The work was co-led by a fantastic grad student in my and @cncurtis lab @SusanneTilk and a truly stellar postdoc in my lab @cd_mcfarland. For me this work finally solved some troubling puzzles about how evolution works in cancer. We would really love to know what you think! 2/n
-
-
Prikaži ovu nit
-
Note that
@SusanneTilk posted a very clear description of the main results here: https://twitter.com/SusanneTilk/status/1173693668793831424?s=20 … I will just add some of my personal musings. 3/nPrikaži ovu nit -
This is our first foray into cancer genomics and evolution. Cancer biology has been a big focus for us but so far we have focused primarily on modeling cancer in genetically modified mouse models as part of a truly great collaboration with http://med.stanford.edu/winslowlab.html 4/n
Prikaži ovu nit -
That collaboration was also co-led by
@cd_mcfarland, as well as Zoe Rogers and Ian Winters who all developed Tuba-seq: http://petrov.stanford.edu/pdfs/0136.pdf , http://petrov.stanford.edu/pdfs/0143.pdf , https://news.stanford.edu/2018/04/03/stanford-scientists-track-cancer-growth-crispr/ …. That work is still going strong but Chris and Susanne here opened a new chapter. 5/nPrikaži ovu nit -
PSA: Chris McFarland is on the job market this year. Don’t miss your chance! 6/n
Prikaži ovu nit -
Ok, so what is the puzzle? First some background. Cancer sequencing showed that most cancers accumulate many mutations in many genes. Some of these mutations are likely to be advantageous for the tumors (they generated drivers) but many are thought to be mere passengers. 7/n
Prikaži ovu nit -
The word passenger is good here as it connotes hitching a ride with a driver. This is particularly appropriate in clonal, somatic growth without recombination. Once a mutation lands on the genome of an expanding clone it got it made, even it doesn't help the expansion. 8/n
Prikaži ovu nit -
Passenger mutations by definition can be either neutral or harmful. So what are they in cancer? The standard approach to test this is to ask whether natural selection weeds out putatively functional passengers at a higher rate than the mutations more likely to be neutral. 9/n
Prikaži ovu nit -
These analyses suggested that almost none of the putatively functional, amino acid replacement mutations are weeded out by selection in cancer unlike in germ-line evolution where the majority (often vast majority) of replacement mutations are eliminated 10/n
Prikaži ovu nit -
These fantastic papers from 2017 by
@imartincorena and colleagues as well as Donate Weghorn and colleagues established this pattern quite clearly: https://www.ncbi.nlm.nih.gov/pubmed/29056346 and https://www.ncbi.nlm.nih.gov/pubmed/29106416 11/nPrikaži ovu nit -
Why would this be the case? There are two possibilities. One is that these replacement mutations are neutral. They do nothing at all possibly because most genes in humans are only important for multicellular function and development and dispensable in clonal growth. 12/n
Prikaži ovu nit -
This is a fascinating hypothesis but never seemed likely to me because most selection against germ-line protein mutations is due to their effect on folding and unrelated to function. This was shown convincingly by
@dallandrummond and colleagues https://www.pnas.org/content/102/40/14338 … 13/nPrikaži ovu nit -
I don't see a reason why somatic cells should escape the deleterious effects of protein misfolding as long as the putatively dispensable genes are expressed in cancer. And they are. So, something does not quite make sense here. 14/n
Prikaži ovu nit -
The other possibility is that passengers are often quite harmful but selection can't remove them. One key possibility is the lack of recombination allows deleterious passengers to hitchhike with the drivers as long as the drivers are stronger than the passengers. 15/n
Prikaži ovu nit -
@cd_mcfarland and@SusanneTilk decided to test this possibility in a very clever way. It is known that selection is particularly inefficient in the absence of recombination when mutation rate is high. 16/nPrikaži ovu nit -
If mutations are rare then they also rarely end up being linked and selection can notice marginal effects of mutations quite easily. If mutation rate is high, marginal effects of mutations are less important. Both purifying and positive selection become inefficient. 17/n
Prikaži ovu nit -
Here
@SusanneTilk and@cd_mcfarland show that these predictions hold. Cancers that have very low mutational burden show strong evidence of purifying selection against passengers (~60% of replacement mutations are eliminated) and very strong positive selection for drivers. 18/npic.twitter.com/Z5HkruXEnX
Prikaži ovu nit -
These patterns hold across multiple datasets, mutational calling algorithms, cancer types, and clonal and sub-clonal mutations. Similar patterns are seen in copy-number mutations. An aside - while the paper is short and readable it describes a ridiculous amount of work! 19/n
Prikaži ovu nit -
Having established the pattern of attenuation of efficacy of selection in the high mutation burden tumors,
@SusanneTilk and@cd_mcfarland used simulations to demonstrate consistency with predictions of simple models of clonal evolution and estimated some key parameters 20/nPrikaži ovu nit -
They showed that the data are consistent with drivers having on average ~20% advantage (which is roughly what we see in mouse models for Lkb1 loss http://petrov.stanford.edu/pdfs/0136.pdf ) and passengers having an average individual cost of ~1% and total cost of ~40%. 21/n
Prikaži ovu nit -
So it looks like the tumors carry a large deleterious load of many likely mis-folded proteins. How do they manage this? And does this generate a key vulnerability in tumors that can be exploited therapeutically as was suggested previously? https://www.ncbi.nlm.nih.gov/pubmed/23388632 22/n
Prikaži ovu nit -
I am very excited about this work and for the authors. It is particular fun that this work confirmed Chris's Ph.D. work with
@leonidmirny where they did argue that cancer progression is the tug of war between advantageous drivers and deleterious passengers. 23/nPrikaži ovu nit -
I would like to also thank
@cncurtis and both of our labs for great discussions! Please do let us know what you think - we would love to know. n/nPrikaži ovu nit
Kraj razgovora
Novi razgovor -
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.