Figure 1.
Labeling, measurable parameters and data obtained from pcSILAC experiments.
A) Protein labeling scheme for pcSILAC experiments. Two cell cultures were fully labelled only on Arg residues before cell treatments. For example, cells to be treated later with HSP90 inhibitor were fully labelled with 13C615N4-L-arginine (R10/K0) (“heavy” cells) while cells to be used as control were fully labelled with 13C6-L-arginine (R6/KO) (“medium” cells). At the start of the experiment, “heavy” cells were transferred into a medium containing light R (R0) and “heavy” K (13C615N2-L-Lys, referred to as R0/K8 medium), while “medium” cells were transferred to a medium containing light R (R0) and “medium” K (2H4-L-lysine, referred to as R0/K4 medium). GA or DMSO as a control were added either simultaneously or later, depending on the experiment. Cells were harvested and lysed at various time points to produce extracts that were mixed before analysis. B) Conceptual view of the levels of a hypothetical protein in a mixture of two (control and treated) pcSILAC-labeled samples. Pre-existing protein is fully labeled R6/K0 and R10/K0 (light and dark brown), respectively for the control and treated sample. Newly synthesized protein is labeled R0/K4 (control, pink) and R0/K8 (treated, blue). The SILAC (H/M) ratios for R- and K-containing peptides therefore measure the ratios of preexisting and newly synthesized proteins at time t. C) examples of pcSILAC spectra for peptides from the tyrosine kinase LCK (left, peptide IPYPGMTNPEVIQNLER at t = 6h, z = 2) and the chaperone DNAJB1 ( = Hsp40) (right, peptide EGDQTSNNIPADIVFVLK at t = 20h, z = 2).
Figure 2.
Raw and normalized (H/M)K, (H/M)R ratios measured in pcSILAC experiments.
A),B) Scatter plots of global ratios of pre-existing protein ((H/M)R) and newly synthesized protein ((H/M)K) after correction for mixing ratio and normalization around the median. Values shown are for pcSILAC experiment 1 (average of two replicates) at t = 6h (A) and t = 20h (B). Reference (red) and other proteins discussed in the text (blue) are indicated. C) Evolution in time after start of GA treatment of normalized ratios of total protein (stSILAC, blue), newly synthesized proteins ((H/M)K, green) and pre-existing protein ((H/M)R, red). Data points refer to values at t = 6,12,20h after addition of GA (t = 0). pcSILAC ratios were normalized (i.e. centered around population median) to facilitate comparison with changes in total protein levels. Fitting of the values for DNAJB1 to the model was very poor due to its complex behavior, therefore decay and synthesis rates could not be calculated for this early induced protein. D) Box plots (experiment 1) of global log2(H/M)K and log2(H/M)R ratios after mixing ratio correction but before normalization E) same as D), but values are shown after correction for mixing ratio and normalization.
Figure 3.
Computational workflow for the determination of kinetic parameters kd and VB/VA from pcSILAC data.
The individual steps of the computational workflow are described in detail in the supplementary information S1. Briefly, (1) using the total peak intensities (similar to iBAQ scores), the global mixing ratio between both samples is computed. (2) The equations describing the evolution of the protein concentrations in the cells are converted into equations describing the evolution of the protein abundances using the mixing ratio values. (3) These equations can then be used for a global fit of the experimental data for multiple proteins at the same time. (4) The global fit results in estimates for the coefficients describing the peptide cleavage rates (or efficiencies) - these coefficients are then set to be identical for all protein species. (5) The estimates of the cleavage efficiencies are inserted into the model describing the evolution of protein abundances in the samples; the experimental data from each protein species can then be fitted, for each protein species independently. (6) Instead of directly doing these fits to the experimental data, some random artificial noise is added through a bootstrap-like method - this increases the confidence in the results (as noise cannot be avoided in the experimental measurements, and we repeat the fit with various simulated noise values to compare the outcomes). (7) The fit of the model to the bootstrapped data results in one set of estimates of synthesis and degradation rate constants for the normal and drug-treated conditions for each protein. (8) The final estimates are the average values from the different estimates that were obtained by repeating the fitting step multiple times with different bootstrapped data values for each protein species.
Figure 4.
Results from calculations of decay constants, synthesis rates and evolution of total protein levels.
Indexes of variables are A (as in kA, VA) for control (+DMSO) and B (as in kB, VB) for treated (+GA) cells. A) Scatter plot of the values of the degradation constants for the control and treated sample (experiment 2, 911 proteins). The position of reference proteins is indicated. The dashed line indicates a 1:1 relationship B) Scatter plot of VA and VB (same dataset as A). Other heat shock proteins are shown in pink. The dashed line indicates a 1:1 relationship C) Kernel density estimate of log2 of ratios of synthesis rates (VB/VA) and degradation constants (kB,d/kA,d) after correction for cell growth. D) Comparison of distributions of log2 of ratios S of net protein levels (treated/control) calculated from pcSILAC data at t = 6, 12, 20h vs. ratios at steady-state (t = infinite) calculated from the model. S values were corrected for mixing inequalities.
Figure 5.
Changes in degradation and synthesis rates induced by Hsp90 inhibition and relationship with changes in net total protein levels and protein families.
Indexes of variables are A (as in kA, VA) for control (+DMSO) and B (as in kB, VB) for treated (+GA) cells. A) Scatter plot of the values of ratios of intrinsic degradation constants kB,d/kA,d vs. the ratios of synthesis rates VB/VA. The median values of kB,d/kA,d and VB/VA for the population are indicated with dashed lines. Coloring of points is according to the calculated treated /control total protein ratio at t = 20h (ratio S). B) Same as A) but with coloring of ribosomal, proteasome and heat shock proteins. Groups I and II are discussed in the main text. All values are log2.
Figure 6.
Changes in degradation and synthesis rates in subsets of net decreasing, invariant and increasing proteins.
Indexes of variables are A (as in kA, VA) for control (+DMSO) and B (as in kA, VA) for treated (+GA) cells. Probability density functions of decay constant ratios kB,d/kA,d (A) and synthesis rate ratios VB/VA. (B) are shown for the 10% most strongly increased proteins, the 10% most strongly decreased proteins and a 20% of proteins which were around the median of the concentration change (invariant proteins). The ratio of protein levels at steady-state (pB/pA parameter) was used for the selection.