Supplementary Information for “Interplay between partner and ligand in
the binding of an intrinsically disordered protein”
Supplementary Methods
Protein expression and purification
Protein expression and purification of MCL-1 was carried out as described elsewhere (1).
PUMA peptide, 35 amino acids, residues 127-161 (Uniprot Q99ML1) with the mutation
M144A was used as the wild-type peptide in this study, similar to M144I pseudo wild-type
used in the NMR structure (2) and the 34 amino acid peptide used in the previous kinetic
characterization (1). Unlike the M144I peptide (1), M144A was monomeric at all
concentrations studied (3). Wild-type and mutant peptide were expressed as GB1 fusion
proteins (1). Protein expression was carried out in an identical manner to MCL-1 (1),
2+
however, after binding to Ni -agarose resin, the resin was additionally washed with 50 mM
Tris pH 7.4, 100 mM NaCl, then 50 mM Tris pH 7.4, 100 mM NaCl, 5mM CaCl2. Factor Xa
(New England Biolabs) was added to release the peptide from the resin. The supernatant was
then buffer exchanged, using desalting HiTrap (GE Healthcare), into 10 mM Tris pH 7.0 and
bound to HiTrap Q. A linear gradient of NaCl was applied and peptide eluted at 7-10% 1 M
NaCl. PUMA was further purified using gel filtration Superdex G30 (GE Healthcare)
equilibrated in the experimental phosphate buffer. Peptides were filtered before N2(l), freezing
and storage at -80 ˚C. Peptide concentration was determined using an extinction coefficient
determined by amino acid analysis.
Buffers
All experiments were carried out in 50 mM sodium phosphate pH 7.0.
Circular dichroism
CD spectra were taken using Chirascan (Applied Photophysics), using a 2 mm path length
cuvette. Estimates for the percentage helicity (%helix) were calculated using the average
mean residue ellipticity at 222 nm, obtained using three concentrations of peptide, and Eq. 3
(4).
%Helix = 100 / (1+ ((Ellipticity − Ellipticityhelix ) / (Ellipticitycoil − Ellipticity))) [3]
2
-1
A value of -725 deg cm dmol was used for 100% random coil (Ellipticitycoil ) and a value of –
2
-1
34100 deg cm dmol was used for 100% helix in a 35 amino acid peptide (Ellipticityhelix) (4).
Association kinetics Association was monitored using an SX18 or SX20 fluorescence stopped-flow spectrometer
(Applied Photophysics) at 25˚C. Excitation at 280 nm and emission recorded above 305 nm.
Proteins were mixed with PUMA in excess (minimum x10 fold) to set up pseudo-first-order
conditions (5). PUMA was used in excess as, with fewer fluorescent tryptophans, greater
signal to noise ratio was achieved. Further, as the yields of PUMA were generally larger,
higher concentrations could be used without concentrating (which could lead to aggregation if
MCL-1 was used). A minimum of six kinetics traces were averaged and then fit to a single
exponential function, Eq. 4,
F = F∞ +∆ F exp(−kobs t)
[4]
where F is fluorescence at time t, F ∞ the final fluorescence, ∆F is the fluorescence amplitude
of the reaction, kobs is the apparent rate constant. kobs can be interpreted using Eq. 5.
kobs = k− + k+ [A]0
[5]
where k- is the dissociation rate constant, k+ the association rate constant and [A]0 the
concentration of the protein in excess.
Dissociation kinetics
Dissociation kinetics were followed by pre-forming the PUMA peptide MCL-1 complex at 2.5-5
µM and manual mixing with a solution containing an excess (30-1000 fold) concentration of
the peptide PUMA W133F N149A (M144A) which does not show any fluorescence change
upon binding MCL-1. Fluorescence was followed using Cary Eclipse Fluorimeter (Aligent
Technologies) at 25 ˚C. Data were fit to Eq. 4 with kobs=k-. With sufficient excess of PUMA
W133F N149A (M144A) the fit kobs became essentially independent of the concentration of
this peptide. Only k- in this regime were included in the average k-.
Fitting, Figures and Errors
All fitting was performed using ProFit (Quantum Soft). Figures were prepared using ProFit
and Pymol. Errors were propagated using standard equations.
To assess the error in the concentration dependent pseudo-first-order experiments, a number
of controls were carried out.
(1) We found that error in k+ propagated from the fitting of the individual kinetic traces and the
scatter of the pseudo-first-order plot (from the biological and technical repeats) was very
small (under 7%). We decided that the greatest source of error would come from the
concentration determination itself - using UV absorbance, of which the extinction coefficient is
the greatest source of potential error. This has the potential to lead to systematic deviations in
the observed kobs values making the minimal scatter on the pseudo-first-order plot misleading
as to the true error. To determine the extinction coefficient, on a number of occasions,
absorbance at 280 nm for wild-type and mutant PUMA stocks was recorded and samples
sent for amino acid analysis (aaa) (considered the gold-standard for protein concentration
determination). As can be seen in the figure S11A, there is good agreement between the
absorbance and aaa values. The standard deviation of the calculated extinction coefficients
was 7%. This is by far the largest source of potential error in the k+ determination (potential
as the scatter on Fig S11A may be due to the errors in aaa itself). As any difference between
mutants/stocks could produce a systematic error in the pseudo-first order plot this 7% error
was directly applied to the k+ value obtained.
(2) We used a number of biological repeat experiments utilizing new batches of PUMA (the
concentration of which is critical to the experiment) and repeating the stopped-flow
experiments. This was carried out a number of times for the wild-type proteins over a period
of 12 months and twice for a number of mutant proteins. As Fig S11B shows, there is good
agreement between these biological repeats, well within the 7% error quoted in Table S1.
(3) We also carried out a number of “technical repeats” i.e. repeated measurements for the
same protein and peptide samples, at the same and at different concentrations of peptide.
(4) Finally, although most experiments were carried out with PUMA in excess (see above),
repeating the pseudo-first-order experiments using MCL-1 in excess produced the same k+
values, within the 7% error (Fig. S11C). The fact that these approaches gave the same rate
constants also gave us confidence in our analysis of the mechanism.
.
Errors and Φ-values:
There has been much discussion (and some controversy) about the accuracy of Φ-values,
and in particular how this is affected by the change in overall stability (6, 7). It has even been
suggested that Φ-values will only be accurate if they have been obtained from mutations with
-1
-1
∆∆G ≥ 1.67 kcal mol (7 kJ mol ) (6). We note that only three mutations here satisfy this
extreme condition (L141A, L148A and Y152A) and one is borderline (I137A, ∆∆G = 1.61 ±
-1
0.08 kcal mol ) (all marked * in Fig. 2B). However, most obtained Φ-values fall within the
-1
more generally accepted limit of ∆∆G ≥ 0.6 kcal mol (7), and the distribution and magnitude
-1
of Φ-values with 0.38 < ∆∆G < 1.67 kcal mol (propagated error in Φ < 0.2) are in excellent
-1
agreement with those that satisfy ∆∆G ≥ 1.67 kcal mol .
Molecular simulations
To complement the experimental results, in this work we carry out molecular simulations of
the binding of MCL-1 to PUMA using a coarse-grained, structure based model. We first
observe binding events in unbiased simulations starting from the unbound state of the
complex. Then, biased simulations were run using a restraining potential at different values of
the progress binding coordinate. This allows for an accurate calculation of free energy
surfaces and sampling at the top of the binding barrier.
Coarse-grained model
As the starting point for our Gō model of the PUMA-MCL-1 complex, we take as our reference
structure the most representative conformer from the NMR ensemble of the PUMA-MCL-1
complex (PDB entry 2ROC) (8). To make this reference structure as similar as possible to the
protein used in the experimental work, we introduce two mutations (I144A and M156R) and
add missing residues at the N (127R, 128V, 129E) and C (157Q, 158E, 159E, 160Q, 161H)
termini, without adding any additional native contacts. These changes were introduced using
the UCSF Chimera software (9).
We use the Karanicolas-Brooks topology based (i.e. Gō model) model to run the simulations
(10). The model is coarse-grained to a single bead per residue at the position of the alpha
carbons. The energy function contains harmonic terms for bonds and angles, a statistical
dihedral term based on torsional preferences in experimental structures, and non-bonded
interactions of the form
+ $ '12
$ σ '10 $ σ ' 6 .
σ
ij
ij
ij
Vij = ε ij -13&& )) −18&& )) + 4&& )) 0 ,
r
r
-, % rij (
% ij (
% ij ( 0/
[6]
where εij is the pairwise contact energy between amino-acids i and j, σij is the distance at
which the energy between beads ij is at a minimum, and rij is the instantaneous distance
between the residues.
€
In order to match a set of available experimental references, interaction strengths were
optimized independently for intramolecular and intermolecular interactions for MCL-1 and
PUMA. First, the intramolecular interactions of MCL-1 were used to produce a melting
temperature (Tm = 355 K) comparable to that obtained in the CD experiments. Second, we
tuned the internal contacts of PUMA in order to match the residual helicity at room
temperature. The fraction helix was calculated from the number of consecutive torsions
between 40 and 130º, as described before (11). For a residual helix to be considered we
require at least 4 consecutive residues (i.e. at least a helical turn) to be in the defined range.
Finally, intermolecular contacts were scaled in order to reach a dissociation constant (Kd)
comparable to that found in experiments.
Langevin dynamics simulations
Dynamics of the Gō model described above were propagated using the Langevin equation
-1
(12), as implemented in the CHARMM package (13), with a friction coefficient of 0.2 ps .
Simulations were performed in a cubic periodic box of side 100 Å, with a cutoff of 30 Å used
for the calculation of nonbonded interactions. Constraints were applied to all bonds using
SHAKE (14), and all simulations were performed at 298 K, using a timestep of 10 fs.
We run 10 unbiased dynamics simulations summing a total of 4 µs simulation time. Each
equilibrium simulation was started from a different configuration, all corresponding to unbound
PUMA. As a progress variable for the binding process we calculate the fraction of native
contacts (Q) for PUMA-MCL-1. We also monitor Q for intramolecular contacts within PUMA
and MCL-1 independently. In the 10 simulations which we observe a total of three binding
events, of which we report a binding transition path.
In order to obtain accurate potential of mean force estimates were made using restraining
2
(umbrella) potentials of the form V(Q) = kQ (Q - Q0) /2, where Q0 is the target value of the
fraction of native contacts in each simulation and kQ is the force constant of the restraining
potential. We use 16 windows of Q0 to ensure sufficient sampling of all Q values between 0
and 1. We start the simulations from unbound PUMA to avoid large forces in the initial
configurations.
Analysis of the simulations
The weighted histogram analysis method (WHAM) (15) was used to combine the results from
different umbrella windows to produce 1D and 2D potentials of mean force. Kd values were
estimated from these PMFs as
Kd =
pu
2
[Protein]
pb
[7] where [Protein] is the protein concentration, which can be readily calculated from the size of
the simulation box, and pu and pb are the populations of bound and unbound PUMA-MCL-1,
which we obtain from the potential of mean force.
PCA of MCL-1
We analyse the ensemble of 20 NMR structures of the isolated form of MCL-1 protein (PDB
ID: 1wsx (16)) using the principal component analysis (PCA) method (17). PCA is based on
the analysis of the covariance matrix C, which has dimensions of 3Nx3N for a protein of N
residues. C can be written in terms of NxN submatrices of dimension 3x3
∆𝑥𝑖 ∆𝑥𝑗
𝐂𝑖𝑗 =
∆𝑦𝑖 ∆𝑥𝑗
∆𝑧𝑖 ∆𝑥𝑗
∆𝑥𝑖 ∆𝑦𝑗
∆𝑥𝑖 ∆𝑧𝑗
∆𝑧𝑖 ∆𝑦𝑗
∆𝑧𝑖 ∆𝑧𝑗
∆𝑦𝑖 ∆𝑦𝑗
∆𝑦𝑖 ∆𝑧𝑗
[8]
where each of the terms <ΔxiΔyj> is the cross-correlation between the X and Y components
of the fluctuation vector ΔR for i and j residues, respectively, which represents deviations
from the mean position. The trace of Cij therefore gives correlations between the fluctuations
of residues i and j. These can be straightforwardly calculated from an ensemble of M (in this
case 20) structures. Principal modes were calculated from decomposing the covariance
matrix. This analysis was conducted using the software package ProDy (18). This procedure
results in a description of the dominant changes in the structural ensemble. Only C-alpha
atoms were used for the calculation of covariance matrices.
Supplementary Figures
A
B
kobs
(s-1)
100 W133F
80
60
40
20
I137A
100 L148A
80
60
40
20
Y152A
0
C
10–5
L141A
[PUMA] (M)
1
0.1
0.01
kobs
(s-1)
0.001
1
101
102
101
102
103
0.1
0.01
0.001
103 Excess out-competing
peptide
Figure S1. A) Structure of PUMA (cartoon, blue) and MCL-1 (surface, white), with
large hydrophobic residues mutated in this study shown as spheres; W133 (red), I137
(green), L141 (magenta), L148 (cyan) and Y152 (orange). B) Observed rate
constants for association of PUMA and MCL-1, obtained using pseudo-first-order
conditions with PUMA in excess. In each frame, kobs for wild-type is shown in blue.
The gradient of each linear fit corresponds to k+. C) Observed rate constants for
dissociation of PUMA from MCL-1 induced by adding an excess of an out-competing
peptide. Average of kobs corresponds to k-.
A
B
kobs
(s-1)
100 E132
80
60
40
20
E136
D147
A150
100
80
60
40
20
0
C
10–5
A139
R143
[PUMA]R154
(M)
E158
10–5
[PUMA] (M)
1
0.1
0.01
kobs
(s-1)
0.001
1
101
102
101
102
103
0.1
0.01
0.001
103 Excess out-competing
101
102
103
peptide
Figure S2. A) Structure of PUMA (cartoon, blue) and MCL-1 (surface, white), with
residues chosen for alanine glycine mutations coloured; E132 (red), E136 (green),
A139 (yellow), R143 (magenta), D147 (cyan), A150 (beige), R154 (orange) and E158
(dark green, not visible in the structure). B) Observed rate constants for association
of PUMA and MCL-1, obtained using pseudo-first-order conditions with PUMA in
excess. In each frame, kobs for wild-type is shown in blue, kobs for alanine shown in
white and kobs for glycine colored as for Fig. S2A. The gradient of each linear fit
corresponds to k+. C) Observed rate constants for dissociation of wild-type and
mutant PUMA from MCL-1 induced by adding an excess of an out-competing
peptide. Average of kobs corresponds to k-. Colour scheme identical to Fig. S2B. No
data was collected for R154G (Methods).
0
E132A
E136A
E136
A150
R143A
R143
M.R.E
(deg cm2 dmol-1)
–5000
–10000
–15000
0
D147A
R154A
200 220 240
200 220 240
E158A
–5000
–10000
–15000
200 220 240 260
Wavelength (nm)
0
E132G
E136G
E136
A150
D147G
A150G
200 220 240
200 220 240
A139G
A139
R143G
R143
M.R.E
(deg cm2 dmol-1)
–5000
–10000
–15000
0
E158G
–5000
–10000
–15000
200 220 240 260
Wavelength (nm)
0
W133F
I137A
E136
A150
L141A
A139
L148A
R143
M.R.E
(deg cm2 dmol-1)
–5000
–10000
–15000
0
Y152A
–5000
–10000
A139G+A150G
–15000
200 220 240
200 220 240
Wavelength (nm)
Figure S3. Circular Dichroism (CD) spectra for wild-type (blue) and mutant (black)
PUMA peptides in the absence of MCL-1. The negative M.R.E (mean residue
ellipticity) peak at 222 nm was used to calculate the per cent helical content of each
peptide.
Overall helicity (%)
30
20
10
W133F
I137A
L141A
L148A
Y152A
E132G
E136G
A139G
R143G
D147G
A150G
E158G
A139G+A150G
WId–type
E132A
E136A
R143A
D147A
R154A
E158A
0
PUMA peptide
Figure S4. Calculated per cent helical content for each PUMA peptide determined
using the mean residue ellipticity at 222 nm. Mutation to glycine causes a drop in
overall helical content relative to the corresponding alanine mutation.
G (kcal mol-1)
5
4
3
2
1
0
1.0
0.5
W133F
I137A
L141A
L148A
Y152A
A132G
A136G
A139G
A143G
A147G
A150G
A158G
E132A
E136A
R143A
D147A
R154A
E158A
0
PUMA mutant
Figure S5. Free energy destabilization (of binding MCL-1) caused by PUMA mutation
(top) and the calculated Φ-value (bottom). Shown in black are the values for A150G
in the presence of A139G and vice versa. Shown in grey are the Φ-values discarded
due to their standard deviations being greater than 0.2.
A
error
B
2.0
1.5
error
1.0
1.5
1.0
0.5
0.5
0.0
0.0
0
C
2.0
1
2
3
G (kcal mol-1)
0
4
D
1.5
1.0
1.0
0.5
0.5
0.0
0.0
1
2
3
3
4
2.0
1.5
G (kcal mol-1)
2
G (kcal mol-1)
2.0
0
1
4
0
1
2
3
G (kcal mol-1)
4
Figure S6.. Free energy destabilization of binding and the errors in Φ-values for
PUMA mutants (A) and MCL-1 mutants (B). Lower ∆∆G generally leads to a larger
error in Φ, errors lower than 0.2 shown in red, vertical line shows corresponding ∆G
-1
cut off (0.38 kcal mol ). Φ-values and ∆∆G for conservative PUMA mutants (C) and
MCL-1 mutants (D), Φ-values over the cut off are shown in blue.
A
B
C
D
(s-1) 0.1
kobs 1
(s-1) 0.1
0.01
0.01
kobs
1
0.001
0.001
101
102
103
101
Excess out-competing
peptide
E
Time (min)
-10 0 10 20 30 40 50 60 70 80 90 100110120130140
0.00
F
F300A
-0.15
103
Excess out-competing
peptide
-0.05
-0.10
102
Time (min)
-10 0 10 20 30 40 50 60 70 80 90 100110120130140
0.0
D237A
-0.2
-0.20
-0.25
-0.30
0
-0.4
2
-2
-2
-4
-4
0
-6
-6
-8
-8
-10
-10
-12
-14
-12
-16
-14
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
0.5
Molar Ratio
G
Time (min)
0.05
1.0
1.5
2.0
2.5
Molar Ratio
-10 0 10 20 30 40 50 60 70 80 90 100110120130140
0.00
H
Time (min)
0.05
-10 0 10 20 30 40 50 60 70 80 90 100110120130140
0.00
-0.05
-0.05
-0.10
-0.10
V197A
-0.15
V201A
-0.15
-0.20
-0.20
-0.25
-0.25
-0.30
2
0
0
-2
-2
-4
-4
-6
-6
-8
-8
-10
-12
-10
-14
-12
-16
0.0
0.5
1.0
1.5
Molar Ratio
2.0
2.5
3.0
0.0
0.5
1.0
1.5
2.0
2.5
Molar Ratio
Figure S7. A) Location of MCL-1 residues whose mutation to alanine leads to slower
association. V197 (light green), V201 (brown), L216 (red), V230 (dark green), R244
(magenta), F300 (yellow) and V302 (orange). B) Location of MCL-1 residues whose mutation
to alanine leads to faster association. H205 (light green), M212 (brown), H233 (dark green),
V234 (magenta), D237 (yellow) and T247 (red). C) Observed rate constants for dissociation
of PUMA from wild-type (blue) and mutants (color scheme identical to Fig S7A) MCL-1
induced by adding an excess of an out-competing peptide. Average of kobs corresponds to k-.
D) Dissociation rate constants (color scheme identical to Fig S7B). E) ITC trace for wild-type
PUMA binding MCL-1 mutant F300A. Trace for D237A (F), V197A (G), V201A (H).
G (kcal mol-1)
5
4
3
2
1
1.0
V197A
V201A
H205A
M212A
L216A
V230A
H233A
V234A
D237A
T247A
F300A
V302A
R244A
0
0.5
0
V197A
V201A
H205A
M212A
L216A
V230A
H233A
V234A
D237A
T247A
F300A
V302A
R244A
–0.5
MCL-1 mutant
Figure S8.. Free energy destabilization (of binding PUMA) caused by MCL-1 mutation
(top) and the calculated Φ-value (bottom). Shown in grey are the Φ-values discarded
due to their standard deviations being greater than 0.2.
A
B
C
Figure S9. Structural changes in MCL-1 upon binding. A) Cartoon of unbound MCL-1
(white), based on pdb 1WSX. Residues which upon mutation to alanine speed up
association shown as red spheres. Residues which upon mutation to alanine slow
down (or have no effect on) association shown as green spheres. B) Cartoon of
bound MCL-1 (with PUMA removed), based on pdb 2ROC. C) Cartoon of bound
MCL-1 (with PUMA shown as blue cartoon) based on pdb 2ROC.
th
Figure S10. Graphical representation of the 5 mode of the PCA analysis of 1WSX. We show
the structure of free MCL-1 (pink) overlaid on the PUMA-bound MCL-1 structure (white and
th
blue). The opened state from the 5 mode is shown in red, with arrows indicating the direction
and amplitude of the PCA mode.
2.0.10–4
A
1.5.10–4
[PUMA] (M)
determined using
amino acid analysis 1.0.10–4
5.0.10–5
0
B
0.4
0.8
1.2
A280
100 wild-type
kobs 80
60
(s-1) 40
20
0
A150G
I137A
10–5
10–5
10–5
[PUMA] (M)
C
100 wild-type
kobs 80
60
-1
(s ) 40
20
10–5
100 W133F
80
60
40
20
0
10–5
A150G
I137A
10–5
[PUMA/MCL-1 in excess] (M)
Figure S11 Pseudo first-order experiments were reproducible. A) Amino acid analysis results
plotted against the UV absorbance at 280 nm, used to calculate the extinction coefficient and
the associated errors. PUMA mutants W133F and Y152A were not included in this analysis
B) Biological repeats of the pseudo-first-order experiments, where the proteins were
expressed and purified on different days. Each colour represents a different biological repeat
for wild-type PUMA, A150G and I137A binding wild-type MCL-1. Grey fill represents the 7%
error in k+. C) Pseudo-first-order plots using MCL-1 in excess (white diamonds) and wildtype/mutant PUMA peptide in excess (coloured circles).
Supplementary Video: Principal component analysis of the NMR ensemble of unbound MCL1. We show a cartoon representation of the structure of bound PUMA-MCL-1 (PDB: 2roc;
PUMA in blue and MCL-1 in white). On top of this structure we overlay that of unbound MCL1 (PDB: 1wsx; rainbow). Then, as tube, we show the dynamical mode emerging from the
NMR ensemble of unbound MCL-1 that contains the opening movement of the PUMA binding
groove. Vector arrows represent PCA modes.
Table S1. Association (k+) and dissociation (k-) rate constants for the binding of wildtype/mutant PUMA to wild-type MCL-1. Errors in k- are the standard deviation of multiple (see
Fig S1 and S2) displacement experiments. Errors in k+ reflect the ~7% error in the
concentration determination of PUMA from multiple amino acid analysis experiments (see
Supplementary Methods) (19). Kd calculated from the kinetic rate constants and the errors
propagated using standard equations.
PUMA variant
wild-type
E132A
E136A
R143A
D147A
R154A
E158A
E132G
E136G
A139G
R143G
D147G
A150G
R154G
E158G
A139G+A150G
W133F
I137A
L141A
L148A
Y152A
k+ (M-1s-1)
k- (×10-3 s-1)
Kd (nM)
(7.7 ± 0.6)×106
1.39 ± 0.09
0.181 ± 0.017
6
(5.0 ± 0.4)×10
1.34 ± 0.06
0.27 ± 0.02
6
(6.2 ± 0.5)×10
3.1 ± 0.2
0.5 ± 0.05
(10.6 ± 0.8)×106
1.59 ± 0.04
0.149 ± 0.011
(7.4 ± 0.5)×106
1.31 ± 0.15
0.18 ± 0.02
(7.8 ± 0.6)×106
2.2 ± 0.2
0.28 ± 0.03
6
(8.0 ± 0.6)×10
1.09 ± 0.11
0.136 ± 0.017
6
(4.2 ± 0.3)×10
1.42 ± 0.09
0.34 ± 0.03
(4.2 ± 0.3)×106
6.7 ± 0.3
1.6 ± 0.13
6
(5.0 ± 0.4)×10
2.23 ± 0.13
0.44 ± 0.04
(11.4 ± 0.8)×106
9.7 ± 0.8
0.85 ± 0.09
6
(6.7 ± 0.5)×10
6.8 ± 0.5
1.01 ± 0.011
(7.4 ± 0.5)×106
4.6 ± 0.3
0.62 ± 0.06
6
(8.4 ± 0.6)×10
2.94 ± 0.11
0.35 ± 0.03
6
(8.1 ± 0.6)×10
1.26 ± 0.07
0.155 ± 0.014
(4.3 ± 0.3)×106
7.4 ± 0.3
1.74 ± 0.15
6
(5.5 ± 0.4)×10
3.64 ± 0.19
0.66 ± 0.06
(2.8 ± 0.2)×106
7.6 ± 0.5
2.7 ± 0.3
6
(2.8 ± 0.2)×10
1169 ± 20
423 ± 32
6
(6.1 ± 0.4)×10
620 ± 10
102 ± 8
(6.8 ± 0.5)×106
46 ± 4
6.7 ± 0.8
Table S2. Association (k+) and dissociation (k-) rate constants for the binding of wild-type
PUMA to mutant MCL-1. Errors in k- are the standard deviation of multiple (see Fig S9)
displacement experiments. As identical PUMA solutions were used to obtain k+, the error in
the concentration of PUMA was not explicitly accounted for (as is done for Table S1). For a
truer comparison between the k+ of MCL-1 mutants, the errors in k+ reported are the errors
from the linear fit of the pseudo-first-order observed rate constants (see Fig. 3A). Kd
calculated from the kinetic rate constants and the errors propagated using standard
equations. Exceptions are F300A, D237A, V197A and V201A where the Kd were directly
obtained using ITC (see Fig S9 and main text methods), and k- calculated using this
information and k+.
MCL-1 variant
wild-type
V197A
V201A
H205A
M212A
L216A
V230A
H233A
V234A
D237A
R244A
T247A
F300A
V302A
k+ (M-1s-1)
k- (×10-3 s-1)
Kd (nM)
6
(7.7 ± 0.6)×10
1.39 ± 0.09
0.181 ± 0.017
(5.91 ± 0.06)×106
41 ± 7
7±1
6
(6.8 ± 0.06)×10
37 ± 4
5.4 ± 0.6
(12.24 ± 0.12)×106
9.4 ± 0.9
0.77 ± 0.08
(9.03 ± 0.8)×106
1.10 ± 0.08
0.122 ± 0.009
6
(7.57 ± 0.7)×10
1.3 ± 0.3
0.18 ± 0.04
(7.66 ± 0.7)×106
14.7 ± 1.1
1.93 ± 0.14
6
(11.28 ± 0.11)×10
1.95 ± 0.19
0.173 ± 0.017
(10.19 ± 0.1)×106
16 ± 1
1.55 ± 0.1
6
(11.1 ± 0.1)×10
138 ± 13
12.4 ± 1.2
6
(6.32 ± 0.09)×10
1395 ± 34
221 ± 6
(9.36 ± 0.04)×106
5 ± 0.6
0.54 ± 0.07
(3.81 ± 0.06)×106
69 ± 7
18.0 ± 1.8
(6.48 ± 0.09)×106
2.3 ± 0.2
0.36 ± 0.04
Supplementary References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
Rogers JM, Steward A, & Clarke J (2013) Folding and binding of an intrinsically
disordered protein: fast, but not 'diffusion-limited'. J Am Chem Soc 135(4):1415-1422.
Day CL, et al. (2008) Structure of the BH3 domains from the p53-inducible BH3-only
proteins Noxa and Puma in complex with Mcl-1. J Mol Biol 380(5):958-971.
Rogers JM, Wong CT, & Clarke J (2014) Coupled folding and binding of the
disordered protein PUMA does not require particular residual structure. J Am Chem
Soc 136:5197-5200.
Munoz V & Serrano L (1995) Elucidating the Folding Problem of Helical Peptides
Using Empirical Parameters .3. Temperature and Ph-Dependence. J Mol Biol
245:297-308.
Shammas SL, Rogers JM, Hill SA, & Clarke J (2012) Slow, reversible, coupled
folding and binding of the spectrin tetramerization domain. Biophys J 103(10):22032214.
Sanchez IE & Kiefhaber T (2003) Origin of unusual phi-values in protein folding:
evidence against specific nucleation sites. J Mol Biol 334(5):1077-1085.
Fersht AR & Sato S (2004) Phi-Value analysis and the nature of protein-folding
transition states. Proc Natl Acad Sci USA 101(21):7976-7981.
Day CL, et al. (2008) Structure of the BH3 Domains from the p53-Inducible BH3-Only
Proteins Noxa and Puma in Complex with Mcl-1. J Mol Biol 380:958-971.
Pettersen EF, et al. (2004) UCSF Chimera--a visualization system for exploratory
research and analysis. J Comput Chem 25:1605-1612.
Karanicolas J & Brooks CL, 3rd (2002) The origins of asymmetry in the folding
transition states of protein L and protein G. Protein Sci 11(10):2351-2361.
De Sancho D & Best RB (2012) Modulation of an IDP binding mechanism and rates
by helix propensity and non-native interactions: association of HIF1alpha with CBP.
Mol Biosyst 8:256-267.
Brünger A, Brooks Iii CL, & Karplus M (1984) Stochastic boundary conditions for
molecular dynamics simulations of ST2 water. Chem Phys Lett 105:495-500.
Brooks BR, et al. (1983) CHARMM: A program for macromolecular energy,
minimization, and dynamics calculations. J Comput Chem 4:187-217.
Ryckaert J-P, Ciccotti G, & Berendsen HJC (1977) Numerical integration of the
cartesian equations of motion of a system with constraints: molecular dynamics of nalkanes. J Comput Phys 23:327-341.
Kumar S, Bouzida D, Swendsen RH, Kollman PA, & Rosenberg JM (1992) The
Weighted Histogram Analysis Method for Free-Energy Calculations on Biomolecules
.1. The Method. J Comput Chem 13(8):1011-1021.
Day CL, et al. (2005) Solution Structure of Prosurvival Mcl-1 and Characterization of
Its Binding by Proapoptotic BH3-only Ligands. J Biol Chem 280:4738-4744.
Bakan A & Bahar I (2009) The intrinsic dynamics of enzymes plays a dominant role in
determining the structural changes induced upon inhibitor binding. Proc Natl Acad Sci
USA 106:14349-14354.
Bakan A, Meireles LM, & Bahar I (2011) ProDy: protein dynamics inferred from theory
and experiments. Bioinformatics 27:1575-1577.
Rogers JM, Wong CT, & Clarke J (2014) Coupled folding and binding of the
disordered protein PUMA does not require particular residual structure. J Am Chem
Soc 136(14):5197-5200.