The Origin of Operons'

1
A Database for Comparative Analysis of Gene
Expression in Multiple Bacterial Species Eric
Alm1, Yue Wang2, Jizhong Zhou4 and Adam P.
Arkin1,2,3 1Physical Biosciences Division,
Lawrence Berkeley National Labs, 2Bioengineering
Department, University of California-Berkeley,
3Howard Hughes Medical Institute, 4Oak Ridge
National Laboratory
Abstract
Results
We searched the literature for microarray data
describing gene expression over a wide range of
conditions in over 20 bacterial species. We are
compiling these data in a common format in a
central repository at the Virtual Institute for
Microbial Stress and Survival (VIMSS). In
addition, we are integrating these data with our
Comparative Genomics Database which includes
hand-curated gene annotations as well our own
predictions of the operon and regulon structure
of over 125 fully sequenced genomes.
Searching for generic stress response genes
Comparative analysis of E.coli and S. oneidensis
expression data
We compared data from four different stress
treatments (heat shock, low pH, high pH, salt
shock) to identify genes consistently up or down
regulated in S. oneidensis.
There is little overall correlation in global
gene expression between these two species for any
given condition. However, looking at only those
genes that are significantly up or down regulated
in both species in response to a given treatment
highlights a core of stress-response genes highly
specific to a given treatment. Shown below are
the genes most over and under-expressed in
response to heat shock treatment in both E. coli
and S. oneidensis.
Elucidating transcriptional reglatory networks
using comparative genomics
Motivation
We compared our own predictions of operon and
regulon structure in E. coli to gene expression
data in 14 different conditions obtained from the
Blattner laboratory website (Ref. 2). Shown
(right) is the distribution of correlations
between the expression profiles of 3 classes of
gene pairs (1) genes predicted to be on the same
operon by our operon prediction tool (2)
randomly chosen gene pairs and (3) genes
predicted to be on the same regulon (but not
operon) based on the co-occurrence of those genes
in operons in multiple unrelated species.
Interestingly, genes which tend to occur on the
same operon in many species are also closely
coregulated in E. coli even if they occur on
different operons. These results suggest that
operon structure in distantly related bacteria
can be a good indicator of transcriptional
coregulation.
A wealth of comparative genomic data, less
complicated transcriptional control, smaller
intergenic regions in which to search for
regulatory motifs, and the coregulation of genes
in operons make prokaryotes an ideal target for
building systems-level descriptions of gene
expression. Building such large-scale models
will require rigorous testing and training
against experimental data such as that obtained
from microarray experiments. Although there is
already a large community of researchers
generating microarray data in prokaryotic
systems, there is currently no central data
repository as for model eukaryotic species. We
have obtained a significant fraction of the
published data available for over 20 species, and
plan to release our completed database freely as
a community resource.
Correlation
Overview of a global stress response in E. coli
Shown at left (above) is a view of E. coli
metabolism colored according to changes in gene
expression in response to heat shock. Nodes
represent metabolites, and edges are enzymes
catalyzing reactions among the metabolites.
Enzymes that are down regulatedin response to
heat shock are shown in blue, those upregulated
are shown in red and non-changers are shown in
black.
Comparative view of a global stress response
The global transcriptional response of metabolic
enzymes to low pH treatment (below left) is shown
for comparison to the heat shock treatment above.
Colors are as described above.
Acknowledgements
We would like to acknowledge our collaborators at
Oak Ridge National Laboratory for sharing their
expression data on Shewanella oneidensis Haichun
Gao, Adam Leaphart, Yongqing Liu, Jizhong Zhou,
Matthew Fields, Dorothea Thompson and others.
References

• Bacterial stress response, Storz G and
  Hengge-Aronis R, ASM press 2000
• 2. Blattner laboratory website http//www.genome.w
  isc.edu/functional/microarray.htm