And its contribution to activitydependent changes in nociceptive processing

AMPA receptors

Primary sensory neurons use the amino acid L-glutamate as their principle fast excitatory neurotransmitter. Synaptically released L-glutamate primarily acts on postsynaptic ionotropic glutamate receptors of the (±)-a-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA), kainate, and N-methyl-D-aspartate (NMDA) subtypes and on G-protein-coupled (metabotropic) glutamate receptors. Four different types of AMPA receptor subunits, termed GluR-A through GluR-D or GluR-1 through GluR-4, are known. Although all four types are expressed in the spinal cord, their relative abundance varies considerably among the different laminae. GluR-A is most prevalent in laminae I and II, while GluR-B is rather homogeneously distributed throughout the dorsal horn, and GluR-C and GluR-D are relatively weakly expressed in the superficial laminae (Nagy et al. 2004). The strong expression in the superficial dorsal horn of GluR-A suggests a prominent role in spinal nociceptive processing. In the hippocampus, GluR-A plays an important role in activity-dependent plasticity, namely in long-term potentiation (LTP) at CA3-CA1 pyramidal cell synapses. Induction of LTP increases the phosphorylation of GluR-A by calcium- and calmodulin-dependent kinase II (CaMKII) at Ser 831 (Barria et al. 1997). Phosphorylation at this site increases the single channel conductance of AMPA receptors expressed in human embryonic kidney (HEK) 293 cells by about 40% (Roche et al. 1996; Derkach et al. 1999) and promotes the insertion of AMPA receptors into the postsynaptic membrane of glutamatergic synapses in cultured rat hippocampal neurons (Esteban et al. 2003). The latter process probably constitutes the final step in the generation of activity-dependent increases in synaptic strength at glutamatergic synapses and is essential for the expression of hippocampal LTP (Zamanillo et al. 1999; Mack et al. 2001). Incorporation of GluR-A into AMPA receptor channels hence enables AMPA receptors to function as endpoints in the generation of activity-dependent changes in synaptic transmission (Fig. 2).

The prominent expression of this "plasticity-permitting" GluR-A fits nicely with the fact that among the various afferent inputs to the spinal cord, input from nociceptors is the one that is most sensitive to plastic changes and central sensitization (Wall and Woolf 1984; Cook et al. 1987). Similar to tetanic stimulation in the hippocampus, intense nociceptive input to the dorsal horn leads to phosphorylation of GluR-A at Ser831 (and Ser845; Fang et al. 2003; Nagy et al. 2004). Furthermore, recruitment of GluR-A to the neuronal plasma membrane via an activity- and CaMKII-dependent process has recently been demonstrated in

Fig. 2 Activity-dependent changes at glutamatergic synapses in the superficial dorsal horn. Low-frequency activity of C fibers mainly activates postsynaptic (±)-a-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA) receptor channels. N-methyl-D-aspartate (NMDA) receptors remain inactive due to their blockade by extracellular Mg2+. Under conditions of exaggerated C-fiber activity, glutamate released from C-fiber terminals also activates NMDA receptors and triggers Ca2+ influx through NMDA and, if present, through Ca2+-permeable AMPA receptor channels into the postsynaptic neuron. The subsequent increase in intracellular free Ca2+ activates CaMKII, which phosphorylates GluR-A, leading to its translocation to the subsynaptic membrane and to an increase in channel conductance. Other kinases, including protein kinase A (Zou et al. 2002), protein kinase C (Chen and Huang 1992), and scr (Guo et al. 2004), activated by metabotropic glutamate receptors also phosphorylate NMDA receptor subunits and probably facilitate their activation. Phospho-rylation by protein kinase C reduces the Mg2+ block and thereby further facilitates NMDA receptor activation (Chen and Huang 1992).

Fig. 2 Activity-dependent changes at glutamatergic synapses in the superficial dorsal horn. Low-frequency activity of C fibers mainly activates postsynaptic (±)-a-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA) receptor channels. N-methyl-D-aspartate (NMDA) receptors remain inactive due to their blockade by extracellular Mg2+. Under conditions of exaggerated C-fiber activity, glutamate released from C-fiber terminals also activates NMDA receptors and triggers Ca2+ influx through NMDA and, if present, through Ca2+-permeable AMPA receptor channels into the postsynaptic neuron. The subsequent increase in intracellular free Ca2+ activates CaMKII, which phosphorylates GluR-A, leading to its translocation to the subsynaptic membrane and to an increase in channel conductance. Other kinases, including protein kinase A (Zou et al. 2002), protein kinase C (Chen and Huang 1992), and scr (Guo et al. 2004), activated by metabotropic glutamate receptors also phosphorylate NMDA receptor subunits and probably facilitate their activation. Phospho-rylation by protein kinase C reduces the Mg2+ block and thereby further facilitates NMDA receptor activation (Chen and Huang 1992).

response to intracolonic instillation of capsaicin (Galan et al. 2004). It is therefore probably not too farfetched to speculate that phosphorylation of GluR-A also contributes to activity-dependent pain sensitization, or wind-up (Mendell 1966), seen during intense and prolonged nociceptive input to the dorsal horn. In light of these findings, it is not surprising that a recent study indeed demonstrated that mice lacking the GluR-A subunit exhibit reduced nocicep-tive sensitization in tests of tonic nociceptive stimulation (Hartmann et al. 2004). These results suggest that the generation of central nociceptive sensitization and hippocampal LTP have at least some basic events in common (for a review, see Ji et al. 2003).

Ca2+-permeable AMPA receptors

Despite the similarities discussed above, AMPA receptors in the dorsal horn exhibit a number of rather peculiar features that may also be important for central nociceptive sensitization. Perhaps most striking among these is an unusually high number of Ca2+-permeable AMPA receptors (Engelman et al. 1999), which reside both on y-aminobutyric acid (GABA)ergic interneurons and on excitatory projection neurons (Albuquerque et al.

1999). Because Ca2+ plays an important role in synaptic plasticity, these receptors may critically contribute to LTP-like phenomena in the spinal cord (Gu et al. 1996). The low Ca2+ permeability of most AMPA receptors results from the insertion of the edited form of the GluR-B subunit in the channel complex. In most GluR-B transcripts, a certain adenosine residue is deaminated, leading to an exchange of glutamine to arginine at a critical position in the pore-forming M2 segment (Seeburg et al. 1998). It is not clear whether the high number of Ca2+-permeable AMPA receptors in the dorsal horn results from a low expression of GluR-B (in relation to, for example, GluR-A) or from incomplete RNA editing. Despite this uncertainty, Ca2+-permeable AMPA receptors seem to play an important role in central nociceptive sensitization. Both pharmacological and genetic evidence suggest that Ca2+ influx through dorsal horn AMPA receptors can trigger plastic changes in dorsal horn nociceptive circuits. Intrathecal injection of Joro spider toxin, which selectively blocks Ca2+-permeable AMPA receptors, prevents the generation of mechanical allodynia in response to burn injury in rats (Sorkin et al. 1999, but see also Stanfa et al.

2000), and mice lacking the GluR-B subunit show not only increased Ca2+ influx in the superficial dorsal horn assessed by cobalt uptake but also a facilitation of nociceptive responses in the formalin test (Hartmann et al. 2004).

Presynaptic glutamate receptors

Ionotropic glutamate receptors are widely considered as sole postsynaptic sensors for glutamate, the primary function of which is the transmission of electric signals between neurons across synapses. However, all three types of ionotropic glutamate receptors (AMPA, kainate, and NMDA receptors) are also expressed in presynaptic axon terminals. This presynaptic expression is perhaps nowhere more prominent than in the spinal cord dorsal horn (Liu et al. 1994; Tachibana et al. 1994; Popratiloff et al. 1996; Hwang et al. 2001; Lu et al. 2002). The functional significance of these presynaptic receptors has been investigated for kainate and NMDA receptors. Activation of presynaptic kainate receptors located on primary afferent terminals inhibited glutamate release (Kerchner et al. 2001b), whereas those located at presynaptic terminals of inhibitory interneurons facilitated action-potential-independent release of GABA and glycine (Kerchner et al. 2001a). Similar results have been reported for presynaptic NMDA receptors. Liu et al. (1997) found that intrathecally injected NMDA triggered the release of substance P and glutamate from primary afferents, a process that may promote sensitization. However, more recently, Bardoni et al. (2004) provided evidence that activation of presynaptic NMDA receptors can also inhibit glutamate release. These apparently conflicting results are not as surprising as they seem at first glance. Whether the activation of a depolarizing presynaptic receptor may facilitate or inhibit transmitter release depends on the magnitude and steepness of depolarization. A similar paradox is discussed below for GABA-mediated primary afferent depolarization, which can either reduce transmitter release—as long as it remains subthreshold—or give rise to pronociceptive dorsal root reflexes when it becomes suprathreshold (Willis 1999).

NMDA receptors

The pivotal role of NMDA receptors in the induction of synaptic plasticity in many CNS areas has been so widely accepted that this review will focus on some aspects that might be particularly relevant to nociceptive processing in the dorsal horn. NMDA receptor activation contributes to pain sensation in at least two respects: first, it is important for the transmission of acute pain through the spinal cord, as drugs that block NMDA receptors, such as ketamine, can produce profound analgesia, and, second, it is required for central sensitization and LTP-like phenomena in the dorsal horn (Liu and Sandkuhler 1995). Plastic changes in excitatory synaptic transmission onto dorsal horn projection neurons in response to intense C-fiber input probably contribute to the generation of enduring hyperalgesia after tissue trauma (Sandkuhler 2000). Although the high firing frequency used in initial in vitro studies in spinal cord slices is not observed in C fibers in vivo, other more physiological paradigms used more recently have yielded similar results (Sandkuhler and Liu 1998). Interestingly, the susceptibility to long-lasting changes in synaptic strength varies considerably within dorsal horn neurons. It is most prominent in spinothalamic projection neurons expressing NK1 (substance P) receptors (Ikeda et al. 2003). Because a pivotal role of NK1 receptor-positive neurons for a number of different forms of pain has been demonstrated previously (Mantyh et al. 1997; Nichols et al. 1999; Khasabov et al. 2002), this result endorses an important contribution of dorsal horn synaptic plasticity to the development of chronic pain syndromes.

One important difference between central sensitization in pain pathways and "typical" LTP is that the latter is input-specific. In the hippocampus, increases in synaptic strength are largely restricted to those synapses that have been active during the conditioning stimulation. This feature corresponds to specificity, one of three characteristics that have been postulated by the Canadian psychologist Donald Hebb for neuronal correlates of associative learning (Hebb 1966). Unlike hippocampal LTP, central sensitization in pain pathways is not fully specific. Intense stimulation of cutaneous C fibers elicits secondary (central) hyperalgesia in an area significantly exceeding the field of conditioning stimulation, indicating that primary afferent fibers not activated during the conditioning stimulus can become sensitized. Yet intense C-fiber input can even potentiate A| fiber (touch)-evoked responses of dorsal horn neurons (e.g., Coderre and Melzack 1987). Hence, central sensitization is also het-erosynaptic and could thus be either totally unrelated to LTP or could be made less specific by mechanisms involving LTP in the dorsal horn.

One such possibility could be the spread of diffusible messengers released during intense nociceptive stimulation. Glycine is one such diffusible messenger required for NMDA receptors to become fully active (Johnson and Ascher 1987; Kleckner and Dingledine 1988). In the first years after the discovery of the glycine binding site at NMDA receptors, it was unclear whether glycine binding could contribute to NMDA receptor modulation in vivo or whether this site was permanently saturated by micromolar concentrations of glycine in the cerebrospinal fluid (Westergren et al. 1994). During the last few years, however, increasing evidence has accumulated indicating that plasma membrane glycine transporters, in particular the glial glycine transporter GlyT1, can lower extracellular glycine concentrations in the vicinity of NMDA receptors to subsaturating concentrations (Bergeron et al. 1998; Berger et al. 1998; Gabernet et al. 2005). Perhaps nowhere else in the CNS is a contribution of synaptically released glycine to NMDA receptor facilitation more likely to occur than in the spinal cord, where glycinergic neurons are very abundant. Using nocistatin (Okuda-Ashitaka et al. 1998), a peptide that in the dorsal horn selectively inhibits the release of glycine (and GABA; Zeilhofer et al. 2000), Ahmadi et al. (2003) demonstrated that during intense nociceptive stimulation, glycine synaptically released in the dorsal horn can overcome reuptake by glycine transporters and reach neighboring NMDA receptors to facilitate their activation through a process called spillover. Increased activation of NMDA receptors through glycine spillover apparently contributed to sustained formalin-induced pain behavior and to neu-

Fig. 3 Spillover of synaptically released glycine facilitates NMDA receptor activation in the dorsal horn. a Under conditions of exaggerated nociceptive input to the dorsal horn, glycine released from inhibitory in-terneurons or descending glycinergic fiber tracts can escape the synaptic cleft of the glycinergic synapses and reach nearby NMDA receptors to facilitate their activation (Ahmadi et al. 2003). b Evidence for a physiological role of this process in central sensitization has been obtained in the rat formalin test. Nocistatin (NST), a peptide that in the dorsal horn selectively reduces the release of glycine (and GABA), evokes pro- or antinociceptive effects after intrathecal injection, depending on the dose injected. The antinociceptive effect is specifically antagonized by D-serine, an activator of the glycine binding site of NMDA receptors, indicating that nocistatin suppressed nociception by reducing the availability of glycine at NMDA receptors. Artificial cerebrospinal fluid (ACSF) and L-serine were ineffective (for details, see Ahmadi et al. 2003). **P<0.01; ***P<0.001

Fig. 3 Spillover of synaptically released glycine facilitates NMDA receptor activation in the dorsal horn. a Under conditions of exaggerated nociceptive input to the dorsal horn, glycine released from inhibitory in-terneurons or descending glycinergic fiber tracts can escape the synaptic cleft of the glycinergic synapses and reach nearby NMDA receptors to facilitate their activation (Ahmadi et al. 2003). b Evidence for a physiological role of this process in central sensitization has been obtained in the rat formalin test. Nocistatin (NST), a peptide that in the dorsal horn selectively reduces the release of glycine (and GABA), evokes pro- or antinociceptive effects after intrathecal injection, depending on the dose injected. The antinociceptive effect is specifically antagonized by D-serine, an activator of the glycine binding site of NMDA receptors, indicating that nocistatin suppressed nociception by reducing the availability of glycine at NMDA receptors. Artificial cerebrospinal fluid (ACSF) and L-serine were ineffective (for details, see Ahmadi et al. 2003). **P<0.01; ***P<0.001

ropathic pain in the chronic constriction injury in rats (Muth-Selbach et al. 2004). During intense nociceptive stimulation, spillover of glycine might thus promote the potentiation of synaptic input at sites not fully activated during the conditional stimulation (Fig. 3).

An interesting hypothesis that has again originated from previous findings in the hippocampus postulates that pathological pain might come from the functional activation of previously silent excitatory synapses in the dorsal horn (Li and Zhuo 1998). Early in hip-pocampal development, glutamatergic synapses are silent at resting potential but can be activated when the postsynaptic neuron is depolarized to positive membrane potentials (Durand et al. 1996). The reason behind this unusual behavior is that these silent synapses lack functional AMPA receptors, while NMDA receptors present in these synapses are blocked at negative membrane potentials by extracellular Mg2+. An LTP-like mechanism appears to be required for the recruitment of functional AMPA receptors to these synapses. Similar silent synapses have also been found in the neonatal dorsal horn (Li and Zhuo 1998) but not in adult animals (Baba et al. 2000a). A possible contribution of the activation of previously silent synapses to the generation of pathological pain in the adult therefore remains to be demonstrated.

Besides LTP, NMDA receptor-dependent long-term depression (LTD) can be evoked in the dorsal horn by conditioning stimulation of primary afferent AS fibers both in slices (Randic et al. 1993; Sandkuhler et al. 1997) and in vivo (Liu et al. 1998). Long-lasting depression of AS fiber-mediated postsynaptic responses can also be elicited by activation of group I metabotropic glutamate receptors and subsequent phospholipase C activation (Chen et al. 2000). Very recently, Klein et al. (2004) demonstrated perceptual correlates for LTP and LTD in human volunteers after cutaneous electrical stimulation.

It should finally be noted that NMDA receptors are not only important for the induction of LTP and LTD by permitting the necessary postsynaptic Ca2+ increase but are themselves regulated by neuronal activity and protein kinases. Peripheral inflammation and tonic no-ciceptive stimulation induce PKA-dependent phosphorylation of the NR1 (Zou et al. 2002) and scr-dependent phosphorylation of the NR2B subunit (Guo et al. 2004) of NMDA receptors. scr is activated by group I metabotropic glutamate receptors, a process that explains the contribution of these receptors to dorsal horn LTP (Azkue et al. 2003). Protein kinase C-dependent phosphorylation of NMDA receptors reduces their susceptibility to blockade by extracellular Mg2+ (Chen and Huang 1992). Some mechanisms of activity-dependent plasticity in the dorsal horn are summarized in Fig. 2. A possible contribution of these phos-phorylation events to inflammatory pain is discussed below.

AMPA and NMDA receptors as targets for analgesic drugs

Although both competitive [e.g., 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX)] and noncompetitive (e.g., GYKI 52466) AMPA receptor antagonists exhibit antinociceptive properties in a variety of animal models of pain (Szekely et al. 2002), their use as analgesics is severely hampered by their widespread action in the CNS. However, in contrast to most CNS areas, where fast excitatory neurotransmission is almost exclusively mediated by AMPA receptors, kainate receptors composed of GluR-5, GluR-6, and GluR-7 significantly contribute to primary afferent nociceptive transmission in the spinal cord (Li et al. 1999). Kainate receptor antagonists exert antinociceptive properties in different neuropathic pain models and in the formalin test (for a review, see Ruscheweyh and Sandkuhler 2002). Whether selective kainate receptor blockers are better tolerated than unspecific ones is unknown at present.

N-methyl-D-aspartate receptor antagonists have attracted significantly more attention as possible analgesics than AMPA receptor blockers have. This is probably for two reasons. First, NMDA receptors are typically not required for fast excitatory synaptic transmission under basal conditions. Second, the pivotal role of NMDA receptors for synaptic plasticity has raised the hope that the "preemptive" blockade of NMDA receptors might prevent the generation of chronic pain after tissue injury. Blockers of NMDA receptors are clearly analgesic, as exemplified by the intravenous anesthetic ketamine and a variety of experimental drugs (e.g., Qian et al. 1996). However, sedation and the impairment of motor coordination preclude their long-term use as analgesics. Competitive antagonists at the glycine-binding site (e.g., L-701324) of NMDA receptors show less maximal effect compared with channel blockers (e.g., ketamine or MK-801) or competitive antagonists at the glutamate binding site (e.g., AP-5), but they are probably also less efficient as analgesics. Finally, specific antagonists of NMDA receptors containing the NR2B subunit (such as ifenprodil), which is preferentially expressed in the dorsal horn, have raised new hope but still have not begun clinical application (Chizh et al. 2001).

Was this article helpful?

0 0

Post a comment