High-throughput anti-fibrosis drug screening method based on fluorescence-free image and application of high-throughput anti-fibrosis drug screening method

Abstract

The invention discloses a high-throughput anti-fibrosis drug screening method based on a fluorescence-free image and application of the high-throughput anti-fibrosis drug screening method, relates to the technical field of biological medicine, and aims to obtain a rapid and simple method capable of being used for large-scale drug screening. Correlation analysis is carried out on morphological parameters obtained by self-contained analysis software and the expression quantity of the alpha-SMA protein, 13 morphological parameters highly related to the expression quantity of the alpha-SMA protein are screened out, and the 13 morphological parameters are combined with an artificial intelligence method to be used for large-scale drug screening; the method has the advantages of high throughput, high accuracy, simplicity and convenience in operation, capability of being used for judging whether the medicine has an anti-fibrosis effect or not and the like, can improve the detection efficiency, can be applied to predicting whether the traditional Chinese medicine monomer has the anti-fibrosis effect or not, and proves the accuracy and effectiveness of screening the traditional Chinese medicine monomer as the anti-fibrosis medicine. Important experimental basis and technical support are provided for future drug development and fibrosis disease treatment.

Description

Technical Field

[0001] The present invention belongs to the field of biomedicine technology, and in particular relates to a high-throughput anti-fibrosis drug screening method based on machine learning without the need for fluorescent images and its application. Background Art

[0002] In recent years, fibrosis, as an accompanying problem in the progression of various diseases, has led to an increase in fibrous connective tissue in organ tissues, destruction of organ structure, dysfunction, and even failure, which seriously threatens human health and life. Currently, there is a lack of effective drugs for the treatment of various types of fibrosis, and new treatment strategies and methods are urgently needed. However, existing anti-fibrosis screening is only based on fluorescent staining of fibrosis indicators such as proliferation or α-SMA protein, which is expensive and low-throughput screening. In addition, previous studies have focused on blocking cell proliferation rather than maintaining the original state of fibroblasts; because fibroblasts also play an important role in many tissues and organs. Summary of the invention

[0003] The purpose of the present invention is to provide a high-throughput anti-fibrosis drug screening method based on machine learning without the need for fluorescent images and its application. By providing an anti-fibrosis phenotype screening method that does not require special fluorescent labels, is high-throughput, unbiased, and conforms to the characteristics of fibrotic diseases, the existing anti-fibrosis screening is solved based only on fluorescent staining of fibrosis indicators such as proliferation or α-SMA protein, which is an expensive, low-throughput screening. In addition, previous studies have focused on blocking cell proliferation rather than maintaining the original state of fibroblasts.

[0004] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0005] The present invention provides a high-throughput anti-fibrosis drug screening method based on machine learning without the need for fluorescence images and its application, comprising the following steps:

[0006] Step 1: DPC imaging and fluorescence imaging under stimulation of different concentrations of TGF-βI (0.1ng / ml, 0.5ng / ml, 1ng / ml, 5ng / ml, 10ng / ml);

[0007] Step 2: Under the same TGF-βI concentration (10 ng / mL), DPC imaging parameters and α-SMA fluorescence imaging were collected at different times (3 h, 6 h, 12 h, 24 h, and 36 h);

[0008] Step 3: SD-208 is a selective TGF-βI inhibitor. In this part, DPC imaging technology was used to image NIH / 3T3 at different concentrations of SD-208 (50nM, 25nM, 10nM, 5nM, 1nM), and at the same time, immunofluorescence imaging of α-SMA was performed.

[0009] Step 4: Construction of high-content imaging method;

[0010] Step 5: Establishment of bright field digital phase contrast imaging (DPC) method;

[0011] Step 6: Establish image analysis method based on "machine learning" technology;

[0012] Step 7: Anti-fibrosis testing of 10 natural medicinal active compounds.

[0013] 1Resveratrol 501-36-0

[0014] 2Matairesinol 580-72-3

[0015] 3Catechol 120-80-9

[0016] 4Neo-przewaquinone A 630057-39-5

[0017] 5Salvianolic acid F 158732-59-3

[0018] 6Eupalinolide O 2170228-67-6

[0019] 7Platycodin D Opalinolide O 58479-68-8

[0020] 8Isolinderalactone957-66-4

[0021] 9Alantolactone546-43-0

[0022] 10Eupalinolide A 877822-40-7

[0023] Ten compounds at 1 μM, 5 μM, 10 μM and 20 μM were prepared and acted on NIH / 3T3 cells for 24 hours.

[0024] Furthermore, the high-throughput anti-fibrosis drug screening method based on the absence of fluorescence images and its application are characterized in that, in step one, 33 morphological parameters are analyzed using Harmony 4.9 software, and the 33 morphological parameters are respectively correlated with the fluorescence intensity of α-SMA, and parameters with correlation coefficients greater than 0.3 are selected. Through correlation analysis, 24 DPC parameters such as Cell Area and Cell Eadial Mean (r2>0.3) are correlated with α-SMA expression.

[0025] Furthermore, the high-throughput anti-fibrosis drug screening method based on the absence of fluorescence images and its application are characterized in that, in step 2, 33 morphological parameters are obtained by analysis using the Harmony 4.9 software that comes with high content; the 33 morphological parameters are respectively correlated with the fluorescence intensity of α-SMA, and the parameters with correlation coefficients greater than 0.3 are selected. Through correlation analysis, 19 DPC parameters such as Cell Area and Cell Eadial Mean (r2>0.3) are correlated with α-SMA expression.

[0026] In step three, 33 morphological parameters were obtained by analysis using the high-content built-in Harmony 4.9 software. The 33 morphological parameters were correlated with the fluorescence intensity of α-SMA, and the parameters with correlation coefficients greater than 0.3 were selected. Under the influence of SD-208, 16 morphological parameters such as the cell area of ​​DPC imaging cells and cell DPC SER right 0px were correlated with the fluorescence value of α-SMA (r2>0.3).

[0027] Furthermore, the high-throughput anti-fibrosis drug screening method based on the absence of fluorescence images and its application are characterized in that, according to the above screening criteria, after linear correlation analysis of DPC parameters under the influence of different concentrations of TGF-βI, different time of TGF-βI and different concentrations of target inhibitor SD-208, by analyzing the relationship between different groups of parameters (correlation coefficient greater than 0.3), a total of 13 DPC parameters are common disturbance parameters, including: Cell Area, Cell DigitalPhase Contrast SER Bright 0px, Cell Digital Phase Contrast SER Hole 0px, CellDigital Phase Contrast SER Ridge 0px, Cell Length, Cell Radial Mean, CellRoundness, Cell Symmetry 05, Cell Symmetry 15, Cell Threshold Compactness 30%, Cell Threshold Compactness 40%, Cell Threshold Compactness 50%, CellThreshold Compactness 60%, these parameters will be used for disturbance mechanism prediction;

[0028] Cell Area: An increase in cell area may indicate that cells are expanding, which is common during fibroblast activation and fibrosis.

[0029] Cell Eadial Mean: This refers to the radial mean of the cell, reflecting the average distribution of cell size and shape;

[0030] Cell Digital Phase Contrast SER: These parameters may be related to the complexity and irregularity of the cell edge and may reflect changes in cell activation and migration ability.

[0031] Furthermore, the high-throughput anti-fibrosis drug screening method based on the absence of fluorescent images and its application are characterized by experimental design;

[0032] TGF-βI induction experiment;

[0033] The researchers designed a series of experiments to evaluate the effects of different concentrations of TGF-βI on NIH / 3T3 cells (a commonly used mouse fibroblast cell line). By using different concentrations of TGF-βI, the potential effects of this factor in different pathological conditions in vivo can be simulated;

[0034] SD-208 inhibitor experiments;

[0035] As a selective TGF-βI inhibitor, SD-208 can block the signal transduction of TGF-βI. By observing the changes in cells after SD-208 treatment, we can have a deeper understanding of the role of the TGF-βI signaling pathway in changes in cell morphology and function.

[0036] Experimental methods;

[0037] Digital Phase-Pair (DPC) imaging;

[0038] DPC imaging technology provides a method to observe cell morphology without fluorescent labeling. This technology is particularly suitable for real-time monitoring of the dynamic process of cell response to various drugs or growth factors.

[0039] α-SMA immunofluorescence imaging;

[0040] α-Smooth muscle actin (α-SMA) is a marker of fibroblast activation, and its expression usually increases during fibrosis. Immunofluorescence staining can be used to quantitatively analyze the expression level of α-SMA and thus evaluate the activation state of cells.

[0041] Harmony 4.9 software analysis;

[0042] Harmony 4.9 software, part of high-content analysis (HCA), was used to extract morphological parameters from DPCs and fluorescence imaging, which can be correlated with biological properties of cells (such as α-SMA expression) to discover potential morphological markers.

[0043] Furthermore, the high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application are characterized in that, in step 4, the construction of the high-content imaging method:

[0044] (1) Processing method based on cell morphology analysis software parameters;

[0045] (2) The proposed deep learning and unbiased screening method combines the processing parameters of commercial software with an image-based neural network “machine learning” segmentation network, a preprocessing module, and a classification network to distinguish the activity of the compounds;

[0046] Extraction of cell morphology analysis software parameters:

[0047] This part uses Harmony 4.9 software for analysis; the high-content imaging system generates 3-channel output images, one channel for bright field imaging and the other three channels for fluorescence imaging; the software is used to count, and then the cell morphological parameters are extracted by the built-in algorithm; the DPC images of NIH / 3T3 cells are analyzed using Harmony 4.9 software to obtain 33 morphological parameters; these morphological parameters and a-SMA fluorescence intensity are analyzed for subsequent experiments; this part uses linear regression analysis to distinguish the correlation analysis between DPC images and a-SMA protein fluorescence values; linear correlation analysis is performed between DPC parameters.

[0048] “Machine Learning” Data Processing:

[0049] In this part, we use the Inception V4 neural network to classify images. The initial input is a 16-bit image containing a bright field channel and a DPC channel. We first resize and standardize each channel of the original image, then generate a third channel with the average of the existing two channels to form an RGB image, and input the 299×299×3 image into the Inception V4 network. The output dimension of the second-to-last fully connected layer is modified to 6, and the cell images are divided into 6 types. The Inception v4 neural network is pre-trained on the ImageNet object recognition dataset, which contains 1.2 million non-cell objects in 1000 categories. We use the pre-trained convolutional layers, connect them with randomly initialized fully connected layers, and train the resulting network on our own constructed dataset.

[0050] We used TGF-βI as the modeling agent and added different concentrations of TGF-βI (0.01μM, 0.10μM, 1μM, 5μM, 10, 20μM) to the wells where NIH / 3T3 cells were cultured. The treated cells were labeled as types 1-5, respectively. We defined type 0 and type 1 cells as normal cells, type 2 and type 3 cells as fibrosis process cells, and type 4 and type 5 cells as fibrotic cells. We established a dataset containing 24,000 labeled images, which were evenly distributed in six types. We randomly split the dataset with a training-validation ratio of 4:1, obtaining 19200 and 4800 images, respectively. We used Softmax_cross_entrophy as the loss function and Adam as the optimizer.

[0051] Furthermore, the high-throughput anti-fibrosis drug screening method based on the absence of fluorescence images and its application are characterized in that, in step four, DPC images are collected using the PerkinElmer high-content imaging analysis system Opera Phenix, and 13 morphological parameters are analyzed using the built-in Harmony 4.9 software, and the control group, model group, 1μM, 5μM, 10μM, and 20μM groups are clustered by principal component analysis (PCA) on the above-mentioned multidimensional parameters, and then α-SMA fluorescence images are collected. It can be seen that 29 compounds can reduce the expression of α-SMA, which fully proves the accuracy and effectiveness of this method as a screening anti-fibrosis screening method.

[0052] In step 4, experimental method;

[0053] (1) Cell culture: NIH / 3T3 mouse embryonic fibroblasts were purchased from the National Cell Identification Library of the Chinese Academy of Sciences. NIH / 3T3 cells were inoculated in Dulbecco's Modified Eagle's medium containing 10% newborn calf serum and 1% penicillin and streptomycin and cultured in a 37°C, 5% CO2 incubator.

[0054] (2) Plating and drug administration: After the cells were inoculated in the culture dish, the cells in the logarithmic growth phase were collected and added to a 96-well black transparent bottom culture plate at a concentration of 5000 cells per well, 100 μL per well, and cultured in a 37°C, 5% CO2 incubator for 24 h. Different concentrations of the test compound 1 μM, 5 μM, 10 μM, 20 μM and 10 ng / mL TGF-βI and a control group were added and cultured in a 37°C, 5% CO2 incubator for 24 h before subsequent operations.

[0055] (3) Extraction of cell morphology analysis software parameters: This part was analyzed using Harmony 4.9 software; the high-content imaging system was used to collect DPC channel images; the 13 morphological parameters selected by Harmony 4.9 software were collected and analyzed for DPC images of NIH / 3T3 cells; the C group, model group, 1μM, 5μM, 10μM, and 20μM groups were clustered using principal component analysis (PCA) to analyze the above multidimensional parameters.

[0056] (4) Immunofluorescence staining: NIH / 3T3 cells were fixed with 4% paraformaldehyde for 15 min; then, after washing three times in PBS, the cells were permeabilized with 0.1% Triton X-100 for 15 min, and the samples were blocked with 10% bovine serum albumin for 30 min; taken out; placed in primary antibody solution and incubated at 4°C overnight; after washing, the samples were incubated in FITC-conjugated secondary antibody solution at room temperature for 2 h; then, DAPI was used to stain the cell nucleus; finally, a high-content imaging and analysis system was used for image acquisition; the fluorescence intensity parameters of the α-SMA channel were acquired using Harmony 4.9 software;

[0057] Experimental results:

[0058] (1) The principal component analysis (PCA) clustering effects of 13 parameters of 10 compounds were all able to be distinguished from the model group;

[0059] (2) The α-SMA fluorescence intensity of the 10 compounds was significantly different from that of the model group and all of them could reduce the expression of α-SMA; this fully verified the feasibility and accuracy of the screening method using 13 parameters;

[0060] (3) IC50 toxicity tests were performed on these 10 compounds, and all compounds had anti-fibrotic effects at non-toxic doses.

[0061] Furthermore, the high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application are characterized in that, in step five, a bright field digital phase contrast imaging method is established:

[0062] Microscopic imaging has proven to be an essential tool for quantitatively monitoring biological processes with unprecedented spatial resolution; however, current analytical methods are limited in the number of measurable parameters or image feature extraction; HCS technology, which combines automated fluorescence microscopy with high-throughput automated acquisition technology, has become a powerful tool for studying cell biology and drug development; with significant advances in computing power, robotics and automation technology, as well as image detectors, the amount of information obtained through optical microscopy methods has greatly increased, and high-content imaging (HCS) has become a new and promising technology for the study of cell biology and drug development. Abstract: The application of HCI is becoming more and more extensive; the effectiveness of HCI in precision medicine has been confirmed in recent studies; differential phase contrast (DPC) is a non-interferometric quantitative phase imaging method achieved by an asymmetric imaging procedure; compared with most other methods, DPC is experimentally implemented without the need for fluorescent labels; in DPC, a spatially asymmetric imaging method is used to generate phase gradient images; this technology can observe living cells and analyze bright field images without the use of any fluorescent dyes, and is non-toxic; however, a machine learning approach is needed to extract more information from DPC images; in recent years, several methods have been proposed to predict fluorescence images from transmitted light images, such as bright field images and phase images; although the above methods show the possibility of using computational methods to transform biomarker images, no existing methods are as specific as our work and verify their applicability in HCS applications.

[0063] In this part, the processing parameters of DPC imaging were optimized by comparing the DPC imaging counting method with the cell nucleus fluorescence labeling method. Then, each well was imaged at 10×magnification using the Opera Phenix high-content analysis system equipped with a high-resolution 14-bit charge-coupled device (CCD) camera. The analysis was performed using Harmony 4.9 software, in which 9 images were collected for each well. The resolution of an image was 1360×1024 pixels, corresponding to an actual physical size of 0.88×0.66cm. The imaging system produced two-channel output images, one for bright field imaging and the other for fluorescence imaging. DPC imaging counts and Hoechst 33342 staining counts were compared at 0, 12, 24, and 36 h, respectively. The results showed that there was no significant difference between the two counting methods. Therefore, compared with the traditional Hoechst 33342 cell nucleus staining method, the established method does not require labeling.

[0064] A major advantage of phenotypic screening compared to target-based approaches is that disease-associated proteins are maintained in their natural environment, resulting in more physiological translational outcomes. In this section, deep machine deconvolution analyzed by Harmony 4.9 software outputs 31 cell morphological parameters during DPC imaging. In summary, more parameters can be obtained in a single screening experiment through deep analysis of image morphology.

[0065] (1) Parameter relationship under stimulation of different concentrations of TGF-βI:

[0066] The results showed that the DPC imaging parameters and α-SMA fluorescence imaging under the stimulation of different concentrations of TGF-βI, cell area, cell roundness, cell width, length and aspect ratio showed the same trend; at the same time, the fluorescence intensity of α-SMA increased with the increase of different concentrations of TGF-βI;

[0067] 33 morphological parameters of NIH / 3T3 cells were obtained by DPC image analysis. Through correlation analysis, 24 DPC parameters such as Cell Area and Cell Eadial Mean were correlated with α-SMA expression;

[0068] (2) DPC parameter analysis at different times:

[0069] Subsequently, at the same TGF-βI concentration (10 ng / mL), DPC imaging parameters and α-SMA fluorescence imaging were performed at different times; consistent with the above results, DPC imaging parameters showed the same trend in cell area, cell roundness, cell width, cell length and aspect ratio parameters of α-SMA fluorescence imaging at different times; the expression of α-SMA increased with the extension of administration time;

[0070] Subsequent experiments analyzed these morphological parameters and α-SMA fluorescence intensity; 33 morphological parameters of NIH-3T3 cells were obtained through DPC image analysis; through correlation analysis, 19 DPC parameters (r2>0.3) such as Cell Area and Cell Eadial Mean were correlated with α-SMA expression;

[0071] In step six, an image analysis method assisted by “machine learning” technology was established. In this part, we used a neural network with the Inception v4 architecture to classify images. The initial input was a 16-bit image containing a bright field channel and a DPC channel. We first resized and standardized each channel of the original image, then generated a third channel using the average of the existing two channels to form an RGB (red-green-blue) image, and input the 299×299×3 image into the Inception v4 network. The output dimension of the penultimate fully connected layer was modified to 6, and the cell images were classified into 6 types. The Inception v4 architecture neural network was pre-trained on the ImageNet object recognition dataset, which contains 1.2 million non-cell objects in 1,000 categories. We used pre-trained convolutional layers, connected them with randomly initialized fully connected layers, and trained the resulting network on our own constructed dataset.

[0072] Using TGF-βI as the modeling agent, different concentrations of TGF-βI (0.01μM, 0.10μM, 1μM, 5μM, 10, 20μM) were added to the wells where NIH-3t3 cells were cultured, and the treated cells were labeled as types 1-5 respectively; we defined type 0 and type 1 cells as normal cells, type 2 and type 3 cells as fibrosis process-arrested cells, and type 4 and type 5 cells as fibrotic cells; a dataset containing 24,000 labeled images was established, which were evenly distributed in six types; we randomly split the dataset with a training-validation ratio of 4:1, obtaining 19200 and 4800 respectively; Softmax_cross_entrophy was used as the loss function, and Adam was used as the optimizer.

[0073] After 120 epochs of training, the network loss converged and the accuracy of both the training and validation sets reached 95%; we then tested the model on a test set of 4,200 labeled images, which were independent of the training and validation sets conducted by different operators; the model performed well in predicting non-training data, with a calculated accuracy of approximately 92.30%; the images of the above 1,400 compounds were distinguished using the above machine learning model, and 1,400 × 6 wells × 9 fields of view = 75,600 images were processed.

[0074] Furthermore, the high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application are characterized in that, in step seven, the targeted inhibitor intervenes:

[0075] SD-208 is a selective TGF-βI inhibitor. This part uses DPC imaging technology to image NIH / 3T3 at different concentrations of SD-208 (50nM, 25nM, 10nM, 5nM, 1nM). At the same time, immunofluorescence imaging of α-SMA was performed; as the concentration of SD-208 increased, the fluorescence intensity of α-SMA also decreased, indicating that SD-208 affected the expression of α-SMA protein.

[0076] This part also collected the morphological parameters of NIH / 3T3 cell DPC imaging under the interference of different concentrations of SD-208; under the influence of SD-208, 16 morphological parameters such as cell area and cell DPC SER right 0px of DPC imaging cells were correlated with the fluorescence value of α-SMA.

[0077] In summary, we conducted a linear correlation analysis of DPC parameters under the influence of different concentrations of TGF-βI, the same TGF-βI at different times, and the target inhibitor SD-208. By analyzing the relationship between different groups of parameters (correlation coefficient greater than 0.3), a total of 13 DPC parameters were common perturbation parameters; these parameters will be used to predict the perturbation mechanism.

[0078] According to claim 1, a high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application, characterized in that the anti-fibrosis degree of DPC images is judged based on unbiased automatic deep machine learning:

[0079] In order to process DPC image data more deeply and integrate the image recognition method of image deep learning into our platform; TGF-βI was used as a modeling agent, and the cell images treated with different concentrations of TGF-βI were labeled as types 0-5; we labeled 2000 images for each type. By rotating the augmented data, a final dataset of 48,000 was obtained; the classification network based on Inception V4 was fine-tuned on 48,000 images.

[0080] After 120 epochs of training, the network loss converged and the accuracy of both the training set and the validation set reached 95%; then, we tested the model on a test set of 4200 labeled images, which were independent of the training and validation sets conducted by different operators; the model showed high accuracy in predicting non-training data, with a calculated accuracy of approximately 92.30%; the images of the above 1400 compounds were distinguished using the above machine learning model, and 1400×6 wells×9 fields of view=75600 images were processed.

[0081] Integrating deep learning and unbiased automated discovery of anti-fibrotic lead compounds:

[0082] In summary, through two parts of experiments, 33 lead compounds were screened out by morphological parameters, and 65 lead compounds were screened out by unbiased image learning method; 29 lead compounds were the intersection of the two screening modes; the names and CAS numbers of the 29 compounds are listed in Tables 1-7.

[0083] The number of compounds obtained by the image-based screening mode is greater than that of compounds screened based on morphological parameters. This may be because the results of compound screening based on morphological parameters are overly dependent on phenotypic morphological parameters. Compared with the morphological screening mode, image-based deep learning is an unbiased screening result.

[0084] The above results show that a total of 29 lead compounds can significantly inhibit abnormal proliferation of fibroblasts. It is important to maintain the original state of fibroblasts. This result is in line with our purpose of screening anti-fibrosis lead compounds.

[0085] In addition, the efficacy of the screening results was reconfirmed; immunofluorescence staining was used to verify that the 29 screened compounds all had the ability to downregulate the expression of α-SMA protein; the results showed that all 29 lead compounds inhibited the expression of α-SMA protein, that is, the fluorescence value of α-SMA in the compound group was significantly lower than that in the model group.

[0086] In summary, we rapidly identified 29 TCM compounds from 1400 TCM compounds; these 29 compounds significantly maintained the morphology of fibrocytes and inhibited abnormal fibrocyte proliferation.

[0087] The present invention has the following beneficial effects:

[0088] 1. In the present invention, by performing correlation analysis between these morphological parameters and α-SMA expression levels, researchers can quickly identify cell changes that are closely related to the fibrosis process. This rapid screening method significantly improves the initial screening efficiency of anti-fibrosis drugs and reduces experimental costs. Traditional fluorescent staining technology is prone to errors during operation, while DPC imaging technology provides a more stable and easy-to-quantify analysis method, which reduces subjectivity and errors in the experimental process and improves the accuracy and reliability of the data.

[0089] 2. In the present invention, all 10 compounds can reduce the expression of α-SMA to varying degrees, indicating that they have anti-fibrosis potential. Through the high-content imaging and analysis system and Harmony4.9 software, researchers can quickly screen out compounds with potential therapeutic effects and further understand their specific effects on cell morphology. This method proves its accuracy and effectiveness as an anti-fibrosis drug screening, and provides important experimental basis and technical support for future drug development and treatment of fibrotic diseases. In addition, this high-throughput screening method can accelerate the discovery and development process of new drugs and provide new strategies for disease treatment.

[0090] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0091] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0092] Figure 1 is the correlation between DPC parameters and α-SMA expression under stimulation of different concentrations of TGF-βI;

[0093] Figure 2 is the correlation of DPC parameter α-SMA fluorescence intensity under different stimulation time;

[0094] Figure 3 Linear correlation analysis of DPC parameters under the influence of different concentrations of target inhibitor SD-208

[0095] Figure 4 DPC imaging and α-SMA fluorescence imaging under stimulation of different concentrations of TGF-βI;

[0096] Figure 5 DPC imaging and α-SMA fluorescence imaging under TGF-βI stimulation at different times;

[0097] Figure 6 DPC imaging and α-SMA fluorescence imaging under the intervention of different concentrations of SD-208;

[0098] Figure 7 is the relationship between different groups of parameters;

[0099] Figure 8 Image processing with unbiased automatic deep machine learning

[0100] Fig. 9There are 13 common perturbation parameters;

[0101] Fig.10 This is the principal component analysis diagram of 13 parameters of 10 compounds;

[0102] Fig.11 The difference diagram of α-SMA fluorescence intensity of 10 compounds and the model group;

[0103] Fig.12 This is a trend chart of IC50 toxicity test results of 10 compounds. DETAILED DESCRIPTION

[0104] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0105] See also Figure 1-11 As shown, the present invention is a high-throughput anti-fibrosis drug screening method based on no fluorescence image and its application, comprising the following steps;

[0106] Step 1: DPC imaging and fluorescence imaging under stimulation of different concentrations of TGF-βI (0.1ng / ml, 0.5ng / ml, 1ng / ml, 5ng / ml, 10ng / ml);

[0107] Step 2: Under the same TGF-βI concentration (10 ng / mL), DPC imaging parameters and α-SMA fluorescence imaging were collected at different times (3 h, 6 h, 12 h, 24 h, and 36 h);

[0108] Step 3: SD-208 is a selective TGF-βI inhibitor. In this part, DPC imaging technology was used to image NIH / 3T3 at different concentrations of SD-208 (50nM, 25nM, 10nM, 5nM, 1nM), and at the same time, immunofluorescence imaging of α-SMA was performed.

[0109] Step 4: Construction of high-content imaging method;

[0110] Step 5: Establishment of bright field digital phase contrast imaging (DPC) method;

[0111] Step 6: Establish image analysis method based on "machine learning" technology;

[0112] Step 7: Anti-fibrosis testing of 10 natural medicinal active compounds.

[0113] 1Resveratrol 501-36-0

[0114] 2Matairesinol 580-72-3

[0115] 3Catechol 120-80-9

[0116] 4Neo-przewaquinone A 630057-39-5

[0117] 5Salvianolic acid F 158732-59-3

[0118] 6Eupalinolide O 2170228-67-6

[0119] 7Platycodin D Opalinolide O 58479-68-8

[0120] 8Isolinderalactone957-66-4

[0121] 9Alantolactone546-43-0

[0122] 10Eupalinolide A 877822-40-7

[0123] Ten compounds at 1 μM, 5 μM, 10 μM and 20 μM were prepared and acted on NIH / 3T3 cells for 24 hours.

[0124] A high-throughput anti-fibrosis drug screening method based on the absence of fluorescence images and its application, characterized in that, in step one, 33 morphological parameters are obtained by analysis using Harmony 4.9 software, and correlation analysis is performed between the 33 morphological parameters and the fluorescence intensity of α-SMA, and parameters with correlation coefficients greater than 0.3 are selected. Through correlation analysis, 24 DPC parameters such as Cell Area and Cell Eadial Mean (r2>0.3) are correlated with α-SMA expression.

[0125] A high-throughput anti-fibrosis drug screening method based on the absence of fluorescence images and its application, characterized in that, in step 2, 33 morphological parameters are obtained by analysis using the Harmony 4.9 software that comes with high content; the 33 morphological parameters are respectively correlated with the fluorescence intensity of α-SMA, and the parameters with correlation coefficients greater than 0.3 are selected. Through correlation analysis, 19 DPC parameters such as Cell Area and Cell Eadial Mean (r2>0.3) are correlated with α-SMA expression.

[0126] In step three, 33 morphological parameters were obtained by analysis using the high-content built-in Harmony 4.9 software. The 33 morphological parameters were correlated with the fluorescence intensity of α-SMA, and the parameters with correlation coefficients greater than 0.3 were selected. Under the influence of SD-208, 16 morphological parameters such as the cell area of ​​DPC imaging cells and cell DPC SER right 0px were correlated with the fluorescence value of α-SMA (r2>0.3).

[0127] A high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for images and its application, characterized in that, according to the above screening criteria, after linear correlation analysis of DPC parameters under the influence of different concentrations of TGF-βI, different time periods of TGF-βI and different concentrations of target inhibitor SD-208, by analyzing the relationship between different groups of parameters (correlation coefficient greater than 0.3), a total of 13 DPC parameters are common disturbance parameters, including: Cell Area, Cell Digital Phase ContrastSER Bright 0px, Cell Digital Phase Contrast SER Hole 0px, Cell Digital PhaseContrast SER Ridge 0px, Cell Length, Cell Radial Mean, Cell Roundness, CellSymmetry 05, Cell Symmetry 15, Cell Threshold Compactness 30%, Cell ThresholdCompactness 40%, Cell Threshold Compactness 50%, Cell Threshold Compactness60%, these parameters will be used for disturbance mechanism prediction;

[0128] Cell Area: An increase in cell area may indicate that cells are expanding, which is common during fibroblast activation and fibrosis.

[0129] Cell Eadial Mean: This refers to the radial mean of the cell, reflecting the average distribution of cell size and shape;

[0130] Cell Digital Phase Contrast SER: These parameters may be related to the complexity and irregularity of the cell edge and may reflect changes in cell activation and migration ability.

[0131] A high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application, characterized in that the experimental design;

[0132] TGF-βI induction experiment;

[0133] The researchers designed a series of experiments to evaluate the effects of different concentrations of TGF-βI on NIH / 3T3 cells (a commonly used mouse fibroblast cell line). By using different concentrations of TGF-βI, the potential effects of this factor in different pathological conditions in vivo can be simulated;

[0134] SD-208 inhibitor experiments;

[0135] As a selective TGF-βI inhibitor, SD-208 can block the signal transduction of TGF-βI. By observing the changes in cells after SD-208 treatment, we can have a deeper understanding of the role of the TGF-βI signaling pathway in changes in cell morphology and function.

[0136] Experimental methods;

[0137] Digital Phase-Pair (DPC) imaging;

[0138] DPC imaging technology provides a method to observe cell morphology without fluorescent labeling. This technology is particularly suitable for real-time monitoring of the dynamic process of cell response to various drugs or growth factors.

[0139] α-SMA immunofluorescence imaging;

[0140] α-Smooth muscle actin (α-SMA) is a marker of fibroblast activation, and its expression usually increases during fibrosis. Immunofluorescence staining can be used to quantitatively analyze the expression level of α-SMA and thus evaluate the activation state of cells.

[0141] Harmony 4.9 software analysis;

[0142] Harmony 4.9 software, part of high-content analysis (HCA), was used to extract morphological parameters from DPCs and fluorescence imaging, which can be correlated with biological properties of cells (such as α-SMA expression) to discover potential morphological markers.

[0143] A high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application, characterized in that in step 4, the construction of a high-content imaging method:

[0144] (1) Processing method based on cell morphology analysis software parameters;

[0145] (2) The proposed deep learning and unbiased screening method combines the processing parameters of commercial software with an image-based neural network “machine learning” segmentation network, a preprocessing module, and a classification network to distinguish the activity of the compounds;

[0146] Extraction of cell morphology analysis software parameters:

[0147] This part uses Harmony 4.9 software for analysis; the high-content imaging system generates 3-channel output images, one channel for bright field imaging and the other three channels for fluorescence imaging; the software is used to count, and then the cell morphological parameters are extracted by the built-in algorithm; the DPC images of NIH / 3T3 cells are analyzed using Harmony 4.9 software to obtain 33 morphological parameters; these morphological parameters and a-SMA fluorescence intensity are analyzed for subsequent experiments; this part uses linear regression analysis to distinguish the correlation analysis between DPC images and a-SMA protein fluorescence values; linear correlation analysis is performed between DPC parameters.

[0148] “Machine Learning” Data Processing:

[0149] In this part, we use the Inception V4 neural network to classify images. The initial input is a 16-bit image containing a bright field channel and a DPC channel. We first resize and standardize each channel of the original image, then generate a third channel with the average of the existing two channels to form an RGB image, and input the 299×299×3 image into the Inception V4 network. The output dimension of the second-to-last fully connected layer is modified to 6, and the cell images are divided into 6 types. The Inception v4 neural network is pre-trained on the ImageNet object recognition dataset, which contains 1.2 million non-cell objects in 1000 categories. We use the pre-trained convolutional layers, connect them with randomly initialized fully connected layers, and train the resulting network on our own constructed dataset.

[0150] We used TGF-βI as the modeling agent and added different concentrations of TGF-βI (0.01μM, 0.10μM, 1μM, 5μM, 10, 20μM) to the wells where NIH / 3T3 cells were cultured. The treated cells were labeled as types 1-5, respectively. We defined type 0 and type 1 cells as normal cells, type 2 and type 3 cells as fibrosis process cells, and type 4 and type 5 cells as fibrotic cells. We established a dataset containing 24,000 labeled images, which were evenly distributed in six types. We randomly split the dataset with a training-validation ratio of 4:1, obtaining 19200 and 4800 images, respectively. We used Softmax_cross_entrophy as the loss function and Adam as the optimizer.

[0151] A high-throughput anti-fibrosis drug screening method based on the absence of fluorescence images and its application, characterized in that, in step 4, DPC images are collected using the PerkinElmer high-content imaging analysis system Opera Phenix, and 13 morphological parameters are analyzed using the built-in Harmony 4.9 software, and the control group, model group, 1μM, 5μM, 10μM, and 20μM groups are clustered by principal component analysis (PCA) on the above multidimensional parameters, and then α-SMA fluorescence images are collected, and it can be seen that 29 compounds can reduce the expression of α-SMA, which fully proves the accuracy and effectiveness of this method as a screening anti-fibrosis screening method.

[0152] In step 4, experimental method;

[0153] (1) Cell culture: NIH / 3T3 mouse embryonic fibroblasts were purchased from the National Cell Identification Library of the Chinese Academy of Sciences. NIH / 3T3 cells were inoculated in Dulbecco's Modified Eagle's medium containing 10% newborn calf serum and 1% penicillin and streptomycin and cultured in a 37°C, 5% CO2 incubator.

[0154] (2) Plating and drug administration: After the cells were inoculated in the culture dish, the cells in the logarithmic growth phase were collected and added to a 96-well black transparent bottom culture plate at a concentration of 5000 cells per well, 100 μL per well, and cultured in a 37°C, 5% CO2 incubator for 24 h. Different concentrations of the test compound 1 μM, 5 μM, 10 μM, 20 μM and 10 ng / mL TGF-βI and a control group were added and cultured in a 37°C, 5% CO2 incubator for 24 h before subsequent operations.

[0155] (3) Extraction of cell morphology analysis software parameters: This part was analyzed using Harmony 4.9 software; the high-content imaging system was used to collect DPC channel images; the 13 morphological parameters selected by Harmony 4.9 software were collected and analyzed for DPC images of NIH / 3T3 cells; the C group, model group, 1μM, 5μM, 10μM, and 20μM groups were clustered using principal component analysis (PCA) to analyze the above multidimensional parameters.

[0156] (4) Immunofluorescence staining: NIH / 3T3 cells were fixed with 4% paraformaldehyde for 15 min; then, after washing three times in PBS, the cells were permeabilized with 0.1% Triton X-100 for 15 min, and the samples were blocked with 10% bovine serum albumin for 30 min; taken out; placed in primary antibody solution and incubated at 4°C overnight; after washing, the samples were incubated in FITC-conjugated secondary antibody solution at room temperature for 2 h; then, DAPI was used to stain the cell nucleus; finally, a high-content imaging and analysis system was used for image acquisition; the fluorescence intensity parameters of the α-SMA channel were acquired using Harmony 4.9 software;

[0157] Experimental results:

[0158] (1) The principal component analysis (PCA) clustering effects of 13 parameters of 10 compounds were all able to be distinguished from the model group;

[0159] (2) The α-SMA fluorescence intensity of the 10 compounds was significantly different from that of the model group and all of them could reduce the expression of α-SMA; this fully verified the feasibility and accuracy of the screening method using 13 parameters;

[0160] (3) IC50 toxicity tests were performed on these 10 compounds, and all compounds had anti-fibrotic effects at non-toxic doses.

[0161] A high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application, characterized in that in step five, a bright field digital phase contrast imaging method is established:

[0162] Microscopic imaging has proven to be an essential tool for quantitatively monitoring biological processes with unprecedented spatial resolution; however, current analytical methods are limited in the number of measurable parameters or image feature extraction; HCS technology, which combines automated fluorescence microscopy with high-throughput automated acquisition technology, has become a powerful tool for studying cell biology and drug development; with significant advances in computing power, robotics and automation technology, as well as image detectors, the amount of information obtained through optical microscopy methods has greatly increased, and high-content imaging (HCS) has become a new and promising technology for the study of cell biology and drug development. Abstract: The application of HCI is becoming more and more extensive; the effectiveness of HCI in precision medicine has been confirmed in recent studies; differential phase contrast (DPC) is a non-interferometric quantitative phase imaging method achieved by an asymmetric imaging procedure; compared with most other methods, DPC is experimentally implemented without the need for fluorescent labels; in DPC, a spatially asymmetric imaging method is used to generate phase gradient images; this technology can observe living cells and analyze bright field images without the use of any fluorescent dyes, and is non-toxic; however, a machine learning approach is needed to extract more information from DPC images; in recent years, several methods have been proposed to predict fluorescence images from transmitted light images, such as bright field images and phase images; although the above methods show the possibility of using computational methods to transform biomarker images, no existing methods are as specific as our work and verify their applicability in HCS applications.

[0163] In this part, the processing parameters of DPC imaging were optimized by comparing the DPC imaging counting method with the cell nucleus fluorescence labeling method. Then, each well was imaged at 10×magnification using the Opera Phenix high-content analysis system equipped with a high-resolution 14-bit charge-coupled device (CCD) camera. The analysis was performed using Harmony 4.9 software, in which 9 images were collected for each well. The resolution of an image was 1360×1024 pixels, corresponding to an actual physical size of 0.88×0.66cm. The imaging system produced two-channel output images, one for bright field imaging and the other for fluorescence imaging. DPC imaging counts and Hoechst 33342 staining counts were compared at 0, 12, 24, and 36 h, respectively. The results showed that there was no significant difference between the two counting methods. Therefore, compared with the traditional Hoechst 33342 cell nucleus staining method, the established method does not require labeling.

[0164] A major advantage of phenotypic screening compared to target-based approaches is that disease-associated proteins are maintained in their natural environment, resulting in more physiological translational outcomes. In this section, deep machine deconvolution analyzed by Harmony 4.9 software outputs 31 cell morphological parameters during DPC imaging. In summary, more parameters can be obtained in a single screening experiment through deep analysis of image morphology.

[0165] (1) Parameter relationship under stimulation of different concentrations of TGF-βI:

[0166] The results showed that the DPC imaging parameters and α-SMA fluorescence imaging under the stimulation of different concentrations of TGF-βI, cell area, cell roundness, cell width, length and aspect ratio showed the same trend; at the same time, the fluorescence intensity of α-SMA increased with the increase of different concentrations of TGF-βI;

[0167] 33 morphological parameters of NIH / 3T3 cells were obtained by DPC image analysis. Through correlation analysis, 24 DPC parameters such as Cell Area and Cell Eadial Mean were correlated with α-SMA expression;

[0168] (2) DPC parameter analysis at different times:

[0169] Subsequently, at the same TGF-βI concentration (10 ng / mL), DPC imaging parameters and α-SMA fluorescence imaging were performed at different times; consistent with the above results, DPC imaging parameters showed the same trend in cell area, cell roundness, cell width, cell length and aspect ratio parameters of α-SMA fluorescence imaging at different times; the expression of α-SMA increased with the extension of administration time;

[0170] Subsequent experiments analyzed these morphological parameters and α-SMA fluorescence intensity; 33 morphological parameters of NIH-3T3 cells were obtained through DPC image analysis; through correlation analysis, 19 DPC parameters (r2>0.3) such as Cell Area and Cell Eadial Mean were correlated with α-SMA expression;

[0171] In step six, an image analysis method assisted by “machine learning” technology was established. In this part, we used a neural network with the Inception v4 architecture to classify images. The initial input was a 16-bit image containing a bright field channel and a DPC channel. We first resized and standardized each channel of the original image, then generated a third channel using the average of the existing two channels to form an RGB (red-green-blue) image, and input the 299×299×3 image into the Inception v4 network. The output dimension of the penultimate fully connected layer was modified to 6, and the cell images were classified into 6 types. The Inception v4 architecture neural network was pre-trained on the ImageNet object recognition dataset, which contains 1.2 million non-cell objects in 1,000 categories. We used pre-trained convolutional layers, connected them with randomly initialized fully connected layers, and trained the resulting network on our own constructed dataset.

[0172] Using TGF-βI as the modeling agent, different concentrations of TGF-βI (0.01μM, 0.10μM, 1μM, 5μM, 10, 20μM) were added to the wells where NIH-3t3 cells were cultured, and the treated cells were labeled as types 1-5 respectively; we defined type 0 and type 1 cells as normal cells, type 2 and type 3 cells as fibrosis process-arrested cells, and type 4 and type 5 cells as fibrotic cells; a dataset containing 24,000 labeled images was established, which were evenly distributed in six types; we randomly split the dataset with a training-validation ratio of 4:1, obtaining 19200 and 4800 respectively; Softmax_cross_entrophy was used as the loss function, and Adam was used as the optimizer.

[0173] After 120 epochs of training, the network loss converged and the accuracy of both the training and validation sets reached 95%; we then tested the model on a test set of 4,200 labeled images, which were independent of the training and validation sets conducted by different operators; the model performed well in predicting non-training data, with a calculated accuracy of approximately 92.30%; the images of the above 1,400 compounds were distinguished using the above machine learning model, and 1,400 × 6 wells × 9 fields of view = 75,600 images were processed.

[0174] A high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application, characterized in that in step seven, the targeted inhibitor intervenes:

[0175] SD-208 is a selective TGF-βI inhibitor. This part uses DPC imaging technology to image NIH / 3T3 at different concentrations of SD-208 (50nM, 25nM, 10nM, 5nM, 1nM). At the same time, immunofluorescence imaging of α-SMA was performed; as the concentration of SD-208 increased, the fluorescence intensity of α-SMA also decreased, indicating that SD-208 affected the expression of α-SMA protein.

[0176] This part also collected the morphological parameters of NIH / 3T3 cell DPC imaging under the interference of different concentrations of SD-208; under the influence of SD-208, 16 morphological parameters such as cell area and cell DPC SER right 0px of DPC imaging cells were correlated with the fluorescence value of α-SMA.

[0177] In summary, we conducted a linear correlation analysis of DPC parameters under the influence of different concentrations of TGF-βI, the same TGF-βI at different times, and the target inhibitor SD-208. By analyzing the relationship between different groups of parameters (correlation coefficient greater than 0.3), a total of 13 DPC parameters were common perturbation parameters; these parameters will be used to predict the perturbation mechanism.

[0178] According to claim 1, a high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application, characterized in that the anti-fibrosis degree of DPC images is judged based on unbiased automatic deep machine learning:

[0179] In order to process DPC image data more deeply and integrate the image recognition method of image deep learning into our platform; TGF-βI was used as a modeling agent, and the cell images treated with different concentrations of TGF-βI were labeled as types 0-5; we labeled 2000 images for each type. By rotating the augmented data, a final dataset of 48,000 was obtained; the classification network based on Inception V4 was fine-tuned on 48,000 images.

[0180] After 120 epochs of training, the network loss converged and the accuracy of both the training set and the validation set reached 95%; then, we tested the model on a test set of 4200 labeled images, which were independent of the training and validation sets conducted by different operators; the model showed high accuracy in predicting non-training data, with a calculated accuracy of approximately 92.30%; the images of the above 1400 compounds were distinguished using the above machine learning model, and 1400×6 wells×9 fields of view=75600 images were processed.

[0181] Integrating deep learning and unbiased automated discovery of anti-fibrotic lead compounds:

[0182] In summary, through two parts of experiments, 33 lead compounds were screened out by morphological parameters, and 65 lead compounds were screened out by unbiased image learning method; 29 lead compounds were the intersection of the two screening modes; the names and CAS numbers of the 29 compounds are listed in Tables 1-7.

[0183] The number of compounds obtained by the image-based screening mode is greater than that of compounds screened based on morphological parameters. This may be because the results of compound screening based on morphological parameters are overly dependent on phenotypic morphological parameters. Compared with the morphological screening mode, image-based deep learning is an unbiased screening result.

[0184] The above results show that a total of 29 lead compounds can significantly inhibit abnormal proliferation of fibroblasts. It is important to maintain the original state of fibroblasts. This result is in line with our purpose of screening anti-fibrosis lead compounds.

[0185] In addition, the efficacy of the screening results was reconfirmed; immunofluorescence staining was used to verify that the 29 screened compounds all had the ability to downregulate the expression of α-SMA protein; the results showed that all 29 lead compounds inhibited the expression of α-SMA protein, that is, the fluorescence value of α-SMA in the compound group was significantly lower than that in the model group.

[0186] In summary, we rapidly identified 29 TCM compounds from 1400 TCM compounds; these 29 compounds significantly maintained the morphology of fibrocytes and inhibited abnormal fibrocyte proliferation.

[0187] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0188] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application, characterized in that: comprising the following steps; Step 1: DPC imaging and fluorescence imaging under stimulation of different concentrations of TGF-βI (0.1ng / ml, 0.5ng / ml, 1ng / ml, 5ng / ml, 10ng / ml); Step 2: Under the same TGF-βI concentration (10 ng / mL), DPC imaging parameters and α-SMA fluorescence imaging were collected at different times (3 h, 6 h, 12 h, 24 h, and 36 h); Step 3: SD-208 is a selective TGF-βI inhibitor. In this part, DPC imaging technology was used to image NIH / 3T3 at different concentrations of SD-208 (50nM, 25nM, 10nM, 5nM, 1nM), and at the same time, immunofluorescence imaging of α-SMA was performed. Step 4: Construction of high-content imaging method; Step 5: Establishment of bright field digital phase contrast imaging (DPC) method; Step 6: Establish image analysis method based on "machine learning" technology; Step 7: Anti-fibrosis detection of 29 natural drug active compounds. 1 Resveratrol 501-36-0 2 Matairesinol 580-72-3 3 Catechol 120-80-9 4 Neo-przewaquinone A 630057-39-5 5 Salvianolic acid F 158732-59-3 6 Eupalinolide O 2170228-67-6 7 Platycodin D Opalinolide O 58479-68-8 8 Isolinderalactone957-66-4 9 Alantolactone546-43-0 10 Eupalinolide A 877822-40-7 Ten compounds at 1 μM, 5 μM, 10 μM and 20 μM were prepared and acted on NIH / 3T3 cells for 24 hours.

2. A high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application according to claim 1, characterized in that: In step 1, 33 morphological parameters were obtained by analysis using Harmony 4.9 software. The 33 morphological parameters were correlated with the fluorescence intensity of α-SMA, and the parameters with correlation coefficients greater than 0.3 were selected. Through correlation analysis, 24 DPC parameters such as Cell Area and Cellular Mean (r2>0.3) were correlated with α-SMA expression.

3. The high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application according to claim 1, characterized in that: In step 2, 33 morphological parameters were obtained by analysis using the high-content built-in Harmony4.9 software; the 33 morphological parameters were correlated with the fluorescence intensity of α-SMA, and the parameters with correlation coefficients greater than 0.3 were selected. Through correlation analysis, 19 DPC parameters such as Cell Area and Cell Eadial Mean (r2>0.3) were correlated with α-SMA expression. In step three, 33 morphological parameters were obtained by analysis using the high-content built-in Harmony 4.9 software. The 33 morphological parameters were correlated with the fluorescence intensity of α-SMA, and the parameters with correlation coefficients greater than 0.3 were selected. Under the influence of SD-208, 16 morphological parameters such as the cell area of ​​DPC imaging cells and cell DPC SER right 0px were correlated with the fluorescence value of α-SMA (r2>0.3).

4. The high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application according to claim 1, characterized in that: According to the above screening criteria, after linear correlation analysis of DPC parameters under the influence of different concentrations of TGF-βI, different time of TGF-βI and different concentrations of target inhibitor SD-208, by analyzing the relationship between different groups of parameters (correlation coefficient greater than 0.3), a total of 13 DPC parameters are common perturbation parameters, including: Cell Area, Cell Digital Phase Contrast SER Bright 0px, Cell Digital Phase Contrast SER Hole 0px, Cell Digital Phase Contrast SER Ridge 0px, Cell Length, Cell Radial Mean, Cell Roundness, Cell Symmetry 05, Cell Symmetry 15, Cell Threshold Compactness 30%, Cell Threshold Compactness 40%, Cell Threshold Compactness 50%, Cell Threshold Compactness 60%. These parameters will be used for perturbation mechanism prediction; Cell Area: An increase in cell area may indicate that cells are expanding, which is common during fibroblast activation and fibrosis. Cell Eadial Mean: This refers to the radial mean of the cell, reflecting the average distribution of cell size and shape; Cell Digital Phase Contrast SER: These parameters may be related to the complexity and irregularity of the cell edge and may reflect changes in cell activation and migration ability.

5. The high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application according to claim 1, characterized in that: Experimental design; TGF-βI induction experiment; The researchers designed a series of experiments to evaluate the effects of different concentrations of TGF-βI on NIH / 3T3 cells (a commonly used mouse fibroblast cell line). By using different concentrations of TGF-βI, the potential effects of this factor in different pathological conditions in vivo can be simulated; SD-208 inhibitor experiments; As a selective TGF-βI inhibitor, SD-208 can block the signal transduction of TGF-βI. By observing the changes in cells after SD-208 treatment, we can have a deeper understanding of the role of the TGF-βI signaling pathway in changes in cell morphology and function. Experimental methods; Digital Phase-Channel (DPC) imaging; DPC imaging technology provides a method to observe cell morphology without fluorescent labeling. This technology is particularly suitable for real-time monitoring of the dynamic process of cell response to various drugs or growth factors. α-SMA immunofluorescence imaging; α-Smooth muscle actin (α-SMA) is a marker of fibroblast activation, and its expression usually increases during fibrosis. Immunofluorescence staining can be used to quantitatively analyze the expression level of α-SMA and thus evaluate the activation state of cells. Harmony 4.9 software analysis; Harmony 4.9 software, part of high-content analysis (HCA), was used to extract morphological parameters from DPCs and fluorescence imaging, which can be correlated with biological properties of cells (such as α-SMA expression) to discover potential morphological markers.

6. The high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application according to claim 1, characterized in that: In step 4, the construction of high-content imaging method: (1) Processing method based on cell morphology analysis software parameters; (2) The proposed deep learning and unbiased screening method combines the processing parameters of commercial software with an image-based neural network "machine learning" segmentation network, preprocessing module and classification network to distinguish the activity of compounds; Extraction of cell morphology analysis software parameters: This part uses Harmony 4.9 software for analysis; the high-content imaging system generates 3-channel output images, one channel for bright field imaging and the other three channels for fluorescence imaging; the software is used to count, and then the cell morphological parameters are extracted by the built-in algorithm; the DPC images of NIH / 3T3 cells are analyzed using Harmony 4.9 software to obtain 33 morphological parameters; these morphological parameters and a-SMA fluorescence intensity are analyzed for subsequent experiments; this part uses linear regression analysis to distinguish the correlation analysis between DPC images and a-SMA protein fluorescence values; linear correlation analysis is performed between DPC parameters. "Machine Learning" Data Processing: In this part, we use the Inception V4 neural network to classify images. The initial input is a 16-bit image containing a bright field channel and a DPC channel. We first resize and standardize each channel of the original image, then generate a third channel with the average of the existing two channels to form an RGB image, and input the 299×299×3 image into the Inception V4 network. The output dimension of the second-to-last fully connected layer is modified to 6, and the cell images are divided into 6 types. The Inception v4 neural network is pre-trained on the ImageNet object recognition dataset, which contains 1.2 million non-cell objects in 1000 categories. We use the pre-trained convolutional layers, connect them with randomly initialized fully connected layers, and train the resulting network on our own constructed dataset. We used TGF-βI as the modeling agent and added different concentrations of TGF-βI (0.01μM, 0.10μM, 1μM, 5μM, 10, 20μM) to the wells where NIH / 3T3 cells were cultured. The treated cells were labeled as types 1-5, respectively. We defined type 0 and type 1 cells as normal cells, type 2 and type 3 cells as fibrosis process cells, and type 4 and type 5 cells as fibrotic cells. We established a dataset containing 24,000 labeled images, which were evenly distributed in six types. We randomly split the dataset with a training-validation ratio of 4:1, obtaining 19200 and 4800 images, respectively. We used Softmax_cross_entrophy as the loss function and Adam as the optimizer.

7. The high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application according to claim 1, characterized in that: In step 4, the DPC images were collected using the PerkinElmer high-content imaging analysis system Opera Phenix, and 13 morphological parameters were analyzed using the built-in Harmony 4.9 software. The control group, model group, 1μM, 5μM, 10μM, and 20μM groups were clustered using principal component analysis (PCA) to analyze the above multidimensional parameters, and then α-SMA fluorescence images were collected. It can be seen that all 29 compounds can reduce the expression of α-SMA, which fully demonstrates the accuracy and effectiveness of this method as an anti-fibrosis screening method. In step 4, experimental method; (1) Cell culture: NIH / 3T3 mouse embryonic fibroblasts were purchased from the National Cell Identification Library of the Chinese Academy of Sciences. NIH / 3T3 cells were inoculated in Dulbecco's Modified Eagle's medium containing 10% newborn calf serum and 1% penicillin and streptomycin and cultured in a 37°C, 5% CO2 incubator. (2) Plating and drug administration: After the cells were inoculated in the culture dish, the cells in the logarithmic growth phase were collected and added to a 96-well black transparent bottom culture plate at a concentration of 5000 cells per well, 100 μL per well, and cultured in a 37°C, 5% CO2 incubator for 24 h. Different concentrations of the test compound 1 μM, 5 μM, 10 μM, 20 μM and 10 ng / mL TGF-βI and a control group were added and cultured in a 37°C, 5% CO2 incubator for 24 h before subsequent operations. (3) Extraction of cell morphology analysis software parameters: This part was analyzed using Harmony 4.9 software; the high-content imaging system was used to collect DPC channel images; the 13 morphological parameters selected by Harmony 4.9 software were collected and analyzed for DPC images of NIH / 3T3 cells; the C group, model group, 1μM, 5μM, 10μM, and 20μM groups were clustered using principal component analysis (PCA) to analyze the above multidimensional parameters. (4) Immunofluorescence staining: NIH / 3T3 cells were fixed with 4% paraformaldehyde for 15 min; then, after washing three times in PBS, the cells were permeabilized with 0.1% Triton X-100 for 15 min, and the samples were blocked with 10% bovine serum albumin for 30 min; taken out; placed in primary antibody solution and incubated at 4°C overnight; after washing, the samples were incubated in FITC-conjugated secondary antibody solution at room temperature for 2 h; then, DAPI was used to stain the cell nucleus; finally, a high-content imaging and analysis system was used for image acquisition; the fluorescence intensity parameters of the α-SMA channel were acquired using Harmony 4.9 software; Experimental results: (1) The principal component analysis (PCA) clustering effects of 13 parameters of 10 compounds were all able to be distinguished from the model group; (2) The α-SMA fluorescence intensity of the 10 compounds was significantly different from that of the model group and all of them could reduce the expression of α-SMA; this fully verified the feasibility and accuracy of the screening method using 13 parameters; (3) IC50 toxicity tests were performed on these 10 compounds, and all compounds had anti-fibrotic effects at non-toxic doses.

8. The high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application according to claim 1, characterized in that: In step 5, the bright field digital phase contrast imaging method is established: Microscopic imaging has proven to be an essential tool for quantitative monitoring of biological processes with unprecedented spatial resolution; however, current analytical methods are limited in the number of measurable parameters or image feature extraction; HCS technology combines automated fluorescence microscopy with high-throughput automated acquisition technology and has become a powerful tool for studying cell biology and drug development; With significant advances in computing power, robotics and automation technology, as well as image detectors, the amount of information obtained through optical microscopy methods has greatly increased, and the application of high-content imaging (HCI) has become more and more extensive; The effectiveness of HCI in precision medicine has been confirmed in recent studies; Differential phase contrast (DPC) is a non-interferometric quantitative phase imaging method that uses an asymmetric imaging procedure; Compared with most other methods, DPC is experimentally implemented without the need for fluorescent labeling; In DPC, a spatially asymmetric imaging method is used to generate phase gradient images; this technique can observe living cells, analyze bright field images, does not use any fluorescent dyes, and is non-toxic; however, a machine learning approach is needed to extract more information from DPC images; in recent years, several methods have been proposed to predict fluorescence images from transmitted light images, such as bright field images and phase images; although the above methods show the possibility of converting biomarker images using computational methods, no existing methods are as specifically targeted as our work and verify their applicability in HCS applications. In this part, the processing parameters of DPC imaging were optimized by comparing the DPC imaging counting method with the cell nucleus fluorescence labeling method. Then, each well was imaged at 10×magnification using the Opera Phenix high-content analysis system equipped with a high-resolution 14-bit charge-coupled device (CCD) camera. The analysis was performed using Harmony 4.9 software, in which 9 images were collected for each well. The resolution of an image was 1360×1024 pixels, corresponding to an actual physical size of 0.88×0.66cm. The imaging system produced two-channel output images, one for bright field imaging and the other for fluorescence imaging. DPC imaging counts and Hoechst 33342 staining counts were compared at 0, 12, 24, and 36 h, respectively. The results showed that there was no significant difference between the two counting methods. Therefore, compared with the traditional Hoechst 33342 cell nucleus staining method, the established method does not require labeling. A major advantage of phenotypic screening compared to target-based approaches is that disease-associated proteins are maintained in their natural environment, resulting in more physiological translational outcomes. In this section, deep machine deconvolution analyzed by Harmony 4.9 software outputs 31 cell morphological parameters during DPC imaging. In summary, more parameters can be obtained in a single screening experiment through deep analysis of image morphology. (1) Parameter relationship under stimulation of different concentrations of TGF-βI: The results showed that the DPC imaging parameters and α-SMA fluorescence imaging under the stimulation of different concentrations of TGF-βI, cell area, cell roundness, cell width, length and aspect ratio showed the same trend; at the same time, the fluorescence intensity of α-SMA increased with the increase of different concentrations of TGF-βI; 33 morphological parameters of NIH / 3T3 cells were obtained by DPC image analysis. Through correlation analysis, 24 DPC parameters such as Cell Area and Cell Eadial Mean were correlated with α-SMA expression; (2) DPC parameter analysis at different times: Subsequently, at the same TGF-βI concentration (10 ng / mL), DPC imaging parameters and α-SMA fluorescence imaging were performed at different times; consistent with the above results, DPC imaging parameters showed the same trend in cell area, cell roundness, cell width, cell length and aspect ratio parameters of α-SMA fluorescence imaging at different times; the expression of α-SMA increased with the extension of administration time; Subsequent experiments analyzed these morphological parameters and α-SMA fluorescence intensity; 33 morphological parameters of NIH-3T3 cells were obtained through DPC image analysis; through correlation analysis, 19 DPC parameters (r2>0.3) such as Cell Area and Cell Eadial Mean were correlated with α-SMA expression; In step six, an image analysis method assisted by "machine learning" technology is established. In this part, we use the Inceptionv4 architecture neural network to classify the images. The initial input is a 16-bit image, containing a bright field channel and a DPC channel. We first resize and standardize each channel of the original image, and then generate a third channel with the average of the existing two channels to form an RGB (red-green-blue) image, and input the 299×299×3 image into the Inceptionv4 network. The output dimension of the second-to-last fully connected layer is modified to 6, and the cell images are classified into 6 types. The Inception v4 architecture neural network is pre-trained on the ImageNet object recognition dataset, which contains 1.2 million non-cell objects in 1000 categories. We use pre-trained convolutional layers, connect them with randomly initialized fully connected layers, and train the resulting network on our own constructed dataset. Using TGF-βI as the modeling agent, different concentrations of TGF-βI (0.01μM, 0.10μM, 1μM, 5μM, 10, 20μM) were added to the wells where NIH-3t3 cells were cultured, and the treated cells were labeled as types 1-5 respectively; we defined type 0 and type 1 cells as normal cells, type 2 and type 3 cells as fibrosis process-arrested cells, and type 4 and type 5 cells as fibrotic cells; a dataset containing 24,000 labeled images was established, which were evenly distributed in six types; we randomly split the dataset with a training-validation ratio of 4:1, obtaining 19200 and 4800 respectively; Softmax_cross_entrophy was used as the loss function, and Adam was used as the optimizer. After 120 epochs of training, the network loss converged and the accuracy of both the training and validation sets reached 95%; we then tested the model on a test set of 4,200 labeled images, which were independent of the training and validation sets conducted by different operators; the model performed well in predicting non-training data, with a calculated accuracy of approximately 92.30%; the images of the above 1,400 compounds were distinguished using the above machine learning model, and 1,400 × 6 wells × 9 fields of view = 75,600 images were processed.

9. The high-throughput anti-fibrosis drug screening method based on fluorescence imaging without the need for fluorescence imaging and its application according to claim 1, characterized in that: In step seven, targeted inhibitor intervention: SD-208 is a selective TGF-βI inhibitor. This part uses DPC imaging technology to image NIH / 3T3 at different concentrations of SD-208 (50nM, 25nM, 10nM, 5nM, 1nM). At the same time, immunofluorescence imaging of α-SMA was performed; as the concentration of SD-208 increased, the fluorescence intensity of α-SMA also decreased, indicating that SD-208 affected the expression of α-SMA protein. This part also collected the morphological parameters of NIH / 3T3 cell DPC imaging under the interference of different concentrations of SD-208; under the influence of SD-208, 16 morphological parameters such as cell area and cell DPC SER right 0px of DPC imaging cells were correlated with the fluorescence value of α-SMA. In summary, we conducted a linear correlation analysis of DPC parameters under the influence of different concentrations of TGF-βI, the same TGF-βI at different times, and the target inhibitor SD-208. By analyzing the relationship between different groups of parameters (correlation coefficient greater than 0.3), a total of 13 DPC parameters were common perturbation parameters; these parameters will be used to predict the perturbation mechanism. The high-throughput anti-fibrosis drug screening method based on the method without fluorescence imaging and its application according to claim 1 is characterized in that: Determination of anti-fibrosis degree of DPC images based on unbiased automatic deep machine learning: In order to process DPC image data more deeply and integrate image recognition methods based on deep learning into our platform; Using TGF-βI as a modeling agent, cell images treated with different concentrations of TGF-βI were labeled as types 0-5; we labeled 2,000 images for each type. By rotating the augmented data, a final dataset of 48,000 images was obtained; the classification network based on InceptionV4 was fine-tuned on 48,000 images. After 120 epochs of training, the network loss converged and the accuracy of both the training set and the validation set reached 95%; then, we tested the model on a test set of 4200 labeled images, which were independent of the training and validation sets conducted by different operators; the model showed high accuracy in predicting non-training data, with a calculated accuracy of approximately 92.30%; the images of the above 1400 compounds were distinguished using the above machine learning model, and 1400×6 wells×9 fields of view=75600 images were processed. Integrating deep learning and unbiased automated discovery of anti-fibrotic lead compounds: In summary, through two parts of experiments, 33 lead compounds were screened out by morphological parameters, and 65 lead compounds were screened out by unbiased image learning method; 29 lead compounds were the intersection of the two screening modes; the names and CAS numbers of the 29 compounds are listed in Tables 1-7. The number of compounds obtained by the image-based screening mode is greater than that of compounds screened based on morphological parameters. This may be because the results of compound screening based on morphological parameters are overly dependent on phenotypic morphological parameters. Compared with the morphological screening mode, image-based deep learning is an unbiased screening result. The above results show that a total of 29 lead compounds can significantly inhibit abnormal proliferation of fibroblasts. It is important to maintain the original state of fibroblasts. This result is in line with our purpose of screening anti-fibrosis lead compounds. In addition, the efficacy of the screening results was reconfirmed; immunofluorescence staining was used to verify that the 29 screened compounds all had the ability to downregulate the expression of α-SMA protein; the results showed that all 29 lead compounds inhibited the expression of α-SMA protein, that is, the fluorescence value of α-SMA in the compound group was significantly lower than that in the model group. In summary, we rapidly identified 29 TCM compounds from 1400 TCM compounds; these 29 compounds significantly maintained the morphology of fibrocytes and inhibited abnormal fibrocyte proliferation.

Images