HCN channels segregate stimulation‐evoked movement responses in neocortex and allow for coordinated forelimb movements in rodents

Abstract

Key points

• The present study tested whether HCN channels contribute to the organization of motor cortex and to skilled motor behaviour during a forelimb reaching task.

• Experimental reductions in HCN channel signalling increase the representation of complex multiple forelimb movements in motor cortex as assessed by intracortical microstimulation.

• Global HCN1^KO mice exhibit reduced reaching accuracy and atypical movements during a single‐pellet reaching task relative to wild‐type controls.

• Acute pharmacological inhibition of HCN channels in forelimb motor cortex decreases reaching accuracy and increases atypical movements during forelimb reaching.

Abstract

The mechanisms by which distinct movements of a forelimb are generated from the same area of motor cortex have remained elusive. Here we examined a role for HCN channels, given their ability to alter synaptic integration, in the expression of forelimb movement responses during intracortical microstimulation (ICMS) and movements of the forelimb on a skilled reaching task. We used short‐duration high‐resolution ICMS to evoke forelimb movements following pharmacological (ZD7288), experimental (electrically induced cortical seizures) or genetic approaches that we confirmed with whole‐cell patch clamp to substantially reduce I _h current. We observed significant increases in the number of multiple movement responses evoked at single sites in motor maps to all three experimental manipulations in rats or mice. Global HCN1 knockout mice were less successful and exhibited atypical movements on a skilled‐motor learning task relative to wild‐type controls. Furthermore, in reaching‐proficient rats, reaching accuracy was reduced and forelimb movements were altered during infusion of ZD7288 within motor cortex. Thus, HCN channels play a critical role in the separation of overlapping movement responses and allow for successful reaching behaviours. These data provide a novel mechanism for the encoding of multiple movement responses within shared networks of motor cortex. This mechanism supports a viewpoint of primary motor cortex as a site of dynamic integration for behavioural output.

Key points

• The present study tested whether HCN channels contribute to the organization of motor cortex and to skilled motor behaviour during a forelimb reaching task.

• Experimental reductions in HCN channel signalling increase the representation of complex multiple forelimb movements in motor cortex as assessed by intracortical microstimulation.

• Global HCN1^KO mice exhibit reduced reaching accuracy and atypical movements during a single‐pellet reaching task relative to wild‐type controls.

• Acute pharmacological inhibition of HCN channels in forelimb motor cortex decreases reaching accuracy and increases atypical movements during forelimb reaching.

Abbreviations

aCSF

artificial cerebrospinal fluid

CFA

caudal forelimb area

GABA_A receptor

type A gamma‐aminobutyric acid receptor

HCN channel

hyperpolarization‐activated cyclic nucleotide‐gated non‐selective cation channel

ICMS

intracortical microstimulation

I_h

hyperpolarization‐activated cyclic nucleotide‐gated non‐selective cation current

KO

knockout

L5PC

layer 5 pyramidal cell

PFA

paraformaldehyde

Introduction

Sherrington made the seminal observation more than 100 years ago that single areas within motor cortex are highly labile and can code for distinctly different movements, depending on immediate prior experience (Grunbaum & Sherrington, 1901; Sherrington, 1939). However, it is still unknown how these distinct movements are represented in motor cortex and how those representations are related to coordinated movements (Graziano, 2009). One approach to determine the functional architecture of motor cortex has been to apply intracortical microstimulation (ICMS) paradigms to layer 5 of motor cortex in whole‐animal preparations and characterize the evoked motor responses (Asanuma & Sakata, 1967; Young et al. 2011). Altering ICMS variables, such as intensity, elicits separate types of movements at single cortical sites (Fig. 1) (Teskey & Kolb, 2011), indicating that cellular networks in primary regions of cortex are overlapping and are complexly organized to generate behaviour.

Figure 1. Representative image of ICMS forelimb movement responses presented as an overlay on rodent motor cortex.

Under baseline conditions, the majority of sites within motor cortex respond to short‐train ICMS with single contralateral (primary) forelimb movement. ICMS responses at a small number of sites present as combinations of two or more contralateral forelimb/hindlimb movements or bilateral forelimb/hindlimb movements, which are collectively termed multiple movements. The bilateral forelimb movements include but are not exclusively mirror‐type movements. Maps are the appropriate size for 37–39 day old rats (Young et al. 2012). Dashed lines indicate multiple movement responses.

This study examines the contribution of hyperpolarization‐activated cyclic nucleotide‐gated non‐selective cation (HCN) channels (Pape, 1996) in guiding the expression of multiple behavioural representations within motor cortex. HCN channels and their associated mixed cation current I _h (Spain et al. 1987) regulate synaptic integration to enable key adaptations important in the control of chronic pain (Tu et al. 2004), spatial working memory (Nolan et al. 2004) and cerebellar motor learning (Nolan et al. 2003; Rinaldi et al. 2013). Given the high levels of HCN1 expression in layer 5 pyramidal cells (L5PCs) of motor cortex (Lörincz et al. 2002), especially in corticospinal neurons (Sheets et al. 2011), we hypothesized that HCN channels normally segregate overlapping cortical movement representations and allow for coordinated movement. To examine this possibility, we employed several experimental manipulations of HCN channels while assessing evoked movement responses in vivo using ICMS.

This study targeted the neocortical representation of forelimb movements so that the contribution of HCN channels could be assessed using measures of both neocortical physiology and skilled forelimb behaviour. We first tested whether pharmacological blockade of HCN channels using ZD7288 in neocortex affected cortical movement representations using two different anaesthetic agents that have non‐overlapping mechanisms of action. We then employed an in vivo experimental seizure protocol, which we confirmed reduced I _h in L5PCs, to assess the effects of this experience‐dependent reduction of HCN channels on cortical movement representations. ICMS‐derived movement representations were then analysed in mice with genetic deletion of the HCN1 channel subunit (HCN1^KO mice). HCN1^KO mice were also assessed on a skilled forelimb reaching task to determine whether this genetic channelopathy affected forelimb motor behaviour. Finally, rats proficient in this same reaching task were then assessed for reaching success and movement components following local infusion of ZD7288 or vehicle within motor cortex. We report that experimentally reducing I _h increases the expression of multi‐joint unilateral forelimb and bilateral limb movements (i.e. multiple movements) at single penetration sites within the forelimb area of motor cortex and results in poorer reaching success and altered coordination of the forelimb during performance of a skilled reaching task. Together our findings indicate that I _h has a central role in the expression of movement representations and the generation of successful coordinated forelimb movements.

Methods

Ethical approval

All animals were housed and handled according to the Canadian Council on Animal Care guidelines based on protocols approved by the University of Calgary Health Sciences Animal Care Committee. All efforts were made to adhere to the principles of reduction, replacement and refinement in experimental design (Russell & Burch, 1959), with every attempt made to limit the number of subjects and minimize animal suffering. Animals were given access to food (Prolab RMH 2500 lab diet, PMI Nutrition International, Brentwood, MO, USA) and water ad libitum unless otherwise stated.

Animals

Young male Hooded Long‐Evans (LE) rats (n = 88) (Charles River, Montreal, QC, Canada), adult male C57BL/6J mice (n = 13; Jackson Laboratory, Bar Harbor, ME) and non‐littermate adult homozygous C57BL/6J male mice (n = 12) lacking the p region and S6 transmembrane (pore‐S6) domain of the hyperpolarization‐activated, cyclic nucleotide‐gated K⁺ 1 (Hcn1) gene (B6.129S‐Hcn1^tm2kndl/J; Jackson Laboratory, Bar Harbor, ME, USA) (HCN1^KO mice) (Nolan et al. 2003) were used in this study. Rats used for acute effects of ZD7288 were 38–43 days old. Rats used for effects of seizures were age‐matched for ICMS and cell electrophysiology experiments and data were collected at 37–39 days of age. This time‐point corresponds to when GABA levels are at or near their peak (Carpenter et al. 1988), motor maps have emerged (Young et al. 2011) and when animals are easily amenable to the patch clamping approach (Scullion et al. 2013; Hussin et al. 2015). HCN1^KO and C57BL/6J mice underwent ICMS assessment between 5 and 7 weeks of age. Rats were housed individually whereas mice were group housed (maximum five per cage). At appropriate experimental time‐points, all animals were humanely killed by an overdose of anaesthetic followed by decapitation.

Electrode implantation

Twisted‐wire bipolar stimulating electrodes were constructed and implanted according to an established methodology (Young et al. 2011). Briefly, each rat [postnatal day (P)30–31] was anaesthetized by continuous inhalation of 2% isoflurane and surgical areas were prepared by shaving hair and applying betadine antiseptic solution. Lidocaine (2%) was administered to the tissue surrounding a midline incision. Burr holes were created to access the stereotaxic coordinates described below. An electrode was implanted with one recording electrode in motor cortex at 1.0 mm anterior and 3.5 mm lateral to Bregma and 1.3 mm ventral to skull surface and one stimulating electrode in corpus callosum at 1.0 mm anterior and 0.5 mm lateral to Bregma and –3.8 mm ventral to skull surface. Dental cement was used to secure the electrode and the skin was sutured closed. Animals were given lactated Ringer solution (subcutaneous injection, 0.5–1 ml). Their health was monitored daily and stimulation sessions commenced on the fourth day after surgery.

3 Hz seizures

Four stimulation sessions were given over two days (two sessions per day). Each session consisted of 120 s of 1 ms pulse width, 3 Hz, 1000 μA balanced biphasic current delivered via the chronically implanted callosal electrode; this stimulation results in epileptiform activity that propagates throughout the neocortex (Corcoran & Cain, 1980; Teskey & Racine, 1993). Control rats received no current through the electrode. Behavioural seizures observed during the stimulation sessions were scored using Racine's scale (Racine, 1972). Animals that exhibited a minimum of one stage‐4 seizure were included in cell electrophysiology or ICMS experiments performed 1–3 days following the last stimulation session.

Spread of ZD7288 following cortical surface application and intracortical infusion

To estimate the spread of ZD7288 within motor cortex and other structures, we applied a 1% solution of fluorescein (MW: 332.31; Sigma Aldrich, St Louis, MO, USA) as a substitute for ZD7288 (MW: 292.81). The dye was allowed to diffuse for 15, 30 or 45 min following cortical surface application (30 μl) to match the timing of ZD7288 in the ICMS experiments. The dye was allowed to diffuse for 45 min following intracortical infusion (5 μl) in order to match the timing of ZD7288 in the reaching experiments. Immediately after the determined time interval had elapsed, rats were deeply anaesthetized with sodium pentobarbital and transcardially perfused with 150 ml of ice‐cold saline followed by 150 ml of 4% paraformaldehyde (PFA). Brains were post‐fixed for 24 h in 4% PFA, transferred to 30% sucrose for 48 h for cryoprotection and then sections were cut at 50 μm. Large, stitched images were collected on an Olympus slide scanner to determine the spread of the dye.

The distribution of fluorescein dye following these two routes of administration is shown in Fig. 2. For cortical application (Fig. 2 B), fluorescein dye was present between +4.0 mm anterior and –2.0 mm posterior to Bregma and exhibited the maximum ventral penetration at the level of the caudal forelimb area where it reached the corpus callosum. All three time‐points (15, 30 and 45 min) for the cortical application resulted in similar distribution patterns for fluorescein. In the experiment designed to target the caudal forelimb area using intracortical cannulation (Fig. 2 C), the dye spread extended between +3.0 mm anterior and 0.0 mm posterior to Bregma and exhibited ventral penetration within the corpus callosum. In both types of administration, dye spread was predominantly within neocortex and covered the entire extent of motor cortex.

Figure 2. Spread of fluorescein dye across motor cortex following application at cortical surface (30 μl) or intracortical infusion (5 μl).

A, modified rat atlas from Paxinos & Watson (2006). Each section is labelled as distance from Bregma (mm). Scale bar = 2 mm. B, spread of fluorescein dye (green) following cortical surface application as performed for ZD7288 in ICMS experiments. Representative slices are from 30 min after application and were similar to the 15 and 45 min time‐points. C, spread of fluorescein dye following intracortical infusion as performed for ZD7288 in skilled reaching task experiments. Representative slices are from 45 min after infusion to match the duration of a single behavioural session. Vertical bands within fluorescence are artefacts of image acquisition.

Single pellet reaching task

All apparatus, training methods and analysis were used/performed with minor modification to previous studies (Whishaw et al. 1991; Boychuk et al. 2011; Brown & Teskey, 2014). Reach training was conducted in clear Plexiglas test boxes (height: 45 cm × width: 14 wide × depth: 35 cm). A 1 cm vertical aperture was removed from the front wall extending from 2 cm above the floor to a height of 15 cm. A 4 cm wide shelf was fixed to the outside front wall at a distance of 3 cm from the floor. The shelf contained two indentations located 2 cm in front of the front wall and aligned with the edges of the aperture. Mice or rats were placed on a restricted diet and familiarized to flavoured sucrose pellets that were used throughout this task (Bioserve, Frenchtown, NJ, USA). The pellets were placed on shelf indentations to promote reaching through the front wall aperture. During each trial, animals started at the rear of the test box and approached the front to reach through the aperture in order to obtain a pellet placed in the shelf indentations. There were two possible locations for these indentations, which were both offset from the midline of the aperture, in order to encourage use by only one limb. During pre‐training, pellets were placed in both indentations whereas only the appropriate indentation was used for each animal once a preferred limb was identified. Only one reach attempt was permitted per trial. The reach was considered successful if the animal was able to successfully grasp the pellet from the shelf and transfer it into its mouth without dropping the pellet. Following each reach attempt, the animals were shaped to return to the rear of the box for the next trial. During initial training sessions, the animals were rewarded with a pellet placed in the back of the box after each trial to facilitate shaping. As training progressed, the animals were only rewarded for successful reach attempts. In all sessions, the animals could perform as many trials as possible with the number of successful and unsuccessful reach attempts recorded. The number of reach attempts and percentage success were determined for each training session (Whishaw et al. 1991). On the final training session for mice, or the drug infusion sessions for rats, the qualitative aspects of the 10 discrete subcomponents of the reaching behaviour were also assessed using video recordings (60 frames s^–1) of these sessions (Whishaw et al. 1991).

Mice were given daily 30 min sessions of pre‐training (5–10 days based on individual performance) until they performed their first reach attempt and the forelimb used during this attempt was designated the preferred limb for subsequent training. Subsequent mouse training/testing occurred daily for 15 min for a total of 15 days. For rats, limb preference was established when five consecutive reach attempts were made using one limb during pre‐training. Starting at P23, rats were reach‐trained twice per day (15 min per session) to determine limb preference and to improve reaching performance prior to cannula implant within motor cortex. Only rats that reached a minimum success rate of 50% by P30 were chosen for implantation.

ZD7288 infusions and reaching performance

On P31 or P32, under isoflurane anaesthesia, rats were chronically implanted with a 22 G stainless steel guide cannula (Plastics One, Inc., Roanoke, VA, USA) over the caudal forelimb area (CFA) (from Bregma: 0.5 mm anterior, 2.5 mm lateral; from dura: 1.0 mm ventral) contralateral to the preferred reaching forelimb. Following surgery, rats were allowed to recover for 3 days and reaching performance was reassessed on the following days until they were of testing age (P38–43; as used in mapping experiments). Baseline performance was established by averaging the final three training/testing sessions and did not differ from pre‐surgical performance. Subsequently, either 30 μm ZD 7288 or saline was infused (5 μl at a rate of 0.5 μl min^–1) into the CFA in a counterbalanced order allowing 24 h between each infusion.

Standard ICMS procedure

ICMS was performed as previously described (Young et al. 2011) under either ketamine/xylazine or alpha‐chloralose anaesthesia. All animals were food restricted 24 h prior to surgery to ensure consistent effects of anaesthetics. Anaesthetic dosages of ketamine/xylazine were different for rats and mice, as the rat dosage was fatal to mice. Rats initially received intraperitoneal (i.p.) injections of ketamine (100 mg kg⁻¹) and xylazine (5 mg kg⁻¹). Supplemental injections of ketamine alone (25 mg kg⁻¹), or a cocktail of both ketamine (17 mg kg⁻¹) and xylazine (2 mg kg⁻¹) were delivered i.p. as required throughout the surgical procedure to maintain a constant level of anaesthesia as indicated by breathing rate, vibrissae whisking, and a foot reflex (in rats) or a tail reflex (in mice) in response to a gentle pinch. Mice initially received i.p. injections of ketamine (20 mg kg⁻¹) and xylazine (1 mg kg⁻¹). Supplemental injections of ketamine alone (5 mg kg⁻¹), or a cocktail of both ketamine (3.4 mg kg⁻¹) and xylazine (0.4 mg kg⁻¹) were delivered i.p. as required throughout the surgical procedure to maintain a constant level of anaesthesia as indicated by the same behavioural indicators as in the rats.

Rats initially received i.p. injections of alpha‐chloralose at a dose of 120 mg kg⁻¹ dissolved in 100% DMSO at a concentration of 80 mg ml⁻¹ and xylazine at a dose of 5 mg kg⁻¹ with a concentration of 20 mg ml⁻¹. Supplemental injections of a 2/3 alpha‐chloralose and 1/3 xylazine cocktail were delivered i.p. as required throughout the surgical procedure to maintain a constant level of anaesthesia as indicated by breathing rate, vibrissae whisking and a foot reflex.

Surgical areas were prepared by shaving hair and applying betadine antiseptic solution. A midline incision and craniotomy were performed to expose sensorimotor cortex and a small puncture in the cisterna magna was created using an 18‐gauge needle to reduce pressure due to oedema. Dura was carefully removed and 37.4°C silicone fluid (Factor II Inc., Lakeside, AZ, USA) was used to cover the neo‐cortical surface. In experiments where the HCN blocker ZD7288 (30 μm) or artificial cerebrospinal fluid (aCSF) were used to assess acute changes, physiological saline was used to cover the neocortical surface instead of the silicone fluid because the latter prevented drug absorption. A digital image of the exposed portion of the brain was captured using a Stemi 2000‐C stereomicroscope (Carl Zeiss, Thornwood, NY, USA), digital camera (Canon Canada Inc., Mississauga, ON, Canada) and displayed on a computer. A grid composed of 500 μm squares was overlaid on the digital image. Penetration points were chosen at the intersections of the grid lines and at a central point in the middle of each square (interpenetration distance of 353 μm), except when located over a blood vessel. Microelectrodes were made from tungsten wire or borosilicate glass capillaries filled with 3.5 m NaCl and bevelled at a 30 deg angle to yield a 3 μm tip diameter. The microelectrodes, with impedance values between 1 and 1.5 MΩ, were lowered from the brain surface to a depth of 1550 μm for rats and 800 μm for mice. Electrical stimulation was delivered via an isolated stimulator (A‐M Systems, Carlsborg, WA, USA) and consisted of 13 monophasic cathodal pulses, each 200 μs in duration, delivered at a frequency of 300 Hz, and repeated every second. Rats were maintained in a prone position, with the limb contralateral to the stimulation side being supported by placing a cotton tipped applicator below the elbow joint and elevating the forelimb. This allowed visual detection by an independent observer who was unaware of stimulation variables of all possible forelimb (digit, wrist, elbow or shoulder) movements. Sites that failed to elicit a movement with 60 μA were defined as non‐responsive. The more central map points were then determined in an effort to reduce the likelihood of the intracortical microstimulation affecting the border points of the map (Nudo et al. 1990). Level of anaesthesia was assessed by revisiting positive‐response sites to confirm that movement thresholds were ± 5 μA of original values. In experiments comparing the effects of ZD7288 or aCSF, two complete ICMS motor maps were made for each animal spaced 15 min apart to allow time for drug absorption after surface application (30 μl). The timing and concentration of ZD7288 (30 μm) were determined in a preliminary series of experiments assessing changes in movement threshold at a single positive forelimb site in sensorimotor cortex. This approach has previously been demonstrated as a sufficient delivery method for assessing ICMS changes during pharmacological manipulation (Scullion et al. 2013). ICMS was performed with two experimenters and multiple personnel performed separate experiments. One individual controlled placement of the stimulation electrode and applied ZD7288/aCSF whereas a second individual controlled stimulation intensity and recorded type of evoked movements and movement threshold. The second individual was blind to pharmacological, genetic or seizure conditions given to each animal and recorded the movement type and threshold of each movement. The lowest amount of current to evoke a forelimb movement was considered the primary movement threshold and the movement was considered the primary response. Additional contralateral forelimb, ipsilateral forelimb or non‐forelimb (e.g. tail or hindlimb) movements evoked at the same single site were considered multiple movements, whose threshold was determined by the lowest current intensity to evoke the secondary movement. Primary forelimb response and multiple forelimb responses were defined under pre‐ZD7288 conditions to standardize how responses were measured during HCN channel manipulation. In rare cases when a non‐forelimb movement emerged with HCN channel manipulation, and possessed the lowest threshold at this site, the non‐forelimb was considered multiple and the lowest threshold forelimb movement was considered primary. Canvas (version 9.0.1) imaging software (ACD Systems Inc., Miami, FL, USA) was used to calculate the areal extent of forelimb responses. Animals were humanely killed following ICMS.

Slice preparation

Animals were humanely killed and each brain was placed in ice cold (4°C) slicing solution. Slicing solution contained (in mm) 87 NaCl, 2.5 KCl, 25 NaHCO₃, 0.5 CaCl₂, 7 MgCl₂, 1.25 NaH₂PO₄, 25 glucose and 75 sucrose saturated with 95% O₂/5% CO₂. A vibrating microtome (VT1000S, Leica Microsystems, ON, Canada) was used to prepare tissue sections. Coronal slices (350 μm) containing motor cortex were prepared from male LE rats, age matched to the animals used to assess movement representations after seizures (P37–39), using the same methodology as Hussin et al. (2015).

Electrophysiological recordings

Slices in the recording chamber were superfused at a flow rate of 1–2 ml min⁻¹ with aCSF that contained (in mm): 126 NaCl, 2.5 KCl, 26 NaHCO₃, 2.5 CaCl₂, 1.5 MgCl₂, 1.25 NaH₂PO₄, 10 glucose; saturated with 95% O₂/5% CO₂ at 32–34°C. Whole‐cell somatic recordings were obtained from visualized L5PCs using a previous methodology (Scullion et al. 2013; Hussin et al. 2015). The recording solution contained (in mm) 108 potassium gluconate, 8 sodium gluconate, 2 MgCl₂, 8 KCl, 1 K‐EGTA, 4 K‐ATP, 0.3 Na‐GTP and 10 Hepes that was corrected to pH 7.2 with KOH. Recordings were accepted if series resistance was < 20 MΩ and changed < 20% throughout the experiment. Electrophysiological signals were amplified (Multiclamp 700A) low‐pass filtered at 1 kHz and digitized at 10 kHz (Digidata 1322A). Data were collected and analysed using pClamp 9.0. L5PCs were voltage clamped at –50 mV and I _h responses were evoked in voltage clamp mode by applying hyperpolarizing voltage steps (–50 to –120 mV, –10 mV increments of 2 s duration). This stimulation protocol was then repeated 15 min after application of the I _h blocker ZD7288 (30 μm). I _h responses were calculated as the amplitude of late steady‐state holding current from traces generated by digitally subtracting the pre‐ZD7288 responses from the post‐ZD7288 responses. Each hyperpolarizing voltage step was followed by a step to –50 mV in order to examine the inward I _h tail current at each potential.

Pharmacological manipulation of HCN channels

ZD7288 (Tocris, Burlington, ON, Canada) was dissolved in aCSF at a concentration of 30 μm (Harris & Constanti, 1995) and bath applied during slice patch clamp recording experiments. The same concentration was also topically applied to the cortical surface for ICMS experiments. The concentration of ZD7288 in brain parenchyma was not measured, but this approach provided a more rapid and local blockade of HCN channels relative to other experiments in this study. Preliminary studies using ICMS revealed that ZD7288 concentrations > 100 μm acutely blocked ICMS responses in a manner consistent with reports of blockade of glutamate and calcium signalling, leading us to choose a lower concentration that was effective in vitro (Chen, 2004; Inaba et al. 2006; Sánchez‐Alonso et al. 2008).

Statistical analysis

Group data are expressed as mean ± standard error of the mean (SEM). Normality was assessed (JMP 5.0; SPSS 11.0) using Shapiro–Wilk W testing before choosing relevant comparative tests. Significance was determined using appropriate Student's t test or ANOVA with Bonferroni‐corrected post hoc comparisons. Mann–Whitney U tests (one‐tailed) were used to assess differences in ordinal data (i.e. rating scale for analysis of reaching components). All differences were considered significant at P < 0.05.

Results

HCN channel activity was reduced via three different methods: pharmacological, repeated seizures and genetic knockout of HCN1. The expression of motor maps was then examined using high‐resolution ICMS, which reveals somatotopically organized evoked‐movement responses. Movements arise principally by indirect (transynaptic) activation of L5PCs (Jankowska et al. 1975; Hussin et al. 2015) and provide insight into the cellular networks responsible for skilled motor behaviour (Nudo et al. 1996). Unless stated otherwise, following the derivation of a motor map we applied the I _h blocker ZD7288 (30 μm) (Harris & Constanti, 1995) to the neocortex and re‐derived the map.

Acute ZD7288 treatment reveals multiple movement representations at many sites within motor cortex

Under ketamine/xylazine anaesthesia, pharmacologically blocking I _h increased bilateral forelimb, multi‐joint unilateral forelimb and bilateral limb movements at single penetration sites, which were collectively termed multiple movements (Fig. 1). ZD7288 significantly (F _1,10 = 35.47, P = 0.0001) increased the proportion of sites containing multiple movement responses by 821% (t ₅ = 7.75, P = 0.001) whereas vehicle treatment (aCSF) had no effect (t ₅ = 0.57, P > 0.99) (Fig. 3 A, B). In 82.2 ± 8.8% of cases, multiple movements were a combination of the original movement response, which retained the lowest movement threshold, and a new movement response as opposed to the combination of two new movement responses in rats. Map areas for primary forelimb responses were not significantly altered by ZD7288 (F _1,10 = 4.29, P = 0.065) and separate treatment groups did not significantly differ in area of primary responses (F _1,10 = 0.10, P = 0.76) (Fig. 3 A, C). In addition to the number of responsive sites for separate movements, the stimulation thresholds for separate types of movement at each site were also measured. It is important to note that mean reported values for stimulation threshold were not adjusted to reflect changes in the number of responsive sites for either primary or multiple movements. Instead, mean threshold values were determined by classifying each movement as primary or multiple and then generating separate averages for these two categories under baseline or treatment conditions. With standard ICMS parameters, when a site changes from non‐responsive to responsive it is due to a decrease of threshold from a value above the maximum stimulation intensity of 60 μA to a value below this maximum intensity. Since thresholds above maximum stimulation intensity are excluded from the data analysis, the absolute reduction in stimulation threshold between the non‐responsive state and the responsive state is not measured. As a result, the wide‐scale increase in the number of multiple movement sites represents a wide‐scale reduction in the thresholds needed to evoke these multiple movements; however, mean reported thresholds values here will not reflect this effect. Movement thresholds for primary forelimb responses were not significantly affected by application of ZD7288 (F _1,10 = 0.14, P = 0.71) and there was no significant difference in primary movement threshold between the separate treatment groups (F _1,10 = 0.017, P = 0.90) (Fig. 3 D). Movement thresholds for multiple movements were not significantly affected by application of ZD7288 (F _1,10 = 0.28, P = 0.61) and separate treatment groups did not significantly differ in movement thresholds for these multiple responses (F _1,10 = 0.55, P = 0.48).

Figure 3. Acute blockade of HCN channels with ZD7288 in motor cortex significantly increases expression of multiple movement responses in naïve rats.

A, representative images of forelimb motor maps before (top row) or after (bottom row) surface application of aCSF (left; n = 6) or ZD7288 (right; n = 6). Line = Bregma; ‘A’ = anterior–posterior axis; ‘M’ = medial–lateral axis. B, per cent forelimb motor map area exhibiting multiple movement responses was significantly increased in post‐ZD7288 animals. C, mean primary forelimb map area was not altered by acute ZD7288 treatment. D, mean primary forelimb thresholds were not altered by acute ZD7288 treatment. E, under alpha chloralose, the normalized forelimb motor map area exhibiting multiple movement responses was also significantly increased in post‐ZD7288 animals (n = 9) whereas this was not observed in animals given aCSF (n = 8). F, under alpha chloralose, mean primary forelimb map area was not altered by acute ZD7288 treatment. G, under alpha chloralose, mean primary forelimb thresholds were reduced by acute ZD7288 treatment. ^* P < 0.05.

An additional ICMS experiment was undertaken to test whether the effects of HCN channel inhibition were specific to ketamine/xylazine anaesthesia. Alpha‐chloralose was selected as an anaesthetic agent due to its non‐overlapping mechanisms of action relative to ketamine and xylazine and to avoid interactions between these channels and isoflurane (Garrett & Gan, 1998; Chen et al. 2009 a; Chen et al. 2009 b). Under alpha‐chloralose anaesthesia, pharmacological blockade of I _h with ZD7288 also significantly (F _1,30 = 13.73, P = 0.0009) increased multiple movements by 418% at single penetration sites (t ₁₅ = 5.31, P = 0.0002) (Fig. 3 E). This effect on multiple movements was not observed when aCSF was applied as vehicle treatment (t ₁₅ = 0.47, P > 0.99). Consistent with the previous ICMS experiment, ZD7288 did not affect the area of primary forelimb movements (F _1,30 = 2.69, P = 0.11) and the separate treatment groups did not differ on these area measurements (F _1,30 = 0.51, P = 0.48) (Fig. 3 F). ZD7288 did significantly (F _1,30 = 7.44, P = 0.011) reduce primary movement thresholds (t ₃₀ = 2.87, P = 0.015), whereas aCSF did not affect these thresholds (t ₃₀ = 1.04, P = 0.61), and the separate treatment groups did not differ on this parameter (F _1,30 = 0.0057, P = 0.94) (Fig. 3 G). Thus, whereas ZD7288 can cause subtle changes in movement threshold, these changes are not sufficient to affect the overall map area. Instead, the most dramatic effect of HCN channel inhibition on ICMS responses is the increase in complex multiple movements and this effect occurs with ketamine/xylazine and alpha‐chloralose anaesthesia.

Experimental seizures reduce I _h in L5PCs and increase multiple movement representations in motor cortex

Animal models of epilepsy often exhibit reduced cortical I _h (Albertson et al. 2011). We thus made use of seizures as a non‐pharmacological means of reducing the function of HCN channels in motor cortex. Experimental seizures result in many neural plastic changes (Noebels et al. 2012), and we note that it is difficult to define a role for how one seizure‐induced change affects whole‐animal behaviour. As a result, the goal of this approach was to see whether an in vivo experience, that has previously been shown to reduce I _h current, could alter multiple movement expression during ICMS. To accomplish this, we repeatedly elicited seizures and showed that I _h currents in L5PCs of motor cortex were reduced.

In control animals, L5PCs displayed robust I _h currents (Fig. 4 A–C). L5PCs from animals given seizure and control treatment did not differ in cell capacitance (t ₂₇ = 0.05, P = 0.96). After four brief electrically induced seizures over a 2 day period, whole‐cell somatic recordings of L5PCs from slices of motor cortex revealed that steady‐state ZD7288‐sensitive I _h currents were significantly (F _1,27 = 4.57, P = 0.04) reduced by 38% (–120 mV step, P = 0.0002; Fig. 4 A, B). I _h tail currents were also significantly reduced (F _7,189 = 3.42, P = 0.002) (Fig. 4 A, C). This paralleled a 306% and significant (t ₁₈ = 4.19, P = 0.0006) increase in the expression of multiple movement responses obtained from separate age‐matched rats given the same seizure induction paradigm, when compared with rats with control un‐stimulated electrodes that displayed no seizures (Fig. 5 A, B). There were no significant differences (t ₁₃ = 0.13, P = 0.90) in thresholds for multiple movement responses (Fig. 5 C). Similar to earlier reports (Van Rooyen et al. 2006), repeated experimental seizures did significantly increase motor map area for primary forelimb responses (t ₁₈ = 5.06, P < 0.0001; Fig. 5 A, D). There were no significant differences (t ₁₈ = 0.15, P = 0.88) in movement thresholds for primary forelimb responses (Fig. 5 E).

Figure 4. Repeated experimental seizures reduce I _h in L5PCs in motor cortex.

A, representative responses of L5PCs from motor cortex to a series of hyperpolarizing voltage steps for Control (left; n = 14) and Seizure (right; n = 15) conditions (V _m = –50 mV). Recordings were made before (Top) and during (Middle) bath application of ZD7288 (30 μm); digital subtraction of the post‐drug response from the pre‐drug response is plotted below to identify ZD7288‐sensitive components. B, I _h responses (steady‐state amplitude of digitally subtracted traces) were significantly reduced in L5PCs from seizure animals. C, I _h tail current amplitudes were also significantly reduced in the Seizure condition. ^* P < 0.05.

Figure 5. Repeated experimental seizures increase expression of multiple movement responses in motor cortex.

A, representative motor maps for Control (left; n = 10) and Seizure (right; n = 10) conditions. B, mean per cent of total forelimb motor map occupied by multiple movement responses was significantly increased after seizures. C, mean multiple movement thresholds were not significantly different between control and seizure conditions. D, mean primary forelimb map area was significantly increased in animals given seizures. E, mean primary forelimb thresholds were not altered by seizures. ^* P < 0.05.

Mice that lack HCN1 channels possess greater proportions of multiple movement representations in motor cortex

We then assessed motor maps in C57BL/6J and HCN1 knockout (KO) mice (HCN1^KO) (Nolan et al. 2003). As in the rat, ZD7288 application significantly increased the proportion of multiple movement responses in C57BL/6J mice by 248% (F _2,12 = 8.71, P = 0.01) (t ₁₂ = 5.22, P = 0.0006) (Fig. 6 A). In 52.7 ± 5.5% of cases, multiple movements were a combination of the original movement response, which retained the lowest movement threshold, and a new movement response as opposed to the combination of two new movement responses in C57BL/6J mice. Untreated HCN1^KO mice displayed a 231% greater proportion of multiple movements relative to untreated C57BL/6J mice (t ₂₄ = 4.00, P = 0.002) which was not significantly altered by ZD7288 treatment (t ₁₂ = 0.12, P > 0.99). There was a significant overall effect of ZD7288 on thresholds of multiple movements in these cohorts of mice (F _2,12 = 4.27, P = 0.04) (Fig. 6 B). C57BL/6J mice given ZD7288 exhibited a significant increase in thresholds for multiple movements (t ₁₂ = 4.97, P = 0.001). In contrast, there were no significant changes in thresholds of multiple movements in C57BL/6J mice given aCSF (t ₁₂ = 1.39, P = 0.57) and HCN1^KO mice given ZD7288 (t ₁₂ = 1.42, P = 0.55). Prior to ZD7288 application, C57BL/6J and HCN1^KO mice did not significantly differ in map area occupied by primary forelimb responses (F _2,12 = 1.71, P = 0.22) (Fig. 6 C). ZD7288 resulted in a significant overall effect on map area occupied by primary forelimb responses (F _1,12 = 11.71, P = 0.005). There was a trend of ZD7288 application to reduce map area of primary forelimb movements in C57BL/6J mice that did not reach statistical significance (t ₁₂ = 2.69, P = 0.058). No significant change in primary forelimb map area was detected in C57BL/6J mice given aCSF (t ₁₂ = 2.18, P = 0.15) or HCN1^KO mice given ZD7288 (t ₁₂ = 0.96, P > 0.99). Prior to ZD7288 application, C57BL/6J and HCN1^KO mice did not significantly differ in movement thresholds for primary forelimb responses (F _2,12 = 2.28, P = 0.14) (Fig. 6 D). ZD7288 provided a significant overall effect on thresholds for primary forelimb movements (F _1,12 = 8.98, P = 0.01). In C57BL/6J mice, ZD7288 caused a significant increase in movement thresholds for primary forelimb responses (t ₁₂ = 3.25, P = 0.020) whereas this was not detected in HCN1^KO mice (t ₁₂ = 1.02, P = 0.98) and did not occur in C57BL/6J mice given aCSF (t ₁₂ = 0.96, P > 0.99). These data indicate that HCN channels within motor cortex keep overlapping movement responses functionally separate, and suggest their critical involvement in synaptic integration. As changes in motor maps parallel changes in the acquisition of skilled movement (Nudo et al. 1996; Kleim et al. 1998), the role of I _h in learning and executing a skilled movement was then determined.

Figure 6. HCN1^KO mice exhibit increased expression of multiple movement responses.

ICMS maps were assessed in C57BL/6J mice pre/post‐aCSF (n = 4), C57BL/6J mice pre/post‐ZD7288 (n = 5) and HCN1^KO mice pre/post‐ZD7288 (n = 6). A, a significantly larger proportion of multiple movement responses was detected in HCN1^KO mice pre‐ZD7288 and C57BL/6J mice post‐ZD7288 relative to C57BL/6J mice either pre/post‐aCSF or pre‐ZD7288. B, mean multiple movement thresholds were significantly increased in C57BL/6J mice after ZD7288. C, mean primary forelimb map area was not altered by ZD7288. D, mean primary forelimb thresholds were significantly increased in C57BL/6J mice after ZD7288. ^* P < 0.05.

Mice that lack HCN1 channels exhibit disrupted skilled motor learning

The number of reach attempts, percentage successful reaches and reaching component errors of HCN1^KO and C57BL/6J mice were assessed with the single‐pellet reaching task. HCN1^KO and C57BL/6J mice significantly increased their number of reach attempts over 15 days of training (F _14,112 = 13.36, P < 0.0001) without significant differences between groups (F _1,8 = 1.58, P = 0.24) suggesting that both were motivated to perform the task (Fig. 7 A). HCN1^KO mice, however, exhibited significantly poorer reaching accuracy (F _1,8 = 6.88, P = 0.03) (Fig. 7 B). On the final day of training we assessed the movement components of successful reaching movements and determined that HCN1^KO mice exhibited atypical reaching movements (scored as ‘errors’) on digits to midline (P = 0.007) and elbow to midline (P = 0.002) relative to C57BL/6J mice (Fig. 7 C).

Figure 7. HCN1^KO mice exhibit deficits in reaching accuracy on a skilled reaching task.

A, C57BL/6J (n = 4) and HCN1^KO (n = 6) mice did not significantly differ in the number of reach attempts. B, C57BL/6J mice achieved significantly greater per cent reaching success than HCN1^KO mice on several training days. C, mean ranked scores for reaching components. Significantly greater error scores representing atypical motor movements were observed in HCN1^KO mice on Digits to Midline and Elbow to Midline. ^* P < 0.05.

ZD7288 infusion within motor cortex impairs skilled motor function

The motor impairments exhibited by the global HCN1^KO mice prompted us to ask whether local inhibition of HCN channels in motor cortex could affect skilled motor behaviour in wild‐type rodents. This new experiment was designed to examine the acute effects of HCN channel inhibition in rats that have already received extensive motor training in order to avoid the potential for compensatory plasticity (Nolan et al. 2003; Chen et al. 2010; Bonin et al. 2013). As a result, a new approach was developed to deliver ZD7288 to motor cortex in wake‐behaving rats.

Based on the fluorescein dye tracing, an intracortical cannula was implanted in each rat that had become proficient with the single pellet reaching task (see Methods; Fig. 2). Rats were given post‐surgical behavioural training and the final 3 days of this training were taken as a baseline assessment. In two subsequent testing days, saline vehicle or ZD7288 was administered immediately prior to one session and then the treatment groups were switched in a second session 24 h later in order to provide a counter‐balanced design. Reach attempts were not affected by either saline or ZD7288 infusion within motor cortex even though a non‐significant increase in attempts was observed in animals given ZD7288 (F _2,10 = 1.74, P = 0.24) (Fig. 8 A). Local infusion of ZD7288 resulted in a significant (F _2,10 = 15.54, P = 0.001) reduction in forelimb reaching accuracy relative to performance at baseline (t ₅ = 4.61, P = 0.02) or during saline infusion (t ₅ = 4.87, P = 0.01) (Fig. 8 B). In contrast, saline infusion did not significantly affect reaching accuracy (t ₅ = 0.96, P = 0.78). The reaching behaviour of these animals was also assessed based on individual movement components as performed in the previous experiment with HCN1^KO mice. These movement components were compared between saline and ZD7288 conditions (Fig. 8 C). ZD7288 infusion induced atypical movements in Advance (P = 0.04) and Grasp (P = 0.007) components whereas altered movement in Arpeggio did not reach statistical significance (P = 0.15).

Figure 8. Acute pharmacological inhibition of HCN channels within motor cortex reduces reaching accuracy and increases atypical movements on a skilled reaching task.

A, the number of reaching attempts was not significantly affected by intracortical saline (n = 6) or ZD7288 infusion (n = 6). A non‐significant increase in attempts with ZD7288 was noted. B, per cent reaching accuracy was reduced in animals given intracortical infusion of ZD7288 relative to baseline performance or vehicle infusion. Vehicle did not affect reaching accuracy C, mean ranked scores for reaching components. Significantly greater error scores representing atypical movements were observed during ZD7288 infusion relative to vehicle infusion on Advance and Grasp. ^* P < 0.05.

Discussion

This study made use of three different methods to reduce HCN channel function and for the first time demonstrates that HCN channels are critical for maintaining the separate encoding of movement responses at sites within motor cortex. Local pharmacological application of the I _h blocker ZD7288 resulted in a dramatic increase in the proportion of cortical sites where multiple movement responses could be elicited under two separate anaesthetics. Multiple movement responses were typically the combination of the original movement plus a new movement rather than the combination of two new movements. Likewise, experimentally induced seizures, which have been previously shown to result in reduction in reaching accuracy and coordination (Henry et al. 2008), substantially reduced I _h in somatic recordings of L5PCs in motor cortex and significantly increased the expression of multiple movements. HCN1^KO mice also exhibited abnormally high proportions of multiple movements within their motor maps, learned a skilled reaching task with less accuracy and used abnormal movements during reaching relative to wild‐type controls. Long‐Evans rats that were trained on this task also exhibited reduced reaching accuracy and used abnormal movements during reaching when given single administrations of ZD7288 within motor cortex.

Relationship between HCN channels and organization of motor cortex

High‐resolution ICMS has identified extensive overlapping of boundaries between representations encoding separate single‐joint movements as well as non‐uniform composition of individual representations due to the presence of other movement types within their borders, thereby resulting in a mosaic topography (Nudo et al. 1996). Other studies also note the presence of individual sites where multiple discrete movements are represented (Jacobs & Donoghue, 1991; Li & Waters, 1991; Donoghue et al. 1992; Schneider et al. 2002; Brecht et al. 2004; Henderson et al. 2011, 2012; Seong et al. 2014; Viaro et al. 2014). These findings collectively indicate that multiple overlapping movement responses are present within individual ICMS sites and underscore the need to identify what information is encoded by these multiple responses as well as how this response is regulated.

This new role for HCN channels provides a new mechanistic basis for how multiple output patterns can coexist within cortical networks. Experiments using ICMS with increased duration of pulse trains (i.e. long‐train ICMS) have revealed complex evoked movements involving multiple movement responses (Graziano, 2009). Many of the long‐train evoked movement responses occur as complex motor behaviours that were not thought possible without coordinated activity of pre‐motor and motor areas (Graziano, 2009). Here, experimental reductions of HCN channel activity within motor cortex resulted in the unmasking of a much larger repertoire of multiple movement responses at individual sites than what is observed under baseline conditions. This HCN channel‐dependent separation of multiple movement responses is critical for behaviour as genetic or pharmacological disruptions of these channels cause accuracy errors and a larger number of abnormally coordinated movements on a skilled forelimb reaching task.

Contribution of HCN channels to excitability of motor cortex

While several neocortical cell types possess HCN channels, L5PCs were the focus of this study because they highly express this conductance and are uniquely positioned to modulate motor output (Lörincz et al. 2002; Notomi & Shigemoto, 2004; Graziano, 2009; Sheets et al. 2011). Low levels of HCN2 channels are expressed throughout cortical structures whereas the HCN3 and HCN4 isoforms have little expression in the forebrain (Moosmang et al. 2001; Notomi & Shigemoto, 2004). HCN channels provide L5PCs with a depolarizing mixed cation current along with a shunting leak conductance but also produce a net outward current caused by the rapid inactivation of these channels in response to depolarizing inputs (Schwindt & Crill, 1997; Magee, 1998, 1999; Stuart & Spruston, 1998). HCN channels also contribute to oscillatory activity and frequency tuning in cortical pyramidal neurons as demonstrated in vitro (Ulrich, 2002) and in vivo (Stark et al. 2013). Several studies favour the view that standard ICMS parameters predominantly activate neurons by excitation of synaptic afferents onto L5PCs, a phenomenon termed ‘indirect’ or ‘trans‐synaptic’ activation, as opposed to a mechanism of direct activation (Asanuma & Rosen, 1972; Jankowska et al. 1975; Ranck, 1975; Tehovnik, 1996; Bolay & Dalkara, 1998; Tolias et al. 2005; but see Histed et al. 2009). We have previously found that standard ICMS‐evoked responses observed with in vivo or in vitro preparations are abolished by pharmacological blockade of glutamate and type A GABA (GABA_A) receptors (Hussin et al. 2015). Thus, studies which support a trans‐synaptic mechanism of ICMS activation would suggest that HCN channels’ regulation of multiple movements will depend on their ability to control synaptic integration of L5PC afferents. In line with this, Sheets et al. (2011) performed glutamate photolysis during somatic patch clamp recordings of L5PCs and detected significant ZD7288‐dependent increases in uncaging responses.

Dual patch recordings of pyramidal cells indicate that HCN channels are capable of modifying distal and somatic responses to distal inputs (Williams & Stuart, 2000; Berger et al. 2001; Harnett et al. 2015). Experimental reductions of HCN channels increase membrane input resistance at dendritic and somatic sites (Berger et al. 2001; Albertson et al. 2011; Sheets et al. 2011; Harnett et al. 2015). Blocking HCN channels can reduce the electrotonic compartmentalization across the neuronal dendrite‐to‐action potential initiation zone axis (Larkum et al. 1999; Williams & Stuart, 2000; Berger et al. 2001; Harnett et al. 2015). As a result, the multiple movement expression observed here may arise by inhibition of both proximal and distal HCN channels. A recent study in L5PCs provides evidence that HCN channels reduce excitability at distal apical sites primarily by limiting supralinear dendritic spikes while promoting excitability at proximal sites by depolarizing membrane potential towards action potential threshold (Harnett et al. 2015). As a result, we propose that our multiple movement data are most consistent with enhanced L5PC responses due to trans‐synaptic activation (Sheets et al. 2011; Hussin et al. 2015) but that this effect requires inhibition of HCN channels at distal and proximal sites in order to overcome the hyperpolarization that reduces excitability at proximal sites (Harnett et al. 2015).

There is evidence that HCN channels are expressed in axon terminals, during development, but in mature animals functional expression of these channels in terminals has only been identified in entorhinal cortex (Shah, 2014). HCN channels have recently been detected in the axon initial segment where they were shown to suppress cell firing and to be negatively regulated by serotonin 5‐HT1A receptor signalling (Ko et al. 2016). We have previously identified a pro‐excitatory effect of serotonin on ICMS responses, which was mediated by 5‐HT1A receptors, but we failed to detect a change in multiple movement expression while manipulating this receptor system (Scullion et al. 2013). It remains possible that HCN channels expressed within axon initial segments or terminals affect network excitability in motor cortex. Additionally, it is unlikely that ZD7288 is affecting multiple movement expression by blocking excitatory signalling since ICMS responses require intact glutamatergic signalling (Jacobs & Donoghue, 1991; Schneider et al. 2002; Hussin et al. 2015). Indeed, in preliminary studies here doses of > 100 μm ZD7288 blocked ICMS responses and higher doses of ZD7288 have previously been shown to block excitatory glutamate signalling as well as to reduce sodium and calcium conductance (Chen, 2004; Inaba et al. 2006; Sánchez‐Alonso et al. 2008). Additionally, studies here found no evidence for changes in multiple movement expression when HCN1^KO mice were tested with ZD7288, thereby supporting an HCN channel mechanism.

HCN channels as candidates to restrict ICMS multiple movement expression

It is important to consider whether the reduced contribution of HCN channels affects multiple movements by increasing the region of activated tissue during ICMS. Topical application of bicuculline dramatically increases the amount of tissue containing primary forelimb responsive sites (Young et al. 2012; Viaro et al. 2014). In contrast, proportions of primary forelimb movement sites here were not dramatically changed by genetic or pharmacological reduction of HCN channel activity. The increased ICMS thresholds for primary movements in C57BL/6J mice with ZD7288 could be due to several different possibilities. There may be small effects of non‐selective actions of ZD7288, or interactions between ZD7288 and anaesthesia, despite confirming that all groups received the same amount of anaesthetic in each experiment. In line with this, ZD7288 during ketamine and xylazine anaesthesia was associated with increased primary thresholds, if any effect, whereas ZD7288 during alpha‐chloralose anaesthesia was associated with reduced primary thresholds. Overall, the effects of ZD7288 on primary movement thresholds were never sufficient to overtly change the number of responsive primary sites. As a result, we conclude that HCN channels provide a more specific role in multiple movement generation rather than a general role to excitability within motor cortex.

There has been no evidence for changes in multiple movement expression during experimental manipulation of the forebrain cholinergic system, serotonin, dopamine or during application of diazepam (Conner et al. 2005; Brown et al. 2011; Young et al. 2011, 2012; Scullion et al. 2013). A few studies have identified increased multiple movement expression when the GABA_A antagonist bicuculline is applied to small regions of sensorimotor cortex that are located distant from the ICMS pipette (Jacobs & Donoghue, 1991; Schneider et al. 2002). In these cases, sites at the ICMS pipette adopt the movement responses present at the site of bicuculline application (Jacobs & Donoghue, 1991; Schneider et al. 2002). Widespread application of bicuculline results in greater overlap between regions containing whisker responses and regions containing forelimb responses (Viaro et al. 2014). The relatively lower multiple movement expression with alpha‐chloralose, in comparison to ketamine and xylazine, may be due to the former increasing GABA_A signalling (Garrett & Gan, 1998). Some of the animals used in the present study were comparatively young, leaving the possibility that multiple movement expression possess a developmental component, although this was not detected across the age ranges tested here. A previous report that assessed the effects of topical bicuculline application on ICMS responses failed to detect differences in multiple movement expression across a wider range of ages, but this topic requires further study (Young et al. 2012).

In the present study, animals given experimental seizures exhibited increased expression of primary and multiple movement responses and this is consistent with previous reports (Henderson et al. 2011, 2012). The lack of change in primary movement responses with pharmacological and genetic manipulation of HCN channels (as opposed to experimental seizures) is noteworthy. One possibility is that the manipulations differ in their effects on HCN channels. Another is that seizures result in separate patterns of cellular and molecular changes, that include but are not limited to HCN channels, and this possibility is supported by many documented cellular changes in models of experimental seizures (Noebels et al. 2012).

Additional motor pathways and considerations

It is possible that experimental seizures and genetic deletion of HCN1 resulted in increased multiple movements due to changes in HCN channels at sites remote from neocortex. For example, HCN channel modification in subcortical structures or spinal cord could also affect ICMS responses (Moosmang et al. 2001; Notomi & Shigemoto, 2004; Milligan et al. 2006). Deletion of HCN1 in brain regions other than motor cortex, such as cerebellum, may have contributed to the reaching impairments of HCN1^KO mice (Nolan et al. 2003; Rinaldi et al. 2013). Indeed, global HCN1^KO mice exhibit impairments on the rotarod task which are absent in animals with forebrain‐selective deletion of HCN1 (Nolan et al. 2003). Pharmacological inhibition was performed on rats that had already received extensive training on the reaching task. Inhibition of protein synthesis in motor cortex reduces performance on skilled reaching when the inhibition is targeted to early training days prior to performance reaching a plateau (Luft et al. 2004). It is possible that greater reaching impairments could be elicited with ZD7288 infusion if earlier phases of reach training were targeted. Intracortical infusions of ZD7288 were also performed as single‐event administrations whereas repeated or prolonged disruption of HCN channels in motor cortex may have an additive or supra‐additive effect. The cannula strategy was used for the in vivo work as a necessity of targeting the cortical surface in awake‐behaving animals whereas cannulae were avoided in the ICMS experiments because they would interfere with placement of the stimulating electrode. It is expected that the intracortical infusion approach resulted in a similar drug administration as the surface application approach. Fluorescein dye spread exhibited similar distribution patterns with these two approaches, and in both cases, dye was detected throughout motor cortex.

The experimental seizures used here are anticipated to result in many changes to cell signalling within the nervous system (Noebels et al. 2012). As a result, other possible seizure‐induced changes, in addition to the reduction of I _h in L5PCs that was observed here, may have affected ICMS map expression. It is also possible that HCN1^KO mice possess compensatory changes for the loss of I _h. HCN1^KO mice do exhibit increased HCN3 expression in addition to an increased alpha 5 subunit‐mediated tonic GABA_A conductance (Nolan et al. 2003; Chen et al. 2010). If HCN3 expression compensated for a lack of HCN1, it would be expected that ZD7288 would further increase the expression of multiple movements as HCN3 is sensitive to ZD7288 (Stieber et al. 2005). Instead, HCN1^KO mice exhibited elevated multiple movement expression that was insensitive to local ZD7288 treatment. Given that GABA_A receptor antagonism increases multiple movement expression (Jacobs & Donoghue, 1991; Schneider et al. 2002; Viaro et al. 2014) it is unlikely that a compensatory tonic GABA_A conductance is responsible for the multiple movement expression observed in these mice. Additionally, the wild‐type C57BL/6J mice in experiments here were not littermates, thereby including the possibility that other factors could have contributed to the outcomes tested. Despite these possibilities, the concurrence in multiple movement expression between the genetic manipulation, experimental seizures and application of ZD7288 directly within motor cortex supports an interpretation that HCN channels in this brain structure provide a regulatory contribution to multiple movement expression. The ability to induce motor impairments with global deletion of HCN1 and acute intracortical infusion of ZD7288 within motor cortex is also consistent with this viewpoint.

An important aspect of this HCN channel‐dependent effect in motor cortex is its specificity to multiple movement representations. Single primary ICMS movements are predominantly generated by trans‐synaptic activation of projecting L5PCs that results in signalling to spinal motor neuron pools via decussated fibres within the corticospinal tract (Brösamle & Schwab, 1997). The co‐occurring but unique movement responses observed with multiple response sites are probably mediated by a combination of motor networks and pathways. Separate types of multiple movement responses may be expressed due to activation of crossing cortico‐cortical projections (Koralek et al. 1990), cortico‐subcortical projections or via signalling through uncrossed fibres within the corticospinal tract (Thallmair et al. 1998). Cortical microcircuits involving a small number of neurons may contribute to these separate movement responses (Harnett et al. 2015; Suter & Shepherd, 2015; Yamawaki & Shepherd, 2015). Differences in these motor pathways raise the possibility that HCN channels contribute to signalling within or between L5PCs in a pathway‐specific manner that is selective for multiple movements. This signalling may aid in the types of active changes in motor output patterns that support skilled motor behaviour and can be altered in pathological conditions such as epilepsy.

Additional information

Competing interests

No conflicts of interest, financial or otherwise, are declared by the authors.

Author contributions

J.A.B. contributed to the concept of the project and the experimental design, performed data collection and analysis, performed interpretation of the data and contributed to the preparation of the manuscript. J.S.F., L.A.P. and A.C.S. performed data collection and analysis, performed interpretation of the data and contributed to the preparation of the manuscript. Q.J.P. and G.C.T. contributed to the concept of the project and the experimental design, performed interpretation of the data and contributed to the preparation of the manuscript. All data collection was undertaken in the laboratories of Q.J.P. and G.C.T.

Funding

This work was supported by the Canadian Institutes of Health Research (MOP‐130495 to C.G.T.; MOP‐191318 to Q.J.P.); Natural Sciences and Engineering Research Council (RGPIN03819‐2014 to C.G.T.; RGPIN‐05791‐2014 & RGPIN‐355919‐2009 to Q.J.P.), and an Alberta Heritage Foundation for Medical Research Scientist Award (Q.J.P.).

Translational perspective

These results are the first evidence of a gating mechanism for separate movement responses within individual sites of cortex. Here, HCN channels constrain network activity in motor cortex in a manner that restricts multiple movement responses and loss of this regulation results in skilled motor deficits and altered coordination of movements. These data fit a model of cortical organization where complex and multiple movement representations encoded into neocortical networks are dynamically regulated to form complex behaviour. Moreover, reductions in the function of I _h, in disease states such as epilepsy, may lead to behavioural dysfunction. HCN channels may represent a new therapeutic target to facilitate repair of neocortex after brain insult.