*2.1. Reagents*

UA, DA, NaH2PO4, and Na2HPO4 were purchased from Aladdin Reagents Co., Ltd. (Shanghai, China). K4[Fe(CN)6], K3[Fe(CN)6], MnSO4, KMnO4, NaOH, H3PO4, and absolute ethanol were supplied by Sinopharm Chemical Reagent Co. Ltd. (Shanghai, China). All of the chemicals were analytically pure and used as received. NrGO was supplied by Nanjing Xianfeng NANO Material Tech Co. Ltd. (Nanjing, China). Human serum samples were provided by Zhuzhou People's Hospital (Zhuzhou, China). The human samples are a mixture of residual serum from various individuals after clinical examination. Deionized water with the resistivity of 18.2 MΩ was used in all of the experiments.

#### *2.2. Materials Characterization*

Crystalline structures and surface morphologies of MnO2NFs and MnO2NFs/NrGO were investigated by powder X-ray diffractometry (XRD) and scanning electron microscopy (SEM), respectively. SEM images were taken from a cold field-emission SEM (Hitachi S-4800, Tokyo, Japan). The XRD patterns of MnO2NFs were collected using a powder XRD system (PANalytical, Almelo, The Netherlands) with monochromatized Cu Kα radiation (λ = 0.1542 nm), which was operated at 40 kV and 40 mA.

#### *2.3. Synthsis of MnO2NFs*/*NNrGO Comoposites*

The MnO2NFs was prepared by a slow addition of MnSO4 into KMnO4 solution followed by a simple stirring procedure. Typically, 1 mmol of KMnO4 and 1.5 mmol of MnSO4 were adequately dissolved into 20 mL deionized water, separately. Then, the MnSO4 solution was added dropwise into KMnO4 solution at a rate of 1 mL min−1, and agitated continuously at room temperature for 2 h. The resultant product was collected by centrifugation at 12,000 rpm, followed by cleaning alternately with absolute alcohol and deionized water three times, and dried at 60 ◦C in a vacuum oven overnight. Obviously, this route is time-saving and more convenient when compared with the conventional hydrothermal method.

MnO2NFs/NrGO composites were prepared as follows. Firstly, 10 mg MnO2NFs were uniformly dispersed in 10 mL deionized water under an ultrasonication bath for 0.5 h. Then 0.2 g NrGO nanosheets were added into the above MnO2NFs dispersion (1 mg mL−1) and dispersed under ultrasonication for 1 h. The MnO2NFs/NrGO were stored at 4 ◦C in a refrigerator when not used. To ensure good reproducibility, the MnO2NFs/NrGO were subjected to ultrasonication for 0.5 h before each modification.

#### *2.4. Fabrication of MnO2NFs*/*NrGO Modified Electrodes*

The bare glassy carbon electrodes (GCEs) were carefully polished using 0.3 μm and 0.05 μm alumina slurry, then alternately washed by anhydrous alcohol and deionized water several times, and allowed to dry under an infrared lamp. The MnO2NFs/NrGO-modified GCE (MnO2NFs/NrGO/GCE) was prepared via a simple drop-casting approach. Specifically, 5 μL MnO2NFs/NrGO dispersion was

carefully dropped and casted on the GCE surface with a micropipette, then dried with an infrared lamp to form a sensing film. For comparison, MnO2NFs and NrGO-modified GCEs (MnO2NFs/GCE, NrGO/GCE) were also prepared via similar procedures.

#### *2.5. Procedures for Electrochemical Mesurements*

For all the electrochemical tests, a typical three-electrode assemble was immersed into a 10 mL electrochemical cell, in which a bare or modified GCE was worked as the working electrode. Saturated calomel electrode (SCE) and platinum wire were used as a reference electrode and auxiliary electrode, respectively. To evaluate the electrochemical performance of various modified electrodes, cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) was measured in the 0.1 M phosphate bu ffered solution (PBS, pH 7.0), using 0.5 mM [Fe(CN)6] 3−/4− as redox probe couples. EIS plots for di fferent electrodes were recorded at open circuit potential using 5 mV (rms) AC sinusoid signal at a frequency range from 100,000 Hz to 0.1 Hz. The voltammetric responses of 10 μM DA and UA at di fferent electrodes were tested by CV. After a suitable accumulation, linear scanning voltammetry (LSV) was performed for the determination of DA and UA. The potentials were scanned from 0 V to 0.8 V at 100 mVs−<sup>1</sup> for both CV and LSV.

#### **3. Results and Discussion**

#### *3.1. Physical Chararazation*

The crystalline structure of MnO2 nanoflowers was characterized by XRD. As presented in Figure 1, sharp di ffraction peaks were observed at 2θ of 12.94◦, 18.34◦, 28.78◦, 37.66◦, 42.14◦, 49.90◦, 56.44◦, 60.26◦, 69.74◦, 71.34◦, and 73.72◦, which can be well-indexed into (110), (200), (310), (211), (301), (411), (600), (521), (541), (222), and (730) facets, respectively. It is in good agreemen<sup>t</sup> with XRD standard card JSPDF 44-0141 [16,17], suggesting tetragonal crystalline of α-MnO2 were successfully synthesized. Moreover, no visible peak relating to impurities appears, indicating high-purity of α-MnO2. SEM images of MnO2NFs are shown in Figure 2A, B. Obviously, flower-like nanostructures composed of interconnected nanoflakes suggests MnO2 nanoflowers were successfully synthesized. The porous microstructures indicate that MnO2 nanoflowers have a large specific surface area, which is favorable for electrochemical sensing. As shown in Figure 2C,D, the NrGO nanosheets were warped on the surface of MnO2NFs, suggesting MnO2NFs/NrGO nanocomposites were successfully prepared.

**Figure 1.** X-ray di ffraction (XRD) patterns of MnO2 nanoflowers and standard card JSPDF 44-0141.

**Figure 2.** Scanningelectronmicroscopy (SEM)imagesofMnO2NFs (**A**,**B**) andMnO2NFs/NrGOnanocomposites (**C**,**D**) at different magnifications.

#### *3.2. Evaluation of Electrochemical Performances*

In order to assess the electrochemical performances, CVs for various modified electrodes were measured in a mixture solution of 0.5 mM [Fe(CN)6]<sup>3</sup>−/4− and 0.1 M KCl (Figure 3A). A pair of quasi-reversible redox peaks occurred on all of the electrodes with *<sup>i</sup>*pa/*i*pc ≈ 1.0. At bare GCE, a pair of weak redox appeared with the anodic and cathodic peak current of 12.23 and 9.02 μA, respectively. After the modification of GCE by MnO2NFs or NrGO, the redox peak currents increased by 2-fold approximately. As expected, a well-defined and sharp redox peak was observed at the MnO2NFs/NrGO/GCE, with the highest anodic and cathodic peak currents (*i*pa = 92.41 μA, *i*pc = 87.89 μA). It indicates that MnO2NFs/NrGO significantly improved electrochemical performances. It is well-known that the effective electroactive area is a critical factor that directly influences the electrochemical sensing performances. The effective electroactive areas of different electrodes were also calculated, using the *Randles*–*Sevcik* equation as follows [16,17,20]:

$$i\_{\rm pc} = (2.69 \times 10^5) \,\mathrm{m}^{3/2} \,\mathrm{D}^{1/2} \,\mathrm{v}^{1/2} \,\mathrm{AC} \tag{1}$$

where *i*pc represents the cathodic peak current (A), *n* represents the electron transfer number, D represents the diffusion coefficient of K3[Fe(CN)6] (7.6 × 10−<sup>6</sup> cm<sup>2</sup> s<sup>−</sup><sup>1</sup> [39]), *v* denotes the scanning rate (V s<sup>−</sup>1), A denotes the effective electroactive area (cm2), and *C* denotes the K3[Fe(CN)6] concentration (mol cm<sup>−</sup>3). The effective electroactive areas were estimated to be 0.0770, 0.3183, 0.3958, and 0.7496 cm<sup>2</sup> for the bare GCE, MnO2NFs/GCE, NrGO/GCE, and MnO2NFs/NrGO/GCE, respectively. The effective electroactive area of MnO2NFs/NrGO/GCE is about 9-fold higher than that of the bare GCE approximately. The results sugges<sup>t</sup> that the MnO2NFs/NrGO nanocomposites significantly enlarged the effective electroactive surface area, which promoted the accumulation of target analysts and thus increased the response electrochemical signals.

**Figure 3.** Cyclic voltammetry (CVs) (**A**) and Nyquist plots (**B**) for various electrodes in 0.5 mM [Fe(CN)6]<sup>3</sup>−/4− containing 0.1 M KCl solution. The inset in Figure 3B represents the magnification of Nyquist at a higher frequency.

EIS has intensively been used to investigate interfacial properties of various electrochemical sensors [16,17,40–42]. Nyquist plots for various electrodes are plotted in Figure 3B. Obviously, Nyquist plots comprise of the semicircular at higher frequencies relating to the electron transfer-limited process, and linear portions at lower frequencies corresponding to the diffusion-controlled process. The semicircular diameter represents the charge transfer resistance (*R*ct). The *R*ct values for the bare GCE, MnO2NFs/GCE, NrGO/GCE, and MnO2NFs/NrGO/GCE are 1950, 2551, 72.47, and 25.28 ohm, respectively. After modification with MnO2NFs, the *R*ct value increased by 601 ohm because of the poor electro-conductivity of MnO2. When GCE was modified with NrGO, the *R*ct significantly decreased to 74.25 ohm, which can attribute to the good electro-conductivity and high-specific surface area of NrGO [35,36]. As expected, the lowest *R*ct value was obtained at MnO2NFs/NrGO/GCE, probably due to the existence of abundant electrocatalytic active sites that can greatly accelerate the redox reaction of [Fe(CN)6]<sup>3</sup>−/4<sup>−</sup>. The results demonstrate that the MnO2NFs/NrGO can effectively decrease the *R*ct.

#### *3.3. Voltammetric Responses of DA and UA at Various Electrodes*

CV responses of 10 μM DA and UA (1:1) were measured at different electrodes in 0.1 M PBS (pH 3.93) (Figure 4). When the potentials were scanned from 0 to 0.8 V, two anodic peaks were observed at all electrodes, which are closely related to the oxidation of DA and UA. However, only one peak belonging to the reduction of DA occurred at reverse scanning. These phenomena indicate that the electrooxidation of UA is totally irreversible. On the bare GCE, two very weak anodic peaks appeared (*i*pa(DA) = 1.126 μA, *i*pa(UA) = 0.385 μA), demonstrating sluggish kinetics for the electrooxidation of DA and UA. After modification of the GCE by MnO2NFs, the *i*pa(DA) increased a little (*i*pa(DA) = 3.766 μA) while the ipa(UA) was enhanced significantly (ipa(UA) = 12.73 μA), showing MnO2NFs have good electrocatalytic toward the oxidation of UA because of the presence of Mn<sup>4</sup>+/Mn3<sup>+</sup> as an electron mediator. Moreover, the high-specific surface area also contributed to the obvious enhancement on the *i*pa(UA). When GCE was modified with NrGO, the *i*pa(DA) and *i*pa(UA) increased to 7.029 μA and 16.43 μA respectively, suggesting superb electrocatalytic activity toward the oxidation of DA and UA. The superb electrocatalytic activity of NrGO can explain the following facts. Nitrogen atoms in NrGO sheets may interact with target biomolecules via hydrogen bond, which can activate the amine and hydroxy groups and expedite the charge transfer process. Meanwhile, the π–π interactions between NrGO and target biomolecules can also facilitate the charge transfer process [43]. Two sharp anodic peaks occurred at MnO2NFs/NrGO/GCE, and the anodic peak currents enhanced remarkably (*i*pa(DA) = 14.8 μA, *i*pa(UA) = 36.3 μA). The synergistic effect between MnO2NFs and NrGO sheets was mainly responsible for the enhanced response peak currents. Specially, MnO2NFs had higher catalytic activity toward the oxidation of DA and UA when the electrical conductivity was improved by coupling with NrGO sheets. Meanwhile, the hydrogen bond and π–π interactions between NrGO sheets and target biomolecules can also facilitate the charge transfer process. It is worth noting the biggest peak potential separation (about 150 mV) at MnO2NFs/NrGO/GCE, rendering this composite more selective for the simultaneous detection of UA and DA. Besides, the largest background current was also obtained at the MnO2NFs/NrGO nanocomposites, due to the high-specific capacitance of the MnO2NFs/NrGO nanocomposites [44].

**Figure 4.** CVs of 10 μM Dopamine (DA) and uric acid (UA) mixture solution (1:1) on different electrodes.

*3.4. Optimization of Voltammetrical Parameters*

#### 3.4.1. Effect of pH

As known to all, the voltammetric responses of DA and UA highly depend on the solution pH. Therefore, it's worthwhile to optimize pH. The dependences of pH on the anodic peak currents of DA and UA are shown in Figure 5A. In the pH range of 2.58 to 3.93, the *i*pa(DA) gradually increased with the increase of pH, then decreased slowly as the pH rose to 7.01, and suddenly decreased when the pH exceeded 7.01. Obviously, the maximal *i*pa(DA) was achieved at pH 3.93. As for UA, the *i*pa(UA) show a downward trend, with pH varying from 2.58 to 8.52. To ensure the highest possible anodic peak current for DA and UA, pH 3.93 was selected for the following experiments. Moreover, the anodic peak potentials of DA and UA linearly decreased as pH was rising (Figure 5B). The linear relationships of *<sup>E</sup>*pa versus pH can be expressed as *E*pa(DA) = −0.0685 pH + 0.839 (R<sup>2</sup> = 0.974) and *E*pa(UA) = −0.0639 pH + 0.679 (R<sup>2</sup> = 0.976), respectively. Their slopes (68.5 pH/mV and 63.9 pH/mV) are close to 59 mV/pH, demonstrating the equal numbers of electron (e−) and protons (H+) involved in their electrooxidation processes [16]. As reported, the oxidation of DA and UA are two electron transfer processes [45]. Hence, the electrooxidation of DA and UA involves two electrons (2e−) and two protons (2H+). The electrochemical oxidation process of DA and UA at the MnO2NFs/NrGO/GCE are illustrated in Scheme 1.

**Figure 5.** (**A**) Dependence of pH on the anodic peak currents of DA and UA. (**B**) Linear plots of anodic peak potentials of DA and UA against pH (*n* = 3).

**Scheme 1.** Scheme diagram of the electrochemical oxidation of DA and UA at the MnO2NFs/NrGO/GCE.

#### 3.4.2. Effect of Scanning Rate

In order to give a deep insight into the oxidation of DA and UA, CVs of 10 μM DA and UA were performed at various scanning rates (Figure 6A). As the scanning rates increased, their anodic peaks shifted positively while the cathodic peaks shifted in the reverse direction. Furthermore, their response peak currents increased with the potentials scanning speeding up. It is noteworthy that the background currents also enhanced synchronously. To pursue high-signal to noise (S/N), 0.1 Vs−<sup>1</sup> was recommended as the optimal scanning rate. As shown in Figure 6B, the redox peak currents of DA were proportional to the square root of the scanning rate (*v*1/2), suggesting the electrooxidation of DA is a diffusion-controlled process. There was also a good relationship between the anodic peak currents of UA and the square root of the scanning rate (Figure 6C), indicating a diffusion-limited electrode process for UA oxidation.

**Figure 6.** (**A**) CVs of 10 μM DA and UA (1:1) at different scanning rates. (**B**) Linear relationship between the redox peak currents of DA and square root of scanning rates (*n* = 3); ( **C**) Linear relationship between the anodic peak currents of UA and square root of scanning rates (*n* = 3).

#### 3.4.3. Influence of Accumulation Parameters

Accumulation can e ffectively boost the response peak currents of target species, so the influence of accumulation potential and time were also investigated. The anodic peak currents of DA and UA sharply increased with the accumulation potentials shifting from −0.4 V to −0.3 V, then they gradually decreased with a further increase of the accumulation potential (Figure 7A). The highest anodic peak currents of DA and UA were achieved at −0.3 V, so −0.3 V was chosen as the optimal accumulation potential. As presented in Figure 7B, their anodic peak currents gradually enhanced during the first 150 s, then decreased with the prolonging of the accumulation time. Therefore, accumulation was performed at −0.3 V for 150 s in the following experiments.

**Figure 7.** Effect of accumulation potential ( **A**) and time (**B**) on the anodic peak currents of DA and UA.

#### *3.5. Individual and Simultaneous Determination of DA and UA*

For individual detection of DA and UA at the MnO2NFs/NrGO/GCE, LSVs were measured in the potential range of 0–0.8 V in 0.1 M PBS (pH 3.93). In this case, only the concentrations of the target substance were varied, while the concentrations of the other substance were kept unchanged. As illustrated in Figure 8A,B, there was a good linear relationship between the *i*pa(DA) and DA concentrations ranging from 0.4 μM to 10 μM. The linear equation is *i*pa(DA) = 1.3090 *C*DA − 0.1953 (R<sup>2</sup> = 0.989). However, the *i*pa(DA) were positively proportional to the Napierian logarithm of DA concentrations (ln *C*DA) in the concentration range of 10 μM–100 μM (Figure 8C,D). The linear regression equation can be expressed as *i*pa(DA) = 19.1371ln *C*DA − 32.3044 (R<sup>2</sup> = 0.993). The LOD was estimated as 0.054 μM. Regarding the individual determination of UA, the *i*pa(UA) were well-linear to the working concentrations, varying from 0.4 μM to 6.0 μM (Figure 9A,B), with the linear equation of *i*pa(UA) = 2.4934 *C*DA − 0.9302 (R<sup>2</sup> = 0.989). At the higher concentration region (6.0 μM to 100 μM), the *i*pa(UA) were positively correlated to the Napierian logarithm of UA concentrations (ln *C*UA). The corresponding linear equation is *i*pa(UA) = 44.3228ln*C*UA − 65.7789 (R<sup>2</sup> = 0.995). The LOD is 0.062 μM for the individual determination of UA. It is noteworthy that the addition of the target biomolecule does not have a notable interference on the electrochemical response signals of the other biomolecule. The results firmly imply that DA and UA can be sensitively and selectively detected at MnO2NFs/NrGO/GCE in the mixture of DA and UA.

**Figure 8.** Linear scanning voltammetry (LSVs) at the MnO2NFs/NrGO/GCE in 0.1 M PBS (pH 3.93) containing 1 μM UA and various DA concentrations from 0.4 μM to 10 μM (**A**) and from 10 μM to 100 μM (**C**); Linear plots of the *i*pa(DA) versus the DA concentrations varying from 0.4 μM to 10 μM (**B**) and from 10 μM to 100 μM (**D**).

The remarkable electrocatalytic activity of MnO2NFs/NrGO enables simultaneous detection of DA and UA using the LSV method (Figure 10). Two well-separated anodic peaks belonging to the electrooxidation of DA and UA were observed on LSV curves using MnO2NFs/NrGO/GCE. Furthermore, LSV responses were resolved into two peaks at 0.420 V and 0.554 V, respectively. The results demonstrate an excellent discriminability from the two biomolecules in mixture solutions. At the lower concentration region (0.02 μM–6.0 μM), the *i*pa(DA) and *i*pa(UA) enhanced linearly with their concentrations increasing (Figure 10A, B). The linear plots can be expressed as *i*pa(DA) = 3.0627*C*DA − 0.2848 (R<sup>2</sup> = 0.991), and *i*pa(UA) = 3.0627*C*UA − 0.2848 (R<sup>2</sup> = 0.990), respectively. However, the *i*pa(DA) and *i*pa(UA) are positively proportional to the Napierian logarithm of the DA and UA concentrations (ln*C*DA and ln*C*UA) at the higher concentration region from 6.0 μM to 100 μM (Figure 10C, D). The corresponding linear regression equations are *i*pa(DA) = 16.2222ln*C*DA − 12.4506 and *i*pa(UA) = 37.7032ln*C*UA − 48.2926, respectively. The correlation coefficient is 0.990 for both DA and UA. The LODs are calculated to be 0.036 μM and 0.029 μM for DA and UA, respectively. All of the results indicate that the proposed MnO2NFs/NrGO/GCE featured wider linear detection ranges and a lower LOD for the electrochemical oxidation of DA and UA. The analytical performances were compared to those in previous works (Table 1). Obviously, the sensing parameters of the proposed sensor are comparable to, or even better than, previously reported modified electrodes [7,46–54].

**Figure 9.** (**A**) LSVs at the MnO2NFs/NrGO/GCE in 0.1 M phosphate buffered solution (PBS) (pH 3.93) containing 1 μM DA and various UA concentrations from 0.4 μM to 6.0 μM (**A**) and from 6.0 μM to 100 μM (**C**) (*n* = 3); Linear plots of the *i*pa(UA) versus the UA concentrations varying from 0.4 μM to 6.0 μM (**B**) and from 6.0 μM to 100 μM (**D**) (*n* = 3).

**Figure 10.** (**A**) LSVs on theMnO2NFs/NrGO/GCE in 0.1M PBS (pH 3.93) containing various concentrations of DA and UA ranging from 0.2 μM to 6.0 μM (**A**) and from 6.0 μM to 100 μM (**C**); (**B**) Plots of the anodic peak currents as the function of DA and UA concentrations in the range of 0.2 μM –6.0 μM (*n* = 3); (**D**) Plots of the anodic peak currents as the function of DA and UA concentrations in the range of 6.0 μM–100 μM (*n* = 3).


**Table 1.** Comparison sensing performance between previous reports and the proposedMnO2NFs/NrGO/GCE for the simultaneous detection of DA and UA.

DPV: differential pulse voltammetry.

#### *3.6. Selectivity, Repeatability, and Reproducbility Assay*

a

Before real sample detection, the selectivity and repeatability, as well as reproducibility were also evaluated. To assess the anti-interfering ability of the proposed MnO2NFs/NrGO/GCE, the LSV responses of the DA and UA in the coexistence of common interfering substances (i.e., ascorbic acid, alanine, citric acid, glutamic acid, cysteine, and lysine) were compared. The relative errors (less than 5%) were accepted even in the presence of 100-fold the above interfering species, demonstrating good selectivity (Figure S1). It is noted that the 100-fold AA have no obvious interfering because of the well-separated peak potential between AA and DA (Δ*E*p = 0.260 V). To check the repeatability, eight successive determinations of 1 μM DA and UA (1:1) were also performed. The relative standard deviation (RSD) for DA and UA were 6.35% and 5.32%, respectively, indicating good repeatability. To examine electrode reproducibility, the anodic peak currents of 1 μM DA and UA were recorded at five MnO2NFs/NrGO/GCEs, which were prepared by similar procedure. The RSD of the anodic peaks were 6.02% and 4.70% for DA and UA respectively, showing that the electrode preparation have excellent reproducibility.

#### *3.7. Determination of DA and UA in Human Serum Samples*

To validate practicability, the concentrations of DA and UA in human serum samples were also detected on MnO2NFs/NrGO/GCE. The determination results were calculated from the calibration curves (Table 2). To further validate the precision of the proposed sensor, a series of known concentration solutions of DA and UA were spiked with the serum samples to figure out the recovery. The recoveries are 96.2–105.6% and 96.2–104.9% for DA and UA respectively, verifying that biological matrixes, like human serum, do not influence the simultaneous detection of DA and UA.

**Table 2.** Detection results of DA and UA in human serum samples using MnO2NFs/NrGO/GCE (*n* = 3).


a The human serum samples were detected at 10-fold dilution. b Not detected.

## **4. Conclusions**

In summary, this paper reported the facile synthesis of MnO2NFs/NrGO nanocomposites and their application on the simultaneous determination of DA and UA. The electrochemical measurements indicated that the MnO2NFs/NrGO composites possess a large, e ffective electroactive area and low-charge transfer resistance. MnO2NFs/NrGO/GCE showed superb catalytic capacity toward the electrooxidation of DA and UA, attributing to the synergistic e ffect from MnO2NFs and NrGO sheets. The anodic peak currents of DA and UA increased linearly with their concentrations varying from 0.2 μM to 6.0 μM. However, their anodic peak currents were highly correlated to the Napierian logarithm of their concentrations, ranging from 6.0 μM to 100 μM. The LODs were 0.036 μM and 0.029 μM for DA and UA, respectively. Furthermore, the proposed sensor successfully realized DA and UA detection in human serum samples with satisfactory recovery. Combining with prominent advantages such as facile preparation, high sensitivity, good selectivity, repeatability, and reproducibility, the proposed MnO2NFs/NrGO nanocomposites have become the most competitive candidates for the simultaneous determination of DA and UA in various real samples.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2079-4991/9/6/847/s1, Figure S1: The anodic peak currents of 1 μM DA and UA in the presence of 100-fold alanine (AL), glutamic acid (GA), ascorbic acid (AA), lysine (LY) and citric acid (CA).

**Author Contributions:** X.W., S.Y. conceived and designed the experiments; X.W., S.Y., and Z.C. performed the experiments; X.W., S.Y., Z.C., Y.Y., and Y.X. analyzed the data; Q.H., G.L., and J.L. contributed reagents/materials/analysis tools; X.W. and S.Y. wrote the original draft; G.L. revised the manuscript.

**Funding:** This work was supported by the Undergraduates' Innovation Experiment Program of Hunan Province (No. 2018649), Natural Science Foundation of Hunan Province (No. 2019JJ50127, 2018JJ3134), National Natural Science Foundation of China (No. 61703152), Scientific Research Foundation of Hunan Provincial Education Department (18A273, 18C0522), Project of Science and Technology Plan of Zhuzhou (201707201806).

**Acknowledgments:** We sincerely express our thanks to Zhuzhou Peoples Hospital for o ffering human serum samples.

**Conflicts of Interest:** The authors declare no conflicts of interest.
