Propagation of aggregated p53: Cross-reaction and coaggregation vs. seeding

Significance

Aggregation of tumor suppressor p53 oncogenic mutants not only loses their activity but may also lead to gain of oncogenic function, possibly by coaggregation with other proteins. We explored how a destabilized p53 mutant may coaggregate with WT p53 and its homologs p63 and p73. The results are explained by the simple two-step initiation mechanism we proposed. Rather than the oncogenic mutant seeding coaggregation by a prion-like process, coaggregation results from the simultaneous unfolding and cross-reaction of WT and mutant molecules. Consequently, preformed p53 aggregate displays little seeding of aggregation of p53, and high concentrations of p53 mutants are required to trap WT p53 into aggregate. Coaggregation is predominantly by trapping rather than seeding and induced propagation.

Abstract

Destabilized mutant p53s coaggregate with WT p53, p63, and p73 in cancer cell lines. We found that stoichiometric amounts of aggregation-prone mutants induced only small amounts of WT p53 to coaggregate, and preformed aggregates did not significantly seed the aggregation of bulk protein. Similarly, p53 mutants trapped only small amounts of p63 and p73 into their p53 aggregates. Tetrameric full-length protein aggregated at similar rates and kinetics to isolated core domains, but there was some induced aggregation of WT by mutants in hetero-tetramers. p53 aggregation thus differs from the usual formation of amyloid fibril or prion aggregates where tiny amounts of preformed aggregate rapidly seed further aggregation. The proposed aggregation mechanism of p53 of rate-determining sequential unfolding and combination of two molecules accounts for the difference. A molecule of fast-unfolding mutant preferentially reacts with another molecule of mutant and only occasionally traps a slower unfolding WT molecule. The mutant population rapidly self-aggregates before much WT protein is depleted. Subsequently, WT protein self-aggregates at its normal rate. However, the continual production of mutant p53 in a cancer cell would gradually trap more and more WT and other proteins, accounting for the observations of coaggregates in vivo. The mechanism corresponds more to trapping by cross-reaction and coaggregation rather than classical seeding and growth.

The mechanism of aggregation of p53 (1) is different from that of the classical nucleation—growth of formation of amyloid fibrils (2–5). The classical mechanisms involve relatively slow nucleation events followed by rapid growth. The fibrils or their fragments act as seeds to accelerate greatly the polymerization of molecules from solution. However, the initiation of aggregation of p53 is relatively rapid, there appears to be little seeding from already polymerized molecules (6, 7), and stirring does not significantly speed up aggregation (1). An extensive Φ-value analysis of the aggregation of core domains of p53 suggests that the initial events involve the sequential unfolding of two molecules, involving a combination of first-order and second-order kinetics (1). Full-length p53, however, has a tetramerization domain, which gives a mixture of dimers and tetramers at normally encountered concentrations of protein. At 37 °C, tetrameric p53 dissociates into dimers with a dissociation constant of 50 nM, and the dimers to monomers with a dissociation constant of 0.55 nM, with half lives of 20 and 50 min, respectively, for dissociation (8). Tetramerization of p53 may complicate the aggregation kinetics. The possibilities of intramolecular core-domain interactions could, for example, change the kinetic mechanism and accelerate the initial aggregation events by proximity effects of core domains in tetramers. Further, WT protein may form mixed hybrids with a destabilized, oncogenic mutant and facilitate seeding of the aggregation of WT domains, which may be a cause of negative dominance (9–11). Aggregation-prone mutant p53 is also proposed to coaggregate with and inactivate p63 and p73 via residues 251–257 (strand S9) in the core domain acting as an aggregation nucleus (9). p53 does not form heterotetramers with p63 or p73 (12) but binds via core domain interactions (13–15).

Here, we first probed the consequences of tetramerization on the aggregation of full-length p53 and its mutants. Then, we examined the effects of mutant p53 seeding the aggregation of WT p53 and family members p63 and p73. We found that the kinetics of aggregation of full-length p53 is very similar to that of isolated core domains, and therefore the mechanistic data on monomers can be transferred to oligomers. We verified that coprecipitation of mutant p53 with p63 and p73 can be replicated in vitro, as well with WT p53, but that it occurs by a process of coaggregation and trapping by cross-reaction rather than seeding.

Results

Kinetics of Aggregation.

We compared the aggregation kinetics characteristics of different full-length p53 variants with full-length p53 at 37 °C (Table 1). The strategy was to perform measurements on the parent p53 over a wide range of concentrations and a smaller number of experiments on the mutants.

Table 1.

Full-length p53 variants

FL p53 variants	Mutant	-∆∆GD-N, kcal/mol*	Slope†
FLWT	—	0	1.58 ± 0.04
FLG245S	G245S	1.2	1.42 ± 0.05
QMFLYC	Y220C	4.2	1.45 ± 0.03
QMFLYCI195V	I195V	1.7	1.34 ± 0.03

^*ΔΔG_D-N was measured on mutation in the core domain of the construct in column 1.

^†Slope of the plots of logarithms of initial rates vs. protein concentration from ThT data.

Aggregation of full-length WT p53.

Full-length WT p53 (FLWT) forms a thioflavine T (ThT)-binding species before formation of large light scattering aggregates at 37 °C (Fig. 1). We fitted the ThT binding curves to the standard two-step equation over a wider range of concentrations (0.25–4 µM) than previously described (6) (Fig. 2 A and B). The rate constants for both k₁ and k₂ decrease below 1–2 µM protein (protein concentrations measured as monomers). The rate constants could be fitted to curves with dissociation constants of 1.3 ± 0.4 and 0.75 ± 0.11 µM, respectively. Those values are significantly above the equilibrium constants for tetrameric protein dissociating to dimers (50 nM in terms of tetramers or 200 nM in concentration of monomers) (8). Further, the concentrated stock solution of p53 was diluted 25-fold or more to initiate the reaction and would not have had time to dissociate during the fast phase of the reaction (16). Those constants presumably relate to the reaction schemes previously described (1) and do not represent actual dissociation constants.

Fig. 1.

Aggregation time courses of FLWT at 37 °C. Aggregation was monitored by (A) ThT binding and (B) light scattering. The solid lines through the data in A are the best fits to two-step sequential first-order reactions (6).

Fig. 2.

Concentration dependence of the aggregation kinetics of FLWT. Apparent hyperbolic concentration dependence of rate constants of (A) k₁ and (B) k₂ with protein concentrations. (C) Concentration dependence of initial rate of ThT fluorescence calculated from k₁, k₂, and amplitude, A, for first-order consecutive reactions. (D) Concentration dependence of initial aggregation rate measured directly from initial rates V₀ in arbitrary units.

Plots of logarithms of initial rates, V₀ (measured against t²) vs. protein concentration are used to give information about the molecularity of the nucleus (17). For ThT fluorescence following two-step sequential kinetics, a plot of log(k₁k₂A) vs. log[p53] is the equivalent to log(V₀) vs. log[p53] for the initial rate of ThT binding (where A is the amplitude of the signal) (1). Such plots for ThT had slopes of 1.5–1.6, showing that there is indeed a concentration dependence of the processes. The slope is between a first-order process, where the slope should be 1 (expected at very high concentrations of protein that favor dimerization), and a second order, where the slope should be 2 (expected at low concentrations).

Aggregation of full-length p53 Y220C in quadruple-mutant construct.

The full-length protein in quadruple-mutant construct (QMFL) contains four stabilizing mutations. The mutation Y220C destabilizes it by 4.2 kcal/mol (1, 18) to give full-length p53 Y220C in quadruple-mutant construct (QMFLYC), which was studied over only the concentration range of 1–4 µM (measured in single-chain units). At first sight, the rate constants seem independent of concentration, but the slope of a plot of log(k₁k₂A) vs. log[p53] was 1.45 ± 0.03 (Fig. 3C), indicating again a transition between a first-order process at high concentrations to a second-order process at low concentrations.

Fig. 3.

Concentration dependence of the aggregation kinetics of QMFLYC measured by ThT binding. Rate constants of (A) k₁ and (B) k₂ were measured at protein concentrations higher than 1 µM. (C) Concentration dependence of initial aggregation rate of QMFLYC calculated from k₁, k₂, and amplitude, A, for first-order consecutive reactions. (D) Time courses of QMFLYC aggregation, fitted to equation for first-order consecutive reactions.

The magnitude of k₁k₂A could increase with concentration of protein if the amplitude A increases faster than linearly while k₁k₂ remains constant. Plots of A vs. [FLWT] or vs. [QMFLYC] are nearly linear, as are k₁k₂ (Fig. 4). The product of k₁k₂ did increase with increasing concentration. The mutation I195V destabilizes the core domain of QMFLYC by 1.7 kcal/mol and increased both k₁ and k₂ of QMFLYC six- to sevenfold (Fig. 5).

Fig. 4.

Concentration dependence of the amplitude, A (A and C), and rate constants k₁, k₂ (B and D) of FLWT and QMFLYC monitored by ThT binding.

Fig. 5.

Concentration dependence of the aggregation kinetics of QMFLYCI195V monitored by ThT binding. Rate constants of (A) k₁ and (B) k₂ were measured at protein concentrations higher than 1 µM. (C) Concentration dependence of initial aggregation rate calculated from k₁, k₂, and amplitude, A, for first-order consecutive reactions. (D) Aggregation traces of QMFLYCI195V fitted to equation for first-order consecutive reactions.

Aggregation of WT p53 with mutation G245S.

The oncogenic mutation G245S in the core domain of WT p53 destabilizes it by 1.2 kcal/mol (18). k₁ increased fivefold with a slight decrease in k₂ (Fig. 6). Similar to other full-length p53 variants, the slope of a plot of log(k₁k₂A) vs. log[p53] was 1.42 ± 0.05 (Fig. 6C).

Fig. 6.

Concentration dependence of aggregation kinetics of FLG245S monitored by ThT binding. Rate constants of (A) k₁ and (B) k₂ were measured at protein concentrations higher than 1 µM. (C) Concentration dependence of initial aggregation rate calculated from k₁, k₂, and amplitude, A, for first-order consecutive reactions. (D) Aggregation traces of FLG245S fitted to equation for first-order consecutive reactions.

Comparison of p53 WT full-length and core domain construct.

The monomeric core domain construct WTLC (LC = long core; residues 89–312) had a sevenfold higher value of k₁ than for p53, whereas k₂ was similar but was still increasing linearly at 14 µM protein (Fig. 7 A and B). The slopes of plots of log(k₁k₂A) vs. log[WTLC] (2.68 ±0.11; Fig. 7C) and log(V₀) vs. log[WTLC] for scattering (1.96 ± 0.03; Fig. 7D), were fully indicative of a second-order process. Light scattering showed faster aggregation of WTLC than FLWT (Fig. S1).

Fig. 7.

Concentration dependence of the aggregation kinetics of WTLC. Both rate constants of (A) k₁ and (B) k₂ are concentration dependent. Double logarithmic plots of the product of (C) rate constants and amplitude (k₁k₂A) from ThT binding and (D) initial rates V₀ of aggregation from light scattering vs. the initial protein concentration.

Aggregation of Mixtures of WT and Mutants.

Full-length p53 hetero- and homo-tetramers.

Full-length tetramers undergo subunit exchange only slowly (16). After 24-h incubation at 0 °C, an equimolar mixture of WT₄ and mut₄ gives a 1:2:1 ratio of WT₄:WT₂mut₂:mut₄ (16). We compared the aggregation of various WT and mutant constructs at 2 µM of each. They were allowed to aggregate either separately; mixed together and immediately incubated at 37 °C (unexchanged, and only a small amount of subunit exchange occurs over the time course of aggregation); or preincubated together overnight on ice to form 1:2:1 complexes (subunit exchanged). Two micromolars together of WT and mutant full-length p53 (4 µM in total) that are unexchanged aggregated more rapidly than the sum of the two separate 2-µM samples because of the second-order component in the kinetics (Fig. 8). The best comparison for the effects of hetero-oligomerization is between the 4 µM solutions of subunit-exchanged and unexchanged mixtures. The aggregation of subunit-exchanged mixtures was only slightly faster than that of the unexchanged mixture detected by ThT binding and light scattering for FLG245S and FLWT, and QMFLYCI195V and QMFLYC (Fig. 8). However, exchanged vs. unexchanged QMFLYC and QMFL showed very significant effects (Fig. 8 B and C). The I195V and G245S mutations cause much smaller changes in stability (1.7 and 1.2 kcal/mol, respectively) than does Y220C (4.2 kcal/mol), and therefore the effects of hetero-oligomerization are much less marked.

Fig. 8.

Effect of mutant and WT hetero-tetramer formation on aggregation of full-length p53. Comparison of the aggregation of subunit exchanged mixture, subunit unexchanged mixture, and separate aggregation of (A) FLG245S and FLWT, (B) QMFLYC and QMFL, and (C) QMFLYC and QMFL monitored by ThT binding and comparison of subunit exchanged with unexchanged mixture by (D) light scattering. Proteins are all at 2 µM (in single chains).

Mutant p53 core domain coaggregates with WT full-length.

We tested the seeding effects of the highly destabilized oncogenic mutations R175H (LCR175H; Fig. 9 A and B) and Y220C (LCYC; Fig. 9 C and D) in the construct of the long core domain on FLWT. One micromolar of destabilized core did cause a burst of seeding when mixed with 2 µM WT p53. This was most clearly seen in Fig. 9 B and D, where the black curve, the difference in ThT fluorescence between 1 µM mutant + 2 µM FLWT aggregating together and the sum of 1 µM mutant and 2 µM FLWT aggregating separately, gave a transient increase in the aggregation. Mixing 1 µM of preaggregated LCR175H (Fig. S2) accelerated aggregation of 2 µM FLWT much less than found for fresh LCR175H (Fig. 9), showing that the reaction of a state before the formation of large aggregates is the major causative agent of coaggregation.

Fig. 9.

Aggregation of FLWT seeded by (A and B) LCR175H and (C and D) LCYC monitored by ThT fluorescence at 37 °C. FLWT and (A) LCR175H or (C) LCYC mixture (2 µM of FLWT +1 µM LCR175H or LCYC) aggregates faster than the sum of their separate aggregations, (2 µM of FLWT) + (1 µM LCR175H or LCYC). Deducting the separate aggregation from the aggregation of the mixture shows a burst in the aggregation process of FLWT in the presence of (B) LCR175H or (D) LCYC.

Simulation of coaggregation.

We simulated the coaggregation of WT and mutant p53 according to the mechanism in Fig. 10 and the legend in Fig. 11. We did not attempt a quantitative fitting of experimental data but made just an illustration of the consequences of the proposed mechanism. There is an initial burst of aggregation of WT as it encounters the unfolded mutant molecules that are rapidly formed via k_+1mut. During that time, the mutants are rapidly depleted and therefore after a few minutes are no longer present and no longer affect the aggregation of WT.

Fig. 10.

Scheme for analyzing coaggregation data.

Fig. 11.

Simulation of seeding with same parameters for WT as in figure 3 of ref. 1 and mutants of different values of k₊₁ and k₊₂. Mutants are assumed to aggregate with both k_+1mut and k_+2mut (A) 10× higher and (B) 100× higher than the corresponding values for WT. WT is assumed to react at its normal k_+2wt if it encounters unfolded mutant and vice versa. The reverse rate constants k_-1 and k_-2 are assumed to be the same for WT and mutant.

Coaggregation of p53 mutants with p63 and p73.

With melting temperatures of 62.0 and 49.4 °C, respectively, p63 and p73 are very stable at 37 °C (19). However, during the fast aggregation process of 1 µM LCR175H, small amounts of p73 and p63 were rapidly trapped into the aggregate indicated by the higher ThT and scattering fluorescence intensity (Fig. 12). SDS/PAGE also detected that FLG245S trapped tiny amounts of p73 into the aggregate (Fig. S3). Increasing the concentration of FLG245S from 3 to 10 µM caused more p73 to coaggregate.

Fig. 12.

Coaggregation of p53 LCR175H with p73 and p63. Aggregation of p73 and p63 at 37 °C monitored by (A and B) ThT binding and (C and D) light scattering shows initial transient increase in the presence of 1 µM p53 LCR175H.

Discussion

Tetrameric p53 Aggregates Similarly to Monomeric Core Domains.

NMR and small-angle X-ray scattering (SAXS) studies show that FLp53 occupies an extended cross-like conformation, anchored at the center by the tet domains and the four core domains having a high degree of mobility (20). The isolated WTLC domain has similar stability to WTFLp53, with very similar melting temperature (46.2 °C) (19, 21), as do other core domain mutants compared with full-length proteins (22). This high degree of independence in the tetramer was reflected here by the kinetics of aggregation of FLp53 and its mutants being very similar to that of isolated core domains, with early initiation events having second-order components, showing separate tetrameric FLp53 units reacting intermolecularly (Figs. 1 and 2). Formation of tetramers does not significantly enhance, if at all, the rate of aggregation of core domains.

WT and Mutant Hetero-Tetramers Aggregate Faster Than Mixture of Homotetramers.

Coexpression of mutant and WT p53 leads to heterotetramers (23), possibly causing a dominant negative effect (10) by directly impairing DNA binding activity (23–25) or inducing a structural change in WT (11). We took advantage of the slow exchange of subunits of full-length proteins (16), which is slower than aggregation at 37 °C, to compare combinations of WT and other constructs with destabilized mutants that were either mixtures of homotetramers (unexchanged) or had been allowed to undergo prior subunit exchange (exchanged; Fig. 8). The effects varied from exchanged and unexchanged following very similar aggregation curves for small differences in stability to larger changes for a highly destabilized mutant (Fig. 8 from A to C and from top to bottom in D). The intramolecular effects in hetero-oligomers compete with intermolecular reactions in the concentration ranges studied (∼4 µM).

Aggregating Unstable Mutants Trap WT p53 Family Members into Aggregate.

Often, very small amounts of aggregates or fragments thereof, as low as 0.05%, seed aggregation of bulk protein (26–28). Here, stoichiometric quantities of unstable mutant or its aggregate were required to detect enhancement of the rate of aggregation of WT or other stable variants (Figs. 8 and 9), as indicated previously for core domains (7). Similarly, only a tiny amount of p63 or p73 was trapped into the aggregate by a rapidly aggregating mutant p53 (Fig. 12 and Fig. S3).

The oncogenic mutant LCR175H was much more effective at causing aggregation of FLp53 than was the aggregated mutant (Fig. 8 and Fig. S2). The stimulation of aggregation is seen as a burst of trapping of WT protein into the mutant aggregates (Figs. 9 and 12). We can simulate the kinetics of aggregation by the model in Fig. 10, which is the model used to describe the self-aggregation of p53 (1) expanded to include cross-reactions of unfolded WT and mutant p53, which is described in Fig. 11. There is an initial burst of aggregation of WT as it encounters the unfolded mutants that are rapidly formed via k_+1mut. During this time, the mutant molecules are rapidly depleted and therefore after a few minutes are no longer present and no longer affect the aggregation of WT.

Cross-Reaction and Coaggregation vs. Seeding.

Preformed aggregate has little stimulating effect on the rate of aggregation of WT protein, and aggregation is stimulated by fresh unaggregated mutant molecules as they are in the process of unfolding and aggregating, in a trapping process, consistent with the model in Fig. 10. This mechanism is quite different from the classical nucleation growth model of aggregation where formation of the nucleus is very slow and growth is relatively rapid. This process is consistent with the finding that excess amounts of mutant p53 is required to suppress WT p53 function (29). It also explains why simultaneously expression of destabilizing p53 mutants (V143A and R175H) and p73 in a dual-inducible system, where protein expression levels are better controlled than transient cotransfection, failed to inhibit the ability of p73 to suppress cell proliferation (30), whereas these p53 mutants coaggregate (9) and abrogate the activity of p73 in transient cotransfection experiments (13, 15, 31, 32). WT p53 naturally aggregates at an appreciable rate, and the initial nucleation events are fast. Accordingly, rapidly aggregating oncogenic mutants have only a small effect on the aggregation of WT protein. In vitro, the effects of mixing oncogenic mutant with WT give only a transient burst of trapping of WT protein and then the supply of mutant is depleted. However, in cancer cells, mutant p53 is continually produced and continues to trap WT protein, leading to the observation of coaggregates (9). Further, the concentration of p53 has about a 20- to 30-fold excess over p63 or p73 (33). Again, continual production of unfolded p53 will cause mutant p53, p63, and p73 to coaggregate because of the continued input into the system. Gain of function may be caused by the overwhelming amount of mutant present relative to that of WT p53 and homologs. The process is predominantly cross-reaction and coaggregation rather than classical seeding.

Methods

Protein Expression and Purification.

Mutants I195V and G245S were constructed using QuikChange site-directed mutagenesis (Stratagene) as described (1). Recombinant p53 proteins including core domain (94–312), long core domain (89–312), and full-length variants were expressed and purified as described previously (7, 34). For full-length p53 variants except superstable quadruple mutant (M133L/V203A/N239Y/N268D) of p53 (QMFL), DsbA within the lysate was removed by washing with denatured Escherichia coli protein after adsorption of target protein to a Ni column.

Subunit Exchange.

Subunit exchange of full-length p53 was performed based on the method used before (16). Equal amounts of two different full-length p53 variants were incubated on ice for 24 h to get an equilibrated subunit-exchanged mixture.

Kinetics of Aggregation.

Aggregation kinetics of p53 variants was generally monitored by light scattering and ThT fluorescence using a Horiba FluoroMax-3 spectrophotometer as previously described (6, 7). Proteins were added into temperature-equilibrated buffer at 37 °C with a final buffer composition of 25 mM potassium phosphate, 150 mM NaCl, 1 mM tris(2-carboxyethyl)phosphine hydrochloride (TCEP), 0.625% glycerol, 20 µM ThT, and 5% (vol/vol) DMSO, pH 7.2. For full-length p53, proteins were quickly prewarmed to 20 °C before addition into prewarmed buffer.

Seeding of WT p53, p63, and p73 Aggregation.

Aggregation was monitored as described in the previous paragraph. For the seeding effect of p53 long core R175H (LCR175H) and long core Y220C (LCYC) on FLWT, fresh LCR175H or LCYC was added together with fresh FLWT into preequilibrated buffer at 37 °C. For coaggregation of LCR175H with p63 and p73, the LCR175H/p63 or LCR175H/p73 mixture was preincubated on ice for 1 h and then added into 37 °C buffer to initiate the aggregation experiment. To check the effect of aggregated LCR175H on aggregation of FLWT, 1 µM LCR175H was aggregated at 37 °C for 70 min before addition of fresh FLWT.

Coaggregation of p73 and FLG245S Measured by SDS/PAGE.

Before initiating aggregation, 3 µM p73 was kept on ice for 1 h with 0, 3, and 10 µM FLG245S in 25 mM potassium phosphate, 150 mM NaCl, 1 mM TCEP, and 0.625% glycerol, pH 7.2. Then, the samples were incubated at 37 °C with shaking to start the aggregation. After aggregation for 100 min, aggregate was collected by centrifugation at 14,300 × g for 30 min at 4 °C. Aggregates were dissolved by 8 M urea and then analyzed together with supernatants using NuPAGE 4–12% Bis-Tris gel (Invitrogen).