I am here at Heidelberg after a great day at EMBL attending a symposium on the role of variability and noise in biological decision making. As usual, I made a few scribbles which I know I will not be able to decipher once I get home, so here is a quick transcription and elaboration. Of course, being a newbie to this subject area, what I’ve written below is bound to be riddled with inaccuracies, I’d be very grateful to any reader who can correct any misconception below.
It all started with an enthusiastic talk by Arjun Raj with an interesting question: how does DNA know how big the cell is? This information is needed so that it can produce the right amount of specific proteins, and the right quantity of these is controlled by the control of transcription rate (rather than degradation rate, apparently). One thing I learned: a typical volume for cells of 4-5 pico litres. Arjun also showed great microscopic images of cells, with chemicals clearly localized to cytoplasm and also dna localization.
This idea of nuclear compartmentalization was followed by Nico Battich, who again started with an interesting effect: transcription is noisy, but cytoplasmic concentration is uniform. How is this achieved? His idea is that nuclear membrane works as a leaking tap, with diffusion rate allowing nucleus to be used as a buffer. A takeaway concept: concentration in terms of cell surface vs cell volume, particularly useful when thinking of membrane proteins.
Talking of nucleus, another thing I learned, but over dinner from a new acquaintance from Macau, is that some hormones, specifically estrogen, work on nuclear receptors, rather on cellular membrane like other hormones. All these insights on cellular compartments seem very relevant to my student Faiz’s work on developing a compartmentalised cell simulation model.
Another theme of the conference is noise, and the talk by Eric van Minwegen explored this theme very nicely. In developing synthetic promoters, the best ones they developed present low noise levels, though standard natural promoters have a much higher level of noise. Moreover, higher levels of noise exist when promoters are regulated by a transcription factor, and even more when regulated by two transcription factors. Which points to an interesting role for noise in the interplay with regulators.
These are just some of the talks, others have been equally interesting but I can’t type more now. A few general observations: biologists seem extremely polite. All speakers invariably start by thanking the organisers, some for inviting them to give a talk and some being even more subtly polite by thanking for being allowed to give a talk. Most comments and questions from the audience invariably sttart with a comment about how interesting the talk was. All very polite. Of course, the talks were all very good.
At the start, the organisers made a plea for audience not to take photos or to live-tweet the talks, which I thought was very sensible. But I can’t resist jotting down these notes, for myself as well as those who couldn’t benefit from being here.
Edited to add the following…
The rest of the conference was equally interesting. As you may have realized, I am very interested in spatial heterogeneity, and Bernd Bodenmiller talked about the structure in tumors, and in particular about the micro-environment and the importance of cell-cell interactions. I couldn’t help but notice the similarity with biofilm structures. Following the leads here took me to this image, which shows in a very striking way how the micro-environment is structured.
That image is all very well as a cartoon, but how can that structure be obtained from real samples? Chris Bakal answered this question, by showing how each pixel in a microscopy image can be analyzed in terms of up to 32 markers, in order to build a detailed map. What Chris Bakal showed next was even more exciting: how the images can then be used to classify the cells in terms of many characteristics – such as nuclear eccentricity and ruffle-ness, cell roundedness, and so on. I am keen to play with PhenoPlot, a visualization software that can generate simpler glyphs from imaging data. He was mainly trying to address the chicken-and-egg question, of the relationship between biochemical processes and cell shape changes. His main result, from what I understood, is that NF-kB triggers changes to the nuclear shape, which then results in many other effects. More details in paper.
I had never heard of NF-kB before this conference, but it seems to be an important component in the understanding of cancer.
From the start of the conference I noticed that Sabrina Spenser seemd to be highly regarded, and when I heard her talk I could understand why. She gave an exciting, very interesting talk on cell division – again something that I haven’t really thought about much before. For example, it seems the cell division cycle is divided into 4 phases: G1, S, G2 andM, and it is in the transition between G1 and S that the cell commits to division. or if not it balances with the quiescent state G0. More precisely, the commitment is associated with therestriction point R, and it is here that the famous mitogen system of ERK, etc – which many computational biologists have modelled – comes into play. She showed how CDK2 translocates from nucleus to cytoplasm, and she showed this through images where the nucleus was visualised by linking histones’ H2B markers with Turquoise and the CDK2 itself to Venus fluorescence. I am finally understanding a little bit about how imaging data is being used to drive model building.
Another excellent visualization of translocation was provided by Alexander Loewer. From the abstract, he is looking at how `upon binding of TGF-beta tp its receptors, SMAD proteins are translocated to the nucleus and activate numerous genes’. This is done by looking at live-cell microscopy, where SMAD2 is bound to YFP and nuclear H2B is bound to CFP. One of the points made is that p21 is an effector protein for the system. What was very nice in this talk is the cohesion between perturbation experiments and mathematical modeling in developing a detailed and well-founded description of a spatially complex system.
In fact, one of the nicest things in this conference was the openness that many of the biologists expressed to mathematical and computational modeling: I found them much more open to the potential contribution, with several people I chatted to saying that the community needed more computational work. I expressed my surprise at this openness and one person – I can’t remember who – said this is natural because the whole subject of heterogeneity and noise arose from a very physics-based vision, and there has always been an acknowledgement that the phenomena are so complex that models are really required. On the other hand, there were some people I talked to who were not particularly open to the role of modeling. But overall, it seemed a great opportunity for truly interdisciplinary work, so I may actually step out of mycomfort zone of microbiology and try to work with these complex eukaryotic systems.