Case Study 1
To evaluate the workability of the pipeline of analyzing small non-coding RNA from NGS, the public domain data is used. This data (GEO accession ID:
GSE19833) is downloaded from GEO (Gene Expression Omnibus) which is a repository collecting array- and sequence-based high-throughput data. This data contains small RNA data from NGS and gene expression data from microarray in human normal and cancer cells [34]. The normal cell is peripheral blood mononuclear (PBMC) cell. The cancer cell lines are K562 and HL60.
For monitoring the expression of miRNAs from NGS, 760 expressed miRNAs are detected in normal and cancer cells. Then, 21 differentially expressed miRNAs are identified by selecting both up- or down-regulated in two cancer cell lines (fold change>=1.5 or <=0.67). Table 5 demonstrates these 21 differentially expressed miRNAs. Among them, only hsa-miR-25 is up-regulated and other 20 miRNAs are down-regulated. In addition to profiling the expression level of known miRNAs, 16 novel miRNA candidates are identified. Figure 11~18 show their RNA secondary structure, chromosome locations and cross-species conservation. Most of them are highly conserved in the stem (mature miRNAs).
In 16 novel miRNAs, five of them are opposite strand of known miRNAs in the stem. They are hsa-miR-382 (the second novel miRNA), hsa-miR-1185 (the fifth), hsa-miR-365 (the sixth), hsa-miR-642 (the tenth) and hsa-miR-539 (the fifteenth). Some of these five novel miRNAs are annotated in other species such as mouse, chimp, rat, cow and dog.
60
Table 5. The lists of differentially expressed miRNAs in two cancer cell lines Normalized miRNA expression Fold change
miRNA name PBMC (Normal) K562 HL60 K562/PBMC HL60/PBMC
hsa-miR-25 6525.68 10451.74 10090.54 1.60 1.55
Figure 11.The secondary structure and cross-species conservation of novel miRNA 1 and 2 (case study 1)
61
Figure 12. The secondary structure and cross-species conservation of novel miRNA 3 and 4 (case study 1)
Figure 13. The secondary structure and cross-species conservation of novel miRNA 5 and 6 (case study 1)
62
Figure 14. The secondary structure and cross-species conservation of novel miRNA 7 and 8 (case study 1)
Figure 15. The secondary structure and cross-species conservation of novel miRNA 9 and 10 (case study 1)
63
Figure 16. The secondary structure and cross-species conservation of novel miRNA 11 and 12 (case study 1)
Figure 17. The secondary structure and cross-species conservation of novel miRNA 13 and 14 (case study 1)
64
Figure 18. The secondary structure and cross-species conservation of novel miRNA 15 and 16 (case study 1)
In profiling gene expression from microarray, 3078 differentially expressed genes are identified (fold change >=2 or <=0.5). The criteria are the same with selecting differentially expressed miRNAs. These genes are all up or down-regulated in two cancer cell lines. Then, this gene profile is used to reduce false positive rate of predicting TFs which regulate miRNAs and miRNA target sites predictions.
Table 6 lists the predicted transcription factors which regulate miRNAs. The trend of fold change of these transcription factors are the same with miRNAs. For example, hsa-miR-146a is down-regulated in cancer cells. The TF which regulates hsa-miR-146a is also down-regulated in cancer cells. hsa-miR-25 is up-regulated in cancer cells. So, the TF which regulates it is also up-regulated. Among these predicted TFs, NR3C1 id most joint TF which regulates 13 down-regulated miRNAs (total 20 down-regulated miRNAs).
65
Table 6. The predicted lists of transcription factor which regulate miRNAs miRNA (fold change) TF lists (fold change) regulate miRNAs
hsa-miR-146a (0.43) RELA (0.44), MEIS1 (0.46), GATA3 (0.07), USF1 (0.35), REL (0.13), NFKB2 (0.11), JUN (0.19), AHR (0.01), ETS1 (0.01), NR3C1 (0.33) hsa-miR-21 (0.45) ZEB1 (0.23), TCF7 (0.07), RUNX3 (0.16), ETS1 (0.01), CEBPB (0.24),
NR3C1 (0.33)
hsa-miR-101 (0.44) RELA (0.44), GATA3 (0.07), EGR2 (0.30), ZEB1 (0.23), EGR3 (0.20), TCF7 (0.07), USF1 (0.35), HIF1A (0.12), AHR (0.01), TCF7L2 (0.16), SP4 (0.43), NR3C1 (0.33), STAT1 (0.33)
hsa-miR-361-3p (0.11) MAFB (0.10), TFEC (0.12), BCL6 (0.16), TCF7 (0.07), BACH2 (0.09), MAFF (0.18), ARID5B (0.04), ETS1 (0.01), CEBPB (0.24), BACH1 (0.35), MAF (0.49), NR3C1 (0.33)
hsa-miR-27a (0.35) GATA3 (0.07), SMAD3 (0.30)
hsa-miR-22 (0.27) STAT5A (0.14), EGR2 (0.30), GFI1 (0.36), EGR3 (0.20), KLF12 (0.02), ARID5B (0.04), STAT4 (0.01), PAX5 (0.29), SP4 (0.43), NR3C1 (0.33), STAT1 (0.33)
hsa-miR-25 (1.54) BRCA1 (22.29), WT1 (6.20), TFDP1 (2.10), NFYB (2.19), MYC (6.19) hsa-miR-192 (0.65) SMAD7 (0.18), BCL6 (0.16), JUND (0.17), FOSL2 (0.03), FOS (0.02),
FOSB (0.09), JUNB (0.08), ARID5B (0.04), JUN (0.19), PAX5 (0.29), ETS1 (0.01), RORA (0.01), SMAD3 (0.30)
hsa-miR-148a (0.61) EGR2 (0.30), EGR3 (0.20), TP53 (0.10), USF1 (0.35), KLF12 (0.02), HIF1A (0.12), ARID5B (0.04), PAX5 (0.29), MXD1 (0.16), SP4 (0.43), CEBPB (0.24)
hsa-miR-320d (0.37) STAT5A (0.14), IRF4 (0.30), BACH2 (0.09), MAFF (0.18), TCF7L2 (0.16), SP4 (0.43), ETS1 (0.01), CEBPB (0.24), BACH1 (0.35), MAF (0.49), IRF1 (0.04), SMAD3 (0.30), MAFB (0.10), ZEB1 (0.23), IRF9 (0.15), TCF7 (0.07), USF1 (0.35), STAT4 (0.01), PAX5 (0.29), NR3C1 (0.33), STAT1 (0.33)
hsa-miR-26b (0.61) STAT5A (0.14), RELA (0.44), EGR3 (0.20), REL (0.13), AHR (0.01), ETS1 (0.01), EGR2 (0.30), STAT4 (0.01), PAX5 (0.29), STAT1 (0.33) hsa-miR-27b (0.41) GCM1 (0.34), ARID5B (0.04), AHR (0.01), ETS1 (0.01), CEBPB (0.24),
SMAD3 (0.30), ZEB1 (0.23), JUN (0.19), NR3C1 (0.33)
hsa-miR-181d (0.50) EGR3 (0.20), SP4 (0.43), ETS1 (0.01), SMAD3 (0.30), EGR2 (0.30), USF1 (0.35), HIF1A (0.12), NR3C1 (0.33)
hsa-miR-152 (0.32) EGR3 (0.20), ETS1 (0.01), EGR2 (0.30), FOS (0.02), JUN (0.19), PAX5 (0.29), NR3C1 (0.33)
hsa-miR-34c-5p (0.44) EGR3 (0.20), AHR (0.01), SP4 (0.43), ETS1 (0.01), EGR2 (0.30), PAX5 (0.29)
hsa-miR-106b (0.48) GFI1 (0.36), ARID5B (0.04), AHR (0.01), SP4 (0.43), ETS1 (0.01), CEBPB (0.24), USF1 (0.35), HIF1A (0.12), MXD1 (0.16), NR3C1 (0.33) hsa-miR-151-3p (0.61) STAT5A (0.14), EGR3 (0.20), ARID5B (0.04), AHR (0.01), SP4 (0.43),
SMAD3 (0.30), EGR2 (0.30), USF1 (0.35), HIF1A (0.12), STAT4 (0.01), PAX5 (0.29), MXD1 (0.16), STAT1 (0.33)
hsa-miR-24 (0.64) GATA3 (0.07), GCM1 (0.34), ARID5B (0.04), AHR (0.01), ETS1 (0.01), CEBPB (0.24), SMAD3 (0.30), ZEB1 (0.23), JUN (0.19), NR3C1 (0.33) hsa-miR-26a (0.42) STAT5A (0.14), EGR3 (0.20), TP53 (0.10), AHR (0.01), SP4 (0.43), ETS1
(0.01), EGR2 (0.30), HIF1A (0.12), STAT4 (0.01), PAX5 (0.29), NR3C1 (0.33), STAT1 (0.33)
hsa-miR-16 (0.48) STAT5A (0.14), GFI1 (0.36), EGR3 (0.20), KLF12 (0.02), AHR (0.01), SP4 (0.43), ETS1 (0.01), IRF1 (0.04), EGR2 (0.30), USF1 (0.35), HIF1A (0.12), STAT4 (0.01), PAX5 (0.29), NR3C1 (0.33), STAT1 (0.33) hsa-miR-1 (0.18) EGR3 (0.20), TP53 (0.10), KLF12 (0.02), AHR (0.01), SP4 (0.43), CEBPB
(0.24), IRF1 (0.04), SMAD3 (0.30), EGR2 (0.30), PAX5 (0.29), NR3C1 (0.33)
66
In miRNA target site predictions, the target sites of 21 differentially expressed miRNAs are identified by using miRanda and TargetScans to predict first. Then, gene expression profile is combined. If the trend of fold change between miRNAs and their target genes are the same, these genes are removed.
For example, if the miRNA is down-regulated, its target gene should be up-regulated. 2864 MTIs (miRNA-target interactions) are identified from 1088 genes based on this rule. These MTIs and target genes are used to do GO and pathway enrichment analysis.
1088 predicted miRNA targets are used to do GO enrichment analysis (p-value<0.0001) by using the web-based tool (DAVID Bioinformatics Resources: http://david.abcc.ncifcrf.gov/home.jsp ). Table 7 lists the analysis results of these genes. For example, 106 genes involved in DNA metabolic process, 133 genes involved in cell cycle, 57 genes involved in DNA replication, 53 genes involved in mitosis and 56 genes involved in DNA repair. These gene functions have high association with carcinogensis.
Table 7. The function of miRNA target genes by GO enrichment analysis
GO terms Number of involved genes p-value
GO:0006259~DNA metabolic process 106 1.03E-27
GO:0007049~cell cycle 133 6.75E-26
GO:0006260~DNA replication 57 8.12E-23
GO:0007067~mitosis 53 1.04E-16
GO:0000280~nuclear division 53 1.04E-16
GO:0006281~DNA repair 56 1.52E-13
GO:0051301~cell division 56 7.71E-13
GO:0006310~DNA recombination 25 4.41E-08
GO:0034660~ncRNA metabolic process 38 2.51E-07
GO:0006412~translation 44 9.94E-06
GO:0065003~macromolecular complex assembly 73 1.10E-05
GO:0022613~ribonucleoprotein complex biogenesis 29 1.37E-05
GO:0006886~intracellular protein transport 47 2.04E-05
GO:0032259~methylation 17 2.35E-05
GO:0007005~mitochondrion organization 24 2.58E-05
GO:0006396~RNA processing 61 4.14E-05
GO:0046907~intracellular transport 70 4.50E-05
67
2864 MTIs are applied for pathway enrichment analysis. Table 8 shows 17 pathways which are identified by using phyper function in R package (P-value
<=0.05). For example, there are 12 miRNA target genes involved in the pathway
“DNA replication”. The total number of gene in this pathway is 36. 27 miRNA target genes involve in the pathway “cell cycle”. There are total 173 genes in this pathway. Comparing the results between GO and pathway enrichment analysis, some biological processes are both identified such as DNA replication and Cell cycle. In general, the analysis result of pathway is more specific than GO enrichment analysis. For example, the pathway “Mismatch repair” is the subclass of DNA repair which is identified from GO analysis. Moreover, the pathway can offer more detailed relationship of gene and gene interaction. Figure 19 shows the miRNA target genes involve in cell cycle. The involved target genes are colored according to their fold change (red: up-regulated, green:
down-regulated). In each gene text, if there are multiple genes in the same text and some of these genes are miRNA targets, pink ball is marked (blue ball means only one miRNA target gene exist in gene text). For example, the gene text “MCM”
contains MCM family. Among them, MCM3, MCM4 and MCM6 are miRNA target genes. Therefore, pink ball is marked at the gene text “MCM”. MCM3 is put in the gene text. The fold changes and MTIs of MCM4 and MCM6 are put in the right side table of pathway figure. Another multiple miRNA genes in the same gene text are also marked such as YWHAG, ORC6L, CDK4 and CCNE1. Through figure 19, how these miRNA target genes lead to great influence in cell cycle can be understood more clearly.
68
Table 8. The pathway lists of miRNA target genes by pathway enrichment analysis
Pathway name P-value # of matched genes # of genes in pathway
DNA replication 1.66E-06 12 36
Cell cycle 1.92E-05 27 173
RNA polymerase 6.55E-05 9 29
Aminoacyl-tRNA biosynthesis 4.50E-04 10 44
RNA degradation 8.09E-04 12 64
Homologous recombination 1.95E-03 8 36
Valine, leucine and isoleucine biosynthesis 2.10E-03 5 15
Alanine, aspartate and glutamate metabolism 5.83E-03 7 34
Mismatch repair 1.03E-02 6 29
Non-homologous end-joining 1.42E-02 4 15
Synthesis and degradation of ketone bodies 1.77E-02 3 9
Steroid biosynthesis 1.92E-02 6 33
p53 signaling pathway 2.44E-02 10 75
One carbon pool by folate 2.53E-02 5 26
Butanoate metabolism 2.85E-02 6 36
Pyruvate metabolism 2.95E-02 7 46
Terpenoid backbone biosynthesis 3.27E-02 4 19
Figure 19. miRNA target genes involve in cell cycle
In case study 1, the full analysis pipeline and results are demonstrated by using public domain available data. The results contain miRNA expression profile, novel miRNAs discovery, TF list of miRNAs identification, miRNA target genes
69
prediction, GO enrichment analysis of miRNA target genes and pathway enrichment analysis of MTIs.
Case Study 2
The data for case study 2 are from my cooperator, Dr. John Shyy, who is from Biomedical Sciences, University of California Riverside. The small RNA NGS sequencing data are under normoxia and hypoxia in human vascular endothelial cells (HUVECs). Hypoxic stress is an important factor which activates various physiological or pathophysiological responses in whole kind of cells. After processing sequencing data, 35 known differentially expressed miRNAs (fold change >=2 or <=0.5) are identified under hypoxia (Table 9). Among them, hsa-let-7 family and hsa-miR-103/107 are up-regulated. In addition, four novel miRNA candidates are found (Figure 20~21). These four novel miRNAs are conserved in human, chimp, rhesus, mouse, rat, cow and dog. HIF1A, hyroxia inducible factor 1α, is a key transcription factor induced by oxygen deprivation. It regulates hundreds of proteins involved in angiogenesis, erythropiesis, cell cycle and metastasis. To investigate the relationship between HIF1A and differentially expressed miRNAs, the binding profile of HIF1A from TRANSFAC is used to scan the promoter region of miRNAs. Transcription factor binding sites of HIFIA are identified in the promoter region of hsa-let-7 family and miR-103/107 (Figure 22). These prediction results are validated by quantifying the miRNA expression by Taqman qRCR under infecting Ad-HIF1A for 72 hr (data not shown). After verifying the relationship between HIF1A and miRNAs, the target genes of these miRNAs are predicted. 169 and 148 target genes are predicted for hsa-let-7 family and miR-103/107, respectively. Among these target genes, six genes are both targeted by hsa-let-7 family and miR-103/107 such as RPS6KB2, TARBP2,
70
RGPD5, RGPD6, RGPD8 and EIF2C1 (Ago1). AGO1 is an important protein which anchors the miRNA-induced silencing complex (miRISC). The target sites in 3’- UTR of AGO1 are conserved in 15 species (Figure 23). Then, these two target sites are experimentally validated by Luciferase reporter assay (data not shown).
Table 9. The lists of differentially expressed miRNAs in hypoxia HUVECs Normalizated miRNA expression Fold change miRNA name normoxia hypoxia hypoxia/normoxia
hsa-miR-30a* 603.94 3977.29 6.59
71
Figure 20. The secondary structure and cross-species conservation of novel miRNA 1 and 2 (case study 2)
Figure 21. The secondary structure and cross-species conservation of novel miRNA 3 and 4 (case study 2)
72
Figure 22. HIFIA binding sites in the promoter regions of let-7 family and miR-103/107
Figure 23. The target sites of hsa-let-7 family and hsa-miR-103/107 in the 3’
UTR of Ago1. These target sites are conserved in 15 species (red part is seed region).
73
According to in silico bioinformatics analysis and in vitro validation results, the model “miRNA-mediated transcriptional de-suppression” is proposed (Figure 24). In this model, HIF1A is induced under hypoxia. Then, HIF1A induces the expression of hsa-let-7 family and miR-103/107. These miRNAs down-regulate the expression level of Ago1 by targeting the 3’-UTR of it. The quantities of miRISC formation are reduced due to the low expression level of Ago1. So, the translational repression by miRNA targeting is reduced. The genes which are targeted by miRNAs are up-regulated. To verify this model, VEGF (vascular endothelial growth factor), induced under hypoxia, is validated about
“transcriptional de-suppression” in virto and in vivo (data not shown).
Figure 24. The model of translational suppression under normoxia and de-suppression under hypoxia
74
Case Study 3
The small RNA NGS sequencing data are from my cooperator, Dr. Hailing Jin, who is from Department of Plant Pathology and Microbiology, University of California Riverside [173]. There are four immunoprecipitation sequencing data sets from Arabidopsis leaves. The selected proteins are Ago1 and Ago2. These four sets are AGO1-IP-mock, AGO1-IP-Pst (Pseudomonas syringae pv. tomato), AGO2-IP-mock and AGO2-IP-pst. Excluding these four sets, there are two sequencing data about mutant Ago2 (ago2-mock and ago2-pst). Argonaute (AGO) proteins are important protein family in plants. They are RNAi effectors that bind miRNAs or siRNAs and mediate gene silencing by targeting the mRNAs. In plants, there are various immune responses including pathogen-associated molecular pattern-triggered immunity (PTI) and bacterial effector-triggered immunity (ETI). In plants, AGO1 is primarily associated with miRNAs and regulates PTI by several stress-related miRNAs. In figure 25, the mRNA level of AGO2 is highly induced by virulent Pst (EV) and avirulent Pst (avrRpt2) at 6 and 14 hours post inoculation (hpi). Therefore, AGO2 is the RNA silencing effector in both ETI and PTI by bounding miRNAs or siRNAs. To investigate the function of AGO2 in antibacterial immune responses, AGO1- and AGO2-IP SBS sequencing data are generated. AGO1-IP libraries are used as control. ago2 mutant sequencing data by the same treatment are also constructed.
75
Figure 25. The mRNA expression level of AGO2 in mock, virulent Pst (EV) and avirulent Pst (avrRpt2)
After processing the sequencing data and profiling the expression level of miRNAs and various kind siRNAs, an interesting thing is found. The expression level of various miRNA and miRNA* are opposite in AGO1- and AGO2-IP sets. For example, ath-miR165a is abundant in AGO1-IP but is lowly expressed in AGO2-IP.
ath-miR165a* is abundant in AGO2-IP but is lowly expressed in AGO1-IP. In table 10, there are five miRNA and miRNA* pairs which have great different expression level between AGO1- and AGO2-IP sets. In previous studies, only one strand of miRNA is expressed in many experimental conditions and another strand of miRNA is usually considered as non-function.
76
Table 10. The lists of differentially expressed miRNAs (comparing AGO1- and AGO2-IP)
To investigate the function of the miRNA*, the target sites of these five miRNA* are predicted by the modified plant target prediction guideline (described in the “Materials and Methods” section). In table 11, there are 41 predicted genes which are targeted by these five miRNA* (16 target genes for miR165a*, 13 target genes for miR393b*, 6 target genes for miR396a*, 3 target genes for miR396b* and 3 target genes for miR472*). The function of miR393b has been reported that it contributes to PTI by silencing auxin receptors TIR1, AFB2, and AFB3. To test whether miR393b* also involves in PTI, AT5G50440 encodes MEMB12, an SDS-resistant soluble N-ethylmaleimide sensitive factor attachment protein receptor (SNARE) that is mainly localized in cis-Golgi cisternae. The relationship between MEMB12 and miR393* is validated by detecting the protein level of wild-type MEMB12 (MEMB12-wt) and the miR393*
target site mutated in MEMB12 (MEMB12-mu). In figure 26, the protein level of MEMB12 is reduced by miR393* targeting in MEMB12-wt but the protein level of MEMB12 is recovered by mutating miR393* target site in MEMB12-mu.
77
Table 11. The predicted target genes of selected miRNA*
miRNA name Locus ID Description Score
ath-miR165a* AT2G30490 C4H (CINNAMATE-4-HYDROXYLASE); trans-cinnamate 4-monooxygenase
4 ath-miR165a* AT1G66150 TMK1 (TRANSMEMBRANE KINASE 1); transmembrane
receptor protein serine/threonine kinase
4
ath-miR165a* AT1G74790 catalytic 4
ath-miR165a* AT2G02790 IQD29 (IQ-domain 29); calmodulin binding 4.5
ath-miR165a* AT5G06350 binding 4.5
ath-miR165a* AT5G52620 F-box family protein 5
ath-miR165a* AT3G17850 protein kinase, putative 5
ath-miR165a* AT3G19650 cyclin-related 5
ath-miR165a* AT4G31600 UDP-glucuronic acid/UDP-N-acetylgalactosamine transporter-related
5.5 ath-miR165a* AT3G28340 GATL10 (Galacturonosyltransferase-like 10); polygalacturonate
4-alpha-galacturonosyltransferase/ transferase, transferring hexosyl groups
5.5
ath-miR165a* AT2G42300 basic helix-loop-helix (bHLH) family protein 5.5
ath-miR165a* AT5G44400 FAD-binding domain-containing protein 5.5
ath-miR165a* AT5G61780 tudor domain-containing protein / nuclease family protein 5.5
ath-miR165a* AT5G01030 unknown protein 5.5
ath-miR165a* AT5G54690 GAUT12 (GALACTURONOSYLTRANSFERASE 12);
polygalacturonate 4-alpha-galacturonosyltransferase/ transferase, transferring glycosyl groups / transferase, transferring hexosyl groups
5.5
ath-miR165a* AT3G48580 xyloglucan:xyloglucosyl transferase, putative / xyloglucan endotransglycosylase, putative / endo-xyloglucan transferase, putative
5.5
ath-miR393b* AT3G63180 TKL (TIC-LIKE) 4.5
ath-miR393b* AT3G19420 ATPEN2 (ARABIDOPSIS THALIANA PTEN 2); phosphatase/
protein tyrosine phosphatase
4.5
ath-miR393b* AT1G61350 armadillo/beta-catenin repeat family protein 5
ath-miR393b* AT3G48350 cysteine proteinase, putative 5
ath-miR393b* AT4G19490 protein binding 5
ath-miR393b* AT4G09040 RNA recognition motif (RRM)-containing protein 5
ath-miR393b* AT2G39300 unknown protein 5
ath-miR393b* AT5G58660 oxidoreductase, 2OG-Fe(II) oxygenase family protein 5
ath-miR393b* AT2G19640 ASHR2 (ASH1-RELATED PROTEIN 2) 5.5
ath-miR393b* AT2G25140 CLPB4 (CASEIN LYTIC PROTEINASE B4); ATP binding / ATPase/ nucleoside-triphosphatase/ nucleotide binding / protein binding
5.5
ath-miR393b* AT5G05900 UDP-glucoronosyl/UDP-glucosyl transferase family protein 5.5
ath-miR393b* AT5G50440 MEMB12 (MEMBRIN 12); SNAP receptor 5.5
ath-miR393b* AT3G09530 ATEXO70H3 (exocyst subunit EXO70 family protein H3);
protein binding
5.5 ath-miR396a* AT2G19590 ACO1 (ACC OXIDASE 1); 1-aminocyclopropane-1-carboxylate
oxidase
4.5
ath-miR396a* AT5G15610 proteasome family protein 5
ath-miR396a* AT2G18710 SCY1 (SecY Homolog 1); P-P-bond-hydrolysis-driven protein transmembrane transporter
5
ath-miR396a* AT1G51340 MATE efflux family protein 5.5
ath-miR396a* AT5G14130 peroxidase, putative 5.5
ath-miR396a* AT3G18220 phosphatidic acid phosphatase family protein / PAP2 family protein
5.5 ath-miR396b* AT1G19720 pentatricopeptide (PPR) repeat-containing protein 4.5
ath-miR396b* AT1G09610 unknown protein 5
ath-miR396b* AT5G60610 F-box family protein 5.5
ath-miR472* AT4G13395 RTFL12 (ROTUNDIFOLIA LIKE 12) 5
ath-miR472* AT1G12290 disease resistance protein (CC-NBS-LRR class), putative 5
ath-miR472* AT2G28010 aspartyl protease family protein 5.5
78
Figure 26. The protein level of MEMB12 with miR393b* (MEMB12-wt and mu) The model of miRNA* and miRNA pair associating with different AGO proteins targeting different regulators for the same cellcular process (PTI) is proposed based on in silico bioinformatics and in virto experimental validated results. Figure 27 demonstrates this model. miR393b associated with AGO1 targets auxin receptors TIR1, AFB2, and AFB3 for regulating pattern-triggered immunity (PTI). miR393b* associated with AGO2 targets MEMB12. Reduced MEMB12 leads to increased exocytosis of antimicrobial protein PR1. Therefore, AGO2 regulates PTI by binding miR393b* and subsequently modulating exocytosis of antimicrobial molecules.
Excluding miR393b and miR393b* pair, there are other miRNA nad miRNA*
pairs identified in this study. AGO2 is induced by ETI and PTI. Therefore, the function of miRNA* associated with AGO2 may regulate a group of genes involved in various pathways which are correlated with plant immunity.
79
Figure 27. miRNA* and miRNA pair, each of which targets different regulators within the same cellular process – immunity through two distinct RNAi effectors.
80
Case Study 4
The small RNA sequencing data are from my cooperator, Dr. Lin Na-Sheng, who is from Institute of Plant and Microbial Biology, ACADEMIA SINICA. In plants, viruses can induce post-transcriptional gene silencing (PTGS) through virus-specific small interfering RNAs (vsiRNAs) targeting the mRNA of host.
Satellite RNAs (satRNAs) can also trigger PTGS to produce satRNA-derived siRNAs (satsiRNAs). In Dr. Lin’s previous studies, two BaMV (Bamboo mosaic virus)-associated Satellite RNAs (satBaMVs) are identified [174]. They are BSL6 and BSF4. The similarity between BSL6 and BSF4 is very high (93%). BSL6 can greatly reduce BaMV accumulation and attenuate BaMV-induced symptoms in N.
benthamiana and Chenopodium quinoa. However, BSF4 does not have this function. The key component of satBaMV determines this function is the apical hairpin stem loop (AHSL) located in the 5’-UTR of BSL6. To further investigate the relationship between BaMV and satBaMVs, total eleven sequencing sets from Arabidposis thaliana and Nicotiana benthamiana are generated [175]. There are 4 sets from inoculated leaves of A. thaliana (Mock, BaMV, BaMV+BSF4 and BaMV+BSF6) and 7 sets from inoculated (I) and systemic (S) leaves of N.
benthamiana (Mock, BaMV(I), BaMV(S), BaMV+BSF4(I), BaMV+BSF4(S), BaMV+BSL6(I) and BaMV+BSL6(S)).
After processing sequencing data, the amount of vsiRNAs and satsiRNAs are counted in each library. Table 12 lists the amount of vsiRNAs and satsiRNAs in N.
benthamiana. In inoculated and systemic leaves with BSL6, very few vsiRNAs and satsiRNAs are detected. It is corresponded to the results of previous study (low expression level of BaMV and BSL6 RNAs) [176]. In A. thaliana, the expression level of vsiRNAs and satsiRNAs is also very low (Table 13). The detected
81
expression level of vsiRNAs and satsiRNAs in BSF4 is higher than in BSL6. It is because BSF4 can not reduce BaMV accumulation (the higher expression level of BaMV and BSF4 RNAs). According the statistics of length of vsiRNAs and satsiRNAs, the major length is 21 nt in A. thaliana (Figure 28). The major length is 21 and 22 nt in N. benthamiana (Figure 29).
Table 12. The amount of small RNAs in 7 libraries in N. benthamiana
Mock BaMV (I) BaMV (S) BaMV Total 2,489,506 3,293,585 3,411,741 5,247,031 3,567,466 3,272,499 4,853,059
BaMV 123,901
Table 13. The amount of small RNAs in 4 libraries in A. thaliana
Mock BaMV BaMV+BSF4 BaMV+BSL6
Total 4,676,816 3,437,925 2,037,033 2,221,999 BaMV 23,714 (0.7%) 29,880 (1.5%) 4,555 (0.2%)
BSF4 1,281(0.1%)
BSF4 1,281(0.1%)