Na_V1.5 sodium channel window currents contribute to spontaneous firing in olfactory sensory neurons

Abstract

Olfactory sensory neurons (OSNs) fire spontaneously as well as in response to odor; both forms of firing are physiologically important. We studied voltage-gated Na⁺ channels in OSNs to assess their role in spontaneous activity. Whole cell patch-clamp recordings from OSNs demonstrated both tetrodotoxin-sensitive and tetrodotoxin-resistant components of Na⁺ current. RT-PCR showed mRNAs for five of the nine different Na⁺ channel α-subunits in olfactory tissue; only one was tetrodotoxin resistant, the so-called cardiac subtype Na_V1.5. Immunohistochemical analysis indicated that Na_V1.5 is present in the apical knob of OSN dendrites but not in the axon. The Na_V1.5 channels in OSNs exhibited two important features: 1) a half-inactivation potential near −100 mV, well below the resting potential, and 2) a window current centered near the resting potential. The negative half-inactivation potential renders most Na_V1.5 channels in OSNs inactivated at the resting potential, while the window current indicates that the minor fraction of noninactivated Na_V1.5 channels have a small probability of opening spontaneously at the resting potential. When the tetrodotoxin-sensitive Na⁺ channels were blocked by nanomolar tetrodotoxin at the resting potential, spontaneous firing was suppressed as expected. Furthermore, selectively blocking Na_V1.5 channels with Zn²⁺ in the absence of tetrodotoxin also suppressed spontaneous firing, indicating that Na_V1.5 channels are required for spontaneous activity despite resting inactivation. We propose that window currents produced by noninactivated Na_V1.5 channels are one source of the generator potentials that trigger spontaneous firing, while the upstroke and propagation of action potentials in OSNs are borne by the tetrodotoxin-sensitive Na⁺ channel subtypes.

neurons express a diverse set of membrane ion channels that modulate and control cell signaling. “Signaling” implies a certain precision, but ion channels are functionally independent and their activity, even when coordinated by a stimulus, is inherently stochastic, a feature that introduces noise and degrades precision. In this context, utilizing stochastic behavior to facilitate signaling would be an asset.

Olfactory sensory neurons (OSNs) transduce odors using a pathway that links G protein-coupled odor receptors (ORs; Buck and Axel 1991) to adenylyl cyclase III (AC_III), which generates intracellular cAMP. cAMP opens cyclic nucleotide-gated (CNG) channels permeable to Na⁺ and Ca²⁺, allowing intracellular Ca²⁺ to rise and gate Cl⁻ channels (Gonzalez-Silva et al. 2013; Stephan et al. 2009). Together, the CNG and Cl⁻ channels generate a receptor potential that modulates OSN firing. OSNs fire not only in response to odor but spontaneously in its absence (Frings and Lindemann 1991; Gesteland et al. 1965; Lowe and Gold 1995; Maue and Dionne 1987; Reisert and Matthews 2001). Like odor-evoked firing, spontaneous firing is important. Spontaneous activity during early development allows OSNs to connect accurately with glomerular targets in the olfactory bulb (Imai et al. 2006; Ma et al. 2014; Yu et al. 2004), while suppression of spontaneous firing in mature OSNs mediates inhibitory odor responses (Reisert 2010).

Action potentials are generated by the opening and closing of intrinsic ion channels. Their rapid upstroke is preceded by an initial depolarization that raises the local membrane potential to the firing threshold. This initial depolarization, referred to as a generator potential, is driven by inward current from diverse sources depending on conditions. The term “generator potential” is used in various contexts with reference to propagation of action potentials, repetitive firing, and infrequent action potentials, although the mechanisms in each case may differ. In this report we use “generator potential” to describe the initial depolarization that generates the random, infrequent firing of OSNs. Despite the importance of spontaneous firing in OSNs, the identity of the channels that pass the current responsible for the underlying generator potentials is unknown. Initial studies implicated the odor transduction cascade, since knocking out AC_III disrupts glomerular targeting (Col et al. 2007; Zou et al. 2007). However, neither knocking out the CNG2A subunit of CNG channels (Brunet et al. 1996; Lin et al. 2000) nor blocking CNG channels with amiloride (Nakashima et al. 2013) suppresses spontaneous firing, indicating that CNG channels cannot be responsible for this activity. Another candidate for the genesis of spontaneous firing is the cAMP-modulated, hyperpolarization-activated cation (HCN) channel (Nakashima et al. 2013). Selectively blocking HCN channels with ZD7288 suppresses spontaneous and odor-evoked firing (Nakashima et al. 2013); however, spontaneous activity was restored in the presence of blocker by depolarizing field stimulation, and both spontaneous and odor-evoked firing were restored by increasing the external [K⁺] to 25 mM. These results indicate that spontaneous firing is sensitive to membrane potential. In addition, they suggest that HCN channels, while mediating the cAMP-dependent regulation of spontaneous firing (Nakashima et al. 2013), may not be the predominant source of the underlying generator potentials since firing persists in the presence of an HCN channel blocker.

Whatever the origin of spontaneous activity, it is sensitive to AC_III stimulation. In OSNs, AC_III occurs in the cilia, where it is activated by odor, and in the soma, where it interacts with β₂ adrenergic receptors (β₂ARs) (Kawai et al. 1999; Nakashima et al. 2013). Like other G protein-coupled receptors, both β₂ARs and ORs are constitutively active (de Ligt et al. 2000; Reisert 2010), resulting in a basal cAMP level that in OSNs activates a basal conductance, principally via HCN channels. In this context, HCN channels serve as negative feedback regulators, opening with increasing probability upon hyperpolarization but passing a depolarizing current, thereby establishing an equilibrium that stabilizes the membrane potential. Cyclic nucleotides bind to HCN channels and shift the equilibrium to more positive values. The rate of spontaneous firing in OSNs correlates with the basal cAMP level, and odors that elevate intracellular cAMP increase firing while “antagonist odors” that suppress cAMP decrease firing (Reisert 2010). But if neither the CNG nor the HCN channels are the primary source of the generator potentials that elicit firing, what channels are responsible? During a study of signaling in OSNs, we found that a subtype of voltage-gated Na⁺ channel (VGSC), heretofore undetected in OSNs, appears to play a critical role in this process.

METHODS

Molecular and physiological experiments were conducted with male CD-1 mice (Charles River Laboratories, Wilmington, MA) housed in Boston University's animal care facility. Anatomical experiments were conducted on male and female mice of both CD-1 and C57BL/6 lines (stock no. 000664, The Jackson Laboratory, Bar Harbor, ME) as well as two transgenic lines.

In the first transgenic line, 129P2-Omp^tm3Mom/MomJ (stock no. 006667), referred to as OMP-GFP, the coding region as well as part of the 3′ nontranslated region of the olfactory marker protein (OMP) gene is replaced by green fluorescent protein (GFP). This mutation creates a knockout in which the mature OSNs highly express GFP. Hemizygous animals were used for the anatomical experiments, so these mice had one intact OMP allele.

In the second transgenic line (MMRRC STOCK Itpr3tm1.1Rmnc) a tau-GFP construct (tauGFP) placed under the control of the Itpr3 regulatory elements is expressed by a population of nonneuronal microvillous cells situated high in the olfactory epithelium but that do not extend an axon to the olfactory bulb and are not OSNs (Hegg et al. 2010). The animals were housed on a 14-h light cycle with access to standard chow ad libitum. All animals were used in accordance with protocols approved by the local Institutional Animal Care and Use Committees.

Cell Isolation

Four- to six-week-old male CD-1 mice were killed by CO₂ inhalation followed by decapitation. Olfactory turbinates and nasal septal tissue were quickly removed and placed in chilled mammalian extracellular saline solution (MES, in mM: 145 NaCl, 5 KCl, 20 HEPES Na⁺ salt, 1 MgCl₂, 2 CaCl₂, 1 Na⁺-pyruvate, and 5 glucose). The olfactory epithelium was removed from the turbinates, chopped into small pieces with fine scissors, and transferred to 3 ml of Ca²⁺-free dissociation solution (in mM: 145 NaCl, 5 KCl, 20 HEPES Na⁺ salt, 1 Na⁺-pyruvate, and 2 EDTA) containing 2.5 mM l-cysteine and 0.25 mg/ml papain (Calbiochem, La Jolla, CA) for 8 min. The cell suspension was gently triturated with a fire-polished Pasteur pipette and then filtered through a nylon mesh to remove undissociated pieces of epithelial tissue. Finally, 2 volumes of stop solution (in mM: 145 NaCl, 5 KCl, 20 HEPES Na⁺ salt, 1 MgCl₂, 3 CaCl₂, 1 Na⁺-pyruvate, and 5 glucose, with 5 mg/ml leupeptin) were added to halt the dissociation process. All solutions were adjusted to 7.2 pH with an osmolarity of ∼320 mosM.

Electrophysiology

For whole cell recording from dissociated cells, ∼200 μl of solution containing OSNs was placed on a glass slide, where the cells settled to the surface. Cells were visualized with a ×40 DIC water immersion objective (Zeiss, Thornwood, NY) on a Zeiss Axioskop upright microscope. Dissociated OSNs were identified by their characteristic shape. Only OSNs with intact cilia were used for recording. Electrodes were pulled from borosilicate glass (type 8250, A-M Systems, Carlsborg, WA) with a P-80/PC Flaming Brown micropipette puller (Sutter Instruments, Novato, CA), fire-polished to a final resistance of 6–8 MΩ, and filled with mammalian intracellular saline (MIS, in mM 110 K⁺-gluconate, 30 KCl, 10 HEPES free acid, 1 MgCl₂, 0.023 CaCl₂, 1 EGTA, and 10 NaCl, pH 7.2) containing 0.5 mM GTP and 2.5 mM Mg-ATP. Recordings were made with an Axopatch 200A amplifier (Molecular Devices, Sunnyvale, CA) and digitized with a DigiData 1200 series A/D converter (Molecular Devices). Data were analyzed with pCLAMP 8.0 software (Molecular Devices). Pipette series resistance, pipette capacitance, whole cell capacitance, and series resistance were electronically compensated as appropriate. For some recordings the K⁺ conductance was blocked by the addition of 25 mM tetraethylammonium chloride (TEA) to the extracellular solution or the use of a Cs-intracellular saline (in mM: 150 MeSO₄, 150 CsOH, 25 CsCl₂, 10 HEPES free acid, 1 MgSO₄, 0.4 CaCl₂, and 1 EGTA). Tetrodotoxin (TTX, 10 nM–2 μM) was used to block VGSCs, while TTX-resistant VGSCs were blocked with 1 mM Zn²⁺. All chemicals were obtained from Sigma (St. Louis, MO) unless otherwise noted. All electrophysiological data are presented as means ± SE; statistical significance was determined with a Student's t-test. For characterization, Na⁺ currents were normalized to each cell's peak current, and then activation and inactivation data were fitted with a Boltzmann function to estimate half-activation and half-inactivation potentials.

To monitor spontaneous firing, extracellular recordings were made from the cilia of OSNs in excised, intact olfactory epithelial tissue with a loose-patch recording configuration (Frings and Lindemann 1990). Olfactory cilia exposed on the apical surface of the tissue were drawn into the lumen of 5- to 10-MΩ patch pipettes filled with MES, and the extracellular currents generated by action potentials were recorded with loose-patch amplifiers. Pipettes normally drew up cilia from two to four OSNs as indicated by the different amplitudes and firing rates of the extracellular currents. Ten-second-long epochs sampled at 10 kHz were recorded at 1-min intervals for periods up to 45 min. Data shown in the figures give the number of action potentials per 10-s interval. The recording chamber, with a volume of ∼0.5 ml, was continuously perfused with MES at a rate of 0.5–1 ml/min. To assess the effects of TTX and Zn²⁺ on firing rate, 5-ml aliquots of MES containing TTX (nominally 2–50 nM) or zinc salicylate, ZnCl₂, or ZnSO₄ (1 mM each) were applied by perfusion.

Whole Tissue mRNA Isolation and RT-PCR

RNA was isolated from 100 mg of olfactory epithelial tissue taken from 6- to 8-wk-old male CD-1 mice (Charles River Laboratories). Olfactory epithelial tissue was homogenized in 1 ml of TRIzol (Invitrogen, Carlsbad, CA; catalog no. 15596-026) with a Wheaton homogenizer and then transferred to a 1.5-ml microcentrifuge tube along with 0.2 ml of chloroform (Fisher Scientific, catalog no. C298-4). The tubes were shaken and then incubated for 2–3 min at room temperature to allow for phase separation. After a 15-min centrifugation (10,000 g at 4°C), the top aqueous layer containing the RNA was transferred to a new 1.5-ml tube. Slowly, 0.5 ml of isopropyl alcohol (MP Biomedicals, catalog no. 194006) was added to precipitate the RNA and mixed by inversion. After a 10-min incubation at room temperature, the samples were centrifuged for 10 min (10,000 g at 4°C). The supernatant was decanted, 1 ml of 75% ethanol (Acros, catalog no. 61509-0020) was added to wash the RNA pellet, and the tubes were centrifuged for 5 min (7,500 g at 4°C). The RNA pellet was resuspended in 100 μl of RNase-free water and purified further with the RNeasy mini kit (Qiagen, catalog no. 74104) according to the manufacturer's instructions. Genomic DNA (gDNA) contamination was removed from the RNA with the Turbo DNA-free kit (Ambion, catalog no. AM1907). First-strand cDNA synthesis was performed on 2 μg of RNA with oligo(dT) primers and the RETROscript kit (Ambion, catalog no. AM1710). Controls for gDNA contamination were run without reverse transcriptase (RT). After heat inactivation of the RT enzyme, samples were stored at −80°C for later use.

Polymerase chain reaction (PCR) was performed on 100 ng of cDNA per 50-μl reaction with Platinum Taq DNA polymerase (Invitrogen, catalog no. 10966-034) and a Bio-Rad MyCycler thermal cycler (Bio-Rad, Hercules, CA) to detect VGSC transcripts. The primers used (Eurofins MWG Operon, Huntsville, AL) are listed in Table 1. The following thermal cycler conditions were used: for Na_V1.1, Na_V1.8, and Na_V1.9: 94°-2 min, 35 cycles (94°-1 min, 62°-1 min, 72°-1 min), 72°-10 min, 4°-hold; for Na_V1.2 through Na_V1.7: 94°-2 min, 35 cycles (94°-1 min, 68°-2 min), 68°-10 min, 4°-hold. PCR products were separated on a 1.5% Ultra-Pure agarose TBE gel (Invitrogen, catalog no. 15510-027), containing 0.2 μg/ml ethidium bromide (Sigma, catalog no. E1510), and visualized on a Gel Logic 200 imaging system (Kodak, Rochester, NY) with Kodak Molecular Imaging software. Band sizes were determined by comparison with a 1-kb Plus DNA ladder (Invitrogen, catalog no. 10787-018), and their identities were confirmed through DNA sequencing. All primers were designed to span at least one intron so the resulting bands were absolutely the result of cDNA and not gDNA amplification, but negative controls (no RT, no template) were still run at each stage of amplification. RT-PCR experiments were performed in triplicate.

Table 1.

RT-PCR primer pairs for the 9 VGSC subtypes with expected size of transcript

Subtype	Sequence 5′ to 3′	Sense	bp
NaV1.1	TAATAGATAAGCCAGCTACTGATGACAATG	S	277
	CACCAGGTTGACAATATGTTTAACTTTCAG	AS
NaV1.2	ATGGAGAAAGGCGTCCCAGCAACGTTAGCCAG	S	383
	TGGCTTACAACAGTCCCAAATCAGGCACATAT	AS
NaV1.3	GATTCTGCTTTTGAGATCAACACTACTTCCTAC	S	590
	TTCTCCTAACCCTCCTATCCCACTGAAGTCTC	AS
NaV1.4	AACAGCCAGGAGAGCTGGGTCAGCAACTCTAC	S	436
	TGTTGGAACTCCTCCTCTTTCTCCTGGTCTTC	AS
NaV1.5	TATAGTGCTGGACAGACCCCCAGACACGAC	S	767
	AGGCCAGAGTCGCTGATTCGGTGCCTCAG	AS
NaV1.6	GTGGACATTCACAGGAACGGCGACTTCCAG	S	317
	ATCCAAGTATTCCTCAGGCTGCTCCACTG	AS
NaV1.7	TGATAATAGATAAGGCAACTTCCGACGACAG	S	524
	AAAGCTCCACCAAACTCAACGTCACAATCAG	AS
NaV1.8	AATCAGAGCGAGGAGAAGACG	S	196
	CTAGTGAGCTAAGGATCGCAGA	AS
NaV1.9	AGCCCAACGAAGTGAAGAAA	S	183
	TCTCCAAGCCAGAAACCAAG	AS

RT-PCR primer pairs for the 9 voltage-gated Na⁺ channel (VGSC) subtypes are shown together with the expected size of the transcript (bp). Primers for Na_V1.1–Na_V1.7 are from Lou et al. (2005); those for Na_V1.8 and Na_V1.9 are from Zhao et al. (2010).

Immunohistochemistry

The various lines of mice of both sexes aged between 2 wk and 6 mo were anesthetized with 20% chloral hydrate (2 mg/g body wt), Nembutal (100 mg/kg), or Fatal Plus solution (50 mg/kg) injected intraperitoneally and then perfused transcardially with 0.9% saline followed by PLP fixative (75 mM lysine, 1.6% paraformaldehyde, 10 mM sodium periodate, pH 7.2–7.4).

The olfactory organs were dissected and postfixed in the same fixative for 15 min to overnight. Cryoprotection was carried out in 20% sucrose overnight. The tissue was embedded in Tissue Tek OCT (Sakura Finetek, Torrance, CA). Cryosections (12–14 μm) were mounted on Superfrost Plus slides (VWR, West Chester, PA) or Tanner Scientific charged microscope slides (Light Labs, Dallas, TX) and frozen at −80°C until further use.

For single label of wild-type mice or GFP lines, standard immunocytochemical procedures were used. Briefly, cryosections were rinsed in 0.1 M phosphate-buffered saline (PBS), blocked in blocking solution containing 1% BSA, 2–3% normal donkey serum, and 0.3% Triton X-100 in PBS for 2 h, and then incubated in the primary antisera overnight to 3 days at 4°C. After three washes in buffer, 10–20 min each, the sections were incubated in the appropriate secondary antibodies (Alexa 488, Alexa 568, 1:400, Invitrogen; DL549, DL649, Jackson ImmunoResearch, West Grove, PA) for 2 h at room temperature. In some cases, Draq5, a far-red nuclear counterstain (Abcam, Cambridge, MA, ab108410) was applied at a 1:5,000 dilution along with secondary antisera. After incubation, sections were washed three times for 20 min and coverslipped with Fluoromount-G (Fisher Biotech, Birmingham, AL). Control slides were treated either without the primary antibody or with normal rabbit serum replacing the primary antiserum. In some cases, antiserum against GFP was used to enhance the signal from the expressed GFP protein. In all cases, GFP immunoreactivity (IR) corresponded to the intrinsic GFP fluorescence. For the NaV antisera, specificity was tested by adsorption with the corresponding peptide as indicated in Table 1. These control sections showed no labeling. To test the specificity of secondary antibodies, the secondary antibody was applied to the tissue without a primary antibody; no staining was observed in these controls.

For double-label preparations of Na_V1.5 and OMP, we used a hybrid detection method utilizing tyramide for the first antiserum and indirect immunofluorescence for the second. Sections were incubated with anti-Nav1.5b antibody (Table 2), diluted 1:500 and processed with TSA A568 amplification (Invitrogen). Sections were then washed, and antigen retrieval was performed consisting of 10 mM sodium citrate pH 9 for 10 min at 80°C whereupon the sections were incubated with anti-OMP (Sigma) diluted 1:1,000. Sections were then washed and reacted with secondary antiserum, Alexa 488 donkey anti rabbit IgG (1:400; Molecular Probes, Invitrogen) for 2 h at room temperature. Sections were viewed under an epifluorescence microscope or a Fluoview laser scanning confocal microscope (Olympus, Center Valley, PA).

Table 2.

Antisera used for immunohistochemistry experiments

Antisera Against	Source	Catalog No.	NIF ID	Lot	Dilution
GFP	Abcam	AB290	AB_303395	207431	1:1,000
NaV1.5	Drs. J. Caldwell and S. R. Levinson, University of Colorado Denver	AP1010-2.2		01/29/07	1:200
NaV1.5	Millipore	AB5493	AB_177503	LV1390273	1:500
NaV1.7	Drs. J. Caldwell and S. R. Levinson, University of Colorado Denver	AP2882-3.2		02/20/03	1:200
		AP468.4.1		09/28/06	1:100
OMP	Sigma	O7889	AB_796160		1:1,000
OMP (rat)	Dr. F. Margolis, University of Maryland	Goat no. 255	AB_2315009	2/4/76	1:5,000

GFP, green fluorescent protein; OMP, olfactory marker protein.

RESULTS

Three Subtypes of VGSCs Are Expressed in OSNs

Sodium currents elicited by brief depolarizing voltage pulses were recorded from dissociated OSNs in the presence and absence of TTX, a specific and selective blocker of VGSCs. Cells were held at −150 mV between test pulses to minimize steady-state inactivation. Ten nanomolar TTX blocked 23 ± 7% (n = 9) of the maximal Na⁺ current, while 1 μM TTX blocked 72 ± 5% (Fig. 1A). Ten nanomolar TTX is sufficient to block a high percentage of TTX-sensitive (TTX-S) VGSCs (EC₅₀ = 1–10 nM) without significantly attenuating the TTX-resistant (TTX-R) (EC₅₀ ≥ 1 μM) current (Goldin 2001). In contrast, 1 μM TTX should completely block the TTX-S component but only partially block the TTX-R current.

Fig. 1.

Tetrodotoxin (TTX)-resistant (TTX-R) and TTX-sensitive (TTX-S) Na⁺ current components. A: whole cell Na⁺ currents were recorded from a dissociated mouse olfactory sensory neuron (OSN) while K⁺ currents were blocked by Cs⁺ and TEA; currents elicited by voltage steps ranging from −140 mV to 0 mV in 10-mV increments, holding potential −150 mV. Ten nanomolar TTX blocked ∼25% of the maximal Na⁺ current, while 1 μM TTX blocked ∼75%. B: activation and steady-state inactivation current data were normalized and fit with Boltzmann functions. Ten nanomolar TTX shifted the mean half-activation voltage 8 mV negative and the half-inactivation voltage 10 mV negative, as expected if the TTX-S component were selectively blocked. Shaded area indicates the presence of a window current in the absence of TTX. Control: ●, activation; ▲, inactivation; 10 nM TTX: ○, activation; △, inactivation.

Low concentrations of TTX also shifted the activation and inactivation curves of VGSCs (Fig. 1B); since TTX does not modify VGSC kinetics, this indicates that the kinetics of the TTX-S and TTX-R currents differ. To characterize the kinetics of the Na⁺ current, normalized conductance was plotted against membrane potential and the resulting activation and inactivation curves were fit with Boltzmann functions. The half-activation potential shifted from −51 ± 2 mV (control) to −59 ± 3 mV in the presence of 10 nM TTX (P < 0.05, n = 9), while the half-inactivation potential shifted from −89 ± 2 mV (control) to −99 ± 3 mV (P < 0.05, n = 9). The more negative values reflect the properties of the TTX-R channels, suggesting in particular that the steady-state half-inactivation potential of the TTX-R channels in OSNs is −100 mV or more negative. The overlap of the activation and inactivation curves indicates that the TTX-R channels in particular have a substantial window current (see below). These results show that functional TTX-S and TTX-R VGSCs are present in mouse OSNs, and suggest that each type can carry a substantial portion of the Na⁺ current that OSNs are capable of producing.

To identify which VGSC subtypes underlie the TTX-S and TTX-R currents, we screened olfactory epithelial tissue for all nine VGSC α-subunits through RT-PCR. Messenger RNAs for five VGSC α-subunits were detected: four TTX-S subtypes, Na_V1.2, Na_V1.3, Na_V1.6, and Na_V1.7, together with one TTX-R subtype, Na_V1.5 (Fig. 2). Positive controls confirmed the specificity of the primers, and the identity of the PCR fragments was confirmed by sequencing. The results suggest that the Na_V1.5 Na⁺ channel subtype accounts for the large TTX-R component of voltage-gated Na⁺ current in OSNs, while one or more of the TTX-S VGSC subtypes carry the remainder of the current.

Fig. 2.

RT-PCR. Olfactory epithelial tissue was tested for the presence of voltage-gated Na⁺ channel (VGSC) mRNAs with RT-PCR. A: 5 VGSC α-subunit mRNAs were detected: Na_V1.2, Na_V1.3, Na_V1.5, Na_V1.6, and Na_V1.7. B: RT-PCR was performed on a variety of control tissues to confirm that the primers amplified the correct gene. All products were sequenced to confirm their identity. The control tissues were brain (Na_V1.1, Na_V1.2, Na_V1.3, and Na_V1.6), skeletal muscle (Na_V1.4), heart (Na_V1.5), and trigeminal nerve (Na_V1.7, Na_V1.8, and Na_V1.9). Column labels: 1 = Na_V1.1, 2 = Na_V1.2, = Na_V1.3, etc. Expected band sizes were Na_V1.1 = 277 bp, Na_V1.2 = 383 bp, Na_V1.3 = 590 bp, Na_V1.4 = 436 bp, Na_V1.5 = 767 bp, Na_V1.6 = 317 bp, Na_V1.7 = 524 bp, Na_V1.8 = 196 bp, and Na_V1.9 = 183 bp.

Immunohistochemical labeling with subtype-specific antibodies was used to examine the distribution of Na_V1.5 and compare it with that of Na_V1.7. Confirming previous reports, we found Na_V1.7-like IR in the cell bodies and axons of OSNs (see below). In addition, we detected Na_V1.5-like IR in the dendritic endings of mature OSNs.

Na_V1.7.

Immunohistochemistry employing a subtype-specific antibody to Na_V1.7 (Table 2) revealed substantial Na_V1.7-like IR in cells of the olfactory epithelium including in the somata of OSNs (Fig. 3), as well as in a population of microvillous epithelial cells with cell bodies situated high in the olfactory epithelium (Fig. 3, A–C). In agreement with previous reports (Ahn et al. 2011; Weiss et al. 2011), the intense Na_V1.7-like IR appeared in the cell bodies, in bundles of OSN axons visible in the lamina propria beneath the epithelium (Fig. 3, D and E), and in the nerve layer and glomeruli of the olfactory bulb (Fig. 3D and as reported by Weiss et al. 2011). In the apical compartment of the olfactory epithelium, numerous short immunoreactive cells were apparent, and these all expressed GFP driven by the IP₃ receptor type 3 (IP₃R3) promoter (Fig. 3, A–C). The presence of IP₃R3-positive nonsensory, microvillous epithelial cells in the apical compartment of the olfactory epithelium is documented elsewhere (Hegg et al. 2010). This apical compartment also contains the apical dendrites and dendritic knobs of OSNs. Little or no Na_V1.7-like IR is evident in these structures (Fig. 4C). Support cells and basal cells in the olfactory epithelium did not appear to be labeled by the Na_V1.7 antibody, although the microvillous IP₃R3-GFP cells were labeled.

Fig. 3.

Na_V1.7-like immunoreactivity (IR) in the olfactory epithelium and olfactory bulb. A–C: olfactory epithelia from IP₃R3-GFP mice showing prominent microvillous cells (mv) labeled both with the antibody for Na_V1.7 (A, red in C) and green fluorescent protein (GFP; B, green in C). The somata (OSN) and axons (ax) of OSNs (A and C) are labeled by Na_V1.7 antiserum but do not display GFP. The microvillous cells that sit above the layer of OSN somata extend an apical process to the surface of the epithelium, with a basal process extending to the base of the epithelium. Note that relatively little Na_V1.7 IR is present in the apical compartment of the OSNs at and above the level of the microvillous cell somata. D: sagittal section through the olfactory epithelium (OE) showing olfactory nerve fascicles (ON) beneath the epithelium coalescing into the olfactory nerve layer (ONL) reaching the glomeruli (GL) of the olfactory bulb. Na_V1.7 label is intense in the olfactory nerve as well as in the glomeruli, which contain the olfactory nerve terminals. Pseudocolor blue counterstain is Draq5. E and F: typical Na_V1.7 IR (E) in the olfactory epithelium (OE) and nerve (ON) fascicles is absent after preabsorption of the antiserum with the cognate peptide (F).

Fig. 4.

Na_V1.5-like IR in cardiac myocytes and olfactory epithelium. A and B: Na_V1.5 IR in cardiac myocytes (A) is eliminated after preabsorption with the cognate peptide (B). C and D: similarly, punctate apical labeling of olfactory epithelium (C) is absent after preabsorption of the antiserum with peptide (D).

Na_V1.5.

With a subtype-specific antibody for Na_V1.5 (Table 2), intense Na_V1.5-like IR was apparent in cardiac muscle (Fig. 4A) and in a punctate pattern in the apical compartment of the olfactory epithelium (Fig. 4C). Immunostaining of both cardiac tissue and olfactory epithelium was abolished by preabsorption of the primary antiserum with the cognate peptide (Fig. 4, B and D).

Longitudinal sections through the olfactory epithelium show that the apical punctate Na_V1.5 immunostaining in the olfactory epithelium corresponds to the apical knobs of mature OSNs expressing OMP. This is apparent both in tissue colabeled with OMP antiserum (Fig. 5D) and in tissue from OMP-GFP mice (Fig. 5, C, E, and F). All OMP-positive profiles of OSNs exhibit bright staining for Na_V1.5. En face views of the surface of the olfactory epithelium (Fig. 5, B and D) reveal a similar colabeling of OMP-positive and Na_V1.5-positive punctate structures. Higher-magnification views from OMP-GFP tissues show that Na_V1.5 IR is strongly localized to the apical knob without extending into the olfactory cilia (Fig. 5F). In no case was Na_V1.5 localized to the IP₃R3-GFP microvillous cells of the epithelium (Fig. 5, A and B). Fine, scattered punctate Na_V1.5 IR appeared in deeper layers of the olfactory epithelium, but these small puncta were not obviously associated with any particular cellular compartment (Fig. 5E). The cellular origin of the fine, punctate labeling was unclear.

Fig. 5.

Na_V1.5-like IR in olfactory epithelium. A: olfactory epithelia from IP₃R3-GFP mice (cf. Fig. 3C) showing GFP-positive microvillous cells (mv, green) are not associated with punctate Na_V1.5 IR at the surface of the epithelium (red). In transverse sections Na_V1.5 labeling was largely confined to the dendritic knobs (dk) of OSNs at the apical surface of the olfactory epithelium. The cell bodies of the IP₃R3-GFP+ microvillous cells lie in the upper compartment of the epithelium, and their basal processes extend through the layer of OSN somata to reach the basement membrane. B: en face view of the surface of the olfactory epithelium showing Na_V1.5-labeled dendritic knobs (dk, red) amidst GFP-labeled apical processes from the IP₃R3-GFP-labeled microvillous cells (mv, green). Confocal image processed with a 2-pixel-diameter dust and scratches filter from Photoshop to reduce background graininess. C: cross section of the olfactory epithelium of an OMP-GFP mouse showing the layer of apical cilia (ac, green), Na_V1.5+ dendritic knobs (dk, red), and deeper olfactory marker protein (OMP)-GFP+ OSN somata (OSN). A deep fascicle of OMP-GFP+ axons that lack Na_V1.5 staining is apparent below the epithelium (ax). Scale is same as in A. D: en face view of the epithelium of a C57BL/6 mouse similar to B colabeled for OMP protein (green) and Na_V1.5 (red). All Na_V1.5+ endings also are positive for OMP. Magnification identical to B. E: section through the olfactory epithelium of an OMP-GFP mouse showing the localization of Na_V1.5 staining to the dendritic knobs of GFP-labeled OSNs. Scattered Na_V1.5+ puncta are evident within the OSN cell body layer of the epithelium but are not obviously associated with particular regions of the OSNs. F: high-magnification en face view of the surface of the olfactory epithelium of an OMP-GFP mouse showing localization of Na_V1.5 IR to the dendritic knob of the OSNs. GFP-labeled cilia can be seen radiating outward from the dendritic knobs, but the cilia are not obviously labeled for Na_V1.5. The few GFP-positive profiles near bottom left edge are below the level of the olfactory knobs and show the paucity of Na_V1.5 IR in even the apical portions of the dendrite.

Differential Properties of VGSC Subtypes in OSNs

Our immunohistochemical results indicate that Na_V1.5 channels are localized principally in the dendritic endings of OSNs, while the Na_V1.7 subtype is distributed mainly in OSN somata and axons. The presence of Na_V1.5-like IR in the apical knobs of OSNs suggests that this VGSC subtype might account for the TTX-R component of voltage-gated Na⁺ current in these cells. The location of label in the dendritic apical knob suggests that the channel may have a unique physiological function that does not involve the propagation of action potentials to the olfactory bulb.

To address whether the distribution of Na_V1.5 channels reflects a specialized role for this subtype, we examined its properties compared with those of the TTX-S subtypes. Measurements were made with dissociated OSNs held at −120 mV or more negative (unless otherwise indicated) to allow all Na⁺ channels to recover from inactivation.

Na_V1.5 channels in OSNs are selectively blocked by Zn²⁺.

Zinc is an effective blocker of TTX-R VGSCs, its selectivity being due to the configuration of the TTX binding site (Ravindran et al. 1991; Schild et al. 1991). Whole cell patch-clamp recordings were used to assess the effect of Zn²⁺ on Na⁺ currents in dissociated OSNs (OSNs lacking an axon). To isolate the Na⁺ current, outward K⁺ currents were attenuated by recording from cells in baths containing 25 mM TEA and using Cs⁺ intracellular saline in the whole cell pipette. One millimolar Zn²⁺ suppressed maximal peak Na⁺ currents to one-quarter of their control size, producing an average block of 71 ± 2% (n = 9) (Fig. 6, A and B); this effect was reversible (not shown). Application of 1 mM Zn²⁺ together with 200 nM TTX blocked 90 ± 1% of the peak Na⁺ current (n = 6), while 1 mM Zn²⁺ applied together with 2 μM TTX blocked 94 ± 1% of the peak Na⁺ current (n = 3). These results suggest that most of the Na⁺ current resistant to 1 mM Zn²⁺ was carried by TTX-S Na⁺ channels.

Fig. 6.

Zn²⁺-sensitive component of Na⁺ current. A: whole cell Na⁺ currents recorded from a dissociated mouse OSN while K⁺ currents were blocked by Cs⁺ and TEA. Families of currents were elicited by voltage steps ranging from −140 mV to 0 mV in 10-mV increments from an OSN held at −150 mV. Zn²⁺ (1 mM) blocked ∼70% of the Na⁺ current; the remaining current was almost completely blocked by the addition of 200 nM TTX, identifying it as the TTX-sensitive component. B: data from 9 cells: 71 ± 2% of the Na⁺ current (I_Na) was blocked by 1 mM Zn²⁺. The combination of 200 nM TTX + 1 mM Zn²⁺ blocked 90 ± 1% of the current, while 2 μM TTX + 1 mM Zn²⁺ blocked 94 ± 2% of the Na⁺ current. C: activation and steady-state inactivation current data were normalized and fit with Boltzmann functions. Zn²⁺ (1 mM) shifted the mean half-activation voltage 28 mV positive and the half-inactivation voltage 6 mV positive, as expected if the TTX-resistant component were blocked by Zn²⁺. Shaded area represents the potential window current under control conditions. Control: ●, activation; ▲, inactivation; 1 mM Zn²⁺: ○, activation; △, inactivation. D: current-clamp recordings of overshooting action potentials indicate that the noninactivated VGSCs are capable of supporting action potentials initiated at the nominal resting potential of OSNs. Action potentials were elicited by 2- to 3-ms depolarizing current pulses in OSNs held at −70 mV, a potential at which Na_V1.5 channels are largely inactivated.

Na_V1.5 channels in OSNs are largely inactivated at rest.

The steady-state inactivation characteristics of whole cell Na⁺ currents in rat and mouse OSNs show half-inactivation values ranging from −75 to −108 mV (Lagostena and Menini 2003; Rajendra et al. 1992). The data in Fig. 1 suggest that Na_V1.5 channels may account for the negative values, because blocking the TTX-S channels with 10 nM TTX produced a hyperpolarizing shift of the activation and inactivation curves. In contrast, selectively blocking Na_V1.5 channels with Zn²⁺ produced a depolarizing shift (Fig. 6C). With Na_V1.5 channels blocked by 1 mM Zn²⁺, the whole cell Na⁺ current that is carried by the TTX-S channels showed a significant depolarizing shift in both the half-activation potential (−25 ± 3 mV vs. −53 ± 2 mV, P < 0.05, n = 10) and the steady-state half-inactivation potential (−83 ± 3 mV vs. −89 ± 2 mV, P < 0.05, n = 10) compared with values measured in the same cells in the absence of Zn²⁺; these changes were reversible. The magnitudes of these shifts are much greater than what would be expected for an effect only on surface potential. The activation and steady-state inactivation properties of the TTX-S Na⁺ current were also assessed when OSNs were held at −70 mV to inactivate Na_V1.5 channels. Under this condition, the mean half-activation potential of the remaining whole cell Na⁺ current was −35.1 ± 3.5 mV (n = 7) and the mean steady-state half-inactivation potential was −58.0 ± 1.0 mV (n = 24). To confirm that in resting OSNs the noninactivated VGSCs were capable of supporting action potentials, we made current-clamp recordings from OSNs held at −70 mV; brief, depolarizing current pulses evoked overshooting action potentials in the cells (Fig. 6D).

These data suggest that olfactory Na_V1.5 channels have a steady-state half-inactivation potential near −100 mV. As a consequence, most Na_V1.5 channels should be inactivated at the nominal resting potential of OSNs near −70 mV. Nevertheless, despite inactivation of Na_V1.5 channels, resting OSNs can and do fire action potentials, leading one to question the role of Na_V1.5 channels in OSNs.

Na_V1.5 channels in OSNs have a substantial window current.

In general, the term “window current” is used to describe a small Na⁺ current that may be seen in a “window” of voltages where the activation and steady-state inactivation curves overlap. The activation and inactivation curves represent the probability densities that channels are either in the open activated state or in the closed inactivated state. Within the window, the inactivation curve indicates that most of the Na⁺ channels are inactivated; the tiny fraction of the channels that are not inactivated will be in the closed resting state. The overlapping activation curve that forms the window means that there is a small but real probability that any noninactivated, resting Na⁺ channel may open spontaneously at these voltages. Given the small fraction of noninactivated channels within the window and the small likelihood that any of these channels will open, only a few channels are likely to be open at any moment, and the number of open channels will fluctuate. It is the net current that these few open channels carry from moment to moment that defines the “window current.”

As noted above, the presence of a Na⁺ “window current” in mouse OSNs is indicated in Fig. 1B and Fig. 6C, where it is represented by the gray shaded area of overlap between the activation and inactivation curves. The window peaks near −70 mV, around the nominal resting potential of OSNs. With the application of 10 nM TTX to block the TTX-S VGSC subtypes, the peak of the window was reduced only slightly and its center was shifted from −70 mV to −75 mV (Fig. 1B). By contrast, when Zn²⁺ was applied to block Na_V1.5 channels the center of the window shifted from −70 mV to −55 mV and the peak was reduced by ∼75% (Fig. 6C). These results suggest that Na_V1.5 channels are principally responsible for the window current in OSNs.

Spontaneous Activity

At rest, intact OSNs show considerable spontaneous activity, typically firing at rates of several hertz. Because the Na_V1.5 channels are largely inactivated in resting OSNs, the upstroke and propagation of spontaneous action potentials appear to be sustained primarily by the TTX-S VGSC subtypes. We monitored spontaneous firing from mature OSNs on intact, freshly excised olfactory turbinates by recording extracellular currents from olfactory cilia in vitro with a loose-patch amplifier. The tissue was bathed in continuously flowing saline to which specific inhibitors could be added without disrupting the flow.

We first verified that blocking olfactory phosphodiesterase with the membrane-permeant 3-isobutyl-1-methylxanthine (IBMX) increased the spontaneous firing rate as reported by Frings and Lindemann (1991). In OSNs, intracellular cAMP is kept in check by a phosphodiesterase (PDE1C) that hydrolyzes cAMP to AMP. IBMX (250 μM) caused a rapid increase in the rate of spontaneous firing in 12 of 14 cells (Fig. 7A); the average increase was 340 ± 110%, excluding one outlier with a 1,500% increase. We also established that inhibiting olfactory AC_III with membrane-permeant SQ 22,536 reduced the spontaneous firing rate. In 12 of 12 cells, the spontaneous firing rate was suppressed by 79 ± 3% by 250 μM SQ 22,536 (Fig. 7B). Both the IBMX and SQ 22,536 results are to be expected because of the constitutive activity of ORs (Reisert 2010) and β₂ARs (de Ligt et al. 2000). Constitutive activity produces a low level of cAMP that consequently activates cAMP-dependent cation channels that depolarize OSNs slightly; greater depolarization as with IBMX increases spontaneous firing, while reversing the depolarization, as shown by Yu et al. (2004) or here with SQ 22,536, reduces the spontaneous firing rate.

Fig. 7.

Manipulating intracellular cAMP alters spontaneous firing. A: application of 250 μM IBMX, a potent membrane-permeant phosphodiesterase inhibitor, increased the rate of spontaneous firing in 1 cell from a baseline of 3–4 action potentials (APs) per 10 s to ∼30 per 10 s after IBMX was bath applied. The effect washed out. B: application of 250 μM SQ 22,536, a potent membrane-permeant inhibitor of adenylyl cyclase III (AC_III), completely suppressed firing in a cell from a baseline rate of ∼20 APs per 10 s. After washout there was a partial but transient recovery.

Blocking TTX-S VGSCs in OSNs suppresses spontaneous firing.

Spontaneous firing was suppressed by the addition of nanomolar concentrations of TTX to block the TTX-S VGSC subtypes (Fig. 8, A and C). In 19 of 21 cells, 2–50 nM TTX suppressed spontaneous firing; suppression was partial at the lowest concentrations and complete at concentrations above ∼20 nM. Of 11 cells tested with ∼10 nM TTX, the average spontaneous firing rate was reduced to a value of 18 ± 3% of baseline firing. In light of the observation that the TTX-S channels can sustain action potentials in OSNs even when Na_V1.5 channels are inactivated, these data suggest that the TTX-S channels are responsible for the upstroke and propagation of action potentials in OSNs.

Fig. 8.

Suppression of spontaneous firing. Spontaneous firing was recorded from the cilia of OSNs in pieces of olfactory epithelium with a loose-patch amplifier. The cells were exposed to either TTX or Zn²⁺ in a constantly flowing bath. A and C: nanomolar concentrations of TTX suppressed spontaneous firing. Exemplar recordings from 6 OSNs whose cilia were aspirated together (3 of the 6 units can be distinguished in top trace) are shown in A. The 2 traces (each 2.5 s long) illustrate that 10 nM TTX markedly suppressed spontaneous firing. Top trace was recorded in normal saline and bottom trace after exposure to 10 nM TTX at minute 28 of the experiment; the time course is shown in C. Three concentrations of TTX (5-ml aliquots each) were applied in succession: ∼2, ∼10, and ∼20 nM. The firing rate began to subside soon after beginning perfusion with 2 nM TTX; it subsided further with the application of 10 and then 20 nM TTX. The mean firing rate in 20 nM TTX was 0.6/s compared with 6.6/s before application of TTX. Each point in the time course is the total number of APs fired by the 6 OSNs in a 10-s interval. B and D: Zn²⁺ (1 mM) suppressed spontaneous firing. In a different experiment, 3 OSNs were recorded simultaneously. The data trace in B is shown with greater temporal resolution than the traces in A to better illustrate how different units could be distinguished by the different size currents. D illustrates the time course of suppression caused by 1 mM Zn²⁺. Washout produced partial recovery. Each point represents the total number of APs fired by the 3 OSNs in a 10-s interval.

Blocking Na_V1.5 VGSCs in OSNs suppresses spontaneous firing.

Because Na_V1.5 channels are largely inactivated in resting OSNs, we expected that a selective blocker of Na_V1.5 channels would have little effect on spontaneous firing in resting cells. With the same recording protocol as for TTX above, Zn²⁺ was bath applied while monitoring activity. Contrary to expectations, Zn²⁺ inhibited spontaneous firing in OSNs (Fig. 8, B and D). Application of 1 mM zinc salicylate completely suppressed firing in six of six cells; washing out the Zn²⁺ partially reversed the suppression. Since salicylate has been reported to affect VGSCs in its own right (Liu et al. 2007), we tested two other zinc salts, ZnCl₂ and ZnSO₄; both suppressed firing with onset and recovery rates indistinguishable from those seen with zinc salicylate, suggesting it was Zn²⁺ that suppressed firing. Overall, spontaneous activity was recorded from 19 OSNs exposed to 1 mM Zn²⁺. Spontaneous firing was fully suppressed in 17 cells, partially suppressed in 1 other cell, and not sustained in 1 cell. Since most Na_V1.5 channels should be closed and inactivated under these recording conditions, the primary effect of Zn²⁺ should be to block the few noninactivated Na_V1.5 channels that carry the window current. These results lead us to suggest that Na_V1.5 window currents are responsible, at least in part, for the generator potentials that initiate spontaneous firing in OSNs.

Quantitative evaluation.

Early whole cell recordings confirmed that the opening of a single channel was capable of eliciting action potentials in rodent OSNs (Lynch and Barry 1989; Maue and Dionne 1987), although at the time the channel was not identified. To evaluate the feasibility of our proposal that Na_V1.5 channels may elicit spontaneous firing, we modeled the depolarization induced by the opening of a single Na⁺ channel in a small cell. The issue is whether opening a single Na⁺ channel can produce sufficient current to depolarize an OSN to its firing threshold. In a small cell with a resting membrane conductance g_m, opening a single Na⁺ channel with conductance γ admits charge driven by the voltage difference between the membrane potential E_m and the Na⁺ equilibrium potential E_Na. When the channel opens, current flows first to charge the membrane capacitance; if the channel does not close, the current will shift the membrane potential from an initial E_m(0) to an equilibrium value, E_m(∞), where

$$E_{\text{m}}\left(\infty \right)=[g_{\mathrm{m}}{E}_{\text{m}}\left(0\right)+\gamma E_{\mathrm{N}\mathrm{a}}]/[g_{\mathrm{m}}+\gamma ]$$

The change in potential will have an exponential time course with a time constant τ = C_m/(g_m + γ), where C_m is the membrane capacitance:

$$E_{\text{m}}\left(t\right)=E_{\text{m}}\left(0\right)+\Delta E\left[1-\exp \left(-t/\tau \right)\right]$$

where

$$\Delta E=E_{\text{m}}\left(\infty \right)-E_{\text{m}}\left(0\right)=\gamma \left[E_{\text{Na}}-E\left(0\right)\right]/\left(g_{\text{m}}+\gamma \right)$$

We used this expression to estimate the time t′ needed for the membrane potential to reach a firing threshold E_m(t′). Writing ΔE(t′) = E_m(t′) − E_m(0), the expression is

$$t\prime =\tau \hspace{0.17em}\ln \left\{\Delta E/\left[\Delta E-\Delta E\left(t\prime \right)\right]\right\}$$

For an OSN with nominal membrane and channel properties [input resistance 10 GΩ, g_m = 100 pS, C_m = 2 pF, E_m(0) = −70 mV, E_Na = +60 mV; Na_V1.5 channel with γ = 20 pS, 2 ms mean open duration], we estimate that a single Na_V1.5 channel would generate a steady-state depolarization [relative to E_m(0)] of ∼21.7 mV that rose with a time constant of 16.7 ms. Under these conditions one Na_V1.5 channel would bring the membrane potential to a voltage of −55 mV (∼firing threshold) in ∼19.6 ms. An opening of this duration would be very rare. Single-channel open durations of rat cardiac Na⁺ channels are exponentially distributed with a mean value of ∼2 ms (Patlak and Ortiz 1985). Using this value, open durations lasting 19.6 ms or longer should occur only once in ∼18,000 openings. However, bursts of openings or overlapping openings of several channels could substantially increase the likelihood of firing. For example, Patlak and Ortiz (1985) noted that rat cardiac Na⁺ channels occasionally showed brief bursts of openings that resulted in a second, slower current decay lifetime of 8–14 ms. For nominal bursts with lifetimes in this range, say 10 ms, action potentials should be generated once in every 7 bursts, decidedly more frequent. Given the peak maximal Na⁺ currents we observed in OSNs, and attributing 50–70% of the maximal current to Na_V1.5 channels, we estimate there may be on the order of 2,000–3,000 Na_V1.5 channels in each OSN. Thus, while the probability of an Na_V1.5 channel opening spontaneously at the resting potential is unknown, the presence of a large number of these channels in each OSN suggests that long-duration openings, bursts, or coincident openings of several channels may not be infrequent events, and that such events could depolarize OSNs to threshold.

DISCUSSION

Diverse mechanisms are known to modulate spontaneous firing in neurons, but the precise origins of the generator potentials that initiate spontaneous activity have not been examined widely. For example, in hippocampal interneurons (Maccaferri and McBain 1996), somatosensory neurons (Momin et al. 2008), and vestibular neurons (Horwitz et al. 2014) HCN channels modulate the interspike interval, while in hypothalamic tuberomammillary neurons a small persistent Na⁺ current appears to be the dominant modulator (Taddese and Bean 2002). But whether either of these currents also produces the generator potentials is not clear. In neurons that show high-frequency repetitive firing such as cerebellar Purkinje and granule cells, a resurgent Na⁺ current (Khaliq et al. 2003; Raman and Bean 1997) associated with Na_V1.6 VGSCs (Afshari et al. 2004) appears to both set the interspike interval and produce the generator potentials. High-frequency repetitive firing in which one action potential can drive the next differs from low-frequency, random firing, and the underlying mechanisms are likely to differ as well.

In OSNs, spontaneous firing is dominated by low-frequency, random events interspersed with occasional bursts. OSNs express CNG and HCN channels, various K⁺, Ca²⁺, and Cl⁻ channels, and at least three types of VGSCs; the uncoordinated basal activity of these channels causes the resting membrane potential to fluctuate. As such, identifying one channel type as the sole origin of the generator potential may be futile. We have approached this issue by positing that OSNs exhibit a level of basal channel activity that modulates the probability of firing by shifting the mean resting potential and that occasionally results in transient perturbations that bring the membrane to firing threshold. Our data suggest that the Na_V1.5 channels could contribute significantly to the perturbations since blocking them almost completely suppresses spontaneous firing. Other data suggest that the K⁺, CNG, and HCN channels may dominate the basal activity but do not serve as the proximate trigger of an action potential. Nevertheless, the matter is not clear. Possibly generator potentials in OSNs arise as a serendipitous, coincident activation of several channel types.

The sensitivity of spontaneous firing to intracellular cAMP focused attention on two cyclic nucleotide-dependent ion channels as possible mediators of the generator potential. The CNG channel, a part of the odor transduction pathway expressed mainly in the cilia of OSNs, was quickly ruled out since neither knocking out the CNG2A subunit (Brunet et al. 1996; Lin et al. 2000) nor blocking the channel (Nakashima et al. 2013) suppresses spontaneous firing. Unlike CNG channels, HCN channels are expressed primarily in OSN somata (Mobley et al. 2010). The HCN channels are activated by hyperpolarization, although with relatively slow kinetics, and when open they pass small depolarizing currents through low-conductance channels (Thon et al. 2013). This promotes a feedback equilibrium between the hyperpolarization-induced channel opening and the depolarizing current that stabilizes the membrane potential. When cAMP binds to HCN channels, the equilibrium point shifts to more depolarized potentials.

The basal cAMP level in OSNs is maintained by the constitutive activity of β₂ARs acting primarily in the soma (Nakashima et al. 2013) and by ORs acting locally in the cilia (Reisert 2010; Takeuchi and Kurahashi 2008). cAMP causes OSNs to depolarize when its level rises or hyperpolarize when it falls, and coincident with this rise or fall the rate of spontaneous firing also rises or falls (Nakashima et al. 2013). The regulation of spontaneous firing by cAMP-sensitive HCN channels marks them as a potential source of the generator potentials that elicit spontaneous firing. However, their slow activation kinetics, small conductance, and distance from the cilia where odor transduction occurs and the observation that spontaneous firing persists in the presence of an HCN channel blocker (Nakashima et al. 2013) suggest that HCN channels may contribute to but not be the sole source of the generator potentials.

Our study indicates that a previously undetected VGSC subtype, Na_V1.5, may be an important factor in generating OSN action potentials. We find that Na_V1.5 channels are localized in the dendritic knob; that a blocker of the channels suppresses spontaneous firing; that the channels have a window current allowing them to open spontaneously at the resting potential where spontaneous firing occurs; and that the cAMP-dependent modulation of spontaneous firing can be accounted for by the kinetic properties of the channel.

VGSCs are composed of a primary α-subunit and one or more accessory β-subunits (Catterall et al. 2005). Expression of the α-subunit alone produces a functional VGSC, while the accessory β-subunits modulate channel kinetics, gating, and localization (Brackenbury and Isom 2011). Nine genes encode the α-subunits; six express TTX-S channels blocked by nanomolar TTX (ED₅₀ ≤ 10 nM), and three express TTX-R channels requiring micromolar TTX for block (ED₅₀ ≥ 1,000 nM) (Goldin 2001). While the TTX-S subtypes are expressed widely in excitable cells, the TTX-R subtypes have more limited distributions, being found primarily in peripheral neurons and cardiac myocytes (Goldin 2001). Many neurons and other excitable cells express multiple VGSC subtypes, but the rationale for this elaboration is seldom understood.

Using RT-PCR, we detected RNA for five different VGSC α-subunits in mouse olfactory epithelial tissue, the same five subunits found previously by expression profiling (Sammeta et al. 2007). Immunohistochemistry previously revealed the expression of Na_V1.7 (Ahn et al. 2011; Weiss et al. 2011) and Na_V1.3 (Weiss et al. 2011) in mouse OSNs. Our study confirms the expression of Na_V1.7 but adds that Na_V1.5, the TTX-R cardiac subtype, is also expressed in mouse OSNs. While Na_V1.5 is not widely expressed in neurons, low levels have been detected in brain (Wang et al. 2009), macrophages (Carrithers et al. 2007), and embryonic dorsal root ganglia (Rush et al. 2007), and a truncated isoform is expressed in adult mouse dorsal root ganglia (Kerr et al. 2007).

The Na_V1.5 channels in OSNs exhibit two distinct properties. First, the channels have a strongly negative inactivation characteristic with a steady-state half-inactivation potential near −100 mV. As a result, most Na_V1.5 channels are inactivated (closed, unresponsive to depolarization) at nominal OSN resting potentials (about −70 mV). In mature, dissociated OSNs (identified by morphology and the presence of apical cilia), Na_V1.5 channels carry ∼70% of the maximal Na⁺ current elicited when inactivation is relieved by hyperpolarization, the balance of current being carried by other VGSC subtypes. When OSNs are held at −70 mV, the peak Na⁺ currents are much smaller than maximal because of Na_V1.5 inactivation and the voltage sensitivity of the remaining Na⁺ current reflects the loss of the Na_V1.5 component. Na⁺ currents with similar properties were reported by Rajendra et al. (1992) in dissociated rat OSNs, although the channel subtypes were not identified.

The second distinct property of olfactory Na_V1.5 channels is the presence of a window current centered at the nominal resting potential. This is seen by overlap of the foot of the activation curve with the tail of the steady-state inactivation curve and, as explained above, establishes a small but real probability that closed, noninactivated Na_V1.5 channels can open spontaneously at membrane potentials within the “window.” To assess the feasibility of the hypothesis that window currents may elicit potentially significant depolarization, we estimated the effects of single-channel openings using physiological values appropriate for Na_V1.5 channels in OSNs. The results indicate that long-duration openings, bursts of openings, or the coincident opening of several channels can produce generator potentials of sufficient magnitude to depolarize OSNs to threshold. Thus Na_V1.5 channels can support spontaneous generator potentials.

VGSCs serve two widely recognized roles in neurons, accounting for the upstroke and propagation of action potentials; we ruled out both roles for Na_V1.5 channels in OSNs. By immunohistochemistry, Na_V1.5-like IR was found to be highly enriched in the apical dendritic knobs of mature OSNs but absent from their axons. This distribution differs markedly from that of Na_V1.7, which is present on the cell soma and axon and which propagates the action potentials that carry odor information down the olfactory nerve (Ahn et al. 2011; Weiss et al. 2011). The absence of Na_V1.5 channels from axons precludes such a conductive function. In addition, knocking out SCN9A, the gene that encodes the Na_V1.7 α-subunit, does not prevent OSNs from firing in response to odor (Weiss et al. 2011), although it does inhibit signal propagation. Since Na_V1.5 channels are largely inactivated during odor responses, VGSCs other than Na_V1.5 and Na_V1.7 must be responsible for the upstroke of the action potential. Nevertheless, blocking Na_V1.5 channels with Zn²⁺ suppresses spontaneous firing in OSNs. The selectivity of Zn²⁺ for Na_V1.5 channels relative to the TTX-S subtypes is due to the configuration of the TTX binding site (Ravindran et al. 1991; Schild et al. 1991). Zinc may also affect other voltage-sensitive channels by screening surface charge, suggesting that its capacity to suppress spontaneous firing may not be limited solely to Na_V1.5 channels. Nevertheless, these results suggest that Na_V1.5 channels have a significant role in generating spontaneous activity.

If Na_V1.5 window currents contribute to the generator potentials that underlie spontaneous firing in OSNs, how could the cAMP-dependent transduction pathway affect the rate of Na_V1.5-dependent firing? The role of a generator potential is to shift membrane potential transiently from its initial resting value to the firing threshold to induce an action potential. If the potential transient is too small, no action potential will be generated. Suppose for a moment that the generator potential is due solely to Na_V1.5 channels. For an OSN with a resting potential near −70 mV, the required perturbation would be ∼15 mV. As discussed above, an Na_V1.5-induced potential transient of this magnitude would be a rare event because most of the open durations of the channels are too short. VGSC open durations are exponentially distributed, with more short-duration than long-duration openings (Aldrich and Stevens 1987). Because the OSN membrane potential is sensitive to intracellular cAMP through actions on CNG and HCN channels, a small change in cAMP will shift the membrane potential, thereby altering the magnitude of the potential transient needed to elicit firing. It will also change the rate of firing nonlinearly because the distribution of VGSC open durations is exponential. Depolarization that brings the membrane potential closer to threshold will allow a greater proportion of the Na_V1.5-induced transients to reach threshold, resulting in an exponential increase in the rate of firing. In the same way, hyperpolarization will exponentially reduce firing. This hypothetical example illustrates how a change in membrane voltage brought about by a shift in the level of intracellular cAMP can alter the rate of spontaneous firing generated by Na_V1.5 window currents. The effect would be similar if Na_V1.5 channels and HCN channels worked in concert to produce generator currents, the Na_V1.5 window currents acting as the trigger and the HCN channels serving to amplify the requisite current. Membrane potential as the factor that couples cAMP to the firing rate in OSNs can explain many observations: expression of exogenous K⁺ channels (Yu et al. 2004), which increases G_K, hyperpolarizes OSNs, and lowers the firing rate; expression of inactive ORs (Connelly et al. 2013), which should lower intracellular cAMP and hyperpolarize OSNs, suppresses firing; overexpression of HCN channels (Nakashima et al. 2013) depolarizes OSNs and increases firing; knockout of HCN channels (Mobley et al. 2010), which should hyperpolarize OSNs and suppress activity, interferes with glomerular targeting; block of Ano2 Cl⁻ channels (Reisert 2010), which should hyperpolarize OSNs, suppresses firing.

The location of Na_V1.5-like IR at the end of the OSN dendrite places the Na_V1.5 channels close to the site of odor transduction. OSNs have a single dendrite that terminates in a swelling from which sensory cilia bearing ORs and CNG channels extend into the overlying mucus. The ORs, CNG channels, and Na_V1.5 channels are thus located close to one another in OSNs, allowing any interaction to be most acute. Under physiological conditions, constitutive OR activity contributes to the basal cAMP level, gates CNG channels to modulate membrane potential locally, and affects the basal firing rate (Reisert 2010). This raises the intriguing possibility that changes in firing accompanying excitatory as well as inhibitory odor transduction may be linked directly to an Na_V1.5-dependent spontaneous firing mechanism. The capacity to initiate action potentials near the site of odor transduction may be crucial in facilitating the sensory response of OSNs.

In conclusion, OSNs express at least three VGSC subtypes, each with a different primary role. Our data suggest that the Na_V1.5 subtype localized in the dendritic knob may contribute to the generator potentials underlying spontaneous firing. The Na_V1.7 subtype expressed in the axon and soma propagates action potentials to the olfactory bulb, while a third VGSC subtype, likely Na_V1.3, is needed to produce the initial upstroke of the action potential. This division of responsibilities may be oversimplified but underscores the principle that different VGSC subtypes may work cooperatively to generate, sustain, and propagate action potentials in OSNs. Whether the functional relationship among the different VGSC subtypes extends beyond spontaneous firing to involve the response to odor and the encoding of odor information remains a possibility that must yet be explored.

GRANTS

This work was aided by support from Boston University, the Rocky Mountain Taste and Smell Center Core for Cellular Visualization and Analysis [National Institute on Deafness and Other Communication Disorders (NIDCD) P30 DC-04657; D. Restrepo, principal investigator], and NIDCD Grants DC-04863 to V. Dionne and DC-006070 to D. Restrepo and T. E. Finger.

DISCLOSURES

Additional support was received from Allergan, Inc. (44960 to S. R. Levinson). No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: C.T.F., A.H., N.D.D., N.S., S.R.L., and V.E.D. performed experiments; C.T.F., N.D.D., N.S., T.E.F., and V.E.D. analyzed data; C.T.F., A.H., T.E.F., and V.E.D. interpreted results of experiments; C.T.F. and V.E.D. drafted manuscript; C.T.F., A.H., N.D.D., N.S., S.R.L., T.E.F., and V.E.D. approved final version of manuscript; T.E.F. and V.E.D. prepared figures; T.E.F. and V.E.D. edited and revised manuscript; V.E.D. conception and design of research.