Collective sensing by communicating cells

Speaker: Andrew Mugler
Organising Department: Bionanoscience Department
Subject: Collective sensing by communicating cells
Location: Building 58, TU Delft
Date: 17-12-2016

Author:  Nemo Andrea

Andrew Mugler obtained his PhD at Columbia University for his research in biological networks and models for gene expressions. He has also worked at AMOLF, a well known research laboratory within the Netherlands. In this seminar, he gave a broad talk on environment sensing of both independent cells and collective cell groups.

The start of the seminar focused on introducing the topic and defining the research questions. The speaker covered chemotaxis in E. Coli, a method of sensing a concentration gradient by means of a run-and-tumble method. This was followed up by a brief recap section on diffusion and introducing a concept not yet covered in the nanobiology course, namely the Berg-Purcell limit. This limit relates to how many particles a cell must reasonably count in order to be certain that an observed concentration is not due to low number noise, something that would perhaps set in motion a wasteful process in the cell. This limit is based on the assumption that the membrane has no influence and that the particles diffuse freely.

Naturally, such simplifications can be valuable in certain applications, but in order to test whether this limit is a good reflection of biological reality, a more realistic approach is required. Research by Bialek & Sateyeshgar (PNAS, 2005) incorporated membrane interaction using the fluctuation dissipation theorem, while Kaize et al. (Biophysical Journal, 2014) used reaction diffusion theory. Both these models are closer to the biological system, and in both models, the Berg-Purcell term sets the noise floor. In other words, while the other models did have some additional noise terms, the minimal noise was still determined by the Berg-Purcell limit. This Berg-Purcell limit has been applied to both E. Coli chemotaxis, wherein the Berg-Purcell limit gives an estimate of the minimum time required for a cell to make a reasonable assessment of the concentration around it matches the observed parameters in experiments. A similar predictive power of the limit was observed in experiments with amoeba.

The question Mugler is interested in is whether cells could surpass this limit by communicating. They created a model of cells that could use short range communication to exchange information and solved this model using differential equations. These differential equations also included noise terms and solutions were obtained by working in Fourier space. These solutions showed that whereas a single cell has a prefactor of 1/2 before the noise term, two cells had a factor of 3/8. This shows that indeed cells can reduce noise by communicating, but the increase in efficiency does decrease the more cells are added. This is because it is possible that a cell may measure a particle that another cell has already measured, making that new observation not true new information. This was repeated for long range communication between cells. In this scenario an optimisation problem arises for cells. One the one hand they want to be apart to minimise the double counting of the same particle, but they also want to be close together to minimise noise from their internal communication. This long range model may explain why some tissues like to be a particular distance apart. It should be mentioned that this particular spacing between cells may also be due to other factors.

Left: [1] phase plot of results in 3D

Lastly, they also set out to test concentration gradient sensing experiments. They had built a custom setup wherein organoids (multiple cell clusters) and single cells were placed in a container with a variable concentration gradient. Here, by setting the right steepness of gradient, it was observed that single cells could not sense a concentration gradient, whereas the organoids could. This could be explained by the cells communicating and thereby being able to surpass the limit that the single cells could not. They modelled this communication by a local excitation – global inhibition (LEGI) system. This revised model suggests that beyond a certain number of cells, there is no advantage in terms of sensory precision. The speaker illustrated this by a rather elegant analogy; the children’s game Telephone. As the signal passes around through various cells it gets more and more distorted by noise until eventually it no longer contains the original information.

Left: [2] Precision of gradient sensing with temporal integration      

I really enjoyed how this seminar covered various models related to cell sensing, starting at the very naive theoretical models and moving on to models that increase in complexity. Seeing how the rather naive theoretical could even predict some of the observed parameters in living systems really suggested these models could reflect some aspects of the systems. I think that studying the exchange of information between cells could teach us a lot about the rise of multicellularity and specialisation of cell types.

Besides the points mentioned above, seeing how various topics covered in the nanobiology course were used in current day research. We have extensively covered and even naively modelled bacterial chemotaxis, and we have studied fourier transforms and its applications. Our courses also covered key concepts related to this research such as diffusion, making this seminar a good example of the kind of research nanobiology students could consider doing during their research career.

[1] Fancher, Mugler, arXiv:1603.04108
[2] Mugler, Levchenko, Nemenman, PNAS, 2016



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s