IGF-1 Microinjection in the Prefrontal Cortex Attenuates Fentanyl-Seeking Behavior in Mice

Abstract

Background

Opioid use disorder (OUD) is a chronic relapsing psychiatric disorder with an enormous socioeconomic burden. Opioid overdose deaths have reached an epidemic level, especially for fentanyl. One of the biggest challenges to treat OUD is the relapse to drug seeking after prolonged abstinence. Abnormalities in insulin-like growth factor-1 (IGF-1) have been reported in various neurological and psychiatric disorders, including OUD. However, whether IGF-1 and its downstream signaling pathways are associated with relapse to fentanyl seeking remains unclear.

Methods

Mice were subjected to daily 2-hour fentanyl (10 μg/mL, 27 μL/infusion) oral self-administration training for 14 days, followed by 14-day fentanyl cessation. Expression levels of IGF-1/IGF-1 receptor and downstream signaling pathways in the dorsomedial prefrontal cortex (dmPFC) were detected. Then, IGF-1 was bilaterally microinjected into the dmPFC from fentanyl cessation day 9 to day 13. Fentanyl-seeking behavior and excitatory synaptic transmission of pyramidal neurons in PFC were evaluated.

Results

We found that 14-day cessation from fentanyl oral self-administration caused significant downregulation of IGF-1 and IGF-1 receptor phosphorylation in the dmPFC. These changes were accompanied by inhibition of the downstream Akt and S6 signaling pathway. In addition, local administration of IGF-1 in the dmPFC attenuated context-induced fentanyl-seeking behavior. Furthermore, electrophysiology and immunohistochemistry analyses showed that IGF-1 blocked fentanyl-induced reduction of a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid and N-methyl-D-aspartate receptors-mediated excitatory synaptic transmission as well as synaptic expression of a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor and N-methyl-D-aspartate receptor subunits.

Conclusions

These results suggest that IGF-1 in the PFC plays a pivotal role in regulating fentanyl seeking after prolonged cessation from fentanyl oral self-administration.

Significance Statement.

Fentanyl-related overdose deaths have increased exponentially over the past few years. Yet, the neurobiology underlying fentanyl addiction is largely unknown. Here we used a mouse fentanyl oral self-administration (SA) model to examine the molecular mechanisms underlying fentanyl addiction, with a focus on the relapse to fentanyl seeking. This paper revealed that prolonged cessation (14 days) from fentanyl oral SA induces downregulation of IGF-1/IGF-1R and downstream Akt and S6 signaling activities in dorsomedial prefrontal cortex (dmPFC), one of the key brain regions regulating drug-seeking behavior. Intriguingly, IGF-1 infusion in the dmPFC during fentanyl cessation attenuates fentanyl seeking. Concomitantly, fentanyl cessation-induced disruption of excitatory synaptic transmission mediated by NMDA and AMPA receptors in PFC pyramidal neurons is normalized by IGF-1 infusion. In addition to bringing new insight into the understanding of fentanyl addiction, these findings may provide biological targets for developing novel therapies for opioid addiction in the future.

INTRODUCTION

Opioid use disorder (OUD) is characterized by compulsive drug seeking and use despite harmful consequences. The global prevalence of OUD dramatically increased from 1990 to 2017, reaching 40.5 million cases in 2017 (James et al., 2018; Roth et al., 2018). The number of opioid overdose deaths in the United States continued to increase to 68 630 cases in 2020 (NIDA, 2022). This trend is largely attributed to the fast-growing consumption of fentanyl and its derivatives (CDC, 2022a). Fentanyl is a potent synthetic opioid, which is highly addictive and has a high risk for overdose. Compared with other opioids, fentanyl has distinct pharmacological properties (e.g., 50–100 times more potent than morphine [Finch and DeKornfeld, 1967; Kliewer et al., 2019]; fast onset but short duration of action due to high lipophilicity [Pathan and Williams, 2012]), suggesting that nonidentical neurobiological mechanisms underlying fentanyl addiction may exist (Everett and Baracz, 2020). Therefore, it is important to investigate the neurobiological basis underlying relapse to fentanyl seeking.

The prefrontal cortex (PFC) is one of the key areas within the neurocircuitry involved in drug addiction and relapse (Shalev et al., 2002; Koob and Volkow, 2016). PFC controls high-order cognitive processes, including motivation, attention, memory, and decision-making (Carlén, 2017), which are disrupted in OUD patients (Goldstein and Volkow, 2011). Fentanyl and its derivatives induce unique biochemical and functional alterations in PFC (Ezeomah et al., 2020; Anderson et al., 2021). For example, a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR)-mediated excitatory synaptic transmission in pyramidal neurons from dorsomedial PFC (dmPFC) is disrupted after prolonged abstinence from remifentanil exposure (Anderson et al., 2021). However, the molecular mechanisms underlying fentanyl-induced PFC dysfunction are still unclear.

Insulin-like growth factor-1 (IGF-1) is a polypeptide hormone locally produced in neurons and glial cells of the brain or derived from the peripheral circulation (Rotwein et al., 1988). IGF-1 receptor (IGF-1R) is highly expressed in the neocortex, thalamus, and choroid plexus (Bondy and Lee, 1993). IGF-1 binds to IGF-1R and activates downstream phosphoinositide 3-kinase (PI3K)-AKT or rat sarcoma virus-mitogen-activated protein kinase pathways (Dobolyi and Lékó, 2019), which are all involved in the regulation of neuronal and synaptic function (Thomas and Huganir, 2004; Hoeffer and Klann, 2010). Therefore, IGF-1 and IGF-1R directly regulate neuronal excitability and synaptic transmission (Gazit et al., 2016; Pristerà et al., 2019; Fetterly et al., 2021). Both clinical (Reece, 2013; Valverde-Filho et al., 2015) and preclinical (Beitner-Johnson and Nestler, 1993; Brolin et al., 2018) studies reported changes in IGF-1 levels after chronic opioid exposure. However, the specific role of IGF-1 in fentanyl addiction remains vague.

This study aimed to examine changes in IGF-1/IGF-1R signaling pathways in dmPFC from mice that underwent prolonged cessation of fentanyl oral self-administration. We chose a fentanyl oral self-administration model because, first, it is clinically relevant. Most opioid abusers start drug use by orally taking prescription opioid analgesics (Kolodny et al., 2015; Gasior et al., 2016). Additionally, fentanyl is often mixed with other drugs and made into pills that can be taken orally (CDC, 2022b). Moreover, oral fentanyl misuse from various pharmaceutical forms is increasing (Nunez-Olarte and Alvarez-Jimenez, 2011; Passik et al., 2011). The absorption of oral fentanyl starts with rapid buccal mucosa absorption (25%), and the majority (75%) is absorbed via the gastrointestinal tract after swallowing (FDA, 2011). Although only 30% of gastrointestinal tract–absorbed fentanyl escapes first-pass metabolism (Darwish et al., 2007), oral fentanyl misuse can still induce increased blood fentanyl levels and cause death (Woodall et al., 2008). Second, the oral fentanyl self-administration model has been widely used to establish stable fentanyl-maintained behavior in preclinical studies (Shaham et al., 1993; Klein et al., 1997; Wade et al., 2008; Monroe and Radke, 2021). We also investigated the potential roles of IGF-1 in regulating fentanyl-seeking behavior and N-methyl-D-aspartate receptor (NMDAR)- and AMPAR-mediated excitatory synaptic transmission. Our results indicate that IGF-1 microinjection in the dmPFC could attenuate fentanyl seeking and restore fentanyl-induced inhibition of excitatory synaptic transmission in the dmPFC.

METHODS

Animals and Chemicals

C57BL/6J mice were obtained from Jackson Laboratory (Bar Harbor, ME, USA). Animals were group housed and bred in a temperature- and humidity-controlled animal care unit under a 12-hour-light/-dark cycle (lights on at 11:00 am and lights off at 11:00 pm) with food and water available ad libitum. Eight- to 10-week-old male and female mice were used in this study. All behavioral experiments were conducted during the light off period. All animal studies were approved by the Institutional Animal Care and Use Committee of the University of Kansas, which was accredited by the American Association for the Accreditation of Laboratory Animal Care. Fentanyl citrate was provided by the National Institute of Drug Abuse drug supply program and dissolved in water or saline for use. Recombinant human IGF-I was purchased from PeproTech Inc. (Cranbury, NJ, USA).

Oral Self-Administration

Oral self-administration was conducted in operant chambers (Med Associates, Fairfax, VT, USA) equipped with 2 nose-poke holes (1 active, 1 inactive) mounted with 2 stimulus lights in the same panel, a house light, and a metal rod floor (Gancarz et al., 2015; Singh et al., 2022). According to publications (Wade et al., 2008; Monroe and Radke, 2021), mice were subjected to daily 2-hour oral self-administration training, during which time each active nose-poke response resulted in the delivery of 27 μL water or fentanyl solution (10 μg/mL, a concentration inducing reliable operant responding that is distinguishable from non-drug cues under a fixed ratio 1 schedule; Wade et al., 2008) or sucrose solution (10%) on an fixed ratio 1 schedule. Reward delivery was done from the active nose-poke hole connected with a drug line and a syringe pump (0.735 mL/min using a 5-mL syringe) outside the box; therefore, the active nose pokes reflect both reward seeking and reward consumption. Each drug delivery was accompanied by a 5-second illumination of the stimulus light (discrete cue) inside the active nose-poke hole and a 5-second timeout period (no reward was available). Water restriction (approximately 12 hours) was performed prior to the self-administration session during the first 2–3 sessions to facilitate acquisition. All mice were trained to reach stable responses for fentanyl and sucrose: the active nose-poke responses varied below 10% over the last 3 days of training (Lai et al., 2013), and a minimum 10 active responses were made per session. Because there was no sex difference in fentanyl deliveries in the operant training (Monroe and Radke, 2021), male and female mice were combined for data analysis.

Drug-Seeking Test

Mice were subjected to cessation from self-administration for 14 days. This better represents the human scenario where relapse typically occurs after a drug-free abstinence period rather than extinction training (Fuchs et al., 2008). Then mice were placed back in the same chambers for a 1-hour context-induced seeking test (i.e., drug-seeking test), during which time the active responses produced discrete cues (light) previously paired with reward delivery (water, or fentanyl or sucrose) but without reward on-board.

Stereotaxic Surgery and IGF-1 Microinjection

Following oral self-administration, mice underwent 14-day cessation from self-administration, during which time bilateral guide cannulae were implanted. Surgery was performed according to previous publications (Wang et al., 2015, 2016). Mice were anesthetized with 100/5 mg/kg ketamine/xylazine and positioned in a stereotaxic frame (RWD Instruments, San Diego, CA, USA). A double guide cannula with a dummy (RWD Instruments) was implanted into the dmPFC (AP: +2.1, ML: ±0.3, DV: −1.7) and fixed with dental cement. A microinjector with a 0.5-mm protrusion below the guide cannula aiming at dmPFC (−2.2 mm) was used for infusion. Mice were counterbalanced according to self-administration performance during the last 4 sessions and assigned to receive vehicle (saline) or IGF-1 (100 ng/µL/hemisphere) infusion (0.5 µL/min using 100-μL Hamilton syringes) controlled by a syringe pump (RWD Instruments) during self-administration cessation day 9–13. After infusion, the needle was kept at the injection site for another 5 minutes to allow complete diffusion. Thirty-seven mice (20 males and 17 females) were included in a water- or fentanyl-seeking test; 13 mice (6 male and 7 female) were included in sucrose-seeking test (supplementary Figure 9).

Mechanical Pain Sensitivity

Twenty mice (10 males and 10 females) underwent water or fentanyl oral self-administration. Immediately after the last session, mice were subjected to mechanical pain sensitivity using a dynamic plantar aesthesiometer (Ugo Basile, Gemonio, Italy) (Rodriguez et al., 2021). A movable force actuator was placed below the plantar surface of mice. Paw withdrawal force was automatically recorded when an increasing force (up to 10 g) was exerted on the hind paw. Measurements of paw withdrawal force were performed 5 times with an interval of 5 minutes. An averaged value of 5 measurements represented the paw withdrawal force of each mouse.

Locomotor Activity

Immediately after the drug-seeking test, mice were taken out of the operant chamber and placed into a corner of a transparent plastic apparatus (length: 70 cm, width: 50 cm, height: 35 cm) to freely explore for 15 minutes. Locomotor activity was recorded via iSpy (Margaret River, Western Australia, Australia). The numbers of line crossings over the 15-minute period were measured to reflect locomotor activity.

Western-Blot Analysis

Twelve mice (6 males and 6 females) were killed (1%–3% isoflurane inhalation followed by decapitation) 14 days after the last water or fentanyl self-administration session; another 12 mice (6 males and 6 females) were killed 1 day after the last self-administration session. PFC tissue punches were harvested and homogenized with radioimmunoprecipitation assay lysis buffer with 1 mM phenylmethylsulfonyl fluoride and protease and phosphatase inhibitor cocktail (Roche, CA, USA). Proteins (30–50 µg) were loaded onto 7.5%–15% Tris-SDS polyacrylamide gels for electrophoresis separation and then transferred to nitrocellulose membranes (Millipore, Burlington, MA, USA). Membranes were blocked with 5% non-fat dry milk in tris-buffered saline with 0.1% Tween 20 detergent for 1 hour and incubated overnight at 4°C with primary antibodies against IGF-1 (1:500, Novus), p-IGF-1R beta (Tyr1161) (1:500, ThermoFisher), IGF-1R (1:1000, Cell Signaling Technology [CST]), p-Akt (Ser473) (1:1000, CST), Akt (1:1000, CST), p-ERK1/2 (Thr202/Tyr204) (1:1000, CST), ERK1/2 (1:1000, CST), p-S6 (Ser235/236) (1:1000, CST), S6 (1:1000, CST), and Tubulin (1:3000, Sigma-Aldrich). Then membranes were incubated with horseradish peroxidase–labeled goat anti-rabbit or goat anti-mouse secondary antibodies for 2 hours. Protein bands were visualized and analyzed using the ChemiDoc Imaging System and ImageLab software (Bio-Rad Laboratories, Hercules, CA, USA). Full blots for representative blots are provided in supplementary Figure 4.

Noncontingent Fentanyl Administration

Twelve mice (6 males and 6 females) received daily saline or fentanyl i.p. injections, and another 12 mice (6 males and 6 females) received daily water or fentanyl oral gavage once per day for 14 days. The daily increasing doses of i.p. or oral gavage fentanyl administration were equivalent to averaged daily fentanyl deliveries during oral self-administration (0.27–0.41 mg/kg; see supplemental Table 1 for details). Body weight was measured daily, and maximum dosing volume was 10 mL/kg. Fourteen days after the last administration, mice were killed (via 1%–%3% isoflurane followed by decapitation), and PFC tissues were collected for the western-blot experiment.

Immunofluorescence Staining

Twelve mice (6 males and 6 females) were subjected to water or fentanyl self-administration and received IGF-1 infusions during day 9–13 cessation from self-administration and then were perfused 2 hours after last IGF-1 infusion to capture IGF-1 treatment-induced transient changes in p-S6 signaling. Another 12 mice (6 males and 6 females) received similar self-administration training and IGF-1 treatment and then were perfused 24 hours after last IGF-1 infusion for the study of glutamate receptor A1 subunit-PSD95 and glutamate receptor N1 subunit-PSD95 co-staining. Immunohistochemistry and analysis were performed as previously described (Wang et al., 2017). Mice were anesthetized (100 mg/kg sodium pentobarbital i.p.) and transcardially perfused with 0.01 M phosphate-buffered saline followed by cold 4% paraformaldehyde. Brains were postfixed for 24 hours. Coronal sections of fixed brains encompassing dmPFC (bregma 2.8–1.8) were cut serially at 45 μm (using a vibratome [VT1000s, Leica, Wetzlar, Germany]) and then blocked in 5% normal donkey serum with 0.5% Triton-X for 1 hour, followed by incubation overnight at 4°C with primary antibodies against p-S6 (Ser235/236) (1:100, CST), GluN1 (1:50, ThermoFisher), GluA1 (1:50, ThermoFisher), PSD95 (1:100, CST), and NeuN (1:100, Sigma-Aldrich). Then sections were incubated with Alexa Fluor 488-conjugated donkey anti-rabbit or Alexa Fluor 555-conjugated goat anti-mouse secondary antibodies (1:400, ThermoFisher) for 2 hours. Sections were mounted on slides with 4ʹ,6-diamidino-2-phenylindole (DAPI)-containing mounting media (Vector Lab, Newark, CA, USA). Images were acquired under identical conditions using a Nikon Ti2 fluorescent microscope or Leica confocal microscopy and analyzed using ImageJ.

Electrophysiology

Fourteen mice (8 males and 6 females) were subjected to oral fentanyl self-administration training and received IGF-1 treatment during cessation day 9–13. Then mice were killed (1%–3% isoflurane inhalation followed by decapitation) 24 hours after last IGF-1 infusion, and brains were immediately collected for electrophysiology study. The whole-cell patch-clamp recording was used to measure synaptic currents in layer V dmPFC pyramidal neurons as previously described (Wang et al., 2020, 2021, 2022; Yue et al., 2022). Coronal slices (300 μm) were submerged in continuously flowing oxygenated (95% O₂ and 5% CO₂) artificial cerebrospinal fluid (in nM: 10 glucose, 130 NaCl, 26 NaHCO₃, 1.25 NaH₂PO₄, 3 KCl, 1 CaCl₂, 5 MgCl₂, pH 7.60, 300 mOsm). Cells were visualized with a 40× water immersed lens with an upright microscope (Scientifica, East Sussex, UK). Patch pipettes (resistance at 3–6 MΩ) were filled with internal solution (in nM: 4 NaCl, 1 MgCl₂, 130 Cs-methanesulfonate, 10 CsCl, 5 ethylene glycol tetraacetic acid, 2 QX-314, 12 phosphocreatine, 5 MgAdenosine triphosphate, 0.2 Na₂guanosine triphosphate, 10 N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid, pH 7.2–7.3, 265–270 mOsm). Evoked excitatory postsynaptic currents (EPSC) were generated with a pulse from a stimulation isolation unit controlled by a pulse generator (Digitimer, Fort Lauderdale, FL, USA) and delivered via a bipolar stimulating electrode (FHC Inc., Bowdoin, ME, USA) placed approximately 100 μm from the recorded neurons. A Multiclamp 700B amplifier and Digidata 1550B were used for recordings (Molecular Devices, San Jose, CA, USA). Bicuculline (20 μM) was added to isolate the excitatory transmission. Membrane potential was maintained at −70 mV for AMPAR-EPSC recordings. For NMDAR-EPSC, 6,7-Dinitroquinoxaline-2,3-dione (20 μM) was applied, and the cell (clamped at −70 mV) was depolarized to +40 mV for 3 seconds before stimulation to fully relieve the voltage-dependent Mg²⁺ block. For input-output responses, evoked EPSCs were elicited by a series of pulses with different stimulation intensities delivered at 0.05 Hz. Patched cells were discarded if the series resistance changed by more than 10%.

Statistical Analysis

An unpaired 2-tailed Student’s t test was used to compare differences between 2 groups. Experiments with more than 2 dependent variables were subjected to multi-factor ANOVA followed by Bonferroni correction for multiple post-hoc comparisons. Pearson correlation coefficient was calculated to analyze colocalization between GluN1 or GluA1 and PSD95 using the JACoP plugin for ImageJ software. The data were expressed as mean ± SEM. P < .05 was statistically significant. Statistical analyses were performed using SPSS software (SPSS, Chicago, IL, USA) and GraphPad Prism software (GraphPad, San Diego, CA, USA). Statistical details are provided in supplementary Table 2.

RESULTS

Acquisition of Fentanyl Oral Self-Administration

Based on previous publications (Wade et al., 2008; Monroe and Radke, 2021), we first validated a protocol for fentanyl oral self-administration in mice (Figure 1A). Multi-factor repeated-measures ANOVA analysis showed significant main effects for drug (water or fentanyl) in drug deliveries (Figure 1B; F_1,18(drug) = 12.54, P < .01) and total active responses (Figure 1C; F_1,18(drug) = 13.28, P < .01). We also found significant main effects for session in drug deliveries (Figure 1B; F_{13,234(session)} = 1.897, P < .05) and total active responses (Figure 1C; F_{13,234(session)} = 1.794, P < .05). Additionally, there was a significant main effect for interactions in drug deliveries (Figure 1B; F_{13,234(drug × session)} = 3.522, P < .001) and total active responses (Figure 1C; F_{13,234(drug × session)} = 3.598, P < .001). Furthermore, post-hoc Bonferroni analysis showed that fentanyl deliveries were significantly higher than water deliveries during sessions 11–14 (Figure 1B). Additionally, the total active responses in the fentanyl group were significantly higher than in the water group in sessions 10–14 (Figure 1C). These results indicate an increased acquisition of operant responding to oral fentanyl.

Figure 1.

Acquisition of fentanyl oral self-administration. (A) Timeline for the behavioral test. Mean numbers of drug deliveries (B), total active responses (C), and total inactive responses (D) per session during fentanyl oral self-administration training. (E) Assessment of mechanical pain sensitivity immediately after the last session. Data are expressed as mean ± SEM, n = 10 mice/group. ^*P < .05, ^**P < .01, ^***P < .001. FEN, fentanyl; H2O, water.

To verify the oral consumption of fentanyl, mechanical pain sensitivity was measured immediately after the last session in a different group of mice. Similarly, this group of mice showed increased acquisition of operant responding to oral fentanyl (supplementary Figure 1). An unpaired Student’s t test showed that the fentanyl group had a greater paw withdrawal force than the water group (Figure 1E;, t = 2.317, P < .05, t test), suggesting a valid analgesic effect of fentanyl after oral consumption.

Context-Induced Drug Seeking

Next, we sought to determine whether fentanyl oral self-administration leads to drug-seeking behavior. Mice that underwent water or fentanyl self-administration were subjected to drug cessation for 14 days. Then a 1-hour drug-seeking test was performed (refer to timeline in Figure 1A). Compared with the water group, the fentanyl group had significantly higher total active responses (t = 4.532, P < .001; Figure 2A) and active responses with (supplementary Figure 2A) and without (supplementary Figure 2B) contingency in place. No significant difference was observed between groups in terms of total inactive responses (Figure 2B), inactive responses (with and without contingency on board [supplementary Figure 2C–D]), and locomotion activity (Figure 2C), suggesting increased active responses are not due to changes in overall nose-poking behavior or locomotion. Together, these results suggest that prior experience with fentanyl oral self-administration can lead to fentanyl seeking.

Figure 2.

Context-induced drug-seeking test after 14-day cessation of self-administration. Mean numbers of total active responses (A) and total inactive responses (B) during fentanyl-seeking test. (C) Mean numbers of line crossing during the locomotion test. Data are expressed as mean ± SEM, n = 10 mice/group. ^***P < .001. FEN, fentanyl; H₂O, water; NS, not significant.

IGF-1/Akt/S6 Signaling Pathway Was Downregulated After Cessation of Fentanyl Oral Self-Administration

Next, we sought to examine the changes of IGF-1/IGF-1R signaling pathways in dmPFC after fentanyl cessation. Similarly, mice were subjected to water or fentanyl self-administration (supplementary Figure 3) and then sacrificed 14 days after last self-administration session. We found that the protein levels of IGF-1 (t = 2.384, P < .05) and p-IGF-1R (t = 2.301, P < .05) in dmPFC were significantly reduced after 14-day cessation of fentanyl self-administration (Figure 3; t test). Total IGF-1R levels were not altered. Furthermore, we found no changes in IGF-1, p-IGF-1R, and IGF-1R expression levels in dmPFC after 1-day cessation of fentanyl self-administration (supplementary Figure 5). Additionally, 14-day cessation of noncontingent fentanyl (via daily i.p. injection or oral gavage) did not alter IGF-1, p-IGF-1R, and IGF-1R expression levels in dmPFC (supplementary Figure 6). Collectively, these data suggest that IGF-1/IGF-1R signaling changes in dmPFC are related to the prolonged cessation of fentanyl reinforcement.

Figure 3.

IGF-1/IGF-1R/Akt/S6 signaling pathway was suppressed after fentanyl cessation. Representative immunoblots (A) and quantification (B) of IGF-1, p-IGF-1R, IGF-1R, p-Akt, Akt, p-S6, S6, p-ERK, and ERK protein levels in the dmPFC from mice underwent water or fentanyl self-administration followed by 14 days of no access to drug. (C) Schematic overview of IGF-1/IGF-1R system and the downstream Akt/S6 and MEK/ERK signaling pathways. Data are expressed as mean ± SEM, n = 6 mice/group. ^*P < .05, ^**P < .01. Akt, Protein kinase B; ERK, extracellular signal-regulated kinase; FEN, fentanyl; H₂O, water; IGF-1, insulin-like growth factor 1; IGF-1R, insulin-like growth factor 1 receptor, S6, ribosomal protein S6.

Then we examined the changes in 2 major IGF-1/IGF-1R downstream cascades: Akt/S6 and MEK/ERK signaling pathways. As shown in Figure 3, 14-day cessation of fentanyl self-administration induced a significant decrease in the protein levels of p-Akt (t = 2.394, P < .05, t test) and p-S6 (t = 4.145, P < .01, t test), a substrate for p70S6K. Moreover, p-ERK protein levels did not change. These results indicate that fentanyl cessation inhibits downstream Akt/S6 but not ERK signaling pathway in dmPFC.

IGF-1 Microinjection in dmPFC Attenuated Fentanyl Seeking and Blocked the Changes of Akt/S6 Signaling Pathway

To find out the potential role of IGF-1 in fentanyl-seeking behavior, mice received a 5-day IGF-1 microinjection in dmPFC during days 9–13 of cessation from water or fentanyl oral self-administration (supplementary Figure 7). A drug-seeking test was then performed 24 hours after the last IGF-1 infusion (Figure 4A). Multi-factor ANOVA analysis showed a significant main effect for treatment (vehicle or IGF-1) in total active responses (Figure 4B; F_{1,33(treatment)} = 11.51, P < .01). Additionally, post-hoc Bonferroni analysis revealed that mice self-administered fentanyl and received vehicle microinjection in dmPFC (FEN+VEH group) showed significantly higher total active responses compared with mice that self-administered water and received vehicle treatment (H₂O + VEH group) (Figure 4B). Moreover, IGF-1 microinjection in dmPFC attenuated total active responses in mice that self-administered fentanyl (Figure 4B). Similarly, compared to the H₂O+VEH group, the active responses with (supplementary Figure 8A) and without (supplementary Figure 8B) contingency in place were higher in the Fen+VEH group, which were attenuated by IGF-1 treatment. Interestingly, IGF-1 treatment did not change the total active responses in mice that underwent water self-administration (Figure 4B). Moreover, there were no significant differences in total inactive responses (Figure 4C), inactive responses (with and without contingency on board [supplementary Figure 8C–D]), and locomotion activity among all 4 groups (Figure 4D). These results suggest that IGF-1 attenuates context-induced fentanyl-seeking behaviors without affecting general behavioral responses (i.e., nose poke) and baseline locomotion activity. Intriguingly, we found that IGF-1 infusions in dmPFC during days 9–13 after the last session of sucrose oral self-administration (supplementary Figure 9A–D) failed to alter total active and inactive responses during the sucrose-seeking test (supplementary Figure 9E–G). These data suggest that IGF-1 specifically attenuates fentanyl seeking.

Figure 4.

IGF-1 microinjection in the dmPFC attenuated fentanyl seeking. (A) Left: timeline for vehicle or IGF-1 infusion (100 ng/µL/hemisphere, once per day for 5 days) and drug-seeking test. Right: a representative image of guide cannula and injector position. Mean numbers of total active responses (B) and total inactive responses (C) during fentanyl-seeking test with or without repeated IGF-1 infusions during drug cessation. (D) Mean numbers of line crossing during the locomotion test. Data are expressed as mean ± SEM, n = 7–12 mice/group. ^***P < .001 FEN + VEH group vs H₂O + VEH group. ^###P < .001, FEN + IGF-1 group vs FEN + VEH group. AI, agranular insular cortex; DP, dorsal peduncular cortex; FEN, fentanyl; H₂O, water; IGF-1, insulin-like growth factor 1; M1, primary motor cortex; M2, secondary motor cortex; MO, medial orbital cortex; PrL, prelimbic cortex, which is equivalent to dmPFC; VEH, vehicle.

To validate the IGF-1 treatment effect, we measured the level of p-S6 (2 hours after the last IGF-1 infusion [Figure 5A; supplementary Figure 10]), which reflects the activity of the IGF-1/Akt/S6 signaling pathway. Multi-factor ANOVA analysis revealed significant main effects for drug (F_1,48(drug) = 66.04, P < .001), treatment (F_{1,48(treatment)} = 5.234, P < .05), and interaction (F_{1,48(drug × treatment)} = 6.306, P < .05) in the number of p-S6–positive neurons (Figure 5B–C). Post-hoc analysis showed that the number of p-S6–positive neurons was reduced after the 14-day cessation of fentanyl oral self-administration, and IGF-1 microinjection in dmPFC brought it back to normal level.

Figure 5.

IGF-1 microinjection in the dmPFC blocked the reduced activity of signaling molecule S6. (A) Schematic of the experimental procedure. Representative images (B) and mean numbers (C) of p-S6 positive neurons in the dmPFC from mice underwent water or fentanyl self-administration and received vehicle or IGF-1 infusion (100 ng/µL/hemisphere, once per day for 5 days) during cessation. Data are expressed as mean ± SEM, n = 9–15 slices/3 mice/group. ^***P < .001, FEN + VEH group vs H₂O + VEH group. ^##P < .01, FEN + IGF-1 group vs FEN + VEH group. DAPI, 4',6-diamidino-2-phenylindole; FEN, fentanyl; H₂O, water; IGF-1, insulin-like growth factor 1; NeuN, neuronal nuclei protein; PFC, prefrontal cortex ; S6, ribosomal protein S6; VEH, vehicle.

IGF-1 Microinjection in dmPFC Blocked the Changes in Excitatory Synaptic Transmission of PFC Pyramidal Neurons

Next, we investigated the potential neuronal mechanisms underlying the IGF-1–mediated attenuation of fentanyl seeking with a focus on AMPAR- and NMDAR-mediated EPSC in dmPFC pyramidal neurons. Experimental paradigm was shown in Figure 6A. First, we recorded AMPAR-EPSC in deep-layer dmPFC pyramidal neurons, which mainly send projections to other subcortical regions, including reward pathways (Carr and Sesack, 2000). Multi-factor repeated-measures ANOVA revealed significant main effects for drug (F_3,46(drug)=6.115, P < .01), stimulation intensity (F_{4,184(stimulation intensity)} = 111.3, P < .001), and interaction (F_{12,184(drug × stimulation intensity)} = 2.757, P < .01) in AMPAR-EPSC amplitudes (Figure 6B–C). Post-hoc Bonferroni analysis showed that fentanyl cessation resulted in significant decreases of AMPAR-EPSC amplitudes at multiple stimulation intensities, which were blocked by IGF-1 microinjection in dmPFC. Moreover, IGF-1 treatment did not induce any changes in AMPAR-EPSC amplitudes in pyramidal neurons from mice that underwent water self-administration.

Figure 6.

IGF-1 microinjection in the dmPFC blocked the reduction of AMPAR synaptic function in dmPFC pyramidal neurons. (A) Schematic of the experimental procedure. (B) Input-output curves of AMPAR-EPSC in PFC pyramidal neurons from mice underwent water or fentanyl self-administration and received vehicle or IGF-1 infusion (100 ng/µL/hemisphere, once per day for 5 days) during cessation (n = 11–15 cells/3–4 mice/group). (C) Representative traces for AMPAR-EPSC. (D) Representative images for imaging area. (E) Representative images of GluA1-PSD95 puncta in the dmPFC from mice underwent water or fentanyl self-administration and received vehicle or IGF-1 infusion (100 ng/µL/hemisphere, once per day for 5 days) during cessation (n = 13–6 slices/3 mice/group). (F) Quantification of Pearson correlation coefficient between GluA1 and PSD-95 colocalization. Data are expressed as mean ± SEM. ^*P < .05, ^**P < .01, FEN + VEH group vs H₂O + VEH group. ^##P < .01, ^###P < .001, FEN + IGF-1 group vs FEN + VEH group. AMPAR, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor; dmPFC, dorsomedial prefrontal cortex; EPSC, excitatory postsynaptic current; FEN, fentanyl; GluA1, glutamate receptor A1; H₂O, water; IGF-1, insulin-like growth factor 1; PSD95, postsynaptic density protein 95; VEH, vehicle.

To further examine whether changes in AMPAR-mediated synaptic transmission were related to synaptic AMPAR expression, we performed co-staining of GluA1 with excitatory synaptic marker postsynaptic density 95 (PSD95). A significant main effect for IGF-1 treatment in Pearson coefficient between GluA1 and PSD95 colocalization was observed (Figure 6D–F; multi-factor ANOVA, F_{1,54(treatment)} = 6.607, P < .05). Post-hoc Bonferroni analysis showed that fentanyl cessation significantly reduced the Pearson coefficient between GluA1 and PSD95 colocalization in dmPFC. Moreover, IGF-1 treatment blocked the fentanyl-induced inhibition of GluA1 and PSD95 colocalization (Figure 6F). These results, together with electrophysiological data, suggest that IGF-1 microinjection in dmPFC blocks the reduction of fentanyl cessation–induced AMPAR-mediated synaptic function.

Next, we examined changes in NMDAR-mediated excitatory synaptic transmission. We found that there were main effects for drug (F_3,42(drug)=14.86, P < .001), stimulation intensity (F_{4,168(stimulation intensity)} = 107.9, P < .001), and interaction (F_{12,168(drug × stimulation intensity)} = 4.341, P < .001) in the amplitudes of NMDAR-EPSC (Figure 7A–B; multi-factor repeated-measures ANOVA). Post-hoc Bonferroni analysis showed that the FEN+VEH group had lower NMDAR-EPSC amplitudes at multiple stimulation intensities compared with the H₂O + VEH group. Furthermore, IGF-1 treatment blocked the fentanyl-induced reduction of NMDAR-EPSC amplitudes at the stimulation intensity of 30 μA, 40 μA, and 50 μA. Moreover, IGF-1 treatment did not alter NMDAR-EPSC amplitudes in pyramidal neurons from mice self-administered water.

Figure 7.

IGF-1 microinjection in the dmPFC prevented the decrease of NMDAR synaptic function in dmPFC pyramidal neurons. (A) Input-output curves of NMDAR-EPSC in PFC pyramidal neurons from mice underwent water or fentanyl self-administration and received vehicle or IGF-1 infusion (100 ng/µL/hemisphere, once per day for 5 days) during cessation (n = 11–12 cells/3–4 mice/group). (B) Representative traces for NMDAR-EPSC. (C) Representative images of GluN1-PSD95 puncta in the dmPFC from mice underwent water or fentanyl self-administration and received vehicle or IGF-1 infusion (100 ng/µL/hemisphere, once per day for 5 days) during cessation (n = 11–12 slices/3 mice/group). (D) Quantification of Pearson correlation coefficient between GluN1 and PSD-95 colocalization. Data are expressed as mean ± SEM. ^**P < .01, ^***P < .001, FEN + VEH group vs H₂O + VEH group. ^#P < .05, ^##P < .01, ^###P < .001, FEN + IGF-1 group vs FEN + VEH group. dmPFC, dorsomedial prefrontal cortex; EPSC, excitatory postsynaptic current; FEN, fentanyl; GluN1, glutamate receptor N1; H₂O, water; IGF-1, insulin-like growth factor 1; NMDAR, N-methyl-D-aspartate receptor; PSD95, postsynaptic density protein 95; VEH, vehicle.

Next, we performed co-staining of GluN1 and PSD95 to directly visualize changes in synaptic NMDARs after fentanyl cessation and IGF-1 treatment. As shown in Figure 7C–D, there was a significant main effect for IGF-1 treatment in the Pearson coefficient of GluN1 and PSD95 colocalization (multi-factor ANOVA, F_{1,43(treatment)} = 4.348, P < .05). Post-hoc Bonferroni analysis revealed that the GluN1 and PSD95 colocalization level in dmPFC was significantly decreased after fentanyl cessation and that IGF-1 treatment recovered GluN1 and PSD95 colocalization level (Figure 7D). These data, together with electrophysiological results, suggest that fentanyl cessation diminishes synaptic NMDAR function in dmPFC, which can be prevented by IGF-1 microinjection in dmPFC.

DISCUSSION

Use of Fentanyl Oral Self-Administration Model

In the current study, we first validated that mice could reliably acquire fentanyl through oral self-administration (Figure 1B–D). Moreover, the increased paw withdrawal force during the mechanical pain sensitivity test further confirmed the consumption of fentanyl during self-administration (Figure 1E). Furthermore, after 14 days of no access to contingent fentanyl, mice showed context-induced fentanyl-seeking behavior (Figure 2). All these data highlight the feasibility of using the fentanyl oral self-administration model to study the neurobiology of fentanyl addiction. However, the model we used has limitations, for example, uncontrolled drug intake (not all fentanyl was consumed) and longer duration of drug absorption may result in different pharmacodynamics and pharmacokinetics compared to inhaled and i.v. infused drugs (Dershwitz et al., 2000). Additionally, active responses represent both reward seeking and reward consumption as fentanyl was dispensed directly in the active nose-poke hole. Concomitantly, the active responses during timeout may reflect compulsive-like behavior (Anderson et al., 2018) or simply fentanyl consumption. Future studies using the design of fentanyl delivery in a port separated from nose-poke holes may tease apart these two effects.

Opioid-Induced Changes in IGF-1/Akt/S6 Signaling Pathway

We found decreased IGF-1 and p-IGF-1R levels in mouse dmPFC after 14-day (Figure 3) but not 1-day (supplementary Figure 5) cessation of contingent fentanyl nor after cessation of noncontingent fentanyl (supplementary Figure 6), suggesting this pathway is involved in fentanyl reinforcement that may be related to addiction. In fact, IGF-1/IGF-1R signaling has been implicated in opioid addiction. For example, IGF-1 levels were decreased in rat ventral tegmental area (VTA) and plasma after 1-day morphine withdrawal (75 mg/d, 5 days, s.c.) (Beitner-Johnson and Nestler, 1993). Additionally, IGF-1 levels showed a reducing trend in rat frontal cortex after 27-day morphine exposure (s.c. mini-osmotic pump, 17.5 mg/kg/d) (Brolin et al., 2018). However, a clinical study reported that blood IGF-1 levels were increased in OUD patients (Reece, 2013). Another study showed that IGF-1 and p-IGF-1R levels were increased in rat frontal cortex, hippocampus, and midbrain after 40-h morphine withdrawal (10–100 mg/kg increasing dose over 6 days, i.p.) (Peregud et al., 2016). Together, these results suggest that opioid-induced changes in IGF-1 levels may vary in different species and organs (peripheral, central, or different brain areas) and may depend on the specific paradigm of opioid administration.

IGF-1 binding to IGF-1 receptor leads to autophosphorylation of IGF-1R. This will induce the phosphorylation of juxtamembrane tyrosines and carboxyl-terminal serines to promote the recruitment of adaptor proteins (Chitnis et al., 2008), such as insulin receptor substrates 1–4 (activating the PI3K/Akt signaling pathway and further disinhibiting mTORC1/p70S6K/S6 cascades; Shepherd, 2005), and Src homology and collagen domain proteins A–C (activating rat sarcoma virus/mitogen-activated protein kinase signaling branches; Chong et al., 2003; Chitnis et al., 2008). Here, we found that IGF-1 downregulation further led to inhibition of Akt/S6 signaling after fentanyl cessation. Indeed, PI3K/Akt/mTOR/p70S6K/S6 signaling in different brain regions was involved in drug addiction. For example, chronic morphine administration inhibited the IR substrate 2-Akt signaling pathway in VTA (Russo et al., 2007; Mazei-Robison et al., 2011), while others reported increased Akt phosphorylation in nucleus accumbens after heroin cessation and downregulation of Akt activity attenuated heroin-seeking behavior (Zhu et al., 2021). Of note, Mazei-Robison et al. found that morphine increased mTORC1/p70S6K/S6 activity but decreased mTORC2 activity in VTA (Mazei-Robison et al., 2011). Moreover, different times of cessation from opioid exposure may change the activity of the IGF-1R singling pathway. For example, heroin self-administration induced elevation of IGF-1R/mTORC2 signaling in basolateral amygdala, which returned to normal after 30-day abstinence (Ucha et al., 2022). All these studies together with our data suggest that the IGF-1/PI3K/Akt/mTOR/p70S6K/S6 signaling pathway is related to opioid addiction, and it may play distinct roles in different brain regions.

Opioid-Induced Changes in Excitatory Synaptic Transmission

Drug-induced changes in excitatory synaptic transmission and/or synaptic plasticity in the mesocorticolimbic system play an important role in the neuronal adaptations underlying the long-lasting behavioral plasticity induced by drugs, especially for psychostimulants such as cocaine (Wolf, 1998; Kalivas, 2009; Wolf and Ferrario, 2010). Similarly, opioids also alter excitatory synaptic function in multiple brain regions. For example, after extinction training following heroin self-administration, long-term potentiation and long-term depression are blunted in nucleus accumbens core excitatory synapses projected from prelimbic cortex (Shen and Kalivas, 2013). Yet, studies examining opioid-evoked excitatory synaptic function changes in PFC neurons are sparse. One study showed that reexposure to heroin-associated cues induced GluA2 endocytosis in ventromedial PFC, which was parallel with a decreased AMPAR/NMDAR ratio (Van den Oever et al., 2008). Another study reported that prolonged cessation from remifentanil self-administration resulted in hypoactive states in dmPFC neurons, which was driven by a reduction in AMPAR-mediated excitatory synaptic transmission in female mice (Anderson et al., 2021). Here we further showed that both AMPAR- and NMDAR-mediated synaptic transmission in dmPFC pyramidal neurons was inhibited after a 14-day cessation from fentanyl oral self-administration (Figures 6 and 7).

More studies using biochemical approaches reveal altered expression levels of glutamate receptor subunits after opioid exposure. The surface expression of GluA1 and/or GluA2 in medial PFC was downregulated, and spontaneous firing of neurons was blunted after acute morphine administration (Giacchino and Henriksen, 1996; Mickiewicz and Napier, 2011; Herrold et al., 2013). Furthermore, GluN1 and GluN2B subunit expression was increased in postmortem PFC tissue from opioid users (Daneshparvar et al., 2019). In another preclinical study, GluN2B subunit expression in rat medial PFC was increased after 7-day heroin injection but returned to normal after 7-day heroin cessation (Zhu et al., 2020). In contrast, we found that expression levels of GluA1 and GluN1 on excitatory synapses (colocalized with PSD95) were reduced after fentanyl cessation (Figures 6 and 7). The discrepancies among these studies may result from multiple possibilities such as exposure to various types of opioid and distinct periods of cessation. Moreover, one main difference is that our study focused on GluA1 and GluN1 expression levels on excitatory synapses from dmPFC pyramidal neurons, whereas other studies were using bulk tissue. This suggests that beyond glutamate receptor subunit gene transcription, other processes regulating excitatory synaptic transmission may also play key roles in opioid-induced excitatory synaptic dysfunction (e.g., local mRNA translation at the synapse and glutamate receptor trafficking).

Intriguingly, we found that IGF-1 treatment prevented reduction of excitatory synaptic transmission in dmPFC neurons and attenuated fentanyl seeking (Figure 4). Excitatory synaptic transmission changes in dmPFC were related to cognitive flexibility deficits after prolonged cessation from opioid exposure (Anderson et al., 2021). Disrupted cognitive function is associated with a high risk for drug addiction, including increased vulnerability for relapse (Gould, 2010). Therefore, it is worthwhile to investigate whether there is cognitive dysfunction in this fentanyl oral self-administration model and whether it can be recovered by IGF-1 treatment.

All the studies above highlight that excitatory synaptic transmission changes in PFC are key components for the neurobiology underlying opioid addiction, including fentanyl. Additionally, considering the role of dmPFC and ventromedial PFC in heroin seeking is inconsistent throughout the literature [e.g., inactivation of PFC enhances cue-induced heroin reinstatement (Schmidt et al., 2005), other studies have found an attenuation of cue-, heroin-, or context-induced heroin reinstatement (Rogers et al., 2008; Bossert et al., 2011, 2012], future studies are needed to elucidate whether dmPFC or ventromedial PFC plays distinct roles in fentanyl seeking.

We found that IGF-1 treatment blocked the fentanyl cessation–induced inhibition of downstream signaling molecule S6 (Figure 5). Ribosomal protein S6 is part of the small ribosomal subunit 40S forming 43S preinitiation complex, which enhances mRNAs translation at synapses and elsewhere to modulate synaptic plasticity (Costa-Mattioli and Sonenberg, 2006). Furthermore, PI3K/Akt/mTOR signaling interacts with other pathways to directly modulate AMPAR and NMDAR phosphorylation (Zhao et al., 2019; Ma et al., 2020). Therefore, it is possible that IGF-1 treatment prevented the inhibition of AMPAR and NMDAR function through PI3K/Akt/mTOR/p70S6K/S6 signaling-mediated translation and phosphorylation of GluA1 and GluN1. Future studies are needed to test this hypothesis and elucidate the possible role of other signaling pathways.

In conclusion, our study demonstrates that IGF-1/IGF-1R activity is decreased after prolonged cessation of fentanyl oral self-administration, which is accompanied by suppression of the Akt/S6 signaling pathway. IGF-1 microinjection in the PFC prevents a fentanyl-induced decrease of excitatory synaptic transmission and attenuates fentanyl-seeking behavior. Our study provides novel insight into the neurobiology underlying fentanyl addiction.

Author Contributions

G.L. performed experiments, analyzed data, and drafted the manuscript. S.Y., Y.W., and A.S. performed experiments, and analyzed data. Z.J.W. designed experiments, supervised the project, and wrote the manuscript.

Interest Statement

The authors declare no competing interests.

Data Availability

The data supporting the findings of this study are available in the article and in its online supplementary material.