BRCA1

• Introduction
• In the news
• In the primary literature
• News vs. primary literature
• Gene
• Protein
• Organism phenotypes
• Conclusions
• Further directions
• About this website

Search Tool for Retrieval of Interacting Genes/Proteins (STRING) 8.0

Searching BRCA1 in STRING yielded many results from a variety of organisms. I chose just to look at the human data, as between the different protein interaction databases and software, there was a wealth of information for humans alone, though a quick look at the very distantly related brc-1 from C. elegans shows that it interacts with the worm homologs for BARD1, FACD2 and RAD51, as well as a number of other DNA repair proteins. For more information on the sequence homology of the human interaction network to other species, please see the protein homology page.

After looking at the function of some of these proteins, it is clear that many of them are also involved in DNA repair, indicating that there may be some complex formation with BRCA1.

From STRING 8.0, 2009. BRCA1 evidence map. Retrieved from http://string.embl.de/newstring_cgi/show_network_section.pl.

OSPREY

OSPREY is downloaded through the OSPREY website as a computer program [1]. The program allows you to search a gene in a number of model organisms, then add interactions by connecting to a number of databases through the internet. Because of the vast number of proteins that have been associated with BRCA1, I chose to limit my results in a number of ways. I ran the results in two ways: coding the edges by experimental process and coding the edges manually to highlight the data found in STRING. I used the same initial parameters for all runs. These included taking experimental evidence from only Affinity capture-Western, Affinity capture-MS, Co-purification, and Two-hybrids. I chose these because I feel they are the most reliable to give more physiological interactions between proteins, as in vitro studies cannot achieve, as well as more precise, as co-localization fails to do. I also removed proteins linked to GO processes less commonly associated with BRCA1, including Carbohydrate metabolism, Protein biosynthesis, and Protein amino acid dephosphorylation. Initial results obtained are shown in Figure 1.

Next, I further limited GO processes to only those dealing with DNA repair and the cell cycle: cell cycle, stress response, signal transduction, DNA recombination, DNA repair, DNA metabolism, DNA damage response, DNA replication, and protein degradation. These were more closely related to BRCA1's major functions. The results, shown in Figure 3, do not look strikingly different from those in Figure 2, indicating that the original interactions were probably more useful than I first thought.

In attempt to single out some trends or specific proteins, I next searched for the proteins found in STRING using their default settings, and used manual edge coloring to show the interactions that were found using STRING. These are shown in red in Figure 4. Some interactions found in STRING were not found in OSPREY, with the vast majority of these being interactions found by textmining. These missing interactions were those between RBBP8 and RAD50, RBBP8 and RAD51, RAD51 and FAD2, RAD51 and RAD50, RAD51 and ESR1, p53 and RAD50, p53 and FAD2, BARD1 and RAD50, BARD1 and ATM, and FAD2 and RAD50. Also two proteins were completely missing from OSPREY: H2AFX and ENSP00000259008.

Finally I managed to narrow the OSPREY results down to a legible number of interactions. This was done by only including interactions between the proteins found in STRING. While I feel that this may exclude meaningful interactions not shown in STRING, it does show that these interactions can be derived in more than one program. The obvious difference between Figure 5 and the STRING Figure 1 is that OSPREY gives interactions that were not found in STRING, though they are looking at the same proteins. For example, BRCA2 was not included in the STRING figure, but in OSPREY is shown to interact with BRCA1 as well as other proteins that BRCA1 interacts with.

Molecular Interactions Database (MINT)

MINT allows you to search by gene name to get protein interaction results in an interactive viewer [2]. Using a cut off score to limit results to the most likely interactions, MINT found 17 proteins (including viral proteins in squares) that interact with BRCA1. From the way MINT presents the results, it is unclear whether or not any of the interacting proteins also interact with one another. The number in yellow indicates the number of experiments showing the interaction. Clicking on the plus sign in the original viewer opens the interactions associated with that protein, as well as those already shown with BRCA1. Unfortuantely, most of the proteins BRCA1 associates with also interact with a large number of proteins as well, and opening these made the viewer impossible to decipher.

From MINT, 2009. BRCA1 MINT viewer. Retrieved from http://mint.bio.uniroma2.it/mint/Welcome.do.

Table 1. Scores associated with the interactions in Figure 6 above. Scores take into account only experimental evidence. Data was taken from MINT database in association with the figure above.

BioCarta

BioCarta provides pre-illustrated pathways, searchable by gene name. I will not go into detail how these pathways work, but I would like to include them here so they can be compared to the above interactions. This shows how these protein interactions integrate into biological pathways in a cell.

From BioCarta, 2009. Role of BRCA1, BRCA2, and ATR in Cancer Susceptibility. Retrieved from http://www.biocarta.com/pathfiles/h_atrbrcaPathway.asp. Copyright BioCarta, 2009.

From BioCarta, 2009. BRCA1-dependent Ub-ligase Activity. Retrieved from http://www.biocarta.com/pathfiles/h_bard1Pathway.asp. Copyright BioCarta, 2009.

From BioCarta, 2009. ATM Signaling Pathway. Retrieved from http://www.biocarta.com/pathfiles/h_atmPathway.asp. Copyright BioCarta, 2009.

Analysis

BRCA1 has a wealth of protein interactions to work with (OSPREY found 376 proteins that interact with BRCA1 by various lines of evidence). The challenge is to find more valuable interactions to be able to better characterize them in more detail. The end result would be cellular pathways that show the interactions between proteins and complexes that work to accomplish a specific function, as can be seen in BioCarta figures above. I feel that the best way to narrow down search results is to use multiple databases to find multiple search hits.

STRING was the obvious first choice for looking for protein interactions. Because this database picks only the top hits based on scores that take into account the experimental evidence used, it gives the best-guess estimate as to which protein interactions are most important physiologically. OSPREY does not give scores for protein interactions, which makes narrowing down the number of interactions more challenging. As you can see in Figures 2-4 above, even by narrowing down by the most meaningful experimental processes and GO terms, the figures were still very difficult to interpret. When I only look at the interactions STRING found, Figure 5 becomes much easier to read. It is also interesting in that additional interactions are included if they also are found to interact with this network. MINT did not add to the interactions found in STRING and OSPREY, except for the presence of viral protein interactions. Interestingly, MINT uses a different and more stringent scoring method than STRING, which may be useful, though I did not look into this further.

[1] Breitkreutz, BJ., Stark, C., and Tyers M. (2003). Osprey: A Network Visualization System. Genome Biology,
     4(3):R22. doi:10.1186/gb-2003-4-3-r22.
[2] Chatr-aryamontri, A., Ceol, A., Palazzi, L.M., Nardelli, G., Schneider, M.V., Castagnoli, L., and Cesareni, G.
    (2007). MINT: the Molecular INTeraction database. Nucleic Acids Research, 35(Database issue):D572-D574.
    doi:10.1093/nar/gkl950.

Site created by Jessica D. Kueck
Genetics 677 Assignment, Spring 2009
University of Wisconsin-Madison