Transcript Gen660_Lecture9B_GeneExpressionEvo_2014

What forces constrain/drive protein evolution?
Looking at all coding sequences across multiple genomes
can shed considerable light on
which forces contribute how much to the rates of protein evolution.
1
What features explain the variation in rates of protein evolution?
Insights from Genomics:
1.
Rate of mutation/recombination of the locus (more recombination =
more efficient selection = easier to select adaptive alleles)
2.
Number of constrained residues (‘functional density’)
3.
Protein fold (structure, stability, folding)
4.
Protein essentiality (i.e. essential proteins evolve slower)
…. explains very little of the variation
4.
Number of protein-protein interactions (‘connectivity)
Initially reported, but now largely refuted as a global constraint
5.
Pleiotropy (i.e. number of processes in which protein is involved)
… explains only 1% of variation in evo. rates
6.
And the # 1 best predictor is ……
Expression Level of the underlying transcript
explains 30 - 50% of the variation in protein evo. rates!
2
Assessed the %variation explained by:
Previous studies: linear and multiple regressions
* expression level
Here:
* dispensibility
They argue the inter-dependence of these
* protein abundance
features makes multiple regression inappropriate
* codon bias
* gene length
… use principal component analysis instead
* # protein-protein interactions
3
* centrality in protein-protein networks
Principal Component Analysis (PCA)
* Each item (e.g. gene, protein,
dog skull) can be plotted as
a point in PC space.
Takes complex (perhaps related) measurements for each item* and identifies
independent ‘components’ (= abstract summaries of the data points) that best distinguish your
items into subgroups. The first component (PC1) is the plane that explains the most of the
4
variance in your groups (i.e. is the best predictor of subgroups).
Gene expression/Codon Bias/Protein Abundance (*all related)
explain 43% of variation in Ka and 52% variation in Ks!
% var in Ka explained by 7 Principal Components
Same holds for Ka and Ks, but less so for Ka/Ks …
because selection is likely acting on BOTH Ka AND Ks
Their model: selection is acting on translation to minimize protein unfolding5
From Pal et al. Integrated View of Protein Evolution
6
Of course, phenotypes can also evolve through regulatory changes
Seminal paper by King & Wilson:
# of genes can’t be the only answer … must involve regulatory differences
7
Of course, phenotypes can also evolve through regulatory changes
i.e. When, Where, How much, and in what context a protein is present
RNAi
Affect translation rates,
RNA decay, RNA localization
RBP (some affect splice sites)
AAAAAAA
RBP
TF
TF
RNAP
ORF
RNAP
anti-sense RNA
8
Of course, phenotypes can also evolve through regulatory changes
i.e. When, Where, How much, and in what context is a protein is present
RNAi
Affect translation rates,
RNA decay, RNA localization
AAAAAAA
TF
TF
RNAP
ORF
RNAP
anti-sense RNA
Some effectors are encoded at the gene affected (local or cis effectors)
9
Of course, phenotypes can also evolve through regulatory changes
i.e. When, Where, How much, and in what context a protein is present
RNAi
Affect translation rates,
RNA decay, RNA localization
AAAAAAA
TF
TF
RNAP
ORF
RNAP
anti-sense RNA
Other effectors are encoded far from the gene affected (trans effectors)
10
The Coding vs. Noncoding Debate
Which type of change is ‘more important’ in evolution?
Are some genes/processes/functions more likely to evolve by one or the other?
What are the features that dictate coding vs. noncoding evolution?
A major advantage of non-coding regulatory changes:
Minimizing Pleiotropic Effects
Because cis-regulatory information is often modular.
11
EVE regulatory elements in D. melanogaster:
a model of modularity
12
From Developmental Biology, 6th Edition
The Coding vs. Noncoding Debate
Which type of change is ‘more important’ in evolution?
Are some genes/processes/functions more likely to evolve by one or the other?
What are the features that dictate coding vs. noncoding evolution?
What evolutionary forces act on gene expression regulation?
Before considering selection, it’s important to characterize
how gene expression varies within and between species.
13
Next-generation (‘deep’) sequencing can also be applied to
quantify mRNA (or other RNA) levels
Seq reads
cDNA
AAAAAAA
ORF
RNA
DNA
14
What facilitates regulatory evolution?
By now, many studies have looked at natural variation in transcript
abundance, simply to look qualitatively at which genes vary more/less.
Features that influence how variable a gene’s expression is across individuals:
* Gene dispensibility
Genes with variable expression within species are heavily enriched
for non-essential genes
* Genes with upstream TATA elements
TATA regulation in yeast (and other organisms?) is associated
with variable expression
* Redundancy
Either gene or regulatory redundancy
* Modularity in regulation
Genes with more upstream elements or greater environmental responsiveness
15
What facilitates regulatory evolution?
But some genes may not vary in expression because of constraint
(i.e. purifying selection)
while others may not vary in expression due to low rates of mutation/change
These cases can be distinguished by measuring the:
Mutational variance (Vm) = how much expression of a given gene varies
in response to mutation but in the ABSENCE of selection?
Genetic variance (Vg) = how much expression of a given gene varies
in natural populations (i.e. influenced by mutation + selection)
Vg/Vm = 1 means no constraint (expression variation in nature is the same
as in lab-derived ‘mutation lines’ … must be little selection in nature)
Vg/Vm <<1 means much less variation in natural population than mutation lines
… this must mean there has been purifying selection to reduce Vg
16
Generated ‘mutation accumulation’ lines in C. elegans
For each line:
- grew cells 280 generations
- each generation randomly picked 1 individual to generate next gen.
Measured whole-genome expression differences in each MA line
- calculated Vm
Measured whole-genome expression differences in each of 5 natural isolates
- calculated Vm
All genes had Vg/Vm < 1 … pervasive purifying selection on expression
Genes with the lowest Vg/Vm: enriched for signaling proteins and TFs
Genes with the highest Vg/Vm: enriched for carbon and amino acid metabolism
17
Expression can vary by the single gene (due to cis polymorphisms)
or for modules of coregulated genes (due to trans-acting effects)
TF
upstream
ORFs
TF
TF
TF
18