Transmitter System

Glutamate is found in 65–80% of DRG and trigeminal ganglion neurons (Battaglia and Rustioni 1988, Tracey, De Biasi et al 1991). Although aspartate was at first considered to be an afferent neurotransmitter, there is no functional evidence for this and only glutamate will be considered. Many sensory neurons exhibiting glutamate immunoreactivity have small perikarya that link them to small primary afferent fibers. Electromicrographic studies using afferent transport markers have shown glutamate to be present in the dorsal horn terminals of large fractions of both myelinated and unmyelinated axons (Broman et al 1993). Specific activation of small afferents with capsaicin evokes the release of glutamate from primary afferent neurons (Jeftinija et al 1991, Sorkin et al 1993). Recovery of glutamate in microdialysates of the dorsal spinal cord in vivo is increased several-fold after the injection of noxious chemicals into the periphery (Skilling et al 1988, Sluka and Westlund 1992, Sorkin et al 1992, Malmberg and Yaksh 1995b, Marsala et al 1995), thus providing additional support for the hypothesis that glutamate is released from afferent nociceptors, although other cellular sources of excitatory amino acids are not excluded by these studies. These findings are consistent with the observation of vesicular glutamate transporters in Aβ, Aδ, and C fibers (Oliveira et al 2003, Todd et al 2003, Hughes et al 2004). Subtypes of glutamate transporters are located predominantly, perhaps exclusively, on specific cell types; for example, excitatory amino acid carrier 1 (EAAC1) is found on dorsal horn and DRG neurons and axonal terminals, whereas glutamate aspartate transporter (GLAST) and glutamate transporter-1 (GLT-1) are found on astrocytes and microglia in the spinal cord. Astrocytic transporters are thought to be important in the newly appreciated tripartite synapse, where they transport excess glutamate into astrocytes, which is then converted into glutamine by the enzyme phosphate-activated glutaminase. Glutamine is released into the synapse, where it is picked up by axon terminals and converted back into glutamate by the resident mitochondria. Under basal conditions, transporter inhibition results in increased levels of extracellular glutamate, spontaneous pain behavior, and evoked hypersensitivity. The latter two phenomena are reversed by glutamate receptor antagonists (Liaw et al 2005). Decreased dorsal horn expression of GLT-1 and GLAST is observed following partial sciatic nerve ligation (Xin et al 2009), chronic constriction injury (CCI; Ramos et al 2010), and paclitaxel neuropathy (Weng et al 2005, Wang et al 2010), thus suggesting that injury induces the loss of astrocytic transporters and the resultant glutamate-mediated excitotoxicity. Interestingly, Ramos and co-authors (2010) reported that administration of ceftriaxone, an agent that up-regulates GLT-1 expression, reverses both the loss of GLT-1 and the pain behavior seen after a variety of injury states. These data conflict with those seen after intraplantar injection of formalin or complete Freund’s adjuvant, where glutamate transporter blockade or knockdown is reported to enhance pain behavior (Niederberger et al 2003, Yaster et al 2011).

Receptors

The post-synaptic excitatory effects of spinal excitatory amino acids are reflected by their potent ability to initiate pain behavior in animals after spinal delivery. These effects are mediated by specific interactions with a variety of glutamate receptors that are broadly divided into ionotropic and metabotropic subtypes.

The ionotropic glutamate AMPA, kainate, and NMDA receptors will be considered first. Receptors in each class are constituted from multiple subunits from different gene families to form transmembrane glutamate-activated pores. Details of assembly are provided elsewhere (Mayer and Armstrong 2004). Intrathecal injections of glutamate receptor agonists have emphasized the importance of both NMDA and non-NMDA sites on dorsal horn neurons in producing powerful algogenic behavior (Aanonsen and Wilcox 1987, Sun and Larson 1991, Coderre and Melzack 1992, Malmberg and Yaksh 1992a, Kontinen and Meert 2002). Equally important is the fact that presynaptic ionotropic autoreceptors, found on primary afferent terminals, regulate the release of glutamate (see Fig. 28-3).

Transmitter System

SP was the first peptide identified as being specific for small sensory afferents and remains the best characterized. It, along with several sequence-similar peptides (e.g., neurokinin A [NKA]), are widely distributed among small IB4-negative DRG neurons whose central terminals synapse in spinal laminae I and inner II (Pan et al 2003) (see Fig. 28-2). Based on axons identified by conduction velocity, about half of C fibers and 20% of Aδ fibers contain SP (McCarthy and Lawson 1989). In addition, populations of bulbospinal-projecting neurons also contain and probably release SP (Hokfelt et al 2000). Spinal cord release of SP is secondary to direct stimulation of central C-fiber terminals by capsaicin (Jhamandas et al 1984), by acute activation of C fibers (Yaksh et al 1980, Go and Yaksh 1987), and by noxious mechanical (Oku et al 1987, Kuraishi et al 1989) and cold (Tiseo et al 1990) stimuli. Using antibody-coated microelectrodes, SP and NKA were found to be released in the superficial dorsal horn in response to noxious thermal, mechanical, and chemical stimuli (Diez Guerra et al 1988; Duggan et al 1988, 1990). Using NK1 receptor internalization as an index of synaptic activity, peripheral noxious stimuli were found to initiate a stimulation intensity–dependent release of SP (Mantyh et al 1995, Allen et al 1997, Honor et al 1999).

Receptor

Several classes of NK receptors have been identified (Almeida et al 2004). These G protein–coupled receptors stimulate PLC, thereby leading to breakdown of phosphoinositol and elevation of intracellular calcium levels. As with other G protein–coupled receptors, when this receptor is occupied, it undergoes internalization (Mantyh 2002). NK1 receptors are densely distributed on superficial dorsal horn neurons, many of which project to the brain stem (rostroventral medulla) and diencephalon (nucleus parabrachialis) (Todd 2002, Spike et al 2003) and to a lesser degree to deeper dorsal horn neurons (Stucky et al 1993). NK3 receptors are also found superficially in the dorsal horn (Ding et al 2002).

Physiology

Spinal delivery of neurokinins, particularly SP, has been shown to (1) evoke activity in nociceptive dorsal horn neurons (Salter and Henry 1991), (2) produce mild agitation (Hylden and Wilcox 1981, Seybold et al 1982), and (3) induce potent hyperalgesia (Yashpal et al 1982, Papir-Kricheli et al 1987, Malmberg and Yaksh 1992a, Hua et al 1999) in unanesthetized animals. At the several tachykinin receptors it appears that NK1 and perhaps NK2 receptors are of most importance in nociception (Fleetwood-Walker et al 1988, Laneuville et al 1988). Spinal NK1 receptor antagonists reduce the afterdischarge in dorsal horn neurons evoked by acute noxious stimulation (Radhakrishnan and Henry 1991).

Behavior

Behavioral studies in animal models have emphasized that intrathecal neurokinin antagonists fail to alter acute nociceptive behavior (e.g., hot plate test) but do diminish the hyperalgesia induced by persistent stimuli, such as in the phase 2 formalin test (Yamamoto and Yaksh 1991, Yashpal et al 1993, Hua et al 1998), carrageenan-evoked thermal hyperalgesia (Gao et al 2003), and visceral nociception (Okano et al 2002, Gaudreau and Plourde 2003). Convergent results have been reported in rats with reduced expression of NK1 protein because of intrathecal injection of antisense oligonucleotides (Hua et al 1998) and in mice lacking the NK1 receptor (Laird et al 2001). NK3-preferring antagonists depress spinal wind-up (Barbieri and Nistri 2001) and central sensitization of a spinal withdrawal reflex (Houghton et al 2000) and reduce hyperalgesia in arthritic models (Zaratin et al 2000).

Transmitter System

CGRP-like immunoreactivity is expressed in approximately 45–70% of lumbar DRG neurons (McCarthy and Lawson 1990, Verge et al 1993). Based on identification of axonal conduction velocity, the majority of CGRP-containing neurons were classified as nociceptive (e.g., CGRP in 46% of C fibers, in 33% of Aδ fibers, and in 17% of Aβ fibers) (McCarthy and Lawson 1990). CGRP is released from the spinal terminals of primary afferent neurons by high-intensity mechanical and thermal stimuli, as well as by local injection of irritants (Morton and Hutchison 1989, Garry and Hargreaves 1992).

Receptors

The effects of CGRP are believed to be mediated by the calcitonin-like receptor, that is, a G_s-coupled seven-transmembrane–spanning receptor (Hay Conner et al 2004).

Physiology

Application of CGRP induces spinal facilitation of dorsal horn responses that were blocked by putative CGRP antagonism (Sun et al 2003). Iontophoretic application of CGRP potentiates the depolarizing effects of SP (Biella 1991).

Behavior

Intrathecal delivery of partial CGRP sequences believed to be antagonistic resulted in a reduction in the hyperalgesia induced by intradermal capsaicin (Sun et al 2003) and carrageenan (Bird et al 2006). Spinal delivery of a CGRP antagonist increased thermal escape latency with and without tissue inflammation (Yu et al 1996). In addition, CGRP antagonism diminished the writhing response induced by phenylbenzoquinone (Saxen et al 1994) and the thermal hyperalgesia and tactile allodynia otherwise observed after cord hemisection (Bennett et al 2000).

Transmitter System

SST immunoreactivity is limited to populations of small cells in DRGs and small dorsal horn neurons (Tessler et al 1986, O’Brien et al 1989, Kiyama and Emson 1990, Zhang et al 1993). SST has also been identified in populations of bulbospinal-projecting cells (Krisch 1981). Early work showed that SST is released from the spinal cord by capsaicin (Gamse et al 1981). Subsequent work indicated differential release of SST in the spinal cord in response to noxious thermal but not noxious mechanical stimuli (Kuraishi et al 1985, Morton et al 1989, Tiseo et al 1990).

Receptors

SST and its analogues act through a family of G protein–coupled receptors (SST1–5) that are widely distributed in the brain and periphery. SST1, 2, and 5 inhibit the opening of voltage-sensitive calcium channels (Olias et al 2004). Binding and parallel immunohistochemistry showed SST receptor subtypes 1, 2, and 3 in laminae I–III and in the ventral horn (Segond von Banchet et al 1999). Some of this immunoreactivity is probably present on interneurons and on terminals of sensory afferents. Immunoreactivity for the SST3 receptor is also present on a large percentage of DRG neurons and motoneurons (Senaris et al 1995).

Physiology

STT has been shown to inhibit spinal dorsal horn neuronal firing in response to noxious stimuli (Randic and Miletic 1978, Sandkuhler et al 1990, Chapman and Dickenson 1992) through a decrease in post-synaptic membrane excitability by activation of inwardly rectifying K⁺ conductance (Kim et al 2002). Other work has emphasized a biphasic concentration-dependent activation of neurons and long-lasting depression suggesting toxicity (Delfs and Dichter 1983). After intrathecal application, SST increased the hindpaw electromyographic reflex (Wiesenfeld-Hallin 1985) and facilitated thermal nociception (Wiesenfeld-Hallin 1986).

Behavior

Considerable controversy exists regarding the effects of spinal SST and its analogues. Early work suggested that it was antinociceptive. However, other reports indicated that antinociception was observed at doses that resulted in pronounced motor dysfunction (Gaumann and Yaksh 1988, 1989; Mollenholt et al 1988; Spampinato and Ferri 1991). It is probable that important differences are related to the nature of the multiple receptors being activated by the several agonists. The spinal pharmacology of these excitatory and inhibitory receptor-mediated effects has not been fully studied to date.

Transmitter System

VIP and pituitary adenylate cyclase–activating peptide (PACAP) are both structurally related members of the glucagon/secretin superfamily (Dickinson and Fleetwood-Walker 1999). VIP-positive neurons are numerous in primary afferent neurons of the thoracic and, in particular, the sacral spinal nerves, as well as in cranial nerves that innervate viscera (Kuo et al 1985, Kawatani et al 1986, Yaksh et al 1988). VIP protein and mRNA expression are localized primarily in small to medium-sized DRG neurons (Nahin et al 1994, Dun et al 1996). Afferent stimulation, but not spinal capsaicin, releases VIP from the spinal cord (Yaksh et al 1982a, Takano et al 1993).

Receptors

VIP binding is concentrated in spinal laminae I and II (Yashpal et al 1991). PACAP has also been identified in small afferents, which unlike VIP, are capsaicin sensitive. Capsaicin results in the release and depletion of PACAP in the spinal cord (Zhang et al 1997). VIP and PACAP are both recognized by a family of three receptors. Cloning reveals them to be G protein–coupled, adenylate cyclase–activating receptors (Lutz et al 1993). Message for each of the three receptors is present in the spinal dorsal horn, particularly in laminae II–IV (Dickinson et al 1999).

Physiology/Behavior

Iontophoretic VIP and PACAP evoke the excitation of dorsal horn neurons (Xu and Wiesenfeld-Hallin 1991; Dickinson et al 1997, 1999). Intrathecal VIP initiates the facilitation of spinal flexor reflexes, but spinal delivery of a VIP antagonist was without effect on this reflex (Wiesenfeld-Hallin 1989). Application of PACAP or a putative PACAP agonist (maxadilan) resulted in long-lasting spinal depolarization (Xu and Wiesenfeld-Hallin 1996) and hyperalgesia (Narita et al 1996). Conversely, application of a putative PACAP antagonist was found to induce a slow depolarizing response and reduce stimulation-evoked activation in spinal cord slices. Others have reported PACAP-induced inhibition of the C fiber–evoked flexor reflex (Zhang et al 1993), block of the tail flick (Narita et al 1996), and a reduction in formalin-induced flinching (Zhang et al 1993). Accordingly, whether PACAP is nociceptive or antinociceptive is controversial and doubtless depends on the specific receptors and systems examined (Dickinson and Fleetwood-Walker 1999).

Transmitter System

Galanin is expressed in DRGs and the spinal dorsal horn (Hokfelt et al 1987, Michener et al 1990). In the dorsal horn, galanin is primarily located in small GABAergic and enkephalinergic cells (Zhang et al 1993, Simmons et al 1995). In the DRG, neither the fiber caliber associated with galanin-positive neurons (Lawson et al 1993) nor the stimuli to which they respond have been characterized. Galanin staining density in the superficial dorsal horn decreases with C- but not with A-fiber stimulation, probably indicating release (Klein et al 1992). The physiological stimuli that evoke spinal galanin release in normal animals have not been defined. However, the peptide does not appear to be released in response to noxious thermal or mechanical stimulation (Morton and Hutchison 1989).

Receptors

Three receptors have been cloned for galanin (Gal_1–3) and belong to the superfamily of G protein–coupled receptors (Branchek et al 2000). Activation of either the Gal₁ or Gal₃ receptor produces hyperpolarization via G_i/o, whereas Gal₂ receptor activation leads to stimulation of G_q/11, thereby producing mobilization of calcium (Branchek et al 2000). All three receptor transcripts are present in the DRG and spinal cord (Waters and Krause 2000). Gal₁ receptor mRNA is present in lamina II local neurons (Parker et al 1995).

Physiology

Early work indicated that intrathecal galanin facilitates the flexor reflex in response to noxious stimulation at low doses and inhibits it at higher doses (Wiesenfeld-Hallin et al 1988). It is now known that intrathecal Gal₁ receptor (Gal_1–29) but not Gal₂ receptor (Gal_2–13)–preferring agonists inhibit spinal SP release, as assessed by NK1 receptor internalization evoked by paw compression. Spinal release of prostaglandin E₂ (PGE₂) evoked by intrathecal SP was blocked by both Gal₁ and Gal₂ receptor–preferring agonists. These data were taken to support both a pre- and post-synaptic action for Gal₁ receptor and a post-synaptic action for Gal₂ receptor at the level of the spinal dorsal horn (Hua et al 2005).

Behavior

Intrathecal low doses of galanin produce a significant reduction in the mechanical threshold (Kerr et al 2000, Liu et al 2001), whereas higher doses are reported to produce vocalization (Cridland and Henry 1988). Based on Gal₁ versus Gal₂ receptor–preferring agonists, this enhanced sensitivity is believed to be mediated by the Gal₂ receptor. Spinal Gal_1–29 but not Gal_2–11 markedly inhibited the flinching behavior induced by paw formalin, whereas both agents blocked the hyperalgesia induced by intrathecal SP (Hua et al 2004).

Transmitter System

Adenosine triphosphate (ATP) is believed to be released, in part, from primary afferent terminals (Stevens and Fields 2000, Matsuka et al 2001, Gu 2003). In culture, ATP is released from DRG axons following electrical stimulation (Stevens and Fields 2000).

Receptors

Given the multiple subunits, at least 10 functional R-homomeric and heteromeric P2X receptors have been identified (Khakh et al 2001, North 2002). P2X receptors are expressed at a variety of sites on neurons and non-neuronal cells (Kennedy et al 2003, Fields 2004). These effects are antagonized by the local application of antagonists. An important effect on the primary afferent terminal has also been postulated based on the ability of P2X agonists to initiate afferent transmitter release (see below). Current thinking points to an important role of such afferent-evoked ATP release in activating adjacent glia (Stevens and Fields 2000). Further discussion on purines in pain transmission and the results of manipulating its effects on behavior are considered below.

Amino Acid Projections

Glutamate has been extensively identified in neurons of the spinothalamic and its trigeminal homologue tracts, thus suggesting the probable role of this excitatory amino acid (Magnusson et al 1987, Ericson et al 1995, Persson and Broman 2004).

Peptidergic Projections

Immunohistochemical investigations examining the content of dorsal horn neurons labeled after the injection of a retrogradely transported substance into various brain stem sites have demonstrated spinal neurons containing cholecystokinin (CCK)-like immunoreactivity (LI), dynorphin 1–8, SST, bombesin, VIP, and SP projecting into the bulbar reticular formation (Standaert et al 1986, Nahin 1987, Leah et al 1988). Spinifugal cells containing CCK and dynorphin–LI labeling have been found in and around the central canal. Ascending tract cells located in lamina I and projecting into the spinomesencephalic and diencephalic pathways contain galanin, CCK, dynorphin, and VIP, whereas lamina V cells projecting in a spinoreticular component contain SST (Ju et al 1987a, 1987b; Leah et al 1988; Nahin et al 1989). SP-positive neurons or neurons containing message for preprotachykinin are sparse, but such cells projecting to the thalamus have been found in lamina I, in lamina V, and around the central canal (Battaglia and Rustioni 1992, Battaglia et al 1992, Noguchi and Ruda 1992, Nishiyama et al 1995).

Neuronal Receptive Field Size

The effective receptive field of a dorsal horn cell is not invariant. Classic studies have shown that section of the lateral portion of Lissauer’s tract (an intrasegmental projection system arising in part from the substantia gelatinosa [SG]) or topical application of strychnine (a glycine receptor antagonist) increases the size of the sensory dermatome in primates (Denny-Brown et al 1973). Iontophoretic delivery of glycine and GABA antagonists similarly increases the receptive field size of dorsal horn projection neurons (Zieglgansberger and Herz 1971, Lin et al 1996). Repetitive activation of small, typically high-threshold afferent input leads to a significant increase in the size of the receptive field of a given dorsal horn neuron. In contrast, other systems may decrease the size or components of the receptive field that activate a given dorsal horn neuron. μ-Opioid agonists diminish the size of the high-threshold (C-fiber) component of the receptive field but have little or no effect on the low-threshold component (Yaksh 1978).

Neuronal Response to Afferent Input

The magnitude of the response may be altered in the absence of a change in stimulus magnitude. Thus, as noted above, repetitive activation of C fibers will lead to an augmented response to subsequent afferent input, a phenomenon referred to as “wind-up” (Mendell 1966). In addition to modifying the magnitude of response to a given noxious stimulus, local application of glycine or GABA antagonists augments dorsal horn WDR neuron responses to low-threshold (Aβ) afferent input (Khayyat et al 1975, Yokota et al 1979). Conversely, agonists of specific dorsal horn receptor classes, such as those for the μ- and δ-opioid and α₂-adrenergic receptors, induce powerful suppression of the small afferent-induced excitation of these cells (see below). Furthermore, consistent with the effects of activating these specific receptor systems, considerable evidence points to a complex set of bulbospinal modulatory substrates that, by acting through these receptor systems, produce corresponding changes in dorsal horn output. Thus, brain stem stimulation can diminish the slope of the response (frequency of discharge)–versus–stimulus intensity curve of dorsal horn neurons, as well as shift the intercept of the stimulus intensity–response curve to the left, indicative of a reduction in the threshold stimulus intensity necessary to evoke activity in the cell (Gebhart et al 1983, 1984). These shifts may be modality specific, thus implicating presynaptic inhibition. Conversely, other input facilitates the response of the dorsal horn to afferent traffic (Suzuki et al 2002). These bidirectional effects on the input–output relationships of the dorsal horn mediated by spinal and supraspinally organized systems indeed form the core property of the original “gate control” formalization proposed by Melzack and Wall (1965; see also Yaksh 1999).

Spinobulbospinal

Thus, small afferent input evokes spinal release of 5-hydroxytryptamine (5-HT) and norepinephrine, indicative of the activation of bulbospinal projections (Tyce and Yaksh 1981). This linkage is mediated, in part, by spinal activation of lamina I neurons, which send projections into the caudal brain stem, particularly the caudal raphe nuclei in the rostroventral medulla (Todd et al 2000). Behavioral and electrophysiological studies have shown that the noradrenergic projections exert potent analgesic effects, as evidenced by reversal of these effects with intrathecal noradrenergic antagonists (Sagen and Proudfit 1984; see Jones 1991 for review). In contrast, the serotonergic projections have potent inhibitory or facilitatory effects on dorsal horn neurons, notably lamina V neurons, depending on the subtype of serotonin receptor involved (Willcockson et al 1984, Sorkin et al 1993), as shown by loss of this effect after destruction of the bulbospinal 5-HT pathways with the intrathecal application of a serotonin neurotoxin (5,6-dihydroxytryptamine) (Suzuki et al 2002) or by blocking one of several serotonin receptors (Zeitz et al 2002). Direct support for the functional significance of these spinobulbospinal serotonergic systems on nociception is provided by the observations that these treatments have been shown to diminish a variety of hyperpathic states associated with inflammation and nerve injury (Porreca et al 2001, Rahman et al 2006, Zhang et al 2009; but see Leong et al 2011). It should be noted that components of these descending pathways also project into the thoracic intermediolateral cell column synapsing onto preganglionic sympathetic neurons (see Fig. 28-5). These bulbospinal projections contribute to the sympathetic response initiated by spinal nociceptive input (e.g., the spinobulbospinal loop) (Ross et al 1984, Minson et al 2002).

Supraspinal–Bulbospinal

The bed nuclei from which the bulbospinal projections arise receive robust input from the rostral systems. Although space is insufficient to review this connectivity in detail, the limbic forebrain systems (septum, nucleus accumbens, hypothalamus) form a multisynaptic pathway projecting through the ventral diencephalon and mesencephalic core into the pons and rostroventral medulla to activate the bulbospinal projections described above and exert a modulatory effect on the spinal systems (see Fig. 28-1), and such linkage has been well documented (Behbehani and Fields 1979). Thus, stimulation in specific zones of the mesencephalic PAG can initiate an aversive response and produce potent effects on blood pressure (e.g., reflecting an excitatory effect on spinal preganglionic sympathetic outflow) (Blessing 2003). These linkages are particularly important because they connect regulation of spinal activation to forebrain systems known to underlie elements of anxiety and emotionality. Thus, classic conditioning studies have demonstrated that increased autonomic outflow and reflex function can result from behavioral paradigms that, for example, pair a painful stimulus with an otherwise innocuous cue (e.g., a light or brief sound). These conditioning paradigms have potent effects on forebrain function, and they are also known to activate the bulbospinal links discussed above. In short, this spinobulbospinal circuit represents a feedforward facilitatory system that can mediate a robust state of facilitated processing reflecting the role played by the supraspinal systems regulating spinal processing.

Glutamate Ionotropic Receptors

As reviewed above, the AMPA but not the NMDA receptor plays a pivotal role in the acute excitation initiated by release of glutamate from afferent terminals. The focus of this section is on the role of the NMDA ionophore in facilitating afferent-evoked excitation. An important component is its ability to produce significant increases in intracellular calcium. Previous comments above on Ca²⁺-permeable AMPA and kainate receptors should accordingly also be considered in the context of the present discussion focusing on the facilitatory effects of glutamate; however, it should be kept in mind that the subcellular machinery downstream of these various receptors differs.

Metabotropic Glutamate Receptors

Endogenous System

Bulbospinal serotonin-containing projections arise from the midline caudal raphe nuclei and project to the spinal dorsal and ventral horns (Dahlstrom and Fuxe 1964). These cells contain and release a variety of transmitters, including serotonin and SP (Bowker et al 1983). High-intensity lumbar afferent stimulation has been shown to activate the spinobulbospinal pathways and lead to spinal release of serotonin (Tyce and Yaksh 1981). Current evidence suggests that activation of this bulbospinal circuit is dependent on local circuits (e.g., from the PAG; Behbehani and Fields 1979), as well as on spinobulbar input arising from NK1 receptor–bearing spinal marginal cells that project into the medullary brain stem (Todd et al 2000).

Receptors

At least seven major subtypes of serotonin receptors exist (Hoyer et al 1994), several of which have been identified in the spinal cord (5-HT₁, 5-HT₂, 5-HT₃, and 5-HT₄). Message for a variety of the 5-HT subtypes has also been identified in DRGs, including the 5-HT_1B, 5-HT_1D, 5-HT_2A, 5-HT_2C, 5-HT₃, and 5-HT₇ receptors (Pierce et al 1996). Autoradiographic, electrophysiological, and/or pharmacological studies show the presence of 5-HT_1A, 5-HT_1D, 5-HT_2A, and 5-HT₃ receptors on sensory neurons (Hamon et al 1989; Todorovic and Anderson 1990, 1992). Specifically, 5-HT_1A and 5-HT_1D receptors are present on unmyelinated peptidergic DRG neurons (Cardenas et al 1997, Potrebic et al 2003). 5-HT_2A receptors are distributed in the spinal cord in lamina II inner and are particularly dense in lamina IX. Immunoreactive cell bodies were found to be numerous in lamina IX. Immunoreactivity is mainly post-synaptic on dendrites and cell bodies with a significant incidence of “non-synaptic” localization (Doly et al 2004). 5-HT₃ receptors are expressed by small myelinated (Aδ) afferents and a small population of largely non-peptidergic C fibers (Zeitz et al 2002), as well as by dorsal horn interneurons. Immunohistochemical and autoradiographic studies have revealed that 5-HT₃ receptors are located mainly in the central terminals of primary afferents and, in part, are post-synaptic on dorsal horn neurons (Kidd et al 1993, Morales et al 1998), fibers, and terminals. In situ hybridization studies have shown that DRG cells contain mRNA sequences that code for both the 5-HT_3A and 5-HT_3B subunits (Kia et al 1995). Most of the dorsal horn axons immunoreactive for 5-HT_3A subunits are associated with capsaicin-sensitive terminals that do not bind IB4 or contain CGRP (Maxwell et al 2003). 5-HT₇ receptors have been found on peptidergic small afferents and are concentrated in laminae I and II (Meuser et al 2002).

Physiological Effects

A variety of these 5-HT receptors are G protein coupled, activate PLC, increase intracellular calcium (5-HT_2A–2C, 5-HT₇), and stimulate cyclic adenosine monophosphate (cAMP; 5-HT₄, 5-HT₆, 5-HT₇). Others depress cAMP (5-HT_{1A, 1B, 1D}). Yet other 5-HT subtypes are known to be excitatory cation channels (5-HT₃) (see Hoyer and Martin 1997, Barnes and Sharp 1999 for review). Given the several functionally distinct populations of receptors at the spinal level, the effects of endogenous or exogenous serotonin will clearly be complex. It is evident that a number of these receptors can indeed couple in such a manner to facilitate transmission through the dorsal horn. In some instances this effect could be mediated by a direct excitatory effect on the membrane of primary afferent fibers or projection neurons. Thus, intrathecal injection of 5-HT₃ and 5-HT₇, agonists that depolarize membranes, has been reported to enhance and 5-HT₃ antagonists to reduce dorsal horn nociceptive responses (Meuser et al 2002, Zeitz et al 2002). Conversely, activation of raphe–spinal projections electrically or by glutamate has been reported to facilitate and to inhibit spinal nociceptive processing. The facilitatory effect is thought to be mediated by 5-HT_1A receptor–mediated inhibition of spinal GABAergic neurons (Zhuo and Gebhart 1991, Zemlan et al 1994). An interesting adjunct to the role of bulbospinal serotonin projections is the possibility that they may potentiate excitability by recruiting so-called silent AMPA receptors on dorsal horn neurons to the post-synaptic membrane (Kerchner et al 1999).

Behavior

Manipulation of a variety of 5-HT receptor systems has indeed been shown to alter behavior. 5-HT₃ receptor antagonism or knockdown of 5-HT₃ receptor expression has no effect on acute nociception but serves to reduce facilitated states. The second-phase behavior in the formalin test is attenuated after intrathecal injection of a 5-HT₃ receptor antagonist (ondansetron) (Oyama et al 1996, Zeitz et al 2002). 5-HT₃ receptor antagonists reduce the mechanical hyperalgesia evoked by carrageenan (Eschalier et al 1989). Given the importance of spinal 5-HT₃ receptor–mediated facilitation of dorsal horn processing and pain behavior and the apparent role of marginal cell input in driving this spinobulbospinal serotonin loop, it is important to note that destruction of NK1 receptor–bearing marginal cells by the intrathecal delivery of a neurotoxin (SP–saporin) reduces the small afferent–evoked facilitation that is otherwise observed in deep dorsal horn neurons (Suzuki et al 2002). These data jointly emphasize that the physiological and behavioral effects of bulbospinal serotonin are mediated by an effect on the spinal 5-HT₃ receptor. Alternatively, spinal delivery of 5-HT₃ agonists has also been shown to be antinociceptive, and the effects of PAG stimulation on spinal nociceptive processing is reversed by spinal 5-HT₃ antagonism, perhaps reflecting 5-HT₃–mediated excitation of GABAergic/glycinergic interneurons (Alhaider et al 1991). Spinal 5-HT₇ but not 5-HT₃ antagonists blocked the antinociceptive effects of rostroventral medulla application of morphine. In contrast, hyperalgesia was blocked by spinal 5-HT₃ but not by 5-HT₇ antagonism, thus suggesting that descending inhibitory or facilitatory pathways arising in the rostroventral medulla act at the spinal level through 5-HT₇ and 5-HT₃ receptors, respectively (Dogrul et al 2009).

Tumor Necrosis Factor

TNF is a cytokine released from astrocytes and microglia through activation of a metalloproteinase that results in proteolytic cleavage of it from the cell membrane ((McGeehan et al 1994, Zhou et al 2010). TNF binds to TNFR1 and TNFR2, both having been identified in astrocytes and microglia (Dopp et al 1997) and in neurons (Zhang et al 2010). In nerve injury and in the presence of chronic inflammation there is a prominent increase in TNF. TNF has been shown to activate DRG cells and initiate a hyperpathic state after intrathecal delivery (Leung and Cahill 2010). In the DRG, TNF serves to increase the TTX-resistant Na⁺ current through a p38 MAPK–mediated cascade initiated by TNFR1 (Jin and Gereau 2006). Importantly, block of TNF with the use of knockout mice or the intrathecal delivery of antibodies can produce a significant reduction in these hyperpathic states (Lee et al 2009, Zhang et al 2011).

Fractalkine

This protein, also known as CX3CL1, binds to the chemokine receptor CX3CR1. It is constitutively expressed on the neuronal membranes from whence it is released; its receptor is constitutively expressed on spinal microglia (Clark et al 2011). Intrathecal application of fractalkine results in microglia activation and a facilitated pain state. Following peripheral nerve injury and persistent inflammation, fractalkine is up-regulated. The functional contribution of fractalkine systems to nociceptive processing is supported by the observation that hyperalgesic states can be attenuated by the intrathecal delivery of a neutralizing antibody to the fractalkine receptor (Milligan et al 2008, Clark et al 2011).

Cyclooxygenase

Current work has shown that at least two constitutively expressed prostaglandin synthases called cyclooxygenases (COX-1 and COX-2) are present in spinal neurons and non-neuronal cells, such as astrocytes (O’Banion 1999). COX-1 is predominantly expressed in naïve microglia (Yermakova et al 1999), but COX-2 expression has been noted in microglia after stimulation (Bauer et al 1997). In vivo perfusion studies have shown that local depolarization (Yaksh 1982), afferent stimulation (Ramwell et al 1966; Coderre et al 1990; Malmberg and Yaksh 1995a, 1995b), or direct activation of spinal neurons with SP or NMDA results in increased extracellular levels of prostanoids in the spinal cord (Dirig and Yaksh 1999, Hua et al 1999, Yaksh et al 2001a, Svensson et al 2003) (see Fig. 28-3).

Importantly, it should be emphasized that although peripheral COX-2 is inducible (Vane et al 1998, O’Banion 1999), spinal COX-2 is constitutively expressed and appears to be engaged immediately in the presence of the appropriate stimulus (e.g., intrathecal SP/NMDA) (see Svensson and Yaksh 2002, Ghilardi et al 2004). After intracellular formation, these lipidic acids are exported to the extracellular space, where they can then exert powerful effects on adjacent neuronal elements through a family of prostaglandin receptors (DP, EP, FP, IP, and TP). These G protein–coupled receptors with seven transmembrane domains (Armstrong and Wilson 1995, Negishi et al 1995, Versteeg et al 1999) trigger intracellular signals that can be stimulatory as evidenced by activation of adenyl cyclase (Negishi et al 1995, Narumiya et al 1999) and activation of PLC (Yousufzai et al 1988, Birnbaumer et al 1990). Conversely, inhibitory effects have been demonstrated as evidenced by depression of cAMP production (Melien et al 1988, Negishi et al 1989). In situ hybridization and immunohistochemical studies have localized the receptor proteins EP1, EP2, EP3, EP4 (Kawamura et al 1997, Donaldson et al 2001), and IP (Matsumura et al 1995) to the superficial layers of spinal cord, and DP, EP1, EP3, and IP receptors have been detected on DRG neurons (Oida et al 1995, Wright et al 1999).

On the primary afferent terminal, prostaglandins will, via EP receptors, increase Ca²⁺ conductance through voltage-sensitive calcium channels in DRG neurons (Makhinson et al 1999) and increase secretion of primary afferent peptides such as SP (Nicol et al 1992). Similarly, block of spinal COX-2 will significantly but incompletely decrease the release of SP evoked by strong, tissue-injuring stimuli (Ghilardi et al 2004) or the release of glutamate evoked by intraplantar formalin (Malmberg and Yaksh 1995b).

Post-synaptically, activation of spinal EP2 receptors by PGE₂ reduces glycinergic inhibition by phosphorylation of the α₃ subunit of the glycine receptor (Harvey et al 2004). As discussed above, glycine plays an important role in regulating afferent traffic in the dorsal horn, and block of spinal glycinergic function can initiate a potent behavioral allodynia (Yaksh 1989).

Behaviorally, intrathecal prostaglandins delivered to the unanesthetized rat evoke hyperalgesia (Yaksh 1982, Taiwo and Levine 1986, Uda et al 1990), whereas spinal COX inhibitors, specifically those selective for COX-2 but not COX-1, suppress the thermal hyperalgesia induced by spinally injected SP or NMDA (Malmberg and Yaksh 1992b, Yaksh et al 2001a), the augmented flexor reflex in adjuvant-treated rats (Malmberg and Yaksh 1992a, Seybold et al 2003), and the behavioral hyperalgesia resulting from intra-abdominal acetic acid (Yaksh 1982), intraplantar formalin (Yamamoto and Nozaki-Taguchi 2002), and peripheral tissue injury (Malmberg and Yaksh 1992a, Yaksh et al 2001b, Du et al 2004). These observations suggest a role for COX products in mediating the spinal facilitated nociceptive processing leading to hyperalgesia. These results further support the assertion that COX inhibitors exert their anti-hyperalgesic effects through inhibition of spinal prostaglandin release.

Lipoxygenase

LOX catalyzes O₂ insertion into polyunsaturated fatty acids. Several families of LOX enzymes have been classified according to their activity: 5-LOX, 12-LOX, and 12/15-LOX. Frequently, the several enzyme families have a number of isozymes (Buczynski et al 2009). In the presence of inflammation, there are significant increases in the expression of a variety of LOX products, including 12-HETE and Hepolillin B₃ (HXB₃) (Buczynski et al 2010). Importantly, intrathecal delivery of 12-LOX– but not 5- or 15-LOX–preferring inhibitors prevented increases in spinal HXB₃ and the associated hyperpathia. These agents increased SP release, thus suggesting a presynaptic effect and increased intracellular calcium in DRGs through a TRPV1/TRPA1-mediated action (Gregus et al in revision).

It should be noted that the contribution by lipid mediators is only just beginning to be understood. Thus, it has been shown that spinal cord depolarization enhances the release of linoleic acid metabolites and that when examined, these substances were 12-LOX products found to be potent activators of spinal TRP receptors (Patwardhan et al 2009).

Phosphorylation

There is expansive evidence that in the presence of repetitive input, phosphorylation plays an important role in enhancing the input–output function of dorsal horn neurons. Space is inadequate to include an in-depth discussion of this topic, their targets, or the results of phosphorylation on nociceptive processing (see Mao et al 1994, Willis 2002, Obata and Noguchi 2004; see also Chapter 3 for further discussion). These enzymes play important roles in a variety of signal transduction cascades, and several examples will be noted.

Ubiquitination

Mono-ubiquitination (tagging a protein with a single ubiquitin molecule) of plasma membrane receptors, predominantly G coupled-protein receptors, initiates agonist-induced receptor endocytosis, which usually culminates in recycling. In contrast, poly-ubiquitination of receptors results in endocytosis, followed by proteosome lysis. This is exemplified by the CXCR4 chemokine receptor (Marchese and Benovic 2001). δ-Opiate receptors are tagged with ubiquitin for proteolytic processing via a different pathway (Hislop et al 2009). Proteins other than membrane receptors are also targeted by ubiquitin for proteosomal lysis; an important example is the PKA regulatory subunit. Nerve injury leads to ubiquitination and proteolytic destruction of the PKA regulatory subunit (Chain et al 1999). Local administration of selective ubiquitin–proteosome antagonists blocks the CCI-induced increase in PKA activity while concurrently reversing the CCI-induced behavioral hyperalgesia (Moss et al 2002). Ossipov and colleagues (2007) identified a role for the ubiquitin–protease system in enhanced spinal release of CGRP and dynorphin, probably through actions on proteins involved in synaptic vesicle release, and others have recently shown that sphingosine-1-phosphate–induced activation of the mammalian target of rapamycin (mTor) is also dependent on E3 ubiquitin (Maeurer et al 2009).

Endogenous Systems

Terminals containing norepinephrine and epinephrine are present in the spinal gray and in axons that arise from neurons in the pontine A5, A6 (nucleus locus coeruleus), and A7 (subcoeruleus) cell groups (Westlund et al 1983, Rajaofetra et al 1992). Activation of these systems by direct stimulation of the bulbar catecholamine nuclei with microinjections of glutamate (Hammond et al 1985) or by activation of local input into these nuclei from the PAG (Cui et al 1999) or from ascending pathways (Tyce and Yaksh 1981) will lead to increases in spinal extracellular norepinephrine concentrations.

Adrenergic Receptors

Adrenaline and noradrenaline act through two major classes of α-adrenergic receptors: α₁ and α₂. There are three principal α₁ receptor subtypes (Hieble et al 1995). All three subtypes have been identified in DRGs and the spinal dorsal horn, α_1A, α_1B, and α_1D (Xie et al 2001). Three distinct subtypes of α₂-adrenergic receptors are distinguished: α_2A, α_2B, and α_2C (see Aantaa et al 1995, Bylund 1995). Species differences have been identified. An α_2D receptor was found in the rat and is the rodent homologue of the human α_2A receptor (MacKinnon et al 1994). Characterization of the distribution of mRNA for the three subtypes in the DRG revealed that the ordering of prevalence in a normal animal was α_2C (80%), α_2A (20%), and α_2B adrenoceptor (rare). Importantly, with nerve injury there was a prominent increase in the number of DRG neurons expressing α_2A with no change in α_2C (Shi et al 2000). Distribution of the α₂-subtype receptor protein is more variable. In terms of protein, more DRG cells express α_2A than α_2C (Stone et al 1998, Birder and Perl 1999). Thus, spinal α_2A receptor protein in the spinal cord is present on the terminals of SP-containing, capsaicin-sensitive afferent fibers. In other studies, α_2C receptor protein was found to be densely distributed on axons adjacent to cell bodies and the proximal dendrites of NK1 receptor–bearing lamina I cells and on distal dendrites from laminae III/IV neurons (Olave and Maxwell 2003). Importantly, these terminals were positive for glutamate transporter type 2 protein. Because this transporter is principally present in interneurons (Todd et al 2003), it was concluded that the α_2C protein was present on the terminals of spinal interneurons (Olave and Maxwell 2003).

Metabotropic Receptors

Group II receptors are distributed widely over presynaptic terminals in the dorsal horn (Petralia et al 1996, Lujan et al 1997, Testa et al 1998). mGlu3 receptors are expressed predominantly in the superficial dorsal horn (Ohishi et al 1993a, 1993b; Petralia et al 1996; Lujan et al 1997; Boxall et al 1998; Yung 1998; Jia et al 1999) and preferentially in axon terminals (Fagni et al 2004). Immunostaining for mGlu2/3 and mGlu7 receptors in the DRG has been reported (Ohishi et al 1995, Jia et al 1999). mGlu3 receptors are also found on glial cells (Ohishi et al 1993a, Petralia et al 1996, Boxall et al 1998, Berthele et al 1999, Jia et al 1999).

Physiology

Group II mGlu receptors negatively modulate the release of glutamate (Battaglia et al 1997) through inhibition of voltage-sensitive Ca²⁺ channels (Fagni et al 2000). This implies that group II mGlu receptors reduce the hyperexcitable states associated with hyperalgesia and allodynia. Groups II and III mGlu receptor agonists diminish the post-synaptic potentials evoked by primary afferent stimulation in dorsal horn neurons, thus suggesting that mGlu receptors expressed at primary afferent synapses exert a presynaptic inhibitory effect (Gerber et al 2000b). Activation of both group II and group III mGlu receptors can reverse capsaicin-induced facilitation of spinothalamic neurons (Neugebauer et al 2000). Similarly, group II mGlu receptor agonists diminish the C fiber–evoked discharges of WDR spinal dorsal horn neurons otherwise facilitated in the presence of carrageenan inflammation (Stanfa and Dickenson 1998).

Behavior

Intrathecal delivery of group II mGlu receptor agonists reversed the hyperalgesia induced by acute inflammation in rats and sheep (Fisher and Coderre 1996, Dolan and Nolan 2002).

GABA/Glycine System

GABA and glycine are the principal inhibitory neurotransmitters in the spinal cord. Up to 30% of neurons in laminae I and II and 45% of those in lamina III express GABA, and the majority also display glycine immunoreactivity (Todd and Sullivan 1990). It is currently thought that GABA and glycine in fact act as co-transmitters of interneurons at many synapses in the dorsal horn (Taal and Holstege 1994, Todd et al 1996, Keller et al 2001). Systematic examination has suggested that synaptic glomeruli that are presynaptic to small unmyelinated axons may display only GABA whereas glomeruli associated presynaptically with larger myelinated axons possess both neurotransmitters (Todd 1996).

Receptors

GABA and glycine both act through ligand-gated channels (GABA_A and glycine receptors, respectively), whereas GABA also acts on a G protein–coupled receptor (GABA_B). Details of the receptor structure and pentameric subunit composition of the GABA_A ionophore (Chebib and Johnston 1999, Steiger and Russek 2004) and glycine (Breitinger and Becker 2002) are considered elsewhere. The GABA_A receptor is a chloride ionophore, and when activated, Cl⁻ moves along its concentration gradient. The normal transmembrane Cl⁻ distribution is regulated by the cation–chloride co-transporters NKCC1 and NKCC2, which import and export Cl⁻ ions, respectively. In the dorsal horn, the import/export activity is such that under basal conditions, intracellular Cl⁻ is lower than extracellular Cl⁻; as a result, when the GABA_A ionophore is activated, there is an influx of Cl⁻ that results in a negative charge and leads to inhibition. Conversely, slightly more intracellular Cl⁻ is present in the primary afferent terminal, and here, GABA_A activation actually leads to depolarization. Though seemingly paradoxical, this modest depolarization suppresses transmitter release by inactivating calcium channels; the mechanism underlying the inhibition is referred to as primary afferent depolarization (PAD) (Rudomin 2002). Interestingly, after injury there is reduced Cl⁻ export activity in dorsal horn neurons, which leads to accumulation of intracellular Cl⁻ such that activation of the ionophore may now lead to an exit of Cl⁻ (i.e., depolarization). This results in turning the GABA/glycine effect into excitation of the second-order neuron (Price et al 2005). This shift has been suggested to contribute to the hyperpathia that occurs after inflammation and nerve injury (Morales-Aza et al 2011).

GABA subunit composition defines the role of other ligands, such as benzodiazepines or neurosteroids, that can alter the effects of GABA at the GABA_A ionophore (Whiting 2003). Agonist occupancy of the benzodiazepine or neurosteroid binding site enhances the activity of GABA at the GABA_A receptor (Hevers and Luddens 1998). GABA_A binding as well as message is present in large and small DRG cells and is found in high concentration in the superficial dorsal horn on terminals and cell bodies. Benzodiazepine subunit expression is present in DRGs and on spinal neurons (Wisden et al 1991, Bohlhalter et al 1996). The glycine ionophore, when activated, increases Cl⁻ conductance in the post-synaptic membrane and reduces the excitability of secondary-order neurons. Glycine binding and protein are present throughout the spinal gray. It has been shown in particular, however, that receptors composed of the α₃ subunit are present in high quantity in the superficial lamina II of the dorsal horn (Harvey et al 2004). The GABA_B receptor is a member of the seven-transmembrane–spanning superfamily, which when coupled through G_i/o protein linkage, serves to diminish the opening of voltage-sensitive calcium channels and to hyperpolarize the membrane (Hammond 2001). GABA_B binding is maximal in lamina II, with about half the binding lost after capsaicin application or rhizotomy (Price et al 1987).

Physiological Effect

Given the organizational connectivity of GABA and glycine receptors within the dorsal horn, it is not surprising that activity in dorsal horn nociceptive and non-nociceptive neurons and terminals is under powerful tonic regulation by GABA_A and glycine receptors. As noted above, the principal effect of activating these receptors is to initiate inhibitory control given the transmembrane distribution of Cl⁻ in the normal neuron. An anomalous finding is that following nerve injury, there is a change in this transmembrane gradient such that activation of GABA/glycine ionophores results in depolarization. In the following sections we will consider several examples of the effects of these inhibitory and excitatory actions.

Behavior

Intrathecal delivery of GABA_A receptor agonists has modest effects on acute escape responses (Hammond and Washington 1993) but attenuates hyperalgesia in a number of models (Malan et al 2002). With regard to benzodiazepines that interact with the GABA_A site, a similar profile of effects is noted. Early work showed that intrathecal benzodiazepines reduce small afferent–evoked somatosympathetic reflexes (Niv et al 1983, Gaumann et al 1990). In acute behavioral reflex models, such as the tail flick test, there is a dose-dependent increase in latency (Allen and Yaksh 2004). Activity is also observed in models of facilitated processing, such as the formalin test (Nishiyama and Hanaoka 2003), or in models of inflammation (Kyles et al 1995). Intrathecal delivery of GABA_B agonists has a moderate effect on acute nociceptive thresholds, but the maximum analgesic effects may not occur until doses are achieved that alter motor function (Wilson and Yaksh 1978, Yaksh and Reddy 1981, Aran and Hammond 1991, Malan et al 2002). In models of facilitated processing, a dose-dependent effect on hyperalgesia has been noted (Dirig and Yaksh 1995, Kaneko and Hammond 1997, Malan et al 2002).

An important aspect of GABAergic and glycinergic spinal system function is the apparent tonic role they play in regulating afferent processing that would otherwise be non-noxious in character. Transient block of spinal GABA_A (bicuculline/picrotoxin) or glycine (strychnine) receptors reveals the prominent tonic role played by the GABA and glycine systems. Early work demonstrated that low-threshold tactile stimuli, typically ineffective in producing evidence of escape behavior, were able to evoke powerful pain behavior after spinal antagonism of GABA and glycine receptors (Yaksh 1989). Such observations are in concert with yet earlier studies on the activity of spinal/trigeminal single units noted above. These observations raise the likelihood that the encoding of low-threshold mechanical stimuli as innocuous depends completely on the presence of tonic activation of intrinsic glycine/GABAergic neurons that are known to exist within the spinal dorsal horn. Several lines of evidence substantiate the relevance of these dorsal horn inhibitory amino acids in regulating the behavior generated by low-threshold afferent transmission. Thus, genetic variants such as the poll Hereford calf (Gundlach et al 1988) and the spastic mouse (White and Heller 1982) have been shown to display particular sensitivity to even modest stimulation, and these models exhibit up to a 10-fold decrease in glycine binding. Second, GABA_A knockout mice exhibit tactile allodynia similar to that observed with spinal GABA_A receptor antagonism (Yaksh 1989, McGowan and Hammond 1993, Onaka et al 1996). The mechanical allodynia probably reflects the loss of GABA_A receptors presynaptic to low-threshold Aβ (Gmelin and Zimmermann 1983) and Aδ (De Koninck and Henry 1994) primary afferent fibers. Such behavioral evidence of the effects of hypersensitivity would be consistent with loss of post-synaptic GABA_A receptors on spinofugal neurons (Lin et al 1996). Third, in humans, strychnine intoxication is characterized by hypersensitivity to light touch (Arena 1970), and the role of such interneurons in the encoding of afferent input has been suggested as an important mechanism involved in the allodynia and hyperesthesia evoked following spinal cord ischemia (Hao et al 1992a, 1992b; Marsala and Yaksh 1992) and peripheral nerve injury (Yaksh et al 1992).

Adenosine Systems

Adenosine is generated from ATP by ectonucleotidases that dephosphorylate ATP, adenosine diphosphate (ADP), and adenosine monophosphate (AMP) (Zimmermann et al 1998). Extracellular adenosine levels can be elevated at the spinal cord level by local depolarization (K⁺ and capsaicin) (Sweeney et al 1989) and by a wide variety of pharmacological interventions, including the direct effects of 5-HT, noradrenaline, and opiates.

Receptors

Four receptor subtypes (A₁, A_2A, A_2B, and A₃) have been identified based on their pharmacology and cloning (Fredholm et al 2001). The three classes of receptors are all G protein coupled, with A₁ and A₃ typically being inhibitory (reduced cAMP) and the A₂ families being excitatory (increased cAMP) (Schulte and Fredholm 2003). The spinal distribution of these adenosine receptors has been established with autoradiography and immunohistochemistry. Both approaches have shown adenosine A₁ binding receptor protein in the superficial dorsal horn spinal cord (laminae I and II). This binding/protein was reduced by local kainate, but not by rhizotomy (Geiger et al 1984, Choca et al 1988, Deuchars et al 2001b), thus suggesting localization on interneurons. Adenosine A_2A receptor message is present in DRGs (Kaelin-Lang et al 1998) but not in the spinal cord (Kaelin-Lang et al 1999), thus suggesting their presence on the spinal terminals of sensory afferents.

Physiological Effect

Adenosine A₁ receptor activation increases K⁺ conductance, thereby leading to hyperpolarization and to reduced frequency of opening of voltage-sensitive calcium channels (Dunwiddie and Masino 2001). Examination of the effects of A₁ receptor agonists on primary afferent terminal–releasing properties indicates inhibition of release of the primary afferent stores of CGRP, but not SP (Santicioli et al 1992, 1993; Mauborgne et al 2002). Electrophysiological studies have further shown that adenosine acts presynaptically through A₁ receptors to diminish monosynaptic Aδ and C fiber–evoked depolarization in some superficial dorsal horn neurons (Lao et al 2001, Patel et al 2001).

Behavior

The intrathecal delivery of adenosine mediated by A₁-type receptor pharmacology produces modest increases in acute nociceptive thresholds (Sosnowski et al 1989) but also induces significant anti-hyperalgesic action in models of post–inflammation-induced thermal hyperalgesia (Poon and Sawynok 1998), in the formalin model (Malmberg and Yaksh 1993a), and in models of neuropathic pain associated with glycine inhibition (Sosnowski and Yaksh 1989) and nerve injury (Lee and Yaksh 1996, Khandwala et al 1998, Poon and Sawynok 1998). When examined, these effects were not observed with adenosine A₂–preferring agonists. Consistent with these behavioral effects, intrathecal adenosine A₁ activation decreased the incidence of c-Fos–positive cells in the superficial and deep dorsal horn neurons evoked by peripheral inflammation (Sorkin et al 2003).

Important aspects of the actions of endogenous adenosine are its rapid uptake and inactivation. Accordingly, it is important to note that spinal delivery of adenosine kinase or adenosine deaminase inhibitors will of itself have significant anti-hyperalgesic effects that appear to possess an adenosine A₁ antagonist pharmacology (Keil and DeLander 1992, Poon and Sawynok 1998, Lavand’homme and Eisenach 1999, McGaraughty et al 2001). These observations provide confirmation of the probable contribution of the endogenous release of adenosine to local spinal modulation.

Endogenous Agents

Identification of the presence of cannabinoid binding sites has led to discovery of endogenous agents that act on these receptors. N-Arachidonyl ethanolamide (anandamide) was identified first (Devane et al 1992), followed by a second endocannabinoid, 2-arachidonylglycerol (Mechoulam et al 1995, Sugiura et al 1995). These molecules exhibit affinity for both the CB1 and CB2 receptors (Guindon and Hohmann 2009). Local microdialysis during brain stem stimulation revealed that parameters found to be antinociceptive and antagonized by CB antagonists were associated with an increase in the release of several endocannabinoids (Guindon and Hohmann 2009). Similarly, injections of formalin into the hindpaws of rats induced the release of anandamide in the PAG (Walker et al 1999). These molecules are long-chain lipids, and there has been interest in the hypothesis that they may act at additional sites. In particular, the vanilloid (TRPV1) receptor can be activated by anandamide in a manner blocked by TRPV1 receptor antagonists (Zygmunt et al 1999).

Receptors

Based on expression and pharmacology, two cannabinoid receptor families have been identified. Although both CB1 and CB2 are negatively coupled to adenylyl cyclase, they show downstream differences. CB1 is negatively coupled through G_i/o (Felder et al 1995) to N- and P/Q-type calcium channels (Caulfield and Brown 1992, Mackie et al 1995) and positively coupled to inwardly rectifying potassium channels (Deadwyler et al 1995, Mackie et al 1995). CB2 is not coupled to either calcium Q-type or inwardly rectifying potassium channels (Felder et al 1995). In DRGs, CB1 receptors appear to be coupled primarily to N-type calcium channels (Khasabova et al 2004). The presence of cannabinoid binding in the spinal cord has been well characterized (Herkenham et al 1991). Capsaicin treatment produces a modest reduction and rhizotomy produces a small yet greater reduction in binding (Hohmann and Herkenham 1998, Hohmann et al 1999). Consistent with this proposed distribution, CB1 mRNA and protein are highly expressed in DRG cells of heterogeneous size (Sanudo-Pena et al 1999, Hohmann 2002), including TRPV1-positive cells (Ahluwalia et al 2000). CB2 receptor protein has been identified in DRG cell cultures (Ross et al 2001). These results suggest for the CB1 receptor a modest presynaptic localization in small capsaicin-sensitive and non–capsaicin-sensitive afferents, with the preponderance of post-synaptic expression in the spinal dorsal horn. In the spinal cord, CB1 message has been identified in all spinal laminae except IX (Mailleux and Vanderhaeghen 1992). Immunohistochemistry has indicated CB1 receptor protein in laminae I and II inner and lamina X (Farquhar-Smith et al 2000, Salio et al 2002). At least some dorsal horn immunoreactivity is located on GABA-positive interneurons (Salio 2002). The CB1 receptor, though constitutively expressed at a lower level in the spinal cord, displays marked up-regulation in the spinal cord and DRG after nerve injury and persistent inflammation (Hsieh et al 2011).

Physiological Effects

In culture, CB1 agonists were observed to reduce the entry of Ca²⁺ in DRG cultures, thus suggesting a probable effect on terminal release (Ross et al 2001). With regard to primary afferent release, in isolated DRG cell cultures, opiates but not CB1 agonists diminish release of CGRP (Khasabova et al 2004). In vitro recording suggests that CB1 activation results in inhibition of dorsal horn glutamate-mediated activation through inhibition of its release onto lamina II glycine/GABAergic interneurons (Morisset and Urban 2001). No evidence of a post-synaptic effect was noted given that AMPA-kainate–mediated post-synaptic current was unaltered. These results are consistent with the demonstrated co-localization of CB1 receptor protein with GABA-positive neurons (Jennings et al 2001). Intrathecal delivery of CB1 agonists suppressed noxious heat–evoked activity in dorsal horn WDR neurons. These results are consistent with the ability of CB1 agonists to suppress stimulus-evoked Fos expression in the lumbar spinal cord (Tsou et al 1996). Conversely, CB1 but not CB2 receptor antagonism exaggerated acute nociceptive but not non-nociceptive transmission at the level of the spinal cord, thus suggesting a tonic role of the endogenous cannabinoids at spinal CB1 receptors (Chapman 1999).

Behavioral Effects

Intrathecal and supraspinal (see below) delivery of a CB agonist produces a significant antinociceptive effect in a variety of models, including thermal escape (Yaksh 1981, Smith and Martin 1992; for an extensive review see Pertwee 2001). Current work emphasizes the clear importance of the CB1 receptor in mediating antinociception after spinal or systemic delivery. Knockout of the CB1 receptor (Ledent et al 1999, Zimmer et al 1999) or knockdown with antisense oligonucleotides (Edsall et al 1996) significantly reduces the effects of antinociceptive CB treatment. Importantly, consistent with up-regulation of the CB2 receptor after nerve injury and persistent inflammation, intrathecal CB2a agonists were effective in attenuating injury-evoked hyperalgesia (Hsieh et al 2011).

Cholinergic Systems

In the spinal dorsal horn, cholinergic soma, as defined by choline acetyltransferase immunoreactivity, are present in laminae III, V, and X (Borges and Iversen 1986). A dense plexus of cholinergic terminals extends through laminae II and III (Sherriff and Henderson 1994). Pharmacological studies have suggested that the dorsal horn effects of acetylcholine are mediated through nicotinic and muscarinic receptors.

Cholinergic Receptors

Physiological Effects

Behavioral Effects

Neurotensin System

Neurotensin is a tridecapeptide. Neurotensin-positive terminals and cell bodies are present in small non-GABAergic interneurons in laminae I and II (Proudlock et al 1993).

Neurotensin Receptors

Neurotensin is believed to exert its effects through at least three cloned receptors (Vincent et al 1999).

Physiological Effects

Local delivery of neurotensin in the spinal dorsal horn excites spinal nociceptors (Stanzione and Zieglgansberger 1983).

Behavioral Effects

Intrathecal delivery of this peptide has been reported to yield signs of both excitation (Seybold et al 1982) and antinociceptive actions in thermal and chemical nociceptive models (Yaksh et al 1982b).