The anterior cingulate cortex and central nucleus of the amygdala connect widely with brainstem nuclei involved in descending modulation, including the rostral ventromedial medulla. Endogenous opioids in these circuits participate in pain modulation. The hypothesis was that a differential opioidergic role for the brain nuclei listed in regulation of spinal neuronal responses because separable effects on pain behaviors in awake animals were previously observed.
This study utilized in vivo electrophysiology to determine the effects of morphine microinjection into the anterior cingulate cortex, right or left central nucleus of the amygdala, or the rostral ventromedial medulla on spinal wide dynamic range neuronal responses in isoflurane-anesthetized, male Sprague–Dawley rats. Ongoing activity in the ventrobasal thalamus was also measured. In total, 33 spinal nerve ligated and 26 control age- and weight-matched control rats were used.
Brainstem morphine reduced neuronal firing to 60-g von Frey stimulation in control rats (to 65 ± 12% of control response (means ± 95% CI), P < 0.001) with a greater inhibition in neuropathic rats (to 53 ± 17% of control response, P < 0.001). Contrasting anterior cingulate cortex morphine had only marginal modulatory effects on spinal neuronal responses with limited variance in effect between control and neuropathic rats. The inhibitory effects of morphine in the central nucleus of the amygdala were dependent on pain state and laterality; only right-side morphine reduced neuronal firing to 60-g stimulation in neuropathic rats (to 65 ± 14% of control response, P = 0.001). In addition, in neuropathic rats elevated ongoing neuronal activity in the ventral posterolateral thalamus was not inhibited by anterior cingulate cortex morphine, in contrast to evoked responses.
Cumulatively the data support opioid modulation of evoked responses predominately through a lateralized output from the right amygdala, as well as from the brainstem that is enhanced in injured conditions. Minimal modulation of dorsal horn responses was observed after anterior cingulate cortex opioid administration regardless of injury state.
Descending control from supraspinal neuronal networks onto spinal cord neurons can modulate nociception
Endogenous opioids in these brain circuits participate in pain modulation
A differential opioidergic role for brain nuclei involved in supraspinal pain modulation has not been previously reported
In vivo electrophysiologic recordings from the dorsal horn of the spinal cord in male rats reveal differential effects of morphine at the anterior cingulate cortex, right amygdala, and the ventromedial medulla on evoked pain responses
These data differentiate supraspinal opioid circuit regulation of spinal nociceptive processing and suggest that the regulation of sensory and affective components of pain are likely separate
Cumulative complex peripheral and central processes combine to drive nociceptive transmission, and modulation occurs at multiple levels in the central nervous system. After transduction of a potentially noxious message at the periphery, nociceptive afferents project to the brain via the spinal cord. Brain circuits can subsequently modulate spinal neuronal activity via descending control systems that modify pain perception.1–3 The pain detection system is highly plastic; injury-induced synaptic and molecular changes may occur at peripheral and/or central loci, promoting chronic pain.4 Interactions between sensory and affective brain regions and the impact of morphine on these areas are unclear, as is the relationship between the pain experience and activity in ascending/descending circuits.
The anterior cingulate cortex has a key role in pain processing and pain-related emotion5,6 and widely connects with regions of the descending control system including the rostral ventromedial medulla.7 This brainstem modulatory circuit exerts excitatory and inhibitory influences over spinal nociceptive processing.8 Plasticity within the amygdala contributes to affective dimensions of pain, and concerning nociceptive processing, the right central nucleus of the amygdala demonstrates increased neuronal activation in varied pain states including neuropathy.9
Systemic morphine reduces nociceptive responses in rodents,10 and a high density of opioid receptors are found in the three brain nuclei discussed.11–13 Brainstem neurons have clear sensitivities to µ opioid receptor agonists; rostral ventromedial medulla microinjection of opiates is known to be analgesic.14 Despite this, neither direct nor indirect activation of µ opioid receptor-positive neurons in this brain nucleus is required for analgesia.15 Opioidergic manipulation in the right central nucleus of the amygdala inhibits stress-induced pain,16,17 and endogenous opioid signaling in the anterior cingulate cortex is proposed to be necessary for the relief of pain aversiveness.18 Formulation of optimal therapies for chronic pain patients requires a more complete understanding of the potentially different roles played by spinal and supraspinal circuits that govern the pain experience.
The impact of morphine on central processing of ongoing aversive states and evoked responses is not fully understood. We hypothesized that different supraspinal circuits are likely involved in the regulation of evoked hypersensitivities, and we investigated this by pharmacologically manipulating brain-amplified spinal pain processes. Behavioral studies reveal differential anterior cingulate cortex and rostral ventromedial medulla control of the ongoing aversive state and evoked hypersensitivities,19 but the brain networks described may have differential effects on suprathreshold responses that relate to the high levels of pain reported by patients. We utilize in vivo electrophysiology techniques to assess the effect of rostral ventromedial medulla, anterior cingulate cortex, or central nucleus of the amygdala (right or left) morphine microinjection on spinal wide dynamic range neuronal responses. We examine the injury-induced plasticity observed at the spinal level by repeating the experiments in spinal nerve–ligated rats. Finally, because the ventrobasal thalamus is a key relay in sensory pathways from spinal cord to the brain, we examined the possible association of effects of anterior cingulate cortex morphine on ascending and descending inhibition.
Materials and Methods
Approval from the institutional animal care committee (University College London, London, United Kingdom) was obtained before all experimental procedures. Male Sprague–Dawley rats weighing 120 to 140 g (for spinal nerve ligation surgery) or 250 to 300 g (weight for age- and weight-matched control rats) were obtained from Biological Services, University College London (London, United Kingdom). The animals were group-housed on a 12-h:12-h light–dark cycle. Food and water were available ad libitum. All procedures described received approval from the United Kingdom Home Office and Biological Services of University College London. The animals were monitored throughout the duration of the study to reduce unnecessary stress and/or pain, and the number of animals was used in accordance with the International Association for the Study of Pain ethical guidelines. All data are reported according to Animal Research: Reporting of In Vivo Experiments guidelines.
Spinal Nerve Ligation Surgery
Spinal nerve ligation surgery was performed as described previously.20 The rats (120 to 140 g at the time of surgery) were maintained under 2% (v/v) isoflurane anesthesia delivered in a 3:2 ratio of nitrous oxide and oxygen. Under aseptic conditions, a paraspinal incision was made, and the left tail muscle was excised. Part of the L5 transverse process was removed to expose the L5 and L6 spinal nerves, which were then isolated with a glass nerve hook (Ski-Ry Ltd., United Kingdom) and ligated with a nonabsorbable 6-0 braided silk thread proximal to the formation of the sciatic nerve. The surrounding skin and muscle were closed with absorbable 3-0 sutures. Sham surgery was performed in an identical manner omitting the ligation step. All rats were monitored for normal behaviors (grooming and mobility) and for general health and weight gain postsurgery.
Dorsal Horn Recordings.
In vivo electrophysiology experiments were conducted in rats weighing 250 to 300 g on postoperative days 14 to 18 (sham and spinal nerve ligated-operated animals) or weight- and age-matched naive rats as previously described.20 Briefly, the animals were anesthetized and maintained for the duration of the experiment with isoflurane (1.5%) delivered in a gaseous mix of N2O (66%) and O2 (33%). The animals were secured to a stereotaxic frame. A laminectomy was performed to expose the L4 and L5 segments of the spinal cord. Extracellular recordings were made from deep dorsal horn neurons (lamina V and VI) using two MΩ 127-µm-diameter parylene-coated tungsten electrodes (A-M Systems, USA). All the neurons recorded were wide dynamic range and responded to natural stimuli including brush, low- and high-intensity mechanical and thermal stimuli in a graded manner with coding of increasing intensity.
A train of 16 transcutaneous electrical stimuli, delivered via stimulating needles inserted into the peripheral receptive field, was applied at three times the threshold current for C-fiber activation of the dorsal horn wide dynamic range cell. A poststimulus histogram was constructed, and responses evoked by Aβ-fibers (0 to 20 ms), Aδ-fibers (20 to 90 ms), and C-fibers (90 to 350 ms) were separated and quantified on the basis of latency. Neuronal responses occurring after the C-fiber latency band were quantified as postdischarge (350 to 800 ms). Activity-dependent hyperexcitability was measured as “wind-up,” calculated as the difference between the total number of action potentials at C-fiber latency produced by the train of 16 electrical stimuli, and ““input,” which represented the postsynaptic C-fiber–evoked dorsal horn neuronal response after the first of the 16 electrical stimuli in the electrical train.
The peripheral receptive field was stimulated using mechanical (brush and von Frey filaments 2, 8, 26, and 60 g) and thermal stimuli (42, 45, and 48°C applied with a constant water jet). All natural stimuli were applied for 10 s each. The data were captured and analyzed by a Cambridge Electronic Design 1401 interface coupled to a Pentium computer with Spike 2 software (rate functions).
Three baseline responses to peripheral stimuli as detailed above were characterized for each neuron before pharmacologic assessment (a drug study was carried out on one neuron per animal only) after three consecutive stable control trials (10% variation for C-fiber evoked, less than 20% variation for all other parameters). Neuron values were averaged to give the predrug control values.
Ventral Posterolateral Thalamus Recordings.
In vivo electrophysiologic recordings in the ventral posterolateral thalamus were performed as previously described.21 As with spinal recordings, the animals were maintained on 1.5% (v/v) isoflurane anesthesia for the entire experiment. The rats were secured in a stereotaxic frame, a midline incision was made across the scalp, and after the skull was exposed, coordinates for the right ventral posterolateral thalamus (contralateral to injury) were calculated in relation to bregma (2.28 mm caudal, 3.2 mm lateral).22 A small craniotomy was performed with a high-speed surgical microdrill. Extracellular recordings were made from ventral posterolateral thalamic neurones (between −2.2 and −2.5 mm caudal from bregma).
The data were captured and analyzed by a Cambridge Electronic Design 1401 interface coupled to a computer with Spike2 software (United Kingdom) with rate functions. Spike sorting was performed post hoc with Spike2 using fast Fourier transform followed by principal component analysis of waveform feature measurements for multiunit discrimination. Stimulus-evoked neuronal responses were determined by subtracting total spontaneous neuronal activity in the 10-s period immediately preceding stimulation. Spontaneous firing of individual neurones (number of spikes/s) is expressed as the mean of these 10-s periods.
Neurons were recorded from one site per rat; one to three neurons were characterized at each site. A total of nine neurons were recorded from seven spinal nerve–ligated rats, and seven neurons were recorded from five control rats.
Intrabrain Injections for In Vivo Electrophysiology
The animals were anaesthetized and secured into a stereotaxic frame. After collection of baseline control neuronal data, morphine (10 μg/0.5 μl, dissolved in saline) was microinjected into specific brain regions using a Hamilton syringe (Thames Restek United Kingdom) as detailed: rostral ventromedial medulla −0.5 mediolateral, −11 caudal from bregma, 10 mm dorsal from bregma; anterior cingulate cortex +2.6 rostral and −0.6 lateral from bregma and 1.8 mm dorsal from bregma (anterior cingulate cortex site of injection comparable to that used previously);19 right central nucleus of the amygdala + 1.6 to 3.2 rostral, −4 lateral from bregma, 8 dorsal from bregma; left central nucleus of the amygdala + 1.6 to 3.2 rostral, 4 lateral from bregma, 8 dorsal from bregma. The drug effect was followed for up to 60 min with tests carried out at 10, 30, and 60 min. For the postdrug effects, maximal changes from predrug baseline values are plotted. A dose of 10 µg/0.5 µl morphine was chosen to ensure parity with previous studies where effects were observed using this dose.19 After the conclusion of the experiment, the brains were dissected and sliced, and the site of injection was verified for anterior cingulate cortex, central nucleus of the amygdala, and rostral ventromedial medulla microinjections.
Experimental Design and Statistical Analysis
The animal group (control or spinal nerve–ligated rats) is the between-subject variable. Each animal in each group received morphine or saline microinjection into the anterior cingulate cortex, right or left central nucleus of the amygdala, or rostral ventromedial medulla, and these are within-subject variables. The critical variables for independent replication are as follows: male Sprague–Dawley rats, 26 control rats in total, and 33 neuropathic rats in total. The rats were randomly assigned into cages by an independent experimenter and then were designated to the naïve, sham, or neuropathic groups. The only endpoint used in this study is an objective measure free from subjective bias (i.e., total evoked spikes to a given stimulus). Blinding to the experimental condition was not feasible because of the presence of a surgical scar, and spinal nerve ligation surgery is frequently associated with some muscle atrophy and foot guarding. The experimenter was not blinded to microinjection substance (saline vs. morphine) because saline was always administered first, such that morphine drug effects did not impact the vehicle control effect (in keeping with the National Center for the Replacement, Refinement and Reduction of Animals in research, we used the same rat for saline and subsequent morphine microinjection analyses). Monitoring neuronal activity provides a completely objective measure. For each experimental group the investigation was carried out in the following manner: all electrophysiology procedures commenced approximately 2 h into the light cycle and lasted around 3 to 4 h in total; the animals were anesthetized and maintained for the duration of the experiment with isoflurane (1.5%) delivered in a gaseous mix of N2O (66%) and O2 (33%); upon isolating a single wide dynamic range spinal neuron, morphine (10 µg/0.5 µl) was microinjected into the anterior cingulate cortex, the central nucleus of the amygdala, or the rostral ventromedial medulla; after data collection the rat was overdosed on isoflurane (5%, delivered in a gaseous mix of N2O (66%) and O2 (33%)), and upon cessation of heartbeat, cervical dislocation was performed. For every spinal nerve ligation procedure, the surgeries were carried out at 9:30 am. All in vivo electrophysiology experimental animals were group-housed (sham-operated and spinal nerve–ligated rats shared a cage, whereas naïve rats shared a cage). Rats were chosen at random from their respective cages the morning of each experiment. All were within a weight range of 250 to 300 g. All recovery surgical procedures (nerve ligation) were performed in designated theatres within the Biological Services Unit of University College London. All nonrecovery procedures (electrophysiology) were performed in designated laboratories within the Medical Sciences Department of University College London. There were no animals missing from analysis; all rats survived the surgeries and experimental protocol. We did not exclude any outliers from the analysis. At the time of spinal nerve ligation surgery, male rats weighing 120 to 140 g were approximately 10 to 12 weeks old. At the time of in vivo electrophysiology experimentation, male rats weighing 250 to 300 g were approximately 12 to 14 weeks old. Minimum group sizes were determined by a priori calculations using the following assumptions (α 0.05, 1-β 0.8, ε 1, effect size range d = 0.5 to 0.8). Effect sizes of inhibitory activity of known analgesics on mechanical and heat-evoked responses were determined from historical data sets; the typical means and variation of neuronal sample sizes ranging from 5 to 10 were comparable with this study. Each experimental group contained between 5 and 9 animals (and therefore 5 to 9 single unit recorded cells) to ensure statistical robustness while adhering to the “3 Rs” (refine, reduce, replace; https://www.nc3rs.org.uk/the-3rs; accessed January 8, 2020). Statistical analyses were performed using SPSSv25 (IBM, USA) to determine drug effects on neuronal excitability. The primary outcome measure was a change in the neuronal response; no prior assumptions about the directionality were made. Statistical differences in electrical parameters and neuronal responses to dynamic brush stimulation were determined using two-tailed paired Student’s t test. Drug effects on mechanical and thermal coding were determined using a two-way repeated-measure ANOVA followed by Bonferroni correction (which is free of dependence and distributional assumptions) to control for multiple paired comparisons; the group effect is reported throughout. Where appropriate, sphericity was tested using Mauchly’s test; the Greenhouse–Geisser correction was applied if violated. The data represent means ± 95% CI. Asterisks and plus signs denote statistically significant differences (+/*P < 0.05, ++/**P < 0.01, ***P < 0.001).
There Is a Statistically Significant Inhibition of Spinal Wide Dynamic Range Neuronal Responses to a Range of Modalities after Microinjection of Morphine into the Rostral Ventromedial Medulla
Descending pathways from brainstem structures influence nociceptive signaling in the spinal cord dorsal horn.23 To ensure that a potential modulatory effect of a supraspinal brain region on spinal neuronal activity could be revealed, we microinjected morphine (10 µg/0.5 µl) directly into the rostral ventromedial medulla and compared spinal neuronal responses before and after injection (data depicted as means ± 95% CI). After rostral ventromedial medulla morphine, there was a statistically significant reduction in the number of action potentials fired by spinal wide dynamic range neurons upon stimulation of the peripheral receptive field, and no difference in the level of neuronal inhibition was observed between naïve (n = 4) and sham-operated (n = 3) animals; thus these groups are pooled as the “control” group throughout (n = 7 rats and 7 single unit cell recordings). In the control animal group, microinjection of morphine into the rostral ventromedial medulla statistically significantly inhibited spinal neuronal responses to a range of noxious natural stimuli (mechanical responses: two-way repeated-measure ANOVA, group main effects P = 0.001, F1, 6 = 29.02, dose × von Frey interaction, P < 0.001, F3, 15 = 17.66; thermal responses: two-way repeated-measure ANOVA group main effects P = 0.018, F1, 6 = 11.88, dose × temperature interaction P = 0.319, F2, 10 = 1.29; fig. 1, A and B, respectively). Input was also statistically significantly reduced (paired Student’s t test; P = 0.0012; fig. 1C). Morphine injection data were compared with the baseline after vehicle injection (microinjection of saline into the rostral ventromedial medulla of control or neuropathic animals did not affect spinal neuronal responses to natural or electrical stimulation as depicted in a subsequent figure).
There Is an Enhanced Inhibition of Spinal Neuronal Responses after Microinjection of Morphine into the Rostral Ventromedial Medulla of Neuropathic Rats
There is a key role for the rostral ventromedial medulla in the pathogenesis and maintenance of neuropathic pain.24 To observe whether injury-induced plasticity impacts the inhibitory effect of rostral ventromedial medulla morphine at the spinal level, we repeated the experiments outlined above in spinal nerve–ligated rats (n = 6 rats and 6 single unit cell recordings). Relative to the inhibition observed in control animals, there was an enhanced inhibition produced after morphine microinjection into the rostral ventromedial medulla of neuropathic animals across natural modalities. Spinal neuronal responses to innocuous mechanical stimulation were now also statistically significantly reduced (mechanical responses: two-way repeated-measure ANOVA group main effects P < 0.001, F1, 5 = 75.26, dose × von Frey interaction P = 0.037, F3, 15 = 3.64; for thermal responses: two-way repeated-measure ANOVA group main effects P = 0.027, F1, 5 = 9.63, dose × temperature interaction P = 0.070, F2, 10 = 3.50; fig. 1, D and E, respectively). In neuropathic animals, input was statistically significantly inhibited once again, as well as Aδ, C-fiber, and postdischarge values (paired Student’s t test, P = 0.022, 0.015, 0.049, and 0.007, respectively (fig. 1F).
Spinal Wide Dynamic Range Neuronal Responses Are Marginally Reduced after Microinjection of Morphine into the Anterior Cingulate Cortex to a Comparable Level in Control and Spinal Nerve Ligated Animals
Sensory qualities of pain are likely to be differentially influenced by brain opioid receptor circuits, and endogenous opioid activity in the anterior cingulate cortex is implicated in the relief of pain aversiveness.18 To determine whether the supraspinal control of spinal neuronal responses is differentially regulated according to the opioid brain circuit in question, we microinjected morphine (10 µg/0.5 µl) into the anterior cingulate cortex of control, uninjured animals (n = 7 rats and 7 single unit cell recordings). We now observe attenuated spinal neuronal responses to a subset of mechanical (two-way repeated-measure ANOVA group main effects P = 0.004, F1, 6 = 20.04, dose × von Frey interaction P = 0.003, F3, 18 = 6.64) and thermal (two-way repeated-measure ANOVA group main effects P = 0.013, F1, 6 = 12.01, dose × temperature interaction, P = 0.602, F2, 12 = 0.53) stimuli (fig. 2, A and B). In neuropathic animals (n = 9 and 9 single unit cell recordings), a statistically significant yet small inhibition of evoked neuronal responses is also observed (mechanical responses: two-way repeated-measure ANOVA group main effect P = 0.016, F1, 8 = 9.03, dose × von Frey interaction P = 0.107, F3, 24 = 2.27; for thermal responses: two-way repeated-measure ANOVA group main effects P = 0.016, F1, 8 = 9.14, dose × temperature interaction P = 0.109, F2, 16 = 2.56), as well as for input and brush-evoked responses (paired Student’s t test, P = 0.045 and 0.042, respectively; fig. 2, D, E, and F). Morphine injection data were compared with the baseline after vehicle injection (microinjection of saline into the anterior cingulate cortex of control or neuropathic animals did not affect spinal neuronal responses to natural or electrical stimulation as depicted in the subsequent figure).
A direct and simple comparison of the inhibitory effect of rostral ventromedial medulla or anterior cingulate cortex morphine microinjection in spinal nerve ligated compared with control group animals is achieved by plotting mechanical responses for all animal groups as a percentage of their baseline responses. For example, morphine microinjection into the rostral ventromedial medulla of control animals statistically significantly reduces evoked spinal neuronal responses to 60-g von Frey filaments to 65 ± 12% (means ± 95% CI) of control responses (two-way repeated-measure ANOVA group main effects P = 0.004, F1, 6 = 23.95; P < 0.001; Bonferroni post hoc). This percentage of reduction in evoked spinal neuronal responses compared with the control response is of a higher magnitude in neuropathic rats, now 53 ± 17% (two-way repeated-measure ANOVA group main effects, P < 0.001, F1, 5 = 128.10; P < 0.001; Bonferroni post hoc). Variability for single unit in vivo electrophysiology recordings was as observed previously (fig. 3, top left graph).19
Comparatively, morphine microinjection into the anterior cingulate cortex of control animals statistically significantly reduces evoked spinal neuronal responses to 8- and 26-g vF filaments alone, and this is only to 71 and 83% of control responses (two-way repeated-measure ANOVA; P = 0.010, F1, 6 = 21.56; P < 0.01 for both respectively; Bonferroni post hoc). In neuropathic animals, evoked neuronal responses to 8- and 26-g vF stimulation are reduced to 82 and 75% of baseline neuronal firing (two-way repeated-measure ANOVA; P = 0.018, F1, 6 = 14.79; P < 0.05 and P < 0.01, respectively; Bonferroni post hoc; fig. 3, top right graph). The reductions are considerably less than those seen with the rostral ventromedial medulla injections.
Microinjection of saline into the rostral ventromedial medulla of control or neuropathic animals did not affect spinal neuronal responses to natural or electrical stimulation; all morphine injection data were compared with the baselines collected after vehicle injection. These two animal groups are therefore combined as “baseline” for the sake of data presentation (mechanical modality shown; two-way repeated-measure ANOVA group main effects P = 0.460, F1, 12 = 0.66; fig. 3, lower graph). Microinjection of saline into the anterior cingulate cortex of control or neuropathic animals had no effect spinal neuronal responses to natural or electrical stimulation (mechanical response analysis: two-way repeated-measure ANOVA group main effects P = 0.564, F1, 12 = 0.39; data not shown).
Microinjection of Morphine into the Right Central Nucleus of the Amygdala Inhibits Spinal Neuronal Responses to Natural Stimuli in Spinal Nerve Ligated Animals Only
The amygdala, connected to both ascending and descending nociceptive pathways, can modulate pain processing.25 We investigated right central nucleus of the amygdala morphine on spinal neuronal responses in control uninjured and neuropathic rats (n = 7 rats and 7 single unit cell recordings and 6 rats and 6 single unit cell recordings, respectively). In control animals, microinjection of morphine into the right central nucleus of the amygdala did not statistically significantly reduce evoked spinal neuronal responses to the range of natural stimuli (mechanical responses: two-way repeated-measure ANOVA group main effects P = 0.045, F1, 5 = 6.40, dose × von Frey interaction P = 0.081, F3, 18 = 2.63; for thermal responses: two-way repeated-measure ANOVA group main effects P = 0.160, F1, 5 = 2.72, dose × temperature interaction P = 0.842, F2, 10 = 0.18), whereas Aβ and Aδ values were statistically significantly inhibited (paired Student’s t test, P = 0.024 and 0.039, respectively; fig. 4, A, B, and C). In contrast, in neuropathic animals, the right central nucleus of the amygdala morphine statistically significantly inhibited spinal neuronal responses to innocuous and noxious mechanical (two-way repeated-measure ANOVA group main effects P = 0.001, F1, 5 = 46.40, dose × von Frey interaction P < 0.001, F3, 15 = 25.71), thermal (two-way repeated-measure ANOVA group main effects P = 0.044, F1, 5 = 7.10, dose × temperature interaction P = 0.114, F2, 10 = 0.27) and C-fiber and brush-evoked responses (paired Student’s t test, P = 0.006 and 0.013, respectively; fig. 4, D, E, and F). As a percentage of baseline, the neuronal response to 60-g von Frey stimulation was significantly reduced (65 ± 14%; mean ± 95% CI; two-way repeated-measure ANOVA, P = 0.001; data not shown).
Microinjection of Morphine into the Left Central Nucleus of the Amygdala Does Not Inhibit Spinal Neuronal Responses
Given that the amygdala can modulate both ascending and descending pain signals, it is feasible that global modulation of pain processing is possible from the central nucleus. We investigated the function of the left central nucleus of the amygdala in the context of the neuropathic (persistent) pain model by microinjecting morphine in this brain region and observing the effect on spinal neuronal responses (n = 5 rats and 5 single unit cell recordings). Interestingly, the evoked neuronal responses were not statistically significantly inhibited after this pharmacologic manipulation to mechanical (two-way repeated-measure ANOVA group main effects P = 0.571, F1, 5 = 0.38, dose × von Frey interaction P = 0.874, F3, 12 = 0.23) or thermal stimuli (two-way repeated-measure ANOVA group main effects P = 0.378, F1, 5 = 0.98, dose × temperature interaction P = 0.540, F2, 8 = 0.67); nor were Aδ, Aβ, C-fiber, postdischarge, wind-up input, or brush responses statistically significantly inhibited (fig. 4, G, H, and I). Microinjection sites for rostral ventromedial medulla, anterior cingulate cortex, and right and left central nucleus of the amygdala morphine are depicted in figure 5.
Microinjection of Morphine into the Anterior Cingulate Cortex Reduces Evoked but Not Spontaneous Neuronal Activity in the Ventral Posterolateral Thalamus Selectively in Spinal Nerve–ligated Rats
The spinothalamic tract is a key ascending pathway for the sensory components of pain. In addition, neuropathy produces both ongoing and enhanced evoked responses in the ventral posterolateral thalamus.21 We examined the effects of intraanterior cingulate cortex morphine on thalamic neuronal responses. In sham/naïve rats (n = 5 rats and 7 single unit cell recordings), discrete delivery of morphine into the anterior cingulate cortex had no effect on ventral posterolateral thalamus neuronal responses to low intensity (2 g) or noxious (26 and 60 g) punctate mechanical stimuli (two-way repeated-measure ANOVA group main effects, P = 287, F1,6=1.37) or heat-evoked responses (two-way repeated-measure ANOVA group main effects, P = 0.706, F1,6 = 0.16; fig. 6, A and B, respectively). Further there was no modulation of dynamic brush responses (paired Student’s t test, P = 0.287) or spontaneous activity (paired Student’s t test, P = 0.38; fig. 6, E and F, respectively). In contrast, in neuropathic rats (n = 7 rats and 9 single unit cell recordings), morphine inhibited ventral posterolateral thalamus neuronal responses to punctate mechanical stimuli (two-way repeated-measure ANOVA group main effects, P = 0.002, F1,8 = 21.94, dose × von Frey interaction, P = 0.069, F3, 24 = 2.70; fig. 6C) and to heat-evoked responses at noxious intensities of stimulation (48°C; two-way repeated-measure ANOVA group main effects, P = 0.044, F1,8=5.74, dose × temperature interaction P = 0.041, F2, 16 = 3.94; fig. 6D). Likewise, dynamic brush evoked responses were inhibited (paired Student’s t test, spinal nerve ligated, P = 0.018; fig. 6E). Spontaneous activity was not modulated (paired Student’s t test, P = 0.438; fig. 6F).
We investigated the effect of morphine microinjection in the rostral ventromedial medulla, right or left central nucleus of the amygdala or anterior cingulate cortex on dorsal horn and thalamic sensory nociceptive responses in uninjured or spinal nerve–ligated rats. We reveal that morphine microinjection in the rostral ventromedial medulla of uninjured animals inhibits spinal neuronal responses to a range of modalities and that this inhibition is enhanced in neuropathic rats. In contrast morphine microinjection in the anterior cingulate cortex produces limited inhibition of spinal neuronal responses and at a comparable level in uninjured and neuropathic animals. Meanwhile microinjection of morphine in the right but not left central nucleus of the amygdala inhibits spinal neuronal responses in neuropathic but not control rats. We reveal that in the ventral posterolateral thalamus, anterior cingulate cortex morphine reduces innocuous and noxious evoked responses with no effect on ongoing activity. Cumulatively, these data differentiate supraspinal opioid circuit regulation of spinal nociceptive processing and support previous studies demonstrating that the regulation of sensory and affective components of pain are likely separate.18
Morphine inhibits spinal processing and impacts descending modulatory pathway output. The rostral ventromedial medulla, with high concentrations of opioid receptors and endogenous opioids, regulates descending modulation.8 Rostral ventromedial medulla-derived descending controls are mostly dual γ-aminobutyric acid– and opioidergic13 with a clear influence on pain thresholds.26 Rostral ventromedial medulla morphine microinjection successfully demonstrated that brain circuits modulate evoked spinal neuronal activity through descending controls under these conditions. Neuropathy comprises ongoing and stimulus-evoked pain elements, and the anterior cingulate cortex contributes to the ongoing aversive state in nerve-injured rodents18 with a differential effect on sensory and affective pain responses.19,27 In the clinic, patients with neuropathic pain can present with ongoing pain, allodynia, and/or hyperalgesia.28 Distinct underlying mechanisms may govern ongoing versus evoked pain because hyperalgesia can be alleviated pharmacologic in the absence of the relief of ongoing pain.29 Preclinically, somatosensory signals in the dorsal horn of the spinal cord are pathologic altered after nerve injury and thalamus30 and pre- and postsynaptic plasticity in the anterior cingulate cortex is observed,27 as well as in the amygdala.31 Reduced µ opioid receptor density in the spinal cord after peripheral nerve injury is a possible explanation for opioid-insensitive components of neuropathic pain.32 Whether or not µ opioid receptor density is altered in spinal nerve–ligated rats is unclear. The anterior cingulate cortex may influence nociceptive sensory transmission by explicit regulation of spinal neuronal activity.33 We showed that morphine microinjection here had no effect on threshold pain responses to mechanical or thermal stimuli but in contrast led to a small reduction in evoked spinal neuronal responses to suprathreshold mechanical and thermal stimuli. The inhibitory effects were of a smaller magnitude than those seen with rostral ventromedial medulla morphine and did not vary between control and neuropathic animals. Endogenous opioid receptor activity in the anterior cingulate cortex was previously strongly implicated in pain relief as assessed using conditioned place preference.18 Because we have investigated spinal neuronal evoked responses in the present study, we project that we are observing a separate opioid-led interaction. Of note, only male rats were used in this study. Although this allows a direct comparison between data collected previously, we acknowledge that this is a limitation because sex is a biologic variable that could impact the results gathered. A further limitation is that although descriptive, the present study does not investigate neuronal markers; this could have provided mechanistic insight relating to brain region–specific activity.
Sensory and affective areas of the brain exert modulatory actions on spinal cord activity, where the peripheral barrage first meets central circuits. Despite morphine’s central actions, it does not adequately relieve the ongoing pain and/or evoked hypersensitivities experienced by neuropathic patients; investigating the underlying pathologic mechanisms and neural sites in neuropathy, with the aim of aiding improved therapy development, is important because the successful treatment of neuropathic pain in patients is a large unmet clinical need.34 Behavioral studies evaluate the ability of an animal to detect a noxious threshold. Models of neuropathy (after spinal nerve ligation for example) allow behavioral changes and subsequent analgesic “relief” to be assessed, whereas subsets of components of neuropathic pain—for example, evoked versus ongoing pains—can be dissected using conditioned place preference assays.35 Recently pain behaviors were analyzed in control and spinal nerve–ligated rats after morphine microinjection in the anterior cingulate cortex or rostral ventromedial medulla. The study revealed separable effects of rostral ventromedial medulla versus anterior cingulate cortex morphine in naïve animals because only the former inhibited the tail flick response. Meanwhile in neuropathic rats rostral ventromedial medulla morphine (1) reduced tactile allodynia, (2) produced conditioned place preference, and (3) produced antihyperalgesic and analgesic effects against natural stimuli.19
In comparison with behavioral tests the in vivo electrophysiology technique described here allows the recording of deep dorsal horn wide dynamic range spinal neuronal responses to suprathreshold stimuli in anaesthetized rats. By comparing the mechanically and thermally evoked activity of rat spinal neurons with human psychophysical responses to thermal stimuli, it is evident that rodent spinal neuronal responses code in a similar way to that observed with human perception.36 Regarding anterior cingulate cortex morphine in our electrophysiology study, the inhibitory effect observed was most obvious only for midrange stimuli and not the lowest forces that would likely reflect conscious behavioral responses.
We conclude that anterior cingulate cortex morphine has minimal effects on evoked spinal responses in both control and neuropathic conditions. A differential role of the anterior cingulate cortex in the regulation of different types of pain conditions has previously been postulated with a lack of control of evoked responses.37 Other studies33,38 have suggested bidirectional control of nociception by anterior cingulate cortex circuits or facilitations. Regarding facilitations, this action seemed to act mostly on lamina I neurons rather than the deep neurons we recorded here and was independent of the rostral ventromedial medulla, involving direct spinal projections but was occluded after nerve injury. Clearly complex top-down controls from the anterior cingulate cortex exist. Another pathway from the midcingulate zone to the posterior insula controls a descending 5HT3 facilitation that relays in the rostral ventromedial medulla area.39 Thus, cingulate cortex outputs acting on the spinal cord may be direct or indirect; given the clinical importance of opioids in pain management, it seems clear that understanding whether or not opioid circuits play a role in these outputs could lead to improved therapeutic outcomes for pain control.
The limbic brain central nucleus of the amygdala also impacts descending modulatory controls,40,41 and hemispheric lateralization of this brain region in terms of impact on nociception has been previously documented.42–44 After peripheral neuropathy there is an assymetric time-dependent activation of the right central nucleus of the amygdala.9 The consensus that right, but not left, central nucleus of the amygdala is pronociceptive is challenged by recent research suggesting that the latter does indeed contribute to mechanical allodynia after a contralateral peripheral neuropathy.31 Under the conditions of our study, opioidergic manipulation of the right central nucleus of the amygdala only caused inhibition of spinal neuronal responses to innocuous and noxious natural stimuli after spinal nerve ligation surgery. This indicates a laterality of central control that is altered by neuropathy.
After nerve injury, pain transmission in the central nervous system is enhanced by multiple spinal and brain mechanisms. Interestingly, in the spinal cord wide dynamic range neurons with intact connections to the periphery do not display ongoing activity, yet deafferented neurons do.45 Furthermore, the baseline responses of these neurons are not enhanced to peripheral stimuli despite the neuropathy yet have enlarged receptive fields so that a given stimulus should activate more neurons.46 Indeed, the consequences of these changed spinal neuronal populations are reflected in changes seen in the ventrobasal thalamus, a major sensory relay for onward transmission of spinal nociceptive projections in that neurons here have both enhanced responses and ongoing activity.21 However, anterior cingulate cortex morphine microinjection had no effects in sham-operated animals or on ongoing activity in and only marginally modulated thermal and mechanical evoked activity in neuropathic animals, similar to what was seen with spinal neuronal recordings. The ability of the brain to modulate spinal neuronal responses under the anesthetic conditions of in vivo electrophysiology was verified by the rostral ventromedial medulla morphine injections, where 10 μg was highly effective.
In conclusion, the data reveal differential effects of morphine at distinct brain sites on evoked pain responses. Descending controls from the rostral ventromedial medulla modulate evoked activities in the spinal cord and are altered after nerve injury; this plasticity creates an environment whereby morphine injection into the right central nucleus of the amygdala can modulate spinal neuronal activity. We suggest that whereas anterior cingulate cortex morphine is able to preferentially reduce the ongoing aversive state, it fails to markedly modulate evoked responses in sensory pathways such as the spinal cord and thalamus, consistent with clinical observation.
Supported in part by a strategic award to the Wellcome Trust Pain Consortium (London Pain Consortium, London, United Kingdom; to Drs. Bannister, Dickenson, Patel, and Porreca) and by National Institute on Drug Abuse grant No. R01DA041809 (North Bethesda, Maryland; to Dr. Navratilova).
The authors declare no competing interests.