molecular level Hervé Seitz IGH du CNRS, Montpellier, France March 3, 2016
microrna target prediction
. microrna target prediction mirna: target: 2 7 5 N NNNNNNNNNNNNNN 3 NNNNNN the seed
microrna target prediction Computational programs for target prediction: look for seed matches in 3 UTRs, select the ones that were conserved in evolution.
microrna target prediction Computational programs for target prediction: look for seed matches in 3 UTRs, select the ones that were conserved in evolution. Such short matches are very frequent (60 % of human coding genes seem to be targeted: Friedman et al., 2009).
microrna target prediction Computational programs for target prediction: look for seed matches in 3 UTRs, select the ones that were conserved in evolution. Such short matches are very frequent (60 % of human coding genes seem to be targeted: Friedman et al., 2009). = mirnas are implicated in every physiological process in animals.
microrna target prediction Computational programs for target prediction: look for seed matches in 3 UTRs, select the ones that were conserved in evolution. Such short matches are very frequent (60 % of human coding genes seem to be targeted: Friedman et al., 2009). = mirnas are implicated in every physiological process in animals. mirna-mediated repression is very modest (usually < 2-fold): lower than well tolerated fluctuations in gene expression (e.g., haplosufficiency). Why have these sites been conserved if they are not functional?
Baek et al., 2008: quantification of mir-223-mediated repression in mouse neutrophils.
Blood collection Neutrophil isolation RNA extraction cdna labeling, array hybridization
Blood collection Neutrophil isolation RNA extraction cdna labeling, array hybridization Blood collection Pooled blood Split in 5 replicates Neutrophil isolation RNA extraction cdna labeling, array hybridization
Experimental details
For 150 out of 192: inter-individual fluctuations across 5 wild-type mice exceeds mirna-mediated regulation (p-value < 0.05).
USP9X mrna Mus musculus Homo sapiens Sus scrofa Bos taurus mir-134 site Sarcophilus harrisii Monodelphis domestica Gallus gallus Taeniopygia guttata Danio rerio
USP9X mrna Mus musculus Homo sapiens Sus scrofa Bos taurus Sarcophilus harrisii mir-134 site 5 3 GACAUAACCAGCAAUGAAGA-UUUU-AGUCUCA GGCAUGACACUG-AUUCAGG-AUUUCAGUCACA GGCAUGAC----AAUUCAGG-CGUUCAGUCACA GACAUGACCCUGAAUUCAGG-AGUUCAGUCACA AGCAUGACACUGAAUUCAGGAAUUUCAGUCACA Monodelphis domestica AGCAUGACACUGAAUUCAGGAAUUUCAGUCACA Gallus gallus GGCACUAUGCUGAGUUCUGGAACAACAGUCACA Taeniopygia guttata GGCACUAUGCUGAAUUCUGGAACAACAGUCACA Danio rerio GAUGUGACGCUGAAUUCUGAAGCCACAGUCACA : mir-134: 3 GGGGAGACCAGUUGGUCAGUGU 5 : conserved in 8 or 9 species out of 9 : conserved in 6 or 7 species out of 9 : conserved in less than 6 species out of 9
USP9X mrna mir-134 site mir-134 Mus musculus Homo sapiens Sus scrofa Bos taurus 5 3 GACAUAACCAGCAAUGAAGA-UUUU-AGUCUCA GGCAUGACACUG-AUUCAGG-AUUUCAGUCACA GGCAUGAC----AAUUCAGG-CGUUCAGUCACA GACAUGACCCUGAAUUCAGG-AGUUCAGUCACA Sarcophilus harrisii AGCAUGACACUGAAUUCAGGAAUUUCAGUCACA Monodelphis domestica AGCAUGACACUGAAUUCAGGAAUUUCAGUCACA Gallus gallus GGCACUAUGCUGAGUUCUGGAACAACAGUCACA Taeniopygia guttata GGCACUAUGCUGAAUUCUGGAACAACAGUCACA Danio rerio GAUGUGACGCUGAAUUCUGAAGCCACAGUCACA : mir-134: 3 GGGGAGACCAGUUGGUCAGUGU 5 : conserved in 8 or 9 species out of 9 : conserved in 6 or 7 species out of 9 : conserved in less than 6 species out of 9
Hominidae Catarrhini Boreoeutheria Euteleostomi
3 UTR seed matches: Proportion of over-conserved seed matches 1.0 0.8 0.6 0.4 0.2 0.0 (n=10 mirna families) (n=14 mirna families) (n=14 mirna families) Hominidae Catarrhini Boreoeutheria (n=10 mirna families) Euteleostomi Hominidae Catarrhini Boreoeutheria Euteleostomi Clade-specific mirna families Comparison to prediction programs Effect of tree architecture
: revisiting mirna target definition
: revisiting mirna target definition Every measurable change in gene expression does not translate into a, evolutionarily selectable phenotype.
: revisiting mirna target definition Every measurable change in gene expression does not translate into a, evolutionarily selectable phenotype. A central feature of biological systems: their robustness to external insults. Hard to reconcile with the extreme sensitivity required for fine-tuning (the butterfly effect has probably been counter-selected).
Acknowledgements Anna Sergeeva: Laura Martinez: Blaise Li: Natalia Pinzo n: Isabelle Busseau: Delphine Maze :
Acknowledgements Anna Sergeeva: Laura Martinez: Blaise Li: Natalia Pinzo n: Isabelle Busseau: Delphine Maze : Jessy Presumey and Florence Apparailly (INM, Montpellier)
Acknowledgements Anna Sergeeva: Laura Martinez: Blaise Li: Natalia Pinzo n: Isabelle Busseau: Delphine Maze : Jessy Presumey and Florence Apparailly (INM, Montpellier)
: Mechanism of target repression Results: biological pathways mir-223 experiment: experimental details Over-conserved sites in prediction programs Published evidence for genome-wide targeting Issues with published pseudo-targets Absolute RNA quantification results RNA-Seq statistics : Pseudo-targets for other regulators? Propagation of gene expression perturbation
mirna target repression (adapted from Huntzinger and Izaurralde, 2011)
Biological robustness enzyme 1 enzyme 2 enzyme 3 Substrate Product 1 Product 2 Product 3
Experimental details Return
Experimental details Return
Experimental details Return
Experimental details p: probability that the difference between two individual mice is smaller than repression Return
Over-conservation in prediction programs Proportion of over-conserved seed matches Proportion of over-conserved seed matches Predictions by microt 5.0: 1.0 0.8 0.6 0.4 0.2 0.0 Hominidae Catarrhini Boreoeutheria Euteleostomi Clade-specific mirna families Predictions by miranda (aug2010): 1.0 0.8 0.6 0.4 0.2 0.0 Hominidae Catarrhini Boreoeutheria Euteleostomi Clade-specific mirna families Proportion of over-conserved seed matches Proportion of over-conserved seed matches Predictions by PicTar2: 1.0 0.8 0.6 0.4 0.2 0.0 Hominidae Catarrhini Boreoeutheria Euteleostomi Clade-specific mirna families Predictions by TargetScan 7.0: 1.0 0.8 0.6 0.4 0.2 0.0 Hominidae Catarrhini Boreoeutheria Euteleostomi Clade-specific mirna families Return
Effect of tree architecture Proportion of seed matches conserved outside clade of interest 1.0 0.8 0.6 0.4 0.2 3 UTR seed matches: Non-seed hexamers Hominidae-specific seeds Catarrhini-specific seeds Boreoeutheria-specific seeds Euteleostomi-specific seeds 0.0 Hominidae Catarrhini Boreoeutheria Euteleostomi Clade of interest Return
Published evidence
Published evidence Targets for a given mirna often belong to the same biological pathways. mrna for gene 1 mirna mrna for gene 2 mrna for gene 3 mrna for gene 4
Published evidence Targets for a given mirna often belong to the same biological pathways. mrna for gene 1 mrna for gene 2 mrna for gene 3 mirna mrna for gene 4
Published evidence Expression domains for mirnas and their overlap at their boundaries. mirna mirna and mrna expression mrna sharp boundary of mrna activity domain spatial or temporal axis
Published evidence Expression domains for mirnas and their overlap at their boundaries. mrna mirna and mrna expression mirna sharp boundary of mirna activity domain spatial or temporal axis
Published evidence House-keeping genes are rarely predicted to be targeted. mirna avoidance mrna for tissue specific gene cell type 1 mrna for house keeping gene mirna mrna for tissue specific gene avoidance cell type 2 mrna for house keeping gene
Published evidence House-keeping genes are rarely predicted to be targeted. mirna avoidance mrna for tissue specific gene cell type 1 mrna for house keeping gene mirna mrna for tissue specific gene avoidance cell type 2 mrna for house keeping gene
Reported examples of pseudo-targets A proposed pseudo-target: PTENP1, that de-repressed PTEN (Poliseno et al., 2010).
Reported examples of pseudo-targets A proposed pseudo-target: PTENP1, that de-repressed PTEN (Poliseno et al., 2010). But PTENP1 mrna is 100 times less abundant than the PTEN mrna (Ebert and Sharp, 2010).
Reported examples of pseudo-targets PTEN mrna (NM_008960) nt: CDS sirnas against PTENP1 1 869 2080 8229
Reported examples of pseudo-targets Proposed by Poliseno et al (2010): sirnas More probably: sirnas PTENP1 mrna PTEN mrna mirnas PTEN mrna
Reported examples of pseudo-targets sirna #2 against PTENP1: PTEN mrna (nt 948-961): 5 3 UAAUAAUCAUCAUUCUGGC GUUAUUA----UAA-ACCU 3 5 sirna #2 against PTENP1: PTEN mrna (nt 8189-8208): 5 3 UAAUAAUC-AUCAUUCUGGC : :: UUUAUUACUUGGAAAAAUUA 3 5 sirna #2 against PTENP1: PTEN mrna (nt 1383-1401): 5 3 UAAUAAUCAUCAUUCUGGC : AUUAUUAUAUGUAUCGCGG 3 5 sirna #3 against PTENP1: PTEN mrna (nt 2192-2212): 5 3 UCCUAUA--UGAUCUCUGAUG :: AGGAUAUUGACGUUAGACUGU 3 5 sirna #2 against PTENP1: PTEN mrna (nt 3769-3782): 5 3 UAAUAAUCAUCAUUCUGGC AUUAUUACC-----GACCU 3 5
Reported examples of pseudo-targets sirnas against AFF1 sirnas against RBM9 PTEN mrna CDS (NM_008960) nt: 1 869 2080 8229 sirnas against CNOT6L PTEN mrna CDS (NM_008960) nt: 1 869 2080 8229 sirnas against DCBLD2 PTEN mrna CDS (NM_008960) nt: 1 869 2080 8229 sirnas against JARID2 PTEN mrna CDS (NM_008960) nt: 1 869 2080 8229 sirnas against TNRC6a PTEN mrna CDS (NM_008960) nt: 1 869 2080 8229 sirnas against TNRC6b PTEN mrna CDS (NM_008960) nt: 1 869 2080 8229 sirnas against ZEB2 PTEN mrna (NM_008960) CDS PTEN mrna (NM_008960) CDS nt: 1 869 2080 8229 nt: 1 869 2080 8229 sirnas against MBNL1 PTEN mrna (NM_008960) CDS nt: 1 869 2080 8229
mirna quantification fmol synthetic oligo 0.5 1 2 4 8 15 20 M days of differentiation 0 1 2 3 4 5 6 nt: 38 23 21 18 (quantification of mir-1 and mir-206)
mirna quantification mirna abundance during C2C12 differentiation mirna molecules per cell 0 5000 10000 15000 20000 25000 mir 1 + mir 206 mir 133 0 1 2 3 4 5 6 Day of differentiation
mrna quantification
mrna quantification Very deep sequencing: three time points (day 0, day 3, day 6) in triplicate; each replicate: between 267 and 333 million transcriptome-matching reads. Statistics
mrna quantification Very deep sequencing: three time points (day 0, day 3, day 6) in triplicate; each replicate: between 267 and 333 million transcriptome-matching reads. Statistics 27 synthetic spike-ins, for calibration.
mrna quantification Day 0, replicate 1 Read abundance (fpkm) 0 1000 2000 3000 4000 0 2000 4000 6000 8000 Molecules per cell
mrna quantification Day 0, replicate 1 Read abundance (fpkm) 0 1000 2000 3000 4000 mir-1/mir-206 sites mir-133 sites 0 2000 4000 6000 8000 Molecules per cell
mrna quantification Molecules per cell 0 20000 40000 60000 80000 mir-1 + mir-206 0 1 2 3 4 5 6 Day
mrna quantification Molecules per cell 0 20000 40000 60000 80000 mir-1 + mir-206 3.2-fold excess of targets 0 1 2 3 4 5 6 Day
mrna quantification Molecules per cell 0 10000 30000 50000 mir-133 0 1 2 3 4 5 6 Day
mrna quantification Molecules per cell 0 10000 30000 50000 mir-133 15.3-fold excess of targets 0 1 2 3 4 5 6 Day
Deep-sequencing statistics Number of reads per kb (median transcript, excluding spike-ins): Replicate 1 159.26 Day 0 Replicate 2 140.53 Replicate 3 150.03 Replicate 1 213.09 Day 3 Replicate 2 167.77 Replicate 3 198.48 Replicate 1 223.57 Day 6 Replicate 2 243.86 Replicate 3 220.23 Return
Pseudo-targets for other regulators? Transcription factors: experimentally identified binding sites are poorly conserved among vertebrates (Odom et al., 2007 and Schmidt et al., 2010).
Pseudo-targets for other regulators? Transcription factors: experimentally identified binding sites are poorly conserved among vertebrates (Odom et al., 2007 and Schmidt et al., 2010). RNA-binding proteins are poorly specific (thousands of experimentally validated targets for each analyzed protein: Hafner et al., 2010; Lebedeva et al., 2011 and Hafner et al., 2013).
Pseudo-targets for other regulators? Transcription factors: experimentally identified binding sites are poorly conserved among vertebrates (Odom et al., 2007 and Schmidt et al., 2010). RNA-binding proteins are poorly specific (thousands of experimentally validated targets for each analyzed protein: Hafner et al., 2010; Lebedeva et al., 2011 and Hafner et al., 2013). Real molecular events, which are neutral in evolutionary terms? Return
Propagation of gene expression perturbation + Gene Z1 mirna Gene Y1 Gene Z2 Gene Z3 Gene X Gene Y2 Gene Z4 Gene Z5 Gene Z6 Gene Z7 Gene Y3 Gene Z8 Gene Z9 Direct mirna target Indirect mirna targets
Propagation of gene expression perturbation log 2 (fold-change) (blue boxplot) 10 8 6 4 2 Zfy overexpression Sept12 Tsarg2 Pebp4 Miwi 1500 1000 500 Number of affected genes (red line) 0 0 Direct targets 13.5 dpp Indirect targets at 13.5 dpp Indirect targets at 15.5 dpp (in collaboration with H. Royo and J. Turner, MRC, London)