How do regulatory networks evolve?

1

How do regulatory networks evolve?

Module = group of genes co-regulated by the same regulatory system

* Evolution of individual gene targets Gain or loss of genes from a module

* Evolution of activating signals Change in responsiveness but not regulators

* Wholesale evolution of the entire moduleTranscription factor sites occur upstream of totally different genes, responding to totally different signals

2

How do regulatory networks evolve?

Short time-scales: gene target turnover (gain and loss)

Time

3

ChIP-chip of two cooperatively-acting TFs in 3 species(S. cerevisiae, S. mikatae, S. bayanus ~20 my diverged)

Tec1 Ste12 Ste12 Ste12

Pseudohyphal growthGenes

Mating genes(haploid cells only)

Science 2007

4

Scer Smik Sbay

Ste12

Tec1

380 167 250

348 185 126

21% bound in all 3 species

20% bound in all 3 species

Only ~20% of orthologous regions bound in all 3 species

5* non-S. cer but otherwise conserved binding: enriched for Mating Genes

6

Only 20% of bound fragments conserved over 20 my(75% of these have underlying binding sites conserved)

Tec1 Ste12

Borneman et al. Science 2007

Substantial ‘rewiring’ of transcriptional circuits:

* Gain and loss of individual gene targets

* S.cer evolution of the module of mating genes

How common will these trends be? Different trends for different functional processes?

7

How common will these trends be? Different trends for different functional processes?

Science April 2010

PLoS Biol. April 2010

ChIP-seq (NFB and RNA-Pol II) and RNA-seq in 10 humans from 3 different populations

Lots of variation (up to 25% variation in binding levels)

ChIP-chip’d Ste12 in 43 S. cerevisiae segregants

Nature. March 2010

ChIP-chip of 6 developmental TFs in D. mel vs. D. yakuba (5 my)* only 1-5% of genes are variable targets (gene target turnover)

* lots of evidence of TF binding site turnover within CONSERVED target regions

Cell. March 2013

40% mouse TF binding sites conserved over < 6my

10-22% of TF binding is conserved in mammals(diverged ~80 my)

Science. 2010

8

How do regulatory networks evolve?

Short time-scales: gene target turnover (gain and loss)

Time Time

Evolved Responsiveness

Cooption of existing network

9

Ancestral GAL control likely by Cph1 … S. cerevisiae lineage picked up Gal4 and Mig1 sites upstream of GAL genes

10

In addition to changes in upstream cis-elements …

Major changes in the Gal4 transcription factor & upstream along S. cerevisiae lineage:

* Gained a domain that interacts with the Galactose-responsive Gal80 protein

* Other changes in the upstream response (Gal1-Gal3 duplication) contributedto sensitized pathway

11

How do regulatory networks evolve?

Short time-scales: gene target turnover (gain and loss)

Time Time

Evolved Responsiveness

12

How do regulatory networks evolve?

Sub/neo-functionalization through TF duplication & divergence

Time Time

TF duplicationEvolved TF sensitivity, binding specificity,

and ultimately targets

Gene targets can also duplicate (especially in WGD)

13

Example: Arg80 and Mcm1 duplicationTuch et al. 2008. PLoS Biol.

Mcm1 is a co-factor that works with many different site-specific TFs

Tuch. et al. performed ChIP-chip on Mcm1 orthologs in multiple fungi.

* Found dramatic differences in inferred Mcm1-TF interactions and modules

One case in particular: Arginine biosynthesis genes

Mcm1 + Arg81 at arg genes is ancestral

Duplication of Mcm1(Arg80) at WGD

Loss of Mcm1 binding at arg genesPresumably taken over by Arg80

Time (>150 my)

14

How do regulatory networks evolve?

If co-regulation is so important, then how can tolerate many independent changes in upstream cis-elements?

Conundrum:

Clearest cases of regulatory switches are often for highly co-regulated genes, whose co-regulation is high conserved.

15

Here they ChIP’d 6 TFs implicated inRP regulation in

S. cerevisiae and/or C. albicans

Ifh1-Fhl1 co-activators are conservedin Sc-Ca (>200 my)

Required co-factors have evolved:

Hmo1 and Rap1 required for Ifh1-Fhl1 bindingin S. cerevisiae

* Hmo1 is a ‘generalist’ in C. albicans

In C. albicans, Cbf1 (generalist) andTbf1 (specialist) are required for

Ifh1-Fhl1 binding

16

* They propose that ‘generalist’ factors can readily ‘specialize’ to regulate

a specific module

* Species-specific connection of regulons:selection for alternate co-regulation

Figure 6

18

* They propose that ‘generalist’ factors can readily ‘specialize’ to regulate

a specific module

* Species-specific connection of regulons:selection for alternate co-regulation

* Co-evolution of binding sites AND interactions of regulators

* They propose cycles of neutral accumulation of mutations (and binding

sites) followed by deleterious mutation that is rapidly ‘corrected’ to rebalance co-

regulation

19

They raise the question:

Is wholesale rewiring commonto all modules

Or

facilitated byvery strong pressures to keep

genes co-regulated?