RNA Self-Assembly Contributes to Stress Granule Formation and Defining the Stress Granule Transcriptome

RNA self-assembly contributes to stress granule formation and defining the stress granule transcriptome

Briana Van Treecka, David S. W. Prottera, Tyler Mathenya, Anthony Khonga,b, Christopher D. Linkc, and Roy Parkera,b,1

aDepartment of Chemistry and Biochemistry, University of Colorado Boulder, Boulder, CO 80309; bHoward Hughes Medical Institute, University of Colorado Boulder, Boulder, CO 80309; and cDepartment of Integrative Physiology, University of Colorado Boulder, Boulder, CO 80309

Contributed by Roy Parker, February 4, 2018 (sent for review January 2, 2018; reviewed by Paul Anderson and Imed E. Gallouzi)

Stress granules are higher order assemblies of nontranslating mRNAs find that various RNAs are capable of self-interactions, even in the and proteins that form when translation initiation is inhibited. Stress absence of Watson–Crick base pairing. More importantly, we granules are thought to form by protein–protein interactions of RNA- show that mixtures of cellular RNAs partition into self-assemblies binding proteins. We demonstrate RNA homopolymers or purified in vitro under physiologically relevant conditions and that the cellular RNA forms assemblies in vitro analogous to stress gran- RNAs enriched in these assemblies reflect the RNAs enriched in ules. Remarkably, under conditions representative of an intracel- stress granules in vivo. This argues that RNA–RNA interactions lular stress response, the mRNAs enriched in assemblies from total contribute to both stress granule assembly and to the determination of yeast RNA largely recapitulate the stress granule transcriptome. which RNAs are enriched in stress granules. Strikingly, we also ob- We suggest stress granules are formed by a summation of pro- served that RNA self-assembly is promoted by the pathogenic di- tein–protein and RNA–RNA interactions, with RNA self-assembly peptide repeats GR and PR. This work highlights that RNPs exchange likely to contribute to other RNP assemblies wherever there is a high between monomeric and multimeric states within the cell and that the local concentration of RNA. RNA assembly in vitro is also increased by cell utilizes a variety of mechanisms to regulate this equilibrium. GR and PR dipeptide repeats, which are known to increase stress Results granule formation in cells. Since GR and PR dipeptides are involved – in neurodegenerative diseases, this suggests that perturbations in- Four points led us to hypothesize that RNA RNA interactions in creasing RNA–RNA assembly in cells could lead to disease. trans might contribute to stress granule formation. First, stress granules form when there is ribosome run-off, exposing the previously ribosome-occupied coding regions that would be expected RNP granules | dipeptides | RNA self-assembly | stress granules to form RNA–RNA interactions both in cis and in trans.Second, long mRNAs partition highly into stress granules but only show a ibonucleoprotein (RNP) granules are eukaryotic non– modest increase in the binding sites of stress granule proteins, R membrane-bound organelles composed of RNA and protein. suggesting length might contribute to the partitioning of mRNAs RNP granules are located in the cytoplasm or the nucleus and into stress granules through trans-RNA–RNA interactions (28). include P-bodies, germ granules, neuronal granules, paraspeckles, Third, stress granule cores in lysates are resistant to high salt or Cajal bodies, and stress granules (1–6). Stress granules are cytoplasmic assemblies of nontranslating mRNAs and RNA-binding Significance proteins that form during stress responses where bulk translation initiation is inhibited. Stress granules, which are ubiquitous, non–membrane-bound Stress granules are of interest for four reasons. First, stress gran- assemblies of protein and RNA, form when translation initia- ules are thought to play a role in the stress response and gene regulation (7, 8). Second, related RNP granules exist in neurons tion is inhibited, contribute to the regulation of gene expression, and can affect synaptic plasticity (9, 10). Third, mutations in RNA- and are implicated in the pathologies of cancer and neurode- binding proteins, or stress granule-remodeling complexes, which generative disease. Understanding the mechanisms of stress lead to constitutive or increased stress granule formation, are granule assembly is crucial to gaining greater insight into their causative in amyotrophic lateral sclerosis, frontotemporal lobar biological function and pathological misregulation. We provide degeneration, inclusion body myopathy, and other degenerative evidence that RNA–RNA interactions contribute to the assembly disorders (11–13). Fourth, stress granule formation can influence of stress granules. Furthermore, we show that pathogenic di- both tumor progression and viral infection (14–17). peptides increase the propensity of RNA to assemble. Together, Stress granules, as well as other RNP granules, are thought to this argues that RNAs are assembly prone and must be carefully form by RNA providing a scaffold for RNA-binding proteins that regulated. A summative model of stress granule assembly, which form protein–protein interactions in trans to drive assembly. For includes trans-RNA–RNA interactions, can be extended to other example, mammalian stress granule assembly is promoted by inter- ribonucleoprotein granules in the cell. actions of the TIA1, G3BP1, and ATXN2 proteins (18–20). Simi- larly, assembly of yeast P-bodies is promoted by the dimerization of Author contributions: B.V.T., D.S.W.P., A.K., and R.P. designed research; B.V.T., D.S.W.P., the RNA-binding protein Edc3, as well as an intrinsically disordered and C.D.L. performed research; B.V.T., D.S.W.P., T.M., and A.K. analyzed data; and B.V.T. region (IDR) on Lsm4p (21). Consistent with interactions between and R.P. wrote the paper. RNA-binding proteins promoting RNP granule assembly, several Reviewers: P.A., Brigham and Women’s Hospital; and I.E.G., McGill University. purified RNA-binding proteins, in isolation or with RNA, can form The authors declare no conflict of interest. self-assemblies in vitro as coacervates, fibers, or hydrogels, although Published under the PNAS license. the specific relationship of these in vitro assemblies to RNP granules Data deposition: The data reported in this paper have been deposited in the Gene Ex- remains unclear (22–27). The role of trans-RNA–RNA interactions pression Omnibus (GEO) database, https://www.ncbi.nlm.nih.gov/geo (accession no. in stress granule formation remains relatively unexplored. GSE99170). Herein, we illustrate that RNA–RNA interactions contribute 1To whom correspondence should be addressed. Email: [email protected]. to the assembly of stress granules and the targeting of RNAs to This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. stress granules. We show that trans-RNA–RNA interactions play 1073/pnas.1800038115/-/DCSupplemental. a role in the recruitment of given RNAs to stress granules. We Published online February 26, 2018.

2734–2739 | PNAS | March 13, 2018 | vol. 115 | no. 11 www.pnas.org/cgi/doi/10.1073/pnas.1800038115 Downloaded by guest on October 3, 2021 aliphatic alcohols, which are known to disrupt many protein–protein, AB but not RNA–RNA, interactions (29). Stress granule cores have ncRNAs (antisense excluded) Antisense ncRNAs been suggested to be independent of RNA based on their resistance 30 to RNase treatment (29), but a reexamination of this observation 300 showed that extensive RNase treatment fails to degrade the RNA within stress granule cores (Fig. S1). Fourth, trans-RNA–RNA in- 20 teractions contribute to the assembly of multimeric RNPs in Dro- 200 sophila embryos, as well as RNA foci containing transcripts with 10 repeat expansions (30, 31). 100 Additional evidence that RNA–RNA interactions could contribute to stress granule assembly came from the partitioning of Frequency (counts) Frequency (counts) noncoding RNAs (ncRNAs) into stress granules in mammalian 0 0 -10 -5 0 5 10 -10 -5 0 5 10 cells (28). By data-mining the mammalian stress granule tran- Log (Fold Change) Log (Fold Change) scriptome (28), we observed that ncRNAs shorter than 3,000 bases 22 were generally depleted from stress granules, as predicted by length C All mRNAs D Enriched asRNA E Depleted asRNA (Fig. 1A). However, antisense ncRNAs shorter than 3,000 bases binding partner localization binding partner localization showed a bimodal distribution, with one population enriched in 1 stress granules and one depleted (Fig. 1B). Of the 14 antisense 1780 1626 4 ncRNAs significantly enriched greater than twofold in stress gran- 10 ules and with fragments per kilobase of transcript per million p=0.00045 10* 12 mapped reads (FPKM) values above 1 in total RNA samples, 7789 p<0.0001 10 were antisense to long mRNAs that were enriched in stress granules (Fig. 1D), five of which are illustrated (Fig. 1F). This is Enriched (SG) *1 has a secondary binding striking, since only 14.5% of mRNAs in the cell are enriched in partner that is depleted stress granules (Fig. 1C). In contrast, of the 23 antisense ncRNAs Neither (SG) significantly depleted greater than twofold from stress granules and Depleted (SG) Transcript SG Length with FPKM values above 1 in total RNA samples, only one had an Enrichment (bases) antisensepartnerthatwasenrichedinstressgranules(Fig.1E). Thus, F there is a correlation between the localization of antisense ncRNAs PEG3 5.5 8375 and their ability to base-pair to longer mRNAs, implying that RNA– PEG3-AS1 7.7 1314 RNA interactions may partition specific RNAs into stress granules. – MDM2 2.9 12633 To determine if RNA RNA interactions could contribute to RP11-611O2.5 4.5 785 stress granule formation, we calculated the approximate concentration of the coding region of mRNAs exposed during polysome col- IRF2BP2 4.1 4663 lapse. For yeast, we estimated the concentration of exposed coding RP4-781K5.2 3.6 693 μ regions to be between 170 and 800 g/mL (Materials and Methods), DYRK2 2.1 8912 while in the mammalian U-2 OS cell line, the exposed ORFs were RP11-335O4.3 4.6 777 estimated at ∼180 μg/mL (Materials and Methods). If RNA–RNA interactions contribute to stress granule formation, then we pre- PLCD4 3.7 4350 dicted that RNA at these concentrations would spontaneously as- RP11-548H3.1 6.4 759 semble under conditions mimicking the intracellular milieu. Fig. 1. RNA–RNA base-pairing influences RNA localization. All ncRNAs To test this prediction, we assessed whether purified protein- (excluding antisense) (A) and antisense ncRNAs (B) with lengths <3,000 nt free total RNA from Saccharomyces cerevisiae at 150 μg/mL would and significant sequencing reads were plotted in a histogram showing their

form assemblies under conditions where we varied the salt and log2(fold change) enrichment in stress granule/total RNA. A red bracket used PEG to mimic molecular crowding. We observed that total highlights the second peak for RNAs enriched in stress granules. (C) Pie chart BIOCHEMISTRY yeast RNA readily self-assembled and formed two types of as- illustrating the proportion of all mammalian mRNAs enriched (red), de- semblies (Fig. 2A). At higher PEG and lower salt, the RNA was pleted (blue), or neither (gray) in stress granules. Stress granule localization observed to form small droplets (Fig. 2 A, a). With increasing salt, of binding partners to enriched (D) and depleted (E) antisense ncRNAs is we observed the formation of more amorphous assemblies that shown. *One enriched antisense ncRNA transcript has a secondary binding contained a larger percentage of the RNA, which, due to their partner to a depleted mRNA. (F) Examples of enriched antisense ncRNAs and morphology, we refer to as RNA tangles (Fig. 2 A, b and c and Fig. their binding partners. SG, stress granule. S2 A and B). Based on fluorescence recovery after photobleaching (FRAP) of spiked-in fluorescent RNAs or dilution into lower ionic strength, droplets were more dynamic and less stable than RNA to GFP, the IDRs of yeast Lsm4 and eIF4GII, both of which bind tangles (Fig. S2 C and D). Specifically, 97% of the RNA in tangles RNA (24), are recruited to RNA droplets more than the IDR of formed at high salt is immobile (Fig. S2C), and dilution takes FUS or GFP alone (Fig. S3 A and B). Similarly, RNA droplets nearly an hour to disrupt tangles, whereas droplets disperse on the recruit the RNA-binding protein hnRNPA1 (Fig. S3C). Thus, order of seconds (Fig. S2D). Importantly, both droplets and tan- assemblies formed by RNA interactions in cells would be expected gles were enriched for the RNA-specific dye SYTO RNASelect to recruit RNA-binding proteins. – (Fig. S2E) and depleted for PEG (Fig. S2F). This is consistent with RNA RNA interactions contributing to RNA self-assemblies PEG functioning primarily as a crowding agent, which is further could be helical stacking (33), specific base-pairing (31), or supported by Ficoll-promoting RNA self-assembly (Fig. S2G). promiscuous interactions between mRNA involving both tradi- Moreover, in the absence of crowding agents, RNA also self- tional Watson–Crick interactions and additional interactions assembled under physiological concentrations of salt (150 mM) (34). Evidence that non-Watson–Crick interactions can promote and the polyamines spermine (223 μM) and spermidine (1,339 μM), RNA self-assembly is that all four homopolymers self-assemble, whichareknowntostabilizeRNA–RNA interactions (32) (Fig. 2B). with polyU forming rapidly relaxing droplets [as previously ob- Thus, under in vitro conditions analogous to the cytosol during a served (32)], polyC forming slower relaxing droplets, polyA stress response, purified RNA undergoes self-assembly. forming asymmetrical assemblies with very slow relaxation rates, Stress granules recruit a diversity of RNA-binding proteins in and polyG forming an aggregate that is presumably based on addition to RNA. Similarly, we observed that RNA self-assemblies G-quadruplexes (Fig. 2C and Fig. S4). Thus, RNA sequences recruited RNA-binding proteins (Fig. S3). Specifically, when fused may impart biophysical characteristics to their assemblies, but,

Van Treeck et al. PNAS | March 13, 2018 | vol. 115 | no. 11 | 2735 Downloaded by guest on October 3, 2021 ABYeast total RNA Total Yeast RNA Phase Diagram 12 ab a c 10

8 b Nothing - polyamines 6 Droplets c d In between PEG (%) 4 d Tangles 2 Fig. 2. Various RNAs self-assemble in vitro. (A) Phase diagram of RNA assembly morphology under 0 + polyamines 0 200 400 600 800 varying PEG and NaCl concentrations. Images corre- C NaCl (mM) 10 µm 5µm spond to labeled positions in the phase diagram: polyU polyC polyA polyG Droplets formed at 0 mM NaCl, 10% PEG (a); droplet/ tangles formed at 300 mM NaCl, 7.5% PEG (b); tangles formed at 750 mM NaCl, 10% PEG (c); and no assemblies at 300 mM NaCl, 2.5% PEG (d). All con-

ditions contain 1 mM MgCl2 and 150 μg/mL yeast total RNA. (B) Total yeast RNA in 150 mM NaCl and

1 mM MgCl2 with and without physiologically relevant conditions of spermine (223 μM) and spermidine (1,339 μM). (C) polyU, polyC, polyA, and polyG self- partition in vitro at 500 μg/mL (respective) homo- 10µm polymers, 10% PEG, and 750 mM NaCl.

more importantly, RNA self-assembly is a general property of affect mRNAs partitioning into stress granules in vivo but not to diverse RNAs and is not restricted to Watson–Crick interactions. affect RNA self-assembly in vitro, where there is no translational If the self-assembly of RNA in vitro is relevant to stress granule apparatus. Taken together, these observations suggest that the assembly, we predicted that similar mRNAs would assemble into partitioning of mRNAs into stress granules is modulated, in part, RNAdropletsinvitroandstressgranulesinvivo.Totestthis by the self-assembly capabilities of RNA. prediction, we purified RNA droplets formed under physiological We suggest a working model where stress granules form when salt (150 mM NaCl) and PEG from total yeast RNA in triplicate, the summation of protein–protein, protein–RNA, and RNA–RNA sequenced the assembled RNA, and compared that RNA pop- interactions increase over a threshold for assembly. This model can ulation with the mRNAs that partition into stress granules (28). be illustrated in a phase diagram, with transitions between “phases” This experiment revealed several important observations. explaining stress granule formation (Fig. 4). For example, increasing First, triplicates of in vitro assemblies and total RNA agreed the concentration of exposed RNA, either by inhibiting translation within themselves but showed clear differences (Fig. S5A), allowing initiation leading to ribosome run-off or by transfection or injection the identification of both enriched and depleted RNAs in the in of RNA into cells (35–37), could cause a transition in the cell that vitro assemblies (Fig. 3A). RNAs significantly (P < 0.01) and two- leads toward stress granule assembly via increased RNA–RNA in- fold enriched and depleted under these conditions will be referred teractions (Fig. 4, yellow arrow). Similarly, deletion of G3BP1 and to throughout the remainder of this paper as in vitro-enriched G3BP2, which are abundant proteins that contribute to stress and -depleted, respectively. On average, the in vitro-enriched RNAs granule assembly (19), prevents assembly under most stresses and were significantly longer, while the in vitro-depleted RNAs were moves components back into the nonassembled phase in our model shorter (Fig. 3B). This closely mirrored the length effect seen in the (Fig. 4, red arrow). Conversely, overexpression of TIA1, G3BP1, or analysis of the yeast stress granule transcriptome (28) (Fig. 3C). FMR1, all of which can contribute to stress granule assembly (18, A second key observation was a strong overlap between in vitro- 38–40), may drive the cell into a regime of assembled stress granules and stress granule-enriched mRNAs (Fig. 3D and Fig. S5B). by increasing the protein–protein interactions (Fig. 4, black arrow). Specifically, of the 916 mRNAs enriched in yeast stress granules Since RNA–RNA interactions contribute to stress granule as- (28), 634 are enriched in vitro. Conversely, of the 1,111 mRNAs sembly, we hypothesized that any small molecule or peptide that depleted from yeast stress granules (28), only 56 are enriched stabilizes RNA–RNA interactions would promote stress granule in vitro. This correlation extends to RNAs depleted from in vitro assembly. Strikingly, prior work has shown that arginine-containing assemblies and stress granules in vivo (Fig. 3E and Fig. S5C). Of dipeptides GR and PR produced by RAN translation of hex- the 1,456 mRNAs depleted from in vitro assemblies, only 10 are anucleotide (G4C2) repeat expansions of C9orf72 (41, 42) are both enriched in stress granules, and of the 1,111 mRNAs depleted toxic to cells and trigger stress granule formation (43, 44). Al- from stress granules, 573 are also depleted from the RNA self- though these dipeptides perturb many cellular processes (45–47), assembly in vitro. This indicates that the biophysical properties of we predicted they may directly induce stress granules by stabilizing RNA that drive RNA self-assembly in vitro correlate with a critical RNA–RNA interactions. Indeed, we observed that (PR)10 and mRNA feature for partitioning mRNAs into stress granules in vivo. (GR)10 robustly stimulated RNA self-assembly in vitro, while (GP)10 Plotting the relative enrichment of RNAs in in vitro assemblies had little effect (Fig. 5 and Fig. S6). Assemblies with (PR)10 or vs. in stress granules shows a correlation in the degree of enrich- (GR)10 were enriched for both RNA and dipeptides (Fig. 5A), and ment in both cases, with a Pearson’s correlation of 0.53 (Fig. 3F). neither RNA nor protein alone was sufficient for robust assembly In general, RNAs more enriched in stress granules are also more under these conditions (Fig. 5A and Fig. S6A). Since the most toxic enriched in assemblies in vitro, and vice versa. The RNAs that dipeptides are the same dipeptides that stimulate the assembly of show less correlation between in vivo and in vitro recruitment can RNA in vitro, we suggest that dipeptides may exert some of their be explained by the effects of translation efficiency on mRNA re- toxic effects by promoting RNA–RNA assemblies in the cell, such as cruitment into stress granules, where efficient translation has been through the formation of stress granules. shown to correlate with depletion from stress granules (28). Re- moval of transcripts with fractions of optimal codons below 0.4 or Discussion above 0.6 increases the Pearson’s correlation coefficient to 0.61 We present several lines of evidence that stress granules form, in (Fig. S5E). This makes sense, as translation would be expected to part, by RNA–RNA interactions and that those interactions can

2736 | www.pnas.org/cgi/doi/10.1073/pnas.1800038115 Van Treeck et al. Downloaded by guest on October 3, 2021 Protein dominated assembly Stress granules Assembly driven by protein-protein and RNA-RNA interactions ABCYeast Assemblies stress granules Overexpress 105 in vitro (Khong et al., 2017) multivalent RBPs *** *** 4 16 16 10 10 10 No stress Stress 4 4 granules granules 103 4 4 free RNA 102 2 2 Depletion of

assemblies (reads) Enriched 101 key RBPs

Neither protein-protein interactions RNA length (kb) RNA length (kb) 0 Depleted 0 0 RNA dominated 10 No assembly In vitro assembly No stress 100 101 102 103 104 105 granules

Total RNA (reads) Neither Neither Enriched Enriched Depleted Depleted RNA-RNA interactions D In vitro assembly enriched (150 mM NaCl)

Enriched Enriched (SG) Enriched Depleted (SG) Fig. 4. Four-phase model of stress granule assembly. Model illustrating how n=1488 n= 916 n=1488 n=1111 assemblies may be RNA-dominated (Bottom Right) or protein-dominated (Top Left); however, a combination of interactions is often responsible for assembly within cells (Top Right). Arrows denote examples from the literature that lead to stress granule formation or dissolution. The yellow arrow 854 634 282 1432 1055 shows the formation of stress granules through a large influx of non- 56 translating RNAs. The red arrow signifies a lack of stress granules when key proteins are deleted. The black arrow denotes the formation of stress -15 granules through overexpression of certain RNA-binding proteins. Overlap: p<1x10 Overlap: n.s. -15 Lack of overlap: n.s. Lack of overlap: p<1x10 E In vitro assembly depleted (150 mM NaCl) influence the partitioning of mRNA. First, we show that a por- Depleted Enriched (SG) Depleted Depleted (SG) tion of total yeast RNA effectively self-assembles in vitro under n= 1456 n= 916 n= 1456 n=1111 conditions mimicking intracellular stress conditions (Fig. 2). Im- portantly, this self-assembly of yeast RNA largely reproduces the stress granule transcriptome (Fig. 3). Moreover, the enrichment of short antisense ncRNAs in mammalian stress granules correlated 1446 906 883 573 538 10 with their ability to base-pair to a longer mRNA enriched in stress granules (Fig. 1). Finally, dipeptide repeats, which are known to induce stress granules in cells, strongly increase RNA self-assembly Overlap: n.s. Overlap: p<1x10-15 in vitro (Fig. 5). In combination with genetic experiments showing -15 Lack of overlap: p<1x10 Lack of overlap: n.s. proteins can enhance stress granule assembly, we conclude that F stress granules assemble by a summation of protein–RNA, protein– 4 protein, and RNA–RNA interactions (Fig. 4). The precise set of R=0.53 interactions that drive stress granule formation can vary under dif- 3 ferent conditions as long as the total summation is above the critical 2 threshold for assembly. assemblies – 1 There is growing evidence that RNA RNA interactions can BIOCHEMISTRY help drive the assembly of stress granules. A role for RNA–RNA in vitro 0 interactions could explain why purified stress granule cores are -1 stable against many insults (29). In addition, a role for RNA–RNA Codon interactions could suggest why the ATPase activity of Ded1p is -2 Optimality required to disassemble stress granules (48). Our results also high- -3 0.30 0.80 light the importance of charge shielding in promoting RNA self- 2 assembly, as increasing salt concentrations or adding positively

Log (fold change) -4 charged molecules, such as polyamines or arginine-containing -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 dipeptides, greatly increases assembly. This is consistent with ob- Log (fold change) in stress granules 2 servations in the literature suggesting that ionic strength, which has a strong impact on RNA self-assembly in vitro (Fig. 2 and Fig. Fig. 3. RNAs in self-assemblies in vitro largely recapitulate the stress gran- S2), could be an important regulator of stress granule formation. ule transcriptome. (A) RNAs from in vitro assemblies formed at 150 mM NaCl Stress granule assembly can be triggered by hypertonic shock and, identified as significantly (P < 0.01) and twofold enriched (1,488 RNAs, red conversely, can fail to form in conditions of low intracellular os- dots) and depleted (1,456 RNAs, blue dots) compared to total yeast RNA. The molality (49). Moreover, sorbitol, another osmotic stressor, can box plots show the correlation between transcript length and RNA locali- ΔΔ < partially rescue formation of stress granules in G3BP1/2 cells zation to in vitro assemblies (B) and yeast stress granules (C) (28). ***P (19), perhaps because the sorbitol increases the intracellular os- 0.001 between any three box plots. (D) In vitro-enriched mRNAs significantly – overlap with mRNAs identified in the yeast stress granule transcriptome and motic strength, thereby stabilizing RNA RNA interactions. In exhibit significant lack of overlap with RNAs depleted from stress granules. another line of evidence, the addition of G4C2 RNA to U-2 OS n.s., not significant. (E) In vitro-depleted RNAs significantly overlap with cell lysates condenses an assembly containing many stress granule RNAs depleted from stress granules and exhibit a significant lack of overlap components and cellular RNAs, perhaps by nucleating interac- with RNAs identified in the yeast stress granule transcriptome. (F) Degree of tions between various mRNAs in the lysates (37). Finally, the enrichment and depletion between in vitro assemblies and stress granules prevalence of RNA–RNA interactions helps explain why injection correlates (Pearson’s correlation, R = 0.53). RNAs that are more enriched or transfection of concentrated RNA into cells leads to the for- in vitro than in vivo (Upper Left) tend to have greater proportions of optimal mation of large, higher order assemblies (35–37).Thisisanalo- codons (orange). gous to the huge influx of exposed RNAs during a stress response,

Van Treeck et al. PNAS | March 13, 2018 | vol. 115 | no. 11 | 2737 Downloaded by guest on October 3, 2021 specificity to assemblies, as has been seen with Oscar and Bicoid A Bright Field Bright Field Dipeptides SYTO 17 mRNAs in Drosophila embryos (30). Although promiscuous RNA–RNA interactions are capable of

10 forming between any two RNAs, it is important to note that there will be a gradient of any given RNA’s propensity to assemble. We

(GR) show that longer RNAs are more enriched in assemblies in vitro and in stress granules (Fig. 3). In addition, one anticipates that the No dipeptides ability of RNAs to form base-pairing interactions will increase their self-assembly properties. Given these two inputs, short-structured RNAs, such as tRNAs (originally “soluble” RNAs), will be excluded 10 10 from RNA assemblies. Alternatively, longer RNAs, particularly – (PR) (GP) those with repeat sequences capable of self base-pairing, will be highly efficient at self-assembly. Recent reports have illustrated that longer repeat RNAs more effectively form intracellular and in vitro 5 RNA assemblies (31) and that transfection of G-quadruplex–capable B Phase Diagram (GR)10 and RNA G4C2 RNA elicits robust stress granule assembly, whereas its an- 160 tisense counterpart C4G2 does not (37). In this light, it is notable 140 that many repeat-containing RNAs, including both pathogenic toxic RNAs and satellite RNAs, form specific nuclear foci (43, 51– 120 53). This suggests that formation of hyperstable RNA assemblies in cells is toxic, and this provides a possible explanation for why re- 100 peat RNAs with a strong tendency to base-pair cause disease once 80 they are expanded beyond a certain length. This work suggests that RNPs exist within cells at an equilibrium 60 between monomeric RNPs and multimeric RNP granules that is [RNA] influenced by many parameters. For example, RNA helicases, ri- 40 bosome association, and monovalent RNA-binding proteins are expected to play a role in maintaining RNPs in the monomeric 20 state. Consistent with this view, it is known that depletion of the 0 abundant Tdp-43 ortholog in Caenorhabditis elegans leads to the 01020 30 40 50 accumulation of dsRNA foci in the nucleus (54). In contrast, longer [Dipeptides] ( RNA lengths, high local or transient RNA concentrations, and the propensity of RNA to interact with itself will promote RNA–RNA Legend association. Taken together, we suggest that RNA–RNA assemblies No assembly Moderate assembly* Protein only in cells may be a default state and that cells prevent this RNA aggregation by active means, ribosomes, and RNA-binding proteins. Sparse assembly Robust assembly Protein and RNA Materials and Methods Fig. 5. Pathogenic dipeptides increase RNA assembly. (A) Dipeptides (GR)10 Materials. Homopolymer RNAs were purchased from Amersham Pharmacia Bio- and (PR)10 promote assembly, while (GP)10 does not. Fluorescent dipeptides tech, Inc. (polyA, polyC, polyU, and polyG) and Sigma (polyU). Fluorescently labeled (green) and RNA (SYTO 17, red) are both enriched in assemblies. (B)Phase RNA containing consensus sequences for polypyrimidine tract-binding protein (PTB diagram illustrating the assembly of (GR) and RNA. The squares signify lack 10 RNA) (24) and short homopolymer oligos were purchased from IDT (Dataset S1). of assembly. +,sparseandsmallassembly;++, moderate assemblies (consti- Dipeptides were ordered through New England Peptide (Dataset S1). tuting either frequent but smaller assemblies or larger assemblies that were more sparse); +++, robust assembly. Green indicates protein-only assembly. Antisense RNA Analysis. Briefly, ncRNAs from the mammalian stress granule transcriptome (28) were split into those defined as antisense and all other ncRNAs (55). Binding partners were found on the UCSC Genome Browser, which may allow for the emergence of interactions that are nor- and their enrichment in stress granule was graphed. mally outcompeted by more specific, high-affinity interactions. The role of RNA–RNA interactions in stress granule assembly – Preparation of Stress Granule-Enriched Fraction and RNase Treatment. Stress gran- has two broader implications. First, one anticipates that RNA ules were enriched from BY4741 cells transformed with Ded1Δ141–150 as de- RNA interactions will contribute to other RNP granules. Indeed, scribed previously (29). RNase mixture (AM2286; Ambion) or RNase III (M0245L; given the thermodynamic strength of RNA–RNA interactions, they New England Biolabs) was added to the granule-enriched fraction as per the should be expected to be in a stable state in cells and will form in relevant manufacturer’s directions and placed at 37 °C for 2 h before imaging on trans whenever there is a sufficiently high local concentration of a microscope. Additional details are provided in Supporting Information. RNA. For example, we suggest that very efficient transcription of long RNAs would be expected to drive the formation of RNA– In Vitro Assembly and Dynamics. Specific information regarding in vitro as- RNA interactions between nascent transcripts. One possible ex- sembly formation (total RNA, homopolymer, RNA-binding protein re- ample of this process would be the assembly of paraspeckles, which cruitment, and dipeptide/RNA assemblies) as well as dynamics (relaxation form at sites of transcription of NEAT1 RNA and contain multiple times, FRAP, and dilution) can be found in Supporting Information. NEAT1 copies (50). A second implication is that RNA–RNA interactions can be Microscopy of RNA Assemblies. Mixtures were placed in 96-well glass-bottomed plates with high-performance no. 1.5 cover glass (Fisher Scientific). Images were promiscuous and form between any two RNAs with single-stranded × regions. For example, by chance, the average mammalian mRNA in acquired on a DeltaVision epifluorescence microscope with a 100 objective (Applied Biosystems) equipped with a sCMOS camera. Images for FRAP analysis stress granules (7.5 kb) (28) should have over 300 possible sites of were acquired using a Nikon A1R laser scanning confocal microscope. six consecutive base pairs with another 7.5 kb of mRNA. Even if the vast majority of these sites are lost to intramolecular RNA – Total RNA Pelleting for Tape Station and Sequencing Analysis. Total yeast RNA folding or RNA-binding proteins, numerous possible sites for RNA was mixed with 150 mM NaCl and 10% PEG, and then pelleted for se- RNA interaction will remain. Thus, one anticipates that RNA–RNA quencing. RNA pellets and total RNA were treated with a DNA-free DNA interactions can arise between many different RNAs. In some removal kit (Thermo Fisher Scientific) and sent to the University of Colorado cases, evolution will have created definitive interaction sites to give BioFrontiers Institute Next-Gen Sequencing Core Facility for Ribo-Zero

2738 | www.pnas.org/cgi/doi/10.1073/pnas.1800038115 Van Treeck et al. Downloaded by guest on October 3, 2021 treatment, library construction, and NextSeq run. Read quality was assessed Numerical Calculations. All numerical manipulations, including estimation of exposed using fastqc. Illumina adaptors were trimmed with Trimmomatic 0.32 (56). RNA during stress, and statistical tests can be found in Supporting Information. An index genome was built with Bowtie 0.12.7 using the S288C reference genome (57). Reads were aligned using Bowtie 0.12.7, and mapped reads ACKNOWLEDGMENTS. We thank all members of the R.P. laboratory and Olke were counted using HTSeq (58). Normalization and differential expres- Uhlenbeck for valuable conversations. We thank Yuan Lin for purifying and sion was performed with DESeq 1.22.1 (59). Sequences can be viewed in sending all MBP-GFP-IDR-HIS protein constructs. We thank the University of Col- the National Center for Biotechnology Information Gene Expression orado BioFrontiers Institute Next-Gen Sequencing Core Facility, which performed Omnibus database (accession no. GSE99170). More details are provided in the Illumina sequencing and library construction. Some of the imaging in this Supporting Information. work was performed at the BioFrontiers Institute Advanced Light Microscopy Core. Laser scanning confocal microscopy was performed on a Nikon A1R microscope supported by National Institute of Standards and Technology-University of Length and Optimal Codon Analysis. Sequencing data for in vitro RNA as- Colorado Cooperative Agreement Award 70NANB15H226. National Science semblies were analyzed for length and codon optimality based on previous Foundation SCR Training Grant T32GM08759 (to B.V.T.) and the Howard Hughes reports (60, 61). Specific details are provided in Supporting Information. Medical Institute (R.P.) funded this work.

1. Buchan JR (2014) mRNP granules. Assembly, function, and connections with disease. 33. Zanchetta G, Bellini T, Nakata M, Clark NA (2008) Physical polymerization and liquid RNA Biol 11:1019–1030. crystallization of RNA oligomers. J Am Chem Soc 130:12864–12865. 2. Voronina E, Paix A, Seydoux G (2012) The P granule component PGL-1 promotes the 34. Leontis NB, Stombaugh J, Westhof E (2002) The non-Watson-Crick base pairs and their localization and silencing activity of the PUF protein FBF-2 in germline stem cells. associated isostericity matrices. Nucleic Acids Res 30:3497–3531. Development 139:3732–3740. 35. Bounedjah O, et al. (2014) Free mRNA in excess upon polysome dissociation is a 3. Kiebler MA, Bassell GJ (2006) Neuronal RNA granules: Movers and makers. Neuron 51: scaffold for protein multimerization to form stress granules. Nucleic Acids Res 42: – 685 690. 8678–8691. – 4. Anderson P, Kedersha N (2006) RNA granules. J Cell Biol 172:803 808. 36. Mahadevan K, et al. (2013) RanBP2/Nup358 potentiates the translation of a subset of 5. Fox AH, et al. (2002) Paraspeckles: A novel nuclear domain. Curr Biol 12:13–25. mRNAs encoding secretory proteins. PLoS Biol 11:e1001545. 6. Gall JG (2000) Cajal bodies: The first 100 years. Annu Rev Cell Dev Biol 16:273–300. 37. Fay MM, Anderson PJ, Ivanov P (2017) ALS/FTD-associated C9ORF72 repeat RNA 7. Buchan JR, Parker R (2009) Eukaryotic stress granules: The ins and outs of translation. promotes phase transitions in vitro and in cells. Cell Rep 21:3573–3584. Mol Cell 36:932–941. 38. Tourrière H, et al. (2003) The RasGAP-associated endoribonuclease G3BP assembles 8. Anderson P, Kedersha N (2009) RNA granules: Post-transcriptional and epigenetic stress granules. J Cell Biol 160:823–831. modulators of gene expression. Nat Rev Mol Cell Biol 10:430–436. 39. Anderson P, Kedersha N (2008) Stress granules: The Tao of RNA triage. Trends 9. Barbee SA, et al. (2006) Staufen- and FMRP-containing neuronal RNPs are structurally – and functionally related to somatic P bodies. Neuron 52:997–1009. Biochem Sci 33:141 150. 10. McCann C, et al. (2011) The Ataxin-2 protein is required for microRNA function and 40. Mazroui R, et al. (2002) Trapping of messenger RNA by fragile X mental retardation synapse-specific long-term olfactory habituation. Proc Natl Acad Sci USA 108: protein into cytoplasmic granules induces translation repression. Hum Mol Genet 11: E655–E662. 3007–3017. 11. Kim HJ, et al. (2013) Mutations in prion-like domains in hnRNPA2B1 and 41. Ash PEA, et al. (2013) Unconventional translation of C9ORF72 GGGGCC expansion hnRNPA1 cause multisystem proteinopathy and ALS. Nature 495:467–473. generates insoluble polypeptides specific to c9FTD/ALS. Neuron 77:639–646. 12. Buchan JR, Kolaitis R-M, Taylor JP, Parker R (2013) Eukaryotic stress granules are 42. Zu T, et al. (2013) RAN proteins and RNA foci from antisense transcripts in C9ORF72 cleared by autophagy and Cdc48/VCP function. Cell 153:1461–1474. ALS and frontotemporal dementia. Proc Natl Acad Sci USA 110:E4968–E4977. 13. Mackenzie IR, et al. (2017) TIA1 mutations in amyotrophic lateral sclerosis and 43. Lee Y-B, et al. (2013) Hexanucleotide repeats in ALS/FTD form length-dependent RNA frontotemporal dementia promote phase separation and alter stress granule dynamics. foci, sequester RNA binding proteins, and are neurotoxic. Cell Rep 5:1178–1186. Neuron 95:808–816.e9. 44. Boeynaems S, et al. (2017) Phase separation of C9orf72 dipeptide repeats perturbs 14. Anderson P, Kedersha N, Ivanov P (2015) Stress granules, P-bodies and cancer. Biochim stress granule dynamics. Mol Cell 65:1044–1055.e5. Biophys Acta 1849:861–870. 45. Kwon I, et al. (2014) Poly-dipeptides encoded by the C9orf72 repeats bind nucleoli, 15. Somasekharan SP, et al. (2015) YB-1 regulates stress granule formation and tumor impede RNA biogenesis, and kill cells. Science 345:1139–1145. – progression by translationally activating G3BP1. J Cell Biol 208:913 929. 46. Lin Y, et al. (2016) Toxic PR poly-dipeptides encoded by the C9orf72 repeat expansion 16. Reineke LC, Lloyd RE (2013) Diversion of stress granules and P-bodies during viral target LC domain polymers. Cell 167:789–802.e12. infection. Virology 436:255–267. 47. Shi KY, et al. (2017) Toxic PRn poly-dipeptides encoded by the C9orf72 repeat ex- 17. Beckham CJ, Parker R (2008) P bodies, stress granules, and viral life cycles. Cell Host pansion block nuclear import and export. Proc Natl Acad Sci USA 114:E1111–E1117. Microbe 3:206–212. 48. Hilliker A, Gao Z, Jankowsky E, Parker R (2011) The DEAD-box protein Ded1 modulates 18. Gilks N, et al. (2004) Stress granule assembly is mediated by prion-like aggregation of translation by the formation and resolution of an eIF4F-mRNA complex. Mol Cell 43: TIA-1. Mol Biol Cell 15:5383–5398. – 19. Kedersha N, et al. (2016) G3BP-Caprin1-USP10 complexes mediate stress granule 962 972. condensation and associate with 40S subunits. J Cell Biol 212:845–860. 49. Bounedjah O, et al. (2012) Macromolecular crowding regulates assembly of mRNA BIOCHEMISTRY 20. Nonhoff U, et al. (2007) Ataxin-2 interacts with the DEAD/H-box RNA helicase stress granules after osmotic stress: New role for compatible osmolytes. J Biol Chem DDX6 and interferes with P-bodies and stress granules. Mol Biol Cell 18:1385–1396. 287:2446–2458. 21. Decker CJ, Teixeira D, Parker R (2007) Edc3p and a glutamine/asparagine-rich domain 50. Clemson CM, et al. (2009) An architectural role for a nuclear noncoding RNA: NEAT1 of Lsm4p function in processing body assembly in Saccharomyces cerevisiae. J Cell Biol RNA is essential for the structure of paraspeckles. Mol Cell 33:717–726. 179:437–449. 51. de Mezer M, Wojciechowska M, Napierala M, Sobczak K, Krzyzosiak WJ (2011) Mu- 22. Hennig S, et al. (2015) Prion-like domains in RNA binding proteins are essential for tant CAG repeats of Huntingtin transcript fold into hairpins, form nuclear foci and are building subnuclear paraspeckles. J Cell Biol 210:529–539. targets for RNA interference. Nucleic Acids Res 39:3852–3863. 23. Kato M, et al. (2012) Cell-free formation of RNA granules: Low complexity sequence 52. Mankodi A, Lin X, Blaxall BC, Swanson MS, Thornton CA (2005) Nuclear RNA foci in domains form dynamic fibers within hydrogels. Cell 149:753–767. the heart in myotonic dystrophy. Circ Res 97:1152–1155. 24. Lin Y, Protter DSW, Rosen MK, Parker R (2015) Formation and maturation of phase- 53. Hall LL, et al. (2017) Demethylated HSATII DNA and HSATII RNA foci sequester – separated liquid droplets by RNA-binding proteins. Mol Cell 60:208 219. PRC1 and MeCP2 into cancer-specific nuclear bodies. Cell Rep 18:2943–2956. 25. Nott TJ, et al. (2015) Phase transition of a disordered nuage protein generates en- 54. Saldi TK, et al. (2014) TDP-1, the Caenorhabditis elegans ortholog of TDP-43, limits the – vironmentally responsive membraneless organelles. Mol Cell 57:936 947. accumulation of double-stranded RNA. EMBO J 33:2947–2966. 26. Elbaum-Garfinkle S, et al. (2015) The disordered P granule protein LAF-1 drives phase 55. Kinsella RJ, et al. (2011) Ensembl BioMarts: A hub for data retrieval across taxonomic separation into droplets with tunable viscosity and dynamics. Proc Natl Acad Sci USA space. Database (Oxford) 2011:bar030. 112:7189–7194. 56. Bolger AM, Lohse M, Usadel B (2014) Trimmomatic: A flexible trimmer for Illumina 27. Zhang H, et al. (2015) RNA controls PolyQ protein phase transitions. Mol Cell 60: sequence data. Bioinformatics 30:2114–2120. 220–230. 57. Langmead B, Trapnell C, Pop M, Salzberg SL (2009) Ultrafast and memory-efficient 28. Khong A, et al. (2017) The stress granule transcriptome reveals principles of mRNA alignment of short DNA sequences to the human genome. Genome Biol 10:R25. accumulation in stress granules. Mol Cell 68:808–820.e5. 58. Anders S, Pyl PT, Huber W (2015) HTSeq–A Python framework to work with high- 29. Jain S, et al. (2016) ATPase-modulated stress granules contain a diverse proteome and – substructure. Cell 164:487–498. throughput sequencing data. Bioinformatics 31:166 169. 30. Ferrandon D, Koch I, Westhof E, Nüsslein-Volhard C (1997) RNA-RNA interaction is 59. Anders S, Huber W (2010) Differential expression analysis for sequence count data. required for the formation of specific bicoid mRNA 3′ UTR-STAUFEN ribonucleopro- Genome Biol 11:R106. tein particles. EMBO J 16:1751–1758. 60. Nagalakshmi U, et al. (2008) The transcriptional landscape of the yeast genome de- 31. Jain A, Vale RD (2017) RNA phase transitions in repeat expansion disorders. Nature fined by RNA sequencing. Science 320:1344–1349. 546:243–247. 61. Presnyak V, et al. (2015) Codon optimality is a major determinant of mRNA stability. 32. Aumiller WM, Jr, Pir Cakmak F, Davis BW, Keating CD (2016) RNA-based coacervates Cell 160:1111–1124. as a model for membraneless organelles: Formation, properties, and interfacial li- 62. Schindelin J, et al. (2012) Fiji: An open-source platform for biological-image analysis. posome assembly. Langmuir 32:10042–10053. Nat Methods 9:676–682.

Van Treeck et al. PNAS | March 13, 2018 | vol. 115 | no. 11 | 2739 Downloaded by guest on October 3, 2021