Preparation method and application of SERS (Surface Enhanced Raman Scattering) substrate with sunken structure for prostatic cancer exosome detection

Abstract

The invention discloses a preparation method and application of a concave-structure SERS (Surface Enhanced Raman Scattering) substrate for prostatic cancer exosome detection, and relates to the field of Raman spectrum detection. The invention aims to solve the problems that the quantity of electromagnetic field hot spots of the currently reported SERS substrate with the concave structure is small, the SERS substrate does not have good uniformity, the size is not matched with EVs, the detected EVs signal is relatively weak, and the parting detection requirement cannot be met. The preparation method of the SERS substrate with the concave structure comprises the following steps: 1, preparing a PDMS concave trap structure array; 2, preparing a concave well array of which the surface is covered with densely arranged Au NPs; according to the invention, Au nanoparticles are grown in situ in a PDMS concave structure, a concave trap array of which the size is matched with that of EVs and the surface is covered with densely arranged Au NPs is constructed, high-sensitivity detection (lower detection limit of 105 particles / mL) and high-accuracy typing of prostatic cancer EVs are realized, and the accuracy rate can reach 98% or more.

Description

Technical Field

[0001] The present invention relates to the field of Raman spectroscopy detection, and in particular to a preparation method and application of a SERS substrate with a concave structure for prostate cancer exosome detection. Background Art

[0002] Prostate cancer has extremely high metastatic and invasive characteristics, which is the main reason for its high mortality rate. By developing a variety of liquid biopsy methods, active health testing of prostate cancer patients in the early localized stage of the disease can be carried out to avoid the side effects of puncture biopsy, which has become the key to improving the quality of life and survival rate of patients.

[0003] Exosomes, also known as extracellular vesicles (EVs), are stable structures with a diameter of about 30-200nm, wrapped by a phospholipid bilayer, and are actively released by cells. They contain proteins, phospholipids, nucleic acids, and a variety of metabolites. Since EVs are continuously released by tumor cells, the membrane proteins on their surface and the substances inside can constantly reflect the progression of cancer. Detecting EVs in patients has become the key to the diagnosis and treatment of patients. Existing EVs detection methods, such as mass spectrometry, fluorescence, and flow cytometry, rely on large and expensive detection equipment and specialized labor costs, with a long detection cycle, which cannot meet the needs of rapid clinical detection.

[0004] Surface enhanced Raman spectroscopy (SERS) can provide significantly amplified Raman signals, with the advantages of high sensitivity, high information content and no labeling. The acquisition of EVs fingerprint-level molecular signals through label-free SERS spectroscopy, as well as quantitative detection and rapid typing of EVs, has attracted widespread attention from scientists in recent years. However, the existing label-free SERS detection methods for EVs have problems such as low sensitivity and poor typing effect. The main reason is that unlike some small molecules, EVs have a spherical structure with a diameter of about 50-100nm and a large volume, making it difficult for them to fully contact the electromagnetic field hotspot area. The SERS substrate with concave structure characteristics has a strong light trapping ability and can distribute the electromagnetic field into a three-dimensional spatial structure, which is helpful for the detection of large detection targets such as EVs. It has attracted widespread attention from scientists in recent years. However, it has been reported that the EVs signal detected by the concave structure SERS substrate is weak, and effective information cannot be obtained to meet the typing detection needs. The main reason is that on the one hand, the number of electromagnetic field hotspots of the existing concave structure SERS substrate is small and does not have good uniformity. For example, the existing recessed structure SERS substrates are all prepared by physical methods and are composed of smooth gold films. Their SERS enhancement performance is very limited. Although the number of SERS hotspots can be significantly increased by growing Au nanoparticles (Au NPs) by chemical methods, it is impossible to construct dense and uniform Au NPs inside the recessed structure due to the limitations of existing methods. On the other hand, the existing recessed structure SERS substrates have not yet matched the size of EVs, making it impossible for EVs to fully contact the SERS hotspots.

[0005] Therefore, through a new in situ growth method, a SERS substrate covered by a dense and uniform Au NPs array with a well size that matches the size of EVs is constructed, thereby improving its SERS detection effect on EVs. Combined with machine learning and other algorithms, high-sensitivity and high-accuracy prostate cancer exosome typing can be achieved, which is of great research significance. Summary of the invention

[0006] The purpose of the present invention is to solve the problems that the currently reported recessed structure SERS substrates have a small number of electromagnetic field hotspots, lack good uniformity, and have a size that does not match EVs, and the detected EVs signals are weak and cannot meet the needs of typing detection, and to provide a preparation method and application of a recessed structure SERS substrate for prostate cancer exosome detection.

[0007] The present invention provides a method for preparing a novel concave structure SERS substrate, through which SERS signals are obtained, and in conjunction with a machine learning algorithm, high-accuracy rapid detection and typing of prostate cancer EVs are achieved.

[0008] The present invention achieves in situ growth of AuNPs on the PDMS concave structure by regulating the in situ growth conditions, such as using hydrogen peroxide (H2O2) as a mild reducing agent and introducing sodium oleate to adjust the pH of the system, thereby avoiding the problems of self-nucleation and fast reaction rate of the system, and constructing a concave well array with densely arranged Au NPs on the surface, excellent uniformity, and size matching that of EVs, to achieve high-sensitivity detection of prostate cancer EVs (lower limit of detection 10 5 particles / mL) and high-accuracy typing (accuracy 98%).

[0009] A method for preparing a concave structure SERS substrate for prostate cancer exosome detection is specifically completed by the following steps:

[0010] 1. Preparation of PDMS concave well structure array:

[0011] ①, construct a single-layer polystyrene microsphere array on the surface of the substrate by liquid surface self-assembly method, and then heat and solidify;

[0012] ②, pouring PDMS glue on the single-layer polystyrene microsphere array, heating it for a certain period of time to solidify it, peeling off the film, cleaning it and blowing it with nitrogen to obtain a PDMS concave well structure array;

[0013] 2. Preparation of a surface-covered well array with densely packed Au NPs:

[0014] ① First, use a plasma cleaning machine to perform surface hydrophilization treatment on the PDMS concave well structure array, then place it in an APTES anhydrous ethanol solution for heating, take it out and wash it with anhydrous ethanol, blow dry it with nitrogen, and obtain an amino-modified concave well array;

[0015] ②, immerse the amino-modified concave well array in the gold seed solution for a period of time, and then rinse it with deionized water to obtain a gold seed-modified concave well array;

[0016] ③. Immerse the gold seed-modified concave well array into the growth solution, react at room temperature for a period of time, take it out and then wash it, and obtain a concave well array with densely arranged Au NPs on the surface, which is the concave structure SERS substrate for prostate cancer exosome detection;

[0017] The growth solution described in step 2③ is composed of water, sodium oleate solution, tetrachloroauric acid solution, polyvinyl pyrrolidone solution, and hydrogen peroxide.

[0018] The recessed structure SERS substrate for prostate cancer exosome detection was used to detect prostate cancer exosomes.

[0019] Beneficial effects of the present invention:

[0020] 1. The present invention achieves in-situ growth of Au nanoparticles inside the PDMS recessed structure by regulating in-situ growth conditions, such as using hydrogen peroxide (H2O2) as a mild reducing agent, introducing sodium oleate to adjust the pH of the system, and using PVP as a surfactant, avoiding the problems of self-nucleation and fast reaction rate of the system, and constructing a recessed well array with densely arranged Au NPs on the surface, excellent uniformity, and a size matching that of EVs;

[0021] Second, the SERS substrate prepared by the present invention can fit single EVs tightly inside the concave well structure, providing a more effective three-dimensional SERS hotspot, which helps to better obtain the SERS signal of EVs and achieve high-sensitivity detection of prostate cancer EVs (lower limit of detection 10 5 particles / mL), which is three orders of magnitude higher than that of the gold-film-coated concave-structure SERS substrate prepared by magnetron sputtering, and is helpful for the spectral detection of low-concentration EVs;

[0022] 3. The present invention achieves high-accuracy identification of four types of prostate cancer cell EVs by deep learning modeling of the SERS spectra of prostate cancer EVs. The accuracy, precision, and recall rates of the model are 98.86%, 98.85%, and 98.87%, respectively.

[0023] 4. The preparation method provided by the present invention does not need to rely on large-scale equipment and has low cost; the detection process is convenient and fast, and no labeling is required, which can realize rapid detection and typing of EVs; the present invention provides a new solution for SERS detection of EVs, which can be easily extended to the detection of other target substances similar to EVs, such as viruses. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A schematic diagram of a preparation method and application of a concave structure SERS substrate for prostate cancer exosome detection according to the present invention;

[0025] Figure 2 Scanning electron microscope images of the concave structure SERS substrates prepared in experiments 1 to 4, wherein Nanowell-Au NPs 1 is the SERS substrate prepared in experiment 1, Nanowell-Au NPs 2 is the SERS substrate prepared in experiment 2; Nanowell-Au NPs 3 is the SERS substrate prepared in experiment 3; Nanowell-Au NPs 4 is the SERS substrate prepared in experiment 4;

[0026] Figure 3Atomic force microscopy images of the concave structure SERS substrates prepared in Experiments 1 to 4, where (a) is Nanowell-Au NPs 1, (b) is Nanowell-Au NPs 2, (c) is Nanowell-Au NPs 3, and (d) is Nanowell-Au NPs 4;

[0027] Figure 4 Transmission electron microscopy images of four types of prostate cancer cell EVs;

[0028] Figure 5 Nanoparticle tracer analysis images of four prostate cancer cell EVs;

[0029] Figure 6 Scanning electron microscopy images of LNCaP cell EVs in the Nanowell-Au NPs 2 structure;

[0030] Figure 7 (a) is the SERS comparison spectra of Nanowell-Au NPs prepared in experiments 1 to 4 on LNCaP cell EVs, (b) is the SERS comparison spectra of Nanowell-Au NPs 2, Au NPs, and Nanowell-Au film prepared in experiments 2, 5, and 6 on LNCaP cell EVs;

[0031] Figure 8 (a) is the SERS spectra of Nanowell-Au film prepared in experiment 6 to LNCaP cell EVs with different concentrations, (b) is the SERS spectra of Nanowell-Au NPs 2 prepared in experiment 2 to LNCaP cell EVs with different concentrations;

[0032] Fig. 9 (a) is the SERS comparison spectra of Nanowell-Au NPs 2 prepared in Experiment 2 on four prostate cancer cells (LNCaP, RWPE1, PC3, DU145); (b) is the t-distributed stochastic neighbor embedding (t-SNE) scatter plot;

[0033] Fig.10 This is the deep learning model architecture diagram corresponding to Experiment 1, Experiment 8, and Experiment 9;

[0034] Fig.11 Diagram of the model training process of SERS spectra of four EVs obtained by different deep learning models. DETAILED DESCRIPTION

[0035] Specific embodiment 1: In this embodiment, a method for preparing a concave structure SERS substrate for prostate cancer exosome detection is specifically completed according to the following steps:

[0036] 1. Preparation of PDMS concave well structure array:

[0037] ①, construct a single-layer polystyrene microsphere array on the surface of the substrate by liquid surface self-assembly method, and then heat and solidify;

[0038] ②, pouring PDMS glue on the single-layer polystyrene microsphere array, heating it for a certain period of time to solidify it, peeling off the film, cleaning it and blowing it with nitrogen to obtain a PDMS concave well structure array;

[0039] 2. Preparation of a surface-covered well array with densely packed Au NPs:

[0040] ① First, use a plasma cleaning machine to perform surface hydrophilization treatment on the PDMS concave well structure array, then place it in an APTES anhydrous ethanol solution for heating, take it out and wash it with anhydrous ethanol, blow dry it with nitrogen, and obtain an amino-modified concave well array;

[0041] ②, immerse the amino-modified concave well array in the gold seed solution for a period of time, and then rinse it with deionized water to obtain a gold seed-modified concave well array;

[0042] ③. Immerse the gold seed-modified concave well array into the growth solution, react at room temperature for a period of time, take it out and then wash it, and obtain a concave well array with densely arranged Au NPs on the surface, which is the concave structure SERS substrate for prostate cancer exosome detection;

[0043] The growth solution described in step 2③ is composed of water, sodium oleate solution, tetrachloroauric acid solution, polyvinyl pyrrolidone solution, and hydrogen peroxide.

[0044] Specific implementation method 2: The difference between this implementation method and specific implementation method 1 is that the method of liquid surface self-assembly described in step 1① is as follows: add polystyrene microspheres to a mixture of water and ethanol, mix them evenly by ultrasonication, and obtain a polystyrene microsphere dispersion with a mass fraction of 1.25% to 2.5%; use a syringe to transfer the polystyrene microsphere dispersion with a mass fraction of 1.25% to 2.5% to a substrate, and construct a single-layer polystyrene microsphere array on the substrate surface; the volume ratio of water to ethanol in the water and ethanol mixture is 1:1; the particle size of the polystyrene microsphere is 670nm; the substrate described in step 1① is a hydrophilic silicon wafer with a size of 0.5cm×0.5cm. The preparation method is: clean the substrate in O2 plasma for 60s to 120s, with a power of 100W to 200W; the temperature of the heating and curing described in step 1① is 90℃ to 110℃, and the heating and curing time is 60min to 90min. The other steps are the same as those in specific implementation method 1.

[0045] Specific implementation method three: The difference between this implementation method and specific implementation method one or two is that: the mass ratio of PDMS main agent to curing agent in the PDMS glue described in step one ② is (8.5-9.5):1, and the model of the PDMS glue is Dow Corning SYLGARD184; the amount of PDMS glue described in step one ② is 5g-10g; the heating temperature described in step one ② is 60℃-90℃, and the heating time is 3h-5h; the cleaning described in step one ② is to use dichloromethane for cleaning 2-4 times. The other steps are the same as specific implementation method one or two.

[0046] Specific embodiment 4: This embodiment differs from specific embodiments 1 to 3 in that: the plasma cleaning machine described in step 2① is 100W to 200W; the surface hydrophilization treatment time is 60s to 120s; the mass fraction of the anhydrous ethanol solution of APTES described in step 2① is 2% to 5%; the heating temperature described in step 2① is 60℃ to 65℃, and the heating time is 3h to 6h. The other steps are the same as those of specific embodiments 1 to 3.

[0047] Specific implementation mode 5: The difference between this implementation mode and specific implementation modes 1 to 4 is that the preparation method of the gold seed solution described in step 2 ② is as follows: 2.5 mL of sodium borohydride solution with a concentration of 20 mmol / L to 25 mmol / L is added to 100 mL of chloroauric acid and sodium citrate solution, and stirred for 2 h to 4 h; the chloroauric acid and sodium citrate solution is obtained by dissolving chloroauric acid and sodium citrate in deionized water, wherein the mass fraction of chloroauric acid is 0.1%, and the mass fraction of sodium citrate is 0.01%; the ultrasonic time described in step 2 ② is 30 min to 60 min; after taking out in step 2 ②, rinse with deionized water for 3 to 5 times. The other steps are the same as those of specific implementation modes 1 to 4.

[0048] Specific implementation method 6: The difference between this implementation method and specific implementation methods 1 to 5 is that the reaction time at room temperature described in step 2 ③ is 20 min to 30 min; after taking out in step 2 ③, it is first washed with anhydrous ethanol 2 to 3 times, and then washed with deionized water 2 to 3 times; the concentration of sodium oleate described in step 2 ③ is 8 mmol / L to 10 mmol / L; the concentration of tetrachloroauric acid is 8 mmol / L to 10 mmol / L; the concentration of polyvinyl pyrrolidone is 1% to 5%; the concentration of hydrogen peroxide is (35% to 37%); the volume ratio of water, sodium oleate solution, tetrachloroauric acid solution, polyvinyl pyrrolidone solution and hydrogen peroxide in the growth solution is 50: (1 to 5): (5 to 35): 5: (2 to 5). The other steps are the same as those of specific implementation methods 1 to 5.

[0049] Specific embodiment seven: This embodiment is a concave structure SERS substrate for detecting prostate cancer exosomes.

[0050] Specific embodiment eight: This embodiment differs from specific embodiments one to seven in that: the concave structure SERS substrate for prostate cancer exosome detection is used to detect prostate cancer exosomes, which is specifically completed according to the following steps:

[0051] 1. Pipette the prostate cancer exosome solution onto the surface of the concave structure SERS substrate for prostate cancer exosome detection, and then wait for the solution to evaporate at room temperature;

[0052] The solvent in the prostate cancer exosome solution described in step 1 is PBS buffer or water with a pH of 7.4;

[0053] The concentration of the prostate cancer exosome solution described in step 1 is 10 5 ~10 10 Number of particles / mL;

[0054] Second, the SERS spectrum of the prostate cancer exosomes in the concave well is detected by a confocal Raman spectrometer to obtain a SERS spectrum;

[0055] If the collected SERS characteristic peak is 656cm -1 、736cm -1 、877cm -1 、1169cm -1 、1319cm -1 、1472cm -1 、1586cm -1 、1620cm -1 , that is, the SERS signal of the exosomes is obtained. The other steps are the same as those in the first to seventh embodiments.

[0056] Specific implementation method 9: The difference between this implementation method and specific implementation methods 1 to 8 is that: the confocal Raman spectrometer objective lens magnification described in step 2 is 100X, the laser wavelength is 785nm, the power is 2mW, and the integration time is 10s; 5 to 50 sample points are collected for each SERS detection sample; the SERS spectrum obtained in step 2 is preprocessed to obtain a preprocessed SERS spectrum; the preprocessing is divided into three steps: smoothing, baseline correction and standardization, wherein smoothing adopts Savitzky-Golay filtering, the number of windows is 20, and the baseline correction adopts asymmetric reweighted penalty least squares smoothing method. The other steps are the same as specific implementation methods 1 to 8.

[0057] Specific embodiment ten: The difference between this embodiment and specific embodiments one to nine is that: the pre-processed SERS spectrum is introduced into the deep learning model for training to achieve non-labeled SERS spectrum classification of prostate cancer exosomes;

[0058] The deep learning model is obtained by improving the residual network Resnet architecture, which is divided into six residual blocks, each of which is divided into two residual units, wherein the number of output channels of each residual block is 64, 64, 128, 128, 256, 256, respectively, and connected by the Relu function; the model finally imports 19200 neurons in the fully connected layer, and the final result is converted into the prostate cancer exosome category by the Softmax function; the model learning rate optimizer type is Adam, and the number of iterations is 200 times. The other steps are the same as the specific implementation methods one to nine.

[0059] Figure 1 A schematic diagram of a preparation method and application of a concave structure SERS substrate for prostate cancer exosome detection according to the present invention;

[0060] Depend on Figure 1It can be seen that: the present invention provides a new non-labeling SERS spectroscopy detection strategy for prostate cancer EVs. A concave SERS substrate loaded with densely arranged gold nanoparticles was constructed through an in situ chemical growth strategy. On the one hand, the opening size and depth of the concave well can be flexibly adjusted by controlling the reaction conditions, which is conducive to contact with the surface of EVs; on the other hand, the densely arranged Au NPs provide abundant electromagnetic field hotspots, which help to stimulate the membrane protein signals of EVs. It greatly improves the problems of the concave structure loaded with gold film prepared by physical methods such as magnetron sputtering, such as dependence on large equipment, complex preparation process, limited electromagnetic field hotspots and weak signals. By deeply studying the SERS spectra of prostate cancer cell EVs with different degrees of invasiveness obtained by ultracentrifugation, typing detection was achieved.

[0061] The following examples are used to verify the beneficial effects of the present invention:

[0062] Experiment 1: A method for preparing a concave structure SERS substrate for prostate cancer exosome detection is completed by the following steps:

[0063] 1. Preparation of PDMS-Nanowell structure array:

[0064] ①, construct a single-layer polystyrene microsphere (PS sphere) array on the substrate surface by liquid surface self-assembly method, and then heat and solidify;

[0065] The method of liquid surface self-assembly described in step 1① is as follows: adding polystyrene microspheres to a mixture of water and ethanol, and uniformly mixing by ultrasonication to obtain a polystyrene microsphere dispersion with a mass fraction of 1.25%; transferring the polystyrene microsphere dispersion with a mass fraction of 1.25% to a substrate using a syringe, and constructing a single-layer polystyrene microsphere array on the substrate surface; the volume ratio of water to ethanol in the mixture of water and ethanol is 1:1; the particle size of the polystyrene microspheres is 670nm;

[0066] The substrate described in step 1① is a hydrophilized silicon wafer with a size of 0.5 cm×0.5 cm. The preparation method is: cleaning the substrate in O2 plasma for 75 seconds at a power of 150 W;

[0067] The temperature of the heating and curing described in step 1① is 100°C, and the heating and curing time is 70 minutes;

[0068] ②, pouring PDMS glue on the single-layer polystyrene microsphere array, heating it for a certain period of time to solidify it, peeling off the film, cleaning it and blowing it with nitrogen to obtain a PDMS concave well structure array;

[0069] The model of the PDMS glue is Dow Corning SYLGARD184;

[0070] The amount of PDMS glue used in step 1② is 7.5 g; the heating temperature in step 1② is 65° C., and the heating time is 4.5 hours;

[0071] The cleaning described in step 1② is to use dichloromethane to clean twice;

[0072] 2. Preparation of a surface-covered well array with densely packed Au NPs (Nanowell-Au NPs):

[0073] ① First, the surface of the PDMS well structure array was hydrophilized using a plasma cleaner, and then placed in an APTES anhydrous ethanol solution for heating, and then taken out and cleaned with anhydrous ethanol and dried with nitrogen to obtain an amino-modified well array (Nanowell-APTES);

[0074] The plasma cleaning machine described in step 2① is 150W; the surface hydrophilization treatment time is 75s;

[0075] The mass fraction of the anhydrous ethanol solution of APTES described in step 2① is 5%;

[0076] The heating temperature in step 2① is 65°C and the heating time is 3h;

[0077] ②, immerse the amino-modified nanowell array into the gold seed solution for a period of time, take it out and rinse it with deionized water to obtain the nanowell array modified with gold seeds (Nanowell-Au seeds);

[0078] The preparation method of the gold seed solution described in step 2② is as follows: add 2.5mL of sodium borohydride solution with a concentration of 20mmol / L to 100mL of chloroauric acid and sodium citrate solution, and stir for 3h; the chloroauric acid and sodium citrate solution is obtained by dissolving chloroauric acid and sodium citrate in deionized water, wherein the mass fraction of chloroauric acid is 0.1%, and the mass fraction of sodium citrate is 0.01%;

[0079] The ultrasonic time in step 2② is 30 minutes; after taking out in step 2②, rinse with deionized water 3 times;

[0080] ③. Immerse the gold seed-modified concave well array in the growth solution, react at room temperature for a period of time, take it out and wash it, and obtain a concave well array with densely arranged Au NPs on the surface, which is the concave structure SERS substrate for prostate cancer exosome detection (Nanowell-Au NPs 1);

[0081] The growth solution described in step 2③ is composed of water, sodium oleate solution, tetrachloroauric acid solution, polyvinyl pyrrolidone solution, and hydrogen peroxide; the concentration of the sodium oleate is 10mmol / L; the concentration of tetrachloroauric acid is 10mmol / L; the concentration of polyvinyl pyrrolidone is 1%; the concentration of hydrogen peroxide is 37%; the volume ratio of water, sodium oleate solution, tetrachloroauric acid solution, polyvinyl pyrrolidone solution and hydrogen peroxide in the growth solution is 50:1:5:5:2;

[0082] The reaction time at room temperature described in step 2③ is 30 minutes; after being taken out in step 2③, it is first washed 3 times with anhydrous ethanol and then washed 3 times with deionized water;

[0083] 3. Label-free SERS spectroscopy for prostate cancer EVs:

[0084] ①, drip the EVs solution (prostate cancer exosome solution) onto the surface of a 0.5 cm × 0.5 cm concave structure SERS substrate (Nanowell-Au NPs 1) for prostate cancer exosome detection through a pipette, and then wait for the solution to evaporate at room temperature;

[0085] The solvent in step 3① is PBS or water, the volume is 5 μL, and the concentration of EVs solution is 10 7 Number of particles / mL;

[0086] ②, Use confocal Raman spectrometer to detect the SERS spectrum of EVs in the concave well and pre-process the collected spectrum;

[0087] The confocal Raman spectrometer described in step 3② has an objective lens magnification of 100X, a laser wavelength of 785nm, a power of 2mW, and an integration time of 10s;

[0088] The spectrum preprocessing described in step 3② is divided into three steps: smoothing, baseline correction and standardization. The smoothing adopts Savitzky-Golay filtering (the number of windows is 20), and the baseline correction adopts asymmetric reweighted penalized least squares smoothing method;

[0089] ③. Import the obtained SERS spectra into the deep learning model for training to achieve non-labeled SERS spectral classification of prostate cancer EVs;

[0090] The deep learning model described in step 3③ is obtained by improving the residual network (Resnet) architecture and is divided into six residual blocks. Each residual block is divided into two residual units, where the number of output channels of each residual block is 64, 64, 128, 128, 256, 256 respectively, and they are connected by the Relu function;

[0091] In step 3③, the model is finally imported into the fully connected layer (19200 neurons), and the final result is converted into EVs category through the Softmax function;

[0092] The model learning rate optimizer type described in step 3③ is Adam, and the number of iterations is 200;

[0093] The model described in step 3③ is cross-validated by five-fold, and the average values ​​of parameters such as accuracy, precision, ROC curve, etc. of the model are analyzed to evaluate the model effect.

[0094] Experiment 2: The difference between this experiment and Experiment 1 is that the volume ratio of water, sodium oleate solution, tetrachloroauric acid solution, polyvinyl pyrrolidone solution and hydrogen peroxide in the growth solution described in step 2③ is 50:3:15:5:3. The other steps and parameters are the same as those in Experiment 1.

[0095] Experiment 3: The difference between this experiment and Experiment 1 is that the volume ratio of water, sodium oleate solution, tetrachloroauric acid solution, polyvinyl pyrrolidone solution and hydrogen peroxide in the growth solution described in step 2③ is 50:5:25:5:4. The other steps and parameters are the same as those in Experiment 1.

[0096] Experiment 4: The difference between this experiment and Experiment 1 is that the volume ratio of water, sodium oleate solution, tetrachloroauric acid solution, polyvinyl pyrrolidone solution and hydrogen peroxide in the growth solution described in step 2③ is 50:5:35:5:5. The other steps and parameters are the same as those in Experiment 1.

[0097] Experiment 5: The preparation method of the SERS substrate covered with Au NPs is completed in the following steps:

[0098] ① Immerse the blank silicon wafer in the gold seed solution for a period of time, then rinse it with deionized water to obtain a silicon wafer modified with gold seeds;

[0099] The preparation method of the gold seed solution described in step ① is as follows: 2.5 mL of a sodium borohydride solution with a concentration of 20 mmol / L is added to 100 mL of a chloroauric acid and sodium citrate solution, and stirred for 3 hours; the chloroauric acid and sodium citrate solution is obtained by dissolving chloroauric acid and sodium citrate in deionized water, wherein the mass fraction of chloroauric acid is 0.1%, and the mass fraction of sodium citrate is 0.01%;

[0100] The ultrasonic time in step ① is 30 minutes; after taking out in step 2②, rinse with deionized water 3 times;

[0101] ②, immerse the gold seed-modified silicon wafer into the growth solution, react for a period of time at room temperature, take it out and then wash it to obtain a SERS substrate covered with Au NPs (Au NPs);

[0102] The growth solution described in step ② is composed of water, sodium oleate solution, tetrachloroauric acid solution, polyvinyl pyrrolidone solution, and hydrogen peroxide; the concentration of the sodium oleate is 10mmol / L; the concentration of tetrachloroauric acid is 10mmol / L; the concentration of polyvinyl pyrrolidone is 1%; the concentration of hydrogen peroxide is 37%; the volume ratio of water, sodium oleate solution, tetrachloroauric acid solution, polyvinyl pyrrolidone solution and hydrogen peroxide in the growth solution is 50:1:5:5:2;

[0103] The reaction time at room temperature described in step ② is 30 minutes;

[0104] After taking out in step ②, first use anhydrous ethanol to wash 3 times, and then use deionized water to wash 3 times.

[0105] Experiment 6: The Au film was deposited on a blank silicon wafer using a magnetron sputtering deposition method to obtain a SERS substrate with a recessed structure coated with a gold film (Nanowell-Aufilm).

[0106] Figure 2 Scanning electron microscope images of the concave structure SERS substrates prepared in experiments 1 to 4, wherein Nanowell-Au NPs 1 is the SERS substrate prepared in experiment 1, Nanowell-Au NPs 2 is the SERS substrate prepared in experiment 2; Nanowell-Au NPs 3 is the SERS substrate prepared in experiment 3; Nanowell-Au NPs 4 is the SERS substrate prepared in experiment 4;

[0107] from Figure 2 It can be seen that a uniform and densely arranged Au NPs array was successfully prepared on the surface of the PDMS concave well structure, in which the Au NPs have uniform particle size, no agglomeration, and the Au NPs are fully covered without defects. Among them, the gold nanoparticle sizes corresponding to different Nanowell-Au NPs are 30nm, 70nm, 85nm and 95nm, respectively. As the diameter of the gold nanoparticles increases, the diameter of the concave structure gradually shortens, and the diameters corresponding to different Nanowell-Au NPs are 334nm, 227nm, 155nm, and 91nm, respectively. In theory, the diameter of the concave structure can be continuously adjusted according to the volume ratio of the growth solution. Since the diameter of exosomes is in the range of 30-200nm, the concave structure SERS substrate prepared in this experiment can match the size of EVs in the nanoscale, capture one EV in a single concave structure, and maximize the contact area between EVs and the sidewalls of the concave structure.

[0108] Figure 3Atomic force microscopy images of the concave structure SERS substrates prepared in Experiments 1 to 4, where (a) is Nanowell-Au NPs 1, (b) is Nanowell-Au NPs 2, (c) is Nanowell-Au NPs 3, and (d) is Nanowell-Au NPs 4;

[0109] from Figure 3 It can be seen that the average depths of the concave structures prepared in experiments 1 to 4 are 105nm, 151nm, 120nm and 82nm respectively, and the depth shows a pattern of first increasing and then decreasing. This is due to the growth of gold nanoparticles. In the early stage, the growth rate of gold nanoparticles at the top of the concave structure is faster, which increases the depth of the concave structure. However, in the later stage, as the gold particles at the bottom continue to grow, the depth gradually decreases.

[0110] Experiment 7: Extraction of EVs from prostate cancer cells. First, four prostate cancer cells with different invasiveness were cultured, among which RWPE-1 cells are normal prostate epithelial cells; LNCaP cells are androgen-dependent prostate lymphatic metastasis cell lines with moderate invasiveness; PC3 is androgen-independent, bone metastasis cell line with strong invasiveness; DU145 is a prostate brain metastasis cell line with strong invasiveness. When the cells grow to a density of 70%, the existing culture medium is discarded, and the cells are rinsed twice with a PBS solution with a pH value of 7.4, and the serum-free medium is replaced. The culture is continued for 48 hours, and the culture supernatant is collected, followed by gradient centrifugation, and a speed of 300g is applied. After 10 minutes, cells and other substances are precipitated and removed; 2000g speed for 10 minutes, dead cells and other substances are precipitated and removed; 10000g speed for 30 minutes to remove cell debris and microvesicles and other substances. The supernatant was centrifuged twice at 120,000 g for 70 minutes using an ultracentrifuge, and the resulting EVs were dispersed in PBS (pH = 7.4) to obtain an EVs sample (prostate cancer exosome solution).

[0111] Figure 4 Transmission electron microscopy images of four types of prostate cancer cell EVs;

[0112] like Figure 4 As shown, the sizes of the four types of EVs are all below 200 nm, showing a "tray-shaped" structure, and the phospholipid bilayer wrapping can be clearly seen.

[0113] Figure 5 Nanoparticle tracer analysis images of four prostate cancer cell EVs;

[0114] The test results showed that the average particle size of EVs ranged from 100 to 200 nm, and there were no particles with larger sizes, indicating that the EVs obtained in the experiment were of high purity.

[0115] Figure 6 Scanning electron microscopy images of LNCaP cell EVs in the Nanowell-Au NPs 2 structure;

[0116] from Figure 6 It can be seen that a single EV with a size of about 150nm is confined inside a single concave structure, and the EVs are in full contact with the Au nanoparticles on the sidewalls. Therefore, Nanowell-Au NPs 2 is proven to have excellent EV capture ability, and its size matches EVs very well, which can achieve the capture of single EVs inside the concave structure.

[0117] Figure 7 (a) is the SERS comparison spectra of Nanowell-Au NPs prepared in experiments 1 to 4 on LNCaP cell EVs, (b) is the SERS comparison spectra of Nanowell-Au NPs 2, Au NPs, and Nanowell-Au film prepared in experiments 2, 5, and 6 on LNCaP cell EVs;

[0118] from Figure 7 (a) shows that the Nanowell-Au NPs prepared in Experiment 2 achieved the strongest SERS signal intensity for EVs testing, mainly because Nanowell-Au NPs have higher SERS hotspot intensity and their size matches that of EVs.

[0119] from Figure 7 (b) The study found that Nanowell-Au NPs had the strongest signal intensity, which was about 20 times higher than that of smooth gold film.

[0120] Figure 8 (a) is the SERS spectra of Nanowell-Au film prepared in experiment 6 to LNCaP cell EVs with different concentrations, (b) is the SERS spectra of Nanowell-Au NPs 2 prepared in experiment 2 to LNCaP cell EVs with different concentrations;

[0121] The study found that Nanowell-Au NPs can achieve a concentration of 10 5 The Nanowell-Au film can only achieve a concentration of 10 8The detection limit of EVs was improved by 3 orders of magnitude in the test of 1000 particles / mL EVs. Nanowell-Au NPs have excellent EVs capture function, high detection sensitivity and good detection uniformity.

[0122] Fig. 9 (a) is the SERS comparison spectra of Nanowell-Au NPs 2 prepared in Experiment 2 on four prostate cancer cells (LNCaP, RWPE1, PC3, DU145); (b) is the t-distributed stochastic neighbor embedding (t-SNE) scatter plot;

[0123] Fig. 9 (a) By comparing the SERS signal images between EVs of different cells, it is found that the EVs signals of different cells are slightly different, only a few Raman peaks have different positions, most of the Raman peaks have the same peak position, and only a few differences exist in the intensity value. Among them, t-SNE is a common high-dimensional data visualization method. The difference from PCA is that it only retains small pairwise distances or local similarities, better retains local features, and each cell is almost completely separated in the t-SNE map.

[0124] Experiment 8: The difference between this experiment and Experiment 1 is that the deep learning model described in step 3③ is a deep learning model based on artificial neural network (ANN), which consists of four hidden layers, and the number of neurons on each layer is 256, 64, 16, and 4 respectively. The neurons are connected by Relu activation function, and the neural network structure is relatively simple. The other steps and parameters are the same as those in Experiment 1.

[0125] Experiment 9: The difference between this experiment and Experiment 1 is that the deep learning model described in step 3③ is a deep learning model based on the VGG-16 architecture. The model consists of three layers, each of which includes several convolutional units and a maximum pooling layer. The model finally outputs the final result through a fully connected layer. The other steps and parameters are the same as those in Experiment 1.

[0126] The deep learning model architecture diagrams corresponding to Experiment 1, Experiment 8, and Experiment 9 are as follows: Fig.10 shown.

[0127] Fig.11It is a relationship curve between the loss value, accuracy and number of iterations in the training process of different models. By observing the picture, it can be found that as the number of training iterations increases, the model accuracy gradually increases. Among them, the ANN model corresponding to test eight has a fast training speed. After convergence, the model accuracy no longer increases. Instead, the model accuracy decreases, indicating that the ANN model has insufficient learning ability for the test samples. The VGG16 model corresponding to test nine has the slowest training speed. After a certain number of trainings, the accuracy reaches the ideal value. However, there is instability in the model training process, which is manifested in the oscillation of the accuracy within a certain range of values. The model of test one is more stable and smooth during training, and can quickly converge to a range with a smaller loss value. The model has achieved the highest accuracy, specificity and recall. The model of the present invention can achieve high-accuracy typing of 4 types of prostate cancer cell EVs.

Claims

1. A method for preparing a concave structure SERS substrate for prostate cancer exosome detection, characterized in that The preparation method is specifically completed according to the following steps:

1. Preparation of PDMS concave well structure array: ①, construct a single-layer polystyrene microsphere array on the surface of the substrate by liquid surface self-assembly method, and then heat and solidify; ②, pouring PDMS glue on the single-layer polystyrene microsphere array, heating it for a certain period of time to solidify it, peeling off the film, cleaning it and blowing it with nitrogen to obtain a PDMS concave well structure array; 2. Preparation of a surface-covered well array with densely packed Au NPs: ① First, use a plasma cleaning machine to perform surface hydrophilization treatment on the PDMS concave well structure array, then place it in an APTES anhydrous ethanol solution for heating, take it out and wash it with anhydrous ethanol, blow dry it with nitrogen, and obtain an amino-modified concave well array; ②, immerse the amino-modified concave well array in the gold seed solution for a period of time, and then rinse it with deionized water to obtain a gold seed-modified concave well array; ③. Immerse the gold seed-modified concave well array into the growth solution, react at room temperature for a period of time, take it out and then wash it, and obtain a concave well array with densely arranged Au NPs on the surface, which is the concave structure SERS substrate for prostate cancer exosome detection; The growth solution described in step 2③ is composed of water, sodium oleate solution, tetrachloroauric acid solution, polyvinyl pyrrolidone solution, and hydrogen peroxide.

2. The method for preparing a concave structure SERS substrate for prostate cancer exosome detection according to claim 1, characterized in that The method of liquid surface self-assembly described in step 1① is as follows: add polystyrene microspheres to a mixture of water and ethanol, mix them evenly by ultrasonication, and obtain a polystyrene microsphere dispersion with a mass fraction of 1.25% to 2.5%; use a syringe to transfer the polystyrene microsphere dispersion with a mass fraction of 1.25% to 2.5% to a substrate, and construct a single-layer polystyrene microsphere array on the substrate surface; the volume ratio of water to ethanol in the water and ethanol mixture is 1:1; the particle size of the polystyrene microsphere is 670nm; the substrate described in step 1① is a hydrophilic silicon wafer with a size of 0.5cm×0.5cm. The preparation method is: clean the substrate in O2 plasma for 60s to 120s, with a power of 100W to 200W; the temperature of the heating and curing described in step 1① is 90℃ to 110℃, and the heating and curing time is 60min to 90min.

3. The method for preparing a concave structure SERS substrate for prostate cancer exosome detection according to claim 1, characterized in that The mass ratio of PDMS main agent to curing agent in the PDMS glue described in step 1② is (8.5-9.5):1, and the model of the PDMS glue is Dow Corning SYLGARD184; the amount of PDMS glue described in step 1② is 5g-10g; the heating temperature described in step 1② is 60℃-90℃, and the heating time is 3h-5h; the cleaning described in step 1② is 2-4 times of cleaning with dichloromethane.

4. The method for preparing a concave structure SERS substrate for prostate cancer exosome detection according to claim 1, characterized in that The plasma cleaning machine described in step 2① is 100W~200W; the surface hydrophilization treatment time is 60s~120s; the mass fraction of the anhydrous ethanol solution of APTES described in step 2① is 2%~5%; the heating temperature described in step 2① is 60℃~65℃, and the heating time is 3h~6h.

5. The method for preparing a concave structure SERS substrate for prostate cancer exosome detection according to claim 1, characterized in that The preparation method of the gold seed solution described in step 2② is as follows: add 2.5mL of sodium borohydride solution with a concentration of 20mmol / L to 25mmol / L to 100mL of chloroauric acid and sodium citrate solution, and stir for 2h to 4h; the chloroauric acid and sodium citrate solution is obtained by dissolving chloroauric acid and sodium citrate in deionized water, wherein the mass fraction of chloroauric acid is 0.1%, and the mass fraction of sodium citrate is 0.01%; the ultrasonic time described in step 2② is 30min to 60min; after taking out in step 2②, rinse with deionized water 3 times to 5 times.

6. The method for preparing a concave structure SERS substrate for prostate cancer exosome detection according to claim 1, characterized in that The reaction time at room temperature described in step 2③ is 20min to 30min; after being taken out in step 2③, it is first washed with anhydrous ethanol 2 to 3 times, and then washed with deionized water 2 to 3 times; the concentration of sodium oleate described in step 2③ is 8mmol / L to 10mmol / L; the concentration of tetrachloroauric acid is 8mmol / L to 10mmol / L; the concentration of polyvinyl pyrrolidone is 1% to 5%; the concentration of hydrogen peroxide is (35% to 37%); the volume ratio of water, sodium oleate solution, tetrachloroauric acid solution, polyvinyl pyrrolidone solution and hydrogen peroxide in the growth solution is 50:(1 to 5):(5 to 35):5:(2 to 5).

7. Application of the concave structure SERS substrate for prostate cancer exosome detection prepared by the preparation method according to claim 1, characterized in that The recessed structure SERS substrate for prostate cancer exosome detection was used to detect prostate cancer exosomes.

8. Application of the concave structure SERS substrate for prostate cancer exosome detection prepared by the preparation method according to claim 1, characterized in that The concave structure SERS substrate for prostate cancer exosome detection is used to detect prostate cancer exosomes, which is completed in the following steps:

1. Pipette the prostate cancer exosome solution onto the surface of the concave structure SERS substrate for prostate cancer exosome detection, and then wait for the solution to evaporate at room temperature; The solvent in the prostate cancer exosome solution described in step 1 is PBS buffer or water with a pH of 7.4; The concentration of the prostate cancer exosome solution described in step 1 is 10 5 ~10 10 Number of particles / mL; Second, the SERS spectrum of the prostate cancer exosomes in the concave well is detected by a confocal Raman spectrometer to obtain a SERS spectrum; If the collected SERS characteristic peak is 656cm -1 、736cm -1 、877cm -1 、1169cm -1 、1319cm -1 、1472cm -1 、1586cm -1 、1620cm -1 , that is, the SERS signal of exosomes is obtained.

9. Application of the concave structure SERS substrate for prostate cancer exosome detection prepared by the preparation method according to claim 1, characterized in that The confocal Raman spectrometer described in step 2 has an objective lens magnification of 100X, a laser wavelength of 785nm, a power of 2mW, and an integration time of 10s; 5 to 50 sample points are collected for each SERS detection sample; the SERS spectrum obtained in step 2 is preprocessed to obtain a preprocessed SERS spectrum; The preprocessing is divided into three steps: smoothing, baseline correction and standardization, wherein the smoothing adopts Savitzky-Golay filtering with a window number of 20, and the baseline correction adopts an asymmetric reweighted penalty least squares smoothing method.

10. Application of the concave structure SERS substrate for prostate cancer exosome detection prepared by the preparation method according to claim 1, characterized in that The pre-processed SERS spectra were imported into the deep learning model for training to achieve label-free SERS spectral classification of prostate cancer exosomes; The deep learning model is obtained by improving the residual network Resnet architecture, which is divided into six residual blocks, each residual block is divided into two residual units, wherein the number of output channels of each residual block is 64, 64, 128, 128, 256, 256, respectively, and connected by the Relu function; the model finally imports 19200 neurons in the fully connected layer, and the final result is converted into the prostate cancer exosome category by the Softmax function; the learning rate optimizer type of the model is Adam, and the number of iterations is 200 times.

Images