Introduction

Cardiomyocyte Ca2+ overload occurs in many pathological conditions such as hypoxia, ischemia, oxidative stress, cardiac hypertrophy, and heart failure1,2,3,4. Intracellular Ca2+([Ca2+]i) overload causes cardiac arrhythmias and myocardial dysfunction5. Extensive reports show that the late or persistent sodium current (INaL) in ventricular myocytes is increased in many pathological conditions that lead to [Ca2+]i overload6,7,8,9,10. An increase in the amplitude of INaL prolongs the action potential duration, increases the transmural dispersion of repolarization, and causes cardiac arrhythmias11. An increase in INaL also increases the intracellular sodium concentration and subsequently raises [Ca2+]i via the reverse-mode Na+-Ca2+ exchanger (NCX)7,12. Inhibition of INaL was reported to attenuate the increase in [Ca2+]i13,14,15. Inhibition of INaL is therefore a potential therapeutic target for the treatment of heart diseases associated with [Ca2+]i overload16,17.

Ketamine is an intravenous and intramuscular anesthetic that is widely used in both humans and animals. In vitro study data show that ketamine has antiarrhythmic effects and decreases the incidence of reperfusion-induced arrhythmias18,19,20,21 and that it enhances the recovery of force of contraction during reperfusion22. Furthermore, ketamine suppresses the activity of neutrophils and decreases their postischemic adhesion in the coronary artery23,24,25. In clinical studies, ketamine has reduced the incidence of ventricular arrhythmias and clinical markers of myocardial injury in cardiac surgery patients26,27,28. These results suggest that ketamine may have cardioprotective effects, but the underlying mechanisms are still unknown. Ketamine has been reported to inhibit various ionic currents, including the L-type Ca2+ current (ICaL)29,30,31,32,33,34, peak sodium current (INa)32, inward rectifier K+ current (IK1)30,35, delayed rectifier K+ current (IK)30, ATP-sensitive K+ current (IKATP)36,37, and human ether-a-go-go-related gene (hERG) channel38. However, there is no study regarding the effects of ketamine on INaL. In previous studies, we found that inhibition of INaL attenuates augmented INaL-induced [Ca2+]i overload39,40. Thus, this study investigated the effects of ketamine on INaL, the NCX current (INCX), myocyte shortening and [Ca2+]i transients in the presence of sea anemone toxin II (ATX), an opener of INaL channels.

Materials and methods

Isolation of ventricular myocytes

This study adheres to the Guidance for Ethical Treatment of Laboratory Animals (the Ministry of Science and Technology of China, 2006) and is approved by the Institutional Animal Care and Use Committee of the Medical College of Wuhan University of Science and Technology (Wuhan, China).

Myocytes were isolated enzymatically from the hearts of rabbits of both sexes (1.7–2 kg; Wuhan Institute of Biological Products Co, Ltd, Wuhan, China), as previously described 41. In brief, adult New Zealand white rabbits were heparinized (2000 U) and anesthetized with ketamine (iv; 30 mg/kg) and xylazine (im; 7.5 mg/kg). Hearts were quickly excised and retrogradely perfused with Ca2+-free Tyrode's solution for 5 min, followed by an enzyme-containing solution (0.1 g/L collagenase type I, 0.01 g/L protease E and 0.5 g/L bovine serum albumin) for a further 40–50 min. The perfusate was finally switched to KB solution for 5 min. All solutions were bubbled with 100% O2 and maintained at 37 °C. The left ventricle was cut into small chunks and gently agitated in KB solution. The cells were filtered through nylon mesh and stored in KB solution at 4 °C until used.

Solution

For cell isolation, the regular Tyrode's solution contained the following (in mmol/L): 135 NaCl, 0.33 NaH2PO4, 5.4 KCl, 1.8 CaCl2, 1 MgCl2, 10 glucose, and 10 HEPES (pH 7.4). The KB solution contained the following (in mmol/L): 70 KOH, 40 KCl, 20 KH2PO4, 1 MgCl2, 20 taurine, 50 glutamic acid, 0.5 EGTA, 10 glucose, and 10 HEPES (pH 7.4). For INaL recordings, the intracellular (pipette) solution contained the following (in mmol/L): 120 CsCl, 1 CaCl2, 11 EGTA, 5 MgCl2, 5 Na2ATP, 10 TEA-Cl, and 10 HEPES (pH 7.3). The bath solution contained the following (in mmol/L): 135 NaCl, 0.33 NaH2PO4, 5.4 CsCl, 1.8 CaCl2, 1 MgCl2, 0.05 CdCl2, 0.3 BaCl2, 10 glucose, and 10 HEPES (pH 7.4). For the hypoxia experiment, the modified bath solution in which glucose was omitted was pre-equilibrated with 100% N2 for at least 1 h. Hypoxia was induced using a previously described method42. For INCX recordings, the pipette solution included the following (in mmol/L): 20 NaCl, 10 CaCl2, 3 MgCl2, 5 MgATP, 50 aspartic acid, 20 EGTA, 10 HEPES, and 120 CsOH (pH 7.4). The bath solution contained the following (in mmol/L): 140 NaCl, 2 CsCl, 2 CaCl2, 1 BaCl2, 2 MgCl2, 5 HEPES, and 10 glucose (pH 7.4). In addition, 20 μmol/L ouabain and 1 μmol/L nicardipine were added to block the Na+-K+ pump and ICaL, respectively. For ICaL recordings, the pipette solution contained the following (in mmol/L): 80 CsCl, 60 CsOH, 0.65 CaCl2, 5 disodium creatine phosphate, 5 MgATP, 40 aspartic acid, 10 EGTA, and 5 HEPES (pH 7.3). The bath solution was the Tyrode's solution. For cell shortening and [Ca2+]i transient recordings, the bath solution contained the following (in mmol/L): 131 NaCl, 4 KCl, 1.8 CaCl2, 1 MgCl2, 10 HEPES, and 10 glucose (pH 7.4).

Current recordings

All experiments were conducted at 22–25 °C. The electrode resistance (when filled with pipette solution) was 1.5–2 MΩ. Cell capacitance and series resistances were electronically compensated by 60%–80%. Currents were recorded using an EPC-9 amplifier (HEKA Electronic, Lambrecht, Pfalz, Germany), filtered at 2 kHz and sampled at 10 kHz. Current measurements were normalized using the cell capacitance.

INaL was recorded by a 300-ms depolarizing pulse to −20 mV from a holding potential of −120 at a frequency of 0.2 Hz. The amplitude of INaL was determined from the average current measured during a time interval of 190 to 210 ms after initiation of the depolarizing pulse to eliminate any contribution of INa 2. To record the current-voltage relationship of INaL, 300-ms depolarizing pulses to membrane potentials from −80 to +50 mV were applied at 0.5 Hz from a potential of −120 mV.

INCX was elicited by a 10-ms prepulse to +60 mV from a holding potential of −40 mV followed by a 2-s ramp pulse from +60 to −120 mV (with a speed of −90 mV/s). INCX was measured as the current sensitive to 5 mmol/L Ni2+ at +50 and −100 mV.

ICaL was elicited by a 150-ms prepulse to −40 mV from a holding potential of −80 mV followed by a 300-ms depolarizing pulse from −40 mV to 0 mV (0.2 Hz). ICaL was measured as the difference between peak inward current and the current remaining at the end of the 300-ms pulse.

Measurements of myocyte cell shortening and [Ca2+]i transients

Fura-2 was loaded by incubating cell suspensions with 1 μmol/L Fura-2/AM for 30 min at 25 °C in the dark. Fura-2-loaded myocytes mounted in a chamber situated on the stage of an Olympus IX-70 inverted microscope were field stimulated to contract between platinum electrodes (0.5 Hz, 37 °C). Cardiomyocytes that possessed an appropriate morphological appearance (rod shaped with clean edges, clear striations, and no large blebs), a resting sarcomere length >1.70 μm, and no spontaneous contraction was selected for experimentation. Myocyte shortening and [Ca2+]i transients were measured simultaneously using a video-based edge detection and dual excitation fluorescence photomultiplier system (IonOptix, Milton, MA, USA). A xenon lamp provided the excitation light. Alternating excitation wavelengths of either 340 nm or 380 nm were obtained at a frequency of 250 Hz. A photomultiplier collected the emitted fluorescence signals. The ratio of both Fura-2 fluorescence signals (340/380 ratio) was continuously measured after background fluorescence subtraction. Drugs were applied after a 10-min stable period. Contractile variables included peak shortening amplitude (PS; μm), maximal rate of shortening (+dL/dt; μm/ms), maximal rate of relaxation (−dL/dt; μm/ms) and time to peak shortening (TPS; s). For [Ca2+]i transients, the following parameters were measured: diastolic [Ca2+]i (340/380 ratio), amplitude of the [Ca2+]i transient (Δ[Ca2+]i; ratio), maximal rate of [Ca2+]i rise (+d[Ca2+]i/dt; 1/ms), maximal rate of [Ca2+]i decay (−d[Ca2+]i/dt; 1/ms), time to peak (TP; s) and time to 90% decay (TD90; s).

Chemicals

Ketamine, ATX, tetrodotoxin (TTX) and protease E were purchased from Sigma Chemical (Saint Louis, MO, USA). Hydrogen peroxide (H2O2) was obtained from Wuhan Zhongnan Chemical Reagent Co (Wuhan, China). Fura-2/AM and collagenase type I were obtained from Tocris (Ellisville, MO, USA) and Gibco (GIBCOTM, Invitrogen, Paisley, UK), respectively.

Statistical analysis

The data are expressed as the mean±SD. A paired Student's t-test was performed for comparisons between two groups. Statistical analysis was performed using repeated measures analysis of variance (ANOVA), followed by the Scheffé test for multiple comparisons. Statistical analyses were performed using Origin software v7.0 (OriginLab, Northampton, MA). All statistical tests were two-tailed. A value of P<0.05 was considered statistically significant.

Results

Ketamine inhibited INaL in rabbit ventricular myocytes in a concentration-dependent manner

The whole-cell patch-clamp technique was used to record INaL. Ketamine (20, 40, 80 μmol/L) decreased the current density of INaL from 0.34±0.08 to 0.28±0.08, 0.22±0.07, and 0.14±0.06 pA/pF (P<0.01 vs control for all; n=11) in a concentration-dependent manner, respectively (Figure 1). This inhibitory effect of ketamine was reversible upon washout (0.34±0.09 vs 0.33±0.11 pA/pF, P>0.05, n=6). Figures 1A and 1B show the representative current records and summary data for the INaL current density.

Figure 1
figure 1

Ketamine (Ket) inhibited INaL in rabbit ventricular myocytes in a concentration-dependent manner. (A) Representative whole-cell recordings of INaL in the absence of drug (Control) and after the application and washout of Ket (20, 40, or 80 μmol/L). (B) Summary data for the mean current density of INaL under different conditions. The data are expressed as the mean±SD (n=11). cP<0.01 vs Control. fP<0.01 vs 20 μmol/L Ket. iP<0.01 vs 40 μmol/L Ket.

PowerPoint slide

Full size image

Effect of ketamine on ATX-augmented INaL

The INaL channel opener ATX (30 nmol/L) increased INaL (at −20 mV) from 0.29±0.01 to 1.23±0.05 pA/pF (P<0.01; n=8). In the presence of ATX, ketamine (20, 40, 80 μmol/L) decreased it to 1.09±0.08, 0.88±0.08, and 0.72±0.06 pA/pF (P<0.01 vs ATX for all; n=8) in a concentration-dependent manner, respectively (Figure 2). Figures 2A and 2B show the original current records and current-voltage curves of INaL according to the current-voltage relationship protocol, respectively. Figure 2C shows the summary data for the INaL current density recorded at −20 mV.

Figure 2
figure 2

Ketamine (Ket) inhibited ATX-augmented INaL in a concentration-dependent manner. (A) Representative whole-cell recordings of INaL in the absence of drug (Control) and in the presence of ATX (30 nmol/L) before or after the application of Ket (20, 40, or 80 μmol/L). (B) The current-voltage relationship for INaL. (C) Summary data for the mean current density of INaL (at −20 mV) in the absence (Control) and presence of ATX and ATX plus Ket. The data are expressed as the mean±SD (n=8). cP<0.01 vs Control; fP<0.01 vs 30 nmol/L ATX; iP<0.01 vs ATX plus Ket 20 μmol/L; lP<0.01 vs ATX plus Ket 40 μmol/L.

PowerPoint slide

Full size image

Effect of ketamine on the enhanced INaL induced by hypoxia or H2O2

Previous reports show that INaL is increased under hypoxia and oxidative stress conditions. Thus, we studied the effects of ketamine on INaL after exposure to hypoxia or H2O2. Hypoxia (10 min) and 300 μmol/L H2O2 increased INaL from 0.33±0.05 to 0.64±0.06 pA/pF (P<0.01; n=6) and 0.28±0.03 to 0.64±0.10 pA/pF (P<0.01; n=7), respectively. Ketamine (40 μmol/L) decreased it to 0.45±0.06 and 0.43±0.11, respectively (Figure 3). After washing, INaL returned to the predrug level (hypoxia: 0.67±0.05 vs 0.63±0.08 pA/pF, n=4; H2O2: 0.68±0.04 vs 0.67±0.03 pA/pF, n=3; both P>0.05).

Figure 3
figure 3

Ketamine (Ket) inhibited the enhanced INaL induced by hypoxia (A, B) or H2O2 (C, D). (A) and (C), Representative whole-cell recordings of INaL in the absence of drug (Control) and before and after the application and washout of Ket (40 μmol/L) after treatment with 10 min of hypoxia (A) or 300 μmol/L H2O2 (C), respectively. (B) and (D), Summary data for the mean current density of INaL under different conditions. The data are expressed as the mean±SD (n=6 or 7). cP<0.01 vs Control; fP<0.01 vs hypoxia (B) or H2O2 (D).

PowerPoint slide

Full size image

Effects of ketamine on the increased reverse-mode INCX induced by ATX

ATX (30 nmol/L) increased the reverse-mode INCX (at +50 mV) from 0.88±0.08 to 2.76±0.21 pA/pF (P<0.01; n=8). In the presence of ATX, TTX (4 μmol/L) decreased it to the control level (1.04±0.12 pA/pF, P>0.05 vs Control; Figure 4A–4C). Similarly, ketamine (20, 40, 80 μmol/L) decreased the ATX-stimulated INCX from 2.61±0.22 to 2.21±0.22, 1.71±0.16, and 1.20±0.22 pA/pF in a concentration-dependent manner, respectively (n=8, P<0.01 vs ATX for all; Figure 4D–4F).

Figure 4
figure 4

Effects of TTX and ketamine (Ket) on the enhanced INCX induced by ATX. (A) and (D), Representative original currents recorded by a ramp pulse (inset) in the absence (Control) and presence of ATX 30 nmol/L before and after exposure to TTX 4 μmol/L (A) or Ket at 20, 40, or 80 μmol/L (D). (B) and (E), The Ni2+-sensitive INCX was obtained by subtracting the sweep after the application of 5 mmol/L NiCl2 (trace d for A; trace f for D) from sweeps before exposure to NiCl2. (C) and (F), Summary data for the current density of INCX measured at +50 mV and −100 mV. The data are expressed as the mean±SD (n=8). cP<0.01 vs Control; fP<0.01 vs ATX 30 nmol/L (ATX); iP<0.01 vs ATX plus Ket 20 μmol/L (ATX-K20, F); lP<0.01 ATX plus Ket 80 μmol/L (ATX-K80, F) vs ATX plus Ket 40 μmol/L (ATX-K40, F).

PowerPoint slide

Full size image

Effects of TTX and ketamine on enhanced cell shortening and [Ca2+]i transients induced by ATX

ATX (3 nmol/L) enhanced cell shortening and [Ca2+]i transients (Figures 5 and 6). PS, +dL/dt, −dL/dt, diastolic [Ca2+]i, Δ[Ca2+]i, and +d[Ca2+]i/dt, and −d[Ca2+]i/dt were increased to 160%, 180%, 191%, 114%, 165%, 159%, and 162% of control, respectively, and TD90 was decreased to 82% of control by ATX (P<0.01 vs control for all; n=7). In the presence of ATX, TTX (2 μmol/L) reverted these measures to 99%, 97%, 101%, 100%, 104%, 98%, 106%, and 98% of control, respectively (P>0.05 vs control for all; Figure 5). Similarly, ketamine (40 μmol/L) reverted the above parameters from 169% to 99%, 169% to 99%, 224% to 114%, 110% to 99%, 165% to 104%, 159% to 98%, 162% to 106%, and 82% to 98% of control, respectively (n=7, P<0.01 ATX vs ketamine group for all; Figure 6).

Figure 5
figure 5

Effects of TTX on enhanced myocyte shortening and [Ca2+]i transients induced by ATX. (A) Representative recordings of cell shortening (upper) and [Ca2+]i transients (lower) in the absence (Control) and presence of 3 nmol/L ATX before and after exposure to TTX 2 μmol/L. (B) Summary data for representative parameters of cell shortening and [Ca2+]i transients. The data are expressed as the mean±SD (n=7). cP<0.01 vs Control; fP<0.01 vs ATX group.

PowerPoint slide

Full size image
Figure 6
figure 6

Effects of ketamine (Ket) on enhanced myocyte shortening and [Ca2+]i transients induced by ATX. (A) Representative recordings of cell shortening (upper) and [Ca2+]i transients (lower) in the absence (Control) and presence of ATX 3 nmol/L before and after exposure to Ket 40 μmol/L. (B) Summary data for representative parameters of cell shortening and [Ca2+]i transients. The results are expressed as the mean±SD (n=7). cP<0.01 vs Control; fP<0.01 vs ATX group.

PowerPoint slide

Full size image

Effect of ketamine on ICaL

ICaL plays an important role in cell shortening and [Ca2+]i transients. Previous studies show that ketamine inhibits ICaL in the cardiomyocytes of some species29,30,31,32,33,34. However, no studies have investigated the effect of ketamine on ICaL in rabbit ventricular myocytes. The effect of ketamine on ICaL could be attributed to ketamine's suppressive effect on myocyte shortening in this study. Thus, in this study, we examined the effect of ketamine on ICaL and observed that ketamine (40 μmol/L) significantly inhibited ICaL in rabbit ventricular myocytes. The current density of ICaL was decreased from 4.41±1.15 to 2.94±1.06 pA/pF (P<0.01, n=9; Figure 7). After washing, ICaL returned to the predrug control level (4.65±0.74 vs 4.54±0.67 pA/pF, P>0.05, n=3).

Figure 7
figure 7

Ketamine (Ket) inhibited ICaL in rabbit ventricular myocytes. (A) Representative whole-cell recordings of ICaL in the absence of drug (Control) and after the application and washout of Ket (40 μmol/L). (B) Summary data for the mean current density of ICaL. The data are expressed as the mean±SD (n=9). cP<0.01 vs Control.

PowerPoint slide

Full size image

Discussion

Cardiomyocyte [Ca2+]i overload occurs in many pathological conditions such as hypoxia, ischemia, oxidative stress, cardiac hypertrophy, and heart failure1,2,3,4 and results in cardiac arrhythmias and myocardial dysfunction5. Previous reports show that INaL is an important contributing factor to [Ca2+]i overload in many pathological conditions. An increase in INaL can increase the intracellular sodium concentration and subsequently raise [Ca2+]i through reverse NCX7,12. Extensive studies have reported that inhibition of INaL attenuates the increase in [Ca2+]i13,14,15. Inhibition of INaL is therefore a potential therapeutic target for the treatment of heart diseases associated with [Ca2+]i overload16,17.

Ketamine is an intravenous and intramuscular anesthetic that is widely used in pediatric and adult cardiac surgery. In clinical studies, the serum concentrations of ketamine reach 60 μmol/L 5 min after an intravenous administration of 2 mg/kg43. The protein binding of ketamine is 12%–50% 44,45, and thus, the free concentration of ketamine may reach 30–50 μmol/L at 5 min. However, greater plasma concentrations can be expected at 1–3 min after induction and may reach 100–150 μmol/L. Therefore, our experimental concentrations (20–80 μmol/L) appear likely to be within the clinical range.

Ketamine has been reported to inhibit various ionic currents29,30,31,32,33,34,35,36,37,38. Hara et al reported that ketamine (30, 100, 300 μmol/L) dose-dependently blocked peak INa in guinea pig ventricular myocytes32. However, no studies have investigated the effects of ketamine on INaL in cardiomyocytes until now. In this study, we found that ketamine inhibited control and the enhanced INaL induced by ATX, hypoxia, or H2O2 (Figures 1,2,3). The inhibitory effect of ketamine was reversible (Figures 1, 3). In our present and previous studies39, the ATX-augmented reverse-mode INCX, myocyte shortening and [Ca2+]i transients were reversed completely by TTX (Figures 4,5), which shows that an increase in INaL causes these changes. The inhibition of INaL is predicted to reverse the ATX-induced changes described above. Consistent with this hypothesis, ketamine inhibited ATX-augmented reverse INCX, myocyte shortening and [Ca2+]i overload (Figures 4, 6). This result is similar to that obtained in our previous reports, which describe the inhibition of reverse INCX by sophocarpine and resveratrol, cell shortening and [Ca2+]i overload by inhibiting INaL39,40. Therefore, the inhibition of INaL contributes to ketamine's suppressive effects on [Ca2+]i transients and myocyte shortening.

In this study, we observed an interesting phenomenon: ketamine partly inhibited ATX-augmented INaL and INCX (Figures 2, 4) but completely, although partly not expectedly, reversed the changes in [Ca2+]i transients and myocyte shortening induced by ATX (Figure 6). This result shows that there are additional mechanisms other than inhibiting INaL underlying ketamine's suppressive effects. Previous studies show that ketamine inhibits ICaL in guinea pig and rat ventricular myocytes, in human right atrial myocytes and bullfrog atrial myocytes29,30,31,32,33,34. Hara et al reported that ketamine (30 μmol/L) decreased ICaL by 26.1% in guinea pig ventricular myocytes 32. Endou et al reported that ketamine (100 μmol/L) decreased ICaL by 10.8% in rat ventricular myocytes30. However, no study had yet investigated the effect of ketamine on ICaL in rabbit ventricular myocytes. In this study, 40 μmol/L ketamine, a concentration used in our myocyte shortening and [Ca2+]i transient recordings, inhibited ICaL by 33.4% in rabbit ventricular myocytes (Figure 7), which is similar to the findings of Hara et al32 but different from those of Endou et al30. These discrepancies seem to originate from the species differences. The present result suggested that the inhibition of ICaL could also contribute to ketamine's suppressive effects on [Ca2+]i transients and myocyte shortening.

Extensive studies demonstrate that perioperative myocardial ischemia is common and is associated with serious cardiac morbidities and mortality. Its incidence in noncardiac surgery patients at risk of or with known coronary artery disease is 20%–63% 46. The incidence of intraoperative myocardial ischemia in patients undergoing coronary artery bypass grafting surgery is 26%–78%47. A previous report has shown that hypoxia (8 min) increases INaL and reverses INCX in rabbit ventricular myocytes42, which suggests an increase in [Ca2+]i at that time. Furthermore, a burst of H2O2 is generated in cardiomyocytes during ischemia, and H2O2 also increases INaL in cardiomyocytes. As such, cardiomyocyte [Ca2+]i overload will probably occur in the perioperative period.

Ketamine is widely used in pediatric and adult cardiac surgery. Previous studies show that ketamine has antiarrhythmic effects and decreases the incidence of reperfusion-induced arrhythmias in animal models18,19,20,21. Hanouz et al reported that ketamine preconditions human myocardium and enhances the recovery of force of contraction during reperfusion22. Furthermore, ketamine has been shown to suppress the activity of neutrophils and decrease their postischemic adhesion in the coronary artery in in vitro studies23,24,25. In clinical studies, ketamine has reduced the incidence of ventricular arrhythmias and clinical markers of myocardial injury in cardiac surgery patients26,27,28. These results suggest that ketamine may have cardioprotective effects. In this study, ketamine inhibited the augmented INaL induced by hypoxia and H2O2 (Figure 3) and attenuated the augmented INaL-induced [Ca2+]i overload (Figure 6). Therefore, ketamine could protect the heart against ischemia/reperfusion injury in the perioperative period by preventing [Ca2+]i overload and may be a good candidate anesthetic for clinic use.

In summary, ketamine inhibited INaL and ICaL and decreased the enhanced reverse INCX, myocyte shortening and [Ca2+]i transients induced by ATX in rabbit ventricular myocytes. Ketamine could protect ventricular myocytes against increased INaL-induced [Ca2+]i overload in anesthetic use.

Author contribution

An-tao LUO, Zhen-zhen CAO, and Ji-hua MA designed the research; An-tao LUO, Zhen-zhen CAO, Yu XIANG, Shuo ZHANG, Chun-ping QIAN, Chen FU, and Pei-hua ZHANG performed the experiments; An-tao LUO, Zhen-zhen CAO, Yu XIANG, Shuo ZHANG, Chun-ping QIAN, and Chen FU analyzed the data; An-tao LUO, Zhen-zhen CAO, and Ji-hua MA wrote the paper.