Preparation method of gold-plated SERS (Surface Enhanced Raman Scattering) probe with Ag NCs modified on surface for detecting free basic group and application of gold-plated SERS probe in real-time monitoring of quality of frozen tuna

Abstract

The invention discloses a preparation method of a gold-plated SERS (Surface Enhanced Raman Scattering) probe for detecting free basic groups and modifying Ag NCs on the surface and application of the gold-plated SERS probe to real-time monitoring of the quality of frozen tunas, and is characterized in that a gold-plated needle is cleaned, soaked in a sodium citrate solution and then soaked in an AgNCs solution to obtain the gold-plated SERS probe; the application for monitoring the quality of the frozen tuna in real time comprises the following steps: puncturing the gold-plated SERS probe into a frozen fish sample to be detected to obtain Raman spectrum data, preprocessing the Raman spectrum data, and respectively inputting the preprocessed Raman spectrum data into PLS models of five substances, namely adenine, guanine, cytosine, thymine and uracil; outputting concentration predicted values corresponding to the five basic groups in the frozen sample to be detected; and calculating the concentration values of adenine, guanine, cytosine, thymine and uracil in the frozen fish sample to be detected according to the linear relationship between the predicted value and the true value. The method has the advantages of simplicity, rapidness, and high sensitivity and accuracy.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of food detection, and particularly relates to a preparation method of a gold-plated SERS probe of surface-modified Ag NCs for detecting free bases and application of the gold-plated SERS probe in real-time monitoring of the quality of frozen tuna. BACKGROUND

[0002] Generally, the average content of nucleic acid in the nucleus of cells of the same organ of the same species is relatively constant and is not easily affected by subjective factors and environmental factors, and is a stable biochemical index. For frozen aquatic products which are difficult to distinguish from appearance, a convenient detection method for nucleic acid degradation related products can be established for quality change evaluation during the frozen storage period.

[0003] After the death of animals, the nucleic acids (DNA and RNA) of their tissues and cells are gradually degraded by nucleases and nucleotidases into nucleic acid fragments and free bases, including adenine (A), guanine (G), cytosine (C), thymine (T), uracil (U) and ribose. The common quantitative detection methods for free bases at present mainly include high performance liquid chromatography (HPLC), fluorescence analysis, electrochemical method, mass spectrometry and colorimetric method. However, these methods still have certain limitations in the quantitative analysis of purines and pyrimidines in animal foods due to the disadvantages of complicated processing, long time consumption, expensive instruments and low sensitivity. Surface-enhanced Raman scattering (SERS) is a light scattering technology based on the interaction between light and chemical bonds in materials. By analyzing the wavelength and intensity changes of the non-elastic scattering light generated by the measured sample under laser irradiation, the unique "fingerprint" spectrum of the molecule can be distinguished, and SERS has the advantages of high specificity, high sensitivity and fast analysis speed.

[0004] In recent years, research on Raman spectroscopy data analysis has been widely reported, including spectral preprocessing techniques such as first derivative (D1st), wavelet transform (WT), and standard normal variable transformation (SNV); and feature variable selection techniques such as Monte Carlo-no-information variable elimination (MC-UVE), competitive adaptive reweighted sampling (CARS), and bidirectional interval partial least squares (biPLS). Hybrid spectral preprocessing methods mean using multiple different spectral preprocessing methods simultaneously in spectral analysis to optimize the raw spectral data, thereby improving the quality and interpretability of the spectrum, and ultimately enhancing the predictive performance of the model. The bidirectional interval partial least squares (biPLS) method is an improved interval selection algorithm based on the traditional interval partial least squares (iPLS). It adopts an "out-of-range" strategy when selecting intervals and uses joint intervals to build the model. By searching for the optimal joint interval for modeling, it can improve the stability and accuracy of the model. Currently, no SERS-based methods for detecting the content of free bases in fish meat have been found. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a simple, rapid, sensitive and accurate method for preparing a gold-plated SERS probe for detecting free bases by surface modification Ag NCs and its application in real-time monitoring of the quality of frozen tuna.

[0006] The technical solution adopted by this invention to solve the above-mentioned technical problems is as follows: a method for preparing a gold-plated SERS probe with surface-modified AgNCs for detecting free bases, comprising the following steps: Step 1: Synthesize Ag NCs via a hydrothermal method to obtain an Ag NCs solution; Step 2: After cleaning the gold-plated needle, immerse it in sodium citrate solution, and then immerse it in AgNCs solution to obtain a gold-plated SERS probe with AgNCs surface modification.

[0007] Further, step 1 is as follows: Mix 70-80 mM CTAB solution, 8-12 mM silver ammonia solution and 1-2 mM glucose solution in a volume ratio of 1:1:2 and stir until homogeneous. Heat the mixture at 100-150°C for 6-10 hours. After cooling, remove the mixture and centrifuge and wash it with deionized water to obtain AgNCs solution.

[0008] Further, step 2 is as follows: After ultrasonic treatment in acetone and ethanol to remove surface impurities, the cleaned needle is immersed in a sodium citrate solution with a concentration of 0.5-1.5wt% for 10-14 hours, then cleaned with ethanol. The gold-plated needle is then placed in an AgNCs solution and soaked for 10-14 hours. After washing with ultrapure water and ethanol, it is dried in a nitrogen atmosphere to obtain a gold-plated SERS probe with AgNCs surface modification.

[0009] This invention also provides the application of the gold-plated SERS probe prepared by the above method for real-time monitoring of the quality of frozen tuna, wherein the free base is at least one of adenine, guanine, cytosine, thymine and uracil.

[0010] Furthermore, the method for detecting the content of free bases in frozen fish includes the following steps: Step 1: Insert the gold-plated SERS probe 1.5-2 cm into the frozen fish sample to be tested, leave it for 30 seconds, then remove it and perform Raman spectroscopy detection, collecting samples from 400-1800 cm⁻¹. -1 Raman spectral data of the range; Step 2: After preprocessing the Raman spectral data collected in Step 1, input them into the PLS models of five substances: adenine, guanine, cytosine, thymine, and uracil, and output the predicted concentration values ​​x of adenine, guanine, cytosine, thymine, and uracil in the frozen sample to be tested. Step 3: Substitute the five predicted concentration values ​​x obtained in Step 2 into the following formula for the linear relationship between predicted and true values ​​to calculate the concentration values ​​y of adenine, guanine, cytosine, thymine and uracil in the frozen fish sample to be tested. The linear relationship between the predicted and true values ​​of adenine is y = 1.0140x + 0.2175. The linear relationship between the predicted and actual values ​​of guanine is y = 0.9956x + 0.0533; The linear relationship between the predicted and actual values ​​of thymine is y = 1.0010x - 0.2818; The linear relationship between the predicted and actual values ​​of cytosine is y = 1.0111x + 0.1717; The linear relationship between the predicted and actual values ​​of uracil is y = 1.1017x + 0.0982.

[0011] Furthermore, in step 2, for the adenine PLS model, the Raman spectral data preprocessing method is to first perform first derivative preprocessing, select 19 smoothing points, and then perform wavelet transform, select db1 as the wavelet basis function, and set the number of decomposition layers to 3; the number of intervals for the biPLS model is set to 19.

[0012] Furthermore, in step 2, for the guanine PLS model, the Raman spectral data preprocessing method is to first perform wavelet transform, select db2 as the wavelet basis function, and set the number of decomposition layers to 4; then perform first derivative preprocessing, select the number of smoothing points to 23; and then set the number of intervals for the biPLS model to 15.

[0013] Furthermore, in step 2, for the thymine PLS model, the Raman spectral data preprocessing method is to first perform first derivative preprocessing, select 17 smoothing points, then perform wavelet transform, select db3 as the wavelet basis function, and set the number of decomposition layers to 3; the number of intervals for the biPLS model is set to 18.

[0014] Furthermore, in step 2, for the cytosine PLS model, the Raman spectral data preprocessing method is to first perform first derivative preprocessing, select 7 smoothing points, then perform wavelet transform, select db4 as the wavelet basis function, and set the number of decomposition layers to 4; the number of intervals for the biPLS model is set to 19.

[0015] Furthermore, in step 2, for the uracil PLS model, the Raman spectral data preprocessing method is to first perform wavelet transform, select db1 as the wavelet basis function, and set the number of decomposition layers to 4; then perform first derivative preprocessing, select the number of smoothing points to 19; and then set the number of intervals of the biPLS model to 14.

[0016] Compared with the prior art, the advantages of the present invention are as follows: 1. Probe preparation: High sensitivity and stability: Silver nanoclusters (Ag NCs) synthesized via a hydrothermal method possess uniform size and abundant "hot spots." Modifying these nanoclusters onto the surface of a gold-plated needle cleverly combines the excellent biocompatibility and chemical stability of the gold substrate with the superior surface-enhanced Raman scattering (SERS) activity of the silver nanoclusters, constructing a high-performance SERS probe. This probe significantly enhances the Raman signal and improves detection sensitivity. The preparation method is controllable and reproducible: the specific synthesis parameters (such as concentration, ratio, temperature and time) and surface modification steps (such as sodium citrate activation and soaking time) have been further clarified, which standardizes the preparation process, ensures the consistency and reproducibility of probe performance, and is conducive to large-scale production and application.

[0017] 2. Applications in detecting free base content in frozen fish: It enables rapid, in-situ, and non-destructive testing: sampling can be completed simply by inserting the probe into the sample and allowing it to remain there briefly. There is no need for complex pretreatment such as extraction and purification of the sample, which greatly simplifies the operation process, shortens the detection time, and enables in-situ, minimally invasive, or even non-destructive testing of frozen fish samples. It is suitable for on-site or online analysis.

[0018] Highly targeted and accurate quantitative analysis models were established: optimized PLS quantitative models were constructed for five different free bases (adenine, guanine, cytosine, thymine, and uracil). Each model defined a specific combination of spectral preprocessing (first derivative, wavelet transform) and a characteristic wavelength selection method (biPLS), effectively eliminating background interference, improving the signal-to-noise ratio, and extracting the most relevant spectral information to the target analyte, thereby significantly improving the model's prediction accuracy and robustness. It provides a reliable quantitative correction relationship: it clearly provides a validated linear correction formula between the predicted and actual values ​​of five bases, so that the predicted values ​​output by the model can be accurately converted into actual concentration values, ensuring the accuracy and reliability of the final test results, and providing accurate quantitative indicators for the evaluation of the freshness or quality of frozen fish.

[0019] 3. Innovative integration of SERS probe technology with chemometrics models: It not only provides high-performance detection hardware (probes), but also comes with a series of optimized models for specific application scenarios (multiple free bases in frozen fish), forming a complete and efficient analytical detection solution.

[0020] In summary, this invention presents a method for preparing a gold-plated SERS probe with surface-modified Ag NCs for detecting free bases and its application in real-time monitoring of frozen tuna quality. Using a gold-plated probe substrate material based on surface-modified Ag NCs, combined with optimized spectral preprocessing methods and a PLS mathematical model, a highly efficient and sensitive quantitative analysis method for five free bases (adenine, guanine, thymine, cytosine, and uracil) in tuna meat was established. This method provides a rapid, sensitive, and multi-indicator simultaneous detection technique for freshness assessment, quality monitoring, and shelf-life prediction of frozen aquatic products (taking fish as an example). It has significant application potential and promotional value in food safety monitoring, cold chain logistics quality control, and related scientific research fields. Attached Figure Description

[0021] Figure 1 shows the morphology of the gold-plated SERS probe with Ag NCs surface modification. a is the overall view, b is a magnified view of a part, and c is the morphology under a scanning electron microscope. Figure 2 shows the Raman spectra of five purines and pyrimidines as single and mixed standards, where a is the SERS intensity value of adenine at different Raman shifts, b is the SERS intensity value of guanine at different Raman shifts, c is the SERS intensity value of thymine at different Raman shifts, d is the SERS intensity value of cytosine at different Raman shifts, e is the SERS intensity value of uracil at different Raman shifts, and f is the SERS intensity value of the mixture of the five nucleic acid bases at different Raman shifts. Figure 3Raman images of multiple mixed samples superimposed; Figure 4 The PLS calibration model prediction results for five substances based on different smoothing points of the first derivative preprocessing method D1st are shown. Among them, a is adenine, b is guanine, c is thymine, d is cytosine, and e is uracil. Figure 5 The prediction results of the PLS calibration model for five substances based on different separation layers and wavelet basis functions of wavelet transform (WT) are shown. Among them, a is adenine, b is guanine, c is thymine, d is cytosine, and e is uracil. Figure 6 shows the prediction performance of the PLS calibration model based on different combinations of preprocessing methods, where a is RMSECV and b is R 2 CV ; Figure 7 The prediction performance of the biPLS calibration model based on different interval numbers is given, where a is RMSECV and b is Ri. 2 CV; Figure 8 The relationship between the true and predicted values ​​obtained from the PLS calibration model for different substances is shown, where a is adenine, b is guanine, c is thymine, d is cytosine, and e is uracil. Figure 9 shows the HPLC quantitative analysis results of five nucleic acid bases, where 1, 2, 3, 4, and 5 are Adenine, Guanine, Thymine, Cytosine, and Uracyl, respectively. Figure 10 Gel electrophoresis images of total DNA from tuna meat stored for different times; Figure 11 (a) shows the correlation analysis between TBA content and nucleic acid bases detected by SERS, and (b) shows the correlation analysis between TBA content and ATGCU content detected by HPLC. Detailed Implementation

[0022] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0023] Reagents: Guanine (G, 99% purity), Adenine (A, ≥99.5% purity), Cytosine (C, 99% purity), Thymine (T, 99% purity), Uracil (U, ≥99.5% purity), Silver nitrate (AgNO3, 99.9% purity), Cetyltrimethylammonium bromide (CTAB, 98% purity), and glucose (C6H2O). 12 All reagents (O6) were purchased from Shanghai Aladdin Biochemical Technology Co., Ltd. (Shandong, China). Sodium citrate (C6H5Na3O7) was from Sigma-Aladdin. Sodium hydroxide (NaOH), ammonia (NH3-H2O), and sodium chloride (NaCl) were from Sinopharm Chemical Reagent Co., Ltd. (Shenyang, China). Carrageenan was purchased from Shanghai Maclean's Biochemical Co., Ltd. Needles and sucrose were purchased from local supermarkets. All reagents were of analytical grade, and solutions were prepared using Milli-Q water (18.2 MΩ-cm) or ethanol. Yellowfin tuna (caught in September 2024 by the tuna purse seine vessel "MARIBO 61" in the Pacific Ocean, weighing 20±2 kg, frozen at -60℃) was purchased from Zhoushan Yinghai Ocean Fishery Co., Ltd. The -60℃ frozen yellowfin tuna was transferred to the -18℃ constant temperature freezer of Ningbo Jinri Food Co., Ltd. in December 2024 for a 120-day tuna storage experiment.

[0024] Instrumentation: The morphology was examined using a field emission scanning electron microscope (SEM, SU-70, Hitachi). Surface-enhanced Raman spectroscopy (SERS) was performed using a compact Raman spectrometer (BWS 415, B&W Tek) equipped with a 532 nm semiconductor laser. The integration time was 10 seconds, and the laser power was 20 mW or 30 mW.

[0025] Specific Example 1: A method for preparing a gold-plated SERS probe with surface-modified Ag NCs for detecting free bases, comprising the following steps: Step 1: Synthesis of Ag NCs via hydrothermal method: 75 mM CTAB solution, 10 mM silver ammonia (AgOH·2NH3) solution, and 1.5 mM glucose solution were added to a beaker in a volume ratio of 1:1:2 and stirred until homogeneous. The resulting mixture was placed in the lining of a reaction vessel and heated in an oven at 120°C for 8 hours. After natural cooling, the mixture was removed and centrifuged with deionized water (6000 r / min for 10 min) to obtain the AgNCs solution, which was then properly stored at 4°C for later use.

[0026] Step 2: Pretreatment of gold-plated needles and preparation of gold-plated probe substrates with surface-modified Ag NCs.

[0027] The gold-plated needle was sequentially placed in acetone and ethanol, and ultrasonically treated for 1 hour each to remove surface impurities. After cleaning, the needle was immersed in a 0.5-1.5% sodium citrate solution for 12 hours, followed by two rinses with ethanol. Then, the needle was immersed in an AgNCs solution for 12 hours, washed sequentially with ultrapure water and ethanol, and finally dried under nitrogen to obtain the gold-plated SERS probe with AgNCs surface modification. Figure 1 As can be seen from the ac, Ag NCs were successfully attached to the gold-plated SERS probe.

[0028] Specific Example 2: Raman Spectroscopic Analysis of Free Bases (A, T, G, C, U) 1. Raman spectroscopy analysis of single free bases (A, T, G, C, U) Using a gold-plated SERS probe with surface-modified Ag NCs prepared in Specific Example 1, five free bases in a simulated fish matrix—adenine (A), thymine (T), guanine (G), cytosine (C), and uracil (U)—were detected.

[0029] Prepare separate standard solutions of free adenine, guanine, cytosine, thymine, and guanine; the concentration of each single standard solution is 10. -2 M; Take 5 mL of each single standard solution and stir with the gold-plated SERS probe for 5 minutes. After standing and drying, excite with a 532 nm Raman laser at a power of 30 mW for 10 s to measure the Raman intensity of the SERS probe. Collect the 500-1800 cm⁻¹ values ​​of adenine, guanine, cytosine, thymine, and guanine standard solutions. -1 The SERS spectra of the range were obtained, and the SERS intensity values ​​of adenine, guanine, cytosine, thymine, and guanine at different Raman shifts were obtained, such as... Figure 2 As shown.

[0030] like Figure 2 As shown in (a), the characteristic Raman band of adenine is located at 733 cm⁻¹. -1 1333cm -1 and 1455cm -1 The Raman signal responses generated are caused by the whole molecular ring breathing mode, the stretching vibration mode of C=C and CN, and the bending vibration mode of CH. like Figure 2 As shown in (b), the characteristic Raman band of guanine is located at 655 cm⁻¹. -1 960cm -1 1269cm -1 and 1538 cm-1 The Raman signal responses generated are the Raman response signals generated by the whole molecular ring breathing mode, the bending vibration mode of CH, and the stretching vibration mode of CN. like Figure 2 As shown in (c), the characteristic Raman band of thymine is located at 760 cm⁻¹. -1 1366cm -1 1444cm -1 and 1497cm -1 Raman signal response caused by the CH rocking vibration mode of the whole molecular ring breathing mode; like Figure 2 As shown in (d), the characteristic Raman bands of cytosine include those generated by the NH bending vibration mode at 795 cm⁻¹. -1 1306cm -1 and 1365cm -1 The Raman response signal induced at the location; like Figure 2 As shown in (e), the characteristic Raman band of uracil is located at 795 cm⁻¹. -1 1050cm -1 1274cm -1 1386cm -1 and 1641cm -1 The Raman response signals are generated by the planar rocking vibration mode and the C=O stretching vibration mode of CH, respectively.

[0031] This study demonstrates that gold-plated probe substrates based on surface-modified Ag NCs exhibit significant characteristic Raman signal enhancement effects on all five bases in a simulated food matrix using surface-enhanced Raman spectroscopy (SERS), enabling quantitative and qualitative analysis.

[0032] 2. Raman spectroscopy analysis of mixed free bases (A, T, G, C, U) SERS detection was performed on a mixture of the five nucleotide components: 25.00 mg each of cytosine, uracil, thymine, guanine, and adenine were accurately weighed into a 25 mL volumetric flask, dissolved and diluted to volume with 0.20 mol / L NaOH solution to prepare a 1000 mg / L mixed standard stock solution. This solution was then diluted to obtain a concentration of 10... -2 For the mixed standard solution M, 5 mL of the mixed standard solution was added to the gold-plated SERS probe and stirred for 5 minutes. After standing and drying, the probe was excited with a 532 nm Raman laser at a power of 30 mW for 10 s. The Raman intensity of the SERS probe was measured, and samples of adenine, guanine, cytosine, thymine, and guanine standard solutions were collected from 400 to 1800 cm⁻¹. -1The SERS spectra of adenine, guanine, cytosine, thymine, and guanine were obtained at different Raman shifts, and the results are as follows: Figure 2 As shown in (f), at 655 cm -1 731 cm -1 784cm -1 1078 cm -1 1182 cm -1 1333 cm -1 1376 cm -1 1429 cm -1 1481 cm -1 and 1646 cm -1 Raman characteristic signal peaks are generated at locations such as [locations not specified]. Figure 3 As shown, the mixing of the five nucleotide components resulted in some characteristic peaks showing enhanced overlap and others showing weakened overlap. Therefore, relying solely on univariate analysis methods may not meet the requirements for accurate quantitative and qualitative analysis of the target object.

[0033] Specific Implementation Example 3: Construction of the PLS Calibration Model The complexity of nucleic acid degradation products (including ribose, purines, and pyrimidines) in fish tissue cells introduces a lot of redundant information that affects quantitative analysis results. Therefore, this study uses a mixed spectral preprocessing method and variable selection strategy to establish a partial least squares (PLS) calibration model to determine the content of A, T, G, U, and C in fish meat. First, 21 samples were prepared for simulation, and their Raman spectral data were collected. All samples were reasonably divided into calibration and prediction sets. The specific impact of the mixed spectral preprocessing method and its integration strategy on the predictive performance of the partial least squares (PLS) calibration model was discussed in detail. By comparing the model performance under different preprocessing methods and integration strategies, the impact of the biPLS variable selection strategy on the performance of the PLS calibration model was evaluated. Based on the characteristic variables determined in the above optimization process, a PLS calibration model was constructed, which can rapidly and accurately quantify purines and pyrimidines in fish meat using Raman spectral data. The specific steps are as follows: Step 1: Preparation of food matrix simulating the environment of fish meat Because the target for detection is free bases, water activity was used as the indicator for preparing the simulated fish meat matrix gel. Carrageenan and sucrose (water:carrageenan:ribose = 100g:10g:5g, Aw = 0.99) were dissolved in warm water (50-60℃). Then, different amounts of A, T, G, C, and U (Table 1) were added to these solutions, and the mixture was shaken to aid dissolution. The solutions were then poured into square molds and allowed to gel, thus obtaining model fish meat gels containing purines and pyrimidines. For free base detection using the simulated fish meat gel, the surface-modified Ag NCs prepared in Example 1 were inserted 1.5-2 cm into the simulated fish meat gel, left for 30 seconds, and then withdrawn for Raman spectroscopy. To ensure the accuracy and reliability of the measurement data, three parallel sets were set up for each sample, and each gel sample underwent three independent needle penetration measurements at different orientations.

[0034] Table 1. Reference concentrations of nucleic acid bases in food models

[0035] Step 2: Raman spectroscopy acquisition In this experiment, a 532 nm Raman laser was used as the excitation source, with a power set to 20 mW. Raman spectral data of 21 samples were collected as independent variables, and sample concentration was used as the dependent variable to establish a spectral data matrix. The Raman spectral data acquisition time for all samples was 10 seconds. To ensure the accuracy and reliability of the experimental data, multiple measurements were taken to reduce errors. The Raman spectral data of each sample was independently recorded 5 times. Statistical analysis was performed on the obtained data, and the results are expressed as mean ± standard deviation.

[0036] Step 3: Spectral data preprocessing and optimization During Raman spectroscopy data acquisition, various factors can interfere with the accuracy of the concentrations and characteristic band intensities of purines (including adenine A and guanine G) and pyrimidines (including thymine T, cytosine C, and uracil U) in food. For example, fluctuations in laser energy, the presence of background noise, and the non-uniformity of the baseline in portable Raman spectrometers are major interfering factors. These factors can all lead to a decrease in spectral data quality, further affecting the accuracy of subsequent analytical results. Furthermore, due to the certain similarity in chemical structure among nucleic acid bases, their characteristic Raman spectral response signals may overlap or shift. This phenomenon not only increases the difficulty of spectral interpretation but also significantly increases the complexity of quantitative analysis. Especially when constructing partial least squares (PLS) calibration models, the overlap or shift of characteristic peak positions can lead to a decrease in model accuracy and stability, thus affecting the quantitative analysis results. Therefore, spectral preprocessing is necessary to improve model accuracy and stability.

[0037] (1) The Raman spectral data collected in step 2 were preprocessed using the first derivative (d1st). The preprocessed data were then used to construct a PLS calibration model using partial least squares regression. After preprocessing the data by setting different smoothing points, the PLS calibration model for adenine A, guanine G, thymine T, cytosine C, and uracil U was input, and the root mean square error of cross-validation (RMSECV) and the coefficient of determination (R²) were output. CV Using R² as the evaluation index, the optimal smoothing point is selected to obtain the optimal PLS calibration model. CV The larger the value and the smaller the RMSE, the better the model's predictive performance.

[0038] The results are as follows Figure 4 As shown in (a)-(e), for adenine analysis, RMSECV initially increases with the number of smoothed points, then decreases when the number of smoothed points is 15, but shows an upward trend again after the number of smoothed points is 19. When the number of smoothed points is 19, R² CV The maximum value was 0.9895, and the RESSMCV was 0.1512. For guanine analysis, the PLS calibration model was more effective when the smoothing point was 23, with R²... CV The R² value was 0.9907, and the RMSCV was 0.0676. For Tymine analysis, after optimization with a smoothing point of 17, the PLS calibration model had a lower RMSCV, with a RMSCV of 0.1794 and an R² value of 0.9907. CV The value is 0.9883; for cytosine analysis, when the smoothing point is 7, R² is... CV The R² value was 0.9877, and the RESSMCV was 0.2385; for uracil analysis, when the smoothing point was 19, the R² value of the PLS calibration model was 0.9877. CV The highest value was obtained, 0.9863, and the RESMCV was 0.3566.

[0039] (2) The Raman spectral data collected in step 2 were preprocessed using wavelet transform (WT), and the preprocessed data were used to construct a PLS calibration model using partial least squares regression. The surface-enhanced Raman scattering (SERS) spectral bands were decomposed into wavelet components of different frequencies and positions using wavelet transform (WT) preprocessing, serving as an effective means of noise removal and smoothing the spectral curves. The influence of different combinations of wavelet basis functions and decomposition levels on the performance of the calibration model was systematically evaluated. Different wavelet transform (WT) parameters were set, including wavelet basis functions (from db1 to db5) and decomposition levels (from 1 to 7). After preprocessing the data, PLS calibration models for adenine A, guanine G, thymine T, cytosine C, and uracil U were input, and the root mean square error of cross-validation (RMSECV) and coefficient of determination (R²) were output.CV By selecting the optimal wavelet basis function and the number of decomposition layers, the optimal PLS calibration model can be obtained.

[0040] The results are as follows Figure 5 As shown in (a)-(e), for the Adenine analysis, when db1 is selected as the wavelet basis function and the number of decomposition levels is set to 3, the WT-PLS calibration model exhibits good predictive performance, with a coefficient of determination (R²CV) of 0.9899 and a root mean square error of cross-validation (RMSECV) of 0.16. For the wavelet transform preprocessing analysis of Guanine, it was found that when db2 is used as the wavelet basis function and the number of decomposition levels is set to 4, the calibration model exhibits the best predictive performance, with R²CV reaching a high level. CV The R² value increased to 0.9909, while the RSMCV decreased to 0.1085. For Tymine analysis, when db3 was chosen as the wavelet basis function and the decomposition level was 3, R²... CV The maximum value of 0.9910 was reached, with a RESMCV of 0.1375. In the Cytosine analysis, when db4 was used as the wavelet basis function and the decomposition level was 4, the RESMCV of the PLS calibration model reached a lower value of 0.123, while R²... CV The R²CV remains at a high level of 0.9903. For the analysis of Uracil, when db1 is selected as the wavelet basis function and the number of decomposition layers is set to 4, the PLS calibration model can also obtain satisfactory prediction results. Under these conditions, the R²CV is as high as 0.9910 and the RESMCV is 0.1240.

[0041] (3) To improve model performance, a hybrid data preprocessing method with different combinations was adopted. For each nucleic acid base, the preprocessing combination that produced the best prediction results was selected. The impact of various preprocessing methods and their order on the model's predictive ability was compared. The results are as follows: Figure 6 As shown, the overall impact trend of different preprocessing methods and their combination order on model performance can be observed. With the increase of preprocessing steps and the optimization of the combination order, RMSECV gradually decreases, while R²... CV The values ​​gradually increase. This trend indicates that a reasonable combination and order of preprocessing steps are crucial for improving the prediction accuracy of the PLS calibration model.

[0042] Specifically, in the analysis of A, T, and C, it was found that the combination of D1st (first derivative) and wavelet transform (WT) (D1st+WT) is the optimal preprocessing method, with corresponding R² values ​​of 1 / 2 and 1 / 2 / 3 respectively. CVThe values ​​were 0.9892, 0.9911, and 0.9903, while the RMSECV values ​​were 0.1680, 0.1317, and 0.1093, respectively. These results indicate that the D1st+WT combination significantly improves the prediction accuracy of the PLS calibration model for A, T, and C. For substances G and U, the WT+D1st combination was also found to be the optimal pretreatment method, with corresponding R² values ​​of 0.9892, 0.9911, and 0.9903, respectively. CV The values ​​were 0.9902 and 0.9901, while the RMSECV values ​​were 0.1382 and 0.1087, respectively. This demonstrates that by employing a hybrid preprocessing method with different combinations and selecting the optimal preprocessing combination for each nucleic acid base, the prediction accuracy of the PLS calibration model was successfully improved. Step 4: Input variable optimization based on biPLS Building upon the optimal preprocessing combination described above, the input variables for biPLS are further optimized. The biPLS variable selection method aims to optimize the spectral data of each substance by selecting the optimal number of band intervals and eliminating intervals that are useless or have an adverse effect on the analysis results, thereby saving modeling time while improving prediction accuracy. This method also verifies the overlap between the substance's characteristic peaks and the selected bands, ensuring that the model can focus on the spectral regions containing key information.

[0043] To enable the PLS model to accurately locate intervals containing characteristic peaks, the full spectrum was divided into 10 to 20 intervals. This step was implemented based on in-depth analysis of the spectral data characteristics to ensure that each interval represents specific spectral features. The biPLS method was used to systematically screen each interval, removing those that did not contribute to the analysis and prediction results or had a negative impact.

[0044] The results are as follows Figure 7As shown, the prediction performance of the biPLS calibration model based on different interval numbers is compared. The results show that compared with the full-spectrum PLS model, the biPLS model significantly improves the prediction results in both R²cv and RMSECV. Further analysis revealed that when the number of intervals is set to 19, the prediction results for substances A and C are optimal, with R²cv of 0.9907 and RMSECV of 0.154 for A; and R²cv of 0.9901 and RMSECV of 0.087 for C. For G... Analysis shows that the PLS calibration model exhibits the best predictive performance when the number of intervals is set to 15, with an R²cv of 0.9909 and a RESMCV of 0.139. In the T analysis, the model also achieves good predictive performance when the number of intervals is 18, with an R²cv of 0.9904 and a RESMCV of 0.124. For the U analysis, when the number of intervals is set to 14, the PLS calibration model has an R²cv of 0.9899 and a relatively high RESMCV value of 0.0902. In summary, by employing the biPLS variable selection method, the input data of the PLS calibration model was successfully optimized, the predictive accuracy of the model was improved, and the overlap between the material characteristic peaks and the selected bands was verified.

[0045] Step 5: Verify the predictive performance of the PLS calibration model. Table 2 Prediction results based on the PLS calibration model

[0046] The detailed data in Table 2 further confirms the necessity of using biPLS variable selection after the preprocessing step. This method plays a crucial role in improving the model's predictive performance. For the predictive analysis of A in food, the D1st-WT-biPLS-PLS calibration model exhibits good predictive ability, with a prediction determination coefficient (R²p) of 0.9903 and a prediction root mean square error (MREP) of 0.0851. Analysis of G and U reveals that the PLS calibration model, which first applies WT for smoothing and denoising, then performs D1st processing to enhance spectral features, and finally uses biPLS variable selection, achieves the best predictive performance. The R²p for G and U reaches 0.9937 and 0.9911, respectively, while the MREPs are 0.0366 and 0.0572, respectively. This indicates that this preprocessing and variable selection strategy is particularly effective in improving the model's predictive accuracy. The PLS calibration model based on biPLS and D1st-WT exhibits better predictive performance for T and C, with R²p values ​​of 0.9920 and 0.9925, and MREP values ​​of 0.1247 and 0.0947, respectively. The RPD value is inversely proportional to the model prediction error; a larger RPD value indicates a smaller prediction error and higher reliability of the prediction results. These results demonstrate that a calibration model with high-precision predictive capability for specific substances can be successfully constructed through a reasonable combination of preprocessing methods and variable selection strategies.

[0047] Step 6: Establish the linear relationship between predicted and true values ​​under the PLS calibration model for the five substances. The spectral data of the samples collected in Table 1 were input into the optimal PLS calibration model for the five substances. The values ​​of the five substances were output as predicted values ​​(x) and the actual concentrations of the corresponding samples were used as y values. A linear relationship between the predicted and actual values ​​under the PLS calibration model for the five substances was established, and the results are as follows: Figure 8 As shown.

[0048] Figure 8 In the equation (a), adenine is represented, and its linear relationship is y = 1.0140x + 0.2175. Figure 8 In the middle (b), guanine is represented, and its linear relationship is y = 0.9956x + 0.0533; Figure 8 In the middle (c), thymine is represented, and its linear relationship is y = 1.0010x - 0.2818; Figure 8 In the equation (d), cytosine is used, and its linear relationship is y = 1.0111x + 0.1717. Figure 8In the equation (e), uracil is represented by the linear relationship y = 1.1017x + 0.0982.

[0049] In summary, this study successfully constructed a quantitative calibration model for five components (A, G, T, C, and U) in food by cleverly combining WT and D1st preprocessing techniques with the biPLS variable selection method. This model not only possesses high predictive accuracy but also enables precise quantitative analysis of target components, providing a powerful tool and new ideas and references for food quality control and component analysis.

[0050] Specific Example 4: HPLC detection of free bases (A, T, G, C, U).

[0051] High-performance liquid chromatography (HPLC) was used as an auxiliary method to analyze free bases in fish meat. This method aims to compare and verify the accuracy and convenience of Raman detection results.

[0052] Take 50 g of fish meat, mince and mix evenly. Weigh 1 g of the mixed fish meat (accurate to 0.01 g) into a 15 mL glass test tube, add 4 mL of perchloric acid, seal the tube, and shake for 2 min. Then, adjust the pH value to about 5.0 with 10 mol / L NaOH solution, and then adjust the pH value to 6.0 with 1.0 mol / L NaOH solution. Make up the volume to 25 mL with ultrapure water, centrifuge at 10000 rpm / min for 5 min, filter the supernatant through a 0.22 μm aqueous phase filter membrane, and wait for analysis (Veciana-Nogues et al., 1997). Chromatographic conditions: A Shimadzu Shim-pack Gis C18 column (250 mm × 4.6 mm ID, particle size 5 μm) was used; the DAD detection wavelength was 262 nm; the flow rate was 0.70 mL / min; the column temperature was 26℃; the injection volume was 10 μL; mobile phase A was a pH 7.25, 20 mmol / L sodium acetate solution, and mobile phase B was methanol. Results are as follows: Figure 9 As shown.

[0053] Specific Example 5: Analysis of Nucleic Acid Degradation and Free Bases (ATGCU) During Tuna Refrigeration Yellowfin tuna (caught in the Pacific Ocean in September 2024 by the tuna purse seiner "MARIBO 61", weighing 20±2 kg, frozen at -60℃ on board) was purchased from Zhoushan Yinghai Ocean Fishery Co., Ltd. The -60℃ ship-frozen yellowfin tuna was transferred to the -18℃ constant temperature freezer of Ningbo Jinri Food Co., Ltd. in December 2024 for a 120-day tuna storage experiment.

[0054] Six frozen tuna were collected at 0, 20, 40, 60, 80, 100 and 120 days respectively. After thawing at 4℃ for 24 h, three pieces of dorsal muscle (100g / piece) were taken from each fish, for a total of 18 samples. They were randomly divided into three groups of 6 samples each. Nucleic acid extraction and agarose gel electrophoresis analysis were performed in each group; free base extraction and HPLC detection were performed in the other group; and substrate puncture sampling with pluggable probes was performed and SERS detection was performed.

[0055] 1. Monitoring DNA degradation by agarose gel electrophoresis Total nucleic acid was extracted from fish tissue using a lysis method. A nucleic acid extraction kit (FastPure DNA Isolation Mini Kit) was used, and the procedure was followed according to the kit instructions. First, the fish tissue was crushed on ice, then lysis buffer and proteinase K were added, and the mixture was incubated overnight at 56°C to ensure complete tissue lysis. The tissue was then purified by centrifugation, and finally, the nucleic acids were eluted with elution buffer to obtain the total nucleic acid sample from the fish meat. Agarose gel electrophoresis was used to detect the total nucleic acid extracted from the fish samples. The samples were temporarily stored at -80°C for comparative analysis at the end of the experiment.

[0056] like Figure 10 As shown, in a 120-day tuna storage experiment where tuna was transferred from a -60℃ freezer to a -18℃ freezer, the gel electrophoresis bands of total DNA from the tuna's back flesh were concentrated and bright on day 0. At -18℃, with prolonged storage, the brightness of the tuna flesh's total DNA bands gradually decreased, and the tailing phenomenon intensified; this change is a typical characteristic of DNA degradation. At -18℃, some endogenous enzymes (such as nucleases) still retain some activity, which contributes to DNA degradation to some extent with prolonged storage. Furthermore, although low temperature can slow down oxidation reactions, the presence of oxygen leads to oxidative stress, which also damages DNA with prolonged storage.

[0057] Relationship between AGTCU content and quality changes in tuna during cold storage (Analysis of DNA degradation and lipid oxidation changes during tuna storage) Table 3. TBA content and ATGCU content detected by SERS and HPLC at different freezing times

[0058] Table 3 shows that the contents of free A, T, G, C, and U in fish meat detected by HPLC and SERS methods all gradually increased with prolonged storage time. SERS method showed higher sensitivity and accuracy than HPLC. This is mainly because HPLC detection of free A, T, G, C, and U requires the fish meat sample to undergo grinding, perchloric acid extraction, centrifugation, filtration, and instrumental analysis, taking at least 2.5 hours. Furthermore, perchloric acid has a certain acidic hydrolytic effect during the entire extraction process of free purines. Using the gold-plated probe substrate material with surface-modified AgNCs, as described in this study, the fish meat sample was directly punctured for Raman spectroscopy analysis of A, T, G, C, and U, which was completed in only 5 minutes. The entire operation was direct and time-efficient.

[0059] As shown in Table 3, in this experiment, under -18℃ conditions and frozen storage for 120 days, the TBA content in the dorsal muscle of tuna increased from 0.53 mg / 100g to 6.87 mg / 100g. This is because tuna muscle is rich in highly unsaturated fatty acids such as EPA and DHA, which undergo auto-oxidation and hydrolysis during frozen storage. The thiobarbituric acid (TBA) method for determining the oxidation of fat in fish meat is a classic method for evaluating tuna quality, indicating that the quality of tuna meat also gradually deteriorates under -18℃ conditions.

[0060] Linear regression analysis was used in statistics to calculate the correlation coefficient (R²) between TBA values ​​and the content of individual bases (A, T, G, C, U) and the content of base combinations (e.g., A+T, G+C+U, A+T+G+C+U). This quantifies the strength of the correlation. The closer R² is to 1, the stronger the linear relationship between the two and the better the predictive ability.

[0061] A comparative analysis was conducted on the correlation between changes in TBA and free bases in tuna flesh during 120 days of frozen storage. For example... Figure 11 (a) shows the free base content detected by SERS. The correlation coefficients (R²) between TBA content and A, T, G, C, U, A+T, G+C+U and A+T+G+C+U are shown. 2 All are greater than 0.8, and the order of size is: R 2 G+C+U >R 2 A+T+G+C+U >R 2 A+T >R 2 C >R 2 A >R 2 G >R 2 T >R 2U .

[0062] like Figure 11 (b) shows the free base content detected by HPLC. The correlation coefficients (R²) between TBA content and A, T, G, C, U, A+T, G+C+U, and A+T+G+C+U are also shown. 2 All are greater than 0.75, and the order of size is: R 2 G+C+U >R 2 A+T+G+C+U >R 2 G >R 2 T >R 2 A >R 2 U >R 2 C >R 2 A+T .

[0063] comprehensive Figure 11 (a) and Figure 11 The analysis results in (b) show that the changes in thiobarbituric acid (TBA) content and free base content in tuna stored at -18℃ for 120 days were positively correlated. In particular, the correlation coefficients (R²) for G+C+U reached 0.9695 and 0.9382, respectively, indicating that the deterioration of tuna meat quality during frozen storage includes oil oxidation and nucleic acid degradation. Among these, the change in the total content of the three free bases (G, C, and U) was most strongly correlated with TBA, an indicator of tuna oil deterioration, and can be used as an effective indicator for evaluating tuna quality. Based on the gold-plated probe substrate material with surface-modified Ag NCs prepared in this experiment, a rapid SERS detection method for free bases in fish meat was constructed, which can quickly and effectively detect tuna quality.

[0064] In summary, this study innovatively combines Raman spectroscopy with chemometrics, successfully applying gold-plated pluggable probes modified with Ag NCs to the detection of five free bases (adenine, guanine, thymine, cytosine, and uracil) in tuna meat. The binding of these five free bases (adenine, guanine, thymine, cytosine, and uracil) to the gold-plated probe substrate modified with Ag NCs significantly enhanced the Raman response, yielding SERS spectra with clear characteristic peaks. When the SERS spectra of the simulated mixture of five nucleic acid base components in the tuna matrix were bound to the gold-plated probe substrate modified with Ag NCs, some characteristic peaks showed enhanced overlap, while others showed weakened overlap.

[0065] To address the problem of mutual interference between mixed bases in fish meat, this study explored the impact of integrated strategies for single and mixed preprocessing methods on the predictive performance of calibration models. By optimizing the combination and order of spectral preprocessing methods and incorporating a two-way interval partial least squares (biPLS) variable selection method to process Raman spectral data of mixed bases, useless intervals and redundant variables were effectively removed. Results showed that the PLS calibration model established after optimizing the spectral preprocessing methods exhibited good predictive performance for the five free bases. The D1st-WT-biPLS-PLS calibration model performed well for A, T, and C, with prediction determination coefficients (R²p) of 0.9903, 0.9920, and 0.9925, respectively, and prediction root mean square errors (MREP) of 0.0851, 0.1247, and 0.0947, respectively. For G and U, WT-D1st-biPLS-PLS is a better calibration model, with R²p reaching 0.9937 and 0.9911 respectively, and MREP reaching 0.0366 and 0.0572 respectively. Validation and optimization of the combination and order of preprocessing methods further improved the predictive ability of the PLS calibration model.

[0066] Nucleic acid gel electrophoresis experiments revealed that nucleic acids in tuna meat stored at -18℃ for 120 days exhibited gradual degradation with prolonged storage. Based on the SERS spectroscopy and HPLC methods developed in this paper, the free bases generated from the degradation of five nucleic acids in tuna meat were detected. The results showed that the SERS spectroscopy method developed in this paper is more sensitive and convenient. During tuna storage at -18℃ for 120 days, the changes in the content of the five free bases were positively correlated with the TBA value of the fish meat, and the correlation coefficient between the content of the five free bases in the fish meat measured by the SERS spectroscopy method established in this paper and the TBA value was 0.9695. Therefore, the SERS rapid detection method for free bases in fish meat and the PLS mathematical model constructed based on the gold-plated probe substrate material modified with Ag NCs surface can achieve real-time detection of tuna quality.

[0067] The foregoing description is not intended to limit the invention, nor is the invention limited to the examples given. Any changes, modifications, additions, or substitutions made by those skilled in the art within the scope of the invention should also be considered within the protection scope of the invention.

Claims

1. A method for preparing a gold-plated SERS probe for detecting surface-modified Ag NCs of free bases, characterized in that... Includes the following steps: Step 1: Synthesize Ag NCs via a hydrothermal method to obtain an Ag NCs solution; Step 2: After cleaning the gold-plated needle, immerse it in sodium citrate solution, and then immerse it in AgNCs solution to obtain a gold-plated SERS probe with AgNCs surface modification.

2. The method for preparing a gold-plated SERS probe for detecting free bases using surface-modified Ag NCs according to claim 1, characterized in that... Step 1 is as follows: Mix 70-80 mM CTAB solution, 8-12 mM silver ammonia solution and 1-2 mM glucose solution in a volume ratio of 1:1:2 and stir until homogeneous. Heat the mixture at 100-150°C for 6-10 hours. After cooling, remove the mixture and centrifuge and wash it with deionized water to obtain AgNCs solution.

3. The method for preparing a gold-plated SERS probe for detecting free bases using surface-modified Ag NCs according to claim 1, characterized in that... Step 2 is as follows: After ultrasonic treatment in acetone and ethanol to remove surface impurities, the cleaned needle is immersed in a 0.5-1.5 wt% sodium citrate solution for 10-14 hours, then cleaned with ethanol. The gold-plated needle is then placed in an AgNCs solution and soaked for 10-14 hours. After washing with ultrapure water and ethanol, it is dried in a nitrogen atmosphere to obtain a gold-plated SERS probe with AgNCs surface modification.

4. An application of a gold-plated SERS probe for detecting free bases, prepared by the method of any one of claims 1-3, for real-time monitoring of the quality of frozen tuna, characterized in that: The free base is at least one of adenine, guanine, cytosine, thymine, and uracil.

5. The application according to claim 4, characterized in that... The method for detecting the free base content in frozen fish includes the following steps: Step 1: Insert the gold-plated SERS probe 1.5-2 cm into the frozen fish sample to be tested, leave it for 30 seconds, then remove it and perform Raman spectroscopy detection, collecting samples from 400-1800 cm⁻¹. -1 Raman spectral data of the range; Step 2: After preprocessing the Raman spectral data collected in Step 1, input them into the PLS models of five substances: adenine, guanine, cytosine, thymine, and uracil, and output the predicted concentration values ​​x of adenine, guanine, cytosine, thymine, and uracil in the frozen sample to be tested. Step 3: Substitute the five predicted concentration values ​​x obtained in Step 2 into the following formula for the linear relationship between predicted and true values ​​to calculate the concentration values ​​y of adenine, guanine, cytosine, thymine and uracil in the frozen fish sample to be tested. The linear relationship between the predicted and true values ​​of adenine is y = 1.0140x + 0.2175. The linear relationship between the predicted and actual values ​​of guanine is y = 0.9956x + 0.0533; The linear relationship between the predicted and actual values ​​of thymine is y = 1.0010x - 0.2818; The linear relationship between the predicted and actual values ​​of cytosine is y = 1.0111x + 0.1717; The linear relationship between the predicted and actual values ​​of uracil is y = 1.1017x + 0.0982.

6. The application according to claim 5, characterized in that: In step 2, for the adenine PLS model, the Raman spectral data preprocessing method is to first perform first derivative preprocessing, select 19 smoothing points, then perform wavelet transform, select db1 as the wavelet basis function, and set the number of decomposition layers to 3; the number of intervals for the biPLS model is set to 19.

7. The application according to claim 5, characterized in that: In step 2, for the guanine PLS model, the Raman spectral data preprocessing method is as follows: first, wavelet transform is performed, db2 is selected as the wavelet basis function, and the number of decomposition layers is set to 4; then, first derivative preprocessing is performed, and the number of smoothing points is selected to be 23; then, the number of intervals for the biPLS model is set to 15.

8. The application according to claim 5, characterized in that: In step 2, for the thymine PLS model, the Raman spectral data preprocessing method is to first perform first derivative preprocessing, select 17 smoothing points, then perform wavelet transform, select db3 as the wavelet basis function, and set the number of decomposition layers to 3; the number of intervals for the biPLS model is set to 18.

9. The application according to claim 5, characterized in that: In step 2, for the cytosine PLS model, the Raman spectral data preprocessing method is to first perform first derivative preprocessing, select 7 smoothing points, then perform wavelet transform, select db4 as the wavelet basis function, and set the number of decomposition layers to 4; the number of intervals for the biPLS model is set to 19.

10. The application according to claim 5, characterized in that: In step 2, for the uracil PLS model, the Raman spectral data preprocessing method is as follows: first, wavelet transform is performed, db1 is selected as the wavelet basis function, and the number of decomposition layers is set to 4; then, first derivative preprocessing is performed, and the number of smoothing points is selected to be 19; then, the number of intervals of the biPLS model is set to 14.

Images