Molecularist

I have a thing about noise (see my ramblings here). And as folks try to turn elements of biological circuits into well-behaving engineered parts, I’ve been trying to put my finger on why biological noise means so much to me.

A series of articles in Nature back in September covered the cost of feedback control, the fundamental limits on the suppression of molecular fluctuations, the functional roles for noise in genetic circuits (subscription required for all three).

The main review mentions that noise is a nuisance in the design of deterministic (engineered) circuits. But goes on to show that noise in biological systems provides critical functions that are hard to achieve with deterministic circuits.

Noise in biological systems is unavoidable, there are limits to how much a feedback system can reduce noise. And the review goes on to discuss many areas where noise is integral to the stability or responsiveness of a process, such as gene expression coordination, state-switching, positive feedback, differentiation, and in development.

How might we understand this noise and actually engineer it into our deterministic circuits?

One thing, though, that tempers my bias towards keeping noise in biologically engineered circuits is that digital electronics also had their start in a noisy analog world. Will engineering biological circuits be forever mired in the analog noisy world of biology or will these circuits eventually be complex enough to exhibit the precise nature that biological engineers seem to want? Should biological engineers seek to incorporate noise into their calculations or strive to limit stochastic fluctuations?

Really, that’s a bit beyond my ability to understand circuit design and noise and all, so I leave it to smarter folk than I. 🙂

Image from BarelyFitz

Here’s a paper from September I haven’t had a chance to comment on.

Zhang et al reported in Science (review, paper – subscription required) how two almost identical protein sequences can have different translational modifications. In this case, β-actin and γ-actin are 98% identical, the differences are marginal and don’t explain why one protein is modified with arginine and the other is not.

The difference is actually quite subtle – there is one amino acid in the protein sequence that is lysine in both proteins, but the codons in the RNA are different. That difference is enough to change the relative translation speed of β-actin so that it rapidly folds and gets arginylated but not degraded as happens in the slower translation of γ-actin. In the end, this affects the relative lifetime of each protein in the cell, leading to differential functions.

That’s so cool.

It seems like every so often, things we view as well understood turn out to have another layer of subtlety built in. For example, differential codon usage is usually considered a synonymous change, nothing momentous. Sure, there’s been studies of codon usage relative to the abundance of corresponding tRNAs and how that might be used to modulate abundance of a protein (a quick Google search revealed some interesting codon-usage papers). But I think this case is novel, differential codon usage actually affecting the post-translational modification of a protein simply by tweaking the speed of translation.

How might we create a codon usage table that takes into account tRNA abundance, contribution to translational speed of different codons, and speed of post-translational processes to be able to model and predict things like the differences between the functions of β-actin and γ-actin.

Also, how prevalent is this subtle effect on post-translational modification and how susceptible is it to breaking and causing trouble in a cell?

image from Science review (found via Google images, mind you)