(gray, see Table S2 in Text S3 for mean and CV) distributions in endogenous protein concentrations for untreated HeLa cells. Bar. Bid. Mcl1. Bax. Bcl-2. Rcpt.
Rcpt
Flip
C8
Bar
Bid
Mcl1
Bax
20
tMOMP sensitivity
5
0
0
0
Bcl-2
20
-52 4 6 2 4 6 2 4 6 2 4 6 2 4 6 2 4 6 -202 4 6 -202 4 6 log10 mol/cell log10 mol/cell log10 mol/cell log10 mol/cell log10 mol/cell log10 mol/cell log10 mol/cell log10 mol/cell XIAP
Cyto c
Smac
Apaf
C9
C3
C6
PARP
tMOMP sensitivity
5
0
-52
4
6
log10 mol/cell
2
4
6
log10 mol/cell
2
4
6
log10 mol/cell
2
4
6
log10 mol/cell
2
4
6
log10 mol/cell
2
4
6
log10 mol/cell
2
4
6
log10 mol/cell
2
4
6
log10 mol/cell
Figure S1. Analytically calculated sensitivity of tMOMP to changes in protein initial concentrations. Scatter plots show the simulated relationship between initial protein concentration and tMOMP sensitivity, as calculated using the analytical form described in Equation 1 (see also Figure Box 1 for the specific expression for tMOMP. The initial concentration for the indicated protein was uniformly sampled in the exponent for values between 102 to 107 proteins per cell while all other initial protein concentrations and rate constants were set at their default value. Vertical bars represent the 5th and 95th percentiles of the measured (orange, see Figure 3 and Figure S8 in this Text S1) or assumed (gray, see Table S2 in Text S3 for mean and CV) distributions in endogenous protein concentrations for untreated HeLa cells.
Rcpt
Flip
C8
Bar
Bid
Mcl1
Bax
Bcl-2
10
0 2 4 6 log10 mol/cell XIAP
2 4 6 log10 mol/cell Cyto c
2 4 6 log10 mol/cell Smac
2 4 6 log10 mol/cell Apaf
2 4 6 log10 mol/cell C9
2 4 6 log10 mol/cell C3
2 4 6 2 4 6 log10 mol/cell log10 mol/cell C6
PARP
-10
10
tPARP (hr)
10 0
5
0 2 4 6 log10 mol/cell
2 4 6 log10 mol/cell
2 4 6 log10 mol/cell
2 4 6 log10 mol/cell
4 6 2 2 4 6 2 4 6 log10 mol/cell log10 mol/cell log10 mol/cell
tPARP sensitivity
0
5
tPARP sensitivity
tPARP (hr)
10
2 4 6 -10 log10 mol/cell
Figure S2. Sensitivity of tPARP to changes in protein initial concentrations. Scatter plots show the simulated relationship between initial protein concentration and tPARP (green) or numerically calculated tPARP sensitivity (blue) following TRAIL addition, for the indicated proteins. The initial concentration for the indicated protein was uniformly sampled in the exponent for values between 102 to 107 proteins per cell while all other initial protein concentrations and rate constants were set at their default value. Vertical bars represent the 5th and 95th percentiles of the measured (orange, see Figure 3 and Figure S8 in this Text S1) or assumed (gray, see Table S2 in Text S3 for mean and CV) distributions in endogenous protein concentrations for untreated HeLa cells.
Rcpt
Flip
C8
Bar
Bid
Mcl1
Bax
Bcl-2
10
fPARP
0.8
5
0.6 0.4
0
0.2
fPARP sensitivity
1.0
-5
0 2 4 6 2 4 6 log10 mol/cell log10 mol/cell XIAP
Cyto c
2 4 6 2 4 6 log10 mol/cell log10 mol/cell Smac
Apaf
2 4 6 2 4 6 2 4 6 2 4 6 log10 mol/cell log10 mol/cell log10 mol/cell log10 mol/cell C9
C3
C6
PARP
10
fPARP
0.8
5
0.6 0.4
0
0.2 -5
0 2 4 6 2 4 6 log10 mol/cell log10 mol/cell
2 4 6 log10 mol/cell
2 4 6 log10 mol/cell
2 4 6 2 4 6 2 4 6 2 4 6 log10 mol/cell log10 mol/cell log10 mol/cell log10 mol/cell
Figure S3. Sensitivity of fPARP to changes in protein initial concentrations. Scatter plots show the simulated relationship between initial protein concentration and fPARP (green) or numerically calculated fPARP sensitivity (blue) following TRAIL addition, for the indicated proteins. The initial concentration for the indicated protein was uniformly sampled in the exponent for values between 102 to 107 proteins per cell while all other initial protein concentrations and rate constants were set at their default value. Vertical bars represent the 5th and 95th percentiles of the measured (orange, see Figure 3 and Figure S8 in this Text S1) or assumed (gray, see Table S2 in Text S3 for mean and CV) distributions in endogenous protein concentrations for untreated HeLa cells.
fPARP sensitivity
1.0
Bar
Bid
Mcl1
Bax
Bcl-2
6
0.05
4
0
2 0
-0.05 2 4 6 log10 mol/cell
2 4 6 log10 mol/cell Cyto c
2 4 6 log10 mol/cell Smac
2 4 6 log10 mol/cell Apaf
2 4 6 log10 mol/cell
2 4 6 log10 mol/cell
C9
C3
2 4 6 2 4 6 log10 mol/cell log10 mol/cell C6
PARP
-0.1
8
0.1
6
0.05
4
0
2
-0.05
0 2 4 6 log10 mol/cell
2 4 6 log10 mol/cell
2 4 6 log10 mol/cell
2 4 6 log10 mol/cell
2 4 6 log10 mol/cell
2 4 6 log10 mol/cell
2 4 6 log10 mol/cell
tswitch sensitivity
C8
0.1
x10-3 XIAP tswitch (sec/(cPARP molec/cell))
Flip
2 4 6 -0.1 log10 mol/cell
Figure S4. Sensitivity of tswitch to changes in protein initial concentrations. Scatter plots show the simulated relationship between initial protein concentration and tswitch (green) or numerically calculated tswitch sensitivity (blue) following TRAIL addition, for the indicated proteins. The initial concentration for the indicated protein was uniformly sampled in the exponent for values between 102 to 107 proteins per cell while all other initial protein concentrations and rate constants were set at their default value. Vertical bars represent the 5th and 95th percentiles of the measured (orange, see Figure 3 and Figure S8 in this Text S1) or assumed (gray, see Table S2 in Text S3 for mean and CV) distributions in endogenous protein concentrations for untreated HeLa cells. Note that when a simulated cell does not die (see fPARP response curves, Figure S3 in this Text S1), tswitch is assigned an arbitrarily high value.
tswitch sensitivity
tswitch (sec/(cPARP molec/cell))
x10-3 Rcpt 8
75kD 50kD
46kD
37kD 36,000 25kD endogenous Bcl-2/cell
Background subtracted signal
10
5
GST-Bcl2 loaded (ng) 4 1 0.25
B
w tH eL M a ar ke rs
A Endogenous Bcl-2
30 25 20
10
Pure Bcl-2 Linear fit
5 0
20kD 15kD
y = 6547.5x R2 = 0.99
15
0
1 2 3 Amount of pure Bcl-2 loaded (ng)
4
Background subtracted signal: 25.9
7.6
1.9
1.0
D
160
75kD 127,000 GFP-Bcl-2/cell
50kD 37kD 25kD
Background subtracted signal: 175.2
53.6
8.1
Probe with anti-GFP antibody
200
Background subtracted signal
rs ke ar M
GFP loaded (ng) 4.5 1.1 0.28
4 G x10 4 FP -B IMS cl -R -2 H P+ eL a
Probe with anti-Bcl-2 antibody
C GFP-Bcl-2
120 y = 39397x R2= 0.99
80
Pure GFP Linear fit
40 0 0
1
2 3 4 Amount of pure GFP loaded (ng)
5
9.0
Figure S5. Immunoblot quantification of endogenous Bcl-2 and GFP-Bcl-2 in HeLa cells. Pure protein and HeLa cell lysate were loaded on a 10% Tricine SDS-PAGE gel. After transfer to a PVDF membrane, blots were probed, scanned on a LI-COR Odyssey, and quantified digitally. A) The membrane was probed with rabbit anti-Bcl-2 (Santa Cruz Biotechnology SC783) followed by AF680-conjugated anti-rabbit. Pure Bcl-2 is a 46 kDa fusion protein (Santa Cruz Biotechnology SC4096). B) From the standard curve, we calculate that, on average, a single HeLa cell has 1.57x10-15 g Bcl-2. Using a molecular weight of 26,135 Da for Bcl-2, we find 36,000 Bcl-2/cell. The average of 14 such measurements yields 30,000 Bcl-2/HeLa cell; s.e.m = 10,000. C) The membrane was probed with mouse anti-GFP (Roche #11814460001) followed by IRDye 800-conjugated anti-mouse. Pure GFP was purchased from Biovision (#4999-100). D) From the standard curve, we calculate that, on average, an IMS-RP GFP-Bcl-2 HeLa cell has 5.68x10-15 g GFP-Bcl-2. Using a molecular weight of 27,000 Da for GFP, we find 127,000 GFP-Bcl-2/cell. The average of 5 such measurements yields 133,000 GFP-Bcl-2/HeLa cell, s.e.m = 18,000. We set the background-subtracted average GFP-Bcl-2 fluorescence intensity of the population of cells in the first frame of the movie used in Figure 7 equal to the average number of GFP-Bcl-2 in the HeLa cells (calculated above). By adding 30,000 endogenous Bcl-2 to this amount, the x-axis of Figure 7A was rescaled into total Bcl-2 proteins per cell.
B
120 80 40 0
0
300
0.001 0.002 0.003 0.004 0.005 [IgG] ug/ul Bax
200 100 0
0
0.03
Bcl-2
120 Mean signal
0.01 0.02 [IgG] ug/ul
90 60 30 0
0
0.01 0.02 0.03 0.04 0.05 [IgG] (ug/ul)
Mean signal
C3 80 60 40 20 0
0
0.002 0.004 0.006 0.008 0.01 [IgG] ug/ul
9 8 7 6 5 4 3 2 1 0
% of Max
80
40 20
0
0
0.0004 0.0008 [IgG] ug/ul
100
101
102 Bid
103
104
101
102 103 XIAP
104
101
102 Bax
104
100
XIAP
80 60 40 20 0
0 100
0.001 0.002 0.003 0.004 0.005 [IgG] ug/ul Bax
6
100
5
80
4 3 2 1 0
60
% of Max
XIAP
100
Bid
% of Max
Mean signal
160
0.0004 0.0008 [IgG] ug/ul
8 7 6 5 4 3 2 1 0
Antibody Isotype control
60 40 20
0
0.01 0.02 [IgG] ug/ul
3.5 3 2.5 2 1.5 1 0.5 0 0
Bcl-2
2.5
0.03
0
103
80 60 40 20 0.01 0.02 0.03 0.04 0.05 [IgG] ug/ul C3
0
100
101
102 103 Bcl-2
104
100
101
102 C3
104
100
2
80
1.5 1 0.5 0
100
100 % of Max
0
Signal ratio (Ab/isotype)
100
Signal ratio (Ab/isotype)
200
Signal ratio (Ab/isotype)
Mean signal
300
0
Mean signal
Signal ratio (Ab/isotype)
Bid
400
C
% of Max
Antibody Isotype control
Signal ratio (Ab/isotype)
A
0
60 40 20
0.002 0.004 0.006 0.008 0.01 [IgG] ug/ul
0
103
Figure S6. Optimization and validation of antibody dilutions for flow cytometry. A) Plots showing average flow cytometry signal as a function of increasing concentrations of antibody and isotype control. For Bid, the matched isotype control was rabbit IgG; for XIAP, Bax, and Bcl2, mouse IgG1; for C3, mouse IgG2a (Santa Cruz Biotechnology). B) Plots showing the ratio in average signal from antibody staining versus background isotype control. Arrows indicate the concentrations of antibodies used in all other experiments. C) Histograms showing the degree of overlap of distributions of isotype control and antibody-stained cells. All the antibodies used are described in the Methods section of the main text.
100
B
HeLa 2o only HeLa + Bid siRNA, 1o + 2o Ab HeLa + NT siRNA, 1o + 2o Ab
4
10
% of Max
80
103
60
102
40
101
20
100
C 100
101
102 Bid
103
% of Max
101
D
HCT116 2o only HCT116 Bax-/- ,1o + 2o Ab HCT116 parental,1o + 2o Ab
104
80
102 Bid-GFP
103
4
10
HeLa + YFP-Bax, 1o + 2o Ab
103
60
102
40
101
20
E 100
101
102 Bax
103
104
101
102 XIAP
103
100 100
104
H 4
10 anti-Bcl-2 + anti-rabbit-AF647
anti-Bcl-2 + anti-mouse-AF647
HeLa + GFP-Bcl-2, 1o + 2o Ab HeLa + GFP-Bcl-2, 2o only
103 102 101 100 0 10
4
10
HeLa + C3-GFP, 1o + 2o Ab
101
20
10
103
102
40
4
102 YFP-Bax
103
60
G
101
F
HCT116 2o only HCT116 XIAP-/-, 1o + 2o Ab HCT116 parental, 1o + 2o Ab
80
0 100
100 0 10
104
anti-C3 + anti-mouse-AF647
0 100
% of Max
100 0 10
104
anti-Bax + anti-mouse-AF647
0
HeLa + Bid-GFP, 1o + 2o Ab HeLa + Bid-GFP, 2o only
anti-Bid + anti-rabbit-AF647
A
101
102 GFP-Bcl-2
103
104
101
102 C3-GFP
103
104
HeLa + GFP-Bcl-2, 1o + 2o Ab HeLa + GFP-Bcl2-, 2o only
103 102 101 100 0 10
101
102 GFP-Bcl-2
103
104
Figure S7. Validation of antibodies used for measuring protein distributions by flow cytometry. A-B) Validation of a rabbit anti-Bid antibody (Atlas Antibodies HPA000722). A) Histograms of anti-Bid stained non-targeting (NT) siRNA treated HeLa cells (blue) compared to anti-Bid stained, Bid siRNA treated cells (green) and cells stained with secondary antibody only (orange). B) 2D scatter plot of anti-Bid vs. Bid-GFP signal in Bid-GFP transfected HeLa cells stained with anti-Bid followed by Alexa Fluor 647 (AF647) conjugated anti-rabbit secondary antibody (red), or secondary antibody only (blue). C-D). Validation of mouse anti-Bax antibody (Chemicon International MAB4601). C) Histograms of parental HCT116 cells stained with anti-Bax followed by a secondary antibody (blue) or with secondary antibody alone (orange) and of HCT116 Bax-/derivatives, stained with both anti-Bax and secondary antibody (green). D) 2D scatter plot of anti-Bax vs. YFP-Bax signal in YFPBax transfected HeLa stained with anti-Bax and AF647-conjugated anti-mouse antibody. E) Validation of mouse anti-XIAP antibody (BD Biosciences 610717); histograms of parental HCT116 cells stained with antiXIAP and secondary antibody (blue) or with secondary antibody alone (orange) and HCT116 XIAP-/- cells stained with antiXIAP and secondary antibody. F) Validation of mouse anti-C3 antibody (Santa Cruz Biotechnology SC7272); 2D scatter plot of anti-C3 vs. C3-GFP signal in C3-GFP transfected HeLa cells stained with anti-C3 and AF647 conjugated secondary antibody. G-H) Validation of mouse anti-Bcl-2 antibody (Santa Cruz Biotechnology SC7382; (G)) and rabbit anti-Bcl-2 antibody (Santa Cruz Biotechnology SC783; (H)) in 2D scatter plots of anti-Bcl-2 vs. GFP-Bcl-2 signal for GFP-Bcl-2 transfected HeLa cells stained with anti-Bcl-2 and AF647 conjugated secondary antibody.
Frequency
A
0.02 0.015
B
Bax Histogram of data Normal fit
0.8
0 0
0.4 0.2 100
200
300 400 FL1H
500
600
Bid
Frequency
0.015
0
100
200
300 FL1H
400
500
Bid
0.8
Empirical CDF Fitted CDF
0.6
0.01
0.4
0 0
0.2 100
200
300 400 FL4H
500
0
600
Bcl-2 0.015 Frequency
0
1
Histogram of data Normal fit
0.005
0
100
200
300 400 FL4H
500
600
Bcl-2
1
Histogram of data Normal fit
0.8
Empirical CDF Fitted CDF
0.6
0.01
0.4 0.005 0 0
0.016 Frequency
Empirical CDF Fitted CDF
0.6 0.01 0.005
0.012
0.2 100
200
300 FL1H
400
500
XIAP
0
0
100
200
300 FL1H
400
500
XIAP
1
Histogram of data Normal fit
0.8
Empirical CDF Fitted CDF
0.6 0.008
0.4
0.004 0 0
0.2 100
200
300 400 FL1H
500
600
Caspase-3 0.012 Frequency
Figure S8. Protein level distributions are well fit by lognormal distributions. Fitted probability density functions and measured histograms (A), and empirical and fitted cumulative density functions (B) of fluorescence intensity in HeLa cells stained with validated 600 antibodies against the indicated proteins, as measured by flow cytometry. The data, recorded in a .fcs file, were read into MATLAB in linear form and gated to select cells of similar size as described in Methods. All distributions were found to pass the KolmogorovSmirnov test for normality indicating that the underlying fluorescence distributions are well fit by lognormal distributions.
Bax
1
0
0
200
300 400 FL1H
500
600
Caspase-3
1
Histogram of data Normal fit
100
0.8
Empirical CDF Fitted CDF
0.6
0.008
0.4 0.004 0 0
0.2 100
200
300 400 FL1H
500
600
0
0
100
200
300 400 FL1H
500
600
Rcpt
10 8 tPARP (hr)
Flip
R=−0.28
C8
R=0.03
Bar
Bid
R=0.12
R=−0.34
Mcl1
R=−0.34
Bax
R=0.20
Bcl-2
R=−0.40
R=0.37
6 4 2 0 10
500
1500
1000
3000
0.5
1 1.5 2
1500
5
10
R=0.16
R=-0.41
R=0.06
R=-0.34
500
15 1
2
3
4
0.5
1
R=0.25
R=-0.41
1.5
2
4
R=-0.32
6
R=0.002
tPARP (hr)
8 6 4 2 0
10
500 1500 mol/cell XIAP R= 0.25
Cyto c R=0.01
8 tPARP (hr)
0.5 1 1.5 2 mol/cell x 104
1000 3000 mol/cell
Smac R=−0.08
500
1500 mol/cell
5 10 15 1 2 3 4 4 mol/cell x 104 mol/cell x 10
Apaf R=−0.07
C9 R=−0.06
C3 R=−0.07
0.5 1 1.5 mol/cell x 105 C6 R=−0.01
2 4 6 mol/cell x 104 PARP R=−0.01
6 4 2 0 10
tPARP (hr)
8
0.5 1
1.5 2 2 R=0.11
4
6
8
10 0.5 1
R=0.01
1.5 2
0.5 1
1.5 2
R=-0.13
R=-0.08
0..5 1 1..5 2 2 4 6 8 10 0..5 1 1..5 2 mol/cell x 105 mol/cell x 105 mol/cell x 105
0..5 1 1.5 2 mol/cell x 105
0.5
1
1.5
2 0.5 1
1.5 2
0.5 1
1.5 2
0.5 1
1.5 2
R=-0.25
R=-0.03
R=-7x10-4
0..5 1 1..5 2 0..5 1 1..5 2 mol/cell x 104 mol/cell x 105
0..5 1 1..5 2 mol/cell x 104
0..5 1 1..5 2 mol/cell x 106
R=-0.08
6 4 2 0
Figure S9. Correlations in initial conditions qualitatively and quantitatively impact the relationship of time of death to protein concentration. Scatter plots of time of death (tPARP) vs. initial concentration for the indicated proteins for 200 simulations of the model with independently sampled initial conditions (blue) or initial conditions correlated as measured for Bax, Bcl2, Bid, caspase-3 and XIAP and all other proteins varying independently (brown). Black lines are linear regressions of the data points, and R values are the Pearson correlation coefficients.
-0.2 -0.3
AP
XI
l-2
Independent Correlated
ip Ba r M cl -1 XI AP Bc l-2
C 9 C 6 PA R C P yt o c Fl
3E-07
ac ip Ba r M cl Bc 1 l-2
Sm
o C c 9
C yt
Fl
C 9 Sm ac
af
3
Ap
C
x
tswitch
2E-07
-1E-07
AP
0
l-2
1E-07
XI
R C P 3 Ap af C 9 Ba x C 8 C 6 R cp t M cl -1 Bi d Ba r C yt o Fl c ip
PA
Bc
-0.1
ac
0.2
Sm
R
0.3
4E-07
Bc
0.4
-12E-06
C 9 Ba x C 8 C 6 R cp t M cl -1 Bi d Ba r C yt o Fl c ip
tswitch
Independent Correlated
Sm a PA c R C P 3
-0.7
-8E-06 -10E-06 Slope (sec/(cPARP molec/cell)*molec)
Independent Correlated
-0.6
0
0
-6E-06
-0.5
0.1
2E-06
-2E-06
-0.4
0.5
4E-06
-4E-06
-0.3
0.6
fPARP
Ba
9
Sm
ac
af
C
x
3
Ap
C
Ba
P R PA
R
-0.2
Ba x Bi d C 8
-3.4 8E-06
Slope (%/molec)
l-2 8 M cl -1 R cp t C 6 Fl ip Bi d Ba r C yt o c
-0.1
Independent Correlated
6E-06
C
0
Bc
0.1
-0.4 -0.8
fPARP AP
0.2
XI
0.3
0
-0.2 -0.6
Independent Correlated
-0.8
0.2
P
-0.6
0.4
R
-0.4
0.6
PA
ip Ba r M cl Bc 1 l-2
Fl
-0.2
Slope (sec/molec)
ac Sm
o C c 9
C yt
R
0
Ba x Bi d C 8
0.2
tMOMP
0.8
C 3
0.4
1.6
tMOMP R cp t PA R P XI AP Ap af C 6
0.6
C 3
-0.5
Independent Correlated
R cp t PA R P XI AP Ap af C 6
Independent Correlated
-0.4
l-2 8 M cl -1 R cp t C 6 Fl ip Bi d Ba r C yt o c
-0.3
C
-0.2
Bc
M
Ba r cl -1 XI AP Bc l-2
Fl ip
-0.1
tPARP
Ba x C 8 Bi d R cp t Sm ac Ap af C 3
R C P yt o c
PA
C 9 C 6
0
R cp t Sm ac Ap af C 3
R
0.1
Ba x C 8 Bi d
0.2
Slope (sec/molec)
0.3
1.7 0.5 0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.3 -0.4 -0.5 -3.6
AP
0.4
tPARP
XI
0.5
-2E-07 -3E-07 -4E-07
Independent Correlated
Figure S10. Correlation in protein levels affects the rank ordering of impact of species on all four features. Bar graphs of the Pearson correlation coefficients (R; left) and slopes of the linear regression (right) of each feature against endogenous variability in each protein listed as obtained for simulations sampling from fully independent distributions (blue), from joint distribution for Bax, Bcl-2, Bid, caspase-3 and XIAP and all other proteins sampled independently (brown). For each feature, the protein species were ordered by the value of R. The bar graph showing the Pearson correlation coefficients for tPARP (top left) is reproduced from Figure 5E to allow comparisons of the effects across different features.
Bax 12
Bax, C8
12 MSE = 0.264
Bax, C8, C3
12 MSE = 0.228
12 MSE = 0.134
MSE = 0.146
10
10
10
8
8
8
8
6
6
6
6
4
4
4
4
2
2
2
2
tMOMP (hr)
10
0
2
4
6 Bid/cell
8
0 x 104
2
Bax, C8, C3, Rcpt 12
4
6 Bid/cell
8
0 x 104
Bax, C8, C3, Rcpt, Mcl1 12
MSE = 0.098
MSE = 0.066
2
4
6 Bid/cell
8
0 x 104
Bax, C8, C3, Rcpt, Mcl1, Bcl-2 12 MSE = 0.044 10 8
8
6
6
6
6
4
4
4
4
2
2
2
2
tMOMP (hr)
8
6 Bid/cell
8
0 x 104
2
4
6 Bid/cell
8
0 x 104
2
4
6 Bid/cell
8
x 104
10
0 x 104
8
MSE = 0.034
8
4
6 Bid/cell
12
10
2
4
Bax, C8, C3, Rcpt, Mcl1, Bcl-2, Bar
10
0
2
2
4
6 Bid/cell
8
x 104
Figure S11. Predictability of death time is improved by knowledge of key protein concentrations. Scatter plots of predicted tMOMP as a function of Bid initial protein levels either with perfect knowledge of the concentrations of all model species (black points) or knowing only the levels of the most influential proteins listed above the plot (ranked by R2 for tMOMP as in Figure S10 in this Text S1; brown points). Simulations were selected from a series for which initial protein concentrations were sampled from joint distribution for Bax, Bcl-2, Bid, caspase-3 and XIAP (all other proteins sampled independently) by defining “knowledge” of a protein as having an initial concentration within the range of average ± 12.5%. MSE is the mean squared error relative to perfect knowledge (black points). These graphs show example results from a single run while the curves in Figure 6B show average results and standard deviations for n = 10 runs.
250 ng/ml TRAIL
50 ng/ml TRAIL
30K Bcl2 60K Bcl2
2000
10 ng/ml TRAIL 30K Bcl2 60K Bcl2
30K Bcl2 60K Bcl2
1600 1200 800 400 00
2
4
6 8 tMOMP (hr)
10
12 0
2
4
6 8 tMOMP (hr)
10
12 0
2
4
6 8 tMOMP (hr)
10
12
Figure S12. Sensitivity of tMOMP variability to doubling the initial level of Bcl-2. Histograms of the distributions of tMOMP as observed in simulation with average Bcl-2 levels set to 30,000 per cell (light bars) or 60,000 per cell (dark bars), for three doses of TRAIL (2.5ug/ml cycloheximide). Bax, Bcl2, Bid, Caspase-3 and XIAP initial protein initial concentrations were sampled from the measured joint distribution and all other initial concentrations were independently from lognormal distributions parametrized as described in Table S2 in Text S3. Distributions generated using 30,000 Bcl-2 per cell better recapitulate those observed experimentally (see Spencer et al., (2009), Ref. 15 in main text for experimentally observed tMOMP distributions).