Although sodium-activated potassium channels (KNa) have been suggested to shape various firing patterns in neurons, including action potential repolarization, their requirement for high concentrations of Na+ to gate conflicts with this view. We characterized KNa channels in adult rat dorsal root ganglion (DRG) neurons. Using immunohistochemistry, we found ubiquitous expression of the Slack KNa channel subunit in small-, medium-, and large-diameter DRG neurons. Basal KNa channel activity could be recorded from cell-attached patches of acutely dissociated neurons bathed in physiological saline, and yet in excised inside-out membrane patches, the Na+ EC50 for KNa channels was typically high, ∼50 mm. In some cases, however, KNa channel activity remained considerable after initial patch excision but decreased rapidly over time. Channel activity was restored in patches with high Na+. The channel rundown after initial excision suggested that modulation of channels might be occurring through a diffusible cytoplasmic factor. Sequence analysis indicated that the Slack channel contains a putative nicotinamide adenine dinucleotide (NAD+)-binding site; accordingly, we examined the modulation of native KNa and Slack channels by NAD+. In inside-out-excised neuronal patch recordings, we found a decrease in the Na+ EC50 for KNa channels from ∼50 to ∼20 mm when NAD+ was included in the perfusate. NAD+ also potentiated recombinant Slack channel activity. NAD+ modulation may allow KNa channels to operate under physiologically relevant levels of intracellular Na+ and hence provides an explanation as to how KNa channel can control normal neuronal excitability.
Numerous investigations on sodium-activated potassium channels (KNa) have put forward putative physiological roles for these channels including action potential repolarization, slow afterhyperpolarization, burst firing, and adaptation after repetitive firing (Bhattacharjee and Kaczmarek, 2005). In some neurons, these channels have been suggested to control the resting membrane potential (Haimann et al., 1992; Koh et al., 1994; Bischoff et al., 1998). Recent evidence indicates that in adult medium-sized dorsal root ganglion (DRG) neurons, a single Na+-dependent action potential can elicit a KNa current, and KNa channels were suggested to regulate the depolarizing after potential (DAP) (Gao et al., 2008). However, it has been difficult to reconcile these diverse physiological functions with one apparent property of KNa channels: their requirement for high concentrations of [Na+]i to activate. The normal resting level of [Na+]i in neurons lies between 4 and 15 mm (Rose, 2002), and the effective concentrations required to activate 50% of channels (EC50) have mostly been reported to range between 40 and 80 mm (Dryer, 1994), although some groups have reported much lower EC50s (Haimann et al., 1990; Dale, 1993). The two genes that encode KNa channels, Slack (Slo 2.2, kcnt1) and Slick (Slo 2.1, kcnt2) (Bhattacharjee and Kaczmarek, 2005; Salkoff et al., 2006), when expressed in CHO (Chinese hamster ovary) cells have EC50s of ∼40 and ∼80 mm, respectively (Bhattacharjee et al., 2003). Although it has been suggested that transient elevations of Na+ could reach in the tens of millimolars if a high density of Na+ channels are present and the space is diffusionally restricted such as the node of Ranvier (Koh et al., 1994), quantitative models based on the diffusion equation suggest that the somatic concentration of Na+ in the vicinity of the KNa channels cannot reach this level, even if KNa channels are assumed to be closely associated with Na+ channels (Dryer, 1994). This raises the following question: are KNa channels physiologically relevant, or are they only important during pathological states where [Na+]i can accumulate rapidly such as hypoxia (Yuan et al., 2003)?
Na+ dose–response relationships for KNa channels have been established primarily through excised-patch recordings, because the presumed Na+ -binding site likely resides on the cytoplasmic face of the channel. By nature, the excised-patch configuration removes channels out from their native environment; in the case of KNa channels, this experimental approach may not allow for an accurate assessment of what the normal Na+ sensitivity of these channels might be. For example, when recording KNa channels from intact cultured trigeminal neurons using the cell-attached configuration, KNa channels exhibited a higher open probability compared with excised-patch recordings (Haimann et al., 1992). In addition, the high initial activity seen in cell-attached patches rapidly decreased after patch excision suggesting that a diffusible cytoplasmic factor likely regulates the activity of channels in intact neurons (Haimann et al., 1992). Similarly, in both rat and chick olfactory neurons, KNa channels exhibited a rapid “rundown” after patch excision (Egan et al., 1992; Dryer, 1993), and this rundown was also thought to occur as a result of the diffusion away of some unknown factor modulating KNa channels (Egan et al., 1992). Thus, it is very plausible that KNa channels in their native environment are activated by lower, physiological-relevant [Na+]i compared with channels recorded in excised patches (Dryer, 2003).
Among DRG neurons, KNa channels have only been studied in rat neonatal small DRG neurons (Bischoff et al., 1998) or in adult medium-sized DRG neurons (Gao et al., 2008). In this study, we show that Slack KNa channels are ubiquitously expressed in adult rat DRG neurons. We also determine that the Na+ dose–response relationship of KNa channels in both small and large DRG neurons exhibited rundown during initial patch excision. Using sequence analysis, we discovered an NAD+-binding site in the cytoplasmic terminal of the Slack channel and found that NAD+ can activate both neuronal KNa channels and Slack channels expressed in human embryonic kidney-293 cells. Modulation of KNa channels by NAD+ may now explain how these channels can participate in normal neuronal signaling.
Materials and Methods
Adult rat DRG neurons were dissected and dissociated in 0.28 Wünsch units/ml Liberase Blendzyme (Roche Diagnostics Corporation) for 1 h at 37°C followed by washes in Hanks' buffered saline and trituration. Neurons were plated on poly-d-lysine and laminin (Sigma-Aldrich)-coated coverslips. Dissociated neurons were allowed to settle for 2 h before recording and were then used for up to 2 d. Neurons were cultured in Neurobasal-A medium (Invitrogen) supplemented with B-27 supplement (Invitrogen), l-glutamine (Invitrogen), and nerve growth factor (Harlan) and stored in a humidified incubator in 5% CO2 at 37°C. Large neurons were classified with diameters >50 μm, and small neurons were classified with diameters <25 μm. Proprioceptive and nociceptive DRG neurons are usually designated as large and small, respectively. Overall assumptions in this classification are based on previous work examining conduction velocity versus the size of the neuron (Harper and Lawson, 1985). Neurons that fell outside this range were not recorded because of the possibility the neuron could be either proprioceptive or nociceptive.
Lumbar and thoracic DRGs were isolated from adult male (200 g) Sprague Dawley rats. Animals were anesthetized with sodium pentobarbital (60 mg/kg), perfused transcardially with 60 ml of PBS containing heparin (50 μg/ml) and sodium nitrite (5 mg/ml) followed by 60 ml of cold 4% paraformaldehyde. DRGs were removed, cleaned of surrounding tissue, postfixed in 4% paraformaldehyde, and transferred to 20% sucrose. After embedding in freezing media, frozen 20 μm sections of DRG were made. Slices were permeabilized with a PBS solution containing 0.4% Triton X-100. Sections were then blocked for 2 h at room temperature with PBS containing 5% BSA. Then sections were incubated with a mixture of primary antibodies in PBS containing 5% BSA overnight at 4°C. Primary antibodies included mouse anti-neurofilament antibody (1:400; Millipore Bioscience Research Reagents), calcitonin gene-related peptide (CGRP) antibody (1:2500; Sigma), and chicken anti-Slack antibody (1:1000) (Bhattacharjee et al., 2002). After several rinses, secondary antibodies Alexa Fluor 633 goat anti-mouse, Alexa Fluor 488 goat anti-rabbit, and Alexa Fluor 546 goat anti-chicken were added (1:1000) for 2 h. Coverslips were then mounted on slides using Prolong Gold antifade reagent with 4′,6′-diamidino-2-phenylindole dihydrochloride.
Human embryonic kidney (HEK-293) cells that stably express Slack channels (Yang et al., 2006) were a kind gift from Dr. Fred Sigworth (Yale University, New Haven, CT). Cells were cultured on 35 mm dishes in modified low Na+ DMEM medium (Yang et al., 2006) containing 10% fetal bovine serum and penicillin-streptomycin (Invitrogen) and maintained in a 5% CO2 incubator at 37°C.
Electrophysiology and analysis of data.
All recordings were performed at room temperature. Electrodes resistances for excised-patch recordings ranged between 9 and 13 MΩ. Data was acquired using an Axopatch 200B (Molecular Devices), digitized, and filtered at 2 kHz and by using pCLAMP 9.2 (Molecular Devices) acquisition software. Data analysis was performed using Clampfit (Molecular Devices) and Origin 6.0 (OriginLab). Statistical analysis was performed using GraphPad Prism 4. Data are expressed as mean ± SE. Unpaired Student's t tests and one-way ANOVA analysis were used to determine statistical significance. For inside-out patch-clamp recordings, the pipette solution contained (in mm, unless specified otherwise) 10 NaCl, 130 KCl, 10 HEPES, 5 EGTA, 1 MgCl2, and 1 tetraethylammonium chloride (TEA-Cl). The bath solution contained 130 KCl, 10 NaCl, 10 HEPES, and 5 EGTA. Solutions for patch-perfusion experiments to determine Na+ dependency of channels were prepared as follows: equivalent solutions of 140 KCl, 140 NaCl, and 140 N-methyl glucamine (NMG)-Cl, and corresponding constituents (i.e., HEPES, EGTA) were made first. Starting solutions were then mixed to obtain the desired final ion concentration. The K+ concentration was kept constant at 40 mm; NMG was used as cationic substitute to maintain osmolarity. Therefore, the highest Na+ concentration we could use in our system was 100 mm. Cl− concentrations were kept constant (140 mm) throughout the experiments. The pH of all solutions was adjusted to 7.3 with KOH. Inside-out patches were perfused with the SmartSquirt small volume delivery system (Automate Scientific) using a 100 μm perfusion tip with a flow rate of 0.01 ml/min, and varying concentrations of Na+ were used in the perfusion. The free-acid form of β-nicotinamide adenine dinucleotide (β-NAD) was purchased from Fluka Biochemika and similarly β-NADP, β-NADH α-NAD, and oxidized glutathione (GSSG) from Sigma-Aldrich. Dose–response curves of Na+, Na+, and NAD+ were fit to the equation I = Imin + Imax/[1 + (C50/C)n], where I is the measured current density, Imin is the minimal current density, Imax is he peak current density, C is the concentration of agonist used, C50 is the concentration of agonist required to achieve half maximal activation (i.e., EC50), and n is the degree of cooperativity. For tracking channel rundown time courses, the Savitzky–Golay smoothing algorithm was used to filter irregularities in data and perform a local polynomial regression for several data points. Relative maxima, minima, and width are preserved using this filtering algorithm.
For cell-attached recordings, the bath solution contained 140 NaCl, 3 KCl, 1 CaCl2, 1 MgCl2, and 25 HEPES. Pipette solution contained 10 NaCl, 140 KCl, 1 TEA-Cl, 1 μm tetrodotoxin (BIOMOL International), and 10 HEPES.
Mutation of the putative NAD+ site was performed using sense and antisense primers designed against the putative NAD+-binding region of the Slack channel cDNA in pTRACER. In the primer pair, one base mismatch was included to alter the final sequence that was obtained by PCR of a given cDNA by Pfu Turbo. PCR products were digested with DpnI to eliminate nonamplified cDNA, and the remaining products were used to transform Escherichia coli. Miniprep DNA was then sequenced to verify that the designated changes had been made. Glycine 792 was mutated to alanine (G792A).
HEK-293 cells were cultured in DMEM containing 10% fetal bovine serum and penicillin-streptomycin. Cells were plated on plastic 35 mm dishes at a confluence of 20% and then transiently transfected with 0.9 μg of Slack G792A cDNA and 0.1 μg of CD8 a (lymphocyte cell-surface antigen) cDNA and 5 μl of Lipofectamine (Invitrogen). Recordings were performed 1–3 d after transfection. Transfected cells were identified by their binding to CD8-coated beads (Dyna-beads M-450 CD8).
Ubiquitous Slack expression in adult DRG neurons
Using a previously characterized antibody against Slack channels (Bhattacharjee et al., 2002), we found Slack immunoreactivity in all types of rat DRG neurons (Fig. 1). Using confocal microscopy, we sampled an entire DRG and found Slack labeling in >90% of the neurons (Fig. 1A). Staining was positive for all neurons examined; for example, in small and medium CGRP-positive neurons, equally high levels of Slack immunoreactivity were found as in small CGRP-negative neurons. Slack labeling was also strong in large diameter neurons that were positive for neurofilamin (a marker of myelinated neurons), as demonstrated by the pinkish-purple labeling in overlapping images (Fig. 1A,B). During examination at higher magnification, we found that Slack channels localized at the cell surface of DRG somata and within axonal tracts (Fig. 1B). Control experiments included incubating DRGs with the secondary antibody alone and with the antibody preabsorbed with blocking peptide (data not shown). Slack is widely distributed and not confined to one DRG neuronal subtype. We also found Slick channel immunoreactivity in adult DRG somata; however, there was differential labeling in small- and medium-sized DRG neurons versus large-sized DRG neurons (data not shown), suggesting that Slick channels may have a unique physiological role in nociception.
KNa channels in DRG neurons
Adult DRG neurons were acutely dissociated and recorded within 2–24 h after plating. Neurons were categorized as either “large” or “small” based on diameter size (large >50 μm; small <25 μm). Using the cell-attached configuration, we recorded large conductance KNa-like channel activity from acutely dissociated small (Fig. 2A) and large DRG neurons (data not shown) bathed in physiological saline (n = 4 for both). Channels were confirmed to be KNa channels after excision and demonstration of Na+ dependence (Fig. 2A). Channels typically had large unitary conductance (∼200 pS) (Fig. 2B) and exhibited multiple subconductance states. Our data are very similar to that reported by Haimann et al. (1992) who recorded KNa channels at rest in avian trigeminal neurons using cell-attached patches. These would suggest KNa channels could contribute to physiological excitability.
To further characterize Na+ dependence, inside-out patches were excised from neuronal somata and perfused with increasing concentrations of Na+. Large conductance, Na+-dependent K+ channels were found in nearly three-quarters of patches excised from both small and large DRG neurons. Patches typically contained 3–5 channels. The pipette solution contained 1 mm TEA and was devoid of Ca2+ (contained 5 mm EGTA) as was the bath solution. This high expression correlates with our Slack immunohistochemical data and is similar to what has been reported earlier (Bischoff et al., 1998; Gao et al., 2008). The KNa channel EC50s for Na+ was determined to be 53 mm (n = 6) and 55 mm (n = 6) in small neurons (Fig. 2C) and large neurons (supplemental Fig. 1, available at www.jneurosci.org as supplemental material), respectively, and each with a Hill coefficients of 2.4.
KNa channels exhibit rundown after patches are excised from somata
In some of the patches that were excised from small DRG neurons, we observed a high basal activity followed by a time-dependent decline in activity (Fig. 3A). Neurons were initially bathed in a 130 mm K+, 10 mm Na+-containing solution. We observed this rundown, which occurred within ∼90 s after excision (Fig. 3B), in at least half of the patches recorded (3 of 6) from small neurons and in 1 of 6 large neurons. We retrospectively analyzed the data by averaging the three recordings taken from the small DRG neurons during initial excision and tracked the time course of rundown followed by the perfusion of patches with varying concentrations of Na+. Because the initial activity was very different for each of the three patches, and had differing time course of rundown, to adjust for the scatter, the data was filtered using the Savitzky–Golay smoothing algorithm. Note that channel activity could be restored with increasing concentrations of Na+, although very high concentrations of Na+ were required to match the activity that was seen during the initial excision (Fig. 3B). This rundown phenomenon is similar to the rundown of KNa channels recorded from olfactory neurons (Egan et al., 1992; Dryer, 1993) and trigeminal neurons (Haimann et al., 1992). In these previous reports, rundown was suggested to occur as the result of the diffusion away of some unknown factor modulating KNa channels allowing the channels to exhibit activity at basal [Na+]i conditions.
Slack and Slick channels contain a putative NAD+-binding site
A modulating factor that increases the sensitivity of KNa channels to Na+ is an attractive idea, but a variety of cytoplasmic factors including cAMP, cGMP, ATP, cAMP-dependent protein kinase with ATP, protein kinase C, and pertussis toxin have been tested and none of these caused facilitory effects on native KNa channels (Egan et al., 1992). We examined the secondary structure of the C terminal of both Slack and Slick using the secondary prediction program JPred3 from the University of Dundee (Dundee, UK; http://www.compbio.dundee.ac.uk/∼www-jpred/) and the domain prediction software SBASE (Trieste, Italy; http://hydra.icgeb.trieste.it/∼kristian/SBASE/). SBASE predicted that both Slack and Slick channels contain TrkA-N-like domains in their respective distil C termini. TrkA is a prokaryotic transporter that that binds NAD+ at its N-terminal domain (TrkA N) (Schlösser et al., 1993). Canonical NAD+-binding motifs have particular criteria for binding NAD+. This binding motif includes a “fingerprint region” containing a βαβαβ motif of protein folding. In addition, the nucleotide-binding site must fit the following criteria: (1) A glycine-rich, phosphate-binding consensus sequence of GXGXXG which connects the β1 stand to the α1 helix; (2) six positions usually occupied by small hydrophobic amino acids; (3) a conserved, negatively charged residue at the end of the second β strand; and (4) a conserved positively charged residue at the beginning of the first β strand (Bellamacina, 1996). Using the predicted structural information from JPred3 and the hallmarks for nucleotide binding, we have compiled the structure of putative NAD+-binding sites residing within the second RCK (regulate the conductance of K+) domains of the Slack and Slick subunits, respectively, and compared them to known NAD+-binding sites from prokaryotic K+ transporters (Roosild et al., 2002) (Fig. 4). The third glycine of the GXGXXG phosphate consensus sequence is missing from Slack and Slick putative NAD+-binding sites; however, this third glycine has been replaced by bulkier residues in other proteins that have the capability of binding the phosphorylated form of NAD+, namely NADP+. It is believed that a bulkier amino acid residue is required here to allow the phosphate group of NADP+ to bind (Bellamacina, 1996).
NAD+, but not NADH, modulates KNa channels in DRG neurons
We subsequently tested if NAD+ regulates native KNa channels in DRG neurons. We assessed the KNa channel Na+ dose–response relationship in excised patches from small DRG neurons in the presence and absence of 1 mm NAD+ in the perfusate (Fig. 5) (n = 6). Again, we identified KNa channels using the criteria of conductance, subconductance states, and Na+ dependence. Patches were first perfused with increasing concentrations of Na+ and then the same patches were perfused with the same increasing Na+ concentrations plus 1 mm NAD+ (Fig. 5A). We consistently observed higher open probabilities in the presence of NAD+ versus the absence of NAD+. During these experiments, we did not obtain single channels; instead, patches contained multiple channels per patch; however, we did not observe any changes in unitary conductance nor did we see changes in open times (210 ± 40 ms in 10 mm Na; 192 ± 37 ms in 10 mm Na+ + 1 mm NAD+). The Na+ dose-dependent relationship and Na+/NAD+ dose-dependent relationship was established in small DRG neurons, and we observed a leftward shift in the presence of NAD+ (Fig. 5B). In these experiments, EC50 was estimated to be ∼52 mm, whereas in the presence of NAD+, the EC50 was calculated to be ∼20 mm. We similarly observe NAD+ facilitation of KNa channels recorded from large DRG neurons (supplemental Fig. 2, available at www.jneurosci.org as supplemental material). Note that NAD+ in the absence of Na+ did not affect channel activity.
We next determined if the reduced form of NAD+, namely NADH, could also modulate the open probability of the channel. Although NAD+ and NADH are structurally very similar, the NADH consensus-binding site is considered to be GXGXXGXXXGXXXXXG (Scrutton et al., 1990). We perfused several inside-out-excised patches from small neurons (n = 5) with solutions containing 30 mm Na+ with and without NADH. We found no significant change in open probability versus the control, whereas native channels perfused with NAD+ exhibited a significant 2–2.5-fold increase in open probability (Fig. 5C) (p < 0.01).
NAD+ and NADP+, but not NADH nor αNAD, modulate Slack channels
Since at present the precise molecular composition of native KNa channels in DRG neurons is unknown, we tested if NAD+ also modulated recombinant Slack channels. We performed similar inside-out-excised patch-perfusion experiments on Slack channels recorded from a Slack-stable HEK-293 cell line. Inside-out-excised patches were perfused with 10 mm Na+, followed by 10 mm Na+ and 1 mm NAD+, or 1 mm NADP+ (Fig. 6A,B). Because of the high expression of Slack channels in this cell line, we used a lower concentration of Na+ (10 mm) to study Slack channel open probability. We also did not observe Slack channel run down after patch excision. Similar to the neurons, however, we did see a statistically significant twofold increase in Slack channel open probability in the presence of NAD+ and a statistically significant 1.5-fold increase in the presence of NADP+. We did not observe a change in open probability when NADH was included in the perfusate (Fig. 6A,B,E) (n = 6 for NAD+, n = 5 for NADP+, and n = 5 for NADH), nor was their an effect using the αNAD+ stereo isomer of NAD+ (Fig. 6E) (n = 3). These data confirm that NAD+ and NADP+ act on Slack channels as well.
NAD+ effects on Slack channels are likely not attributable to oxidation
Although our evidence to this point strongly suggested that NAD+ was acting by allosterically modulating Slack channels, we could not be sure that NAD+ was acting by oxidizing Slack channels. To confirm that the NAD+ effects are direct and not attributable to oxidation, we applied 1 mm of the oxidized form of GSSG to patches. At this concentration, GSSG is a strong oxidizing agent (Ruddock et al., 1996). We found that 1 mm GSSG did not alter Slack channel open probability (Fig. 6C,E) (n = 5), suggesting that NAD+ is probably allosterically affecting channels.
In general, NAD+ concentrations within cells are much higher than NADH and NADP+. The cellular concentration of NAD+ has been estimated to be between 0.3 to 0.4 mm (Yamada et al., 2006; Yang et al., 2007), although the concentrations within neurons are not well established. We tested various concentrations of NAD+ to evaluate the minimal concentrations of NAD+ needed to cause a significant facilitation in open probability. We performed experiments on inside-out-excised patches of Slack channels from the same Slack-stable cell line using a fixed 10 mm Na+ concentration and applying varying concentrations of NAD (0.001, 0.01, 0.01, 1, and 10 mm) (Fig. 6D) (n = 5). We observed a significant change in open probability at 0.1 mm NAD and 1 mm NAD (n = 5, p < 0.005 for each). Note, at 10 mm NAD, there was a significant and reversible decrease in channel activity (n = 5, p < 0.001). It seemed that the higher NAD+ concentration might have been interfering with Na+ binding and not blocking the channel because open probability was diminished and unitary conductance was not lowered (data not shown).
Finally, to confirm that NAD+ was modulating the channel directly, we used site-directed mutagenesis to mutate the second glycine in the phosphate-binding region “GXGXXG” to an alanine, “GXAXXG.” The second glycine, because of its missing side chain, allows for close contact of the main chain to the diphosphate of NAD(P)+. It is thought that any side chain in this position would protrude into the binding site of NAD(P)+ and disrupt cofactor binding (Bellamacina, 1996). We chose alanine accordingly to disrupt the binding site and not a bulky amino acid or charged amino acid that could have distorted overall protein conformation affecting channel functioning.
We transfected HEK-293 cells with the Slack G792A and conducted Na+ and Na+/NAD+ perfusion experiments using inside-out-excised patches from transfected cells. We found no significant change in channel activity between the Na+ perfusion and the Na+/NAD+ perfusion on channels with the mutated NAD+-binding site (Fig. 6E) (n = 5).
In this study, using immunolabeling and electrophysiological analyses, we demonstrated that KNa channels are abundantly expressed in adult rat DRG neurons. Previous work has documented that KNa channels can be recorded in small- and medium-sized DRG neurons (Bischoff et al., 1998; Gao et al., 2008), and we have now demonstrated that KNa channels are also present in large, proprioceptive DRG neurons. The Na+-dose–response relationship for KNa channels for both small and large DRG neurons was determined to be similar. We also showed that KNa channels are active at rest and are modulated by the metabolic coenzyme NAD+. The modulation by NAD+ resulted in a KNa channel Na+ dependence that falls in the physiological [Na+]i range for neurons. Because DRG neurons fire at modest frequencies (0.5–10 Hz), this modulation is highly pertinent; it will allow channels to operate under conditions where appreciable [Na+]i accumulation is not expected to normally occur. NAD+ did not modulate channels in the absence of Na+. In the presence of Na+, NAD+ did not increase KNa channel unitary conductance. It seemed that NAD+ acted at the level of Na+ binding; however, extensive single-channel analysis will be required to determine how NAD+ modulates KNa channels.
The results of our findings and those by Haimann et al. (1992) and Egan et al. (1992) underscore the idea that Na+ sensitivity as measured by excised-patch recordings of KNa channels likely does not reflect what the real Na+ sensitivity might be when channels are situated in their native environment. We presume that channel rundown is the result of NAD+ diffusing away from the patch. However, KNa channel rundown has not been observed in all neurons (Dryer, 1993), and we certainly did not see rundown in all DRG neurons. There may be other factors influencing NAD+ modulation such as the molecular composition of native KNa channels and/or channel phosphorylation status. Indeed, there is a consensus protein kinase C (PKC) phosphorylation site in the heart of the NAD+-binding site of Slack (Fig. 4), and whole-cell perforated-patch-recorded Slack currents are facilitated after PKC activation (Santi et al., 2006). It is not well established how PKC potentiates Slack channels. It could be that phosphorylation of the NAD+-binding site enhances NAD+ binding. This remains to be explored.
Beyond its role in metabolism, an integrative protective role for NAD+ in neurons is starting to emerge. Numerous investigations have shown that when neurons are injured, they respond by upregulating NAD+ biosynthetic enzymes, and NAD+ protects neurons from subsequent neurodegeneration (Wang et al., 2005; Sasaki et al., 2006; Ying, 2007). Similarly, DRG neuronal dysfunction during nerve injury and/or diabetes has been attributed to decreases in NAD+ levels (Ido, 2007; Sharma et al., 2008), which can result in neuropathic pain. Neuropathic pain is pain that arises from abnormal nervous system physiology, at times completely removed from ongoing tissue damage or inflammation (Stacey, 2005). In many instances, neuropathic pain occurs as a direct result of hyperexcitable DRG neurons. Indeed, streptozotocin, an NAD+-depleting agent and pancreatic β-cell toxin, used to induce experimental diabetes in animal models, is directly toxic to sensory neurons resulting in measurable thermal hyperalgesia before hyperglycemia ensues (Pabbidi et al., 2008). These would suggest that a direct coupling might exist between resting NAD+ levels and the intrinsic firing properties of DRG neurons.
Our data suggests that a decrease in DRG neuronal NAD+ should result in an impaired modulation of KNa channels and concomitant decreased Na+ sensitivity of KNa channels. KNa channels, however, are not the only K+ channels modulated by metabolic coenzymes. Precedence for modulation by NAD+, NADP+, or NADH has been set with a variety of K+ channels, including the K+ transport-nucleotide-binding channel, the ATP-sensitive K+ channel, and the Ca2+-activated K+ channel (Lee et al., 1994; Roosild et al., 2002; Dabrowski et al., 2003; Tipparaju et al., 2005). If KNa channels control DRG neuronal resting membrane potential (Bischoff et al., 1998) and/or limit the DAP (Gao et al., 2008), then reduced KNa activity as a result of decreased NAD+ modulation will lead to DRG neuronal hyperexcitability and spontaneous firing. As we have now shown, KNa channels are abundantly and ubiquitously expressed in DRG neurons. Since both propioceptors and nociceptors are believed to contribute to painful neuropathy (Amir et al., 1999; Ma and LaMotte, 2007), KNa channel dysfunctioning may be a likely culprit in the pathology of neuropathic pain. Alternatively, increasing KNa channel activity may also represent a therapeutic strategy for the treatment of neuropathic pain.
This study was supported by a Junior Faculty Award from the American Diabetes Association to A.B. We thank Dr. Elsa Daurignac for critical reading of this manuscript.
- Correspondence should be addressed to Dr. Arin Bhattacharjee, The State University of New York at Buffalo, 102 Farber Hall, 3435 Main Street, Buffalo, NY 14214.