Introduction
The present study focused on the role of intracellular signaling mechanisms in the amygdala in pain-related plasticity and behavior. The amygdala is now recognized as an important player in the emotional-affective dimension of pain [
1‐
9]. The laterocapsular division of the central nucleus of the amygdala (CeLC) is of particular importance, because it receives nociceptive ("pain-related") information directly from spinal cord and brainstem and indirectly, through the lateral-basolateral amygdala circuitry, from thalamus and cortex [
1,
8]. Our previous studies demonstrated central sensitization [
10‐
15] and synaptic plasticity [
10,
16‐
19] in the CeLC in the kaolin/carrageenan-induced arthritis pain model. Recent imaging data also showed increased amygdala activation related to knee pain in patients with osteoarthritis [
20]. Pain-related synaptic plasticity in the CeLC was confirmed in a model of chronic neuropathic pain [
3] and was mimicked by tetanic stimulation of presumed nociceptive inputs from the brainstem [
21].
A consequence of pain-related amygdala activation is increased pain behavior. Deactivation of the central nucleus decreased nocifensive and affective behavior associated with arthritic [
9,
10,
22], formalin-induced [[
2]; but see Tanimoto et al., 2003], visceral [
23‐
25], and neuropathic pain [
4]. However, the amygdala is also important for pain inhibition, particularly in the context of stress-induced and conditioned forms of analgesia [
26‐
32]. The conditions under which the amygdala assumes pro- or anti-nociceptive functions and the underlying mechanisms remain to be determined.
Arthritis pain-related synaptic plasticity and central sensitization in the CeLC require the upregulation of presynaptic metabotropic glutamate receptors [
12,
16] and increased postsynaptic NMDA receptor function through a mechanism that involves NR1 phosphorylation by PKA [
13,
17]. Pain-related PKA activation in the CeLC appears to occur downstream of calcitonin gene-related peptide receptor CGRP1 [
10] and corticotropin-releasing factor receptor CRF1[
11,
33]. Protein kinases such as PKA, PKC, and ERK, play important roles in the central sensitization of spinal cord neurons [
34‐
40]. The effects of PKA and PKC activators on spinal transmission and excitability were blocked by inhibitors of ERK signaling, suggesting that PKA and PKC are upstream activators of ERK in the spinal cord [
39,
40].
Pain-related functions and interactions of protein kinases, including PKA, PKC, and ERK, in the amygdala are largely unknown. A recent biochemical and behavioral study showed ERK activation in the CeLC in the formalin pain model and antinociceptive effects of inhibiting ERK activation in the CeLC [
2]. The present study used a multidisciplinary approach at the cellular and system levels to determine the effects of selective inhibitors of PKA, PKC, and ERK in the amygdala on pain-related synaptic plasticity and behavior. We focused on these protein kinases because they are important for spinal central sensitization and can phosphorylate the NMDA receptor [
41‐
43], which is a critical mechanism of arthritis pain-related plasticity in the amygdala [
17].
Discussion
The key findings of this study are as follows. Inhibition of PKA or ERK, but not PKC, in the CeLC decreases NMDA receptor-mediated synaptic plasticity in the arthritis pain model but has no effect on basal transmission under normal conditions. PKA and ERK inhibitors administered together do not occlude each other's action but have additive effects, suggesting independent signaling pathways for PKA and ERK. PKA activation by forskolin under normal conditions induces an NMDA receptor-mediated synaptic component that mimics synaptic facilitation observed in the arthritis model. This effect is not blocked by the inhibition of ERK activation, arguing against a role of ERK downstream of PKA. Consequently, inhibitors of PKA and ERK, but not PKC, in the CeLC decrease supraspinally (vocalizations) and spinally (withdrawal reflexes) organized pain behaviors in animals with arthritis but not in normal animals.
The significance of these results is that in the amygdala PKA and ERK, but not PKC, modulate information processing and behavior through separate (not serially arranged) signaling pathways. This is different from pain-related plasticity in the spinal cord [
39,
40] and from other models of plasticity such as hippocampal long-term potentiation (LTP) [
61], where PKA and PKC act in concert to activate ERK. In dorsal horn neurons activation of PKA, PKC, or ERK increased neuronal excitability and inhibited transient potassium (A-type) currents. The effects of PKA and PKC activators were blocked by inhibitors of ERK signaling, suggesting that PKA and PKC act as upstream activators of ERK [
39,
40]. Spinal PKA and PKC activation has also been implicated in central sensitization [
62] and behavioral hypersensitivity [
63‐
65] in different pain models. More recent studies showed ERK activation and antinociceptive effects of ERK inhibition in the spinal cord in several pain models [reviewed by [
35]]. The lack of evidence for the involvement of PKC in the present study was somewhat surprising. However it has been pointed out before that "studies on the effects of PKC on NMDA receptors have yielded conflicting results, probably because PKC has multiple effects depending on cell type, sites of action, and variable associations of NMDA receptors with other proteins" [
42].
Our data suggest that NMDA receptors are the target of PKA and ERK. NMDA receptors have been shown to function as "upstream" activators of protein kinases. NMDA receptors couple directly [
66] or via PKA and PKC [
61,
67] to ERK activation and are involved in pain-related ERK activation in the spinal dorsal horn [see [
35]]. NMDA receptor dependent ERK activation plays an important role in the central sensitization of dorsal horn neurons [
68]. However, NMDA receptors are also "downstream" targets of protein kinases. PKA, PKC, and ERK can phosphorylate NMDA receptors to enhance current flow through the receptor and accelerate the kinetics of the ion channel [
41‐
43,
69‐
71]. PKC mediated NMDA receptor phosphorylation removes the magnesium block [
72], rendering the channel functional even at normal resting membrane potentials as observed in the present study. Pain-related NMDA receptor phosphorylation of spinothalamic tract (STT) cells in the deep dorsal horn requires both PKC and PKA, whereas phosphorylation in superficial dorsal horn STT cells is due to the action of PKA [
73]. The contribution of ERK-mediated NMDA receptor phosphorylation to pain-related neuronal and behavioral changes remains to be determined, but a recent study showed ERK-mediated NMDA receptor phosphorylation by brain-derived neurotrophic factor (BDNF), which can modulate nociceptive transmission in the spinal dorsal horn [
41].
The effectiveness of protein kinase inhibitors in the present study suggests tonic NMDA receptor phosphorylation in amygdala neurons in the arthritis pain state. Kinetics of phosphorylation by PKA and ERK are fast (1–5 min) [
41,
41,
43]. PKA can overcome constitutive protein phosphatase activity and rapidly enhance NMDA receptor currents [see [
42]]. Blocking phosphorylation with PKA and ERK inhibitors would shift the balance from phosphorylation toward dephosphorylation by constitutively active phosphatases [
71]. For example, type I protein phosphatase (PP1) binds to an NMDA receptor-associated protein and decreases current flow through the channel [
70]. Striatal enriched tyrosine phosphatase (STEP) is a component of the NMDA receptor complex and can prevent hippocampal LTP without affecting normal synaptic transmission [
74]. STEP immunoreactivity is found in cell bodies in several brain areas, including the amygdala [
75]. Therefore, the negative regulation of NMDA receptor function by protein kinase inhibitors in the present study can be explained by the relative dominance of constitutively active phosphates.
The mechanisms leading to pain-related PKA and ERK activation in the amygdala remain to be determined. A variety of neuromodulator/neurotransmitter receptors, including metabotropic glutamate receptors that are important for pain-related plasticity in the amygdala [
12,
16], have been shown to couple to ERK activation via PKA and PKC [
61]. Evidence from our previous studies suggests that neuropeptide receptors CGRP1 and CRF1 contribute to pain-related changes in the amygdala through a mechanism that involves PKA activation [
10,
33]. If PKA and ERK are indeed activated through different mechanisms as the present study may suggest, neuropeptide receptors could activate PKA whereas metabotropic glutamate receptors could couple to ERK activation.
Some methodological aspects need to be considered. The conclusions of this study rely on the selectivity of the protein kinase inhibitors. The role of PKA was determined by using two compounds that inhibit PKA activation through different mechanisms. KT5720 is a widely used selective PKA inhibitor (at nanomolar to low micromolar concentrations) that binds to the catalytic subunits of PKA, causing the displacement of the regulatory subunit and thereby inhibiting the phosphorylating activity of the kinase [
50,
60]. cAMPS-Rp is a competitive antagonist of cAMP-induced activation of PKA (selective in the low to mid micromolar range) by interacting with cAMP binding sites on the regulatory subunits to prevent cAMP-induced dissociation and activation of the enzyme [
51]. Both inhibitors had similar effects. Although these compounds are membrane permeable, we showed that direct intracellular injection of KT5720 had the same effect as perfusion of the slice, confirming an intracellular site of action. U0126 is a well established, membrane-permeable and highly selective inhibitor of ERK activation (at nanomolar to low micromolar concentrations) by directly inhibiting the mitogen-activated protein kinase kinase family members, MEK-1 and MEK-2 [
52]. The MEK/ERK-selectivity of U0126 is supported by the fact that the inactive structural analogue U0124 had no effect. PKA and ERK inhibitors had additive effects that were not mimicked by a selective PKC inhibitor (GF109203x (Toullec et al., 1991), further arguing against non-specific effects.
In this study we used protein kinase inhibitors rather than activators, because we sought to determine the role of endogenously activated kinases. Exogenous activation of PKA with forskolin was used to determine the interaction with ERK. We did not test phorbol esters, which are commonly used to activate ERK, because they do so through PKC activation [
2,
61], which does not appear to be involved in arthritis pain related plasticity in our studies. Therefore, phorbol esters would not mimic the endogenous situation but possibly confound the analysis of ERK function.
Another issue concerns the use of microdialysis for drug application in the behavioral studies. Microdialysis offers several advantages, including continued drug delivery and steady state levels without a volume effect [
76]. However, the dose delivered by microdialysis is not known. Based on our previous microdialysis studies of similar-sized non-peptide compounds, we used drug concentrations in the microdialysis fiber that were 100 times higher than the target concentration in the tissue [
50,
51,
53] because of the concentration gradient across the dialysis membrane and diffusion in the tissue[
10‐
14,
33]. A "dilution" factor of 100 is further supported by the qualitatively and quantitatively similar effects of drug concentrations applied to the brain slices in the electrophysiological studies and those administered by microdialysis in the behavioral studies.
Finally, it may be surprising that the kinetics of the NMDA component and the compound EPSC were largely similar, whereas data in the literature suggest that NMDA receptors mediate slow EPSCs of relatively long duration [for recent review see [
77]]. In addition, NMDA receptor-mediated EPSCs could be recorded at a holding potential of -60 mV, where NMDA receptor channels are normally blocked by magnesium. The NMDA component was isolated pharmacologically with NBQX and bicuculline and was only present in slices from arthritic animals, which is consistent with our previous study [
17] that showed similar characteristics of NMDA receptor-mediated synaptic transmission in the amygdala in the arthritis pain model. The results can be explained by the effects of receptor phosphorylation. NMDA receptor phosphorylation relieves the magnesium block and renders the channel functional even at -60 mV [
72]. NMDA receptor phosphorylation by PKA or PKC also accelerates the rise and decay times of the ion channel [
69,
78], which explains the absence of apparent differences in the kinetics of NMDA EPSC and compound EPSC in the present study. However, our finding that an NMDA receptor-mediated component was difficult to detect under normal conditions even at depolarized membrane potentials (Figure
6) may suggest that PKA modulates NMDA receptor function through additional mechanisms such as synaptic targeting [
78].
In conclusion, the present study shows that PKA and ERK, but not PKC, are important for pain-related plasticity in the amygdala and for the behavioral consequences of this activity change. PKA and ERK target the NMDA receptor possibly through independent signaling cascades. PKA and ERK render normally "silent" NMDA receptors functional in the arthritis pain model. The independence of PKA and ERK signaling and the lack of PKC effects in this study are different from spinal central sensitization and hippocampal LTP and suggest that the role of protein kinases may be more specific than previously thought.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
YF and JH performed patch-clamp recordings, analyzed electrophysiology data, and provided figures and manuscript drafts. JH, TI, MS, HA, and CR obtained and analyzed behavioral data and provided figures and results in abstract form. TI also helped finalize the manuscript. VN conceptualized the hypothesis, designed and supervised the experiments, directed the data analysis, and finalized the manuscript. All authors read and approved the manuscript.