The HCN4 channel modulates external tufted cell burst firing

Abstract

Hyperpolarization-activated cyclic nucleotide-gated cation (HCN) channels are expressed widely in various brain regions and cardiac tissue. The HCN channel conductance, named as I_h (in neurons), plays a controlling role in neuronal excitability and rhythmic oscillatory activity in individual neurons and neuronal networks. Immunohistochemical studies in the literature show HCN channel isoforms are expressed in the olfactory bulb, but relative isoform expression levels are unknown. HCN channels are encoded by four genes (HCN1-4). The expression and function of those isoforms in the five types of mitral/tufted cells are not well known. Here, we examined HCN1-4 protein and channel conductance in olfactory bulb output neurons. HCN1, 2, 3 and 4 isoforms were expressed in the olfactory bulb, with HCN4 the most abundant. HCN channel mediated I_h conductance was present in the different classes of output neurons with highest level in external tufted cells and low to none in mitral cells. To isolate specific contributions of the isoforms to the rhythmic firing of external tufted cells, we used cell specific pharmacology infusing antagonists or isoform specific blocking antisera into single neurons. Only the HCN4 blocking antisera terminated external tufted cell burst firing, while other antisera had no significant effect on spiking. Those results indicate HCN4 in external tufted cells controls burst firing in these cells. Since burst firing of external tufted cells is a key regulator of activity in the olfactory bulb, this suggests HCN4 and the resulting I_h conductance are key circuit regulators.

NEW & NOTEWORTHY

External tufted cells in olfactory bulb are pacemaker neurons that fire rhythmic bursts. This study demonstrates that pacemaker action of external tufted cells is critically dependent on I_h from HCN4 channel providing feedforward excitation to generate long-lasting depolarizations in the mitral cells. Since mitral/tufted cells play important roles in odor discrimination and olfactory learning, we speculate that HCN4 is a key regulatory channel in driving overall network activity and function of the olfactory bulb.

Graphical Abstract

INTRODUCTION

The olfactory bulb is the first brain processing center for odor, functioning in sensory detection, identification, and discrimination of odors (1–3), regulating key feeding behaviors such as food intake and energy balance (4). Mitral/tufted cells are the primary excitatory output neurons of the olfactory bulb, interact with inhibitory interneurons to form local circuits (5–9). Odor signals are conveyed from olfactory bulb mitral/tufted cell axons out to olfactory cortex (10–13). Thus, olfactory information is dynamically tuned by local bulbar circuits and converted to a spatiotemporal pattern code for high-order olfactory processing in the olfactory cortex (14–16). Correctly timed bulbar output plays an essential role in olfaction and the factors disrupting mitral/tufted cell firing activity could readily lead to olfactory sensory dysfunction.

Mitral/tufted cells are categorized into five distinct subgroups based on anatomical location of their soma within the laminar structure of bulb (mitral cells in the deepest layer and four types of tufted cells progressively more superficial through to the external tufted cells in the outermost layer). External tufted cells are considered as the primary ‘pacemaker’ of the olfactory bulb because they intrinsically generate spontaneously, rhythmic bursts of activity (17, 18). These bursts function as an essential driver, coordinating glomerular network activity and synchronizing the output of mitral/tufted cells (17, 19–22). A key determinant of the intrinsic pacemaker activity in external tufted cells is the I_h (in neurons) current, driven by the hyperpolarization-activated cyclic nucleotide-gated cation (HCN) channels (18). HCN channels, encoded by four genes (HCN1-4), are highly region specifically expressed in brain (23–28). Immunohistochemical studies confirmed HCN channels are abundantly expressed through the olfactory bulb (29, 30), but the isoform expression levels were not compared. I_h plays a controlling role in neuronal excitability and rhythmic oscillatory activity in individual neurons and neuronal networks (31–33). The roles of individual isoforms in driving the resulting I_h conductance and external tufted cell activity are unknown.

Since HCN channels in external tufted cells may be key drivers of the function of external tufted cell pacemaker activity and resulting mitral cell firing, we hypothesize a key role of one or more HCN isoform(s) in mitral/tufted cell firing in olfactory bulb. We examined the four isoforms by protein expression and assessed HCN channel conductance through pharmacologically manipulated HCN channels using specific HCN channel blocker, ZD7288 and HCN1-4 blocking antisera. These were applied using intracellular infusion to provide single cell pharmacokinetic control of each of the subunits to examine whether HCN channels modulated output neuron spiking. We demonstrate a critical role for I_h in external tufted cell rhythmic firing, and that this I_h conductance is selectively dependent on the HCN4 channel in external tufted cells.

METHODS AND MATERIALS

Ethics statement

All experimental procedures were ethically conducted in accordance with protocols approved by the University of Maryland Institutional Animal Care and Use Committee (IACUC).

Animals

Three- to four-month-old C57BL/6J mice were used, sourced from Jackson Laboratory. Animals were housed on a 12:12 h light/ dark cycle with ad libitum access to food and water. Both male and female mice were used and no sex difference in data was observed. All data presented as the mean of both sexes combined.

Chemicals and reagents

ZD7288 and Tetrodotoxin (TTX) were purchased from TOCRIS (Bristol, UK) and other chemicals were from Sigma-Aldrich (MO, USA). Blocking antisera were obtained from Alomone Labs (Jerusalem, Israel). Suppliers of primary and secondary antibodies for western blotting were described in the preceding text.

In vitro slice preparation and whole cell recording

Horizontal olfactory bulb slices were cut as previous described (34). Briefly, the brain was removed, and 350-μm horizontal slices cut with a VT1200S Vibratome (Leica Biosystems Nussloch GmbH, Nussloch, Germany) in 4 °C oxygenated (95% O₂-5% CO₂) cutting solution containing (in mM) 204.5 sucrose, 2.5 KCl, 1.25 NaH₂PO₄, 25 NaHCO₃, 10 MgCl₂, 0.5 CaCl₂ and 10 D-glucose. The slices were then transferred to a holding chamber with oxygenated artificial cerebrospinal fluid (in mM): 125 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 25 NaHCO₃, 1.3 MgSO₄, 1.3 CaCl₂, and 10 D-glucose for at least one-hour incubation at 23 °C before recording.

Recording pipettes were pulled from thick-wall borosilicate glass with filament (O.D. 1.5 mm, I.D. 0.75 mm; Sutter Instrument, Novato, CA) in a horizontal pipette puller P-97 (Sutter Instrument, Novato, CA, USA). Patch electrodes had resistance of 4 to 7 MΩ (4-5 MΩ for mitral cells and 5-7 MΩ for external tufted cells) when filled with an internal solution containing (in mM) 120 K-gluconate, 4 MgCl₂, 5 EGTA, 10 HEPES, 3 Na₂-ATP, 0.3 Na₃-GTP, 4 Na₂-phosphocreatine and 0.1% biocytin (290 mOsm/Kg H₂O and pH was adjusted to 7.3 with KOH) for current clamp for firing rate and sag recording. K-gluconate was replaced with CsCH₃SO₃ (115 mM), and TTX (0.5 μM) was added for voltage clamp for I_h recording (pH was adjusted to 7.3 with CsOH).

Slices were transferred to the recording chamber and perfused at 3 ml/min with artificial cerebrospinal fluid at 30 °C (Bipolar Temperature Controller, VA, USA). Whole cell recordings were performed under visual guidance of a BX50WI (Olympus Corporation, Japan) fixed-stage upright microscope equipped with near-infrared differential interference contrast optics. MultiClamp 700A amplifiers, pClamp 9.2 software and Digidata 1322A interface (Molecular Devices, Axon Instruments, CA, USA) were used to acquire and analyze data. Signals were digitized at 20 kHz and analyzed offline.

In vivo single-unit recordings

Animals were anesthetized with intraperitoneal injection of 1.5 to 2 g/Kg (in total) urethane. The anesthesia level was judged by the dose at which the paw withdrawal reflex ceased and that the amplitude/rate of respiration (measured by chest transducer) was stable during the experiment. Anesthetized animals were held in a stereotaxic device and body temperature maintained at 37°C with an insulated Peltier warming platform controlled and monitored by a thermoregulation device (Bipolar Temperature Controller, Norfolk, VA, USA). A craniotomy opening was made, and the olfactory bulb exposed using a dental drill and fine dura forceps.

In vivo recordings of single-unit activity were performed (35). Recording pipettes were pulled from borosilicate glass with filament (O.D. 1.5 mm, I.D. 0.75 mm; Sutter Instrument, Novato, CA). Patch electrodes had resistance of 2 to 2.5 MΩ when filled with 154 mM NaCl. The electrode was positioned with a 90° angle to dorsal surface of the olfactory bulb and inserted gradually from lateral dorsal surface to the mitral cell layer on medial side of the bulb. The layer was determined both by advancing the predicted distance from the lateral bulb surface and by assessment of characteristic unit activity of each layer traversed. A silver wire was inserted into the occipital cortex to serve as the ground reference electrode. MultiClamp 700A/Digidata 1322A (Axon Instruments, CA, USA) were used to acquire and analyze data. Signals were low-pass filtered at 2 kHz. Single-unit spiking signals were extracted by high-pass filter (>300 Hz).

Single unit cell identify was extrapolated from a combination of physical position (layer) and characteristic spiking pattern. External tufted cells are located in the glomerular layer, positioned in the bulb region we record at a depth between 100 to 170 μm, while mitral cells at this point are situated 390 to 470 μm from olfactory surface (35). While these two populations both show bursting activity, the positional segregation allows us to ensure we are not recording mitral cells by not advancing more than 200 μm from the surface. In the glomerular layer (100-170 μm depth) there are external tufted cells and superficial tufted cells that have significant spatial overlap (external tufted cells tend to be slightly more shallow with superficial tufted cells slightly deeper, but there is significant overlap spatially). However, the spiking pattern of external tufted cells and superficial tufted cells differ sufficiently to have reasonable confidence in identity of the cell types. External tufted cells have burst firing of 2–7 action potentials (APs)/burst with higher intraburst frequency, but superficial tufted cells fire 8- 15 APs/burst with lower intraburst frequency (35). Other cells in the glomerular layer, including periglomerular cells and short axon cells do not exhibit burst spiking. Thus, we have reasonable confidence in segregating those populations on their control activity.

Odor presentation

A representative odorant, amyl acetate (Sigma-Aldrich, Cleveland, OH), was mixed with mineral oil to 10% (vol/vol) and placed in a 500-mL glass bottle. A charcoal-filtered airstream was flowed over the surface of the solution to draw volatized amyl acetate into an air handling system. Odorized air stream (4 liters/min) was delivered with 7 s of odor switched into the stream every one minute (for 15 trials for control) controlled by a MultiClamp 700A commander and PG4000A digital stimulator and solenoid switching valves (35). ZD7288 was topically applied to the craniotomy opening of olfactory bulb for in vivo recording. After 10 min, odor was delivered for another 15 trails.

Western blotting

Olfactory bulb was quickly removed, tissue lysed in RIPA lysis buffer (Cell Signaling, #9806) and homogenized with a bullet blender (Next Advance, Troy, NY, USA). After centrifuging, the supernatant was removed. Samples were prepared with 1:1 denaturing sample buffer (Bio-Rad, Hercules, CA, USA), boiled for 5 min, and run on Novex 4–12% Bis-Tris Protein Gels (Thermo Fisher, Waltham, MA, USA) in NuPage running buffer (Thermo Fisher, Waltham, MA, USA) with 150 volts for 1 h. Proteins were transferred to Novex nitrocellulose-blotting membranes (Thermo Fisher, Waltham, MA, USA). Membrane blots were incubated in TBST buffer with 5% powder milk for 1 h, and then probed in 4 °C overnight with one of the following primary antibodies in same buffer: rabbit anti-HCN1 (#APC-056, 1:200), -HCN2 (#APC-030, 1:200), -HCN3 (#APC-057, 1:200) and -HCN4 (#APC-052, 1:200) antibody (Alomone Labs, Jerusalem, Israel) and rabbit anti-β-actin (1:2500; Cell Signaling Technology, Danvers, MA, USA). Wet blots were incubated in secondary HRP-linked anti-rabbit IgG (Cell Signaling Technology, Danvers, MA, USA) in 5% milk for 1 h and visualized using Pierce Biotechnology ECL Western blotting substrate kit (Thermo Fisher, Waltham, MA, USA). Images were acquired using ChemiDoc imaging system (BioRad, Hercules, CA, USA). Quantitative analysis was performed with the ImageJ program (NIH, v.1.52p, Bethesda, MD, USA). The ratios of band density of HCN1-4 over β-actin was calculated.

Immunohistochemistry

After recording, slices with biocytin-filled cells were fixed in 4% paraformaldehyde in 0.1 M phosphate buffer at 4 °C overnight or longer. Slices were incubated with 4 μg/mL Alexa Fluor 546 streptavidin for 2 hours at room temperature and imaged. Digital microscopy images were captured using Leica TCS SP8 MP Confocal Microscope (Buffalo Grove, IL, USA).

Data analysis and statistics

In vitro and in vivo electrophysiological recording data were analyzed off-line with Clampfit 10.2. Data were expressed as means ± SEM. Statistical analysis, graphs and plotting were completed with Origin 2024b (OriginLab Corporation, Northampton, MA, USA). ANOVA were used to compare group data unless otherwise stated. When ANOVA gave a significant (P < 0.05) result, the post-hoc Tukey-Kramer test was employed to assess differences between groups.

RESULTS

HCN expression and function in olfactory bulb output neurons

To examine protein expression in olfactory bulbs from wild type mice, western blotting was performed with antibodies specific to HCN 1, 2, 3, and 4 isoforms. The ratio of HCN isoform intensity of immunoreactivity to β-actin was 23.4 ± 5.1% (HCN1), 37.9 ± 8.1% (HCN2), 16.9 ± 4.4% (HCN3) and 89.1 ± 9.3% (HCN4) (Fig. 1, n = 5 for HCN1, 2, 3 and 4). This showed that HCN4 expression in the bulb was significantly higher than other isoforms (P < 0.01). HCN2 was visually slightly higher than HCN 1 and 3, but there were no statistically significant differences between HCN1, 2 and 3 in the population data (n = 5).

Figure 1.

Western blot study showing differential HCN isoform expression in olfactory bulb. A: representative western blot image of the expression of HCN1, 2, 3, and 4 and β-actin. B: The ratio of isoforms HCN1, 2, 3 and 4 band intensity over β-actin in wild mice.

Hyperpolarization-activated cyclic nucleotide-gated cation (HCN) channels generate a conductance named I_h (in neurons). These I_h hyperpolarizing currents evoke a stereotypic depolarizing “sag” response in traces, occurring as HCN channels open with hyperpolarization generating the sag. This I_h sag is measured as the difference between the peak and steady state voltage at the end of a current injection step. We measured hyperpolarization sag in all five classes of olfactory bulb output neuron, classified by somatic position and dendrite architecture as per our previous studies (35, 36). Hyperpolarization-activated sag was largest in external tufted cells (eTC, 21.89 ± 1.83 mV, n = 8), and progressively lower in the deeper output neuron classes from superficial (sTCs, 10.96 ± 2.52 mV, n = 7), middle (mTCs, 4.59 ± 1.41 mV, n = 5), deep tufted cells (dTCs, 0.84 ± 0.99 mV, n = 5) and mitral cells (MCs, 0.19 ± 0.67 mV, n = 9; P < 0.01) (Fig. 2A, B). These data show that the peripheral output neurons, primarily the external tufted cell, exhibits the greatest I_h conductance.

Figure 2.

Functional HCN channels are differentially expressed in mitral/tufted cells (M/TCs). A: representative traces showing the large hyperpolarization (−150 pA)-activated sag (Arrow) observed in external tufted cell (eTC), but small to absent sag in mitral cell (MC) in current clamp; representative biocytin filled eTC and MC shown. B: Group data for sag amplitude in external (eTCs, n = 8), superficial (sTCs, n = 7), middle (mTCs, n = 5), deep tufted cells (dTCs, n = 5) and mitral cells (MCs, n = 9). C: Hyperpolarization (−50 mV from holding potential of −50 mV)-activated inward current I_h (Arrow) in control (Black), and ZD7288 (100 μM, Red) and ZD7288-sensitive current I_h (from subtraction of ZD7288 from control. Blue) in voltage clamp mode. Large ZD7288-sensitive current was present in external tufted cells (Left), but small to absent in mitral cells (Right, the dot line is at 0 PA). D: Group data for ZD-7288 sensitive currents for external tufted cells (eTC, n = 7) and mitral cells (MC, n = 6).

Depolarizing sag amplitude is an indirect measurement of hyperpolarization-activated inward cation current (I_h). To verify these observations are due to HCN conductance, hyperpolarizing steps were applied with and without the selective I_h blocker ZD7288 (100 μM). ZD7288-sensitive current I_h was obtained by subtraction of current amplitude in ZD7288 from control (in artificial cerebrospinal fluid). The amplitude of the ZD7288-sensitive current (hyperpolarization of −50 mV from the holding potential of −50 mV, 500 ms) in external tufted cells was large (142.08 ± 14.39 pA, n = 7, P < 0.01) while negligible in mitral cells (3.60 ± 1.85 pA, n = 5; Fig. 2C and D) in agreement with our hyperpolarization-sag measures of I_h in these cells.

The I_h conductance is strongly implicated in rhythmic oscillatory activity in individual neurons and neuronal networks including the olfactory bulb (37). External tufted cells are intrinsic ‘pacemaker’ neurons, exhibiting rhythmic bursts of activity. To confirm that the I_h conductance measured here underly these bursts, we recorded external tufted cell spiking in the presence of the I_h antagonist ZD7288. Bath application of ZD7288 decreased the burst firing rates/frequency in a dose-response manner (from 3 – 200 μM, n = 3 - 5). Each cell was tested at one concentration of ZD7288. ZD7288 block was unchanged after >30 min (the longest one hour) washout (not shown), consistent with literature of this drug as irreversible (38–40). ZD7288 gradually decreased burst firing rates with ZD7288 at 3 μM (by ~25% of baseline) and 10 μM (by ~50%) and at 30 μM (by 80~100%) (Fig. 3A, C). Bath application of high dose ZD7288 (100 μM or higher) effectively silenced intrinsic burst activity of all external tufted cells (Fig. 3B, n = 5).

Figure 3.

Blockade of HCN channels terminates spontaneous bursting of external tufted cells. A: Representative trace showing reduced external tufted cell bursting in the presence of ZD7288 (10 μM) after 5 min of bath application. B: Representative trace showing effect of high dose ZD7288 (100 μM) to terminate burst spiking. C: Dose-dependent ZD7288-induced inhibition of external tufted cell burst firing rates. ZD7288 was added with final concentrations of 3 μM to 200 μM. The perfusing system flow rate was 3 mL/min with a calibrated 30 seconds for the drug to travel from the reservoir to the recording chamber. The start time (0 min) was the time at which ZD7288 reached the chamber. Fluid exchange estimations (2 ml chamber volume) are approximately 2 min for the chamber fluid to reach final concentrations at this flow-exchange rate.

In the previous experiments we demonstrated that HCN channel blockade reduces external tufted cell activity in physiological slices. We next tested if spiking observations in vitro also occur in vivo, where there is an intact bulbar output network and cortical feedback circuitry. This includes the presence of olfactory epithelium and reciprocal modulation between mitral/tufted cells and olfactory cortex (e.g., piriform). To test if disruption to HCN channels underlies changes in external tufted cell spike timing in vivo, we performed single unit recording of external tufted cells of olfactory bulb in anesthetized mice with or without topical application of ZD7288 to the surface of the bulb.

External tufted cells exhibited rhythmic respiration locked spike activity, as expected from the extensive literature on spike recordings from whole animal olfactory bulbs (Fig. 4A – C). Odor stimulation (amyl acetate) did not significantly change the frequency of respiration and bursts. However, odor exposure increased the number of spiking per burst and resulted in an average increase of 31.38 ± 5.27% in spiking activity (mean spiking/s) from baseline (percentage increase = (Odor–Baseline)/Baseline ×100; n = 5, paired-samples t test, all P > 0.01 vs. baseline; Fig. 4 B, C), returning to baseline in ~50 seconds after odor stimulation. Topical application of ZD7288 (100 μM) to the surface of the olfactory bulb gradually decreased burst frequency and completely terminated the spiking in all external tufted cells within ~ 4 to 6 min (n=5, Fig. 4C).

Figure 4.

Blockade of HCN channels terminate external tufted cell bursting in vivo. Representative traces for single-unit recording of external tufted cell spiking (bottom) and respiration (top) at baseline (A), odor stimulation (B) and topical application of ZD7288 (C, 100 μM, 5 min after topical application). Inserts in A and B show bursts on an expanded scale.

In vitro bath or in vivo topical application ZD7288 could be acting upon other cells in the circuit resulting in potential network activity suppression of external tufted cell spiking rather than intrinsic I_h conductance changes on the individual neuron. To address this, we performed a patch-infusion experiment. ZD7288 (100 μM) was applied, using an intracellular infusion technique to yield single cell pharmacology (41–43). This approach involves backloading a layer of intracellular solution with pharmacological agent and waiting for diffusion into the cell. After obtaining whole-cell patch-clamp configuration, the cell was held for long enough time so that sufficient ZD7288 could diffuse from the pipette tip into the cell, resulting in a blockage of HCN channels only in the recorded cell, i.e. single cell pharmacological block of I_h.

Intracellular infusion of ZD7288 (100 μM) gradually hyperpolarized the membrane potential as the dose of ZD7288 increased during a diffusion equilibration into the cell and when equilibrated, and it terminated spontaneous burst firing in external tufted cells in 6 -7 min (n = 6, Fig. 5). Those results show that the I_h conductance in external tufted cells is the critical driving role for external tufted cells bursting independent of any network activity.

Figure 5.

Effect of intracellular infusion of ZD7288 terminates external tufted cell firing. ZD7288 (100 μM). External tufted cell at 1 minute after whole cell configuration baseline (top). After 6 min intracellular diffusion, ZD7288 completely terminated external tufted cell spiking (bottom).

HCN4 mediates bursting spikes in external tufted cells

Hyperpolarization-activated cyclic nucleotide-gated cation (HCN) channels are formed by four subunits integrally embedded in the cell membrane to create an ion conducting pore. These subunits can be monomeric (all the same unit) or heteromeric (mix of subunits). Our western blot data indicated that the predominant isoform in the bulb was HCN4, but HCN1-3 were also present at lower levels and could be contributing to channel formation. Thus, any or all of these isoforms may be present in external tufted cells resulting in the I_h conductance that could control bursting.

To isolate the role of each HCN isoform, we conducted intracellular infusion of isoform specific blocking antibodies (44–47), to selectively inhibit function of each isoform at a single cell pharmacology level. After obtaining whole-cell patch-clamp configuration, the cell was held for 25-30 min to identify whether antisera could block spiking. Blocking antisera could gradually diffuse from the pipette tip into the cell body, reaching a high enough intracellular concentration to visibly reduce burst spiking by quick blocking the target channels. Infusion of blocking antisera to HCN1, 2, or 3 had no significant effect upon external tufted cell membrane potential, number of spikes per burst and spiking for the full 25-30 min after initial patch (Fig. 6; n = 4 for HCN1,2 and 3). In contrast, intracellular antisera to HCN4 completely abolished external tufted cell spontaneous spiking within 6-9 min after initial patching (n = 5). The percentage of baseline firing rate (%) with HCN1, 2, 3 and 4 blocking antisera was: 100.25 ± 2.69, 96.5 ± 2.84, 99.5 ± 1.32 and 0 ± 0, respectively. This shows that the I_h conductance in external tufted cells is critically dependent upon the presence of the HCN4 subunit of the hyperpolarization-activated cyclic nucleotide-gated cation channel.

Figure 6.

Effect of intracellular delivery of HCN blocking antisera on external tufted cell firing. A: External tufted cell firing 1 minute after whole cell configuration baseline (top). After 6 min intracellular diffusion with HCN4 blocking antisera external tufted cell spiking was terminated in this cell (bottom); representative biocytin filled external tufted cell shown. B: External tufted cell firing 1 minute after whole cell configuration baseline (top). After 6 min (shown here, to longer time 30 min recorded in this cell) intracellular diffusion with HCN2 blocking antisera external tufted cell spiking was unaffected (bottom). C: Group data showing firing rates with blocking antisera represented as a percent of the pre-equilibration baseline.

DISCUSSION

Hyperpolarization-activated cyclic nucleotide-gated cation (HCN) channels which generate the I_h conductance are encoded by four genes (HCN1-4), differentially expressed in brain (23, 24). In external tufted cells, a large I_h in mice generates intrinsic rhythmic bursting from 1 to 8 Hz, acting as a “pacemaker” current. We show that subclasses of mitral/tufted output cells in olfactory bulb exhibit much smaller I_h with negligible current in mitral cells themselves. This observation in mitral cells using somatic whole cell recording is consistent with previous studies showing that mitral cells show no or a very small I_h current (48) or membrane potential sag (49, 50). One interesting study shows a broad range of sag amplitude across mitral cells (−3.4 mV to 22 mV in vivo and −14 mV to 29 mV in vitro), but a small I_h (3.5 pA in average) in soma (51). This study suggested the I_h current might be slightly higher in the distal apical dendrite, 26.5 pA in average (51), but still small compared to the large I_h in external tufted cells that we see (142 pA in average in this study). The distribution of HCN channel subunits contributing to these currents is partially characterized. Immunobiological data shows HCN2 expression across the glomerular layer, but very spare in other layers (52). This study could not tell if this HCN2 expression was mitral cell distal apical dendrites or other cell types. HCN1 immunoreactivity is not detectable in mitral cells including somata, main apical and lateral dendritic processes of mitral cells (30). HCN3 expression is low in mouse brain including main olfactory bulb mitral cells; in contrast, HCN4 is highly expressed in olfactory bulb (29). These qualitative immunohistochemical observations are consistent with our quantitative relative abundance measures from western blots. The interneurons in local circuits that are connected with mitral/tufted cells have previously been shown to have very low to no HCN expression (37, 53, 54). Since external tufted cells have an important role in individual mitral/tufted cell firing activity and network function, we explored the role specific HCN subunits play in external tufted cell function.

The I_h conductance is widely implicated in multiple neuron classes for contribution to generating “pacemaker” potentials (55, 56). The hyperpolarization that follows each action potential or burst activates I_h in different brain regions, such as in ventral tegmental area dopamine neurons (57), substantia nigra pars compacta (58), thalamus (59) and external tufted cells in olfactory bulb (17, 18) and I_f (identical to I_h with a cardiac specific terminology of I_f) in sinoatrial node cells and Purkinje fibers of the heart, (55). Activation of I_h or I_f contributes to the slow depolarization (pacemaker potential) that brings the membrane voltage to action potential threshold, generating intrinsically spontaneously individual spiking or burst. There is broad evidence indicating HCN4 as the main isoform being able to control spontaneously rhythmic activity in heart (60). The HCN4 blocker, ivabradine, is clinically approved to treat heart failure and angina pectoris (61, 62). Ivabradine does not cross the blood brain barrier due to poor penetration, so the cardiac agent has minimal CNS side-effects (63). However, the dominant functional isoform(s) in brain controlling neuronal activity and behaviors are not well documented.

HCN1-4 isoforms are differentially expressed in brain neurons (23, 24, 60). Because there are no isoform-selective HCN antagonists, pharmacological manipulation at the single isoform level has been challenging. Genetic manipulation with knockout animals targeting HCN isoforms show HCN2 is involved in protection from ischemic damage in the hippocampus (64) and HCN1 in hippocampal activity (65); HCN2 modulates in the ventral tegmental area-involved behavioral responses (66); hippocampal HCN4 promotes anxiety-like behavior (67) and HCN2 overexpression in a thalamocortical circuit mediates allodynia in mice (68). Thus, the specific subunit of HCN and resulting neural activity regulation is diverse throughout the nervous system.

Our present study showed that four isoforms are differentially expressed in olfactory bulb with overall highest expression of the HCN4 isoform. Since protein abundance measures do not necessarily reflect individual cell functional characteristics, it was important to elucidate the role of each isoform in external tufted cell firing. Using a novel intracellular pharmacology approach, where we infused isoform selective blocking antisera into individual cells, we found that HCN4 was the only isoform that could terminate external tufted cells pacemaker activity. This result showed HCN4 is the key channel component generating the I_h current, and subsequent conductance control of external tufted cell pacemaker activity.

Multiple voltage-dependent conductance cooperatively regulate external tufted cell spontaneous bursting (18). I_h plays a key role in setting membrane potential and depolarizes the cell toward the threshold of persistent Na⁺ current (I_NaP) and low-voltage-activated calcium current (I_L/T), which are required to generate the depolarizing envelope that drives burst spiking. Action potential depolarizes the membrane and induces Ca²⁺ influx via high-voltage-activated Ca²⁺ channels (I_HVA), activating large-conductance Ca²⁺-dependent K⁺ current (I_K(Ca)) that terminates the burst by hyperpolarizing the membrane. Hyperpolarization activates I_h and the cycle is regenerated. As expected, blocking I_h by ZD7288 terminates external tufted cell burst spiking. ZD7288 is widely used as a selective blocker for HCN channels. However, ZD7288 has been reported to inhibit low-voltage-activated T-type calcium channels (I_L/T) (69, 70) and transient Na_V1.4 currents (71). ZD7288 has no significant effect on the inwardly rectifying potassium currents (I_Kir, stabilizing the membrane) (40, 72); while ZD7288 strongly alters afterhyperpolarization by suppressing the HCN channels that are putatively related to K⁺ channels shaping neuronal activity (73). HCN channels are voltage-dependent channels activated at potential near or below resting potential of external tufted cells (~−50 mV). I_NaP, I_L/T and I_K(Ca) all activate at membrane potential of −50 mV or above. We recorded ZD7288-sensitive current using Cs⁺-based intracellular solution that blocked K⁺ channels and TTX in extracellular solutions to block Na⁺ channels. ZD7288-sensitive current at −50 mV (by subtraction of current in ZD7288 from control) is only ~10 pA (34), and so, ZD7288 off-target effects (from Ca²⁺ channels) at −50 mV would be very limited. We reported ZD7288-sensitive current at membrane potential of −100 mV (Fig. 2C and D), at which other channels are mostly inactivated and there were almost no off-target effects. Thus, we can reasonably conclude ZD7288 sensitive currents are principally HCN mediated I_h and not off-target secondary effects on other channels.

We recorded spontaneous burst spiking in external tufted cells with resting potential of approximately −50 mV (Fig. 3 and 5) and in vivo single-unit spiking (Fig. 4), in which HCN channels and other channels (Na⁺, Ca²⁺ and K⁺ channels) might be open. Thus, ZD7288 could exert blocking effects on all activated channels (I_h and off-target channels) at ~−50 mV that quickly hyperpolarized the cell to more negative membrane potentials to slow/terminate burst spiking, at which HCN channels would be more strongly activated with larger I_h but other channels more deactivated. Together with strong support that intracellular antisera to HCN4 (specific targeting only the HCN4 channel) completely abolished external tufted cell spontaneous spiking, we collectively conclude a driving role for HCN4 in the I_h generated bursting cycle as described above.

This study demonstrates dysfunction of HCN channels by pharmacological manipulation leads to disruption of cell firing in vitro and in vivo. The pacemaker action of external tufted cells that is critically dependent on I_h, mainly generated from the opening of HCN4 channel (demonstrated in this study) in providing feedforward excitation to generate long-lasting depolarizations (LLDs) present in the mitral cells. I_h changes affect intrinsic bursting rate of external tufted cells via reduced I_h magnitude which could regulate the magnitude of feedforward excitation from external tufted cells to mitral cells. Since mitral/tufted cells play important roles in odor discrimination and olfactory learning (74–78), we speculate that HCN4 channel expression in the bulb is a key regulatory channel in driving overall network activity and function of the bulb, and the role of specifically genetic manipulation of HCN4 and other HCN isoforms on mitral/tufted cell spiking and olfactory function will be further investigated.