Tuina therapy promotes behavioral improvement and brain plasticity in rats with peripheral nerve injury and repair

Abstract Introduction Tuina is currently one of the popular complementary and alternative methods of rehabilitation therapy. Tuina can improve patients' pain and mobility function. However, the underlying physiological mechanism remains largely unknown, which might limit its further popularization in clinical practice. The aim of this study is to explore the short‐term and long‐term changes in brain functional activity following Tuina intervention for peripheral nerve injury repair. Methods A total of 16 rats were equally divided into the intervention group and the control group. Rats in the intervention group received Tuina therapy applying on the gastrocnemius muscle of the right side for 4 months following sciatic nerve transection and immediate repair, while the control group received nerve transection and repair only. The block‐design functional magnetic resonance imaging scan was applied in both groups at 1 and 4 months after the surgery. During the scan, both the injured and intact hindpaw was electrically stimulated according to a “boxcar” paradigm. Results When stimulating the intact hindpaw, the intervention group exhibited significantly lower activation in the somatosensory area, limbic/paralimbic areas, pain‐regulation areas, and basal ganglia compared to the control group, with only the prefrontal area showing higher activation. After 4 months of sciatic nerve injury, the control group exhibited decreased motor cortex activity compared to the activity observed at 1 month, and the intervention group demonstrated stronger bilateral motor cortex activity compared to the control group. Conclusion Tuina therapy on the gastrocnemius muscle of rats with sciatic nerve injury can effectively alleviate pain and maintain the motor function of the affected limb. In addition, Tuina therapy reduced the activation level of pain‐related brain regions and inhibited the decreased activity of the motor cortex caused by nerve injury, reflecting the impact of peripheral stimulation on brain plasticity.


INTRODUCTION
Tuina, as the traditional Chinese massage, serves as a complementary and alternative method in the rehabilitation therapy. It consists of many classic manipulation methods which are grouped into four distinct categories: pushing-rolling, squeezing/pressing, moving joints, and vibrating (Fang et al., 2018). Accumulating literature have reported its application in multiple diseases including back pain, cervical vertigo, insomnia, headaches, and hypertension (F. Huang et al., 2020;T. Li et al., 2021;Nie et al., 2019;X. Yang et al., 2014;. According to existing clinical trials, application of Tuina in several diseases has been proven to be inspiringly effective, especially in pain relieving (Happe et al., 2016;Tang et al., 2016;Wang, 2012). However, the underlying physiological mechanism still remains largely unaddressed.
Early clinical studies have shown that Tuina therapy can effectively alleviate chronic neck pain and exhibit long-term effects after 12 weeks (Pach et al., 2018). In addition, the combination of Tuina therapy and traditional Chinese exercise is also beneficial in reducing pain and improving disability (Zhou et al., 2022). Studies have shown that Tuina therapy intervention in rats with sciatic nerve pathological pain reduced the spinal dorsal horn C-fiber response, suggesting that the analgesic effect of Tuina therapy is related to the increase in pain threshold of C-fiber-induced field potential of ipsilateral and contralateral nerves (Jiang et al., 2016). The analgesic mechanism of Tuina therapy in peripheral pathological pain is mainly manifested by regulating the TLR4 pathway and miRNA to inhibit peripheral inflammation, regulating ion channels, inhibiting the activation of neuroglia cells, and regulating brain function changes. Tuina therapy has analgesic effects by acting on different levels of targets and is an effective treatment for peripheral neuropathic pain (Z. F. Liu et al., 2022).
A systematic review of the effects of Tuina therapy on poststroke sequelae improvement showed that, in addition to traditional therapies, therapeutic massage, especially Tuina therapy, has a significant effect on improving the motor function and reducing spasticity of stroke survivors (Cabanas-Valdes et al., 2021). The use of Tuina has shown positive effects in promoting the recovery of patients' motor function after injury (Kang et al., 2022).
As Tuina is a therapy performed locally, researchers used to focus on the physiological mechanism of its peripheral effects. For exam-ple, a previous study revealed that Tuina might decrease the activation level of peripheral nociceptive C-fiber (Jiang et al., 2016). It reasoned that Tuina may induce long nerve fiber signaling and thereby inhibit transmission of pain signal to the central nervous system by activating inhibitory neurons. Some other researchers suggested that neurotransmitters might also play an important role in the therapeutic effect of Tuina (Guo et al., 2016;S. R. Huang et al, 2003;Sousa, Moreira, et al., 2015). In a biomolecular study, the gene expression at the point of nerve injury and the myelin integrity was modulated by Tuina, which was related to the functional recovery following peripheral nerve injury repair (T. . However, the effect of Tuina on the central nervous system following peripheral nerve injury is rarely referred to.
Peripheral nerve injury is one of the major neuropathies that lead to lifelong disabilities (R. Li et al., 2014). Current therapeutic theory suggests that nerve regeneration and target muscle atrophy are two most important factors that influence functional outcomes (Q. Bao et al., 2021;He et al., 2022;Ruven et al., 2017;Sun et al., 2022;Zainul et al., 2022). Previous studies demonstrated that brain plasticity was also an important factor that involves in the recovery other than peripheral regeneration (Navarro et al., 2007). As more studies regarding to Tuina therapy in the treatment of peripheral nerve injury have been reported, investigations on brain plasticity have become essential for a better understanding of its effects and appropriate application in clinical practice (Z. Liu et al., 2021;.
Functional magnetic resonance imaging (fMRI) is a safe and noninvasive method widely used in neuroscience research (Fox, 2018;Rocca & Filippi, 2006). It is often used to explore the central nervous system mechanisms of both central nervous system diseases (Shan et al., 2023) and other peripheral diseases (Bhat et al., 2017;Xing et al., 2020Xing et al., , 2021. Previous fMRI studies have found that the anterior cingulate cortex (ACC), insular cortex, and primary and secondary somatosensory cortices (S1 and S2) are involved in the complex experience of pain (H. Li et al., 2022). In our previous study on electroacupuncture intervention for knee osteoarthritis, we observed that electroacupuncture reduced the centrality and nodal efficiency of the right ACC, which is a central component of the reward circuitry (J. P. Zhang et al., 2023).
This inhibition of activity in the reward circuitry contributed to pain suppression and reduced the transmission of pain signals from the source to the central nervous system. Currently, research on the brain mechanisms of pain has expanded from the study of a single brain region to the level of functional connectivity and functional networks (B. B. Bao et al., 2022). Researchers hope to use changes in the cortical function and plasticity as prognostic factors related to pain (Jenkins et al., 2023).
In the present study, the fMRI study was used to describe the alteration of brain activity induced by Tuina therapy as a rehabilitation treatment following peripheral nerve injury repair. The Tuina therapy was applied on the gastrocnemius muscle of the right side for 4 months following sciatic nerve transection and immediate repair, while the fMRI scan and the behavior test were applied at the first and fourth months after the surgery. The aim of this study is to explore the short-and long-term changes in brain functional activity induced by Tuina therapy in peripheral nerve stimulation tasks, to understand the effects of Tuina on pain-related circuits and motor-related cortex function regulation, and to further understand the central neural plasticity mechanisms of Tuina in the recovery of peripheral nerve injury.

Animals
A total of 16 male Sprague-Dawley rats weighted from 200 to 250 g and aged from 6 to 8 weeks were used in the study. These rats were randomly divided into the experimental and control groups (eight rats in each group). All of the rats were obtained from Shanghai Slack Laboratory Animal Company. They were housed under a condition of 12 h dark/12 h light cycle, with unrestricted food and water supply. They were kept for 1 week before any intervention started.

Peripheral nerve injury model
The peripheral nerve injury model was established by cutting and immediately repairing the sciatic nerve ( Figure 1). Specifically, after an intraperitoneal injection of 0.2% pentobarbital, the sciatic nerve on the right side was completely transected at midthigh level. Epineurium of the sciatic nerve was immediately anastomosed with 12-0 sutures (Ren et al., 2016). During the surgery, bipolar coagulation was utilized for hemostasis, in order to reduce postsurgical mortality. The incision area was treated with antibiotic powder and sealed with Michel clips.

Tuina therapy
A customized Tuina manipulation emulator (patent No. ZL201420 482075.3, State Intellectual Property Office) was used in the present study. The emulator's key structural elements consisted of a controller, a microengine, a pair of mobile arms, several pressure transducers, and a silicone tube ( Figure 2). The mobile arms with periodic movement were designed to simulate manipulation by human hands. The intervention parameters including pressure, frequency, and duration F I G U R E 1 Peripheral nerve injury model. The peripheral nerve injury model was established by cutting and immediately repairing the sciatic nerve of the rat.

F I G U R E 2
Illustration of the Tuina manipulation emulator. The photograph shows the Tuina manipulation emulator and its application on the rat. The emulator's key structural elements consisted of a controller, a microengine, a pair of mobile arms, several pressure transducers, and a silicone tube. The mobile arms with periodic movement were designed to simulate manipulation by human hands. The intervention parameters including pressure, frequency, and duration were set before the procedure started.
were set before the procedure started. During the procedure, the pressure could be regulated by the tension adjusting screw, the tension spring, and sensor be monitored on the screen, and the frequency be controlled by the adjusting knob. The emulator was used to simulate twirling and kneading manipulation on the gastrocnemius muscle of the injured hindlimb. The force was set at 0.45 N, and the frequency was 60 times/min, 10 min a day. Tuina manipulation started on the seventh postoperative day.
To guarantee the consistency of models, one specific technician performed all the surgery and Tuina intervention.

CatWalk gait analysis
The animal gait analysis system is a popular tool for the quantitative assessment of footsteps and natural gait in animal models of nerve disease, neurotic atrophy, nerve trauma, and pain symptom groups.
In the present study, the Catwalk XT animal gait analysis system (Noldus Information Technology, Wageningen, The Netherlands) was used to evaluate motor function in rats. The system included a 1.5-m long walking platform with a glass walkway at the bottom. The light-emitting diode emitted light scattering into the glass plate of the walkway. When the rat passed the walkway, the footprints were captured by a high-speed camera placed underneath and the speed and intensity were recorded.
Two weeks before the surgery, adaptive training was carried out every day in the morning and afternoon. During the training, 12 g/day of food was given daily to cause slight starvation. An eligible training requires at least three consecutive uninterrupted runs. The conditional parameters for a qualified run were as follows: passing time 1.00-8.00 s, speed variation rate less than 60%. The gait CatWalk analysis was carried out every month for 4 months after the surgery. The maximum contact mean intensity (MCMI), stride length (SL), and swing speed (SS) were recorded.

Statistical analysis of behavioral tests
The behavioral results were expressed as mean ± standard deviation (mean ± SD) and analyzed using SPSS 22.0 software package (SPSS Inc., Chicago, IL). A multifactor repeated measures analysis of variance was performed to compare the behavioral data of the control group and intervention group at the four time points for TWL, MCMI, SL, and SS. Correlation analysis was performed between the behavioral data and fMRI results separately. Values of p < .05 were considered to be statistically significant.

Functional MRI scan
All the fMRI scans were performed in a 7.0-T horizontal-bore Bruker

Stimulation tasks
A dummy scan lasted for 8 s in the time series was applied before any task was performed. The dummy scan was performed for magnetic equilibrium and the initial 8 s of data would not be included in the analysis. A "boxcar" model for the stimulation paradigm was used, which sequentially contained an epoch of ON and an epoch of OFF. Both "ON" and "OFF" epochs lasted for 30 s and these two epochs formed one cycle. In order to acquire significantly positive results, a total of eight cycles were arranged in one stimulation session. During each session, only one side hindpaw was stimulated. The electrical stimulation was applied with needles inserted beneath the skin of each hindpaw.
One needle electrode was located between the first and second digits, while the other one was located between the third and fourth digits. In order to avoid habituation of sensory stimulation, the stimulus was performed in a pseudorandom pattern. The fMRI scans were performed at the first and fourth months after surgery.

Data processing and statistical analysis
The preprocessing and statistical analysis were mainly performed by In the first-level analysis, a general linear model was determined according to the stimulation paradigm and the parameters were estimated through a Bayesian approach by SPM8. The contrast images containing information on β1-β2 were then acquired. In the second-level analysis, a two-sample t-test was performed between the experimental and the control groups at the same time point. In addition, paired t-tests were conducted to compare the brain activity between the fourth month and the first month within each group. After the t-value map was generated with contrast vectors, the threshold was set as p < .05 and the results were then interrogated with the false discovery rate (FDR) correction. The statistical map was visualized by projecting to a standard rodent template. The analysis and results were reported based on the Montreal Neurological Institute (MNI) atlas standard.

Behavioral analysis
There was a significant main effect of the group on the TWL, MCMI, were performed between two groups per time point. From the first to fourth postoperative months, the TWL of the two groups both decreased gradually. There was no significant difference between the two groups at first month after surgery (p > .05). From the second month, the TWL of the Tuina group was significantly lower than the control group (p < .05) ( Figure 3a and Table 1). MCMI, SL, and SS of the two groups increased gradually from the first to fourth month. For the MCMI, there was no significant difference between the two groups from the first to third month (p > .05). At the fourth month, the MCMI of the Tuina group was significantly higher than the control group (p < .05) ( Figure 3b and Table 1). For the SL, there was no significant difference between the two groups at the first and the second month (p > .05). The SL of the Tuina group was significantly higher than the control group at the third and fourth month (p < .05) ( Figure 2 and Table 1; Figure 3c and Table 1). For the SS, there was also no significant difference between the two groups at the first and the second month (p > .05). At the third and fourth months, the SS of the Tuina group was significantly higher than the control group (p < .05) ( Figure 3d and Table 1).

fMRI results
The two-sample T-test was performed between the two groups at each time point. In addition, paired t-tests were conducted to compare the brain activity between the fourth month and the first month within each group. And the threshold of statistical analysis was set at p < .05 (FDR correction).

Right (injured) hindpaw stimulation task (intervention group-control group)
At the first postoperative month, multiple areas including limbic/paralimbic areas, pain-regulating areas, and basal ganglia presented lower activation in the intervention group, compared with the control group. Specifically, subregions of the hippocampus, retrosplenial cortex, medial prefrontal cortex, orbitofrontal cortex, hypothalamus, cingulated cortex, piriform cortex, insular, and subregions of thalamus presented significantly lower activation (Figure 4 and Table 2).
At the fourth postoperative month, the activation pattern was similar to that of the first postoperative month. Areas including the limbic/paralimbic areas, pain-regulating areas, and basal ganglia presented lower activation in the intervention group. Specific regions of the somatosensory cortex, subregions of the hippocampus, amygdala, AcbC, AcbSh, cingulate cortex, subregions of thalamus, pallidum, and putamen were involved. Additionally, activation in the prefrontal area was higher in the intervention group compared with the control group ( Figure 4 and Table 3).

3.2.2
Left (intact)hindpaw stimulation task (intervention group-control group) At the first postoperative month, activation in multiple areas was lower in the intervention group, including the limbic/paralimbic areas, painregulating areas, and basal ganglia, compared with the control group.
At the fourth postoperative month, activation in brain regions in the intervention group was also lower, involving the limbic/paralimbic areas, pain-regulating areas, and basal ganglia, compared with the control group. Specifically, the retrosplenial cortex, AcbSh, subregions of the hypothalamus, insular cortex, globus pallidus, cingulate cortex, subregions of thalamus, pallidum, and putamen were involved ( Figure 5 and Table 5).

F I G U R E 3
Comparison of behavioral tests between the intervention group and control group. Thermal withdrawal latency (TWL), maximum contact mean intensity (MCMI), stride length(SL), and swing speed (SS) of rats of the intervention group and control group at four time points. *p < .05 (at individual time points between the intervention group and control group).

Group
Time point In the control group, compared to the first month, the following brain regions showed increased activity in the fourth month: right posterodorsal hippocampus, the right interstitial nucleus of the posterior limb of the anterior commissure (IPAC); The following brain regions showed decreased activity: left retrosplenial cortex, right entorhinal cortex, left superior colliculus, left motor cortex, right mesencephalic region (Table 6).
In the intervention group, the following brain region showed decreased activity in the fourth month compared to the first month: right superior frontal (Table 6).

F I G U R E 4
Comparison of cortical activation between the experimental and control groups during the right (injured) hindpaw stimulation task. Each column in the figure displays the difference in the limbic/paralimbic system, pain-related brain regions, and somatosensory cortex between experimental and control groups. Each row in the figure displays the difference in activation between experimental and control groups at two observation points. The warm tone represents higher activation in the intervention group than that in the control group, while the cold tone represents lower activation.

F I G U R E 5 Comparison of cortical activation between the experimental and control groups during the left (intact) hindpaw stimulation task.
Each column in the figure displays different activated brain areas between experimental and control groups. Each row in the figure displays the difference in activation between two observation points in the same group. The warm tone represents higher activation in the intervention group than that in the control group, while the cold tone represents lower activation.

3.2.4
Motor-related brain regions during right (injured) hindpaw stimulation task (intervention group-control group) At the first postoperative month, the control group showed increased activity in the primary motor cortex and somatosensory cortex of the right hemisphere compared to the intervention group (Table 7), p < .001.
At the fourth postoperative month, the intervention group showed increased activity in both primary motor cortices and the somatosen-sory cortex of the right hemisphere compared to the control group (Table 7), p < .001.

Correlation
At the first postoperative month, the Pearson correlation coefficients between the TWL of the injured hindpaw and the functional activity of the left insular cortex was -.03210 (p > .05), indicating no statistical significance; the Pearson correlation coefficients between TA B L E 2 Difference of activation between the experimental and control groups during the right (injured) hindpaw stimulation task at the first postoperative month.

DISCUSSION
The sciatic nerve is a mixed nerve that composes of sensory and motor fibers. After injury, both motor and sensory functions of the hindlimb are partially or completely lost except for the saphenous innervations area (Shin & Howard, 2012).
Several measures could be used to evaluate sensorimotor function recovery following sciatic nerve injury repair. The TWL is one of the classic methods to evaluate the recovery of sensory function after TA B L E 3 Difference of activation between the experimental and control groups during the right (injured) hindpaw stimulation task at the fourth postoperative month.

Cluster centroid (MNI)
Brain Gait analysis is an automated behavior testing method, which is easy to control the speed of animal movement and allows the evaluation of animal locomotion function in free status (Bozkurt et al., 2008;Xu et al., 2019) and has been widely used in animal models of peripheral nervous diseases. It is also an important method for evaluating static and dynamic functions after nerve injury (Deumens et al., 2007;Gabriel et al., 2007;Kappos et al., 2017;Wu, Lu, Hua, Ma, Shan, et al., 2018;. In the present study, gait analysis parameters, MCMI, SL, and SS, were used to evaluate the motor function after sciatic nerve injury repair. The motor function recovered to various degrees in both groups. The MCMI, SL, and SS were significantly bet-ter in the Tuina group. Therefore, Tuina is also related to better motor recovery after nerve repair. The effect was more obvious at the third and fourth months after the surgery. Previous studies have demonstrated that peripheral nerve injuries would result in changes not only at the local site of injury but also in long-lasting cortical plasticity. These changes are caused by permanent loss of sensation and the misdirection of the newly formed axons after the nerve repair, the newly formed sensory and motor axons reinnervated do not match their original organs (Darian-Smith, 1994;Lundborg, 2007;Rosen & Lundborg, 2004;Yamahachi et al., 2009).
Reports have suggested that Tuina would improve functional outcomes following peripheral nerve injury repair (Z. Liu et al., 2021;. However, it is not well understood whether the long-lasting cortical plasticity is involved in the improvement of function after Tuina therapy. To our knowledge, the present study is the first longitudinal study on cortical plasticity in a rat model of peripheral nerve injury treated by Tuina therapy.

TA B L E 6
Difference of activation between the fourth month and the first month during the right (injured) hindpaw stimulation task in the brain regions.

Cluster centroid (MNI)
Brain Our study found that 1 month after modeling, compared to the intervention group, the primary motor cortex activity in the right In the cortical activation regions during right (injured) hindpaw stimulation tasks, several brain areas were less activated in the intervention group than that in the control group. As the somatosensory cortex mainly receives a projection of sensory fibers from the peripheral, and the enhancement of afferent signal is a key factor that impacts the excitability of the somatosensory cortex. A study on monkeys has proved this view. They found that high-intensity stimulation to the monkey's fingers can make the somatosensory cortex more activated (Qi et al., 2016). It was also reported that Tuina therapy may decrease the activation level of peripheral nociceptive C-fiber, which might reduce the sensory afferentation signal (Jiang et al., 2016). Therefore, the somatosensory cortex of the intervention group showed lower activation during the peripheral stimulating task.
Neuropathies following transection usually induce paresthesia or pain before reinnervation of targets (Alexander et al., 2019). Tuina therapy would alleviate paresthesia or pain after the peripheral neural pathway was reconnected. This reduction of sensory afferents following Tuina therapy consequently led to a reduction in the extent and strength of activation in the sensory cortex. Therefore, the somatosensory cortex in the intervention group was less activated.
In addition, the pain-related areas, basal ganglia, and limbic/paralimbic areas also presented lower activation in the Tuina group. Previous studies have reported regions of modules in the "pain network," including the primary somatosensory cortex, medial frontal cortical structures (such as the ACC, middle cingulate cortex (MCC), and supplementary motor area (SMA)), and the parasylvian cortical structures (such as the insula and opercular) (C. C. Liu et al., 2011). The insula is most strongly associated with other structures in the brain network, the anterior insula is connected with ACC and the frontal cortices, while the posterior insula is associated with the primary somatosensory cortex, motor cortex, and temporal cortex (Abram et al., 2015). The ACC is a key structure that processes information relating to pain-induced unpleasantness. The thalamus is an important sensory relay station that transmits sensory signals to the cerebral cortex and is highly associated with pain (Zitnik et al., 2014). The basal ganglia (such as caudate, putamen, and pallidum) is a major site for adaptive plasticity in the brain, affecting the normal state in a broad range of behaviors as well as neurological and psychiatric conditions including pain (Graybiel, 2004). The limbic system includes multiple brain regions involving memory, emotion, and sensation. The hippocampus and amygdala transmit cortical signals to the hypothalamus and brainstem to regulate the arousal and motivational state of the whole neuraxis (Cardinal, 2002). It is recognized that the hippocampus, amygdala, and parahippocampus play an important role in regulating motivation, emotions, memory, and pain (Kirby et al., 2013;Y. Yang et al., 2016;Zubieta et al., 2001). Moreover, the inactivation of the hippocampus and amygdala was found to be related to the regulation of pain threshold (W. T. Zhang et al., 2003). In the present study, the activation pattern of pain-related areas, basal ganglia, and limbic/paralimbic areas was quite synchronized. Therefore, these areas also played an important role in Tuina therapy-induced brain plasticity.
The prefrontal cortex has been suggested to participate in controlling functional interactions between areas of the pain-related brain regions (Tracey & Mantyh, 2007). Along with the basal ganglia, the prefrontal cortex is also closely associated with attention and executive during tasks (Yuan & Raz, 2014). Considering the integration of sensory and motor functions, the sensory input is essential for motor initialization and adjustment in walking. After 4 months of Tuina therapy, the higher activation of the prefrontal cortex implied restoration of the motor regulating function. Rather than reactivating some isolated areas alone, it demonstrated a gradual recovery of more advanced brain function.
According to the brain activation regions during the left (intact) hindpaw stimulation task, lower activation of pain-related areas, basal ganglia, and limbic/paralimbic was observed in the intervention group, which was similar to the right (injured)hindpaw stimulation task. We assumed that lower activation in pain-related areas and advanced modulation network was related to relieving effects of Tuina therapy on decreasing the excitability of sensory, pain, and emotion-related areas. This suggested that Tuina therapy potentially leads to more extended changes in the brain other than the local effect.

CONCLUSION
Tuina therapy on the gastrocnemius muscle of rats with sciatic nerve injury can effectively alleviate pain and maintain the motor function of the affected limb, thus avoiding the decline in activity caused by nerve injury. In addition, Tuina therapy changes the activation level of painrelated brain regions and the decreased activity of the motor cortex caused by nerve injury, providing a central nervous system mechanism for Tuina therapy in peripheral nerve injury and reflecting the impact of peripheral stimulation on brain plasticity.

Limitation
Further investigations should be conducted before extrapolating the present conclusions due to the differences between humans and rodents. More investment is needed to study the central nervous system mechanisms of Tuina therapy.

AUTHOR CONTRIBUTIONS
S

CONFLICT OF INTEREST STATEMENT
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.