Team:Edinburgh/Artificial Selection

Suppose, like most people, you want to create E. coli that is really efficient at degrading cellulose. Rather than do a bunch of difficult genetic engineering work, why not let Darwin do the work for you? If you simply get your E. coli working at 1% efficiency, can't you just use artificial selection to get you the remaining 99% of the way?

If cellulose is available as a food source for your culture, then it should work: the cells will only get food if they can break down cellulose, and so there is a selection pressure to get better at doing so. But there's a problem noted by [http://www.springerlink.com/content/4l4m28lp06120253/ Van Zyl et al (2007)]: if your E. coli simply secrete cellulases into the media, then a rare mutant that produces more or better cellulases will not gain much benefit, since these cellulases float away and will probably benefit other cells instead.

However, if the cellulases are attached to the cell (for example, because they are part of a fusion to Ice Nucleation Protein...) then this problem does not arise; the rare mutant gains all of the benefit, since the sugar that is created is right there beside it.

To demonstrate this, enter Team Synergy's Celluvolve computer simulation!

Rules

The rules are as follows:

• An array of bacteria exists.
• Each bacterium has a "genotype": its ability to produce cellulase.
• The genotype is a number between 0 and 255.
• Each iteration, some random bacteria are chosen.
• These bacteria have a chance to reproduce, based on their food supply.
• Food supply is defined either as:

• (LEFT) The ability of the cell to produce cellulase.
• (RIGHT) The average ability of nearby cells (including itself).

• Reproduction means:

• Overwriting a nearby cell with one's own genotype.
• However, reproduction is not exact. There is "mutation".
• The offspring's genotype is a number close to the parent's.

The left half of the simulation models the case where the cellulase produced by a cell can only benefit itself. The right half models the case where the cellulase can float away and benefit neighbours instead of the cell that produced it.

Software

C code for the simulator is available. Note that the program spits out .bmp files every 50th generation, and runs for 10,000 generations. A complete run will generate more than 100 MB of files...

Simulation run

In this simulation, each pixel represents a bacterium. Its brightness indicates its ability to produce cellulase. In the left half of the simulation, the cellulase benefits only the producer, whereas in the right half, it floats a small distance and so also benefits 4 of the producer's neighbours.

Generation 1000: Hardly a noticable difference...

Generation 2000: Evolution is making more of a difference on the left side...

Generation 4000: Foom... bacteria on the left side are now much better at degrading cellulose.

Dispersal and kin selection

How quickly the "floating cellulase" bacteria evolve depends on the details of dispersal after reproduction. If the daughter cell moves to a random location, it will probably not be able to provide a benefit to its parent. But if it moves to a location nearby, it may well be able to.

Here's a graph of the extreme case, where the daughter can move to any location:

Graph of average genotype (i.e. ability to make cellulase) over time. Blue is the case where cellulases attach to the cell producing them; red is the case where cellulases can float away.

Since E. coli are motile, it's probably true that daughter cells will not remain close to their parents, and so kin selection will not help the "floating cellulase" bacteria evolve.

References

• Van Zyl WH, Lynd LR, Den Haan R, McBride EJ (2007) [http://www.springerlink.com/content/4l4m28lp06120253/ Consolidated bioprocessing for bioethanol production using Saccharomyces cerevisiae]. Advances in Biochemical Engineering/Biotechnology 108: 205-235 (doi: 10.1007/10_2007_061).