Intercellular mRNA transfer through membrane nanotubes – behind the scenes.

My paper was recently published. I suggest that you read it before reading this post (it is an open access paper). In this paper we show that full-length mRNA molecules can be transferred between mammalian cells through membrane nanotube-like extensions that connect the cells.

This project was conceived as early as 7 years ago. I met Rob Singer at an EMBO workshop towards the end of 2010 and discussed the option of a postdoc position in his lab. At the time I was exposed to some literature on extracellular RNA and I thought it will be a cool idea to use Singer lab imaging approaches to visualize it and quantify. I was extremely excited by the possibility that neurons might send RNA via exosomes to other neurons at the synapse – thus increasing the complexity of synaptic plasticity. This idea found its way as a high-risk high-gain section in the research proposal that I submitted to postdoc fellowship applications. The idea was to label the mRNA and follow it by live imaging. I will also get quantitative data of mRNA transfer, such as number of molecules, kinetics etc.

The bulk of the proposal, by the way, was dedicated to developing single-molecule translation imaging tools and study local translation in neurons, which I didn’t do eventually (but others in the lab did: Jeff Chao’s TRICK ; Bin Wu’s SINAPS ; Adina Buxbaum’s in situ study of mRNA in neurons and Young Yoon’s paper on localized translation in neurons).

When I joined Rob’s lab, I wanted to start looking on mRNA in neurons. Rob suggested, justly, that we should start with a simpler system, like mouse embryonic fibroblasts (MEFs). He also suggested that I should start with beta-actin mRNA since we already had in the lab the MEFs with MS2-labeled beta-actin (MBS MEFs). Actually, we thought that beta-actin will be a good negative control – i.e. an example for an mRNA which doesn’t transfer, since it is so prevalent and why should cells share beta-actin mRNA? Thus, we thought this will be my “background” state to compare to other mRNAs.

So I co-cultured wild-type (WT) MEFs with MBS MEFs and performed my very first FISH.  Imagine my surprise when I got this result:

1st result

My first graph as I put it in my notebook, Oct 2012 [yes, it is presented badly. I’ve learned better since then]. MBS MEFs (donors) and WT MEFs were plated together and co-cultured for 24hr. FISH with MBS-specific probes was performed and images were analyzed  by in-house Matlab program written by Timothee Lionnet. X-axis – confluence of the co-culture (50%/50% in both cases); Y-axis – average number of cytoplasmic FISH spots in WT (acceptor) cells.

 So, beta-actin mRNA can transfer, and in nice numbers! In order to start understanding this phenomenon, I tried different things: primary cells (it works!), several types of stress (variable, and interesting, effect), mouse/human co-culture, time course, other mRNA species, etc.

The key experiment was the tripod experiment.  The theory that guided my work was that mRNA transfers through exosomes. But instead of isolating exosomes, a rather tedious experiment, I started with something simpler: I collected “conditioned” media – media from MBS MEF culture, centrifuged at low speed to get read of any large cell fragments and put it on the acceptor cells.  The result was that there is very slight, if any, mRNA in the media that cells can take in. I thought it was a problem of stability.

Adina Buxbaum,  then a PhD student at the lab, suggested that I form these tripods (initially we called them “Bankers” because it was taken from a classical book called “Culturing Nerve Cells” by Banker & Goslin). In this setup, the cells share media – so they can freely exchange materials by diffusion, but there is no direct contact between cells.


Illustration of the tripod setup.

The results surprised me – no mRNA transfer except in co-culture that maintains contact between cells. For almost two months I tried to think of a mechanism of mRNA transfer which isn’t through exosomes. Using the tripods, I also ruled-out apoptotic bodies. What could it be?

Then, I attended a departmental seminar, I don’t even remember what about. It was already the end of the talk and time for questions. I was tired and wsn’t paying much attention. Then, from behind me, someone* asked the speaker “Can this be related to membrane nanotubes?”

Something in my brain clicked. I rushed to the lab in excitement and searched pubmed, what are these nanotubes he mentioned?

After reading some papers, this seemed like a plausible mechanism of transfer. I found a paper in which nanotube formation was increased in fibronectin coated glass compared to poly-lysine. This was a simple experiment to do and the results were really amazing: reduced mRNA transfer with poly-lysine compared to fibronectin. I had to thank my good luck – I was doing all of my experiments from the beginning on fibronectin coated glass.

On the other hand…

The live imaging! The live imaging, using the MS2 system, was supposed to be the greatest way to look at mRNA transfer. To SEE a molecule move from one cell to another.

It should have been so simple. Alas, here I ran out of luck.

I would plate the cells, go to the microscope, see these wonderful nanotubes form. I would image one cell at a time, at a range of 1 minute to an hour; with intervals of 1sec to 1minute or more. I imaged and imaged and just could not understand – WHY CAN’T I SEE mRNA TRANSFER?

I had a bad feeling for a while, but only in early 2014 I actually did an experiment that proved it: the MCP-GFP protein, which binds the MS2 loops, inhibits mRNA transfer.


I put that image in my ppt presentation for GRC meeting I attended after I got that result. I stoped doing that.


There were other frustrating experiments. For instance, I wanted to measure the half-life of the transfered mRNA. Here’s a simple experiment: get GFP labeled acceptor cells. Co-culture with unlabled donor cells. Then harvest, sort GFP+ cells by FACS, re-plate and then FISH at different time points. Can you guess what happened?

Answer: All the transfered mRNA was gone by the first time point (30min). What happened? I found that simply trypsinizing the cells and immediately re-plating them – even without sorting – leads to massive loss of the transfered mRNA. I can only guess that becuase the cells change their shape (from flat to round) during harvest – followed by plating and re-flatening, there was also massive degradation of the transfred beta-actin mRNA.

The next best thing was to selectively kill donor cells after co-culture. What could go wrong?

Well, it seems that it is not that simple to selectively kill only donor cells in a reasonable short time. Translation inhibitors (e.g. puromycin, Diphtheria toxin) kills the cells really slow (>24hr). Not useful in our case. I tried chromophore-assisted light inactivation (CALI) with KillerRed. The cells didn’t die, even with CMV promoter and long exposures to light.

I then found a wonder drug: raptinal. It releases cytochrome c from mitochondria and bypasses half the apoptotic pathway – directly inducing APAF-1 which activates caspases. It can kill any mammalian cell (well, of those tested) in less then two hours. This is exactly what I needed! Moreover, APAF-1 knockout cells are resistant to the drug (for 1-hour treatment. More than that and they die by another pathway). So I could use APAF-1 k/o cells as acceptors. Hooray!

Well, not so much hooray. It seems that transfered beta-actin mRNA is sensitive to massive cytochrome c release (alternatively, raptinal also does something else in those cells which affected the transfered mRNA). Most of the transfered mRNA disappears within 30min of treatment. At least I could use this experiment to proved that apoptotic bodies contain mRNA but their contribution to mRNA transfer is very minor.

The last thing I wanted to mention is the part of Arjun Raj‘s lab. This was not a planned collaboration. It started when I gave a talk at the Transcription Imaging Consortium (TIC) meeting in HHMI Jannelia towards the end of 2014. It turns out that one of Arjun’s students heard my talk, and realized that I describe similar findings to what they have.

So Arjun contacted Rob and we discussed this. I was both happy that another lab confirmed my findings and scared of being scooped. But Arjun is a very nice guy.  We decided to either publish back-to-back or combine our data (which we eventually did).  He waited over two years(!) for me. This is not something to take for granted, and for that I am very thankful. The data that was provided by Arjun’s team (Chris, Maggie & Elliot) strengthened the paper. [Since PNAS didn’t allow me to do proper acknowledgments of their contributions: Chris did the work on GFP mRNA, including transwell & exosomes; Maggie & Elliot performed the human-mouse co-culture experiments for transfer of endogenous human mRNA].

Last, here’s a cool video of two cells connected by a nanotube. Look at the blob moving from one cell to another (alas, no mRNA there. You can see a donor MBS MEF at the bottom of the field).

Composite_cell_14_14-6-14 10fps

Movie at 10fps showing two acceptor cells connected by membrane nanotubes. At the bottom of the field – a MBS MEF cell. Red – membrane-targeted TagRFP-T. Green – tdMCP-GFP.  Images taken at 7sec intevals.


* It was Ben Ovryn, which later published a paper on imaging of membrane nanotubes. I wrote about it here.




One response to “Intercellular mRNA transfer through membrane nanotubes – behind the scenes.

  1. Pingback: Separating cells is hard | greenfluorescentblog

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 )

Google photo

You are commenting using your Google 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 )

Connecting to %s