Cancer cells have heterogeneous fitness, and this heterogeneity stems from genetic and epigenetic sources. Here, we sought to assess the contribution of asymmetric mitosis (AM) and time on the variability of fitness in sister cells. Around one quarter of sisters had differences in fitness, assessed as the intermitotic time (IMT), from 330 to 510 min. Phenotypes related to fitness, such as ERK activity (herein referring to ERK1 and ERK2, also known as MAPK3 and MAPK1, respectively), DNA damage and nuclear morphological phenotypes were also asymmetric at mitosis or turned asymmetric over the course of the cell cycle. The ERK activity of mother cell was found to influence the ERK activity and the IMT of the daughter cells, and cells with ERK asymmetry at mitosis produced more offspring with AMs, suggesting heritability of the AM phenotype for ERK activity. Our findings demonstrate how variabilities in sister cells can be generated, contributing to the phenotype heterogeneities in tumor cells.

Non-genetic heterogeneity in a population of tumor cells is key to explaining several aspects of cancer biology and specially the ineffectiveness of cancer treatments (Wilting and Dannenberg, 2012). Genetically identical cells present individual variability in gene expression profiles and drug metabolism, promoting the generation of a momentary state of tolerance to treatment, which contributes to the survival and eventual stable resistance of that cell or its descendants (Brock et al., 2009; Ramirez et al., 2016). Instability and stochasticity during the production of cellular substrates in individual cells favor the emergence of genetic alterations producing phenotypes that confer resistance to therapy (Marusyk et al., 2020). Single-cell studies provide evidence that single cells are highly variable at different levels, from transcription and signaling of a single protein to the general state of the cell (phenotype) (Batchelor et al., 2009, 2011; Lahav et al., 2004; Lenz et al., 2021). The contribution of genetic and all the other non-genetic processes to the generation of heterogeneity are still far from clear, but the importance of the non-genetic processes are relevant (Gookin et al., 2017; Granada et al., 2020; Lawson et al., 2018; Lenz et al., 2022).

Asymmetric mitosis (AM) is a classic event that contributes to cellular heterogeneity given that it produces two daughter cells with different characteristics (Venkei and Yamashita, 2018). After undergoing an AM, daughter cells might present a difference in terms of fitness, both in prokaryotes and eukaryotes (Higuchi-Sanabria et al., 2014; Lindner et al., 2008). Through AM, glioblastoma stem cells segregate more p75NTR (also known as NGFR) and EGFR to one of the sister cells promoting resistance to therapy with EGFR inhibitors (Hitomi et al., 2021; Lathia et al., 2011). Although AM is a well described feature of cancer stem cells (CSCs), it can also occur in non-CSCs, and thus is not exclusive of this cell state (Kaseb et al., 2016). The ERK (herein referring to ERK1 and ERK2, also known as MAPK3 and MAPK1, respectively) pathway is classically known to regulate cell proliferation and survival, integrating different signal sources with multiple functions (Jacques-Silva et al., 2004; Lenz et al., 2000). Technologies that allow the monitoring of ERK activity in live cells (Kudo et al., 2018; Pargett et al., 2017; Regot et al., 2014), have shown that ERK activity is highly dynamic (Chavez-Abiega et al., 2022; Ryu et al., 2016) and influences cell outcome (De et al., 2020). We have previously shown that fitness is highly dynamic in cancer cells and two to four generations are enough to produce the ERK variability in a colony of cells found among randomly grouped cells in the population (Lenz et al., 2021).

In this work, we assessed the fitness difference in sister cells and determined how much of it is acquired through AM or develops over the progression of the cell cycle, focusing on measuring ERK signaling activity, which is linked to fitness. Furthermore, knowing that the phenotype of daughter cells can be affected by the phenotypic status of the mother cell (Lenz et al., 2021; Min et al., 2020), we also questioned whether mother cell ERK activity can influence the behavior of daughter cells.

Asymmetric mitosis contributes to fitness heterogeneity in cancer

Cell fitness can be considered the most important phenotype in cancer biology, as cancer is a disease of excess cell fitness (Dey-Guha et al., 2015; Lenz et al., 2021; Silva et al., 2016). Therefore, we evaluated whether mitosis is able to generate fitness heterogeneity. For this we measured the difference (Δ) in intermitotic time (IMT) in pairs of sister cells after mitosis. If values of the Δ distribution rejected normality after a normal Shapiro–Wilk distribution test, they were considered asymmetric. In this case, sister cells with Δ values above the third quartile (>Q3) were considered asymmetric (Fig. 1A). Confirming the fitness variability among sister cells, a bimodal distribution was observed for IMT in sister cells (Fig. 1B). Whereas the symmetric sisters had a ΔIMT range from 0 to 270 min, the asymmetric sisters had a ΔIMT from 330 to 510 min, confirming that different fitness occurs in a proportion of sister cells. However, the difference observed in sister cells is still smaller than the difference observed in randomly paired cells from the same set of cells (non-sister) (Fig. S1A). A similar separation of the cells considered asymmetric was obtained with the threshold of one standard deviation above the mean (Fig. S1B). Tracking the expression of endogenous Ki-67 (also known as MKI67) through YFP tagging indicated that two out of five sister pairs had different progression through the cell cycle (Fig. 1C).

Fig. 1.

Variability of sister cells after mitosis. (A) Distribution analysis performed for each difference (Δ) between pairs of sister cells after mitosis for a given phenotype. Phenotypes that rejected normality (Shapiro–Wilk P<0.05) were separated in quartiles and the pairs above the third quartile (Q3) were considered asymmetric. An unpaired two-tailed Mann–Whitney test was performed to confirm the separation of asymmetric sister cells from symmetric sister cells. (B) ΔIMT in minutes in sister after mitosis of A172ERK-53BP1 glioma cells. (C) Mean fluorescence intensity obtained through measuring the expression of YFP integrated in the endogenous locus of the Ki-67 gene in U251 cells over time. Sisters with the same Ki67 behavior are painted with the same gray color. Δ number of 53BP1 foci (D) Δ cell area (E), Δ nuclear area (F) and nuclear roundness (G). For box plots, the box represents the 25–75th percentiles, and the median is indicated. The whiskers show the range (n=31 pairs of sister cells). P-values shown on B, D–G were calculated with a Mann–Whitney test.

Fig. 1.

Variability of sister cells after mitosis. (A) Distribution analysis performed for each difference (Δ) between pairs of sister cells after mitosis for a given phenotype. Phenotypes that rejected normality (Shapiro–Wilk P<0.05) were separated in quartiles and the pairs above the third quartile (Q3) were considered asymmetric. An unpaired two-tailed Mann–Whitney test was performed to confirm the separation of asymmetric sister cells from symmetric sister cells. (B) ΔIMT in minutes in sister after mitosis of A172ERK-53BP1 glioma cells. (C) Mean fluorescence intensity obtained through measuring the expression of YFP integrated in the endogenous locus of the Ki-67 gene in U251 cells over time. Sisters with the same Ki67 behavior are painted with the same gray color. Δ number of 53BP1 foci (D) Δ cell area (E), Δ nuclear area (F) and nuclear roundness (G). For box plots, the box represents the 25–75th percentiles, and the median is indicated. The whiskers show the range (n=31 pairs of sister cells). P-values shown on B, D–G were calculated with a Mann–Whitney test.

Cell phenotypes linked to fitness, such as number of 53BP1 (also known as TP53BP1) foci, an indicator of hyper-replication and DNA double-strand breaks (DSBs) (Mirman and de Lange, 2020) were also asymmetric (Fig. 1D). Morphological phenotypes, such as cell area, nuclear morphometrical parameters (area, roundness and perimeter) also showed a bimodal distribution (Fig. 1E–G; Fig. S1C). However, not all morphometric parameters showed high variability in sister cells after mitosis, as the nuclear irregularity index (NII) (Filippi-Chiela et al., 2012), aspect, extent and radius ratio had a normal distribution according to the normality test (Fig. S1D). In addition, it is important to mention that only one pair of sisters was asymmetric for all the measured features, indicating that asymmetry is not a phenomenon applied to all phenotypes (Fig. S1E).

Mitotic and temporal variability of ERK activity in sister cells

Given that IMT is a phenotype that can only be assessed at the completion of the cell cycle, we opted to focus our attention on the ERK signaling status, as this signaling pathway is linked to cell fitness and shows significant variability in isogenic cells (Filippi et al., 2016). ERK activity was measured in live cells through the use of a kinase translocation reporter (KTR) (Kudo et al., 2018; Regot et al., 2014). ERK activity in the last 5 h before mitosis correlated negatively with the mean IMT of the daughter cells (Pearson r: −0.5074, P<0.05) (Fig. 2A), whereas the ERK activity just after mitosis did not correlate with the IMT of the cell itself (Fig. 2B), indicating that the ERK activity of the mother cell is a predictor of the average daughter cell fitness and confirming the importance of the activity of this pathway in the mother cells for the decisions of the daughter cells in the next cell cycle (Min et al., 2020; Stern et al., 2021 preprint). Sister cells underwent AM for ERK activity, showing a bimodal distribution for Δ values (Fig. 2C,D). The degree of asymmetry in the sister cells was similar to the asymmetry observed a non-tumoral cell line, suggesting that having ERK asymmetries is not cancer cell specific (Fig. S2A). The degree of difference in ERK activity was similar in asymmetric sister cells and asymmetric non-sister cells in both cell lines (Fig. S2B). However, sisters asymmetric for ERK activity are not necessarily asymmetric for IMT (Fig. S2C).

Fig. 2.

Contribution of mitosis and time on ERK variability in sister cells. (A) ERK activity of the mother cell in the last 5 h (green in the diagram at the top) correlates negatively with the average IMT of the daughter cells. (B) ERK activity just after mitosis (green in the diagram at the top) does not correlate with the IMT of the cell. (C) Example of symmetric (left) and asymmetric (right) sisters for ERK activity. (D) Each point on the graph represents the difference (Δ) of the C/N ratio in sister cells pairs right after mitosis (sister with the highest C/N ratio−sister with the lowest C/N ratio). Asymmetric sisters were considered pairs of sisters that had differences (Δ ERK activity) above the third quartile (Q3). An unpaired two-tailed Mann–Whitney test was performed to confirm the separation of asymmetric sister cells from symmetric sister cells. For box plots, the box represents the 25–75th percentiles, and the median is indicated. The whiskers show the range (n=31 sister cell pairs). (E) Contribution of mitosis and time to the generation of ERK variability in sister cells. (F,G) The cell cycle was aligned from cell birth (0) to cell division (1). (F) Shows one pair of sisters that underwent AM, thus having high variability of ERK activity (Δ ERK activity) in the first frame (1st frame) after mitosis. (G) Shows one pair of sisters that underwent symmetric mitosis but showed variability over time, representing sisters with temporal variability. (H,I) ΔERK activity was calculated for sister cells that had more mitotic variability (green) (H) and for sisters that had more temporal variability (red) (I). The amplitude (A) in the graphs was obtained by determining the maximum – minimum Δ value of sisters observed in the first frame after mitosis and in the last cell cycle block.

Fig. 2.

Contribution of mitosis and time on ERK variability in sister cells. (A) ERK activity of the mother cell in the last 5 h (green in the diagram at the top) correlates negatively with the average IMT of the daughter cells. (B) ERK activity just after mitosis (green in the diagram at the top) does not correlate with the IMT of the cell. (C) Example of symmetric (left) and asymmetric (right) sisters for ERK activity. (D) Each point on the graph represents the difference (Δ) of the C/N ratio in sister cells pairs right after mitosis (sister with the highest C/N ratio−sister with the lowest C/N ratio). Asymmetric sisters were considered pairs of sisters that had differences (Δ ERK activity) above the third quartile (Q3). An unpaired two-tailed Mann–Whitney test was performed to confirm the separation of asymmetric sister cells from symmetric sister cells. For box plots, the box represents the 25–75th percentiles, and the median is indicated. The whiskers show the range (n=31 sister cell pairs). (E) Contribution of mitosis and time to the generation of ERK variability in sister cells. (F,G) The cell cycle was aligned from cell birth (0) to cell division (1). (F) Shows one pair of sisters that underwent AM, thus having high variability of ERK activity (Δ ERK activity) in the first frame (1st frame) after mitosis. (G) Shows one pair of sisters that underwent symmetric mitosis but showed variability over time, representing sisters with temporal variability. (H,I) ΔERK activity was calculated for sister cells that had more mitotic variability (green) (H) and for sisters that had more temporal variability (red) (I). The amplitude (A) in the graphs was obtained by determining the maximum – minimum Δ value of sisters observed in the first frame after mitosis and in the last cell cycle block.

Given our focus is on ERK activity of the last 5 h of the cell cycle, we asked whether the differences in ERK activity at the end of the cell cycle is due to asymmetry produced by mitosis or was produced over the progression through the cell cycle. Thus, we imaged cells every 30 min along the whole cell cycle of the two sister cells and observed 23% of sisters had a maximum difference (Δmax) right after mitosis (mitotic variability), whereas 33% attained the Δmax over the course of the cell cycle (temporal variability) (Fig. 2E, see Movies 1 and 2); 40% of sisters never attained ERK asymmetric status over one cell cycle. We also analyzed the nuclear area of sisters after mitosis and the variability was almost exclusively temporal, as expected due to the increase in nuclear area with the progression of the cell cycle (Fig. S2D) (Filippi-Chiela et al., 2012). The difference in ERK activity observed in sister cells right after mitosis is still smaller than the difference observed in non-sister cells (Fig. S2E). However, sister cells become more different from each other as they move through the cell cycle, turning them as different as non-sister cells (Fig. S2F), confirming the role of time in generating differences in the ERK activity status of sister cells.

To better understand the behavior of sisters that underwent AM on ERK activity status along the cell cycle, we separated pairs of cells into two groups: pairs with large mitotic variability (Fig. 2F,H) and pairs with large temporal variability (Fig. 2G,I). The ΔERK activity throughout the cell cycle was obtained by defining sister 1 as the cell with higher ERK activity and sister 2 as the cell with lower ERK activity at the first frame and maintaining the order ΔSis1-Sis2 for the whole cell cycle. We found that pairs of sister cells with large mitotic differences reduced their ΔERK during the cell cycle (Fig. 2H). We also observed that around 60% of sisters with symmetric mitosis developed differences over the course of the cell cycle (Fig. 2I) eventually having higher differences at the end of the cell cycle when compared to sister cells with AM, showing amplitude of 0.55 and 0.21, respectively (Fig. 2I; Fig. S2E,F). It is important to mention that the mean ERK activity of asymmetric sister cell pairs is equal to the average ERK activity of symmetric sister cell pairs over the cycle (Fig. S2G), indicating that symmetry is not due to very low or very high ERK activity. Finally, it is important to add that the differences found in pairs of sister cells does not correlate with the average ERK activity (Fig. S2H). Taken together, these data confirm the importance of time in increasing ERK heterogeneity in a subpopulation of sister cells, especially in producing large differences in ERK activity prior to mitosis.

Temporal variability influences the fitness of sister cells

Considering the existence of temporal variability in terms of ERK activity, next we asked whether mitotic variability also could generate asymmetrical sister cells for ERK activity in the next generation. We found that sister cells initially asymmetric for ERK activity had an ∼69% (9 of 13) AM in their next generation (Fig. 3A), whereas sister cell pairs initially symmetric for ERK activity had only 21% (6 of 28) AM (Fig. 3B). In addition, mitosis performed by initially asymmetrical sister cell pairs had a mean value of ΔERK activity after mitosis above the third quartile (Q3=0.1071, threshold defined for asymmetric sisters), when compared to mitoses performed by pairs of initially symmetrical sister cells (Fig. 3C). Sister cells that had an AM for ERK activity did not appear to produce the greater inter-sister variability (Fig. S3A), just as the mean IMT of sisters who inherited higher ERK activity was similar to the mean IMT of sisters who inherited lower ERK activity (Fig. S3B). Moreover, the variability acquired throughout the cell cycle did not produce AM in the next generation (Fig. 3D). These data show that AM, in terms of ERK activity, is partially heritable whereas temporal variability does not generate heritability. In addition, we observed that the greater the difference in ERK activity peaks in sister cell pairs, the greater the difference in IMT between sisters (Fig. S3C), suggesting that the IMT variability found in sister cells can also be explained, at least in part, by the variability of ERK activity throughout the cycle.

Fig. 3.

Heritability and variability of ERK on fitness of sister cells. (A,B) Each point on the graphs represents the difference (Δ) of ERK activity for each daughter generated from initially asymmetric (A) or symmetric (B) mothers for ERK activity. The dashed line represents the threshold defined for asymmetric sisters (Q3=0.1071). (C) Δ 2nd generation obtained by determining the value of maximum delta−minimum delta of (Δ) of ERK activity for each daughter generated from initially asymmetric. (D) Δ 2nd generation obtained by determining the value of maximum delta−minimum delta of (Δ) of ERK activity for each daughter generated from sisters with temporal asymmetry for ERK activity. (E) ΔERK activity, in absolute values, of the mother and each daughter cell compared to ΔERK activity observed in pairs of sister cells. (F) ERK activity average in pairs of sisters compared to the maximum ERK activity from mother cell in the last 5 h before mitosis. (G) IMT Conservation (IMTCon) and similarity of IMT (SisSymmetry) in sister cells. (H) ERK activity of cells treated or not with MEKi (trametinib) at 20 nM. (I,J) Average IMT in daughter cells (I) and the difference (Δ) of IMT (slower-faster) of daughter cells (J) that had mother cells treated or not with MEKi in the last block of the cell cycle. The data is expressed as mean±s.d. P-values shown in C–F, H–J were calculated with an unpaired two-tailed t-test (***P≤0.001; ****P≤0.0001; ns, not significant).

Fig. 3.

Heritability and variability of ERK on fitness of sister cells. (A,B) Each point on the graphs represents the difference (Δ) of ERK activity for each daughter generated from initially asymmetric (A) or symmetric (B) mothers for ERK activity. The dashed line represents the threshold defined for asymmetric sisters (Q3=0.1071). (C) Δ 2nd generation obtained by determining the value of maximum delta−minimum delta of (Δ) of ERK activity for each daughter generated from initially asymmetric. (D) Δ 2nd generation obtained by determining the value of maximum delta−minimum delta of (Δ) of ERK activity for each daughter generated from sisters with temporal asymmetry for ERK activity. (E) ΔERK activity, in absolute values, of the mother and each daughter cell compared to ΔERK activity observed in pairs of sister cells. (F) ERK activity average in pairs of sisters compared to the maximum ERK activity from mother cell in the last 5 h before mitosis. (G) IMT Conservation (IMTCon) and similarity of IMT (SisSymmetry) in sister cells. (H) ERK activity of cells treated or not with MEKi (trametinib) at 20 nM. (I,J) Average IMT in daughter cells (I) and the difference (Δ) of IMT (slower-faster) of daughter cells (J) that had mother cells treated or not with MEKi in the last block of the cell cycle. The data is expressed as mean±s.d. P-values shown in C–F, H–J were calculated with an unpaired two-tailed t-test (***P≤0.001; ****P≤0.0001; ns, not significant).

Influence of the mother cell on the fitness of daughter cells

Heritability of phenotypes, including asymmetries, is key for understanding the generation of heterogeneity in cancer cells (Lenz et al., 2022). We found that sister cells are more similar to each other than to the mother cell regarding ERK activity (Fig. 3E). However, maximum ERK activity of the mother cell in the last 5 h before mitosis correlated with the average ERK activity in daughter cell pairs (Pearson r: 0.43, P<0.05) (Fig. 3F), which is in line with the impact of ERK activity at the end of mitosis on IMT (Fig. 2A). If the ERK activity of the whole cell cycle of the mother is considered, the correlation with the daughter IMTs does not reach statistical significance (Fig. S3D), reinforcing the importance of the last hours of the cell cycle of the mother for the IMT of the daughters. As observed for ERK activity, sister cells were much more similar to each other, in terms of IMT, than to their mother cells, showing sister symmetry (SisSymmetry) values that ranged from 0.59 to 0.99, whereas IMT conservation (IMTCon) values ranged from 0.68 to 6.7 (Fig. 3G). As expected, cells with lower IMT than their mother (IMTCon<1) progressed more homogeneously through the cell cycle than those with IMTCon>1, as one or two of the sister cells in this case did not progress directly through the G1 phase (Fig. 3G). Corroborating this, there was no correlation between the IMT of the mother cell and the average IMT of sister cell pairs (Fig. S3E). On the other hand, as expected, there was a correlation of IMT between pairs of sister cells (Spearman r: 0.8, P<0.01) (Fig. S3F). In addition, cousin cells also showed correlation to IMT values (Fig. S3G), supporting the cousin–mother inequality hypothesis (Chakrabarti et al., 2018). Taken together, these data suggest that the IMT of mother cells is not heritable and does not predict the symmetry of IMT in sister cells. However, the IMT in sister cells can be affected both by the ERK activity of the mother cell (Fig. 2A), and by the temporal variability of ERK among sister cells (Fig. S3C).

In order to confirm that the ERK activity of the mother cell in the last block of the cell cycle (more precisely the last 5 h before division) influences the IMT of the daughter cells, mother cells were treated with the MEK inhibitor trametinib (MEKi) during the last block of the cell cycle and the IMT of daughter cells was evaluated. MEKi reduced the average ERK activity (Fig. 3H) and this inhibition in the mother cell resulted in daughter cells with a higher average IMT when compared to untreated cells (Fig. 3I). Furthermore, after ERK inhibition in mother cells, daughter cells showed a greater difference in IMT among themselves (Fig. 3J), suggesting that a high ERK activity of mother cells normally led to reduced differences in IMT in the daughter cells.

Non-genetic heterogeneity has been considered an important event for drug resistance in cancer (Bell and Gilan, 2020) and AM is part of the mechanism that produces this heterogeneity (Berge et al., 2019; Inaba and Yamashita, 2012). Phenotype variability and plasticity favor the emergence of phenotypic states that are resistant to treatments, resulting in its ineffectiveness (Pisco et al., 2013). Sister cells presented differences in IMT and in endogenous Ki-67 levels over the progression of the cell cycle. Ki-67 variability is associated with cell cycle regulation, and low levels of Ki-67 indicates G0 entry (Sobecki et al., 2017). This suggests that the larger differences in IMT of the sister cells is due to the entry of one sister in G0, whereas the other sister progresses through G1 without arrest.

In glioblastoma cells, the increase in EGFR and neurotrophin receptor (p75NTR) in one of the daughter cells through AM results in resistance to inhibition of growth factor receptors (Hitomi et al., 2021). In this work, we show that cells can undergo AM and produce asymmetric daughter cells for fitness and for phenotypes related to fitness, such as ERK activity and DNA damage. Proliferative heterogeneity has been described in sister cells and can be driven by AKT, producing AKTlow and AKTnormal daughter cells (Dey-Guha et al., 2011, 2015). Thus, we add that AM contributes to fitness heterogeneity in glioblastoma sister cells but does not seem to be a master regulator of asymmetry for the analyzed phenotypes. We also added that AM can be heritable to the next generation, ensuring ERK heterogeneity in subsequent generations.

Additionally, we found that the difference in ERK activity observed in sister cells from normal and cancer cell lines, after mitosis, is still smaller than the difference observed in non-sister cells, suggesting that, although mitosis contributes to the generation of differences in sister cells, some information is transmitted from the mother cell to the daughter cells. It has already been demonstrated that sister cells show high correlation in terms of mean transcriptional activity, which is inherited from mother cells to daughter cell (Phillips et al., 2019). However, we observe that the average ERK activity of the mother cells is not inherited by their daughter cells. Notwithstanding, the maximum ERK activity in the mother cell in the last 5 h before mitosis was correlated with the average ERK activity in daughter cell pairs and the fitness of the daughter cells. This agrees with the time frame of during which MEK inhibition in the mother cell cycle can affects the percentage of proliferative daughter cells (CDK2inc) found previously (Min et al., 2020). Surprisingly, inhibition of the ERK pathway at the end of G2, besides increasing the IMT of the next cycle, also increased the differences between sister cells, suggesting that high activity at the end of the cycle leads to more homogeneous sisters regarding fitness.

Besides AM, phenotypic dynamics over time can also influence the outcome of single cells (Lane et al., 2017). ERK signaling is widely variable even in isogenic cells (Filippi et al., 2016) and its dynamics influence the outcome of the cell (De et al., 2020). We previously demonstrated that two to four generations are enough to cause the ERK variability found in a colony of cells to be similar to the variability of randomly grouped cells (Lenz et al., 2021). Here, we observed that sister cells turned asymmetric over the course of the cell cycle, highlighting the role of time in generating differences in the ERK activity status of sister cells. Therefore, we add that ERK heterogeneity, in addition to being acquired through AM, can be developed throughout cell cycle in sister cells. Moreover, we add that ERK dynamics in sisters was correlated with ΔIMT in sister cells, suggesting that the fitness variability found in sister cells can also be explained, at least in part, by the variability of ERK activity throughout the cycle.

Taken together, our findings demonstrate that although mitosis contributes to ERK differences in sister cells, the variability in ERK activity during the progression of the cell cycle is mainly responsible for producing the ERK differences relevant for IMT in sister cells. Our data also further reinforce the importance of understanding the role of mitosis and time in generating variabilities, especially in cancer, in order to reduce the heterogeneity of phenotypes.

Cell culture, fluorescent labels and reagents

A172, MRC-5 and U251 glioma cell lines obtained from American Tissue Culture Collection (ATCC) were cultured in DMEM LOW (Gibco #31600-034) supplemented with 10% fetal bovine serum (Laborclin #630111) and maintained at 37°C and 5% CO2 in a humidified incubator. A172 and MRC-5 cells were transduced with ERK-KTR fluorescent reporter (Kudo et al., 2018; Regot et al., 2014) and with the nuclear marker Apple53BP1trunc (Yang et al., 2015), producing A172ERK-53BP1 and MRC5ERK-53BP1. U251Ki-67 cells were constructed through tagging endogenously locus of Ki-67 with YFP through CRISPR-cas9 (Lenz et al., 2021; Stewart-Ornstein and Lahav, 2016). During all experiments, cell culture medium was supplemented with 1% penicillin and streptomycin (Gibco #15140-122) and 0.1% amphotericin B (Gibco #15290-018). Cells were regularly tested for Mycoplasma using the Mycoplasma Detection Kit MycoAlert (Lonza). Trametinib (MEKi, MedChem #HY-10999) was diluted in DMSO and used at 20 nM for 6 h.

Time-lapse microscopy, live-cell tracking and fluorescence quantification

Images of A172ERK-53BP1, MRC5ERK-53BP1 and U251Ki-67 cells were taken every 30 min with the 20× objective on an Incucyte® S3 System (Sartorius). The tracking of individual cells was performed manually. The ERK activity was quantified semi-automatically through software developed by our group (available upon request) that measured the relative cytoplasmic to nuclear fluorescence (C/N) ratio (Kudo et al., 2018; Regot et al., 2014). ERK activity peaks were considered values above the average cell ERK activity plus one standard deviation (s.d.). Ki-67 activity was quantified by the average yellow fluorescence intensity (YFP) in each cell after mitosis and nucleolus formation.

IMT, IMTconservation and SisSymmetry

Intermitotic time (IMT) was calculated in minutes from the birth of a cell to its next division. IMTCon in A172ERK-53BP1 cells were obtained by averaging the sister cell IMT divided by the mother cell IMT. The SisSymmetry values were obtained by determining the ratio between the sister with the lowest IMT (faster sister) and the sister with the highest IMT (slower sister).

Mitosis variability and asymmetric sisters

To perform the mitotic variability, sister cells were analyzed at the first frame after mitosis for the morphological phenotypes: cellular area, nuclear area, roundness, nuclear irregularity index (NII), aspect, extent, radius ratio, for a signaling phenotype (number of 53BP1 foci), ERK activity (ERK-KTR fluorescent reporter) and the fitness phenotype as intermitotic time (IMT). The difference of sister cell pairs (Δ) for each phenotype was calculated and represents the variability of sister cells after mitosis. The (Δ) values were submitted to the Shapiro–Wilk test to assess the normality of the data. In this work, phenotypes that accepted normality (Shapiro–Wilk P>0.05) represent phenotypes in which mitosis was not enough to generate variability in sister cells, that is, phenotypes that did not undergo AM. Phenotypes that rejected normality (Shapiro –Wilk P<0.05), in other words that present a bimodal distribution, represent phenotypes that performed AM, generating pairs of sister cells with greater variability among themselves (asymmetric sisters) in relation to other pairs of more similar sisters (symmetric sisters). Asymmetric sisters were considered pairs of sister cells that had differences (Δ values) above the third quartile (Q3). To confirm the separation in two groups, an unpaired two-tailed Mann–Whitney test was performed between ‘symmetrics' and ‘asymmetrics' in the graphs. ‘Symmetrics’ were composed of more homogeneous sister cell pairs (with lower Δ values) whereas ‘asymmetrics’ were composed of more heterogeneous sister cell pairs (with higher Δ values).

Cell cycle alignment, mitotic and temporal variability

Sister cell pairs were synchronized according to the cell cycle at values from 0 to 1, where 0 was the time the cell was born and 1 the time the cell divided. Then, each cell had its life cycle divided into 5 blocks (block 1: frames from 0 to 0.2; block2: from 0.2 to 0.4; block 3: from 0.4 to 0.6; block4: from 0.6 to 0.8 and block5: frames from 0.8 to 1). As a control, the cell cycle was also aligned considering block 1 larger than the others, simulating the G1 phase, and the results obtained were the same (data not shown). For ERK activity and nuclear area, the average of values obtained within each block was calculated, obtaining 5 values for each single cell. The difference between the averages of each block was calculated between pairs of sister cells (average of block1 sister 1 - average of block1 - sister 2) and the maximum difference (Δmax) found over the 5 blocks was established. Mitotic variability was calculated by the Δ first frame after mitosis and temporal variability was calculated by the Δmax – Δ first frame. The same threshold established to identify asymmetric sisters (Δ values above the third quartile - Q3) was used to obtain the percentage of cells that showed the mitotic (green) or temporal variability (red) in the Fig. 2. To assess ERK behavior throughout the cell cycle, the difference (Δ) of ERK activity in pairs of cells was obtained by the cell that had the greater ERK activity - the cell that had less ERK activity in the first frame after mitosis. The amplitude (A) in the graphs was obtained by maximum – minimum Δ value between sisters or non-sisters observed in the 1st frame after mitosis and in the last cell cycle block.

Asymmetrics in the second generation

Analyses of descendants of cells that had initially symmetrical or asymmetrical mitosis were performed. The difference (Δ) in ERK activity for each daughter generated from initially asymmetric or symmetric mothers for ERK activity were obtained. Then, the Δ2nd generation was obtained by determining the value of maximum delta−minimum delta of (Δ) of ERK activity for each daughter generated from initially asymmetric divisions. Similarly, the Δ 2nd generation was obtained by determining the value of maximum delta−minimum delta of (Δ) of ERK activity for each daughter generated from sisters with temporal asymmetry for ERK activity, that is, pairs of sisters that had Δ values above the third quartile in the last block of the cell cycle.

We wish to thank Ralph Weissleder for the Apple-53BP1trunc (Addgene plasmid #69531) and Markus Covert for the pLentiPGK Puro DEST ERK-KTR-Clover (Addgene plasmid #90227). We also wish to thank Dr Gallit Lahav (Harvard University) for the plasmid used for tagging Ki-67, and the Multi BioCell Lab of CBiot of image acquisition.

Author contributions

Conceptualization: J.H.B., K.R.B., G.L.; Methodology: J.H.B., L.S.L., D.T.; Software: L.C.P.; Validation: J.H.B.; Formal analysis: J.H.B., L.S.L., D.T., J.M.; Investigation: J.H.B., L.S.L., D.T.; Data curation: J.H.B., L.S.L., D.T., K.R.B., G.L.; Writing - original draft: J.H.B., L.S.L., D.T., K.R.B., G.L.; Writing - review & editing: J.H.B.; Visualization: J.H.B., G.L.; Supervision: G.L.; Project administration: J.H.B., G.L.

Funding

This work was supported by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPQ), FAPERGS-FAPESP (2019/15477-3) and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (Capes) funding agencies. J.H.B., L.S.L. and G.L. received CNPQ fellowships. L.C.P., D.T., J.M. and K.R.B. received CAPES fellowships.

Data availability

All relevant data can be found within the article and its supplementary information.

Batchelor
,
E.
,
Loewer
,
A.
and
Lahav
,
G.
(
2009
).
The ups and downs of p53: understanding protein dynamics in single cells
.
Nat. Rev. Cancer
9
,
371
-
377
.
Batchelor
,
E.
,
Loewer
,
A.
,
Mock
,
C.
and
Lahav
,
G.
(
2011
).
Stimulus-dependent dynamics of p53 in single cells
.
Mol. Syst. Biol.
7
,
488
.
Bell
,
C. C.
and
Gilan
,
O.
(
2020
).
Principles and mechanisms of non-genetic resistance in cancer
.
Br. J. Cancer
122
,
465
-
472
.
Berge
,
U.
,
Bochenek
,
D.
,
Schnabel
,
R.
,
Wehling
,
A.
,
Schroeder
,
T.
,
Stadler
,
T.
and
Kroschewski
,
R.
(
2019
).
Asymmetric division events promote variability in cell cycle duration in animal cells and Escherichia coli
.
Nat. Commun.
10
,
1901
.
Brock
,
A.
,
Chang
,
H.
and
Huang
,
S.
(
2009
).
Non-genetic heterogeneity – a mutation-independent driving force for the somatic evolution of tumours
.
Nat. Rev. Genet.
10
,
336
-
342
.
Chakrabarti
,
S.
,
Paek
,
A. L.
,
Reyes
,
J.
,
Lasick
,
K. A.
,
Lahav
,
G.
and
Michor
,
F.
(
2018
).
Hidden heterogeneity and circadian-controlled cell fate inferred from single cell lineages
.
Nat. Commun.
9
,
5372
.
Chavez-Abiega
,
S.
,
Grönloh
,
M. L. B.
,
Gadella
,
T. W. J.
, Jr
,
Bruggeman
,
F. J.
and
Goedhart
,
J.
(
2022
).
Single-cell imaging of ERK and Akt activation dynamics and heterogeneity induced by G-protein-coupled receptors
.
J. Cell Sci.
135
,
jcs259685
.
De
,
S.
,
Campbell
,
C.
,
Venkitaraman
,
A. R.
and
Esposito
,
A.
(
2020
).
Pulsatile MAPK signaling modulates p53 activity to control cell fate decisions at the G2 checkpoint for DNA damage
.
Cell Rep.
30
,
2083
-
2093.e5
.
Dey-Guha
,
I.
,
Wolfer
,
A.
,
Yeh
,
A. C.
,
G Albeck
,
J.
,
Darp
,
R.
,
Leon
,
E.
,
Wulfkuhle
,
J.
,
Petricoin
,
E. F.
,
Wittner
,
B. S.
and
Ramaswamy
,
S.
(
2011
).
Asymmetric cancer cell division regulated by AKT
.
Proc. Natl. Acad. Sci. USA
108
,
12845
-
12850
.
Dey-Guha
,
I.
,
Alves
,
C. P.
,
Yeh
,
A. C.
,
Salony
,
Sole
,
X.
,
Darp
,
R.
and
Ramaswamy
,
S.
(
2015
).
A mechanism for asymmetric cell division resulting in proliferative asynchronicity
.
Mol. Cancer Res.
13
,
223
-
230
.
Filippi-Chiela
,
E. C.
,
Oliveira
,
M. M.
,
Jurkovski
,
B.
,
Callegari-Jacques
,
S. M.
,
Da Silva
,
V. D.
and
Lenz
,
G.
(
2012
).
Nuclear morphometric analysis (NMA): screening of senescence, apoptosis and nuclear irregularities
.
PLoS One
7
,
e42522
.
Filippi
,
S.
,
Barnes
,
C. P.
,
Kirk
,
P. D. W.
,
Kudo
,
T.
,
Kunida
,
K.
,
Mcmahon
,
S. S.
,
Tsuchiya
,
T.
,
Wada
,
T.
,
Kuroda
,
S.
and
Stumpf
,
M. P. H.
(
2016
).
Robustness of MEK-ERK dynamics and origins of cell-to-cell variability in MAPK signaling
.
Cell Rep.
15
,
2524
-
2535
.
Gookin
,
S.
,
Min
,
M.
,
Phadke
,
H.
,
Chung
,
M.
,
Moser
,
J.
,
Miller
,
I.
,
Carter
,
D.
and
Spencer
,
S. L.
(
2017
).
A map of protein dynamics during cell-cycle progression and cell-cycle exit
.
PLoS Biol.
15
,
e2003268
.
Granada
,
A. E.
,
Jiménez
,
A.
,
Stewart-Ornstein
,
J.
,
Blüthgen
,
N.
,
Reber
,
S.
,
Jambhekar
,
A.
and
Lahav
,
G.
(
2020
).
The effects of proliferation status and cell cycle phase on the responses of single cells to chemotherapy
.
Mol. Biol. Cell
31
,
845
-
857
.
Higuchi-Sanabria
,
R.
,
Pernice
,
W. M. A.
,
Vevea
,
J. D.
,
Alessi Wolken
,
D. M.
,
Boldogh
,
I. R.
and
Pon
,
L. A.
(
2014
).
Role of asymmetric cell division in lifespan control in Saccharomyces cerevisiae
.
FEMS Yeast Res.
14
,
1133
-
1146
.
Hitomi
,
M.
,
Chumakova
,
A. P.
,
Silver
,
D. J.
,
Knudsen
,
A. M.
,
Pontius
,
W. D.
,
Murphy
,
S.
,
Anand
,
N.
,
Kristensen
,
B. W.
and
Lathia
,
J. D.
(
2021
).
Asymmetric cell division promotes therapeutic resistance in glioblastoma stem cells
.
JCI Insight
6
,
130510
.
Inaba
,
M.
and
Yamashita
,
Y. M.
(
2012
).
Asymmetric stem cell division: precision for robustness
.
Cell Stem Cell
11
,
461
-
469
.
Jacques-Silva
,
M. C.
,
Bernardi
,
A.
,
Rodnight
,
R.
and
Lenz
,
G.
(
2004
).
ERK, PKC and PI3K/Akt pathways mediate extracellular ATP and adenosine-induced proliferation of U138-MG human glioma cell line
.
Oncology
67
,
450
-
459
.
Kaseb
,
H. O.
,
Lewis
,
D. W.
,
Saunders
,
W. S.
and
Gollin
,
S. M.
(
2016
).
Cell division patterns and chromosomal segregation defects in oral cancer stem cells
.
Genes Chromosomes Cancer
55
,
694
-
709
.
Kudo
,
T.
,
Jeknić
,
S.
,
Macklin
,
D. N.
,
Akhter
,
S.
,
Hughey
,
J. J.
,
Regot
,
S.
and
Covert
,
M. W.
(
2018
).
Live-cell measurements of kinase activity in single cells using translocation reporters
.
Nat. Protoc.
13
,
155
-
169
.
Lahav
,
G.
,
Rosenfeld
,
N.
,
Sigal
,
A.
,
Geva-Zatorsky
,
N.
,
Levine
,
A. J.
,
Elowitz
,
M. B.
and
Alon
,
U.
(
2004
).
Dynamics of the p53-Mdm2 feedback loop in individual cells
.
Nat. Genet.
36
,
147
-
150
.
Lane
,
K.
,
Van Valen
,
D.
,
Defelice
,
M. M.
,
Macklin
,
D. N.
,
Kudo
,
T.
,
Jaimovich
,
A.
,
Carr
,
A.
,
Meyer
,
T.
,
Pe'er
,
D.
,
Boutet
,
S. C.
et al. 
(
2017
).
Measuring signaling and RNA-seq in the same cell links gene expression to dynamic patterns of NF-κB activation
.
Cell Syst.
4
,
458
-
469.e5
.
Lathia
,
J. D.
,
Hitomi
,
M.
,
Gallagher
,
J.
,
Gadani
,
S. P.
,
Adkins
,
J.
,
Vasanji
,
A.
,
Liu
,
L.
,
Eyler
,
C. E.
,
Heddleston
,
J. M.
,
Wu
,
Q.
et al. 
(
2011
).
Distribution of CD133 reveals glioma stem cells self-renew through symmetric and asymmetric cell divisions
.
Cell Death Dis.
2
,
e200
.
Lawson
,
D. A.
,
Kessenbrock
,
K.
,
Davis
,
R. T.
,
Pervolarakis
,
N.
and
Werb
,
Z.
(
2018
).
Tumour heterogeneity and metastasis at single-cell resolution
.
Nat. Cell Biol.
20
,
1349
-
1360
.
Lenz
,
G.
,
Gottfried
,
C.
,
Luo
,
Z.
,
Avruch
,
J.
,
Rodnight
,
R.
,
Nie
,
W.
and
Neary
,
J. T.
(
2000
).
P2Y purinoceptor subtypes recruit different MEK activators in astrocytes
.
Br. J. Pharmacol.
129
,
927
-
936
.
Lenz
,
L. S.
,
Faccioni
,
J. L.
,
Bracco
,
P. A.
,
Santos
,
J. A. F.
,
Pereira
,
L. C.
,
Buss
,
J. H.
,
Tamborindeguy
,
M. T.
,
Torgo
,
D.
,
Monteiro
,
T.
,
Mantovani
,
G. B.
et al. 
(
2021
).
Cancer cell fitness is dynamic
.
Cancer Res.
81
,
1040
-
1051
.
Lenz
,
G.
,
Onzi
,
G. R.
,
Lenz
,
L. S.
,
Buss
,
J. H.
,
Dos Santos
,
J. A.
and
Begnini
,
K. R.
(
2022
).
The origins of phenotypic heterogeneity in cancer
.
Cancer Res.
82
,
3
-
11
.
Lindner
,
A. B.
,
Madden
,
R.
,
Demarez
,
A.
,
Stewart
,
E. J.
and
Taddei
,
F.
(
2008
).
Asymmetric segregation of protein aggregates is associated with cellular aging and rejuvenation
.
Proc. Natl. Acad. Sci. USA
105
,
3076
-
3081
.
Marusyk
,
A.
,
Janiszewska
,
M.
and
Polyak
,
K.
(
2020
).
Intratumor heterogeneity: the Rosetta stone of therapy resistance
.
Cancer Cell
37
,
471
-
484
.
Min
,
M.
,
Rong
,
Y.
,
Tian
,
C.
and
Spencer
,
S. L.
(
2020
).
Temporal integration of mitogen history in mother cells controls proliferation of daughter cells
.
Science
368
,
1261
-
1265
.
Mirman
,
Z.
and
De Lange
,
T.
(
2020
).
53BP1: a DSB escort
.
Genes Dev.
34
,
7
-
23
.
Pargett
,
M.
,
Gillies
,
T. E.
,
Teragawa
,
C. K.
,
Sparta
,
B.
and
Albeck
,
J. G.
(
2017
).
Single-cell imaging of ERK signaling using fluorescent biosensors
.
Methods Mol. Biol.
1636
,
35
-
59
.
Phillips
,
N. E.
,
Mandic
,
A.
,
Omidi
,
S.
,
Naef
,
F.
and
Suter
,
D. M.
(
2019
).
Memory and relatedness of transcriptional activity in mammalian cell lineages
.
Nat. Commun.
10
,
1208
.
Pisco
,
A. O.
,
Brock
,
A.
,
Zhou
,
J. Y.
,
Moor
,
A. E.
,
Mojtahedi
,
M.
,
Jackson
,
D.
and
Huang
,
X.-S.
(
2013
).
Non-Darwinian dynamics in therapy-induced cancer drug resistance
.
Nat. Commun.
4
,
2467
.
Ramirez
,
M.
,
Rajaram
,
S.
,
Steininger
,
R. J.
,
Osipchuk
,
D.
,
Roth
,
M. A.
,
Morinishi
,
L. S.
,
Evans
,
L.
,
Ji
,
W.
,
Hsu
,
C.-H.
,
Thurley
,
K.
et al. 
(
2016
).
Diverse drug-resistance mechanisms can emerge from drug-tolerant cancer persister cells
.
Nat. Commun.
7
,
10690
.
Regot
,
S.
,
Hughey
,
J. J.
,
Bajar
,
B. T.
,
Carrasco
,
S.
and
Covert
,
M. W.
(
2014
).
High-sensitivity measurements of multiple kinase activities in live single cells
.
Cell
157
,
1724
-
1734
.
Ryu
,
H.
,
Chung
,
M.
,
Dobrzyński
,
M.
,
Fey
,
D.
,
Blum
,
Y.
,
Sik Lee
,
S.
,
Peter
,
M.
,
Kholodenko
,
B. N.
,
Li Jeon
,
N.
and
Pertz
,
O.
(
2016
).
Frequency modulation of ERK activation dynamics rewires cell fate
.
Mol. Syst. Biol.
12
,
866
.
Silva
,
A. O.
,
Dalsin
,
E.
,
Onzi
,
G. R.
,
Filippi-Chiela
,
E. C.
and
Lenz
,
G.
(
2016
).
The regrowth kinetic of the surviving population is independent of acute and chronic responses to temozolomide in glioblastoma cell lines
.
Exp. Cell Res.
348
,
177
-
183
.
Sobecki
,
M.
,
Mrouj
,
K.
,
Colinge
,
J.
,
Gerbe
,
F.
,
Jay
,
P.
,
Krasinska
,
L.
,
Dulic
,
V.
and
Fisher
,
D.
(
2017
).
Cell-cycle regulation accounts for variability in Ki-67 expression levels
.
Cancer Res.
77
,
2722
-
2734
.
Stern
,
A. D.
,
Smith
,
G. R.
,
Santos
,
L. C.
,
Sarmah
,
D.
,
Zhang
,
X.
,
Lu
,
X.
,
Iuricich
,
F.
,
Pandey
,
G.
,
Iyengar
,
R.
and
Birtwistle
,
M. R.
(
2021
).
Predicting individual cell division events from single-cell ERK and Akt dynamics
.
BioRxiv
2021.09.14.460349
.
Stewart-Ornstein
,
J.
and
Lahav
,
G.
(
2016
).
Dynamics of CDKN1A in single cells defined by an endogenous fluorescent tagging toolkit
.
Cell Rep.
14
,
1800
-
1811
.
Venkei
,
Z. G.
and
Yamashita
,
Y. M.
(
2018
).
Emerging mechanisms of asymmetric stem cell division
.
J. Cell Biol.
217
,
3785
-
3795
.
Wilting
,
R. H.
and
Dannenberg
,
J.-H.
(
2012
).
Epigenetic mechanisms in tumorigenesis, tumor cell heterogeneity and drug resistance
.
Drug Resist. Updat.
15
,
21
-
38
.
Yang
,
K. S.
,
Kohler
,
R. H.
,
Landon
,
M.
,
Giedt
,
R.
and
Weissleder
,
R.
(
2015
).
Single cell resolution in vivo imaging of DNA damage following PARP inhibition
.
Sci. Rep.
5
,
10129
.

Competing interests

The authors declare no competing or financial interests.

Supplementary information