Skeletal muscle is the most abundant tissue of mammals, comprising 40–50 % of the total body mass. It is a heterogeneous tissue comprised of a variety of functionally diverse muscle fibre types with contractile and metabolic properties(Reference Choi and Kim1). Skeletal muscle fibres are generally classified into two major groups: type I muscle fibres expressing myosin heavy chain I (MyHC I) and type II muscle fibres expressing MyHC IIa, MyHC IIx and MyHC IIb(Reference Zhang, Liu and Fu2). It is well known that type I muscle fibres have more mitochondria and type IIb muscle fibres contain fewer mitochondria(Reference Schiaffino and Reggiani3). Mitochondrial biogenesis and function may be a new mediator of skeletal muscle fibre type(Reference Venhoff, Lebrecht and Pfeifer4). Improvement of mitochondrial biogenesis and function has been reported to promote the formation of type I muscle fibres(Reference Lin, Wu and Tarr5,Reference Zhang, Zhou and Wu6) .
l-Arginine (Arg) is a nutritionally essential amino acid for animals(Reference Jobgen, Fried and Fu7). Previous studies have suggested that Arg plays multiple physiological functions in animals, such as antioxidant activity, reproductive performance and fat deposition(Reference Mateo, Wu and Bazer8–Reference Zheng, Yu and He11). A few recent studies have shown that Arg promotes more type I muscle fibres formation in mice and MyHC I expression in porcine skeletal muscle satellite cells(Reference Chen, Guo and Jia12–Reference Chen, Jia and Zhao14). However, no information has been reported about the effect of Arg on type I muscle fibre type formation in weaning piglets. It has been reported that Arg modulates mitochondrial function(Reference Mabalirajan, Ahmad and Leishangthem15,Reference Kurhalyuk and Tkachenko16) . But it remains unclear whether Arg affects type I muscle fibres formation by improving mitochondrial biogenesis and function.
It is well known that the increase of the proportion of type I muscle fibres is beneficial to the improvement of meat quality in animal husbandry(Reference Chang, da Costa and Blackley17). Here, the aim of the present study was to investigate the role of Arg in mitochondrial biogenesis and function-related marker expression, mitochondrial DNA (mtDNA) content and muscle fibre type composition in weaning piglets. For further studies, mitochondrial complex I inhibitor rotenone (Rot) was used to investigate the mechanism of Arg affecting type I muscle fibres formation.
Materials and methods
Ethics statement
All animal procedures were performed according to protocols approved by the Animal Care Advisory Committee of Sichuan Agricultural University.
Animal experimental design
Sixty Duroc × Landrace × Yorkshire piglets with an average body weight of 6·55 (sd 0·36) kg were randomly assigned to four treatments. Each treatment consisted of five replicate pens of three piglets per pen. The piglets were provided a basal diet or a basal diet supplemented with 0·5, 1·0 or 1·5 % of Arg during a 28-d experimental period, while adjusting alanine to maintain the diets isonitrogenous. The basal diet was formulated according to the nutrient requirements for 7–10 kg and 11–25 kg pigs (Nutrient Requirements of Swine, 11th revised edition, 2012). The composition and nutrient levels of the diets are listed in Table 1. l-Arg (catalogue no. A8094) was obtained from Sigma. Water and feedstuff were given ad libitum during the experimental period. Individual body weight was measured at the beginning and end of the experimental period. Feed intake was recorded every day.
Table 1. Composition and nutrient levels of the diets
* The vitamin premix supplied the following per kg diet: vitamin A, 9000 IU; vitamin D3, 3000 IU; vitamin E, 24 IU; vitamin K, 3 mg; vitamin B1, 3 mg; vitamin B2, 7·5 mg; vitamin B3, 30 mg; vitamin B6, 3·6 mg; vitamin B12, 0·036 mg; biotin, 0·15 mg and folic acid, 1·5 mg.
† The trace mineral premix supplied the following per kg diet: Fe, 100 mg; Cu, 6 mg; Mn, 4 mg; Zn, 100 mg; Se, 0·3 mg and iodine, 0·14 mg.
‡ Nutrient levels were calculated values.
§ To convert Mcal to MJ, multiply by 4·184.
Sample collection
At the end of the experiment, one piglet from each pen was slaughtered in a humane manner. Then, the longissimus dorsi muscle was collected, immediately frozen in liquid N2 and stored at −80°C until further use.
Succinic dehydrogenase enzyme assay
The activity of succinic dehydrogenase (SDH) in longissimus dorsi muscle was determined using a commercial kit from Nanjing Jiancheng Bioengineering Institute. SDH activity was normalised to the total protein concentration.
Assay of nitric oxide synthase activity and nitric oxide content
The nitric oxide synthase (NOS) activity and nitric oxide (NO) content in longissimus dorsi muscle were detected by commercial biochemical kits (Nanjing Jiancheng Bioengineering Institute) following the manufacturer’s instructions. The total protein concentration was measured using the Coomassie brilliant blue method (Nanjing Jiancheng Bioengineering Institute).
Analysis of mitochondrial DNA content
The relative mtDNA content was determined by real-timequantitative PCR as described above. The genomic DNA was isolated from the longissimus dorsi muscle of weaning piglet using a DNAiso reagent according to the manufacturer’s instructions (TaKaRa). The primers for amplification of mtDNA (accession No. AF276923) were as follows: forward primer 5′-ACACCCTATAACGCCTTGCC-3′ and reverse primer 5′-AGGTGCCTGCTTTCGTAGC-3′. The primers for amplification of β-actin (accession No. DQ452569) were as follows: forward primer 5′-CAAAGCCAACCGTGAGAAGATG-3′ and reverse primer 5′-TGGCAAAGAGAGGCAAGAGAG-3′.
ATPase analysis
Longissimus dorsi muscle was collected and frozen in an optimal cutting temperature compound (Cat No. 4583, Sakura) before sectioning. The transverse serial sections were incubated with calcium chloride solution (pH 10·4) for 5 min and calcium chloride (pH 9·4) solution for 30 min. Then, the sections were stained with calcium chloride, cobalt nitrate and ammonium sulphide solutions. Finally, the sections were dehydrated with absolute ethyl alcohol. Sections were photographed with a Nikon Eclipse E100 microscope equipped with a Nikon DS-U3 digital camera (Nikon Incorporation). The ratio of type I:type II fibres was calculated by using Motic Images Advanced 3.2 analysing systematical software. Light grey-stained fibres are slow type I, and dark fibres are fast type II.
Real-time quantitative PCR
The total RNA from longissimus dorsi muscle was extracted using the RNAiso Plus reagent (TaKaRa) according to the manufacturer’s instructions. The RNA concentration and purity were determined using a Beckman DU-800 spectrophotometer (Beckman Coulter). Subsequently, reverse transcription was performed using a PrimeScript RT reagent Kit with gDNA Eraser (TaKaRa) according to the manufacturer’s instructions. Real-time quantitative PCR was performed using SYBR select Master Mix (Applied Biosystems) by a 7900HT Real-Time PCR Detection System (384-cell standard block) (Applied Biosystems). Sequences of the primers used for real-time quantitative PCR are listed in Table 2. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA was used as an internal control, and the relative gene expression was calculated using the 2–ΔΔCt analysis method(Reference Livak and Schmittgen18).
Table 2. List of genes, primer sequences, GenBank accession numbers and product sizes
MyHC I, myosin heavy chain I; Tnni1, troponin I1; Tnnc1, troponin C1; Tnnt1, troponin T1; PGC-1α, PPAR-γ coactivator-1α; NRF1, nuclear respiratory factor 1; TFB1M, mitochondria transcription factor B1; Cytc, cytochrome c; ATP5G, ATP synthase subunit C1; Sirt1, sirtuin 1; GAPDH, glyceraldehyde 3-phosphate dehydrogenase.
Western blot analysis
Western blot analysis was performed as described before(Reference Chen, Guo and Jia12). In brief, protein was isolated from longissimus dorsi muscle using radio immunoprecipitation assay lysis buffer (Pierce). Protein concentrations were determined using the bicinchoninic acid protein assay kit (Pierce). Equal amounts of protein were separated by 8 % SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Millipore) using a wet Trans-Blot System (Bio-Rad). After blocking with 5 % fat-free milk in Tris-buffered saline-0·1 % Tween-20 for 1 h, the membrane was incubated overnight with primary antibodies at 4°C. Subsequently, the membrane was incubated with corresponding horseradish-peroxidase-conjugated secondary antibody for 1 h at 37°C. The primary antibodies used were: anti-Cytc (catalogue no. 10993-1-AP, ProteinTech Biotechnology), anti-MyHC I (catalogue no. M8421, Sigma), anti-sirtuin 1 (Sirt1) (catalogue no. 8469S, Cell Signaling), anti-PGC-1α (PPAR-γ coactivator-1α) (catalogue no. 2178S, Cell Signaling) and anti-β-actin (catalogue no. 4967S, Cell Signaling). The protein bands were visualised by the Clarity Western ECL Substrate (Bio-Rad). The densities of bands were determined using the Gel-Pro Analyzer version 4.2 (Media Cybernetics). β-Actin protein was used as a loading control.
Cell culture and treatments
The isolation and identification of porcine skeletal muscle satellite cells were performed as described in our previous study(Reference Chen, Guo and Jia13). Porcine skeletal muscle satellite cells were grown in Dulbecco's modified Eagle’s medium/F12 supplemented with 15 % fetal bovine serum and with 100 U/ml penicillin and 100 μg/ml streptomycin at 37°C in a 5 % CO2 atmosphere. The isolated cells were induced to differentiate in Dulbecco's modified Eagle’s medium/F12 medium supplemented with 2 % of horse serum (Hyclone), when cells were grown to about 80 % confluence. The differentiation medium was replaced with a fresh medium every 24 h. After 3 d of differentiation, the cells were treated with different concentrations (0, 50, 100 and 200 μg/ml) of Arg for 3 d. To explore the molecular mechanism, the cells were pretreated with Rot (catalogue no. R8875, Sigma) (dissolved in dimethyl sulfoxide) for 1 h before being treated with Arg.
Statistical analysis
Data are expressed as mean values with their standard errors. Statistical analysis was performed using SPSS 22.0 statistical software. All data were analysed by one-way ANOVA followed by Tukey’s test. A value P < 0·05 was considered to be statistically significant.
Results
Effect of arginine on growth performance of pigs
Compared with the control group, dietary Arg supplementation had no effect on growth performance of pigs, including final body weight, average daily gain, average daily feed intake and feed-gain ratio (Table 3).
Table 3. Effect of arginine (Arg) on growth performance of piglets (n 15) (Mean values with their standard errors)
ADG, average daily gain; ADFI, average daily feed intake; F:G, feed-gain ratio.
Effect of arginine on succinic dehydrogenase activity of skeletal muscle of pigs
As shown in Fig. 1, 1·0 and 1·5 % Arg supplementation significantly increased (P < 0·05) the activity of SDH in the longissimus dorsi muscle and there was no significant difference between 0·5, 1·0 and 1·5 % Arg treatment groups.
Fig. 1. Effect of arginine (Arg) on succinic dehydrogenase (SDH) activity in longissimus dorsi muscle of weaning piglets. Results are mean values with their standard errors from five piglets. a,bValues with unlike letters are significantly different (P < 0·05).
Effect of arginine on nitric oxide synthase activity and nitric oxide content of skeletal muscle of pigs
The NOS activity was significantly increased (P < 0·05) in 0·5 and 1·0 % Arg treatment groups compared with the control group (Fig. 2(A)). As shown in Fig. 2(B), 1·0 and 1·5 % Arg treatment groups significantly increased (P < 0·05) the NO content compared with the control group.
Fig. 2. Effect of arginine (Arg) on nitric oxide synthase (NOS) activity and nitric oxide (NO) content in longissimus dorsi muscle of weaning piglets. (A) NOS activity. (B) NO content. Results are mean values with their standard errors from five piglets. a,b,cValues with unlike letters are significantly different (P < 0·05).
Effect of arginine on muscle fibre type composition of skeletal muscle of pigs
As shown in Fig. 3, 1·0 % Arg supplementation significantly increased (P < 0·05) the MyHC I, troponin I1, C1 and T1 (Tnni1, Tnnc1 and Tnnt1) mRNA levels. Furthermore, ATPase staining analysis showed that 1·0 % Arg supplementation significantly increased (P < 0·05) the number of type I muscle fibres and significantly decreased (P < 0·05) the number of type II muscle fibres compared with the control group (Fig. 4). Compared with the control group, dietary supplementation of 1·0 or 1·5 % Arg significantly increased (P < 0·05) the protein expression of MyHC I (Fig. 6).
Fig. 3. Effect of arginine (Arg) on mRNA expressions of type I muscle fibres-related genes in longissimus dorsi muscle of weaning piglets. Results are mean values with their standard errors from five piglets. a,b,cValues with unlike letters are significantly different (P < 0·05). 
, 0 % Arg; 
, 0·5 % Arg; 
, 1·0 % Arg; and 
, 1·5 % Arg. MyHC I, myosin heavy chain I; Tnni1, troponin I1; Tnnc1, troponin C1; Tnnt1, troponin T1.
Fig. 4. Effect of arginine (Arg) on the proportion of type I and type II muscle fibres by ATPase staining analysis. Results are mean values with their standard errors from five piglets. a,bValues with unlike letters are significantly different (P < 0·05). 
, 0 % Arg; 
, 0·5 % Arg; 
, 1·0 % Arg; and 
, 1·5 % Arg.
Effects of arginine on mitochondrial biogenesis and function-related marker expression and mitochondrial DNA content of skeletal muscle of pigs
As shown in Fig. 5(A), 1·0 % Arg supplementation significantly increased (P < 0·05) the mRNA expressions of nuclear respiratory factor 1 (NRF1), mitochondria transcription factor B1 (TFB1M) and cytochrome c (Cytc) compared with the control group. Cytc protein expression (Fig. 6) and ATP synthase subunit C1 (ATP5G) mRNA expression (Fig. 5(A)) were also significantly increased (P < 0·05) in the 1·0 % Arg and 1·5 % Arg treatment groups. As shown in Fig. 5(B), mtDNA content of Arg-treated groups was significantly increased (P < 0·05) compared with the control group.
Fig. 5. Effect of arginine (Arg) on mRNA expressions of mitochondrial biogenesis and function-related genes and mitochondrial DNA (mtDNA) content in longissimus dorsi muscle of weaning piglets. (A) mRNA expressions of mitochondrial biogenesis and function-related genes. (B) mtDNA content. Results are mean values with their standard errors from five piglets. a,b,cValues with unlike letters are significantly different (P < 0·05). 
, 0 % Arg; 
, 0·5 % Arg; 
, 1·0 % Arg; and 
, 1·5 % Arg. PGC-1α, PPAR-γ coactivator-1α; NRF1, nuclear respiratory factor 1; TFB1M, mitochondria transcription factor B1; Cytc, cytochrome c; ATP5G, ATP synthase subunit C1.
Fig. 6. Effect of arginine (Arg) on the protein expressions of myosin heavy chain I (MyHC I), cytochrome c (Cytc), PPAR-γ coactivator-1α (PGC-1α) and sirtuin 1 (Sirt1) in longissimus dorsi muscle of weaning piglets. Results are mean values with their standard errors from three piglets. a,b,cValues with unlike letters are significantly different (P < 0·05). 
, 0 % Arg; 
, 0·5 % Arg; 
, 1·0 % Arg; and 
, 1·5 % Arg.
In addition, the PGC-1α mRNA level in longissimus dorsi muscle was significantly increased (P < 0·05) in 1·0 % Arg treatment group compared with the control group (Fig. 5(A)). As shown in Fig. 6, PGC-1α protein expression of Arg-treated groups was significantly increased (P < 0·05) compared with the control group. In addition, protein expressions of Sirt1 of 0·5 and 1·0 % Arg treatment groups was significantly increased (P < 0·05) compared with the control group (Fig. 6).
Rotenone inhibits mitochondrial biogenesis and function-related protein expression in porcine skeletal muscle satellite cells
We found that mitochondrial complex I inhibitor Rot significantly decreased the expression of mitochondrial biogenesis and function-related proteins PGC-1α and Cytc (Fig. 7). We also observed that 0·5 and 1·0 μm Rot could damage cell morphology (data not shown). Therefore, Rot was supplemented with 0·1 μm in the following studies.
Fig. 7. Effect of rotenone (Rot) on PPAR-γ coactivator-1α (PGC-1α) and cytochrome c (Cytc) protein expression in porcine skeletal muscle satellite cells. About 80 % confluent porcine skeletal muscle satellite cells were cultured in the differentiation medium (Dulbecco's modified Eagle’s medium/F12, 2 % horse serum) for 3 d and then treated with different concentrations of Rot (0, 0·1, 0·5 and 1·0 μm) for 24 h. PGC-1α and Cytc protein levels were determined by Western blot analysis. The amount of PGC-1α and Cytc was normalised to the amount of β-actin. The mean values with their standard errors of the densitometry results from three independent experiments are shown in the lower panel. a,b,c,dValues with unlike letters are significantly different (P < 0·05). 
, 0 μm Rot; 
, 0·1 μm Rot; 
, 0·5 μm Rot and 
, 1·0 μm Rot.
Arginine affects the formation of type I muscle fibres by improving mitochondrial biogenesis in porcine skeletal muscle satellite cells
To investigate the effect of Arg on mitochondrial biogenesis and function-related protein expression and type I muscle fibres formation, porcine skeletal muscle satellite cells were treated with different concentrations of Arg (0, 50, 100 and 200 μg/ml). Western blot result showed that the maximal up-regulation of MyHC I, Cytc, PGC-1α and Sirt1 protein expression was observed in 100 μg/ml Arg treatment group (Fig. 8). In the following studies, Arg was supplemented with 100 μg/ml.
Fig. 8. Effect of arginine (Arg) on myosin heavy chain I (MyHC I), PPAR-γ coactivator-1α (PGC-1α), sirtuin 1 (Sirt1) and cytochrome c (Cytc) protein expression in porcine skeletal muscle satellite cells. About 80 % confluent porcine skeletal muscle satellite cells were cultured in the differentiation medium (Dulbecco’s modified Eagle’s medium/F12, 2 % horse serum) for 3 d and then treated with different concentrations of Arg (0, 50, 100 and 200 μg/ml) for 3 d. MyHC I, PGC-1α, Sirt1 and Cytc protein levels were determined by Western blot analysis. The amount of MyHC I, PGC-1α, Sirt1 and Cytc was normalised to the amount of β-actin. The mean values with their standard errors of the densitometry results from three independent experiments are shown in the lower panel. a,b,cValues with unlike letters are significantly different (P < 0·05). , 0 μg/ml Arg; , 50 μg/ml Arg; , 100 μg/ml Arg and , 200 μg/ml Arg.
To determine whether mitochondrial biogenesis contributes to Arg-induced type I muscle fibres formation, mitochondrial complex I inhibitor Rot was used. As shown in Fig. 9, Arg increased MyHC I protein level, whereas Rot annulled the positive effect of Arg on MyHC I protein expression.
Fig. 9. Mitochondrial complex I inhibitor rotenone (Rot) annuls the effect of arginine (Arg) on protein expression of myosin heavy chain I (MyHC I) in porcine skeletal muscle satellite cells. About 80 % confluent porcine skeletal muscle satellite cells were cultured in the differentiation medium (Dulbecco's modified Eagle’s medium/F12, 2 % horse serum) for 3 d and then pretreated with 0·1 μm Rot for 1 h followed by Arg (100 μg/ml) treatment for 3 d. MyHC I protein level was determined by Western blot analysis. The amount of MyHC I was normalised to the amount of β-actin. The mean values with their standard errors of the densitometry results from three independent experiments are shown in the lower panel. a,b,cValues with unlike letters are significantly different (P < 0·05).
Discussion
Arg is classified as a semi-essential or conditionally essential amino acid in mammals. Arg is the unique natural precursor in the biosynthesis of NO by NOS(Reference Palmer, Ashton and Moncada19). A large number of studies showed that NO or NOS plays an important role in regulating muscle fibre type composition(Reference Martins, St-Louis and Murdoch20–Reference Moon, Balke and Madorma22). In the present study, we found that dietary supplementation of 1·0 % Arg increased NO content, NOS activity, protein expression of MyHC I and mRNA expressions of type I muscle fibres-related genes (MyHC I, Tnni1, Tnnc1 and Tnnt1) in longissimus dorsi muscle of weaning piglets. Some studies had indicated that type I and type II muscle fibres can be identified by myosin ATPase staining(Reference Zhang, Wang and Li23,Reference Reyes, Banks and Tsang24) . Our data showed that 1·0 % Arg supplementation significantly increased the number of type I muscle fibres and significantly decreased the number of type II muscle fibres by ATPase staining analysis. Previous reports showed that the type I muscle fibres had higher SDH enzyme activity(Reference Tang, Wagner and Breen25). In the present study, the SDH activity was significantly increased by Arg in longissimus dorsi muscle. Taken together, these findings showed that dietary supplementation of Arg promoted type I muscle fibres formation in weaning piglets.
Mitochondria are well-known organelles that provide energy for eukaryotic cells(Reference Nunnari and Suomalainen26). PGC-1α, a major regulator of mitochondrial biogenesis, is expressed in skeletal muscle(Reference Lehman, Barger and Kovacs27). Sirt1 modulates PGC-1α expression and activity(Reference Reznick and Shulman28) and plays an important role in regulating mitochondrial function(Reference Price, Gomes and Ling29,Reference Chen, Xiang and Jia30) . PGC-1α and Sirt1 control mitochondrial biogenesis and function via the induction and activation of several downstream nuclear transcription factors, such as NRF1, mitochondria transcription factor B1 (TFB1M), ATP5G and cytochrome c (Cytc)(Reference Zhang, Zhou and Wu6,Reference Price, Gomes and Ling29,Reference Vazquez, Lim and Chim31) . Furthermore, there is some evidence that improvement of mitochondria biogenesis and function may play a key role in promotion of type I muscle fibres formation and may be a new mediator of skeletal muscle fibre type(Reference Venhoff, Lebrecht and Pfeifer4–Reference Zhang, Zhou and Wu6). Type I (slow-twitch) muscle fibres are rich in the mitochondria(Reference Schiaffino and Reggiani3,Reference Zierath and Hawley32) . It has been reported that some nutrients (such as leucine and resveratrol) can regulate type I (slow-twitch) muscle fibres expression by improving the mitochondrial biogenesis and function(Reference Price, Gomes and Ling29,Reference Chen, Xiang and Jia30,Reference Peng, Tao and Zhang33) . In the present study, our data found that dietary supplementation of 1·0 % Arg significantly increased the PGC-1α, Sirt1 and Cytc protein expressions, the PGC-1α, NRF1, TFB1M, Cytc and ATP5G mRNA expressions and the mtDNA content, suggesting that Arg plays an important role in regulating mitochondrial biogenesis and function.
It should be noted that there might be an optimal dietary supplementation of Arg in weaning piglets. In the present study, we showed that for many of the outcome measures, 1·5 % Arg was statistically lower than 1·0 % Arg, suggesting that the effectiveness of Arg might be optimal at 1·0 %. The reason might be that dietary supplementation with 1·0 % Arg could meet the needs of pigs, but excessive supplementation was disadvantageous. The exact reasons need to be further studied in the future.
Roteone (Rot) is a common pesticide and has been reported as a specific inhibitor of mitochondrial electron transport chain complex I, the first step in the electron transport chain(Reference Yuhki, Miyauchi and Kakinuma34). There is evidence that the inhibition of mitochondrial electron transport chain can lead to mitochondrial dysfunction(Reference Peng, Tao and Zhang33). Rot has been proven to induce mitochondrial dysfunction(Reference Yee, Freestone and Bai35). Here, we observed that the cellular morphology changed gradually following the increase in Rot concentration, and no obvious damage effects were observed in porcine skeletal muscle satellite cells with 0·1 μm Rot treatment, which was consistent with the result of the previous study in PC12 cells(Reference Wang, Dong and Liu36). Our data indicated that 0·1 μm Rot could effectively inhibit mitochondrial electron transport chain. Interestingly, inhibition of mitochondrial electron transport chain by Rot down-regulated the expression of PGC-1α (a major regulator of mitochondrial biogenesis) and annulled Arg-induced type I muscle fibres formation.
In conclusion, we reported that Arg promotes the formation of porcine type I muscle fibres, which is dependent on the improvement of mitochondrial biogenesis. Our study not only brings new information on the nutritional function of Arg but also deeply understands the mechanism of Arg regulating the formation of type I muscle fibres.
Acknowledgements
This work was supported by the National Key R&D Program of China (no. 2018YFD0500403) and the Sichuan Youth Science and Technology Foundation (no. 2017JQ0008).
X. C. and Z. H. conceived the study and designed the experiments. X. L. carried out the experiments and analysed the data. X. C. wrote the manuscript. X. C., D. C., B. Y., J. H. and Z. H. contributed reagents/materials/analysis tools. Z. H. revised the manuscript. All authors read and approved the final manuscript.
The authors declare that there are no conflicts of interest.
