This study was performed on 46 adult C57BL/6 virgin female mice (6–8 weeks old, CERJ, Le Genest St. Isle, France). Their oestrous phase is about 12 h every 3–9 days, and the majority of mice are synchronized in our breeding house, so that the observed differences in this study are probably not due to hormonal differences in the two groups. The care and use of mice conformed to institutional policies and guidelines and was approved by the Animal Experimentation Ethics Committee of the University of Montpellier. Mice were housed in cages with a 12 h light/dark cycle and fed food and water ad libitum.
All animals were deeply anesthetized by isoflurane inhalation. Their left gluteal regions were shaved and cleaned with betadine. The left sciatic nerve was exposed at the mid-thigh without any damage to the muscle tissue and a 1 mm length of the sciatic nerve was crushed with Dumont #5 forceps for 15 sec. The animals were randomly divided into two groups. In group 1 there were 16 controls and 10 mice that were electro-acupuncture-treated 3 days after the nerve crush. To investigate the effects of electro-acupuncture during the period of pain-related behaviour (i.e. 15–28 days post-operation), we used a second group of mice, termed group 2, with 10 controls and 10 mice that were electro-acupuncture-treated two weeks after nerve crush. Group 1 was included for sensory tests and CatWalk analysis. Group 2 was assessed only for sensory testing.
The acupoints used were Hoantiao GB30 and Yanglinquan GB34. In humans, GB30 and GB34 are used to treat sciatic nerve pain and paralysis. They have been demonstrated to be analgesic acupoints in animal models of peripheral inflammatory pain  and molecular mechanisms involved in the analgesic effects of acupuncture have been postulated to contribute to nerve regeneration . According to World Health Organization standards, GB30 is located in the buttocks region, at the junction of the lateral one third and medial two thirds distance between the prominence of the greater trochanter and the hiatus of the sacrum . In the transpositional animal acupoint system for mouse and rat models, it is located at the depression superior to the greater trochanter of the femur . GB34 is located at the depression anterior and inferior to the fibular head.
For the group receiving electro-acupuncture, the following steps were performed: under isoflurane anaesthesia, electrical stimulation (2 Hz, rectangular pulse, 0.5 ms duration, intensity 0.8- 1 mA for 15 minutes) was applied via two acupuncture needles (diameter 0.2 mm, 2 cm length) inserted to a depth of 3 mm into the acupoints using an electrical stimulator (Improved KWD-808-II apparatus). The intensity of the stimulation was enough to produce a twitching of the hind leg. The low frequency stimulation used in the present study is reported to promote analgesic effects . The anode electrode was connected to GB 30 (the lateral side of the thigh, proximal to the crushed zone, “proximal” needle) and the cathode electrode was connected to GB 34 (between the crushed zone and distal end of the sciatic nerve, “distal” needle). This disposition was considered the most appropriate for an accelerated regeneration . For the control group, mice were anaesthetized for 15 minutes without acupuncture needles. In a study on peripheral nerve regeneration , no differences were reported between controls without needles and those having implanted needles without electricity. We did not stimulate non-acupoints because this generates a transcutaneous electrical stimulation, known to promote analgesia  and to interfere with nerve regeneration .
Following nerve crush (day post operation, DPO = 0), EA application for group 1 was performed at DPO 5, 7, 9 and 11, i.e. during the period of sensory-motor recovery. For group 2, EA application was performed at DPO 21, 23, 25 and 27, i.e. during the period of pain development.
Behavioural responsiveness of the mice was tested following one week of habituation to the testing environment and the observer. Two baseline measurements were taken on two separate days preceding the surgery. The mice were then tested every two days after surgery for 6 weeks.
Mechanical withdrawal thresholds
The paw withdrawal threshold in response to a mechanical stimulus was measured every two days using a series of graded von Frey filaments. Pressure applied ranged from 0.008 to 4 g beginning with the minimum intensity filament giving a positive response in the scale range for any animal. An ‘up-down’ method was then applied to determine the 50% withdrawal threshold, T50 . According to Dixon , optimal threshold calculation by this method requires six responses in the immediate vicinity of the 50% estimated threshold.
Nociceptive threshold to acute thermal stimulation was measured using the paw retrieval test. Focused light from a 12.5 W projection bulb was applied to the middle of the plantar surface of the hind paw (3 mm diameter) . The projection bulb was turned off as soon as the mouse removed its paw, and a digital timer connected in series measured the paw withdrawal latency to an accuracy of 0.1 s. We used a cut-off latency of 15 s to avoid the possibility of tissue damage.
Walking track analysis
To assess sensorimotor functional recovery, we used the CatWalk method for both static and dynamic gait analysis . Briefly, animals crossing a walkway with a glass floor are videotaped using a computer-assisted setup and digitized data of paw-floor contact area are used for off-line analysis. One week prior to the left sciatic nerve crush, mice were trained daily to cross the walkway. A normal run was defined as: the mouse crossing the walkway without any interruptions or hitches, the presence of footfall patterns for all four pads, and a running time of around one second to cover 45 cm. Three consecutive runs were recorded. A run in which only three pads were used (i.e. the left hind appears without any recorded trace on the foot floor) was not considered as a “normal” run. In the present study, only mice displaying a normal run were analysed. Gait was monitored every two days. Static parameters (intensity of the paw prints, print width and print length) and dynamic parameters (stance phase, swing phase, swing speed and duty cycle) were measured using CatWalk software 7.1 (Noldus Information Technology, St Louis, France). To eliminate any contribution of weight to the effects observed on the CatWalk, data are expressed as a percentage of respective contralateral values (i.e. compared to the right hind paw).
Results are presented as mean ± SEM. For comparison between the two groups, the data were analysed by two-way ANOVA followed by Tukey’s post-hoc test. The X
2 test was used for run analysis. A Mann–Whitney U test was used to compare independent values between groups. A p value < 0.05 was considered statistically significant.
To analyse any possible correlations between the sensory tests and the CatWalk method, a Pearson correlation test was performed.