2-APV

AN N-METHYL-D-ASPARTATE RECEPTOR MEDIATED LARGE, LOW-FREQUENCY, SPONTANEOUS EXCITATORY POSTSYNAPTIC CURRENT IN NEONATAL RAT SPINAL DORSAL HORN NEURONS

Abstract—Examples of spontaneous oscillating neural activ- ity contributing to both pathological and physiological states are abundant throughout the CNS. Here we report a spontane- ous oscillating intermittent synaptic current located in lamina I of the neonatal rat spinal cord dorsal horn. The spontaneous oscillating intermittent synaptic current is characterized by its large amplitude, slow decay time, and low-frequency. We dem- onstrate that post-synaptic N-methyl-D-aspartate receptors (NMDARs) mediate the spontaneous oscillating intermittent synaptic current, as it is inhibited by magnesium, bath-applied d-2-amino-5-phosphonovalerate (APV), or intracellular MK-801. The NR2B subunit of the NMDAR appears important to this phenomenon, as the NR2B subunit selective NMDAR an- tagonist, alpha-(4-hydroxphenyl)-beta-methyl-4-benzyl-1-piper- idineethanol tartrate (ifenprodil), also partially inhibited the spontaneous oscillating intermittent synaptic current. Inhibition of spontaneous glutamate release by the AMPA/kainate recep- tor antagonist 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) or the mu-opioid receptor agonist [D-Ala2, N-Me-Phe4, Gly5] en- kephalin-ol (DAMGO) inhibited the spontaneous oscillating in- termittent synaptic current frequency. Marked inhibition of spontaneous oscillating intermittent synaptic current frequency by tetrodotoxin (TTX), but not post-synaptic N-(2,6-dimethyl- phenylcarbamoylmethyl)triethylammonium bromide (QX-314), suggests that the glutamate release important to the spontane- ous oscillating intermittent synaptic current is dependent on active neural processes. Conversely, increasing dorsal horn synaptic glutamate release by GABAA or glycine inhibition in- creased spontaneous oscillating intermittent synaptic current frequency. Moreover, inhibiting glutamate transporters with threo-beta-benzyloxyaspartic acid (DL-TBOA) increased sponta- neous oscillating intermittent synaptic current frequency and decay time. A possible functional role of this spontaneous NMDAR–mediated excitatory postsynaptic current in modulat- ing nociceptive transmission within the spinal cord is discussed.

Key words: spontaneous oscillation, voltage-clamp, electro- physiology, in vitro.

Spontaneous oscillating neural activity, which contributes both to pathological states such as epilepsy (see Jefferys, 2003 for review) and to physiological processes necessary to healthy functioning, such as sleep (see Steriade and Timofeev, 2003 for review) and memory storage (see Dra- guhn et al., 2000 for review), has been observed in several regions of the CNS, including: the brain stem (Benardo and Foster, 1986; Pettigrew et al., 1988; Ishimatsu and Wil- liams, 1996; Yajima and Hayashi, 1999; Parkis et al., 2003; Verdier et al., 2004); cerebellum (Lawrie et al., 1993; Lang, 2001); olfactory bulb (McQuiston and Katz, 2001; Nusser et al., 2001; Puopolo and Belluzzi, 2001); hippocampus (Ogura et al., 1987; Schneiderman and MacDonald, 1989; Leung and Yim, 1991; Soltesz and Deschenes, 1993; Traub et al., 1994; Cobb et al., 1995; Strata et al., 1997; Garaschuk et al., 1998; Staley et al., 1998; Bacci et al., 1999; Fischer et al., 1999; Khazipov et al., 2001; Lamsa et al., 2000; Palva et al., 2000; Maier et al., 2003) and spinal cord (Legendre et al., 1985; Spanswick and Logan, 1990; Chub,1991; Streit, 1993; Hochman et al., 1994; Bracci et al., 1996a,b; Lewis and Faber, 1996; Lidierth and Wall, 1996; Ballerini et al., 1997, 1999; Keefer et al., 2001; Demir et al., 2002; Rozzo et al., 2002).

Within the spinal cord, spontaneous oscillating neural activity has been identified in the dorsal and ventral roots (Chub, 1991; Bracci et al., 1996b; Lidierth and Wall, 1996 Ballerini et al., 1997), motoneurons (Streit, 1993; Hochman et al., 1994; Bracci et al., 1996a,b; Ballerini et al., 1997), ventral horn interneurones (Ballerini et al., 1999; Rozzo et al., 2002) intermediolateral cell nuclei (Spanswick and Lo- gan, 1990), dorsal horn (Ruscheweyh and Sandkühler, 2003, 2005; Asghar et al., 2005), as well as in different cultured spinal cord preparations (Legendre et al., 1985; Lewis and Faber, 1996; Keefer et al., 2001). Spontaneous neural oscillations have been identified extracellularly in the form of rhythmic bursts of action potentials (Legendre et al., 1985; Spanswick and Logan, 1990; Chub, 1991; Streit, 1993; Hochman et al., 1994; Bracci et al.,1996a,b; Lidierth and Wall, 1996; Ballerini et al., 1997, 1999; Keefer et al., 2001; Demir et al., 2002; Rozzo et al., 2002) and intracellularly in the form of spontaneous intermittent syn- aptic currents (Ballerini et al., 1999; Ruscheweyh and Sandkühler, 2003, 2005; Asghar et al., 2005). Pharmaco- logical manipulations of these spontaneous neural oscilla- tions have yielded several common features, including inhibition by tetrodotoxin (TTX) (Ballerini et al., 1999; Rozzo et al., 2002), glutamate receptor antagonists (Keefer et al., 2001), magnesium (Chub, 1991; Keefer et al., 2001; Rozzo et al., 2002), and GABA and glycine (Streit, 1993; Bracci et al., 1996b; Ballerini et al., 1999; Keefer et al., 2001). These properties point to neuronal excitatory network-driven oscillations that, under basal conditions, are suppressed by inhibitory neurotransmis- sion.

Here we report a novel large amplitude, long duration spontaneous oscillating excitatory postsynaptic current (sEPSCosc) in lamina I of the neonatal rat dorsal horn of the spinal cord. We further characterize this sEPSCosc with regard to its pharmacological modulation in comparison to other patterns of rhythmic synaptic activity in the CNS.

EXPERIMENTAL PROCEDURES

All experimental methods were approved by the University of Washington animal care and use committee and adhere to the Committee for Research Ethical Issues guidelines published by the International Association for the Study of Pain. All efforts were made to minimize both the number of rats used in the experiments and the distress involved in the experimental procedures.

Spinal cord slice preparation

Ten- to 17-day-old Sprague–Dawley rats were anesthetized with halothane, and a laminectomy was performed from mid-thoracic to low lumbar levels. Cold buffer (220 mM sucrose, 3 mM KCL, 8 mM MgCl2, 1.25 mMNaH2PO4, 26 mM NaHCO3, and 10 mM D-glucose bubbled with 95% O2–5% CO2) was placed on the spinal cord, decapitation was performed, and the cord quickly removed and placed in 35 °C, 2% low melting point agar. The spinal cord embedded in agar was then chilled, blocked, and transversely sliced (300 – 400 µm) using a Vibratome (Ted Pella Inc., Redding, CA, USA). Spinal cord slices were held in O2-bubbled modified artificial cerebrospinal fluid (aCSF) (113 mM NaCl, 3 mM KCl, 1 mM NaH2PO4, 25 mM NaHCO3, 11 mM glucose, 2 mM CaCl2, and, except where otherwise indicated, bicuculline methiodide (bicuculline) (10 µM) and strychnine hydrochloride (strychnine) (1 µM)) for approximately 1 h prior to electrophysiological record- ings. MgCl2 concentration was varied (0 –2 mM) in different experiments.

Whole-cell voltage clamp

Spinal cord slices were transferred to a Plexiglas recording cham- ber and held in place by a harp constructed of platinum wire in the shape of a “U” with nylon threads glued across the open end of the U. A Zeiss Axioskop FS microscope with a 770 (±40) nm band- pass filter was used to identify specific cells from which to record. A CCD camera (Dage-MTI) projected an image of the slice onto a black and white monitor (Sony), which was then used to identify lamina I dorsal horn cells (identified as cells no farther than 10 µm from the dorsal edge of the slice). A glass recording electrode (approximately 15 Mohm resistance when filled with 125 mM KMeSO4, 8 mM NaCl, 10 mM HEPES, 2 mM MgATP, 0.5 mM NaGTP, and 5 mM EGTA at pH 7.3 and approximately 285 mOsm) was advanced using a micro-manipulator (Sutter) until whole cell configuration was obtained using suction on the tar- geted cell. Except where otherwise indicated, whole cell voltage clamp was maintained at a resting potential of —70 mV using an Axopatch 200B amplifier (Axon Instruments) and Pulse software (HEKA). Initial series resistance and capacitance were noted and a change of greater than 20% was used as the upper limit for inclusion of data in analysis.

Quantification and analysis of drug effects

Baseline data were collected following the first 5 min post-whole cell configuration. After 5 min of baseline data collection, drug was added to the perfusate at a flow rate of 1–2 ml/min and recordings were obtained for a minimum of 10 min. Except where otherwise indicated, drug was subsequently washed out of the slice by application of the original perfusate at the same flow rate.

Mini-Analysis software (Synaptosoft) was used to analyze the frequency, maximal amplitude, decay time (measured as time for response amplitude to decrement from 90% to 37% of maximum), and rise time (measured as time for response amplitude to in- crease from 10% to 90% of maximum)of the sEPSCs. Drug effects were quantified as the mean response during the last 5 min of drug administration compared with the mean response during the 5 min baseline period. Drug washout effects were quantified as the mean response during each 5 min epoch of washout com- pared with baseline. Data were transformed [(post-drug/pre- drug)×100] to illustrate the % change from baseline response. Data were analyzed using between-subjects and/or within-sub- jects analyses of variance (GB-STAT for Macintosh) as appropri- ate. Fisher’s LSD test (Carmer and Swanson, 1973) was used for post hoc comparisons. A statistically significant effect was based on a P value of less than 0.05.

Drugs

Bicuculline, strychnine, 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), d-2-amino-5-phosphonovalerate (APV), MK-801, N-(2,6- dimethylphenylcarbamoylmethyl)triethylammonium bromide (QX- 314), and alpha-(4-hydroxphenyl)-beta-methyl-4-benzyl-1-piper- idineethanol tartrate (ifenprodil) were purchased from Sigma-Al- drich. DL-Threo-beta-benzyloxyaspartic acid (DL-TBOA) was purchased from Tocris Cookson, Inc. TTX, Fugu sp was pur- chased from CalBioChem. Naloxone and [D-Ala2, N-Me-Phe4, Gly5] enkephalin-ol (DAMGO) were gifts from the National Insti- tute of Drug Abuse (NIDA) Drug Supply System. Except where otherwise indicated, all drugs were added to the perfusate in a 1:1000 dilution.(P=0.03) and increased the rate of the sEPSCosc (mean frequency=0.03±0.005 Hz) (P=0.0006) (Fig. 1D–F).
Throughout the rest of this report, we use the term sEPSCosc to refer to this large, slow, intermittent sEPSC that is present in 1 mM and 0 mM Mg2+ aCSF.

The sEPSCosc was so large and of such long duration that it was initially interpreted as a loss of voltage clamp, except that it repeatedly returned to baseline. By increas- ing the recording time and using a 0 Mg2+ bath it be- came apparent that the sEPSCosc occurred intermit- tently at a low frequency (Fig. 2A). In current clamp, a rhythmic series of action potentials appeared at a similar rate as the sEPSCosc (Fig. 2B).

Post-synaptic N-methyl-D-aspartate receptor (NMDAR) mediation of the sEPSCosc

Because of its very long rise time (Perouansky and Yaari, 1993) and its block with Mg2+, we investigated the role of NMDARs in mediating the sEPSCosc. These receptors are well known to be blocked by Mg2+ at resting membrane potentials. The competitive NMDAR antagonist APV (25 µM) blocked the sEPSCosc in both the 1 mM Mg2+- containing aCSF (P=0.024) (Fig. 3A) and the 0 mM Mg2+ aCSF (P=0.036) (Fig. 3B). NMDAR mediation of sEPSCs in lamina I is unusual. In 2 mM Mg2+ perfusate, APV.

RESULTS

Characterization of the sEPSCosc and its sensitivity to Mg2+

An unusually large and slow sEPSC was observed to occur intermittently in lamina I of the spinal dorsal horn of the neonatal rat when the bath Mg2+ concentration was decreased from 2 to 1 mM. In comparison to the sEPSCs present in the aCSF containing 2 mM Mg2+, the sEPSC that occurred in aCSF containing 1 mM Mg2+ was larger (mean maximal amplitude=462±169 pA vs. 50±2 pA, n=7 cells), had a longer rise time (mean rise time=110±6 ms vs. 1±0. 2 ms, n=7 cells), was slower to decay (mean decay time=97±27 ms vs. 3±0.3 ms, n=8 cells), and had a much lower frequency (mean frequency=0.0012±0.0003 Hz vs. 2±0.5 Hz, n=8 cells) (Fig. 1A and B). Complete removal of Mg2+ from the aCSF (e.g. Fig. 1C) significantly potentiated the amplitude (mean maximal amplitude=1296±124 pA) usually has no effect on amplitude or frequency of sEPSCs or mEPSCs (J. Zeng, unpublished observations). In contrast, CNQX (10 µM), the AMPA/kainate glutamate receptor antag- onist, completely eliminates the sEPSCs and mEPSCs in the same preparation (Terman et al., 2001). In this report, CNQX (10 µM) reversibly decreased the sEPSCosc frequency (P=0.001) (Fig. 3E) but did not eliminate it. Although CNQX only minimally inhibited sEPSCosc amplitude, a significant decrease in amplitude was observed during washout of CNQX (P=0.0001) (Fig. 3C), perhaps due to rundown of the sEPSCosc in this study. CNQX had no significant effect on the sEPSCosc decay time (P=0.130) (Fig. 3D). Taken together, these data indicate that the sEPSCosc is primarily mediated by the NMDAR and that CNQX inhibits the sEPSCosc fre- quency, not by blocking AMPA and/or kainate receptors on the cells we are studying, but by decreasing glutamate re- lease at pre-synaptic or polysynaptic sites.

After having demonstrated NMDAR mediation of the sEPSCosc, we investigated whether pre-or-post-synaptic

NMDARs were involved in its generation. Because the sEPSCosc occurred at a higher rate in 0 Mg2+ perfusate and was thus more readily investigated, the remaining experiments were done in 0 Mg2+ aCSF. We added the non-competitive NMDAR antagonist MK-801 (1 mM) to the intracellular pipette solution to preferentially block post- synaptic NMDARs. This method has previously been used to selectively block post-synaptic NMDARs (see Arvanian and Mendell, 2001; Garraway et al., 2003). In our hands, MK-801 in the intracellular pipette solution greatly attenu- ated (approximately 90% reduction) maximal EPSC ampli- tudes from bath-applied NMDA (50 µM in 0 mM Mg2+, n=9 cells) without having any effect on sEPSC frequency, amplitude, or decay time in 2 mM Mg2+ (n=13 cells) (data not shown). In the present experiment, blocking post-syn- aptic NMDARs by intracellular MK-801 (1 mM) dialysis decreased the amplitude (P=0.012) (Fig. 4A,B) and decay time (P=0.012) (Fig. 4C), but did not affect the frequency of the sEPSCosc (P=0.289) (Fig. 4A, D). The inhibition of sEPSCosc amplitude is characteristic of post-synaptic block suggesting that the applied MK-801 is blocking post- synaptic NMDARs which are important in mediating the sEPSCosc. On the other hand, the absence of an effect of MK-801 on sEPSCosc frequency supports the notion that post-synaptic NMDARs are activated by glutamate from pre-synaptic stores.

Neuronal pre-synaptic glutamate release mediates the sEPSCosc

To address the question of whether the sEPSCosc is de- pendent on active neural transmission, we recorded the sEPSCosc in the presence of the sodium channel blocker TTX (0.2 µM). Bath-application of TTX dramatically inhib- ited sEPSCosc frequency (P=0.003, n=5 cells) (Fig. 5C). A mean of only one (±1) sEPSCsosc was recorded over a 5 min period during TTX administration compared with a mean of 10 (±2) sEPSCsosc in those cells pre-TTX admin- istration. We believe that the dramatic TTX-induced inhi- bition of sEPSCosc frequency indicates a neural source of glutamate in the generation of the sEPSCsosc. This does not discount a possible role for glia in mediating this phe- nomenon, as glia have recently been implicated in both the release of glutamate (Benz et al., 2004) and the modula- tion of neural excitability (Hayden, 2001). However, there is no evidence as yet that TTX can directly block glial release of glutamate (c.f. Morita et al., 2003), pointing to the importance of voltage-dependent sodium channel-me- diated neural transmission in producing sEPSCsosc.

We went on to add the lidocaine derivative QX-314 (1 mM) to the intracellular pipette solution to selectively block the voltage dependent
Na+ channels in the recorded cell (see also Connors and Prince, 1982; Hu et al., 2002). Intracellular administration of QX-314 did not significantly inhibit sEPSCosc frequency (P=0.937) (Fig. 6A, D) or de- cay time (P=0.801) (Fig. 6C). Intracellular QX-314 did, however, significantly inhibit the sEPSCosc maximal ampli- tude (P=0.049) (Fig. 6A, B). The inhibitory effect of both TTX and intracellular QX-314 on amplitude suggests that post-synaptic voltage dependent Na+ channels contribute to the NMDAR-mediated sEPSCsosc in these studies, per- haps due to space clamp limitations imposed by our whole cell voltage clamp methods (that is, an inability to voltage clamp the entire cell at —70 mV). Nonetheless, the failure of intracellular QX-314 to inhibit sEPSCosc frequency dem- onstrates that the Na+ channel-dependent active neuro- transmission that is so important in mediating sEPSCosc in our TTX experiments takes place in a cell (or cells) pre- synaptic to the cells from which we record.

If pre-synaptic neural activity releases glutamate which binds to post-synaptic NMDARs to produce sEPSCsosc, then compounds that inhibit and potentiate neural gluta- mate release should inhibit and potentiate, respectively, the sEPSCosc frequency. For example, it is likely that CNQX inhibits sEPSCosc frequency by decreasing neural excitability in the dorsal horn and thereby decreasing glu- tamate release. Similarly, GABA and glycine are well known to inhibit glutamatergic-mediated excitatory neural activity in the spinal dorsal horn (e.g. Yoshimura and Nishi, 1995). In the next study, we examined the effects of block- ing GABA and glycine inhibition on sEPSCsosc. Bath-ap- plication of bicuculline (10 µM), a GABAA receptor antag- onist, and strychnine (1 µM), a glycine antagonist, revers- ibly potentiated the sEPSCosc frequency by approximately four-fold (pbic=0.003, pstrych=0.013) (Fig. 7C). Spontane- ous EPSCosc decay time was not significantly affected by either bicuculline (P=0.952) or strychnine (P=0.079) (Fig. 7B). In these studies, bicuculline also reversibly potential glutamate transporters usually work together to keep glu- tamate concentrations under tight control (Asztely et al., 1997; Huang, 1998; Kullmann and Asztely, 1998; Mitchell and Silver, 2000; Piet et al., 2004). We examined whether modulating glutamate concentrations by inhibiting gluta- mate transporters would affect the sEPSCosc. DL-TBOA is a non-selective glutamate transporter inhibitor which has been shown to increase synaptic glutamate concentrations (Diamond, 2001). We applied TBOA (100 µM; 1:100 dilu- tion) and found that sEPSCosc frequency showed a signif- icant reversible potentiation (P=0.031) (Fig. 9A,D). Spon- taneous EPSCosc decay time did not show a significant reversible potentiation during the 10 min of TBOA applica- tion. However, the potentiation continued to increase dur- ing initial TBOA washout such that after 5 min of washout there was a potentiation in decay time which was signifi- cantly reversed 10 min later (P=0.028) (Fig. 9A, C). TBOA had no significant effect on sEPSCosc amplitude (P= 0.266) (Fig. 9A, B). Potentiation of the sEPSCosc fre- quency and decay time by TBOA parallels the findings of others (e.g. Diamond, 2001) demonstrating the effects of high concentrations of glutamate in the synaptic cleft.

High synaptic concentrations of glutamate can cause glutamate to “spill over” onto neighboring synapses and onto extrasynaptic membranes (Asztely et al., 1997; Huang, 1998; Kullmann and Asztely, 1998; Mitchell and Silver, 2000; Piet et al., 2004). The NR2B receptor subunit of the NMDAR is often implicated in glutamate spillover effects, probably due to the preferentially extrasynaptic are depicted for sEPSCosc amplitude (A), decay time (B), and fre- quency (C) during bath-application of TTX (0.2 µM) and washout of TTX (Wash). TTX significantly inhibits sEPSCosc amplitude and fre- quency (* indicates significant difference from Wash, P<0.05; ** indi- cates significant difference from Wash, P<0.01). Mu opioids also inhibit glutamate release in the spinal cord dorsal horn, including glutamate release from primary afferents in lamina I (Jeftinija, 1988; Kangrga and Randic, 1991; Terman et al., 2001). We used the mu-opioid receptor agonist DAMGO to test whether inhibition of glutamate re- lease by this means would similarly inhibit the sEPSCsosc. Bath-application of DAMGO (1 µM) induced a significant naloxone (1 µM)-reversible inhibition of the frequency (P=0.002) (Fig. 8A,D), but not amplitude (P=0.059) (Fig. 8A, B) nor decay time (P=0.621) (Fig. 8C) of the sEPSCosc. Thus, modulation of glutamate release by numerous mech- anisms correlates with the frequency of the sEPSCsosc. Synaptic glutamate concentration modulates the sEPSCosc Glutamate release is not the only mechanism that deter- mines synaptic glutamate concentration. Neural and glial inhibited the sEPSCosc frequency (P=0.003) and ampli- tude (P=0.045), but had no effect on decay time (P=0.271) (Fig. 10A–D). Notably, the inhibitory effect of ifenprodil on sEPSCsosc was not as great as with APV, which completely eliminated the sEPSCosc in both the 1 mM Mg2+ and 0 Mg2+ preparations. In sum, although NMDARs containing the NR2B subunit appear to mediate at least part of the sEPSCosc, this subtype of the NMDAR is not solely responsible for generating these events. DISCUSSION In this report we have characterized a novel large sponta- neous low frequency oscillating EPSC (sEPSCosc) in lam- ina I of the neonatal rat spinal cord dorsal horn. This sEPSCosc is: blocked by2 mM Mg2+ and APV; inhibited by ifenprodil, 1 mM Mg2+, TTX, CNQX, and DAMGO; and potentiated by strychnine, bicuculline, and TBOA. Taken together, these data suggest that this sEPSCosc is gener- ated by glutamate released from pre-synaptic neuronal sites that act on post-synaptic NMDARs, at least some of which contain the NR2B subunit. Fig. 6. Blockade of post-synaptic Na+ channels with QX-314 inhibits the amplitude, but neither decay time nor frequency of sEPSCosc. (A) Representative traces showing inhibition of amplitude by QX-314. (A) Means and S.E.M.s are depicted for sEPSCosc amplitude with and without QX-314 (1 mM) in the intracellular pipette solution (* indicates significant decrease compared with control (no QX-314 in the intracellular pipette solution, P<0.05). (B) Means and S.E.M.s are depicted for sEPSCosc decay time with and without intracellular QX-314. (C) Means and S.E.M.s are depicted for sEPSCosc frequency with and without intracellular QX-314. No significant differences exist between groups. sEPSCosc is NMDAR-mediated The observation that the sEPSCosc is sensitive to magne- sium block (i.e. absent in 2 mM magnesium and potenti- ated in 0 mM magnesium) parallels observations that pre- viously identified spontaneous membrane oscillations in the ventral root and ventral spinal cord (Chub, 1991; Bal- lerini et al., 1999; Keefer et al., 2001) and spontaneous synaptic currents identified in the olfactory bulb (Puopolo and Belluzzi, 2001) are sensitive to magnesium blockade. Notably, the spontaneous low frequency (0.1– 0.3 Hz), relatively large (approximately 100 pA) synaptic currents identified by Puopolo and Belluzzi (2001) in the olfactory bulb were induced only after removal of bath magnesium. The sEPSCosc reported here has a similar frequency but is present in the dorsal horn in 1 mM Mg2+ aCSF, well within the normal range for physiological magnesium. Magnesium sensitivity of the sEPSCosc prompted us to investigate the role of NMDARs in mediating this phenom- enon. Again our findings of sEPSCosc block by a NMDAR antagonist parallels previous observations that spontane- ous membrane voltage oscillations in the ventral root (Chub, 1991; Bracci et al.,1996b); ventral spinal cord (St- reit, 1993; Hochman et al., 1994; Bracci et al., 1996a); and dorsal root (Chub, 1991 (chick embryos)); as well as spon- taneous synaptic currents in the ventral spinal cord (Bracci et al., 1996a), cultured embryonic spinal cord (Lewis and Faber, 1996) and other CNS areas (Robinson et al., 1993 in cultured cortical neurons; Puopolo and Belluzzi, 2001 in olfactory bulb neurons) are dependent on NMDARs. Inhibition of the sEPSCosc amplitude by intracellular MK-801 administration suggests that the NMDARs impor- tant for mediating sEPSCsosc are located on lamina I neu- rons. This does not rule out a role for pre-synaptic NMDARs in modulating sEPSCosc frequency, as pre- synaptic NMDARs (including those on primary afferent terminals) have recently been reported to influence pre- synaptic glutamate release (Bardoni et al., 2004; J. Zeng, unpublished observations). However, we can nei- ther confirm nor reject such a role, as APV completely eliminates all sEPSCsosc, making further inquiry into the locus of its action impossible. Synaptic glutamate levels regulate the sEPSCsosc Glutamate release was strongly correlated with sEPSCosc frequency in our studies. For instance, CNQX, which greatly diminishes dorsal horn excitability by blocking AMPA and kainate glutamate receptors, significantly inhib- its sEPSCosc frequency. Similarly, the mu-opiate receptor agonist DAMGO, which inhibits primary afferent glutamate release in lamina I (Kangrga and Randic, 1991; Terman et al., 2001), also inhibits sEPSCosc frequency. Conversely, increases in dorsal horn excitability by blockade of GABAA and glycine inhibition are correlated with an increase in sEPSCosc frequency. Unfortunately, more specific gluta- mate release studies using mEPSC analysis are made impossible by the near total elimination of sEPSCsosc by TTX. Nonetheless, this TTX-sensitivity also points to a correlation of sEPSCosc frequency and glutamate release, with a certain minimum level of glutamate release neces- sary to produce any sEPSCsosc. Moreover, the observa- tion that bath-applied TTX virtually abolishes sEPSCsosc demonstrates that glutamate from glial stores is not suffi- cient to produce sEPSCsosc, pointing instead to neuronal stores of glutamate in mediating sEPSCsosc. The failure of intracellular administration of the sodium channel blocker QX-314 to inhibit sEPSCosc frequency again supports the importance of pre-synaptic (and likely polysynaptic) gluta- mate release in generating the sEPSCosc. However, QX- 314 did inhibit sEPSCosc amplitude, suggesting that volt- age-activated sodium channels are involved in mediating a portion of this large post-synaptic response despite our efforts to voltage clamp the neurons at —70 mV. Nevertheless, synaptic concentrations of glutamate appear to importantly influence sEPSCosc frequency. Nor- mally, neuronal and glial glutamate transporters act to limit glutamate in the synaptic cleft in order to maintain non- toxic synaptic glutamate concentrations (see Danbolt, 2001 for review). Following bath-application of the non- specific glutamate transporter inhibitor TBOA, we ob- served a potentiation of the sEPSCosc frequency. Physio- logical and/or pharmacological disruption of glutamate transporter function is thought to allow glutamate to stay in the synaptic cleft longer and perhaps even spill over onto neighboring synapses or even extra-synaptic targets (Asztely et al., 1997; Huang, 1998; Kullmann and Asztely, 1998; Mitchell and Silver, 2000; Piet et al., 2004; Takayasu et al., 2004). TBOA’s potentiating effects on sEPSCosc decay time parallel previous observations of TBOA-induced glutamate transporter inhibition, and are consistent with the potentiating effects of glutamate spillover on EPSCs (Diamond, 2001). Because TBOA inhibits both neuronal and glial gluta- mate transporters, it is impossible to differentiate the im- portance of glial versus neuronal transporters in regulating synaptic glutamate concentrations in this preparation. However, bath application of dihydrokainic acid (DHK), a selective inhibitor of the glial glutamate transporter GLT-1, did not significantly potentiate the frequency or decay time of the sEPSCosc (n=7, data not shown). We suspect, therefore, that both neuronal and glial glutamate transport- ers are involved in the glutamate uptake that, when inhib- ited, facilitates sEPSCsosc. NR2B NMDAR subunit involvement If the sEPSCosc is partly induced by increases in glutamate in the synaptic cleft which can cause spillover onto extra- synaptic regions, then the often extra-synaptic location of spinal dorsal horn NMDARs containing the NR2B subunit is ideal to participate in these events (Mutel et al., 1998; Boyce et al., 1999; Chazot, 2004 (review), see also Lozovaya et al., 2004; Scimemi et al., 2004). In this report, bath-application of ifenprodil, a selective antag- onist of the NR2B subunit, significantly inhibited the amplitude of sEPSCsosc, suggesting that post-synaptic NR2B subunits are involved in generating sEPSCsosc. However, the effects of ifenprodil on sEPSCsosc did not induce nearly as much inhibition as the less selective antagonist APV. Thus, we do not believe that NR2B-con- taining NMDARs are solely involved in sEPSCosc genera- tion. Notably, bath-application of PPDA, a non-selective competitive inhibitor of the NR2D subunit (Feng et al., 2004; Lozovaya et al., 2004; Kinarsky et al., 2005) also partially blocked the sEPSCosc (data not shown). Further study of the NMDAR subunits involved in mediating the sEPSCosc awaits more selective pharmacological agents. Also of note, was the inhibition of sEPSCosc frequency by ifenprodil, suggesting a role for pre-synaptic NR2B-con- taining NMDARs in modulating sEPSCsosc. Unfortunately, we are unable to distinguish whether these pre-synaptic NMDARs are specifically on adjacent axon terminals or instead on dendrites or cell bodies of other, perhaps wide- spread, dorsal horn neurons because miniature EPSC studies of this phenomenon are not feasible. Physiological relevance The sEPSCosc characterized in this report could be spe- cific to the developing neonatal spinal cord. The neonatal rat spinal cord is characterized by both greater NMDAR- mediated excitation (Kalb et al., 1992; Baba et al., 2000) and greater NR2B expression (Shibata et al., 2003) than the adult spinal cord. On the other hand, Ruscheweyh and Sandkühler (2005) have recently reported NMDAR-depen- dent oscillations of neural activity induced by the potas- sium channel blocker 4-AP in superficial dorsal horn of slices from rats up to 19 days of age. Moreover, Rus- cheweyh and Sandkühler (2003) reported similar 4-AP- induced oscillations in rats as old as 28 days, although, these oscillations were not NMDAR dependent, perhaps due to the 2 mM Mg2+ aCSF concentration used in those studies. Characterizing the ontogeny of the sEPSCosc would require considerably more study. However, we have observed sEPSCsosc in slices from older rats (3 weeks- old) and mice (6 weeks-old), albeit at a lower frequency (see Table 1). How our findings in lamina I generalize to other areas of the spinal cord is uncertain. As mentioned above, others have identified spontaneous NMDAR-mediated neural os- cillations in the spinal cord ventral horn (Streit, 1993; Hoch- man et al., 1994; Bracci et al., 1996a,b; Ballerini et al., 1997, 1999; Rozzo et al., 2002). In contrast, the 4-AP induced Ca2+ oscillations reported by Ruscheweyh and Sandkühler (2005) were absent in the ventral horn and decreased in deeper dorsal horn. Asghar et al. (2005) reported an oscillation in neural activity in substantia ge- latinosa caused by focal application of potassium. This oscillation was not NMDAR-mediated, although such substantia gelatinosa neural activity could certainly drive the more superficial activity in lamina I through glutamate re- lease. Whatever the driving force for the lamina I oscilla- tions, it is interesting to note that the 4-AP-induced oscil- lations in intracellular Ca2+ identified by Ruscheweyh and Sandkühler. (2005) appear synchronously in numerous superficial dorsal horn neurons, sometimes, even on both sides of the spinal midline. Preliminary data in our labora- tory suggest that sEPSCsosc may also occur synchro- nously in numerous lamina I neurons, perhaps with impor- tant implications for nociception. We can only speculate about the physiological rele- vance of a synaptic current that is clearly under tight phys- iological control by Mg2+ and synaptic glutamate concen- tration, as well as GABA and glycine levels of inhibitory neurotransmission. Nonetheless, we observe sEPSCsosc in physiological concentrations of magnesium. Also, gluta- mate transporter inhibition (and presumed increases in synaptic glutamate concentrations) has been reported in several physiological and pathophysiological states, in- cluding NMDAR-dependent increases in nociception (Mao et al., 2002; Minami et al., 2001; Sung et al., 2003; Liaw et al., 2005). Moreover, some authors have recently found a decrease in resting inhibitory tone in the dorsal horn in a number of chronic pain models (Zeilhofer et al., 2000; Moore et al., 2002; Muller et al., 2003), many of which are reversed by NMDAR antagonists, including selective NR2B subunit blockers (see Petrenko et al., 2003 for re- view). Thus, in spinal cords of adult animals with experi- mentally induced pains, neuroplastic changes can occur in which characteristics of the neonatal spinal cord—includ- ing greater NMDAR activity—are once again expressed. In this regard, our observations of NMDAR-mediated large spontaneous EPSCs (and resultant action potentials) in neonatal Lamina I neurons—which we and others have previously found to be associated with 2-APV nociception—may have important implications for spontaneous in vivo pain states.