p53 Transactivation Domain Mediates Binding and Phase Separation with Poly-PR/GR

The most common genetic cause of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) is the presence of poly-PR/GR dipeptide repeats, which are encoded by the chromosome 9 open reading frame 72 (C9orf72) gene. Recently, it was shown that poly-PR/GR alters chromatin accessibility, which results in the stabilization and enhancement of transcriptional activity of the tumor suppressor p53 in several neurodegenerative disease models. A reduction in p53 protein levels protects against poly-PR and partially against poly-GR neurotoxicity in cells. Moreover, in model organisms, a reduction of p53 protein levels protects against neurotoxicity of poly-PR. Here, we aimed to study the detailed molecular mechanisms of how p53 contributes to poly-PR/GR-mediated neurodegeneration. Using a combination of biophysical techniques such as nuclear magnetic resonance (NMR) spectroscopy, fluorescence polarization, turbidity assays, and differential interference contrast (DIC) microscopy, we found that p53 physically interacts with poly-PR/GR and triggers liquid–liquid phase separation of p53. We identified the p53 transactivation domain 2 (TAD2) as the main binding site for PR25/GR25 and showed that binding of poly-PR/GR to p53 is mediated by a network of electrostatic and/or hydrophobic interactions. Our findings might help to understand the mechanistic role of p53 in poly-PR/GR-associated neurodegeneration.


Introduction
Amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) are neurodegenerative diseases with many common neuropathological, genetic, and clinical features [1,2]. The most frequent genetic cause of ALS and FTD is a GGGGCC hexanucleotide repeat expansion in the first non-coding region of the chromosome 9 open reading frame 72 gene (C9orf72) [3,4]. This mutation results in loss of function of the C9orf72 by inhibition of transcription [3,5,6]; generation of RNA foci [7], which sequester critical RNA-binding proteins [8,9] and/or interfere with essential cellular processes [10]; and repeat-associated non-AUG (RAN) translation of dipeptide repeat proteins (DPRs) [11][12][13][14]. Five DPRs can be translated, including glycine-alanine (GA) and glycine-arginine (GR) repeats from the sense strand, proline-alanine (PA) and proline-arginine (PR) repeats from the anti-sense strand, and glycine-proline (GP) from both strands [12,13]. Accumulation of these DPRs in the brains of patients with C9orf72 mutants has been shown [15]. Among them, poly-PR and poly-GR are the most toxic ones. The addition of recombinant PR or GR polymers to HeLa or human astrocytes cells and expressing 50 repeats of GR or PR in Drosophila result in RNA processing alterations and early lethality, respectively [15][16][17][18][19][20][21]. Moreover, expression of poly(GR) 100 or poly(PR) 28 causes early-onset and severe neurodegeneration with motor dysfunction and memory impairment in mice [14,22,23]. Toxicity 2. Results 2.1. Poly(PR) 25 /Poly(GR) 25 Interact Directly with p53 As previous studies have shown a link between DPRs, p53 activation, and stabilization [40], we wanted to assess whether the model DPRs poly(PR) 25 and poly(GR) 25 directly interact with p53 in vitro using the purified proteins and NMR spectroscopy.
To this end, we purified recombinant isotope-labelled p53 1−312 (residues 1 to 312), which includes the N-terminal TAD; the poly-proline region; the DBD; and performed titrations with either synthetic poly(PR) 25 or poly(GR) 25 dipeptide repeats named PR25 and GR25, respectively. We excluded the C-terminal tetramerization domain that would prevent obtaining high-quality NMR data ( Figure 1A). The addition of one stoichiometric equivalent of either unlabeled PR25 or GR25 caused disappearance and chemical shift perturbations (CSPs) of the p53 1−312 1 H-15 N HSQC cross-peaks ( Figure 1B,C), which indicates direct binding of PR25/GR25 to the corresponding residues of p53 1−312 in the fast to intermediate NMR exchange time scale. Inspection of the peaks showed that, mostly, the intense peaks belonging to the disordered N-terminal region seemed to be affected by PR25/GR25.  In order to localize the binding site of PR25/GR25 on p53, we carried out a divideand-conquer approach and purified the isolated N-terminal disordered region containing the transactivation domains 1 and 2 as well as the poly-proline region (p53 1−94 , residues 1 to 94) and a construct harboring the p53 DBD (p53 DBD ; residues 95 to 312). 1 H-15 N HSQC NMR spectra of isotope-labelled p53 1−94 and p53 DBD were recorded in the absence and presence of one stoichiometric equivalent of PR25/GR25. The addition of PR25/GR25 to the p53 DBD did not change the 1 H-15 N HSQC NMR spectra of p53 DBD ( Figure 1D,E), which indicates that this p53 domain is not involved in the interaction with PR25/GR25. In contrast, and in line with the observations for p53 1−312 , p53 1−94 1 H-15 N HSQC crosspeaks showed CSPs in the presence of PR25/GR25 ( Figure 1F,G), which indicates that the interaction between p53 and PR25/GR25 is mediated by residues localized within this N-terminal intrinsically disordered region. Inspection of the residues affected by the addition of PR25/GR25 showed that the strongest CSPs cluster within p53 TAD2 , for both PR25 and GR25 ( Figure 1F,G). In summary, our NMR analysis shows that PR25/GR25 directly bind to p53, involving residues clustering in p53 TAD2 .

p53 Transactivation Domain 2 Is the Main PR25/GR25 Binding Site
To specify the role of p53 TAD1 and p53 TAD2 in PR25/GR25 binding, isotope-labelled p53 TAD1 and p53 TAD2 were purified separately, and 1 H-15 N HSQC NMR experiments were performed in presence or absence of PR25/GR25. p53 TAD1 1 H-15 N HSQC cross-peaks showed no CSPs upon the addition of one stoichiometric equivalent of PR25/GR25 ( Figure S1A,B). In contrast, and in line with the results obtained for the longer constructs, p53 TAD2 1 H-15 N HSQC cross-peaks showed CSPs upon PR25/GR25 addition (Figure 2A,B). This shows that TAD2, but not TAD1, is sufficient to mediate PR25/GR25 binding. Inspection of the TAD2 residues affected upon addition of PR25/GR25 showed that the C-terminal end (51-55) is mainly involved. Addition of PR25 or GR25 yielded similar CSPs. To specify the role of p53 TAD1 and p53 TAD2 in PR25/GR25 binding, isotope-labelled p53 TAD1 and p53 TAD2 were purified separately, and 1 H-15 N HSQC NMR experiments were performed in presence or absence of PR25/GR25. p53 TAD1 1 H-15 N HSQC cross-peaks showed no CSPs upon the addition of one stoichiometric equivalent of PR25/GR25 ( Figure S1A,B). In contrast, and in line with the results obtained for the longer constructs, p53 TAD2 1 H-15 N HSQC crosspeaks showed CSPs upon PR25/GR25 addition (Figure 2A,B). This shows that TAD2, but not TAD1, is sufficient to mediate PR25/GR25 binding. Inspection of the TAD2 residues affected upon addition of PR25/GR25 showed that the C-terminal end (51-55) is mainly involved. Addition of PR25 or GR25 yielded similar CSPs.
To complement the NMR data, we performed FP experiments in order to determine binding affinities of p53 1−94 and p53 TAD2 for fluorescein isothiocyanate (FITC)-labeled PR25 and GR25 peptides ( Figure 2C,D). We found that p53 1−94 has similar low micro molar range affinity for PR25 and GR25 with equilibrium dissociation constants (Kds) of 1.84 ± 0.72 and 0.43 ± 0.17 µM for PR25 and GR25, respectively. Similarly, p53 TAD2 has low micro molar range affinity for both PR25 and GR25 with Kds of 3.64 ± 1.16 and 2.85 ± 0.51 µM for PR25 and GR25 binding, respectively. Compared with p53 TAD2 , p53 1−94 has approximately twofold/sixfold higher affinities for PR25 and GR25 binding, respectively ( Figure 2C,D). Even though p53 TAD1 is insufficient to bind PR25 and GR25 ( Figure S1A,B), it enhances binding to PR25/GR25 in the p53 1−94 construct ( Figure 2C,D). Overall, we observed that the binding affinities of PR25/GR25 for p53 are similar comparing the results obtained for p53 1−94 and p53 TAD2 , with slightly enhanced affinity of GR25. Given the similar CSPs observed for both GR25 and PR25, and the absence of binding to isolated p53 TAD1 , we conclude that p53 TAD2 is sufficient for PR25/GR25 binding.  To complement the NMR data, we performed FP experiments in order to determine binding affinities of p53 1−94 and p53 TAD2 for fluorescein isothiocyanate (FITC)-labeled PR25 and GR25 peptides ( Figure 2C,D). We found that p53 1−94 has similar low micro molar range affinity for PR25 and GR25 with equilibrium dissociation constants (K d s) of 1.84 ± 0.72 and 0.43 ± 0.17 µM for PR25 and GR25, respectively. Similarly, p53 TAD2 has low micro molar range affinity for both PR25 and GR25 with K d s of 3.64 ± 1.16 and 2.85 ± 0.51 µM for PR25 and GR25 binding, respectively. Compared with p53 TAD2 , p53 1−94 has approximately twofold/sixfold higher affinities for PR25 and GR25 binding, respectively ( Figure 2C,D). Even though p53 TAD1 is insufficient to bind PR25 and GR25 ( Figure S1A,B), it enhances binding to PR25/GR25 in the p53 1−94 construct ( Figure 2C,D). Overall, we observed that the binding affinities of PR25/GR25 for p53 are similar comparing the results obtained for p53 1−94 and p53 TAD2 , with slightly enhanced affinity of GR25. Given the similar CSPs observed for both GR25 and PR25, and the absence of binding to isolated p53 TAD1 , we conclude that p53 TAD2 is sufficient for PR25/GR25 binding.

Binding of PR25/GR25 to p53 1−94 Increases Its Rigidity and α-Helical Secondary Structure Propensity
The TAD of p53 is intrinsically disordered, but has been shown previously to adopt secondary α-helical structure upon binding to different protein partners [41][42][43]. In order to obtain information on p53 1−94 flexibility in the presence of DPRs, we recorded 15 N{ 1 H} heteronuclear NOE experiments of isotope-labeled p53 1−94 in the presence and absence of PR25/GR25. 15 N{ 1 H} heteronuclear NOEs report the motion of individual N-H bond vectors and provide information about the rigidity of the protein backbone, in which positive values correspond to rigidity and negative values represent flexible disordered regions. We observed that HetNOEs increased in the presence of PR25/GR25, in the regions comprising TAD1 (from 17 to 37) and TAD2 (from 38 to 55), which indicates that binding of PR25/GR25 enhances the rigidity of p53 TAD1 and TAD2 ( Figure 3A,B). However, there are few residues that exhibit higher flexibility when bound to PR25, namely Gln38 and 51Glu. In complex with GR25, residues 21Asp, 24Lys, and 56Glu become more flexible ( Figure 3A,B). Binding of PR25/GR25 to p53 1−94 increases its rigidity and α-helical secondary structure propensity via electrostatic and/or hydrophobic interactions. (A,B) 15 N{ 1 H} heteronuclear NOEs are plotted for p53 1−94 residues in the absence (black) or presence of GR25 (cyan in A) or PR25 (orange in B), respectively. Error bars for heteronuclear NOE values were derived from error propagation calculation using standard deviation of 10 arbitrarily chosen noise peaks in saturated and unsaturated spectra; (C,D) Plot of difference between the secondary 13 C chemical shifts of the Cα (noted as ΔδCα) and the Cβ (noted as ΔδCβ) nuclei of p53 1−94 in the absence (black) or presence of GR25 (cyan in C) and PR25 (orange in D), respectively. Secondary chemical shifts were obtained by subtracting the random coil chemical shifts (predicted by ncIDP) from observed chemical shifts. Differences between the secondary chemical shift deviations (ΔδCα -ΔδCβ) were plotted against the amino acid residue numbers, taking into account next neighbor effects; (E,F) 1 H-13 C HSQC spectra of 13 C-labeled p53 1−94 in the absence (black) or presence of either GR25 (cyan in E) or PR25 (orange in F). Residues with NOE cross peaks are indicated by black squares; (G) Slices from 1 H 1 H planes of a 3D 13 C/ 15 N filtered, 13 C-edited NOESY-HSQC spectrum recorded on 13 C, 15 N-labelled p53 1−94 in the presence of 1.33 stoichiometric equivalents of unlabeled GR25. NOESY cross peaks are shown in cyan. Representative two-dimensional planes are shown for different 13 C chemical shifts and  B), respectively. Error bars for heteronuclear NOE values were derived from error propagation calculation using standard deviation of 10 arbitrarily chosen noise peaks in saturated and unsaturated spectra; (C,D) Plot of difference between the secondary 13 C chemical shifts of the Cα (noted as ∆δCα) and the Cβ (noted as ∆δCβ) nuclei of p53 1−94 in the absence (black) or presence of GR25 (cyan in C) and PR25 (orange in D), respectively. Secondary chemical shifts were obtained by subtracting the random coil chemical shifts (predicted by ncIDP) from observed chemical shifts. Differences between the secondary chemical shift deviations (∆δCα -∆δCβ) were plotted against the amino acid residue numbers, taking into account next neighbor effects; (E,F) 1 H-13 C HSQC spectra of 13 C-labeled p53 1−94 in the absence (black) or presence of either GR25 (cyan in E) or PR25 (orange in F). Residues with NOE cross peaks are indicated by black squares; (G) Slices from 1 H 1 H planes of a 3D 13 C/ 15 N filtered, 13 C-edited NOESY-HSQC spectrum recorded on 13 C, 15 N-labelled p53 1−94 in the presence of 1.33 stoichiometric equivalents of unlabeled GR25. NOESY cross peaks are shown in cyan. Representative two-dimensional planes are shown for different 13 C chemical shifts and illustrate intermolecular NOEs involving Glu-Hγ*Cγ, Glu/Gln-Hβ*Cβ, Pro-Hγ*Cγ, Leu-Hδ*Cδ, Ala-Hβ*Cβ, and Met-Hε*Cε (from left to right) and their corresponding signals are indicated by dashed lines; cross peaks of these residues to 1 H chemical shifts characteristic for glycine and arginine residues are labeled. Black and cyan peaks correspond to positive and negative peaks, respectively; (H) Slices from 1 H 1 H planes of a 3D 13 C/ 15 N filtered, 13 C-edited NOESY-HSQC spectrum recorded on 13 C, 15 N-labelled p53 1−94 in the presence of 1.33 stoichiometric equivalents of unlabeled PR25. NOESY cross peaks are shown in orange. Representative two-dimensional planes are shown at different 13 C chemical shifts and illustrate intermolecular NOEs involving Leu-Hδ*Cδ, Thr-Hγ*Cγ, and Met-Hε*Cε (from left to right) and their corresponding signals are indicated by dashed lines. Cross peaks of these residues to 1 H chemical shifts characteristic for glycine and arginine residues are labeled. Black and orange peaks correspond to positive and negative peaks, respectively.
To test if PR25/GR25 binding affects the secondary structure of p53 1−94 , CBCA(CO)NH NMR experiments of isotope-labeled p53 1−94 were recorded in the absence and presence of PR25/GR25. From these experiments, we obtained the 13 C chemical shifts of the alpha (α) and beta (β) carbon atoms of the p53 residues, which can be used to obtain information on the propensity of the corresponding residues to adopt either a α-helical or β-stranded secondary structure. The p53 TAD1 region (residues 17-28) harbors a stretch of positive ∆δ(Cα-Cβ) values in the presence and absence of PR25/GR25, showing that this p53 domain is prone to form α-helical secondary structure on its own. This propensity remains unaltered upon addition of PR25/GR25, which is in line with the absence of binding to this region ( Figure 3C,D). In contrast, we observed a slight increase in the ∆δ(Cα-Cβ) values in the TAD2 region, especially for residues 48Asp, 49Asp, 50Ile, and 51Glu in the presence of PR25, as well as residue 48Asp in the presence of GR25, suggesting that this p53 domain harbors increased α-helical propensity upon DPR binding ( Figure 3C,D). It is worth noting that binding of p53 to GR25 induces extensive line broadening; therefore, several of the 13 C chemical shifts of the TAD2 could not be assigned. In summary, the interaction of PR25/GR25 with the p53 N-terminal disordered region increases the overall rigidity of the region encompassing the TAD1 and TAD2 sub-domains. While the TAD1 is already α-helical in the absence of DPRs and helical propensity is unchanged upon DPR binding, the TAD2 becomes α-helical only upon binding.

Interaction between PR25/GR25 and p53 1−94 Is Mediated by Electrostatic and/or Hydrophobic Interactions
To obtain insight into the molecular details of how p53 binds PR25 or GR25, 13 C and 15 N filtered and 13 C-edited 3D nuclear Overhauser effect spectroscopy (NOESY) experiments were recorded with 13 C-15 N isotope-labelled p53 1−94 in the presence of either PR25 or GR25. These types of experiments provide information about intermolecular interactions by eliminating NOEs from intramolecular interactions in each of the components of the complex [44]. We observed NOE cross peaks from each carbon-attached 1 H position involving ambiguous p53 1−94 Glu-Cγ, Glu/Gln-Cβ, Pro-Cγ, Leu-Cδ, Ala-Cβ, and Met-Cε for the GR25-bound complex, indicating that these p53 residues are in close proximity with GR25 protons (Figure 3E,G). In the same way, we observed NOE cross from each carbon-attached 1 H position involving p53 1−94 Leu-Cδ, Thr-Cγ, and Met-Cε for the PR25-bound complex ( Figure 3F,H). Although the poor chemical shift dispersion in the 1 H, 13 C HSQC spectrum of p53 1−94 did not allow for residue-specific assignment of NOE cross peaks, we can, based on the CSP results, tentatively assign these residues to be in close proximity with residues clustered in p53 TAD2 such as Glu51/Glu56, Gln38/Gln52, Pro47, Leu43/Leu45, Ala39, Met40/Met44, and Thr55. Inspection of the NOESY cross peaks revealed that these p53 residues showed cross peaks to 1 H chemical shifts characteristic for arginine, proline, and glycine residues. The nature of these interactions was mostly hydrophobic, as indicated by cross peaks between p53 leucines and methionines to either PR25 arginines or PR25 prolines. In a similar manner, the interaction between p53 and GR25 is mostly mediated via electrostatic interactions from p53 glutamates and arginines from GR25, but also hydrophobic contact between (i) p53 methionines and prolines to arginines of GR25; (ii) p53 leucines and prolines to either glycines or arginines of GR25; and (iii) p53 alanines to glycines of GR25.
In summary, we demonstrated that binding and interaction of p53 with poly-PR/GR are mediated mainly by p53 TAD2 and involve a network of electrostatic and/or hydrophobic interactions. It has been previously shown that poly-PR/GR can regulate phase separation of proteins with low complexity domains (LCDs) [45]. Furthermore, other studies have shown that p53 can phase separate in vitro and associate with MLOs in cells [46,47]. To understand whether PR25/GR25 affect phase separation of p53, we performed turbidity assays and DIC microscopy with different p53 constructs harboring the (i) N-terminal disordered region and the DBD (p53 1−312 ), (ii) the N-terminal disordered region (p53 1−94 ), (iii) TAD2 and (iv) DBD. Titration of increasing concentrations of PR25/GR25 with fixed (50 µM) concentrations of p53 1−312 resulted in increasing turbidity at low PR25/GR25 concentrations ( Figure 4A,B). In contrast, higher concentrations of PR25/GR25 (especially PR25) suppressed phase separation of p53 1−312 ( Figure 4A,B). Phase separation of fixed (50 µM) concentrations of p53 1−94 reached a maximum at 150 µM GR25 and 100 µM PR25, respectively. Higher concentrations of PR25/GR25 suppressed LLPS ( Figure 4A,B). Among the p53 constructs tested, p53 TAD2 shows a slight propensity to phase separate only in the presence of 40 to 50 µM GR25, whereas its phase separation was not induced by PR25 (Figure 4A,B). As expected, addition of increasing concentration of PR25/GR25 with p53 DBD did not change the phase separation propensity of p53 DBD because there is no direct binding between PR25/GR25 and p53 DBD .

1, 22, 11431 9 of 18
As expected, addition of increasing concentration of PR25/GR25 with p53 DBD did not change the phase separation propensity of p53 DBD because there is no direct binding between PR25/GR25 and p53 DBD . We further validated our findings by monitoring the phase separation of p53 1−312 , p53 1−94 , and p53 TAD2 by DIC microscope. To record the formation and fusion of droplets, different concentrations of PR25/GR25 were used for different constructs in order to observe the light phase and avoid the dense phase [48]. Droplet formation and their fusions were observed and recorded over time. In line with the turbidity assays, we observed droplet formation of p53 constructs except p53 DBD upon addition of PR25/GR25 ( Figure  4C-E,G). There was no droplet formation for p53 TAD2 in the presence of PR25 ( Figure 4F). In summary, our turbidity and DIC results showed that poly-PR/GR regulate p53 LLPS via direct interactions.   We further validated our findings by monitoring the phase separation of p53 1−312 , p53 1−94 , and p53 TAD2 by DIC microscope. To record the formation and fusion of droplets, different concentrations of PR25/GR25 were used for different constructs in order to observe the light phase and avoid the dense phase [48]. Droplet formation and their fusions were observed and recorded over time. In line with the turbidity assays, we observed droplet formation of p53 constructs except p53 DBD upon addition of PR25/GR25 (Figure 4C-E,G). There was no droplet formation for p53 TAD2 in the presence of PR25 ( Figure 4F). In summary, our turbidity and DIC results showed that poly-PR/GR regulate p53 LLPS via direct interactions.

Discussion
In this work, we found that p53 directly interacts with poly-PR/GR. We showed that the interaction between poly-PR/GR and p53 is mediated via the N-terminal TAD of p53 harboring two sub-domains, namely TAD1 (residues 1-39) and TAD2 (residues 40-61) [49] ( Figure 1B-G, Figure 2A,B and Figure S1A,B). We observed that p53 TAD2 is sufficient to bind to PR25/GR25 and that the presence of other regions within the N-terminal TAD slightly enhanced the affinity towards PR25/GR25 ( Figure 2C,D). The intrinsically disordered N-terminal p53 TAD is involved in several protein-protein interactions, and p53 TAD2 alone has been reported to bind to CBP/p300 (CREB (cAMP-response-element-binding protein)-binding protein), high mobility group B1 (HMGB1), replication protein A, transcription factor B1 subunits of human and yeast transcription factor II H complex, and metastasis-associated protein S100A4 [41,[50][51][52][53][54][55]. S100A4 is overexpressed in a range of different tumor types [56]. Its oncogenic property involved inhibition of p53-dependent growth arrest and apoptosis through enhancing p53 degradation. The presence of poly-PR/GR might compete with the binding of S100A4, which might promote p53 function and stability [56]. CBP/p300 stabilizes p53 in response to DNA damage by preventing MDM2-mediated p53 degradation [57][58][59][60]. Binding of poly-PR/GR to p53 TAD2 might interfere with binding to CBP/p300 during DNA damage or under stress conditions and, in turn, might modulate p53 stability and function.
We demonstrated that binding of PR25/GR25 to p53 1−94 enhances the rigidity and formation of α-helical propensity ( Figure 3A-D). For both TADs, it has been reported that they show α-helical propensity in the absence of binding partners [61], and that binding to folded proteins, such as the nuclear coactivator binding domain of CBP, results in folding and increased helicity of p53 TAD2 [41]. We showed that interactions between p53 1−94 and PR25/GR25 are mediated by a network of electrostatic and/or hydrophobic interactions ( Figure 3E-H and Figure 5B). It is known that p53 TAD interactions with others are also mediated by both electrostatic and/or hydrophobic interactions [61]. Interactions between TAD2 and the DNA-binding HMG box are both hydrophobic and electrostatic, involving the minor-groove-binding residues and basic DNA-binding residues in the HMG box, respectively [62]. Thus, it is tempting to speculate that PR25/GR25 could compete with proteins binding to p53 TAD2, such as HMGB1, and modulate their function. [40].
Aberrant LLPS through poly-PR/GR has been reported for several disease-related RNA-binding proteins, including FUS, TDP-43, hnRNPA1, hnRNP2B1, TIA1, Ataxin-2, and Matrin-3 [33,45,67]. DPRs can alter the liquid-to-solid transition and aggregation of RNA-binding proteins and, in turn, MLO formation in cells [45]. Overexpression of poly-PR/GR, for example, leads to disruption of nucleolar phase dynamics by binding of poly-PR/GR to nucleolar components like nucleophosmin-1. Given that p53 stability is regulated in nucleoli and that functional nucleoli are required for MDM2-mediated p53 degradation [38,68], disruption of nucleoli by poly-PR/GR might explain the accumulation of p53 [36,38,45]. In the future, it will be interesting to investigate how p53 LLPS is involved in these (patho)physiological processes.
Poly-PR/GR mediated LLPS of p53 observed here might be of general importance in the regulation of transcriptional condensates. Activation domains from transcription factors, including p53, form phase-separated condensates with the mediator to activate gene expression [69]. In the future, it will be interesting to reveal if poly-PR/GR modulates the formation of p53 transcriptional condensates and, through this, regulates the expression of p53 target genes.
As a summary, we demonstrate that binding of p53 to PR25/GR25 is mediated by mainly p53 TAD2 and that phase separation of p53 is regulated by PR25/GR25 in vitro. This might help to understand the mechanistic role of p53 in poly-PR/GR dipeptide repeat related diseases such as ALS and FTD.  [64][65][66]. The addition of PR25/GR25 induces phase separation of p53, which is mediated by a combination of intermolecular interactions between Figure 5. Interaction between p53 and PR25/GR25 is mediated by mainly TAD2 and phase separation of p53 is induced by PR25/GR25. (A) p53 is found in two forms, which are open and p53 DBD bound to p53 TAD2 (closed state, auto-inhibited) [64][65][66]. The addition of PR25/GR25 induces phase separation of p53, which is mediated by a combination of intermolecular interactions between p53 DBD and p53 TAD2 and intermolecular interactions between p53 TAD2 and PR25/GR25; (B) Schematic representation of the network of hydrophobic and electrostatic interactions between p53 TAD2 and PR25/GR25. In addition to the important role of p53 in cancer, p53 stabilization and activation by C9orf72 poly-PR seem to be key mechanisms causing the neurodegenerative phenotype in primary cortical neurons expressing C9orf72 poly-PR, iPSCs from ALS patients with C9orf72 GGGGCC, transgenic fly expressing C9orf72 GGGGCC, and transgenic mouse expressing C9orf72 poly-PR [40]. Moreover, it has been shown that poly(GR) 80 expression in iPSC-derived motor neurons increased DNA damage, which leads to a pro-apoptotic response by activating p53 [24,39,63]. However, it remained unclear whether a direct interaction between p53 and poly-PR/GR could explain the phenotype. Our findings help to understand how stabilization of p53 and activation of the p53 transcriptional pathway are regulated and how these interactions could be targeted in neurodegenerative disease related to C9orf72 dipeptide repeat proteins.
Both poly-PR/GR and p53 have been reported to undergo LLPS individually; however, whether poly-PR/GR can directly interact with p53 and thereby regulate p53 LLPS remained elusive. We showed that phase separation of p53 can be triggered by PR25/GR25, and that an interaction between p53 DBD and p53 TAD2 contributes strongly to LLPS ( Figure 4A,B,D and Figure S2). This suggests that LLPS of p53 1−312 is driven by a network of intermolecular interactions between p53 TAD2 and PR25/GR25 and intermolecular interactions between p53 TAD2 and p53 DBD ( Figure 5A). The dissociation of droplets observed at high concentrations of PR25/GR25 is likely due to saturation of the inter-molecular interactions between p53 TAD2 and p53 DBD . This model also implies that PR25/GR25 competition with the intra/intermolecular p53 TAD2 -p53 DBD interaction could lead to an activation of p53, through loss of the previously described auto-inhibition [64][65][66], which is in line with the enhanced p53 transcriptional activity observed in poly-PR disease models [40].
Aberrant LLPS through poly-PR/GR has been reported for several disease-related RNA-binding proteins, including FUS, TDP-43, hnRNPA1, hnRNP2B1, TIA1, Ataxin-2, and Matrin-3 [33,45,67]. DPRs can alter the liquid-to-solid transition and aggregation of RNA-binding proteins and, in turn, MLO formation in cells [45]. Overexpression of poly-PR/GR, for example, leads to disruption of nucleolar phase dynamics by binding of poly-PR/GR to nucleolar components like nucleophosmin-1. Given that p53 stability is regulated in nucleoli and that functional nucleoli are required for MDM2-mediated p53 degradation [38,68], disruption of nucleoli by poly-PR/GR might explain the accumulation of p53 [36,38,45]. In the future, it will be interesting to investigate how p53 LLPS is involved in these (patho)physiological processes.
Poly-PR/GR mediated LLPS of p53 observed here might be of general importance in the regulation of transcriptional condensates. Activation domains from transcription factors, including p53, form phase-separated condensates with the mediator to activate gene expression [69]. In the future, it will be interesting to reveal if poly-PR/GR modulates the formation of p53 transcriptional condensates and, through this, regulates the expression of p53 target genes.
As a summary, we demonstrate that binding of p53 to PR25/GR25 is mediated by mainly p53 TAD2 and that phase separation of p53 is regulated by PR25/GR25 in vitro. This might help to understand the mechanistic role of p53 in poly-PR/GR dipeptide repeat related diseases such as ALS and FTD.
Synthetic peptides were synthesized and HPLC-purified by Peptide Specialty Laboratories GmbH at Heidelberg, Germany. Peptides were obtained as lyophilized powder.

Protein Expression and Purification
To express the recombinant unlabeled or 15 N labeled or 15 N 13 C labeled ZZ-His 6 proteins, different bacterial expression pETM11-ZZ-His 6 vectors were transformed into E. coli BL21-DE3 Star strain. On the next day, a single colony was picked and grown overnight in 20 mL lysogeny broth (LB) medium supplemented with 50 mg/L kanamycin. Then, 10 mL of the pre-culture was transferred to 1 L LB or into minimal medium supplemented with either 6 g of unlabeled glucose or 2 g of 13 C 6 H 12 O 6 (Cambridge Isotope Laboratories, Saarbrücken, Germany) and 3 g of either unlabeled NH 4 Cl or 15 NH 4 Cl (Merck, Darmstadt, Germany). When OD (600 nm) reached 0.8, 1 L of cells was induced with 0.5 mM Isopropyl β-D-1-thiogalactopyranoside (IPTG, BLDpharm, Shanghai, China). Protein expression was performed at 20 • C for 16 h. Cell pellets corresponding to disordered protein fragments (p53 1−94 , p53 TAD1 , and p53 TAD2 ) were harvested and sonicated in denaturating lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 20 mM Imidazole, 6 M urea), and cell pellets corresponding to folded fragments (p53 DBD and p53 1−312 ) were harvested and sonicated in non-denaturating lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 20 mM Imidazole, 2 mM tris(2-carboxyethyl)phosphine(TCEP, BLDpharm, Shanghai, China). After sonication, samples were centrifuged at 6198 rcf for 45 min at 4 • C. ZZ-His 6 tag proteins were purified from the lysate using Ni-NTA agarose beads (Qiagen, Hilden, Germany), and then ZZ-His 6 tag was cleaved by the addition of 2 (w/w) % His 6 -tagged TEV protease for 16 h at 4 • C. Desalting of proteins into low imidazole buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 2 mM Imidazol) was performed using a desalting column (HiPrep 26/10, GE Healthcare, Chicago, IL, USA) on an ÄKTA Pure system (GE Healthcare, Chicago, IL, USA). Untagged proteins were separated from uncleaved protein by performing a second affinity purification using Ni-NTA beads. Lastly, size exclusion chromatography purification was performed in the buffer of interest using a gel filtration column (Superdex 75 Increase, GE Healthcare (Chicago, IL, USA) for p53 1−94 , p53 DBD , p53 1−312 ; and Superdex Peptide, GE Healthcare (Chicago, IL, USA) for p53 TAD1 and p53 TAD2 ). Their absorbance at 280 nm was used to estimate protein concentrations by assuming that the ε at 280 nm was equal to the theoretical ε value.

Resonance Assignment and NOE Experiments
The following 3D spectra were acquired for resonance assignment of the 300 µM 13 C, 15 N-labeled p53 1−94 and the 300 µM 13 C, 15 N-labeled p53 1−94 in complex with 400 µM unlabeled PR25 and GR25: noesyhsqcgpwgx13d with 200 ms NOESY mixing time experiment for 13 16 scans, 256 points in F1, 1024 points in F2) were recorded for unbound p53 1−94, and p53 1−94 in the presence of either PR25 or GR25 with a saturation period/total interscan delay of 3.0 s. HetNOE values were calculated by division of intensity of saturated spectra to intensity of unsaturated spectra. Standard deviations of saturated and unsaturated spectra were calculated using 10 additional random peaks. Error bars for heteronuclear NOE values were derived from error propagation calculation using a standard deviation of 10 arbitrarily chosen noise peaks in saturated and unsaturated spectra.
Random coil chemical shifts were used as reference values to calculate secondary chemical shift based on ncIDP (neighbor corrected IDP library) using the following formula [71][72][73]:

Fluorescence Polarization Measurements
N-terminally fluorescein isothiocyanate (FITC)-labelled PR25/GR25 were dissolved into 20 mM Hepes pH 7.0, 50 mM NaCl, 2 mM TCEP, and 0.04% (w/v) NaN 3 . Measurements were taken at room temperature in black 384-well plates using a ClarioStar Plus (BMG labtech, Ortenberg, Germany) spectrophotometer. Filters were selected as a function of FITC optical characteristics (λ ex = 495 nm, and λ em = 530 nm). Then, 100 nM FITC-labelled PR25/GR25 were incubated with increasing concentrations of purified p53 1−94 and p53 TAD2 in a final volume of 35 µL. The polarization data were fitted using graphpad prism 8 with the following equation: Here, P 0 represents the polarization of FITC-labelled peptides in absence of p53 1−94 and p53 TAD2 , and P max to the highest polarization of the binding curve corresponding to the saturation of the interaction. L corresponds to the concentration of p53 1-94 and p53 TAD2 proteins, and K d is the dissociation constant.