Transcript General

Identification of a Novel cis-Regulatory Element Involved in
the Heat Shock Response in Caenorhabditis elegans Using
Microarray Gene Expression and Computational Methods
Debraj Guha Thakurta, Lisanne Palomar, Bary D. Stormo, Pat Tedesco, Thomas E.
Johnson, Davis W. Walker, Gordon Lithgow, Stuart Kim, and Christopher D. Link
Presented by
Abel G. Gezahegne
ECS 289A
February 24, 2003
Overview

Monitor ~12,000 genes from C. elegans to determine genes
up-regulated on heat shock (HS).

Analyze the upstream regions of these genes using
computational DNA pattern recognition methods to identify
any cis-regulatory motifs.

Determine the significance of these motifs using statistical
methods.

Perform comparative sequence analysis to determine if any
cross-species conservations exist.
Microarray Experiment

Determine Gene expression patterns before and after HS using
DNA Microarray for 11,917 known and predicted C. elegans
genes.

Animals were harvested as young adults and then split in two
halves: HS population and control population.

5 independent HS experiments at 35OC: In two experiment
animals were harvested after 1 hr of HS. In three experiments
animals were heat shocked for 2 hrs and allowed to recover at
20OC for 2 hrs then harvested.
Software Tools

Consensus – a greedy algorithm that searches for a matrix with a
low probability of occurring by chance.

ANN-Spec – an algorithm based on Artificial Neural Network
and Gibbs sampling method to discover un-gapped
patterns in DNA sequences

GLASS – Graphical Language for Assembly of Secondary
Structures: a sequence alignment algorithm.

Patser –
given weight matrix identifies high scoring subsequences
and calculates p values.
Gene Identification


Identified 28 genes induced in at least four of the
five experiments and over-expressed by a factor
of two or more.
Because of noise in DNA Microarray considered
only genes up-regulated by an average factor of
four or more.
Gene Identification (cont.)




Used 500 bp upstream from transcription start site to select candidates for
promoter elements.
Two DNA motifs identified by Consensus and ANN-Spec.
HSE - TTCTAGAA, a well known DNA binding site for HS Transcription
Factors (HSF).
HSAS - GGGTGTC, un unknown motif that does not correspond to any
known TF binding site.
Mathematical Model

Probability of a protein binding to a site with a score s:
P(bound|s)  es

When multiple binding sites exist, probability of binding:
Pmseq = sites es

Geometric Mean of the pp-values:

< Pmseq > = [ Sseq sites es ] 1/N
Difference of the log geometric means of the pp-values:
DLGM = log < Pmseq >HS - < Pmseq >Rand
Statistical Significance




Use the DLGM to determine the cutoff scores using the 13 up-regulated genes
and 3000 random genes from the C. elegans genome.
DLGM = log < Pmseq >HS - log < Pmseq >Rand
At a low cutoff value there are substantial amount of low scoring sequences
thus DLGM is low.
At a high cutoff even the high scoring sequences are not being used thus
DLGM drops.
The cutoff score that maximizes DLGM is chosen as the appropriate cutoff
value.
Cross-Species Conservation

To study conservation of regulatory sites
across related species two orthologous
gene pairs were examined between C.
elegans and C. briggsae.

The pattern of HSE and HSAS sites on
the promoters indicate conservation
across closely related species.

Output from VISTA (VISualization
Tools for Alignment.
Cross-Species C. (cont)

The gene structure and distances
between the genes are similar in both
organisms.

The two genes share 450 nt in the
upstream DNA sequence.

Output from GLASS alignment
algorithm.
Mutant Promoter Construct

A single mutation of HSE or HSAS still results in a
significant expression level of GFP (green fluorescence
protein).

Mutation of all three or two sites of HSE’s or one HSE’s and
the HSAS results in dramatic reduction is expression level.
Remarks and Conclusion

Since Microarray data was conducted
for ~2/3 of the C. elegans genes, there
may exist other HS induced genes.

Through experiments and statistical
methods the novel cis-regulatory
element discovered has been shown to
play a significant role in heat shock
response.

This has also shown computational
methods can be a valuable tool in
discovery of novel regulatory elements.