Transcript Recent retrotransposition events have not affected

Transposable elements: insertion pattern and
impact on gene expression evolution in hominids
Maria Warnefors, Vini Pereira, Adam Eyre-Walker
University of Sussex, Brighton, United Kingdom
SMBE, Lyon, 2010
Species-specific transposable elements (TEs)
7786 insertions
2933 insertions
~95% of these TEs are Alu, L1 or SVA elements
(Mills et al, Am J Hum Genet, 2006)
Images: Wikimedia Commons
TEs can influence gene expression via several
mechanisms
Me Me Me Me
TEs provide transcription start sites (TSSs) for 30% of human transcripts
(Faulkner et al, Nature Genetics, 2009)
Intronic L1 insertion reduces reporter gene expression in HeLa cells
(Han et al, Nature, 2004)
TE silencing downregulates neighbouring genes in Arabidopsis
(Hollister and Gaut, Genome Research, 2009)
Have TEs increased expression divergence (ED)
between humans and chimpanzees?
Low ED
Have TEs increased expression divergence (ED)
between humans and chimpanzees?
Low ED
High ED
Previous studies of TEs and expression divergence
20% of mouse/rat ED is due to TEs
(Pereira et al, PLoS ONE, 2009)
Alu elements correlate with mouse/human ED, but the
direction depends on the definition of ED
(Urrutia et al, Genome Biology, 2008)
Materials and methods
Expression divergence was calculated for ~9000 genes based on
expression in five tissues (data from Khaitovich et al, Science, 2005)
Expression divergence is the Euclidean distance between expression
levels in one or more tissues
Species-specific TEs (Mills et al, Am J Hum Genet, 2006) were grouped
according to insertion site:
UPSTREAM
20 kb
10 kb
INTRONIC
2 kb
DOWNSTREAM
2 kb
10 kb
20 kb
Species-specific TEs are associated with an increase
in expression divergence
ED
TE
No TE
Combined p = 0.024
Z transformation method (Whitlock, J Evol Biol, 2005)
Do TEs cause the change in expression?
H. sapiens
P. troglodytes
M. mulatta
SPECIFIC
Contribute to ED
SHARED
Do not contribute
to ED
Images: Wikimedia Commons
Shared TEs are also associated with an increase in
expression divergence
SPECIFIC (p = 0.024)
ED
TE
No TE
ED
SHARED (p = 0.00003)
Recent TE insertions did not cause an increase in
expression divergence
Relative effect =
EDSPECIFIC / EDNO SPECIFIC
EDSHARED / EDNO SHARED
Second approach: Human-macaque versus chimpmacaque expression divergence
H. sapiens
Expression data:
~3700 genes in the prefrontal cortex
(Somel et al, PNAS, 2009)
P. troglodytes
M. mulatta
Images: Wikimedia Commons
Species-specific TEs have not caused increased ED
p = 0.32
H/M
C/M
p = 0.13
Do TEs act to increase transcript diversity?
New TE
No new TE
p value (M-W)
Human
2.7
2.3
2 x 10-16
Chimp
2.0
1.8
5 x 10-10
Do TEs act to increase transcript diversity?
New TE
No new TE
p value (M-W)
Human
2.7
2.3
2 x 10-16
Chimp
2.0
1.8
5 x 10-10
Expected transcript div.
Transcripts per gene
Transcripts per gene
No evidence that recently inserted TEs have
contributed to hominid transcript diversity
TE insertions are more common upstream than downstream of genes
496 TEs
561 TEs
Summary of ~9000 genes (from the Khaitovich dataset)
The upstream bias is just significant (p = 0.049).
TE insertions are more common upstream than downstream of genes
496 TEs
561 TEs
Summary of ~9000 genes (from the Khaitovich dataset)
The upstream bias is just significant (p = 0.049).
TEs may accumulate upstream because they have a role in chromatin
regulation. (Huda et al, Gene, 2009)
TEs may preferentially insert upstream (as known for Drosophila P elements)
Hominid TEs are subject to insertion bias
TE upstream
TE downstream
Active in germ line
255
181
Inactive in germ line
306
315
Expression state from eGenetics data (Kelso et al, Genome Research, 2003)
Only germ line expressed genes exhibit upstream bias (p = 0.003,
c2 test)
The enrichment of TEs in upstream sequences can be explained
in terms of insertion bias and does not indicate a role for TEs in
gene regulation
Summary
New TEs are correlated with, but not causal to, an increase in
expression divergence between human and chimpanzee. (However,
TEs may still contribute substantially to within-species variation).
We do not find evidence for a TE-induced increase in transcript
diversity.
Excess of upstream insertions can be explained by neutral insertion
bias.
To what extent do regulatory regions evolve within TEs, simply
because TEs are so abundant?
Acknowledgements
Adam Eyre-Walker
Vini Pereira
John Brookfield