Synaptic Plasticity in Spinal Lamina I Projection Neurons That Mediate Hyperalgesia

See allHide authors and affiliations

Science  21 Feb 2003:
Vol. 299, Issue 5610, pp. 1237-1240
DOI: 10.1126/science.1080659


Inflammation, trauma, or nerve injury may cause enduring hyperalgesia, an enhanced sensitivity to painful stimuli. Neurons in lamina I of the spinal dorsal horn that express the neurokinin 1 receptor for substance P mediate this abnormal pain sensitivity by an unknown cellular mechanism. We report that in these, but not in other nociceptive lamina I cells, neurokinin 1 receptor–activated signal transduction pathways and activation of low-threshold (T-type) voltage-gated calcium channels synergistically facilitate activity- and calcium-dependent long-term potentiation at synapses from nociceptive nerve fibers. Thereby, memory traces of painful events are retained.

Lamina I neurons that mediate hyperalgesia express the neurokinin 1 (NK1) receptor for substance P (1, 2), which is released from primary afferent Αδ- and C-fibers upon intense noxious stimulation (3, 4). Selective destruction of these neurons by saporin conjugated to substance P attenuates the development of hyperalgesia following inflammation of a hind paw or tight ligation of L5 and L6 spinal nerves, a model of neuropathic pain, in behaving rats (1, 2). In anesthetized rats, sensitization of spinal nociceptive neurons by chemical excitation of C-fibers with capsaicin (5) is diminished. Many neurons in lamina I that express the NK1 receptor have an ascending projection to the brain (6), consistent with the involvement of a supraspinal loop in central sensitization (7). The cellular mechanisms by which these lamina I neurons mediate hyperalgesia and central sensitization are unknown.

Recordings were made from 516 lamina I neurons (8). Because only a subset of lamina I neurons express the NK1 receptor (9), we took advantage of the fact that 80% of all lamina I neurons that send a projection to the parabrachial area (6) express the NK1 receptor. We performed whole-cell patch-clamp recordings from 355 lamina I projection neurons (PNs) that were retrogradely labeled with DiI (Fig. 1, A and B). For comparison we also recorded from 161 unidentified neurons (UNs) in lamina I. Sixty-nine of 71 PNs and 37 of 48 UNs tested received mono- and/or polysynaptic input from fine primary afferent nerve fibers, most of which are nociceptive. This result is consistent with the finding that all spinoparabrachial neurons (10) and most lamina I neurons are nociceptive specific. To determine whether PNs express functional NK1 receptors, we added substance P to the bath solution in the presence of tetrodotoxin. This addition induced a transient inward current in 77% of 27 PNs tested (Fig. 1D). Responses were prevented by addition of the specific substance P receptor antagonist L-703,606 (n = 9) to the bath solution. In contrast, seven of nine UNs failed to respond to substance P (Fig. 1E). Lamina I projection neurons retrogradely labeled with DiI (n = 85) displayed significantly different active and passive membrane properties as compared with UNs (n = 70). These differences included larger membrane capacitance (69 ± 3 pF versus 37 ± 3 pF), more negative resting membrane potentials (mean ± SE, −63 ± 1 mV versus −58 ± 1 mV; P < 0.01, t test), and more negative thresholds for action potential (AP) firing (−43 ± 1 mV versus −37 ± 3 mV). To exclude the possibility that these differences were due to the incorporation of DiI, we also tested Fluoro Gold as a retrograde marker, which produced similar results (table S1).

Figure 1

Projection neurons in lamina I express the NK1 receptor. (A) The distribution of 200 nl DiI (2.5%) injected into the parabrachial area in one representative animal. KF, Kölliker Fuse subnucleus; LPB, lateral parabrachial area; MPB, medial parabrachial area; mcp, middle cerebellar peduncle; scp, superior cerebellar peduncle. Black indicates the area damaged by the injection; orange indicates the spread of the tracer into adjacent regions. The numbers under each section indicate the distance (in millimeters) posterior to bregma. (B) Lamina I projection neuron retrogradely labeled with DiI in transmission and in fluorescence mode. The location of this neuron is shown in (C) by an orange dot in the slice outline. The black dot indicates the UN from (E). PNs (D) but not UNs (E) respond with an inward current to bath application of substance P (2 μM) in the presence of tetrodotoxin (0.5 μM). (D) is from the same neuron as (B). EPSCs were measured in lamina I neurons in response to electrical stimulation of dorsal roots at C-fiber strength. Mean times courses EPSC amplitudes ± SE in PNs (F) and UNs (G) before and after conditioning HFS of the dorsal root at 100 Hz for 1 s three times at 10-s intervals.

We tested whether activity-dependent synaptic long-term potentiation (LTP), which is a potential cellular mechanism of afferent-induced hyperalgesia (11–14), can preferentially be induced in PNs. We measured mono- and polysynaptic C-fiber–evoked excitatory postsynaptic currents (EPSCs) from PNs and UNs before and after applying conditioning high-frequency stimulation (HFS) to the spinal dorsal root. In 15 of 23 (65%) PNs tested, but in none of 10 UNs (Fig. 1, F and G), HFS induced LTP of C-fiber–evoked EPSCs. When results from all 23 recorded PNs were pooled, mean EPSC amplitudes significantly increased to 128 ± 8% (P < 0.01) and did not decrease within 30 min after HFS (Fig. 1F). We tested whether the supraspinal projection or the expression of NK1 receptors is a better marker for neurons that express synaptic plasticity. HFS induced synaptic LTP in all PNs that responded to bath application of substance P (2 μM) (mean EPSC amplitude 138 ± 6% of control at 10 min after HFS; n = 5). In these neurons, substance P induced a desensitizing inward current (21 ± 7 pA, holding potential V hold = –55 mV), indicating that these neurons expressed functional NK1 receptors. In contrast, HFS failed to change synaptic strength in all five PNs that did not respond to substance P (supporting online text). We asked whether the expression of NK1 receptors is only a marker or whether it has any functional role for synaptic plasticity. In the presence of the NK1 receptor antagonist L-703,606, HFS of dorsal roots failed to induce synaptic LTP in all PNs tested (Fig. 2A). Activation of NK1 receptors may lead to a rise in the concentration of free cytosolic Ca2+ by Ca2+ release from inositol 1,4,5-trisphosphate (IP3)–sensitive intracellular Ca2+stores through the phospholipase C (PLC) pathway. Blockade of Ca2+ rise by BAPTA added to the pipette solution (Fig. 2B), blockade of PLC (Fig. 2C), or blockade of IP3 receptors (Fig. 2D) all prevented LTP induction. Substance P may also enhance Ca2+ influx through NMDA (N-methyld-aspartate) receptor channels in some cells. Bath application of substance P significantly and supra-additively increased NMDA receptor–mediated inward currents in PNs but not in UNs (fig. S1A) through the PLC pathway (fig. S1B). Pharmacological blockade of NMDA receptors prevented LTP induction by HFS (Fig. 2E). Thus, the presence and activation of NK1 receptors on spinal lamina I projection neurons are essential for the induction of activity-dependent synaptic LTP that requires a substance P–induced rise in Ca2+, likely by Ca2+ release from intracellular stores, and a substance P–facilitated Ca2+ influx through NMDA receptor channels.

Figure 2

Induction of LTP of synaptic strength between afferent C-fibers and PNs by HFS is blocked by the NK1 receptor antagonist L-703,606 (10 μM) (A), by the Ca2+chelator BAPTA (20 mM, included in the pipette solution) (B), by the PLC inhibitor U73122 (10 μM) (C), by blocking IP3 receptors with 2-APB (100 μM) (D), or by blocking NMDA receptor channels with D-AP5 (50 μM) (E), added to the bath solution. (F) Mean changes (+SE) in synaptic strength induced by HFS in PNs and UNs under control conditions and during application of drugs (PNs, control: 128 ± 8% of control; n = 23, P< 0.01; L-703,606: 95 ± 1% of control, n = 11; BAPTA: 105 ± 3% of control, n = 11; U73122: 97 ± 1% of control, n = 7; 2-APB: 104 ± 2% of control, n = 7; D-AP5: 94 ± 1% of control, n = 4; UNs: 102 ± 5% of control,n = 10). Traces show original EPSC recordings evoked by stimulation of a dorsal root; spontaneous EPSCs are occasionally superimposed. ** P < 0.01.

Upon excitation, 62 of 85 (73%) PNs tested, but virtually none (3%) of 70 UNs, discharged APs with a hump in the falling phase leading to AP broadening (Fig. 3Aa). To determine the ionic basis for AP broadening, we added Cd2+ to the bath solution, which nonselectively blocks voltage-gated calcium channels (VGCCs). Cd2+ abolished or strongly reduced AP broadening (n = 5, P < 0.05). To identify the type of VGCC involved, we added the L-type VGCC blockers nicardipine or verapamil to the bath solution, which failed to affect AP broadening (n = 5 each, P > 0.05). In contrast, Ni2+, which blocks T/R-type VGCCs when applied at micromolar concentrations, abolished or strongly reduced AP broadening (n = 8, P < 0.01), suggesting that opening of T/R-type VGCCs is involved (Fig. 3, B and C). To examine whether low voltage–activated T-type and/or high voltage–activated R-type Ni2+-sensitive calcium currents (15) are differentially expressed in neurons with or without AP broadening, we determined the activation curves (8). In all 11 neurons with AP broadening, T-type VGCCs with low voltage-activation thresholds (–65 mV from aV hold of –90 mV) were identified. None of seven neurons without AP broadening displayed a T-type calcium current. The expression of T-type VGCCs in PNs with AP broadening, but not in other lamina I neurons, suggests that these neurons have a stronger activity-dependent influx of Ca2+. We thus measured cytosolic Ca2+ transients in the somata of lamina I neurons, using Fura-2 in the pipette solution during controlled AP discharges (8). Short depolarizing current pulses were adjusted to elicit a single AP per pulse and were applied at 40 Hz for 1 s. Neurons with T-type VGCCs exhibited significantly higher Ca2+ transients during AP discharges than other lamina I neurons (Fig. 3, A and C). In the presence of Ni2+, however, the Ca2+ rise in neurons with T-type VGCCs was not different from those without (Fig. 3C). The size of AP broadening quantified as the area under the hump of the AP waveform was positively and linearly correlated with Ca2+ rise (Fig. 3D). Thus, opening of T-type VGCCs during AP firing enhanced Ca2+influx in these neurons.

Figure 3

AP broadening in PNs is correlated with the magnitude of activity-dependent rise in the intracellular Ca2+ concentration ([Ca2+]i). (A) Upper traces show APs; lower traces show rise in [Ca2+]i in response to a train of 40 action potentials in 1 s for a PN with (a) and without (b) T-type VGCCs and a UN (c). (B) Blockade of T-type VGCCs by addition of Ni2+ (100 μM) to the bath solution abolished AP broadening and reduced the rise in [Ca2+]i in response to AP firing. Upper traces show APs and lower traces show [Ca2+]irise before (control) and during Ni2+ application. (C) Mean amplitudes (+SE) of [Ca2+]i rise in response to trains of 40 APs in PNs with (722 ± 88 nM, n = 16) or without T-type VGCCs (273 ± 47 nM; n = 18,P < 0.05) and UNs (366 ± 80 nM;n = 12, P < 0.05) under the control conditions. During superfusion with Ni2+, [Ca2+]i in PNs with T-type VGCCs increased to 365 ± 62 nM (n = 6), similar to the rise in PNs without T-type VGCCs (308 ± 112 nM, n = 5). (D) The magnitude of AP broadening was quantified by measuring the area under the hump of the AP waveform, here called area under the curve. A strong linear correlation exists between the size of the area under the curve in PNs and the amplitude of the [Ca2+]i rise in response to AP firing.

An indirect indication that T-type VGCCs are involved in LTP induction is the observation that in only one of five PNs without T-type VGCCs, but in all five PNs with T-type VGCCs, LTP could be induced by HFS (Fig. 4, A and B). The expression of T-type VGCCs in lamina I neurons was associated with a steeper Ca2+ rise during conditioning HFS (Fig. 4, C and D). The strength in Ca2+ rise was a good predictor of the magnitude of LTP (Fig. 4E). To directly examine whether the induction of this form of synaptic LTP requires Ca2+ influx into postsynaptic cells through T-type VGCCs, we included Ni2+ in the bath solution, which abolished LTP induction (Fig. 4G).

Figure 4

LTP induction of C-fiber–evoked EPSCs requires activation of T-type VGCCs. LTP is reliably evoked by HFS in PNs with (A) but not without (B) T-type VGCCs. (C) Examples of Ca2+ signals of PNs with and without T-type VGCCs during HFS (F340/F380: ratio of the intensities of fluorescence measured at 340 nm and 380 nm). (D) The [Ca2+]i rise during HFS was quantified by addition of the F340/F380 values of the three Ca2+ peaks evoked by HFS. The [Ca2+]i rise during HFS in PNs with T-type VGCCs was significantly larger (1.2 ± 0.3,n = 5) than that in PNs without T-type VGCCs (0.3 ± 0.1; n = 5, P < 0.05). (E) A linear correlation exists between the change of EPSC amplitude and the [Ca2+]i rise during HFS [(•), PNs with, (○), PNs without T-type VGCCs]. (F) A linear correlation exists between the change of EPSC amplitude following HFS and the size of the area under the curve (hump of AP) in the five PNs with T-type VGCCs. (G) Induction of LTP is prevented by the presence of T-type VGCC blocker Ni2+ (100 μM) in the recording solution (n = 8). * P < 0.05.

The present study identified synaptic mechanisms that lead to activity-dependent sensitization of spinal dorsal horn lamina I neurons that mediate abnormal sensitivity to pain. Induction of LTP by conditioning stimulation of C-fibers required coactivation of NK1 receptors for substance P, Ni2+-sensitive T-type calcium currents, and NMDA receptors, all contributing to a steep rise in postsynaptic Ca2+ levels that may be required for LTP induction (16, 17). All three cloned members of the T-type VGCC family (α1G, α1H, and α1I) are present in the spinal dorsal horn, with the α1G and α1H subtypes being most prominent in lamina I (18). Blocking spinal T-type VGCCs with ethosuximide depressed all types of sensory responses of spinal dorsal horn neurons in vivo (19). A role of T-type VGCCs in the sensitization of spinal nociceptive neurons had not been shown previously. Hyperalgesia in behaving animals involves calcium-dependent signal transduction pathways: Animals with neuropathic pain have enhanced cytosolic Ca2+ concentrations in spinal neurons (20). Neuropathic but not acute pain requires activation of calcium-activated protein kinase C or calcium-calmodulin–dependent protein kinase II. This activation may lead to phosphorylation of GluR1 subunit of the AMPA receptor, which enhances glutamatergic synaptic transmission (11, 21, 22). LTP can also be induced in intact rats by electrical stimulation of dorsal roots (23) or by natural noxious stimulation including inflammation, trauma, and nerve injury (24). LTP induction in vivo also requires coactivation of NK1 and NMDA receptors (23). An LTP-like, NMDA receptor–dependent enhancement of pain sensitivity has recently been induced in human volunteers by high-frequency conditioning transcutaneous electrical nerve stimuli (25). Thus, LTP at synapses between nociceptive afferents and a well-defined group of neurons in the spinal dorsal horn underlies some forms of abnormal sensitivity to pain.

Supporting Online Material

Materials and Methods

SOM Text

Fig. S1

Table S1


  • * Present address: Department of Human and Artificial Intelligence Systems, Fukui University, Fukui, Japan.

  • To whom correspondence should be addressed. E-mail: juergen.sandkuehler{at}


Navigate This Article