?Finding the Weaknesses of Cancer
By xerro five, Section Biology Posted on Wed Apr 28th, 2010 at 06:23:04 PM PST
At this point, we know quite a lot about what mutations occur in cancer and what effect many of these have on the affected cell's performance (in terms of alteration of key cancer-related molecular pathways such as apoptosis and DNA repair). What we don't know much about yet, however, is what vulnerabilities these altered cellular states open up that could be used for therapeutic targets. For example, while it is beneficial for a cancer cell to have self-sufficiency in growth factor signalling, maybe this characteristic also means that the cell is more sensitive to attack with something related to the signalling pathway. To find potential synthetic lethal options, perhaps computational prediction coupled with in vitro validation could be used - providing new cancer cell specific therapeutic targets.

Synthetic lethality as a means for treating cancer is not a new concept, although it has attracted more attention recently. For example, treating BRCA negative breast cancer cells with chemicals that inhibit PARP1 which is a necessary component of the base excision repair pathway creates synthetic lethality only in the cancer cells because only they have the lack of BRCA [1]. The lethality comes from an inability to repair replication-stalling DNA damage by either homologous recombination (no BRCA) or nucleotide excision repair (nor PARP since it is being inhibited), but the normal cells can survive this DNA damage since they still have homologous recombination repair [1]. Although the above example is a good proof of principle, there has been little other progress with synthetic lethality in cancer. So it would be helpful to discover other potential synthetic lethal interactions to use in various cancer types (since each would probably have some different ones) based on what is wrong with them. To do so, an unbiased screening technique could be employed in theory (as has been proposed previously [2]), but while this is fine for systems like yeast, working with human cancer cell lines means that time, resources, and money become limiting factors. So then it would be best to establish targets first through computational prediction based on gene network information. One such algorithm for doing so was published in 2009 which was designed specifically for predicting synthetic lethal relationships, and showed a 95% true positive rate when they compared the results to C. elegans and S. cerevisiae known interactions [3]. While previously such prediction methods and network knowledge were not available, now represents a good time to combine them with high-throughput screens for drugs that work with these targets specifically in different cancer cell lines for validation of using these interactions as new targets of cancer therapy based on synthetic lethality exploitation. It is perhaps due to the feasibility issue mentioned that unbiased screens have either not been attempted or seen much success in this area so far, but this approach should help. Summary: Predict synthetic lethal gene relationships relevant to cancer using an algorithm Screen as many cancer cell lines as possible with as many chemicals as possible that are predicted to work with the predicted relationships to selectively kill cancer cells. Presumably this activity would be through inhibition of a particular protein that would cause synthetic lethality, although activation is not an impossible mechanism. Further, in-depth clinical characterization of positive results Hopefully this information will lead to a new view of cancer where, rather than the more mutations, the worse the prognosis, we would see the situation as something similar to "the bigger they are, the harder they fall" (more mutations = more vulnerabilities). [1] Synthetic lethality--a new direction in cancer-drug development. Iglehart JD, Silver DP. N Engl J Med. 2009 Jul 9;361(2):189-91. [2] The concept of synthetic lethality in the context of anticancer therapy. Kaelin WG Jr. Nat Rev Cancer. 2005 Sep;5(9):689-98. [3] Predicting genetic interactions with random walks on biological networks. Chipman KC, Singh AK. BMC Bioinformatics. 2009 Jan 12;10:17.

< Characterising Lung Cancer Metastases (5 comments) | Let's cure leukemia - oncolytic virus at action again (8 comments) >