Unstructured mRNAs form multivalent RNA-RNA interactions to generate TIS granule networks

Weirui Ma, Gang Zhen, Wei Xie, Christine Mayr

Preprint posted on February 14, 2020

Unstructured RNAs provide structure to phase separated TIS granule networks.

Selected by Christian Bates

Categories: molecular biology


Phase separated bodies, or biomolecular condensates, are a hot topic in current biology after being implicated in an array of diverse and essential cellular functions[1]. These subcellular structures represent regions containing high concentrations of specific biomolecules that exist in a distinct phase: often compared to the droplets that form when oil and water are mixed. Organizing the cell in this way allows transfer of components between the condensate and the surrounding cellular milieu, but not mixing. As such, it is thought that phase separated bodies function to provide order within the crowded cellular environment (Figure 1). 


Figure 1|A computer simulation of the crowded cellular cytoplasm of E.coli. Source: doi:10.7295/W9CIL28234. Reposted under CC-BY-NC-SA 3.0 from the cell image library.


In order to form, these complex phase-separated structures rely on multiple weak interactions, or multivalency. In proteins, this multivalency can be conferred through the presence of promiscuous structured domains capable of binding several partners, or, more commonly, poorly structured low complexity domains. The contributions made by proteins to phase-separation are relatively well understood, and a so-called ‘molecular grammar’ is being developed determining the contribution of specific amino acids to phase-separation capability[2]. Interestingly, in a cellular context these condensates frequently contain large amounts of nucleic acids, such as RNA. The specific RNAs present within phase separated compartments can massively impact the formation of these condensates, modifying key aspects such as viscosity, size and composition[3]. Moreover, testament to its role in phase separation, RNA can even phase separate in isolation. Yet, despite the clear functional importance of RNA in phase separated bodies, the precise properties of RNA that endow it with this ability are less clear and are a topic of great interest.


The TIS granule is a recently identified organelle, identified by Christine Mayr’s group in 2018 (Figure 2). This organelle intertwines with the endoplasmic reticulum, a classic and well studied example of a cellular organelle, to promote increased trafficking of specific proteins to the cell surface[4]. However, TIS granules differ from the classical view of organelles as membrane bound subregions of the cell, as it has no membrane and instead is formed by phase-separation. Like many other condensates, TIS granules contain large amounts of RNA, in particular AU-rich RNA. This study set out to understand the contribution made by this AU-rich RNA to TIS granule formation and structure.


Figure 2|Example of wild-type TIS11B network (Magenta) and ER membrane network (green). Generated from Figure 1 of the preprint. Made available under a CC-BY 4.0 International license.


Key findings

TIS granules are typified by the presence of the TIS11B protein, an RNA binding protein expressed in a wide range of tissues[4]. As such, the authors began by assessing the impact of mutating the RNA binding domains (RBDs) within TIS11B. Mutating these RBDs altered the TIS granule structure, switching from a mesh- to a blob-like structure. The authors then provide an elegant control, whereby the RBDs from TIS11B were switched for RBDs from a different RNA binding protein that also binds AU-rich RNA, HuR. With this TIS-HuR chimera, the TIS granules reformed into a mesh-like structure again. This suggested that the RNAs bound by these proteins are vital for the appropriate structural formation of the TIS granule.


Figure 3|mRNA structure underpins proper TIS granule formation. (A) Left: RNA structure prediction from RNA fold, the more red the structure, the more confident the structure prediction is. Right: dot plot showing that RNAs that form spherical TIS granules have a lower NED score (more rigid structure). (B) Top: diagrammatic overview of the RNA stapling method. Bottom: Data showing that when stapled together, the two RNAs generate a TIS granule with a meshwork structure. Figure generated using data from Figures 3 and 4 from the preprint. Made available under a CC-BY 4.0 International license.


As both TIS11B and HuR bind AU-rich RNA, the article went on to examine the impact of different RNA species on TIS granule structure. TIS11B alone formed spherical droplets in-vitro, a property that is commonly observed for phase separating proteins. Interestingly, the majority of phase-separated bodies appear as sphere-like structures. In this case then, the TIS granule appears to be the odd one out, by forming a meshwork structure instead. After adding the 3’-untranslated regions (3-’UTRs) from RNAs that are known to localise to TIS granules, TIS11B reverted to the mesh-like structure seen in-vivo. Thus, reaffirming the role of RNA in defining the structure of this condensate. Importantly, the authors controlled for the effects of RNA length and GC content, features that have been described previously to impact condensate formation[5,6].


Due to the fact that neither RNA length, or GC content appeared to impact TIS11B network formation, Ma and colleagues next looked at RNA structure using in-silico simulations to predict the probability of base-pairing within the RNA molecule. Ultimately, they arrived at a length-normalized score (NED) for the number of structures predicted for a molecule of RNA. This revealed that RNAs with a more defined structure (a lower NED score) are less likely to induce network formation, and vice-versa (Figure 3A). Perhaps then, unstructured regions of RNA are free to interact with neighbouring RNA molecules through Watson-Crick base-pairing, whereas structured regions have already saturated much of their base-pairing potential by binding to themselves in cis. This seemingly reflects the finding in proteins whereby proteins with disordered regions scaffold and promote phase separation due to their increased multivalency.


Using this information, the paper continued to investigate whether network formation indeed relied on trans-RNA interactions using a deft experiment. Here, the 3’-UTRs from two RNAs that do not normally induce network formation, TLR8 and MYC, were added to purified TIS11B in-vitro. These RNAs are highly structured and are therefore less likely to take part in trans-RNA interactions. Indeed, when added to TIS11B together, these RNAs did not induce network formation. However, when they were artificially ‘stapled’ together through the addition of ‘dimerisation elements’ (sequences of RNA that are the reverse complement of one another), the TIS11B protein displayed the mesh-like structure seen in-vivo. Therefore, under conditions driving trans-RNA interactions, TIS11B adopts to a mesh-like structure (Figure 3B).

Why this work is important

This work builds upon a previous publication from the same lab[4], which defined this novel TIS granule. To date, very little is known about this membraneless organelle, despite the fact that it seems to play a role in and augment important processes such as membrane trafficking. This pre-print looks at TIS granules at the molecular level, determining how the chemistry of biomolecular components drives the unique mesh-like structure that the TIS granule displays. By doing this, it provides not only a more detailed view of this poorly understood cellular compartment, but also a range of tools that can be used in-vivo to further interrogate the function of this granule. For example, it is now possible to use these findings to ask questions about what happens to the cell when the formation of this granule is impeded, or when specific RNAs are excluded from it, by altering RNA structure artificially.

Open questions

  1. Typically, it is thought that 80S ribosomes, and in particular actively translating ribosomes, are excluded from a range of phase-separated compartments (although a recent pre-print from Jeffrey Chao’s group contends this point[7]). It remains to be proven that translation occurs within this granule (although I may have missed something!); is there any direct evidence for translation at TIS granules?
  2. TIS granules are thought to be a gel-like condensate, which is typically attributed with slower diffusion rates and a more compact molecular structure. Do you think that if translation does occur within the TIS granule that its biophysical environment is likely to have an impact on the rate of translation?
  3. Is the mesh-like network that TIS granules display important for their function? Or is it possible that spherical TIS granules would be equally functional?
  4. Historically phase-separated compartments have been difficult to purify and assess with ‘Omic methods. However, recently a breakthrough in this field has occurred with several labs reporting the purification and interrogation of condensates such as stress granules and P-bodies. Given this, along with the gel-like nature of TIS granules, is it possible to pull down TIS11B and assess the transcriptome and proteome of these TIS granules?


  1. Banani, S. F., Lee, H. O., Hyman, A. A. & Rosen, M. K. Biomolecular condensates: Organizers of cellular biochemistry. Nat. Rev. Mol. Cell Biol. 18, 285–298 (2017).
  2. Wang, J. et al. A Molecular Grammar Governing the Driving Forces for Phase Separation of Prion-like RNA Binding Proteins. Cell 0, (2018).
  3. Boeynaems, S. et al. Spontaneous driving forces give rise to protein-RNA condensates with coexisting phases and complex material properties. Proc. Natl. Acad. Sci. U. S. A. 116, 7889–7898 (2019).
  4. Ma, W. & Mayr, C. A Membraneless Organelle Associated with the Endoplasmic Reticulum Enables 3’UTR-Mediated Protein-Protein Interactions. Cell175, 1492–1506.e19 (2018).
  5. Jain, A. & Vale, R. D. RNA phase transitions in repeat expansion disorders. Nature 546, 243–247 (2017).
  6. Courel, M. et al. GC content shapes mRNA storage and decay in human cells. Elife 8, (2019).
  7. Mateju, D., Eichenberger, B., Eglinger, J., Roth, G. & Chao, J. A. Single-molecule imaging reveals translation of mRNAs localized to stress granules. bioRxiv 2020.03.31.018093 (2020).


Tags: condensate, llps, phase separation, rna

Posted on: 8th April 2020 , updated on: 13th April 2020


Read preprint (No Ratings Yet)

Author's response

Weirui Ma shared

  1. Thanks for your interest in our work and your excellent summary.

    To address your questions:

    1. We think translation happens in the TIS granule-ER domain that we called TIGER domain. This is the surface of the endoplasmic reticulum (ER) covered by the TIS granules. We have some indirect evidence to support this. First of all, TIS granules localize to the perinuclear ER. The perinuclear ER is the rough ER and represents the major site of protein translation on the ER. Therefore, it is very likely that translation happens in the TIGER domain. Second, we performed live-cell imaging of CD47 protein, a membrane protein that is translated on the surface of the ER and trafficked to the plasma membrane. We indeed observed CD47 protein that localized to the surface of the ER that was associated with TIS granules. As ER-localized CD47 protein represents newly synthesized protein, our data suggests that translation happens in the TIGER domain.

    We currently have no idea whether translation also happens inside TIS granules. In the FRAP (fluorescence recovery after photobleaching) experiment, TIS11B only shows slow recovery after photobleaching. However, molecules enriched in TIS granules, for example the chaperone HSPA8, are still highly dynamic in TIS granules. So it is possible that lots of biochemical reactions, including translation, could indeed happen inside TIS granules.

    1. It is a nice point that the biophysical environment of the TIS granule has an impact on translation. We think TIS granules with their own biophysical and biochemical properties could provide a specific environment for biochemical reactions that happen in the granule. Moreover, TIS granules also create a special environment on the surface of the ER, which has an impact on the biochemical reactions that happen on the surface of the ER, including translation. Our lab is currently investigating what reactions are promoted in TIS granules.
    2. It is an open question whether the mesh-like shape of TIS granules is required for its functions. The major problem is that we currently know very little about TIS granules. We still don’t know what biochemical reactions happen inside the TIS granule and in the TIS granule-ER domain. We think if a reaction happens inside the TIS granule, the shape should not be critical for this reaction to occur. However, if a reaction happens on the interface between the ER and TIS granules, for example, co-translational protein complex formation, the mesh-like shape should be important. The mesh-like shape allows TIS granules to intertwine with the ER. This feature enables the two organelles to share a large amount of surface area to provide a place and an environment for biological reactions to happen. If TIS granules are sphere-like, they cannot intertwine with the ER, as demonstrated by the TIS11B RNA-binding domain mutants. As a result, the interface between the ER and TIS granules is largely gone, and thus the sphere-like TIS granule will not be fully functional. We previously showed that overexpression of wild-type TIS11B, which forms mesh-like TIS granules, is able to promote CD47 cell surface localization. However, TIS11B RNA-binding domain mutants, which form sphere-like granules have lost this ability. This piece of evidence suggests that the mesh-like shape is important for at least some of the functions of TIS granules.
    3. We have purified TIS granules and are currently analyzing their transcriptome and proteome.

    Thanks again for highlighting our work and your questions. I hope that soon we will have more answers.

    Best regards,

    Weirui Ma

Have your say

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Sign up to customise the site to your preferences and to receive alerts

Register here