HCN-Channel-Dependent Hyperexcitability of the Layer V Pyramidal Neurons in IL-mPFC Contributes to Fentanyl-Induced Hyperalgesia in Male Rats

Opioids are often first-line analgesics in pain therapy. However, prolonged use of opioids causes paradoxical pain, termed “opioid-induced hyperalgesia (OIH).” The infralimbic medial prefrontal cortex (IL-mPFC) has been suggested to be critical in inflammatory and neuropathic pain processing through its dynamic output from layer V pyramidal neurons. Whether OIH condition induces excitability changes of these output neurons and what mechanisms underlie these changes remains elusive. Here, with combination of patch-clamp recording, immunohistochemistry, as well as optogenetics, we revealed that IL-mPFC layer V pyramidal neurons exhibited hyperexcitability together with higher input resistance. In line with this, optogenetic and chemogenetic activation of these neurons aggravates behavioral hyperalgesia in male OIH rats. Inhibition of these neurons alleviates hyperalgesia in male OIH rats but exerts an opposite effect in male control rats. Electrophysiological analysis of hyperpolarization-activated cation current (Ih) demonstrated that decreased Ih is a prerequisite for the hyperexcitability of IL-mPFC output neurons. This decreased Ih was accompanied by a decrease in HCN1, but not HCN2, immunolabeling, in these neurons. In contrast, the application of HCN channel blocker increased the hyperalgesia threshold of male OIH rats. Consequently, we identified an HCN-channel-dependent hyperexcitability of IL-mPFC output neurons, which governs the development and maintenance of OIH in male rats.


Introduction
Opioids are the first-line analgesics in pain therapy. Longterm opioid use, however, might result in paradoxical pain known as "opioid-induced hyperalgesia (OIH)." The emergence of OIH may provide significant new obstacles to the successful management of pain. More evidence of OIH in clinical populations has been reported [1]. In addition, individuals with OIH differed in how they responded to various noxious stimulation [1]. This paradoxically increased pain sensitivity could be duplicated not only in opioid addicts but also in healthy human volunteers [2][3][4][5][6][7][8]. There are probably numerous causes for OIH, even though it is generally believed that the brain or brainstem is where the pain processes begin before moving through the spinal cord to the portion of the body that hurts [9]. So, a better comprehension of the underlying mechanism of OIH is prerequired for solving this dilemma.
Pain pathways are complex and dynamic systems that may play different roles in multiple pain models. In general, nociceptive signals are transmitted to the spinal dorsal horn for integration, causing sensitization of neurons in the spinal dorsal horn. The nociceptive signal is in turn transmitted to the cortex via the brainstem and thalamus, among other multilevel relays, to produce pain sensation [10]. Xixi Wang and Sifei Gan contributed equally to this work.
Although the mechanisms of OIH are not yet completely understood, multiple evidence has been revealed for this paradoxical phenomenon including peripheral and central plasticity [11]. Pain-related behaviors are caused by hyperactivity in the laterocapsular division of central nucleus of the amygdala (CeLC, commonly known as the "nociceptive amygdala") [12]. And many previous works showed that the maladaptive plasticity in nociceptive amygdala and periaqueductal gray (PAG) may play a pivotal role in the development and maintenance of OIH [13][14][15][16]. Meanwhile, there is emerging evidence that the medial prefrontal cortex (mPFC) is a central region for pain processing [17,18]. For instance, an increased activity of mPFC has been reported in inflammatory pain model, whereas neuropathic pain was associated with a state of hypo-excitability of mPFC [19][20][21]. Interestingly, mPFC is often interconnected with nociceptive amygdala and periaqueductal gray, the two proved supraspinal OIH modulating centers [22]. Thus, it could be speculated that mPFC might take a role in OIH modulation.
Early, prefrontal cortex in rodents was defined as having connections with the mediodorsal nucleus of the thalamus [23]. Today, there is a broad consensus that mammals have prefrontal lobes that resemble the primate classic prefrontal cortex, so the prefrontal cortex is present in rodents [24]. The mPFC is subdivided into the anterior cingulate cortex (ACC), the prelimbic (PL), and the infralimbic cortex (IL) according to its function and anatomy [19,25]. Most pain studies focus on PL-mPFC and ACC subregions, but little is known about IL-mPFC. Compared with PL-mPFC, IL-mPFC drives stronger direct connections with nociceptive amygdala, a key OIH modulation center [15,22]. So, we postulate that the IL-mPFC might be critically involved in OIH modulation. Anatomically, each region of the mPFC includes five cellular layers [19,26]. The primary output neurons of the mPFC are excitatory pyramidal neurons in layer V, which is a major source of excitatory input to the amygdala [27,28]. These facts raise the possibility that layer V excitatory pyramidal neurons of IL-mPFC are responsible for the pain-control-function. In the current investigation, we reported specific excitatory changes of pyramidal neurons in the IL-mPFC layer V and examine the molecular mechanisms by which these neurons may be involved in OIH regulation.
Recently an important regulator of pain processing has been revealed, which is the hyperpolarization-activated cationic depolarizing current (Ih), mediated by hyperpolarization-activated cyclic nucleotide-gated (HCN) channels [29,30]. Usually, HCN channels have four subtypes which are predominantly expressed in pyramidal neurons of the mPFC, where they control resting membrane properties and modulate neuronal excitability [25,[31][32][33][34]. HCN channels can facilitate DRG neuron excitability in neuropathic pain [35]; blocking it with ZD7288 can provide analgesic effect [36][37][38][39]. On the contrary, in supraspinal central, there are reports that decreased open probability of HCN causes an enhanced input resistance and neuronal excitability observed in mPFC after neuropathic pain [32,33]. Further, it is uncertain what changes of HCN in IL-mPFC layer V output neurons during OIH conditions. Based on the studies reported that repeated opioid exposure can alter HCN channel expression and high HCN channel expression exhibited in mPFC [17,33,40], we hypothesize that HCN channel may also play a role in OIH modulation through its effect on the excitability of layer V pyramidal neurons.
To test this hypothesis, we first determined the hyperexcitability of layer V pyramidal neurons in IL-mPFC by electrophysiological and immunochemical methods in OIH rats. Then, the effects of the positive and negative modulation of these neurons on behavioral hyperalgesia were tested by optogenetic and chemogenetic methods, respectively. Next, to explore the underlying mechanism of the hyperexcitability, the expression of HCN channel was tested. What is more, we used its agonists and antagonists to explore the effects of HCN channel on excitability and input resistance of IL-mPFC layer V pyramidal neurons. Last, HCN blocker ZD7288 was microinjected into the right IL-mPFC to learn more about how it affected behavior in OIH rats.

Animals
All experimental rats were provided by the Laboratory Animal and Biomedical of South-Central University for Nationalities. Male Sprague-Dawley rats weighing 70-320 g were kept in a temperature-controlled room with a 12-h light/ dark cycle and free access to food and drink. Despite the growing recognition of gender variations in pain sensitivity, the trials reported here were conducted only on male rats, as in our and most other earlier studies [15,16]. The rats were 4 weeks old (70-80 g) at the start of viral vector injection. Animals for brain slice physiology and behavioral experiments were 7 to 8 weeks old (240-320 g) waiting for the successful expression of the virus. To acclimate to the laboratory conditions, all animals were transported from the animal facility to the laboratory for at least 1 h before experiment. The Laboratory Animal and Biomedical Ethics Committee of South-Central University for Nationalities sets ethical criteria for all animal utilization procedures. We followed the International Association for the Study of Pain's rules to the letter.

Fiber Implants
For optogenetic activation behavioral experiments, fiber optic cannula (core: 200 μm, length: 5.5 mm, Inper Company, China) was implanted into IL-mPFC 2 or 3 weeks after injection of viral. The fiber tips were 20 μm above the site of IL-mPFC (2.9 mm anterior to bregma; 0.4 mm lateral to midline; depth, 4.7 mm). After the optical fiber implanted, the rats were housed individually in a single cage, and the rearing conditions were described above. And all the rats were given at least 1 week to recover for subsequent optogenetic behavioral experiment.

Cannulation and Microinjection into the IL-mPFC
Ketamine (50 mg/kg) and xylazine (7.5 mg/kg) were used to make rats unconscious while they were placed in a stereotaxic frame (RWD Life Science, Shenzhen, China). A 33-gauge prosthetic cannula was then inserted into the right IL-mPFC (coordinates: 2.9 mm anterior to the bregma; 0.4 mm lateral to the midline; and 4.7 mm ventral from the surface of the skull). All rats were individually housed and given at least a week to recover after cannulation before receiving microinjections. The individual performing the experimental groupings is different from the people performing the behavioral tests. Firstly, rats were randomly divided into DMSO and ZD7288 groups, each group was then randomly divided into control (saline) and OIH (fentanyl) group, a total four groups (control + DMSO; OIH + DMSO; control + ZD7288; OIH + ZD7288). Baseline of mechanical and thermal pain thresholds were measured in all groups prior to saline/fentanyl injection. After 6.5 h fentanyl or saline injection, mechanical and thermal pain thresholds of all rats were obtained again. HCN blocker ZD7288 (MedChemExpress, Shanghai, China) was dissolved in 2% DMSO. Then, the right IL-mPFC received a total of 0.5 μL of medication using an automated dosing pump through a 10-μL Hamilton syringe at 0.25 μL/min. The injector was left in the area for an additional 10 min after injection to allow the medication to disseminate. The same volume of a vehicle solution of DMSO (2%, 0.5 μL) was administered to rats in the control group. Ten to 15 min after microinjection, we measured the mechanical and thermal pain thresholds again. Rat brains were cut into sections after each experiment to verify the precision of the microinjection.

OIH Model
OIH was established in rats as described previously [15,41]. Fentanyl (60 μg/kg) was administered to rats at 15-min intervals for four times for a cumulative dose of 240 μg/ kg to induce OIH. The control rats received the same volume of 0.9% NaCl. The mechanical and thermal nociceptive thresholds were tested at 0 (baseline), 1, 3, 5, 6.5, 9, and 10 h and on days 1, 2, 3, 4, 5, 6, and 7 after the last injection of fentanyl (time 0). A successful OIH model was confirmed at 6.5 h by mechanical threshold. Individuals performing the behavioral testing were blinded to treatment group. In addition, OIH induced by intermittent fentanyl injections is different from chronic administration, which can also induce opioid tolerance [42].

Nociceptive Behavioral Testing
The individual performing the injection is different from the person performing all the behavioral tests. The sensitivity to harmful stimuli was determined by measuring the mechanical paw withdrawal thresholds (PWT) and thermal paw withdrawal latencies (PWL) at the central left hind paw. To adapt to experimental environment, rats were exposed to the measurement room for 30 min before measurement.

Mechanical Paw Withdrawal Thresholds
To evaluate mechanical allodynia, mechanical paw withdrawal threshold was measured using von Frey filaments (North Coast, San Jose, CA, USA) according to the upand-down method described by previous study [15,16]. Before the trial, rats were placed alone in a plexiglass chamber above a mesh table for 30 min to acclimate. Starting from 1.4 g, a series of von Frey filaments were sequentially applied vertically to the middle of the left hind paw of the rats until an explicit paw withdraw, indicating a positive response, or lasting for 15-20 s to confirm a negative response. Based on the response to the stimulation, the 50% probability of PWT was calculated by up-and-down algorithm.

Thermal Paw Withdrawal Latency
To evaluate thermal hyperalgesia, Hargreaves test was applied with a modified thermal stimulator system (BME-410 C Thermal Pain Stimulator, Bern Technology Co., Ltd., Tianjin, China). Rats were placed in a single plexiglass chamber for 30 min prior to measurement, and the intensity of the illumination light was adjusted to maintain a basic pain threshold of 10 to 13 s. The laser beam vertically irradiates the center of the left hind paw through the glass bottom plate. The positive reaction is defined as an explicit paw withdrawal. Record the reaction time 3 times, and take the average value, which is PWL. The interval between each test should be at least 5 min, with a 15-s cut-off time to avoid excessive tissue damage.

Optogenetic Behavioral Experiments
The individual performing the experimental groupings is different from the people performing the behavioral tests. The light from a patch-cord attached to a 473-nm laser was used to deliver optical stimulation through the fiber-optic cannula. Laser output was controlled using an Inper pulse stimulator (Inper Company, China). The excitatory neurons of IL-mPFC were optically activated by laser light pulses (5 ms, 20 Hz, 8-10 mW) produced by a blue laser (473 nm; Inper Company, China).

Chemogenetic Behavioral Experiments
DMSO at a concentration of 0.1 mg/μL was used to dissolve Clozapine-N-Oxide (CNO, Brain VTA, China), then stored the stock at − 20 °C. Immediately prior to the experiment, the stock was dissolved in 0.9% NaCl to produce a working solution of 0.3 mg/mL. Similarly, the individual performing the experimental groupings is different from the people performing the behavioral tests. Baseline of mechanical and thermal pain thresholds were measured in all rats prior to saline/fentanyl injection. OIH rats were subcutaneously received fentanyl to induce OIH model, while the control rats received the same volume of saline. 6.5 h after injection, all rats received measurement in PWT and PWL. Finally, rats injected with 3 mg/kg CNO intraperitoneally were measured for PWT and PWL half an hour later.
For chemogenetic activation, rats received hM3D (Gq)-EGFP and EGFP which were randomly divided into EGFP group, EGFP + CNO group, hM3D-EGPF group, and hM3D-EGPF + CNO group according to whether they received intraperitoneal injection of CNO or not. Then, each group was randomly divided into control group and OIH group according to the injection of saline/fentanyl.
For chemogenetic inactivation, the grouping method is the same as above.

Patch-Clamp Recording
All patch-clamp recordings were made from IL-mPFC layer V excitatory neurons. IL-mPFC excitatory cells were visualized with a Nikon microscope (Nikon Company, Japan). Neurons expressing ChR 2 were activated by blue light (473 nm, 1 Hz). After identifying the virus-labeled cells, we could observe the luminescent cells on the display and then record the electrophysiological properties of these cells. Electrodes (impedance was 4-6 MΩ) were pulled with a Flaming/Brown Micropipette Puller (Model P-97, Sutter Instrument, USA), which were filled with internal solution (in mM, pH 7.3, osmolality 295 mOsm/kg): 145 KCl, 10 HEPES, 5 EGTA, 5 NaCl, 4 Mg-ATP, and 0.3 Na 3 GTP. Data acquisition and analysis were performed using a dual four-pole Bessel filter (Warner Instruments, Hamden, CT), a low-noise Digidata 1322 interface (Molecular Devices, Sunnyvale, CA), HEKA EPC-10 amplifier (HEKA, Lambrecht, Germany), a Pentium PC, and patch master software (Molecular Devices). The excitability of neurons was determined by action potentials through recording the input-output curve measured from the changes in membrane potential. Action potentials of layer V IL-mPFC excitatory neurons were evoked by a series of current steps (ranging from 0 to 140 pA in 20 pA increment) for 1.2 s from a potential of − 70 mV. Injecting a negative current pulse (starting from − 300 pA), then determine the input resistance (Rin) in 30 pA increments in current-clamp mode by dividing the voltage value by the corresponding current value. BaCl 2 (200 μM) was bathed into the extracellular solution in chamber to suppress K + conductance, and hyperpolarizationactivated cyclic nucleotide-gated (HCN) channels (Ih) were recorded in voltage-clamp mode at a negative membrane potential. The membrane potential changes through a series of hyperpolarization voltage steps (lasting for 2.5 s), which decrease from the holding potential of − 50 mV to the final voltage of − 130 MV (− 10 mV increments). The maximal available Ih current was determined from the − 130 mV step.

Immunohistochemical Staining
The neuronal excitability in IL-mPFC layer V excitatory neurons was assessed by staining with c-Fos. HCN1 protein's change was also assessed in IL-mPFC layer V excitatory neurons. Ketamine (50 mg/kg) and xylazine (7.5 mg/kg) were used after 6.5 h fentanyl or saline injection to anesthetize rats, then the left ventricle was perfused with 0.1 M cold phosphate-buffered saline (PBS) and 4% paraformaldehyde (PFA) in PBS, respectively. The brains were extracted and postfixed in 4% PFA overnight at 4 °C. Then, brains were cryoprotected in 20% sucrose for 24 h at 4 °C followed by 30% sucrose for 24 h. Brain free-floating Sects. (30 μm) cut on vibration microtome (LEICA VT1200S) in chilled 0.1 M PBS were prepared for immunohistochemical staining. QuickBlockTM Blocking Buffer for Immunol Staining (Beyotime, Wuhan, China) was used to block the brain slices in a 24-well Tissue Culture Plate (Biofil, Guangzhou, China) for 10 min at room temperature, then incubating the Sects. (12-24 h at 4 °C) with the primary antibodies: anti c-Fos (1:300, rabbit, CST technology), anti-HCN1 (1:100, rabbit, Abclonal), anti-HCN2 (1:100, rabbit, Proteintech), and anti-CaMKIIα (1:300, mouse, CST). QuickBlockTM Primary Antibody Dilution Buffer for Immunol Staining (Beyotime, Wuhan, China) was used to dilute the primary antibodies. After being rinsed with 0.1 M PBS (pH 7.4), the brain slices were incubated with fluorophore-conjugated secondary antibodies for 3 h at room temperature (1:300, Alexa Fluor 488 Affinipure Donkey Ati-Rabbit lgG (H + L), Cy3 Goat Anti-Mouse IgG (H + L)). QuickBlockTM Secondary Antibody Dilution Buffer (Beyotime, Wuhan, China) was used to dilute the secondary antibodies. After immunohistochemistry, nuclei were stained with DAPI Staining Solution for 10 min (Beyotime, Wuhan, China). Subsequently, washing above brain sections three times with 0.1 M PBS for 10 min each, briefly, confocal images were captured using a 40 × objective with a Nikon TE-200 U inverted fluorescence microscope. Cell counting were analyzed using the ImageJ analysis system (National Institutes of Health [NIH], Bethesda, MD).

Statistical Analyses
Means (standard error of the mean) are used to express all numerical data. We employed two-way analysis of variance (ANOVA) with Bonferroni adjustments for behavioral tests. An unpaired Student's t test was used to compare two groups of immunohistochemistry results. We utilized GraphPad Prism 5.0 software (GraphPad Software, San Diego, CA) for statistical analysis and defined statistical significance as P < 0.05. Figure 1A is the schematic diagram of behavioral experimental design. We subcutaneously administered fentanyl into rats to study the fentanyl-induced hyperalgesia. Following the last fentanyl injection, behavioral thresholds were measured. As shown in Fig. 1B, C, increased mechanical and thermal hyperalgesia thresholds were found at 1 h and 3 h after exposed to fentanyl compared to saline injection. In contrast, mechanical and thermal hyperalgesia reached their nadir at 6.5 h after fentanyl exposure, with hypersensitivity states gradually returning to basal levels by day 7. These results corroborate findings from previous reports that rats exposed to opioids experience hyperreflexia symptoms in a time-dependent manner [15,34,43]. Figure 2A is the schematic diagram of immunohistochemical experimental design used to assess the neuronal excitability in IL-mPFC layer V excitatory neurons. To investigate whether layer V excitatory neurons in IL-mPFC changes in neuronal activity in OIH rats, we initially measured c-Fospositive and Ca 2+ /calmodulin-dependent protein kinase IIα (CaMKIIα, a putative marker of glutamatergic neurons) co-expressing cells (Fig. 2B); yellow arrows indicated co-labeled cells. We discovered that the number of c-Fospositive glutamatergic neurons was increased in OIH rats compared to control rats (Fig. 2C). Figure 3A (top) is a schematic diagram of the experimental design of virus injection and expression. To further directly examine the activity of the layer V glutamatergic neurons in IL-mPFC, we labeled these neurons by injecting AAV viral vector under controlled CaMKIIα promotor into IL-mPFC (Fig. 3A, bottom). Measuring the number of action potentials (AP) and the rheobase at 6.5 h after the last fentanyl injection, which are often used to evaluate the  green: c-Fos; blue: DAPI (bar: 50 μm). Yellow arrows indicate the co-labeled cells. C The bar graphs show summaries of the co-staining c-Fos and CaMKIIα data in the two groups (unpaired t-test; n = 4, P = 0.0070) intrinsic membrane excitability, analysis of the AP features demonstrated that the spikes in OIH rats was significantly increased compared to control rats (Fig. 3B). The rheobase, the minimum current required to elicit an action potential, was decreased in OIH rats (Fig. 3C). Representative traces from layer V glutamatergic neuron recordings were taken from saline-or fentanyl-treated rats in 40 pA, 80 pA, and 120 pA steps, respectively (Fig. 3D). The initial membrane potential each trace is − 70 mV. Above we observed that fentanyl-exposed rats had a significant increase in intrinsic excitability. The observed fentanyl-induced increase in AP spikes and decrease in the rheobase suggest a likelihood of change in input resistance (R in ). To identify this point, R in was determined by injecting a negative current pulse (− 300 pA) and then increasing 30 pA in current clamp mode. Figure 3E is its I/V curve. Compared with control group, R in of neurons in OIH group increased (Fig. 3F), which was calculated based on injection of 300 pA current step. The mean R in of neurons was 32.19 ± 1.592 MΩ from OIH rats and 28.68 ± 0.8677 MΩ from control rats. Figure 3G shows the family of whole cell voltage traces (vertical scale bar is 20 mV) in response to a series of current steps.

Photoactivation of IL-mPFC Glutamatergic Neurons Aggravates Fentanyl-Induced Hyperalgesia
The combined results of the c-Fos immunohistochemistry and electrophysiological experiments in IL-mPFC slices demonstrate that sensitization of IL-mPFC output neurons is necessary for OIH-related mechanical and thermal hypersensitivity. Experiments were then conducted to determine whether activation of IL-mPFC output neurons was sufficient to induce hypersensitivity to mechanical and thermal stimulation of the hind paw. For this purpose, we injected the AAV-CaMKIIα-ChR2-eYFP into the right IL-mPFC of rats and performed the behavioral tests before and after laser light (Fig. 4A, B). Three weeks after virus expression, we observed ChR2 expression in the IL-mPFC and validated its function. Current-clamp recordings from IL-mPFC acute brain slices demonstrated that blue laser pulses elicited time-locked action potential firing in ChR2-eYFP-infected neurons, establishing the expression and function of ChR2 in these neurons (Fig. 4C). Then, rats were randomized into two groups: control and OIH. OIH rats were subcutaneously received fentanyl to induce OIH model. The control rats received the same volume of saline. Rats were kept in plexiglass chamber for 30 min of habituation prior to the experiments. The Von Frey and Hargreaves tests were performed to evaluate the mechanical and thermal paw-withdrawal threshold. We found that the photoactivation of IL-mPFC glutamatergic neurons in OIH rats with blue laser light (473 nm, 20 Hz, 8 mW) aggravated mechanical hyperalgesia (Fig. 4D). No significant difference in thermal threshold between light off and light on in OIH group was found (Fig. 4E). Overall, our results indicated that optical activation of IL-mPFC glutamatergic neurons in control rats was not sufficient to induce hyperalgesia neither for mechanical or thermal behavior. neurons decreases the mechanical paw withdrawal threshold, compared to before optic activation (light off) in OIH rats (paired t-test, P = 0.0158). No significant differences were found between before (light off) and after (light on) light stimulation in control rats. E Optic activation (473 nm, 20 Hz, 5-ms pulse width, 8 mW) of IL-mPFC glutamatergic neurons did not change the thermal paw withdrawal threshold injected with AAV-CaMKIIα-ChR2-eYFP, compared to before optic activation in OIH rats (paired t-test, P = 0.0901). No significant differences were found between before (light off) and after (light on) light stimulation in control rats

Chemogenetic Activation of IL-mPFC Glutamatergic Neurons Aggravates Fentanyl-Induced Hyperalgesia
The above results suggest that optogenetic activation of glutamatergic IL-mPFC cells aggravates fentanyl-induced mechanical hypersensitivity. To identify whether prolonged pharmacogenetic activation of these neurons has the same effect on fentanyl-induced hyperalgesia, we injected rAAV 2/9-CaMKIIα-hM3D (Gq)-EGFP-WPREs-pA (hM3D (Gq)-EGFP) and a control viral vector rAAV 2/9-CaMKIIα-EGFP-WPRE-hGH pA (EGFP) into the right IL-mPFC (Fig. 5A). Control virus was used in the present study to exclude clozapine-N-oxide (CNO) effects on the central nervous system. Both rats received hM3D (Gq)-EGFP and EGFP which were randomized into control and OIH groups. OIH rats were subcutaneously received fentanyl to induce OIH model, while the control rats received the same volume of saline. Three weeks after virus expression (Fig. 5B), the baseline of mechanical and thermal threshold was tested. Whole-cell patch clamp recordings performed on IL-mPFC slices from rats expressing hM3Dq in the IL-mPFC layer V neurons acknowledged that bathing with clozapine-Noxide (CNO; 40 μM), but not saline, significantly activated neuronal activity in hM3Dq transduced cells (Fig. 5C). The results of selective activation of IL-mPFC glutamatergic neurons on mechanical and thermal behavior were tested pre-and postintraperitoneal injection of CNO (3 mg/kg) in control and OIH rats (Fig. 5D, E). The results indicated that CNO-mediated chemogenetic activation of IL-mPFC glutamatergic neurons aggravated OIH-induced hypersensitivity to the mechanical (Fig. 5D) and thermal (Fig. 5E) paw withdrawal threshold/latency in OIH rats injected with hM3D (Gq)-EGFP. Significantly, CNO application did not affect behavioral responses in control rats which received hM3D (Gq)-EGFP and EGFP, demonstrating that the effects of activating IL-mPFC glutamatergic cells have specificity for pain state. The effects of CNO are hM3Dq-specific because CNO application did not influence OIH-induced hyperalgesia in rats transduced with the EGFP control virus.

Chemogenetic Inactivation of IL-mPFC Glutamatergic Neurons Alleviates Fentanyl-Induced Hyperalgesia
To establish a causal relationship between IL-mPFC glutamatergic cell activity and OIH induced mechanical and thermal hyperalgesia, we stereotaxically microinjected rAAV 2/9-CaMKIIα-hM4D (Gi)-EGFP-WPRE-hGh pA (hM4D (Gi)-EGFP) (encoding the inhibitory domain) or rAAV-CaMkIIα-EGFP-WPRE-hGh pA (EGFP) (null control) into the right IL-mPFC (Fig. 6A). Both rats received hM4D (Gi)-EGFP and EGFP which were randomized into control and OIH groups, respectively. The remaining steps are the same as chemogenetic activation. Whole-cell patch clamp recordings performed on IL-mPFC slices from rats expressing hM4Di in the IL-mPFC layer V neurons bathing with clozapine-N-oxide (CNO; 40 μM), but not saline, significantly inactivated neuronal activity (Fig. 6B). CNOmediated chemogenetic inhibition of these neurons is only sufficient to reverse mechanical (Fig. 6C), but not thermal (Fig. 6D) hypersensitivity in OIH rats received hM4D (Gi)-EGFP. The effects of CNO are hM4Di-specific because CNO application had no effect on OIH-induced hypersensitivity in rats transduced with the EGFP control virus. Interestingly, unlike ineffectively of chemogenetic activation on behavior in control rats, chemogenetic deactivation of IL-mPFC glutamatergic neurons induces mechanical and thermal hyperalgesia in control rats transduced with hM4D (Gi)-EGFP but not EGFP.

HCN Channel-Mediated Current Decreased in IL-mPFC Layer V Glutamatergic Neurons in Fentanyl-Induced Hyperalgesia Rats
The HCN channel is widely expressed on IL-mPFC layer V glutamatergic neurons, which hold a key position in controlling neuronal excitability. Thus, we next investigated whether HCN channel-mediated current (Ih) is altered on IL-mPFC layer V glutamatergic neurons 6.5 h after inducing of OIH model. The virus injection method was the same as above; labeled cells were identified under blue laser light (473 nm) and recorded with light off (Fig. 7A). Ih was recorded on IL-mPFC layer V glutamatergic neurons from rats injected with saline or fentanyl. In the presence of Ba 2+ (200 μm) to inhibit IKir (inward rectified potassium channel), Ih currents were tested on IL-mPFC layer V glutamatergic neurons of both groups (n = 6 in control group; n = 8 in OIH group) in response to a series of hyperpolarizing voltage steps (− 50-140 mV). We measured Ih density (Ih amplitude/ the mechanical threshold in control rats received hM4D (Gi)-EGFP (hM4D-EGFP and hM4D-EGFP + CNO groups, P = 0.0205; 0.632 ± 0.1627 vs. 0.146 ± 0.0449, n = 5 for each) but not EGFP (EGFP and EGFP + CNO groups, P = 0.52, n = 6 for each). D Chemogenetic deactivation of the glutamatergic neurons in the IL-mPFC did not change the thermal threshold in OIH rats received hM4D (Gi)-EGFP (hM4D-EGFP and hM4D-EGFP + CNO groups, P = 0.7004, n = 5 for each) and EGFP (EGFP and EGFP + CNO groups, P = 0.1556, n = 7 for each). Chemogenetic deactivation of the glutamatergic neurons in the IL-mPFC decreased the thermal threshold in control rats received hM4D (Gi)-EGFP (hM4D-EGFP and hM4D-EGFP + CNO groups, P = 0.0002; 14.57 ± 0.698, n = 5; 9.274 ± 0.421, n = 5 for each) but not EGFP (EGFP and EGFP + CNO groups, P = 0.1479, n = 6 for each). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 surface area: PA/PF) to exclude the effects of neuronal surface area. Compared to the control groups, results indicated that Ih current density decreased in the OIH groups at hyperpolarizing voltage steps from − 60 to − 140 mV (Fig. 7B). The maximal Ih amplitude (we assessed Ih from the − 130-mV step) significantly decreased in OIH rats compared to control rats (Fig. 7C). Figure 7D is the trajectories of the Ih current of the control group and the OIH group, respectively. Ih current density and maximal Ih amplitude were blocked by the application of ZD7288 (50 μM, selective HCN channel antagonist) in control rats (Fig. 7E). Figure 7F is the trajectories of the Ih current of the control group and the control + ZD7288 group. Similar changes occurred in the OIH group (Fig. 7G). Figure 7H is the trajectories of the Ih current of the OIH group and the OIH + ZD7288 group.

Decreased HCN1 Not HCN2 Was Involved in the Increased Excitability of Layer V Glutamatergic Neurons of IL-mPFC in OIH Rats
Then, we performed immunohistochemical studies to identify the kind of HCN channel that was involved. The immunohistochemistry identifies the expression of HCN1 in the IL-mPFC layer V glutamatergic neurons (Fig. 8A), demonstrating a significant decrease in expression of HCN1 levels in OIH rats compared to control rats (Fig. 8B). The expression of HCN2 in the IL-mPFC layer V glutamatergic neurons (Fig. 8C) demonstrates that there were no changes in the expression of HCN2 levels in OIH rats compared to control rats (Fig. 8D).

Effect of HCN Channel Agonist cAMP on Action Potentials and Input Resistance in Layer V Glutamatergic Neurons
To determine whether the observed changes in action potentials (AP) and input resistance (IR) were mediated by decreased expression of HCN channel, we measured AP and IR before and after application of HCN agonist (cAMP, 100 μM) in control and OIH group. The AP spikes at injected current 140 pA was decreased in control group after application of cAMP (Fig. 9A). Figure 9B is the AP trace diagram of control group. The spikes of AP decreased significantly in OIH group after cAMP application (Fig. 9C). Figure 9D is the AP trace diagram of OIH group. Notably, there was no statistical difference in the number of AP between control group (without cAMP) and OIH + cAMP (curve: Fig. 9E; trace diagram: Fig. 9F), indicating a latent therapy effect of cAMP. No difference between before and after cAMP application for IR in control group was found (curve: Fig. 9G; IR: Fig. 9H). Rin in Fig. 9 was calculated based on injection of 300 pA current step. Figure 9I is the family of whole cell voltage traces (vertical scale bar is 20 mV) in response to a series of current steps of control group. The initial holding potential is − 70 mV. However, after application of cAMP, IR was decreased in OIH group (curve: Fig. 9J; IR: Fig. 9K); Fig. 9L is the family of whole cell voltage traces (vertical scale bar is 20 mV) in response to a series of current steps of OIH group. We found no difference in control group (without cAMP) and OIH + cAMP group (curve: Fig. 9M; IR: Fig. 9N). Figure 9O is the family of whole cell voltage traces (vertical scale bar is 20 mV) in response to a series of current steps of them.

Effect of HCN Channel Blocker ZD7288 on Action Potentials and Input Resistance in Layer V Glutamatergic Neurons
In a similar vein, we want to investigate how HCN blocker (ZD7288, 50 μM) affects the aforementioned factors in Fig. 8 Decreased HCN1 but not HCN2 was involved in the increased excitability of layer V glutamatergic neurons of IL-mPFC in OIH rats. A HCN1 immunofluorescent co-staining with CaMKIIα in control and OIH groups. Red: CaMKIIα. Green: HCN1. Blue: DAPI. White: merge. B The expression of HCN1 protein in IL-mPFC layer V glutamatergic neurons decreased in OIH rats compared to control rats by immunochemistry (unpaired t-test, P = 0.0027). C HCN2 immunofluorescent co-staining with CaMKIIα in control and OIH groups. Red: CaMKIIα. Green: HCN1. Blue: DAPI. White: merge. D The expression of HCN2 protein in IL-mPFC layer V glutamatergic neurons not changed in OIH compared to control by immunochemistry (unpaired t-test, P > 0.05) control and OIH group. Differently, the results showed that ZD7288 had no effect on spikes of AP both in control (curve: Fig. 10A; trace diagram: Fig. 10B) and OIH group (curve: Fig. 10C; trace diagram: Fig. 10D). The spikes of AP in OIH were still more than that in the control + ZD7288 group (curve: Fig. 10E; trace diagram: Fig. 10F). There was no difference of IR before and after ZD7288 application in control group (curve: Fig. 10G; IR: Fig. 10H). Figure 10I is the IR trace diagram of it. Same result was obtained in OIH group (curve: Fig. 10J; IR: Fig. 10K); Fig. 10L is the IR trace diagram of it. Compared to control group (without ZD7288), there was an increase for IR in OIH + ZD7288 group (curve: Fig. 10M; IR: Fig. 10N). Figure 10O is the IR trace diagram of them.

HCN Channel Blocker Has an Analgesic Effect on Fentanyl-Induced Hyperalgesia Rats
The results of patch-clamp experiments found that there were statistical differences in the AP spikes between the control group (without ZD7288) and the OIH + ZD7288 group. Therefore, we eager to further explore the effect of ZD7288 on behavior in OIH in vivo. Figure 11A is the schematic diagram of cannulation and subsequent behavioral experimental design. Both DMSO and ZD7288 groups were randomly divided into two groups (control + DMSO; OIH + DMSO; control + ZD7288; OIH + ZD7288). As shown in Fig. 11B, C, neither DMSO nor ZD7288 had any impact on the basal pain thresholds in the control groups and that DMSO also had no effect on the pain thresholds of OIH rats. Interestingly, ZD7288 microinjection markedly raised the mechanical and thermal hyperalgesia thresholds in OIH group.

Discussion
In the present study, electrophysiological and immunohistochemistry approaches were utilized to assess intrinsic excitability of IL-mPFC pyramidal neurons in OIH rat model with a special focus on layer V output neurons. Specifically, we found that IL-mPFC layer V glutamatergic neurons showed an increased excitability manifested by increased AP firing number, input resistance, and decreased rheobase. The coexpression of c-Fos and CaMKIIα further demonstrated this hyperexcitability of IL-mPFC output neurons. What would be the behavioral effects of activating or inhibiting the activity of these neurons in vivo? To achieve this, we performed optogenetic (activation) and chemogenetic (activation and inactivation) experiments separately. Remarkably, activating these neurons exacerbates mechanical and thermal hypersensitivity in OIH rats. In contrast, inhibiting these neurons is necessary for alleviating mechanical hypersensitivity but not thermal allodynia. Furthermore, the hyperexcitability of IL-mPFC output neurons is attenuated by activating HCN through the nucleotide cyclic adenosine 3′, 5′-monophosphate (cAMP). Thus, combined with the above immunohistochemistry results, we anticipate that the downregulation of HCN1 channel revealed in this study might bias IL-mPFC output neurons toward hyperexcitability, thereby inducing OIH. To further explore the effect of HCN channel on the behavior of OIH, we microinjected HCN blockers ZD7288 into the right IL-mPFC. Paradoxically, OIH-induced hyperalgesia was ameliorated or even reversed. The mechanisms responsible for the opposite results in behavioral, molecular, and cellular deserve further study.
It has been established that PL-mPFC is critically involved in pain processing [25,44]. The IL-mPFC may differentially participate in opioid abuse [45,46] and play critical role in cocaine abuse and drug seeking [47]. Its role in pain process is emerging recently, and it seems opposite to PL-mPFC [48][49][50]. For the first time, our study identified IL-mPFC as a key center for OIH modulating. Photogenetic and chemogenetic approaches revealed that the OIH behaviors are attributable to specific activation of IL-mPFC pyramidal neurons. More specifically, selective activation or inhibition of IL-mPFC glutamatergic neurons strengthens or alleviates hyperalgesia behaviors in OIH rats. Chemogenomic inactivation of these neurons mimicked hyperalgesia in normal rats, which is similar to what is observed after lidocaine microinjection [51]. It means that these neurons are required for preventing nociception in normal rats. Importantly, our behavioral results proposed that there was a causal link between hyperactivity of IL-mPFC glutamatergic neurons and fentanyl induced hyperalgesia. Taken together, the activity of IL-mPFC glutamatergic neurons presents in a dynamical form according to different pathophysiological conditions. Several mechanisms have been proposed; first, it has been proved that peripheral nociceptive information transmitted to amygdala wherein it activates amygdala, and the hyperactivity is relayed to inhibitory interneurons in IL-mPFC, which inactivate the layer V output neurons in IL-mPFC [52,53]; we call this as feedforward inhibitory mechanism. Second, it is of interest to note that the loss of parvalbumin expressing (PV + ) neurons and reduction of axon initial segment length have been reported in layer V/ VI neurons of IL-mPFC in neuropathic pain [54], which is termed as a major structural change's mechanism. Third, local cortical networks in the prefrontal cortex have the ability to generate sustained activity for seconds or longer on their own [55], which may facilitate the state of hyperalgesia. In the current study, we sought to explore the intrinsic electrophysiological mechanisms which underline the augmentation of neuronal excitability in the layer V of IL-mPFC and found that the hyperexcitable state is correlated with decreased Ih, which is mediated by the down expression of HCN1 channel.
HCN channels have been reported to play a major role in pain via promoting hyperexcitability and ectopic firing in neurons [56], and it has attracted a lot of attention as a therapeutic target for pain treatment [57,58]. For instance, under situations of chronic pain, opening channel probability of HCN at physiological membrane potential is decreased [59]. Increased cAMP is associated with increased conductance of Ih current and opening probability of HCN channels [60], indicating that HCN channel contributes to neuronal excitability via increased input resistance [32,33,56,61,62], which is consistent with our observation that HCN agonist 8-bro-cAMP prevented the accumulated-fentanyl-induced alterations in increased input resistance and the excitability of layer V pyramidal neurons in IL-mPFC. Repeated opioid exposure has been shown to downregulate cAMP signaling [34], which would further contribute to decreased HCN currents Ih. Thus, accumulated-fentanyl-induced suppression of Ih in layer V pyramidal neurons in IL-mPFC observed here may cause enhanced activation of these neurons. In agreement, our findings suggest that decreased HCN, by increasing the propensity for action potentials to occur in layer V pyramidal neurons, would lead to greater IL-mPFC neuronal output and hence hyperalgesia after exposed to fentanyl. Though application of ZD7288 has been shown previously to boost action potential firing in prefrontal cortex [63], we did not observe any changes on excitability and input resistance after its antagonist ZD7288 application in OIH condition. It is worth noting that the role of ZD7288 needs to open the channel and then play an inhibitory role by closing the channel [64]. It is possible that all HCN channels exist in hyperpolarized state in OIH neurons, with concomitant no response after application of its antagonist. In fact, we logically hypothesize that HCN may be a key node for changes in neuronal excitability because numerous Fig. 11 HCN channel blocker has an analgesic effect on fentanylinduced hyperalgesia rats. A The schematic diagram of cannulation and behavioral experimental design. B OIH groups successfully developed mechanical hyperalgesia after fentanyl injection. Neither DMSO nor ZD7288 had any effect on the basal mechanical pain thresholds in the control groups, and DMSO also had no effect on the mechanical pain thresholds of OIH rats. Microinjection of ZD7288 into the right IL-mPFC could significantly increase mechanical hyperalgesia thresholds in OIH group (P = 0.0066; OIH + DMSO = 7; OIH + ZD7288 = 9), but not the control group (P = 0.7327, control + DMSO = 6; control + ZD7288 = 8). C Microinjection of ZD7288 into the right IL-mPFC could significantly increase thermal hyperalgesia thresholds in OIH group (P < 0.0001; OIH + DMSO = 7; OIH + ZD7288 = 10), but not the control group (P = 0.5310). Neither DMSO nor ZD7288 had any effect on the basal thermal pain thresholds in the control groups, and DMSO also had no effect on the thermal pain thresholds of OIH rats neurotransmitters and neuromodulators have been demonstrated to influence neuronal excitability via Gq-coupled metabotropic receptors and, more importantly, because HCN channels have been demonstrated to be regulated by neurotransmitters like dopamine and norepinephrine [59,65]. Then, how HCN channels regulate neuronal excitability is worthy of further investigation. Previous analysis showed that inhibiting of HCN channels facilitates somatic-dendritic excitatory postsynaptic potential (EPSP) amplitudes, thus boosting neuronal excitability [62]. Another study reported that loss of HCN1 channel enhances glutamatergic synaptic transmission and alters T-type Ca 2+ channel activity, which may further facilitate neuron excitability [66]. Other studies also have shown that HCN channel dysfunction plays a pivotal role in regulating dopamine receptor 1 (D1R), thereby promoting cortical hyperexcitability in neuropathic conditions [33,59]. However, one report showed that blockade of HCN channel activity results in potentiated inhibitory control over layer V pyramidal cells by enhanced presynaptic Ca 2+ entry [67]. This seems contradict to our experimental results. Of note, that layer V pyramidal neurons are driven not only by local inhibitory input but also excitatory input [19]. These findings back with the theory that accumulated-fentanyl-induced suppression of HCN leads to a net increased excitatory input to layer V pyramidal neurons in IL-mPFC.
Changes in both function and expression of HCN channel are likely to affect cellular excitability. It is indispensable to explore whether HCN channel changed, if so, which kind(s) of HCN channel changed in OIH rats. HCN protein expressions are impaired in OIH rats, as evidenced by decreased maximal available current (measured at − 140 mV) because current density reflects functional protein levels [33]. Four members of HCN channels have been identified (HCN 1-HCN 4) [57,68]. The lack of known off-target effects of ZD7288 and cAMP allowed all isoforms of HCN be blocked or activated by ZD7288 or cAMP [60,69]. Therefore, the results of electrophysiological experiments cannot determine which type of HCN has changed. The HCN1 subunit is predominantly expressed in the neocortex and mPFC, in which it is primarily located in pyramidal cells [4,58,70]. Indeed, HCN1 expression is significantly reduced in the IL-mPFC layer V pyramidal neurons after OIH by immunohistochemistry experiment. Although HCN2 also plays a role in pain processing [33], our experiments did not find it to be altered in OIH. The frequency of miniature excitatory postsynaptic currents (mEPSCs) increased when HCN1 channels were inhibited pharmacologically or genetically, demonstrating that presynaptic HCN1 channels regulate the basal glutamatergic synaptic release [66]. Our analysis did not show the contribution of HCN1 to synaptic release and transmission, even though it is possible that there are OIH-dependent increases of synaptic release and transmission.
Finally, we explored the behavioral significance of decreased Ih in the IL-mPFC by measuring mechanical and thermal threshold after acute microinjection of ZD7288 into the IL-mPFC. It is interesting to note that ZD7288 microinjection leads to an analgesic effect in OIH rats, which is in agreement with earlier findings showing that ZD7288 microinjection into mPFC has analgesic effects in different rat models of neuropathic pain [33]. In reality, treatment with ZD7288 in the peripheral nervous system suppresses cationic currents to achieve analgesia as well as to suppress spontaneous discharges [35]. And the effects of ZD7288 may depend on depression of synaptic transmission, as shown in the presence of Ih blocking treatment for the epileptiform hyperexcitability [71]. Conversely, ZD7288 can also increase neuronal excitability by rasing input resistance [72]. Therefore, we hypothesize that the elevated neuronal excitability caused by the lower Ih current at the electrophysiological level is caused by an increase in input resistance. On the other hand, ZD7288 causes analgesia at the behavioral level by lowering neuronal excitability via reduction of cation currents during OIH. It must be pointed out that we did not employ HCN-null rats like other previous studies [73] since our patch-clamp experiments require both activation and inhibition of HCN channels. To clarify this discrepancy between electrophysiological and behavior, more research is necessary.
One consequence of the hyperexcitability of the layer V neurons is that the glutamatergic output to the amygdala and vlPAG is enhanced [74], wherein OIH would be facilitated [15,16]. Also, IL-mPFC output neurons could inhibit PL-mPFC neuronal activity. And hypo-excitability of PL-mPFC neuron has been revealed in acute and chronic pain models [25,26,74,75]. So, the increased excitability of IL-mPFC glutamatergic neurons might potentially lead to lower PL-mPFC activity and lead to behavior hyperalgesia in OIH rats in the present study. The actual neural circuits through which IL-mPFC modulates OIH is beyond the scope of this investigation and needs to be elucidated.

Conclusion
In conclusion, our study uncovered a novel hypothesis that mechanical and thermal hyperalgesia in a model of OIH in rats are mediated by increased excitability in IL-mPFC layer V pyramidal neurons, which is dependent on decreased Ih current and HCN1 expression. Therefore, targeting Ih current in IL-mPFC layer V pyramidal neurons might represent a new potential therapeutic candidate for the treatment of OIH. However, blockade of HCN channel alleviate fentanyl-induced hyperalgesia. These results indicate that HCNchannel-dependent hyperexcitability of the layer V pyramidal neurons in IL-mPFC contributes to fentanyl-induced hyperalgesia in rats.