The Anterior Insula and its Projection to the Prelimbic Cortex are Involved in the Regulation of 5-HT-Induced Itch

Animal surgeries and virus injections were performed according to similar procedures described previously [36‐38]. Three percent pentobarbital sodium was administrated to anesthetize the rats (40 mg/kg, dissolved in saline, i.p.). After fixation in a stereotaxic apparatus, viruses were injected into the brain at 0.05 µL/min with a glass micropipette (tip diameter 10–20 µm). Then, the micropipette was left in place for 5 additional minutes and slowly withdrawn.

Traditionally, calcium/calmodulin-dependent protein kinase (CaMKIIα) is a marker for glutamatergic cells [39‐42]. Thus, to target glutaminergic neurons, we used CaMKIIα as a promoter. For fiber photometry recording of the activity of glutaminergic AIC and PIC neurons during itch behaviors, rAAV2/9-CaMKIIα-jGCaMP6s (200 nL, titer: 5.90 × 10¹² genome copies (GC)/mL), or rAAV2/9-CaMKIIα-EGFP (200 nL, titer: 4.05 × 10¹² GC/mL as control) were microinjected into the left AIC (anteroposterior (AP) + 3.72 mm to bregma, mediolateral (ML) ±4.4 mm to midline suture; dorsoventral (DV) –5.6 mm to skull surface) or the left PIC (AP + 0.48 mm, ML + 5.6 mm, DV –7.4 mm) (Fig. 2B).

For pharmacogenetic inhibition of global AIC or PIC neuronal activity, rAAV2/9-hSyn-hM4Di-mCherry (titer: 6.96 × 10¹² GC/mL) or rAAV2/9-hSyn- mCherry (titer: 6.25 × 10¹² GC/mL as control) were bilaterally microinjected into the AIC (Fig. 3B) or PIC, at the above coordinates, 200 nL per side.

For selective pharmacogenetic inhibition of the glutaminergic AIC or PIC neurons, rAAV2/9-CaMKIIα-hM4Di-mCherry (titer: 5.89 × 10¹² GC/mL) or rAAV 2/9-CaMKIIα-mCherry (titer: 5.04 × 10¹² GC/mL as control) were bilaterally microinjected into the AIC or PIC (Fig. 4B) (coordinates as above), 200 nL per side.

For selective inhibition of the glutaminergic AIC neurons with projections to the PrL, either rAAV2/9-EF1α-DIO-hM4Di-mCherry (titer: 2.42 × 10¹² GC/mL) or rAAV2/9-EF1α-DIO-mCherry virus (titer: 2.71 × 10¹² GC/mL as control) was microinjected bilaterally into the AIC, 150 nL per side. Moreover, rAAV2/retro-CaMKIIα-Cre (titer: 6.62 × 10¹² GC/mL) was bilaterally microinjected into the PrL (AP + 3.1 mm, ML ±0.70 mm, DV –4.30 mm), 200 nL per side (Fig. 6B).

After virus injection, the optical fibers (optical fiber with 200 µm core diameter and ceramic ferrules with 2.50 mm diameter, Thorlabs) for photometric recording of neuronal activity were implanted in the brain at 50 µm above the viral injection sites in the unilateral AIC or PIC and were fixed with dental cement. Animals were allowed to recover for 4 weeks.

To record bilateral AIC and PIC glutaminergic neuronal activity during itch, rats received an intradermal injection of 5-HT unilaterally into the nape. During recording, both scratching behaviors and fluorescence signals in the ipsilateral AIC or PIC were simultaneously recorded by a fiber photometry system for 30 min after 5-HT injection. The contralateral fluorescence signals in the AIC or PIC were recorded by repeating this procedure when 5-HT was contralaterally administrated into the nape. A 488 nm laser (blue) was used to excite fluorescence signals. Moreover, a 405-nm laser (purple) was concurrently provided to isolate the movement-corrected signals from the channel. To minimize the bleaching of jGCaMP6s and EGFP, the laser power was adjusted to a low level of 20–40 µW at the tip of the optical fiber.

The fluorescence signals were analyzed offline with the built-in software of the fiber photometry system. The formula (F − F0)/F0 was used to calculate the change of the fluorescence values (ΔF/F), in which F indicates the fluorescence value at each time point during the period from −4 s to 10 s and F0 indexes the baseline median of the fluorescence value during the period from −4 s to −2 s, relative to the onset of each scratching train. To analyze the fluorescence changes across the itch behaviors, the average ΔF/F during the early (1–5 s), and later period (6–10 s), relative to the onset of each scratching train were calculated and finally averaged from all scratching trains for each animal and plotted as heatmaps (Fig. 2).

To pharmacogenetically inhibit the neuronal activity in the AIC or PIC, clozapine-N-oxide (CNO; 4 mg/kg, diluted in 5% DMSO, i.p.) was administered to rats having expressed hM4Di-mCherry or mCherry. Thirty minutes after the CNO injection, behavioral tests for itch, pain, or CPA [20, 21] were conducted.

Before itch-induced scratching tests, a small magnet was attached to one of the rat’s hind limbs. Each rat was placed in a transparent plastic chamber (40 cm long, 40 cm deep, and 60 cm high) and allowed to move freely. The scratching behaviors were monitored by a digital video camera fastened to the roof of the chamber and recorded by a magnetic induction method. At first, a 15-min baseline of scratching behavior was recorded. Then, the rat was gently removed from the chamber and 5-HT (5.00 mmol/L, 50 µL) or compound 48/80 (65 mmol/L, 50 µL) was injected intradermally in the unilateral nape ipsilateral to the hind limb with the small magnet. Scratching behaviors were continuously recorded for 30 min after pruritogen administration. A custom-written software was used to automatically analyze the itch-scratching data. Details and criteria to identify the scratching behaviors have been described previously [38, 43].

During itching, rats repeatedly moved their unilateral hind limb attaching with a small magnet to scratch the injection site. This action induced robust inductive coil voltage fluctuations. A 16-channel amplifier (AM Systems) was used to amplify and digitalize (sampling rate of 1000 Hz) the voltage signals. The amplified signals were bandpass filtered (3−100 Hz) and recorded with a data acquisition system (AD Instruments). Both the cumulative number and duration of scratching trains during the 15 min (baseline) and 30 min (after injection) were calculated for analysis. Data for each 5 min were averaged and analyzed.

Animals were injected with rAAV2/9-CaMKIIα-hM4Di-mCherry or rAAV2/9- CaMKIIα-mCherry (control) into the bilateral AIC. Four weeks after the virus injection, the CPA tests were applied using an unbiased, counterbalanced three-compartment conditioning apparatus and conducted under red light and sound-attenuated conditions [20, 21]. Each chamber had a unique combination of visual properties (one side had black walls, whereas the other side had black and white striped walls). Behavioral activity in each compartment was monitored and recorded with a video camera and analyzed using SA213 Conditioned Place Aversion v 1.0.

On day 5 (post-phase), rats were given CNO (4 mg/kg) to selectively inhibit AIC glutaminergic neurons. Thirty minutes after the CNO injection, the rats were allowed to freely explore all three chambers for 15 min. Durations of stays in each chamber were recorded and analyzed (Fig. 5). The preference for the itch-paired chamber was calculated by the ratio of the time an animal spent in the itch-paired chamber to that in the itch-unpaired chamber.

A digital video camera was used to record pain-related behaviors. At first, the baseline of behavioral activity was recorded for 15 min. To evoke pain-related behaviors, the allyl isothiocyanate (AITC, 10% / 25 µL diluted with 7% Tween80, W203408, Sigma) was administered intradermally into the cheek [44]. Then the behaviors were recorded for 30 min after the AITC injection.

Digital video data were evaluated by two observers blinded to the behavioral procedures. The pain behavior was defined as a caudal-to-rostral wiping movement by the unilateral forelimb across the 5-HT injected location. Grooming with bilateral forelimbs and scratching with the hindlimb were excluded. Wipings are recognized as the effective measurement of pain-related behaviors [45]. The number of wipes was scored by the observers.

A long beam (2 m × 20 mm) with a flat surface was used to examine the changes in motor ability. The beam was placed 50 cm above the floor, with a light at one end to provide aversive stimulation, and a black plastic box at the other end to motivate animals to cross the beam.

The balance beam tests were implemented on three successive days. In the first 2 days, rats were trained to cross the beam three times at minimal 10-min intervals. On day 3, the durations of three successful tests in which the rat crossed the beam without halt were averaged.

The locomotor activity was evaluated by an OFT. Rats were initially placed in the center of the testing box (60 cm × 60 cm × 60 cm) and videotaped individually. The center area was defined as the centric 30 cm × 30 cm. Total travel distance and average speed were recorded in a 10-min period. The movement tracks of the rats were automatically analyzed by the software (XR-XZ301, Xinruan, China).

To detect insular neuronal c-fos expression and verify the expression of target genes, rats were given 5-HT, compound 48/80, or saline by intradermal injection (50 µL) into the unilateral nape before behavioral tests. Sixty minutes after the itch behavioral test or at the end of pharmacogenetic or fiber photometry experiments, rats were deeply anesthetized with an overdose of 3% pentobarbital sodium (i.p.) and perfused transcardiacally with physiological saline, followed by 4% paraformaldehyde (PFA; pH 7.4). The brain was removed and stored in 4% PFA (4°C, 24 h). The brain was then dehydrated successively in 10%, 20%, and 30% sucrose solutions for 24 h, 24 h, and 48 h, respectively. Coronal sections (30 µm thick) were cut on a freezing microtome. The sections were stored in cold PBS (0.01 mol/L, pH 7.4). Sections were placed in PBST (PBS + 0.3% Triton X-100) with 2% normal bovine serum albumin for 1 h, then incubated with primary antibody (4°C, 24 h, rabbit anti-c-Fos 1:500; mouse anti-NeuN 1:500; mouse anti-CaMKIIα 1:100). Then the sections underwent three wash steps for 10 min each in PBST, followed by 2 h incubation with secondary antibody (goat anti-rabbit conjugated to AlexaFluor488, 1:500; goat anti-mouse conjugated to AlexaFluor568, 1:500; goat anti-mouse conjugated to AlexaFluor488, 1:500, Invitrogen). Finally, the sections were washed with PBST (once, 10 min) and incubated for 10 min with DAPI (1:2000, D9542, Sigma-Aldrich). The sections then underwent three more wash steps of 10 min each in PBST, followed by mounting and coverslipping on microscope slides. Confocal fluorescence imaging with the antibodies and the fluorescent proteins were acquired using a slide view VS200 (Olympus, Japan). Coronal sections from 3 rats (2 sections per rat) were examined for statistical analysis.

The extent of target gene expression and the placement of the optical fiber were examined by procedures as described previously [38, 43]. In brief, the deeply anesthetized rats were perfused; their brains were removed, stored, dehydrated, and coronally sectioned at 30 µm thick. The collected sections were washed with PBS (0.01 mol/L, pH 7.4) for 10 min and incubated with DAPI (1:2000) for another 10 min, followed by an additional three wash steps of 10 min each in PBS. Histological images were acquired by an Olympus BX53F fluorescence microscope (Japan) or an Olympus VS200 virtual slide system (Japan).

Data are expressed as the mean ± standard error (SEM). The statistical significance was determined by two-way ANOVA with repeated measures followed by the Tukey post hoc test or by two-tailed unpaired Student’s t-test using the SPSS for Windows package (v. 25.0). A value of P <0.05 was considered statistically significant.

To assess the role of the AIC and PIC in itch processing, we checked the c-fos expression in the AIC and PIC after 5-HT and compound 48/80 administration (Fig. 1). Immunohistochemistry results indicated that the number of c-Fos positive (c-Fos⁺) neurons in the AIC and PIC significantly increased after 5-HT or compound 48/80 injection compared with the saline group (F_{(2, 17)} = 69.935, 5-HT: P <0.001, compound 48/80: P <0.001 for AIC; F_{(2, 17)} = 53.626, 5-HT: P < 0.001, compound 48/80: P <0.001 for PIC).

To further confirm the activation of the AIC and PIC neurons and analyze the temporal relationship between insular neuronal activation and scratching behavior during itch-scratching, fiber photometry of intracellular Ca²⁺ signals in the AIC and PIC glutaminergic neurons was applied during 5-HT-induced itch (Fig. 2A). We microinjected rAAV expressing a Ca²⁺ indicator (jGCaMP6s) or EGFP under control of the CaMKIIα promoter (rAAV2/9-CaMKIIα-jGCaMP6s or rAAV2/9-CaMKIIα- EGFP) into the left AIC or PIC. Optic fibers were implanted above the injection sites (Fig. 2B). Four weeks later, fluorescence signals were recorded after ipsilateral and contralateral injections of 5-HT (2 days). We found that bilateral fluorescent signals of both AIC and PIC glutaminergic neurons significantly increased during 5-HT-induced itching (Fig. 2G and H show the ipsilateral result. P = 0.001 for 1−5 s and P = 0.007 for 6−10 s in AIC; P = 0.813 for 1−5 s and P = 0.011 for 6−10 s in PIC). Notably, the fluorescent signal of the AIC was significantly higher and occurred earlier than that of the PIC (Fig. 2I, P = 0.004 for 1−5 s; P = 0.083 for 6−10 s), and the elevation preceded the scratching behavior by ~1−2 s in the AIC (Fig. 2D), which indicates possible involvement of AIC neuronal activation in itch sensation or the aversive emotion accompanying itching.

Although we had revealed that the AIC and PIC neurons were activated during itch, direct evidence to confirm the role of the insula in itch regulation was still insufficient. To functionally identify the involvement of the AIC and PIC neurons in the modulation of acute itch, we used the pharmacogenetic approach of designer receptors exclusively activated by designer drugs (DREADDs) (Fig. 3A). To conduct the pharmacogenetic manipulations on global AIC or PIC neurons, recombined adeno-associated virus (rAAV) encoding the inhibitory DREADD receptor hM4Di or control (rAAV2/9-hsyn-hM4Di-mCherry or rAAV2/9-hsyn-mCherry as control), were microinjected bilaterally into the AIC or PIC (Fig. 3B). Four weeks after virus injection, the expression of hM4Di-mcherry was confirmed by immunohistochemistry (Fig. 3C, 94% of NeuN-immunopositive cells expressed hM4Di-mcherry). The itch behavioral test was applied after administration of CNO (i.p.), and the OFT and balance beam test were applied to measure the motor coordination after pharmacogenetic inhibition of the AIC or PIC neurons.

The results indicated that global inhibition of the AIC neurons reduced the number and cumulative duration of scratching bouts elicited by 5-HT (Fig. 3D and E, F_{(1, 19)} = 33.262, P <0.001; F_{(1, 19)} = 42.182, P <0.001 for the number of scratching bouts, P <0.001 for the total duration of scratching bouts), but not compound 48/80 (Fig. 3F and G, F_{(1, 20)} = 0.076, P = 0.785 for number of scratching bouts, P = 0.785 for total number for scratching bouts; F_{(1, 20)} = 0.25, P = 0.623 for duration of scratching bouts, P = 0.623 for total duration of scratching bouts).

Global pharmacogenetic inhibition of the PIC neurons did not affect the number and cumulative duration of scratching bouts elicited by 5-HT (Fig. 3H and I, F_{(1, 20)} = 0.602, P = 0.447 for the number of scratching bouts, P = 0.447 for the total number of scratching bouts; F_{(1, 20)} = 4.117, P = 0.056 for the duration of scratching bouts, P = 0.056 for the total duration of scratching bouts), or compound 48/80 (Fig. 3J and K, F_{(1, 15)} = 0.246, P = 0.627 for number of scratching bouts, P = 0.627 for total number for scratching bouts; F_{(1, 15)} = 0.035, P = 0.855 for duration of scratching bouts, P = 0.855 for total duration of scratching bouts). In addition, the pain-related wiping responses were also measured. The results showed that inhibition of global AIC or PIC failed to affect the pain-related wiping (Fig. 7B and C, P = 0.503 for AIC, P = 0.926 for PIC). Pharmacogenetic inhibition of the AIC or PIC neurons did not affect the motor coordination (Fig. 3O and S, P = 0.941 for AIC, P = 0.622 for PIC), and did not significantly affect the distance moved and velocity in the OFT (Fig. 3M–R, P = 0.41 for AIC, P = 0.924 for PIC).

To dissect the involvement of subpopulations of the AIC and PIC neurons in the modulation of acute itch, cell type-specific pharmacogenetic manipulations were used to inhibit glutaminergic neurons in both the AIC and PIC during 5-HT-induced itch (Fig. 4A). The results demonstrated that pharmacogenetic inhibition of AIC glutaminergic neurons reduced the number and the cumulative duration of scratching bouts elicited by 5-HT (Fig. 4D and E, F_(1,17) = 33.368, P <0.001 for number of scratching bouts, P <0.001 for total number for scratching bouts; F_{(1, 17)} = 39.655, P <0.001 for duration of scratching bouts, P <0.001 for total duration of scratching bouts). Nevertheless, inhibition of the PIC glutaminergic neurons had no significant effect on 5-HT-induced scratching behavior (Fig. 4F and G, F_{(1, 21)} = 0.346, P = 0.563 for number of scratching bouts, P = 0.563 for total number for scratching bouts; F_{(1, 21)} = 2.008, P = 0.171 for duration of scratching bouts, P = 0.171 for total duration of scratching bouts).

In addition, selective inhibition of glutaminergic neurons in the AIC, but not PIC, significantly decreased the pain-related wiping by the forelimb elicited by AITC (Fig. 7D and E, P = 0.02 for AIC, P = 0.869 for PIC). Pharmacogenetic inhibition of both AIC and PIC glutaminergic neurons did not affect motor coordination (Fig. 4K and O, P = 0.184 for AIC, P = 0.734 for PIC), and did not significantly affect the distance moved and velocity in OFT (Fig. 4I–N, P = 0.669 for AIC, P = 0.865 for PIC).

To investigate the contributions of glutaminergic AIC neurons to mediating the aversive aspects of itch-scratching, the effects of selective pharmacogenetic inhibition of glutaminergic AIC neurons on CPA behavior were examined (Fig. 5A). During CPA pre-training on day 1, rats were allowed free access to both chambers for 15 min without administration of pruritogens, and animals that had spent similar times in both chambers in the inhibition (AIC::hM4Di) and control (AIC::mCherry) groups were selected for further experiments. After 3 days of CPA with 5-HT administration, control rats exhibited an evident avoidance of the itch-paired chamber, indicated by the shorter time spent in the itch-paired chamber during post-training than that during pre-training (Fig. 5B and D, P = 0.001). Interestingly, rats in the inhibition group did not show an aversive response to the itch-paired chamber, in which they spent time similar to that in the itch-paired chamber during pre- and post-training (Fig. 5C and E, P = 0.476). In addition, the preference for the itch-paired chamber during post-training was significantly lower in the control group than that in the inhibition group (Fig. 5F, P <0.001), suggesting that pharmacogenetic inhibition of glutaminergic AIC neurons effectively blocks the itch-associated CPA during the post-training test.

It has been reported in a series of functional imaging studies that the prelimbic cortex, a subregion of the prefrontal cortex closely associated with cognition, emotion, and motivation, is activated during itching [13, 46, 47]. Pharmacogenetic inhibition of the PrL reduces both 5-HT- and compound 48/80-induced scratching behaviors [31]. Given the verified role of glutaminergic AIC neurons in itch regulation in this study and the close anatomical connection between the AIC and PrL [48], we then assessed the regulatory effect of the AIC-PrL projections during itch processing. To selectively express hM4Di in glutaminergic AIC neurons descending to the PrL, a Cre-dependent AAV (rAAV2/9-EF1α-DIO-hM4Di-mCherry) and a retrograde AAV expressing Cre recombinase (rAAV2/retro-CaMKIIα-Cre) were bilaterally microinjected into the AIC and the PrL, respectively (Fig. 6A). Five weeks after virus injection, an itch behavior test was implemented to examine the effect of these neurons in regulating itch-scratching. The results showed that selective inhibition of the AIC-PrL projection reduced the number and cumulative duration of scratching bouts elicited by 5-HT (Fig. 6C and D, F_(1,19) = 12.847, P = 0.003 for number of scratching bouts, P = 0.002 for total number for scratching bouts; F_(1,19) = 9.787, P = 0.04 for duration of scratching bouts, P = 0.005 for total duration of scratching bouts). Conversely, inhibition of this circuit significantly increased the pain-related wiping with the forelimb elicited by AITC (Fig. 7F, P = 0.014). Pharmacogenetic inhibition of the AIC-PrL projection did not affect motor coordination (Fig. 6H, P = 0.571), and did not significantly affect the distance moved (Fig. 6F, P = 0.738) and velocity (Fig. 6G, P = 0.735) in the OFT test.

Although cumulative studies have identified the insula as a core region related to multiple brain functions and a series of psychiatric and neurological disorders [48‐51], imaging studies on itch have yielded extensive correlations with the insular lobe [12, 16‐18], the role and functional specificity of the IC in itch modulation remain ambiguous. Here, we demonstrated that bilateral inhibition of global AIC neurons or selective inhibition of the glutaminergic neurons in the AIC reduced the scratching behaviors induced by intradermal injection of 5-HT but not those induced by compound 48/80. Nevertheless, both inhibition of global PIC neurons and selective inhibition of the glutaminergic neurons in the PIC failed to affect itching-scratching behaviors induced by both 5-HT and compound 48/80. The current results provide preliminary evidence that the AIC is involved in the regulation of 5-HT-induced itch.

Most imaging studies have indicated that the AIC, but not the PIC, are activated during the itching processes [16‐18]. Given that the immunohistochemistry and fiber photometry results showed activation of PIC neurons during itching in the present study, the involvement of the PIC in itch regulation cannot be excluded. Conservative speculation suggests that the role of the AIC is far more crucial than that of the PIC in itch regulation. It should be noted that neuronal activation of the PIC occurred slightly later than that of the AIC. This differential temporal pattern between the AIC and PIC suggests that they might be involved in different aspects of itch processing.

Differences between AIC and PIC activation during pain processes have been extensively addressed by imaging studies, in which most of the results indicated that pain stimulation activates the AIC rather than the PIC [52‐56]. The current results show that inactivating the glutaminergic neurons in the AIC, but not the PIC, significantly reduced pain responses, supporting the previous imaging findings. However, global inhibition of the AIC in this study failed to affect pain behavior. Since only nociceptive responses to mustard oil were examined, a reasonable interpretation of the AIC involvement in pain needs more evidence.

Recent experimental studies have revealed that specific neuronal subtypes in the PAG and VTA of the midbrain are responsible for mediating negative or positive itch-scratching emotions [20, 21]. Here, we found that pharmacogenetic inhibition of glutaminergic AIC neurons effectively blocked itch-associated CPA behavior, supporting the roles of the insula in processing itch-associated aversive emotion [12, 50, 57‐59]. The insula has been previously acknowledged to be critical in linking pain sensation with aversive emotional responses. Patients with insula lesions can recognize pain but lack an appropriate aversive emotional response [50, 57‐59]. Our current results support the conclusion that the AIC is involved in itch regulation, at least partially, through its mediation of itch-associated aversion. As multiple brain regions, such as the PAG [21], VTA [20], and amygdala [23], are involved in itch-associated aversion, the AIC may function together with these emotion-regulatory brain areas to modulate the negative emotion accompanying itch.

The AIC reciprocally connects with frontal brain regions, including the medial prefrontal cortex (mPFC) [48], a key region implicated in cognitive, emotional, and executive functions. The PFC has also been identified as closely associated with itch regulation in functional imaging studies [12‐18]. Inhibition of the PrL, a subregion of the mPFC, reduced both 5-HT- and 48/80-induced itch in rats [31]. Based on these previous findings and the close anatomical connection between the PrL and AIC [48], we examined the role of the AIC-PrL projection in itch processing. The results revealed that selective inhibition of AIC neurons projecting to the PrL reduced itch scratching but increased pain swiping behaviors, suggesting its differential involvement in the modulation of itch and pain. Unusually, inhibition of the total glutaminergic AIC neurons and the AIC glutaminergic neurons projecting to the PrL showed opposite effects on pain regulation. Perhaps different subpopulations of glutaminergic neurons in the AIC are differentially involved in pain modulation.