News |  Sitemap |  Contact
PDF Export  | 

Testing non-random spatial arrangements in the nucleus using image-based models


Gene expression is not only based on DNA sequence information, but is also tightly regulated by the deposition of molecular marks that target genes for activation or silencing. Furthermore, the position of a gene within the cell nucleus may influence its expression, as many non-membranous nuclear compartments exist that are enriched for particular transcription and processing factors. Namely, it has been suggested that all transcription occurs within foci of RNA polymerase, and therefore genes must associate with these foci to become active. Conversely, the proximity to DNA-dense, heterochromatic regions, such as the nuclear periphery and centromeric clusters, has been correlated with gene silencing in mammals. Finally, long-range interactions between genes within the same chromosome or across different chromosomes have been shown to occur and may have a role in gene regulation.

To detect non-random, potentially functional, associations between genes and nuclear landmarks, good controls that highlight the non-randomness of the event are critical. However, controls are often far from ideal or hard to find, and a simple computational model that would yield the basal, random level of association would be desirable.

I have developed different computational models on ImageJ that aim to generate randomly positioned genetic loci to help assess whether the experimental measurements are non-random. Most of these models use experimental nuclear shapes to define spatial boundaries, i.e., they use the nuclear counterstaining as a ‘canvas’ for the computer-generated loci. This makes the models cell-type-independent, which is particularly advantageous given the distinct nuclear morphologies landscapes of different tissues and cell lines. Staining of other nuclear structures can be used to impose restrictions in the positioning of the loci (e.g., to exclude loci from nucleoli). I also developed a different, image-independent model that is useful to test a set of minimal assumptions that explain non-random distances between two different genomic loci.

These models were used in different systems to test whether distances between gene loci and transcription factories, centromeric clusters or the nuclear periphery were non-random. The models also tested the non-randomness of distances between two different gene loci and whether preferential radial positions of specific loci within the nucleus could justify non-random distances between them.

Nucleus, transcription, centromere, random model

Miguel Branco

MRC Clinical Sciences Centre, London, UK / The Babraham Institute, Cambridge, UK


Short Biography   
I graduated in Biochemistry in 2003, in Portugal, during which I started my research career and published my first paper. I then went to London to do my PhD at the MRC Clinical Sciences Centre, with Dr. Ana Pombo. I investigated the role of the spatial organisation of the genome in nuclear function, particularly in transcription. This involved extensive microscopy and image analysis, and it was during this time that I started developing my own analysis tools using ImageJ, and also designing models using both ImageJ and Matlab. During my PhD I published three more first-author papers, two reviews and a book chapter. I'm now working at The Babraham Institute with Dr. Wolf Reik, on the epigenetics of lineage specification during mouse embryogenesis.

© Luxembourg Institute of Science and Technology | Legal Notice