Functional Regulation of T-Type Calcium Channels by S-Nitrosothiols in the Rat Thalamus

Pavle M. Joksovic, Allan Doctor, Benjamin Gaston, Slobodan M. Todorovic


Although T-type Ca2+ channels in the reticular thalamic nucleus (nRT) have a central function in tuning neuronal excitability and are implicated in sensory processing, sleep, and epilepsy, the mechanisms involved in their regulation are poorly understood. Here we recorded T-type Ca2+ currents from intact nRT neurons in brain slices from young rats and investigated the mechanisms of T-type channel modulation by S-nitrosothiols (SNOs). We found that extracellular application of S-nitrosoglutathione (GSNO), S-nitrosocysteine (CSNO) and S-nitroso-N-acetyl-penicillamin (SNAP) rapidly and reversibly reduced T-type currents. The effects of SNOs are strongly stereoselective at physiological concentrations: L-CSNO was fourfold more effective in inhibiting T-type current than was D-CSNO. The effects of GSNO were abolished if cells had been treated with free hemoglobin or N-ethylmaleimide, an irreversible alkylating agent but not by 8-bromoguanosine-3′,5′-cyclomonophosphate sodium salt, a membrane-permeant cGMP analogue or 1H-(1,2,4) oxadiazolo (4,3-a) quinoxalin-1-one, a specific soluble guanylyl cyclase inhibitor. In addition, bath applications of GSNO inhibited T-type currents in nucleated outside-out patches and whole cell recordings to a similar extent, with minimal effect on cell-attached recordings, suggesting a direct effect of GSNO on putative extracellular thiol residues on T-type channels. Biophysical studies indicate that GSNO decreased the availability of T-type channels at physiological potentials by modifying gating and stabilizing inactive states of the channels. In current-clamp experiments, GSNO diminished the amplitude of low-threshold calcium spikes and frequency of spike firing with minimal effects on the passive membrane properties. Collectively, the results indicate that SNOs may be a class of endogenous agents that control the functional states of the thalamus.


T-type or low-voltage-activated (LVA) currents in the thalamus are crucial in regulating the neuronal excitability that underlies rhythmic thalamo-cortical oscillations with small membrane depolarizations. They also have been implicated in the various stages of the sleep-and-wake cycle, sensory processing, and the pathophysiology of absence epilepsy (Steriade 2005). Studies using anatomical lesions (Jones 1985) and physiological methods (Steriade et al. 1985) have established that GABAergic reticular thalamic neurons (nRT) have a key function in initiating these low-amplitude oscillations. However, the molecular mechanisms for the regulation of ionic conductances that are responsible for these oscillations with naturally occurring modulators have not been extensively studied. One such mechanism involves the production of nitric oxide (NO) and related compounds. NO, an important signaling molecule in the peripheral as well as the CNS (Lipton 1999; Stamler et al. 2001), is known to signal via the widely studied cGMP-dependent stimulation. More recently, however, a parallel signaling network has been described in which endogenous redox-activated NO congeners signal via regulated S-nitrosylation of critical cysteine residues of proteins (forming S-nitrosothiols or SNOs); this novel, redox-responsive signaling pathway has been demonstrated in neurons (Stamler et al. 2001). There is strong immunohistologic evidence that NO is endogenously produced in thalamus-projecting cholinergic neurons (Usunoff et al. 1999), which in turn have a major function in the regulation of oscillatory modes of thalamic neurons associated with sleep and arousal (McCormick and Bal 1997). However, the precise cellular mechanisms of redox-activated forms of NO (particularly SNOs) on neuronal signaling and, more specifically, their effects on rhythmic oscillatory activity dependent on T-type calcium channels, have not previously been studied. Such agents could provide an important intrinsic mechanism for the control of neuronal excitability in both physiological and pathological conditions and be novel targets for therapeutic interventions.

Here we have used several NO-generating compounds and thiol-modifying agents to evaluate the possible mechanisms and importance of their action in regulating T-type channels and underlying low-threshold calcium spikes (LTS) in nRT neurons in intact brain slices. Our results demonstrate that these agents downregulate function of T-type channels in nRT neurons and underlying burst firing via a direct neuronal membrane effect that is independent of cGMP and likely involves S-nitrosylation of critical cysteine resides on external side of the channel. Additionally, our data strongly suggest that alterations of tissue redox states in the thalamus with endogenous or exogenous NO-generating agents may be a novel approach for treating pathological states associated with abnormal oscillations of thalamo-cortical networks.


In vitro tissue slice preparation

We performed most of the experiments on transverse rat brain slices cut through the middle anterior portion of the nRT (Paxinos and Watson 1982) at a thickness of 250–300 μm. Gravid Sprague-Dawley rats were housed in a local animal facility in accordance with protocols approved by the University of Virginia Animal Use and Care Committee. We adhered to the guidelines in the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Rats at the ages of 7–14 days of either sex were anesthetized with halothane and decapitated. The brains were rapidly removed and placed in chilled (4°C) cutting solution consisting of (in mM) 2 CaCl2, 260 sucrose, 26 NaHCO3, 10 glucose, 3 KCl, 1.25 NaH2PO4, and 2 MgCl2 equilibrated with a mixture of 95% O2-5% CO2. We glued a block of tissue containing the thalamus to the chuck of a vibrotome (TPI, St. Louis, MO) and cut 250–300 μm slices in a transverse plane. We incubated the slices in 36°C oxygenated saline for 1 h, then placed them in a recording chamber that was superfused at a rate of 1.5 ml/min with, in mM, 124 NaCl, 4 KCl, 26 NaHCO3, 1.25 NaH2PO4, 2 MgCl2, 10 glucose, and 2 CaCl2 equilibrated with a mixture of 95% O2-5% CO2. Slices were maintained in the recording chamber at room temperature on room air, where they remained viable for ≥1 h.

Recording procedures

The extracellular saline solution typically used for recording Ca2+ currents in whole cell and outside-out experiments and as the extracellular solution in current-clamp experiments, consisted of (in mM) 2 CaCl2, 130 NaCl, 2.5 MgCl2, 10 glucose, 26 NaHCO3, 1.25 NaH2PO4, and 0.001 tetrodotoxin (TTX). In some current-clamp experiments, TTX was omitted to study spike firing of nRT neurons. For recording T-type currents in brain slices, we used an internal solution (solution 1) of (in mM) 135–140 tetramethylammonium hydroxide (TMA-OH), 10 EGTA, 40 HEPES, and 2 MgCl2 titrated to pH 7.15–7.25 with hydrofluoric acid (HF) (Todorovic and Lingle 1998). For some experiments, we altered this internal solution by decreasing TMA-OH to 90 and adding, in mM, 3 MgATP, 0.3 Tris-GTP, 45 Cs methane-sulfonate titrated with HF to pH 7.15–7.25 (solution 2). For recording the effects of S-nitrosoglutathione (GSNO) on G-protein-mediated processes in Ca2+ currents in brain slices, the internal solution contained (in mM) 110 Cs-methane sulfonate, 14 phosphocreatine, 10 HEPES, 9 EGTA, 5 MgATP, and 0.3 Tris-GTP adjusted to pH 7.15–7.20 with CsOH (solution 3). In some experiments, GTP was replaced with equimolar GTP-γ-S (guanosine-5′-O-(3-thio)triphosphate). Recording electrodes for current-clamp studies contained (in mM) 130 KCl, 5 MgCl2, 1 EGTA, 40 Na-HEPES, 2 MgATP, and 0.1 Na3-GTP at pH 7.2. For the data presented membrance potential values were corrected for the measured liquid junction potential of −10 mV (solution 1), −2 mV (solution 2), and −3 mV (solution 3) in voltage-clamp experiments and −5 mV in current-clamp experiments. For cell-attached recordings of Ca2+ currents, electrodes contained 140 mM TEA-OH, 10 mM BaCl2, 2 mM MgCl2, 1 mM CsCl2, 3 mM 4-aminopyridine, 1 μM TTX, and 10 mM HEPES, pH 7.2 adjusted with Tris-base solution (Joksovic et al. 2005a).

All recordings were obtained from thalamic neurons visualized with an infrared differential interference contrast camera (C2400; Hammamatsu, Hammamatsu City, Japan) on the Zeiss 2 FS Axioscope (Jena, Germany) with a ×40 lens and patch-clamp pipette using a Sutter micromanipulator MP-285 (Sutter Instrument, Novato, CA).

Electrophysiological recordings

We recorded Ca2+ currents in thalamic slices from 188 visually identified rat nRT neurons. Recordings were made with standard whole cell, outside-out, and cell-attached voltage-clamp techniques (Hamill et al. 1981) or the nucleated patch technique (Sather et al. 1992). Electrodes were fabricated from thin-walled microcapillary tubes (Drummond Scientific, Broomall, PA) and had final resistances of 3–6 MΩ. We recorded membrane currents with an Axoclamp 200B patch-clamp amplifier (Molecular Devices, Foster City, CA). Voltage commands and the digitization of membrane currents were done with Clampex 8.2 of the pClamp software package (Molecular Devices, Foster City, CA). Neurons were typically held (Vh) at −90 mV and depolarized to test potential (Vt) of −50 mV every 10–20 s to evoke inward Ca2+ currents. Data were analyzed using Clampfit (Molecular Devices, Foster City, CA) and Origin 7.0 (OriginLab, Northampton, MA). For whole cell recordings, we filtered currents at 5–10 kHz; for cell-attached, cell-free outside-out, and nucleated patch recordings, we filtered currents at 2–5 kHz. We typically compensated for 50–80% of series resistance (Rs). In some experiments, a P/5 protocol was used for on-line leakage subtraction.

Methodological considerations

By necessity, thalamic slices were obtained from young rats. With current technology, direct visualization of nRT neurons in this heavily reticulated region of the thalamus of adult animals is difficult. Therefore although this is the first description of T-type channel modulation by SNOs in nRT neurons, it is not certain that the same mechanism would be identical in the same neurons from adult rats.

Because voltage control is compromised in whole cell recordings from slices due to the presence of extensive cell processes, we paid close attention to the following signs of good voltage control: there was no extensive delay in the onset of current and the onset and offset kinetics depended on voltage, not on the amplitude of current. In the kinetic study, we included only cells in which, according to these criteria, adequate clamp conditions were obtained. In whole cell experiments, because intact nRT neurons have long processes, rapid components of recorded current, such as fast-activation kinetics or tail currents, are not likely to reflect the true amplitude and time course of Ca2+ current behavior. However, all our measurements of amplitudes from holding, peak, and steady-state currents were made at time points sufficient to ensure reasonably well-clamped current conditions. Furthermore, using brain slices from young animals, in which dendritic processes are not fully developed, ameliorated the space-clamp problem.

By using cell-attached, outside-out, and nucleated patch recordings, we avoided space-clamp problems, but measured currents are of substantially smaller amplitudes than in whole cell recordings. Thus we set specific criteria for the inclusion of data from the nucleated and cell-attached patch recordings in our analysis: recordings had to be stable for ≥5 min; electrode capacitance had to be sufficiently well compensated; and seal resistance had to be sufficiently high (range, 3–25 GΩ) to allow unequivocal identification of ensemble channel currents. T-type channel activity was recognized by characteristic near-complete inactivation of current at negative voltages, which could be well described with a single-exponential time course. The steps we used to activate T-type channels in the cell-attached patches were similar to those used in whole cell experiments and nucleated patch experiments. T-type currents were presented conventionally as inward currents.

Drug delivery to intact tissue slices is compromised by a decline in drug concentrations along the length of the bath and diffusion through the tissue, which contains endogenous metals and redox-sensitive buffers. Consistent with this we noticed significant cell-to-cell variations in the effects of the same concentrations of SNOs. Thus although our method allows investigation of the effects of redox agents in intact native cells, all quantitative assessments should be taken with caution. Actual effective concentrations of all drugs are likely to be much lower than those reported.

Analysis of current

The voltage dependence of steady-state activation was described with a single Boltzmann distribution Math where Gmax is the maximal conductance, V50 is the half-maximal voltage, and k (units of mV) represents the voltage dependence of the distribution. The voltage dependence of steady-state inactivation was described with a single Boltzmann distribution Math where Imax is the maximal current, V50 is the half-maximal voltage, and k (units of mV) represents the voltage dependence of the distribution. The time course of current inactivation and recovery from inactivation was fitted using a single-exponential equation, f(t) = A1exp(-t/τ1), yielding one-time constants (τ1) and their amplitude (A1).

The amplitude of T-type current was measured from the peak, which was subtracted from the current at the end of the depolarizing test potential to avoid small contamination with residual high-voltage-activated (HVA) currents. For all current-voltage (I-V) curves and steady-state inactivation curves, fitted values typically were reported with 95% linear confidence limits. Input resistance (Rin) was determined from the slope of the peak voltage versus the current plot that resulted from injecting 80- to 160-ms-long current ranging from 100 to 500 pA. Statistical analysis was performed with either paired or unpaired Student's t-test, where appropriate with statistical significance at P < 0.05.

Drugs and chemicals

Tetrodotoxin (TTX) was obtained from Alomone Lab (Jerusalem, Israel) and GSNO from Axxora LLC (San Diego, CA). l-S-nitrosocysteine (L-CSNO) and D-CSNO were freshly synthesized by mixing equal volumes of 8 mM of L- or D-cysteine, in 250 mM HCl, 0.5 mM EDTA, and 100 mM NaNO2 in PBS at room temperature as described previously (Lipton et al. 2001); these preparations were stored in the dark on ice or at −80°C until used. All other salts and chemicals were obtained from Sigma Chemical (St. Louis, MO). Stock solutions of 100 mM GSNO, 100 mM 8-bromoguanosine-3′,5′-cyclomonophosphate sodium salt (8-Br-cGMP), 100 mM of SNAP were made in distilled water, while 100 mM stock solutions of N-ethylmaleimide (NEM), 100 mM 2-(trimethylammonium) ethyl methanethio-sulfonate (MTSET), and 10 mM 1H-(1,2,4) oxadiazolo (4,3-a) quinoxalin-1-one (ODQ) were made in dimethylsulfoxide (DMSO). The maximal final concentration of DMSO used in our experiments was 0.5%, which does not significantly affect native thalamic Ca2+ currents (Joksovic et al. 2005b). Stock solutions (1 mM) of human oxygenated hemoglobin were prepared in normal saline, dialyzed overnight as described previously (Doctor et al. 2005), and diluted in external solution just before recordings.


Multiple independently controlled glass syringes attached to the common PVC tubing served as reservoirs for a gravity-driven perfusion system. Manually controlled valves were used to switch solutions. In most of the experiments, a stock solution of GSNO was applied to the bath directly with a pipette, giving a calculated nominal concentration in the bath of about 1 mM. All experiments were done at room temperature (20–24°C). All drugs were prepared as stock solutions and freshly diluted to appropriate concentrations at the time of the experiment to avoid any chemical interaction of redox agents with trace metal ions in external solutions. During each experiment, solution was removed from the end of the chamber opposite the tubing by constant suction. Changes in Ca2+ current amplitude in response to rapidly acting drugs or ionic changes typically were complete in 1–2 min. Switch of flow between separate perfusion syringes, each containing control saline, resulted in no significant changes in the amplitude and kinetics of T-type Ca2+ current.


Effects of SNOs on T-type calcium currents in intact nRT neurons in brain slices

GSNO is a naturally occurring low-mass SNO, which supports NO+ (nitrosonium) transfer redox chemistry. Thus we initially tested the effects of this agent on isolated T-type currents in nRT neurons. Figure 1, A–C, shows that local extracellular application of GSNO in nRT neurons reversibly inhibited peak T-type currents in concentration-dependent fashion, giving almost complete maximal inhibition and an estimated IC50 level of 1.2 ± 0.1 mM. No apparent changes in macroscopic current inactivation kinetics were observed during the application of GSNO. No desensitization of the response was observed when the same concentration of GSNO was repeated in the same cells within one minute (1 ± 5% difference, P > 0.05, paired t-test, n = 4, data not shown). We found that GSNO also inhibited T-type current in thalamic relay cells of the ventrobasal (VB) nucleus. Figure 1D shows that 1 mM GSNO reversibly inhibited about 15% of T-type current in a VB neuron. On average, 1 mM GSNO inhibited 21 ± 2% of peak T-type current in VB neurons (n = 6, P < 0.001, paired t-test).

FIG. 1.

Inhibitory effects of S-nitrosoglutathione (GSNO) on T-type currents in thalamic neurons. A: representative raw T-type current traces in an reticular thalamic nucleus (nRT) neuron (Vh −90 mV, Vt −50 mV) before (control), during, and after (wash) application of 1 mM GSNO. The application resulted in depression of the current amplitude but had no significant effect on the apparent macroscopic kinetics of the current. B: time course of inhibition of T-type current with the same concentration of GSNO in the same cell demonstrates rapid current inhibition. The return to baseline was rapid and almost complete after washout of GSNO. The horizontal bar indicates the time of GSNO application. C: concentration-response curve for T-type current inhibition by GSNO. The solid symbols indicate the average of multiple determinations (n = 6–22) and vertical lines are ± SE (visible only if bigger than symbols). The solid line represents the best fit of the Hill equation. The fitted value for the curve is IC50 of 1.2 ± 0.1 mM, n = 3.35 ± 1.3; max fitted effect = 100% inhibition. D: representative time course of inhibition of peak T-type current in a VB neuron by 1 mM GSNO. Inset: average traces from 5 consecutive sweeps from the same experiment before (black line) and during (gray line) application of GSNO from the time point labeled by a gray circle.

To determine whether the inhibitory effect of GSNO is mechanistically based on S-transnitrosylation and/or NO· release, we used oxygenated human hemoglobin. Hemoglobin from red blood cells commonly serves as a trap for both NO radicals and NO+ (Carver et al. 2005; Doctor et al. 2005). Figure 2, A and B, shows the time course and traces of GSNO inhibition of T-type current in nRT neurons in the presence of 100 μM of oxygenated hemoglobin, respectively. Application of hemoglobin alone inhibited T-type current to the similar degree and abolished effect of GSNO that was subsequently applied. Figure 2C summarizes results from similar experiments, showing no significant further effect of GSNO on peak T-type current in the presence of hemoglobin in the same cells (40 ± 6% inhibition before and 50 ± 5% during hemoglobin applications, n = 7, P > 0.05, paired t-test). These experiments strongly suggest that GSNO could interact with T-type channels via NO+ and/or NO· molecules. Next we examined the effects of other SNOs such as SNAP and CSNO. SNAP mimicked the effects of GSNO but was less effective and, at 1 mM, reversibly inhibited in average 22 ± 4% of T-type current (P < 0.01, paired t-test, n = 5 cells, Fig. 3A). Stereoselective effects of some endogenous SNOs have been implicated in physiological redox signaling. For example, the strong stereospecificity of CSNO in its effects on biological systems is presumably related to the ability to support NO+ transfer chemistry in vivo within sterically restricted signaling regions protecting redox-responsive cysteine residues in regulatory proteins (Lipton et al. 2001) and/or due to existence of stereoselective membrane transporters for CSNO (Li and Whorton 2005). We therefore tested the effects of L-CSNO and D-CSNO on T-type currents in nRT neurons (Fig. 3, B–D). We found that L-CSNO was the most potent among SNOs that inhibited peak T-type current in 14 tested cells with 50 μM inhibiting 22 ± 3% and 100 μM inhibiting 37 ± 11% of T-type current. Importantly, in the next set of experiments, we included in internal solution 1 mM MTSET, membrane-impermeant thiol-modifying oxidizing agent that blocks nRT T-type currents when applied in external solution (Joksovic et al. 2006). We found that in the presence of internal MTSET, 100 μM L-CSNO applied externally still inhibited T-type current to similar extent by 28 ± 5% (n = 6 cells, P > 0.05). In contrast, D-CSNO was not significantly effective at 50 μM (5 ± 2% change in peak current, n = 5, P > 0.05). At 100 μM, D-CSNO inhibited only ∼10% of T-type current (n = 5, P < 0.01). The histogram in Fig. 3D compares the effects of stereoisomers of CSNO on peak T-type current in nRT neurons. Because both isomers release the NO radical (NO·) at the same rate (Lipton et al. 2001), more prominent inhibition of T-type current by L-CSNO strongly suggests direct trans S-nitrosylation underlies our observed inhibition of T-type channels by endogenous low-mass SNOs. Furthermore, experiments with MTSET strongly suggest that observed stereoselectivity is due to direct effects on the channels rather than differences in the uptake through the cell's membrane.

FIG. 2.

Oxygenated hemoglobin abrogates inhibitory effects of GSNO on T-type currents in nRT neurons. A: time course of the inhibition of T-type current by GSNO in the presence and absence of oxygenated human hemoglobin (Hb). Note the similar degree of reversible depression of peak T-type current by GSNO and Hb when these agents are applied alone. However, there was minimal additional effect of GSNO in the presence of Hb. B: raw T-type current traces from the same experiment depicted on A. C: histogram compares results from 7 cells in experiments similar to one depicted on A and B. The average magnitude of T-type current inhibition (means ± SE) by 1 mM GSNO (▒) was 40 ± 6% (P < 0.01). Hb given alone (▒) similarly inhibited T-type current by 40 ± 7% (P < 0.01). However, co-application of GSNO and Hb in the same cells (▤) did not result in further significant inhibition of T-type current (50 ± 5% inhibition, P > 0.05). Paired t-test was used for statistical analysis.

FIG. 3.

Inhibitory effects of other SNOs on T-type current in nRT neurons. A–C: representative traces show T-type currents before (control), during the application, and after (wash) the application of 1 mM SNAP (A, averages of 3 consecutive sweeps), 0.05 mM L-CSNO (B, averages of 6 consecutive sweeps), and 0.05 mM D-CSNO (C, raw traces). All agents except D-CSNO reversibly inhibited peak T-type current and had minimal effects on current kinetics. Calibration bars are scaled to the same value representing 50 ms for horizontal and 100 pA for vertical bars. D: histogram for T-type current inhibition by 2 concentrations each of L-CSNO (□) and D-CSNO (▒). The columns indicate the average inhibition of T-type current from 5 to 8 determinations ± SE. Both 50 and 100 μM L-CSNO (□) were fourfold more effective than was D-CSNO (▒). *, statistically significant difference in inhibitory effects (P < 0.001, Student's t-test). There is no statistically significant difference (n.s.) in inhibition of T-type current with 100 μM L-CSNO regardless whether MTSET was included in internal solution.

Mechanisms of S-nitrosothiol inhibition of T-type currents in nRT neurons

In studying these mechanisms, we used GSNO to probe for potential direct effects on neuronal membrane versus indirect effects (e.g., G-protein-mediated intracellular signaling) because it is both stable and generally cell-impermeant. If the effects of GSNO are related to generated diffusible intracellular second messengers such as cGMP, we expected that GSNO would exert some effect on T-type currents in the cell-attached mode. In this recording configuration, access of bath-applied GSNO to the channels in the patch would require transport across the cell membrane or diffusion of putative second messenger.

As shown in the typical cell-attached experiment depicted in Fig. 4A, we found that bath application of GSNO (1 mM) for ≤3 min minimally affected ensemble channel currents (6 ± 1% block, n = 8, P < 0.01). In contrast, Fig. 4B shows that the same concentration of GSNO in nucleated outside-out patches inhibited 39 ± 3% of ensemble channel currents (n = 7, P < 0.001); this is similar to the results of previous experiments in whole cell recordings. Figure 4C summarizes these experiments, comparing the effects of GSNO in cell-attached and nucleated patches. These data strongly suggest that putative GSNO-sensitive sites on T-type channels are located on the extracellular side of the neuronal membrane and, moreover, are not dependent on diffusible intracellular second messengers. Furthermore, using different internal solutions to minimize other intracellular biochemical modifications of the channel, such as phosphorylation and G-protein-mediated reactions, did not significantly affect the ability of GSNO to inhibit T-type current in whole cell experiments. For example, with an internal solution without ATP and ∼80 mM F in the internal solution (Fig. 4D, open bar, n = 12) or where GTP was replaced with GTP-γ-S (Fig. 4D, middle bar, n = 5), an irreversible activator of G proteins did not significantly alter the inhibitory effect of GSNO when compared with experiments in which the internal GTP and ATP levels were not manipulated (Fig. 4D, right bar, n = 5). Finally, we used outside-out recordings in cell-free patches from nRT neurons and found that 100 μM L-CSNO reversibly inhibited 31 ± 4% (n = 3 patches, P < 0.05, paired t-test) of ensemble channel current (Fig. 4, E and F).

FIG. 4.

Different effects of SNOs on T-type current in nRT neurons in recordings from cell-attached and outside-out patches and similar effects in whole cell recordings with different internal solutions. A: ensemble T-type current traces recorded in cell-attached mode before (control, black trace) and during (gray trace) the application of GSNO. 1 mM GSNO applied in external solution had very little effect. Traces in each condition are averaged from 5 consecutive sweeps. B: in contrast, in traces on the right (Vh −90 mV, Vt −50 mV) recorded in nucleated outside-out patch configuration, peak T-type current during application of 1 mM GSNO (gray trace) was diminished ∼50% compared with those in the control traces obtained before and after drug application (black traces). Traces in each condition are averaged from 6 consecutive sweeps. C: histogram compares the effect of 1 mM GSNO in cell-attached and nucleated patch recording mode yielding, respectively, 6 ± 1% (n = 8 cells, P < 0.01, paired Student's t-test) and 39 ± 3% inhibition of peak current (n = 7 cells, P < 0.001, paired Student's t-test). Asterisk indicates significant difference in nucleated patch vs. cell-attached recordings (P < 0.001, unpaired Student's t-test). D: histogram compares the effects of 1 mM GSNO on peak T-type current in whole cell recording in the presence of different internal solutions. Tetra methyl ammonium fluoride (TMAF; open square), GTP (right-hatch square), and GTP-γ-S (left-hatch square) resulted, respectively, in 37 ± 3, 46 ± 5, and 45 ± 3% inhibition of baseline T-type current by GSNO (P < 0.001 for each experiment, paired Student's t-test). Vertical bars represent ± SE; the number of experiments is indicated in parentheses. The inhibitory effect of GSNO on baseline T-type current amplitude was not significantly different (n.s.) regardless of the internal solution used (unpaired Student's t-test). E: ensemble T-type current traces recorded in cell-free outside-out mode before (control, black trace) and during (gray trace) the application of L-CSNO. L-CSNO at 100 μM applied in external solution inhibited ∼33% of the peak current. Traces in each condition are averaged from 5 consecutive sweeps. F: time course of the effects of L-CSNO in outside-out recordings from the the same cell depicted on E of this figure. Note that effect is almost fully reversible.

These data strongly suggest that the effects of SNOs do not occur via diffusible second messengers and G proteins but may be consistent with the existence of redox-sensitive sites on the neuronal membrane that can modulate T-type channel behavior in nRT neurons. Thus we wished to determine whether irreversible thiol modification of T-type channels alters the effects of GSNO on nRT currents. NEM has been a commonly used agent for studies of redox reactions since it was demonstrated covalently to modify protein thiol groups by alkylation, it thus may serve to block thiol-based redox reactions on regulatory cysteines in proteins of interest (Gaston 1999; Stamler et al. 2001). Thus if GSNO changes a cell's redox states by acting on putative thiol groups on the channel, it should be possible to eliminate or greatly attenuate its effect by prior bath application of NEM. For these experiments, we applied GSNO first and measured the response of the peak T-type current. Then we applied 0.3 mM NEM for 8–15 min until an apparent steady-state effect was achieved (Fig. 5A). Then, using the same cells (n = 7), we applied the same concentration of GSNO again after NEM was washed out to avoid direct chemical interaction of these agents. NEM at this concentration slowly blocked T-type currents on its own by ∼50%, but the effect of GSNO measured after NEM was greatly abolished inhibiting only additional 4% of T-type current. This strongly suggests that the effect of GSNO on T-type current is due to alteration of thiol groups on channels. Indeed, the experiments summarized in Fig. 5A (right) indicate that NEM diminished peak T-type current when applied after GSNO and completely abrogated the effects of GSNO when repeatedly applied to cells. In other experiments, we demonstrated that 0.3 mM NEM also abrogated the inhibitory effects of another thiol-modifying oxidizing agent, 1 mM of 5,5′ dithio-bis(2-nitrobenzoic acid) (DTNB): DTNB alone inhibited 51 ± 5% peak T-type current, P < 0.001, paired t-test; DTNB after NEM treatment in the same cells inhibited only 1 ± 2% of remaining current (P > 0.05, paired t-test; n = 5, data not shown). In contrast, NEM did not affect the ability of 3 μM mibefradil, a traditional T-type channel blocker (Ertel and Clozel 1997), to inhibit T-type Ca2+ currents in nRT cells (mibefradil blocked 67 ± 1% of remaining current, n = 3, P < 0.01, data not shown). Similarly, we showed (Fig. 5B) that 8-Br-cGMP (1 mM), a cell-permeant cGMP analogue, minimally affected the response of cells to GSNO even though, when given alone, 8-Br-cGMP also inhibited T-type current. Furthermore, we tested ODQ, a specific soluble guanylyl cyclase inhibitor at concentrations of 10 μM and found that similarly to 8-Br-cGMP, inhibited T-type current when given alone but minimally affected the response of cells to GSNO (Fig. 5C). At these concentrations, ODQ was reported to abolish the effects of another SNOs (e.g., SNAP) on inhibitory synaptic currents in brain slices (Li et al. 2003).

FIG. 5.

N-ethylmaleimide (NEM) but not 8Br-cGMP and 1H-(1,2,4) oxadiazolo (4,3-a) quinoxalin-1-1 (ODQ) abrogated the effects of GSNO on T-type currents in nRT neurons. A, left: representative time course showing peak T-type current amplitudes (Vh −90 mV, Vt −50 mV) before (control) and during applications of 1 mM GSNO (1) and 0.3 mM NEM (2). Horizontal bars indicate the duration of application of GSNO and NEM. Initial GSNO application reversibly inhibited ∼50% and NEM inhibited ∼45% of the peak current. When the same concentration of GSNO was applied after NEM treatment, it induced only minimal noticeable inhibition of the inward current. Note that the inhibitory effect of NEM on peak T-type current had a slower time course than did that of GSNO. Right: histogram from similar experiments where vertical columns illustrate the effects of GSNO on T-type currents in cells before (control) and after exposure to NEM (n = 7). These agents decreased T-type current 47 ± 10% (1st application of GSNO), 48 ± 4% (NEM), and 54 ± 6% (2nd application of GSNO that was following NEM). Note that NEM completely occluded the effect of GSNO on peak T-type current. B: representative time course showing peak T-type current amplitudes in control conditions and during applications of GSNO (1) and 1 mM 8-bromoguanosine-3′,5′-cyclomonophosphate sodium salt (8-Br-cGMP; 2) is depicted on the left with horizontal bars indicating duration of application of GSNO and 8Br-cGMP. Initial GSNO application reversibly inhibited ∼25% and 8Br-cGMP similarly inhibited ∼26% of the peak current. When the same concentration of GSNO was applied with 8Br-cGMP, it still induced ∼30% inhibition of the remaining inward current with the similar fast onset and wash. Right: columns illustrate the average effects of GSNO and 8Br-cGMP applied separately, and the effect of GSNO with 8Br-cGMP on T-type currents in the same cells (n = 5). These agents decreased T-type current 27 ± 3% (P < 0.01), 25 ± 7% (P < 0.05), and 41 ± 3% (P < 0.001), respectively. Note that GSNO in the presence of 8Br-cGMP still significantly depressed T-type current (P < 0.05). C, left: representative T-type current time course in control conditions and during applications of GSNO (1) and 10 μM ODQ (2) with the horizontal bars indicating duration of application of GSNO and ODQ. First GSNO application reversibly inhibited ∼200 pA and ODQ similarly inhibited ∼210 pA of the peak current. When the same concentration of GSNO was applied with ODQ, it again inhibited ∼190 pA of the remaining inward current. Right: average effects of GSNO and ODQ applied individually, and the effect of GSNO in the presence of ODQ on T-type currents (n = 8 cells). Initial application of GSNO in these cells produced ∼25 ± 2% (P < 0.01) inhibition of the T-type current. Application of ODQ alone inhibited ∼37 ± 6% (P < 0.01) of current and subsequently applied GSNO with ODQ further inhibited T-type current (59 ± 7% of baseline T-type current amplitude, P <0.01). When compared with ODQ alone, GSNO in the presence of ODQ still significantly depressed T-type current (P < 0.05). *, statistically significant effect; n.s., not significant (P > 0.05) effect by paired Student's t-test.

Many blockers of ion channels may modify the kinetic properties of currents. Thus we studied alterations in biophysical properties associated with the inhibition of T-type currents by GSNO in nRT neurons (Fig. 6). For all experiments, we obtained control data first, then applied GSNO to the same cells. All data were collected after achieving the maximal apparent steady-state effect of GSNO. First we examined the current-voltage (I-V) relationship and found that 1 mM GSNO depressed current amplitude at all tested potentials (Fig. 6A). Figure 6B shows a representative family of traces in control conditions and in the presence of GSNO elicited from a holding potential (Vh) of −90 mV to test potentials (Vt) ranging from −80 to −45 mV.

FIG. 6.

The effects of GSNO on the kinetic properties of T-type currents in nRT neurons. A: average current-voltage (I-V) relationships before (control, •) and during the application of 1 mM GSNO (○). Currents were elicited by progressing from −80 to +20 mV in 5-mV increments from holding potential of −90 mV (n = 5 cells). Note slight shift of I-V curve to the right in the presence of GSNO. B: family of raw T-type current traces elicited with test potentials from −80 to −45 mV in 5-mV increments from holding potential of −90 mV in control conditions (top) and in the presence of 1 mM GSNO (bottom). C: normalized conductance in whole cell recordings from current-voltage experiments depicted in this figure. GSNO shifted half conductance (V50) to the right from −70 ± 1 mV (k = 4.3 ± 0.5 mV) in control to −65 ± 1 mV (k = 6.3 ± 1.0 mV). Fits were done using the Boltzmann equation. The extrapolated reversal potential (Er) was taken to be +45 mV. The estimates of V50 and k show only small shifts, with even ≤20 mV differences in assumed reversal potential values. Inset: normalized activation curves with control (—) and the effect of GSNO (- - -). D: steady-state inactivation kinetics of T-type currents in nRT neurons were examined by paired-pulse protocols at different conditioning potentials. Double-pulse protocols to −50 mV were used to elicit T-type currents, separated by 3.5-s prepulse to potentials ranging from −110 to −40 mV before (•) and during (○) the application of GSNO. The data were best fitted with the Boltzmann equation, yielding V50 of −82.0 ± 0.3 mV (k = 8.8 ± 0.3 mV) in control and −86.0 ± 1.3 mV (k = 10.2 ± 1.6 mV) in the presence of 1 mM GSNO (n = 5 cells). Inset: normalized inactivation curves with controls (—) and the effect of GSNO (- - -). E and F: effects of GSNO on recovery from inactivation of T-type channels in nRT neurons. Reactivation was examined with a paired-pulse protocol in which a 500-ms step to −50 mV was first used to inactivate most of the T-type current. After a recovery interval ranging from 2 to 10,000 ms at −90 mV (E) or −130 mV (F), a 2nd step to −50 mV was used to assay the amount of current that had recovered. The percentage of recovery in the absence and presence of 1 mM GSNO was then plotted as a function of recovery duration and fit with a single exponential function. Interestingly, GSNO prolonged recovery time at both recovery potentials from 750 ± 37 ms in control to 1,386 ± 119 ms at −90 mV (n = 5 cells) and from 555 ± 23 to 788 ± 74 ms at −130 mV (n = 4 cells). Insets: normalized single-exponential curves with controls (—) and the effect of GSNO (- - -).

The effects of GSNO on steady-state activation and inactivation are presented in Fig. 6, C and D, respectively. Interestingly, GSNO slightly modified voltage-dependent gating by shifting the activation curve to the right by 5 mV (Fig. 6C). In contrast, GSNO shifted steady-state inactivation (n = 5 cells) toward more negative potentials by 4 mV (Fig. 6D). Because T-type channels are inactivated at depolarized potentials and can be deinactivated after inhibitory postsynaptic potentials (IPSPs) in CNS, we performed a series of recordings to explore the effect of GSNO on recovery from inactivation (deinactivation) at two different potentials. We found that GSNO at a repolarizing potential of −90 mV (Fig. 6E) slowed recovery from inactivation for ∼85%, and that a repolarizing potential of −130 mV (Fig. 6F) GSNO did so for 42% (n = 5 cells and 4 cells, respectively). Overall, these data suggest that GSNO decreases the availability of T-type channels at physiological membrane potentials by inhibiting gating, slowing recovery from inactivation and stabilizing inactive states of the channels.

Effects of GSNO on lLTS and spike firing in nRT neurons

The ability of thalamic neurons to fire LTS allows burst firing of these neurons with small depolarizations of the neuronal membrane such as those caused by excitatory postsynaptic potentials (EPSPs) in CNS. It has been well established that the LTS of nRT neurons is the key factor leading to synchronization of low-amplitude oscillation in the loop of mutually connected nRT, thalamic relay, and cortical neurons (Steriade 2005). Thus we tested the functional effect of GSNO inhibition of T-type currents on modulation of spike firing in nRT neurons. For these current-clamp experiments, depicted on Fig. 7, we hyperpolarized membrane with constant-current injections to membrane potentials more than −75 mV to remove inactivation of T-type channels. We then injected a small depolarizing pulse via recording electrode to evoke LTS and accompanied repetitive spike firing.

FIG. 7.

GSNO inhibited LTS and burst firing in whole cell current-clamp recordings from nRT neurons. A: representative traces recorded in current-clamp mode from an nRT neuron at a holding potential of −85 mV as indicated by - - -. Low-threshold spike (LTS) was evoked on injection of a small depolarizing pulse (duration: 160 ms, amplitude: 60 pA) and elicited a burst of 5 action potentials that was greatly diminished in the presence of 1 mM GSNO (1 action potential) and almost completely recovered on washout of GSNO (4 action potentials). B: protocol similar to that in A was applied in another nRT neuron but in the presence of 1 μM TTX to isolate LTS. GSNO at 1 mM reversibly decreased the amplitude of LTS ∼42%. C: current-clamp traces showing the reversible inhibitory effect of 1 mM GSNO on LTS that was elicited as a rebound response of cell membrane after the injection of a hyperpolarizing pulse (duration: 240 ms, amplitude: 245 pA). The amplitude of the hyperpolarizing prepulse was little affected by GSNO, which inhibited ∼70% of rebound LTS. Bars are scaled to 15 mV and 50 ms in Fig. 6, A–C. In all experiments, constant current injection was delivered via the recording electrode to achieve membrane potentials (- - -) at which LTS was most prominent. D: histogram compares number of action potential (AP) spikes in control (3.8 ± 0.6) and in the presence of GSNO (0.4 ± 0.25). *, P < 0.001 (n = 5 cells, paired Student's t-test). |, SE. E: histogram depicts the effect of GSNO on LTS amplitude in the presence of TTX, which was significantly depressed by 53 ± 4% at 1 mM GSNO (P < 0.001, n = 6 cells, paired Student's t-test). The amplitude of LTS was measured from the point of inflection to the maximal peak.

Figure 7A (left) shows a representative experiment in which initial small depolarization of the membrane evoked prominent LTS crowned with five action potentials (APs). When GSNO was applied, the same depolarizing pulse evoked reduced amplitude LTS and only one AP. The effect was reversible; a similar spike-firing pattern returned after GSNO was removed (Fig. 7A, right). Figure 7D shows that, on average, GSNO decreased the number of APs in a burst from about 4 to <1. To isolate LTS, in the next series of experiments we included 1 μM TTX in the recording chamber to block sodium-channel-dependent APs. Figure 7B shows a representative cell in which GSNO reversibly diminished the amplitude of LTS. In similar experiments summarized in Fig. 7E, we found that, on average, GSNO reduced the amplitude of LTS by ∼50%. Importantly, Fig. 7C shows that, as estimated by injecting a hyperpolarizing pulse, GSNO caused little change in the passive membrane properties of the neurons. On average, GSNO caused a 4.4 ± 0.8-mV hyperpolarizing membrane potential shift (n = 5, P > 0.05, paired t-test) and increased Rin from 500 ± 70 to 570 ± 80 MΩ (n = 5, P > 0.05, paired t-test).


Inhibition of T-type currents in nRT neurons by endogenous low molecular weight (LMW) SNOs is largely independent of diffusible intracellular second messengers but involves extracellular sites of action

Our previous studies of peripheral sensory neurons indicate that thiol-modifying reducing agent l-cysteine increases the T-type channel-dependent excitability of these cells (Nelson et al. 2005) and enhance pain perception in vivo (Todorovic et al. 2001). Similarly, we have found that DTT (dithiotreitol) and L-cysteine increased the T-type channel-dependent subthreshold excitability of intact nRT neurons in brain slices, whereas the oxidizing agent DTNB decreased it (Joksovic et al. 2006). In the present study, we found that SNOs modulate the function of native T-type Ca2+ channels in intact nRT neurons in brain slices. Inhibition of the current with SNOs is accompanied by alterations in the channel's gating, recovery from inactivation and steady-state inactivation, resulting in diminished availability of the channels to generate LTS in these neurons. The effects of GSNO, a representative SNO, are not fully reproduced in the cell-attached configuration of recording and do not depend on the procedures that manipulate the level of activity of G proteins and ATP in the cells. In contrast, the inhibitory effects of GSNO and CSNO are similar in whole cell and outside-out recordings where membrane has been excised from the cell soma, suggesting that the critical SNO-responsive elements are on the external side of the channel membrane. The effects of GSNO are mimicked by L-CSNO and SNAP, inhibited in solution enriched with oxygenated hemoglobin and are abrogated by the treatment of cells with NEM, a relatively specific and irreversible modulator of thiol groups. Most significantly, stereoselectivity for L-CSNO strongly suggests the presence of a redox-responsive regulatory cysteine(s) protected within the tertiary protein structure. It is unlikely that inhibitory effects of SNOs on T-type current in nRT neurons are via their reducing equivalents given that strong reducing agents like DTT and l-cysteine increase T-type current in these cells (Joksovic et al. 2006). It is interesting that we found that hemoglobin as another endogenous agent also inhibits T-type currents in nRT neurons. Hemoglobin is commonly used in electrophysiological experiments to probe for the involvement of NO-related molecules because it can trap both NO radicals and NO+ (Carver et al. 2005; Doctor et al. 2005). However, recently hemoglobin has been demonstrated to both donate and trap NO and multiple NO metabolites (reviewed in Singel and Stamler 2005). It is not clear whether donation of NO-related molecules may underlie hemoglobin inhibition of T-type current, but it may provide a possible explanation for this effect.

Together, these results strongly suggest that, via trans S-nitrosylation reactions, endogenous LMW SNOs modify the sulfhydryl (thiol) moiety of key regulatory cysteines in native T-type Ca2+ channels, which in turn inhibit Ca2+ currents. Yoshimura et al. (2001) found that N-type HVA Ca2+ currents in peripheral sensory neurons are modulated by SNAP via a cGMP-dependent signaling pathway. Similarly, SNAP inhibits L-type Ca2+ channels in cell-attached patches of bovine chromaffin cells in cGMP-dependent manner (Carabelli et al. 2002). In contrast, S-nitrosylation is implicated in inhibition of L-type Ca2+ channels by GSNO in smooth muscle cells (Poteser et al. 2001). Interestingly, Campbell and colleagues (1996) found evidence for dual modulation of L-type Ca2+ current in isolated cardiac myocytes: GSNO enhanced current in NEM-sensitive manner, suggesting direct extracellular membrane modulation by S-nitrosylation but inhibition of current was sensitive to agents that alter level of intracellular cGMP. Thus different subtypes of voltage-gated Ca2+ channels may be differently regulated by SNOs either by S-nitrosylation or by cGMP-dependent signaling pathways. Indeed recent data indicate that S-nitrosylation may play an important role in regulation of neuronal excitability by affecting multiple targets such as voltage- and ligand-gated ion channels as well as release of intracellular calcium by interaction with ryanodine receptors (reviewed in Ahern et al. 2002).

At least three forms of NO exist, including the nitric oxide radical (NO·), nitrosonium ion equivalents (NO+) with one fewer electrons than does NO, and nitroxyl anion (NO), which has one more electron than does NO. The balance among these species is governed by the redox milieu of tissue, pH, temperature, and the presence of catalytic amounts of transitional metals (Gaston 1999; Lipton 1999). The latter two forms of NO can readily be transferred to endogenous thiol (sulfur) groups on proteins to form nitrosothiols. This reaction, S-nitrosylation or transnitrosation, may be important in the biological regulation of function of many proteins, including ion channels (Lipton 1999; Stamler et al. 2001). Based on the strong stereoselective effects of CSNO, we conclude that the effects of SNOs on T-type channels in nRT neurons are predominantly mediated by NO+.

Thiol sites have been considered to be unique bioregulatory sites (redox switches) for the NO/NO-related species. Many protein targets of S-nitrosylation contain a consensus nitrosylation motif consisting of a single cysteine residue flanked by charged amino acids (Stamler et al. 2001). Cloning the α1 subunits of T-type channels has revealed the existence of at least three subtypes having different kinetic and pharmacological properties. It is likely that these subtypes, α1G (Cav3.1) (Perez-Reyes et al. 1998), α1H (Cav3.2) (Cribbs et al. 1998), and α1I (Cav3.3) (Lee et al. 1999) provide a molecular basis for the heterogeneity of T-type currents observed in native cells (Herrington and Lingle 1992; Todorovic and Lingle 1998). Molecular studies have further indicated that mRNA for both the CaV3.2 (α1H) and CaV3.3 (α1I) channels is abundant in nRT neurons (Talley et al. 1999), whereas exclusively CaV3.1 (α1G) in VB neurons, suggesting that they contribute to native currents. Thus even tough tertiary protein structure of T-type channels is not known, it is interesting to speculate about possible sites on T-type channels where S-nitrosylation may take place. We found that bath application of SNOs diminished current flow though T-type channels in both rat nRT and VB neurons and had a prominent effect in our outside-out, but not cell-attached, nRT recordings. We also found that GSNO inhibits all three subtypes of recombinant T-type channels (M. T. Nelson and S. M. Todorovic, unpublished observation). Thus it is possible that the effects of GSNO are due to S-nitrosylation of conserved cysteine residue(s) located on the extracellular surface of T-type Ca2+ channels. For example there is a well-conserved cysteine flanked by charged residues in repeat II between S5 and the pore-loop domains. Another possibility exists in IVS1, where a single cysteine residue resides near a putative salt bridge. Alternatively, the thiol-containing sites may be located on other adjacent membrane-bound molecules that can directly interact with the T-type channel in nRT neurons. In comparison, GSNO and CSNO have been shown to modulate N-methyl-d-aspartate (NMDA) channel function by donating NO+ (S-nitrosylation) to a single cysteine residue on the extracellular side of the channel (Choi et al. 2000). Because there are >50 conserved extracellular cysteines in T-type Ca2+ channels (Perez-Reyes 2003) extensive molecular, cellular, and electrophysiological studies will be needed for precise deciphering of molecular sites for interactions between T-type channels and NO-related molecules.

Possible functional implication of SNOs in the thalamus

It is well established that in these cells the amplitude of T-type channels is directly proportional to the size of LTS, which in turn regulates burst-firing associated with small depolarizations of neuronal membrane. Furthermore, in our current-clamp experiments we directly tested the possibility that SNOs modulate the excitability of nRT cells. We found that GSNO diminished T-type channel-dependent burst firing and LTS at concentrations that blocked isolated T-type currents in our voltage-clamp experiments.

It has been hypothesized that, in addition to their well-established role in sleep and the absence seizure, LTS of nRT neurons have an important function in the neuronal synaptic plasticity of cortical and thalamic neurons (Llinas et al. 1999; Steriade 2005). Thus it is possible that generation of NO-related molecules in vivo may lead to alterations of T-type channel-dependent neuronal excitability in the thalamus in both physiological and pathological conditions. For example, NO serves as a key signaling molecule in physiological processes as diverse as host-defense reactions, neuronal communication, and vascular tone regulation (Gaston 1999; Stamler 2001). However, disturbance of the balance among different NO-related species has been implicated in inflammatory processes, ischemia of CNS neurons, and neurodegenerative disorders (Lipton 1999). Various NO donors, including SNOs, can exert neuroprotective effects in cortical cell cultures in vitro (Vidwans et al. 1999) and GSNO, reduces inflammation and protects the brain against focal cerebral ischemia in an experimental stroke model in rats (Khan et al. 2005). Furthermore, Nikonenko et al. (2005) reported that the inhibition of T-type Ca2+ channels in vitro protects neurons from delayed ischemic injury. Because ischemic strokes are often accompanied by abnormal neuronal excitability (e.g., seizures), which can aggravate brain damage, it would be a desirable therapeutic intervention to downregulate the function of T-type channels in nRT neurons during ischemic episodes in the CNS. Thus exogenously applied SNOs could be useful for dampening the T-type channel-dependent neuronal excitability in the thalamus that underlies seizure activity.

In conclusion, SNOs regulate T-type Ca2+ channels and underlying burst firing in rat nRT neurons in brain slices via a direct effect on membrane thiol residues. Because the thalamus is the main sensory gate that controls the flow of information into the CNS, we propose that these substances may serve as important endogenous modulators of thalamic functional modes and may be suitable tools for the development of therapies for pathological conditions such as ischemia in which the fine balance of NO-related molecules has been altered.


This work was supported by National Institute of General Medical Sciences Grants GM-075229 and GM-070726 to S. M. Todorovic.


We thank Dr. Lisa Palmer for helpful comments on this manuscript and E. Henderson for preparation of hemoglobin.


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


View Abstract