Identification of Spinal Inhibitory Interneurons Required for Attenuating Effect of Duloxetine on Neuropathic Allodynia-like Signs in Rats

Neuropathic pain is a chronic pain condition that occurs after nerve damage; allodynia, which refers to pain caused by generally innocuous stimuli, is a hallmark symptom. Although allodynia is often resistant to analgesics, the antidepressant duloxetine has been used as an effective therapeutic option. Duloxetine increases spinal noradrenaline (NA) levels by inhibiting its transporter at NAergic terminals in the spinal dorsal horn (SDH), which has been proposed to contribute to its pain-relieving effect. However, the mechanism through which duloxetine suppresses neuropathic allodynia remains unclear. Here, we identified an SDH inhibitory interneuron subset (captured by adeno-associated viral (AAV) vectors incorporating a rat neuropeptide Y promoter; AAV-NpyP+ neurons) that is mostly depolarized by NA. Furthermore, this excitatory effect was suppressed by pharmacological blockade or genetic knockdown of α1B-adrenoceptors (ARs) in AAV-NpyP+ SDH neurons. We found that duloxetine suppressed Aβ fiber-mediated allodynia-like behavioral responses after nerve injury and that this effect was not observed in AAV-NpyP+ SDH neuron-selective α1B-AR-knockdown. These results indicate that α1B-AR and AAV-NpyP+ neurons are critical targets for spinal NA and are necessary for the therapeutic effect of duloxetine on neuropathic pain, which can support the development of novel analgesics.


Introduction
Neuropathic pain develops due to a lesion or disease of the somatosensory system [1]. Mechanical allodynia, which refers to pain caused by otherwise innocuous mechanical stimuli, is a symptom of neuropathic pain. Mechanical information from the skin is transmitted to the spinal dorsal horn (SDH) via low-threshold mechanoreceptors such as Aβ fibers [2]. Normally, Aβ fibers do not activate brain-projecting pain transmission neurons located in lamina I of the SDH, but after peripheral nerve injury (PNI), activation occurs and pain is induced (neuropathic allodynia) [3,4]. An increasing body of evidence indicates that dysfunction of inhibitory interneurons in the SDH after PNI critically contributes to Aβ fiber-mediated excitation of nociceptive lamina I neurons and neuropathic allodynia [5][6][7][8][9]. Thus, the enhancement of inhibitory interneuron activity would be an effective way to mitigate neuropathic allodynia.
Somatosensory information processing in the SDH is regulated by serotonin and noradrenaline (NA) released from spinal terminals of descending neurons in the brain [10][11][12]. The descending NAergic pathway has been proposed as a target of duloxetine [13][14][15], a 2 of 11 serotonin and NA reuptake inhibitor (SNRI) that has shown clinical efficacy for treating neuropathic pain [1,16]. Duloxetine increases spinal NA levels by inhibiting its transporters at NAergic terminals [17][18][19]. Within the SDH, NA inhibits glutamate release from the presynaptic terminal of C fibers and excitation of SDH interneurons via α 2 -adrenoceptors (ARs) [20]. However, the role of α 2 -ARs in the duloxetine-mediated mitigation of neuropathic pain remains controversial. The effect of duloxetine is reportedly reduced by yohimbine, an α 2 -AR antagonist; however, the selectivity ratio between α 2 -AR and α 1 -AR is much lower than that of the α 2 -AR-selective antagonist idazoxan [21]. It has also been shown that idazoxan reduces the suppressive effect of duloxetine repeatedly administered on neuropathic behavioral hypersensitivity [19] but not that of acutely administered duloxetine [22]. In addition to α 2 -AR, NA also activates α 1 -AR and has an excitatory effect on inhibitory interneurons in the SDH [23][24][25][26]. However, the role of duloxetine in attenuating neuropathic pain-like behavior is poorly understood. Furthermore, despite recent advances in our understanding of the heterogeneity of inhibitory interneurons in the SDH [7,27,28], the neuronal subset that targets spinal NA necessary for the effect of duloxetine remains unclear.
In the present study, we investigated the mechanism by which duloxetine modulates neuropathic pain by focusing on an SDH inhibitory interneuron subset (captured by an adeno-associated viral [AAV] vector including a rat neuropeptide Y promoter; AAV-NpyP + neurons), which has previously been identified as a key player for Aβ fiber-evoked allodynia-like behavior after PNI [9]. Using multiple approaches (electrophysiology, subsetselective RNA interference, and optogenetics), we identified AAV-NpyP + SDH neurons as critical targets of spinal NA that contribute to the analgesic effect of duloxetine on neuropathic allodynia.

Intra-SDH Injection of rAAV Vector
According to our previous methods [9,31], rats were intraperitoneally (i.p.) injected with pentobarbital (65 mg/kg) under isoflurane (2%) anesthesia. In brief, we inserted the microcapillary backfilled with rAAV solution into the SDH between Th13 and L1 vertebrae (500 µm lateral from the midline and 250 µm in depth from the surface of the dorsal root entry zone) and microinjected 800 nl rAAV solution (3 × 10 12 GC/mL) using FemtoJet Express (Eppendorf, Hamburg, Germany). After microinjection, we slowly removed the inserted microcapillary from the SDH and the skin was sutured with 3-0 silk.

Neuropathic Pain Model
We used the spinal nerve injury model with some modifications [32,33]. In brief, under isoflurane (2%) anesthesia, the L5 spinal nerve was tightly ligated with 5-0 silk and cut just distal to the ligature. The wound and the surrounding skin were sutured with 3-0 silk.

Light Illumination of Hind paw
According to our previous methods [4,9], we placed rats on a transparent acrylic plate and habituated them for 30-60 min. The plantar surface of the hind paw (touching the acrylic plate floor) was illuminated with a blue laser diode (COME2-LB473/532/100, Lucir, Osaka, Japan: wavelength, 470 nm; frequency, 5 Hz; interval, 10 s; 10 times per each ipsilateral and contralateral hind paw) over the acrylic plate. Using a thermopile (COME2-LPM-NOVA, Lucir, Osaka, Japan), we measured the light power intensity (1 mV), which was a laser power meter with 1 mW at the skin. Withdrawal responses of the hind paw to light illumination (0: no reaction, 1: mild movement without any lifting and flinching behaviors, 2: hind paw lifting and flinching) were calculated from total scores of 10 times per hind paw. To investigate the light-induced responses of animals, we were careful to establish experimental conditions prior to light stimuli. First, the animals were awake. Second, both hind paws were clearly attached to the floor of the acrylic plate. Third, the animals were at rest without moving or walking.

Von Frey Test
As previously described [4,9], calibrated von Frey filaments (0.4-15 g, Stoelting, Wood Dale, IL, USA) were applied to the plantar surface of the rat hind paw, and the 50% paw withdrawal threshold was determined.

Intraperitoneally Administration of Duloxetine
Two weeks after PNI, W-TChR2V4 rats were intraperitoneally administrated duloxetine (30 mg/kg) dissolved by saline.

NA Excites the Majority of AAV-NpyP + Neurons Via α 1B -AR
To examine the activity of AAV-NpyP + SDH neurons, we visualized AAV-NpyP + neurons by intra-SDH microinjection of the AAV-NpyP vector including the gene encoding tdTomato (tdT) in wild-type (WT) rats ( Figure 1A). AAV-NpyP + neurons (tdT + cells) were located in lamina IIo and immunolabeled with paired box 2 (PAX2) as a marker of inhibitory neurons ( Figure 1A). Using spinal cord slices from tdT-expressing WT rats, whole-cell recordings from AAV-NpyP + neurons were performed under the current clamp mode. We measured averaged resting membrane potential (RMP) for 1 min of pre-drug and post-drug application, and a change in RMP (∆RMP) of 5 mV or more was judged to be depolarization or hyperpolarization. NA application produced depolarization in 64.3% of the AAV-NpyP + neurons (n = 18/28), and more than half of the depolarizing neurons (n = 11/18) evoked action potential firing ( Figure 1B,C). The remaining neurons (n = 9/28) exhibited hyperpolarization ( Figure 1B,C). The average changes in the RMPs of NA-depolarizing and hyperpolarizing neurons were 12.43 and −7.67 mV, respectively ( Figure 1D). These data indicate that almost all AAV-NpyP + neurons responded to NA, with depolarization being the predominant response.
To examine the activity of AAV-NpyP + SDH neurons, we visualized AAV-NpyP + neurons by intra-SDH microinjection of the AAV-NpyP vector including the gene encoding tdTomato (tdT) in wild-type (WT) rats ( Figure 1A). AAV-NpyP + neurons (tdT + cells) were located in lamina IIo and immunolabeled with paired box 2 (PAX2) as a marker of inhibitory neurons ( Figure 1A). Using spinal cord slices from tdT-expressing WT rats, whole-cell recordings from AAV-NpyP + neurons were performed under the current clamp mode. We measured averaged resting membrane potential (RMP) for 1 min of predrug and post-drug application, and a change in RMP (ΔRMP) of 5 mV or more was judged to be depolarization or hyperpolarization. NA application produced depolarization in 64.3% of the AAV-NpyP + neurons (n = 18/28), and more than half of the depolarizing neurons (n = 11/18) evoked action potential firing ( Figure 1B,C). The remaining neurons (n = 9/28) exhibited hyperpolarization ( Figure 1B,C). The average changes in the RMPs of NA-depolarizing and hyperpolarizing neurons were 12.43 and −7.67 mV, respectively ( Figure 1D). These data indicate that almost all AAV-NpyP + neurons responded to NA, with depolarization being the predominant response.

Knockdown of α 1B -AR in AAV-NpyP + Neurons Suppresses NA-Evoked Depolarization
To determine the AR subtype responsible for the effects of NA, spinal slices were treated with subtype-specific antagonists. NA-induced depolarization in AAV-NpyP + neurons was prevented by the α 1B -AR antagonist L-765,314 ( Figure 1C) but not by silodosin (α 1A -AR) or A-315456 (α 1D -AR). When L-765,314 was treated, the proportion of AAV-NpyP + neurons exhibiting hyperpolarization increased, and co-treatment with atipamezole (α 2 -AR antagonist) abolished the NA-induced responses ( Figure 1C,D). To determine the role of α 1B -ARs expressed in AAV-NpyP + neurons, short hairpin RNA (shRNA) targeting α 1B -ARs (shmirAdra1b) was used to knockdown gene expression in AAV-NpyP + neurons by intra-SDH microinjection of AAV-NpyP-mCherry-shmirAdra1b. We confirmed mCherry expression in the PAX2 + lamina IIo neurons, which is consistent with our previous findings [9] (Figure 2A). Electrophysiological recordings revealed that the NA failed to depolarize almost all mCherry + neurons ( Figure 2B,C). These results indicated that NA-induced depolarization and hyperpolarization are mediated by α 1B -ARs and α 2 -ARs, respectively. Furthermore, AAV-NpyP + neurons depolarized by NA would also express α 2 -ARs, but the net effect of NA in these neurons is excitatory.

Knockdown of α1B-AR in AAV-NpyP + Neurons Suppresses NA-Evoked Depolarization
To determine the AR subtype responsible for the effects of NA, spinal slices were treated with subtype-specific antagonists. NA-induced depolarization in AAV-NpyP + neurons was prevented by the α1B-AR antagonist L-765,314 ( Figure 1C) but not by silodosin (α1A-AR) or A-315456 (α1D-AR). When L-765,314 was treated, the proportion of AAV-NpyP + neurons exhibiting hyperpolarization increased, and co-treatment with atipamezole (α2-AR antagonist) abolished the NA-induced responses ( Figure 1C,D). To determine the role of α1B-ARs expressed in AAV-NpyP + neurons, short hairpin RNA (shRNA) targeting α1B-ARs (shmirAdra1b) was used to knockdown gene expression in AAV-NpyP + neurons by intra-SDH microinjection of AAV-NpyP-mCherry-shmirAdra1b. We confirmed mCherry expression in the PAX2 + lamina IIo neurons, which is consistent with our previous findings [9] (Figure 2A). Electrophysiological recordings revealed that the NA failed to depolarize almost all mCherry + neurons ( Figure 2B,C). These results indicated that NA-induced depolarization and hyperpolarization are mediated by α1B-ARs and α2-ARs, respectively. Furthermore, AAV-NpyP + neurons depolarized by NA would also express α2-ARs, but the net effect of NA in these neurons is excitatory.

Duloxetine Alleviates Neuropathic Allodynia-like Behavior Via α 1B -AR in AAV-NpyP + Neurons
As duloxetine has been shown to increase spinal NA [17][18][19], we predicted that α 1B -ARs in AAV-NpyP + neurons could be involved in its effect on neuropathic pain. To determine this, we examined the effect of duloxetine on neuropathic allodynia using transgenic rats (W-TChR2V4) expressing channelrhodopsin-2 (ChR2) at the nerve endings of touch-sensing Aβ fibers [4,30]. We found that the pain-like withdrawal behavior elicited by photostimulation of Aβ fibers by applying light to the plantar skin 2 weeks after PNI was suppressed by intraperitoneal administration of duloxetine (30 mg/kg) ( Figure 3A). A similar effect was observed in PNI-induced behavioral hypersensitivity to mechanical Cells 2022, 11, 4051 7 of 11 stimulation by von Frey filaments ( Figure 3A). We further found that the suppressive effects of duloxetine were abolished in AAV-NpyP + neuron-specific α 1B -AR-knockdown ( Figure 3B). In addition, we confirmed that the excitatory effect of NA on AAV-NpyP + neurons was retained after PNI, and in spinal slices from WT rats 2 weeks after PNI; NA caused depolarization of 65.0% of AAV-NpyP + neurons (n = 13/20), and the average change in RMPs was 10.73 mV. These results indicate that α 1B -ARs expressed in AAV-NpyP + neurons are necessary for the analgesic effect of duloxetine on neuropathic allodynia-like behavior. mine this, we examined the effect of duloxetine on neuropathic allodynia using transgenic rats (W-TChR2V4) expressing channelrhodopsin-2 (ChR2) at the nerve endings of touchsensing Aβ fibers [4,30]. We found that the pain-like withdrawal behavior elicited by photostimulation of Aβ fibers by applying light to the plantar skin 2 weeks after PNI was suppressed by intraperitoneal administration of duloxetine (30 mg/kg) ( Figure 3A). A similar effect was observed in PNI-induced behavioral hypersensitivity to mechanical stimulation by von Frey filaments ( Figure 3A). We further found that the suppressive effects of duloxetine were abolished in AAV-NpyP + neuron-specific α1B-AR-knockdown ( Figure  3B). In addition, we confirmed that the excitatory effect of NA on AAV-NpyP + neurons was retained after PNI, and in spinal slices from WT rats 2 weeks after PNI; NA caused depolarization of 65.0% of AAV-NpyP + neurons (n = 13/20), and the average change in RMPs was 10.73 mV. These results indicate that α1B-ARs expressed in AAV-NpyP + neurons are necessary for the analgesic effect of duloxetine on neuropathic allodynia-like behavior.

Discussion
In this study, we identified AAV-NpyP + SDH neuron subset that predominantly exhibited depolarization to spinal NA via α 1B -ARs. We further found that the suppressive effect of duloxetine on Aβ fiber-mediated allodynia-like behavioral responses of a model of neuropathic pain was not observed in AAV-NpyP + SDH neuron-selective α 1B -AR-knockdown. From these findings, this study demonstrates that α 1B -AR and AAV-NpyP + neurons are critical targets for spinal NA and are necessary for the therapeutic effect of duloxetine on neuropathic pain. Thus, α 1B -AR expressed in AAV-NpyP + neurons may be targets for the development of novel analgesics.
Spinal NA has been previously shown to directly excite a part of inhibitory interneurons in SDH lamina II [23][24][25]34,35], but the subset of inhibitory interneurons responsible for pain modulation by spinal NA remains unknown. In this study, we showed that AAV-NpyP + neurons, a recently identified inhibitory interneuron subset located selectively in lamina IIo [9], mostly respond to NA, and its effect is excitable in most of these interneurons. Indeed, over 60% of these neurons are depolarized by NA. This proportion is higher than that of GAD67-expressing lamina II inhibitory interneurons (approximately 40%) [35], suggesting that AAV-NpyP + neurons are major targets of spinal NA. Furthermore, our data obtained from pharmacological and genetic interventions using selective antagonists for AR subtypes and AAV-NpyP + neuron-specific α 1B -AR shRNA expression indicated that NA directly excites AAV-NpyP + interneurons via α 1B -ARs. However, the responsible α 1 -AR subtype for the effect of NA on inhibitory neurons in rats may be different from that in mice. Our previous study using mouse spinal cord slices showed that NA increased the frequency of inhibitory postsynaptic currents in substantia gelatinosa neurons via α 1A -ARs [26,36]. The reason for the difference in the α 1 -AR subtypes involved remains unclear, but some possibilities may be considered: the differences in the species (rat vs. mouse), population of inhibitory interneurons (AAV-NpyP + neurons vs. Vgat-Cre + neurons), and the measured responses (depolarization of cell body vs. inhibitory postsynaptic currents in SDH neurons received inputs from NA-responding neurons). Nevertheless, both α 1 -AR subtypes are coupled with the Gq protein, and their effect on neuronal activity is consistently excitatory.
In addition to the excitatory response, NA also produces hyperpolarization in a small number of AAV-NpyP + neurons. A similar hyperpolarizing effect of NA has also been reported in a part of GAD67-expressing SDH interneurons in mice [35]. Notably, the proportion of AAV-NpyP + neurons with hyperpolarization increased when α 1B -ARs were blocked or knocked down. The occlusion of NA-induced hyperpolarization by the α 2 -AR antagonist atipamezole leads to the possibility that some AAV-NpyP + neurons co-express both α 1B -ARs and α 2 -ARs and that the net activity of AAV-NpyP + neurons by NA stimulation could be determined by the balance between excitatory and inhibitory responses via α 1B -ARs and α 2 -ARs, respectively. Our data showing that NA caused depolarization in the majority of AAV-NpyP + neurons indicate that the net effect of NA in these neurons is excitatory via α 1B -AR activation.
Within the SDH, AAV-NpyP + interneurons receive excitatory inputs from Aβ fibers and transmit inhibitory signals to lamina I neurons that project to the brain [9]. The role of AAV-NpyP + neurons in spinal pain transmission and processing is evident in neuropathic pain conditions. After PNI, the RMP of AAV-NpyP + neurons deepen, and excitability is impaired. This alteration causes lamina I neurons to be excited by Aβ fiber stimulation, thereby leading to neuropathic allodynia. Conversely, chemogenetically depolarizing AAV-NpyP + neurons after PNI suppresses Aβ fiber-derived neuropathic allodynia-like behavior [9]. Thus, a depolarizing stimulus to AAV-NpyP + neurons could suppress neuropathic pain. This study demonstrated that duloxetine suppressed Aβ fiberevoked neuropathic allodynia-like behavior and that α 1B -ARs of AAV-NpyP + neurons are necessary for this effect. Considering that duloxetine increases spinal NA levels [17][18][19] and that the excitatory effect of NA on AAV-NpyP + neurons is retained after PNI, excitation of AAV-NpyP + neurons by spinal NA via α 1B -ARs is considered critical in the therapeutic effect of duloxetine on neuropathic allodynia-like behavior. NA has been reported to have multiple sites of action in SDH to modulate pain transmission and processing [11,37]. NA inhibits the presynaptic release of glutamate from C-fibers and excitability of SDH neurons via α 2 -ARs [20], although the role of α 2 -AR in the attenuating effect of duloxetine on neuropathic hypersensitivity remains controversial [19,22,38] (also see Introduction). Furthermore, NA also activates SDH astrocytes, a type of glial cells, via α 1A -ARs, and astrocytic α 1A -AR knockout enhances the attenuating effect of duloxetine on PNI-induced hypersensitivity [39]. The role of these ARs in the effects of duloxetine on neuropathic allodynia elicited by Aβ fibers will be an important subject for future research.

Conclusions
We identified AAV-NpyP + SDH neurons and α 1B -ARs as the critical targets of spinal NA, which contribute to the attenuating effect of duloxetine on neuropathic allodynia-like behavior and can be targets for the development of novel analgesics. As duloxetine is used to treat allodynia and other pain symptoms in patients with neuropathic pain [1,16], this study advanced our understanding of its pain-relieving mechanism and strengthens our view that enhancing the activity of AAV-NpyP + SDH interneurons is a potential pharmacological strategy for treating neuropathic pain.  Data Availability Statement: All data generated or analyzed during this study are included in the paper.