robustness: revisting the significance of mirna-mediated regulation

Transcription

1 : revisting the significance of mirna-mediated regulation Hervé Seitz IGH (CNRS), Montpellier, France October 13, 2012

2 microrna target identification

3 .. microrna target identification mir: target: N NNNNNNNNNNNNNN NNNNNN 3 the seed

4 Identification of mirna targets Computational programs for target prediction: look for seed matches in 3 UTRs, select the ones that were conserved in evolution.

5 Identification of mirna targets 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).

6 Identification of mirna targets 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.

7 Identification of mirna targets 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?

8

9 Most computationally predicted targets are not functionally targeted (not repressed enough).

10 Most computationally predicted targets are not functionally targeted (not repressed enough). Their phylogenetic conservation means that these binding sites have a function.

11 Most computationally predicted targets are not functionally targeted (not repressed enough). Their phylogenetic conservation means that these binding sites have a function. That function could be to repress the mirna by titrating it.

12 Most computationally predicted targets are not functionally targeted (not repressed enough). Their phylogenetic conservation means that these binding sites have a function. That function could be to repress the mirna by titrating it..

13 Most computationally predicted targets are not functionally targeted (not repressed enough). Their phylogenetic conservation means that these binding sites have a function. That function could be to repress the mirna by titrating it..

14 Most computationally predicted targets are not functionally targeted (not repressed enough). Their phylogenetic conservation means that these binding sites have a function. That function could be to repress the mirna by titrating it.. pseudo target (insensitive) real target (sensitive)

15 A new interpretation The molecular event is the same (interaction between an mrna and a mirna), but the interpretation of that interaction is different: Current theory: the mirna represses the mrna. New : the mrna represses the mirna (except for a few real targets).

16 Discriminative predictions According to the new : According to the current theory:

17 Discriminative predictions According to the new : Abundantly expressed pseudo-targets have a stronger mirna-modulating effect, so their interaction with the mirna should be better conserved According to the current theory:

18 Discriminative predictions According to the new : Abundantly expressed pseudo-targets have a stronger mirna-modulating effect, so their interaction with the mirna should be better conserved = mrna abundance should be positively correlated with mirna binding site conservation. According to the current theory:

19 Discriminative predictions According to the new : Abundantly expressed pseudo-targets have a stronger mirna-modulating effect, so their interaction with the mirna should be better conserved = mrna abundance should be positively correlated with mirna binding site conservation. According to the current theory: Each target is a real target, its interaction is more or less conserved depending on the functional importance of the regulation (it is not expected to correlate with gene expression).

20 mrna abundance and seed match conservation Does mrna abundance correlate positively with mirna binding site conservation?

21 mrna abundance and seed match conservation Does mrna abundance correlate positively with mirna binding site conservation? mirna binding site conservation: measured by TargetScan s P CT score. mrna abundance in published microarray experiments.

22 mrna abundance and seed match conservation Kendall's τ = (p-value = ) Conservation score (P CT ) of binding sites to mir Gene expression in hypothalamus

23 mrna abundance and seed match conservation Adjusted p value 1e 16 1e 12 1e 08 1e 04 1 hypothalamus Adjusted p value 1e 16 1e 12 1e 08 1e 04 1 kidney Kendall s τ Kendall s τ

24 mrna abundance and seed match conservation Adjusted p value 1e 16 1e 12 1e 08 1e 04 1 hypothalamus Adjusted p value 1e 16 1e 12 1e 08 1e 04 1 kidney Kendall s τ Kendall s τ +34 other mouse tissues, same results: +33 fly tissues, same results:

25 Discriminative predictions According to the new : According to the current theory:

26 Discriminative predictions According to the new : Pseudo-target expression levels can fluctuate between individuals in a natural population (phenotype is robust). According to the current theory:

27 Discriminative predictions According to the new : Pseudo-target expression levels can fluctuate between individuals in a natural population (phenotype is robust). According to the current theory: mirna targets are tightly regulated (inter-individual fluctuation should not exceed repression).

28 Natural variability vs. repression

29 Natural variability vs. repression Baek et al., 2008: quantification of mir-223-mediated repression in mouse neutrophils.

30 Natural variability vs. repression Baek et al., 2008: quantification of mir-223-mediated repression in mouse neutrophils. Blood collection Neutrophil isolation RNA extraction cdna labeling, array hybridization

31 Natural variability vs. repression Baek et al., 2008: quantification of mir-223-mediated repression in mouse neutrophils. Blood collection Neutrophil isolation Blood collection Pooled blood Split in 5 replicates Neutrophil isolation RNA extraction RNA extraction cdna labeling, array hybridization cdna labeling, array hybridization

32 Natural variability vs. repression

33 Natural variability vs. repression

34 Natural variability vs. repression

35 Natural variability vs. repression p: probability that the difference between two individual mice is smaller than repression

36 Natural variability vs. repression

37 Natural variability vs. repression For 168 predicted targets out of 189: inter-individual fluctuations across 5 wild-type mice exceeds mirna-mediated regulation (p-value < 0.05).

38 Natural variability vs. repression For 168 predicted targets out of 189: inter-individual fluctuations across 5 wild-type mice exceeds mirna-mediated regulation (p-value < 0.05).

39 : revisiting mirna target definition

40 : revisiting mirna target definition Every measurable change in gene expression does not translate into a macroscopic, evolutionarily selectable phenotype.

41 : revisiting gene regulation definition High-throughput identification of transcription factor or RNA-binding protein targets: thousands of genes are bound, many of them are not under selective pressure to keep these binding sites.

42 : revisiting gene regulation definition High-throughput identification of transcription factor or RNA-binding protein targets: thousands of genes are bound, many of them are not under selective pressure to keep these binding sites. Microscopic events which are neutral in evolutionary terms.

43 : revisiting gene regulation definition High-throughput identification of transcription factor or RNA-binding protein targets: thousands of genes are bound, many of them are not under selective pressure to keep these binding sites. Microscopic events which are neutral in evolutionary terms. Are there pseudo-targets for these other regulators?

44 Acknowledgements Anna Sergeeva Laura Martinez Natalia Pinzo n Restrepo

45 Acknowledgements Anna Sergeeva Laura Martinez Natalia Pinzo n Restrepo Jessy Presumey and Florence Apparailly (INM, Montpellier, France)

46 Acknowledgements Anna Sergeeva Laura Martinez Natalia Pinzo n Restrepo Jessy Presumey and Florence Apparailly (INM, Montpellier, France)

47 Robustness of pathways Conservative approximations in the assessment of abundance/conservation correlation Positive correlation between gene expression and conservation of mirna binding sites

48 Second paradox enzyme 1 enzyme 2 enzyme 3 Substrate Product 1 Product 2 Product 3

49 mrna abundance and seed match conservation Most predicted targets are expected to be pseudo-targets (conservative approximation: we will consider every predicted target). mirna binding site should correlate with that mrna s abundance in the cells where mirna titration is beneficial (conservative approximation: we will consider whole tissues and organs). Poorly abundant mrnas (without a real titration effect on the mirna) should not exhibit such correlation (conservative approximation: we will consider every mrna). Return

50 mrna abundance and seed match conservation Adjusted p value 1e 16 1e 12 1e 08 1e 04 1 hypothalamus Adjusted p value 1e 16 1e 12 1e 08 1e 04 1 spleen Adjusted p value 1e 16 1e 12 1e 08 1e 04 1 naive B cells Kendall s τ Kendall s τ Kendall s τ kidney ES cells ovary Adjusted p value 1e 16 1e 12 1e 08 1e 04 1 Adjusted p value 1e 16 1e 12 1e 08 1e 04 1 Adjusted p value 1e 16 1e 12 1e 08 1e Kendall s τ Kendall s τ Kendall s τ Return

51 mrna abundance and seed match conservation Adjusted p value 1e 16 1e 12 1e 08 1e 04 1 adult ovary Adjusted p value 1e 16 1e 12 1e 08 1e 04 1 adult eye Adjusted p value 1e 16 1e 12 1e 08 1e 04 1 adult heart Expression of deeply conserved/expression of poorly conserved Expression of deeply conserved/expression of poorly conserved Expression of deeply conserved/expression of poorly conserved adult hind gut adult salivary gland larval feeding trachea Adjusted p value 1e 16 1e 12 1e 08 1e 04 1 Adjusted p value 1e 16 1e 12 1e 08 1e 04 1 Adjusted p value 1e 16 1e 12 1e 08 1e Expression of deeply conserved/expression of poorly conserved Expression of deeply conserved/expression of poorly conserved Expression of deeply conserved/expression of poorly conserved Return