Prolonged DNMT1 absence induced proliferation defects and cell cycle arrest in the G1-S phase of the cell cycle in normal human cells

To study DNA hypomethylation effects on normal cells, we took advantage of the RPE-1 cell line (Retinal Pigment Epithelial cells) engineered to express endogenous DNMT1 tagged with both mNeonGreen and auxin-inducible degron AID (RPE-1 NADNMT1) [32]. Upon the addition of auxin indole Acetic Acid (IAA), the system allowed DNMT1 degradation (Fig. 1A, Suppl. Fig. 1A, B). Nevertheless, by live microscopy observation, we found that approximately 0.8% of the cells did not respond to IAA (escapers) (data not shown), which could be responsible for the low residual level of DNMT1 in the immunoblot (Suppl. Fig. 1B).

We first assessed DNMT1's importance in cell proliferation and cell cycle over a timeframe of 7 days upon IAA treatment. RPE-1 NADNMT1 cells underwent a progressive slowing-down in cell proliferation, which became significant over time (Fig. 1B). To ensure that the presence of escapers in the cell population did not alter our results, we analyzed the growth of one clone obtained from a single cell of the population (clone5). The clone5 treated with IAA showed a slowing down of cell proliferation as well as the entire population (Supp. Fig. 1C).

To assess whether the observed slow cell proliferation was due to cell cycle arrest, flow cytometry analysis with Propidium Iodide staining was performed. The histogram in Fig. 1C shows the progressive accumulation of cells with 2n DNA content starting from the third day of treatment. This result suggests possible cell cycle arrest in the G1 phase or at the G1/S transition. To quantify cells actively replicating DNA (S phase) EdU assay was performed on RPE-1 NADNMT1 cells upon IAA treatment. As shown in Supp. Fig. 1D, the percentage of cells in the S phase did not change in IAA-treated cells compared with that in the untreated control. However, we took advantage of the pattern of the incorporated EdU to identify cells in different stages of the S phase [42] (Fig. 1D). The graph in Fig. 1E reveals a significant increase in the percentage of cells in the early S phase, starting from the third day of treatment (Fig. 1E, Supp. Fig. 1E), which is consistent with the accumulation of cells with a 2n DNA content (Fig. 1C). To further confirm this, we synchronized the cells in the G1 phase (2n DNA content) by serum starvation for 48 h, and then released the cells in the presence or absence of IAA for an additional 4 and 6 days (Supp. Fig. 2A), and evaluated the cell cycle progression with flow cytometry (Supp. Fig. 2B). These results showed that once synchronized, RPE-1 NADNMT1 cells had mainly a 2n DNA content (~ 80%) after release in IAA for 4 and 6 days (Release IAA 4d and 6d). In contrast, the cells released for 4 days in normal media (Release NT) re-entered the cell cycle with a profile similar to that of asynchronous cells (NT async).

Taken together, these results strongly suggest that RPE-1 NADNMT1 cells arrest their proliferation at the G1/S transition when DNMT1 absence is prolonged.

To confirm the cell cycle arrest of RPE-1 NADNMT1 cells, the mitotic index was also evaluated (Fig. 1F) which showed a significant reduction in the RPE-1 NADNMT1 cell population in mitosis after 4 days of IAA treatment (Fig. 1G).

Cell cycle arrest induced by DNMT1 prolonged absence is due to p21 p53-dependent activation

It was previously observed that DNMT1 acute depletion induces activation of the p53-p21(waf1/cip1) pathway via p14ARF upregulation [12]. Similarly, we observed an increase in p21, both at the transcriptional and protein levels, in RPE-1 NADNMT1 cells from the third day of treatment (Fig. 2A, B, D). Instead, while the total fraction of p53 protein did not change upon the treatment (Suppl. Fig. 1F), the fraction of activated p53 phosphorylated at S15 increased from the fourth day of IAA treatment (Fig. 2A and C). However, the p14ARF transcript level increased only on the 7th day of treatment (Suppl. Fig. 1G), suggesting that its activation and involvement in p53 and p21 triggering is not an early event following DNMT1 protein degradation.

To test the involvement of p53 in p21 upregulation, we knocked down p53 by RNA interference during IAA treatment (Fig. 2E). Following p53 reduction, the p21 protein increase in IAA-treated cells was reduced (from 2.2 of IAA siLuc to 1.5 of IAA sip53) (Fig. 2F–G and Suppl. Fig. 1H). However, the p21 protein level was not lowered down to that of the untreated condition, suggesting that either the residual p53 (Fig. 2F) keeps activating p21, or p53 could not be the only regulator of p21 activation in a DNMT1-depleted context. Nevertheless, MPM2 staining revealed that, while only 1.7% of RPE-1 NADNMT1 control cells treated with IAA were in G2 phase/mitosis (G2/M) (in accordance with the mitotic index shown in Fig. 1G), 8.7% of the p53-depleted cells (sip53) were in G2/M when treated with IAA (Fig. 2H, I). Importantly, the percentage of IAA-treated RPE-1 NADNMT1 sip53 cells in the G2/M was similar to that of the untreated control cells (Fig. 2I). These data indicated that p53 reduction was sufficient to rescue the proliferation arrest observed following DNMT1 prolonged degradation.

Fig. 2

DNMT1 degradation induces cell proliferation arrest that is dependent on p53 and p21. A Immunoblot showing the level of p21 protein and phosphorylated p53 on serine 15 (P-p53(S15)) at the indicated time points during IAA treatment. B Quantification of A showing p21 increase during IAA treatment. The graph shows the mean of at least 3 independent experiments. The error bars represent the standard error of the mean, SEM. Unpaired t-test between the samples: **** < 0.0001; ** < 0.01; * ≤ 0.05. C Quantification of A showing P-p53(S15) increase upon IAA treatment. The graph shows the mean of at least 3 independent experiments. The error bars represent the standard error of the mean, SEM. Unpaired t-test between the samples: * ≤ 0.05. D Histogram summarising RT-qPCR results of p21 transcript levels in RPE-1 NADNMT1 cells during IAA treatment. The graph shows the mean of at least 3 independent experiments. The error bars represent the standard error of the mean, SEM. Unpaired t-test between the samples: ** < 0.01; * ≤ 0.05. E Schematics showing how sip53 experiments were performed. F Representative immunoblot of p21 and p53 in the indicated samples. G Graph summarising the results of F shows that sip53 transfection reduces p21 protein level in RPE-1 NADNMT1 cells upon IAA treatment. The graph shows the mean of at least 3 independent experiments. The error bars represent the standard error of the mean, SEM. Unpaired t-test between the samples: **** < 0.0001; * ≤ 0.05. H Representative images of immunofluorescence assay targeting MPM2 in IAA-treated RPE-1 NADNMT1 cells transfected with siLuc (control) or with sip53. Nuclei were counterstained with DAPI. I The graph is the quantification of H. The graph shows the mean of 3 independent experiments. At least 100 nuclei for each sample were evaluated. The error bars represent the standard error of the mean, SEM. Unpaired t-test between the samples: ** < 0.01; * ≤ 0.05; ns > 0.05

DNMT1 rescue restores normal levels of p21 and cell proliferation arrest

We then sought to evaluate whether or not cell proliferation defects were reversible. We, thus, washed out IAA and checked the effect of NADNMT1 rescue after 24 h (Fig. 3A, B and Supp. Fig. 2C). We observed the reduction of p21 protein in the washout (WO) condition with respect to both IAA- and untreated samples suggesting that DNMT1 recovery even for only 24 h is sufficient to shut down p21 activation (Fig. 3C, D).

Fig. 3

DNMT1 restoration rescues cell proliferation. A Schematics showing how washout experiments were performed. B Immunoblot shows NADNMT1 successfully recovered upon 1 day of IAA washout. C Immunoblot of p21 level in the untreated, IAA-treated, and washed-out RPE-1 NADNMT1 cells. D Graph summarising the results in C highlights that NADNMT1 rescue reduces p21 level in RPE-1 NADNMT1 cells. The graph shows the mean of 3 independent experiments. The error bars represent the standard error of the mean, SEM. Un-paired t-test between samples ** < 0.01; * ≤ 0.05; ns > 0.05. E Graph summarising EdU staining data in the washout experiments. The graph shows the mean of 3 independent experiments. The error bars represent the standard error of the mean, SEM. At least 200 nuclei for each condition were evaluated. Unpaired t-test between samples: * ≤ 0.05; ns > 0.05. F Graph summarising data of MPM2 staining in the untreated, IAA-treated, and washed-out conditions. MPM2-positive cells increase after NADNMT1 rescue. The graph shows the mean of 3 independent experiments. At least 150 cells were analyzed for each sample. The error bars represent the standard error of the mean, SEM. Unpaired t-test between the samples: * ≤ 0.05; ns > 0.05. G Representative images of mitotic spreads of untreated, IAA-treated, and washed-out RPE-1 NADNMT1 cells. The nuclei and chromosomes were stained with Giemsa. Red arrows indicate cells in mitosis. H Graph summarising the data in G. Increased mitoses are observed in the IAA-washout condition. The graph shows the mean of 3 independent experiments. The error bars represent the standard error of the mean, SEM. At least 450 cells were evaluated. Unpaired t-test between the samples: ** < 0.01; ns > 0.05

We, then, analyzed EdU staining and observed no significant difference in the percentage of cells in the S phase or early S after IAA washout with respect to the untreated cells, suggestive of no changes in the cell cycle progression (Fig. 3E and Supp. Fig. 2D). However, the reduction of the percentage of washed-out cells in the early S phase was not statistically different from the one of IAA treatment, probably suggesting that 1 day of DNMT1 rescue might be not enough to completely recover the G1/S arrest.

To have a better understanding of the progression of the cell cycle after IAA washout, we evaluated G2 and mitosis. MPM2 staining showed no significant difference in the percentage of cells in G2/M between the washout and the untreated samples, suggesting a recovery of cell proliferation when DNMT1 is restored (Fig. 3F). However, there was no significant increase in the percentage of cells in G2/M between the washout and the IAA-treated conditions (Fig. 3F). On the other hand, by analyzing the mitotic index, we observed a full recovery of the cells in mitosis after IAA washout with respect to the IAA treatment (Fig. 3G, H). Also, the mitotic index was comparable to the one of untreated RPE-1 NADNMT1 cells. To evaluate the possibility that 1 day of release from IAA might be not enough to fully restore the cell cycle, we repeated the experiments on the RPE-1 NADNMT1 cells clone5 extending the wash out to 2 days and we named this condition WO2d (Supp. Fig. 2E). Immunoblotting confirmed the complete rescue of DNMT1 (Supp. Fig. 2F) and MPM2 staining also revealed a significant increase in the percentage of cells in G2/M upon two days of release from IAA (WO2d) with respect to the IAA-treated condition (Supp. Fig. 2G). These results suggest that the proliferation arrest induced by prolonged DNMT1 absence can be progressively rescued by DNMT1 restoration.

DNMT1 inducible degradation causes global DNA demethylation

To evaluate whether DNMT1 inducible degradation affects global DNA methylation levels, a 7-day time-course analysis was carried out on RPE-1 NADNMT1 cells (Fig. 4A, B). Similar to previous observations in DLD-1 NADNMT1 cells [32], 5-methylcytosine (5MeC) immunostaining showed a progressive reduction in global DNA methylation starting from 2 days after IAA addition. The maximum degree of methylation loss is reached after 4 days of treatment, remaining unchanged up to 7 days of treatment. This finding is in accordance with the G1/S cell cycle arrest at both 4 and 7 days after IAA addition (Fig. 1C, E).

Fig. 4

DNMT1 inducible degradation causes global DNA demethylation. A Representative images of immunostaining for 5-MethylCytosine (5MeC) displaying global DNA methylation reduction from 2 days up to 4 days of IAA treatment. Nuclei were counterstained with DAPI. B Single-cell nucleus quantification of 5MeC signal (AU: arbitrary unit) without (0 days) or with IAA (2, 3, 4 and 7 days) treatment. The scatter plot is cumulative of 6 independent experiments with at least 150 cells counted for each condition. Error bars represent the standard error of the mean SEM. Un-paired t-test between the samples: **** < 0.0001; ** < 0.01; * ≤ 0.05; ns > 0.05

Given the association between global DNA demethylation and DNA damage signaling [11, 16, 43], we also wanted to address whether decreased levels of 5MeC could possibly trigger phosphorylation of histone H2AX on serine 139 (γH2AX), one of the earliest markers of the DNA damage response. However, we did not find any increase of γH2AX by immunoblotting (Supp. Fig. 2H), as we previously reported [12], suggesting no occurrence of DNA damage.