Alternative Splicing in Plants

Nodes are exons from selected gene models (Delledonne 2009), (URGI 2010) and edges are the probes (Delledonne 2009) that cover them.

Description

Alternative splicing (AS) is a process by which the exons of the RNA produced by transcription of a gene are reconnected in multiple ways during RNA splicing. AS is estimated to occur in over 90% of all genes in humans, where it greatly increases the diversity of proteins that can be encoded by the genome. AS also occurs in 20-30% of plant genes, however the majority of AS variants use a different AS mechanism, intron retention. Studies have shown that splicing variants are present and may play key roles in abiotic stress conditions such as temperature and biotic stress such as pathogen infection.

Most current work for assessing AS in plants has been done using EST and cDNA collections which do not adequately represent conditions such as stress. This proposal seeks to exploit the large amount of existing expression data in the form of micro, exon, and tiling array data sets and use it to integrate with the avalanche of short-read sequence datasets becoming available to find and confirm exon skipping AS. The discipline of signal processing focuses on estimating signal parameters and separating signals from noise. We will apply model-based approaches such as generative mixture models to estimate the key parameters of the exon and intron expression in RNAseq data for determining when plants use alternative splicing and in what tissues. Existing models for human systems only take exon signal levels into account.

This proposal has three objectives: 1) Integrate plant expression platforms for AS detection, 2) develop novel model-based versions of alternative splicing detection and exon expression level estimation methods based on signal processing methods from communication systems, and 3) Combine splice rewiring events and traditional gene summarizations together in a biochemical pathway context.  As a meaningful proof of concept, we intend to apply our methods to two important but different species: Zea mays, an important and well-studied and highly domesticated crop and Vitis vinifera, a less domesticated plant that is highly reactive to environmental effects such as climate change (Kast et al. 2009; Duchene et al. 2010; Jorquera-Fontena et al. 2010; Kapur et al. 2010; Pugliese et al. 2010). In plants, AS can be influenced by abiotic stresses (i.e., temperature fluctuations) (Barbazuk et al. 2008, Egawa et al. 2006, Palusa et al. 2007, Palusa et al. 2010) and can play a key role in resistance genes (Zhang et al. 2007).

This work will focus on methods for combining the existing datasets of previous gene expression data from Affymetrix microarrays from our database for PLant and Pathogen EXpression data (PLEXdb) with the deluge of RNA-Seq data to discover new splice sites and develop methods for estimating the relative transcript abundances for each splice variant using detection methods from the area of signal processing. All of the data will be publicly available via the Gene Atlas page at PLEXdb.org as well as in an AS splice track at MaizeGDB and VitisCyc.

The specific objectives of the grant are:

Objective 1.   Expression Platform Integration. Analyze existing EST- and gene model-based chip designs for plant species to determine the probe alignments on the gene models to estimate the potential for exon expression using microarray data. Process existing data sets in PLEXdb to determine the occurrence of differential splicing. Integrate and compare these results to alternative splicing data developed from short-read, high-throughput experiments, focusing on data from the species Vitis vinifera and Zea mays.

Objective 2.   Estimate alternative splicing variants using a model-based approach. Develop novel model-based versions of alternative splicing detection and isoform expression level estimation methods based on signal processing methods. These methods will be combined with those in Objective 1 using gene atlas RNA-seq data provided by our collaborators at MaizeGDB and the University of Verona.

Objective 3.   Pathway Integration and Hypothesis Generation. Develop novel methods to integrate splicing and expression networks with biochemical pathway networks and partition the results into experiment-dependent sub-networks of interest.

Objective 4.   Education and Outreach. The above objectives will be integrated into training workshops and opportunities for students, postdocs, and summer interns. We will also provide focused workshops for the grape and maize communities through MaizeGDB and the Grape Research Coordination Network on the interaction between the isoforms and the metabolic pathways.

Project Personnel

• Julie A Dickerson (PI) Electrical and Computer Engineering
• John Van Hemert, Bioinformatics and Computational Biology.

Publications

Funding Sources: