Identification of TopBP1 condensation inhibitors by high throughput screening.

(A) Schematic description of the high-throughput screening system. Doxycycline-inducible TopBP1 was fused to mCherry and the light-sensitive cryptochrome 2 (Cry2) at its C-terminus and stably integrated in Flp-In 293 T-Rex cells. A 5-min blue light exposure (cycles of 4 s on and 10 s off) allows inducing optoTopBP1 condensation in the absence of DNA damage. Cells were grown in 384-well plates, incubated with 10 µM TargetMol molecules for 2 h before inducting optoTopBP1 condensate formation. (B) Graphical representation of the screening results. Drugs with a z-score lower than –2 were considered inhibitors, and drugs with a z-score higher than 2 were considered activators of TopBP1 condensate formation. The screening was performed in triplicate and each dot represents the mean z-score of a molecule. (C) Graphical representation of the screening confirmation in HCT116 cells. Chk1 phosphorylation at S345 was assessed using Celigo® immunofluorescence imaging aver 2 h co-incubation with SN-38 and each of the 131 drugs selected from the first screening. Drugs leading to Chk1 phosphorylation (pChk1) inhibition (>1.5-fold change) were considered promising candidates and the ten molecules leading to the highest inhibition were selected for the next screening. (D) Viability and synergy matrices obtained aver HCT116 cell incubation with increasing concentrations of FOLFIRI (5FU from 0.009 to 0.148 µM, SN38 from 0.077 to 1.235 nM) and each of the ten drugs (MCB-613 13.3 to 850 nM, OTSSP167 7.8 to 500 nM, AZD2858 13.3 to 850 nM, P005091 31.5 to 1000 nM, Tubercidin 0.15 to 10 nM, Dactolisib 1.58 to 100 nM, Physalin F 6.25 to 400 nM, PFK158 1.2 to 300 nM, PF5622711.2 to 300 nM, Torkinib 0.4 to 300 nM) for 96 h. Cell viability was assessed with the SRB assay. Blue matrices represent cell viability. The black/red/green matrices represent additivity/synergy/antagonism, respectively. The synergy matrices were calculated with an R script (see Materials and Methods).

AZD2858 inhibits the formation of endogenous TopBP1 condensates and the ATR/Chk1 signaling pathway.

(A) Representative immunofluorescence images (upper panels) of TopBP1 condensates in HCT116 cells incubated with AZD2858 (100 nM) or/and SN-38 (300 nM) for 2 h, and the corresponding quantification (lower panel). Scale bars: 10 µm. The experiment was replicated 3 times. Cell profiler was used for quantifying TopBP1 foci (>1000 cells analyzed per condition). Statistical significance was first assessed using ANOVA (p-value < 2.2e-16). The Mann-Whitney test, represented in the figure, was then used to specifically compared SN-38 and AZD2858+SN-38 conditions (****: p < 0.0001). For more details on the analyses, refer to the Materials and Methods section. (B) Immunoblot of the indicated proteins aver incubation of HCT116 cells with AZD2858 at the indicated concentrations and/or SN-38 (300 nM) for 2 h. The experiment was replicated 3 times, and a representative replicate is shown. (C) Percentage of HCT116 cells positive for phosphorylated Chk1 (S345) or γH2AX expression in function of the AZD2858 concentration (0 nM, 100 nM, 500 nM, 1000 nM) in the presence (blue) or not (black) of FOLFIRI (dilution: 1/2, corresponding to 6 µM of 5-FU and 50 nM of SN-38). Cells were incubated for 2 h and positive cells were identified using Celigo® immunofluorescence imaging. The experiment was replicated 3 times, and each point represents a biological replicate. The statistical significance was determined by linear modeling interrogating the effect of each drug separately and their interaction (p-values given in the inset). (D) Immunoblotting of TopBP1 and ATR isolated with streptavidin beads from optoTopBP1-expressing cells incubated with doxycycline for 16 h to induce optoTopBP1 expression, and also with AZD2858 (1 µM) and/or SN-38 (300 nM) for the last 2 h. Biotin was added to the medium in all conditions for the last 30 min. Bands that correspond to endogenous and optoTopBP1 proteins are indicated with one and two stars (* and **), respectively. The experiment was replicated twice, and a representative result is shown. (E) Chromatin was extracted at the indicated time points (min) following the assembly of nuclei from sperm DNA incubated in X. laevis egg extracts incubated with DMSO (control), camptothecin (CPT; 55 µM) or/and AZD2858 (1 µM). Chromatin samples were analyzed by western blotting.

AZD258 inactivates the S-phase checkpoint in SN-38-treated cells.

(A) Cell cycle analysis aver a 2-h incubation with AZD2858 (100 nM) and/or SN-38 (300 nM); 10 µM BrdU was added in the last 15 min of the 2-h incubation. Cell debris were gated out and BrdU incorporation was plotted against DNA content (stained with PI). Red arrows indicate BrdU incorporation. The gates for the S-phase population (BrdU-positive cells) were set broadly to prevent bias and ensure inclusion of cells with weak BrdU incorporation, particularly in the SN-38-only condition. The percentage of cells within this gate remains comparable across conditions, even though the FACS plot’s overall shape changes, reflecting a shiv in BrdU incorporation distribution and highlighting a heterogeneous population with varying levels of BrdU incorporation, especially in the combination treatment. 20,000 events per condition were analyzed. The G0/G1, S and G2/M gates are shown. The experiment was replicated 3 times, and a representative replicate is shown. More information is available in Table S2. (B) DNA fiber analysis of replication tracts labeled with two sequential 15-min pulses of IdU and CldU added at the end of the 2-h incubation with AZD2858 (100 nM) +/-SN-38 (300 nM). The dot plots show the CldU tract length of individual replication forks. Data are pooled from n = 3 biological replicates (yellow, blue, green). Data distribution is shown as half-violin plots. Circles represent the mean of each replicate, and the error bars represent the SEM of the means of the three replicates. Statistical significance was first assessed using ANOVA (p-value < 2.2e-16). The t-test, represented in the figure, was then used to specifically compared the indicated conditions (ns: non-significant; *: p<0.05; ***: p<0.001). For more details on the analyses, refer to the Materials and Methods section. (C) Same as (A) but aver 6 h or 12 h of incubation with AZD2858 (100 nM) and/or SN-38 (300 nM). The gate of interest (S-phase) and the corresponding cell percentages are outlined in red. More information is available in Table S3 and Figure S5C and S5D (D) PFGE analysis of DNA damage in HCT116 cells incubated with AZD2858 (100 nM) and/or SN-38 (300 nM) for the indicated times. The experiment was replicated 3 times, and a representative replicate is shown. (E) Immunoblot of the indicated proteins aver incubation with AZD2858 (100 nM) or/and SN-38 (300 nM) for the indicated times. The experiment was replicated 3 times, and a representative replicate is shown.

AZD2858 synergizes with SN-38 and FOLFIRI by inducing DNA damage and cell apoptosis.

(A) Flow cytometry analysis to quantify sub-G1 HCT116 cells aver 48-h incubation with sub-optimal doses of AZD2858 (100 nM) alone or with SN-38 (1.5 nM) or FOLFIRI (SN-38 1.5 nM; 5-FU 222 nM). The x-axis shows the DNA content (PI staining) and the y-axis the cell count. The presence of sub-G1 cells gated under the G1 peak suggests DNA fragmentation, a characteristic feature of apoptotic cell death. The experiment was replicated 3 times, and a representative replicate is shown. (B) Graph showing the number of sub-G1 cells (considered as dead cells) versus the number of live cells, according to the indicated treatments (from A). Error bars represent 3 individual biological replicates. The statistical significance was determined by linear regression analysis, more information is available in Table S4 (****: p<0.0001). (C) Immunoblot of the indicated proteins in HCT116 cells aver 48-h incubation as described in A. (D) PFGE analysis of DNA damage in HCT116 cells incubated for 48 h as described in A. (E) HCT116 cells were incubated with increasing concentrations of SN-38 (0 to 1.23 nM) and AZD2858 (0 to 1000 nM) for 96 h (2D culture system). Cell viability was assessed with the SRB assay. The synergy matrix was calculated as described in Materials and Methods. The experiment was replicated 3 times, and a representative replicate is shown.

AZD2858 synergizes with FOLFIRI on a panel of CRC cell lines, including HCT116 resistant to SN-38.

(A) HCT116 cells were incubated with increasing concentrations of FOLFIRI (5-FU from 0.009 to 0.148 µM and SN-38 from 0.077 to 1.235 nM) and AZD2858 (from 15.6 to 1000 nM). Cell viability was assessed with the SRB assay in 2D cultures to generate the viability matrix (blue). The synergy matrices (black and red) were calculated as described in Materials and Methods. (B) HCT116 cells were cultured in 3D to form spheroids and incubated with increasing concentrations of FOLFIRI (5-FU from 0.009 to 0.148 µM and SN-38 from 0.077 to 1.235 nM) and AZD2858 (from 23.4 to 1500 nM). Cell viability was assessed with CellTiter-Glo to obtain the viability matrix (blue). The synergy matrices (black and red) were calculated as described in Materials and Methods. Representative brighvield images of spheroids are shown: untreated and aver incubation with the drug concentrations giving the highest synergy score. (C) HT29, SW620 and SW480 CRC cells were incubated (2D culture) with increasing concentrations of FOLFIRI (HT29 cells: 5-FU from 0.03 to 1 µM and SN-38 from 0.07 to 2.35 nM; SW620 and SW480 cells: 5-FU: from 0.13 to 4.4 µM, SN-38: from 0.007 to 0.25 nM) and AZD2858 (from 15.6 to 1000 nM). (D) SN-38-resistant HCT116 SN-6 (six times more resistant to SN-38 than wild-type HCT116, 2D culture) were incubated with increasing concentrations of FOLFIRI (5-FU from 0.46 to 14.85 µM and SN-38 from 0.1 to 3.375 nM) and AZD2858 (from 15.6 to 1000 nM). (E) Same as in B but with SN-38-resistant HCT116 SN-6 cells. Drug concentrations were as follows: FOLFIRI (5-FU from 0.083 to 1.333 µM and SN-38 from 0.694 to 11.1 nM) and AZD2858 (from 46.8 to 750 nM). (F) Same as (D) but with SN-38-resistant HCT116 SN-50 cells (fivy times more resistant to SN-38 than wild-type HCT116). (G) Same as in B but with SN-38-resistant HCT116 SN-50 cells. Drug concentrations were as follows: FOLFIRI (5-FU from 0.75 to 12 µM and SN-38 from 6.25 to 100 nM) and AZD2858 (from 46.8 to 750 nM). Experiments were replicated 3 times, and representative replicates are shown.

TopBP1 forms more foci in SN-38-resistant HCT116 cells.

(A) Number of TopBP1 foci per cell in parental and SN-38-resistant HCT116 cells. Data were pooled from n = 3 biological replicates (yellow, blue, green). Data distribution is shown as a half-violin plots on the right. Circles represent the mean of each replicate, and the error bars represent the SEM of the means of the three replicates. Significance of the differences was tested using an ANOVA on a generalized linear model interrogating both the effect of cell line genotype and replicate identity (modeling TopBP1 foci number by a Poisson distribution). Foci numbers are significantly different between the two cell lines (p-value<2.2e-16) and across the 3 replicates of the experiment (p-value<2.2e-16). (B) Representative images of one replicate.

Impact of the AZD2858 and SN-38 combination on two nuclear biomolecular condensates.

(A) The effect of AZD2858 (100nM) or/and SN-38 (300 nM) on 53BP1 and (B) PML condensates were tested aver 2 h of incubation. Arsenic (6 µM, 2 h) was used as a positive control for PML nuclear bodies9 induction. Scale bars: 10 µm; ****: p<0.0001 (Mann-Whitney test for 53BP1 foci, Kruskal-Wallis for multiple comparison for PML foci).

Effects of AZD2858 on components of the DNA damage response.

(A) Quantification of the percentage of phosphorylated Chk2 (pChk2) (T68), and pATM (S1981)-positive cells in HCT116 CRC cells incubated with FOLFIRI (5-FU: 6 µM and SN-38: 50 nM) and increasing con-centrations of AZD2858 (from 250 to 1000 nM) for 2 h and (B) of pChk1 and γH2AX-positive cells aver incubation for 20 h. Data were obtained by immunofluorescence analysis using the <Expression analy-sis= application of the Celigo Imaging Cytometer (Nexcelom). The combination of FOLFIRI and AZD2858 is shaded in pink. The experiment was replicated 3 times, and each point represents a biological repli-cate. The statistical significance was determined by linear modeling analysis interrogating the effect of each drug separately, and of their interaction (p-values given in the inset).

AZD2858 effect on ATR signaling is not related to GSK-3β activity.

(A) Immunoblot of the indicated proteins aver incubation of SW620 cells that express shLuc or shGSK-3β with AZD2858 (100 nM) or/and SN-38 (300 nM) for 2 h. (B) Immunoblot of indicated proteins aver incubation of HCT116 cells with AZD2858 (100 nM) or/and SN-38 (300 nM), or insulin (0.5 µM) (positive control of GSK-3β inhibition).

Cell cycle profiling of HCT116 cells incubated with AZD2858 or/and SN-38.

(A) Cells were pulsed with BrdU for 15 min before the treatments, then chased without BrdU for the 6 hours of treatment, thereby named Pulse chase experiment (Non treated, AZD2858 100nM, SN-38 300nM, combination of AZD2858 and SN-38). Cellular debris were gated out and the level of BrdU incorporation was plotted against the DNA content. 20,000 events per condition were analyzed. BrdU-positive cells include “MP” (mitotic passage, cells that were in S phase during the pulse, completed mitosis, and returned to a 2N DNA content), “S” (cells in S phase during the pulse and still in S phase at the endpoint), and “G2” (cells in S phase during the pulse that progressed to 4N DNA). Among the BrdU-negative cells, G0/G1 cells had a 2N DNA content aver the chase. BrdU-negative cells with >2N DNA content entered S-phase aver the BrdU pulse and were gated <SE= for S-phase entry. “G2-blocked” 4N cells were already in G2/M during the BrdU pulse and were still blocked at the endpoint. (B) Same as in A) but aver 12 h of treatment. The gate of interest, S entry and G2 phase, are outlined in green and red. The experiment was replicated 3 times, and a representative replicate is shown. (C) Graphical representation of the percentage of BrdU-positive cells aver 6 hours of treatment from three independent biological replicates (“Endpoint” refers to the protocol when BrdU is added at the end of the treatment, as indicated in the schematic illustration and corresponding to the representative replicate shown in Figure 3C). (D) Same as C) but aver 12 h of treatment. (E) Graphical representation of three independent biological replicates of the percentage of cells in the S phase entry gate in Pulse chase experiment at 12 h of treatment (corresponding to green squares in Figure S5B). (F) Same as E but for the G2 gate (corresponding to red squares in Figure S5B). Statistical significance was first assessed using ANOVA (cf Materials and Mathod for detailed analysis). The t-test, represented in thefigure, was then used to specifically compared the indicated conditions (ns: non-significant; *: p<0.05; **: p<0.01 (t-test)).

The synergistic effect of the AZD2858 + FOLFIRI combination is due to cytotoxicity.

(A) HCT116 cells were incubated with increasing concentrations of each drug for 72 h. Then, dual stain-ing with Hoechst/IP was performed to visualize dead cells (red) and live cells (green), respectively, in each well. Results were quantified using the dedicated Celigo® software and shown as percentages of dead cells and live cells relative to untreated controls. (B) HCT116 cells were incubated with increasing concentrations of FOLFIRI (5-FU from 0.009 to 0.148 µM and SN-38 from 0.077 to 1.235 nM) and AZD2858 (from 15.6 to 1000 nM) for 96 h. Then, dual staining with PI and Hoechst was performed to visualize dead cells (red) and live cells (khaki), respectively, in each well. Results were quantified using the dedicated Celigo® software: the khaki matrix represents cell survival and the orange matrix cell death, relative to untreated controls. (C) CT−26 cells were incubated with increasing concentrations of FOLFIRI (5−FU from 0.083 to 1.333 μM and SN−38 from 0.694 to 11.11 nM) and AZD2858 (from 7.8 to 500 nM) for 96 h. Then, dual staining with PI and Hoechst was performed to visualize dead cells (red) and live cells (khaki), respectively, in each well. Results were quantified using the dedicated Celigo® software; the khaki matrix represents cell survival and the orange matric cell death relative to un− treated controls. (D) CT−26 cells were incubated with increasing concentrations of FOLFIRI (5−FU from 0.083 to 1.333 μM and SN−38 from 0.694 to 11.11 nM) and AZD2858 (from 7.8 to 500 nM). Cell viability was assessed with the SRB assay in 2D cultures to generate the viability matrix (blue). The synergy matrices (black, red and green) were calculated as described in Materials and Methods. (E) CT−26 murine cell line cells were cultured in 3D. Drug concentrations were as follows: FOLFIRI (5−FU from 0.083 to 1.333 μM and SN−38 from 0.694 to 11.1 nM) and AZD2858 (from 15.6 to 1000 nM). Cell via− bility was assessed with CellTiter−Glo to obtain the viability matrix (blue). The synergy matrices (black and red) were calculated as described in Materials and Methods. Representative brightfield images of spheroids are shown: untreated and after incubation with the drug concentrations giving the highest synergy score. Experiments were replicated 3 times, and representative replicates are shown.

GSK-3 inhibitors in the TargetMol library.

The Specificity column indicates the IC50 values available in the literature corresponding to 50% of the maximal concentration needed to inhibit the GSK-3 target (except for GSK3i XIII, where only the inhibition constant, Ki, value was available). The GSK-3β specificity column indicates the specificity toward this isoform, if available. If specificity was determined, but the GSK-3β isoform was not clearly specified, a question mark is used to indicate this uncertainty (for VP3.15 dihydrobromide and AT 7519 hydro-chloride salt). ++++: < 1 nM; +++: between 1 nM and 10 nM; ++: between 10 nM and 40 nM; +: > 40 nM. Molecules that inhibited light-induced optoTopBP1 foci in the present screen are highlighted in red, and the z-score is indicated. The SN-38-induced Chk1 phosphorylation (pChk1) inhibition column indicates whether the potential GSK-3 inhibitors from the initial screen inhibit SN-38-induced Chk1 phosphorylation at S345 in HCT116 cells. N/A: Not Available.

Three biological replicates results of flow cytometry experiments of the 2-hour condition (endpoint), including the representative one shown in Figure 3A.

AZD2858: 100 nM. SN-38: 300 nM

Three biological replicates results of flow cytometry experiments of the 6– and 12-hour condition (endpoint and pulse chase), including the representative one shown in Figure 3C and Figure S5C-D.

AZD2858: 100 nM. SN-38: 300 nM

Results of the linear regression analysis (Figure 4A-B).

NT: Non−Treated. A: AZD2858. S: SN−38. F: FOLFIRI (5−FU + SN−38). The p−values of interest from Figure 4B are indicated in red.

FOLFIRI concentration ranges used for the synergy matrix assays.

All the dilutions are based on the dilution 1/1 corresponding to 12 µM 5-fluorouracil (5-FU) and 100 nM SN-38.

References of the antibodies used in the study.

WB, western blotting, IF, immunofluorescence analyses.