Mesoscopic p53-rich clusters represent a new class of protein condensates

Abstract

The protein p53 is an important tumor suppressor, which transforms, after mutation, into a potent cancer promotor. Both mutant and wild-type p53 form amyloid fibrils, and fibrillization is considered one of the pathways of the mutants' oncogenicity. p53 incorporates structured domains, essential to its function, and extensive disordered regions. Here, we address the roles of the ordered (where the vast majority of oncogenic mutations localize) and disordered (implicated in aggregation and condensation of numerous other proteins) domains in p53 aggregation. We show that in the cytosol of model breast cancer cells, the mutant p53 R248Q reproducibly forms fluid aggregates with narrow size distribution centered at approximately 40 nm. Similar aggregates were observed in experiments with purified p53 R248Q, which identified the aggregates as mesoscopic protein-rich clusters, a unique protein condensate. Direct TEM imaging demonstrates that the mesoscopic clusters host and facilitate the nucleation of amyloid fibrils. We show that in solutions of stand-alone ordered domain of WT p53 clusters form and support fibril nucleation, whereas the disordered N-terminus domain forms common dense liquid and no fibrils. These results highlight two unique features of the mesoscopic protein-rich clusters: their role in amyloid fibrillization that may have implications for the oncogenicity of p53 mutants and the defining role of the ordered protein domains in their formation. In a broader context, these findings demonstrate that mutations in the DBD domain, which underlie the loss of cancer-protective transcription function, are also responsible for fibrillization and, thus, the gain of oncogenic function of p53 mutants.

INTRODUCTION

Wild-type (WT) p53 is a transcription factor, which is inactivated in almost every tumor, mostly through mutations in the TP53 gene.^1–6 The importance of p53 in tumor suppression is demonstrated by the susceptibility to cancer of individuals who inherit a mutant TP53 allele (Li–Fraumeni syndrome) and the spontaneous tumor predisposition of Trp53-null mice.¹ Compared to humans, elephants suffer a ca. Fourfold lower cancer mortality, potentially related to multiple copies of TP53.⁵ The p53 protein is such a powerful tumor suppressor that it is inactivated in almost every tumor, through mutations in the TP53 gene, in about 50% of human cancers, or deregulation of its associated pathways.^1,2 The mutations induce structural destabilization of the mutant p53s resulting in loss of transcription function. The mutants also readily aggregate into degradation-resistant amyloid fibrils.^2,7–9 The ability of the mutant p53 aggregates to sequester WT p53, its paralogs, p63 and p73,^2,8,10,11 and other essential cellular proteins represents a potent dominant-negative and gain-of-oncogenic-function mechanisms.^2,8,10–12 Importantly, fibril suppression, for instance, by stabilization of the mutant p53 conformation, has been identified as a general way to fight cancer.^2,13 Despite the well-demonstrated oncogenic relevance of the aggregation of p53 and its mutants, numerous questions about the molecular pathway of this aggregation remain elusive.

The functional form of p53 combines four identical chains which incorporate both structured domains, essential for binding to DNA, and extensive disordered regions targeted by the p53 negative regulator, MDMX [Fig. 1(A)].^14,15 The rich structural diversity of p53 affords the opportunity to elucidate general questions of protein aggregation. In the last 10 years, liquid condensates of proteins with significant disordered regions, which carry out crucial functions as common membraneless organelles such as nucleoli, Cajal and P-bodies, and stress granules, have drawn much interest.^17–21 The propensity to form dense liquids at concentrations approximately μM, orders of magnitude lower than fully structured proteins and comparable to the physiological content of the condensing proteins, has been attributed to the strong intermolecular attraction between the disordered domains.^19,20 These studies pose the question whether the presence of large disordered regions is a necessary prerequisite for condensation and aggregation at physiologically relevant micromolar concentrations.^19–22 A related question is whether liquid phases facilitate the formation of ordered biological solids.²¹ p53 is a suitable system to address these general questions since it only forms a thermodynamically stable dense liquids^17–24 if truncated²⁵ or at low pHs.²⁶ In this way, data obtained with p53 promote discrimination between the roles of the dense liquids and the distinct recently discovered condensates, the mesoscopic p53-rich clusters,^16,27,28 in aggregation and coaggregation. Another important question is whether the disordered p53 regions enhance the protein's propensity to aggregate and, in this way, promote the gain of oncogenic function of the mutants, or the mutations in the DBD domain, which underlie the loss of cancer-protective transcription function, are solely responsible for aggregation.

FIG. 1.

The protein p53, its mutation and aggregation of the mutant p53 R248Q in breast cancer cells. (A) The structure of a p53 monomer. The transactivation domain (TAD, red), proline-rich region (PRR, yellow), DNA binding domain (DBD, purple), and tetramerization domain (TED, green). TAD and PRR are intrinsically disordered; DBD and TED are structured. Most of the oncogenic mutations localize in the DBD. Purple bars designate mutations' frequencies. Arginine at site 248 is labeled. (B) Combined staining of breast cancer cells HCC70 with Pab240, which binds to unfolded or aggregated p53, and ThT, which detects amyloid structures. The cell nucleus is stained with a Hoechst dye. (C) Staining with Pab240 before and after treatment with 1,6-hexanediol, known to disperse dense liquid droplets of disordered proteins. (D) Distributions of the volumes of the puncta found in HCC70 cells treated with Pab240. Each trace represents the volume distribution of puncta from a single cell. Data in (B)–(D) are from Yang et al.¹⁶

The kinetics of protein fibrillization have been monitored in bulk assays, which convolve molecularly distinct events, such as nucleation, fibril growth, fracture, merging, and secondary nucleation, and yield sparse details on the molecular mechanisms. In consequence, the mechanisms suggested by data obtained in bulk assays incorporate numerous inconsistencies.^{2,7,8,29–38} To probe the mechanisms that govern phase behaviors of the mutants across multiple length scales, from cellular to molecular, here we rely on immunofluorescent 3D confocal microscopy of breast cancer cells expressing the p53 mutants combined with light scattering from solutions of the purified protein and time-resolved in situ atomic force microscopy (AFM) of growing fibrils.^27,28

In the studies presented here, we employed WT p53, the mutant R248Q, and stand-alone ordered DBD and disordered N-terminus p53 domains [Fig. 1(A)]. As p53 binds to DNA, the positive arginine (R) 248 residue embeds in the minor groove of the double helix to support the contact.^2,36 Mutations at this site are the most frequent cancerogenic p53 mutations, in particular in ovarian and breast cancers.^2,36,39,40 To evaluate the roles of the ordered and disordered p53 domains in mesoscopic aggregation and fibrillization and the related gain of oncogenic function, we compared the behaviors of two truncated proteins, the p53 DBD and the disordered region at its N-terminus.

P53 R248Q AGGREGATES IN THE CYTOPLASM OF CANCER CELLS

We investigated the phase behaviors of p53 in the HCC70 breast cancer cell line, which carries the p53 R248Q mutation. Using immune-fluorescent 3D confocal microscopy, we quantified both aggregated and unaggregated p53.¹⁶ This technique involves fluorescently tagged antibodies binding to specific cellular targets, allowing us to map the 3D distribution of the target molecule.⁴¹ Uniformly distributed targets show diffuse staining, while target molecule aggregates appear as puncta.

Staining of the HCC70 cells with the dye Pab240, which specifically binds to denatured and aggregated p53, reveals exclusively cytosolic puncta and lack of detectable p53 staining in the nucleus [Fig. 1(B)]. Staining with Thioflavin T (ThT, a probe for amyloid structures) is exclusive to the nucleus, where it indicates the presence of non-p53 amyloid formations; no ThT staining is detectable in the cytoplasm.^16,27 Exposure of HCC70 cells to DO1, an antibody targeting the N-terminal region of p53, results in widespread nuclear staining; the results are discussed in detail in Yang et al.^15,26 The contrasting nuclear and cytoplasmic staining patterns observed with the three different probes suggest that the non-aggregated form of p53 R248Q is primarily localized within the nucleus (exhibiting diffuse DO1 staining), whereas the aggregated or structurally altered variant of p53 R248Q, detected through Pab240 punctate staining, is predominantly found in the cytoplasm.

HCC70 cells treated with 1,6-hexanediol, which destabilizes liquid droplets,^18,42 showed no reduction in the number of p53 puncta [Fig. 1(C)], indicating that the p53 R248Q aggregates are not dense liquid droplets. The distributions of volumes of the aggregates, evaluated from their total fluorescent intensity, in 32 cells are similar and relatively narrow with average volumes approximately 0.2 μm³ [Fig. 1(D)]. The narrow and weighted to small sizes volume distributions are not consistent with behaviors expected for both disordered agglomerates and dense liquid droplets, whose stochastic nucleation at low driving forces may result in broad distributions and greater variability between cells.

Combining the findings from Pab240 and ThT imaging alongside the effects of 1,6-hexanediol, it is evident that the mutant p53 R248Q assembles into cytoplasmic aggregates with a uniform size distribution in breast cancer cells. Separate investigations have shown that, in contrast to the mutant, wild-type (WT) p53 does not undergo aggregation within these cancer cells.¹⁶ These observations suggest that the cytoplasmic aggregates formed by p53 R248Q are neither amyloid fibrils nor stable dense liquid droplets. To further explore the characteristics and formation processes of these aggregates, we conducted in vitro experiments, in which we could vary the parameters that control aggregation, using purified WT p53 and R248Q p53.

WT P53 AND R248Q P53 FORM MESOSCOPIC PROTEIN-RICH CLUSTERS

To characterize the aggregation in WT p53 and R248Q p53 solutions, we employed oblique illumination microscopy (OIM) [Figs. 2(A)–2(C)].^43–46 Solution inhomogeneities are detected from the light that they scatter toward the objective lens [Fig. 2(A)]. We apply a Stokes–Einstein procedure to the recorded Brownian trajectories of individual aggregates, extracted from sequences of OIM images [Fig. 2(C)]^43,45 and obtain the aggregates' sizes. With this procedure, OIM detects and measures sizes in the range from 20 nm to 5 μm, much smaller than the diffraction limit of conventional optical microscopy.^28,43–46 In experiments with WT p53, OIM revealed no aggregates at temperatures below 18 °C.²⁸ By contrast, the mutant p53 R248Q formed aggregates at temperatures as low as 15 °C [Fig. 2(B)], indicating a stronger aggregation propensity of this mutant.²⁷

FIG. 2.

Mesoscopic protein-rich clusters of p53 R248Q. (A) Schematic of oblique illumination microscopy (OIM). (B) OIM micrographs of a 2 μM p53 R248Q solution at the beginning of monitoring by OIM and after 30 min. T = 15 °C. The clusters appear as bright speckles. (C) The Brownian trajectory of one cluster. (D) Number density distribution of the cluster radii R determined by OIM. The averages of five measurements are displayed. The error bars represent the respective standard deviations. (E) The evolution of the average radius R and number concentration N of clusters determined at 15 °C by OIM from images as in (B). Horizontal lines denote the mean values of R and N. (F) The concentration dependence of R and N 15 °C. In (E) and (F), the averages of five measurements are displayed. The error bars represent the respective standard deviations and are smaller than the symbol size for most measurements. (G) Schematic of formation of mesoscopic p53-rich clusters owing to accumulation transient misassembled oligomers, tentatively represented as pentamers and highlighted in oval. Data in (B) and (D)–(F) are from Yang et al.¹⁶

To investigate whether the aggregates possess amyloid-like characteristics, we employed the fluorescent probe anilinonaphthalenesulfonate (ANS). These assays demonstrated that under comparable conditions, specifically at 2 μM, amyloid fibril formation typically requires several hours to initiate^27,28 Furthermore, the aggregates differ from classic amyloid behavior, displaying a relatively uniform size distribution [Fig. 2(D)], with an average R of 45 ± 5 nm at 2 μM and 15 °C, comprising roughly 1000 p53 molecules per aggregate. Both R and N (aggregate concentration) remained stable over at least 2 h [Fig. 2(E)], which contrasts with the dynamics expected from liquid–liquid phase separation—a first-order phase transition^22,23—where continuous nucleation and droplet growth would lead to increases in both R and N over time.^47,48

The reversibility of the aggregates is demonstrated by the concentration-dependent behavior of N, which decreases approximately 20-fold, from 8.7 × 10⁸  to 0.4 × 10⁸ cm⁻³, when the protein concentration is reduced seven times, from 2.0 to 0.3 μM [Fig. 2(F)]. This pronounced decline in N with dilution suggests that the aggregates are not permanently disordered agglomerates, which would scale proportionally with protein concentration. Instead, the data imply that the structures are condensates maintained in a dynamic balance with the surrounding solution. At 37 °C, the aggregate radius R expands to approximately 150 nm,^26,27 aligning well with the typical size of intracellular clusters observed in HCC70 cells (Fig. 1). This similarity in size supports the interpretation that the cytoplasmic aggregates seen in HCC70 breast cancer cells likely represent mesoscopic p53-rich clusters.

The behaviors exhibited by both WT p53 and p53 R248Q aggregates align with earlier reports describing mesoscopic clusters enriched in globular proteins.^28,43,49,50 Based on these similarities, we interpret the observed aggregates as mesoscopic p53-rich clusters. Recent theoretical models suggest that such clusters, particularly in multi-subunit proteins like tetrameric p53, arise from the transient accumulation of misassembled oligomers [Fig. 2(G)].^28,49,50 The average cluster size is governed by the interplay between the lifetime of these misassembled intermediates and their diffusion-driven escape from the cluster's interior. As a result, cluster size remains constant over time and is largely unaffected by protein concentration.^50,52 The overall amount of protein captured in the clusters and the related parameters such as number of clusters and the total volume they occupy increase exponentially with protein concentration. This behavior reflects a thermodynamic equilibrium between the clusters and the surrounding solution.^49,51,52 Notably, the clusters formed by both p53 R248Q and WT p53 conform closely to the predictions of this model.^27,28

The mesoscopic p53-rich clusters observed in vitro and in breast cancer cells represent a phase distinct from dense protein liquid.^17–21 The strongest distinction from dense liquids relates to their sizes, which are limited to about 50 nm, independent of the protein concentration, and, together with the numbers of clusters, remain steady in time (Fig. 2). In another distinction from dense liquids, the mechanism of cluster formation requires the presence of a modified protein species [Fig. 2(G)]; the addition of a chemical transformation to a phase transformation connects the equilibrium concentration to the initial p53 concentration, a further difference with the solution-dense liquid dynamics.^16,27 On the other hand, the clusters share some features of the dense liquid droplets. Their spherical shapes [Figs. 3(A)–3(C)]¹⁶ suggest that the cluster phase is liquid, which allows individual clusters to succumb to minimization of their total surface free energy. Another manifestation of the role of surface free energy in the cluster evolution is the recorded slow, over several hours, increase in the cluster size with a time exponent of ca. 0.3,²⁸ indicative of Ostwald ripening driven by the minimization of the surface free energy of the cluster population.^51,53,54

FIG. 3.

Fibrillization of wild type and mutant p53. (A)–(C) Negative staining EM micrographs of aggregates of WT p53 and p53 R248Q. (A) Clusters and fibrils. Gold arrows highlight empty clusters; red arrows, clusters that spawn fibrils; and blue arrows, fibrils. (B) Branched fibrils coated with amorphous agglomerates. (C) Amorphous agglomerates. (D) Schematic of two-step nucleation of fibrils. Step 1. Mesoscopic p53-rich clusters, highlighted in ovals, form from misassembled p53 oligomers and native tetramers. Step 2. Fibrils nucleate within the mesoscopic clusters. Fibril growth proceeds classically, via sequential association of p53 monomers from the solution, as seen in (A).⁴⁰

NUCLEATION OF P53 FIBRILS IS HOSTED BY MESOSCOPIC CLUSTERS

Using negative-stain transmission electron microscopy (TEM), we visualized the assemblies present in solutions containing WT p53 and the p53 R248Q variant [Figs. 3(A)–3(C)].⁵⁵ The TEM images revealed three distinct types of p53 condensates: (i) spherical aggregates approximately 60–80 nm in diameter with a relatively uniform size distribution [Fig. 3(A)], similar to the mesoscopic p53-rich clusters identified through scattering methods [Figs. 2(D) and 2(E)]; (ii) elongated, filamentous structures indicative of amyloid fibrils [Fig. 3(A)]; and (iii) irregular, amorphous agglomerates that either surround fibrils [Fig. 3(B)] or exist as separate entities [Fig. 3(C)]. The finding of mutant and wild-type p53 aggregates of distinct morphologies concurs with previous in vitro observations.^7,56,57 Notably, many fibrils appeared to originate from the spherical aggregates, which are presumed to correspond to the mesoscopic p53-rich clusters [Fig. 3(A)]. This observation supports a two-step nucleation process for amyloid-like p53 fibrils, wherein mesoscopic p53-rich clusters act as sites that facilitate fibril nucleation, as depicted in Fig. 3(D).

P53 FIBRIL GROWTH

To monitor the dynamics of WT p53 fibril growth, we employed time-resolved in situ atomic force microscopy (AFM).⁵⁸ An advantage of AFM methods to recently applied super-resolution microscopy is the absence of fluorescent tags attached to the fibrillating molecules^59–61 and the ability to measure fibril thickness to characterize fibril structure.^59,62,63 AFM micrographs [Fig. 4(A)] demonstrate that the thicknesses (measured as in Ref. 59) of the WT p53 fibrils range from approximately 1.5 to 7 nm and suggest the individual fibrils may have distinct structures, i.e., different fibril polymorphs may coexist in the same solution. Importantly, the fibril thickness varies substantially along a fibril's length, implying that divergent structures may coexist in the same fibril and the fibril structure may vary as the fibril grows, in contrast to Aβ40 fibrils that maintain their structures and thicknesses during growth.^64–66 Individual fibrils alternate between growth (before 2.5 min), rapid dissolution (between 2.5 and 6 min), and slow growth thereafter [Figs. 4(B) and 4(C)] despite the steady solution conditions. These unusual dynamics may manifest stop-and-go kinetics at the fibril tip attributable to interactions between incoming monomers during docking and restructuring^61,67–69 or transient impurity poisoning,⁶⁹ to varying fibril structure, as suggested by the thicknesses in Fig. 4(A), or, more trivially, to convective flows in the solution that supply varying p53 monomer concentration to the fibril tip.⁷⁰

FIG. 4.

The structure and dynamics of WT p53 fibrils by AFM. (A) Image of six fibrils. Side panels illustrate thickness determinations along cross sections marked with distinct colors. (B)–(C) Dynamics of growth and dissolution of a fibril, images in (B) and length in (C).

DOMINANT ROLE OF P53 DBD FOR THE MESOSCOPIC PROTEIN-RICH CLUSTERS

The above experiments compare the propensities to form mesoscopic p53-rich clusters of p53 R248Q and WT p53. The mutant exhibits greater numbers of clusters and forms them at a lower temperature, indicating that it forms clusters more readily. The mutant differs from WT p53 by a single destabilizing mutation in the ordered DNA-binding domain (DBD). This observation suggests that observed mesoscopic aggregation may be due to interactions of the DBD,^71,72 in contrast to formation of a second stable dense liquid (liquid–liquid separation) which relies on attraction between disordered protein chains.^17–21 For further tests of this conclusion, we explore the aggregation behaviors of purified p53 DBD solutions with OIM [Fig. 2(A)].

We measured the radii R and numbers N of aggregates per unit volume by OIM [Figs. 2(A)–2(C)]. Similar to WT p53 and p53 R248Q, the p53 DBD aggregates at 15 °C exhibit narrow size distributions with an average radius of $45\pm 10$ nm [Fig. 5(A)]. The aggregate sizes are independent of the monomer concentration, indicating a phase behavior not typical of dense liquid droplets and amorphous agglomerates.^16,28 Accordingly, we identify these aggregated as mesoscopic clusters. Interestingly, p53-DBD-rich clusters form at 15 °C, where full length WT p53 does not aggregate.²⁸ The second peak in the N(R) size distribution [Fig. 5(A)] becomes prominent at concentrations 5 μM and higher and may reflect nuclei of p53 DBD fibrils; this possibility is explored below.

FIG. 5.

Protein-rich clusters of p53 DBD. (A) Size distributions of clusters measured by OIM at three different concentrations of p53 DBD. (B) The radius and numbers of protein rich cluster of 2 mM full length and DBD of p53 at 25 °C. Red bar corresponds to p53 DBD and cyan bar, to full length p53.

To further explore the consequences of removing the disordered domains for the mesoscopic aggregation of DBD p53, we compare their size R and number concentrations N to those of full length p53 at the same concentration, 2 μM, and temperature, 25 °C [Fig. 5(B)]. The sizes of the p53-DBD-rich clusters are, within the measurement error, similar to those of full-length WT p53. Surprisingly, the number of p53 DBD clusters is somewhat lower than that of full-length WT p53 clusters.

Similar to clusters of full-length WT p53 and p53 R248Q, the clusters of p53 DBD were reversible. Decreasing in solution concentration from 5 to 2 μM, approximately 2.5-fold, reduces the number of clusters from 1.2 × 10⁸ to 0.16 × 10⁸ cm⁻³, approximately sevenfold [Fig. 5(A)]. The ability of p53 DBD to form mesoscopic protein-rich clusters supports the conclusion that the clusters of wild type and mutant p53 are mainly driven by DBD and not by the disordered domains.

FIBRILLIZATION OF P53 DBD

The nucleation of amyloid fibrils of full length WT p53 and p53 R248Q employs a two-step pathway, whereby the fibrils assemble within mesoscopic p53-rich clusters [Fig. 3(D)].^16,28 What remains unclear, however, is which domains are involved in the two-step nucleation of fibrils. To elucidate this issue, we monitored the kinetics and imaged the morphology of aggregates forming in solutions of p53 DBD fibrils. To identify the fibrils and monitor their growth, we added to a p53 DBD solution the dye ThT, which binds to the hydrophobic parts of the beta sheet stacks that constitute the fibrils.⁷³ Thus, the growth of amyloid fibrils is reflected by an increase in ThT fluorescence intensity. In solutions with p53 DBD concentrations of 40 and 20 μM, fibrillization exhibits a lag time of about 20 min, with growth completing in about 5 h [Fig. 6(A)]; both of these times are significantly shorter than the corresponding values for full-length WT p53 and other proteins forming amyloid fibrils.^7,8,12,16,74 Solutions with lower concentrations of p53 DBD exhibits somewhat longer lag times [Fig. 6(A)], which are still shorter than those of other proteins.^7,8,12,74

FIG. 6.

Fibrillization of p53 DBD. (A)–(B) The evolution of thioflavin T (ThT) fluorescence intensity I_ThT as a measure of the fibrillization of p53 DBD at 37 °C. (A) Fibrillization evolution at five p53 DBD concentrations. (B) Ficoll slows down the fibril nucleation and increases the degree of fibrillization. (C) Representative OIM images without and with crossed polarizers. Parallel arrows indicate normal OIM and crossed arrows indicates crossed polarizers. (D) Size distributions of aggregates seen in parallel polarizers (gold) and crossed polarizers (green); the latter was multiplied by a factor of 10 for better visibility. (E) Negative staining TEM microscopy detect p53 DBD clusters (indicated with gold arrows), fibrils (red arrows), and clusters hosting nucleation of fibrils (blue arrows).

The addition of Ficoll 70 kDa, a cross-linked compact polysaccharide with radius 4.7 nm,⁷⁵ to a solution of p53 DBD, extends the delay time and increases the degree of fibrillization [Fig. 6(B)]. Ficoll is a crowding agent, which increases the chemical potential of p53 DBD in the solution by restricting the volume available to the protein molecules and thus reducing their entropy.^76,77 The resulting greater fibrilization driving force leads to higher degree of fibrillization in the presence of Ficoll. With enhanced protein chemical potential, however, the suppressed fibril nucleation, seen as extended delay time, appears counterintuitive. A possible explanation may be found in an analogy with fibrils of full-length WT p53, whose nucleation is also delayed by the presence of Ficoll.²⁸ Ficoll was found to sequester in the mesoscopic p53-rich clusters, the location where fibrils nucleate and the increased viscosity was identified as the cause for delayed nucleation.²⁸ Thus, the extended fibrillization delay of p53 DBD in the presence of Ficoll is compatible with the two-step nucleation of these protein fibrils.

For further tests of the role of clusters on fibril nucleation, we monitored p53 DBD solutions with OIM, in which we placed in front of the objective lens of the microscope [Fig. 2(A)] a polarizer that could be oriented parallel or perpendicular to the plane of polarization of the incident light beam. The fibrils are highly anisotropic and rotate that plane of polarization. Thus, the light scattered from them would be detectable even if the polarizer is oriented perpendicular to the original polarization plane, a position often called crossed polarizer.⁷⁸ OIM observations with this polarizer orientation revealed that the number of detected aggregated decreased significantly [Figs. 6(C) and 6(D)]. Importantly, the locations of fibrils, detected with a crossed polarizer, overlapped with locations of spots from clusters, observed when the polarizer was parallel to the polarization plane of the incident light, suggesting that the fibril form within clusters. The average size of fibrils is larger than that of p53 DBD-rich clusters because the fibrils can grow out of a cluster [Fig. 6(D)], whereas the cluster size is restricted (Fig. 2). The number of clusters containing fibrils is significantly lower than the total number of clusters since not all clusters host fibril nucleation.

Finally, we examined the dynamics between p53 DBD-rich cluster and fibrils nuclei with negative staining transmission electron microscopy (TEM) [Fig. 6(E)]. The TEM micrographs reveal three types of aggregates: clusters with diameters 60 to 80 nm, fibrils with linear structures, and amorphous agglomerates. Importantly, these images captured fibrils that originate at clusters. The TEM observations support the two-step nucleation of p53 DBD amyloid fibrils, similar to observations with full-length WT p53 and p53 R248Q.^16,28

P53 DISORDERED DOMAIN DOES NOT FORM CLUSTERS AND FIBRILS

It has been suggested that fibrillization of p53 is not fully attributable to its DBD, but the disordered transactivation domain also plays a role.^79,80 The SEM and optical microscopy observations above (Fig. 6) indicate that p53 DBD alone is sufficient for the fibrillization of full length p53 by hosting fibrils in p53-rich clusters. To test a potential role of the p53 N-terminus segment, which contains transactivation and proline-rich domains [Fig. 1(A)], on mesoscopic aggregation and fibrillization, we monitored the aggregation of p53 N-terminus segment (Fig. 7).

FIG. 7.

Fibrillization of p53 disordered domain. (A) The radius of aggregates formed by p53 N-terminus with different concentrations at 15 °C. (B) The evolution of thioflavin T (ThT) florescence in the presence 40 μM of p53 N-terminus or p53 DBD at 37 °C.

The sizes of the aggregates of p53 N-terminus segment, monitored by OIM, increase in time [Fig. 7(A)], in contrast to the mesoscopic clusters observed with full-length WT p53, p53 R248Q, and p53 DBD. The initial sizes of the aggregates are ca. 30 nm [Fig. 7(A)]. In approximately 1 h, the aggregates' sizes reach from 35 to 60 nm. Increased size with time is a characteristic behavior of dense liquid droplets formed by liquid–liquid phase separation.^81,82 The observed size growth suggests that the disordered p53 N-terminus segment does not form clusters and the ordered p53 DBD dominates the formation of mesoscopic protein-rich clusters.

With p53 DBD, fibril nucleation is hosted by the mesoscopic clusters. Fibrillization of other proteins, however, may be facilitated by stable dense liquid droplets.^83,84 To test whether p53 N-terminus segment forms fibrils, we monitored the evolution of ThT intensity of p53 N-terminus segment and compare it with that of p53 DBD under identical conditions. Although the fluorescence intensity of solutions of p53 DBD exponentially increases after an approximately 20 min lag time, the intensity emitted by p53 N-terminus segment is steady for 12 h [Fig. 7(B)]. This lack of ThT signal indicates that p53 N-terminus segment does not form fibrils. We conclude that the disordered segment of p53 plays no role in its fibrillization is likely not a part of the dominant-negative and gain-of-function behaviors of mutant p53.

CONCLUSIONS

We explored mesoscopic aggregation and fibrillization of the p53 mutant R248Q and the stand-alone ordered DBD and disordered N-terminus domains. We establish that p53 R248Q forms mesoscopic protein-rich clusters, an anomalous liquid with several unique properties, both in breast cancer cells and in solutions of purified protein. We show that the mesoscopic clusters represent a novel protein phase, distinct from the dense liquid droplets. In further contrast to dense liquids, our findings with the p53 ordered and disordered segments demonstrate that the cluster formation relies on destabilization of the ordered domain of p53 and not on interactions between the disordered protein segments.

These findings demonstrate that the disordered p53 regions do not play essential roles in p53 aggregation. Thus, the mutations in the DBD domain, which underlie the loss of cancer-protective transcription function, are also responsible for p53 aggregation, seen as a part of the gain of oncogenic function mechanism of mutated p53.

AUTHOR DECLARATIONS

Conflict of Interest

The authors have no conflicts to disclose.

Author Contributions

David Yang: Formal analysis (lead); Investigation (lead); Methodology (lead); Writing – original draft (lead). Alexander Tilson: Investigation (supporting); Methodology (supporting). Michael Sherman: Investigation (lead); Methodology (lead). Navin Varadarajan: Conceptualization (lead); Formal analysis (lead); Methodology (lead); Resources (equal). Peter G. Vekilov: Conceptualization (lead); Data curation (equal); Formal analysis (equal); Funding acquisition (lead); Investigation (equal); Methodology (supporting); Project administration (lead); Resources (lead); Supervision (lead); Validation (lead); Writing – original draft (lead); Writing – review & editing (equal).

DATA AVAILABILITY

The data that supports the findings of this study are available within the article.