Method for enriching and screening sirna for gene silencing

Abstract

Provided is a method for enriching and screening siRNA, wherein the siRNA can be used for gene silencing. By means of a bioinformatics method and on the basis of sequence information of known siRNAs which have gene silencing effects and which have been clinically applied or are under research, the sequence characteristics of siRNA drugs that have clinically applied are obtained by means of a computational algorithm, and an siRNA sequence evaluation model is established on the basis of a computational method. The model is suitable for enriching and screening for siRNA sequences with clinical development value for different target genes.

Description

A method for enriching and screening siRNAs for gene silencing Technical Field

[0001] This invention relates to a method for enriching and screening siRNAs with gene silencing effects. Specifically, this invention uses bioinformatics methods to obtain the sequence characteristics of siRNA drugs successfully applied in clinical practice based on known clinical applications and research-produced siRNA sequences with gene silencing effects through computational algorithms. Based on these computational methods, an siRNA sequence evaluation model is established. This model is suitable for enriching and screening siRNA sequences with clinical development value targeting different genes. Background Technology

[0002] Small interfering RNA (siRNA) consists of two single-stranded RNA molecules, each 15 to 27 nucleotides in length, with a partially or completely complementary double-stranded structure. siRNA acts as a gene silencing or gene expression repressor within cells. Its mechanism of action involves binding to the mRNA of a target gene, leading to mRNA degradation or inhibition of translation. This gene-silencing ability holds significant potential for treating human diseases, and many researchers and commercial entities are currently investing heavily in developing therapies based on this technology.

[0003] In recent years, RNAi technology has made significant progress in the life sciences and medical fields, especially in drug development. As of October 2024, six RNAi drugs have been launched, such as Patisiran from Alnylam and Nedosiran from Dicerna. siRNA drugs are characterized by high specificity, high selectivity, and long-lasting efficacy, providing novel drug development strategies for multiple disease areas, including metabolic, neurological, and cardiovascular diseases, and have broad clinical application prospects.

[0004] Despite the enormous potential of siRNA drugs, designing siRNA sequences with clinical development value remains a challenge.

[0005] Although various siRNA design tools and rules exist, such as those identified by Reynolds et al. (who discovered multiple features influencing siRNA functionality, including avoiding inverted repeat sequences, lower G / C content (30%-52%), and base preference at positions 3, 10, 13, and 19 of the 19bp sense strand (Reynolds et al., 2004, Nat Biotechnol 22, 326-330); and Amarzguioui et al. (who reported a correlation between silencing efficiency and nucleic acid asymmetry at the ends of the siRNA, finding that siRNAs rich in AU at the 3' end and poor in AU at the 5' end had the best function (Amarzguioui et al., 2004, BIOCHEM BIOPH RES CO 316, 1050-1058), online tools like siRNA Wizard and siRNA Target Finder also exist. However, design rules reported on different websites and in the literature vary considerably, and the importance of different rules is unclear. Furthermore, these tools and rules primarily serve scientific research, and the clinical development potential of the sequences provided is unknown.

[0006] Therefore, this paper proposes a new siRNA design method with a drug development orientation. This method uses marketed and clinically-entering siRNA drug sequences as the training set, and comprehensively considers multiple characteristics such as siRNA nucleotide sequence, siRNA / miRNA structure, and off-target effects to screen out siRNA sequences with high drug development potential, thereby improving the efficiency of siRNA drug development. Summary of the Invention

[0007] This invention aims to improve the screening efficiency of siRNAs with gene silencing effects by using bioinformatics methods.

[0008] On one hand, the present invention provides a method for developing an algorithm for enriching and screening siRNAs, the method comprising: (a) selecting a set of siRNAs already in clinical application or clinical development as a training set; (b) selecting a set of siRNA sequence features and extracting feature data based on the training set; (c) based on (a) and (b), obtaining the correlation between the features and clinical development potential through statistical calculation, and screening features with high correlation; (d) based on the features obtained in (c), establishing a computational method model; and (e) selecting another set of siRNAs already in clinical application or clinical development as a test set to verify the effectiveness of the above model, wherein the siRNAs in the test set are all different from the siRNAs in the training set.

[0009] In some implementations, the method is a way to develop algorithms for enriching and screening siRNAs for gene silencing.

[0010] In some implementations, the training or test set contains siRNAs with gene silencing effects targeting at least two different targets.

[0011] In some implementations, the siRNA sequence characteristics may include one or more of the following characteristic types: content of various nucleotides, positional preference of various nucleotides, siRNA / mRNA structure, and siRNA specificity for potential targets.

[0012] In some embodiments, the siRNA sequence characteristics are selected from: GC content (%), G content (%), U content (%), G content (%), AU count (antisense strand 1-7), GC count (antisense strand 9-14), GC count (sense strand 1-5), antisense strand terminal nucleotide, sense strand terminal nucleotide, sense strand 3rd nucleotide, sense strand 10th nucleotide, sense strand 13th nucleotide, siRNA asymmetry, siRNA secondary structure, siRNA single nucleotide repeat, G repeatability, target mRNA region openness, target mRNA location, potential off-target effects, species homology, and sequence immunogenicity.

[0013] In some implementations, the computational method model may be: feature weighting, logistic regression, support vector machine, decision tree, random forest, Bayesian classification, or neural network.

[0014] Secondly, the present invention provides a method for enriching and screening siRNAs, the method comprising: (a) generating all siRNAs of a specified length for a target mRNA sequence; (b) performing sequence feature description on the siRNAs; (c) obtaining siRNA scores using a sequence feature-based computational model; and (d) screening siRNAs with potential gene silencing effects based on the siRNA score. In some embodiments, the screened siRNAs with potential gene silencing effects have clinical development potential, not only possessing target gene silencing effects but also meeting clinical development requirements in terms of efficacy in subjects, safety, off-target effects, immunogenicity, and synthetic difficulty.

[0015] In some implementations, the method is used for enriching and screening siRNAs for gene silencing.

[0016] In some implementations, the target is selected from TMPRSS6, AGER, MUC5B, DNAJC15 / MCJ, HMGCR, ANGPTL3, ALK, APOC3, TRPV1, KLKB1, complement component 3 (C3), ASGR1, DDIT4, Factor VIII, Factor IX, DPP4, STAT3, EPHA2, GPR146, APOA, MMP7, COL3A1, APP, Factor XII, ANGPTL8, MASP2, BCL2L12, SREBF2, CASP2, LDHA, PNPLA3, ALAS1, TLR7 / 8 / RIG-1, DUX4, HBV, ANGPTL4, DMD, CTGF, complement group The following are listed: C5, CD8, KIF11, ITGV6, DGAT2, COX-2, PRDM14, GPR75, KLK1, CD71, TP53, APOA1, Agtr1, TGFB1, CD4, PTGS2, MUC5AC, lipoprotein(a) (LPA), NPC1L1, ALDH2, AGT, KRAS, HSD17B13, DMPK, FactorX, SCAP, VEGF, TTR, CIDEB, GRB10 / 14, SERPINA1, INHBE, HAO1, KHK, MTARC1, complement factor B (CFB), SERPINAC1, CD40, PCSK9, XDH, SERPINAF2, FactorXI or COL1A1.

[0017] In some embodiments, the two nucleotide strands of the specified-length siRNA are each 15 to 27 nucleotides in length. In some embodiments, the sense strand of the siRNA is 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, or 27 nucleotides in length; the antisense strand of the siRNA is also 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, or 27 nucleotides in length. In some preferred embodiments, the double-stranded length of the siRNA is 19, 20, 21, 22, 23, 24, or 25 nucleotides. In some preferred embodiments, the double-stranded length of the siRNA is always 19. In some preferred embodiments, the double-stranded lengths of the siRNA are 19 and 21 nucleotides, respectively. In some preferred embodiments, the double-stranded lengths of the siRNA are 21 and 23 nucleotides, respectively.

[0018] In some implementations, the sequence features in step (b) above include one or more of the following feature types: content of various nucleotides, positional preference specificity of various nucleotides, siRNA / mRNA structure, and siRNA specificity to potential targets.

[0019] In some embodiments, the sequence features are selected from: GC content (%), G content (%), U content (%), AU count (antisense strand 1-7), GC count (antisense strand 9-14), GC count (sense strand 1-5), antisense strand terminal nucleotide, sense strand terminal nucleotide, sense strand 3rd nucleotide, sense strand 10th nucleotide, sense strand 13th nucleotide, siRNA asymmetry, siRNA secondary structure, siRNA single nucleotide repeat, G repeatability, target mRNA region openness, target mRNA location, potential off-target effects, species homology, and sequence immunogenicity. In some preferred embodiments, the sequence features are selected from: GC content (%), G content (%), U content (%), AU count (antisense strand 1-7), GC count (antisense strand 9-14), GC count (sense strand 1-5), antisense strand terminal nucleotide, sense strand terminal nucleotide, sense strand 10th nucleotide, sense strand 13th nucleotide, siRNA asymmetry, siRNA secondary structure, siRNA single nucleotide repeat, G repeatability, target mRNA region openness, target mRNA location, potential off-target effects, and species homology.

[0020] In some implementations, the sequence features are described according to the following rules:

[0021] (1) The GC content is between 30% and 52%. If this condition is met, the characteristic value is 1; otherwise, it is 0.

[0022] (2) If the U content is ≤40%, the characteristic value is 1; otherwise, it is 0.

[0023] (3) AU count (antisense chain 1-7): The A+U count of the first to seventh positions of the antisense chain is ≥4. If this condition is met, the characteristic value is 1; otherwise, it is 0.

[0024] (4) GC count (antisense chain 9-14): If the G+C count of the 9th to 14th positions of the antisense chain is ≤3, the characteristic value is 1; otherwise, it is 0.

[0025] (5) GC count (sense chain 1-5): The G+C count of the first to fifth positions of the sense chain is ≥2. If this condition is met, the characteristic value is 1; otherwise, it is 0.

[0026] (6) The antisense strand ends with a G or C nucleotide. If this condition is met, the characteristic value is 1; otherwise, it is 0.

[0027] (7) The 10th nucleotide of the sense strand is either A or U. If this condition is met, the characteristic value is 1; otherwise, it is 0.

[0028] (8) The nucleotide at the 13th position of the sense strand is not G. If this condition is met, the characteristic value is 1; otherwise, it is 0.

[0029] (9) The sense chain ends with nucleotides A or U. If this condition is met, the characteristic value is 1; otherwise, it is 0.

[0030] (10) siRNA asymmetry: asymmetry between the 3' and 5' ends of the sense strand, calculated by subtracting the A / U count of the first 3 positions of the sense strand from the A / U count of the first 3 positions of the sense strand at the 5' end;

[0031] (11) siRNA secondary structure: siRNA does not have an internal secondary structure and does not have an inverted repeat sequence of length ≥10. If it meets these conditions, the characteristic value is 0; otherwise, it is 1.

[0032] (12) siRNA single nucleotide repeat: There are no A / U single nucleotide repeats with a length > 8 and G / C single nucleotide repeats with a length > 4 in the siRNA sequence. If the condition is met, the characteristic value is 0; otherwise, it is 1.

[0033] (13) G repetition: The number of times G with a length greater than 2 is <2. If the characteristic value is 0, it is 1 otherwise.

[0034] (14) Openness of the target mRNA region: The mRNA region targeted by siRNA has good openness and is a single-stranded part that has not formed a secondary structure. If it meets the condition, the characteristic value is 1; otherwise, it is 0.

[0035] (15) The target mRNA is located 50 nucleotides downstream of the transcription start site and within the coding region. If this condition is met, the characteristic value is 1; otherwise, it is 0.

[0036] (16) Potential off-target: There are no potential off-target genes, that is, there are no genes other than the target gene that differ from the siRNA sequence by no more than three nucleotides. If this condition is met, the characteristic value is 1; otherwise, it is 0.

[0037] (17) Species homology: The homology of the siRNA target region in the listed species, including cynomolgus monkey, house mouse or brown rat. If the symptom is satisfied, the characteristic value is 1; otherwise, it is 0.

[0038] (18) If the G content is ≤40%, the characteristic value is 1; otherwise, it is 0.

[0039] In some implementations, the computational method model based on sequence features is

[0040] Where f iFor the sequence features described in step (b), w i These are the weight values ​​corresponding to the sequence features.

[0041] In some implementations, the weight values ​​corresponding to the sequence features are obtained through iterative optimization using heuristic methods based on the training set data. In some implementations, the weight values ​​corresponding to the sequence features are obtained through parameter tuning optimization based on the training set data. In some implementations, the weight values ​​corresponding to the sequence features are each independently selected from any value between -3 and 3. In some implementations, the weight values ​​can be any integer or decimal selected from -3 to 3. In some implementations, the weight values ​​corresponding to the sequence features are each independently selected from -3, -2.5, -2, -1.5, -1, -0.5, 0, 0.5, 1, 1.5, 2, 2.5, or 3.

[0042] In some implementations, the above method further includes testing the bioactivity of the screened siRNA.

[0043] Thirdly, the present invention provides a gene silencing siRNA molecule comprising a double-stranded region of 15 to 27 base pairs, wherein the double-stranded region consists of two antisense strands and two anti-complementary sense strands, and the siRNA has at least one or more of the following characteristics: (1) GC content between 30-60%, preferably between 30-52%; (2) U content ≤40%; (3) a high proportion of A and U at positions 1-7 of the antisense strand, preferably ≥4. (4) Fewer G and C at positions 9-14 of the antisense strand, preferably ≤3; (5) More G and C at positions 1-5 of the sense strand, preferably ≥2; (6) The terminal nucleotide of the antisense strand is G or C; (7) The nucleotide at position 10 of the sense strand is A or U; (8) The nucleotide at position 13 of the sense strand is not G; (9) The terminal nucleotide of the sense strand is A or U; (10) Asymmetry between the 3' and 5' ends of the sense strand, calculated by subtracting the A / U count from the first 3 positions of the 3' end of the sense strand. (11) The siRNA has no internal secondary structure and no inverted repeat sequences of length ≥10. (12) The siRNA sequence has no A / U single nucleotide repeats of length >8 and no G / C single nucleotide repeats of length >4. (13) The number of G repeats of length >2 is small, preferably the number of consecutive G repeats of length >2 is <2. (14) The mRNA region targeted by the siRNA has good openness and is a single strand that has not formed a secondary structure. (15) The target mRNA is located 50 nucleotides downstream of the transcription start site and within the coding region. (16) There are no potential off-target genes, that is, there are no other genes other than the target gene that differ from the siRNA sequence by no more than three nucleotides. (17) The siRNA target region is homologous to the listed species, including cynomolgus monkeys, house mice or brown rats. (18) The G content is ≤40%.

[0044] In some embodiments, the sense strand length of the siRNA is 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, or 27 nucleotides; the antisense strand length of the siRNA is 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, or 27 nucleotides. In some preferred embodiments, the double-strand length of the siRNA is 19, 20, 21, 22, 23, 24, or 25 nucleotides. In some preferred embodiments, the double-strand length of the siRNA is always 19 nucleotides. In some preferred embodiments, the double-strand lengths of the siRNA are 19 and 21 nucleotides, respectively. In some preferred embodiments, the double-strand lengths of the siRNA are 21 and 23 nucleotides, respectively.

[0045] In some embodiments, the target of the siRNA is selected from TMPRSS6, AGER, MUC5B, DNAJC15 / MCJ, HMGCR, ANGPTL3, ALK, APOC3, TRPV1, KLKB1, complement component 3 (C3), ASGR1, DDIT4, Factor VIII, Factor IX, DPP4, STAT3, EPHA2, GPR146, APOA, MMP7, COL3A1, APP, Factor XII, ANGPTL8, MASP2, BCL2L12, SREBF2, CASP2, LDHA, PNPLA3, ALAS1, TLR7 / 8 / RIG-1, DUX4, HBV, ANGPTL4, DMD, CTGF, Complement component 5 (C5), CD8, KIF11, ITGV6, DGAT2, COX-2, PRDM14, GPR75, KLK1, CD71, TP53, APOA1, Agtr1, TGFB1, CD4, PTGS2, MUC5AC, lipoprotein (a) (LPA), NPC1L1, ALDH2, AGT, KRAS, HSD17B13, DMPK, FactorX, SCAP, VEGF, TTR, CIDEB, GRB10 / 14, SERPINA1, INHBE, HAO1, KHK, MTARC1, complement factor B (CFB), SERPINAC1, CD40, PCSK9, XDH, SERPINAF2, FactorXI or COL1A1.

[0046] Fourthly, the present invention provides a gene silencing siRNA library, the library containing multiple different siRNAs that can produce gene silencing effects against at least one target gene, the siRNA having at least one or more of the following characteristics: (1) GC content between 30-60%, preferably between 30-52%; (2) U content ≤40%; (3) a large number of A and U at positions 1-7 of the antisense strand, preferably ≥4; (4) a small number of G and C at positions 9-14 of the antisense strand, preferably ≤3; (5) a large number of G and C at positions 1-5 of the sense strand, preferably ≥2; (6) the terminal nucleotide of the antisense strand is G or C; (7) the nucleotide at position 10 of the sense strand is A or U; (8) the nucleotide at position 13 of the sense strand is not G; (9) the terminal nucleotide of the sense strand is A or U; (10) asymmetry between the 3' and 5' ends of the sense strand, calculated by subtracting the A / U count of the first 3 positions of the sense strand from the 5' end of the sense strand. 'The A / U count of the first 3 positions is preferably ≥0, (11) the siRNA does not have internal secondary structure and does not have inverted repeat sequences of length ≥10, (12) the siRNA sequence does not have A / U single nucleotide repeats of length >8 and G / C single nucleotide repeats of length >4, (13) the number of G repeats of length >2 is small, preferably the number of consecutive G repeats of length >2 is <2, (14) the mRNA region targeted by the siRNA has good openness and is a single strand that has not formed secondary structure, (15) the target mRNA is located 50 nucleotides downstream of the transcription start site and within the coding region, (16) there are no potential off-target genes, that is, there are no other genes besides the target gene that differ from the siRNA sequence by no more than three nucleotides, (17) the siRNA target region is homologous to the listed species, including cynomolgus monkeys, house mice or brown rats, (18) the G content is ≤40%.

[0047] In some embodiments, the siRNA library contains siRNAs having at least one, two, three, four, five, six, seven, eight, nine, ten, eleven, twelve, thirteen, fourteen, fifteen, sixteen, seventeen, or all of the following characteristics: (1) GC content between 30-60%, preferably between 30-52%; (2) U content ≤40%; (3) a high proportion of A and U at positions 1-7 of the antisense strand, preferably... ≥4, (4) fewer G and C at positions 9-14 of the antisense strand, preferably ≤3, (5) more G and C at positions 1-5 of the sense strand, preferably ≥2, (6) the terminal nucleotide of the antisense strand is G or C, (7) the nucleotide at position 10 of the sense strand is A or U, (8) the nucleotide at position 13 of the sense strand is not G, (9) the terminal nucleotide of the sense strand is A or U, (10) asymmetry between the 3' and 5' ends of the sense strand, calculated by subtracting the A / U count from the first 3 positions of the 3' end of the sense strand. (11) The A / U count of the first 3 positions of the sense strand at the 5' end is preferably ≥0. (12) The siRNA does not have internal secondary structures and does not have inverted repeat sequences of length ≥10. (13) The siRNA sequence does not have A / U single nucleotide repeats of length >8 and G / C single nucleotide repeats of length >4. (14) The number of G repeats of length >2 is relatively small, preferably the number of consecutive G repeats of length >2 is <2. (15) The mRNA region targeted by the siRNA has good openness and is a single strand that has not formed secondary structures. (16) The target mRNA is located 50 nucleotides downstream of the transcription start site and within the coding region. (17) There are no potential off-target genes, that is, there are no other genes other than the target gene that differ from the siRNA sequence by no more than three nucleotides. (18) The siRNA target region is homologous to the listed species, including cynomolgus monkeys, house mice or brown rats. (19) The G content is ≤40%.

[0048] In some embodiments, the sense strand length of the siRNA is 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, or 27 nucleotides; the antisense strand length of the siRNA is 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, or 27 nucleotides. In some preferred embodiments, the double-strand length of the siRNA is 19, 20, 21, 22, 23, 24, or 25 nucleotides. In some preferred embodiments, the double-strand length of the siRNA is always 19 nucleotides. In some preferred embodiments, the double-strand lengths of the siRNA are 19 and 21 nucleotides, respectively. In some preferred embodiments, the double-strand lengths of the siRNA are 21 and 23 nucleotides, respectively.

[0049] In some implementations, the siRNA library contains siRNAs that have gene silencing effects on different target genes and have clinical development potential, and can be used for screening siRNA drugs.

[0050] In some embodiments, the target of the siRNA is selected from TMPRSS6, AGER, MUC5B, DNAJC15 / MCJ, HMGCR, ANGPTL3, ALK, APOC3, TRPV1, KLKB1, complement component 3 (C3), ASGR1, DDIT4, Factor VIII, Factor IX, DPP4, STAT3, EPHA2, GPR146, APOA, MMP7, COL3A1, APP, Factor XII, ANGPTL8, MASP2, BCL2L12, SREBF2, CASP2, LDHA, PNPLA3, ALAS1, TLR7 / 8 / RIG-1, DUX4, HBV, ANGPTL4, DMD, CTGF, Complement component 5 (C5), CD8, KIF11, ITGV6, DGAT2, COX-2, PRDM14, GPR75, KLK1, CD71, TP53, APOA1, Agtr1, TGFB1, CD4, PTGS2, MUC5AC, lipoprotein (a) (LPA), NPC1L1, ALDH2, AGT, KRAS, HSD17B13, DMPK, FactorX, SCAP, VEGF, TTR, CIDEB, GRB10 / 14, SERPINA1, INHBE, HAO1, KHK, MTARC1, complement factor B (CFB), SERPINAC1, CD40, PCSK9, XDH, SERPINAF2, FactorXI or COL1A1.

[0051] the term

[0052] To facilitate understanding of this disclosure, certain technical and scientific terms are specifically defined below. Unless otherwise expressly defined herein, all other technical and scientific terms used herein have the meanings commonly understood by one of ordinary skill in the art to which this disclosure pertains.

[0053] The term "siRNA" refers to small interfering RNA (siRNA), which consists of two single-stranded RNA molecules, each approximately 15 to 27 nucleotides in length, forming a partially or completely complementary double-stranded structure. siRNA acts as a gene silencing or gene expression inhibitor in cells by binding to the mRNA of a target gene, leading to mRNA degradation or inhibition of translation. The siRNA described herein can be a blunt-ended double-stranded RNA structure composed of two single-stranded RNA molecules of 15-27 nucleotides in length, or a structure with a protruding end consisting of 1-6 consecutive nucleotides at at least one end of the double-stranded structure. The double strands constituting the siRNA are a sense strand and an antisense strand. The nucleotide sequence of the sense strand is partially or identical to the nucleotide sequence of the siRNA's action site in the target mRNA; the nucleotide sequence of the antisense strand is partially or completely complementary to the siRNA's action site in the target mRNA. The sense and antisense strands of the siRNA mentioned herein can form a partially or completely complementary double-stranded structure. The term "siRNA" as used in this article generally refers to the naked sequence of siRNA, but may also include chemically modified siRNA sequences, siRNA molecules linked to delivery systems such as GalNAc derivatives, and further include compositions containing siRNA molecules and other auxiliary components, which can be directly used for clinical treatment, disease prevention, and basic research.

[0054] The term "complementarity" or "complementarity" is used in the general sense in this field and typically refers to the formation or presence of hydrogen bonds between one nucleic acid sequence and another through conventional Watson-Crick bonding or other non-traditional types of bonding. Complete complementarity can mean that all consecutive residues in one nucleic acid sequence form hydrogen bonds with the same number of consecutive residues in a second nucleic acid sequence. Partial complementarity within a nucleic acid molecule can include multiple mismatches or non-base-pair nucleotides (e.g., 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more mismatches, such as 1 to 3 mismatches, non-nucleotide linkers, or non-base-pair nucleotides).

[0055] The term "gene silencing" refers to the phenomenon in eukaryotes where specific genes are not expressed or their expression is reduced due to various reasons. It mainly occurs at two levels: transcriptional gene silencing (TGS), caused by DNA methylation, heterochromatinization, and position effects; and post-transcriptional gene silencing (PTGS), which occurs at the post-transcriptional level, inactivating genes through the specific degradation of target RNA. This process mainly includes co-inhibition and RNA interference (RNAi)-induced gene silencing. In RNA interference, small RNA molecules (such as siRNA or miRNA) bind to mRNA, preventing its translation or causing its degradation. The "gene silencing" discussed in this article mainly refers to the degradation of target RNA or the inhibition of translation caused by siRNA. The terms "target RNA," "target gene," "target mRNA," or "target mRNA" may include, but are not limited to: TMPRSS6, AGER, MUC5B, DNAJC15 / MCJ, HMGCR, ANGPTL3, ALK, APOC3, TRPV1, KLKB1, complement component 3 (C3), ASGR1, DDIT4, Factor VIII, Factor IX, DPP4, STAT3, EPHA2, GPR146, APOA, MMP7, COL3A1, APP, Factor XII, ANGPTL8, MASP2, BCL2L12, SREBF2, CASP2, LDHA, PNPLA3, ALAS1, TLR7 / 8 / RIG-1, DUX4, HBV, and ANGPTL. 4. DMD, CTGF, complement component 5 (C5), CD8, KIF11, ITGV6, DGAT2, COX-2, PRDM14, GPR75, KLK1, CD71, TP53, APOA1, Agtr1, TGFB1, CD4, PTGS2, MUC5AC, lipoprotein (a) (LPA), NPC1L1, ALDH2, AGT, KRAS, HSD17B13, DMPK, FactorX, SCAP, VEGF, TTR, CIDEB, GRB10 / 14, SERPINA1, INHBE, HAO1, KHK, MTARC1, complement factor B (CFB), SERPINAC1, CD40, PCSK9, XDH, SERPINAF2, FactorXI or COL1A1.

[0056] As described herein, "having a gene silencing effect" refers to the inhibitory or reducing effect of siRNA on the expression of target genes. Specifically, it means that siRNA-mediated reduction in target gene expression compared to the untreated group or normal control group. "Significant reduction" means a reduction in target gene expression of at least approximately 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 99%, or higher compared to normal or pre-treatment levels. siRNAs with gene silencing effects as described herein can be further developed into drugs for preventing, reducing, or improving diseases or symptoms in subjects using conventional techniques in this field. siRNAs with potential gene silencing effects as described herein not only have target gene silencing effects but also possess clinical development potential in terms of efficacy in subjects, safety, off-target effects, immunogenicity, and synthetic difficulty. For example, the siRNA with clinical development potential has the following characteristics: (1) high specificity, that is, it can specifically silence the expression of target genes and minimize off-target effects on non-target genes; (2) low immunogenicity, that is, through optimized sequence and modification, the possibility of it inducing an immune response in the human body is significantly reduced, ensuring good tolerability in long-term treatment; (3) synthetic feasibility, that is, the synthesis process is convenient, the purity is high, and it can be produced on a large scale; (4) high stability, that is, the chemically modified siRNA has good stability in vivo, is not easily degraded, and has a long half-life, thereby ensuring a more lasting therapeutic effect. It will also be readily understood by those skilled in the art that the siRNA with gene silencing effect selected and / or designed according to the methods described herein can be prepared according to any available technology, including but not limited to in vivo or in vitro chemical synthesis, enzymatic or chemical cleavage, or in vivo or in vitro template transcription. In some embodiments, the inhibition of gene expression and / or target gene expression can be determined by real-time quantitative PCR, or by other methods known in the art.

[0057] As used herein, "computational methods" or "algorithms" refer to methods that apply mathematical and computer science approaches to solve problems and perform calculations. Commonly used computational methods in the biomedical field mainly include machine learning, specifically including but not limited to feature weighting, logistic regression, support vector machines, decision trees, random forests, Bayesian classification, or neural networks. The computational methods and algorithms described herein can be interpreted according to the conventional understanding of those skilled in the art.

[0058] The term "subject" refers to an individual to whom siRNA will be delivered, for example, for experimental and / or therapeutic purposes. Preferred subjects are mammals, particularly domesticated mammals (e.g., dogs, cats, etc.), primates, or humans.

[0059] The terms “G”, “C”, “A”, and “U” typically represent nucleotides with bases of guanine, cytosine, adenine, and uracil, respectively. However, it is understood that the terms “ribonucleotide” or “nucleotide” can also refer to modified nucleotides, and those skilled in the art will clearly understand that guanine, cytosine, adenine, and uracil can be substituted with other moieties without substantially altering the base-pairing properties of the oligonucleotide containing such substituted moieties. In some cases, the "nucleotide" can be modified by 2'-sugar modification, such as replacing the -OH group at the 2' carbon position of the sugar structure of at least one nucleotide with -H, -CH3 (methyl), -OCH3 (-O-methyl), -NH2, -F, -O-2-methoxyethyl-O-propyl, -O-2-methylthioethyl, -O-3-aminopropyl, or -O-3-dimethylaminopropyl. It can also be modified in other ways, such as cholesterol conjugation, peptide nucleic acids, locked nucleic acids (LNAs), base modifications (e.g., introduction of 5-bromouracil, 5-iodouracil), terminal modifications (e.g., N-acetylgalactosamine modification), and thiophosphate nucleoside linkage modifications. As those skilled in the art will understand, chemical modifications can be determined using conventional techniques or other suitable assays to select those chemical modifications that effectively reduce the expression of the target gene. The extent to which the presence of such chemical modifications affects the efficiency of siRNA-mediated gene silencing remains to be determined, but regardless of the specific chemical modification applied, the method of the present invention can be used in any case to select preferred siRNA sequences.

[0060] The term "GC content" generally refers to the percentage of G and C nucleotides relative to the total number of nucleotides in the siRNA. In some embodiments, a GC content between 30% and 65% is advantageous for the gene silencing effect of the siRNA; in some preferred embodiments, the GC content is between 30% and 60%, and more preferably, the GC content is between 30% and 52%.

[0061] The term "G content" refers to the percentage of G nucleotides on the antisense strand relative to the total number of nucleotides on the antisense strand. Based on the inventors' research and development experience, a G content ≤40% is advantageous for the clinical development of siRNA.

[0062] The term "U content" refers to the percentage of U nucleotides on the antisense strand relative to the total number of nucleotides on the antisense strand. In some implementations, a U content of ≤40% is advantageous for the gene silencing effect of siRNA.

[0063] As described herein, nucleotide positions are typically counted starting from the 5' end of the sense or antisense strand of RNA, according to conventional understanding in the art. For example, "antesis chain terminal nucleotide" refers to the 3' terminal nucleotide of the antisense chain; "sense chain terminal nucleotide" refers to the 3' terminal nucleotide of the sense chain; "sense chain 3rd nucleotide" refers to the 3rd nucleotide starting from the 5' end of the sense chain; "sense chain 10th nucleotide" refers to the 10th nucleotide starting from the 5' end of the sense chain; "sense chain 13th nucleotide" refers to the 13th nucleotide starting from the 5' end of the sense chain; "AU count (antesis chain 1-7)" refers to the total number of A and U bases at positions 1-7 starting from the 5' end of the antisense chain; "GC count (antesis chain 9-14)" refers to the total number of G and C bases at positions 9-14 starting from the 5' end of the antisense chain; "GC count (sense chain 1-5)" refers to the total number of G and C bases at positions 1-5 starting from the 5' end of the sense chain.

[0064] The term "siRNA asymmetry" in this paper refers to the distributional bias of G / C and A / U nucleotides at the 3' and 5' ends of the siRNA, i.e., the difference in the number of A / U nucleotides at both ends. This characteristic can affect the functional efficiency and selectivity of the siRNA molecule. When designing and applying siRNA, terminal asymmetry can influence which strand will act as the guide strand to participate in the activity of the RNA-induced silencing complex. In this paper, this is calculated by subtracting the A / U count at the first three positions of the sense strand's 5' end from the A / U count at the first three positions of the sense strand's 3' end; in some preferred embodiments, this count difference is ≥0.

[0065] The term "siRNA secondary structure" refers to the local folding structure that an siRNA molecule may form in space; in this article, it refers to inverted repeat sequences where the sense or antisense strand of the siRNA exists. An inverted repeat sequence refers to a sequence in an RNA molecule where the bases of one sequence are inversely complementary to those of another sequence. When the base sequence of one sequence is complementary from 5' to 3' to the base sequence of another sequence from 3' to 5', it is called an inverted repeat. This structure causes RNA to self-pair and fold, forming a hairpin structure, which affects siRNA function. These sequences can be calculated using dynamic programming methods; in some preferred embodiments, the length of the inverted repeat sequence is <10.

[0066] The term "siRNA single nucleotide repeat" refers to the continuous repetition of the same nucleotide in an siRNA sequence. In this paper, it refers to the continuous repetition of any one of the nucleotides A, U, C, or G. This indicator is also a common feature used to describe siRNA. However, the types and lengths of repeating nucleotides that need to be restricted reported in the literature vary. Based on the inventors' research and development experience, this paper sets the following in some preferred embodiments: A / U single nucleotide repeat length ≤ 8, and G / C single nucleotide repeat length ≤ 4.

[0067] In this article, the term "G repetition" refers to the G appearing more than twice consecutively, such as GGG, GGGG, etc.

[0068] The term "G repeatability" in this paper refers to the number of G repeats on the strand with the most G repeats in the two strands of siRNA. Based on the inventors' research and development experience, this paper sets the frequency of occurrence to <2 in some preferred embodiments.

[0069] The term "openness of the targeted mRNA region" refers to the accessibility of the target mRNA sequence when siRNA or other RNA interference molecules bind to and act on mRNA. In this document, it refers to whether the siRNA-targeted mRNA region forms a secondary structure, facilitating siRNA recognition and binding. Tools used to calculate openness include, but are not limited to, Mfold / UnaFold and Vienna RNApackage. In some preferred embodiments, the siRNA-targeted mRNA region has good openness, consisting of single-stranded portions without secondary structures, facilitating siRNA recognition and binding.

[0070] The term "targeting mRNA site" refers to the binding site selected by siRNA or other RNA interference molecules on the mRNA sequence. The siRNA binding site described herein is assessed by distance from the first base at the 5' end of the mRNA. In some preferred embodiments, the targeting mRNA site is located 50 nucleotides downstream of the transcription start site and within the coding region. Here, the term "transcription start site" refers to the location of the first base transcribed at the 5' end of a gene. The term "coding region" refers to the portion of a gene that can be translated into a protein. Structurally, a gene is divided into a coding region and a non-coding region. The coding region is the portion that can be transcribed into messenger RNA and synthesized into the corresponding protein. On mRNA, the coding region is the sequence from the start codon to the stop codon. This sequence is recognized by ribosomes and used for translation into a polypeptide chain, i.e., the primary structure of a protein.

[0071] The term "potential off-target effect" is known in the art and refers to the non-specificity of siRNA during its action, where it may interact with genes other than the target gene to block gene expression instead of specifically blocking gene expression, producing unexpected effects. For example, it may lead to the degradation of mRNAs other than the intended target mRNA due to overlap and / or homology with secondary mRNA information. In this paper, the method used to measure the characteristics of "potential off-target effect" is to predict by computer whether there are other genes in the human genome that differ from the siRNA sequence by no more than three nucleotides, other than the target gene. If they are present, it is considered to have potential off-target effect. In practice, although some siRNAs have shown very good target gene silencing effects in in vivo or in vitro studies, they can also reduce or weaken the effects on non-target genes, leading to undesirable adverse reactions and limiting their clinical application. Therefore, assessing the potential off-target effect of siRNA in the early stages of drug screening is also very important for determining its clinical development potential.

[0072] The term "species homology" refers to the similarity of genes between different species. In this paper, species homology is used to measure whether gene sequences are identical between different species. In some preferred embodiments, the target gene exists in multiple species, and the target sequence corresponding to the siRNA is present and identical in the genes of multiple species. In preferred embodiments, the siRNA target region is homologous in the listed species such as cynomolgus monkeys, house mice, or brown rats. In this paper, "species homology" is a descriptive indicator set up in the early drug screening stage to fully consider subsequent preclinical development trials, avoiding the problem of greater difficulty in subsequent development due to differences in species homology. For example, some siRNAs are not homologous to conventional laboratory animals such as cynomolgus monkeys, house mice, or brown rats, and only transgenic humanized mice can be selected for experiments in subsequent studies. Therefore, the "species homology" characteristic introduced in this paper is used as a consideration indicator for screening siRNAs suitable for clinical development. Attached Figure Description

[0073] Figure 1 shows the nucleotide ratio distribution at each position of the 20 siRNAs in the training set.

[0074] Figure 2 shows the frequency distribution of different features (GC content (%), U content (%), G content (%), AU count (antisense strands 1-7), GC count (antisense strands 9-14), GC count (sense strands 1-5)) in the 20 siRNAs in the training set.

[0075] Figure 3 shows the frequency distribution of different features (siRNA asymmetry, siRNA secondary structure, siRNA single nucleotide repeats, and G repeatability) among the 20 siRNAs in the training set.

[0076] Figure 4A shows the dose-response curves of the attenuation effect of nine selected siRNAs on C3 mRNA in HepG2 cells.

[0077] Figure 4B shows the detection results of the attenuation effect of nine selected siRNAs on C3 mRNA in HepG2 cells at concentrations of 100 nM and 0.16 nM.

[0078] Figure 5A shows the detection results of the attenuation effect of nine selected siRNAs on C3 mRNA in PHH cells at concentrations of 1000 nM and 1 nM.

[0079] Figure 5B shows the dose-response curves of the attenuation effect of four selected siRNAs on C3 mRNA in PHH cells.

[0080] Figure 6 shows the results of testing different doses of siRNA to reduce human C3 protein in mice. Detailed Implementation

[0081] This invention will be described in conjunction with the embodiments described below. Unless otherwise specified herein, the sources of reagents and other experimental materials can be obtained from any molecular biology reagent supplier, provided their quality / purity meets the standards for molecular biology applications. The embodiments in this application are for illustrative purposes only and are not intended to limit the substance and scope of this application.

[0082] Example 1. Sequences and sequence features used to develop algorithms

[0083] Based on literature and patent reports, information on small nucleic acid drugs that have entered the clinical research stage was collected, including sequences of marketed small nucleic acid drugs and preferred sequences from related patents of small nucleic acid drugs under development. These were divided into a training set and a test set, as shown in Tables 1 and 2. The training set contains 20 siRNA drugs, including 6 marketed drugs, involving 15 different target genes; the test set contains 14 siRNA drugs, involving 11 different target genes.

[0084] Table 1 Training Set

[0085] Table 2 Test Set

[0086] Example 2. Construction of siRNA screening and enrichment algorithm

[0087] 2.1 Selection of Sequence Features

[0088] Based on the literature, there are various siRNA design rules, such as avoiding inverted repeat sequences, having a low G / C content (30%-52%), a base preference at positions 3, 10, 13, and 19 of the sense strand in 19bp, and nucleic acid asymmetry at both ends of the siRNA. By selecting some effective features reported in the literature and supplementing these features with the inventor's research and development experience, the final feature description set is shown in Table 3.

[0089] Table 3 Sequence Feature Descriptors

[0090] 2.2 Feature Filtering

[0091] The model utilizes Logomaker to perform statistical analysis on base position specificity, generating a distribution map of bases at different positions. This visually demonstrates the frequency differences of bases at different positions, revealing the importance and potential biological significance of base features at specific positions (Figure 1). Simultaneously, based on these features, the model generates a feature frequency matrix, quantifying the global frequency of features. Based on the feature matrix and distribution map (Figures 2 and 3), the model performs more refined feature selection, excluding features with minimal contribution to prediction or insignificant positional frequency, thus forming a more representative feature set and reducing model complexity. For example, a feature where the third nucleotide of the sense strand is A shows a low frequency in the statistical distribution and is therefore excluded.

[0092] 2.3 Construction of Algorithm Model

[0093] For the selected features, the model assigns weights based on their frequency distribution and saliency information in the distribution plot, using a feature-weighted approach to build the algorithm model:

[0094] Where f i For the sequence features selected in 2.2, w i These are the weight values ​​corresponding to the sequence features. Based on the training set data, the model iteratively adjusts the weights for each feature through a heuristic optimization process. Each weight value for a sequence feature is independently selected from a value between -3 and 3, and can be -3, -2.5, -2, -1.5, -1, -0.5, 0, 0.5, 1, 1.5, 2, 2.5, or 3.

[0095] 2.4 Validation of the Algorithm Model

[0096] The training set was divided according to a score of 5, 6, 7, 8, or 9 as a threshold, and the results are shown in Table 4.

[0097] Table 4. Evaluation Results of the Training Set

[0098] A score of 6 was chosen as the final threshold, and the model was evaluated using an independent test dataset. The results are shown in Table 5.

[0099] Table 5. Test Set Evaluation Results

[0100] The model's positive prediction value on the training set is 90%, indicating that the model has a strong screening ability during the learning process of the training data; at the same time, the positive prediction value on the test set is 78.6%, which still maintains a strong screening ability on unseen data, indicating that the model has good generalization ability and can screen out siRNAs with clinical development value for different genes.

[0101] Example 3. Validation of the siRNA screening and enrichment algorithm using CFB literature data.

[0102] To further verify the effectiveness of the siRNA design method constructed in Example 2, the publicly disclosed siRNAs targeting CFB (derived from patent US20230257749A1) were screened and evaluated using the algorithm described in Example 2.

[0103] The patent discloses over 700 naked siRNA sequences. The applicant screened 18 sequences from in vitro experiments for in vivo validation, deducing that these 18 sequences possess gene silencing effects and are suitable for further preclinical research. These 18 sequences were evaluated using the calculation method described herein, and 83.3% of the sequences could be screened using the method in Example 2. The scores for each molecule are shown in Table 6.

[0104] Table 6. The effect of CFB siRNA on the reduction of human CFB protein levels in mice and the computational model score.

[0105] Example 4. Validation of the siRNA screening and enrichment algorithm using C3

[0106] To further verify the effectiveness and adaptability of the siRNA screening and enrichment method constructed in Example 2, this paper also screened and analyzed siRNAs targeting C3.

[0107] To obtain a double-stranded nucleic acid molecule that induces highly efficient RNA interference targeting C3, a large number of siRNA molecules were initially designed using a 19 bp nucleotide sequence within the 1-5231 bp range of the relatively conserved human C3 gene mRNA sequence (Genbank ID: NM_000064, SEQ ID No. 87). These molecules were then screened using both wet and dry experiments as described in Example 2, and the experimental results were compared.

[0108] The method, steps, and results of the wet test are as described in patent application PCT / CN2024 / 100673:

[0109] First, 94 pairs of siRNA molecules (C3-1 to C3-94) were designed, and their preliminary inhibitory effects on C3 gene expression were tested in HepG2 cells (transfection concentration: 20 nM). The results showed that most molecules could effectively reduce the level of human C3 mRNA. Then, considering in vitro experimental data, nine optimal molecules (transfection concentration: 0.128 pM–100 nM) were further selected to verify the dose-dependent relationship of their inhibitory effects on C3 gene expression in HepG2 cells. The results are shown in Figures 4A and 4B.

[0110] Subsequently, the siRNA was chemically modified based on the corresponding naked sequence to obtain nine modified molecules, C3-95 to C3-103. The inhibitory effect of these modified siRNAs on C3 gene expression was verified in primary human hepatocytes (PHH). The results are shown in Figures 5A and 5B and Table 7. These molecules all showed good inhibitory effects on C3 mRNA, with C3-98, C3-96, C3-103 and C3-95 showing the best inhibitory effects, with IC50 values ​​of 1.6, 0.6, 0.3 and 1.0 nM, respectively.

[0111] Table 7. In vitro free uptake studies in PHH cells to test the C3 mRNA attenuation effect of siRNA.

[0112] After further chemical modification and optimization and screening of the GalNAc delivery system, two superior molecules (C3-127 and C3-130, both corresponding to the naked sequence of C3-96) were finally tested for in vivo activity in mice. The results showed that the selected siRNA molecules significantly reduced the level of human C3 protein in mouse serum at D7, D14 and D21, and the reducing effect could be sustained until D35 after administration, indicating that the siRNA has strong persistence in reducing the circulating C3 protein level in aged mice (as shown in Figure 6 and Table 8).

[0113] Table 8. In vivo studies in human C3 gene transgenic mice showed that different doses of GalNAc-siRNA conjugate were effective in reducing human C3 protein levels.

[0114] Similarly, the designed siRNA sequences were screened using the computational model described in Example 2. Data showed that 88.9% of the nine siRNAs selected through wet-enzyme experiments met the threshold of Score ≥ 6. Furthermore, the two siRNAs with the best inhibition efficiency in the free-uptake experiment both scored 10 points, significantly higher than the scores of other siRNAs (Table 9). The results of mouse activity tests also indicated that siRNA molecules with strong and sustained ability to reduce target genes / proteins in animals can be screened using the algorithm described herein.

[0115] Table 9. Wet experimental results and calculated scores of siRNA sequences targeting C3.

[0116] Example 5. Validation of siRNA screening and enrichment algorithm using CFB

[0117] To further verify the effectiveness and adaptability of the siRNA screening and enrichment method constructed in Example 2, this paper also uses this method to design, screen, and experimentally verify siRNAs for CFB targets from scratch.

[0118] To obtain a highly efficient double-stranded nucleic acid molecule that induces RNA interference targeting CFB, a large number of siRNA molecules were first designed by traversing a 19 bp nucleotide sequence within the 1-2476 bp range of the relatively conserved human CFB gene mRNA sequence (Genbank ID: NM_001710.6). Then, the molecules were screened using the dry experiment method described in Example 2, and the screening results were verified using a wet experiment.

[0119] Using the method described in Example 2, 510 siRNAs meeting a score ≥ 6 were screened from 2458 siRNAs as candidate siRNAs and sorted from highest to lowest score. The results were then validated using a wet assay. Specifically, for siRNA transfection of human primary hepatocytes (PHH cells, from BioIVT, HPCH10+), the cells were reverse-transfected in 384-well plates treated with type I collagen (Corning, catalog number #356705). The transfection process was as follows: First, 5 μl of Opti-MEM and 0.1 μl of Lipofectamine were added to each well of the 384-well plate. TM RNAiMAX transfection reagent (Invitrogen) TM (Catalog number #13778150), then add 5 μl of double-stranded siRNA and incubate at room temperature for 15 minutes. The final concentrations of all siRNAs were 0.1 nM and 1 nM, respectively. Silencer Thermo Fisher (s533333) and Silencer were used. TMNegative Control No. 2 siRNA (catalog number #AM4613) was selected as both positive and negative controls. Next, 40 μl of Invitrogro CP medium (BioIVT, catalog number #Z99029) containing approximately 10,000 cells was added to the siRNA mixture. After culturing the cells at 37°C and 5% CO2 for 24 hours, RNA purification was performed. Dynabeads was used... TM mRNA DIRECT™ Purification Kit (Invitrogen) TM mRNA was isolated from PHH cells (catalog number #61012). In short, after removing the culture medium from the cells, 50 μl of lysis / binding buffer was added. The cell culture plate was placed in a shaker and vortexed at 300 rpm for 30 minutes at room temperature. Then, 10 μl of cell lysis buffer was added to a 384-well PCR plate containing 10 μl of lysis buffer (containing 3 μl of magnetic beads) and incubated at room temperature for 5 minutes. Subsequently, the magnetic beads were captured, and the supernatant was removed from the 384-well magnetic plate (Permagen). The RNA bound to the magnetic beads was washed twice with 20 μl of wash buffer A, and then once with 20 μl of wash buffer B. Immediately after RNA extraction, reverse transcription was performed using the SuperScript VILO cDNA Synthesis Kit (Invitrogen, catalog number #11754050). CFB and TBP mRNA were measured using a Quantstudio 5 real-time PCR system (Invitrogen Life Sciences) and TaqMan qPCR primers (CFB-FAM, detection ID: Hs00156060_m1; TBP-VIC, detection ID: Hs00427620_m1). TBP mRNA was used as an internal control to normalize CFB mRNA in both siRNA-treated and untreated cells. The activity of a given CFB siRNA was expressed as the percentage of siRNA-normalized CFB mRNA expression divided by the percentage of siRNA-normalized CFB mRNA expression in the untreated control group. Data in Table 10 were analyzed using the comparative CT value method and are expressed as the remaining percentage of CFB mRNA relative to the untreated control group. Error is expressed as standard deviation (SD).

[0120] The results showed that most siRNAs could effectively reduce CFB mRNA levels. Among the top 100 siRNA molecules with the best overall performance (experimental results are shown in Table 10), 49 were enriched in the top 100 candidate siRNAs, 32 were enriched in candidate siRNAs ranked 101-300, and 19 were derived from candidate siRNAs ranked 301-510. The experimental results indicate that the top-ranked siRNAs are more likely to pass the experimental screening and have higher drug development potential.

[0121] Table 10. In vitro studies conducted in PHH cells to test the CFB mRNA attenuation effect of siRNA

[0122] In summary, the method for enriching and screening siRNA sequences for gene silencing provided by this invention is effective. It eliminates the need to test a large number of candidates to select effective siRNAs, making it more efficient and cost-effective than conventional experimental screening methods. Moreover, it can achieve good screening results for different target genes, greatly improving the efficiency of siRNA drug development.

Claims

1. A method for developing an algorithm for enriching and screening siRNA, the method comprising: (a) Select a set of siRNAs that are already in clinical use or clinical development as the training set; (b) Select a set of siRNA sequence features and extract feature data based on the training set; (c) Based on (a) and (b), the correlation between features and clinical development potential is obtained through statistical calculations, and features with high correlation are screened. (d) Based on the features obtained in (c), establish a computational method model; (e) Select another set of siRNAs that are already in clinical application or clinical development as a test set to verify the effectiveness of the above model, wherein the siRNAs in the test set are different from the siRNAs in the training set.

2. The method according to claim 1, wherein the method is a method for developing algorithms for enriching and screening siRNAs for gene silencing.

3. The method according to claim 1 or 2, wherein the training set or test set contains siRNAs with gene silencing effects targeting at least two different targets.

4. The method according to any one of claims 1-3, wherein the siRNA sequence features may include one or more of the following feature types: content of various nucleotides, positional preference of various nucleotides, siRNA / mRNA structure, and siRNA specificity for potential targets.

5. The method according to claim 4, wherein the siRNA sequence features are selected from: GC content (%), G content (%), U content (%), AU count (antisense strand 1-7), GC count (antisense strand 9-14), GC count (sense strand 1-5), antisense strand terminal nucleotide, sense strand terminal nucleotide, sense strand 3rd nucleotide, sense strand 10th nucleotide, sense strand 13th nucleotide, siRNA asymmetry, siRNA secondary structure, siRNA single nucleotide repeat, G repeatability, target mRNA region openness, target mRNA location, potential off-target effects, species homology, and sequence immunogenicity.

6. The method according to any one of claims 1-5, wherein the calculation method model may be: feature weighting, logistic regression, support vector machine, decision tree, random forest, Bayesian classification or neural network.

7. A method for enriching and screening siRNA, the method comprising: (a) Generate all siRNAs of a specified length by iterating through the target mRNA sequence; (b) Describe the sequence characteristics of the above siRNAs; (c) Obtain the siRNA score using a computational model based on sequence features; (d) Based on the siRNA score, siRNAs with potential gene silencing effects were screened.

8. The method of claim 7, wherein the method is a method for enriching and screening siRNAs for gene silencing.

9. The method according to claim 7 or 8, wherein the target is selected from TMPRSS6, AGER, MUC5B, DNAJC15 / MCJ, HMGCR, ANGPTL3, ALK, APOC3, TRPV1, KLKB1, complement component 3 (C3), ASGR1, DDIT4, Factor VIII, Factor IX, DPP4, STAT3, EPHA2, GPR146, APOA, MMP7, COL3A1, APP, Factor XII, ANGPTL8, MASP2, BCL2L12, SREBF2, CASP2, LDHA, PNPLA3, ALAS1, TLR7 / 8 / RIG-1, DUX4, HBV, ANGPTL4, DMD, CTGF Complement component 5 (C5), CD8, KIF11, ITGV6, DGAT2, COX-2, PRDM14, GPR75, KLK1, CD71, TP53, APOA1, Agtr1, TGFB1, CD4, PTGS2, MUC5AC, lipoprotein (a) (LPA), NPC1L1, ALDH2, AGT, KRAS, HSD17B13, DMPK, FactorX, SCAP, VEGF, TTR, CIDEB, GRB10 / 14, SERPINA1, INHBE, HAO1, KHK, MTARC1, complement factor B (CFB), SERPINAC1, CD40, PCSK9, XDH, SERPINAF2, FactorXI or COL1A1.

10. The method according to any one of claims 7-9, wherein the lengths of the two nucleotide chains of the specified length siRNA are 15 to 27 nucleotides each.

11. The method according to any one of claims 7-10, wherein the sequence features in step (b) include one or more of the following feature types: content of various nucleotides, positional preference specificity of various nucleotides, siRNA / mRNA structure, and siRNA specificity to potential targets.

12. The method according to claim 11, wherein the sequence features are selected from: GC content (%), G content (%), U content (%), AU count (antisense strand 1-7), GC count (antisense strand 9-14), GC count (sense strand 1-5), antisense strand terminal nucleotide, sense strand terminal nucleotide, sense strand 3rd nucleotide, sense strand 10th nucleotide, sense strand 13th nucleotide, siRNA asymmetry, siRNA secondary structure, siRNA single nucleotide repeat, G repeatability, target mRNA region openness, target mRNA location, potential off-target effects, species homology, and sequence immunogenicity; wherein, Preferred features include: GC content (%), G content (%), U content (%), AU count (antisense strand 1-7), GC count (antisense strand 9-14), GC count (sense strand 1-5), antisense strand terminal nucleotide, sense strand terminal nucleotide, sense strand 10th nucleotide, sense strand 13th nucleotide, siRNA asymmetry, siRNA secondary structure, siRNA single nucleotide repeat, G repeatability, target mRNA region openness, target mRNA location, potential off-target effects, and species homology.

13. The method according to claim 12, wherein the sequence features are described according to the following rules: (1) The GC content is between 30% and 52%. If this condition is met, the characteristic value is 1; otherwise, it is 0. (2) If the U content is ≤40%, the characteristic value is 1; otherwise, it is 0. (3) AU count (antisense chain 1-7): The A+U count of the first to seventh positions of the antisense chain is ≥4. If this condition is met, the characteristic value is 1; otherwise, it is 0. (4) GC count (antisense chain 9-14): If the G+C count of the 9th to 14th positions of the antisense chain is ≤3, the characteristic value is 1; otherwise, it is 0. (5) GC count (sense chain 1-5): The G+C count of the first to fifth positions of the sense chain is ≥2. If this condition is met, the characteristic value is 1; otherwise, it is 0. (6) The antisense strand ends with a G or C nucleotide. If this condition is met, the characteristic value is 1; otherwise, it is 0. (7) The 10th nucleotide of the sense strand is either A or U. If this condition is met, the characteristic value is 1; otherwise, it is 0. (8) The nucleotide at the 13th position of the sense strand is not G. If this condition is met, the characteristic value is 1; otherwise, it is 0. (9) The sense chain ends with nucleotides A or U. If this condition is met, the characteristic value is 1; otherwise, it is 0. (10) siRNA asymmetry: asymmetry between the 3' and 5' ends of the sense strand, calculated by subtracting the A / U count of the first 3 positions of the sense strand from the A / U count of the first 3 positions of the sense strand at the 5' end; (11) siRNA secondary structure: siRNA does not have an internal secondary structure and does not have an inverted repeat sequence of length ≥10. If it meets these conditions, the characteristic value is 0; otherwise, it is 1. (12) siRNA single nucleotide repeat: There are no A / U single nucleotide repeats with a length > 8 and G / C single nucleotide repeats with a length > 4 in the siRNA sequence. If the condition is met, the characteristic value is 0; otherwise, it is 1. (13) G repetition: The number of times G with a length greater than 2 is <2. If the characteristic value is 0, it is 1 otherwise. (14) Openness of the target mRNA region: The mRNA region targeted by siRNA has good openness and is a single-stranded part that has not formed a secondary structure. If it meets the condition, the characteristic value is 1; otherwise, it is 0. (15) The target mRNA is located 50 nucleotides downstream of the transcription start site and within the coding region. If this condition is met, the characteristic value is 1; otherwise, it is 0. (16) Potential off-target: There are no potential off-target genes, that is, there are no genes other than the target gene that differ from the siRNA sequence by no more than three nucleotides. If this condition is met, the characteristic value is 1; otherwise, it is 0. (17) Species homology: The homology of the siRNA target region in the listed species, including cynomolgus monkeys, house mice, or brown rats. If the symptom is met, the eigenvalue is 1; otherwise, it is 0. (18) If the G content is ≤40%, the characteristic value is 1; otherwise, it is 0.

14. The method according to any one of claims 7-13, wherein the computational method model based on sequence features is Where f i For the sequence features described in step (b), w i These are the weight values ​​corresponding to the sequence features.

15. The method according to claim 14, wherein the weight value corresponding to the sequence feature is obtained by parameter tuning optimization based on the training set data.

16. The method according to any one of claims 7-15, the method further comprising performing a bioactivity test on the screened siRNA.

17. A siRNA with gene silencing effect, comprising a double-stranded region of 15 to 27 base pairs in length, wherein, The double-stranded region consists of two antisense strands, one antisense and one antisense strand, and the siRNA has at least one or more of the following characteristics: (1) The GC content is between 30-60%, preferably between 30-52%. (2) U content ≤ 40%, (3) The antisense chain has a large number of A and U in positions 1-7, preferably ≥4. (4) The antisense chain has fewer G and C in positions 9-14, preferably ≤3. (5) The first 1-5 positions of the sense chain contain a large number of G and C, preferably ≥2. (6) The antisense strand ends with a G or C nucleotide. (7) The 10th nucleotide of the sense strand is A or U. (8) The nucleotide at position 13 of the sense strand is not G. (9) The sense chain ends with an A or U nucleotide. (10) Asymmetry between the 3' and 5' ends of the sense chain is calculated by subtracting the A / U count of the first 3 bits of the sense chain from the A / U count of the first 3 bits of the sense chain at the 5' end. Preferably, the A / U count is ≥0. (11) siRNA does not have an internal secondary structure and does not contain inverted repeat sequences of length ≥10. (12) The siRNA sequence does not contain A / U single nucleotide repeats of length >8 or G / C single nucleotide repeats of length >4. (13) The number of G repetitions with a length greater than 2 is relatively small, preferably the number of consecutive G repetitions with a length greater than 2 is < 2. (14) The mRNA region targeted by siRNA has good openness and is a single-stranded portion that has not formed a secondary structure. (15) The target mRNA location is 50 nucleotides downstream of the transcription start site and within the coding region. (16) There are no potential off-target genes, that is, there are no other genes, besides the target gene, that differ from the siRNA sequence by no more than three nucleotides. (17) Homology of the siRNA target region in the listed species, including cynomolgus monkeys, house mice, or brown rats. (18) G content ≤ 40%.

18. A library of siRNAs with gene silencing effects, said library comprising a plurality of different siRNAs capable of producing gene silencing effects against at least one target gene, said siRNAs having at least one or more of the following characteristics: (1) The GC content is between 30-60%, preferably between 30-52%. (2) U content ≤ 40%, (3) The antisense chain has a large number of A and U in positions 1-7, preferably ≥4. (4) The antisense chain has fewer G and C in positions 9-14, preferably ≤3. (5) The GC counts of the first 1-5 bits of the meaningful chain are relatively high, preferably ≥2. (6) The antisense strand ends with a G or C nucleotide. (7) The 10th nucleotide of the sense strand is A or U. (8) The nucleotide at position 13 of the sense strand is not G. (9) The sense chain ends with an A or U nucleotide. (10) Asymmetry between the 3' and 5' ends of the sense chain is calculated by subtracting the A / U count of the first 3 bits of the sense chain from the A / U count of the first 3 bits of the sense chain at the 5' end. Preferably, the A / U count is ≥0. (11) siRNA does not have an internal secondary structure and does not contain inverted repeat sequences of length ≥10. (12) The siRNA sequence does not contain A / U single nucleotide repeats of length >8 or G / C single nucleotide repeats of length >4. (13) The number of G repetitions with a length greater than 2 is relatively small, preferably the number of consecutive G repetitions with a length greater than 2 is < 2. (14) The mRNA region targeted by siRNA has good openness and is a single-stranded portion that has not formed a secondary structure. (15) The target mRNA location is 50 nucleotides downstream of the transcription start site and within the coding region. (16) There are no potential off-target genes, that is, there are no other genes, besides the target gene, that differ from the siRNA sequence by no more than three nucleotides. (17) Homology of the siRNA target region in the listed species, including cynomolgus monkeys, house mice, or brown rats. (18) G content ≤ 40%.

Images