Heterosynaptic long-term potentiation of non-nociceptive synapses requires endocannabinoids, NMDARs, CamKII, and PKCζ

Abstract

Noxious stimuli or injury can trigger long-lasting sensitization to non-nociceptive stimuli (referred to as allodynia in mammals). Long-term potentiation (LTP) of nociceptive synapses has been shown to contribute to nociceptive sensitization (hyperalgesia) and there is even evidence of heterosynaptic spread of LTP contributing to this type of sensitization. This study will focus on how activation of nociceptors elicits heterosynaptic LTP (hetLTP) in non-nociceptive synapses. Previous studies in the medicinal leech (Hirudo verbana) have demonstrated that high-frequency stimulation (HFS) of nociceptors produces both homosynaptic LTP as well as hetLTP in non-nociceptive afferent synapses. This hetLTP involves endocannabinoid-mediated disinhibition of non-nociceptive synapses at the presynaptic level, but it is not clear if there are additional processes contributing to this synaptic potentiation. In this study, we found evidence for the involvement of postsynaptic level change and observed that postsynaptic N-methyl-d-aspartate (NMDA) receptors (NMDARs) were required for this potentiation. Next, Hirudo orthologs for known LTP signaling proteins, CamKII and PKCζ, were identified based on sequences from humans, mice, and the marine mollusk Aplysia. In electrophysiological experiments, inhibitors of CamKII (AIP) and PKCζ (ZIP) were found to interfere with hetLTP. Interestingly, CamKII was found to be necessary for both induction and maintenance of hetLTP, whereas PKCζ was only necessary for maintenance. These findings show that activation of nociceptors can elicit a potentiation of non-nociceptive synapses through a process that involves both endocannabinoid-mediated disinhibition and NMDAR-initiated signaling pathways.

NEW & NOTEWORTHY Pain-related sensitization involves increases in signaling by non-nociceptive sensory neurons. This can allow non-nociceptive afferents to have access to nociceptive circuitry. In this study, we examine a form of synaptic potentiation in which nociceptor activity elicits increases in non-nociceptive synapses. This process involves endocannabinoids, “gating” the activation of NMDA receptors, which in turn activate CamKII and PKCζ. This study provides an important link in how nociceptive stimuli can enhance non-nociceptive signaling related to pain.

INTRODUCTION

Noxious stimuli can have a persistent effect on nociceptive and non-nociceptive neural pathways (1). At the physiological level, these long-term effects are mediated by both neuromodulatory transmitters and activity-dependent changes in excitatory postsynaptic potentials (EPSPs). An example of the latter is long-term potentiation (LTP) in which noxious stimuli enhance signaling in nociceptive and non-nociceptive synapses via homosynaptic mechanisms (2, 3). However, there are also examples of LTP “spreading” from activated synapses to previously inactive connections, e.g., heterosynaptic LTP (hetLTP) in nociceptive synapses mediated by glial signaling (4). In this study, we will look at the potential spread of LTP from nociceptors to non-nociceptive synapses.

In Hirudo verbana, we have previously used high-frequency stimulation (HFS) to induce homosynaptic LTP in nociceptive (N) synapses that are N-methyl-d-aspartate receptor (NMDAR)-mediated (5), similar to LTP in the C fibers of rodent models (6). In addition, N cell HFS also produces heterosynaptic potentiation in non-nociceptive pressure (P) synapses (7). Heterosynaptic LTP (hetLTP) in the P synapses is due, in part, to an endocannabinoid-mediated decrease in tonic GABA input onto the P cell resulting in disinhibition that may also promote or gate LTP in these synapses (7–9) (Fig. 1). Prior studies have shown that endocannabinoids can promote LTP in the central nervous system via disinhibition (12–14). However, cannabinoid-mediated promotion of LTP has not been examined in the context of signaling by primary somatosensory afferents.

Figure 1.

Hirudo neural circuit used to study heterosynaptic long-term potentiation (hetLTP) of non-nociceptive synapses. High-frequency stimulation (HFS) of the nociceptive afferent (N cell) leads to endocannabinoid production. These endocannabinoids depress tonic GABAergic inhibition onto the presynaptic pressure sensory neuron (P cell), resulting in disinhibition of the P synapse. This disinhibition ultimately leads hetLTP of the P synapse (8–11). 2-AG, 2-arachidonoylglycerol; Postsyn, postsynaptic.

Hirudo possess the same endocannabinoid transmitters found in vertebrates, e.g., 2-arachidonoylglycerol (2-AG) and anandamide (15), and there appears to be considerable conservation in the mechanisms mediating other forms of endocannabinoid-mediated synaptic plasticity between Hirudo and vertebrates (16). As already mentioned, Hirudo also have the capacity for NMDAR-mediated LTP in nociceptive synapses, and homosynaptic NMDAR-LTP is also present in the non-nociceptive P synapses (17, 18). These, in addition to the advantages of the well-defined, stereotypic nervous system of H. verbana, lend themselves for further investigation of the features of LTP influenced/elicited by endocannabinoids. The fact that this is in the context of plasticity elicited by nociceptive activity and expressed in non-nociceptive synapses is a further advantage.

In this study, we examined how HFS of nociceptors in H. verbana leads to potentiation of the non-nociceptive pressure (P) cell synapses. We found evidence of NMDAR involvement in this hetLTP in addition to the presynaptic disinhibition mechanisms identified earlier. Furthermore, we observed a role for calmodulin-dependent kinase II (CamKII) and the zeta isoform of atypical protein kinase C (PKCζ) in the induction and maintenance of this hetLTP.

METHODS

General

H. verbana leeches of ∼3 g (Leech.com, Perris, CA or North America BioPharma, Erie, CO) were kept in artificial pond water (0.5 g/L Instant Ocean Sea Salt, Aquarium Systems) on a 12-h light/dark cycle in a 15°C incubator. Before dissection, animals were lightly anesthetized by chilling at 4°C for 30 min. Individual ganglia were dissected and pinned into 35-mm Sylgard-lined dishes for electrophysiology experiments under constant perfusion Hirudo saline solution (110 mM NaCl, 4 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, 5 mM NaOH, 10 mM glucose, and 10 mM HEPES; pH = 7.4) at a rate of 2 mL/min.

Dual Intracellular Recording Technique

Within a single isolated ganglion, nociceptive neurons (N cells, HFS), pressure-sensing neurons (P cells, the presynaptic neuron), and anterior pagoda neurons (AP cells, postsynaptic neuron) were identified under darkfield illumination by their characteristic position, size, and electrophysiological characteristics (19). Borosilicate glass electrodes with inner diameter (ID) of 0.5 mm and outer diameter (OD) of 1.0 mm were pulled on a Sutter Instrument (Novato, CA) Model P-97 Flaming Brown micropipette puller to have a consistent shape and form that elicited a 20–40 MΩ tip resistance when loaded with 3 M potassium acetate. The electrodes were placed in electrode holders that were connected to micromanipulators (Model 1480; Siskiyou Inc., Grants Pass, OR). Current clamp recordings of all neurons were made with a bridge amplifier (BA-1S; NPI, Tamm, Germany) and signals digitally converted for analysis using a DigiData 1322 A (Molecular Devices, Sunnyvale, CA). Current injection into the cells was accomplished using a digital stimulator (Multichannel Systems STG1004; Reutlingen, Germany). Twin P cell action potentials were elicited with a 300-ms interval. The first EPSP was used to measure changes in amplitude between the pre- and post-test. The second EPSP was used to measure the paired-pulse ratio (PPR = 2nd EPSP/1st EPSP), which provides an indication of whether LTP-related changes in synaptic transmission have a pre- or postsynaptic loci (20, 21). The 300-ms interval was chosen because this minimized the level of temporal summation between the first EPSP and second EPSP. This combined EPSP/PPR recording was repeated at 20-s intervals until 5–10 recordings were obtained and the EPSPs averaged to obtain the first EPSP amplitude and the PPR. To ensure that postsynaptic recordings were stable and that changes in EPSP amplitude were not due to changes in passive electric properties of the postsynaptic neuron soma (this includes damage to the cell during the recording process), the postsynaptic input resistance (IR) was monitored during the pre- and posttest recordings. This was accomplished by measuring the change in membrane potential during a 500 ms, 1 nA negative current pulse delivered at 20-s intervals (alternated with the EPSP/PPR tests). The postsynaptic neuron was hyperpolarized to approximately −70 mV during EPSP and input IR recordings to prevent postsynaptic action potentials. Electrodes were then removed from the P and AP cells to prevent damage due to osmotic stress and re-inserted an hour later for the posttest. After the pretest recordings, an intracellular electrode was used to deliver high-frequency stimulation (HFS) to a lateral N cell. The HFS consisted of 20 trains delivered at 10-s intertrain interval (ITI), with each train consisting of 10 action potentials at 25 Hz. The P and AP electrodes were reinserted 45 min later and posttest recordings of the EPSP, PPR, and IR were carried out. Measures of synaptic transmission were based on the peak amplitude from the average of 5–10 EPSP pairs during the pre- and posttest recordings. Changes in synaptic transmission in a given experiment were based on the percent change in the posttest EPSP amplitude relative to the pretest level [i.e., 100 × (EPSP_post/EPSP_pre)]. The percent change in PPR between the pre- and posttest records was calculated in the same way and changes in the PPR were used to assess whether observed synaptic potentiation or depression was due to pre- or postsynaptic changes (21). Percent change of the IR between the pre- and posttest recordings was also calculated and only recordings that exhibited less than a 15% change in input resistance were included for final analysis.

Pharmacological Treatments

Drugs were stored in frozen aliquots and diluted to the desired concentration with Hirudo normal saline immediately before use. (2R)-Amino-5-phosphonovaleric acid (AP5) (Tocris Biosciences No. 0106), tetrahydrolipstatin (THL or Orlistat) (Tocris Biosciences No. 3540), and MK801 (Tocris Biosciences No. 0924) stocks were dissolved in DMSO. Autocamtide-2-related inhibitory peptide (AIP) (Tocris Biosciences No. 5959) and zeta inhibitory peptide (ZIP) (Tocris Biosciences No. 2549) stocks were dissolved in Hirudo normal saline. AP5 (100 µM) and MK801 (250 μm) are N-methyl-d-aspartate (NMDA) receptor antagonists. THL (10 µM) inhibits the endocannabinoid-synthesizing enzyme diacylglycerol lipase. AIP (1 µM) inhibits calmodulin-dependent kinase II. ZIP (1 µM) inhibits atypical protein kinase C (PKC) ζ. Drugs applied during the induction phase of long-term potentiation (LTP) were bath-applied for 1 min before and during the 3-min HFS protocol. Drugs applied during the maintenance phase of LTP were either applied for 15 min immediately following HFS (early maintenance) or after 45 min post-HFS right before the posttest EPSP measurements (late maintenance). Bath application of the peptide drugs AIP and ZIP was performed at the whole ganglion level to block CaMKII and PKCζ, respectively. Both ZIP and AIP have been previously shown to be effective when applied via extracellular/bath application (22–25). Control experiments were carried out to show that AIP and ZIP alone (i.e., HFS omitted) do not affect EPSP amplitude. MK801 was intracellularly injected to determine whether synaptic plasticity required pre- and postsynaptic NMDARs. This was accomplished by 250 µM MK801 in the electrode filling solution during the pretest recordings which allowed the solution to enter the P or AP cells via diffusion from the electrode for 10–15 min before HFS delivery. The MK801 concentration used in this study was based on pilot experiments with a range of MK801 concentrations (data not shown) to find a concentration that affected LTP, but did not affect synaptic transmission when the HFS was omitted. These pilot studies also indicated that MK801 could be effectively applied (meaning that the drug interfered with hetLTP) in this manner without using current to ionophoretically drive the drug into the cell.

Sequence Identification in Hirudo verbena

Human amino acid sequences for CamKII and PKCζ were entered into the National Center for Biotechnology’s (NCBI) Basic Local Alignment Search Tool (BLAST), specifically the tBLASTn tool, to identify translated nucleotides. A transcriptome shotgun analysis against Hirudo was performed. Identified transcripts were translated to amino acids with Expasy’s Translate tool. The best match for each protein was then analyzed via InterProScan and Clustal Omega to ensure it was the correct protein type and to inspect homology of known protein isoforms with other organisms including Homo sapiens, Mus musculus, and Aplysia californica.

Statistics

Each replicate within an experimental group was performed using a ganglion from a different animal, so sample size refers to number of animals tested. All values are presented graphically and in the text as the means ± SE. All statistical analyses were done as one-way analysis of variance (ANOVA) with Tukey’s multiple comparison post hoc tests to confirm the ANOVA results. All analyses were done with GraphPad Prism 9. Asterisks indicate a statistically significant difference to post hoc analysis (Tukey’s multiple comparison) as follows: P < 0.05 (*), P < 0.005 (**), P < 0.001 (***), and P < 0.0001 (****).

RESULTS

Endocannabinoid Synthesis and NMDA Receptors Are Necessary for Heterosynaptic LTP of Non-Nociceptive Synapses

The first set of experiments was to confirm the previous findings that heterosynaptic LTP (hetLTP) in the P cell is endocannabinoid-dependent. In normal saline, N-cell HFS potentiated the P-AP synapse (Fig. 2, A and B). No change in paired-pulse ratio (PPR) was observed in these potentiated synapses (Fig. 2C), suggesting that potentiation occurred at the postsynaptic level and was not due to presynaptic increases in neurotransmitter release (20, 26). There were also no changes in postsynaptic input resistance indicating that synaptic potentiation was not due to increases in the passive electrical properties of the postsynaptic cell, at least as measured at the neuron soma (Fig. 2D).

Figure 2.

N-Methyl-d-aspartate (NMDA) receptors (NMDARs) and endocannabinoid synthesis mediate heterosynaptic potentiation of non-nociceptive pressure (P) synapses. A: sample P-to-anterior pagoda (AP) excitatory postsynaptic potential (EPSP) traces for before (black) and after (gray) nociceptive (N) cell high-frequency stimulation (HFS) (n = 9), HFS + (2R)-amino-5-phosphonovaleric acid (AP5) (n = 8), and AP5 alone (n = 9). A sample trace for HFS + tetrahydrolipstatin (THL) (n = 8) is not shown. Pairs of EPSPs were elicited at a 300-ms interval to measure the paired-pulse ratio (PPR) as well as EPSP amplitude. B: N cell HFS elicited significant potentiation of P synapses that was NMDAR- and 2-arachidonoylglycerol (2-AG) synthesis-dependent [one-way ANOVA F(3,30) =20.91, P < 0.0001]. HFS-induced potentiation that was blocked by AP5 (HFS vs. HFS + AP5, P < 0.0001; HFS vs. AP5 control, P < 0.001). HFS in the presence of AP5 elicited significant depression of P synapses (HFS + AP5 vs. AP5 Control, P < 0.05). No change in EPSP amplitude was observed when AP5 was applied alone. HFS in the presence of THL (inhibits 2-AG synthesis) also blocked potentiation in agreement with prior findings (10) (HFS + THL vs. HFS, P < 0.001). C: in terms of PPR, a significant treatment effect was observed [one-way ANOVA F(3,30) = 6.15, P < 0.01]. Post hoc analysis showed that there was a significant difference between the HFS vs. HFS + AP5 (P < 0.005), but no changes in PPR compared with the AP5 and HFS + THL groups. D: no changes in postsynaptic input resistance were observed in any of the four groups (one-way ANOVA F(3,30) = 1.71, P = 0.19). Asterisks indicate a statistically significant difference to post hoc analysis (Tukey’s multiple comparison) as follows: *P < 0.05, **P < 0.005, ***P < 0.001, and ****P < 0.0001.

To confirm the role of endocannabinoid production in this form of LTP, 2-arachidonyl glycerol (2-AG) synthesis was inhibited using tetrahydrolipstatin (THL, 10 µM). THL did prevent synaptic potentiation (Fig. 2B) in agreement with prior studies showing that this hetLTP is endocannabinoid-dependent (7, 10). No change in the PPR of the P-AP synapse, nor AP input resistance was observed in the HFS + THL group (Fig. 2, C and D).

Next, we investigated the potential role of NMDARs. We were motivated by the observation that N cell HFS produced no change in paired-pulse facilitation in the P synapses even though hetLTP was observed (Fig. 2C). This was surprising because our prior studies had indicated that 1) disinhibition increased P synaptic transmission at the presynaptic level, 2) that N cell HFS decreased tonic GABAergic input to the P cell in an endocannabinoid-dependent manner, and 3) that endocannabinoid-mediated potentiation of P synapses required presynaptic disinhibition (7, 8, 11). The lack of change in the PPR suggests that presynaptic disinhibition alone is not capable of eliciting hetLTP, but rather may initiate the potentiation process that ultimately requires changes at the postsynaptic level (21, 26). When 100 µM AP5, a competitive NMDAR inhibitor, was bath-applied during N cell HFS, hetLTP in the P synapse was blocked. Interestingly, not only was LTP blocked but also a statistically significant long-term depression (LTD) of this synapse was observed in the HFS + AP5 group when compared with the AP5-only controls where HFS was omitted (Fig. 2, A and B). This LTD was accompanied by a significant increase in PPR indicating changes at the presynaptic level. AP5 in the absence of HFS had no effect on the P synapse’s EPSP or PPR (Fig. 2C). No changes in input resistances were observed (Fig. 2D), indicating that potentiation or depression of EPSP amplitude was not due to changes in the passive membrane properties of the postsynaptic cell, at least at the soma level.

Next, experiments were carried out to determine whether the NMDARs involved in this heterosynaptic LTP were pre- or postsynaptically located. This was accomplished using MK801, a use-dependent, noncompetitive inhibitor of the NMDAR that binds inside the ion channel and prevents the inflow of calcium. MK801 (250 µM) was included in the electrode-filling solution loaded into the microelectrodes used for either the presynaptic P cell or postsynaptic AP cell recordings. This concentration of MK801 was chosen because it was found to not alter EPSP amplitude on its own (data from other concentrations not shown) and is actually lower than the concentration used in other studies (27). Presynaptic MK801 did not prevent hetLTP following N cell HFS (Fig. 3, A and B). In contrast, postsynaptic MK801 did block LTP indicating the presence of the relevant NMDARs. Unlike in the AP5 experiments, no LTD was uncovered in the HFS + MK801 groups regardless of whether MK801 was in the pre- or postsynaptic cell (Fig. 3B). The PPRs in these experiments identified no change among any groups (Fig. 3C). Input resistances for the postsynaptic cells also did not change in any of the treatment groups (Fig. 3D). To summarize, HFS of nociceptors elicited LTP in non-nociceptive P synapses that required endocannabinoid synthesis but also required activation of postsynaptic NMDARs.

Figure 3.

Postsynaptic and not presynaptic N-methyl-d-aspartate (NMDA) receptors are involved in heterosynaptic long-term potentiation (hetLTP) of non-nociceptive synapses. A: sample traces showing that hetLTP was still observed following presynaptic injection of MK801 (high-frequency stimulation, HFS + MK801_pre) but was blocked when MK801 was injected into the postsynaptic cell (HFS + MK801_post). B: one-way ANOVA showed a significant effect of treatment group on changes in excitatory postsynaptic potential (EPSP) amplitude [F(3,16) = 12.11, P < 0.001]. Pressure (P) synapses were potentiated in the HFS + MK801_pre group (n = 5) (HFS + MKpre vs. MK pre control, P < 0.005), whereas hetLTP was blocked in the HFS + MK801_post group (n = 6) (HFS + MKpost vs. MK post control, P = 0.56). No changes from baseline in EPSP amplitude were observed in either the MK_pre control (n = 4) and MK_post control (n = 5) groups where the HFS was omitted. C: no changes in paired-pulse ratio (PPR) were observed in any of the MK801 groups [F(3,16) = 0.45, P = 0.72]. D: no changes in input resistance were observed in any of the MK801 groups [F(3,16) = 1.134, P = 0.37]. Asterisks indicate a statistically significant difference to post hoc analysis (Tukey’s multiple comparison) as follows: **P < 0.005, ***P < 0.001.

Calmodulin Kinase II Is Required for Heterosynaptic LTP of Non-Nociceptive Synapses

Given the role of NMDARs, we next investigated the role of CamKII, a protein kinase that has been shown to contribute to induction and maintenance of LTP (28). We first determined that Hirudo has an ortholog of CamKII that may also be acted upon by autocamtide-2-related inhibitory peptide (AIP), an inhibitory pseudosubstrate specific for CamKII. Human CamKII isoforms were protein BLASTed against Hirudo transcriptomes with a transcriptome shotgun assembly to reveal a potential sequence. Sequence homology was completed for this proposed sequence of hirCamKII, which is believed to be most similar to the α or β isoform, with known Homo (68.13% amino acid similarity), Mus (68.20% similarity), and Aplysia (75.24% similarity) sequences (Fig. 4). The ATP binding site for CamKII had 8/9 identical amino acids with the remaining residue being strongly similar between Homo and Hirudo sequences. The pseudosubstrate domain had 8/10 identical amino acids 1/10 strongly similar, and 1/10 weakly similar amino acids. This pseudosubstrate domain is where CamKII autophosphorylates to sustain its function and is also the region the peptide inhibitor, AIP, mimics.

Figure 4.

Hirudo calmodulin-dependent kinase II (CamKII) sequence homology. Hirudo sequence was determined by BLASTing (tblastn) the human amino acid sequence from UniProt against Hirudo transcriptomes using transcriptome shotgun assembly. The proposed sequence in Hirudo (NCBI: GGIQ01080724.1) was aligned with human (UniPROT: Q9UQM7), mouse (UniPROT: P11798), and Aplysia (NCBI: NP_001191514.1) sequences. The bolded sequences indicate the ATP binding area. The bolded and italicized sequences indicate the pseudosubstrate domain. An asterisk below the sequence indicates a fully conserved residue. A colon indicates conservation between groups with a residue that has strongly similar properties. A period indicates conservation between groups with a residue that has weakly similar properties.

Compared with HFS in normal saline, 1 µM AIP applied during the N cell HFS (induction period) blocked hetLTP (Fig. 5, A and B). AIP application during the first 15 min after HFS (early maintenance phase) also disrupted hetLTP relative to the HFS group. Application of AIP 45 min after HFS (late maintenance phase) disrupted LTP relative to the HFS group, as well. This suggests that CamKII was required not only for induction of LTP but also the early and late maintenance phase (Fig. 5B). The PPR and input resistance for these groups did not exhibit any statistically significant changes (Fig. 5, C and D). Together this suggests that CamKII is required both for induction and maintenance of hetLTP.

Figure 5.

Calmodulin-dependent kinase II (CamKII) has a role in the induction and maintenance of heterosynaptic long-term potentiation (hetLTP) in non-nociceptive synapses. A: sample traces of pressure (P) synapses before (black) and after (gray) high-frequency stimulation (HFS) in a group where autocamtide-II related inhibitory peptide (AIP) was applied during nociceptive (N) cell HFS (AIP Induction) and where AIP was applied immediately after HFS (AIP Early Maintenance). B: one-way ANOVA detected a significant effect of treatment group on changes in excitatory postsynaptic potential (EPSP) amplitude [F(5,32) = 9.98, P < 0.0001]. N cell HFS elicited significant potentiation (HFS only, n = 6) compared with synapses where no HFS was delivered (Saline Only, n = 8) (P < 0.0001). HetLTP was blocked in groups where AIP was applied during HFS (AIP induction, n = 5; HFS vs. AIP Induction, P < 0.01), immediately after HFS (AIP early maintenance, n = 5; HFS vs. AIP Early Maintenance, P < 0.05) or 45 min after HFS (AIP late maintenance, n = 6; HFS vs. AIP Late Maintenance, P < 0.05). AIP alone (n = 8) had no effect on EPSP amplitude (Saline Only vs. AIP Only, P = 0.82). C: no changes in paired-pulse ratio (PPR) were observed in any of the AIP-related groups [F(5,32) = 0.66, P = 0.65]. D: no changes in input resistance were observed [F(5,32) = 0.74, P = 0.60]. E: timeline below illustrates order of pre- and post-HFS synaptic recordings, delivery of the HFS and timing of drug application during induction, early maintenance and late maintenance. Asterisks indicate a statistically significant difference to post hoc analysis (Tukey’s multiple comparison) as follows: *P < 0.05, **P < 0.005, and ****P < 0.0001.

Atypical Protein Kinase C Zeta Is Required for Heterosynaptic LTP of Non-Nociceptive Synapses

We next investigated the role of a potential invertebrate ortholog of PKMζ, PKCζ, in the induction and maintenance of hetLTP in Hirudo. Again, we identified a sequence of PKCζ in Hirudo and aligned it with other known sequences in human (61.16% amino acid similarity), mouse (61.55% similarity), and Aplysia (69.01% similarity) (Fig. 6). The sequence homology in the ATP binding site between humans and Hirudo had 9/9 identical amino acids and the pseudosubstrate domain had 11/11 identical amino acids providing confidence that zeta inhibitory peptide (ZIP), a PKCζ pseudosubstrate-mimetic inhibitor, would be effective. This high homology between the Homo and Hirudo proteins, in addition to the Aplysia protein that has been experimentally shown to function similarly to PKMζ (23, 29), gives us confidence to study it in the same manner.

Figure 6.

Hirudo protein kinase C zeta (PKCζ) sequence homology. Hirudo sequence was determined by BLASTing (tblastn) the human amino acid sequence from UniProt against Hirudo transcriptomes using transcriptome shotgun assembly. The proposed Hirudo sequence (NCBI: GGIQ01050911) was aligned with human (UniPROT: Q055130), mouse (NCBI: NP_032886.2), and Aplysia (NCBI: XP_012944127.1) sequences. The bolded sequences indicate the ATP binding area. The bolded and italicized sequences indicate the pseudosubstrate domain. An asterisk below the sequence indicates a fully conserved residue. A colon indicates conservation between groups with a residue that has strongly similar properties. A period indicates conservation between groups with a residue that has weakly similar properties.

ZIP (1 µM) application during the HFS induction phase did not prevent synaptic potentiation (Fig. 7, A and B). However, ZIP application during the first 15 min after HFS (early maintenance phase) did prevent potentiation measured during the posttest. ZIP was also effective at blocking hetLTP when applied 45 min after HFS, right before the posttest (late maintenance phase). PPR and input resistance did not change for any of the experimental groups (Fig. 7, C and D). Together these results indicate that Hirudo PKCζ is not required for induction of hetLTP but is required for maintenance.

Figure 7.

Protein kinase C zeta (PKCζ) has a role in the maintenance of heterosynaptic long-term potentiation (hetLTP) but not induction in non-nociceptive synapses. A: sample traces for showing pressure (P) synapse excitatory postsynaptic potentials (EPSPs) recorded before (black) and after (gray) nociceptive (N) cell high-frequency stimulation (HFS) where zeta inhibitory peptide (ZIP) was applied either during induction or the early or late maintenance periods. B: one-way ANOVA detected a significant effect of treatment group on changes in EPSP amplitude [F(5,29) =15.98, P < 0.0001]. N cell HFS (HFS only, n = 6) elicited hetLTP in P synapses that was not observed when HFS was omitted (Saline Only, n = 8; HFS Only vs. Saline Only, P < 0.0001). ZIP applied during HFS (HFS + ZIP induction, n = 5) did not block hetLTP (HFS Only vs. ZIP Induction, P = 0.9144). However, no hetLTP was observed in synapses in which ZIP applied immediately after HFS (HFS + ZIP early maintenance, n = 5; HFS Only vs. ZIP Early Maintenance, P < 0.0001) or 45 min after HFS (HFS + ZIP late maintenance, n = 5; HFS Only vs. ZIP Late Maintenance, P < 0.0001). ZIP alone had no effect on EPSP amplitude (ZIP Only vs. Saline Only, P = 0.65). C: no effect on paired-pulse ratio (PPR) was observed across all treatment groups [F(5,29) = 1.37, P = 0.26). D: no change in input resistance (IR) was observed across all treatment groups [F(5,29) = 1.29, P = 0.30]. E: timeline below illustrates order of pre- and post-HFS synaptic recordings, delivery of the HFS and timing of drug application during induction, early maintenance, and late maintenance. Asterisks indicate a statistically significant difference to post hoc analysis (Tukey’s multiple comparison) as follows: ***P < 0.001 and ****P < 0.0001.

DISCUSSION

This hetLTP in non-nociceptive synapses is of interest due to potential associations with chronic pain. In our model, HFS of the N cell mimics a noxious stimulus that potentiates in not only nociceptive synapses (homosynaptic LTP) (5) but also non-nociceptive synapses as well (hetLTP). Previous studies have shown hetLTP between nociceptive pathways (1, 4, 30, 31), but not between nociceptive and non-nociceptive pathways. In Hirudo, we believe that nociceptive cell HFS drives endocannabinoid production in the local area of the postsynaptic cell (Fig. 1). This endocannabinoid production then leads to depression of tonic GABAergic input to the presynaptic P cell (7, 8). This disinhibition, in turn, appears to trigger LTP of the P synapse in an NMDAR-dependent manner that also required CamKII and PKCζ activity (Fig. 8). Prior studies have shown that endocannabinoids can enhance LTP via a disinhibition mechanism (12–14). What is unusual here is that endocannabinoid-mediated disinhibition appears to “gate” NMDAR-dependent LTP in the P synapse, i.e., potentiation appeared to occur as a result of background activity in the synapse. A similar observation has also been reported in the hippocampus where cannabinoid receptor activation produced widespread LTP without any external stimulation (32).

Figure 8.

Proposed cellular model of endocannabinoid-mediated heterosynaptic long-term potentiation (hetLTP). 2-Arachidonoylglycerol (2-AG) is produced as a result of nociceptive (N) cell high-frequency stimulation (HFS) (see Fig. 1) and depresses tonic inhibition to the pressure (P) cell. The resulting disinhibition of the P cell synapses leads to increased activation of postsynaptic N-methyl-d-aspartate receptors (NMDARs). NMDAR activity presumably permits a Ca²⁺ influx that in turn promotes activation of calmodulin-dependent kinase II (CamKII), and protein kinase C zeta (PKMζ) that mediates postsynaptic changes responsible for LTP.

First, we wanted to understand if the hetLTP was due to endocannabinoid-mediated disinhibition alone or if other mechanisms were involved. From our earlier studies, we found that GABA could presynaptically inhibit P synapses, that disrupting GABAergic inhibition blocked endocannabinoid-mediated LTP of P synapses, and that HFS of N cells produced a depression on tonic inhibition of the P cell (7, 8, 11). Using measurements of pair-pulse facilitation before and after HFS, we expected to observe decreases in the PPR which would be consistent with hetLTP having a presynaptic locus (21, 26). However, we found that there was no change in PPR during hetLTP. This suggests that endocannabinoid-mediated disinhibition of the presynaptic P cell on its own cannot explain hetLTP and that expression of potentiation involved an additional postsynaptic process. To determine what mediated these postsynaptic changes, we focused on the potential involvement of NMDARs for this heterosynaptic form of LTP since prior studies have shown the presence of NMDARs in mediating homosynaptic LTP in P synapses (17, 33). Bath-applied AP5 did block HFS-induced hetLTP and injected MK801 provided evidence that the NMDARs are located in the postsynaptic cell. Interestingly, the EPSP depression was observed in the HFS + AP5 group when compared with the AP5 Control group (where the HFS was omitted) suggesting that blocking NMDARs during HFS uncovers a form of heterosynaptic LTD in the P synapses. This depression was coupled with an increase in the PPR indicating a presynaptic mechanism since decreases in neurotransmitter release are associated with increases in PPR (19, 20). Similar results were observed with AP5 disruption of homosynaptic LTP in Hirudo N cell synapses (5). This depression in the P synapses was not observed when MK801 was injected into the pre- or postsynaptic cell, and it is unclear why this is the case. One possibility is that there is a contribution by NMDARs on other neurons that are exposed to bath-applied AP5, but not to intracellular-delivered MK801, e.g., the N cell or the unknown GABAergic neurons involved in disinhibition of the P cells (Fig. 1).

Next, we investigated the role of CamKII in this heterosynaptic potentiation. CamKII is well known for its role in LTP by contributing to AMPA receptor exocytosis (34) and increasing AMPA receptor conductance (35). In addition, CamKII’s ability to autophosphorylate and thus prolong its activity indicates a role in LTP maintenance and potentially in memory formation (25). Human CamKII isoforms were searched against Hirudo transcriptomes to identify a potential ortholog. A candidate Hirudo CamKII was found with well-conserved homology in the ATP binding site as well as in the pseudosubstrate domain. The CamKII inhibitor AIP was able to prevent hetLTP during the induction phase (N cell HFS) and when applied during the early and late maintenance phase.

In addition to CamKII, we looked at another kinase proposed to have a role in LTP maintenance, PKCζ. The truncated version of this protein, PKMζ, has garnered significant interest in terms of synaptic plasticity, memory formation, and chronic pain (36, 37). Similar to CamKII, it is believed to exhibit its function through AMPA receptor trafficking (38) and increased AMPA receptor activity (36). The time course for the function of these two kinases are different however with CamKII believed to be involved with induction and maintenance and PKMζ being only, or at least primarily, involved in maintenance. Although CamKII can have long-acting effects through its autophosphorylation, PKMζ has the potential for even longer-acting effects due to lacking a regulatory domain and will therefore be active until degradation.

In Hirudo, we identified a homolog by comparing the Homo sequence to the Hirudo transcriptome. This revealed a protein in Hirudo that appeared to be a high similarity homolog with PKMζ/PKCζ in Homo, Mus, and Aplysia in that it had identical amino acid sequences at the critical ATP binding and pseudosubstrate sites lending credence to being the homolog of the mammalian version. The established terminology for these two proteins can be complicated. In mammals, they both originate from the PRKCZ gene with PKCζ being produced from the full-length transcript, whereas PKMζ is produced from a truncated transcript due to an internal promoter and therefore does not contain the regulatory domain (39). In our electrophysiology experiments, we found that hetLTP was not affected by ZIP application during the induction phase, however, no hetLTP was observed following ZIP treatment during the early and late maintenance phases.

There are two major criticisms with the use of ZIP to assess PKMζ and PKCζ function. The first is that in PKMζ null mice, ZIP still has an effect (40). This may be explained by the fact that mammalian organisms have a duplication of the gene and that the very similar PKCι/λ are able to compensate for PKMζ (29). This compensatory mechanism is only seen in genetic knockouts and therefore should not affect the experiments performed in this study (41). The second is that ZIP has been shown to have poor specificity by acting on other kinases (36). To address this issue, 1 µM ZIP was used in these experiments, which is below the level reported to have significant off-target effects.

Previous experiments in mammals have investigated inhibition of GABAergic signaling in the spinal cord leading to increased signaling in Aδ/C and Aβ fibers synapses (42). Importantly, this disinhibition “opens the gate” for input of non-nociceptive signaling onto nociceptive circuitry in the spinal cord. Other mammalian studies have identified inhibitory interneurons as gatekeepers of mechanical allodynia with their ablation leading to allodynia and their stimulation causing neuropathic pain alleviation (43). A recent in silico model extensively detailed the neuronal connectome of the spinal dorsal horn and was able to replicate the effects of disinhibition leading to mechanical allodynia (44). Interestingly, evidence in mammals indicates that disinhibition is transient and promotes subsequent changes responsible for the maintenance of persistent pain (45). Our studies in Hirudo support this idea with endocannabinoid-mediated disinhibition of P synapses ultimately triggering LTP of these synapses via NMDAR-mediated mechanisms.

A largely unaddressed question is how does prior injury trigger disinhibition within these spinal circuits. Endocannabinoids represent a potentially important mechanism based on their well-established capacity to depress inhibitory synapses (46). Using semi-intact preparations in Hirudo, we have found that direct activation of nociceptors or noxious stimuli applied to the skin elicits hetLTP in P synapses and behavioral sensitization to non-nociceptive stimuli, both of which are endocannabinoid-dependent (10). Evidence of this endocannabinoid-mediated sensitization was also observed in vivo 1) where response thresholds to non-nociceptive stimuli were reduced following 2-AG injections (47) and 2) where injury-induced sensitization to non-nociceptive stimuli was reduced by inhibiting 2-AG synthesis (48). These studies mirror preclinical findings in rodents demonstrating a pro-nociceptive effect of cannabinoids via synaptic disinhibition (49, 50).

These findings of pro-nociceptive effects by cannabinoids in both vertebrates and invertebrates are important given attempts to utilize the cannabinoid system as an analgesic. Definitive clinical evidence of cannabinoids to treat pain is still lacking (51, 52) and the International Association for the Study of Pain does not recommend cannabinoid-based treatments for pain at this time (53). A likely reason for this lack of clarity is that the cannabinoid system has both pro- and antinociceptive effects based on both preclinical and clinical studies (10, 49, 54–61). The key then is understanding how the anti- versus pro-nociceptive effects of endocannabinoids are selectively deployed. Studies from our laboratory suggest that different patterns of afferent activity elicit different endocannabinoid effects (7, 16, 62). The key next steps, as the current study demonstrates, are to determine the cellular mechanisms that translate different patterns of afferent signaling into either pro- or anti-nociceptive endocannabinoid effects.

DATA AVAILABILITY

Data will be made available upon reasonable request.

GRANTS

Funding provided by National Institute of Neurological Disorders and Stroke (NINDS) under Grant No. 1R01NS092716-01A1 (to B. D. Burrell), National Institute of General Medical Science (NIGMS) under Grant No. T32GM136503 (to B. D. Burrell), and National Science Foundation (NSF)-DGE Grant No. DGE-1545679 (to R. T. Paulsen).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

A.D.F. and B.D.B. conceived and designed research; A.D.F., R.T.P., E.J.K., and B.D.B. performed experiments; A.D.F., R.T.P., E.J.K., and B.D.B. analyzed data; A.D.F. and B.D.B. interpreted results of experiments; A.D.F. prepared figures; A.D.F. drafted manuscript; B.D.B. edited and revised manuscript; B.D.B. approved final version of manuscript.