Small G-Protein Rheb Gates Mammalian Target of Rapamycin Signaling to Regulate Morphine Tolerance in Mice

Adult C57BL/6J male and female mice (Animal Center, Chinese Academy of Sciences, Shanghai, China) weighing 20 to 25 g were used in this study. The mice were housed (three to five animals per cage) in a room maintained at a constant ambient temperature of 22 to 23°C and humidity of 50 to 60%, with an alternating 12-h light/12-h dark cycle. Food and water were provided ad libitum. All efforts were made to minimize the number of animals used and animal suffering. The animals were randomly assigned to groups using simple randomization, and mice of different groups were cohoused. Age-matched littermates without the Cre gene served as control mice in each experiment.

Mice with nestin Cre and the Rheb S16H knock-in allele were generated in the laboratory of Dr. Worley.¹⁶  The LoxP-flanked tPA (transcriptional stop) was located upstream of the Rheb S16H knock-in allele. Rheb cDNA transcription started once the floxed tPA (transcriptional stop) was removed by the Cre recombinase and central nervous system–specific knock-in of Rheb (Rheb knock-in: Rheb k/k; Nestin cre) was generated. The genotype of the transgenic mice was determined by PCR using the following primers to distinguish between the wild-type and knock-in allele and cre-mediated excision of the stop signal: WTF1, 5′-GCA CTT GCT CTC CCA AAG TC-3′; WTR1, 5′-GCG GGA GAA ATG GAT ATG AA-3′ to amplify wild-type allele (596 bp); FloxF, 5′-GCA CTT GCT CTC CCA AAG TC-3′; FloxR, 5′-GGG GAA CTT CCT GAC TAG GG-3′ to amplify the knock-in allele (395 bp); 5′-TGC CAC GAC CAA GTG ACA GCA ATG-3′ (forward), and 5′-ACC AGA GAC GGA AAT CCA TGG CTC-3′ (reverse). The primers generated a 400-bp amplicon and were used for amplification of nestin Cre. Both male and female adult mice were used for behavioral studies.

Mice with CaMKII cre and loxP-flanked Rheb allele were generated in the laboratory of Dr. Worley.¹⁶  Floxed Rheb mice were crossed with CaMKII Cre transgenic mice to generate mice with neuron-specific Rheb deletion (Rheb conditional knockout [cKO]: Rheb f/f; CaMKII cre). Genomic DNA was isolated from the ears, and genotyping was performed using the following primers to amplify wild-type (650 bp) and the floxed allele (850 bp): 5′-GCC CAG AAC ATC TGT TCC AT-3′ (forward) and 5′-GGT ACC CAC AAC CTG ACA CC-3′ (reverse). The primers used for the amplification of CaMKII Cre were 5′-GAC AGG CAG GCC TTC TCT GAA-3′ (forward), and 5′-CCT CTC CAC ACC AGC TGT GGA-3′ (reverse), with an amplicon of 500 bp. Both male and female adult mice were used for behavioral studies.

pAAV-CaMKII-GFP-Cre and pAAV-VGAT-EGFP-Cre were obtained from Genechem Company (China). pAAV-CBG-DIO-EGFP-miR30shRNA(Rheb) (target: NM_053075.3) was designed and synthesized by Obio Company (China). AAV9 vector was injected at a maximum feasible dose, with a total volume of 5 μl per animal.

An mTOR small interfering RNA (siRNA; catalog no. 6548) and its control scrambled siRNA (catalog no. 6332) were purchased from Cell Signaling Technology Inc. (USA). TurboFect in vivo transfection reagent (Thermo Scientific Inc., USA) was used as a delivery vehicle for siRNA to prevent degradation and enhance cell membrane penetration. The mice were injected intrathecally with siRNA or vehicle once daily for 3 days. On day 7, the mice were administered intrathecal injections of saline or morphine, as described below.

Morphine (Shenyang First Pharmaceutical Factory, China) was dissolved in saline at a final concentration of 1 µg/µL. Metformin was purchased from MedChemExpress (USA; catalog no. HY-17471A) and dissolved in saline at a final concentration of 20 µg/µL. Rapamycin (V900930; Sigma–Aldrich, USA) was dissolved in 0.1% dimethyl sulfoxide in saline at a final concentration of 1 µg/µl. All drugs were delivered using an insulin syringe (BD Biosciences, USA). The injection was performed by lumbar puncture into the subarachnoid space of the lumbar thecal at approximately the lumbar 4/5 level, as previously described.¹⁷ 

To set the morphine tolerance model, the mice were injected intrathecally with saline (10 µl) or morphine (10 µg/10 µl) twice daily at 12-h intervals for 5 consecutive days. Analgesia was assessed 30 and 60 min later via the tail-flick assay.

To determine the analgesic effect of morphine in Rheb knock-in mice, nociceptive behavior was examined in the following groups: control with 10 µl of saline, control with 10 µg of morphine, knock-in with 10 µl of saline, and knock-in with 10 µg of morphine. For dose–response analysis, the doses tested were 1, 2, 4, 8, 16, 32, and 64 μg at day 6.

To determine the analgesic effect of morphine in Rheb cKO mice, nociceptive behavior was examined in the control with morphine (control + morphine 1 µg) and cKO with morphine (cKO + morphine 1 µg) groups. To determine the development of morphine tolerance in cKO mice, behavioral changes were examined in the following four groups: control with 10 µl of saline, control with 10 µg of morphine, cKO with 10 µl of saline, and cKO with 10 µg of morphine.

To determine the effect of rapamycin on the process of morphine tolerance, vehicle or rapamycin was intrathecally injected 30 min before morphine treatment for 5 days, and the nociceptive behavioral changes were examined in the following groups every single day: 10 µl of saline with 10 µl of vehicle, 10 µl of saline with 10 µg of rapamycin, 10 µg of morphine with 10 µl of vehicle, and 10 µg of morphine with 10 µg of rapamycin. mTOR siRNA (catalog no. 6332) and control scrambled siRNA (catalog no. 6568) were purchased from Cell Signaling Technology Inc. TurboFect in vivo transfection reagent (Thermo Scientific Inc.) was used as a delivery vehicle for siRNA to prevent degradation and enhance cell membrane penetration. The mice were injected intrathecally with siRNA or vehicle once daily for 3 days. On day 7, the mice were administered intrathecal injections of saline or morphine as described above. To determine the effect of metformin on the development of morphine tolerance, saline or metformin (200 mg/kg, ip) was administered 30 min before the intrathecal injection of morphine for 5 days, and tail-flick latencies were examined in all groups every day.

The tail-flick test was used to evaluate the antinociceptive effect of the drugs.^18,19  Briefly, the tip of a mouse’s tail was submerged into hot water (52.5 ± 0.5°C), and the time until it was lifted from the water was recorded, which was defined as tail-flick latency. To avoid tissue injury, a cut-off time of 10 s was set. Response latencies were recorded three times, with a 10-min interval between each reading. Response latency was measured before (baseline) and at the indicated time after drug administration. All behavioral tests were carried out by a technician who was blinded to the experimental groups.

After the behavioral tests, the mice were euthanized under deep anesthesia using pentobarbital (50 mg/kg, ip) infused with ice-cold saline containing heparin, and the lumbar spinal dorsal horn was removed. The spinal dorsal horn was immediately homogenized in an ice-cold tissue protein extraction reagent. The samples were prepared as previously described.²⁰  The membranes were incubated with the following primary antibodies: mouse anti-Rheb (1:500; Santa Cruz Biotechnology, USA), mouse anti-actin (1:1,000; Cell Signaling Technology ), rabbit anti-p-mTOR-Ser²⁴⁴⁸ (1:1,000; Cell Signaling Technology), rabbit anti-mTOR (1:1,000; Cell Signaling Technology), rabbit anti-p-S6-Ser^235/236 (1:1,000; Cell Signaling Technology), rabbit anti-S6 (1:1,000; Cell Signaling Technology), rabbit anti-P-4E-BP1-Thr^37/46 (1:1,000; Cell Signaling Technology), rabbit anti-4E-BP1 (1:1,000; Cell Signaling Technology), rabbit anti-p-AMPKα1-Thr¹⁸³/ AMPKα2-Thr¹⁷²(1:1,000, Abcam, USA), and rabbit anti-AMPK (1:1,000; Abcam, USA) at 4°C overnight. The blots were washed in Tris-buffered saline with Tween 20 and incubated with secondary antibodies (1:5,000; Huaan Biotechnology, China). Signals were detected by Image Quant Ai600 (General Electric Co., USA) using an enhanced chemiluminescence reagent (Thermo Fisher Scientific, USA) and visualized with the ChemiDocXRS system (Bio-Rad, USA). The results were analyzed and quantified using ImageJ software (version 2.0.0; National Institutes of Health, USA).

The mice were transcardially perfused first with cold saline, followed by 4% cold paraformaldehyde under deep anesthesia. The lumbar spinal cord was harvested, postfixed with 4% paraformaldehyde for 4 h, and dehydrated overnight in sucrose at 4°C. Frozen sections (10 μm) were cut on microscope slides before further detection. Double staining was used to identify the colocalization of pS6 with Rheb; pS6 and Rheb with NeuN, CD11b, or Iba1; and GFAP and pS6 with GAD67 and VGLUT2 in the spinal cord. The slides were incubated overnight with primary antibodies (rabbit anti-pS6, 1:500, Cell Signaling Technology; mouse anti-Rheb, 1:50, Santa Cruz; rabbit anti-NeuN, 1:1,000, Huaan; mouse anti-NeuN, 1:1,000, Thermo Fisher; mouse anti-CD11b, 1:100, Abcam; rabbit anti-Iba1, 1:200, Huaan; mouse anti-GFAP, 1:200, Abcam; rabbit anti-GFAP, 1:200, Huaan; rabbit anti-GAD67, 1:200, Abcam; and rabbit anti-VGLUT2, 1:200, Abcam). On the second day, the slides were washed in phosphate-buffered saline and incubated with the following secondary antibodies for 2 h at 25 ± 1°C: goat anti-rabbit IgG H&L, Alexa Fluor 594, 1:500, Abcam; goat anti-mouse IgG H&L, Alexa Fluor 594, 1:500, Abcam; goat anti-mouse IgG H&L, Alexa Fluor 488, 1:500, Abcam; goat anti-mouse IgG H&L, and Alexa Fluor 488, 1:500, Abcam). Images were acquired using a fluorescence microscope (DM IL LED; Leica, USA).

Spinal cord slices were prepared as described previously.²¹  The mice (postnatal days 15 to 25) were anesthetized with pentobarbital (50 mg/kg, ip); the spinal cords were rapidly excised and placed in ice-cold cutting solution containing the following and oxygenated with 95% O₂ and 5% CO₂ (310 to 320 mOsm): 95 mM NaCl, 1.8 mM KCl, 1.2 mM KH₂PO₄, 7 mM MgSO₄, 0.5 mM CaCl₂, 26 mM NaHCO₃,50 mM sucrose, and 15 mM glucose (pH 7.4). Transverse slices (300 μm) were cut from the lumbar spinal cord using a vibratome (VT1200; Leica, Germany). The slices were incubated in artificial cerebrospinal fluid containing the following: 127 mM NaCl, 1.8 mM KCl, 1.3 mM MgSO₄, 1.2 mM KH₂PO₄,2.4 mM CaCl₂, 26 mM NaHCO₃, and 15 mM glucose, followed by bubbling with 95% O₂ and 5% CO₂ (310 to 320 mOsm) for 40 min at 34°C. The slices were transferred to a recording chamber and continuously perfused with oxygenated artificial cerebrospinal fluid at a rate of 3 ml/min (22 to 26°C).

Whole-cell patch-clamp recordings were acquired using an EPC-10 triple amplifier (HEKA, German), and the signals were filtered at 2.9 kHz and sampled at 15 kHz. The recording micropipettes were made from borosilicate glass capillaries (Sutter, USA) and had a resistance of 5 to 8 MΩ. The internal solution contained the following: 3 mM Na₂ATP, 125 mM potassium gluconate, 0.5 mM NaGTP, 2 mM CaCl₂, 2 mM MgCl₂, 10 mM HEPES, and 10 mM EGTA (pH 7.5). Neurons were randomly picked from the laminae II of the dorsal horn.

No statistical analyses were used to predetermine the sample size; however, our sample sizes were based on a previous report.²²  At least four animals were used for each recording protocol. Data on behavioral tests were converted to a percentage maximum possible effect calculated as follows: the percentage maximum possible effect = 100 × (postdrug latency threshold – predrug latency threshold)/(cut-off latency threshold – pre-drug latency threshold). The relative expression of proteins was normalized to that of β-actin in different groups, and the phosphorylation levels were compared with the total level of target proteins. The data are presented as means ± SD. Statistical analysis was performed using GraphPad Prism 9 (GraphPad Software Inc., USA). The investigators performing the behavioral, cell counting, protein quantitation, and electrophysiologic experiments were blinded to the treatment and genotypes. Outliers were not evaluated, and no data were excluded from statistical analyses. Before analysis, all data were tested using the Shapiro–Wilk normal distribution test. Parametric tests or nonparametric tests were used according to the results of the normal distribution tests. All statistical analyses were two-tailed. Unpaired Student’s t tests were used to compare differences between two groups. Multiple group comparisons were performed using one-way or two-way ANOVA with mix design or repeated measures followed by Bonferroni’s post-tests. Statistical significance was set at P < 0.05.

The animal experiments were in compliance with the guiding principles of the Care and Use of Animals and the Animal Management Rule of the Ministry of Public Health, People’s Republic of China (document No. 545, 2001), and were approved by the Animal Care and Use Committee of the Sixth People’s Hospital Affiliated with Shanghai Jiao Tong University (document No. SYXK-2016-0020). All the transgenic mice were provided by Dr. Paul Worley and transported from his lab to Dr. Xu’s lab in Shanghai by World Courier under suitable conditions: less than three males or five females per cage at a constant ambient temperature of 22 to 23°C. Food and water were provided ad libitum.

To explore the role of Rheb in acute antinociception and chronic morphine tolerance using 1- and 5-day treatments (fig. 1, A and C), we tested whether Rheb overexpression or conditional knockout affects morphine-induced acute antinociception or chronic tolerance. To overexpress Rheb, we utilized Rosa26-Rheb S16H transgenic mice (Rheb knock-in),¹³  the establishment of which was validated via PCR (Supplemental Content 1A, https://links.lww.com/ALN/D418). The Rheb knock-in mice exhibited a 3.47 ± 0.883-fold overexpression of spinal Rheb (Supplemental Content 1B, https://links.lww.com/ALN/D418; P = 0.001, n = 6) and significant phosphorylated levels of mTOR by 1.879 ± 0.392-fold (P = 0.0012), S6 by 1.553 ± 0.274-fold (P = 0.0011), and 4EBP1 by 2.445 ± 0.724-fold (P = 0.0009; Supplemental Content 1C, https://links.lww.com/ALN/D418; n = 6), indicating persistent activation of mTORC1 signaling. Regarding tail withdrawal latency from warm water, morphine-induced antinociception after intrathecal injection was impaired in Rheb knock-in mice (fig. 1B, left panel, tail-flick latency: 4.65 ± 1.10 s in Rheb knock-in mice vs. 10 ± 0 s in littermate control mice, means ± SD, P < 0.0001, n = 7), with an efficacy of 42.66% compared to the control group (fig. 1B, right panel, maximum possible effect: 100% in control mice vs. 42.66 ± 11.46% in Rheb knock-in mice, P < 0.0001, n = 7). During the revision of the article, as per editorial request to investigate sex as a biologic variable, we performed these experiments in female mice and obtained similar results (Supplemental Content 2A and 2B, https://links.lww.com/ALN/D418, tail-flick latency: 4.62 ± 0.73 s in Rheb knock-in mice vs. 9.05 ± 0.45 s in littermate control mice, P < 0.0001; maximum possible effect: 89.42 ± 4.93% in control mice vs. 4.98 ± 8.00% in Rheb knock-in mice, P < 0.0001, n = 6).

To investigate the effects of Rheb knockout on morphine-induced chronic tolerance, we crossed floxed Rheb mice with CaMKII-Cre transgenic mice to generate central nervous system–specific deletion of Rheb (Rheb cKO, Supplemental Content 1D, https://links.lww.com/ALN/D418). Immunoblotting analysis using total proteins showed that Rheb levels in the spinal cord (Supplemental Content 1E, https://links.lww.com/ALN/D418, Rheb level: 0.81 ± 0.27 in control mice vs. 0.15 ± 0.08 in Rheb cKO mice, P = 0.0001, n = 6) were significantly lower in Rheb cKO mice than in control mice, which indicated that Rheb was successfully deleted in Rheb cKO mice. We first determined whether the morphine tolerance model was successfully established in wild-type control and Rheb cKO mice. Thermal nociceptive thresholds were measured 1 h after each daily injection to evaluate tolerance. Consistent with our previous work,²³  we found that in wild-type mice, chronic morphine treatment produced significant antinociceptive tolerance (Supplemental Content 3A and 3B, https://links.lww.com/ALN/D418; maximum possible effect: 99.04 ± 1.66% on day 1 vs. 15.58 ± 5.79% on day 5, P < 0.0001, n = 6) with a fixed twice daily dose of 10 µg, which significantly upregulated the excitability of spinal neurons (fig. 2, A to G; saline + vehicle vs. morphine + vehicle). Remarkably, we observed that while morphine antinociception progressively diminished in littermate controls, morphine retained nearly full antinociceptive efficacy across all days in Rheb cKO mice (fig. 1D). Morphine-induced antinociception in Rheb cKO mice decreased gradually similar to that in the littermate control during the 5-day treatment course; notably, the Rheb cKO group retained a substantially higher sensitivity to morphine-induced antinociception than the control group (fig. 1, E and F, maximum possible effect on day 5: 16.13 ± 6.16% in control mice vs. 49.08 ± 6.32% Rheb cKO mice, P < 0.0001; change in maximum possible effect: −82.81 ± 5.62% in control mice vs. Rheb cKO mice −48.16 ± 9.13% Rheb cKO mice, P < 0.0001, n = 6). These results demonstrate that Rheb, through gain and loss of function, bidirectionally controls nociception and its adaptation to morphine, highlighting its critical role in spinal cord nociception processing. Similar results were observed in female mice (Supplemental Content 2, C–E, https://links.lww.com/ALN/D418; maximum possible effect on day 5: 23.51 ± 5.99% in control mice vs. 54.15 ± 7.96% in Rheb cKO mice, P < 0.0001; change in maximum possible effect: −71.44 ± 6.61% in control mice vs. −42.02 ± 9.68% in Rheb cKO mice, P = 0.0001, n = 6).

Because modulation of Rheb affected the analgesic effect of acute morphine and the development of morphine-induced tolerance, we examined the functional impact and Rheb expression in the spinal cord after the establishment of morphine tolerance. After 5 days of morphine administration, the mice were euthanized for immunohistochemistry analyses. The mRNA level of Rheb did not increase after chronic morphine treatment (fig. 3, A and B, Rheb mRNA level: 1.7 ± 0.30 in saline group vs. 1.348 ± 0.1052 in morphine group, P = 0.3011, n = 6); however, there was significant increase in the protein level by 4.27 ± 0.195-fold (fig. 3C, P = 0.0036, n = 6) and fluorescent immunoreactivity of Rheb (fig. 3D, Rheb positive cells percentage: 19.50 ± 0.4.18% in saline group vs. 63.67 ± 12.55% in morphine group, P < 0.0001, n = 6) in the spinal dorsal horn of mice that received morphine treatment.

To map out the neuronal population expressing increased levels of Rheb, we performed immunolabeling experiments with the neuronal marker NeuN, astrocyte marker GFAP, and microglia marker Iba1 (fig. 3E). Rheb was expressed predominantly in neurons and astrocytes, but not in microglia. Moreover, double-staining experiments of Rheb with vGLUT2 (fig. 3F) and CaMKIIα (fig. 3G) showed that Rheb was expressed in excitatory neurons in the spinal cord.

To determine the role of spinal Rheb in the development of morphine tolerance, we intrathecally injected Rheb shRNA AAV into vGLUT2 Cre mice to generate spinal cord–specific Rheb knockdown from the excitatory neurons to rule in or out potential contributions of other central excitatory neurons (fig. 3H). The efficiency of Rheb knockdown was confirmed by immunoblotting (fig. 3I, Rheb level: 0.76 ± 0.18 in control mice vs. 0.47 ± 0.23 in knockdown mice, P = 0.03, n = 6). The excitatory neuron-specific Rheb knockdown mice were resistant to morphine-induced chronic tolerance (fig. 3, J to L; maximum possible effect on day 5: 5.98 ± 2.97% in control mice vs. 18.94 ± 2.48% in Rheb knockdown mice, P = 0.0001; change in maximum possible effect: −90.11 ± 6.90% in control mice vs. −78.82 ± 3.82% in Rheb knockdown mice, P = 0.0049, n = 6). These results demonstrate that Rheb upregulation in excitatory neurons is causally linked to the development of morphine tolerance.

Because CaMKII-Cre-mediated manipulation is not restricted to the spinal cord, we further introduced CaMKII-Cre and VGAT-Cre viruses to the sacral spinal cord of Rheb flox/flox mice to specifically knock down Rheb in excitatory and inhibitory neurons, respectively, to assess which type of neuron is involved in mediating morphine tolerance. Figure 4A shows a schematic diagram of the experimental procedure and successful expression of the Cre virus in the spinal dorsal horn. Surprisingly, only Rheb f/f;CaMKII-Cre mice, but not Rheb f/f;VGAT-Cre mice, delayed the occurrence of morphine tolerance (fig. 4, B to D; maximum possible effect on day 5: 18.53 ± 8.99% in control mice vs. 57.78 ± 9.87% in Rheb f/f;CaMKII-Cre mice, P < 0.0001; change in maximum possible effect: −76.54 ± 9.54% in control mice vs. −38.84 ± 10.53% in Rheb f/f;CaMKII-Cre mice, P < 0.0001, n = 6). These data further indicate that Rheb, specifically in spinal excitatory neurons, plays an important role in morphine tolerance.

Located downstream of Rheb, the mTOR signaling pathway has been previously implicated in various functions, such as cell growth and proliferation and synaptic plasticity.^24–26  Inhibition of mTOR by rapamycin has resulted in autophagy upregulation, reduced neuroinflammation, a neuroprotective effect in multiple neurodegenerative disorders, and a potent immunosuppressant effect.^27,28 

Consistent with previous research,¹²  repeated morphine treatment induced mTORC1 activation (fig. 5A), as evidenced by significant phosphorylation of spinal mTOR (phosphorylation level: 1 ± 0.42 in saline group vs. 1.63 ± 0.29 in morphine group, P = 0.0133, n = 6), 4EBP1 (phosphorylation level: 1 ± 0.69 in saline group vs. 1.99 ± 0.59 in morphine group, P = 0.0235, n = 6), and S6 (phosphorylation level: 1 ± 0.33 in saline group vs. 3.11 ± 1.01 in morphine group, P = 0.0006, n = 6) after the establishment of morphine tolerance. Increased immunoreactivity of p-S6 was detected in the spinal dorsal horn of mice after chronic morphine treatment (fig. 5B); however, mTORC2 signaling remained unaffected (fig. 5E; phosphorylation level: 1 ± 0.13 in saline group vs. 1.02 ± 0.18 in morphine group, P = 0.808, n = 6). These data confirmed that chronic morphine-induced tolerance results in the spinal activation of mTORC1 signaling but not mTORC2.

We examined the localization of p-S6 in different neuronal cell types. Chronic morphine-induced p-S6 was mainly localized in the neurons but not in the astrocytes or microglia (fig. 5C). To identify the specific subset of neurons that are involved in chronic morphine-induced mTORC1 signaling, we investigated the portion of p-S6 expressed in excitatory and inhibitory neurons in the spinal dorsal horn. Double-immunofluorescence staining results showed a strong overlap of p-S6 with vGLUT2 (percentage, 56.21 ± 5.53%), a specific excitatory neuronal marker, and partial colocalization with GAD67 (percentage, 29.59 ± 8.96%), a specific inhibitory neuronal marker (fig. 5D; P = 0.0008, n = 6). These overlapping expression patterns indicate that chronic morphine-induced Rheb expression and the activation of mTORC1 signaling occurred in the same neurons.

We intrathecally injected rapamycin,²⁹  an inhibitor of mTOR activity, to observe its effect on the development of morphine-induced tolerance. During the development of morphine-induced tolerance, the reduction in morphine-induced antinociceptive effect was significantly mitigated by coadministration of rapamycin (fig. 6A), retaining approximately 50% of the maximum possible effect on day 5 in the morphine + rapamycin group compared to the morphine + vehicle group (fig. 6B; maximum possible effect on day 5: 52.60 ± 9.56% in morphine + rapamycin group vs. 16.60 ± 8.54% in morphine + vehicle group, P < 0.0001, n ≥ 4). The maximum possible effect on day 5 was reduced to approximately 80% in the morphine group compared to that on day 1, and coadministration of rapamycin partially prevented the decrease, resulting in 40% of morphine-induced maximum possible effect (fig. 6C; change in maximum possible effect: −41.91 ± 10.63% in morphine + rapamycin group vs. −79.98 ± 6.17% morphine + vehicle group, P < 0.0001, n ≥ 4). Moreover, coadministration of rapamycin significantly blocked the rightward shift of the dose–response curve caused by chronic morphine treatment (fig. 6D), with a lower EC₅₀ (3.617 µg) in the morphine + rapamycin group than in the vehicle group (9.145 µg) on day 6 (fig. 6E; P < 0.0001, n = 6). To further enhance the activity of spinal mTOR, we knocked down mTOR with intrathecal mTOR siRNA. In mice treated with mTOR siRNA, mTOR expression was specifically and selectively reduced compared with saline-treated controls (fig. 6, F and G; mTOR level: 0.98 ± 0.23 in saline + vehicle group vs. 0.34 ± 0.05 in saline + siRNA group, P < 0.0001; 1.18 ± 0.15 in morphine + siRNA scramble group vs. 0.34 ± 0.05 in morphine + siRNA group, P < 0.0001, n = 6). Similar to rapamycin, mTOR siRNA substantially prevented the morphine-induced rightward shift of the dose–response curve (fig. 6H), and the EC₅₀ in mTOR siRNA group (3.678 µg) was much lower than that in the scrambled mTOR siRNA group (7.814 µg). Vehicle and scrambled mTOR siRNA had no effect on spinal mTOR expression or morphine-induced tolerance. To investigate the neural excitability before and after rapamycin treatment, we performed whole-cell patch-clamp recordings in laminae II dorsal horn neurons (fig. 2A). Rapamycin effectively prevented morphine-induced upregulation of firing evoked by current injection (fig. 2, B to F; saline + vehicle vs. morphine + vehicle: P = 0.0461; morphine + vehicle vs. morphine + rapamycin: P = 0.0018) and resting membrane potential in the spinal dorsal horn neurons (fig. 2G; saline + vehicle vs. morphine + vehicle: 51.90 ± 3.51 vs. 55.58 ± 3.85, P = 0.0263; morphine + vehicle vs. morphine + rapamycin: 55.58 ± 3.85 vs. 52.21 ± 2.68, P = 0.0252). These results suggest that inhibition of spinal mTOR signaling can attenuate morphine-induced tolerance. Collectively, our results indicate that chronic morphine-induced tolerance was dependent on Rheb and activation of mTORC1 signaling in spinal excitatory neurons, implicating their roles in priming the developmental trajectories of morphine tolerance.

Our results thus far have demonstrated that Rheb upregulation is related to the development of morphine-induced tolerance through mTORC1 activation, prompting us to explore related upstream signaling mechanisms. Previous studies suggest that as a negative regulator of the mTORC1 pathway, AMPK may be a candidate for the induction of spinal Rheb expression after repeated intrathecal morphine treatment.

We used metformin, an activator of AMPK, and injected it intraperitoneally to observe its effect on the development of morphine-induced tolerance. During the development of morphine-induced tolerance, reduction in morphine-induced antinociceptive effect was significantly mitigated by coadministration of metformin (fig. 7, A and B), retaining approximately 40% of the maximum possible effect on day 5 in the morphine + metformin group (fig. 7C, maximum possible effect on day 5: 39.51 ± 7.40% in morphine + metformin group vs. 15.58 ± 5.79% in morphine group, P < 0.0001, n = 6). The maximum possible effect on day 5 was reduced to approximately 83% in the morphine group in comparison with day 1, and coadministration of metformin partially prevented the decrease, resulting in approximately 60% of morphine-induced maximum possible effect (fig. 7D; change in maximum possible effect: −59.73 ± 6.47% in morphine + metformin group vs. −83.46 ± 5.03% in morphine group, P < 0.0001, P < 0.0001, n = 6).

Additionally, we observed that phosphorylation of spinal AMPK was decreased significantly after repeated intrathecal morphine treatment (fig. 7E; phosphorylation level: 2.08 ± 0.29 in saline group vs. 1.17 ± 0.35 in morphine group, P = 0.0006, n = 6). Activation of AMPK with intraperitoneal metformin effectively prevented the chronic morphine-induced downregulation of spinal AMPK phosphorylation (fig. 7F; phosphorylation level: 0.07 ± 0.087 in morphine group vs. 0.76 ± 0.63 in morphine + metformin group, P = 0.0246, n = 6). Interestingly, metformin significantly inhibited the upregulation of chronic morphine-induced spinal Rheb (fig. 7G; Rheb level: 0.99 ± 0.70 in morphine group vs. 0.27 ± 0.21 in morphine + metformin group, P = 0.0145, n = 7). Furthermore, coadministration of metformin decreased the level of phosphorylation of spinal mTOR (fig. 7H; phosphorylation level: 1.07 ± 0.61 in morphine group vs. 0.29 ± 0.13 in morphine + metformin group, P = 0.005, n = 6), 4EBP1 (fig. 7I; phosphorylation level: 1.0 ± 0.18 in morphine group vs. 0.70 ± 0.20 in morphine + metformin group, P = 0.0204, n = 6), and S6 (fig. 7J; phosphorylation level: 1.0 ± 0.11 in morphine group vs. 0.47 ± 0.15 in morphine + metformin group, P < 0.0001, n = 6). Double-immunofluorescence staining results revealed that the localization of p-AMPK overlapped with that of neuronal markers NeuN (fig. 7L) and Rheb (fig. 7K), consistent with the expression patterns of Rheb and p-S6. These results demonstrate that Rheb is regulated by the endogenous level and/or activity of AMPK, possibly in a phosphorylation manner, thereby serving as an indispensable molecular switch for bidirectional regulation of morphine efficacy and tolerance.

This study showed that chronic morphine treatment decreased the phosphorylation of spinal AMPK and increased the expression of spinal Rheb, leading to the activation of spinal mTOR signaling, which underlies the development of morphine-induced tolerance. Elevation of AMPK levels and spinal Rheb knockout alleviated these effects. Conversely, overexpression of Rheb in the spinal cord had the opposite effects. Based on these results and those of our previous study,¹²  we propose that the small GTPase Rheb is involved in the AMPK–Rheb–mTOR signaling pathway and acts as a key endogenous switch axis for bidirectional regulation of the development and maintenance of morphine-induced tolerance in the spinal dorsal horn. To this end, targeting spinal Rheb may present a new therapeutic strategy for potentially preventing and even reversing chronic morphine-induced tolerance.

Previous studies have demonstrated that Rheb mRNA levels are significantly increased 2 h after carrageenan injection during hypersensitivity induction by peripheral inflammation³⁰  and are modulated in a morphine-induced conditioned manner.³¹  However, its potential roles in morphine tolerance remain unknown. We demonstrated that Rheb overexpression impaired intrathecal morphine-induced analgesia, as evidenced by the phenotype of Rheb S16H mice observed in this study and the opposite effects detected by its deletion in excitatory neurons in the spinal cord. Chronic morphine treatment induces antinociceptive tolerance³² ; hence, our finding on the profound increase in spinal Rheb after repeated morphine exposure indicates that Rheb plays a vital role in pain modulation and the development of chronic morphine-induced tolerance. Rheb mRNA levels did not change after chronic morphine treatment in this study. The increase in Rheb protein may be attributed to a reduction in degradation or an enhancement in protein translation. These findings provide unequivocal evidence for the causal link between endogenous Rheb-dependent signaling and morphine-induced analgesia.

As a negative regulator of the Rheb-mTOR signaling pathway, activated AMPK promotes the inhibition of Rheb, subsequently affecting mTORC1 activity. Therefore, AMPK is likely responsible for the increase in spinal Rheb during the development of morphine-induced tolerance. In contrast to the results of previous studies,^15,33,34  we showed that AMPK-Rheb signaling in a subset of spinal excitatory neurons likely accounts for the development of morphine-induced antinociceptive tolerance. Activation of AMPK suppresses neuroinflammation and ameliorates bone cancer pain³⁵  and other types of pathologic pain.³⁶  In addition to modulating pain transduction, AMPK-modulated Rheb expression has been reported in functional dyspepsia treatment at the protein and mRNA levels.³⁷  These observations suggest that AMPK is a potent negative regulator of Rheb expression. In our study, chronic morphine exposure induced the inhibition of spinal AMPK activity (as evidenced by its lowered phosphorylation levels), disinhibiting spinal Rheb expression, which in turn activated mTOR signaling. Metformin activated AMPK to counteract the increase in Rheb levels induced by morphine, thereby mitigating its tolerance. Previous studies have demonstrated the antinociceptive effect of metformin in rodent pain models; metformin relieved spinal nerve ligation–induced tactile allodynia in rats and mice by activating AMPK and inhibiting the mTORC1 pathway.^38,39  Additionally, metformin prevented tactile allodynia in other neuropathic pain models, such as those of spinal cord injury⁴⁰  and bortezomib-, paclitaxel-, and cisplatin-induced hyperalgesia^41,42  in rodents. Activation of AMPK reduced morphine tolerance by inhibiting microglial-mediated neuroinflammation.^15,33  These results demonstrate that AMPK is an important target in the regulation of pain by metformin, although it has many other AMPK-independent effectors.

mTORC1 signaling is known to control growth by balancing anabolic processes, such as protein, lipid, and nucleotide synthesis. Dysregulation of mTORC1 signaling leads to abnormalities in many diseases, including cancer, diabetes, neurodegeneration conditions, and epilepsy. Although rapamycin, a potent inhibitor of mTORC1, has demonstrated efficacy, it may also have unexpected off-target effects owing to the broad involvement of mTORC1 signaling in the synthesis of various proteins involved in many physiologic and pathologic processes. In this study, we demonstrated that Rheb regulates the levels of key protein substrates specifically associated with morphine-dependent tolerance. Decreased expression of Rheb in excitatory neurons may potentiate morphine-induced acute analgesia and reduce chronic morphine-induced antinociceptive tolerance. Conversely, overexpression of Rheb impaired the antinociceptive effect of acute morphine. These completely opposite effects provided a proof of concept for the rational use of Rheb as an ideal drug target, as it is uniquely situated immediately upstream of the mTORC1 signaling cascade, thus affecting a subset of specific key proteins critical for morphine-induced tolerance/hyperalgesia.

Morphine increases the release of excitatory peptides⁴³  and induces neuroplastic changes, which underlie spinal excitability reflected as thermal and tactile hypersensitivity to peripheral stimuli.⁹  Notably, Rheb regulates the expression of excitatory amino acid transporter 4 and limited extracellular glutamate levels,⁴⁴  whereas a loss of its downstream effector in excitatory neurons reduces evoked excitatory postsynaptic current amplitudes.⁴⁵  Opioid-induced plasticity has been reported in both acute and chronic morphine-induced analgesic tolerance.⁴⁶  The activation of AMPK signaling inhibited spinal synaptic plasticity, alleviating acute pain.⁴⁷  In persistent postsurgical pain, indirect AMPK activators prevent long-term neuronal plasticity.⁴⁸  Additionally, spinal Rheb-mTOR signaling regulates spinal sensitization and inhibition, because blocking spinal mTOR could attenuate inflammation-induced thermal and tactile hypersensitivity.³⁰  mTOR signaling in the spinal cord is required for neuronal plasticity and behavioral hypersensitivity associated with neuropathy, neuronal circuits of facilitated pain processing in inflammation-induced hyperalgesia,⁴⁹  and the development and maintenance of bone cancer–induced pain hypersensitivities.⁵⁰  Our experiments showing site- and cell-specific overexpression or knockout of Rheb in transgenic mice provided compelling evidence that Rheb is necessary for bridging AMPK and mTOR-dependent protein translation in spinal excitatory neurons to causally underpin the development of morphine tolerance, hyperalgesia, and possibly other sensory maladaptations. Nevertheless, the data from this study cannot absolutely rule out the supraspinal impact of Rheb on morphine tolerance, which is one of the limitations of this study.

Overall, we propose a new working model in which the AMPK-Rheb-mTOR signaling pathway in the dorsal horn excitatory neurons regulates morphine tolerance. After long-term opioid administration, chronic morphine decreases the phosphorylation of spinal AMPK, subsequently disinhibiting the expression of spinal Rheb. This is followed by activation of downstream mTOR signaling to stimulate S6K and 4E-BP activities, resulting in the initiation of mRNA translation and adaptive changes in protein translation in the dorsal horn.¹²  Therefore, Rheb signaling is a key regulator of the aberrant plasticity of nociceptive circuits and adaptation during chronic morphine exposure. In addition to opioid-induced tolerance and hyperalgesia, patients with neuropathic and inflammatory pain may benefit from Rheb inhibition. Thus, intrathecal use of mTOR inhibitors or metformin clinically may have additional benefits, such as antinociception and anti-tolerance, and serve as a potentially superior strategy in managing opioid-induced tolerance.

It is with the deepest gratitude that the authors thank Paul F. Worley, Ph.D., (Solomon H. Snyder Department of Neuroscience, Johns Hopkins University School of Medicine, Baltimore, Maryland) for his kindness and the invaluable transgenic mice. They also thank Editage for English language editing.

Supported by the General Program of the National Natural Science Foundation of China (Beijing, China) under grant Nos. 82171486 (to Dr. Xu) and 82171262 (to Dr. Jiang); the Shanghai Natural Science Foundation (Shanghai, China) under grant No. 21ZR1448400 (to Dr. Xu); the Key Project of Medical Engineering Intersection of Shanghai Jiaotong University (Shanghai, China) under grant No. YG2021ZD23 (to Dr. Xu); the Incubation Project of Shanghai Sixth People’s Hospital (Shanghai, China) under grant No. YNMS202114 (to Dr. Xu); the Shanghai Sailing Program (Shanghai, China) under grant No. 21YF1434200 (to Dr. Wang); and the Youth Program of the National Natural Science Foundation of China under grant Nos. 82001455 (to Dr. Du) and 82201366 (to Dr. Ma).

The authors declare no competing interests.

Supplementary Content 1. Validation of transgenic mice, https://links.lww.com/ALN/D418

Supplementary Content 2. Modulation of spinal Rheb expression in female mice, https://links.lww.com/ALN/D418

Supplementary Content 3. Validation of the morphine-induced tolerance model, https://links.lww.com/ALN/D418

Supplementary Content 4. Full-length Western blotting images, https://links.lww.com/ALN/D419