Author Archives: rnorman2013

Final Project: Rene Norman-MATLAB modeling of hnRNP A2 RNA granules

RNA granules are non-membrane bound organelles comprising of multivalent RNA and protein interactions. One subset of RNA granules contains the scaffolding protein Tumor Overexpressed Gene (TOG) with seven binding domains for the RNA binding protein hnRNP A2. hnRNP A2 then binds specifically to A2 Response Element (A2RE) containing RNAs or CGG repeat containing RNAs. Single molecule imaging studies reveal that on average a granule contains 20 TOG molecules, more than 100 hnRNP A2 molecules and 60 RNA molecules. We can therefore define a subunit of an RNA granule as a TOG molecule saturated with hnRNP A2 molecules and with 3 specific RNAs bound. The only interaction remaining is the non-specific interaction of RNA with hnRNP A2 sites. Therefore, we can use the Gillespie algorithm to stochastically model the assembly of RNA granules using probabilities based on the kinetic constants for non-specific RNA: A2 interaction. From this, we can gain insight into what regulates size and dynamics of RNA granules. Furthermore, by adding CGG expansions to the model which increases the number of available binding sites, we can gain insight into structural granule differences that may expedite CGG repeat disorders.


Project Idea: CGG RNA affects granule assembly, translation and calcium transients – Rene Norman [Smith]

RNA granules are non-membrane-bound organelles consisting of RNA and protein components. The base of the granule is a number of multivalent scaffolding proteins, which bind to an RNA binding protein. In one subset of RNA granules, the scaffolding protein is Tumor Overexpressed Gene (TOG), which has seven binding domains for the RNA binding protein hnRNP A2.  hnRNP A2 binds to an eleven nucleotide sequence in some RNAs called the A2 Response element (A2RE). The RNA can also bind to hnRNP A2 non-specifically.  Single molecule imaging shows that there are on average 20 TOG molecules, more than 100 hnRNP A2 molecules and 60 RNA molecules per granule. Therefore, I can define a subunit of an RNA granule as one TOG molecule saturated with hnRNP A2 molecules bound to three RNA molecules.  The only interaction left is the non-specific interaction between RNA and hnRNP A2. I propose using MATLAB to model connections (non-specific RNA interactions) between subunits in the granules as one process, from outside the granule as a second process and removing a connection as a third process. These processes will be modeled using the Gollespie algorithm. I will then track the size of the granule, the number of connections and which subunits are connected using an adjacency matrix.  Once the stochastic process becomes stationary, that is the distribution of the number of subunits over time no longer changes, I will study the dynamic behavior of the stochastic system. I plan to compare this to in vivo time lapse data of RNA fluorescence  and potentially compare this to bursts of translation in the same cell type.

A second goal for this project is to include a second type of RNA, CGG repeat RNA, which will simply change the number of available binding sites per subunit. I hope to determine if there should be variability in the dynamic behavior when CGG RNA is added with and without CGG expansions. Finally, I would like to see if this difference correlates with changes that we see in translation, which would warrant a second model, which I have started in Virtual Cell regarding translation and calcium transients, all of which are affected by CGG repeat expansions in granules in rat hippocampal neurons and human fibroblasts from patients with CGG repeat expansion disorders.


Rene Norman, University of Connecticut Health Center, Farmington, CT

Screen Shot 2013-07-24 at 6.08.01 AM