On 7/5/07, Schutter, Kyle <Kyle_Schutter@brown.edu> wrote: 

Hi Prof. Nicola Neretti

This is Kyle from the Brown iGEM team.  We came and talked with you about the tri stable switch a few weeks ago and you showed us the presentation that Jason had made.  I have since talked to Jason and had him send me the model matlab code, which I have run and reseearched so that I have a good idea of what is going on in the code.  Naturally I now have a lot more questions, especially about how to implement the data and I was wondering if you could send me on the right path to interpret the data.  I am not sure whether you would rather respond by email or have me drop by ship street (fourth floor?), but I will just fire away with my questions.  The matlab code is included. 

One of our variables is cooperativity of our repressors which is an exponent which if it is one then repression will increase linearly with the increase in of repressors.  if the variable is two then the relative repression increases with the square of the number of repressors in the system.  One of the tests we are running is having an inducible promoter that will increase the amount of repressor in the system.  We will do this with several concentration of repressor so that we will know what concentration of repressor represses it promoter.  So, my question is, if we see increase in repression with the square (cube, fourth, etc) does this mean that the cooperativity variable is 2 (or 3, 4, etc)?  Wont we have the problem that we see increased repression over time anyways? Can we take teh curve that best fits the plots at each measured point in our experiment?  Or does the cooperativity have to be normalized against some standard value? 

Another question I have is about our alpha values (the relative transcription/translation rates of the promoter and gene).  The model says that if our alpha values are significantly close together, we should have a stable switch.  But since these values are relative, how do we normalize them. Also, how do we measure them.   and are they really so important.  It seems to me that the transcription/translation rates should be less of a factor than the protein (repressor and reporter (gfp)) folding and the protein binding constants.  If we did include all of these into one variable, could we say that if we started with no reporter protein in cells of a fixed number of plasmids under standardized conditions that if one showed x amount of flourence twice as fast as another it would be a relative 2 to the other at 1. Or should we say that if one shows twice as much flourence after a certain time that it is twice the relative rate of the other or is there some standardized measurement to be aware of. 

Thanks for considering all of our questions. I am sure that we will have more. ;-)

The attachments are a slightly modified matlab code, a picture of our testing constructs, a graph of the tristable region (The legend isnt quite right, the last two entries are switched)  and Jason's presentation. 

If I am not really very clear with my questions maybe we can set up a time that we (tri stable team) can meet with you.  We thought that you seemed to have to most knowledge and might be able to help us here.

Best,

Kyle Schutter

Kyle,

as I mentioned when we last met, I strongly suggest you start from the bistable switch. If you remember, we said the it was possible to decompose the tristable into three bistable  switches connected to each other, and that one of the things to test  would be to see the effect of the relative orders of the two repressors.

If the tristable fails for some reason, at least you would have built and characterized a useful switch that can be the starting point for a bigger project in the future. I believe Jason has matlab code for that, or you can simply modify your code to generate the bistable switch and generate plots similar to those in the original paper. We know that the bistable switch should work, since it did in the past. The tricky part is to build it with new components from the registry and characterize it well. 

This week is quite busy for me since we have a visiting scientist joining our lab. I might have some time to meet in the second half of the week. Let's just do this via e-mail for now.

-Nicola

Kyle,

After we spoke today, it came to my mind that you do not actually need to put the 2 repressors next to each other, but you can simply create another construct with the same prromoter/RBS combination. I.e. instead of 

promoterX-RBSX-repressorA-repressorB-GFP

you will have:

promoterX-RBSX-repressorA-GFP

and

promoterX-RBSX-repressorB-GFP

This should have pretty much the same effect, but will greatly simplify your part's characterization in terms of alpha and beta. The tristable will then be built on top of the elementary parts that have already been used in the bistable. An n-stable will be built in pretty much the same way, with no need to build more complex constructs. 

-Nicola
On 7/18/07, Schutter, Kyle <Kyle_Schutter@brown.edu> wrote: 
Nicola,

Thats a great idea!!!  Hadnt thought of that and it might be less ligations.

Just curious about how high copy plasmids work, if you have 2 100 coppy plasmids in the cell do you have 50 or  100 of each.  If it is 100 then that would be great, if it is 100 total (50+50) then we might have problems putting 6 plasmids in the cell rather than just 3.  Also we might run out of antibiotic resistances (amp, kan, tet, chlor).  I guess we could put them all in the same plasmid 
ie ptet>rbs>araC>gfp>term>ptet>rbs>lacI>gfp>term.  but how many base pairs can we fit in one plasmid say it is 3000 bp.  How much DNA can we stuff in there?

A bit off the topic, but the questoin was posed today how we interpret the GFP readout.  does flourescence increase linearly with increased concentration of GFP or does it drop off?  I dont know if you know the answer or can point me in the right direction.  Maybe there is some paper on GFP output out there somewhere you ran across. 

Thanks for sending me that idea.  We never really thought about that before.

Best, Kyle

Kyle,


I think it should be somehow linear within a range of concentrations that is not too large (see http://www.biocompare.com/technicalarticle/1226/Quantitative-Analysis-Of-GFP-In-Plant-Extracts-Using-The-VersaFluor(tm)-Fluorometer-System-from-Bio-Rad.html
 and 
http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=ShowDetailView&TermToSearch=12845471&ordinalpos=2&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_RVDocSum
)
I suggest you check with Brodsky as well, since he surely knows much more about this. 


Well, with respect to the previous design, you just need to add an additional promoter and RBS. You can probably get away with just one GFP instead of two. So you would have something like: 

promoterX-RBSX-repressorA-GFP-terminator-promoterX-RBSX-repressorB-terminator

instead of

promoterX-RBSX-repressorA-repressorB-GFP-terminator

Still, you can probably test each promoter-rbs-repressor element separately to begin with. 

Just do a quick calculation on how many bps that would add and see if that's reasonable.

-Nicola