Dissecting the architecture of a quantitative trait locus in yeast.

1
Dissecting the architecture of a quantitative
trait locus in yeast.

• Steinmetz LM, Sinha H, Richards DR, Spiegelman
  JI, Oefner PJ, McCusker JH, Davis RW. Nature.
  2002 Mar 21416(6878)326-30.

BioNetworks October 13, 2003 Presentation by
Alison Hottes
2
Model Identification By Perturbation

• Reverse Genetics
• Genotype? Phenotype
• Alter a network link in known way and see what
  happens.
• E.g., delete a gene
• Forward Genetics
• Phenotype? Genotype
• Look for mutations or strains with a phenotype of
  interest and try to identify the underlying cause.

3
QTL (Quantitative Trait Locus)

• Many traits take on a continuum of values and are
  influenced by multiple genes (and the
  environment)
• Height, weight, intelligence, fish fin length
• Want to find chromosomal marker(s) whose presence
  is correlated with the trait.
• Screen many markers
• Identify important genomic regions.
• Many algorithms exist both for determining
  productive crosses and mapping the chromosomal
  regions.
• Identify which gene(s) in each region is
  responsible.

4
Htg- High temperature growth

• YJM145
• grows at 41C.
• S288c
• Poor growth at 41C.
• Diploid hybrid
• Grows better than either

• What is the genetic cause of the Htg phenotype?

5
Crosses
S96-Htg-

• Sporulated the hybrid
• Tested each segregant for Htg phenotype
• Selected 104 segregants that were at least as
  Htg as the original Htg strain

YJM789-Htg
6
Finding Markers

• Used Affymetrix arrays
• Identified 3444 markers
• Hybridized DNA from 19 Htg segregants to Affy
  chips to determine the parent of origin.

DNA from S96 (Htg- ) Array was designed from this
strain.
DNA from YJM789 (Htg ) Array was designed from
this strain.
7
Mapping

• Found 2 intervals that dont segregate randomly
  among the Htg progedy
• Mapped further using Denaturing High Performance
  Liquid Chromatography (DHPLC)
• Found 6kb region with 100 Htg association

8
Sequence Analysis

• Sequenced 32 kb in 5 more Htg and 6 more Htg-
  strains
• 83 variations between original strains
• 24 non-synonymous
• Data is ambiguous

9
Now what?

• Trait doesnt completely segregate with locus
• Suggests interactions of some kind
• No genes in locus had 3-fold expression
  difference between strains at 30C or 37C

10
Reciprocal-hemizygosity analysis

• Deleted each allele of all 15 genes in the hybrid
  background
• Works even for essential genes
• Accounts for interactions by putting alleles in
  background of both strains.
• Compared growth of reciprocal strains
• Could be extended to combinatorial analysis

11
Results

• 3 genes cause differences
• 2 from Htg strain
• 1 from Htg- strain
• Partially explains heterosis
• Confirmed that results are not a dosage effect

12
Sequence of Alleles

• Sequences dont suggest cause of Htg effects

13
Summary

• Went all the way from phenotype to 3 of the genes
  involved
• Tightly linked genes that contribute to the same
  phenotype can complicate mapping
• Suggests mapping to increasingly fine resolutions
  may be unproductive
• Interactions between alleles is complicated
• Used Affymetrix arrays to develop a set of
  markers

14
Thoughts for Discussion

• In some organisms, constructing the mutants for
  reciprocal hemizygosity analysis is prohibitively
  difficult
• What utility does QTL data have for modeling?
  What kind of modeling?
• Naturally perturbed networks can be harder to
  interpret than experimentally perturbed ones, but
  may give some insight into the complexity of the
  phenomenon under study.