A Large Family of Conserved Noncoding Elements

1
A Large Family of Conserved Non-coding Elements

• Michael Kamal
• Broad Institute

2
Key Points

• 1. Discovery Some classes of ancient repetitive
  elements show significant sequence conservation
  at orthologous position.
• 2. MER121 is the best conserved AR class.
• 1000 copies in human genome
• Largest CNE family found so far.
• 4. Orthologous copies are similar, paralogous
  are not!
• 5. Clues to possible biological function

3
Motivation CNEs

• 5 of the human genome is under purifying
  selection
• 3 CNEs (conserved non-coding elements)
• Generally do not cluster based on sequence
  similarity
• Bejerano(2004) 700K CNEs
• 96 of CNEs are unique in the human genome
• 250 groups with more than 10 CNEs
• Largest group 800 CNEs, but each similar to
  only 2 other CNEs
• CNEs are better conserved than background
  Ancient Repeats

4
Ancient repetitive elements

• Ancient Repeats (ARs) transposon relics, predate
  mammalian radiation.
• 22 of human genome, 780 classes
• Mostly non-functional sequence neutral evolution
• UCSC multiple alignments human, dog, mouse, rat
  (HDMR)
• ARs often deleted
• ARs often mutated
• Probability AR orthologous base, present in all
  species
• is perfectly conserved 1/2

5
Large conserved words in ARs

• Look for gt50bp of perfect conservation across
  HDMR
• Expect none. Naively
• Find 116 instances across 22 classes.
• 26/115 lie within MER121 only 1/4000th of total
  AR bases
• Top 10 ranked by density conserved words/Mb
  sequence

6
What is MER121?

• medium frequency reiterated repeat 121
• RepeatMasker 900 copies in human, consensus
  412bp
• Median divergence 26, size 180bp
• Repbase possible nonautonomous DNA transposon
• Not clearly related to any other known repeat

7
What is MER121?

• medium frequency reiterated repeat 121
• RepeatMasker 900 copies in human, consensus
  412bp
• Median divergence 26, size 180bp
• Repbase possible nonautonomous DNA transposon
• Not clearly related to any other known repeat

8
MER121 copies are rarely deleted

• Compare to MER119
• Unremarkable, known DNA transposon
• 1168 copies in human
• 23 diverged from consensus
• Consensus size 586bp
• 4-way retention rate
• How often does a human instance align to gt50bp
  in HDMR?
• MER121 82
• MER119 28

9
MER121 aligns at high identity

• Within retained copies, how often does a human
  base align to bases in all other species?
• MER121 96
• MER119 73

10
MER121 aligns at high identity

• Within retained copies, how often does a human
  base align to bases in all other species?
• MER121 96
• MER119 73
• How often is a four-way aligned base perfectly
  conserved?
• MER121 72
• MER119 49

11
MER121 aligns at high identity

• Within retained copies, how often does a human
  base align to bases in all other species?
• MER121 96
• MER119 73
• How often is a four-way aligned base perfectly
  conserved?
• MER121 72
• MER119 49
• Exons 78 (coding)

12
Orthologous conservation for top 75 instances

• Multiple alignment of top 75 human retained
  instances of most similar to consensus
• Show how each human base is conserved in HDMR

MER121
MER119
13
MER121 in other species

• Opossum (Monodelphis domestica)
• 1000 copies.
• At least 600 have orthologous copies in human
• Chicken
• 2 small copies
• Map to same central 70bp within the consensus
• 1 appears to be orthologous to a human copy
• MER121 dispersal postdated split from chicken

14
Fine-scale conservation
Project HDMR 4-way and human-mono pairwise
aligned instances onto repeat consensus
Record how often an orthologous base present in
all aligned species maps to a given position
along the consensus
position 10
15
MER121 center is retained more than flanks

How often does an aligned column map to a given
position along the consensus?
16

MER121 center is not better conserved
Rate of perfect four-way and pairwise identity
along the consensus
Placement
17




MER121 is not a protein coding gene
Size of indels in human-dog aligned
regions Multiples of size 3 not favored for
MER121
18




Little evidence of transcription

• MER121 consensus aligns significantly to 1 cDNA
  in large public databases
• Riken Fantom3 Mammalian Gene Collection
  databases
• 150K sequences, 246 Mb
• Significant overlap with only 19 human and mouse
  ESTs in Genbank
• (Genbank 11 million ESTs, 5.53Gb )
• (Require E-value of at least 1e-03, gt50bp
  aligning)

19




Little evidence of RNA secondary structure

• Evofold predictions (Pedersen, et al)
• 51 cases of small overlap for 49K predictions
• Median overlap 24bp, max 49bp
• RNAz predictions (Washietl, et al, 2005)
• No cases of significant overlap within strict set
  
• 36K predictions (Pgt0.9)
• 2. 2 cases of large overlap within permissive
  set
• 91K predictions in permissive set (Pgt0.5)
• Overlap size (261bp, 320bp)

20




MER121 generally found in gene-poor regions

• Compare to 241 classes of ARs with 500-4000
  copies in the human genome
• Rank class by typical distance to an Ensembl Gene
  start
• MER121 8 farthest (138Kb)
• MER119 130 farthest(59Kb)
• Rank class by typical density of exons in 500Kb
  neighborhood
• MER121 20 lowest
• MER119 175 lowest close to genome median

21




Clusters of nearby MER121 elements
Expected distance between copies D
3.3Mb Cluster elements that lie within region
of size D/2
5 clusters with gt8 elements
22


Two large clusters


12 elements Inhibin beta-A gene
8 elements TBX3/TBX5 T-box transcription factors
23
Summary

• Some Ancient Repeats appear to be under purifying
  selection
• MER121 is the best conserved AR class, largest
  CNE family with 1000 copies
• Unusual conservation profile Orthologous copies
  are similar, paralogous are not!
• Not a protein coding gene, unlikely to be an RNA
  gene family
• Perhaps cis-regulatory element mobilized by a
  transposon around the time of the mammalian
  radiation

24
Next steps Experiments

• Is MER121 transcribed?
• Enhancer/Insulator assay.
• Chromatin modifications (methylation H3K9?).
• Do proteins bind?

25
Acknowledgments

• Co-authors
• Xiaohui Xie
• Eric S Lander
• Many colleagues