Something that scientists would love to do is to look at a live working cell (ideally human, but we’d be happy with anything) in all its glorious detail, see all its proteins at work and moving, while being able to monitor the concentration of every substrate and metabolite.
Apart from the fact that you can’t get much detail while looking at the whole cell (see earlier blog post ‘How do we know’), by the time you have prepared the cell for an electron microscope (fixating, dehydrating and placing in a vacuum) you have also killed it three times over.
The best we can do while keeping the cell alive is light microscopy. The resolution limit of light microscopy means that you can only see details that are larger than 200 nm, i.e. medium sized organelles like mitochondria. There are a few tricks we can employ to circumvent this resolution limit by using FRET [1] and super resolution techniques [2] to measure small distances and observe the motion of individual proteins.
left: conventional light microscopy, right: super resolution microscopy [2]
We can get a lot of detail and scan across the whole cell by using an electron microscope, which gets down to a 1 nm resolution, i.e. less than the resolution required to see molecules (check out the living cell gallery). But unfortunately, the cell is no longer alive, and we can also not see it in colour, so for many of the specks we see on the image it’s anyone’s guess what they represent.
Thus, to see RNA polymerase copy information from the DNA to make RNA, or actin move along the cytoskeleton or even just lipid rafts floating across the surface, we need to come up with something else.
This is the point where computer modelling joins the team. We are already pretty good at knowing roughly what is going on. We have seen still images of ribosomes attached to mRNA and we have been able to work out the shape of many proteins, sometimes even in different conditions, using X-ray crystallography. We also know what many proteins do, which substrates they interact with and how they are activated. Feeding all this information to a computer and adding some laws of physics, how far can simulations get in filling in the gaps in our knowledge?
Computer simulations are already standard for modelling single aspects of a cell, like its chemotactic behaviour [3] or transcription [4]. These usually require information about the cell environment (e.g. chemical gradients) as well as a thorough knowledge of the function of the relevant internal mechanisms. These models are usually not ruled by physics, but by a phenomenological description of the interactions between substances, drawing on our knowledge of enzyme kinetics.
Many models focus on the cell as the most natural unit of biology, and from there go one scale out or in to include information about the cell environment or its neighbours or alternatively some relevant features of the internal state [5]. Technically, if we threw all of these models together, we should get pretty close to having a model of the whole cell. Unfortunately it is not that easy. Firstly, many of the aspects of the cell live are not covered by existing models and secondly, combining two computational models can be as difficult as mixing water and oil, especially if they were designed to represent different organisms or environments. Finally, each of the existing models carries a level of uncertainty, which when lumped together into one big model with nearly 2000 parameters [6] is just asking for trouble. The E-cell project has taken a more minimalistic approach, by incorporating only the 127 genes absolutely necessary for survival (which is still a lot) and allowing the user to add additional genes, proteins and metabolomic pathways where necessary [7].
Unfortunately for us, most of these models only produce graphs showing concentrations over time, rather than proteins interacting with each other as if we were watching a video. These kinds of things are done using molecular dynamics simulation, but they are much more computationally expensive, and so far unfeasible for the whole cell. As simulations become more efficient and computer power more available, we will hopefully soon be able to ‘see’ a ‘real’ live cell in action, but for now we will just have to live with artists’ impressions of the inner life of the cell.
References
all of the papers cited below are open access, so get going diving into more details, or just looking at the pretty pictures
feature image of Chloroplasts:
By Juan Carlos Fonseca Mata – Own work, CC BY-SA 4.0,
https://commons.wikimedia.org/w/index.php?curid=68616603
[1]
Sekar, Rajesh Babu, and Ammasi Periasamy. “Fluorescence resonance energy transfer (FRET) microscopy imaging of live cell protein localizations.” The Journal of cell biology 160.5 (2003): 629-633.
http://jcb.rupress.org/content/160/5/629/tab-pdf
[2] image of super resolution microscopy
Huang, Bo, Hazen Babcock, and Xiaowei Zhuang. “Breaking the diffraction barrier: super-resolution imaging of cells.” Cell143.7 (2010): 1047-1058.
https://www.sciencedirect.com/science/article/pii/S0092867410014200
[3]
Neilson, Matthew P., et al. “Chemotaxis: a feedback-based computational model robustly predicts multiple aspects of real cell behaviour.” PLoS biology 9.5 (2011): e1000618.
http://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.1000618
[4]
Bonneau, Richard, et al. “A predictive model for transcriptional control of physiology in a free living cell.” Cell 131.7 (2007): 1354-1365.
https://www.sciencedirect.com/science/article/pii/S009286740701416X
[5]
Walker, Dawn C., and Jennifer Southgate. “The virtual cell—a candidate co-ordinator for ‘middle-out’modelling of biological systems.” Briefings in bioinformatics 10.4 (2009): 450-461.
https://academic.oup.com/bib/article/10/4/450/297153
[6]
Karr, Jonathan R., et al. “A whole-cell computational model predicts phenotype from genotype.” Cell 150.2 (2012): 389-401.
https://www.sciencedirect.com/science/article/pii/S0092867412007763
[7]
Tomita, Masaru, et al. “E-CELL: software environment for whole-cell simulation.” Bioinformatics (Oxford, England) 15.1 (1999): 72-84.
https://academic.oup.com/bioinformatics/article/15/1/72/218375