Heterotypic phase separation in aggregation: Driver or deterrent?

Liquid-liquid phase separation (LLPS) in cells is known as a complex physicochemical process causing the formation of membrane-less organelles (MLOs). Cells have well-defined different membrane-surrounded organelles like mitochondria, endoplasmic reticulum, lysosomes, peroxisomes, etc., however, on demand they can create MLOs as stress granules, nucleoli and P bodies to cover vital functions and regulatory activities. However, the mechanism of intracellular molecule assembly into functional compartments within a living cell remains till now not fully understood. in vitro and in vivo investigations unveiled that MLOs emerge after preceding liquid-liquid, liquid-gel, liquid-semi-crystalline, or liquid-crystalline phase separations. Liquid-liquid and liquid-gel MLOs form the majority of cellular phase separation events, while the occurrence of micro-sized crystals in cells was only rarely observed, however can be considered as a result of a preceding protein phase separation event. In vivo, also known and termed as in cellulo crystals, are reported since 1853. In some cases, they have been linked to vital cellular functions, such as storage and detoxification. However, the occurrence of in cellulo crystals is also associated to diseases like cataract, hemoglobin C diseases, etc. Therefore, better knowledge about the involved molecular processes will support drug discovery investigations to cure diseases related to in cellulo crystallization. We summarize physical and chemical determinants known today required for phase separation initiation and formation and in cellulo crystal growth. In recent years it has been demonstrated that LLPS plays a crucial role in cell compartmentalization and formation of MLOs. Here we discuss potential mechanisms and potential crowding agents involved in protein phase separation and in cellulo crystallization.

Article Figures and data Abstract Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract The initiation of DNA replication in metazoans occurs at thousands of chromosomal sites known as origins. At each origin, the Origin Recognition Complex (ORC), Cdc6, and Cdt1 co-assemble to load the Mcm2-7 replicative helicase onto chromatin. Current replication models envisage a linear arrangement of isolated origins functioning autonomously; the extent of inter-origin organization and communication is unknown. Here, we report that the replication initiation machinery of D. melanogaster unexpectedly undergoes liquid-liquid phase separation (LLPS) upon binding DNA in vitro. We find that ORC, Cdc6, and Cdt1 contain intrinsically disordered regions (IDRs) that drive LLPS and constitute a new class of phase separating elements. Initiator IDRs are shown to regulate multiple functions, including chromosome recruitment, initiator-specific co-assembly, and Mcm2-7 loading. These data help explain how CDK activity controls replication initiation and suggest that replication programs are subject to higher-order levels of inter-origin organization. https://doi.org/10.7554/eLife.48562.001 Introduction The appropriate spatiotemporal regulation of DNA replication is essential to genetic integrity and cell proliferation. In eukaryotes, the initiation of DNA replication requires the coordinated action of three proteinaceous factors – the Origin Recognition Complex (ORC), Cdc6, and Cdt1 – which co-assemble on DNA origins in the G1 phase of the cell cycle to catalyze loading of the Mcm2-7 replicative helicase onto chromatin. Once activated during the transition to S phase, Mcm2-7 helps promote origin melting, replisome assembly, and translocation of the replication fork. ORC consists of a heterohexameric complex comprising the subunits Orc1-6. Five ORC subunits (Orc1-5), as well as Cdc6 and the six subunits of Mcm2-7, possess an ATPases Associated with diverse cellular Activities (AAA+) domain; Orc6 and Cdt1 are the only non-AAA+ proteins used for Mcm2-7 loading. Despite sharing a high overall degree of conservation across eukaryotes, certain aspects of initiator subunit sequence and function have nonetheless diversified during evolution. For example, S. pombe Orc4 contains a unique domain not found in other ORCs that endows the protein with a preference for A/T rich regions of DNA (Chuang and Kelly, 1999; Lee et al., 2001). Conversely, S. cerevisiae ORC is able to recognize specific origin sequences (Bell and Stillman, 1992; Li et al., 2018), whereas origin specification for metazoan ORCs appears more contextual (Remus et al., 2004; Vashee et al., 2003). S. cerevisiae Cdt1 additionally possesses a catalytically-inactive dioxygenase domain at its N-terminus that is necessary for yeast viability and Mcm2-7 loading (Frigola et al., 2017; Takara and Bell, 2011); this fold is absent in S. pombe and metazoan Cdt1s. Much of our current understanding of initiator mechanism derives from reconstitution studies using budding yeast initiation factors. These efforts have helped define an orchestrated set of ATP-dependent molecular exchanges between ORC, Cdc6, and Cdt1 that culminate in the loading of two copies of Mcm2-7 onto origin DNA in the form of a stable double-hexamer (Duzdevich et al., 2015; Evrin et al., 2009; Remus et al., 2009; Ticau et al., 2015). One aspect of initiation that has been difficult to probe in vitro, however, is the potential for interactions between initiation factors associated with different origins. Evidence for mesoscale coordination of origin activity derives from multiple sources. For example, genomic studies aimed at defining sites of ORC binding and their relation to replication origins have revealed evidence of origin clustering across chromosomes at nucleosome-free regions (Cayrou et al., 2011; Miotto et al., 2016; Vaughn et al., 1990). Cellular patterns of ORC localization are also strikingly non-uniform, often presenting as concentrated foci on chromatin whether it be in D. melanogaster follicle cells (Austin et al., 1999) or human cells in tissue culture (Lidonnici et al., 2004; Prasanth et al., 2010; Shen et al., 2010). Outside of initiation, there exists a well-characterized phenomenon of origin interference that has been taken as evidence of inter-origin communication (Cayrou et al., 2011); the clustering of co-replicating regions into so-called Replication Factories or replication domains (RD) similarly has been observed in both fixed and live cells (Cook, 1999; Xiang et al., 2018). Finally, the replication timing profile of origins, which fire asynchronously throughout S-phase, partition to specific chromosome territories (Cremer and Cremer, 2001; Gilbert et al., 2005; Pope et al., 2014). It is currently unclear whether the clustering patterns observed for ORC binding sites, origin communication, and the timing of replication domains is coincidence or dependent upon an as-yet-discovered set of physiological properties of replication factors. Present-day views of mesoscale organization within cells typically invoke processes such as protein gradients or membrane compartmentalization. Recently, a rapidly expanding body of work has begun to recognize protein/protein and protein/RNA liquid–liquid phase separation (LLPS) as playing critical functions in generating membraneless pseudo-organelles and co-localized bodies (Boeynaems et al., 2018). Both nuclear and cytoplasmic liquid phase condensates (also known as biomolecular condensates) have been observed and implicated in a panoply of functions, including cellular signaling (Li et al., 2012; Su et al., 2016), centrosome assembly (Woodruff et al., 2017), and chromatin/heterochromatin assembly and maintenance (Larson et al., 2017; Strom et al., 2017; reviewed in Maeshima et al., 2016). A functional role for biological condensates is only beginning to emerge; such entities may help sequester and/or co-localize certain factors to modulate biochemical output and response (Shin and Brangwynne, 2017). Intrinsically disordered amino acid regions (IDRs) are often found in proteins that phase separate and can underpin multivalent interactions that help drive phase separation (Li et al., 2012). Many phase-separating IDRs are additionally enriched for certain amino acids and are hence referred to as low-complexity domains (LCD) (Hennig et al., 2015; Kato et al., 2012). Although multiple nuclear events are now being scrutinized through the lens of phase separation, how these processes integrate and communicate with other LLPS or dispersive-state pathways is currently unknown. Here we report that the Drosophila melanogaster replication initiation factors ORC, Cdc6, and Cdt1 possess N-terminal IDRs that facilitate DNA-dependent liquid-liquid phase separation in vitro. Bioinformatic analyses reveal that these initiator IDRs possess a high-complexity sequence signature that is preserved in metazoan homologs, but that is distinct from other condensate-promoting IDRs in other cellular pathways. Biochemical studies show that these IDRs drive the selective co-assembly and enrichment of Drosophila initiation factors into liquid phases in the presence of DNA, while simultaneously excluding non-partner proteins that also phase separate. Although Drosophila Mcm2-7 does not appear to phase separate on its own, initiation factors can recruit the complex into condensates in a DNA-, ATP-, and ORC/Cdc6/Cdt1-dependent manner. In addition, cellular and genetic studies establish that the Orc1 IDR is critical for its recruitment to mitotic chromosomes and is essential for viability. Collectively, our observations not only reveal a new class of high-complexity sequences that undergo phase separation, but also provide a model for how these elements promote physical interactions between replication initiation factors that could help promote long-range chromosomal communication observed in replication programs. Results D. melanogaster Cdt1 undergoes DNA-dependent liquid-liquid phase separation (LLPS) The architecture of D. melanogaster Cdt1 (DmCdt1) is markedly different from its budding yeast counterpart (Figure 1A). Sequence analysis with the disorder prediction server DISOPRED (Jones and Cozzetto, 2015) reveals that the N-terminal sequence of DmCdt1 is predicted to be an extended intrinsically disordered region (IDR) (Figure 1A). Although the sequence of this region is not conserved per se, an N-terminal IDR is present in other metazoan Cdt1s (Table 1). Figure 1 with 1 supplement see all Download asset Open asset D. melanogaster Cdt1 undergoes DNA-dependent phase separation. (A) Architecture of D. melanogaster and S. cerevisiae Cdt1. The per-residue DISOPRED (Jones and Cozzetto, 2015) disorder prediction score is shown in the plot below each gene, with a cutoff value of 0.5 indicated by the dashed line. Residues scored above this cutoff are predicted to be disordered. (B) dsDNA-coupled (‘DNA Beads’) and control (‘Ctrl Beads’) agarose beads were used to pull-down DmCdt1 and ScCdt1. DmCdt1 but not ScCdt1 bound to the DNA-coupled beads. (C) Fluorescence anisotropy measurements of Cdt1 binding to a Cy5-labeled duplex oligonucleotide. ScCdt1 showed no evidence of binding. DmCdt1 bound with a Kd, app = 83 ± 17 nM. (D) EMSA analysis of DmCdt1 binding to duplex DNA. The complex between DNA and DmCdt1 is heterogeneous and large, and is fully well-shifted at the highest concentrations. In good agreement with anisotropy measurements, the calculated Kd, app for DmCdt1 is ~100 nM. (E) Mixing concentrated DmCdt1 with duplex DNA results in a visible increase in solution turbidity that can be reversed with the addition of KCl. (F) DIC microscopy analysis of solutions of Cdt1, DNA, and a Cdt1/DNA mixture (scale bar = 5 µm). Phase-separated droplets were evident when Cdt1 was mixed with DNA. Gel images are representative from three independent experiments. https://doi.org/10.7554/eLife.48562.002 Table 1 Conservation of initiator IDR sequence features. Analysis of eukaryotic initiator homologs for the presence of an N-terminal IDR, as well as IDR sequence features (‘pI’=isoelectric point; ‘FCR’=fraction charged residues). https://doi.org/10.7554/eLife.48562.004 Orc1Cdc6Cdt1D. melanogaster N-term IDR length: pI: FCR:362 aa 10.2 0.32246 aa 9.4 0.35294 aa 10.1 0.32Human N-term IDR length: pI: FCR:300 aa 10.7 0.33136 aa 10.6 0.29175 aa 10.6 0.29Mouse N-term IDR length: pI: FCR:298 aa 10.2 0.34141 aa 10.0 0.31178 aa 9.8 0.24X. laevis N-term IDR length: pI: FCR:327 aa 9.7 0.33140 aa 10.8 0.26248 aa 10.1 0.31C. elegans N-term IDR length: pI: FCR:242 aa 9.1 0.39172 aa 9.9 0.38194 aa 10.3 0.35D. rerio N-term IDR length: pI: FCR:391 aa 9.8 0.31156 aa 11.1 0.22309 aa 9.9 0.31S. cerevisiae N-term IDR length: pI: FCR:143 aa 4.7 0.4931 aa 6.2 0.32N/A N/A N/AS. pombe N-term IDR length: pI: FCR:117 aa 10.6 0.36133 aa 10.2 0.22N/A N/A N/A The N-terminal IDR of metazoan Cdt1 contains multiple conserved short linear motifs (SLiMs) (Davey et al., 2012) necessary for Cdt1 regulation, such as a PCNA interacting peptide (PIP) box and kinase consensus sequences (Pozo and Cook, 2016). However, known SLiMs only account for a small fraction (<25%) of the total length of the metazoan Cdt1 N-terminal IDR, so we asked whether this domain might have more general functional significance. The IDR of DmCdt1 has a predicted pI that is relatively basic (pI = 10.2); this feature, combined with the presence of tandem WH domains in the protein, suggested that DmCdt1 might bind DNA. We found that agarose beads coupled with a random 60 bp double-stranded DNA (dsDNA) were able to efficiently pull down DmCdt1, whereas no such interaction was observed for ScCdt1 (Figure 1B). Quantitative analysis of dsDNA binding using a fluorescence polarization (FP)-based assay also revealed DNA binding by DmCdt1 (Kd, app = 83 ± 17 nM) but not ScCdt1 (Figure 1C). Mouse and S. pombe Cdt1 have similarly been shown to bind DNA in vitro (Houchens et al., 2008; Yanagi et al., 2002). Electrophoretic mobility assays confirmed DNA binding by DmCdt1 (Kd, app ≈ 100 nM) but also showed that the complex is highly heterogenous (Figure 1D). Indeed, elevated concentrations of the protein (>190 nM) resulted in a well-shifted species, suggestive of increasingly higher-order assemblies; however, size-exclusion chromatography showed that purified DmCdt1 is monodisperse in solution (Figure 1—figure supplement 1A). Interestingly, in the course of conducting these studies, we found that upon mixing 20 µM DmCdt1 with stoichiometric amounts of a 60 bp duplex oligo, the solution became visibly turbid, and that this turbidity was reversible by the addition of salt (400 mM KCl) (Figure 1E). Inspection of the turbid solution by differential interference contrast (DIC) microscopy unexpectedly revealed the presence of phase-separated droplets up to 4 µm in diameter (Figure 1F). These droplets were fully absent from the DNA alone sample and barely detectable in both number and size in the DmCdt1 alone sample. Notably, Cdt1 phase separation was dependent on the length, but not the sequence, of the dsDNA substrate, with maximal LLPS occurring in the presence of oligonucleotides longer than 25 basepairs (Figure 1—figure supplement 1B–C). To confirm that the affinity determined for Cdt1 in binding DNA was not confounded by the phase transition event, we determined whether a rapid binding/exchange equilibrium was maintained in the FP DNA-binding assay using a competition assay that titrated ‘cold’ dsDNA against pre-formed Cdt1/DNA complexes (Figure 1—figure supplement 1D). The unlabeled oligonucleotide was able to fully reduce DNA binding by Cdt1 to background levels, with an inhibition constant (Kc, app = 62 ± 4 nM) comparable to the observed Kd, app (83 nM), demonstrating that the labeled DNA in droplets is freely exchangeable and therefore in equilibrium between bound and unbound states. Together, these data demonstrate not only that DmCdt1 binds DNA, but that DNA-binding in turn induces the protein to phase separate. DNA-dependent phase separation of DmCdt1 requires the N-terminal IDR and occurs at physiologic concentrations The DNA-dependency of Cdt1 phase separation predicts that droplets should be enriched for both protein and nucleic acid. To test this prediction, DmCdt1 was expressed and purified with an N-terminal enhanced Green Fluorescent Protein (eGFP) tag (eGFP-Cdt1) (Figure 2A). A Cy5-labeled duplex oligonucleotide (Cy5-dsDNA) was then mixed with an equimolar amount of protein (5:5 µM) and imaged by two-color fluorescence imaging (Figure 2B). When eGFP-Cdt1 was mixed with Cy5-dsDNA, droplets up to 3 µm in diameter appeared, all containing both protein and nucleic acid. Samples with eGFP-Cdt1 alone did not show such droplets, nor were they observed for samples with Cy5-dsDNA only. These data demonstrate that phase separation by Cdt1 is driven, at least in part, by dsDNA-induced coacervation, and further show that dye-labeled oligonucleotides can serve as a proxy for labeled protein when assessing DNA-dependent protein phase separation by fluorescence microscopy. Figure 2 with 1 supplement see all Download asset Open asset DmCdt1 phase separation is facilitated by an N-terminal IDR. (A) SDS-PAGE analysis and Coomassie stain of purified eGFP-Cdt1. (B) Samples containing eGFP-Cdt1, Cy5-dsDNA (60 bp), and a mixture of eGFP-Cdt1 and Cy5-dsDNA (‘Mixed’) were prepared and analyzed by two-color fluorescent microscopy. Droplets were observed in the Mixed sample enriched for both protein (green) and nucleic acid (red). (C) Cy5-dsDNA was imaged alone or mixed with either wild-type Cdt1 (WT) or a Cdt1 construct lacking the N-terminal IDR (∆IDR). Only WT Cdt1 could induce droplet formation. (D) Schematic of a condensate depletion assay, a method for assessing phase separation (see Materials and methods for details). (E) The depletion assay was utilized to assess the role of the Cdt1 IDR in phase separation. DNA-induced the depletion of WT Cdt1 but not ∆IDR. (F) Phase diagram for Cdt1 in the presence of equimolar amounts of sixty basepair dsDNA (filled markers = phase separation observed, unfilled markers = phase separation not observed). (G) Depletion assay to assess phase separation at sub-physiological concentrations of Cdt1. DNA-induced phase separation of Cdt1 is seen at the lowest concentration tested (50 nM). Gel images are representative of three independent experiments. https://doi.org/10.7554/eLife.48562.005 Given the known importance of protein IDRs in facilitating phase separation (Mitrea and Kriwacki, 2016), we hypothesized that the IDR of Cdt1 might likewise be critical for the condensates we observed. We therefore purified an N-terminal IDR deletion of Cdt1 (Cdt1ΔIDR) and determined whether, upon mixing with DNA, a condensed phase could form (Figure 2C). No droplets were observed when Cdt1ΔIDR was mixed with Cy5-dsDNA, even at elevated levels (10 µM protein, 10 µM DNA). To confirm these data with unlabeled DNA, we devised a simple depletion assay (Figure 2D). In this approach, protein is incubated with or without DNA, after which the denser, phase-separated material is pelleted by centrifugation, and the degree to which protein is depleted from the supernatant is assessed by SDS-PAGE. In agreement with the microscopy data, no depletion of full-length Cdt1 was observed relative to the load control in the absence of DNA (Figure 2E), indicating that no phase separation occurred. Moreover, the full-length Cdt1 signal was fully lost from the supernatant in the presence of DNA, indicating that a near complete partitioning of Cdt1 into the condensed (pelleted) phase took place. When the DmCdt1 IDR was removed, the protein was again fully retained in the supernatant, regardless of whether DNA was present or not. Notably, a construct containing only the Cdt1 N-terminal IDR residues (Cdt1IDR) bound DNA (Kd, app = 158 ± 32 nM) and underwent DNA-dependent phase separation on its own (Figure 2—figure supplement 1A–C), demonstrating that the Cdt1 IDR is necessary and sufficient for and that it with yeast Cdt1, which has an IDR (Table nor binds DNA (Figure did not phase separation in vitro (Figure 2—figure supplement However, all other metazoan Cdt1 homologs possess an N-terminal with human Cdt1 revealed to DmCdt1, it also can undergo DNA-induced liquid phase (Figure 2—figure supplement Collectively, these data demonstrate the essential role of metazoan Cdt1 IDRs in facilitating phase separation. replication in the D. melanogaster Cdt1 concentration at (Figure 2—figure supplement Given the concentration of Cdt1, it was unclear whether our biochemical which were at protein Cdt1 in To this we used fluorescence microscopy to a phase diagram for DmCdt1, protein and salt and for the presence or absence of a condensed phase (Figure Droplets were observed down to the lowest concentration of Cdt1 tested nM) when salt concentrations were set at or below physiological levels salt concentration to physiological levels the concentration of protein necessary to see phase separation to nM. When the concentration of was to no phase separation was observed for concentration of Cdt1 The of Cdt1 to phase separate at concentrations was confirmed by the depletion assay, stoichiometric of DmCdt1 and duplex DNA were prepared from to and assessed for DNA-dependent depletion of protein in the supernatant (Figure DNA-coupled of Cdt1 signal was seen at all concentrations Together, these data demonstrate that DmCdt1 undergoes DNA-dependent phase separation at physiological protein and salt concentrations. ORC and Cdc6 also partition into DNA-dependent liquid phases Given the role of the DmCdt1 N-terminal IDR in phase separation, we whether other replication initiation factors might possess disordered regions of DISOPRED (Jones and Cozzetto, we calculated the of predicted disordered as well as the disordered for each Drosophila protein for and to predicted or high disorder (Figure In addition to Cdt1, Cdc6 and two subunits of ORC and were found to possess a high of disordered as a of their total length, and each a region of disorder longer than amino Conversely, the Mcm2-7 contain than predicted disordered sequence with no disordered region amino patterns were seen for other metazoan replication initiation but not S. cerevisiae proteins (Table 1). Figure 3 with 2 see all Download asset Open asset and undergo DNA-dependent phase separation. (A) of the disorder for each Drosophila replication initiation Cdc6, and Cdt1 each contain IDRs by the each and a high of overall disordered sequence by (B) Analysis of ORC phase separation by the depletion ORC nM) phase in a DNA-dependent in the presence and absence of (C) Cy5-dsDNA µM) was imaged alone and as a mixture with ORC In the presence of ORC, phase-separated droplets (D) ORC Orc1 and have N-terminal The IDR of Orc1 is longer and is enriched for (E) Analysis of nM) and nM) phase separation by depletion of the Orc1 IDR phase separation but of the IDR has no (F) Droplets form when Cy5-dsDNA µM) is mixed with µM) but not when mixed with (G) Cdc6 phase separation was assessed by depletion assay at and 5 µM concentrations. DNA Cdc6 phase separation but this was in the presence of 1 mM Fluorescence imaging reveals phase-separated droplets when Cy5-dsDNA µM) is mixed with Cdc6 Phase separation analysis for a Cdc6 construct lacking the N-terminal IDR nM) no depletion in the presence of DNA nM). µM) is to induce droplet as assessed by fluorescence microscopy with Cy5-dsDNA Gel images are representative of three independent experiments. For proteins with a high fraction of disordered we assessed the of the disordered regions relative to known protein domains (Figure supplement 1A). In all the regions of disorder N-terminal to the of the domain of the of the Cdt1 WH domains and of the and WH domains of and Cdc6 possesses the N-terminal IDR amino and Orc1 the residues). also regions of disorder that serve as sequences between tandem such as between the and WH domain of Cdc6 amino acid or the two WH domains of Cdt1 amino acid The that Cdc6 and two subunits of ORC possess N-terminal IDRs suggested that these Cdt1, might also undergo phase separation in a DNA-dependent and that this might promote their functional within a condensed was tested by the of to phase separate using the depletion assay (Figure to the of DmCdt1 when it ORC was found to be depleted from the supernatant in the but not in the of DNA. The of did not have an on phase separation by ORC (Figure and fluorescence microscopy with Cy5-dsDNA confirmed that the DNA-dependent depletion of ORC from the supernatant was to phase separation (Figure no droplets are observed for DNA ORC phase separation a of protein concentrations (50 to nM) revealed ORC depletion when DNA was present but not when it was absent (Figure supplement 1B). Depletion regardless of oligonucleotide sequence (Figure supplement Cdt1, maximal phase separation with oligonucleotides longer than 25 basepairs (Figure supplement 1D). Cdt1, ORC is able to undergo LLPS at physiological concentrations in the presence of a of DNA without DNA sequence We whether to undergo phase separation could be to the IDR of a specific or whether the IDRs of both Orc1 and are for this In of length, the Orc1 IDR, at amino is 100 residues longer than the IDR (Figure of sequence between Orc1 and IDRs – by the relative of charged and residues – show that the IDRs of Orc1 and are highly in with the for Orc1 and within for each amino acid Interestingly, Orc1 and IDR amino acids are near across and charged this of residues is the predicted of these which is within the is often enriched in protein for both Orc1 and is than for such a region et al., 2018). Although the Orc1 and IDRs are highly in of amino acid they show a in their the Orc1 IDR (pI = is enriched for residues and whereas the IDR (pI = is enriched for residues and that this in might be for facilitating LLPS by ORC, we two with the Orc1 IDR and the other lacking the IDR Phase separation for both was then assessed by the depletion assay and fluorescence imaging with showed no evidence for phase separation in either assay, whereas the construct lacking the IDR wild-type in both assays (Figure These demonstrate that Drosophila ORC phase using interactions that the N-terminal IDR of Orc1 but not Given the predicted of in the Orc1 IDR we were to that this region is relatively across the Drosophila Figure supplement and that the length, and pI of this region is conserved across the metazoan (Table 1). Although S. cerevisiae Orc1 possesses an N-terminal IDR, this region is than observed in and also has an pI (Table 1). with this budding yeast ORC does not phase separate (Figure supplement 1E). We asked whether Cdc6 can undergo DNA-dependent to budding yeast Cdc6 et al., was found to with DNA by dsDNA-coupled agarose pull-down the presence of no on Cdc6 DNA-binding (Figure supplement 1F). whether Cdc6 phase in the presence of DNA using the depletion assay (Figure we observed a DNA-dependent depletion of Cdc6 from the supernatant (Figure however, in contrast to ORC, this depletion was by the presence of (Figure A report that can function as a to protein may provide an for its on Cdc6 phase separation et al., 2017). the assay at concentrations of that DmCdt1, which to fully partition into phases at concentrations (Figure 2E), Cdc6 was only depleted from the supernatant (Figure When a of protein concentrations and DNA sequences (Figure supplement Cdc6 showed phase separating the protein into the Notably, Cdc6 underwent complete partitioning into

Liquid-liquid phase separation (LLPS) regulates diverse biological functions by mediating the assembly of biomolecular condensates. However, it remains unclear how host LLPS is targeted by viruses during infection. Here we show that a plant bunyaviral protein, the disease-specific protein (SP) encoded by rice stripe virus (RSV), possesses phase separation potential through its N-terminal intrinsically disordered region 1 (IDR1). In vivo, however, SP does not form phase-separated biomolecular condensates independently but utilizes its phase separation properties to interfere with the phase separation of the SERRATE protein (SE), a key component of Dicing bodies essential for microRNA processing. By disrupting SE phase separation, SP inhibits D-body assembly and miRNA biogenesis. Our study demonstrates that a viral protein can modulate host microRNA processing by targeting LLPS, revealing a previously uncharacterized mechanism involved in viral infection strategies and miRNA biogenesis regulation in plants.

Liquid-liquid phase separation (LLPS) of proteins into biomolecular condensates has emerged as a fundamental principle underpinning cellular function and malfunction. Indeed, many human pathologies, including protein misfolding diseases, are linked to aberrant liquid-to-solid phase transitions, and disease-associated protein aggregates often nucleate through phase separation. The molecular level determinants that promote pathological phase transitions remain, however, poorly understood. Here we study LLPS of the microtubule-associated protein Tau, whose aberrant aggregation is associated with a number of neurodegenerative diseases, including Alzheimer's disease. Using single molecule spectroscopy, we probe directly the conformational changes that the protein undergoes as a result of LLPS. We perform single-molecule FRET and fluorescence correlation spectroscopy experiments to monitor the intra- and intermolecular changes and demonstrate that the N- and C-terminal regions of Tau become extended, thus exposing the microtubule-binding region. These changes facilitate intermolecular interactions and allow for the formation of nanoscale clusters of Tau. Our results suggest that these clusters can promote the fibrillization of Tau, which can be dramatically accelerated by disease-related mutations P301L and P301S. Our findings thus provide important molecular insights into the mechanism of protein phase separation and the conversion of protein condensates from functional liquid assemblies to pathological aggregates.

A model for identification of potential phase-separated proteins based on protein sequence, structure and cellular distribution

Liquid–Liquid Phase Separation and Its Mechanistic Role in Pathological Protein Aggregation

Liquid-liquid phase separation (LLPS) of proteins and nucleic acids into membraneless organelles (MLOs) plays a critical role in sustaining fundamental physiological processes. However, creating artificial coacervate droplets in living cells from exogenous molecules and modulating the functions of MLOs remain challenging. To address this concern, here we reported enzyme-induced in situ phase separation of peptides into droplets targeting MLO stress granule (SG) for enhanced cancer chemotherapy. The peptide YSO4F containing two sulfated tyrosine residues undergoes sulfatase-responsive LLPS into droplets. Cellular studies confirm in situ phase separation of YSO4F selectively in sulfatase-overexpressing cancer cells. By integrating with appropriate ligands, the in situ-formed droplets d-YF-LSG coacervate with SGs driven by association between the ligand with SG key component protein G3BP2. Mechanistic studies illustrate that the in situ-formed droplets enhance the cytotoxicity of sorafenib via activating caspase-dependent apoptosis. Furthermore, animal experiments confirm that administration of the in situ-formed droplets with sorafenib significantly inhibits tumor growth in murine models bearing tumors, accompanied by an excellent biosafety profile. The findings in this studyelucidate an innovative approach for in situ formulation of coacervate droplets within tumor cells and a new material for targeting membraneless organelles, thus providing a promising new strategy for disease organelle-targeted therapy in the future.

Protein liquid-liquid phase separation (LLPS) is driven by intrinsically disordered regions and multivalent binding domains, both of which are common features of diverse microtubule (MT) regulators. Many in vitro studies have dissected the mechanisms by which MT-binding proteins (MBPs) regulate MT nucleation, stabilization and dynamics, and investigated whether LLPS plays a role in these processes. However, more recent in vivo studies have focused on how MBP LLPS affects biological functions throughout neuronal development. Dysregulation of MBP LLPS can lead to formation of aggregates - an underlying feature in many neurodegenerative diseases - such as the tau neurofibrillary tangles present in Alzheimer's disease. In this Review, we highlight progress towards understanding the regulation of MT dynamics through the lens of phase separation of MBPs and associated cytoskeletal regulators, from both in vitro and in vivo studies. We also discuss how LLPS of MBPs regulates neuronal development and maintains homeostasis in mature neurons.

Hagfish intermediate filament (IF) proteins, which are released from their cells in response to external stimuli, exhibit excellent mechanical properties and biodegradability, with broad material development applications. This study aims to construct hagfish IF-based hydrogels by modulating intermolecular interactions through the coexistence of electrolytes and polymers that are capable of inducing phase separation and subsequent phase transition. Analyses of monomerized and purified IF proteins revealed condensate formation in the presence of appropriate polymer and electrolyte concentrations. Furthermore, the condensate formed by liquid-liquid phase separation and/or liquid-liquid crystalline phase separation underwent a liquid-solid phase transition to form a macroscopic hydrogel. During hydrogelation, the proteins exhibited a structural shift toward β-sheet-rich conformations, indicating that intermolecular β-sheet interactions serve as the primary driving force for structuring. These findings suggest that condensates and hydrogels can be generated via phase separation and phase transitions of hagfish IF proteins, offering a promising route toward sustainable biomaterial development.

Liquid-liquid phase separation (LLPS) of proteins and nucleic acids is a phenomenon that underlies membraneless compartmentalization of the cell. The underlying molecular interactions that underpin biomolecular LLPS have been of increased interest due to the importance of membraneless organelles in facilitating various biological processes and the disease association of several of the proteins that mediate LLPS. Proteins that are able to undergo LLPS often contain intrinsically disordered regions and remain dynamic in solution. Solution-state NMR spectroscopy has emerged as a leading structural technique to characterize protein LLPS due to the variety and specificity of information that can be obtained about intrinsically disordered sequences. This review discusses practical aspects of studying LLPS by NMR, summarizes recent work on the molecular aspects of LLPS of various protein systems, and discusses future opportunities for characterizing the molecular details of LLPS to modulate phase separation.

Biomolecular condensates formed through liquid-liquid phase separation regulate cellular processes, and their dysregulation causes disease. Current methods for identifying endogenous phase-separating proteins have low throughput and cannot capture dynamic responses to stimuli. Here we present a protocol combining osmotic compression or transforming growth factor-β (TGF-β) treatment to induce condensation with sucrose density gradient centrifugation and quantitative mass spectrometry to enable systematic, high-throughput identification of endogenous condensates and phase-separating proteins. The method exploits the density changes that occur when phase-separating proteins undergo oligomerization during condensate formation. In H1975 cells, we identified over 1,500 phase-separating proteins under osmotic compression or TGF-β treatment; 538 of these candidates were not present in PhaSepDB, a database that compiles in vivo, in vitro and omics-derived proteins. The approach detects constitutive condensates and proteins that dynamically phase-separate in response to osmotic stress or TGF-β signaling. This protocol provides proteome-wide analysis of fractions of proteins having different densities and enables temporal resolution of phase-separation events. The procedure takes ~9 d and requires expertise in cell culture, biochemistry and mass spectrometry. This method enables systematic study of biomolecular condensates and disease-associated phase-separation mechanisms.

Phase separation has emerged as an important mechanism explaining the formation of certain biomolecular condensates. Biological phase separation is often driven by the multivalent interactions of modular protein domains. Beyond valency, the physical features of folded domains that promote phase separation are poorly understood. We used a model system─the small ubiquitin modifier (SUMO) and its peptide ligand, the SUMO interaction motif (SIM)─to examine how domain surface charge influences multivalency-driven phase separation. Phase separation of polySUMO and polySIM was altered by pH via a change in the protonation state of SUMO surface histidines. These effects were recapitulated by histidine mutations, which modulated SUMO solubility and polySUMO-polySIM phase separation in parallel and were quantitatively explained by atomistic modeling of weak interactions among proteins in the system. Thus, surface charge can tune the phase separation of multivalent proteins, suggesting a means of controlling phase separation biologically, evolutionarily, and therapeutically.

Physiological, Pathological, and Targetable Membraneless Organelles in Neurons

Liquid-liquid phase separation plays a critical role in cellular processes, including protein aggregation and RNA metabolism, by forming membraneless subcellular structures. Accurate identification of phase-separated proteins is essential for understanding and controlling these processes. Traditional identification methods are effective but often costly and time-consuming. The recent machine learning methods have reduced these costs, but most models are restricted to classifying scaffold and client proteins with limited experimental conditions. To address this limitation, we developed a Mamba-based encoder using contrastive learning that incorporates separation probability, protein type, and experimental conditions. Our model achieved 95.2% accuracy in predicting phase-separated proteins and an ROCAUC score of 0.87 in classifying scaffold and client proteins. Further validation in the DgHBP-2 drug delivery system demonstrated its potential for condition modulation in drug development. This study provides an effective framework for the accurate identification and control of phase separation, facilitating advancements in biomedical research and therapeutic applications.

Liquid Liquid Phase Separation (LLPS) has emerged as a mechanism for the assembly of membraneless organelles in eukaryotes, but little is known about this process in bacteria. LLPS refers to the ability of macromolecules to demix into a dilute phase and a dense phase, called a ‘biomolecular condensate’, which can be observed as clusters or foci in the cell. The major challenge for the study of LLPS in bacteria is the poor spatial resolution of foci in such tiny cells. As a result, it is difficult to demonstrate the liquid‐like nature of a focus in bacterial cells using the conventional approaches for studying large condensates in eukaryotic cells. Here, we developed a rigorous experimental framework for the characterization of LLPS in bacteria, using Escherichia coli as the host organism and the intrinsically disordered protein McdB, which robustly forms liquid‐like droplets in vitro. McdB is a protein that coats a bacterial organelle called the carboxysome. This coating demarcates the carboxysome as cargo for its positioning system, which equally distributes carboxysomes along the cell length of rod‐shaped cyanobacteria. We developed a suite of experiments to investigate the LLPS activity of McdB in vivo, based on the ability of biomolecular condensates to tune their size and shape, fuse, dissolve, and transition between phase states. We used both overexpression and tunable promoters to express fluorescent fusions of McdB and cIEP8, a well‐known aggregator protein. We found that fluorescent fusions of McdB formed nucleoid‐excluded foci in E. coli, but also maintained a soluble phase in the cytoplasm, consistent with LLPS theory. The aggregator protein cIEP8, on the other hand, lacked a soluble fraction in the cytoplasm. Condensates form at a saturation concentration threshold, called Csat. A hallmark of LLPS is that condensates will dissolve if the concentration drops below Csat, while insoluble aggregates should remain as stable foci even after dilution. We decreased protein concentration in vivo by increasing cell volume and by generational dilution via cell division. In both methods, McdB foci dissolved while cIEP8 foci remained intact as insoluble aggregates in response to decreased concentration in the cell. Finally, we also discovered that a well‐established marker for insoluble protein aggregates in vivo, IbpA, does not colocalize with McdB foci. The result suggests that the colocalization of IbpA foci can be used as a broad‐use sensor for the material state of protein complexes in bacterial cells. Our results provide multiple lines of evidence in support of LLPS of McdB in vivo. More broadly, our experimental framework for studying LLPS in bacteria overcomes current limitations in the field and can be used to assess the LLPS activity of other proteins of interest in bacterial cells.