Multivalent interactions with RNA drive recruitment and dynamics in biomolecular condensates in Xenopus oocytes.

RNA localization and biomolecular condensate formation are key biological strategies for organizing the cytoplasm and generating cellular polarity. In Xenopus oocytes, RNAs required for germ layer patterning localize in biomolecular condensates, termed Localization bodies (L-bodies). Here, we have used an L-body RNA-binding protein, PTBP3, to test the role of RNA-protein interactions in regulating the biophysical characteristics of L-bodies in vivo and PTBP3-RNA condensates in vitro. Our results reveal that RNA-protein interactions drive recruitment of PTBP3 and localized RNA to L-bodies and that multivalent interactions tune the dynamics of the PTBP3 after localization. In a concentration-dependent manner, RNA becomes non-dynamic and interactions with the RNA determine PTBP3 dynamics within these biomolecular condensates in vivo and in vitro. Importantly, RNA, and not protein, is required for maintenance of the PTBP3-RNA condensates in vitro, pointing to a model where RNA serves as a non-dynamic substructure in these condensates.

Introduction

Subcellular compartmentalization is a conserved mechanism by which cells enrich biomolecules for particular processes, allowing for spatial control of biological activity. Many of these compartments, including stress granules, nucleoli, and germ granules, are biomolecular condensates, enriching proteins and RNAs relative to their surroundings without a lipid membrane (reviewed in Boeynaems et al., 2018). The formation of biomolecular condensates is thought to be driven by multivalent interactions, either between “sticker” domains in protein intrinsically disordered regions (IDRs) or between ordered interaction domains, such as multivalent signaling or RNA-binding proteins (RBPs) (reviewed in Mittag and Parker, 2018). These interactions lead to biomolecular condensates with a wide range of biophysical states, with varying dynamics from liquid to gel to solid (reviewed in Alberti et al., 2019). However, the relative contributions of each of these interaction domains to the formation and dynamics of different types of condensates in vivo is unclear.

In addition to being enriched for certain types of proteins, a conserved feature of many classes of biomolecular condensates is the incorporation of RNA (reviewed in Fay and Anderson, 2018). Emerging research, particularly in vitro studies, suggests that RNA – through both RNA–RNA and RNA–RBP intermolecular interactions – may play a critical role in the structure and assembly of biomolecular condensates (reviewed in Van Treeck and Parker, 2018). However, the role of RNAs and multivalent RBPs in the formation and maintenance of in vivo condensates remains unclear and may vary based on the concentration of the RNA and the type of condensate. In vitro, low concentrations of RNA often promote liquid-liquid phase separation (LLPS) of proteins (Burke et al., 2015; Lin et al., 2015; Molliex et al., 2015; Patel et al., 2015), whereas high concentrations can inhibit LLPS (Banerjee et al., 2017; Maharana et al., 2018). Conversely, many RNAs are capable of protein-free self-assembly (Jain and Vale, 2017; Langdon et al., 2018; Neil et al., 2021; Van Treeck et al., 2018). In vivo, recent studies have shown that RNA may be acting to form a non-dynamic or structural component within a variety of biomolecular condensates (Clemson et al., 2009; Neil et al., 2021; Niepielko et al., 2018; Trcek et al., 2020; Van Treeck et al., 2018).

In Xenopus oocytes, translationally silenced RNAs, including vg1 mRNA, are transported to the vegetal cortex in large cytoplasmic ribonucleoprotein (RNP) granules (reviewed in Cabral and Mowry, 2020). Proper localization of these RNPs is required for germ layer patterning and mislocalization of vegetally localized mRNAs, such as vg1, leads to embryonic lethality (Birsoy et al., 2006; Dale et al., 1993; Thomsen and Melton, 1993). Cis-elements within the vg1 RNA, including polypyrimidine-rich protein binding sites, are required for proper packaging of the RNA into vegetal transport RNPs and localization to the vegetal cortex (Cote et al., 1999; Lewis et al., 2008). Recent work from our laboratory has characterized these RNPs as biomolecular condensates, termed Localization bodies (L-bodies) (Neil et al., 2021). L-bodies are large, irregularly shaped biomolecular condensates that are very highly enriched for localized RNAs. In vivo, L-bodies are comprised of a non-dynamic, RNA-containing component enmeshed by a comparatively dynamic protein layer. However, the mechanisms underlying the formation and maintenance of the biophysical state of L-bodies are not known.

Proteomic analysis of L-bodies revealed a strong enrichment for proteins containing RNA binding domains (RBDs), IDRs, or both (Neil et al., 2021), a conserved feature of biomolecular condensates (reviewed in Banani et al., 2017). Many roles have been described for proteins with IDRs in biomolecular condensates, but the functions of ordered, multivalent interaction domains are less well understood. Given the striking non-dynamic state of L-body RNAs, we were particularly interested in characterizing the role of multivalent RBPs in L-body assembly and dynamics. For this, we focused on polypyrimidine tract binding protein 3 (PTBP3), a previously uncharacterized L-body protein that we show in this work to be highly colocalized with L-bodies. PTBP3 is a paralog of the well-characterized RBP, PTBP1 (Yamamoto et al., 1999). In addition to their roles in RNA localization, PTB proteins, such as PTBP1 (hnRNPI), PTBP2 (nPTB), and PTBP3 (ROD1), are involved in many steps in RNA metabolism depending on their subcellular localization and binding partners, including splicing, polyadenylation, mRNA stability, and translation initiation (reviewed in Hu et al., 2018; Sawicka et al., 2008). In vitro, PTBP1 phase transitions in the presence of its RNA ligand (Banani et al., 2016; Li et al., 2012). PTB proteins contain four RNA Recognition Motifs (RRMs) that each bind polypyrimidine-rich sequences in RNA (Oberstrass et al., 2005), making PTB an ideal model protein for studying multivalent interactions of well-folded interaction domains within biomolecular condensates.

In this work, we elucidate the role of RNA–RBP interactions in the recruitment and dynamics of components of biomolecular condensates using PTBP3, both in vivo and in vitro. First, we demonstrate that PTBP3 is an L-body RNA binding protein with moderate in vivo mobility and show that PTB–RNA binding is required for the localization and dynamics of both protein and RNA in L-bodies. Next, we show that recombinant PTBP3 phase transitions in vitro into non-spherical, solid- or gel-like condensates in an RNA-dependent manner. In vitro, as in vivo, RNA becomes non-dynamic and PTBP3 dynamics are driven by binding to RNA. Whereas RNA-protein interactions are required for in vitro condensate formation, we find that RNA, and not protein, is necessary for condensate maintenance after the degradation of the RNA or protein component, respectively. Finally, we show that a single RNA–RRM interaction is sufficient to target PTBP3 to L-bodies in vivo, whereas multivalent interactions between RNA and protein work in concert to tune the mobility of PTBP3 after localization to L-bodies. Taken together, our results indicate that sequence-specific PTBP3–RNA interactions regulate recruitment of both RNA and protein into L-bodies, but that formation of the non-dynamic RNA component is concentration-dependent, rather than sequence-dependent in vitro. Moreover, this non-dynamic RNA component is required for condensate maintenance in vitro. In L-bodies, it is the strength and number of protein interactions with RNA that drive protein dynamics, suggesting a role for multivalent RNA–RBP interactions in regulating the physical properties of biomolecular condensates.

Results

Localization and dynamics of LE RNA in L-bodies require PTB binding sites

Localization of RNAs to L-bodies can be recapitulated by a minimal localization element (LE RNA) derived from sequences within the 3′ UTR of the vg1 mRNA (Cote et al., 1999; Lewis et al., 2004; Neil et al., 2021). LE RNA contains two pairs of PTB binding sites and strongly localizes within L-bodies (Figures 1A–1C, S1A, and S1B). This localization is not owing to selective degradation of the RNA, as the microinjected LE RNA is stable over 48 h (Figure S1G), and is instead owing to active transport of the RNA (Gagnon et al., 2013; Messitt et al., 2008). To test the role of PTB binding in the localization of RNA to L-bodies, we employed a mutated form of LE RNA, termed mut PTB LE RNA. The mut PTB LE RNA contains three U to A point mutations in each of the four PTB binding sites, but is otherwise identical to the LE RNA (Figure S1A), and has been shown to no longer bind to PTBP1 (Lewis et al., 2004). When microinjected into oocytes that contain endogenous vg1 mRNA, and thus contain L-bodies, mut PTB LE RNA is only slightly localized to endogenous L-bodies and, unlike the LE RNA, is also observed throughout the oocyte cytoplasm (Figures 1C–1C″). Likewise, as seen at higher magnification, whereas the LE RNA is highly localized within L-bodies (Figures 1D, S1C–S1C″), localization of mut PTB LE RNA is significantly reduced (Figures 1D′ and 1D″). The mut PTB LE RNA is significantly less localized in L-bodies, but is still more colocalized than a non-localizing control, XBM RNA, which is neither enriched in nor excluded from L-bodies (Figures 1E, S1D, and S1E). This low, but significant, level of colocalization is likely owing to the binding of mut PTB LE RNA to other L-body RBPs, such as Vera, which bind to sites not affected by the mut PTB LE RNA mutations (Lewis et al., 2004). Taken together, these data demonstrate that PTB binding to the LE RNA is required for robust localization of the RNA into L-bodies.

Figure 1

Localization and dynamics of LE RNA in L-bodies requires PTB binding sites

(A) Schematic of LE RNA (magenta) with four polypyrimidine tracts (PTB Sites 1–4) indicated in green.

(B) Schematic of a stage II Xenopus oocyte with LE RNA (magenta) localization, as shown in whole oocyte images (as in C); the vegetal cortex is at the bottom. The portion of cytoplasm shown in high magnification images (as in D) is denoted by a black box.

(C) Stage II oocytes were microinjected with fluorescently labeled LE RNA (C, magenta) and mut PTB LE RNA, with all PTB binding sites mutated, (C′, green). The overlap is shown in C′′; scale bars = 100 μm.

(D) High magnification view of L-bodies in a stage II oocyte microinjected with LE RNA (D, magenta) and mut PTB LE RNA (D′, green). The overlap is shown in D′′; scale bars = 10 μm.

(E) Normalized Pearson correlation coefficient of Cy5-labeled LE RNA (magenta), mut PTB LE RNA (green), and XBM RNA (blue) with Cy3-labeled LE RNA in stage II oocytes, as in Figures S1A–S1A″, 1C–1C″, and S1C–S1C′′. LE RNA colocalization with LE RNA is set to 1. n = 30 oocytes per RNA and error bars represent SEM. ∗∗∗ indicates p < 0.01.

(F) An image of the vegetal cytoplasm of an oocyte microinjected with Cy5-labeled LE RNA is shown, with a 10 μm2 ROI (white); scale bar = 10 μm. F′ and F″ show the post-bleach and 500-s time points, respectively.

(G) Stage II oocytes were microinjected with Cy5-labeled LE RNA to mark L-bodies, along with either Cy3-labeled LE RNA (magenta), mut PTB LE RNA (green), or XBM RNA (blue). Normalized FRAP recovery curves are shown. n = 21 oocytes and error bars represent SEM. See also Figure S1 and Table S4.

As LE RNA is known to be non-dynamic in L-bodies (Neil et al., 2021), we next tested the dynamics of the mut PTB LE RNA in L-bodies. We microinjected oocytes with Cy3-labeled LE RNA, mut PTB LE RNA, or XBM RNA, along with Cy5-labeled LE RNA to mark endogenous L-bodies, and performed fluorescence recovery after photo bleaching (FRAP) on the Cy3-labeled RNAs in L-bodies (Figures 1F and 1G). As expected, LE RNA is almost entirely non-dynamic within L-bodies, with an immobile fraction of 93.9% (Figure S1F). By contrast, mut PTB LE RNA is significantly more dynamic than the LE RNA, with an immobile fraction of 67.7%, and was indistinguishable from XBM RNA (72.4% immobile fraction). These results indicate that PTB binding sites are required for both localization of LE RNA to L-bodies and the non-dynamic nature of the RNA in L-bodies.

PTBP3 localization and dynamics in L-bodies require RNA binding

As PTB binding sites in LE RNA are necessary for RNA localization, we next tested if RNA binding by PTB is required for protein localization in L-bodies. L-bodies contain two paralogs of PTB, PTBP1 and PTBP3 (Neil et al., 2021). First, we tested the distribution of PTBP1 by expressing mCherry tagged PTBP1 (mCh-PTBP1) in oocytes and assaying the subcellular localization via anti-mCh immunofluorescence (IF) (Figure S2). We found that PTBP1, which shuttles between the nucleus and cytoplasm of Xenopus oocytes (Lewis et al., 2008; Xie et al., 2003), is only slightly colocalized with L-bodies in stage II oocytes (Figure S2A), and only becomes significantly colocalized in L-bodies in later stages of oogenesis (Figures S2B and S2C). Interestingly, unlike its paralog PTBP1, mCh-PTBP3 was not detected in the nucleus and is highly localized to L-bodies, when they are first evident in stage II of oogenesis (Figure 2B). Therefore, we focused on PTBP3 for our experiments.

Figure 2

PTBP3 localization and dynamics in L-bodies require RNA binding

(A) Schematic of PTBP3 (green) with four RRMs (RRM1-4).

(B) Fluorescently labeled LE RNA (B, magenta) was microinjected into stage II oocytes expressing mCh-PTBP3, as detected by anti-mCh IF (B′, green). The overlap is shown in B′′; scale bar = 100 μm.

(C) High magnification view of L-bodies in a stage II oocyte microinjected with LE RNA (C, magenta) and expressing mCh-PTBP3, as detected by anti-mCh IF (C′, green). The overlap is shown in C′′; scale bar = 10 μm.

(D) Schematic of PTBP3 mut1234 (gray) with three point mutations (green X marks) introduced into each of the four RRMs (Table S1).

(E) Lysates prepared from oocytes expressing Vera-mCh, mCh-PTBP3, or mCh-PTBP3 mut1234 were immunoprecipitated using anti-mCh and IgG. Following the isolation of bound RNAs, vg1 mRNA was detected via qRT-PCR, with normalization to a luciferase extraction control. Shown is fold enrichment for vg1 mRNA over the IgG control. n = 3 and error bars represent SEM. ns indicates p > 0.5, ∗ indicates p < 0.5. Statistics shown are an Ordinary one-way ANOVA with Tukey’s multiple comparison correction.

(F) Fluorescently labeled LE RNA (F, magenta) was microinjected into stage II oocytes expressing mCh-PTBP3 mut1234, as detected by anti-mCh IF (F′, green). The overlap is shown in F′′; scale bar = 100 μm.

(G) High magnification view of L-bodies in a stage II oocyte microinjected with LE RNA (G, magenta) and expressing mCh-PTBP3 mut1234, as detected by anti-mCh IF (G′, green). The overlap is shown in G′′; scale bar = 10 μm.

(H) Oocytes expressing mCh-PTBP3 (green) or mCh-PTBP3 mut1234 (gray) were microinjected with LE, mut PTB LE, and XBM RNAs. Oocyte lysates were immunoprecipitated using anti-mCh and IgG. Following the isolation, bound RNAs were detected by qRT-PCR, with normalization to a luciferase RNA extraction control. Shown is the fold enrichment for each anti-mCh IP over the IgG control. n = 3 and error bars represent SEM. ns indicates p > 0.5, ∗∗∗ indicates p < 0.01. Statistics shown are an Ordinary one-way ANOVA with Tukey’s multiple comparison correction.

(I) Stage II oocytes expressing mCh (black), mCh-PTBP3 (green), or mCh-PTBP3 mut1234 (gray) were microinjected with Cy5-labeled LE RNA to mark L-bodies. Normalized FRAP recovery curves are shown. n = 21 oocytes and error bars represent the SEM.

(J) An image of the vegetal cytoplasm of an oocyte expressing mCh-PTBP3 and microinjected with Cy5-labeled LE RNA to mark L-bodies is shown, with a 10 μm2 ROI (white box); scale bar = 10 μm. J′ and J″ show the postbleach and 500-s time points, respectively.

(K) An image of the vegetal cytoplasm of an oocyte expressing mCh-PTBP3mut1234 and microinjected with Cy5-labeled LE RNA to mark L-bodies is shown, with a 10 μm2 ROI (white box); scale bar = 10 μm. J′ and J″ show the postbleach and 500-s time points, respectively. See also Figures S2, S3 and Tables S1, S4.

PTBP3 is an RNA-binding protein containing four well-folded RNA recognition motifs (RRMs; Figure 2A). The PTBP3 RRMs bind to polypyrimidine tracts in RNA, here termed PTB sites (Yamamoto et al., 1999). To determine the distribution of PTBP3 in oocytes, we expressed mCherry-tagged PTBP3 (mCh-PTBP3) in stage II-III oocytes and assayed the subcellular localization using anti-mCh IF. At low magnification, mCh-PTBP3 is highly restricted to the vegetal cytoplasm of oocytes, particularly at the vegetal cortex and in the lower vegetal cytoplasm and is colocalized with LE RNA (Figures 2B–2B″). At higher magnification, mCh-PTBP3 is observed to be highly localized to L-bodies (Figures 2C–2C″). As PTBP3 is highly colocalized within L-bodies in vivo, PTB binding sites are required for LE RNA localization, and it has a clear, multivalent RBD domain structure, we used PTBP3 as a model protein to probe the role of RNA binding in the localization of proteins to L-bodies and their dynamics after localization.

To determine whether RNA-binding is required for the localization of mCh-PTBP3 to L-bodies, point mutations analogous to those that have been demonstrated to disrupt RNA-binding in human PTBP1 (Kafasla et al., 2011) were introduced into Xenopus PTBP3 (Figure 2D and Table S1). Three specific point mutations were engineered into each of the four mCh-PTBP3 RRMs to create a quadruple RRM mutant (mCh-PTBP3 mut1234), which expresses in oocytes containing endogenous wild type PTBP3 at comparable levels to the mCh-PTBP3 (Figure S3). First, to test whether the RRM mutations affected the interaction between PTBP3 and vg1 RNA, we expressed mCh-PTBP3 or mCh-PTBP3 mut1234 in oocytes and performed RNA immunoprecipitation (RIP) experiments using Vera-mCh, a well-established vg1 mRNA binding protein as a positive control (Lewis et al., 2004) (Figure 2E). mCh-PTBP3 immunoprecipitated endogenous vg1 mRNA comparably to Vera-mCh, showing that PTBP3 interacts with vg1 mRNA in oocytes. Conversely, immunoprecipitation of vg1 mRNA was significantly decreased in the mCh-PTBP3 mut1234 injected oocytes, demonstrating that the mutations inserted into PTBP3 significantly reduce binding to endogenous vg1 mRNA in vivo. In addition to the RNA-binding deficiency, we cannot exclude the possibility that the mutations inserted into PTBP3 also disrupt some protein–protein interactions either directly or through RNA-binding-dependent interactions.

We also used RIP experiments to determine whether the mutation of the PTB binding sites in mut PTB LE RNA blocked binding to PTBP3. Oocytes were microinjected with either mCh-PTBP3 or mCh-PTBP3 mut1234, along with LE RNA, mut PTB LE RNA, and XBM RNA. Following immunoprecipitation with anti-mCh and IgG as a negative control, the fold enrichment of each RNA was assayed via RT-qPCR (Figure 2H). As with the endogenous vg1 mRNA, mCh-PTBP3 strongly immunoprecipitated the LE RNA, whereas the mCh-PTBP3 mut1234 did not; further demonstrating that the PTBP3 RRM mutations disrupt RNA binding. Importantly, neither mCh-PTBP3 nor mCh-PTBP3 mut1234 immunoprecipitated the mut PTB LE or XBM RNAs significantly over IgG controls. We conclude from these results that PTBP3 binds to one or more of the PTB binding sites in the LE RNA.

To determine whether the binding of PTBP3 to RNA is required for localization of PTBP3 to L-bodies, we tested the subcellular distribution of mCh-PTBP3 mut1234 in stage II-III oocytes containing wild-type endogenous PTBP3 and thus forming L-bodies. In contrast to wild-type mCh-PTBP3 (Figures 2B–2B″), mCh-PTBP3 mut1234 is distributed throughout the oocyte cytoplasm, and is not colocalized with LE RNA in L-bodies (Figures 2F–2F″). At high magnification, mCh-PTBP3 mut1234 is neither enriched nor excluded from L-bodies, but is nearly ubiquitous in the vegetal cytoplasm (Figures 2G–2G″), in marked contrast to wild-type PTBP3 (Figures 2C–2C″). These results indicate that the mutations inserted into PTBP3 mut1234 that block RNA binding also disrupt localization to L-bodies.

Our previous work suggested a model for an L-body structure in which localizing RNAs form a non-dynamic component, enmeshed by proteins exhibiting a range of moderate to high mobilities (Neil et al., 2021). Here, we hypothesize that the dynamics of L-body proteins are regulated by direct binding to RNA. Thus, PTBP3, which binds vg1 mRNA (Figure 2E), should be moderately dynamic within L-bodies, whereas the quadruple RRM mutant should be significantly more dynamic. To test this hypothesis, we expressed mCh, mCh-PTBP3, or mCh-PTBP3 mut1234 in stage II oocytes and analyzed the protein dynamics in vivo using FRAP (Figures 2I–2K). As predicted, mCh-PTBP3 is moderately dynamic, with an immobile fraction of 52.4%, whereas mCh-PTBP3 mut1234 was significantly more dynamic, with an immobile fraction of 6.3%, and more closely resembled the mCh alone control. Taken together, these data indicate that the binding of PTBP3 to RNA is necessary for its localization to L-bodies, and that binding to non-dynamic L-body RNA(s) regulates the dynamics of the protein within L-bodies.

Recombinant PTBP3 and LE RNA phase transition into solid or gel-like condensates in vitro

To probe the potential role of interactions between PTBP3 and RNA in phase transitions, we developed an in vitro assay using recombinant PTBP3, purified under high salt conditions to minimize co-purification of contaminating nucleic acids (Figures S4A and S4B) and in vitro transcribed LE RNA. Purified PTBP3 was mixed with LE RNA and incubated at room temperature for 1 h without a crowding agent. To quantify the degree of phase transitions across different conditions, turbidity was assayed (Sanulli and Narlikar, 2021). First, to assess the role of electrostatic interactions on phase transitions, the turbidity of PTBP3 and LE RNA or PTBP3 alone was tested across increasing salt concentrations (Figure 3A). Across all NaCl concentrations tested, no significant turbidity was detected with PTBP3 alone, indicating that RNA is required for phase transition in vitro. However, with PTBP3 and LE RNA, turbidity was detected in a salt-dependent manner; maximum phase transition was observed at intermediate levels of NaCl, but was inhibited at both low and high concentrations. Next, to assess the role of RNA on phase transitions, the turbidity of PTBP3 and LE RNA or LE RNA alone was tested across an RNA concentration series (Figure 3B). No turbidity was observed with LE RNA alone even at the highest concentrations tested, indicating that the observed phase transition requires PTBP3 and is not due solely to RNA–RNA interactions. With both PTBP3 and LE RNA, phase transition was detected in an RNA-dependent manner, indicating that RNA is required for phase transitions. Importantly, phase transition was inhibited by high concentrations of LE RNA, indicating that multivalent interactions between PTBP3 and LE RNA may be important for in vitro condensate formation.

Figure 3

Recombinant PTBP3 and LE RNA phase transition into solid or gel-like condensates in vitro

(A) PTBP3 and LE RNA together (green) or PTBP3 alone (black) were incubated in the indicated concentrations of NaCl for 1 h at room temperature. Phase transitions were monitored by turbidity, measured by OD600. Error bars represent the SEM n = 3. ns indicates p > 0.5, ∗ indicates p < 0.5, ∗∗ indicates p < 0.1.

(B) PTBP3 and LE RNA together (green) or LE RNA alone (magenta) were incubated in the indicated concentrations of LE RNA for 1 h at room temperature. Phase transitions were monitored by turbidity, measured by OD600. Error bars represent the SEM n = 3. ns indicates p > 0.5, ∗ indicates p < 0.5. Statistics shown are multiple paired t tests using a two-stage step-up method.

(C) 12.5-μM AF647-labeled PTBP3 was incubated with 15 ng/μL AF488-labeled LE RNA for 1 h at room temperature. The resulting condensates are shown by DIC (C), and fluorescent imaging for LE RNA (C′, magenta) and PTBP3 (C″, green). The overlay is shown in C‴; scale bars = 10 μm.

(D–E) An image of PTBP3-LE condensates is shown with a 2 μm2 FRAP ROI (white).

(D) LE RNA fluorescence pre-bleach, postbleach (D′), and 4 min post-recovery (D″). Scale bars = 2 μm.

(E) AF647-PTBP3 fluorescence pre-bleach, post-bleach (E′), and 4 min post-recovery (E″).

(F) Normalized FRAP recovery curves are shown for PTBP3-LE RNA condensates, carried out as in D–E. n = 21 condensates per RNA and error bars represent SEM. See also Figure S4 and Table S4.

To further analyze in vitro phase transition of PTBP3 and LE RNA, condensates formed at intermediate salt and RNA concentrations were visualized by microscopy. When incubated together, PTBP3 and LE RNA formed non-spherical structures that were highly enriched for both the protein and RNA components (Figures 3C–3C‴). Whereas some condensates appeared round, as would be expected for liquid condensates owing to surface tension (Hyman et al., 2014), other condensates appeared to be comprised of multiple small droplets that had interacted to form larger structures, but not fused and reformed into a larger sphere. Under these conditions, neither PTBP3 nor LE RNA formed condensates alone (Figures S4D and S4E), supporting a critical role for protein-RNA interactions in the phase transition. To characterize the dynamics of the protein and RNA in the in vitro condensates, we performed FRAP. LE RNA (Figures 3D–3D″) was non-dynamic in vitro (immobile fraction = 99.8%), indicating that in in vitro condensates, LE RNA forms a solid or gel-like component (Figures 3F and S4C). PTBP3 (Figures 3E–3E″), whereas significantly more dynamic than LE RNA, was only moderately dynamic within the in vitro condensates (immobile fraction = 82.0%) (Figures 3F and S4C). These results demonstrate that interactions between LE RNA and PTBP3 are sufficient to drive phase transition into non-spherical condensates in vitro, with trends in RNA and protein dynamics mirroring the findings in in vivo L-bodies: both in vivo and in vitro, LE RNA is non-dynamic and PTBP3 is significantly more dynamic than the RNA.

Maintenance of PTBP3 and LE RNA condensates requires RNA, but not the protein component

Whereas PTBP3-LE RNA in vitro condensate formation requires both protein and RNA, the components required for the maintenance of the condensates were not known. To test this, we formed in vitro condensates under the same conditions as above, followed by treatment with DNase, RNase, or Proteinase K (Figure 4). With DNase treatment, no significant change was observed in the fluorescence of PTBP3 or LE RNA (Figures 4A and 4B, Video S1). With RNase treatment, however, both the LE RNA and PTBP3 fluorescence decreased rapidly, suggesting that the RNA, facilitating either RNA–RNA or RNA–protein interactions, is required for the maintenance of the condensates (Figures 4C and 4D, Video S2). Conversely, with Proteinase K treatment, whereas the PTBP3 fluorescence decreased rapidly, the LE RNA fluorescence was substantially resistant to degradation of the protein component and remained in the condensates for several minutes after protein degradation was complete (Figures 4E and 4F, Video S3). Importantly, these results were not owing to incomplete proteolysis as no large molecular weight products were observed after Proteinase K treatment (Figure S5). Taken together, these results suggest that, whereas RNA–protein interactions are required for in vitro condensate formation, RNA–RNA interactions are sufficient to maintain the RNA within the condensates following degradation of the protein.

Figure 4

Maintenance of PTBP3 and LE RNA condensates requires RNA, but not protein

(A) 12.5-μM AF647-labeled PTBP3 was incubated with 15 ng/μL AF488-labeled LE RNA for 1 h at room temperature and then treated with 0.18 U/μL DNase. Shown is representative fluorescent imaging for LE RNA (A, magenta) and PTBP3 (A′, green) immediately after enzyme addition (t = 0), after 50, 150, and 250 s; scale bars = 10 μm. See Video S1.

(B) Fluorescence over time relative to t = 0 s is shown for LE RNA (magenta) and PTBP3 (green) following DNase treatment. Fluorescence over time was analyzed for 50 × 50 ROIs (n = 10) from three replicates; error bars show the SEM.

(C) 12.5-μM AF647-labeled PTBP3 was incubated with 15 ng/μL AF488-labeled LE RNA for 1 h at room temperature and then treated with 0.05 U/μL RNase A and 1.82 U/μL RNase T1. Shown is representative fluorescent imaging for LE RNA (C, magenta) and PTBP3 (C′, green) immediately after enzyme addition (t = 0), after 50, 150, and 250 s; scale bars = 10 μm. See Video S2.

(D) Fluorescence over time relative to t = 0 s is shown for LE RNA (magenta) and PTBP3 (green) following RNase treatment. Fluorescence over time was analyzed for 50 × 50 ROIs (n = 10) from three replicates; error bars show the SEM.

(E) 12.5-μM AF647-labeled PTBP3 was incubated with 15 ng/μL AF488-labeled LE RNA for 1 h at room temperature and then treated with 1.14 mg/mL Proteinase K. Shown is representative fluorescent imaging for LE RNA (E, magenta) and PTBP3 (E′, green) immediately after enzyme addition (t = 0), after 50, 150, and 250 s; scale bars = 10 μm. See Video S3.

(F) Fluorescence over time relative to t = 0 s is shown for LE RNA (magenta) and PTBP3 (green) following Proteinase K treatment. Fluorescence over time was analyzed for 50 × 50 ROIs (n = 10) from three replicates; error bars show SEM. See also Figure S5.

Video S1. DNase treatment of PTBP3-LE RNA condensates, related to Fig. 4

Condensates were formed by incubation of 12.5-μM AF647-labeled PTBP3 with 15 ng/μL AF488-labeled LE RNA for 1 h at room temperature, followed by DNase treatment (final concentration of 0.18 U/μL DNase). Condensates were imaged on an Olympus FV3000 confocal microscope, using a 60× UPlan Super Apochromat objective (silicon oil, NA = 1.3) and 5× digital zoom, with frames collected every 5 s. Shown is a video of LE RNA (magenta) and PTBP3 (green), beginning immediately after enzyme addition; the display rate is 5 frames per sec.

Video S2. RNase treatment of PTBP3-LE RNA condensates, related to Fig. 4

Condensates were formed by incubation of 12.5-μM AF647-labeled PTBP3 with 15 ng/μL AF488-labeled LE RNA for 1 h at room temperature, followed by RNase treatment (final concentration of 0.05 U/μL RNase A and 1.82 U/μL RNase T1). Condensates were imaged on an Olympus FV3000 confocal microscope, using a 60× UPlan Super Apochromat objective (silicon oil, NA = 1.3) and 5× digital zoom, with frames collected every 5 s. Shown is a video of LE RNA (magenta) and PTBP3 (green), beginning immediately after enzyme addition; the display rate is 5 frames per sec.

Video S3. Proteinase K treatment of PTBP3-LE RNA condensates, related to Fig. 4

Condensates were formed by incubation of 12.5-μM AF647-labeled PTBP3 with 15 ng/μL AF488-labeled LE RNA for 1 h at room temperature, followed by Proteinase K treatment (final concentration of 1.14 mg/mL). Condensates were imaged on an Olympus FV3000 confocal microscope, using a 60× UPlan Super Apochromat objective (silicon oil, NA = 1.3) and 5× digital zoom, with frames collected every 5 s. Shown is a video of LE RNA (magenta) and PTBP3 (green), beginning immediately after enzyme addition; the display rate is 5 frames per sec.

In vitro condensate morphology and PTBP3 dynamics are dependent on RNA-binding

To test whether the phase transition of PTBP3 in vitro is dependent on specific RNA sequences, we compared the phase transition of PTBP3 in the presence of mut PTB LE RNA, XBM RNA, and LE RNA, as above. For each of the RNAs, PTBP3 formed condensates that incorporated both PTBP3 and the RNA (Figures 5A–5C), demonstrating that although in vitro condensate formation requires RNA, it does not require the interaction of PTBP3 specifically with PTB sites that are present only in LE RNA. For in vitro condensates, unlike in vivo L-bodies, there was no difference in partitioning of LE RNA, mut PTB LE RNA, or XBM RNA into PTBP3 condensates (Figure S6B). However, the morphologies of the in vitro condensates formed by PTBP3 with LE RNA, mut PTB LE RNA, and XBM RNA are distinct from one another (Figures S6C–S6E). PTBP3 and LE RNA (Figures 5A, S6C–S6E) formed fewer, larger, and more irregularly shaped structures, whereas condensates formed with PTBP3 and XBM RNA were smaller, more numerous, and significantly more circular (Figures 5C, S6C–S6E). Condensates formed in vitro with PTBP3 and mut PTB LE RNA displayed an intermediate morphology, but more closely resembled XBM RNA condensates (Figures 5B, S6C–S6E). Importantly, none of these differences were owing to differences in RNA size as all in vitro transcribed RNAs were precisely length matched.

Figure 5

In vitro condensate morphology and PTBP3 dynamics are dependent on RNA-binding

(A) 12.5-μM AF647-labeled PTBP3 was incubated with 15 ng/μL AF488-labeled LE RNA for 1 h at room temperature. The resulting condensates are shown in DIC (A), and by fluorescence imaging for LE RNA (A′, magenta), and PTBP3 (A″, green). The overlay is shown in A‴; scale bars = 10 μm.

(B) 12.5-μM AF647-labeled PTBP3 was incubated with 15 ng/μL AF488-labeled mut PTB LE RNA for 1 h at room temperature. The resulting condensates are shown in DIC (B), and by fluorescence imaging for mut PTB LE RNA (B′, magenta), and PTBP3 (B″, green). The overlay is shown in B‴; scale bars = 10 μm.

(C) 12.5-μM AF647-labeled PTBP3 was incubated with 15 ng/μL AF488-labeled XBM RNA for 1 h at room temperature. The resulting condensates are shown in DIC (C), and by fluorescence imaging for XBM RNA (C′, magenta), and PTBP3 (C″, green). The overlay is shown in C‴; scale bar = 10 μm.

(D) % immobile fraction for RNA FRAP of in vitro condensates. Error bars represent the SEM. ns indicates p > 0.5.

(E) Normalized FRAP recovery curves for are shown for PTBP3 in condensates containing PTBP3-LE RNA (magenta), PTBP3-mut PTB LE RNA (green), and PTBP3-XBM RNA (blue). n = 21 condensates per RNA and error bars represent SEM. See also Figures S6, S7, S8 and Table S4.

The differences in in vitro condensate morphology led us to hypothesize that these different types of condensates may have varying protein and/or RNA dynamics. To test this, we performed FRAP on the protein and RNA in each type of in vitro condensate. Surprisingly, each of the RNAs was non-dynamic, exhibiting mobilities that are not significantly different from one another (Figure 5D), with immobile fractions for whole condensate FRAP of 99.8%, 99.7%, and 99.5% for the LE, mut PTB LE, and XBM RNA, respectively. Similarly, all RNAs were also non-dynamic via partial condensate FRAP (Figure S7). These data suggest that in vitro, as in L-bodies in vivo, RNA is non-dynamic. However, unlike in vivo, RNA is non-dynamic in vitro regardless of the primary sequence of the RNA. Conversely, dynamics of PTBP3 protein were dependent on the sequence of the RNA incorporated into the in vitro condensates; PTBP3 was the least mobile in condensates formed with LE RNA (immobile fraction = 82.0%), and significantly more mobile in condensates formed with mut PTB LE (immobile fraction = 71.5%) or XBM RNA (immobile fraction = 64.8%) (Figures 5E and S6A). These data suggest that PTBP3 dynamics in vitro are dependent on the ability to interact with the non-dynamic RNA, consistent with our in vivo results. When treated with DNase, RNase, or protease, the PTBP3-mut PTB LE and PTBP3-XBM in vitro condensates responded similarly to the PTBP3-LE condensates: no change was observed with DNase treatment, both protein and RNA were concurrently released from the condensates with RNase treatment, and RNA was partially resistant to loss of the protein with protease treatment (Figures S8 and S9).

PTBP3 dynamics in L-bodies are dependent on multivalent interactions with RNA

Because PTBP3 dynamics depend on interaction with RNA both in vivo and in vitro, we next asked whether this relies on specific PTBP3 RRMs or combinations of PTBP3 RRMs. To test the role of multivalency, we took an in vivo approach as recombinant PTBP3 RRM mutants exhibited poor solubility in vitro, particularly after cleavage of the maltose binding protein (MBP) tag. However, in vivo, the expression and solubility of the mutant PTBP3 proteins in oocytes were comparable to that of WT PTBP3 (Figure S3). First, we engineered PTBP3 double RRM mutants (mut12 and mut34), expressed them in stage II oocytes, and tested the double RRM mutants for their ability to bind LE RNA in vivo by RIP (Figure 6A). mCh-PTBP3 mut12 immunoprecipitated LE RNA comparably to the wild-type protein, suggesting that RRMs one and two are not required for binding LE RNA in vivo. By contrast, mCh-PTBP3 mut34 immunoprecipitated LE RNA comparably to mCh-PTBP3 mut1234, suggesting that RRM3, RRM4, or both RRMs are required for binding to LE RNA. To distinguish between these possibilities, we created single RRM mutants (mut3 and mut4) and tested their ability to immunoprecipitate LE RNA. Both mCh-PTBP3 mut3 and mCh-PTBP3 mut4 immunoprecipitated LE RNA comparably to the wild-type and mCh-PTBP3 mut12 proteins, demonstrating that LE RNA can bind to RRM3 and RRM4. Hence, intact forms of RRM3 or RRM4 are both necessary and sufficient for robust LE RNA binding in vivo.

Figure 6

PTBP3 dynamics in L-bodies are dependent on multivalent interactions with RNA

(A) Oocytes expressing mCh-PTBP3 (green), mCh-PTBP3 mut12 (blue), mCh-PTBP3 mut34 (yellow), mCh-PTBP3 mut3 (magenta), mCh-PTBP3 mut4 (purple), and mCh-PTBP3 mut1234 (gray) were microinjected with LE RNA. Oocyte lysates were immunoprecipitated using anti-mCh and IgG. Following the isolation of bound RNAs, LE RNA was detected via qRT-PCR, with normalization to a luciferase RNA extraction control. Fold enrichment for mCh-PTBP3 WT over the IgG control is set to 1. n = 3 and error bars represent the SEM. ns indicates p > 0.5, ∗∗∗∗ indicates p < 0.001. Statistics shown are an Ordinary one-way ANOVA with Tukey’s multiple comparison correction.

(B) High magnification view of L-bodies in a stage II oocyte microinjected with LE RNA (B, magenta) and expressing mCh-PTBP3 mut12, as detected by anti-mCh IF (B′, green). The overlap is shown in B′′; scale bars = 10 μm.

(C) High magnification view of L-bodies in a stage II oocyte microinjected with LE RNA (C, magenta) and expressing mCh-PTBP3 mut34, as detected by anti-mCh IF (C′, green). The overlap is shown in C′′; scale bars = 10 μm.

(D) High magnification view of L-bodies in a stage II oocyte microinjected with LE RNA (D, magenta) and expressing mCh-PTBP3 mut3, as detected by anti-mCh IF (D′, green). The overlap is shown in D′′; scale bars = 10 μm.

(E) High magnification view of L-bodies in a stage II oocyte microinjected with LE RNA (E, magenta) and expressing mCh-PTBP3 mut4, as detected by anti-mCh IF (E′, green). The overlap is shown in E′′; scale bars = 10 μm.

(F) Normalized Pearson correlation coefficient of mCh-PTBP3 WT (green), mCh-PTBP3 mut12 (blue), mCh-PTBP3 mut34 (yellow), mCh-PTBP3 mut3 (magenta), mCh-PTBP3 mut4 (purple), and mCh-PTBP3 mut1234 (gray) with LE RNA in stage II oocytes. mCh-PTBP3 WT colocalization with LE RNA is set to 1. n = 30 oocytes per protein and error bars represent SEM. ns indicates p > 0.5, ∗∗∗∗ indicates p < 0.001.

(G) Stage II oocytes expressing mCh-PTBP3 WT (green), mCh-PTBP3 mut12 (blue), mCh-PTBP3 mut34 (yellow), mCh-PTBP3 mut3 (magenta), mCh-PTBP3 mut4 (purple), and mCh-PTBP3 mut1234 (gray) were microinjected with Cy5-labeled LE RNA to mark L-bodies. Normalized FRAP recovery curves are shown. n = 21 oocytes per protein and error bars represent the SEM.

(H) % immobile fraction for protein FRAP experiments shown in (G). Error bars represent the SEM. ns indicates p > 0.5, ∗∗∗∗ indicates p < 0.001. See also Figure S3 and Tables S1, S4.

To test whether specific PTBP3 RRMs drive localization to L-bodies, we expressed each of the double and single PTBP3 RRM mutants in vivo and assayed their subcellular distribution using mCh IF. We found that LE RNA binding by PTBP3 RRM mutants predicts localization to L-bodies: mCh-PTBP3 mut12 is colocalized with L-bodies (Figures 6B–6B″, 6F), whereas mCh-PTBP3 mut34 showed no colocalization (Figures 6C–6C″, 6F); mCh-PTBP3 mut3 and mCh-PTBP3 mut4 single RRM mutants strongly colocalize with L-bodies (Figures 6D–6D″, 6E–6E″, 6F). These results show that binding to RNA by either RRM3, RRM4, or RRM3 and 4 is sufficient to drive the localization of PTBP3 in L-bodies and that this localization does not require multivalent interactions between protein and RNA.

Whereas RNA binding by a single RRM is sufficient to drive protein localization, we reasoned that interaction between the non-dynamic RNA and multiple RRMs may be required for the moderate dynamics of the wild-type PTBP3 protein in L-bodies. To test the role of multivalent interactions in protein dynamics, we tested each of the PTBP3 mutants by FRAP (Figures 6G and 6H). The dynamics of mCh-PTBP mut12 (immobile fraction = 50.1%), was indistinguishable from the wild-type protein, and mCh-PTBP3 mut34 (immobile fraction = 10.2%) was indistinguishable from mCh-PTBP3 mut1234, as in the RIP and colocalization experiments. However, as predicted by our model, mCh-PTBP3 mut3 and mCh-PTBP3 mut four showed intermediate dynamics with immobile fractions of 23.9 and 26.7%, respectively, and were indistinguishable from one another. These results demonstrate that RNA–PTBP3 binding to a single RRM is sufficient for localization to L-bodies, whereas multivalent interactions with RNA act in combination to regulate the dynamics of the protein within L-bodies.

Discussion

Proposed model for L-body component recruitment and dynamics

In this work, we have dissected the role of RBP–RNA binding in L-bodies using both in vivo and in vitro techniques. L-bodies are recently identified, irregularly shaped biomolecular condensates with a non-dynamic RNA component and a comparatively dynamic protein component (Neil et al., 2021). However, the mechanisms underlying the range of biophysical states observed in L-body components were unclear. Here, we propose a multistep model for L-body component recruitment and dynamics based on both specific RNA–RBP binding and RNA concentration-dependent effects using the protein PTBP3 as a model (Figure 7). Whereas PTBP3 binding is a central player in LE RNA localization, many other L-body RBPs may follow similar patterns of RNA-binding dependent dynamics within the oocyte.

Figure 7

Model for recruitment and dynamics of PTBP3 and localized RNAs in L-bodies

RBPs, such as PTBP3 (green), bind localizing RNAs (magenta) in a sequence-dependent manner that is required for both RNA and protein localization to L-bodies (tan), such that PTBP3 RNA-binding mutants (gray) are neither enriched nor excluded from L-bodies. Once localized, locally high concentrations cause RNAs to become solid or gel-like in a sequence-independent manner within the L-body. Arrows denote movement of proteins or RNAs into and out of the L-body, with the arrow thickness indicating the relative amount of this movement. RNA-protein interaction with one RRM is sufficient to target PTBP3 to L-bodies, whereas the strength of number of interactions with non-dynamic RNA tunes the protein dynamics within the L-body such that PTBP3 with a single RNA-binding RRM is more dynamic than PTBP3 with two RNA-binding RRMs.

First, RBPs bind localizing RNAs in the oocyte cytoplasm, driving the localization of the RBP and RNA within L-bodies. Accordingly, PTBP3 strongly colocalizes with L-bodies, but PTBP3 mut1234, which no longer binds to the LE RNA, is nearly ubiquitous throughout the cytoplasm. Similarly, LE RNA localization is a defining feature of L-bodies, but mut PTB LE and XBM RNAs, which do not bind to PTB, are both distributed throughout the cytoplasm. Therefore, sequence-specific RNA–protein binding is required for both the RNA and the RBP to localize to L-bodies in Xenopus oocytes. These results point to PTBP3 as a novel regulator of L-body RNA incorporation, adding mechanistic insight into the compositional control of condensates that are required to pattern the developing embryo. Unfortunately, direct depletion of the endogenous PTBP3 protein to assess its role in LE RNA localization to L-bodies was not possible within this in vivo system. The in vitro RNA–PTBP3 condensates, however, enrich equally for all RNAs as this step of regulation is not present in the minimal in vitro system. Instead, non-specific interactions between PTBP3 and mut PTB LE or XBM RNA are sufficient to drive phase transition in vitro.

Next, following RBP and RNA localization, locally high concentrations of RNA within the L-body facilitate intermolecular RNA–RNA interactions, leading the RNA to exhibit solid or gel-like dynamics. In vivo, LE RNA, which is highly concentrated within the L-body, is almost entirely non-dynamic, whereas mut PTB LE and XBM RNAs are not enriched within L-bodies and are much more dynamic. These dynamics are not a property of RNA size, as LE, mut PTB LE, and XBM RNAs are precisely length-matched, and previous work has demonstrated that RNA mobility in in vivo L-bodies also does not correlate with RNA length (Neil et al., 2021). Whereas the in vivo data do not distinguish between differences in RNA dynamics owing to a lack of PTBP3 binding and RNA concentration-dependent effects, the in vitro data suggests that the process may be RNA concentration dependent; RNA gelation occurs in the in vitro PTBP3–RNA condensates, driving LE, mut PTB LE, and XBM RNAs to all be non-dynamic regardless of PTB binding. In vitro, the non-dynamic RNA is indispensable for the maintenance of the condensate as RNase treatment causes both protein and RNA to dissolve, but RNA–RNA interactions are partially sufficient to maintain the in vitro condensates after the degradation of PTBP3 protein.

Finally, it is both the strength and number of interactions with the solid or gel-like RNA that tune protein dynamics. In oocytes, a single PTBP3 RRM–RNA interaction is sufficient to drive the localization of PTBP3 into L-bodies, but multivalent interactions between multiple RRMs and the RNA work in combination to regulate the dynamics of the protein after localization in vivo. In accordance with this idea, another direct LE RNA binding protein, Vera, was also found to be only moderately dynamic in vivo, with an immobile fraction of 47.1% (Neil et al., 2021). In vitro, it is the strength of the interaction between PTBP3 and the RNA component that determines both condensate morphology and PTBP3 dynamics: PTBP3–LE RNA condensates are non-spherical and have the lowest PTBP3 dynamics, whereas PTBP3–XBM RNA condensates are more spherical and have the highest PTBP3 dynamics. Whereas the precise % immobile fractions differ between the in vitro and in vivo systems, in each environment the LE RNA was significantly less dynamic than PTBP3.

RBP binding to non-dynamic RNA facilitates the formation of non-spherical condensates

In addition to L-bodies, recent studies have identified other non-spherical biomolecular condensates, including TIS granules that form through the phase transition of an RBP, TIS11B, near the ER(Ma and Mayr, 2018, reviwed in Fare et al., 2021). The existence of non-spherical condensates is somewhat counterintuitive as the effects of surface tension drive many liquid-like condensates to be spherical in shape (reviewed in Hyman et al., 2014). However, non-dynamic RNAs may be a conserved mechanism to build non-spherical condensates that still have some liquid-like properties, such as highly dynamic proteins. Such non-spherical condensates are hypothesized to result from multiple small droplets flocculating together (Jawerth et al., 2020; Ranganathan and Shakhnovich, 2020). In this case, all of the available valencies within viscous condensates are hypothesized to be exhausted, preventing the ready coalescence into a larger spherical condensate. In the case of L-bodies, individual PTBP3-LE RNA clusters may be flocculating together, leading to the formation of the irregular shapes observed.

In vitro studies have provided insights into the mechanisms underlying the observed condensate morphologies. Here, we have shown the PTBP3 and LE RNA phase transition in vitro into non-spherical condensates. However, PTBP3 also phase transitions into more spherical condensates in the presence of RNAs without clear PTB binding sites, including the mut PTB LE and XBM RNAs, indicating that specific RNA–RBP interactions may drive the formation of the non-spherical condensates in vitro. In all cases, regardless of the morphology, RNA is non-dynamic in vitro and we hypothesize this to be owing to concentration-dependent rather than sequence-specific effects. Recent in vitro studies using a fusion protein of the RBD of TIS11B and the IDR of FUS (FUS-TIS) and various RNAs also produced condensates with varying morphologies (Ma et al., 2021). However, rather than observing morphological differences based on specific RNA–RBD binding as seen with PTBP3, FUS-TIS condensate morphology varied based on the ability of the RNA to form intermolecular RNA–RNA interactions. Additionally, RNAs in spherical FUS-TIS condensates were found to be dynamic via FRAP, whereas RNAs in non-spherical condensates were non-dynamic. The differences between the in vitro condensates may be owing to variation in the rate of aging of the condensates in vitro or differences in the type of protein tested – PTBP3 is well-folded and is only moderately dynamic via FRAP, whereas FUS-TIS contains a well-established IDR and is highly dynamic via FRAP. However, both studies demonstrate that a stable RNA component formed by certain RNAs can lead to failed condensate fusion events in vitro, driving the formation of a non-spherical condensate.

Although the functions of non-spherical biomolecular condensates remain unclear in vivo, these morphologies increase the surface area to volume ratio, increasing the interaction interface with the cytoplasm. In L-bodies, this may be beneficial for efficiently capturing RNPs that have not yet been incorporated into L-bodies and are diffusing in the cytoplasm. As the later translation of RNAs incorporated into L-bodies is required to pattern the embryo, a high degree of enrichment of localizing RNAs may be required to prevent misexpression of these transcripts. Furthermore, as Xenopus oogenesis occurs over extended timescales, the formation of a more solid- or gel-like condensate may be necessary for the long-term function of the condensate. As a growing number of biomolecular condensates in oocytes and embryos, including P granules in Caenorhabditis elegans (Putnam et al., 2019), germ granules, and oskar RNP granules in Drosophila (Bose et al., 2022; Niepielko et al., 2018; Trcek et al., 2020), and the Balbiani body and L-bodies in Xenopus (Boke et al., 2016; Neil et al., 2021) contain a solid or gel-like component, this may be a conserved feature of condensates in germ cells.

Emergence of biomolecular condensates with non-dynamic RNAs and dynamic proteins

Biomolecular condensates exist on a continuum of biophysical states from a demixed liquid to a solid state, often with different components of the condensate displaying varying dynamics (reviewed in Alberti et al., 2019; Fare et al., 2021). Recent studies have demonstrated a growing variety of biomolecular condensates that have non-dynamic RNAs and comparatively dynamic proteins, including Drosophila germ granules, paraspeckles, and Xenopus L-bodies (Mao et al., 2011; Neil et al., 2021; Niepielko et al., 2018; Trcek et al., 2020). As each of these types of biomolecular condensates is enriched for RNAs, intermolecular RNA–RNA interactions may be particularly thermodynamically favorable owing to the high local concentration of RNA (reviewed in Van Treeck and Parker, 2018), driving the formation of the non-dynamic RNA. Both in vivo and in vitro, PTBP3 drives the localization of RNA into condensates, and therefore is necessary to achieve the locally high concentrations of RNA that can facilitate RNA–RNA interactions. The growing number of condensates with stable RNA components suggests that intermolecular RNA–RNA interactions driving the formation of a stable RNA substructure may be a common feature of many biomolecular condensates, giving broad relevance to the insights into the role of RNA binding in regulating protein dynamics in L-bodies.

In L-bodies, the strength and number of the interactions with the stable RNA determine PTBP3 protein dynamics in vivo. However, the role of other proteins, particularly proteins containing IDRs, in maintaining the biophysical state of the L-body is not yet understood. Whereas in vitro PTBP3–RNA condensates and in vivo L-bodies contain non-dynamic RNA and moderately dynamic PTBP3, both LE RNA and PTBP3 were less dynamic in vitro than in vivo, suggesting that other factors contribute to the maintenance of the biophysical state of L-bodies in vivo. These differences highlight the limitations of minimal in vitro systems in studying the complex biology of phase separation, reinforcing the need to pair the mechanistic insights from in vitro studies with in vivo characterization. As L-bodies, like many other biomolecular condensates, are highly enriched for proteins with multiple RNA binding domains and IDRs, the relative contribution of each of these types of interactions to the dynamics of the condensate is an important outstanding question. For example, in FXR1 assemblies, RNA binding drives phase separation and IDRs of various lengths tune the dynamics of the condensate (Smith et al., 2020). Similarly, in L-bodies, IDR-containing proteins, which are more dynamic than PTBP3 (Neil et al., 2021), may function to keep the condensate in a more liquid-like state, perhaps preventing an irreversible transition to a solid-like state. Additionally, L-bodies contain helicases and post-translational modifying enzymes (Neil et al., 2021), which may function to remodel interactions in vivo to facilitate assembly and/or disassembly.

Limitations of the study

As additional biomolecular condensates continue to be identified across an ever-growing diversity of cells and subcellular locations, it remains an important challenge to characterize the principles that are conserved or divergent across many classes of condensates. One hallmark of biomolecular condensates is the enrichment of RNA and multivalent RNA binding proteins, highlighting the importance of understanding how RNAs, proteins, and RNA–protein interactions contribute to both the composition and characteristics of condensates. Our results, which indicate that protein dynamics are tuned by multivalent interactions with a non-dynamic RNA, may provide a paradigm that is applicable to other classes of biomolecular condensates, with particular importance to solid- or gel-like condensates that regulate post-transcriptional gene expression in oocytes and early embryos. We anticipate that future studies will elucidate the range of principles underlying the functional contributions of such condensates to development and cellular function in health and disease.

STAR★Methods

Key resources table

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Kimberly Mowry (kimberly_mowry@brown.edu).

Materials availability

This study did not generate new unique reagents. Plasmids generated in this study are available from the lead contact upon completion of an MTA.

Experimental model and subject details

Oocytes used in this study were surgically harvested from wild-type Xenopus laevis females (Nasco). All animal experiments were approved by the Brown University Institutional Animal Care and Use Committee.

Method details

Oocyte isolation and culture

Oocytes were defolliculated using 2 mg/mL collagenase (Sigma), and washed in MBSH [88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO3, 0.82 mM MgSO4, 0.33 mM Ca(NO3)2, 0.41 mM CaCl2, 100mM HEPES (pH 7.6)]. Stage II-III oocytes were manually sorted, and cultured at 18°C in OCM+ [50% Leibovitz L-15 medium, 15 mM HEPES (pH 7.6), 1 mg/mL insulin, 50 U/mL nystatin (Sigma), 100 U/mL penicillin/streptomycin (ThermoFisher), 0.1 mg/mL gentamicin (ThermoFisher), 0.1 mg/mL ciprofloxacin] (Sigma).

Cloning and mutagenesis

RNA from stage II-III oocytes was isolated by Trizol extraction and reverse transcribed into cDNA using iScript cDNA synthesis kit. PTBP3 specific primers (Table S2) were used to amplify PTBP3 cDNA using Phusion high fidelity master mix and cloned in the pSP64TSNRLMCS:mCherry (Neil et al., 2021) and pET:THMT (Peti and Page, 2007) vectors. A gBlock gene fragment for the PTBP3 quadruple mutant (IDT) was cloned into pSP64TSNRLMCS:mCherry vector using Gibson Assembly master mix (New England Biolabs). Single and double mutants were created by ligating portions of the WT and quadruple mutant gene together using Gibson Assembly Master Mix and cloning into the pSP64TSNRLMCS:mCherry vector.

RNA transcription

For protein coding RNAs, RNAs were transcribed in vitro with the mMessage machine SP6 kit (ThermoFisher) using the following linearized plasmids as the DNA template: pSP64TSNRLMCS:mCh-PTBP3, pSP64TSNRLMCS:mCh-PTBP3 mut12, pSP64TSNRLMCS:mCh-PTBP3 mut34, pSP64TSNRLMCS:mCh-PTBP3 mut3, pSP64TSNRLMCS:mCh-PTBP3 mut4, pSP64TSNRLMCS:mCh-PTBP3 mut1234, pSP64TSNRLMCS:mCh, pSP64:mCh-PTBP1, and pSP64TSNRLMCS:Vera-mCh (Neil et al., 2021). RNAs were extracted with phenol/chloform/isoamyl alcohol (25:24:1; ThermoFisher) and precipitated with 1 volume of isopropanol as in (Jeschonek and Mowry, 2018). The concentration of the RNA was measured via Qubit RNA broad range assay (ThermoFisher).

For the barcoded RNAs, RNAs were transcribed in vitro with the MEGAscript T7 kit (ThermoFisher) using length matched PCR products as the DNA templates. PCR products were generated from the following plasmids, with the forward primer for each construct containing the T7 promoter sequence and the unique barcode sequence as shown in Table S2: pSP73:2x135 (LE) (Gautreau et al., 1997), pSP73:2 × 135ΔVM1 (mut PTB LE) (Lewis et al., 2004), pSP64:XBM (XBM) (Krieg and Melton, 1984). To synthesize fluorescently labelled RNAs for oocyte microinjection, RNAs were transcribed as above in the presence of 250 nM Cy3- or Cy5-UTP (ThermoFisher). For fluorescently labelled RNAs for in vitro experiments, RNAs were transcribed as above in the presence of 500 nM Alexa Fluor 488-UTP (ThermoFisher). T7 transcribed RNAs were cleaned up using MEGAclear transcription clean up kits (ThermoFisher) and the concentration of RNA was measured via Qubit RNA broad range assay (ThermoFisher).

Whole mount immunofluorescence (IF)

Oocytes were microinjected with 2 nL of 500 nM RNA encoding an mCh-PTBP3 construct and 250 nM Cy5-UTP labelled LE RNA to mark L-bodies. Oocytes were cultured for 48 h in OCM+, fixed for 1 h in fixation buffer [80 mM PIPES (pH 6.8), 1 mM MgCl2, 5 mM EGTA, 0.2% Triton X-100, 3.8% formaldehyde], and washed 3 times for 15 min. each in PBT [137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4, 0.2% BSA, 0.1% Triton X-100]. Oocytes were blocked for 4 h at room temperature in PBT+ [PBT supplemented with 2% goat serum and 2% BSA], incubated overnight at 4°C in a 1:500 dilution of anti-mCh primary antibody (Abcam) in PBT+, and washed 3 times for 2 h each in PBT. Oocytes were incubated overnight at 4°C in a 1:1000 dilution of goat, anti-rabbit AF647 conjugated secondary antibody (ThermoFisher), washed 3 times for 2 h each, dehydrated in anhydrous methanol, and frozen at −20°C until imaging. Immediately prior to imaging, oocytes were cleared in BABB solution (1:2 benzyl alcohol:benzyl benzoate). Oocytes were imaged on an inverted Olympus FV3000 confocal microscope using 20× UPlan Super Apochromat objective (air, NA = 0.75) and 60× UPlan Super Apochromat objective (silicon oil, NA = 1.3) using GaAsP detectors.

Analysis of RNA stability in vivo

Stage II oocytes were microinjected with 250 nM barcoded LE RNA and LE RNA levels were measured at 0-, 24-, and 48-h post-injection by RT-qPCR. Analysis was completed using the delta delta CT method (Livak and Schmittgen, 2001), normalizing to 0 h, and using GAPDH as a control.

RNA immunoprecipitation (RIP)

For the endogenous RIPs, approximately 600 stage II-III oocytes per protein condition were microinjected with 2 nL of 500 nM Vera-mCh, mCh-PTBP3 WT, or mCh-PTBP3 mut1234 RNA. Oocytes were cultured for 48 h at 18°C in OCM+. Oocytes were then crosslinked with 0.1% formaldehyde in PBS for 10 min. at room temperature and then quenched for 5 min. in 250 mM glycine in 25 mM Tris (pH 7.4). Oocytes were then lysed in RIP buffer and clarified twice by centrifugation at 10,000×g for 10 min. at 4°C. 5 mg of antibody [anti- mCh (Abcam) or Normal rabbit IgG (Sigma] was added to each reaction and incubated for 1 h at 4°C with rotation. Next, 15 μL of Pierce Protein A/G magnetic beads in RIP buffer were added to each reaction and incubated for an additional 4 h at 4°C, the beads were washed 3 times in RIP buffer, and bound proteins and RNAs were eluted from the beads in Pierce IgG Elution Buffer (ThermoFisher) via shaking (1200 rpm) at 24°C for 20 min. After removal of the eluent from the beads, crosslinking was reversed by incubation at 70°C for 45 min. To control for RNA extraction efficiency, 2.5 pg of luciferase control RNA (Promega) was added to each sample prior to RNA extraction using RNeasy Plus Micro Kit (Qiagen). cDNA synthesis was performed using iScript cDNA synthesis kit (Biorad), and qRT-PCR was performed for luciferase and vg1 RNAs using PowerUp SybrGreen Master Mix (ThermoFisher) per the manufacturer’s protocol using primers shown in Table S3.

For the barcoded RNA RIPs, 600 stage II-III oocytes per protein expressed were microinjected with 2 nL of 500 nM mCh-PTBP3 WT or mCh-PTBP3 mut1234 RNA and 150 nM Barcode A- vg1 LE, 150 nM Barcode B- mut PTB LE, and 150 nM Barcode C- XBM RNAs. Oocytes were cultured and RIPs were performed as described above without crosslinking and decrosslinking of the samples. qRT-PCR was performed for luciferase, Barcode A, Barcode B, and Barcode C using primers shown in Table S3.

For the combinatorial PTBP3 RRM mutant RIPs, 200 stage II-III oocytes per protein expressed condition were microinjected with 2 nL of 500 nM mCh-PTBP3 WT, mCh-PTBP3 mut12, mCh-PTBP3 mut34, mCh-PTBP3 mut 3, mCh-PTBP3 mut 4, or mCh-PTBP3 mut1234 RNA and 150 nM Barcode A- vg1 LE. Oocytes were cultured and RIPs were completed as described above without crosslinking and decrosslinking of the samples. qRT-PCR was performed for luciferase and Barcode A using primers shown in Table S3.

Recombinant protein expression and purification

MBP-PTBP3 fusion protein was expressed in BL21(DE3) E. coli (New England Biolabs) transformed with pET:THMT:PTBP3 at 15°C overnight with 400 μM IPTG (Sigma). Induced pellets were resuspended in low imidazole purification buffer [20 mM sodium phosphate (pH 7.4), 1 M NaCl, 10 mM imidazole, and 1× Halt protease inhibitors (ThermoFisher)] and lysed via sonication. The suspension was cleared by centrifugation at 28,960 RCF for 30 min. at 4°C, and the resulting supernatant was bound to Ni-NTA agarose (Qiagen) for 1 h at 4°C, washed in 5 bed volumes of low imidazole purification buffer, and eluted in high imidazole purification buffer [20 mM sodium phosphate (pH 7.4), 1 M NaCl, 300 mM imidazole, and 1× Halt protease inhibitor cocktail (ThermoFisher)]. Fractions containing MBP-PTBP3 were pooled, concentrated, buffer exchanged into storage buffer [20 mM sodium phosphate (pH 7.4), 250 mM NaCl], and flash frozen. A260/280 ratios were taken using a Nanodrop One to measure nucleic acid contamination.

In vitro phase transition

Purified MBP-PTBP3 protein was incubated with TEV protease for 2 h at room temperature to cleave off the MBP (maltose binding protein) tag. Protein for fluorescent imaging was labelled with AF647-NHS Ester (ThermoFisher) by resuspending the AF-647 in DMSO and incubating with cleaved PTBP3 protein for 1 h at room temperature. Excess AF-647 was removed in a G50 micro column (GE Healthcare). Labelled protein was buffer exchanged into 2× PTBP3 phase buffer [100 mM Tris (pH 7.5), 200 mM NaCl, 2 mM DTT]. Cleaved PTBP3 protein was diluted to 25 μM in 2× PTBP3 phase buffer with 10% fluorescent labelling for microscopy assays. 30 ng/μL stocks of in vitro transcribed RNA with 25% Alexa Fluor 488-UTP labelled RNA in DEPC-treated H2O for microscopy assays were denatured at 72°C for 10 min. and stored on ice. In a 20 μL reaction, 10 μL of PTBP3 protein (12.5 μM final protein concentration) and 10 μL of RNA (15 ng/μL final RNA concentration) were mixed and incubated for 1 h at room temperature.

Condensate imaging

Condensates were imaged by placing 15 μL of the phase transition reaction onto a #1.5 coverslip with imaging spacers (Grace Bio-labs) and sealed with a slide (total imaging depth of ∼0.24 mM). Condensates were imaged on an inverted Olympus FV3000 confocal microscope using a 60× UPlan Super Apochromat objective (silicon oil, NA = 1.3) and 5× digital zoom.

Condensate enzyme treatments

20 μL of the PTBP3-RNA condensates were prepared as described above and then pipetted onto a #1.5 coverslip. Samples were then treated with 2 μL of TURBO DNase (Invitrogen), 2 μL of RNase Cocktail (Invitrogen), or 2 μL of 12.5 μg/μL Proteinase K (Sigma, 12.5 mg/mL) and live imaged every 5 s for 75 cycles on an inverted Olympus FV3000 confocal microscope using a 60× UPlan Super Apochromat objective (silicon oil, NA = 1.3) and 5× digital zoom.

Turbidity assays

For the NaCl concentration series, in vitro reactions were prepared as described above in 50 mM Tris (pH 7.5), 1 mM DTT, and the indicated NaCl concentration (25, 100, 250, 500 mM, or 1 M). 12.5 μM PTBP3 was incubated with 0 ng/μL RNA (PTBP3 alone) or 15 ng/μL LE RNA (PTBP3 & vg1 LE). For the RNA concentration series, in vitro reactions were prepared as described above in 50 mM Tris (pH 7.5), 100 mM NaCl, 1 mM DTT. 12.5 μM PTBP3 (PTBP3 & LE) or buffer controls (LE alone,) were incubated with the indicated RNA concentration of LE RNA (0 ng/μL, 15 ng/μL, 37.5 ng/μL, and 75 ng/μL). Turbidity was assayed in a 384-well glass bottom plate (MatTek Corporation) with 20 μL samples sealed with clear optical film (Excel Scientific) to prevent evaporation. Absorbance of the samples at 600 nm was read using a Cytation 5 Multi-Mode Reader (BioTek) after 1 h of incubation at room temperature. Absorbance data was normalized to buffer and TEV protease controls. In the turbidity assays, no fluorescent labels were used for either the protein or RNA.

Immunoblotting

25 oocytes per mCh-PTBP3 construct and uninjected controls were homogenized in 50 μL of RIP buffer [25 mM Tris (pH 7.4), 0.5% NP40, 0.5 mM DTT, 150 mM KCl, 5 mM EDTA, 10 mM C4H6MgO4, 1× Halt protease inhibitors (ThermoFisher), and 2 nU/mL Ribolock RNase inhibitor (ThermoFisher)]. Lysates were clarified via centrifugation at 10,000×g for 10 min. at 4°C and then boiled in Laemmli sample buffer (Laemmli, 1970). For immunoblotting, anti-mCh (Abcam) and anti-tubulin (Abcam) primary antibodies were used at 1:1000. Secondary goat anti-rabbit IgG-HRP antibody (Abcam) was used at 1:15,000. Blots were developed using SuperSignal West Pico PLUS chemiluminescent substrate (ThermoFisher) and imaged using a Bio-Rad ChemiDoc.

Fluorescence recovery after photobleaching (FRAP)

For the oocyte FRAP, Stage II oocytes were microinjected with 2 nL of 500 nM RNA encoding mCh-PTBP3 and 250 nM Cy5-UTP labelled LE RNA for FRAP of PTBP3 wild-type and mutant proteins. For RNA FRAP, 2 nL of 250 nM Cy3-UTP labelled test RNA (LE, mut PTB LE, or XBM) was microinjected into stage II oocytes, along with 250 nM Cy5-UTP labelled LE RNA to mark the L-bodies. Microinjected oocytes were cultured for 48 h in OCM+. Seven oocytes per biological replicate (n = 21 oocytes total per construct tested) were analyzed. A 10 μm2 ROI was bleached using the 488 nm laser at 100% for 2 s. Fluorescence recovery was monitored every 5 s for 100 iterations.

For the in vitro condensate FRAP, 7 condensates per replicate (n = 21 condensates total per RNA) were analyzed. For whole condensate FRAP, a 2 μm2 ROI was bleached using the 405 nm laser at 50% and the 561 nm laser at 100% for 0.8 s. For partial condensate FRAP, a 1 μm2 ROI was bleached using the 405 nm laser at 50% and the 561 nm laser at 100% for 0.4 s. Fluorescence recovery was monitored every 5 s for 50 iterations.

Quantification and statistical analysis

Statistical details and analysis

Statistical details of experiments, including exact value of n, what n represents, definition of center and dispersion, and the statistical tests used are detailed in the figure legends. Significance was defined according to the following: ns indicates p > 0.5, ∗ indicates 0.1 < p < 0.5, ∗∗ indicates 0.01 < p < 0.1, ∗∗∗ indicated 0.01 < p < 0.001, and ∗∗∗∗ indicates p < 0.001.

Colocalization analysis

For each of the mCh-PTBP3 wild-type and mutant constructs, images of 10 oocytes in each of 3 biological replicates (n = 30 oocytes) were collected for analysis using a 20× air objective with a 1.2× digital zoom. Working from the top, left corner of the imaging dish, the first 10 oocytes for which LE localization was observed in the perinuclear cup, in L-bodies, and at the vegetal cortex were selected. Colocalization analysis was completed in ImageJ using the Colocalization threshold plugin using an ROI surrounding the oocyte. Statistics shown are an Ordinary one-way ANOVA with Tukey’s multiple comparison correction.

Condensate morphology analysis

Condensate morphology was analyzed on 5 fields of view per replicate (n = 15 fields of view total per RNA). Condensate number, size, and circularity were calculated based on manual thresholding of AF-647 labelled PTBP3 fluorescence using the “Analyze Particles” plugin in ImageJ. Condensates of any circularity and greater than 5 pixels in area were analyzed, excluding all condensates on the edge of the image. Partition coefficients (MCC) were calculated from the same analysis images using the “Colocalization Threshold” plugin in ImageJ. Statistics shown are an Ordinary one-way ANOVA with Tukey’s multiple comparison correction.

Analysis of protein and RNA levels in condensates after enzymatic treatment

To analyze protein and RNA over time after treatment with DNase, RNase, and Proteinase K, 50 × 50 ROIs were placed over stationary condensates, and fluorescence was quantified using the Time Series Analyzer V3 (https://imagej.nih.gov/ij/plugins/time-series.html).

FRAP analysis

FRAP calculations were performed as detailed (Gagnon et al., 2013; Powrie et al., 2016). First, we adjusted for photobleaching by calculating A(t) for the region of interest (ROI) of the time series fluorescence data with the following equation:where F(t) is the background subtracted fluorescence at time t, F is the background subtracted average intensity for the prebleach frames, F(t) is the background subtracted fluorescence intensity value of a neighboring region at the time t, ROI(t) represents the raw fluorescence data from the photobleached ROI at time t, ROI(t) and ROI(t) are fluorescence data from the non-photobleached regions outside (o) and inside (n) the oocyte at time t. Second, the adjusted data A(t) was normalized further against the prebleach values by calculating the normalize adjusted fluorescence time series A(t):

We then further calculated the further normalized time series N(t):where the photobleach corrected fluorescence intensity at the first time point after bleaching is A(1) and the average photobleach corrected fluorescence intensity at the plateau is A. The data was analyzed via one phase non-linear regression and two phase non-linear regression analyses using GraphPad Prism 9. Statistics shown in the main text for the percent immobile fractions are an Ordinary one-way ANOVA with Tukey’s multiple comparison correction of the one phase association data. All percent immobile fraction data is summarized in Table S4.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-alpha tubulin antibody	Abcam	Cat#ab4074, RRID: AB_2288001
Anti-mCherry antibody	Abcam	Cat#ab62341, RRID: AB_945213
Anti-RFP antibody	Abcam	Cat#ab62341, RRID: AB_945213
Goat anti-rabbit IgG-HRP	Abcam	Cat#ab97200, RRID: AB_10679899
Goat anti-rabbit AF647 conjugated secondary	ThermoFisher Scientific	Cat#21244, RRID: AB_2535812
Normal rabbit IgG	Millipore Sigma	Cat#NI01, RRID: AB_490574
Bacterial and virus strains
BL21(DE3) E. coli	New England Biolabs	Cat#C2527H
Chemicals, peptides, and recombinant proteins
Alexa Fluor 488-5-UTP	ThermoFisher Scientific	Cat#C11403
Alexa Fluor 647 NHS ester	Thermo Fisher Scientific	Cat#A37573
Amicon ultra 0.5 centrifugal filter unit	Millipore Sigma	Cat#UFC505024
AnykD criterion TGX stain-free protein gels	Bio-Rad	Cat#5678125
BL21(DE3) competent E. coli	New England Biolabs	Cat#C2527H
Black 384-well glass bottom plate	MatTek Corporation	PBK384G-1.5-C
ChromaTide alexa fluor 488-5-UTP	ThermoFisher Scientific	Cat#C11403
Ciprofloxacin	Millipore Sigma	Cat#17850
Collagenase from Clostridium histolyticum	Millipore Sigma	Cat#C0130
Coomassie brilliant blue R-250 dye	ThermoFisher Scientific	Cat#20278
Cytiva amersham CyDye fluorescent nucleotides, Cy3-UTP	Fisher Scientific	Cat# PA55026
Cytiva amersham CyDye fluorescent nucleotides, Cy5-UTP	Fisher Scientific	Cat#PA55026
Formaldehyde, 37%, microfiltered	Electron Microscopy Science	Cat#15686
Illustra ProbeQuant G-50 micro columns	Millipore Sigma	Cat#28-9034-08
Gentamicin	ThermoFisher Scientific	Cat#15750060
Gibson assembly master mix	New England Biolabs	Cat#E2611L
Halt protease inhibitor cocktail, EDTA-free (100X)	ThermoFisher Scientific	Cat#78439
Invitrogen RNase cocktail enzyme mix	ThermoFisher Scientific	Cat#AM2286
IPTG	Millipore Sigma	Cat#I6758
Leibovitz’s L-15 medium	ThermoFisher Scientific	Cat#11415064
Ni-NTA agarose	Qiagen	Cat#30210
Nystatin	Millipore Sigma	Cat#N1638
Penicillin streptomycin	ThermoFisher Scientific	Cat#15140122
Phusion high-fidelity PCR master mix	New England Biolabs	Cat#M051S
Pierce IgG elution buffer	ThermoFisher Scientific	Cat#21028
Pierce protein A/G magnetic beads	ThermoFisher Scientific	Cat#88803
PowerUp SybrGreen Master Mix	ThermoFisher Scientific	Cat#A25742
Precision plus protein kaliedoscope protein standards	Bio-Rad	Cat#1610375
Proteinase K from Tritachium album	Millipore Sigma	Cat#P2308
Ribolock RNAse inhibitor	ThermoFisher Scientific	Cat#EO0381
SecureSeal imaging spacers	Grace Bio-labs	Cat#470352
ssRNA ladder	New England Biolabs	Cat#N0362S
Super-Signal west pico plus chemiluminescent substrate	ThermoFisher Scientific	Cat#34580
ThermalSeal	Excel Scientific	TS-RT2-100
Turbo DNase	ThermoFisher Scientific	Cat#AM2238
Critical commercial assays
iScript cDNA synthesis kit	Bio-Rad	Cat#1708891
MEGAclear transcription clean-up kit	ThermoFisher Scientific	Cat#AM1908
MEGAscript T7 transcription kit	ThermoFisher Scientific	Cat#AM1334
mMESSAGE mMACHINE SP6 transcription kit	ThermoFisher Scientific	Cat#AM1340
Qubit RNA broad range assay kit	ThermoFisher Scientific	Cat#Q1021
RNeasy Plus Micro Kit	Qiagen	Cat#74034
Experimental models: Organisms/strains
Xenopus laevis, Nasco background, female	Nasco	Cat#LM00535MX
Oligonucleotides
Barcode A vg1 LE forward	Neil et al. (2021)	PCR primer
Barcode B vg1 LE forward	Neil et al. (2021)	PCR primer
Barcode C XBM forward	Neil et al. (2021)	PCR primer
vg1 LE reverse	Neil et al. (2021)	PCR primer
XBM reverse	Neil et al. (2021)	PCR primer
Barcode A qPCR forward	Neil et al. (2021)	qPCR primer
Barcode B qPCR forward	Neil et al. (2021)	qPCR primer
Barcode C qPCR forward	Neil et al. (2021)	qPCR primer
GAPDH qPCR forward	Neil et al. (2021)	qPCR primer
GAPDH qPCR reverse	Neil et al. (2021)	qPCR primer
LE qPCR reverse	Neil et al. (2021)	qPCR primer
luciferase qPCR forward	Neil et al. (2021)	qPCR primer
luciferase qPCR reverse	Neil et al. (2021)	qPCR primer
vg1 qPCR forward	Neil et al. (2021)	qPCR primer
XBM qPCR reverse	Neil et al. (2021)	qPCR primer
Luciferase control RNA	Promega	Cat#L4561
Recombinant DNA
pET:THMT	Peti and Page (2007)	N/A
pET:THMT:PTBP3	This study	N/A
pSP64TSNRLMCS:mCherry	Neil et al. (2021)	N/A
pSP64TSNRLMCS:mCh-PTBP1	This study	N/A
pSP64TSNRLMCS:mCh-PTBP3	This study	N/A
pSP64TSNRLMCS:mCh-PTBP3 mut12	This study	N/A
pSP64TSNRLMCS:mCh-PTBP3 mut1234	This study	N/A
pSP64TSNRLMCS:mCh-PTBP3 mut3	This study	N/A
pSP64TSNRLMCS:mCh-PTBP3 mut34	This study	N/A
pSP64TSNRLMCS:mCh-PTBP3 mut4	This study	N/A
pSP64TSNRLMCS:Vera-mCh	Neil et al. (2021)	N/A
pSP64TSNRLMCS:XBM5	Krieg and Melton (1984)	N/A
pSP73:2X135	Gautreau et al. (1997)	N/A
pSP73:2X135 ΔVM1	Lewis et al. (2004)	N/A
Software and algorithms
ImageJ	NIH	https://imagej.nih.gov/ij/
Analyze particle ImageJ plugin	N/A	https://imagej.nih.gov/ij/docs/menus/analyze.html
Colocalization threshold ImageJ plugin	Costes et al., 2004	https://imagej.net/plugins/colocalization-threshold
Time series analyzer V3 ImageJ plugin	N/A	https://imagej.nih.gov/ij/plugins/time-series.html
Prism 9	GraphPad	https://www.graphpad.com/scientific-software/prism/
Other
Illustra ProbeQuant G50 micro column	GE Healthcare	Cat#28-9034-08
PTBP3 gBlock gene fragment	Integrated DNA Technologies	PTBP3 gBlock