Eight-week-old 150–180 g SD rats were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. The operation was carried out following the Guidelines for Ethical Review of Animal Welfare in China (GB/T 35892-2018). All experiments were approved by the Medical Ethics Committee of Peking University People's Hospital (Permit Number: 2020PHC015, 3/1/2019). The rats were anesthetized with sodium pentobarbital intraperitoneally (Merck, Darmstadt, Germany; 30 mg/kg) before random grouping. In Group I (n = 9), the distal end of the common peroneal nerve was coapted end-to-side to the fenestrated tibial nerve adventitia, and the proximal end was ligated and fixed in the nearby muscle. In Group II (n = 9), the tibial nerve adventitia was fenestrated and epineurial end-to-end anastomosis was performed to anastomose the proximal and distal ends of the common peroneal nerve. Rats in Group III (n = 9) were taken as control and underwent sham operation. After awakening, the rats were fed in groups, and food and drinking water were automatically supplied. A 12:12 light/dark natural circle was adopted in the feeding environment.

The motor function of rats was measured by the CatWalk XT gait analysis system (Nodus, Netherlands), which can automatically record the motion parameters. The camera was placed at the correct position and corresponding rat parameters were setup. Before recording, the rats were trained to be familiar with the running environment. All rats in different groups were recorded automatically. The calculation formula of sciatic functional index (SFI) is SFI = 109.5 (ETS–NTS)/NTS – 38.3 (EPL–NPL)/NPL + 13.3 (EIT–NIT)/NIT – 8.8. SFI is a reliable method for the assessment of the extent of injury and the degree of recovery of the sciatic nerve of the rat. E: experimental; N: normal; PL: print length from heel to longest toe; TS: total spread, or the transverse distance between the 1st and the 5th toes; IT: intermediate toes, or the transverse distance between the 2nd and 4th toes.

The compound action potential of gastrocnemius and tibialis anterior muscles was detected by the Medlec synergy electrophysiological system (Oxford Instruments Inc., Oxford, UK) in each group. After the rats in each group were anesthetized with sodium pentobarbital, the repaired nerves were exposed. Then the stimulation electrodes were placed at positions A, B, C, D, and E of the nerve (Figure 1), and the electrical signals of gastrocnemius and tibialis anterior muscles were recoded through receiver electrode, separately. In the meantime, the action potentials of gastrocnemius and tibialis anterior muscles recorded at point A were taken as the maximum action potentials of Group I and Group II. The action potential of the gastrocnemius muscle recorded at point A and the action potential of the tibialis anterior muscle recorded at point B were adopted as the maximum action potentials of Group III. Statistical analysis was subsequently made to evaluate the size of the compound muscle action potential (CMAP) of each group. Rectangular pulse (0.1 ms duration, 0.06 mA intensity, 5 Hz frequency) was used to acquire the CMAP.

The tibial nerve and common peroneal nerve 2 mm away from the anastomosed common peroneal nerve were collected. The nerve sample was soaked in 4% paraformaldehyde for 12 h and then dyed with 1% osmic acid for another 12 h. After the sample was dehydrated with 50, 70, 90, and 100% ethanol successively, it was embedded in paraffin, and sliced into a 2 μm-thick piece (Leica RM2135, Wetzlar, Germany). The piece was placed on the glass slide. After sealing the slide, it was observed under the microscope [Leica DM4B with Leica Application Suite X (LAS X) software, Wetzlar, Germany]. The number of myelinated nerve fibers per unit area in each group was manually counted.

The gastrocnemius and tibialis anterior muscles of the model and opposite sides were collected from SD rats and weighed. The ratio of the wet weight of the model side muscle to that of the opposite side muscle was calculated. After trimming, the muscle was placed in 4% paraformaldehyde for fixation overnight, and then dehydrated successively with 50, 70, 90, and 100% ethanol. The muscle was embedded in paraffin, and sliced into a 7 μm-thick piece. Then Masson staining was performed. Those for Masson staining were treated with Weigert stain for 8 min and rinsed with distilled water. The sections were then treated with spring red acid complex red for 5 min, immersed in acetic acid for 1 min, treated with phosphomolybdic acid for 2 min, and immersed in acetic acid for 1 min. Next, the sections were treated with aniline blue for 2 min, immersed in acetic acid for 1 min, and rinsed with water. After sealing the slide, it was imaged under a microscope [Leica DM4B with Leica Application Suite X (LAS X) software, Wetzlar, Germany]. The cross-sectional area of gastrocnemius and anterior tibial muscle fibers was calculated using Image J software. Three visual fields of each sample were selected, and the average value was taken as the ultimate muscle fiber cross-sectional area of the sample.

Three SD rats were taken from each group for the tracing and labeling experiment, and the method proposed in (12) was adopted. The regenerated sciatic nerve of rats was re-exposed and cut off. The distal ends of the common peroneal nerve and tibial nerve were immersed in 5% Fluoro-Ruby (FR, Life Technologies) and 4% Fluoro-Gold (FG, Life Technologies) for 2 h, respectively. The wound was washed and sutured layer by layer. Then rats were fed for 7 days so that the retrograde tracer could label the spinal cord neurons. After anesthetizing, the heart of rats was isolated and perfused. L4–L6 lumbar spinal cord segments containing sciatic nerves were cut off and fixed in 4% paraformaldehyde for 12 h. The sample was dehydrated in 10, 20, and 30% sucrose for 12 h, and then sliced to a 40 μm-thick piece by a frozen slicer (Leica CM 1950, Germany). Subsequently, the sample was placed on the glass slide and sealed, and the slice was imaged by the 20 × microscope objective. The excitation wavelength was FG330nm and FR561nm. Finally, the acquired fluorescence image was analyzed using Image J software to detect the neurons labeled by FG only, FR only, and both FG and FR.

All measurement data were expressed by (mean ± standard deviation). One-way ANOVA was employed to analyze data in different groups, followed by Tukey's post-hoc multiple comparison test. SPSS 18.0 was used for statistical processing. P < 0.05 meant that the difference was statistically significant.

The proximal and distal common peroneal nerves at the repair site were observed in vivo and in vitro. In Group I, only the distal common peroneal nerve remained intact, while in Group II, both the proximal and distal ones were complete (Figure 2).

Since the common peroneal nerve and tibial nerve constitute the sciatic nerve, the sciatic functional index can be used to detect the recovery of nerve function. In this paper, the regeneration and repair of the common peroneal nerve were analyzed using the sciatic functional index. The results showed that the sciatic functional indexes of Group I, Group II and Group III were (−30.09 ± 5.9, n = 6), (−11.64 ± 5.12, n = 6), and (−0.05 ± 4.09, n = 6), respectively. Significant statistical difference was found (p < 0.05). The sciatic functional index of Group II was better than that of Group I and closer to that of Group III (Figure 3).

In clinical practice, electromyography is often used to record the electrophysiological characteristics of nerves and muscles. In this study, the functional recovery of the repaired nerves and their innervating effectors were detected by electrophysiology. The CMAP size depends on the number of functional muscle fibers. The detection results suggested that the waveforms of gastrocnemius and tibialis anterior muscles in Group I, Group II, and Group III were not different. The CMAP amplitudes of the gastrocnemius muscle in Group I, Group II and Group III were (26.80 ± 3.10) mV, (27.30 ± 1.80) mV, and (29.00 ± 3.20) mV, respectively, and no significant difference was found (p > 0.05). The CMAP amplitude of the tibialis anterior muscle in Group I, Group II and Group III was (8.70 ± 1.60) mV, (12.20 ± 2.10) mV, and (19.70 ± 1.90) mV, respectively, and there was significant statistical difference (p < 0.05). The CMAP amplitude of the tibialis anterior muscle in Group II is closer to Group III than Group I is to Group III (Figure 4).

Stimulation electrodes were placed at different positions to check whether the proximal end of the common peroneal nerve was fostered on the tibial nerve, and whether the proximal common peroneal nerve could control the anterior tibial muscle or even the gastrocnemius muscle. According to the results, stimulating point A could cause the contraction of gastrocnemius and anterior tibial muscles in Group I and Group II. When stimulating point B, the gastrocnemius and anterior tibial muscles in Group II tightened. The gastrocnemius and anterior tibial muscles in Group I and Group II constricted as the point C was stimulated. Stimulating point D could lead to the contraction of the gastrocnemius muscle in Group I, Group II and Group III. When the point E was stimulated in Group I, Group II and Group III, the anterior tibial muscle tightened (Table 1).

The numbers of myelinated tibial and common peroneal nerve fibers per unit area were calculated using osmium acid staining in each group. The number of myelinated tibial nerve fibers per unit area in Group I, Group II and Group III was (14,883 ± 835) mm²/piece, (14,783 ± 1,091) mm²/piece and (15,017 ± 608) mm²/piece, respectively, and no significant difference is observed (p > 0.05). The number of myelinated common peroneal nerve fibers per unit area in Group I, Group II and Group III was (6,750 ± 903) mm²/piece, (8,967 ± 1,269) mm²/piece, and (14,917 ± 488) mm²/piece, respectively, and there was significant statistical difference (p < 0.05). The number of myelinated common peroneal nerve fibers per unit area in Group II is closer to Group III than Group I is to Group III (Figure 5).

The wet weight ratio and the cross-sectional area of gastrocnemius and tibialis anterior muscle fibers were measured using gross specimen and Masson staining methods. The wet weight ratio of gastrocnemius muscle in Group I, Group II and Group III was (0.99 ± 0.02), (0.99 ± 0.02), and (1.00 ± 0.02), respectively. The cross-sectional area of gastrocnemius muscle fibers was (2,025 ± 121) μm², (2,011 ± 76) μm², and (2,061 ± 107) μm², respectively. Both the wet weight ratio and fiber cross-sectional area of gastrocnemius muscle were not significantly different between Group I and Group II (p > 0.05). The wet weight ratio of tibialis anterior muscle in Group I, Group II and Group III was (0.46 ± 0.08), (0.64 ± 0.08), and (0.99 ± 0.02), respectively. The cross-sectional area of tibialis anterior muscle fibers was (1,131 ± 130) μm², (1,325 ± 103) μm², and (1,971 ± 90) μm², respectively. Significant statistical difference was found between Group I and Group II in the wet weight ratio and fiber cross-sectional area of gastrocnemius muscle (p < 0.05). The wet weight ratio and fiber cross-sectional area of tibialis anterior muscle in Group II is closer to Group III than Group I is to Group III (Figures 6, 7).

Retrograde tracing of neuron was carried out to explore the reinnervation connection between neurons and effectors. There were only FG-labeled, both FG- and FR-labeled fluorescent neurons in the spinal cord of Group I. But there were no only FR-labeled fluorescent neurons. In Group II, only FG-labeled, only FR-labeled and both FG- and FR-labeled fluorescent neurons were observed in the spinal cord. In Group III, only FG-labeled and only FR-labeled fluorescent neurons were found in the spinal cord, but there were no fluorescent neurons labeled by both FG and FR (Figure 8).