Transcription and Translation

Friday I was supposed to meet up with Mike Springer from the Kirchner lab. At some point Mike and I had set up a collaboration in order to figure out what was so special about little regions of the genome that encode signal sequences. (To read more on my paper and what we did click here).

In any case Mike had emailed me that Alex van Oudenaarden was giving a Systems Biology “Theory Lunch” and that he had to postpone our lunch. Having heard Alex once before and being impressed, I decided to check it out.

It was one of the best seminars I’ve attended in quite a while.

Now I’m not going to give you any of the specifics but instead give you a summary of the approach that Alex’s lab used to decipher how a regulatory loop that regulates osmotic pressure in budding yeast operates.

Alex is very much an engineer at heart. He started his talk by considering a CD player. If you were given one, how would you figure out how it worked? What you would not do is deconstruct the player in to nuts and bolts and wires and model the entire mess of primary components up.
You would use a volt meter and figure out how subunits are linked. You would send in an input and read the output. Then you could try to model how this player is put together using smaller components. In other words you would reduce your black box down to smaller black boxes and describe each smaller component in terms of how it manipulates an input into an output.

The idea is to treat the organism like a black box. Give it some input, record the output, then fit the input/output to some equation. Using this method you can derive the nature of the cellular machinery (smaller black boxes) needed to produce such a behavior. Now this is in stark contrast to other mathematical modelers who attempt to write down equations that describe the interaction between EVERY molecule in a system.

Using the first approach you can figure out what parameters are important. You can then try to break or alter the cellular machinery by making mutants. Eventually you can describe the feedback loops in terms of small components that have specific tasks. Thus a map kinase cascade can act as an amplifier o some other group of components can act as a transistor. In the end you get some insight into how the system is built. In the more standard modeling approach you end up with … chaos.

My biggest problem with modeling (and Systems Biology) using the standard approach is that you must know and understand all the components. But what happens if you are missing some key protein or don’t know of some key feedback loop? Your model becomes worthless. In addition the more equations you write the more free parameters there are. In the end most of the exercises seem useless. But in Alex’s case you get the opposite phenomenon. By modeling the SIMPLEST set of equation that represents how the system works, you can figure out how the mechanism should work. In addition these models have strong predictive power as in “there should be two feedback loops, one fast, the other slow, and they should converge at point X”. Now you can go back into the organism and find whether there are two loops and how they are made.

Excellent stuff.


  1. #1 PhysioProf
    May 11, 2008

    That is, indeed, a beautiful metaphor!

    Hey, question for you: Can you point me towards some literature that analyzes whether different secretory signal sequences have different biological consequences? Or is it just the case that a signal sequence is a signal sequence, and there is no biological relevance to any particular sequence, so long as it is sufficient to target the SRP?

  2. #2 MadGenius
    May 12, 2008

    Hi Alex – probably a stupid question, but I’ll ask anyway.

    How much can a ‘regular’ biochemist/cell biologist contribute in Systems biology? Most of what I need requires one to have a really good maths/engineering background.

  3. #3 Alex Palazzo
    May 12, 2008

    The Hegde group has looked at this question in several papers, here’s a post on it. Also from my work, it would seem that the signal sequence coding region promotes mRNA export of the transcript that encodes the secreted protein. It also impacts whether the mRNA is properly targeted to the ER. These activity requires that the coding region has a low adenine content. We found that different signal sequence coding regions promote different export rates. These varying rates will have an impact on the level of protein production.