15th International Conference on Arabidopsis Research - TAIR

T11-007
Computational comparison of eukaryotic SNF1 and
plant-specific SnRK1 protein kinase phosphorylation
motifs on the basis of mutual information (MI).
Jan Hummel(1), Nima Keshvari(1), Wolfram Weckwerth(1), Joachim Selbig(1)
1-Max-Planck-Institute of Molecular Plant Physiology, Am Mühlenberg 1, 14476 Golm, Germany
The SNF1 (sucrose non-fermenting-1) family of protein kinases consists of
the homologous yeast SNF1, animal AMPK and plant SnRK1 counterparts.
First identified in Fungi the gene for SNF1 encodes a protein kinase in S.
cerevisiae that is activated in response to low cellular glucose levels. The
animal homolog of SNF1 is the aforementioned AMP-activated protein kinase
(AMPK), while the plant homolog is SNF1-related protein kinase-1 (SnRK1).
Despite assuming approximately 1.5 billion years of evolutionary divergence
the three members of the SNF1 protein kinase family have a remarkable
homology in eukaryotes where they are involved in regulating key aspects of
cellular function including cell division, metabolism, and responses to external
signals [1]. Today the completed genomic sequences of manifold eukaryotic
organisms provide new potentials for comparative structural analysis of
diverse SNF1 phosphorylation consensus sequences. On the basis of several
suggested common consensus sequences [1, 2] we compared putative
phosphorylation sites in the predicted proteomes of organisms belonging
to different eukaryotic kingdoms. The systematic analysis was done by
adopting a statistical approach using mutual information (MI), a measure of
association to reveal species-specific characteristics of conserved sequence
motifs [3]. The derived MI-profiles indicate evolutionary differences of known
and putative substrate specificities in the SNF1 family of protein kinases. In
addition to the kingdom-based approach we also examine resemblances in
the SnRK1 motifs of different plants likewise clarifying evolutionary distances
by structural and functional distinction of putative phosphorylation motifs.
[1]Halford et al. 2004. JXB 394,35-42
[2]Lunn, MacRae 2003. COPB 6,208-214
[3]Weckwerth, Selbig 2003. BBRC 307,516-521
T11 Modeling the Virtual Plant / Bioinformatics
T11-008
DIAGNOSIS OF PLANT METABOLISM
Yves Gibon(1), Jan Hannemann(1), Oliver Bläsing(1), Joachim Selbig(1), Oliver
Thimm(1), Melanie Höhne(1), Mark Stitt(1)
1-Max-Planck-Institute of Molecular Plant Physiology, Am Mühlenberg 1, 14476 Golm, Germany
Diagnosis of plant metabolism is defined as the process of identifying the
physiological state by the analysis of a minimal number of robust markers,
e.g. metabolite levels and enzyme activities. Applications range from phenotyping
of mutants and natural diversity to evaluation of plant performance.
The establishment of a diagnostic platform necessitates the 3 following
approaches:
1. Phenotyping through a broad range of growth conditions. Various
genotypes are grown under controlled conditions where parameters like
temperature, light intensity, nutrient availability or time are varied. Targeted
analyses as well as profiling experiments are then performed. In addition we
started to collect existing data, from various species.
2. Development of high throughput assays: we have adapted or conceived
robust microplate-based assays for metabolites and enzyme activities.
3. Data mining: we build a database to record growth conditions
and all the parameters determined for every sample. A module allows the
selection of classes of growth conditions and/or genotypes. The classes are
then “separated” via a machine learning system (decision trees), providing
templates for diagnosis. The ImageAnnotator module will be used to visualise
the diagnostic classifications.
To illustrate this, we will present the diagnosis of carbon-starvation in the
Arabidopsis starchless mutant pgm (lacking the plastidial phosphoglucomutase).
15 th <strong>International</strong> <strong>Conference</strong> on Arabidopsis Research 2004 · Berlin