2026 年 51 巻 1 号 p. 37-53
In higher eukaryotic cells, genomic DNA is packaged into dynamic chromatin domains whose physical behavior is coupled to DNA transactions such as transcription and DNA repair. Although chromatin organization is altered in cancer, how oncogenic signals modulate chromatin dynamics over time remains unclear. To address this issue, we established a doxycycline-inducible carcinogenesis model in hTERT-immortalized human RPE-1 cells expressing HPV16 E6/E7, MYC(T58A), and KRAS(G12V) (EMR cells) and investigated chromatin behavior during oncogene-driven transformation. Upon induction, EMR cells displayed accelerated proliferation, loss of contact inhibition, anchorage-independent growth in soft agar, and tumor formation in nude mice. Using time-resolved single-nucleosome imaging to track local chromatin dynamics over days to weeks of oncogene induction, we found that local nucleosome motion was unchanged at 1–3 days, significantly increased at 5–7 days, and returned to parental levels by 4 weeks, despite sustained oncogene expression and stable malignant growth. To explore the basis of this transient increase, we quantified DNA damage, histone marks, and transcription. γH2AX foci were elevated in EMR cells, but ATM/ATR inhibition had only minor effects on local chromatin motion, indicating that the DNA damage response is not the principal driver. By contrast, H3/H4 acetylation and nascent RNA synthesis were upregulated specifically during the early window of enhanced dynamics, whereas the heterochromatin mark H3K9me3 decreased, consistent with transient chromatin loosening associated with increased transcription. These findings reveal a biphasic change in local chromatin dynamics during human oncogene-driven transformation and provide a physical and temporal framework for understanding how oncogenic pathways reorganize chromatin.
Key words: cancer, oncogenesis, single-nucleosome imaging, chromatin dynamics
Graphical Abstract
In human cells, ~2 meters of genomic DNA is wrapped around core histones to form nucleosomes (Koyama and Kurumizaka, 2018; Luger et al., 1997; Maeshima et al., 2014; Olins and Olins, 2003). Recent studies in higher eukaryotic cells suggest that nucleosomes are irregularly folded into variously sized chromatin domains (~100–300 nm) not only in heterochromatin, which is generally silenced with repressive histone marks, but also in transcriptionally active euchromatic regions (Iida et al., 2025; Maeshima et al., 2024; Miron et al., 2020; Nozaki et al., 2023; Paggi and Zhang, 2025; Zakirov et al., 2021). In these domains, nucleosomes fluctuate and behave like a liquid, especially in euchromatin regions (Iida et al., 2025; Maeshima, 2025; Minami et al., 2025; Rippe, 2022). Hi-C methods, which generate fine genome-wide contact probability maps (Dekker and Heard, 2015), have also revealed similar chromatin domains (contact domains or topologically associating domains, TADs) (Dixon et al., 2012; Nora et al., 2012; Rao et al., 2014; Sexton et al., 2012), which are often organized into two distinct, megabase-scale compartments (A and B), corresponding roughly to transcriptionally active (euchromatin) and inactive (heterochromatin) regions, respectively (Lieberman-Aiden et al., 2009).
The behavior of chromatin is closely coupled to DNA transactions such as transcription (Germier et al., 2017; Nagashima et al., 2019; Ochiai et al., 2015) and DNA repair (Iida et al., 2022; Seeber et al., 2018). Indeed, dysregulation of such activities is often found during carcinogenesis (Bradner et al., 2017; Flavahan et al., 2017; Hanahan, 2022; Jackson and Bartek, 2009; Lord and Ashworth, 2012; Timp and Feinberg, 2013). Both chromatin states and 3D genome organization are extensively altered during carcinogenesis, including large-scale changes in heterochromatin domains, enhancer/promoter usage, A/B compartments, and chromatin loops (Corces et al., 2018; Della Chiara et al., 2021; Fiziev et al., 2017; Johnstone et al., 2020; Liu et al., 2025; Wen et al., 2009). In our previous study, we reported that oncogenic-HRAS-transformed mouse fibroblasts (CIRAS-3 cells) have more constrained chromatin with increased amount of heterochromatin (Otsuka et al., 2024). However, more than 30 years have passed since CIRAS-3 cells were established via the introduction of oncogenic HRAS (Egan et al., 1987). Thus, we do not know when and how chromatin dynamics changed during transformation/carcinogenesis; in fact, this remains a general open question.
To approach this question, we established an inducible in vitro carcinogenesis model in human RPE-1 cells (Bodnar et al., 1998). Previous studies showed that oncogenic human papillomavirus (HPV) E6/E7, MYC(T58A), and KRAS(G12V) robustly transform human cells (Inagawa et al., 2014; Narisawa-Saito et al., 2008) and have also been used to develop carcinogenesis models in various human cells (Inagawa et al., 2014; Zhang et al., 2022). These alterations are recurrently observed in cervical adenocarcinoma (ADC) (Cancer Genome Atlas Research et al., 2017; Zhang et al., 2022). The RPE-1 cells used here express hTERT, the catalytic subunit of telomerase, which maintains telomere length and enables immortalization (Bodnar et al., 1998). KRAS(G12V) impairs GTPase activity, locking KRAS in the GTP-bound active state and driving downstream proliferative signaling (e.g., RAF–MEK–ERK, PI3K) (Singhal et al., 2024). HPV16 E6 promotes the degradation of p53 via an E3 ligase pathway (E6–E6AP/UBE3A), while E7 binds RB1 and releases E2F, pushing cells into S phase. Together, they disable key tumor-suppressor pathways and promote immortalization/oncogenesis (Moody and Laimins, 2010). MYC is a transcription factor that activates gene expression. The T58A mutation stabilizes MYC and enhances transforming activity (Das et al., 2023).
Using a doxycycline-dependent inducible system (Tet-On Advanced) (Urlinger et al., 2000), we controlled the expression of HPV16 E6/E7, MYC(T58A), and KRAS(G12V). With these inducible transformed cells (EMR cells), we tracked changes in chromatin state, including local chromatin dynamics, by live-cell single-nucleosome imaging and tracking (Ide et al., 2022; Iida et al., 2022; Itoh et al., 2021; Lakadamyali, 2022; Lerner et al., 2020). We found that local chromatin motion was transiently increased, coincident with increased histone acetylation and transcription, and then returned to the original level during carcinogenesis. These findings imply that chromatin state undergoes a transient change during the carcinogenesis process and may help to elucidate the multi-step nature of tumor development.
To establish an inducible in vitro carcinogenesis model, we chose hTERT RPE-1 cells (Bodnar et al., 1998) that stably express H2B-HaloTag (H2B-Halo) (Shimazoe et al., 2025), because these widely used immortalized non-cancer cells have a nearly diploid karyotype (Bodnar et al., 1998). We introduced a Tet-On Advanced system (Urlinger et al., 2000) with inducible HPV16 E6/E7, MYC(T58A), and KRAS(G12V) oncogenes (Inagawa et al., 2014; Narisawa-Saito et al., 2008) into these cells (Fig. 1A). Induction was confirmed by western blotting, which showed robust expression of E6/E7, MYC(T58A), and KRAS(G12V) after 24 h of doxycycline (Dox) treatment (Fig. 1B). The established cell line was designated as EMR (HPV16 E6/E7, MYC(T58A), and KRAS(G12V)). Next, we examined the growth of EMR cells upon Dox induction. Because RPE-1 cells are non-cancerous, parental cells ceased proliferation at ~100% confluence due to contact inhibition (Fig. 1C). By contrast, Dox-treated EMR cells continued to proliferate after reaching confluence, whereas vehicle-treated (Milli-Q) EMR cells did not (Fig. 1D), indicating a bypass of contact inhibition in Dox-treated EMR cells.
Establishment of a doxycycline-inducible carcinogenesis model (EMR) in hTERT RPE-1 cells
(A) Schematic of the strategy used to establish a doxycycline-inducible carcinogenesis model in hTERT RPE-1 cells (EMR cells). Downstream signaling of HPV16 E6/E7, MYC(T58A), and KRAS(G12V) are also shown. (B) Western blots showing the expression of HPV16 E6, MYC, KRAS, and Lamin A/C in parental and EMR cells treated with vehicle (Milli-Q) or 1 μg/mL Dox for the indicated days (0, 1, 3, 5, and 7). The asterisk indicates the uncleaved fusion protein (HPV16 E6/E7–T2A–MYC–P2A–KRAS) before self-cleavage at the T2A and P2A sites. Uncropped data are provided in Fig. S4. (C and D) Cell growth curves of parental cells (C) and EMR cells (D) treated with vehicle (Milli-Q) or 1 μg/mL Dox. Dox or vehicle was added 2 days after seeding (black arrowhead).
Dox-treated EMR cells formed colonies in soft agar within 2 weeks, while parental and vehicle-treated EMR cells did not (Fig. 2A, B). These results suggest that induced EMR cells acquired transformation-associated growth properties. To evaluate the tumor-forming ability of EMR cells in vivo, we injected parental or EMR cells into the dorsal flank of BALB/c nu/nu mice (Fig. 2C). Mice received drinking water containing Milli-Q (0.1%) or Dox (1 mg/mL) throughout the study. Over 27 days, Dox-treated, EMR cell-injected mice developed progressively larger tumors (Fig. 2C, D). By contrast, EMR cell-injected mice that did not receive Dox and mice injected with parental RPE-1 cells did not develop detectable tumors during the observation period (Fig. 2D). These results indicate that Dox-induced EMR cells are tumorigenic in vivo and support the utility of this inducible carcinogenesis model in RPE-1 cells.
Doxycycline-induced tumorigenic properties of EMR cells in vitro and in vivo
(A) Representative images of parental and EMR cells cultured in soft agar with vehicle (Milli-Q)(upper panels) or 1 μg/mL Dox (lower panels). Images were taken on the day of seeding (0 weeks) and after 2 weeks of culture. (B) Quantification of colonies containing three or more cells after 2 weeks in soft agar. Horizontal bars indicate the mean of triplicate experiments. Differences between vehicle- and Dox-treated EMR cells were significant by Welch’s two-sample t-test (p = 0.017); N.S., not significant. (C) Representative nude mice 42 days after subcutaneous injection of parental (left) or EMR (right) cells. Mice were given drinking water containing vehicle (Milli-Q) (upper panels) or 1 mg/mL Dox (lower panels). (D) Tumor-forming ability of parental and EMR cells in nude mice. Parental cells are plotted with broken lines and EMR cells with solid lines; vehicle-treated tumors are shown in black and Dox-treated tumors in red. Cells were subcutaneously injected on day 0, and tumor size was measured every 3–4 days. Mice were supplied with drinking water with or without 1 mg/mL Dox from day 0. Each point represents the mean tumor volume ± standard deviation (n = 6).
Next, we examined the growth of EMR cells upon Dox induction in more detail (Fig. S1). After 5 days of Dox treatment, EMR cells grew faster than parental cells and vehicle-treated (Milli-Q) EMR cells (Fig. S1A). Flow cytometry was then performed (Fig. S1B). Whereas the cell-cycle profile of parental cells was similar with or without Dox, the G1-phase fraction decreased and the S/G2 fraction increased in Dox-induced EMR cells (Fig. S1B). Estimated durations of each cell-cycle stage showed pronounced changes in Dox-induced EMR cells (Fig. S1C). The mitotic index also increased (Fig. S1B, C), suggesting a delay in mitotic progression for Dox-induced EMR cells. Together, these data indicate that although the doubling time of Dox-induced EMR cells was shorter, their G1 phase shortened while their G2 and M phases were prolonged.
We then used confocal microscopy and 3D reconstruction to measure nuclear volume (Fig. S1D), which has long been recognized as a useful morphological indicator of carcinogenesis and transformation status (Zink et al., 2004) (Fig. S1D). Nuclear volume was significantly increased in Dox-treated EMR cells (Fig. S1E). Because nuclear volume increases as cells progress from G1 to S and G2 (Iida et al., 2022; Maeshima et al., 2010), we adjusted nuclear volumes for cell-cycle composition (Fig. S1B). Following previous reports (Iida et al., 2022; Maeshima et al., 2010), we assumed relative volumes of 1, 1.5, and 2 for the G1/G0, S, and G2 phases, respectively. Based on this assumption, the cycle-weighted volume scores (“normalized nuclear volume ratio” in Fig. S1B) for parental cells without and with Dox and for EMR cells without and with Dox were 1.14, 1.12, 1.15, and 1.17, respectively. These analyses suggest comparable nuclear volumes across conditions.
Single-nucleosome imaging and tracking to study local chromatin motion in transformed cellsWe performed single-nucleosome imaging and tracking (Ide et al., 2022; Iida et al., 2022; Lakadamyali, 2022; Lerner et al., 2020) in living Dox-induced EMR cells to investigate when and how local chromatin behavior was affected. H2B-Halo was incorporated genome-wide into nucleosomes, including euchromatic and heterochromatic regions (Kimura and Cook, 2001; Semeigazin et al., 2024; Shimazoe et al., 2025), because H2A–H2B dimers can also undergo replication-independent exchange within a few hours (Kimura and Cook, 2001) (presumably via chaperones such as FACT/NAP1) (Aguilar-Gurrieri et al., 2016; Belotserkovskaya et al., 2003). As in our previous studies with human RPE-1 cells (Nagashima et al., 2019), we used oblique illumination microscopy (Fig. 3A) (Tokunaga et al., 2008) and sparse nucleosome labeling (Fig. 3B). Tetramethylrhodamine (TMR)-labeled nucleosomes were imaged at 50 ms per frame in asynchronous cells (~100 frames, 5 s total) (Movie S1) and appeared as clear dots (Fig. 3C). These dots showed single-step photobleaching (Fig. 3D), indicating that each dot represents a single H2B-Halo–TMR molecule in a single nucleosome. Notably, at this frame rate only nucleosome-incorporated H2B-Halo–TMR could be tracked; free H2B-Halo diffused too rapidly to appear as discrete dots. Individual dots were fit with a 2D Gaussian function to estimate precise nucleosome positions (Betzig et al., 2006; Rust et al., 2006). The position determination accuracy for H2B-Halo dots was 8.9 nm (Shimazoe et al., 2025). Tracks were generated using u-track (Jaqaman et al., 2008), yielding nucleosome trajectories (e.g., Fig. 3E). Based on these trajectories, we quantified nucleosome motion.
Single-nucleosome imaging and local chromatin dynamics after short-term induction in parental and EMR cells
(A) Scheme for oblique illumination microscopy (HILO) (Tokunaga et al., 2008). The illumination laser (green) can excite fluorescent molecules within a thin optical layer (dotted gray line) of the nucleus and reduce background noise. (B) Scheme for sparse HaloTag labeling with TMR. (C) Single-nucleosome (H2B-Halo-TMR) images of living parental and EMR cells after subtracting background. (D) Single-step photobleaching of two representative nucleosome (H2B-Halo-TMR) dots for parental and EMR cells. The vertical axis represents the fluorescence intensity of individual TMR dots. (E) Representative three trajectories of tracked single nucleosomes in EMR cells. (F, upper) Mean MSD curves (± SD across cells) of nucleosome motion in parental cells from 0.05 to 0.5 s. Milli-Q-treated cells are shown in black and Dox-treated cells in red. For each condition, n = 20 cells from three independent experiments. Differences between Milli-Q and Dox were not significant (N.S.) by Kolmogorov–Smirnov test (p = 0.98 at 1 day, 0.34 at 3 days, 0.17 at 5 days, and 0.57 at 7 days). (F, lower) Mean MSD curves (± SD across cells) of nucleosome motion in EMR cells from 0.05 to 0.5 s. Milli-Q-treated cells are shown in black and Dox-treated cells in red (n = 20 cells per condition, three independent experiments). Differences between Milli-Q and Dox were not significant by Kolmogorov–Smirnov test at 1 and 3 days (p = 0.34 and 0.57, respectively), but were significant at 5 and 7 days (p = 0.035 and 0.034, respectively).
Movie S1
Parental cell, 0.1% Milli-Q, 1 day
Download VideoWe conducted time-course imaging and tracking in EMR and parental cells at 1 day, 3 days, 5 days, and 7 days post-induction. From the trajectory data, displacement distributions and the mean square displacement (MSD) of individual nucleosomes were calculated in EMR and parental cells with or without Dox (Fig. 3F). MSD quantifies how far molecules move over a given time interval. MSD curves in both cell types appeared subdiffusive, consistent with constrained nucleosome motion of the chromatin polymer and in line with previous theoretical predictions (Shinkai et al., 2016, 2017; Tortora et al., 2020) and experimental observations (Iida et al., 2022; Lerner et al., 2020; Minami et al., 2025; Nagashima et al., 2019). In parental cells, there were no significant differences between Dox-treated and vehicle-treated (Milli-Q) conditions (top, Fig. 3F), indicating that Dox itself did not affect local chromatin dynamics. In EMR cells, no significant changes were observed at 1 or 3 days of induction, whereas local nucleosome motion increased significantly at 5 and 7 days (bottom, Fig. 3F; Movies S2 and S3).
Movie S2
EMR cell, 1 μg/mL Dox, 1 day
Download VideoMovie S3
EMR cell, 1 μg/mL Dox, 5 days
Download VideoTo better understand how oncogenic induction affects chromatin dynamics, we classified the trajectory data with the Bayesian-based Richardson-Lucy algorithm (RL algorithm) (Ashwin et al., 2019; Iida et al., 2025; Shimazoe et al., 2025). H2B-Halo trajectories were classified into four motion types: “Super-slow,” “Slow,” “Fast,” and “Super-fast” subpopulations (Fig. 4A). The “Fast” and “Super-fast” subpopulations are likely euchromatic (Iida et al., 2025). Dox induction in EMR cells shifted these populations rightward (Fig. 4B, orange and red), increasing their mobility, while the other two subpopulations remained similar (Fig. 4B, light blue and light green), implying that oncogenic induction increases movements of nucleosomes mainly in euchromatin.
Nucleosome motion classified into subpopulations with the RL algorithm
(A) Distributions of MSD (Δt = 0.5 s) from H2B-Halo trajectories of Milli-Q-treated parental (left) and EMR (right) cells were denoised using the RL algorithm and fitted with four Gaussian components: “Super-slow” (light blue), “Slow” (light green), “Fast” (orange), and “Super-fast” (red). Blue lines indicate the denoised data, while dotted curves represent the fitted Gaussians. The peak of each Gaussian is marked with a dotted line. (B) MSD distributions of Dox-treated parental (left) and EMR (right) cells. Notably, only the “Fast” (orange) and “Super-fast” (red) peaks in EMR cells shifted rightward upon Dox treatment, indicating further acceleration of these subpopulations.
To explore why local nucleosome motion increased at 5–7 days (Fig. 3F), we considered two possible mechanisms: (i) increased DNA damage and activation of the DNA-damage response (DDR), and (ii) increased acetylation in euchromatin/active chromatin.
First, we asked whether the DDR contributes to the increased nucleosome motion. Immunofluorescence for γH2AX, a DNA damage marker (Rogakou et al., 1998), was elevated in Dox-induced EMR cells (Fig. S2A, B), indicating enhanced DNA damage/DDR signaling. Previous studies have shown that activation of DDR signaling increases local chromatin dynamics, and that inhibition of the key DDR kinases ATM and ATR suppresses this increase (Iida et al., 2022; Seeber et al., 2018). We therefore treated Dox-induced EMR cells with KU-55933 (ATM inhibitor) and VE-821 (ATR inhibitor) and measured MSD (Fig. S2C). MSD values were only slightly reduced and not significantly different from Dox + DMSO controls, indicating that DDR is not the primary driver of the increased local chromatin motion.
Next, we examined histone acetylation and transcription in Dox-induced EMR cells. Immunofluorescence for pan–acetyl-H3 and pan–acetyl-H4 showed increased nuclear signal after 5 days of Dox treatment (Fig. 5A–D), whereas the repressive heterochromatin mark histone H3 lysine 9 trimethylation (H3K9me3) decreased (Fig. S2D, E). Furthermore, 5-ethynyl uridine (EU) incorporation likewise increased after 5 days, indicating elevated transcription at this short-term time point (Fig. 5E, F). Histone acetylation has been reported to increase local nucleosome motion (Nozaki et al., 2017), and histone acetylation at promoters and enhancers is tightly associated with transcriptional activation (Kouzarides, 2007; Kuo et al., 1998). Together, these results suggest that progressive histone acetylation is likely to underlie the enhanced local chromatin dynamics.
Increased histone acetylation and transcription in Dox-induced EMR cells
(A) Representative immunofluorescence images of the euchromatin marker H3 acetylation (H3ac) in parental and EMR cells after 5 days of treatment with Milli-Q or Dox. Scale bar, 10 μm. (B) Quantification of nuclear H3ac intensity. N.S., not significant by Wilcoxon’s rank-sum test for Milli-Q versus Dox in parental cells (p = 0.35); significant for Milli-Q versus Dox in EMR cells (p<2 × 10–16). (C) Representative immunofluorescence images of the euchromatin marker H4 acetylation (H4ac) in parental and EMR cells after 5 days of treatment with Milli-Q or Dox. Scale bar, 10 μm. (D) Quantification of nuclear H4ac intensity. N.S., not significant by Wilcoxon’s rank-sum test for Milli-Q versus Dox in parental cells (p = 0.12); significant for Milli-Q versus Dox in EMR cells (p<2 × 10–16). (E) Representative images of EU labeling (nascent RNA) in parental and EMR cells after 5 days of treatment with Milli-Q or Dox. Scale bar, 10 μm. (F) Quantification of nuclear EU intensity. N.S., not significant by Wilcoxon’s rank-sum test for Milli-Q versus Dox in parental cells (p = 0.17); significant for Milli-Q versus Dox in EMR cells (p<4.7 × 10–8).
We extended the induction period to 4 weeks to examine how local chromatin dynamics evolve over time. First, expressions of E6/E7, MYC(T58A), and KRAS(G12V) were monitored (Fig. 6A). Protein levels remained elevated in Dox-induced EMR cells, indicating stable induction for at least 4 weeks. Cell growth analysis after 4 weeks (Fig. 6B) showed that only Dox-induced EMR cells proliferated faster than parental cells and vehicle-treated (Milli-Q) EMR cells, consistent with the short-term induction results (Fig. S1A). Flow-cytometric analysis (Fig. S3A) and derived metrics for doubling time and cell-cycle distribution (Fig. S3B) confirmed a persistent shift toward a shorter G1 phase and prolonged G2/M phases, which was a similar tendency observed for the short-term induction.
Long-term Dox induction and sustained oncogene expression in EMR cells
(A) Western blots showing expression of HPV16 E6, MYC, KRAS, and Lamin A/C in parental and EMR cells treated with 0.1% Milli-Q or 1 μg/mL Dox for the indicated weeks (0–4 weeks). The asterisk indicates the uncleaved fusion protein (HPV16 E6/E7–T2A–MYC–P2A–KRAS) before self-cleavage at the T2A and P2A sites. Uncropped data are provided in Fig. S5. (B) Growth curves of parental (broken lines) and EMR (solid lines) cells treated with Milli-Q (black) or Dox (red) for the indicated times. Mean ± SD of three independent experiments is shown. Doubling times at 28 days (4 weeks), calculated from exponentially fitted curves using data from days 27–29, were 21.37 h for parental + Milli-Q, 21.36 h for parental + Dox, 21.84 h for EMR + Milli-Q, and 16.85 h for EMR + Dox.
Finally, time-course single-nucleosome imaging revealed no significant differences between vehicle- and Dox-treated EMR cells after 2 weeks (Fig. 7A; Movies S4 and S5). Thus, despite sustained oncogene expression and altered growth and cell-cycle profiles, local chromatin dynamics returned to baseline during long-term induction.
Single-nucleosome imaging after long-term Dox induction and model scheme of biphasic chromatin dynamics in EMR cells
(A, upper) Mean MSD curves (± SD across cells) of nucleosome motion in parental cells from 0.05 to 0.5 s after 1, 2, 3, and 4 weeks of treatment with Milli-Q (black) or Dox (red). Differences between Milli-Q and Dox were not significant (N.S.) by the Kolmogorov–Smirnov test (p = 0.34 at 1 week, 0.081 at 2 weeks, 0.83 at 3 weeks, and 0.83 at 4 weeks). For each condition, n = 20 cells from three independent experiments. (A, lower) Mean MSD curves (± SD across cells) of nucleosome motion in EMR cells from 0.05 to 0.5 s after 1, 2, 3, and 4 weeks of treatment with Milli-Q (black) or Dox (red). Differences between Milli-Q and Dox were significant by the Kolmogorov–Smirnov test at 1 week (p = 1.1 × 10–3) but were not significant at 2 weeks (p = 0.34), at 3 weeks (p = 0.57), and at 4 weeks (p = 0.83). For each condition, n = 20 cells from three independent experiments. (B) Schematic model summarizing the biphasic change in local chromatin dynamics during Dox induction. Expression of HPV16 E6/E7, MYC(T58A), and KRAS(G12V) is induced and maintained, driving tumor growth. Local chromatin dynamics transiently increase at early time points, likely associated with increased amounts of active histone marks and transcription, and then return to baseline levels during long-term induction, possibly accompanied by metabolic reprogramming.
Movie S4
EMR cell, 0.1% Milli-Q, 4 weeks
Download VideoMovie S5
EMR cell, 1 μg/mL Dox, 4 weeks
Download VideoIn this study, we established an inducible carcinogenesis model in human RPE-1 cells expressing HPV16 E6/E7, MYC(T58A), and KRAS(G12V) (Fig. 7B). Upon Dox induction, EMR cells acquired transformation-associated growth phenotypes in vitro and became tumorigenic in vivo. Time-resolved single-nucleosome imaging revealed that local chromatin dynamics did not change at 1–3 days, increased at 5–7 days, and returned to baseline by 4 weeks despite sustained oncogene expression (Fig. 7B). During this early phase (5–7 days), histone acetylation and transcription increased, and the repressive mark H3K9me3 decreased, whereas ATM/ATR inhibition produced only minor effects on nucleosome MSD. These results reveal a biphasic temporal change in which chromatin first loosens and later re-stabilizes under persistent oncogenic signaling.
A direct interpretation is that oncogene induction triggers a rapid transcriptional surge accompanied by widespread histone acetylation, transiently fluidizing local chromatin to facilitate transcription. Over subsequent days to weeks, cells adapt their gene expression programs, and epigenetic pathways re-equilibrate, possibly through metabolic reprogramming (see below), restoring local constraints even as the proliferative phenotype persists (Fig. 7B). Here, elevated local mobility may not be required for tumor maintenance. Thus, chromatin behavior may follow a biphasic temporal change toward a more stable transformed state.
Although we observed a shift in cell-cycle composition at days 5–7, we consider that this shift is not the main driver of the increased nucleosome motion, for three reasons. First, prior single-nucleosome work in synchronized human cells showed that local nucleosome motion remains essentially constant, on average, across G1, S, and G2 (Iida et al., 2022). Second, ongoing transcription tends to constrain nucleosome motion rather than increase it (Nagashima et al., 2019). Finally, the Dox-dependent cell-cycle shift persists even at 4 weeks post-induction, when nucleosome motion has returned to pre-induction levels.
In the observed biphasic temporal change in chromatin dynamics, alterations in cellular metabolism may be involved (Fig. 7B). During carcinogenesis, many cells shift ATP production toward elevated glycolytic flux while mitochondrial oxidative phosphorylation is still retained (aerobic glycolysis/Warburg effect) (Hanahan, 2022; Liberti and Locasale, 2016; Vander Heiden et al., 2009; Ward and Thompson, 2012). As reported previously, ATP levels can modulate local chromatin dynamics (Iida et al., 2022; Maeshima et al., 2018; Nozaki et al., 2017). One possibility is that oncogene induction initially enhances glycolysis and ATP supply, transiently fluidizing local chromatin. Over time, as metabolic programs adapt and the relative contribution of oxidative phosphorylation decreases and/or ATP homeostasis is reset, chromatin dynamics may return toward baseline despite persistent oncogenic signaling. This possibility should be directly tested by future measurements of metabolic fluxes and ATP levels in EMR cells.
Finally, several limitations and future directions merit consideration. First, in this system, the three drivers were induced simultaneously in hTERT RPE-1 cells (Bodnar et al., 1998). While human carcinogenesis is widely considered a multistep process (Hanahan, 2022), it will be important to delineate the individual and combined contributions of E6/E7, MYC(T58A), and KRAS(G12V) to transformation and to the observed biphasic chromatin dynamics. Second, because H2B-Halo labels chromatin genome-wide, it will be informative to examine euchromatin and heterochromatin more specifically (Iida et al., 2025; Minami et al., 2025; Semeigazin et al., 2024), which may provide additional insight into the steps of transformation. Third, DNA damage was detected in Dox-induced EMR cells, and replication stress may underlie part of this damage (Igarashi et al., 2023). Because such replication stress can modulate chromatin state and dynamics, its contribution should be assessed explicitly in future work.
hTERT RPE-1 cells (CRL-4000; ATCC) were cultured at 37°C in 5% CO2 in DMEM (D5796; Sigma-Aldrich) supplemented with 10% fetal bovine serum (FBS) (F7524; Sigma-Aldrich). GP2-293 cells (HEK 293-based retroviral packaging cell line part of the retroviral packaging system, Takara Bio) were cultured according to the manufacturer’s instructions.
To establish RPE-1 cell lines stably expressing the H2B-HaloTag (H2B-Halo), the piggyBac transposon system was used (Shimazoe et al., 2025). The plasmid pPB-EF1α-IB-H2B-HaloTag (Nagashima et al., 2019) was co-transfected with pCMV-hyPBase (provided by the Sanger Institute under a materials transfer agreement) into RPE-1 cells. Transfected RPE-1 H2B-Halo cells were selected with blasticidin S.
The Tet-On Advanced system was introduced into the RPE-1 H2B-Halo cell line using a retrovirus transduction system. The plasmid pRetroX-Tet-On-Advanced (a gift from Dr. Yosuke Funato, Kyoto University) was co-transfected with pVSV-G (#138479, Addgene) into GP2-293 cells. After 24 hours, the medium was replaced. After an additional 48 hours, the medium containing retrovirus was collected and used to transduce RPE-1 H2B-Halo cells. Transduced cells (RPE-1 H2B-Halo, Tet-On Advanced) were selected with G418 (500 μg/mL; 16512-81, Nacalai).
Similarly, human papillomavirus 16 E6/E7, MYC(T58A), and KRAS(G12V) oncogenes were introduced into the RPE-1 H2B-Halo, Tet-On Advanced cell line with a lentivirus transduction system. The plasmid pCSII-TRE-Tight-16E6E7-T2A-MYCT58A-P2A-KRASG12V (Inagawa et al., 2014; Zhang et al., 2022) was co-transfected with psPAX2 (#12260, Addgene) and pMD2.G (#12259, Addgene) into GP2-293 cells. After 24 hours, the medium was replaced. After an additional 48 hours, the medium containing lentivirus was collected and used to transduce RPE-1 H2B-Halo, Tet-On Advanced cells, creating the EMR cell line.
Western blotTo examine the expression levels of H2B-HaloTag, TetR, HPV16 E6, MYC, and KRAS, established RPE-1 cells were lysed in Laemmli sample buffer (Laemmli, 1970) supplemented with 10% 2-mercaptoethanol (133-1457; Wako) and incubated at 95°C for 10 min to denature proteins. The cell lysates, corresponding to RPE-1 cells per lane, were subjected to 14% SDS–polyacrylamide gel electrophoresis (SDS–PAGE). Fractionated proteins were transferred to a polyvinylidene difluoride (PVDF) membrane (IPVH00010, Millipore) for western blotting using a semidry blotter (BE-320, Bio Craft). After blocking with 5–10% skim milk (190-12865, Fujifilm), membranes were incubated with primary antibodies against H2B (rabbit, 1:10,000; ab1790, Abcam), HaloTag (mouse, 1:1,000; G9211, Promega), TetR (mouse, 631108, Clontech), HPV16 E6 (mouse, 46B6, a gift from Dr. Tohru Kiyono at the National Cancer Center, Japan), MYC (rabbit, ab32072, Abcam), KRAS (rabbit, 33197S, Cell Signaling), and lamin A/C (mouse, sc-7292, Santacruz), followed by horseradish peroxidase (HRP)–conjugated secondary antibodies (goat anti-rabbit IgG, 1:10,000; 170-6515, Bio-Rad; goat anti-mouse IgG, 1:5,000; 170-6516, Bio-Rad). Chemiluminescence (WBKLS0100, Millipore) was detected using an EZ-Capture MG imaging system (AE-9300H-CSP, ATTO).
Mouse experimentAll mouse experiments were approved by the National Cancer Center Animal Ethical Committee (no. T24-016-M01). A 100-μL suspension containing 1.0 × 106 cells in a 1:1 mixture with Matrigel (356234; Corning) was subcutaneously injected at six sites on the back of male BALB/c nu/nu mice (5–6 weeks old, n = 8; Jackson Laboratory Japan, Inc.). Mice were maintained under a 12-h light/12-h dark cycle with free access to food and water. Doxycycline (Dox) was administered in the drinking water (1 mg/mL in 5% sucrose). Dox-containing water was provided from the day of cell injection and replaced every 2–3 days. Tumor size was measured twice per week using calipers, and tumor volume was estimated using the formula V = (L × W2)/2, where V is volume (mm3), L is the largest diameter (mm), and W is the smallest diameter (mm). Mice were monitored for 45 days after injection, and tumor volumes at day 42 are shown in Fig. 2. If the largest tumor diameter (L) exceeded 10 mm before day 45, mice were euthanized to minimize discomfort.
Imaging and quantification of HaloTag-labelled cellsTo examine H2B-HaloTag localization in RPE-1 cells, cells grown on poly-L-lysine-coated coverslips (P1524-500MG, Sigma-Aldrich; C018001, Matsunami) were incubated with 5 nM HaloTag TMR ligand (8251, Promega) overnight at 37°C in 5% CO2. Cells were then fixed with 1.85% formaldehyde (064-00406, Wako) at room temperature for 15 min, quenched with 50 mM glycine (077-00735, Wako) for 5 min, permeabilized with 0.5% Triton X-100 (T-9284, Sigma-Aldrich) for 5 min, washed with 1× HMK buffer (20 mM HEPES pH 7.5, 1 mM MgCl2, 100 mM KCl), and stained with DAPI (0.5 μg/mL; 10236276001, Roche) for 5 min. Samples were then mounted in PPDI mounting medium (20 mM HEPES pH 7.4, 1 mM MgCl2, 100 mM KCl, 78% glycerol, and p-phenylenediamine 1 mg/mL; 695106-1G, Sigma-Aldrich). Optical section images were acquired with a 0.2-μm z-step using a DeltaVision Personal Microscope (Applied Precision) equipped with an Olympus PlanApo N 60× oil-immersion objective (NA 1.42), a scientific CMOS (sCMOS) camera, and a standard four-color filter set. Deconvolution and projection of z-stacks covering the whole nucleus were performed using SoftWoRx acquisition and analysis software.
ImmunofluorescenceFor immunofluorescence, cells were grown on poly-L-lysine-coated coverslips and, where indicated, incubated with 5 nM HaloTag TMR ligand overnight at 37°C in 5% CO2. Fixation, quenching, permeabilization, washing in HMK buffer, DAPI staining, mounting in PPDI, and imaging conditions were the same as described above for HaloTag-labelled cells. After permeabilization and washing, cells were blocked with 3% BSA in PBS and incubated with primary antibodies against MYC (1:1,000, rabbit, Abcam, ab32072), γH2AX (1:1,000, rabbit, Abcam, ab2893), pan-acetyl H3 (1:1,000, rabbit, Millipore, 06-599), pan-acetyl H4 (1:1,000, rabbit, Millipore, 06-866), or H3K9me3 (1:1,000, gifted by Dr. Hiroshi Kimura at the Institute of Science Tokyo, Japan), followed by Alexa Fluor–conjugated secondary antibodies (1:1,000). Nuclei were counterstained with DAPI, and images were acquired and processed as described for HaloTag imaging.
Anchorage-independent cell growth analysisAnchorage-independent growth in soft agar was assessed as previously described (Zhao et al., 2015). A base layer of 0.5% agar (010-08725, Fujifilm) in DMEM containing 10% FBS and penicillin/streptomycin was poured into wells, followed by a 0.35% agar growth layer in DMEM with 10% FBS and penicillin/streptomycin. A top layer of DMEM containing 10% FBS was then added. Images were taken with an Olympus phase-contrast microscope every 7 days, and colonies containing three or more cells were scored 14 days after cell seeding.
Nuclear volume analysisNuclear volumes were measured as described previously (Iida et al., 2022). Cells were grown on poly-L-lysine-coated coverslips (P1524-500MG, Sigma-Aldrich; C018001, Matsunami). All subsequent procedures were performed at room temperature. Cells were washed with 1× PBS, fixed with 1.85% formaldehyde for 15 min, quenched with 50 mM glycine (077-00735, Wako) for 5 min, permeabilized with 0.5% Triton X-100 for 5 min, washed with 1× HMK buffer (20 mM HEPES pH 7.5, 1 mM MgCl2, 100 mM KCl), stained with DAPI (0.5 μg/mL) for 5 min, and washed three times with 1× HMK for 3 min each before being mounted in PPDI. Z-stack images (0.4-μm step, typically 20–40 sections) were acquired using a FLUOVIEW FV1000 confocal laser scanning microscope (Olympus) equipped with an UPLANSAPO 60× water-immersion objective (NA 1.20). The z-stacks were imported into Imaris (Bitplane AG, Zurich, Switzerland) and converted to Imaris 3D image files (.ims). Nuclear volumes were calculated using the “Surface” module, with intensity thresholds automatically determined to include all DAPI signals. Only well-isolated nuclei were recorded and analyzed.
Cell growth and cell cycleCells were seeded into 6-well plates at 1.0 × 104 cells/mL. The number of cells was examined microscopically at several time points, and growth rates of parental and EMR cells were calculated from exponentially fitted growth curves.
Flow cytometry was performed to analyze the cell-cycle profiles of parental and EMR cells. Collected cells were fixed in 70% ethanol at –30°C for more than 30 min. After fixation, cells were centrifuged at 603 × g for 1 min, and the pellets were washed with PBS containing 1% BSA (bovine serum albumin; A9647-100G, Sigma-Aldrich). After centrifugation, the pellets were resuspended in 600 μL PBS containing 1% BSA, ribonuclease A (50 μg/mL; 10109169001, Sigma-Aldrich), and propidium iodide (40 μg/mL; P4170-10MG, Sigma-Aldrich), and incubated for 30 min at 37°C in the dark. Samples were analyzed using a BD Accuri C6 Plus flow cytometer (BD Biosciences, San Jose, CA, USA). At least 20,000 cells were acquired per sample, and DNA content histograms were generated. The percentages of cells in G0/G1, S, and G2/M phases were determined using ModFit LT software (Verity Software House, Topsham, ME, USA).
Single-nucleosome imagingParental and EMR cells stably expressing H2B-Halo were cultured on 35-mm dishes (153066, Nunc). H2B-Halo incorporated into nucleosomes was fluorescently labelled with HaloTag TMR ligand (8251, Promega) at a low concentration (100 pM) for 20 min at 37°C in 5% CO2 and then washed once with 1× PBS. Cells were trypsinized, resuspended in phenol red–free DMEM (21063-029, Thermo Fisher Scientific) supplemented with 10% FBS, and seeded onto fibronectin-coated glass-bottom dishes (3970-035, IWAKI) (fibronectin 354008, Corning).
During live-cell imaging, a stage-top incubator (INU-TIZ-F1, Tokai Hit) and digital gas mixer (GM-8000, Tokai Hit) were used to maintain culture conditions (37°C, 5% CO2, and humidity). Single nucleosomes were observed using an inverted Nikon Eclipse Ti microscope equipped with a 100-mW Sapphire 561-nm laser (Coherent) and an sCMOS camera (ORCA-Flash 4.0 or ORCA-Fusion BT; Hamamatsu Photonics). H2B-Halo–TMR–labelled nucleosomes were excited with the 561-nm laser through a 100× PlanApo TIRF objective (NA 1.49; Nikon), and emitted fluorescence was collected between 575 and 710 nm. An oblique illumination system with a TIRF unit (Nikon) was used to excite fluorescent nucleosomes within a thin optical section of the nucleus and to reduce background (Fig. 3A and C). Sequential images were acquired at 50 ms per frame under continuous illumination using MetaMorph (Molecular Devices) or NIS-Elements AR (Nikon).
Single-nucleosome tracking analysisImage processing, single-nucleosome tracking, and movement analysis were performed as described previously (Iida et al., 2022; Nozaki et al., 2017), with minor modifications. Sequential images were converted to a 16-bit grayscale, and background signals were subtracted using the rolling-ball algorithm (radius 50 pixels) in Fiji/ImageJ (Schindelin et al., 2012). Nuclear regions were manually selected. The positions of fluorescent dots in each frame were determined and their trajectories were reconstructed using the u-track MATLAB package (Jaqaman et al., 2008).
We previously estimated the localization accuracy of H2B-Halo nucleosomes by analyzing immobilized nucleosomes in formaldehyde-fixed RPE-1 cells (Shimazoe et al., 2025), and found a standard deviation of 2D displacement of 8.9 nm per 50 ms. For single-nucleosome movement analysis, displacement distributions and mean square displacement (MSD) as a function of time lag were calculated from the trajectories using in-house Python scripts. MSD was initially calculated in 2D; to obtain an apparent 3D MSD value, the 2D MSD was multiplied by 1.5 (corresponding to 4–6Dt). Graphs and statistical analyses of MSD under various conditions were performed using R.
Bayesian-based Richardson-Lucy algorithm (RL algorithm)To increase the number of tracked nucleosomes when applying the RL algorithm for motion classification, H2B-Halo was labeled with 50 nM PA-JF646 (provided by the Lavis Lab, Janelia Research Campus, VA, USA) (Grimm et al., 2016) overnight using the same labeling procedure. We obtained a 10-fold increase in H2B-Halo nucleosome trajectory data using H2B-Halo-PA-JF646 and constant photoactivation with weak 405 nm illumination.
The RL algorithm is the iterative algorithm of Richardson (Richardson, 1972) and Lucy (Lucy, 1974) (RL) to derive smooth distributions from the noisy data used in image processing (Dey et al., 2006; Laasmaa et al., 2011). With sufficient observation, the algorithm converts raw, noisy trajectories into a smooth distribution of MSD, providing a denoised, population-level view of heterogeneous nucleosome mobility. A detailed description of the algorithm has been reported previously (Ashwin et al., 2019). The obtained distribution curves were fitted with multiple Gaussian functions, and the peak position and area of each Gaussian component were calculated. For MSD analysis, trajectories within each cell were categorized based on their MSD at Δt = 0.5 s, and MSD was calculated separately for each subpopulation.
This work was supported by the Japan Society for the Promotion of Science (JSPS) and MEXT KAKENHI grants (JP23K17398, JP22H04925 (PAGS), and JP24H00061), and the Takeda Science Foundation to K Maeshima. AO is a SOKENDAI Special Researcher (JST SPRING JPMJSP2104). K Minami was a SOKENDAI Special Researcher (JST SPRING JPMJSP2104) and a JSPS Fellow (JP23KJ0998). MAS is a JSPS Fellow (JP24KJ1161).
Conflict of Interest StatementThe authors declare that they have no competing interests, financial, or otherwise.
Data Availability StatementAll data supporting the findings of this study are included in the article and its Supplementary Information. Source data underlying the graphs (MSD values and subpopulation fits), imaging parameters, and representative raw image sequences are available from the corresponding author upon reasonable request. The scripts used for RL-algorithm classification are available at https://doi.org/10.5281/zenodo.16959115
Author Contribution StatementAO, MAS, and K Maeshima designed the research. AO performed most of the experiments, including cell generation, imaging, and analyses. MAS performed the RL analysis. K Minami provided methodological expertise in single-nucleosome imaging and data analysis. ST contributed to immunofluorescence experiments and prepared illustrations. AO, SW and FT performed mouse experiments. TK contributed to the development of EMR cells. AO, MAS, K Minami and K Maeshima wrote the manuscript with input from all authors.
Ethics Approval and Consent to ParticipateAll animal experiments were approved by the National Cancer Center Animal Ethics Committee (approval no. T24-016-M01) and conducted in accordance with institutional guidelines. This study used established human cell lines only and did not involve human participants.
Patient Consent for PublicationNot applicable.
We are grateful to Dr. T. Yugawa for invaluable advice, Dr. K. M. Marshall and Mr. Y. Nagata for critical reading of this manuscript. We thank Dr. H. Kimura for providing antibodies, and Dr. S. Ide, Dr. S. Iida and Dr. K. Hibino for technical assistance and advice throughout this study. Finally, we thank Dr. K. Saito, Dr. M. T. Kanemaki, Dr. K. Higashi, Dr. A. Kawaguchi, Dr. J. Kitano, Dr. A. Kimura, Dr. K. Tanaka and all members of the Maeshima laboratory for valuable discussions and continuous support.
