A total of 30 female Sprague-Dawley rats (age, 2 months; weight, 180–220 g) were purchased from The Experimental Animal Center of Nantong University (Nantong, China) and subjected to sciatic nerve crush injury as previously described (11). Rats were anaesthetized by injecting 2 ml/kg Equithesin (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany), as previously described (12), corresponding to 17 mg/kg sodium pentobarbital, 42 mg/kg magnesium sulfate and 85 mg/kg trichloroacetaldehyde monohydrate) (11,16). Anesthesia was assessed by loss of pedal reflexes. Rat left sciatic nerves were crushed with forceps for 10 sec, three times. To minimize the discomfort and possible painful mechanical stimulation, the rats were housed in large cages with sawdust bedding post-surgery and maintained in a controlled environment under constant temperature of 25°C and relative humidity of 40–70%, with a 12 h light/dark cycle and free access to food and water. Rats were sacrificed 24 h, 4 days, or 1, 2, 3, 4 or 8 weeks following surgery. Gastrocnemius muscles from the left (injured) side and the right (uninjured) side were collected. Control rats were anesthetized without sciatic nerve crushing (sham surgery). All animal procedures were ethically approved by The Administration Committee of Experimental Animals of The Jiangsu Province (Jiangsu, China).

CMAP recording was performed at 1, 2, 4 and 8 weeks following surgery. Following anesthesia, the left sciatic nerves of the rats were analyzed. The active and reference electrodes were placed into the medial belly and Achilles tendon of gastrocnemius, respectively. The stimulating electrode was placed sequentially in the proximal and distal sciatic nerves. A 2-mA electric stimulus was applied to the stimulating electrode to stimulate CMAP responses. CMAP readings at the proximal and distal sides were recorded using a portable digital MYTO electromyographic machine (Esaote SpA, Genoa, Italy) with Galileo NT System Software (EB Neuro SpA, Florence, Italy).

Gastrocnemius muscles on the left (injured) side and the right (uninjured) side were collected and weighed in healthy control rats and nerve-injured rats at 24 h, 4 days, or 1, 2, 3, 4 or 8 weeks following nerve crush injury. The wet weight of the muscle on the left side was divided by the wet weight of the muscle on the right side to calculate the ratio of injured to uninjured muscle wet weight.

Left-side gastrocnemius muscles were collected and fixed in 4% paraformaldehyde for 12 h at 4°C, embedded in Tissue-Tek^® optimal cutting temperature compound (Sakura Finetek USA, Inc., Torrance, CA, USA), and cut into 25-µm thick longitudinal sections using a cryostat microtome at −15°C (Leica Microsystems GmbH, Wetzlar, Germany). Sections were washed with PBS, blocked with normal goat serum (Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA) for 1 h at 37°C and incubated with anti-neurofilament H (NEFH) antibody (1:200; cat. no. SAB4200747; Sigma-Aldrich; Merck KGaA) overnight at 4°C. Tissue sections were washed, incubated with secondary antibodies conjugated with Alexa Fluor 488 (1:400; cat. no. ab150113; Abcam, Cambridge, MA, USA) and tetramethylrhodamine-labeled α-Bungarotoxin (1:300; cat. no. 203980; Sigma-Aldrich; Merck KGaA) for 2 h at room temperature and observed under a fluorescence microscope at ×100 magnification (Axio Imager M2; Carl Zeiss AG, Oberkochen, Germany).

In addition, after fixation in 4% paraformaldehyde for 12 h at 4°C, gastrocnemius muscles were embedded in paraffin and cut into transverse sections (6-µm thick). Tissue sections were subjected to Masson's trichrome staining at room temperature for about 1 h, to stain muscle fibers in red, collagen fibers in blue and cell nuclei in black, and observed using light microscopy at ×200 magnification (Axio Imager M2; Carl Zeiss AG, Oberkochen, Germany). Fiber cross-sectional area (CSA) was calculated using Leica QWin version 3 Image Analysis System Software (Leica Microsystems GmbH).

Total RNA was isolated from left-side gastrocnemius muscles using TRIzol^® Reagent (Thermo Fisher Scientific, Inc.), purified with RNeasy spin columns (Qiagen, Inc., Valencia, CA, USA) to remove DNA and reverse transcribed to cDNA with the PrimeScript reagent kit, 37°C for 15 min and 85°C for 5 sec, according to the manufacturer's protocol (Takara Biotechnology Co., Ltd., Dalian, China). RT-qPCR was performed using the QuantiNova SYBR Green PCR kit (Qiagen, Inc.) and the StepOne™ real-time PCR System (Applied Biosystems; Thermo Fisher Scientific, Inc.). An initial denaturation of 10 min at 95°C was followed by 40 cycles of 95°C for 15 sec and 60°C for 30 sec. The Cq values for the candidate reference genes 18S, ACTB, ANKRD27, CYPA, GAPDH, HPRT1, MRPL10, PGK1, RICTOR, TBP, UBC, UBXN11 and YWHAZ were determined for subsequent geNorm and NormFinder analyses. The Cq values for the target genes F-box protein 32 (FBXO32) and tripartite motif containing 63 (TRIM63) were additionally assessed and the expression levels of FBXO32 and TRIM63 were calculated using the 2^−ΔΔCq method (17). The quality of the RT-qPCR analysis was confirmed by a single melt-curve peak, representing a single amplification product. Primer sequences (listed in Table I) were obtained from previous publications (18–23) or designed using Primer Express^® software (v3.0.1; Thermo Fisher Scientific, Inc.).

The amplification efficiency analysis was performed by producing serial dilutions of the DNA sample. Subsequently, the Cq values were plotted on a logarithmic scale along with corresponding concentrations. Using a linear regression curve through, the slope of the trend line was calculated. Finally, efficiency was defined using the equation: E=−1+10^(−1/slope). Typically, desired amplification efficiencies range from 90 to 110% (https://biosistemika.com/blog/qpcr-efficiency-over-100/).

Cq values for candidate reference genes were subjected to geNorm analysis (v3.4; http://genorm.cmgg.be/), a Visual Basic for plug-in for Microsoft Excel 2007 (Microsoft Corporation, Redmond, WA, USA), that was used to calculate the stability of the expression levels of the reference genes. Gene expression stability (M value) was determined on the basis of non-normalized expression levels of internal reference genes according to a previous study (6). The normalization factor (pairwise variation value, Vn/n+1) was calculated based on the geometric mean of multiple reference genes (6). The stability values of candidate reference genes were also measured using the NormFinder analysis (v19, http://moma.dk/normfinder-software), a model-based variance estimation approach of Microsoft Excel (Microsoft Corporation), according to a previous study (3).

Data are presented as the mean ± standard deviation. Each independent experiment was repeated three times. Statistical analysis was performed using GraphPad Prism 6.0 (GraphPad Software, Inc., La Jolla, CA, USA). The linear association between two variables was assessed by calculating the correlation coefficient R², ranging from 0 to 1. Differences between rats in the injury and normal control groups were tested using one-way analysis of variance followed by Dunnett's multiple comparisons test. P<0.05 was considered to indicate a statistically significant difference.

To investigate muscle denervation and reinnervation, morphological and physiological parameters of gastrocnemius muscles were examined in normal control rats and in rats subjected to sciatic nerve injury. Staining of the axon marker NEFH and the motor end-plate marker α-Bungarotoxin (Fig. 1A) were detected, and the structure of muscle fibers (Fig. 1B) appeared to be intact. Furthermore, normal CMAP amplitude (Fig. 1C) was observed in the gastrocnemius muscles of uninjured rats. At 1 week following sciatic nerve injury, degenerated axon fragments were not present, and axons were not detectable between muscle fibers (Fig. 1D), muscle fibers were atrophied (Fig. 1E) and CMAP was not be detected (Fig. 1F). At 2 weeks following sciatic nerve injury, a small number of regenerated axons were observed, and a subset of axons initiated reinnervation (Fig. 1G). Masson's trichrome staining results suggested the presence of muscle atrophy and a large number of hyperplastic collagen fibers were observed (Fig. 1H). Low CMAP amplitude was recorded (Fig. 1I) at 2 weeks following injury. Multiple regenerated axons and reinnervated fibers were observed at 4 weeks following nerve injury (Fig. 1J). Additionally, muscle atrophy (Fig. 1K) and CMAP (Fig. 1L) were partially recovered. At 8 weeks following sciatic nerve injury, the number of regenerated axons and reinnervated fibers increased (Fig. 1M), the morphology of muscles (Fig. 1N) and the CMAP profile (Fig. 1O) were similar to the control rats (Fig. 1A-C).

Additionally, the ratio of injured to uninjured muscle wet weight decreased significantly following nerve injury at week 1 and 2, and improved at 4 and 8 weeks (Fig. 1P). CSA exhibited a similar trend (Fig. 1Q), suggesting that the gastrocnemius muscles underwent muscle atrophy following nerve injury. Quantification of the CMAP results suggested that the decreased CMAP amplitude detected at 2 weeks was increased at 4 weeks following nerve injury and was restored to normal levels at 8 weeks following injury (Fig. 1R). Collectively, the present data suggested that the physiological functions of muscles were gradually restored during the process of reinnervation of target muscles.

Prior to the determination of the Cq values of the reference genes, the efficiencies of the primers selected for RT-qPCR were examined. Following a serial dilution, the amplifications of the 13 candidate housekeeping genes exhibited linear standard curves with slope values of ~-3.322 and a linear correlation coefficient (R²) of ~1 (Table II). The amplification efficiencies of the primer pairs of all housekeeping genes were calculated based on the slope of the linear standard curves. The amplification efficiencies of all primer pairs were between 90 and 102% (Table II).

Subsequently, RT-qPCR experiments were performed to measure the raw Cq values of the reference genes. 18S exhibited the lowest Cq value, indicating that the abundance of 18S ribosomal RNA gene was the highest in rat gastrocnemius muscles. The Cq values of GAPDH, a gene encoding an enzyme involved in glycolysis, were decreased compared with other genes. The Cq values of UBXN11 were increased compared with other housekeeping genes, suggesting that its endogenous expression level was the lowest. The Cq values of the other housekeeping genes ranged between 20 and 30 (Fig. 2). In addition, the Cq values of the housekeeping genes in gastrocnemius muscles were assessed at various time points following sciatic nerve injury. The Cq values of ANKRD27, HPRT1, MRPL10, RICTOR and TBP exhibited a stable level in normal control rats and nerve-injured rats at various time points. By contrast, the Cq values of 18S, ACTB, CYPA, GAPDH, PGK1, UBC, UBXN11 and YWHAZ exhibited variability among time points (Fig. 2).

The stabilities of the gene expression levels of the candidate reference genes were examined using geNorm analysis. Average expression stability was calculated and M values were assigned to each reference gene. The M value calculated for each gene was inversely proportional to the stability of the expression level of the corresponding gene. Our previous study identified that all reference genes examined exhibited M<1.5, corresponding to the default limit (6). The reference genes ANKRD27 and HPRT1 exhibited decreased M values compared with the other reference genes, suggesting that the expression levels of these two reference genes were the most stable (Fig. 3A). PGK1 presented the highest M value, corresponding to the lowest stability (Fig. 3A). The normalization factor (pairwise variation value, V_n/n+1) was calculated to assess the optimal numbers of reference genes to use for gene expression normalization. The pairwise variation value V_2/3 calculated was 0.103, below the default limit of V_n/n+1=0.15 (24), suggesting that the two reference genes ANKRD27 and HPRT1 were sufficient to normalize gene expression levels in rat gastrocnemius muscles (Fig. 3B).

The stabilities of the expression levels of the reference genes were further analyzed and ranked using NormFinder. Low stability values were associated with stable expression levels. NormFinder analysis suggested that HPRT1 was the reference gene with the most stable expression level, followed by ANKRD27. The expression level of PGK1 exhibited the lowest stability among the genes examined (Fig. 4). The NormFinder results were consistent with the geNorm results. Collectively, the present results suggested that, in rat gastrocnemius muscles following sciatic nerve injury, the expression levels of HPRT1 and PGK1 exhibited the highest and the lowest stability, respectively.

The influence of distinct reference genes was examined by determining the expression levels of FBXO32 and TRIM63, two genes encoding skeletal muscle atrophy markers (23,25,26). The primer pairs of FBXO32 and TRIM63 exhibited linear standard curves with high efficiency values (94 and 95%, respectively; Table II). HPRT1 and PGK1 exhibited the lowest and the highest stability, respectively. Therefore, these two genes were used as internal controls to perform RT-qPCR experiments (Fig. 5).

Using HPRT1 as the internal control, the expression levels of FBXO32 were observed to be significantly upregulated in rat gastrocnemius muscles following sciatic nerve injury, exhibiting a maximum at 1 day following injury (Fig. 5A). Similarly, using PGK1 as internal control, the expression levels of FBXO32 were observed to be upregulated following nerve injury. However, the highest Cq value of FBXO32 was detected at 1 week following injury. Notably, the fold changes of FBXO32 were identified to vary depending on the reference gene selected (Fig. 5B).

Results from geNorm analysis suggested that two reference genes were sufficient to normalize gene expression levels. In addition to HPRT1, ANKRD27 and TBP were identified to exhibit stable expression levels. Therefore, the geometric mean of ANKRD27 and TBP was used as an internal control (Fig. 5C and F). Using ANKRD27+TBP or HPRT1 as the internal controls did not affect the detected expression level of FBXO32, which exhibited similar fold changes at all time points following peripheral nerve injury (Fig. 5C).

Additionally, TRIM63 exhibited similar expression patterns at various time points following sciatic nerve injury, regardless of the reference gene selected (Fig. 5D-F). However, similarly to FBXO32, the calculated relative fold changes of TRIM63 varied depending on the reference gene selected. The expression level of TRIM63 at 4 days following nerve injury increased by ~5 fold when compared with normal control, using HPRT1 or ANKRD27+TBP as the internal control (Fig. 5D and F). By contrast, using PGK1 as the internal control, the detected relative expression level of TRIM63 at 4 days following nerve injury increased by ~30 folds when compared with the normal control (Fig. 5E).