Introduction

The spinal dorsal horn represents the entry point into the CNS for somatosensory information from the trunk and limbs. This information is relayed via projection cells to supraspinal sites, where it leads to perceptions, including pain and itch^1–3. However, projection cells represent only ∼1% of dorsal horn neurons, with the vast majority comprising excitatory and inhibitory interneurons that are arranged into local microcircuits that process somatosensory information. Altered function of these circuits contributes to chronic pain and pruritus (itch)^{4, 5}. Dysregulation of inhibitory circuits has attracted particular interest, as broad disruption of spinal inhibitory signalling produces behaviours reminiscent of symptoms in patients suffering from chronic pain or pruritus^6–9.

We have described a molecular classification scheme that assigns the inhibitory interneurons in mouse superficial dorsal horn (SDH; laminae I-II) to five largely non-overlapping populations, based on expression of calretinin (CR), parvalbumin (PV), neuronal nitric oxide synthase (nNOS), dynorphin and galanin (Dyn/Gal) or neuropeptide Y (NPY)¹⁰. This scheme has since been validated and extended by large-scale transcriptomic studies^{11, 12}. These molecularly-defined interneuron populations appear to be functionally distinct as they display differential activation profiles in response to noxious and innocuous stimuli^11–13. A major advantage of this approach is that it allows investigation of the function of different populations through targeted manipulation with techniques such as chemogenetics, optogenetetics and toxin-mediated silencing or ablation. Studies of this type have implicated the PV interneurons in preventing tactile allodynia^{14, 15}, the nNOS interneurons in gating both mechanical and thermal inputs¹⁶, and the Dyn/Gal population in suppressing mechanical pain and pruritogen-evoked itch^{16, 17}. Ablation or silencing of dorsal horn NPY-lineage neurons (i.e. cells that express NPY transiently during development or persistently into adulthood) has been reported to cause spontaneous itching behaviours and enhancement of touch-evoked (mechanical) itch, without affecting pruritogen-evoked itch or pain behaviours¹⁸. This has led to the view that the main function of the NPY cells is suppression of mechanical itch^19–24. This limited role for NPY interneurons is surprising for several reasons: (1) they account for one-third of all inhibitory interneurons in SDH¹⁰, (2) they innervate a population of nociceptive projection cells that belong to the anterolateral system (ALS)^{13, 25–27}, (3) while mechanical itch is restricted to hairy skin, NPY-expressing neurons are present throughout the dorsal horn, including areas innervated from glabrous skin^{10, 28} and (4) NPY itself has a role in modulating chronic pain²⁹. As noted above, the approach used by Bourane et al¹⁸ targeted a broad population of inhibitory interneurons that express NPY during development, as well as those that express NPY in adulthood. Tashima et al³⁰ recently attempted to target NPY-INs by injecting adeno-associated viruses (AAVs) with a Npy promoter into the rat spinal cord. However, expression was largely restricted to lamina IIo (even though many NPY cells are found in other laminae) and fewer than half of the targeted cells contained either NPY or its mRNA. Therefore, the role of those dorsal horn interneurons that continue to express NPY (NPY-INs) remains unclear.

Here we use intraspinal injections of AAVs carrying Cre-dependent expression cassettes into young adult NPY::Cre mice to target dorsal horn NPY-INs. We demonstrate that this technique can be used to manipulate inhibitory interneurons that express NPY in adulthood, while avoiding those cells that transiently express NPY during development. We show that chemogenetic activation of dorsal horn NPY-INs suppresses acute mechanical and thermal nocifensive behaviours, as well as those resulting from pruritogen-evoked itch, and reduces activity in spinal networks that process nociceptive and pruritoceptive information. Furthermore, NPY- IN activation abolishes mechanical and thermal hypersensitivity in models of inflammatory and neuropathic pain. Finally, we show that silencing of NPY-INs results in spontaneous itch and an exaggerated response to pruritogens, and that this depends on a circuit involving GABAergic input from NPY-INs to excitatory interneurons that express the gastrin-releasing peptide receptor (GRPR). Together these results demonstrate that dorsal horn NPY-INs have a far broader role than previously suggested, since they gate transmission of nociceptive and pruriceptive information. They therefore represent a potential target for the development of new treatments for pain and itch.

Results

Cre-dependent AAV injections in young adult NPY::Cre mice target dorsal horn inhibitory NPY interneurons and avoid transient NPY-expressing cells

We initially assessed the suitability of using intra-spinal injection of AAVs encoding Cre-dependent constructs in RH26 NPY::Cre mice (the line used by Bourane et al¹⁸) to target NPY-INs in the dorsal horn. We first performed RNAscope fluorescent in situ hybridisation (FISH) to compare Cre and Npy mRNA expression in lumbar spinal cord sections from young adult NPY::Cre mice. Across laminae I-III 91.6% ± 0.3% of cells classed as Cre-positive cells were also Npy-positive, and these accounted for 62.1% ± 0.6% of Npy-positive cells, demonstrating that Cre expression in the NPY::Cre line faithfully captures the majority of NPY-INs in the adult dorsal horn (Figures 1A and 1B). Accordingly, injection of either AAV.flex.tdTomato (tdTom) or AAV.flex.eGFP (both serotype 1 with CAG promoter, see Key Resources Table) into the lumbar dorsal horn of adult NPY::Cre mice (Figure 1C) resulted in fluorescent protein (FP) expression matching that previously reported for NPY neurons^{10, 26}, with cell bodies concentrated in laminae I-III (Figure 1D). The great majority of FP-expressing neurons in laminae I-III were immunoreactive (IR) for NPY (78.5% ± 3.6%), and these accounted for 74.6% ± 1.9% of the NPY-IR neurons in this area (Figures 1D and 1E).

Figure 1.

We then crossed NPY::Cre mice with the Cre-dependent reporter line Ai9 (to label all NPY lineage neurons with tdTomato) and injected AAV.flex.eGFP into the lumbar dorsal horn of these mice, to target cells that continued to express NPY (Figures 1F and Figure 1 – figure supplement 1A). In these animals, tdTom-positive cells were seen throughout the dorsal horn, and were much more numerous than eGFP- expressing cells in the region of the injection site (Figure 1G). All tdTom-labelled cells were immunoreactive for the transcription factor Pax2 (Figure 1 – figure supplement 1B), which is expressed by all dorsal horn inhibitory neurons in rodents^{7, 31}, and these accounted for 40.8% ± 10.0% of Pax2 cells in laminae I-III. Virtually all eGFP-expressing neurons within this region were tdTom-positive (97.8% ± 0.2%), but these only accounted for 51.1% ± 3.5% of the tdTom-positive population (Figures 1H and 1I). In agreement with the data presented above, the great majority tdTom+;eGFP+ neurons were found to be NPY-IR (85.2% ± 0.2%), and these accounted for 67.0% ± 6.3% of the NPY-IR cells in laminae I-III. In contrast, only 32.1% ± 2.1% of the tdTom+;eGFP-negative cells displayed NPY immunoreactivity, corresponding to just 24.1% ± 2.7% of all NPY-IR interneurons (Figures 1H and 1I). Overall, 58.3% ± 0.9% of tdTom+ neurons were NPY-immunoreactive.

These results suggest that transient Cre expression occurs in a broad population of dorsal horn inhibitory interneurons in NPY::Cre mice, presumably driven by NPY expression during development. However, Cre expression in adult mice occurs in a much more restricted population of inhibitory interneurons that express NPY persistently. To characterise these cells in relation to the neurochemical populations of inhibitory interneurons that we had identified in the SDH¹⁰, we focussed our analysis on laminae I-II and compared the expression of the different neurochemical markers between the tdTom+;eGFP+ and tdTom+;eGFP-negative cells. Within this region, 85.8% ± 2.5% of tdTom+;eGFP+ cells were NPY-IR, while approximately 10%, 3% and 1% expressed galanin, nNOS or PV, respectively (Figure 1 – figure supplement 1C). This is in good agreement with the degree of overlap between these markers and NPY that we have previously described¹⁰. Surprisingly, 23.7% ± 7.0% of tdTom+;eGFP+ cells expressed CR (Figure 1 – figure supplement 1C), which has previously been reported to show minimal overlap with NPY-IR³². Nonetheless tdTom+;eGFP+ cells are largely restricted to the NPY+ population of inhibitory interneurons. In contrast, tdTom+;eGFP-negative neurons were much more broadly spread across four of the populations, with approximately 28% expressing NPY, 44% expressing CR and 24% each expressing galanin or nNOS, although virtually none (1.4% ± 1.4%) expressed PV (Figure 1 – figure supplement 1C).

Because we intended to use Cre-dependent expression of the excitatory DREADD hM3Dq^{7, 16} to activate NPY-INs, we also assessed targeting of this receptor to the appropriate interneurons following injection of AAV.flex.hM3Dq-mCherry into the lumbar dorsal horn of adult NPY::Cre mice (Figure 1 – figure supplement 1D). hM3Dq-mCherry-expressing cells in the SDH displayed a near-identical pattern in terms of co-localisation with inhibitory interneuron markers to that of the eGFP+ cells following AAV.flex.eGFP injection (Figure1 – figure supplement 1E and 1F). For the mCherry+ cells, 90.5% ± 6.2% co-expressed NPY (accounting for 66.1% ± 4.8% of all NPY interneurons) and there was little or no overlap with the galanin, nNOS and PV populations (Figure 1 – figure supplement 1E and 1F). Again, significant overlap was observed between hM3Dq-mCherry-and CR-positive cells, with 28.0% ± 1.5% of mCherry-labelled cells displaying CR-IR. NPY-IR was detected in 93.1% ± 3.7% of mCherry+;CR+ cells in sections co-stained with NPY antibody (Figure 1 – figure supplement 1E), demonstrating that this represents NPY::Cre-driven recombination in interneurons co-expressing NPY and CR, rather than ectopic recombination in CR+;NPY-negative interneurons. The level of NPY-IR in the CR cells was generally very low (Figure 1 – figure supplement 1E), which probably explains why this overlap was not detected previously³². As expected, virtually all mCherry-labelled cells were inhibitory (97.4% ± 2.6% Pax2-positive), and these accounted for a quarter of all inhibitory interneurons in the SDH (Figure 1 – figure supplement 1F). Crucially, no mCherry-labelled cells were observed in the ipsi- or contralateral L3, L4 or L5 DRG of four AAV.flex.hM3Dq-mCherry-injected NPY::Cre mice (data not shown), as would be expected from the lack of NPY expression in uninjured mouse DRG neurons³³.

Collectively these results demonstrate that injection of AAVs encoding Cre-dependent constructs into the dorsal horn of adult NPY::Cre mice allows specific targeting of most inhibitory interneurons that persistently express NPY, and avoids capturing a broader population of inhibitory interneurons that express NPY transiently during development.

Activation of inhibitory NPY interneurons reduces activity in dorsal horn circuits recruited by nociceptive and pruritic stimuli

We initially assessed the efficacy of our chemogenetic strategy to activate NPY- INs in NPY::Cre mice injected with AAV.flex.hM3Dq-mCherry by comparing expression of the activity marker Fos two hours after administration of the hM3Dq ligand clozapine-N-oxide (CNO) or vehicle (Figure 2 – figure supplement 1A). Only 8.8% ± 4.7% of mCherry-labelled cells in laminae I-III displayed Fos-IR in vehicle-treated animals, but this rose dramatically to 82.9% ± 2.5% in CNO-treated mice (Figure 2 – figure supplement 1B and 1C). Of these mCherry+;Fos+ cells, 87.5% ± 3.1% displayed detectable NPY immunoreactivity, and these accounted for 53.9% ± 1.2% of all NPY-IR neurons (Figure 2 – figure supplement 1D and 1E). Surprisingly, we also observed a small but significant increase in the proportion of mCherry-negative cells that expressed Fos following CNO (from 2.7% ± 0.3% in vehicle-treated to 5.8% ± 1.0% in CNO-treated mice), and this increase was entirely restricted to inhibitory interneurons (Figure 2 – figure supplement 1C). A significant proportion of these mCherry-negative;Fos+ cells were also NPY-IR (31.3% ± 4.8%; Figure 2 – figure supplement 1E), and these may represent NPY-INs that express hM3Dq-mCherry at a level sufficient for direct CNO-mediated activation, but that is too low for immunohistochemical detection. Alternatively, they may have been indirectly recruited via disinhibition following CNO-mediated activation of hM3Dq- mCherry-expressing NPY-INs. Overall, CNO treatment resulted in Fos expression in 65.9% ± 2.8% of all NPY-INs, and these comprised 65.5% ± 3.8% of Fos-expressing cells (Figure 2 – figure supplement 1E). In summary, intraspinal injection of AAV.flex.hM3Dq-mCherry into adult NPY::Cre mice allows chemogenetic activation of two thirds of dorsal horn NPY-INs, and a small proportion of other dorsal inhibitory interneurons.

Figure 2.

We then assessed the ability of dorsal horn NPY-INs to suppress the transmission of pain- and itch-related information at the circuit level. NPY::Cre mice that had had intraspinal injections of AAV.flex.hM3Dq-mCherry were injected with vehicle or CNO, and then received a noxious heat (hindpaw immersion in 52⁰C water) or pruritic (intradermal injection of chloroquine, CQ, in the calf) stimulus ipsilateral to the viral injection under brief general anaesthesia (Figures 2A and 2B). Mice that received the pruritic stimulus were fitted with an Elizabethan collar to prevent Fos induction due to itch-related biting of the leg. Following a two-hour survival period, spinal cord sections were processed for Fos-IR. In vehicle treated animals, the noxious heat and pruritic stimuli resulted in ipsilateral Fos expression in the somatotopically relevant areas of the dorsal horn. Fos+ cells were particularly clustered in the medial half of the SDH following noxious heat, and the middle third of the SDH after CQ, as previously reported^{34, 35} (Figures 2C and 2D). Accordingly, analysis of Fos expression was performed within these regions of the SDH. In CNO-treated mice there was a clear increase in the proportion of mCherry-labelled cells expressing Fos (Figures 2C-2F), presumably due to direct chemogenetic activation of these cells (as described above). However, for both the noxious heat and pruritic stimuli, we observed a significant decrease in the proportion of mCherry-negative cells that were Fos-positive, when compared to vehicle-treated mice (noxious heat: vehicle = 26.7% ± 4.7% vs. CNO = 6.7% ± 1.2%; CQ injection: vehicle = 12.9% ± 4.2% vs. CNO = 2.1% ± 0.3%) (Figures 2C-2F). In both cases the decrease was largely restricted to Pax2-negative (excitatory) neurons, although a significant decrease was also observed in Pax2-positive cells in heat-stimulated mice, with a similar trend in CQ- stimulated mice (Figures 2E and 2F).

These results demonstrate that activation of NPY-INs inhibits neurons that are normally recruited by noxious or pruritic stimuli in dorsal horn circuits. To investigate the nature of this inhibition in more detail, we performed optogenetic experiments in spinal cord slices from NPY::Cre mice that had received intraspinal injections of AAV.flex.ChR2-eYFP, resulting in expression of eYFP-tagged channelrhodopsin in NPY-INs (Figure 2G). Short pulses of blue light reliably evoked inward currents and action potential firing in all (10/10) eYFP-ChR2+ cells tested (Figure 2 – figure supplement 2A-C). Recordings were made from 41 ChR2-eYFP-negative cells in the SDH (Figures 2G and Figure 2 – figure supplement 2D), with an optogenetically-evoked postsynaptic current (oPSC) being seen in 29 cells (70.7%). In 7 of the cells with oPSCs, bath application of the AMPAr and NMDAr antagonists, NBQX and D- APV, respectively, did not alter the peak amplitude of the current (baseline = −779.8 pA ± 267.7 pA vs. NBQX / D-APV = −756.6 pA ± 285.8 pA, p = 0.578, Wilcoxon matched-pairs signed rank test) (Figure 2 – figure supplement 2E and 2F), demonstrating that these were not mediated by glutamate and were therefore optogenetically-evoked IPSCs (oIPSCs). The GABAergic / glycinergic nature of these oIPSCs was investigated by bath application of gabazine and strychnine, respectively (in the presence of NBQX and D-APV). All oIPSCs tested (6/6) were reduced by gabazine, but not strychnine (Figures 2H and 2I), indicating that inhibition is predominantly mediated by GABA. This is consistent with the finding that NPY neurons in laminae I-III are all GABA-IR, but are not enriched with glycine³⁶. Taken together, these findings demonstrate that NPY-INs provide a powerful GABAergic inhibitory input to surrounding dorsal horn neurons and can reduce the activation of excitatory neurons that are normally recruited by noxious or pruritic stimuli, suggesting that when activated they supress the transmission of pain-and itch-related information in the dorsal horn.

Activation of inhibitory NPY interneurons increases acute nocifensive reflex thresholds and reduces pruritogen-evoked itch behaviour

We then looked for behavioural correlates of the NPY-IN-mediated suppression of dorsal horn pain and itch circuits. To do this, we assessed mechanical and thermal nocifensive reflexes, as well as CQ-induced itch behaviour, in NPY::Cre mice that had received unilateral spinal injections of AAV.flex.hM3Dq-mCherry and were then treated with vehicle or CNO (Figure 3A). In vehicle-treated mice, as expected, there were no significant differences between the hindpaws contralateral and ipsilateral to the AAV injection for the 50% mechanical withdrawal threshold (MWT) or for withdrawal latencies to noxious heat or cold. However, in CNO-treated mice nocifensive thresholds/latencies in the ipsilateral paw were significantly increased across all three modalities, demonstrating a generalised anti-nociceptive effect of NPY-IN activation (Figures 3B-3D). CNO-treated mice also spent significantly less time biting the calf area in the 30 minutes following CQ injection than vehicle-treated controls, demonstrating a reduction in pruritogen-evoked itch upon NPY-IN activation (Figure 3E). It has been proposed that when used at high doses, systemic CNO may have off-target effects as a result of conversion to clozapine³⁷. We therefore tested the effect of 5mg/kg CNO (the dose used throughout our study) on naïve wild-type mice, and found no change in mechanical or thermal nocifensive thresholds, or on locomotor performance (Figure 3 – figure supplement 1). This confirms that the effects observed in AAV.flex.hM3Dq-mCherry-injected NPY::Cre mice are due to DREADD activation, and not the result of off-target effects of CNO. These findings show that chemogenetic activation of dorsal horn NPY-INs has a broad anti-nociceptive effect across a range of modalities and suppresses pruritogen-evoked itch.

Figure 3.

Activation of NPY interneurons blocks mechanical and thermal hypersensitivity in models of inflammatory and neuropathic pain

We next assessed the effects of chemogenetically activating NPY-INs in the context of inflammatory and neuropathic pain (Figures 4A, 4D and 4G). Intraplantar CFA resulted in punctate mechanical and heat hypersensitivity of the ipsilateral paw of AAV.flex.hM3Dq-mCherry-injected NPY::Cre mice that received i.p. injection of vehicle prior to behavioural testing. However, the mechanical and heat hypersensitivity were completely blocked in mice treated with CNO, and the heat latencies were significantly increased above the pre-CFA baseline values (Figure 4B and 4C). Vehicle-treated AAV.flex.hM3Dq-mCherry-injected NPY::Cre mice that had undergone spared nerve injury (SNI) also displayed mechanical and heat hypersensitivity of the ipsilateral paw, compared to pre-surgery thresholds. Both the mechanical and heat hypersensitivity were blocked in CNO-treated mice (Figures 4E and 4F). Because de novo expression of NPY is known to occur in injured A-fibre afferents following nerve injury^{33, 38–40}, this could result in expression of hM3Dq in these afferents, thus confounding interpretation of our results. We therefore quantified the number of mCherry-labelled cells in the somatotopically-relevant L4 and L5 DRG four weeks following SNI surgery in 4 mice (Figure 4 – figure supplement 1). As expected, no labelled cells were observed contralateral to the AAV injection and SNI surgery in either DRG in any of the mice. A few mCherry-labelled cells were observed in both L4 and L5 DRG on the ipsilateral side (cells per DRG: L4 = 12.5 ± 2.3, L5 = 17.5 ± 3.6; Figure 4 – figure supplement 1). Because the numbers of A-fibre sensory neurons within the mid-lumbar DRG are estimated to be in the thousands in mice^{41, 42}, it is highly unlikely that CNO-mediated activation of the very few hM3Dq-expressing cells observed in the L4 and L5 DRG following SNI would contribute to the blockade of neuropathic pain that we observed. We therefore conclude that this effect is due to activation of spinal inhibitory NPY-INs. We also assessed mCherry expression in the L4 and L5 DRG of 5 CFA-treated AAV.flex.hM3Dq-mCherry-injected NPY::Cre mice, 3 days following CFA injection. In contrast to nerve injury, neuropeptide upregulation is not observed in rodent DRG under inflammatory conditions^{33, 40}. As expected, we observed no mCherry-labelled cells in the contra- or ipsilateral L4 or L5 DRG of these mice (data not shown).

Figure 4.

Spinal NPY signalling has been implicated in the suppression of neuropathic pain through inhibition of NPY Y1 receptor (Y1R)-expressing excitatory interneurons in the dorsal horn^{29, 43, 44}. Therefore the suppression of neuropathic hypersensitivity that we observed during chemogenetic activation of NPY-INs could be due to GABAergic transmission, NPY signalling, or a combination of both. To assess the potential role of Y1R signalling, we systemically co-administered CNO and the Y1R-selective antagonist BMS 193885²¹ prior to behavioural testing in AAV.flex.hM3Dq-mCherry-injected NPY::Cre mice that had undergone SNI surgery. Administration of the Y1R antagonist had no effect on the CNO-mediated suppression of tactile and heat hypersensitivity in these mice (Figures 4E and 4F), suggesting that action of NPY on Y1 receptors is not required for this effect.

In addition to evoked hypersensitivity, peripheral nerve injury induces ongoing neuropathic pain in rodents, as well as engaging affective-emotional responses to pain⁴⁵. To determine the contribution of NPY-INs to ongoing pain we tested whether CNO induced conditioned place preference (CPP) in a separate cohort of AAV.flex.hM3Dq-mCherry-injected NPY::Cre mice following SNI surgery (Figure 4G). A wildtype control group that had undergone SNI was also included to test for any possible preference of (or aversion to) the effects of CNO that could have resulted from off-target effects independent of DREADD activation. CNO did not induce preference or aversion in either of these experimental groups (Figures 4H and 4I, Figure 4 – figure supplement 2A and 2B). However, using the same experimental setup we observed preference for a chamber paired with gabapentin in mice that had undergone SNI (Figure 4 – figure supplement 2C and 2D), showing that the CPP method was sufficiently sensitive to detect ongoing neuropathic pain. Together, these findings suggest that activating NPY-INs may not alleviate ongoing pain in the SNI model. We also assessed SNI-induced cold hypersensitivity in the cohort of AAV.flex.hM3Dq-mCherry-injected NPY::Cre mice that were used for CPP testing (Figure 4G). We observed a marked increase in the duration of the response to an acetone droplet applied to the ipsilateral hindpaw relative to the pre-SNI baseline when the mice had been dosed with a vehicle control. Administration of CNO completely blocked this hypersensitivity, demonstrating a reversal of SNI-induced cold allodynia when NPY-INs are activated. This result also demonstrates that the lack of CPP in the chemogenetic experiments was not due to a failure to activate NPY-INs.

In summary, chemogenetic activation of NPY-INs supresses both mechanical and thermal hypersensitivity in models of inflammatory and neuropathic pain, and the suppression of neuropathic hypersensitivity appears to be mediated predominantly by GABAergic transmission from NPY-INs. However, NPY-IN activation does not appear to affect ongoing pain in the neuropathic model.

Toxin-mediated silencing of NPY interneurons causes spontaneous itch and enhances pruritogen-evoked itch but does not alter nocifensive reflexes

We then tested whether tetanus toxin light chain (TeLC)-mediated silencing of NPY-INs following spinal injection of AAV.flex.TeLC.eGFP into NPY::Cre mice altered pain- or itch-related behaviours (Figure 5A and Figure 5 – figure supplement 1A). Immunohistochemical assessment of the overlap of NPY and GFP expression in these mice demonstrated a very similar specificity and efficacy of expression in NPY-INs to that described above for other viral constructs. In animals injected with AAV.flex.TeLC.eGFP 82.6% ± 5.0% of GFP-positive cells co-expressed NPY and 61.7% ± 3.5% of NPY-positive cells co-expressed GFP (Figure 5 – figure supplement 1F and 1G).

Figure 5.

Compared to AAV.flex.eGFP-injected controls, AAV.flex.TeLC.eGFP-injected NPY::Cre mice displayed significant enhancement of CQ-induced itch when tested 4 to 6 days after AAV injection (P<0.0001, 2-way ANOVA with Tukey’s post-test, Figure 5B). Approximately two thirds of AAV.flex.TeLC.eGFP-injected mice also developed skin lesions on the ipsilateral hindlimb, within the corresponding dermatomes, by day 7 (Figures 5D and 5E). This phenotype was never observed in the AAV.flex.eGFP-injected controls, and strongly suggests development of spontaneous itch following silencing of NPY-INs. Consistent with this interpretation, we observed a significant increase in the time spent biting the calf prior to CQ injection in AAV.flex.TeLC.eGFP-injected NPY::Cre mice compared to AAV.flex.eGFP-injected mice of the same genotype (P=0.0014, 2-way ANOVA with Tukey’s post-test, Figure 5C). In contrast to these marked effects on itch-related behaviours, silencing of NPY-INs did not significantly alter punctate tactile or thermal nocifensive thresholds at 4-6 days after AAV injection (Figure 5 – figure supplement 1A-1D). Motor co-ordination, as assessed by rotarod, was also unaffected by NPY- IN silencing; however, a small but significant improvement was detected in AAV.flex.eGFP-injected mice relative to their baseline pre-surgery performance, most likely reflecting a mild training effect (Figure 5 – figure supplement 1E). Taken together these findings indicate that tonic activity of NPY-INs suppresses itch, but has no obvious impact on nociceptive thresholds.

Increased itch caused by silencing NPY interneurons operates through a circuit involving GRPR-expressing excitatory interneurons

Several studies have shown that GRPR-expressing excitatory dorsal horn interneurons (GRPR-INs) are crucial for pruritogen-evoked itch^46–48, while it has been proposed that they are not required for mechanical itch^{18, 21} (but see Chen et al²³). To assess whether signalling through GRPR-INs was required for the itch-related behaviours that we observed when NPY-INs were silenced, we crossed NPY::Cre and GRPR^CreERT2 mice and concomitantly silenced NPY-INs and GRPR-INs through spinal injection of AAV.flex.TeLC.eGFP (Figure 5A). AAV.flex.eGFP-injected mice of the same genotype were again used as a control group. NPY::Cre;GRPR^CreERT2 mice that received injections of AAV.flex.TeLC.eGFP showed no significant difference in CQ-induced itch, compared to AAV.flex.eGFP-injected controls (P=0.34, 2-way ANOVA with Tukey’s post-test, Figure 5B). However, when comparing NPY::Cre and NPY::Cre;GRPR^CreERT2 mice that had received injections of AAV.flex.TeLC.eGFP, we found that the NPY::Cre;GRPR^CreERT2 mice showed significantly less CQ-induced itch behaviour than NPY::Cre mice (P<0.0001, 2-way ANOVA with Tukey’s post-test, Figure 5B). Furthermore, AAV.flex.TeLC.eGFP-injected NPY::Cre;GRPR^CreERT2 mice did not display a significant increase in spontaneous biting prior to CQ administration (compared to AAV.flex.eGFP-injected controls; P=0.82, 2-way ANOVA with Tukey’s post-test, Figure 5C) and never developed skin lesions (Figures 5D and 5E). These data demonstrate that both the spontaneous itch and the increased pruritogen-evoked itch observed following silencing of NPY-INs are at least partly transmitted via GRPR-INs.

This led us to ask whether NPY-INs provide direct inhibitory synaptic input to GRPR-INs. To investigate this we performed ex vivo patch clamp experiments in spinal cord slices from NPY::Cre;GRPR^FlpO mice that had received intraspinal injections of AAV.flex.ChR2-eYFP together with AAV.FRT.mCherry, resulting in expression of eYFP-tagged channelrhodopsin in NPY-INs and mCherry in GRPR- INs (Figure 5F). Recordings were made from 11 mCherry+ cells and all of these exhibited an oIPSC (with no failures) when the slice was illuminated with brief pulses of blue light, with a mean peak oIPSC amplitude of 250.1 pA ± 58.9 pA (Figure 5G). In 3 / 3 of these cells oIPSCs were abolished by the application of TTX and rescued by the addition of 4-AP (Figure 5H), confirming that they were monosynaptic, and therefore that NPY-INs directly inhibit GRPR-INs. The GABAergic / glycinergic nature of the oIPSCs was assessed in three of the cells (Figure 5 – figure supplement 2A-2C). In 2 / 3 cells the oIPSC was gabazine sensitive / strychnine insensitive, indicating GABA-mediated inhibition, while in the other cell the oIPSC was sensitive to both gabazine and strychnine, indicating mixed GABA and glycine inhibition (Figure 5 – figure supplement 2B and 2C). Although Acton et al²¹ provided evidence that GRPR-INs lack the Y1R, Chen et al²³ reported that 35% of GRPR cells had Y1R mRNA. We therefore tested the effect of bath-applying the Y1R agonist [Leu³¹,Pro³⁴]-neuropeptide Y, while recording from GRPR-INs (Figure 5 – figure supplement 2D). We found that all 10 cells tested failed to show an outward current in response to [Leu³¹,Pro³⁴]-neuropeptide Y (Figure 5 – figure supplement 2E and 2F), suggesting that NPY acting on the Y1R is unlikely to have made a significant contribution to the suppression of GRPR cells by NPY-INs. We also investigated inhibitory NPY-IN input to GRPR-INs anatomically. To do this, we quantified the proportion of inhibitory synaptic contacts onto GRPR-INs (labelled through spinal injection of Cre-dependent AAV-Brainbow2 into GRPR^CreERT2 mice; Figure 5I) at which NPY was present in the presynaptic bouton. Inhibitory synapses were identified by the presence of VGAT-positive presynaptic boutons apposed to puncta of the postsynaptic protein gephyrin. Many of the gephyrin puncta on the GRPR-INs were contacted by NPY-IR boutons, and these accounted for 45.0% ± 1.9% of all inhibitory synapses on the GRPR-INs (Figures 5J and 5K). This was significantly higher than the proportion of inhibitory (VGAT+) boutons in the vicinity of the analysed cells that contained NPY (36.2% ± 2.6%; Figure 5K). Together these data provide strong evidence that NPY-INs selectively target GRPR-INs and generate a powerful GABAergic inhibition of these cells.

Toxin-mediated silencing of dynorphin interneurons enhances pruritogen-evoked itch

Dorsal horn inhibitory interneurons that co-express dynorphin and galanin have been implicated in suppression of itch as their activation reduces pruritogen-evoked itch^{16, 22}, while constitutive loss of B5-I neurons, which include this population, results in enhanced pruritogen-evoked itch and skin lesions due to spontaneous scratching^{49, 50}. In addition, Brewer et al⁵¹ reported that chemogenetic inhibition of dynorphin lineage cells increases pruritogen-evoked itch. Given the well-established role of dynorphin-expressing interneurons (Dyn-INs) in suppressing itch, we compared the effect of silencing these cells with that of silencing the NPY-INs, in order to explore a potential overlap of function. Although dynorphin is also expressed by a subset of dorsal horn excitatory interneurons, we have shown that these are largely restricted to areas innervated by afferents from glabrous skin¹⁶. We injected AAV.flex.TeLC.eGFP (or AAV.flex.eGFP as a control) into the dorsal horn of the L3 segment of Pdyn^Cre mice for these experiments (Figure 6A). This segment was chosen for 2 reasons: (1) it receives input from the region of calf that we used to test the effect of pruritogens, and (2) it receives input exclusively from hairy skin, and therefore the great majority of virally-transfected Dyn-INs are likely to be inhibitory cells (those that co-express dynorphin and galanin)¹⁶.

Figure 6.

In contrast to the effects of silencing NPY-INs, silencing of Dyn-INs never resulted in skin lesions, and the time spent biting the calf prior to CQ injection appeared to be unaffected (Figure 6B; P=0.7431, 2-way repeated measures ANOVA with Šidák’s post-test). This suggests that silencing Dyn-INs does not result in spontaneous itch. Silencing of Dyn-INs did result in enhanced CQ-evoked itch, when compared to AAV.flex.eGFP-injected controls (Figure 6B; P=0.0379, 2-way repeated measures ANOVA with Šidák’s post-test). However, this enhancement was markedly less pronounced than that observed following silencing of NPY-INs (compare Figure 6B with Figure 5B). To confirm that the effects of TeLC-mediated silencing resulted from targeting of Cre-expressing cells, we also injected either AAV.flex.TeLC.eGFP or AAV.flex.eGFP into the L3 segments of wild-type mice, and assessed CQ-evoked itch behaviours 4-6 days later (Figure 6 – figure supplement 1A). As expected in these control animals there was no significant difference in the time spent biting the calf between AAV.flex.TeLC.eGFP- or AAV.flex.eGFP-injected mice either before or after injection of CQ (Figure 6 – figure supplement 1B).

These findings suggest that while both NPY- and Dyn-INs can suppress itch, the NPY population has a more substantial role in this mechanism. One explanation for this could be that although Dyn-INs form inhibitory synapses onto GRPR cells²², the density of these synapses is less than that of those arising from the NPY-INs. To test this, we assessed contacts from inhibitory dynorphin cells onto GRPR-INs (Fig 6C-6E), using an antibody against dynorphin B (DynB) and found that these constituted only 20.9% ± 2.2% of the inhibitory synapses on these cells. This did not differ significantly from the proportion of inhibitory boutons that contained DynB in the vicinity of the analysed cells (22.6% ± 1.6%; Figures 6D and 6E). Collectively these results suggest that unlike NPY-INs, Dyn-INs do not preferentially target GRPR-INs. In addition, they contribute a far lower proportion of inhibitory synapses on the GRPR-INs (compared to the NPY-INs), and loss of this input has a much less dramatic effect on itch.

Discussion

Inhibitory interneurons in the SDH play an important role in suppressing pain and itch. NPY-expressing cells constitute around a third of the inhibitory neurons in this region and are also present in deeper laminae. Previous studies in which NPY- lineage neurons were ablated demonstrated that these cells are responsible for preventing mechanical itch through a mechanism involving NPY and the Y1 receptor^{18, 21}. Here we show, by selectively activating those cells that continue to express the peptide, that the NPY cells inhibit acute nocifensive reflexes and reduce mechanical and thermal hypersensitivity in both inflammatory and neuropathic pain models. In addition, they strongly suppress itch evoked by chloroquine. Silencing the NPY cells causes spontaneous itch and exaggerated responses to chloroquine, and both of these effects are reduced by simultaneously silencing GRPR-expressing excitatory interneurons, indicating that suppression of itch by the NPY cells operates through downstream GRPR-INs.

A broad inhibitory role for NPY cells

Our findings indicate that NPY-INs have a far broader role in suppressing pain-and itch-related behaviours than had been suggested by previous studies that used the same NPY::Cre line^{18, 20, 21}, despite the fact that we were targeting a more restricted neuronal population. The differences in experimental findings are likely to result from two methodological issues: (1) the technique used to target cells, and (2) the use of loss-of-function or gain-of-function approaches. In each of these other studies, cells were targeted by an intersectional genetic approach that limited expression to spinal cord and brainstem, but would have included a large additional group of inhibitory neurons that expressed NPY only during development¹⁸. Here, we used an alternative strategy to restrict expression: intraspinal injection of AAVs coding for Cre-dependent constructs^{7, 16}. While this approach failed to capture a minority of NPY-expressing neurons, it enabled us to target a large number of these cells. Importantly, expression was restricted to those cells that continue to express NPY. This was confirmed by our finding that up to 85% of the virally transfected cells contained detectable levels of NPY.

The main differences in interpreting the roles of NPY cells are likely to depend on whether the cells were inactivated (through ablation or synaptic silencing) or chemogenetically activated. In agreement with Bourane et al¹⁸, we found that silencing NPY cells had no effect on acute nociceptive thresholds. However, chemogenetically activating these cells increased thresholds for both thermal and mechanical nocifensive reflexes, and reduced hypersensitivity in neuropathic and inflammatory pain models. Interestingly, Acton et al²¹ also observed an anti-nociceptive effect on mechanical stimuli when they chemogenetically activated NPY- lineage neurons, but attributed this to ectopic activation of Y1 receptors on primary sensory neurons. However, although Y1 is present in cell bodies of some primary sensory cells, it is not thought to traffic to their central terminals^{52, 53}. Acton et al²¹ did not test whether activating NPY-lineage neurons had any effect on responses to thermal stimuli, or on neuropathic/inflammatory hypersensitivity, so it is not possible to compare our findings in these contexts. The most likely explanation for discrepancies between the findings of loss-of-function and gain-of-function studies is that although NPY cells have an antinociceptive action, other interneurons provide sufficient inhibition to maintain nocifensive reflexes when NPY cells are silenced. Our findings therefore indicate that NPY-INs have a far broader role in somatosensory processing than was previously recognised.

NPY cells suppress spontaneous and pruritogen-evoked itch

Our previous studies^{16, 49} had implicated dynorphin/galanin cells in suppression of pruritogen-evoked itch. This was based on the findings that Bhlhb5^-/- mice (which lack these cells) show exaggerated responses to pruritogens⁴⁹, and that chemogenetic activation of Dyn-INs suppressed CQ-evoked itch¹⁶. In support of this, Liu et al²² subsequently showed that activating galanin-expressing cells also suppresses pruritogen-evoked itch. This anti-pruritic action is likely to involve dynorphin acting on κ-opioid receptors⁴⁹ as well as direct inhibition of GRPR cells (which are an integral part of the spinal itch pathway) by GABA and/or glycine released from the dynorphin/galanin cells²². Liu et al also showed that ablating galanin-expressing cells enhances pruritogen-evoked itch, and consistent with this we find enhancement of CQ-evoked itch when cells belonging to this population are silenced by injecting AAV.flex.TeLC into Pdyn^Cre mice.

Here we show that activating NPY cells also strongly suppresses CQ-evoked itch. This is at odds with findings of Acton et al²¹, who reported that chemogenetic activation of NPY-lineage neurons failed to alter scratching in response to CQ. There are technical differences between these studies, since Acton et al used a reporter mouse line to express hM3Dq, and injected CQ intradermally behind the ear. The discrepancy between the results of these studies is most likely to result from higher levels of DREADD expression following viral transfection, and therefore more effective neuronal activation. However, there may also have been a contribution from regional differences in the itch tests used (hindlimb versus head), as well as in the neuronal populations targeted (as noted above). Although Bourane et al¹⁸ reported that ablating ∼70% of NPY-lineage neurons had no effect on itch evoked by CQ, we found that synaptic silencing of the NPY cells with TeLC increased CQ-evoked itch, and often resulted in development of skin lesions, presumably secondary to the spontaneous itch-related biting that was also observed. In fact, the antipruritic action of the NPY-INs may be more powerful than that of the dynorphin/galanin cells, since TeLC silencing in the Pdyn^Cre mouse caused less of an increase in chloroquine-evoked itch behaviour (compared to silencing in the NPY::Cre line) and did not result in the development of spontaneous itch or associated skin lesions. Nonetheless, our findings demonstrate that both NPY-INs and Dyn-INs contribute to the suppression of pruritogen-evoked itch thus revealing an overlap of function of these neurochemically distinct inhibitory interneuron populations.

NPY cells operate through a circuit involving GRPR neurons

Both spontaneous and CQ-evoked itch behaviours were suppressed when GRPR and NPY cells were silenced simultaneously, and we show directly, using both anatomical and electrophysiological methods, that the NPY cells provide a strong inhibitory input to GRPR-INs. This indicates that GRPR cells are downstream of the NPY cells (Figure 7A). This inhibitory input to GRPR cells appears to be even more powerful than that originating from the dynorphin/galanin cells, since NPY- immunoreactive boutons accounted for 45% of the inhibitory synapses on the GRPR cells, compared to the 21% from dynorphin-immunoreactive boutons. Consistent with this we found that optogenetic activation of NPY cells elicited oIPSCs in all of the GRPR cells tested. Interestingly, these were of much higher mean amplitude (∼250 pA), than the ∼80 pA oIPSCs reported by Liu et al²² in GRPR cells when galanin cells were optogenetically activated using a very similar experimental approach. The inhibition of GRPR cells by NPY-INs is likely to be predominantly GABAergic, since oIPSCs were reduced by gabazine in all cells (with one also sensitive to strychnine). Also, consistent with previous evidence showing that the majority of GRPR cells lack Y1 receptors^{21, 23}, we did not detect outward currents in any of the GRPR cells that were tested with a Y1 agonist.

Figure 7.

Although GRPR-expressing excitatory interneurons have been strongly implicated in itch, we have recently shown that these cells respond to noxious as well as pruritic stimuli, that they correspond morphologically to a class of SDH excitatory interneurons known as vertical cells, and that chemogenetically activating them results in behaviours reflecting both pain and itch⁵⁴. Vertical cells provide input to lamina I projection neurons⁵⁵, and are thought to form an integral part of circuits that underlie both normal and pathological pain^{17, 56, 57}. It is already known that axons of NPY cells directly innervate lamina I projection cells, as well as a population of nociceptive projection neurons in laminae III-V of the dorsal horn^{27, 58}. This direct input to ALS projection neurons will presumably contribute to the antinociceptive action of the NPY cells. The present findings raise the possibility that the powerful inhibitory GABAergic NPY-GRPR circuit that we have identified contributes not only to the alleviation of itch, but also to the suppression of nocifensive reflexes and the reduction of hypersensitivity in persistent pain states (Figure 7B).

Activating NPY cells suppresses hypersensitivity in persistent pain states

Importantly, in addition to its effect on acute nocifensive reflexes and itch, activating NPY cells also blocked thermal and mechanical hypersensitivity in both inflammatory and neuropathic pain states. In the SNI model, we found that administration of a Y1 antagonist had no effect on the reversal of mechanical and heat hypersensitivity when NPY cells were activated. NPY acting on Y1 receptors expressed by spinal neurons is known to reduce signs of neuropathic pain^{29, 43, 59}; however, it appears that chemogenetic activation of NPY cells generated GABAergic inhibition that was sufficiently powerful to reverse the hypersensitivity independently of Y1 signalling. Interestingly, our CPP findings suggest that activating NPY neurons may not suppress on-going pain in the SNI model, implying that on-going and evoked components of neuropathic pain operate through different circuits at the spinal cord level.

Previous studies have tested the effects of chemogenetically activating other inhibitory interneuron populations on persistent pain states. Glycinergic cells account for the majority of inhibitory interneurons in deep dorsal horn and are largely separate from the NPY-INs^{36, 60}. Foster et al⁷ showed that activating these cells reduced responses to acute thermal and mechanical noxious stimuli and suppressed mechanical hypersensitivity in the chronic constriction injury model. Activation of PV- expressing inhibitory interneurons reduced inflammatory and neuropathic mechanical allodynia, but had no effect on heat hypersensitivity¹⁵. A recent study by Albisetti et al⁶¹ found that activating a population of dorsal horn inhibitory interneurons defined by expression of Kcnip2 suppressed cold allodynia in a neuropathic model. However, our findings apparently provide the first evidence that activating dorsal horn inhibitory interneurons can suppress heat hypersensitivity, in addition to cold and mechanical allodynia, in persistent pain states. A population of NPY-expressing inhibitory interneurons with a similar laminar location has recently been identified in human spinal cord⁶². These cells therefore provide an attractive target for the treatment of neuropathic pain, particularly for the significant cohort of patients who experience thermal hyperalgesia⁶³.

Materials and methods

Key resources table

Experimental Model and Subject Details

All experiments were approved by the Ethical Review Process Applications Panel of the University of Glasgow, and were carried out in accordance with the European Community directive 86/609/EC, the UK Animals (Scientific Procedures) Act 1986 and ARRIVE guidelines. The following transgenic mouse lines were utilised in this study: the GENSAT BAC transgenic RH26 NPY::Cre line, which express Cre recombinase under control of the NPY promoter⁶⁴; the Ai9 Cre reporter line, in which a loxP-flanked STOP cassette prevents CAG promoter-driven transcription of tdTomato; the GRPR^CreERT2 line, in which 2A-linked optimized Cre recombinase fused with the ligand-binding domain of the estrogen receptor is inserted into the 3’UTR of the Grpr gene⁶⁵; the GRPR^FlpO line, in which 2A-linked FlpO recombinase is fused with the last exon of the Grpr gene²²; and the Pdyn^Cre line, in which an IRES site fused to Cre recombinase is inserted downstream of the stop codon of the Pdyn gene⁶⁶. Further details of these lines can be found in the Key Resources table. NPY::Cre and Ai9, NPY::Cre and GRPR^CreERT2, NPY::Cre and GRPR^FlpO or GRPR^CreERT2 and Ai9 mice were crossed to produce NPY::Cre;Ai9, NPY::Cre;GRPR^CreERT2, NPY::Cre;GRPR^FlpO and GRPR^CreERT2;Ai9 experimental animals, respectively. For experiments inolving the GRPR^CreERT2 line, mice received 6 mg of tamoxifen (2x i.p. injections of 3 mg on consecutive days). For the NPY::Cre;GRPR^CreERT2 mice, this was administered on the day of surgery and on the next day. Wildtype C57BL/6 mice were used for assessment of possible off-target effects of systemic CNO administration, and also to test for Cre-independent effects following injection of AAV.flex.TeLC.eGFP. Mice weighed 15 g to 28 g and animals of both sexes were used, with care taken to include approximately equal numbers of males and females. The animals were between 5 and 14 weeks old at the time of tissue harvest (for anatomy), electrophysiological recording, or behavioural testing. Where drug treatment was given (except for CPP experiments), the treatment type, or the order in which mice received the drug or vehicle was randomised. For TeLC experiments, the viral construct used for each animal was randomised. In all of these cases, the experimenter was blind to the treatment or the viral construct used.

Intraspinal AAV injections

Mice were anaesthetised with 1% to 2% isoflurane and placed in a stereotaxic frame. The skin was incised in the midline over the upper back and superficial muscle was removed from the vertebral column at the level of the T12 to L1 vertebrae, which were then clamped. The L3 and L5 spinal segments were injected through the T12/T13 and T13/L1 intervertebral spaces, respectively, whereas the L4 segment was injected via a hole drilled through the lamina of the T13 vertebra. Injections were performed by making a small slit in the dura and inserting a glass micropipette (outer/inner tip diameter: 60/40 µm) attached to a 10 µL Hamilton syringe, 400 µm lateral to the midline and 300 µm below the pial surface. The following AAV constructs were used: AAV.flex.eGFP (1.72 x 10⁹ GC), AAV.flex.tdTomato (1.76 x 10⁹ GC) and AAV.flex.ChR2-eYFP (5.09 x 10⁸ GC), all from Penn Vector Core, PA, USA; AAV.flex.hM3Dq-mCherry, University of North Carolina Vector Core, NC, USA or University of Zurich Viral Vector Facility, Switzerland, (3.8 x 10⁸ or 7.65 x 10⁸ GC, respectively); AAV.flex.eGFP (2 x 10⁸ GC), AAV.flex.TeLC.eGFP (2 x 10⁸ GC) and AAV.FRT.mCherry, (8.7 x 10⁸ GC), all from University of Zurich Viral Vector Facility, Switzerland; and AAV-Brainbow2 (1.5 - 5.96 x 10⁷ GC), Addgene, MA, USA. Further details of the viruses used can be found in the Key Resources table. 300 nL of virus was infused per injection site (or 500nl for AAV-Brainbow2) at a rate of 30-40 nL/minute using a syringe pump (Harvard Apparatus, MA, USA). Pipettes were left within the spinal cord for 5 mins to minimise leakage of injectate. Once injections were complete the wound was closed and animals recovered with appropriate analgesic administration (0.3 mg/kg buprenorphine and 5 mg/kg carprofen). The success of spinal AAV injections was assessed by post-hoc immunohistochemical staining for the appropriate fluorescent marker protein. Mice were only included for behavioural analyses if the AAV injection(s) into the spinal segments relevant to the dermatome(s) being tested were successful (L4 and L5 for plantar hindpaw-directed tests, L3 for calf skin-directed tests; see below). In some experiments NPY::Cre;GRPR^CreERT2 mice were injected with AAV.flex.eGFP or AAV.flex.TeLC.eGFP, which should result in both NPY- expressing and GRPR-expressing virus-infected cells being labelled with eGFP. This was confirmed for each injection site in each animal by two methods: 1) by assessing the distribution of eGFP-labelled cells, as NPY-INs are located throughout laminae I-III, whereas GRPR-INs are almost entirely restricted to laminae I and IIo, and 2) by antibody co-staining for the inhibitory marker Pax2, as NPY-INs in laminae I-III are exclusively inhibitory, whereas GRPR-INs in laminae I and IIo are exclusively excitatory. Post-AAV-injection behavioural testing was performed within 1-5 weeks for experiments using AAV.flex.hM3Dq-mCherry, and within 4-6 days for experiments using AAV.flex.TeLC.eGFP. For anatomical analyses of inhibitory synaptic input on to GRPR-INs AAV-Brainbow2 injections were performed unilaterally into the L3 and L5 segments and mice were perfused 2–3 weeks post-surgery. For electrophysiological studies, AAV.flex.ChR2-eYFP, on some occasions combined with AAV.FRT.mCherry, was injected unilaterally or bilaterally into the L3 and/or L5 segments and spinal cord slices were prepared from the mice 1–3 weeks post-surgery.

Intraplantar complete Freund’s adjuvant (CFA) injections

Mice were briefly anaesthetised with 1% to 2% isoflurane, the plantar surface of the hindpaw ipsilateral to the spinal AAV injection was wiped with 70% ethanol and 20 µl of 1 mg/ml CFA was injected subcutaneously. Behavioural testing was performed prior to (pre-CFA baseline) and 2 days following CFA injections.

Spared nerve injury (SNI) surgery

Mice were anaesthetised with 1% to 2% isoflurane, an incision was made in the skin over the thigh ipsilateral to the spinal AAV injection and the underlying muscle was blunt-dissected to reveal the sciatic nerve. The tibial and common peroneal branches were identified, and 7-0 Mersilk (Ethicon, Puerto Rico) was used to apply two tight ligatures 2-3 mm apart on each nerve branch. The length of nerve between the ligatures was then removed and the wound was closed. Great care was taken to avoid damage to the sural branch of the sciatic nerve during the surgery. Behavioural testing was performed prior to (pre-SNI baseline) and from 2-4 weeks following SNI surgery.

Drug administration

Clozapine-N-Oxide (CNO; Tocris Bioscience, Bristol, UK) dissolved in a 10% DMSO / 90% sterile saline mixture was injected intraperitoneally (i.p.) at a dose of 5 mg/kg; 10% DMSO / 90% sterile saline mixture alone was used as a vehicle control. In some cases we used CNO-dihydrochloride (Tocris Bioscience), dissolved in 10% water / 90% sterile saline at a dose of 5 mg/kg, with 10% water / 90% sterile saline as a vehicle control. For some experiments the Y1 antagonist BMS 193885 (Bio-Techne, Abingdon, UK) dissolved in a 40% PEG-400 / 60% sterile saline mixture was co-injected i.p. at a dose of 10 mg/kg with CNO; co-injection of the respective vehicles for CNO and BMS 193885 was used as a control. Gabapentin (Sigma Aldrich) was injected i.p.at a dose of 100mg/kg.The timing of CNO or gabapentin injections for conditioned place preference (CPP) testing are described in the relevant section below. For all other behavioural testing, CNO, CNO + BMS 193885 or vehicle were injected a minimum of 30 minutes prior to the start of testing, and all testing was completed within a maximum of 5 hours following injection.

Behavioural testing

von Frey test (noxious punctate mechanical sensitivity)

Mice were placed in a plastic enclosure with mesh flooring and allowed to acclimatise for at least 45 minutes. von Frey filaments of logarithmically incremental stiffness (range 0.01 g to 4 g) were applied to the plantar surface of the hindpaw and the 50% mechanical withdrawal threshold (MWT) was determined using Dixon’s up-down method^{67, 68}. Briefly, filaments were applied sequentially, beginning with the mid-range filament (0.4 g), and the presence or absence of a withdrawal response (lifting and/or shaking of the paw) was noted. If a withdrawal response was observed, the next lowest filament was used subsequently; if no response was observed, the next highest filament was used subsequently. Testing continued until a series of six filaments had been applied from the point when the response threshold was first crossed, and the 50% MWT was calculated using the formula 50% MWT = (10[Xf+kδ])/10,000, where Xf = log value of the final filament applied, k = tabular value (taken from Chaplan et al.⁶⁷) based on the pattern of six positive/negative responses and δ = mean difference (in log units) of the range of filaments used (0.323). When testing mice that had undergone spared nerve injury, care was taken to apply von Frey filaments to the sural territory of the hindpaw.

Hargreaves test (noxious heat sensitivity)

Mice were placed in plastic enclosures on a raised glass platform warmed to 25⁰C and allowed to acclimatise for at least 30 minutes. A radiant heat source (IITC, CA, USA) set to 25% active intensity was targeted to the plantar surface of the hindpaw to be tested (using an angled mirror and guide light), and the time until paw withdrawal from the heat source (withdrawal latency) was noted. Testing of ipsilateral and contralateral paws was alternated with at least 3 minutes interval between consecutive tests, and a cut-off time of 25s was used to prevent tissue damage. Each hindpaw was tested 5 times, and the average withdrawal latency calculated. When testing mice that had undergone spared nerve injury, care was taken to target the heat source to the sural territory of the hindpaw.

Dry ice test (noxious cold sensitivity)

Mice were placed in plastic enclosures on a raised 5 mm-thick glass platform at room temperature and allowed to acclimatise for at least 45 minutes. A dry ice pellet of ∼1 cm diameter was applied to the underside of the glass directly below the hindpaw to be tested, and the withdrawal latency was recorded⁶⁹. Care was taken to ensure that the plantar surface of the hindpaw was in direct contact with the glass prior to testing. Testing of ipsilateral and contralateral paws was alternated with at least 3 minutes interval between consecutive tests, and a cut-off time of 25 s was used to prevent tissue damage. Each hindpaw was tested 5 times, and the average withdrawal latency calculated.

Acetone evaporation test (noxious cold sensitivity)

Mice were placed in plastic enclosure with mesh flooring and allowed to acclimatise for at least 30 minutes. A 10 µl droplet of acetone was applied to the sural territory of the plantar hindpaw and the total amount of time spent shaking, lifting and licking the paw within 30 seconds of acetone application was recorded using a stopwatch. Each hindpaw was tested 3 times, with at least 3 minutes interval between consecutive tests, and the average total response time calculated.

Rotarod test (motor co-ordination)

Mice were placed into the rotarod apparatus (IITC, CA, USA), which was programmed to accelerate from 4 to 40 rpm over 5 minutes. Mice were allowed two trial runs prior to performing four test runs, and the average maximum rpm attained was calculated from the test runs.

Pruritogen-evoked itch test

Mice were acclimatised for 30 mins in plastic enclosures surrounded by angled mirrors to provide unobstructed views of the targeted hindlimb, and were then video recorded for 30 mins (pre-CQ). They then received 10 µl of 1% chloroquine (CQ) dissolved in PBS via intradermal injection into the calf ipsilateral to the spinal AAV injection (which had been shaved 24 hours prior to testing). Successful intradermal injection of CQ was assessed by the appearance of a skin bleb at the injection site. Mice were video recorded for 30 minutes following CQ injection (post-CQ). The amount of time spent biting the injected area was scored offline either manually with a stopwatch or using BORIS event logging software (freely available, https://www.boris.unito.it/)⁷⁰. Videos were viewed at one-quarter speed for analysis.

Conditioned place preference (CPP) test

To test for ongoing neuropathic pain, a three-day conditioning protocol using a biased chamber assignment was used for conditioned place preference (CPP) testing as described previously⁷¹. The custom 3-chamber CPP apparatus consisted of two conditioning side chambers (170 x 150 mm) connected by a centre chamber (70 x 75 mm), 180 mm tall, with infrared-transparent plastic lids (QD Plastics, Dumbarton, UK). Mice were able to discriminate between chambers using visual (vertical versus horizontal black-and-white striped walls) and sensory (rough versus smooth textured floor) cues. On day 1 (acclimation, 7 days after SNI surgery), mice had free access to explore all chambers for 30 minutes. On days 2 and 3 (preconditioning), mice were again allowed to freely explore for 30 min whilst their position was recorded using an infrared camera and AnyMaze 7.16 software (Stoelting, USA). To avoid pre-existing chamber bias, mice spending more than 90% or less than 5% of time in either side chamber during preconditioning were excluded (1 mouse from each experimental group). For conditioning (days 4 to 6), each morning, mice received i.p. vehicle injection, were returned to their home cage for 5 min, then confined to their preferred side chamber for 30 min. Four hours later, mice received i.p. CNO (5 mg/kg) or gabapentin (100 mg/kg), were returned to their home cage for 5 min, and then placed in their non-preferred chamber for 30 min. On test day (day 7), mice could freely explore all chambers whilst their position was recorded, as during pre-conditioning, for 30 min. Difference scores were calculated as the time spent in each chamber on test day minus the mean time spent during pre-conditioning. We have previously shown that CNO given at a much lower dose (0.2 mg/kg) to mice in which spinal GRPR neurons expressed hM3Dq resulted in itch- and pain-related behaviours that started within 5 minutes of administration⁵⁴. It is therefore very likely that NPY neurons would have been activated throughout the conditioning period for CNO. In addition, the timecourse of action of CNO and gabapentin are likely to be similar, since both were administered i.p., and there was a clear preference for the chamber paired with gabapentin.

Noxious heat and pruritic induction of Fos

Mice were injected i.p. with CNO or vehicle 30 minutes prior to receiving a noxious heat or pruritic stimulation under brief isoflurane anaesthesia, ipsilateral to spinal AAV.flex.hM3Dq-mCherry injection. The noxious heat stimulus was immersion of the hindpaw into 52⁰C water for 15 s. The pruritic stimulus was intradermal injection of CQ into the calf as described above, following shaving of the leg 24 hours prior to CQ injection. These mice were fitted with Elizabethan collars to prevent Fos induction through scratching and/or biting of the injected area. Mice were transcardially perfused under deep terminal anaesthesia 2 hours after stimulation, and spinal cord tissue was processed for imaging and analysis as described below.

Electrophysiology

Electrophysiological studies were performed on spinal cord slices from NPY::Cre or NPY::Cre;GRPRFlpO mice that had received an intraspinal injection of AAV.flex.ChR2-eYFP or AAV.flex.ChR2-eYFP combined with AAV.FRT.mCherry, respectively, 1 to 3 weeks prior to recordings. Additional recordings were performed on spinal cord slices from GRPR^CreERT2;Ai9 mice. Recordings were made from 51 cells (39 from female, 12 from male mice) from spinal cord slices obtained from 21 NPY::Cre mice (16 female, 5 male) and an additional 11 cells from 5 female NPY::Cre;GRPR^FlpO mice and 10 cells from 5 female GRPR^CreERT2;Ai9 mice, that were aged 5 to 11 weeks. Spinal cord slices were prepared as described previously⁷². Mice were decapitated under isoflurane anaesthesia and the spinal cord removed in ice-cold dissection solution. In some cases decapitation was performed following transcardial perfusion with ice-cold dissection solution under terminal anaesthesia with pentobarbital (20 mg i.p.). The lumbar region was embedded in an agarose block and 300 µm parasagittal or 350 µm transverse slices were cut with a vibrating blade microtome (Thermo Scientific Microm HM 650V, Loughborough, UK; Leica VT1200s, Milton Keynes, UK; or Campden Instruments 7000smz-2, Loughborough, UK). Slices were then allowed to recover for at least 30 minutes in recording solution at room temperature. In some cases slices were placed in an N-methyl-D-glucamine (NMDG)-based recovery solution at ∼32°C for 15 minutes⁷³ before being placed in a modified recording solution at room temperature for at least 30 minutes. The solutions used contained the following (in mM); dissection, 251.6 sucrose, 3.0 KCl, 1.2 NaH₂PO₄, 0.5 CaCl₂, 7.0 MgCl₂, 26.0 NaHCO₃, 15.0 glucose; NMDG recovery, 93.0 NMDG, 2.5 KCl, 1.2 NaH₂PO₄, 0.5 CaCl₂, 10.0 MgSO₄, 30.0 NaHCO₃, 25.0 glucose, 5.0 Na-ascorbate, 2.0 thiourea, 3.0 Na-pyruvate, and 20.0 HEPES; modified recording, 92.0 NaCl, 2.5 KCl, 1.2 NaH₂PO₄, 2.0 CaCl₂, 2.0 MgSO₄, 30.0 NaHCO₃, 25.0 glucose, 5.0 Na-ascorbate, 2.0 thiourea, 3.0 Na-pyruvate, and 20.0 HEPES; and recording, 125.8 NaCl, 3.0 KCl, 1.2 NaH₂PO₄, 2.4 CaCl₂, 1.3 MgCl₂, 26.0 NaHCO₃, 15.0 glucose. All solutions were bubbled with 95% O₂ / 5% CO₂.

Whole-cell patch-clamp recordings were made from ChR2-eYFP+ or ChR2-eYFP-neurons in NPY::Cre tissue, from mCherry+ cells in NPY::Cre;GRPR^FlpO tissue, or from tdTom+ cells in GRPR^CreERT2;Ai9 tissue in the superficial dorsal horn, using patch pipettes that had a typical resistance of 3 – 7 MΩ. Data were recorded and acquired with a Multiclamp 700B amplifier and pClamp 10 software (both Molecular Devices, Wokingham, UK), and were filtered at 4 kHz and digitised at 10 kHz.

To validate the optogenetic activation of NPY neurons, recordings were made from ChR2-eYFP+ cells using a K-based intracellular solution. The presence of optogenetically-activated currents were observed in voltage clamp mode, from a holding potential of −70 mV, by illuminating the slice with brief (1 – 4 ms) pulses of blue light, generated by a 470 nm LED (CoolLED pE-100, Andover, UK) and delivered via the microscope objective. The ability to optogenetically drive action potential firing in NPY neurons was similarly tested by applying brief pulses of blue light while the membrane potential was held around −60 mV in current clamp mode.

Inhibition from NPY cells to other cells in the superficial dorsal horn was investigated in tissue from NPY::Cre mice, by recording from ChR2-eYFP-negative cells that were within the region of the viral injection, as determined by eYFP expression, and applying pulses of blue light as detailed above. Inhibition of GRPR cells by NPY cells was similarly assessed by recording from mCherry+ GRPR cells in slices from NPY::Cre;GRPR^FlpO mice. Cells were classed as receiving optogenetically-evoked IPSCs (olPSCs) if there was a clear reliably evoked inward current (CsCl-based intracellular, V_hold −70 mV), or outward current (K-based intracellular, V_hold −40 mV; Cs-methanesulfonate-based intracellular, V_hold 0mV), that was time-locked to the light pulse. In a subset of recordings, using CsCl-based intracellular solution (V_hold −70 mV), the optogenetically-evoked postsynaptic currents were confirmed to be non-glutamatergic by bath application of NBQX (10 µM) and D- APV (30 µM). The nature of the oIPSCs recorded in unlabelled (CsCl-based intracellular, V_hold −70 mV) and GRPR (Cs-methanesulfonate-based intracellular, V_hold 0mV) cells was investigated by bath application of the GABA antagonist, gabazine (300 nM) and the glycine antagonist, strychnine (300 nM); in the case of unlabelled cells this was done in the presence of NBQX and D-APV. oIPSCs were evoked 6 times (0.05 Hz) (Baseline) prior to the application of gabazine or strychnine, which was added to the recording solution and washed into the recording chamber for 5 minutes before and during the recording of a further 6 oIPSCs. Strychnine was then added to gabazine or gabazine added to strychnine and following a further 5 minute wash in period 6 oIPSCs were recorded. oIPSCs were classified as ‘sensitive’ to gabazine and/or strychnine if the drug reduced the mean peak amplitude of the oIPSC to a level that was less than 2 SD of the mean peak amplitude recorded during baseline or the previous drug application, and ‘insensitive’ if this threshold was not met. The monosynaptic nature of the oIPSCs was tested in a subset of GRPR cells, by investigating whether oIPSCs that were abolished by the application of tetrodotoxin (TTX; 0.5 µM) could be rescued by the addition of 4-aminopyridine (4- AP; 100 µM)⁷⁴. To test whether the release of NPY from NPY cells may contribute to the inhibition of GRPR cells, patch clamp recordings were made from tdTom+ GRPR cells from GRPR^CreERT2;Ai9 mice to assess responses to the NPY Y₁ receptor agonist, [Leu³¹,Pro³⁴]-Neuropeptide Y (300 nM), which was bath applied in the presence of TTX (0.5 µM), bicuculline (10 µM) and strychnine (1 µM) (K-based intracellular, V_hold −50 mV). Following a 1 minute baseline period, [Leu³¹,Pro³⁴]- Neuropeptide Y was applied for 4 minutes, and cells where classified as responders if [Leu³¹,Pro³⁴]-Neuropeptide Y resulted in an outward current of 5 pA or greater and were classified as non-responders if this threshold was not reached.

The intracellular solutions used contained the following (in mM); K-based, 130.0 K-gluconate, 10.0 KCl, 2.0 MgCl2, 10.0 HEPES, 0.5 EGTA, 2.0 ATP-Na2, 0.5 GTP-Na, and 0.2% Neurobiotin, pH adjusted to 7.3 with 1.0M KOH; CsCl-based, 130.0 CsCl, 1.0 MgCl2, 10.0 HEPES, 10.0 EGTA, 5.0 N-(2,6-dimethylphenylcarbamoylmethyl) triethylammonium bromide (QX-314-Br), 2.0 ATP- Na2, 0.3 GTP-Na, and 0.2% Neurobiotin, pH adjusted to 7.35 with 1.0M CsOH; Cs- methylsulfonate-based, 120.0 Cs-methylsulfonate, 10.0 Na-methylsulfonate, 10.0 EGTA, 1.0 CaCl2, 10.0 HEPES, 5.0 N-(2,6-dimethylphenylcarbamoylmethyl) triethylammonium chloride (QX-314-Cl), 2.0 Mg2-ATP, and 0.2% Neurobiotin, pH adjusted to 7.2 with 1.0M CsOH.

All chemicals were obtained from Sigma except: TTX, QX-314-Br, QX-314-Cl (Alomone, Jerusalem, Israel), Gabazine, NBQX (Abcam, Cambridge, UK), sucrose, glucose, NaH₂PO₄ (VWR, Lutterworth, UK), D-APV, [Leu³¹,Pro³⁴]-Neuropeptide Y (Tocris, Abingdon, UK) and Neurobiotin (Vector Labs, Peterborough, UK).

Immunohistochemistry (IHC)

Animals were terminally anaesthetised with pentobarbital (20 mg i.p.) and transcardially perfused with 4% freshly depolymerised formaldehyde. The spinal cord was then dissected out and post-fixed in the same fixative for 2 hours, and 60 µm- thick transverse or sagittal sections from appropriate lumbar sections were cut on a vibrating microtome. Sections were immersed in 50% ethanol for 30 minutes to enhance antibody penetration before incubation in appropriate primary and secondary antibodies at 4⁰C for 72 and 24 hours, respectively. Details of the antibodies used in this study can be found in the Key Resources table. Sections were continuously agitated during antibody incubation and washed 3 times in PBS that contained 0.3 M NaCl following each incubation. Following the final PBS wash, sections were mounted on slides in Vectashield anti-fade mounting medium (Vector Laboratories, CA, USA). In some cases lumbar dorsal root ganglia (DRGs) were also removed and processed intact for IHC before whole-mounting on slides.

Fluorescent in situ hybridisation (FISH)

Multiple-labelling FISH was performed using RNAscope probes and RNAscope fluorescent multiplex reagent kit 320850 (ACD BioTechne, CA, USA). Mice were deeply anaesthetised, the spinal cord was rapidly removed by hydraulic extrusion and the lumbar enlargement was excised and snap-frozen on dry ice. Lumbar segments were then embedded in OCT mounting medium and 12 µm-thick transverse sections were cut on a Leica CM 1950 cryostat (Leica, Milton Keynes, UK). Sections were mounted non-sequentially (such that sections on the same slide were at least 4 apart) onto SuperFrost Plus slides (VWR, Lutterworth, UK) and air- dried. The slides were then reacted according to the manufacturer’s recommended protocol, using probes against Cre and Npy that were revealed with Atto 550 and Alexa 647, respectively. Further details of the probes used can be found in the Key Resources table. Sections were mounted using Prolong-Glass anti-fade medium containing NucBlue (Hoescht 33342) nuclear stain (ThermoFisher Scientific, Paisley, UK).

Image acquisition and analysis

IHC and FISH slides were imaged with a Zeiss LSM 710 confocal microscope system equipped with Ar multi-line, 405-nm diode, 561-nm solid-state and HeNe lasers. For analyses of AAV.flex.FP, AAV.flex.hM3Dq-mCherry or AAV.flex.TeLC.eGFP-labelled cells image stacks were taken through a 40x objective (NA = 1.3) at a z-separation of 1 µm. For assessment of spinal injection sites and analysis of mCherry expression in DRGs of AAV.flex.hM3Dq-mCherry-injected mice, image stacks were taken through a 10x objective (NA = 0.3) at a z-separation of 2 µm. Image analysis was performed using Neurolucida software (MBF Bioscience, VT, USA).

For comparison of tdTomato or eGFP expression with NPY immunoreactivity in AAV.flex.tdTomato-or AAV.flex.eGFP-injected NPY::Cre mice, and in AAV.flex.eGFP-injected NPY::Cre;Ai9 mice, a modified optical disector method was used⁷⁵. All structures labelled with the neuronal marker NeuN that had their bottom surface between reference and look-up sections separated by 10 µm (10 optical sections) were marked within the injection site in laminae I-III. The NPY-IR, eGFP and/or tdTomato channels were then viewed separately and sequentially, and the NeuN profiles were marked as positive or negative as appropriate. As NPY-IR varies greatly from cell to cell and is often observed as discrete clumps within the perikaryal cytoplasm^{10, 26}, cells were classified as NPY-positive if clear above-background signal was observed in at least 3 consecutive optical sections. The same optical disector method was used to identify NeuN-positive cells, and compare mCherry, Fos and Pax2 or mCherry, Fos and NPY immunoreactivity in these cells in vehicle-or CNO-dosed AAV.flex.hM3Dq-mCherry spinal-injected NPY::Cre mice, as well as eGFP and NPY immunoreactivity in the AAV.flex.TeLC.eGFP spinal-injected NPY::Cre mice. A similar method was used to compare mCherry, Fos and Pax2 in NeuN-positive cells in the vehicle- or CNO-dosed AAV.flex.hM3Dq-mCherry spinal-injected NPY::Cre mice that had received noxious heat or pruritic stimuli 2 hours prior to perfusion; however, in this case the reference and look-up sections were separated by 20 µm (20 optical sections) and analyses were restricted to the somatotopically-relevant areas of laminae I and II (medial half for noxious heat stimulation of hindpaw, middle third for CQ injection into calf). For all of these analyses, markers from previously-assessed channels were hidden as the channels were analysed in sequence, to prevent bias.

For comparison of tdTomato and eGFP or mCherry expression with neurochemical markers of inhibitory interneuron populations in AAV.flex.eGFP-injected NPY::Cre;Ai9 or AAV.flex.hM3Dq-mCherry-injected NPY::Cre mice, respectively, all cells expressing the neurochemical markers and fluorescent protein(s) in laminae I and II were marked throughout the whole section thickness. Each channel was again assessed separately and sequentially, with markers from previously-assessed channels hidden to prevent bias. Only cells with the maximal profile of their soma contained within the z stack were counted. mCherry-positive cells in whole-mounted DRGs from AAV.flex.hM3Dq-mCherry-injected NPY::Cre mice that underwent SNI surgery were counted using the same method.

For analysis of inhibitory synapses onto GRPR-INs, 9 mTFP-labelled cells in total were selected from tissue of 3 AAV-Brainbow2-injected GRPR^CreERT2 mice immunostained for mTFP, VGAT, gephyrin and NPY or Dynorphin B (3 per animal for Dynorphin B and 2, 3 and 4 cells for NPY analyses). Cell selection was performed before the staining for axonal markers was visualised and was based on the completeness of dendritic labelling and separation from other nearby mTFP- labelled neurons. The selected cells were scanned through a 63x oil-immersion lens (numerical aperture 1.4) at a z-separation of 0.3 µm. Z-series were obtained from as much of the dendritic tree as was visible in the section. For analysis, the mTFP and gephyrin channels were initially viewed and the cell bodies and dendritic trees were traced based on the mTFP signal. The locations of all gephyrin puncta associated with the cell body and dendritic tree were then plotted. The VGAT channel was then viewed and the presence or absence of an apposed VGAT-immunoreactive bouton was noted for each gephyrin punctum. Finally, the remaining channel (corresponding to NPY or Dynorphin B) was revealed and the presence or absence of peptide staining was noted for each of the VGAT boutons that contacted a gephyrin punctum on the selected cell. To determine the frequency of all boutons arising from inhibitory interneurons that were positive for each of the neuropeptides, we sampled from those VGAT-immunoreactive boutons in the vicinity of the mTFP-labelled cell. A 4 x 4 µm grid was applied within a box drawn to include the entire dendritic tree of the cell. Only the VGAT channel was viewed initially and in each successive grid square, the VGAT-immunoreactive bouton nearest the bottom right of the square was selected. The presence or absence of NPY or DynorphinB was then recorded for each of these selected VGAT-immunoreactive boutons.

For comparison of Cre and Npy mRNA in FISH sections, all NucBlue profiles within a single optical section were marked throughout laminae I-III. The Cre and Npy channels were then viewed separately and sequentially and cells were marked as positive if they contained ≥4 labelled transcript particles. Markers from the mRNA channel that was assessed first were hidden during assessment of the second channel to prevent bias.

Quantification and statistical analysis

Data are reported as mean ± SEM, unless stated otherwise. Statistical analyses were performed using Prism software (v7,v8 & v9,GraphPad Software, CA, USA). Behavioural and anatomical analyses involving drug treatments were analysed blind to treatment group. Behavioural analyses involving AAV.flex.TeLC.eGFP-mediated silencing and AAV.flex.eGFP-injected controls were analysed blind to the AAV injected. The statistical tests used for each experiment, including tests for multiple comparisons, are given in the appropriate figure legends. A p value of <0.05 was considered significant, and significance markers are denoted within figures as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Data and materials availability

Upon acceptance of this manuscript, all data will be made accessible from the Enlighten: Research Data open repository hosted at the University of Glasgow. This study did not generate any new materials, reagents or code.

Acknowledgements

This research was funded in whole, or in part, by the Wellcome Trust (Grant numbers 102645/Z/13/Z and 219433/Z/19/Z), the Biotechnology and Biological Sciences Research Council (Grant numbers BB/N006119/1, BB/P007996/1) and the Medical Research Council (Grant numbers MR/S002987/1, MR/T01072X/1 and MR/V033638/1). For the purpose of Open Access, the authors have applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission. We are grateful to R Kerr, C Watt and I Plenderleith for expert technical assistance, to Yan-Gang Sun for the gift of GRPR^CrERT2 and GRPR^FlpO mice, to Dr Philippe Ciofi for the Dynorphin B antibody and to Dr Mark Hoon for helpful discussion.

Author contributions

K.A.B., E.P., M.G.-M., A.C.D., A.H.C., A.M.B., G.A.W., J.S.R. and A.J.T. conceived the project and designed experiments. A.J.T. and J.S.R. obtained funding. K.A.B., E.P., M.G.-M., A.C.D., A.H.C., A.M.B. and D.I.H. performed experiments. K.A.B., E.P., M.G.-M., A.C.D., A.H.C., A.M.B., M.E.J. and A.C.-B. analysed data. M.W. provided reagents. K.A.B., A.C.D., A.H.C., A.M.B., G.A.W. and A.J.T. wrote the paper. All of the authors provided feedback and contributed to editing of the manuscript.

Competing interests

The authors declare no competing interests.

Figure supplements

Figure 1 – figure supplement 1.

Figure 2 – figure supplement 1.

Figure 2 – figure supplement 2.

Figure 3 – figure supplement 1.

Figure 4 – figure supplement 1.

Figure 4 – figure supplement 2.

Figure 5 – figure supplement 1.

Figure 5 – figure supplement 2.

Figure 6 – figure supplement 1.