Norepinephrine Facilitates Inhibitory Transmission in Substantia Gelatinosa of Adult Rat Spinal Cord (Part 1): Effects on Axon Terminals of GABAergic and Glycinergic Neurons

This study was approved by the Animal Care and Use Committee at Niigata University School of Medicine. A portion of the lumbosacral spinal cord (2.0–2.5 cm) was removed from an adult rat (9–10 weeks of age, 250–300 g) during urethane anesthesia (1.5–2.0 g/kg, intraperitoneal). The isolated spinal cord was then placed in preoxygenated ice-cold Krebs solution (2–4°C). After removal of the dura mater, all ventral and dorsal roots were cut and the pia-arachnoid membrane was removed. The spinal cord was placed in a shallow groove formed in an agar block and glued to the bottom of the microslicer stage with cyanoacrylate adhesive. After immersion in ice-cold Krebs solution, a 450- to 500-μm thick transverse slice (L3—L5 level, fig. 1A) was cut on a vibrating microslicer (DTK1500; Dosaka, Kyoto, Japan). The spinal cord slice was then placed on nylon mesh in the recording chamber and perfused with Krebs solution (10 ml/min) saturated with 95% oxygen and 5% carbon dioxide at 36 to 37°C. The Krebs solution contained NaCl, 117 mM; KCl, 3.6 mM; CaCl², 2.5 mM; MgCl², 1.2 mM; NaH²PO⁴, 1.2 mM; NaHCO³, 25 mM; and glucose, 11 mM.

Under a dissecting microscope with transmitted illumination, the SG was clearly discernible as a relatively translucent band across the dorsal horn. However, the contours of individual SG neurons cannot be visualized under these conditions; therefore, gigaohm sealing (attaching electrode to the cell with a resistance of at least 1 GΩ) was performed blindly. 14Patch pipettes were fabricated from thin-walled, borosilicate, glass-capillary tubing (1.5 mm OD; World Precision Instruments, Sarasota, FL). After establishing the whole-cell configuration, voltage-clamped neurons were held at either −70 or 0 mV for recording mEPSCs and mIPSCs, respectively (fig. 1B). The reversal potentials of EPSCs and IPSCs in SG neurons are 0 and −70 mV, respectively 15–17; therefore, only EPSCs are recorded at −70 mV and only IPSCs at 0 mV. Under these conditions, EPSCs are recorded as downward deflections in the membrane current trace, and IPSCs are recorded as upward deflections (fig. 2). The pipette solution contained Cs-sulfate (Cs²SO⁴), 110 mM; CaCl², 0.5 mM; MgCl², 2 mM; EGTA, 5 mM; HEPES, 5 mM; tetraethyl ammonium chloride (TEA), 5 mM; adenosine triphosphate–magnesium salt, 5 mM; and guanosine–5′-O-(2-thiodiphosphate) (GDP-β-S), 1 mM. The resistance of a typical patch pipette was 5–10 MΩ. In all experiments, GDP-β-S (a G-protein blocker) was added to the pipette solution to block postsynaptic norepinephrine effects mediated through G proteins. In the current study, norepinephrine and other adrenergic agonists were superfused at least 10 min after establishing the whole-cell configuration (unless otherwise stated) to avoid possible postsynaptic effects of norepinephrine. Membrane currents were amplified using an Axopatch 200A amplifier (Axon Instruments, Foster City, CA) in voltage-clamp mode. Signals were filtered at 2 kHz and digitized at 5 kHz. Data were stored with a personal computer using pCLAMP 6 software (Axon Instruments). Frequencies and amplitudes of miniature postsynaptic currents were measured using Axo-Graph 3 software (Axon Instruments). The amplitude of each postsynaptic current was measured from the initial inflection point (not from the baseline) to the peak, to avoid the effects of temporal summation on the amplitude distribution.

Drugs were applied by exchanging the perfusion solution with one containing a known drug concentration, without altering the perfusion rate and the temperature. The time necessary for the drug-containing solution to flow from the three-way stopcock to the recording chamber was approximately 3 s. Drugs used were norepinephrine (WAKO, Osaka, Japan), 6-cyano-7-nitroquinoxaline-2,3-dione (Tocris Cookson, Ballwin, MO), strychnine (Sigma, St. Louis, MO), bicuculline (Sigma), tetrodotoxin (WAKO), GDP-β-S (Sigma), chloroethylclonidine (RBI, Natick, MA), prazosin (Sigma), clonidine (Sigma), phenylephrine (Sigma), isoproterenol (Sigma), yohimbine (WAKO), 2-(2,6-dimethoxyphenoxyethyl)aminomethyl-1,4-benzodioxane (WB-4101; Sigma), γ-aminobutyric acid (GABA; WAKO), glycine (WAKO), and propranolol (Sigma).

The elementary unit of neurotransmitter release is the content of a single synaptic vesicle. The amplitude of an evoked EPSC or IPSC is some multiple of the postsynaptic current in response to the transmitter content of a single vesicle. The notion of quantal transmission originally was derived from synaptic release at the neuromuscular junction, 18and considerable evidence has accumulated that also supports the quantal hypothesis at central synapses. 19–21At many synapses, exocytosis of synaptic vesicles occurs spontaneously at a low rate, even in the absence of presynaptic stimulation. In the presence of tetrodotoxin, postsynaptic responses to spontaneously released transmitter can be detected as relatively small amplitude miniature postsynaptic currents (mPSCs) (fig. 1B).

The strength (efficacy) of synaptic transmission can be altered through modulation of both “transmitter release probability” and “postsynaptic responsiveness.” Analysis of frequency and amplitude distributions of mPSCs has been used to distinguish between pre- and postsynaptic loci of experimental manipulations, 22,23including volatile anesthetics. 24From the quantal hypothesis, only presynaptic actions can affect the probability of release. Thus, changes in the frequency of mPSCs indicate a presynaptic effect, if recruitment of latent (silent) receptors can be ruled out. 23In this study, we used GDP-β-S in the intrapipette solution to eliminate possible postsynaptic effects (changes in postsynaptic responsiveness or recruitment of silent synapses) by norepinephrine; therefore, changes in the frequency of mPSCs can be attributed only to a presynaptic effect of norepinephrine. Alterations in mPSC peak amplitude can be explained only by changes in postsynaptic responsiveness. The mean (or median) amplitude can also be altered by changes in postsynaptic responsiveness. However, the shape of the amplitude distribution can be skewed by the existence of synapses that are unaffected by the treatment, resulting only in an apparent  change in the mean (median) amplitude.

Numeric data are presented as the mean ± SD (unless otherwise stated). Analyses of the modulation of the frequency of mPSCs were performed using a paired t  test. The effects of selective agonists and antagonists were analyzed using the one-way analysis of variance, and statistical significance was further evaluated using the Scheffé test for post hoc  comparison. The Kolmogorov–Smirnov test was used to compare the effect norepinephrine on the amplitude distribution of postsynaptic currents. Differences for which P < 0.05 were considered significant and are indicated by asterisks in the figures. Dose–response data were fitted using logistic equations. Curve fitting was accomplished using Origin 4.1 software (Microcal Software, Northampton, MA).

Whole-cell patch-clamp recordings were made from 109 SG neurons. In the presence of tetrodotoxin (1 μM), all SG neurons tested exhibited mEPSCs and mIPSCs. At holding potentials of −70 mV, only mEPSCs were observed (figs. 2A and D; top). 14,25At holding potentials more positive than −60 mV, mEPSCs and mIPSCs both were observed (fig. 2B). However, at 0 mV, only mIPSCs could be observed (fig. 2C). 25,26mEPSCs observed at −70 mV were completely blocked by 6-cyano-7-nitroquinoxaline-2,3-dione, suggesting that they were mediated by non-NMDA (AMPA/kainate) receptors. Two distinct types of mIPSCs could be distinguished based on the decay time course (fig. 2C). 14,16,17One type of mIPSC had a short duration (10–20 ms) and was antagonized by a glycine receptor antagonist: strychnine (1–2 μM). The other had a relatively long duration (50–100 ms) and was antagonized by a GABA^Areceptor antagonist, bicuculline (10 μM), suggesting that the two types of mIPSCs were mediated by glycine and GABA^Areceptors, respectively. Shortly after establishing the whole-cell patch-clamp configuration, norepinephrine elicited outward currents at −70 mV in the majority of SG neurons (18 of 24 cells;fig. 2D), in agreement with previous reports. 13This effect was mimicked by clonidine (10 μM) and blocked by yohimbine (0.5 μM) (data not shown), indicating that the outward currents were mediated by activation of α²receptors. These currents disappeared when diffusion of GDP-β-S was complete (a few minutes after rupturing the patch of membrane;fig. 3A), suggesting that the outward currents were mediated by G-protein–coupled potassium (K⁺) currents, as reported previously. 27 

In 26 SG neurons, membrane potentials were clamped to −70 mV, and the effect of norepinephrine on the frequency of mEPSCs was evaluated. The baseline frequency of mEPSCs was 20.9 ± 15.2 Hz (n = 26; range, 3.7–64.5 Hz). The frequency of mEPSCs was not significantly affected by norepinephrine (10–100 μM, 92 ± 19% of control for norepinephrine 20 μM;P = 0.08, n = 17, paired t  test;figs. 2D and E). In sharp contrast, norepinephrine markedly increased the frequency of mIPSCs (figs. 2D and F). The baseline frequency of mIPSCs was 2.9 ± 1.5 Hz (n = 59; range, 0.5–6.2 Hz). Norepinephrine (10–100 μM) increased the frequency of IPSCs in 55 of 59 SG neurons tested (459 ± 167% of control for norepinephrine 20 μM;P < 0.0001, n = 18; paired t  test;fig. 2F). In 8 of 18 cells in which norepinephrine elicited outward currents at −70 mV, the membrane potentials were also clamped at 0 mV, and the effect of norepinephrine on mIPSCs was evaluated. In all eight cells, norepinephrine elicited an increase in the frequency of mIPSCs. The effect of norepinephrine on mIPSC frequency reached a steady state within 2 min and showed no evidence of desensitization over at least 30 min. After norepinephrine washout, IPSC frequency slowly returned to baseline (recovery time was usually > 5 min).

The effect of norepinephrine was mimicked by the α¹agonist phenylephrine (10–60 μM; n = 11), but not by the α²agonist clonidine (10–40 μM; n = 10) or the β agonist isoproterenol (40 μM; n = 6) (fig. 3B). The α¹-receptor antagonists prazosin (0.5 μM; n = 5) and WB-4101 (0.5 μM; n = 6) reversibly antagonized the norepinephrine effect; the α^1Band α^1Dantagonist chloroethylclonidine (10 μM; n = 6) had no significant effect (fig. 3C). In addition, the α²- and β-receptor antagonists yohimbine (1 μM; n = 6) and propranolol (1 μM; n = 5), respectively, were without effect (fig. 3C).

Because mIPSCs consist of glycinergic and GABAergic components, we evaluated which type of mIPSCs were facilitated by norepinephrine. In the presence of strychnine (2 μM), norepinephrine increased the remaining mIPSCs (n = 12). These mIPSCs were completely abolished by the simultaneous application of strychnine (2 μM) and bicuculline (10 μM), confirming that facilitated mIPSCs were GABAergic (fig. 4A). Conversely, mIPSCs recorded in the presence of bicuculline were facilitated by norepinephrine (n = 10;fig. 4B). Thus, norepinephrine increased the frequency of both GABAergic and glycinergic mIPSCs.

Facilitation of GABAergic and glycinergic mIPSC frequency by norepinephrine was concentration dependent (fig. 5). Concentration–response curves for GABAergic (n = 7) and glycinergic (n = 8) mIPSCs were well-fitted by logistic equations. The estimated EC⁵⁰values for the effect of norepinephrine on GABAergic (29.5 ± 12.2 μM) and glycinergic (38.8 ± 37.3 μM) IPSCs were similar, as were the Hill coefficients (GABAergic: 2.8 ± 1.4; glycinergic: 2.8 ± 1.5). Norepinephrine facilitates GABAergic mIPSCs with an approximately twofold greater efficacy than for glycinergic mIPSCs (GABAergic: 1,094 ± 342%; glycinergic: 541 ± 403%).

Amplitude histograms were constructed for each set of GABAergic and glycinergic mIPSCs. In all cells tested, amplitude distribution analyses showed a slight but statistically significant norepinephrine-induced increase in the median amplitude of GABAergic (n = 7) and glycinergic (n = 5) mIPSCs (fig. 6). However, the peak amplitude of both types of mIPSCs was not increased by norepinephrine. The effect of norepinephrine was particularly prominent between 10 and 15 pA for GABAergic and glycinergic mIPSCs. This change in the amplitude distribution is clearly shown in the cumulative histogram (figs. 6B and D). In the presence of norepinephrine, the relative frequency curves were significantly shifted to the right in all cells tested.

The norepinephrine-induced changes in amplitude distributions might be attributable to an increase in postsynaptic sensitivity to GABA and glycine under conditions in which GDP-β-S diffusion was incomplete. Alternatively, an unknown mechanism not mediated by G protein might mediate a postsynaptic effect of norepinephrine. To rule out these possibilities, we evaluated whether norepinephrine affected postsynaptic sensitivity to exogenously applied GABA and glycine under our recording conditions (i.e. , with an internal solution containing GDP-β-S). Bath-applied GABA and glycine elicited outward currents at a holding potential of 0 mV (figs. 7A and B). Although the frequency of GABAergic or glycinergic mIPSCs was increased by norepinephrine, the amplitude of GABA- or glycine-evoked (n = 6 for both) outward currents was not significantly affected by norepinephrine (fig. 7C). We also tested whether norepinephrine affected the kinetics of mIPSCs. The kinetics of GABAergic (n = 3) and glycinergic (n = 3) mIPSCs before and during norepinephrine application were identical, as shown by the superimposed averaged records (fig. 7D). Therefore, norepinephrine does not appear to alter the postsynaptic responsiveness of SG neurons to GABA and glycine.

We showed that norepinephrine increases the frequency of mIPSCs in the majority (> 90%) of SG neurons tested, without any significant effect on postsynaptic responsiveness. In contrast, norepinephrine did not change or slightly decreased the frequency of mEPSCs. These data indicate that norepinephrine selectively facilitates the quantal release of inhibitory transmitters from presynaptic terminals of inhibitory interneurons. This study shows that norepinephrine increases the frequency of both GABAergic and glycinergic mIPSCs. This may not be surprising in light of the observation that GABA and glycine can be colocalized, at least in a subpopulation of dorsal horn neurons, and cotransmission of GABA and glycine can occur. 28However, norepinephrine facilitated GABAergic mIPSCs with greater efficacy than glycinergic mIPSCs. The reason for this difference has yet to be determined, but it may result from a differential distribution of α¹receptors on GABAergic and glycinergic terminals.

The facilitatory effect of norepinephrine was mediated via  adrenergic α¹receptors, because phenylephrine (an α¹-receptor agonist) mimicked norepinephrine, and prazosin and WB-4101 (α¹-receptor antagonists) inhibited the norepinephrine effect. Thus, GABA- and glycine-containing neurons are endowed with α¹receptors on the axon terminals, and activation of these receptors enhances GABAergic and glycinergic inhibitory transmission in the SG. Currently, at least three native α¹-adrenergic receptor subtypes (α^1A, α^1B, and α^1D) have been identified pharmacologically (α^1Cis lacking). 29Both α^1Band α^1Dreceptors are characterized by a high sensitivity to chloroethylclonidine. The action of norepinephrine on mIPSC frequency was highly sensitive to WB-4101 (an α^1A-receptor antagonist), but resistant to a relatively high concentration (10 μM) of chloroethylclonidine (an α^1B- and α^1D-receptor antagonist). Therefore, the most likely α¹-adrenergic receptor responsible for norepinephrine action is the α^1Asubtype.

The norepinephrine-induced increase in mIPSC frequency could be secondary to changes in the postsynaptic neuron. If norepinephrine produces an increase in the sensitivity of postsynaptic GABA^Aand glycine receptors, it could potentiate a population of subliminal mIPSCs, such that they would become visible, yielding an apparent increase in mIPSC frequency. Furthermore, if norepinephrine results in the recruitment of “silent” inhibitory synapses in a manner analogous to that described for excitatory synapses, 23,30,31the apparent mIPSC frequency would increase. It is unlikely that these types of postsynaptic changes mediate the facilitatory effects of norepinephrine because all known adrenergic receptors are coupled to ion channels via  G proteins, and G proteins were inhibited in postsynaptic (recorded) neurons by GDP-β-S. In addition, the amplitude of outward currents elicited by exogenously applied GABA and glycine was not altered by norepinephrine, nor were kinetics of mIPSCs. Together, these observations indicate that norepinephrine does not alter postsynaptic responsiveness to GABA and glycine. Therefore, it is surprising that the amplitude–distribution analysis showed a significant norepinephrine-induced increase in the median amplitude of GABAergic and glycinergic mIPSCs and a shift in the cumulative histogram curves to the right. This most likely results, however, from a relatively selective increase in a subpopulation of mIPSCs with amplitudes between 10 and 15 pA (figs. 6A and C). SG neurons receive numerous types of inhibitory inputs from interneurons in the SG and the surrounding laminae (laminae I, III, IV, and so forth). 32The effect of norepinephrine may be selective for a subpopulation of these inputs. If norepinephrine facilitates quantal release, especially in a population of inhibitory synapses that generate mIPSCs with an amplitude of 10–15 pA, the amplitude distribution could be skewed. Although norepinephrine facilitates inhibitory transmitter release via  presynaptic α¹receptors, it may not influence all presynaptic terminals of inhibitory interneurons in the SG in a homogeneous fashion. Alternatively, temporal summation of mIPSCs might result in an increase in the median amplitude of mIPSCs. We measured the amplitude of mIPSCs from the initial inflection point (not from the baseline) to avoid effects of temporal summation on the amplitude distribution. However, in cases in which two (or more) quantal events occur at exactly the same moment in time, they cannot be distinguished; therefore, they may have been measured as a single mIPSC. This type of error is more likely to occur during conditions in which the frequency of mIPSCs is increased. Nonetheless, we believe this to be rare because carbachol, a muscarinic agonist, also increased the frequency of mIPSCs but did not affect the amplitude distribution. 26 

Almost all norepinephrine-containing terminals in the dorsal horn of the spinal cord are supraspinal in origin. 33Currently, our knowledge of how these descending pain-control pathways inhibit nociceptive transmission at the spinal level is not defined clearly, but there are several possible mechanisms. First, activation of noradrenergic descending systems releases norepinephrine, which can directly hyperpolarize a proportion of SG neurons that may be excitatory interneurons in the pain pathway (postsynaptic inhibition). 13Second, norepinephrine can inhibit excitatory transmitter (glutamate, substance P, calcitonin gene-related peptide [CGRP], and so forth) release from primary afferent terminals or presynaptic terminals of excitatory interneurons. 34,35However, our data do not support this type of mechanism in the SG. Third, norepinephrine could depolarize inhibitory interneurons that contain GABA, glycine, or other inhibitory peptides. Iontophoretic application of norepinephrine near nociceptive dorsal horn neurons generally inhibits background activity of these cells and the responsiveness to excitatory amino acids. 36–38This inhibition most likely results from α²-receptor activation, which increases K⁺conductance, thereby evoking a membrane hyperpolarization. 13However, norepinephrine (and brain stem stimulation) also has been reported to produce excitatory effects. 38–40The neurons excited by iontophoretically applied norepinephrine and electrical stimulation of the periaqueductal gray were low-threshold cells, possibly inhibitory interneurons that synapse onto high-threshold and wide-dynamic-range neurons. 40In the accompanying article, we directly address whether norepinephrine depolarizes inhibitory interneurons that synapse onto SG neurons.

Here, we describe a distinct type of mechanism: norepinephrine acts at presynaptic axon terminals of GABAergic and glycinergic interneurons to facilitate inhibitory transmitter release. We also demonstrated that some SG neurons display both an α²-receptor–mediated outward current and an α¹-receptor–mediated increase in mIPSC frequency (fig. 2D). Because 80% of SG neurons are hyperpolarized (or display an outward current) via α²-receptor activation, 13and more than 90% show an increase in mIPSC frequency (current study), it is likely that the excitability of most SG neurons is inhibited by exogeneously applied norepinephrine (and probably by epinephrine) through α²- and α¹-receptor–mediated mechanisms. Finally, it should be noted that norepinephrine may change postsynaptic sensitivity to GABA and glycine. In the current study, we blocked postsynaptic effects of norepinephrine using GDP-β-S to evaluate the presynaptic action. However, in intact SG neurons, norepinephrine may increase postsynaptic responsiveness to inhibitory neurotransmitters by G protein–coupled mechanisms. 41 

γ-Aminobutyric acid and glycine have been shown to be present in cell bodies and terminals in the superficial dorsal horn 42and are thought to be involved in spinal antinociception. 14,16,17Electrical stimulation of primary afferent fibers evokes mono- and polysynaptic excitatory postsynaptic potentials (EPSPs) that are then augmented in both amplitude and duration by perfusion with GABA^Aand glycine receptor antagonists. 16,17In addition, intrathecal administration of GABA^Aagonists or glycine is antinociceptive, 43,44and antagonists of GABA^Aand glycine receptors produce allodynia and hyperalgesia in animals. 45,46These observations suggest that GABA- and glycine-containing interneurons are responsible for discontinuing evoked excitatory responses. The facilitatory effect of norepinephrine on inhibitory interneurons may decrease excitability of dorsal horn neurons, which in turn would increase the threshold for transmission of noxious information.

As reviewed by Willis and Coggsehall, 32SG neurons receive substantial primary afferent input from nociceptive Aδ and C fibers (these fibers can, however, also form synapses directly onto projection neurons). In addition to nociceptive inputs, the SG receives descending adrenergic input from a number of pontine nuclei. 47With respect to output from this region, the majority of SG neurons are local interneurons and do not project to the thalamus, 32although there are a few exceptions. 48The main projections of SG neurons are to lamina I and to deep dorsal horn neurons, with cell bodies in laminae IV and V, where the projection neurons to the thalamus are located. 32Unfortunately, in the current study, the phenotype of the SG neurons from which we recorded is unknown; some SG neurons might produce excitatory, and others inhibitory, effects on projection neurons in laminae IV and V. In the companion article, we address whether α¹-receptor activation inhibits transmission of nociceptive inputs to deep dorsal horn neurons.

Although the vasoconstrictive effects of epinephrine and phenylephrine in the spinal cord have not been established, these drugs are sometimes added to local anesthetics in spinal anesthesia to prolong the duration of analgesia. 5,6It has been reported that subarachnoid epinephrine and phenylephrine do not affect spinal cord blood flow. 49,50Another study suggested that at a relatively high dose, phenylephrine decreased blood flow. 51Furthermore, it has been reported that epinephrine and phenylephrine do not significantly affect the clearance of local anesthetics from the subarachnoid space. 9,10Thus, it remains controversial whether the vasoconstrictive actions of epinephrine or phenylephrine contribute to the prolongation of analgesia in spinal anesthesia. The mechanism of epinephrine action in spinal anesthesia can be accounted for, at least in part, by activation of α²receptors. 7,8,11As discussed previously, exogenously applied epinephrine (a mixed α¹and α²agonist) can decrease the excitability of SG neurons via α¹and α²receptors. Interestingly, phenylephrine (a pure α¹agonist) can also enhance spinal anesthesia. Considering the evidence that phenylephrine does not affect the clearance of local anesthetics, the action of phenylephrine in spinal anesthesia must be caused by a direct action on spinal neurons via α¹receptors. The presynaptic α¹action reported here may account for the prolongation of analgesia by α¹agonists in spinal anesthesia.

The authors thank Dr. K. A. Moore for critically reading the manuscript.