All animal procedures were conducted in strict adherence to the guidelines of the International Association for the Study of Pain and were approved by the Animal Care and Use Committee of Health Science Center at Shenzhen University. 80 male C57BL/6 mice (5–8 weeks of age) were purchased from Guangdong Province Laboratory Animal Center (Guangzhou, China). 20 vGAT-ires-cre mice and 20 td-Tomato (Ai9) mice were purchased from Jackson Laboratory. The animals were housed in plastic cages (5 per cage) in a temperature-controlled environment on a 12 h/12 h light/dark cycle. Food and water were available ad libitum.

Oxytocin (catalogue: H-2510) and [d(CH2)51,Tyr(Me)2, Thr4,Orn8,des–Gly–NH29]–vasotocin (dVOT, catalogue: H-2510) were purchased from Bachem AG (Bubendorf, Switzerland). TC OT39 (catalogue: 1078) was obtained from Tocris (Minnesota, United States).

The partial sciatic nerve ligation (pSNL) pain model was established according to previously described procedures (Seltzer et al., 1990). Briefly, the animals were anaesthetized with sodium pentobarbital (50 mg/kg, i.p.) and a tight ligation of approximately one-third to one-half the diameter of the right sciatic nerve (ipsilateral) was performed with 6–0 silk suture. In sham-operated mice, the nerve was exposed without ligation.

Von Frey testing was performed to assess mechanical allodynia. The mice were habituated to the environment for 2 days before the testing began. All the behaviours were tested blindly. For testing mechanical allodynia, the mice were confined separately in boxes (14 × 18 × 12 cm) placed on an elevated metal mesh floor, and their hind paws were stimulated with a series of von Frey hairs with logarithmically increasing stiffness (0.16–2.00 g, Stoelting) situated perpendicularly to the central plantar surface. The 50% paw withdrawal threshold was determined by Dixon’s up-down method. The hot plate test (Hot/Cold Plate, Cat. 35150, Ugo Basile, Italy) was used to examine thermal hyperalgesia. Each mouse was placed on the hot plate, and the latency of paw withdrawal from the heat stimulus was measured twice separated by a 5-min interval. The average value was used as the latency of response. All behavioural testing was done with the experimenters blinded to the treatment conditions.

OT (0.1 μg in 10 μL) or dVOT (0.1 μg/10 μL) was injected into the subarachnoid space through the intervertebral foramen between L4 and L6 (Hylden and Wilcox, 1980). For the intrathecal infusion of drugs, an osmotic minipump (model 1003D, ALZET, Cupertino, CA, United States) connected with a polyethylene catheter was deposited in a subcutaneous pocket following partial sciatic nerve ligation. The other end of the catheter was inserted from the atlanto-occipital membrane into the subarachnoid space until the tip of the catheter reached the lumbar spinal enlargement. OT and other reagents were then delivered continuously with a flow rate of 1 μL/h for 3 days from days 0 to 2 after pSNL surgery. The final dose of OT intrathecal infusion is 0.3 μg in 100 μL. (The volume delivery rate and the delivery duration of ALZET pumps are fixed at manufacture).

The animals were sacrificed and L4–6 spinal cord segments were collected in tubes with RNAlater (Qiagen Inc., Valencia, CA, United States) and stored at −80°C until RNA isolation. Total RNA was isolated from these tissues according to Chomczynski’s method (Chomczynski and Sacchi, 1987) and reverse transcribed using Omniscript reverse transcriptase (Qiagen Inc., Valencia, CA, United States) at 37°C for 60 min. The reaction was performed in the presence of the RNase inhibitor rRNAsin (Promega, Madison, WI, United States) and an oligo (dT16) primer (Qiagen) to selectively amplify the mRNA. For quantitative PCR, 45 ng of cDNA was used as a template. Reactions were performed using Assay-On-Demand TaqMan probes and TaqMan Universal PCR Master Mix (Applied Biosystems, Foster, CA, United States) according to the manufacturer’s protocol. Reactions were run on a Real-Time PCR iCycler IQ (Bio-Rad, Hercules, CA, United States) with software version 3.0. The expression levels of Kcc2 were normalized to ß-actin.

The animals were sacrificed, and the L4-6 spinal cord segments were removed and stored at −80°C until assayed. The samples were homogenized and centrifuged to extract the protein, and the resulting preparations were saved. Equal amounts of protein were separated by 10% Tris-Tricine SDS-PAGE and transferred onto polyvinylidene difluoride membranes. The membranes were then blocked in 5% non-fat milk for 1 h at room temperature, followed by overnight incubation with rabbit anti-KCC2 antibody (1:1000; ab49917, Abcam, United States) and ß-actin (1:2000; Sigma, United States) primary antibody. Immunoblots were then incubated for 1 h at room temperature with goat anti-rabbit polyclonal IgG (1:3000, ab205718, Abcam, MA, United States). Immunoblots were developed by chemiluminescent substrate and quantified using ImageJ software.

The mice were deeply anesthetized with isoflurane and transcardially perfused with PBS followed by 4% PFA. Lumbar L4-6 spinal cord segments sections were blocked and then incubated overnight at 4°C with rabbit antibodies against KCC2 (Abcam, ab49917, United States). The sections were then incubated for 30 min at 37°C with AF488-conjugated secondary antibodies (donkey, 1:500, Jackson Immuno-Research, West Grove, PA, United States), and the nuclei were stained with DAPI. The sections were viewed under Zeiss 880 inverted confocal microscopy, and images were collected using identical acquisition parameters and quantified using Image-Pro Plus 6.0 software (Media Cybernetics, Silver spring, MD, United States) by experimenters blinded to treatment groups.

In situ hybridization was performed using the RNAscope system (Advanced Cell Diagnostics) following the manufacturer’s protocol. Pre-treatment consisted of dehydration, followed by incubation with hydrogen peroxide and protease IV at room temperature. The Multiplex Fluorescent Kit v2 protocol was followed using commercial probes for the OT receptor (Oxtr, NM_001081147.1, #402658-C3). Images were captured by Zeiss 880 inverted confocal microscopy. Visualized cells with more than 5 puncta per cell were classified as positive neurons.

Adult (5–7 weeks) male mice were anaesthetized with urethane (1.5–2.0 g/kg, i.p.). The lumbosacral spinal cord was removed and submerged into ice-cold dissection solution saturated with 95% O₂ and 5% CO₂ at room temperature. Transverse slices (300–400 μm) were cut in a vibrating microslicer (VT1200s Leica). The slices were incubated at 32°C for at least 30 min in regular artificial cerebrospinal fluid (aCSF) equilibrated with 95% O₂ and 5% CO₂.

The following solutions were used: dissection solution containing (in mM) 240 sucrose, 25 NaHCO₃, 2.5 KCl, 1.25 NaH₂PO₄, 0.5 CaCl₂, and 3.5 MgCl₂ at pH 7.4; regular artificial CSF containing 135 NaCl, 2.5 KCl, 3 MgCl₂, 1 CaCl₂, 10 HEPES, 1 NaH₂PO₄, and 10 glucose at pH 7.4; and normal intrapipette solution for perforated recording containing 115 K-methylsulfate, 25 KCl, 2 MgCl₂, 10 HEPES, 0.4 GTP-Na and 5 Mg-ATP at pH 7.2 and 310 mOsm.

To measure the reversal potential of GABA-evoked currents, a slice was placed in the recording chamber and completely submerged and superfused at a rate of 2–4 ml/min with aCSF. A perforated patch-clamp was applied to avoid changes in the [Cl⁻]_i. To measure the chloride equilibrium potential (E_Cl), gramicidin D (80 μg/ml with an 0.8% DMSO final concentration from an 8 mg/ml stock in DMSO) was added to the intrapipette solution, and 6-cyano-7- nitroquinoxaline-2,3-dione (CNQX, 10 μM), DL-2-amino-5-phosphonovaleric acid (APV, 50 μM) and tetrodotoxin (TTX, 0.5 μM) were added to the aCSF solution. The tip of the patch pipette was filled with the normal intrapipette solution, while the rest of the pipette contained the gramicidin-containing solution. After forming a seal on the membrane, we waited 30 min for the gramicidin to effectively reduce the series resistance to below 100 MΩ. Membrane potential measurements were corrected for liquid junction potential, which was measured as in(Guo et al., 2014). GABA (1 mM) was puffed locally and instantaneously, and the puff pipette was aimed toward the recording pipette. Voltage ramps were applied from +8 to −92 mV over 200 ms at a holding potential of −42 mV. Since the voltage ramp might evoke a basal current, a control voltage ramp was first applied to record the basal current; 1 min later, GABA was puffed, followed by another voltage ramp, and then the GABA-evoked currents were recorded (Billups and Attwell, 2002). The reversal potential was analysed as in (Billups and Attwell, 2002).

Excitatory and inhibitory post-synaptic currents (EPSCs and IPSCs) recordings were made from lamina II inhibitory neurons. The patch-pipette solution contained (in mM) K-gluconate 135, KCl 5, CaCl₂ 0.5, MgCl₂ 2, EGTA 5, HEPES 5, an Mg-ATP 5; or Cs₂SO₄ 110, CaCl₂ 0.5, MgCl₂ 2, EGTA 5, HEPES 5, Mg-ATP5, tetraethylammonium (TEA)-Cl 5 (pH = 7.2) (Jiang et al., 2014). The former and latter solutions were used to record EPSCs and IPSCs, respectively. EPSC recordings were made at a holding potential (V_H) of −70 mV, where no IPSCs were observed, since the reversal potential for IPSCs was near −70 mV. IPSCs were recorded at a V_H of 0 mV, where EPSCs were invisible as reversal potential for EPSCs was close to 0 mV. Cs⁺ and TEA were used to block K⁺ channels expressed in the recorded neurons, and thus to easily shift V_H from −70 to 0 mV. GABAergic IPSCs were obtained in the presence of the glycine-receptor antagonist strychnine (1 mM). EPSC and IPSC events were detected and analysed using Mini Analysis Program 6.0. Signals were acquired using an Axopatch 700B amplifier and analysed with pCLAMP 10.3 software. Only neurons with resting membrane potential < −50 mV and stable access resistance were included.

The data are expressed as means ± SEM and analysed with a t-test or variance (ANOVA) using one-way or mixed factorial designs as appropriate, followed by Bonferroni’s post hoc test or simple-effects ANOVA. All statistical analyses were performed using GraphPad Prism 8.0. (GraphPad Inc., La Jolla, CA, United States). Significance was defined as p < 0.05.

pSNL-induced nerve injury produced mechanical allodynia and thermal hyperalgesia in mice. This mechanical and thermal hypersensitivity started on day 1 and remained relatively stable from days 3 to 14 after nerve ligation (Supplementary Figures S1A,B).

An osmotic minipump was implanted immediately following partial sciatic nerve ligation. OT was then delivered with a flow rate of 1 μL/h for 3 days from days 0–2 after pSNL surgery. Mechanical allodynia and thermal hyperalgesia were tested at days 3, 5, 7 and 14 after pSNL surgery (Figure 1A). As shown in Figures 1B,C, infusion of OT (0.3 μg, 100 μL) for 3 days before the behavioural tests decreased nerve injury-induced nociceptive behaviours in mice. Compared with the vehicle, 3-days continuous infusion of OT increased the mechanical threshold in the von Frey test [F(1,14) = 61.57, p < 0.001; Figure 1B, n = 8] and paw withdrawal latency in the hot-plate test [F(1,14) = 50.74, p < 0.001; Figure 1C, n = 8] for 14 days, which was the longest period we tested, indicating that 3-days continuous intrathecal OT infusion may attenuate the establishment and development of nerve injury-induced neuropathic pain.

In comparison, the effect of a single injection of OT on pSNL-induced mechanical and thermal hypersensitivity was also tested on day 3 after nerve ligation, when the pain behaviours were well established (Figure 1D). Single intrathecal OT (0.1 μg/10 μL) significantly alleviated pSNL-induced mechanical allodynia [F(1,14) = 42.59, p < 0.001; Figure 1E] and thermal hyperalgesia [F(1,14) = 29.66, p < 0.001; Figure 1F] at 10 [p < 0.001] and 30 min [p < 0.001] after injection. This effect of OT was not observed at 60 min after the injection [p > 0.05; Figures 1E,F], indicating that the analgesic effect of a single intrathecal OT administration on nerve injury-induced pain behaviours is transient. OT at the doses used in the present study had no effect on the locomotor activity or motor coordination in mice (date not shown).

We found no significant differences between male and female mice in the analgesic effects of oxytocin [p > 0.05; Supplementary Figure S4].

To determine whether the effects of 3-days OT infusion on neuropathic pain were mediated by Oxtrs, its agonist or antagonist was administrated (Figure 2A). Co-intrathecal infusion (100 μL) of a selective Oxtr antagonist, dVOT (0.3 μg), with OT (0.3 μg) blocked the analgesic effect of OT on nerve injury-induced mechanical [F(1,13) = 25.04, p = 0.0002; Figure 2B, n = 7–8] and thermal hypersensitivity [F(1,12) = 28.92, p < 0.001; Figure 2C, n = 7]. The selective Oxtr agonists TC OT (0.3 μg/100 μL) produced significant analgesic effects which were equivalent to OT [von Frey test F(1,14) = 15.42, p = 0.0015; Hot-plat test F(1,14) = 29.80, p < 0.0001; Figures 2D,E; n = 8]. There results suggested that the 3-days intrathecal infusion of OT induced analgesic effect is mediated by the Oxtrs in the spinal cord.

It was reported that neuronal intracellular chloride concentration was increased in the superficial dorsal horn after nerve injury (Yeo et al., 2021), we performed perforated patch-clamp recording in spinal cord slices derived from each group to investigate the effects of OT on chloride homeostasis (Figure 3A). Since GABA_A receptor (GABA_AR) is the dominant chloride ion channel on the membrane of neurons in the superficial dorsal horn, GABA was puffed briefly to the recorded neuron to trigger transient chloride influx or efflux.

As voltage ramps were applied from +8 to −92 mV (Figure 3C), the GABA-evoked currents were recorded to evaluate chloride equilibrium potential (E _Cl ^- ). These currents were completely blocked by a selective GABA_AR antagonist, bicuculline (10 μM), confirming that they were mediated by GABA_AR (data not shown). The E _Cl ^- in sham mice was −66.68 ± 1.22 mV (Figures 3B–E, n = 5–6, 3 mice per group), whereas that value in pSNL mice shifted to a more positive value of −43.54 ± 1.67 mV [ p < 0.001 vs. sham group; F(2,12) = 36.26, p < 0.001; Figures 3B–E, n = 5 from 3-4 mice]. Continuous intrathecal infusion of OT reversed the value of E _Cl ^- to −59.02 ± 2.69 mV, which was much closer to that of the sham mice [p > 0.05 vs. sham; Figures 3B–E, n = 5 from 3-4 mice], suggesting that 3-days infusion of OT was able to restore [Cl⁻]_i in pSNL mice.

In comparison, we also recorded the E _Cl ^- using the spinal cord slices incubated with saline or OT for 30 min (short-term application, Figure 4A), and the reversal potentials were −44.34 ± 2.91 mV and −46.10 ± 3.10 mV, respectively [p > 0.05 vs. saline; F(2,12) = 31.71, p < 0.0001; Figures 4B–E]. Incubation of the spinal cord slices with OT for a relatively short time failed to restore the value of E _Cl ^- in pSNL mice, suggesting that the effect of OT on E _Cl ^- required relatively long-term application.

Given that the shift of E _Cl ^- in pSNL animals may be due to depressed function of KCC2, we analysed the transcriptional and expression levels of KCC2 in the spinal cord. Compared with the sham group, quantitative PCR data revealed a significant decrease in spinal Kcc2 mRNA levels at both days 7 and 14 after pSNL surgery [p < 0.001 vs. sham; F(2,16) = 3.818, p = 0.0441; Figure 5A, n = 5 per group]. Intrathecal infusion of OT increased spinal Kcc2 mRNA levels in pSNL mice compared with saline group [p < 0.01; F(2,16) = 3.818, p = 0.0441; Figure 5A, n = 5 per group].

Western blotting data also showed that nerve injury-induced a significant decrease in the protein levels of KCC2 in the spinal dorsal horn at days 7 and 14 after pSNL surgery [p < 0.0001 vs. sham; F(2,16) = 8.982, p = 0.0024; Figures 5B,C, n = 5 per group]. Intrathecal infusion of OT restored the protein levels of KCC2 but did not completely reverse this decrease [p < 0.01 vs. saline; F(2,16) = 8.982, p = 0.0024; Figures 5B,C, n = 5 per group]. Immunohistochemistry (IHC) of spinal slices from laminae II further supported the western blotting data, which showed that the KCC2 signal was widely expressed throughout the spinal dorsal horn in sham mice (Figure 5D). Nerve injury-induced a reduction in KCC2 expression at days 7 and 14 after pSNL surgery [p < 0.0001 vs. sham; F(2,12) = 8.119, p = 0.0059; Figures 5D,E]. Infusion of OT reversed this reduction [p < 0.01 vs. sham; F(2,12) = 8.119, p = 0.0059; Figures 5D,E, n = 4 per group] to some extent.

To further explore the underlying mechanism of OT on the regulation of E _Cl ^- , we performed a novel in situ hybridization assay (RNAscope) to investigate the feature of Oxtr mRNA expression. Firstly, we used a novel in situ hybridization assay (RNAscope) to detect the properties of Otxr mRNA distributions in the superficial dorsal horn. As shown in Supplementary Figure S2, Oxtrs mRNA (white) were not expressed on microglia (green) and astrocytes (red), suggesting that majority of Oxtrs are located in the neurons. To test whether that Oxtrs were expressed on the inhibitory neurons in the spinal dorsal horn. Spinal cord slices derived from the vGAT-tdTomato mice were used, in which the inhibitory neurons were visualized by red fluorescence. As shown in Figures 6A,B, about 30% of vGAT + neurons (inhibitory neurons) expressed Oxtrs mRNA signalling in the in the spinal dorsal horn. Oxtr mRNAs were also found expressed in vGAT negative interneurons in the superficial dorsal horn.

We then performed whole-cell voltage clamp on the vGAT positive interneurons in the superficial dorsal horn. About 72% recorded vGAT⁺ neurons (n = 18) produced an inward current when OT (0.5 μM) was perfused for 3 min at the V_H of −70 mV with an average of −10.40 ± 1.27 pA (upper trace in Figures 6C–E), but OT did not change the frequency and amplitude of spontaneous EPSCs in all of the examined vGAT⁺ neurons [t-test, p = 0.0663, t (34) = 1.963 for frequency; p = 0.6311, t (34) = 0.4890 for amplitude; Figure 6F]. In the presence of the Oxtr antagonist dVOT (1 μM), OT failed to induce an inward current in all recorded vGAT positive interneurons in the superficial dorsal horn (Figures 6G,H, n = 12). In comparison, OT perfusion produced an inward current in 38% recorded vGAT negative neurons (Supplementary Figures 3B,C, n = 13).

Due to OT produced inward currents in some vGAT positive interneurons, we tested the effects of OT on GABAergic transmission in the spinal cord in the presence of a glycine-receptor antagonist, strychnine (1 μM). OT (0.5 μM) perfusion for 3 min increased the frequency and amplitude of spontaneous GABAergic IPSCs at the V_H of 0 mV from 5.02 ± 0.49 Hz to 13.61 ± 1.72 Hz and 9.40 ± 0.68 pA to 13.17 ± 1.30 pA, respectively (t-test, p = 0.0009, t (12) = 6.026 for frequency; p = 0.0080, t (12) = 3.899 for amplitude; n = 7; Figures 7A,B). Expectedly, OT enhanced GABAergic spontaneous transmission was total blocked by pre-treatment with a selective Oxtr antagonist, dVOT (1 μM, p = 0.2498, t (12) = 1.274 for frequency; p = 0.2987, t (12) = 1.138 for amplitude; n = 7; Figures 7C,D).