At nanomolar levels, NO binds to a prosthetic heme of NO-sensitive guanylyl cyclase (NO-GC; also referred to as soluble guanylyl cyclase, sGC) and causes the conversion of GTP to cGMP (Francis et al. 2010). NO-GC is a heterodimer consisting of two different subunits termed α and β. Two catalytically active isoforms have been identified (α₁β₁ and α₂β₁) in which the β₁ subunit acts as the dimerizing partner for the α₁ or α₂ subunit (Friebe et al. 2007). There is considerable evidence that NO-GC is a major NO target during pain processing. Mice deficient for the β₁ subunit (GC-KO mice), which are completely devoid of NO-GC activity, failed to develop pain sensitization induced by intrathecal administration of NO donors. GC-KO mice also demonstrated considerably reduced pain behaviors in inflammatory and neuropathic pain models, whereas the immediate responses to acute nociceptive stimuli were normal (Schmidtko et al. 2008a). The important role of NO-GC for persistent pain processing in the spinal cord is further supported by antinociceptive effects of the NO-GC inhibitor ODQ after intrathecal injection in models of inflammatory and neuropathic pain (Ferreira et al. 1999; Kawamata and Omote 1999; Tao and Johns 2002; Song et al. 2006). Moreover, similar to NO donors, both pronociceptive and antinociceptive effects were observed after i.t. administration of cGMP analogs (Garry et al. 1994; Iwamoto and Marion 1994; Ferreira et al. 1999; Song et al. 2006), and again the administered dose seems to be a determinant for this dual effect (Tegeder et al. 2002, 2004; Schmidtko et al. 2008b). Dual effects of NO and cGMP were also observed in electrophysiological studies with spinal cord slices, in which superfusion with both NO donors and cGMP analogs inhibited ~50 % but activated ~30 % of dorsal horn neurons (Pehl and Schmid 1997).

The most likely reason for the dual effects of NO donors and cGMP analogs is the presence of different pronociceptive and antinociceptive NO/cGMP downstream signaling mechanisms. Unlike the membrane-permeable gas NO, cGMP mainly acts in intracellular compartments at its site of production. Interestingly, the expression pattern of NO-GC in the spinal cord and in DRGs suggests that NO-mediated cGMP production can modulate pain processing at different sites. In the spinal cord, NO-GC immunoreactivity is enriched in inhibitory interneurons in laminae II and III, i.e., in the area of highest nNOS expression (see above). NO-GC is also expressed in neurokinin 1 (NK₁) receptor-positive projection neurons in lamina I (Ding and Weinberg 2006; Ruscheweyh et al. 2006; Schmidtko et al. 2008a). These cells not only contribute to the ascending conduction of pain but are also essential for NO-dependent long-term potentiation (LTP) at the first synapse in pain pathways (Mantyh and Hunt 2004; Ikeda et al. 2006). In DRGs, however, specific NO-GC immunoreactivity was unexpectedly not detected in neurons. Instead thereof, NO-GC protein seems to be present only in satellite cells and vascular cells (Schmidtko et al. 2008a). This finding is supported by observations that axotomy of the sciatic nerve or incubation of DRG sections with an NO donor initiated cGMP production selectively in non-neuronal DRG cells (Morris et al. 1992; Shi et al. 1998). Considering that peripheral nerve injury leads to nNOS upregulation in somata of DRG neurons (see above) and that satellite cells contain NO-GC, it is likely that NO acts as a paracrine messenger from DRG neurons to satellite cells, thereby possibly contributing to the satellite cell proliferation in response to peripheral nerve injury (Zhuang et al. 2005; Scholz and Woolf 2007; Zhang et al. 2007; Kawasaki et al. 2008). Importantly, the observation that NO-GC is not expressed in primary afferent neurons challenges an earlier hypothesis that NO might act as “retrograde” transmitter which is released by spinal cord neurons and stimulates cGMP production via NO-GC activation in primary afferent neurons (Meller and Gebhart 1993; Luo and Cizkova 2000). Instead thereof, NO seems to be primarily a transmitter that (1) is released from nNOS-positive DRG neurons and dorsal horn interneurons (and possibly from so far unidentified iNOS-positive cells) and (2) induces cGMP production in DRG satellite cells, in lamina I projection neurons, and in laminae II/III inhibitory interneurons (Schmidtko et al. 2009).

The elucidation of downstream mechanisms of NO/cGMP signaling in the nociceptive system has been complicated by at least two facts: First, cGMP in general signals by various mechanisms including activation of cGMP-dependent protein kinase (cGK; also referred to as protein kinase G, PKG), activation of cyclic-nucleotide-gated (CNG) channels, modulation of hyperpolarization-activated and cyclic-nucleotide-gated (HCN) channels, and modulation of phosphodiesterases (PDEs) (Craven and Zagotta 2006; Feil and Kleppisch 2008). Recent data indicate that all these cGMP targets are present in the nociceptive system. Second, cGMP is produced not only by NO-GC but also in a NO-independent manner by particulate guanylyl cyclases in response to stimulation by natriuretic peptides. Seven particulate guanylyl cyclase isoforms activated by different ligands have been identified in rodents (Garbers et al. 2006), and particulate guanylyl cyclases A and B (GC-A and GC-B; also referred to as natriuretic peptide receptor A [NPR-A] and natriuretic peptide receptor B [NPR-B], respectively) have been detected in DRG neurons (Schmidt et al. 2007; Kishimoto et al. 2008; Schmidtko et al. 2008a; Zhang et al. 2010; Loo et al. 2012).

After the discovery of pain-relevant NO/cGMP signaling in the 1990s, it was initially thought that most effects of NO and cGMP are mediated by cGKI (Qian et al. 1996), corresponding to the functional NO/NO-GC/cGMP/cGKI signaling pathway that exists in many other tissues (Feil and Kleppisch 2008). More recent studies confirmed the important pain-relevant role of cGKI, but cGKI seems to be mainly activated by NO-independent mechanisms during pain processing (see below). Several immunohistochemical studies detected the α-isoform of cGKI in the majority of DRG neurons and their nerve terminals in the spinal cord and in some dorsal horn neurons (Qian et al. 1996; Tao et al. 2000; Sung et al. 2006; Schmidtko et al. 2008b; Luo et al. 2012; Lorenz et al. 2014). After peripheral nerve injury and inflammation, cGKIα is activated in DRG neurons (Sung et al. 2004, 2006; Lorenz et al. 2014), and its expression increases in the spinal cord (Tao et al. 2000; Tegeder et al. 2002; Schmidtko et al. 2003). The essential contribution of cGKIα to persistent pain processing is reflected by the reduced inflammatory and/or neuropathic pain behavior in global or nociceptor-specific cGKI mutants (Tegeder et al. 2004; Luo et al. 2012; Lorenz et al. 2014) and by profound antinociceptive effects of intrathecally administered cGKI inhibitors (Tao et al. 2000; Schmidtko et al. 2003, 2009; Luo et al. 2012; Lorenz et al. 2014). So far identified targets that are phosphorylated by cGKIα in DRG neurons include cysteine-rich protein 4 (CRP4; initially named CRP2, Schmidtko et al. 2008b), vasodilator-stimulated phosphoprotein (VASP), myosin light chains (MLC), inositol 1,4,5-triphosphate receptor 1 (IP₃R1) (Luo et al. 2012), and possibly large-conductance Ca²⁺-activated K⁺ channels (BK_Ca) (Zhang et al. 2010; Lu et al. 2014).

However, consistent with the cellular distribution of cGKIα and NO-GC described above, double-immunohistochemical stainings confirmed that cGKI and NO-GC are not colocalized in DRGs and only partially colocalized in the spinal cord (Schmidtko et al. 2008a). This implicates that upstream mechanisms different from NO and NO-GC may activate cGKIα during pain processing. Indeed, several studies demonstrated that the particulate guanylyl cyclases GC-A and GC-B are colocalized with cGKIα in DRG neurons and mediate cGKIα activation after stimulation with natriuretic peptides (Schmidt et al. 2007; Kishimoto et al. 2008; Schmidtko et al. 2008a; Zhang et al. 2010). In addition, an alternate mechanism of cGMP-independent cGKIα activation has been recently discovered in DRG neurons: Oxidants such as hydrogen peroxide (H₂O₂) can cause interprotein disulfide bond formation between two cGKIα cysteine residues, rendering the kinase catalytically active, independently of cGMP (Burgoyne et al. 2007). Interestingly, H₂O₂-induced cGKIα disulfide bond formation was increased in DRGs after peripheral nerve injury, and knock-in mice with impaired H₂O₂ activation but normal cGMP activation of cGKIα demonstrated reduced neuropathic pain behaviors (Lorenz et al. 2014). Hence, both cGMP derived from particulate guanylyl cyclases and H₂O₂ derived from so far unidentified sources activate cGKIα in DRGs during pain processing. In contrast, NO and NO-GC seem to use targets different from cGKIα to mediate their pain-relevant effects in DRGs.

In a recent study, CNG channels were identified as a novel target of NO signaling during pain processing: Using in situ hybridization experiments, the CNG channel subunit CNGA3 was detected in inhibitory neurons of the dorsal horn and in DRG satellite cells. After hindpaw inflammation, CNGA3 expression was upregulated in the dorsal horn and in DRGs, and mice lacking CNGA3 (CNGA3^−/−) showed increased inflammatory pain behaviors. Moreover, the pain hypersensitivity evoked by i.t. delivery of cGMP analogs and NO donors was increased in CNGA3^−/− mice (Heine et al. 2011), indicating that CNGA3-positive CNG channels are a downstream target of NO signaling that contributes in an inhibitory manner to persistent pain processing. Further studies are required to identify additional downstream targets that mediate the pro- and antinociceptive effects of NO-mediated cGMP production.

A cGMP-independent mechanism of NO signaling is S-nitrosylation, i.e., the covalent and reversible attachment of NO to a reactive cysteine thiol (Hess et al. 2005). Several recent in vitro and ex vivo studies indicate that S-nitrosylation is a signaling mechanism of NO during pain processing (for review, see Tegeder et al. 2011). For example, whole-cell recordings of rat spinal cord slices revealed that NO may S-nitrosylate voltage-activated Ca²⁺ channels, thereby reducing glutamate release from primary afferent terminals (Jin et al. 2011). Unlike this antinociceptive mechanism, S-nitrosylation of actin was reported to ameliorate inhibitory postsynaptic currents in the spinal dorsal horn (Lu et al. 2011). In DRG neurons, NO was found to activate ATP-sensitive potassium channels by S-nitrosylation of cysteine residues in the SURI subunit, and this effect was not blocked by inhibitors of NO-GC or cGKI (Kawano et al. 2009). Furthermore, NO directly activated TRPV1 and TRPA1 channels in isolated inside-out patch recordings (Miyamoto et al. 2009), and it seems likely that this effect is also mediated by S-nitrosylation (Yoshida et al. 2006). In a recent proteomic approach using two-dimensional S-nitrosothiol difference gel electrophoresis and S-nitrosylation-site identification in spinal cord extracts, more than 50 proteins with modified S-nitrosylation in response to peripheral nerve injury were detected. The modified proteins are involved in synaptic signaling, protein folding and transport, mitochondrial function, and redox control (Scheving et al. 2012). The functional contribution of most of these proteins to pain processing is currently unknown; however, it seems very likely that S-nitrosylation essentially contributes to cGMP-independent NO signaling in the nociceptive system.