burst during
response to cold

Control

ZD7288

P-value

73.0 8.6

67.0 10.0

0.36

32.0 0.3

33.0 0.2

0.32

5.3 1.4

6.6 1.1

0.44

997
998

Values are reported as mean SEM. n=8. P-value is the result of a paired t-test between

999

data in the absence and in the presence of 20 M ZD7288.

1000
1001

49

Control

500 ms
30
20
10
0
0.8
0.4
0.0
8
6
4
2
35

T (C)

25
15
0

Cycle time (s)

(impulses/s)

8
6
4
2
0

ol 288
ntr
C o ZD 7

D
20 V

10

Time (min)
0.8
0.6

15

**

0.4
0.2
0.0

ol 288
ntr
C o ZD 7

20

Spikes per burst

Sp/Burst

ISI (s)

impulses/s

Mean Freq.

20 M ZD7288

4
3

2
1
0

ol 288
ntr
C o ZD 7

Control
ZD7288

1 ms

20 M ZD7288

Control

500 ms
20
10
0
1.0
0.5
0.0
14
10
6
2
35
25
15
15

15

25

Cycle time (s)

Mean Freq.

(impulses/s)

20

10
5
0

ol 288
ntr
C o ZD 7

0.8
0.6

30

Time (min)

**

0.4
0.2
0.0

ol 288
ntr
C o ZD 7

35

40

Spikes per burst

T (C)

Sp/Burst

ISI (s)

impulses/s

8
6

45

4
2
0

ol 288
ntr
C o ZD 7

15

15



10
5
0
15



10
5
0






24 26 28 30 32 34

 

6
4
2



10
5
0
15



10
5
0
15



10
5
0

0.0 0.2 0.4 0.6 0.8

 



0
24 26 28 30 32 34

 

fraction of
skipped events

Spikes per burst

12

15

 

 


16

Event count
(sq. root)

0.0 0.2 0.4 0.6 0.8

mean firing rate
(impulses/s)



10

 

Event count
(sq.root)




Event count
(sq. root)

Event count
(sq. root)

Event count
(sq. root)

Event count
(sq. root)




0.4
0.3
0.2
0.1
0.0


 

24 26 28 30 32 34

 

normalized
firing rate

4
3
2
1
24 26 28 30 32 34

 

20 M ZD7288

20
10
0
1.2
0.8
0.4
0.0
6
4
2
35
25
15

10
 

 

8
6

4
2
0

ol 288
ntr
C o ZD 7

10

0.8
0.6

15

Time (min)

0.4
0.2
0.0

ol 288
ntr
C o ZD 7

20

25

   

   

 





HCN1-/-

30

4
3
2
1
0

ol 288
ntr
C o ZD 7



30 M Ivabradine

20
10

0
1.2
0.8
0.4



0.0
8
6
4
2

 

35
25
15

   

6
 

 

20

4
2
0

ol
br
ntr Iva
Co

0.4
0.3

Time (min)

**

0.2
0.1
0.0

ol
br
ntr Iva
Co

40

60

   

ol
br
ntr Iva
Co





0.8

 



150
100
50
10
0
1.2

3 mM CsCl

0.4
0.0
8
6
4
2
35
25
15
10

20

30

Time (min)

40

50

60

'wild type'
model
100
event count

V (mV)

-40

event count

V (mV)

30C

100

-40
-80

50

0
400

event count

24C

200

-40
-80

500 ms

Mean freq.
(impulses/s)

16
12
8

model
exp. data

4
0

fraction of skip

Spikes per burst

6
4
2
0

Sp/Burst impulses/s

gh

22

26

0.0

0.2

0.4

0.6

ISI (s)

0.4
0.3
0.2
0.1
0.0

ISI (s)

cycle time (s)

V (mV)

50

0
150

-80

34C

30

temperature (C)

0.8
0.6
0.4
0.2
0.0

34

22

26

30

temperature (C)

0.6
0.0
10
5
0
4
2
0
0.8
0.4
0.0

40

80

120

Time (s)

160

200

34

model with slow Ih

100

event count

V (mV)

0
-40
-80

event count

V (mV)

-40

50

0
400

event count

24C

200

-40
-80

500 ms

12
8

model
exp. data

4
0

4
0

0.6

30

34

0.5
0.4
0.3
0.2
0.1
0.0

0.00

22

gh

0.4

ISI (s)

0.05

impulses/s

0.2

0.10

Sp/Burst

0.0

fraction of skip

ISI (s)

cycle time (s)

Mean freq.
(impulses/s)

16

Spikes per burst

30C

100

-80
V (mV)

50

0
150

34C

26

30

temperature (C)

34

22

26

temperature (C)

0.6
0.0
10
5
0
4
2
0
0.8
0.4
0.0

40

80

120

Time (s)

160

200

Mean freq.
(impulses/s)

16
12
(+/+)

HCN1

HCN1
wt model
slow Ih model

(-/-)

Spikes per burst

6
4
2

Cycle Time (ms)

0
500
400
300
200
100
0
22

26

30

Temperature (C)

34

Cycle time (ms)

Mean rate
(impulses/s)

18
12
6

500

Experimental
HCN1+/+
HCN1-/-

250

0.8

Model

fraction of skips

Spikes per burst

wt

4
2
0
28

32

Temperature (C)

gh
impulses/s
Sp/Burst

0.6
0.4
0.2
0.0

24

slow Ih
slow Ih gh=0.2

36

24

28

32

Temperature (C)

36

0.2
0.0
10
5
0
4
2
0

ISI (s)

0.8
0.6
0.4
0.2
0.0

50

100

Time (s)

150

200

V (mV)

-50

0
-50

0.10

dV/dt

0.00

0.3

0.0
0.5
0.0
-0.5

asd

0.2

0.00

0.1
0.0

0.2
0.1
0.0

50 ms

0.2
0.0
0.5
0.0
-0.5
0.3

gmemb

gmemb

0.3

0.2

-50

0.4
0.05

dV/dt

0.0
0.5
0.0
-0.5

asd

0.2

ah

asd
dV/dt

0.4
0.05

0.10

0.4

gmemb

gh=0

ah

V (mV)

slow Ih model

V (mV)

wt model

0.2
0.1
0.0

50 ms

100 ms