All chemicals were of analytical grade and used without further purification. Citric acid, ethylenediamine, sodium borohydride (NaBH₄), Chloroauric acid tetrahydrate (HAuCl₄∙4H₂O), bovine serum albumin (BSA), 6-mercaptopurine monohydrate, NaH₂PO₄, Na₂HPO₄, NaCl, KCl, CaCl₂, NH₃∙H₂O, NaOH, HCl, MgSO₄, Zn(NO₃)₂, Ni(NO₃)₂, Hg(NO₃)₂, CuSO₄, Fe(NO₃)₃, Fe(NO₃)₂, Pb(NO₃)₂, CdCl₂, CrCl₃, Isolucine,, Uracil, Glucosuria, Aspartic acid, Tryptophan, Tyrosine, Lysine, Adenine, Cytosine, Cystine, Thiophenol, glutathione were purchased from Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). All reagents were used as-received without additional purification. All the reagent solutions were prepared by the water purified through a Millipore system with a resistance of 18.2 MΩ·cm.

Preparation for stock solution of 6-MP [25]. N₂ was bubbled through an aqueous solution of 0.1 M NaOH until saturation to remove dissolved oxygen. 86 mg 6-MP was added into 500 μL of the aforementioned solution. 4.5 mL boiled ultrapure deionized water was added after the solid had dissolved completely. Then 100 mM 6-MP stock solution was prepared. The stock solution was diluted 5 times and stored as 100 μL per batch independent solution. The as-prepared solutions were stored in a −20˚C refrigerator. 1.90 mL boiled ultrapure deionized water per batch was added into the thawed solution before use.

The fluorescence spectra were recorded on an RF-6000 spectrofluorometer (Shimadzu, Japan) with 1 cm quartz cells. The light source used in the spectrofluorometer was a 150 W Xe arc lamp (Ushio Inc, Japan), and the emitted power density was approximately 20 - 32 mW∙cm⁻² in its wavelength range. The slits for excitation and emission monochromators width were both 5 nm. The transmission electron microscopy (TEM) images were recorded using a JEOL 2010 transmission electron microscope. Fourier transform infrared (FT-IR) spectra were acquired from a Thermo Fisher Nicolet iS10 FT-IR spectrometer.

The CDots were prepared by a simple one-pot hydrothermal method with minor modifications [41]. Typically, 4.204 g of citric acid and 1.34 mL of 1,2-ethylenediamine were mixed and dissolved into 40 mL of water to form a clear solution. The mixture was put into a 50 mL poly(tetrafluoroethylene) Teflon-lined autoclave tube and the solution was sealed and treated at 200˚C for 4 h. The resulting brown solution was cooled to room temperature naturally and filtered through 0.45 μM Suporfilters to remove the large or agglomerated particles. Then the CDots solution was purified by dialyzing against pure water using a membrane (M_W = 3.5 kDa) for 12 h and then storing at 4˚C for further use.

All the glasswares were first washed with aqua regia and then rinsed with ultrapure water, several times before use. AuNCs were synthesized and purified according to the literature [42]. In a typical experiment, HAuCl₄ solution (10 mL, 10 mM) was added to the BSA solution (10 mL, 50 mg/mL) under vigorous stirring. After 5 min, suitable NaOH (1 M) solution was introduced to the mixture to adjust the pH to 11.0 and then the mixture was kept under stirring for 12 h at 37˚C. The solution color changed from pale yellow to brown. Then the resulting brown solution was purified by dialyzing against pure water using a membrane (M_W = 12 kDa) for 24 h and then storing at 4˚C for further use.

The nanohybrid system was prepared by the following procedure. First, in order to adjust the fluorescent intensity ratio of CDots and AuNCs to be 1:1, suitable CDots solution was added to 10 mL AuNCs solution. The mixture was kept under vigorous stirring for surface hybridization through reaction and interaction. The fluorescence spectra were recorded from 375 to 775 nm under excitation at 365 nm. This nanohybrid probe shows a good physical stability [38].

To evaluate the sensitivity of the RF probe for the 6-MP, in 1.5 mL centrifuge tube, 100 μL nanohybrid solution and 800 μL PB buffer (pH 8.0, 50 mM) were added, then 100 μL various concentrations of 6-MP were also added. The mixture was stirred. Subsequently, the fluorescence spectra were recorded from 375 to 775 nm under excitation at 365 nm.

To further study the specificity of the sensing system towards detecting 6-MP, the interferences of co-existing foreign substances were tested under the above-selected conditions, spiked with different substances of a known concentration individually.

A schematic illustration of the ratiometric fluorescence bioassay platform for the detection of 6-MP based on the nanohybrid of fluorescence carbon dots and gold nanoclusters was shown in Figure 1. In brief, we first prepared the CDots and AuNCs. In the presence of 6-MP, the sulfhydryl group of 6-MP was preferentially bound with AuNCs through thiol-Au bond [23] and caused the fluorescence signal of AuNCs can be effectively quenched, whereas the fluorescence intensity of CDots is unaffected, which can serve as a better reference signal for 6-MP assay. By combining the two fluorescence behaviors, the nanohybrid represented an ideal platform for the ratiometric determination of 6-MP, with the AuNCs serving as the 6-MP recognition component and the CDots acting as the reference fluorophore.

The properties of the as-prepared Au NCs and CDots were investigated by fluorescence spectroscopy, TEM, and FT-IR spectra. Figure 2 shows the maximum emission center of AuNCs at 656 nm (Figure 2(a)), and the CDots emission band at 450 nm (Figure 2(b)). The fluorescence spectra of Au NCs and CDots were further recorded every 5 min for 1 h under ultraviolet irradiation at 365 nm, and the fluorescence intensities exhibit no distinct change, implying that both the AuNCs and CDots exhibit good stability against photobleaching in aqueous solutions. The nanohybrid solution emits two emission bands at 450 and 656 nm.

The morphologies and the sizes of the two components were characterized by TEM as shown in Figure 3 The diameter of CDots was estimated to be ~4 nm with a good dispersity (Figure 3(a)). The AuNCs were readily dispersed in water and possessed a good monodispersity with a particle size of about ~3 nm (Figure 3(b)). Additionally, CDots, AuNCs and CDots-AuNCs hybrid were characterized by FT-IR as shown in Figure 4 CDots have many characteristic peaks, such as −OH and N−H (3100 - 3500 cm⁻¹), C−H (2930 cm⁻¹), C=ONR

(1640 cm⁻¹), C−N (1290 cm⁻¹), and C−O−C (1084 cm⁻¹). The AuNCs showed stretching vibrations of O−H at 3457 cm⁻¹, C=O stretching vibrations of carboxyl groups at 1661 and 1524 cm⁻¹. For the CDots-AuNCs hybrid, three absorption bands at 3457 cm⁻¹, 1661 and 1524 cm⁻¹ were assigned to AuNCs, the characteristic peak at 1290 cm⁻¹ could be assigned to the stretching vibration of the C−N groups and 1084 cm⁻¹ was related to C−O stretching vibration from the CDots [39]. The results revealed that the CDots-AuNCs hybrid showed signals of both CDots and AuNCs.

Figure 5 represents the fluorescence detection of 6-MP by the nanohybrid sensor. When the I₆₅₆/I₄₅₀ intensity ratio was adjusted to 1:1, as shown in Figure 3, the emission at 656 nm from the AuNCs gradually decreased upon the addition of 6-MP, but the fluorescence at 450 nm from the CDots was unchanged with the increase of 6-MP.

Figure 6 shows that the fluorescence intensity ratio, I₆₅₆/I₄₅₀ of the nanohybrid system decreased proportionately with increasing amounts of 6-MP, and a relationship can be set up between I₆₅₆/I₄₅₀ and the 6-MP concentration. The linear curve equation was I₆₅₆/I₄₅₀ = 0.992 – 0.023 × [6-MP] with a correlation coefficient R² of 0.998, indicating a good linear correlation between I₆₅₆/I₄₅₀ and the concentration of 6 MP. The detection limit for 6-MP was determined to be 54 nM based on the definition of 3 times deviation of the blank signal (3σ). The decline of the fluorescence ratio (I₆₅₆/I₄₅₀) of the nanohybrid probe was attributed to the quenching of the fluorescence of the AuNCs by 6-MP. When the 6-MP was added, 6-MP was adsorbed on the BSA-AuNCs because the thiol-Au bond

can form. The formation of such bonds removed the protection effect of the BSA and neutralized the surface charge of the AuNCs. Without this protection effect of the BSA, the detached AuNCs coated with thiols, could aggregate, because there would be an increase in the van der Waals attraction forces between them [23].

In this study, precision was measured by intra-day and inter-day variability and expressed as the RSD, which was calculated from three replicate determinations of reference standard solution concentration of 6-MP at three concentrations (5.5, 22.5, 30.2 μM) within one day (intra-day precision) and three replicates over three days (inter-day precision). The RSDs for the intra-day and inter-day precision were 2.3% - 5.4% and 2.5% - 6.8%, respectively.

The interferences of co-existing foreign substances were tested under the above-selected conditions, spiked with different substances of a known concentration individually. An error of ±5% in the relative fluorescence intensity was considered tolerable. The results are summarized in Table 2. It can be seen from Table 2 that most of the tested substances including those of metal ions, some protein-forming amino acids, uracil, dextrose and glucosuria scarcely interfere with the determination at high tolerance levels. It is concluded that the method is free from many interferences of foreign substances. However, the thiol-containing compounds, cystine, thiophenol, glutathione, which were similar in structure to 6-MP, could be tolerated only at relative low levels. This suggested that the sensor responded differently with different thiol compounds. These compounds often have other different functional groups attached. Such as Yu et al. constructed a FRET assembly by using gold nanoclusters and carbon dots and their application as a ratiometric probe for cysteine [40]. Though AuNCs have been successfully used in the determination of various metal ions such as Hg²⁺, Fe³⁺, Cu²⁺, Pb²⁺ and Cr³⁺ [34] [36] [37] [38] [43] [44] [45] [46], these metal ions did not affect the analytical results, the reason was that all these metal ions did not exist at alkaline solutions (pH 8.0), they could form the corresponding precipitation of hydroxide with OH^–. Thus, satisfactory analytical selectivity may be expected for different bio-thiols, and hence, the novel method has satisfactory selectivity for the analysis of 6-MP.

The proposed method was applied to the determination of 6-MP in spiked human serum. The serum samples, obtained from healthy volunteers, were spiked with 6-MP at different concentrations, and treated as the recommended procedure. The concentrations of 6-MP were calculated from the calibration graph. The results obtained for the determination of 6-MP in spiked human serum are given in Table 3. The satisfactory recoveries obtained with such a simple sample

^aMean values of 11 determinations. ^bRelative standard deviation.

procedure are in the range of 99.6% - 101.6%.