November 18, 2013
Imagine you have the instructions for building a car but you don’t know what any of the specific parts do. In other words, you can build a working car but you don’t understand how it works.
One way to figure out how the car works would be to remove a part and see what happens. You would then know what role that part played in getting a car to run.
So if you remove the steering wheel, you’d see that the thing runs into a wall. That part must be for steering. When you take out the radiator, the car overheats so that part must be for cooling the engine. And so on.
Sounds like a silly way to figure out how the car works, but this is essentially one of the key ways we try to figure out how a cell works. Instead of parts, we knock out genes and see what happens. A new study by Teng and coworkers is making us rethink this approach.
See, one of the big differences between a machine and a cell is that the cell can react and adapt to the loss of one of its parts. And in fact, it not only can but it almost certainly will.
Each cell has gone through millions of years of evolution to adapt perfectly to its situation. If you tweak that, the cell is going to adapt through mutation of other genes. It is as if we remove the radiator from the car and it evolves an air cooling system like the one in old Volkswagen Bugs.
Teng and coworkers decided to investigate whether or not knocking out a gene causes an organism to adapt in a consistent way. In other words, does removing a gene cause a selection pressure for the same subset of mutations that allows the organism to deal with the loss of the gene. The yeast knockout (YKO) collection, which contains S. cerevisiae strains that individually have complete deletions of each nonessential gene, gave them the perfect opportunity to ask this question.
There have long been anecdotal reports of the YKO strains containing additional, secondary mutations, but the authors first needed to assess this systematically. They came up with an assay that could detect whether secondary mutations were occurring, and if so, whether separate isolates of any given YKO strain would adapt to the loss of that gene in a similar way. The assay they developed had two steps.
The first step was to fish out individual substrains from a culture of yeast that started from a single cell in which a single gene had been knocked out. This was simply done by plating the culture and picking six different, individual colonies. Each colony would have started from a single cell in the original culture.
The second step involved coming up with a way to distinguish differently adapted substrains. The first approach was to see how well each substrain responds to increasing temperatures. To do this, they looked for differences in growth at gradually increasing temperatures using a thermocycler.
They randomly selected 250 YKO strains and found that 105 of them had at least one substrain that reproducibly responded differently from the other substrains in the assay. In contrast, when they looked at 26 isolates of several different wild type strains, including the background strain for the YKO collection, there were no differences between them. This tells us that the variation they saw in the knockout substrains was due to the presence of the original knockout.
So this tells us that strains can pretty quickly develop mutations but it doesn’t tell us that they are necessarily adapting to the knocked out gene. To see if parallel evolution was indeed taking place, the authors chose to look at forty strains in which the same gene was independently knocked out. They found that 26 of these strains that had at least one substrain with the same phenotype, and fifteen of those had mutations that were in the same complementation group. So these 15 strains had evolved in similar ways to adapt to the loss of the same gene.
Teng and coworkers designed a second assay independent of the original heat sensitivity assay and tested a variety of single knockout strains. They obtained similar results that support the idea that knocking out a gene can lead cells to adapt in similar ways. This is both good and bad news.
The bad news is that it makes interpreting knockout experiments a bit trickier. Are we seeing the effect of knocking out the gene or the effect of the secondary mutations that resulted from the knockout? Are we seeing the loss of the radiator in the car or the reshaping that resulted in air cooling? We may need to revisit some earlier conclusions based on knockout phenotypes.
The good news is that not only does this help us to better understand and interpret the results from yeast and mouse (and any other model organism) knockout experiments, it also gives us an insight into evolution and maybe even into the parallel evolution that happens in cancer cells, where mutations frequently co-occur in specific pairs of genes. And while we may never be able to predict if that knock you hear in your engine really needs that $1000 repair your mechanic says it does, we may one day be able to use results like these to predict which cells containing certain mutated genes will go on to cause cancer and which ones won’t.
by D. Barry Starr, Ph.D., Director of Outreach Activities, Stanford Genetics
Categories: Research Spotlight