Human%20microRNAs%20co-silence%20in%20well-separated%20groups%20and%20have%20different%20essentialities

1
Human microRNAs co-silence in well-separated
groups and have different essentialities
Gábor Boross, Katalin Orosz, Illés J. Farkas
Eotvos University and Hung. Acad. of Sci.,
Budapest, Hungary
Support Hungarian Sci. Res. Fund (OTKA)
Hung. National Office for Res. and Techn.
(NKTH)
Nov, 2008
2

• Short regulatory RNAs
• RNA silencing vs. transcription regulation

• Databases listing miRNA target pairs huge
  differences
• With growing sample size increasing similarity
• Baek et.al., Selbach et.al (2008) ? use
  TargetScan (and PicTar)
• Modules of co-silencing miRNAs are well
  separated
• Example Seed sequence variation within and
  between co-regulating modules of miRNAs
• Application Essentiality of human miRNAs
• Data filtering Co-regulation modules,
  essentialities more reliable than miRNA-target
  scores

• Well-separated modules, group-to-group control
• may explain low number of individually essential
  miRNAs (Miska et.al. 2008)
• Essentiality of miRNAs is a network property
  position within a module
• may explain large role of ubiquitously expressed
  miRNAs (Landgraf et.al. 2008)
• Computational prediction Group of 9 most vs. 6
  least essential miRNAs ? lethality

3
Short regulatory RNAs Pathways and targets
He et.al. 2004 Science
4
Control of protein concentrations by
transcription factor proteins and ncRNAs
Hobert et.al. 2008 Science
5
Control of protein concentrations by
transcription factor proteins and ncRNAs
6
Computationally predicted lists of miRNA-target
pairs differ strongly
Example from C.elegans Computationally predicted
relative strength of two well-known miRNA-target
interactions
miRBase PITA PicTar TargetScan
lin-4 - LIN-14 No No Yes 11.20 Yes
let-7 - LIN-41 Yes Yes Yes 4.18 No
7
With growing sample size computationally
predicted interaction lists become increasingly
similar
(1) No benchmark data set (2) Baek et.al.,
Selbach et.al. (2008) TargetScan PicTar can
most precisely predict messenger RNA and protein
level fold changes upon the insertion / removal
of selected miRNAs
use TargetScan (control PicTar)
8
Modules of co-silencing miRNAs are well separated
9
Seed sequence variation within and between
co-regulating modules of miRNAs
Seed match best indicator of interaction (Lewis
et.al, 2003)
10
Seed sequence variation within and between
co-regulating modules of miRNAs
Seed match best indicator of interaction (Lewis
et.al, 2003)
11
Seed sequence variation within and between
co-regulating modules of miRNAs
Seed match best indicator of interaction (Lewis
et.al, 2003)
12
Essentiality of human miRNAs A network property
with two networks co-regulation co-expression
13
Essentiality of human miRNAs A network property
with two networks co-regulation co-expression
let-7a is co-expressed with the miRNAs that it
co-regulates with
14
Essentiality of human miRNAs A network property
with two networks co-regulation co-expression
let-7a is co-expressed with the miRNAs that it
co-regulates with
let-7g is not co-expressed with the miRNAs that
it co-regulates with
15
Essentiality of human miRNAs A network property
with two networks co-regulation co-expression
let-7a is co-expressed with the miRNAs that it
co-regulates with
let-7g is not co-expressed with the miRNAs that
it co-regulates with
Knock-out let-7g vs. let-7a If let-7g is
knocked out, then its functions can be replaced
with much smaller precision ( bigger errors
) let-7g is more essential then let-7a Not
co-expressed with co-regulating partners
16
Co-regulation scores, modules and essentialities
extract high-confidence information from
miRNA-target scores
17
Discussion
Group-to-group control miRNA group each miRNA
silences almost the same set of target genes may
explain low number of individually essential
miRNAs in C.elegans (Miska et.al.
2008) Essentiality of miRNAs is a network
property position within a module differently
expressed from strongly co-regulating other
miRNAs may explain large role of ubiquitously
expressed miRNAs (Landgraf et.al. 2008) Data
filtering 6 (9) out of 10 most (least) essential
miRNAs with TargetScan vs. PicTar data are
identical Computational prediction
Experimentally testable hypotheses List of
individually most and least essential human miRNAs
18
Predicted (relative) essentialities of human
miRNAs
19
Human microRNAs co-silence in well-separated
groups and have different essentialities
Gábor Boross, Katalin Orosz, Illés J. Farkas
Eotvos University and Hung. Acad. of Sci.,
Budapest, Hungary
20
Human microRNAs co-silence in well-separated
groups and have different essentialities
Gábor Boross, Katalin Orosz, Illés J. Farkas
Eotvos University and Hung. Acad. of Sci.,
Budapest, Hungary
Support Hungarian Sci. Res. Fund (OTKA)
Hung. National Office for Res. and Techn.
(NKTH)
Nov, 2008