biomolecules Article In-Silico Analysis of pH-Dependent Liquid-Liquid Phase Separation in Intrinsically Disordered Proteins Carlos Pintado-Grima † , Oriol Bárcenas † and Salvador Ventura * Institut de Biotecnologia i Biomedicina, Departament de Bioquímica i Biologia Molecular, Universitat Autònoma de Barcelona, Bellaterra, 08193 Barcelona, Spain; carlos.pintado@uab.cat (C.P.-G.); oriol.barcenas@autonoma.cat (O.B.) * Correspondence: salvador.ventura@uab.cat † These authors contributed equally to this work. Abstract: Intrinsically disordered proteins (IDPs) are essential players in the assembly of biomolecular condensates during liquid–liquid phase separation (LLPS). Disordered regions (IDRs) are significantly exposed to the solvent and, therefore, highly influenced by fluctuations in the microenvironment. Extrinsic factors, such as pH, modify the solubility and disorder state of IDPs, which in turn may impact the formation of liquid condensates. However, little attention has been paid to how the solution pH influences LLPS, despite knowing that this process is context-dependent. Here, we have conducted a large-scale in-silico analysis of pH-dependent solubility and disorder in IDRs known to be involved in LLPS (LLPS-DRs). We found that LLPS-DRs present maximum solubility around physiological pH, where LLPS often occurs, and identified significant differences in solubility and disorder between proteins that can phase-separate by themselves or those that require a partner. We also analyzed the effect of mutations in the resulting solubility profiles of LLPS-DRs and discussed how, as a general trend, LLPS-DRs display physicochemical properties that permit their LLPS at physiologically relevant pHs. Citation: Pintado-Grima, C.; Keywords: pH; liquid-liquid phase separation; intrinsically disordered proteins; protein solubility; Bárcenas, O.; Ventura, S. In-Silico protein disorder; mutations; bioinformatics Analysis of pH-Dependent Liquid-Liquid Phase Separation in Intrinsically Disordered Proteins. Biomolecules 2022, 12, 974. https:// 1. Introduction doi.org/10.3390/biom12070974 Our view of cellular organization has been progressively changing over the last decade. Academic Editor: Sonia Longhi Since the discovery of the first liquid droplets in C. elegans embryo’s cells [1], numerous Received: 22 June 2022 membrane-less organelles (MLOs) with a wide variety of biological functions have been Accepted: 8 July 2022 described in different organisms [2–5]. In contrast to their classic membrane-enclosed Published: 12 July 2022 counterparts, MLOs are dynamic supramolecular structures that can undergo reversible liquid–liquid phase separation (LLPS) in response to specific stimuli [6,7]. The reversible Publisher’s Note: MDPI stays neutral and tunable nature of LLPS has turned into an effective compartmentalization mechanism with regard to jurisdictional claims in for the always-changing cellular milieu, allowing the selective spatiotemporal formation of published maps and institutional affil- biomolecular condensates [8]. MLOs are often enriched in intrinsically disordered proteins iations. (IDPs) and/or proteins bearing unstructured regions and low-complexity domains [9–11]. These sequences play an important role in droplet formation by establishing multivalent weak intermolecular interactions [12,13]. Indeed, mutations in some of these unstructured Copyright: © 2022 by the authors. regions might deregulate the equilibrium of LLPS and lead to the onset of neurodegen- Licensee MDPI, Basel, Switzerland. erative diseases such as amyotrophic lateral sclerosis (ALS), frontotemporal dementia This article is an open access article (FTD) [10,14,15], or muscular dystrophies [16,17]. distributed under the terms and Intrinsically disordered regions (IDRs) are exposed to the solvent and thus influ- conditions of the Creative Commons enced by the protein microenvironment, whose fluctuation can trigger conformational Attribution (CC BY) license (https:// switches [18]. This context-dependency also applies to LLPS processes, and parameters creativecommons.org/licenses/by/ such as ionic strength, temperature, or pH have been described as regulatory elements 4.0/). in these reactions [6]. Other factors apart from solvent conditions can also influence the Biomolecules 2022, 12, 974. https://doi.org/10.3390/biom12070974 https://www.mdpi.com/journal/biomolecules Biomolecules 2022, 12, 974 2 of 12 behavior of IDPs in LLPS, including the protein concentration and the presence of other biomolecules, mainly RNA and partner proteins [19]. Therefore, the capacity of a protein to phase-separate is not binary and cannot be univocally attributed to the intrinsic properties of the sequence as it strongly depends on the specific conditions of the cellular milieu at any given time. In recent work, we accounted for the effect of pH on the solubility [20] and disorder state [21] of IDPs using different equations that simultaneously consider the impact of the pH on sequence hydrophobicity and net charge. This allowed us to develop two novel bioinformatic tools to predict pH-dependent solubility and disorder: SolupHred [22] and DispHScan [23], respectively, which recapitulate previous and novel experimental data [24–26]. The solubility and degree of disorder of a protein at given pH influence its propensity to phase-separate, but large-scale analyses exploring these connections are scarce. To provide insights on the role of pH in LLPS, we conducted an in-silico study of solubility and disorder at different pH values for LLPS-involved disordered regions (LLPS-DRs), accounting for a total of 1600 sequences, using the SolupHred and DispHScan algorithms. 2. Materials and Methods 2.1. Dataset Generation LLPS regions were extracted from PhaSepDB, a manually curated database of liquid- liquid phase separation-related proteins and MLOs [27]. Afterward, the disordered nature of these regions was surveyed using the IUPRED2A server [28]. High-confidence LLPS-DRs (sequences longer than 20 amino acids with an IUPRED score ≥ 0.5) were saved for the analyses (n = 1600). Two different subgroups were created according to the ability of each LLPS region to phase-separate by itself (psself) or with the help of a partner (psother). A second dataset of general disordered segments was generated for comparison purposes. IDRs longer than 20 amino acids were obtained from DisProt (release 2021_08), a manually curated database of experimentally validated IDPs [29]. 2.2. pH-Dependent Solubility and Disorder Analyses The effect of pH on both LLPS-DRs’ solubility and disorder was predicted by two in-house algorithms considering this key variable in their pipelines. SolupHred and DispHScan web servers recalculate protein lipophilicity and net charge as a pH function to predict IDPs’ solubility and disorder in a specific pH context [22,23]. All disordered sequences were run with SolupHred and DispHScan in a pH range between 4 and 9, using a step size of 0.1 to account for small variations. The software output solubility and disorder scores, respectively. The disorder score is named the DispH score, with lower values indicating a more disordered state. Statistical significance between variables and/or datasets was assessed with Mann-Whitney-Wilcoxon two-sided test with Bonferroni correction. p-values were marked with asterisks to better convey statistical significance (p > 0.05 (ns), p ≤ 0.05 (*), p ≤ 0.01 (**), p ≤ 0.001 (***), p ≤ 0.0001 (****)). As for the statistical significance of the linear regression (whether the distribution follows a slope of 0 or not), a Wald Test with t-distribution of the test statistic was used. Statistical tests are described in the Supplementary Material of this article. 3. Results 3.1. LLPS-DRs Present Maximum Solubility around Neutral pH One of the main uncertainties when dealing with IDPs and LLPS is the conformational state that disordered segments present during the phase-separation process. Some studies suggest that LLPS-DRs contain aggregation-prone regions (APRs), which are needed for condensate formation, and that they act by reducing the solubility in the conditions in which LLPS occurs [30,31]. To explore this hypothesis, we analyzed the pH-dependent solubility of LLPS-DRs in the 4 to 9 pH range, comparing the obtained results with the pHs at which LLPS was described. Given that similar solubility scores may span various pH units, with the pHs at which LLPS was described. Given that similar solubility scores may span various pH units, we assumed 10% maximum/minimum solubilities as the most repre- Biomolecules 2022, 12, 974 sentative variables for the analysis. The results revealed that solubility in LLPS-DRs tends 3 of 12 to be maximum around neutral pH (μ = 6.96, σ = 1.29, Figure 1A), whereas minimum solubilities are mainly achieved at the extremes of the analyzed pH interval (Figure 1B). When studying buffer conditions described for LLPS, a similar result was observed (n = we 181, assumed 10% μ = 7.50; σ =maximum/minimum 0.43). This suggests that solubilities LLPS-DRs areas the mostnear soluble representative physiologicalvariables pH for values the analysis. and thatThe results LLPS occursrevealed at pHs atthat solubility which, in LLPS-DRs as a group, tends these protein to bedisplay regions maximum around neutral significant pH (µ To solubility. = 6.96, = 1.29, these furtherσ support Figure 1A), whereas findings, minimum the absolute solubilities differences in maxi- are mumachieved mainly and minimum at thesolubility extremes scores were of the individually analyzed compared pH interval with those (Figure 1B). obtained at When studying the conditions buffer pH where LLPS was described described for LLPS,(Figure 2A). The a similar analysis result was confirms observedthat, (n =as181, a trend, µ = 7.50; σ = effective LLPS-promoting 0.43). This suggests thatpHs resembleare LLPS-DRs more those at soluble which near the correspondent physiological proteins pH values and that are maximally soluble than those with minimal solubility. LLPS occurs at pHs at which, as a group, these protein regions display significant solubility. Overall, the results indicate that pH solution conditions favoring high solubility over- To further support these findings, the absolute differences in maximum and minimum lap with those at which condensation is experimentally observed, suggesting that these solubility scores were individually compared with those obtained at the pH where LLPS molecular signatures might be important for LLPS and relevant at the cellular physiolog- wasical described (Figure 2A). The analysis confirms that, as a trend, effective LLPS-promoting pH. This is surprising since it seems to contradict the general assumption that pro- pHsaggregational resemble more those at conditions arewhich the correspondent necessary for LLPS. proteins are maximally soluble than those with minimal solubility. Figure 1. 10% maximum (A) and 10% minimum (B) solubility distribution of LLPS-DRs in the range Figure 1. 10% maximum (A) and 10% minimum (B) solubility distribution of LLPS-DRs in the range between pH 4 and 9 using step size 0.1. Maximum solubility is mostly attained around neutral pH between Biomolecules 2022, 12, x FOR PEER REVIEW pH whereas (μ = 6.95), 4 and 9 using stepsolubilities minimum size 0.1. Maximum are found at solubility is of the extremes mostly the pHattained interval, around neutral 4 of 13 pH in more acidic (µ =or 6.95), basicwhereas minimum solubilities are found at the extremes of the pH interval, in more acidic conditions. or basic conditions. Figure 2. Absolute differences in LLPS-DR solubility (A) and disorder (B) scores between the Figure 2. Absolute differences in LLPS-DR solubility (A) and disorder (B) scores between the pre- predicted maximum dicted maximum or minimum or minimum valuevalue and and the the at score score at the the pH pHLLPS where where LLPS was was described. described. pH condi- pH conditions in LLPS tions in LLPS buffersbuffers are to are closer closer to LLPS-DR LLPS-DR maximum maximum solubilities solubilities (p = 6.080 ×(p10 ) and× =−316.080 10−31 ) and minimum DispH scores minimum (p =scores DispH −15). Four 3.905 ×(p10= 3.905 − 15 × asterisks 10 ). Four(****)asterisks indicate p ≤ 0.0001. (****) indicate p ≤ 0.0001. 3.2. Overall, the results A Link between indicate that pH-Dependent pH solution Solubility conditionsDisorder and pH-Dependent favoringin high solubility over- LLPS-DRs lap withWe performed the same kind of analysis as above to assess whether pHs atthat those at which condensation is experimentally observed, suggesting whichthese LLPS is reported coincide with conditions that favor disorder or, on the contrary, promote compactness of LLPS-DRs (Figure 2B). The data indicate that minimum DispH scores and thus larger disorder content match better conditions at which LLPS experimentally occurs. The relationship between solubility and conformational disorder in the LLPS-DR da- Biomolecules 2022, 12, 974 4 of 12 molecular signatures might be important for LLPS and relevant at the cellular physio- logical pH. This is surprising since it seems to contradict the general assumption that pro-aggregational conditions are necessary for LLPS. 3.2. A Link between pH-Dependent Solubility and pH-Dependent Disorder in LLPS-DRs We performed the same kind of analysis as above to assess whether pHs at which LLPS is reported coincide with conditions that favor disorder or, on the contrary, promote compactness of LLPS-DRs (Figure 2B). The data indicate that minimum DispH scores and thus larger disorder content match better conditions at which LLPS experimentally occurs. The relationship between solubility and conformational disorder in the LLPS-DR dataset was then analyzed. When plotting the solubility and disorder scores at neutral pH (pH = 7.0), a highly significant linear correlation was observed between these two vari- ables (R2 = 0.90, p < 1−10 ) (Figure 3), suggesting that solubility and disorder are intimately ligated in this dataset. Furthermore, higher solubility scores were correlated with minimum DispH scores (maximum disorder) and vice-versa. This indicates that regions that populate a more soluble sequence space are also more prone to be disordered. This makes sense since the presence of low-soluble sequences with a high degree of disorder would render them aggregation-prone and thus harmful at the analyzed physiological pH, whereas a higher degree of compactness would protect their hydrophobic residues from exposure to the solvent, at least transiently. Conversely, highly soluble sequences would find it challenging to attain a compact Biomolecules 2022, 12, x FOR PEERconformation, REVIEW and their presence in an unfolded state would not represent a signifi- 5 of 13 cant risk of establishing aberrant hydrophobic interactions with other cellular components. Figure 3. Correlation Figure 3. Correlation betweenbetween LLPS-DR disorder LLPS-DR and solubility disorder and scores at pH 7.0.scores solubility A significant linear7.0. A significant linear at pH correlation can be observed (R2 = 0.90, p < 1−10). Higher solubility scores are associated with lower correlation can DispHbescores observed 2 (R =states). (more disordered − 10 0.90, p < 1 ). Higher solubility scores are associated with lower DispH scores (more disordered states). 3.3. Different Datasets in LLPS-DRs Present Distinct Property Distributions Not all proteins present in MLOs can undergo independent LLPS [19]. As we said, 3.3. Different thisDatasets in LLPS-DRs process is highly Present context-dependent [6], andDistinct Property many proteins require a Distributions partner, often a nucleic acid molecule, to phase-separate. Therefore, the properties of LLPS-DRs may vary Not allaccording proteins present to their ability in MLOsLLPS to undergo can by undergo themselvesindependent (psself) or with aLLPS partner [19]. As we said, this process is highly context-dependent [6], and many proteins require a ob- (psother). Indeed, no differences in calculated solubility and disorder at pH 7.0 were partner, often a nucleic served between the entire LLPS-DRs dataset (without differentiating psself and psother acid molecule, to phase-separate. sequences) and the IDRs present inTherefore, the properties DisProt. Conversely, when the studyof LLPS-DRs considered psself may vary according to their ability to undergo and psother LLPS by regions separately, thethemselves significance level (psself) or increased dramatically with a when partnercom- (psother). Indeed, no pared with DisProt and between them. This indicates that psself and psother sequences differences need in calculated solubility and disorder at pH 7.0 were observed between the entire to be studied independently. LLPS-DRs dataset Psself(without regions showeddifferentiating psself lower solubility scores thanand psother psother sequences) and DisProt (Figure 4A), and the IDRs present which, not surprisingly, was associated with higher DispH scores (more ordered) (Figure in DisProt. 4B). Conversely, when the study considered psself and psother These differences are more apparent in the distribution plot of solubility and disorder regions separately, the significancescores levelfordramatically psself and psother datasets at pH increased 7.0 (Figure when 4C,D). This compared indicates with that psself DisProt and between them. This regions tend to be less soluble and disordered than psother at pH 7.0. This allows us to indicates that psself reconcile ourand psother analysis with the sequences needpropensity view that aggregation to be studied independently. is a player in LLPS and to reformulate it. For psself proteins, which can separate efficiently and autonomously, the situation is not that they experiment LLPS at pHs at which their solubility is at a min- imum, but instead that its intrinsic solubility at physiological pH, despite significant, is below the average solubility of disordered proteins, which facilitate homotypic self-as- sembly. An inverse trend applies to disorder. Still, this association only applies for psself Biomolecules 2022, 12, 974 5 of 12 Psself regions showed lower solubility scores than psother and DisProt (Figure 4A), which, not surprisingly, was associated with higher DispH scores (more ordered) (Figure 4B). These differences are more apparent in the distribution plot of solubility and disorder scores for psself and psother datasets at pH 7.0 (Figure 4C,D). This indicates that psself regions tend to be less soluble and disordered than psother at pH 7.0. This allows us to reconcile our analysis with the view that aggregation propensity is a player in LLPS and to reformulate it. For psself proteins, which can separate efficiently and autonomously, the situation is not that they experiment LLPS at pHs at which their solubility is at a minimum, Biomolecules 2022, 12, x FOR PEER REVIEW 6 of 13 but instead that its intrinsic solubility at physiological pH, despite significant, is below the average solubility of disordered proteins, which facilitate homotypic self-assembly. An inverse trend applies to disorder. Still, this association only applies for psself proteins since proteins since proteins that require an interactor to be incorporated in MLOs are not only proteins morethat require soluble an interactor than psself to bethat ones, but also incorporated inIDRs the conjunct of MLOs are notand in DisProt only willmore find soluble than difficulties psself ones, but also that the conjunct of IDRs in DisProt and will find difficulties to self-assemble without a scaffolding molecule. All these biophysical connec- to self-assemble without tions are masked a scaffolding when studying themolecule. Allas LLPS dataset these biophysical a homogeneous connections protein group. are masked when studying the LLPS dataset as a homogeneous protein group. FigureFigure 4. Comparison 4. Comparison betweensolubility between solubility (A) (A)and anddisorder disorder(B) scores of DisProt, (B) scores PhaSepDB, of DisProt, psself, PhaSepDB, psself, and psother datasets at pH 7.0. Differences in disorder and solubility are non-significant between and psother datasets at pH 7.0. Differences in disorder and solubility are non-significant between DisProt and PhaSepDB. However, when psself and psother are treated independently, major differ- DisProt and ences PhaSepDB. can be observedHowever, with DisProtwhen and psself among andthesepsother sub-sets.areThetreated independently, psself dataset major differ- exhibits lower encessolubilities (C) and lower can be observed withdisorder DisProt (D)and thanamong psother these sequences *. The density sub-sets. coordinate The psself provides dataset a exhibits lower representation of the number of observed sequences for each given score. * (A) Solubility: DisProt– solubilities (C) and lower disorder (D) than psother sequences *. The density coordinate provides a psself (2.321 × 10−05); DisProt–psother (2.000 × 10−07); psself-psother (2.744 × 10−12). * (B) Disorder: representation of the DisProt–psself number (3.282 × 10−07);of observed sequences DisProt–psother for (2.902 × 10 −05);each given score. psself-psother (4.457* ×(A) Solubility: 10−13). Four as- DisProt– psselfterisks (2.321(****) −05 ); DisProt–psother × 10indicate 10−07 ); psself-psother (2.744 × 10−12 ). * (B) Disorder: (2.000p×> 0.05. p ≤ 0.0001. “ns” indicate DisProt–psself (3.282 × 10−07 ); DisProt–psother (2.902 × 10−05 ); psself-psother (4.457 × 10−13 ). Four 3.4. Psself Regions present Lower Dispersion in Solubility and Disorder in the Physiological pH asterisks (****) indicate p ≤ 0.0001. “ns” indicate p > 0.05. Range When the differences between maximum and minimum solubility and disorder scores were analyzed in the entire selected pH range (from pH 4.0 to pH 9.0 with a step size of 0.1, as detailed in the Methods section), DisProt exhibited the broader dispersion Biomolecules 2022, 12, 974 6 of 12 Biomolecules 2022, 12, x FOR PEER REVIEW 7 of 13 3.4. Psself Regions Present Lower Dispersion in Solubility and Disorder in the Physiological pH Range When the differences between maximum and minimum solubility and disorder scores were analyzed in the entire selected pH range (from pH 4.0 to pH 9.0 with a step size of 0.1, of bothassolubility detailed and disorder in the Sectionparameters, 2), DisProtwith psself the exhibited being the dataset broader with the dispersion lowest of both solubility dispersionand disorder (Figure parameters, 5A,B). with psself This suggests being the that psself dataset regions arewith less the lowest sensible todispersion pH fluctua- (Figure tions and5A,B). This suggests populate a narrowerthatsolubility-disorder psself regions are lessspace,sensible likely to pH fluctuations compatible and with cellular populate conditions in which LLPS can still occur even if pH deviations arise, as long as theyinare a narrower solubility-disorder space, likely compatible with cellular conditions which not veryLLPS can On large. still the occur even other if pHthe hand, deviations dispersionarise, of as long asregions psother they areisnot very large.wider significantly On the forother both hand, the dispersion parameters, of psother suggesting a lowerregions is significantly selective pressure towider keep for theirboth parameters, solubility/disor- suggesting a lower der properties selective restricted inpressure to keep their a cell-compatible pHsolubility/disorder gradient, which isproperties expected, restricted as they are in a cell-compatible pH gradient, not competent for phase separation alone.which is expected, as they are not competent for phase separation alone. Figure 5. Difference between maximum and minimum solubility (A) and disorder (B) scores for Figure 5. Difference between maximum and minimum solubility (A) and disorder (B) scores for DisProt, PhaSepDB, psself, and psother datasets. The DisProt dataset exhibits a wider dispersion, DisProt, PhaSepDB, psself, and psother datasets. The DisProt dataset exhibits a wider dispersion, understood as the difference between the maximum and minimum values, in comparison with the understood as the difference between the maximum and minimum values, in comparison with the PhaSepDB, psself, and psother datasets. In consonance with previous results, psother and psself PhaSepDB, psself, and psother datasets. In consonance with previous results, psother and psself datasets datasetsare aresignificantly significantlydifferent different *. *. * (A) Solubility: * (A) Solubility:DisProt-PhaSepDB DisProt-PhaSepDB (1.526 × 10×−10 (1.526 19 ); DisProt– −19); DisProt– psself × − 31 × − 06 × − 14 psself (2.064 × 10 ); DisProt–psother (5.508 × 10 ); psself-psother (2.009 × 10 ). * (B)Disorder: (2.064 10 −31); DisProt–psother (5.508 10 ); −06 psself-psother (2.009 10 ). −14 * (B) Disorder: DisProt-PhaSepDB × ×1010 −14 × 10 −29 ); DisProt–psother (5.526 × 10−03 ); DisProt-PhaSepDB(4.496 (4.496 ););DisProt–psself −14 DisProt–psself (3.325 (3.325 × 10−29 ); DisProt–psother (5.526 × 10−03); psself-psother (8.586×× 10 psself-psother(8.586 10−−1717).).Two Twoasterisks asterisks(**) indicate pp ≤≤0.01. (**)indicate 0.01.Four Fourasterisks asterisks(****) (****)indicate indicatep ≤ 0.0001. p ≤ 0.0001. 3.5. 3.5.Case CaseStudy StudyofofIndependent IndependentLLPSLLPSHappening HappeningatatPhysiological Physiological pHpH To Tocontextualize contextualizethe previous the previousresults with results withdefined cases defined of LLPS, cases we sought of LLPS, the litera- we sought the lit- ture for IDPs whose LLPS was experimentally demonstrated to occur at physiological pH. erature for IDPs whose LLPS was experimentally demonstrated to occur at physiological An example of a well-characterized LLPS-DR is the low complexity domain (LCD) of pH. TDP-43, a protein associated with neurodegenerative disorders such as ALS or FTD [10,14,32]. An example of a well-characterized LLPS-DR is the low complexity domain (LCD) of Experimental studies indicate that LLPS of the TDP-43 LCD is pH-dependent [33]. LLPS was TDP-43, a protein associated with neurodegenerative disorders such as ALS or FTD tested at three different pH values (4.0, 6.0, and 7.0) with different salt concentrations (from 0 [10,14,32]. Experimental studies indicate that LLPS of the TDP-43 LCD is pH-dependent to 300 mM NaCl). The presence of liquid droplets was observed in all salt conditions at pH 7.0 [33].progressively and LLPS was tested at three dissipated different as the pH values pH decreased, (4.0, higher requiring 6.0, and 7.0) with different concentrations salt of salt to concentrations observe (from LLPS. These 0 to reveal results 300 mM thatNaCl). The presence physiological of liquid conditions droplets was are conductive observed of LLPS for the in all salt conditions at pH 7.0 and progressively dissipated as the pH decreased, TDP-43 LCD. When analyzing the resulting solubility curve predicted by SolupHred for this requiring higher concentrations LLPS-DR in this pH range of (Figure salt to observe LLPS. These 6A), maximum results solubility reveal that is achieved physiological at pH 7.0. The LCDs con- ditions are conductive of LLPS for the TDP-43 LCD. When analyzing of the hnRNPA1 [34] or the U1-70K proteins [35] have been shown to form liquid droplets the resulting solu- bility at pH curve predicted 7.0, where by SolupHred the solubility score isfor thistoLLPS-DR close in this(Figure the maximum pH range (Figure 6B,C). 6A), max- Importantly, imum solubility dysfunctional LLPS is of achieved at pH 7.0. these proteins leadThe LCDs ofaggregation to aberrant the hnRNPA1 and [34] or the U1-70K pro- neurodegeneration. teins [35] have been shown to form liquid droplets at pH 7.0, where the solubility score is close to the maximum (Figure 6B,C). Importantly, dysfunctional LLPS of these proteins lead to aberrant aggregation and neurodegeneration. Biomolecules Biomolecules 2022, 2022, 12,12, 974x FOR PEER REVIEW 8 7ofof 13 12 Figure6.6.Predicted Figure Predictedsolubility solubilitycurves curvesatatdifferent differentpHs pHsfor for the the LCDs LCDs of of TDP-43 (A), hnRNPA1 TDP-43 (A), hnRNPA1 (B) (B) and and U1-70K U1-70K(C).(C).InInall allthree threecases, cases,homotypic homotypicLLPS LLPSwas wasobserved observed atat pH pH 7.0, 7.0, a condition in which which they they displayrelatively display relativelyhighhighsolubility solubilityscores. scores. Altogether, Altogether,for these for proteins, these LLPS proteins, LLPSoccurs occurs in in a high a highsolubility range solubility that range is compatible that is compat- with ible with physiological conditions. This does not imply that LLPS necessarily occurs physiological conditions. This does not imply that LLPS necessarily more occurs more efficiently efficientlyatata apH pHininwhich whichprotein proteinsolubility solubilityisisstrictly strictly at at its its maximum, maximum, but rather that that to to form formbiomolecular biomolecular condensates, proteins condensates, should proteins should be inbeain solubility regime a solubility that enables regime them that enables tothem diffuse and interact to diffuse homotypically and interact and, and, homotypically if needed, withwith if needed, theirtheir partners. Conditions partners. Conditionsof very low low of very solubility can can solubility alsoalso promote LLPS, promote but but LLPS, theythey are are often associated often with associated withaggregation aggrega- and tionpathogenicity. and pathogenicity. 3.6. 3.6.Mutations MutationsininLLPS LLPSFormation Formationand andDisease Disease Mutations MutationsininLLPS-DRs LLPS-DRscan canhinder hinderthe thecapacity capacity of of proteins proteins bearing bearing these these sequences sequences totophase-separate. phase-separate.This Thishashas been been studied studied in in vitro, vitro, analyzing analyzing the formation of the formation of liquid liquid con- con- densates using different variants of a given sequence. Unveiling densates using different variants of a given sequence. Unveiling how these mutations af- how these mutations affect the solubility fect the solubility pattern pattern of of these these regions regions is is vital vital to to understanding understanding their their connection connection with with LLPS deregulation. LLPS deregulation. ItItisiswell wellestablished establishedthat thattyrosines tyrosinesare are important important residues residues in in LLPS LLPS since since their theirinvolve- involve- ment in π-π [36] or cation-π [11,37] interactions are key for the ment in π-π [36] or cation-π [11,37] interactions are key for the weak multivalent contact weak multivalent contact that thatsustain sustainliquidliquiddroplets. droplets.InInthis thisway, way, anan increasing increasing number number ofofTyr-to-Ser Tyr-to-Ser substitutions substitutions in the in LCD the LCD of FUS has has of FUS beenbeen linked withwith linked a reduction a reduction in the formation in the formationof droplets of dropletsat pH at pH7.5 7.5 [9]. Therefore, [9]. Therefore, we plotted the relative we plotted fluorescence the relative fluorescenceintensity after 10 intensity min, after 10 amin, measure of LLPS a measure of inLLPS this study, against the solubility scores obtained by SolupHred in this study, against the solubility scores obtained by SolupHred at pH 7.5 for at pH 7.5 for all Y→ allS variants. Y→S variants.We observed that thethat We observed solubility of the variants the solubility was proportional of the variants to the number was proportional to the ofnumber Ser residues introduced in the sequence (R 2 = 0.95) (Figure 7A) and thus that LLPS and of Ser residues introduced in the sequence (R = 0.95) (Figure 7A) and thus that 2 solubility LLPS andappear again solubility to be again appear correlated. to be correlated. The Themost moststudied studiedaminoaminoacid acidsubstitutions substitutions ininLLPS-DRs LLPS-DRs areare those thoseleading leading to to patholog- patho- ical aggregation. logical aggregation.For example, For example, mutations mutations of the well-conserved of the well-conserved Asp314 Asp314 at the C-terminal at the C-termi- LCD of hnRNPA1 nal LCD of hnRNPA1 to either Asn Asn to either or Val ordivert the process Val divert the processof phase separation of phase separationtowards towards the formation of pathogenic amyloid fibrils, a process that is associated the formation of pathogenic amyloid fibrils, a process that is associated with ALS onset with ALS onset [38,39]. When [38,39].studying the solubility When studying curves of the solubility theseoftwo curves thesevariants, a general two variants, decrease a general in maxi- decrease in mum solubility was observed (Figure 7B), consistent with maximum solubility was observed (Figure 7B), consistent with aggregation occurring aggregation occurring inside in- liquid droplets, side liquid resulting droplets, in their resulting rigidification in their rigidificationandand ultimately in the ultimately formation in the formation of of stable sta- amyloid ble amyloid assemblies. assemblies. Overall, tight solubility conditions are required for LLPS, and sequence modifications that alter this property, either towards more or less soluble states, impact the efficacy of the process. Biomolecules 2022, 12, x FOR PEER REVIEW 9 of 13 Biomolecules 2022, 12, 974 8 of 12 Figure 7. Predicted solubility scores for LLPS-impacting mutations in the LCD of FUS at pH 7.5 (A), Figure 7. Predicted solubility scores for LLPS-impacting mutations in the LCD of FUS at pH 7.5 (A), and hnRNPA1 (B) in the 10% maximum solubility pH interval. Both solubilizing (A) and aggregating and hnRNPA1 (B) in the 10% maximum solubility pH interval. Both solubilizing (A) and aggregat- (B) mutations may alter LLPS equilibrium. ing (B) mutations may alter LLPS equilibrium. 3.7. pH-Dependent LLPS: Optimal Condition Evaluation Overall, tight solubility conditions are required for LLPS, and sequence modifica- Finding the optimal conditions at which LLPS may occur for a given protein is not tions that trivial; alter this multiple property, variables either must towards when be assessed more or less soluble studying states, impactOne this phenomenon. the efficacy recent of the process. study tried to establish a generic approach to study LLPS under near-native conditions [40]. In this work, the authors induced the formation of liquid condensates by the LCDs of 3.7. pH-Dependent hnRNPA2, TDP-43, LLPS: Optimal NUP98, and Condition ERD14 proteins Evaluation by a pH jump from extreme pHs (3.0 or 11.0),Finding optimal conditions where the proteins did not phase at which separate, LLPS may occur forpH to physiological a given whereprotein is not they found trivial; multiple that liquid variables droplets must be assessed when studying this phenomenon. One recent were formed. studyThe triedsolubility curves to establish obtained a generic by SolupHred approach to study forLLPS theseunder LLPS-DRs (Figureconditions near-native 8) in the analyzed pH regime indicate that in the case of hnRNPA2, [40]. In this work, the authors induced the formation of liquid condensates by the LCDs TDP-43, and ERD14, the condensation-promoting of hnRNPA2, TDP-43, NUP98, pH precisely and ERD14 mapsproteins within theby range a pH jumpin whichfrom theextreme LCDs manifest pHs (3.0 their or maximum 11.0), where the solubility. proteinsThis didisnot notphase the case with NUP98, separate, for which the to physiological pH solubility where they at phys- found iological pH is high, but that liquid droplets were formed. not maximum, as this value is attained at very low pHs, where, in fact, The LLPSsolubility was not observed. An inspection of the sequence of this protein curves obtained by SolupHred for these LLPS-DRs (Figure 8) in the indicates that it is highly cationic, and the high net charge of the region at acidic analyzed pH regime indicate that in the case of hnRNPA2, TDP-43, and ERD14, the con- pH would compromise LLPS because of electrostatic densation-promoting pH repulsion. preciselyDecreasing maps within thisthe effect by moving range in whichtoward the LCDs neutral pHs manifest would allow phase separation to occur in conditions where the their maximum solubility. This is not the case with NUP98, for which the solubility atsolubility is still significantly high. An important corollary of this analysis is that LLPS does not ineludibly occur in physiological pH is high, but not maximum, as this value is attained at very low pHs, conditions where the solubility is very low. It is important to note that according to our where, in fact, LLPS was not observed. An inspection of the sequence of this protein indi- algorithm, the pI of a given protein does not necessarily coincide with the pH conditions at cates that it is highly cationic, and the high net charge of the region at acidic pH would which the minimum solubility score is found, or on the other way around, that the pHs compromise LLPS because of electrostatic repulsion. Decreasing this effect by moving to- more distant from the pI are not always those at which the protein would exhibit maximum ward neutral pHs would allow phase separation to occur in conditions where the solubil- solubility. This results from considering simultaneously the impact of the pH in sequence ity is still significantly hydrophobicity and net high. An important charge, not only this corollary of this last factor. analysis Indeed, for theis that LLPS does not above-mentioned ineludibly occur in conditions where the solubility is proteins, the LLPS-DR pI is a poor predictor of conditions eliciting LLPS.very low. It is important to note that according to our algorithm, The discussed thedopInot results still of allow a given forprotein does not necessarily the standardization of a method coincide with to predict the pH conditions at which the minimum solubility score is found, or enhance LLPS just considering the solution pH independently. However, obtaining a pH- or on the other way around, that the pHs more distant from the pI are not always dependent solubility profile allows for an evaluation of the physicochemical parameters those at which the protein would that may exhibit maximum influence solubility.ofThis the formation results liquid from considering condensates. simultaneously As a general the im- trend, naturally pact of the pH occurring in sequence LLPS-DRs hydrophobicity can do it around neutral and pH, net charge, not onlymatches which usually this lastwithfactor. Indeed, intervals for the above-mentioned of significant solubility. proteins, Moving away the LLPS-DR from these pI isregions a poor destabilizes predictor of the conditions elicit- multivalent- ing weak LLPS. interactions required for LLPS, increasing the repulsive net charge or over-stabilizes Biomolecules 2022, 12, 974 9 of 12 Biomolecules 2022, 12, x FOR PEER REVIEW 10 of 13 them, permitting LLPS, but also the evolution of liquid droplets towards the formation of pathogenic aggregates. Figure8.8.Predicted Figure Predictedsolubility solubility curves curves in in thethe pHpH jump jump interval interval forfor thethe LCDs LCDs of hnRNPA2 of hnRNPA2 (A),(A), TDP- TDP-43 43 (B), NUP98 (C) and ERD14 (D). The jump from an extreme pH (dotted red (B), NUP98 (C) and ERD14 (D). The jump from an extreme pH (dotted red line) to near neutral pH line) to near neutral pH (dotted (dotted greengreen line) induces line) induces the formation the formation of liquid of liquid droplets droplets in conditions in conditions withinwithin -or to- -or close close 10%to- 10% maximum solubilities. maximum solubilities. The discussed results still do not allow for the standardization of a method to predict 4. Discussion or enhance LLPS just The intriguing consideringofthe phenomenon solution LLPS pH independently. has been However, attracting the attention of obtaining biologists a and biophysicists in the last years. Many different observations of MLOs associatedparame- pH-dependent solubility profile allows for an evaluation of the physicochemical with a ters that wide mayofinfluence variety IDPs have thebeen formation reportedof liquid condensates. in the literature As a general [5,12,41]. However, trend, naturally large-scale occurring analyses LLPS-DRs of the properties canofdo it around these regionsneutral pH,towhich in context usually matches their surrounding with intervals environment have remained elusive, despite the importance of extrinsic factors in modulating this multivalent- of significant solubility. Moving away from these regions destabilizes the process [6]. Inweak interactions this work, we haverequired for LLPS, conducted increasing the a bioinformatics repulsive survey net charge to investigate theoreffect over-stabilizes of pH on the solubility and disorder of LLPS-DRs, allowing for better elucidation of theformation them, permitting LLPS, but also the evolution of liquid droplets towards the role of thisof pathogenic solvent aggregates. condition in the outcome of LLPS. Our results indicate that LLPS-DRs present maximum solubility around neutral 4. Discussion pH. This was initially intriguing, as previous studies suggested that APRs endorsing IDRs with lower local The intriguing solubilities of phenomenon may be necessary LLPS to drive the has been attracting the formation attention of ofbiologists protein condensates [30,31].inIndeed, and biophysicists the lastLLPS years.isMany a highly dynamic different and reversible observations of MLOsprocess that likely associated with a wide variety of IDPs have been reported in the literature [5,12,41]. However, large-scale analyses of the properties of these regions in context to their surrounding environment Biomolecules 2022, 12, 974 10 of 12 requires weaker interactions than those provided by highly aggregating patches. Therefore, LLPS-DRs need to display a significant degree of solubility to phase-separate into liquid droplets near physiological pH conditions. Intrinsic solubility is, however, an important determinant of autonomous phase separation, since psself IDRs exhibit, as a group, less soluble sequences than psother and DisProt IDRs, a property that is accompanied by a higher propensity to populate compact states. Still, for these interaction-prone sequences, LLPS occurs at pH values where they are predicted to exhibit significant solubility, a condition that would be compatible with their functioning and preclude the aggregation of these regions. This might be biologically important since psself IDRs’ solubility and disorder properties seem to have evolved to be more resistant to small pH perturbations. Our analysis provides a plausible answer to why some disordered regions phase-separate under specific pH conditions and others do not, although it is not intended to predict their behavior individually. Mutations in psself LCDs can shift the equilibrium that sustains LLPS both to a more soluble or aggregated state, which may eventually lead to disease onset. In contrast, regions in the psother dataset need a partner to compensate for their lack of intrinsic condensation propensity. In these proteins, studying the contribution of mutations is much less straightforward, as pathogenicity may stem from poor interaction with its phase-separating partner or enhanced interaction with non-intended binders. With this work, we aimed to start elucidating the role of pH in context-dependent phenomena such as LLPS and describe the general tendencies that arise from large-scale analysis. Given that many cellular processes and disease-associated pathways are sensible to cellular milieu fluctuations, we believe that by understanding the contribution of pH in both IDPs’ solubility and disorder at a large scale, we will be a step closer to understanding the impact of the environment in such processes. Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/biom12070974/s1, Table S1: LLPS-DRs pH data predictions. Author Contributions: Conceptualization: S.V., C.P.-G. and O.B. Dataset generation and data cu- ration: C.P.-G. Investigation and formal analysis: C.P.-G. and O.B. Writing: C.P.-G., O.B. and S.V. Funding acquisition and supervision: S.V. All authors have read and agreed to the published version of the manuscript. Funding: This work was funded by European Union’s Horizon 2020 research and innovation pro- gramme under GA 952334 (PhasAGE). S.V. was supported by the Spanish Ministry of Science and Innovation (PID 2019-105017RB-I00) and by ICREA Academia 2020. C.P.-G. was supported by the Secretariat of Universities and Research of the Catalan Government and the European Social Fund (2021 FI_B 00087). Data Availability Statement: Data supporting the results of this study can be accessed in the Supple- mentary Table S1 of this article. Conflicts of Interest: The authors declare no conflict of interest. Abbreviations IDPs Intrinsically disordered proteins IDRs Intrinsically disordered regions LCD Low complexity domain LLPS Liquid-liquid phase separation LLPS-DRs Liquid-liquid phase separation-disordered regions MLOs Membraneless organelles psself LLPS region that can phase-separate by itself psother LLPS region that requires a partner to phase-separate ALS Amyotrophic lateral sclerosis FTD Frontotemporal dementia APRs Aggregation-prone regions Biomolecules 2022, 12, 974 11 of 12 References 1. Brangwynne, C.P.; Eckmann, C.R.; Courson, D.S.; Rybarska, A.; Hoege, C.; Gharakhani, J.; Julicher, F.; Hyman, A.A. Germline P granules are liquid droplets that localize by controlled dissolution/condensation. Science 2009, 324, 1729–1732. [CrossRef] 2. Banani, S.F.; Lee, H.O.; Hyman, A.A.; Rosen, M.K. Biomolecular condensates: Organizers of cellular biochemistry. Nat. Rev. Mol. Cell Biol. 2017, 18, 285–298. [CrossRef] 3. Mao, Y.S.; Zhang, B.; Spector, D.L. Biogenesis and function of nuclear bodies. Trends Genet. TIG 2011, 27, 295–306. [CrossRef] 4. Decker, C.J.; Parker, R. P-bodies and stress granules: Possible roles in the control of translation and mRNA degradation. Cold Spring Harb. Perspect. Biol. 2012, 4, a012286. [CrossRef] [PubMed] 5. Alberti, S.; Hyman, A.A. Biomolecular condensates at the nexus of cellular stress, protein aggregation disease and ageing. Nat. Rev. Mol. Cell Biol. 2021, 22, 196–213. [CrossRef] 6. Alberti, S.; Gladfelter, A.; Mittag, T. Considerations and Challenges in Studying Liquid-Liquid Phase Separation and Biomolecular Condensates. Cell 2019, 176, 419–434. [CrossRef] 7. Holehouse, A.S.; Pappu, R.V. Protein polymers: Encoding phase transitions. Nat. Mater. 2015, 14, 1083–1084. [CrossRef] 8. Shin, Y.; Brangwynne, C.P. Liquid phase condensation in cell physiology and disease. Science 2017, 357, eaaf4382. [CrossRef] 9. Kato, M.; Han, T.W.; Xie, S.; Shi, K.; Du, X.; Wu, L.C.; Mirzaei, H.; Goldsmith, E.J.; Longgood, J.; Pei, J.; et al. Cell-free formation of RNA granules: Low complexity sequence domains form dynamic fibers within hydrogels. Cell 2012, 149, 753–767. [CrossRef] 10. Purice, M.D.; Taylor, J.P. Linking hnRNP Function to ALS and FTD Pathology. Front. Neurosci. 2018, 12, 326. [CrossRef] 11. Nott, T.J.; Petsalaki, E.; Farber, P.; Jervis, D.; Fussner, E.; Plochowietz, A.; Craggs, T.D.; Bazett-Jones, D.P.; Pawson, T.; Forman-Kay, J.D.; et al. Phase transition of a disordered nuage protein generates environmentally responsive membraneless organelles. Mol. Cell 2015, 57, 936–947. [CrossRef] [PubMed] 12. Ryan, V.H.; Fawzi, N.L. Physiological, Pathological, and Targetable Membraneless Organelles in Neurons. Trends Neurosci. 2019, 42, 693–708. [CrossRef] [PubMed] 13. Das, S.; Lin, Y.H.; Vernon, R.M.; Forman-Kay, J.D.; Chan, H.S. Comparative roles of charge, pi, and hydrophobic interactions in sequence-dependent phase separation of intrinsically disordered proteins. Proc. Natl. Acad. Sci. USA 2020, 117, 28795–28805. [CrossRef] [PubMed] 14. Borroni, B.; Bonvicini, C.; Alberici, A.; Buratti, E.; Agosti, C.; Archetti, S.; Papetti, A.; Stuani, C.; Di Luca, M.; Gennarelli, M.; et al. Mutation within TARDBP leads to frontotemporal dementia without motor neuron disease. Hum. Mutat. 2009, 30, E974–E983. [CrossRef] [PubMed] 15. Vance, C.; Rogelj, B.; Hortobagyi, T.; De Vos, K.J.; Nishimura, A.L.; Sreedharan, J.; Hu, X.; Smith, B.; Ruddy, D.; Wright, P.; et al. Mutations in FUS, an RNA processing protein, cause familial amyotrophic lateral sclerosis type 6. Science 2009, 323, 1208–1211. [CrossRef] 16. Batlle, C.; Ventura, S. Prion-like domain disease-causing mutations and misregulation of alternative splicing relevance in limb-girdle muscular dystrophy (LGMD) 1G. Neural Regen. Res. 2020, 15, 2239–2240. [CrossRef] 17. Batlle, C.; Yang, P.; Coughlin, M.; Messing, J.; Pesarrodona, M.; Szulc, E.; Salvatella, X.; Kim, H.J.; Taylor, J.P.; Ventura, S. hnRNPDL Phase Separation Is Regulated by Alternative Splicing and Disease-Causing Mutations Accelerate Its Aggregation. Cell Rep. 2020, 30, 1117–1128. [CrossRef] 18. Uversky, V.N. Intrinsically disordered proteins and their environment: Effects of strong denaturants, temperature, pH, counter ions, membranes, binding partners, osmolytes, and macromolecular crowding. Protein J. 2009, 28, 305–325. [CrossRef] 19. Farahi, N.; Lazar, T.; Wodak, S.J.; Tompa, P.; Pancsa, R. Integration of Data from Liquid-Liquid Phase Separation Databases Highlights Concentration and Dosage Sensitivity of LLPS Drivers. Int. J. Mol. Sci. 2021, 22, 3017. [CrossRef] 20. Santos, J.; Iglesias, V.; Santos-Suarez, J.; Mangiagalli, M.; Brocca, S.; Pallares, I.; Ventura, S. pH-Dependent Aggregation in Intrinsically Disordered Proteins Is Determined by Charge and Lipophilicity. Cells 2020, 9, 145. [CrossRef] 21. Santos, J.; Iglesias, V.; Pintado, C.; Santos-Suarez, J.; Ventura, S. DispHred: A Server to Predict pH-Dependent Order-Disorder Transitions in Intrinsically Disordered Proteins. Int. J. Mol. Sci. 2020, 21, 5814. [CrossRef] 22. Pintado, C.; Santos, J.; Iglesias, V.; Ventura, S. SolupHred: A server to predict the pH-dependent aggregation of intrinsically disordered proteins. Bioinformatics 2021, 37, 1602–1603. [CrossRef] 23. Pintado-Grima, C.; Iglesias, V.; Santos, J.; Uversky, V.N.; Ventura, S. DispHScan: A Multi-Sequence Web Tool for Predicting Protein Disorder as a Function of pH. Biomolecules 2021, 11, 1596. [CrossRef] 24. Jacoby, G.; Segal Asher, M.; Ehm, T.; Abutbul Ionita, I.; Shinar, H.; Azoulay-Ginsburg, S.; Zemach, I.; Koren, G.; Danino, D.; Kozlov, M.M.; et al. Order from Disorder with Intrinsically Disordered Peptide Amphiphiles. J. Am. Chem. Soc. 2021, 143, 11879–11888. [CrossRef] 25. Uversky, V.N.; Gillespie, J.R.; Millett, I.S.; Khodyakova, A.V.; Vasiliev, A.M.; Chernovskaya, T.V.; Vasilenko, R.N.; Kozlovskaya, G.D.; Dolgikh, D.A.; Fink, A.L.; et al. Natively unfolded human prothymosin alpha adopts partially folded collapsed conformation at acidic pH. Biochemistry 1999, 38, 15009–15016. [CrossRef] 26. Wu, K.P.; Weinstock, D.S.; Narayanan, C.; Levy, R.M.; Baum, J. Structural reorganization of alpha-synuclein at low pH observed by NMR and REMD simulations. J. Mol. Biol. 2009, 391, 784–796. [CrossRef] [PubMed] 27. You, K.; Huang, Q.; Yu, C.; Shen, B.; Sevilla, C.; Shi, M.; Hermjakob, H.; Chen, Y.; Li, T. PhaSepDB: A database of liquid-liquid phase separation related proteins. Nucleic Acids Res. 2020, 48, D354–D359. [CrossRef] Biomolecules 2022, 12, 974 12 of 12 28. Meszaros, B.; Erdos, G.; Dosztanyi, Z. IUPred2A: Context-dependent prediction of protein disorder as a function of redox state and protein binding. Nucleic Acids Res. 2018, 46, W329–W337. [CrossRef] 29. Quaglia, F.; Meszaros, B.; Salladini, E.; Hatos, A.; Pancsa, R.; Chemes, L.B.; Pajkos, M.; Lazar, T.; Pena-Diaz, S.; Santos, J.; et al. DisProt in 2022: Improved quality and accessibility of protein intrinsic disorder annotation. Nucleic Acids Res. 2021, 50, D480–D487. [CrossRef] 30. Iglesias, V.; Santos, J.; Santos-Suarez, J.; Pintado-Grima, C.; Ventura, S. SGnn: A Web Server for the Prediction of Prion-Like Domains Recruitment to Stress Granules Upon Heat Stress. Front. Mol. Biosci. 2021, 8, 718301. [CrossRef] 31. Wallace, E.W.; Kear-Scott, J.L.; Pilipenko, E.V.; Schwartz, M.H.; Laskowski, P.R.; Rojek, A.E.; Katanski, C.D.; Riback, J.A.; Dion, M.F.; Franks, A.M.; et al. Reversible, Specific, Active Aggregates of Endogenous Proteins Assemble upon Heat Stress. Cell 2015, 162, 1286–1298. [CrossRef] [PubMed] 32. Neumann, M.; Sampathu, D.M.; Kwong, L.K.; Truax, A.C.; Micsenyi, M.C.; Chou, T.T.; Bruce, J.; Schuck, T.; Grossman, M.; Clark, C.M.; et al. Ubiquitinated TDP-43 in frontotemporal lobar degeneration and amyotrophic lateral sclerosis. Science 2006, 314, 130–133. [CrossRef] [PubMed] 33. Babinchak, W.M.; Haider, R.; Dumm, B.K.; Sarkar, P.; Surewicz, K.; Choi, J.K.; Surewicz, W.K. The role of liquid-liquid phase separation in aggregation of the TDP-43 low-complexity domain. J. Biol. Chem. 2019, 294, 6306–6317. [CrossRef] 34. Tsoi, P.S.; Quan, M.D.; Choi, K.J.; Dao, K.M.; Ferreon, J.C.; Ferreon, A.C.M. Electrostatic modulation of hnRNPA1 low-complexity domain liquid-liquid phase separation and aggregation. Protein Sci. Publ. Protein Soc. 2021, 30, 1408–1417. [CrossRef] [PubMed] 35. Xue, S.; Gong, R.; He, F.; Li, Y.; Wang, Y.; Tan, T.; Luo, S.Z. Low-complexity domain of U1-70K modulates phase separation and aggregation through distinctive basic-acidic motifs. Sci. Adv. 2019, 5, eaax5349. [CrossRef] 36. Vernon, R.M.; Chong, P.A.; Tsang, B.; Kim, T.H.; Bah, A.; Farber, P.; Lin, H.; Forman-Kay, J.D. Pi-Pi contacts are an overlooked protein feature relevant to phase separation. eLife 2018, 7, e31486. [CrossRef] 37. Van Lindt, J.; Lazar, T.; Pakravan, D.; Demulder, M.; Meszaros, A.; Van Den Bosch, L.; Maes, D.; Tompa, P. F/YGG-motif is an intrinsically disordered nucleic-acid binding motif. RNA Biol. 2022, 19, 622–635. [CrossRef] 38. Iglesias, V.; Conchillo-Sole, O.; Batlle, C.; Ventura, S. AMYCO: Evaluation of mutational impact on prion-like proteins aggregation propensity. BMC Bioinform. 2019, 20, 24. [CrossRef] 39. Kim, H.J.; Kim, N.C.; Wang, Y.D.; Scarborough, E.A.; Moore, J.; Diaz, Z.; MacLea, K.S.; Freibaum, B.; Li, S.; Molliex, A.; et al. Mutations in prion-like domains in hnRNPA2B1 and hnRNPA1 cause multisystem proteinopathy and ALS. Nature 2013, 495, 467–473. [CrossRef] 40. Van Lindt, J.; Bratek-Skicki, A.; Nguyen, P.N.; Pakravan, D.; Duran-Armenta, L.F.; Tantos, A.; Pancsa, R.; Van Den Bosch, L.; Maes, D.; Tompa, P. A generic approach to study the kinetics of liquid-liquid phase separation under near-native conditions. Commun. Biol. 2021, 4, 77. [CrossRef] 41. Taylor, J.P.; Brown, R.H., Jr.; Cleveland, D.W. Decoding ALS: From genes to mechanism. Nature 2016, 539, 197–206. [CrossRef] [PubMed]