Fig 1.
Variability in basal and cisplatin-induced gene expression in HepG2 cells compared to PHHs from donor samples.
(A) Graphical representation of DNA damage-induced protein signaling upon cisplatin exposure. Kinases ATM, ATR, CHEK1 and CHEK2 are activated upon DNA damage and phosphorylate transcription factor p53. Phosphorylated p53 enters the nucleus and induces transcription of downstream targets, among which are MDM2, p21 and BTG2. (B) TP53 expression in HepG2 and 50 PHH donor samples without and after cisplatin exposure. HepG2 cells cluster with one low TP53-expression PHH donor sample. (C-F) Basal expression of TP53 (C), MDM2 (D), CDKN1A (E) and BTG2 (F) in PHHs and the HepG2 cell line within their corresponding low-, intermediate- and high-expression clusters. HepG2 cell line has lower basal expression levels for TP53, CDKN1A and BTG2, but not for MDM2, compared to average expression in PHHs within the same cluster. Contour lines are violin plots with individual samples marked by small grey dots, and large colored dots are cluster means.
Fig 2.
Expression patterns and correlations between TP53 and its downstream targets in HepG2 cells and 50 PHHs at increasing cisplatin levels.
(A-D) TP53 (A), MDM2 (B), CDKN1A (C) and BTG2 (D) gene expression patterns as a function of cisplatin concentration in HepG2 cells (3 replicates (dots) and their mean (line segments), left panels) and PHHs (right panels) at the 8-hour time point. Note that at high cisplatin concentrations, gene expression declines, which is likely explained by cytotoxicity onset. (E) Basal downstream target correlations with basal TP53 expression at the 8-hour timepoint. (F-G) Correlation strengths between TP53 and the downstream targets MDM2, CDKN1A and BTG2 after exposure to 0, 0.1, 1 and 3.3 μM cisplatin at 8 (F) and 24 (G) hours. * p-value < 0.1, ** p-value < 0.01, *** p-value < 0.001.
Fig 3.
ODE modeling of cisplatin-induced p53 signaling in HepG2 cells calibrated on live cell confocal imaging data.
(A) Graphical representation of the model. Cisplatin causes DNA damage, which triggers phosphorylation of p53. Phosphorylated p53 induces expression of its downstream targets MDM2, p21 and BTG2. (B) Example images of nuclear GFP expression in p53, and cytoplasmic GFP expression in BTG2 HepG2 BAC-GFP reporter cells at 1 and 48 hours. Blue, Hoechst-stained nuclei; green, GFP signal. (C) Mean of single-cell protein expression data measured over a 65-hour period after 1, 2.5 and 5 μM cisplatin exposure. Experimental replicates (exp. 1–4) are shown in different colors. The model simulation after parameter calibration is shown as a black solid line. (D) Delay in peak MDM2, p21 and BTG2 expression levels with respect to maximum p53 expression. (E) Response latency of MDM2, p21 and BTG2 for different cisplatin concentrations. (F) Model-predicted effective cisplatin concentrations (ECi) over time. Corresponding applied concentrations are 1, 2.5 and 5 μM for EC1, EC2 and EC3, respectively. (G) Induction of p53-GFP in three replicates (Exp. 1–3) after exposure of non-treated HepG2 cells to conditioned medium collected from cells previously exposed to cisplatin for 72 hours.
Fig 4.
Reduced p53-MDM2 feedback promotes p53 and MDM2 expression.
(A) Model simulations upon disruption of the MDM2-p53 feedback by weakening the feedback strength of MDM2 on p53 degradation to 50% (dashed green) or 20% (dashed yellow) of the original strength; simulations without disruption are shown as solid purple lines. (B-C) p53-GFP (B) and MDM2-GFP (C) expression in HepG2 cells after exposure to increasing concentrations of cisplatin (purple) and to cisplatin in combination with 5 (green) or 10 (yellow) μM Nutlin, both administered at t = 0 hours.
Fig 5.
Comparison between correlations for virtual donor samples and PHHs.
(A) Graphical explanation of the creation of virtual donor samples. For all fitted parameters Pi, we added a value, drawn from a normal distribution with mean 0 and standard deviation pi ∙ c, to the parameter value. We executed this 50 times to generate one set of 50 virtual donor samples and repeated this 1000 times to obtain 1000 sets of 50 virtual donor samples. (B) Correlations between basal TP53 expression and the downstream targets MDM2, CDKN1A and BTG2 at 8 hours for HepG2-derived virtual samples (colored boxes) and PHHs (grey shaded areas) in basal expression conditions (left) and after 3.3 μM cisplatin exposure (right) with variability factor c = 0.2 and a small amount of measurement noise, chosen such that the correlation of basal TP53 with itself after cisplatin exposure under variability factor c = 0.2 decreased to the experimentally observed correlations in PHHs. The grey solid lines represent the observed correlations in PHHs and the grey shaded areas represent the 95% confidence interval of correlations found for alternative PHH donor sets acquired with 1000 times bootstrapping. (C) Effect of changes in MDM2 feedback strength (top), dephosphorylation rate (middle) and p53-dependent MDM2 synthesis rate (bottom) on correlation of basal TP53 expression with basal MDM2 expression (medium) or after 3.3 μM cisplatin exposure (cisplatin) at the 8-hour timepoint and with measurement noise. A value r = 1 implies no adjustment in parameter value with respect to the fitted values, r > 1 implies stronger feedback and r < 1 implies weaker feedback. Grey solid lines and shaded areas are the same as in (B). (D) Adjustments in steady state values of p53 and MDM2 upon changes in factor r.