Establishment of a facial nerve trunk crush injury model and evaluation of facial nerve self‐healing in rats

Abstract Introduction/aims To facilitate further investigation into the mechanisms of facial nerve regeneration, a simple and reliable model of facial nerve crush injury is essential. Nevertheless, the establishment of such models lacks standardization and repeatability, while the healing capacity of the nerve is often overlooked, potentially affecting future studies. Methods We made facial nerve trunk crush injury models with different pressing times and detected the changes from the distal nerves to the motoneurons via behavior analysis, electrophysiological test, and histomorphometry analysis. Results It revealed a particular capacity for self‐healing following facial nerve crush damage because there was almost no facial motoneuron apoptosis in the MC group during the observation period, and rats in MC group had total facial paralysis in behavioral tests following surgery and varying degrees of recovery 28 days postoperatively with no treatments. As the pressing time increased, the latency, wave amplitude, nerve fiber damage degree, nerve axon ratio, myelin thickness, electroneurograph (ENoG) value, ultrastructural damage, abnormal morphological changes, and the buccal muscle atrophy of each MC group gradually increased or got worse during the observation period. However, after 28 postoperative days, only the ENoG values of the M10min and M12min groups were beyond 90%, indicating no self‐healing. Discussion It suggests that a stable model of peripheral facial palsy may be created by applying a 12.5 cm mosquito clamped to the facial nerve trunk for at least 10 min, which laid the foundation for the subsequent research to objectively evaluate facial nerve regeneration.


INTRODUCTION
As it affects 20-30 per 100,000 persons each year, the prevalence of facial nerve dysfunction can be rather high (Ali et al., 2020). In China, the incidence of facial palsy is 49.77 per 100,000 people 1 year (Neurophysiological Monitoring Group, 2022). In those who are affected, the loss of facial nerve function results in emotional and practical issues, including embarrassment, functional limitations, low self-esteem, and a poor quality of life (Toraman et al., 2021). Although many studies have investigated the mechanisms of facial nerve damage and restoration, more research are necessary to achieve better outcomes. Developing a standardized facial nerve injury model would be of significant value to the otolaryngologists or plastic surgeons interested in facial nerve regeneration.
Currently, there is no standardized approach for creating facial palsy models, which can be produced using various methods, including compression (Cai et al., 2021), transection (Fujii et al., 2020), cold stimulation (Joko et al., 2020), and inoculation virus method (Mu et al., 2022). Peripheral nerve injury in maxillofacial surgery most frequently manifested as compression and contusion (Mourad et al., 2022). Hence, we believe that, unlike transection injury, the crush injury model produces consistent damage that is reproducible and clinically relevant.
However, there is currently no standardized protocol regarding the age of animals used for modeling or the duration and extent of facial nerve entrapment.
On the other hand, most experimental researches on facial nerve injury rely solely on the compressed section of the facial nerve that was obtained during the modeling process or on the animal's symptoms to evaluate the effectiveness of the model (Zhang et al., 2012). If the selfhealing of the facial nerve compression model is not taken into account, the therapeutic efficacy of the model may be assessed solely based on the presence or absence of symptoms, which can affect subsequent stages of the study. Therefore, it is crucial to determine whether the facial palsy model undergoes self-healing during the experiment.
A comprehensive analysis about the self-healing of facial nerves after the compression injury from these aspects, including animal behavior, neuron electrophysiology, neurohistopathology, atrophy of target muscles, and apoptosis of central motoneurons, has not been reported at present. According to the previous literature, in our experiment, we set up six groups of different compression durations of the facial nerve trunk for 10 s, 1 min, 5 min, 8 min, 10 min, and 12 min.
Through this approach, we were able to establish a reproducible and reliable compression model for further facial palsy research. This model will serve as a basis for studying the mechanisms underlying facial palsy recovery.

Animals and groups
A total of 104 Sprague Dawley (SD) rats, both male and female, aged 4-5 months, weighing 250-300 g/each were provided by the Exper-

Surgical procedure
The surgical procedures were carried out in an aseptic environment.
Ketamine (40 mg/kg) and xylazine (5 mg/kg) were used to anesthetize the rats intraperitoneally. After that, the facial nerve, which emerges from the stylomastoid foramen, was tightly clamped 2-mm-long section with a 12.5-cm mosquito forcep held perpendicular to the nerve's longitudinal axis (Figure 1a). The trunk was continuously pressed for 10 s, 1 min, 5 min, 8 min, 10 min, and 12 min, respectively. After multilayer suture and disinfection, the incision was closed. Only the right facial nerve trunk was exposed in the NC group during surgery, and the incision was stitched together in layers.

Behavioral analysis
The behavior assessment was performed in each group on 1, 7, 14, 21, and 28 days postoperatively. Twice blowing air over the eyelids caused the blink reflex to occur with a 5-mL syringe connected to an 18-gauge needle, placed around 3 cm from the eyes. On a scale of 0-2, the blink reflex intensity was evaluated (0, no difference between the two sides; 1, a delay in the blink reflex in comparison to the unaffected side; 2, a complete disappearance of the blink reflex). After 30 s of observation, the vibrissae's movement was rated on a scale from 0 to 2 (0 signifies no change between the two sides; 1, a weaker vibrissae movement than on the healthy side; 2, a disappearance of vibrissae movement). A score between 0 and 1 was assigned for the positioning of the apex nasi (0 meant it was in the center; 1 told it was toward the unaffected side) (Cai et al., 2021). The final scores were calculated as the sum of all scores.
The NC group had a total score of 0, showing no signs of facial nerve paralysis. Symptoms of facial paralysis were taken into account when the scores on aggregate were higher than 3 ( Figure 1c).

Electrophysiological assessment
At 10 min and 28 days postoperatively, animals in each group were anesthetized for the electrophysiological tests. At the intersection of F I G U R E 1 (a) The right facial nerve trunk was exposed in surgery. (b) Functional scores at different time points in each group. Note: The animal behavior in the normal control (NC) group received a score of 0 at all time points, while each MC group showed a gradual decrease in scores at each postoperative time point. As the pressing time was prolonged, the scores of each MC group gradually increased and showed significant differences between groups at each time point. There were no significant differences between the MC group at any time point. #p < 0.05 versus NC. (c) The performance of peripheral facial palsy in each group. Note: The placement of the apex nasi in NC group rats was in the center, and the beards on both sides are symmetrical and move normally. After surgery, rats in each MC group showed peripheral facial palsy with crooked mouth, skewed apex nasi, and weakened beard movement. At 28 days after surgery, the symptoms of facial palsy in each MC group nearly disappeared.
the facial nerve's buccal branch and the canthus vertical line, two stimulating electrodes were placed, and the collecting electrodes were positioned subcutaneously on the right side of the orbicularis oris muscle in order to measure the compound muscle action potentials (CMAPs) of the injured facial nerves. A reference electrode was placed subcutaneously on the tail. The parameters were adjusted as follows: a square wave with a 0.2 ms wavelength and a 2.5 V stimulation intensity.
Then the peak amplitude and latent period in each rat were recorded.
Normal CMAPs were obtained from the NC group.
The latency stage before facial nerve trunk injury was recorded as La. The latency stage after surgery was recorded as Lb, and 28 days postoperatively it was recorded as Lc. The peak value of Mwave was represented as A. Before surgery, it was recorded as Aa, and it was recorded as Ab after injury. The peak value of the M-wave was recorded as Ac 28 days postoperatively. The ENoG value after surgery equals (1 − Ab/Aa) × 100%, and the ENoG value 28 days postoperatively equals (1 − Ac/Aa) × 100% .

Tissue sample preparation
Rats (n = 4) were randomly selected from each group immediately following the surgery and 28 days postoperatively. Animals were transcardially perfused with 4% paraformaldehyde to be pre-fixed under deep anesthesia. The facial nerve and buccal muscle tissues were collected, drained, made translucent, waxed, and embedded into paraffin blocks after being treated in 4% paraformaldehyde. Then the rats were placed on a brain stereotaxic instrument to determine the location of the facial motoneurons. The neck was broken, the temporal bone was opened, the brainstem was removed, and the brainstem tissues F I G U R E 3 HE staining of facial nerve in each group after surgery and 28 days postoperatively. Note: In the normal control (NC) group, the nerve membranes were intact, and the axons were neatly arranged (a). After surgery, a series of pathological changes such as nerve fiber degeneration, necrosis, hemorrhage, nerve bundle membrane fracture, axon fracture, and myelin vacuolization appeared in each MC group (b). At 28 days postoperatively, the facial nerve damage did not recover completely. The degree of nerve damage gradually increased with the prolongation of the facial nerve compression time. The red arrow in the picture points to the outer nerve membrane, the green containing the facial motoneurons (up to below the inferior colliculus and down to the pontine sulcus) were cut off with a blade. Frozen sections were made, followed by continuous coronal sectioning and observation under a light microscope until the facial motoneurons appeared. The remaining rats (n = 4) from each group were transcardially perfused with 2.5% glutaraldehyde under deep anesthesia immediately following surgery and 28 days postoperatively for transmission electron microscopy (TEM). The facial nerves that had been removed were treated in 2.5% glutaraldehyde.

Pathological evaluation of facial nerve
Routine procedures with Hematoxylin-Eosin (HE) staining included dewaxing and hydrating paraffin sections of the facial nerve, staining with hematoxylin for 10 min, rinsing in tap water for 5 min, fractionating in hydrochloric acid alcohol fractionation solution for 30 s, soaking in tap water for 3 min, returning to blue in warm water at 50 • C, staining with 85% alcohol for 5 min, staining with eosin for 5 min, soaking in tap water for 5 min, dehydrating, transparenting, and sealing.

Nissl staining of facial motoneuron
Frozen sections were defrosted, baked for 30 min at 37 • C, submerged in 1% toluidine blue solution heated to 50 • C, and stained for 20 min at 56 • C. Then they were rinsed with distilled water, immersed in 70% alcohol for 1 min, differentiated in 95% alcohol, transparented, sealed, and inspected under the light microscopy. hematoxylin re-staining. To calculate the optical density, area, and proportion of all captured pictures with a positive expression, three sample sections from each animal were randomly chosen. Slices were chosen 400 times to acquire photographs with the aid of the image-processing software Image-Pro Plus 6.0.

Masson staining of buccal muscle
Sections were routinely dewaxed to water, stained with modified Masson trichrome staining solution (0311A20; Hefei Bomei Biotechnology Co., Ltd.) according to the instructions, and then they were dehydrated, transparent, sealed, and microscopically examined. Utilizing the Image-Pro Plus 6.0 picture analysis system, the optical density and area of every collected image were measured, and the percentage of fibrous tissue expression area was computed.

Statistical analysis
The analysis was carried out using the SPSS 25.0 program. Mean ± SD was used to express all quantitative data. Analysis of variance was employed to compare the results of each group, and the least significant difference post hoc test was carried out. Nonparametric Wilcoxon signed-rank test was perpormed to compare the degree of axonal ultrastructure damage. Statistics were deemed significant at p < 0.05.

Electrophysiologic findings
Before the surgery, there were no discernible changes between each MC group in M-wave amplitude and latency (p > 0.05). All the rats were able to generate action potential from the orbicularis oris muscle after surgery and 28 days postoperatively. Except for the M10s and M1min groups, the M wave latency was longer postoperatively, and the wave amplitude was lower in the M5min, M8min, M10min, and M12min groups than that preoperatively. All of these variations were statistically significant (p < 0.05). At 28 days postoperatively, the M-wave latency was substantially longer.

Apoptosis and Nissl body changes in facial motoneurons
In Tunel staining, the negative expression (normal nuclei) was blue or light blue, whereas the apoptotic nuclei were light yellow or brownish yellow. There was no obvious facial motoneuron apoptosis following surgery in the MC groups. Statistical significance was not observed in

Buccal muscle atrophy analysis
Masson staining of normal muscular tissue revealed that collagenous connective tissue turned blue-purple and muscle fibers turned red. Following a muscle tissue injury, the collagenous tissue tends to become more stained as a result of scar repair or muscular fibrosis. All the MC groups experienced a decrease of buccal muscle fiber and an increase of collagen fiber 28 days postoperatively. With longer pressing time, the buccal muscle's atrophy grew more pronounced, and the area ratio of collagen fiber expression fraction gradually increased, which was notably distinct from the NC group (p < 0.01) (Figure 8).

DISCUSSION
The facial nerve injury model has been widely investigated for decades to assess mechanisms relating to nerve injury, survival, and regeneration as well as the assessment of neurotoxins and neurotrophic factors.
It is a simple and reproducible peripheral nerve injury model (Byram et al., 2020). Nerve compression models have been extensively used to study facial nerve injury and regenerative repair pathophysiological changes. There is no agreement on the methods, locations, or nerve compression duration in the facial nerve compression models, and the self-healing factors of the facial nerve had not been widely discussed.
The buccal branch of the facial nerve (Wang et al., 2020) or the facial nerve's main trunk  was traditionally used to determine where the facial nerves were compressed. Compared to other branches, the buccal branch of the facial nerve was thicker and more superficial. However, unlike humans, rats had only three branches that made up the facial terminal branch: the auricular nerve, the buccal nerve, and the mandibular rim nerve (Shan et al., 2005). Additionally, F I G U R E 5 (a) Comparison of the myelin sheath thickness of facial nerve in each group after surgery and 28 days postoperatively. Note: In all MC groups, the myelin thickness of the facial nerve was smaller than that of the normal control (NC) group after surgery and 28 days postoperatively. Furthermore, there was a decreasing trend between each group. Compared with the NC group post-surgery, # p < 0.05. Compared with the NC group 28 days postoperatively, *p < 0.05. (b) The standard axon ratio in each group after surgery and 28 days postoperatively. Note: In each MC group after surgery and 28 days postoperatively, the standard axon ratio was smaller than that in the NC group. There was also a decreasing trend observed between the MC groups. Compared with the NC group post-surgery, # p < 0.05. Compared with the NC group 28 days postoperatively, *p < 0.05. (c and d) The ultrastructural damage ratio of facial nerve in each group under electron microscope after surgery and 28 days postoperatively. Note: After surgery, the extent of ultrastructural damage to the facial nerves differed among the MC groups. Additionally, the damage was more severe with longer compression times. At 28 days postoperatively, the degree of axonal and myelin lesions in the NC group was mainly concentrated in "-," but the lesions in each MC group were all concentrated in "1+" and "2+," with axonal vacuolation and fissured myelin sheath. Compared with the NC group, # p < 0.05. some rats had variations in the buccal branch, which made it challenging to ensure consistency and reproducibility of modeling. However, the location of the facial nerve trunk was more constant and can avoid the influence of a branch innervating facial muscle movement Therefore, the main trunk of the facial nerve was more suitable for modeling.
The determination of the facial nerve trunk compression time is currently controversial, ranging from 10 s (Cao et al., 2021), 30 s (Hyeok , 1 min (Jang et al., 2014), twice for 30 s (10-s interval)  to 3 min 10 , 5 min (Li et al., 2018),10 min (Li et al., 2019). Accordingly, we set up 10 s, 1 min, 5 min, 8 min, 10 min, and 12 min groups and allowed them to heal naturally. Then we used animal behavioral observation, neurophysiological testing, and tissue staining techniques to assess the damaged facial nerve regeneration. higher capacity for regeneration of the rodent nervous system (Ali et al., 2020), and the motor deficits of the nose and mouth frequently showed transient changes, with the majority of cases recovering on their own within hours or days (Joko et al., 2020). This high frequency of active movements might be the primary cause of the rapid disappearance of appearance problems because of the high frequency of nose and lip use in rats' faces. A total score of more than 3 indicated the existence of peripheral facial palsy in this trial, which employed the

F I G U R E 8 (a and b)
Masson staining of buccinator in each group at 28 days postoperatively. Note: In the normal control (NC) group, Masson staining is manifested as red muscle staining and blue collagen fiber staining. All the MC groups experienced a decrease of buccal muscle fiber and an increase of collagen fiber at 28 days postoperatively. As the crushing time increases, the above changes become more pronounced. (c) The collagen fiber expression fraction in each group at 28 days postoperatively. Note: With longer pressing time, the buccal muscle showed more severe atrophy, and the area ratio of collagen fiber expression also increased gradually. Compared with the NC group, ##p < 0.01. In the image, the collagenous connective tissues are marked with red arrows, whereas the buccal muscle tissues are marked with green arrows.
lesion was still complete facial palsy. So the animal behavioral assessment could not be used as an index to determine facial nerve healing but could be used as a macroscopic secondary criterion to evaluate the success of facial palsy model.
There was no unified standard for grading the pathology of successful facial palsy models. Most scholars used the Sunderland III degree as the standard for successful facial nerve injury modeling (Liu et al., 2017 Rapid nerve impulse conduction relied on the myelination of nerves.
The thickness of myelin sheath was closely related to myelination. As a result, the facial nerve regeneration could be evaluated based on the typical number of axons and the thickness of the myelin sheath. Under the TEM, the standard axon ratio and myelin thickness decreased with the compression time lasted. Twenty-eight days postoperatively, the standard axon ratio and myelin thickness reduced again in the MC groups. Taking the facial nerve pathology into account, it could be said that the severity of the nerve damage increased with the pressing time lasted. As a result, the facial nerve damage was more severe 28 days postoperatively than that after surgery. Thus, the pathohistological staining performed immediately after surgery was not a reliable method for assessing the extent of facial nerve damage.
Nerve electrophysiology was one of the most sensitive and earliest indicators to determine nerve injury and repair (Bain et al., 1989), which was widely used to diagnose peripheral nerve injury. The commonly used methods were electromyography (ENoG), nerve excitability experiment, and facial nerve electrogram. In this experiment, the M-wave latency showed an increasing tendency both immediately after surgery and 28 days postoperatively between each MC group.
Additionally, the M-wave amplitude of the MC groups showed a tendency to decline. The M-wave latency and amplitude showed the same trend with the facial nerve trunk ultrastructural changes. The lesions were all aggravated 28 days postoperatively than that after surgery.
Therefore, ENoG could quantitatively estimate the degree of nerve degeneration. However, Wallerian degeneration did not occur immediately following facial nerve injury; rather, it developed gradually from the proximal end to the distal end over roughly a week. The inflammation of the facial nerve caused early damage that could not be accurately measured through ENoG, which did not meet the necessary standards for early diagnosis. This could lead to the possibility of incorrect or missed diagnoses (Kimura, 2006). It was assumed that axonal degeneration was the leading cause of facial nerve injury within a short period. However, only the M10min and M12min groups showed ENoG values beyond 90% 28 days postoperatively, and Li et al. (2003) noted that when ENoG levels were beyond 90%, the natural healing ability of facial nerve was subpar. It is suggested that the optimal length of time for compressing the facial nerve with a 12.5 cm micromosquito clamp is a minimum of 10 min.
Following the crush surgery, the entrapment segment experienced inhibited axoplasmic transport, which prevented the target muscle from receiving muscle neurotrophic factor produced by the neuronal cytosol. This ultimately caused muscle atrophy. Masson staining was used to quantify the collagen area/bundle area value. We discovered that the buccal muscle atrophy was more apparent with the pressing time lasted. The MC groups demonstrated substantial statistical differences compared to the NC group, except for the M10s and M1min groups. The validity of the crush model has been verified again. Accord-ing to the earlier studies on peripheral nerve injury, a facial nerve transection in mice causes a loss of motoneurons of 75%, while crushing the same nerve caused negligible cell apoptosis (Lee et al., 2017). Our findings were consistent with the previous investigation, which found no evidence of significant central facial motoneuron apoptosis following facial nerve trunk crush injury for varying lengths of time. Even cytosolic crinkling and deformation were visible in Nissl staining, demonstrating pathomorphological alterations in facial motoneurons. Cytoplasmic Nissl bodies gradually changed from blocky to granular, then to powdery or even disappeared. With the lengthening of the pressing time, the severity of the damage mentioned above gradually grew.
In summary, the study supported the possibility of using animal behavioral evaluation as an additional criterion to assess the early- This study also had several limitations worthing further efforts.
Referring to our study, to exclude the influence of facial nerve selfhealing factors on the next experiment, we needed to press the facial nerve trunk for at least 10 min. Compared with the previous literature, the time was relatively longer. We assumed that under the premise of achieving the same degree of nerve damage, whether the pressing tool or the pressing force could be changed to shorten the pressing time.
Moreover, what factors affected the self-healing of facial nerves and its molecular mechanisms needed to be further studied.

AUTHOR CONTRIBUTIONS
Jing Fei designed the study and drafted the manuscript. Xirui Guan and Hongdi Zheng performed animal surgery, histology, and morphometry experiments. Lin Gao and Ping Ni collected the data. Lin Gao and Kunling Duan analyzed and interpreted the data. Na Liao and Leiji Li conceived and designed the experiment and edited the review.