It was a beautiful Saturday morning here in San Francisco, as ASCB 2008 got underway. The full program doesn't really kick off until the Sunday, but there was still plenty to see today, including 16 different 'Special Interest Subgroups' on topics ranging from the technical (Atomic Force Microscopy & Optical Trapping, single molecule imaging...) to the biological (nuclear actin, retrograde membrane trafficking...).
I decided to pop in on a session on "Building The Cell", and ended up staying the whole afternoon. The session was organized by Wallace Marshall from UCSF, and he'd invited a mix of speakers ranging from cell biologists to physicists and computational biologists. Why? Marshall's idea was that scientists can directly study things at the molecular scale, and directly study things at the cellular level. But it's harder to see what's going on in between: and this is where he thinks modeling will prove to be important. Rather than running a session purely about modeling however, Marshall wanted to engage 'regular' cell biologists by 'mixing it up a little' and illustrating how modeling is being used to address a fundamental biological question: how do you build cellular structures?
By and large, I think this approach worked and made for an interesting session. A trio of talks dealt with how membranous organelles are formed and shaped. Gia Voeltz (U. of Colorado, Boulder) takes a straight, biological approach to studying how proteins called reticulons form the tubular structures of the ER. Orna Cohen-Fix (NIH, Bethesda) gave a talk on the importance of lipid production to forming a nucleus with the correct size and shape - a potentially important question, given that many tumor cells have irregularly-shaped nuclei.
Jian Liu (UC Berkeley) also looks at membrane curvature, but he takes a modeling approach to studying how endocytic vesicles form. Many of the molecular players involved in endocytosis are already known, but Liu is able to get insights to how they interact and are regulated by generating mathematical models. Liu showed a movie of a theoretically generated endocytic vesicle forming (basically a straight line invaginating with the same kinetics as a real endocytic vesicle), which was actually pretty cool, given that the movie was generated from his mathematical model of 25 parameters. Liu says we get a lot of insights from this, such as the existence of positive feedback loops which determine the assembly and disassembly of the proteins involved in vesicle formation and membrane fission.
Attention then switched to the cytoskeleton. Ed Munro (U. of Washington) models how early C. elegans embryos generate and maintain polarity, which involves actomyosin contractility around the cell cortex and the action of Cdc42 on 3 different pathways: Par6 recruitment to the membrane, Arp2/3-mediated actin-branching, and myosin recruitment and activation. Again, his mathematical models and simulations trying to recreate polarity yielded new insights, such as the presence of feedback loops in all 3 of the Cdc42 pathways.
Next-up, Margaret Gardel (U. of Chicago) looked at how myosin motors affect the actin cytoskeleton to generate traction, Jennifer Ross (U. of Massachusetts, Amherst) flipped things round and explained how the cytoskeleton affects motors. Ross, a physicist, gave a great talk on how the microtubule-based motors kinesin and dynein/dynactin behave under different circumstances. For example, if microtubules are overlayed with one another in vitro, dynein/dynactin carrying cargo may switch tracks, reverse, or fall off when it reaches a crossover point. But if the density of dynein/dynactin molecules is increased, the motors and their cargo will simply pause at the crossover. This is just one way in which cytoskeletal architecture regulates motor behavior: microtubule-associated proteins, post-translational modifications, and actin-microtubule interactions all have their effects too.
Finally, Alex Mogilner (UC, Davis) gave a nice talk about mitotic spindle assembly. He tries to recreate the process of microtubule 'search and capture' of kinetochores in simulated models. He adds in various factors and scenarios, and sees whether he can capture all the kinetochores in silico in about the same amount of time that real spindles form (approximately 20 minutes). The key factors seem to be chromosome movement - more dynamic chromosomes in the simulation induced faster spindle assembly - and having microtubules grow from the chromosomes as well as from the spindle poles (This has been proposed from biological studies, so the model suggests it might be critical). Problem solved? Not quite - in this simulation, the kinetochores were often incorrectly attached, with sister chromatids often attached to the same spindle pole. To get correct attachment, Mogilner had to model a scenario in which the attachment of one chromatid results in the chromosome rotating so that the sister chromatid faces the opposite pole.
It remains to be seen whether this last scenario happens in vitro, but Mogilner's talk, along with the session as a whole, nicely showed how computational models can provide fresh insights to cell biological questions. We might not all have the ability to do the modeling ourselves, but we can certainly pay attention to the results they yield.