Cross-talk between Human Spinal Cord μ-opioid Receptor 1Y Isoform and Gastrin-releasing Peptide Receptor Mediates Opioid-induced Scratching Behavior

It is more than 40 yr since Snyder’s landmark discovery of spinal cord opioid receptors in rats,¹  followed within 3 yr by the clinical description of spinal opioids to treat pain in humans.²  Spinal opioids have become a mainstay of clinical pain management in labor,^3–5  in perioperative care,^6–11  and in chronic pain management.^12,13  Spinal opioids are frequently coadministered with local anesthetics to provide analgesia requiring lower doses of each agent, thus minimizing most undesirable side effects.^14,15  Most opioid-induced side effects such as nausea, vomiting, and respiratory depression are more commonly observed following systemic administration.⁸  By contrast, pruritus is more commonly observed following intrathecal or epidural administration,^16,17  particularly in obstetric patients.^18,19  Spinal opioid-induced pruritus is commonly described for all spinal opioids and for both intrathecal and epidural routes. It is one of the most irritating side effects following spinal opioid administration and markedly decreases patient satisfaction. Nevertheless, the mechanisms by which opioids induce pruritus in human remain poorly understood.

The Oprm1 gene that encodes the μ-opioid receptor (MOR) comprises dozen of alternative splicing variants or isoforms in rodents and humans, and some might be responsible for physiologic function unrelated to analgesia^20–22  and for individual variability in sensitivity to opioid drugs.^23,24  Most Oprm1 isoforms differ in the structurally distinct C-terminals, which may underlie their functional differences.²²  Gastrin-releasing peptide receptor (GRPR) is a G_q protein-coupled receptor; it is expressed in the dorsal spinal cord, whereas its endogenous neuropeptide ligand, gastrin-releasing peptide, is expressed in sensory neurons, but not in the spinal cord.^25–27  GRPR is a pivotal receptor in the spinal cord for relaying nonhistaminergic itch transmission from the periphery to the brain.^28–30  Unlike Oprm1 isoforms, no GRPR splicing variant has been reported in mice or in humans. In mice, we demonstrated that spinal opioids induce pruritus by activating the MOR1D isoform which cross-activates GRPR in the spinal cord GRPR neurons.³¹  By contrast, opioids induced analgesia through the MOR1 isoform that is not expressed in GRPR neurons.³¹  These findings provide a molecular and neural basis for potentially uncoupling opioid-induced analgesia from opioid-induced itch in the spinal cord. In primates, it has been shown that intrathecal gastrin-releasing peptide induces scratching but does not affect pain sensation.^32,33  Furthermore, aging primates with chronic itch exhibit increased expression of gastrin-releasing peptide in sensory neurons and increased expression of GRPR in the spinal cord.³⁴  These studies strongly imply that the gastrin-releasing peptide–GRPR itch-specific signaling pathway is evolutionarily conserved between rodents and primates. Nevertheless, it was not clear that this was the mechanism for spinal opioid-induced pruritus in humans.

In this study, we tested the hypothesis that spinal morphine induces pruritus in humans through human Oprm1 splicing variants that interact with GRPR, in a manner similar to the established interaction between MOR1D and GRPR in mice. We demonstrate the existence of multiple Oprm1 isoforms in human spinal cord, and our studies suggest that MOR1Y cross-activates GRPR to cause opioid-induced pruritus in humans. These findings have important implications for therapeutic inhibition of pruritus without compromising the efficacy of opioid analgesia.

The study protocol was approved by the Research Advisory Committee of Mid-American Transplant (St. Louis, Missouri). Specimens of human spinal cord were obtained from fresh, cold-stored cadavers. Donors were a 73-yr-old male and a 67-yr-old male. Both had sustained sudden cardiac arrest with resuscitation and subsequent withdrawal of care occurring in early evening, approximately 16 h and 18 h before sample collection, respectively. Both donors had coronary artery disease, hyperlipidemia, type 2 diabetes mellitus, and hypertension. The first donor also had end-stage pulmonary fibrosis, pulmonary hypertension, cardiac failure, renal failure, and chronic steroid medication. A midline posterior incision was made in the thoracolumbar area, carried down to the fascia, and then subperiosteally to expose the lamina, pars, and facet joints. Lamina from T9 to T12 (first specimen) and T8 to T11 (second specimen) were exposed. An oscillating saw was used to cut the lamina laterally at the level of the posterolateral paraspinal gutters. Resection of the lamina was performed using a Kerrison rongeur. On the left side, the pars and facets were resected, exposing the nerve roots to a point beyond their dorsal root ganglion; nerves were cut distal to the dorsal root ganglion. The dura was opened in the midline and the spinal cord was cut proximally and then distally. Sutures were placed to identify the dorsal rostral and dorsal caudal aspects of each specimen. The right sided nerve roots were cut intradurally. A sleeve of dura was left around the nerve roots and the dorsal root ganglion on the left side. The specimens were placed in in RNA stabilizer (RNAlater; QIAGEN, Germany) and kept frozen on dry ice immediately.

The linear amino acid sequences alignments of human MOR1 (hMOR1; NP_000905.3), mouse MOR1 (mMOR1;NP_001289722.1), human GRPR (hGRPR; NP_005305.1), and mouse GRPR (mGRPR; NP_032203.1) were performed using William Pearson’s lalign program (version 2.1.30).³⁵ 

We designed isoform-specific reverse transcription polymerase chain reaction primers and performed reverse transcriptase polymerase chain reaction using the human spinal cord samples. Sequences of splicing variants specific primers are listed in table 1. Reverse transcriptase polymerase chain reaction was performed as previously described.^31,36  Briefly, total RNA was isolated and genomic DNA was removed in accordance with manufacturer’s instructions (RNeasy plus mini kit; QIAGEN). RNA was quantified using Eppendorf BioPhotometer and stored at −80°C. Single-stranded cDNA was synthesized from 1 μg of total RNA using High Capacity cDNA Reverse Transcription Kit (Life Technologies, USA) and stored at −20°C until analysis. Thermal cycling was initiated with denaturation at 95°C for 10 min. After this initial step, 35 cycles of polymerase chain reaction (heating at 95°C for 10 s, 56°C for 30 s, and 70°C for 1 min) were performed.

The plasmids pcDNA3.1/hMOR1, hMOR1B1, hMOR1B3, hMOR1B5, hMOR1O, hMOR1X, and hMOR1Y were provided by Dr. Y. X. Pan, M.D., Ph.D.; and pcDNA/hGRPR was provided by Dr. James Battey, M.D., Ph.D. Splicing variants-specific 3’ end sequence of hMOR1A, hMOR1B2, and hMOR1B4 were integrated into reverse primers and synthesized by Integrated DNA Technologies (USA). Full cDNA segments were amplified by polymerase chain reaction from hMOR1 template and sub cloned into pcDNA3.1 between Nhe I and Xba I site. All the plasmids were verified by DNA sequencing.

HEK 293 cells were grown in Dulbecco modified Eagle medium supplemented with 10% fetal bovine serum in a humidified atmosphere containing 5% CO₂. The cells were first transfected with pcDNA3.1/hGRPR (20 μg/10⁷ cells) containing the neomycin resistance by electroporation (GenePulser Xcell; Bio-Rad, USA). Stable transfectants were selected in the presence of 500 μg/ml G418 (Invitrogen, USA). To generate lines coexpressing two receptors, the cells were subjected to a second round of transfection and selected in the presence of 500 μg/ml G418 and 100 μg/ml hygromycin (Roche, Switzerland). Clones expressing hGRPR, hMOR1B1, hMOR1Y, hMOR1B1/hGRPR, and hMOR1Y/hGRPR were generated.

To identify a human MOR1 splicing variant that interacts with GRPR, we performed calcium imaging on HEK 293 cells and examined functional interactions between hGRPR and different human Oprm1 C terminal splicing variants. HEK 293 cells stably expressing hGRPR were planted on 12-mm round coverslips coated with poly-D-lysine. Plasmids of different human Oprm1 C terminal splicing variants (1 μg per coverslip) were transfected using Lipofectamine 3000 (ThermoFisher Scientific, USA) in accordance to manufacturer’s instructions. Twenty-four hours after transfection, cells were used for calcium imaging as previously described.³⁷  Fura-2 acetoxymethyl ester (Molecular Probes; ThermoFisher Scientific) was diluted to 2 mM stock in dimethyl sulfoxide/20% pluronic acid. Cells were loaded with fura 2-acetomethoxy ester (Molecular Probes) for 30 min at 37°C. After washing, the coverslips were mounted on a Warner Instruments (USA) recording chamber (RC 26G) perfused with calcium imaging buffer at a rate of ~2 ml/min. An inverted microscope (Eclipse Ti 10X objective; Nikon, Japan) with Roper CoolSNAP HQ2 digital camera was used for fura-2 calcium imaging after 340/380-nm laser excitations (sampling interval, 2 s; exposure time adjusted for each experiment to ~40 ms for 340 nm and to ~30 ms for 380 nm). Responsive cells were identified as regions of interest and F340/F380 ratios were measured using NISElements (version 3.1, Nikon). Calibration of intracellular calcium concentrations ([Ca²⁺]_i) was performed using Fura-2 Calcium Imaging Calibration Kit (Invitrogen) following manufacturer instruction. Each experiment was done at least three times, and at least 30 cells were analyzed each time. Cells were incubated first with morphine (1 μM), and then with gastrin-releasing peptide (1 nM; Bachem, Switzerland), which were dissolved in sterile saline.

For antagonists and transactivator of transcription peptides studies, stable cell lines expressing hMOR1Y/hGRPR were used. Naltrexone (10 μM) and GRPR antagonist (D-Phe-6-Bn(6 to 13)OMe; 1 μM) were preincubated for 15 min before imaging. Transactivator of transcription-hMOR1Y was added into cell cultures 1 h before fura-2 and present in the incubation buffer for all of the following steps.

C-terminal splicing variants of hMOR1 share identical amino acid sequence from M1 to Q388. Each variant has a unique C terminal tip that may account for their specific functions. To selectively inhibit specific Oprm1 isoform, we synthesized a transactivator of transcription–fusion peptide containing a transactivator of transcription (YGRKKRRQRRR), a transactivating domain of human immunodeficiency virus protein that can permeate cell membrane,³⁸  fused to the unique C-terminal tip of the hMOR1Y isoform (IRDPISNLPRVSVF). We only created a transactivator of transcription–fusion peptide for Oprm1 isoforms that were detected in the spinal cord and that exhibit GRPR specific co-hybridization in the calcium imaging studies. Introduction of transactivator of transcription–fusion peptides permits their competition with corresponding receptors for interaction with hGRPR both in vitro and in vivo.

Male C57BL/6J mice of 25 ± 3 g body weight between 7 and 12 weeks old were purchased from the Jackson Laboratory. All mice were housed under a 12 h light/dark cycle with food and water provided ad libitum. All experiments were performed in accordance with the guidelines of the National Institutes of Health (Bethesda, Maryland) and the International Association for the Study of Pain (Washington, D.C.) and were approved by the Animal Studies Committee at Washington University (St. Louis, Missouri).

Nineteen mice were randomly assigned to three groups (0, 0.1, and 0.4 nmol of transactivator of transcription-hMOR1Y) for morphine-induced scratching behaviors and analgesia using simple randomization. Twenty-one mice were randomly assigned to three groups (0, 0.1, and 0.4 nmol of transactivator of transcription-hMOR1Y) for gastrin-releasing peptide-induced scratching behaviors. Behavioral experiments were performed between 9:00 and 11:00 am. The injection area was shaved 2 days before experiments. To examine the effect of transactivator of transcription–fusion peptide at the spinal cord level, saline or different doses of transactivator of transcription–hMOR1Y (0.1 nmol and 0.4 nmol) were administered via intrathecal injections. Animals were treated in sequential order. After 30 min, each mouse was placed in a plastic arena (10 × 11 × 15 cm) for another 30 min to acclimatize. Mice were then briefly removed from the chamber for intrathecal injections of morphine or gastrin-releasing peptide. Morphine was injected at the dose of 0.3 nmol per mouse that induces robust scratching behaviors and thermal analgesic effect.³¹ 

Scratching behaviors were videotaped (HDR-CX190; Sony, Japan) for 30 min. The videos were played back on a computer and the quantification of mice behaviors were performed by investigators who were blinded to the treatments. One scratching episode is defined as a lifting of the hind limb toward the body and then a replacing of the limb back to the floor or the mouth, regardless of how many scratching strokes take place between those two movements.

Warm water tail immersion assay was conducted as described previously.³⁹  Thirty minutes after morphine injection, mice tails were dipped beneath the 50°C water in a temperature-controlled water bath (IITC Inc., USA). The latency to withdrawal was measured with a 15-s cutoff. For opioid analgesia, tail-flick results were expressed as percentage of maximum possible effect [%MPE = (post drug latency – pre drug latency) × 100/ (cutoff time – pre drug latency)].

One-way ANOVA with one between-subject factor of drug treatment followed by post hoc analysis (Dunnett) was used for multiple group comparisons. Two-tailed independent t test was used for two group comparisons. Normality and equal variance assumptions were tested with the Kolmogorov–Smirnov and the Brown-Forsythe median test, respectively. All statistical analyses were carried out with Prism 8 (v8.0.1; GraphPad, USA). The data are presented as mean ± SD, and P < 0.05 was considered statistically significant. No a priori statistical power calculation was conducted to guide sample size; sample size was based on our previous studies. No data points were omitted as outliers.

To identify Oprm1 splicing variants from human spinal cord, we first isolated cDNA from the dorsal horn and ventral horn of human spinal cords. Using reverse transcription polymerase chain reaction, we were able to detect distinct bands corresponding to hMOR1, hMOR1A, hMOR1B5, and hMOR1Y (fig. 1A). In the ventral horn, reverse transcription polymerase chain reaction bands of hMOR1 and hMOR1A were present, suggesting that these two variants are broadly expressed (fig. 1B). A band corresponding to hGrpr was also present in the dorsal horn of human spinal cord (fig. 1C). Reverse transcription polymerase chain reaction results demonstrate that various hMOR splicing variants are expressed in human spinal cord (fig. 1A).

We previously showed that morphine induces itch signaling by activating MOR1D–GRPR cross-talk³¹  in mice. Of note, the amino acid sequences of MOR and GRPR are highly conserved in human and mice (fig. 1, D and E). As such, a key criterion for identifying a human Oprm1 splicing variant that interacts with GRPR in a similar manner to the behavior of MOR1D in mice, is that the isoform should mediate calcium transients in HEK 293 cells expressing GRPR when morphine is added.³¹  Of five splicing variants that were detected in human dorsal horn, only hMOR1Y cross-activated calcium signaling of hGRPR after morphine incubation (fig. 2A). Morphine failed to evoke calcium responses when other splicing variants were coexpressed with hGRPR (fig. 2B and Supplemental Digital Content 1, https://links.lww.com/ALN/B956). In contrast, morphine did not induce calcium responses in HEK 293 cells that were singly transfected with hMOR1Y or hGRPR (fig. 2C). When cells were preincubated with naltrexone, a MOR antagonist that inhibits morphine-induced pruritus in primates,³³  morphine-induced calcium spikes were completely blocked in hGRPR/hMOR1Y cells (mean difference, 56 ± 3 nM) (fig. 3, A and B, red). Morphine-induced calcium spike was also blocked by the GRPR antagonist (mean difference, 54 ± 3 nM) (fig. 3, A and B, blue). These data suggest that hMOR1Y is capable of mediating the effect of morphine on cross activation of hGRPR.

Our previous studies using a transactivator of transcription–peptide approach showed that the seven amino acid sequence in the C terminal of MOR1D in mice is critical for cross-talk between MOR1D and GRPR because Tat-peptide could attenuate MOR1D–GRPR cross-signaling.³¹  To test whether hMOR1Y and hMOR1B1 may cross-activate hGRPR through their C-terminal, we synthesized transactivator of transcription–fusion peptides containing a transactivator of transcription (YGRKKRRQRRR), fused to the C terminal of hMOR1Y (IRDPISNLPRVSVF) that we demonstrated to interact with hGRPR, and tested whether the transactivator of transcription–peptide could attenuate the cross-talk with hGRPR. We first incubated transactivator of transcription–hMOR1Y with HEK 293 cells expressing hGRPR/hMOR1Y. Remarkably, preincubation of transactivator of transcription–hMOR1Y for 2 h dose-dependently blocked morphine-induced calcium spikes in hGRPR/hMOR1Y-expressing cells (mean difference, 28 ± 2 nM, vehicle vs. 50 μM; 66 ± 2 nM, vehicle vs. 200 μM) (fig. 3, C and D). High concentration of Tat-hMOR1Y (200 μM) also showed inhibiting effect on gastrin-releasing peptide-induced calcium responses (mean difference, 12 ± 2 nM) (fig. 3, C and D). These results suggest that the C-terminus of hMOR1Y may direct interact with hGRPR.

The finding that the transactivator of transcription–hMOR1Y could block hMOR1Y and hGRPR cross-talk supports two important notions. First, GRPR between human and mouse is conserved (fig. 1E). Second, the C-terminal of hMOR1Y and mMOR1D may share important biochemical characteristics enabling their cross-activation of GRPR, irrespective of species (table 2). To further examine this, we tested the effect of transactivator of transcription–hMOR1Y on mouse scratching behaviors induced by intrathecal injections of morphine or gastrin-releasing peptide. Remarkably, transactivator of transcription–hMOR1Y dose-dependently attenuated morphine-induced scratching behaviors (mean difference, 24 ± 16, 0 nmol vs. 0.1 nmol; 54 ± 17, 0 nmol vs. 0.4 nmol) (n = 6 to 7 mice per group; P = 0.011, one-way ANOVA) (fig. 4A) with no effect on morphine-induced antinociception (fig. 4B). Interestingly, transactivator of transcription–hMOR1Y also blocked gastrin-releasing peptide-induced scratching behaviors (mean difference, 20 ± 18, 0 nmol vs. 0.1 nmol; 6 ± 18, 0 nmol vs. 0.4 nmol) (n = 7 mice per group; P = 0.013, one-way ANOVA) (fig. 4C). These results suggest that transactivator of transcription–hMOR1Y may directly block mouse GRPR function. To test this, we performed calcium imaging on HEK 293 cells coexpressing mMOR1D/mGRPR. Preincubation of transactivator of transcription–hMOR1Y for 2 h significantly blocked morphine and gastrin-releasing peptide-induced calcium spikes (mean difference, 101 ± 6 nM for morphine; 103 ± 4 nM for gastrin-releasing peptide) (fig. 4, D and E). Thus, transactivator of transcription–hMOR1Y could function as a GRPR antagonist in MOR1D/GRPR neurons. Taken together, the expression of hMOR1Y in the dorsal horn of the spinal cord and its interactions with hGRPR suggests that hMOR1Y may play important roles in morphine-induced pruritus.

This work presents several major findings that are directly relevant to our understanding of itch transmission in humans. First, we demonstrate the existence of hGRPR in human spinal cord. GRPR was discovered to be an itch receptor in mice in 2007.²⁸  The present work represents the evidence suggesting that GRPR also functions as an itch receptor that mediates opioid-induced pruritus in human spinal cord. Second, we demonstrate that the Oprm1 gene undergoes extensive alternative splicing in human spinal cord, with marked regional differences between the ventral and dorsal horns. Strikingly, the regional variation of expression patterns of splicing variants of Oprm1 in the human spinal cord mirrors that in mouse spinal cord,^40–42  further supporting the notion that the Oprm1 splicing variants are highly conserved across animal species.²²  Third, our data suggest that hMOR1Y and hGRPR can cross-talk, and the C-terminal of hMOR1Y is required for cross-activation of GRPR-mediating itch transmission. This finding is reminiscent of mouse MOR1D–GRPR cross-talk in the spinal cord,³¹  strongly indicating a functional conservation of opioid-induced itch pathways between mice and humans. Lastly, we show that transactivator of transcription–hMOR1Y can function as a GRPR antagonist and exert its anti-GRPR signaling in the absence of morphine but the presence of hMOR1Y. These findings imply that the mere presence of hMOR1Y may alter the conformation and functionality of hGRPR, and thereby enable transactivator of transcription–hMOR1Y to interfere the binding of gastrin-releasing peptide to GRPR.

We previously demonstrated uncoupling between the desired spinal opioid-induced antinociceptive effects mediated via the MOR1 isoform and the unwanted spinal opioid-induced scratching behavior mediated via the MOR1D isoform in mice.³¹  This double disassociation of function occurs at different cell types in mouse spinal cord because MOR1D is expressed in GRPR neurons which are dedicated to itch but not pain transmission, while MOR1 is expressed in non-GRPR neurons, most likely nociceptive neurons.²⁹  Interfering with the seven amino acids at the C-terminal of MOR1D using a transactivator of transcription–peptide approach blocks MOR1D–GRPR crosstalk and thereby attenuates the scratching behavior following spinal opioids.³¹  Therefore, opioids induce itch and analgesia through completely distinct neural pathways in the mouse spinal cord.

In this report, we have identified the sequence for each of the Oprm1 isoforms of the human spinal cord. Assessment of the free cytosolic calcium concentration in transfected HEK cells showed that the human MOR1Y isoform can be activated by both morphine and gastrin-releasing peptide, and this activation can be blocked by a transactivator of transcription–peptide we designed to bind specifically to the unique C-terminal of the hMOR1Y isoform and by the MOR antagonist naltrexone (fig. 4F). Furthermore, we showed that transactivator of transcription–hMOR1Y dose-dependently attenuated scratching behaviors induced both by morphine and by gastrin-releasing peptide in mice, with no effect on morphine-induced analgesia. To our surprise, we could not find similarity between the C-terminals of hMOR1Y and mMOR1D (12 vs. 7 amino acids in length). This suggests that the precise mechanisms by which the C-terminals interact with GRPR may differ between mice and humans. Alternatively, it is possible that the crosstalk may occur following conformational change of one or two receptors rather than a direct physical heteromeric interaction. The C-terminal could exert the effect through interfering in downstream signaling events rather than direct heteromeric interaction in humans. Because specific antibody against hMOR1Y is not available, we are unable to assess this possibility in this study. Future function-structural characterization is required to identify minimal amino acids of the C-terminals required for crosstalk and signaling components affected.

Owing to technical difficulties in human studies, particularly the lack of a specific hMOR1Y antibody, we were unable to determine whether hMOR1Y and hGRPR are expressed in distinct subpopulations of neurons in the spinal cord, nor to functionally verify the role of hMOR1Y by siRNA knockdown as we did in mice. However, the observations that hMOR1Y and GRPR engage in crosstalk and that perturbation of the C-terminal of hMOR1Y impairs GRPR signaling and scratching behavior in mice strongly argues that structurally and functionally the crosstalk between splicing Oprm1 isoform and GRPR is evolutionarily conserved between human and mouse.

There are two important clinical applications of our findings. First, the study provides a potential pharmacologic target to treat or prevent spinal opioid-induced pruritus in human, resulting in a drug which could be coadministered together with spinal opioids. It is conceivable that two novel pharmacologic strategies could be employed to tackle opioid-induced pruritus. First, one may target GRPR by the coadministration of morphine with a GRPR antagonist. In addition, opioid-induced pruritus may be specifically alleviated by the coapplication of morphine and a transactivator of transcription–peptide targeting the hMOR1Y C-terminal.⁴³  These approaches would be preferable to the current practice where opioid antagonists^44,45  or histamine antagonists^46,47  have been used to attempt to tackle severe opioid-induced pruritus, since these drugs are—at best—only partially effective, and since opioid antagonists compromise the desired analgesic effects of spinal opioids. Second, the discovery of multiple splicing variants in the human spinal cord raises an important question as to their specific function in humans, which has been largely ignored and remain unknown to date. A major challenge for overcoming the limitation of opioid drugs in pain management is to diminish their unwanted side effects without affecting their analgesic action.^48,49  Our findings should significantly strengthen the notion that activation of distinct Oprm1 isoforms may account for some side effects associated with opioid drugs in pain management. As such, these findings may herald future efforts to uncouple desired opioid analgesia from other adverse side effects, such as addiction and respiratory depression.

The authors gratefully appreciate the assistance of Mid-American Transplant, St Louis, Missouri, who facilitated our harvesting of human spinal cord samples.

Supported by the National Institutes of Health (Bethesda, Maryland) grant Nos. AR056318-06, NS094344, and RO1DA037261-01A1 (to Dr. Chen).

The authors declare no competing interests.