Post-trancriptional Regulation by microRNA

1
Post-trancriptional Regulation by microRNAs

• Herbert Levine
• Center for Theoretical Biological Physics, UCSD
• with E. Levine, P. Mchale, and E. Ben Jacob
  (Tel-Aviv)
• Outline Introduction
• Basic model
• Spatial sharpening
• Temporal Sequencing

2
What are MicroRNAs?

• MicroRNAs (miRNAs) are small noncoding RNA
  molecules that regulate eukaryotic gene
  expression at the translation level

RISC RNA-induced Silencing Complex
3
MicroRNA formation
miRNAs are processed from several precursor
stages Mammalian genomes seem to have 100s of
miRNAs
4
This talk

• Basic molecular model
• Local vs global parameters
• Spatial sharpening
• Temporal control

5
Basic silencing model
Bare messenger RNA Bound miRNA-mRNA Processed
state
Second step reflects the fact that complex is not
just degraded directly, but is targeted to a
specialized location (a cytoplasmic P-body) to
stop translation Binding- local rates transport
- global rates
6
Basic silencing model II

• Simple to analyze this in steady-state
• Critical parameter q - how much miRna is degraded
  per degraded mRNA (in processed state)
• q0 miRNA is completely recycled (catalytic
  mode)
• qgt0 miRNA is partially degraded
  (stoichiometeric)
• qlt0 amplification (occurs for siRNA)

7
Results
Effective equations
with
Effective silencing requires that ?as gt Qam?,
where ??m?s/?. Sharp silencing threshold
8
Threshold Effect

• Cartoon vs Reality

• RyhB miRNA regulation of sodB
• Threshold-linear units, similar to some neuron
  models
• Also, fluctuations reduced in silenced state
• From E. Levine, T. Hwa lab

9
Local vs. Global parameters

• Data on silencing has been very controversial,
  with disagreements as to whether there is both
  mRNA and protein repression or only protein
  repression
• In our model, the repression ratio can be altered
  by cell state (global) variables such as the
  transport into and out of the processed state,
  and miRNA loss (q)

10
Local vs. Global parameters
Global control through the effective parameter
Gives different repression ratios for same system
of miRNA and target, different cellular context
11
Local vs. Global parameters
Complex interplay of local and global parameters

• Different protocols can give opposite answers if
  these are not carefully controlled
• Simple physics but complex biology

12
Spatial sharpening

• What happens if we have a miRNA expressed with
  the opposite spatial pattern from its target
  mRNA?
• Motivation Complementary expression patterns
• And, the miRNA might diffuse from cell to cell
• Motivation - intercellular transport of siRNA in
  plants
• Could this be an actively maintained front with
  qlt0?

Voinnet (2005)
D Kosman et al, Science (2006)
Iba4 vs Hoxb8 - Ronshaugen et al. Genes Dev.
2005
13
Conceptual idea
The model predicts that mobile microRNA (red)
fine-tune this pattern by establishing a sharp
interface in the target expression profile
(green).
Morphogens set up a poorly defined expression
domain, where mRNA levels (green) vary smoothly
across the length of the embryo.
14
Spatial model

• Note - eq has been rescaled using
• We will assume that the transcription profiles
  are 1d functions, decaying in opposite
  directions, and investigate what are the
  resultant mRNA and miRNA
• The relevant parameters are the annihilation
  rate k and the miRNA diffusion constant D
  (compared to the scale established by
  transcription)

15
Zero diffusion, large k
Crossing point at
16
Adding miRNA diffusion

• K10000
• Dark line is analytic calculation
• Interface is sharpened
• Crossing point is shifted to left

17
Effect of increasing rate k
In the large k and/or small D limit, there is a
sharp transition layer Diffusion of miRNA eats
into m profile, and m has a sharp drop
18
Analytic solution
No miRNA flux is allowed into the region xltxt The
zero flux Greens function is clearly The
miRNA profile is given by And the interface is
determined by setting miRNa 0 (no
fluctuations). Once this position is determined,
we still have
to the left
19
Comments

• Sharp stripes are also possible

20
Comments

• Can be tested with genetic mosaics

21
Stability Analysis

• Can extend analysis to time-dependent case
• Now, miRNA equation becomes
• Linearizing around steady-state gives simple
  result

implies
22
Response to 2d quenched noise
Analytically Low-pass filter due to diffusion
23
C. Elegans development
Lin4 and Let7 miRNAs control differentiation As
usual, they act by silencing targets Is there
any good reason why miRNAs are used for this
task?
24
miRNA as temporal regulator

• Lin-28 needed for start of L2 phase needs to be
  turned off later than Lin-14
• Basic idea - one miRNA target has 5 binding sites
  (lin-14) and one has only 1 (lin-28)
• If miRNA act stoichiometrically, first target
  will soak up all the miRNAs and the second one
  will not be repressed until later

25
The complete circuit

• Direct positive feedback
• Indirect positive feedback
• Double-negative feedback
• miRNA switches g5 into off state and this then
  makes g1 also switch to off state
• This works better in stoichiometric mode, as g1
  is not repressed until g5 stops absorbing s

Experimentally, lin-14 inhibits an inhibitor of
lin-28 which is independent of lin-4 and vice
versa
26
Positive feedback
Stoichiometric mode
Catalytic mode
Thin lines - simple miRNA repression Thick lines
- with bistable behavior due to feedback Dashed
lines - reduced feedback note temporal ordering
27
The complete circuit

• Direct positive feedback
• Indirect positive feedback
• Double-negative feedback
• miRNA switches g5 into off state and this then
  makes g1 also switch to off state
• This works better in stoichiometric mode, as g1
  is not repressed until g5 stops absorbing s

Experimentally, lin-14 inhibits an inhibitor of
lin-28 which is independent of lin-4 and vice
versa
28
Final results
Solid lines catalytic Dashed Stoichiometric
Precise temporal staging is made easier by miRNA
29
Summary

• microRNAs are yet another level of genetic
  regulation
• In nature, miRNAs seem to be able to regulate
  both spatial and temporal aspects of development
• We have argued that the stoichiometric mode of
  operation seems to be an enabling factor
• Is this easier to arrange and control (via cell
  state) than equivalent transcription circuits??
  Is it easier to target many genes
  simultaneously??