Method for screening efficient low-toxicity mRNA delivery vectors based on a library of ionizable lipids
By using an automated microfluidic platform and a live bioluminescence imaging system to screen for highly efficient and low-toxicity mRNA delivery vectors, the problems of high cytotoxicity and low throughput in existing technologies have been solved, enabling efficient and safe screening and optimization of mRNA delivery vectors.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HEFEI AFANA BIOTECHNOLOGY CO LTD
- Filing Date
- 2025-10-29
- Publication Date
- 2026-06-19
AI Technical Summary
Existing lipid library-based mRNA delivery vector screening methods suffer from the problem that high expression is often accompanied by high cytotoxicity, and the in vivo validation process is time-consuming and has low throughput, making it difficult to optimize key properties such as liver/spleen targeting.
A lipid nanoparticle library was prepared using an automated microfluidic platform. In vivo rapid validation was achieved through high-throughput in vitro screening combined with an in vivo bioluminescence imaging system. Highly efficient and low-toxic lipids were screened using efficacy scores and liver-spleen ratio indicators. The lipid library design was optimized by combining machine learning algorithms.
This enabled the screening of highly efficient and low-toxicity mRNA delivery vectors, shortened the in vivo validation cycle, improved screening throughput and targeting, and ensured the safety and efficacy of the candidates.
Smart Images
Figure SMS_1
Abstract
Description
Technical Field
[0001] This invention relates to the field of biotechnology, specifically to a method for screening highly efficient and low-toxicity mRNA delivery vectors based on an ionizable lipid library. Background Art
[0002] In recent years, significant progress has been made in the research of RNA therapy. However, RNA itself is easily degraded by nucleases, and its large molecular weight and negative charge make it difficult for RNA to directly cross the cell membrane and enter the cell. The emergence of lipid nanoparticles (LNPs) has solved the problem of RNA delivery.
[0003] Ionizable lipid libraries, as a core tool for mRNA delivery vector development, systematically construct structurally diverse collections of lipid molecules through combinatorial chemistry strategies. The establishment of ionizable lipid libraries can accelerate lead compound discovery, guide structure-activity relationship analysis, and support breakthroughs in key delivery systems. However, existing lipid library-based screening methods still have significant drawbacks: First, traditional screening paradigms only focus on reporter gene expression intensity (such as RLU value), ignoring the problem that high expression is often accompanied by high cytotoxicity. The lack of simultaneous quantification and correlation analysis of cytotoxicity leads to a large number of "false positive" candidates mixed in with the initial screening results, which have a very high risk of failure in subsequent development due to safety issues. Second, in the in vivo validation stage, the initial screening hits require multiple rounds of long-term animal experiments, involving dissection and separation of organs and quantification of mRNA expression using methods such as qPCR / ELISA. A single evaluation cycle exceeds 2 weeks and has extremely low throughput, which severely restricts the efficiency of optimizing key attributes such as liver / spleen targeting.
[0004] To address the aforementioned issues, we propose an improvement by screening efficient and low-toxicity mRNA delivery vectors based on ionizable lipid libraries. Summary of the Invention
[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:
[0006] This invention provides a method for screening efficient and low-toxicity mRNA delivery vectors based on an ionizable lipid library, comprising the following steps:
[0007] S1. Provides a lipid library containing at least 100 structurally diverse ionizable lipids, wherein the structural differences of each ionizable lipid are selected from at least two of the following: head group type, linker chemical bond type, and hydrophobic tail chain length;
[0008] S2. Using an automated microfluidic platform, each ionizable lipid in the lipid library is mixed with helper lipids, cholesterol, PEG-lipids and reporter gene mRNA, respectively, to prepare a lipid nanoparticle (LNP) library in parallel.
[0009] S3. Perform high-throughput in vitro screening on the LNP library:
[0010] a. Transfect each LNP into the target cell line.
[0011] b. Simultaneously detect reporter gene expression intensity and cell viability.
[0012] c. Calculate the efficacy score of each LNP (normalized efficiency = reporter gene expression intensity / cell viability value);
[0013] S4. Select the ionizable lipids corresponding to the LNPs that rank in the top 10% of efficacy scores and have cell viability values greater than 80% as Hit (initial screening positive candidates).
[0014] S5. Perform rapid in vivo verification of the Hit:
[0015] a. The Hit lipid was prepared into LNPs encapsulating luciferase mRNA and injected into animal models.
[0016] b. Quantitative analysis of liver / spleen / lung mRNA expression distribution at 6h and 24h time points using an in vivo bioluminescence imaging system.
[0017] c. Detect the luminescence intensity value (RLU / mg) of a unit protein in the target organ tissue.
[0018] S6. The final lipids were determined based on the criteria that the liver expression intensity in vivo was 200% higher than that of the positive control and ALT was less than 100 U / L.
[0019] As a preferred embodiment of the present invention, the reporter gene expression intensity is the luciferase chemiluminescence value (RLU), and the cell viability value is obtained by the CellTiter-Glo ATP assay.
[0020] As a preferred embodiment of the present invention, the efficacy score calculation includes filtering conditions: LNP particle size of 70-150nm, polydispersity index (PDI) of less than 0.3, and mRNA encapsulation efficiency of greater than 85%.
[0021] As a preferred technical solution of the present invention, the rapid in vivo verification includes calculating liver-targeting specific indicators: the selection criterion for Hit lipids is a liver-spleen ratio greater than 5, where liver-spleen ratio = liver RLU / mg ÷ spleen RLU / mg.
[0022] As a preferred embodiment of the present invention, the lipid molecules in the ionizable lipid library contain biodegradable linkers, which are selected from ester bonds, acetal bonds, or disulfide bonds.
[0023] As a preferred technical solution of the present invention, based on the structural parameters and performance data of the final selected lipids, it also includes using machine learning algorithms to establish a structure-activity relationship (SAR) model to generate design rules for a new generation of lipid libraries.
[0024] As a preferred embodiment of the present invention, the structure of the ionizable lipids in the lipid library is [head group]-[linker]-[hydrophobic tail chain], wherein:
[0025] The head group is selected from: primary amine, secondary amine, tertiary amine, piperazine or imidazole ring;
[0026] The linker is selected from: biodegradable ester bonds, acetal bonds, or disulfide bonds;
[0027] The hydrophobic tail chain comprises one or two independent alkyl or alkenyl chains with a carbon chain length of C8-C18, wherein the alkyl or alkenyl chains may be straight chains or contain one or more methyl branches.
[0028] As a preferred embodiment of the present invention, a structure-activity relationship (SAR) model is established using machine learning algorithms to generate design rules for a new generation of lipid libraries, specifically including:
[0029] A. Feature Engineering: The structural features of the final selected lipids and other lipids in the library are parameterized and encoded. The structural features include: head group type and pKa calculation value, linker type, hydrophobic tail chain length, unsaturation degree and branching degree.
[0030] B. Model Training: Using the parameterized encoded structural features as input variables and the corresponding in vivo liver expression intensity and cell viability values as output variables, the SAR model is trained using the random forest regression algorithm.
[0031] C. Virtual screening: The performance of the virtual lipid library was predicted using a trained SAR model, and virtual lipids that were predicted to have a liver expression intensity greater than 150% of the positive control and a predicted cell viability greater than 85% were screened out.
[0032] D. Library iteration: Based on the screening results, the virtual lipids are synthesized to form a second-generation lipid library, and the screening process described in claim 1 is repeated.
[0033] As a preferred embodiment of the present invention, a system for screening highly efficient and low-toxicity mRNA delivery vectors based on an ionizable lipid library includes:
[0034] Lipid library management module: Stores ionizable lipid libraries containing structurally parameterized descriptions;
[0035] Automated LNP fabrication module: integrates microfluidic chip array and liquid handling robot to achieve parallel LNP assembly at the 96-well or 384-well plate level;
[0036] Dual-modal detection module: Includes a high-throughput microplate reader to simultaneously detect chemiluminescent signals (reporter gene expression) and fluorescence signals (cell viability) in cell culture plates.
[0037] In vivo bioluminescence imaging system, equipped with a high-sensitivity CCD camera and organ segmentation analysis software;
[0038] AI analytics server: performs performance scoring calculations, in vivo distribution quantification, and SAR modeling.
[0039] As a preferred embodiment of the present invention, the automated LNP preparation module includes:
[0040] The flow rate control accuracy of the microfluidic chip is ±0.1 μL / min;
[0041] An integrated tangential flow filtration unit automates LNP buffer exchange.
[0042] The dynamic light scattering instrument is connected online to provide real-time feedback of LNP particle size data to the preparation parameter control system.
[0043] The beneficial effects of this invention are:
[0044] At the algorithm level, this invention innovatively defines "efficacy score" (normalized efficiency = reporter gene expression intensity / cell viability value) as the core screening indicator. Through mathematical normalization, this algorithm fundamentally corrects the "false positive" bias caused by the "efficiency-only" approach in traditional methods, and realizes the intrinsic correlation and synchronous optimization of efficiency and toxicity.
[0045] At the evaluation criteria level, this invention has for the first time created a final selection criterion that quantitatively integrates the dual attributes of high efficacy and low toxicity: "in vivo liver expression intensity > 200% of positive control and ALT < 100 U / L". It also introduced "liver-spleen ratio > 5" as a quantitative indicator of liver-targeting specificity, providing a clear and objective decision-making basis for the safety and targeting of candidate products.
[0046] In terms of technical means, in the in vivo validation process, a live bioluminescence imaging system was creatively adopted to replace the traditional tissue dissection and quantitative methods, realizing non-invasive, in vivo, and multi-time point dynamic observation. This revolutionarily shortened the in vivo efficacy evaluation cycle from the traditional more than 14 days to less than 48 hours, greatly improving the screening throughput and iteration speed. Detailed Implementation
[0047] The preferred embodiments of the present invention are described below. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.
[0048] Example: A method for screening highly efficient and low-toxicity mRNA delivery vectors based on an ionizable lipid library, comprising the following steps:
[0049] S1. Provides a lipid library containing at least 100 structurally diverse ionizable lipids, wherein the structural differences of each ionizable lipid are selected from at least two of the following: head group type, linker chemical bond type, and hydrophobic tail chain length. The head group is a primary / secondary / tertiary amine or a cyclic amine (piperazine / imidazolium). The linker is degradable (ester bond / acetal bond / disulfide bond) or non-degradable (ether bond). The hydrophobic tail chain is a C8-C18 alkyl chain containing branches or double bonds.
[0050] S2. Using an automated microfluidic platform, lipids and mRNA were mixed at a constant flow rate using a microfluidic chip. Each ionizable lipid in the lipid library was mixed with helper lipids, cholesterol, PEG-lipids, and reporter gene mRNA, respectively, to prepare a lipid nanoparticle (LNP) library in parallel. The ratio of ionizable lipids, helper lipids, cholesterol, and PEG-lipids was 50:10:38.5:1.5.
[0051] S3. Perform high-throughput in vitro screening of the LNP library:
[0052] a. Transfect each LNP into the target cell line.
[0053] b. Simultaneously detect reporter gene expression intensity and cell viability.
[0054] c. Calculate the efficacy score of each LNP (normalized efficiency = reporter gene expression intensity / cell viability value). This step is one of the key innovations of this invention. Its purpose is to overcome the technical bias of traditional screening that "emphasizes efficiency and neglects toxicity". By introducing a normalization algorithm, it actively eliminates the interference of high toxicity on the screening results and ensures that the initial screening candidates have both high expression and high safety.
[0055] S4. Select the ionizable lipids corresponding to the LNPs that rank in the top 10% of efficacy scores and have cell viability values greater than 80% as Hit (initial screening positive candidates).
[0056] S5. Perform rapid in vivo validation of Hit:
[0057] a. The Hit lipid was prepared into LNPs encapsulating luciferase mRNA and injected into animal models.
[0058] bb used an in vivo bioluminescence imaging system to quantify the mRNA expression distribution in the liver, spleen, and lungs at 6h and 24h time points. This method is a revolutionary alternative to traditional anatomical quantification techniques, shortening the in vivo validation cycle from several weeks to 1-2 days and providing richer information on in vivo distribution dynamics.
[0059] c. Detect the luminescence intensity value (RLU / mg) of a unit protein in the target organ tissue.
[0060] S6. Based on the criteria of liver expression intensity > 200% of the positive control and ALT < 100 U / L, the final lipids were selected. This invention is the first to combine significant efficacy advantages (> 200% of the positive control) with clear preclinical liver safety indicators (ALT < 100 U / L), forming a strict final selection "double threshold", supplemented by the targeting requirement of "liver-spleen ratio > 5", thereby systematically ensuring the comprehensive excellent properties of the final lipids.
[0061] Table 1. Cell-level experimental data (initial screening → structure optimization → final selection and validation)
[0062]
[0063] Efficacy score = RLU / cell viability value × 100 (the higher the value, the better the overall performance; ×100 is only a normalization factor)
[0064] Note: The physical parameter filtration conditions are not met (particle size > 150nm, PDI > 0.3, encapsulation efficiency < 85%).
[0065] Hit criteria: Top 10% in efficacy score + cell viability >80% + qualified physical parameters.
[0066] Furthermore, the reporter gene expression intensity was measured as the luciferase chemiluminescence value (RLU), and cell viability was obtained using the CellTiter-Glo ATP assay. The reporter gene assay used firefly luciferase mRNA with added D-luciferin substrate, and then the chemiluminescence signal was collected at a wavelength of 560 nm.
[0067] Specifically, the efficacy score calculation includes filtering conditions: LNP particle size of 70-150nm, polydispersity index (PDI) of less than 0.3, and mRNA encapsulation rate of greater than 85%. LNP particle size is detected by dynamic light scattering measurement. Qualified LNP particle size can ensure the permeability of the sinusoidal endothelial space, PDI of less than 0.3 can ensure batch consistency, and mRNA encapsulation rate of greater than 85% can prevent mRNA degradation and immune activation.
[0068] Specifically, rapid in vivo validation includes calculating liver-targeted specific indicators: the selection criteria for Hit lipids is a liver-spleen ratio greater than 5, where liver-spleen ratio = liver RLU / mg ÷ spleen RLU / mg, and the auxiliary indicator is a lung-liver ratio less than 0.3, where lung-liver ratio = lung RLU / mg ÷ liver RLU / mg, which can avoid non-target accumulation.
[0069] Specifically, the lipid molecules in the ionizable lipid library contain biodegradable linkers selected from ester bonds, acetal bonds, or disulfide bonds.
[0070] Furthermore, it also includes establishing a structure-activity relationship (SAR) model based on the structural parameters and performance data of the final selected lipids using machine learning algorithms, generating design rules for a new generation of lipid libraries, and predicting liver expression efficiency based on random forest regression.
[0071] Furthermore, machine learning algorithms are used to establish structure-activity relationship (SAR) models and generate design rules for a new generation of lipid libraries, specifically including:
[0072] A. Feature Engineering: The structural features of the final selected lipids and other lipids in the library are parameterized and encoded. The structural features include: head group type and pKa calculation value, linker type, hydrophobic tail chain length, unsaturation degree and branching degree;
[0073]
[0074] C. Virtual screening: The performance of the virtual lipid library was predicted using a trained SAR model, and virtual lipids that were predicted to have a liver expression intensity greater than 150% of the positive control and a predicted cell viability greater than 85% were screened out.
[0075] D. Library Iteration: Based on the screening results, the virtual lipids are synthesized to form a second-generation lipid library, and the previous screening process is repeated.
[0076] Furthermore, the structure of the ionizable lipids in the lipid library is [head group]-[linker]-[hydrophobic tail chain], wherein:
[0077] The head group is selected from: primary amine, secondary amine, tertiary amine, piperazine, or imidazole ring;
[0078] Linkers are selected from: biodegradable ester bonds, acetal bonds, or disulfide bonds;
[0079] The hydrophobic tail chain contains one or two independent alkyl or alkenyl chains with a carbon chain length of C8-C18. The alkyl or alkenyl chains may be straight chains or contain one or more methyl branches.
[0080] A screening system based on ionizable lipid libraries for screening efficient and low-toxicity mRNA delivery vectors includes:
[0081] Lipid library management module: Stores ionizable lipid libraries with structural parameterization descriptions, including lipid ID, SMILES, pKa, etc.
[0082] Automated LNP fabrication module: integrates microfluidic chip array and liquid handling robot to achieve parallel LNP assembly at the 96-well or 384-well plate level;
[0083] Dual-modal detection module: Includes a high-throughput microplate reader to simultaneously detect chemiluminescent signals (reporter gene expression) and fluorescence signals (cell viability) in cell culture plates.
[0084] In vivo bioluminescence imaging system, equipped with a high-sensitivity CCD camera and organ segmentation analysis software;
[0085] AI analytics server: performs performance scoring calculations, in vivo distribution quantification, and SAR modeling.
[0086] Furthermore, the automated LNP preparation module includes:
[0087] The flow rate control accuracy of the microfluidic chip is ±0.1 μL / min;
[0088] An integrated tangential flow filtration unit automates LNP buffer exchange.
[0089] The dynamic light scattering instrument is connected online to provide real-time feedback of LNP particle size data to the preparation parameter control system.
[0090] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for screening highly efficient and low-toxicity mRNA delivery vectors based on an ionizable lipid library, characterized in that, Includes the following steps: S1. Provides a lipid library containing at least 100 structurally diverse ionizable lipids, wherein the structural differences of each ionizable lipid are selected from at least two of the following: head group type, linker chemical bond type, and hydrophobic tail chain length; S2. Using an automated microfluidic platform, each ionizable lipid in the lipid library is mixed with helper lipids, cholesterol, PEG-lipids and reporter gene mRNA respectively to prepare a lipid nanoparticle (LNP) library in parallel. S3. Perform high-throughput in vitro screening on the LNP library: a. Transfect each LNP into the target cell line. b. Simultaneously detect reporter gene expression intensity and cell viability. c. Calculate the efficacy score of each LNP (normalized efficiency = reporter gene expression intensity / cell viability value); S4. Select the ionizable lipids corresponding to LNPs that rank in the top 10% of efficacy scores and have a cell viability value greater than 80% as Hit (initial screening positive candidates). S5. Perform rapid in vivo validation of the Hit: a. The Hit lipid was prepared into LNPs encapsulating luciferase mRNA and injected into animal models. b. Quantitative analysis of liver / spleen / lung mRNA expression distribution at 6h and 24h time points using an in vivo bioluminescence imaging system. c. Detect the luminescence intensity value (RLU / mg) of a unit protein in the target organ tissue. S6. The final lipids were determined based on the criteria that the liver expression intensity in vivo was 200% higher than that of the positive control and that ALT was less than 100 U / L.
2. The method for screening efficient and low-toxicity mRNA delivery vectors based on an ionizable lipid library according to claim 1, characterized in that, The expression intensity of the reporter gene was quantified by detecting the luciferase chemiluminescence value (RLU), and the cell viability value was obtained by the CellTiter-Glo ATP assay.
3. The method for screening efficient and low-toxicity mRNA delivery vectors based on an ionizable lipid library according to claim 1, characterized in that, Before calculating the efficacy score, the method further includes a step of filtering the LNPs based on physical parameters, wherein the filtering conditions are: LNP particle size of 70-150 nm, polydispersity index (PDI) of less than 0.3 and mRNA encapsulation efficiency of greater than 85%.
4. The method for screening efficient and low-toxicity mRNA delivery vectors based on an ionizable lipid library according to claim 1, characterized in that, The rapid in vivo validation includes calculating liver-targeting specific indicators: the selection criteria for Hit lipids is a liver-spleen ratio greater than 5, where liver-spleen ratio = liver RLU / mg ÷ spleen RLU / mg.
5. The method for screening efficient and low-toxicity mRNA delivery vectors based on an ionizable lipid library according to claim 1, characterized in that, The lipid molecules in the ionizable lipid library contain biodegradable linkers selected from ester bonds, acetal bonds, or disulfide bonds.
6. The method for screening efficient and low-toxicity mRNA delivery vectors based on an ionizable lipid library according to claim 1, characterized in that, It also includes the use of machine learning algorithms to establish structure-activity relationship (SAR) models based on the structural parameters and performance data of the final selected lipids, and to generate design rules for a new generation of lipid libraries.
7. The system for screening efficient and low-toxicity mRNA delivery vectors based on an ionizable lipid library according to claim 6, characterized in that, Machine learning algorithms are used to establish structure-activity relationship (SAR) models and generate design rules for next-generation lipid libraries, specifically including: A. Feature Engineering: The structural features of the final selected lipids and other lipids in the library are parameterized and encoded. The structural features include: head group type and pKa calculation value, linker type, hydrophobic tail chain length, unsaturation degree and branching degree. B. Model training: Using the parameterized encoded structural features as input variables and the corresponding in vivo liver expression intensity and cell viability values as output variables, the SAR model is trained using the random forest regression algorithm; C. Virtual screening: The performance of the virtual lipid library was predicted using a trained SAR model, and virtual lipids that were predicted to have a liver expression intensity greater than 150% of the positive control and a predicted cell viability greater than 85% were screened. D. Library iteration: Based on the screening results, the virtual lipids are synthesized to form a second-generation lipid library, and the screening process described in claim 1 is repeated.
8. The system for screening efficient and low-toxicity mRNA delivery vectors based on an ionizable lipid library according to claim 1, characterized in that, The structure of the ionizable lipids in the lipid library is [head group]-[linker]-[hydrophobic tail chain], wherein: The head group is selected from: primary amine, secondary amine, tertiary amine, piperazine or imidazole ring; The linker is selected from: biodegradable ester bonds, acetal bonds, or disulfide bonds; The hydrophobic tail chain comprises one or two independent alkyl or alkenyl chains with a carbon chain length of C8-C18, wherein the alkyl or alkenyl chains may be straight chains or contain one or more methyl branches.
9. The system for screening highly efficient and low-toxicity mRNA delivery vectors based on an ionizable lipid library according to the method of claims 1-7, characterized in that, include: Lipid library management module: Stores ionizable lipid libraries containing structurally parameterized descriptions; Automated LNP fabrication module: integrates microfluidic chip array and liquid handling robot to achieve parallel LNP assembly at the 96-well or 384-well plate level; Dual-modal detection module: Includes a high-throughput microplate reader to simultaneously detect chemiluminescent signals (reporter gene expression) and fluorescence signals (cell viability) in cell culture plates. In vivo bioluminescence imaging system, equipped with a high-sensitivity CCD camera and organ segmentation analysis software; AI analytics server: performs performance scoring calculations, in vivo distribution quantification, and SAR modeling.
10. The system for screening efficient and low-toxicity mRNA delivery vectors based on an ionizable lipid library according to claim 9, characterized in that, The automated LNP preparation module includes: The flow rate control accuracy of the microfluidic chip is ±0.1 μL / min; An integrated tangential flow filtration unit automates LNP buffer exchange. The dynamic light scattering instrument is connected online to provide real-time feedback of LNP particle size data to the preparation parameter control system.