Advertisement

Evolving Methods in Defining the Role of RNA in RNP Assembly

  • Jaya Sarkar
  • Jong Chan Lee
  • Sua MyongEmail author
Chapter
Part of the Biological and Medical Physics, Biomedical Engineering book series (BIOMEDICAL)

Abstract

The causative connection between aberrant liquid–liquid phase separation (LLPS) of ribonucleoprotein (RNP) granules and neurodegeneration is well known. LLPS driven by intrinsically disordered proteins has been intensely investigated. However, the role of RNA and RNA–protein interaction in RNP granule assembly, properties, maintenance, and eventual onset and progression of neurodegeneration remains poorly understood. A critical hurdle in addressing this question is the shortcomings of currently employed ensemble methods in probing very early-stage molecular events. Here, we present a unique combination of single-molecule biophysical and ensemble methods that can dissect single protein–RNA interaction. The advantage of this approach is that it provides a potential tool to identify early-stage molecular defects that may contribute to the onset of neurodegenerative diseases.

Keywords

Ribonucleoprotein (RNP) Granules Liquid–liquid phase separation RNA–protein interaction Single-molecule approach Early assembly of RNP 

3.1 Introduction

The role of liquid–liquid phase separation (LLPS) in biology has received intense attention over the past few years. Its biological relevance continues to grow from being the basis behind the formation of ribonucleoprotein (RNP) granules [1], heterochromatin compaction [2, 3] to microtubule assembly [4]. Owing to their composition, certain types of RNP granules, such as stress granules (SGs), have the potential to act as the melting pot of misfolded proteins and protein aggregates that can lead to the formation of pathological bodies found in neurodegeneration. In fact, mutations in several SG proteins accelerate aberrant aging of these RNP bodies and are causative of neurodegeneration. Our focus, in this chapter, is on the potential role of RNA as an essential component of these RNP granules, more specifically: What is the molecular basis of RNA–protein interaction involved in the assembly, maintenance, and pathological progression of SGs? To address this question, here, we try to consolidate some of the myriad of recent findings in the field; discuss some current methodologies in their strengths and weaknesses; and finally put forth our methods and insights in an attempt to tackle some of the gaps and outstanding questions in the field. Together, these approaches may lead to a better understanding of disease pathogenesis and developing therapeutic interventions.

3.1.1 Composition of RNP Granules

Broadly, RNP granules are a general term used for membraneless phase-separated organelles containing a high local concentration of proteins and RNA. In eukaryotes, some of these are nuclear (such as Cajal bodies and PML bodies), while some are cytoplasmic (such as SGs and P bodies) [5]. A prominent example of RNP granules in other organisms includes germ cell granules (P granules) in Caenorhabditis elegans. In our discussion here, we focus on two widely used models in probing RNP granule mechanisms—SGs and P granules, using them as examples when appropriate. SGs are sites of RNA triage, formed from untranslated mRNAs and RNA-binding proteins (RBPs), when eukaryotic cells are under stress [6]. P granules play a key role in germ cell development in C. elegans [7].

RNP granules contain RNA, RBPs, and also non-RNA-binding proteins. The RBPs present in RNP granules contain signature motifs or domains—RNA recognition motifs (RRMs) and intrinsically disordered regions (IDRs). IDRs are also termed low complexity domains that are structurally disordered. While some IDRs feature uncharged polar amino acid residues infused with bulky aromatic residues (such as Gln-Gly-Ser-Tyr or Gly-rich patches), others may have charged residues (such as Arg-Gly-rich patches). Such amino acid composition renders the granule-forming RBPs interactive, thus making them ideal agents for nucleating (homotypic interaction) and recruiting others (heterotypic interaction) to promote large assemblies such as RNP granules (discussed in the next section). Taken together, the RNA-binding ability and self- and cross-interactive nature of RNP forming proteins enable them to establish multivalent yet dynamic RNP network (Fig. 3.1).
Fig. 3.1

RNP components and interaction. a Simplified list of RNP constituents. b Different types of molecule-to-molecule interaction mode. c Multivalent interactions that may occur in different RNP context. d Defective molecular assembly that may depict liquid-to-solid transition of RNP components in pathogenic conditions

Under healthy conditions, assembly and disassembly of SGs are all a part of regular cellular dynamics, designed to protect untranslated mRNAs during stress. Components of SG identified by the earlier study include stalled preinitiation complex containing ribosomal subunits; translation associated factors, such as initiation factors eIF2, eIF3, PABP; and mRNA structure/function regulating proteins such as Staufen and G3BP [8]. Recent proteomic analyses of SGs isolated from yeast and mammalian cells have revealed a more diverse composition of these granules [9], identifying the presence of novel and conserved classes of proteins that include: ATP-dependent RNA and DNA helicases, and numerous DEAD-box proteins; ATP-dependent protein and nucleic acid remodeling factors, such as heat shock proteins and chaperones; ribosome biogenesis proteins; and housekeeping proteins such as aminoacyl-tRNA synthetases.

Neuronal inclusions from amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) patients contain two RNA-processing nuclear proteins namely FUS and TDP-43 [10]. In vitro studies showed that disease-associated mutations in these proteins exhibited signs of accelerated aging or maturation of protein droplets characterized by loss of liquid-like property and subsequent appearance of solid-like crystals [11]. Such drastic change indicates aberrations in IDR–protein interaction that disrupts the inherent LLPS mechanism that is believed to control the assembly and dynamics of RNP granules. Interestingly, for mammalian cells, stress-specific differences have been highlighted in the composition, assembly, and dynamics of SGs and SG-like cytoplasmic foci that are also induced by stress. For example, while sorbitol stress recruits TDP-43 to canonical SGs, sodium arsenite stress does not [12, 13].

Similar to mammalian SGs, C. elegans P granules are also composed of RNA and certain conserved classes of proteins, many of which are disordered [7]. These include DEAD-box helicases LAF-1, GLH-1 through 4; RNA-binding proteins MEG-3, PGL-1 through 3. Overall, the conserved presence of intrinsically disordered and RNA-binding proteins in both SGs and P granules underscores the importance of RNA remodeling activities in RNP granules.

3.1.2 Mechanisms of RNP Granule Formation: Protein–Protein and Protein–RNA Interactions Are Both Driving Forces

In the recent past, molecular mechanisms underlying RNP granule assembly have been intensely investigated, from both the biophysics and material science perspective. In in vitro studies, granule/pathological inclusion forming proteins, such as FUS, have been shown to access different material states [11, 14]. Initially, FUS phase separates into metastable droplets that behave as liquids, then with time and increased local protein concentration, this liquid phase quickly matures to an intermediate hydrogel-like state, finally converting to stable solid-like fibers (structurally resembling amyloid fibers in neurodegeneration), and this conversion is accelerated by ALS-implicated protein mutations in FUS. In vivo, however, in the healthy state, RNP granules are thought to maintain liquid-like state; for example, liquid properties of P granules have been established, based on fusion, dripping, and rapid diffusion rate of components between inside and outside of these granules [15]. In vivo SGs, on the other hand, have been envisioned as a coexistence of densely packed state and liquid states, evidenced by the presence of a stable core surrounded by a dynamic shell [9].

Mostly derived from cell-free in vitro studies, at present, the converging understanding is that IDR protein-driven LLPS is the key behind liquid–liquid demixing and formation of reversible RNP granules that coexist with its surrounding components [6]. The critical requirement here is multivalent weak and transient interactions (provided by the labile interactions between IDR proteins), that are strong enough to hold the RNP assemblies together, but are not so strong as to arrest dynamics and reversibility of these structures. Because these RNP granules contain many different proteins, protein–protein interaction between non-IDR proteins also contributes toward building and maintenance of RNP granules. Heat shock proteins and chaperones that constitute SG of yeast and mammalian cells [9] are thought to counteract the aggregation-prone tendency of IDR-containing proteins by resolving misfolded proteins and dispersing granule components [16]. A recent study shows that ATP can also act as a hydrotrope to solubilize the molecularly crowded and compacted state of cellular granules [17].

Increasingly critical role of RNA in RNP granule nucleation and dynamics: Many RBPs present in RNP granules, apart from possessing RRMs, also possess IDRs, for example, LAF-1, FUS, TDP-43, MEG-3, and Whi3. A central question here is: Does RNA play an active role in nucleation and assembly of RNP granules, or is it just an inevitable consequence of RNA tagging along the RBP? Recent reports allude to both the active and regulatory roles for RNA in RNP assembly, albeit through multi-faceted molecular mechanisms, as highlighted below: (i) RNA impacts assembly of LLPS droplets. RNA seeds FUS higher-order assemblies (visible as ropey structures in transmission electron microscopy images) even at low protein concentrations, suggesting that RNA promotes the phase separation of FUS [18]. In vivo, in Drosophila fly model and in mammalian neuronal cells, RNA-binding ability of FUS is essential for ALS mutation containing FUS to show neurodegenerative phenotype or to localize to cytoplasmic SGs, respectively [19]. Similarly, disrupting RNA-binding ability of TDP-43 mutants (ALS-linked) rescued TDP-43 mediated cellular toxicity [20]. Whi3 is a fungal RNA-binding IDR protein (responsible for asynchronous nuclear division via spatial patterning of RNA transcripts), whose phase separation is driven by its cellular mRNA target sequences [21]. Furthermore, mRNA structure is critical for assembling of distinct Whi3 droplets and protein-driven RNA conformational changes for maintaining such identity [22]. (ii) Strikingly, a recent pioneering study reports on how RNA by itself can phase separate. This particular RNA arises from repeat expansion at C9 or f72 (chromosome 9 open reading frame 72) which is responsible for causing high percentage of both familial and sporadic ALS [23, 24]. The expanded form of RNA was shown to phase transition into nuclear foci with strength directly proportional to the length and secondary structure of the repeat RNA [25]. (iii) RNA also tunes dynamics of the RNP granules post-nucleation. RNA fluidizes LAF-1 liquid droplets in vitro, with a concurrent increase in the protein–RNA dynamics probed by single-molecule assay [26]. In contrast, Whi3 droplet viscosity was increased and dynamics decreased in the presence of a specific mRNA [21]. Therefore, RNA can up or down regulate the granule fluidity depending on the molecular context. (iv) RNA regulates phase separation of IDR proteins in cells. Maharana et al. [27] showed that high RNA:protein ratio keeps these proteins soluble in nucleus, while low ratio promotes LLPS in the cytoplasm, a hallmark in ALS patients.

The overall current understanding is that at the granule assembly stage, RNA provides a platform which leads to recruitment of multiple RBPs, thus enabling multivalent interactions among RNPs (Fig. 3.1). These multivalent interactions, in turn, increase the local protein concentration, which increases the propensity of interaction between the IDRs of these RBPs, allowing for clustering into a stable nucleation core. Once established, such core can incorporate other proteins and phase separate into cellular granules [28]. Post-assembly, RNA may continue to have a critical role in modulating granule fluidity, plausibly via tuning dynamics of RNA–RBP interaction. Thus, a synergistic role of RNA and protein in regulating RNP granule assembly, properties, and maintenance is increasingly been acknowledged.

3.1.3 Stages of RNP Granule Life and Implications in Disease

As mentioned in the previous section, evidence suggests that cellular RNP granules such as SGs and P granules are liquid-like reversible structures. In vitro however, the disordered constituent proteins of SGs, such as FUS and TDP-43, can quickly convert from liquid to hydrogel to fibers. In agreement with the finding that SGs absorb pathogenic inclusions [29], several SG proteins, including FUS and TDP-43, are present in aggregates/inclusions that are present in the motor neurons of ALS and FTD patients.

In fact, pathological inclusions are believed to originate from misregulated SGs [10]. Aberrations in interactions between IDR-containing RBPs and other SG proteins potentially can convert the weak transient interactions (that are responsible for LLPS and normal liquid-like SG dynamics) into more ordered solid-like interactions that cause loss/misregulation of SG fluidity and/or disassembly. Although there is lack of direct evidence for existence of the gel and fiber states in vivo for FUS and TDP-43, it is generally perceived that while the liquid-like state of granules represents the normal/default situation in cells, the fiber-like states resemble the beta-sheet structure of amyloid fibers that are found in the aggregates that occur in diseased individuals. Many ALS and FTD patient mutations in FUS and TDP-43 map to their RRM and IDR domains [30], strongly suggesting the deleterious effect caused by RNA-binding defect and IDR-driven aggregation in diseased state. On the other hand, the ALS-associated long multiple repeat RNAs, C9 or f72, promote multivalent intermolecular interactions responsible for LLPS of the RNAs into liquid droplets, and then into gels that manifest as RNA foci, a hallmark of C9 or f72 associated ALS [25].

While the quick coalescence of RNA and RBPs into SGs is critical for preserving cellular processes during stress, it comes with the high risk of aggregating these proteins into pathological inclusions. How does the cell manage this risk and prevent the abnormal aggregation? Two proposed avenues by which cells do this are: (i) Balanced cross-talk between RNA and RBP quality control [31]. Spinocerebellar ataxia type 31 (SCA31) is characterized by toxicity arising from RNA foci formed by expanded repeats of UGGAA and the aberrant proteins produced from non-canonical RAN translation (non-AUG translation) from the expanded RNAs. Ishiguro et al. showed that FUS and TDP-43 act as RNA chaperones by directly binding to the expanded UGGAA RNAs, resolving the folded structures, leading to reduced RNA foci and suppressed neurotoxicity [32]. ATP-dependent RNA helicases such as DEAD-box proteins were also found to unwind expanded RNA repeats and rescue toxicity. This group proposed a model in which not only RBPs can mitigate RNA toxicity, but also non-expanded RNAs can rescue mutant RBP-mediated toxicity. Thus, mutation in either the RNAs or the RBPs can perturb the balance in protein–RNA homeostasis, causing aggregation and toxicity. In addition, the protein quality control pathways such as molecular chaperones, protein degradation pathways, and prevention of mistranslation at the levels of aminoacyl-tRNA synthetases and ribosomes also contribute toward RNP homeostasis [33]. (ii) Small molecules in cells, such as ATP. Early study by Brangwynne et al. demonstrated that ATP removal from C. elegans induced loss of liquid-like property in P granules, suggesting the role of ATP or ATP-mediated processes in fluidizing RNP granules [15]. ATP depletion experiments in mammalian cells showed that ATP is required during SG assembly and also in maintaining granule fluidity [9]. A recent in vitro analyses shed insight into how cellular ATP in high concentrations (similar to the physiological concentration of 5–10 mM) may act as a biological hydrotrope whereby the amphipathic property of ATP induces solubilization of aggregation-prone cellular proteins such as FUS [17].

3.2 Current Methods in Probing RNP Granules: Strengths and Limitations

Early genetic and cellular studies on stress response provided clear evidence of the existence of SGs in cells. Evidence indicates that SG formation is vital to cellular survival under stressed condition [34]. Genetic studies also identified key players and mutations in the RNP components that can cause neurodegenerative diseases. Recent research effort has focused on understanding the material properties of in vitro droplets and cellular granules, using a combination of simple and sophisticated methods, as outlined here (Fig. 3.2): (i) Bright-field imaging of granule-forming IDR protein droplets, turbidity measurements by optical density (300–600 nm) (Fig. 3.2a, b) under different conditions such as temperature, salt, and protein concentration have provided valuable information about the propensity and size of protein droplets, enabling construction of a phase diagram which defines conditions that promote LLPS-driven droplet formation [26, 35]. Droplet fusion events have also been studied more precisely using FUS protein and optical tweezers [11]. (ii) Microrheology experiments have been developed to measure the viscosity and elasticity of droplets (Fig. 3.2d), shedding light on fluidity of the droplets under varying conditions such as the presence or absence of RNA [21]. (iii) Fluorescence recovery after photobleaching (FRAP) of both in vitro droplets and cellular granules (Fig. 3.2c) has also yielded information about the diffusion kinetics [11, 15, 21, 26]. (iv) Conventional biochemical methods including SDS-PAGE and Western blotting have been employed to distinguish the hydrogel-like state from liquid droplet and solid fiber [14]. In combination with electron microscopy and X-ray diffraction, these hydrogels were deduced to contain homotypic polymerized fibers that were dynamic, unlike disease-featuring amyloid fibers. Also, oligomerization stoichiometry of IDR proteins has been probed by electrophoretic mobility shift assay (EMSA) (Fig. 3.2f) [36]. (v) Recently, Roy Parker’s lab has devised a SG isolation method [37]. While mass-spectrometric analyses (Fig. 3.2e) of isolated granules have revealed the diverse proteome of mammalian SGs [9], time-course fluorescence microscopy analyses have revealed that SG assembly is a multi-step process in which the stable core forms first, followed by the dynamic shell [28].
Fig. 3.2

Biophysical ensemble measurement. a Formation of LLPS can be measured by optical density at 350–600 nm. b Bright-field imaging can be used to track the growth and fusion kinetic of in vitro droplets. c FRAP analysis indicates fluidity of droplet content. d Microrheology measures viscoelasticity of droplets. e Mass spectrometry reveals constituents of RNP granules. f Electrophoresis probes’ stages of protein/RNA multimerization

The above-mentioned methods have undoubtedly provided valuable mesoscale information such as droplet formation conditions, size, fusion kinetics, viscosity, and diffusion parameters, some molecular mechanisms underlying RNP nucleation and changes in granule properties post-nucleation. However, most of our current knowledge derived from these methods lack molecular details of RNA–protein interaction involved in the early stages of RNP granule nucleation and assembly. In the next part of this chapter, we introduce a few such biophysical methods, including single-molecule fluorescence detection, which are ideally suited to address the molecular basis of RNP formation [26, 36].

3.3 Methods to Probe Initial Phases of RNP Assembly

3.3.1 Measuring RNA–Protein Interaction Across the Phase Boundary: Single-Molecule FRET and EMSA

One of the very early stages in RNP granule assembly likely involves discrete steps of RNA–protein and protein–protein interaction in the soluble phase which transitions to the liquid-like phase separation, which ages to more mature forms of hydrogel-like and solid state such as fibers. In vitro, the conditions that lead to the onset of phase separations such as temperature, salt, and protein concentrations can be tuned for each protein or protein/RNA system to generate a phase diagram which can display the clear partition between the soluble and the LLPS space. Such analyses have been done for proteins including FUS, LAF1, and Whi3 [21, 26, 35].

Understanding RNA–protein interaction at the onset of granule assembly necessitates first understanding how the protein interacts with RNA in its soluble phase. To probe the interaction between single RNA and single protein, we used single-molecule FRET (smFRET) assay based on total internal reflection fluorescence (TIRF) microscopy (Fig. 3.3) [38]. In addition, we applied EMSA (Fig. 3.2f) [39] to determine the stoichiometry of protein–RNA complex. Combination of these two approaches allows one to probe single RNA–protein interaction as a function of granule promoting parameters (protein and salt concentration) and map it to corresponding phase space as demonstrated in our previous work [26, 36].
Fig. 3.3

smFRET detection. a FRET-RNA or DNA substrate immobilized to PEG surface. b Cy3 and Cy5 signal from same set of molecules (circle). c FRET values collected from thousands of molecules are built into FRET histogram. d Individual smFRET traces report on time-dependent change in FRET

In our previous study, we employed LAF-1, an IDR-containing DEAD-box RNA helicase present in P granules of C. elegans as a model protein. Purified LAF-1 phase separates in vitro, driven by its intrinsically disordered N-terminal RGG-rich domain that is also an RNA-binding domain. EMSA experiments revealed that LAF-1 binds specifically to single-stranded (ss)RNA. So, our model RNA substrates consisted of ssRNA overhang of 15–50 nucleotides (poly U sequence) in a format of partially duplexed RNA (Fig. 3.3a). We refer to these substrates as U15, U30, U40, U50, depending on the length of the poly U overhang. One of the RNA strands is biotinylated so that the RNA substrate can be surface immobilized on the PEG-passivated quartz slide to be used on the TIR microscope set up. Each RNA substrate is dual labeled with a pair of FRET-suitable fluorophores which are arranged such that FRET reports on how LAF-1-binding impacts the conformation of ssRNA (Fig. 3.3b–d).

Varying LAF-1 concentration from low to high (corresponding to the transition from soluble phase to LLPS regime based on the LAF-1 phase diagram) was applied to fluorophore-labeled RNA substrates for EMSA analysis (Fig. 3.4a, b). Only one shifted band (relative to the unbound RNA only band) was observed for U30 across the protein concentrations, indicating a monomer protein binding to RNA. For U40 and U50, in addition to this band, a super-shifted band was observed in high protein concentrations, indicating multimer protein binding to RNA (Fig. 3.4c). In light of the applied LAF-1 concentrations required for soluble to LLPS transition, this set of data suggests that U30 may not be long enough to accommodate more than one protein, yet U40 and U50 have sufficient length to recruit multimers of proteins which can promote the droplet assembly.
Fig. 3.4

LAF-1 induces dynamics on ssRNA in droplet-forming condition. a Color and shape key for droplet versus non-droplet-forming conditions. b Experimental conditions cutting across phase boundary in [LAF-1] and [NACl]. c Droplet-forming conditions coincide with dimerization of LAF-1 denoted by double red asterix. d, e In droplet-forming condition of high [LAF-1], LAF-1 induces dynamic mobility on ssRNA, evidenced by FRET fluctuation

Is there any change in these proteinRNA interactions going from soluble to phase separation? Here is where the smFRET measurements offer unique advantage. We applied different protein concentrations to surface immobilized FRET-labeled RNA. The intensities of donor (Cy3) and acceptor (Cy5) are collected from approximately 300–400 RNA molecules per field of view, i.e., in one movie. Such data can be analyzed in two ways: (i) FRET efficiencies collected from thousands of U50 molecules (from 10 to 20 short movies) are built into a histogram which displays the overall FRET distribution; (ii) individual smFRET traces taken for 2–3 min displays how FRET (calculated from intensity of the donor and acceptor dyes) changes over time, which is interpreted as the conformational changes within individual RNA molecules as the proteins act upon them. These two evaluations together give us a clear picture regarding not only the binding mode of the protein to the RNA, but also the rare glimpses of molecular details of this interaction intractable by ensemble methods. The U50 RNA alone yields a low FRET peak due to the distance between the two dyes separated by 50 ribonucleotides. Application of low LAF-1 or high salt concentrations (that represent soluble phase) to U50 RNA shifts the FRET histogram peak from low FRET (unbound U50 RNA) to high FRET (representing LAF-1-bound U50 RNA) (Fig. 3.4d, e). The time traces of individual U50 molecules show a shift from low to high FRET immediately after the protein addition and the signal remaining stable over time. The EMSA data taken in the same condition shows a single band shift, representing monomer-bound U50 fraction. This indicates that a monomer LAF-1 binding induces tight compaction of the RNA (bringing the two dyes into a close proximity) that is stable over time. As LAF-1 protein concentration increases or salt concentration decreases, approaching and crossing the phase boundary, a broad mid-FRET peak appears in addition to the high FRET peak. The single-molecule traces exhibit dynamic FRET fluctuation interspersed with a static high FRET state (Fig. 3.4d, e). In this condition, EMSA analysis reveals a mixture of a monomer-bound (single shift) and multimer-bound (double shift) stoichiometric states (Fig. 3.4c), reflecting the coexistence of a monomer and multimer-bound states generating static high FRET and dynamic fluctuating FRET, respectively. When the protein concentrations correspond to the inside of phase boundary, EMSA showed primarily double shift and smFRET traces displayed majority of molecules exhibiting FRET fluctuations.

Thus, an ensemble biochemical assay such as EMSA, biophysical measurements including microrheology, viscoelasticity, and the smFRET assay can be combined to extract unique material properties of RNP droplet and the underlying molecular details involved in the formation of LLPS, especially at the early stages of RNP assembly. In case of LAF-1, such a strategy helped us understand that as conditions (concentrations of protein, salt, etc.) transition from soluble to phase-separated LAF-1 RNP droplet, monomer-bound and tightly wrapped RNA evolves to multimer protein which dynamically interacts with RNA, likely representing a state that is ready to assemble into RNP droplets. The dynamicity may also contribute to the droplet fluidizing effect that RNA has been observed to have on LAF-1 RNP droplets. We envision this strategy to be effective with other IDR granule-forming proteins as well, such as FUS and MEG-3 (unpublished data).

3.3.2 RNA Annealing Assay as a Proxy for RNP–RNP Interaction in RNP Granules

In the context of RNP granules, RNP–RNP interaction is undoubtedly a key factor in regulating all stages of granule life. We devised an assay that can potentially test for this level of interaction [36]. We posited that for two complementary strands of RNA to hybridize, two sets of RNA–protein complexes need to come together, i.e., requiring RNP–RNP contact. We established an annealing assay in which we immobilized a partially duplexed RNA (with ss overhang of mixed sequence) that is FRET dye (Cy3, Cy5) labeled, exhibiting high FRET. We apply pre-incubated mixture of complementary ssRNA and LAF-1 (Fig. 3.5a). In the pre-incubated mix, while some LAF-1 is expected to be in complex with the ssRNA, some protein is expected to be free to interact with the immobilized RNA on surface. The annealing between the two complementary RNA is expected to result in a decrease in FRET since the dyes in the annealed substrate will now be separated by duplexed RNA (Fig. 3.5b). We subjected various conditions of LAF-1 including its N- and C-terminal truncation mutants and found that RNA annealing was greatly enhanced in the conditions that promote droplet formation and dynamic RNA–protein interactions, but substantially diminished for out-of-LLPS conditions that induce tight RNA compaction by the protein (Fig. 3.5c, d). Thus, dynamic RNA–protein interactions (when approaching granule-forming conditions) promote RNP–RNP interactions between LAF-1-RNA complexes. The correlation between (i) monomer to multimer stoichiometric transition, (ii) static to dynamic change in RNA–protein interaction, (iii) defective to efficient RNA annealing, and the (iv) soluble to LLPS reflect that this set of measurement could serve as reporter assays that define the underlying molecular transactions that contribute to the RNP droplet assembly.
Fig. 3.5

LAF-1-RNA dynamics promote RNA annealing. a High FRET converts to low FRET upon RNA annealing. b FRET histogram before (top) and after (bottom) annealing. c Kinetic analysis of RNA annealing reaction. d Annealing rate for various mutants that represent static versus dynamic LAF-1-RNA interaction

3.3.3 Measuring Size of In Vitro Droplets and Cellular Granules at Nucleation: Dynamic Light Scattering and Single-Molecule Pull-Down Assay

In vitro droplets from purified IDR, proteins have been evaluated by DIC imaging, their fluidity measured by FRAP and microrheology analysis as discussed above. These methods, however, do not offer insight into the size of the assemblies at the very early stages at the onset of nucleation. Dynamic light scattering (DLS) is sensitive enough to detect monomer to multimer transition of protein condensation at the onset of droplet assembly, but becomes unsuitable once stable droplets (>1 µm in radius) have formed. DLS has been used to probe oligomer size growth and kinetics of assembly for poly A binding protein (Pab1 in yeast) which form into granules [40]. For such purposes, DLS can yield two parameters: (i) hydrodynamic radius (Rh) for the IDR protein in the soluble phase, by batch mode DLS which estimates size and size distribution of oligomers (in the range of 0.5–1000 nm radius). The highly sensitivity capturing of light scattering pattern that changes according to the size of protein particles makes DLS an apt method to track size and growth of granules at the very early stages of droplet nucleation, far beyond the detection limit of DIC imaging (Fig. 3.6a, red arrow indicating increasing protein concentration, moving to LLPS favorable condition). (ii) Continuous thermal scans of IDR proteins can reveal the temperature of aggregation onset, defining the lower critical solution temperature (LCST) above which the protein will phase separate. This measurement mode is useful for comparing disease mutants of granule proteins, and also for assessing how the presence of RNA may impact the LCST and hence the onset of aggregation. However, as mentioned before, beyond a certain radius (1000 nm), detection of aggregates by DLS is not reliable, thus making it unsuitable to probe droplets that have already assembled. Furthermore, the resolution limit of batch DLS is a factor of 2–5 in size, making it difficult to assign precise stoichiometric state, which can be assessed better by the method introduced below.
Fig. 3.6

Probing molecular assembly of granules. a Dynamic light scattering is useful in measuring early phase of molecular assembly in vitro droplets of purified protein. b Single-molecule pull-down assay can reveal the multimeric state of target proteins in cellular granules by photobleaching

Single-molecule pull-down (SiMPull) assay [41, 42] is a unique and powerful single-molecule technique that combines traditional pull-down assay principles with single-molecule fluorescence microscopy, permitting direct visualization of individual protein complexes directly pulled down from cell lysate, thus constituting a method which is non-perturbing and preserving native cellular context. Upon expressing a target protein fused to a fluorescence marker protein such as GFP or RFP, SiMPull analysis can reveal how many of the labeled proteins are present in cellular protein complexes. The key in this assay is the selective capture of the protein of interest from a cell lysate via an antibody. We discuss an experimental design here to illustrate how SiMPull can be applied to reveal protein oligomeric state in cellular granules, using SGs as an example. Mammalian cells typically used in SG studies such as HEK293, U2OS, or HeLa cells may also be used for this assay. A fluorescently tagged version (such a GFP, YFP, RFP) of the known SG protein of interest can be expressed in the mammalian cells of choice. The cells can be subjected to stressed or non-stressed conditions and SG formation (bright fluorescence puncta) can be checked by fluorescence microscopy. The cell lysate can then be applied onto the single-molecule imaging surface (composed of flow chambers constructed on a sandwich of PEG-passivated slide and coverslip). The imaging surface can be coated with the specific antibody against the protein of interest (anti-GFP antibody, for example) and the cell lysate can be applied. The target protein and protein-containing complexes will be captured by the antibody. Because the target protein is tagged with GFP at either N- or C-terminus, counting the number of photobleaching steps in each spot can yield stoichiometric information about the protein-containing complex unit. In the absence of stress, when there is no granule formation, the protein is expected to be in soluble phase (Fig. 3.6b). The SiMPull image is likely to be occupied by low-intensity spots, depicting monomeric state of the protein (deduced from analyses that are described below). In the presence of stress, we expect to also capture bright fluorescent spots representing clusters of proteins on route to assemble into granules. The total intensity of each of the high-intensity spots will be proportional to the number of fluorescent protein units present in that granule. Therefore, intensities of individual spots alone can reflect the oligomeric status of the proteins. In addition, tracking the number of photobleaching steps for conditions with and without stress can lead to more accurate analysis to distinguish mono- di-, trimeric, and higher oligomeric cellular granules that may represent clusters that form in the early stages of granule assembly. In addition, the SiMPull assay may also be expanded to probe the granule assembly of multiple granule-forming proteins by co-expressing with different fluorescent proteins in cells.

3.3.4 Super-Resolution Imaging to Reveal Granule Structure

Imaging cellular granules by expressing fluorescently tagged protein or in vitro imaging of LLPS droplets formed by purified protein have been a simple but powerful tool for initial studies in the field. The length scale of RNP granules ranges from 100 nm to several micrometers, which is roughly around the limit of conventional diffraction limited microscopy (~300 nm). Recently, new findings suggest that the RNP granules may have substructures: within the nucleolus, and two proteins’ phase separate into two layers of LLPS, driven by different surface tension [43]; a proteomic analysis also revealed that SGs have substructures, consisting of a stable core and a dynamic shell [9]. These studies clearly indicate a more complex level of RNP granule architecture that requires further investigation. However, the substructures are often too small to be observed clearly with a conventional microscope.

Super-resolution imaging techniques developed in the last two decades provides an ideal tool to resolve the substructures within RNP granules which are inaccessible with a conventional microscope. Two important branches of super-resolution imaging technique are stimulated emission depletion (STED) microscopy and localization microscopy. Localization microscopy includes photoactivated localization microscopy (PALM) [44] and stochastic optical reconstruction microscopy (STORM) [45]. STED microscopy was developed by Hell, S. group [46]. Here, we will briefly discuss the applicability and the potential of STED microscopy to reveal granule substructures.

Briefly, STED microscopy uses two lasers instead of one to achieve super-resolution above the optical diffraction limit. The excitation laser is used to excite fluorophores in the same manner as the conventional confocal imaging, and the other STED laser is used to deplete the fluorophore in the shape of a donut, making the fluorophores emit photons only at the center of the donut, hence achieving super-resolution. For STED imaging, the GFP tagged protein is less ideal than the immunolabeling using antibodies conjugated with organic fluorophores. This method is typically better for higher signal-to-noise ratio due to the brightness of the organic fluorophores. The choice of fluorophores is subject to the specific experimental design and scheme. STED microscopy can be beneficial in capturing structural details of granules because the typical size of membraneless granules in cells is sub-micron.

3.4 Concluding Thoughts

In light of the tremendous diversity that has been observed in methods to study granules, we reiterate two important points: Firstly, methods to study granules must be carefully selected according to the stage of the granule’s life that is being targeted. Secondly, a strategic combination of methods is often more powerful to extract the maximum and most accurate information about a particular stage of granule life. While bulk and mesoscale methods will continue to hold an important place in the granule field, because they report on the material property of granules by relatively simple means, methods accessing finer molecular details of interactions will expand and grow, shedding insight about the granule assembly and dynamics. Although we did not discuss here, atomic level probing for the IDR proteins using NMR and hydrogen exchange-coupled mass spectrometry has also been employed to define molecular coordinate of protein conformations and dynamics. However, we believe single-molecule methods, including smFRET and SiMPull, which have found novel applications in granule studies, have the potential to access unprecedented fine molecular details that, until recently, have remained inaccessible by other methods.

Notes

Acknowledgements

This project was supported by Catalyst Research Award from Johns Hopkins University, R01 GM115631 and R01 CA 207342 for all members at Johns Hopkins University.

References

  1. 1.
    Guo, L., & Shorter, J. (2015). It’s raining liquids: RNA tunes viscoelasticity and dynamics of membraneless organelles. Molecular Cell, 60, 189–192.CrossRefGoogle Scholar
  2. 2.
    Larson, A. G., Elnatan, D., Keenen, M. M., Trnka, M. J., Johnston, J. B., Burlingame, A. L., et al. (2017). Liquid droplet formation by HP1α suggests a role for phase separation in heterochromatin. Nature, 547, 236–240.ADSCrossRefGoogle Scholar
  3. 3.
    Strom, A. R., Emelyanov, A. V., Mir, M., Fyodorov, D. V., Darzacq, X., & Karpen, G. H. (2017). Phase separation drives heterochromatin domain formation. Nature, 547, 241–245.ADSCrossRefGoogle Scholar
  4. 4.
    Woodruff, J. B., Ferreira Gomes, B., Widlund, P. O., Mahamid, J., Honigmann, A., & Hyman, P. O. (2017). The centrosome is a selective condensate that nucleates microtubules by concentrating tubulin. Cell, 169, 1066–1077 e1010.Google Scholar
  5. 5.
    Spector, D. L. (2006). SnapShot: Cellular bodies. Cell, 127, 1071.CrossRefGoogle Scholar
  6. 6.
    Protter, D. S., & Parker, R. (2016). Principles and properties of stress granules. Trends in Cell Biology, 26, 668–679.CrossRefGoogle Scholar
  7. 7.
    Updike, D., & Strome, S. (2010). P granule assembly and function in Caenorhabditis elegans germ cells. Journal of Andrology, 31, 53–60.CrossRefGoogle Scholar
  8. 8.
    Anderson, P., & Kedersha, N. (2006). RNA granules. The Journal of Cell Biology, 172, 803–808.CrossRefGoogle Scholar
  9. 9.
    Jain, S., Wheeler, J. R., Walters, R. W., Agrawal, A., Barsic, A., & Parker, R. (2016). ATPase-modulated stress granules contain a diverse proteome and substructure. Cell, 164, 487–498.CrossRefGoogle Scholar
  10. 10.
    Aulas, A., & Vande Velde, C. (2015). Alterations in stress granule dynamics driven by TDP-43 and FUS: A link to pathological inclusions in ALS? Frontiers in Cellular Neuroscience, 9, 423.CrossRefGoogle Scholar
  11. 11.
    Patel, A., Lee, H. O., Jawerth, L., Maharana, S., Jahnel, M., Hein, M. Y., et al. (2015). A liquid-to-solid phase transition of the ALS protein FUS accelerated by disease mutation. Cell, 162, 1066–1077.CrossRefGoogle Scholar
  12. 12.
    Aulas, A., Fay, M. M., Lyons, S. M., Achorn, C. A., Kedersha, N., Anderson, P., et al. (2017). Stress-specific differences in assembly and composition of stress granules and related foci. Journal of Cell Science, 130, 927–937.CrossRefGoogle Scholar
  13. 13.
    Dewey, C. M., Cenik, B., Sephton, C. F., Dries, D. R., Mayer, P., 3rd, Good, S. K., et al. (2011). TDP-43 is directed to stress granules by sorbitol, a novel physiological osmotic and oxidative stressor. Molecular and Cellular Biology, 31, 1098–1108.CrossRefGoogle Scholar
  14. 14.
    Kato, M., Han, T. W., Xie, S., Shi, K., Du, X., Wu, L. C., et al. (2012). Cell-free formation of RNA granules: Low complexity sequence domains form dynamic fibers within hydrogels. Cell, 149, 753–767.CrossRefGoogle Scholar
  15. 15.
    Brangwynne, C. P., Eckmann, C. R., Courson, D. S., Rybarska, A., Hoege, C., Gharakhani, J., et al. (2009). Germline P granules are liquid droplets that localize by controlled dissolution/condensation. Science, 324, 1729–1732.ADSCrossRefGoogle Scholar
  16. 16.
    Kroschwald, S., Maharana, S., Mateju, D., Malinovska, L., Nuske, E., Poser, I., et al. (2015). Promiscuous interactions and protein disaggregases determine the material state of stress-inducible RNP granules. eLife, 4, e06807.Google Scholar
  17. 17.
    Patel, A., Malinovska, L., Saha, S., Wang, J., Alberti, S., Krishnan, Y., et al. (2017). ATP as a biological hydrotrope. Science, 356, 753–756.ADSCrossRefGoogle Scholar
  18. 18.
    Schwartz, J. C., Wang, X., Podell, E. R., & Cech, T. R. (2013). RNA seeds higher-order assembly of FUS protein. Cell Reports, 5, 918–925.CrossRefGoogle Scholar
  19. 19.
    Daigle, J. G., Lanson, N. A., Jr., Smith, R. B., Casci, I., Maltare, A., Monaghan, J., et al. (2013). RNA-binding ability of FUS regulates neurodegeneration, cytoplasmic mislocalization and incorporation into stress granules associated with FUS carrying ALS-linked mutations. Human Molecular Genetics, 22, 1193–1205.CrossRefGoogle Scholar
  20. 20.
    Voigt, A., Herholz, D., Fiesel, F. C., Kaur, K., Muller, D., Karsten, P., et al. (2010). TDP-43-mediated neuron loss in vivo requires RNA-binding activity. PloS One, 5, e12247.ADSCrossRefGoogle Scholar
  21. 21.
    Zhang, H., Elbaum-Garfinkle, S., Langdon, E. M., Taylor, N., Occhipinti, P., Bridges, A. A., et al. (2015). RNA controls PolyQ protein phase transitions. Molecular Cell, 60, 220–230.CrossRefGoogle Scholar
  22. 22.
    Langdon, E. M., Qiu, Y., Ghanbari Niaki, A., McLaughlin, G. A., Weidmann, C., Gerbich, T. M., et al. (2018). mRNA structure determines specificity of a polyQ-driven phase separation. Science.Google Scholar
  23. 23.
    DeJesus-Hernandez, M., Mackenzie, I. R., Boeve, B. F., Boxer, A. L., Baker, M., Rutherford, N. J., et al. (2011). Expanded GGGGCC hexanucleotide repeat in noncoding region of C9ORF72 causes chromosome 9p-linked FTD and ALS. Neuron, 72, 245–256.CrossRefGoogle Scholar
  24. 24.
    Renton, A. E., Majounie, E., Waite, A., Simon-Sanchez, J., Rollinson, S., Gibbs, J. R., et al. (2011). A hexanucleotide repeat expansion in C9ORF72 is the cause of chromosome 9p21-linked ALS-FTD. Neuron, 72, 257–268.Google Scholar
  25. 25.
    Jain, A., & Vale, R. D. (2017). RNA phase transitions in repeat expansion disorders. Nature, 546, 243–247.ADSCrossRefGoogle Scholar
  26. 26.
    Elbaum-Garfinkle, S., Kim, Y., Szczepaniak, K., Chen, C. C., Eckmann, C. R., Myong, S., et al. (2015). The disordered P granule protein LAF-1 drives phase separation into droplets with tunable viscosity and dynamics. Proceedings of the National Academy of Sciences of the United States of America, 112, 7189–7194.ADSCrossRefGoogle Scholar
  27. 27.
    Maharana, S., Wang, J., Papadopoulos, D. K., Richter, D., Pozniakovsky, A., Poser, I., et al. (2018). RNA buffers the phase separation behavior of prion-like RNA binding proteins. Science.Google Scholar
  28. 28.
    Wheeler, J. R., Matheny, T., Jain, S., Abrisch, R., & Parker, R. (2016). Distinct stages in stress granule assembly and disassembly. eLife, 5.Google Scholar
  29. 29.
    Li, Y. R., King, O. D., Shorter, J., & Gitler, A. D. (2013). Stress granules as crucibles of ALS pathogenesis. The Journal of Cell Biology, 201, 361–372.CrossRefGoogle Scholar
  30. 30.
    Lagier-Tourenne, C., Polymenidou, M., & Cleveland, D. W. (2010). TDP-43 and FUS/TLS: emerging roles in RNA processing and neurodegeneration. Human Molecular Genetics, 19, R46–R64.CrossRefGoogle Scholar
  31. 31.
    Ishiguro, T., Sato, N., Ueyama, M., Fujikake, N., Sellier, C., Kanegami, A., et al. (2017). Regulatory role of RNA chaperone TDP-43 for RNA misfolding and repeat-associated translation in SCA31. Neuron, 94, 108–124 e107.CrossRefGoogle Scholar
  32. 32.
    Mateju, D., Franzmann, T. M., Patel, A., Kopach, A., Boczek, E. E., Maharana, S., et al. (2017). An aberrant phase transition of stress granules triggered by misfolded protein and prevented by chaperone function. EMBO Journal, 36, 1669–1687.CrossRefGoogle Scholar
  33. 33.
    Alberti, S., Mateju, D., Mediani, L., & Carra, S. (2017). Granulostasis: Protein quality control of RNP granules. Frontiers in Molecular Neuroscience, 10, 84.CrossRefGoogle Scholar
  34. 34.
    Lavut, A., & Raveh, D. (2012). Sequestration of highly expressed mRNAs in cytoplasmic granules, P-bodies, and stress granules enhances cell viability. PLoS Genetics, 8, e1002527.CrossRefGoogle Scholar
  35. 35.
    Burke, K. A., Janke, A. M., Rhine, C. L., & Fawzi, N. L. (2015). Residue-by-residue view of in vitro FUS granules that bind the C-terminal domain of RNA polymerase II. Molecular Cell, 60, 231–241.CrossRefGoogle Scholar
  36. 36.
    Kim, Y., & Myong, S. (2016). RNA remodeling activity of DEAD box proteins tuned by protein concentration, RNA length, and ATP. Molecular Cell, 63, 865–876.CrossRefGoogle Scholar
  37. 37.
    Wheeler, J. R., Jain, S., Khong, A, & Parker, R. (2017). Isolation of yeast and mammalian stress granule cores. Methods.Google Scholar
  38. 38.
    Roy, R., Hohng, S., & Ha, T. (2008). A practical guide to single-molecule FRET. Nature Methods, 5, 507–516.CrossRefGoogle Scholar
  39. 39.
    Hellman, L. M., & Fried, M. G. (2007). Electrophoretic mobility shift assay (EMSA) for detecting protein-nucleic acid interactions. Nature Protocols, 2, 1849–1861.CrossRefGoogle Scholar
  40. 40.
    Riback, J. A., Katanski, C. D., Kear-Scott, J. L., Pilipenko, E. V., Rojek, A. E., Sosnick, T. R., & Drummond, D. A. (2017). Stress-triggered phase separation is an adaptive, evolutionarily tuned response. Cell, 168, 1028–1040 e1019.Google Scholar
  41. 41.
    Jain, A., Liu, R., Ramani, B., Arauz, E., Ishitsuka, Y., Ragunathan, K., et al. (2011). Probing cellular protein complexes using single-molecule pull-down. Nature, 473, 484–488.ADSCrossRefGoogle Scholar
  42. 42.
    Jain, A., Liu, R., Xiang, Y. K., & Ha, T. (2012). Single-molecule pull-down for studying protein interactions. Nature Protocols, 7, 445–452.CrossRefGoogle Scholar
  43. 43.
    Feric, M., Vaidya, N., Harmon, T. S., Mitrea, D. M., Zhu, L., Richardson, T. M., et al. (2016). Coexisting liquid phases underlie nucleolar subcompartments. Cell, 165, 1686–1697.CrossRefGoogle Scholar
  44. 44.
    Betzig, E., Patterson, G. H., Sougrat, R., Lindwasser, O. W., Olenych, S., Bonifacino, J. S., et al. (2006). Imaging intracellular fluorescent proteins at nanometer resolution. Science, 313, 1642–1645.ADSCrossRefGoogle Scholar
  45. 45.
    Rust, M. J., Bates, M., & Zhuang, X. (2006). Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy (STORM). Nature Methods, 3, 793–795.CrossRefGoogle Scholar
  46. 46.
    Klar, T. A., & Hell, S. W. (1999). Subdiffraction resolution in far-field fluorescence microscopy. Optics Letters, 24, 954–956.ADSCrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC, part of Springer Nature 2019

Authors and Affiliations

  1. 1.Biophysics DepartmentJohns Hopkins UniversityBaltimoreUSA
  2. 2.Department of Biophysics and Biophysical ChemistryJohns Hopkins UniversityBaltimoreUSA
  3. 3.Center for Physics of Living Cells, University of IllinoisUrbanaUSA

Personalised recommendations