|Home | About | Journals | Submit | Contact Us | Français|
In addition to the HCN1 channels that mediate the h current, the Kx current also performs signal filtering in rod photoreceptors. This current is known to be mediated by potassium channels and has similarities to the neuronal M current and EAG potassium channels. Although it is known that in filtering the light response of rods, Ih and IKx undergo complementary conductance changes, the qualities and significance of these changes are not clear. Here we present an analysis demonstrating the filtering effect of HCN1 channels in salamander rods when IKx is blocked, and a simulation of the rod light response showing the magnitude and time course of the conductance changes by both currents. From this analysis, we propose that the purpose of opposing conductance changes by Ih and IKx may be to optimize the lateral propagation of signals through gap junctions in the rod network.
During a photoreceptor light response many simultaneous membrane currents are summed by the cell to generate the voltage response. In turn, the voltage change of the cell alters the gating of each conductance through the voltage sensors of the channel. Through this nonlinear action of voltage on gating, and the action of gating on voltage, different classes of ion channels influence one another’s activity. Therefore, it is generally imprecise to classify the function of a single class of ion channel in a neuron outside the context of its peers. In this research addendum we describe one such interaction between Ih, mediated by HCN1 channels, and IKx, mediated by M-like potassium channels1,2 in influencing the membrane impedance of a rod during its light response.
In our work on HCN channels in salamander photoreceptors, we characterized their biophysical properties and investigated their functional role in tiger salamander (ambystoma tigrinum) rods and cones.3 From investigations of the gating kinetics and immunohistochemical staining, we showed that HCN1 channels are responsible for the Ih current observed in these cells. We used the HCN antagonist ZD7288 to block HCN1 channels in rods and cones, and demonstrated that this increases the amplitude and duration of rod and cone light responses. When considered in the frequency domain, HCN block reveals the low-pass filter characteristic of the photocurrent.3 This low-pass characteristic comes from the slowness of the photocurrent, whose gating depends on a complex cascade of molecular interactions. In contrast, the opening of voltage gated channels such as HCN channels depends on the motion of charged voltage sensors and their interactions with the pore, which is generally a much faster process.4,5 In normal physiological conditions, HCN channels reduce response amplitudes at low frequencies, flattening the frequency response of the cell to light stimuli.3 This flattening allows the voltage response to be frequency independent over a wider range of stimulus frequencies, enabling the synapse to avoid saturation at low frequencies while still passing higher frequency signals. From a signal processing standpoint, this compensatory effect by HCN channels is analogous to high-pass filtering. Other studies previously described high-pass filtering in the rod network and postulated that it could be a way for the network to increase the signal to noise ratio for transient signals by spreading them over a larger area.6–8 By dividing transient signals into networked rods, multiple parallel rod to bipolar cell synapses can be used, increasing the signal-to-noise ratio at the bipolar cell layer. In addition to their role in the rod network, other studies have also shown how high-pass filtering due to Ih is important in increasing the speed of the the individual photoreceptor light response.9,10
Although HCN1 channels appear to be major players in the rod and cone light response, they are not the only source of filtering by voltage-gated ion channels. Another current called IKx has also been shown to be involved in high-pass filtering of rod light responses, especially responses to dim stimuli.11 IKx is similar to the M-current in neurons in that it is a potassium conductance that is partially activated at resting potential, is further activated by depolarization, and is largely non-inactivating.2 Kx channels also appear to be similar to EAG and Kcnv2 potassium channels, but its exact molecular origin is presently unknown.12,13 While both IKx and Ih are known to mediate high-pass filtering in rods, an important difference between the two is that during a light response (in which the rod membrane hyperpolarizes), Kx conductance decreases, while h conductance increases. Due to their different reversal potentials (−30 mV for Ih and −75 mV for IKx), the net current change caused by Ih and IKx gating is inward during a light response, tending to counteract the initial hyperpolarization phase of the response.11,14 This reactive depolarizing effect leads to high-pass filtering of the input signal.
Patch clamp recordings were made from dark adapted tiger salamander (ambystoma tigrinum) rods at room temperature. External solution consisted of (in mM) 108 NaCl, 2.5 KCl, 1.2 MgCl2, 2 CaCl2, 5 HEPES (Sigma-Aldrich) titrated to pH 7.7 with HCl. 20 TEA-Cl, 5 BaCl2, and 5 CoCl2 were added to the bath to block voltage gated potassium, Kx, and calcium conductances, respectively.14–16 Where indicated, 50 μM ZD 7288 (Tocris) was added to the bath to block HCN channels.17 Internal solution consisted of 106 K-gluconate, 5 NaCl, 2 MgCl2, 5 EGTA and 5 HEPES, titrated to pH 7.4 with KOH. Whole-cell recordings were made from salamander rods in the whole mount retina, using an EPC-10 patch-clamp amplifier (HEKA) in current clamp mode. Light stimuli were generated with a custom voltage to current source buffer circuit driving a 530 nm green Luxeon V LED (Phillips).
Simulations were computed in MATLAB (Mathworks) using the ode15s numerical solver. The differential equations and constants for the model are given in appendix 1. Equations for voltage gated channels come from studies previously published by Liu and Kourennyi with the Ih model as published by Barnes and Hille.18,19 Although more accurate models of HCN channel kinetics exist,20 the Barnes model describes Ih kinetics relatively well (unpublished data), and requires only one state variable, greatly decreasing computational complexity. Equations and constants for the photocurrent come from studies by RD Hamer, originally from Nikonov et al.21 Initial conditions for the model were determined by allowing the system to relax to steady-state without any light input.
To extend our previous studies of the h-current, we examined the contribution of Ih to the rod light response when the Kx conductance, calcium conductance, and other potassium conductances are blocked. To do this, we recorded the rod light response with 5mM Co, 5 mM Ba and 20 mM TEA present in the bath.14–16 Five flashes of light of increasing intensity were delivered to a rod and the voltage response was recorded. Then Ih was blocked with 50 μM ZD7288 and light responses were recorded again.
In the presence of Co, Ba, and TEA, the rod’s light response is smaller and occurs from a more depolarized potential than in normal ringer solution (figure 1A). Compare this with the simulation of a normal light response in figure 2A. The depolarization in darkness is consistent with a reduction in the outward potassium current (due to IKx and other uncharacterized potassium conductances) that normally counteract the inward dark current. A small transient “nose” in the light response is seen due to the presence of Ih (figure 1A).
When Ih currents are then blocked with ZD7288, the light response is seen to increase in magnitude and the transient “nose” is abolished (figure 2B). This demonstrates that Ih can play a role light response recovery even when IKx is blocked. In both the solutions with and without ZD7288, an overshoot is seen following the recovery phase of the light response (figures 1A and B). This is a known effect of TEA on the rod light response first observed by Fain et al., but unlike in other studies, the overshoots we observed failed to generate regenerative spikes due to the block of calcium currents with Co.18,22 The ionic current that causes this overshoot is not completely clear. Although our model could account for much of the shape of the waveforms in figure 1A and B when Kx and h-conductances were blocked (data not shown), it failed to account for this overshoot. One potential source of the overshoot could be an uncharacterized effect of TEA and/or Co on the photocurrent, which the model did not include.
Although others have noted the complementary conductance changes by Ih and IKx during a light response, the magnitude and time courses of these changes are unknown. To evaluate the simultaneous contributions of Ih and IKx to the rod light response, we simulated the rod light response by solving differential equations describing voltage gated channels and the photocurrent numerically (see Methods and Appendix 1). The model was stimulated with five flashes of light of increasing intensity, and the time courses of the voltage, h, and Kx conductances at each flash intensity were evaluated. During a light response the voltage (figure 2A) causes an increase in h conductance and a decrease in Kx conductance (figure 2B). These complimentary conductance changes tend to counterbalance one another during the flash response, resulting in a reduced net conductance change whose amplitude is time dependent (figure 2B, green traces). With large stimuli, the faster response kinetics of Ih cause a small transient conductance increase, followed by a longer lived conductance decrease due to IKx. Smaller stimuli cause a more synchronous activation of Ih and IKx (figure 2B, green traces). In our model, the net conductance change due to both currents deviates no more than 0.3 nS from the resting level, whereas each individual conductance changes by nearly 0.6 nS. Although the conductance increase by Ih and decrease by IKx are not perfectly synchronized, together they halve the maximal conductance change of one current individually.
It has been observed using current pulse injection that, in contrast to cones, rods do not undergo an appreciable conductance change during a light response.23 One explanation for this observation was that during a light response, an increase in h conductance counteracts the conductance decrease from the photocurrent.23 This hypothesis does not account for the then unknown Kx conductance, and overlooks an important property of the photocurrent. During a light response, the photocurrent, which is actually a shutting off of the inward dark current, causes hyperpolarization of the cell membrane. The dark current’s instantaneous I–V relation is nearly flat at throughout the rod’s physiological voltage range, from −20 to −80 mV, in both light and darkness (Baylor and Nunn, figure 6).24 This property means that from the standpoint of the rod, the photocurrent acts as a current source whose magnitude depends on light, and not on the membrane potential. The counterintuitive consequence is that although the photocurrent is mediated by a closing of the ion channels carrying the dark current, the voltage-independence of the current through these channels means that it does not contribute to a membrane conductance change during a light response. There may, however, be some slow conductance change associated with the voltage dependance of the Na-Ca-K exchange pump.25 It is important to note that unlike the rod, the cone dark current I–V relation is not flat, and therefore cones do undergo a conductance decrease when exposed to light.26
With the dark current ruled out as a source of conductance change, we conclude that the lack of observed net conductance change during a rod light response is likely due to the coordinated counterbalancing of h and Kx conductances, as we show with our simulation (figure 2). While the opposite conductance changes by Ih and IKx were first investigated some time ago,7,8 the question remains as to what, if any, advantage these complementary changes would confer during a light response. One theory is that the two different conductance changes are a consequence of having two separate mechanisms (IKx and Ih) for filtering small and large signals.11 Alternatively, we propose that the answer to this question may lie with the fact that rod photoreceptors are coupled to one another through gap junctions.
In the rod network (figure 3), signals propagate to adjacent rods through gap junctions in order to cancel random noise in individual cells and increase the number of parallel channels used in the rod to bipolar cell synapse.27 One commonly overlooked aspect of the rod network is that the degree to which signals propagate through the network is dependent not only on the strength of the signal itself and the coupling impedance, but also on the membrane impedances of the cells in the network. With high membrane impedance (low conductance), signals tend to dissipate less (figure 3, green arrow) and propagate further (figure 3, red arrow), and with low membrane impedance (high conductance), signals dissipate more readily. If IKx was absent and only Ih was present in rods, not only would high-pass filtering of input signals be reduced, but signals would dissipate more quickly due to unopposed h conductance increase (figure 3). If only IKx was present, then the propagation of signals in the network would be weighted to favor larger responses that completely turn off the Kx conductance. By having both h and Kx conductance, the cell achieves a high degree of filtering of input signals while minimizing the distortion of signal propagation in the network that would be a consequence of membrane conductance change.
We demonstrate that Ih filters the light response even when IKx, ICa and other potassium currents are blocked. This is further evidence that both currents are necessary for filtering of input signals to rods. While it has been previously shown Ih and IKx cause opposite conductance changes during a light response, the potential advantage of these complementary conductance changes has been unclear. Results from our membrane model of the rod show that the conductance changes from Ih and IKx do largely cancel one another, and that the time course of this net conductance change depends on the stimulus flash intensity. Further, we propose a potential advantage of the complementary conductances may be to maximize the amount of filtering by voltage gated channels while minimizing any perturbation of signal spread in the rod network. This would allow the cell to optimally spread signals from illuminated cells into adjacent rods for better use of the synapses between rods and bipolar cells.
This work was supported by grants from NIH (EY 04446), NIH Vision Core (EY 02520), the Retina Research Foundation (Houston), and Research to Prevent Blindness, Inc.