WO2023085837A1 - Method and device for screening for osteoporosis drug by using bone tissue mimetic image analysis - Google Patents

Method and device for screening for osteoporosis drug by using bone tissue mimetic image analysis Download PDF

Info

Publication number
WO2023085837A1
WO2023085837A1 PCT/KR2022/017713 KR2022017713W WO2023085837A1 WO 2023085837 A1 WO2023085837 A1 WO 2023085837A1 KR 2022017713 W KR2022017713 W KR 2022017713W WO 2023085837 A1 WO2023085837 A1 WO 2023085837A1
Authority
WO
WIPO (PCT)
Prior art keywords
bone
bone tissue
drug
image
cell
Prior art date
Application number
PCT/KR2022/017713
Other languages
French (fr)
Korean (ko)
Inventor
김정아
탁성호
백규림
Original Assignee
한국기초과학지원연구원
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from KR1020220053246A external-priority patent/KR102531759B1/en
Application filed by 한국기초과학지원연구원 filed Critical 한국기초과학지원연구원
Publication of WO2023085837A1 publication Critical patent/WO2023085837A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
    • G01N33/48—Biological material, e.g. blood, urine; Haemocytometers
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
    • G01N33/48—Biological material, e.g. blood, urine; Haemocytometers
    • G01N33/50—Chemical analysis of biological material, e.g. blood, urine; Testing involving biospecific ligand binding methods; Immunological testing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis

Definitions

  • the present invention relates to a method and apparatus for screening an osteoporosis drug by analyzing an image obtained from a bone tissue mimetic capable of mimicking the function and structure of bone tissue.
  • Bone diseases are rapidly increasing due to the aging of the population.
  • osteoporosis is a skeletal disease in which fractures increase due to weakening of bone strength, is a senile disease, and has emerged as an important public health problem worldwide.
  • Korea is also one of the countries where the population is aging rapidly, and the number of osteoporosis patients is also increasing at a rapid pace.
  • Korean Patent Publication No. 10-2015-0020702 discloses a three-dimensional biological connective tissue construct that mimics the shape of human connective tissue including bone connective tissue cells, but it is composed of a simple three-dimensional scaffold.
  • an analysis method for observing cell-cell interactions in bone tissue or confirming functional improvement of bone is not easy to use for an analysis method for observing cell-cell interactions in bone tissue or confirming functional improvement of bone.
  • the present inventors have completed the present invention by evaluating and screening drugs for treating or preventing bone diseases such as osteoporosis by analyzing images obtained from a three-dimensional bone tissue mimic by artificial intelligence. .
  • the present invention comprises the steps of treating a bone tissue mimic with a drug related to bone disease; Labeling the bone tissue matrix with a marker to obtain an image or image from the drug-treated bone tissue matrix; Obtaining a bone tissue matrix image or image including the labeled marker; Calculating a feature map by learning the bone tissue mimetic image or image as a deep learning model; and providing drug efficacy evaluation information using the feature map.
  • the bone tissue mimetic lower plate As an example according to the present invention, the bone tissue mimetic lower plate; an upper plate located above the lower plate; a barrier portion formed of a protruding structure derived from either the lower plate or the upper plate and located between the upper plate and the lower plate; One or more cell sample injection guides provided on one side of the top plate, wherein the size of the top plate is such that only the inside of the barrier part is closed, so that the outer part of the barrier part has an open structure. It may be cultured in
  • the bone tissue matrix includes a bone cell (osteocyte), filled in the center of the gel (gel); and a cell mixture disposed to surround the outer portion of the gel.
  • osteocyte bone cell
  • gel gel
  • cell mixture disposed to surround the outer portion of the gel.
  • the bone disease is osteoporosis, incomplete osteogenesis imperfecta, hyperossification, hypercalcemia, hyperparathyroidism, osteomalacia, dissolving bone disease, osteonecrosis, Paget's disease of bone, bone fracture, It may be selected from the group consisting of rheumatoid arthritis, osteomyelitis, periodontal bone loss, bone loss due to cancer, senile bone loss, and rickets.
  • the cell mixture may include at least one or more selected from the group consisting of osteogenic cells, osteoblasts, osteoclasts, immune cells, and vascular cells.
  • the deep learning model uses images or images of a bone mimic treated with a drug known to have a preventive or therapeutic effect on bone disease and an image or image of a bone mimic not treated with the drug as learning data. It may have been learned using
  • the drug is used for DNA synthesis, RNA synthesis, gene expression, gene structure, protein synthesis, protein modification, In the group consisting of protein secretion, protein structure, sugar synthesis, sugar secretion, sugar modification, lipid secretion, cell membrane synthesis, intracellular signal transduction, intercellular signal transduction, intracellular organelle, cell differentiation, cell division, and cell movement It may be one that affects one or more of the selected ones.
  • the drug is a natural product, synthetic drug, combination drug, herbal medicine, herbal medicine extract, herbal preparation, herbal medicine, protein medicine, gene recombinant medicine, cell culture medicine, enzyme medicine, It may be selected from the group consisting of microbial drugs, antibody drugs, hormone drugs, radiation drugs, antibody-drug conjugates, cell therapy products, and gene therapy products.
  • the marker may be obtained by further adding any one or more selected from the group consisting of a dye, a fluorescent material, a phosphorescent material, and a radioactive material.
  • the bone tissue matrix image or image may be obtained by processing two or more markers for identifying different objects.
  • the region labeled with the marker is a cell, nucleus, ribosome, lysosome, Golgi body, centrosome, cell membrane, mitochondria, microplatelet, microfilament, cytoplasm, DNA, RNA, nucleic acid, histone of the bone tissue mimetic , It may represent at least one selected from the group consisting of proteins, glycoproteins, membrane proteins, carbohydrates, membrane carbohydrates, lipids, cholesterol, glycolipids, sugars, and collagen.
  • the area marked with the marker may represent at least one selected from the group consisting of the shape, shape, position and movement of the bone tissue mimetic.
  • the feature map may indicate the result of calculating the image or image of the bone tissue mimetic using a convolutional neural network (CNN).
  • CNN convolutional neural network
  • an input device for receiving a drug-treated bone tissue mimetic image or image; a storage device for storing a deep learning model that can be learned using the bone tissue mimetic image or image and evaluate drug efficacy; an arithmetic device that receives a bone mimic image or image from the input device, inputs the image to the deep learning model stored in the storage device, and determines the effect of the drug on the bone mimetic according to a value output from the deep learning model; It provides a device for evaluating drugs for preventing or treating bone diseases, including a.
  • the bone tissue matrix includes a bone cell (osteocyte), filled in the center of the gel (gel); and a cell mixture disposed to surround the outer portion of the gel.
  • osteocyte bone cell
  • gel gel
  • cell mixture disposed to surround the outer portion of the gel.
  • the bone tissue mimetic image or image may include a region marked with a marker on the drug-treated bone tissue mimetic.
  • the cell mixture may include at least one selected from the group consisting of osteogenic cells, osteoblasts, osteoclasts, immune cells, and vascular cells.
  • the marker may be obtained by further adding any one or more selected from the group consisting of a dye, a fluorescent material, a phosphorescent material, and a radioactive material.
  • the bone tissue matrix image or image may be obtained by processing two or more markers for identifying different objects.
  • the gel may be a hydrogel.
  • the hydrogel is Matrigel, Puramatrix, collagen, fibrin gel, polyethylene glycol diacrylate (PEG-DA), polyethylene glycol dimesa Composed of PEG-DMA, PNIPAM, Poloxamer, Chitosan, Agarose, Gelatin, Hyaluronic acid and Alginate It may include any one or more selected from the group.
  • the hydrogel may further include an extracellular matrix (ECM).
  • ECM extracellular matrix
  • the cell mixture may include at least one of osteoblasts and osteoclasts and a culture medium.
  • (S1) mixing bone cells (Osteocyte), extracellular matrix (ECM) and hydrogel; (S2) dropwise addition of the mixture onto a support and then gelation; (S3) seeding osteoblasts to surround the outer edge of the gelled hydrogel; And (S4) injecting the culture solution; provides a method for producing a bone tissue mimic, including.
  • step (S1) matrigel may be further included for mixing.
  • the hydrogel may be collagen.
  • the gelation in step (S2) may be performed in the range of pH 6 to 8.
  • the step (S3) may be inoculating osteoclasts together with osteoblasts to surround the outer edge of the gelled hydrogel.
  • a method for evaluating drugs for preventing or treating bone diseases such as osteoporosis or screening drugs for developing new drugs can be performed efficiently and with high accuracy.
  • the method of the present invention is easy to use as an image-based evaluation or screening method.
  • conventional co-culture models or trans-swell-type models in which bone cells and osteoblasts are arranged in a vertical direction have difficulties in imaging analysis due to their thickness, whereas osteocytes can mimic the functional characteristics of actual bone tissue.
  • Easy and effective observation of cell-cell and cell-ECM interactions occurring in bone tissue similar to reality by using a bone tissue mimetic whose main composition is the arrangement relationship in which the included gel and cell mixture are placed horizontally can do.
  • FIG. 1 is an example showing a method and apparatus for screening a drug for bone disease using bone tissue mimetic image (image) analysis according to the present invention.
  • Figure 2 is an example showing the structure of the bone tissue matrix.
  • FIG. 3 is an example illustrating a process of extracting an image from a bone tissue matrix.
  • FIG. 4 is an example illustrating a process of constructing a deep learning model for evaluating the efficacy of a drug for bone disease using a bone tissue mimetic image.
  • FIG. 5 is an example showing an image of a bone tissue mimetic used in the process of building a deep learning model.
  • 6 is an example showing the configuration of a deep learning model.
  • 9 is an example showing ROC curves of the BN model ( ⁇ -catenin+nucleus) and the BNM model ( ⁇ -catenin+nuclear image merge).
  • 10 is an example showing the configuration of an evaluation device for drug efficacy using a bone tissue mimetic.
  • FIG. 11 is an example illustrating a process of manufacturing a structure of a bone tissue mimic.
  • FIG. 12 is an example illustrating a process of extracting an osteoblast-derived decellularized extracellular matrix (OB-dECM) from osteoblasts.
  • OB-dECM osteoblast-derived decellularized extracellular matrix
  • FIG. 13 is an example showing the reaction in osteoblasts by SOST secreted from osteocytes and the state of treatment with a SOST monoclonal antibody as an osteoporosis-inducing drug.
  • Figure 14 is an example showing the timing of inoculation of osteoblasts according to the maturation of osteoblasts in the bone tissue matrix, the timing of drug treatment, and the photographing timing.
  • 15 is an example of a schematic diagram of a prior art system and a system (Bone-on-a-chip) of the present invention.
  • 17 is a result of measuring cell maturity of bone cells in the system of the prior art and the system of the present invention using osteogenic differentiation markers, respectively.
  • FIG. 18 illustrates a cell culture container for culturing bone tissue mimics according to the present invention.
  • units used without particular notice are based on weight, and for example, % or a unit of ratio means weight% or weight ratio.
  • the expression “comprises” is an open description having a meaning equivalent to expressions such as “includes”, “includes”, “has” or “characterized by”, and is further listed It does not exclude elements, materials or processes that do not exist. Also, “actually... The expression “consists of” means that the specified element, material or process together with other elements, materials or processes not listed do not significantly affect at least one basic and novel technical idea of the invention in an unacceptably significant manner. It means that it can exist in quantity. Also, the expression “consisting of” means that only the described elements, materials or processes are present.
  • the terms “ingredient”, “composition”, “composition of a compound”, “compound”, “drug”, “pharmaceutical active”, “active agent”, “healing” “treatment” “treatment” or “medicament” is used interchangeably to mean a compound or compound(s) or composition of matter that, when administered to a subject (human or animal), induces a desired pharmacological and/or physiological effect by local and/or systemic action. used interchangeably.
  • the present invention includes the steps of treating a drug to a three-dimensional bone tissue mimetic; Searching for and pre-processing intracellular markers related to bone diseases such as osteoporosis from the drug-treated three-dimensional bone tissue mimetic; obtaining a large number of images from the 3-dimensional bone tissue mimetic containing the pre-processed intracellular marker; pre-processing the image into a data set for application to a deep learning algorithm; Calculating a feature map using a convolutional neural network (CNN); integrating the calculated feature maps; and providing information on evaluating drug efficacy using the integrated feature map.
  • CNN convolutional neural network
  • processing a drug candidate to the three-dimensional bone tissue mimetic pre-processing to find intracellular markers related to bone diseases such as osteoporosis from the treated 3-dimensional bone tissue mimetic; obtaining a large number of images from the 3-dimensional bone tissue mimetic containing the pre-processed intracellular marker; pre-processing the image into a data set for application to a deep learning algorithm; Calculating a feature map using a convolutional neural network (CNN); integrating the calculated feature maps; and providing information on evaluating drug efficacy using the integrated feature map.
  • CNN convolutional neural network
  • the step of treating the 3-dimensional bone tissue mimetic with the drug is the step of injecting the drug or drug candidate into the 3-dimensional bone tissue mimetic.
  • the pretreatment of the intracellular marker in the drug evaluation method and the drug screening method may be staining the intracellular marker.
  • the intracellular marker may be, for example, ⁇ -catenin.
  • the intracellular marker is, for example, ⁇ -catenin, and the pretreated intracellular marker may be a drug evaluation indicator.
  • the bone tissue matrix includes a bone cell (Osteocyte) and filled in the center gel (gel); and a cell mixture disposed to surround the outer portion of the gel.
  • the bone tissue mimetic is a biomimetic structure that well mimics bone tissue, and can be manufactured in a well plate shape with a multi-well structure that is friendly to high-speed analysis. Therefore, in conjunction with high-speed equipment, it can be applied to various biological analyzes (eg, cell activity, toxicity, imaging analysis, etc.) in a large amount of samples, and the analysis efficiency can be increased. method is not limited.
  • the bone tissue mimetic may be formed in a well plate having a multi-well structure, and a method for screening an active substance for preventing or treating bone diseases using the same is provided. do.
  • the gel may be a hydrogel, and the hydrogel is Matrigel, Puramatrix, collagen, fibrin gel, polyethylene glycol diacrylate (PEG-DA), polyethylene glycol diacrylate Mesacrylate (PEG-DMA), PNIPAM, Poloxamer, Chitosan, Agarose, Gelatin, Hyaluronic acid and Alginate It may be one or more selected from the group consisting of, preferably a mixture of collagen and Matrigel. When collagen and Matrigel are mixed and used, the differentiation and maturity of the bone tissue mimetic according to the present invention can be improved.
  • the hydrogel may further include an extracellular matrix (ECM).
  • ECM extracellular matrix
  • the ECM may be extracted from cells or tissues or biochemically synthesized.
  • the cell mixture may include, but is not limited to, any one or more of osteoblasts and osteoclasts and a culture medium, and furthermore, other tissue cells such as blood vessels or immune cells and a culture medium. can include more.
  • the bone tissue mimetic is not limited, but the lower plate 620; an upper plate 630 located above the lower plate; a barrier portion 610 having a protruding structure derived from either the lower plate or the upper plate and positioned between the upper plate and the lower plate; One or more cell sample injection guides 720 provided on one side of the top plate; the size of the top plate is such that only the inside of the barrier part is closed, so that the outer part of the barrier part has an open structure. It may be cultured in a cell culture container. An example of the structure of the cell culture vessel according to an embodiment of the present invention is shown in FIG. 18.
  • osteocytes are basic cells constituting the bone tissue of vertebrates, which are flat oval cells
  • osteoblasts are cells that synthesize and secrete bone matrix of vertebrates
  • osteoclasts In vertebrates, refers to a multinucleated cell that destroys and absorbs bone tissue that becomes unnecessary along with bone growth.
  • the types of bone cells, osteoblasts, and osteoclasts are not particularly limited, and include cell lines derived from vertebrates including humans and mice, or animal cells such as primary cultured cells and stem cells. can do.
  • the manufacturing method of the bone tissue mimetic is (S1) mixing the bone cells (Osteocyte), extracellular matrix (ECM) and hydrogel; (S2) dropwise addition of the mixture onto a support and then gelation; (S3) seeding osteoblasts to surround the outer edge of the gelled hydrogel; and (S4) injecting the culture medium.
  • At least one gel may be further included, and for example, a hydrogel extracted from tissues such as matrigel and ECM may be further included and mixed.
  • Matrigel is rich in hyaluronic acid and cytokine, so it can improve the differentiation of bone tissue matrix and increase its maturity.
  • the support may be made of glass, ceramic, silicone rubber, silicon, polystyrene, polymethylmethacrylate, polypropylene, or polycarbonate. , It may be composed of one or more materials selected from the group consisting of polyurethane, photocurable plastics, thermoplastics, and metals. In addition, it may include a chemical variant thereof, and may include a material that can be used for 3D printing.
  • silicone rubber examples include, but are not limited to, polydimethylsiloxane (PDMS) and ecoflex.
  • the hydrogel is Matrigel, Puramatrix, collagen, fibrin gel, polyethylene glycol diacrylate (PEG-DA), polyethylene glycol dimesacrylate (PEG-DMA) It may be at least one selected from the group consisting of PNIPAM, Poloxamer, Chitosan, Agarose, Gelatin, Hyaluronic acid, and Alginate. And, preferably, it may be a mixture of collagen and Matrigel, or a mixture of collagen, Matrigel, and ECM, but is not limited thereto.
  • the differentiation and maturity of the bone tissue mimetic according to the present invention can be improved.
  • the ECM may be extracted from bone cells or bone tissue or biochemically synthesized, but is not limited thereto.
  • Gelation in the step (S2) may be performed in the range of pH 6 to 9.
  • gelation is made in the pH range of 7 to 8, the morphology and viability of bone cells in the bone tissue matrix can be improved.
  • the step (S3) may be seeding osteoclasts together with osteoblasts so as to surround the outer edge of the gelated hydrogel.
  • the bone disease is caused by osteoporosis, incomplete osteogenesis imperfecta, hyperostosis, hypercalcemia, hyperparathyroidism, osteomalacia, lytic bone disease, osteonecrosis, Paget's disease of bone, bone development disease, bone fracture, and rheumatoid arthritis It may be any one selected from the group consisting of bone loss, inflammatory rheumatoid arthritis, osteomyelitis, metastatic bone disease, periodontal bone loss, rickets, bone loss due to cancer and senile bone loss, preferably osteoporosis. It is not limited.
  • Prevention of the present invention means any action that suppresses or delays the onset of a disease by administering the composition according to the present invention.
  • treatment includes therapeutic effects such as inhibiting or stopping the progression of a disease or disease, reducing the rate of progression, improving or alleviating symptoms, or preventing a disease for a biological subject (eg, human or animal).
  • bone tissue matrix image means an image of a bone tissue matrix photographed.
  • the image includes a marker region, and the marker region represents a region marked with a marker for displaying a specific drug response or mechanism on the bone tissue mimetic.
  • “learning model” refers to a machine learning model
  • the machine learning model may include various types of models.
  • machine learning models include decision trees, random forests (RFs), K-nearest neighbors (KNNs), Naive Bayes, support vector machines (SVMs), and artificial neural networks (ANNs). It is not limited.
  • the machine learning model is associated with various neural network models, and among them, ANN is a statistical learning algorithm that imitates a biological neural network.
  • DNN Deep Neural Network
  • Various types of DNN models have been studied, such as CNN (Convolutional Neural Network), RNN (Recurrent Neural Network), RBM (Restricted Boltzmann Machine), DBN (Deep Belief Network), GAN (Generative Adversarial Network), RL ( Relation Networks), etc.
  • CNN Convolutional Neural Network
  • RNN Recurrent Neural Network
  • RBM Restricted Boltzmann Machine
  • DBN Deep Belief Network
  • GAN Geneerative Adversarial Network
  • RL Relation Networks
  • 1 is an example of a bone disease drug screening method using bone tissue mimetic image analysis.
  • Bone tissue mimetic 100 of Figure 1 mimics internal bone tissue.
  • the body bone tissue includes human or animal bone tissue. Through this, it is possible to evaluate the efficacy of the drug 110 for bone diseases without direct experiments on humans or animals.
  • the in vivo bone tissue may be derived from humans or non-human animals.
  • Bone mimic 100 can be used for rapid drug screening.
  • the bone tissue mimic 100 may be further processed with a marker 120 for confirming a specific substance or reaction.
  • the imaging device 130 of FIG. 1 acquires the bone tissue matrix image 200 by photographing the bone tissue matrix 100.
  • the bone tissue matrix image 200 of FIG. 1 is an image for viewing the reaction of the bone tissue matrix according to drug treatment.
  • the bone tissue matrix image 200 generated by the imaging device 130 may be stored in a separate database (DB).
  • the analysis device 300 of FIG. 1 receives the bone tissue mimetic image 200 generated by the imaging device 130 and inputs it to a pre-learning model.
  • the analysis device 300 outputs a result value for the degree to which the drug 110 gives to the bone tissue mimetic 100 according to the value output by the learning model.
  • the learning model is a model trained to predict drug effects in advance.
  • the learning model outputs a probability value for the effect of the drug 110 to be analyzed.
  • the analysis device 300 may determine whether or not the drug is effective based on the probability value output by the learning model.
  • the structure of the bone tissue parent body 100 will be described.
  • Figure 2 is an example showing a bone tissue mimic (100).
  • Bone tissue mimetic 100 mimics the bone tissue in the body.
  • the bone tissue in the body includes human or animal bone tissue. Through this, the efficacy of drugs for bone diseases can be evaluated without direct experiments on humans or animals.
  • the bone tissue mimetic 100 has a hydrogel 104 containing osteocytes 101 and a decellularized extracellular matrix 102 derived from osteoblasts located at the center, and a cell mixture 103 around it. can exist
  • the bone tissue matrix 100 has a structure in which the cell mixture is horizontally disposed around the hydrogel 104. This has the characteristic of being able to easily observe cell-cell interactions optically, unlike models that are disposed in a vertical direction or models that have different focal planes, such as transwells.
  • the bone tissue mimetic 100 is a biomimetic structure that well mimics bone tissue, and can be manufactured in a well plate shape with a multi-well structure that is friendly to high-speed analysis. , with structural features favorable for high-speed drug efficacy evaluation assays.
  • the bone tissue mimetic 100 can be used as a model for drug screening, particularly high-speed screening or drug evaluation, for the development of new drugs for bone-related diseases such as osteoporosis, and can also be used for various studies on bone diseases. .
  • the cell mixture 103 may include at least one of osteoblasts, osteoclasts, and osteogenic cells. It may also contain other tissue cells such as blood vessels or immune cells and culture medium.
  • Figure 3 shows the process of obtaining the bone tissue matrix image 200 from the bone tissue matrix 100.
  • the bone tissue matrix image 200 can be obtained by using the imaging device 130 after processing the marker 120 on the bone tissue matrix 100.
  • the image 200 of the bone tissue mimetic may be an image of the bone tissue mimetic 100 treated with the drug 110.
  • the drug 110 may be a drug for treating bone disease.
  • the drug 110 may be a combination of several drugs.
  • the drug 110 is applied to DNA synthesis, RNA synthesis, gene expression, gene structure, protein synthesis, protein modification, protein secretion, protein structure in at least one of the bone cells in the center or the cell mixture in the outer part of the bone tissue mimetic 100.
  • a drug that affects at least one of glycosynthesis, glucose secretion, sugar modification, lipid secretion, cell membrane synthesis, intracellular signaling, intercellular signaling, intracellular organelles, cell differentiation, cell division, and cell movement may be
  • the drug 110 includes natural products, synthetic drugs, combination drugs, herbal medicines, herbal medicine extracts, herbal medicines, herbal medicines, protein medicines, genetically recombinant medicines, cell culture medicines, enzyme medicines, microbial medicines, antibody medicines,
  • the drug may be at least one of a monoclonal antibody drug, a hormone drug, a radiopharmaceutical, an antibody-drug conjugate, a cell therapy, and a gene therapy.
  • the marker 120 is used to identify a specific substance or reaction in the bone tissue mimetic.
  • the marker 120 may emit various signals that the imaging device 130 can measure.
  • it may be a dye that exhibits a specific color, a luminescent material that generates light, a fluorescent or phosphorescent material that generates fluorescence or phosphorescence, or a material that generates radioactivity.
  • the imaging device 130 may acquire an image of the bone tissue matrix by photographing the bone tissue matrix 100.
  • the imaging device 130 refers to any device capable of measuring an image required for image analysis.
  • it may be at least one of a general microscope, a fluorescence microscope, a confocal microscope, an electron microscope, a scanning probe microscope, and a tomography microscope.
  • the imaging device 130 may transmit the captured image 200 of the bone tissue matrix to the image analysis device or store it in a separate image DB 230.
  • the bone tissue matrix image 200 means an image of the bone tissue matrix 100 photographed by the imaging device 130.
  • the bone tissue matrix image 200 includes a marker region.
  • the marker region represents a region marked with a marker for displaying a specific location or reaction of a material on the bone tissue mimic.
  • the bone tissue mimetic image 200 may be an image including a single marker for identifying one specific target or an image including a plurality of markers for identifying different targets.
  • the marker region may be a region in which a learning model is centrally learned in an image of a bone tissue mimic and a drug is evaluated based thereon.
  • the marker area may be displayed differently according to the type of the processed marker 120 .
  • it may be a region that emits color by treating dye, emits light by treating a luminescent material, emits fluorescence or phosphorescence by treating a fluorescent/phosphorescent material, or shows radioactivity by treating a radioactive element.
  • the marker area may indicate the shape, shape, position, or movement of the bone tissue mimetic depending on the type of the marker 120 .
  • the marker region may be a cell, a cell, a nucleus, an endoplasmic reticulum, a ribosome, a lysosome, a Golgi apparatus, a centrosome, a cell membrane, mitochondria, microtubules, microfilaments, cytoplasm, DNA, RNA, nucleic acids, histones, proteins, glycoproteins, membrane proteins, carbohydrates, membrane carbohydrates, lipids, cholesterol, glycolipids, sugars, collagen, and the like may be labeled.
  • Figure 4 shows a process of implementing a deep learning model for evaluating the efficacy of a bone disease drug using an image of a bone mimic.
  • the process of building a learning model for a deep learning model is divided into a process of building learning data (310) and a process of learning a model using learning data (330).
  • the learning data construction process 310 and the model learning process 330 may be performed in separate devices.
  • the learning device performs a process of building a learning model.
  • the learning device uses the bone tissue mimetic image 200 as learning data for implementing a learning model.
  • the learning device may receive an image of the bone tissue mimetic to be used as the learning data from the image DB 230 or transmitted from the imaging device 130 .
  • the image of the bone tissue mimic to be used as the learning data may include an image of the drug-treated bone tissue mimic (positive data) and an image of the drug-untreated bone tissue mimic (negative data).
  • the learning data may include an image treated with an effective drug (positive data) and an image treated with an ineffective drug (negative data).
  • the learning data may include label value (positive or negative) information for the corresponding image.
  • the image of the bone tissue matrix to be used as the learning data may be an image including a single marker for identifying a specific target or an image including a plurality of markers for identifying different targets.
  • the different objects are cells in the bone tissue matrix, cells of the bone tissue matrix, nucleus, endoplasmic reticulum, ribosomes, lysosomes, Golgi apparatus, centrosome, cell membrane, mitochondria, microtubules, microfilaments, cytoplasm, DNA, RNA, nucleic acids, histones , proteins, glycoproteins, membrane proteins, carbohydrates, membrane carbohydrates, lipids, cholesterol, glycolipids, sugars, collagen, and the like may be labeled.
  • the bone tissue matrix image to be used as learning data may be directly used as learning data, but may also be used as learning data after augmenting the data.
  • data augmentation may be performed by dividing the parent image into n pieces (n>2), randomly enlarging or reducing each image, or horizontally or vertically flipping each image.
  • the learning device may divide the image into two left and right parts and horizontally flip only one of them to prepare a bone tissue matrix image. In this way, the learning device may generate n x m images of the bone tissue matrix by processing m images of the bone tissue matrix.
  • Bone tissue mimetic image to be used as the learning data may be stored in the learning DB (320).
  • the learning device performs a model learning process 330 using learning data using the bone tissue mimetic image stored in the learning DB 320.
  • the learning device acquires at least one or more bone tissue matrix images from the learning DB (320).
  • the learning device inputs the corresponding image to the learning model.
  • the learning model extracts features from the input image and calculates a probability value whether there is a change due to drug treatment.
  • the learning device updates parameters of the learning model by comparing a probability value output from the learning model with a previously known label value of the corresponding image.
  • the learning device repeats the process of learning the learning model using a plurality of learning data (330).
  • the learning model may calculate a probability value by analyzing the characteristics of the marker region as main features.
  • the learning model may be a model in which a marker region is previously segmented and an image is analyzed centering on the corresponding region.
  • the learning model may be a model that gives attention to the marker region.
  • the analysis device may determine whether or not the drug is effective.
  • FIG. 5 is an example showing an image of a bone tissue mimetic used by a researcher as learning data and verification data.
  • the researcher treated one bone mimic with a drug (+drug), and did not treat the other mimic with a drug (NC).
  • each bone mimic with a fluorescent marker capable of detecting ⁇ -catenin and cell nuclei.
  • the researcher obtained images displaying ⁇ -catenin and nuclei in each bone tissue mimetic through an image acquisition device.
  • a model created with learning data and verification data is called a BN model.
  • the researcher made an image set using an image acquisition device with images in which ⁇ -catenin and nuclei were displayed in each bone tissue mimic and fluorescent images merged with the images.
  • a model created using this as learning data and verification data is called a BNM model.
  • the researcher used 420 sets of drug-treated and 424 sets of non-drug-treated bone tissue mimetic images to implement a learning model after going through an augmentation process.
  • the researcher used a method of dividing the bone tissue mimetic image into four parts as an augmentation process and randomly inverting left and right.
  • the researcher used 90% of the images as training data for deep learning model implementation and used the remaining 10% as verification data.
  • the researcher performed 10 cross validations to evaluate deep learning.
  • 6 is an example showing the structure of an implemented deep learning model.
  • the deep learning model consists of 3 convolution layers, 3 pooling layers, 1 drop layer, 1 flatten layer and 2 fully connected layers. -connected layer).
  • the convolution layer extracts features from an image and creates a feature map therefrom.
  • the pooling layer extracts the largest value or average value of the feature map values in order to reduce the size of the feature map created by the convolution layer or to emphasize specific data.
  • the drop layer uses only a part of the neural network model during training to prevent overfitting in the deep learning model.
  • the platen layer makes the features of the extracted data into one dimension.
  • the deep learning model analyzes the input bone tissue mimetic image and outputs a probability value, and the analysis device can determine whether or not the drug is effective based on the probability value.
  • ROC curves receiver operating characteristics curve
  • FPR (1-selectivity) is shown on the horizontal axis (X-axis), and sensitivity is shown on the vertical axis (Y-axis).
  • the analysis device 500 is an example of an analysis device 500 for determining the efficacy of a drug for bone disease by analyzing an image of a bone tissue mimetic.
  • the analysis device 500 corresponds to the above-described analysis device (300 in FIG. 1).
  • the analysis device 500 may be physically implemented in various forms.
  • the analysis device 500 may have a form of a computer device such as a PC, a network server, and a chipset dedicated to data processing.
  • the analysis device 500 may include a storage device 510, a memory 520, an arithmetic device 530, an interface device 540, a communication device 550, and an output device 560.
  • the storage device 510 may store an image of a bone tissue mimetic.
  • the storage device 510 may store a code or program for determining the efficacy of a drug by analyzing an image of a bone tissue mimetic.
  • the storage device 510 may store the learning model learned using the bone tissue mimetic image.
  • the memory 520 may store data and information generated during the process of the analysis device 500 analyzing the bone tissue mimetic image.
  • the arithmetic unit 530 may remove unnecessary regions by binarizing the input image of the bone tissue mimetic and performing erosion and expansion processing on the binarized image.
  • the arithmetic unit 530 may divide the input bone tissue matrix image into left and right parts based on the midpoint of the left end point and the right end point of the image.
  • the arithmetic device 530 may process the image of the bone tissue model by inverting it left and right.
  • the arithmetic device 530 may be a device such as a processor, an AP, or a chip in which a program is embedded that processes data and performs certain arithmetic operations.
  • the interface device 540 is a device that receives certain commands and data from the outside.
  • the interface device 540 may receive a bone tissue mimetic image from a physically connected input device or an external storage device.
  • the interface device 540 may transmit the result of evaluating the efficacy of the drug to an external object by analyzing the image of the bone tissue mimetic.
  • the communication device 550 refers to a component that receives or transmits certain information through a wired or wireless network.
  • the communication device 550 may receive a bone tissue matrix image from an external object. In addition, the communication device 550 may analyze the bone tissue mimetic image and transmit the result of evaluating the efficacy of the drug to an external object such as a user terminal.
  • the interface device 540 and the communication device 550 are components for exchanging certain data from a user or other physical object, they can also be collectively referred to as input/output devices. When limited to information or data input functions, the interface device 540 and the communication device 550 may also be referred to as input devices.
  • the output device 560 is a device that outputs certain information.
  • the output device 560 may output an interface required for data processing, an input bone tissue matrix image, analysis results, and the like.
  • the above-described method for constructing learning data and learning a learning model using augmented learning data may be implemented as a program (or application) including an executable algorithm that can be executed on a computer.
  • the program may be stored and provided in a temporary or non-transitory computer readable medium.
  • a non-transitory readable medium is not a medium that stores data for a short moment, such as a register, cache, or memory, but a medium that stores data semi-permanently and can be read by a device.
  • the various applications or programs described above are CD, DVD, hard disk, Blu-ray disk, USB, memory card, ROM (read-only memory), PROM (programmable read-only memory), EPROM (erasable PROM) or EEPROM It may be stored and provided in a non-transitory readable medium such as (electrically EPROM) or flash memory.
  • Temporary readable media include static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), and enhanced SDRAM (enhanced SDRAM). SDRAM, ESDRAM), sync link DRAM (SLDRAM), and direct rambus RAM (DRRAM).
  • Figure 11 shows an example of a structure manufacturing process for bone tissue mimetic.
  • a bone tissue replica (aka bone-on-a-chip) was fabricated using a hydrogel integrated unit based on a well plate.
  • the elastomer PDMS polydimethyl siloxane
  • the PDMS replica was removed, and an injection port having a diameter of 1 mm was formed for injecting the gel.
  • the PDMS replica was then split into a dumbbell shape using a custom punch.
  • a PDMS replica was placed in the center of the well plate using a custom jig made of polyether ether ketone (PEEK). Then, the PDMS replica was fixed on a jig and bonded by plasma treatment.
  • PEEK polyether ether ketone
  • the inner surface of the chip was coated with a Tris buffer solution containing 1 mg/mL dopamine hydrochloride and washed 5 times with distilled water.
  • the fabricated device was used after drying in an oven at 80 °C.
  • Mouse bone cells (IDG-SW3; Kerafast) were inactivated by heat treatment in a collagen-coated flask in 10% (v/v) FBS (fetal bovine serum; Gibco) and 50 U/mL recombinant mouse interferon gamma (IFN- ⁇ ).
  • FBS fetal bovine serum
  • IFN- ⁇ U/mL recombinant mouse interferon gamma
  • Protein Gel
  • 1% w/v
  • penicillin/streptomycin Thermo Fisher Scientific
  • Alpha Minimum Essential Medium Alpha-MEM
  • Gibco Alpha Minimum Essential Medium
  • Mouse osteoblasts (MC3T3-E1; ATCC) were cultured in Alpha minimum essential medium (Alpha-MEM) supplemented with 1% (w/v) penicillin/streptomycin (Thermo Fisher Scientific) in 10% (v/v) FBS (Gibco). ; Gibco) at 37°C, 5% (v/v) CO 2 and cultured in an incubator environment.
  • Alpha-MEM Alpha minimum essential medium
  • penicillin/streptomycin Thermo Fisher Scientific
  • FBS Gibco
  • Gibco 37°C, 5% (v/v) CO 2
  • interferon gamma was removed and 10% (v/v) FBS (Gibco), 50 ug/mL ascorbic acid, 4 mM beta-glycerophosphate ( ⁇ -glycerophosphate) and 1% (w/v) penicillin/streptomycin (Thermo Fisher Scientific) were added to the differentiation medium (alpha-MEM) at 37°C.
  • FIG. 12 is an example showing the process of extracting OB-dECM derived from osteoblasts.
  • mouse bone cells (MC3T3-E1) were cultured for 2 weeks, and then the cell sheet was removed.
  • the lyophilized OB-dECM was finely cut and mixed with 1 mg/mL pepsin (Sigma-Aldrich) in 0.1 M acetic acid, and then stirred for 12 hours at 500 rpm at 4°C using a magnetic bar.
  • Col/OB-dECM composite hydrogel was prepared by mixing 3-4 mg/mL rat tail collagen type I (Corning) and the OB-dECM solution in storage.
  • the optimal composition of the hydrogel is 2 mg/mL collagen and 1 mg/mL OB-dECM.
  • osteoblasts (Osteoblast, MC3T3-E1) were seeded to surround the outer edge of the gelated hydrogel, cultured in an osteogenic medium, and cultured in the medium every 2 days. was replaced to prepare a bone tissue mimetic.
  • the bone tissue mimic according to the present invention has a structure in which osteoblasts are present inside the hydrogel on the central support and osteoblasts are horizontally arranged around the hydrogel, and cell-cell interactions can be observed with an optical image. It can be seen that the structure has a
  • a conventional system Transwell system
  • the bone tissue simulating system of the present invention were constructed and a co-culture experiment was conducted.
  • 15 shows a schematic diagram of the system of the prior art and the system of the present invention (Bone-on-a-chip) used in this co-culture experiment.
  • the growth of osteoblasts in each system was confirmed by measuring cell activity over time by a Proliferation assay (CCK assay), where the activity was measured from absorbance (OD 450 ) using color development. The results are shown in FIG. 16 . From this, it was confirmed that the growth potential of osteoblast cells was significantly higher in the bone tissue mimetic system according to the present invention than in the prior art system during co-culture.
  • CCK assay Proliferation assay
  • SOST sclerostin
  • FIG. 13 is a schematic diagram of the action of SOST.
  • SOST secreted from osteocytes is delivered to osteoblasts, then inhibits the Frizzled/WNT signaling pathway, and finally promotes ⁇ -catenin degradation to induce transcription from genes. ) and, as a result, suppress the proliferation of osteoblasts.
  • a therapeutic agent for osteoporosis can be prepared as a monoclonal antibody that effectively binds to a specific site of SOST.
  • the osteoporosis treatment binds to SOST to prevent its function, ultimately prevents ⁇ -catenin from being decomposed, promotes the movement of ⁇ -catenin to the nucleus, and ultimately promotes bone growth through the proliferation of osteoblasts. play a shaping role.
  • the concentration of SOST released from bone cells was quantified by ELISA (enzyme-linked immunosorbent assay) as follows to determine the concentration of the drug to be treated (SOST-inhibiting monoclonal antibody).
  • the culture medium of bone cells (IDG-SW3) cultured in the bone tissue mimetic prepared in Examples 1-4 was collected every 2-3 days for 14 days, and then measured with the SOST ELISA Kit (ALPCO, USA). .
  • osteoblasts M3T3-E1
  • drugs SOST-inhibiting monoclonal antibody
  • osteoblasts IDG-SW3 and osteoblasts (MC3T3-E1) in the hydrogel were fixed for 20 minutes using 4% paraformaldehyde, followed by 0.1% (v/v) ) Triton X-100 was used for permeabilization, and blocking was performed using a PBS solution containing 5% BSA (Bovine serum albumin).
  • BSA Bovine serum albumin
  • ⁇ -Catenin was immunostained using a ⁇ -catenin antibody (1:50, Santa cruz Biotechnology, USA) attached with the fluorescent marker Alexa Fluor 488, and the nucleus was immunostained with the fluorescent marker Hoechst 33342 (1:500, 1:500). Thermo Fisher Scientific, USA).
  • Fluorescent images were obtained using a fluorescence microscope (Celena X high content imaging system, Logos Biosystems, Korea) and a confocal microscope (LSM-710, Carl Zeiss, Germany).
  • ⁇ -catenin nuclear translocation rates were calculated using the following formula.
  • ⁇ -catenin nuclear translocation rate I n / I t x 100, (I t : intensity of ⁇ -catenin in the cell, I n : intensity of ⁇ -catenin in the nucleus).
  • the image that can be obtained from this is learned through artificial intelligence with a high accuracy of 95% or more, whether or not there is an effect of drug treatment It was confirmed that it can determine, in particular, when using the bone tissue mimetic according to the present invention, compared to the bone tissue mimetic prepared by culturing cells by the existing method, by better mimicking the bone production mechanism, the drug candidate The present invention was completed by confirming that the model was very suitable for activity evaluation and screening.

Abstract

The present invention relates to a method and a device for screening for an osteoporosis drug by analyzing an image obtained from a bone tissue mimic capable of mimicking the function and structure of bone tissue. Particularly, according to one aspect of the present invention, provided is a method for evaluating a drug for preventing or treating bone disease, comprising the steps of: treating a three-dimensional bone tissue mimic with a drug; finding an intracellular marker related to bone disease such as osteoporosis from the drug-treated three-dimensional bone tissue mimic and pre-processing same; obtaining many images from the three-dimensional bone tissue mimic including the pre-processed intracellular marker; using the images so as to pre-process same into a data set for application to a deep learning algorithm; calculating feature maps by using a convolutional neural network (CNN); integrating the calculated feature maps; and using the integrated feature maps so as to provide information in which drug efficacy has been evaluated.

Description

골 조직 모사체 이미지 분석을 이용한 골다공증 약물 스크리닝 방법 및 장치Osteoporosis drug screening method and device using bone tissue mimetic image analysis
본 발명은 골 조직의 기능과 구조를 모방할 수 있는 골 조직 모사체로부터 얻어진 이미지를 분석하여 골다공증 약물을 스크리닝하는 방법 및 장치에 관한 것이다.The present invention relates to a method and apparatus for screening an osteoporosis drug by analyzing an image obtained from a bone tissue mimetic capable of mimicking the function and structure of bone tissue.
인구의 고령화로 인하여 골 질환이 급속도로 늘어나는 추세이다. 특히, 골다공증은 골 강도의 약화로 골절이 증가하게 되는 골격계 질환으로서 노인성 질환에 해당되며 전 세계적으로 중요한 보건학적 문제로 대두되고 있다. 한국 또한 고령화가 빨리 진행되고 있는 나라 중 하나로 골다공증 환자 역시 빠른 속도로 증가하고 있다. 전세계 인구 중 2억 명이 골다공증과 같은 골 질환을 겪고 있는 것으로 추산된다. 50대 이상은 20~30% 정도가 최소 한 번 이상의 골 질환을 겪는 것으로 조사되었으며, 향후 그 수가 더 늘어날 것으로 예상된다. 따라서, 골 질환 환자의 증가에 따라 효능이 뛰어난 치료 약제의 개발이 필요한 실정이다.Bone diseases are rapidly increasing due to the aging of the population. In particular, osteoporosis is a skeletal disease in which fractures increase due to weakening of bone strength, is a senile disease, and has emerged as an important public health problem worldwide. Korea is also one of the countries where the population is aging rapidly, and the number of osteoporosis patients is also increasing at a rapid pace. It is estimated that 200 million of the world's population suffer from bone diseases such as osteoporosis. About 20 to 30% of those over 50 years old have been found to suffer from at least one bone disease, and the number is expected to increase further in the future. Therefore, with the increase in patients with bone diseases, there is a need to develop therapeutic agents with excellent efficacy.
한편, 최근 미세 유체 기술의 발전으로 장기의 주요 기능 단위를 모델링 할 수 있는 장기 칩(organ-on-a-chip)이 개발되고 있다. 이와 관련하여, 한국공개특허 10-2015-0020702 뼈 결합 조직 세포를 포함하는 인간의 결합 조직의 형상을 모방한 3차원 생체 결합 조직 구성체를 개시하고 있으나, 이는 단순한 3차원 스케폴드 형태로 구성되는 것에 불과하여 골 조직 내의 세포와 세포 간의 상호작용을 관찰하거나 뼈의 기능적 향상 등을 확인하기 위한 분석법에 활용하기에 용이하지 않다는 문제가 있다.Meanwhile, with the recent development of microfluidic technology, an organ-on-a-chip capable of modeling major functional units of an organ is being developed. In this regard, Korean Patent Publication No. 10-2015-0020702 discloses a three-dimensional biological connective tissue construct that mimics the shape of human connective tissue including bone connective tissue cells, but it is composed of a simple three-dimensional scaffold. However, there is a problem in that it is not easy to use for an analysis method for observing cell-cell interactions in bone tissue or confirming functional improvement of bone.
따라서, 효과가 우수한 골 질환 치료 약물을 효율적으로 개발하고, 이의 효능 분석 및 평가를 고효율로 수행할 수 있기 위해서는 실제 골 조직의 구성적 배치, 기능적 특징을 정확히 모사할 수 있는 3차원 골 조직 모사체의 개발 및 이를 이용한 효율적인 후보 약물의 평가 방법이 절실히 필요한 상황이다.Therefore, in order to efficiently develop highly effective bone disease treatment drugs and to perform efficacy analysis and evaluation with high efficiency, a three-dimensional bone tissue mimetic capable of accurately replicating the structural arrangement and functional characteristics of actual bone tissue There is an urgent need for the development of a drug candidate and an evaluation method for an efficient candidate drug using the same.
상기와 같은 문제의 해결을 위해, 본 발명자들은 3차원 골조직 모사체로부터 얻어진 이미지를 인공 지능에 의해 분석하는 방식으로 골다공증과 같은 골 질환의 치료 또는 예방용 약물을 평가 및 스크리닝함으로써 본 발명을 완성하였다.In order to solve the above problems, the present inventors have completed the present invention by evaluating and screening drugs for treating or preventing bone diseases such as osteoporosis by analyzing images obtained from a three-dimensional bone tissue mimic by artificial intelligence. .
그러나 본 발명이 이루고자 하는 기술적 과제는 이상에서 언급한 과제에 제한되지 않으며, 언급되지 않은 또 다른 과제들은 아래의 기재로부터 당업자에게 명확하게 이해될 수 있을 것이다.However, the technical problem to be achieved by the present invention is not limited to the above-mentioned problems, and other problems not mentioned will be clearly understood by those skilled in the art from the following description.
본 발명은 골조직 모사체에 골질환에 관한 약물을 처리하는 단계; 약물을 처리한 골조직 모사체로부터 영상 또는 이미지 확보를 위해, 골조직 모사체를 마커로 표지하는 단계; 표지된 마커를 포함하는 골조직 모사체 영상 또는 이미지를 확보하는 단계; 상기 골조직 모사체 영상 또는 이미지를 딥러닝 모델로 학습시켜 특징 맵(feature map)을 산출하는 단계; 및 상기 특징 맵을 이용하여, 약물 효능 평가 정보를 제공하는 단계;를 포함하는, 골 질환의 예방 또는 치료용 약물의 평가 방법을 제공한다.The present invention comprises the steps of treating a bone tissue mimic with a drug related to bone disease; Labeling the bone tissue matrix with a marker to obtain an image or image from the drug-treated bone tissue matrix; Obtaining a bone tissue matrix image or image including the labeled marker; Calculating a feature map by learning the bone tissue mimetic image or image as a deep learning model; and providing drug efficacy evaluation information using the feature map.
본 발명에 따른 일 예로, 상기 골조직 모사체는 하판; 상기 하판의 상측에 위치한 상판; 상기 하판 또는 상판 중 어느 하나로부터 유래하여 상판과 하판 사이에 위치하는 돌기구조로 이루어진 배리어부; 상기 상판의 일측에 구비된 하나 이상의 세포시료 주입 가이드부;를 포함하고, 상기 상판의 크기는 상기 배리어부의 내측만을 폐쇄하는 크기여서, 배리어부의 외측 부분은 개방된 구조인 것을 특징으로 하는 세포배양용기에서 배양되는 것일 수 있다.As an example according to the present invention, the bone tissue mimetic lower plate; an upper plate located above the lower plate; a barrier portion formed of a protruding structure derived from either the lower plate or the upper plate and located between the upper plate and the lower plate; One or more cell sample injection guides provided on one side of the top plate, wherein the size of the top plate is such that only the inside of the barrier part is closed, so that the outer part of the barrier part has an open structure. It may be cultured in
본 발명에 따른 일 예로, 상기 골조직 모사체는 골세포(osteocyte)를 포함하며, 중심부에 충진된 젤(gel); 및 상기 젤의 외측부를 둘러싸도록 배치된 세포 혼합물;을 포함할 수 있다.As an example according to the present invention, the bone tissue matrix includes a bone cell (osteocyte), filled in the center of the gel (gel); and a cell mixture disposed to surround the outer portion of the gel.
본 발명에 따른 일 예로, 상기 골 질환은 골다공증, 불완전 골형성증(osteogenesis imperfecta), 과골화증, 고칼슘혈증, 부갑상선 기능 항진증, 골연화증, 융해성 골질환, 골괴사증, 뼈의 파젯병, 골 골절, 류마티스 관절염, 골수염, 치주성 골 손실, 암에 의한 골 손실, 노인성 골 손실 및 구루병으로 이루어진 군에서 선택되는 것일 수 있다.As an example according to the present invention, the bone disease is osteoporosis, incomplete osteogenesis imperfecta, hyperossification, hypercalcemia, hyperparathyroidism, osteomalacia, dissolving bone disease, osteonecrosis, Paget's disease of bone, bone fracture, It may be selected from the group consisting of rheumatoid arthritis, osteomyelitis, periodontal bone loss, bone loss due to cancer, senile bone loss, and rickets.
본 발명에 따른 일 예로, 상기 세포 혼합물은 골원성세포, 조골세포, 파골세포, 면역세포 및 혈관세포로 이루어진 군에서 선택되는 적어도 하나 이상을 포함하는 것일 수 있다.As an example according to the present invention, the cell mixture may include at least one or more selected from the group consisting of osteogenic cells, osteoblasts, osteoclasts, immune cells, and vascular cells.
본 발명에 따른 일 예로, 상기 딥러닝 모델은 골 질환의 예방 또는 치료 효과가 알려진 약물이 처리된 골조직 모사체의 영상 또는 이미지와 상기 약물이 처리되지 않은 골조직 모사체의 영상 또는 이미지를 학습데이터로 이용하여 학습된 것일 수 있다.As an example according to the present invention, the deep learning model uses images or images of a bone mimic treated with a drug known to have a preventive or therapeutic effect on bone disease and an image or image of a bone mimic not treated with the drug as learning data. It may have been learned using
본 발명에 따른 일 예로, 상기 약물은 상기 골조직 모사체 중심부의 골세포(osteocyte) 또는 외측부의 세포 혼합물 중 어느 하나 이상에 대해 DNA 합성, RNA 합성, 유전자 발현, 유전자 구조, 단백질 합성, 단백질 변형, 단백질 분비, 단백질 구조, 당 합성, 당 분비, 당 변형, 지질 분비, 세포막 합성, 세포 내 신호 전달, 세포 간 신호 전달, 세포 내 소기관, 세포의 분화, 세포의 분열 및 세포의 운동으로 이루어진 군에서 선택되는 어느 하나 이상에 영향을 주는 것일 수 있다.As an example according to the present invention, the drug is used for DNA synthesis, RNA synthesis, gene expression, gene structure, protein synthesis, protein modification, In the group consisting of protein secretion, protein structure, sugar synthesis, sugar secretion, sugar modification, lipid secretion, cell membrane synthesis, intracellular signal transduction, intercellular signal transduction, intracellular organelle, cell differentiation, cell division, and cell movement It may be one that affects one or more of the selected ones.
본 발명에 따른 일 예로, 상기 약물은 골 질환의 예방 또는 치료를 위한 천연물, 합성의약품, 복합의약품, 한약, 한약 추출물, 한약 제제, 생약, 단백질 의약품, 유전자 재조합 의약품, 세포 배양 의약품, 효소 의약품, 미생물 의약품, 항체 의약품, 호르몬 의약품, 방사선 의약품, 항체-약물 접합체, 세포 치료제 및 유전자 치료제로 이루어진 군에서 선택되는 것일 수 있다.As an example according to the present invention, the drug is a natural product, synthetic drug, combination drug, herbal medicine, herbal medicine extract, herbal preparation, herbal medicine, protein medicine, gene recombinant medicine, cell culture medicine, enzyme medicine, It may be selected from the group consisting of microbial drugs, antibody drugs, hormone drugs, radiation drugs, antibody-drug conjugates, cell therapy products, and gene therapy products.
본 발명에 따른 일 예로, 상기 마커는 염색약, 형광 물질, 인광 물질 및 방사능 물질로 이루어진 군에서 선택되는 어느 하나 이상을 더 첨가하여 얻는 것일 수 있다.As an example according to the present invention, the marker may be obtained by further adding any one or more selected from the group consisting of a dye, a fluorescent material, a phosphorescent material, and a radioactive material.
본 발명에 따른 일 예로, 상기 골조직 모사체 영상 또는 이미지는 서로 다른 대상을 식별하는 둘 이상의 마커를 처리하여 얻는 것일 수 있다.As an example according to the present invention, the bone tissue matrix image or image may be obtained by processing two or more markers for identifying different objects.
본 발명에 따른 일 예로, 상기 마커로 표지된 영역은 상기 골조직 모사체의 세포, 핵, 리보솜, 리소좀, 골지체, 중심체, 세포막, 미토콘드리아, 미세소판, 미세섬유, 세포질, DNA, RNA, 핵산, 히스톤, 단백질, 당 단백질, 막 단백질, 탄수화물, 막 탄수화물, 지질, 콜레스테롤, 당 지질, 당 및 콜라겐으로 이루어진 군에서 선택되는 적어도 하나 이상을 나타내는 것일 수 있다.As an example according to the present invention, the region labeled with the marker is a cell, nucleus, ribosome, lysosome, Golgi body, centrosome, cell membrane, mitochondria, microplatelet, microfilament, cytoplasm, DNA, RNA, nucleic acid, histone of the bone tissue mimetic , It may represent at least one selected from the group consisting of proteins, glycoproteins, membrane proteins, carbohydrates, membrane carbohydrates, lipids, cholesterol, glycolipids, sugars, and collagen.
본 발명에 따른 일 예로, 상기 마커로 표지된 영역은 상기 골조직 모사체의 형상, 모양, 위치 및 움직임으로 이루어진 군에서 선택되는 적어도 하나 이상을 나타내는 것일 수 있다.As an example according to the present invention, the area marked with the marker may represent at least one selected from the group consisting of the shape, shape, position and movement of the bone tissue mimetic.
본 발명에 따른 일 예로, 상기 특징 맵(feature map)은 상기 골조직 모사체의 영상 또는 이미지를 컨볼루셔널 신경망(CNN)을 이용하여 연산한 결과를 나타내는 것일 수 있다.As an example according to the present invention, the feature map (feature map) may indicate the result of calculating the image or image of the bone tissue mimetic using a convolutional neural network (CNN).
본 발명의 다른 양태로, 약물이 처리된 골조직 모사체 영상 또는 이미지를 입력 받는 입력장치; 상기 골조직 모사체 영상 또는 이미지를 이용하여 학습되어, 약물 효능을 평가할 수 있는 딥러닝 모델을 저장하는 저장장치; 상기 입력장치로부터 골조직 모사체 영상 또는 이미지를 전송받아, 상기 저장장치에 저장된 딥러닝 모델에 입력하고, 상기 딥러닝 모델이 출력하는 값에 따라 약물의 골조직 모사체에 대한 영향을 판단하는 연산장치;를 포함하는, 골 질환의 예방 또는 치료용 약물의 평가 장치를 제공한다.In another aspect of the present invention, an input device for receiving a drug-treated bone tissue mimetic image or image; a storage device for storing a deep learning model that can be learned using the bone tissue mimetic image or image and evaluate drug efficacy; an arithmetic device that receives a bone mimic image or image from the input device, inputs the image to the deep learning model stored in the storage device, and determines the effect of the drug on the bone mimetic according to a value output from the deep learning model; It provides a device for evaluating drugs for preventing or treating bone diseases, including a.
본 발명에 따른 일 예로, 상기 골조직 모사체는 골세포(osteocyte)를 포함하며, 중심부에 충진된 젤(gel); 및 상기 젤의 외측부를 둘러싸도록 배치된 세포 혼합물;을 포함할 수 있다.As an example according to the present invention, the bone tissue matrix includes a bone cell (osteocyte), filled in the center of the gel (gel); and a cell mixture disposed to surround the outer portion of the gel.
본 발명에 따른 일 예로, 상기 골조직 모사체 영상 또는 이미지는 약물이 처리된 골조직 모사체에 마커로 표지된 영역을 포함할 수 있다.As an example according to the present invention, the bone tissue mimetic image or image may include a region marked with a marker on the drug-treated bone tissue mimetic.
본 발명에 따른 일 예로, 상기 세포 혼합물은 골원성세포, 조골세포, 파골세포, 면역세포 및 혈관세포로 이루어진 군에서 선택되는 적어도 하나 이상을 포함할 수 있다.As an example according to the present invention, the cell mixture may include at least one selected from the group consisting of osteogenic cells, osteoblasts, osteoclasts, immune cells, and vascular cells.
본 발명에 따른 일 예로, 상기 마커는 염색약, 형광 물질, 인광 물질 및 방사능 물질로 이루어진 군에서 선택되는 어느 하나 이상을 더 첨가하여 얻는 것일 수 있다.As an example according to the present invention, the marker may be obtained by further adding any one or more selected from the group consisting of a dye, a fluorescent material, a phosphorescent material, and a radioactive material.
본 발명에 따른 일 예로, 상기 골조직 모사체 영상 또는 이미지는 서로 다른 대상을 식별하는 둘 이상의 마커를 처리하여 얻는 것일 수 있다.As an example according to the present invention, the bone tissue matrix image or image may be obtained by processing two or more markers for identifying different objects.
본 발명의 다른 양태로, 골세포(osteocyte)를 포함하며 중심부에 충진된 젤(gel); 및 상기 젤의 외측부를 둘러싸도록 배치된 세포 혼합물; 을 포함하는, 골조직 모사체를 제공한다.In another aspect of the present invention, a gel containing bone cells (osteocyte) and filled in the center (gel); and a cell mixture arranged to surround the outer portion of the gel. Containing, it provides a bone tissue mimic.
본 발명에 따른 일 예로, 상기 젤은 하이드로 젤일 수 있다.As an example according to the present invention, the gel may be a hydrogel.
본 발명에 따른 일 예로, 상기 하이드로 젤은 마트리젤(Matrigel), 푸라매트릭스(Puramatrix), 콜라겐(Collagen), 피브린젤(Fibrin gel), 폴리에틸렌글리콜 디아크릴레이트(PEG-DA), 폴리에틸렌글리콜 디메사크릴레이트(PEG-DMA), 폴리나이팜 (PNIPAM), 폴록세이머(Poloxamer), 키토산(Chitosan), 아가로스(Agarose), 젤라틴(Gelatin), 히알루론산(Hyaluronic acid) 및 알지네이트(Alginate)로 이루어진 군에서 선택되는 어느 하나 이상을 포함할 수 있다.As an example according to the present invention, the hydrogel is Matrigel, Puramatrix, collagen, fibrin gel, polyethylene glycol diacrylate (PEG-DA), polyethylene glycol dimesa Composed of PEG-DMA, PNIPAM, Poloxamer, Chitosan, Agarose, Gelatin, Hyaluronic acid and Alginate It may include any one or more selected from the group.
본 발명에 따른 일 예로, 상기 하이드로 젤은 세포외기질(extracellular matrix; ECM)을 더 포함할 수 있다.As an example according to the present invention, the hydrogel may further include an extracellular matrix (ECM).
본 발명에 따른 일 예로, 상기 세포 혼합물은 조골세포 및 파골세포 중 어느 하나 이상과 배양액을 포함할 수 있다.As an example according to the present invention, the cell mixture may include at least one of osteoblasts and osteoclasts and a culture medium.
본 발명의 다른 양태로, (S1) 골세포(Osteocyte), 세포외기질(extracellular matrix; ECM) 및 하이드로젤을 혼합하는 단계; (S2) 상기 혼합물을 지지체 상에 적가한 후 젤화(gelation)하는 단계; (S3) 젤화된 하이드로젤 외측부 가장자리를 둘러싸도록 조골세포(Osteoblast)를 시딩(seeding)하는 단계; 및 (S4) 배양액을 주입하는 단계;를 포함하는, 골조직 모사체의 제조방법을 제공한다.In another aspect of the present invention, (S1) mixing bone cells (Osteocyte), extracellular matrix (ECM) and hydrogel; (S2) dropwise addition of the mixture onto a support and then gelation; (S3) seeding osteoblasts to surround the outer edge of the gelled hydrogel; And (S4) injecting the culture solution; provides a method for producing a bone tissue mimic, including.
본 발명에 따른 일 예로, 상기 (S1) 단계에서 마트리젤(matrigel)을 더 포함하여 혼합하는 것일 수 있다.As an example according to the present invention, in step (S1), matrigel may be further included for mixing.
본 발명에 따른 일 예로, 상기 하이드로젤은 콜라겐(collagen)일 수 있다.As an example according to the present invention, the hydrogel may be collagen.
본 발명에 따른 일 예로, 상기 (S2) 단계의 젤화는 pH 6 내지 8의 범위에서 이루어지는 것일 수 있다.As an example according to the present invention, the gelation in step (S2) may be performed in the range of pH 6 to 8.
본 발명에 따른 일 예로, 상기 (S3) 단계는 젤화된 하이드로젤 외측부 가장자리를 둘러싸도록 조골세포(osteoblast)와 함께 파골세포(osteoclast)를 접종하는 것일 수 있다.As an example according to the present invention, the step (S3) may be inoculating osteoclasts together with osteoblasts to surround the outer edge of the gelled hydrogel.
본 발명에 따르면 딥러닝 알고리즘에 기반하여 얻어진 정보를 이용할 수 있으므로, 골다공증과 같은 골질환의 예방 또는 치료용 의약품을 평가하거나 신규 약물의 개발을 위한 약물을 스크리닝하는 방법이 효율적이고 높은 정확도로 수행될 수 있다. 또한, 본 발명의 방법은 이미지에 기반한 평가 또는 스크리닝 방법으로써 사용이 간편하다. 더욱이, 골세포와 조골세포를 수직 방향으로 나열하여 키우는 종래의 공배양 모델이나 Tran swell형 모델은 두께가 있어서 이미징 분석에 어려움이 있었는데 반해, 실제 뼈 조직의 기능적 특징을 모사할 수 있도록 골세포가 포함된 겔과 세포혼합물을 수평적으로 배치하는 배치관계를 주요 구성으로 하는 골 조직 모사체를 이용함으로써 실제와 유사하게 골 조직 내에서 일어나는 세포-세포, 세포-ECM 간 상호 작용을 용이하고 효과적으로 관찰할 수 있다.According to the present invention, since information obtained based on a deep learning algorithm can be used, a method for evaluating drugs for preventing or treating bone diseases such as osteoporosis or screening drugs for developing new drugs can be performed efficiently and with high accuracy. can In addition, the method of the present invention is easy to use as an image-based evaluation or screening method. Moreover, conventional co-culture models or trans-swell-type models in which bone cells and osteoblasts are arranged in a vertical direction have difficulties in imaging analysis due to their thickness, whereas osteocytes can mimic the functional characteristics of actual bone tissue. Easy and effective observation of cell-cell and cell-ECM interactions occurring in bone tissue similar to reality by using a bone tissue mimetic whose main composition is the arrangement relationship in which the included gel and cell mixture are placed horizontally can do.
도 1은 본 발명에 따른 골조직 모사체 이미지(영상) 분석을 이용한 골질환 약물 스크리닝 방법 및 장치를 도시한 예이다.1 is an example showing a method and apparatus for screening a drug for bone disease using bone tissue mimetic image (image) analysis according to the present invention.
도 2는 골조직 모사체의 구조를 도시한 예이다.Figure 2 is an example showing the structure of the bone tissue matrix.
도 3은 골조직 모사체에서 영상을 추출하는 과정을 도시한 예이다.3 is an example illustrating a process of extracting an image from a bone tissue matrix.
도 4는 골조직 모사체 영상을 이용하여 골질환 약물의 효능을 평가하는 딥러닝 모델을 구축하는 과정을 도시한 예이다.4 is an example illustrating a process of constructing a deep learning model for evaluating the efficacy of a drug for bone disease using a bone tissue mimetic image.
도 5는 딥러닝 모델을 구축하는 과정에서 사용된 골조직 모사체 영상을 도시한 예이다.5 is an example showing an image of a bone tissue mimetic used in the process of building a deep learning model.
도 6은 딥러닝 모델의 구성을 도시한 예이다.6 is an example showing the configuration of a deep learning model.
도 7은 구현된 BN 모델에서 훈련횟수에 따른 정확도 및 손실도를 나타낸 결과이다.7 is a result showing accuracy and loss according to the number of trainings in the implemented BN model.
도 8은 구현된 BNM 모델에서 훈련횟수에 따른 정확도 및 손실도를 나타낸 결과이다.8 is a result showing accuracy and loss according to the number of trainings in the implemented BNM model.
도 9는 상기 BN 모델 (β-카테닌+핵)및 BNM 모델 (β-카테닌+핵 이미지 병합)의 ROC 곡선을 도시한 예이다.9 is an example showing ROC curves of the BN model (β-catenin+nucleus) and the BNM model (β-catenin+nuclear image merge).
도 10은 골조직 모사체를 이용한 약물 효능에 대한 평가장치에 대한 구성을 도시한 예이다.10 is an example showing the configuration of an evaluation device for drug efficacy using a bone tissue mimetic.
도 11은 골조직 모사체의 구조물을 제작하는 과정을 도시한 예이다.11 is an example illustrating a process of manufacturing a structure of a bone tissue mimic.
도 12는 조골세포에서 조골세포 유래의 탈세포된 세포외 기질(osteoblast-derived decellularized extracellular matrix; OB-dECM) 추출 과정을 도시한 예이다.12 is an example illustrating a process of extracting an osteoblast-derived decellularized extracellular matrix (OB-dECM) from osteoblasts.
도 13은 골세포(osteocyte)에서 분비된 SOST에 의한 조골세포(osteoblast) 내에서의 반응 및 골다공증 유발 약물로서 SOST 단일 클론 항체를 처리하였을 때의 모습을 도시한 예이다.13 is an example showing the reaction in osteoblasts by SOST secreted from osteocytes and the state of treatment with a SOST monoclonal antibody as an osteoporosis-inducing drug.
도 14는 골조직 모사체에서 골세포의 성숙에 따른 조골세포의 접종 시기, 약물의 처리 시기 및 촬영 시기를 도시한 예이다.Figure 14 is an example showing the timing of inoculation of osteoblasts according to the maturation of osteoblasts in the bone tissue matrix, the timing of drug treatment, and the photographing timing.
도 15는 종래 기술의 시스템과 본 발명의 시스템(Bone-on-a-chip)에 대한 모식도를 나타낸 예이다.15 is an example of a schematic diagram of a prior art system and a system (Bone-on-a-chip) of the present invention.
도 16은 조골 세포의 성장을 Proliferation assay (CCK assay)에 의해 시간의 경과에 따른 세포 활성도를 측정하여 확인한 결과이다.16 is a result of confirming the growth of osteoblasts by measuring cell activity over time by a Proliferation assay (CCK assay).
도 17은 종래 기술의 시스템과 본 발명의 시스템에서 골 세포의 세포 성숙도를 골분화 마커를 이용하여 각각 측정한 결과이다.17 is a result of measuring cell maturity of bone cells in the system of the prior art and the system of the present invention using osteogenic differentiation markers, respectively.
도 18은 본 발명에 따른 골조직 모사체의 배양을 위한 세포배양용기를 예시한 것이다.18 illustrates a cell culture container for culturing bone tissue mimics according to the present invention.
이하 첨부한 표 또는 도면들을 참조하여 본 발명의 골 조직 모사체 이미지 분석을 이용한 골다공증 약물 스크리닝 방법 및 장치에 대해 상세히 설명한다.Hereinafter, the osteoporosis drug screening method and apparatus using the bone tissue mimetic image analysis of the present invention will be described in detail with reference to the accompanying tables or drawings.
도면이 기재되어 있을 경우, 이는 당업자에게 본 발명의 사상이 충분히 전달될 수 있도록 하기 위해 예로서 제공되는 것이다. 따라서 본 발명은 제시되는 도면들에 한정되지 않고 다른 형태로 구체화될 수도 있으며, 상기 도면들은 본 발명의 사상을 명확히 하기 위해 과장되어 도시될 수 있다.When drawings are described, they are provided as examples so that the spirit of the present invention can be sufficiently conveyed to those skilled in the art. Therefore, the present invention may be embodied in other forms without being limited to the drawings presented, and the drawings may be exaggerated to clarify the spirit of the present invention.
이때, 사용되는 기술 용어 및 과학 용어에 있어서 다른 정의가 없다면, 이 발명이 속하는 기술분야에서 통상의 지식을 가진 자가 통상적으로 이해하고 있는 의미를 가지며, 하기의 설명 및 첨부 도면에서 본 발명의 요지를 불필요하게 흐릴 수 있는 공지 기능 및 구성에 대한 설명은 생략한다.At this time, unless there is another definition in the technical terms and scientific terms used, they have meanings commonly understood by those of ordinary skill in the art to which this invention belongs, and the gist of the present invention is described in the following description and accompanying drawings. Descriptions of well-known functions and configurations that may be unnecessarily obscure are omitted.
또한 본 발명의 명세서에서 사용되는 단수 형태는 문맥에서 특별한 지시가 없는 한 복수 형태도 포함하는 것으로 의도할 수 있다. In addition, singular forms used in the specification of the present invention may be intended to include plural forms as well unless otherwise indicated in the context.
또한 본 발명의 명세서에서 특별한 언급 없이 사용된 단위는 중량을 기준으로 하며, 일 예로 % 또는 비의 단위는 중량% 또는 중량비를 의미한다.Also, in the specification of the present invention, units used without particular notice are based on weight, and for example, % or a unit of ratio means weight% or weight ratio.
또한 본 발명의 명세서에서, “포함한다”는 표현은 “구비한다”, “함유한다”, “가진다” 또는 “특징으로 한다” 등의 표현과 등가의 의미를 가지는 개방형 기재이며, 추가로 열거되어 있지 않은 요소, 재료 또는 공정을 배제하지 않는다. 또한 “실질적으로…로 구성된다”는 표현은 특정된 요소, 재료 또는 공정과 함께 열거되어 있지 않은 다른 요소, 재료 또는 공정이 발명의 적어도 하나의 기본적이고 신규한 기술적 사상에 허용할 수 없을 만큼의 현저한 영향을 미치지 않는 양으로 존재할 수 있는 것을 의미한다. 또한 “구성된다”는 표현은 기재된 요소, 재료 또는 공정만이 존재하는 것을 의미한다.In addition, in the specification of the present invention, the expression "comprises" is an open description having a meaning equivalent to expressions such as "includes", "includes", "has" or "characterized by", and is further listed It does not exclude elements, materials or processes that do not exist. Also, “actually… The expression "consists of" means that the specified element, material or process together with other elements, materials or processes not listed do not significantly affect at least one basic and novel technical idea of the invention in an unacceptably significant manner. It means that it can exist in quantity. Also, the expression “consisting of” means that only the described elements, materials or processes are present.
본 발명의 명세서에서 사용된 용어, "성분", "조성물", "화합물의 조성물", "화합물", "약물", "약학적 활성제", "활성제", "치유" "치료법" "치료" 또는 "약제"는 대상체(인간 또는 동물)에 투여될 때 국소 및/또는 전신 작용에 의해 원하는 약학적 및/또는 생리학적 효과를 유도하는 화합물 또는 화합물(들) 또는 물질의 조성물을 의미하기 위해 상호교환적으로 사용된다.As used herein, the terms "ingredient", "composition", "composition of a compound", "compound", "drug", "pharmaceutical active", "active agent", "healing" "treatment" "treatment" or “medicament” is used interchangeably to mean a compound or compound(s) or composition of matter that, when administered to a subject (human or animal), induces a desired pharmacological and/or physiological effect by local and/or systemic action. used interchangeably.
본 발명은 하나의 양태로, 3차원 골 조직 모사체에 약물을 처리하는 단계; 약물을 처리한 3차원 골 조직 모사체로부터 골다공증과 같은 골질환에 관련된 세포내 마커를 찾아 전처리하는 단계; 상기 전처리된 세포내 마커를 포함하는 3차원 골 조직 모사체로부터 대량의 이미지를 확보하는 단계; 상기 이미지를 이용하여 딥러닝 알고리즘에 적용하기 위한 데이터 세트로 전처리하는 단계; 콘볼루셔널 신경망(CNN: convolutional neural network)을 이용하여 특징 맵(Feature map)을 산출하는 단계; 산출된 특징맵을 통합하는 단계; 및 통합된 특징 맵을 이용하여 약물 효능을 평가한 정보를 제공하는 단계를 포함하는, 골 질환의 예방 또는 치료용 약물의 평가 방법이 제공된다.In one aspect, the present invention includes the steps of treating a drug to a three-dimensional bone tissue mimetic; Searching for and pre-processing intracellular markers related to bone diseases such as osteoporosis from the drug-treated three-dimensional bone tissue mimetic; obtaining a large number of images from the 3-dimensional bone tissue mimetic containing the pre-processed intracellular marker; pre-processing the image into a data set for application to a deep learning algorithm; Calculating a feature map using a convolutional neural network (CNN); integrating the calculated feature maps; and providing information on evaluating drug efficacy using the integrated feature map.
본 발명의 다른 일 양태로, 3차원 골 조직 모사체에 약물 후보 물질을 처리하는 단계; 상기 처리한 3차원 골 조직 모사체로부터 골다공증과 같은 골질환에 관련된 세포내 마커를 찾아 전처리하는 단계; 상기 전처리된 세포내 마커를 포함하는 3차원 골 조직 모사체로부터 대량의 이미지를 확보하는 단계; 상기 이미지를 이용하여 딥러닝 알고리즘에 적용하기 위한 데이터 세트로 전처리하는 단계; 콘볼루셔널 신경망(CNN: convolutional neural network)을 이용하여 특징 맵(Feature map)을 산출하는 단계; 산출된 특징맵을 통합하는 단계; 및 통합된 특징 맵을 이용하여 약물 효능을 평가한 정보를 제공하는 단계를 포함하는, 골 질환의 예방 또는 치료 활성 약물의 스크리닝 방법이 제공된다.In another aspect of the present invention, processing a drug candidate to the three-dimensional bone tissue mimetic; pre-processing to find intracellular markers related to bone diseases such as osteoporosis from the treated 3-dimensional bone tissue mimetic; obtaining a large number of images from the 3-dimensional bone tissue mimetic containing the pre-processed intracellular marker; pre-processing the image into a data set for application to a deep learning algorithm; Calculating a feature map using a convolutional neural network (CNN); integrating the calculated feature maps; and providing information on evaluating drug efficacy using the integrated feature map.
본 발명의 일 구현예에 따르면, 상기 약물 평가 방법 및 상기 약물 스크리닝 방법에서 상기 3차원 골 조직 모사체에 약물을 처리하는 단계는 상기 3차원 골 조직 모사체에 약물 또는 약물 후보 물질을 주입하는 단계일 수 있다.According to one embodiment of the present invention, in the drug evaluation method and the drug screening method, the step of treating the 3-dimensional bone tissue mimetic with the drug is the step of injecting the drug or drug candidate into the 3-dimensional bone tissue mimetic. can be
본 발명의 다른 일 구현예에 따르면, 상기 약물 평가 방법 및 상기 약물 스크리닝 방법에서 상기 세포내 마커의 전처리는 상기 세포내 마커를 염색하는 것일 수 있다.According to another embodiment of the present invention, the pretreatment of the intracellular marker in the drug evaluation method and the drug screening method may be staining the intracellular marker.
본 발명의 다른 일 구현예에 따르면, 상기 약물 평가 방법 및 상기 약물 스크리닝 방법에서 상기 세포내 마커는 예를 들어 β-카테닌일 수 있다.According to another embodiment of the present invention, in the drug evaluation method and the drug screening method, the intracellular marker may be, for example, β-catenin.
본 발명의 다른 일 구현예에 따르면, 상기 약물 평가 방법 및 상기 약물 스크리닝 방법에서 상기 세포내 마커는 예를 들어 β-카테닌이고, 상기 전처리된 세포내 마커가 약물 평가 지표일 수 있다.According to another embodiment of the present invention, in the drug evaluation method and the drug screening method, the intracellular marker is, for example, β-catenin, and the pretreated intracellular marker may be a drug evaluation indicator.
본 발명의 일 구현예에 따르면 상기 골조직 모사체는 골세포(Osteocyte)를 포함하며 중심부에 충진된 젤(gel); 및 상기 젤의 외측부를 둘러싸도록 배치된 세포 혼합물;을 포함하는 것일 수 있다. 상기 골조직 모사체는 뼈 조직을 잘 모사하는 생체 모방 구조체이면서, 고속 분석에 친화적인 다중 웰 구조의 웰플레이트 형상으로 제작할 수가 있다. 따라서, 고속 장비와 연계하여 실제로 대량의 샘플에서 다양한 생물학적 분석(예를 들어 세포 활성, 독성, 이미징 분석 등)에 적용 가능하며 분석 효율을 높일 수 있으나, 본 발명에서 골 조직 모사체의 제작이 상기 방법으로 한정되는 것은 아니다.According to one embodiment of the present invention, the bone tissue matrix includes a bone cell (Osteocyte) and filled in the center gel (gel); and a cell mixture disposed to surround the outer portion of the gel. The bone tissue mimetic is a biomimetic structure that well mimics bone tissue, and can be manufactured in a well plate shape with a multi-well structure that is friendly to high-speed analysis. Therefore, in conjunction with high-speed equipment, it can be applied to various biological analyzes (eg, cell activity, toxicity, imaging analysis, etc.) in a large amount of samples, and the analysis efficiency can be increased. method is not limited.
이에 따라, 이하 설명하는 기술의 다른 구현예에 따르면 상기 골조직 모사체는 다중 웰(multi-well) 구조의 웰플레이트로 형성될 수 있으며, 이를 이용한 골 질환의 예방 또는 치료 활성 물질의 스크리닝 방법을 제공한다.Accordingly, according to another embodiment of the technology described below, the bone tissue mimetic may be formed in a well plate having a multi-well structure, and a method for screening an active substance for preventing or treating bone diseases using the same is provided. do.
상기 젤은 하이드로젤일 수 있으며, 상기 하이드로젤은 마트리젤(Matrigel), 푸라매트릭스(Puramatrix), 콜라겐(Collagen), 피브린젤(Fibrin gel), 폴리에틸렌글리콜 디아크릴레이트(PEG-DA), 폴리에틸렌글리콜 디메사크릴레이트(PEG-DMA), 폴리나이팜 (PNIPAM), 폴록세이머 (Poloxamer), 키토산(Chitosan), 아가로스(Agarose), 젤라틴(Gelatin), 히알루론산 (Hyaluronic acid) 및 알지네이트(Alginate)로 이루어진 군으로부터 선택되는 하나 이상일 수 있으며, 바람직하게는 콜라겐 및 마트리젤의 혼합물이다. 콜라겐 및 마트리젤을 혼합 사용할 경우 본 발명에 따른 골 조직 모사체의 분화를 향상시키고 성숙도를 높일 수 있다.The gel may be a hydrogel, and the hydrogel is Matrigel, Puramatrix, collagen, fibrin gel, polyethylene glycol diacrylate (PEG-DA), polyethylene glycol diacrylate Mesacrylate (PEG-DMA), PNIPAM, Poloxamer, Chitosan, Agarose, Gelatin, Hyaluronic acid and Alginate It may be one or more selected from the group consisting of, preferably a mixture of collagen and Matrigel. When collagen and Matrigel are mixed and used, the differentiation and maturity of the bone tissue mimetic according to the present invention can be improved.
상기 하이드로젤은 세포외기질(extracellular matrix; ECM)을 더 포함할 수 있다. 상기 ECM은 세포 또는 조직으로부터 추출되거나 또는 생화학적 합성된 것일 수 있다.The hydrogel may further include an extracellular matrix (ECM). The ECM may be extracted from cells or tissues or biochemically synthesized.
상기 세포 혼합물은 상기 세포 혼합물은 조골세포(Osteoblast) 및 파골세포(Osteoclast) 중 어느 하나 이상과 배양액을 포함할 수 있으나 이에 한정되는 것은 아니며, 나아가 혈관 또는 면역 세포 등의 타 조직세포 및 배양액 등을 더 포함할 수 있다.The cell mixture may include, but is not limited to, any one or more of osteoblasts and osteoclasts and a culture medium, and furthermore, other tissue cells such as blood vessels or immune cells and a culture medium. can include more.
상기 골조직 모사체는 제한되는 것은 아니나, 하판(620); 상기 하판의 상측에 위치한 상판(630); 상기 하판 또는 상판 중 어느 하나로부터 유래하여 상판과 하판 사이에 위치하는 돌기구조로 이루어진 배리어부(610); 상기 상판의 일측에 구비된 하나 이상의 세포시료 주입 가이드부(720);를 포함하고, 상기 상판의 크기는 상기 배리어부의 내측만을 폐쇄하는 크기여서, 배리어부의 외측 부분은 개방된 구조인 것을 특징으로 하는 세포배양용기에서 배양되는 것일 수 있다. 본 발명의 일 예에 따른 상기 세포배양용기 구조의 예시를 도 18에 도시하였다. 상기와 같은 세포배양용기를 이용하여 본 발명의 골조직 모사체를 배양한 결과, 골 생성 메커니즘을 보다 잘 구현할 뿐만 아니라, 약물 처리 후 변화에 따른 영상 및 이미지를 확보하여 약물 활성을 평가하거나, 약물 후보 물질을 스크리닝하기 위한 목적으로 활용하는데 있어 현저히 유용한 것을 확인할 수 있었다.The bone tissue mimetic is not limited, but the lower plate 620; an upper plate 630 located above the lower plate; a barrier portion 610 having a protruding structure derived from either the lower plate or the upper plate and positioned between the upper plate and the lower plate; One or more cell sample injection guides 720 provided on one side of the top plate; the size of the top plate is such that only the inside of the barrier part is closed, so that the outer part of the barrier part has an open structure. It may be cultured in a cell culture container. An example of the structure of the cell culture vessel according to an embodiment of the present invention is shown in FIG. 18. As a result of culturing the bone tissue mimetic of the present invention using the cell culture container as described above, not only the bone production mechanism is better implemented, but also the drug activity is evaluated by securing images and images according to the change after drug treatment, or drug candidate It was confirmed that it is remarkably useful in utilizing the material for the purpose of screening.
본 발명에 있어서, "골세포"는 척추동물의 골조직을 이루는 기본 세포로서, 편평한 타원형의 세포이며, "조골세포"는 척추동물의 골기질을 합성하고 분비하는 역할을 하는 세포이고, "파골세포"는 척추동물에서, 뼈의 성장에 수반되어 불필요하게 된 골조직을 파괴ㆍ흡수하는 다핵 세포를 의미한다.In the present invention, "osteocytes" are basic cells constituting the bone tissue of vertebrates, which are flat oval cells, "osteoblasts" are cells that synthesize and secrete bone matrix of vertebrates, and "osteoclasts" In vertebrates, refers to a multinucleated cell that destroys and absorbs bone tissue that becomes unnecessary along with bone growth.
본 발명의 방법에 적용될 수 있다면, 골세포, 조골세포, 파골세포의 종류는 특별히 제한되지 않고, 인간, 마우스 등을 포함하는 척추동물 유래의 세포주 혹은 일차배양세포, 줄기세포 등의 동물세포를 포함할 수 있다.If applicable to the method of the present invention, the types of bone cells, osteoblasts, and osteoclasts are not particularly limited, and include cell lines derived from vertebrates including humans and mice, or animal cells such as primary cultured cells and stem cells. can do.
한편, 상기 골 조직 모사체의 제조방법은 (S1) 골세포(Osteocyte), 세포외기질(extracellular matrix; ECM) 및 하이드로젤을 혼합하는 단계; (S2) 상기 혼합물을 지지체 상에 적가한 후 젤화(gelation)하는 단계; (S3) 젤화된 하이드로젤 외측부 가장자리를 둘러싸도록 조골세포(Osteoblast)를 시딩(seeding)하는 단계; 및 (S4) 배양액을 주입하는 단계;를 포함한다.On the other hand, the manufacturing method of the bone tissue mimetic is (S1) mixing the bone cells (Osteocyte), extracellular matrix (ECM) and hydrogel; (S2) dropwise addition of the mixture onto a support and then gelation; (S3) seeding osteoblasts to surround the outer edge of the gelled hydrogel; and (S4) injecting the culture medium.
상기 (S1) 단계에서는 적어도 하나 이상의 젤을 더 포함할 수 있으며, 일예로, 마트리젤(matrigel), ECM과 같이 조직에서 추출된 하이드로젤 등을 더 포함하여 혼합할 수 있다. 마트리젤은 히알루론산(Hyaluronic acid)과 사이토카인(cytokine)이 풍부하여 골 조직 모사체의 분화를 향상시키고 성숙도를 높일 수 있도록 한다.In the step (S1), at least one gel may be further included, and for example, a hydrogel extracted from tissues such as matrigel and ECM may be further included and mixed. Matrigel is rich in hyaluronic acid and cytokine, so it can improve the differentiation of bone tissue matrix and increase its maturity.
상기 지지체는 유리(glass), 세라믹(ceramic), 실리콘 고무(silicone rubber), 실리콘(silicon), 폴리스티렌(polystyrene), 폴리메틸메타크릴레이트(polymethylmethacrylate), 폴리프로필렌(polypropylene), 폴리카보네이트(polycarbonate), 폴리우레탄(polyurethane), 광경화성 플라스틱, 열가소성 플라스틱 및 금속으로 이루어진 군으로부터 선택되는 어느 하나 이상의 소재로 구성될 수 있다. 또한, 이의 화학적 변형체를 포함할 수 있으며, 3D 프린팅에 이용될 수 있는 소재를 포함할 수 있다.The support may be made of glass, ceramic, silicone rubber, silicon, polystyrene, polymethylmethacrylate, polypropylene, or polycarbonate. , It may be composed of one or more materials selected from the group consisting of polyurethane, photocurable plastics, thermoplastics, and metals. In addition, it may include a chemical variant thereof, and may include a material that can be used for 3D printing.
상기 실리콘 고무로는 이에 제한되는 것은 아니나, 폴리디메틸실록세인(PolyDiMethyl Siloxane; PDMS), 에코플렉스(ecoflex) 등을 예시할 수 있다.Examples of the silicone rubber include, but are not limited to, polydimethylsiloxane (PDMS) and ecoflex.
상기 하이드로젤은 마트리젤(Matrigel), 푸라매트릭스(Puramatrix), 콜라겐(Collagen), 피브린젤(Fibrin gel), 폴리에틸렌글리콜 디아크릴레이트(PEG-DA), 폴리에틸렌글리콜 디메사크릴레이트(PEG-DMA), 폴리나이팜(PNIPAM), 폴록세이머 (Poloxamer), 키토산(Chitosan), 아가로스(Agarose), 젤라틴(Gelatin), 히알루론산(Hyaluronic acid) 및 알지네이트(Alginate)로 이루어진 군으로부터 선택되는 하나 이상일 수 있으며, 바람직하게는 콜라겐 및 마트리젤의 혼합물, 또는 콜라겐, 마트리젤 및 ECM의 혼합물일 수 있으나 이에 한정되는 것은 아니다.The hydrogel is Matrigel, Puramatrix, collagen, fibrin gel, polyethylene glycol diacrylate (PEG-DA), polyethylene glycol dimesacrylate (PEG-DMA) It may be at least one selected from the group consisting of PNIPAM, Poloxamer, Chitosan, Agarose, Gelatin, Hyaluronic acid, and Alginate. And, preferably, it may be a mixture of collagen and Matrigel, or a mixture of collagen, Matrigel, and ECM, but is not limited thereto.
콜라겐 및 마트리젤의 혼합물 또는 콜라겐, 마트리젤 및 ECM의 혼합물을 사용할 경우 본 발명에 따른 골 조직 모사체의 분화를 향상시키고 성숙도를 높일 수 있다. 한편, 상기 ECM은 골 세포 또는 골 조직으로부터 추출되거나 또는 생화학적 합성된 것일 수 있으나 이에 한정되는 것은 아니다.When a mixture of collagen and Matrigel or a mixture of collagen, Matrigel, and ECM is used, the differentiation and maturity of the bone tissue mimetic according to the present invention can be improved. Meanwhile, the ECM may be extracted from bone cells or bone tissue or biochemically synthesized, but is not limited thereto.
상기 (S2) 단계의 젤화는 pH 6 내지 9의 범위에서 이루어지는 것일 수 있다. 젤화가 pH 7 내지 8의 범위에서 이루어지는 경우, 골 조직 모사체 내에서 골세포의 형태와 생존력이 향상될 수 있다.Gelation in the step (S2) may be performed in the range of pH 6 to 9. When gelation is made in the pH range of 7 to 8, the morphology and viability of bone cells in the bone tissue matrix can be improved.
상기 (S3) 단계는 젤화된 하이드로젤 외측부 가장자리를 둘러싸도록 조골세포(Osteoblast)와 함께 파골세포(Osteoclast)를 시딩(seeding)하는 것일 수 있다. 파골세포를 추가적으로 시딩하는 경우, 보다 더 실제의 골 조직과 유사한 특성을 나타내는 골 조직 모사체를 제조할 수 있다.The step (S3) may be seeding osteoclasts together with osteoblasts so as to surround the outer edge of the gelated hydrogel. In the case of additionally seeding osteoclasts, it is possible to prepare a bone tissue mimetic exhibiting properties similar to those of actual bone tissue.
상기 골 질환은 골다공증, 불완전 골형성증(osteogenesis imperfecta), 과골화증, 고칼슘혈증, 부갑상선 기능항진증, 골연화증, 용해성 골질환, 골괴사증, 뼈의 파젯병, 골 발생 질환, 골 골절, 류마티스 관절염에 의한 골손실, 염증성 류마티스 관절염, 골수염, 전이성 골질환, 치주성 골 소실, 구루병, 암에 의한 골 손실 및 골의 노인성 손실로 이루어진 군으로부터 선택되는 어느 하나일 수 있으며, 바람직하게는 골다공증일 수 있으나 이에 한정되는 것은 아니다.The bone disease is caused by osteoporosis, incomplete osteogenesis imperfecta, hyperostosis, hypercalcemia, hyperparathyroidism, osteomalacia, lytic bone disease, osteonecrosis, Paget's disease of bone, bone development disease, bone fracture, and rheumatoid arthritis It may be any one selected from the group consisting of bone loss, inflammatory rheumatoid arthritis, osteomyelitis, metastatic bone disease, periodontal bone loss, rickets, bone loss due to cancer and senile bone loss, preferably osteoporosis. It is not limited.
본 발명의 "예방"이란 본 발명에 따른 조성물의 투여에 의해 질환의 발병을 억제 또는 지연시키는 모든 행위를 의미한다."Prevention" of the present invention means any action that suppresses or delays the onset of a disease by administering the composition according to the present invention.
본 발명에서, "치료"는 생물학적 대상체 (예: 인간 또는 동물)을 목적으로 하는 질병이나 질환의 진행 억제 혹은 중지, 진행 속도 감소, 증상 개선 및 완화, 또는 질병예방 등의 치료 효과를 포함한다.In the present invention, "treatment" includes therapeutic effects such as inhibiting or stopping the progression of a disease or disease, reducing the rate of progression, improving or alleviating symptoms, or preventing a disease for a biological subject (eg, human or animal).
본 발명에서, “골 조직 모사체 영상”은 골조직 모사체를 촬영한 영상을 의미한다. 상기 영상은 마커영역을 포함하며, 상기 마커영역은 골 조직 모사체에 특정 약물 반응이나 기작을 표시하기 위한 마커가 표시된 영역을 나타낸다.In the present invention, “bone tissue matrix image” means an image of a bone tissue matrix photographed. The image includes a marker region, and the marker region represents a region marked with a marker for displaying a specific drug response or mechanism on the bone tissue mimetic.
본 발명에서, “학습 모델”은 기계학습 모델을 말하며, 기계 학습모델은 다양한 유형의 모델을 들 수 있다. 예컨데 기계 학습모델은 결정 트리, RF(random forest), KNN(K-nearest neighbor), 나이브 베이즈(Naive Bayes), SVM(support vector machine), ANN(artificial neural network) 등을 들 수 있으나, 이에 한정되는 것은 아니다.In the present invention, “learning model” refers to a machine learning model, and the machine learning model may include various types of models. For example, machine learning models include decision trees, random forests (RFs), K-nearest neighbors (KNNs), Naive Bayes, support vector machines (SVMs), and artificial neural networks (ANNs). It is not limited.
상기 기계 학습모델은 다양한 신경망 모델과 연관되며 이 중 ANN은 생물의 신경망을 모방한 통계학적 학습 알고리즘이다. 또한, DNN(Deep Neural Network)은 일반적인 인공신경망과 마찬가지로 복잡한 비선형 관계(non-linear relationship)들을 모델링할 수 있다. DNN은 다양한 유형의 모델이 연구되어 왔으며, 예컨대, CNN(Convolutional Neural Network), RNN(Recurrent Neural Network), RBM(Restricted Boltzmann Machine), DBN(Deep Belief Network), GAN(Generative Adversarial Network), RL(Relation Networks) 등을 들 수 있다. 일예로, 이하 설명에서는 CNN 기반의 모델을 기반으로 설명한다.The machine learning model is associated with various neural network models, and among them, ANN is a statistical learning algorithm that imitates a biological neural network. In addition, DNN (Deep Neural Network) can model complex non-linear relationships like general artificial neural networks. Various types of DNN models have been studied, such as CNN (Convolutional Neural Network), RNN (Recurrent Neural Network), RBM (Restricted Boltzmann Machine), DBN (Deep Belief Network), GAN (Generative Adversarial Network), RL ( Relation Networks), etc. As an example, the following description is based on a CNN-based model.
이하, 골조직 모사체의 이미지 분석을 이용한 약물 스크리닝 방법 및 장치에 대해 도면을 참조하여 상세히 설명한다.Hereinafter, a drug screening method and apparatus using image analysis of a bone mimic will be described in detail with reference to the drawings.
0. 골조직 모사체 이미지 분석을 이용한 골질환 약물 스크리닝 방법 및 장치0. Bone disease drug screening method and device using bone tissue mimetic image analysis
골조직 모사체 이미지 분석을 이용한 골질환 약물 스크리닝 방법 및 장치의 전체적인 구조에 대해 설명한다.The overall structure of a bone disease drug screening method and device using bone tissue mimetic image analysis will be described.
도 1은 골조직 모사체 이미지 분석을 이용한 골질환 약물 스크리닝 방법에 관한 일 예시이다.1 is an example of a bone disease drug screening method using bone tissue mimetic image analysis.
도 1의 골조직 모사체(100)는 체내 골조직을 모방한다. 상기 체내 골조직은 사람 또는 동물의 골조직을 포함한다. 이를 통해 사람 또는 동물에 대한 직접적인 실험 없이도 골질환에 대한 약물(110) 효능을 평가할 수 있다. 상기 체내 골조직은 사람 또는 사람을 제외한 동물에서 유래될 수 있다. 골조직 모사체(100)는 빠른 약물 스크리닝에 사용될 수 있다. 골조직 모사체(100)에는 특정 물질 또는 반응을 확인하기 위한 마커(120)가 더 처리될 수 있다.Bone tissue mimetic 100 of Figure 1 mimics internal bone tissue. The body bone tissue includes human or animal bone tissue. Through this, it is possible to evaluate the efficacy of the drug 110 for bone diseases without direct experiments on humans or animals. The in vivo bone tissue may be derived from humans or non-human animals. Bone mimic 100 can be used for rapid drug screening. The bone tissue mimic 100 may be further processed with a marker 120 for confirming a specific substance or reaction.
도 1의 영상장치(130)는 상기 골조직 모사체(100)를 촬영하여 골조직 모사체 영상(200)을 획득한다.The imaging device 130 of FIG. 1 acquires the bone tissue matrix image 200 by photographing the bone tissue matrix 100.
도 1의 골조직 모사체 영상(200)은 약물 처리에 따른 골조직 모사체의 반응을 보기 위한 영상이다. 영상장치(130)가 생성한 골조직 모사체 영상(200)은 별도 데이터베이스(DB)에 저장될 수 있다.The bone tissue matrix image 200 of FIG. 1 is an image for viewing the reaction of the bone tissue matrix according to drug treatment. The bone tissue matrix image 200 generated by the imaging device 130 may be stored in a separate database (DB).
도 1의 분석장치(300)는 영상장치(130)가 생성한 상기 골조직 모사체 영상(200)을 입력 받아, 기 학습모델에 입력한다. 분석장치(300)는 상기 학습모델이 출력하는 값에 따라 약물(110)이 골조직 모사체(100)에 주는 정도에 대한 결과값을 출력한다. 학습모델은 사전에 약물 효과를 예측하도록 학습된 모델이다. 학습모델은 분석 대상인 약물(110)의 효과에 대한 확률값을 출력한다. 분석장치(300)는 학습 모델이 출력하는 확률값을 기준으로 해당 약물의 유효성 여부를 결정할 수 있다.The analysis device 300 of FIG. 1 receives the bone tissue mimetic image 200 generated by the imaging device 130 and inputs it to a pre-learning model. The analysis device 300 outputs a result value for the degree to which the drug 110 gives to the bone tissue mimetic 100 according to the value output by the learning model. The learning model is a model trained to predict drug effects in advance. The learning model outputs a probability value for the effect of the drug 110 to be analyzed. The analysis device 300 may determine whether or not the drug is effective based on the probability value output by the learning model.
1. 골조직 모사체 구조1. Bone tissue matrix structure
골조직 모사체(100)의 구조에 대해 설명한다.The structure of the bone tissue parent body 100 will be described.
도 2는 골조직 모사체(100)를 도시한 예이다.Figure 2 is an example showing a bone tissue mimic (100).
골조직 모사체(100)는 체내의 골조직을 모방한다. 상기 체내의 골조직은 사람 또는 동물의 골조직을 포함한다. 이를 통해 사람 또는 동물에 대한 직접적인 실험 없이도 골질환에 대한 약물의 효능을 평가할 수 있다.Bone tissue mimetic 100 mimics the bone tissue in the body. The bone tissue in the body includes human or animal bone tissue. Through this, the efficacy of drugs for bone diseases can be evaluated without direct experiments on humans or animals.
상기 골조직 모사체(100)는 중심부에 골세포(101)를 포함하는 하이드로젤(104)과 조골세포로부터 유래한 탈세포화된 세포외 기질(102)이 위치하고, 그 주위에 세포 혼합물(103)이 존재할 수 있다.The bone tissue mimetic 100 has a hydrogel 104 containing osteocytes 101 and a decellularized extracellular matrix 102 derived from osteoblasts located at the center, and a cell mixture 103 around it. can exist
상기 골조직 모사체(100)는 하이드로젤(104) 주변부로 세포혼합물이 수평적으로 배치되는 구조를 갖는다. 이는 기존의 수직 방향으로 배치되는 모델 또는 트랜스웰(Transwell)과 같이 초점평면이 다른 모델과 달리 광학적으로 세포-세포간 상호작용을 용이하게 관찰할 수 있는 특징을 갖는다.The bone tissue matrix 100 has a structure in which the cell mixture is horizontally disposed around the hydrogel 104. This has the characteristic of being able to easily observe cell-cell interactions optically, unlike models that are disposed in a vertical direction or models that have different focal planes, such as transwells.
상기 골조직 모사체(100)는 종래의 스캐폴드식 모사체와는 달리 뼈 조직을 잘 모사하는 생체 모방 구조체이면서, 고속 분석에 친화적인 다중 웰(multi-well) 구조의 웰플레이트 형상으로 제작할 수 있어, 고속 약물 효능 평가 분석에 유리한 구조적 특징을 갖는다.Unlike conventional scaffold-type mimics, the bone tissue mimetic 100 is a biomimetic structure that well mimics bone tissue, and can be manufactured in a well plate shape with a multi-well structure that is friendly to high-speed analysis. , with structural features favorable for high-speed drug efficacy evaluation assays.
상기 골조직 모사체(100)는 골다공증과 같은 골 관련 질환에 대한 신규 약물의 개발을 위한 약물 스크리닝, 특히 고속 스크리닝 또는 의약품 평가를 위한 모델로 사용될 수 있으며, 또한 골 질환에 관한 다양한 연구에 사용될 수 있다.The bone tissue mimetic 100 can be used as a model for drug screening, particularly high-speed screening or drug evaluation, for the development of new drugs for bone-related diseases such as osteoporosis, and can also be used for various studies on bone diseases. .
상기 세포혼합물(103)은 조골세포(osteoblast), 파골세포(osteoclast) 및 골원성세포(osteogenic cell) 중 적어도 하나가 포함될 수도 있다. 또한 혈관 또는 면역세포 등의 타 조직세포와 배양액을 포함할 수도 있다.The cell mixture 103 may include at least one of osteoblasts, osteoclasts, and osteogenic cells. It may also contain other tissue cells such as blood vessels or immune cells and culture medium.
2. 골조직 모사체 영상 획득2. Bone tissue matrix image acquisition
골조직 모사체에서 영상을 획득하는 과정에 대해서 설명한다.The process of acquiring images from the bone tissue matrix will be described.
도 3은 골조직 모사체(100)에서 골조직 모사체 영상(200)을 획득하는 과정을 나타낸 것이다.Figure 3 shows the process of obtaining the bone tissue matrix image 200 from the bone tissue matrix 100.
상기 골조직 모사체 영상(200)은 골조직 모사체(100)에 마커(120)를 처리한 후 영상장치(130)를 이용하여 얻을 수 있다.The bone tissue matrix image 200 can be obtained by using the imaging device 130 after processing the marker 120 on the bone tissue matrix 100.
상기 골조직 모사체의 영상(200)은 약물(110)이 처리된 골조직 모사체(100)에 대한 영상일 수도 있다.The image 200 of the bone tissue mimetic may be an image of the bone tissue mimetic 100 treated with the drug 110.
상기 약물(110)은 골질환을 치료하기 위한 약물일 수 있다.The drug 110 may be a drug for treating bone disease.
상기 약물(110)은 여러 개의 약물이 병용투여 된 것일 수도 있다.The drug 110 may be a combination of several drugs.
상기 약물(110)은 상기 골조직 모사체(100)의 중심부의 골세포 또는 외측부의 세포 혼합물 중 적어도 하나에 DNA합성, RNA합성, 유전자 발현, 유전자 구조, 단백질 합성, 단백질 변형, 단백질 분비, 단백질 구조, 당 합성, 당 분비, 당 변형, 지질 분비, 세포막 합성, 세포내 신호전달, 세포간 신호전달, 세포내 소기관, 세포의 분화, 세포의 분열, 세포의 운동 중 적어도 하나에 영향을 미치는 약물일 수도 있다.The drug 110 is applied to DNA synthesis, RNA synthesis, gene expression, gene structure, protein synthesis, protein modification, protein secretion, protein structure in at least one of the bone cells in the center or the cell mixture in the outer part of the bone tissue mimetic 100. A drug that affects at least one of glycosynthesis, glucose secretion, sugar modification, lipid secretion, cell membrane synthesis, intracellular signaling, intercellular signaling, intracellular organelles, cell differentiation, cell division, and cell movement may be
상기 약물(110)은 골질환을 치료하기 위한 천연물, 합성의약품, 복합의약품, 한약, 한약추출물, 한약제제, 생약, 단백질의약품, 유전자 재조합 의약품, 세포 배양 의약품, 효소 의약품, 미생물 의약품, 항체 의약품, 단일클론항체 의약품, 호르몬 의약품, 방사선 의약품, 항체-약물 접합체 의약품(Antibody-drug conjugate), 세포 치료제, 유전자 치료제 중 적어도 하나인 것을 특징으로 하는 약물 일 수도 있다.The drug 110 includes natural products, synthetic drugs, combination drugs, herbal medicines, herbal medicine extracts, herbal medicines, herbal medicines, protein medicines, genetically recombinant medicines, cell culture medicines, enzyme medicines, microbial medicines, antibody medicines, The drug may be at least one of a monoclonal antibody drug, a hormone drug, a radiopharmaceutical, an antibody-drug conjugate, a cell therapy, and a gene therapy.
상기 마커(120)는 골조직 모사체에서 특정물질이나 반응을 확인하기 위하여 이용된다.The marker 120 is used to identify a specific substance or reaction in the bone tissue mimetic.
상기 마커(120)는 영상장치(130)가 측정할 수 있는 여러 신호를 발할 수 있다. 예를 들어 특정색상을 나타내는 염색약이나, 빛을 발생시키는 발광물질이나, 형광이나 인광을 발생시키는 형광물질 또는 인광물질이거나, 방사능을 발생시키는 물질일 수도 있다.The marker 120 may emit various signals that the imaging device 130 can measure. For example, it may be a dye that exhibits a specific color, a luminescent material that generates light, a fluorescent or phosphorescent material that generates fluorescence or phosphorescence, or a material that generates radioactivity.
영상장치(130)는 골조직 모사체(100)를 촬영하여 골조직 모사체 영상을 획득할 수 있다.The imaging device 130 may acquire an image of the bone tissue matrix by photographing the bone tissue matrix 100.
영상장치(130)는 이미지분석에 필요한 영상을 측정할 수 있는 모든 장치를 말한다. 예를 들어 일반현미경, 형광현미경, 공초점현미경, 전자현미경, 주사탐침 현미경, 토모그래피 현미경 중 적어도 하나 일 수 있다.The imaging device 130 refers to any device capable of measuring an image required for image analysis. For example, it may be at least one of a general microscope, a fluorescence microscope, a confocal microscope, an electron microscope, a scanning probe microscope, and a tomography microscope.
영상장치(130)는 촬영한 골조직 모사체의 영상(200)을 영상분석장치에 전송하거나, 별도의 영상DB(230)에 저장할 수도 있다..The imaging device 130 may transmit the captured image 200 of the bone tissue matrix to the image analysis device or store it in a separate image DB 230.
상기 골조직 모사체 영상(200)은 영상장치(130)가 골조직 모사체(100)를 촬영한 영상을 의미한다.The bone tissue matrix image 200 means an image of the bone tissue matrix 100 photographed by the imaging device 130.
상기 골조직 모사체 영상(200)은 마커영역을 포함한다. 마커영역은 골조직 모사체에 물질의 특정위치나 반응을 표시하기 위한 마커가 표시된 영역을 나타낸다.The bone tissue matrix image 200 includes a marker region. The marker region represents a region marked with a marker for displaying a specific location or reaction of a material on the bone tissue mimic.
상기 골조직 모사체 영상(200)은 하나의 특정대상을 식별하는 단일 마커가 포함된 영상 또는 서로 다른 대상들을 식별하는 복수의 마커들이 포함된 영상 일 수도 있다.The bone tissue mimetic image 200 may be an image including a single marker for identifying one specific target or an image including a plurality of markers for identifying different targets.
상기 마커 영역은 이하 설명하는 발명에서 학습모델이 골조직 모사체의 이미지에서 중심적으로 학습 부분이며 이를 중심으로 약물을 평가하는 영역일 수 있다.In the invention described below, the marker region may be a region in which a learning model is centrally learned in an image of a bone tissue mimic and a drug is evaluated based thereon.
상기 마커 영역은 처리한 마커(120)의 종류에 따라 다르게 표시될 수 있다. 예를 들어 염색약을 처리하여 색상을 띠거나 발광물질을 처리하여 빛을 발하거나, 형광/인광물질을 처리하여 형광 또는 인광을 발하거나, 방사능원소를 처리하여 방사능이 나타나는 영역일 수도 있다.The marker area may be displayed differently according to the type of the processed marker 120 . For example, it may be a region that emits color by treating dye, emits light by treating a luminescent material, emits fluorescence or phosphorescence by treating a fluorescent/phosphorescent material, or shows radioactivity by treating a radioactive element.
상기 마커영역은 마커(120)의 종류에 따라서 골조직 모사체의 형상이나 모양 위치 움직임 등을 나타낸 것일 수도 있다.The marker area may indicate the shape, shape, position, or movement of the bone tissue mimetic depending on the type of the marker 120 .
상기 마커영역은 마커(120)의 종류에 따라서 골조직 모사체 내에서 세포, 상기 골조직 모사체의 세포, 핵, 소포체, 리보솜, 리소좀, 골지체, 중심체, 세포막, 미토콘드리아, 미세소관, 미세섬유, 세포질, DNA, RNA, 핵산, 히스톤, 단백질, 당 단백질, 막 단백질, 탄수화물, 막 탄수화물, 지질, 콜레스테롤, 당 지질, 당, 및 콜라겐 등을 표지한 것일 수도 있다.Depending on the type of the marker 120, the marker region may be a cell, a cell, a nucleus, an endoplasmic reticulum, a ribosome, a lysosome, a Golgi apparatus, a centrosome, a cell membrane, mitochondria, microtubules, microfilaments, cytoplasm, DNA, RNA, nucleic acids, histones, proteins, glycoproteins, membrane proteins, carbohydrates, membrane carbohydrates, lipids, cholesterol, glycolipids, sugars, collagen, and the like may be labeled.
3. 딥러닝 모델 구현3. Deep learning model implementation
골조직 모사체의 이미지를 이용하여 골질환 약물을 스크리닝 하는 딥러닝 모델을 구현하는 과정에 대해서 설명한다.The process of implementing a deep learning model for screening drugs for bone diseases using images of bone tissue mimics will be described.
도4는 골조직 모사체의 영상을 이용하여 골질환 약물의 효능을 평가하는 하는 딥러닝 모델을 구현하는 과정을 도시한 것이다.Figure 4 shows a process of implementing a deep learning model for evaluating the efficacy of a bone disease drug using an image of a bone mimic.
딥러닝 모델을 위한 학습모델 구축 과정은 학습데이터 구축 과정(310)과 학습데이터를 이용한 모델 학습 과정(330)으로 구분된다. 학습데이터 구축 과정(310)과 모델 학습 과정(330)은 별도의 장치에서 수행될 수 있다.The process of building a learning model for a deep learning model is divided into a process of building learning data (310) and a process of learning a model using learning data (330). The learning data construction process 310 and the model learning process 330 may be performed in separate devices.
설명의 편의를 위하여 학습장치가 학습모델 구축 과정을 수행한다고 설명한다.For convenience of explanation, it is explained that the learning device performs a process of building a learning model.
상기 학습장치는 학습모델을 구현하기 위한 학습데이터로 골조직 모사체 영상(200)을 이용한다.The learning device uses the bone tissue mimetic image 200 as learning data for implementing a learning model.
상기 학습장치는 상기 학습데이터로 이용될 골조직 모사체의 영상은 영상DB(230)에서 받거나, 영상장치(130)에서 전송받을 수 있다.The learning device may receive an image of the bone tissue mimetic to be used as the learning data from the image DB 230 or transmitted from the imaging device 130 .
상기 학습데이터로 이용될 골조직 모사체의 영상은 약물이 처리된 골조직 모사체의 영상(양성 데이터) 및 약물이 처리되지 않은 골조직 모사체의 영상(음성 데이터)을 포함할 수 있다. 또는 학습 데이터는 유효한 약물이 처리된 영상(양성 데이터) 및 유효하지 않는 약물이 처리된 영상(음성 데이터)를 포함할 수 있다. 학습 데이터는 해당 영상에 대한 라벨값(양성 또는 음성) 정보를 포함할 수 있다.The image of the bone tissue mimic to be used as the learning data may include an image of the drug-treated bone tissue mimic (positive data) and an image of the drug-untreated bone tissue mimic (negative data). Alternatively, the learning data may include an image treated with an effective drug (positive data) and an image treated with an ineffective drug (negative data). The learning data may include label value (positive or negative) information for the corresponding image.
상기 학습데이터로 이용될 골조직 모사체의 영상은 하나 특정 대상을 식별하는 단일 마커가 포함된 영상 또는 서로 다른 대상들을 식별하는 복수의 마커들이 포함된 영상일 수도 있다. 여기서 서로 다른 대상들은 골조직 모사체 내에서 세포, 상기 골조직 모사체의 세포, 핵, 소포체, 리보솜, 리소좀, 골지체, 중심체, 세포막, 미토콘드리아, 미세소관, 미세섬유, 세포질, DNA, RNA, 핵산, 히스톤, 단백질, 당 단백질, 막 단백질, 탄수화물, 막 탄수화물, 지질, 콜레스테롤, 당 지질, 당, 콜라겐 등을 표지한 것일 수도 있다.The image of the bone tissue matrix to be used as the learning data may be an image including a single marker for identifying a specific target or an image including a plurality of markers for identifying different targets. Here, the different objects are cells in the bone tissue matrix, cells of the bone tissue matrix, nucleus, endoplasmic reticulum, ribosomes, lysosomes, Golgi apparatus, centrosome, cell membrane, mitochondria, microtubules, microfilaments, cytoplasm, DNA, RNA, nucleic acids, histones , proteins, glycoproteins, membrane proteins, carbohydrates, membrane carbohydrates, lipids, cholesterol, glycolipids, sugars, collagen, and the like may be labeled.
상기 학습데이터 구축 과정(310)에서, 학습데이터로 이용될 골조직 모사체 영상은 바로 학습데이터로 이용될 수도 있지만 데이터를 증강한 후 학습데이터로 이용될 수도 있다.In the learning data construction process 310, the bone tissue matrix image to be used as learning data may be directly used as learning data, but may also be used as learning data after augmenting the data.
상기 학습데이터 구축 과정(310)에서 데이터의 증강은 모사체의 영상을 n개로(n>2) 분할하거나 각 이미지를 무작위 하게 확대 또는 축소하거나 수평 또는 수직 뒤집기하는 방법이 이용될 수 있다. 예컨데 학습장치는 영상을 좌우 2개로 분리한 뒤 그 중 하나에 대해서만 수평 뒤집기하여 골조직 모사체 영상을 마련할 수 있다. 이와 같이 학습장치는 m개의 골조직 모사체의 영상들을 처리하여 n x m개의 골조직 모사체 영상을 생성할 수 있다.In the learning data construction process 310, data augmentation may be performed by dividing the parent image into n pieces (n>2), randomly enlarging or reducing each image, or horizontally or vertically flipping each image. For example, the learning device may divide the image into two left and right parts and horizontally flip only one of them to prepare a bone tissue matrix image. In this way, the learning device may generate n x m images of the bone tissue matrix by processing m images of the bone tissue matrix.
상기 학습데이터로 이용될 골조직 모사체 영상은 학습DB(320)에 저장될 수도 있다.Bone tissue mimetic image to be used as the learning data may be stored in the learning DB (320).
상기 학습장치는 학습DB(320)에 저장된 골조직 모사체 영상을 이용하여 학습데이터를 이용한 모델 학습 과정(330)을 수행한다.The learning device performs a model learning process 330 using learning data using the bone tissue mimetic image stored in the learning DB 320.
상기 학습장치는 학습DB(320)에서 적어도 하나이상의 골조직 모사체 영상을 획득한다. 학습장치는 해당 영상을 학습모델에 입력한다. 학습모델은 입력된 영상에서 특징을 추출하여 약물처리에 따른 변화가 있는지 확률값을 산출한다. 학습장치는 학습모델이 출력하는 확률값과 사전에 알고 있는 해당 영상의 라벨값을 비교하여 학습모델의 파라미터를 갱신한다. 학습장치는 다수의 학습데이터를 이용하여 학습모델을 학습하는 과정을 반복한다(330).The learning device acquires at least one or more bone tissue matrix images from the learning DB (320). The learning device inputs the corresponding image to the learning model. The learning model extracts features from the input image and calculates a probability value whether there is a change due to drug treatment. The learning device updates parameters of the learning model by comparing a probability value output from the learning model with a previously known label value of the corresponding image. The learning device repeats the process of learning the learning model using a plurality of learning data (330).
한편, 학습 모델은 마커 영역의 특징을 주된 특징으로 분석하여 확률값을 산출할 수 있다. 따라서, 학습 모델은 마커 영역을 사전에 분할하여 해당 영역을 중심으로 영상을 분석하는 모델일 수 있다. 또는 학습 모델은 마커 영역에 어텐션(attention)을 부여하는 모델일 수도 있다.Meanwhile, the learning model may calculate a probability value by analyzing the characteristics of the marker region as main features. Accordingly, the learning model may be a model in which a marker region is previously segmented and an image is analyzed centering on the corresponding region. Alternatively, the learning model may be a model that gives attention to the marker region.
상기 학습장치가 학습한 모델로부터 출력한 확률값을 기준으로 분석장치가 해당 약물의 유효성 여부를 결정할 수 있다.Based on the probability value output from the model learned by the learning device, the analysis device may determine whether or not the drug is effective.
이하, 본 발명에 따라 연구자가 학습 모델을 구축한 과정 및 검증 결과에 대해 설명한다.Hereinafter, the process and verification results of the researcher building the learning model according to the present invention will be described.
도 5는 연구자가 학습데이터 및 검증데이터로 이용한 골조직 모사체 영상을 도시한 예이다.5 is an example showing an image of a bone tissue mimetic used by a researcher as learning data and verification data.
연구자는 하나의 골조직 모사체에는 약물을 처리했고(+drug), 나머지 골조직 모사체에는 약물을 처리하지 않았다(NC).The researcher treated one bone mimic with a drug (+drug), and did not treat the other mimic with a drug (NC).
연구자는 각각의 골조직 모사체에 β-카테닌(β-catenin)과 세포 핵을 탐지할 수 있는 형광 마커를 처리하였다.The researcher treated each bone mimic with a fluorescent marker capable of detecting β-catenin and cell nuclei.
연구자는 영상획득 장치를 통하여 각 골조직 모사체에 β-카테닌과 핵을 표시한 영상을 각각 획득하였다. 이를 학습데이터 및 검증데이터로 만든 모델을 BN 모델이라 한다.The researcher obtained images displaying β-catenin and nuclei in each bone tissue mimetic through an image acquisition device. A model created with learning data and verification data is called a BN model.
연구자는 영상획득 장치를 통하여 각 골조직 모사체에 β-카테닌과 핵을 표시한 영상 및 이미지를 병합(merged)한 형광 이미지로 이미지 세트를 만들었다. 이를 학습데이터 및 검증데이터로 하여 만든 모델을 BNM모델이라 한다.The researcher made an image set using an image acquisition device with images in which β-catenin and nuclei were displayed in each bone tissue mimic and fluorescent images merged with the images. A model created using this as learning data and verification data is called a BNM model.
연구자는 약물이 처리된 골조직 모사체 영상 420 세트 및 약물이 처리되지 아니한 골조직 모사체 영상 424 세트를 증강 과정을 거친 후 학습모델 구현에 이용하였다.The researcher used 420 sets of drug-treated and 424 sets of non-drug-treated bone tissue mimetic images to implement a learning model after going through an augmentation process.
연구자는 증강과정으로 상기 골조직 모사체 영상을 4개로 분할하고, 무작위하게 좌우 반전하는 방법을 이용하였다.The researcher used a method of dividing the bone tissue mimetic image into four parts as an augmentation process and randomly inverting left and right.
연구자는 상기 이미지 중 90%을 딥러닝 모델 구현을 위한 학습데이터로 이용하였으며 나머지 10%는 검증데이터로 이용하였다.The researcher used 90% of the images as training data for deep learning model implementation and used the remaining 10% as verification data.
연구자는 딥러닝의 평가를 위해서 10번의 교차검증(cross validation)을 수행하였다.The researcher performed 10 cross validations to evaluate deep learning.
도 6은 구현된 딥러닝 모델의 구조를 도시한 예이다.6 is an example showing the structure of an implemented deep learning model.
상기 딥러닝 모델은 3개의 컨볼루션 층(convolution layer), 3개의 풀링 층(pooling layer), 1개의 드랍 층(drop layer), 1개의 플래튼 층(flatten layer) 및 2개의 완전한 연결 층(fully-connected layer)로 구성되어 있다.The deep learning model consists of 3 convolution layers, 3 pooling layers, 1 drop layer, 1 flatten layer and 2 fully connected layers. -connected layer).
상기 딥러닝 모델에서 컨볼루션 층은 이미지에서 특징을 추출하고 이로부터 특성 맵을 만들게 된다.In the deep learning model, the convolution layer extracts features from an image and creates a feature map therefrom.
상기 딥러닝 모델에서 풀링 층은 컨볼루션 층에서 만든 특성 맵의 크기를 줄이거나, 특정 데이터를 강조하기 위해 특성 맵의 값 중 가장 큰 값 또는 평균 값을 추출한다.In the deep learning model, the pooling layer extracts the largest value or average value of the feature map values in order to reduce the size of the feature map created by the convolution layer or to emphasize specific data.
상기 딥러닝 모델에서 드랍 층은 딥러닝 모델에서 과적합(overfitting)을 방지하기 위해 훈련 과정 중에 신경망 모델(neural network model)의 일부분만을 이용한다.In the deep learning model, the drop layer uses only a part of the neural network model during training to prevent overfitting in the deep learning model.
상기 딥러닝 모델에서 플래튼 층은 추출한 데이터의 특징을 하나의 차원으로 만들게 된다.In the deep learning model, the platen layer makes the features of the extracted data into one dimension.
상기 딥러닝 모델은 입력된 골조직 모사체 영상을 분석하여 확률 값을 출력하며, 분석장치가 상기 확률 값을 기준으로 해당 약물의 유효성 여부를 결정할 수 있다.The deep learning model analyzes the input bone tissue mimetic image and outputs a probability value, and the analysis device can determine whether or not the drug is effective based on the probability value.
도 7은 BN 데이터를 이용하여 딥러닝 모델을 구현한 BN 모델에서 훈련 횟수에 따른 정확도와 손실도에 관한 결과를 나타낸다. 훈련 횟수가 60회에 가까워질수록 정확도는 1에 가까워지고, 손실도는 0에 가까워진다. 이를 이하 표 1에 도시하였다. BN 모델의 정확도는 97.2%인 것으로 확인되었다.7 shows the results of accuracy and loss according to the number of trainings in the BN model implementing the deep learning model using BN data. As the number of training approaches 60, the accuracy approaches 1 and the loss approaches 0. This is shown in Table 1 below. The accuracy of the BN model was found to be 97.2%.
도 8은 BNM 데이터를 이용하여 딥러닝 모델을 구현한 BNM 모델에서 훈련 횟수에 따른 정확도와 손실도에 관한 결과를 나타낸다. 훈련횟수가 60회에 가까워질수록 정확도는 1에 가까워지고, 손실도는 0에 가까워진다. 이를 이하 표 1에 도시하였다. BNM 모델의 정확도는 99.5%인 것으로 확인되었다.8 shows the results of accuracy and loss according to the number of times of training in the BNM model implementing the deep learning model using BNM data. As the number of training approaches 60, the accuracy approaches 1 and the loss approaches 0. This is shown in Table 1 below. The accuracy of the BNM model was found to be 99.5%.
BN 모델 및 BNM 모델의 정확도 및 손실도 평가Evaluate accuracy and loss of BN model and BNM model
검증(Validation)Validation 학습테스트(Test)Learning test (Test)
손실도(%)Degree of loss (%) 정확도(%)accuracy(%) 손실도(%)Degree of loss (%) 정확도(%)accuracy(%)
BN 모델
(β-카테닌+핵)BN model
(β-catenin + nucleus) 		0.1010.101 0.9750.975 0.0940.094 0.9720.972
BNM 모델(β-카테닌+핵
이미지 병합)BNM model (β-catenin + nucleus
merge images) 		0.0500.050 0.9870.987 0.0190.019 0.9950.995
도 9는 상기 BN 모델 및 BNM 모델로부터 이미지 분류를 수행한 후, 진단 평가한 결과에 대한 ROC 곡선(receiver operating characteristics curve)을 나타낸 예시이다.9 is an example of ROC curves (receiver operating characteristics curve) for diagnostic evaluation results after performing image classification from the BN model and the BNM model.
가로축(X축)에는 FPR(1-선택도(selectivity))을, 세로축(Y축)에는 민감도(sensitivity)를 도시하였다.FPR (1-selectivity) is shown on the horizontal axis (X-axis), and sensitivity is shown on the vertical axis (Y-axis).
그 결과, BN 모델 및 BNM 모델 모두 AUC(area under curve)가 1에 가까운 값을 나타내는 것을 확인하였다.As a result, it was confirmed that both the BN model and the BNM model showed an area under curve (AUC) value close to 1.
이와 같은 결과는 본 발명에 따른 학습 모델이 골질환 관련 약물 테스트에 정확도가 높으며 효율적이라는 것을 나타낸다.These results indicate that the learning model according to the present invention is highly accurate and efficient for bone disease-related drug testing.
도 10은 골조직 모사체 영상을 분석하여 골질환 약물의 효능을 판별하는 분석장치(500)의 예시이다. 분석장치(500)는 전술한 분석장치(도 1의 (300))에 해당한다. 분석장치(500)는 물리적으로 다양한 형태로 구현될 수 있다. 예컨대, 분석장치(500)는 PC와 같은 컴퓨터 장치, 네트워크 서버, 데이터 처리 전용 칩셋 등의 형태를 가질 수 있다.10 is an example of an analysis device 500 for determining the efficacy of a drug for bone disease by analyzing an image of a bone tissue mimetic. The analysis device 500 corresponds to the above-described analysis device (300 in FIG. 1). The analysis device 500 may be physically implemented in various forms. For example, the analysis device 500 may have a form of a computer device such as a PC, a network server, and a chipset dedicated to data processing.
분석장치(500)는 저장장치(510), 메모리(520), 연산장치(530), 인터페이스 장치(540), 통신장치(550) 및 출력장치(560)를 포함할 수 있다.The analysis device 500 may include a storage device 510, a memory 520, an arithmetic device 530, an interface device 540, a communication device 550, and an output device 560.
저장장치(510)는 골조직 모사체 영상을 저장할 수 있다.The storage device 510 may store an image of a bone tissue mimetic.
저장장치(510)는 골조직 모사체 영상을 분석하여 약물의 효능을 판별하는 코드 또는 프로그램을 저장할 수 있다.The storage device 510 may store a code or program for determining the efficacy of a drug by analyzing an image of a bone tissue mimetic.
저장장치(510)는 골조직 모사체 영상을 이용하여 학습된 학습모델을 저장할 수 있다.The storage device 510 may store the learning model learned using the bone tissue mimetic image.
메모리(520)는 상기 분석장치(500)가 골조직 모사체 영상을 분석하는 과정 중에 생성되는 데이터 및 정보 등을 저장할 수 있다.The memory 520 may store data and information generated during the process of the analysis device 500 analyzing the bone tissue mimetic image.
연산장치(530)는 입력된 골조직 모사체의 영상을 이진화하고, 이진화된 영상에 대한 침식 및 팽창 처리를 하여 불필요한 영역을 제거할 수 있다. 연산장치(530)는 영상의 왼쪽 끝점과 오른쪽 끝점의 중점을 기준으로 입력된 골조직 모사체 영상을 왼쪽 및 오른쪽으로 분할할 수도 있다.The arithmetic unit 530 may remove unnecessary regions by binarizing the input image of the bone tissue mimetic and performing erosion and expansion processing on the binarized image. The arithmetic unit 530 may divide the input bone tissue matrix image into left and right parts based on the midpoint of the left end point and the right end point of the image.
또한, 연산장치(530)는 골조직 모사체 영상을 좌우 반전하여 처리할 수 있다.In addition, the arithmetic device 530 may process the image of the bone tissue model by inverting it left and right.
연산장치(530)는 데이터를 처리하고, 일정한 연산을 처리하는 프로세서, AP, 프로그램이 임베디드된 칩과 같은 장치일 수 있다.The arithmetic device 530 may be a device such as a processor, an AP, or a chip in which a program is embedded that processes data and performs certain arithmetic operations.
인터페이스 장치(540)는 외부로부터 일정한 명령 및 데이터를 입력받는 장치이다.The interface device 540 is a device that receives certain commands and data from the outside.
인터페이스 장치(540)는 물리적으로 연결된 입력장치 또는 외부 저장장치로부터 골조직 모사체 영상을 입력받을 수 있다.The interface device 540 may receive a bone tissue mimetic image from a physically connected input device or an external storage device.
인터페이스 장치(540)는 골조직 모사체 영상을 분석하여 약물의 효능 평가한 결과를 외부 객체에 전달할 수도 있다.The interface device 540 may transmit the result of evaluating the efficacy of the drug to an external object by analyzing the image of the bone tissue mimetic.
통신장치(550)는 유선 또는 무선 네트워크를 통해 일정한 정보를 수신하거나 전송하는 구성을 의미한다.The communication device 550 refers to a component that receives or transmits certain information through a wired or wireless network.
통신장치(550)는 외부 객체로부터 골조직 모사체 영상을 수신할 수 있다. 또한 통신장치(550)는 골조직 모사체 영상을 분석하여 약물의 효능을 평가한 결과를 사용자 단말과 같은 외부 객체에 송신할 수도 있다.The communication device 550 may receive a bone tissue matrix image from an external object. In addition, the communication device 550 may analyze the bone tissue mimetic image and transmit the result of evaluating the efficacy of the drug to an external object such as a user terminal.
인터페이스 장치(540) 및 통신장치(550)는 사용자 또는 다른 물리적 객체로부터 일정한 데이터를 주고받는 구성이므로, 포괄적으로 입출력장치라고도 명명할 수 있다. 정보 내지 데이터 입력 기능에 한정하면 인터페이스 장치(540) 및 통신장치(550)는 입력장치라고 할 수도 있다.Since the interface device 540 and the communication device 550 are components for exchanging certain data from a user or other physical object, they can also be collectively referred to as input/output devices. When limited to information or data input functions, the interface device 540 and the communication device 550 may also be referred to as input devices.
출력장치(560)는 일정한 정보를 출력하는 장치이다. 출력장치(560)는 데이터 처리 과정에 필요한 인터페이스, 입력된 골조직 모사체 영상, 분석 결과 등을 출력할 수 있다.The output device 560 is a device that outputs certain information. The output device 560 may output an interface required for data processing, an input bone tissue matrix image, analysis results, and the like.
또한, 상술한 바와 같은 학습 데이터 구축 방법 및 증강된 학습 데이터를 이용한 학습모델의 학습 방법은 컴퓨터에서 실행될 수 있는 실행가능한 알고리즘을 포함하는 프로그램(또는 어플리케이션)으로 구현될 수 있다. 상기 프로그램은 일시적 또는 비일시적 판독 가능 매체(non-transitory computer readable medium)에 저장되어 제공될 수 있다. 비일시적 판독 가능 매체란 레지스터, 캐쉬, 메모리 등과 같이 짧은 순간 동안 데이터를 저장하는 매체가 아니라, 반영구적으로 데이터를 저장하며, 기기에 의해 판독(reading)이 가능한 매체를 의미한다. 구체적으로는 상술한 다양한 어플리케이션 또는 프로그램들은 CD, DVD, 하드 디스크, 블루레이 디스크, USB, 메모리 카드, ROM(read-only memory), PROM(programmable read-only memory), EPROM(erasable PROM) 또는 EEPROM(electrically EPROM) 또는 플래쉬 메모리 등과 같은 비일시적 판독 가능 매체에 저장되어 제공될 수 있다. 일시적 판독 가능 매체는 스태틱 램(static RAM, SRAM), 다이나믹 램(dynamic RAM, DRAM), 싱크로너스 디램(synchronous DRAM, SDRAM), 2배속 SDRAM(double data rate SDRAM, DDR SDRAM), 증강형 SDRAM(enhanced SDRAM, ESDRAM), 동기화 SDRAM(sync link DRAM, SLDRAM) 및 직접 램버스 램(direct rambus RAM, DRRAM)과 같은 다양한 RAM을 의미한다.In addition, the above-described method for constructing learning data and learning a learning model using augmented learning data may be implemented as a program (or application) including an executable algorithm that can be executed on a computer. The program may be stored and provided in a temporary or non-transitory computer readable medium. A non-transitory readable medium is not a medium that stores data for a short moment, such as a register, cache, or memory, but a medium that stores data semi-permanently and can be read by a device. Specifically, the various applications or programs described above are CD, DVD, hard disk, Blu-ray disk, USB, memory card, ROM (read-only memory), PROM (programmable read-only memory), EPROM (erasable PROM) or EEPROM It may be stored and provided in a non-transitory readable medium such as (electrically EPROM) or flash memory. Temporary readable media include static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), and enhanced SDRAM (enhanced SDRAM). SDRAM, ESDRAM), sync link DRAM (SLDRAM), and direct rambus RAM (DRRAM).
이하, 본 발명의 이해를 돕기 위하여 골조직 모사체 이미지 분석을 위해 필요한 골조직 모사체의 구조 및 이미지 획득 방법과 딥러닝 모델을 이용한 골질환 약의 스크리닝 방법에 대한 바람직한 실시예를 제시한다.Hereinafter, in order to facilitate the understanding of the present invention, a preferred embodiment of a method for obtaining a structure and image of a bone tissue matrix necessary for image analysis of a bone tissue matrix and a method for screening a drug for bone disease using a deep learning model will be presented.
그러나 하기의 실시예는 본 발명을 보다 쉽게 이해하기 위하여 제공되는 것일 뿐, 권리범위를 특정한 실시 형태에 대해 한정하려는 것은 아니며, 이하 설명하는 기술적 사상 및 기술 범위에 포함되는 모든 변경, 균등물 내지 대체물을 포함하는 것으로 이해되어야 한다.However, the following examples are provided for easier understanding of the present invention, and are not intended to limit the scope of rights to specific embodiments, and all changes, equivalents or substitutes included in the technical spirit and technical scope described below It should be understood to include.
[실시예][Example]
[실시예 1] 골조직 모사체의 제조[Example 1] Preparation of bone tissue matrix
1-1. 구조물의 제작1-1. fabrication of structures
도 11은 골조직 모사체를 위한 구조물 제작 공정의 예시를 나타낸 것이다.Figure 11 shows an example of a structure manufacturing process for bone tissue mimetic.
골조직 모사체(일명 뼈칩, Bone-on-a-chip)는 웰플레이트에 기반한 하이드로젤 통합 유닛을 이용하여 제작하였다.A bone tissue replica (aka bone-on-a-chip) was fabricated using a hydrogel integrated unit based on a well plate.
포토공정(photolithography) 방식으로 두께가 2mm가 되도록 제작된 형틀 위에 탄성중합체 PDMS(polydimethyl siloxane)을 부어 주입한 후 중합시켰다.The elastomer PDMS (polydimethyl siloxane) was poured onto a mold made to have a thickness of 2 mm by photolithography, injected, and then polymerized.
중합 후, PDMS 복제물(replica)을 떼어 내고, 젤을 주입하기 위한 직경 1 mm의 주입구를 형성하였다. 이후 맞춤형 펀치를 이용하여 PDMS 복제물이 덤벨 모양이 되도록 갈라내었다.After polymerization, the PDMS replica was removed, and an injection port having a diameter of 1 mm was formed for injecting the gel. The PDMS replica was then split into a dumbbell shape using a custom punch.
PDMS 복제물을 PEEK(polyether ether ketone)으로 만든 맞춤형 지그(jig)를 이용하여 웰플레이트 중앙에 배치시켰다. 이어서, PDMS 복제물을 지그 위에 고정시킨 후 플라즈마처리하여 접합시켰다.A PDMS replica was placed in the center of the well plate using a custom jig made of polyether ether ketone (PEEK). Then, the PDMS replica was fixed on a jig and bonded by plasma treatment.
접합 후, 칩 표면에 하이드로젤의 접착력을 향상시키기 위하여, 칩 내부 표면을 1 mg/mL의 도파민 염산염(dopamine hydrochloride)이 들어 있는 Tris 버퍼 용액으로 코팅하고, 증류수로 5회 세척하였다.After bonding, in order to improve the adhesion of the hydrogel to the chip surface, the inner surface of the chip was coated with a Tris buffer solution containing 1 mg/mL dopamine hydrochloride and washed 5 times with distilled water.
제작된 장치는 80℃의 오븐에서 건조한 후 사용하였다.The fabricated device was used after drying in an oven at 80 °C.
1-2. 골조직 모사체를 위한 세포 배양1-2. Cell culture for bone tissue mimetics
마우스 골세포(IDG-SW3; Kerafast)는 콜라겐으로 코팅한 플라스크에서 열처리하여 불활성화시킨 10%(v/v) FBS(fetal bovine serum; Gibco) 및 50 U/mL 재조합 마우스 인터페론 감마(IFN-γ) 단백질(Gibco)에 1%(w/v) 페니실린/스트렙토마이신(Thermo Fisher Scientific)이 첨가된 알파 최소 필수 배지(alpha Minimum Essential Medium, Alpha-MEM; Gibco)에서 33℃, 5%(v/v) CO2 인큐베이터 환경에서 배양하였다.Mouse bone cells (IDG-SW3; Kerafast) were inactivated by heat treatment in a collagen-coated flask in 10% (v/v) FBS (fetal bovine serum; Gibco) and 50 U/mL recombinant mouse interferon gamma (IFN-γ). ) Protein (Gibco) plus 1% (w/v) penicillin/streptomycin (Thermo Fisher Scientific) in Alpha Minimum Essential Medium (Alpha-MEM; Gibco) at 33°C at 5% (v/v). v) Cultivated in a CO 2 incubator environment.
마우스 조골세포(MC3T3-E1; ATCC)는 10%(v/v) FBS(Gibco)에 1%(w/v) 페니실린/스트렙토마이신(Thermo Fisher Scientific)이 첨가된 알파 최소 필수 배지(Alpha-MEM; Gibco)에서 37℃, 5%(v/v) CO2 인큐베이터 환경에서 배양하였다.Mouse osteoblasts (MC3T3-E1; ATCC) were cultured in Alpha minimum essential medium (Alpha-MEM) supplemented with 1% (w/v) penicillin/streptomycin (Thermo Fisher Scientific) in 10% (v/v) FBS (Gibco). ; Gibco) at 37°C, 5% (v/v) CO 2 and cultured in an incubator environment.
세포는 80% 컨플루언시(confluency)에 도달하였을 때, 0.25%(w/v) 트립신-EDTA(Gibco)를 이용하여 분리하였으며, 전체 배지는 2~3일마다 교체하였다.When the cells reached 80% confluency, they were detached using 0.25% (w/v) trypsin-EDTA (Gibco), and the entire medium was replaced every 2-3 days.
마우스 골세포(IDG-SW3)의 골형성 분화(osteogenic differentiation)를 위해, 인터페론 감마가 제거되고 10%(v/v) FBS(Gibco), 50 ug/mL 아스코르브산, 4 mM 베타-글리세로포스페이트(β-glycerophosphate) 및 1%(w/v) 페니실린/스트렙토마이신(Thermo Fisher Scientific)이 첨가된 분화 배지(differentiation medium; alpha-MEM)에서 37℃ 환경에서 배양하였다.For osteogenic differentiation of mouse osteocytes (IDG-SW3), interferon gamma was removed and 10% (v/v) FBS (Gibco), 50 ug/mL ascorbic acid, 4 mM beta-glycerophosphate (β-glycerophosphate) and 1% (w/v) penicillin/streptomycin (Thermo Fisher Scientific) were added to the differentiation medium (alpha-MEM) at 37°C.
마우스 조골세포(MC3T3-E1)의 분화를 위한 배양은 마우스 골세포(IDG-SW3)과 같은 조건에서 수행되었다.Culture for the differentiation of mouse osteoblasts (MC3T3-E1) was performed under the same conditions as mouse osteoblasts (IDG-SW3).
1-3. 조골세포 유래 탈세포화된 세포외기질(osteoblast-derived decellularized extracellular matrix; OB-dECM) 추출 및 하이드로젤 제조1-3. Osteoblast-derived decellularized extracellular matrix (OB-dECM) extraction and hydrogel preparation
도 12는 조골세포 유래의 OB-dECM 추출 과정을 도시한 일 예이다.12 is an example showing the process of extracting OB-dECM derived from osteoblasts.
먼저 마우스 골세포(MC3T3-E1)을 2주 간 배양한 후, 세포막(cell sheet)를 걷어내었다.First, mouse bone cells (MC3T3-E1) were cultured for 2 weeks, and then the cell sheet was removed.
걷어낸 세포막에 트립신-EDTA(trypsin-EDTA)을 첨가하여 3분 간 처리 후, 0.1%의 트리톤 X-100(Triton X-100)으로 30분 동안 처리하여 분리된 세포를 제거하였다.After treatment for 3 minutes by adding trypsin-EDTA to the skimmed cell membrane, the detached cells were removed by treatment with 0.1% Triton X-100 for 30 minutes.
세포를 제거한 잔여물에 DNase I을 30분 동안 처리하여 DNA 및 불순물 등을 완전히 제거하였다. 최종적으로 수득한 ECM을 동결건조 시킨 후 -20℃에서 보관하였다. 각 단계마다 인산완충식염수(phosphate buffer saline, PBS)를 이용하여 3회 내지 5회의 세척하였다.The residue from which cells were removed was treated with DNase I for 30 minutes to completely remove DNA and impurities. Finally, the obtained ECM was lyophilized and stored at -20 °C. Each step was washed 3 to 5 times using phosphate buffered saline (PBS).
동결건조된 OB-dECM은 잘게 잘라 0.1 M 아세트산에 1 mg/mL 펩신(Sigma-Aldrich)과 혼합한 후, 마그네틱바를 이용하여 4℃에서 500 rpm으로 12시간 동안 교반하였다.The lyophilized OB-dECM was finely cut and mixed with 1 mg/mL pepsin (Sigma-Aldrich) in 0.1 M acetic acid, and then stirred for 12 hours at 500 rpm at 4°C using a magnetic bar.
OB-dECM 용액에 함유된 펩신을 50 ug/mL 펩스타틴(pepstatin; Sigma-Aldrich)을 넣어 불활성화시킨 후, 용액은 이후 사용할 때까지 4℃에 보관하였다.After inactivating pepsin contained in the OB-dECM solution by adding 50 ug/mL pepstatin (Sigma-Aldrich), the solution was stored at 4°C until further use.
이어서, 3~4 mg/mL 마우스 꼬리 콜라겐 I형(Rat tail collagen type I; Corning)과 보관중인 OB-dECM 용액을 혼합하여, Col/OB-dECM 복합 하이드로젤(composite hydrogel)을 제조하였다. 하이드로젤의 최적의 조성은 2 mg/mL 콜라겐 및 1 mg/mL OB-dECM이다.Subsequently, a Col/OB-dECM composite hydrogel was prepared by mixing 3-4 mg/mL rat tail collagen type I (Corning) and the OB-dECM solution in storage. The optimal composition of the hydrogel is 2 mg/mL collagen and 1 mg/mL OB-dECM.
겔화를 위해 0.5 M 수산화나트륨을 이용하여 중화시키고, 10 x PBS를 이용하여 하이드로젤의 삼투압을 조절하였다.For gelation, neutralization was performed using 0.5 M sodium hydroxide, and osmotic pressure of the hydrogel was adjusted using 10 x PBS.
1-4. 골 조직 모사체의 제조1-4. Preparation of bone tissue mimics
Col/OB-dECM과 1 x 106 cell/ml의 마우스 골세포(osteocyte, IDG-SW3)를 혼합한 후, 상기 하이드로젤을 지지체(칩) 상에 적가(loading)한 후, pH 7.5 및 37℃에서 30분 동안 젤화(gelation)를 진행하였다.After mixing Col/OB-dECM and 1 x 10 6 cell/ml of mouse bone cells (osteocyte, IDG-SW3), the hydrogel was loaded onto a scaffold (chip), and pH 7.5 and 37 Gelation was performed at °C for 30 minutes.
젤화가 완전히 완료된 후 젤화된 하이드로젤 외측부 가장자리를 둘러싸도록 5 x 102 cell/ml의 조골세포(Osteoblast, MC3T3-E1)를 접종(seeding)한 후, 골형성 배지에서 배양하고, 2일마다 배지를 교체하여 골 조직 모사체를 제조하였다.After complete gelation, 5 x 10 2 cell/ml of osteoblasts (Osteoblast, MC3T3-E1) were seeded to surround the outer edge of the gelated hydrogel, cultured in an osteogenic medium, and cultured in the medium every 2 days. was replaced to prepare a bone tissue mimetic.
본 발명에 따른 골조직 모사체는 중심부 지지체상에 하이드로 젤 내부에 골세포가 존재하고, 하이드로젤 주변부로 조골세포가 수평적으로 배치되는 구조로 광학적인 이미지로 세포-세포간 상호작용을 관찰할 수 있는 구조를 갖는 것을 확인할 수 있다.The bone tissue mimic according to the present invention has a structure in which osteoblasts are present inside the hydrogel on the central support and osteoblasts are horizontally arranged around the hydrogel, and cell-cell interactions can be observed with an optical image. It can be seen that the structure has a
[실시예 2] 종래의 골 조직 모사 플랫폼과의 비교 실험[Example 2] Comparative experiment with conventional bone tissue simulation platform
종래의 골 조직 모사 플랫폼과의 성능 비교를 위해 종래 기술의 시스템(Transwell 시스템)과 본 발명의 골조직 모사체 시스템을 구성하여 공배양 실험을 진행하였다. 본 공배양 실험에서 사용한 상기 종래 기술의 시스템과 본 발명의 시스템(Bone-on-a-chip)에 대한 모식도를 도 15에 나타내었다.For performance comparison with the conventional bone tissue simulating platform, a conventional system (Transwell system) and the bone tissue simulating system of the present invention were constructed and a co-culture experiment was conducted. 15 shows a schematic diagram of the system of the prior art and the system of the present invention (Bone-on-a-chip) used in this co-culture experiment.
양 시스템에서 실험 조건을 동일하게 진행하였다. 즉, 골세포로서 IDG-SW3 을 1 x 106 cells/ml 의 세포 농도로 하여 2 mg/ml의 콜라겐 및 1 mg/ml ECM과 함께 포함하는 겔을 준비하고, pH 7.5 및 37℃에서 30분 동안 젤화(gelation) 하였으며, 그 후 조골 세포로 MC3T3-E1을 5 x 102 cells/well 의 세포 농도로 배치하여 각각의 시스템을 구성하였다. 6일간 공배양한 후 다음과 같이 조골 세포의 성장성 및 골 세포의 분화력을 확인하였다.Experimental conditions were performed identically in both systems. That is, a gel containing IDG-SW3 as bone cells at a cell concentration of 1 x 10 6 cells/ml together with 2 mg/ml collagen and 1 mg/ml ECM was prepared, and the gel was maintained at pH 7.5 and 37°C for 30 minutes. During gelation, and then MC3T3-E1 as osteoblasts was placed at a cell concentration of 5 x 10 2 cells/well to configure each system. After co-culture for 6 days, the growth potential of osteoblasts and the differentiation ability of osteoblasts were confirmed as follows.
2-1. 조골 세포의 성장성2-1. growth of osteoblasts
각 시스템에서 조골 세포의 성장은 Proliferation assay (CCK assay)에 의해 시간의 경과에 따른 세포 활성도를 측정함으로써 확인하였으며, 여기서 상기 활성도는 발색을 이용한 흡광도 (OD450)로부터 측정된 것이다. 그 결과를 도 16에 나타내었다. 이로부터 공배양시 종래 기술의 시스템에서보다, 본 발명에 따른 골조직 모사체 시스템에서 조골 세포의 성장성이 현저히 높다는 것을 확인하였다.The growth of osteoblasts in each system was confirmed by measuring cell activity over time by a Proliferation assay (CCK assay), where the activity was measured from absorbance (OD 450 ) using color development. The results are shown in FIG. 16 . From this, it was confirmed that the growth potential of osteoblast cells was significantly higher in the bone tissue mimetic system according to the present invention than in the prior art system during co-culture.
2-2. 골 세포에서의 세포 성숙도2-2. Cell maturity in osteocytes
각 시스템에서 골 세포의 세포 성숙도를 골분화 마커를 이용하여 확인하였다. 세포 성숙도의 확인을 위해 RNA 50 ng을 이용하여 RT-PCR을 수행하였고, 어닐링 온도는 59℃로 하였다. 결과는 도 17에 나타내었다. 이로부터 공배양시 종래 기술의 Transwell 시스템에서보다, 본 발명에 따른 시스템에서 골 세포의 분화력이 현저히 높다는 것을 확인할 수 있다.Cell maturity of bone cells in each system was confirmed using an osteogenic differentiation marker. To confirm cell maturity, RT-PCR was performed using 50 ng of RNA, and the annealing temperature was 59°C. Results are shown in FIG. 17 . From this, it can be confirmed that the differentiation capacity of bone cells is significantly higher in the system according to the present invention than in the prior art Transwell system during co-culture.
[실시예 3] 약물 스크리닝을 위한 골조직 모사체 영상 획득 [Example 3] Acquisition of bone mimetic image for drug screening
3-1. 골질환 관련 약물 선택3-1. Bone disease related drug choice
SOST(sclerostin)는 골세포에서 생성되어 분비되며 골밀도를 감소시키는 물질이다.SOST (sclerostin) is a substance that is produced and secreted from bone cells and reduces bone density.
도 13은 SOST의 작용에 관한 모식도로, 골세포에서 분비된 SOST는 조골세포로 전달되고, 이후 Frizzled/WNT 신호 전달 경로를 억제하며, 최종적으로 β-카테닌 분해를 촉진하여 유전자로부터의 전사(transcription)을 방해하고, 결과적으로 조골세포의 증식을 억제한다.13 is a schematic diagram of the action of SOST. SOST secreted from osteocytes is delivered to osteoblasts, then inhibits the Frizzled/WNT signaling pathway, and finally promotes β-catenin degradation to induce transcription from genes. ) and, as a result, suppress the proliferation of osteoblasts.
상기 억제 기전의 원리로부터, 골다공증 치료제는 SOST의 특정부위에 효과적으로 결합하는 단일클론항체로 제작될 수 있다. 즉 골다공증 치료제는 SOST에 결합하여 기능을 하지 못하게 하고, 궁극적으로는 β-카테닌이 분해되지 않도록 하여 β-카테닌의 핵(nucleus)으로의 이동을 촉진시키며, 궁극적으로 조골세포의 증식을 통해 뼈를 형성시키는 역할을 수행한다.Based on the principle of the inhibition mechanism, a therapeutic agent for osteoporosis can be prepared as a monoclonal antibody that effectively binds to a specific site of SOST. In other words, the osteoporosis treatment binds to SOST to prevent its function, ultimately prevents β-catenin from being decomposed, promotes the movement of β-catenin to the nucleus, and ultimately promotes bone growth through the proliferation of osteoblasts. play a shaping role.
3-2. 골조직 모사체에 대한 약물 투여3-2. Drug Administration to Bone Tissue Mimics
골세포에서 방출되는 SOST의 농도를 이하와 같이 ELISA(enzyme-linked immunosorbent assay)로 정량하여, 처리할 약물(SOST-억제 단일클론항체)의 농도를 결정하였다.The concentration of SOST released from bone cells was quantified by ELISA (enzyme-linked immunosorbent assay) as follows to determine the concentration of the drug to be treated (SOST-inhibiting monoclonal antibody).
먼저, 실시예 1-4를 통해 제조한 골조직 모사체에서 배양된 골세포(IDG-SW3)의 배양액을 14일 동안 2~3일 주기로 수집한 후, SOST ELISA Kit(ALPCO, USA)로 측정하였다.First, the culture medium of bone cells (IDG-SW3) cultured in the bone tissue mimetic prepared in Examples 1-4 was collected every 2-3 days for 14 days, and then measured with the SOST ELISA Kit (ALPCO, USA). .
이로부터, 세포가 성숙함에 따라 골 세포가 분비하는 SOST 양이 증가하였고, 특히 10일 경과 후 급격히 상승하는 것을 확인하였다.From this, it was confirmed that the amount of SOST secreted by the bone cells increased as the cells matured, and in particular, it increased rapidly after 10 days.
상기 결과를 바탕으로, 도 14에 도시된 절차와 같이 먼저 10일 동안 골세포를 성숙시킨 후, 조골세포(MC3T3-E1)와 약물(SOST-억제 단일클론항체)을 처리한 후, 14일 경과 시 약물에 대한 조골세포의 반응을 분석하였다.Based on the above results, as in the procedure shown in FIG. 14, bone cells were first matured for 10 days, then osteoblasts (MC3T3-E1) and drugs (SOST-inhibiting monoclonal antibody) were treated, and then 14 days elapsed. The response of osteoblasts to the drug was analyzed.
형광 이미지의 획득을 위해, 하이드로젤 내 골세포(IDG-SW3)와 조골세포(MC3T3-E1)를 4% 파라포름 알데하이드(paraformaldehyde)을 이용하여 20분동안 고정시킨 뒤, 0.1%(v/v) Triton X-100을 이용하여 투과화(permeabilization)하고, 5% BSA(Bovine serum albumin)가 들어있는 PBS용액으로 이용하여 블록(blocking)하였다.To acquire fluorescence images, osteoblasts (IDG-SW3) and osteoblasts (MC3T3-E1) in the hydrogel were fixed for 20 minutes using 4% paraformaldehyde, followed by 0.1% (v/v) ) Triton X-100 was used for permeabilization, and blocking was performed using a PBS solution containing 5% BSA (Bovine serum albumin).
β-카테닌은 형광마커인 Alexa Fluor 488가 부착된 β-카테닌 항체(1:50, Santa cruz Biotechnology, USA)를 사용하여 면역염색하였고, 핵(Nucleus)은 형광마커인 Hoechst 33342(1:500, Thermo Fisher Scientific, USA)로 염색하였다.β-Catenin was immunostained using a β-catenin antibody (1:50, Santa cruz Biotechnology, USA) attached with the fluorescent marker Alexa Fluor 488, and the nucleus was immunostained with the fluorescent marker Hoechst 33342 (1:500, 1:500). Thermo Fisher Scientific, USA).
형광 이미지는 형광 현미경(Celena X high content imaging system, Logos Biosystems, Korea)과 공초점 현미경(LSM-710, Carl Zeiss, Germany)을 사용하여 확보하였다.Fluorescent images were obtained using a fluorescence microscope (Celena X high content imaging system, Logos Biosystems, Korea) and a confocal microscope (LSM-710, Carl Zeiss, Germany).
3-3. 골다공증 약물 테스트 및 이미지 분석3-3. Osteoporosis drug testing and image analysis
상기 실시예에서 제조한 골조직 모사체에서 IDG-SW3 세포를 OB-dECM 하이드로젤에서 10일 동안 배양한 후, MC3T3-E1 세포를 접종하여 공배양한 후, 세포에 20 ng/mL의 항-SOST 단일클론항체(Clone AbD09097_h/mlgG2a; Bio-Rad)를 골다공증 치료 약물로 같은 날 처리하였다.After culturing IDG-SW3 cells in the OB-dECM hydrogel for 10 days in the bone tissue mimetic prepared in the above example, inoculating MC3T3-E1 cells and co-cultivating them, the cells were then treated with 20 ng/mL of anti-SOST. A monoclonal antibody (Clone AbD09097_h/mlgG2a; Bio-Rad) was treated with an osteoporosis treatment drug on the same day.
CellProfiler-기반 모듈 (Celena X cell analyzer; Logos Biosystems)을 이용하여, MC3T3-E1 세포에서 β-카테닌의 강도 및 핵 전위를 측정하였다(N은 독립웰의 수이고, 웰당 3개의 이미지를 평균적으로 사용하였다).Using a CellProfiler-based module (Celena X cell analyzer; Logos Biosystems), the intensity and nuclear translocation of β-catenin were measured in MC3T3-E1 cells (N is the number of independent wells, and three images per well were averaged). did).
파이프라인 시퀀스(pipeline sequences)의 경우, 전체 및 겹친 부분에 대해 PrimaryObjects 모듈 식별(핵과 β-카테닌의 위치 식별), MaskImage 모듈(핵과 겹친 β-카테닌의 분할) 및 MeasureObjectIntensity 모듈(핵과 β-카테닌의 강도 결정)을 사용하였다.For pipeline sequences, identify the PrimaryObjects module (identify the location of nuclei and β-catenin), MaskImage module (segmentation of nuclei and overlapping β-catenin), and MeasureObjectIntensity modules (intensity of nuclei and β-catenin) for total and overlapping parts. crystal) was used.
β-카테닌 핵 전위율(nuclear translocation rates)은 다음 공식을 사용하여 계산하였다.β-catenin nuclear translocation rates were calculated using the following formula.
β-카테닌 핵 전위율 = In / It x 100, (It: 세포 내에서 β-카테닌의 강도, In: 핵에서 β-카테닌의 강도).β-catenin nuclear translocation rate = I n / I t x 100, (I t : intensity of β-catenin in the cell, I n : intensity of β-catenin in the nucleus).
3-4. 딥러닝을 활용한 이미지 분석에 의한 약물 후보군 평가 결과3-4. Results of drug candidate group evaluation by image analysis using deep learning
상기 과정을 통해 확보한 이미지에 대해, 상기 딥러닝 모델 구현 방법에 따라, 약물 처리 후 얻은 420개의 이미지와 약물 비처리 후 얻은 424개의 이미지에 대한 학습을 수행한 후, 항-SOST 단일클론항체 처리 결과의 정확도를 평가하였다.For the images obtained through the above process, according to the deep learning model implementation method, learning is performed on 420 images obtained after drug treatment and 424 images obtained after non-drug treatment, and then anti-SOST monoclonal antibody treatment The accuracy of the results was evaluated.
그 결과, 상기 표 1에 도시한 바와 같이, BN 모델의 경우 97.2%, BNM 모델의 경우 99.5%에 해당하는 테스트 세트가 정확한 판단을 내린 것을 확인하였다.As a result, as shown in Table 1 above, it was confirmed that the test set corresponding to 97.2% for the BN model and 99.5% for the BNM model made an accurate decision.
이 결과로부터, 본 발명에 따른 골조직 모사체에 대해 골 질환 관련 약물을 처리한 후, 이로부터 확보할 수 있는 이미지를 인공지능을 통한 학습을 통해 95% 이상의 높은 정확도로 약물 처리에 따른 효과의 유무를 판단할 수 있음을 확인하였으며, 특히, 본 발명에 따른 골조직 모사체를 이용한 경우, 기존 방법으로 세포를 배양하여 제작한 골조직 모사체 대비하여, 골 생성 메카니즘을 보다 잘 모사함으로써, 약물 후보물질의 활성 평가 및 스크리닝에 매우 적합한 모델인 것을 확인하여 본 발명을 완성하였다.From this result, after processing the bone disease-related drug for the bone tissue mimetic according to the present invention, the image that can be obtained from this is learned through artificial intelligence with a high accuracy of 95% or more, whether or not there is an effect of drug treatment It was confirmed that it can determine, in particular, when using the bone tissue mimetic according to the present invention, compared to the bone tissue mimetic prepared by culturing cells by the existing method, by better mimicking the bone production mechanism, the drug candidate The present invention was completed by confirming that the model was very suitable for activity evaluation and screening.
전술한 본 발명의 설명은 예시를 위한 것이며, 본 발명이 속하는 기술분야의 통상의 지식을 가진 자는 본 발명의 기술적 사상이나 필수적인 특징을 변경하지 않고서 다른 구체적인 형태로 쉽게 변형이 가능하다는 것을 이해할 수 있을 것이다. 그러므로 이상에서 기술한 실시예들은 모든 면에서 예시적인 것이며 한정적이 아닌 것으로 이해해야만 한다.The above description of the present invention is for illustrative purposes, and those skilled in the art can understand that it can be easily modified into other specific forms without changing the technical spirit or essential features of the present invention. will be. Therefore, the embodiments described above should be understood as illustrative in all respects and not limiting.

Claims (29)

  1. 골조직 모사체에 골질환에 관한 약물을 처리하는 단계;Treating the bone tissue matrix with a drug related to bone disease;
    약물을 처리한 골조직 모사체로부터 영상 또는 이미지 확보를 위해, 골조직 모사체를 마커로 표지하는 단계;Labeling the bone tissue matrix with a marker to obtain an image or image from the drug-treated bone tissue matrix;
    표지된 마커를 포함하는 골조직 모사체 영상 또는 이미지를 확보하는 단계;Obtaining a bone tissue matrix image or image including the labeled marker;
    상기 골조직 모사체 영상 또는 이미지를 딥러닝 모델로 학습시켜 특징 맵(feature map)을 산출하는 단계; 및Calculating a feature map by learning the bone tissue mimetic image or image as a deep learning model; and
    상기 특징 맵을 이용하여, 약물 효능 평가 정보를 제공하는 단계;providing drug efficacy evaluation information using the feature map;
    를 포함하는, 골 질환의 예방 또는 치료용 약물의 평가 방법.Including, a method for evaluating a drug for preventing or treating bone disease.
  2. 제 1항에 있어서,According to claim 1,
    상기 골조직 모사체는The bone tissue matrix is
    하판;lower plate;
    상기 하판의 상측에 위치한 상판;an upper plate located above the lower plate;
    상기 하판 또는 상판 중 어느 하나로부터 유래하여 상판과 하판 사이에 위치하는 돌기구조로 이루어진 배리어부;a barrier portion formed of a protruding structure derived from either the lower plate or the upper plate and located between the upper plate and the lower plate;
    상기 상판의 일측에 구비된 하나 이상의 세포시료 주입 가이드부;를 포함하고,Including; one or more cell sample injection guides provided on one side of the top plate,
    상기 상판의 크기는 상기 배리어부의 내측만을 폐쇄하는 크기여서, 배리어부의 외측 부분은 개방된 구조인 것을 특징으로 하는 세포배양용기에서 배양되는 것인, 골 질환의 예방 또는 치료용 약물의 평가 방법.The size of the upper plate is such that only the inner side of the barrier portion is closed, and the outer portion of the barrier portion is cultured in a cell culture vessel, characterized in that it has an open structure.
  3. 제 1항에 있어서,According to claim 1,
    상기 골조직 모사체는 골세포(osteocyte)를 포함하며, 중심부에 충진된 젤(gel); 및 상기 젤의 외측부를 둘러싸도록 배치된 세포 혼합물;을 포함하는, 골 질환의 예방 또는 치료용 약물의 평가 방법.The bone tissue matrix includes bone cells (osteocyte), filled in the center of the gel (gel); and a cell mixture arranged to surround the outer portion of the gel.
  4. 제 1항에 있어서,According to claim 1,
    상기 골 질환은 골다공증, 불완전 골형성증(osteogenesis imperfecta), 과골화증, 고칼슘혈증, 부갑상선 기능 항진증, 골연화증, 융해성 골질환, 골괴사증, 뼈의 파젯병, 골 골절, 류마티스 관절염, 골수염, 치주성 골 손실, 암에 의한 골 손실, 노인성 골 손실 및 구루병으로 이루어진 군에서 선택되는 것인, 골 질환의 예방 또는 치료용 약물의 평가 방법.The bone disease is osteoporosis, incomplete osteogenesis imperfecta, hyperostosis, hypercalcemia, hyperparathyroidism, osteomalacia, dissolving bone disease, osteonecrosis, Paget's disease of bone, bone fracture, rheumatoid arthritis, osteomyelitis, periodontal A method for evaluating a drug for preventing or treating bone disease, which is selected from the group consisting of bone loss, bone loss due to cancer, senile bone loss, and rickets.
  5. 제 3항에 있어서,According to claim 3,
    상기 세포 혼합물은 골원성세포, 조골세포, 파골세포, 면역세포 및 혈관세포로 이루어진 군에서 선택되는 적어도 하나 이상을 포함하는 것인, 골 질환의 예방 또는 치료용 약물의 평가 방법.The cell mixture comprises at least one or more selected from the group consisting of osteogenic cells, osteoblasts, osteoclasts, immune cells and vascular cells, a method for evaluating a drug for preventing or treating bone diseases.
  6. 제 1항에 있어서,According to claim 1,
    상기 딥러닝 모델은 골 질환의 예방 또는 치료 효과가 알려진 약물이 처리된 골조직 모사체의 영상 또는 이미지와 상기 약물이 처리되지 않은 골조직 모사체의 영상 또는 이미지를 학습데이터로 이용하여 학습된 것인, 골 질환의 예방 또는 치료용 약물의 평가 방법.The deep learning model is learned using, as learning data, an image or image of a bone mimic treated with a drug known to have a preventive or therapeutic effect on bone disease and an image or image of a bone mimic not treated with the drug as learning data. A method for evaluating a drug for preventing or treating bone disease.
  7. 제 1항에 있어서,According to claim 1,
    상기 약물은 상기 골조직 모사체 중심부의 골세포(osteocyte) 또는 외측부의 세포 혼합물 중 어느 하나 이상에 대해 DNA 합성, RNA 합성, 유전자 발현, 유전자 구조, 단백질 합성, 단백질 변형, 단백질 분비, 단백질 구조, 당 합성, 당 분비, 당 변형, 지질 분비, 세포막 합성, 세포 내 신호 전달, 세포 간 신호 전달, 세포 내 소기관, 세포의 분화, 세포의 분열 및 세포의 운동으로 이루어진 군에서 선택되는 어느 하나 이상에 영향을 주는 것인, 골 질환의 예방 또는 치료용 약물의 평가 방법.The drug is used for DNA synthesis, RNA synthesis, gene expression, gene structure, protein synthesis, protein modification, protein secretion, protein structure, and sugar for any one or more of the osteocytes in the center of the bone tissue matrix or the cell mixture in the outer part. Affects at least one selected from the group consisting of synthesis, secretion of sugar, modification of sugar, lipid secretion, cell membrane synthesis, intracellular signal transduction, intercellular signal transduction, intracellular organelles, cell differentiation, cell division, and cell movement To give a method for evaluating a drug for preventing or treating bone disease.
  8. 제 1항에 있어서,According to claim 1,
    상기 약물은 골 질환의 예방 또는 치료를 위한 천연물, 합성의약품, 복합의약품, 한약, 한약 추출물, 한약 제제, 생약, 단백질 의약품, 유전자 재조합 의약품, 세포 배양 의약품, 효소 의약품, 미생물 의약품, 항체 의약품, 호르몬 의약품, 방사선 의약품, 항체-약물 접합체, 세포 치료제 및 유전자 치료제로 이루어진 군에서 선택되는 것인, 골 질환의 예방 또는 치료용 약물의 평가 방법.The above drugs are natural products, synthetic drugs, combination drugs, herbal medicines, herbal medicine extracts, herbal preparations, herbal medicines, protein medicines, genetically recombinant medicines, cell culture medicines, enzyme medicines, microbial medicines, antibody medicines, hormones for the prevention or treatment of bone disease. A method for evaluating a drug for preventing or treating bone disease, which is selected from the group consisting of pharmaceuticals, radiopharmaceuticals, antibody-drug conjugates, cell therapy products, and gene therapy products.
  9. 제 1항에 있어서,According to claim 1,
    상기 마커는 염색약, 형광 물질, 인광 물질 및 방사능 물질로 이루어진 군에서 선택되는 어느 하나 이상을 더 첨가하여 얻는 것인, 골 질환의 예방 또는 치료용 약물의 평가 방법.The marker is obtained by further adding any one or more selected from the group consisting of a dye, a fluorescent material, a phosphorescent material and a radioactive material, a method for evaluating a drug for preventing or treating bone disease.
  10. 제 1항에 있어서,According to claim 1,
    상기 골조직 모사체 영상 또는 이미지는 서로 다른 대상을 식별하는 둘 이상의 마커를 처리하여 얻는 것인, 골 질환의 예방 또는 치료용 약물의 평가 방법.The bone tissue mimetic image or image is obtained by processing two or more markers that identify different objects, a method for evaluating a drug for preventing or treating bone disease.
  11. 제 1항에 있어서,According to claim 1,
    상기 마커로 표지된 영역은 상기 골조직 모사체의 세포, 핵, 리보솜, 리소좀, 골지체, 중심체, 세포막, 미토콘드리아, 미세소판, 미세섬유, 세포질, DNA, RNA, 핵산, 히스톤, 단백질, 당 단백질, 막 단백질, 탄수화물, 막 탄수화물, 지질, 콜레스테롤, 당 지질, 당 및 콜라겐으로 이루어진 군에서 선택되는 적어도 하나 이상을 나타내는 것인, 골 질환의 예방 또는 치료용 약물의 평가 방법.The region labeled with the marker is a cell, nucleus, ribosome, lysosome, Golgi apparatus, centrosome, cell membrane, mitochondria, microplatelet, microfilament, cytoplasm, DNA, RNA, nucleic acid, histone, protein, glycoprotein, membrane of the bone tissue mimetic. A method for evaluating a drug for preventing or treating bone disease, which represents at least one selected from the group consisting of proteins, carbohydrates, membrane carbohydrates, lipids, cholesterol, glycolipids, sugars, and collagen.
  12. 제 1항에 있어서,According to claim 1,
    상기 마커로 표지된 영역은 상기 골조직 모사체의 형상, 모양, 위치 및 움직임으로 이루어진 군에서 선택되는 적어도 하나 이상을 나타내는 것인, 골 질환의 예방 또는 치료용 약물의 평가 방법.The area labeled with the marker represents at least one or more selected from the group consisting of shape, shape, position and movement of the bone tissue mimetic, a method for evaluating a drug for preventing or treating bone disease.
  13. 제 1항에 있어서,According to claim 1,
    상기 특징 맵(feature map)은 상기 골조직 모사체의 영상 또는 이미지를 컨볼루셔널 신경망(CNN)을 이용하여 연산한 결과를 나타내는 것인, 골 질환의 예방 또는 치료용 약물의 평가 방법.The feature map (feature map) represents the result of calculating the image or image of the bone tissue mimetic using a convolutional neural network (CNN), a method for evaluating a drug for preventing or treating bone disease.
  14. 약물이 처리된 골조직 모사체 영상 또는 이미지를 입력받는 입력장치;an input device for receiving a drug-treated bone tissue mimetic image or image;
    상기 골조직 모사체 영상 또는 이미지를 이용하여 학습되어, 약물 효능을 평가할 수 있는 딥러닝 모델을 저장하는 저장장치;a storage device for storing a deep learning model that can be learned using the bone tissue mimetic image or image and evaluate drug efficacy;
    상기 입력장치로부터 골조직 모사체 영상 또는 이미지를 전송받아, 상기 저장장치에 저장된 딥러닝 모델에 입력하고, 상기 딥러닝 모델이 출력하는 값에 따라 약물의 골조직 모사체에 대한 영향을 판단하는 연산장치;를 포함하는, 골 질환의 예방 또는 치료용 약물의 평가 장치. an arithmetic device that receives a bone mimic image or image from the input device, inputs the image to the deep learning model stored in the storage device, and determines the effect of the drug on the bone mimetic according to a value output from the deep learning model; A device for evaluating drugs for the prevention or treatment of bone diseases comprising a.
  15. 제 14항에 있어서,According to claim 14,
    상기 골조직 모사체는 골세포(osteocyte)를 포함하며, 중심부에 충진된 젤(gel); 및 상기 젤의 외측부를 둘러싸도록 배치된 세포 혼합물;을 포함하는, 골 질환의 예방 또는 치료용 약물의 평가 장치.The bone tissue matrix includes bone cells (osteocyte), filled in the center of the gel (gel); and a cell mixture disposed to surround the outer portion of the gel; an apparatus for evaluating a drug for preventing or treating bone disease.
  16. 제 14항에 있어서,According to claim 14,
    상기 골조직 모사체 영상 또는 이미지는 약물이 처리된 골조직 모사체에 마커로 표지된 영역을 포함하는 것인, 골 질환의 예방 또는 치료용 약물의 평가 장치.The bone tissue matrix image or image includes a region marked with a marker on the drug-treated bone tissue matrix, the evaluation device for preventing or treating bone diseases.
  17. 제 15항에 있어서,According to claim 15,
    상기 세포 혼합물은 골원성세포, 조골세포, 파골세포, 면역세포 및 혈관세포로 이루어진 군에서 선택되는 적어도 하나 이상을 포함하는 것인, 골 질환의 예방 또는 치료용 약물의 평가 장치.The cell mixture comprises at least one or more selected from the group consisting of osteogenic cells, osteoblasts, osteoclasts, immune cells and vascular cells, a device for preventing or treating bone diseases.
  18. 제 16항에 있어서,According to claim 16,
    상기 마커는 염색약, 형광 물질, 인광 물질 및 방사능 물질로 이루어진 군에서 선택되는 어느 하나 이상을 더 첨가하여 얻는 것인, 골 질환의 예방 또는 치료용 약물의 평가 장치.The marker is obtained by further adding any one or more selected from the group consisting of dye, fluorescent material, phosphorescent material and radioactive material, a drug evaluation device for preventing or treating bone disease.
  19. 제 16항에 있어서,According to claim 16,
    상기 골조직 모사체 영상 또는 이미지는 서로 다른 대상을 식별하는 둘 이상의 마커를 처리하여 얻는 것인, 골 질환의 예방 또는 치료용 약물의 평가 장치.The bone tissue mimetic image or image is obtained by processing two or more markers for identifying different objects, a device for evaluating a drug for preventing or treating bone diseases.
  20. 골세포(osteocyte)를 포함하며 중심부에 충진된 젤(gel); 및A gel containing osteocytes and filled in the center; and
    상기 젤의 외측부를 둘러싸도록 배치된 세포 혼합물;a cell mixture arranged to surround the outer portion of the gel;
    을 포함하는, 골조직 모사체.Containing, bone tissue matrix.
  21. 제 20항에 있어서,21. The method of claim 20,
    상기 젤은 하이드로 젤인, 골조직 모사체.The gel is a hydrogel, a bone tissue mimetic.
  22. 제 20항에 있어서,21. The method of claim 20,
    상기 하이드로 젤은 마트리젤(Matrigel), 푸라매트릭스(Puramatrix), 콜라겐(Collagen), 피브린젤(Fibrin gel), 폴리에틸렌글리콜 디아크릴레이트(PEG-DA), 폴리에틸렌글리콜 디메사크릴레이트(PEG-DMA), 폴리나이팜 (PNIPAM), 폴록세이머(Poloxamer), 키토산(Chitosan), 아가로스(Agarose), 젤라틴(Gelatin), 히알루론산(Hyaluronic acid) 및 알지네이트(Alginate)로 이루어진 군에서 선택되는 어느 하나 이상을 포함하는, 골조직 모사체.The hydrogel is Matrigel, Puramatrix, collagen, fibrin gel, polyethylene glycol diacrylate (PEG-DA), polyethylene glycol dimethacrylate (PEG-DMA) At least one selected from the group consisting of PNIPAM, Poloxamer, Chitosan, Agarose, Gelatin, Hyaluronic acid, and Alginate. Containing, bone tissue matrix.
  23. 제 22항에 있어서,23. The method of claim 22,
    상기 하이드로 젤은 세포외기질(extracellular matrix; ECM)을 더 포함하는, 골조직 모사체.The hydrogel further comprises an extracellular matrix (ECM), a bone tissue mimic.
  24. 제 20항에 있어서,21. The method of claim 20,
    상기 세포 혼합물은 조골세포 및 파골세포 중 어느 하나 이상과 배양액을 포함하는, 골조직 모사체.Wherein the cell mixture comprises a culture solution with any one or more of osteoblasts and osteoclasts, bone tissue mimics.
  25. (S1) 골세포(Osteocyte), 세포외기질(extracellular matrix; ECM) 및 하이드로젤을 혼합하는 단계;(S1) mixing osteocytes, extracellular matrix (ECM) and hydrogel;
    (S2) 상기 혼합물을 지지체 상에 적가한 후 젤화(gelation)하는 단계;(S2) dropwise addition of the mixture onto a support and then gelation;
    (S3) 젤화된 하이드로젤 외측부 가장자리를 둘러싸도록 조골세포(Osteoblast)를 시딩(seeding)하는 단계; 및(S3) seeding osteoblasts to surround the outer edge of the gelled hydrogel; and
    (S4) 배양액을 주입하는 단계;(S4) injecting the culture medium;
    를 포함하는, 골조직 모사체의 제조방법.Containing, the method of producing a bone tissue mimic.
  26. 제 25항에 있어서,26. The method of claim 25,
    상기 (S1) 단계에서 마트리젤(matrigel)을 더 포함하여 혼합하는 것인, 골조직 모사체의 제조방법.In the (S1) step, matrigel (matrigel) is further included and mixed, a method for producing a bone tissue mimic.
  27. 제 25항에 있어서,26. The method of claim 25,
    상기 하이드로젤은 콜라겐(collagen)인, 골조직 모사체의 제조방법.The hydrogel is collagen (collagen), a method for producing a bone tissue mimic.
  28. 제 25항에 있어서,26. The method of claim 25,
    상기 (S2) 단계의 젤화는 pH 6 내지 8의 범위에서 이루어지는 것인, 골조직 모사체의 제조방법.Gelation of the step (S2) is a method for producing a bone tissue mimic which is made in the range of pH 6 to 8.
  29. 제 25항에 있어서,26. The method of claim 25,
    상기 (S3) 단계는 젤화된 하이드로젤 외측부 가장자리를 둘러싸도록 조골세포(osteoblast)와 함께 파골세포(osteoclast)를 접종하는 것인, 골조직 모사체의 제조방법.The step (S3) is to inoculate osteoclasts together with osteoblasts to surround the outer edge of the gelated hydrogel, a method for producing a bone tissue mimic.
PCT/KR2022/017713 2021-11-11 2022-11-11 Method and device for screening for osteoporosis drug by using bone tissue mimetic image analysis WO2023085837A1 (en)

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
KR10-2021-0155192 2021-11-11
KR20210155192 2021-11-11
KR1020220053246A KR102531759B1 (en) 2021-11-11 2022-04-29 Method and Apparatus for Screening Osteoporosis Drug Based on Image Analysis of Bone Tissue Mimetics
KR10-2022-0053246 2022-04-29

Publications (1)

Publication Number Publication Date
WO2023085837A1 true WO2023085837A1 (en) 2023-05-19

Family

ID=86336241

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2022/017713 WO2023085837A1 (en) 2021-11-11 2022-11-11 Method and device for screening for osteoporosis drug by using bone tissue mimetic image analysis

Country Status (1)

Country Link
WO (1) WO2023085837A1 (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090061479A1 (en) * 2006-03-30 2009-03-05 Clarke Mark S F Mineralized three-dimensional bone constructs
KR101846370B1 (en) * 2015-08-04 2018-04-06 재단법인 아산사회복지재단 Method and program for computing bone age by deep neural network
KR102095485B1 (en) * 2019-07-25 2020-03-31 주식회사 메디팹 An osteoporosis model comprising a calcium phosphate hydrogel composition and uses thereof
KR20210139191A (en) * 2020-05-13 2021-11-22 한국기초과학지원연구원 Bone tissue mimetics and method for preparing the same

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090061479A1 (en) * 2006-03-30 2009-03-05 Clarke Mark S F Mineralized three-dimensional bone constructs
KR101846370B1 (en) * 2015-08-04 2018-04-06 재단법인 아산사회복지재단 Method and program for computing bone age by deep neural network
KR102095485B1 (en) * 2019-07-25 2020-03-31 주식회사 메디팹 An osteoporosis model comprising a calcium phosphate hydrogel composition and uses thereof
KR20210139191A (en) * 2020-05-13 2021-11-22 한국기초과학지원연구원 Bone tissue mimetics and method for preparing the same

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
AVNET S, LIPRERI MV, POMPO GD, GRAZIANI G, BOANINI E, BALDINI N. : "DEVELOPMENT OF A BONE-ON-A-CHIP BASED ON A 3D OSTEOCYTIC NETWORK FOR THE SCREENING OF ANTI-OSTEOPOROTIC DRUGS | Bone & Joint", ORTHOP PROCS., vol. 103-B, no. supp_13, 1 January 2021 (2021-01-01), pages 28, XP093066649, DOI: 10.1302/1358-992X.2021.13.028 *
PAEK KYURIM, KIM SEULHA, TAK SUNGHO, KIM MIN KYEONG, PARK JUBIN, CHUNG SEOK, PARK TAI HYUN, KIM JEONG AH: "A high‐throughput biomimetic bone‐on‐a‐chip platform with artificial intelligence‐assisted image analysis for osteoporosis drug testing", BIOENGINEERING & TRANSLATIONAL MEDICINE, vol. 8, no. 1, 1 January 2023 (2023-01-01), XP093066655, ISSN: 2380-6761, DOI: 10.1002/btm2.10313 *

Similar Documents

Publication Publication Date Title
Ma et al. Deterministically patterned biomimetic human iPSC-derived hepatic model via rapid 3D bioprinting
Cosgrove et al. Nuclear envelope wrinkling predicts mesenchymal progenitor cell mechano-response in 2D and 3D microenvironments
Vermeulen et al. Generation of organized porcine testicular organoids in solubilized hydrogels from decellularized extracellular matrix
Yam et al. Characterization of human transition zone reveals a putative progenitor-enriched niche of corneal endothelium
Vazquez et al. Human bone derived collagen for the development of an artificial corneal endothelial graft. In vivo results in a rabbit model
Park et al. Development of a novel dual reproductive organ on a chip: recapitulating bidirectional endocrine crosstalk between the uterine endometrium and the ovary
WO2023085837A1 (en) Method and device for screening for osteoporosis drug by using bone tissue mimetic image analysis
Zander et al. In vitro studies of primary explosive blast loading on neurons
Qi et al. IL-1β decreases the elastic modulus of human tenocytes
Edwards et al. Three-dimensional models of the cervicovaginal epithelia to study host–microbiome interactions and sexually transmitted infections
Yavitt et al. In situ modulation of intestinal organoid epithelial curvature through photoinduced viscoelasticity directs crypt morphogenesis
Mearns-Spragg et al. The biological evaluation of jellyfish collagen as a new research tool for the growth and culture of iPSC derived microglia
Huang et al. Lipopolysaccharide induced intestinal epithelial injury: a novel organoids-based model for sepsis in vitro
WO2020060222A2 (en) Microfluidic device for cerebrovascular simulation and high-efficiency blood-brain barrier simulation system comprising same
Mbanefo et al. IPSE, an abundant egg-secreted protein of the carcinogenic helminth Schistosoma haematobium, promotes proliferation of bladder cancer cells and angiogenesis
Wodzanowski et al. Multiscale invasion assay for probing macrophage response to gram-negative bacteria
Elomaa et al. Bioactive photocrosslinkable resin solely based on refined decellularized small intestine submucosa for vat photopolymerization of in vitro tissue mimics
WO2022250406A1 (en) Non-alcoholic fatty liver artificial tissue model
WO2023244055A1 (en) Novel use of pbmc-derived cytotoxic t cells
WO2021256723A1 (en) Transparentizing pretreatment method of biological sample having size of at most 1 mm, and transparentizing method of biological sample including same
Sudo et al. Bile canalicular formation in hepatic organoid reconstructed by rat small hepatocytes and nonparenchymal cells
WO2018135891A1 (en) Three-dimensional cell culturing scaffold comprising mammal-derived mesenchymal stem cell or cell differentiated therefrom, three-dimensional cell culturing method using same, drug screening method using same, and use thereof
WO2021162530A1 (en) Decellularized heart extracellular matrix-derived scaffold for culturing and transplanting heart organoid, and preparation method therefor
KR102531759B1 (en) Method and Apparatus for Screening Osteoporosis Drug Based on Image Analysis of Bone Tissue Mimetics
Lotowska et al. Ultrastructural characteristics of the respective forms of hepatic stellate cells in chronic hepatitis B as an example of high fibroblastic cell plasticity. The first assessment in children

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22893256

Country of ref document: EP

Kind code of ref document: A1