Kinetic mechanism of p53 oncogenic mutant

aggregation and its inhibition

Rainer Wilcken1, GuoZhen Wang1, Frank M. Boeckler2, and Alan R. Fersht3
MRC Laboratory of Molecular Biology, Hills Road, Cambridge CB2 0QH, United Kingdom
Contributed by Alan R. Fersht, July 9, 2012 (sent for review April 25, 2012)

Aggregation of destabilized mutants of the tumor suppressor p53

is a major route for its loss of activity. In order to assay drugs that
inhibit aggregation of p53, we established the basic kinetics of
aggregation of its core domain, using the mutant Y220C that has
a mutation-induced, druggable cavity. Aggregation monitored by
light scattering followed lag kinetics. Electron microscopy revealed
the formation of small aggregates that subsequently grew to larger amorphous aggregates. The kinetics of aggregation produced
surprising results: progress curves followed either by the binding
of Thioflavin T or the fluorescence of the protein at 340 nm fitted
well to simple two-step sequential first-order lag kinetics with rate
constants k 1 and k 2 that were independent of protein concentration, and not to classical nucleation-growth. We suggest a mechanism of first-order formation of an aggregation competent state as
being rate determining followed by rapid polymerization with the
higher order kinetics. By measuring the inhibition kinetics of k 1
and k 2 , we resolved that the process with the higher rate constant
followed that of the lower. Further, there was only partial inhibition of k 1 and k 2 , which showed two parallel pathways of aggregation, one via a state that requires unfolding of the protein and
the other of partial unfolding with the ligand still bound. Inhibition
kinetics of ligands provides a useful tool for probing an aggregation mechanism.
amyloid misfolding folding cancer

he most frequently mutated gene in cancer is that of the tumor

suppressor p53. Some 70% of types of human cancers have
their p53 directly inactivated by mutation, and so its reactivation
is an important therapeutic goal (1). The most common oncogenic mutations are of residues in the DNA binding domain
(DBD, residues 94312) that make direct contact with DNA
(2). But about 30%40% of oncogenic mutants are inactivated
because the mutations lower the stability of the DBD so that
it denatures at body temperature, although it can be fully active
at lower temperatures (36). We are attempting to rescue those
temperature-sensitive mutants of p53 by designing small molecules that bind specifically to the native state of the protein
and stabilize it (7). The rationale is that small molecules that bind
to the native state of a temperature-sensitive mutant and not the
denatured states will raise the melting temperature by a mass action effect. Such a molecule will slow down the rate of unfolding if
it binds weakly to the transition state for unfolding. The obvious
binding sites to target are those already present that bind natural
ligands. In theory, they can be targeted in a chaperone strategy in
which a rescue drug will compete for a binding site (7). We have
chosen, instead, as a particularly useful paradigm for these studies, the stability of the p53 mutant Y220C, because the mutation
induces a druggable cavity that is remote from functional areas of
the protein (8, 9). We have designed small molecules that raise
the apparent T m of Y220C and slow down the initial rate of
denaturation by a factor of 3 or 4 (8). But the loss of activity of
p53 is more complicated than simple denaturation, and the
further design and refinement of putative leads requires a deeper
understanding of the process and how to target it.
Early cell-based studies found that p53 forms high molecular
weight aggregates in transformed cells and other cells (1012).
1358413589 PNAS August 21, 2012 vol. 109 no. 34

Biophysical studies in vitro found the inactivation of p53 and

its mutants by denaturation involves the initial reversible unfolding of the core domain to give an intermediate, which then aggregates irreversibly by classical nucleation-growth lag kinetics
(3, 4, 6, 13) and gives aggregates of various morphologies, including amorphous, fibrillar, and prionoid, according to the means of
preparation (14, 15). The fluorescence of the single tryptophan
residue, Trp146, in the native state is highly quenched, but in the
denatured state has a strong emission at 356 nm and an aggregated state even stronger at 340 nm (3). The aggregate of wildtype p53 forms reversibly during urea-mediated denaturation
or irreversibly during thermal denaturation or the presence of
EDTA, and more readily with destabilized mutants (3). As destabilized mutants unfold and then aggregate faster than wild type, it
was suggested that the phenomenon of negative dominance of
unstable mutants cotranslated with wild-type protein (16) results
from the denatured mutant nucleating the aggregation of wild
type in mixed hybrids (5). An in-depth study has characterized
nucleation of p53 in cells, identified a nucleation-prone sequence
in its core domain, which also nucleates the aggregation of p63
and p73, and leads to coaggregates in cell lines (17).
Native-state stabilization can inhibit aggregation of p53 (3, 4)
and proteins in general. (18) In order to design and assay novel,
mutant-specific anticancer drugs that can inhibit the aggregation
of Y220C, we have analyzed the kinetics of aggregation of the
mutant and found it fits a very simple scheme. We have used
novel binding ligands (19) as tools to probe the biophysics of
denaturation and aggregation in vitro and the structural requirements for inhibition of aggregation. We have found candidates
for molecules that can reactivate Y220C in a cancer cell line.
Results
All experiments were performed on the core domain of Y220C in
the framework of the stabilized quadruple mutant of p53 (9). It
has an approximate midpoint for denaturation of 44.5 C on rapid
heating by differential scanning calorimetry (8). We measured the
aggregation of Y220C at 37 C by a combination of methods.
Kinetics of Aggregation. Kinetics of aggregation from light scattering.

Aggregation was measured at 37 C, by monitoring light scattering at 500 nm (Fig. 1A). There was an initial lag period, which is
usually taken as indicating nucleation events taking place; a
growth period; and then a leveling off as substrate was depleted.
The lag time shortened with increasing concentrations of p53.
This was also shown by monitoring the absorbance at 360 nm
(Fig. 1C). Light scattering is greatly weighted to the contribution
Author contributions: R.W., G.W., and A.R.F. designed research; R.W. and G.W. performed
research; F.M.B. contributed new reagents/analytic tools; R.W., G.W., and A.R.F. analyzed
data; and A.R.F. wrote the paper.
The authors declare no conflict of interest.
1

R.W. and G.W. contributed equally to this work.

Present address: Department of Medicinal Chemistry, Institute of Pharmacy, EberhardKarls-University Tbingen, Auf der Morgenstelle 8, 72076 Tbingen, Germany.

To whom correspondence should be addressed. E-mail: arf25@cam.ac.uk.

This article contains supporting information online at www.pnas.org/lookup/suppl/

doi:10.1073/pnas.1211550109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1211550109

10

20

30

40

50

60

70

40
20
0

80

20

40

Time (min)
3 M

9 M

D 12

6 M

ThT fluorescence (482 nm)/[p53]

Absorbance (360 nm)

C 1.2

12 M

0.8
0.6
0.4
0.2
0

10

20

30

40

50

Time (min)

60

80

100

120

100

120

Time (min)

60

70

80

10

3 M
6 M
9 M
12 M

8
6
4
2
0

20

40

60

80

Time (min)

Fig. 1. Aggregation kinetics of the core domain of p53Y220C at different

concentrations at 37 C. Panes (A), (B), and (C) monitor aggregation by rightangle light scattering, ThT binding, and absorbance, respectively. (D) Kinetics
monitored by ThT values where the ThT intensity values are normalized to
concentration of p53. Visible particulates appear after the lag phase and
cause random noise in the signals. The time courses for ThT (20 M) binding
were fitted to Eq. 1 with a very small linear drift term (solid lines through
curves). ThT was not present in the scattering experiments, but controls show
it has insignificant effects (Fig. S2B). The y-axes for (A), (B), and (D) are in
arbitrary units, and absorbance units for (C).

of large aggregates, as the magnitude of Rayleigh scattering varies as the size raised to the power of 6, which overemphasizes
the lag.
Kinetics of aggregation from thioflavin T binding. Thioflavin T (ThT)
binds to amyloid fibrils, and can give a good signal that is relatively independent of size and so shows early aggregation events
that precede the formation of large aggregates (14, 20, 21). Indeed, the lag was considerably reduced (Fig. 1B). Normalizing
the ThT signal to the concentration of p53 (Fig. 1D) showed that
the amplitude of the signal was constant, and not dependent on
the size of particles, in contrast to light scattering. Remarkably,
the kinetics fitted to the very simple Scheme 1 of a two-step reaction of sequential first-order reactions (where A is the monomeric protein, B is a monomeric intermediate, and C is a product
that has the ThT-fluorescence signal), which has the solution
Eq. 1 for the formation of C (22)

C
t A
0 k1 k2 k2 expk1 t k1 expk2 tk1 k2 :

[1]
In practice, we fitted the data to:
Ft mk1 k2 k2 expk1 t k1 expk2 tk1 k2 k3 t;
[2]
where F t is the intensity of ThT fluorescence at time t, and m, the
amplitude, A
0 f , where f is the specific fluorescence of ThT

Scheme 1.

Wilcken et al.

Fluorescence spectra and kinetics. Fluorescence spectra at various

time points during aggregation (Fig. 2A) showed apparently only
two major fluorescent species during the process: the native protein (max 305 nm) being lost and the aggregate state (max
340 nm) being formed with an isoemission wavelength of 314 nm
and with the same rate constants (Fig. 2C). The process is not
simple denaturation of Y220C, since the experiments were at
7.5 C below its T m and the fluorescence spectrum of the product
is not that of the denatured state. [The denatured state has max
356 nm (3, 4) and Fig. 2B.] The kinetics of fluorescence 340 nm
also fitted reasonably well to Eq. 1 to give similar values of k2 and
k1 in min 1 : 1 M p53, 0.20  0.01 and 0.084  0.0009; 2 M,
0.47  0.02 and 0.067  0.0006; 3 M, 0.45  0.02 and 0.066
0.0006; and 5 M, 0.61  0.04 and 0.074  0.006. The formation
of large aggregates causes the noise in the latter stages of all the
kinetic runs and a slow decrease in signal (Fig. 2D). The mean

A 4x10

6
0 sec
20
50
81
125
177
224
272
313
358
381
487
594
896
1,198
1,513
1,798
2,101
2,398

3x106

2x106

1x106

0
300

320

340

360

380

400

4x106

3x106

2x106

1x106

300

Wavelength (nm)

305nm

340

360

380

400

D 140

340

120

3.5x10

3.0x106

2.5x106

2.0x106

1.5x106

320

Wavelength (nm)

4.0x106

500

1,000

1,500

2,000

2,500

1
2
3
5

100

M
M
M
M

80
60
40
20
0

Time (sec)

50

100

150

Time (min)

Fig. 2. Changes in fluorescence spectra of p53 with time on incubation at

37 C. (A) Repetitive scans on excitation at 280 nm show an isofluorescence
point at 314 nm and movement to max 340 nm with time. (B) The fluorescence spectrum of denatured p53 Y220C. (C) The loss of native tyrosine
fluorescence at 305 nm paralleled the increase in aggregated state fluorescence at 340 nm during those scans for different concentrations of p53.
(D) The increase in fluorescence monitored at 340 nm was fitted to Eq. 1, with
a linear drift term (solid lines).
PNAS August 21, 2012

vol. 109

no. 34

13585

BIOPHYSICS AND
COMPUTATIONAL BIOLOGY

60

Fluorescence intensity (cps)

50

80

bound to the aggregate. A small linear drift term of k3 t was added

to Eq. 1 to allow for any drift in signal strength because of machine drift or slow settling of particles. k3 was close to zero. Note
that Eq. 1 is unchanged if k1 and k2 are interchanged, so we cannot assign from the curve fitting whether the numerically greater
rate constant precedes the lower or vice versa. We solve the
assignment problem later in this paper from inhibition studies.
The values of k2 and k1 in min 1 derived from the fits to Eq. 2
were, respectively: 3 M p53, 0.32  0.03 and 0.064  0.0016;
6 M, 0.36  0.02 and 0.067  0.001; 9 M, 0.37  0.03 and
0.063  0.0007; and 12 M, 0.37  0.015 and 0.071  0.007
(means and standard errors for single runs). Both rate constants
were virtually independent of protein concentration. The residuals between calculated curves and data are good (Fig. S1). There
may be a low amplitude faster phase that we missed in the analysis, possibly that of the initial unfolding of the protein prior to
slower aggregation events. But the above equation accounts for
the major phases of aggregation. The mean and standard error
for 9 measurements at 3 M protein in the inhibition experiments
below gave 0.314  0.018 and 0.0545  0.0017 min 1 .

Fluorescence 340 nm

100

3 M
6 M
9 M
12 M

Fluorescence intensity (cps)

150

100

Fluorescence intensity (cps)

B120

1 M
2 M
3 M
6 M
9 M
12 M

ThT fluorescence (482 nm)

Scatter (500 nm)

A200

V 0L V 0 max K I K I L
:

Scatter Intensity (500 nm)

The slopes of the inhibition plots plotted against concentrations of ligand fitted saturation curves, with V 0BL being undetectably small. Variation of 5174 (Fig. 3A) gave a K I of 19  4 M
(Fig. 3C). Compound 5201 was particularly effective: the calculated K I 6  1 M, and the lag period for aggregation at
120 M drug was increased to nearly an hour (Fig. 3B and D).
Similar plots were performed for other ligands (Fig. S6).
13586

www.pnas.org/cgi/doi/10.1073/pnas.1211550109

100

200
150
100
50
0

120

10

20

30

50

60

70

80

D 0.07

y = m1 - m1*m0/(M3+M0)
Value
Error
0.060077 0.0027664
5.9043
1.0501
9.3081e-5
NA
0.98754
NA

0.06

m1
m3
Chisq
R

0.05
-2

0.040
0.030

0.04

0.03

0.050

0.02

0.020

0.01

0.010
0.000

40

Time (min)

V (a.u. min )

0.060

20

40

60

80

0.00

100 120 140

20

5174 ( M)

40

60

80

100 120 140

5201 ( M)

Fig. 3. Inhibition of light scattering kinetics by ligands 5174 (A, C) and 5201
(B, D); measurements done with 3 M protein, 37 C, standard buffer.

Thioflavin T binding. The rate of increase of fluorescence of ThT

was inhibited by the ligands and gave curves that could individually be well fitted to Eq. 2 (Fig. S7). The inhibition of k1 (Fig. 4)
and k2 (Fig. 5) fitted to simple binding isotherms that did not tend
to zero as the concentration of ligand increased. We first fitted
the data to Eq. 4 (substituting k for V 0 ) without any constraints.
The curves generated large standard errors (Table 1), so we then
refitted assuming a common value for k1 in the absence of ligand,
the average of all the curves (0.0557  0.002 min 1 ), and for
the value for ligand-bound protein (0.0187  0.0027 min 1 ),
so that the only variable was the constant K I (Fig. 4). The values
thus derived were similar to the free-fitted, but with much better
standard errors (Table 1). These values were very similar to those
measured directly by isothermal titration calorimetry (ITC). (19)
The same was done for k2 using values of (0.319  0.001 min 1 )
0.07

0.07

y = 0.0187 + 0.037*m3/(m3+M0...
Value
Error
m3
15.178
3.8028
Chisq
8.9432e-5
NA
R
0.97041
NA

0.06

y = 0.0187 + 0.037*m3/(m3+M0...
Value
Error
m3
30.685
10.3
Chisq
0.00045383
NA
R
0.90064
NA

0.06

0.05

0.05

0.04
0.03

0.04
0.03

0.02

0.02

0.01

0.01

0.00

0.00
0

50

100

150

50

0.06

0.07
y = 0.0187 + 0.037*m3/(m3+M0...
Value
Error
m3
8.6094
0.99278
Chisq
4.6846e-5
NA
R
0.98089
NA

150

200

250

y = 0.0187 + 0.037*m3/(m3+M0...
Value
Error
m3
4.9834
1.067
Chisq
0.00029042
NA
R
0.91015
NA

0.06
0.05
-1

-1

k1 (min )

0.04

0.03
0.02

0.04
0.03
0.02

0.01
0.00

100

5176 ( M)

5174 ( M)

k (min )

[4]

80

y = m1 - m1*m0/(M3+M0)
Value
Error
0.072172 0.0043558
m1
19.493
4.052
m3
NA
Chisq 0.00023796
0.9741
NA
R

0.05

V 0L V 0BL V 0 max V 0BL K I K I L
:

60

0.070

[3]

More generally, if the ligand-bound protein also aggregates

with an initial rate of V 0BL , then:

40

-1

were monitored at 500 nm to avoid any light absorbance by them.

The addition of 5174 at up to 120 M significantly inhibited
aggregation (Fig. 3A). However, the shape of the curve changed
with increasing concentration, having a sloping phase before
leveling off. Such curves can be fitted to empirical equations. Following the procedure of Wetzel and coworkers (25, 26), we used a
simple model-free method and plotted the initial rates of scattering against t 2 , where t is time [a generally useful plot (27)]. The
rate of addition of monomer to growing oligomers varies as t 2 and
the slope of the plot is V 0 . If a ligand binds to the protein with
dissociation constant K I and the protein-ligand complex is fully
inhibited from aggregation, V 0 depends on the concentration of
unbound protein and is given by:

20

Light scattering. The effects of potential drugs on aggregation

20

k (min )

Kinetic Analysis of Inhibition of Aggregation. We analyzed the inhibition of aggregation by examples from two classes of drug leads:
PK083 (8), a carbazole derivative, and four new leads, PK5174,
PK5176, PK5196, and PK5201 (abbreviated to 83, 5174 etc.),
which are designed to occupy a more extended binding site (19)
(Fig. S5).

40

C 0.080
-2

Electron microscopy studies. During the lag period, we saw only tiny
particles in electron micrographs. These covered much of the
grid, but during the growth period they progressively coalesced
to form amorphous aggregates that grew larger with increasing
time (Fig. S4A, and see electron microscopy studies of inhibition
below, Fig. S4B).

60

0 M
5 M
10 M
120 M

250

Time (min)

-1

characterized by being seeded by already aggregated protein. (24)

The addition of 1 M aggregated Y220C to 2 M fresh p53
(Fig. S3A) had only a small effect on light scattering relative to
that of 2 M fresh protein alone, and the initial rate was similar
to that of 3 M fresh protein. The addition of 1 M aggregated
Y220C to 2 M fresh p53 monitored by ThT fluorescence
(Fig. S3B) gave no detectable change in the kinetics.

80

k1 (min )

Lack of seeding by aggregated Y220C. Nucleation mechanisms are

B 300

0 M
10 M
40 M
120 M

V (a.u. min )

ThT has minimal perturbation of kinetics. The near coincidence of

ThT and 340-nm kinetic data strongly suggests that ThT does not
affect the kinetics significantly. We showed directly that varying
ThT between 520 M has minimal effects on 340 nm-monitored
kinetics (Fig. S2A) and on light scattering (Fig. S2B). The rate
constants for ThT kinetics remained constant within experimental error between 5 and 25 M (Fig. S2C), but the amplitudes
increased, fitting a binding isotherm of 10  0.6 M (Fig. S2D),
indicating a reversible binding of ThT. We discuss further in the
accompanying paper why different techniques give different
kinetics (23).

A 100

Scatter Intensity (500 nm)

rate constants are 0.43  0.08 and 0.073  0.004 min 1 , averaged from the singles runs at each concentration.

0.01

20

40

60

80

5196 ( M)

100

120

0.00

10

20

30

40

50

60

5201 ( M)

Fig. 4. Inhibition of the rate constant k1 in ThT binding kinetics by ligands

fitted to the equation k1 k1BL k1 max k1BL K I K I L
, where k1 max is
constrained to 0.0557 and k1BL to 0.0187 min 1 , respectively.
Wilcken et al.

0.10

0.005

100

0.00

150

50

100

150

200

0.000
0

250

50

0.020

0.40

y = 0.055 + 0.264*m3/(m3+M0)
Value
Error
m3
22.977
5.817
Chisq
0.016302
NA
R
0.8747
NA

0.35

0.15
0.10

0.20

150

200

250

y = m1 + m2*m3/(m3+M0)
Value
Error
0.0010643 0.0017071
0.020446 0.0019925
4.4322
1.3739
2.1663e-5
NA
0.97361
NA

0.010

0.015

0.010

0.15
0.005

0.10

0.05

100

m1
m2
m3
Chisq
R

0.020

-2

0.25

50

0.025

-1

k2 (min )

0.20

5176 ( M)

y = m1 + m2*m3/(m3+M0)
Value
Error
m1 0.00077552 0.00086468
m2
0.016286 0.00092494
m3
12.097
2.3291
Chisq
3.903e-6
NA
R
0.99215
NA

0.015

k1k (min )

0.30

0.000

150

-2

y = 0.055 + 0.264*m3/(m3+M0)
Value
Error
m3
35.075
4.739
Chisq
0.0041041
NA
R
0.95591
NA

100

5174 ( M)

5176 ( M)

0.25

0.005

0.05
0.000
0

20

40

60

80

100

120

0.00

10

5196 ( M)

20

30

40

50

20

40

60

60

80

100

120

0.000

10

20

5196 ( M)

30

40

50

60

5201 ( M)

5201 ( M)

Fig. 5. Inhibition of the rate constant k2 in ThT binding kinetics by ligands

fitted to the equation k2 k2BL k2 max k2BL K I K I L
, where k2 max is
constrained to 0.319 and k2BL to 0.055 min 1 respectively.

and for the value for ligand-bound protein (0.055  0.015 min 1 ,
Fig. 5), to generate values of K I that were, generally, significantly
higher, apart from 5176, where the data were poorer.
The product of the two rate constants k1 k2 also fitted a simple
isotherm. The initial rate of aggregation, according to Eq. 1 as t
tends to 0, is given by: V 0 0.5k1 k2 t 2 A
0 . That equation is analogous to that for the initial rate of the light scatter plots. The
initial rate of increase of ThT fluorescence versus t 2 may be
derived from the product of k1 and k2 measured from the lag
kinetics. A plot of k1 k2 versus [ligand] is analogous to Eqs. 3
and 4 for calculating a K D for inhibition (Fig. 6). The derived
values were close to those obtained from ITC (Table 1). We could
not analyze the data for the binding of 83 because it binds to ThT.
340 nm fluorescence. The kinetics of appearance of aggregate
was followed by fluorescence at 340 nm. The absorbance of the
ligands at the excitation wavelength as well as the emission caused
significant loss of signal at higher concentrations, so the data were
less accurate. But the same trends as for the ThT-monitored
kinetics were observed, and the limited analysis of k1 k2 gave dissociation constants consistent with the other methods (Table 1).
Electron microscopy. 140 M 5174 and 400 M 83 inhibited the
formation and growth of aggregates in good qualitative agreement with the light scattering studies (Fig. S4B).

Discussion

Fig. 6. Inhibition of the rate constant k1 k2 in ThT binding kinetics by ligands

fitted to the equation k1 k2 k1 k2 BL k1 k2 max k1 k2 BL K I K I L
.

kinetics of ThT fluorescence and intrinsic tryptophan fluorescence at 340 nm monitoring the aggregation of p53Y220C fitted
well to two apparent sequential first-order rate constants, 0.3 and
0.06 min 1 , which are independent of concentration in the concentration range measured (112 M protein) (Figs. 1 and 2).
That behavior is strikingly different from standard nucleationgrowth kinetics for protein aggregation, which is usually complex
and involves rate constants for formation of a nucleus and with
higher-order concentration terms for growth. But simple firstorder aggregation kinetics has been observed previously. (2830)
Electron microscopy revealed an initial formation of small particles that increased in size to form an amorphous aggregate over
the same time frame as the spectral changes (Fig. S4), rather than
well-formed fibrils. Neither the individual progress curves nor
their dependence on concentration fit a standard nucleationgrowth model as the rate constants are simple first order and do
not increase with concentration of protein.
We tried fitting it to the simplest model of protein aggregation,
the 2-step mechanism of slow, continuous nucleation followed by
fast growth (Scheme 2), analyzed by Finke and Watzky (31), who
have comprehensively compared it with other models (32).
The analytical solution of Scheme 2 is (31):
Bt A0 k1 k2 A0 1 k1 k2 A0 expk1 k2 A0 t [5]
It can be convenient to eliminate A from the equations, so let
k2 A0 k2app
Bt k2app

Complex Kinetics Fits to a Simple Equation Other Than Classical

Nucleation Growth. p53Y220C core domain aggregates at 37 C,

k1 k2app 1 k1 k2app expk1 k2app t [6]

some 7 C below its temperature of reversible denaturation. The

Table 1. Inhibition of aggregation Y220C measured by different techniques

Inhibitor
PK83
PK5174
PK5176
PK5196
PK5201

K I [M] k1
(ThT) free fit
-*
15
61
6.4
3

9
60
1.7
2

K I [M] k1
(ThT) constrained
-*
15
31
8.6
5

4
10
1
1

K I [M] k2
(ThT) free fit
27
10
42
15

15
11
22
8

K I [M] k2
(ThT) constrained
25
16
35
23

5
8
5
6

K I [M] k1 k2
(ThT)
-*
15
13
13
5.1

1
7
1
0.8

K I [M] k1 k2
(340 nm)

K I [M] V 0
(scatter)

95 30
31 4
37 7
20 4
73

96
19
35
10
6

K D [M] native
state (ITC)

3
4
4
1
1

125
16
21
10
8

*PK83 interacts with Thioflavin T, which masks its binding. ITC measured at 20 C. Stoichiometry 1 mol ligandmol protein.

Wilcken et al.

PNAS August 21, 2012

vol. 109

no. 34

13587

BIOPHYSICS AND
COMPUTATIONAL BIOLOGY

-1

0.010

50

0.30

0.015

0.005

k1k (min )

0.35

k (min )

0.010

0.05

5174 ( M)

0.00

0.020

-2

0.15
0.10

0.05
0.00

0.20

0.15

0.25

y = m1 + m2*m3/(m3+M0)
Value
Error
0.0022457 0.0015148
0.012686
0.002032
7.9331
6.8441
7.9769e-5
NA
0.90448
NA

m1
m2
m3
Chisq
R

-1

k (min )

-1

k2 (min )

0.20

0.015

-2

0.30

0.25

0.025

y = m1 + m2*m3/(m3+M0)
Value
Error
m1 0.00078618 0.0005601
m2
0.016452 0.00065108
m3
11.946
2.2454
Chisq
1.0203e-6
NA
R
0.99812
NA

0.30

0.020

y = 0.055 + 0.264*m3/(m3+M0)
Value
Error
m3
16.032
7.9992
Chisq
0.0443
NA
R
0.73386
NA

0.35

k1k (min )

0.40

y = 0.055 + 0.264*m3/(m3+M0)
Value
Error
m3
25.394
4.6872
Chisq
0.0035619
NA
R
0.9703
NA

k1k (min )

0.35

Scheme 2.

The aggregation kinetics of p53 Y220C monitored by ThT

fluorescence appears to fit to Eq. 6 at each individual concentration (Fig. S8A), but not as well as for the simple 2-step lag
kinetics, as shown from the residuals to the fits (Fig. S8B). We
can rule out conclusively the FinkeWatzky mechanism from
the concentration dependence of the rate constants: the term
k2app ; k2 A0 , in Eq. 6 should increase linearly from 0 with
the concentration of A0 , but is independent of concentration
from 3 to 12 M protein (Fig. S8D). There is also no significant
seeding of aggregation (Fig. S3).
Vitalis and Pappu (27) have analyzed the significance of slopes
of initial rates of scattering, etc., versus t 2 , V 0 , versus concentration of monomer. The slope for homogeneous nucleation with n
molecules in the nucleus should be n 2. The measured slope
of 1 (Fig. S9) is indicative of either a heterogeneous distribution
of nuclei or of a secondary process that we have identified to be
a first-order conformational transition to a species that is capable of supporting polymerization. And more importantly, the
data and the inferred rate constants are consistent with such a
unimolecular process.
Kinetics of Inhibition. The basic conclusion is that ligands that bind
to the mutation-induced cavity in Y220C inhibit its aggregation,
with initial rates being as expected from the dissociation constants. In addition, the kinetics of inhibition has provided invaluable mechanistic information. The rate constants k1 and k2 for
ThT binding are inhibited by compounds 5174, 5176, 5196,
and 5201 such that each inhibition profile fits a simple binding
isotherm with a finite limiting value at saturating concentrations
of ligand. The lack of complete inhibition could result from there
being two populations of protein, one of which does not bind the
ligand. But ITC shows a stoichiometry of 11 ligand bound to
protein. (19) The remaining alternative is that ligand-bound protein can still aggregate, albeit at a much reduced rate.
Basic Aggregation Mechanism. The first-order kinetics implies that
the rate determining steps in the process are the first-order formation of an aggregation-competent state that can rapidly polymerize so that the higher-order steps are after the rate-determining
ones. We have analyzed simple schemes (SI Text, Schemes S1S3)
to see if they could account for the inhibition behavior and the
first-order sequential kinetics. The simplest is given in Scheme 3.
The ligands can bind to states A and B, and the A.L and
B.L states also aggregate via an aggregation-competent state C,
which may possibly also polymerize when bound to L. So p53
can aggregate by complete unfolding, which is inhibited by the
binding of ligands, or by partial unfolding with the ligand still
bound.
Order of Rate Constants. Eq. 3 for sequential kinetics is symmet-

rical with respect to exchange of k1 and k2 , and so unless the concentration of B is also measured, we do not know whether the
aggregation

A
KA

k2 competent
fast
C polymerisation

k1
B
L

A.L

KB

k1A.L

k2B.L

B.L

Scheme 3.

13588

www.pnas.org/cgi/doi/10.1073/pnas.1211550109

higher rate constant precedes the lower or vice versa. But the
values of K A and K B can be used to deconvolute the order:
K A should be the same as the independently measured value
of the dissociation constant from the protein. As seen in Table 1,
it is the lower rate constant that is inhibited, with a value of K A
similar to that measured by ITC, and so represents the first process. Accordingly, k1 , k1A:L , k2 , k1B:L in Scheme 3, are 0.0557 
0.002, 0.0187  0.0027, 0.319  0.001, 0.055  0.015 min 1 ,
respectively.
In the accompanying paper, (23) we explore the consequences
of the proposed mechanism for the aggregation of the core
domain of Y220C and show the mechanism extends to the fulllength protein.
Methods
Chemical Compounds. PK083 (83, EN30014607) was purchased from Enamine and >95% pure. PK5174, PK5176, PK5196 and PK5201 were synthesized
within the framework of a custom synthesis contract by Roowin S.A. For all
compounds, compound identity and >95% purity were guaranteed by the
supplier.
Protein Expression and Purification. The thermostabilized Y220C protein was
expressed and purified as described (8).
Isothermal Titration Calorimetry (ITC). ITC experiments were conducted using
a MicroCal iTC200 calorimeter. Protein samples used in the cell unit were prepared to a final concentration of 50200 M in 25 mM potassium phosphate,
pH 7.2, 150 mM NaCl, 1 mM tris(2-carboxyethyl)phosphine hydrochloride
(TCEP) in 5% (vv) DMSO. Compounds for use in the syringe unit were dissolved in the same buffer at 5% (vv) DMSO. Measurements were performed
at 20 C using injection steps of 2 L at 0.5 Ls (initial injection: 0.5 L) and
120 s spacing. Data analysis was performed using MicroCal Origin software.
Kinetics of Aggregation. Light scattering. To measure the effect of compounds
on the kinetics of aggregation of Y220C, we monitored protein aggregation
by measuring light scattering at 37 C at 500 nm as excitation and emission
wavelengths (excitation slit width 0.8 nm, emission slit width 2 nm), using a
Horiba FluoroMax-3 spectrophotometer. Experiments were generally performed with a protein concentration of 3 M in 25 mM potassium or sodium
phosphate, pH 7.2, 150 mM NaCl, 5 mM DTT or 1 mM TCEP, and 5% DMSO.
The kinetics at the single wavelength of 340 nm was measured using the
same setup as light scattering above, but excitation and emission wavelengths were set to 285 nm and 340 nm, respectively, with slit widths set
to 3 nm (excitation) and 4 nm (emission). To obtain homogenous mixing,
Scienceware (Bel-Art Products) spectrometer cell spinbars were used. Effects
of aggregation seeds were minimized using disposable Fisher Scientific UV
grade PMMA cuvettes and storing the spinbars in nitric acid when not in
use. Data analysis was performed using KaleidaGraph (Synergy Software).

Thioflavin T assays. As for detecting other amyloid aggregates, we measured

the ThT fluorescence at 482 nm upon excitation at 450 nm (33) (excitation/
emission slits are 3 nm4 nm) using a Horiba FluoroMax-3 Spectrofluorometer. Time-resolved fluorescence was recorded immediately after adding
3 M Y220C to pre-equilibrated buffer (25 mM sodium phosphate, pH 7.2,
150 mM NaCl, 1 mM TCEP, 5% DMSO, 20 M ThT).
Time-resolved fluorescence spectra. The intrinsic fluorescence spectra of
Y220C (excited at 280 nm, emission between 300 nm and 500 nm) were recorded with a Horiba FluoroMax-4 Spectrofluorometer immediately after the
addition of protein to pre-equilibrated buffer (25 mM sodium phosphate,
pH 7.2, 150 mM NaCl, 1 mM TCEP, 5% DMSO). Both the slits for excitation
and emission were 5 nm.
Transmission Electron Microscopy. Y220C (3 M) was incubated at 37 C with or
without the compounds in 25 mM Tris-HCl, pH 7.2, 150 mM NaCl, 1 mM TCEP,
and 5% DMSO. Samples were taken at different time points. The samples
(5 L) were adsorbed to freshly glow-discharged formvar or carbon film
400 mesh copper grids, rinsed with deionised water, and stained with 1%
uranyl acetate or sodium phosphotungstate. Images were taken using a
Phillips 208S transmission electron microscope at 80 kV.
ACKNOWLEDGMENTS. This work was funded by an ERC Advanced Grant
to A.R.F.

Wilcken et al.

19. Wilcken R, et al. (2012) Halogen-enriched fragment libraries as leads for drug rescue of
mutant p53. J. Am. Chem. Soc 134:68106818.
20. Manno M, et al. (2007) Kinetics of different processes in human insulin amyloid
formation. J Mol Biol 366:258274.
21. Meuvis J, Gerard M, Desender L, Baekelandt V, Engelborghs Y (2010) The conformation and the aggregation kinetics of alpha-synuclein depend on the proline residues
in its C-terminal region. Biochemistry 49:93459352.
22. Fersht A (1999) Structure and Mechanism in Protein Science: A Guide to Enzyme
Catalysis and Protein Folding (W. H. Freeman, New York).
23. Wang G, Fersht AR (2012) First-order rate-determining aggregation mechanism of
p53 and its implications. Proc Natl Acad Sci USA, 10.1073/pnas.1211557109.
24. Frieden C (2007) Protein aggregation processes: In search of the mechanism. Protein
Sci 16:23342344.
25. Chen S, Ferrone FA, Wetzel R (2002) Huntingtons disease age-of-onset linked to
polyglutamine aggregation nucleation. Proc Natl Acad Sci USA 99:1188411889.
26. Bhattacharyya AM, Thakur AK, Wetzel R (2005) polyglutamine aggregation nucleation: Thermodynamics of a highly unfavorable protein folding reaction. Proc Natl
Acad Sci USA 102:1540015405.
27. Vitalis A, Pappu RV (2011) Assessing the contribution of heterogeneous distributions
of oligomers to aggregation mechanisms of polyglutamine peptides. Biophys Chem
159:1423.
28. Kendrick BS, et al. (1998) Aggregation of recombinant human interferon gamma:
Kinetics and structural transitions. J Pharm Sci 87:10691076.
29. Shukla AA, Gupta P, Han X (2007) Protein aggregation kineticsduring Protein A chromatography. Case study for an Fc fusion protein. J Chromatogr A 1171:2228.
30. Weijers M, Barneveld PA, Cohen Stuart MA, Visschers RW (2003) Heat-induced
denaturation and aggregation of ovalbumin at neutral pH described by irreversible
first-order kinetics. Protein Sci 12:26932703.
31. Morris AM, Watzky MA, Agar JN, Finke RG (2008) Fitting neurological protein aggregation kinetic data via a 2-step, minimal/Ockhams razor model: The FinkeWatzky
mechanism of nucleation followed by autocatalytic surface growth. Biochemistry
47:24132427.
32. Morris AM, Watzky MA, Finke RG (2009) Protein aggregation kinetics, mechanism, and
curve-fitting: A review of the literature. Biochim Biophys Acta 1794:375397.
33. Solomon B, Koppel R, Hanan E, Katzav T (1996) Monoclonal antibodies inhibit in vitro
fibrillar aggregation of the Alzheimer beta-amyloid peptide. Proc Natl Acad Sci USA
93:452455.

BIOPHYSICS AND
COMPUTATIONAL BIOLOGY

1. Brown CJ, Lain S, Verma CS, Fersht AR, Lane DP (2009) Awakening guardian angels:
Drugging the p53 pathway. Nat Rev Cancer 9:862873.
2. Cho Y, Gorina S, Jeffrey PD, Pavletich NP (1994) Crystal structure of a p53 tumor suppressor-DNA complex: Understanding tumorigenic mutations. Science 265:346355.
3. Bullock AN, et al. (1997) Thermodynamic stability of wild-type and mutant p53 core
domain. Proc Natl Acad Sci USA 94:1433814342.
4. Bullock AN, Henckel J, Fersht AR (2000) Quantitative analysis of residual folding and
DNA binding in mutant p53 core domain: Definition of mutant states for rescue in
cancer therapy. Oncogene 19:12451256.
5. Bullock AN, Fersht AR (2001) Rescuing the function of mutant p53. Nat Rev Cancer
1:6876.
6. Friedler A, Veprintsev DB, Hansson LO, Fersht AR (2003) Kinetic instability of p53
core domain mutants: Implications for rescue by small molecules. J Biol Chem
278:2410824112.
7. Friedler A, et al. (2002) A peptide that binds and stabilizes p53 core domain: Chaperone strategy for rescue of oncogenic mutants. Proc Natl Acad Sci USA 99:937942.
8. Boeckler FM, et al. (2008) Targeted rescue of a destabilized mutant of p53 by an in
silico screened drug. Proc Natl Acad Sci USA 105:1036010365.
9. Joerger AC, Ang HC, Fersht AR (2006) Structural basis for understanding oncogenic
p53 mutations and designing rescue drugs. Proc Natl Acad Sci USA 103:1505615061.
10. Kraiss S, Spiess S, Reihsaus E, Montenarh M (1991) Correlation of metabolic stability
and altered quaternary structure of oncoprotein p53 with cell transformation. Exp
Cell Res 192:157164.
11. Kraiss S, Quaiser A, Oren M, Montenarh M (1988) Oligomerization of oncoprotein
p53. J Virol 62:47374744.
12. McCormick F, Clark R, Harlow E, Tjian R (1981) SV40 T antigen binds specifically to a
cellular 53 K protein in vitro. Nature 292:6365.
13. Butler JS, Loh SN (2003) Structure, function, and aggregation of the zinc-free form of
the p53 DNA binding domain. Biochemistry 42:23962403.
14. Ishimaru D, et al. (2003) Fibrillar aggregates of the tumor suppressor p53 core domain.
Biochemistry 42:90229027.
15. Ano Bom AP, et al. (2012) Mutant p53 aggregates into prion-like amyloid oligomers
and fibrils: Implications for cancer. J Biol Chem.
16. Milner J, Medcalf EA (1991) Cotranslation of activated mutant p53 with wild type
drives the wild-type p53 protein into the mutant conformation. Cell 65:765774.
17. Xu J, et al. (2011) Gain of function of mutant p53 by coaggregation with multiple
tumor suppressors. Nat Chem Biol 7:285295.
18. Johnson SM, et al. (2005) Native state kinetic stabilization as a strategy to ameliorate
protein misfolding diseases: A focus on the transthyretin amyloidoses. Acc Chem Res
38:911921.

Wilcken et al.

PNAS August 21, 2012

vol. 109

no. 34

13589