I just finished a textbook that Drew Endy gave me on the design principles of biological circuits.
There is a fascinating section describing how the transcription networks of E.Coli (the common bacteria in your gut) robustly build electric flagella motors on demand, and a navigation system that senses food gradients across distances larger than the bacteria itself. It then moves at 30 body lengths per second.
It’s an interesting case study in gene transcription networks. When E.Coli is bathed in nutrients, it focuses energy on cell division (growth) and not movement. With a lack of nutrients, a genetic trigger induces the manufacture of several helical propellers (flagella) to enable it to swim to a better life.
In the diagram above, you see the 45nm wide, 100 Hz proton-pump rotary motor, as it is assembled in stages from 30 different proteins (text labels above). The motor and flagellum tail are hollow allowing each protein to self-assemble in sequence as they move down the assembly line straw.
Two transcription factors, let’s call them X and Y, regulate the six operons (gene clusters Z1 to Z6) to produce all of the proteins in the proper sequence. Master regulator X activates Y and both jointly activate the sequence of six operons Z1 – Z6, in order, with logical OR gate inputs. The signaling network is a classic multi-output feed-forward loop, common in biological networks, especially sensory transcription networks. By reversing the activation thresholds (K) of X and Y, a FIFO order is achieved in protein production, a just-in-time manufacturing sequence. See diagram below.
Like all feed-forward loops, the network embeds other valuable information processing. Each of the output nodes has a sign-sensitive input filter, ignoring intermittent absence of the X signal and noise from a fluctuating environment. Deactivation only occurs when X has been off for about the time period of one cell generation, the time needed to complete assembly of the flagellar motor. See more examples of network topologies and benefits below.
The author proposes convergent evolution across many information networks, from genes to protein kinase cascades, to neurons, which is quite plausible.
The premise that I wrestle with is his claim that these networks are readily understandable. Working from the bottom up, and from the incredibly sparse networks and topologies, I can see why he’s excited. But I wonder if this scales. Jumping to neuronal circuits, the easy modularity is a bit more elusive, and I wonder if the simplified networks of parasitic organisms are a simple tier in the hierarchy of abstractions, just a few steps more complex than codon encoding and epigenetics. Perhaps evolved information networks embed much more accumulated computational complexity and offer fewer pattern matches to our engineered artifacts.
The information systems of biology have so many interesting lessons on robustness and distributed action. Here is a fascinating talk on bacterial communication (chemical quorum sensing and nanowires).
Leave a Reply