Translation Repressors, an RNA Helicase, and Developmental Cues Control RNP Phase Transitions during Early Development

Developmental Cell

Article Translation Repressors, an RNA Helicase, and Developmental Cues Control RNP Phase Transitions during Early Development Arnaud Hubstenberger,1 Scott L. Noble,1,2 Cristiana Cameron,1 and Thomas C. Evans1,* 1Department

of Cell and Developmental Biology Program in Molecular Biology University of Colorado Anschutz Medical Campus, Aurora, CO 80045, USA *Correspondence: [email protected] http://dx.doi.org/10.1016/j.devcel.2013.09.024 2Graduate

SUMMARY

Like membranous organelles, large-scale coassembly of macromolecules can organize functions in cells. Ribonucleoproteins (RNPs) can form liquid or solid aggregates, but control and consequences of these RNP states in living, developing tissue are poorly understood. Here, we show that regulated RNP factor interactions drive transitions among diffuse, semiliquid, or solid states to modulate RNP sorting and exchange in the Caenorhabditis elegans oocyte cytoplasm. Translation repressors induce an intrinsic capacity of RNP components to coassemble into either large semiliquids or solid lattices, whereas a conserved RNA helicase prevents polymerization into nondynamic solids. Developmental cues dramatically alter both fluidity and sorting within large RNP assemblies, inducing a transition from RNP segregation in quiescent oocytes to dynamic exchange in the early embryo. Therefore, large-scale organization of gene expression extends to the cytoplasm, where regulation of supramolecular states imparts specific patterns of RNP dynamics.

INTRODUCTION Living cells organize functions not only by membrane compartmentalization, but also by assembling supramolecular structures within aqueous environments. Small-scale molecular complexes are built by stereospecific interactions, many of which have been defined at angstrom resolution. By contrast, assembly, functions, and regulation of higher order superstructures are poorly understood. Supramolecular assemblies are emerging as a prominent feature of gene expression pathways. Chromatin is organized into specific domains within the nucleus in a cell-type-specific manner (Bickmore and van Steensel, 2013). Upon transcription, RNAs associate with proteins to form ribonucleoprotein complexes (RNPs), and specific RNPs often coassemble into a remarkable diversity of large RNP granules or domains. In the nucleus, these structures include the nucleolus, Cajal bodies,

and a variety of other nuclear RNP particles (Mao et al., 2011). Diverse RNP assemblies are also common in the cytoplasm and include P-bodies (PBs), stress granules (SGs), neuronal granules, U-bodies, germ granules, and a variety of PB/SGrelated granule types that form in early development (Liu and Gall, 2007; Anderson and Kedersha, 2009; Voronina et al., 2011; Decker and Parker, 2012). Like chromatin, these RNP assemblies are regulated by developmental programs and cell state changes, suggesting important roles in controlling cell fates. Phase transition theory has recently been used to explain large-scale organization of RNPs and other complexes (Weber and Brangwynne, 2012). Driven by reversible multivalent interactions, RNPs or other molecules can transition among diffuse, liquid, or solid states (Figure 1D). Reconstituted in vitro experiments revealed that various RNA-binding proteins can undergo liquid-liquid or liquid-solid demixing to form dynamic hydrogels (Han et al., 2012; Kato et al., 2012; Li et al., 2012). Liquid-like condensation is strongly supported for germ granules in Caenorhabditis elegans embryos and for nucleoli in Xenopus oocytes (Brangwynne et al., 2009, 2011). However, RNPs can also polymerize into solid structures. Prion-like domains are relatively common in RNP factors, and some may induce stable aggregation (Alberti et al., 2009; Si et al., 2010; Heinrich and Lindquist, 2011). Solid RNP aggregates are often associated with neurological disorders (King et al., 2012). These findings suggest that many RNPs can assemble into a variety of supramolecular states, which might be carefully controlled to promote specific functions. This hypothesis predicts that different native RNP assemblies have variable dynamics that are modulated by RNP regulators and cellular control pathways. In this study, we find that precise regulation of RNP phase transitions leads to dramatic control of supramolecular states in vivo during early C. elegans development. Repressors of mRNA translation induce competence of specific conserved RNP components to coassemble into large viscoelastic semiliquids. The DEAD-box RNA helicase CGH-1/RCK/DDX6 controls repressor-modified RNP components in part to prevent phase transition to nondynamic solids. Developmental cues modulate these pathways to induce or repress semiliquid assembly and to control both demixing specificity and dynamics. Changes in RNP assemblies allow shifts from segregation in arrested oocytes to dynamic exchange, transformation, and localization during active early development.

Developmental Cell 27, 161–173, October 28, 2013 ª2013 Elsevier Inc. 161

Developmental Cell Control of RNP Phase Transitions

Figure 1. RNP Granules in C. elegans and Phase Transition Model of RNP Coassembly (A–C) Diagram (A) and live images (B and C) of RNP transformations during the germ cell to embryo program. Germline stem cells generate meiotic nuclei that progress (large arrows) into oocytes and embryos, which generate distinct RNP granules (grPBs, germ granules, and embryo PBs). (B) During oogenesis, GFP:CAR-1 (green) labeled grPBs and PGL-1:RFP (red) labeled germ granules. Germ granules were nuclear attached in early meiosis (large arrowhead) remained distinct from grPBs in oogonia (small arrowhead) but merged in late oocytes (small arrow). (C) In a 6- to 8-cell embryo, GFP:CAR-1 and PGL-1:RFP both entered germ granules (arrow) in germ precursor cell. GFP:CAR-1 also labeled somatic granules (embryo PBs, small arrowhead) that are distinct from grPBs and germ granules. Green granules in germ blastomere (large arrowhead) are likely embryo PB precursors (Gallo et al., 2008). (D) Altered RNP (M) interactions cause transitions to different states. Inhibited interactions promote diffuse ‘‘gas-like’’ RNP distribution. Dynamic multivalent interactions induce demixing to liquid droplet. Stable polymerization forms solids, which could be ordered (depicted) or disordered (as in prion aggregates). (E–G) In live C. elegans oocytes, GFP:CAR-1 distributions suggest three distinct states. (E) In active oocytes, GFP:CAR-1 suggests a diffuse state. (F) In arrested oocytes, GFP:CAR-1 labeled grPBs suggestive of liquid droplets. (G) After CGH1/RCK/DDX6 loss, GFP:CAR-1 labeled square granules suggestive of a solid state. Scale bars, 5 mm.

RESULTS Cytoplasmic RNPs control a progression of germ cell to early embryo development (Lasko, 2009). In C. elegans, several different RNP granule types assemble during this program (Figure 1A) (Schisa et al., 2001; Boag et al., 2005, 2008; Gallo et al., 2008; Jud et al., 2008; Noble et al., 2008; Voronina et al., 2011). All of these assemblies have similarities to PBs and SGs of other cells and share some components with each other. However, all differ in composition, morphology, and dynamics. Germ granules (P granules) carry germline-specific factors like PGL-1 and shift from nucleus-bound to cytoplasm during oogenesis (Figures 1A and 1B) (Updike and Strome, 2010). Distinct large RNP assemblies (here called grP-bodies or grPBs) form in arrested oocyte cytoplasm, where they recruit repressed mRNAs but exclude translationally activated mRNAs (Figures 1A and 1B) (Schisa et al., 2001; Noble et al., 2008). These grPBs contain several PB proteins and mRNA control factors including the Lsm protein CAR-1 (RAP55 in human, Trailer Hitch in fruit fly), the DEAD box RNA helicase CGH-1 (RCK/DDX6 in human, Dhh1p in yeast, Me31b in fruit fly), and mRNA-binding translational repressors such as PUF-5 and MEX-3 (Boag et al., 2008; Jud et al., 2008; Noble et al., 2008). Germ granules remain distinct from grPBs through midstages of oogenesis but merge

with grPBs as oocytes differentiate (Figures 1A and 1B) (Noble et al., 2008). Remarkably, the assembly of large oocyte grPBs is inhibited by sperm signals, which cause RNPs to disperse (Schisa et al., 2001; Jud et al., 2008; Noble et al., 2008). During embryogenesis, RNP factors form PB-like assemblies distinct from grPBs (Figures 1A and 1C) (Audhya et al., 2005; Boag et al., 2005, 2008; Lall et al., 2005; Squirrell et al., 2006; Gallo et al., 2008; Noble et al., 2008). These observations suggest specific control of large-scale RNP coassembly during early development. To investigate RNP coassembly, we examined dynamics of GFP and RFP protein fusions in gonads of living nematodes. In live animals, GFP:CAR-1 and PGL-1:RFP assembled into RNP granules in patterns indistinguishable from endogenous proteins, and fusions did not alter abundance, sizes, or transformations of RNP assemblies (Figures 1B, 1C, and 1E–1G; Supplemental Experimental Procedures available online). In gonads activated by sperm, GFP:CAR-1 labeled small (<250 nm) and diffusely distributed particles in all oocytes (Figure 1E). Thus, ‘‘active oocytes’’ with small RNPs include all oocytes in diakinesis of meiotic prophase (analyzed here), and not just the last oocyte undergoing meiotic maturation (excluded from analysis) (McCarter et al., 1999). In arrested oocytes of live animals lacking sperm, GFP:CAR-1 condensed into large grPBs of up

162 Developmental Cell 27, 161–173, October 28, 2013 ª2013 Elsevier Inc.

Developmental Cell Control of RNP Phase Transitions

Figure 2. RNPs in Activated Oocytes Have Regulated Mobility, Leading to a Diffuse State (A) GFP:CAR-1 before and after photobleaching a 2 mm zone ‘‘B’’ in active oocytes (N = nucleus). (B–D) FRAP of GFP:CAR-1 and GFP:CCF-1 suggests complete mobility of RNP complexes. A two-phase exponential fitted recovery of both GFP:CAR-1 and GFP:CCF-1 in active oocytes (B), yielding two half-times for both proteins (D), whereas a far faster single exponential fitted FRAP of GFP alone (C and D) (note exponential scale in D). (E) Spatiotemporal analysis (kymograph) showed GFP:CAR-1 recovery from periphery to bleach zone ‘‘B’’ center, supporting diffusion-coupled mobility. (F) GFP:CAR-1 was nearly homogenous from cortex to nucleus (near center) of activated oocytes suggesting diffuse state. Error bars represent SEM (B and C) or 95% confidence intervals (D and F). Scale bars, 2 mm.

to 10 mm (Figure 1F), similar to endogenous CAR-1 in fixed gonads (Jud et al., 2008; Noble et al., 2008). However, live arrested oocytes revealed grPBs with smooth spherical surfaces, subcompartments, and protrusions, suggestive of a liquid-like state (Figure 1F). GFP:CCF-1 also localized to grPBs in live arrested oocytes, suggesting that this RNA deadenylase enzyme coassembles with CAR-1 and other grPB factors (Figure 4C). Interestingly, after loss of the RNA helicase CGH-1, GFP:CAR-1 accumulated into large sheet-like granules (Figure 1G), very similar to structures seen in fixed cgh-1(lf) gonads (Audhya et al., 2005; Boag et al., 2008; Noble et al., 2008). These results suggest that oocyte grPB RNPs may undergo regulated phase transitions from diffuse to different condensed states that can be either liquid-like or solid-like (Figure 1D). Small RNP Particles of Activated Oocytes Distribute Broadly by Regulated Diffusion RNP distribution in sperm-activated oocytes is suggestive of a decondensed ‘‘freely mixing’’ state, in which soluble RNP complexes diffuse to occupy available cytoplasmic space (Figures 1D and 1E). To test this, we measured fluorescence recovery after photobleaching (FRAP) of GFP:CAR-1 (Figures 2A and 2B). As predicted, recovery was total suggesting that most, if not all, GFP:CAR-1 is mobile (Figure 2B). Surprisingly however, GFP:CAR-1 FRAP rates were biphasic and
50-fold slower than untagged GFP (Figures 2B–2D). GFP:CCF-1 FRAP rates were similar to CAR-1 supporting that these rates reflect mobility of RNP complexes (Figures 2B and 2D). However, these FRAP rates may not result from free diffusion alone. Observed RNP sizes (<250 nm) were far smaller than those predicted from simple diffusion (
3 mm), suggesting that reversible interactions or

exclusion effects may suppress RNP mobility (Supplemental Experimental Procedures). Two recovery rates likely reflect heterogeneity of RNP complexes or other interacting components. FRAP occurred progressively from the periphery toward the center of bleach zones, at rates dependent on bleach area size (Figure 2E). These dynamics are consistent with a diffusion-coupled mobility, in which interactions or exclusion effects delay effective diffusion (Luby-Phelps et al., 1986; Sprague and McNally, 2005). As predicted for a diffuse state, GFP:CAR-1 intensities were mostly constant throughout the oocyte cytoplasm, with slight concentration in the cortex (Figure 2F). Therefore, small RNPs of activated oocytes retain access with the cytosol due to a diffusive noncondensed or ‘‘soluble’’ state. This ‘‘soluble state’’ could consist of individual RNP complexes in a ‘‘gas-like’’ distribution, small multimeric RNP coassemblies or more likely a combination. Oocyte Arrest Induces a Phase Transition to Viscoelastic Droplets that Segregate RNPs In gonads lacking sperm, GFP:CAR-1 and other RNP factors concentrate into large grPB assemblies (Figures 1F and 3A). RNP superassembly could result from a phase transition, where RNPs condense into liquid grPB droplets by a demixing process, as for germ granules (Figure 1D) (Brangwynne et al., 2009). Alternatively, grPB factors could assemble into solid polymers, similar to viral particles or ribosomes. Liquid-liquid demixing typically produces a system of coalescing droplets that lead to a scale-independent, broad size distribution (Weitz and Lin, 1986; Takayasu et al., 1988). Indeed, within each gonad, the proportion (P) of grPBs of various volumes (V) was consistent with an inverse ‘‘power-law’’ relationship (P(V)
1/V
V1) that was scale

Developmental Cell 27, 161–173, October 28, 2013 ª2013 Elsevier Inc. 163

Developmental Cell Control of RNP Phase Transitions

Figure 3. RNPs Segregate into Viscoelastic, Semiliquid Droplets in Arrested Oocytes (A) Most grPBs are quasistatic in arrested oocytes of live worms. False colored z stack projections of approximately three to five oocytes are shown at t = 0 (green), t = 2 hr (red), and with two images merged. (B) Size distribution of grPBs support liquid-liquid demixing and coalescence. Green line is theoretical distribution for pure liquid condensation (slope of 1.5) (Brangwynne et al., 2011). Red line fits observed distribution, with a slope of 1 suggesting bias toward large grPBs. (C) grPBs (>1 mm) tended toward a spherical aspect ratio (AR) of 1, as predicted for liquids. (D and E) After gonad extrusion, grPBs fused and relaxed to spherical shapes like liquids, but with high viscosity. Plots of large (D) and small (E) fusing droplet AR over time yielded single exponential relaxation to spheres. (legend continued on next page)

164 Developmental Cell 27, 161–173, October 28, 2013 ª2013 Elsevier Inc.

Developmental Cell Control of RNP Phase Transitions

independent (Figure 3B). This distribution is similar to other RNP liquid condensations but with a slightly steeper slope, suggesting bias toward large grPBs (Figure 3B) (Brangwynne et al., 2011). GFP:CAR-1 droplets in live animals also had aspect ratios that tended toward
1, consistent with a liquid-like state driven toward spherical shapes by surface tension (Figure 3C). These observations support the possibility that grPBs are formed by a liquid phase transition. The induction of grPBs from active to arrested oocytes could involve condensation of ‘‘gas-like’’ individual RNP complexes, coalescence of small grPB droplets to large, or some combination. Regardless, these findings are consistent with a shift in steady state toward phase transition upon oocyte arrest. Macroscopic features suggest grPBs have both liquid and solid characteristics. Several grPBs exhibited protrusions, suggestive of fusion or fission events (Figures 1F and 3A). Surprisingly, however, grPB shapes remained stable, and many granules did not move over 1 hr in intact live animals indicating very high effective viscosities combined with possible tethering to cytosolic structures (Figure 3A). After extruding intact gonads from dissected animals, granules became slowly mobile, and fusion events were sometimes detected. Merging grPBs relaxed toward a larger rounded droplet, as expected for liquid droplets (Figures 3D and 3E). However, slow relaxation kinetics estimated a relatively high viscosity (
103 Pa 3 s), roughly similar to thick peanut butter, two orders of magnitude more viscous than previous estimates of C. elegans germ granules but similar to Xenopus nucleoli (Figure 3F; Supplemental Experimental Procedures) (Brangwynne et al., 2009, 2011). Moreover, moderate mechanical stress induced grPB elongation that rapidly recovered in seconds, in contrast to slow droplet fusion relaxation of tens of minutes (Figures 3G and 3H). This result indicates elasticity of a transient polymeric state. High stress loads induced grPB elongation that often failed to recover, consistent with a partly solid-like character (Figure 3I). Together, these properties suggest that collective polymeric interactions among many CAR-1 and CCF-1 RNP complexes induce a semiliquid/semisolid, viscoelastic state. If large grPB droplets have liquid characteristics, RNP components should move randomly within them. To test this, we measured FRAP of GFP:CAR-1 in a 2 mm area within larger grPBs (Figure 3J). Indeed, photobleached GFP:CAR-1 completely exchanged with fluorescent GFP:CAR-1 within grPBs (Figures 3J and 3K). Moreover, spatiotemporal pattern

and rate analysis showed that recovery progressed from the periphery to the center with single-phase kinetics (Figures 3K and 3L). This behavior indicates diffusion-like mobility within grPBs, consistent with a liquid-like state. However, mobilities were 20–50 times slower than in activated oocytes, and at least an order of magnitude slower than for all other cytoplasmic RNP granules examined thus far (Figure 3N) (Kedersha et al., 2000, 2005; Andrei et al., 2005; Leung et al., 2006; Aizer et al., 2008; Chalupnı´kova´ et al., 2008; Mollet et al., 2008; Zhang et al., 2011). We estimated viscosity from FRAP, which roughly and independently agreed with fusion relaxation kinetics (
102–103 Pa 3 s, Supplemental Experimental Procedures). Slowed mobility was not due to imaging conditions since arrested and sperm-activated nematodes were identically treated. Mobility was also not influenced by anesthetics used for most experiments and did not decrease with animal age (Figure S1). Taken together with the viscous and elastic nature of grPB droplets, these results show that large-scale RNP interactions greatly suppress RNP exchange, with dynamic reversibility that maintains some RNP mobility. Furthermore, FRAP, fusion kinetics, and elastic behavior indicate that the Lsm protein CAR-1/ RAP55 and the deadenylase CCF-1 are components of a dynamically polymerizing/depolymerizing matrix within these RNP assemblies. Large grPBs with slow dynamics suggest they effectively inhibit movements of RNPs between grPBs and the cytosol. To test this, we bleached whole granules of
2 mm (Figure 3J). Whole granule FRAP rates were approximately three times slower than same-sized areas within larger grPBs (Figures 3K and 3N). In addition, whole granule recovery was spatially uniform over time (Figure 3M). Therefore, GFP:CAR-1 exchange at the grPB-cytosol interface is rate limiting. If so, large granules with smaller surface to volume ratios should have very slow turn over. Indeed, granules >4 mm failed to recover cytosolic GFP:CAR-1 within 1 hr supporting this prediction (Figure 3K). Moreover, GFP:CAR-1 and other factors are highly concentrated in grPBs; GFP:CAR-1 was almost undetectable in cytosol, and similar localization was seen for several other grPB factors in fixed gonads, including PUF-5, CGH-1, and some repressed mRNAs (Figures 5M–5O) (Noble et al., 2008). Therefore, the macroscopic properties of grPBs influence individual RNPs: dynamic polymeric interactions within large viscoelastic droplets sharply segregate some RNP components from the cytosol. For largest grPBs (up to 10 mm), an RNP component in their center

(F) Length scales (L[mm]) of fusions against relaxation time (t[s]) yielded capillary velocity (slope), which is surface tension over viscosity (g/h
0.008 mm/s) supporting a high viscosity similar to thick peanut butter (
103 Pa 3 s). (G) After moderate pressure stress, grPB AR relaxed in seconds suggesting grPBs are elastic. (H) Elastic recovery was far faster than fusion relaxation, supporting elasticity of transient polymeric state. (I) Higher stresses induced stretch to long thin grPBs that failed to recover, consistent with semisolid character. (J–N) GFP:CAR-1 inside grPBs has slow random mobility, and limited exchange with cytosol. (J) Time-lapse image of GFP:CAR-1 before and after photobleaching of 2 mm zones with whole small grPBs (red arrow) or within large grPBs (light arrow). (K) GFP:CAR-1 FRAP was assayed within large grPBs (intra-grPB, blue; n = 7) normalized to whole grPB. GFP:CAR-1 exchange from cytosol to small 1–2 mm grPBs (grPB-cyto 2 mm, red, see red arrow in A) or cytosol to large (grPB-cyto 5 mm, green) grPBs was determined by whole grPB FRAP normalized to unbleached grPBs in imaged field. Single exponentials fit FRAP curves. (L) Kymograph shows spatiotemporal GFP:CAR-1 movements from grPB periphery to center indicating diffusion-coupled mobility. (M) Kymograph of whole small gPB FRAP revealed homogenous recovery suggesting cytosol to grPB mobility is rate limiting. (N) FRAP half-times t(1/2) revealed strong inhibition of RNP mobility upon oocyte arrest. GFP:CCF-1 and GFP:CAR-1 dynamics were similar, independently supporting mobility of native RNPs. (O) In arrested oocytes, grPBs are enriched at oocyte cortex (Compare with Figure 2F). Whisker plots (H) show minimum to maximum values (whiskers), and quartiles (boxes). Graphs show mean ± SEM (K), or 95% confidence interval (N and O). Scale bars, 5 mm (A and J); 1 mm (D, E, G, and I). See also Figure S1.

Developmental Cell 27, 161–173, October 28, 2013 ª2013 Elsevier Inc. 165

Developmental Cell Control of RNP Phase Transitions

will rarely enter the cytosol. Thus, the bias toward large assemblies favors segregation in arrested oocytes. Several observations suggest that grPBs have additional higher order organization by external cytoplasmic elements. First, grPBs were largely immobile in intact animals (Figure 3A). Second, grPBs were significantly enriched in the arrested oocyte cortex (Figure 3O). Third, grPBs had relatively regular spacing within the oocyte cortical regions, in contrast to square sheet particles of cgh-1(lf) worms (Figure 4F, see below). These results suggest that grPBs are specifically tethered to cortical elements, which further supports a role in segregation of specific RNPs from other cytosolic activities. The DEAD Box RNA Helicase CGH-1/RCK Prevents RNP Factor Polymerization to a Structured Solid State The induction of sheet structures after CGH-1 loss suggests that this conserved DEAD box RNA helicase modulates granule fluidity or organization (Figures 1G and 4A) (Audhya et al., 2005; Boag et al., 2008; Noble et al., 2008). We therefore investigated GFP:CAR-1 dynamics after cgh-1(RNAi) or in a cgh1(tn691) mutant (both referred to as cgh-1(lf)). Sheet formation occurred both in the presence and absence of sperm (Figure 4B). Remarkably, cgh-1(lf) sheets had precise geometric features: almost all sheets were nearly perfect thin squares, with mean length/width ratio of 0.99, cornices with 90.3 angles, and depths of <300 nm (Figures 4A–4D). Sheet size varied up to
10 um, but all sheets retained a square geometry, suggesting regular geometrically constrained growth (Figures 4D and 4E). In addition, some GFP:CCF-1 also localized to the square sheets (Figure 4C). At least some specific mRNAs also associate with cgh-1(lf) sheets in fixed gonads (Noble et al., 2008). By contrast, another grPB factor, GFP:PATR-1 (see below) became completely diffuse in the cytosol after cgh-1(RNAi) and PGL1:RFP only associated with sheets transiently during oogenesis (Figure S2). These results suggest that CGH-1 loss directly or indirectly induces transformations of a subset of RNP factors into regular polymeric arrays that grow by highly ordered component assembly. Their precise geometry strongly suggests these arrays result from specific interactions and not random aggregation. Unlike grPBs, cgh-1(lf) sheets were resistant to mechanical stress, as expected for relatively strong interactions within a solid (Figure 4I). Moreover, after photobleaching one-half of individual sheets,
75% of GFP:CAR-1 failed to recover (>1 hr), whereas the remaining
25% recovered with rapid kinetics (t[1/2] = 58 s) and a spatiotemporal pattern suggesting recovery from cytosol (Figures 4G and 4H). We conclude that an immobile pool of CAR-1 resides within stable solid sheets, whereas a second pool rapidly exchanges with the solid surface. Cytosolic levels of GFP:CAR-1 were higher in cgh-1(RNAi) arrested oocytes than in controls, and GFP:CAR-1 mobility within the cytosol was similar to the fastest pool in activated oocytes (t[1/2] = 40 s) (Figure 4G). Collectively, these results suggest that this RNA helicase inhibits stable polymerization of CAR-1 or a CAR-1 complex into a solid state and controls grPB composition, either as a consequence or independently of polymerization. Therefore, CAR-1 or an RNP factor complex has the intrinsic capacity to polymerize into nondynamic and ordered solid square lattices. This capacity could contribute to coassembly of dynamic grPB semiliquids in wild-type oocytes.

Translational Repressors Promote RNP Phase Transitions that Are Regulated by the CGH-1 Helicase During oogenesis, many mRNAs are translationally repressed and selectively coassemble with their regulators in grPB droplets (Noble et al., 2008). Therefore, repressed mRNPs may influence grPB condensation. PUF-5 and PUF-6/7 bind specific 30 untranslated mRNA elements and redundantly repress a set of mRNAs during late oogenesis (Lublin and Evans, 2007; Stumpf et al., 2008; Hubstenberger et al., 2012). We therefore asked if PUF-5/6/7 influences grPB dynamics. Surprisingly, loss of PUF-5 alone strongly inhibited condensation and activated mobility of GFP:CAR-1 (Figures 5A–5F and 5J–5L). We previously found that PUF-5 depletion alone also caused derepression of several mRNAs, but most oocytes formed and arrested (due to PUF-6/7 redundancy) indicating that grPB disruption is not due to defective oocyte arrest (Hubstenberger et al., 2012). Loss of grPB assembly was specific to late oogenesis, where PUF-5 is expressed and functions and was not seen in distal gonads, indicating that grPB dynamics depend on PUF-5 protein and not puf-5 mRNA per se (data not shown, see Figure 5S). In fixed arrested oocytes, depletion of PUF-5 and PUF-6/7 disrupted coassembly of glp-1 mRNA and CGH-1 in grPBs, showing that PUF-5/6/7 are required for coassembly of multiple grPB components (Figures 5M–5R). GFP:CAR-1 FRAP rates in puf-5(RNAi) arrested oocytes were 100 times faster than control RNAi (Figures 5J and 5L). Even in activated oocytes, PUF-5 loss induced a significant mobility increase of GFP:CAR-1, suggesting that PUF-5 influences CAR-1 RNPs in the diffuse state (Figures 5K and 5L). Therefore, PUF-5 is required for suppression of RNP dynamics, and for grPB coassembly. To test if other mRNA repressors controlled grPB dynamics, we also knocked down PUF-3/11 proteins. PUF-3/ 11 proteins bind different 30 UTR elements to repress a distinct subset of mRNAs during late oogenesis (Stumpf et al., 2008; Hubstenberger et al., 2012). Loss of PUF-3/11 significantly reduced grPB size and increased the fraction of diffuse cytosolic GFP:CAR-1 (Figures 5G–5I). Together, these results show that grPB assembly, growth, and dynamics depend on mRNA repressors, consistent with a model where repressed mRNP states or repression of specific mRNAs are essential for condensation. The apparently opposing functions of PUF proteins and CGH-1/RCK in controlling CAR-1 assembly and mobility prompted us to ask how these activities may interact. Remarkably, whereas cgh-1(tn691) mutants produced only large square sheets, loss of PUF-5 in cgh-1(tn691) animals prevented square sheet formation and instead produced very small GFP:CAR-1 puncta (Figures 5S and 5T). Similarly, PUF-3/11 depletion in cgh-1(tn691) worms transformed square sheets to small, rounded assemblies (Figures 5S and 5T). Furthermore, sheet transformations were specific to oogonia and later oocytes, where both PUFs function, as sheets still formed in the distal gonad (Figure 5S). Collectively, these findings suggest that these PUFs are required for RNP competency to coassemble into either semiliquid droplets or solid polymers; the CGH-1 RNA helicase could act on repressor-altered RNPs in part to prevent a nondynamic solid state. If so, other RNA-binding repressors may have similar interactions with CGH-1. In support of this prediction, we found that loss of the KH domain repressor

166 Developmental Cell 27, 161–173, October 28, 2013 ª2013 Elsevier Inc.

Developmental Cell Control of RNP Phase Transitions

Figure 4. The RNA Helicase CGH-1/RCK/DDX6 Prevents a Nondynamic Solid State (A–C) Loss of CGH-1 induced GFP:CAR-1 (A,B) and GFP:CCF-1 (C) into thin square sheets in live animals. Z stack projections (A) and single confocal sections (B and C) are shown. Either temperature upshift of cgh-1(tn691) or cgh-1(RNAi) induced granules that are square (top/bottom view in A, restrictive) and very thin (<300 nm; lateral views in (B) and (C), cgh-1(RNAi)). (B) Square cgh-1(lf) sheets form in both the presence and absence of sperm. (D) Sheet granules are nearly precise squares. Side length ratios (l/l = s), and intersection angles (q), were narrowly distributed around 1 and 90 respectively. (E) Square sheet sizes varied around
6 mm peak. (F) RNP granule distribution is regulated. After cgh-1(RNAi), sheet granules had random distribution, whereas control grPB distributions suggest regular spacing. (G and H) Majority of GFP:CAR-1 (
75%) in square sheets is immobile, whereas minor fraction rapidly exchanges with sheet surface. (G) Half of cgh-1(lf) square sheets were photobleached and FRAP rates determined (‘‘granule’’ curve in red). Only 25% recovered (t[1/2] = 58 s) over 2 hr. Nonbleached zones of the same granules (‘‘unbleached’’ in purple) did not lose GFP:CAR-1, supporting immobility of CAR-1 within the sheet. Complete recovery of diffuse cytosolic GFP:CAR-1 (‘‘cytosol’’ in brown) revealed high mobility similar to normal activated oocytes (t[1/2] = 40 s). (H) Kymograph of GFP:CAR-1 FRAP shows spatially uniform, partial recovery whereas nonbleached zones did not change, suggesting rapid exchange of a minor CAR-1 fraction between cytosol and sheet surface. (I) Square sheet granules after high mechanical stress. Error bars represent SEM. Scale bars, 5 mm (A–C, G, and I). See also Figure S2.

GLD-1 also suppressed square sheet formation in earlier stages of cgh-1(lf) gonads, where GLD-1 functions (Figure 5U) (Jones and Schedl, 1995). Therefore, divergent translation repressors

induce the capacity of RNP factors to assemble into either liquid or solid supramolecular phases, the properties of which depend directly or indirectly on the CGH-1 RNA helicase.

Developmental Cell 27, 161–173, October 28, 2013 ª2013 Elsevier Inc. 167

Developmental Cell Control of RNP Phase Transitions

Figure 5. Translation Repressors Induce RNP Factor Competence for Semiliquid or Solid Assembly, which Is Regulated by the CGH-1/RCK Helicase (A–I) Depletion of PUF translation repressors inhibited grPB condensation. Control RNAi (mock(RNAi)) did not alter GFP:CAR-1 (A) and PGL-1:RFP (B) assembly in large grPBs of arrested oocytes (A–C). Following PUF-5 RNAi, GFP:CAR-1 and PGL-1:RFP were diffuse or in small particles (D–F). Following PUF-3 depletion, GFP:CAR-1 particles (G) became small, whereas PGL:RFP (H) remained in distinct small aggregates. (J–L) PUF-5 depletion strongly increased RNP mobility rates. (J and K) FRAP of 2 mm zones are shown for mock(RNAi) (blue) or puf-5(RNAi) (red) in arrested (J) or active (K) oocytes (mock(RNAi) FRAP in (J) was for intra-grPB as in Figure 3K). (L) FRAP half-times (t[1/2]) revealed strong activation of RNP mobility following PUF-5 depletion in arrested oocytes, and significant mobility increase in active oocytes in each of two half-times, with the fast pool showing a strong increase (n = 6–24). (M–R) Fixed arrested gonads were costained for glp-1 mRNA and CGH-1 after mock(RNAi) or puf-5/6/7(RNAi). (S and T) Depletion of PUF-5 or PUF-3/11 in cgh-1(tn691) animals block solid square sheet induction. Whole gonad sections (S) show disruption of sheets was specific to late oogenesis. Oocyte sections (T) show small rounded puncta in puf-5(RNAi);cgh-1(tn691) and puf-3/11(RNAi);cgh-1(tn691) gonads. (U) Depletion of the GLD-1 repressor impairs square sheet formation in distal gonad of cgh-1(tn691) animals (epifluorescence image). Error bars are SEM (J and K) or 95% confidence intervals (L). Scale bars, 5 mm (A–I and M–U).

168 Developmental Cell 27, 161–173, October 28, 2013 ª2013 Elsevier Inc.

Developmental Cell Control of RNP Phase Transitions

Figure 6. RNP Assembly Dynamics and Liquid Demixing Are Regulated during Development (A–I) Within grPBs of arrested oocytes, demixing specificity segregates RNP components into several grPB subdomains. (A–C) GFP:CAR-1 tightly overlapped with the RNA dye SYTO-60 in grPBs. (D–I) PGL-1:RFP localized to subdomains (red arrowheads) that were distinct from GFP:PATR-1 subdomains (D–F; green arrowheads), and also distinct from GFP:CAR-1 domain (G–I). GFP:PATR-1 occupied subregions within CAR-1 domain (compare D to A and G). Spherical subdomains were depleted of RNP components (blue arrowheads). (J–L) By contrast, GFP:CAR-1 and PGL-1:RFP were mixed in embryonic germ granules. (M and N) GFP:CAR-1 and PGL:RFP fluorescence measured across granules (dashed lines in I and L) revealed segregation in grPBs (M), and mixing in embryo germ granules (N). (O) Ratio of GFP:CAR-1 in granules relative to cytosol was higher for grPBs than embryo germ granules. (P) GFP:CAR-1 exchange between granules and cytosol is
100 times faster for embryonic germ granules than for similar size grPBs (
2 mm). Error bars represent 95% confidence intervals (O and P). Scale bars, 2 mm.

Embryogenesis Modulates RNP Droplet Dynamics and Demixing Specificity During development, RNP factors rearrange into different granule types (Figures 1A–1C). Control of RNP coassembly dynamics may underlie these transformations. To test this idea, we analyzed various grPB factors and the germ granule factor PGL-1 in live animals. As oogonia formed, PGL-1 particles detached from nuclei into the cytosol and eventually merged with grPBs in arrested oocytes (Figures 1B and 6D–6I). Strikingly however, PGL-1 within grPBs remained in distinct puncta that did not overlap with a large surrounding CAR-1 domain (Figures 6G–6I and 6M). The RNA-binding dye SYTO-60 tightly colocalized with CAR-1, suggesting that at least the CAR-1 domain is rich with RNAs (Figures 6A–6C). Surprisingly, the mRNA decay factor GFP:PATR-1 also localized to subregions within the CAR-1 domain of grPBs, but many PATR-1 subregions were distinct from PGL-1 puncta (Figures 6D–6F). Furthermore,

GFP:CAR-1 and SYTO-60 surrounded numerous spherical regions devoid of grPB factors (Figures 1F and 6A–6I). Therefore, specific RNP components remain segregated within several grPB subdomains in live animals, revealing clear molecular specificity in substructure that was previously suggested by imaging of fixed gonads (Jud et al., 2008; Noble et al., 2008; Patterson et al., 2011). RNP condensation into ‘‘disordered’’ grPB assemblies is nonetheless driven by molecularly specific interactions, which can prevent mixing of different condensates even within a common coassembly. After sperm-driven activation and fertilization, grPBs disassemble and RNP factors reorganize into coassemblies in the embryo (Figures 1A and 1C) (Boag et al., 2008; Gallo et al., 2008). Embryonic germ granules continually localize to posterior cytoplasm at each cell division and appear to be more liquid than oocyte grPBs (Figure 1C) (Brangwynne et al., 2009; Updike and Strome, 2010). We therefore directly tested whether dynamics of

Developmental Cell 27, 161–173, October 28, 2013 ª2013 Elsevier Inc. 169

Developmental Cell Control of RNP Phase Transitions

Figure 7. Models for Control of RNP Phase Transitions (A) Translation repressors, like PUF proteins, bind mRNAs to generate repressed mRNPs. Repressor-dependent recruitment or alteration of corepressive factors like CAR-1 may be a prerequisite for supramolecular coassembly, or repression of a regulator may indirectly promote coassembly. The CGH-1 (RCK/DDX6, Dhh1p) RNA helicase acts specifically on repressormodified RNPs, to directly or indirectly inhibit assembly into nondynamic solids. (B) Developmental cues modulate RNP phase transitions. Sperm signals activate oocytes, induce grPB decondensation, and activate RNP dynamics. Fertilization produces embryo with far more liquid RNP assemblies (embryo germ granules) and altered mixing specificities of RNP components.

a single component (CAR-1) that is shared by both assemblies changes after the oocyte to embryo transition. Strikingly, GFP:CAR-1 exchange between embryonic germ granules and cytosol was two orders of magnitude faster than grPB-cytosol exchange in arrested oocytes (Figure 6P). GFP:CAR-1 FRAP rates in embryonic germ granules were similar to previously reported PGL-1 diffusion rates (Brangwynne et al., 2009). Thus, FRAP rates and droplet fusion kinetics independently demonstrate that embryonic germ granules are far more fluid (
100fold) than grPBs (Figures 3D–3F, 3J–3N, and 6P) (Brangwynne et al., 2009). In addition, the arrested oocyte to embryo transition induced large shifts in CAR-1 distribution: GFP:CAR-1 was
10fold concentrated in embryonic germ granules over cytosol, compared to >50-fold concentration in oocyte grPBs (Figure 6O). Moreover, embryogenesis triggered alterations in demixing specificity and granule organization. Although GFP:CAR-1 and PGL-1:RFP were segregated within grPBs in arrested oocytes, they tightly overlapped within germ granule domains in embryonic germ blastomeres (compare Figure 6M with 6N). Therefore, phase transitions of CAR-1 complexes are modulated in two ways upon transition from arrested oocyte to early embryo. First, CAR-1 dynamics within and from RNP assemblies are activated, leading to far more extensive droplet-cytosol exchange. Second, RNP modifications allow mixing of components that are segregated in oocyte grPBs. DISCUSSION This study reveals that RNP phase transitions are controlled with surprising precision in early development, leading to starkly different supramolecular states that alter RNP organization and dynamics. In the cytoplasm, a subset of RNP components can transition among diffuse, semiliquid, or solid polymer forms.

Pathways of mRNA regulation control these transitions (Figure 7A). Moreover, developmental cues dramatically alter liquid-like assemblies by regulating coassembly, viscosity, and demixing specificity within droplets (Figure 7B). RNP assemblies transform from viscoelastic semiliquids with subdomains to far more dynamic liquid droplets during early development. Therefore, reversible interactions among thousands of RNP complexes impart regulated patterns of RNP dynamics, and large-scale organization of gene expression pathways in the cytoplasm. Alterations in RNP states mirror shifts in RNP regulation that are critical to early development. In arrested oocytes, high viscosity of RNP droplets strongly inhibits access of repressed RNPs to the cytosol compartment where translation is known to be active (Noble et al., 2008). Because tightly regulated mRNAs are repressed in both active and arrested oocytes, grPB condensation is not essential for repression. However the sharp segregation of RNP factors likely plays a reinforcing, redundant, or modulatory role. Inducers (PUFs and CAR-1) and modulators (CGH-1) of assembly promote mRNA repression and stability, consistent with links among these processes (Lublin and Evans, 2007; Boag et al., 2008; Noble et al., 2008; Hubstenberger et al., 2012). In activated oocytes, RNP droplet dissolution and mobilization may be important to promote rapid changes in RNP organization after fertilization. In the embryo, far more liquid granules favor continuous RNP localization and remodeling that are tied to new patterns of mRNA activity important to cell specification. Collectively, these results suggest that mRNP modification pathways control an underlying capacity of specific RNP components to polymerize, thereby driving or suppressing phase transition to semiliquid granules, and preventing assembly of nondynamic solids (Figure 7A). In one model, RNA-binding proteins induce repressed mRNP states (or RNP factor states) that are then competent for coassembly, and the CGH-1/RCK RNA helicase could modulate assembly-competent RNPs in part to prevent transition to solid polymers and maintain semiliquid

170 Developmental Cell 27, 161–173, October 28, 2013 ª2013 Elsevier Inc.

Developmental Cell Control of RNP Phase Transitions

subdomains. Alternatively, RNA-binding repressors could act indirectly by repressing mRNAs for negative regulators of assembly. Because several different repressors promote coassembly, a simpler view is that they transform various RNPs to an assembly competent state. RNA helicases like CGH-1 remodel RNP structures, which could directly stimulate superstructure dynamics or indirectly maintain RNP components essential for dynamics (Fairman et al., 2004; ErnoultLange et al., 2012; Hilliker, 2012). Regardless, these findings support the idea that RNP remodeling is a prerequisite for RNP supramolecular coassembly and determines biophysical properties of these structures in vivo. Regulated RNP phase transitions appear to be prevalent in the nucleus and cytoplasm of diverse eukaryotes and many cell types, from oocytes to brain neurons. Several common themes are emerging. Although cytoplasmic RNP particles vary in their properties, many specifically recruit repressed mRNPs and several shared protein types, including CAR-1 and CGH-1 homologs (Anderson and Kedersha, 2009; Decker and Parker, 2012). Thus, related mechanisms of RNP transformation may be a common requirement for supramolecular condensation and control. To induce coassembly, some RNP components must have multivalent interaction potential, which is known to drive liquid-liquid demixing in vitro (Han et al., 2012; Kato et al., 2012; Li et al., 2012). These could be RNAs with their multiple sites for binding factors. Multivalent factors could also be proteins: PGL proteins may promote germ granule condensation through multivalent interactions, and many RNP proteins have low complexity or prionlike sequences that can induce sol-gel or liquid-liquid demixing phenomena (Gilks et al., 2004; Alberti et al., 2009; Hanazawa et al., 2011; Updike et al., 2011; Han et al., 2012; Kato et al., 2012; King et al., 2012). So far, the few RNP assemblies examined in vivo have liquid or semiliquid properties, suggesting that liquid-like states may be typical (this study) (Brangwynne et al., 2009, 2011). However, our studies suggest such states are modulated, which can dramatically slow or stimulate RNP dynamics, and also inhibit or promote mixing of subdomains. ATPdependent RNA helicases may commonly regulate such RNP dynamics. In addition to CGH-1/DDX6, Vasa-type RNA helicases promote germ granule integrity in several systems, and nucleolus viscosity increases after ATP depletion (Hay et al., 1988; Kawasaki et al., 1998; Kuznicki et al., 2000; Brangwynne et al., 2011; Updike et al., 2011). Together, these findings suggest that many RNP complexes can inherently coassemble into large superstructures by phase transition. Repressors, helicases, and other factors likely modulate these states through control of reversible interaction kinetics. Defects in these pathways could lie at the heart of solid RNP aggregates common to several neurological disorders (King et al., 2012). EXPERIMENTAL PROCEDURES RNAi and Genetic Manipulations RNAi was done by the feeding method using nonspecific sequence as a control (mock RNAi), as described (Timmons et al., 2001; Hubstenberger et al., 2012). L4 larvae were grown on test or control RNAi plates at 20 C for 48 hr. To induce square sheet formation in cgh-1(tn691ts), animals were grown at 15 up to 36 hr post-L4 stage and then shifted to 25 for 12–20 hr. For RNAi of cgh-1(tn691ts), L4 larvae were grown on RNAi plates at 15 for 36 hr and then shifted to 25 for 12 hr.

Live Imaging Live adult nematodes were anaesthetized in M9 buffer with 2% Tricaine methanesulfonate (Sigma), 0.2% tetramisole hydrochloride (Sigma), mounted on 2% agarose pads with coverslip, and imaged on a Leica TCS SP5 II confocal microscope (using HCX PL APO CS 403 oil objective, 1.25 aperture). Unless stated otherwise, single confocal sections (1 Airy unit pinhole) are shown. For image analysis, ImageJ, and the plugin collection MBF ImageJ for microscopy by Tony Collins, was used (Schneider et al., 2012). The HiLo lookup table was used to ensure below saturation fluorescence and full background subtraction. Nematodes with activated (+sperm) or arrested (sperm) oogenesis were generated in parallel using gravid hermaphrodites (N2) or virgin fog2(q71) females, respectively. Unmated fog-2(q71) females only lack sperm and were indistinguishable from sperm-depleted N2 hermaphrodites (Schedl and Kimble, 1988; Jud et al., 2007, 2008; Noble et al., 2008). To stain RNA in live worms, 100 ml of 0.25 mM SYTO 60 (Invitrogen), 118 mM NaCl, and 48 mM KCl were added on bacterial lawns and fed to nematodes for 15 hr at 20 C. Worms were incubated 15 hr at 20 C, washed, and then mounted on slides in M9 buffer with 30 mM NaN3 or with Tricaine/ tetramisole (as above); both conditions revealed tight SYTO 60 colocalization with GFP:CAR-1 (Figures 6A–6C), but Tricaine/tetramisole had higher cytoplasmic staining (data not shown). SYTO 60 labeled nucleoli but not chromosomal nuclear compartments, supporting specificity for RNA (data not shown). Fluorescence Recovery after Photobleaching Analysis For fluorescence recovery after photobleaching (FRAP), circular zones 2 mm in diameter were photobleached and confocal time series were recorded at minimal laser intensities, using 2 Airy unit pinhole size. Only differentiated oocytes still in meiotic prophase (diakinesis) were imaged (2 to 3 active oocytes, 2 to 5 arrested oocytes). Embryos were imaged at 4- to 8-cell stages. For all FRAP series, imaging-induced bleaching was normalized by dividing fluorescence intensities of bleach zones by unbleached fluorescent areas or by fluorescence of entire image field, which gave similar FRAP rates. Normalized FRAP curves were quantified according to FRAP(t) = [I(t)  I(t0)]/[I(t  1)  I(t0)], where FRAP(t) is relative fluorescence recovered at time t, I(t) is normalized fluorescence at time t, I(t0) is normalized intensity at time 0, and I(t  1) is normalized intensity prior to photobleaching. To determine spatiotemporal FRAP patterns, kymographs were generated by measuring fluorescence across a zone line of interest over time, using ImageJ ‘‘reslice.’’ To mitigate movements during imaging, time series were aligned to an image at t = 0, using ImageJ, Stacks-Shuffling-MultiStackReg plugin. RNP Granule Size Distribution, Shape, and Gonad Organization Analyses To analyze RNP granule size distributions, maximum intensity projections of confocal z stacks were generated and late oocyte regions were manually defined. To delimit granules from cytosol and each other, ImageJ ‘‘smooth,’’ ‘‘threshold,’’ ‘‘watershed’’ functions were combined. Automatic particle analysis of granules >1 mm3 was then used to calculate grPB area, spherical volumes were estimated, granules were binned by volume, and proportion of all granules in each bin was determined (Figure 3B). To analyze normal grPB shape, particle analysis function was used to calculate aspect ratios (ARs), the ratio of major and minor axes lengths of best fitting ellipses (Figure 3C). To analyze cgh-1(lf) square sheet shapes, manually measured side lengths and intersection angles of adjacent sides of individual granules were determined from maximum projections in ImageJ volume viewer. Closest neighbor distance analyses were conducted in two dimensions. To plot GFP:CAR-1 distribution from oocyte cortex to oocyte center (near nucleus), fluorescence was measured along lines drawn from plasma membrane to nuclear envelope. Relative distances were normalized by binning each line in 50 equal units, and average bin fluorescence was plotted against normalized position. Analyses of RNP Granule Fusion and Mechanical Stress Responses To measure fusion events, gonads were extruded from worms dissected on a coverslip in a modified embryo culture medium: 5% (w/v) sucrose, 90% (v/v) L-15 medium (Leibovitz L15 with L-glutamine, without Phenol Red; Gibco), 10% (v/v) fetal calf serum (Gibco) (Wolke et al., 2007). The coverslip was inverted onto a 4% agarose pad in L-15 buffer. Time-lapse movies were

Developmental Cell 27, 161–173, October 28, 2013 ª2013 Elsevier Inc. 171

Developmental Cell Control of RNP Phase Transitions

recorded with 2 Airy unit pinholes, and ARs of pairs of fusing granules were analyzed over time. To impose mechanical stress by pressure over time, dissected worms and extruded gonads were placed in 10 ml dissection buffer between coverslip and slide. Distorted GFP:CAR-1 particle aspect ratios were determined. At early times, elastic responses were detected in oocytes leaking through gonad openings. Higher pressures induced at later times caused grPB extensions passed a tipping point that remained distended. In Situ Detection of RNA and Proteins Dissected and fixed gonads were simultaneously stained for glp-1 mRNA and CGH-1 protein, by combined fluorescent in situ hybridization (FISH) and immunofluorescence, as described (Noble et al., 2008). Antisense and sense RNA probes derived from glp-1 cDNA were made as described (Barbee and Evans, 2006; Noble et al., 2008); only antisense probes stained grPBs and small RNPs ((Noble et al., 2008; data not shown). Similar results were seen with RNA probes to pos-1 mRNA (data not shown). Statistical Analyses and Calculations All statistics analyses were conducted using Prism 5 (GraphPad Software). FRAP recovery curves were fitted using the least square method to determine fit to single or 2 phase exponentials as follows, FRAP = P 3 (1-e(t/t)), or FRAP = P 3 [1-%fast 3 0.01 3 e(t/tfast)-(100-%fast) 3 0.01 3 e(t/tslow)] respectively, where t is time, P the plateau reached at infinite time, and t the time constant. Half-times were calculated as t(1/2) = t 3 ln(2). To compare best-fit values, we used the extra sum-of-squares F test. Diffusion rates and size estimates were estimated from FRAP and the Stokes-Einstein relationship (calculations in Supplemental Experimental Procedures) (Soumpasis, 1983). Viscosities were estimated by two independent approaches: (1) grPB fusion kinetics and (2) FRAP-based diffusion rates within grPBs, as previously described (calculations in Supplemental Experimental Procedures) (Brangwynne et al., 2009, 2011).

Andrei, M.A., Ingelfinger, D., Heintzmann, R., Achsel, T., Rivera-Pomar, R., and Lu¨hrmann, R. (2005). A role for eIF4E and eIF4E-transporter in targeting mRNPs to mammalian processing bodies. RNA 11, 717–727. Audhya, A., Hyndman, F., McLeod, I.X., Maddox, A.S., Yates, J.R., 3rd, Desai, A., and Oegema, K. (2005). A complex containing the Sm protein CAR-1 and the RNA helicase CGH-1 is required for embryonic cytokinesis in Caenorhabditis elegans. J. Cell Biol. 171, 267–279. Barbee, S.A., and Evans, T.C. (2006). The Sm proteins regulate germ cell specification during early C. elegans embryogenesis. Dev. Biol. 291, 132–143. Bickmore, W.A., and van Steensel, B. (2013). Genome architecture: domain organization of interphase chromosomes. Cell 152, 1270–1284. Boag, P.R., Nakamura, A., and Blackwell, T.K. (2005). A conserved RNA-protein complex component involved in physiological germline apoptosis regulation in C. elegans. Development 132, 4975–4986. Boag, P.R., Atalay, A., Robida, S., Reinke, V., and Blackwell, T.K. (2008). Protection of specific maternal messenger RNAs by the P body protein CGH-1 (Dhh1/RCK) during Caenorhabditis elegans oogenesis. J. Cell Biol. 182, 543–557. Brangwynne, C.P., Eckmann, C.R., Courson, D.S., Rybarska, A., Hoege, C., Gharakhani, J., Ju¨licher, F., and Hyman, A.A. (2009). Germline P granules are liquid droplets that localize by controlled dissolution/condensation. Science 324, 1729–1732. Brangwynne, C.P., Mitchison, T.J., and Hyman, A.A. (2011). Active liquid-like behavior of nucleoli determines their size and shape in Xenopus laevis oocytes. Proc. Natl. Acad. Sci. USA 108, 4334–4339. Chalupnı´kova´, K., Lattmann, S., Selak, N., Iwamoto, F., Fujiki, Y., and Nagamine, Y. (2008). Recruitment of the RNA helicase RHAU to stress granules via a unique RNA-binding domain. J. Biol. Chem. 283, 35186–35198.

SUPPLEMENTAL INFORMATION

Decker, C.J., and Parker, R. (2012). P-bodies and stress granules: possible roles in the control of translation and mRNA degradation. Cold Spring Harb. Perspect. Biol. 4, a012286.

Supplemental Information includes Supplemental Experimental Procedures and two figures and can be found with this article online at http://dx.doi.org/ 10.1016/j.devcel.2013.09.024.

Ernoult-Lange, M., Baconnais, S., Harper, M., Minshall, N., Souquere, S., Boudier, T., Be´nard, M., Andrey, P., Pierron, G., Kress, M., et al. (2012). Multiple binding of repressed mRNAs by the P-body protein Rck/p54. RNA 18, 1702–1715.

ACKNOWLEDGMENTS

Fairman, M.E., Maroney, P.A., Wang, W., Bowers, H.A., Gollnick, P., Nilsen, T.W., and Jankowsky, E. (2004). Protein displacement by DExH/D ‘‘RNA helicases’’ without duplex unwinding. Science 304, 730–734.

We thank Geraldine Seydoux, Keith Blackwell, Jayne Squirrell, Karen Oegema, Jim Priess, and the Caenorhabditis Genetics Center (CGC; University of Minnesota) for providing transgenic strains and reagents used in this study. Some strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). We wish to particularly thank David Greenstein for providing several strains, antibodies, and detailed information prior to publication. We also thank Rytis Prekeris, Chad Pearson, Natalia Toledo Melendez, and Richard Davis for valuable discussions. This work was supported by a grant from the National Institutes of Health (GM079682 to T.C.E.). Received: May 31, 2013 Revised: August 27, 2013 Accepted: September 26, 2013 Published: October 28, 2013 REFERENCES Aizer, A., Brody, Y., Ler, L.W., Sonenberg, N., Singer, R.H., and Shav-Tal, Y. (2008). The dynamics of mammalian P body transport, assembly, and disassembly in vivo. Mol. Biol. Cell 19, 4154–4166. Alberti, S., Halfmann, R., King, O., Kapila, A., and Lindquist, S. (2009). A systematic survey identifies prions and illuminates sequence features of prionogenic proteins. Cell 137, 146–158. Anderson, P., and Kedersha, N. (2009). RNA granules: post-transcriptional and epigenetic modulators of gene expression. Nat. Rev. Mol. Cell Biol. 10, 430–436.

Gallo, C.M., Munro, E., Rasoloson, D., Merritt, C., and Seydoux, G. (2008). Processing bodies and germ granules are distinct RNA granules that interact in C. elegans embryos. Dev. Biol. 323, 76–87. Gilks, N., Kedersha, N., Ayodele, M., Shen, L., Stoecklin, G., Dember, L.M., and Anderson, P. (2004). Stress granule assembly is mediated by prion-like aggregation of TIA-1. Mol. Biol. Cell 15, 5383–5398. Han, T.W., Kato, M., Xie, S., Wu, L.C., Mirzaei, H., Pei, J., Chen, M., Xie, Y., Allen, J., Xiao, G., and McKnight, S.L. (2012). Cell-free formation of RNA granules: bound RNAs identify features and components of cellular assemblies. Cell 149, 768–779. Hanazawa, M., Yonetani, M., and Sugimoto, A. (2011). PGL proteins self associate and bind RNPs to mediate germ granule assembly in C. elegans. J. Cell Biol. 192, 929–937. Hay, B., Jan, L.Y., and Jan, Y.N. (1988). A protein component of Drosophila polar granules is encoded by vasa and has extensive sequence similarity to ATP-dependent helicases. Cell 55, 577–587. Heinrich, S.U., and Lindquist, S. (2011). Protein-only mechanism induces selfperpetuating changes in the activity of neuronal Aplysia cytoplasmic polyadenylation element binding protein (CPEB). Proc. Natl. Acad. Sci. USA 108, 2999–3004. Hilliker, A. (2012). Analysis of RNA helicases in P-bodies and stress granules. Methods Enzymol. 511, 323–346. Hubstenberger, A., Cameron, C., Shtofman, R., Gutman, S., and Evans, T.C. (2012). A network of PUF proteins and Ras signaling promote mRNA repression and oogenesis in C. elegans. Dev. Biol. 366, 218–231.

172 Developmental Cell 27, 161–173, October 28, 2013 ª2013 Elsevier Inc.

Developmental Cell Control of RNP Phase Transitions

Jones, A.R., and Schedl, T. (1995). Mutations in gld-1, a female germ cell-specific tumor suppressor gene in Caenorhabditis elegans, affect a conserved domain also found in Src-associated protein Sam68. Genes Dev. 9, 1491– 1504. Jud, M., Razelun, J., Bickel, J., Czerwinski, M., and Schisa, J.A. (2007). Conservation of large foci formation in arrested oocytes of Caenorhabditis nematodes. Dev. Genes Evol. 217, 221–226. Jud, M.C., Czerwinski, M.J., Wood, M.P., Young, R.A., Gallo, C.M., Bickel, J.S., Petty, E.L., Mason, J.M., Little, B.A., Padilla, P.A., and Schisa, J.A. (2008). Large P body-like RNPs form in C. elegans oocytes in response to arrested ovulation, heat shock, osmotic stress, and anoxia and are regulated by the major sperm protein pathway. Dev. Biol. 318, 38–51. Kato, M., Han, T.W., Xie, S., Shi, K., Du, X., Wu, L.C., Mirzaei, H., Goldsmith, E.J., Longgood, J., Pei, J., et al. (2012). Cell-free formation of RNA granules: low complexity sequence domains form dynamic fibers within hydrogels. Cell 149, 753–767.

McCarter, J., Bartlett, B., Dang, T., and Schedl, T. (1999). On the control of oocyte meiotic maturation and ovulation in Caenorhabditis elegans. Dev. Biol. 205, 111–128. Mollet, S., Cougot, N., Wilczynska, A., Dautry, F., Kress, M., Bertrand, E., and Weil, D. (2008). Translationally repressed mRNA transiently cycles through stress granules during stress. Mol. Biol. Cell 19, 4469–4479. Noble, S.L., Allen, B.L., Goh, L.K., Nordick, K., and Evans, T.C. (2008). Maternal mRNAs are regulated by diverse P body-related mRNP granules during early Caenorhabditis elegans development. J. Cell Biol. 182, 559–572. Patterson, J.R., Wood, M.P., and Schisa, J.A. (2011). Assembly of RNP granules in stressed and aging oocytes requires nucleoporins and is coordinated with nuclear membrane blebbing. Dev. Biol. 353, 173–185. Schedl, T., and Kimble, J. (1988). fog-2, a germ-line-specific sex determination gene required for hermaphrodite spermatogenesis in Caenorhabditis elegans. Genetics 119, 43–61. Schisa, J.A., Pitt, J.N., and Priess, J.R. (2001). Analysis of RNA associated with P granules in germ cells of C. elegans adults. Development 128, 1287–1298.

Kawasaki, I., Shim, Y.H., Kirchner, J., Kaminker, J., Wood, W.B., and Strome, S. (1998). PGL-1, a predicted RNA-binding component of germ granules, is essential for fertility in C. elegans. Cell 94, 635–645.

Schneider, C.A., Rasband, W.S., and Eliceiri, K.W. (2012). NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 9, 671–675.

Kedersha, N., Cho, M.R., Li, W., Yacono, P.W., Chen, S., Gilks, N., Golan, D.E., and Anderson, P. (2000). Dynamic shuttling of TIA-1 accompanies the recruitment of mRNA to mammalian stress granules. J. Cell Biol. 151, 1257–1268.

Si, K., Choi, Y.B., White-Grindley, E., Majumdar, A., and Kandel, E.R. (2010). Aplysia CPEB can form prion-like multimers in sensory neurons that contribute to long-term facilitation. Cell 140, 421–435.

Kedersha, N., Stoecklin, G., Ayodele, M., Yacono, P., Lykke-Andersen, J., Fritzler, M.J., Scheuner, D., Kaufman, R.J., Golan, D.E., and Anderson, P. (2005). Stress granules and processing bodies are dynamically linked sites of mRNP remodeling. J. Cell Biol. 169, 871–884.

Soumpasis, D.M. (1983). Theoretical analysis of fluorescence photobleaching recovery experiments. Biophys. J. 41, 95–97.

King, O.D., Gitler, A.D., and Shorter, J. (2012). The tip of the iceberg: RNAbinding proteins with prion-like domains in neurodegenerative disease. Brain Res. 1462, 61–80.

Squirrell, J.M., Eggers, Z.T., Luedke, N., Saari, B., Grimson, A., Lyons, G.E., Anderson, P., and White, J.G. (2006). CAR-1, a protein that localizes with the mRNA decapping component DCAP-1, is required for cytokinesis and ER organization in Caenorhabditis elegans embryos. Mol. Biol. Cell 17, 336–344.

Kuznicki, K.A., Smith, P.A., Leung-Chiu, W.M., Estevez, A.O., Scott, H.C., and Bennett, K.L. (2000). Combinatorial RNA interference indicates GLH-4 can compensate for GLH-1; these two P granule components are critical for fertility in C. elegans. Development 127, 2907–2916. Lall, S., Piano, F., and Davis, R.E. (2005). Caenorhabditis elegans decapping proteins: localization and functional analysis of Dcp1, Dcp2, and DcpS during embryogenesis. Mol. Biol. Cell 16, 5880–5890. Lasko, P. (2009). Translational control during early development. Prog. Mol. Biol. Transl. Sci. 90, 211–254. Leung, A.K., Calabrese, J.M., and Sharp, P.A. (2006). Quantitative analysis of Argonaute protein reveals microRNA-dependent localization to stress granules. Proc. Natl. Acad. Sci. USA 103, 18125–18130. Li, P., Banjade, S., Cheng, H.C., Kim, S., Chen, B., Guo, L., Llaguno, M., Hollingsworth, J.V., King, D.S., Banani, S.F., et al. (2012). Phase transitions in the assembly of multivalent signalling proteins. Nature 483, 336–340. Liu, J.L., and Gall, J.G. (2007). U bodies are cytoplasmic structures that contain uridine-rich small nuclear ribonucleoproteins and associate with P bodies. Proc. Natl. Acad. Sci. USA 104, 11655–11659. Lublin, A.L., and Evans, T.C. (2007). The RNA-binding proteins PUF-5, PUF-6, and PUF-7 reveal multiple systems for maternal mRNA regulation during C. elegans oogenesis. Dev. Biol. 303, 635–649. Luby-Phelps, K., Taylor, D.L., and Lanni, F. (1986). Probing the structure of cytoplasm. J. Cell Biol. 102, 2015–2022. Mao, Y.S., Zhang, B., and Spector, D.L. (2011). Biogenesis and function of nuclear bodies. Trends Genet. 27, 295–306.

Sprague, B.L., and McNally, J.G. (2005). FRAP analysis of binding: proper and fitting. Trends Cell Biol. 15, 84–91.

Stumpf, C.R., Kimble, J., and Wickens, M. (2008). A Caenorhabditis elegans PUF protein family with distinct RNA binding specificity. RNA 14, 1550–1557. Takayasu, H., Nishikawa, I.I., and Tasaki, H. (1988). Power-law mass distribution of aggregation systems with injection. Phys. Rev. A 37, 3110–3117. Timmons, L., Court, D.L., and Fire, A. (2001). Ingestion of bacterially expressed dsRNAs can produce specific and potent genetic interference in Caenorhabditis elegans. Gene 263, 103–112. Updike, D., and Strome, S. (2010). P granule assembly and function in Caenorhabditis elegans germ cells. J. Androl. 31, 53–60. Updike, D.L., Hachey, S.J., Kreher, J., and Strome, S. (2011). P granules extend the nuclear pore complex environment in the C. elegans germ line. J. Cell Biol. 192, 939–948. Voronina, E., Seydoux, G., Sassone-Corsi, P., and Nagamori, I. (2011). RNA granules in germ cells. Cold Spring Harb. Perspect. Biol. 3, 3. Weber, S.C., and Brangwynne, C.P. (2012). Getting RNA and protein in phase. Cell 149, 1188–1191. Weitz, D.A., and Lin, M.Y. (1986). Dynamic scaling of cluster-mass distributions in kinetic colloid aggregation. Phys. Rev. Lett. 57, 2037–2040. Wolke, U., Jezuit, E.A., and Priess, J.R. (2007). Actin-dependent cytoplasmic streaming in C. elegans oogenesis. Development 134, 2227–2236. Zhang, J., Okabe, K., Tani, T., and Funatsu, T. (2011). Dynamic associationdissociation and harboring of endogenous mRNAs in stress granules. J. Cell Sci. 124, 4087–4095.

Developmental Cell 27, 161–173, October 28, 2013 ª2013 Elsevier Inc. 173