Live-cell p53 single-molecule binding is modulated by C-terminal acetylation and correlates with transcriptional activity

Abstract

Live-cell microscopy has highlighted that transcription factors bind transiently to chromatin but it is not clear if the duration of these binding interactions can be modulated in response to an activation stimulus, and if such modulation can be controlled by post-translational modifications of the transcription factor. We address this question for the tumor suppressor p53 by combining live-cell single-molecule microscopy and single cell in situ measurements of transcription and we show that p53-binding kinetics are modulated following genotoxic stress. The modulation of p53 residence times on chromatin requires C-terminal acetylation-a classical mark for transcriptionally active p53-and correlates with the induction of transcription of target genes such as CDKN1a. We propose a model in which the modification state of the transcription factor determines the coupling between transcription factor abundance and transcriptional activity by tuning the transcription factor residence time on target sites.Both transcription binding kinetics and post-translational modifications of transcription factors are thought to play a role in the modulation of transcription. Here the authors use single-molecule tracking to directly demonstrate that p53 acetylation modulates promoter residence time and transcriptional activity.

Fig. 1

Single-molecule tracking of HaloTag-p53 in response to DNA damage. a, b Characterization of MCF-7/6/Hp53 cell line a Western blot of p53 and HaloTag-p53 at different times following the induction of DNA damage by 10 Gy IR in MCF-7/6/Hp53, a stable cell line expressing HaloTag-p53, and in the parental breast cancer cell line MCF-7 (2 replicates). b Representative confocal microscopy fields of MCF-7/6/HaloTag-p53 cells before and after exposure to 10 Gy IR. HaloTag-p53 was labeled with HaloTag-TMR fluorescent ligand. A small fraction of cells displays high HaloTag-p53 levels even when unstimulated. Scale bar 15 µm. A similar fraction of p53-positive cells can be identified in the parental cell line by immunofluorescence. Scale bar 15 µm. c–e Single-molecule tracking of HaloTag-p53. c The average projection of the images allows the identification of the cell nucleus and of the sites of relatively stable immobilization of the single molecules, which appear as bright spots. Scale bar 5 µm. d The movies were tracked to compute the distribution of single-molecule displacements (n _cells  = 8, n _{displacements}  = 8876 for 0 h and 7973 for 2 h). The distributions were fitted with a three-component diffusion model, where the slowest diffusion component is representative of chromatin-bound molecules. The insets show the cumulative distribution of displacements. Fitted parameters are shown in Table 1. e The experiments above were repeated at different times following the induction of DNA damage to measure the diffusion coefficients and the fraction of molecules in the bound state and in the two free states (3 replicates, n _cells = 17, 9, 23, 26, 22, n _{displacements} = 18 609, 8704, 41 854, 38 155, 22 889 for 0, 1, 2, 3, and 4 h after DNA damage, error bars: SD)

Fig. 2

Measurement of the p53 residence time by kymograph analysis. a In kymograph representation, chromatin-bound transcription factors (white rectangles) appear as horizontal segments parallel to the temporal axis: the duration of each binding event is provided by the length of the segment. Typically, 2 h after irradiation p53 displays longer binding events than in unstimulated conditions. b The measured binding events are used to populate the complement cumulative distribution function (3 replicates, n _cells = 18, 23, and 20, n _bound = 108, 171, 81 for 0 h, 2 h, and 4 h, respectively) which is then corrected for the observational photobleaching and fit by a bi-exponential distribution (See Supplementary Methods). c The model provides estimates for the average residence time of p53 to chromatin and combined with the information on p53 bound fraction (Fig. 1) allows estimating the average free time between binding events. d The bi-exponential fit also provides estimates for the residence time of the short-lived population, for the long-lived residence time, and for the fraction of molecules in each of these states. The time that p53 spends searching for these stable sites is calculated as described in the Supplementary Methods (error bars: 95% CI)

Fig. 3

An increase in p53 levels is not sufficient in modulating p53-binding kinetics but p53-CTD acetylation is also necessary. a Western blot of HaloTag-p53 total protein and acetylated at lysine 382 upon exposure to 10 Gy IR or Doxycycline. Quantifications of western blots are provided in Supplementary Fig. 4. b Complement cumulative distribution of single-molecule displacements at different times upon induction of HaloTag-p53 expression by doxycycline and measured bound fraction (inset, 3 replicates, n _cells = 25, 19, 25, 9, n _{displacements} = 14 278, 14 420, 10 807, 11 227 for 0 h, 2 h, 4 h, and 24 h after doxycycline, respectively. c Distribution of residence times and average residence times (inset) following doxycycline induction (n _bound = 80, 180, 67, 80). d Western blot of HaloTag-p53 total protein and acetylated at lysine 382 upon irradiation and upon the combination of IR and Wortmannin. Quantifications of western blots are provided in Supplementary Fig. 4. e Cumulative distribution of displacements (2 replicates, n _cells = 17, 15, 15, n _{displacements} = 17 592, 9846, 17 272 for Ctrl, IR, IR+Wort, respectively) and f the complement cumulative distribution of residence times (n _bound = 115, 107, 172) upon irradiation and treatment with Wortmannin (error bars: SD for bound fractions, 95% CI for average residence times)

Fig. 4

Dissecting the role of CTD acetylation in binding. a Scheme of the treatments and p53 mutants. Expression levels of (HaloTag)-p53 upon the different treatments are shown. b Cumulative distribution of displacements (top) and complement cumulative distribution function of the single-molecule residence times (bottom) for control MCF-7/6/Hp53 cells, and cells transfected with a short-hairpin inhibiting the methyltransferase Set8 (3 replicates, n _cells = 22, 23 for CTRL and Set8 Sh1, respectively). c Distribution of displacements (top) and complement cumulative distribution function of the single-molecule residence times (bottom) for HaloTag-p53-wt, HaloTag-p53-K382Q, or HaloTag-p53-6Q (acetylation mimicking mutants) transiently transfected in H1299 cells (insets, 3 replicates, n _cells = 28, 24, and 18 for WT, K382Q, and 6Q, respectively). d Distribution of displacements for HaloTag-p53-wt (top, 2 replicates, n _cells = 19 for no irradiation and 16 for 10 Gy IR) and p53-K382R (acetylation inhibiting mutant, bottom, 2 replicates, n _cells = 16 for no irradiation and 15 for 10 Gy IR) transiently transfected in H1299 cells before and 2 h after irradiation (error bars: SD for bound fractions, 95% CI for average residence times)

Fig. 5

Population and single-cell transcriptional response to the modulation of p53-binding kinetics. a qPCR of p53 targets following irradiation (3 replicates, error bars: SD). b qPCR of p53 targets following induction of HaloTag-p53 expression by doxycycline (3 replicates, error bars: SD). c–g smFISH imaging of CDKN1a. c smFISH is performed by hybridizing multiple labeled oligonucleotides to the specific RNA and acquiring 3D stacks to count mature RNAs (white square) and nascent RNAs at transcription sites (red square, maximum projection displayed). d Average amount of mature (top panel) and nascent (bottom panel) RNA per cell (top panel, 2 replicates, n _cells = 91, 121, 64 for 0 h, 2 h and 4 h, respectively; ANOVA-Tukey test). The measurement of nascent CDKN1a RNA was validated by qPCR, using primers targeting pre-spliced CDKN1a RNA (See Supplementary Fig. 4) e Correlation of cell-by-cell number of nascent transcripts vs. nuclear intensity of HaloTag-p53. Nascent CDKN1a is correlated with p53 levels 2 h after the induction of damage (Pearson correlation, r = 0.55, p < 0.0001, slope = 0.0043) but not before the induction of damage (r = 0.12, p > 0.1), nor 4 h after (r = 0.27, p > 0.1), f Number of detected active CDKN1a transcription sites per cell (left panel) and amount of nascent CDKN1a RNA per active site (right panel) (2 replicates, n _cells  = 91, 121, 64 for 0 h, 2 h, and 4 h, respectively; ANOVA-Tukey test). g Exemplary smFISH for the treatments with Doxycycline and IR+Wortmannin (top-left). Correlation between p53 residence time, number of active CDKN1a transcription sites per cell and p53-CTD acetylation across the tested conditions (top-right). The radius of the displayed dots is proportional to the amount of HaloTag-p53-Ac382. Correlations coefficients are shown in Table 2 (error bars: SEM for FISH data, 95% CI for residence times)