Weekly Discussions
January - February
19 January 2010
Why make something noisy?
- The benefit of sampling
- or maybe trying every ombination along a spectrum
- a maze
- Minesweeper? - except a wrong turn can't be completely detrimental to the whole system .... maybe though - aging / death discussion
Positive feedback loops --> bistable, bimodal phenotype
- ex. S. cerevisiae with galactose (Acar paper) - have 2 positive, 1 negative - negative allows switching between states
- perhaps the two states can be complementary to each other
- random on/off transition
- barcodes? decoding, decrypting - combination lock-ish - bacteria as a random number generator - stops when it fits the "slot"
Meeting with Dr. Campbell
- went through papers - paper simulating how noise can continue an oscillatory system
- positive feedback loops are important
- Minesweeper --> maybe Battleship
- the aging papers Dr. C sent - thinking can somehow randomly switch something on, if beneficial to survival @ that time, will select for those and kill the others
- maybe some cell differentiation using +/- feedback
- maybe some antenna-ish like sampling - a Biosensor - cell has lots of sensors, randomly sample for something, if detect something with one of the sensors then report...do something else while other cells can keep sensing
25 January 2010
I still like the oscillatory idea - noise driving oscillation --> What is oscillation good for? Combining oscillation and quorum sensing - synchronize system
Biosensors = about the promoter - want sensor up/down (on/off, like an antenna) from cell to cell to sense environment --> can maybe start positive feedback loop to amplify report / signal
Possible design: sensor promoters that are repressed by LexA, like a cap. Have RecA (protease) that cleaves off LexA from some of the sensor promoters and not others, noisily. Or perhaps hin/hix system can randomly flip some promoters on (correct direction), some off (incorrect direction).
Perhaps use quorum sensing to get coordinated response from cell population - if one cell senses, Lux operon activated - tell other cells to... do something Maybe turn off one of their sensors/reporters?
One of the problems with fluorescent proteins according to the biosensor articles is the detection level - need a lot of it. Perhaps some positive feedback loop here could induce differentiation by noise level? (bistable noise threshold)
Report sensing by inhibition / loss of signal - could have sensor promoter transcribe a protein to degrade reporter - if after a certain time, protein not degraded, then the degrader is degraded (kind of like the noise aspect of the bacteriophage lambda system)
Could have biosensors to mark bacterial communication or movement?
In meeting: Dr. C - what would be the purpose of making this a noise-based system versus always having all sensors on? Don't know. He suggested I read the aging articles for next week.
10 February 2010
I read the aging articles this week, and I thought particularly the Nystrom review was helpful/ thought-provoking. I thought perhaps it would be interesting and useful to create a system that could be applied to a number of different synthetic devices in order to control the noise threshold, and any cells running the device outside of that noise threshold would die, based on the error catastrophe theory.
We talked about this and ppGpp in meeting. Sigma factor competitive binding could have a strong and useful noise component. Random number generator is also a viable idea. Perhaps using sigma factors with this idea, having competitive binding between two reporters as a binary system. Dr. C suggested in a binary number system, cells could be run through a flow cytometer.
One of the oscillatory system articles I read mentioned superconducting junctions - which on the surface sounds interesting and perhaps something cells could do. I'll look into it.
17 February 2010
I read more about the error catastrophe theory and ppGpp this week. (p)ppGpp is very important to the cell, in many species - which means it can be very useful, or very detrimental to mess around with and very tightly regulated, which to me does not lend itself nicely to a noise study. It's still classified as a potential useful part on the chalkboard page.
We began the meeting talking about ppGpp. Dr. Campbell agrees it is still a potentially useful result, and since it has a cofactor DksA, perhaps even though induction of ppGpp is a stringent response, there is a noise factor in the expression/availability of DksA. We should look more into the regulation of DksA and the synthesizers SpoT and RelA. We discussed cells "flipping" into senescence vs. growth and division cycle:
- What if we introduced a toxin that could kill all rapidly dividing or growing cells --> selecting for cells that have randomly flipped into senescence
- Perhaps could synchronize the cultures - would involve getting them all to the same Time 0 point in the cell cycle (I'm pretty sure I've read a yeast paper like that, or there's also something like that in the Genomics textbook, probably around the topic oscillation - will look up)
- Can we select cells based on their flip rate (senescence / growth)? Purpose?
- Make Baby Boomer Bacteria - cells that we somehow stop at a certain point in cell cycle so that they never divide --> get really old cells - perhaps use some sort of negative feedback loop?
- There is a drug that causes cells to keep elongating without dividing (get from Dr. C or look up)
Could we do something that involves elongation, the separation of some plasmid & transcription factor randomly into dividing cells - those that have both... Or conjugation - when cells elongate or move around and make contact with another cell something happens...
Some ideas of Dr. Campbell's on the way home....
1) What if we utilized transposons for random events? Perhaps placing Tet resistance downstream of random genomic loci?
See this link http://www.bio.davidson.edu/Courses/Molbio/MolStudents/spring2010/DesPain/Transposase.html
2) What if we had 20 strains of knockout E. coli and each one could not make one amino acid or nucleotide? Then as the cells got near each other, they might be able to absorb the missing metabolite and grow faster than normal due to limited nutrient agar? As cells congregated, they would make more of each others missing metabolite and the largest, most diverse mass of cells would grow the quickest?
24 February 2010
Discussed the elongation papers (Goksor, Sturgeon, Gustaffsson) and Danchin. Dr. Campbell suggested it may be best at this point to pick one or two topics to hone in on and try to formulate a proposal about it for after spring break. I'll work on the noise threshold device and elongation for my proposals.
March - April
10 March 2010
I finished a proposal for the noise tuner device, and I started the introduction of an elongation proposal outline, but didn't get very far; I'm still not sure (and neither is Dr. C) what to do with noise and elongation. I did note an idea about making chains that act as bacterial nerves and send an impulse down the chain of connected cells. That does not have a strong noise component, although it would be very fun to do.
Dr. Campbell liked the versatility of the design, that it could utilize Lux, which is used in so many synthetic systems that a noise tuner could perhaps be relatively easily added into previous and future devices. He did have some ideas:
- Instead of tuning the system to some specific degree (as tuning the destabilization of the repressor protein in Isaacs paper), it would still be a good innovation to design and construct a noise tuner(s) that tolerates a specific non-tunable window of noise, have perhaps three different window options
- Perhpas could utilize a longer cascade idea (because noise increases in longer cascades) and the end of the noisy activation cascade could be a repressor for Tet, if there's a certain threshold of noise to start the cascade, it'll be really noisy by the time it gets to managing Tet
- "Tet" could be a toxin instead
I'll work first with the abstractions of the concept, maybe figure out how to simulate it, and then figure out the actual biological parts that would help with constructing it.
During lab meeting, Dr. Heyer raised the question whether cells that express outside of the noise window would at a later time point perhaps fall back into the window, or is that expression state stable. Positive feedback bimodal distribution, from what I've read sounds like it is bistable, but perhaps that depends on the protein being expressed - if it were something like Tet that affected livelihood, it may be different. Should look into it.
She also noted that this type of device is called a band-pass filter, used a lot in optics. So that might be a useful search term.
17 March 2010
This past week, I searched for papers about band-pass filters and I found two articles by Sohka et al (2009) detailing how they constructed a tunable bacterial band-pass filter. I read through the papers with Dr. Campbell, and we decided that these do not impede on my project idea, but do provide validation for the importance of this project, and leave room for improvement.
One thing from the Sohka et al paper(s) is that their tolerable window size does not change, but can slide, so different abundances of enzyme are output from the system. Dr. Campbell suggested that perhaps one way to make a window that slides with my project would be varying the plasmid copy number. In lab meeting the point was raised that if I'm using a positive feedback loop to cause a bistable distribution, the effect can't be so extreme that there are no intermediates that survive the positive feedback loop. For that, I thought perhaps putting some repressor somewhere into the design might be effective.
I still have to have a meeting with Dr. Heyer. I am working on figuring out what I would want to use to make the two thresholds for the noise window.
I'm still kind of toying with an idea similar to the Isaacs paper sort of model - of thinking of a protein whose activity or expression can be destabilized/driven by a certain noise threshold (so it can tolerate a certain noise window) and it can also be destabilized by some other controllable factor, like heat or some concentration of a substrate that could be added to media in order to minimize the tolerable noise threshold. This protein is responsible for starting a positive feedback loop that either kills the cell or lets it live (perhaps by not starting the positive feedback loop).
Although I would really like that idea - which I think views the noise aspect as more of a threshold than a window with both sides to be accounted for (but is not really "embracing" noise like amplification by positive feedback would) - discussions right now are focused more on a window. So, again, i'm working on figuring out what I could do to rule out all cells that are noisily above a desired output, and noisily below a desired output.
7 April 2010
I had a meeting with Drs. Campbell and Heyer to because I'm having some difficulty sorting out the issues with the noise tuner and figuring out how to effectively communicate and illustrate it in a paper. We covered two main issues:
- How do you control when the cells will die? If there's a lower threshold, how do you not kill all cells before they build up output above that threshold? How do you know all cells won't die if they fall out of the thresholds once (or do you want this)?
- What system would be better than using Lux - as a quorum sensing mechanism it would cause a population response instead of single cell survival/death. What if the existing system has LuxR?
Things to discuss in paper:
- Review of noise
- Design
- Communication issues with design
- Death issues with design
- Hin/hix?
- Promoter activation
- Cre/lox
