While PKMζ is well-recognized as a potential molecular mechanism for the maintenance of LTP and long-term memory [8, 40] and its important role in pain plasticity has recently been elucidated [11–15], neurotransmitter systems involved in the regulation of PKMζ have not been described in detail. Moreover, the specific role of PKMζ in CNS plasticity has recently been called into question with PKCλ emerging as a potential redundant mechanism in CNS plasticity [9, 10, 29]. Here we demonstrate that BDNF promotes persistent sensitization via a ZIP-reversible mechanism. Moreover, we show that BDNF is critical for both the initiation and maintenance of persistent sensitization, a role that it may uniquely share with an aPKC-dependent process [11, 13]. Linked to these in vivo findings, we further demonstrate that BDNF regulates PKCλ and PKMζ synthesis via an mTORC1-dependent pathway and PKMζ phosphorylation via PDK1 at spinal and cortical synapses. Importantly, we show definitively, for the first time, that both PKCλ and PKMζ are synthesized in an activity-dependent fashion at synaptic sites. Therefore, BDNF plays a key role in regulating aPKCs in the pain pathway elucidating a hitherto unrecognized pathway regulating the maintenance of a centralized chronic pain state.
PKMζ is an atypical PKC that was first recognized as a constitutively active kinase that may play a role in maintenance of late-LTP [8, 41]. Because PKMζ lacks a regulatory region, once translated, and phosphorylated by PDK1, the kinase has the potential to maintain autonomous activity over extended periods of time, satisfying theoretical considerations for a kinase-mediated mechanism maintaining late-LTP . This hypothesis has been borne-out by a body of subsequent work demonstrating a key role for PKMζ in maintaining late-LTP and also long-term memory [42, 43]. While parallels between molecular mechanisms of long-term memory and pain plasticity have long been recognized, only recently has PKMζ been elucidated as a potential target for maintenance of chronic pain states. PKMζ appears to play different roles in different anatomical locations in the pain pathway. PKMζ in sensory neurons is important for nerve growth-factor mediated hyperexcitability . PKMζ in the anterior cingulate cortex plays a key role in regulating tonic-aversive aspects of chronic neuropathic pain [12, 15]. Interestingly, a ZIP-reversible process in the spinal cord appears to play little, if any role in maintaining chronic neuropathic pain [12, 15], perhaps because this chronic pain state is critically dependent on ongoing afferent input to the spinal dorsal horn . In contrast, in chronic pain states wherein afferent input resolves but hypersensitivity either persists or can be rekindled by a normally subthreshold stimulus (e.g. persistent sensitization , CFA-induced inflammation  or chronic post ischemic pain [13, 46]) the maintenance of this pain state is reversed by spinal injection of ZIP. Our present findings expand on these previous results demonstrating that while CaMKIIα and MEK/ERK signaling is required for initiation of persistent sensitization, these kinases do not play an active role in the maintenance phase of persistent sensitization. These findings can be viewed as in contrast to other models, such as CFA, formalin, and/or incision, wherein ERK [47, 48] and CaMKIIα  play an important role in initiation and maintenance of a continuous hypersensitive pain state. Such differences, as mentioned above, may be related to afferent input engaged by these stimuli, which presumably resolves during the maintenance phase of the persistent sensitization model. We feel it is important to point out that i.t. drug applications during the maintenance phase are made when the mice show no overt signs of mechanical hypersensitivity. If these compounds were to be given at the same time as PGE2 injection an inhibitory effect might be expected because afferent input would be re-engaged, likely utilizing priming-dependent peripheral mechanisms that have recently been elucidated . These results, combined with our previous findings, strongly implicate aPKCs as the sole family of kinases responsible for the maintenance of persistent sensitization.
Despite the emerging role of PKMζ and potentially PKCλ in pain plasticity, mechanisms involved in aPKC regulation in the pain pathway are nearly completely unknown. We hypothesized that BDNF might play a key role in regulating aPKCs. This hypothesis was based on a known role of BDNF in pain states consistent [18–20] with the known effects consistent with an involvement of aPKCs. While BDNF can have several sources in the spinal dorsal horn, acutely it is released from nociceptors synapsing in the outer lamina of the dorsal horn [17, 19] where it regulates inflammatory but not neuropathic pain . BDNF also plays an important role in regulating LTP at dorsal horn synapses  consistent with the known role of BDNF in LTP in other CNS regions . These findings, combined with our present results, are consistent with a model wherein BDNF released from nociceptive endings in the spinal dorsal horn initiates signaling cascades that lead to the formation and phosphorylation of aPKCs at these synapses. Although spinal BDNF plays a role in neuropathic pain, as mentioned below, this has been linked to release from microglia [50, 51] and not nociceptor terminals because neuropathic pain develops normally in mice lacking BDNF expression in nociceptors . This finding is consistent with previous findings showing a limited role of a spinal ZIP-reversible process in neuropathic pain. We cannot, however, rule out an effect of microglial BDNF in our experiments. In that regard, BDNF is also known to play an important role in microglial activity  and neuropathic pain where it regulates GABAergic modulation of spinal circuits through disruption of Cl- homeostasis [51, 52]. Interestingly, this mechanism appears to be shared in morphine-induced hyperalgesia . Our findings from spinal SNS experiments clearly demonstrate that BDNF applied exogenously is capable of stimulating synthesis of PKCλ and PKMζ and phosphorylation of PKMζ. Whether BDNF released from microglia is incapable of achieving these effects at spinal synapses will have to await further experimentation. Although BDNF can have trkB-independent actions , we surmise that the effects of BDNF in our experiments were mediated by trkB due to the effect of the trkB antagonist ANA-12 .
An important implication of our current findings is that BDNF not only plays a role in initiating a centralized chronic pain state but that it also plays an active role in maintaining such a pain state via regulation of aPKCs. If this is the case, what is the source of BDNF? It is unlikely to be derived from presynaptic release from nociceptors because these sensory neurons are unlikely to be active after the resolution of IL-6-induced allodynia. It is also unlikely that microglia are the source because this would be inconsistent with the neuropathic pain findings for ZIP [12, 13, 15]. Important clues might be gleamed from the LTP literature wherein both pre- and post-synaptic release of BDNF regulates consolidation of late-LTP [55–57]. Interestingly, this likely involves alternatively spliced isoforms of BDNF in hippocampus  facilitating the possible recognition of such a mechanism being engaged in the spinal dorsal horn. While these experiments are outside of the scope of the present findings, this is likely to be a fruitful area of future research to gain a better understanding of maintenance mechanisms of a centralized chronic pain state.
Another important question raised by our findings relates to the dependence of maintenance of persistent sensitization on aPKCs but not protein synthesis. If BDNF regulates both PKCλ and PKMζ synthesis and PKMζ phosphorylation and initiation and maintenance of persistent sensitization are dependent on both aPKCs and BDNF but only initiation is dependent on protein synthesis, how is this seeming contradiction resolved? One possible explanation is that in the absence of protein synthesis, BDNF regulation of PKMζ phosphorylation is sufficient to maintain the chronic pain state. Interestingly, in spinal SNSs, BDNF stimulation of mTORC1 activity was transient whereas PDK1 mediated phosphorylation of both AKT and PKMζ was persistent. Hence, it is physiologically feasible that in the absence of protein synthesis, BDNF-mediated phosphorylation of PKMζ is sufficient to maintain persistent sensitization. Another possibility is that PKMζ, and possibly PKCλ, has an exceptionally long half-life at synapses. In this scenario, despite blockade of protein synthesis over long periods, aPKCs formed via previous protein synthesis would be capable of overcoming a lack of new protein availability due to its long half-life. Our preliminary observations (Melemedjian, Ghosh and Price, unpublished observations) support this model but ultimately require further experimentation. However, it is clear that BDNF can maintain late-LTP when protein synthesis is inhibited via a PKMζ-dependent mechanism  suggesting that similar mechanisms may be at play in the setting of persistent sensitization.
Importantly, we demonstrate that BDNF regulates aPKC formation in cortical SNSs in an analogous fashion to spinal SNSs. Insofar as both the maintenance of a centralized chronic pain state and long-term memory require both BDNF  and PKMζ , and considering that we demonstrate that BDNF regulates aPKCs across CNS structures, this illustrates the potential existence of a conserved pathway for the maintenance of synaptic plasticity from pain to memory. We propose that this has profound implications for understanding how mechanisms of plasticity evolved in central nervous systems and we suggest that these mechanisms might have first evolved for the most rudimentary neural function: protecting the organism against potentially lethal tissue injury. An important point moving forward will be to unveil how different aPKC isoforms contribute to pain plasticity through genetic models, as we have recently reviewed . This need is highlighted by the recent findings from the learning and memory literature showing that genetic removal of PKMζ fails to affect learning and memory despite the continued efficacy of ZIP in these animals, suggesting a potential redundant role of PKCλ in these pathways [9, 10]. It will be crucial to carefully examine the role of PKCλ in plasticity moving ahead.
In closing, we reveal that BDNF regulates the formation of PKCλ and PKMζ and phosphorylation of PKMζ and that BDNF/aPKC signaling forms a signaling axis required for the maintenance of a centralized chronic pain state. Our results imply that spinally directed therapeutics targeting BNDF and/or aPKCs might offer disease-modifying effects on certain chronic pain states in humans that are currently only treated by palliative management. The generation of such a class of therapeutics would have profound implications for the treatment of chronic pain.