Suppression by an h Current of Spontaneous Na⁺ Action Potentials in Human Cone and Rod Photoreceptors

Abstract

purpose. The sense of vision in humans is robust, and visual flickering is rarely experienced. To investigate this mechanism, electrophysiological and molecular biological techniques were used on human cone and rod photoreceptors.

methods. Voltage-gated currents were recorded using the patch-clamp technique on isolated human cones, and especially their voltage-gated Na⁺ currents were analyzed in detail. Whether Na⁺ channel transcripts could be detected in single photoreceptors using RT-PCR was also examined, to test the expression of voltage-gated Na⁺ channels in cones and/or rods.

results. Under current-clamp conditions, blocking h currents (hyperpolarization-activated cationic currents) with Cs⁺, Tl⁺, or ZD7288 hyperpolarized the resting potentials of cones and rods by ∼10 to 15 mV, and surprisingly generated spontaneous action potentials. The spontaneous spikes were blocked by 1 μM tetrodotoxin, but not by 1 mM Co²⁺, suggesting that they were Na⁺ spikes rather than Ca²⁺ spikes. Under voltage-clamp conditions, application of Cs⁺ and ZD7288 markedly decreased the steady inward current through the h channel. This is consistent with Cs⁺-induced hyperpolarization under a current-clamp condition. SCN2 Na⁺ channel was observed in both cones and rods by single-cell RT-PCR analysis, suggesting that human photoreceptors express the SCN2 Na⁺ channel.

conclusions. The data confirmed that voltage-gated Na⁺ channels were expressed not only in human rods but also in cones by electrophysiological and molecular biological experiments. These results suggest that the h current may contribute to preventing visual flickering by inhibiting the generation of spontaneous Na⁺ spikes in human photoreceptors.

The retina captures an image and converts it to neural messages that ultimately result in visual experience. ^{1 2 3 4 5 6} The photocurrent of the photoreceptors primarily reflects light-activated biochemical processes in the outer segment. ^{1 7 8 9} Photoexcited visual pigment activates the GTP-binding protein transducin, which stimulates cGMP phosphodiesterase. This enzyme hydrolyzes cGMP, allowing cGMP-activated cationic channels in the outer segment to close. ^{10 11} In contrast, the photovoltage can be shaped by voltage-gated channels in the inner segment. ^{12 13 14 15 16} The voltage-gated currents in the inner segments are of particular importance because the photovoltage, not the photocurrent, controls synaptic transmission. ¹⁶ A light stimulus hyperpolarizes the photoreceptors and reduces the release of the neurotransmitter by decreasing the Ca²⁺ influx at their synaptic terminals. ^{17 18}  

Although mammalian photoreceptors have been thought to be nonspiking neurons, ^{13 14 15 16 19 20 21 22} recently we reported that human rod photoreceptors express voltage-gated Na⁺ channels and can generate Na⁺ action potentials in response to membrane depolarization from membrane potentials of −60 or −70 mV. ²³ Na⁺ spikes in human rods would be elicited at the termination of a light response that hyperpolarized the potential well below −50 mV. This may serve to amplify the release of a neurotransmitter when a bright light is turned off, and thus selectively amplify the off response to the light signal. In addition, signaling via Na⁺ spikes in human rods may serve to accelerate the termination of photovoltage signals, thereby securing the transmission of rapidly changing visual signal. ²³  

If human photoreceptors generated Na⁺ spikes incidentally regardless of a light stimulus, one should experience visual flickering. Indeed, it is known that patients with angina pectoris who take a bradycardiac agent, an h channel blocker, experience adverse visual flickering. ²⁴ However, the visual sense in healthy or normal persons is quite robust and one rarely experiences visual flickering. The mechanism behind this is unknown. We discovered that blockage of a hyperpolarization-activated cationic current (h current) generates spontaneous action potentials in isolated human rods and cones. These action potentials were blocked by bath application of 1 μM tetrodotoxin, a voltage-gated Na⁺ channel blocker, suggesting that they were Na⁺ spikes. We confirmed that voltage-gated Na⁺ channels were expressed not only in rods but also in cones by electrophysiological and molecular biological experiments. We suggest that the h current may contribute to preventing visual flickering by inhibiting the generation of spontaneous Na⁺ spikes in human rods and cones. 

Materials and Methods

Preparations

A small piece of retina was excised from 15 adult (39–71 years) patients with fresh retinal detachment during the surgical procedure to reattach the retina. ²⁵ In a standard vitreous surgery for retinal detachment, the fluid between the detached retina and the retinal pigment epithelium is removed through an artificial retinal hole that is made during surgery. ²⁵ To make the artificial hole, a small piece of retina (diameter, <1 mm) must be exercised. This sample can be used in experiments, as we have reported. ²³ The estimated time passed after the detachment of the photoreceptors is <2 days before surgery. The functional recovery of patients after retinal surgery indicates that there was no severe disease in the retina used in the present experiments. In addition, electron microscopy revealed no abnormality in the inner segment of photoreceptors 7 days after the onset of retinal detachment. ²⁶ All experiments were performed in compliance with the guidelines of the Society for Neuroscience and the Declaration of Helsinki and also were approved by the ethics committee of our institute. 

The dissociation procedure of human cones and rods was similar to that of monkey photoreceptors reported by Yagi and Macleish. ¹⁵ In short, a small piece of human retina was incubated for 5 minutes at 37°C in a solution containing 7 U/mL of papain (Sigma-Aldrich, St. Louis, MO) with no added Ca²⁺ and Mg²⁺. The tissue was then rinsed twice with Ames’ medium and triturated. Isolated cells were plated on a concanavalin-A-coated glass coverslip. All electrophysiological recordings were performed within 7 hours after the surgical procedure. 

Electrophysiology

Membrane voltage and currents were recorded in the whole-cell configuration with a patch-clamp amplifier (Axopatch 200B; Axon Instruments, Foster City, CA) linked to a computer. ^{27 28} The voltage clamp procedures were controlled by the accompanying software (pCLAMP; Axon Instruments). Data were low-pass filtered (4-pole Bessel type) with a cutoff frequency of 5 kHz and then digitized at 10 kHz by an analog-to-digital interface. Isolated photoreceptors were perfused at 1 mL/min with Ames’ medium (which is buffered with 22.6 mM bicarbonate), equilibrated with 95% O₂-5% CO₂, and maintained at 37°C. The recording pipette was filled with pseudointracellular solution (in mM): potassium gluconate, 125; KCl, 15; CaCl₂, 1; EGTA, and 5; HEPES, 10, or Cs⁺ solution: CsCl, 140; CaCl₂, 1; EGTA, 5; and HEPES, 10. The solution was adjusted with KOH or CsOH to pH 7.4. Pipette resistance was 5 to 7 MΩ. Series resistance was neutralized (60%–70%) during recording of the membrane currents. The voltage decrease across the remaining series resistance was not corrected. All test substances (CsCl, TlCl, ZD7288, BaCl₂, tetraethylammonium chloride, CoCl₂, and tetrodotoxin) were applied through the bath. 

RT-PCR Analysis

Total RNA was extracted from a small piece of human retina (RNeasy; Micro kit; Qiagen, Valencia, CA), and stored at −70°C until use. Reverse transcription was performed with reverse transcriptase (Sensiscript; Qiagen) in a total volume of 10 μL. The reaction mixture was incubated for 1 hour at 37°C, followed by heating at 93°C for 5 minutes. RT-PCR was performed with Ex Taq DNA polymerase (TaKaRa Biotech, Ltd., Kyoto, Japan). PCR buffer and the cDNA were mixed to a total volume of 50 μL. The amplification protocol consisted of 3 minutes at 94°C, followed by 45 cycles of 94°C for 15 seconds, 60°C for 15 seconds, and 72°C for 30 seconds, ending with 72°C for 5 minutes. Each primer is shown in Table 1 . We designed PCR primers in different exons of each target gene, so as to distinguish amplified products derived from cDNA and those from genomic DNA. ²⁹  

For single-cell RT-PCR, photoreceptor cells were isolated enzymatically from a small piece of retina. ^{15 30} Isolated cones and rods were identified on the basis of their distinctive morphology. Isolated cells were transferred into a PCR tube separately with a small volume of the pipette solution made with RNase-free water and stored at −70°C until use. Each cell was lysed in the buffer for reverse-transcription containing RNase inhibitor isolated from human placenta (TaKaRa Biotech, Ltd.), and subjected to reaction with the reverse transcriptase (Qiagen). The PCR reaction was performed as just described. The SCN2A F2 and R2 primer sets were used for single-cell RT-PCR (Table 1) . The predicted size of the amplified cDNA from the primers is 160 bp. The amount of cDNA template used for PCR was 4 μL for amplification of Na⁺ channel and 1 μL for amplification of GAPDH. 

Results

Spontaneous Na⁺ Action Potentials in Human Cones and Rods

An isolated human cone and rod photoreceptor can be clearly distinguished from other retinal neurons by the shape. ^{31 32} After dissociation, a cone’s inner segment, soma, and axon were visible (Fig. 1A) . Figure 1Bshows an isolated complete rod with an outer segment, inner segment, soma, and axon. However, the outer segment was lost in most of the photoreceptors, probably due to a partial loss of outer segments caused by the retinal detachment. The lack of an outer segment, consequently the smaller cell capacitance and relatively simple geometry, allowed for excellent space- and voltage-clamp conditions with tight-seal electrodes in the whole-cell configuration. Thus, we recorded mainly from cones and rods that lacked an outer segment. A human rod with no outer segment is clearly distinguished morphologically from an isolated bipolar cell with no dendrite, because all the rods recorded in this experiment have the distinctive inner segment (see Fig. 1B , black arrow), whereas the bipolar cells have no inner segment. ³² In addition, the somatic diameter of human rods (6.3 ± 0.1 μm, n = 20) was smaller than that of human bipolar cells (8.3 ± 0.2 μm, n = 12). 

Under the current-clamp, we examined the effects of an h current on the membrane potential of the isolated rod. An isolated rod that lacks an outer segment had a resting membrane potential of −53 ± 4 mV (mean ± SEM, n = 16) in the normal (Ames’) medium (Fig. 2A , thin line). When 1 mM CsCl, an h channel blocker, ^{13 14 33} was added to the medium, the resting potential hyperpolarized by ∼10 to 15 mV, and surprisingly, the rod fired spontaneous action potentials (Fig. 2A , thick line). A similar result was obtained in five of eight cells, and was also observed when 1 mM TlCl, another h channel blocker, ³³ was applied (Fig. 2B , thick trace; two of three cells). However, 200 μM Ba²⁺, an anomalous rectifier K⁺ channel blocker, was ineffective (Fig. 2B , dotted trace; n = 8). Spikes were blocked by bath application of 1 μM tetrodotoxin (TTX), a voltage-gated Na⁺ channel blocker (Fig. 2A , dotted trace; n = 4), but not by 1 mM Co²⁺, a Ca²⁺ channel blocker (not shown; n = 4). This suggests that they were Na⁺ spikes rather than Ca²⁺ spikes. Although application of CsCl generated spontaneous action potentials in an isolated rod, hyperpolarization of the same rod by 15 mV with extrinsic current did not generate spontaneous spikes (n = 5). 

A similar result was also obtained in intact solitary rods. An intact rod with its outer segment attached had a resting membrane potential of −48 ± 8 mV in the normal medium (n = 3). When 1 mM CsCl was added to the medium, the resting potential hyperpolarized by ∼15 mV, and the rod fired spontaneous action potentials, which were blocked by 1 μM TTX (n = 2). 

An isolated cone that lacks an outer segment had a resting membrane potential of −53 ± 3 mV (n = 24) in the normal medium (Fig. 2C , thin trace). When 1 mM CsCl was applied to an isolated cone, its resting potential also hyperpolarized by ∼10 to 15 mV, and the cone fired spontaneous action potentials (Fig. 2C , thick line). A similar result was obtained in 5 of 24 cells, and was also observed in the presence of 100 μM ZD7288, another h channel blocker (Fig. 2D , thick trace; two of seven cells). However, 200 μM Ba²⁺ was ineffective (Fig. 2D , dotted trace; n = 10). Spikes were also blocked by 1 μM TTX (Fig. 2C , dotted trace; n = 4), suggesting that human cones also generate spontaneous Na⁺ spikes. A similar result was obtained in intact cones. An intact cone had a resting membrane potential of −50 ± 7 mV in the normal medium (n = 4). When 1 mM CsCl was added to the medium, the cone also fired spontaneous action potentials, which were blocked by 1 μM TTX. Hyperpolarization by 15 mV with extrinsic current of an isolated cone, which generated spontaneous spikes by application of CsCl, did not generate spontaneous spikes (n = 5). 

Voltage-Gated Na⁺ Current in Human Cones

To understand the mechanism underlying spike generations in human cones, we recorded their membrane currents under voltage-clamp conditions. A steady inward current of 240 pA was observed at a holding potential (V _h) of −100 mV (Fig. 3A) . Depolarizing voltage steps induced a transient inward current (Fig. 3A , arrow) immediately after the fast capacitive current. The transient inward current began to be activated at −60 mV and was at its maximum at −40 mV (Figs. 3B 3D) . A similar current was recorded in 8 of 43 cones. Mean peak amplitude was 474 ± 25 pA (n = 8). However, when V _h was −50 mV, depolarizing voltage steps did not evoke the transient inward current (not shown), probably because of inactivation of the current. This current was reversibly and dose-dependently blocked by 1 nM to 3 μM TTX (Figs. 3C 3E) ; however, 1 mM Co²⁺ was ineffective. The dose-response curve against TTX was fitted by the Hill equation with an IC₅₀ of 34 nM and a Hill coefficient of 1.1 (Fig. 3E) . These findings suggest that the transient inward current is carried through voltage-gated Na⁺ channels, and also support that action potentials in human cones are Na⁺ spikes. 

To relate the voltage-gated Na⁺ currents to the fast action potentials in human cones, we analyzed their kinetics in detail. The decay phase of the Na⁺ currents induced by depolarization to −40 mV could be fitted by a single exponential function of a time constant of 1.6 ms (Fig. 3F) . The decaying time constant of the Na⁺ currents was maximum at −50 mV (1.8 ± 0.1 ms, n = 5) and decreased as the membrane voltage was depolarized (Fig. 3G) . The kinetics of the voltage-gated Na⁺ current is much faster than that of voltage-gated Ca²⁺ currents in other preparations, ^{34 35 36} suggesting that the voltage-gated Na⁺ currents are responsible for generating fast action potentials in human cones. Recovery from inactivation followed a double exponential time course with time constants of 11 and 84 ms (Fig. 3H) . These time constants are similar to those reported for recovery of voltage-gated Na⁺ current from inactivation in rat and goldfish bipolar cells (6 and 81 ms; ³⁷ 12 and 80 ms ³⁸ ). 

The activation and inactivation curves of voltage-gated Na⁺ currents were fitted by a single Boltzmann function (Fig. 3I) .  

\[1/(1\ {+}\ \mathrm{exp}{[}(V\ {-}\ V_{\mathrm{half}})/K_{h}{]})\]

, where V represents the membrane potential, V _half the half-activation and the half-inactivation voltage, and K _h a coefficient. In the activation curve, a cone was depolarized from V _h of −100 mV, and the conductance was normalized to its maximum value. The half-activation voltage was −35 mV and the K _h was −7.9 mV (n = 5). Inactivation was also voltage dependent. Voltage-gated Na⁺ current was inactivated by a conditioning polarization more positive than −90 mV and became almost zero by a conditioning polarization more positive than −30 mV. The half-inactivation voltage was −64 mV and K _h 7.7 mV (n = 5). These values are similar to those of voltage-gated Na⁺ currents in other preparations. ^{34 35 36 38}  

Hyperpolarization-Activated Cationic Current in Human Cones

Because a light stimulus hyperpolarizes cones, ^{1 7 9} we also investigated the membrane currents activated by hyperpolarization from the resting potential of cones in darkness (−40 mV ¹⁶ ). At a V _h of −40 mV, hyperpolarizing voltage steps evoked a slow inward current (Fig. 4A) . After the termination of the command pulse, an inward tail current was also recorded. These currents were markedly reduced by 1 mM Cs⁺, a blocker of a hyperpolarization-activated cationic current (h current ^{13 14 15} ), in the bath (Figs. 4B 4C) , but not by 20 mM tetraethylammonium chloride (TEA, a delayed rectifier K⁺ current blocker) or 200 μM Ba²⁺ (an anomalous rectifier K⁺ current blocker ³⁹ ). At the resting potential (−53 mV) of cones that lacked an outer segment, the amplitude of the h current was −23 ± 3 pA (n = 11) in the control solution. This indicates that a small portion of the h current is activated at the resting potentials. 

This current was dose-dependently blocked by 1 μM to 0.1 mM ZD7288 and 10 μM to 3 mM Cs⁺ (Fig. 4D) . The dose-response curve against ZD7288 was fitted by the Hill equation with an IC₅₀ of 5.7 μM and a Hill coefficient of 0.98 (Fig. 4D) . The curve against Cs⁺ was also fitted with an IC₅₀ of 32 μM and a Hill coefficient of 0.94. The results suggest that the inward currents during and after the command pulses are an h current. We determined the reversal potential of the h current by measuring the I-V relationship of the fully activated channel (Figs. 4E 4F) . The reversal potential was −32 ± 2 mV (n = 12). The h current is also found in photoreceptors of lower vertebrates ^{13 14} and monkey ¹⁵ retinas, and in human rods, ²³ but is absent in chicken photoreceptors. ⁴⁰  

Effects of Cs⁺, Tl⁺, and ZD7288 on K⁺ Currents in Human Cones and Rods

To determine whether spontaneous Na⁺ spikes induced by application of Cs⁺, Tl⁺, and ZD7288 (Fig. 2)were caused by the inhibition of the h current or caused by the inhibition on K⁺ currents, we investigated the effects of these agents on K⁺ currents in human cones and rods under voltage-clamp conditions. At a V _h of −100 mV, depolarizing voltage steps evoked an outward current in an isolated cone (Fig. 5A) . This current was markedly reduced by 20 mM TEA (a delayed rectifier K⁺ current blocker; Figs. 5C and 5D , filled circles), but not significantly changed by 1 mM Cs⁺ (Fig. 5B) , 1 mM Tl⁺, or 100 μM ZD7288 (Figs. 5C 5D) . A similar result was obtained in an isolated rod (Figs. 5E 5F) . These results suggest that spontaneous Na⁺ spikes of human cones and rods induced by application of Cs⁺, Tl⁺, and ZD7288 were caused by the inhibition of the h current rather than by the inhibition of K⁺ currents. 

Molecular Characterization of Voltage-Gated Na⁺ Channels in Human Cones and Rods

As a further test of the expression of voltage-gated Na⁺ channels in photoreceptors, we examined whether Na⁺ channel transcripts could be detected in single photoreceptors, by using RT-PCR. First, we determined which type of Na⁺ channel is expressed in the human retina. We designed PCR primers corresponding to α subunits of type SCN1, -2, -3, or -8 Na⁺ channels (Table 1 , SCN1A, -2A, -3A, and -8A mRNAs), because these types of Na⁺ channels are expressed in the human central nervous system. ⁴¹ Total RNA was extracted from a small piece of retina and subjected to RT-PCR analysis. PCR products from the primer set of SCN1A, -2A, -3A, and -8A were electrophoresed and their molecular size determined (Fig. 6A) . We detected an amplified product of the predicted size only with a primer set of SCN2A. This result indicates that the SCN2 Na⁺ channel was dominantly expressed in the human retina. 

Next, we performed a single-cell RT-PCR analysis to test whether cones and/or rods express the SCN2A transcript. The SCN2 Na⁺ channel identified from retinal mRNA was observed in 16 of 21 cones and 2 of 4 rods, as illustrated by the example shown in Figure 6B . This suggests that human cones and rods express the SCN2 Na⁺ channel. 

Discussion

The h Currents in Human Cones and Rods

The h currents can be distinguished from anomalous rectifier K⁺ currents because of their poor Ba²⁺ sensitivity. Anomalous rectifier K⁺ currents are nearly completely blocked by Ba²⁺ in the low-micromolar range, whereas h currents are not changed significantly by 200 μM Ba²⁺. ^{33 39} In human cones (this report) and rods, ²³ slow inward-rectifying currents induced by membrane hyperpolarizations were reduced by 1 mM Cs⁺, but not by 200 μM Ba²⁺, suggesting that the inward-rectifying currents are h currents. The pharmacology and kinetics of the h currents in human cones and rods are similar to those of h currents in photoreceptors of cold-blooded vertebrates ^{13 14 33 42} and monkey ¹⁵ retina and are also similar to those of h currents in other preparations. ^{43 44}  

The basic physiological function of h currents in various preparations is thought to be to depolarize cells after periods of hyperpolarization. In thalamic and cardiac cells, an h current contributes to rhythmic firing by depolarizing the membrane potential after the hyperpolarization of an action potential. ^{43 44} In photoreceptors of cold-blooded vertebrates, h currents are thought to initiate the recovery from a strong hyperpolarization arising from bright flashes of light. ^{13 14 45} Fain et al. ⁴⁵ have reported that superfusing rods with a Cs⁺-containing solution eliminates a rapid depolarizing sag in the membrane potential and increases the amplitude of the response to a bright flash of light. Thus, the h current would oppose large hyperpolarizations and act to drive the membrane potential toward depolarization. 

It is likely that the h current in human cones and rods acts to depolarize their resting potentials as well. In the present experiment, at the V _h of the cone’s resting potentials (−53 mV), the amplitude of the h current was −23 ± 3 pA. Under current-clamp conditions, blockage by 1 mM CsCl of the h current hyperpolarized the cone’s and rod’s resting potentials by ∼10 to 15 mV. Therefore, the h current plays an important role in the resting state of human photoreceptors. 

Voltage-Gated Na⁺ Currents in Human Cones

We observed voltage-gated Na⁺ currents in 8 of 43 isolated human cones (19%). This ratio was much smaller than that (68%) obtained in rods of the human retinal slices in our previous experiment. ²³ The reason that voltage-gated Na⁺ currents were observed in only a subset of human cones is still unclear. Wen et al. ⁴⁶ reported a TTX-sensitive Na⁺ current in cultured human and monkey retinal pigment epithelial (RPE) cells but none in RPE cells from freshly isolated tissue. They speculate that a wound-healing type response may initiate expression of the TTX-sensitive Na⁺ channels in RPE cells. This raises the possibility that the isolation procedure of photoreceptors may also initiate expression of voltage-gated Na⁺ channels in human cones. However, in the present experiments, all electrophysiological recordings were obtained within only 7 hours after the surgical procedure, and thus it would be unlikely that the isolation procedure initiates and completes expression of the voltage-gated Na⁺ channels in cones within 7 hours. Thus, this would rule out an injury-induced change in phenotypic expression of the Na⁺ channels in human cones. Therefore, only certain types of human cones may express voltage-gated Na⁺ channels in vivo. Further study is needed to identify which types of cones express the Na⁺ channels. 

We have shown that bath application of Cs⁺ generates spontaneous Na⁺ spikes in human cones and rods. When the bright light is turned off, spikelike current responses are observed in the intact monkey cones using suction electrode recordings. ¹⁹ Schnapf et al. ¹⁹ recorded the membrane current from cone outer segments protruding from small pieces of monkey retina. They observed spikelike current responses in the cone, when a narrow suction electrode squeezed its inner segment. This may suggest that action potentials can be generated in the inner segment of intact primate cones. 

Availability of a Voltage-Gated Na⁺ Channel at Resting Potential and Physiological Role of the h Current

Because activation and inactivation of voltage-gated Na⁺ currents are strongly voltage dependent, it is important to know how many Na⁺ channels are available at the cone resting potentials to understand their contribution to generation of the action potential. Because the inactivation curve for the voltage-gated Na⁺ current in human cones was fitted by a single Boltzmann function, the fraction of Na⁺ channels to be activated by depolarization from the resting potential (−53 ± 3 mV) can be estimated from the function, as described in the Results section. At the resting potential (−53 mV) only 19% of Na⁺ channels can be activated by depolarization. This may explain why the generation of spontaneous Na⁺ spikes in human cones is inhibited under normal (physiological) conditions. 

However, we found that blocking the cone’s h current hyperpolarizes the membrane potential by ∼10 to 15 mV and causes spontaneous Na⁺ spikes. The steady hyperpolarization of 10 to 15 mV caused by blocking the h current would reduce inactivation of the voltage-gated Na⁺ channels. At this voltage (from −63 to −68 mV), 46% to 62% of Na⁺ channels can be activated by depolarization. 

Because the activation curve of the voltage-gated Na⁺ current in human cones was also fitted by a single Boltzmann function, the fraction of Na⁺ channels that are still activated at those membrane potentials can be estimated from the Boltzmann function. In the activation curve, V _half was −35 mV and K _h −7.9 mV. Thus, at membrane potentials of −53, −63, and −68 mV, the estimated values of the activation curve are 0.093, 0.028, and 0.015, respectively. In the voltage-gated Na⁺ current of the Hodgkin-Huxley model, ⁴⁷ they settled on three m-gating particles to control activation and one h-gating particle for inactivation. Thus, the probability that all particles are in the permissive position is m ³ · h. Therefore, at membrane potentials of −53, −63, and −68 mV, the estimated fraction of Na⁺ channels that are still open is 1.5 × 10⁻⁴, 1.0 × 10⁻⁵, and 2.1 × 10⁻⁶, respectively. This suggests that at those steady membrane potentials, the fraction of open Na⁺ channels is decreased, as the membrane potentials of photoreceptors are more hyperpolarized. This result appears to contradict the spontaneous Na⁺ spike generation in human photoreceptors. However, when membrane depolarization activates voltage-gated Na⁺ channels, the fraction of activated Na⁺ channels is mainly determined by the resting value of the h parameter rather than that of the m parameter, because the kinetics of the h parameter is much slower than that of the m parameter. ⁴⁷ Thus, when the resting potentials of photoreceptors were hyperpolarized to between −63 and −68 mV by application of h channel blockers, 46% to 62% of Na⁺ channels were activated by depolarization. Therefore, it is likely that the fluctuations in the membrane voltage activate a significant amount of voltage-gated Na⁺ channel and generate spontaneous Na⁺ spikes in human photoreceptors. 

In contrast, although application of CsCl generated spontaneous action potentials in isolated rods and cones, hyperpolarization of the same cells by 15 mV with extrinsic current did not generate spontaneous spikes (n = 5). The exact reasons that the extrinsic current failed to generate the spikes are still unclear within the present experiment. However, because hyperpolarization of photoreceptors by 15 mV markedly increases the conductance of h channels, the reduction of fluctuations in the membrane voltage due to the increase of the h channel’s conductance or membrane conductance may fail to activate a significant amount of voltage-gated Na⁺ channel and also may fail to generate spontaneous Na⁺ spikes. Simulation studies are needed to elucidate the mechanisms underlying spontaneous Na⁺ spike generation in human cones and rods. 

We suggest that the presence of the h current may contribute to prevent visual flickering by suppressing the spontaneous Na⁺ spikes in human cones. This mechanism can also explain the inhibition of spontaneous Na⁺ spikes in human rods. Indeed, it is known that patients with angina pectoris who take a bradycardiac agent, an h channel blocker, experience adverse visual flickering. ²⁴  

In the dark, the resting potentials of intact cones and rods with an outer segment in vivo are more depolarized than those (−53 mV) of the isolated photoreceptors recorded in this experiment. At dark resting potentials (∼ −40 mV), ¹⁶ the h channel is less activated, and the voltage-gated Na⁺ channel more inactivated, compared with the results obtained in the present experiment. Thus, it seems to be unlikely that the spontaneous Na⁺ spikes are generated in human photoreceptors in the dark. In mesopic vision, however, the resting potentials of photoreceptors can be more negative in voltage than −50 mV. Therefore, the presence of the h current may contribute to the prevention of visual flickering by suppressing the spontaneous Na⁺ spikes in human photoreceptors in vivo in mesopic vision. 

 Supported by Japanese Society for the Promotion of Science Grants 14704075 (FK), 13470372 (MH), and 14580779 (EM); The Promotion and Mutual Aid Corporation for Private Schools of Japan; and The Hori Information Science Promotion Foundation.

 Submitted for publication June 18, 2004; revised August 19, 2004; accepted September 26, 2004.

 Disclosure: F. Kawai, None; M. Horiguchi, None; H. Ichinose, None; M. Ohkuma, None; R. Isobe, None; E. Miyachi, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Primer Sequences Used for Amplification of Human Na⁺ Channels

mRNA Species	Sequence	Predicted Size (bp)
SCN1A F	5′-TCCACCACCCTGTTGCTGTA-3′	145
SCN1A R	5′-CCAATCAGGAGGGTTACTCATT-3′
SCN2A F	5′-CTCTCGATTCAGTGCCACCC-3′	145
SCN2A R	5′-CAGTCTGGAGGGTTACTCATG-3′
SCN3A F	5′-ATTTTCCGATTCAGTGCCACCT-3′	144
SCN3A R	5′-CAGTCAGGAGGGTTGCTCAAG-3′
SCN8A F	5′-CTCTTCAGATTTAGTGCCACGC-3′	130
SCN8A R	5′-CCAGTCAGGAGGGTTACTAAAA-3′
SCN2A F2	5′-TTGGAAGAGGCTGAACAGAAGG-3′	160
SCN2A R2	5′-CTGAAAAAACTCCTATCCCACCAG-3′
GAPDH F	5′-TGAACGGGAAGCTCACTGG-3′	307
GAPDH R	5′-TCCACCACCCTGTTGCTGTA-3′

 F, forward primer; R, reverse primer.

Figure 1.

 

A Nomarski micrograph of a solitary human cone and rod photoreceptor. (A) A solitary cone that lacks an outer segment. Inner segment (solid arrow), soma (solid arrowhead), and axon (open arrowhead) are clearly visible. (B) Solitary rod’s outer segment (open arrow), inner segment (solid arrow), soma (solid arrowhead), and axon (open arrowhead). Bar, 20 μm.

Figure 2.

 

Blockage of a hyperpolarization-activated cationic current, the h current, generated spontaneous Na⁺ action potentials in human rods and cones. (A) Membrane potentials of an isolated human rod in control Ames’ solution (thin trace), 1 mM CsCl (thick trace), and 1 mM CsCl plus 1 μM TTX (dotted trace), under current-clamp conditions of 0 pA injection. The recording pipette was filled with pseudointracellular solution (in mM): potassium gluconate, 125; KCl, 15; CaCl₂, 1; EGTA, 5; and HEPES, 10. (B) Membrane potentials of another rod in control solution (thin trace), 1 mM TlCl (thick trace), and 200 μM Ba²⁺ (dotted trace). (C) Membrane potentials of an isolated cone in control Ames’ solution (thin trace), 1 mM CsCl (thick trace), and 1 mM CsCl plus 1 μM TTX (dotted trace), under current-clamp conditions. (D) Membrane potentials of another cone in control solution (thin trace), 100 μM ZD7288 (thick trace), and 200 μM Ba²⁺ (dotted trace).

Figure 3.

 

Human cones expressed voltage-gated Na⁺ channels. (A) Membrane currents of a human cone induced by depolarizing voltage steps from a holding potential (V _h) of −100 mV. Command voltages were increased in 10-mV steps, from −100 to +40 mV. The recording pipette was filled with K⁺ solution. (B) Current responses in (A) are shown at a faster time scale. Command voltages were increased in 10-mV steps, from −60 to 0 mV. (C) Current responses recorded from the same cell as in (B) in the bath solution containing 1 μM TTX. Command voltages were also the same as in (B). (D) I-V relationship of voltage-gated Na⁺ currents. Peak Na⁺ currents were plotted against test-pulse voltage. (E) Relationship between the inhibition of Na⁺ current and TTX concentration (n = 5). The trace represents the Hill equation obtained by least-squares nonlinear fit to the data. (F) Voltage-gated Na⁺ currents induced by depolarization to −40 mV (V _h = −100 mV). Bath solution contained Cs⁺ (5 mM), TEA (20 mM), and Co²⁺ (1 mM), and the recording pipette was filled with Cs⁺ solution. Thick trace: a single exponential function obtained by least-square nonlinear fit to the data. (G) The mean decay rate of voltage-gated Na⁺ currents was plotted against voltage (n = 5). Error bars: SEM. (H) Recovery from inactivation. After inactivation by a 50-ms pulse to 0 mV, voltage was returned to −100 mV for various times, and recovery was tested by a second voltage-clamp step to 0 mV. The curve through the mean data points represents the best-fitting sum of two exponentials (n = 5). Error bars: SEM. Current for each cell was normalized to the control current before the inactivating prepulse. (I) Activation (•) and inactivate (▴) curves of the voltage-gated Na⁺ currents (n = 5). Relative conductance of the activation curve at a specific membrane voltage was estimated as a ratio of the recorded current amplitude to that expected from the maximum conductance. Relative conductance of the inactivation curve was estimated as a ratio of the current amplitude induced by depolarization to 0 mV after a 1-second conditioning pulse of various voltages to that induced by the same depolarization without conditioning pulses. Traces represent a single Boltzmann function obtained by least-square nonlinear fit to the data. Short vertical bars represent SEM.

Figure 4.

 

Hyperpolarization-activated cationic currents (h currents, I_h) in a human cone. (A) Membrane currents induced by hyperpolarization from a V _h of −40 mV. Command voltages were increased in 10-mV steps from −100 to −40 mV. (B) Current responses recorded from the same cone as in (A), in the normal bath solution (thin trace) and in the presence of 1 mM Cs⁺ (thick trace). The membrane potential was stepped from −40 to −100 mV for 500 ms. (C) I-V relationship of I_h in the normal bath solution (•) and in the presence of 1 mM Cs⁺ (▴). Steady state currents (measured just before the end of the voltage steps) were plotted against test-pulse voltage. (D) A concentration-response curve for block by Cs⁺ (▴) and ZD7288 (▪) of I_h. The smooth curves represent the Hill equation and short bars SEM (n = 4). (E) Reversal of I_h. The membrane potential was stepped from −40 to −140 mV, to activate I_h. After I_h reached its maximum, the membrane potential was stepped from −100 to 0 mV in increments of 10 mV. (F) The fully activated I-V relationship of I_h measured 10 ms after the start of the test step in (E).

Figure 5.

 

Cs⁺, Tl⁺, and ZD7288 did not significantly change K⁺ currents in human cones and rods. (A) Membrane currents of an isolated cone induced by depolarization from a V _h of −100 mV in control solution containing 1 μM TTX. Command voltages were increased in 10-mV steps from −30 to +40 mV. The recording pipette was filled with K⁺ solution. (B) Current responses recorded from the same cone as in (A), in the bath solution containing 1 mM Cs⁺. Command voltages were also the same as in (A). (C) Membrane currents of the cone in the control solution and in the presence of 100 μM ZD7288 or 20 mM TEA. The membrane potential was stepped from −100 to +40 mV for 500 ms. All bath solutions contained 1 μM TTX. (D) I-V relationship recorded from the same cone as in (C) in the control solution (○) and in the presence of 1 mM Tl⁺ (▪), 100 μM ZD7288 (▴) or 20 mM TEA (•). (E) Membrane currents of an isolated rod in the control solution and in the presence of 1 mM Cs⁺ or 20 mM TEA. The membrane potential was stepped from −100 to +40 mV for 500 ms. All bath solutions contained 1 μM TTX. (F) I-V relationship recorded from the same rod as in (E) in the control solution (○) and in the presence of 1 mM Cs⁺ (▪) or 20 mM TEA (•).

Figure 6.

 

Human cones and rods expressed an SCN2 Na⁺ Channel. (A) The SCN2 Na⁺ channel was dominantly expressed in the human retina. Agarose gel electrophoresis of Na⁺ channel cDNAs obtained by RT-PCR from whole retina. Each pair of lanes shows the result from the primer set of SCN1A (1A), SCN2A (2A), SCN3A (3A), and SCN8A (8A) Na⁺ channels. The predicted size of each amplified Na⁺ channel cDNA was approximately 150 bp. Complementary DNA of the predicted size was detected only in the SCN2A lane. M, molecular weight marker; +, retinal cDNA; −, negative control (RNase-free water). (B) Human cones and rods express the SCN2 Na⁺ channel. Agarose gel electrophoresis of SCN2A cDNAs obtained by RT-PCR from single photoreceptor cells. Each lane shows the result from a single cone or rod. The predicted size of the amplified cDNA for SCN2A was approximately 160 bp, for GAPDH approximately 300 bp. In the cone and rod lanes, cDNA of the predicted size was detected in three of six cones and in one of two rods. N, no cell was added in RT-PCR (negative control); P, total retina was added in RT-PCR (positive control).

The authors thank John McReynolds and Norio Ozaki for comments and advice and Noriko Ihira for technical assistance with the RT-PCR experiment.