Po j Neurophysiol Role of Ih in the Firing Pattern of Mammalian Cold

7/23/2019 Po j Neurophysiol Role of Ih in the Firing Pattern of Mammalian Cold

1/60

ROLE OF Ih IN THE FIRING PATTERN OF MAMMALIAN COLD1

THERMORECEPTOR ENDINGS2

3

Patricio Orio1, Andrs Parra

2, Rodolfo Madrid

3, Omar Gonzlez

2,4, Carlos Belmonte

2 and4

Flix Viana2.5

1Centro Interdisciplinario de Neurociencia de Valparaso CINV, Facultad de Ciencias,6

Universidad de Valparaso, 2360102 Valparaso, Chile.

7

2Instituto de Neurociencias de Alicante, Universidad Miguel Hernndez-CSIC, 035508

Alicante, Spain.9

3 Departamento de Biologa, Facultad de Qumica y Biologa, Universidad de Santiago de10

Chile, 9160000 Santiago, Chile.11

4Fundacin de Investigacin Oftalmolgica, Instituto Fernandez-Vega, Oviedo (Spain).12

13

Running Head: Role of h-currents in cold thermoreceptor endings14

15

Author contributions: P.O., R.M., F.V., and C.B. designed the experiments. A.P., O.G.16and R.M. performed the experiments. P.O. and A.P. analyzed the experimental data. P.O.17

performed and analyzed mathematical simulations. P.O. and F.V. wrote the paper.1819

20

21

2223

24Contact Information:25Dr. Patricio Orio26

Centro Interdisciplinario de Neurociencia de Valparaso27

Universidad de Valparaso28Gran Bretaa 111129

Articles in PresS. J Neurophysiol (September 5, 2012). doi:10.1152/jn.01033.2011


2/60

ABSTRACT34

Mammalian peripheral cold thermoreceptors respond to cooling of their sensory endings35

with an increase in firing rate and a modification of their discharge pattern. We recently36

showed that cultured trigeminal cold-sensitive (CS) neurons express a prominent37

hyperpolarization activated current (Ih), mainly carried by HCN1 channels, supporting38

subthreshold resonance in the soma without participating in the response to acute cooling.39

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

mice,40

still reveals a deficit in cold perception. Here we investigated the role of Ih in CS nerve41

endings, where cold sensory transduction actually takes place. Corneal CS nerve endings in42

mice show a rhythmic spiking activity at neutral skin temperature that switches to bursting43

mode when the temperature is lowered. The Ihblockers ZD7288 and Ivabradine alter the44

firing patterns of CS nerve endings, lengthening the inter-spike intervals and inducing45

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

46

mouse corneas and found that they behave similar to the wild type, although with a lower47

slope in the firing frequency versus temperature relationship thus explaining the deficit in48

cold perception of HCN1-/-

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

mice49

was also affected by ZD7288, a result that we attribute to the presence of HCN2 channels50

in the place of HCN1. Mathematical modeling shows that the firing phenotype of CS nerve51

endings from HCN-/-

mice can be reproduced by replacing HCN1 channels with the slower52

HCN2 h l th th b b li hi I W th t I i d b HCN1 h l53


3/60

Keywords: Hyperpolarization-activated current, HCN channel, sub-threshold oscillation,57bursting, cold thermoreceptors.58


4/60

INTRODUCTION59

Mammalian cold thermoreceptors are sensory terminals specialized in the detection of60

innocuous cold/cool temperatures impinging on the skin (Hensel 1981). The transduction of61

cold stimuli takes place in the free nerve endings of the peripheral axons of cold62

thermoreceptor neurons from trigeminal and dorsal root ganglia. They display spontaneous,63

tonic or bursting, spike activity at neutral skin temperatures that is accelerated by cooling64

the receptive field and suppressed by warming (Braun et al. 1980; Brock et al. 2001; Dykes65

1975; Iggo 1969). The average frequency of their static discharge has a bell-shaped66

relationship with respect to the skin temperature, with the maximum frequency of single67

cold thermoreceptors ranging from 18C to 34C. Therefore, the action potential rate cannot68

encode unambiguously the peripheral temperature under static conditions (Bade et al. 1979;69

Dykes 1975; Hensel and Wurster 1970) suggesting that discrimination depends on the70

temporal structure of the impulse sequence (Braun et al. 1980; Hensel 1981). At static71

temperatures above ~30C the activity consists mainly in regularly fired single spikes,72

while at lower temperatures the bursting activity prevails (Iggo 1969). The transitions73

between these different spiking patterns are continuous, and their temporal structure74

suggests that all the different patterns can be attributed to oscillations of the membrane75

potential which are systematically altered by temperature (Braunet al., 1980; Braunet al.,76

1990; Schafer et al., 1991). Acute cooling leads to a transient rise in mean frequency77

d b i di t i i th f f b t f ll d b i i b t78


5/60

Gabbiani 2004) and it has been proposed that bursting enhances the sensitivity and working82

range of cold receptors (Iggo 1969). However, the contribution of this temporal firing code83

to the psychophysical capacities for cold discrimination in vivohas not been established.84

The cellular and molecular mechanisms that confer specificity for thermal detection at cold85

nerve endings have been unveiled in recent times. Cooling and cooling compounds, like86

menthol, activate TRPM8, a non-selective, calcium-permeable, cation channel of the87

transient receptor potential family (reviewed in Babes et al. 2011; Latorre et al. 2011;88

McKemy et al. 2002; Peier et al. 2002). This channel is selectively expressed in cold89

sensitive (CS) neurons and is a major determinant of their cold sensitivity (Bautista et al.90

2007; Colburn et al. 2007; Dhaka et al. 2007). CS neurons also possess background K+91

channels that are closed by cooling contributing to depolarization and firing (Reid and92

Flonta 2001; Viana et al. 2002), and it has been shown that these channels participate in93

cold perception (Noel et al. 2009). In contrast, the ionic mechanisms underlying the94

rhythmic firing activity exhibited by peripheral cold receptors are still unknown.95

Experimental evidence suggests the involvement of slow, TTX-resistant persistent sodium96

currents (Brock et al., 1998; Herzog et al., 2001) and low-threshold calcium channels97

(Schafer et al., 1982; Schafer et al., 1991) in the underlying oscillations that generate98

rhythmic firing. Mathematical simulations (Braun et al. 1998; Longtin and Hinzer 1996)99

have shown that the usual effect of temperature on ion channel gating kinetics (Q10~3,100

(Hill 2001)) i h t d th t t i d d h f iki tt i101


6/60

(Orio et al. 2009). The Ih is a key regulator of cellular excitability and contributes to105

rhythmic firing in neurons at various levels of the nervous system (Pape 1996) and the heart106

(DiFrancesco 2006; Schreiber 2003). For example, Ih participates in the generation of107

spindle waves and single cell oscillations in the thalamus (Luthi et al. 1998; McCormick108

and Pape 1990), and subthreshold resonance behavior in cortical cells (Hutcheon et al.109

1996; Richardson et al. 2003) and hypoglossal motoneurons (van Brederode and Berger110

2011). Ih is carried through channels of the hyperpolarization-activated and cyclic111

nucleotide-gated (HCN) ion channel subfamily (reviewed by Hofmann et al. 2005; Pape112

1996; Robinson and Siegelbaum 2003), which include four members (HCN1-4)113

characterized by six membrane-spanning domains and one cyclic nucleotide-binding114

domain (Ludwig et al. 1998; Santoro et al. 1998). Individual HCN subunits assemble as115

homotetrameric channels with distinct voltage dependency and kinetic properties (reviewed116

by Kaupp and Seifert 2001). Further diversity is obtained by co-assembly of the different117

HCN subunits (Chen et al. 2001; Ulens and Tytgat 2001). In CS neurons, the kinetic and118

activation properties of Ih suggest a major involvement of HCN1 channels with some119

participation of the HCN2 isoform (Orio et al. 2009). Indeed, CS neurons from HCN1 gene120

knock-out mice show no Ih or a slower Ihof reduced amplitude. Behavioral responses to121

cooling are impaired in HCN1 null mice and after peripheral pharmacological blockade of122

Ih (Orio et al. 2009), suggesting an important role of Ih in cold thermoreception. However,123

the actual role of Ihin the process of thermotransduction and coding of cold temperatures124


7/60

Fortunately, a preparation that allows long-term extracellular recordings from corneal129

sensory nerve endings in-vitro allowed us to pursue this goal (Brock et al. 1998; Parra et al.130

2010). Here, we characterize the effects of pharmacological blockade of Ih in the131

spontaneous and temperature-dependent firing patterns of cold sensitive nerve endings in132

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

mice. A133

change of the firing pattern upon blockade of Ih with ZD7288 or Ivabradine suggests an134

important role of HCN channels in tuning and maintaining the slow oscillation underlying135

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

mice corneas, firing patterns136

show less marked changes than those we observed under pharmacological suppression of137

HCN channels, when compared to wild-type data. Based on previous results (Orio et al.138

2009), we propose that this is due to functional expression of HCN2 channels in CS nerve139

endings, a hypothesis that is further supported by mathematical modeling. We conclude that140

Ih plays a critical role in shaping the temperature-dependent firing patterns in cold141

thermoreceptors, determining the oscillation period and the length of bursts that underlie142

their rhythmic activity.143


8/60

MATERIALS AND METHODS144

Animals145

All animal experiments were conducted at the Institute of Neurosciences in Alicante146

(Spain). Experimental protocols were supervised and approved by the Director of Animal147

Research Services, and were performed according to EC animal use guidelines148

(2007/526/EC). 20 HCN1+/+

mice and 7 HCN1-/-

mice were used. Animals were sacrificed149

by exposure to CO2 and decapitated. Eyes were carefully removed and placed in a150

recording chamber. The mouse line HCN1-/-

was generated by the laboratory of E. Kandel151

(Columbia University) and obtained from The Jackson Laboratory (stock number 101045).152

Mice were genotyped by PCR.153

Extracellular recording of single nerve terminal impulses154

To characterize the firing of cold-sensitive nerve endings, we used an in vitropreparation155

of mouse cornea developed in the laboratory, as described recently in Parra et al. (2010).156

Briefly, the optic nerve and associated tissues of the eye were drawn into a suction tube at157

the bottom of a small recording chamber and eyes were continuously perfused (1 ml/min)158

with a solution of the following composition (mM): NaCl (128), KCl (5), NaH2PO4 (1),159

NaHCO3(26), and glucose (10). The solution was gassed with carbogen (95%O2, 5%CO2)160

and maintained at the desired temperature with a manually-controlled Peltier device. Glass161

micropipette electrodes filled with the same solution were applied to the cornea with light162


9/60

set at 0.1 Hz). Data were captured and analyzed using a CED 1401 interface coupled to a167

computer running Spike 2 software (Cambridge Electronic Design, Cambridge, UK).168

Temperature was simultaneously acquired at 10 samples/s.169

Analysis of burst patterns170

Nerve terminal impulses (NTIs) were detected during acquisition with a threshold-based171

criterion and further filtered manually to remove artifacts. Only single-unit recordings172

(evidenced by a spikes having the same shape and similar size), or multi-unit recordings173

where spike sorting was easily achieved by different spike shapes, were used. Following174

sorting, inter-spike intervals (ISIs) for individual units were calculated. A burst is defined175

to begin when the ratio of two consecutive ISIs (ISIn-1:ISIn) is higher than 2.5. Conversely,176

a burst ends when the ratio ISIn-1:ISIn is lower than 0.4. Also, a maximum intra-burst177

interval (MaxIntra) and a minimum inter-burst interval (MinInter) were defined such that if178

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


10/60

was defined as the main cycle of the oscillation (1stCT). Skip events. A skip event (Braun et189

al. 1980; Longtin and Hinzer 1996) was defined when ISI > 1stCT * 0.6.190

Mathematical modeling of cold-sensitive nerve endings191

The electrical activity of cold-sensitive nerve endings was modeled using a modified192

version of the Huber-Braun model (Braun et al. 1998) that includes an hyperpolarization-193

activated current (Ih) and a TREK1-like potassium current. The change in membrane194

voltage (V) is given by195

wtrekhsrsdrdlm gIIIIIIIdt

dVC = , (1)196

where Cmis membrane capacitance,Iiare ionic membrane currents andgwis a noise term.Il197

stands for leak current, such that Il=gl(V-El), being g the conductance and E the reversal198

potential. Id and Isd are depolarizing and slow depolarizing currents, respectively. Id199

represents a fast sodium channel that generates action potentials while Isdis a mixed sodium200

and calcium slow conductance (Longtin and Hinzer 1996; Plant 1981); a simplification that201

assumes separate calcium and sodium channels with similar activation properties. The202

activation properties ofIsdresembles NaV1.9 channels (Herzog et al. 2001) and the fraction203

of calcium conductance is controlled by the parameter (see below). Ir and Isr are204

repolarizing and slow repolarizing currents (i.e. potassium currents), respectively. Itrekis a205

temperature-sensitive potassium current. Currents are given by206

)( iiii EVagI = , (2)207


11/60

where is a temperature factor for channel kinetics. ad is considered to be instantaneous211

such that

= dd aa , defined as in eq. (4)212

asrresembles a calcium activated potassium channel with the expressions213

n

d

n

n

srKCa

Caa

+= and

Ca

sd CaI

dt

dCa

= . (5),(6)214

Finally,215

+

+=

10exp1

8.02.0

10

TTq

am

trek (7)216

Temperature scaling factors are given by217

10

25

10

25

0.3,3.1

==TT

(8)218

The noise component (gw) was modeled as an exponentially correlated noise in the form of219

an Ornstein-Uhlenbeck process:220

w

ww gdt

dg

+= (9)221

where is drawn at each integration step from a normal distribution with mean 0 and222

variance D.223

Parameters used for the wt model are:224

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=50,225

Er=Esr=Etrek=-90, Eh=-30, El=-60, Vd0=Vr

0=-25, Vsd

0=-40, Vh

0=-85, sd=sr=0.25, ssd=0.11,226


12/60

The model was implemented and solved in the NEURON simulation environment231

(http://www.neuron.yale.edu) (Hines and Carnevale 1997). All the files necessary to run the232

simulations presented here are available in ModelDB233

(http://senselab.med.yale.edu/ModelDB/, accession number 141063).234

Statistical Analysis235

Unless stated otherwise, data is expressed as mean S.E.M. Data was compared using236

Students t-test. Paired tests were applied when appropriate.237

Drugs238

ZD7288 (4-Ethylphenylamino-1,2-dimethyl-6-methylaminopyrimidinium chloride) was239

purchased from Tocris Bioscience (UK). Ivabradine (3-(3-{[((7S)-3,4-240

dimethoxybicyclo[4,2,0]octa-1,3,5-trien-7-yl)methyl] methylamino}propyl)-1,3,4,5-241

tetrahydro-7,8-dimethoxy-2H-3-benzazepin-2-one, hydrochloride) was a generous gift of242

the Institut de Recherches Internationales Servier (France).243

244

245


13/60

RESULTS246

Blockade of Ihalters spontaneous firing activity but not acute response to cold in nerve247

terminals.248

To investigate the function of Ih at sensory nerve terminals, we took advantage of an in249

vitro preparation of the adult mouse cornea that allows long term recordings of cold250

thermoreceptor endings (Parra et al. 2010). Figure 1A shows extracellular recordings of251

single nerve terminal impulses in a CS nerve terminal. The time course of the recorded252

electrical activity in the same nerve ending can be seen in Figure 1B, which plots the firing253

frequency, the inter-spike intervals, the number of spikes per burst and the temperature of254

the bathing solution during the complete experiment. In the control condition (Figure 1A,255

top left) the terminal shows spontaneous nerve impulses with a frequency around 7256

impulses/s. When cooled transiently to 20C, the terminal responded with an increase in257

mean firing frequency up to 30 impulses/s (Figure 1B). Following application of 20 M258

ZD7288, a selective blocker of HCN channels (Harris and Constanti 1995), to the bath259

solution perfusing the cornea, the most obvious effect was a slowdown of the mean spiking260

frequency (an increase of the mean ISI) shortly followed by the appearance of bursts261

(evidenced by the appearance of shorter ISIs and an increase in the number of spikes per262

bursts, see Figure 1B and Figure 1A, right). On the other hand, application of ZD7288 did263

not induce significant changes in the shape of NTI spikes (Figure 1D and Table 1),264

i di ti th t ZD7288 i t ff ti th ti h i ibl f ti265


14/60

pulses). The maximum firing frequency during the cooling stimulus showed no difference269

between the control condition and when ZD7288 was present (Table 2). Moreover, Table 2270

shows that application of the drug did not change significantly the threshold for the increase271

in firing frequency in response to cold, in agreement with our previous findings in cultured272

CS trigeminal neurons where ZD7288 did not change the temperature threshold for action273

potential firing (Orio et al. 2009). Finally, ZD7288 did not change the bursting274

characteristics of the response to cooling ramps (Table2).275

From these observations we conclude that blockade of HCN channels in CS nerve terminals276

does not affect the response to an acute cold stimulus, but alters profoundly the temporal277

pattern of the spontaneous firing of action potentials in a steady-state temperature278

condition. We also tested the effect of ZD7288 in CS nerve endings from guinea pig279

corneas (Brock et al. 1998) and found very similar results (not shown).280

Blockade of Ihalters stationary firing pattern at low temperatures281

We wanted to examine whether the changes of firing pattern observed upon Ih blockade282

could also be observed at temperatures lower than the resting temperature of 34C. Figure 2283

shows the behavior of a mouse CS nerve ending that was exposed to 20 M ZD7288 while284

maintained at a constant temperature of 27C. In this case, the terminal showed a bursting285

firing pattern right from the beginning of the recording (Figure 2A, left, and 2B), typical for286

temperatures lower than 30C. After some minutes in the presence of ZD7288, it becomes287

evident that the firing pattern suffers the same alteration as seen at 34C; namely the288


15/60

used in Figure 1: there is a significant increase in cycle time and in the number of spikes292

per burst, while the firing frequency is not significantly affected (Figure 2C).293

Overall, these results indicate that the participation of Ihin the oscillatory properties of CS294

nerve endings is also required for their characteristic bursting behaviors at lower295

temperatures, and therefore may be critical for the neural coding of stationary temperatures.296

Firing patterns of corneal cold sensitive nerve endings in wild type and HCN1 null mice297

To investigate further the potential role of HCN channels in the coding of cold-evoked298

responses, we used an alternative approach to the pharmacological blockade of Ih, taking299

advantage of an HCN1 null mouse line (Nolan et al. 2003). Figure 3A shows representative300

recording traces and inter-spike intervals (ISI) histograms for the steady state firing of NTIs301

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

(left) and HCN1-/-

302

(right) animals. In both cases, the decrease in temperature is associated to longer cycle303

times (the larger ISIs) and bursting becoming more prominent (higher number of events304

with short ISI). Notably, bursting is absent at 34C in the wild type nerve ending while it is305

evident at the same temperature in the HCN1 null animal. Figure 3B-E summarizes the306

firing pattern parameters of several nerve endings over the entire temperature range (24-307

34C). Although there was notable variability between individual nerve endings in terms of308

firing patterns and mean firing rates, the overall trends are maintained within the309

population. Nerve endings from HCN1 null mice have a higher mean firing frequency at310

temperatures between 33C and 35C than those from wild type animals, but below 32C311


16/60

means that at 30C and above, there is a higher probability in wild typenerve endings of315

skipping a cycle of the oscillation (Braun et al. 1980), a phenomenon that is evident in the316

ISI histograms as peaks at approximately the double of the main ISI values observed (see317

Figure 3A at 34C, arrow).318

The data presented above suggests that corneal CS nerve endings from HCN1 null mice319

have subtle but significant differences in firing pattern compared to wild type siblings.320

Although the firing frequency versus temperature curves (Figure 3B) are not significantly321

different (P=0.16 in a two-way ANOVA test), in the HCN1 null nerve endings there is a322

significantly higher probability of firing action potentials at normal body surface323

temperature (p=0.02 non paired t-test for the data at 34C). Moreover, the slope of the firing324

frequency versus temperature relationship is lower in HCN1 null animals compared to wild325

type (Figure 3B, fit lines). In other words, during moderate cooling from 34C to 28C a CS326

nerve ending in a wild type animal increases its firing frequency by a factor of 3.1, while in327

the case of a HCN1-/- animal this factor is only 1.7. This difference grows at lower328

temperatures (Figure 3E).329

Effect of Ihblockade in CS nerve endings from HCN1-/-

mice330

The results so far indicate that pharmacological blockade of Ih in CS nerve endings has a331

more profound effect on the firing pattern than genetic deletion of HCN1. One possibility332

to explain this observation is that HCN1-/-

mice still have Ihin their CS nerve ending due to333

the expression of the HCN2 channel (Orio et al. 2009). Thus, we tested the effect of334


17/60

ZD7288 also induces a longer cycle time (Figure 4A grey rectangle and see also Figure 4B,338

center) with an increased number of spikes per burst (Figure 4B, right). In the average,339

however, there was a significant decrease of the mean firing frequency (Figure 4B left);340

probably because the longer cycle time was accompanied by a modest increase in the341

number of spikes per burst (Figure 4B, right) that failed to be significant. The latter may be342

related to the fact that, as already mentioned, cold receptors from KO animals showed343

bursting in the control condition. As in wild type nerve endings, ZD7288 produced no344

significant change in the shape of the nerve terminal impulses recorded (see Table 1).345

The effect of ZD7288 on CS nerve endings from HCN1-/-

animals also raises the possibility346

that this drug is acting on a molecular target other than HCN channels, as it has been347

suggested previously at mossy fiber synapses (Chevaleyre and Castillo 2002).348

Alternatively, in cold-sensitive terminals from HCN1 null animals there may still be a349

significant Ihcurrent, supported by expression of HCN2 channels, as it was observed in a350

fraction of cultured CS trigeminal neurons obtained from these animals (Orio et al. 2009).351

This in turn may explain why the firing pattern of HCN1-/-

CS terminals is not much352

different from that of HCN+/+

CS terminals, as there is still a hyperpolarization-activated353

current albeit with a more negative half-activation voltage and slower kinetics (Santoro et354

al. 2000). In an effort to discriminate between these two alternatives, we tested the effect of355

Ivabradine, another HCN channel blocker (Bois et al. 1996; Bucchi et al. 2006), on the356

electrical activity of wild type CS nerve endings. In this case we also found an increase of357


18/60

Ivabradine are only quantitative, with the same qualitative result (lengthening of cycle362

time) produced by both blockers. The partial effect of Ivabradine can be explained by363

differences in their blocking mechanisms, specifically, because Ivabradine needs repeated364

open/close cycles to effectively block HCN1 channels while ZD7288 can readily block365

open channels (Bucchi et al. 2006; Shin et al. 2001). Moreover we confirm that the effect of366

Ivabradine can be attributed to a partial block of Ih using mathematical simulations (see367

below). On the other hand, Cs+ion can also block HCN channels (DiFrancesco 1982), with368

additional effects on inward rectifier K+ channels (Hille 2001). We tested the effect of 3369

mM CsCl on the firing pattern of the spontaneous activity of CS terminals and found that it370

rapidly increased the cycle time and promoted bursts, resembling the effects of ZD7288 and371

Ivabradine. However, this was followed by a dramatic increase of activity where bursts372

could hardly be distinguished (Figure 6). This effect was poorly reversible, suggesting that373

Cs+may be acting intracellularly on inward rectifier potassium channels. Still, the overall374

results obtained are consistent with the idea that the principal mechanism of action of375

ZD7288 and Ivabradine is a blockade of Ih.376

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

terminals would come377

from the direct assessment of the activity of specific ionic conductances at mammalian378

nerve terminals; however, this is currently not feasible due to their small size. To further379

explore whether the compensation of HCN1 function by expression of HCN2 channels is a380

plausible explanation for the results described above, we developed and analyzed381


19/60

Mathematical models of cold receptors bearing HCN1 or HCN2 channels384

The rhythmic spiking activity of vertebrate cold thermoreceptors is thought to arise from an385

underlying oscillation of the membrane potential, upon which an action potential is386

generated each time firing threshold is reached (Braun et al. 1980; Schafer et al. 1990).387

Cold temperatures would slow down the intrinsic oscillation, allowing more than one action388

potential per cycle and thus giving rise to the bursting nature of the response. Experimental389

evidence points to the involvement of slow, TTX-resistant persistent sodium currents390

(Brock et al. 1998; Herzog et al. 2001) and low-threshold calcium channels (Schafer et al.391

1982; Schafer et al. 1991) in the generation of rhythmic firing.392

Mathematical models that account for the temperature-dependent change in spiking pattern393

of cold-sensitive nerve fibers have been published and analyzed (Braun et al. 1998; Huber394

et al. 2000; Longtin and Hinzer 1996), all based on Plants ionic model of slow wave395

bursting (Plant 1981). In this model, a slow oscillation is driven by a slow voltage-396

dependent sodium current that acts as positive feedback. Calcium entry through voltage-397

activated calcium channels, and the consequent activation of calcium-activated potassium398

channels, provides a negative feedback current. On top of this oscillation, fast voltage-399

activated sodium and potassium currents generate action potentials. By adding the typical400

effect of temperature on ion channel gating kinetics (a 3-fold speed up for every 10C401

temperature rise), a model is obtained in which the spiking pattern changes with402

temperature in a similar way as the cold-sensitive nerve endings and fibers.403


20/60

by a Q10 factor of 3). Thereafter, the parameters were hand tuned to reproduce two key408

observations of the firing pattern at static temperature in wtmice: the effect we observe in409

the firing pattern after Ihblockade (Figures 1,2) and the firing pattern versus temperature410

curves (Figures 3B-D). In the first case, the most important parameter change with respect411

to the original Brauns model was an increase in the voltage dependency of the slow412

depolarizing current activation (zsd), from an original value of 0.09 mV

-1

to 0.11 mV

-1

(at413

34C, this is equivalent to 2.4 and 2.9 electron equivalents, respectively). Persistent, TTX-414

resistant sodium currents have been described in trigeminal neurons with a voltage415

dependency a little bit higher than 0.1 mV-1

(Herzog et al. 2001). In the second case, to fine416

tune the model to fit the experimentally observed behavior of cold receptors we found it417

necessary to add a background K+ current specifically modulated by temperature in the418

form of a TREK1-like potassium channel (Maingret et al. 2000). Otherwise, a lower slope419

of the mean firing frequency versus temperature relationship was obtained (not shown).420

Figure 7A shows the firing patterns simulated by the model at three different temperatures,421

34C, 30C and 24C, and the corresponding ISI histograms. As shown in Figure 7B, the422

firing pattern parameters in the 24C 34C range of the model match very well the mean423

curves obtained in CS terminals from wild type mice. In the case of the percentage of skip424

events, the experimental curve can be regarded as a smoothed version of the simulated425

curve. This result is typically obtained when stochastic contributions are not modeled426

properly (Faisal et al. 2008), but tuning that aspect of the model is beyond the purpose of427


21/60

induction of bursts, as observed in mice terminals (see Figure 1B grey rectangle). Finally, a432

partial block of Ih(to a 16% of the original value) produces a behavior that resembles the433

effect of Ivabradine: a modest increase of cycle time (from 123 ms to 180 ms) and very434

little bursting (from 1 to 1.14 spikes/burst).435

To simulate a HCN1-/-

mouse nerve ending, rather than eliminating Ih we modified its436

parameters to make it resemble HCN2 channels. The changes consisted in a shift of the437

open probability curve towards negative potentials (Vh0

from -85 mV to -95 mV) and438

slower channel kinetics (hat 25C changed from 125 ms to 900 ms). This is in agreement439

with the properties of HCN2 channels in heterologous expression systems (Chen et al.440

2001) and the characteristics of Ih recorded in cultured CS neurons from mice lacking441

HCN1 (Orio et al. 2009). No other parameter of the model was changed.442

Remarkably, as shown in Figure 8A and 8B, the sole change of HCN1- to HCN2-like Ihis443

sufficient to make the model match the firing patterns of CS nerve endings from HCN1 null444

mice between 24C and 34C. The progressive reduction of Ih conductance in the model445

also reproduces the effect of ZD7288 on terminals lacking HCN1 (Figure 8C), namely an446

increase of the cycle time, which combined with a small increase in the number of spikes447

per burst results in a reduction of the mean firing rate. The effect of changing Ihparameters448

is more strikingly evidenced in Figure 9, where the experimental data of Figure 3B-D449

(circles) are plotted together with the data from both simulations (fast and slow Ih;450

triangles).451


22/60

conductance (gh changed from 0.6 to 0.2) could also reproduce the experimental results455

obtained in HCN1 ko terminals. As shown in Figure 10, expression of one third of Ih456

current with slow properties (HCN2-like) is enough to obtain a behavior similar to the457

observed in HCN1-/-

, and thus overexpression of HCN2 is not necessary for functional458

compensation.459

Analysis of the time course of model variables (see Appendix) shows that Ih provides a460

critical depolarizing drive at the beginning of the oscillations. The apparently subtle change461

in kinetic properties from HCN1 to HCN2 results in a diminished Ihcurrent, thus delaying462

the start of a new oscillation. At the same time, the reduced membrane conductance results463

in a lengthening of the burst period. From these results, we can conclude that the cold464

receptor phenotype observed in HCN1 null mice can be reasonably explained by a465

compensation of HCN1 channels by the remaining HCN2 related subunit.466

467


23/60

DISCUSSION468

In a previous study (Orio et al. 2009) we characterized extensively a prominent h current469

present in cultured cold-sensitive trigeminal neurons, hypothesizing about its potential role470

in the rhythmic generation of action potentials in cold-sensitive nerve endings.471

Pharmacological or genetic suppression of Ihin mice produced an altered cold perception in472

behavioral studies, further supporting a possible role for Ih in the stimulus coding of cold473

thermoreceptors (Orio et al. 2009). In this work, by taking advantage of the recently474

described development of extracellular recordings in mice corneal sensory endings (Parra et475

al. 2010), we have confirmed this hypothesis by directly determining the effect of476

pharmacological Ihsuppression on the firing patterns of cold-sensitive nerve endings from477

mouse cornea. Moreover, we have characterized the firing pattern under static temperature478

conditions of cold-sensitive nerve endings from knock-out mice for HCN1 channel479

subunits, the principal molecular determinant of Ihin cold thermoreceptor neurons.480

HCN1 channels are not only present in CS peripheral nerve endings and neurons, but also481

in many other neurons within the central nervous system. Thus, it is possible that other482

neurons involved in the behavioral response to cold are also affected by the lack of this483

channel. Therefore, the behavioral deficits observed can be the result of changes in the484

function of several neuronal types, rather than the sole changes in nerve ending firing485

patterns. While we cannot rule out this possibility with the present evidence, still our results486

h i t t l f I i th i iti l di f i ld t t487


24/60

Effects of Ihblockers on firing pattern491

The most notorious effect of Ih block on CS terminal NTIs is a rapid alteration in their492

firing pattern, occurring within a few minutes of exposure of the preparation to ZD7288 or493

Ivabradine, widely used Ihblockers. This change in firing pattern consisted in a lengthening494

of the mean cycle time and the induction of bursts. There were quantitative differences in495

the effect of both blockers, characterized by a lesser effect of Ivabradine compared to496

ZD7288. Differences between the effect of both compounds are to be expected, since they497

have a different blocking mechanism (Bucchi et al. 2006; Shin et al. 2001). Specifically, a498

lesser potency of Ivabradine is plausible, since it needs repeated open/close cycles to499

effectively block the channel, a situation that is not likely to occur if, according to the500

mathematical simulations, the nerve endings spend most of the time at less than -60mV.501

The observed change in firing pattern reveals an important role for Ih in the rhythmic502

generation of action potentials, which is a distinctive feature of CS nerve endings and fibers503

(Braun et al. 1980; Carr et al. 2003). A longer cycle time and the appearance of bursting504

suggest that the slow membrane potential oscillation thought to underlie this rhythmic505

activity (Braun et al. 1980; Braun et al. 1998) becomes slower when Ihis blocked, as has506

been shown in the inferior olive (Bal and McCormick 1997). Until direct recording of507

membrane potential at the tiny nerve endings becomes experimentally feasible this508

suggestion will remain unconfirmed, but the computer simulations that reproduce the effect509

of ZD7288 (Figures 7C and also Figure 11, Appendix) also show this effect, supporting this510


25/60

Role of HCN1 versus HCN2515

Our study shows that application of ZD7288 to CS nerve endings affects their firing pattern516

regardless of the expression of the HCN1 protein, which we showed to be responsible for a517

major fraction of the Ihobserved in the soma of cultured trigeminal CS neurons (Orio et al.518

2009). This result could be interpreted as ZD7288 acting on another (unspecific) target, an519

interpretation we cannot fully rule out. On the other hand, this finding could be explained520

by the presence of another member of the HCN channel family in CS nerve endings,521

partially fulfilling the role of HCN1. We think that the last possibility is a very plausible522

explanation, as the Ihin CS neurons resembles a mixture of HCN1 and HCN2 channels and523

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

mice524

(Orio et al. 2009). Moreover, our computational simulations show (Figure 8) that the525

behavior of HCN1-/-

CS nerve endings can be fully reproduced just by changing Ih526

activation parameters to make it resemble HCN2 channels.527

This result shows that the normal rhythmic firing of CS nerve endings requires the528

expression of a hyperpolarization-activated current with specific activating kinetics.529

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

mice530

may seem small, it can produce a significant difference in neural encoding of temperature.531

Just from the point of view of mean static firing frequency, CS terminals from HCN1 null532

animals have a lower slope of the firing frequency versus temperature relationship (see533

Figure 4B). As the neural coding underlying cold perception is thought to involve changes534


26/60

New kinetic model of CS nerve terminals537

The temperature-dependent firing patterns of CS nerve endings and fibers have been538

previously modeled using conductance-based models, notably in the studies of Longtin &539

Hinzer (1996) and Huber & Braun (Braun et al. 1998; Huber et al. 2000). These models are540

based on Plants model of slow wave bursting (Plant 1981) in which a slow oscillation of541

the membrane is driven by a slow persistent sodium/calcium current and a calcium-542

activated potassium channel. By adding the usual temperature dependencies of channel543

kinetics and conductance, the firing patterns of CS nerve endings at different static544

temperatures can be qualitatively reproduced. Huber & Brauns model stands out for545

simplified equations and low number of parameters, leaving only the essential while still546

being biologically plausible.547

Our results prompted the inclusion of a hyperpolarization-activated current in the model in548

order to explain the results with HCN channel blockers as well as the differences observed549

in the firing patterns between wtand HCN1-/-mice. Previous models omitted this current in550

simulating the overall behavior of CS nerve endings while still being adequate for551

explaining the qualitative aspect of firing pattern change with temperature. Our expanded552

model reproduces quantitative aspects of the firing patterns during cooling (see Figures 7B553

and 8B) as well as the effects of Ih blockade on the firing patterns. Furthermore, in the554

process of tuning the model to make it dependent on the expression of Ih, we found that this555

behavior is quite sensitive to the voltage dependency of the persistent sodium current (INa,p).556


27/60

part of the oscillation. With the higher voltage dependency we use (0.11, equivalent to 2.8561

charges per channel at 25C), that agrees with the values reported by Herzog et al. (2001),562

there are less sodium channels open at negative voltages and a new oscillation will take563

longer to start unless some other conductance (e.g. Ih in our model) helps in depolarizing564

the membrane again.565

As a result, we present a model that comprehensively reproduces all the experimental566

evidence here discussed. In doing so, it provides a plausible explanation for the effect of the567

Ih blocker on the CS nerve ending from HCN1(-/-)

animals, a result that at first appeared568

contradictory. The model also shows that the Ihis important for tuning the correct period of569

the slow oscillation: the opening of HCN channels adds a depolarizing drive and helps in570

the onset of the oscillation while their contribution to the total membrane conductance help571

to determine the burst duration. Replacing HCN1 channels with an HCN2-like572

conductance, with a different time constant and half-activation voltage, results in longer573

cycles with augmented bursting. Finally, the absence of Ih results in the longest cycles574

because of a more profound hyperpolarization of the membrane during the resting phase of575

the cycle.576

The role of cold-modulated conductances577

Previous models did not take into account the presence of a cold-inhibited hyperpolarizing578

(i.e. potassium) current, such as TREK1 and/or TRAAK channels (Kang et al. 2005; Reid579

and Flonta 2001; Viana et al. 2002), in CS neurons. We found that incorporating this type580


28/60

However, it is noteworthy that current models for temperature-induced changes in firing584

pattern (including ours) do not consider the presence of TRMP8, a channel member of the585

TRP channel family activated by decreases in temperature and menthol (Brauchi et al.586

2004; McKemy et al. 2002; Peier et al. 2002; Voets et al. 2004). One may wonder how this587

omission affects the interpretation of the results we obtained, considering the critical588

importance of this current in cold thermoreceptor function (Bautista et al. 2007; Colburn et589

al. 2007; Dhaka et al. 2007; Parra et al. 2010). TRPM8 is unspecific for monovalent590

cations: its activation causes depolarization of the membrane and a consequent increase in591

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

),592

the cold-induced inward current that can be recorded in cultured CS neurons and that is593

blocked by the TRPM8 blocker BCTC (Madrid et al. 2006), shows a fast desensitization594

upon a sustained cold stimulus (see Reid et al. 2002; Viana et al. 2002). This595

desensitization depends on phospholipase C activation, caused in turn by an increase in596

intracellular calcium (Daniels et al. 2009) and shows a time course similar to the adaptation597

of the dynamic response of CS nerve endings. Thus, it is possible that TRPM8 activity is598

mostly responsible for the transient responses to sudden temperature changes, playing a599

lesser role in establishing the firing patterns observed at static temperatures. Considering600

that the experimental work and the model here presented focus primarily on static601

responses of CS nerve terminals, the absence of TRPM8 should be of minor impact.602

Obviously, future models, trying to replicate dynamic aspects of thermoreceptor activity,603


29/60

Concluding remarks605

The hyperpolarization-activated current (Ih) has been implicated in pacemaking and606

oscillatory activity throughout the nervous system (Chan et al. 2004; Hutcheon et al. 1996;607

Luthi et al. 1998; McCormick and Pape 1990; Thoby-Brisson et al. 2000). In addition, I h608

has been recorded in sensory neurons (Doan et al. 2004; Momin et al. 2008; Orio et al.609

2009; Scroggs et al. 1994) and the expression of HCN channels has been demonstrated in610

sensory neurons and axon terminals (Antal et al. 2004; Kouranova et al. 2008; Momin et al.611

2008; Tu et al. 2004) with its abnormal expression associated with pathologic pain (Emery612

et al. 2011; Luo et al. 2007). The high-pass filter properties of Ihhave been implicated in613

sensory coding by retinal networks (Cangiano et al. 2007; Publio et al. 2009) and in the614

phase-locked response of hypoglossal motoneurons to sinusoidal stimulus (van Brederode615

and Berger 2011). Here, we demonstrate a role of Ihin rhythmic and oscillatory activity of616

cold thermoreceptor terminals. When put together with our previous observation of a617

behavioral deficit in cold sensing in mice lacking HCN1 or in the presence of Ihblockers618

(Orio et al. 2009), our results support the idea that the rhythmic discharge of action619

potentials in CS nerve endings plays a significant role in the sensory coding process, as was620

originally hypothesized more than 30 years ago by Braun and colleagues (Braun et al.621

1980). Finally, our results emphasize the notion that cold temperature perception is a622

complex sensory process that involves additional sensory events beyond the initial623

transduction step.624


30/60

APPENDIX627

The role of Ih in the membrane potential oscillation628

We examined closely the time course of the voltage and activation variables in the629

simulation, trying to understand what the role of Ihcan be and why a change in Ihactivation630

parameters can affect the bursting pattern and firing frequency. Figure 11A-C presents this631

examination graphically. In the wt model (Figure 11A), at the beginning of a voltage632

oscillation (dV/dt=0, vertical segmented line) the open probability of the h current (ah) is633

increasing because of membrane hyperpolarization. Since the reversal potential for the h634

current is -30mV, at negative membrane potentials it provides a depolarizing force that635

precedes the positive feedback of the slow depolarizing current Isd, thus accelerating the636

beginning of a new cycle (note that asd the open probility of Isd starts to grow when ah is637

almost at its maximum). In the slow Ih model (Figure 11B), ah displays a lower value638

during the cycle, because of a more negative half-activation voltage and the slower639

activation kinetics, limiting the opening of h channels during the hyperpolarization. This640

results in a weaker depolarizing drive for the onset of a new firing cycle. Finally, in the641

absence of Ih (Figure 11C), the new cycle in the model is initiated very slowly by the642

depolarizing positive feedback provided by Isd alone. This causes a very long inter-burst643

period and cycle times as seen in the experimental results with Ihblockade. It is also worth644

noting that the absence or reduction of Ihchanges dramatically the voltage reached during645

th h l i ti h f th ill ti hil th t d l h i646


31/60

A third important observation in the simulations is the increase in burst duration produced649

by the reduction or absence of Ih. This may result counter-intuitive, considering that Ih is650

depolarizing for most of the time during the cycle and thus should contribute in maintaining651

the firing of action potentials. However, we like to note that when the voltage is near -50652

mV, Ih is only weakly depolarizing and as an extra membrane conductance it actually653

dampens the depolarizing force of Isd. In the absence of Ih, more potassium conductance654

has to build up to dampen Isd and this delays the end of the firing phase. The bottom row of655

Figure 11 shows that the bursting phase always ends when the total membrane conductance656

reaches a similar value of 0.3 mS/cm2. When the Ih is decreased or absent, it takes more657

action potentials to reach this value because the initial conductance is lower.658

This analysis not only explains the dramatic change in firing pattern when Ih is fully659

blocked; it also shows how subtle changes in channel kinetics and activation curve given by660

a shift from HCN1 to HCN2 channels introduce significant alterations in the firing661

properties of cold thermoreceptor endings. In turn, this may explain the behavioral deficits662

in cold perception observed in HCN1-/-

animals (Orio et al. 2009).663


32/60

ACKNOWLEDGMENTS664

The authors thank E. Quintero, A. Perez and A. Miralles for excellent technical assistance665

and Tansy Donovan-Rodriguez for participating in preliminary pharmacological666

experiments.667


33/60

GRANTS668

This work was supported by funds from the Chilean Comisin Nacional de Investigacin669

Cientfica y Tecnolgica CONICYT (PBCT PSD20 and FONDECYT 11090308 to P.O.,670

FONDECYT 1100983 to R.M. and ACT-1113 to P.O. and R.M.), the Universidad de671

Valparaso (DIPUV 51/2007 to P.O.), the Universidad de Santiago de Chile (VRID-672

USACH to R.M.) and the Spanish MICIIN (BFU2008-04425 and CONSOLIDER-673

INGENIO 2010 CSD2007-00023 to C.B., SAF2010-14990 to F.V.). The CINV is funded674

by Grant P09-022-F of the Millennium Science Initiative (Ministerio de Economa, Chile).675

676


34/60

References677

678

Antal M, Papp I, Bahaerguli N, Veress G, and Vereb G. Expression of679

hyperpolarization-activated and cyclic nucleotide-gated cation channel subunit 2 in axon680

terminals of peptidergic nociceptive primary sensory neurons in the superficial spinal681

dorsal horn of rats.Eur J Neurosci 19: 1336-1342, 2004.682

Babes A, Ciobanu AC, Neacsu C, and Babes RM. TRPM8, a sensor for mild cooling in683

mammalian sensory nerve endings. Curr Pharm Biotechnol 12: 78-88, 2011.684

Bade H, Braun HA, and Hensel H. Parameters of the static burst discharge of lingual cold685

receptors in the cat.Pflugers Arch 382: 1-5, 1979.686

Bal T, and McCormick DA. Synchronized oscillations in the inferior olive are controlled687

by the hyperpolarization-activated cation current I(h). Journal of neurophysiology 77:688

3145-3156, 1997.689

Bautista DM, Siemens J, Glazer JM, Tsuruda PR, Basbaum AI, Stucky CL, Jordt SE,690

and Julius D. The menthol receptor TRPM8 is the principal detector of environmental691

cold.Nature 448: 204-208, 2007.692

Bois P, Bescond J, Renaudon B, and Lenfant J. Mode of action of bradycardic agent, S693

16257, on ionic currents of rabbit sinoatrial node cells. British journal of pharmacology694

118: 1051-1057, 1996.695

Brauchi S, Orio P, and Latorre R. Clues to understanding cold sensation:696


35/60

Braun HA, Huber MT, Dewald M, Schafer K, and Voigt K. Computer simulations of701

neuronal signal transduction: The role of nonlinear dynamics and noise. Int J Bifurcat702

Chaos 8: 881-889, 1998.703

Brock JA, McLachlan EM, and Belmonte C. Tetrodotoxin-resistant impulses in single704

nociceptor nerve terminals in guinea-pig cornea.J Physiol 512 ( Pt 1): 211-217, 1998.705

Brock JA, Pianova S, and Belmonte C. Differences between nerve terminal impulses of706

polymodal nociceptors and cold sensory receptors of the guinea-pig cornea. J Physiol 533:707

493-501, 2001.708

Bucchi A, Tognati A, Milanesi R, Baruscotti M, and DiFrancesco D. Properties of709

ivabradine-induced block of HCN1 and HCN4 pacemaker channels. The Journal of710

physiology 572: 335-346, 2006.711

Cangiano L, Gargini C, Della Santina L, Demontis GC, and Cervetto L. High-Pass712

Filtering of Input Signals by the Ih Current in a Non-Spiking Neuron, the Retinal Rod713

Bipolar Cell.PLoS ONE 2: e1327, 2007.714

Carr RW, Pianova S, Fernandez J, Fallon JB, Belmonte C, and Brock JA . Effects of715

heating and cooling on nerve terminal impulses recorded from cold-sensitive receptors in716

the guinea-pig cornea.J Gen Physiol 121: 427-439, 2003.717

Colburn RW, Lubin ML, Stone DJ, Jr., Wang Y, Lawrence D, D'Andrea MR, Brandt718

MR, Liu Y, Flores CM, and Qin N. Attenuated cold sensitivity in TRPM8 null mice.719

Neuron 54: 379-386, 2007.720


36/60

Chen S, Wang J, and Siegelbaum SA. Properties of hyperpolarization-activated724

pacemaker current defined by coassembly of HCN1 and HCN2 subunits and basal725

modulation by cyclic nucleotide.J Gen Physiol 117: 491-504, 2001.726

Chevaleyre V, and Castillo PE. Assessing the role of Ih channels in synaptic transmission727

and mossy fiber LTP.Proc Natl Acad Sci U S A 99: 9538-9543, 2002.728

Daniels RL, Takashima Y, and McKemy DD. Activity of the Neuronal Cold Sensor729

TRPM8 Is Regulated by Phospholipase C via the Phospholipid Phosphoinositol 4,5-730

Bisphosphate.Journal of Biological Chemistry 284: 1570-1582, 2009.731

Dhaka A, Murray AN, Mathur J, Earley TJ, Petrus MJ, and Patapoutian A. TRPM8732

is required for cold sensation in mice.Neuron 54: 371-378, 2007.733

DiFrancesco D. Block and activation of the pace-maker channel in calf purkinje fibres:734

effects of potassium, caesium and rubidium. The Journal of physiology 329: 485-507, 1982.735

DiFrancesco D. Serious workings of the funny current.Prog Biophys Mol Biol 90: 13-25,736

2006.737

Doan TN, Stephans K, Ramirez AN, Glazebrook PA, Andresen MC, and Kunze DL.738

Differential distribution and function of hyperpolarization-activated channels in sensory739

neurons and mechanosensitive fibers.J Neurosci 24: 3335-3343, 2004.740

Dykes RW. Coding of steady and transient temperatures by cutaneous "cold" fibers serving741

the hand of monkeys.Brain Res 98: 485-500, 1975.742

Emery EC, Young GT, Berrocoso EM, Chen L, and McNaughton PA. HCN2 Ion743


37/60

Harris NC, and Constanti A. Mechanism of block by ZD7288 of the hyperpolarization-748

activated inward rectifying current in guinea pig substantia nigra neurons in vitro. J749

Neurophysiol 74: 2366-2378, 1995.750

Hensel H. Cutaneous Thermoreceptors. In: Thermoreception and Temperature Regulation,751

edited by Hensel H. London: Academic Press, 1981, p. 33-61.752

Hensel H, and Wurster RD. Static properties of cold receptors in nasal area of cats. J753

Neurophysiol 33: 271-275, 1970.754

Herzog RI, Cummins TR, and Waxman SG. Persistent TTX-resistant Na+current affects755

resting potential and response to depolarization in simulated spinal sensory neurons. J756

Neurophysiol 86: 1351-1364, 2001.757

Hille B.Ion Channels of Excitable Membranes. Sunderland, MA, USA: Sinauer Associates758

Inc., 2001.759

Hines ML, and Carnevale NT. The NEURON simulation environment.Neural Comput 9:760

1179-1209, 1997.761

Hofmann F, Biel M, and Kaupp UB. International Union of Pharmacology. LI.762

Nomenclature and structure-function relationships of cyclic nucleotide-regulated channels.763

Pharmacol Rev 57: 455-462, 2005.764

Huber MT, Krieg JC, Dewald M, Voigt K, and Braun HA . Stochastic encoding in765

sensory neurons: impulse patterns of mammalian cold receptors. Chaos Solitons & Fractals766

11: 1895-1903, 2000.767


38/60

Kang D, Choe C, and Kim D. Thermosensitivity of the two-pore domain K+ channels772

TREK-2 and TRAAK.J Physiol 564: 103-116, 2005.773

Kaupp UB, and Seifert R. Molecular diversity of pacemaker ion channels. Annu Rev774

Physiol 63: 235-257, 2001.775

Kouranova EV, Strassle BW, Ring RH, Bowlby MR, and Vasilyev DV.776

Hyperpolarization-activated cyclic nucleotide-gated channel mRNA and protein expression777

in large versus small diameter dorsal root ganglion neurons: correlation with778

hyperpolarization-activated current gating.Neuroscience 153: 1008-1019, 2008.779

Krahe R, and Gabbiani F. Burst firing in sensory systems. Nat Rev Neurosci 5: 13-23,780

2004.781

Latorre R, Brauchi S, Madrid R, and Orio P. A cool channel in cold transduction.782

Physiology (Bethesda) 26: 273-285, 2011.783

Longtin A, and Hinzer K. Encoding with bursting, subthreshold oscillations, and noise in784

mammalian cold receptors.Neural Comput 8: 215-255, 1996.785

Ludwig A, Zong X, Jeglitsch M, Hofmann F, and Biel M. A family of hyperpolarization-786

activated mammalian cation channels.Nature 393: 587-591, 1998.787

Luo L, Chang L, Brown SM, Ao H, Lee DH, Higuera ES, Dubin AE, and Chaplan SR.788

Role of peripheral hyperpolarization-activated cyclic nucleotide-modulated channel789

pacemaker channels in acute and chronic pain models in the rat.Neuroscience 144: 1477-790

1485, 2007.791


39/60

Madrid R, Donovan-Rodriguez T, Meseguer V, Acosta MC, Belmonte C, and Viana F.794

Contribution of TRPM8 channels to cold transduction in primary sensory neurons and795

peripheral nerve terminals.J Neurosci 26: 12512-12525, 2006.796

Maingret F, Lauritzen I, Patel AJ, Heurteaux C, Reyes R, Lesage F, Lazdunski M,797

and Honore E. TREK-1 is a heat-activated background K+

channel. Embo J 19: 2483-798

2491, 2000.799

McCormick DA, and Pape HC. Properties of a hyperpolarization-activated cation current800

and its role in rhythmic oscillation in thalamic relay neurones. J Physiol 431: 291-318,801

1990.802

McKemy DD, Neuhausser WM, and Julius D. Identification of a cold receptor reveals a803

general role for TRP channels in thermosensation.Nature 416: 52-58, 2002.804

Momin A, Cadiou H, Mason A, and McNaughton PA. Role of the hyperpolarization-805

activated current Ih in somatosensory neurons.J Physiol 2008.806

Noel J, Zimmermann K, Busserolles J, Deval E, Alloui A, Diochot S, Guy N, Borsotto807

M, Reeh P, Eschalier A, and Lazdunski M. The mechano-activated K+channels TRAAK808

and TREK-1 control both warm and cold perception.EMBO J 2009.809

Nolan MF, Malleret G, Lee KH, Gibbs E, Dudman JT, Santoro B, Yin D, Thompson810

RF, Siegelbaum SA, Kandel ER, and Morozov A. The hyperpolarization-activated811

HCN1 channel is important for motor learning and neuronal integration by cerebellar812

Purkinje cells. Cell 115: 551-564, 2003.813


40/60

Pape HC. Queer current and pacemaker: the hyperpolarization-activated cation current in817

neurons.AnnuRevPhysiol 58: 299-327, 1996.818

Parra A, Madrid R, Echevarria D, del Olmo S, Morenilla-Palao C, Acosta MC, Gallar819

J, Dhaka A, Viana F, and Belmonte C. Ocular surface wetness is regulated by TRPM8-820

dependent cold thermoreceptors of the cornea.Nat Med 16: 1396-1399, 2010.821

Peier AM, Moqrich A, Hergarden AC, Reeve AJ, Andersson DA, Story GM, Earley822

TJ, Dragoni I, McIntyre P, Bevan S, and Patapoutian A. A TRP channel that senses823

cold stimuli and menthol. Cell 108: 705-715, 2002.824

Plant RE. Bifurcation and resonance in a model for bursting nerve cells.J Math Biol 11:825

15-32, 1981.826

Publio R, Oliveira RF, and Roque AC. A Computational Study on the Role of Gap827

Junctions and Rod Ih Conductance in the Enhancement of the Dynamic Range of the828

Retina.PLoS ONE 4: e6970, 2009.829

Reid G, Babes A, and Pluteanu F. A cold- and menthol-activated current in rat dorsal root830

ganglion neurones: properties and role in cold transduction.J Physiol 545: 595-614, 2002.831

Reid G, and Flonta M. Cold transduction by inhibition of a background potassium832

conductance in rat primary sensory neurones.Neurosci Lett 297: 171-174, 2001.833

Richardson MJE, Brunel N, and Hakim V. From Subthreshold to Firing-Rate834

Resonance.JNeurophysiol 89: 2538-2554, 2003.835

Robinson RB, and Siegelbaum SA. Hyperpolarization-activated cation currents: from836


41/60

Santoro B, Liu DT, Yao H, Bartsch D, Kandel ER, Siegelbaum SA, and Tibbs GR.841

Identification of a gene encoding a hyperpolarization-activated pacemaker channel of brain.842

Cell 93: 717-729, 1998.843

Scroggs RS, Todorovic SM, Anderson EG, and Fox AP. Variation in IH, IIR, and844

ILEAK between acutely isolated adult rat dorsal root ganglion neurons of different size. J845

Neurophysiol 71: 271-279, 1994.846

Schafer K, Braun HA, and Hensel H. Static and dynamic activity of cold receptors at847

various calcium levels.J Neurophysiol 47: 1017-1028, 1982.848

Schafer K, Braun HA, and Rempe L. Discharge pattern analysis suggests existence of a849

low-threshold calcium channel in cold receptors.Experientia 47: 47-50, 1991.850

Schafer K, Braun HA, and Rempe L. Mechanism of Sensory Transduction in Cold851

Receptors. In: Thermoreception and Thermoregulation, edited by Bligh J, and Voigt K.852

Berlin: Springer Verlag, 1990, p. 30-36.853

Schreiber S. Influence of Ionic Conductances on Spike Timing Reliability of Cortical854

Neurons for Suprathreshold Rhythmic Inputs.J Neurophysiol 91: 194-205, 2003.855

Shin KS, Rothberg BS, and Yellen G. Blocker state dependence and trapping in856

hyperpolarization-activated cation channels: evidence for an intracellular activation gate.857

The Journal of general physiology 117: 91-101, 2001.858

Spray DC. Cutaneous temperature receptors.Annu Rev Physiol 48: 625-638, 1986.859

Thoby-Brisson M, Telgkamp P, and Ramirez JM. The role of the hyperpolarization-860


42/60

Tu H, Deng L, Sun Q, Yao L, Han JS, and Wan Y . Hyperpolarization-activated, cyclic863

nucleotide-gated cation channels: roles in the differential electrophysiological properties of864

rat primary afferent neurons.J Neurosci Res 76: 713-722, 2004.865

Ulens C, and Tytgat J. Functional heteromerization of HCN1 and HCN2 pacemaker866

channels.JBiolChem 276: 6069-6072, 2001.867

van Brederode JF, and Berger AJ. GAD67-GFP+ neurons in the Nucleus of Roller. II.868

Subthreshold and firing resonance properties. Journal of neurophysiology 105: 249-278,869

2011.870

Viana F, de la Pea E, and Belmonte C. Specificity of cold thermotransduction is871

determined by differential ionic channel expression.NatNeurosci 5: 254-260, 2002.872

Voets T, Droogmans G, Wissenbach U, Janssens A, Flockerzi V, and Nilius B. The873

principle of temperature-dependent gating in cold- and heat-sensitive TRP channels.Nature874

430: 748-754, 2004.875

876

877

878


43/60

FIGURE CAPTIONS879

Figure 1. ZD7288 alters the firing pattern at 34C in CS nerve terminals from the mouse880

cornea without impairing the acute response to cooling.881

A. Extracellular recordings of single nerve terminal impulses from a CS nerve terminal882

from the adult mouse cornea at 34C. Sample recordings of 1.5 seconds duration are shown883

for the control condition (left) and after 10 min of exposure to 20 M ZD7288 (right). B.884

Firing frequency, ISI values, number of spikes per burst and temperature trace for the same885

CS terminal shown in A. Black and grey line boxes indicate the periods analyzed for graphs886

in Figure 1C. In the spikes per burst plot, the thick gray line indicates the moving average.887

C. Mean firing frequency, cycle time and number of spikes per burst at baseline888

temperature (34C) in the control condition and after addition of 20 M ZD7288. Bars889

represent mean SEM. The experimental conditions were paired with their respective890

controls and were compared with a paired t-test. * P


44/60

thick gray line indicates moving average. C. Mean firing frequency, cycle time and902

number of spikes per burst at 27C in the control condition and after addition of 20 M903

ZD7288. Bars represent mean SEM. The experimental conditions were paired with their904

respective controls and were compared with a paired t-test. *P


45/60

spikes per burst in experiments performed with 6 HCN1-/-

animals. In each experiment, a925

150-200 s time interval of regular spiking before (black) or after (grey) application of926

ZD7288 was analyzed. Bars represent mean SEM. * p


46/60

After t=120 and until the end of the simulation,ghwas 0 (see top trace). Firing parameters948

before and afterghreduction are, respectively: mean firing rate: 7.9 and 3.9 s-1

; cycle time:949

123 and 332 ms; spikes per burst: 1 and 1.7.950

Figure 8. Simulations of NTI activity with a model bearing a HCN2-like Ih.951

A, sample 2-seconds voltage trace (left) and ISI histograms (right) obtained from952

simulations at three temperatures. The only differences with the model shown in Fig. 7 are953

the parameters of the Ih(see text). B,Plots of different firing pattern parameters (defined in954

Methods section) versus temperature, comparing modeled data with the mean of955

experimental values obtained in corneal cold receptors from HCN1 null mice. C,ISI plot,956

number of spikes per burst and firing frequency of a 200 seconds simulation at 33C where957

the Ihmaximum conductance, gh, was progressively reduced until elimination (top trace).958

Firing parameters before and afterghreduction are, respectively: mean firing rate: 7.35 and959

3.9 s-1

; cycle time: 180 and 332 ms; spikes per burst: 1.34 and 1.7.960

Figure 9. Comparison of experimental and simulated firing patterns.961

Mean firing frequency, mean number of spikes per burst and cycle time at different962

temperatures from 24C to 34C are plotted for experimental recordings (circles) and963

simulated terminals (triangles). Note the similarity between HCN1+/+

mice and the wt964

model (filled symbols) and between HCN1-/-

mice and the slow Ih model (empty965

symbols).966


47/60

(Figure 8) is theghvalue. A, comparison of the firing rate, cycle time, spikes per burst and971

fraction of skip events at different temperatures. Note that the model with reduced gh972

behaves almost identically to the slow Ih model. B, effect of blocking Ih in the model at973

33C. Note that, as occurs with the slow Ihmodel, the reduction ofghcauses an increase of974

the cycle time without a noticeable induction of bursting, resulting in a reduction of the975

firing rate. Firing parameters before and after gh reduction are, respectively: mean firing976

rate: 7.9 and 3.9 s-1

; cycle time: 220 and 332 ms; spikes per burst: 1.55 and 1.7.977

Figure 11. Role of Ihin the membrane voltage oscillation.978

Time course of a simulated voltage trace (top row), the activation variable of the slow979

depolarizing current asd and the hyperpolarization activated current ah (continuous and980

dashed lines, respectively, second row), the first derivative of the voltage (third row) and981

the total membrane conductance (bottom row). A,wtmodel. B, slow Ihmodel. C, model982

without Ih(gh=0). Traces were simulated with the temperature set at 28C to emphasize the983

role of Ih in bursting behavior. The dashed vertical line marks the time when dV/dt=0,984

representing the start of a new oscillation. The plot of total membrane conductance985

excludes the fast, spike generating conductances gd and gr. Note the different time scale986

used in C.987

988


48/60

TABLES989

Table 1. Effect of ZD7288 on NTI shape parameters.990

+Amplitude AmplitudeMaximum

dV/dt

Minimum

dV/dt

+Half-

width

Half-

width

HCN1(+/+) 0.97 0.03 1.02 0.03 0.99 0.06 1.04 0.04 1.00 0.02 0.97 0.03

HCN1(-/-) 1.01 0.02 0.94 0.03 1.07 0.02 0.99 0.03 0.99 0.01 0.98 0.01

991

NTI parameters are expressed as the ratio between ZD7288 (20 M) and the control992

condition. In all cases, the paired differences were not statistically significant (P> 0.15).993

Data is presented as mean S.E.M.994

995


49/60

Table 2. Effect of ZD7288 on the dynamic response parameters of CS nerve endings.996

Parameter Control ZD7288 P-value

Max. Firing Rateduring response to

cold (impulses/s)

73.0 8.6 67.0 10.0 0.36

Temperature

Threshold (C)32.0 0.3 33.0 0.2 0.32

Spikes/Burst duringresponse to cold

5.3 1.4 6.6 1.1 0.44

997

Values are reported as mean SEM. n=8. P-value is the result of a paired t-test between998

data in the absence and in the presence of 20 M ZD7288.999

1000

1001


50/60

0.0

0.4

0.8

0

10

20

30

0 5 10 15 20

15

25

35

Contr

ol

ZD72

88

0

2

4

6

8

Contr

ol

ZD72

88

0.0

0.2

0.4

0.6

0.8

Contr

ol

ZD72

88

0

1

2

3

4

2

4

6

8


51/60

0.0

0.5

1.0

0

10

20

15 20 25 30 35 40 45

15

25

35

Control

ZD7288

0

5

10

15

Control

ZD7288

0.0

0.2

0.4

0.6

0.8

Contro

l

ZD7288

0

2

4

6

8

2

6

10

14


52/60

24 26 28 30 32 34

0.0

0.1

0.2

0.3

0.4

24 26 28 30 32 34

1

2

3

4

0

5

10

15

Eventcount

(sq.root)

24 26 28 30 32 34

0

4

8

12

16

meanfiringrate

(impulses/s)

fractionof

skippedevents

24 26 28 30 32 34

0

2

4

6

8

Spikesperburst

0

5

10

15

Eventcou

nt

(sq.root)

0. 0 0. 2 0. 4 0. 6 0. 8

0

5

10

15

Eventcount

(sq.root)

0

5

10

15

Eventcount

(sq.root)

0

5

10

15

Eventcou

nt

(sq.root)

0 .0 0 .2 0 .4 0. 6 0. 8

0

5

10

15

Eventcount

(sq.root)

normalized

firingrate


53/60

0.0

0.4

0.8

1.2

0

10

20

2

4

6

0 5 10 15 20 25 30

15

25

35

Contr

ol

ZD72

88

0

2

4

6

8

10

Contr

ol

ZD72

88

0.0

0.2

0.4

0.6

0.8

Contr

ol

ZD72

88

0

1

2

3

4


54/60

0.0

0.4

0.8

1.2

0

10

20

2

4

6

8

0 20 40 60

15

25

35

Contr

olIva

br

0

2

4

6

Contr

olIva

br

0.0

0.1

0.2

0.3

0.4

Contr

olIva

br

0

1

2


55/60

3 mM CsCl

0.0

0.4

0.8

1.2

0

1050

100150

2

4

6

8

10 20 30 40 50 60

15

25

35

Time (min)


56/60

-80

-40

0

V(mV)

-80

-40

0

V(mV)

-80

-40

0

V(mV)

0

50

100

even

tcount

0

50

100

150

even

tcoun

t

0.0 0.2 0.4 0.6

0

200

400

even

tcoun

t

0

4

8

12

16

Mean

freq

.

(impu

lses/s

)

0.0

0.1

0.2

0.3

0.4

cyc

letime

(s)

22 26 30 34

0

2

4

6

8

Sp

ike

sper

burs

t

22 26 30 34

0.0

0.2

0.4

0.6

0.8

fract

iono

fs

kip

0

5

10

impu

lses

/s

0.0

0.6g

h

0

2

4

Sp

/Burs

t


57/60

22 26 30 34

0.00

0.05

0.10

0 8

-80

-40

0

V(mV)

-80

-40

0

V(mV)

-80

-40

0

V(mV)

0

50

100

eventcount

0

50

100

150

e

ventcount

0.0 0.2 0.4 0.6

0

200

400

eventcount

0

4

8

12

16

Meanfreq.

(impulses/s)

0.0

0.1

0.2

0.3

0.4

0.5

cycletime(s

)

22 26 30 34

0

2

4

6

8

Spikes

perburst

fractio

nofskip

0

5

10

impu

lses/s

0.0

0.6g

h

0

2

4

Sp/Burst


58/60

0

4

8

12

16

0

2

4

6

22 26 30 34

0

100

200

300

400

500

HCN1(+/+)

HCN1(-/-)

wtmodel

slow Ihmodel

Meanfreq.

(impulses/s)

Spikesperburst

C

ycleTime(ms)

Temperature (C)


59/60

0

250

500

0

6

12

18

24 28 32 36

0

2

4

6

24 28 32 36

0.0

0.2

0.4

0.6

0.8

Cycletime(ms)

wt

Me

anrate

(imp

ulses/s)

Spikesperburst

fractionofskips

0 50 100 150 200

0.0

0.2

0.4

0.6

0.8

ISI(s)

0

5

10

impuls

es/s

0.0

0.2g

h

0

2

4

Sp/Burst


60/60

-50

0

V

(mV)

0.0

0.2

0.4

asd

0.00

0.05

0.10

ah

50 ms

wtmodel slow Ihmodel gh=0

-0.5

0.0

0.5

d

V/dt

-50

0

V

(mV)

0.0

0.2

0.4

asd

0.00

0.05

0.10

ah

50 ms

A B C

-0.5

0.0

0.5

d

V/dt

-50

0

V

(mV)

0.0

0.2

0.4

asd

-0.5

0.0

0.5

d

V/dt

100 ms

0.0

0.1

0.2

0.3

gmemb

0.0

0.1

0.2

0.3

gmemb

0.0

0.1

0.2

0.3

gmemb

Po j Neurophysiol Role of Ih in the Firing Pattern of Mammalian Cold

Documents