The protective function of acute pain is lost when pain persists long after the initial injury is healed and changes into a chronic pain state. As a mechanistic basis, plasticity of neural substrates which mediate pain has been proposed, i.e. pain pathways are modifiable in a use-dependent manner or are subject to modulatory influences. Indeed, somatosensory nociceptive pathways can change dynamically over several scales of time as well as over scales of complexity, ranging from molecular, synaptic, cellular and network levels (Basbaum et al., 2009; Kuner, 2010; Prescott et al., 2014). Furthermore, tremendous complexity and dynamic range is given by structural reorganisation, which is complementary to and potentially causally associated with plasticity occuring at a functional level (Kuner and Flor, 2016).
Functional plasticity at spinal synapses and its determinants
A major task in understanding chronic pain is to study the integration in peripheral, spinal and supraspinal networks (Basbaum et al., 2009; Prescott et al., 2014). Excitability of dorsal horn neurons increases in several pathological pain states, and this increase has been demonstrated to be associated with a state of hypersensibility towards nociceptive stimuli (hyperalgesia) (Kuner, 2010; Sandkühler, 2009). Here, glutamate serves as the primary nociceptive neurotransmitter at the first synapse in the pain pathway, namely between the primary afferent and the second order neuron (Fig. 1). There is a large amount of evidence showing that the NMDA subtype of glutamate receptors (NMDAR) are key mediators of pathological hypersensitivity to pain, owing to their high calcium-permeability and Mg2+-block under physiological conditions. Unlike NMDAR, which always permit calcium entry upon activation, the AMPA subtype of glutamate receptors (AMPAR) encode a regulatable switch for controlling glutamate-evoked entry of calcium into neurons via inclusion of the subunit GluA2 (GluR-B or GluR2), which imparts low calcium-permeability to AMPAR channels because it carries an Arginine (R) residue in its pore-forming M2 segment via Q/R site RNA editing (Fig. 1) (Seeburg et al., 2001). Although relatively rare in nature, we observed that calcium-permeable AMPAR are expressed at a high density in the spinal dorsal horn, particularly in laminae 1 and 2, where primary afferents carry nociceptive and thermoreceptive inputs terminate and synapse onto spinal cord projection neurons and interneurons (Hartmann et al., 2004). This is consistent with the highly dense expression of the GluA1 subunit (GluR-A or GluR1 subunit) in lamina 1 and 2, and accordingly, in mice lacking GluA1, we observed a loss of nociceptive plasticity and a marked reduction in acute inflammatory hyperalgesia (Hartmann et al., 2004). In contrast, mice lacking the GluA2 subunit showed a facilitation of nociceptive plasticity and inflammatory hyperalgesia (Hartmann et al., 2004), consistent with the ability of GluA2 to impart reduced permeability to calcium in AMPAR and modification of current rectification and microscopic channel conductance. These and other studies have revealed that spinal calcium-permeable AMPAR are important in mediating activity-dependent changes in spinal processing of pain. Importantly, spinal calcium transients are known to be enhanced in pathological pain states, such as peripheral inflammation, and spread spatially across somatotopic borders, and these phenomena are also dependent upon the expression of calcium-permeable AMPAR (Luo et al., 2008; Sandkühler, 2009). Intense nociceptive activity also triggers phosphorylation of NMDAR in spinal dorsal horn neurons and thereby recruit several downstream plasticity-related kinases, such as ERK and CamKII (Luo et al., 2014). Activation of metabotropic glutamate receptors (mGluRs) results in activation of PLC and PKC-mediated signaling in spinal neurons (Fig. 1; (Luo et al., 2014). Collectively, these kinases bring about increased insertion of AMPAR subunits in the postsynaptic membrane (Fig. 1), resulting in increased calcium influx and increased excitability.
This scenario is very reminiscent of long term potentiation at hippocampal synapses and indeed long-term potentiation has also been reported at nociceptor synapses, particularly at neurons that project to key brain centers in pain pathway (Luo et al., 2012; Sandkühler, 2009). Driving C nociceptors at a frequency of 2 Hz, which is plausible in chronic pain states as well as upon acute injury, can lead to an increase in the amplitude of postsynaptic responses. Moreover, previous studies have demonstrated that low-frequency asynchronous nociceptor activity can also induce synaptic potentiation (Sandkühler, 2009). As a consequence, nitric oxide (NO) synthases, which are activated by calcium postsynaptically and can also be expressed in activated glia, can generate NO as a retrograde messenger. Our laboratory demonstrated that NO activates soluble guanylyl cyclase leading to cGMP production in the presynaptic terminals of nociceptors at spinal synapses (Luo et al., 2012); Fig. 1). We further went on to show that protein kinase G1 recruited by cGMP plays a key role in inducing long-term potentiation by increasing the probability of neurotransmitter release at synapses between C fibers and lamina 1 PAG projection neurons (Luo et al., 2012). In models of inflammatory pain, presynapse-specific deletion of PKG1 in nociceptor terminals led to a marked attenuation in nociceptive hypersensitivity in parallel to an abrogation of synaptic long-term potentiation. A long series of molecular and biochemical experiments revealed that these functions of PKG1 were mediated via phosphorylation of key presynaptic substrates, such as the IP3 receptor on intracellular calcium stores as well as the myosin light chain subunits (Luo et al., 2012) – the latter have been implicated in the recruitment of synaptic vesicles leading to enhanced neurotransmitter release.
A structural basis for functional plasticity
Thus, at spinal synapses between projection neurons and nociceptors, both pre- as well as postsynaptic mechanisms of synaptic potentiation can contribute to long lasting changes in synaptic strength. Structurally, these could be accounted by an increase in size of postsynaptic spines upon persistent nociceptive inputs or via increased spine density on postsynaptic neurons, thereby in both scenarios accounting for increased insertion of postsynaptic AMPA receptors. We and others have indeed found evidence for increased spine density on spinal neurons associated with enhanced nociceptive activity (Tan et al., 2012). For example, in models of inflammatory pain as well as several types of neuropathic pain, it has been demonstrated that lamina 2 neurons show increased spine remodeling, leading to an increase in the number of synaptic contacts (Fig. 2; (Simonetti et al., 2013). Mechanistically, dendritic spine stability as well as spine remodeling are closely associated with the activity of molecules called RhoGTPases, which include the RhoGTPase RhoA as well as Rac1 (Tolias et al., 2011). Whereas RhoA serves to disrupt dendritic spine stability via actomyosin coupling downstream of the Rho-dependent kinase (Rock), Rac1 leads to phosphorylation of cofilin thereby promoting dendritic spine stability. In the past years, pharmacological studies have implicated Rac in maintaining dendritic spine stability in the spinal dorsal horn in chronic pain states (Tan et al., 2012). However, given that pharmacological agents can act on a variety of different cells in the spinal cord, all of which ubiquitously express Rac1, we performed specific viral-mediated manipulations in excitatory neurons of the spinal dorsal horn. Our analyses demonstrated that a knockdown of Rac1 led to a decrease in spine density on neurons in the superficial spinal cord, while at the same time decreasing the intensity or magnitude of inflammatory pain in vivo (Lu et al., 2015). In contrast, over expression of Rac1, which presumably leads to increased constitutive activation of Rac1, led to an increase in spine density and thereby evoked pain hypersensitivity in mice already in the basal state (Lu et al., 2015), i.e. in the absence of nociceptive insults, such as inflammation or neuropathy. These data thereby indicate a tight correlation between structural changes induced by Rac1 and functional plasticity seen in the spinal dorsal horn in the context of nociceptive hypersensitivity.
It is well appreciated that glutamatergic receptors are anchored at synapses by a variety of anchoring and interacting proteins which further serve as links to signaling entities. Indeed, a variety of such protein-protein interactions have been described to be functionally active and important in nociceptive hypersensitivity at spinal nociceptive synapses (Kuner, 2010). However, one component of synaptic spines, which has not received appreciable attention, is actin itself. Indeed, actin rearrangement is required for any type of structural plasticity to occur at synapses. Both RhoA and Rac1 have been implicated in dendritic spine remodeling by active reorganisation of synaptic actin (Tolias et al., 2011). The question therefore arises as to how the glutamatergic plasticity mechanisms that were outlined in the sections above bring about actin rearrangement to remodel spinal synapses. The molecular link is given by proteins of the family of guanine exchange nucleotide factors (GEFs), which are recruited by activated receptors in response to extracellular signals and lead to activation of Rac via a transfer between the GDP-bound state (inactive) to a GTP-bound state (active). We therefore explored the molecular link between excitatory receptors and their signaling to Rac activation at spinal nociceptive neurons. As a prerequisite, we hypothesized that ideal candidate proteins must carry domains which enable physical interactions with glutamatergic receptors or their signaling effectors on the one hand and also carry domains that lead to Rac activation via GEF activity on the other hand. A detailed literature survey revealed that the Kalirin family of synaptic proteins may serve as a promising bridge linking glutamatergic receptors to RhoA and Rac modulation in neurons (Rabiner et al., 2005). The kalirin gene leads to the expression of a variety of splice variants, which differentially carry GEF domains for Rac or RhoA activation, as well as other types of domains, such as serine threonine kinase activity (Rabiner et al., 2005). One important splice variant of the kalirin gene, Kal-7, has the unique additional perspective of interacting with the NMDA receptor–PSD95 complex by virtue of its PDZ-binding motif (Rabiner et al., 2005). Our expression analyses show that the Kal-7 isoform is indeed well-expressed in the spinal dorsal horn, in addition to its prominent and restricted expression in the cortex and hippocampus. Using the approach of viral-mediated gene delivery to express the Cre recombinase unilaterally in spinal dorsal horn neurons, we achieved a specific knockout of the Kal-7 splice variant of the kalirin gene in excitatory neurons of the spinal dorsal horn (Lu et al., 2015). These mice did not show a compensatory change in the expression of the other kalirin splice variants, such as Kal-9 and Kal-12. We were particularly intrigued by Kal-7 protein because it has been recently demonstrated to modulate the synaptic insertion of AMPA receptor subunits (Rabiner et al., 2005), thereby fulfilling all criteria for linking functional and structural changes at spinal synapses. We observed that a lack of Kal-7 in excitatory neurons of the spinal cord suppresses the activity-dependent development of new spines in vitro as well as in vivo. In spinal neurons lacking Kal-7, we observed a decrease in the number of total spines which upon more detailed morphological analyses emerged as a specific attenuation in the number of new immature spines, namely thin and stubby spines (Lu et al., 2015). In contrast, the highly developed mushroom spines were not changed in their density. The same observations were made in vivo with the help of Golgi staining in mice with peripheral paw inflammation. Nociceptive activity-induced spine remodeling was thus entirely missing in mice lacking the Kal-7 protein in excitatory spinal neurons (Lu et al., 2015). Behaviorally, these mice showed a decreased duration in nocifensive responses in the phase 2 of the formalin test (an indicator of a lack of short-term plasticity in spinal dorsal horn neurons) as well as an attenuation in the development of inflammatory hypersensitivity (Lu et al., 2015). The circle was complete when we observed in electrophysiological recordings that infusion of a peptide disrupting interactions between Kal-7 and the NMDA receptor-PSD95 complex (Fig. 3) into lamina 1 PAG projection neurons led to a complete inability to develop synaptic potentiation upon persistent activation of C fibers (Lu et al., 2015). In neurons infused with a control peptide, synaptic potentiation could be elicited upon persistent C-fiber stimulation. Thus, taken together a specific deletion of Kal-7 in excitatory neurons of the spinal dorsal horn led to a decrease in inflammatory hypersensitivity and attenuation of spine remodeling while at the same time abrogating nociceptive activity-dependent synaptic potentiation. These results thereby indicate a tight link between structural and functional plasticity at spinal synapses and suggest a key role for Kal-7 in coordinating these structural and functional changes.
Long term structural modulation of spinal synapses
A key question that arises at this point is whether these acute synaptic changes can account for the really long-term nature of chronic pain. At this context, I do believe that genomic programming is an important aspect of pain chronicity. Indeed, novel omics technologies have revealed that chronic pain states are associated with very prominent transcriptional and epigenetic regulation in spinal dorsal horn neurons. However, these mechanisms are not entirely independent of synaptic activity, but are likely to rather represent a consequence of synaptic changes. Thus, we and others have shown that a variety of synapse-to-nucleus messengers can relay changes in synaptic activity to genomic coding programs in the cell nucleus of spinal neurons, which include kinases such as the extracellular receptor-regulated kinase (ERK), and second messengers such as cAMP (Kuner, 2010; Simonetti et al., 2013). One important synapse-to-nucleus messenger, which has emerged from our recent work, is calcium – indeed, upon an excessive increase in synaptic levels via activation of glutamatergic signaling, calcium can travel along the dendrites to the cell nucleus in spinal neurons and diffuse via the nuclear pores, leading to the activation of cAMP response element-binding (CREB) protein in the nucleus. We have recently shown that intense activation of C and Aδ nociceptors activates nuclear calcium signaling, thereby triggering a unique genomic program in spinal nociceptive neurons (Simonetti et al., 2013). This includes a variety of well-known pain-related genes, e.g. upregulation of cyclooxygenases leading to prostaglandin signaling, diverse synaptic molecules such as the AMPAR-binding protein GRIP, diverse proteases, GABA receptors, chemokines and several other targets that have been implicated in central sensitization as well as synaptic potentiation (Simonetti et al., 2013). However, the true beauty of such an approach is that it can uncover entirely novel molecules and mechanisms – indeed we observed that several components of the complement pathway were regulated by the unique genomic program triggered by nuclear calcium in spinal neurons. Amongst these, we studied C1q, the initiator of the complement cascade, which has been recently proposed to act as a synapse pruning factor (Simonetti et al., 2013). Our data indicate that upon persistent activation of nociceptors, nuclear calcium signaling in spinal neurons suppresses C1q expression, thereby enabling synaptic remodeling and permitting the transition from acute to subacute or chronic nociceptive hypersensitivity (Simonetti et al., 2013); Fig. 2).We have also previously demonstrated that a defense program against pain chronicity is encoded by nociceptive activity-induced expression of Homer1a in spinal neurons, which leads to synaptic pruning and remodeling, thereby alleviating inflammatory pain (Tappe et al., 2006).
These results indicate again that structural and functional plasticity go hand in hand in the pathophysiology of chronic pain and secondly, that genomic regulation can alter not only function but also the structure of synapses in a long-term manner.
Is dendritic spine plasticity in spinal neurons also relevant to pain therapy?
We analyzed this question in the context of analgesic actions of spinal cannabinoids. Using a proteomics approach, we pulled down the signaling complex of the key neuronal cannabinoid receptor, cannabinoid receptor-1 (CB1) (Njoo et al., 2015). Unexpectedly, we pulled down several key players of the so-called WAVE1 complex, which is a key regulator of actin assembly (Fig. 3). Activation of the WAVE1 complex leads to a conversion of G-actin into filamentous actin (F-actin) via the ARP2/3- complex, leading thereby to dendritic stabilization (Takenawa and Suetsugu, 2007). Rac1 is a key activator of the WAVE1 complex and we observed that CB1 interacts physically with the Rac1 as well as with the WAVE1 complex. Using a series of biochemical and enzymatic activity assays, we observed that cannabinoid-induced CB1 activation led to an inhibition of Rac1 and subsequently to a downregulation of WAVE1 activity in spinal neurons (Njoo et al., 2015). In contrast, inverse agonists at the CB1 receptor elicited an increase in basal Rac1 activity as well as WAVE1 activation. Both effects were lost in neurons derived from CB1 knockout mice, thereby indicating specificity. In assays in which the lifetime of filamentous fluorescent actin was measured in dendritic spines on mature spinal neurons in photobleaching experiments, we observed that CB1 activation suppresses actin turnover and actin assembly in synaptic spines (Njoo et al., 2015). Consequently, prolonged exposure of spinal excitatory neurons to cannabinoids markedly reduced the density of existing mature synaptic spines.
This was also confirmed in vivo in spinal neurons in the context of inflammatory pain – the increase in synaptic density induced by paw inflammation was abrogated when mice were treated intrathecally with a cannabinoid agonist (Njoo et al., 2015). In the same mice, we observed that intrathecal cannabinoid agonist application reversed inflammatory hypersensitivity (Njoo et al. 2015). Thus, a decrease in inflammatory hypersensitivity via exogenously-applied cannabinoids was directly associated with an abrogation of inflammation-induced increase in synaptic spine density. Using pharmacological as well as genetic approaches, we then went on to show that both structural and functional effects on spinal neurons induced by spinal cannabinoids were dependent upon the WAVE1 complex (Njoo et al., 2015). These data indicate that counteracting nociceptive activity-induced increase in synaptic spine density in the spinal cord is causally associated with the spinal analgesic action of cannabinoids.
It is particularly noteworthy that new data are also coming in on established therapeutics, such as Gabapentin (Matsumura et al., 2015), which support the relation between suppression of spine dynamics and analgesia.
In summary, the above studies indicate that structural plasticity of dendritic spines on dorsal horn neurons and synaptic as well as behavioral hypersensitivity to nociceptive and non-nociceptive stimuli go hand-in-hand. Glutamatergic signaling can recruit Rac1 activation via Kal-7 signaling at spinal synapses, leading to acute and subacute regulation of spine density. Chronic structural and functional regulation is achieved via synapse-to-nucleus translocation of messenger molecules such as ERKs and calcium. Nuclear calcium signaling triggers unique genomic programs in spinal dorsal horn neurons, which dynamically regulate the density of synaptic spines, including the downregulation of the pruning factor C1q.
Another take-home message is that dendritic spine remodeling in spinal neurons is relevant to pain therapy. Thus, inhibiting or reversing structural plasticity may offer viable options for novel forms of pain therapy.
Hartmann, B., Ahmadi, S., Heppenstall, P.A., Lewin, G.R., Schott, C., Borchardt, T., Seeburg, P.H., Zeilhofer, H.U., Sprengel, R., and Kuner, R. (2004). The AMPA receptor subunits GluR-A and GluR-B reciprocally modulate spinal synaptic plasticity and inflammatory pain. Neuron 44, 637–650. CrossrefPubMedGoogle Scholar
Lu, J., Luo, C., Bali, K.K., Xie, R.G., Mains, R.E., Eipper, B.A., and Kuner, R. (2015). A role for Kalirin-7 in nociceptive sensitization via activity-dependent modulation of spinal synapses. Nat. Commun. 6, 6820. Web of ScienceGoogle Scholar
Luo, C., Gangadharan, V., Bali, K.K., Xie, R.G., Agarwal, N., Kurejova, M., Tappe-Theodor, A., Tegeder, I., Feil, S., Lewin, G., et al. (2012). Presynaptically localized cyclic GMP-dependent protein kinase 1 is a key determinant of spinal synaptic potentiation and pain hypersensitivity. PLoS Biol. 10, e1001283. Web of SciencePubMedCrossrefGoogle Scholar
Luo, C., Seeburg, P.H., Sprengel, R., and Kuner, R. (2008). Activity-dependent potentiation of calcium signals in spinal sensory networks in inflammatory pain states. Pain 140, 358–367. Web of SciencePubMedCrossrefGoogle Scholar
Matsumura, S., Taniguchi, W., Nishida, K., Nakatsuka, T., and Ito, S. (2015). In vivo two-photon imaging of structural dynamics in the spinal dorsal horn in an inflammatory pain model. European J. of Neurosci. 41, 989–997. Web of ScienceGoogle Scholar
Njoo, C., Agarwal, N., Lutz, B., and Kuner, R. (2015). The Cannabinoid Receptor CB1 Interacts with the WAVE1 Complex and Plays a Role in Actin Dynamics and Structural Plasticity in Neurons. PLoS Biol. 13, e1002286. Web of ScienceCrossrefGoogle Scholar
Rabiner, C.A., Mains, R.E., and Eipper, B.A. (2005). Kalirin: a dual Rho guanine nucleotide exchange factor that is so much more than the sum of its many parts. Neuroscientist 11, 148–160. CrossrefGoogle Scholar
Sandkühler, J. (2009). Models and mechanisms of hyperalgesia and allodynia. Physio. Rev. 89, 707–758. Google Scholar
Seeburg, P.H., Single, F., Kuner, T., Higuchi, M., and Sprengel, R. (2001). Genetic manipulation of key determinants of ion flow in glutamate receptor channels in the mouse. Brain Res. 907, 233–243. CrossrefPubMedGoogle Scholar
Simonetti, M., Hagenston, A.M., Vardeh, D., Freitag, H.E., Mauceri, D., Lu, J., Satagopam, V.P., Schneider, R., Costigan, M., Bading, H., et al. (2013). Nuclear calcium signaling in spinal neurons drives a genomic program required for persistent inflammatory pain. Neuron 77, 43–57. Web of SciencePubMedCrossrefGoogle Scholar
Tan, A.M., Samad, O.A., Fischer, T.Z., Zhao, P., Persson, A.K., and Waxman, S.G. (2012). Maladaptive dendritic spine remodeling contributes to diabetic neuropathic pain. J. of Neurosci. 32, 6795–6807. Web of ScienceGoogle Scholar
Tappe, A., Klugmann, M., Luo, C., Hirlinger, D., Agarwal, N., Benrath, J., Ehrengruber, M.U., During, M.J., and Kuner, R. (2006). Synaptic scaffolding protein Homer1a protects against chronic inflammatory pain. Nat. Med. 12, 677–681. CrossrefGoogle Scholar
The author thanks Rose Lefaucheur for secretarial assistance and members of the author’s laboratory as well as colleagues in the Heidelberg Pain Consortium (SFB1158) for scholarly discussions. The author thanks Paul Naser, Dr. Pooja Gupta and Dr. Céline Heinl for help with the manuscript. The author’s work is supported by an ERC Advanced Investigator grant (PAIN PLASTICITY) and by the Collaborative Research Center 1158 on pain (SFB 1158) of the Deutsche Forschungsgemeinschaft.
German version available under https://doi.org/10.1515/nf-2017-0016
About the article
Rohini Kuner is a Full Professor of Pharmacology at Heidelberg University. She is the Spokesperson of the Heidelberg Pain Consortium funded by the German Research Foundation. She was trained in pharmacology and neuroscience at the University of Iowa City, USA, the Max Planck Institute for Medical Research and Heidelberg University.
Published Online: 2017-09-11
Published in Print: 2017-08-28