Title Page
Abstract
Contents
List of Abbreviations 16
1. Introduction 18
1.1. The function of microRNA in gene regulation 18
1.2. Metazoan microRNA biogenesis 19
1.3. Structural insights into microRNA biogenesis 24
1.3.1. Primary microRNA processing by DROSHA 24
1.3.2. Precursor microRNA processing by DICER 28
2. Sequence determinant of small RNA production by DICER 31
2.1. Background 31
2.2. Result 35
2.2.1. Massively parallel assay to identify a sequence motif 35
2.2.2. The GYM motif enhances dsRNA processing 45
2.2.3. The C-terminal dsRBD of DICER recognizes the motif 54
2.2.4. miRNA processing relies on the interaction between the GYM motif and dsRBD 57
2.2.5. Evolutionary implications of the GYM motif 67
2.2.6. The GYM motif improves the design of shRNA and DsiRNA 70
2.3. Discussion 72
2.4. Methods 74
3. Structure of the human DICER-pre-miRNA complex in a dicing state 83
3.1. Background 83
3.2. Results 84
3.2.1. Structural determination 84
3.2.2. Overall structure of hDICER in a dicing state 90
3.2.3. Stem recognition by dsRBD and RIIID 107
3.2.4. PAZ helix reorients to accommodate RNA in a dicing state 114
3.2.5. Architecture of the 5' and 3' end pockets 119
3.2.6. 5' end recognition is disrupted in cancer-associated DICER mutations 123
3.2.7. Identity of the 5' end base affects the 5' counting mechanism 129
3.3. Discussion 133
3.4. Methods 138
4. Conclusion 146
국문초록 148
Bibliography 150
Table 2.1. Oligonucleotides used in this study. 81
Table 2.2. Oligonucleotides used in this study. 82
Table 3.1. Cryo-EM data collection, refinement and validation statistics. 145
Figure 1.1. Pri-miRNA processing by DROSHA-DGCR8 complex. 20
Figure 1.2. Sequence motifs of pri-miRNAs that facilitate DROSHA processing. 21
Figure 1.3. Pre-miRNA processing by DICER. DICER recognizes the 2-nt 3' overhang to follow 5' and 3' counting rules. 23
Figure 1.4. Structure of human DROSHA (Kwon et al., 2016). Domain compositions of human DROSHA and DGCR8 are shown. Colored boxes indicate regions... 25
Figure 1.5. Structure of human DROSHA and DGCR8 in complex with pri-miRNA. Domain compositions of human DROSHA and... 27
Figure 1.6. Structure of human DICER in complex with pre-miRNA (Liu et al.,2018). Domain composition of human DICER is shown. 29
Figure 2.1. Illustration of the mechanism of cleavage site choice by DICER. 32
Figure 2.2. Cleavage site decision of pre-let-7a-1 and pre-miR-324. 33
Figure 2.3. In vitro processing of a pre-miR-324 variant by DICER. "No-bulge pre-miR-324" was used for this assay to avoid the influence of the bulge. A mismatch... 34
Figure 2.4. Schematic outline of the massively parallel assay. 35
Figure 2.5. Massively parallel assay results from the 1st screening. Structural (left) and sequence (middle and right) preference of DICER substrates at the... 36
Figure 2.6. A structural model of human DICER in a dicing state. The dsRNA was modeled into the cryo-EM structure of human DICER, based on the the crystal... 36
Figure 2.7. Pre-miRNAs used in the massively parallel assay. The 5-bp and 3-bp windows (-1-to-3, -1-to-1, 1-to-3 positions relative to the starting position... 37
Figure 2.8. SDS-PAGE and Size-exclusion chromatography of purified proteins. 38
Figure 2.9. Distribution of the cleavage scores measured from the 2nd screening. Top 5 motifs in each backbone are shown. Analyses were done using data from... 39
Figure 2.10. Distribution of read counts and cleavage scores of variants. 39
Figure 2.11. Correlation of cleavege scores of variants between different conditions of varying reaction time. 40
Figure 2.12. Distribution of the cleavage scores measured from the 2nd screening with 1-to-3 randomization. 41
Figure 2.13. Enrichment of sequences of the top 1% variants. Expected proportions were calculated by assuming that 4 bases have the equal contributions to... 41
Figure 2.14. Structural impact on cleavage scores. G-U pair was considered as mismatch only when it is in between mismatches. p, pair; m, mismatch. 42
Figure 2.15. Structural impact on cleavage scores of variants (positions ≠1-to-1) with 10% or 30% cleavage. G-U pair was considered as mismatch only when it is... 42
Figure 2.16. Structural impact on cleavage scores of variants (positions 1-to-3) with 10%, 20%, or 30% cleavage. G-U pair was considered as mismatch only... 43
Figure 2.17. Impact of the base combinations at position 1 on cleavage scores with 20% cleavage. Variants with base-pairs at all but position 1 were included in... 44
Figure 2.18. Impact of the base combinations at position 1 on cleavage scores with 10%, 20%, or 30% cleavage. Variants with base-pairs at all but position 1... 44
Figure 2.19. In vitro processing of pre-let-7a-1 variants by human DICER. Bars indicate mean ± SD (n = 3). **p 〈0.01, ***p 〈0.001 by two-sided Student's t... 46
Figure 2.20. ITime-course in vitro processing of pre-let-7a-1 variants by human DICER. †, nicked products at the 3p positions. 47
Figure 2.21. In vitro processing of pre-let-7a-1 variants by human DICER-TRBP complex. Substrates were radiolabeled at their 5' ends. †, nicked products at the... 48
Figure 2.22. DROSHA processing assay and miRNA abundance measurement of pre-miR-A1 variants in HEK293T cells. Left: Schematic outline of this experiment.... 49
Figure 2.23. In vitro processing of duplex variants by human DICER. Cleavage products and their corresponding cleavage sites are marked with arrowheads. 50
Figure 2.24. In vitro processing of duplex variants by human DICER. Cleavage products and their corresponding cleavage sites are marked with arrowheads. 51
Figure 2.25. In vitro processing of pre-let-7a-1 variants by fly Dcr-1. Substrates were radiolabeled at their 5' ends. †, nicked products at the 3p positions. 51
Figure 2.26. In vitro processing of duplex variants by fly Dcr-1. Cleavage products and their corresponding cleavage sites are marked with arrowheads. 52
Figure 2.27. In vitro processing of duplex variants with a terminal mismatch by human DICER. The duplex had a mismatch at its terminus (marked in orange)... 53
Figure 2.28. In vitro processing of duplex variants with a terminal base-pair by human DICER. The duplex had a base-pair at its terminus (marked in orange) so... 53
Figure 2.29. A structural model of human DICER in a dicing state. The dsRBD (marked in red) is in the vicinity of the GYM motif (marked in yellow). 54
Figure 2.30. In vitro processing of pre-let-7a-1 variants by DICER △dsRBD. Bars indicate mean ± SD (n = 3). ***p 〈0.001 by two-sided Student's t test... 55
Figure 2.31. In vitro processing of duplex variants by either DICER △dsRBD or R1855L mutant. The cleavage product and its corresponding cleavage site marked... 55
Figure 2.32. Amino acid sequence alignment of dsRBDs of metazoan DICERs. 56
Figure 2.33. In vitro processing of duplex variants by human DICER point mutants at the indicated position. Cleavage products and their corresponding... 56
Figure 2.34. In vitro processing of pre-let-7a-1 variants by either DICER R1855L or R1855A/E1859A (AA). Bars indicate mean ± SD (n = 2 or 3, respectively). *p... 58
Figure 2.35. In vitro processing of duplex variants by either DICER R1855L or R1855A/E1859A (AA). The cleavage product and its corresponding cleavage site... 59
Figure 2.36. Schematic outline of the rescue experiment. 59
Figure 2.37. Comparison of miRNA expressions and cleavage accuracy in either HCT116 or HEK293T. Spike-ins were used for normalization. RPM, reads per... 60
Figure 2.38. Comparison of miRNA abundance. Spike-ins were used for normalization. RPM, reads per million. 61
Figure 2.39. In vitro processing of variants of pre-miR-27b by either DICER WT or △dsRBD. Pre-miRNAs were radiolabeled at their 5' end. Reactions were... 61
Figure 2.40. In vitro processing of variants of pre-miR-21 by either DICER WT or △dsRBD. Pre-miRNAs were radiolabeled at their 5' end. Reactions were... 62
Figure 2.41. In vitro processing of variants of pre-let-7d/f-1/i by either DICER WT or △dsRBD. Reactions were performed with different time points as indicated.... 63
Figure 2.42. Alternative processing of pre-miR-324. 64
Figure 2.43. Comparison of cleavage accuracy. For a given miRNA, the most abundant 5'-isomiR was identified in the WT sample. Then the fold change of... 64
Figure 2.44. The usage of 5' ends of miRNAs in the DICER-null HCT116 cells rescued with indicated DICER. The annotation in miRBase release 21 was used... 65
Figure 2.45. GYM motif-mediated cleavage site decision of miR-34a and let-7e. The usage of 5' ends of miR-34a-3p or 3' ends of let-7e-5p and 5' ends of let-7e-3p... 66
Figure 2.46. Distribution of the GYM motif at the indicated positions in natural pre-miRNAs across species. 67
Figure 2.47. The association between the GYM score and alternative processing. miRNAs registered in miRGeneDB were included in this analysis. For a given... 68
Figure 2.48. The GYM scores at the position -1 of human pre-miRNAs. miRNAs registered in miRGeneDB (n = 383) were included in this analysis. The dashed... 69
Figure 2.49. The GYM scores at the position -1 of human pre-miRNAs. miRNAs registered in miRGeneDB (n = 383) were included in this analysis. The dashed... 69
Figure 2.50. The effect of GYM motif in RNA interference by shRNA. Design of shRNAs with optimal (GUm), suboptimal (CUp), and poor (CUp) GYM... 71
Figure 2.51. Points and bars indicate mean ± SD (n = 3, biological replicates). 71
Figure 2.52. A proposed model for the role of GYM motif in DICER processing. 73
Figure 3.1. The sequence of pre-let-7a-1GYM used for structural determination.[이미지참조] 85
Figure 3.2. SDS-PAGE and size-exclusion chromatography of purified proteins. 86
Figure 3.3. In vitro processing of pre-let-7a-1 with human DICER. 87
Figure 3.4. Overview of the image processing procedure for structural determination of apo-DICER. 88
Figure 3.5. Representative micrograph and 2D class averages of the apo-DICER. Scale bar represents 50 nm. 89
Figure 3.6. Gold standard Fourier Shell Correlation (FSC) and angular particle distribution heatmaps of the apo-DICER. 89
Figure 3.7. Consensus maps and local resolution analysis of apo-DICER. Each domain is indicated in different colors. 90
Figure 3.8. Domain organization of hDICER. Schematics for the apo state shows amino acid residues included in the model (solid lines) and not included in the... 91
Figure 3.9. 3D maps of individual domains of the apo-DICER. 92
Figure 3.10. Size-exclusion chromatography of the DICER-RNA complex. 93
Figure 3.11. SDS-PAGE and urea-PAGE of the DICER-RNA complex. 93
Figure 3.12. Overview of the image processing procedure for structural determination of DICER in a dicing state. 94
Figure 3.13. Representative micrograph and 2D class averages of the DICER-pre-let-7a-1GYM. Scale bar represents 50 nm.[이미지참조] 95
Figure 3.14. Gold standard Fourier Shell Correlation (FSC) and angular particle distribution heatmaps of the DICER-pre-let-7a-1GYM.[이미지참조] 95
Figure 3.15. Protein-RNA interactions in the dicing state at the domain level. Sequences of pre-let-7a-1GYM that are not included in the model are not shown.[이미지참조] 96
Figure 3.16. Cryo-EM map of hDICER in a dicing state (color, 3.0 Å) overlaying that of hDICER in an apo state (grey, 4.0 Å). 96
Figure 3.17. Overall structure of the hDICER with a pre-miRNA in a cleavage-competent state. Black arrowheads point to the DICER cleavage sites. 97
Figure 3.18. 3D maps of the pre-let-7a-1GYM.[이미지참조] 98
Figure 3.19. 3D maps of the individual domains of DICER-pre-let-7a-1GYM.[이미지참조] 98
Figure 3.20. Cryo-EM map of the catalytic site created by RIIIDa and RIIIDb. 99
Figure 3.21. B-factor and Q-score plots for active site residues in the hDICER-let-7a-1GYM complex structure. Q-scores for each residue were derived from MapQ...[이미지참조] 99
Figure 3.22. Overall structure of the hDICER with a pre-miRNA in a cleavage-competent state. Black arrowheads point to the DICER cleavage sites. 100
Figure 3.23. Close-up views of the catalytic sites in RIIIDa and RIIIDb domains and electrostatic potential surface model of the catalytic valley along the protein-... 101
Figure 3.24. Superposition of RIIID domains of hDICER and Aa RNase III (left) and active sites of hDICER and Aa RNase III (right). 101
Figure 3.25. Buried surface area of hDICER in a pre-dicing state and a dicing state. 102
Figure 3.26. RMSD of hDICER-pre-let-7a-1GYM (this study) compared to hDICER-TRBP-pre-let-7a-1mutant (PDB: 5ZAL). Residues not resolved in the...[이미지참조] 103
Figure 3.27. Interdomain interactions in apo-hDICER. 103
Figure 3.28. Steric clash between pre-let-7a-1GYM and apo-hDICER.[이미지참조] 104
Figure 3.29. Cryo-EM map of the hDICER-pre-miR-3121GYM complex in a dicing state.[이미지참조] 105
Figure 3.30. Selected 2D class averages and 3D maps showing heterogeneity in the helicase domain. White arrowhead indicates the location of the helicase... 106
Figure 3.31. Urea-PAGE of hDICER-pre-let-7a-1GYM complex incubated with or without MgCl₂ for 10 min at RT, visualized by SYBR gold staining.[이미지참조] 107
Figure 3.32. Selected 2D class averages and 3D maps showing heterogeneity of the helicase domain of hDICER-pre-let-7a-1GYM complex.[이미지참조] 108
Figure 3.33. Conformational change in dsRBD during the transition from an apo state to a dicing state. Black arrowheads indicate cleavage sites. 108
Figure 3.34. Conformational changes of the dsRBD in the apo, dicing and pre-dicing states (PDB: 5ZAL). 109
Figure 3.35. Comparison between the structures of ideal A-form dsRNA helix and pre-let-7a-1GYM.[이미지참조] 109
Figure 3.36. Protein-RNA interactions near the cleavage sites in the minor groove. 110
Figure 3.37. Superposition of the dsRBDs of hDICER and Arabidopsis DCL3 (AtDCL3) (PDB: 7VG2). 111
Figure 3.38. Protein-RNA interactions in the interface between RIIID domains and RNA backbon. Left: RIIIDa, Right: RIIIDb. 112
Figure 3.39. Surface charge of the dsRBD, with dsRNA-dsRBD interface in dicing and pre-dicing states (PDB: 5ZAL). 112
Figure 3.40. Non-sequence-specific interactions between dsRBD and the RNA phosphate backbone. 113
Figure 3.41. Cryo-EM map and model of the hDICER dsRBD with dsRNA. 113
Figure 3.42. Sequence-specific interactions between dsRBD and the C-C mismatch of the GYM motif. 114
Figure 3.43. Superposition of hDICER PAZ-platform domain in the cryo-EM structure and in the crystal structure (PDB: 4NHA, grey). 115
Figure 3.44. Conformational change of the PAZ helix between a dicing state and a pre-dicing state. 116
Figure 3.45. Changes in the position of the pre-miRNA in a dicing state and a pre-dicing state. 116
Figure 3.46. Electrostatic interactions between the positively charged PAZ helix and the negatively charged RNA phosphate backbone. 117
Figure 3.47. In vitro processing of pre-let-7a-1 with a 2-nt 3' overhang. Points and bars indicate mean ± SD (n = 2). Relative cleavage was calculated by... 118
Figure 3.48. Time-course in vitro processing of pre-let-7a-1 with a 2-nt 3' overhang. 119
Figure 3.49. In vitro processing of pre-let-7a-1 with a 1-nt 3' overhang or a 3-nt 3' overhang. Points and bars indicate mean ± SD (n = 2). Relative cleavage was... 120
Figure 3.50. Schematic outline of the rescue experiment and comparison of miRNA abundance. Spike-ins were used for normalization. RPM, reads per million. 121
Figure 3.51. The 5' and 3' ends of pre-miRNA anchored in the basic 5' and 3' pockets in the platform and PAZ domains, respectively. 121
Figure 3.52. Pre-miRNA end recognition by the 5' and 3' pocket via hydrogen bonding. 122
Figure 3.53. Cryo-EM map of the 5' pocket (left) and 3' pocket (right). 122
Figure 3.54. Superposition of hDICER-pre-let-7a-1GYM and Platform-PAZ-Connector Helix.[이미지참조] 123
Figure 3.55. Superposition of dsRNAs complexed with hDICER and AtDCL3 (PDB: 7VG2). 124
Figure 3.56. Superposition of hDICER in a dicing state and AtDCL3-pre-siRNA complex (PDB: 7VG3, magenta). 124
Figure 3.57. In vitro processing of duplex RNAs with either a 2-nt overhang and a varying sequence at the 5' end. The nucleotide in the 3p strand opposite to the... 125
Figure 3.58. In vitro processing of duplex RNAs with either a 3-nt overhang and a varying sequence at the 5' end. The nucleotide in the 3p strand opposite to the... 125
Figure 3.59. In vitro DICER processing of pre-miRNA-like duplex with a 2-nt or 3-nt 3' overhang. The base opposite to the varying sequence is A on the 3p strand. 126
Figure 3.60. The effect of cancer-associated 5' pocket mutations in cleavage accuracy. Changes in cleavage accuracy was estimated with the fold change of... 127
Figure 3.61. The effect of 5' pocket mutations in cleavage accuracy. Changes in cleavage accuracy was estimated with the fold change of the proportion of the... 128
Figure 3.62. DROSHA/DICER cleavage sites dictated by 5' ends of mature miRNAs. 128
Figure 3.63. Abundance change of endogenous miRNAs. Spike-ins were used for normalization. 129
Figure 3.64. Predicted structural impact of the 5' end base substitutions on the interaction with the 5' pocket. 130
Figure 3.65. The nucleotide composition of human pre-miRNAs. The proportion of each base at the indicated position relative to the 5' end of pre-miRNAs is... 131
Figure 3.66. Examples of altered processing sites observed in the rescue experiments. Note that the DICER cleavage sites can be inferred from the 5' end of 3p... 132
Figure 3.67. The 5'-terminal base identity of pre-miRNAs. Compare the miRNA groups whose major 5'-isomiRs were affected by the R821 mutation with those... 133
Figure 3.68. Model of the structural transition and substrate recognition of hDICER during the pre-miRNA processing cycle. In the apo state (this study,... 134
Figure 3.69. Comparison of the substrate RNA movement during DICER processing between two small RNA pathways. In the miRNA pathway, a hairpin-shaped... 136
Figure 3.70. A phylogenetic tree of Dicer homologs. The scale bar indicates the length for the indicated frequency of amino acid variation. 137