Here’s a cool paper1:

Paper zhuang

This is a different subject to what I’m used to in terms of the biology, but I thought I might be able to communicate a bit of the super resolution imaging, because that’s my bag at the moment. I’ve included mostly images from the paper, no graphs – if you are interested in those, just google the paper: its a free one.

What’s the transcriptome? RNA and why we measure it

RNA stands for Ribonucleic acid. It is produced or ‘transcribed’ as an intermediate between gene (your DNA) and protein: the complex blob of ordered molecules that performs a specific function in a cell. In an average human cell there are trillions of different proteins.

The amount of a particular RNA coding for a particular protein is an important readout. Researchers can take this number and compare it between conditions: for example normal cells versus mutant cells. Mutant cells might be making less of a particular protein, and more of another – in this way we can take educated guesses at the intricate little pathway of signals that is being affected, and thus learn more about the mutants. The levels of all of the different RNA molecules in a cell is one way to define the ‘transcriptome’, but really its a pretty loose definition.

The problem is that more RNA doesn’t always equal more functional protein: the process is tightly controlled in many different ways.

One of the ways it is regulated is ‘spatially’, as in, where exactly it is in the cell that these little RNAs are located. Mostly we (researchers) study proteins spatially: often they contain targeting domains that say: “take me to the membrane” or “take me near the nucleus”. Mutation of this region of the protein leads to its continued function but lack of targeting. In the same way, mutation of the DNA (that makes the RNA) can alter the ‘post code’ region in the RNA, so it can no longer be directed to the right place in the cell.

When RNAs reach the right part of the cell, they are locally translated: turned into protein. As it turns out, you see local translation in useful places: RNAs coding for proteins that make pores – little tunnels – in the nuclear membrane are taken there during or before translation. RNAs that make membrane proteins are directed to the cell membrane.

Along with the RNAs themselves, there are the translation machinery: ribosomes, and the degradation machinery: degradosomes. These little proteins control the RNAs, and conventional fluorescent microscopy methods have provided researchers with the broad brush strokes: by simply looking at the light intensity given off by the fluorescently labelled proteins, we can see that they are non-uniformly distributed…

What is new and exciting at the moment is that we now have the capability, using clever microscopy techniques, to access the “nanoscale” (in part conceived by Xiaowei Zhuang, the lab this paper came from). At this level, a new universe of very precise, very subtle spatial mechanisms are enacted by proteins, that up until now were hidden behind the diffraction limit of visible light. Instead of seeing blurs of fluorescent intensity, we now get coordinate maps of precise molecular positions, with resolution of around 20 nm.

This is perfect for bacteria, which are generally much smaller than eukaryotic cells at about 5 um on average. RNAs are ridiculously small: an oligonucleotide (a length of RNA) is ~1 billionth of a metre, or 1 nanometre. With super resolution, the dynamics of clusters of these RNAs is accessible for study and quantitation.

Imaging RNA

To quantitate RNA in terms of its spatiality, the authors of this paper have taken a “global approach”, and looked at a large proportion of the bacteria’s total RNA. Their technique allows them to label RNAs that code for proteins of known function, and differentiate them on the nanoscale.

These RNAs were split into different ‘roles’ such as ‘adhesion’ or ‘proliferation’ or ‘movement’ and the targeting strands labelled with a different fluorophore for each one. In this way, they were able to carry out multi-colour super resolution microscopy on a representative ‘75%’ of the total RNA in a bacterial cell.

RNAs are targeted to the membrane through the SRP pathway, and while they are there they are actively being turned into protein
The basic finding was that RNAs encoding ‘inner membrane’ proteins are located near the membrane, whereas cytoplasmic and outer membrane protein coding RNAs are located all over the cell, fairly uniformly. Its certainly true that super resolution gave them access to this information, as the width of the damned beasties is only 1000 nm


Figure 1: Super resolution images of RNAs that code for inner membrane proteins (left) or cytoplasmic proteins (right). Adapted from1.

Secondly, they identified how these things are targeted to the membrane. In E.coli there are two RNA targeting protein pathways called SRP and SecB. All they did here, was to label a protein that everyone knows uses SRP (it doesn’t matter which protein), and one that definitely uses SecB, to see what happens. The super resolution images (figure 2) clearly show that SRP dependent proteins generally inhabit the same region as the inner membrane RNAs… So far so interesting.


Figure 2: Super resolution images of an SRP targeted protein, which goes to the inner membrane, and a SecB targeted protein: everywhere else. Adapted from 1

Next, they proved that the positions of these RNAs at the inner membrane are dependent on translation happening – ie that it is an active protein. To do this, they added a drug called ‘kasugamycin’, to stop translation into protein. Sure enough, when you do this, the RNAs lose their targeting and are simply everywhere – no longer localised to the inner membrane.


Figure 3: Targeted RNAs are detargeted when you stop translation with a drug called Kasagumycin. (compare this figure with 1C). This means that the RNAs targeted to the membrane are being actively translated into protein. Adapted from1

RNAs are nano-grouped with degradosomes

Finally, the researchers discovered that these inner membrane proteins are degraded faster than cytoplasmic ones*, and this may be because they are in close nanoscale proximity to the very machinery required for the breakdown!
*how they measure RNA decay is the topic for another article


Figure 4: Degradation machinery is also localised to the inner membrane region. RNase E, RhIB, PNPase and PAP I are all examples of degradors. Adapted from1

So what they are saying is that these RNAs are so tightly controlled that they are locally clustered right next to the thing that degrades them. Thus, these guys describe local and spatial control over membrane protein co-translation and insertion into the membrane by a nanospatial mechanism in tiny, enigmatic E. coli. Papers like these often end with a simple model, and here’s theirs:


Figure 5: These guys have concluded from this work that: CERTAIN RNAs are transported and actively translated to the membrane via SRP. Their transcription is controlled LOCALLY, by degradosome proteins in the same region. Adapted  from1

The reason I think this is super interesting is that it adds to the building body of knowledge that suggests that the especially labile fast moving parts of the cell (bacterial or otherwise) must be highly regulated in a local way on the nanoscale.

Ie, the events that control are so fast that all of the machinery, the little cogs, must be there already. Clusters of proteins must create their own nano-niche to control protein levels and function at a local level, in addition to controlling the slower coarser level of gene transcription in the nucleus.


The way the super resolution was done was to provide a global answer. Fair enough – the conclusions reached here are pretty sound. They could have gone deeper though (perhaps this will be the topic of their next paper). When you have that length scale of spatial detail, there are great algorithms that allow you to do unbiased identification and analysis of clusters, providing potentially more information about the way they are organised.2

One great thing about this paper was the high n numbers. Most super resolution work to date has been done on a few cells, because the images take a long time to analyse and the researchers were looking at a particular phenomenon that is well conserved, normally using a plasmid borne fluorescent protein attached to something of interest. Here, they looked at 511 bacteria, and used computational alignment of the well defined shape of the rod like organisms to aid their averaging, and hence the validity of their results.


  1. Moffitt, J. R., Pandey, S., Boettiger, A. N., Wang, S. & Zhuang, X. Spatial organization shapes the turnover of a bacterial transcriptome. Elife 5, (2016).
  2. Rubin-Delanchy, P. et al. Bayesian cluster identification in single-molecule localization microscopy data. Nat. Methods advance on, (2015).

Written by: Michael Shannon